Intercellular adhesion molecules (ICAMs) and spermatogenesis

Abstract

BACKGROUND

During the seminiferous epithelial cycle, restructuring takes places at the Sertoli–Sertoli and Sertoli–germ cell interface to accommodate spermatogonia/spermatogonial stem cell renewal via mitosis, cell cycle progression and meiosis, spermiogenesis and spermiation since developing germ cells, in particular spermatids, move ‘up and down’ the seminiferous epithelium. Furthermore, preleptotene spermatocytes differentiated from type B spermatogonia residing at the basal compartment must traverse the blood–testis barrier (BTB) to enter the adluminal compartment to prepare for meiosis at Stage VIII of the epithelial cycle, a process also accompanied by the release of sperm at spermiation. These cellular events that take place at the opposite ends of the epithelium are co-ordinated by a functional axis designated the apical ectoplasmic specialization (ES)—BTB—basement membrane. However, the regulatory molecules that co-ordinate cellular events in this axis are not known.

METHODS

Literature was searched at http://www.pubmed.org and http://scholar.google.com to identify published findings regarding intercellular adhesion molecules (ICAMs) and the regulation of this axis.

RESULTS

Members of the ICAM family, namely ICAM-1 and ICAM-2, and the biologically active soluble ICAM-1 (sICAM-1) are the likely regulatory molecules that co-ordinate these events. sICAM-1 and ICAM-1 have antagonistic effects on the Sertoli cell tight junction-permeability barrier, involved in Sertoli cell BTB restructuring, whereas ICAM-2 is restricted to the apical ES, regulating spermatid adhesion during the epithelial cycle. Studies in other epithelia/endothelia on the role of the ICAM family in regulating cell movement are discussed and this information has been evaluated and integrated into studies of these proteins in the testis to create a hypothetical model, depicting how ICAMs regulate junction restructuring events during spermatogenesis.

CONCLUSIONS

ICAMs are crucial regulatory molecules of spermatogenesis. The proposed hypothetical model serves as a framework in designing functional experiments for future studies.

Introduction

In mammalian tissues, including the seminiferous epithelium of the adult testis and the epididymis, cell–cell and cell–matrix interactions are indispensable to multiple cellular events, such as spermatogenesis in the testis, sperm maturation in the epididymis, and maintenance of cell/tissue homeostasis and survival (Dym and Cavicchia, 1978; Parvinen, 1982; Mruk and Cheng, 2004b; Hess and de Franca, 2008; Hermo et al., 2010a, b; Cyr, 2011; Ivanov, 2011; Shen et al., 2011; Dube and Cyr, 2012; Taddei et al., 2012) (Fig. 1). In some cellular processes, such as cell migration and morphogenesis during development, which alter cell adhesive states continuously (Gardel et al., 2010; Parsons et al., 2010), maintenance of cell adhesion to neighboring cells and/or to the extracellular matrix (ECM) is crucial to maintain tissue homeostasis and functionality. In the testis, spermatogenesis (which includes the cellular events of mitosis, meiosis, spermiogenesis and spermiation), which takes place in the epithelium of the seminiferous tubule, also involves extensive restructuring of cell junctions because developing germ cells must traverse the epithelium associated with their extensive morphological changes (Leblond and Clermont, 1952b; Fawcett, 1975; Setchell, 1978; de Kretser and Kerr, 1988). Germ cells per se, unlike other mammalian cells such as fibroblasts, macrophages, neutrophils and metastatic cancer cells, are non-motile: instead, they rely on Sertoli cells for their movement migrating from near the basement membrane to the luminal edge of the adluminal compartment via extensive morphological transformation, and can be defined into 14, 12 and 6 stages of the seminiferous epithelial cycle of spermatogenesis in the rat, mouse and human, respectively (Leblond and Clermont, 1952a; Clermont, 1963, 1972; Parvinen, 1982; de Kretser and Kerr, 1988; Mruk and Cheng, 2004b; Amann, 2008; Hess and de Franca, 2008; Mruk et al., 2008). Furthermore, preleptotene spermatocytes transformed from type B spermatogonia residing in the basal compartment must traverse the blood–testis barrier (BTB), which is created by Sertoli cell–cell junctions near the basement membrane (Dym and Fawcett, 1970; Fawcett et al., 1970; Fawcett, 1975; Setchell and Waites, 1975; Setchell, 2008; Mital et al., 2011). Thus, it is conceivable that extensive restructuring and turnover take place at the Sertoli cell–cell and Sertoli–germ cell interface during spermatogenesis.

Figure 1.

A schematic diagram illustrating cellular events pertinent to transepithelial migration of developing germ cells during the epithelial cycle of spermatogenesis in the seminiferous epithelium of the adult rat testis. Adjacent Sertoli cells (green background, with the nucleus located near the basement membrane) extend from the tunica propria (consists of four layers: beginning with the basement membrane, type I collagen layer, the peritubular myoid cell layer and the lymphatic endothelium, from inside to outside of the seminiferous tubule) towards the tubule lumen in the rat testis. The seminiferous epithelium is divided by BTB into the basal and the adluminal (apical) compartments. Several co-existing junction types, namely tight junction (TJ), basal ES, and GJ, which together with the desmosome, constitute the BTB. However, these junctions undergo extensive restructuring at Stages VIII–IX to facilitate the transit of preleptotene spermatocytes residing at the basal compartment, crossing the BTB to enter the adluminal compartment while differentiating into leptotene spermatocytes to prepare for meiosis I/II. Before the ‘old’ BTB is being disassembled, Sertoli cells assemble a ‘new’ BTB behind transiting spermatocytes as shown in this diagram. Furthermore, developing germ cells, such as spermatogonia, spermatocytes and pre-step 8 spermatids attach to Sertoli cells via desmosomes and GJs, whereas elongating/elongated spermatids (step 8–19 spermatids) attach to Sertoli cells via the apical ES (typified by the presence of actin filament bundles sandwiched in between cisternae of endoplasmic reticulum and the apposed plasma members of the Sertoli cell and the spermatid) and must traverse the seminiferous epithelium moving from the basal to the adluminal compartment. Additionally, elongating spermatids also move ‘down’ towards the tunica propria, almost touching the Sertoli cell nuclei at Stage V of the epithelial cycle, before they move back to the adluminal compartment again and reach the luminal edge of the tubule lumen at Stages VI–VIII to prepare for spermiation (as shown by large gray arrows). At the same time of BTB restructuring at Stages VIII–IX, spermiation takes place at late Stage VIII to release the mature spermatozoa (step 19 spermatids) into the tubule lumen, with late-stage apical ES undergoing disintegration. In short, there are extensive restructuring at the Sertoli–Sertoli and Sertoli–germ cell interface during the seminiferous epithelial cycle of spermatogenesis.

In mammalian tissues, including the testis, cell adhesion is achieved via cell junctions, wherein cell adhesion molecules (CAMs) are crucial to cell adhesion in particular to elicit appropriate changes in cell adhesion in response to environmental stimuli (Byers et al., 1993a, b; Cheng and Mruk, 2002; Cavallaro and Dejana, 2011). CAMs are integral membrane proteins and each CAM is equipped with an extracellular, a transmembrane and a short cytoplasmic domain. In most cases, peripheral adaptors interact with the intracellular tail of CAMs and link them to the underlying cytoskeleton (Parsons et al., 2010; Gibson, 2011).

In mammalian cells, several major groups of CAMs have been identified, including cadherins, integrins and the immunoglobulin superfamily (IgSF) CAMs (Parsons et al., 2010; Cavallaro and Dejana, 2011; Gibson, 2011). In the testis, cadherins are the major components of the intercellular actin-based adherens junctions (AJs) connecting Sertoli–Sertoli and Sertoli–germ cells. They consist of classical cadherins, such as E-, P-, N- and VE cadherins, which associate with adaptors α-, β- or γ-catenin to constitute cell adhesion protein complexes. Cadherins/catenins are also one of the functional units of the actin-based cell–cell anchoring junctions known as ectoplasmic specialization (ES) (a testis-specific atypical AJ type) in the mammalian testis (note: the other functional units are the nectin-afadin and the integrin-laminin adhesion protein complexes) (Vogl et al., 1991, 1993, 2008; Cheng and Mruk, 2002, 2012 Wong et al., 2008; Yan et al., 2008). Cadherins are also components of desmosomes in the testis, which are intermediate filament-based cell–cell anchoring junctions, consisting of non-classical cadherins such as desmogleins and desmocollins (Russell and Peterson, 1985; Byers et al., 1993a; Lie et al., 2011; Mruk and Cheng, 2011, 2012). Recent studies have shown that there is extensive crosstalk between classical and non-classical cadherins in the testis, such as those at the BTB (Lie et al., 2011; Mruk and Cheng, 2011, 2012) (Fig. 1). In epithelial cells, integrins mainly connect ECM proteins (e.g. laminins, fibronectins and collagens) residing outside of a cell to the cytoskeleton within the cell (Luo et al., 2007; Weber et al., 2011). However, in the seminiferous epithelium of adult rat testes, integrins confer cell adhesion between the Sertoli and germ cells at the apical ES, as well as between the Sertoli cell and the basement membrane at the hemidesmosome (Cheng and Mruk, 2010). IgSF CAMs, probably the most populous family of cell surface molecules (with 765 members in humans), are a collection of integral membrane glycoproteins that possess at least one Ig-like domain in their extracellular region, capable of heterophilic and/or homophilic binding to integrins and/or among themselves (Brummendorf and Lemmon, 2001; Aricescu and Jones, 2007; Wang and Cheng, 2007; Cavallaro and Dejana, 2011). Its members include ICAM-1, sICAM-1, ICAM-2 and also CAR (coxsackievirus and adenovirus receptor) and nectin-3, which are also CAMs in the testis (see Table 1). Recent studies have shown that cell adhesion systems based on different CAMs are functionally inter-dependent rather than independent (Takai et al., 2008; Weber et al., 2011), illustrating considerable interplays between them. Besides direct heterophilic interactions between cadherins and integrins (Cepek et al., 1994; Whittard et al., 2002), as well as between integrins and IgSF CAMs, CAMs share common cytoskeletal networks (e.g. actin cytoskeleton), and there is accumulating evidence that they function beyond mere adhesion but are also involved in recruiting signaling molecules (e.g. non-receptor protein kinases, actin-binding proteins) and thus mediating multiple signal transduction cascades, regulating numerous cellular functions (Cavallaro and Dejana, 2011; Gibson, 2011).

