Follicular dendritic cells
Periarteriolar lymphocyte sheath
Chimeric (mAb 5D12)
p85 cleavage product of poly-adenosine ribose polymerase
Interactions between CD40 on APC and CD154 (CD40L) expressed by activated CD4+ T cells are crucially involved in formation and function of germinal centers (GC), but mechanistic insight into these interactions remains limited. Functional studies have mostly been restricted to experimental immunization of young-adult inbred SPF rodents that are often genetically manipulated, while studies in humans disallow in vivo manipulation. Therefore, we asked whether a functional antagonist of CD40 interferes with natural GC formation in adult cynomolgus monkeys (Macaca fascicularis) exposed to the environmental antigens of their conventional housing in captivity. Animals were treated with different doses of a unique chimeric antagonist anti-CD40 mAb (ch5D12) and analyzed 1 week or 7 weeks after last injection. Detailed in situ analysis showed that high-dose anti-CD40 treatment increased the ratio of primary over secondary follicles compared to PBS or low-dose treatment, indicative of impairment of the CG reaction. This impairment was reversible since recovery animals, except those with residual anti-CD40 levels, had normalized ratios. Anti-CD40 treatment was associated with decreased antibody production and increased numbers of apoptotic cells in GC. These data demonstrate that CD40-CD154 interactions are pivotal in physiological GC formation in primates responding to environmental antigens, and they support immunotherapeutic strategies using antagonist anti-CD40.
Extensive evidence indicates that interactions between CD40 on APC and CD154 (CD40L) expressed by recently activated antigen-specific CD4+ T helper cells are critical for the formation, maintenance, and proper functioning of germinal centers (GC) for the generation of B cell memory and plasma cell formation 1–5. However, a number of inherent constraints limit extrapolation of these studies to physiological GC reactions in humans. First, many studies have been performed in young-adult rodents housed under SPF conditions and using model thymus-dependent and -independent antigens in combination with antibodies against CD40 and CD154 1, 3–12. Secondly, some studies used highly informative and elegant but non-physiological transgenic and knockout rodent systems 11, 13–17. Thirdly, many studies employed artificial stimulation by anti-Ig or immunization with model antigens, often in strong adjuvants. Finally, analysis of the GC reaction in humans is by definition limited to in vitro functional experiments or in situ analysis of tissue sections.
Hence, several aspects of CD40-CD154 interactions in physiological GC function warrant further investigation. For instance, do environmental antigens induce detectable GC responses in adult primates housed under non-SPF conditions, and how is CD40 ligation involved? Can GC function be modulated in vivo using novel immunotherapeutics designed to antagonize CD40 ligation in humans? At which anatomical locations and between which cell types in the secondary lymphoid organs are CD40-CD154 interactions critical to GC function?
We therefore took advantage of a unique mAb that blocks CD40. mAb 5D12 against human CD40 was originally generated as a high-affinity mouse IgG2b mAb and effectively antagonizes CD40-CD154 interactions in vitro in a number of different assays including T-B interaction and macrophage activation 18. This antagonistic activity stands in contrast to most other anti-CD40 mAb, which ligate CD40 leading to intracellular signal transduction and general activation of professional APC and other CD40-expressing cell types. A recombinant chimeric human IgG4 antibody containing the variable domains of the heavy and light chains from mAb mu5D12 was subsequently engineered to reduce potential immunogenicity, to limit Fc receptor effector functions and to enhance in vivo half life in humans. Treatment of marmoset monkeys with experimental autoimmune encephalomyelitis (EAE) as a model for MS with either the mouse (murine) or chimeric (ch) version of mAb 5D12 provides clinical benefit and a diminution of autoantibody responses 19, 20, confirming that mAb 5D12 also has antagonist activity in vivo.
Since mAb 5D12 does not bind rodent CD40, the use of non-human primates as a relevant species allowing functional binding of the mAb to its epitope is obligatory to assess safety, tolerability and an initial safe dose following Food and Drug Administration guidelines. The safety study of mAb ch5D12 21 preceded phase I clinical testing. The current study used the treatment regimen of this safety assessment to determine whether adult non-human primates develop GC reactions in response to antigens present in their non-SPF environment. In addition, it was assessed whether a candidate therapeutic antagonistic anti-CD40 mAb can interfere with GC function and hence determined how CD40-CD154 interactions are involved.
