Seminiferous Tubule Basement Membrane

Seminiferous tubule basement membrane (STBM) plays an important role in spermatogenesis. In the present study, the composition and structural organization of type IV collagen of bovine STBM was investigated. STBM was found to be composed of all six α-chains of type IV collagen based upon immunocytochemical and biochemical analysis. The content of α3(IV) chain (40%) and the α4(IV) chain (18%) was substantially higher than in any other basement membrane collagen. The supramolecular structure of the six α(IV) chains was investigated using pseudolysin (EC 3.4.24.26) digestion to excise triple-helical molecules, subsequent collagenase digestion to produce NC1 hexamers and antibody affinity chromatography to resolve populations of NC1 hexamers. The hexamers, which reflect specific arrangements of α(IV) chains, were characterized for their α(IV) chain composition using high performance liquid chromatography, two-dimensional electrophoresis, and immunoblotting with α(IV) chain-specific antibodies. Three major hexamer populations were found that represent the classical network of the α1(IV) and α2(IV) chains and two novel networks, one composed of the α1(IV)-α6(IV) chains and the other composed of the α3(IV)-α6(IV) chains. The results establish a structural linkage between the α3(IV) and α5(IV) chains, suggesting a molecular basis for the conundrum in which mutations in the gene encoding the α5(IV) chain cause defective assembly of the α3(IV) chain in the glomerular basement membrane of patients with Alport syndrome.

Basement membranes (BMs) 1 are thin sheet-like extracellular structures that compartmentalize tissues. They are substrata for cells of various organs and provide important signals for differentiation, maintenance, and remodeling of tissues. BM function is altered in acquired and genetic diseases, such as Goodpasture syndrome, an autoimmune disorder, Alport syndrome, a form of hereditary nephritis, and diffuse leiomyomatosis, a hereditary disease characterized by benign proliferation of smooth muscle. Type IV collagen, the major constituent of BMs has been linked to the pathogenesis of each of these disorders (1).
Type IV collagen has recently emerged as a family of triplehelical isoforms consisting of six genetically-distinct chains, designated ␣1(IV) to ␣6(IV) (1)(2)(3)(4)(5)(6)(7)(8). The entire coding sequences for all six human ␣(IV) chains and certain ␣(IV) chains from other species have now been determined . Their primary structures are similar. Each is characterized by a ϳ25-residue noncollagenous sequence at the amino terminus, a ϳ230-residue noncollagenous (NC1) sequence at the carboxyl terminus, and, between these sequences, a long collagenous domain of ϳ1400 residues of Gly-Xaa-Yaa repeats that, together with two other ␣(IV) chains, forms the triple helix. The first ϳ130 residues of the collagenous domain is called the 7 S domain and is involved in tetramerization of triple helical protomers, whereas the NC1 domain is involved in their dimerization. The collagenous domain is interrupted by more than 20 short nontriplehelical regions which are thought to increase flexibility of this collagenous region. Analysis of the ␣1(IV) and ␣2(IV) chains reveals that the Gly-X-Y sequences flanking these interruptions are atypical when compared with the remainder of the collagenous domain. The atypical flanking sequences may be important in stabilizing the triple helix and in the formation of polygonal networks in BMs (23).
The six ␣(IV) chains differ considerably with respect to tissue distribution. At the protein level, immunochemical studies have shown that the ␣1(IV) and ␣2(IV) chains have a ubiquitous distribution whereas the ␣3(IV), ␣4(IV), and ␣5(IV) chains have a restricted distribution in both human and rodent tissues (26 -29). At the mRNA level, the relative expression of the ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) chains varies greatly among a variety of human tissues, including a variation in the ratio of expression of the ␣3(IV) and ␣4(IV) chains (16,17). Thus, the chain composition of a BM may be tissue specific. Likewise, the kind of triple-helical isoform(s) and their supramolecular organization in BM may be tissue-specific.
The ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) chains have all been implicated in the pathogenesis of human diseases (1, 7, 30 -33). In Goodpasture syndrome, the ␣3(IV) chain is the target for the pathogenic autoantibodies. In Alport syndrome, the COL4A5 gene encoding the ␣5(IV) chain is mutated in the common X-linked form of the disease, and the COL4A3 and COL4A4 genes are mutated in the autosomal recessive form (1). In leiomyomatosis, the COL4A5 and COL4A6 genes are deleted in some patients. These mutations together with the restricted tissue expression of chains indicate specific biological functions of the ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) chains, including a vital role in the molecular sieve function of the glomerular BM and in differentiation of smooth muscle cells.
Knowledge of the tissue-specific composition and organization of ␣(IV) chains of several distinct BMs is of fundamental importance for elucidating the structure/function relationships of these chains and their role in mechanisms underlying diseases. To date, studies on organization have focused primarily on the BM of the renal glomerulus, owing to the well established role of glomerular BM in the molecular sieve function of the kidney and the loss of this function in Goodpasture and Alport syndromes. In the present study, the composition and organization of the ␣(IV) chains of seminiferous tubule BM (STBM) was investigated. STBM was chosen for study because it appears to have a special role in spermatogenesis, ultrastructural abnormalities are thought to lead to infertility, and temporal studies of the expression of type IV collagen chains show the unique expression of the ␣3(IV) chain at the initiation of spermatogenesis (34). The results reveal that STBM contains the highest percentage of the ␣3(IV) chain of any BM collagen thus far studied and that it exists in a novel supramolecular complex comprised of the ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) chains.
Immunocytochemical Localization of ␣(IV) Chains in Bovine Testis-Fresh bovine testes were obtained at a slaughterhouse and quickly frozen on dry ice. Cryostat sections of bovine testis were placed on polylysine-coated slides. The sections were fixed for 10 -20 min with cold acetone. To gain access to antigenic epitopes, the cryostat sections were washed in 0.15 M NaCl, 10 mM sodium phosphate, pH 7.4, and then treated with 6 M urea, 0.1 M glycine-HCl, pH 3.6, for 1 h at room temperature. The sections were washed in 0.15 M NaCl, 10 mM sodium phosphate, pH 7.4, and then blocked with 1:1 Blotto (40):10% normal goat or rabbit serum plus 0.1% protease inhibitor mixture (41), 0.1% Tween 20 in 0.15 M NaCl, 10 mM sodium phosphate, pH 7.4, for 20 min at room temperature. Primary antibodies were added for 2 h at room temperature. Normal serum of the same species as the secondary an-tibody was added at 2% along with 0.1% Tween 20 to the diluted primary antibody. Primary antibodies were rabbit anti-bovine ␣1(IV)/ ␣2(IV) NC1 (diluted 1/100 to 1/500), Alport syndrome alloantibodies eluted from a rejected transplanted kidney (36) (diluted 1/500 to 1/1000), and Goodpasture syndrome antibody (37) (diluted 1/50 to 1/100). Anti-peptide antisera specific for ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) NC1 were used at a dilution of 1/150 to 1/500. The sections were washed three times in 0.15 M NaCl, 10 mM sodium phosphate, pH 7.4, and then incubated in peroxidase-labeled goat-anti rabbit IgG (Sigma; diluted 1/1000 to 1/2500) or rabbit-anti human IgG (diluted 1/2000) for 2 h at room temperature. Again 2% normal serum was added to reduce nonspecific binding. The sections were washed three times in 0.15 M NaCl, 10 mM sodium phosphate, pH 7.4. Diaminobenzidine was used to develop a brown reaction product. The sections were then lightly counterstained with hematoxylin. Controls included the substitution of normal rabbit serum for anti-␣1(IV)/␣2(IV), or the use of no primary antibody instead of Alport syndrome alloantibodies. Sections were examined with a Zeiss Axioskop and photographs made with Kodak Ektar 25 film using an 80A filter.