Table I.

IgSF CAMs in the mammalian testis.

IgSF CAM	KO male	KO phenotype	Cellular localization in the testis	Regulator (s)	Putative binding partner(s)	Reference(s)
Ceacam1	Fertile	No apparent defects	SC/GC interface			Lauke et al. (2004); Kuespert et al. (2006)
Ceacam6			SC; interstitial cells		Vimentin	Kurio et al. (2011)
Ceacam6-L			Apical ES			Kurio et al. (2008)
Nectin-2	Infertile	Disrupted spermatid morphogenesis; disorganized nectin-3	BTB; apical ES (SC side)		Nectin-3	Bouchard et al. (2000); Inagaki et al. (2006)
Nectin-3	Infertile	Disrupted spermatid morphogenesis; disappearing localization of nectin-2 at the apical ES	Apical ES (GC side)		Afadin; Nectin-1, -2; Necl-1, -2, -5; sGCβ1, ICAM-2	Bouchard et al. (2000); Ozaki-Kuroda et al. (2002); Inagaki et al. (2006); Sarkar et al. (2006)
VSIG1			SC/GC interface (GC side)		ZO-1	Kim et al. (2010)
JAM-A	Subfertile	Sperm motility defects	BTB; GC	Gonadotropin, CNP	P-glycoprotein, sGCβ1, Dynamin II	Lie et al. (2006); Sarkar et al. (2006); Xia et al. (2007); Shao et al. (2008); Tarulli et al. (2008); Su et al. (2009)
JAM-B	Fertile	No apparent defects	BTB; apical ES (SC side)	IL-1α, TGF-β2	JAM-C	Gliki et al. (2004); Sakaguchi et al. (2006); Wang and Lui (2009)
JAM-C	Infertile	Lacked elongated spermatids; abnormal distribution of F-actin	Spermatocytes; round, elongating and elongated spermatids; apical ES (GC side)		Necl-2, Par3, Par6, CAR, JAM-B, Cdc42, PKCλ, PATJ	Gliki et al. (2004); Fujita et al. (2007); Mirza et al. (2006)
JAM-D	Fertile	No apparent defects	GC			Nagamatsu et al. (2006)
Necl-2	Infertile	Vacuoles in seminiferous epithelium; abnormal distribution of F-actin; disrupted spermatid morphogenesis	SC and spermatogonia/spermatocytes/elongating spermatids cell-cell interface; residual bodies		4.1G, Necl-5, Par3, JAM-C	Fujita et al. (2006); Fujita et al. (2007); Terada et al. (2010); Maekawa et al. (2011)
Necl-4					4.1G	Yang et al. (2011)
Necl-5			SC		Necl-2	Wakayama et al. (2007)
ICAM-1	Fertile	Granulocytosis	BTB; apical ES	IL-1, TNFα, IFN-γ, LPS, sICAM-1	See Table II	Sligh et al. (1993); Riccioli et al. (1995); De Cesaris et al. (1998)
ICAM-2	Fertile	Impaired angiogenesis; airway hyper-responsiveness	SC/GC interface		See Table II	Riccioli et al. (1995); De Cesaris et al. (1998); Gerwin et al. (1999); Huang et al. (2005)
VCAM-1	Fertile	No apparent defects	SC	IL-1, TNFα, IFN-γ, LPS	ICAM-1	Gurtner et al. (1995); Riccioli et al. (1995); De Cesaris et al. (1998)
NCAM	Fertile	No apparent defects	Gonocyte/SC interface (immature SC side); Leydig cells	T3	GFRα1	Cremer et al. (1994); Laslett et al. (2000); Orth et al. (2000); Yang and Han (2010)
ALCAM	Fertile	Axon fasciculation defects; retinal dysplasias	Gonocyte/SC interface (gonocyte side); SC; myoid cells	sALCAM		McKinnon et al. (2000); Ohbo et al. (2003); Ikeda and Quertermous (2004)
TCAM1	Fertile	No apparent defects	SC and spermatocyte/round spermatid cell-cell interface			Sakatani et al. (2000); Nalam et al. (2010)
CAR		Embryonic death by E12 day	SC/GC interface (GC side), including BTB and apical ES	FSH, TNFα	JAM-C, Vinculin, β-Catenin, c-Src	Lie et al. (2006); Mirza et al. (2006); Mirza et al. (2007); Wang et al. (2007)
Basigin	Infertile	Azoospermia; spermatogenesis arrested at the metaphase of the first meiosis; abnormal ES	SC, GC, Leydig cells		MCT1, MCT2, MMP-2	Igakura et al. (1998); Toyama et al. (1999); Yuasa et al. (2001); Chen et al. (2010); Chen et al. (2011); Mannowetz et al. (2012)

Ceacam, carcinoembryonic antigen-related cell adhesion molecule; JAM, junctional adhesion molecule; VSIG1, V-set and Ig domain containing 1; Necl, nectin-like molecule; ICAM, intercellular adhesion molecule; VCAM, vascular cell adhesion molecule; NCAM, neural cell adhesion molecule; ALCAM, activated leukocyte cell adhesion molecule; TCAM, testicular cell adhesion molecule; CAR, coxsackievirus and adenovirus receptor; LPS, lipopolysaccharide; IFN, interferon; JNK/SAPK, c-Jun N-terminal kinase/stress-activated protein kinase; T3, thyroid hormone; CNP, C-type natriuretic peptide; sGCβ1, soluble guanylate cyclase β1; MCT, monocarboxylate transporter; Par3, partitioning-defective protein 3; c-Src, cellular transforming protein of Rous sarcoma virus; apical ES, apical ectoplasmic specialization (a testis-specific actin filament rich-adherens junction, AJ); TGF, transforming growth factor; IL-1, interleukin 1; ZO-1, zonula occludens-1; PALS1, protein associated with Lin Seven 1; PATJ, PALS1-associated tight junction protein; MMP, matrix metalloproteinase; SC, Sertoli cell(s); GC, germ cell(s).

In this review, we focus on the roles of intercellular adhesion molecule (ICAM) family (Fig. 2) in junction dynamics with emphasis on junction restructuring events in the seminiferous epithelium of the testis; however, studies in other epithelia and/or endothelia are also discussed since this information is helpful to us to better understand the role of ICAMs in spermatogenesis. However, other adhesion proteins pertinent to the regulation of spermatogenesis in the testis are not discussed herein since they have recently been reviewed and discussed elsewhere, and readers can refer to these earlier excellent reviews (Setchell, 2008; Morrow et al., 2010; Mital et al., 2011; Pelletier, 2011; Franca et al., 2012). The ICAM family is a paradigmatic subgroup of the IgSF CAM family that are known to be expressed in a number of epithelial and endothelial cells, fibroblasts, neurons, leukocytes, platelets, erythrocytes, as well as testicular cells (Aricescu and Jones, 2007; Wang and Cheng, 2007; Kasprzak et al., 2012; Thomas and Baumgart, 2012) (Tables I and II). IgSF CAMs are known to take part in various biological processes with widely studied examples, such as neural cell adhesion molecule (NCAM) in neural development (Maness and Schachner, 2007) and ICAM-1 in transendothelial/transepithelial migration (TEM) of leukocytes during inflammation (Rahman and Fazal, 2009; Long, 2011). To date, several IgSF CAMs have been identified to be differentially expressed during spermatogenesis (see Table I) and it is increasingly clear that their loss and/or defects may cause infertility in both animals and humans (Bouchard et al., 2000; Yuasa et al., 2001; Maekawa et al., 2011). However, their roles at cell junctions and the underlying mechanism(s) that regulates cell adhesion in the testis during spermatogenesis remain elusive. Herein we provide an update on the possible roles of ICAM family members, in particular ICAM-1 and -2, during spermatogenesis based on recent findings in the field (Riccioli et al., 1995; De Cesaris et al., 1998; Xiao et al., 2012a, b). We also provide a hypothetical model regarding the mechanism(s) by which ICAMs regulate junction restructuring events during spermatogenesis. This model was prepared based on recent studies in the testis (Riccioli et al., 1995; De Cesaris et al., 1998; Xiao et al., 2012a, b) and other studies in the field that illustrate their roles in transmigration of leukocytes across the endothelial/epithelial barrier. This model thus serves as a framework upon which functional experiments can be designed. This review also sheds some lights on the role of ICAMs in fertility in men.

Figure 2.

Structure of ICAMs. Schematic drawing of the five ICAM family members illustrating different numbers of the extracellular Ig-like domains (D).

Table II.

Structure and function of ICAMs in mammalian cells/tissues.

