B Cell Memory Is Directed toward Conformational Epitopes of Parvovirus B19 Capsid Proteins and the Unique Region of VP1

Parvovirus B19 B Cell Memory •JID 2004:189 (15 May) •1873 MAJOR ARTICLE B Cell Memory Is Directed toward Conformational Epitopes of Parvovirus B19 Capsid Proteins and the Unique Region of VP1 Amanda Corcoran, 1 Bernard P. Mahon, 2 and Sean Doyle 1 1 National Institute for Cellular Biotechnology, Department of Biology, and 2 Institute of Immunology, National University of Ireland, Maynooth, Maynooth, County Kildare, Ireland Background. Loss of antibody reactivity against linear epitopes of parvovirus B19 (B19) capsid proteins VP1 and VP2 occurs after infection; however, it is unclear whether B cell memory is established against linear epitopes. Methods. B cell enzyme-linked immunospot assay was used to evaluate B19-specific B cell memory in volunteer donors ( ).np22 Results. B cell memory is maintained against conformational epitopes of VP2 and is absent against linear epitopes of VP2. Individuals seronegative for IgG against the unique region of VP1 have detectable B cell memory, with the potential to mount a humoral response on reexposure to B19. Conversely, in mice immunized with VP2, long-lasting IgG against linear epitopes of VP2 and a strong B cell–memory response are observed. Conclusions. B cell memory is established and maintained against conformational epitopes of VP2 and against linear epitopes of VP1 but not against linear epitopes of VP2. These findings further our understanding of the immune response to B19 and suggest that analysis of B19-specific B cell memory merits consideration for future B19-vaccine studies. Parvovirus B19 (B19) is a small, nonenveloped erythrovirus that usually causes a mild, self-limiting disease (called “erythema infectiosum”) in children. However, immunocompromised patients (such as patients with HIV or patients with cancer who are undergoing chemotherapy), pregnant women, and, increasingly, transplant recipients are at significant risk for B19 infection [1, 2]. Many such individuals are unable to produce neutralizing antibodies to clear the virus, and this can lead to a persistent infection resulting in anemia [1]. It is significant that many blood-product companies have begun to implement screening protocols to minimize the B19 viral load in such blood-derived products as packed red blood cells, albumin, and intravenous immunoglobulin [3]. Received 25 August 2003; accepted 1 November 2003; electronically published 26 April 2004. Financial support: Commission of the European Communities (grant QLK2-CT2001-00877); B.P.M. is a Wellcome Trust New Blood Fellow. Reprints or correspondence: Dr. Sean Doyle, National Institute for Cellular Biotechnology, Dept. of Biology, National University of Ireland, Maynooth, Maynooth, County Kildare, Ireland ([email protected]). The Journal of Infectious Diseases 2004;189:1873–80 2004 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2004/18910-0011$15.00 B19 encodes 3 main proteins: a nonstructural protein, NS1 (77 kDa), and 2 structural proteins, VP1 (83 kDa) and VP2 (58 kDa) [4]. The latter 2 proteins, which compose the viral capsid (5% VP1; 95% VP2), are colinear, except that VP1 contains a 227-residue aminoterminal extension (hereafter, “unique region”), compared with VP2 [5]. A transient, high viremia is observed in individuals after infection with B19, followed by the development, within a number of days, of B19specific IgM [6]; within a number of weeks, the presence of B19-specific IgG directed against linear and conformational epitopes of viral capsid proteins VP1 and VP2, in addition to the presence of IgG directed against NS1, is evident [3, 7, 8]. B19 infection leads to an intriguing immunological phenomenon—a specific loss of IgG, initially reactive against linear epitopes of VP1 and VP2, occurs after infection [8, 9]. In fact, specific IgG against linear epitopes of VP2 disappears within 6–12 months after infection [8]. More dramatically, there is a near-total diminution of NS1-specific IgG in previously infected individuals [10]. It is, therefore, firmly established that detection of B19-specific IgG against conformational epitopes is currently the most reliable method for the detection of previous infection with, and probable immunity to, B19. However, 1874 •JID 2004:189 (15 May) •Corcoran et al. the reason for the postinfection loss of IgG reactivity against linear epitopes of B19 capsid proteins has not yet been elucidated, and few mechanisms that would explain this loss have been suggested [8]. Recently, considerable effort has been directed toward the elucidation of the nature of the T cell response to B19 and toward the definition of T cell epitopes, yet the nature of B cell memory directed against B19 remains unclear [11, 12]. Memory