Phenotypic Characterization of Chicken Bursal Stromal Elements

Many, if not all, of the different phases of intrabursal B-cell maturation are controlled by the stromal components. We have used an extensive panel of mAb to provide a detailed phenotypic profile of these cells. Antigenic specificities were defined for the entire surface epithelium, interfollicular surface epithelium, follicle-associated epithelium, basement membrane, basement membrane-associated epithelium. Several mAb were specific for the medulla, including those reactive with the stellate network of epithelial cells, isolated macrophages, and granular, apparently secreted antigens. One of these, MUI-92, appears to be bursa-specific. Two mAb reacted strongly with stellate cortical macrophages, one of which weakly stained similar cells in the medulla. MHC-class II antigens were expressed on endothelium of the corticomedullary junction, macrophagelike cells in the cortex, and medulla and B lymphocytes predominantly in the cortex. Collectively, these mAb have demonstrated the antigenically distinct nature of discrete regions in the bursa, but also the continuity of the surface epithelium with the corticomedullary junction and medulla. They represent excellent reagents for defining the stromal cell contribution to B-cell development.


INTRODUCTION
Despite the central importance of the avian bursa of Fabricius in formulating the dichotomy model of the immune system and the fact that this organ remains the only known centralized site for the generation of B lymphocytes, the exact mechanisms by which it functions are not known. As a primary lymphoid organ, the bursa appears to have the capacity to attract bloodborne precursors (Le Douarin, 1986), induce their commitment to the B-cell lineage (Weber, 1982;Boyd et al., 1983), stimulate extensive proliferation of B lymphocytes within the follicles (Pink et al., 1985) associated with which is the generation of antigen-receptor diversity probably through gene hyperconversion (Weill et al., 1986;Lassila et al., 1988), localize intestinal-derived antigens in the vicinity of developing B cells *Corresponding author. Present address: Department of Pathology and Immunology, Monash Medical School, Commercial Road, Prahran, 3181, Victoria, Australia. (Bockman and Cooper, 1973), and confer upon mature B lymphocytes the capacity to produce specific antibodies (Granfors et al., 1982). In addition, the bursa can function as a secondary lymphoid organ since histologically plasma cells are evident (Naukkarinen and Sorvari, 1982), and plaque-forming cells can be found following immunization (Van Alten and Meuwissen, 1972). In this regard, T cells have been identified near the bursa (Odend'hal and Breazile, 1980) and within individual follicles (Chen, Cooper, Wilson, and Boyd, unpublished observations). The bursa also does not function in isolation, but has receptors for steroids (Ylickomi et al., 1987) and can directly influence the development of the thymus (Droege, 1976). Clearly, the functional capacities of the bursa are very complex, and although it is widely accepted that the controlling elements are encompassed within the nonlymphocytic stromal components of this organ, their exact nature is poorly understood. Histological and ultrastructural studies on the bursal stroma have demonstrated the interfollicular surface epithelial cells (IFE) to be relatively undifferentiated and continuous with those of the corticomedullary junction; these epithelial cells are supported by a basement membrane (Olah et al., 1975). The follicle-associated epithelial cells (FAE), which facilitate contact between medullary lymphocytes and lumenal contents (Sorvari et al., 1975), have specialized features, including a villous surface and a stratified network of stellate cells (Holbrook et al., 1974;Glick, 1983). During embryogenesis, the plical surface epithelium (SE) is infiltrated by mesenchymal "dark" cells, resulting in follicle development, the medullary component of which is supported by a stellate network of epithelial cells (Frazier, 1974;Boyd, Barr, et al., 1976) and dendritic reticulum cells, phagocytic macrophages (Naukkarinen and Sorvari, 1982) and "secretory" cells (Olah and Glick, 1978). The cortex is in contact with the vasculature and connective tissue, has only rare epithelial cells, but does have macrophages and "dark" reticular cells (Frazier, 1974;Glick, 1983). A more refined approach we have used to defining stromal elements is the production of monoclonal antibodies to these cell types. This study represents an extension to our earlier work, which utilized polyvalent antisera to identify a reticulinfiber framework specific for the bursal cortex , a bursa-dependent reticularepithelial cell antigen(s) present in the bursa medulla and B-dependent areas in the spleen, particularly periellipsoida! sheaths (Boyd, Barr et al., 1976), and a gut-associated mucin present in the bursal SE and medulla (Boyd and Ward, 1984). Houssaint et al. (1986) described two mAb to bursal stromal antigens, BEP-1 reactive with the SE and basement membrane-associated epithelium (BMAE) and BEP-2 reactive with a cytoplasmic antigen of medullary and SE cells.
