Extracellular matrix components and testicular peritubular cells influence the rate and pattern of Sertoli cell migration in vitro

DEVELOPMENTAL

BIOLOGY

113,119-134

(1986)

Extracellular Matrix Components and Testicular Peritubular Cells Influence the Rate and Pattern of Sertoli Cell Migration in Vitro P. S. TUNG AND I. B. FRITZ Bunting

and Best

Department

of Medical

Research,, Received

University April

of Toronto,

10, 19X5; accepted

112 College July

Street,

Toronto,

Olztam’o

MsG

lL6,

Canada

5, 1985

We report the patterns of migration of Sertoli cells plated on specific substrata, and the influences of testicular peritubular cells on these processes. Data presented indicate that while peritubular cells readily spread when explanted onto Type I collagen, Sertoli cells do not. A delay of 4 to 6 days occurs after Sertoli cells are plated before they begin to migrate randomly to form plaque-like monolayers on Type I collagen. These processes are dependent upon the synthesis and subsequent deposition of laminin and/or Type IV collagen by Sertoli cells, and are independent of fibronectin. A different behavior occurs when reconstituted mixtures of purified Sertoli cells and peritubular cells are sparsely plated onto Type I collagen. Perituhular cells rapidly spread to form chains of cells between Sertoli cell aggregates. Sertoli cells then migrate on the surfaces of the peritubular cells, culminating in the formation of cable-like structures between aggregates. Evidence is presented that the Sertoli cell migration to form “cables” under these conditions is dependent upon fibronectin synthesized by peritubular cells, and is independent of the presence of laminin or Type IV collagen. We discuss the possible relevance of these data to the role which precursors of peritubular cells may play in determining the behavior of Sertoli cell precursors in viva during tubulogenesis, or in the remodelling of the seminiferous tubule which occurs during different stages of the cycle of the seminiferous epithelium in spermatogenesis. o 1986 Academic

Press, Inc.

both in vivo and in vitro. Peritubular myoid cells of rodents are thought to be derived from fibroblast-like cells in the fetal gonadal interstitium under the influence of adjacent Sertoli cells (Bressler and Ross, 1973). Peritubular cells and Sertoli cells in coculture interact with each other in a variety of ways, resulting in prolonged mutual survival in a medium containing no added proteins under conditions in which neither cell type would otherwise survive in monoculture (Tung and Fritz, 1980). Peritubular cells in coculture with Sertoli cells also sustain the production of androgen binding protein (ABP), a marker for Sertoli cell functions (Tung and Fritz, 1980; Hutson and Stocco, 1981). The addition of peritubular cell conditioned medium to Sertoli cells in culture has recently been found to increase the formation of ABP and transferrin, indicating that some of the influences on Sertoli cells can be mediated by proteins secreted by peritubular cells (Skinner and Fritz, 1985a). When Sertoli cells are plated on top of preexisting multilayers of peritubular cells, they spread along the upper surfaces of peritubular cells, forming plaque-like monolayers. Subsequently, a remarkable restructuring takes place, during which Sertoli cells reaggregate to form multinodular mounds. A limiting membrane is generated between individual nodules, comprised of Sertoli cell aggregates, and the surrounding ribbon of peritubular cells. This limiting membrane has morphological characteristics almost indistinguishable from those of a normal basal lamina. Further, the structure

INTRODUCTION

During mammalian embryogenesis, precursors of epithelial cells which are undergoing histogenesis often remain in direct apposition with mesenchymal cells throughout the induction period. Mesenchymal cell-epithelial cell interactions are apparently required not only for the differentiation of the tissue (Grobstein, 1967), but also for the stabilization of the phenotype at later stages of development (Bernfield and Banerjee, 1978). Mesenchymal-epithelial cell interactions which regulate development during embryogenesis could possibly continue to operate in modified form in adult life, particularly in organs such as the testis in which continuing cytodifferentiation is evident. Advanced germinal cells are continuously developing from stem cells in adult mammalian seminiferous tubules (Clermont, 1972), and considerable restructuring takes place as clones of germinal cells in early meiosis are translocated from the basal to the adluminal compartments of the seminiferous tubule (Russell, 1977). In this sense, processes occurring during spermatogenesis appear to be homologous to some of the cytodifferentiating processes taking place during organogenesis in fetal development (Roosen-Runge, 1977). Within this general context, mesenchymal cells in the boundary tissue of testicular tubules (peritubular myoid cells and fibroblasts) have been observed to interact specifically with neighboring epithelial-type cells (Sertoli cells) in a complex manner, 119

0012-1606/86 Copyright All rights

$3.00

0 1986 by Academic Press. Inc. of reproduction in any form reserved.

120

DEVELOPMENTAL BIOLOGY

of the nodules formed in vitro is grossly similar to that of germ cell-depleted seminiferous tubules in vivo (Tung and Fritz, 1980). The two cell types function cooperatively in the formation of components present in extracellular matrix (ECM) and in the basal lamina. Peritubular cells in monoculture synthesize collagens (Types I, III, and IV), and fibronectin, but no detectable laminin. In contrast, Sertoli cells in monoculture produce only Type IV collagen and laminin, but no detectable fibronectin (Tung et al., 1984, 1985b; Skinner et al., 1985). Each cell type synthesizes proteoglycans. However, different kinds are produced, and the form synthesized by Sertoli cells appears to be unique, since it contains both chondroitin sulfate and heparin sulfate glycosaminoglycans attached to the same protein core (Skinner and Fritz, 1985b). Deposition of all ECM components synthesized and secreted occurs to a greater extent during coculture of Sertoli cells and peritubular cells than during monoculture of either cell type (Skinner et al., 1985). These data are consistent with the observation that formation of a basal lamina-like structure takes place in vitro only when cells are in coculture (Tung and Fritz, 1980). In extending our examination of the interactions between Sertoli cells and peritubular cells, we have investigated the migration, pattern formation, and remodelling of Sertoli cells cultured on various substrata in the presence and absence of peritubular cells. In this communication, we present data demonstrating that the patterns of Sertoli cell migration are radically altered by the presence of peritubular cells. We offer interpretations of these observations, based largely upon data demonstrating the dependency of Sertoli cell migration on specific ECM components. Finally, we speculate upon the possible role of mesenchymal peritubular cells in the behavior of the epithelial-type Sertoli cell precursors during different developmental states in vivo. MATERIALS

