Eur. J. Immunol. 1991. 21: 2951-2962

Yong-Jun Liu,

Jun Zhang, Peter J. L. Lane, Eric Y.-T. Chan a d Ian C. M. MacLennan Department of Immunology, The Medical School, University of Birmingham, Birmingham

Sites of B cell activation

2951

Sites of specific Bcell activation in primary and secondary responses to T cell-dependent and T cell-independent antigens* Techniques which identify hapten-specific B cells in tissues have been used to determine the sites of Bcell activation in rat spleens in response to Tcelldependent (TD) antigens and T cell-independenttype-1 (TI-1) antigens. Surfaceassociated hapten binding by specific memory B cells and B blasts was distinguished from the strong cytoplasmichapten binding by specific plasma cells and plasmablasts. Blast cells in S phase were identified in tissue sections by staining cells which had been pulse labeled in vivo with 5-bromo-2’-deoxyuridine. Hapten-specific B blast cells are found in three sites: (a) around interdigitating cells in the T cell-rich zones; (b) in the follicular dendritic cell network and (c) in association with macrophages in the red pulp. Hapten-binding memory B cells, which are not in cell cycle, accumulatein the marginalzones and to a lesser extent the follicular mantles in response to TI3 and TI-1 antigens. The hapten-specific sponse inTzones is confined to the first few days after antigen is given and but massive on secondary challenge, for primary responses toTD a to theTmnes. Both the primary and when marginal zone memory B cells ndary Tzone responses toTI-1 apten-specific B blasts are also response to TD antigens starts with a small number of B blasts entering each follicle.These increase in number exponentially so ay after immunization they fill the follicle. The oligoclonalityof the response is shown in simultaneousresponses to two haptens where 6%-31% of the follicles on day 3 after immunization con blasts specific for only one of pattern of germinal centers the two haptens. During the 4th day classical develops.The surface immunoglobulin-posit asts are lost from the follicle center, while one pole of the follicular dendritic cell network fills with surface immunoglobulin-negativecent but divide to give rise to centro follicular dendritic cell network population is renewed from cen the centroblasts

Some follicular response is seen to TI-1 antigens but this is much less dramatic than that seen during TD responses.

by mAb and in many cases the genes encoding them have been identified and sequenced. Immunohistological analyConsiderable progress has been made in understanding the sis indicates that the expression of molecules involved in molecules involved in the induction of proliferation and B cell activation can vary in different locations. Thus, the differentiationof B cells [l-31. Most of these are recognized B cells in the dark and light zones of germinal centers differ from each other and these in turn are phenotypically [I 95521 distinct from Bcells in the follicular mantle and in the marginal zone [4, 51. These differences in phenotype are * This work was supported by a programme grant from the British likely to reflect differences in capacity to respond to Medical Research Council. activation signals.

1 Introduction

Comspondence: Ian C. M. MacLennan, Department of Immunology, The Medical School, Edgbaston, Birmingham B152’IT, GB

If knowledge of the molecular basis of B cell activation obtained in vitro is to be related to events occurring in vivo Abbreviations: TD: T cell dependent TI-1:T cell independent it is essential to know where and when B cells are activated type-1 TI-% Tcell independent type-2 BrdUrd: 5-bromo-2’- during immune responses. Studies by van Rooijen et al. [6] deoxyuridine MSH Spider crab hernocyanin On: phenyloxaza- and Kosco et al. [7] have described the appearance of antigen-specific plasma cells and plasmablasts during relone FDC: follicular dendritic cell @ VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1991

+

0014-2980/91/1212-2951$3.50 .25/0

2952

Y.-J. Liu, J. Zhang,

P. J. L. Lane et al.

sponses to a range of antigens. Although these studies were able to recognize plasmablasts emerging in the T zones and from follicles they did not identify memory B cells and the proliferation of hapten-specific cells in follicles. A very recent report by Jacob et al. [8] has identified the sites of B cell proliferation in mice in primary responses to hapten protein. The study reported here complements these studies in defining the sites of Bcell activation in rat spleen during responses to T cell-dependent (TD) and T cellindependent type-1 (TI-1) antigens. The TD antigens used are hapten-protein conjugates, while the TI-1 antigens are conjugates of haptens and bacterial LPS. The B cells

Eur. J. Immunol. 1991.21: 2951-2962

responding to these antigens have been identified in tissue sections by enzyme histochemistry using haptens conjugated to horseradish peroxidase or alkaline phosphatase [9]. Cells which are in active cell cycle are identified by their uptake of 5-bromo-2’-deoxyuridine (BrdUrd) [lo]. This report considers the sites of Bcell activation in the spleen, as assessed by the entry of hapten-specific Bcells into cell cycle.The main splenic compartments are shown in Fig. l a and b. Hapten-binding cells are identified in four distinct areas: the follicles; the periarteriolar T cell-rich zones; the marginal zones and the red pulp. Each of these

Figure 1. Photomicrographs of frozen sections of rat spleen at different stages of responses to hapten-protein o r hapten-LPS stained by various immunohistological procedures. (a) Spleen section ( x 160) from a rat not undergoing active B cell proliferation taken from a rat primed with MSH 4 weeks previously and given BrdUrd in the drinking water for the last 48 h. B cells in the marginal zone (MZ) and follicle (F) are identified by HIS14 binding (brown staining).The nuclei of cells which have been through S phase of the cell cycle in the previous 48 h and have taken up BrdUrd into their DNA are stained red. There is no focus of proliferation in either the follicle or the marginal zone. R = red pulp and T = T zone. (b) Spleen section ( x 120) showing a well-developed germinal center taken from a rat primed with MSH and re-immunized 4 weeks later with DNP-MSH and Ox-MSH. The spleen was taken on the 5th day after re-immunization. The rat was given 5 mg BrdUrd i.p. 6 h and 3 h before the spleen was taken. Staining pattern: IgM-expressing cells brown, IgD-expressing cells blue, cells which had taken up BrdUrd red.The germinal center is surrounded at its base by theTcell zone (T) and at its apex and sides by the follicular mantle which is identified by its content of IgD-expressing B cells. Most of the cells in the base of the germinal center (dark zone, D ) have taken up BrdUrd,while only a minority of cells in its upper portion (light zone, L) are labeled with BrdUrd. (c) Clonality of the follicular reaction: section ( X 160) from the spleen of an MSH-primed rat 72 h after re-immunizing with DNP-MSH and Ox-MSH. One follicle is filled with DNP-binding cells (blue) and the other with Ox-binding cells (red).This is towards the end of the first stage of the follicular reaction when colonizing primary B blasts increase in numbers exponentially to fill the FDC network. At this stage no zonal pattern with centroblasts and centrocytes is apparent. (d) Centrocytes express sIg and are mostly not in cell cycle while centroblasts are in cell cycle but do not express sIg. Photomicrograph ( X 160) is of a follicle of a rat primed with MSH and re-immunized with DNP-MSH and Ox-MSH. The spleen was taken 2 h after an injection of BrdUrd and 5 days after boosting. Many centrocytes are expressing anti-Ox Ab (red) but few have taken up BrdUrd (blue). Conversely, many centroblasts have taken up BrdUrd but few are expressing anti-Ox.

