Proximal tubular injury and rapid formation of atubular glomeruli in mice with unilateral ureteral obstruction: a new look at an old model

Abstract

Unilateral ureteral obstruction (UUO), employed extensively as a model of progressive renal interstitial fibrosis, results in rapid parenchymal deterioration. Atubular glomeruli are formed in many renal disorders, but their identification has been limited by labor-intensive available techniques. The formation of atubular glomeruli was therefore investigated in adult male mice subjected to complete UUO under general anesthesia. In this species, the urinary pole of Bowman's capsule is normally lined by tall parietal epithelial cells similar to those of the proximal tubule, and both avidly bind Lotus tetragonolobus lectin. Following UUO, these cells became flattened, lost their affinity for Lotus lectin, and no longer generated superoxide (revealed by nitroblue tetrazolium infusion). Based on Lotus lectin staining, stereological measurements, and serial section analysis, over 80% of glomeruli underwent marked transformation after 14 days of UUO. The glomerulotubular junction became stenotic and atrophic due to cell death by apoptosis and autophagy, with concomitant remodeling of Bowman's capsule to form atubular glomeruli. In this degenerative process, transformed epithelial cells sealing the urinary pole expressed α-smooth muscle actin, vimentin, and nestin. Although atubular glomeruli remained perfused, renin immunostaining was markedly increased along afferent arterioles, and associated maculae densae disappeared. Numerous progressive kidney disorders, including diabetic nephropathy, are characterized by the formation of atubular glomeruli. The rapidity with which glomerulotubular junctions degenerate, coupled with Lotus lectin as a marker of glomerular integrity, points to new investigative uses for the model of murine UUO focusing on mechanisms of epithelial cell injury and remodeling in addition to fibrogenesis.

regardless of its etiology, progressive renal disease follows a predictable course leading ultimately to renal parenchymal loss and interstitial fibrosis (57). Over the past decade, attention has focused on excessive interstitial extracellular matrix as playing a central role in this process (57). The associated cellular events, including tubular atrophy, interstitial macrophage infiltration, and myofibroblast accumulation, have been elucidated through the study of the renal response to unilateral ureteral obstruction (UUO) (10). Developed several decades ago (37), this model leads to progressive renal parenchymal injury very similar to that developing in many clinical renal disorders and has the advantages of localizing the lesions to one kidney without surgical renal ablation or systemic effects resulting from toxins (10). Consequent to the development of numerous targeted mutants, the mouse has become the most widely used species (1), and a literature search for “ureteral obstruction or UUO” and “mice” for the past 10 yr yields over 280 papers (Medline). In the majority of these studies, interstitial fibrosis was used as the primary end point, the goal being attenuation or reversal of interstitial collagen accumulation.

Using a model of partial UUO in the neonatal mouse, we recently reported another pathway for functional nephron loss: the formation of atubular glomeruli (ATG) (52). We described similar lesions arising spontaneously in kidneys of mice with deletion of endothelial nitric oxide synthase (eNOS), in which there is progressive destruction of the glomerulotubular junction (16). In a comprehensive review of the ATG literature to date, we determined that selective injury to the glomerulotubular junction develops in a wide spectrum of clinical renal disorders, including renal artery stenosis, proteinuric type I diabetes, IgA and membranous nephropathy, obstructive nephropathy, pyelonephritis, and renal allograft rejection (9). Notably, a more recent study of type 2 diabetics revealed significant numbers of ATG even in patients with low levels of proteinuria (54).

While there is significant evidence for the importance of glomerulotubular disruption in many renal disorders, this pathway remains underappreciated, attributable in large part to the difficulties inherent in demonstrating the presence of ATG in kidney tissue (microdissection or serial sectioning) (9). In the present study of the adult mouse, we demonstrate that destruction of the glomerulotubular junction and formation of ATG develop in 80% of nephrons after only 2 wk of ureteral obstruction. In addition, we present a method by which rapid quantitation of glomerulotubular degeneration can be made.

MATERIALS AND METHODS

Experimental animals and surgical procedures.

Male mice of the C57Bl strain were subjected to complete UUO (30 animals) or sham operation (10 animals) at ∼6 wk of age. All surgery was performed in accordance with an animal protocol approved by the University of Virginia Animal Care and Use Committee. Animals were anesthetized with isoflurane plus oxygen, and the left ureter was exposed through a flank incision. In animals undergoing UUO, the ureter was ligated with 8–0 nylon, while in sham-operated mice the ureter was left undisturbed.