AND

METHODS

Cell preparations and culture conditions. Sertoli cellenriched preparations, containing about 5% peritubular cells, were isolated from testes of ZO-day-old Wistar rats by sequential enzymatic digestion with trypsin and collagenase (Dorrington and Fritz, 1975; Tung et ah, 1975). In our discussion, we call this preparation a “conventional” Sertoli cell-enriched preparation. In contrast, purified Sertoli cell preparations, almost free of peritubular cells, were isolated by modified procedures recently described (Tung et al, 1984). Briefly, decapsulated testicular tissues from 20-day-old Wistar rats were minced, dispersed in Hanks’ buffer containing DNase, and trypsinized (Dorrington and Fritz, 1975). The re-

VOLUME 113,1986

sultant seminiferous tubule segments were then digested with a mixture of collagenase (1 mg/ml, Type I, Sigma, St. Louis, MO.) and bovine testicular hyaluronidase (1 mg/ml Type I-S, Sigma) in Hanks’ buffer for 30 min under gentle agitation at 32°C. Segments were washed vigorously in Hanks’ buffer using a large-bore Pasteur pipet, harvested by brief (5 min) low-speed (60 g) centrifugation, and further digested with hyaluronidase (1 mg/ml) alone for another 30 min. The resulting cell aggregates were washed twice more with Hanks’ buffer containing 1% bovine serum albumin (BSA, Sigma) and were then dispersed by vigorous agitation with a Pasteur pipet in Ca”‘-free Hanks’ buffer containing 0.1 mM ethyleneglycol bis (fi-aminoethyl-ether)-N,N’-tetraacetate (EGTA, Sigma). Finally, cell aggregates were washed with modified MEM (Tung et al, 1975) containing 1% BSA, and then suspended in MEM at desired cell densities before plating. Peritibular cells were isolated from testes of ZO-dayold Wistar rats according to procedures previously published (Tung and Fritz, 1977). Confluent primary cultures were maintained in modified MEM containing 10% calf serum for 4-6 days. Secondary cultures were then obtained by procedures previously described (Tung and Fritz, 1977; 1980). Sertoli cells were usually sparsely explanted at a density of loo-250 aggregates/100 mm’. In the purified preparations treated with hyaluronidase, each aggregate contained an average of 35 cells. In “conventional” preparations not treated with hyaluronidase, each aggregate contained an average of 100 cells. For confluent or dense cultures, we plated at a density of 5000 aggregates1100 mm’. Culture plates used included Linbro multiwell culture plates (Linbro, FB-16-24-7C), or bacterial culture dishes (Falcon), as specified in experiments under Results. Substrata employed for cell cultures included polystyrene, glass (coverslips), Type I and Type IV collagens, laminin, and fibronectin. Culture conditions were identical to those previously described (Tung et al., 1975). Cultures were maintained in modified Eagle’s minimal essential medium (MEM), supplemented with 10% calf serum or 5% heat-inactivated rabbit serum, as indicated under Results. Preparation of substrata. Type I collagen was solubilized by stirring rat tail tendon for 48 hr at 4°C in 1:lOO (v/v) acetic acid at a ratio of 100 ml/g tendon. The resulting solution was filtered through a sterile gauze and centrifuged at 20,000~ for 1 hr at 4°C. The supernatant fraction was collected, lyophilized, and stored at -30°C. For gelation, the collagen was reconstituted in 1:lOOO (v/v) acetic acid to water, at a final concentration of 0.5% (wt/vol). The Type I collagen solution (60 ~1) was evenly spread onto a 100 mm2 surface, and then exposed

TIJNG

AND

FRITZ

Sertoli

to ammonia fumes for 15 min. The gels in plates were washed, neutralized with Hanks’ buffer containing phenol red, sterilized under uv, and saturated with modified MEM. Laminin was a gift from Dr. H. Kleinman and Dr. G. Martin of the National Institutes of Health. Human plasma fibronectin was purchased from Calbiochem. Type IV collagen, generously provided by Dr. Jaro Sodek (Medical Research Council Group on Periodontal Physiology, University of Toronto) was dissolved in dilute acetic acid (1:lOOO V/V acetic acid to water). A fixed amount of the Type IV collagen solution (60 ~1, 1 mg/ ml, in Hanks’ buffer) was spread for each 100 mm2 surface area. Fibronectin and laminin were each dissolved in Hanks’ buffer at the same concentrations, and were plated in a similar fashion. The vessels were incubated overnight at 4”C, and surfaces were then washed with Hanks’ buffer prior to being used for cell culture. Antisera. Rabbit antisera directed against laminin and Type IV collagen were gifts from Dr. H. Kleinman and Dr. G. Martin of NIH. Rabbit antiserum against fibronectin (human plasma) was purchased from Calbiochem. For positive and negative controls, bulk absorption was carried out with lyophilized laminin, Type IV collagen, or fibronectin. Undiluted antiserum was absorbed as previously described (Tung and Fritz, 1978). Morphological techniques. Cell cultures were examined periodically using a Nikon inverted microscope with phase optics, and representative fields were photographed. For scanning electron microscopy (SEM), samples were fixed in 2% glutaraldehyde in 0.1 M Millonig phosphate buffer, postfixed in 1% osmium, criticalpoint dehydrated, and processed as previously described (Tung et al., 1976). For transmission electron microscopy (TEM), the samples were fixed in 3% glutaraldehyde in the same buffer and processed as in previous investigations (Tung et al., 1975). Immunolocalizatiorz. Deposition of ECM components was detected using the peroxidase-anti-peroxidase (PAP) procedures originally described by Sternberger (1979), and modified by Tung et al. (1985a). Briefly, cells supported by coverslips were lysed with distilled water containing 0.1 MNH40H for 5 min at room temperature. Coverslips were gently washed with several changes of phosphate-buffered saline (PBS), and then fixed with 3% paraformaldehyde in PBS for 30 min. Specimens were subsequently washed, preincubated for 30 min in PBS containing 1% normal goat serum, blotted, incubated with rabbit antisera in serial dilutions, reconstituted in PBS, and then washed (3 X 5 min in PBS). Affinity-purified F(ab’)z fragments of goat anti-rabbit IgG (Cappel Laboratories), diluted 1:50, were added onto the specimen, and incubated for 30 min. Coverslips were washed, and incubated for 30 min with PAP raised in rabbits