Eur. J. Immunol. 1991. 21: 2951-2962 areas has its own putative accessory cells involved in B cell activation: (a) follicular dendritic cells (FDC) have the capacity to take up antigen as an immune complex and retain thisinitsnative form formanymonths [ l l , 121.In the present study follicles were the only site where continued hapten-specific B cell activation was observed during the established phase of TD antibody responses. (b) Interdigitating cells, also referred to as dendritic cells, in theTzones show high constitutive expressio culating Tcells and to some ex these cells [13, 141. They are antigens and efficiently induce T cells [15,161. In the present stud differentiation in the T zones was immediately following the administration of antigen. (c) The Bcells of the marginal zone are p primary blood sinusoidal network of the such they are well placed to interact with blood. Following interaction with antigen, hapten-specific memory B cells migrate from the marginal zone to T zones before becoming B blasts. MQ,in the selectively take up and retain neutral be of relevance toTI-2 antibody respo pulp of the spleen contains abundant MQ, on the cords of Billroth and within the venous sinusoids. Proliferation of hapten-specific B cells in the red pulp was found to be a feature of responses to TI-1 antigens.

2 Materials and methods 2.1 Antigens Spider crab hemocyanin (MSH) was a kind gift of the late Prof. J. H. Humphrey. MSH was used for carrier priming in the form of an alum precipitate and was given as 50 pg i.p. with 5 x lo9 chemically killed Bordetella pertussis organisms (Per Vac, Wellcome, Beckenham, GB). Two haptencarrier conjugates were used as soluble antigens administered i.v.: 2-phenyloxazalane (Ox) conjugated to MSH and DNP also conjugated to MSH. Ox-MSH was prepared as described by Miikela et al. [19]. DNP-MSH was prepared by adding 1 ml of 2,4-dinitrofluorobenzeneto 10 ml of a 2.5 mg/ml solution of MSH in 0.1 M borate buffer, pH 8.4.This was mixed for 2 h at room temperature before the conjugate in the aqueous phase was separated from unreacted 2,4-dinitrofluorobenzeneand dialyzed extensively against PBS. Alum-precipitatedDNF-MSH mixed with chemically killed Bordetella pertussis was given i.p. to study the fully primary response to this antigen. LPS derived from serotype 055:B5 and a conjugate of TNP and LPS were obtained from Sigma Chemical Poole, GB. TNP-LPS was administered i.v. at 20 w

Sites of B cell activation

2953

2.3 lmmunohistology 2.3.1 Conjugates and antisera DNP-alkaline phosphatase was prepared as follows: 5 pl nzene (Sigma) was added to 500 U calf intestinal alkaline phosphatase type VII (Sigma) in 2 ml 0.1 M borate buffer, pH 8.5. The mixture was stirred at room temperature for 1h and then dialyzed against PBS at h "C before use. The preparation of Ox-alkaline phosphatase was as for Ox-MSH. Ox-peroxidase was prepared as follows: horseradish peroxidase (Sigma) was first conjugated with poly-L-lysine (Sigma) to form polymers as described by Claassen and Van Rooijen [20]. It was then dialyzed against 5% NaHC03, 4 mg 4-ethoxymethylene2-phenyloxazalone was added per ml dialysate and the mixture was stirred at 4°C for 24 h. The resulting suspension was dialyzed against PBS, any remaining precipitate was then removed by centrifugation at loo00 rpm for 20 min and the SN, containing Ox-horseradish peroxidase, again dialyzed against PBS before use. mAb HIS14 (pan B reactive) and HIS22 (follicular mantle cell reactive) were kind gifts from Dr. Kroese and Prof. Nieuwenhuis, Department of Histology, University of Groningen, The Netherlands. These mAb were prepared and characterized as described [21].The mAb BU20a which recognizes BrdUrd incorporated into DNA was kindly supplied by Dr, D.Y. Mason of the John RadcliffeHospital, Oxford, GB.W3.13, an anti-leukosialin antibody which has a panT cell distribution but is not expressed on B cells [22], was kindly supplied by Dr. A.Williams of the MRC Cellular Immunology Unit, Oxford, GB. mAb MARD-1 was a kind gift of Prof. Bazin University of Louvain Medical School, Belgium. Rabbit anti-rat IgM was supplied by Dako (Glostrup, Denmark). Alkaline-phosphatase-conjugated sheep anti-mouse Ig, peroxidase-conjugated sheep antimouse Ig and peroxidase-conjugated sheep anti-rabbit Ig were kindly provided by the Binding Site, Birmingham, GB. 2.3.2 Tissue processing Portions of spleen for immunohistological examination were snap frozen in liquid nitrogen and stored at - 70°C. Five-micrometer frozen sections were cut and mounted on glass slides. The sections were thoroughly dried at room temperature for at least 1h, and were fixed in acetone at 4 "C for 15 min. They were then dried at room temperature and stored in sealed plastic bags at - 20°C. The sections were warmed to room temperature and kept dry before staining. 2.3.3 Simultaneous detection o€ DNI-binding and Ox-binding cells by double immunohistology (Fig. 1c; 2 c,d)

2.2 Animals have been maintained as a

minimal.

Sections were washed in 50 m~ Tfis-HC1 buffered saline, pH 7.6, and then h a mixture of DNP-alkaline phosphatase and e in this buffer for 1h. After a further wash, alkaline phosphatase was developed using the substrate naphthol AS MX phosphate (0.4 mg/ml) and color reagent Fast Blue BB salt (1 mg/ml) in 50mM

2954

Y.-J. Liu, J. Zhang, I? J. L. Lane et al.

Eur. J. Immunol. 1991.21: 2951-2962

Eur. J. Immunol. 1991. 21: 2951-2962

Tris-HC1 buffered saline, pH 9.2, containing 0.8 mg/ml levamisole to inhibit endogenous alkaline phosphatase. Peroxidase was then developed using 3-amino-9-ethylcarbazole (4 mg/lO ml50 m~ sodium acetate buffer, pH 4.5, was added immediately before to which 1drop 30% HZOZ use) or 3,3'-diaminobenzidine (5 mg/lO ml Tris-buffered saline containing 1 drop HzOz) if sections were to be subsequently examined by triple immunohistology [9] to detect BrdUrd. Slides were then washed in running water and mounted in glycerol jelly.