B cells make a significant contribution to protective immunity and are characterized in terms of (1) a rapid proliferative response, accompanied by cellular differentiation after antigen reexposure, to produce affinity-matured, antibody-secreting plasma cells; (2) a lower activation threshold, compared with that of naive B cells, in response to cytokine and antigen; and (3) an absence of spontaneous immunoglobulin secretion. Recently, the identification of antigen-specific memory B cells against vaccine antigens (diphtheria and tetanus), whereby isolated peripheral-blood mononuclear cells (PBMCs) were nonspecifically stimulated by bacterial antigens and interleukin-2 (IL-2), followed by the specific detection of antigen-specific memory B cells, has been described [13]. This approach may also prove to be useful in the elucidation of virus-specific B cell–mediated immunity. Although B19 linear epitope–specific IgG is absent in previously infected individuals, it is currently unclear whether B cell–memory response is ever established against linear or conformational epitopes of B19 proteins in infected individuals. Furthermore, in the absence of reliable cell-culture techniques for the detection of B19, no method is currently available for the confirmation of either past exposure to the virus or serology results for plasma cell–derived B19-specific IgG. Here we describe the investigation of B cell memory against B19, an investigation using a B cell enzyme-linked immunospot(ELISpot) assay to enumerate circulating B cells that are specific to either conformational or linearized B19 capsid proteins. This technique could also be used to define the immune status achievable with candidate B19 vaccines and to aid in the design of immunization protocols that adequately induce mature immune responses and so ensure long-lasting protection. Data are presented that further our knowledge of B cell memory directed against specific viral antigens; the use of a B19-specific B cell ELISpot assay offers the possibility that past exposure to, and infection with, B19 can be confirmed. SUBJECTS, MATERIALS, AND METHODS Study population. Heparinized venous-blood samples (10 mL; ) were drawn from a cohort of healthy adult volunteersnp22 (mean age, 26 years; age range, 21–41 years) and were processed ⭐2 h after collection. Prior to the collection of samples, consent was obtained from all volunteers, and approval was obtained from the ethics committee of the National University of Ireland, Maynooth. The human-experimentation guidelines of the authors’ institution were followed in the conduct of this research, as were the animal-experimentation guidelines of thegovernment of Ireland’s Department of Health and Children. Antigens. B19 recombinant proteins VP1, VP2, and NS1 were produced in, and purified from, Spodoptera frugiperda (Sf9) insect cells infected with recombinant baculovirus encoding either VP1, VP2, or NS1 open-reading frames, as described elsewhere [10, 14–16]. Antibody assays. Plasma obtained from blood samples was used to evaluate B19-specific IgG reactivity to the viral capsid proteins. B19-specific IgG against native VP2 (VP2-N) was determined by a commercial EIA (Biotrin International) and was quantified using the World Health Organization B19-specific IgG international-standard preparation (code 93/724; National Institute for Biological Standards and Control). IgG reactivity against denatured VP1 (VP1-D) and denatured VP2 (VP2-D) was determined by EIA formats developed in-house, as described elsewhere [16]. In brief, microtiter plates (Nalge Nunc) were coated overnight at 4C with either VP1-D or VP2-D. The microtiter plates were then washed and blocked before being incubated with plasma samples that had been diluted 1:100 in PBS containing 0.05% (v/v) Tween-20 (PBST); the plates were then incubated with horseradish peroxidase (HRP)– conjugated anti–human IgG (Dako) for 30 min. After a wash step (4 washes with PBST), tetramethylbenzidine substrate was added to the wells, which were then incubated for 10 min. The enzymatic color reaction was terminated by adding 1 NH 2 SO 4 . The absorbance was read at 450 nm and at 630 nm. Immunoassay cutoffs were defined as the absorbance plus 2 SDs greater than the mean absorbance, as determined from a panel of B19-specific IgG–negative specimens. Results are expressed as index values (IV), calculated as specimen absorbance divided by cutoff absorbance (IV !1, B19-specific IgG negative; IV 11.0, B19-specific IgG positive). Avidity assays. The avidity of B19-specific IgG antibody against VP1 and VP2 was determined in the same manner as was used in the antibody assays described above, but with a modification of the wash step. Microtiter plates were coated as described above and were then incubated with plasma samples in duplicate