Cell

Migration

121

Patterns

(Cappel). For color development, washed specimens were finally incubated in PBS containing (3,3’)-diaminobenzidine tetrahydrochloride (Sigma) (0.5 pg/ml), and Hz02 (0.01%). Control specimens included nonimmunized rabbit serum or preabsorbed antisera instead of first antisera. The preparations were thoroughly washed, dehydrated, cleared, and mounted with or without counterstaining for microscopic examination by procedures previously described (Tung et al., 1985a). The presence of ECM components on cell surfaces was detected by similar immunolocalization procedures, using PAP, except that the cells were fixed intact. In investigations to determine intracellular localization, cells were permeabilized with acetone at -20°C for 7 min after fixation, and then subjected to procedures described above. RESULTS

Migration of Sertoli Cells or Peritubula,r Monocult~~re on Various Substratu

Cells i72

In confirmation of previous observations, enriched or purified Sertoli cell aggregates sparsely explanted onto polystyrene or glass coverslip spread to form plaques of monolayer within 1 day (Tung et al., 1975; Eddy and Kahri, 1976; Tung and Fritz, 1980). In contrast, Sertoli cell aggregates sparsely explanted onto Type I collagen remained within the initial aggregates for 4 to 6 days (Figs. lA, C). Although the spreading of Sertoli cells was delayed when Type I collagen was the substratum, peritubular cells in monoculture readily migrated randomly on this substratum, and rapidly formed a monolayer or multilayers, dependent on cell densities (Figs. lB, D). When explanted onto uncoated bacterial culture dishes, or Nucleopore filters having a pore diameter of 0.45 pm, hyaluronidase-treated preparations of purified Sertoli cell aggregates failed to adhere firmly by 24 hr after plating. When plated on bacterial culture dishes or Nucleopore filters previously coated with Type I collagen, Sertoli cells did attach, but migrated poorly if at all during the next 4 to 6 days. In contrast, Sertoli cells plated on bacterial culture dishes or Nucleopore filters previously coated with fibronectin, laminin, or Type IV collagen firmly adhered, and migrated in a random fashion, forming plaque-like monolayers within 24 hr which were similar to those formed on uncoated polystyrene or glass (Table 1). Luminin and Type IV Collagen Deposition, by Sertoli Cells Explanted onto Type I Collagen In purified Sertoli cell aggregates explanted onto Type I collagen and maintained in culture for 2 days, random

122

DEVELOPMENTAL

BIOLOGY

VOLUME

113, 1986

FIG. 1. Phase-contrast micrographs of conventional Sertoli cell-enriched preparations not treated with hyaluronidase (A, C), and peritubular cell preparations (B, D) maintained on polystyrene (A, B) and Type I collagen (C, D). All cultures were fixed 45 hr after plating. Arrows indicate peritubular cells contaminants spreading out from Sertoli cell-enriched aggregates. Sertoli cells formed contiguous monolayers on polystyrene (A) while Sertoli cells remained within aggregates on Type I collagen (C). Peritubular cells formed monolayers or multilayers on polystyrene (B) or on Type I collagen (D). All cells were prepared from testes of ZO-day-old rats, and were maintained in MEM containing 10% calf serum. Magnifications: X240.

deposits of fibril bundles (0.25 to 1 PMdiameter) became evident in over 90% of the aggregates (Fig. 2A). Few if any such fibrils were observed in comparable preparations of Sertoli cells plated onto uncoated polystyrene (Fig. 2B). Since fibrils were present on top of the gel surface, and on top of occasional germ cells present in the preparation (Fig. 2C), it appears unlikely that the fibrils were part of the Type I collagen gel added as substratum. They appeared to arise from Sertoli cells (Fig. 2D). Using antisera against laminin or against Type IV collagen for the determination of localization in lysed cell preparations, we have observed the deposition of these ECM components in cultures of Sertoli cells plated onto Type I collagen (Figs. 3B, D). Lesser amounts were also evident in Sertoli cells explanted onto uncoated polystyrene or glass surfaces (data not shown). In confirmation of previous observations (Tung et ah, 1984), no fibronectin was deposited by Sertoli cells prior to or after

plaque formation. In contrast, fibronectin, but not laminin, was detectable in peritubular cells cultured either onto Type I collagen, or onto uncoated polystyrene (Tung et ab, 1984; Skinner et al., 1985). Migration of Sertoli Cells Explanted on Type I Collagen is Suppressed by Antisera against Laminin or Type IV Collagen Purified preparations of Sertoli cell aggregates, depleted of peritubular cells, were plated on flat-bottom microtest plates coated with Type I collagen. Cells cultured in MEM containing 5% heat-inactivated rabbit serum began to spread to form plaque-like colonies in 4 to 6 days, in a manner similar to that observed in conventional Sertoli cell preparations plated on Type I collagen in multiwell culture plates (Fig. 1A). Addition to the medium of heat-inactivated antisera (5%) against laminin or Type IV collagen prevented cell migration

