Antiviral Protection and Germinal Center Formation, But Impaired B Cell Memory in the Absence of CD19

CD19^−/−, Monitor, KINDG, 129Sv, and C57/BL6 mice were obtained from the breeding colony of the Institut für Labortierkunde, Veterinary Hospital (Zurich, Switzerland). Breedings were performed under specific pathogen–free conditions, whereas experiments were done in conventional animal facilities. Mice were used at 8–12 wk of age.

VSV serotype Indiana (VSV-IND; Mudd-Summers isolate) was obtained originally from Prof. D. Kolakowsky, University of Geneva (Geneva, Switzerland). It was grown on BHK cells in MEM supplemented with 5% FCS at 37°C. LCMV (isolate WE) had been obtained originally from Dr. F. Lehmann-Grube, Heinrich Pette Institut f  ür Experimentelle Virologie und Immunologie, University of Hamburg (Hamburg, Germany). The recombinant baculoviruses expressing the glycoprotein of VSV (VSV-G) and the nucleoprotein of LCMV (LCMV-NP) were a gift from Dr. D.H.L. Bishop, NERC Institute of Virology (Oxford, UK). They were derived from nuclear polyhedrosis virus, and were grown at 28°C in Spodoptera frugiperda cells in spinner cultures in TC-100 medium (41).

Purified CVF (Naja naja) was obtained from Sigma Chemie (233550; Buchs, Switzerland). We injected twice 1.2 μg of CVF diluted in 200 μl of balanced salt solution (BSS) with an interval of 8–10 h, 16 h before immunization. The anti-CR antibody 7G6 was a gift of Prof. T. Kinoshita, Department of Immunoregulation, Research Institute for Microbial Diseases, Osaka University (Osaka, Japan). It cross-reacts extensively on CR2 and CR1 and induces a downregulation of these receptors for ∼7 d. 200 μg of antibody was diluted in 200 μl of BSS and injected once intravenously 1 d before immunization (25).

For immunizations with live virus or baculovirus-derived VSV-G (Bac VSV-G), 2 × 10⁶ PFU of VSV or 10 μg of Bac VSV-G, respectively, were injected in 200 μl of BSS i.v. or i.p. Alternatively, the same amount of virus or protein was mixed 1:1 with IFA and then injected subcutaneously at the base of tail.

Formalin inactivation of VSV-IND was performed by addition of 16 μl of 4% formalin to 1 ml of VSV-IND at a concentration of 10⁹ PFU/ml (final formalin concentration 0.0625% [42]). The mixture was incubated at 4°C for 16h. Before injection, inactivated virus was diluted with BSS so that 200 μl could be injected intravenously in every experiment.

Neutralizing titers of sera were determined as described (43). In brief, the sera were prediluted 40-fold in MEM supplemented with 5% FCS, then heat inactivated for 30 min at 56°C. Serial twofold dilutions were mixed with equal volumes of VSV diluted to contain 500 PFU/ml. The mixture was incubated for 90 min at 37°C in an atmosphere with 5% CO₂. 100 μl of serum–virus mixture was transferred onto Vero cell monolayers in 96-well plates and incubated for 1 h at 37°C. An overlay of 100 μl DMEM containing 1% methylcellulose was added. After incubation for 24 h at 37°C, the overlay was flicked off, and the monolayer was fixed and stained with 0.5% crystal violet. The highest dilution of serum that reduced the amounts of plaques by 50% was taken as the neutralizing titer. Titers are indicated as −log₂ of 40-fold prediluted sera. To determine IgG titers, undiluted serum was first pretreated with an equal volume of 0.1 M 2-mercaptoethanol in saline (44).

We used a sandwich ELISA with the following steps: (a) coating with baculovirus-derived LCMV-NP (1 μg/ml); (b) blocking with 2% BSA (Fluka AG, Buchs, Switzerland) in PBS; (c) 20-fold prediluted mouse serum, titrated 1:2 over 10 dilution steps; (d) IgM- or IgG-specific horseradish peroxidase–labeled goat anti–mouse antibodies (0.5 μg/ml; Southern Biotechnology Associates, Inc., Birmingham, AL); and (e) substrate ABTS (2.2′-azino-bis-[3-ethylbenzthiazoline-6-sulfonate]); Boehringer Mannheim, Mannheim, Germany) and H₂O₂ (Fluka AG). Plates were coated overnight at 4°C; all other incubations were for 60–90 min at room temperature (RT). Between incubations, plates were washed three times with PBS containing 0.5 ml Tween 20 per liter. OD was measured at 405 nm in an ELISA reader.

