Intercellular adhesion molecule-1 is a regulator of blood–testis barrier function

Throughout spermatogenesis, many critical biochemical, cellular and morphological events take place. These events allow germ cells to traverse the seminiferous epithelium while developing into elongated spermatids that are eventually released into the tubule lumen at late stage VIII of the seminiferous epithelial cycle during spermiation (Cheng and Mruk, 2002; de Kretser and Kerr, 1988; Mruk and Cheng, 2004; O’Donnell et al., 2011). For example, many different junction types are known to constitute the function of the blood–testis barrier (BTB, also known as the Sertoli cell barrier). These include tight junctions (TJs), basal ectoplasmic specializations (ES), gap junctions (GJs) and desmosomes, and they must disassemble intermittently during stages VIII–XI to allow zygotene spermatocytes to enter the adluminal compartment without compromising the homeostasis of the seminiferous epithelium and without affecting spermatogenesis (Cheng and Mruk, 2012; Mruk and Cheng, 2004). To achieve this, the testis has a unique mechanism in place where preleptotene/leptotene spermatocytes are briefly trapped within an intermediate compartment that is sealed at north and south poles by TJs, basal ESs, GJs and desmosomes (Mruk and Cheng, 2008; Russell, 1993a; Yan et al., 2008). The only way that spermatocytes can move upwards in the seminiferous epithelium is if these junctions, which are situated immediately in front of spermatocytes, are disassembled. Before this can occur, though, Sertoli cells must assemble a functional BTB behind spermatocytes, illustrating that an enormous amount of coordinated restructuring is taking place at the BTB (Cheng and Mruk, 2009; Cheng and Mruk, 2012). Unfortunately, the mechanism of junction assembly and disassembly at the BTB still remains incompletely understood.

In this report, we show for the first time that intercellular adhesion molecule-1 (ICAM-1) is a critical regulator of BTB dynamics. ICAMs are immunoglobulin (Ig)-like adhesion proteins residing largely on the cell surface, and five members (ICAM-1, -2, -3, -4 and -5) have been identified to date (Lawson and Wolf, 2009; Long, 2011; Muller, 2011). Of these, ICAM-1 is the best studied with respect to the movement of leukocytes across the endothelium during infection and inflammation where it is known to ‘seal’ the endothelium as migrating leukocytes cross it (somewhat analogous to the intermediate compartment in the testis), thereby maintaining barrier integrity (Dustin et al., 1986; Mamdouh et al., 2009; Nourshargh et al., 2010; Rothlein et al., 1986). A soluble form of ICAM-1 (sICAM-1), which is comprised of the entire or a portion of the extracellular domain of ICAM-1 and most likely derived from proteolytic cleavage of membrane-bound ICAM-1, has also been described to have important biological functions in other epithelia (Gearing et al., 1992; Lawson and Wolf, 2009; Marlin et al., 1990; Rothlein et al., 1991; van de Stolpe and van der Saag, 1996; Witkowska and Borawska, 2004). Herein, we show that Sertoli cell surface-associated ICAM-1 enhanced BTB function by facilitating cross-talk among different junction types, illustrating that ICAM-1 is working below spermatocytes to strengthen barrier function during stages VIII-XI of the seminiferous epithelial cycle. On the other hand, sICAM-1 was found to be a negative regulator of Sertoli cell barrier/BTB function, demonstrating that junction disassembly above migrating spermatocytes is facilitated in part by ICAM-1 cleavage.

A survey of different ICAM family members in selected rat organs was performed using RT–PCR (supplementary material Fig. S1; Table S1). Of these, Icam-1 and Icam-2 were expressed by the testis (supplementary material Fig. S1Aa,B), whereas Icam-4 and Icam-5 were restricted to the liver and brain, respectively (supplementary material Fig. S1C,D). Icam-1 RT–PCR results were verified by immunoblotting (supplementary material Fig. S1Ab). ICAM-1 was then localized in the adult rat testis by using a rabbit polyclonal antibody targeting the cytoplasmic domain of rat ICAM-1 (supplementary material Table S2) for immunohistochemistry (Fig. 1A). It should be noted that this antibody was found to cross-react with full-length (membrane-associated) ICAM-1 (97 kDa, referred to as ICAM-1 in this study), but it did not noticeably cross-react with soluble (extracellular) ICAM-1 (∼70 kDa). ICAM-1 immunoreactivity associated with Sertoli and germ cells at all stages of the seminiferous epithelial cycle (Fig. 1Ab–h). Discrete ICAM-1 staining was found to surround the heads of elongating and elongated spermatids at stages IX–XIII (Fig. 1Af–h). Moreover, ICAM-1 immunoreactivity was stage-specific at the site of the BTB (highest at stage VIII; Fig. 1Ab,d,e). No staining was observed when anti-ICAM-1 IgG was replaced with rabbit IgG (Fig. 1Aa). By immunoblotting, the monospecificity of the ICAM-1 antibody was assessed (Fig. 1B). A protein band corresponding to ICAM-1 was seen in lysates from Sertoli and germ cells, as well as in the testis, and these data were in agreement with RT–PCR results (Fig. 1C), as well as with previous reports from other laboratories (De Cesaris et al., 1998; Riccioli et al., 1995).