2.3.4 Double hnunoenzyme techniques for determining the phenotype of hapten-binding cells

Sites of B cell activation

2955

rated into cellular DNA. The binding of BU20a was then revealed using alkaline phosphatase-conjugated rabbit anti-mouse antibody and mouse mAb antibody against alkaline phosphatase, the APAAP technique [23]. The substrate for alkalinephosphatase used in this situation was Fast Red TR salt which yields a red precipitate. BrdUrd staining was preceded by immunohistological staining to identify mouse mAb bound to the section using peroxidase-conjugatedor alkaline phosphatase-conjugated sheep anti-mouse Ig or rabbit anti-IgM identified with peroxidase-conjugated sheep anti-rabbit antibody. Alkaline phosphatase activity was detected with AS MX ith Fast Blue BB salt, peroxidase activity with phosphat% diaminobe idine.

To demonstrate staining by HIS14 or HIS22, together with hapten-binding cells, sections were incubated with the 2.5 Quantitative microscopy and definition of splenic appropriate mAb, followed by a mixture of sheep anticompartments mouse peroxidase (absorbed with rat serum to prevent cross-reaction with rat tissue) and haptenated alkaline 2.5.1 Determining the percentage of the splenic volume occupied by germinal centers phosphatase. The identification of the bound enzymes was carried out as described in Sect. 2.3.3 This was carried out using frozen sections of spleen stained to reveal HIS22 binding cells and counterstained with .hematoxylin. HIS22 binds to follicular mantle B cells but 2.4 BrdUrd administration and detection not to B blasts, centroblasts or centrocytes in follicle As indicated in the results rats were either given BrdUrd at centers.The relative volume of germinal centers, therefore, 0.8 mg/ml in their drinking-waterover 48 h periods or an i.v. was determined by assessing the HIS22- area within injection of 5 mg BrdUrd 2 h before being killed. The follicles.This area was determined using the point counting schedule of BrdUrd administration used in detailed kinetic technique of Weible [24].A 1-cm2graticule was placed in studies of germinal center cells is specified in the results. one x 10-eyepiece of a conventionallight microscope.This Cells which had incorporated BrdUrd into their DNA were graticule is divided into vertical and horizontal lines at detected in frozen tissue sectionswhich had been processed 1mm intervals.The sectionswere then viewed using a x 10 for immunohistology as described in Sect. 2.2.3. These objective to give 100 x magnification. The number of sections were then treated for 20 min with 1M HC1 to intercepts between vertical and horizontal graticule lines expose and partially degrade the DNA. This also served to falling on (a) the spleen section and (b) the HIS22- areas terminate immunoenzymatic reactions which had taken within follicles was determined. The percentage splenic place on the slide previously without displacing the colored volume occupied by germinal centers is 100 x b/a. Sections precipitates produced by the action of peroxidase on were systematically counted starting at the top left hand diaminobenzidine or alkaline phosphatase on naphthol AS corner of each section and making serial sweeps from left to MX phosphate. Sections were then washed and exposed to right until at least 40mm2 (4000 intercepts) of spleen the mouse mAb BU20a which identifies BrdUrd incorpo- section had been examined.

4 Figure 2. Further immunohistologicalpictures of frozen sections of rat spleen. (a) Phase 3 of the follicular response, where some follicles contain small numbers of hapten-binding secondary B blasts ( X 160): The rat had been primed with MSH and re-immunized with DNP-MSH and Ox-MSH. The spleen was taken 6 weeks after re-immunization and 2 h after an injection of BrdUrd. A BrdUrd+ hapten-binding blast (red nucleus and blue membrane staining) is arrowed in the follicle center, many BrdUrd- DNP-binding memory cells (blue membrane staining) are seen in the marginal zone (MZ). B cells are identified with HIS14 (gold). (b) The massive hapten-specific blast reaction in the Tcell zone associated with the early phase of a secondary response to hapten-protein. The spleen ( X 500) is from a rat 48 h into a secondary immune response to DNP-MSH. DNP-binding cells (blue), cells which had taken up BrdUrd injected 2 h before the spleen was removed (red),Tcells identified by mAb W3.13 (brown). (c) DNP-binding memory B cells (blue) and Ox-binding memory B cells (red) ( X 500) in the marginal zone of an MSH-primed rat which had been re-immunized with DNP-MSH and Ox-MSH 6 weeks before the spleen was taken for immunohistology.The location of marginal sinus,which marks the junction between the Tcell zone and the marginalzone is identified with arrows.TheTcel1 zone (T) contains no hapten-binding cells. (d) Selectivedepletion of DNP-binding memory cells from the marginal zone after challenge with DNP-MSH: This spleen (x 250) is from a rat primed with MSH re-immunized with DNP-MSH and Ox-MSH and 6 weeks later re-challenged with DNP-MSH. The spleen was taken 48 h later. The marginal zone ( M Z ) is depleted of DNP-binding cells (blue) but contains Ox-binding cells (red).The Tcell zone (T) in contrast with the spleen depicted in Fig. 2c contains many DNP-blasts. This section is from the same spleen as the rat shown in Fig. 2b. (e) Nonspecific proliferation of follicular B cells in a rat givenT"-LPS 24 h previously ( x 120). Cells which have taken up BrdUrd in the 2 h before the spleen was taken are stained red. B cells are stained brown with HIS14. Note the large sue of the follicle (F) which appears to have expanded into the outer T cell zone. By contrast the marginal zone (MZ) is markedly depleted of cells."he marginal sinus separating the marginal zone and follicle is arrowed. ( f ) Massive hapten-specific B cell proliferation in the red pulp (R) and Tcellzones (T) in a rat 72 h after secondary challenge withTNP-LPS ( x 120). Cells which had taken up BrdUrd in the 2 h before the spleen was taken are stained red, DNP-binding cells blue and B cells brown.

2956

Y.-J. Liu, J. Zhang, F! J. L. Lane et al.

2.5.2 Determining the relative number of DNP-binding cells in each splenic compartment and the proportion of these cells taking up BrdUrd into their DNA following a 2 h BrdUrd pulse Sections were stained by three-color immunohistology to reveal HIS14+ cells (brown); DNP-binding cells (blue) and cells with BrdUrd in their DNA (red) (Fig. 2f). HIS14 stains all B cells and allows the following splenic compartments to be identified. The marginal zone is an area filled with HIS14+ B cells. It surrounds the T cell-rich periarteriolar Tzone and the follicles.The inner limit of the marginal zone in rats is clearly defined by the marginal sinus. This is the inner limit of the marginal zone blooQinusoidal network. The Tcells of the T zones are HIS14- while the B cells of the follicles are HIS14+.The red pulp is the area outside the marginal zone. These sections were examined using the microscope graticule system described in Sect. 2.5.1 at a magnification of 250 x. Serial sweeps of the microscope stage were made starting at the top left hand corner of each section. The numbers of DNP-binding cells and the number of DNPbinding cells which had taken up BrdUrd in each of the four splenic compartments, follicles, marginal zones, T zones and red pulp in the area of spleen examined were recorded. The number of graticule intercepts falling on the total area of spleen screened was also recorded. Sections were screened until an area of spleen covered by at least 500 graticule line intercepts (2 mm2) had been counted, larger areas being counted where the number of DNP-binding cells per unit area was low. The relative number of cells in each compartment was calculated as: 100 x the number of cells in a compartmentftotal number of intercepts falling on the area of spleen examined.