ICAMs	Structural featuresa	Cellular expression	Inducibility	Binding partnersb	Selected references
ICAM-1 (CD54)	76–114 kDa, D1–D5, 8 N-glycans	Broadly and differentially expressed: Sertoli cells, germ cells, epithelial cells, endothelial cells, leukocytes (low basal level), fibroblasts	Highly inducible: TNFα, IFN-γ, IL-1α, IL-1β (α is more potent), LPS, phorbol ester, shear stress (by blood flow)	Occludin, ZO-1, N-cadherin, β-catenin, αLβ2, αMβ2, αXβ2 integrins, (s)ICAM-1, VCAM-1, fibrinogen, CD9, CD43, SHP-2, Trio, PIP2, Actin, GAPDH, β-tubulin, α-actinin, cortactin, filamins, ezrin, moesin, PKCs, SGEF, Src, Pyk2	Doulet et al. (2006); Federici et al. (1996); Pluskota et al. (2000); Muro et al. (2003); Millan et al. (2006); van Buul et al. (2007); Kanters et al. (2008); Di Lorenzo et al. (2011); van Rijssel et al. (2012); Xiao et al. (2012b)
ICAM-2 (CD102)	55–60 kDa, D1–D2, 6 N-glycans	Broadly and stably expressed: Sertoli cells, germ cells, epithelial cells, endothelial cells (high basal level), leukocytes, platelets	Non-inducible: Down-regulated by TNFα/IL-1β in HUVECs	Nectin-3, afadin, β1, αLβ2, αMβ2 integrins, (s)ICAM-2, DC-SIGN, actin, α-actinin, ezrin, radixin, moesin, PIP2, c-Src, Pyk2, annexin II	Geijtenbeek et al. (2000a); Furutani et al. (2007); Yoon et al. (2008); Xiao et al. (2012b)
ICAM-3 (CD50)	110–160 kDa, D1–D5, 15 N-glycans	Specifically and stably expressed: Leukocytes (high basal level)	Inducible: TNFα, retinoic acid, on endothelial cells (inflammation/ pathogenesis); down-regulated by PMA and Ca2+ ionophore	αLβ2, αDβ2 integrins, DC-SIGN, (s)ICAM-3, ezrin, moesin, Lck, Fyn, Kidins220	Juan et al. (1994); Geijtenbeek et al. (2000b); Serrador et al., 2002; Jean-Mairet et al. (2011)
ICAM-4 (CD242, LW)	42 kDa, D1–D2, 4 N-glycans	Specifically expressed: Erythrocytes	Not determined	αLβ2, αMβ2, αXβ2, α4β1, αVβ1, αIIbβ3, αVβ3, αVβ5 integrins	Spring et al. (2001); Hermand et al. (2003); Zennadi et al. (2004); Ihanus et al. (2007)
ICAM-5 (TLCN)	130 kDa, D1–D9, 15 N-glycans	Specifically expressed: Neurons	Non-inducible	β1, αLβ2 integrins, (s)ICAM-5, α-actinin, ezrin, radixin, moesin, presenilin 1 and 2, UOL	Annaert et al. (2001); Nyman-Huttunen et al. (2006); Furutani et al. (2007); Conant et al. (2011); Tse et al. (2009)

^aStructural features of each ICAM in humans and/or rodents.

^bLFA-1, leukocyte-specific β2-integrin subfamily consists of four family members: LFA-1, also known as CD11a/CD18 and αLβ2); Mac-1, macrophage-1 antigen, also known as CD11b/CD18 and αMβ2; p150, 95, also known as CD11c/CD18 and αXβ2, and CD11d/CD18 (αDβ2); CD, cluster of differentiation (according to the CD nomenclature of leukocyte antigens); D, Ig-like domain; N-glycans, potential N-glycosylation sites; LW, Landsteiner–Wiener (LW) blood group antigen; TLCN, telencephalin; TNFα, tumor necrosis factor α; IL-1, interleukin-1; IFN-γ, interferon-γ; LPS, lipopolysaccharide (a bacterial endotoxin); PMA, phorbol myristate acetate; SHP-2, Src-homology domain 2 (SH2)-containing phosphatase-2; DC-SIGN, dendritic cell-specific ICAM-3 grabbing non-integrin also known as cluster of differentiation 209 (CD209); PIP2, phosphatidylinositol 4,5-bisphosphate; GraP-DH, glyceraldehyde-3-phosphate dehydrogenase; PKC, protein kinase C; SGEF, RhoG-specific SH3-containing guanine-nucleotide exchange factor; Kidins220, kinase D-interacting substrate of 220 kDa; HUVECs, human umbilical vein endothelial cells; UOL, upstream of latency-associated-transcript, a late gene transcripts in Herpes simplex virus type 1 (HSV-1)-infected neurons; ZO-1, zonula occludens-1.

Methods

For this review, a search of the PubMed database at http://www.pubmed.org and http://scholar.google.com were performed to identify the current status of research on ICAM in both reproductive and non-reproductive fields. Articles in English from databases available in PubMed until October 2012 regarding the roles of IgSF CAMs in mammalian testes and ICAMs in cell adhesion dynamics were primarily considered. Individual searches were conducted using keywords and different combinations thereof. Keywords used for these searches include cell adhesion, cell junction, cell adhesion molecules, immunoglobulin superfamily CAMs, ICAM, soluble ICAM, human reproduction, testis, endothelial/epithelial barrier, BTB, seminiferous tubule, seminiferous epithelium, TEM of leukocytes and transmigratory cup. Journal articles were selected on the basis of relevance, quality and timeliness. We are indebted to many investigators in the field who have contributed significantly to the study of ICAMs in mammalian cell movement and early eminent investigators who laid the ground work in studying spermatogenesis, helping us to prepare this article for reproductive biologists interested in understanding the role of ICAMs in spermatogenesis. However, due to the space limit, many original references and important articles could not be cited, even though we have included several landmark earlier articles in this brief review. Also, every effort was made to cite the latest reviews on a specific subject by leading investigators in the field, so that readers can have access to many important original studies in these recent reviews.

Results

ICAMs and cell adhesion

Five ICAM family members, namely ICAM-1 to -5, are known to date, which possess different numbers of Ig-like C2-type domains in their extracellular region, ranging from two to nine (Fig. 2). Each of the Ig-like domains is derived from a separate exon and the deduced polypeptide sequence is differentially glycosylated (Fig. 2, Table II). Apart from the structural similarities, ICAMs differ greatly in tissue distribution and also in regulation. For example, ICAM-1 and -2 are ubiquitously expressed in mammalian cells and both of them (and also the only two of them) are expressed by Sertoli and germ cells in the rat testis, whereas ICAM-3 to -5 display highly restricted tissue/cellular distribution (Table II). Additionally, they are either inducible or non-inducible by different cytokines and other biomolecules (Table II). Human ICAM-1, -3, -4 and -5 genes were mapped to the chromosome region 19p13.2–13.3, with ICAM-2 at 17q23–25 (Hayflick et al., 1998). Genome-wide analysis in different populations has offered a genetic linkage of loci on human chromosome 19 to multiple diseases, reflecting the possibility that ICAMs could be useful diagnostic markers and therapeutic targets. For instance, the ICAM gene cluster on chromosome 19 has been identified as a susceptibility locus for breast and prostate cancer, as well as systemic lupus erythematosus (Kammerer et al., 2004; Kim et al., 2012).

ICAMs were first shown to be involved in leukocyte homotypic aggregation and leukocyte–endothelial cell adhesive interactions via their binding to β2-integrins (Harris et al., 2000). Each member of the leukocyte-specific β2-integrin subfamily (i.e. the CD11 clusters) is capable of binding to one or more member of the ICAM family at the cell–cell interface, and among them the leukocyte function-associated antigen 1 (LFA-1) recognizes all five ICAMs (Harris et al., 2000; Luo et al., 2007) (Table II). It is noted that integrins are pivotal cell adhesion receptors that mediate leukocyte activation and migration and their interactions with ICAMs in leukocyte adhesion were found to have various modulatory effects on the immune system, such as inducing the production of cytokines (e.g. interleukins) and/or activation of proteinases (e.g. matrix metalloproteinases (MMPs)) (Park et al., 2010; Costantini et al., 2011). Thus, much effort in the field has been devoted to understanding how ICAMs contribute to leukocyte recruitment and extravasation into target tissues in pathological conditions (e.g. inflammation) and ICAMs are generally accepted as co-stimulators of leukocyte activation. In the testis, β1-integrin, an adhesion molecule found at the apical ES, hemidesmosome and stem cell niche (Cheng et al., 2011a), was shown to interact with ICAM-2 at the apical ES (Xiao et al., 2012b). It has been demonstrated that activation of MMP-2 is needed for generation of biologically active laminin fragments from the α6β1-integrin–laminin-333 adhesion protein complex at the apical ES to prepare for spermiation (Cheng et al., 2011a). Along these lines of research, it is of interest to know whether ICAM-2 participates in the process, e.g. modulation of the MMP-2 action. Also, consistent with their roles as CAMs, ICAMs are actively engaged in remodeling of the actin cytoskeleton in effector cells (Vicente-Manzanares and Sanchez-Madrid, 2004; van Rijssel et al., 2012), but studies have implied that ICAMs may also associate with the microtubule- and intermediate filament-based cytoskeleton (Federici et al., 1996; Nieminen et al., 2006; Nejmeddine et al., 2009; Jean-Mairet et al., 2011). The cytoplasmic domain of several ICAMs has been shown to interact with various actin-binding and regulatory proteins. For instance, ICAM-1, -2, -3 and -5 were found to bind to ERM proteins, ezrin, radixin and moesin (Serrador et al., 2002; Doulet et al., 2006; Furutani et al., 2007). ICAM-1, -2 and -5 were found to bind to an actin cross-linker α-actinin (Nyman-Huttunen et al., 2006). ICAM-1 is also known to interact with cortactin, a substrate for Src that co-activates actin-related protein (Arp) 2/3 complex with neural Wiskott-Aldrich syndrome protein (N-WASP) to support branched actin filament (F-actin) polymerization to generate branched F-actin network (Allingham et al., 2007; Schnoor et al., 2011; van Rijssel et al., 2012) (Table II). Therefore, ICAMs can efficiently regulate F-actin reorganization in mammalian cells through their differential interactions with various actin-binding and regulatory proteins: cell surface distribution of ICAMs can also be controlled by the underlying cytoskeleton via linker proteins, such as ezrin and α-actinin (Nyman-Huttunen et al., 2006; Yoon et al., 2008). In the rat testis, both ICAM-1 and -2 were found to structurally interact with actin (Xiao et al., 2012a, b). Interestingly, ICAM-2 co-localized with F-actin at the apical ES throughout different stages of the epithelial cycle, except at stages VII–VIII shortly before spermiation (Xiao et al., 2012b), suggesting that ICAM-2 may play a role in F-actin reorganization at Stage VIII of the cycle to facilitate sperm release by causing disruption of F-actin bundles at the apical ES during spermiation. On the other hand, ICAMs also have the ability to activate signal transduction via ectodomain shedding, such as to generate soluble ICAMs, or clustering on plasma membrane to activate Ca²⁺- and/or Rho-mediated signaling, and/or recruit intracellular downstream partners, such as protein kinases and phosphatases, in response to external stimuli, which modulates the cytoskeleton to affect cell adhesion and migration (Etienne-Manneville et al., 2000; Pluskota et al., 2000; Thompson et al., 2002; Park et al., 2010). More notably, ICAM-1 clustering on endothelial cells was shown to recruit vascular cell adhesion molecule-1 (VCAM-1) (van Buul et al., 2010b) to the site to regulate cell adhesion, illustrating that ICAMs possess a broad signaling capacity. In the sections that follow, we elaborate on these points, e.g. findings concerning ICAM shedding and clustering in the regulation of cell adhesion, as well as the cooperative activities among ICAMs and the related cytoskeletal and signaling components during leukocyte TEM, and their implication in germ cell migration across the seminiferous epithelium.

