The Urothelium: Life in a Liquid Environment

Abstract

The urothelium, which lines the renal pelvis, ureters, urinary bladder, and proximal urethra, forms a high-resistance but adaptable barrier that surveils its mechanochemical environment and communicates changes to underlying tissues including afferent nerve fibers and the smooth muscle. The goal of this review is to summarize new insights into urothelial biology and function that have occurred in the past decade. After familiarizing the reader with key aspects of urothelial histology, we describe new insights into urothelial development and regeneration. This is followed by an extended discussion of urothelial barrier function, including information about the roles of the glycocalyx, ion and water transport, tight junctions, and the cellular and tissue shape changes and other adaptations that accompany expansion and contraction of the lower urinary tract. We also explore evidence that the urothelium can alter the water and solute composition of urine during normal physiology and in response to overdistension. We complete the review by providing an overview of our current knowledge about the urothelial environment, discussing the sensor and transducer functions of the urothelium, exploring the role of circadian rhythms in urothelial gene expression, and describing novel research tools that are likely to further advance our understanding of urothelial biology.

In medical school, students are taught that the urothelium (or transitional epithelium) is a distensible epithelium that serves as a barrier to urine and pathogens. This review expands upon these basic insights by providing an up-to-date summary and discussion of several topics including 1) urothelial development and regeneration; 2) evidence that the urothelium can alter the water and solute composition of the urine during normal physiology and in response to overdistension; 3) new insights into how the urothelium maintains patency in the face of expansion, contraction, and other physical forces; 4) the latest developments in urothelial sensor/transducer function and circadian rhythms, including discussion of how these pathways may contribute to lower urinary tract dysfunction; and 5) a discussion of new tools and techniques that are advancing urothelial research.

I. INTRODUCTION

The goal of this review is to summarize insights into urothelial biology and function, emphasizing those findings that have occurred in the intervening decade since we last surveyed the field (358). Because of the abundant literature on the subject, we focus on the urothelium lining the bladder; however, when information is available, we also discuss the urothelium lining the renal pelvis, ureters, and proximal urethra. The review is divided into sections, with the first one after this introduction devoted to familiarizing the uninitiated reader with the urothelium and then discussing recent findings in urothelial development and regeneration (sect. II). The next section examines urothelial barrier function, including information about ion and water transport, and the function of tight junctions (sect. III). We also explore evidence that the urothelium can alter the water and solute composition of urine during normal physiology and in response to overdistension. The subsequent section is devoted to the tissue adaptions that accompany expansion and contraction of the lower urinary tract including general changes in cell and tissue shape, and more specifically, changes in umbrella cell apical surface area and in the diameter of its junctional ring (sect. IV). Finally, the last two sections review our current understanding of urothelial sensor and transducer functions (sect. V), followed by a discussion of tools and techniques that are likely to advance our understanding of urothelial biology (sect. VI). After each of these individual sections, we highlight unanswered questions and areas where future exploration is called for. These serve as natural stopping points for those readers that need to take a break, or for those who prefer to focus on select aspects of urothelial biology and function.

When possible, we highlight areas of clinical relevance, but those readers with specific interests in urothelial-associated pathologies including cancer, cystitis, lower urinary tract dysfunction, or therapeutics are directed to the following excellent review articles on these topics (24, 65, 69, 223, 407, 484, 630, 777). We also note that because gene products often have multiple aliases, we use the gene nomenclature formulated by the HUGO Gene Nomenclature Committee throughout, and which can be accessed at the following website: https://www.genenames.org. The nomenclature for rodent genes closely follows that for humans. Finally, we recommend that anyone interested in the urothelium should read the review by R. Marian Hicks (295), which some 45 yr later still provides elegant and useful insights into this intriguing tissue.

II. HISTOLOGY, DEVELOPMENT, AND REGENERATION OF THE UROTHELIUM

A. Classification of the Urothelium

The urothelium is a stratified epithelium comprised of three distinct cell layers: the superficial layer, populated by a single layer of umbrella cells; the intermediate cell layer, which can be one-to-several layers thick depending on species; and the basal cell layer, which is one cell layer thick (295) (FIGURE 1A). The urothelium is one of the eight or so epithelial tissue types found in the body, a classification system that is based on the morphology of the superficial-most cell layer (i.e., squamous, cuboidal, or columnar) and the number of apparent cell layers: simple, one cell-layer thick; stratified, multiple cell layers thick; or pseudostratified, in which the epithelium appears stratified, but all cells touch the basement membrane. The urothelium was classified by early histologists as a transitional epithelium because its morphological characteristics lay somewhere between a stratified columnar and a stratified squamous epithelium. However, the term transitional epithelium is considered by some to be a misnomer (554), as the urothelium in general, and the superficial cell layer in particular, undergoes dramatic morphological changes in response to fluid movements through the ureters or as the bladder fills and voids. As a result, the morphology of the outermost umbrella cell layer is variable, and depends on whether it is distended or not. In this review we use the term urothelium, coined by Melicow in 1945 (480), and which is often preferred by those biologists, mathematicians, bioengineers, and medical specialists who study this particular epithelium.

FIGURE 1.

Cell layers of the urothelium. A: tissue organization of mouse bladder urothelium visualized using transmission electron microscopy. The urothelium comprises a superficial umbrella cell layer, 1–2 layers of intermediate cells, and a basal cell layer. The image is false-colored to highlight the following cell types: BC, basal cells; bIC, binucleate intermediate cell; Cp, capillary; IC, intermediate cells; Pe, pericyte; UC, umbrella cells. B: en face view of umbrella cell (UC) layer. Three umbrella cells are false-colored to reveal their relationship to one another. The boxed region is magnified in the inset image. Inset: arrowheads label the zippered apical membrane that lies directly above the tight junction. Microplicae (Mp) are marked. C: scanning electron micrograph of a dislodged umbrella cell revealing the tops of newly exposed intermediate cells (labeled with closed white circles). (Electron microscope images provided by Wily G. Ruiz.)

There is some contention in the scientific literature about whether the urothelium is stratified or pseudostratified. This is largely driven by Petry and Amon’s original observation that the outermost umbrella cell layer of multiple species (including cat, dog, guinea pig, monkey, rabbit, and rat), as well as underlying cell layers of the urothelium, send “thin” projections that touch the basement membrane (569). By today’s standards the morphology looks poor, the “stretched” samples were prepared by physically pulling the tissue in fixative, and there was no indication if the reported umbrella cell projections were a rare or a common phenomenon. A more recent report also argues that human bladder umbrella cells extend basement membrane-directed protrusions (575). Again, the morphology of the urothelium in this study is not optimal, and in our own studies we find that the umbrella cell layer of human bladder epithelium is easily damaged, exposing intermediate cells, which as noted below do extend projections basally. A final example occurs in mice with experimentally induced spinal cord injury. Within one day of injury, KRT14⁺ basal cells (discussed in more detail below) send thin projections upwards that terminate at or near the bladder lumen; however, these projections are not observed in sham-treated animals (383). While the function of these projections is unknown, it could represent a mechanism for the interkinetic migration of nuclei from the basal to superficial region of the urothelium (767). These data do demonstrate that under pathological conditions some urothelial cells can span the entirety of the urothelium.

In contrast, we find the evidence against pseudostratification more compelling. For example, an electron microscopic analysis of 31 samples of well-preserved human bladder epithelium revealed that intermediate cells send slender projections that contact the basement membrane; however, no such projections are observed emanating from umbrella cells (343). Likewise, serial sectioning coupled with transmission electron microscopy (TEM) failed to detect any connections between the umbrella cells of carefully processed human ureter, or the bladders of several mammalian species (including pig, rat, sheep, and shrew) (300, 623). Again, the presence of thin projections from intermediate cells, but not from umbrella cells, were observed (623). Most recently, reporter mice, which express membrane-bound green fluorescent protein in UPK2⁺ cells, were used to identify thin projections emanating from a population of uroplakin⁺ intermediate cells that extend toward the basal cell layers (748). These projections may reflect the requirement that epithelial cells form contacts with the basement membrane before undergoing cell division (406, 613). Within this past decade, optimal fixation methods (i.e., freeze substitution) coupled with high-resolution, three-dimensional (3D) TEM tomography, as well as focused ion-beam scanning electron microscopy (SEM), have been exploited to characterize subapical vesicles in umbrella cells and to define the characteristics of the interstitial cells that reside below the urothelium (312, 533). If umbrella cells have connections to the basement membrane, these techniques would likely be revealing.

B. Cell Strata of the Urothelium

In this section, we discuss the individual strata that comprise the urothelium of the bladder, beginning with the superficial umbrella cell layer and ending with the basal cell layer.

1. The superficial umbrella cell layer

The superficial cell layer of the urothelium is in direct contact with the urinary space and comprises a single layer of long-lived (~200 days in rodents) (344), large polyhedral cells (up to ~150 µm diameter in distended bladders), which in rat bladders have a volume of ~6,000 µm³ (703) (FIGURE 1B). For comparison, an erythrocyte has a volume of ~100 µm³, a HEK cell has a volume of ~2,000 µm³, and the soma of a Purkinje cell in the cerebellum has a volume of ~14,000 µm³ (19, 91, 218). Superficial cells are terminally differentiated and have a well-defined apical-basolateral polarity. This is the only cell layer of the urothelium that forms identifiable tight junctions and adherens junctions, specialized regions of cell-cell contact that seal the intercellular space between adjacent epithelial cells (see discussion in sects. III and IV). The cells in this superficial layer are variously referred to as cover cells, facet cells (which reflects their polyhedral shape when viewed from above), superficial cells, or umbrella cells. The latter term, used as early as 1915 (378), and which we prefer, likely refers to the dome-shaped nature of the umbrella cell’s apical pole in its relaxed state, and the observation that each umbrella cell covers multiple underlying intermediate cells (see closed white circles in FIGURE 1C). In cross section, and depending on species, the basal region of the umbrella cell can have protrusions that interdigitate between underlying intermediate cells. The umbrella cells’ one or more nuclei are usually contained in a large central bulge that protrudes toward the underling cell layers. While rabbit umbrella cells have a single nucleus (705), those of mice, rats, and humans have two or more nuclei (295, 748) (see left-most umbrella cell in FIGURE 1A). The nuclei of umbrella cells are typically larger in diameter than those of intermediate and basal cells. This difference in size reflects polyploidy. For example, the binucleate mouse umbrella cell is octoploid (4n + 4n) or greater (417, 748). Wang et al. (748) recently showed in an elegant series of experiments that the precursor for the octoploid umbrella cell is a binucleate intermediate cell (2n + 2n) (see tan-colored intermediate cell in FIGURE 1A). They further showed that the increase in umbrella cell chromosome number from its 2n + 2n precursor likely occurs by way of endoreplication, a process in which the replication of DNA in S-phase occurs in the absence of intervening mitoses.

Umbrella cells have several additional specializations, many of which are observable using scanning electron microscopy (SEM) (FIGURE 1B). First, in partially stretched bladders, the borders of the umbrella cells become apparent, revealing the polyhedral nature of the cells. Second, these border regions are comprised of a remarkable structure called the “zippered apical membrane,” originally described using atomic force microscopy (381), and which comprises very short interdigitating “fingers” of apical membrane that form directly above the tight junction. While these fingers are not readily visible by SEM, the overall position of the tight junction is (see FIGURE 1B, inset). Third, the entire surface of the umbrella cell is covered by anastomosing, raised actin-rich ridges called microplicae (see FIGURE 1B, inset), which are also referred to as “hinges” when the urothelium is cut in cross section. Microplicae are found on a variety of epithelial cell types in the body (including epithelial cells lining the hard palette, nonciliated cells in the respiratory tract, and the intercalated cells of collecting ducts) (26, 31). Although their function remains unknown, they have been proposed to prevent abrasion in some settings. Despite their rigid appearance, the microplicae found on the epithelial cells lining the surface of zebrafish larvae are highly dynamic (395). If this is also true of umbrella cells is unknown. Interestingly, a glycosylated protein called urohingin (85 kDa) is specifically localized to the hinge regions of mammalian umbrella cells (428). However, its identity remains unknown, and there is no available information whether it is required for microplicae formation or dynamics.

Additional apical membrane specializations are observed using TEM and its variants. In cross section, the apical plasma membrane of the umbrella cell is scalloped, with hinge regions occurring at the peaks of the scallops (arrowheads in FIGURE 2A), and so-called “plaques” in the intervening regions (cyan colored in FIGURE 2A). Because of its insolubility in a subset of detergents, including Sarkosyl, the whole of the apical membrane can be readily purified (428). A defining feature of the plaque regions is the presence of an asymmetric unit membrane (AUM), in which the outer leaflet of the apical membrane appears twice as thick as the inner leaflet (FIGURE 2C, inset) (576). This thickening is the result of a closely packed, crystalline array of 16-nm “AUM particles” (FIGURE 2D), which are comprised of uroplakins and are oriented in a manner that places the majority of their mass in the outer leaflet of the plasma membrane (FIGURE 2, E and F) (490, 660). Each plaque region contains ~1,000–3,000 AUM particles.

FIGURE 2.

Hinges, plaques, and discoidal and/or fusiform-shaped vesicles (DFVs) at the apical pole of umbrella cells revealed by transmission electron microscopy (TEM). A: ultrastructure of the apical region of a mouse umbrella cell. The position of “hinges” (cross sections through microplicae) are marked with arrowheads, and the location of intervening plaque regions (Pl) are indicated by thin blue-colored lines. (Image provided by Wily G. Ruiz.) B: morphology of discoidal-shaped vesicles (DV) in rat umbrella cells. C: TEM of the apical surface of a mouse umbrella cell. The boxed region is magnified in the inset. The asymmetric unit membrane (AUM) in the plaque region is marked. Note the presence of a glycocalyx on the luminal surface of the plasma membrane. (Image provided by Steven Truschel.) D: freeze, deep-etch electron microscopy reveals the architecture of plaque regions (Pl) and associated hinges (Hi). The inset is a magnified view of the 16-nm AUM particles that make up the plaque regions. [Image, provided by Dr. John Heuser, is used by permission from Wiley-Blackwell and is from Apodaca (32).] E: cartoon depicting the generalized architecture of the uroplakins embedded in a lipid bilayer. Blue circles on the luminal side of the uroplakins are sites of N-linked glycosylation. F: arrangement of uroplakins in the 16-nm AUM particles. Note that UPK1A/UPK2 heterodimers comprise the inner “ring” of the AUM particles, while UPK3A/UPK1B heterodimers form the outer ring. Interactions between UPK2 and UPK3A bridge the two regions.

The chief and perhaps sole components of these AUM particles are the uroplakins, originally discovered of Dr. Tung-Tien Sun and colleagues (764). However, detergent-purified apical membranes also contain the proteolipid myelin-and-lymphocyte protein MAL (797), and as noted above urohingin. The uroplakins are a small family of transmembrane proteins that in mammals include UPK1A, UPK1B, UPK2 (also referred to as UPK2A in phylogenetic comparisons), UPK3A, UPK3B, and UPK3C, although other variants are found in other species (176, 241). The major uroplakins in the urothelium are UPK1A, UPK1B, UPK2, and UPK3A. UPK1A and UPK1B are members of the tetraspanin family of proteins, while UPK2 and UPK3A are single-pass, type 1 membrane proteins (FIGURE 2E). Structural studies, employing cryoelectron microscopy coupled with biochemistry, have established that AUM particles are comprised of two closely apposed rings: heterodimers of UPK1A with UPK2 form the inner ring, while the outer ring is formed by heterodimers of UPK1B with UPK3A (FIGURE 2F) (174, 309, 490, 707, 708). Since the publication of these early structural determinations, there have been dramatic advances in the hardware and software that allows for cryoelectron microscopy to illuminate structure at the atomic level (with a resolution of 2.5–3.0 Å) (131). As such, it will be intriguing to apply these new tools to gain further insights into uroplakin interactions and structure within AUM particles.

The function of uroplakins remains elusive, but they are proposed to contribute to the apical permeability barrier (311), and knockout (KO) mice lacking expression of Upk1b, Upk2, or Upk3a exhibit a number of abnormalities including small, poorly differentiated umbrella cells, vesicoureteral reflux, and an attendant hydronephrosis (3, 116, 310, 372). The Upk3a KO mice form a small number of plaques, which may result from compensation by Upk3b, which is expressed at low levels in the urothelium. Interestingly, Upk3b KO mice do not have an observable urothelial phenotype, perhaps because Upk3a expression compensates (608). The uroplakins, UPK1A in particular, serve as receptors for uropathogenic Escherichia coli (489). Zebrafish larvae with reduced expression of the UPK3A ortholog, known as upk3b (also referred to as upk3l, and recently reclassified as upk3d in phylogenetic studies), exhibit defects in their pronephros including loss of apical microvilli, loss of apical polarity, and altered ability to eliminate water (493). This supports previous genetic studies in humans that spontaneous mutations in UPK3A may lead to renal agenesis or dysplasia (336, 337). The use of uroplakins as differentiation markers is discussed further below, but those readers seeking additional information about uroplakins are directed to the following review articles (117, 358, 764).

The other striking feature of umbrella cells is the presence of a very large pool of subapical vesicles, which are comprised of two membrane sheets that are arranged to form a flattened disk that resembles a round flat bread (312). In mice, the subapical vesicles are called fusiform vesicles: in cross section, they are very flat and elongate, they have expanded ends, they lack a discernable lumen in the central region of the vesicle, and they can be arranged in stacks (FIGURE 2A) (799). In humans, swine, and rabbits, the subapical vesicles have a somewhat expanded lumen and are referred to as discoidal vesicles (343, 705, 761). Rat vesicles are intermediate in appearance (FIGURE 2B) (743). Regardless of their morphology, all of these subapical vesicles likely perform the same essential function in membrane and AUM delivery. Thus we collectively refer to them as discoidal and/or fusiform-shaped vesicles (DFVs). Serial sectioning, and more recently electron microscopic tomography, confirm that DFVs are individual entities and not connected to one another or to the plasma membrane (312, 705). In a manuscript that includes superb freeze deep-etch microscopy, Terada et al. (692) demonstrate that clusters of DFVs are surrounded by a meshwork of intermediate filaments, which associate with DFVs by way of short, branched actin filaments.

A major functional role of DFVs is to carry cargo to the apical surface of the umbrella cell, allowing for expansion of the apical membrane in response to bladder filling. This function and the underlying mechanics of vesicle fusion are discussed in greater detail in section IV below. The major cargoes of DFVs are the uroplakins, although it is likely that most apical membrane proteins use DFVs to traffic to the apical membrane, but this has not been shown definitively. While not a membrane protein per se, human growth hormone (hGH) is packaged into DFVs and secreted into the urinary space in transgenic mice engineered to express this protein, or when the umbrella cell layer is transduced with adenoviruses encoding hGH (237, 356, 360). Thus it appears that DFVs can also transport apical secretory cargoes. This possibility is further supported by evidence that bovine urothelium can natively secrete tissue-type plasminogen activator and urokinase (173); however, it has not been shown that this is DFV mediated. Other organelles found in the umbrella cell cytoplasm include the Golgi, multivesicular bodies (MVBs, also called late endosomes), lysosomes, autophagosomes, and the cytoskeleton, which are discussed in greater detail in the sections below.

2. The intermediate and basal cell layers

The cell layer directly underneath the umbrella cell layer is referred to as the intermediate cell layer. Depending on species, this cell layer can be one to several layers thick (FIGURE 1A) (295). These cells are sometimes described as being pyriform in shape (i.e., they have the appearance of an inverted pear). Intermediate cells, which are smaller in diameter than umbrella cells (~20 µm), are attached to one another and to the adjacent superficial and basal cell layers by desmosomes. While they express tight junction-associated proteins such as claudins and the cell-cell adhesion protein CDH1 (E-cadherin) (4, 520, 748), they do not form morphologically identifiable tight junctions or adherens junctions. Intermediate cells are typically mononucleate; however, as noted above a recent report describes a population of polyploid, binucleate intermediate cells (2n + 2n) in mice, which give rise to umbrella cells (748). Whether binucleate intermediate cells are present in other species is not known at this time. The final urothelial cell layer is the basal cell layer, which is positioned at the interface of the urothelium and the underlying lamina propria (FIGURE 1). These are smallish cells (~10 µm in diameter), which attach directly to the basement membrane via hemidesmosomes (16, 340, 341, 555), and attach to the overlying intermediate cells via desmosomes. Basal cells are in close proximity to capillary endothelial cells (328). Their potential role as “stem cells” is described below.

3. The urothelium lining the renal pelvis, ureters, and urethra

There is a tendency to speak of the urothelium as a singular tissue; however, the urothelium lining the renal pelvis and ureter is derived from mesoderm, while the urothelium lining the bladder and urethra is derived from endoderm (243). Not surprisingly then, there are differences in these urothelial tissues. We start with the renal pelvis, which ensheaths the kidney papilla (inner medulla) at the apex of the renal pyramid(s) and then doubles back to cover the adjacent regions of the kidney proper, eventually merging with the ureter (FIGURE 3) (190). The papilla is comprised in large part by the terminal segments of the medullary collecting ducts. The surface of the papilla is sieve-like and contains numerous openings, which are the terminal portions of the ducts of Bellini, a structure where the urine-filled collecting ducts converge and then release urine into the space surrounding the pelvis. In unipapillate mammals such as mice, a single papilla extends directly into the pelvis (e.g., see FIGURE 3). In the case of humans, whose kidneys have multiple papilla, each papilla projects into a funnel-shaped calyx, which then opens into the larger chamber-like renal pelvis.

FIGURE 3.

Location of urothelium in the lower urinary tract. Cartoon depicting the organization of the urothelium and underlying tissues in the renal pelvis, ureter, bladder, and proximal urethra. AT, adipose tissue; C, kidney cortex; D, detrusor; Fx, fornix; IM, inner medulla of kidney; IS, inner stripe, outer medulla of kidney; LP, lamina propria; M, mesangial cell layer; OS, outer stripe, outer medulla of kidney; Rs, rhabdosphincter; SE, simple epithelium; SkM, skeletal muscle layers; SM, smooth muscle layers; UO, ureteral orifices; Ut, urothelium. [Adapted from Dalghi et al. (156) and redrawn by Dennis R. Clayton.]

In the pelvis, the epithelium slowly transitions from simple, to pseudostratified, to stratified, to urothelium (FIGURE 3) (392, 601, 728). The kidney papilla itself is covered by a simple cuboidal-to-columnar epithelium (SE in FIGURE 3). A subset of these epithelial cells has apical microvilli, but most lack expression of uroplakins or KRT20 (cytokeratin-20), widely used differentiation markers of umbrella cells. This simple epithelium continues into the region called the fornix (Fx in FIGURE 3), a bend that forms a concavity between the apex of the papilla, the adjacent inner and outer stripe regions of the outer medulla, and the cortex of the kidney. The epithelium of the pelvis overlaying the inner stripe of the outer medulla is initially simple (the majority of cells have apical microvilli), transitions to pseudostratified, and then becomes bilayered as the epithelium enters the outer stripe region. Although some uroplakin-positive cells are found in the region of the simple epithelium, these cells lack DFVs. The superficial cells comprising the pseudostratified epithelium have microvilli, but do not express KRT20 or uroplakins, and thus lack AUM and have no DFVs (601). The superficial cells in the bilayered epithelium are KRT20 positive, but express relatively small amounts of uroplakins and do not form DFVs and do not assemble AUM plaques. Finally, beginning in the region of the pelvis that covers the outer stripe and continuing into the regions overlying the cortex, the bilayered epithelium gives rise to a classical three-layered urothelium that forms AUM, has DFVs, and expresses all four uroplakins and KRT20. Overall, the pelvis exhibits a diversity of epithelial types, which undergo a steady progression until they differentiate into a full-fledged urothelium. The functional roles of these epithelia are poorly understood, but some investigators have proposed that the simple epithelium lining the papilla may contribute to final urine formation (392). Otherwise, the function of these epithelia is likely to be related to forming a distensible and relatively impermeable barrier.

The urothelium lining the ureters is similar to that found in the bladder, in that it is stratified and comprised of three cell types (FIGURE 3). Some regions of the outer, superficial cell layer express uroplakins and form DFVs (86, 300, 727). In humans, there is heterogeneity in the urothelium lining the ureters, even in contiguous regions (300). These differences include the number of strata, the morphology of the cell layers, and the lysosome content and number of nuclei (from 2 to 8) in the umbrella cell layer. Some studies have directly compared the urothelium in the ureter and bladder and report differences in ultrastructure, biochemistry, and growth potential in culture (429, 593, 764). For example, the total uroplakin expression in the bladder urothelium is ~10-fold greater than that in the ureter (429). Moreover, uroplakin expression in the human ureter is variable: some regions are uroplakin⁺, while adjacent regions are not. Furthermore, not all umbrella cells in the ureters express a full complement of uroplakins, despite looking “normal” (593). In contrast, uroplakin expression is much more continuous in the bladder epithelium, and the majority of umbrella cells express all four major uroplakins. Finally, the urothelium from the bladder shows high growth potential in culture, while that of the ureter (and urethra) is relatively meager by comparison (429).

The urethra is a fibromuscular, tubelike structure that transports urine from the bladder to the external urethral orifice, where it is released (FIGURE 3) (1). In males, the region of the urethra nearest the neck of the bladder is called the proximal or prostatic urethra, and is lined by urothelium and intervening patches of stratified columnar epithelium. Distal to the proximal urethra is the membranous urethra, a segment in which the urothelium is replaced by a stratified columnar epithelium. This same epithelium extends into the spongy urethra, eventually transitioning to a nonkeratinizing stratified squamous epithelium at the end of the urethra. In females, the urothelium transitions to stratified columnar and then nonkeratinizing stratified squamous at its terminus. Those portions of the urethra that are lined by urothelium not only expresses uroplakins, but in the mouse they also expresses KRT1 and KRT10, which are absent in the urothelium lining the bladder, ureter, and renal pelvis (TABLE 1) (429).

Table 1.

Expression of cytokeratins in the urothelium lining the lower urinary tract

Species	Urothelial Cell Layer	Renal Pelvis	Ureter	Bladder	Proximal Urethra
Homo sapiens	UC	KRT5†, KRT7, KRT8/18, KRT19, KRT20† (14); KRT7, KRT18 (570)	KRT5, KRT7, KRT8/18, KRT19, KRT20† (14); KRT18, KRT20 (593)	KRT5, KRT7, KRT8/18, KRT19, KRT20 (14); KRT8 (702); KRT20 (283)	KRT7, KRT13, KRT18♂, KRT19 (294)
	IC	KRT4†, KRT5†, KRT6, KRT7, KRT8/18, KRT13†, KRT17, KRT19 (14); KRT7, KRT18 (570)	KRT5†, KRT7, KRT8/18, KRT13, KRT19 (14)	KRT4†, KRT5, KRT7, KRT13, KRT8/18, KRT19 (14); KRT8 (702); KRT20† (283)	KRT7, KRT13†, KRT19 (294)
	BC	KRT4, KRT5, KRT6, KRT7, KRT8/18, KRT13, KRT17†, KRT19 (14); KRT7, KRT8, KRT13, KRT19 (570)	KRT7, KRT8/18, KRT13, KRT17, KRT19 (14)	KRT4†, KRT5, KRT7, KRT8/18, KRT13, KRT17, KRT19 (14); KRT8, KRT17 (702)	KRT7, KRT13, KRT14♀, KRT17, KRT18♂, KRT19 (294)
	2-D PAGE of dissected tissue	KRT4, KRT5, KRT7, KRT8, KRT13, KRT18, KRT19 (5)	KRT4, KRT5, KRT7, KRT8, KRT13, KRT18, KRT19 (5)	KRT4, KRT5, KRT7, KRT8, KRT13, KRT18, KRT19 (5)	KRT4, KRT5, KRT7, KRT8, KRT13, KRT18, KRT19 (5)
Mus musculus	UC	KRT8/18 (387)		KRT7, KRT18*, KRT20 (201); KRT7, KRT20 (731); KRT7, KRT18, KRT20 (618); KRT7 (649)	K1, K10 (429); KRT13† (243)
	IC			KRT18*, KRT7 (201); KRT7 (731); KRT5 (239); KRT7 (618); KRT5 (639); KRT7 (434); KRT7 (649)	KRT1, KRT10 (429); KRT5, KRT13† (243)
	BC		KRT15† (687)	KRT18*, KRT7 (201); KRT7 (731); KRT5 (239); KRT7 (618); KRT5 (639); KRT7 (649); KRT14† (558)	KRT1, KRT10, KRT14 (429); KRT5, KRT14 (243)
Canis familiaris/Ovis aries	UC	KRT18 (462)
	IC	KRT18† (462)
	BC

Keratin expression in urothelium as reported by the indicated reference is shown. BC, basal cell layer; IC, intermediate cell layer; PAGE, polyacrylamide gel electrophoresis; UC, umbrella cell layer. *Reportedly expressed only during development. †Expression is found in scattered cells.

Interestingly, in response to vitamin A deficiency, patches of bladder urothelium undergo metaplasia (a process in which one differentiated tissue is replaced by another), giving rise to a keratinized, stratified squamous epithelium (429). The source of this metaplastic epithelium is highly likely to be the KRT1⁺ and KRT10⁺ urothelium, although lineage tracing experiments would remove any doubt. These results further support the notion that the urothelium is not a singular tissue, but instead it varies in potential and function depending on its location. Obviously, more research is needed to better understand the functional similarities and differences between the urothelium lining the different organs that comprise the lower urinary tract.

We end this section by mentioning a specialized neuroepithelial cell that is interspersed between the urothelial cells lining the proximal urethra (and are also found scattered within the epithelia lining other regions of the urethra). These so-called brush cells (also known as tuft cells or solitary chemosensory cell), originally described in the respiratory and gastrointestinal tracts, are so named because they characteristically have a tuft of short apical microvilli that projects into the urethral lumen (171, 388). Other than the urethra, they are absent in all other regions of the lower urinary tract. Urethral brush cells release acetylcholine and are closely allied with subjacent sensory nerve fibers positive for CHRNA3 (nicotinic cholinergic receptor α3 subunit). Intriguingly, they express the canonical taste transduction machinery required to sense bitter substances and application of denatonium, a bitter compound, to the urethra stimulates detrusor activity (171, 388). This stimulatory activity is partially blocked by an inhibitor of nicotinic acetylcholine receptors (i.e., mecamylamine). In addition to bitter compounds, isolated brush cells also respond to heat-inactivated uropathogenic Escherichia coli (171), to agonists of umami receptors (171), and also express mRNA for the epithelial sodium channel (ENaC) (349). Thus brush cells may play important roles in responding to bacterial infections, to the presence of bitter and umami compounds, and perhaps to the salt composition of the urine. It is unknown if brush cells affect the function of urothelial cells, or vice versa, but the intimate nature of their relationship–they share cell-cell contacts–makes this a likely possibility.

C. Urothelial Differentiation Markers

Cell-type specific markers are used as tools to define stem cells, identify other progenitor populations, and characterize cells that express genes specific to their cell and tissue type. The latter includes cells like umbrella cells, which do not undergo further differentiation. The utility of differentiation markers also extends to pathological conditions such as neoplasms where markers are used to define their likelihood of causing more severe metastatic disease and identifying treatment strategies (327). While potential urothelial differentiation markers are many, and include claudins (described below) and glycans (467), those that have been codified, mostly in mice, fall into a relatively small number of categories. These include uroplakins, cytokeratins, and signaling molecules/transcription factors including SHH, TP63, and FOX2A (FIGURE 4). As is the case for all such studies, they depend on the marker that is targeted, the antibody and sample preparation that are employed, and the species examined. For example, KRT5 marks basal and a subset of intermediate cells in mouse urothelium (239), but using the same antibody to probe rat tissues reveals large swaths of unlabeled cells (G. Apodaca, unpublished observations).

FIGURE 4.

Differentiation markers of the adult mouse urothelium. The differentiation markers expressed by the superficial (umbrella) cell layer, the intermediate cell layers, and the basal cell layer of the mouse urothelium.