To expand the above findings, co-immunoprecipitation (co-IP) was performed to identify protein–protein interactions at the BTB (Fig. 2A). ICAM-1 structurally interacted with occludin (but not with claudin-11 or coxsackie and adenovirus receptor [CAR]), zonula occludens-1 (ZO-1), N-cadherin and β-catenin, as well as with actin, a cytoskeleton protein (Fig. 2A). It is worth noting that previous studies have shown all of these proteins to localize to the Sertoli cell barrier/BTB (Cheng and Mruk, 2012; Cyr et al., 1999; Hellani et al., 2000; Mruk and Cheng, 2004; Su et al., 2010a; Wang et al., 2007; Wong et al., 2004). To validate that ICAM-1 was indeed a constituent protein of the BTB, Sertoli cells (previously cultured for 4 days and having a functional barrier that mimicked the BTB in vivo) were fluorescently immunostained and analyzed by confocal microscopy (Fig. 2B). ICAM-1 was found to partially co-localize with occludin and N-cadherin at the Sertoli cell barrier when corresponding images were merged (Fig. 2Ba–h). This is in agreement with previous reports that used other models (Millán et al., 2006; Yang et al., 2005), as well as with subsequent experiments that used adult testes for immunofluorescent staining. Here, ICAM-1 partially co-localized with occludin and ZO-1, as well as with N-cadherin and β-catenin at the BTB (Fig. 2Ca–p). ICAM-1 immunoreactivity was most intense at the site of the BTB at stage VIII of the seminiferous epithelial cycle (Fig. 2Cb–d,f–h,j–l,n–p), consistent with immunohistochemistry results (Fig. 1A).

Recombinant sICAM-1 embodying rat extracellular ICAM-1 domains 2 and 3 was produced by Escherichia coli (supplementary material Table S1), and protein lysates obtained from soluble and insoluble fractions from both uninduced and induced bacterial cell cultures were resolved by SDS-PAGE. A predominant protein of 24 kDa was noted by Coomassie blue gel staining (Fig. 3A). Production of His₆-tagged recombinant sICAM-1 was verified first by a His₆ antibody (Fig. 3B) and second by a commercially available ICAM-1 antibody (data not shown) (supplementary material Table S2). This commercially available antibody, raised against the extracellular domain of rat ICAM-1, was predicted to cross-react with sICAM-1. Indeed, a protein of 24 kDa, corresponding to recombinant sICAM-1, was noted in agreement with experiments that used the His₆ antibody (data not shown). However, this antibody was not used for any of the other experiments reported in this study because its use yielded results of poor quality. When testis, Sertoli cell (cultured for 5 days in vitro) and germ cell (freshly isolated) lysates were used for immunoblotting in conjunction with the sICAM-1 antibody produced by us (supplementary material Table S2), a ∼70 kDa protein was observed in all samples (Fig. 3C). A 97 kDa protein corresponding to membrane-associated ICAM-1 (see Fig. 1B) was also weakly detected in all samples with the sICAM-1 antibody. It is worth noting that ICAM-1 and sICAM-1 levels were consistently lowest in Sertoli cells. Finally, proteins of lower M_r were also noted, and they may represent other sICAM-1 fragments, which may or may not exert biological activity in the testis. At this point, additional experiments would be needed to support this postulate (Fig. 3C).

When DNA constructs embodying either full-length ICAM-1 or sICAM-1 lacking transmembrane and cytoplasmic domains (Fig. 4A) were transfected into Sertoli cells at time 0, an increase in the Icam-1 mRNA level was detected by semi-quantitative (supplementary material Fig. S2Aa,b) and quantitative (supplementary material Fig. S2Ba,b) PCR when compared with Sertoli cells that were transfected with the pCI-neo/MOCK (control) plasmid. Co-IP and immunoblotting were also used to verify PCR results and to assess changes following ICAM-1 (Fig. 4Ba,b) and sICAM-1 (Fig. 4Ca,b) overexpression. Two days after ICAM-1 transfection, a significant increase in transepithelial electrical resistance (TER) was detected, which was maintained until the next day (Fig. 4Bc). By contrast, sICAM-1 overexpression resulted in a decline in TER, demonstrating a disruption of Sertoli cell barrier function (Fig. 4Cc). Specificity was addressed by overexpression of another ICAM family member, namely ICAM-2 (supplementary material Fig. S3). By immunofluorescent staining, ICAM-2 was found to localize to spermatocytes and elongated spermatids (supplementary material Fig. S3Aa,b). Interestingly, ICAM-2 did not co-localize with ZO-1 (supplementary material Fig. S3Aa,b), illustrating that ICAM-2 is not a constituent protein of the BTB. Following ICAM-2 overexpression, an increase in the ICAM-2 protein level was noted when compared with the corresponding control, illustrating successful overexpression in vitro (supplementary material Fig. S3B,a,b). As expected, ICAM-2 overexpression did not affect the integrity of the Sertoli cell barrier when its function was assessed by TER (supplementary material Fig. S3C). These results illustrate that the changes that we observed following ICAM-1 and sICAM-1 overexpression were specific to ICAM-1 and sICAM-1. Finally, it is worth noting that Sertoli cell viability was not significantly affected by Effectene® transfection reagent because TER readings did not decrease following transfection of pCI-neo/MOCK-containing plasmids (Fig. 4Bc,Cc).