Antibody-forming cell (AFC) frequencies were determined as described (45, 46). In brief, 25–square well polystyrene plates were coated with purified VSV (∼5 × 10⁸ PFU/ml). On the next day, plates were blocked with 2% BSA in PBS (see ELISA, above) for 2 h. Titrated amounts of spleen or bone marrow cells (titration 1:5 over four dilution steps starting with 5 × 10⁶ cells/well) were added in 2% MEM and incubated for 5 h at 37°C. After washing with PBS-Tween (see ELISA, above), goat anti–mouse IgG antibody (2 μg/ml; E·Y Labs, San Mateo, CA) was added, and plates were incubated for 2 h at 37°C. After washing with PBS-Tween, alkaline phosphatase–labeled donkey anti– goat antibody (1 μg/ml; Jackson ImmunoResearch Labs, West Grove, PA) was added, and plates were incubated overnight at RT. On the next day, plates were washed, and the substrate solution (5-bromo-4-chloro-3-indolyl phosphate at 1 mg/ml in 0.6% agarose) was added to develop blue color spots. AFC numbers were then calculated per 10⁶ nucleated cells.

Single cell suspensions of splenocytes from VSV-IND–primed normal or CD19^−/− mice were injected intravenously into irradiated (450 rad) recipients. Donor cells were pooled from two individuals, and 10⁷ spleen cells were transferred to each host (groups of three animals). Recipient mice were then challenged 24 h after transfer with 2 × 10⁶ PFU of formalin-inactivated VSV. Blood was taken on day 3 after challenge to determine neutralizing IgG titers.

Freshly removed organs were immersed in HBSS and snap frozen in liquid nitrogen. Tissue sections of 5-μm thickness were cut in a cryostat, placed on siliconized glass slides, air dried, fixed with acetone for 10 min, and stored at −70°C. Secondary affinity-purified polyclonal anti-Ig antisera were diluted in Tris-buffered saline (TBS, pH 7.4) containing 5% mouse serum. All other dilutions were made in TBS alone. Incubations were done at RT for 30 min; TBS was used for all washing steps. Alkaline phosphatase was visualized using naphthol AS-BI (6-bromo-2-hydroxy-3-naphtholic acid-2-methoxy anilide) phosphate and new fuchsin as substrate. Endogenous alkaline phosphatase was blocked by levamisole. All color reactions were performed at RT for 15 min with reagents from Sigma Chemical Co. (St. Louis, MO). Sections were counterstained with hemalum. Coverslips were mounted with glycerol and gelatin.

Staining for VSV-specific B cells was done as described (47). For staining of cell differentiation markers, rehydrated tissue sections were incubated with the rat primary mAbs anti-FDC (4C11 [48]) and anti-B220 (RA3-6B2; PharMingen, San Diego, CA). Primary rat mAbs were revealed by a twofold sequential incubation with rabbit anti–rat Ig and rat alkaline phosphatase anti–alkaline phosphatase complex (Dakopatts A/S, Glostrup, Denmark). GCs were stained with peanut agglutinin (PNA [47]).

Primary neutralizing antibody responses to four different forms of VSV-G were tested (Fig. 1, A–D). VSV-G exposed in a highly repetitive order on live or formalin-inactivated viral particles has been shown to induce neutralizing IgM responses completely independent of T cells, i.e., as a TI-1 antigen. In contrast, VSV-G on the membrane of cells infected with a recombinant vaccinia virus expressing VSV-G (vacc VSV-G) and Bac VSV-G are TI-2 antigens (49). For all types of VSV-G antigens, the IgG response is TD (50).