2.6 Assessment of serum anti-hapten antibody titers The levels of anti-Ox, anti-DNP and anti-MSH antibodies of the IgM and IgA classes and IgGl, IgGza, IgGzb and IgGzc subclasses were measured by solid-phase RIA as described previously [25].

3 Results 3.1 Experimental design The studies of T D responses were made in three situations: (a) during the primary response to alum-precipitated DNP-MSH given i.p. with B. pertussis; (b) during primary responses to DNP and Ox, whereT cell help had been made non-limiting, by priming the rats i.p. with alum-precipitated MSH with B. pertussis, 1 month before challenge i.v. with soluble DNP-MSH and Ox-MSH; and (c) during the secondary response to soluble DNP-MSH given i.v. 6 weeks after rats primed with MSH as above had been challenged i.v. with soluble DNP-MSH and Ox-MSH. Double immunization with DNP-MSH and Ox-MSH was used to provide an internal control; DNP-binding cells (blue) and Ox-binding B cells (red) (Fig. 2 a and b) can be distinguished from cells which have passively bound antiDNP and anti-Ox antibodies. Hapten-binding cells in S

Eur. J. Immunol. 1991. 21: 2951-2962 phase of the cell cycle were identified by pulse labeling with BrdUrd for 2 h before the spleens were taken for histology. This allowed B blasts to be distinguished from memory B cells and plasmablasts to be distinguished from plasma cells. B blasts are defined as cells in cell cycle which have a staining pattern suggesting mainly membrane-associated Ig (e.g. Fig. 1c and d). Plasmablasts are also in cell cycle but have cytoplasmic Ig (e.g. Fig. 2 b). The double immunization also allows location of DNP-binding memory B cells to be compared with that of Ox-binding memory cells following challenge with DNF-MSH. In a second set of experiments primary and secondary responses to TNP-LPS were studied, the interval between immunizations in this study being 6 weeks. The haptenspecific antibody responses induced by these immunization schedules using hapten-protein and hapten-LPS produced responses similar to those reported previously [25-271.

3.2 The location in the spleen of hapten-specific B blasts, memory B cells and plasma cells during responses to hapten protein.

3.2.1 Primary response to hapten-protein The results of the quantitative immunohistology in these experiments are shown in Figs. 3 and 4. During the primary response to DNP-MSH the number of hapten-binding B cells seen in the spleen was relatively low compared with the numbers seen following immunization whereT cell help was not limiting. Only occasional hapten-binding B blasts WEE seen in the T cell zone during the first week following immunization. Blasts were present in the follicle between 7 and 21 days after immunization. Some hapten-binding cells were seen in the marginal zone from 8 days onwards. The size of the follicular reaction as assessed by the HIS22-ve area within follicles is shown in Fig. 4. The peak of the follicular reaction was at 14 days.

3.2.2 Primary anti-hapten response to hapten-protein when T cell help is non-limiting When T cell help was made non-limiting by priming with carrier, the response to hapten protein was far more dramatic. The first hapten-specific cells were seen in the outer layers of the T cell zone. These were present between 36 h and 60 h from challenge with hapten protein. Many of these cells contained substantial amounts of anti-hapten antibody in their cytoplasm indicating that they were undergoing plasmacytoid differentiation. These cells were found also in the bridging channels which cross the marginal zone, connecting theTcell zone with the red pulp. Haptenbinding cells in the bridging channels appeared to form a continuum with hapten-specific plasma cells in the red pulp suggesting that the latter might be derived from the former. Hapten-specific plasma cells in the red pulp were found between 36 h and 6 days after immunization.These plasma cells showed strong cytoplasmic binding of hapten and did not become labeled in 2 h BrdUrd pulses. By 36 h after immunization small numbers of haptenbinding B blasts were apparent in all follicles. At this stage the blasts were interspersed with small IgD+ B cells.The B

Eur. J. Immunol. 1991.21: 2951-2962

Sites of B cell activation

2957

Figure 3. The relative numbers of DNP-binding cells in various splenic compartments during responses to DNP-MSH (closed circles) and the proportion of these cells which had taken up BrdUrd in the 2 h before the spleen was taken (open circles). In the fully primary response (lry) rats were given alum-precipitated DNP-MSH i.p. with lo9killed B. pertussis. In the primary anti-hapten response whereTcell help was non-limiting (lry Thelp) rats were primed as above and boosted i.v. 1month later with DNP-MSH and Ox-MSH. I n the secondary anti-hapten response (2ndry); rats were primed and boosted as above and re-challenged with DIG-MSH 6 weeks after boosting with the hapten-protein mixture. Each circle indicates the values for a single rat spleen. Not all the rats were given BrdUrd pulses.The values given are in arbitrary units which are directly proportional to the total number of hapten-binding or BrdUrd+ hapten-binding cells in each compartment, see Sec. 2.5.2. Lines are drawn through the median values for each time point.

+

:

The distribution of Ox-binding blasts and DNF-binding blasts in follicles is not uniform. Between 6% and 31% of follicles are monospecific (Fig. 1c). Some of the follicles contain Ox-specificcells on one side and DNP-specificcells on the other. The respective proportions of follicles monospecific for either Ox or DNP in nine rats studied 3 days after immunization was 6,14,20,21,22,25, 26, and 31%, confirming the oligoclonal nature of the follicular responses.