2.1 Anti-CD40 treatment reverses the ratio of primary/secondary follicles
The treatment schedule of cynomolgus monkeys with mAb ch5D12 is shown in Fig. 1. It is of note that a limited number (four in total) of injections was used, amounting to 20 mg/kg or 100 mg/kg in total per animal. As described previously, no harmful (side) effects of mAb treatment were noted in an extensive analysis of all animals in vivo or by postmortem pathology 21. Crucially, however, in hematoxylin-counterstained sections of submandibular lymph nodes from high-dose animals, a strong reduction in the number of GC (secondary follicles) was seen (Fig. 2). To assess whether blocking of CD40-CD154 interactions interferes with the spontaneous occurrence of GC (i.e. in absence of deliberate immunization), extensive in situ analysis of frozen spleen tissue from all animals was performed (Figs. 3–7). The spleen was selected for analysis as this secondary lymphoid organ is connected to the blood stream and is thus optimally exposed to systemically administered anti-CD40 mAb. Fig. 3A demonstrates a reversal of the ratio of primary over secondary follicles in animals treated with the high-dose of mAb ch5D12 (25 mg/kg) and analyzed on day 28 (treatment period). The sizes of primary and secondary follicles in treated versus control animals were also taken into account (see Fig. 5 and compare to Fig. 2). Whereas both groups of animals treated with 5 mg/kg and analyzed at day 28 or day 78 showed a percentage of primary follicles around or well below 50%, the high-dose group killed early (day 28) showed a strongly increased percentage of primary follicles. The highest levels of circulating ch5D12 were detected in this group (Fig. 3B). The high-dose treatment group with recovery demonstrated that the effect of anti-CD40 on the ratio of primary over secondary follicles was reversible. Two of the animals treated with 25 mg/kg and allowed to recover until day 78 showed over 50% secondary follicles (Fig. 3A), and this paralleled the lowest levels of circulating mAb ch5D12 (Fig. 3B).
2.2 Organization of the GC reaction in cynomolgus spleen
To confirm that the GC reaction and plasma cell formation occur similarly in cynomolgus spleen as described for humans and rodents as well as to assess how anti-CD40 treatment interferes with GC function in vivo, the organization of the cynomolgus spleen with respect to the major cell types involved was determined. Fig. 4 shows consecutive spleen sections from a representative animal that received high-dose treatment followed by recovery, showing partial but not complete recovery of the GC reaction. As shown in Fig. 4A, CD3+ T cells (black) were observed scattered in the red pulp, densely packed in two periarteriolar lymphocyte sheath (PALS) areas, scattered in primary B cell follicles directly adjacent to the PALS and in a ring-like organization in a GC. The next section (Fig. 4B) shows resting B cells identified by sIgD expression (black), showing that the large PALS area is in contact with one secondary follicle with a clear GC (right, containing the ring of T cells shown in 4A), a follicle with a very small focus of GC B cells (left) and a primary follicle (top). In Fig. 4C and D, staining for IgM demonstrates resting B cells in follicles with dim surface expression, IgM-producing plasma cells with bright cytoplasmic staining and extracellular trapping of IgM-containing immune complexes by follicular dendritic cells (FDC), which are thought to be crucial for B cell memory formation. This immune complex trapping is evident in the two follicles with small or florid GC but not in the primary follicle (top). As expected, plasma cells were mostly found at the border between the PALS and GC, at the edge of the GC and in the red pulp. In analyses not shown, CD154 expression was detected on T cells found in the same location as plasma cells, consistent with previous findings in mouse spleen and human tonsil 12, 22. After in vivo administration, mAb ch5D12 could be detected bound to resting B cells expressing low levels of CD40 and CD40-expressing DC in the PALS (data not shown). This confirms that upon systemic administration, anti-CD40 readily gains access to different CD40-expressing cell populations in the spleen as previously demonstrated for the mouse version of this mAb in marmoset monkeys 19.
2.3 Spontaneous GC reaction and reversible interference by anti-CD40
Since the study design included PBS-treated animals and recovery animals, we could address the question of whether cynomolgus monkeys develop GC in the absence of infection or immunization with model antigens, i.e. to environmental antigens. In contrast to young (month-old) rodents housed under clean or SPF conditions, these adult (year-old) non-human primates were housed under conventional conditions and were thus chronically exposed to a variety of environmental antigens capable of inciting GC reactions. Indeed, both PBS-treated and recovery animals had florid GC reactions as judged by both the number and the size of individual GC. Fig. 5A shows the number and size of GC in a PBS-treated animal as visualized by sIgD expression on resting B cells. Fig. 5B is a magnification of the lower right GC of 5A, showing the high number of GC B cells. In contrast, high-dose treatment with anti-CD40 and analysis at day 28 resulted in a nearly complete abolishment of GC in the spleen (Fig. 5C). When recovery for 7 weeks was allowed, re-emergence of GC was seen (Fig. 5D) (see also quantification in Fig. 3).