Tissue collection and processing.

Mice were studied 1 (n = 3 UUO), 3 (n = 5 UUO), 7 (n = 10 UUO + 5 sham), and 14 days (n = 12 UUO + 5 sham) after operation. All animals were injected with Euthasol solution (Virbac, Fort Worth, TX). In the majority of animals, kidneys were fixed by immersion or perfusion with 10% phosphate-buffered formalin, embedded in paraffin, and sagittally sectioned at thicknesses ranging from 2 to 10 μm. In some cases, kidneys were perfused with a solution of 1.5% glutaraldehyde in a solution of 3% dextrose, 3% dextran (43,500 avg. MW), and 50 mM CaCl₂. These were cut into 50-μm coronal sections and embedded in plastic by a technique described previously (15). “Semithin” sections (∼0.25 μm) were stained with toluidine blue; in some instances, complete series of semithin sections were examined to confirm that glomerular profiles were “atubular,” that is, had lost their connection to proximal tubules. Ultrathin sections (80–100 nm) were prepared from selected specimens and examined in a Zeiss EM 10CA transmission electron microscope.

Staining.

Fragmented DNA was detected using Apoptag (Chemicon, Temecula, CA), a method based on the TdT-mediated dUTP nick end labeling reaction (20). For immunohistochemistry, sections were pretreated to quench endogenous peroxidase (H₂O₂ in methanol) and endogenous biotin (Avidin-Biotin blocking kit, Vector Laboratories, Burlingame, CA).

Lotus tetragonolobus (asparagus pea) lectin (Vector Laboratories) marks proximal tubule epithelial cells in mouse and human kidney (24, 48). Paraffin sections of formalin-fixed kidney were treated by this staining procedure, which incorporated proteinase K enzymatic digestion before exposure to biotinylated lectin (1:50) and development by the ABC-DAB regimen. Additional immunohistochemical staining procedures utilized antibodies against phospho-histone H3 (ser 10; Cell Signaling Technology, Beverly, MA) at 1:200 dilution for detection of mitotic cells, LC3B (Cell Signaling Technology, Danvers, MA) at 1:200 dilution and beclin (BD Transduction Laboratories, San Diego, CA) at 1:500 dilution for demonstration of autophagy, vimentin (Abcam ab45939, Abcam, Cambridge, MA) at 1:2,100 dilution, α-smooth muscle actin (α-SMA; Sigma #A-2547, St. Louis, MO) at a dilution of 1:800, nestin (Chemicon International #MAB353) at 1:200 dilution, and neuronal NOS (NOS type I: Abcam ab76067) at 1:500 dilution to identify cells of the macula densa. Renin (goat anti-rabbit) antibody (a gift from Dr. Tadashi Inagami of Vanderbilt University) was used at a dilution of 1:10,000.

Oxidative damage.

Perfusion with a 1-mg/ml solution of nitroblue tetrazolium (NBT) in HBSS was carried out following the procedure of Chien et al. (11). In the presence of superoxide, NBT is reduced to insoluble blue formazan crystals. To demonstrate superoxide distribution, the perfused kidneys were fixed in formalin, embedded in paraffin, and sections 4–10 μm were cut and counterstained with 0.1% neutral red.

For examination of vascular patency, 2-μm-thick paraffin sections of equivalent 14-day UUO kidneys, fixed with phosphate-buffered formalin either by immersion or perfusion, were stained with the use of a trichrome staining kit (HT 15 Accustain Trichrome Kit, Sigma Diagnostics, St. Louis, MO). By extending the wash following the Biebrich Scarlet/acid fuchsin stainining step, bright red staining is diminished in the parenchyma but left intact in the erythrocytes.

Quantitation.

Glomerular areas in sham-operated and 14-day UUO contralateral and obstructed kidneys were calculated following a method used previously for rat kidney (56). These area measurements (see Table 2) were used to calculate average glomerular diameters for the purpose of serial section analysis (see Glomerulotubular scoring below).

Table 2.