The present study confirms these findings and describes 15 additional bursal stromal-cell antigens.
This panel of mAb has been very useful for identifying the ontogenic developmental relatio.nship between the individual stromal components (Wilson and Boyd, 1990a) and together with comparative analysis of bursae from cyclophosphamideor testosterone-treated chickens (Wilson and Boyd, 1990b), we have assigned tentative roles for these stromal cells in different phases of B-cell differentiation. Preliminary details of our mAb have been recently published .

RESULTS
A panel of 31 mAb was selected, each of which was thoroughly characterized. It provided an extensive phenotypic profile of the bursal stromal components and clearly delineated discrete regions (Table 1). No two antibodies have identical specificities. The complexity of the SE was revealed by at least nine distinct staining patterns. MUI-80 reacted with the entire SE ( Fig. la), whereas MUI-51 stained only the interfollicular SE, being negative on the FAE (Fig. lb). Phenotypic differences were further revealed within the multiple cell layers that form the SE and the similarities between the SE and medulla. MUI-90 was specific for the underlying basement membrane (Fig. 2a) and MUI-58 with the BMAE (Fig. 2b); whereas the latter was very weak to negative on the SE, MUI-91 strongly labeled both the SE and BMAE (Fig. 2c). MUI-77 was similar to MUI-58, but was distinguished by a different ontogenic expression (Wilson and Boyd, 1990a). MUI-73 FIGURE 1. MUI-80 stained the entire bursal SE (a, x400), whereas MUI-51 labeled only the interfollicular epithelium (bli], x125); note negative FAE (arrow). The double labeling of b(i) with antikeratin (x125) is shown in b(ii). also stained the BMAE, but, in addition, stained the because there was a marked heterogeneity between FAE and some areas of the medullary epithelium the folliclesmsome were strongly positive but others very weakly (Fig. 2d). MUI-53 and MUI-70 both were weak or negative (Fig. 3d). This was not an stained the entire SE, but only stained the basement artefact of the sections because serial sectioning membrane or a single layer of epithelial cells at the through the follicles gave similar results. Whereas corticomedullary junction (Fig. 2e). These antibodies MUI-62 was reactive with thymic epithelial clusters were distinguished by their thymic reactivity; MUIand a mucin-like molecule throughout the respira-70 was specific for pan type 1 epithelium, which tory and gastrointestinal tract, MUI-69 had very lines the subcapsule and perivascular regions, limited nonbursal reactivity, staining a subset of whereas MUI-53 only stained restricted regions of thymic medullary epithelial cell clusters and very the subcapsular epithelium (Boyd, Wilson, et al., infrequent goblet cells in the gastrointestinal tract. submitted).
MUI-92 was particularly interesting because it Three mAb stained the SE and network of medulstained the outer medullary epithelial cells in the lary epithelial cells, but each identified distinct bursal follicles, but was negative on all other tissues determinants. MUI-75 stained the basal layer of the tested (Fig. 3e). SE and the network of medullary epithelial cells In the bursa, three mAb were specific for keratin- (Fig. 3a). MUI-55 and MUI-65 reacted with the negative macrophagelike cells. MUI-66 strongly entire SE and stellate network of medullary stained isolated stellate macrophagelike cells in the epithelial cells; with the former, the staining was cortex that were weaker and less frequent in the more diffuse, extensive, and appeared cytoplasmic medulla (Fig. 4a). MUI-72 stained similar cells in the (Fig. 3b), whereas the latter was more linear (Fig. cortex, but was virtually negative on the medulla 3c). MUI-62 also labeled the SE and medulla, but (Fig. 4b). The heterogeneity of macrophagelike cells the cells in the medulla were keratin-negative and was further exemplified by MUI-79, which stained the staining very granular. 68,and 69 had such cells in the interstitial areas between follicles similar medullary staining patterns to MUI-62, but (Fig. 4c).