Polymorphic and multifunctional molecules: soluble and multimeric ICAMs

Secretion and ectodomain shedding have been reported for several IgSF CAMs, such as JAMs (junctional adhesion molecules) and NCAM-L1 (Cavallaro and Dejana, 2011). It is therefore not surprising that in addition to the membrane-bound full-length ICAM (mICAM), different truncated forms of ICAMs have been found as normal physiological constituents of various biological fluids: they are known as soluble or circulating ICAM (sICAM/cICAM). Since sICAM-1, -2, -3 and -5 have been detected in serum of healthy humans (Mustjoki et al., 2001; Gahmberg et al., 2008; Lopez-Lerma and Estrach, 2009), it is conceivable that sICAM may function as essential biomolecules in mammalian cells/tissues. For instance, sICAM-4 was identified in spent media of COS-7 cells transfected with murine ICAM-4 cDNAs (Lee et al., 2003) and sICAM-1 was detected in mutant mice lacking mICAM-1 (van Den Engel et al., 2000). Also, the age-dependent expression profiles of mICAM-5 and sICAM-5 do not parallel each other during brain development (Gahmberg et al., 2008), indicating their differential roles. Additionally, considerably elevated levels of sICAM were detected in many pathological conditions, implying a crucial role of sICAM in immune surveillance (Gahmberg et al., 2008; Mendez et al., 2008; Lopez-Lerma and Estrach, 2009). Recently, a form of sICAM-1 with an apparent molecular weight (Mr) of ∼70 kDa was discovered in normal rat testis, which was also enriched in germ cells detected using a polyclonal antibody targeting the extracellular domains 2 and 3 of ICAM-1 (Xiao et al., 2012a) (Table II). By using this antibody, however, additional protein fragments with lower Mr were also observed in Sertoli and germ cells, suggesting other truncated forms of sICAM-1 may exist in the testis (Xiao et al., 2012a), but whether they have distinct biological functions in maintaining spermatogenesis remains to be investigated.

While sICAM was viewed as a molecular marker of inflammation, and might be involved in modulating immune responses such as the production of cytokines and proteinases, perhaps overlapping the function of or distinctly different from that of mICAM (Witkowska and Borawska, 2004; Gahmberg et al., 2008; Tse et al., 2009), the origin of sICAM and its significance remains to be elucidated. Nonetheless, it is generally accepted that proteolytic cleavage is the primary mechanism responsible for the ectodomain shedding of ICAM-1, such as via action of (i) serine proteases (e.g. elastase and cathepsin G) (Robledo et al., 2003; Mendez et al., 2008), (ii) MMPs [e.g. MMP-9, -13 (also known as collagenase-3) and -14 (also known as membrane type 1-MMP, i.e. MT1-MMP)] (Fiore et al., 2002; Sithu et al., 2007; Tarin et al., 2009), and/or (iii) zinc-dependent metalloproteinase (MP) [e.g. TNFα-converting enzyme (TACE, also known as a disintegrin and metalloproteinase-17, i.e. ADAM-17) (Tsakadze et al., 2006)]. sICAM-2, -3 and -5 are generated from the corresponding ICAMs via cleavage by MP, caspase-3 and -8, and MMP-2 and -9, respectively (Tian et al., 2007; Sato et al., 2009). These findings are useful since MMP-2 is known to be restricted to the apical ES in the rat testis, while MMP-9 is limited to the basal ES at the BTB (Cheng et al., 2011a), so it becomes desirable to determine whether ICAMs are substrates of MMPs in the testis. Moreover, it appears that the regulatory molecules (e.g. inflammatory mediators such as TNFα), needed to induce cleavage of ICAM extracellular domain, vary among cell types and this event can be modulated by multiple kinases [e.g. Src, p38 mitogen-activated protein kinase (MAPK), phosphatidylinositol 3-kinase], and phosphorylation of Tyr residues within the ICAM cytoplasmic domain can affect the shedding of the extracellular domain (Tsakadze et al., 2004; Sithu et al., 2007; Mendez et al., 2008). It is worth noting that the cytoskeletal association of mICAM may be crucial for its sensitivity to proteolytic cleavage, such as in ICAM-5 (Tian et al., 2007). Alternative RNA splicing can also be a mechanism that accounts for the generation of sICAM-1 and sICAM-4 (Lee et al., 2003; Mendez et al., 2008). It is noted that different cleaved products or splice variants of ICAMs have been identified, such as with or without deletion of one or more extracellular Ig-like repeats (Lee et al., 2003; Witkowska and Borawska, 2004; Tian et al., 2007).

sICAMs are biologically active molecules and their up-regulation by cytokines mirrors that of their corresponding mICAMs (Gahmberg et al., 2008; Mendez et al., 2008; Witkowska and Borawska, 2004). However, mICAM expression is often regulated at the transcriptional level by transcription factors, such as nuclear factor-κB, cAMP-response element-binding protein, E-twenty-six or the RUNX family in response to stimuli, whereas that of sICAM is regulated as a combined result of changes in the mICAM protein level and the proteolytic activity (Witkowska and Borawska, 2004; Hadad et al., 2011; Estecha et al., 2012). sICAM-1 and -4 have been suggested to compete for ligand binding with the native mICAMs present on the cell surface and thus reduce cell adhesion (Wang et al., 2005; Trinh-Trang-Tan et al., 2010). sICAM-1 was found to adversely affect endothelial barrier function, blocking leukocyte TEM (Otto et al., 2000; Yang et al., 2005; Clark et al., 2007; Sumagin and Sarelius, 2010). Indeed, overexpression of the aforementioned ∼70 kDa sICAM-1 containing all five Ig-like domains in Sertoli cells with an established barrier function in vitro was found to display a disruptive effect on the Sertoli cell barrier, which was a sharp contrast to ICAM-1 (Xiao et al., 2012a), illustrating that sICAM-1 and ICAM-1 can have antagonistic effects on the BTB integrity. Considering this and the relative abundance of sICAM-1 produced by germ cells, it is tempting to speculate that while ICAM-1 maintains the barrier function, sICAM-1 may be used by preleptotene spermatocytes to serve as a molecular ‘switch’ to ‘open up’ the BTB at or near their apical region to facilitate their passage at the site. sICAM may also induce the loss of cell adhesion between Sertoli and germ cells, in particular spermatocytes and round spermatids, since in vivo overexpression of the sICAM-1 in rat testes, besides perturbing the BTB, was also found to induce depletion of spermatocytes and round spermatids from the seminiferous epithelium (Xiao et al., 2012a). Cytokines (e.g. TNFα, TGF-βs and IL-1α) and androgens are known to mediate junction restructuring by triggering internalization and/or recycling of adhesion protein complexes at the BTB and apical ES, via the combined effects of protein kinases (e.g. Src family kinases, focal adhesion kinase (FAK)), proteases (e.g. MMPs) and actin-binding and regulatory proteins (e.g. Eps8, Arp3) and perhaps others (e.g. zyxin, vinulin, N-WASP, cortactin (Young et al., 2009a, b, 2012; Cheng and Mruk, 2012; Lie et al., 2012; Young and Vogl, 2012) and cytokines were shown to up-regulate ICAM-1 expression in mouse Sertoli cells (De Cesaris et al., 1998). Interestingly, a non-classical endocytosis pathway directed by ICAM-1 has been reported in endothelial cells, which involves Src- and Rho-dependent actin cytoskeleton remodeling (Muro et al., 2003; Serrano et al., 2012). However, it is not known whether ICAM-1 is involved in assisting intracellular protein trafficking in the seminiferous epithelium, such as the recycling of adhesion protein complexes from the ‘old’ BTB and/or apical ES to the ‘new’ sites (Fig. 1). It is likely that sICAM-1 is involved in TNFα-mediated protein endocytosis/junction disassembly in the testis (Xia et al., 2009), since TNFα treatment is known to increase endothelial sICAM-1 production which contributes to paracellular permeability (De Cesaris et al., 1998; Clark et al., 2007; Mendez et al., 2008). Nevertheless, it remains possible that sICAM-1 promotes TNFα action during spermatogenesis as shown in other cell systems, or, it is a bidirectional regulation that synergistically leads to disintegration of the BTB and apical ES (McCabe et al., 1993; Otto et al., 2000; Tsakadze et al., 2004). On another note, sICAM-5 was shown to stimulate the phosphorylation of cofilin (an actin-depolymerizing factor) to induce F-actin reorganization (Conant et al., 2011), whereas the remnant cytoplasmic tail remaining after cleavage might compete and disrupt the cellular anchorage of neighboring mICAM-5 to the actin cytoskeleton and consequently affect the intracellular signaling (Sithu et al., 2007; Tian et al., 2007). Furthermore, homophilic interactions between sICAMs and mICAMs have been demonstrated (Table II). It was found that the lowered level of mICAMs impeded cell survival (Sato et al., 2009; Trinh-Trang-Tan et al., 2010), but the correlations between different sICAMs in body fluids, as well as that between sICAM and its corresponding mICAM in pathological conditions, such as male infertility, have not been established and are probably more complicated than it was initially conceived (Mustjoki et al., 2001; Lopez-Lerma and Estrach, 2009; Cserti-Gazdewich et al., 2010; Xiao et al., 2012a).