To further investigate the disruptive effects of sICAM-1 overexpression on Sertoli cell barrier function, we investigated the steady-state levels of different integral membrane, scaffolding and regulatory proteins known to be present at the BTB (Fig. 5A,B). Immunoblotting revealed no changes in the steady-state levels of occludin, ZO-1, CAR, junctional adhesion molecule-A (JAM-A), claudin-11, α-catenin, β-catenin, desmocollin-3, p130Cas and focal adhesion kinase (FAK)/p-FAK-Y397/p-FAK-Y407 following sICAM-1 overexpression (Fig. 5A,B). However, sICAM-1 overexpression brought about a decrease in the levels of N-cadherin and γ-catenin, as well as in levels of the GJ protein connexin 43 and two tyrosine kinases, proline-rich tyrosine kinase 2 (Pyk2)/p-Pyk2-Y402 and c-Src/p-Src-Y530 (Fig. 5A,B). Thereafter, these findings were validated by immunofluorescent staining. While no changes were observed in JAM-A, claudin-11, CAR and ZO-1 localization (Fig. 5C), a loss in N-cadherin, γ-catenin and connexin 43 from the Sertoli-Sertoli cell interface was noted following sICAM-1 overexpression (Fig. 5C). These changes are in agreement with immunoblotting results (Fig. 5A,B). Finally, ultrastructural analysis by electron microscopy was performed on Sertoli cells 2 days after transfection with pCI-neo/MOCK- or sICAM-1- containing plasmids (Fig. 5D). Following pCI-neo/MOCK overexpression, the integrity of the Sertoli cell barrier was maintained. On the other hand, several breaks were observed at sites of cell–cell contact following sICAM-1 overexpression, illustrating that the integrity of Sertoli cell barrier was disrupted (Fig. 5D).

We next investigated the physiological role of sICAM-1 at the BTB by assessing barrier integrity following sICAM-1 overexpression in vivo. An increase in Icam-1 mRNA and protein levels was detected following sICAM-1 overexpression by semi-quantitative (Fig. 6A) and real-time (Fig. 6B) PCR, as well as by co-IP and immunoblotting (Fig. 6C,D) when compared with testes that were transfected with the pCI-neo/MOCK plasmid, illustrating successful overexpression in vivo. Following sICAM-1 overexpression, drastic morphological changes were observed 2–3 days after the last administration of DNA (Fig. 6E). These changes included the loss of spermatocytes and round spermatids from the seminiferous epithelium, and an increase in seminiferous tubule diameter (Fig. 6E). Also, elongated spermatids were mis-oriented within the epithelium (Fig. 6F). It is worth noting that obvious morphological changes in testes were noted in ∼30% of animals when cross-sections were examined microscopically. From these animals, cross-sections were obtained from the center of testes (near injection sites), and within each cross-section, ∼30% of tubules showed damage. No changes were observed after the administration of pCI-neo/MOCK-containing plasmids (Fig. 6E,F). A shortening of the seminiferous epithelium (in terms of height) was also observed (Fig. 6E–G). While an increase in seminiferous tubule diameter was observed following sICAM-1 overexpression (Fig. 6E,F), there was no significant increase in testis weight because this was offset by germ cell loss (data not shown). The reason for this increase in seminiferous tubule diameter is not immediately known. Magnified images from this experiment are also shown in supplementary material Fig. S4.

Finally, BTB function was assessed following sICAM-1 overexpression in vivo. For the positive control (CdCl₂), inulin-fluorescein isothiocyanate (FITC) was found to penetrate the BTB and to enter the adluminal compartment of the seminiferous epithelium (Fig. 6Ha,b), illustrating BTB disruption. This is in line with previously published reports, which showed CdCl₂ to disrupt BTB integrity (Durieu-Trautmann et al., 1994; Elkin et al., 2010; Prozialeck et al., 2006; Siu et al., 2009; Wong et al., 2004). For the negative control (pCI-neo/MOCK), inulin-FITC was restricted to the basal compartment (Fig. 6Hc,d). When sICAM-1 constructs were transfected into testicular cells, inulin-FITC was found to enter the adluminal compartment 2 days after the last administration of DNA (Fig. 6He,f). Images were analyzed by measuring the distance traveled by inulin-FITC (D_FITC) and by measuring the radius of the seminiferous tubule (Dr). Data are presented as a ratio of the two measurements (Fig. 6I). The diffusion of inulin-FITC into the adluminal compartment in testes from rats that received sICAM-1-containing plasmids demonstrates BTB disruption (Fig. 6H,I). Ultrastructural analysis by electron microscopy also showed sICAM-1 overexpression to disrupt the BTB in vivo (Fig. 6J).