Glomerular area (μm²) 14 days following sham operation or UUO

Procedure	Results
Sham	4,603 ± 524
Contralateral to UUO	6,198 ± 186*
Obstructed	3,354 ± 331*†

Data are means ± SD.

^*P < 0.05 vs. sham.

^†P < 0.05 vs. contralateral.

The phenomenon of glomerular disconnection from proximal tubules was examined in Lotus lectin-stained kidney sections as described previously (52). In the adult mouse, lectin-staining cuboidal epithelial cells normally form part of Bowman's capsule, but disappear in atubular glomeruli, the capsules of which consist entirely of epithelial cells that lack lectin staining. Using serial sections, we previously showed, in mice with partial neonatal UUO, that the presence of Lotus lectin-positive glomerular capsules is an indicator of intact glomerulotubular junctions (52). To examine this in adult animals, a single Lotus-stained median sagittal section of the contralateral and obstructed kidney was examined from each animal (7-day and 14-day UUO: 7 animals in each category); all glomeruli were counted and scored on the basis of the presence or absence of lectin staining (whether in the capsule or in clearly connected proximal tubules). Regardless of the extent of staining, any Lotus-positive glomerular profiles were scored as positive, and those lacking visible staining were scored as negative. Results were expressed as the percent of lectin-positive glomeruli.

The occurrence in Bowman's capsule of tall parietal epithelial cells (PECs) indistinguishable from proximal tubule epithelial cells has been known for well over a century (2). Sexual dimorphism has also been documented in adult mouse glomeruli, with tall PECs being substantially more numerous in males than females (13). To test this, an additional four female mice (9–10 mo of age) were also examined after Lotus staining, and sexual dimorphism was confirmed. Therefore, only males were used for the portion of this study pertaining to the effects of UUO in adult mice.

Glomerulotubular scoring.

The correlation between Lotus staining and glomerulotubular degeneration was also investigated by use of a more direct method. On the basis of glomerular area measurements (see Table 2) made on adult mice (sham, 14-day UUO contralateral, and 14-day UUO obstructed), and presuming glomeruli to be spherical, average glomerular diameter was calculated (sham = 76 μm, UUO-contralateral = 89 μm, UUO-obstructed = 64 μm). In 4-μm sections, 24–26 serial sections are therefore sufficient to capture entire glomeruli in the obstructed kidney.

In a median sagittal section, ∼125 glomerular profiles are present in an obstructed kidney, but not all are suitable candidates since the section may not pass through their centers.

Categorization of nephrons was based on morphology of their glomerulotubular junctions. As defined for rat (3, 50), normal glomerulotubular junctions are connected to patent proximal tubules composed of typical tall proximal tubule epithelial cells, whereas “atrophic” glomerulotubular junctions constitute the broadest morphological category, including constricted glomerulotubular necks as well as examples in which the capsules and necks are composed of flattened epithelial cells. The unifying feature of all atrophic glomerulotubular junctions is the presence of a connected (but structurally altered) proximal tubule, the lumen of which can be either patent or closed. ATG, in addition to having no demonstrable connection to a proximal tubule, have capsules composed entirely of flat PECs.

In a section series of obstructed kidneys from each of three animals, sections at the midpoint of the series (section #12 or 13) were examined, and all candidate glomeruli identified there were traced backwards and forwards through the series. Glomeruli in the normal and atrophic categories were readily identified by the presence of connected proximal tubules, the morphology of which determined their category. ATG were those that lay entirely within the section series and showed no connection to a proximal tubule. A total of 125 glomeruli were scored by this regimen.

Statistical analysis.

The SigmaStat program v. 3.0 (Aspire Software International, Ashburn, VA) was used. Parameters were compared between contralateral and obstructed kidneys for each experimental category (7 day, 14 day) using Student's t-test for paired observations. Comparisons between groups were made using Kruskal-Wallis one-way ANOVA on ranks with pairwise multiple comparisons by Dunn's method. Statistical significance was defined as P < 0.05.

RESULTS

In the normal Bowman's capsule, tall PECs surround the urinary pole and extend into the proximal tubule at the glomerulotubular junction; toward the vascular pole, these cells make a sharp transition to flattened PECs (Fig. 1a). The tall PECs of the capsule and proximal tubular cells avidly bind Lotus tetragonolobus lectin (Fig. 1b). NBT perfusion results in the deposition of formazan crystals, denoting superoxide production concentrated at the basal surfaces of tall epithelial cells in the capsule and connected proximal tubule (Fig. 1c).