Preparation of STBM-The protocol for preparing STBM sleeves was as described by Enders et al. (41), with minor modifications. Testes were decapsulated, ground with a meat grinder, and blended to a liquified state with distilled water, 0.05% sodium azide and then shaken for 1 h on ice and pelleted by centrifugation at 8,000 ϫ g for 10 min. The pellet was washed three times with the same solution and then suspended in 1 M NaCl containing 200 Kunitz units/ml DNase, 2 mM phenylmethylsulfonyl fluoride, 0.1% protease inhibitors (42) and gently shaken for 90 min at room temperature. The solution was pelleted by centrifugation at 8,000 ϫ g for 10 min and the pellet resuspended in 1% sodium deoxycholate and gently shaken for 1 h on ice. The resulting pellet was washed three times with distilled water.
Digestion of STBM with Pseudomonas aeruginosa Pseudolysin-STBM (1 g) was digested with 0.5% pseudolysin at 4°C for 24 h, and the reaction was arrested by the addition of 20 mM EDTA as described previously (38). The pseudolysin-soluble fraction was separated from enzyme and EDTA by twice precipitating with 20% NaCl. About 20 -30 mg of the resulting precipitate was dissolved in 0.15 M NaCl, 50 mM Tris-HCl, pH 7.5, and fractionated on a Sephacryl S-1000 gel filtration column equilibrated with the same buffer to remove residual amounts of pseudolysin from type IV collagen components. The soluble fraction represented about 16% of the STBM. Both pseudolysin-soluble and pseudolysin-insoluble fractions of STBM were used for further study.
Electron Microscopy-The pseudolysin soluble STBM (20 g/ml) was sprayed in 0.15 M ammonium bicarbonate, 50% glycerol or 0.05 M acetic acid, 50% glycerol and further treated for rotary shadowing following established procedures (43). Samples were examined using a JOEL JEM-100CX II electron microscope.
Electrophoretic Analysis-One-dimensional SDS-PAGE was performed using the discontinuous buffer system described by Laemmli (45) in a 4 -22% gel gradient. For two-dimensional gel electrophoresis, the first dimension was performed according to O'Farrell (46), with some modifications (43), and the second dimension was SDS-PAGE using a 10 -22% gel gradient. Samples for electrophoretic analysis were either lyophilized or ethanol precipitated when necessary. Gels were either stained with Coomassie Brilliant Blue R-250 or silver as described by Morrissey (47).
Amino Acid Sequence Analysis-Amino-terminal amino acid sequence analysis for ␣3(IV) NC1 monomer was performed at the University of Kansas Medical Center Biotechnology Support Facility. A lyophilized ␣3(IV) NC1 monomer preparation was purified by C 18 reversed-phase HPLC, derivatized with phenylthiohydantoin, and then identified by use of a 470A protein sequencer with an on-line 120A PTH analyzer.
Immunoblotting-To perform immunoblotting, the SDS-PAGE separated proteins were transferred electrophoretically to nitrocellulose paper as described previously (43), then blocked with bovine serum albumin and reacted with primary and secondary antibodies by methods previously described (48).
Chromatographic Techniques-Quantitation of hexamer subunits was performed on a C 18 reversed-phase HPLC column as described previously (43). The identification of subunits in different pools from HPLC was performed by silver staining of two-dimensional electrophoretic gels and by immunoblotting with chain-specific antibodies.

Chain Composition of Type IV Collagen
The kind and relative amount of type IV collagen chains in STBM were determined by quantitative analysis of their NC1 domains. Previous studies established that the respective NC1 domains, in a hexameric configuration can be obtained from basement membrane by collagenase digestion, identified by two-dimensional electrophoresis using chain specific antibodies for immunoblots, and resolved and quantitated by C 18 reversed-phase HPLC (43).