As mentioned above, while the release of soluble ectodomains of ICAMs is capable of triggering signal transduction and mediating junction disassembly, the redistribution and/or clustering of ICAMs on the cell surface is equally critical in regulating cellular functions. For instance, ICAM-1 is known to dynamically inter-convert between monomers and dimers on cell surface and the latter assume a more potent and active role in cell adhesion (Miller et al., 1995; Reilly et al., 1995; Oh et al., 2011). Likewise, dimeric immobilized sICAM-1 binds more efficiently to LFA-1 versus monomeric sICAM-1 (Welder et al., 1993; Jun et al., 2001a, b). Structurally, ICAM dimers can further integrate to W-shaped tetramers which are apt for macro-clustering to form larger multimers in response to changes in environment, as well as undergo homophilic and/or heterophilic co-clustering with membrane partners on the adjacent cell surface (Jun et al., 2001a; Shaw et al., 2004; Yang et al., 2004; Millan et al., 2006). Studies have indicated that clustering may facilitate ICAM-1-mediated endocytosis and actin stress-fiber formation but do not necessarily lead to changes in vascular permeability (van Buul et al., 2002; Muro et al., 2003; Millan et al., 2006; Sumagin et al., 2011). Approach of leukocytes to an endothelium layer often suffices to induce clustering of endothelial ICAM-1, laying the foundation for successive leukocyte-endothelial cell interactions, which can be mimicked by cross-linking ICAM-1 on the plasma membrane with specific monoclonal antibodies (Millan et al., 2006; Muller, 2011). It has been found that cross-linking of ICAM-1 on diverse cell types transmits outside signals to inside of the cell (‘outside-in’ signaling), causing redistribution of ICAM-1 on the apical surface and establishes its association with the actin cytoskeleton (Hubbard and Rothlein, 2000; Amos et al., 2001; Rahman and Fazal, 2009). This is accompanied by increases in cytosolic Ca²⁺ and tyrosine phosphorylation of multiple cellular proteins, and an activation of numbers of signaling pathways (Rahman and Fazal, 2009). As a result of these signals, the F-actin cytoskeleton is rearranged and subsequently affects the cell–cell adhesion (Doulet et al., 2006; Wittchen, 2009). More intriguingly, ICAM-1 cross-linking has been shown to promote VCAM-1 expression and recruits VCAM-1 to the same sites (Lawson et al., 1999; van Buul et al., 2010b). Similarly, cross-linking of other ICAMs has been displayed to initiate and/or augment intracellular communication (Berney et al., 1999; Perez et al., 2002).

Apparently, the ‘outside-in’ signaling conveyed by ICAM clustering relies heavily on the short cytoplasmic tail, which interacts with a legion of cytoskeletal binding/regulatory proteins in the cell cytosol (Table II). Tailless ICAM-1 (i.e. sICAM-1) has been shown to fail in mediating intracellular signaling versus intact ICAM-1, and thus intact ICAM-1 is crucial in various biological activities, such as leukocyte TEM and virus infection (Pluskota et al., 2000; Greenwood et al., 2003; Lyck et al., 2003; Yang et al., 2005; Dermody et al., 2009; Sumagin and Sarelius, 2010). On the other hand, clustering of ICAM-1 induces better association between ICAM-1 and the underlying actin cytoskeleton and its disruption may create an ‘inside-out’ signal (Topp et al., 2003; van Buul et al., 2010a). The likely effects of ICAM-1 clustering-induced intracellular signaling in endothelial cells to facilitate leukocyte TEM are discussed below and are summarized in Fig. 3. It is worth noting that there remains a lack of evidence supporting the role of ICAM-1 clustering in signaling during spermatogenesis, which should be carefully evaluated in future studies. Nonetheless, the action of ICAM-mediated signaling function summarized herein can be the basis of designing future experiments.

Figure 3.

A schematic drawing illustrating the molecular mechanisms by which ICAM-1, sICAM-1 and ICAM-2 regulate TEM (transendothelial/transepithelial migration) of leukocytes via paracellular or transcellular pathways. Endothelial cells were shown in either light purple and green. Prior to leukocyte TEM, endothelial ICAM-1 and/or -2, besides acting as adhesion molecules, may serve as a guide for the movement of leukocytes on the endothelium. In paracellular TEM (see ‘red’ arrows), ICAM-1 clustering can recruit non-receptor protein tyrosine kinases (e.g. Src and Pyk2) to phosphorylate junctional protein complexes (e.g. VE-cadherin-catenin adhesion protein complex) and cause their disassociation to open up the endothelial junction. Also, Src recruitment can be used to phosphorylate cortactin (or other actin-binding and regulatory proteins), which results in an F-actin remodeling and/or re-organization to facilitate leukocyte TEM. In transcellular TEM (see ‘blue’ arrows), ICAM-1-mediated phagocytosis/endocytosis, either independent of or dependent on caveolar network, can be used for leukocyte transcellular TEM without perturbing the endothelial junctions.

ICAM in TEM

Throughout the seminiferous epithelial cycle of spermatogenesis, such as in the rat testis, developing germ cells, both spermatocytes and spermatids migrate progressively across the seminiferous epithelium from the basal to the adluminal compartment, including the passage of preleptotene spermatocytes across the BTB (Pelletier and Byers, 1992; Pelletier, 2001, 2011; Hermo et al., 2010b). Developing elongating spermatids that have reached and reside in the adluminal compartment at Stages III–IV also traverse the epithelium, returning to the epithelium near the tunica propria (Fig. 1), until almost ‘touching’ the Sertoli cell nuclei located near the basement membrane at Stage V of the epithelial cycle, before they move back to the adluminal compartment at Stages VI–VIII, illustrating extensive spermatid movement (Clermont, 1963, 1972; Russell, 1977). Yet, germ cells per se are non-motile cells, lacking the apparatus (e.g. lamellipodia) to ‘migrate’ on their own as do fibroblasts, macrophages, neutrophils, and other cells, but rely on the Sertoli cells where the extensive re-organization of actin and tubulin networks provides the protrusive force for their movement (Russell et al., 1989; Lee and Cheng, 2004; Vogl et al., 2008). Importantly, a large body of work is found in the literature delineating the involvement of ICAMs in leukocyte TEM, a process highly reminiscent of germ cell migration during spermatogenesis. We thought it pertinent to provide a brief and critical review herein since this information serves as the basis of some important future experiments.

In an immune response, circulating leukocytes leave the blood vessels by extravasation to reach the site of injury or infection. However, the vascular endothelium acts as a filtration blood–tissue barrier that regulates the leukocyte passage. As has been largely accepted, leukocyte extravasation is composed of a series of sequential steps, including initial capture/tethering, rolling along the vessel wall, arrest on the endothelium, adhesion strengthening and spreading, intravascular crawling, diapedesis/transendothelial migration and finally crossing the endothelial basement membrane to the interstitial tissue and arrival at the intended site. This multistep process demands drastic changes in cytoskeleton and efficient coupling of adhesion receptors on both leukocytes and endothelial cells (Wittchen, 2009; Fernandez-Borja et al., 2010; Nourshargh et al., 2010; Muller, 2011). ICAM-1 has been implicated in almost every course of leukocyte recruitment, including rolling, arrest, adhesion, crawling and TEM (Sumagin and Sarelius, 2007; Rahman and Fazal, 2009). For example, during the adhesive cell–cell interacting phase between leukocytes and endothelial cells prior to TEM, endothelial ICAM-1 and -2 have been proposed to serve as a guide or signal for leukocyte recognition of the sites for TEM, and help leukocyte switch from adhesion/crawling mode to TEM, as well as to optimize endothelial junctions for leukocyte access (Schenkel et al., 2004; Yang et al., 2005; Millan et al., 2006; Porter and Hall, 2009; Woodfin et al., 2009; Steiner et al., 2010; Sumagin and Sarelius, 2010). Leukocyte TEM has been shown to take place via both paracellular and transcellular routes. The use of the ‘paracellular’ route for TEM signifies a transient dissolution of inter-endothelial junctions to allow the passage of leukocytes, while in the ‘transcellular’ route leukocytes directly traverse the cytoplasm of an endothelial cell, without perturbing the endothelial junctions (Wittchen, 2009; Komarova and Malik, 2010; Muller, 2011).