Since sICAM-1 overexpression in vitro decreased the steady-state levels of N-cadherin and γ-catenin (Fig. 5), we decided to expand part of these results to the in vivo level (Fig. 7). Immunoblotting revealed no changes in the steady-state levels of occludin, CAR, desmoglein-2 and desmocollin-3 (Fig. 7A), consistent with in vitro data (Fig. 5). However, sICAM-1 overexpression in vivo brought about a decrease in the levels of N-cadherin and connexin 43 (Fig. 7A,B). Following sICAM-1 overexpression in vivo, no changes in occludin localization at the BTB were observed when compared with testes that were transfected with the pCI-neo/MOCK plasmid (Fig. 7Ce–h versus Fig. 7a–d). On the contrary, N-cadherin staining at the BTB became thinner and discontinuous (Fig. 7Cm–p versus Fig. 7i–l). Immunofluorescent staining of F-actin, a critical scaffolding protein, was also performed. In control testes, F-actin localized to the BTB and apical ES (Fig. 7Da–c) as previously described (Kopera et al., 2009; Sarkar et al., 2008). Following sICAM-1 overexpression, a loss of F-actin was observed at the BTB (Fig. 7Dd–f), illustrating BTB disruption. No changes in F-actin localization were noted at the site of the apical ES (Fig. 7Dd,e). Paraffin-embedded testes from overexpression experiments were also immunostained for desmoglein-2 (Fig. 7E). In control testes, desmoglein-2 localized largely to spermatocytes and early spermatids (Fig. 7Ea,a1). However, no obvious changes in desmoglein-2 localization were noted following sICAM-1 overexpression (Fig. 7Ec,c1,c2). No staining was observed when anti-desmoglein-2 IgG was replaced with rabbit IgG (Fig. 7Eb). Finally, we summarize some of the findings reported in this study in a schematic illustration, which depicts the role of ICAM-1 and sICAM-1 at the BTB (Fig. 7F).

Herein, we report on the role of ICAM-1 in the mammalian testis. ICAM-1 is an Ig-like, membrane-bound adhesion molecule expressed by several cell types, including Sertoli and germ cells in the testis (De Cesaris et al., 1998; Wong et al., 2000), and it is well known to function in the transendothelial migration of leukocytes (Lawson and Wolf, 2009; Long, 2011; Mamdouh et al., 2009; Muller, 2011). Immunolocalization studies showed ICAM-1 to be present at the site of the BTB, where its level was highest at stage VIII of the seminiferous epithelial cycle, coinciding with two critical cellular events that occur at this stage: the movement of preleptotene/leptotene spermatocytes across the BTB and the release of elongated spermatids into the tubule lumen (de Kretser and Kerr, 1988; Mruk and Cheng, 2004; Russell, 1993b). Weaker ICAM-1 immunoreactivity was also observed throughout the remaining seminiferous epithelium, as well as with elongating/elongated spermatids, and these observations are in line with ICAM's proposed roles in cell adhesion and cell migration (Lawson and Wolf, 2009; Long, 2011; Muller, 2011). Additional immunolocalization studies showed ICAM-1 to partially co-localize with BTB constituent proteins such as occludin, ZO-1, N-cadherin and β-catenin. Together with results obtained from co-IP experiments, which illustrated ICAM-1 to structurally associate with occludin-ZO-1 and N-cadherin-β-catenin protein complexes, we conclude that ICAM-1 is an integral component of the BTB where it probably contributes to barrier integrity. Indeed, ICAM-1 overexpression in Sertoli cells in vitro was found to enhance TJ permeability barrier function. These findings are intriguing for several reasons. First, they support the premise that BTB function is not constituted by TJs alone (in contrast to most other blood–tissue barriers) but by co-existing and co-functioning TJs, basal ESs, GJs and desmosomes where ICAM-1 plays a structural and/or signaling role. Second, they indicate that ICAM-1 may upregulate barrier function via cross-talk with occludin-ZO-1 and N-cadherin-β-catenin protein complexes. In addition, ICAM-1 was also found to associate with actin, suggesting that ICAM-1 may be involved in the remodeling of the Sertoli cell cytoskeleton, which is crucial for BTB integrity during preleptotene/leptotene spermatocyte movement. While ICAM-1^−/− mice were viable, relatively healthy (for instance, their immune responses were impaired) and fertile (Sligh et al., 1993; Xu et al., 1994), an in-depth examination of BTB function in these animals is lacking. It is possible that BTB integrity is disrupted in a small percentage of seminiferous tubules such as those at stages VIII–XI but that mice remain fertile. In conclusion, it is possible that ICAM-1 is functioning to ‘seal’ the BTB during the movement of germ cells, somewhat similar to what occurs during the passage of leukocytes across the endothelium, but additional studies would need to be performed to substantiate this hypothesis.