were negative on the SE. These antibodies were MUI-54 reacted with all bursal epithelium and, in distinguished by their ontogenic appearance (Wilson this regard, closely resembled the panepithelium and Boyd, 1990a). MUI-69 was of particular interest antikeratin reagent (Fig. 5a); the major dis-  MUI-78, which reacts with BL (MHC class II; Boyd, Wilson, et al., 1987, submitted), stained predominantly lymphocytes and macrophages in the cortex, the endothelium at the corticomedullary junction, and isolated stellate cells in the medulla; there was no obvious staining of epithelium ( Fig.   FIGURE 3. MUl-75 staining of the basal layer of the SE (arrow) and network of medullary epithelial cells (a, x125).  both stained all layers of the SE in addition to the medullary epithelium; for the former, however, the staining was more diffuse and cytoplasmic (b, x400), whereas the latter appeared restricted to plasma membrane (c, x200). MUI-69 showed granular staining of keratin-negative cells in the medulla of follicles (d[i], x125); the double labeling of d(i) with antikeratin is shown in d(ii) (x125). MUI-92 was bursa-specific and stained outer medullary (M) epithelial cells, the cortex (C) being negative (e, x400). 5b(i)). MUI-36 is a pan-B-cell marker, reacting exclusively with all chicken B lymphocytes and a restricted subpopulation of macrophages at the corticomedullary junction and in the thymic medulla .
All the MUI mAb were compared with antibodies specific for cytoskeletal components (Table 2). Antikeratin stained the entire SE, BMAE, and the stallate interconnecting network of medullary epithelial cells; the cortex was negative. Antiactin and antivimentin also reacted with medullary networks and the muscular layers; vimentin, unlike actin, was also present in the cortex. Stress fibers were only found in the muscle layer.

Reactivity with Nonbursal Tissues
In the initial screening of the fusions, hybridomas showing broad tissue cross-reactivity were discarded. Many, however, reacted with the chicken thymus and these characteristics have been described elsewhere (Boyd, Wilson, et al., submitted). The nonlymphoid reactivities of the mAb were limited and have also been described elsewhere (Boyd, Wilson, et al., submitted). FIGURE 4. MUI-66 staining of stellate, keratin-negative cells predominantly in the cortex and less frequent in the medulla (ali], x125); the double labeling of a(i) with antikeratin is shown in a(ii) (x125). MUI-72 also stained stellate keratin-negative cells, but these were only found in the cortex (b, x125). MUI-79 identified macrophages in the interstitial areas between follicles (F); the cortex and medulla were essentially negative (c, x125).

DISCUSSION
This study provides an extensive phenotypic profile of the stromal components of the bursa of Fabricius and demonstrates the complexity of the components that could potentially contribute to the microenvironment in this region. It complements our earlier data based on heteroantisera, which defined bursal-specific cortical reticulin fibers, a gut-associated mucin in the medulla  and an antigen present on bursa-dependent reticular epithelial cells in the bursal medulla and splenic germinal centers and periellipsoidal sheaths (Boyd and Ward, 1978). Recently, Houssaint et al. (1986) also described two mAb reactive with bursal stromal components: BEP-1, which stains mainly the SE and BMAE, and BEP-2, which binds to an intracytoplasmic antigen apparently secreted by medullary epithelial cells. In contrast to the present study, neither of these showed definite antigenic heterogeneity among the bursal epithelial cells, although the FAE were only weakly positive for BEP-1.