The NC1 hexamer of STBM was purified under nondenaturing conditions by ion-exchange and gel filtration chromatography and then characterized by a variety of approaches. The hexamer eluted from a Sephracryl S-300 column ( Fig. 2A) in a position identical to that of NC1 hexamer from glomerular BM (43). Pool II was examined by rotary shadowing electron microscopy to confirm the integrity of the hexamers (Fig. 2B). The yield of hexamer was 6 -7 mg/g dry weight STBM. In the presence of 6 M guanidine-HCl (Fig. 2C) or 10% SDS (Fig. 2D), the NC1 hexamer dissociated into monomer and dimer subunits. The relative amounts of dimer and monomer were determined from the elution profile (Fig. 2C) to be 60 and 40%, respectively. Upon reduction of disulfide bonds in the presence of SDS, dimers completely dissociated into monomers (Fig. 2D).
STBM-NC1 hexamer was analyzed both by two-dimensional electrophoresis and by C 18 reversed-phase HPLC to identify and quantitate the kinds of NC1 domain subunits. The electrophoresis pattern revealed a complex set of Ͼ20 dimers and Ͼ10 monomers (Fig. 3A). The pattern is qualitatively similar to that of the NCl hexamer from glomerular BM (43), but differs in the relative abundance of the subunits. STBM-NC1 hexamer was dissociated with 0.1% trifluoroacetic acid and the subunits were resolved by HPLC into four fractions (I-IV) (Fig. 3B). NC1 subunits comprising each fraction were identified by two-dimensional electrophoresis on the basis of: 1) their migration positions (Fig. 4, A-D) in relation to that previously established for NC1 subunits of bovine glomerular BM (43); and 2) their binding to chain-specific antibodies (Fig. 4, E-L). Fraction I contained ␣1(IV), ␣2(IV), and ␣5(IV) NC1 monomers; fraction II contained ␣1(IV), ␣2(IV), and ␣5(IV) NC1 dimers; fraction III contained ␣4(IV) NC1 monomers and dimers and ␣6(IV) NC1 dimers; and fraction IV contained ␣3(IV) NC1 monomers and dimers and trace amounts of ␣5(IV) NC1. In each case, dimers have apparent size isoforms, and both monomers and dimers have charge isoforms. The identity of the ␣3(IV) NC1 monomers, designated ␣3aNC1 and ␣3bNC1 in Fig. 4D, was also determined by amino-terminal sequence analysis. The sequence of the first 12 residues of ␣3aNC1 and ␣3bNC1 was identical to that previously established for the ␣3(IV) subunits of bovine glomerular BM (4). Only subunits comprising fraction IV bound Goodpasture autoantibodies and Alport alloantibod- Note that most of the immunoreactivity using anti-peptide ␣3(IV) antisera also appears confined to the STBM (arrows, panel E). Antisynthetic peptide chain antisera specific for ␣4(IV) NC1 (panel F) and ␣5(IV) NC1 (panel G) gave similar restricted localization to the STBM. Anti-synthetic peptide chain antisera specific for ␣6(IV) NC1 localized to the STBM and larger blood vessel basement membrane (panel H).

Organization of Type IV Collagen in STBM
Pseudolysin-soluble STBM-STBM was digested with pseudolysin at 4°C for 24 h, which was previously established to solubilize truncated protomers of type IV collagen that retain a portion of the triple-helical domain and the complete NC1 domain (38). These truncated molecules were characterized by rotary shadowing electron microscopy (Fig. 5). They have a triple-helical (rod-like) segment, 287 nm in length, linked to a globular NC1 domain and are dimerized through NC1-NC1 interactions forming molecules with lengths of approximately 600 nm. The identity of chains comprising these molecules was determined by analysis of respective NC1 domains that were released upon collagenase digestion. The identity of the NC1 domain was determined by HPLC chromatography and two-dimensional gel electrophoresis and immunoblotting (Fig. 6) on the basis of the identity of the components established in Fig. 4B. The two-dimensional immunoblotting pattern indicated that the NC1 domain contains ␣1(IV) and ␣2(IV) NC1 monomers and dimers. The HPLC profile (panel A) supports this identity and shows that ␣1(IV) and ␣2(IV) NC1 domains comprise Ͼ95% of the NC1 domains, reflecting that the triple-helical molecules shown in Fig. 5 are comprised of ␣1(IV) and ␣2(IV) chains.