Currently, the prevailing hypothesis of leukocyte TEM is as follows. Leukocyte TEM is launched and maintained by some specific endothelial structures enriched in ICAM-1, whose expression is rapidly induced upon activation by various inflammatory mediators (Table II). It was first reported that an actin-rich cup-like docking structure analogous to receptor-mediated phagosomes at the leukocyte–endothelial cell interface was found to be quickly vanished as the leukocyte TEM began (Barreiro et al., 2002). A similar ‘transmigratory cup’ was also found during transcellular and paracellular TEM; however, this ultrastructure enwrapped and followed leukocyte crossing all the way during the TEM process and appeared to require the involvement of the microtubule-based, besides the actin-based, cytoskeleton (Carman et al., 2003; Carman and Springer, 2004) (Fig. 3). Formation of these cup-like ultrastructures required dynamic redistribution/clustering of ICAM-1 at the leukocyte–endothelial cell–cell contact site, and involved the recruitment of multiple actin-binding proteins, such as ERM family proteins, α-actinin and vasodilator-stimulated phosphoprotein (an enhancer of Arp2/3/N-WASP-mediated branched actin nucleation to create a branched actin network, also found in phagosomes), as well as focal adhesion proteins such as vinculin, talin and paxillin (Barreiro et al., 2002; Doulet et al., 2006). Other CAMs, including VCAM-1, ICAM-2, platelet endothelial cell adhesion molecule-1 (PECAM-1/CD31), JAM-A, and CD44, were also implicated in the processing of this cup-like docking ultrastructure (Barreiro et al., 2002, 2008; Carman et al., 2003; Carman and Springer, 2004; Doulet et al., 2006). In short, this cup-like docking ultrastructure serves as the receptor for leukocytes and the transmigration of leukocyte-docking/receptor-like ultrastructure across the endothelial barrier, either paracellularly or transcellularly, is analogous to a ‘giant’ phagosome, in which ICAM-1 summons all the necessary elements possibly via its redistribution/clustering ability at the site (Fig. 3).

It was also reported that the onset of TEM associated with the formation of a ring-like ultrastructure (Shaw et al., 2004) which was assembled via a spatiotemporal re-localization of LFA-1 on leukocytes that triggered a co-clustering of ICAM-1. During paracellular TEM, a leukocyte probably inserted a pseudopod into the ‘ring’ and squeezed across the endothelial junctions (Shaw et al., 2004) (Fig. 3). ICAM-1 was shown to be involved in the transient ‘opening’ of intercellular junctions, since clustering of ICAM-1 at the site recruits non-receptor protein tyrosine kinases (e.g. c-Src and Pyk2), which subsequently phosphorylate crucial components of the VE-cadherin-catenin adhesion protein complex, namely VE-cadherin, β- and γ-catenins, leading to a reversible loss of the junction integrity and thus ‘an opening’ in the endothelial tight junction (TJ) barrier for the passage of a leukocyte (Allport et al., 2000; Shaw et al., 2001; van Buul et al., 2007; Nottebaum et al., 2008; Turowski et al., 2008; Di Lorenzo et al., 2011; Pelletier, 2011; Alcaide et al., 2012). ICAM-1 clustering was also found to result in a dramatic remodeling of the actin cytoskeleton, via its interaction with the actin nucleation-promoting factor cortactin, which is phosphorylated by Src upon its association with ICAM-1 (Etienne-Manneville et al., 2000; Tilghman and Hoover, 2002; Yang et al., 2006; Schnoor et al., 2011). The net result of these changes in the ICAM-1-c-Src-cortactin pathway thus destabilizes the integrity of the endothelial TJ barrier to facilitate the transit of leukocytes. In the testis, ICAMs and Src/Pyk2 are also natural partners in regulating junction dynamics, in that the pathways involving either ICAM-1 or -2 were c-Src- and/or Pyk2 dependent (Xiao et al., 2012a, b). Collaboratively, in the TNFα-treated human Sertoli cells, ICAM-1 and c-Src were found to co-localize at the Sertoli cell–cell interface as shown in Fig. 4.

Figure 4.

Co-localization of ICAM-1 with c-Src at the cell–cell interface of human Sertoli cells treated with TNFα. Human Sertoli cells were cultured as earlier described (Chui et al., 2011) at 0.05 × 10⁵ cells/cm² on fibronectin-coated glass coverslips in DMEM/F-12 containing 10 μg/ml insulin, 5 μg/ml human transferrin, 2.5 ng/ml epidermal growth factor and 100 μg/ml penicillin–streptomycin in a humidified atmosphere at 35°C with 95% air/5% CO₂ (vol/vol) and immunostained for ICAM-1 [red fluorescence; rabbit polyclonal antibody from Novus Biologicals (NB100-81977)] and c-Src [green fluorescence; mouse monoclonal antibody from Santa Cruz Biotechnology (sc-8056)] was used at a 1:100 and 1:50 working dilution, respectively. Dual-labeled immunofluorescence analysis was performed essentially as earlier described (Xiao et al., 2011). In normal human Sertoli cells, ICAM-1 was mostly localized in the cell cytosol with relatively weak staining at the cell–cell interface but more c-Src was localized near the human Sertoli cell surface. However, following treatment of these human Sertoli cells with TNFα (10 ng/ml) for 1 day (1d), ICAM-1 was found to be redistributed and localized more to the human Sertoli cell–cell interface, co-localized with Src (‘white’ arrows indicate co-localized ICAM-1 and c-Src at the human Sertoli cell–cell interface). Cell nuclei (blue) were stained with DAPI (4′,6-diamidino-2-phenylindole). Bar = 100 μm, which applies to all other micrographs.

Besides the above-stated pathways for TEM, signaling pathways involving cytosolic Ca²⁺ flux [Etienne-Manneville et al., 2000), endothelial nitric oxide synthase (Kevil et al., 2004; Martinelli et al., 2009), other protein kinases (e.g. FAK and different protein kinase C isoforms (PKCs)] (Etienne et al., 1998; Rodriguez-Fernandez et al., 1999; Strell et al., 2010; Sumagin et al., 2011), small GTPases, including Rho/Rho associated protein kinase/Rac (Adamson et al., 1999; Barreiro et al., 2002; Thompson et al., 2002; Smith et al., 2003; van Buul et al., 2007), and other actin-binding proteins (e.g. filamins) (van Rijssel et al., 2012; Su et al., 2012) may also actively take part in this process. Furthermore, it was shown that PECAM-1 might block the ICAM-1-stimulated Src activation and cytoskeletal rearrangement via the tyrosine phosphatase SHP-2 in order to ‘re-seal’ the endothelial gaps following the passage of migrating leukocytes at the site (Wang et al., 2003; Couty et al., 2007), illustrating that the ICAM-Src interaction is crucial in regulating junction permeability.

Interestingly, there are two reports revealing that ICAM-1 might co-operate with caveolin-1 and/or vimentin to facilitate leukocyte transcellular transmigration (Millan et al., 2006; Nieminen et al., 2006) via an ICAM-1-caveolea-mediated transcytotic pathway. For instance, ICAM-1 was shown to co-cluster with caveolin-1 in actin-rich ‘ring-like’ invaginations in the endothelium, surrounding the invading leukocyte pseudopod and these ultrastructures were found to move together as a complex towards the basal plasma membrane (Fig. 3). Other studies have reported that ICAM-1 might bind indirectly to caveolin-1 via filamin (Kanters et al., 2008), and a vimentin intermediate filament-based cup-like ultrastructure around a migrating leukocyte in the transcellular route was also shown (Nieminen et al., 2006) (Fig. 3). Collectively, these findings illustrate the complexity of leukocyte TEM and the indispensability of ICAM-1 in these junctional rearrangements (Fig. 3). While germ cells are not motile cells per se, during spermatogenesis as they progressively ‘relocate’ across the seminiferous epithelium and move back-and-forth between the basal and the adluminal compartment, these findings will serve as a guide for future functional studies.