Fig. 1.

Mouse kidney structure compared in 0.25-μm “semithin” plastic sections (a, d, g), 4-μm paraffin sections with Lotus tetragonolobus immunostaining (b, e, h, i), and 10-μm sections of nitroblue tetrazolium (NBT)-perfused material (c, f, j). a: Sagittal section through glomerulus which joins with its proximal tubule (PT) at the glomerulotubular junction (GTJ). The PT is composed of tall epithelial cells, continuous with cells (arrows), similar in appearance, which form a substantial portion of Bowman's capsule. The remainder of the capsule is composed of flattened parietal epithelial cells (PECs). b: Similarly oriented glomerulus showing that Lotus lectin stains the tall epithelial cells of both PT and Bowman's capsule, but leaves the flat PECs unstained. c: In this nephron from an unobstructed (contralateral) kidney, NBT staining for the presence of superoxide produces a distinct outlining (arrowheads) concentrated at the basal surfaces of the tall epithelial cells of Bowman's capsule and the adjacent PT. d: Ureteral obstruction leads to rapid and progressive degeneration of nephrons, characterized by shift in the morphology of capsule cells from tall to flattened (arrows) and atrophy of the GTJ and contiguous PT. e: Epithelial cells of the capsule, GTJ, and initial PT segment lose their affinity for Lotus lectin. f: NBT reaction product disappears from the glomerulus and becomes localized to the luminal side of epithelial cells of the downstream PT. g-j: Atubular glomeruli, traced in serial consecutive sections to confirm the lack of attached PTs. In g, capsules now consist entirely of flattened PECs, but the internal glomerular components (podocytes, mesangium, and capillaries) exhibit normal morphology. Two tubule profiles (*) were followed and found to be disconnected, isolated segments that had lost their lumina and were composed of epithelial cells containing numerous autophagic bodies (also see Fig. 3, g and h). h and i: Loss of the tall cell configuration of PECs results in glomerular capsules devoid of Lotus staining. j: Although nearby fragments of PTs are heavily stained by NBT perfusion, the atubular glomerulus shows no evidence of superoxide production. Scale bar in j = 50 μm and applies to all panels.

Following 14 days of UUO, tall PECs become flattened, and similar cells extend down the initial segment of proximal tubule, which becomes atrophied and collapsed (Fig. 1d). Lotus staining is lost from the glomerular capsule, but is retained further down the tubule (Fig. 1e). A similar localization of formazan crystals is present, having disappeared from Bowman's capsule and now concentrated in the luminal portions of downstream proximal tubules (Fig. 1f). In obstructed kidneys, numerous glomeruli have capsules entirely composed of flat cells, but have normal appearing glomerular tufts (Fig. 1g). Such glomerular profiles lack Lotus staining altogether, and many can be proven by serial section analysis to be atubular (Fig. 1, h and i). NBT staining is absent from these ATG (Fig. 1j).

Lotus staining patterns remain constant in the contralateral kidneys, as reflected in the relative percentages of Lotus-positive capsules (Fig. 2). Two quantitative methods were employed to measure nephron degeneration in the obstructed kidney: 1) the percentage of Lotus-positive glomerular capsules was determined in a single median sagittal section of each kidney examined (Fig. 2) and 2) the distribution of “normal,” “atrophic tubule,” and “atubular” glomeruli was determined by serial sectioning (Table 1). In 14-day animals, close correlation was found between these two methods, such that Lotus-positive glomeruli (20%) correspond to the fraction of “normal” glomeruli measured in serial sections of obstructed kidneys (15%), and Lotus-negative glomeruli (80%) correspond to the sum of glomeruli with atrophic tubules plus ATG (85%; Fig. 2c, Table 1). Compared with sham-operated mice, mean glomerular size is significantly reduced in the obstructed kidney and increased in the contralateral kidney, reflecting adaptive growth (Table 2).

Fig. 2.