The present panel of mAb includes two reagents similar to BEP-1 and BEP-2, based on their staining patterns. These are MUI-91 and MUI-62, respectively. The remaining mAb clearly reveal epithelial cell heterogeneity. Whereas MUI-80 reacted exclusively with all the SE, MUI-51 stained only the interfollicular epithelium (non-FAE). This provides tiated nature of the BMAE revealed by transmission further proof for the specialized nature of the FAE, electronmicroscopy (Beezhold et al.,1.983), it may be which have previously been shown to have ultra-the precursor of both the outer SE and the stellate structural and histochemical properties consistent network of reticular epithelial cells forming the supwith their ability to endocytose luminal contents porting framework of the medulla. with subsequent deposition in the underlying inter-The complex nature of the medullary stroma was connecting network of reticular cells and fibers. A clearly revealed in this study. It consists of a likely consequence of this is the presentation to dif-network of stellate epithelial cells interspersed ferentiating B lymphocytes of intestinally derived within which are isolated keratin-negative macroantigens. In this context, the distribution of MHC phagelike cells. It is not yet known if the four mAb class II antigens (MUI-78) in the bursa was interreactive with the latter are detecting different popuesting because the major medullary staining was of lations of cells or different epitopes on the same isolated stellate keratin-negative cells. Therefore, cells. The apparent secretory nature of a subpopulathese may be functioning in an antigen-presenting tion of medullary epithelial cells originally proposed capacity, the additional signals required for B-cell by Olah and Glick (1978) was also demonstrated by activation being provided by the often frequent T MUI-69. This antibody was particularly interesting cells present in and around the follicles (Murthy et because it was negative on the SE and cortex, had al., 1984). MHC class II antigens have been very limited nonbursal reactivity, and showed described on bursal lymphocytes (Ewert et al., 1984), marked variability in staining between different folliwhich were shown here to be predominantly in the cles. It would not be unreasonable to propose a cortex, on the medullary epithelium of cyclophosspecific function in B-C'ell maturation for this molephamide-treated bursae (Hoshi et al., 1988), on cule and its variable intrabursal distribution raises endothelium at the corticomedullary junction (Belo the possibility of follicle heterogeneity. Whether this et al., 1985), and now on medullary stromal cells; indicates different stages of B-cell maturation is however, their role in B-cell differentiation from unclear. BEP-2 of Houssaint et al. (1986) also reacts multipotential precursors to functionally mature with an apparently secreted antigen in the medulla plasma cells remains obscure, and SE, although no mention was made in that As distinct from BEP-1, MUI-55, and 65 reacted study of differences between follicles. MUI-62 with the entire SE and the network of intercon-resembles this mAb; we believe from ontogenic necting medullary epithelial cells, including the studies, however, that the antigen is secreted by the basement-membrane associated epithelium. Since SE and taken up by the underlying medullary chick-quail chimera studies have demonstrated these keratin-negative cells. MUI-92 is arguably the most regions to be endodermally derived (Le Douarin et important determinant in this study, because it is al., 1976), these mAb may be valuable lineage bursa-specific. Its restriction to outer medullary markers. In addition to the distinction between the epithelial cells suggests a direct role in intrafollicular FAE and the interfollicular SE, heterogeneity was B-cell expansion. also revealed within the multiple layers that consti-In contrast to the complexity and specificity of the tute the SE. MUI-90 reacted with the underlying medullary stroma, only two mAb reacted preferenbasement membrane, but is unlikely to be directed tially with the cortex. MUI-72 extensively stained against type IV collagen because of its lack of the stellate cells in the cortex, which would correactivity on intestinal and renal glomeruli-assorespond to the reticular cells defined by electron ciated basement membranes and its species specimicroscopy (Olah et al., 1975). This apparently ficity (Boyd, Wilson, et al., submitted). Although the exclusive nature of the cortex and medulla provides cellular origin of the individual components of basefurther support for their different embryonic ment membranes is still speculative, the contribution origins--the cortex being mesenchymal and the of the epithelial cells was indicated by the reactivity medulla endodermal (Le Douarin et al., 1975). of MUI-53 and 70 with both the basement mem-However, whereas no endodermal (epithelial) cells brane and the entire SE. The distinctive nature of were found in the cortex, the medulla did contain the BMAE was demonstrated by MUI-58 and 93, its mesenchymal-derived macrophages.
The continuity with the basal layer of. the SE by MUI-91, relationship between the cortical and medullary B and with the medullary epithelial network by MUIcells still remains very speculative. No differences in 54, 65, and 75. In view of the apparently undifferenthese two cell compartments were observed with MUI-36, a pan-B-cell reagent. Similar results were lymphocytes discarded. The remaining tissue aggrefound with an equivalent mAb (L22; Pink and Rijn-gates and follicles were digested to a single cell beek, 1983) and heteroantisera to B lymphocytes, suspension by incubation in an enzyme solution although a fetal-associated antigen, CFAA, was consisting of collagenase type IV (0.15%), trypsin found on cortical but not medullary B cells and con-(0.15%), and DNase (0.01%) (all from Boehringerversely, a mature B-cell marker, CMBLA, was found Mannheim). For immunization or antigenic analysis only in the medulla (Boyd and Ward, 1984). This of stromal cells, trypsin was omitted. The stromal suggests that the medulla contains more mature B cells were then collected by repeated sedimentation lymphocytes, but further study on the lineage through newborn calf serum at I g for 30 min, 4C. relationships between these areas is required.