Transepithelial migration of germ cells in the testis

In the seminiferous tubule of mammalian testis where spermatogenesis takes place, Sertoli cells extend from the basement membrane towards the tubule lumen, and together with germ cells at different stages of their development, creating the seminiferous epithelium (Weber et al., 1983; Russell and Peterson, 1985; Russell et al., 1989; Griswold, 1995; Cheng and Mruk, 2002; Mruk and Cheng, 2004b; Vogl et al., 2008) (Fig. 1). During the translocation of developing germ cells from the basal to the adluminal compartment in the rat testis, germ cell contacts with Sertoli cells are mediated via either the desmosome, the gap junction, or the apical ES (Mruk and Cheng, 2004a; Yan et al., 2007; Wong et al., 2008). Specifically, spermatogonial stem cells, spermatogonia, spermatocytes and step 1–7 spermatids attach to Sertoli cells via desmosomes and communicate with each other via gap junctions (Lie et al., 2011; Mruk and Cheng, 2011), while the apical ES connects elongating/elongated spermatids (step 8–19 spermatids) to Sertoli cells (Russell and Peterson, 1985; Vogl et al., 2000; Toyama et al., 2003; Cheng and Mruk, 2010) (Fig. 1). Moreover, at Stage VIII of the seminiferous epithelial cycle, preleptotene spermatocytes residing in the basal compartment must traverse the BTB to the adluminal compartment while transforming to leptotene and then zygote spermatocytes prepare for meiosis (Russell, 1977). Yet the BTB is one of the tightest blood–tissue barriers in the mammalian body, constructed by apposed plasma membranes of adjacent Sertoli cells near the basement membrane (Fawcett et al., 1970; Fawcett, 1975; Setchell and Waites, 1975; Pelletier, 2011; Franca et al., 2012). It partitions the seminiferous epithelium into a basal and an adluminal compartment (Dym and Fawcett, 1970; Setchell, 1980; Pelletier, 2001, 2011; Mital et al., 2011; Cheng and Mruk, 2012; Franca et al., 2012). The BTB comprises multiple junction types, namely TJs, basal ESs, gap junctions (GJs) and desmosomes (Russell and Peterson, 1985; Vogl et al., 2008; Cheng and Mruk, 2012). Importantly, a unique ‘intermediate compartment’ mechanism involving massive cytoskeletal changes and endocytic vesicle-mediated protein endocytosis/recycling has been suggested, wherein the ‘old’ BTB above spermatocytes is disassembled immediately following the assembly of a ‘new’ BTB below, to facilitate the transmigration of preleptotene/leptotene spermatocytes at Stage VIII (Cheng and Mruk, 2012; Russell, 1977, 1978; Yazama, 2008) (Fig. 1). At the same time, spermiation happens near the tubule lumen, with the breakdown of apical ES to release the mature spermatozoa (Russell and Clermont, 1976; O'Donnell et al., 2011). It has been shown that complex adhesive interactions and extensive signaling cascades occur at the Sertoli–Sertoli and Sertoli–germ cell interface; (Russell and Clermont, 1976; Cheng and Mruk, 2012) these are reminiscent of the leukocyte TEM.

CAMs, such as N-cadherin at the BTB, and β1-integrin at the apical ES and hemidesmosome also take part in regulating junction turnover during spermatogenesis. They are essential constituents of the apical ES–BTB–hemidesmosome axis that co-ordinates the BTB restructuring and spermiation—the two cellular events taking place at the opposite ends of the Sertoli cell epithelium (Cheng and Mruk, 2010, 2012). Interestingly, ICAM-1 present on the surface of cultured mouse Sertoli cells was shown to be inducible by inflammatory mediators, such as TNFα, IL-1 and IFN-γ (Riccioli et al., 1995), and there are some novel findings in the field which characterized the role of ICAM family members at the cell junctions in the testis (Xiao et al., 2012a, b). In the following sections, we primarily focus on the role of ICAM-1 and -2 in the rat testis, which have been shown to regulate the dynamics of the BTB and the apical ES, respectively.

ICAM-1 in the testis: a multifunctional molecule in spermatogenesis

ICAM-1 was first shown to be expressed by mouse Sertoli cells cultured in vitro, residing on the Sertoli cell surface and it was up-regulated by treatment of these cells with IL-1, TNFα, lipopolysaccharide or interferon-γ; it was postulated that the interactions of ICAM-1 and the pro-inflammatory cytokines were involved in autoimmune pathologies and other immunological responses in the testis (Riccioli et al., 1995). Subsequent studies have shown that the TNFα-induced ICAM-1 expression is mediated via the JNK MAPK signaling pathway (De Cesaris et al., 1999).  In the rat testis, ICAM-1, similar to the mouse testis (Riccioli et al., 1995), is expressed by Sertoli cells; however, ICAM-1 is also expressed by germ cells, and its expression in the seminiferous epithelium is stage specific (Xiao et al., 2012a). ICAM-1 is highly expressed in the basal region of the seminiferous epithelium including the BTB at Stages VIII–IX of the epithelial cycle (Xiao et al., 2012a), coinciding with the time of BTB restructuring. Using immunohistochemistry and a specific anti-ICAM-1 antibody, ICAM-1 has been shown to associate with most types of germ cells but it is most robust with elongating spermatids at Stages IX–XIII at the apical ES (Xiao et al., 2012a). More strikingly, by using co-immunoprecipitation and confocal microscopy, ICAM-1 has been found to partially co-localize with the TJ-protein occludin and a basal ES-protein N-cadherin (Xiao et al., 2012a). Consistent with its role as an integral membrane protein and a constituent component of the BTB, the overexpression of ICAM-1 in Sertoli cells exhibits a strengthening effect on the Sertoli cell barrier function. Figure 4 depicts the distribution of ICAM-1 in human Sertoli cells. Interestingly, the treatment of human Sertoli cells with TNFα was found to induce the re-distribution of ICAM-1 in these cells with ICAM-1 mostly found at the cell–cell interface and co-localized with c-Src (Fig. 4), suggesting c-Src may be used to alter the phosphorylation status of ICAM-1 at the Sertoli cell BTB, modulating its cell adhesive (or other) function at the site.

In our study, a soluble form of ICAM-1 that encompasses all five Ig-like domains was identified in the normal rat testis and enriched in germ cells (Xiao et al., 2012a). Subsequent analysis to overexpress this sICAM-1 in Sertoli cells in vitro was found to elicit an adverse effect on the Sertoli cell barrier (Xiao et al., 2012a). In short, sICAM-1 was found to down-regulate the expression of a number of BTB constituent proteins, such as N-cadherin, γ-catenin (also known as plakoglobin), as well as a GJ protein connexin 43 at the Sertoli cell BTB in vitro which mimicked the BTB in vivo (Xiao et al., 2012a), likely via a c-Src- and Pyk2-mediated pathway. In line with these observations, when the sICAM-1 construct was overexpressed in the rat testis in vivo, a down-regulation on the expression of N-cadherin and connexin 43 at the BTB was detected, concomitant with a disruption of the BTB function when the barrier integrity was assessed by a functional in vivo assay (Xiao et al., 2012a), analogous to findings in other epithelia/endothelia as discussed above. The F-actin network at the BTB was also severely damaged. Surprisingly, a loss of spermatocytes and round spermatids from the seminiferous epithelium was also observed (Xiao et al., 2012a), illustrating a primary loss of BTB function can lead to a secondary loss of adhesion function at the Sertoli–germ cell interface. It is noted that desmosomes at the site are the likely target of sICAM-1, since spermatocytes and round spermatids are anchored to Sertoli cells in the seminiferous epithelium via desmosomes. However, an evident upsurge in the expression of a desmosome protein desmoglein-2 was demonstrated by immunohistochemistry, despite significant destruction of the adhesion protein complexes between Sertoli and early germ cells, which may represent a physiological response in the testis to prevent further sICMA-1-induced germ cell loss. γ-Catenin, the only known common adaptor to both AJ (i.e. apical and basal ES in the testis) and desmosomes (Lewis et al., 1997), was found to be down-regulated at the Sertoli cell BTB site in vitro following the overexpression of sICAM-1 when examined by immunoblotting and immunofluorescence analysis (Xiao et al., 2012a). It was shown that in epithelial cells c-Src-induced tyrosine phosphorylation of γ-catenin would impede the association of γ-catenin with the AJ but the reverse was shown for desmosomes (Miravet et al., 2003). Therefore, it is important to assess the changes in the γ-catenin expression throughout the seminiferous epithelium, in particular at the sites of desmosomes and apical ESs. It is possible that c-Src-induced γ-catenin phosphorylation results in its dissociation with N-cadherin at the BTB, but a tighter association with the desmosomes at the Sertoli–germ cell interface.

During leukocyte paracellular TEM, the transient VE-cadherin-mediated gaps that form in the endothelial barrier to facilitate leukocyte migration in response to ICAM-1 clustering was shown to be resealed rather quickly, within 5 min (Shaw et al., 2001). Our findings by overexpressing sICAM-1 in the system that led to a severe BTB damage, however, suggest a mechanism that involves the degradation of several BTB components (e.g. the N-cadherin-γ-catenin protein complex) that takes hours before changes in the phenotypes were detected (Xiao et al., 2012a). Interestingly, presenilin-1, an intramembrane-cleaving protease involved in the degradation of the cadherin–catenin protein complex, was shown to be critical to the turnover of ICAM-5 in neurons via an autophagic degradative pathway (Kang et al., 2002; Marambaud et al., 2002; Esselens et al., 2004). In mouse testis, presenilin-1 is expressed by both spermatogonia and Sertoli cells (Dirami et al., 2001). Therefore, it remains to be examined if tailless ICAM-1 is working in concert with a similar intramembranous protein degradation pathway.