TUNG

SUMMARYOF

PATTERNSOF

MIGRATIONOF

AND

FRITZ

Sertoli

Cell

TABLE CELLSORPERITUBULAR

SERTOLI

1. Uncoated tissue 2. Above,

3. IJncoated

4. Bacterial (a) (b)

glass culture coated

coverslips plates with

bacterial

or polystyrene

Type

I collagen

culture

dish

culture dishes Type I collagen

coated

Fibronectin, laminin, type IS’ collagen

with

or

Readily within

Purified

Sertoli

attach, 24 hr

forming

123

Patterns

1 CELLSIN

Attachment Substratum

h4igratifm

MONOCULTURE

and migration

patterns

WHEN

PLATEDON~ARIOUS

of cells

in monocultures

cell aggregates plaque-like

SUBSTRATA

Peritubular

cells

monolayer

Readily attach, forming monolayer/ multilayers within 24 hr

Readily attach, but remain in initial aggregates during next 4-6 days; then slowly spread out randomly to form plaquelike monolayer

Readily attach, forming monolayer/ multilayers within 24 hr

Attach poorly evident

Slowly attach, with low plating efficiency and no spreading

if at all, and no migration

(a)

Slowly next

attach, and 4-6 days

(b)

Readily attach, forming plaque-like monolayers within 24 hr

from the aggregates during the following 14 to 16 days of culture (Table 2). In contrast, fibronectin antiserum did not affect the rate or pattern of migration of Sertoli cells cultured in the absence of peritubular cells. These observations indicate that the migration of Sertoli cells from peritubular cell-depleted aggregates, when explanted onto Type I collagen, is dependent on the presence of laminin and Type IV collagen. This conclusion is consistent with the observed absence of migration of Sertoli cells until ECM fibrils are deposited (Fig. 2). Cell Migration in Sertoli Cell-Peritubular Cell Cocultures

Small numbers of peritubular cells, present as contaminants in conventional Sertoli cell-enriched preparations not treated with hyaluronidase (Tung et al, 1984), readily migrated from aggregates, whereas the Sertoli cells did not (Fig. 1C). In these preparations, explanted in the presence of serum on Type I collagen, migrating and proliferating cells having a morphology characteristic of peritubular cells were frequently observed to orient themselves in a direction pointing toward adjacent aggregates (Figs. 4 and 5). Although the initial migration of peritubular cells from aggregates appeared to be random, subsequent movement of peritubular cells often resulted in the formation of a chain of cells which linked adjacent cell aggregates (Fig. 4). In these cases, the long axis of peritubular cells and their processes appeared to be oriented in such a manner that the shortest distance between two nearest aggregates was connected by a bridge of peritubular cells (Figs. 4 and 5).

do not migrate

during

Readily within

attach, 24 hr

forming

monolayers

Readily attach, forming monolayer/ multilayers within 24 hr

Ultrastructural examination of cells within initially formed bridges demonstrated that only peritubular cells were present in this location (Fig. 6B). In contrast, Sertoli cells constituted the only cell type remaining within the aggregates (Fig. 6A). Identification of both cell types was based upon ultrastructural criteria previously described (Bressler and Ross, 1973; Tung and Fritz, 1975; Tung et al., 1975; Tung and Fritz, 1977). One to two days after initial bridge formation between aggregates of cells plated on Type I collagen, some of the migrating peritubular cells flattened, forming attenuated lamellae (Fig. 7). On the upper surfaces of peritubular cells, cytoplasmic extensions of Sertoli cells became attached, and Sertoli cells then first became evident on the bridges between aggregates (Fig. 8A). In the absence of peritubular cells, few if any migrating Sertoli cells were detectable between aggregates. In the presence of peritubular cells, Sertoli cells spread out, migrated, and maintained a close approximation with adjacent migrating Sertoli cells. This resulted in a thickening of the initial bridge between aggregates, forming ultimately a cable-like structure (Fig. 9). These “cables” consisted of a layer of Sertoli cells lying on top of a carpet of peritubular cells (Fig. 8B). Using antisera against fibronectin in PAP immunolocalization experiments, we detected fibronectin on the plasmalemmal surface (Fig. lOB), as well as in the perinuclear region of peritubular cells (Fig. 10D). Fibronectin deposition was observed at all stages of cable formation. Extracellular deposition was also evident in peritubular cell monocultures or cocultures (Figs. lOE, F), and this deposition was enhanced in peritubular cells plated on Type I collagen. In confirmation of previous

124

DEVELOPMENTAL

BIOLOGY

VOLUME

113. 1986

FIG. 2. Scanning electron micrographs of hyaluronidase-treated Sertoli cell preparation plated on Type I collagen (A, C, D) and on polystyrene (B). Cultures were fixed 48 hr after plating. Note (A), Sertoli cells deposit fibrils when plated on Type I collagen (X340). (B) Sertoli cells readily spread out, but deposit no fibrils on polystyrene (x1530); (C) fibrils deposited on top of germinal cells (G) which had not been dislodged from Sertoli-enriched preparation cell (X6800). (D) Higher magnification of fibrils in (A), showing a fibril originating (arrow) in the cytoplasm of a Sertoli cell (X3400).