CD19^−/− mice immunized with 2 × 10⁶ PFU of live VSV (Fig. 1,A) or vacc VSV-G (Fig. 1,C) exhibited normal neutralizing VSV-G–specific antibody titers, although in the latter case, induction of IgM antibodies was delayed 3 d compared with control mice. CD19^−/− mice survived these infections normally without displaying any morbidity. In contrast, antibodies to formalin-inactivated VSV (Fig. 1,B) and Bac VSV-G protein (Fig. 1 D) were reduced markedly. This result differs from recent experiments with haptenated antigens, where the TI-2 response to DNP-Ficoll (36) but not to NP-Ficoll (17) was enhanced.

Antibodies to LCMV have been shown to be dependent on T helper cells for the IgM and IgG response. Therefore, we tested CD19^−/− mice for their capacity to mount LCMV-NP–specific antibodies after immunization with either live virus intravenously (100 PFU LCMV strain WE; Fig. 1,E) or with UV-inactivated virus subcutaneously (300 μl of purified LCMV in CFA corresponding to ∼20 μg of protein; Fig. 1,F). IgG (Fig. 1, E and F) as well as IgM (data not shown) to both types of antigens were decreased when measured in an NP-specific ELISA. This result confirms the published severe defect of these mice in TD antibody responses (17, 18). But there is a striking difference between a low dose of replicating and a high dose of inactivated virus in CD19^−/− mice; only the latter nonreplicating antigen failed to induce IgG antibodies in CD19^−/− mice.

B cell development in CD19^−/− mice has been shown to be virtually normal except for a severe reduction of peritoneal B-1 cells (17, 18). To see whether this has any consequence for antiviral antibodies, we immunized CD19^−/− and control mice with live VSV, formalin-inactivated VSV, or Bac VSV-G via three different application routes (Table 1): intravenously, subcutaneously in IFA, or intraperitoneally. In the case of Bac VSV-G, the reduction in neutralizing titers in CD19^−/− mice was comparable for all three application routes, although it was slightly more pronounced after subcutaneous immunization; this indicated that IFA could not compensate for a CD19-mediated costimulation of B cells.

In contrast, after intraperitoneal immunization with live or, with a much more pronounced effect, formalin-inactivated virus, neutralizing titers were reduced by a factor of 8–100, suggesting a role for B-1 cells in early anti-VSV immune responses after intraperitoneal infection.

To test the hypothesis that CD19 functions as a signal transducer for CR2, we looked for the ability of CD19^−/− mice to produce VSV-specific neutralizing antibodies after interference with the complement system in two different ways: either the complement component C3 was depleted by CVF (Fig. 2,A), or the complement receptors CR1 and CR2 were blocked by the mAb 7G6 (Fig. 2 B). Mice were then immunized with 2 × 10⁶ PFU of formalin-inactivated VSV.

Both treatments led to more pronounced reduction of IgG than IgM responses, confirming earlier results showing that VSV activates B cells to produce IgM mostly independent of additional signals, whereas switch to IgG depends on CD4⁺ T cells and complement (31). In addition, Fig. 2 shows that (a) the titers in CD19^−/− mice were lower than in C3-depleted or (less pronounced) in anti-CR–treated controls, and (b) C3 depletion led to a further reduction in CD19^−/− mice. This second finding suggests that complement has effects on B cells that are not CD19 mediated, as for example an opsonizing function of complement for enhanced antigen presentation (51). The first finding could be explained by incomplete blocking of the complement system with CVF or anti-CR, or it may indicate a direct complement-independent function of CD19 on B cells. Whether this needs triggering via a separate potential CD19 ligand is not known. But direct association of CD19 with sIgM on B cells has already been demonstrated (13, 14), and may indicate a direct influence on sIg signaling by CD19 (52).