One category of differentiation markers is the uroplakins, which as described above are expressed at high levels in umbrella cells. As such, uroplakins are good markers for the umbrella cell layer across species (FIGURE 4). However, in mice there is a relatively small population of intermediate cells, both mononucleate and binucleate, that also express uroplakins, UPK3A in particular (239, 748). These uroplakin-expressing intermediate cells are typically found in the layer of cells just below the umbrella cell layer and can give rise to umbrella cells when the superficial cell layer is damaged. However, UPK2 is reportedly expressed in all umbrella and intermediate cells of the mouse bladder, but not in basal cells (496). Again, the choice of marker and antibody is important when designing and interpreting such studies.

In the bladder, the majority of uroplakins are expressed almost exclusively in the urothelium, but there is a growing appreciation that uroplakin proteins are also expressed in nonurothelial cell types. These include the following: the collecting ducts of the renal medulla (UPK1B), mesothelial cells and anterior prostate (UPK1B), stomach parietal cells (UPK3A), a small population of lung club cells (UPK3A), epididymis (UPK1B), cornea (UPK1B and UPK3B), testis (UPK1B), sperm (UPK1A, UPK1B, UPK2, UPK3A, UPK3B), and ovaries, including oocytes and eggs (UPK1A, UPK1B, UPK2, UPK3A, UPK3B) (6, 44, 270, 347, 389, 430, 608). Intriguingly, uroplakin knockout mice exhibit decreased in vitro fertilization rates and litter size, indicating that uroplakins may play an important role in egg fertilization (430). This latter finding supports early studies that implicated uroplakins in Xenopus laevis sperm-egg fusion (455a, 455b).

Keratins (also known as cytokeratins) are filament-forming cytoskeletal elements that are expressed in epithelial tissues (181). In humans, these include 28 type I and 26 type II genes (not including 13 pseudogenes). The type I and II keratins associate to form coiled-coil heterodimers and are expressed in a tissue- and differentiation-dependent manner. While some of these genes are hair specific, the cell-expressed keratins include the low-molecular-weight (40–56.6 kDa) acidic type I isoforms (KRT9–28) and the high-molecular-weight (53–68 kDa) basic-to-neutral type II isoforms (KRT1–8 and KRT71–80). The urothelium is reported to express numerous keratins, some in a differentiation specific manner, including KRT1, KRT4, KRT5, KRT6, KRT7, KRT8, KRT10, KRT13, KRT14, KRT17, KRT18, KRT19, and KRT20 (14, 201, 429, 593, 702, 730, 731) (TABLE 1). Keratin expression in the urothelium varies depending on its location. In mice, KRT10 is expressed only in the urothelium lining the urethra (429), while in humans KRT6 is only expressed in the urothelium lining the renal pelvis (14). Furthermore, keratin expression/distribution is not identical between species. For example, KRT5 is not expressed in the umbrella cells of rodents (239), but is avidly expressed in the umbrella cells of humans (14) (TABLE 1).

The keratins most often used to describe mouse urothelial differentiation are KRT5, KRT14, and KRT20, and to a lesser extent KRT7 (239, 283, 618, 639). In mouse, KRT20 is only expressed in umbrella cells, and not in the other two cell layers (FIGURE 4) (283). The only keratin we are aware of that is reported to be solely expressed by a population of mouse intermediate cells is KRT7 (434), although other studies report the KRT7 is expressed throughout the mouse urothelium (TABLE 1) (e.g., Ref. 649). The genes encoding Krt5/Krt14 are routinely expressed in stratified epithelia such as skin, and oftentimes in their basal cell layer (181). Not surprisingly then, KRT5 and KRT14 protein are expressed in the urothelial basal cell layer. In the bladder, KRT14 is only expressed in a small population of basal cells (558), while KRT5 is expressed in all of the cells located in the basal cell layer (including those that are KRT14 positive) (239, 243, 558, 639). Moreover, KRT5 is also expressed by almost all intermediate cells, except for those that express UPK3A (FIGURE 4) (239, 243). Somewhat confusingly, the KRT5-positive cells in the intermediate cell layers are often referred to as KRT5-positive basal cells (239). Because intermediate cells are positionally distinct from basal cells, and basal cells express markers and form protein complexes (i.e., hemidesmosomes) that are not normally found in any intermediate cell population (e.g., KRT14, KRT17, COL17A1, ITGB4) (14, 45, 390, 555, 558), we prefer the terms KRT5-positive basal cells or KRT5-positive intermediate cells.

Other urothelial differentiation markers are associated with signaling and transcription pathways, particularly those involved in mouse development. In this case, expression of three genetic markers have been revealing: Shh (sonic hedgehog), Tp63 (also known as tumor promoter 63 kDa, p63, or Trp63), and Foxa2 (Forkhead box protein A2) (239, 243, 639). SHH is a secreted protein that acts as a morphogen to promote patterning of a variety of tissues during development. It does so by activating the hedgehog signaling pathway: SHH binds to PTC1, liberating SMO to promote GLIA-dependent transcription (308). Originally identified as being expressed solely in basal cells (639), subsequent work has established that SHH is also expressed in all intermediate cell populations (FIGURE 4) (239). TP63 is a member of the p53 family of transcription factors (255). It is particularly important in tissues that result from interactions between stroma and epithelial cells. KO mice lacking Tp63 expression exhibit developmental defects including lack of limbs, teeth, and mammary glands (255). Much like SHH, TP63 is expressed in all cells in the basal and intermediate cell layers (239, 243). FOXA2 (Forkhead box protein A2) is developmentally expressed and has been used primarily in studies of early urothelial development (239, 243). It is a transcription factor that promotes tissue-specific gene expression, not only by binding DNA at specific sites, but also by interacting with linker H1 histones to disengage them from their interactions with enhancers and/or promoter sites (220, 329).

D. Development of the Urothelium

This section is a brief description of new insights into the development of the mouse bladder urothelium, many of which have been described by Dr. Cathy Mendelsohn and colleagues. For a more complete picture of the development of the urothelium (and other tissues) in the human and rodent urinary tract, the reader is directed to the exhaustive analysis performed by Georgas et al. (243), a recent review of ureter growth and differentiation (86), and a review of the development of the human bladder and ureterovesicular junction (431). As a note to the reader, in the following discussion any reference to uroplakins generally means UPK3A, unless otherwise indicated.

Gandhi et al. (239) describe that at mouse embryonic day (E)11–12 a primitive simple-to-stratified epithelium is formed that is largely comprised of a newly described population of progenitor (P-0) cells that express endodermal markers (SHH, TP63, and FOXA2) and uroplakins, but not KRT5 (FIGURE 5). These P-0 cells are not detectable in adult urothelium, but as a result of elegant lineage tracing studies they are known to give rise to the uroplakin-positive intermediate and superficial cell types during early bladder development. The other cell type present at these early stages, but in low number, is an undifferentiated endodermal population of cells (FOXA2⁺, TP63⁺, SHH⁺, uroplakin^–, KRT5^–). Its relationship to the P-0 cells is unclear (239). By E13–E14, the now multilayered urothelium is comprised of a superficial layer of P-0 cells, and subjacent layers filled with their progeny (FOXA2^–, TP63⁺, SHH⁺, uroplakin⁺, KRT5^–), which are the early forefathers of the uroplakin⁺, KRT5^– intermediate cells described in the preceding section (FIGURE 5) (239, 243). Note that at this time in urothelial development, there are no KRT5⁺ basal or intermediate cell types.

FIGURE 5.

Development of the mouse urothelium. The cell types and differentiation markers expressed at embryo stage (E)12, E13, E15–17 of the developing mouse embryo and adult urothelium. The phenotypes of the associated cell types are presented in the bottom half of the figure. BC, basal cell; IC, intermediate cells (binucleate and mononucleate); P-0, early urothelial progenitor cell; SC, superficial cell; UC, umbrella cell.

During the period of E14–E17, a number of events occur, culminating in the formation of a three-layered urothelium similar to that found in the adult mouse (239, 243, 639). During this time, the superficial P-0 cell layer differentiates, giving rise to mononucleate and binucleate intermediate cells (with the following phenotype: FOXA2^–, TP63⁺, SHH⁺, uroplakin⁺, KRT5^–), as well as uroplakin-positive superficial layer comprised of binucleate umbrella cells (FOXA2^–, TP63^–, SHH^–, uroplakin⁺, KRT5^–). During E14–E15, expression of KRT5 is finally observed, giving rise to FOXA2 ^–, SHH⁺, TP63⁺, uroplakin^–, and KRT5⁺ intermediate and basal cell populations. Although ultimately derived from cloacal endoderm, the progenitor that seeds these KRT5-expressing cells is unclear, but they are not derived from P-0 cells. Finally, at ~E16.5, KRT14-positive basal cells are observed, which have the following phenotype: FOXA2^–, TP63⁺, SHH⁺, uroplakin^–, KRT5⁺, KRT14⁺ (243, 558). In the adult urothelium, the latter cells are reported to be bladder stem cells and can give rise to all cell layers when the urothelium is exposed to multiple bouts of cyclophosphamide injury (558). However, as described below, the uroplakin⁺ intermediate cells can also give rise to umbrella cells in other contexts.

E. Regeneration of the Urothelium in Response to Injury: Defining the Urothelial Stem Cell Population

Under normal conditions of homeostasis, urothelial cell turnover is very low and labeling studies employing BrdU or KI67 reveal almost no label-retaining cells or cells undergoing mitosis (146, 383, 417, 463). Not surprisingly then, the lifespan of the rodent umbrella cell is estimated at ~200 days (342, 344). However, in response to acute injuries caused by exposure to chemicals (e.g., chitosan, cyclophosphamide, protamine sulfate, saccharin), mechanical overdistension, subtotal cystectomy, exposure to urine postnatally, or infection with uropathogenic bacteria, the urothelium undergoes rapid repair and regeneration (52, 146, 199, 200, 239, 401, 436, 515, 519, 520, 558, 571, 599, 600, 639, 729). In acute injuries, regeneration begins almost immediately, it is often associated with urothelial proliferation, and it terminates with a functional and normal appearing urothelium within several days to a few weeks. Regeneration requires input from both the stroma and from the urothelium and includes signaling pathways regulated by BMP4, non-canonical and canonical Wnt signaling pathways, Delta-Notch, ELF3, various growth factors, retinoids, Sonic Hedgehog (SHH and GLI1), and TP63 (52, 85, 165, 239, 436, 519, 520, 572, 639). What these pathways are doing to contribute to the regeneration, and their relationship to one another, is an open question.

A long-standing goal of the urology community is to engineer bladder replacements in response to pathologies such as bladder reconstruction after tumor resection (122). Thus an important question is what is the stem cell population(s) that regenerates the urothelium after injury? This decades long search has been hampered by our limited understanding of the distinct cell types that populate the urothelium, the diversity of models used in these studies (including cultured human epithelial cells, mice, rats), and the different types of injury employed. However, a better understanding of urothelial cell populations, the identification of markers that define these cell types, the introduction of genetically modified mouse models, and the focus on a more limited set of urothelial injury models is now moving the field forward in a rapid manner. For those readers seeking additional information about the urothelial cancer stem cells, a recent review by Wang et al. (744) provides an entry point into this topic.

First, we describe two commonly employed and physiologically relevant models of urothelial injury: infection with uropathogenic strains of E. coli, and treatment with cyclophosphamide (Cytoxan). E. coli is the most common cause of urinary tract infections in the United States (647). Uropathogenic strains of these bacteria express pili (fimbrae) on their surfaces, which they use to attach to N-linked carbohydrates covalently attached to UPK1A molecules expressed at the apical surface of umbrella cells (489). A fraction of these adherent bacteria is endocytosed (20, 82, 203, 204). While the majority of these are actively expelled by exocytosis (82), those that remain form intracellular bacterial communities during the subsequent 6–12 h (20). Coincident with formation of these colonies, the bacteria trigger the exfoliation of umbrella cells through a pathway that in a urothelial cell line requires UPK3A and casein kinase 2-dependent phosphorylation (697). This spurs the subsequent proliferation of the remaining urothelial cells and, eventually, regeneration of the urothelium.

On the other hand, cyclophosphamide is a chemotherapeutic agent that is employed to treat a wide variety of cancers (lymphoma, multiple myeloma, ovarian cancer, breast cancer, lung cancer, and neuroblastoma) or to suppress the immune system in diseases such as nephrotic syndrome (7). Of relevance to the bladder, cyclophosphamide is converted in the urine to acrolein, a toxic compound that causes marked irritation and chemical burns (99, 150). Not surprisingly then, complications of cyclophosphamide treatment include hemorrhagic cystitis, which results from urothelial and vascular damage (472). In rats, a single dose of cyclophosphamide also results in hemorrhagic cystitis within 1–2 days, which then resolves over the next several days as the urothelium regenerates. Some studies employ a “chronic” cyclophosphamide treatment (multiple treatments over a 1- to 2-wk period), which is also accompanied by bladder inflammation, urothelial hyperplasia, symptoms of overactive bladder, and pelvic pain (93, 558, 748).

In other stratified epithelia such as the epidermis, the basal cell layer serves as a stem cell population to seed epidermal growth and regeneration. Thus it seems reasonable to assume that the urothelial progenitor is also likely to be found in the basal cell layer. Here, there are several relevant observations. In response to infections by E. coli, Mysorekar et al. (519) were first to observe, followed by others (146, 639), that KRT5⁺ and SHH⁺ basal cells (and possibly intermediate cells with the same phenotype) proliferate, giving rise to the other cell types in the urothelium. More recently, an important study by Papafotiou et al. (558) revealed that it is likely the KRT14⁺ subpopulation of basal cells (which as noted above are also KRT5⁺ and SHH⁺) that may act as urothelial progenitor cells (FIGURE 4). In these studies, a single bout of cyclophosphamide-induced injury stimulates a local proliferation of KRT14⁺ basal cells; however, after five consecutive treatments with cyclophosphamide, KRT14⁺ cells are found in all three urothelial cell layers (558).

What has baffled the field are several reports that cell types other than basal cells are also functional in urothelial regeneration after injury (771). For example, Colopy et al. (146) reported that KRT5^–, uroplakin⁺ intermediate cells undergo proliferation 1 day post exposure to E. coli. Other studies demonstrate that treatment with protamine sulfate or chitosan causes the selective loss of the superficial umbrella cell layer. Within minutes, this loss spurs the differentiation of newly exposed cells in the underlying intermediate layer, stimulating them to express (or redistribute) proteins normally associated with umbrella cells, including TJP1 and uroplakins (200, 401, 729). Gandhi et al. (239) used lineage tracing to establish that it is the KRT5^–, uroplakin⁺ intermediate cells that serve as progenitors for superficial umbrella cells in response to an acute cyclophosphamide-induced injury. They further noted that KRT5⁺ cells could also be triggered to undergo proliferation in response to cyclophosphamide, but they did not differentiate into umbrella cells. Most recently, Wang et al. (748) published that in response to either E. coli or acute cyclophosphamide injury, the generation of 4n + 4n umbrella cells depends on the initial formation of binucleate intermediate cells (2n+ 2n; TP63 ⁺, SHH⁺, uroplakin⁺, KRT5^–; FIGURE 4). As described above, these cells arise from incomplete cytokinesis of mononucleate intermediate cells with the same phenotype. Together, these studies indicate that mononuclear and binucleate KRT5^– and uroplakin⁺ intermediate cells can also regenerate the superficial umbrella cell layer in response to an acute injury.

The crucial question then becomes whether the urothelial lineage progresses in a linear fashion from basal cells to intermediate cells to umbrella cells, or if it is nonlinear, with populations of both basal and intermediate cells serving as urothelial progenitors? The emerging answer is that it may depend on the extent of urothelial and subjacent tissue damage (FIGURE 6). As already noted above, a single dose of cyclophosphamide leads to an acute injury, which results in the local proliferation of KRT5⁺ and KRT14⁺ basal cells and regeneration of the umbrella cell layer by way of mono-/binucleate uroplakin⁺ and KRT5^– (and KRT14^–) intermediate cells (FIGURE 6, left) (239). In contrast, multiple doses of cyclophosphamide leads to the appearance of KRT5⁺ and KRT14⁺ cells throughout all layers of the urothelium (FIGURE 6, right) (558). Tellingly, Wang et al. (748) reported that serial cyclophosphamide treatment depletes the uroplakin⁺ and KRT5^– intermediate cell pool, ultimately giving rise to mono- and binucleate intermediate cell populations that are KRT5⁺. Thus, in response to a chronic injury, a linear urothelial lineage is revealed with KRT5⁺ cells giving rise to mono-/binucleate intermediate cells, which in turn give rise to umbrella cells. While this study did not determine if these KRT5⁺ cells were also KRT14⁺, based on work by Papafotiou et al. (558), this seems likely. It is also unclear if the pool of uroplakin⁺ and KRT5^– intermediate cells re-emerges once the injury is stopped.

FIGURE 6.

Regeneration of the mouse urothelium in response to injury. Left: urothelial response to acute injury. The loss of binucleate (4n+4n) umbrella cells, and/or subjacent cell layers, results in proliferation of the urothelium and regeneration of the superficial umbrella cell layer. This process depends on uroplakin⁺ intermediate cells, which undergo incomplete cytokinesis to generate binucleate (2n+2n) intermediate cells (bIC). In turn, the bIC give rise to superficial umbrella cells, which increase their ploidy by way of endoreplication. Right: urothelial response to serial and/or major injury. In the absence of bIC and their progenitors, KRT5⁺ basal cell, and possibly KRT5⁺, KRT14⁺-derived basal cells, give rise to all of the cell types, including the bICs. The latter give rise to superficial cells, which further differentiate into umbrella cells. BC, basal cell; IC, intermediate cells (binucleate and mononucleate); SC, superficial cell; UC, umbrella cell.

Finally, Schäfer et al. (622) observe that focal mucosal resection (2/3 of the luminal surface) of the bladder leads to proliferation of KRT5⁺ basal cell populations as well as reseeding of the umbrella cell by uroplakin⁺ intermediate cells. In contrast, augmentation cystoplasty, a surgery that involves wholesale resection of the bladder and replacement with acellular grafts (derived from small intestine submucosa), leads to the de novo regeneration of all urothelial lineages by the progeny of KRT5⁺ basal cells (622). Again, these studies indicate that the type and extent of injury likely defines which population of urothelial progenitor cells are responsible for regenerating the urothelium. Furthermore, they indicate that KRT5⁺ basal cells (and possibly the KRT14⁺ subpopulation of basal cells) can serve as stem cells to repopulate the entire urothelium in response to extensive urothelial injury.

F. Pressing Questions in Urothelial Regeneration

With the new insights described above, the field should now be able to address a number of pressing questions.

• 1. Does the involvement of intermediate cell progenitors versus basal cell ones vary by injury? Here, we need studies of other physiologically relevant animal models, and human tissue samples, to assess how different injuries affect urothelial regeneration.
• 2. How is the loss of umbrella cells communicated to basal stem cells? In other tissue systems, terminally differentiated cells as well as so-called transit amplifying cells (partially differentiated cells that undergo rapid proliferation in response to injury or tissue loss) can act as regulatory components of the stem-cell niche, signaling when stem cells should proliferate (306, 307, 792). At present, we do not know if umbrella cells release inhibitory factors that prevent stem cells from proliferating, or if damaged or infected umbrella cells release mediators that promote stem cell proliferation. In the case of E. coli infections of umbrella cells, there is increased mitoses in the basal cell layer and signaling events emanating from the stroma (639). But what triggers these responses is unknown.
• 3. Are KRT14⁺ basal cells the stem cell population in the adult urothelium? It is important that investigators include an analysis of KRT14⁺ cells in their regeneration studies. This will allow the field to better assess whether these cells serve as the sole stem cell population or whether other KRT5⁺ cell populations are involved.
• 4. Are the signaling pathways driving urothelial regeneration similar for every injury? Both E. coli and serial treatments with cyclophosphamide induce a Wnt/β-catenin-dependent proliferation of basal cells (558, 639). Likewise, both cyclophosphamide- and surgery-induced injuries depend on retinoic acid signaling pathways (239, 622). Thus similar signaling cascades can be initiated in response to very different models of injury. If this will be true of other injuries must await further studies.
• 5. An additional unanswered question is how do long-term, chronic injuries affect urothelial differentiation and regeneration? For example, patients with the classical form of interstitial cystitis (those with Hunner’s lesion) as well as cats with feline interstitial cystitis suffer from urothelial defects (208, 339, 403), as do animals with experimental outlet obstruction, or experimental spinal cord injury (SCI) (36, 119, 438, 602). In most cases, there has not been a careful analysis of how these pathologies affect urothelial differentiation. In the case of spinal cord injury, morphological changes include hypertrophy of the bladder, an apparent increase in intermediate cell layers, and a focal loss of umbrella cells within a few hours of insult (36, 383, 488, 642, 724). This loss of umbrella cells is accompanied by the rapid expression of uroplakins and tight junction-associated proteins in newly exposed cells in the underlying intermediate cell layer (36). By 2 wk, the urothelium and its associated tight junction barrier are reformed, and by 4 wk, the surface cells are small (10–20 µm in diameter). It was unknown if this difference in cell size reflects changes in urothelial differentiation as a result of spinal cord injury.
• 6. Kullmann et al. (383) have now examined what happens to urothelial differentiation in response to spinal cord injury. Within a few days, the umbrella cells are replaced by small superficial cells that express uroplakins along with differentiation markers normally associated with cells found in the basal and intermediate cell layers (KRT5, KRT14, and TP63); however, these cells lack expression of KRT20 (383). Spinal cord injury also stimulates proliferation of KRT14^– and KRT14⁺ cells, and even 28-days post injury, small superficial cells persist that coexpress KRT5, KRT14, TP63, and uroplakins, but not KRT20 (383). Thus, unlike the complete regeneration of the urothelium that occurs in response to acute injuries, spinal cord injury results in patches of poorly differentiated urothelium weeks after the initial insult. While the ramifications of this altered differentiation are unknown, it may also be a characteristic of the human condition (51, 152, 723, 724).
• 7. By better understanding the urothelial cell types affected by chronic pathologies, and the signaling cascades that are driving these events, it may eventually be possible to intercede pharmacologically, stimulating the production of a highly differentiated urothelium with its barrier function renewed. This is particularly important, as defects in the urothelial barrier can lead to central nervous system sensitization and bladder-derived pelvic pain (505), which may be avoided if interventions can be made early in the disease process.

III. THE UROTHELIAL BARRIER

In this section, we discuss one of the major functions of the urothelium: to form a barrier to pathogens and to the excess water, metabolic waste products, and other solutes present in urine, a liquid with an osmolality of 300–900 mosmol/kgH₂O. Because urine contacts the renal pelvis, ureters, and urethra, and is stored in the bladder for extended periods of time, the urothelium must prevent the diffusion of urinary components into the underlying tissues. Consequently, the urothelium is exposed to large osmotic and chemical gradients as well as cyclical mechanical stretch as urine is produced, transported, stored, and eventually eliminated from the body. We begin by describing the specialized properties of the umbrella cell apical membrane. We then discuss the water and solute transport pathways, which allow the urothelium to adapt to and possibly even modulate the composition of the urine bathing its mucosal surface. In the last section, we explore some features of the high-resistance tight junctions that prevent unregulated paracellular flux across the urothelium. Again, there is limited information available about the permeability properties of the urothelium lining the renal pelvis, ureters, and urethra. However, there is a report that the chemotherapeutic drug mitomycin C is more permeable across the urothelium lining the ureters versus that lining the bladder (757). The underlying basis of this difference is unknown, but as noted above, the urothelium lining the ureters and bladder exhibit differences in protein expression and production of DFVs.

A. The Glycocalyx at the Apical Surface of Umbrella Cells

Although the urothelium is multilayered, it is the outermost umbrella cell layer, and in particular its mucosal surface, that forms the barrier to pathogens as well as urine and its associated metabolites, solutes, and water. This barrier is multifactorial and includes the apical membrane, the umbrella cell tight junction, and the glycocalyx at the apical membrane of the umbrella cell layer (123, 358).

The glycocalyx is a 50- to 300-nm electron-dense “fuzzy” coat that extends from the apical surfaces of epithelial cells (554). It is prominent in enterocytes lining the gut and in endothelial cells. However, in the urothelium, only a thin, somewhat disorganized glycocalyx is associated with the umbrella cell apical membrane when visualized using TEM (FIGURE 2C). Its presence is more readily detectable using the following stains: periodic acid-Schiff, alcian blue, colloidal iron, or ruthenium red coupled with TEM (180, 295, 418, 501, 705). In general, the glycocalyx comprises glycoconjugates associated with membrane-bound glycoproteins and glycolipids, as well as soluble components including galectins and proteoglycans. The latter are comprised of a core protein to which glycosaminoglycans (GAGs)–unbranched carbohydrates composed of repeating disaccharides–are attached. Well-known GAGs include heparin sulfate, chondroitin sulfate, keratan sulfate, and dermatan sulfate.

This so-called “GAG layer” has previously received significant attention in the urology field. Dr. Lowell Parsons (562) postulated that the bladder GAG layer formed the “blood-urine barrier,” precluding the adherence of bacteria and leakage of urinary solutes and waste materials from entering the bladder interstitium. The permeation aspects of this hypothesis were largely based on studies where protamine sulfate was instilled in the bladders of patients or added to the mucosal surface of rabbit urothelium, which stimulated increased permeability of Ca²⁺, urea, and water (432, 563). Protamine sulfate, a cationic peptide originally purified from salmon sperm, is routinely used to treat over heparinization. Addition of heparin to the protamine sulfate-treated bladder or tissue “cured” the permeability defect. As a result of these studies, a synthetic heparin analog, pentosanpolysulfate, was developed as a treatment for bladder disorders such as interstitial cystitis/bladder pain syndrome (IC/BPS). Unfortunately, this treatment has significant side effects, and its efficacy is mixed (reviewed in Ref. 254). For example, earlier studies reported some benefits associated with pentosanpolysulfate treatment (reviewed in Ref. 254), but the most recent multi-center, double-blind study found no significant effect of using this drug (254, 537).

Other studies have called the central aspect of the “GAG layer hypothesis” in question. For example, protamine sulfate treatment results in a rapid denudation of the umbrella cell layer (401). Moreover, work from Dr. Simon Lewis and colleagues (712, 713) demonstrated that protamine sulfate inserts into in the umbrella cell apical membrane, and upon doing so forms reversible pores that conduct ions. If true, then the addition of heparin to the protamine sulfate-treated urothelium is not restoring the GAG barrier, but instead is more likely functioning by removing protamine sulfate from the membrane. More to the point, Dr. Veer Bhavanandan and colleagues (102, 103, 297) used a large panel of lectins as well as [³H]GlcNH₂ and [³⁵S]sulfate labeling (coupled with well-established fractionation methodologies) to identify the transmembrane mucin MUC1 as the major contributor to the rabbit umbrella cell glycocalyx. Significantly, GAGs only contribute a small fraction to the urothelial glycoconjugates. If this is true of human urothelium is not known. Moreover, biochemical studies have established that expression of MUC1 does not alter the permeability of epithelial cells to water or urea, but slightly increases the permeability to ions (404).

Because a GAG layer does not form a barrier to water flow or ion permeability, we think it is unlikely to be the blood-urine barrier. However, one aspect of the Parson’s hypothesis that has stood the test of time is the probable role that the glycocalyx plays in innate immunity. Here, there is more compelling evidence that the glycocalyx may have intrinsic, nonspecific, anti-adherence properties that protect against bacterial infection (50, 158, 564, 609). Finally, there is a growing appreciation that the glycocalyx of other tissues, including that associated with the endothelium, is a critical component of tissue function, including responses to mechanical forces such as shear stress (450). Thus further study of the urothelial glycocalyx is warranted.

B. Biophysical Properties of the Umbrella Cell Apical Membrane

The membrane properties of the umbrella cell apical membrane, in particular its AUM, have fascinated researchers for decades. However, less is known about the lipid bilayer itself, and its specific properties. As noted above, the umbrella cell apical membrane is insoluble in a subset of detergents, including Sarkosyl (428). This is a property of so-called “lipid rafts,” which in other cells are thought to be microdomains comprised of nanometer-sized clusters of cholesterol and sphingolipids that form a liquid-ordered phase floating in a sea of phospholipids (634). Lipid rafts are often described as being signaling platforms as signaling molecules can be redirected to rafts when activated. In the case of the umbrella cell, it appears that the entire apical membrane is detergent insoluble and is thus more like a large ship than a raft (428). As expected for a raftlike membrane, the lipid constituents of the umbrella cell apical membrane are highly enriched in cholesterol and cerebrosides (113, 357, 667), a subclass of sphingolipids with two acyl chains and a simple sugar residue (glucose or galactose) as a head group. Cerebrosides are enriched in the membranes of neuronal cells, myelin, and skeletal muscle cells. Other major lipid components of the apical membrane include phosphatidylcholine and phosphatidylserine (113, 357, 667). The fatty acid composition of urothelial membranes has also been published (113).

Looking at the apical membrane, studded with plaques and divided by actin-rich microplicae, it may be tempting to assume that the apical membrane is very rigid. To measure the deformability of the umbrella cell apical membrane, Mathai et al. (468) used optical magnetic twisting cytometry, a technique in which ferromagnetic beads are attached to the cell surface and then tracked as they are pulled by a magnetic field. Surprisingly, the umbrella cell apical membrane is highly deformable (shear modulus G ~30 Pa), even more so than the plasma membrane of the red blood cell (G ~90 Pa). Similar, but somewhat higher values were obtained using atomic force microscopy, which has a tip that indents the apical membrane more deeply and is thus more sensitive to cytoskeletal effects. When umbrella cells are fixed with paraformaldehyde, the elastic modulus, expectedly, becomes very large (G ~5,000 Pa). Interestingly, if the umbrella cell layer is removed by protamine sulfate treatment, the newly exposed intermediate cells have a shear modulus that is more than an order of magnitude higher (G ~400 Pa) than umbrella cells (468). This indicates that hypercompliant plasma membrane domains are a feature of only fully differentiated umbrella cells. Mathai et al. (468) also reported that Upk2 KO mice, which are unable to form plaques, exhibit a shear modulus similar to that of the red blood cell (~90 G). While this was interpreted to mean that uroplakins govern the deformability of the apical membrane, the urothelium of Upk2 KO mice is hyperplastic, the umbrella cells of Upk2 KO mice are poorly differentiated, and they exhibit increased amounts of actin under their apical surfaces (372). Thus any effect on deformability may be indirect.

C. Transport of Ions, Water, and Pharmacological Agents Across the Urothelial Barrier

The apical membrane of umbrella cells coupled with the tight junction, which seals the space between neighboring umbrella cells, form a barrier that prevents the unrestricted diffusion of solutes and water across the epithelium (34). Electrophysiological studies have shown that urinary bladders from rabbits exhibit an exceptionally low permeability to ions with a transepithelial resistance exceeding 20,000 Ω·cm² in the quiescent state, and approaching ~75,000 Ω·cm² when transcellular Na⁺ transport is blocked by the inhibitor amiloride (114, 423). Lipid bilayers have inherently low permeability to ions (the resistance of a pure phospholipid bilayer = 10¹⁵ Ω·cm²) (298). Thus the diffusion of ions across biological membranes relies on the presence of ion channels, which are able to move millions of ions per second without compromising selectivity (298).

1. Ion transport in the umbrella cell layer

Umbrella cells express the ENaC (423, 424, 651), a highly selective (permeability of Na⁺/K⁺ >50) and amiloride-sensitive Na⁺ channel, that is the rate-limiting step for Na⁺ transport across the urothelium (424). By analogy to the distal nephron, Na⁺ transport in the urinary bladder is regulated by the mineral corticoid hormone aldosterone (109, 193, 423), which is released from the adrenal gland in response to Na⁺ depletion. The function of ENaC in the urothelium is an open question. However, recent studies show that the mammalian urinary bladder can actively reabsorb salt in its distended state, and hence it is able to alter the composition of urine (509). Other investigators have proposed that ENaC is important in triggering release of ATP from the urothelium (214). Du et al. (184) explored the effects of the ENaC inhibitor amiloride (IC₅₀ of ~0.1 µM for ENaC) on bladder function, but the high concentration of amiloride employed (1 mM) makes it likely that other channels were affected. ENaC is also reported to regulate stretch-induced changes in apical surface area (782).