TIJNG

AND

FRITZ

Sertoli

Cell

Migrcrtiox

Patterns

125

C FIG. 3. Phase-contrast micrographs showing PAP immunolocalization of laminin and Type IV collagen deposition by purified Sertoli cell preparations cultured on Type I collagen. Cells were fixed 120 hr after plating, unlysed (A) or lysed (B-D). Antisera (1:lO) against laminin was added to cells in (B), and antisera against Type IV collagen was added to cells in (D). Preadsorbed anti-sera or nonimmune sera was added to cells in (C). All preparations in (B-D) were washed, and then incubated with PAP reagents, as described under Materials and Methods. (A) Shows untreated non-permeabilized cells (X380).

observations (Skinner et al., 1985), laminin tected in any of these locations. Fiboxectin Antiserum Changes the Pattern Migration of Sertoli Cells in Cocultures Maintained on Type I Collagen

was not de-

of

In Sertoli cell-peritubular cell cocultures maintained on Type I collagen, cable formation was evident when 5% nonimmune rabbit serum was present in the culture medium (Fig. 9). Addition of heat-inactivated antiserum

against fibronectin (5%) to the medium prevented this cable formation. Instead, plaque formation occurred, but only after deposition of laminin and Type IV collagen. Plaques formed in a manner similar to that observed in purified Sertoli cells aggregates plated on Type I collagen (Fig. 2). In contrast, antisera against laminin and Type IV collagen had no effect on cable formation in the cocultures. However, each of these antisera suppressed the formation of plaque-like monolayers by monocultures of Sertoli cells plated on Type I collagen (Table 2). These observations indicate that the migration of Sertoli cells

126

DEVELOPMENTAL BIOLOGY

VOLUME 113,1986

TABLE2 SUMMARVOFEFFECTSOFADDITIONOFANTISERA AGAINSTVARIOUS ECM COMPONENTSONTHE MIGRATIONOF MONOCULTURES OFSERTOLI CELLSAND PERITUBULAR CELLSPLATEDONTYPE I COLLAGEN Migration Antiserum (5%)added ECM components None,

Laminin,

against Purified

or fibronectin

or type

pattern of cells in culture

Sertoli

Remain in aggregates randomly to form IV collagen

Remain in aggregates beginning to spread

cell aggregates 4-6 days; plaque-like for 14-16 randomly

along the surfaces of peritubular cells plated on Type I collagen to form cable-like structures is not dependent on Type IV collagen or laminin. In contrast, in the absence of peritubular cells, Sertoli cell migration to form a plaque-like monolayer is dependent on these components when the substratum is Type I collagen. Migration of Sertoli Cells in Reconstituted Coculture System In cultures of hyaluronidase-treated Sertoli cell preparations depleted of peritubular cells and plated on Type I collagen, only plaque-like monolayer formation occurred (Fig. 2). If, however, peritubular cells were added to these preparations of purified Sertoli cell aggregates at the time of plating on Type I collagen, cable formation took place. The percentage of Sertoli cell aggregates which formed cable-like structures was directly proportional to the relative number of peritubular cells added, up to a ratio of peritubular cells to Sertoli cell aggregates of 0.16 (Fig. 11). In this reconstituted system, the addition of small numbers of peritubular cells (5 to 10%) to the purified Sertoli cell preparation resulted in the same patterns of cell migration as those observed in conventional Sertoli cell preparations not subjected to hyaluronidase treatment (Figs. 4-8). Such preparations are known to contain approximately 5% peritubular cells (Tung et al., 1984). DISCUSSION

Data presented indicate that purified Sertoli cell aggregates explanted onto laminin, Type IV collagen or fibronectin readily spread to form plaque-like monolayers, but that Sertoli cells plated onto Type I collagen remain in the initial aggregates for several days without migration. In the absence of peritubular cells, subsequent random spreading on Type I collagen to form plaque-like monolayers is dependent on the deposition of laminin and/or Type IV collagen. This conclusion is

Peritubular

cells

then spread monolayer

Migrate within

to form 24 hr

monolayer/multilayers

days

Migrate within

to form 24 hr

monolayer/multilayers

before

based upon the observations that laminin and Type IV collagen are deposited by Sertoli cells cultured on Type I collagen as substratum, and the addition to the medium of antiserum against laminin or Type IV collagen inhibits monolayer formation under these conditions. Although Sertoli cells do not readily migrate on Type I collagen, peritubular cells rapidly spread on this substratum. When aggregates containing both peritubular cells and Sertoli cells are explanted onto Type I collagen, peritubular cells initially migrate from the aggregates, and establish chains of cells between adjacent aggregates. Subsequently, Sertoli cells spread along the surface of the chain of peritubular cells, resulting in the formation of a cable-like structure between two adjacent aggregates. The preferential migration of Sertoli cells along this route is in sharp contrast to the random (plaque-forming) migration of purified Sertoli cells explanted onto Type I collagen in the absence of peritubular cells. Addition of antisera against laminin or Type IV collagen to the medium in which cocultures have been plated onto Type I collagen does not inhibit migration of Sertoli cells to form cable-like structures. However, cable formation in the cocultured system is inhibited by addition of fibronectin antiserum. We interpret combined observations to indicate that the preferential migration of Sertoli cells along the surfaces of peritubular cells is not dependent on laminin or Type IV collagen deposition, but it is dependent on fibronectin secreted by peritubular cells. In contrast, random (plaque-forming) migration of Sertoli cell explanted onto Type I collagen in the absence of peritubular cells is dependent on laminin and/or Type IV collagen deposition. In neither case do Sertoli cells migrate directly on Type I collagen (Tables 1 and 2). Peritubular cells in monoculture synthesize Types I, III, and IV collagens and fibronectin, but not laminin (Tung et al., 1984; Skinner et al., 1985). Sertoli cells in monoculture for periods up to 10 days secrete only Type

T~JNC

chain

of cells

between

aggregates

(X1620).