When the FDC network has filled with blasts a dramatic .. change associated with germinal center formation occurs within the follicle. Well-developed germinal centers with 0 2 4 7 10 14 21 dark and light zones are present in all follicles by 96 h after Days after last immunization immunization (Fig. l b and d). B blasts are no longer Figure 4. The relative size of the follicle center reaction during the prominent in the follicle center. A cluster of proliferating fully primary anti-hapten response (triangles), the primary anti- cells (centroblasts) appears in that part of each follicle hapten response when Tcell help is not limiting (closed circles), adjacent to the Tcell zone. This is the dark zone of the and the secondary anti-hapten response (open circles). The germinal center.The centroblasts like the follicular B blasts immunization groups are described in the legend to Fig. 3 and are HIS14+ and HIS22- but they differ from the B blasts in Sec. 3.1. The size of the follicular reaction is assessed as the that they are not hapten-binding cells.The follicle center is percentage of the section which is within follicles and is HIS22- as specified in Sect. 2.5.1. Each point represents the value for a single now filled mainly with hapten-binding centrocytes, which rat. Lines are drawn through median values. The time scale are not in cell cycle (Fig. 1d). At this stage BrdUrd-labeling studies show that the centroblasts were all labeled within represents the time from the last immunization of each rat. 6 h of injecting BrdUrd (Figs. 1b and 5). A small proportion of hapten-binding cells in the centrocyte-containing blasts then increased in numbers exponentially until reach- light zone were labeled within 1h of starting BrdUrd ing a peak level 72-84 h after immunization. At this stage labeling but this proportion did not increase greatly over hapten-binding blasts fill the follicle center, the small the next 5 h. Between 6 and 12 h after injecting BrdUrd follicular B cells are displaced to form a follicular mantle. almost all the centrocytes became labeled. These findings About half of the hapten-binding blasts were labeled 2 h confirm that: (a) centroblasts are in cell cycle and (b) after injecting BrdUrd and almost all were labeled within centrocytes typically are not in cell cycle, but are derived 6 h. Some tingible bodies were seen in M@ in the follicle from cells which have been in cell cycle during the previous centre. BrdUrd-labeling studies indicate that these con- 12 h. The studies also suggest that some hapten-binding sisted of condensed DNA from cells which had been in S cells in the light zone enter cell cycle but either leave the phase within the previous 12 h. light zone or die within the light zone before completing cell

2958

Y-J. Liu, J. Zhang, I? J. L. Lane et al.

E E v) \ v)

3

I

I

20001

Eur. J. Immunol. 1991. 21: 2951-2962

1500 1000

u U

2

500

0)

P

a

J

n ".

0

G\

4

2 L$

6

8 10 12 14 16 18 20 22 24

Hours from start of labeling

Figure5 The relative number of cells which have taken up BrdUrd into their DNA in the light zone (open bars) and dark zones (closed bars) of germinal centers at intervals after giving injections of BrdUrd as indicated by arrows at 0 and 3 h. Lines are drawn through median values.Vertica1 bars represent the range of values for groups of four rats used at each time point, the open bars are set to the right to avoid overlap. Sections were taken from the spleens of MSH-primed rats 5 days after boosting with DNP-MSH. These sections were stained as in Fig. 1b and the light zone was identified by the IgM staining of the FDC network.

cycle. Analysis of the hapten specificity of the centrocytes in the fully developed germinal centers indicates that 5%-25% of the follicles still contain either centrocytes specific for Ox alone (e.g. Fig. 1 d ) or centrocytes only specific for DNP. This finding indicates that the germinal center is derived directly from the small number of B blasts which colonized the follicle early in the response and that exchange of cells between follicles at this stage is minimal. Germinal centers were largest at the time of their formation from B blasts and then gradually decreased in size so that 3 weeks after immunization the germinal centers with dark and light zones were no longer apparent. At this stage small numbers of hapten-binding B blasts were found in follicle centers. These were still apparent 7 weeks after immunization (Fig. 2a).

I

Pndry

1r Y

I

640

160 40

0 DAYS

POST

FINAL

2

3

4

IMMUNISATlON

Hapten-binding cells appeared in large numbers in the marginal zones on the third day after immunization and were still seen in this site 1year after immunization. As we have reported previously [9] these cells do not take up BrdUrd during a 2 h labeling period. The distribution of Ox-specific and DNP-specific blasts in the marginal zones appeared to be random (Fig. 2c).This applied also to those areas of marginal zone immediately over monospecific follicles, indicating that hapen-specific memory B cells are unlikely to migrate directly to the marginal zone from follicles. Although a few hapten-specific B cells are found in the follicular mantles these cells are much less numerous than the marginal-zone memory B cells.

3.2.3 Secondary response to hapten-protein To study secondary antibody responses further groups of carrier-primed rats were challenged i.v. with DNP-MSH plus Ox-MSH and 7 weeks later re-immunized i.v. with DNP-MSH only. The results of quantitative immunohistology on these animals' spleens are also shown in Fig. 3 and 4. By 24 h after challenge the marginal zones are selectively depleted of DNP-binding cells while Ox-binding cells remain in this site (Fig. 2 c and d). This depletion is associated with the appearance of DNP-specific blasts in the follicles and the T zones. The first wave of follicular DNP-specific blasts is over by 36 h while the number of these blasts in the T cell-rich periarteriolar lymphocytic sheath (PALS) continues to rise. This massive PALS resonse (Fig. 2 b) continues for 48 h and is over 72 h after immunization. DNP-specificblasts are seen throughout the red pulp at 48 h but by 72 h all the hapten-binding cells in the red pulp are plasma cells. DNP-specific plasma cells have disappeared from the red pulp within 7 days of immunization. By 72 h after immunization a second wave of the follicular response ensues. This never reaches the level seen in the primary anti-hapten response when Tcell help is nonlimiting, where the peak HIS22- area within follicles is four time as great. DNP-binding cells reappear in the marginal zone by 72 h and persist there for months.

1 Figure 6. The relative numbers of T"-binding cells in various splenic compartments during primary and secondary responses to TNP-LPS (closed circles) and the proportion of these cells which had taken up BrdUrd in the 2 h before the spleen was taken (open circles). Each immunization was with 50 pg TNP-LPS i.v. with 6 weeks between immunizations. Each point indicates the values for a single rat spleen. The values given are in arbitrary units which are directly proportional to the total number of hapten-binding or BrdUrd+ hapten-binding cells in each compartment, see Sec. 2.5.2. Lines are drawn through the median values for each time point.

Sites of B cell activation

Eur. J. Immunol. 1991. 21: 2951-2962 6.4

immunization with DNP-MSH irrespective of whether T cell help is available; (b) proliferation of hapten-specific B blasts in the red pulp is far more extensive in responses to TNP-LPS (Fig. 2f); (c) hapten-specific plasma cells are far more persistent in the red pulp in secondary responses to TNP-LPS; and (d) follicular responses to TNI-LPS are smaller and less uniform and are associated with less proliferation of hapten-specific cells in follicles. Some small germinal centers, however, seem to be formed in a proportion of follicles in response to TNP-LPS.

I

I :h* 0

2

4

7

10

2959

14

21

Days after last immunization Figure 7. The relative size of the follicle center reaction during primary (circles) and secondary (squares) anti-hapten responses to TNP-LPS. The size of the follicular reaction is assessed as the percentage of the section which is within follicles and is HIS22-, as detailed in Sect. 2.5.1. Each point represents the value for a single rat. Lines are drawn through median values. The time scale represents the time from the last immunization of each rat.