While ICAM-1 is known to function in cell adhesion, significantly less is known about the role of sICAM-1. sICAM-1, which is derived from the proteolytic cleavage of membrane-bound ICAM-1, has been suggested in some biological contexts to have a role that is opposite to that of ICAM-1 (Clark et al., 2007). Interestingly, Sertoli cell barrier/BTB function was found to be compromised in vitro and in vivo following sICAM-1 overexpression, illustrating that sICAM-1 can negatively regulate barrier function. This transient disruption in barrier dynamics was mediated by changes in the steady-state levels of basal ES (N-cadherin, γ-catenin) and GJ (connexin 43) proteins, which are known to be critical for BTB function, as well as by changes in tyrosine kinases (Pyk2/p-Pyk2-Y402 and c-Src/p-Src-Y530). For instance, a previous study showed the Sertoli cell TJ permeability barrier to be disrupted following downregulation of connexin 43 and plakophilin-2 (a desmosomal plaque protein that associates with connexin 43) (Li et al., 2009), demonstrating that GJs are critical for BTB function. ICAM-1 is also hypothesized to be regulated by Src and Pyk2. Src and Pyk2 are known to bind each other and to orchestrate the cascade of signaling events leading to cell movement (Gelman, 2003), suggesting that Src-Pyk2 may also be important in germ cell movement in the testis. Moreover, administration of PP1, a c-Src inhibitor, resulted in the loss of spermatocytes and round spermatids (but not elongating/elongated spermatids) from the seminiferous epithelium (Lee and Cheng, 2005). These results are similar to our findings following sICAM-1 overexpression in vivo when germ cell sloughing was noted. Taken collectively, our data illustrate that sICAM-1 may disrupt basal ESs and GJs situated above preleptotene/leptotene spermatocytes, resulting in Sertoli cell cytoskeleton remodeling, BTB disassembly and germ cell movement (Fig. 7F). In the testis, proteolytic cleavage of membrane-bound ICAM-1 may be mediated by matrix metalloprotease-9 (MMP-9), similar to reports in other epithelia and endothelia (Grenier and Bodet, 2008), resulting in the release of sICAM-1 and in the timely disruption of BTB disassembly. This cleavage may be triggered by cytokines such as transforming growth factor-β (TGF-β), tumor necrosis factor α and interleukin-1α (IL-1α) or sICAM-1 may be working with cytokines to regulate BTB dynamics because cytokines have been shown to disrupt barrier function (Lie et al., 2011; Lui et al., 2001; Lui et al., 2003). Indeed, ICAM-1 is known to be upregulated by pro-inflammatory cytokines (Inoue et al., 2011; Turner et al., 2011; Wertheimer et al., 1992). Another pool of sICAM-1, produced by alternative splicing (Wakatsuki et al., 1995), may also exist in the seminiferous epithelium, and this pool may be interfering with ICAM-1-mediated cell adhesion above preleptotene/leptotene spermatocytes (Fig. 7F). While additional studies are needed, our results provide a framework upon which new experiments can be designed with the aim of better understanding the biology of the testis.

A female New Zealand white rabbit (3 kg b.w.) and male Sprague-Dawley rats at 90 (300 g b.w.) and 20 days-of-age were purchased from Charles River Laboratories (Kingston, NY). The use of animals was approved by The Rockefeller University Laboratory Animal Use and Care Committee (Protocol Numbers 09-016 and 12-506), and all experiments were conducted in accordance with ethical guidelines. Seminiferous tubules and total germ cells were isolated from testes of adult rats as previously described (Aravindan et al., 1996; Zwain and Cheng, 1994).

Sertoli cells were isolated from testes of 20-day-old rats and cultured in DMEM/F12 (Sigma-Aldrich, St Louis, MO) supplemented with growth factors and an antibiotic (Cheng et al., 1986; Mruk et al., 2003). Cells were seeded at a density of 0.5×10⁶ cells/cm² in 12-well plates for lysate preparation, RNA isolation and overexpression experiments, at a density of 1.2×10⁶ cells/cm² in Millicell inserts (Millipore Corp., Billerica, MA) for transepithelial electrical resistance (TER) measurements and overexpression experiments, or at a density of 0.05×10⁶ cells/cm² on round glass coverslips (18 mm; Thomas Scientific, Swedesboro, NJ) for immunofluorescent staining. It is worth noting that functional junctions were assembled and maintained in control Sertoli cells cultured at all densities used in this study, a conclusion that is based on nearly two decades of research from this and other laboratories (Chung and Cheng, 2001; Chung et al., 1999; Dym and Papadopoulos, 1992; Florin et al., 2005; Hadley et al., 1987; Janecki et al., 1991; Janecki and Steinberger, 1986; Lie et al., 2011; Lui et al., 2001; Siu et al., 2003; Su et al., 2010b; Tung and Fritz, 1993). All substrata were coated with Matrigel™ basement membrane matrix as previously described (Xiao et al., 2011). Cultures were treated with a hypotonic buffer 36 hr after plating cells to remove contaminating germ cells (Galdieri et al., 1981), thereby yielding Sertoli cells with a purity of ∼98% (Lee et al., 2004). Sertoli cell viability was assessed in random cultures throughout the entire study by using the Cell Proliferation kit (Roche, Indianapolis, IN).