The close clinical monitoring of the animals in this safety analysis protocol excludes the possibility that underlying infections are a confounder. In addition, since PBS/polysorbate without any addition of protein was used as a control treatment, it is highly unlikely that the florid GC reaction in control animals was due to the injection. Furthermore, in recovery animals, the period between the last injection of mAb and analysis was 8 weeks, and known kinetics of the GC reaction predict that GC induced by the last injection of PBS would have long waned. Collectively, these data imply that these cynomolgus monkeys experienced ongoing GC reactions induced by environmental antigens and that anti-CD40 at 25 mg/kg interfered with this reaction.
2.4 Anti-CD40 treatment inhibits antibody production
To determine whether anti-CD40 treatment also affects antibody production, the response against mAb ch5D12 was investigated. Anti-ch5D12 antibodies were detected in the circulation of 6 out of 16 antibody-treated animals at the time of death, with the highest levels found in both low-dose groups (Fig. 3B). Strikingly, none of the animals from the high-dose group killed early (day 28) displayed an antibody response against ch5D12. These data indicate that high doses of anti-CD40 also transiently interfere with the antibody response. Furthermore, circulating anti-ch5D12 antibodies were not found in half of the anti-CD40-treated animals with a ratio of primary over secondary follicles similar to PBS-treated animals (Fig. 3B), suggesting that the GC response was not only due to ch5D12 as antigen.
2.5 Anti-CD40 treatment in relation to apoptosis in GC
In view of published data that one of the functions of CD40 ligation is providing survival signals to GC B cells during selection, we investigated whether injection of anti-CD40 influences apoptosis in the GC. PARP (poly-adenosine ribose polymerase) is a target for caspase-3 in the apoptotic process, resulting in the p85 cleavage product, which is a widely accepted marker for apoptotic cells in tissue sections 23. Fig. 6 demonstrates that anti-CD40 treatment was indeed associated with an increased number of cells expressing p85PARP, whereas such apoptotic cells were lacking in the control animals. The number of apoptotic cells was not directly correlated with the percentage of primary over secondary follicles. Fig. 7A underscores that the maximum number of apoptotic cells found in an individual section of a secondary follicle of an animal treated with anti-CD40 mAb was approximately ten, while a secondary follicle in a control animal showed no apoptotic cells (Fig. 7B). These data are consistent with findings in transgenic mice, where even in systems with massive synchronized apoptosis of GC B cells 17, numbers of apoptotic cells are not overly high 24. This is most likely a reflection of the kinetics of apoptosis in relation to detectability of apoptotic markers. Since these low numbers preclude statistical testing, this analysis suggests, but does not prove, that at least one of the mechanisms by which anti-CD40 treatment reduces the number of secondary follicles is by affecting GC cell survival.
3.1 General remarks
The major findings of this study are, first, that adult cynomolgus monkeys housed under conventional conditions and without any evidence of infection or other afflictions have florid GC reactions, apparently against antigens from their environment in captivity. This is in full accordance with very recent findings of Gommerman et al. 25, who also found GC spleen in untreated control cynomolgus monkeys in a study manipulating FDC and GC by administration of recombinant lymphotoxin. These authors attributed the presence of GC in the unmanipulated situation to ‘considerably greater background immunological activity compared with rodents’. Second, the 25 mg/kg but not the 5 mg/kg dose of the therapeutic antagonist mAb ch5D12 directed against CD40 expressed on APC blocks GC formation and/or maintenance. This is in line with rodent studies using immunization with model Ag and treatment with anti-CD154 1, 3 or genetic ablation of CD154 11 or CD40 13. Third, at least one of the mechanisms involved in this blocking of GC appears to be increased apoptosis of GC cells. Fourth, high-dose treatment with mAb ch5D12 also interferes with antibody production. This result is in line with our previous findings in ch5D12-treated cynomolgus monkeys 21 and marmoset monkeys immunized with a protein antigen 20. The specificity of the effects of the antagonist mAb ch5D12 are underscored by the fact that rhesus monkeys treated in an identically designed safety assessment of an antibody against CD4 that blocks HIV-1 entry but not CD4+ T cell function showed no effect on the size or number of GC 21. Importantly, the IgG4 backbone of this engineered anti-CD4 mAb is identical to that of the mAb ch5D12 used in the current study, such that it serves as an isotype control of sorts. The current study is the first to demonstrate in non-human primates that therapeutic interference with CD40-CD154 interactions using a human-specific anti-CD40 antagonist limits the GC reaction. In full agreement with our findings, Brams et al. 26 refer to data not shown demonstrating that at a high dose (50 mg/kg weekly for 13 weeks), therapeutic mAb 24–31 directed against CD154 on activated CD4+ T cells also blocks GC formation in cynomolgus macaques upon vaccination with thymus-dependent protein antigen.