on 2 separate plates. After being incubated, the plates were washed differently to determine avidity; one plate was washed with 4 Mof urea and PBST for 5 min, whereas the other plate was washed with only PBST for 5 min. Both plates were then washed twice with only PBST for 5 min. Antigen-bound IgG was detected as described above, by use of HRP conjugate. The avidity index (AI) was calculated as the ratio of the absorbance obtained for the plate washed with urea and PBST and that obtained for the plate washed with only PBST; Parvovirus B19 B Cell Memory •JID 2004:189 (15 May) •1875 AI is expressed as a percentage. High and low AIs are indicative of past and recent infection, respectively. Isolation of PBMCs and stimulation of cells. PBMCs were isolated by density-gradient centrifugation, as described elsewhere [17]. PBMCs ( cells/mL) were cultured for 5 days 6 5⫻10 in complete RPMI 1640, in the presence of heat-killed Staphylococcus aureus Cowan strain I cells (SACs) (Calbiochem) diluted 1:5000 and of 10 ng of IL-2/mL (Serotec). SACs and IL2 jointly function to induce generalized antibody production in resting memory B cells [13]. B cell–memory ELISpot assay. After culture, cells were washed and quantified by ELISpot assay using a modification of a protocol described elsewhere [13]. Nitrocellulose-lined 96well plates (Millipore) were coated with either VP2-N (10 mg/ mL), VP2-D (10 mg/mL), VP1-D (10 mg/mL), or rabbit anti– human IgG (10 mg/mL; Dako) in 50 mmol of sodium carbonate buffer/L (pH 9.6) overnight at 4C. Rabbit anti–human IgG facilitated detection of the total number of IgG-secreting B cells for each donor, to ensure that memory B cell activation had occurred. Plates were washed with PBS and then were blocked with 20% (v/v) fetal calf serum in RPMI 1640. After stimulation, cells (at concentrations of , , and 1⫻10 4 65 1⫻10 1 ⫻10 cells/well) were added to the plates, incubated in complete RPMI 1640 for 18 h at 37C, and then removed by washing with PBS. Secreted IgG was detected using biotinylated rabbit anti–human IgG (1 mg/mL) and subsequent use of streptavidinconjugated alkaline phosphatase (Sigma-Aldrich) at a dilution of 1:1000 (1% [w/v] bovine serum albumin in PBS). Spots were developed by the addition of 5-bromo-4-chloro-3-indolyl phosphate/p-nitroblue tetrazolium chloride substrate (SigmaAldrich) to the wells and were counted after 15 min by light microscopy. The numbers of spots in triplicate wells were averaged, and resultant data were reported as the mean number of spot-forming cells (SFCs) per 1 million starting cells (SCs). Murine serology and B cell–memory response. BALB/c mice were immunized intraperitoneally with either 50 mgof VP1-D ( ) or 50 mg of VP2-N ( ) in Freund’s comnp3np3 plete adjuvant (Sigma-Aldrich). The murine IgG response was analyzed at weeks 6 and 32 after immunization, by the antibody assays described above for the human antibody response,except that an anti–mouse IgG–HRP conjugate (Bio-Rad Laboratories) was used. Murine B cell ELISpot assays were also performed on isolated spleen cells taken from individual mice, according to the protocol described above, except that (1) goat anti–mouse IgG (Sigma-Aldrich) was used to capture total B cell–derived IgG and (2) biotinylated rabbit anti-mouse IgG (1 mg/mL) was used to detect secreted IgG. Statistical analysis. Frequencies of the level of SFCs obtained from the ELISpot assays were subjected to comparative analysis by Student’s ttest; comparison of antibody data with ELISpot results was assessed by Pearson’s correlation; titers of mouse antibody were compared by a repeated-measure analysis of variance. The statistical-software package GraphPad InStat was used for all analyses. RESULTS The present study was designed to investigate whether B cell memory is established against B19. It can be seen from figure 1 that all individuals who were seropositive for B19-specific IgG ( ; table 1) exhibit strong evidence of B cell memory,np17 whereas individuals who were seronegative for B19-specific IgG ( ; table 1) do not appear to harbor any significant levelnp5 of B19-specific memory B cells. In fact, the difference between the level of SFCs in seropositive individuals and that in seronegative individuals was statistically significant for both VP2-N and VP1-D ( and , respectively); a total of 15Pp.003 Pp.009 seropositive individuals exhibited B19-specific memory (mean SD, 87 154 SFCs/1 million SCs; range, 7.8–606 SFCs/1 million SCs) against VP2-N, and 12 seropositive individuals exhibited B19-specific memory against VP1-D (mean SD, 69110 SFCs/1 million SCs; range, 4.4–406 SFCs/1 million SCs) (figure 1). There was no statistically significant difference between the number of memory B cells directed against VP1D and that directed against VP2-N ( ); however, thePp.2 greatest