Total RNA from tissues and cells was obtained by using TRIzol® reagent (Invitrogen, Eugene, OR). Genomic DNA was removed from all samples by using deoxyribonuclease I (amplification grade, Invitrogen), and this was followed by an RT reaction. S16 was incorporated into all semi-quantitative PCR experiments as an internal control (Xiao et al., 2011). For quantitative real-time PCR (qPCR), cDNAs were mixed with a SYBR® Green PCR master mix (Applied Biosystems Inc. [ABI], Foster City, CA), and the reaction was carried out by using an ABI 7900HT Fast Real-Time PCR system (Genomic Resource Center, The Rockefeller University). Data were analyzed by ABI SDS 2.3 software. The fold-change in mRNA was calculated according to the comparative C_T method (2^−ΔΔC_T method), and each data point was normalized against Gapdh. Supplementary material Table S1 lists the primer pairs and conditions that were used for all PCR experiments reported in this study.

Co-IP and immunoblotting were performed as previously described (Xiao et al., 2011). Approximately 600–800 µg of total protein derived from testis, seminiferous tubule or Sertoli cell lysate was incubated with 2 (commercially available) or 15 µg (produced by us) anti-ICAM-1 or anti-sICAM-1 IgG, respectively, followed by immunoblotting. Images were captured by using a Fujifilm LAS-4000 mini imaging system and analyzed in Multi Gauge software (version 3.1; Fujifilm, Valhalla, NY). Saturated images were not included in final statistical analyses. Each antibody was pre-screened in preliminary immunoblotting experiments by using lysates obtained from different testes to ensure that each antibody cross-reacted with its corresponding antigen of expected molecular weight. Also, in order to prevent overloading of gels with protein, which may have inadvertently obscured important changes in protein levels, preliminary immunoblotting experiments were used to pinpoint the quantity of lysate to be used for the detection of each antigen. In general, tissue lysates (including seminiferous tubule lysates) were used at 50–100 µg/lane, and cell lysates were used at 15–25 µg/lane, unless stated otherwise. Supplementary material Table S2 lists the antibodies and conditions that were used for all of the co-IP and immunoblotting experiments reported in this study. With regard to ICAM-1 and sICAM antibodies, this table also clearly designates which antibody was used for each experiment.

Immunohistochemistry and immunofluorescent staining were performed as previously described by using the Histostain-Plus detection kit or reagents from Molecular Probes®/Invitrogen (Eugene, OR) with minor modifications to routine laboratory protocols (Xiao et al., 2011). Images were captured by using either a color (DS-Fi1-U2) or a monochromatic (Ds-Qi1Mc-U2) camera attached to a Nikon 90i motorized microscope and NIS-Elements AR software (version 3.2; Nikon Instruments Inc., Melville, NY). Images were compiled in Adobe Photoshop CS3 Extended software (version 10.0.1; Adobe Systems, San Jose, CA). If needed, color and fluorescent images were lightly modified for brightness and/or contrast, and all images within a single experimental group were modified using identical parameters. Supplementary material Table S2 lists the antibodies and conditions that were used for all of the immunohistochemistry and immunofluorescent staining experiments reported in this study.

Confocal microscopy was performed as previously described (Lie et al., 2011) by using Sertoli cells seeded at a density of ∼0.75–1.0×10⁶ cells/cm² on Matrigel™-coated Transwell® permeable supports (24 mm; Corning, Lowell, MA). These cells were used for immunofluorescent staining as described above. Images were captured by using an inverted LSM 510 laser scanning confocal microscope and LSM 510 META software (Carl Zeiss, Thornwood, NY). Three-dimensional images were processed with ImageJ software (version 1.46, NIH), and the Richardson-Lucy deconvolution algorithm was used to reduce blur and/or noise associating with acquired images (Lucy, 1974; Richardson, 1972). Electron microscopy was performed as previously described (Lee and Cheng, 2003; Wong et al., 2004). Confocal and electron microscopy were performed at the Bio-Imaging and Electron Microscopy Resource Centers, respectively, The Rockefeller University.