3.2 Therapeutic aspects
In view of the wide array of cellular interactions and effector functions driven by CD40-CD154 interactions, immunotherapeutics blocking these interactions are under active development for clinical use. Potential targets include diseases (mostly) driven by B cells such as SLE 27, 28 and neutralizing antibody responses against Factor VIII in treatment of hemophilia 2 but also diseases driven by CD4+ T cells with major contributions of B cells and macrophages, such as MS 19, 20, 29, Crohn's disease and rheumatoid arthritis 30. A concern when using this approach would be transient induction of generalized immunosuppression, with a hyper-IgM phenotype as the extreme result. However, this is highly unlikely since all patients over a few years of age have a fully developed memory B cell compartment providing effective continued protection against many prevalent pathogens. An unexpected finding has been that administration of anti-CD154 mAb resulted in thromboembolic complications in two independent clinical trials. It is currently unclear whether this is due to activation of complement by the antibody backbone (huIgG1), interference with functions of CD154 on platelets 31 and/or to other factors. Blocking CD40 using antagonist mAb ch5D12 is not just the mirror approach to anti-CD154 treatment, as these molecules have very different expression characteristics. Many mAb against CD40 are agonists leading to potent activation of antigen presentation as well as effector functions. This characteristic has led to novel vaccination approaches for example against polysaccharides 32. In higher doses (100 μg), agonist anti-CD40 treatment also leads to extensive proliferation of accessory cells and hypersplenism in the mouse, indicating that agonist anti-CD40 does not simply mimic T cell help 33. In contrast, antagonist anti-CD40 mAb 5D12 binds with high affinity (Kd 2.2×10–10) and effectively blocks productive ligation of CD40 by CD154. It should be emphasized here that treatment with mAb ch5D12 according to the current Good Laboratory Practice-compliant toxicity protocol for immunotherapeutics is entirely safe as evidenced by normal values for more than 20 parameters, including body weight, food consumption, hematology, clinical chemistry, urinalysis, platelet aggregation and cellular subsets 21. Although antagonist anti-CD40 potentially interferes with immune function, this analysis also excludes occurrence of infections during the study.
3.3 Mechanism of action of antagonist anti-CD40
Where and how does antagonist anti-CD40 interfere with GC responses against environmental antigen in cynomolgus monkeys? Obviously, functional interpretation of our data is limited to some extent by the complex dynamics of GC formation 34 and the inherent restrictions of the design of Good Laboratory Practice-compliant safety studies for regulatory purposes. The time points available for analyses were 1 week and 8 weeks after the last of four mAb injections. It has been argued that the view of GC organization (division into dark and light zones) based on studies of chronically inflamed human tonsil GC responses is overly static. Furthermore, much of what is known about the role of CD40-CD154 interactions in GC formation is derived from mouse studies, and species differences (in mice versus man) for CD40 signaling in B cells clearly exist. For instance, in contrast to humans, mouse GC B cells have high Bcl-2 expression.
Fig. 8 summarizes where CD40-CD154 interactions relevant to B cell responses are thought to occur in relation to the three previously defined antigen-specific checkpoints in the GC reaction 10 and the extrafollicular plasma cell reaction. In view of the systemic administration of anti-CD40 and its ready access to all APC subsets in the spleen, the mAb may act at all four of these sites. Our data on apoptosis in the GC (Figs. 6 and 7) indicate that blockade of CD40 on light zone B cells, preventing interaction with CD3+CD45RO+CD154+ T cells in the outer GC zone, which contain prestored CD154 that can be rapidly recruited 7, 12 (see Fig. 4 for location), is at least one of the mechanisms involved, probably leading to more rapid dissolution of existing GC 1.