There is also evidence that a nonselective, stretch-activated cation channel, inhibited by Gd³⁺, is expressed apically, and possibly a chloride conductance as well (745). The identity of these channels is unknown, but some TRP family channels are mechanosensitive, function as nonselective cation channels, and are apparently expressed apically (780). Other studies indicate that the urothelium expresses several K⁺ channels including K_ATP and the pore-forming α-subunit of the large-conductance K⁺ channel KCNMA1 (443, 745, 782).

2. Transport of water and urea across the urothelium

Similar to other water-tight membranes, the apical membrane of urothelial cells displays low permeability to water, urea, and NH₃ (123). Interestingly, Hu et al. (311) showed that relative to control mice, the urothelium of Upk3a KO mice is more permeable to water, and to a lesser extent urea as well. This led them to propose that uroplakins regulate or contribute to the permeability of water and urea across biological membranes. However, like the umbrella cells of Upk2 KO mice, the umbrella cells of Upk3a KO mice also exhibit abnormalities in their differentiation and show increased permeability to methylene blue (310). Without information about the lipid composition, parameters such as membrane fluidity and deformability, or the expression of proteins such as aquaporins, it is difficult to parse why the loss of uroplakins leads to apical membrane permeability changes in these KO mice.

Considering its role as a barrier, it is somewhat surprising that the urothelium also expresses several aquaporins, a family of ~10 channels that transport water or small uncharged solutes (NH₃, CO₂, glycerol, urea) across cellular membranes (2, 375). Early studies reported expression of aquaporin (AQP)2 and AQP3 in the rat urothelium, and AQP1 in the endothelial cells lining the capillary and arterioles of the bladder wall (657). Both AQP2 and AQP3 are localized primarily to the basolateral membrane of umbrella cells, and the plasma membranes of intermediate and basal cells (657). In human surgical specimens, transcripts for AQP3, AQP4, AQP7, AQP9 and AQP11, but not of AQP1 or AQP2, are detected (606). Similarly, AQP3, AQP9, and AQP11 are expressed in adult pig bladders (459). Interestingly, AQP2 and AQP3 protein expression is significantly increased by 24–49 and 160–200%, respectively, in dehydrated rats (657). This may indicate that the urothelium plays an active role in water resorption under these conditions. Moreover, Morizawa et al. (509) used shRNA technology to show that a fraction of instilled fluid is reabsorbed in very distended rat bladders, and that this occurs in an AQP2-dependent manner. They also showed that AQP2 expression, but not that of AQP3 or AQP1, increases during distension, further supporting the possibility that the urothelium has active mechanisms to sense distension and increase water transport in response. However, the reported lack of AQP2 expression at the apical surface of the umbrella cells makes it unclear how the water is being transported across this nominally impermeable membrane.

Interestingly, the reabsorption of water in the Morizawa et al. (509) study was independent of vasopressin (a hormone that promotes water retention). This observation is surprising given recent studies that reveal that the rat urinary bladder mucosa expresses the vasopressin receptors AVPR1A and AVPR2 (80, 111), and that AVPR2 expression in the bladder mucosa increases with age (80). If both observations are true, then some other mechanism(s) must regulate AQP2 expression and water resorption in the urothelium. Or, it is possible that vasopressin sensitivity increases with age. In summary, these studies indicate that the urinary bladder can alter the composition of urine during the storage phase of the micturition cycle. However, unlike water reabsorption in the distal nephron of the kidneys, the resorption of water in the bladder may not be subject to hormonal regulation by vasopressin. Finally, we note that there is also evidence that urea transporters are also expressed in the urothelium including that lining the bladder and ureters (446, 656, 658, 659, 742).

3. Transport of pharmacological agents across the urothelium

While little is known about the mechanisms that facilitate the transport of organic molecules across the urothelium, major efforts are underway to treat superficial bladder cancer and other bladder conditions (e.g., interstitial cystitis and neurogenic detrusor overactivity) by the intravesicular route. The systemic administration of therapeutic agents to treat bladder conditions is often ineffective because only a small fraction of the drug reaches the affected site (271). Thus the intravesicular route may provide the following advantages for the delivery of therapeutic agents: 1) increased efficiency in delivering therapeutic agents to the site of action (i.e., urothelial layer), 2) reduced side effects as consequence of the restricted area of contact and limited diffusion in the systemic circulation, and 3) increased bioavailability of molecules with low inherent solubility in aqueous solutions. However, even direct delivery of pharmaceutical agents has issues.

A critical factor that governs the ability of organic molecules to cross biological membranes by passive diffusion is their respective lipophilicity (11). Somewhat surprisingly, when crossing bladder tissue, penetration of hydrophilic molecules (e.g., methylprednisolone hemisuccinate and mitomycin C) is favored over more lipophilic ones (lidocaine hydrochloride) (497). Several physical and chemical methods have been developed to enhance the transport of organic compounds across the urothelium including electromotive drug administration (reviewed in Ref. 247), use of chemical penetration enhancers (e.g., DMSO, chitosan, polycarbophil, protamine) (108, 202, 247, 355), liposome formulations (reviewed in Refs. 331, 550, 710), and protein nanoparticles (445). Further studies are needed to gain a better understanding of the basic mechanisms that mediate the transport of organic molecules across the urothelium and to develop more effective systems for the intravesicular delivery of therapeutic agents.

D. Fluid and Solute Reabsorption from the Urine: A New Function for the Urothelium?

The urothelium has pathways for ion reabsorption, urea transport, and aquaporin-dependent water transport, but the function of these pathways has been unclear. Now, there is a growing awareness that the urothelium can change the fluid and solute composition of the urine. As early as 1959, Levinsky and Berliner (419) observed that when perfusing dog ureters and bladder in situ, movement of water, urea, Na⁺, K⁺, Cl^–, creatinine, and H⁺ occurred, especially at low flow rates. Subsequently, Cahill et al. (112) and Shafik et al. (636) reported differences in the urine composition between the pelvis and bladder. Together, these studies indicate that the urothelium can alter the composition of urine. More recently, Watanabe and Azuma (752) used transabdominal real-time 3D ultrasound (or automatic capacity recordings) to measure bladder volume during sleep. These investigators observed two types of patterns: some patients exhibited a “gradual elevation” type, where the overnight production of urine is less than the functional capacity; however, other patients exhibited a “steep-flat” pattern, where the bladder reaches its functional capacity, and then the urine volume is apparently reduced, in a pattern that is repeated until the patient awakes. While resorption of fluid was not directly measured in this study, if the findings hold true, then reducing excess urine volume may be an important function of the urothelium during sleep. These results fit with observations made some 20 yr ago by Sugaya et al. (671) that saline solution is recovered from the bladder when rat bladders are filled beyond their normal functional capacity and the more recent report by Morizawa et al. (509), described above, that fluid is reabsorbed when rat bladders are overdistended. Finally, Spector et al. (655, 656) describe how hydration status affects urea, Na⁺, K⁺, and Cl^– transport across the urothelium. Thus, in addition to forming a barrier, it seems very likely that the urothelium can alter the composition of the urine in a manner that depends on hydration status, the state of bladder distension, and if the subject is asleep or awake. The latter could have important implications for understanding nocturia (awakening to urinate during the night), nocturnal enuresis (bed wetting), and their treatments. Additional insights into these two conditions and their relationship to the circadian clock are described in section V.

E. The Umbrella Cell Apical Junctional Complex

The adhesion of adjacent epithelial cells to one another and their ability to regulate paracellular transport depend on the formation of a specialized epithelial structure called the apical junctional complex (210, 219, 240). This structure has three components: 1) the apical-most tight junction or zonula occludens, which is formed by transmembrane claudins, and other membrane and cytoplasmic proteins that are linked to the actin cytoskeleton; 2) the subjacent adherens junctions or zonula adherens, which is formed by cadherins and associated catenins that tether the adherens junction to the actin cytoskeleton; and 3) the lateral-most desmosomes or macula adherentes that are comprised of cadherin-like molecules (desmogleins and democollins) and their associated cytoplasmic partners that connect the desmosomes to the intermediate filament cytoskeleton (FIGURE 7, A and B) (219). Additional information about the organization of the umbrella cell junctional complex and its response to filling and voiding are described below. As there has been little research on urothelial adherens junctions and desmosomes, the subsequent discussion is focused on the tight junction. However, those readers interested in the adherens junction and desmosomes are directed to the following review articles (240, 573, 607).

FIGURE 7.

Components of the umbrella cell junctional complex. A: the junctional complex is comprised of the belt-like tight junction, the belt-like adherens junction, and desmosomes, which form spot-like junctions. [Redrawn from Apodaca and Gallo (34).] B: transmission electron micrograph of mouse umbrella cell junctional complex (JC). AJ, adherens junction; De, desomosome; TJ, tight junction. Inset: higher magnification view of umbrella cell tight junction, showing “kissing points.” C: freeze fracture replica of apical tight junction-associated strands from frog bladder epithelium. [From Claude and Goodenough (140), with permission from Rockefeller Press.] D: three-dimensional reconstruction of rat umbrella cell layer immunolabeled with antibodies to CLDN8 and co-stained with TRITC-phalloidin (actin). The 3D image was tilted using Volocity software. One scale bar unit = 12.3 µm. [From Acharya et al. (4).] E: model of claudin architecture. The extracellular segments (ECS-1/ECS-2) are folded into five β-strands, which are numbered. The transmembrane domains are labeled I–IV. The red bar between β-strands three and four is a disulfide bridge. The position of the conserved claudin W-G-L-W-R signature motif is marked. ECH, short extracellular helix, which promotes cis-interactions between claudins on the same membrane domain. F: structure of the CLDN15 protomer. The ECH α-helix is colored cyan, and the β-strands formed by ECS1/2 are colored navy blue (β-strands are numbered). The α-helical transmembrane domains are colored white, and the approximate position of the lipid bilayer is indicated by the transparent blue box. G and H: two stacked rows of claudin polymers arranged in an anti-parallel fashion and viewed en face (G) or in cross section (H). When ECS1/ECS2 from four claudin protomers (colored in red, green, yellow, and blue) interact on adjacent membranes, they form pore-like, β-barrel channels. The location of the paracellular pores is indicated by red asterisks in G, and the red arrow in H indicates the pathway for ion flow through the pore formed by stacked, anti-parallel claudin protomers. F–H were generated using the Swiss PDB viewer software.

1. The tight junction

The tight junction forms a continuous ring that completely circumscribes the apex of adjacent umbrella cells (FIGURE 7, C and D). When viewed face-on by freeze-fracture microscopy, the tight junction comprises a network of anastomosing “junctional strands” (FIGURE 7C) (228). However, when the junctional complex is viewed by conventional TEM, and if the section is cut exactly perpendicular to the plane of the lateral membrane, the strands appear as “kissing points” (FIGURE 7B, inset) (228). Tight junctions have two basic functions. First, they act as a “fence” to separate the apical and basolateral membrane domains. In doing so, they restrict the lateral diffusion of lipids and membrane proteins in the exofacial leaflet of the plasma membrane. This segregation of plasma membrane domains is essential to maintain the vectorial transport of solutes (421, 423, 424). The second function of the tight junction is to regulate the paracellular diffusion of ions and other molecules through the lateral interstitial space (which is known as “gate function”).

Early on, Claude (139) proposed that the resistance of the paracellular pathway is related to the number of strands and the presence of paracellular pores that were proposed to transition stochastically between “open” and “closed” conformations. In several tissues and model epithelia, a correlation is found between the number of junctional strands and transepithelial resistance (TER) (191, 256, 451, 452, 460, 624); however, this association is absent in other cell types (465, 500). For example, freeze-fracture analysis shows no difference in number of junctional strands among two clones of Madin-Darby canine kidney (MDCK) cells (I and II) whose transepithelial resistance differ >30-fold (665). Furthermore, a large increase in transepithelial resistance (~300%), without a change in the number or distribution of junctional strands, is observed when MDCK monolayers are switched from 37 to 4°C (257). Together, these studies indicate that TER is not completely defined by strand number per se, but more likely, it is the permeability properties of the pores within the junctional strand. This notion was reinforced with the discovery of the claudins, the structural and functional units of the junctional strands (229, 231, 706).

2. Claudin structure and oligomeric assembly

There are numerous transmembrane and cytoplasmic proteins associated with the tight junctions (240); however, in this review our focus is on the claudins, a group of 27 transmembrane proteins that share a similar topology and also form the functional backbone of the tight junctions (FIGURE 7E) (27, 376). The claudin family can be divided in two groups based on sequence homology: classical ones include CLDN-1, -2, -3, -4, -5, -6, -7, -8, -9, -14, -17, -19A, -19B, and -20; and nonclassical ones include CLDN-10a, -10b, -11, -12, -15, -16, -18A, -18B, -21, -22, -23, and -24 (394). Note that CLDN13 is present in mice, but not in humans or chimpanzees, and in rats there is a single CLDN10 isoform that is a hybrid of CLDN10A and CLDN10B. In terms of topology, claudins have four transmembrane domains, with their NH₂- and COOH-terminal regions extending into the cytoplasm (FIGURE 7, E and F) (228). Their ability to form strandlike structures when expressed exogenously in nonepithelial tissues indicates that claudins are also the chief structural component of the junctional strands (229, 231, 706).

The crystal structure of four members of this family, CLDN3, CLDN4, CLDN15, and CLDN19, are now resolved (524, 611, 640, 678). The overall structure of the individual protomer resembles the shape of a human left hand with the four transmembrane (TM) segments corresponding to the forearm. Extracellular segment 1 (ECS1), comprised of four β-strands (β1–4), corresponds to the four fingers, and the α3 helix and associated β-strand (β5) of ECS2 form the thumb (FIGURE 7, E and F). As noted above, the NH2 and COOH termini are intracellular. Claudins are linked through a COOH-terminal PDZ domain-binding motif to the tight junction associated scaffolding proteins TJP1, TJP2, and TJP3 (ZO-1 to ZO-3) (25). In turn, these proteins link the tight junction to the actin cytoskeleton. Tight junction formation is dependent on the presence of TJP1 and TJP2 (715).

Claudins form tight junction strands by two types of interactions: cis-interactions link adjacent claudins on the same membrane, whereas trans-interactions take place when claudins on opposing cell membranes bind to one another. Each claudin isoform forms a subset of cis- and trans-interactions. For example, CLDN1 interacts with itself in cis, and with CLDN3 in trans, but not with CLDN2 (231). Based on the resolved crystal structure of CLDN15, Suzuki et al. (679) proposed that tight junction strands are formed by a two stacked rows of claudin protomers arranged in an anti-parallel fashion, while the paracellular pores are formed by the association of ECS1 and ECS2 on adjacent membranes creating pore-like β-barrel channels (FIGURE 7, G and H). Further refinements of this model using molecular dynamic simulations demonstrate that single and double pores can form, that pore size can fluctuate in short time scales, and that the narrowest portion of the pore is in the region of 2.5–3.0 Å (12). The radii of most physiological ions (Na⁺, K⁺, Cl^–) in aqueous solutions are in the range of 2.0–3.0 Å (461).

The permeability across the tight junction varies in a cell-type specific manner. For example, in the proximal tubule of the kidney, the TER is ~200 Ω·cm², while the TER of the urothelium approaches 75,000 Ω·cm² in the presence of amiloride (423). Moreover, the tight junction can discriminate to some extent on the basis of electrical charge and molecular size. It turns out that these functions depend on the combination of claudin isoforms that comprise the tight junction. Based on their effects when overexpressed in epithelial monolayers, claudins are classified as “pore forming” when they make tight monolayers leakier, or “barrier forming” when they make leaky monolayers tighter (21). For instance, CLDN2 is a pore-forming, cation-selective claudin, as its expression causes a very large decrease (~10-fold) in the TER of epithelial monolayers and a concordant rise in paracellular cation transport (778). In contrast, CLDN8 raises the TER of low-resistance cell lines indicating that it is barrier forming (28, 779). It should be noted that many members of the claudin family do not have an ascribed function yet. Functional studies indicate that ECS1 faces the conductive pathway and that residues in this region define the conductance and selectivity of the paracellular pores (29, 143, 144, 778). Point mutations that reverse the charge on specific residues within this loop, and chimeras that exchange the first extracellular loop between different claudin isoforms, alter the charge selectivity of the paracellular pore. ECS2 is believed to regulate the association of claudins expressed on adjacent cell membranes (trans-interactions) (163, 326, 377, 678).

3. Tight junction organization and claudin distribution in the urothelium

The known urothelial expressed claudins are described in TABLE 2. CLDN1 is ubiquitously expressed in most tissues of the body (612) and acts predominantly as a barrier builder (325, 475). Message for CLDN1 is detected in human bladders (615, 616, 726), but not in mouse or rat bladders (TABLE 3) (4). In human bladder urothelium, CLDN1 is found along the plasma membrane of the basal and intermediate urothelial cell layers, but is particularly enriched along the basal surface of the basal cell layer (87, 526, 685). As noted above, CLDN2 forms a high-conductance cation-selective pore (18, 230, 303, 722). It is expressed in “leaky” epithelia such as the proximal tubule of the kidney and the intestinal crypts (198, 299, 518). Somewhat surprisingly, in mouse bladders, Cldn2 message and CLDN2 protein are expressed in perimembranous and cytoplasmic vesicular elements in all three cell layers (4). In rats, CLDN2 is localized to the basal and intermediate cell layers, as well as the lateral surfaces and tight junction region of the umbrella cell layer (4, 504). CLDN3, a ubiquitously expressed barrier forming claudin (487), is restricted to the apicolateral surface of the umbrella cell layer in humans (526, 685, 701) and in mouse urothelium (225). Multiple studies indicate that CLDN4 is barrier forming, increasing TER in model epithelia (107, 415, 485, 574, 617, 720, 721). However, Hou and co-workers (303, 304) used shRNA technologies and site-directed mutagenesis to demonstrate that CLDN4 interacts with CLDN8 and that this association is required for an anion-selective paracellular pathway in cell lines derived from the collecting duct. CLDN4 is present in human and rodent urothelium with higher expression in the umbrella and intermediate cell layers than the basal one (4, 87, 526, 616, 654, 685, 701, 726). In rabbit, rat, and mouse bladders, CLDN4 is detected in the umbrella cell tight junctions and at regions of cell-cell contact in the intermediate and basal cell layers of the urothelium (4, 225).

Table 2.

Claudin expression in the urothelium

CLDN	Functional Property	Tight Junction Charge Selectivity	mRNA	Reference Nos.	Protein	Localization in Urothelium	Reference Nos.
1	Barrier-builder	Decreased permeability for cations	H	615, 701	H (IHC)	Basal and intermediate cell layers	87, 526, 685, 701
2	Pore-forming	Cationic	H/M	4, 615, 701, 726	H (IHC); R/M (IF)	Umbrella cell apical cytoplasm; surface of basal and intermediate cell layers	4, 504, 685, 701
3	Barrier-builder	Decrease permeability for cations	H	701, 726	H (IHC)	Umbrella cell tight junction	526, 685, 726
4	Barrier-builder	Anionic	H/M	4, 615, 701, 726	H (IHC); Rb/R/M (IF)	Umbrella cell tight junction and basolateral surface; surface of basal and intermediate cell layers	4, 87, 114, 380, 526, 654, 685, 726
5	Barrier-builder	Decrease permeability for cations	H	701, 726	H (IHC)	Umbrella cell basolateral surface	654, 701, 726
7	Ambiguous	Anionic/ambiguous	H	701, 726	H (IHC)	Umbrella cell tight junction and basolateral surface; surface of basal and intermediate cell layers	87, 526, 685, 701, 726
8	Barrier-builder	Anionic	H/M	4, 726	Rb/R/M (IF)	Umbrella cell tight junction	4, 114
10	Pore-forming	10a anionic*; 10b cationic	H	726	ND	?
11	Ambiguous	Anionic	H	42	H (IHC)	Nominally expressed in all cell layers of the urothelium	42
12	Ambiguous	ND	M	4	R/M (IF)	Umbrella cell tight junction	4
13	ND	ND	M	4	ND	?

^*Claudin-10 exists in several splice variants, the two most common isoforms are 10a and 10b (272). However, the isoform(s) expressed in the urothelium has not been defined. H, human; R, rat; Rb, rabbit; M, mouse; IHC, immunohistochemistry; IF, immunofluorescence; ND, not determined.

Table 3.

Gene expression analysis of claudins in the rat urothelium

Gene	Primers	Expression in Bladder	Validation
Cldn1	U: ccaacgcggggttgcagctt; L: gcgatcagccccagcaggatg	ND	Kidney
Cldn2	U: ccatggcctcccttggcgtc; L: atgccagtgctgtgggtcgc	ND	Kidney
Cldn3	U: cggtgtcctggtccgccaac; L: gtggatcgcggcgcggaata	ND	Kidney
Cldn4	U: agcaacatcgtcacggcgca; L: ttgctggcgggaacagagcg	+
Cldn5	U: caggcagctcgggcactcac; L: aatcgccgttggccgtggtt	ND	Kidney
Cldn6	U: tgcttggctgggtcaacgcc; L: tgcacttggctccggcaagg	+
Cldn7	U: cgggcctgcaactactgggc; L: aggagcggggtgcacggtat	+
Cldn8	U: gtgctgcgtccgtcctgtcc; L: gcgccgtggtccagcctatg	+
Cldn9	U: cgcaggggtcctcctcctcc; L: cggggcctctcgaagtggga	ND	Brain
Cldn10†
Cldn11	U: ccgaatgggccacgagcctg; L: cacagcaccgatccagcccg	ND	Brain
Cldn12	U: agtgcgcccggtatgacgga; L: gcagcaaggcacccattgcg	+
Cldn13*
Cldn14	U: actatcctgccgcactggcg; L: agggtctgtaggggccctcg	ND	Liver
Cldn15	U: ctaggcatggtggggctccg; L: acagatgccgcccaggatgga	ND	Intestine
Cldn16	U: gccacccggacggactgttg; L: gtgcccgagtcaccaccagc	+
Cldn17	U: agcagcgcgtgcactcatgt; L: agagcagcccgcctgcaatg	+
Cldn18	U: cagcggcatgggtgggatgg; L: ggccacggcaagcaatgcac	ND	Lung
Cldn19	U: tctgttcgtcggctgggcct; L: gccgttgagggtccagagcg	ND	Lung
Cldn20	U: gctgccactctgctgccaaa; L: tagccctcgcagcctgcacat	+
Cldn21*
Cldn22	U: gggagggaccctgctctgga; L: ggaacagggcctccccgaac	ND	Eye
Cldn23	U:gcatggtgctcacgccctgt; L:gggtgtgagtgcgggttccg	+

ND, not detected. *Not present in rat. †Primers could not be validated. Total RNA from the rat urinary bladder, kidney, brain, intestine, eye, lung, or liver was isolated using the RNAqueous-4PCR kit (Ambion) following the manufacturer instructions. The AccuScript PfuUltra II RT-PCR kit (Stratagene) was used for cDNA synthesis and PCR amplification. PCR products were resolved on 2% agarose gels, and bands with the expected molecular size were cut and gel extracted. PCR products were verified by sequencing using the TOPO TA cloning for sequencing kit (Invitrogen).

In human bladder samples, CLDN7 exhibits a similar distribution to that of CLDN4 (87, 526, 685, 701), although CLDN7 is apparently not expressed in the umbrella cell layer (87, 654, 726). While no mRNA for Cldn7 is detected in mouse bladder (4), CLDN7 protein is reported in mouse ureteric umbrella cells (225). CLDN8, which is barrier forming (779), localizes primarily to the tight junctions of the umbrella cells in mouse and rat bladders (4, 114, 380). CLDN12, which has no ascribed function, is similarly localized to the umbrella cells tight junctions (4). This list is not complete, as other claudins are likely to be expressed in the urothelium. For example, in addition to the previously identified claudins in rat (CLDN4, CLDN7, CLDN8, and CLDN12), we have also identified mRNAs for Cldn6, Cldn16, Cldn17, Cldn20, and Cldn23 (TABLE 3). TJP1 and the integral membrane protein OCLN (occludin) are also present at the tight junction of the umbrella cell layer (4, 114, 403). Finally, when cultured human urothelial cells are induced to undergo differentiation by growth in adult bovine serum and Ca²⁺, they express two splice variants of TJP1 (TJP1α^– and TJP1α⁺); however, the functional significance of this expression is not well understood (650).

In addition to the striking number of claudins expressed in the urothelium, there are also reports of regional differences in the distribution of tight junction-associated proteins. For instance, mRNA levels for CLDN1 and CLDN4 are significantly higher in the trigone than in the dome of human bladder (616). However, no difference in mRNA expression levels for TJP1 are observed between these two regions of the human bladder (616). In rabbit bladders, there are regional differences in the expression of TJP1 (4). In the neck and equatorial regions of the bladder, TJP1 is restricted to the tight junctions of the umbrella cell layer, whereas in the bladder dome it is localized to both the apicolateral tight junctions of umbrella cells as well as to the sites of cell-cell contact in the intermediate and basal cell layers (4). The physiological implications of these findings remain to be determined.

4. Regulation of umbrella cell paracellular permeability

The urothelium is subjected to mechanical stretch, to osmotic pressure caused by the difference in tonicity between the urine and bladder interstitium, and to hydrostatic pressure caused by the accumulation of fluid in the closed compartment formed by the bladder. Recently, Carattino et al. (114) investigated the effects of bladder filling and voiding on the structure and function of the umbrella cell tight junction. These authors reported that bladder filling promotes a significant increase in the perimeter of the tight junction ring, which is rapidly reversed upon voiding. They also observed that when rabbit urothelium, mounted in Ussing chambers, is exposed to stretch, there is a significant drop in both the overall TER as well as tight junction-associated resistance. In other words, the umbrella cell tight junction becomes leakier to ions when the urothelium is stretched. Interestingly, even with a 10-fold drop in TER, the integrity of the urothelial barrier is maintained, and no significant permeation of biotin, fluorescein, or ruthenium red is observed across the urothelium (114). The molecular cause of the drop in TER or its functional significance is currently unknown (although one possibility is described below). However, reversible changes in pore opening, movements of pore-forming claudins into and out of the tight junction, or changes in strand number could all account for these differences.

5. The urothelial tight junction and its relation to bladder function and disease

Other than in situations where CLDN2 is overexpressed, CLDN3 expression is knocked down, or Cldn4 gene expression is knocked out (see discussion below), there are few insights into the role of individual claudins in bladder function and disease. What is known is summarized below.

a) cldn2.

Interstitial cystitis/bladder pain syndrome (IC/BPS) is a complex chronic bladder condition that affects more than 4 million adults in the USA (62, 373). Symptoms include urinary urgency, frequency, nocturia, and bladder and/or pelvic pain, in the absence of any identifiable cause (62, 280, 373, 512, 535, 536). Numerous hypotheses have been put forward to explain the pathogenesis of this disease (120). Recently, Sanchez Freire et al. (615) examined the expression of genes involved in urothelial barrier function, bladder contractility, and inflammation in patients with this condition. They found that mRNA levels for genes encoding TJP1, OCLN, F11R (JAM-1), and CLDN1 were significantly downregulated in biopsies taken from patients with IC/BPS, while message for CLDN2 was upregulated by at least 90-fold (615). While urothelial barrier function was not evaluated in these patients, the findings are consistent with previous studies describing the presence of a dysfunctional urothelial barrier in patients with IC/BPS (338, 559–561) as well as altered tight junction function in cultured human urothelium isolated from patients with IC/BPS (793).

The urothelium endogenously expresses low amounts of CLDN2 (504), which as described above forms a cation-selective pore and is normally expressed in leaky epithelia such as the proximal tubule of the kidney and the enterocytes lining the small intestine (27, 398, 778, 790). Recently, Montalbetti and co-workers assessed the impact of overexpressing CLDN2 on urothelial barrier permeability, overall bladder function, and pain behavior in rats. Overexpression of CLDN2 in rat urothelium increases the permeability to cations, but not toward large uncharged organic molecules such as fluorescein (332.3 Da) or sulfo-NHS-biotin (443.4 Da) (504). Significantly, these rats present the cardinal features of IC/BPS including inflammation in the bladder mucosa and lamina propria, increased voiding frequency, and pelvic allodynia (exaggerated response to touch) (504–506). The increased bladder activity and pelvic sensitivity result from the apparent sensitization of a population of bladder sensory neurons with tetrodotoxin-sensitive action potentials (505), likely Aδ in origin (168). Further study revealed that the inflammation, bladder hyperreflexia, and pelvic allodynia results from the leakage of urinary K⁺ into the bladder interstitium (506). Consistent with this notion, CLDN2-overexpressing rats fed a diet with low K⁺ content neither develop inflammation in the urinary bladder nor do they exhibit pelvic allodynia (506).

Finally, combining molecular and electrophysiological approaches, the mechanism of bladder sensory neuron sensitization in this model was assessed (503). In response to CLDN2 overexpression in the umbrella cell layer, there is an increase in the activity of tetrodotoxin-sensitive, voltage-gated Na⁺ channels and voltage-gated Kv2/Kv9.1 K⁺ channels, which maintain repetitive firing in the face of afferent sensitization (503). It remains to be determined if CLDN2 overexpression is causative of IC/BPS, or if it occurs in response to the underlying defect. Regardless, the effects of overexpressing CLDN2 would likely exacerbate the symptomology associated with IC/BPS.

b) cldn3.

Smith et al. (650) investigated the role of CLDN3 in the tight-junction barrier function of normal human urothelial cells. This group reported that CLDN3 expression increased significantly during the differentiation of the urothelial layer when the tight junctions achieve maximal resistance. CLDN3 knockdown using shRNA technology impaired the formation of a tight barrier. While the impact of this finding on native urothelium and/or bladder function are unknown, it does appear that CLDN3 expression is essential for the formation of high resistance tight junctions in human cultured urothelium.

c) cldn4.

Fujita et al. (225) reported that deletion of mouse Cldn4 results in urothelial hyperplasia and thickening of the urothelium in the renal pelvic region and ureters. Not unexpectedly then, Cldn4 KO mice develop a progressive hydronephrosis as they age, leading to increased mortality (225). Overall tight-junction structure is not affected by the deletion of Cldn4. However, the authors observed a significant diminution of CLDN8 expression and the accumulation of CLDN3 at the tight junction of epithelial cells in the distal nephron. In contrast, the urothelium of Cldn4 KO mice exhibits higher levels of CLDN7 expression when compared with wild-type mice, but there are no changes in the expression patterns of CLDN3 or CLDN8. Significantly, the urothelial barrier to large organic molecules is not impaired by deletion of Cldn4. As a whole, the results of this study support the hypothesis that CLDN4 regulates urothelial proliferation and hence plays an important role in maintaining the homeostatic integrity of the urothelium.

F. Pressing Questions Regarding the Regulation and Function of the Urothelial Barrier

The AUM and associated uroplakins have received deserved attention over the past 50+ years, but much less is understood about the umbrella cell apical membrane or its paracellular pathway.