Several of the mAb described have reacted with Production of mAb macrophagelike cells. In addition to MUI-72 on cortical reticular cells, MUI-79 reacted with macro-The experimental details for production of the mAb phages in the interfollicular regions, MUI-36 with follow standard protocols and have been published those at the corticomedullary junction (and thymic elsewhere . Briefly, medulla, Ward et al., in preparation), MUI-66 with BALB/c mice were immunized i.p. with 0.2 ml those found extensively in the cortex and less serum-free RPMI 1640 containing 100/zl of packed abundantly in the medulla, and MUI-78 (MHC class stromal-cell-rich preparations, at weekly intervals. II) with a subpopulation within the medulla. MUI-66 Three days after the third injection, the spleen cells and 79 also react with blood monocytes and macro-were fused with log growth phase P3/NS-1/1-Ag/ phages throughout many tissues (Boyd and Wilson, 4-1 (NS-1) cells, HAT added on day 1 and HT at unpublished observations). None of these reagents day 7. Hybridoma supernatants were initially are truly pan-macrophage in comparison to the screened by indirect immunofluorescence on 4/zm recently described mAb ChNL-681 (Jeurissen et al., frozen sections of composite blocks of snap-frozen 1988), which has more general reactivity, thymus, bursa, and spleen; those showing broad In summary, this panel of mAb reveals the tissue cross-reactivity were discarded. Selected marked heterogeneity of the bursal stroma and facil-hybridomas were cloned at least twice by limiting itates detailed studies on the nature of the microen-dilution. vironment of this organ. Two approaches that have For indirect immunofluorescent staining, sections been very valuable for assigning a functional role of were completely covered in supernatant, incubated the cells involved to distinct phases of intrabursal B for 20 min at room temperature,, washed three times lymphocyte development have been to examine their (1 xl min, 2 x5 min) with PBS (gentle shaking), and ontogenic development and their status in chickens reincubated with conjugate for a further 20 min, and treated with cyclophosphamide as compared to washed as before. Control preparations were treated testosterone, which ablates the functional micro-with NS-1 conditioned medium. The conjugate was environment. The results of former studies are an FITC-labeled affinity purified F(ab')2 sheep antipresented in an accompanying paper and the latter mouse Ig (final dilution, 1:100, DDAF, Silenus Labs, are in press (Wilson and Boyd, 1990b).
Melbourne). Sections were mounted in veronal buffered glycerol or Permafluor (Lipshaw, Detroit), and examined with a Zeiss Axioskop microscope MATERIALS AND METHODS and photographed with a Zeiss MC100 automatic camera and Ektachrome ASA 1600 film.

Chickens
Australorp xWhite Leghorn F1 hybrid chickens (4-8 Specificity Analysis weeks old) obtained from Research Poultry Farm All mAb were tested on cryostat sections of bursa, (Research, Victoria) were used throughout this thymus, spleen, kidney, liver, lung, glandular stostudy, mach, small intestine, brain, heart, and skin. As an indication of species-specificity, the mAb were Stromal-Cell Preparation tested on sections of rabbit liver, mouse kidney, rat Bursae and thymuses were extensively teased apart stomach, and cultured 3T3 fibroblasts. Reactivity in serum-free RPMI 1640 at 4C and the freed with epithelial cells was determined by double labeling with a rabbit antikeratin (broad spectrum, dilution 1:200; Dako, Santa Barbara, California) and developed with a rhodamine conjugated goat-antirabbit Ig (dilution 1:50, Silenus Labs).
To determine reactivity of the mAb with plasma membrane determinants, immunofluorescence analysis was performed on fresh bursal stromal-cell suspensions. The cells (50/1) were incubated with mAb (100/1) for 20 min, 4C, washed twice in 5 ml RPMI 1640 containing 10% newborn bovine serum (150 g, 5 min), and incubated with 50/zl of the FITCantimouse conjugate (final dilution, 1:100) for a further 20 min, 4C. The cells were again washed twice and 25-/1 aliquots transferred to multiwell glass slides (Flow Labs). They were then rapidly air dried, fixed in absolute ethanol for 30 seconds, dried and washed in PBS for 10 min prior to mounting for immunofluorescence analysis.