number of SFCs were specific to VP2-N and were evident in 2 individuals (S008, mean SD, SFCs/222 16.4 1 million SCs; S014, mean SD, SFCs/1 million606 69 SCs). B cell memory could not be detected in the absence of IL-2 and SAC stimulation prior to ELISpot analysis (data not shown); neither was there evidence of memory directed against NS1 in 5 individuals (S008, S009, S010, S011, and S012) seropositive for B19-specific VP2 IgG, all of whom exhibited B cell memory against either VP1-D or VP2-N. When the IgG avidity of the 5 specimens (S003, S008, S010, S013, and S014) positive, by EIA, for B19-specific IgG against VP1-D were analyzed, a statistically significant positive correlation ( ) was found between the level of SFCs andrp0.96 avidity. Such a correlation was not observed for VP2-N–specific avidity ( ); neither was there a correlation between therp0.27 level of SFCs and the endpoint titer of B19-specific IgG (data not shown). Although 3 individuals (S003, S008, and S012) had circulating VP2-D IgG (detectable by EIA in all 3 [table 1] and by Western blot in S003 only), there was no evidence of any significant B cell memory against linear epitopes of VP2 (figure 1). Furthermore, the level of memory B cells observed in this cohort was not significantly different from that observed in seronegative individuals ( ). Induction of memoryP1.999 against epitopes of the B19 capsid proteins is, therefore, confined to 2 regions—linear epitopes in the unique region of VP1 and conformational epitopes present on VP2-N. VP1 cannot 1876 •JID 2004:189 (15 May) •Corcoran et al. Figure 1. Human parvovirus B19 (B19)–specific B cell memory in blood samples obtained from seropositive (SP) and seronegative (SN) volunteer donors. Results of B cell memory are expressed as B19-specific spot-forming cells (SFCs)/1 million starting cells (SCs) and are the mean SD of triplicate wells. ELISpot microplates were coated with either native VP2 (black bars), denatured VP1 (diagonally shaded bars), or denatured VP2 (horizontally shaded bars). be purified in a conformationally intact form, and so no information with respect to the conformational B cell epitopes of this protein is available at present. To test the hypothesis that there is a species-specific difference between the antigenicity of linear and conformational epitopes of the B19 capsid proteins and to explore the appropriateness of animal-model systems in B19 immunogenicity studies, a murine model was employed. The murine B19-specific IgG response against either VP1 ( ) or VP2 ( ) was an-np3np3 alyzed at weeks 6 and 32 after immunization. Equally strong responses were observed against both linear (endpoint titer, log 10 4.0) and conformational (endpoint titer, log 10 4.6) epitopes of VP2 at week 6 (figure 2A); the antibody response against linear epitopes showed no evidence of waning, because analysis of VP2-specific IgG reactivity at week 32 revealed no statistically significant difference ( ) between conformational epi-Pp.1 topes and linear epitopes (VP2-N endpoint titer, log 10 4.9; VP2D endpoint titer, log 10 4.2). In addition, there was equally strong immunoreactivity against VP1-D at week 6 (endpoint titer, log 10 3.9) and at week 32 (endpoint titer, log 10 4.1) (figure 2A). Similar results were obtained in murine B cell ELISpot analysis. However, unlike the human response to B19 infection, where there was no evidence of memory against linear epitopes of VP2, in the murine response there was strong B cell memory against both the conformational and linear epitopes of VP2 (mean  SD, and SFCs/1 million SCs, respectively) at103 55 129 11 week 32 after immunization (figure 2B). B cell memory against VP1-D was also strongly induced (mean SD, SFCs/104 6 1 million SCs) and was not significantly different ( ) fromPp.8 the response induced against VP2-N. These data indicatea strong, long-lasting, intrinsic antigenicity of the B19 capsid proteins that is directed against both the conformational and linear epitopes. DISCUSSION The present study is the first to demonstrate B cell memory against B19 VP2 and against the unique region of VP1. We have shown that B cell memory is maintained against the unique region of VP1, even in the absence of any detectable antibody against the unique region. In addition, we have shown a statistically significant positive correlation ( ) betweenrp0.96 the level of antibody directed against the unique region of VP1 and the level of SFCs. Finally, whereas B cell memory is not maintained against linear epitopes of VP2, we have shown that, in immunized mice, B cell memory does persist against linear epitopes of VP2. This result suggests that interaction between VP2 itself and the human immune system (or, conceivably, between other B19 proteins [e.g., NS1] and the human immune system) during infection