Primer pairs were prepared as previously described (Beck-Schimmer et al., 1998) (supplementary material Table S1), and a cDNA encoding domains 2 and 3 of rat extracellular ICAM-1 was generated by PCR using AccuPrime™ Taq DNA Polymerase High Fidelity (Invitrogen). Based on results presented herein (supplementary material Fig. S1), ICAM-2 is the only other ICAM known to be expressed by the testis so that a homology search was performed. When the amino acid sequence corresponding to domains 2 and 3 of ICAM-1 was compared against the entire ICAM-2 sequence, a ∼38% similarity was noted, thereby illustrating minimal homology between proteins and little likelihood that the sICAM-1 antibody would cross-react with ICAM-2. Thereafter, this cDNA was cloned into the pET-46 Ek/LIC vector and expressed in Escherichia coli BL21 (DE3) cells (EMD Chemicals, Gibbstown, NJ). The authenticity of the insert was verified by Sanger DNA sequencing (Genewiz, South Plainfield, NJ). Expression was induced with 0.4 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) for 3 hr, and the production of His₆-tagged recombinant sICAM-1 was verified first by a His₆ antibody (Roche, Indianapolis, IN) and second by an ICAM-1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) (supplementary material Table S2). It should be noted that this commercially available ICAM-1 antibody was raised against the extracellular domain of rat ICAM-1. Inclusion bodies were extracted from the insoluble cell fraction by using a BugBuster™ Plus Lysonase™ kit (EMD Chemicals), and the final pellet containing purified inclusion bodies was resuspended in guanidine lysis buffer (0.1 M NaH₂PO₄ pH 8.5 at 22°C containing 10 mM Tris-HCl and 6 M guanidine HCl). The insoluble fraction extract was then purified by Ni ion affinity chromatography by using Ni Sepharose™ High Performance support (GE Healthcare) under denaturing conditions (0.1 M NaH₂PO₄ pH 8.0 at 22°C containing 10 mM Tris-HCl and 8 M urea). Recombinant sICAM-1 was stored in denaturing elution buffer (0.1 M NaH₂PO₄ pH 8.0 at 22°C containing 10 mM Tris-HCl, 8 M urea and 200 mM imidazole), and its purity was assessed by Coomassie Blue staining. For antibody production, gel slices containing ∼200 µg recombinant sICAM-1 were equilibrated extensively against PBS (10 mM NaH₂PO₄ pH 7.4 at 22°C containing 0.15 M NaCl). Gel slices were then sonicated/emulsified in Freund's complete adjuvant in a final volume of ∼600 µl, and this was administered subcutaneously to the back of a New Zealand female rabbit at three different sites with ∼200 µl being delivered per site. Three weeks thereafter the rabbit received a booster injection as described above, except that gel slices were sonicated/emulsified in Freund's incomplete adjuvant. The rabbit received yet another booster injection 3 weeks thereafter. Ten days thereafter blood was collected to obtain antiserum (Cheng and Bardin, 1987; Cheng et al., 1988b; Yan and Cheng, 2006), and IgG was isolated by (NH₄)₂SO₄ precipitation and DEAE-chromatography by using Macro-Prep® DEAE support (Bio-Rad Laboratories, Hercules, CA) (Cheng et al., 1988a; Page and Thorpe, 2001a; Page and Thorpe, 2001b).

Rat full-length ICAM-1, sICAM-1 and ICAM-2 cDNAs were amplified by PCR using AccuPrime™ Taq DNA Polymerase High Fidelity (Invitrogen) and corresponding primer pairs (supplementary material Table S1). For ICAM-1/sICAM-1 overexpression plasmids, XhoI and MluI restriction sites were introduced into sense and antisense primers, respectively. On the other hand, sICAM-1 was generated by a single nonsense mutation (G¹⁵¹¹→A¹⁵¹¹) within the ICAM-1 transmembrane domain (Kita et al., 1992), resulting in a stop codon (TGG→TAG). For the ICAM-2 overexpression plasmid, MluI and XbaI restriction sites were introduced into sense and antisense primers, respectively. PCR products were cloned into the pCI-neo mammalian expression vector (Promega, Madison, WI). Thereafter, resulting plasmids were transformed into NovaBlue Singles™ Competent Cells (EMD Chemicals), purified by using a HiSpeed® plasmid Midi kit (Qiagen, Valencia, CA) and sequenced to verify authenticity (GENEWIZ).

Sertoli cells were isolated and cultured as described above. On day 2, cells were transfected with empty pCI-neo vector (MOCK, control), ICAM-1, sICAM-1 or ICAM-2 plasmids using Effectene® transfection reagent (Qiagen). It should be noted that by this time in vitro, Sertoli cells had formed a polarized epithelium with functional TJs, basal ESs, desmosomes and GJs (Lee and Cheng, 2003; Lie et al., 2009; Siu et al., 2005; Wong et al., 2000), thereby mimicking the BTB in vivo (Byers et al., 1986; Chung et al., 2001; Chung et al., 1999; Siu et al., 2005; Steinberger and Jakubowiak, 1993; Yan et al., 2008). Thus, this excellent in vitro model has been, and continues to be, used by many investigators to study BTB dynamics. Pilot experiments were conducted to titrate the optimal quantity of DNA to be used for each application, and a 30-fold concentration range was tested (∼0.1–3 µg) for each plasmid. Approximately 1 µg and 0.3 µg DNA was used for Sertoli cells cultured in 12-well plates and Millicell inserts, respectively. On the following day, media were replaced (this time point was designated as day 1 after transfection), after which cells were harvested daily for three consecutive days in lysis buffer (10 mM Tris pH 7.4 at 22°C containing 0.15 M NaCl, 1% NP-40 [v/v] and 10% glycerol [v/v]; protease and phosphatase inhibitors were added at a 1∶100 dilution immediately before use) or TRIzol® reagent (Invitrogen). For immunofluorescent staining of Sertoli cells, plasmid DNA was labeled by using the Label IT® Tracker™ intracellular nucleic acid localization kit (Mirus Bio, Madison, WI) at a 0.5∶1 (v∶w) ratio of Tracker™ reagent (Cy3™) to DNA to monitor transfection efficiency, and ∼0.1 µg DNA was used for each transfection.