Progressive loss of glomerulotubular integrity resulting from unilateral ureteral obstruction (UUO) as indicated by Lotus staining. a: 14-day obstructed kidney. b: 14-day contralateral kidney. Staining in Bowman's capsules or the connected PTs (arrowheads) is severely diminished in the obstructed kidney, in which the majority of glomeruli appear as lectin-negative profiles (*). Scale bar in b = 250 μm and applies to a and b. c: Fractional distribution of Lotus-positive glomeruli expressed as a percentage of total glomeruli counted in a single sagittal kidney section of obstructed and contralateral kidneys of each animal (n = 7 in each category). The 37% reduction in Lotus-positive glomerular staining from 7 to 14 days of UUO reflects the rapidity with which glomerulotubular integrity is lost in this model of renal injury (filled bars = obstructed kidney; open bars = contralateral kidney; means ± SE; *P < 0.05 vs. contralateral kidney; #P < 0.05 vs. 14-day UUO kidney).

Table 1.

Distribution of glomerulotubular status determined by serial sectioning of obstructed kidneys from mice following 14 days of UUO

Number of Glomeruli per Mouse	Normal Glomerulotubular Junction	Atrophic Tubules at Glomerulotubular Junction	Atubular Glomeruli
45	22.2%	37.8%	40.0%
39	10.3%	59.0%	30.8%
41	12.2%	41.5%	46.3%
Means ± SD	14.9 ± 6.4%	46.1 ± 11.3%	39.0 ± 7.8%

UUO, unilateral ureteral obstruction.

Apoptotic nuclei appear in Bowman's capsule after a single day of UUO, limited primarily to flat PECs (Fig. 3a). By 3 days (Fig. 3, b and c), and increasing by 7 days of UUO (Fig. 3, d and e), both apoptotic and necrotic cells appear in the tall epithelium of the capsule and proximal tubules. Phospho-histone immunostaining showed no evidence of active mitosis in PECs or proximal tubule epithelium (data not shown). Thus, cell death at the glomerulotubular junction is not countered by a proliferative response.

Fig. 3.

TdT-mediated dUTP nick end labeling positivity identifies cell death in 1-day (a), 3-day (b and c), and 7-day (d, e) obstructed kidney. a: Apoptosis (arrows) is initiated in the flat PECs at the vascular pole of Bowman's capsule, but it shifts to tall PECs at the urinary pole by 3 days (b). c: Details in a semithin plastic section of an apoptotic tall PEC (arrow). d and e: Apoptotic figures continue to appear in both PECs and PT epithelial cells at 7 days. Autophagy is also present in PTs (f, g, h). f: Epithelial cells of a collapsed PT (between arrows) stain strongly for the autophagy-related protein LC3B. Inset: inclusions in a PT stain positively for beclin, another protein involved in autophagy (arrowheads). g: Transmission electron micrograph of a collapsed PT in cross section, demonstrating the lack of a lumen and the presence of autolysosomes (arrows). Nuclei (N) in these cells have a normal appearance, and the cytoplasm is unremarkable except for the numerous large inclusions and a paucity of mitochondria. h: Semithin plastic section of a glomerulus in which the GTJ and initial segment of the PT have transformed into a solid cord of dark cells that joins (at *) with a tubule segment that, although collapsed, still consists of large epithelial cells containing autophagic inclusions (arrows). Scale bar in e = 50 μm and applies to a-e. Scale bars in f = 50 μm, in g = 10 μm, and in h = 50 μm.

In addition to apoptosis and necrosis, autophagy has been recently recognized to play a role in cellular dynamics following renal injury. Autophagy, the process by which damaged macromolecules and organelles are sequestered in autophagosomes for regulated degradation, is controlled in part by the proteins beclin-1 and LC3. Deterioration of proximal tubules is accompanied by immunostaining for both proteins (Fig. 3f) and presence of typical autolysosomal bodies (double-membrane vacuoles; Fig. 3g). Autophagic activity is generally absent from the PECs and stenosed glomerulotubular junction, with abrupt transition to a proximal tubule segment of tall epithelial cells containing autolysosomes (Fig. 3h).

The progression of glomerulotubular injury was examined further by immunostaining for proteins associated with progenitor cells and regeneration: vimentin, nestin, and α-SMA. In sham-operated and contralateral kidneys, vimentin is restricted to podocytes and blood vessels (Fig. 4a), but it is strongly expressed in capsules, glomerulotubular junctions, and atrophied proximal tubules of degenerating glomeruli in the obstructed kidney (Fig. 4b). Under normal conditions, nestin is restricted to podocytes, but following UUO, nestin is also present in PECs and atrophied proximal tubules (Fig. 4c). A contractile protein expressed by vascular smooth muscle cells and renal interstitial cells following UUO, α-SMA, also appeared in PECs of the obstructed kidney and was particularly prominent in cells that appeared to be sealing off the glomerulotubular junction (Fig. 4d).