may be responsible for the absence of B cell memory against linear epitopes of VP2. IgG reactivity against conformational epitopes of VP2 is now accepted as the most reliable indicator of past infection with B19 [9, 18, 19]. In the present study, 88% (15/17) of individuals demonstrating IgG reactivity against VP2-N exhibited B cell memory against VP2-N, whereas only 71% (12/17) showed evidence of B cell memory against the unique region of VP1. The absence of detectable B cell memory against linear epitopes of VP2 was quite marked, especially since 18% (3/17) of se- Parvovirus B19 B Cell Memory •JID 2004:189 (15 May) •1877 Table 1. Summary of donor serology: IgG response, in plasma obtained from adult volunteers, against the parvovirus B19 (B19) capsid proteins VP1 and VP2, in native and denatured forms. Donor ( )np22 Titer of B19-specific IgG against VP2-N, IU/mL Index value a VP2-D VP1-D B19-specific IgG seropositive S001 90 0.5 0.5 S002 49 0.5 0.6 S003 122 2.3 1.2 S004 120 0.5 0.6 S005 118 0.5 0.3 S006 21 0.6 0.5 S007 18 0.4 0.4 S008 147 1.8 2.0 S009 27 0.4 0.4 S010 105 1.0 1.2 S011 43 0.6 0.6 S012 53 1.2 0.5 S013 68 0.6 0.8 S014 140 0.6 1.2 S015 160 0.2 0.8 S016 152 0.5 0.7 S017 38 0.2 0.3 B19-specific IgG seronegative S018 1 0.4 0.4 S019 1 0.5 0.7 S020 3 0.4 0.3 S021 2 0.4 0.3 S022 1 0.1 0.4 NOTE. VP1-D, denatured VP1; VP2-D, denatured VP-2; VP2-N, native VP-2. a Index values are calculated as specimen absorbance divided by cutoff absorbance. Figure 2. A, Murine IgG response to parvovirus B19 after a single intraperitoneal immunization with B19 VP1 and VP2 antigens. At weeks 6 and 32 after immunization, thoracic-blood samples were taken from mice and levels of IgG specific for native VP2 (VP2-N), denatured VP2 (VP2-D), and denatured (VP1-D) in serum were determined by ELISA. Results are expressed as the mean log 10 endpoint titers of antibody, calculated by regression of the straight part of the curve of OD 450/630 nm versus serum dilution, to a cutoff of 2 SDs above background control values obtained from serum in naive mice. B, B19–specific B cell memory in immunized mice at week 32 after immunization. Results of B cell memory are expressed as B19-specific spot-forming cells (SFCs)/1 million starting cells (SCs) and are the mean SD of triplicate wells. ropositive individuals exhibited IgG reactivity against linear epitopes of VP2. To our knowledge, the present study provides the first confirmation that B cell memory is never established against linear epitopes of VP2 and adds further credence to the concept that long-lived plasma cells and memory B cells are 2 separate forms of independently controlled immunological memory [20, 21]. A recent report of the enumeration, in the adenoids of children, of IgG memory B cells against 4 pneumococcal antigens has also shown a broad antigen-specific response, which suggests that certain antigens are more likely than others to elicit the establishment of B cell memory [22]. It is also likely that the B19-specific B cell ELISpot assay could find application in the confirmation of past infection with B19. It is well established that the humoral immune response to B19 evolves gradually, from a population of antibodies with reactivity against both linear and conformational epitopes to one with predominantly conformational-epitope specificity [8, 9, 23, 24]. In the present study, although only 5 individuals exhibited, by either microplate immunoassay or Western blot, IgG reactivity against VP1-D, 12 individuals displayed a positive B cell–memory response against the unique region of VP1. It has been well documented that the unique region of VP1 exhibits characteristics associated with immunodominance, including the stimulation of ex vivo T cell proliferation and of cytokine secretion [16]. In addition, this region appears to elicit neutralizing-antibody responses more effectively than do other linear epitopes of B19 [3]—specifically, more effectively than the VP1 region corresponding to aa 60–100 [25]. Our analysis further reveals a statistically significant correlation between antibody avidity against the unique region of VP1 and the level of SFCs specific for VP1-D, indicating that the strength of the antibody response against the unique region of VP1 may be related to the induction of B cell memory. Conceivably, the level of reexposure to B19—or, perhaps, B19 persistence—could account for the different B cell responses to B19 infection in previously infected individuals. Even though 3 of the individuals tested (S003, S008, and S012) displayed IgG reactivity against linear epitopes of VP2 and had high avidity (AIs, 80.7, 93.2, and 40.2, respectively), 1878 •JID 2004:189 (15 May) •Corcoran et al. there was no evidence, by ELISpot assay, of memory against these VP2-specific epitopes. This phenomenon suggests