• 1. If upwards of 90% of the apical surface comprises AUM plaques, then where are all of the non-uroplakin apical membrane proteins? Presumably, these other proteins are localized to hinge regions, but other than localization for urohingin, this has not been reported.
• 2. Are the biophysical properties of umbrella cell apical surface and the properties of the umbrella cell junctional complex identical in the pelvis, ureters, bladder, and urethra? Because these membranes encounter different mechanical stimuli (see discussion below), and are exposed to urine for different periods of time, it is possible they will not be identical.
• 3. Phosphatidylinositols, such as phosphatidylinositol-4,5-bisphosphate, are critical to membrane function and signaling (137). Yet, there are no insights into the phosphatidylinositol content at the apical membrane of umbrella cells (or other cells in the urothelium). Does its content change during expansion and contraction, or in response to disease?
• 4. How do disease processes such as cystitis affect the lipid properties and composition of the umbrella cell apical membrane? As lipids govern events like cell signaling, membrane fusion, and endocytosis, it is important to understand any changes that accompany disease.
• 5. While the urothelium is rightly billed as the tightest barrier in your body, it has been known for decades that the urothelium has the ability to transport ions, and that it expresses channels and transporters for water and urea. With newer data supporting the possibility that the urothelium can alter the composition of urine, we need a better understanding of when these events occur and how they are regulated. For example, what is the nature of the signal(s) that overdistension is occurring, and which signaling pathways are involved? Here we might imagine that the unidentified stretch-activated channel on the umbrella cell apical membrane may play a role. As well, there is likely to be mediator release from the urothelium, as well as input from subjacent tissues and the central nervous system (see discussion below).
• 6. Can urothelial signaling pathways be targeted to treat conditions like enuresis? If it were possible to selectively increase water re-absorption by the bladder urothelium, it may be possible to prevent or lessen nocturia and bed wetting.
• 7. Important advances have been made in understanding ion permeation and charge discrimination in claudin-based paracellular pores and the organization of the tight junctions. However, fundamental questions remain with regard to the umbrella cell tight junction and its associated role in forming the paracellular barrier. For example, are tight junctions associated with lower urinary tract disease? As noted above, urothelial barrier abnormalities are reported in patients with IC/BPS (338, 559–561), and in urothelial cells infected with E. coli (762). There is a growing appreciation that defects in tight junctions contribute to a variety of conditions including inflammatory bowel disease and various genetic abnormalities (63, 621).
• 8. An additional pressing question is what is the function of the myriad of claudins expressed by the umbrella cell-associated tight junction? Here, there are few insights other than the singular report that knockdown of CLDN3 expression lowers TER in cultured human epithelial cells, and the report of abnormalities observed in Cldn4 KO mice (225, 650). By removing claudins from the umbrella cell tight junction, one-by-one, and in combination, it may be possible to begin to understand which are the critical claudins for forming and modulating the urothelial permeability barrier.
• 9. An additional question is what are the non-tight junction-associated claudins doing, i.e., those along the basolateral surfaces of the umbrella cell layer or plasma membrane of intermediate/basal cells? As tight junctions promote cell-cell interactions, they could be promoting cell-cell adhesion. Or, if held in reserve, they would allow the rapid re-establishment of the tight junction barrier in newly exposed intermediate cells. This is a possibility, because as described in section II, when umbrella cells are removed by protamine sulfate treatment (or chitosan), the underlying intermediate cells begin expressing tight junction-associated proteins at the cell periphery within minutes of injury (402, 729).
• 10. There is also a growing awareness that the tight junction may serve as a sensor to changes in osmolality and hydrostatic pressure gradients (699). Could the umbrella cell tight junction have a similar function?
• 11. Finally, in addition to sensing extracellular stimuli, could the umbrella cell tight junction transmit information to subjacent tissues? Remember from the discussion above, stretching the urothelium causes a large increase in ion transport across the urothelium (114). In this case, K⁺ transport across the tight junction could signal subjacent, afferent nerve processes that bladder filling is occurring. It could also tell the subjacent tissues something about the composition of the urine, particularly when it becomes very concentrated.

IV. MAINTAINING UROTHELIAL FUNCTION IN THE FACE OF A DYNAMIC MECHANICAL ENVIRONMENT

The goal of this section is to explore how the urothelium maintains a functional barrier in the dynamic mechanical environment that is characteristic of the lower urinary tract. After describing its mechanical milieu, we discuss the cellular and tissue adaptations that the urothelium makes in response to these physical forces including shape changes, as well as expansion and contraction of the umbrella cell apical membrane and apical junctional complex.

A. The Mechanical Environment of the Urothelium

The movement of urine from the renal pelvis, down the ureters, and into the bladder is not simply due to gravity. Instead, contractions of the smooth muscle cells that undergird the urothelium stimulate the rhythmic contraction of the renal pelvis (FIGURE 3) (190). A group of specialized cells in the proximal pelvis, called atypical smooth muscle cells, act as pacemakers to drive these events (259, 367). The frequency of the pelvic contractions depends on the species in question and whether the kidney is studied in vivo or ex vivo, but ranges from 2–3 contractions per minute in humans to ~13 contractions per minute in pig kidneys (83, 770). In terms of the urothelial mechanical environment, these contractions expose the urothelium lining the renal pelvis and ureters to changes in wall tension and pressure, as well as shear stress when a fraction of the urine is returned to the pelvis in response to its relaxation (190).

The urothelium lining the ureters, bladder, and urethra are also exposed to mechanical forces during normal lower urinary tract function. For example, once urine enters the ureters, it is transported as a bolus to the bladder by way of peristaltic waves driven by the smooth muscle contractions of the ureter’s tunica muscularis (105, 190). Thus the urothelium lining the ureter likely experiences both shear stress and circumferential wall tension. Interestingly, Dubbins et al. (186) and later Burge et al. (106) demonstrated that urine enters the bladder in a rapid stream in a phenomenon called the ureteric jet effect. If this “spray” affects the bladder urothelium is unknown, but it may expose the urothelium to shear. Within the bladder, the urine slowly accumulates and the intravesicular pressure remains relatively low (e.g., ~5–7 cmH₂O in the rodent bladder) and rises very slowly until afferent wall mechanoceptors, emanating from the pelvic (parasympathetic) and hypogastric (sympathetic) nerves, signal bladder fullness (167, 168). In anesthetized rodents, this occurs at a pressure of ~8–10 cmH₂O. During the micturition phase, information arising from the bladder afferents is processed in the brain, which results in detrusor contraction (167, 168). As a result of this contraction, pressure in the bladder rises to ~35–40 cmH₂O, which returns to baseline as the inner and outer urethral sphincters relax (the outer sphincter is under control of the pudendal nerve), allowing urine to rush through the urethra (167, 168). Thus, while the urothelium in the bladder is mainly exposed to wall tension (and perhaps pressure), the urothelium in the urethra is exposed to both wall tension and shear stress as the urine rapidly passes over the urethral mucosa.

B. Cellular and Tissue Adaptations to Expansion and Contraction

How does the urothelium maintain its patency in the face of constant distension and contraction? In the case of the bladder and ureters, the urothelium is an integral part of the mucosa, which also includes the subjacent connective tissue-rich lamina propria. The other component normally associated with the mucosa is a muscularis mucosae, which is present in human bladders (albeit a discontinuous one) and that of guinea pigs (179, 234), but is essentially absent in mice and rats. In its unfilled state, the mucosa of the ureters and bladder are highly folded, forming what are termed rugae. As the ureter is typically contracted, this folding gives rise to the characteristic star shape observed when the ureter is cut in cross section. Upon distension, the rugae flatten out without visible distension of the underlying collagen fibrils. Subsequently, the highly coiled collagen fibrils found in the region between the lamina propria and detrusor smooth muscle layers become straighter (130, 302). Presumably, recoiling of the collagen fibrils is the major force driving refolding of the mouse rugae; however, this has not been experimentally confirmed. Any role for the muscularis mucosae in these events is unknown, although such a role is proposed for the muscularis mucosae of other organs (714).

The most likely function for the rugae is to maintain a reserve of mucosal surface area during bladder filling. Thus unfurling of rugae likely contributes in a major way to the bladder’s characteristically high degree of compliance (i.e., during the filling phase, the bladder exhibits little increase in pressure as the volume increases) (130). Further testing this possible function is a difficult proposition, but could be done if one could prevent rugae from unfolding, perhaps by manipulating the extracellular matrix and/or smooth muscle function. Intriguingly, refolding of the bladder mucosa may have an important role in the umbrella cell responses to voiding, endocytosis in particular, a function described below.

In addition to these macroscopic changes, there are tissue and cell shape changes that accompany bladder filling. For example, as the bladder fills, the urothelium thins, and in species with multiple intermediate cell layers, the urothelium appears to have fewer cell layers (295). What transpires during this transition is unknown, but it is surmised that the intermediate cell layers are sliding past one another, all the while maintaining their cell-cell contacts. However, this has not been determined experimentally. In addition to these tissue changes, the shape of the umbrella cell dramatically changes during filling, assuming a very flat, squamous morphology in the filled bladder, which rapidly reverts to its prestretched shape upon voiding. Accompanying this shape change are large transitions in umbrella cell apical diameter, which is ~30–50 µm in the relaxed state and ~50–150 µm in the stretched state. Other umbrella cell transformations are discussed below.

C. Expansion of the Umbrella Cell Apical Surface Area in Response to Bladder Filling: Role of Exocytosis

The umbrella cell holds another tool that it can use to maintain barrier function in response to bladder expansion: increasing its apical surface area. Here, it is important to remember that biological membranes do not stretch (30), and thus once the rugae unfold, the umbrella cell’s apical surface area becomes an important limiting factor in how much volume the bladder can hold. If we approximate the bladder as a thin-walled sphere, and if we assume no constraints on expansion of the underlying nonurothelial tissues, then a doubling of apical surface area would lead to an eightfold increase in volume capacity. This is because the area of a sphere is related to the square of its radius (A = 4π·r²), while its volume is radius cubed (V = 4/3π·r³). While our discussion is primarily focused on the role that exocytosis plays in relationship to barrier function, it is important to appreciate that exocytosis is the mechanism by which receptors, channels, and other proteins such as uroplakins are deposited at the apical (and basal) plasma membrane domains. Furthermore, release of mediators such as ATP depend, in part, on exocytosis (279, 747). Thus exocytosis also likely plays critical roles in urothelial sensory/transduction function (the subject of sect. V).

Porter et al. (576) were among the first to propose that exocytosis of DFVs may play a role in umbrella cell apical membrane expansion. They posited that the cytoplasmic surface of the hinge region would fuse with the apical surface and then open like a clam shell to insert new membrane. Evidence in support of this hypothesis includes the following. Studies by Minsky and Chlapowski (492) established that the numerical density of DFVs decreases with bladder filling, and Lewis and de Moura (422) were the first to demonstrate that experimental “punching” of native urothelial tissue leads to expansion of umbrella cell apical membrane. More recent studies used stereology to confirm that apical surface area increases in response to bladder filling (with a corresponding decrease in DFV-associated surface area) (705). Moreover, biochemical studies confirm that filling induces the insertion of new apical membrane (782), and when hGH is exogenously expressed in the urothelium, it is packaged into DFVs and secreted in response to stretch and other physiological regulators (356, 359, 360).

1. Physiological regulation of apical exocytosis

In terms of the physiological regulation of exocytosis, there are several insights that have emerged over the past several years. Yu et al. (782) used isolated bladder mucosa mounted in Ussing stretch chambers to demonstrate that it is changes in membrane tension and not pressure per se that trigger apical exocytosis. The population of DFVs have sufficient membrane to expand the apical surface by 300% (425, 705); however, as described below, endocytosis places a limit on these changes. Stretch stimulates production of cAMP, which in turn stimulates exocytosis (173, 705). This effect appears to be dependent on activation of protein kinase A (705), but this has not been established conclusively. Moreover, the isoform(s) of adenylyl cyclase involved in these events is unknown.

Experiments using Ussing chamber-mounted urothelium reveal that exocytosis proceeds in two waves. The first wave, referred to as “early-stage exocytosis,” is triggered by even small changes in tension, proceeds rapidly (within seconds), and results in very large changes in surface area (>100%) (782). It is insensitive to cycloheximide, an inhibitor of protein synthesis, and to brefeldin-A, an inhibitor of membrane traffic along the biosynthetic transport pathway (49, 782). Like other regulated secretory pathways, early-stage exocytosis also requires Ca²⁺ and is inhibited when Ca²⁺ is removed from the apical bathing solution, or when the preparation is treated with inhibitors of inositol trisphosphate- or ryanodine-sensitive Ca²⁺-induced Ca²⁺ release pathways (173, 746, 782). Taken together, these results indicate that early-stage exocytosis is most likely the result of DFV exocytosis, as there is no other source of preformed membrane vesicles that could account for these rapid and massive area changes. Further analysis reveals that “early-stage exocytosis” is likely triggered by cation flux through apical channels, possibly the amiloride-sensitive ENaC as well as a nonselective cation channel sensitive to Gd³⁺ (ENaC is not sensitive to this metal ion) (782). Other features of early-stage exocytosis include a requirement for actin and possibly intermediate filaments, but not microtubules (782). Finally, early-stage exocytosis is followed by a stretch-induced compensatory endocytosis that limits the extent of apical surface area change by recovering apical membrane (705, 782). This brings the area change down, but not to pre-stretch levels as long as some tension is maintained in the system.

The early exocytic events are followed by “late-stage exocytosis,” first described by Balestreire and Apodaca (49), because it trails the early-stage responses (782). It proceeds slowly and gradually, increasing surface area over several hours. It is regulated in a manner that is very different from early stage events. It is inhibited by cycloheximide and brefeldin-A, indicating a need for new protein synthesis as well as biosynthetic traffic (782). It is insensitive to inhibitors of Ca²⁺ release, ENaC, or nonselective cation channels (782); however, it is dependent on apical epidermal growth factor receptor (EGFR) transactivation (49). In this signal transduction cascade, metalloproteinase-dependent cleavage of an EGFR ligand, typically heparin-binding EGF (HB-EGF), leads to autocrine activation of the EGFR and activation of downstream signaling cascades including ERK1/2 (749). The latter triggers changes in gene expression. Indeed, an HB-EGF → EGFR → ERK cascade appears to drive late-stage exocytosis (49). In general, the upstream signal that triggers HB-EGF cleavage, the putative genes that are activated by ERK, and the downstream targets of these genes are currently unknown.

Apical exocytosis can be modulated by external stimuli, independent of stretch. For example, a number of agonists of P2X receptors (e.g., ATP, benzoyl-ATP, α,β-methyl-ATP, ATP-γ-S, and 2-methylthio-ATP), as well as agonists of P2Y receptors (typically ADP and UTP) trigger a slow steady increase in apical exocytosis (747). Furthermore, adenosine also stimulates a slow, gradual rise in exocytosis, and via a pathway that is strikingly similar to that described for late-stage exocytosis (784). Prakasam et al. (577) used adenosine or an A₁ adenosine receptor agonist (2-chloro-N⁶-cyclopentyladenosine; CCPA) to trigger a G_i-G_βγ-phospholipase C-protein kinase C (PKC) cascade that triggered ADAM17-dependent HB-EGF cleavage, EGFR transactivation, and apical exocytosis. The cytoplasmic tail of rat ADAM17 contains a conserved serine residue at position 811, which resides in a canonical PKC phosphorylation site, and that is phosphorylated in an adenosine-dependent manner. Preventing this phosphorylation by expression of ADAM17-S811A (in an ADAM15 knockdown background), or expression of a tail-minus construct impairs A₁ adenosine receptor-induced apical exocytosis (577). Thus, like the late-stage responses to stretch, EGFR transactivation is central to these adenosine-stimulated events. Currently, it is unknown if the other purines act by the same or similar pathways. Interestingly, in response to stretch, the urothelium releases ATP, ADP, AMP, and adenosine (214, 578, 747, 784). Thus it is possible that the late-stage response is regulated by an autocrine purinergic pathway that maintains exocytosis in the face of continuous stretch (747).

2. The membrane trafficking machinery that drives apical exocytosis

The focus of this section is on the apical trafficking pathways and molecular machinery that transports uroplakins (and other proteins) from their site of folding and assembly in the endoplasmic reticulum, through the Golgi apparatus and associated trans-Golgi network (TGN), eventually leading to their exocytosis at the apical surface (FIGURE 8A). The endoplasmic reticulum is where uroplakin heterodimer formation occurs, followed by further processing in the Golgi (reviewed in Ref. 764). The TGN, a major waystation where cargoes are sorted into different vesicular/tubular carriers before being transported to their final destination, is the site where uroplakins and likely other apical cargoes are packaged into nascent DFVs (296, 313). At least in mice, these early vesicles are discoidal in shape, but mature into fusiform vesicles, which form stacks that are originally oriented parallel to the plasma membrane. After movement through a “chicken wire-like” subapical KRT20-rich network (359, 731, 743), the fusiform-shaped vesicles become separated from one another, and a subset of these vesicles become oriented with their long axis perpendicular to the plasma membrane (312). These changes are consistent with the hypothesis put forward by Porter et al. that vesicles fuse end on (576). If a similar maturation of DFVs occurs in other species is difficult to assess morphologically as rabbit DFVs are all discoidal, and in rat the vesicles are mixed discoidal/fusiform, including those vesicles in close proximity to the apical surface. Although described as lacking a cortical actin cytoskeleton at its apical surface by some (381, 603), umbrella cells do indeed have an actin cytoskeleton under this surface (359, 751), which must be traversed for vesicles to fuse with the apical plasma membrane. Moreover, the actin cytoskeleton is a crucial component of microplicae (31, 395), which as noted above protrude from the apical surfaces of umbrella cells. Finally, DFVs undergo regulated secretion, a process in which a pool of release-ready DFVs accumulate under that apical surface until a suitable signal, e.g., stretch, stimulates their Ca²⁺-dependent exocytosis.

FIGURE 8.

Exocytic and endocytic traffic in umbrella cells. A: discoidal and/or fusiform-shaped vesicles (DFVs) and their associated apical cargoes, including uroplakins (UPKs), are assembled in the trans-Golgi network. There appear to be two populations of DFVs: one that is RAB11A/RAB8A⁺ and another one that is RAB27B⁺. Whether DFVs emerge presorted into distinct RAB11A ⁺ or RAB27B⁺ populations of vesicles (option 1), or emerge as a single population that subsequently undergoes sorting (option 2) is unknown. DFVs are transported to the apical pole of the umbrella cell, which is driven by the myosin motor proteins MYO5A and MYO5B. Although not shown, RAB11A/RAB8A can also recruit MYO5A. The final step in exocytosis, fusion, is mediated by SNAREs. While SNAP23 is a likely t-SNARE, the exact identity of the syntaxin isoform(s) involved is unknown. Possible candidates include STX1A, STX2, STX3, and STX1, and these may differ depending on the vesicle pool with which they interact. Those syntaxins not engaged in fusion complexes exist in an autoinhibited conformation, which results from interactions between the molecules NH₂-terminal H_abc, α-helical region, and the COOH-terminal half of the molecule. For clarity, DFV-associated SNAP23 and cognate syntaxins are not shown, but both are known to be localized to vesicles (presumably in an inactive conformation). There are several potential v-SNAREs including VTI1B, VAMP7, and VAMP8. The latter has been shown to cause accumulation of subapical DFVs in mouse knockout models (751), and is indicated as being associated with DFVs. [Adapted from Gallo et al. (237).] B: localization of endocytosed wheat germ agglutinin-FITC (green) and TJP1 (red) in umbrella cells. TJP1 marks the tight junction, but is also associated with the wheat germ-labeled endosomes in response to voiding. [From Khandelwal et al. (361), with permission from EMBO Press.] C: cationized ferritin was instilled into the bladder lumen, and the tissue fixed and processed for TEM after voiding. [From Khandelwal et al. (361), with permission from EMBO Press.] D: endocytic organelles in the rat umbrella cell. EL, endolysosome; ILV, intraluminal vesicle; LYS, lysosome; MVB, multivesicular body. [From Truschel et al. (703), with permission from the Public Library of Science.] E: model for regulation of apical endocytosis in umbrella cells. In response to basolateral stretch (arising from filling-induced distension or voiding-induced refolding), integrins signal through phosphatidylinositol 3-kinase (PI3K) and focal adhesion kinase (PTK2) to stimulate RHOA activity, possibly by activating the RHOA GEF, ARHGEF28. RHOA likely stimulates actin polymerization, which helps drive apical endocytosis. RHOA is inactivated by a GAP, possibly ARHGAP26, which along with endophilins may promote formation of apical endocytic vesicles. The final step, fission, is stimulated by DNM2. A small fraction of endocytosed apical membrane (and a small amount of internalized fluid) may be recycled, but the majority appears to be delivered to lysosomes by way of the MVB. The urothelial-specific protein SNX31 promotes the incorporation of uroplakins into ILVs.

3. RAB GTPases and DFV exocytosis

Driving exocytic traffic is a molecular machinery that selects (i.e., sorts) cargo molecules, packages these cargoes into transport carriers such as DFVs, and then delivers the vesicles to the cell surface where the cargoes are released in response to vesicle fusion. The sorting receptor(s) that recognizes apical cargoes such as uroplakins and facilitates their incorporation into DFVs is unknown, but DFVs are acidified (273), and in other cell types secretory granule acidification is critical for their function (566).

The one machinery that is involved in all of the aforementioned steps is the RAB family of GTPases (FIGURE 8A) (reviewed in Refs. 495, 791). Small in mass (~20–25 kDa), these membrane-associated cytoplasmic proteins form the largest family of GTP binding proteins within the RAS superfamily. For example, 70 isoforms are known in humans. Like all GTPases, RABs cycle between an active GTP-bound state and an inactive GDP-bound state. These states are regulated by accessory proteins. For example, guanine-nucleotide exchange proteins (GEFs) stimulate the RAB protein to exchange GDP for GTP, leading to RAB activation. In contrast, GTPase-activating proteins (GAPs) stimulate GTP hydrolysis, returning the RAB to its inactive, GDP-bound state. GTP binding activates the RAB protein by inducing a conformational change, exposing “switch regions” that are recognized by effector molecules. These effectors are highly varied and can include GEFs/GAPs for other RAB proteins, cargo molecules, molecular motors involved in vesicle transport, and components of the fusion machinery. In general, RABs associate with specific organelles in the cell (e.g., the endoplasmic reticulum, the Golgi, or secretory vesicles), although multiple RABs can be found associated with any one compartment. In the latter case, this may indicate a compartment where sorting of different cargoes is occurring (e.g., the TGN). Alternatively, some RABs act as part of a cascade (reviewed in Refs. 495, 595). In this case, an upstream RAB binds the exchange factor for a downstream RAB, recruiting it to the membrane. In turn, the downstream RAB recruits the GAP for the upstream RAB, triggering its inactivation and in some cases its removal from the membrane. Because each RAB recruits a different subset of effectors, this system allows for the ordered progression through a membrane trafficking pathway.

Several RABs are localized to mouse and rat umbrella cell DFVs including RAB8 (which has A and B isoforms), RAB11A, RAB27A (rats only), and RAB27B (129, 237, 359, 360, 751). Although RAB27B was the original RAB localized to DFVs (129), RAB11A was the first DFV-localized RAB with a documented role in DFV exocytosis (360). Interestingly, the rat urothelium does not express the closely related isoform RAB11B, but it does express the related RAB25 GTPase, which localizes to vesicles in the intermediate cell layers, but not those in umbrella cells (360). Expression of a dominant-active mutant of RAB11A, which cannot hydrolyze GTP, stimulates apical exocytosis and release of hGH (360). In contrast, expression of a dominant-negative (DN) mutant of RAB11A, or use of an RAB11A-targeted shRNA, inhibits exocytosis (237, 360). Although RAB11A is classically described as modulating recycling in the endocytic pathways, there is no evidence that it does so in umbrella cells. For example, expression of DN-RAB11A, which traps endocytic cargo in the recycling endosomes of other cell types (589), does not cause any accumulation of endocytosed fluid or membrane markers in umbrella cell DFVs (360). RAB8 is also localized to DFVs (the antibody used in these studies is not isoform specific) (359), and similar experiments to those described above confirm that RAB8A is also required for DFV exocytosis (359).

In ciliogenesis and apical lumen formation, RAB11A acts upstream of RAB8A in a cascade that regulates vectorial delivery of cargoes to the cilium or plasma membrane (100, 212). There is some indication that a similar cascade may operate in umbrella cells (359). First, RAB11A⁺, RAB8⁺ DFVs are found mostly in the region of the KRT20 network, while RAB11^–, RAB8⁺ DFVs are found very close to the apical membrane. Second, expression of DN-RAB11A prevents the apical accumulation of RAB8-positive DFVs, whereas expression of DN-RAB8A has no effect on the distribution of RAB11A-positive vesicles. Again, this indicates that RAB11A is acting upstream of RAB8A, possibly recruiting it to DFVs. In ciliogenesis and lumen formation, RAB11A recruits RABIN8, which is a GEF for RAB8A (100, 212). Interestingly, RABIN8 is also localized to DFVs, and its overexpression stimulates stretch-induced exocytosis (359). However, a mutant of RABIN8 that lacks GEF activity also stimulates exocytosis. Thus its action is unlikely to be related to activation of RAB8A. Obviously, additional work is needed to better understand the relationship of these RABs in exocytosis and their associated GEFs and GAPs.

The other RAB associated with DFVs is RAB27, a so-called secretory RAB because of its documented role in modulating a variety of regulated secretory pathways (226). RAB27 has A and B isoforms, which are closely related and share a number of effector molecules (226). Although mice only express RAB27B, rats express both RAB27A and RAB27B (237). In rats, RAB27A is enriched in basal and intermediate cells, while RAB27B is enriched in the umbrella cell layer. Somewhat like RAB8, a population RAB27B⁺ DFVs extend past the KRT20 cytoskeleton and thus are in close proximity to the apical cell surface of both rat and mouse umbrella cells (237, 751). Thus RAB27B may act at later steps in the DFV exocytic pathway. In rats, shRNA knockdown of Rab27b expression impairs apical exocytosis, while knockdown of Rab27a is without effect (237). These results support a role for RAB27B in apical exocytosis. Strikingly, the umbrella cells of Rab27b KO mice exhibit an almost complete loss of DFVs, express fewer uroplakins, but have an apparent increase in multivesicular bodies (i.e., late endosomes) (751). In contrast, Rab27a KO mice have no phenotype. The Rab27b KO phenotype is somewhat unexpected as one would predict that if RAB27B acted at steps that preceded fusion, there would be an accumulation of DFVs and not their loss. One possibility is that DFVs are undergoing crinophagy, an autophagic process whereby excess secretory granules are degraded (154, 209, 626). However, an alternative possibility is that in the absence of Rab27b expression a compensatory pathway operates that stimulates exocytosis and recovery of apical membrane by endocytosis. This would account for the decrease in DFVs and the increase in MVBs. A careful analysis of these KO animals, including measurements of exocytosis and endocytosis, would likely be revelatory.

A matter of contention is the relationship of the RAB11A/RAB8A pathway and that regulated by RAB27B. One model is that RAB11A/RAB8A act upstream of RAB27B to promote exocytosis (751). However, a direct test of this hypothesis is more consistent with there being more than one pathway for DFV exocytosis (FIGURE 8A) (237). For example, in rats there is limited colocalization between RAB27B and RAB11A, or between RAB27B and RAB8 (again, an isotype-specific antibody was not available). If RAB11A and/or RAB8A act upstream of RAB27B, then perturbing their function should have an impact on the generation or localization of RAB27B vesicles. However, expressing DA or DN mutants of RAB11A or RAB8A, or downregulating Rab11a using shRNA expression has no effect on the distribution, size, intensity, or number of RAB27B-labeled DFVs (237). Likewise, downregulating Rab27b expression using shRNA does not impact the distribution, size, intensity, or number of RAB11A-labeled DFVs. These results argue against a simple relationship between RAB11A-RAB8A and RAB27B and point to the need for more work to understand the relationship between these GTPases and their associated pathways.

4. RAB effector molecules and DFV exocytosis

As noted above, RABs modify trafficking pathways by recruiting effector molecules. One such molecule is the unconventional myosin motor protein MYO5B, which binds to GTP-bound (active) RAB11A and RAB8A. In umbrella cells, it is localized to DFVs that are RAB11A⁺ and/or RAB8⁺ (359) (FIGURE 8A). Functionally, MYO5B promotes processive short-range, and in some cases, long-range movements along the actin cytoskeleton, which in umbrella cells undergirds the apical plasma membrane (359). The NH₂-terminal region of each MYO5B heavy chain homodimer is capped by a motor domain, whereas the COOH terminus of each intertwined heavy chain binds to cargo vesicles. When expressed in cells, the cargo-binding domain can act in a dominant-negative manner (597). In the case of umbrella cells, expressing this portion of cargo-binding domain inhibits apical exocytosis (359). A closely related molecule to MYO5B is MYO5A, which is also expressed in umbrella cells and partially colocalizes with DFVs (751). Its functional role in DFV exocytosis has not been determined. MYO5A is recruited to RAB27B via MLPH (melanophilin or Slac2A), an actin- and MYO5A-binding protein (227). Mlph KO mice have a phenotype very similar to Rab27b KOs: few DFVs and increased numbers of MVBs (751). Because RAB11A and RAB8A can also interact with MYO5A, it is complicated to resolve the upstream RAB(s) that regulates MYO5A function. The existence of mutants that alter interactions between MYO5A and select RABs may be useful in this case (435). An additional effector of RAB27B expressed in umbrella cells is SYTL2 (751), a synaptotagmin-like protein that in melanocytes binds to RAB27A and promotes association of melanosomes with the plasma membrane via an interaction between plasma membrane-associated phosphatidylserine and its C2A domain (226). In a similar manner, SYTL2 may promote RAB27B-dependent docking of DFVs with the apical plasma membrane before fusion (FIGURE 8A). However, this is not yet shown to be true.

5. Function of MAL in DFV exocytosis

An additional component of the molecular machinery that may regulate DFV exocytosis is the proteolipid MAL, a 17 kDa transmembrane protein with four transmembrane domains that are expressed in T lymphocytes, myelin-forming cells, and polarized epithelial cells (788). MAL was originally implicated in the sorting of those apical cargoes that reside in detergent-insoluble “rafts” before their exit from the TGN of polarized MDCK epithelial cells (580). In umbrella cells, MAL colocalizes with uroplakins in DFVs and as noted above is associated with Sarkosyl-extracted membrane fractions (FIGURE 8A) (797). Interestingly, MAL depletion in MDCK cells had no obvious effect on the subapical accumulation of exogenously expressed uroplakins; however, it inhibits the amount of uroplakins that accumulate at the apical cell surface (797). Thus MAL depletion may not affect exit from the TGN, but instead may act at a later step, possibly fusion. In Mal KO mice, a similar phenotype to that reported in MDCK cells is observed: DFVs accumulate at the apical pole of the umbrella cells, and there is a concordant increase in the amount of uroplakin expression associated with the urothelium (797). To date, no experiments have assessed if apical exocytosis is affected in these mice. However, mice overexpressing MAL exhibit a phenotype somewhat like that observed for Rab27b KO: umbrella cells have fewer DFVs, there is less overall uroplakin content in the cell, and there is a large accumulation of MVBs (797). In this case, the authors argue that MAL overexpression is stimulating exocytosis, and the increased endocytosis leads to the accumulation of MVBs.

6. SNARE expression in umbrella cells

The final step in vesicle fusion is mediated by soluble N-ethylmaleimide sensitive factor attachment protein receptors (SNAREs) (FIGURE 8A). SNAREs can be classified as being vesicle-associated (v-SNAREs) or target membrane-associated (t-SNAREs) (277). In synaptic vesicle fusion, the v-SNARE is the transmembrane protein synaptobrevin (VAMP1), while the t-SNAREs include the transmembrane protein STX1A (syntaxin1A) and the cytoplasmic protein SNAP25, which is membrane bound as a result of its palmitoylation. Prior to fusion, the α-helical SNARE motif present in each v- and t-SNARE assemble to form a four-helix bundle. In this bundle, VAMP1 and STX1A contribute one helix each, while SNAP25 contributes two. Subsequent zippering of this four-helix bundle is thought to provide the energy necessary to promote fusion. Those readers interested in additional information about SNAREs, membrane fusion, and the regulatory proteins that control these events are directed to the following reviews (277, 669, 670).

Born et al. (90) were the first to describe the expression of VAMP2 (synaptobrevin-2), STX1A (syntaxin1A), and SNAP23 (a SNAP25 homolog) in rat urothelium. In addition to Western blotting (which detected a single protein species of the expected molecular weight), they used cryoelectron microscopy to localize each of these molecules to DFV and the apical plasma membrane. The association of t-SNAREs with DFVs may seem odd at first glance, but in neurons, t-SNAREs are a significant component of synaptic vesicles (740). The function of DFV-localized t-SNAREs is unknown, but as apical cargo carriers DFVs may be the mechanism by which these proteins arrive at the apical cell surface. If true, the DFV-associated t-SNAREs are likely to be in an inactive conformation until their arrival at the surface. In the case of syntaxins, an auto-inhibited state occurs when the NH₂-terminal (H_abc) region of syntaxin interacts with the COOH-terminal half of the molecule (449). An alternative explanation is that a pool of DFVs is involved in recycling and may return endocytosed t-SNAREs to the apical surface. However, see the discussion below concerning recycling and DFVs.