More than 20 years ago, an important role of complement for maintenance of B cell memory was postulated (53) and later confirmed (54, 55). By opsonization of antigen or immune complexes, complement would either enhance the persistence of antigen on follicular dendritic cells (FDCs) or deliver special signals to memory cells via CRs. Since CR2 is thought to transduce signals through CD19, we looked at B cell memory in CD19^−/− and control animals. These mice were immunized with a standard dose of 2 × 10⁶ PFU of VSV, which is known to induce life-long immunity in mice (47). To assess B cell memory, we chose three different readouts. (a) First, we looked at maintenance of neutralizing IgG titers over ∼100 d after immunization with live VSV (Fig. 3,A) or Bac VSV-G (Fig. 3,D); after live VSV infection of CD19^−/− mice, neutralizing antibody titers dropped from 10 to 2 within 105 d, and after immunization with Bac VSV-G, they dropped to 1 within 30 and to 0 within 120 d. In both experiments, control animals maintained the titers obtained by day 20 within a factor of two to four (or within one to two titer steps). (b) As a readout for memory B cells, we took spleen cells from animals of the same experiment as in Fig. 3,A at three different time points and adoptively transferred them into irradiated recipient mice. 1 d after transfer, the recipients were challenged with 2 × 10⁶ PFU of formalin-inactivated virus, and neutralizing IgG titers 3 d after challenge were taken as a readout of B cell memory (Fig. 3,B). In control mice, the memory titers after challenge did not drop significantly over 105 d, whereas in CD19^−/− mice, a significantly lower titer than in controls could be measured as early as day 12. This titer had disappeared completely by the later time points tested. In a similar experiment (Fig. 3,E), memory spleen cells were transferred into irradiated transgenic recipients expressing the VSV-G either as a soluble protein in the serum (Monitor [56]) or under a class I promoter on the membrane of most cells (KINDG [57]). The antigenicity of these transgenes has been shown previously to restimulate memory B cells (58). Although in Monitor mice, only CD19-competent cells were able to mount a small neutralizing titer, in KINDG mice, CD19^−/− cells also produced some neutralizing antibody, but at least 100-fold lower than control cells. (c) As a readout for VSV-specific AFCs (Fig. 3,E), enzyme-linked immunospot assays of bone marrow and spleen cells were performed, again from animals of the experiment described in Fig. 3 A. Plasma cell frequencies were determined at three different time points. Although neutralizing antibody titers in the serum did not differ between CD19^−/− and control mice 12 d after immunization, AFC frequencies in CD19^−/− mice were already lower by a factor of eight in the spleen and almost undetectable in the bone marrow. After 46 and 105 d, AFC numbers in CD19^−/− mice were below the detection level, whereas in CD19-competent mice, significant numbers of plasma cells could be measured in spleen and bone marrow for up to 105 d.

A severe defect in the generation of GCs after immunization with TD antigens has been described for CD19^−/− (17, 36) as well as C3^−/− (59) and CR^−/− mice (60). Since maintenance of B cell memory is thought to be governed by persisting antigen in immune complexes on FDCs in GCs (61), we wanted to see whether the loss of VSV-specific B cell memory in CD19^−/− mice was due to the failure to generate GCs. From the same mice shown in Fig. 3, A–C, spleen sections were prepared and stained immunohistochemically for VSV-specific B cells (see Materials and Methods, and reference 47; Fig. 4, A and B), GC B cells (PNA; Fig. 4, C and D), FDCs (4C11; Fig. 4, E and F), and total B cells (B220; Fig. 4, G and H). Fig. 4 shows spleen sections derived from the group of mice killed 46 d after immunization. Surprisingly, in CD19^−/− as well as in CD19-competent mice, VSV-specific GCs could be observed until day 105; despite this, the former mice lost their neutralizing antibody titers. Quantitative analysis (Table 2) showed a difference in numbers of GCs, with the CD19^−/− mice about twofold lower than the controls. In addition, GCs of CD19^−/− mice appeared smaller in size (diameter reduction of ∼30%). Quantitative analysis of overall GC numbers yields a reduction of maximally sevenfold [(1.5)³ × 2 = 6.75]. This reduction is substantial, but cannot explain the total loss of VSV-specific neutralizing antibodies solely on a quantitative basis. In general, CD19 was apparently not necessary for generation of antigen-specific GCs to VSV-G–neutralizing epitopes.