that the observed IgG reactivity is a product of long-lived plasma cells that are established early in the response and is not a product of memory B cells, which are mostly established after undergoing selection in the germinal center [26]. The induction of memory after infection may favor conformational epitopes over linear epitopes because of simple stoichiometry [27]. Linear epitopes are less likely to be multivalent, thereby reducing B cell receptor cross-linking; this reduction, in turn, would lead to the imposition of a selective disadvantage in the competitive environment of the germinal center [28]. This phenomenon is exemplified in equine infectious anemia virus infection—although the initial antibody response is predominantly directed against linear determinants of the envelope protein, within 1– 2 months after infection the IgG reactivity gradually progresses to a more conformationally dependent antibody population [29]. In addition, conformational epitopes are likely to exist on intact capsids, which, when internalized, are processed by specific B cells, resulting in the presentation of a broader range of peptides to specific T cells [16]. Therefore, even though the T cells recognize only linear peptides, assistance would be preferentially directed toward B cells that are specific for conformational or discontinuous epitopes. This situation is thought to occur in immunity against polioviruses, where B cell memory against conformational epitopes is central to protection but requires support by T cells that recognize peptides from either internal or external capsid proteins [30, 31]. On the other hand, there may be other explanations for this phenomenon; for example, it is known that strong immune-activation signals can induce lymphocyte apoptosis [32]. Linear epitopes of VP2 can elicit a strong initial IgG response, and it is conceivable that such stimulation could induce B cell apoptosis; however, this scenario seems unlikely, given that no loss of memory is observed in immunized mice. Alternatively, the lack of establishment of B cell memory against linear epitopes of VP2 could be due to the absence of a “maintenance” signal, essential to the maintenance of mature B cells, from a functional B cell receptor [33], although it is difficult to construct a scenario, beyond those described above, in which this would occur selectively. Despite the absence of memory against linear epitopes of VP2 in any of the human specimens tested, in the murine model there was a strong B cell–memory response induced against VP2-D and VP1-D. The maintenance of high titers of antibody at week 32 after immunization (VP2-D endpoint titer, log 10 4.2; VP1-D endpoint titer, log 10 4.1) (figure 2A), in parallel with a high level of SFCs at week 32 for VP2-D (mean SD, 129 11 SFCs/1 million SCs) (figure 2B), demonstrates that both VP1 and VP2 possess linear epitopes that are intrinsically immunogenic and that facilitate the establishment of B cell memory. It is interesting to note that the level of SFCs observed in the present study after mice were immunized with B19 antigens is comparable to that observed (50–100 SFCs/1 million SCs) when mice were naturally infected with lymphocytechoriomeningitis virus [34]. Although this analogous B cell–memory response in infection and in immunization is significant, it does not help explain why specific epitope reactivity is lost after natural infection in humans. Furthermore, these results illustrate the difference between the immunity generated by infection in the natural host, and the difficulty of using a murine model to assess immunogenicity. The different responses observed in the human and murine models may be partly explained by the fact that Freund’s adjuvant is powerful and, therefore, may have induced stronger immunoreactivity to both linear and conformational epitopes than would be seen in natural infection. In addition, the route of delivery was different in each model: in the murine model, delivery was intraperitoneal (and, therefore, systemic), whereas the delivery was most likely mucosal in natural infection. Replicating virus may be responsible for inducing in humans host-specific cellular factors that cause specific B cell deletion and, therefore, antibody loss against linear regions of VP2. Because NS1 is a cytotoxic protein [35] that is capable of inducing apoptotic effects in the host [36], the possibility that NS1 might play a role in the loss of epitope specificity cannot be ruled out. Initial B19-vaccine studies that have been performed in mice, guinea pigs, and rabbits [37] have focused on the level of circulating B19 antibody and its associated ability to neutralize the virus. Although a correlation has been observed between the development of antibody against the unique region of VP1 and neutralization