Fig. 4.

Immunostaining for vimentin (a, b), nestin (c), and α-smooth muscle actin (SMA; d) in nephrons from kidneys from animals with 14 days of UUO. a: In the contralateral (unobstructed) kidney, vimentin is present primarily in podocytes (POD) and blood vessels (BV). b: In the obstructed kidney, vimentin is expressed not only in POD and the afferent arteriole (AA) but also in the atrophied GTJ and detaching PT. c: Glomeruli in the obstructed kidney show strong nestin immunostaining both in POD and epithelial cells of the capsule (arrowheads) and the attached atrophic PT (*) d: Serial 2-μm section of the nephron in c, showing α-SMA positivity in capsule cells, particularly in a “shelf” of cells extending through the GTJ (between arrows). Scale bar in b = 50 μm and applies to a and b; scale bar in d = 50 μm and applies to c and d.

In the adult kidney, renin is normally restricted to the juxtaglomerular region, a distribution that persists in both kidneys through 7 days of UUO (Fig. 5, a and b). However, after 14 days of UUO, renin extended down the afferent arteriole of the obstructed kidney and nearly disappeared in the contralateral kidney (Fig. 5, c and d). The macula densa was associated with only some glomeruli of the obstructed kidney (Fig. 5e). Serial section analysis proved such glomeruli to be attached to normal or atrophic proximal tubules; ATG lacked any immunological or structural evidence of an associated macula densa. Contralateral kidneys showed the normal complement of maculae densae (Fig. 5f).

Fig. 5.

a–d: Renin staining for juxtaglomerular cells. a and b: After 7 days of UUO, renin expression is similar between the obstructed and contralateral kidneys. By 14 days (c, d), renin has become expressed along the AA of the obstructed kidney, but it is absent or only weakly expressed in the contralateral kidney. e and f: Neuronal nitric oxide synthase (nNOS) staining for macula densa (MD). Each of the 2 glomeruli in e was traced in serial sections to its connection both with an atrophic PT (example shown by PT) and a collapsed MD that retained its specific staining for nNOS. Atubular glomeruli lacked the MD, however. Typical morphology of the MD is seen in the contralateral kidney (f). g and h: Modified Masson trichrome staining to emphasize erythrocytes. Glomeruli in both the obstructed (g) and contralateral kidneys (h) display open capillaries and few erythrocytes (arrowheads) following fixation by vascular perfusion. Scale bar in h = 50 μm and applies to all panels.

To establish patency of glomerular capillaries in obstructed kidneys, sections of immersion fixed kidneys were compared with those fixed by vascular perfusion. In the latter, glomeruli in both contralateral and obstructed kidneys were characterized by open capillaries, with only a few trapped blood cells (arrowheads, Fig. 5, g and h). This, together with the normal appearing structure of intraglomerular components (Fig. 1G), indicates that there is continued blood flow to glomeruli in the obstructed kidney despite the formation of ATG.

DISCUSSION

Surgical ureteral ligation in the adult mouse has become the most widely used animal model for studying mechanisms responsible for progressive renal disease. While most reports focus on renal interstitial cellular changes and fibrosis, the glomerular and tubular epithelium is now receiving increased attention as the primary target of progressive renal injury (10, 40). The present study concentrates on proximal tubular injury following UUO in the mouse, demonstrating rapid and dramatic injury to this segment of the nephron. In addition to widespread proximal tubular cell death, however, Bowman's capsule undergoes significant remodeling, preserving its integrity by sealing the urinary pole with flattened epithelial cells. Glomerular perfusion continues and the number of renin-containing cells is increased. These findings are consistent with a programmed remodeling process that may have an evolutionary origin.

Glomerular and tubular epithelial cell response to UUO.