AND

FRITZ

Sertoli

Cell

Mip-atim

Patterns

127

DEVELOPMENTAL BIOLOGY

128

FIG. 5. Scanning in culture

in MEM

electron micrograph containing 10% calf

VOLUME 113,1986

of conventional Sertoli cell-enriched preparation serum for 12 hr after plating (X1600).

IV collagen and laminin, but not Types I and III collagens, or fibronectin. Deposition of ECM components in extracellular fibrils, and basement membrane formation in vitro, are dependent on cooperation between Sertoli cells and peritubular cells (Tung and Fritz, 1980; Tung et al., 1985b; Skinner et al., 1985). Data presented in this communication suggest that Sertoli cell aggregates explanted onto Type I collagen deposit more laminin and Type IV collagen than do Sertoli cells plated onto poly-

plated

on Type

I collagen.

The cells

were

maintained

styrene or glass. In Sertoli cell-peritubular cell cocultures, fibronectin secreted by peritubular cells may be involved in mechanisms which permit Sertoli cells to spread on peritubular cells. In addition, deposition of Type I collagen by peritubular cells may facilitate subsequent deposition of laminin and Type IV collagen secreted by Sertoli cells. These ECM components, in the presence of proteoglycans, are thought to interact to form organized basal lamina-like structures. It appears

TLJNG

AND FRITZ

Sertoli

Cell

Migrutim

Patterns

129

FIG. 6. Transmission electron micrographs of conventional Sertoli cell-enriched preparation within the cell aggregate (A), and in the chain of cells between aggregates (B). The population of similar cells in the aggregate has epithelial cell borders, and indented nuclei with clumped hereochromatin on the periphery, characteristic of Sertoli cells (A). The population of cells in the bridge between aggregates has dilated cisternae (arrowheads), dark myofibril bands (arrows), and cell borders characteristic of peritubular myoid cells (B). Conventional Sertoli cellenriched preparations were maintained in culture on Type I collagen for 24 hr. They were then fixed and sectioned vertically to the surface of the substratum (X5940).

130

FIG. ‘7. Scanning electron culture on Type I collagen (A), while the overlapping

DEVELOPMENTAL

BIOLOGY

VOLUME

113, 1986

micrographs of chain of cells between aggregates of conventional for 2 days. The extremely attenuated cell lamellae of flattened, of cell borders of attenuated cells is visible in (B) (X2040).

plausible that interactions of this sort may be required to mediate the formation of basal lamina-like limiting membranes observed to occur in peritubular cell-Sertoli cell cocultures (Tung and Fritz, 1980). In the seminiferous tubule, Sertoli cells are in intimate contact with the basal lamina. Maintenance of the characteristic histotype of Sertoli cells in culture is depen-

Sertoli elongated,

cell-enriched biopolar

preparation migrating cells

maintained are evident

in in

dent, at least in part, upon continual contact between Sertoli cells and the seminiferous tubule biomatrix (Tung and Fritz, 1984). Similar observations have recently been reported by Suarez-Quian et al. (1985). Substitution of single species of ECM components for seminiferous tubule biomatrix as substrata, however, does not sustain maintenance of the normal histotype of Ser-

FIG. 8. (A) Scanning electron micrographs of conventional Sertoli cell-enriched aggregates maintained in culture on Type I collagen for 36 hr. Note that cytoplasmmic processes from two adjacent aggregates are preferentially anchored on the upper surface of a peritubular cell (P) but not on Type I collagen (X7600). (B) Transmission electron micrograph of similar preparation maintained in culture for 120 hr showing a section of the interaggregate bridge cut vertically to the substratum surface. Note Sertoli cells (S) migrating on top of peritubular cells (P) and deposition of ECM. A large lipid droplet is indicated by L. (X7600).

TUNG

FIG. 9. A montage of phase-contrast in culture for 120 hr. Note the closely lamellae (X163)

AND

FRITZ

Serf&

Cell

M~~JT&WL

micrographs of conventional Sertoli cell-enriched apposing Sertoli cells in the cable-like structure

toli cells in culture (unpublished observations). Without an appropriately organized ECM, even in the presence of deposited laminin and Type IV collagen, Sertoli cells appear able only to migrate randomly, and to form plaque-like monolayers. The properties of fibronectin, a glycoprotein present in plasma, fibrous connective tissues, and basement membranes, have been extensively reviewed (Hynes, 1981; Kleinman et ab, 1981; Yamada, 1981). One important function of fibronectin involves the ability to direct cell movement. It is generally believed that during embryogenesis, the direction of migrating cells is regulated by tissue interactions, both systemic and local. Components of extracellular matrix, including fibronectin, are thought to be involved in the guidance of cells by providing suitable attachment sites. In vitro studies have shown that fibronectin promotes the adhesion, cytoskeleta1 organization, and migration of many cell types in culture, and that fibronectin is associated with cell migration in several systems during development (for review, see Hynes, 1981). In the current study we have

131

Pnftems

aggregates plated between aggregates.

on Type Arrows

I collagen indicate

and maintained peritubular cell

shown that fibronectin synthesized by peritubular cells can be detected on the upper surfaces of peritubular cell surfaces, and that Sertoli cells migrate on the fibronectin-coated cell surfaces to form cable-like structures. In this sense, the presence of fibronectin has the capacity to guide the migration of Sertoli cells. At the time of sex differentiation in mammals, primordial germinal cells in the genital ridge of the male embryo are clearly associated with precursors of Sertoli cells. Together, these two cell types form aggregates in the form of short, arch-like primary cords which are embedded in the interstitial mesenchyme (Jost et al., 1981; Byskov, 1982; Pelliniemi et al., 1981). Sertoli cell precursors and germinal cells within aggregates must then migrate and proliferate along certain directions to acquire a continuous tubular architecture (secondary cord formation). It is possible that specific guidance is provided by adjacent mesenchymal cells and/or ECM scaffolding deposited cooperatively by interacting Sertoli cell precursors and mesenchymal cells. During early male gonadogenesis, fibronectin and