of the virus [38], the differing natures of the human and murine immune responses to B19 antigens would suggest that caution be taken in the extrapolation and application of the results of nonprimate vaccine studies to humans. More recently, the in vitro efficacy of a candidate vaccine for B19 (MEDI-491; Medimmune) has been demonstrated; in a double-blind, phase 1 trial, 24 volunteers all produced neutralizing antibody after 3 immunizations [39]. Despite the development of a candidate B19 vaccine [37, 39], there has been no analysis of the potential role that memory B cells may play in either the efficacy of B19-subunit vaccines against infection or in the persistence of immunological protection. The B19specific B cell ELISpot assay presented here would be an ideal ex vivo diagnostic tool for the evaluation of whether candidate B19-subunit vaccines have the capacity to establish B cell memory. Such an approach would be particularly important for understanding the interplay between individual titers of B19specific IgG and B19-specific memory B cells, both to establish the relative importance of each parameter in combating B19 infection in immunized individuals and to define correlates of protection. It would also have important implications in the design of booster strategies, as has been the case in vaccine Parvovirus B19 B Cell Memory •JID 2004:189 (15 May) •1879 evaluation for tetanus toxoid and diphtheria, where it has been concluded that the level of antigen-specific SFCs is an indirect measure of immunological memory [13]. To summarize, analysis of the induction of B cell memory after B19 infection shows that, although B cell memory is strongly maintained against conformational epitopes of the major capsid protein VP2, it is not maintained against its linear epitopes. Furthermore, although previously infected individuals may no longer have circulating IgG against the unique region of VP1, memory B cells persist and have the potential to mount a humoral response on reexposure to B19. Additional studies of selected patient cohorts (e.g., persistently infected individuals or individuals with a B19 infection of defined duration) should further elucidate the mechanism and extent of the absence of B cell memory against linear epitopes of B19 capsid proteins. The findings of the present study have implications for both the confirmation of B19 infection and the development of a vaccine to protect against B19 infection. Acknowledgment We thank Susanne Modrow (University of Regensburg, Germany) for helpful comments made during the preparation of the manuscript. References 1. Young NS. Parvoviruses. In: Fields BN, Knipe DM, Howley PM, eds. Fields virology. 3rd ed. Philadelphia: Lippencott-Raven, 1996:2199–220. 2. Heegaard ED, Hornsleth A. Parvovirus: the expanding spectrum of disease. Acta Paediatr 1995; 84:109–17. 3. Aberham C, Pendl C, Gross P, Zerlauth G, Gessner M. Quantitative, internally controlled real-time PCR assay for the detection of parvovirus B19 DNA. J Virol Methods 2001; 92:183–91. 4. Ozawa K, Ayub J, Yu-Shu H, Kurtzman G, Shimada T, Young N. Novel transcription map for the B19 (human) pathogenic parvovirus. J Virol 1987; 61:2395–406. 5. Shade RO, Blundell MC, Cotmore SF, Tattersall P, Astell CR. Nucleotide sequence and genome organization of human parvovirus B19 isolated from the serum of a child during aplastic crisis. J Virol 1986; 58:921–36. 6. Anderson MJ, Higgins PG, Daies LR, et al. Experimental parvovirus infection in humans. J Infect Dis 1985; 152:257–65. 7. Kurtzman GJ, Cohen BJ, Field AM, Oseas R, Blaese RM, Young NS. The immune response to B19 parvovirus and an antibody defect in persistent viral infection. J Clin Invest 1989; 84:1114–23. 8. Soderlund M, Brown CS, Spaan WJ, Hedman L, Hedman K. Epitope type–specific IgG responses to capsid proteins VP1 and VP2 of human parvovirus B19. J Infect Dis 1995; 172:1431–6. 9. Kerr S, O’Keeffe G, Kilty C, Doyle S. Undenatured parvovirus B19 antigens are essential for the accurate detection of parvovirus B19 IgG. J Med Virol 1999; 57:179–85. 10. Ennis O, Corcoran A, Kavanagh K, Mahon BP, Doyle S. Baculovirus expression of parvovirus B19 (B19V) NS1: utility in confirming recent infection. J Clin Virol 2001; 22:55–60. 11. Corcoran A, Mahon BP, McParland P, Davoren A, Doyle S. Ex vivo cytokine responses against parvovirus B19 antigens in previously infected pregnant women. J Med Virol 2003; 70:475–80. 12. Klenerman P, Tolfvenstam T, Price DA, Nixon DF, Broliden K, Oxenius A. T lymphocyte responses against human parvovirus B19: small virus, big response. Pathol Biol (Paris) 2002; 50:317–25. 13. Nanan R, Heinrich D, Frosch M, Kreth HW. Acute and long-term effects of booster immunisation on frequencies of antigen-specific memory Blymphocytes. Vaccine 2001; 20:498–504. 