Following UUO, glomerular PECs and adjacent proximal tubular cells undergo transition from a polarized columnar epithelium to a squamous configuration. Loss of Lotus staining of these transformed cells presumably reflects disruption of the apical tubulovesicular system, the likely target of Lotus affinity (Forbes MS, unpublished observations). There are significant species differences in the location of the transition from flat PECs lining Bowman's capsule to tall epithelial cells lining the proximal tubule (31). The point of transition varies from tall epithelial cells extending well onto the parietal surface of Bowman's capsule in the mouse to flat cells extending down the initial proximal tubule in the rabbit, with the transition occurring at the glomerulotubular junction in rat and human (31). Tall PECs are the product of phenotypic transformation (8), with transitional cells interposed between flat and tall PECs (14). Even within normal glomeruli, PECs vary in their complement of microvilli and cilia (40). Prominent PECs have been described in a spectrum of renal diseases (including diabetic nephropathy and acute glomerulonephritis) (18), while flat epithelial cells extend down the initial proximal tubule of 10% of nephrons in normal human renal tissue (33). In nephropathic cystinosis, proximal tubular cells undergo apoptosis with progressive flattening of the glomerulotubular junction, leading to the formation of ATG (29, 47, 51). Glomerular PECs and proximal tubular cells clearly retain significant phenotypic plasticity in response to a variety of stimuli.

Although UUO induces apoptosis in cells at the glomerulotubular junction, autophagy is prominent further downstream in tubule segments that retain their Lotus positivity even after becoming fragmented. Others recently reported tubular autophagy following UUO (32), and transgenic overexpression of TGF-β1 localized to the proximal tubule also induces extensive autophagy in this segment (27). Complete UUO in the rat leads to selective upregulation of TGF-β1 expression by proximal tubules, while TGF-β1 inhibition reduces UUO-induced tubular injury (17, 36). Although autophagy can contribute to cell survival (42), the stimulus of persistent UUO in the present study favors cell death.

Susceptibility of the glomerulotubular junction.

The glomerulotubular junction is uniquely susceptible to stress from renal injury comprising glomerular, tubulointerstitial, and toxic disorders (9). Chronic UUO increases oxidative stress and obstructive renal injury can be reduced by antioxidants (41). As shown in the present study, superoxide is normally produced at the basal surface of tall PECs of Bowman's capsule and contiguous proximal tubular epithelial cells. Following UUO, these cells become flattened, with loss of mitochondrial mass and disappearance of NBT staining, which is restricted to proximal tubular remnants. Thus, PECs lining Bowman's capsule of ATG no longer produce superoxide, possibly contributing to the survival of glomeruli despite the persistence of UUO.

Animal models resulting in the induction of ATG have included rats with passive Heymann nephritis, polycystic kidney disease, and remnant kidney (3, 19, 50). In these models, the duration required for development of glomerulotubular atrophy and ATG ranged from 10 wk to 8 mo, whereas in the mouse this requires only 2 wk of UUO. The model of murine UUO is therefore ideally suited to future studies of the cellular mechanisms leading to the formation of ATG. Patients with cystinosis (ages 10–24 yr) (29), type 1 (ages 25–47 yr) (38), or type 2 (ages 41–64 yr) diabetic nephropathy (54) also have ATG. High glucose stimulates the generation of reactive oxygen species in proximal tubular cells, and overexpression of catalase in diabetic mice reduces proximal tubular apoptosis, demonstrating the role of oxidant injury in this disease (6, 7). The course of diabetic nephropathy is accelerated in db/db mice with eNOS^−/− genetic background (58), and spontaneous formation of large numbers of ATG in eNOS knockout mice suggests that a similar process is at work in the diabetic mouse (16).

Function of ATG.

The presence of erythrocytes in glomerular capillaries and their displacement by perfusion fixation indicate that there is ongoing circulation in ATG. Collapse of the cortical thick ascending limb and disappearance of the macula densa in ATG may be related to enhanced renin production by the microvasculature of the obstructed kidney: the endocrine or paracrine function of ATG is not only maintained, but augmented. An increase in renin-containing cells has also been reported in ATG of kidneys from children with congenital urinary tract anomalies (28). Increased renin production by these damaged nephrons may accelerate glomerulotubular disconnection, as suggested by a reduction in the formation of ATG by angiotensin inhibition in rats with radiation nephropathy or Heymann nephritis (3, 12). In normal mice, PECs form tight junctions that provide a second glomerular barrier preventing the penetration of dextran tracers filtered by podocytes (39). However, following anti-glomerular basement membrane disease, tracers localize to Bowman's basement membrane and even to the extraglomerular space (39). Transplanted human kidneys contain ATG whose capsules are lined by podocytes rather than the normal flat PECs (21). Since podocytes are not known to proliferate, this suggests that PECs are transformed to podocytes in such ATG and may allow filtration to the extraglomerular space.