DEVELOPMENTAL BIOLOGY

VOLUME113, 1986

FIG. 10. PAP immunolocalization of fibronectin in peritubular cell monocultures (A-E) and cocultures (F). (A and C) Control non-permeabilized cells, respectively, incubated with non-immune rabbit sera; (B) non-permeabilized peritubular cells incubated with fibronectin antiserum (1:lO); (D): permeabilized peritubular cells incubated with fibronectin antiserum (l:lO), (F, and F) lysed pcritubular cell and peritubular-Sertoli cell preparations, respectively, incubated with fibronectin antiserum (1:lO). Fibronectin was localized on the and permeabilized peritubular

TIJNG

a

0 NlJMBER

2 OF

PERITUBULAR

AND

Sertoli

Cell

4

6

8

CELLS

PLATED

FRITZ

Migration

20

10 PER

133

Patterns

SERTOLI

30 CELL

40

50

AGGREGATE

FIG. 11. Influence of relative numbers of peritubular cells on the formation of cable and plaques by purified Sertoli cell preparations. A fixed number of Sertoli cell aggregates (100 aggregates1100 mm’), containing approximately 35 cells/aggregate and a variable number of peritubular cells, indicated on the abscissa, were plated on Type I collagen and maintained in culture for 72 hr. The percentage of aggregates which joined to form cables, or which formed plaques, was calculated by counting 200 aggregates in five plates. Each aggregate was scored as having cable formation (o), plaque formation (O), or indeterminate merging (not included).

Type I collagen appear in the interstitial space long before the formation of a complete basal lamina (Paranko et ab, 1983), and before laminin and Type IV collagen can be immunohistochemically detected (unpublished data). It is possible that fibronectin is deposited along the interface between precursors of peritubular cells and Sertoli cells during primary and secondary cord formation, just as it occurs in cocultures (Skinner et al., 1985). Alternatively, mesenchymal cells or precursors of peritubular cells may present surfaces coated with fibronectin, thereby providing a scaffolding along which Sertoli cell precursors could preferentially spread to generate a cable-like continuous structure during secondary cord formation. In addition, extracellular de-

position of Type I collagen by peritubular cell precursors may facilitate the subsequent deposition of laminin, Type IV collagen, and other ECM components synthesized by Sertoli cell precursors, thereby leading to the formation of a basal lamina and associated extracellular structures in the boundary tissue of the seminiferous tubule. Processes described might also be involved in the lengthening of the seminiferous tubule during gonadal maturation. Insofar as interactions observed between peritubular cells and Sertoli cells in coculture represent a valid model for interactions occurring in vivo, it would appear that both cell-substratum and cell-cell interactions are involved in providing guidance for cell movements and

surfaces of intact peritubular cells (B); in the peritubular cell perinuclear region (D): in ECM maintained on polystyrene (E); and in ECM deposited by peritubular cell-Sertoli cell cocultures were lysed before fixation. In (A-D) and (F), peritubular cells were maintained on Type I collagen. In (F), cells were fixed 48 hr after Sertoli cells had been plated on top of preexisting multilayers 48 hr previously on Type I collagen (x221).

deposited by peritubular cell monocultures (F). Cell preparations shown in (E) and (F) In (A-E), cells were fixed 96 hr after plating. of peritubular cells, which had been plated

134

DEVELOPMENTAL

BIOLOGY

organization required for the completion of a tubulelike histotype (Tung and Fritz, 1980). On the basis of observations and speculations presented, we suggest that Sertoli cells and peritubular cells may form a functional unit that plays an integral role in the maintenance of the cytoarchitectural arrangements and the biochemical environment in the seminiferous tubule required for spermatogenesis to proceed (Fritz, 1978, 1984). These speculations remain to be evaluated in continuing studies on interactions between peritubular cells and Sertoli cells during gonadal morphogenesis, differentiation, and maturation. We are grateful to Dr. H. Kleinman and Dr. J. Sodek for their generous gifts of antibodies used in these investigations. We thank Edna Cartwright for excellent technical assistance, and Donna McCabe and Fern Teodoro for typing the manuscript. This work was supported by a grant (MT3292) from the Medical Research Council of Canada, to whom we express our deep gratitude. REFERENCES BERNFIELD, M. R., and BANERJEE, S. D. (1978). The basal lamina in epithelial-mesenchymal morphogenetic interactions. In “Biology and Chemistry of Basement Membranes” (N. A. Kefalides, ed), pp. 137 148. Academic Press, New York. BRESSLER, R. S., and Ross, M. H. (1973). On the character of the monolayer outgrowth and the fate of the peritubular myoid cells in cultured mouse testis. Exp. Cell Res. 78: 295-302. BYSKOV (1982). Primordial germ cells and regulation of meiosis. In “Germ Cells and Fertilization” (C. R. Austin and R. V. Short, eds.) pp. l-16. Cambridge Univ. Press, Cambridge. CLERMONT, Y. (1972). Kinetics of spermatogenesis in mammals: Seminiferous epithelial cycle and spermatogonial renewal. Physiol. Rev. 52,198-236. DORRINGTON, J. H., and FRITZ, I. B. (1975). Cellular localization of 5areductase and 3ol-hydroxysteroid dehydrogenase in the seminiferous tubules. Endocrinology 96,879-889. EDDY, E. M., and KAHRI, A. J. (1976). Cell association and surface features in cultures of juvenile rat seminiferous tubules. Anat. Rec. 185,333-358. FRITZ, I. B. (1978). Sites of action of androgens and follicle stimulating hormones on cells of the seminiferous tubule. In “Biochemical Actions of Hormones” (G. Litwack, ed), Vol. V, pp. 249-281. Academic Press, New York. FRITZ, I. B. (1984). Past, present and future of molecular and cellular endocrinology of the testis. In “Third European Testis Workshop” (J. Saez, M. G. Forest, A. Dazord, and J. Bertrand, eds.), Vol. 123, pp. 15-54. Les Editions Inserm. HUTSON, J. C., and STOCCO, D. M. (1981). Peritubular cell influence on the efficiency of androgen binding protein secretion by Sertoli cells in culture. Endocrinology 108,1362-1368. GROBSTEIN, C. (1967). Mechanisms of organogenetic tissue interaction. Natl. Cancer Znst. Monogr. 26, 279-294. HYNES, R. 0. (1981). Fibronectin and its relation to cellular structure and behavior. Zn “Cell Biology of Extracellular Matrix” (E. D. Hay, ed.), pp. 295-334. Plenum, New York. JOST, A., MAGRE, S., and AGELOPOULOU, R. (1981). Early stages of testicular differentiation in the rat. Human Genet. 58,59-63. KLEINMAN, H. K., KLEBE, R. J., and MARTIN, G. M. (1981). Role of collagenous matrices in the adhesion and growth of cells. .Z Cell Biol. 88,473-485.