Contamination of plasmid preparations by bacterial endotoxins was removed by using the MiraCLEAN® endotoxin removal kit (Mirus Bio). In brief, adult rats (∼300 g b.w., n = 3–5) received one intratesticular injection of a transfection mixture each day for three consecutive days. Each intratesticular injection consisted of 10 µg sICAM-1 plasmid DNA in 200 µl TransIT-EE Hydrodynamic Delivery Solution (Mirus Bio) administered via a 28-G¹/₂ insulin syringe so that the total amount of DNA that was injected and uptaken by cells in the testis, including Sertoli and germ cells, was 30 µg. The control consisted of administering the same amount of empty pCI-neo vector (MOCK). In this experiment, only the left testis was injected with plasmid DNA, whereas the right testis was left untreated. Rats were sacrificed 2 and 3 days after the last intratesticular injection, and testes were removed for subsequent analysis. Pilot experiments established the optimal quantity of DNA to be used, as well as the time of animal sacrifice.

Adult rats (∼300 g b.w., n = 3–5) received one intratesticular injection of a transfection mixture each day for three consecutive days as previously described. Two days after the last administration of pCI-neo/MOCK- or sICAM-1-containing plasmids, rats received ∼1.5 mg inulin-FITC (Sigma-Aldrich) suspended in 300 µl PBS via the jugular vein under ketamine-HCl (60 mg/kg b.w.) and xylazine-HCl (15 mg/kg b.w.) anesthesia/analgesia as previously described (Kopera et al., 2009; Sarkar et al., 2008). Approximately 30 min thereafter, rats were euthanized by CO₂ asphyxiation, and testes were removed and snap-frozen in liquid nitrogen. Cryosections (relative thickness, 10 µm) were obtained, mounted with ProLong® Gold antifade reagent containing DAPI (4′,6-diamidino-2-phenylindole) (Invitrogen) and examined by fluorescent optics. The positive control consisted of treating rats (n = 2) with CdCl₂ [3 mg/kg body weight (b.w.), intraperitoneally (i.p.)] and using these animals for BTB integrity assays 3 days after the administration of CdCl₂. CdCl₂ was used as a positive control because its disruptive effects on BTB function have been described in previously published reports (Elkin et al., 2010; Wong et al., 2004).

Total protein concentration was determined by using the D_C protein assay kit (Bio-Rad Laboratories) and a microplate reader (model 680, Bio-Rad Laboratories) with BSA as a standard. Quantitative protein data was obtained by using Scion Image (version 4.0.3.2; Scion Corporation, Houston, TX). Hematoxylin and eosin staining of paraffin-embedded testes was performed by a routine laboratory protocol. Sertoli cell barrier function was assessed daily with a Millicell electrical resistance system (Millipore Corp.) as previously described (Chung and Cheng, 2001). TER measurements consisted of passing short pulses (∼2–3 sec) of current (20 µA) through the Sertoli cell epithelium using Ag/AgCl electrodes at 12, 3, 6 and 9 o'clock positions of the Millicell insert which were then averaged. Readings obtained from Matrigel™-coated ‘blank’ Millicell inserts (ones that did not contain any Sertoli cells) were then subtracted from each averaged value and subsequently multiplied by the effective growth surface area (0.6 cm²). F-actin was stained as previously described (Kopera et al., 2009; Mruk and Lau, 2009; Sarkar et al., 2008).

Comparisons between two experimental groups were performed by Student's t-test, whereas comparisons among multiple experimental groups were performed by one-way ANOVA, followed by Dunnett's test. GB-STAT software (version 7.0; Dynamic Microsystems, Silver Spring, MD) was used for all statistical analyses. Each in vitro experiment was repeated 3–5 times using different batches of Sertoli cells. Within each in vitro experiment, each treatment point consisted of Sertoli cells cultured in 12-well plates, Millipore inserts or on glass coverslips in triplicate, unless stated otherwise. All in vivo experiments consisted of n = 3–5 rats for each treatment point, except for rats that received CdCl₂ where n = 2. P<0.05 was taken as statistically significant.

We thank Dr Kunihiro Uryu at the Electron Microscopy Resource Center, The Rockefeller University, for his excellent assistance in electron microscopy.