4 Discussion

4.1 T zone responses

These studies identlfy three sites of antigen-specific B cell proliferationin the spleen: the follicles, theT cell zones and the red pulp. The timing of these responses is markedly different (Table 1).Hapten-specific B cell responses in the Tcell zones occur only in periods immediately following antigen administration.The magnitude of these responses depends on the type of antigen and the availabilityof Tand B cell memory. They seem to start shortly after immunization and are completed in < 4 days except in the primary 3.3 The location of hapten-specific cells in the spleen in response to protein based antigens when the Tcell zone response extends 2 or 3 days 1onger.This may be due to the responses to hapten-LPS persistence of free antigen until the onset of antibody 3.3.1 Early effects of LPS which are not hapten specific production.The duration of responses in the PALS to both hapten-protein and hapten-LPS seems to coincide with the The results of the quantitative analysis of the response to observed period in which virgin B cells are recruited into TNP-LPS are set out in Figs. 6 and 7. These show many responses to these antigens [26, 271. It is tempting to similarities with the results obtained in responses to conclude, therefore, that the PALS and equivalent T cell hapten-protein although there are also some marked zones in other secondary lymphoid organs are the major sites where virgin B cells are recruited into these responses. differences. A number of studies indicate that virgin Bcells, unlike Striking short-term effects on the distribution of B cells in recirculating memory cells, have little or no capacity to the spleen are known to be produced by the administration respond to antigen held on FDC without first being of LPS alone [28]. These are also seen when TNP-LPS is activated outside follicles [29, 301. given.Within hours there is a marked loss of B cells from the marginal zones, while the follicles become engorged with The short duration of Bcell responses to TD and TI-1 IgM+IgD- B cells which outnumber the resident IgD+ antigens in T zones may reflect the availability of antigen Bcells. Within 2 days this process has reversed and the rather than active suppression of B cell activation. The marginal zones have again filled with B cells. One day after ability to elicit renewed proliferation in T cell zones by immunization there is massive proliferative activity of further antigen administration implies that the supply of B cells in the follicles, as assessed by BrdUrd uptake (Fig. available antigen is an important limiting factor for re2 e). Residual B cells in the marginal zone and B cells in the sponses in this site. A plausible working hypothesis is that red pulp also enter cell cycle.This B cell proliferation is not the antigen which drivesTcellzone responsesis taken up by hapten-specific and has subsided by 48 h after immuniza- B cells themselves and is not presented in the first instance by cells in the Tcell zone. Protein-based antigen taken up tion. by B cells and carried to T cell zones may be processed by the B cells [31, 321 or passed to interdigitating cells which process and present this in association with class I1 MHC 3.3.2 Hapten-specific response to TNP-LPS molecules [15, 161. Such processed antigen can be highly Similarities between the TD and TI-1 responses are: (a) immunogenic to Tcells but may have little capacity to hapten-specific proliferation in the T zones is confined to cross-link the surface antigen receptors on B cells. the first 4 days after immunization, (b) the production of memory Bcells in the marginal zone and (c) the extrafollicular response is greater after secondary immuniza- 4.2 Red-pulp plasma cells tion. Plasma cells in the red pulp of the spleen in rats have been The main differences between these responses are: (a) the found to have a life-span of about 3 days [33]. In the primary Tcell zone response to TNP-LPS is remarkably responses to hapten-protein analyzed in the present study similar to that seen in secondary responses to DNP-MSH plasma cells were only found to persist for some 3 days after and is much greater than that seen following primary the anti-hapten response ended in theTcell zones. In these

2960

Eur. J. Immunol. 1991. 21: 2951-2962

Y-J. Liu, J. Zhang, l? J. L. Lane et al.

Table 1. Summary of the location of hapten-binding cells in responses to different antigens Response

Fully lry DNP-MSH Anti-DNP-MSH and Ox-MSHin MSHprimed rats Secondary anti-DNP-MSH

Tcell zone response

Germinal center

Minor, end of fmt week

Small, only in

Moderate, days 2,3and4

reaction some follicles days 7-21 Large. in all follicles days 4-21

Marginal-zone memory B cells

Red-pulp plasma cells

Red-pulp B blasts

Small number from day 6

Some. days 4-12

None

Large numbers from day 3 for months

Many, days 1.5-6

Very few

Massive. days 2 . 3 and 4

Moderate, in most follicles days 3-10

DNP-SpeCifiC, lost in 1st 2 days, return in large numbers day 3 for months

Large, numbers days 1.5-6

Large numbers, days 1.5-3

Extensive,

Moderate, but not in all follicles days 2-7 Moderate, but not in all follicles days 2-7

Large numbers from day 3 for several weeks

Large numbers, days 2-7

Large numbers days 2, 3 and 4

Some lost days 1 and 2 large numbers from day 3 for months

Large numbers day 2, still present day 21

Massive blast reaction, days 2 , 3 and 4

Primary TNP-LPS

days 2, 3 and 4

Secondary TNP-LPS

Extensive, days 2, 3 and 4

responses there appeared to be a continuity between The first phases of the follicular reaction is associated with plasma cells in the red pulp and hapten-specific blasts in the exponential growth of B blasts which subsequently give bridging channels. It seems likely that the red pulp plasma rise to centroblasts and centrocytes. The proliferation of B cells inTD responses come from B blasts activated inTcell blasts in this initial stage of the follicular reaction is clearly zones. This is less clear in relation to the TI-1 responses oligoclonal. Kroese et al. [35] concluded that follicular studied. The appearances suggested that some red pulp responses were oligoclonal in studies using irradiation plasma cells are derived directly from B cells activated in chimeras.The use of the two-hapten system in the present situ in association with red pulp M a . Help from M@ in study has allowed us to confirm this in a more physiological LPS-induced B cell proliferation and differentiation to situation and this has also been confirmed in a recent study plasma cells in vitro is well characterized [34].The origin of in mice [8]. Between 6 and 31% of follicles were found to be the persistent red pulp plasma cells found in the secondary monospecific at the stage when B blasts fill the follicle response to TNP-LPS in the present study remains to be center. Theoretically one would expect to have 25% identified. As was indicated in the introduction detailed monospecific follicles if the numbers of Ox-specific and studies of the sites where plasmablasts and plasma cells DNP-specific blasts capable of colonizing follicles were appear in responses to a range of antigens have been equal and three blasts seeded each follicle. An average of reported [6,7]. Studies of TD and TI-1 antibody responses five seeding blasts per follicle would be expected to result in in rabbit spleens [6] identify hapten-specific antibody- 6% monospecific follicles. At the end of the phase of B containing cells in the outer T zones and the extension of blast proliferation within follicles (3 to 4 days after these cells into the red pulp of the spleen. The number of immunization) the average number of B blasts in each plasma cellsfound in the red pulp in these studies does seem follicleis in the order of 1.2 x 104 to 1.5 X 104[36].Three to to have been lower than that observed in our studies in the five blasts in exponential growth would have to go through 12 divisions to produce this number of blasts. If this number rat. of blasts is reached in 72 h then cell cycle times would have to be in the order of 6 h. This is close to the cell cycle time estimated for centroblastsin mouse follicles using stathmo4.3 Phases of follicular responses kinetic techniques [37].This rate of cell proliferation within A number of factors have allowed us to study the sequence the dark zone of rat follicles can also be deduced from the of development of the follicular response in considerable BrdUrd-uptake studies reported in the results. The factors detail: (a) the background of follicular proliferation in the restricting the number of B blasts which colonize follicles is spleen was negligible; (b) the follicular responses to unknown. It seems unlikely that this is directly related to hapten-protein in carrier-primed rats were remarkably the number of €3 cells activated inT cell zones, for the T cell uniform in terms of magnitude, time of onset and duration; zone reaction yields large numbers of blasts (Fig. 3). and (c) using a two-hapten system it was possible to distinguish endogenously produced Ig on hapten-specific B The first phase of the follicular response ends at the time blasts from antibody taken up by FDC. Using this system to when the B blasts fill the spaces in the FDC network in the identify DNP-specific B cells and Ox-specific B cells in a follicle center. It seems probable that the primary B blasts single section three closely related phases of the follicular give rise to centroblastsas the oligoclonalityof the follicular reaction persists into the second phase of the response.The reaction have been identified.