VOLUME

113, 1986

PARANKO, J., PELLINIEMI, L. J., VAHERI, A., FOIDART, J.-M., and LAKKALA-PARANKO, T. (1983). Morphogenesis and fibronectin in sexual differentiation of rat embryonic gonads. Differentiation 23,572-581. PELLINIEMI, L. J., and LAUTEALA, J. (1981). Development of sexual dimorphism in the embryonic gonad. Human Genet. 58,64-67. ROOSEN-R~JNGE, E. C. (1977). “The Process of Spermatogenesis in Animals,” pp. l-200. Cambridge Univ. Press, Cambridge, UK. RUSSELL, L. D. (1977). Movement of spermatocytes from the basal to the adluminal compartment of the rat testis. Amer. .I Anat. 148, 313-328. SKINNER, M., TLJNG, P. S., and FRITZ, I. B. (1985). Cooperativity between Sertoli cells and testicular peritubular cells in the production and deposition of extracellular matrix components. J. Cell Biol. 100,941947. SKINNER, M., and FRITZ, I. B. (1985a). Testicular peritubular cells secrete a protein under androgen control that modulates Sertoli cell functions. Proc. N&l. Acad. Sci. USA 82,114-118. SKINNER, M., and FRITZ, I. B. (1985b). Structural characterization of proteoglycans produced by testicular peritubular cells and Sertoli cells. J. Biol. Chem. 260,11874-11883. STERNBERGER, L. A. (1979). “Immunocytochemistry,” 2nd ed. John Wiley & Sons, New York. SUAREZ-QUIAN, C. A., HADLEY, M. A., and DYM, M. (1985). Effect of substrate on the shape of Sertoli cells in vitro. Ann. N. Y. Acad. Sci. 438,417-434. TUNG, P. S., and FRITZ, I. B. (1975). Follicle stimulating hormone and cyclic AMP directed changes in ultrastructure and properties of cultured Sertoli cell-enriched preparations: Comparison with cultured testicular peritubular cells. In “Hormonal Regulation of Spermatogenesis” (F. S. French, V. Hansson, E. M. Ritzen, and S. N. Nayfeh, eds.), pp. 495-508. Plenum, New York. TUNG, P. S., DORRINGTON, J. H., and FRITZ, I. B. (1975). Structural changes induced by follicle stimulating hormone or dibutyryl cyclic AMP on presumptive Sertoli cells in culture. Proc. Natl. Acad. Sci. USA 72,1838-1842. TUNG, P. S., LIN, E. Y. C., and FRITZ, I. B. (1976). A scanning electron microscopic study of cultured cells prepared from rat seminiferous tubules. In “Scanning Electron Microscopy” (P. Johari and R. Becker, eds.), pp. 417-424. IIT Research Institute, Chicago. T~JNG, P. S., and FRITZ, I. B. (1977). Isolation and culture of testicular cells: A morphological characterization. Zrl “Techniques of Human Andrology” (E. S. E. Hafez. ed.), pp. 125-146. North Holland, Amsterdam. TUNG, P. S., and FRITZ, I. B. (1978). Specific surface antigens on rat pachytene spermatocytes and successive classes of germinal cells. Dev. Biol. 64, 297-315. TUNG, P. S., and FRITZ, I. B. (1980). Interactions of Sertoli cells with myoid cells in vitro. Biol. Reprod. 23, 207-217. TUNG, P. S., and FRITZ, I. B. (1984). Extracellular matrix promotes rat Sertoli cell histotypic expression in vitro. Biol. Reprod. 30,213-229. TUNG, P. S., SKINNER, M., and FRITZ, I. B. (1984). Fibronectin synthesis is a marker for peritubular cell contaminants in Sertoli cell-enriched cultures. Biol. Reprod. 30,199-211. TUNG, P. S., DOMENICUCCI, C., WASI, S., and SODEK, J. (1985a). Specific immunohistochemistical localization of osteonectin and collagen Types I and III in fetal and adult porcine dental tissues. J. Hi&o&em. Cytochem 33, 531-540. TUNG, P. S., SKINNER, M., and FRITZ, I. B. (1985b). Cooperativity between Sertoli cells and peritubular cells in the formation of the basal lamina in the seminiferous tubule. Ann. N. Y. Acad. Sci. 438,435-446. YAMADA, K. M. (1981). Fibronectin and other structural proteins. Ztz “Cell Biology of Extracellular Matrix” (E. D. Hay, ed.), pp. 95-114. Plenum, New York.