A somewhat different story emerges in a recent elegant study performed by Wankel et al. (751). They report no expression of STX1A or VAMP2 in mice or rats, but instead find expression of the t-SNAREs STX2, STX3, and STX11. There is agreement with Born et al. (90) that SNAP23 is expressed, but not SNAP25. The v-SNAREs identified by Wankel et al. (751) include VAMP7, VAMP8, and VTI1B. Both t-SNAREs, SNAP23 and STX2, appear to localize in part to the plasma membrane, but like Born et al., they observe a significant fraction of SNAP23 and STX11 associate with DFVs. The v-SNARE VAMP8 appears to be primarily vesicular in association. Strikingly, the umbrella cells of Vamp8 KO mice became smaller in diameter, but more columnar in shape (751). In addition, these Vamp8 KO mice have abundant DFVs, and possibly accumulate increased numbers of MVBs. The authors of this study also describe a loss of plaques at the apical surfaces of the KO cells, but this is difficult to assess using SEM. However, it does appear that the KO cells may lack surface microplicae. While the results presented are consistent with a defect in exocytosis, this has not been measured directly. Finally, there is a recent report that VAMP5 is expressed in the urothelium lining the ureters and bladder. Intriguingly, Vamp5 KO mice suffer from early death, lung abnormalities, and duplication of their ureters (324). If these defects relate to altered membrane traffic is unknown.

D. Recovery of Umbrella Cell Apical Membrane: Role of Endocytosis

The previous section described how exocytosis and apical membrane expansion contributes to compliance by allowing the bladder to hold more urine with minimal changes in intravesicular pressure. However, the umbrella cell in particular must recover apical membrane after a bolus of urine passes through the ureters or after bladder voiding. While removal of excess apical membrane by fission has been proposed as a mechanism to accomplish this task (90), a more likely solution is that any additional apical membrane is recovered by endocytosis. In support of this contention are numerous studies that report apical endocytosis in umbrella cells (17, 123, 296, 361, 379, 464, 576, 598, 705), and that umbrella cells have a well-developed lysosomal system including endosomes, MVBs, endolysosomes, lysosomes, and autophagosomes (FIGURE 8, B–D) (300, 343, 576, 703). Interestingly, as rats age there is a dramatic increase in the volume of endolysosomes (the product of MVB fusion with lysosomes) and an accumulation of undegraded macromolecules, leading to the production of lipofuscin pigment (567, 703). An age-related reduction in lysosome degradation is apparently the cause of this defect.

In the case of umbrella cells, endocytosis is likely to have numerous functions including 1) removing excess apical membrane, ensuring turnover of membrane components such as uroplakins; 2) increasing membrane tension, a critical function in umbrella cells, which must fine-tune their responses to extracellular mechanical stimuli (511); and 3) modulating signaling cascades, usually by promoting the degradation of cell surface receptors. This latter function is likely very important in regulating crosstalk between the urothelium and subjacent tissues (described in sect. V). 4) Endocytosis, particularly from the basolateral cell surface, is one of the chief mechanisms by which cells gather nutrients from their extracellular environment, and this is likely to be the case for umbrella cells as well. 5) Although not a normal function per se, endocytosis is a portal of entry for pathogens such as E. coli, which as described above are internalized in urothelial cells (20).

1. Apical endocytosis in umbrella cells

While we have some knowledge about the pathways that regulate DFV exocytosis, the regulation of apical endocytosis in umbrella cells is less well understood. Apical endocytosis is stimulated both during filling (i.e., stretch-induced endocytosis) and upon voiding (361, 420, 705, 782), but the relationship between these pathways and whether they share similar modes of regulation is unknown. Somewhat unexpectedly, the physiological stimulus in either case appears to be increased tension at the basolateral surfaces of the umbrella cells, which a mathematical model of membrane traffic in the umbrella cell predicts and which experimental evidence supports (361, 511, 782).

There are multiple forms of endocytosis including phagocytosis (cell eating; which allows for engulfment of whole cells) and pinocytosis (cell drinking). Examples of the latter include macropinocytosis (big cell drinking), clathrin-mediated endocytosis (also known as receptor-mediated endocytosis, the most well studied pathway), caveolar-dependent endocytosis, and a number of non-clathrin-dependent pathways (305, 473, 474). Umbrella cells are not known to undergo phagocytosis, and their apical surfaces lack clathrin-coated pits or caveolae (361). Instead, Khandelwal et al. (361) showed that apical endocytosis in umbrella cells occurs by way of a non-clathrin pathway that requires DNM2 (dynamin2), a GTPase that promotes endocytic vesicle fission, and RHOA, a small GTPase in the Ras superfamily that regulates the actin cytoskeleton, as well as second messenger cascades (FIGURE 8E). The function for RHOA in this pathway is unclear, but the actin dependency of umbrella cell apical endocytosis, and the requirement for RHO-associated kinases (ROCKs), may point to a requirement for RHOA-dependent modulation of the actin cytoskeleton. The requirement for dynamin and RHOA indicates that the umbrella cell endocytosis pathway may be similar to the so-called “RHOA pathway” for clathrin-independent endocytosis, which was first described for interleukin-1B receptor internalization and later shown to also be present in yeast (396, 579). The RHOA pathway is likely to be a variant of the fast-endophilin-mediated pathway (92). This latter pathway is driven, in part, by the endophilins, a small family of proteins that contain a membrane-sculpting BAR domain and an SH3 domain that allows it to interact with DNM1/2 and SYNJ1/2. The latter are phosphatases with activity against several phosphoinositides including phosphatidylinositol-4,5-bisphosphate. It is unknown if apical endocytosis in umbrella cells requires endophilins or not.

A putative sensor for this increased basolateral tension is the ITGB1 (β₁ integrin) receptor, which can sense stretch and is localized to the basolateral surface of the umbrella cells (FIGURE 8E) (352, 653, 758). Interestingly, Kanasaki et al. (348) have reported that conditional urothelial Itgb1 KO mice have urological defects, but the underlying cause(s) is unknown. Indeed, function-blocking ITGB1 antibodies and RGD peptide impair voiding-induced endocytosis, providing experimental support for a role for integrins in modulating umbrella cell apical endocytosis (361). Furthermore, integrins, PTK2 (focal-adhesion kinase), and phosphatidylinositol-3-kinase act upstream of RHOA, and are required for RHOA activation (361). Thus integrins and their associated signaling pathways are an important link between basolateral tension detection and RHOA-dependent apical endocytosis (FIGURE 8E). The molecule(s) that links PTK2 to RHOA activation is currently unknown, but may be a protein like ARHGEF28 (p190-RhoGEF), which acts downstream of integrins and promotes RHOA activation (433). Moreover, the protein ARHGAP26 (Graf1), which was previously implicated in clathrin-independent endocytosis, can function as a RHOA GAP (447). Intriguingly, like the endophilins, ARHGAP26 contains a membrane sculpting BAR domain, and thus may also be involved in the formation of apical endocytic vesicles.

2. The fate of internalized apical membrane

A question that has engaged the field of urothelial biology for decades is what is the fate of internalized umbrella cell apical membrane? In other epithelial cells, endocytosed membrane has one of three fates (35): 1) the membrane is endocytosed and then recycled back to the cell surface; 2) the membrane is internalized and then delivered by transcytosis to the opposite pole of the cell; and 3) the internalized membrane is endocytosed, and then delivered to an early endosomal compartment, which matures into an MVB (late endosome), and then fuses with a lysosome. This fusion generates endolysosomes, which are an important site of macromolecule degradation and turnover.

The classical model for umbrella cell apical membrane dynamics is that DFVs are recycling vesicles (295). If true, then it should be possible to label DFVs with endocytic tracers (before bladder voiding/contraction). These experiments have been performed by multiple groups and over many decades, and the answer is always the same: a tiny fraction of the DFVs are labeled, even when using membrane tracers such as cationized ferritin or lectins such as wheat germ agglutinin (which binds UPK3A) (17, 123, 361, 705). In the case of rabbit umbrella cells, endocytic tracers (both membrane and fluid) enter endocytic structures that are localized in close proximity to the tight junction and are thus named “peripheral junction-associated endosomes” (PJAEs; FIGURE 8B) (361). In contrast, DFVs are found across the whole apical pole of the cell, and are not labeled by these endocytic tracers (FIGURE 8C). Furthermore, and as noted above, even upon expressing DN mutants of RAB11A (which should prevent the egress of endocytosed macromolecules) and incubating for extended periods of time, DFVs are not labeled with endocytic tracers (360). Finally, Grasso and Calderón (260) used hypotonic-isotonic shock to stimulate rat umbrella cells to endocytose fluid or membrane probes that recognize UPK3A. Subsequently, the bladder was filled, and after 5 min, the fluid was collected and then replaced with new fluid. This treatment was repeated for a total of 10 times over a 50-min period of time. Any probe in the collected fluid was scored as recycled. At least in rats fed a control diet, the fraction of fluid that recycled is ~2%, and recycling of the membrane probe is ~7%. In oleic acid-fed animals this increased to 8 and 26%, respectively. In either case, the amount of recycling is small to modest in scope. It would be interesting to determine if similar results are obtained using a more physiological method to internalize the apical tracers. Furthermore, it should be determined if recycling occurs via DFVs, from PJAEs, or some other endocytic compartment.

In contrast, there is considerable evidence that internalized membrane and fluid are delivered to MVBs and endolysosomes (17, 123, 296, 576, 598), but generally not DFVs. For example, cultured urothelial cells that are relatively undifferentiated (i.e., they express low amounts of uroplakins) exhibit relatively abundant endocytosis, while highly differentiated cells exhibit low amounts of endocytosis, and the fate of internalized membrane cargo is delivery to lysosomes. In addition, when the apical surface of native umbrella cells is biotinylated, the fate of the apical membrane proteins internalized in response to stretch is degradation (705). Likewise, the fate of apical fluid and membrane cargo internalized in response to voiding is delivery to lysosomes (361). Guo et al. (273) observed that the Buff mouse, which has a genetic abnormality in Vps33a, has decreased numbers of DFVs and a dramatic accumulation of uroplakin- and AUM-positive MVBs in the apical cytoplasm of their umbrella cells. VPS33A is one subunit of the HOPS complex, which modulates several trafficking pathways including MVB-lysosome fusion (568). The cause of the decreased number of DFVs in Buff mice is not known, but could result from increases in DFV exocytosis and apical membrane endocytosis, or may be a result of crinophagy (273). Similar to Buff mice, the umbrella cells of Scarb2 (LIMP2) KO mice have few DFVs and accumulate large numbers of MVBs (238). In addition, the apical plasma membrane of Scarb2 KO mice loses its scalloped appearance and lacks AUM (238). While the underlying defect is yet to be described, it appears that the function of the lysosome degradative pathway is integral to umbrella cell function.

Lastly, we discuss a manuscript that provides first insights into the mechanism by which uroplakins are targeted to lysosomes. Using a urothelium-enriched cDNA library, Vieira et al. (734) identified SNX31, a member of the sorting nexin family of proteins. The sorting nexins are cytoplasmic proteins that interact with membrane via a phospholipid-binding PX binding motif. They regulate a diversity of membrane trafficking steps including cargo sorting. Interestingly, SNX31 may be urothelial specific, as it is expressed in the mouse bladder, but not in a wide variety of other tissues, including those rich in epithelial cells (734). When its distribution is probed using cryoelectron microscopy, SNX31 is localized to MVBs and their associated intraluminal vesicles, which are small vesicles that form within the MVB and contain membrane cargoes that are targeted for degradation by lysosomes. Biotinylated apical membrane proteins, primarily uroplakins, are delivered to these SNX31⁺ MVBs, and most strikingly is biochemical evidence that SNX31 binds to UPK3A/UPK1B heterodimers (734). While it remains an open question what SNX31 is doing in the MVB, the authors hypothesize that by binding UPK3A/UPK1B it may promote their disassembly, facilitating their inclusion in intraluminal vesicles as a prelude to degradation (FIGURE 8E).

E. The Umbrella Cell Junctional Complex Accommodates Tissue Expansion and Contraction

Viewed from above, the diameter of the umbrella cell is defined by its apical junctional complex and associated cytoskeletal ring. If umbrella cell apical diameter remained fixed during filling, then adding apical membrane would cause the umbrella cell’s apical surface to project into the bladder lumen, effectively decreasing bladder volume (albeit by a small amount). In contrast, if the umbrella cell’s junctional complex is expandable, then the apical surface would be drawn out and flattened as the bladder fills. This latter possibility is more concordant with the squamous morphology of the stretched umbrella cell when viewed in cross section (see FIGURE 9).

FIGURE 9.

Expansion and contraction of the umbrella cell apical junctional complex. A: rats were catheterized and their bladders filled (fill), or filled and voided (void), or never allowed to fill (quiescent). The animals were perfusion fixed, and the distribution of CDH1, actin, and nuclei was assessed. B: upon bladder filling, the umbrella cell apical junctional complex (AJC) expands in a process that requires formin-dependent actin polymerization and RAB13-dependent exocytosis, likely of tight junction- and adherens junction-associated proteins. Upon voiding, the apical junctional complex contracts in a process that requires non-muscle myosin IIA (NMMIIA)-dependent constriction of the junction-associated actomyosin cytoskeleton and DNM2-dependent endocytosis of junction-associated proteins. [From Eaton et al. (192), with permission from American Society for Cell Biology.]

Carattino et al. (114) were the first to report that during bladder filling the diameter of the umbrella cell, and corresponding length or perimeter of the apical junctional complex, increased significantly from ~170 µm in the quiescent bladder to ~250 µm in the filled bladder. Strikingly, the perimeter of the AJR returned to ~160 µm within 5 min of bladder voiding. Thus AJR expansion and contraction is an additional mechanism the umbrella cell uses to accommodate bladder filling and voiding. In a follow-up to these studies, Eaton et al. (192) recently described the cellular events and molecular players involved in regulating these events (FIGURE 9). 1) They observe a continuous ring of the actin crosslinking protein ACTN4 closely apposed to the actin ring that surrounds the junction complex. In addition, a ring of non-muscle myosin II (NMMII), distributed in linear foci, is also localized along the cytoplasmic surface of the junctional complex-associated actin ring. 2) During bladder filling, expansion of the junctional complex requires a dynamic actin cytoskeleton and is driven by actin-polymerizing formins, but not ARP2/3-regulated actin networks (192). Expansion does not require NMMII activity. However, it does require exocytosis, in this case regulated by RAB13, but not RAB11A or RAB8A. Thus, despite the necessity that apical exocytosis of DFVs and expansion of the junctional complex be coordinated, it is not likely mediated by RAB11A- or RAB8A-dependent events. 3) Upon bladder voiding, the contraction of the junctional ring depends on NMMII, actin dynamics, RHOA, and DNM2-dependent endocytosis (192). Taken together, these results indicate that the mechanisms by which umbrella cells maintain continuity in response to bladder dynamics not only include changes in apical membrane surface, but also expansion and contraction of the apical junctional complex. Now with a better understanding of some of the molecular players, it should be possible to better detail the molecular events involved, define how the apical membrane and junction complex events are coordinated, and assess whether these events are tied to urinary tract function and dysfunction.

F. Pressing Questions in Exocytosis and Endocytosis Research

There are several questions that need to be addressed with regard to the membrane trafficking events described in umbrella cells.

• 1. The underlying hypothesis of this section is that exocytosis is necessary to accommodate the changes in membrane tension imposed by bladder filling, promoting barrier function. Is this true? The availability of KO mice that perturb exocytosis (e.g., the Vamp8 KO mice described above), as well as technologies to knockdown or increase protein expression (see sect. VI) should allow investigators to assess how membrane trafficking pathways contribute to the high-resistance barrier formed by the urothelium. For example, one would predict that blocking exocytosis of DFVs would lead to barrier failure (e.g., loss of TER, increased permeability to water and urea) and mechanical disruption of the urothelium in response to supraphysiological distension of the bladder.
• 2. Generating and regulating tension in the apical membrane is likely to be critical to the umbrella cell’s ability to sense and respond to mechanical stimuli, including release of mediators. Because exocytosis lessens membrane tension, while endocytosis has the opposite effect, what happens if tension is too loose or too tight? Mathematical models of these trafficking events indicate that they will perturb bladder function (511). With the use of the same tools described above, it should be possible to directly test these hypotheses.
• 3. How do early- and late-stage exocytosis contribute to bladder function? The ability to selectively block “early-stage” and “late-stage” exocytosis (e.g., using ENaC inhibitors or brefeldin-A, respectively) may allow investigators to directly determine if bladder function depends on these pathways.
• 4. Which signaling pathways act upstream of the RABs to trigger exocytosis? One likely possibility is that stretch (or other stimuli) activates RAB GEFs, which then initiate a signaling cascade that triggers exocytosis. Along a similar line, what are the pathways that regulate endocytosis in umbrella cells? As described above, there is an interesting link between integrins and apical endocytosis (361); however, we do not know the molecular players or how are they connected to one another.
• 5. As a final comment, when exploring the phenotype of KO mice that impinge on the endocytic and exocytic pathways, it is important to measure biosynthetic flux and/or endocytic flux through the associated membrane trafficking pathways. Quantitative assays are available to measure these events (260, 273, 360, 361, 422, 705). This will allow those of us interested in urothelial biology to better understand why some vesicular compartments appear to increase or decrease in size, whereas others do not. In addition, and particularly in those scenarios where multiple tissues express the protein of interest, investigators should employ conditional urothelial KOs, or tools that selectively impact the urothelium (see sect. VI for some of these tools). Otherwise, it is difficult to understand the contribution of the urothelium versus other tissues in the organ wall. This concern is echoed in the next section repeatedly. Fortunately, both constitutive and inducible Upk2:Cre and Upk3:Cre mice have been described (239, 496, 637, 798).

V. COMMUNICATION BETWEEN THE UROTHELIUM AND OTHER TISSUES: UROTHELIAL SENSOR AND TRANSDUCER FUNCTION

The purpose of this section is to explore how the urothelium communicates the state of its local environment to subjacent tissues. The general hypothesis is that the urothelium functions as an integral component of a broader “urothelial sensory web” that promotes reciprocal communication between the urothelium, the underlying tissues, and the CNS (FIGURE 10) (33, 68, 74). The existence of the sensory web can explain why increasing the concentration of ATP into the bladder lumen stimulates bladder hyperactivity (59), or why bladder smooth muscle strips respond differently depending on whether the urothelium is left attached or not, in some cases stimulating contraction and in others inhibiting it (288, 455, 523, 630, 675, 691). These effects are not limited to bladder epithelium, as the urothelium lining the ureters and proximal urethra appears to have a similar ability to modify smooth muscle contraction as well (216, 269).

FIGURE 10.

The urothelium-associated sensory web. The function of the sensory web is to allow bidirectional communication between the urothelium and underlying tissues, in this case, sensory afferent termini. Input pathways expressed by the urothelium include channels that respond to stretch (step 1), or channels and receptors that are stimulated by mediators present in urine (step 2A), released apically from the umbrella cell layer (step 2A, and which promotes autocrine signaling), or released by afferent nerve termini (step 2B). Activation of urothelial-associated input pathways stimulates outputs in the form of mediators that are released apically (step 3A) or basolaterally (step 3B). The mediators released from the serosal surfaces of the urothelium bind to receptors/channels on afferent nerve termini, signaling the current physiochemical status of the urothelium to the central nervous system (CNS) (steps 4 and 5). The model presented focuses on the interaction between the urothelium and afferent nerve termini; however, similar pathways likely exist between the urothelium and interstitial cells, and between the urothelium and smooth muscle cells.

Within the sensory web, the urothelium is hypothesized to function as both a sensor and transducer (33, 68, 74). When acting as a sensor, so-called “input” pathways are thought to allow the urothelium to detect changes in its extracellular environment. These input pathways include a broad array of receptors (e.g., those for growth factors, neurotransmitters, purines, and pathogen detection) and channels (e.g., PIEZO1, TRP family channels, ENaC, and P2X family purinergic receptors). However, as will be discussed below, the critical receptor(s) and/or channel(s) that promotes urothelial sensory function still remain to be defined.

The urothelium is also thought to function as a transducer, decoding input signals, and responding via so-called “output” pathways, promoting urothelial communication. Examples of these outputs include urothelial produced acetylcholine, purinergic compounds, prostaglandins, and nitric oxide (NO). These mediators are proposed to alter the activity of subjacent tissues, effectively modulating bladder function. Current data point to the possibility that defects in either sensor or transducer function can lead to bladder dysfunction (33, 65, 121, 223, 630). Our goals in this section are fourfold: review the local environment of the urothelium; review our current understanding of the known input and output pathways, with a focus on observations in KO mice and humans; review the literature that describes the emerging role of bladder and urothelial circadian rhythms in the regulation of bladder function; and discuss areas of research that have not been adequately addressed in this important subfield of urothelial biology.

A. The Urinary Space, the Bladder Mucosa, and the Subjacent Tissues that Comprise the Organ Wall

To understand urothelial sensor and transducer function, we must first understand the local environment of this tissue, and which cell/tissue types it communicates with. Thus we begin this section by describing the composition of urine that bathes the mucosal surfaces of the umbrella cells, and then continue on to describe the urothelial basement membrane, the lamina propria, and other underlying cellular components of the pelvis, ureter, bladder, and urethra wall.

1. Urinary space

Bathing the apical surface of the umbrella cells is urine (FIGURE 11), a complex fluid whose pH, osmolality, and composition depend on one’s nutritional intake, hydration, and health status. Urine begins when blood is filtered within the renal corpuscle, generating an ultrafiltrate that is processed by the nephron until it emerges from the collecting ducts as urine. However, as noted above, the composition of the urine is not fixed and can be altered during its passage from the pelvis, or even during its storage in the bladder (112, 636). Surprising to some is that urine contains far more than just water, ions (K⁺, Na⁺, phosphates, Cl^–, Ca²⁺, Mg²⁺), and the end products of metabolism such as urea (605). One can also find purines (e.g., ATP and adenosine) (194, 733, 784), prostaglandins (221), cytokines (541), as well as any drug that is filtered by the kidneys. In addition, more than 3,000 membrane, cytoplasmic, and secreted proteins have been identified in urine including structural proteins, growth factors (e.g., epidermal growth factor, neurotrophins, nerve growth factor), and other bioactive peptides (353, 620, 795). About half of these proteins are estimated to come from plasma, or are secreted from the epithelial cells lining the urogenital tract (54, 693). An additional, almost 50% comes from shed cells (which in some cases can be cultured) (565), and the few percent left are present in “extracellular vesicles” (54, 693).

FIGURE 11.

The urothelium and subjacent tissues in the lower urinary tract. Tissues present in the walls of the renal pelvis, ureters, bladder, and proximal urethra include the urothelium, the lamina propria, a muscularis externa comprised of smooth muscle, and either a serosa (comprised of mesothelial cells) or an adventitia comprised of connective tissue components (not shown). BM, basement membrane; BV, blood vessel; IsC, interstitial cell; ImC, immune cell; LP, lamina propria; Me, mesothelium; ME, muscularis externa; MM, muscularis mucosae; Se, serosa; Ut, urothelium. (Cartoon created by Dennis R. Clayton.)

Extracellular vesicles include three different components: exosomes, microvesicles, and apoptotic bodies (54, 693). Exosomes are derived from the intraluminal vesicles that are released when MVBs fuse with the plasma membrane. To date, almost 40 proteomic studies have been published describing the protein constituents of urinary exosomes (reviewed in Ref. 482). These constituents include proteins associated with MVBs, lysosome-associated proteins, cell-type specific markers, proteins derived from every segment of the nephron (e.g., AQP2 and PODXL), adhesion proteins, a broad array of cytosolic proteins, and even cytoskeletal proteins. In addition, exosomes contain lipids and RNAs (including microRNAs). Microvesicles form by budding off of the plasma membrane and contain cytosolic proteins, mRNAs, and miRNAs. Apoptotic bodies are released in response to cell death and not surprisingly contain nuclear material, cellular organelles, as well as membrane and cytosolic proteins (55). Extracellular vesicles are the focus of a large number of recent studies. Beyond serving as useful biomarkers, they are reportedly integral to numerous cell and tissue functions including intracellular communication, immune responses, angiogenesis, and signal transduction in both normal and pathological states (54, 55, 693).

In sum, urine contains hundreds of potential biologically relevant products that by binding to receptors and channels on the apical surfaces of the umbrella cell layer could trigger the sensory/transducer functions of the urothelium, affecting urinary tract function. This communication would not only allow the urothelium to sense its local environment, but also events occurring in upstream organs (e.g., the kidneys), and the body as a whole by way of substances found in serum and filtered by the kidneys (e.g., hormones such as oxytocin and melatonin). Despite the trove of substances found in urine, there are few insights into how urine impacts urothelial function. Finally, we are taught in school that the urine is sterile. In fact, the lower urinary tract contains a microflora that cannot be detected using routine bacterial culture media (694, 695). Because microflora are implicated in a variety of bodily functions (590), it will not be surprising to learn that the lower urinary tract microflora also impacts urinary tract function in both normal and disease states.

2. The mucosa lining the urinary tract

While the terms mucosa and urothelium are often used interchangeably, they are not synonymous, a distinction that has sowed confusion in the urothelial literature (reviewed in Ref. 121). While the urothelium comprises epithelial cells, the mucosa lining the lower urinary tract comprises the urothelium and associated basement membrane, its subjacent lamina propria, and the muscularis mucosae (when present). This specialized multi-tissue structure supports the epithelium and also serves to integrate its functions with those of the underlying tissues (223) (FIGURE 11). Thus, in addition to receiving cues from the urinary space, the urothelium receives input from the underlying tissues and may also communicate with these tissues through the output pathways described below. Changes in communication between these tissues are proposed to lead to abnormal lower urinary tract function (223).

a) the urothelial basement membrane.

The basement membrane sits directly below the basal cell layer of the urothelium, separating the epithelium from the underlying stroma. Basement membranes in general perform critical roles in adhesion, migration, proliferation, and differentiation, and they are also important barriers that are breached during carcinogenesis (276, 354). Once liberated from the basement membrane, cancer cells dedifferentiate, become migratory, and are free to metastasize. Histologically, the basement membrane has two major components (FIGURE 12A). 1) The first is the basal lamina, which is synthesized and secreted by the overlying urothelial cells. The basal lamina includes an electron-lucent clear zone that is closely apposed to the cell membrane and is called the lamina lucida. This region is rich in laminin, fibronectin, and proteoglycans. In the human ureter, the chief laminin expressed in the lamina lucida is laminin 5 (α₃β₃γ₂), although other laminin subunits are also expressed (287). The wider and more electron-dense region called the lamina densa is rich in collagen IV, several isoforms of which are expressed in the urothelium of the lower urinary tract including COL4A1, COL4A2, COL4A5, and COL4A6 (366). 2) The second major component of the basement membrane is the reticular lamina, which is produced by fibroblasts and contains fibrillar collagen.

FIGURE 12.

Basement membrane and intraepithelial afferent nerve termini revealed by transmission electron microscopy (TEM). A: components of the mouse urothelial basement membrane (BM) include the basal lamina, which is comprised of the lamina lucida (LL) and lamina densa (LD), and the reticular lamina. The latter is chiefly comprised of fibrillar collagens. BC, basal cell. B: what is likely an individual afferent nerve fiber (false-colored in green) runs along the serosal surface of a mouse urothelial basal cell (BC). The lamina densa is false-colored blue. An intraepithelial nerve terminus (boxed), located within the basement membrane of the basal cell, is magnified in the inset. The nerve terminus includes a clathrin-coated vesicle (CCV), numerous small clear vesicles (CV), and dense-core vesicles (DCV). An interstitial cell (IntC) is seen below the nerve fiber. (TEM images provided by Steven Truschel.)

The basal cell layer is attached to the basement membrane via hemidesmosomes, a complex structure formed by basal cell-expressed transmembrane proteins α₆β₄ integrin, CD151, and COL17A1 (collagen XVII α-chain1; bullous pemphigoid antigen 2; 180 kDa) and two cytoplasmic proteins DST (bullous pemphigoid antigen 1; 230/240 kDa) and PLEC (plectin) (741). Interestingly, our early understanding of hemidesmosome function and its constituent parts was driven, in part, by studies in the rat bladder cell line 804G, which forms large numbers of these structures (592). The contribution of the basement membrane to the sensory web is not well understood, but growth factors are associated with this structure, and its sievelike properties indicate that it is likely to regulate communication between the urothelium and underlying tissues (787).

b) the lamina propria.

The lamina propria includes an upper cellular layer and a deeper fibrous layer (234). As the urothelium itself is avascular, components of the cellular layer include closely apposed capillaries (some of which are fenestrated on the urothelial side), capillary-associated pericytes, lymphatics, and scattered small- to medium-sized blood vessels. Other components of the cellular layer include afferent nerve processes, interstitial cells, and mast cells, which are typical residents of the human bladder lamina propria (but rare in mice and rats) (586). Other immune cells, including macrophages, lymphocytes, and plasma cells, are also found below and within the human urothelium (300). In turn, the cells and neuronal processes are embedded in a relatively thin layer of collagen fibrils that are organized in small bundles (FIGURE 11) (234). Below the cellular layer of the lamina propria is the deeper fibrous layer. This region is mostly fibrillar, with only a few scattered interstitial cells. Components of the matrix include large bundles of collagen fibrils, which form collagen fibers.

As described in the previous section, the muscularis mucosae is a thin often discontinuous band of smooth muscle that is present in humans and guinea pigs (FIGURE 11). Its exact function remains unclear, although as noted above it could play a role in refolding the mucosa after voiding. Other studies propose that spontaneous phasic contractions in mucosal tissue strips arise in the muscularis mucosae (292), and its close proximity to the suburothelial blood vessels and nerve fibers may implicate this structure in modulating blood flow and facilitating afferent firing (413). Finally, its nearness to the urothelium makes it a potential target of urothelial-released mediators.

c) nerve processes innervating the lamina propria and urothelium.

Normal bladder function (e.g., urine storage and micturition) and responses to bladder inflammation and injury depend on sensory inputs from afferent nerve fibers located near to and within the bladder mucosa and the underlying musculature. Afferent fibers innervating the urinary bladder are carried by the hypogastric and pelvic nerves, whereas those innervating the urethra are carried by the pudendal nerve. In mice and rats, the cell bodies (somas) of sensory neurons in the pudendal and pelvic nerves reside in dorsal root ganglia (DRG) located at lumbosacral (L6-S2) level, while those in the hypogastric nerve reside in DRG at the thoracolumbar (T13-L2) level. Afferent fibers innervating the lower urinary tract do not form specialized “nerve terminals” as such, but instead exist as free nerve endings or termini. Gabella (235) recently published an elegant TEM analysis describing nerve termini in close apposition to detrusor smooth muscle cells. How these free nerve endings sense the filling status of the bladder still remains an open question.