Eur. J. Immunol. 1991. 21: 2951-2962

link between centroblasts and centrocytes was established >25 years ago by the studies of Fliedner et al. [38]. The results of these experiments analyzing germinal center kinetics by [3H]dThd uptake are very similar to the results obtained in the present study. Hapten-binding studies confirm the lack of Ig expression in centroblasts and demonstrate that this is re-expressed in centrocytes. Recent studies on isolated germinal center cells provide direct evidence that centrocytes are selected on the basis of their affinity for antigen held on FDC [39].The basis for selection is that cells kill themselves by apoptosis unless they are prevented from doing so by a signal dependent upon interaction with antigen. Studies of the time of onset of somatic mutation in Ig V region genes during T D antibody responses indicate that this does not start until after the stage when proliferation in Tcell zones has ended [40]. It seems probable, therefore, that the antigen-based selection process identified in germinal centers selects cells which have undergonev region gene mutation within the follicles. Further studies have shown that soluble CD23 plus IL-la are capable of inducing germinal center cells to differentiate to plasmablasts [41]. CelIs seen to enter cell cycle in the light zone and then leave the germinal center might have been triggered in this way. The third phase of the follicular response to TD antigens, which is the longest but least dramatic, occurs after the classical germinal center reaction has ended. It is associated with small numbers of hapten-specific B blasts which can be identified in follicles for months (Fig. 2 a). These blasts are the only obvious source of the plasma cells which maintain antibody production for more than a year and the marginal zone memory cells which are still present a year from immunization [36]. Available evidence on plasma cell life-span [33] and the life-span of marginal zone memory B cells [9] indicates that these are considerably less than a year. At this stage there may be exchange between recirculating memory cells generated from other follicles and locally derived B blasts. The phenotype of these secondary B blasts is only defined in that they are sIg+ HIS22- and in cell cycle. It will be important to carry out further studies to see if and how the B blasts which characterize the first phase of the follicular response differ in phenotype from these secondary B blasts. The follicular reaction in secondary responses in the present study was smaller than that in primary responses whereT cell help was not limiting. This has been recognized by others [42] and raises the possibility that memory cells may respond to antigen on FDC without giving rise to centroblasts and centrocytes, i. e. they may enter directly in phase 3 of the follicular response. Further studies using the two-hapten system are being carried out in our laboratory to investigate this further and the interim results are consistent with this hyothesis. Several groups have provided evidence which indicates that high-affinity mutants are mainly generated during primary responses and that memory cells can undergo extensive proliferation in secondary responses without further mutation [43-471. Although memory B cells were seen to proliferate in T cell zones, this only occurs for a few days after immunization. Extended proliferation of hapten-specific cells does occur, however, in follicles in stage 3 of the follicular response.This implies that proliferation in follicles per se does not necessarily result in activation of the V region gene-directed hypermutation. Transferred memory cells can respond to antigen

Sites of B cell activation

2961

pre-localized on FDC [29,30,48].This suggests that during the third phase of the follicular response memory B cells enter follicles to be activated by antigen on FDC. By this means the follicular reaction ceases to become oligoclona1. Follicular proliferation in responses t o m - L P S was seen in the present study, although this was considerably less extensive than that seen during TD responses. Also, while all follicles developed germinal centers in the TD responses, provided T cell help was not limiting, follicular proliferation was only apparent in a proportion of follicles following immunization with TNP-LPS. It is not possible from the present studies to deduce to what extent, if any, the follicular response to TNP-LPS is dependent on T cell help.The ability of TNP-LPS to evokeTcell help is implied by the finding that this antigen evokes secondary responses in euthymic mice but not congenic athymic (nude) mice. This was found using a number of TNP-LPS preparations, including that used in the present study [48]. Hoshi et al. [49] on the other hand reported follicular proliferation for at least 3 weeks in popliteal LN of nude mice injected with LPS in the footpad. Further study is required to determine whether this follicular proliferation is equivalent to that seen during responses to protein-based antigens. 4.4 Marginal zone memory B cells The present study confirms the previous observation that specific memory B cells colonize the marginal zone during T D [9] and TI-1 antibody responses [27]. The results show that these memory cells can be specifically reactivated by TD antigen to migrate from the marginal zone. When rats with DNP-specificmemory B cells and Ox-specificmemory B cells in the marginal zone were challengend with DNPMSH DNP-specific blasts were lost selectively from the marginal zones (Fig. 2d). This was associated with the transient appearance of DNP-binding B blasts in the follicles which preceded the DNP-specificT zone reaction (Fig. 3), the main follicular response came later. These findings suggest that a proportion of marginal-zone memory B cells may migrate via follicles to the T zone. In the T zones they enter cell cycle and give rise to plasmablasts which migrate to the red pulp to become plasma cells. LPS-containing preparations induce mass migration of marginal-zone B cells to follicles. This migration is not antigen specific and reverses within a day or two without the cells progressing t o the follicles [28]. There is no evidence that these memory B cells leave the marginal zone spontaneously to enter the established phase of TD responses. In this respect they probably differ from recirculating memory cells which are in a constant state of migration between the follicles of secondary lymphoid organs [50]and can respond to antigen held on FDC [29,30, 511. The importance of the present study has been to extend our understanding of the microenvironments where B cells are activated at different stages of responses to different classes of antigens. This information should assist in the design of detailed experiments to investigate the cellular and molecular interactions which result in ordered B cell proliferation and differentiation during immune responses. Received May 6, 1991; in final revised form August 21, 1991.

2962

Eur. J. Immunol. 1991. 21: 2951-2962

Y.-J. Liu, J. Zhang, F! J. L. Lane et al.