Furthermore, the nearly complete absence of secondary follicles following administration of antagonist anti-CD40 strongly suggests that at high doses, anti-CD40 also effectively interferes with early DC-T cell interactions and B-T cell interactions (checkpoints 1 and 2 in Fig. 8), preventing de novo formation of GC. CD40 ligation in conjunction with antigen triggering also affects B cell migration. For example, in contrast to B cells triggered through anti-IgM, agonist anti-CD40-activated B cells remain competent for migration into follicles and do not undergo major changes in expression of CXCR5 (receptor for B cell zone-produced CXCL13) or CCR5 35.
3.4 Concluding remarks
In conclusion, spontaneous florid GC formation in response to environmental antigens in adult non-human primates housed in a non-SPF environment is an ongoing process that can be blocked by high but not therapeutic dosing of antagonist anti-CD40 mAb. This finding demonstrates functional biology of mAb ch5D12 in a Good Laboratory Practice-compliant setting in cynomolgus monkeys, confirming the appropriateness of this species as required by formal guidelines. It is important to note that although GC seem to be the primary location of B cell affinity maturation and the formation of B cell memory, there is evidence that they are not an absolute requirement. CD154-deficient hyper IgM syndrom (HIGM) patients, who are unable to mount GC responses, have a subset of IgM+IgD+CD27+ B cells with somatic mutations, which seem to be bona fide memory B cells 36. From the points of view of both clinical and basic science, further data on B cell responses from clinical trials with anti-CD154 and anti-CD40 in several types of inflammatory (auto)immune diseases are eagerly awaited.
4 Materials and methods
Twenty (ten male, ten female) purpose-bred cynomolgus monkeys (Macaca fascicularis) were obtained by Inveresk Research from Shamrock Farm Ltd., Sussex, UK. At commencement of the study, the animals were in the weight range of 1.9 to 2.5 kg. The animals were monitored for food consumption and body weight gain during a 3-week pretrial acclimatization period. Each animal was approved for entry into the study based on satisfactory veterinary examination, clinical observation records, body weight profile and clinical pathology investigations. Two male and two female animals were randomly allocated to each dose group and subsequently ascribed individual study numbers by the use of computer-generated number sequences. The animals were housed in couples of the same sex and dose in custom-designed cages equipped with an automatic drinking water system and food hopper. Each animal was offered 200 g Mazuri Diet (Special Diet Services Ltd., Witham, Essex) per day immediately after dosing, and a fruit supplement was offered twice weekly. Temperature, humidity and a 12 h light/dark cycle were automatically controlled in the animal rooms.
4.2 GMP-produced anti-CD40 mAb 5D12
The chimeric anti-CD40 mAb ch5D12 was produced under GMP conditions at the Tanox Inc. production facilities (Houston, TX) as previously described 21. For administration, the mAb was diluted to 5.54 mg/ml in sterile sodium PBS containing 0.02% polysorbate 80 (pH 6.96, endotoxin <0.5 EU/ml) and stored at 4°C prior to use.
4.3 Treatment schedule
All experiments were performed according to UK legislation on animal experimentation. The treatment schedule as outlined in Fig. 1 fulfills the Food and Drug Administration criteria for safety and tolerability evaluation of immunotherapeutics, which requires that a species is used in which the epitope recognized by the immunotherapeutic is expressed identically to humans. This type of Good Laboratory Practice-compliant safety/tolerability and pharmacokinetic assessment allows identification of an initially safe dose for use in humans, identification of potential target organs for toxicity, and estimation of the pharmacokinetic behavior of mAb ch5D12. Toxicology and pharmacokinetics of this mAb have been described previously 21.
Animal groups received four injections with PBS, with mAb ch5D12 at 5 mg/kg (anticipated therapeutic dose for clinical applications) or with mAb at 25 mg/kg (high dose for comparison and tolerability) at days 1, 8, 15 and 22. Animals were followed for 28 days (treatment period) or 78 days (treatment and recovery period). During the study, animals were assessed for a wide range of parameters, including clinical signs, body weight and food consumption, hematology, clinical chemistry, urinalysis, platelet and immune function, pharmacokinetics of ch5D12 and antibody responses against mAb 5D12, lymphocyte subsets (by flow cytometry) and in vivo coating of CD14+ monocytes and CD20+ B cell subsets by mAb ch5D12. Circulating serum ch5D12 and anti-ch5D12 was determined by ELISA 21. Animals were killed for gross necropsy and detailed immunohistopathological analysis 1 week after the last injection (day 28) or after a 7-week recovery period (day 78).