There are two types of afferent fibers found in the bladder wall: Aδ and C. Thinly myelinated Aδ fibers conduct action potentials in a rapid manner (conduction speed 5–40 m/s). They are primarily found in the detrusor muscle, where they are intercalated within the myofibers (582). Aδ fibers express NEFH, the heavy chain component (200 kDa) of heterotrimeric neurofilament bundles. In contrast, nonmyelinated C fibers are relatively slow conducting (conduction speed 0.5–2 m/s), and their axons are smaller than those of Aδ fibers (the diameter of Aδ fibers is 1–5 μm, while that of C fibers 0.2–1.5 μm) (233, 236). Well-characterized markers of peptidergic C fibers include the substance P peptide (derived from cleavage of TAC1) and CALCA (calcitonin gene-related peptide) (582). When stained, afferent fibers have varicosities along their entire length (233, 236). While the function of these varicosities is unknown, some may be nerve branch points. Gabella and Davis (236) first, and later Gabella (233) demonstrated that CALCA⁺ C fibers are generally absent from the suburothelial lamina propria in the dome region of the rat bladder, but are found in greater abundance in the suburothelium of the caudal regions of the bladder (i.e., equatorial region and neck, especially near the trigone). In these regions, the C fibers form a flat plexus. In contrast, Schueth et al. (627) used two-photon microscopy and 3D tracking softwares to show that CALCA⁺ C fibers also form a suburothelial plexus in the dome region of the mouse bladder. Whether this reflects species differences or other parameters such as antibody accessibility, or the type of imaging analysis performed, is unknown. C fibers are also found beneath the urothelium lining the ureters and urethra (233). Moreover, in the more caudal regions of the bladder, intraepithelial nerve fibers are visible that penetrate directly into the bladder urothelium (e.g., see FIGURE 12B, inset), but this is not observed in those regions of the ureters or urethra in close proximity to the bladder (233). Both, Gabella (233) and Rahnama’i et al. (582) also report that C fibers are found deeper into the bladder wall, forming interactions with the vasculature and with the smooth muscle.

The current consensus is that Aδ fiber afferents respond to bladder distension in the physiological range and therefore participate in normal bladder sensation (168, 217). The contribution of C fibers to normal bladder sensation is less clear. Initial reports indicated that bladder C fibers have high thresholds and respond to bladder distension only at elevated pressure (48, 274, 332). However, this early understanding of C fiber function was later challenged by reports describing a subpopulation of bladder C fiber afferents that respond to bladder distension in the physiological range of pressures (47, 182, 215, 632, 756). In addition to these mechanical stimuli, C afferent fibers are also implicated in bladder nociception and are stimulated by noxious stimuli such as intravesicle acetic acid, hydrochloric acid, mustard oil, acetone, or systemic cyclophosphamide (72, 73, 159, 160, 162, 168, 170, 594, 610, 686, 735, 774, 775). However, Montalbetti and co-workers (504, 505) recently reported that the pelvic allodynia and bladder hyperactivity seen in a rat model of IC/BPS (triggered by altering the permeability of the urothelial tight junctions) are driven by sensitized bladder afferent fibers of Aδ origin. In sum, further studies are needed to understand the role of Aδ and C fibers in lower urinary tract function and nociception.

Communication between the urothelium and adjacent afferent termini is proposed to contribute to normal bladder function (33, 75). While this hypothesis remains a matter of conjecture, a number of potential mediators (e.g., ATP and acetylcholine) are known to be released from the urothelium in response to distension, and its respective receptors are reported to be expressed in bladder afferents. For instance, afferents express the TRP family channels TRPV1, TRPA1, TRPM8; the nerve growth factor receptor NTRK1 (TrkA receptor); the purinergic receptors P2RX2, P2RX3, P2RY1, P2RY2, and P2Y4; the nicotinic and muscarinic acetylcholine receptors; acid-sensitive ion channels (502); estrogen receptors; and endothelin receptors (reviewed in Refs. 168, 483). In addition to sensory pathways, afferents can release a number of secretory products by way of clear vesicles or dense-core ones (FIGURE 12B) including substance P, CALC1, leucine enkephalin (a peptide derived from PENK), corticotropin releasing hormone (CRH), and vasoactive intestinal peptide (VIP) (reviewed in Refs. 168, 483). Thus, in addition to sensing their environment, afferents can also release peptides and other transmitters that can modify the activity of cells in their vicinity. A likely target is the urothelium, which expresses receptors for several of these mediators, indicating they might be important input pathways in the urothelium (see discussion below).

d) interstitial cells.

The final cellular component of the lamina propria we discuss are the so-called interstitial cells, a histological term used to describe those cells that reside in the stroma or connective tissue spaces between the “functional cells” of a tissue. Examples include the interstitial cells of Cajal (ICC) found in the gut, neuroglial cells in the nervous system, and Leydig cells in the testes. Unfortunately, there still remains no consensus on what urinary tract-associated interstitial cells should be called, and various names employed in the literature to date include fibroblasts, fibroblast-like cells, ICC-like cells, myofibroblasts, and telocytes (370). Regardless of name, when examined in cross section these cells are characteristically flat and spindled shape, they have a paucity of cytoplasm, they have a discontinuous basement membrane or lack one altogether, and they appear to have very long protrusions that can extend hundreds of microns (FIGURE 13A) (164, 244, 587). They were originally identified in the bladder as targets of NO using cGMP immunohistochemistry (648), but were later found throughout the lower urinary tract (244, 315, 477, 633).

FIGURE 13.

Suburothelial interstitial cells. A: transmission electron microscopy (TEM) of mouse urothelium and subjacent interstitial cells. (TEM image provided by Steven Truschel.) B: localization of platelet-derived growth factor receptor-α (PDGFRA), actin, and nuclei (Nuc) in the mouse mucosa. PDGFRA⁺ interstitial cells reside below the urothelium (shaded blue in the left-most panel).

Interstitial cells are classified according to their location, including populations that are found in the lamina propria (and are therefore suburothelial), and a population that are intermuscular. Neuhaus et al. (533) published a 3D EM reconstruction of human lamina propria-associated interstitial cells. They identified two major subgroups: those that were fibroblast like and those that were myoid like. A third, branched-type cell type was rarer. When examined in 3D, the majority of the fibroblast-like and myoid-like interstitial cells are not spindle shaped, but instead are flat and sheetlike, and always oriented parallel to the long axis of the urothelium. If this is true of the interstitial cells in other species will have to await a similar analysis. The fibroblast-like interstitial cells form homocytic (same cell) gap junctions, but not heterocytic ones (between adjacent cells) (533). In contrast, myoid-like interstitial cells form large numbers of heterocytic junctions. The presence of these gap junctions fits with previous reports of GJA1 expression in human interstitial cells (673), and studies in rodents, which demonstrate interstitial cells express GJA1 and are likely electrically coupled to one another, forming a network (323). Strikingly, Yu et al. (786) just published a remarkable single-cell gene analysis of human and mouse bladder-derived cells that predicts there are up to five different classes of interstitial cells. If true, then there is much more to learn about the different interstitial cell types located in the bladder wall.

Even though lower urinary tract interstitial cells are found in close proximity to the urothelium, blood vessels, afferent nerve processes, and the detrusor, there are no specialized contacts with these tissues (i.e., there are no synapses or gap junctions) (235, 393, 587, 673, 759). For example, Gabella (235) readily observed nerve termini associated with rat detrusor smooth muscle cells, but he could find no nerve termini associated with interstitial cells. This is different from the gut, where gap junctions are observed between ICC and smooth muscle cells (371). There is controversy about the cellular markers expressed by interstitial cells, and in particular whether they express KIT (c-Kit proto-oncogene), the traditional marker used to identify ICCs in the gut (370). However, there is a growing consensus that populations of interstitial cells express PDGFRA (platelet-derived growth factor receptor-α) (e.g., see FIGURE 13B), along with other markers including ANO1 (anoctamin 1), CAV1/2, CD34, cGMP, and VIM (vimentin) to name a few (64, 245, 369, 785). As noted above, because they are not homogeneous in their expression of markers and genes (245, 786), it is likely there are multiple subtypes of interstitial cells in the lower urinary tract.

Despite their abundance, particularly in the lamina propria, there are limited insights into interstitial cell function within the urinary tract. Their proximity to the urothelium and detrusor would seem to indicate an important role in modulating or bridging communication between interstitial cells and other tissues in the bladder wall including the urothelium (370, 476). Because only a small number of bladder interstitial cells generate spontaneous Ca²⁺ transients in situ, and because these Ca²⁺ transients do not coincide with those of smooth muscle cells, some have argued that it is unlikely bladder interstitial cells have pacemaker function (286). Others have argued that in response to purines and stretch, the role of intermuscular interstitial cells is to restrain or dampen the detrusor activity by altering smooth muscle cell membrane potential (408–411). However, such a function would likely depend on release of mediators or direct electrical coupling between interstitial cells and detrusor smooth muscle cells, which as noted above does not appear to occur in the urinary tract. Others have proposed that interstitial cells serve to amplify signals generated by the urothelium (222), or that interstitial cells work in conjunction with nerves to form a suburothelial stretch sensor (759). Consistent with this latter possibility, interstitial cells express PIEZO1 (156, 442), which as noted above is a stretch-sensitive, nonselective cation channel. Despite these hypotheses, at present there are no clear insights into what interstitial cells are doing in the lower urinary tract.

In terms of sensory pathways, interstitial cells in the bladder were first identified as generating cGMP in response to sodium nitroprusside, an NO donor (648). Thus they are likely responsive to NO production. They are also reported to be sensitive to muscarinic agonists and express CHRM2 (muscarinic acetylcholine receptor M2) and CHRM3 (264, 362, 427, 514). They also express P2RX3, P2RY1, P2RY2, P2RY4, P2RY6, and TRPV4 (408, 410, 411, 674, 763). A subset of intermuscular ICs express ENTPD2 (ectonucleoside triphosphate diphosphohydrolase 2) (783), an ectonuclease that hydrolyzes ATP to AMP, indicating that IC cells can modulate their responses to ATP. Interstitial cells also express prostaglandin receptors PTGER1 (prostaglandin EP1 receptor) and PTGER2 (EP2 receptor), and they express PTGS1 (COX1) (581, 583). Thus interstitial cells may respond to prostaglandins and presumably they can generate them. In terms of mediator release, there is limited information. Other than possibly producing prostaglandins, it is unclear how or if interstitial cells communicate with the urothelium and other tissues in the lamina propria or muscularis mucosa/externa.

3. Tissues deep to the mucosa

Deep to the mucosa is the submucosa (only present in those species with a muscularis mucosae), and the muscularis externa, which comprises three layers of smooth muscle: inner and outer longitudinal and middle circular layers (FIGURE 11). In the bladder, this smooth muscle layer is often referred to as the detrusor. The outer surface of the lower urinary tract is lined by a layer of serosa or adventitia depending on species and organ. For example, in rodents, the dome and equatorial regions of the bladder are lined by a serosa, which comprises small amounts of connective tissue and the mesothelium (a lining epithelium) that forms the peritoneum. The retroperitoneal surface of the bladder including the trigone is surrounded by adventitia, which blends into the surrounding connective tissues. In humans, only the superior surface of the bladder is lined by a serosa, and all other surfaces are surrounded by an adventitia. Other than serving as an interface between the bladder wall and the peritoneum, there is little literature about the role of the mesothelium in bladder function. Ureters and urethra are also surrounded by an adventitia, whereas the renal pelvis borders epithelial and fat tissues (which are near the hilus).

B. Urothelial Sensor Function

A major tenet of the urothelial sensory web is that the urothelium is primed to detect changes in its milieu, both at its mucosal and serosal surfaces, by expressing a myriad of receptor and channels (FIGURE 14). While some of these input pathways, particularly those involved in sensing mechanical stretch, are likely to function during normal lower urinary tract function, urothelial damage or mucosal inflammation is likely to lead to overexpression or underexpression of specific receptors (or changes in release of mediators), which may then contribute to symptoms of urgency and diseases such as overactive bladder. Those readers interested in the disease aspects of urothelial sensory function are directed to the following excellent reviews on this subject (71, 166, 217). In this section we will do the following: 1) review the data supporting urothelial expression and localization of receptors/channels in both animal models (typically rodents) and in human tissue samples. As in previous sections, this information is weighted heavily to bladder urothelium, as there is much less information available about these pathways in the urothelium lining the other organs in the lower urinary tract. We will 2) review the literature implicating the designated receptors and channels in lower urinary tract function and dysfunction, with an emphasis on gene-null rodent models and human studies when available; and 3) provide summaries of these data in TABLE 4.

FIGURE 14.

Summary of urothelial input and output pathways. Mediators released by the urothelium (i.e., “outputs”) are indicated in blue text, and hypothesized target tissues for these outputs are indicated by blue arrows. Mediators present in the urinary space, or those released from afferent nerve termini, are marked in green text and serve as inputs, triggering urothelial responses. Putative input pathways present in the urothelium are indicated in green text and include numerous channels and receptors. There is limited understanding of the mediators released by interstitial cells or smooth muscle cells that impinge on urothelial function, thus the question marks. The purple arrow is a hypothetical mechanism whereby mediators released from one urothelial cell type can act as inputs in a different urothelial cell type.

Table 4.

Receptors and channels implicated in urinary tract function/dysfunction

*Channel/Receptor	Localization in Animal Tissues	Rodent Model	Phenotype	Human Tissue Expression and Relation to Disease
TRP channels
TRPA1	mRNA and protein expression in rat urothelium/mucosa and BSM (185, 666). No evidence for functional expression in mouse urothelial cells (206). Protein expressed in unmyelinated [afferent] nerve fibers innervating the urothelium, suburothelium, detrusor, and blood vessels of the rat (666), and nociceptive nerve fibers innervating the mouse urothelium (521). mRNA and protein in L6-S1 DRG in rat; ~50% of DRG neurons innervating the bladder are positive (185).	Global null mouse	No differences in cystometric parameters for urethane-treated (716) or decerebrated mice (346). No changes in bladder response to acute, cold-induced, urinary urgency (716). Prevents changes in VF, VV, and painlike behavior induced by intravesical infusion of LPS (346).	mRNA expressed in bladder mucosa and muscle. Protein expression confirmed in the mucosa (183). Protein expressed in basal cells of urothelium lining urethra and suburothelial nerves (262). Expression increased (~2-fold) in mucosa of patients with BOO (183). mRNA expression increased in bladder of patients with IC/BPS (301).
TRPC1 and TRPC4	mRNA and protein expression for both channels in mouse urothelium (780). In mouse and rat, TRPC1 protein is expressed in UC cytoplasm. TRPC4 is found along the BL membrane of UC, the surface of IC and BC, nerve fibers, and BSM (780). mRNA and protein expressed in rat L1-S6 DRGs (94).	Global double null mouse	No difference in cystometric parameters for restrained, conscious mice (94). In response to cyclophosphamide-induced cystitis, the KO mice have ↓VF, but no increase in sensory innervation (which is observed in wild-type mice) (94).
TRPM7	mRNA and/or protein expressed in mouse and rat urothelium, suburothelium, and BSM (753, 780). Evidence for functional expression in mouse urothelial cells (206).	Generated inducible urothelium-specific KO mouse (753).	KO mice exhibit ↓VV (metabolic cage), immature/abnormal intercellular junctions, bladder inflammation with edema in the submucosal layer, and increased mRNA production for the inflammatory cytokines TNF-α and IL-1β (753).	mRNA detected in human ureter and bladder urothelium (301, 635). No change in expression observed in patients with classical or nonclassical IC/BPS (301).
TRPM8	mRNA expressed in rat bladder; protein expressed in rat urothelium (345, 662). No evidence for functional expression in mouse urothelial cells (206). mRNA and protein expressed in bladder urothelium, suburothelial nerve fibers, BSM, and L6 DRG in guinea pig bladder (242). Protein expressed in bladder afferent L6-S1 DRGs in rats (289, 638, 662).	Global null mouse	No difference in cystometric parameters (urethane-treated mice) (716). Reduced bladder response to acute cold-induced urinary urgency (716).	mRNA expression is observed in human bladder (301), including the mucosa/urothelium but not in the BSM (183, 662). Other studies failed to detect expression in the urothelium (635). Protein expression detected in the urothelium and nerve fibers in the suburothelium (513, 662). mRNA levels are increased in the bladder of patients with IC/BPS (301). Increased number of TRPM8-positive nerve fibers in patients with IDO and IC/BPS (513). Proposed molecular basis for the diagnostic ice water test and the bladder cooling reflex (513, 662).
TRPV1	mRNA and protein expressed in the urothelium of rat bladder (76). Other studies report no expression in the urothelium, but in sensory afferent nerves (118, 700, 780). No evidence for functional expression in mouse urothelial cells (206).	Global null mouse	KO mice have ↑number of small-diameter spots (void-spot assay) (77, 750). In cystometry, awake KO mice exhibit ↑NVC with no differences in VF and PP, while urethane-treated animals exhibit ↑NVC, ↑BlC, and overflow incontinence (77). KOs exhibit ↓ATP release from the urothelium upon stretch and ↓apical exocytosis, compared with controls (77). KOs exhibit ↓response of low-threshold afferents to bladder distension upon high rates of bladder filling (157). Decerebrated KO mice exhibit ↑NVC in cystometry (776); conflicting results in urethane-treated mice: no NVC in KO mice (125), whereas others report ↑NVC (750). Using metabolic cages, KO mice exhibited ↓VV with no difference in VF compared with controls (776).	TRPV1 mRNA expressed in bladder urothelium (635, 662), and protein expressed in the bladder urothelium, BSM, nerve fibers, and interstitial cells (37, 405, 553, 772). mRNA expression increased in bladder of patients with classic IC/BPS (301). Patients with NDO exhibit increased protein expression in bladder urothelium and increased numbers of TRPV1+ nerve fibers (37, 38, 96). The levels of TRPV1 expression in the trigonal mucosa are positively correlated with the symptoms in patients with SU (440). Treatment with capsaicin or resiniferatoxin reduces OAB by desensitizing bladder afferents (153, 453, 454), and decreasing urothelial TRPV1 expression (37, 439). Botulinum toxin treatment decreases TRPV1 expression in suburothelial sensory fibers (38).
TRPV2	mRNA and protein expression in UC, BSM, and sparse nerve fibers of mouse and rat bladders (76, 780). Evidence for functional expression in mouse urothelial cells (206).	Global null mouse	No bladder studies published to date.	mRNA detected in human bladder urothelium (635).
TRPV4	Protein expressed in urothelium of mouse and rat bladders (67, 246, 780). Localized to basolateral surface of UC, plasma membrane of IC and BC (780). Expressed in endothelial cells (246). mRNA and protein expressed in mouse BSM (696). Evidence for functional expression in mouse urothelial cells (206).	Global null mouse	↑Number of total voiding spots and ↑number of voiding spots far from cage corners (246). In awake, restrained KO mice, cystometry reveals ↓VF, ↑VV and ↑NVC (246); while in decerebrated KO mice, the only change observed is ↑NVC (776). In conscious, unrestrained KO mice, cystometry reveals ↑BC and ↑VV but no difference in NVC (696). In metabolic cage studies, KO mice showed ↑VF, ↓VV, and ↑VVT compared with control (776). ↓ATP release from mucosal side of the urothelium upon stretch (246). Isolated urothelial cells from KO mice exhibit ↓intracellular calcium increase and ↓ATP release upon stretch (498). Tight junctions and desmosomes look normal, but adherence junctions are reportedly absent in KO mice. UCs have larger intracellular vesicles and exhibit abnormal vesicular exocytosis (334). KO mice exhibit attenuated symptoms to cyclophosphamide-induced cystitis (207).	mRNA and/or protein detected in human ureter and bladder urothelium (333, 334, 635). TRPV4 is localized to the urothelial and suburothelial layers (596). In UCs, TRPV4 is localized to the adherence junctions (333, 334). mRNA expression is reduced in the bladder of patients with classic IC/BPS (301). 9 out of 17 patients with gain-of-function mutations in TRPV4 (Charcot-Marie-Tooth disease type 2C) complain of bladder urgency and incontinence (399). TRPV4-mediated ATP release from the mucosa (both basal and agonist-induced) is higher in OAB patients (596).
		Global null rat	No differences in normal bladder function between WT and TRPV4−/− rats (cystometry, awake) (175).
TRPV6A				mRNA detected in human ureter urothelial tissue (635).
Purinergic receptors
ADORA2B	mRNA expressed in rat bladder (178, 661). Protein is expressed in rat and mouse urothelium: localized to cytoplasm and BL membrane of UC, cytoplasm of IC and BC; also expressed by BSM (784)	Global null mouse	KO mice have smaller bladders (281). KO mice exhibit ↑number of voiding spots, but ↓voiding spot size (281). In cystometry, urethane-treated KO mice exhibit ↑VF, ↓BC, and ↓PP (281). ADORA2B hypothesized to be major adenosine receptor mediating BSM purinergic contraction: weaker BSM contractions in tissue from KO (281).	mRNA and protein expressed in BSM (556, 645). Functional studies using human BSM strips indicate that ADORA2B has a modest role in the contractile response to adenosine (556), while others argue its primary function is to counter the stimulatory effects of ATP (281).
P2RX1	P2RX1 mRNA and protein expression is restricted to the BSM of mouse and rat bladder: no urothelial expression (145, 717, 732, 739). Protein expressed in urothelium and BSM of cat bladder (79).	Global null mouse	ATP-mediated inward currents (patch-clamp studies on isolated BSM cells) and contractions (bladder strips, urothelium-free) abolished in BSM from KO mice (732). Loss of P2RX1 eliminates action potentials and Ca2+ flashes induced by electrical field stimulation of bladder strips (293).	mRNA expressed in human bladder is the predominant subtype of P2RX receptor present (546, 547, 718); the protein is reportedly expressed in the BSM but absent from the urothelium (197, 547). Protein is expressed in nerve fibers of the bladder wall (507, 588). mRNA and protein expression are reportedly increased in the BSM of patients with BOO (545, 644), but other studies report no difference (138). Age-related decrease in mRNA levels reported in men lacking BOO, but not in those with BOO (138). Almost complete loss of P2RX1 expression in bladder nerve varicosities from patients with SU (588).
P2RX2	Protein expressed in urothelium, suburothelial plexus, BSM, and serosa in mouse and rat bladders (668, 732). Protein expressed in urothelium and BSM of cat bladder (79). mRNA expressed in mouse TL and LS bladder neurons (127).	Global null mouse	Bladder afferents of P2rx2 KO mice exhibit attenuated response to bladder distension; ↑VT for nerve activation (141). KO mice exhibit altered bladder activity (↓VF and ↑VT) in cystometric analysis (urethane anesthetized). No difference in BSM (muscle strips) response to contraction elicited by addition of neurotransmitters (141). Apical exocytosis impaired in KO mice bladders (747).	mRNA and protein expression observed in bladder mucosa (urothelium and suburothelial interstitial cells), blood vessels, BSM, intermuscular interstitial cells, and nerve fibers in human bladder (481, 507, 547, 588, 643, 681, 690). Expression is lost in the urothelium of patients with BOO (643), while increased in the mucosa of patients with IC/BPS (441, 690). It is also increased in the BSM of patients with IDO (547), and in the interstitial cells of women with OAB (481), but almost completely lost in suburothelial and detrusor nerve fibers of patients with SU (588). Prejunctional P2RX2/3 receptors are involved in the increased cholinergic nerve terminal tonus in BOO patients (644).
P2RX3	Protein expressed in urothelium, suburothelial afferents, BSM, and serosa of rat bladder (197, 668). Protein expressed in urothelium and nerves in the cat bladder (79). Protein expressed in the nerve fibers of the mouse bladder (142). mRNA expressed in mouse TL and LS bladder neurons (127) and in fibers innervating the bladder by way of L6-S1 DRGs (796).	Global null mouse	In conscious KO mice, cystometry reveals ↓VF, ↑BC and ↑VT (142). Bladder afferents (pelvic nerves) from KO mice exhibit an attenuated response to bladder distension (736). The VT for whole-nerve afferent activation is increased in KO mice. No difference in BSM (muscle strips) response to contraction elicited by addition of neurotransmitters (141). Apical exocytosis impaired in KO mice bladders (747).	mRNA and protein expression reported in the urothelium (197, 643, 681, 690), with little or no protein present in the suburothelium, and none in the BSM (197, 681). Protein expression is observed in suburothelial and BSM nerve fibers (95, 507, 547, 588, 643, 772), and lamina propria-associated and intermuscular interstitial cells (681). Expression is lost in the urothelium of patients with BOO (643), while increased in the mucosa of patients with IC/BPS (441, 690). P2RX3-positive nerve fibers are increased in patients with NDO and are reduced in those patients that respond to resiniferatoxin treatment (95). Decreased P2RX3 expression in the detrusor-localized nerve fibers of patients with IDO or SU (507, 588). Prejunctional P2RX2/3 receptors promote increased cholinergic nerve terminal tonus in BOO patients (644).
P2RX4	mRNA expressed in rat bladder (84). Protein expressed in the urothelium of cat bladder (79). Protein not expressed in the urothelium or BSM of mouse, but is expressed in cellular structures near BSM bundles, likely nerve fibers (781).	Global null mouse	No phenotype: KO mice exhibit normal voiding function (by void-spot assay and cystometry in urethane-anesthetized animals) and BSM contractility (measured using myography) (781).	mRNA detected in human ureter and bladder urothelial tissue (635). Protein expressed in the BSM (547), and in a small fraction of nerve fibers (507, 588). mRNA expression is decreased in the bladder of patients with IDO (547), and protein expression is decreased in bladder nerve varicosities of SU patients (588).
P2RX7	Protein expressed in lamina propria of mouse bladder, possibly in immune cells (466). Protein expressed in urothelium of cat bladder (79).	Global null mouse	KO mice exhibit diminished nociceptive and inflammatory responses to cyclophosphamide-induced cystitis relative to controls (466).	Protein expressed in UC of the urothelium and BSM cells, but absent in interstitial cells (681). It is also present in a small fraction of bladder nerve varicosities (507, 588). Other studies failed to detect protein expression in human bladder (547). mRNA decreased in bladder of patients with IDO (547).
P2RY1	Protein expressed in urothelium of cat bladder (79). mRNA expressed in rat BSM and blood vessels (548). mRNA expressed in mouse intermuscular interstitial cells (410).	Global null mouse	No studies performed in bladder. ATP-activated SK currents are mostly abolished in PDGFRA+ cells from KO mice (410).	mRNA expression detected in human ureter and bladder urothelium (635).
P2RY2	Protein expressed in urothelium, suburothelial nerve bundles, and BSM in rat bladder (136). Protein expressed in urothelium of cat bladder (79). mRNA expressed in mouse intermuscular interstitial cells (410, 411) and in ~50% of the mouse bladder neurons located in LS and TL DRGs (128). Evidence for functional expression in rat urothelial cells (136).	Global null mouse	No bladder phenotype reported. Neurons from KO mice lose the facilitatory effect of UTP on sustained currents (128). KO mice exhibit altered water and sodium reabsorption resulting in more concentrated urine (794).	mRNA expression detected in human bladder urothelium (635).
P2RY6	mRNA expressed in urothelium, suburothelium, and detrusor of mouse bladder (363). Protein expressed in urothelium, suburothelial interstitial cells of rat bladder (115, 698). Protein expressed in sparse punctate pattern in BSM (224). Protein expressed in guinea pig urothelium and suburothelial interstitial cells (674).	Global null mouse	KO mice exhibit ↑VF and ↓VV in metabolic cages, and in cystometric analyses, decerebrated KO animals exhibit ↓bladder contraction duration, ↓VV, ↓VT, ↓VE, ↑RP and ↓BlC (363). BSM (detrusor strips) exhibit no changes in detrusor contractility (363). No changes in ATP release by the urothelium in response to stretch (363).	Protein reportedly expressed in the urothelium (643), but other studies failed to detect mRNA expression in the urothelium (635). Apparently upregulated in the lamina propria-associated interstitial cells of patients with BOO (643). Activation facilitates ATP and ACh release from the mucosa of human bladders (643).
Cholinergic receptors
CHRM1	mRNA is expressed in mouse urothelium. Protein restricted to BC (789).	Global null mouse	No bladder phenotype: cystometry performed in awake KO mice, but no differences relative to controls (172).	mRNA and/or protein expressed in urothelium, suburothelium, and in the BSM of human bladders (101, 161, 457, 711). Protein expression is decreased in the urothelium and suburothelium of patients with DO, but increases upon botulinum toxin treatment (161). Suburothelial levels of the receptor inversely correlate with VF (161). CHRM1 triggers carbachol-induced H2S production in the urothelium (155).
CHRM2	mRNA is expressed in mouse urothelium and CHRM2 protein is almost exclusively localized to UC layer (789). Expressed in the BSM of rat, rabbit, mouse, and guinea pig (195). mRNA and protein expressed in mouse bladder afferent nerves (528).	Global null mouse	BSM contractility is slightly decreased, being ~95% of that observed in WT animals (196, 469). Cystometry in awake animals reveals ↓VF and ↑VV in female KO mice compared with controls, while male KO mice do not exhibit any changes in CMG parameters (316). Cystometry in awake mice reveals ↓TP (172).	mRNA and/or protein expressed in mucosa and BSM (101, 161, 457, 711). In urothelium, almost exclusively expressed in the UC layer (101); other studies failed to show protein expression in the urothelium (161). Protein expression also reported in bladder nerve fibers and in interstitial cells (514). Increased protein expression in suburothelial interstitial cells in patients with IC/BPS and IDO (514). Expression levels correlate with urgency and frequency scores (514). Increased protein expression in BSM of patients with IC/BPS-protein internalization and recycling are affected (61, 534). Other studies report no changes in mRNA levels of patients with IDO (458), or protein expression in the suburothelium of patients with DO (161). Increased protein expression in the urothelium of human donors with bladders that exhibit an M2-mediated contractile component (97). Expression levels of mRNA and protein are affected by age and gender (41).
CHRM3	mRNA and protein expressed in mouse urothelium (789). mRNA and/or protein expressed in the BSM of rat, rabbit, mouse, and guinea pig (195); in mouse bladder afferent nerves (528); and the lamina propria-associated interstitial cells of the guinea pig bladder (263).	Global null mouse	Male KO mice present with severely distended bladders, urinary retention, and in some cases mild hydronephrosis. Focal lymphocyte infiltration in the lamina propria, but the urothelium appears normal (470). BSM contractility decreased by 95% relative to WT mice (196, 470) with the residual 5% contraction probably due to CHRM2 (196, 469). Awake KO mice exhibit ↓VF, ↑VV, and ↑BlC in cystometry (316), while a different study also performed in awake KO mice showed a ↓ in ICI, but no significant changes in VF (172).	mRNA and protein expressed in mucosa, BSM, bladder nerve fibers, and in interstitial cells (101, 161, 457, 514, 711). Decrease in mRNA levels with age in the BSM of both male and female subjects (457). Patients with homozygous loss-of-function mutations present with bladder malformations, detrusor hyporeflexia, distended bladder with high residual volumes after micturition, variable vesicoureteric reflux, and hydronephrosis (56, 754). Decrease in mRNA in the mucosa of IDO female patients (458). Increase in protein expression in lamina propria-associated interstitial cells in patients with IDO; expression levels correlate with urgency score (514). Decreased protein expression in the suburothelium of patients with DO; immunoreactivity levels inversely correlated with number of urgency episodes and VF (161). Increased protein expression in the BSM of patients with IC/BPS, with altered protein internalization and recycling (61). CHRM3 mediates carbachol-induced H2S production in the urothelium (155).
CHRM4	mRNA and protein expressed in mouse urothelium (789), and mRNA is expressed in mouse bladder afferent nerves (528).	Global null mouse	No bladder phenotype: cystometry performed in awake KO mice reveals no differences compared with WT controls (172). No changes in BSM contractility (bladder strips) is observed in KO mice (663).	mRNA expression in mucosa and BSM reported (155, 711). Other studies did not observe expression (457), or detection was inconsistent (<40% of the samples) (101). Protein detected in the mucosa, including all three layers of the urothelium (101).
CHRM5	mRNA and protein expressed in mouse urothelium (789).	Global null mouse	Awake KO mice exhibit ↓VV, ↓BC, ↑VF in cystometry (172).	mRNA expressed in mucosa and BSM (101, 457, 711). Protein detected in all three layers of the urothelium (101).
CHRNA3	mRNA and protein expressed in rat urothelium (58, 60). mRNA absent in mouse urothelium (789), but reported in mouse bladder afferent neurons (529).	Global null mouse	KO mice exhibit extreme bladder enlargement (megacystis), bladder infection and inflammation, necrotic and/or dysplastic bladder mucosa, and urinary stones (765). KO mice show dribbling urination. Bladder strips from KO mice do not contract in response to nicotine; however, no defect in response to electrical field stimulation. May indicate that bladder impairment is at the level of innervation (765).
CHRNB2 and CHRNB4	CHRNB4, but not CHRNB2 mRNA expressed in rat urothelium (60). CHRNB2 and CHRNB4 mRNA expressed in mouse bladder afferent neurons (529).	Global null mouse	Chrnb2 KO mouse exhibit focal thickening of bladder mucosa. Detrusor strips have normal contractile response to nicotine (766). Chrnb4 KO mouse exhibit focal mucosal hyperplasia and increased mitotic activity. Bladder strips show limited responses to nicotine (766). Chrnb2;Chrnb4 double KO mice have abnormal mucosa with increased thickness, enlarged bladders with dribbling urination. Develop urinary infections and bladder stones. Bladder strips do not respond to nicotine. Mice exhibit early death (90% die within 10 days of birth) (766).