5 References 1 Nadler, L. M., in Reinherz, E. L. (Ed.), Leucocyte Typing II, Springer-Verlag 1986, p. 257. 2 Ling, N. R., MacLennan, I. C. M. and Mason, D. Y., in McMichael et al. (Eds.) Leukocyte Typing III, Oxford University Press 1987, p. 302. 3 Dorken, B., Moller, F!, Pezzutto, A., Schwartz-Albiez, R. and Moldenhauer, G., in Knapp,W. et al. (Eds.), Leucocyte Typing I v Oxford University Press 1989, p. 15. 4 Johnson, G. D., MacLennan, I. C. M., Khan, M., Hardie, D. L. and Richardson, F! R., in Knapp,W. et al. (Eds.), Leucocyte Typing I K Oxford University Press 1989, p. 183. 5 MacLennan, I. C. M., Johnson, G. D., Khan, M. andHardie, Oxford D., in Knapp, W. et al. (Eds.), Leucocyte Typing University Press 1989, p. 215. 6 Van Rooijen, N., Classen, E. and Eikelenboom, €!, Immunol. Today 1986. 7: 193. 7 Kosco, M. H., Burton, G. F., Kapasi, Z. F., Szakal, A. K. and Tew, J. G., Immunology 1989. 68: 312. 8 Jacob, J., Kassir, R. and Kelsoe, G., J. Exp. Med. 1991.173: 1165. 9 Liu, Y.-J., Oldfield, S. and MacLennan, I. C. M., Eur. J. Imrnunol. 1988. 18: 355. 10 Gratzner, H. G., Science 1982. 218: 472. 11 Tew, J. G. and Mandle, T. E., Immunology 1979. 37: 49. 12 Szakal, A. K., Kosco, M. H. and Tew, J. G., J. Immunol. 1988. 140: 341. 13 Inaba, K.,Whitmer, M. D. and Steinman, R. M., J. Exp. Med. 1984. 160: 858. 14 Fossum, S., Curr. Top. Pathol. 1989. 79: 101. 15 Knight, S. C., Balfour, B. M., OBrien, J., Buttifant, L., Sumerska, T. and Clarke, J., Eur. J. Immunol. 1982. 12: 1057. 16 Inaba, K. and Steinman, R. M., J. Exp. Med. 1984. 160: 1717. 17 Herman, I?, Monogr. Allergy 1980. 16: 126. 18 Humphrey, J. H. and Grennan, D., Eur. J. Immunol. 1981.11: 221. 19 Makela, O., Kajalanen, K. and Micklem, H. S., Scan. J. Immunol. 1978. 45: 91. 20 Claassen, E. and Van Rooijen, N., J. Histochem. Cytochem. 1985. 33: 840. 21 Kroese, F. G. M. ,Wabbena, A. S., Opstelten, D., Deenen, G. J., Schwander, E. H., De Leij, L., Vos, H., Poppema, S., Volberda, J. and Nieuwenhuis, P., Eur. J. lrnmunol. 1987. 17: 921. 22 Williams, A. F., Galfre, G. and Milstein, C., Cell 1977. 12: 663. 23 Cordell, J. L., Falini, G., Erber, W. N., Ghosh, A. K.,

lv

Abdulaziz, Z., MacDonald, S.,Pulford, K. A. F., Stein, H. and Mason, D. Y., J. Histochem. Cytochem. 1984.32: 219. 24 Weibel, E. R., Lab. Invest. 1963. 13: 131. 25 Gray, D., Chassoux, D., MacLennan, I. C. M. and Bazin, H., Clin. Exp. Immunol. 1985. 60: 78. 26 Gray, D., MacLennan, I. C. M. and Lane, P. J. L., Eur. J. Immunol. 1986. 16: 641. 27 Zhang, J., Liu,Y.-J., MacLennan, I. C. M., Gray, D. and Lane, F! J. L., Eur. J. Immunol. 1988. 18: 1417. 28 Pettersen, J. C., Borgen, D. F. and Graupner, K. C., Am. J. Anat. 1967. 121: 305. 29 Gray, D., Immunol. 1988. 65: 73. 30 MacLennan, I. C. M., Liu,Y.-J., Oldfield, S., Zhang, J. and Lane, P. J. L., Curr. Top. Microbiol. Immunol. 1990. I59: 537. 31 Lanzavecchia, A., Nature 1985. 314: 537. 32 Ron,Y. and Sprent, J., J. Immunol. 1987. 138: 2848. 33 Ho, F., Lortan, J., Khan, M. and MacLennan, I. C. M., Eur. J. Immunol. 1986. 16: 1297. 34 Corbel, C. and Melchers, F., Eur. J. Immunol. 1983. 13: 528. 35 Kroese, F. G. M., Wubenna, A. S., Seijen, H. G. and Nieuwenhuis, l?, Eur. J. Immunol. 1987. 17: 1069. 36 Liu,Y.-J., University of Birmingham, PhD thesis 1989. 37 Zhang, J., MacLennan, I. C. M., Liu,Y-J. and Lane, P.J. L., Irnmunol. Letters 1988. 18: 297. 38 Fliedner,T. M., Kress, M., Cronkite, E. I? and Robertson, J. S., Ann. N. Y Acad. Sci. 1964. 113: 578. 39 Liu,Y.-J.,Joshua,D. E.,Williams, G.T., Smith, C. A., Gordon, J. and MacLennan, I. C. M., Nature 1989.342: 929. 40 Berek, C. and Milstein, C., Immunol. Rev. 1987. 96: 23. 41 Liu,Y. J., Cairns, J. A., Holder, M. J., Abbot, S. D., Jansen, K. U., Bonnefoy, J.-Y, Gordon, J. and MacLennan, I. C. M., Eur. J. Imrnunol. 1991. 13: 1107. 42 Kroese, G. M., Seijen, H. G. and Nieuwenhuis, P., Res. Immunol. 1991. 142: 249. 43 Siekevitz, M., Kocks, C., Rajewsky, K. and Dildrop, R., Cell 1987. 48: 578. 44 Weiss, U. and Rajewsky, K., J. Exp. Med. 1990. 172: 1681. 45 McHeyzer-Williams, M. G., Nossal, G. J.V. and Lalor, P A., Nature 1991. 350: 502. 46 Fish, S., Zenowich, E., Flemmiung, M. and Manser,T., J. Exp. Med. 1989. 170: 1191. 47 Dell, C. L., Lu, Y. and Caflin, J. L., J. Immunol. 1989. 143: 3364. 48 Zhang, J., University of Birmingham, PhD thesis 1990. 49 Hoshi, H., Nagata, H. and Horie, K., Res. Pract. 1989. 184:

60. 50 Niewenhuis, P. and Ford, W. L., Cell. Immunol. 1976. 23: 254. 51 Vonderheid, R. H. and Hunt, S.V., Eur. J. Immunol. 1990.20: 79.