14. Brown CS, Van Lent JW, Vlak JM, Spaan WJ. Assembly of empty capsids by using baculovirus recombinants expressing human parvovirus B19 structural proteins. J Virol 1991; 65:2702–6. 15. Brown CS, Salimans MMM, Noteborn MHM, Weiland HT. Antigenic parvovirus B19 coat proteins VP1 and VP2 produced in large quantities in a baculovirus expression system. Virus Res 1990; 15:197–212. 16. Corcoran A, Doyle S, Waldron D, Nicholson A, Mahon BP. Impaired gamma interferon responses against parvovirus B19 by recently infected children. J Virol 2000; 74: 9903–10. 17. Ryan M, Murphy G, Ryan E, et al. Distinct T-cell subtypes induced with whole cell and acellular pertussis vaccines in children. Immunology 1998; 93:1–10. 18. Heegaard ED, Rasksen CJ, Christensen J. Detection of parvovirus B19 NS1–specific antibodies by ELISA and western blotting employing recombinant NS1 protein as antigen. J Med Virol 2002; 67:375–83. 19. Jordan JA. Comparison of a baculovirus-based VP2 enzyme immunoassay (EIA) to an Escherichia coli–based VP1 EIA for detection of human parvovirus B19 immunoglobulin M and immunoglobulin G in sera of pregnant women. J Clin Microbiol 2000; 38:1472–5. 20. Leyendeckers H, Odendahl M, Lohndorf A, et al. Correlation analysis between frequencies of circulating antigen-specific IgG–bearing memory B cells and serum titers of antigen-specific IgG. Eur J Immunol 1999; 29:1406–17. 21. Manz RA, Thiel A, Radbruch A. Lifetime of plasma cells in the bone marrow. Nature 1997; 388:133–4. 22. Zhang Q, Choo S, Finn A. Immune responses to novel pneumococcal proteins pneumolysin, PspA, PsaA, and CbpA in adenoidal B cells from children. Infect Immun 2002; 70:5363–9. 23. Kaikkonen L, Lankinen H, Harjunpaa I, et al. Acute-phase–specific heptapeptide epitope for diagnosis of parvovirus B19 infection. J Clin Microbiol 1999; 37:3952–6. 24. Manaresi E, Gallinella G, Zerbini M, Venturoli S, Gentilomi G, Musiani M. IgG immune response to B19 parvovirus VP1 and VP2 linear epitopes by immunoblot assay. J Med Virol 1999; 57:174–8. 25. Zuffi E, Manaresi E, Gallinella G, et al. Identification of an immunodominant peptide in the parvovirus B19 VP1 unique region able to elicit a long-lasting immune response in humans. Viral Immunol 2001; 14:151–8. 26. McHeyzer-Williams MG, Ahmed R. B cell memory and the long-lived plasma cell. Curr Opin Immunol 1999; 11:172–9. 27. Dal Porto JM, Haberman AM, Shlomchik MJ, Kelsoe G. Antigen drives very low affinity B cells to become plasmacytes and enter germinal centers. J Immunol 1998; 161:5373–81. 28. Bachmann MF, Hengartner H, Zinkernagel RM. T helper cell–independent neutralizing B cell response against vesicular stomatitis virus: role of antigen patterns in B cell induction? Eur J Immunol 1995; 25: 3445–51. 29. Hammond SA, Cook SJ, Lichtenstein DL, Issel CJ, Montelaro RC. Maturation of the cellular and humoral immune responses to persistent infection in horses by equine infectious anemia virus is a complex and lengthy process. J Virol 1997; 71:3840–52. 30. Mahon BP, Katrak K, Mills KH. Antigenic sequences of poliovirus recognized by T cells: serotype-specific epitopes on VP1 and VP3 and cross-reactive epitopes on VP4 defined by using CD4+ T-cell clones. JVirol1992; 66:7012–20. 31. Mahon BP, Katrak K, Nomoto A, Macadam AJ, Minor PD, Mills KHG. Poliovirus-specific CD4+ Th1 clones with both cytotoxic and helper activity mediate protective humoral immunity against a lethal poliovirus infection in transgenic mice expressing the human poliovirus re- 1880 •JID 2004:189 (15 May) •Corcoran et al. ceptor. J Exp Med 1995; 181:1285–92. 32. Donjerkovic D, Scott DW. Activation-induced cell death in B lymphocytes. Cell Res 2000; 10:179–92. 33. Lam KP, Kuhn R, Rajewsky K. In vivo ablation of surface immunoglobulin on mature B cells by inducible gene targeting results in rapid cell death. Cell 1997; 90:1073–83. 34. Slifka MK, Ahmed R. Limiting dilution analysis of virus-specific memory B cells by an ELISPOT assay. J Immunol Methods 1996; 199:37–46. 35. Moffatt S, Tanaka N, Tada K, et al. A cytotoxic nonstructural protein, NS1, of human parvovirus B19 induces activation of interleukin-6 gene expression. J Virol 1996; 70:8485–91. 36. Yaegashi N. Pathogenesis of nonimmune hydrops fetalis caused by intrauterine B19 infection. Tohoku J Exp Med 2000; 190:65–82. 37. Bansal GP, Hatfield JA, Dunn FE, et al. Candidate recombinant vaccine for human B19 parvovirus. J Infect Dis 1993; 167:1034–44. 38. Bostic JR, Brown KE, Young NS, Koenig S. Quantitative analysis of neutralizing immune responses to human parvovirus B19 using a novel reverse transcriptase–polymerase chain reaction–based assay. J Infect Dis 1999; 179:619–26. 39. Ballou WR, Reed JL, Noble W, Young NS, Koenig S. Safety and immunogenicity of a recombinant parvovirus B19 vaccine formulated with MF59C.1. J Infect Dis 2003; 187:675–8.