Despite morphologic evidence of ongoing perfusion, the reduced size of ATG from obstructed kidneys likely reflects reduced glomerular blood flow, which is a known consequence of prolonged tubular obstruction (49). A similar reduction in volume of ATG has been reported in eNOS knockout mice, rats with polycystic kidney disease or remnant kidneys, and patients with diabetes, renal artery stenosis, or pyelonephritis (16, 19, 34, 35, 50, 54). This suggests that perfusion and filtration are reduced compared with intact glomeruli regardless of the etiology of formation of ATG. Decreased glomerular size is therefore a useful adjunct to loss of Lotus lectin staining in the identification of ATG in the murine model.

Remodeling of ATG.

The expression of α-SMA by PECs and by cells that intrude into the damaged urinary pole of Bowman's capsule (Fig. 4d) presumably plays a role in the movement of cells contributing to restoration of a completely enclosed capsule. The glomerulotubular junction is a locus of renal progenitor cells: embryonic stem cells injected into mouse embryonic kidneys became localized to the glomerulotubular junction (26). Multipotent progenitor cells are present in Bowman's capsule of adult human kidneys, with PECs contributing to tubular regeneration at the urinary pole, and podocytes at the vascular pole (30, 44, 45). Demonstration of vimentin and nestin expression by PECs of the obstructed kidney, as shown in the present study, is consistent with their role as progenitor cells (55). While chronic UUO in the rat also increases proximal tubular nestin expression, an effect on PECs was not noted in that report (46). The molecular mechanisms responsible for phenotypic transition of glomerular epithelial cells in the formation of ATG remain to be explored in future studies. These may involve the TGF-β superfamily, mTOR signaling, or other pathways involved in kidney development, growth, or response to injury (5, 25).

ATG also develop in neonatal mice following 6 wk of partial UUO (52). Notably, relief of obstruction arrests the process of glomerulotubular disconnection, leading to remodeling of the renal architecture with resolution of interstitial fibrosis. The combination of injurious and regenerative processes in response to obstructive injury may represent the operation of Cannon's “fight-or-flight” reactions to stress, applied to the cellular level by Goligorsky (22). Persistent complete UUO leads to a relentless destruction of nephrons, whereas release of partial UUO tips the balance toward regeneration. As recently reviewed, the renal epithelial cell appears to play a central role in the response to acute renal injury, which can follow a path of either “failed differentiation and atrophy” or successful regeneration (53). This perspective, reinforced by the present study, suggests that therapeutic approaches should be directed to signaling events in epithelial cells, rather than to downstream signaling events in the interstitium (53).

The orchestrated process of glomerulotubular disconnection demonstrated in the present study may have its origins in a common ancestor shared by mammals and fish. Over the past 120 million years, teleost fish have developed aglomerular kidneys on three occasions, each separated by 50 million years, to better adapt to a marine environment (4). The nephrons of the sculpin, a contemporary fish, undergo progressive destruction of the glomerulotubular junction during maturation (23). Conversely, a number of fish (including the zebrafish) can develop new nephrons within weeks of injury (43). A better understanding of the mechanisms underlying these processes may lead to new approaches to optimizing regenerative therapies in the diseased mammalian kidney.

Coupled with the ease of detection (by the lack of Lotus staining) of ATG or glomeruli connected to atrophic tubules in the obstructed mouse kidney, the rapid evolution of injury makes this a highly useful model for the study of progressive glomerulotubular injury. It is likely that the formation of ATG has remained largely under-recognized because of the time and labor required to perform microdissection or serial sectioning to document them. In addition to measurement of renal interstitial fibroblast and collagen accumulation, quantitation of ATG in future studies may lead to novel therapies for slowing or reversing progressive human renal diseases.

GRANTS

This work was supported by National Institutes of Health Grant DK45179 and a grant from the University of Virginia Children's Hospital.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).