To determine which chains are connected through NC1-NC1 interactions, NC1 hexamers 3 released from the truncated molecules by collagenase digestion were fractionated using two monoclonal affinity columns (anti-␣1(IV) NC1 and anti-␣2(IV) 3 The NC1 domain is excised from the BM suprastructure by collagenase digestion isolated as hexamer. The NC1 hexamer is composed of subunits that correspond to the NC1 domain of the six ␣ chains comprising two adjoining isoforms (triple-helical molecules). The kind of stoichiometry of subunits in a hexamer depends on the chain composition of the isoform and the kind of isoforms associated through NC1-NC1 interactions (1).  NC1) using a strategy previously described (3,25). When the NC1 hexamer was applied to an anti-␣3(IV) NC1 affinity column (Fig. 7A), about 10% bound to the column. The chain identity of the respective NC1 domains was determined by two-dimensional gel electrophoresis (Fig. 7B) and immunoblotting (Fig. 7, C-G). The results revealed that the bound hexamer contained monomers of ␣1(IV), ␣2(IV), ␣3(IV), and ␣4(IV) NC1 domains and homodimers of ␣1(IV), ␣3(IV), and ␣4(IV) NC1 domains, but no ␣2(IV) NC1 dimer. Thus, this population(s) of NC1 hexamer is comprised of ␣3(IV) NC1 in association with ␣1(IV), ␣2(IV), and ␣4(IV) NC1. It represents 2% of the total NC1 hexamer of STBM and it is designated hexamer population A.
The unbound fraction was then applied to the anti-␣1(IV) affinity column and all of the sample bound to the column (Fig.   8A). The bound hexamer was analyzed by two-dimensional gel electrophoresis. The pattern of the two-dimensional spots was similar to that of the NCl hexamer composed of ␣1(IV) and ␣2(IV) NCl (Fig. 8B) in the immunoblot analysis. Furthermore, the bound fraction showed strong reactivity with the antibodies to ␣1(IV) and ␣2(IV) NC1 domain (Fig. 8C), but not with the antibodies to ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV)NCl (data not shown). Therefore, this NCl hexamer population, which did not bind to the anti-␣3(IV) affinity column but bound to the anti-␣1(IV) affinity column, contained only ␣1(IV) and ␣2(IV) chains. Thus, this population(s) of NCl hexamer A (Fig. 11) is composed of ␣1(IV) NC1 associated with an ␣2(IV) NC1. It represents 14% of the total NC1 hexamer of STBM and is designated hexamer population B.
Pseudolysin-insoluble STBM-The identity and organization of chains connected by NC1-NC1 interactions that comprise the pseudolysin-insoluble STBM was determined by analysis of the NC1 domains released upon collagenase digestion as described (see above) for the soluble STBM fraction. Insoluble STBM is comprised of 85% ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) NC1 monomers and dimers, and the remainder (Ͻ15%) of ␣1(IV) and (Fig. 3, panel B). HPLC fractions I-IV, shown in panel B (Fig. 3), were analyzed by two-dimensional electrophoresis and the gels stained for proteins by silverstaining (panels A-D) and for reactivity with chain-specific antibodies in panels E-L.  a Values for ␣1 ϩ ␣2, ␣3, and ␣4 were determined from an HPLC profile for TBM (Fig. 3) and includes both monomer and dimers of NC1 subunits. The value for ␣5 is a minimum value based on the amount in an affinity chromatography pool D (Fig. 11) (Fig. 6, C and D).