4.4 In situ analysis of primary and secondary B cell follicles
To assess whether treatment with antagonistic anti-CD40 mAb interferes with natural GC formation, i.e. in the absence of deliberate immunization, detailed histopathology and immunohistochemistry were performed on frozen tissue sections of secondary lymphoid organs using a panel of markers to discriminate primary and secondary B cell follicles. The immunohistochemical methods used have been described in great detail before. Briefly, 6 μm frozen sections were thaw-mounted on glass slides, stored overnight in humidified atmosphere at room temperature, air dried for 1 h the next day and subsequently fixed with fresh acetone containing 0.01% H202 (to block endogenous peroxidase) for 10 min. Sections were then incubated with primary antibodies in PBS/1% BSA overnight, followed by appropriate secondary and tertiary reagents. Antibodies against the following markers were used: CD3 as a pan T cell marker (pAb from DAKO, Glostrup, Denmark); IgD as a surface marker for resting B cells (pAb from DAKO); IgM (mAb 179–1.1-HRP from C. Persoon, TNO-Prevention and Health, Leiden, The Netherlands) as a surface marker for resting B cells, as a cytoplasmic marker for IgM-producing plasma cells and as part of immune complexes trapped in lymphoid follicles by FDC; IgG as a cytoplasmic marker for IgG-producing plasma cells and trapped IgG-containing immune complexes (mAb 343–4.1-HRP from F. Hopstaken, TNO-Prevention and Health); mAb DRC-1 (DAKO) as a marker for FDC; CD21, the complement component 3 receptor, as a marker for FDC and follicular B cells (mAb B-ly4 from PharMingen, San Diego, CA); HLA-DR as a general marker for APC (mAb L243-biotin from Coulter Clone, Hialeah, FL); CD40 expressed by APC as the target for immunotherapeutic mAb ch5D12 (using mAb mu5D12 from Tanox Pharma BV, 18); CD83 as a marker for mature DC (mAb HB15A from Immunotech, Westbrook, ME). The administered mAb ch5D12 was detected in tissue sections by incubation with a secondary antibody (mAb HP6023 from SBA, Birmingham, AL) directed against human IgG4, which does not cross-react with cynomolgus Ig. For detection of apoptotic cells, an antibody against p85PARP was used (pAb from Promega, Madison, WI). This cleavage product was selected as apoptosis marker since the use of TUNEL assay for apoptosis detection in tissue sections is not without complications 23, and death of GC B cells does not regularly include DNA fragmentation 37. In all experiments, human tonsil sections were included on the slides with cynomolgus sections as internal positive control tissues. Negative control stainings included primary antibody omission for all secondary and tertiary reagents used as well as isotype-matched control antibodies of irrelevant specificity.
4.5 Image analysis and quantification
Primary follicles and secondary follicles (containing GC) were scored using light microscopy based on the following criteria (see Fig. 4 and 5): presence and distribution of T cells (CD3; Fig. 4A); presence of strongly surface IgD+ (sIgD) resting B cells in follicles in the absence or presence of a GC (Fig. 4B), since GC B cells lack sIgD expression; confirmation by sIgM expression in adjacent sections (Fig. 4C, D); follicular immune complex trapping identified by network-like patterns of extracellular IgG and/or IgM; presence of plasma cells containing IgG or IgM as evidenced by strongly positive cytoplasm (Fig. 4C), clearly distinct from sIgM expression on resting B cells. At least two and up to five sections per staining and per animal were evaluated by two independent observers unaware of the treatment of individual animals. Inter-observer accordance approximated 100% due to the clear morphology of the cynomolgus spleen and the intense staining of most markers, most notably sIgD, allowing unambiguous identification of primary follicles (homogenously sIgD+) versus secondary follicles (clear center of blastoid cells completely devoid of sIgD, surrounded by a strongly sIgD+ rim of cells). Data are expressed as percentage of primary versus secondary follicles per animal (Fig. 3).
Part of the research was funded by the European Committee Biomed-2 program (grant BMT-4 CT 97–2131) and by the Netherlands Organization for Scientific Research-New Drug Research Foundation (NWO-NDRF, grant 014–80–007). We thank Marjan van Meurs for assistance with image analysis, Tar van Os for photomicrographs and graphs as well as Dr. Leonie Boven and Dr. Can Kesmir for critical reading of the manuscript.