The role of CD19 and associated CR2 for costimulation of B cells has been clearly demonstrated in vitro as well as with CD19^−/− (17, 18) and CR^−/− mice (29, 30) in vivo. Complement has been postulated to have a bridging role between innate and adaptive immunity (35). This study first looked at B cell responses to live virus infections in CD19^−/− mice. Mice were immunized with VSV, which induces TI-1 IgM antibody response and is controlled by the subsequent TD IgG response; alternatively, vacc VSV-G was used, which induces TI-2 antibodies and is controlled by cytokines and T cells, or LCMV, an Arenavirus that induces strictly TD antibody responses and is controlled by cytotoxic T cells. CD19^−/− mice were protected against all of these infections and did not show any morbidity. Consistent with published results from studies with haptenated protein antigens, the TD antibody responses were reduced markedly. But in contrast to another study (15), TI-2 responses were not enhanced, and TI-1 responses were only reduced when nonreplicating antigen was used. This finding may be explained by the following: (a) immunization with live virus, even if only an abortive replication cycle occurs, may produce a greater antigen load. This correlates with published studies that show decreasing dependence of B cells upon costimulation via complement (and probably CD19) with increasing antigen dose (22, 23, 31). (b) In contrast to protein antigens, immunization with live virus always leads to nonspecific inflammatory reactions and secretion of cytokines (62), among which IFN-α has been shown to bind to CR2 (63) and to provide survival signals for GC B cells (64). Whether this mechanism has any relevance during infections needs to be further investigated. Our results at least indicate that such a stimulation via CR2 would be independent of CD19.

CD19^−/− animals have a severe defect in the formation of B-1 B cells, whereas B-2 cells are normal or only slightly reduced. B-1 cells are thought to produce natural antibodies that may also be involved in autoimmunity (65) as well as in early defense mechanisms against pathogens. Whether the absence of CD19 and the defect in peritoneal B-1 cells have consequences for antiviral immunity was investigated here by comparing immunizations with VSV via various infection routes. Indeed, when the virus was given intraperitoneally, neutralizing antibody titers were reduced. This difference was even more drastic when the virus was inactivated. But in the case of live virus infection, it did not have any effect on protection of CD19^−/− animals. These results may indicate a crucial role of the route of antigen application, or they reveal a biological function of the B-1 cell compartment.

GC formation in CD19^−/− mice has been shown to be severely impaired after immunization with haptenated protein antigens (17, 36). But consequences for generation of B cell memory have not been analyzed. These results indicate that memory antibody titers, AFC frequencies, and memory B cells detected by adoptive transfer are severely impaired in these mice. Most probably, the generation of memory B cells is already affected, because—although antibody titers 12 d after infection were comparable to control mice—AFC and memory B cells were already clearly reduced at that time. This finding was independent of the generation of GCs, which could be found in both CD19^−/− and control animals, even at >100 d after infection, as described for the latter (47). These findings question the strict correlation between GC formation and establishment or possibly maintenance of B cell memory (66). GCs are thought to be the structural environment where somatic mutation occurs, high-affinity B cells are selected (67), and potentially autoreactive B cells are eliminated (68). Our results suggest that GC formation may be necessary but not sufficient for generation or maintenance of B cell memory. This supports evidence obtained in earlier studies using haptenated LPS (69), aged mice (70), or blocking of B7-2 in established GCs, which impaired B cell memory (71). CD19 may provide additional signals for generation and survival of memory B cells. Earlier studies that demonstrate the binding of CD19 to CD77, a marker of GC B cells, would be in line with this hypothesis (72, 73).

We would like to thank M.F. Bachmann and U. Karrer for stimulating discussions, A. Oxenius for providing purified LCMV, A. Althage for excellent technical assistance, L. Vek for immunohistology, and H. Nef and N. Wey for photographs.

This work was supported by the Swiss National Science Foundation (grant 31-32195.91 to R.M. Zinkernagel), the Kanton of Zürich, the Deutsche Forschungsgemeinschaft (SFB243), and the Human Frontier Science Program.

AFC

antibody-forming cell

Bac VSV-G

baculovirus-derived VSV-G

BSS

balanced salt solution

CVF

cobra venom factor

FDC

follicular dendritic cell

GC

germinal center

LCMV

lymphocytic choriomeningitis virus

LCMV-NP

LCMV nucleoprotein

PNA

peanut agglutinin

RT

room temperature

sIg

surface Ig

TD

T cell–dependent

TI

T cell–independent

vacc VSV-G

recombinant vaccinia virus expressing VSV-G

VSV

vesicular stomatitis virus

VSV-G

VSV glycoprotein

VSV-IND

VSV Indiana serotype