␣2(IV) NC1 monomers and dimers
The NC1 hexamer populations comprising this insoluble STBM were fractionated using the two affinity columns and the hexamer characterized by two-dimensional gel electrophoresis as was done for the soluble STBM (see above). When the NC1 hexamer was applied to the anti-␣1(IV)NCl affinity column, about 45% of the protein bound to the column (Fig. 9A). The bound fraction was analyzed by two-dimensional gel electrophoresis and the pattern of the spots was essentially identical to that of the whole NCl hexamer (Fig. 3). Furthermore, the two-dimensional immunoblots showed that the bound fraction reacted with antibodies to ␣1/␣2(IV), ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV)NCl at the monomer and dimer regions (Fig. 9, C-G). Thus, the NC1 hexamer that bound to the anti-␣1(IV) affinity column bound NCl hexamer contained all six ␣(IV) chains. It is composed of ␣1(IV) in association with ␣2(IV), ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) NC1 domains. It represents 35% of the total NC1 hexamer of STBM and is designated hexamer population C.
When the unbound fraction from the anti-␣1(IV) NCl affinity column was applied to the anti-␣3(IV) NCl affinity column, about 90% of the total protein bound to the column (Fig. 10A). The two-dimensional gel electrophoresis (Fig. 10B) and the two-dimensional immunoblot (Fig. 10, C-G) analysis revealed the presence of ␣3(IV) NCl-␣6(IV) NCl domains and the absence of ␣1(IV) and ␣2(IV) NC1 domains. Thus, this population(s) of NCl hexamer is composed of ␣3(IV) NC1 domains in association with ␣4(IV), ␣5(IV), ␣6(IV) NC1 domains. It represents 49% of the total STBM and is designated population D (Fig. 11). Analysis of this population by reversed-phase HPLC revealed it to be comprised of 66% ␣3(IV) NC1, 27% ␣4(IV) NC1 and ␣6(IV) NC1, and 7% ␣5(IV) NC1. The ␣4(IV) content represents the majority of the 27% of ␣4(IV) plus ␣6(IV) NC1 domains because ␣6(IV) NC1 represents only a small fraction, as is shown by the relative abundance of spots (Fig. 4). DISCUSSION STBM was found to be composed of all six ␣-chains of type IV collagen. The evidence was based upon immunocytochemical analysis of testis tissue and biochemical analysis of the NC1 domain hexamer that was released upon collagenase digestion of STBM. The ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) chains are usually regarded as the minor chains of BMs in relation to the abundant ␣1(IV) and ␣2(IV) chains. However, in the case of STBM, the ␣3(IV)-␣6(IV) chains collectively represent more than 80% of the chains, with the ␣3(IV) and ␣4(IV) chains as the major components. These results are in agreement with the supposition that BMs have a tissue-specific composition and organization of type IV collagen.
Pseudolysin cleavage of STBM was used to excise soluble triple helical molecules for determination of their chain organization, as reflected by compositions of NC1 hexamers that were subsequently released upon digestion with collagenase. Four distinct populations of NC1 hexamers were found, A and B from pseudolysin-soluble and C and D from pseudolysininsoluble STBM (Fig. 11). Analysis of populations A-D leads to several conclusions about the organization of the six ␣(IV) collagen chains. First, in population A, in an individual hexamer, one chain of the two triple-helical isoforms (connected this will not affect the qualitative interpretation of the results. Accordingly, the simplest interpretation of the affinity chromatography results is that there are seven dimeric species of isoform in the supramolecular structure of STBM collagen IV (Fig. 12, panel A); these dimers are designated NC1-linked isoforms.