As noted above, there are emerging functional roles ascribed to ICAM-1 based on the studies on leukocyte TEM, some of which may be applicable to the transit of germ cells across the seminiferous epithelium during spermatogenesis. For instance, in addition to providing cell adhesive function in the seminiferous epithelium, ICAM-1 can modulate endocytic vesicle-mediated protein trafficking apparatus and cell signaling pathways involving c-Src and Pyk2 that act on the reorganization of actin cytoskeleton and Ca²⁺ inflow, which in turn affects the localization and/or distribution of TJ, basal ES, GJ and desmosome proteins. Furthermore, studies are needed to examine if macroclustering of ICAM-1 occurs in Sertoli cells, such as during the transit of preleptotene spermatocytes at the BTB, or if Sertoli–germ cell interactions would induce cleavage of mICAM-1 on the Sertoli or germ cell surface to generate sICAM-1 by MMPs, which could be used to mediate BTB ‘opening’ above the transiting preleptotene spermatocytes which are connected in clones via intercellular bridges (Fawcett, 1961). As depicted in Fig. 5, it is likely that during the transit of preleptotene spermatocytes at the BTB, ICAM-1 recruits TJ [e.g. occludin, claudins, zonula occludens-1 (ZO-1)], basal ES (e.g, N-cadherin, β-catenin, nectin-2, afadin) and GJ (e.g. connexin-43) proteins to assemble ‘new’ BTB at the basal region of the transiting spermatocytes, whereas sICAM-1 is being used to ‘disassemble’ the ‘old’ BTB at the apical region of the spermatocytes. It is also possible that ICAM-1 can recruit c-Src or Pyk2 to the site to confer proper phosphorylation of integral membrane proteins at the BTB (e.g. occludin, claudins). These kinases, in turn, can also regulate endocytic vesicle-mediated protein trafficking at the BTB (Cheng and Mruk, 2012), such as by inducing endocytosis, transcytosis and/or recycling of proteins at the BTB, so that integral membrane proteins can be recycled from the ‘old’ to the ‘new’ BTB site for junction assembly. In short, utilizing the antagonistic effects of ICAM-1 and sICAM-1 as demonstrated in in vitro and in vivo studies regarding their effects on the Sertoli cell TJ-permeability barrier function, the immunological barrier can be maintained during the transit of preleptotene spermatocytes at the BTB (Fig. 5).

Figure 5.

A schematic drawing illustrating a hypothetical model by which ICAM-1, sICAM-1 and ICAM-2 regulate the restructuring of the BTB to accommodate the transit of preleptotene spermatocytes across the BTB, and apical ES restructuring to facilitate spermatid movement across the seminiferous epithelium. In (A), sICAM-1 generated from ICAM-1 via cleavage by MMPs can modulate spermatid adhesion at the apical ES via its effects on F-actin organization. ICAM-2 can also form stable adhesion complex with nectin-3/afadin; however, ICAM-2 can also be involved in the degradation of the laminin-333 chains, generating the biologically active laminin fragments to induce apical ES degeneration at spermiation or to induce apical ES restructuring to facilitate spermatid transit across the seminiferous epithelium during spermiogenesis. In (B), cleavage of ICAM-1 by MMPs that generates sICAM-1 can facilitate the disruption of the ‘old’ BTB above the preleptotene spermatocytes in transit at the BTB, whereas the endocytosed adhesion protein complexes can be transcytosed and recycled to assemble ‘new’ BTB behind the transiting preleptotene spermatocytes, so that the integrity of the immunological barrier can be maintained. This model thus depicts the crucial physiological role of ICAM-1, sICAM-1 and ICAM-2 at the apical ES (A) and BTB (B) during the epithelial cycle of spermatogenesis.

ICAM-2: a functional molecule at the Sertoli–germ cell interface

In the rat testis, ICAM-2, analogous to ICAM-1, is expressed by the Sertoli and germ cells. However, its localization in the seminiferous epithelium is restricted to the Sertoli–germ cell interface, and is not found at the Sertoli cell–cell interface at the BTB (Xiao et al., 2012b). Subsequent analysis showed that ICAM-2 is an integrated component of the apical ES (Xiao et al., 2012b). Its expression at the apical ES displays highly restrictive spatiotemporal patterns during the epithelial cycle, which shifts from its diffuse localization surrounding the concave and convex sides of spermatid heads in early and late stages of the seminiferous epithelial cycle to the convex side of spermatid heads prior to spermiation (Xiao et al., 2012b). Similar to other epithelial cells, ICAM-2 associates with the actin cytoskeleton in the rat seminiferous epithelium. Surprisingly, it also physically interacts with nectin-3 (an IgSF CAM present in the testis, see Tables I and II), annexin II and c-Src (Xiao et al., 2012b). Annexin II is a Ca²⁺-dependent phospholipid-binding protein and a regulator of actin polymerization, and also a substrate for c-Src (Hayes et al., 2006; Hayes and Moss, 2009). Interestingly, during CdCl₂-induced germ cell sloughing from the seminiferous epithelium and BTB disruption (Setchell and Waites, 1970; Wong et al., 2004; Siu et al., 2009; Elkin et al., 2010), the protein levels of ICAM-2, annexin II and c-Src were all up-regulated after ∼6–16 h of CdCl₂ treatment, prior to extensive changes in the phenotypes in the seminiferous epithelium, such as germ cell exfoliation and BTB disruption (Xiao et al., 2012b), illustrating ICAM-2, together with annexin II and c-Src, may mediate cadmium-induced testis injury.

ICAM-2 is known to be constitutively expressed in endothelial and epithelial cells (Table II), and its deficiency causes reduced leukocyte TEM (Gerwin et al., 1999; Huang et al., 2006; Porter and Hall, 2009). Studies have proposed that ICAM-2 may serve as a ‘guide’ for leukocyte recognition of the endothelial TJ barrier before TEM (Porter and Hall, 2009; Woodfin et al., 2009). Thus, ICAM-2 may also function to direct the spermatid passage through the seminiferous epithelium via its interaction with nectin-3 and/or β1-intergrin (Xiao et al., 2012b).

It was shown that the lack of ICAM-2 in epithelia appeared to elicit cell apoptosis (Huang et al., 2005). In mammalian testes, ∼75% of germ cells undergo apoptosis during spermatogenesis to maintain the physiological homeostasis of the seminiferous epithelium (Billig et al., 1995; Shaha et al., 2010). Treatment of rats with CdCl₂ is known to induce germ cell apoptosis, degeneration of the apical ES, leading to irreversible infertility. Current studies have indicated that CdCl₂ may initially expedite branched actin polymerization activity, causing disruption of spermatid adhesion to Sertoli cells at the apical ES (Cheng et al., 2011b). During CdCl₂ treatment, ICAM-2 was induced and associated less tightly with actin (Xiao et al., 2012b). Recent studies have shown that ICAM-2 may function as an actin stabilizer since it was associated with F-actin until just prior to spermiation (Xiao et al., 2012b). Indeed, ICAM-2 expression in tumor cells has been shown to suppress metastasis by establishing an ICAM-2/α-actinin-actin linkage (Yoon et al., 2008). Moreover, annexin II, an inhibitor of actin polymerization (Hayes et al., 2006; Hayes and Moss, 2009), was found to form a protein complex with ICAM-2 and c-Src (Xiao et al., 2012b). These three proteins in the rat testis were observed to be synchronously induced after CdCl₂ treatment. It has been reported that c-Src phosphorylates adaptor protein(s) connected to ICAM-2, whose clustering would send out a survival signal that is sufficient to block TNFα-mediated apoptosis (Perez et al., 2002). Thus, it is logical to envision that activated c-Src may be recruited to the activated ICAM-2/annexin II complex at the early stage of CdCl₂-induced apical ES damage in order to protect spermatids from being depleted from the epithelium.

Figure 5 also depicts the likely involvement of ICAM-2, an apical ES protein in the rat testis, in coordinating the release of sperm at spermiation during Stage VIII of the epithelial cycle. It is likely that ICAM-2, surrounding the entire head of the elongating spermatid at Stage X–V of the epithelial cycle, and via its association with nectin-3, afadin, β1-integrin, c-Src, Pyk2 and annexin II, are being used to stabilize the F-actin bundles at the apical ES (Fig. 5). Also, ICAM-1, appearing to be an apical ES protein as well, can also stabilize the apical ES integrity. However, the localization of ICAM-2 begins to ‘shift’ spatially to the convex side of the elongating spermatids at late Stages V–VIII, perhaps being used to facilitate re-organization of the underlying actin-based cytoskeleton to prepare for spermiation (Fig. 5). In short, ICAM-2 and ICAM-1/sICAM-1 likely serve as “molecular switches” to co-ordinate the events of spermiation and BTB restructuring at Stage VIII of the epithelial cycle as depicted in the hypothetical model shown in Fig. 5.

Concluding remarks

Herein, we have briefly reviewed the role of ICAMs in epithelial/endothelial physiology in particular ICAM-1/sICAM-1 and ICAM-2 in cell adhesion and cell movement. They likely serve as molecular ‘switches’, triggering changes in the organization of F-actin in Sertoli cells in the seminiferous epithelium, relating outside signals into the cell to elicit appropriate responses to stimuli, such as at different stages of the epithelial cycle, or changes in the environment via their effects on cell cytoskeleton, eliciting changes in cell adhesive function. Based on recent findings in the field regarding the role of ICAMs in leukocyte TEM and their effects on the Sertoli cell TJ-barrier function, a hypothetic model regarding the involvement of ICAM-1/sICAM-1 and ICAM-2 in regulating spermiation and BTB restructuring during the epithelial cycle, such as Stage VIII, is proposed. It is understood that this model (Fig. 5) will be updated when more data become available in the field. Nonetheless, this model will be useful to investigators to design functional experiments to understand the role of ICAMs in the apical ES–BTB–hemidesmosome functional axis during spermatogenesis.

Authors' roles

C.Y.C. conceived the ideas of preparing this review article; X.X. and C.Y.C. performed the literature search, research on the topic and critically evaluated findings in the field; X.X., D.D.M. and C.Y.C. critically discussed and evaluated the latest research on different aspects of ICAMs and testis function; D.D.M. and C.Y.C. provided the first draft of the hypothetical model; X.X. and C.Y.C. prepared the figures; X.X. and C.Y.C. co-wrote the paper and C.Y.C. performed the final editing of the manuscript.

Funding

This work was supported by grants from the National Institutes of Health (NICHD, R01 HD056034 to C.Y.C. and U54 HD029990, Project 5 to C.Y.C.).

Conflict of interest

None declared.