^*All targeted genes are reported to be expressed in the urothelium. BC, basal cell; BCL, bladder compliance; BlC, bladder capacity; BL, basolateral; BOO, bladder outlet obstruction; BSM, bladder smooth muscle or detrusor; DO, detrusor overactivity; DRG, dorsal root ganglia; IC/BPS, interstitial cystitis/bladder pain syndrome; IC, intermediate cell; ICI, intercontraction interval; IDO, idiopathic detrusor overactivity; IL, interleukin; LS, lumbosacral; NDO, neurogenic detrusor overactivity NVC, non-voiding contractions; OAB, overactive bladder; PP, peak (micturition) pressure; RV, post-void residual volume; SK channels, small-conductance Ca²⁺-activated K⁺ channel; SU, sensory urgency; TL, thoracolumbar; TNF, tumor necrosis factor; TP, threshold pressure; VE, voiding efficiency; VF, voiding frequency; VT, volume threshold; VV, voided volume (per void); VVT, voided volume total; UC, umbrella cell; ↓ decreased; ↑ increased.

1. Stretch-activated and TRP family channels

a) piezo channels.

Because the function of the entire lower urinary tract is governed to a large extent by its mechanical environment, it should not be surprising that the urothelium expresses stretch-sensitive channels. One of the most intriguing is PIEZO1, discovered by Dr. Ardem Patapoutian and colleagues using expression cloning (147). They identified two novel multi-pass transmembrane ion channels, PIEZO1 (2,521 amino acids in humans) and PIEZO2 (2,752 amino acids in humans), which form a small family of channels that are directly gated by mechanical stimuli (e.g., cell probing, hydrostatic pressure, or stretch) (147, 148, 391), and regulate a diversity of stretch-sensitive processes in the body (517). A recent analysis by Dalghi et al. (156) used Piezo1^tdT reporter mice to show that the urothelium lining the renal pelvis, ureters, bladder, and urethra expresses PIEZO1-tdTomato, as do interstitial cells and smooth muscle cells. This indicates that shared pathways for mechanosensation are likely to exist for all of these organs. In mouse urothelium, PIEZO1-tdTomato is expressed on the basolateral surfaces of the umbrella cell layer (and plasma membranes of intermediate and basal cells), indicating that PIEZO1 is likely functional as the urothelium responds to increased wall tension during filling (and upon voiding as the rougae refold). At least in isolated urothelial cells, PIEZO1 was recently shown by Miyamoto et al. to be required for stretch-induced ATP release (494), an important output pathway present in the urothelium. If PIEZO1 has a similar function in the native bladder and if it is responsible for triggering the release of other mediators from the urothelium is unknown. Intriguingly, there are recent studies that report increased PIEZO1 expression in experimental models of cyclophosphamide-induced cystitis and partial bladder outlet obstruction (442, 486), but what PIEZO1 is doing in these settings is unknown. However, one might posit that if excess PIEZO1 expression sensitized the urothelium to stretch, then it could contribute to the bladder overactivity associated with these experimental models. However, this has not been explicitly tested experimentally.

There is much less known about Piezo2 expression in the lower urinary tract, but one report indicates expression of a splice variant in the mouse bladder (684), whereas a different study indicates that Piezo2 may not be expressed in the mouse urothelium (494). Interestingly, patients with arthrogryposis multiplex congenita, which includes patients with sporadic or autosomal dominant forms of arthrogryposis, suffer from bladder dysfunction (39). Because mutations in PIEZO2 are one cause of autosomal dominant forms of arthrogryposis (15, 517), and because PIEZO2 is expressed in upwards of 50% of dorsal root ganglia cells (542), mutations in PIEZO2 could impact lower urinary tract function by altering the activity of PIEZO2⁺-afferent nerve fibers.

b) ENaC and other stretch-sensitive channels.

Other stretch-responsive channels expressed by the urothelium include ENaC and stretch-activated K⁺ channels. In addition to potential functions including Na⁺ absorption, stretch-dependent ATP release, and membrane dynamics (213, 214, 423, 424, 782), inhibiting ENaC is also proposed to impair bladder function (184). However, as noted above, the use of 1 mM amiloride in the latter studies (far above the reported IC₅₀ of ~0.1 µM for ENaC) makes it likely that channels other than ENaC are also affected. Despite the availability of ENaC subunit knockout mice (53, 314, 479), we are unaware of any studies that have tested these animals for a lower urinary tract phenotype.

There are also reports that message for stretch-sensitive K⁺ channels are expressed by the urothelium including the calcium-activated potassium channels Kcnma1 (BK channel α subunit) as well as Kcnn1–4 (SK/IK), the two-pore potassium channels Kcnk2 (TREK-1) and Kcnk4 (TRAAK), and the inwardly rectifying potassium channels Kcnj8 (Kir6.1) and Kcnj11 (Kir6.2) (443, 782). The latter two are subunits of ATP-sensitive potassium channels (K_ATP channels), which have reported roles in bladder function, particularly at the level of the detrusor (89, 258). Expression of KCNJ8 was confirmed by Western blot, and immunofluorescence demonstrated that the protein was expressed at the basolateral surface of the umbrella cell (782). While its function is not well understood, the K_ATP channel opener cromakalim potentiates stretch-induced exocytosis, an effect that is antagonized by glyburide (a K_ATP antagonist) (782). At present, our current understanding of how these channels contribute to urothelial function is limited.

c) trp family channels.

Additional stretch-sensitive channels include some members of the transient receptor potential (TRP) family including TRPV1, TRPV4, and PKD2 (9, 70, 77, 110, 246, 662, 780). It is worth noting that unlike PIEZO1 (149, 682), there is no evidence that TRP family channels are directly gated by changes in membrane tension (456, 538). So, while TRP family channels respond to mechanical stimuli, they are not likely to act as a verifiable mechanosensors, which must directly sense, alter their conformation, and trigger responses to mechanical stimuli (40). In addition to responses to mechanical stimuli, TRP family channels have well-described roles in chemosensation, pain, and temperature sensation (483). For examples, TRPV1 is sensitive to capsaicin and heat, while TRPM8 is involved in cold sensation. Thus TRP family channels could be important in promoting urothelial sensory functions in response to mechanical and chemical signals.

Of the nominally stretch-sensitive TRP channels, Birder et al. (77) were the first to implicate TRPV1 in bladder function. They reported that TRPV1 is expressed in the mouse bladder urothelium and afferent termini. However, localization of TRPV1 to the rodent bladder urothelium is somewhat controversial and may be species or antibody dependent (205, 768, 780) (TABLE 4). In humans, TRPV1 expression is reported in the bladder urothelium, detrusor, nerve fibers, and interstitial cells (TABLE 4) (37, 405, 553, 772). Unfortunately, the transcriptomic analysis by Yu et al. (786) sheds little light, as there is limited expression of TRPV1 in human or mouse urothelial cells. TRPV1 KO mice exhibit several voiding abnormalities including increased numbers of nonvoiding contractions and increased numbers of small voids, but also exhibit a nominal increase in their bladder capacity coupled with decreased apical ATP release (TABLE 4). Classifying this phenotype is not straightforward, but the first two characteristics are consistent with KO mice having a hyperactive bladder. Intriguingly, Trpv1 KO mice appear to be less sensitive to the effects of lipopolysaccharide, exhibiting less inflammation, and fewer bladder contractions/unit of time versus control animals (125). Thus Trpv1 could play a role in cystitis and other inflammatory conditions.

In humans, it is not the loss of TRPV1, but its overexpression that appears to result in bladder overactivity. For example, patients with neurogenic detrusor overactivity express increased amounts of TRPV1 in their urothelium and in TRPV1⁺ nerve fibers (TABLE 4). Intriguingly, the TRPV1-desensitizing agents capsaicin and resiniferatoxin are reportedly effective at treating some forms of overactive bladder; however, adverse side effects including hyperthermia and reduced responses to noxious heat have limited their clinical impact to date (22). Taken as a whole, these results support the importance of TRPV1 in bladder function, but do not define which cell type(s) is involved or whether it is the nerve fibers that are primarily affected. Data from conditional urothelial KO mice are not yet available.

TRPV4 is reportedly expressed by the urothelium and detrusor of mouse bladders (134, 246, 384, 780), and the urothelium and interstitial cells of human bladders (333, 334), and the urothelium lining human ureters (635) (TABLE 4). There is evidence that TRPV4 may contribute to human bladder dysfunction including findings that patients with overactive bladders exhibit increased TRPV4-mediated ATP release (596), and a report that 9 of 17 patients suffering with Charcot-Marie-Tooth disease type 2C (which results from gain-of-function mutations in TRPV4) suffer from bladder urgency and incontinence (399) (TABLE 4). If excess TRPV4 expression/activity leads to an overactive bladder, then inhibiting TRPV4 or generating animal models with reduced Trpv4 expression is likely to quell bladder overactivity, or result in an underactive bladder phenotype. Consistent with these possibilities, treatment with HC067047, a potent and selective inhibitor of TRPV4, increases bladder capacity and reduces micturition frequency in mice and rats with cyclophosphamide-induced cystitis (207). Moreover, Trpv4 KO mice have an underactive bladder phenotype (246). In contrast, Trpv4 KO rats do not have a measurable bladder phenotype (175), which may reflect compensation by other channels or species differences. Interestingly, TRPV4 is reported to be associated with the adherens junction of umbrella cells (333). As this is a site of mechanotransduction (124), TRPV4 could be an important player in urothelial responses to mechanical stimuli or permeability. In support of this possibility is a recent report that the paracellular barrier across the gingival junctional epithelium is perturbed in Trpv4 KO mice (364). While the data support a role for TRPV4 in normal and abnormal bladder function, the urothelial contribution to these events is difficult to interpret as TRPV4 is expressed in multiple tissues within the bladder wall, and the phenotype of conditional urothelial Trpv4 KO mice has not been reported.

Currently, there are few insights into what PKD2 may be doing in the urothelium. In the kidney, PKD2 is localized in part to cilia, where it is proposed to sense flow-regulated signaling events in the nephron (414). However, adult urothelial cells lack cilia. Furthermore, mutations in PKD2 are an important cause of autosomal dominant polycystic kidney disease in humans (98); however, we are unaware of any lower urinary tract abnormalities in patients with this disease.

In addition to TRPV1, TRPV4, and PKD2, the mouse urothelium expresses mRNA for an additional 19 TRP family members (out of 33 possible) (780). Of these, PKD1, TRPA1, TRPC1, TRPC4, TRPM7, TRPM8, TRPML1, and TRPV2 were confirmed by immunofluorescence and Western blotting to be expressed in the mouse urothelium (780). In humans, TRPV2, TRPA1, TRPM7A/B, and TRPM8 are reportedly expressed in the bladder and urothelium, as well as suburothelial nerve fibers (TABLE 4). TRPA1 is reportedly expressed in the basal cells of the urothelium lining the urethra, as well as nerve fibers that extend into the urothelium (262). In the case of TRPM8, its expression is increased in a subset of patients with IC/BPS, and there are increased numbers of TRPM8⁺ nerve fibers in patients with idiopathic detrusor overactivity or painful bladder syndrome (301, 513). Bladder function has also been assessed using KO mice models (TABLE 4). Some of the KO mice have no phenotype unless the animals are further stressed. In the case of Trpm7, Watanabe et al. (753) used conditional urothelial KO mice (Trpm7 KO is embryonically lethal). These conditional KO mice not only have smaller void volumes, they also have abnormal intercellular junctions, which likely explains the inflammation and edema observed in their lamina propria. Studies in humans are limited at this time, but in patients with interstitial cystitis there is no detectable change in TRPM7 expression (301). To summarize, other than Trpm7 conditional KO mice, we have limited understanding of the contribution that urothelial-expressed TRP channels make to lower urinary tract function.

2. Purinergic receptors

a) expression of purinergic receptors in the urothelium.

As noted above, Ferguson et al. (214) were the first to show that ATP is released by the urothelium in response to stretch, followed by reports that P2rx2 and P2rx3 KO mice exhibit bladder underactivity (141, 142). Since the publication of these studies 20 years ago, the interest in the purinergic signaling of the lower urinary tract has not abated. Purinergic receptors form a broad class of receptors that include G protein-coupled P1 receptors (also known as adenosine receptors), G protein-coupled P2Y receptors (which bind ATP, ADP, UTP, and UDP), and ligand-gated ion channel P2X receptors (which preferentially bind ATP).

All three pathways are represented in the urothelium. For example, rodent bladder urothelium expresses all four adenosine receptors (ADORA1, ADORA2A, ADORA2B, ADORA3), all seven P2X receptors (P2RX1–7), and multiple P2Y receptors (TABLE 4) (P2RY1,2,4,6) (79, 197, 412, 677, 690, 732, 747, 784). In humans, ADORA2B expression is apparently limited to the detrusor (556, 645), P2RX1 and P2RX4 are expressed in the detrusor and nerve fibers (507, 546, 547, 588, 718), and P2RX2 and P2RX3 are expressed throughout the bladder wall including mucosa (197, 481, 507, 547, 588, 643, 681, 690, 750). P2XR7 is reportedly expressed in bladder umbrella cells (681), message for P2RY1 and P2RY2 is found in the bladder urothelium (635), and P2RY6 is also reportedly expressed in the bladder urothelium (643); however, this is controversial (635) (TABLE 4). Human ureter urothelium expresses P2RX4 and P2Y1 (635) (TABLE 4). Furthermore, and as noted above, interstitial cells express P2RX3, P2RY1, P2RY2, P2RY4, and P2RY6, and afferent nerve termini express P2RX2, P2RX3, P2RY1, P2RY2, and P2Y4. Thus the major cell types (and nerve termini) of the mucosa are well equipped to respond to release of purines from the urothelium (or those present in extracellular fluids such as urine), presumably modulating bladder function (FIGURE 14).

b) role of purinergic receptors in lower urinary tract function and dysfunction.

While there is little doubt that purinergic signaling is integral to urothelial function, it is difficult to unequivocally identify the receptor(s) involved as many of the traditionally employed pharmacological agonists and antagonists are not selective (330, 397, 738). However, progress is being made on this front, and more selective drugs are available or in the offing (330, 397, 543, 738). While genetically modified mouse models could provide greater insights into the relevant purinergic receptors in the urothelium, the lack of urothelial specific KO mice impacts our understanding of the relevant urothelial input pathways. Below, we provide an overview of the current state of our knowledge.

We begin with the P1 receptors. Adora2b KO mice exhibit an overactive bladder phenotype, although somewhat confusingly bladder strips from these KO mice show weaker contractions versus those in control mice (281) (TABLE 4). Using human bladder smooth muscle strips, Pakzad et al. (556) argue that ADORA2B is involved in detrusor contraction, while Hao et al. (281) suggest that ADORA2B counters the stimulatory effects of ATP. We are unaware of studies that have exploited available rodent models deficient in expression of Adora1, Adora2a, or Adora3 (737), although there are pharmacological studies that implicate adenosine receptor pathways in urothelial signaling (365, 578, 747), including a study that describes the signaling cascade that occurs downstream of ADORA1 activation and stimulates apical exocytosis in umbrella cells (577). In general, there is limited understanding of the role of adenosine receptor signaling in urothelial function.

More data are available for P2X receptors (TABLE 4). Cockayne and co-workers (141, 142) were the first to report that P2rx3 KO mice exhibit bladder hyporeflexia, and a later study showed a similar phenotype using P2rx2 or double KO mice. Furthermore, several studies implicate urothelial purinergic signaling in afferent activation. Widely cited studies by Namasivayam et al. (527), Rong et al. (604), and Vlaskovska et al. (736) demonstrate that intravesicular instillation of P2X receptor agonists (e.g., ATP or α/β-methylene ATP), most likely acting on apical umbrella cell P2X receptors, stimulates afferents in rats or in P2rx3 control mice, but not KO ones. However, the serosal-released mediator(s) that promotes afferent signaling in these studies remains unknown. P2rx2 and P2rx3 KO animals also exhibit defects in umbrella cell DFV exocytosis (747). In contrast, and despite using the same mice as those described by Cockayne and co-workers (141, 142), Takezawa et al. (688) failed to observe any bladder defects in P2rx2 or P2rx3 KO mice, except in response to bacterial lipopolysaccharide. P2rx4 KO animals do not have a readily apparent bladder phenotype (781), but P2rx7 KO mice exhibit less nociceptive behavior and develop a smaller inflammatory response to cyclophosphamide-induced cystitis (466). The findings by Takezawa et al. (688), coupled with a lack of data from conditional urothelial KO mice, make it unclear which P2X channels are functional in the normal mouse urothelium.

There are several reports that P2X expression is altered in patients with lower urinary tract dysfunction (TABLE 4). For example, in individuals suffering from bladder outlet obstruction, expression of P2RX2 and P2RX3 is lost in the urothelium (643), while expression of both are increased in the mucosa of patients with interstitial cystitis (441, 690). Other studies report disease-related changes in P2RX2 and P2RX3 expression in the suburothelial and detrusor-associated nerve fibers of patients with idiopathic detrusor overactivity, sensory urgency, or neurogenic detrusor overactivity (95, 507, 588) (TABLE 4).

Few studies have explored the role of urothelial expressed P2Y receptors. P2ry1 KO mice have been used in studies of interstitial cells (410), but no analysis of lower urinary tract function has been reported. Interestingly, P2ry2 KO mice also exhibit altered water and sodium reabsorption, generating a more concentrated urine (794), but to date, no study has examined whether these animals have a lower urinary tract phenotype. P2ry6 KO mice have a hyperactive bladder, but the lack of any effect on contraction of detrusor strips or urothelial release of ATP led these authors to hypothesize that the effect may reside in the CNS and/or dorsal root ganglia (363). P2YR6 activation facilitates ATP and acetylcholine release from the mucosa of human bladders (TABLE 4) (643).

Despite the abundant literature implicating purinergic signaling in lower urinary tract biology (TABLE 4) (335, 689), there remains few hard insights into which specific purinergic input (i.e., receptor) pathways are functional in the urothelium.

3. Receptors for neurotransmitters

a) muscarinic and nicotinic acetylcholine receptors.

Afferent neurons are reported to produce a variety of chemical and peptide neurotransmitters including substance P, CALC1, leucine enkephalin, CRH, VIP, NO, glutamic acid, and aspartic acid (reviewed in Refs. 168, 483). Not coincidently then, the rodent urothelium expresses receptors for many of these including those for acetylcholine. In particular, the rodent urothelium expresses all five G protein-coupled muscarinic receptors (CHRM1–5) (57, 68, 133), the nicotinic receptor subunits CHRNA3 and CHRNA7, and message for Chrna2, Chrna3, Chrna4, Chrna5, Chrna6, Chrna7, Chrna9, Chrna10, Chrnb3, and Chrnb4 (Chrnb2 is apparently expressed in afferent neurons) (58, 60, 789).

There is also ample evidence for muscarinic receptor expression in the human urothelium and other tissues in the bladder wall (TABLE 4). mRNA for CHRM1, CHRM2, and CHRM3 is reported in the human mucosa/urothelium and detrusor layers (101, 161, 457, 711), while CHRM1 and CHRM2 proteins are expressed in the urothelium, in the suburothelium, and in the detrusor of human bladders (101, 161, 457). Bschleipfer et al. (101) describe CHRM2 expression as being exclusively in the umbrella cell layer, but Datta et al. (161) report no CHRM2 expression in the urothelium. Expression of CHRM4 in the mucosa is variable, with some groups reporting mRNA in the mucosa (155, 711), but others describing expression as not present or being inconsistent (101, 457). CHRM5 is expressed in the mucosa, including all three layers of the urothelium, and in the detrusor (101, 457, 711). At present, there is limited information concerning nicotinic receptor expression in the human urothelium.

Global KO mice for all five muscarinic receptors have been generated (see TABLE 4), but only Chrm2, Chrm3, and Chrm5 KO mice have a lower urinary tract phenotype (172, 196, 316, 469, 663). A bladder phenotype has also been reported for nicotinic receptors including Chrna3, Chrnb2, Chrnb4, and double Chrnb2;Chrnb4 KO mice (see TABLE 4) (765, 766). Interestingly, the Chrna3 and Chrnb2;Chrnb4 double KO mice exhibit a phenotype similar to patients with megacystis-microcolon-intestinal hypoperistalsis syndrome, an autosomal recessive disorder (416, 765, 766). In these patients, several variants with high-frequency polymorphisms are found in both the CHRNA3 and CHRNB4 genes (416).

There are reports that expression of urothelial-expressed muscarinic receptors is altered in patients with bladder disorders (TABLE 4). For example, expression of CHRM1 is decreased in the urothelium and suburothelium of patients with detrusor overactivity, but increases upon botulinum toxin treatment (161). Increased CHRM2 expression is reported in the urothelium of human donors with bladders that exhibit an M2-mediated contractile component (97). In addition, expression levels of CHRM2 (mRNA and protein) are affected by age and gender (41). Patients with homozygous loss-of-function mutations of CHRM3 present with bladder malformations, detrusor hyporeflexia, distended bladder with high residual volumes after micturition, variable vesicoureteric reflux, and hydronephrosis (56, 754). The urothelial contribution to this phenotype is unknown.

Finally, it has been hypothesized that urothelial-expressed muscarinic receptors may be druggable targets for conditions such as overactive bladder (166, 484). This is based in part on the following: 1) expression of multiple muscarinic receptors in the urothelium; 2) in several species, the presence of urothelium blocks muscarinic agonist-induced smooth muscle contractions; and 3) detrusor overactivity can be lessened when muscarinic antagonists are administered intravesically (23, 484). However, as of yet, there are no treatments that specifically target urothelial-expressed muscarinic receptors.

b) other neurotransmitter receptors expressed in the urothelium.

Message for the α- and β-adrenergic G protein-coupled receptors Adra1a, Adra1b, Adra1d, and Adrb1, Adrb2, and Adrb3 are expressed in the porcine bladder urothelium (510), and human ureter urothelium expresses message for ADRB1, ADRB2, ADRB3 (471). The ADRB3 agonist Mirabegron is reported to be efficacious in the treatment of overactive bladder symptoms (471, 769). While the focus has been on its effects on detrusor function, it is possible that urothelial expressed ADRB3 may also contribute to its activity. Other urothelial-expressed receptors include those for the pituitary adenylate cyclase-activating polypeptide receptor ADCYAP1R1 (252), the bradykinin receptors BDKRB1 and BDKRB2 (135), the substance P receptor TACR1 (NK1) (291), and the neurokinin receptors NTRK1 and NTRK2 (TrkA and TrkB) (251). Considering the utility of CALC1 in marking afferent nerve termini (236), it is surprising there are no reports if the CALC1 receptor (comprised of RAMP1 and CRLR) is expressed in the urothelium.

4. Additional receptors

Several other receptors are expressed by the urothelium, including one for nerve growth factor (NGFR) (251), a signaling pathway that has been implicated in a number of bladder pathologies. Schnegelsberg et al. (625) have overexpressed nerve growth factor in the urothelium and find that it stimulates neuronal sprouting and inflammation, increases voiding frequency, and stimulates somatic hypersensitivity. Other growth factor receptors expressed by the urothelium include the epidermal growth factor family receptors EGFR, ERBB2, and ERBB3 (49, 725). In addition, the G protein-coupled prostaglandin receptors PTGDR, PTGDR2, PTEGR1, and PTEGR2 are expressed in the bladder, ureters, and urethra (267, 268, 551, 584). Interestingly, basal cells express PTGS1 (cyclooxygenase I), but the umbrella cells do not (169, 581). Thus the basal layer of the urothelium may be an important site of prostaglandin synthesis. The urothelium is also reported to express the cannabinoid receptors CNR1 and CNR2 (261, 290).

Consistent with its role in forming a barrier to bacterial infection, the urothelium expresses multiple Toll-like receptors (TLRs), which function to recognize and mount responses to invading pathogens and products of inflammation and cell death (400). The TLRs recognize pathogen-associated molecular patterns (PAMPs; e.g., lipopolysaccharide), and damage-associated molecular patterns that are generated by cell and tissue damage (DAMPs; e.g., extracellular DNA) (400). Normal human urothelium is reported to express message for TLR2, TLR3, TLR4, TLR5, TLR7, and TLR9, with particularly strong expression of TLR4 and TLR9 (43, 400). The best understood TLR, TLR4, recognizes lipopolysaccharides (i.e., endotoxin), which is particularly relevant to the bladder as E. coli, the major cause of bacterial cystitis, is a gram-negative organism. TLR4 protein is reportedly expressed in the urothelium, including at the apical surface of the umbrella cells (46, 248, 614, 652).

C. Urothelial Transducer Function

The other major role for the urothelium is to respond to input signals by generating mediators that act on other cell types in the mucosa and tissues deeper in the wall of the associated lower urinary tract organs. Potentially important sensory outputs include ATP, adenosine, acetylcholine, NO, H₂S, and prostaglandins (FIGURE 14). For those interested in a more detailed overview of these mediators and associated effects, as well as their roles in lower urinary tract dysfunction, we recommend the following reviews (65, 71, 74, 630).

1. Release of ATP

As previously noted, Ferguson et al. (214) were the first to report that “stretching” bladder strips, or full-thickness preparations of bladder wall mounted in Ussing chambers, stimulated the release of ATP. Subsequent studies confirmed and extended these observations by showing, for example, that the urothelium lining the ureters is also a source of ATP (368), that stretch stimulates ATP release five- to sixfold above baseline from both the mucosal and serosal surfaces of dissected urothelium, and that turnover of ATP is greater at the serosal than mucosal surfaces (426, 747). This difference in hydrolysis results in much larger accumulation of ATP at the mucosal surface of the tissue. Other investigators have established assays that allow one to measure mucosal ATP release in in vivo or ex vivo preparations of animal bladders (59, 246), as well as in patients undergoing bladder filling (132, 646). In the case of the latter, ATP is primarily released early in filling (during the first 200 ml). In addition to stretch, ATP is released from the urothelium in response to other input pathways including β-adrenoceptor agonists, muscarinic agonists, ADCYAP, TRPV1 agonists, EGF, anti-proliferative factor, and even ATP itself (reviewed in Ref. 630). In the latter case, ATP acts in an autocrine manner to stimulate further ATP release (672). There is also a report that adenosine inhibits the release of urothelial ATP (187). Thus an end product of ATP hydrolysis, adenosine, can act as a negative regulator of ATP release.

Altered ATP release may contribute to lower urinary tract dysfunction. For example, the urothelium of patients with interstitial cystitis releases larger amounts of ATP than controls (386, 676), and isolated urothelium derived from patients with overactive bladder exhibits a similar phenotype (81, 385). Presumably, ATP stimulates sensory input pathways (i.e., P2X receptors), which ultimately results in detrusor contraction.

A matter of some interest is the mechanism(s) by which ATP is released from the urothelium. Early studies employed pharmacological and biochemical approaches, which indicated that multiple pathways may be involved including those dependent on connexin/pannexin hemichannels, vesicular release, ABC proteins, and nucleoside transporters (275, 368, 747). However, more recent work is providing additional molecular insights into the process. Nakagomi et al. (522) focused their attention on SLC17A9 (vesicular nucleotide transporter; VNUT), which is required for the membrane-potential/pH dependent uptake and storage of nucleotides in secretory granules, before their release by exocytosis (508). ATP, which is a common constituent of secretory granules, is a primary target of SLC17A9, but other nucleotides are transported with the following efficacy: ATP > UTP > GTP > ITP/ADP >> AMP (508). Nakagomi et al. (522) observed that SLC17A9 is abundantly expressed in the mouse urothelium, where it is localized to small, acidic vesicles in cultured urothelial cells. Consistent with a vesicular transport process, they observed that when cultured urothelial cells were subjected to 10% strain, ATP release was inhibited by the following: brefeldin-A (a secretory inhibitor), bafilomycin A (which inhibits the V-ATPase), or botulinum toxin A (which cleaves and inactivates the t-SNAREs SNAP25/SNAP23). More tellingly, they observed a significant reduction in stretch-induced ATP release from urothelial cells cultured from Slc17a9 KO mice. They further noted that when exposed to higher strains (20%), the mechanism of ATP release is no longer sensitive to these inhibitors (except for brefeldin-A) and may instead occur via a mechanism that requires connexin hemichannels or pannexins (PANX1–3), which form large transmembrane channels that are also implicated in ATP release (see discussion below).

A phenotype of P2rx2 and P2rx3 KO mice is an underactive bladder (141, 142). Unexpectedly, Nakagomi et al. (522) report that Slc17a9 KO mice exhibit increased voiding frequency with smaller urine volumes, and in cystometry, decerebrate animals show decreased intercontraction intervals and reduced compliance. Thus, in the case of SLC17A9-dependent nucleotide release, nucleotides are likely acting to suppress bladder activity, and thus promote bladder compliance. The target sites in this case are unknown but could be afferents or subjacent tissues. While one would likely ascribe these changes to the urothelium, SLC17A9 is also apparently expressed in the lamina propria and detrusor; thus the Slc17a9 KO mouse phenotype is not easily parsed. In addition, it is presumed that ATP is the actor in this system, but it is equally plausible that it is a different nucleotide, but one that also depends on SLC17A9-dependent import into vesicles. Finally, it would be intriguing to understand whether SLC17A9 functions in both mucosal and serosal release of ATP from the urothelium.

As noted above, ATP is released from both the mucosal and serosal surfaces of the urothelium. It may be tempting to assume that the basolateral release is more physiologically relevant; however, there is evidence that mucosal ATP release can also modulate bladder function by way of a “transmural signaling pathway.” For example, instilling ATP into the bladder lumen stimulates bladder overactivity in freely mobile rats (557). The input pathways involved are not clearly defined, but they are sensitive to the purinergic receptor antagonist PPADS, and thus are likely to be purinergic receptor dependent. Beckel et al. (59) demonstrated that instilling apyrase (an enzyme that hydrolyzes ATP) into the bladder lumen decreases bladder activity, while instillation of ARL67156 (an inhibitor of ecto-ATPases) increases bladder activity. They further showed that knocking down expression of Panx1 using an siRNA approach, or treating the mucosal surfaces with pannexin channel antagonists (Brilliant Blue FCF or carbenoxolone) have two broad effects: 1) they significantly reduce mucosal ATP release, and 2) they significantly increase the intercontraction interval and threshold pressure in cystometrograms. Coupled with observations that PANX1 is expressed in the urothelium, including umbrella cells, it appears that ATP release via PANX1 channels may be stimulating bladder activity. This is consistent with work from Negoro et al. (532) who observed decreased luminal release of ATP in Panx1 KO mice (or P2x7r KO mice). However, Negoro et al. (532) did not assess bladder function. It will be intriguing to confirm that PANX1 is localized to the apical surface of the umbrella cells, determine if it is also important for serosal release of ATP, and assess if Panx1 conditional urothelial KO mice have a phenotype.