NC1 hexamer population B reflects the classical organization in which the ␣1(IV) chain is linked to the ␣2(IV) chain. Its precursor is a pair of truncated protomers linked by NC1 domains (Fig. 5). It is similar in these respects to collagen IV from mouse EHS tumor matrix (49), which has a well characterized net-like supramolecular structure (reviewed in Ref. 50). There is published evidence that isoforms containing ␣1(IV) and ␣2(IV) chains can form an independent network in tissues that also contain the ␣3(IV)-␣6(IV) chains. Immunolocalization studies of the kidney show that only ␣1(IV) and ␣2(IV) chains are found in the mesangium and artery, whereas chains ␣1(IV) through ␣5(IV) are found in glomerular and tubular BM (26 -4 In population D, ␣5(IV) NC1 monomers and dimers are 7% by weight of the total NC1 content. A value of 16.6% would be required for all triple helical molecules to contain an ␣5(IV) chain. Thus, the high abundance of the ␣3(IV) NC1 (66%) and ␣4(IV) NC1 (27%) domains reflects a fraction of the triple helical molecules containing exclusively ␣3(IV) and ␣4(IV) chains with a presumed composition of [␣3(IV)] 2 -␣4(IV). 28). Thus, isoforms containing ␣1(IV) and ␣2(IV) chains must form an independent network in the kidney mesangium and artery, and perhaps form an independent network in glomerular and tubular BM. Further, circumstantial, support for an independent ␣1(IV)/␣2(IV) network is that these chains appear to have a ubiquitous occurrence in tissues, in contrast to the restricted distribution of the other ␣(IV) chains (1). In STBM, hexamer population B contains NC1 domains derived only from ␣1(IV) and ␣2(IV) chains. These results are consistent with an independent network that is composed of ␣1(IV) and ␣2(IV) chains. A portion of this hypothetical network is portrayed in Fig. 12, panel B. NC1 hexamer population C reflects a novel organization in which the ␣1(IV) chain is linked to ␣2(IV)-␣6(IV) chains, suggesting a network that is composed of all six ␣(IV) chains. A portion of this hypothetical network is portrayed in Fig. 12, panel C. These results corroborate the immunocytochemical findings ( Fig. 1) that all six ␣(IV) chains exist in STBM. The possibility of this complicated type of network had been realized when the ␣3(IV) and ␣4(IV) chains were discovered (51) and our results provide evidence for it in STBM. Similar results for glomerular BM provide evidence for it in that membrane too (3,25). NC1 hexamer population D reflects a novel organization in which the ␣3(IV) chain is linked to the ␣4(IV), ␣5(IV), and ␣6(IV) chains. At one extreme, the ␣3(IV) chain could be linked to five ␣4(IV) chains in the hexamer, forming two triple-helical isoforms that are connected through their respective NC1 domains, or to five ␣5(IV) chains or five ␣6(IV) chains. On the other extreme, the ␣3(IV) chain could be linked to five chains in the hexamer in which the ␣4(IV), ␣5(IV), and ␣6(IV) chain are components. In either extreme, population D links an ␣3(IV) chain with an ␣5(IV) chain. The results are consistent with a separate network composed of only ␣3(IV), ␣4(IV), ␣5(IV), and ␣6(IV) chains, as portrayed in Fig. 12, panel D. In summary, the three major populations B, C, and D suggest the existence of three distinct collagen IV networks. However, all three could be connected into one supra-complex through association of 7 S domains, i.e. tetramerization of the amino-terminal domain, thus forming a single network. If all three putative networks are, instead, part of a single network, then they would represent regions of that network. In addition, all six ␣(IV) chains contain cysteine residues in their collage- nous domains. This introduces another level of complexity in the STBM collagen supramolecular structure because it is possible that the various isoforms are cross-linked through disulfide bonds.
The structural linkage between an ␣3(IV) chain and an ␣5(IV) chain, as revealed by hexamer population D, is particularly noteworthy in view of the well described conundrum of the abnormality that occurs in patients with X-linked Alport syndrome. In these cases, mutations in the COL4A5 gene, encoding the ␣5(IV) chain, cause defective assembly of the ␣3(IV) chain in the glomerular BM (1,29,36,52,53). A structural linkage between the ␣3(IV) and ␣5(IV) chains suggests that the ␣5(IV) chain may be required for the incorporation of the ␣3(IV) chain into a triple-helical isoform containing both ␣3(IV) and ␣5(IV) chains or into two triple-helical isoforms, connected through NC1 interactions, in which the ␣3(IV) chain is contained within one isoform and the ␣5(IV) chain in the other. The dependence of one chain on the assembly of another chain is well established in the case of osteogenesis imperfecta in which mutations in the ␣1(I) chain cause defective incorporation of the ␣2(I) chain (54).
The structural linkage between the ␣3(IV) and ␣5(IV) chains in STBM raises questions about the molecular and fertility consequences of COL4A5 gene mutations in patients with Alport syndrome. The STBM serves several functions, including: (a) the site for spermatogonial proliferation, (b) support for the phenotypic differentiation of Sertoli cells, and (c) an aid in maintaining the seminiferous epithelium as an immunologically privileged site (55). What roles the various ␣(IV) chains have in these processes remains unknown. In a recent study of mouse testicular development, it was found that the genes for the ␣1(IV), ␣2(IV), and ␣5(IV) chains were expressed during embryonic development, whereas the ␣3(IV) and ␣4(IV) chains were expressed postnatally at a time which coincided with the initiation of spermatogenesis and expansion of the diameter of the STBM (34,41,56). At the protein level, the ␣3(IV) chain was found to be incorporated into a pre-existing STBM composed of the ␣1(IV) and ␣2(IV) chains. Whether the ␣5(IV) chain pre-existed with the ␣1(IV) and ␣2(IV) chains was not determined. These observations suggested that the ␣3(IV) chain is crucial for spermatogonial proliferation. However, recent work in which the ␣3(IV) chain was deleted in male mice resulted in apparently normal fertility (57). Assuming that COL4A5 gene mutations cause defective assembly of the ␣3(IV) chain in STBM, as they do in renal glomerular BM, and that the ␣3(IV) and ␣5(IV) chains are crucial for spermatogenesis, then male Alport patients would be infertile (58). However, most Alport patients are fertile and only class I Alport patients lack offspring. Hence, the fertility of Alport patients raises the question of whether the ␣3(IV) and ␣5(IV) chains are required for spermatogenesis in human beings or whether the structural relationship of the ␣3(IV) and ␣5(IV) chains in STBM differs from that of glomerular BM.
The detection of ␣6(IV) chains in larger blood vessels (not capillaries) and the STBM is consistent with ␣6(IV) chain expression by smooth muscle cells. Peritubular myoid cells have smooth cytoskeletal elements including desmin and smooth muscle ␣-actin (59,60). Defects in the ␣6(IV) chain have been associated with leiomyomatosis, a benign proliferation of smooth muscle cells (7). It should be interesting to determine if the thickening of the peritubular myoid cell layer which is associated with infertility (61-63) is also associated with defective or altered ␣6(IV) chain synthesis.
Acknowledgment-The technical assistance of Parvin Todd is greatly appreciated. isoform (E, unfilled) was obtained by electron microscopy (Fig. 5) of the collagen IV truncated isoform excised by pseudolysin at 4°C, and analysis of the NC1 hexamers derived from it (Fig. 11, populations A and B). Evidence consistent with an [␣3(IV)] 2 ␣4(IV) isoform (E, black filling) and an ␣3(IV)␣5(IV)[␣4(IV) or ␣6(IV)] isoform (D, black and white striped filling) was obtained by analysis of NC1 hexamer populations C and D (Fig. 11). Evidence consistent with the lower three NC1-linked isoforms in panel A was obtained by analysis of NC1 hexamer populations C and D (Fig. 11). In panels B, C, and D are hypothetical collagen IV octamers whose structures are consistent with the analysis of NC1 hexamer populations B, C, and D (Fig. 11), respectively. The octamers could occur either in separate networks or as part of a single network. It should be noted that the chain composition of the isoforms, and thus the isoform composition of the octamers, is not completely established.