2. Release of nucleosides and nucleotides other than ATP

While the general focus of urothelial research has been on ATP release, the urothelium can also release or produce nucleosides such as adenosine and inosine and nucleotides such as AMP, ADP, ADP-ribose, and cADP-ribose in response to stretch (188, 189, 578, 784). The function of these nucleosides/nucleotides in bladder function is not well understood, but mucosal addition of the ADOR1A-receptor agonist CCPA decreases the threshold pressure when cystometry is performed in urethane-anesthetized rats (578). Likewise, inhibitors of pathways important for adenosine conversion to inosine, or uptake by nucleoside transporters, also decrease threshold pressure (578). Work by Kitta et al. (365) demonstrated that CCPA administered intravesicularly significantly increases the intercontraction interval in cytometry of urethane-anesthetized rats. Moreover, they reported a similar effect on bladder function when CCPA or the ADOR2A agonist ZM24138 was injected intravenously, intrathecally, or intracerebroventricularly. Interestingly, there are reports that adenosine (and CCPA) inhibits detrusor responses to electrical field stimulation, impairs P2X-mediated excitatory junctional potentials, and promotes detrusor relaxation (629, 755). Thus adenosine produced by the urothelium could be an important mechanism to regulate function in underlying tissues (FIGURE 14).

3. Release of acetylcholine

Urothelial cells are reported to express the machinery necessary to synthesize and release acetylcholine including the plasma membrane choline transporter CHT1, the organic cation transporter SLC22A3 (OCT3), and the acetylcholine-synthesizing enzymes choline acetyltransferase (CHAT) and carnitine acetyltransferase (CRAT) (278, 773). However, Lips et al. (437) argue that only CRAT, along with the organic cation transporters SLC22A1 and SLC22A5 (OCT1–2), are expressed in the urothelium. In either case, the insensitivity of acetylcholine release to brefeldin-A treatment indicates a different mechanism of release than that observed in neurons. The urothelium releases acetylcholine in response to both mechanical and chemical stimulation including ATP (FIGURE 14) (278, 540, 664, 773). Once released, there are a number of sites where urothelial-derived acetylcholine could exert its effects including smooth muscles, nerves, and urothelial associated-muscarinic and/or nicotinic receptors. In addition, because cholinergic receptor agonists stimulate the release of NO and ATP from the urothelium (78, 382), the effects of acetylcholine could be indirect and via these other output pathways.

4. Release of NO and other gaseous transmitters

NO is a colorless gas synthesized from l-arginine by NO synthases, which are expressed in the urothelium (78, 211, 250). It is released in response to norepinephrine (an α/β-adrenergic receptor agonist), capsaicin (a TRPV1 agonist), isoproterenol (a β-adrenoceptor agonist), and substance P (66, 78, 516). NO produced by neurons has a well described role in relaxing the urethral sphincter during micturition (167). But, it may have other functions in the lower urinary tract including modulation of afferent nerve function and that of interstitial cells, which as noted above appear to be important targets of NO production (8, 249). In addition to NO, there is recent evidence that H₂S is also produced by the urothelium, which expresses the H₂S synthesizing enzymes cystathionine-β-synthase (CBS) and cystathionine-γ-lyase (CTH) (155, 232). The mechanisms that trigger H₂S release by the urothelium are not known, but carbachol stimulates H₂S production in the bladder carcinoma cell line T24, as does Sildenafil (155, 232). In the bladder, H₂S apparently has an inhibitory effect on detrusor function (232, 683).

5. Release of prostaglandins and other mediators

The urothelium lining the bladder and ureters is a site of prostaglandin synthesis, including PGF_2α, PGD₂, PGE₂, and PGI₂ (13, 265, 266, 350, 478). Prostaglandins are released by stretch, as well as in response to ATP and acetylcholine (539, 540). NO may inhibit PGE₂ production (540). Prostaglandins released from the urothelium may play roles in modulation of nerve and detrusor functions (FIGURE 14) (reviewed in Refs. 68, 630).

D. Circadian Rhythms and Lower Urinary Tract Function

We finish this overall section by describing an emerging field of inquiry that is focused on understanding the impact of circadian rhythms on bladder function. Circadian rhythms are those physical, mental, or behavioral functions that follow an ~24-h daily cycle, typically in response to diurnal changes in light and dark (104). Examples of bodily functions that are regulated by circadian rhythms include metabolism, sleep, temperature, blood pressure, and micturition. In the case of the latter, there is smaller production of urine by the kidneys during sound sleep, and more urine is stored by the bladder during sleep than in active states (525, 531, 719, 760). However, in those patients who suffer from nocturnal enuresis or nocturia, there is a mismatch between urine production and storage (531). The early indications of a link between defects in urinary function and circadian rhythms are explored in the following section.

Circadian rhythms are regulated by a “master clock,” which resides in the hypothalamus of mammals within a structure called the suprachiasmatic nucleus (SC) (104). Comprised of ~20,000 neurons, the SC receives input from several sources including the photosensitive ganglia of the retina. In turn, these neurons release substances such as VIP (282), which promotes release of hormones including cortisol and melatonin from other regions of the hypothalamus. These hormones travel to peripheral target tissues where they regulate clock-driven events by way of so-called “clock genes,” whose function is to generate (and regulate) oscillations in gene expression at the tissue level (104). Interestingly, the expression of clock genes in these tissues remains somewhat rhythmic even in the absence of SC input. Thus it is generally hypothesized that the function of the SC is to ensure that all of the bodies’ cells/tissues synchronize to a single time interval.

The rhythmic oscillations generated by the clock genes depend on autoregulatory feedback loops (104). These cyclical loops, which include changes in transcription, as well as in posttranslational events such as nuclear import and phosphorylation, take ~24 h to complete. The clock genes that encode the so-called “core loop,” which is found in most cells/tissues/organs that exhibit circadian rhythmicity, include the transcription factors CLOCK and ARNTL (BMAL1). These core clock genes dimerize in the cytoplasm, and upon nuclear import stimulate expression of genes such as Per1, Per2, Per3, Cry1, Cry2, Clock, and other Clock-controlled genes that contain an E-box enhancer sequence in their promoters. In turn, cytoplasmic PER1–3 and CRY1–2 assemble into heterodimers, and upon import into the nucleus inhibit the activity of CLOCK/ARNTL dimers. By downregulating the activity of the latter, the expression of Per1–3 and Cry1–2 is diminished. Rhythmic oscillations of the core loop are followed by oscillations of other subloops or peripheral clocks, which further regulate physiology at the tissue and cellular levels. Examples of these subloops include the products of the clock genes Dbp and Nfil3, which interact with D-box sites, and the gene products of the clock genes Rora and Nr1d1 (Rev-erbα), which bind to ROR element sites (104).

An emerging literature supports the hypothesis that bladder function is under the control of circadian rhythms, and by extension clock genes (531). Negoro et al. (530) first reported that 184 bladder genes exhibit circadian rhythmicity, including the connexin-family gene Gja1 (connexin43). Intriguingly, oscillations in Gja1 expression are lost in Cry1;Cry2 null mice, which exhibit a dysfunctional clock. Moreover, the Gja1 promoter is modulated by Sp1 transcriptional elements that are under the regulation of the clock gene Nr1d1. Relative to wild-type mice, Gja1^+/− mice exhibit increased void volumes during their active, night cycle, indicating that Gja1 is important in regulating functional bladder capacity (530). Strikingly, the rhythmic nature of the mouse micturition pattern is lost in Cry1;Cry2 null mice, which exhibit clock dysfunction. These observations led the authors to argue that the oscillatory effects are primarily due to changes in smooth muscle cell Gja1 expression (530); however, there are reports that bladder interstitial cells (described above), which surround bundles of smooth muscle cells, as well as the urothelium express GJA1 (245, 323, 631). Thus interstitial cells, and possibly the urothelium, may also contribute to the phenotype described by Negoro et al. An additional connexin, GJB2 (connexin26), also exhibits rhythmic oscillations (275). Its potential role in bladder function is described below.

Other studies support the hypothesis that circadian rhythms regulate bladder function. For example, there is a correlation between melatonin secretion and nocturia in aged patients (549), and Ihara et al. (320) have shown that Clock^∆19/∆19 mice, which express a dominant-negative mutant of the Clock gene, exhibit a phenotype that is reminiscent of patients with nocturia. Clock^∆19/∆19 mice (relative to control mice) exhibit a loss of their diurnal pattern of urine volume per void, show lower functional bladder capacity, and evince greater voiding frequencies and urine volumes in their sleep phase. The same group has established that intermittent restraint, a form of stress, induces what they describe as a circadian “misalignment,” resulting in nocturia (322).

Most relevant to the topic of this review are reports that mucosal/urothelial-expressed genes also exhibit rhythmic diurnal oscillations. Ihara et al. used preparations of mouse bladder mucosa to show that in addition to core loop and subloop clock genes, the following four genes also exhibit diurnal rhythmic oscillations: Piezo1, Trpv4, Gjb2, and Slc17a9 (321). Interestingly, none of these four genes was identified in the initial analysis by Negoro et al. (530), indicating that other urinary tract-associated clock-regulated genes may still yet be discovered. Expression for these four genes is maximal at the beginning of the active phase and bottoms midway in the sleep phase. These oscillations are lost in Clock^∆19/∆19 mice. Interestingly, the promoter region of all four genes contains several E-box-like elements, potential binding sites for CLOCK/ARNTL complexes, and chromatin precipitation assays confirm that CLOCK is associated with the promoters for Piezo1, Gjb2, and Slc17a9 (321). The regulation of Trpv4 is less clear. Subsequent studies by the same group further established that even in isolated urothelial cells, circadian expression of these four genes can be triggered by adding a bolus of serum (317), a common tactic used to study the circadian clock in cultured cells. Moreover, they describe that stretch-induced changes in intracellular Ca²⁺ also exhibits rhythmic oscillations (318), which may be due to the activities of PIEZO1 and TRPV4, although this requires confirmatory testing. Finally, there are reports that urothelial ATP release also exhibits temporal oscillations, with most ATP release occurring during the mouse active (dark) phase (319, 631).

If one accepts that PIEZO1, TRPV4, GJB2, SLC17A9, and ATP are critical components of the urothelial machinery that senses and responds to bladder distension, then enhanced expression of these genes (or release of excess ATP) during the sleep phase is likely to result in heightened bladder sensitivity, increasing the likelihood that nocturia/enuresis will occur. By genetically removing urothelial expression of these proteins, it may be possible to test this hypothesis directly. Other potential areas of future investigation into circadian rhythms and lower urinary tract function and dysfunction are described below.

E. Pressing Questions in Urothelial Sensor and Transducer Function

While the past decade has seen advances in our understanding of urothelial sensory and transducer functions, there are a number of unanswered questions that need to be addressed.

• 1. Which urothelial input pathways are biologically relevant? While pharmacological approaches have their place, off target effects make it critical that investigators use more selective approaches to verify a protein’s role in bladder function. While mouse genetic models are often used, interpreting the phenotype of global KO mice is difficult because the gene products in question are often expressed in multiple tissues. An important example is the P2rx2;P2rx3 KO mice, which were used to establish a role for purinergic signaling in afferent signaling (141, 142). As noted above, P2RX2 and P2RX3 are reportedly expressed by the urothelium (747), which make it difficult to tease apart their urothelial versus afferent nerve contribution to the mouse phenotype. Using conditional urothelial KO mice and/or the transduction/transfection technologies described in section VI, we may be able to better understand which input (and output) pathways are important for lower urinary tract physiology. While there are many good arguments for using mouse models when available, other animal models should be considered. As noted above, TRPV4 KO mice have an underactive bladder (246), but TRPV4 KO rats apparently have no phenotype (175).
• 2. If the urothelium functions as a transducer, how does it convert input pathways to outputs? At present, the urothelium is a proverbial black box, as we know little about the transduction pathways that connect receptor/channel activation to mediator release. Understanding these pathways will likely prove critical if the field is to develop treatments that modulate the release of urothelial mediators.
• 3. Is there a role for intraepithelial communication (purple arrow in FIGURE 14)? This question is particularly relevant to transmural signaling, in which stimulation of input pathways at the apical surface of the umbrella cell leads to changes in bladder function. Do apical receptor pathways trigger mediator release solely from umbrella cells, or does it impact release from the other cell layers in the urothelium? In the case of ATP release, there is evidence for ATP-induced ATP release (672). One can easily imagine how a positive autocrine loop could amplify any urothelial signal, making it more likely to stimulate an effect in underlying tissues. It is also conceivable that mediators released from one urothelial cell layer will block the release of mediators in other urothelial strata. However, we know of no example of this occurring.
• 4. What are the mechanisms by which mediators are released from the urothelium? In the case of ATP, we have evidence that PANX1 and SLC17A9 are involved, and possibly GJA1 (59, 522, 532, 631). But these may not be the sole pathways involved. Other ATP release pathways include those mediated by other connexin hemichannels (e.g., GJB2), the calcium homeostasis modulator CALHM, P2RX7, ABC transporters, and possibly the F₁F₀ ATP synthase (reviewed in Ref. 253). In the case of other urothelial released nucleosides/nucleotides, there is a paucity of insights, although the equilibrative nucleoside transporter SLC29A1 (NBTI) may play a role in adenosine release from the urothelium (578). In the case of acetylcholine, there appears to be a nonvesicular pathway in play, but the nature of this pathway is unknown (278, 773).
• 5. Do pathways that affect the biogenesis, turnover, and reuptake of mediators affect bladder function? The goal in this case is to understand how pathways that alter the amounts of active mediator regulate urothelial signaling. Such insights are also relevant, because they are an additional way to reveal which mediators are most biologically relevant. As an example, ATP is hydrolyzed to AMP by the action of ENTPD1 (CD39), while AMP is converted to adenosine by the action of NT5E (CD73). Despite the well understood importance of these two enzymes, there are few studies that have examined NT5E or ENTPD1 expression in normal urothelial tissues. One study reports that NT5E, but not ENTPD1, is expressed in human urothelium (499). Yu et al. (783) report expression of mRNAs for Nt5e, and for eight ecto-nucleotidases including Ntpd1–8 in the mouse bladder. Of these, only NTPD3 is localized to the urothelium. It is found along the basolateral surface of the umbrella cells, and along the plasma membrane of cells forming the other cell layers. They report that NT5E is not expressed in the urothelium, but is found on smooth muscle cells. This question of pathways for biogenesis and turnover does not solely affect purinergic signaling, as acetylcholine can be rapidly degraded by acetylcholinesterase, and prostaglandins are subject to reuptake by way of SLCO2A1 (prostaglandin transporter) (628). Information about these pathways in the urothelium is limited or nonexistent.
• 6. Which urothelial-released mediators are biologically relevant and what are the downstream targets of these mediators? The focus of research has generally been on ATP, but the release of other nucleosides/nucleotides, as well as NO (and other gaseous compounds), prostaglandins, and acetylcholine make identifying the functionally relevant mediator(s) difficult. However, by selectively blocking input pathways on downstream tissues (a difficult proposition that likely requires generating conditional KO mice for specific receptors), it may be an additional way to understand which mediators are relevant.
• 7. How do circadian rhythms contribute to bladder function and dysfunction? Current data support the likelihood that bladder function is governed by circadian rhythms, and in mice disturbances in these rhythms may lead to lower urinary tract symptoms including nocturia and enuresis. If altered circadian rhythms contribute to patients suffering from nocturia and enuresis remains to be addressed. The cell types affected by circadian rhythms in the bladder include smooth muscle cells and the urothelium, but it will likely be the case that other cell types in the bladder wall (e.g., interstitial cells, endothelial cells, and mesothelium) will also be impacted. The genes under circadian control are not yet fully understood, but in the case of the urothelium, Piezo1, Gjb2, Trpv4, and Slc17a9 exhibit circadian oscillations. Whether other gene products are involved in bladder chronobiology remains to be determined.
• 8. Considering their proximity to the urothelium, a singularly important question is what do interstitial cells do? There appear to be several hypotheses, but their biological function remains mysterious. Most tellingly, are the lack of insights into the mechanisms by which interstitial cells communicate with other tissues. The absence of specialized adhesion complexes or structure such as gap junctions between interstitial cells and other cell types would seem to indicate that intercellular communication, if any, is most likely by way of mediator release. Thus, as a first step, the mediators released by interstitial cells should be illuminated. Fortunately, interstitial cells can be grown as primary cultures (as an example, see Ref. 369), and with several insights into their own input pathways, it should now be possible to identify the mediators by which interstitial cells communicate with other tissues in the bladder/ureter/urethra wall.
• 9. Other than the bladder, does urothelial sensor/transducer function impact other lower urinary tract organs? There are some insights into input and output pathways in the ureter, but there is much less known about the renal pelvis and urothelium lining the urethra. Thus there is much work left to do in this area.
• 10. Finally, are we convinced that the reported expression and localization of receptors, channels, enzymes, and markers in the urothelium and underlying tissues are correct? With the new emphasis by the National Institutes of Health on validation of reagents, particularly antibodies, it is all the more important that we do so. With new techniques available to dissect different populations of urothelial cells without enzymatic disruption (see discussion in sect. VI), it should now be possible to apply single-cell genomic and new single-cell proteomic approaches to confirm expression at the mRNA and protein level (448). Furthermore, as more KO mice and other transgenic animals are employed by urothelial investigators, they should be used to confirm the specificity of antibodies and other reagents. In the absence of these animals, then tools to knockdown gene expression in the native urothelium (also discussed in sect. VI) can be employed in their stead.

VI. TOOLS FOR UROTHELIAL RESEARCH

The future of urothelial research depends in part on applying new tools to study the urothelium. The goal of this section is to call out tools and approaches that are likely to be of interest to the urothelial investigator and are likely to advance our understanding of this fascinating tissue.

A. Urothelial Cultures

One significant limitation when performing urothelial research is the lack of a readily available, physiologically relevant culture model of the urothelium. Existing tumor-derived cell lines (e.g., T24 and their ilk) exhibit almost none of the characteristics of the native urothelium. Our intention here is not to review all of the cell culture models that have been developed over the past 40 yr, but instead to highlight a few that may be of use to current or future investigators. The TEU-2 ureteric cell line, developed by Klumpp and co-workers, is a transformed cell line that expresses the E6 and E7 oncoproteins of human papillomavirus type 16. It is stratified, forms tight junctions, and expresses claudins consistent with native urothelium; however, it has limited uroplakin expression (591). The same group has derived a PD07i human urothelial cell line that appears to be better differentiated and expresses all four major uroplakins (126). Thus it may be more useful to the broader urology community.

Other systems have been described to generate primary cell cultures or cell lines. Truschel et al. (704) described a system in which dissociated rabbit urothelium cells are plated on type IV collagen-coated permeable filter supports (Transwells) and then cultured in serum-free keratinocyte growth medium. Upon switching to medium containing 1 mM Ca²⁺, this stratified cell culture model develops very high TER (>8,000 Ω·cm²). The superficial cell layer expresses tight junction-associated proteins and uroplakins, forms AUM, expresses ENaC activity, and has relatively low permeability to water and urea. We have applied this culture system to cat, dog, human, mouse, and rat urothelial cells with good success (G. Apodaca, unpublished results). Others have reported that human urothelial cells can be expanded by serial culture, and once plated on permeable filter supports, they too can be induced to form highly differentiated urothelial cell cultures (151). A limitation of all of these existing culture models is that they do not form abundant DFVs, and it is difficult to apply physiologically relevant mechanical forces to study stretch-regulated events.

The other advance that may bear fruit is the use of induced pluripotent stem cells (552). Such models hold promise for situations in which bladder replacement must be undertaken (e.g., as a result of cancer). Work to date indicates that these techniques can be employed to generate urothelial cells that express uroplakins. Most recently, Suzuki et al. (680) described techniques to direct differentiation of induced pluripotent stem cells into a mature, stratified urothelium.

Finally, O’Mullane et al. (544) have developed a coculture model to probe interactions between the urothelium and sensory neurons. These could provide new insights into crosstalk between urothelial cells and the nerves.

B. Transfection and Transduction Technologies

A significant limitation of working with animal models of urothelium is the lack of easily applied tools to alter gene expression, and transgenic animals can be hard to come by. As discussed above, while many global KO mice are available, it can be hard to interpret the urothelial-specific aspects of any phenotype without the use of conditional KO animals. Generating the latter requires acquiring (or making) an animal with a floxed gene, followed by breeding, genotyping, and performing animal husbandry. To overcome some of these limitations, there are several reports of investigators using transfection and transduction tools to alter gene expression in the urothelium.

Ramesh et al. (585) reported that pretreating the urothelium, in situ, by instilling the detergent N-dodecyl-β-d-maltoside (or even sodium dodecyl sulfate) into the bladder lumen, made the urothelium receptive to subsequent transduction by adenoviruses. What the detergent does is unknown, but it is unlikely to target the carbohydrate-rich GAG layer, as proposed by the investigators. As the receptor for adenoviruses is CXADR (Coxsackievirus and adenovirus receptor), a protein localized to the tight junction and lateral surfaces of the cell, it is more likely that the detergent is temporarily altering the permeability barrier, exposing this receptor to the virus. Within 12 h of transduction, protein expression is observed. In our own studies we have found the following: 1) this technique can be exploited to express cDNAs and/or shRNAs in the rat bladder (e.g., RAB GTPases, hGH, myosin motor fragments, GEFs, claudins, and ADAM17) (359–361, 504, 577); 2) the efficiency of transduction approaches 70–95% (359, 360); 3) transduction alone does not alter the ultrastructure or expression of differentiation markers, nor does it impact the tightness of the paracellular barrier (360, 504); 4) umbrella cells can be transduced with multiple adenoviruses simultaneously; 5) the expression of proteins can be tuned by changing the titer of viruses (which affects the amount of protein expression), lengthening or shortening the time of incubation before the animal is euthanized (typically 1–2 days for protein expression), use of different promoters, and the use of a tet-regulated system (tet-off) (360); and 6) in rats, transduction is limited to the umbrella cell layer; however, in mice, the whole urothelium is transduced (FIGURE 15A). Other than cells in the urothelium, no other cell type is transduced in the bladder wall using these techniques. One limitation of these techniques is that expression is limited to 2–4 days, and the umbrella cells become smaller in diameter, a change that reverses itself over several days. In addition, adenoviruses can also cause inflammatory responses when used over an extended period of time (177). However, we have not noted any evidence of inflammation in our experiments.

FIGURE 15.

Examples of methods to study the urothelium. A: mouse urothelium transduced with an adenovirus encoding the Ca²⁺ sensor GCAMP5G (10). The entire mouse urothelium is transduced, but no other cell type in the bladder wall expresses the transgene, including subjacent interstitial cells. BC, basal cell; IC, intermediate cell; UC, umbrella cell. B: a “peeled” mouse bladder, connected to a catheter, is filled by a syringe attached to a Luer fitting. C: localization of UPK3A, KRT5, and actin in a peeled mouse bladder. Note that the thickness of the preparation is ~75 µm.

There have also been advances in transfection methodologies that bear mention. One of these, described by Kashyap et al. (351), is the use of DOTAP liposomes to deliver siRNAs to the bladder wall. Beckel et al. (59) used it to transfect the urothelium with siRNAs that targeted pannexin channels, which transport ATP from the cytosol to the extracellular space. Bolenz et al. (88) describe the use of oligodeoxynucleotides as a potential transfection reagent for treatment of transitional cell carcinoma. The reagent not only transduces a number of cell lines, it also works in an ex vivo porcine bladder model. Finally, Tyagi et al. (709) describe the use of antisense peptide nucleic acids tethered to the HIV TAT protein penetrating 11-mer peptide. When a nerve growth factor-targeted antisense probe was added to bladder with cyclophosphamide-induced cystitis, it suppressed bladder overactivity. This indicates that antisense peptide nucleic acids may be potential treatments for bladder disorders.

C. Mucosal Preparations

In this section, we discuss two recent tissue/cell preparations that are likely to give urothelial biologists new insights into the native urothelium tissue or individual isolated urothelial cells.

1. “Peeled bladder” preparation

For the urothelium to communicate with underlying cells/tissues it must release mediators from its serosal surface. However, measuring this release is not trivial as the umbrella cell layer forms an impermeable barrier, and the serosal side of the urothelium faces the lamina propria and detrusor muscle and is therefore inaccessible. Durnin et al. (189) describe a method to dissect the majority of the serosa, musculature, and deep lamina propria, leaving an intact urothelium with a small amount of upper lamina propria attached. In peeled bladders, the urothelial cells remain in a close-to-native state and can be exposed to physiologically relevant stretch as the peeled bladder is filled. Thus this model could be a significant advance over methods that measure release of purines such as ATP from isolated cells. Coupled with ultra-sensitive assays, Durnin et al. (189) could measure release of small quantities of multiple mediators from the serosal surface of the urothelium in a time-resolved manner.

In our own studies, “peeled” bladders are comprised of urothelium and a small amount of underlying connective tissue (total thickness of the preparation is ~60–100 µm) (FIGURE 15), B and C. The urothelium remains intact and highly differentiated (FIGURE 15C), and when mounted in an Ussing chamber retains a high transepithelial resistance of >4,000 Ω·cm². It is easy to imagine how these peeled bladder preparations can be used to explore the release of any known mediator, to identify new mediators, to gain access to receptor pathways present on surfaces of the basal/intermediate cells (and basolateral surface of the umbrella cell), and to perform imaging studies.

2. Nonenzymatic dissection of urothelial tissue and cells

An additional preparation that is worth mentioning was recently described by Lu et al. (444). It is similar to the Durnin preparation, in that the underling tissue layers are dissected from the urothelium. However, in this case the procedure is performed on bladder pieces. After dissection of the detrusor from the mucosa, the entire lamina propria can be removed from the urothelium. Subsequently, the urothelium can be further dissected into its basal, intermediate, and umbrella cell constituents. The possibilities for use of this technique are limited by one’s imagination. The authors describe how such preparations can be used for patch-clamp studies, for polymerase chain reaction and single-cell gene analysis, and for single-cell Western blot applications. Although the technique was worked out in mice, it can apparently be applied to human urothelial tissues as well.

D. Imaging Technologies for Visualizing Urothelial Tissue In Situ

The past two decades have seen enormous advances in live-cell imaging technologies, which are used by cell biologists, physiologists, and bioengineers to gain fundamental insights into biological processes. However, live-cell and intravital imaging of the urothelium has lagged. In part, because of the location of tissue at the inner surface of the bladder wall (which makes it too far from the serosa to easily image using most techniques), and lack of technologies that would allow one to stretch the tissue in a physiological manner while maintaining focus and homeostasis. We discuss three technologies that are beginning to address these limitations.

In the first approach, Sano et al. (619) used intravital imaging of a Förster resonance energy transfer (FRET) biosensor expressed by a reporter mouse (Eisuke-NES mouse) to detect activation of ERK in the bladder wall. In these experiments, the bladder is exteriorized and “attached” to a vacuum-stabilized imaging window. The latter is a custom-made device that attaches to the objective of the microscope, generates suction, and causes the outer surface of the bladder to adhere to a coverslip. ERK activation is detected by a two-photon microscope, which can “penetrate” deep into tissues (~0.25−1.0 mm, depending on tissue type), with minimal photo damage. One limitation of this approach is that images of the urothelium can only be captured if the bladder wall is thin enough. As a result, the investigators filled their bladders to 15–20 cmH₂O. One can easily imagine how a peeled bladder preparation would overcome these limitations. The second approach is by Cheng et al. (130), who used two-photon microscopy, but in this case the bladder is opened, and attached via Biorakes (a tissue piercing system) to a device that allows for biaxial stretching of the bladder specimen. This is a significant advancement over uniaxial stretching devices, which have significant limitations (reviewed in Ref. 374). Using an immersion objective, Cheng et al. (130) measured the effects of stretch on the collagen component of the extracellular matrix. Because the urothelium can be mounted face up in this system, it should be easy to adapt this system to visualize events in this tissue. Finally, Shiwarksi et al. (641) describe a 3D-printed, open-source, biaxial stretching device that includes integrated hardware and software, temperature regulation, and ease of mounting on a microscope stage for sample tracking and time-lapse fluorescent imaging. Imaging of the urothelium is possible using this device (641).

E. Computer Modeling

The final advance we will discuss is work by Moulton et al. (511), who developed a computer model of how exocytosis and endocytosis are regulated in the urothelium and how the rate of exocytosis and rate of filling can affect afferent nerve activity. Among the several predictions made, the following are particularly relevant: 1) increasing apical membrane tension will stimulate afferent signaling (a downstream result of increased mediator release), which will increase bladder activity; and 2) decreasing apical membrane tension will have the opposite effect. Thus we can surmise that any defects in tension generation will have important consequences for bladder function. In the former case, too much tension will enhance signaling well before the bladder is full, which are symptoms typical of overactive bladder, while the latter condition will lead to bladder overfilling before a signal is sent. These advances highlight a new potential direction in understanding bladder pathology and informing experimental directions.

VII. SUMMARY

• 1. Bound by urine at one surface, and the lamina propria and muscularis at the other, the stratified urothelium lines the renal pelvis, ureters, bladder, and the proximal urethra where it forms the blood-urine barrier. While the urothelium is superficially similar throughout the lower urinary tract, and comprised of umbrella, intermediate, and basal cell layers, its embryonic origins and other properties (e.g., expression of differentiation markers) vary with location.
• 2. In many species, including humans, the outermost umbrella cell layer is multinucleate and polyploid. The immediate precursor of the umbrella cell is a small pool of binucleate intermediate cells (2n + 2n) that give rise to polyploid (4n + 4n) umbrella cells through a process called endoreplication. In turn, the binucleate intermediate cells are derived from a population of mononucleate uroplakin⁺ intermediate cells that undergo incomplete cytokinesis.
• 3. In response to relatively minor injuries, binucleate intermediate cells serve as progenitors to repopulate the umbrella cell layer. However, in response to extensive urothelial injury, KRT5⁺ basal cells, and most likely the KRT14⁺ subpopulation of these basal cells, serve as stem cells to repopulate the entire urothelium.
• 4. The urothelial barrier is multifactorial and includes a glycocalyx, comprised primarily of glycoconjugates associated with MUC1, a low permeability apical membrane that is highly deformable, and high-resistance tight junctions that are composed in part of multiple claudin isoforms.
• 5. The composition of urine is different between the renal pelvis and bladder. Furthermore, in response to overdistension of the bladder, the urothelium can reabsorb water and solutes from the urine. Thus the urothelium is not a simple barrier, but one that can actively contribute to water and solute homeostasis.
• 6. The mucosa and associated urothelium undergo large morphological and functional transitions to accommodate lumen expansion and urine storage. These adaptations include unfolding of mucosal rugae; unfurling of the collagen matrix; and the RAB-, actin cytoskeleton-, and SNARE-dependent exocytosis of DFVs, leading to a dramatic increase in umbrella cell apical surface area. Concomitantly, the umbrella cell’s junctional ring expands, further promoting maximal expansion of the urinary tract organ lumen.
• 7. The process is reversed by refolding of the rugae and collagen matrix, and endocytosis of excess apical membrane and junction-associated proteins.
• 8. In addition, the urothelium actively senses its mechanochemical environment and communicates this to subjacent tissues, including afferent nerve processes, promoting proper lower urinary tract function. These functions are made possible by urothelial input pathways, a large family of channels, receptors, and other proteins that allow the urothelium to sense its local environment. While the downstream signal transduction pathways are poorly understood, the urothelium releases several mediators (so-called output pathways), which allow it to communicate with other tissues in the subjacent tissues.
• 9. Some urothelial genes associated with mechanotransduction and mediator release (e.g., Piezo1, Gjb2, TRPV4, and Slc17a9) exhibit circadian rhythmicity. It is hypothesized that defects in their diurnal expression contribute to bladder conditions such as enuresis and nocturia.
• 10. While several mouse KO models exhibit defects in lower urinary tract function (see TABLE 4), the lack of conditional urothelial-specific knockout mice makes it difficult to understand what contribution the urothelium is making, if any, to these phenotypes.
• 11. Exciting new culture models, transfection/transduction technologies, mucosal preparations, imaging technologies, and mathematical modeling are moving urothelial research forward.

GRANTS

This work was supported by the Urology Care Foundation Research Scholar Award Program (to M. Dalghi), National Institutes of Health Grants R01 DK119183 (to G. Apodaca and M. D. Carattino) and R01DK104287 (to G. Apodaca), and the Kidney Imaging Core and Physiology and Model Systems Core of the Pittsburgh Center for Kidney Research (P30DK079307).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.