Central role for B lymphocytes and CD4+ T cells in immunity to infection by the attaching and effacing pathogen Citrobacter rodentium

Citrobacter rodentium, an attaching-effacing bacterial pathogen, establishes an acute infection of the murine colonic epithelium and induces a mild colitis in immunocompetent mice. This study describes the role of T-cell subsets and B lymphocytes in immunity to C. rodentium. C57Bl/6 mice orally infected with C. rodentium resolved infection within 3 to 4 weeks. Conversely, systemic and colonic tissues of RAG1(-/-) mice orally infected with C. rodentium contained high and sustained pathogen loads, and in the colon this resulted in a severe colitis. C57Bl/6 mice depleted of CD4(+) T cells, but not CD8(+) T cells, were highly susceptible to infection and also developed severe colitis. Mice depleted of CD4(+) T cells also had diminished immunoglobulin G (IgG) and IgA antibody responses to two C. rodentium virulence-associated determinants, i.e., EspA and intimin, despite having a massively increased pathogen burden. Mice with an intact T-cell compartment, but lacking B cells ( micro MT mice), were highly susceptible to C. rodentium infection. Systemic immunity, but not mucosal immunity, could be restored by adoptive transfer of convalescent immune sera to infected micro MT mice. Adoptive transfer of immune B cells, but not na?ve B cells, provided highly variable immunity to recipient micro MT mice. The results suggest that B-cell-mediated immune responses are central to resolution of a C. rodentium infection but that the mechanism through which this occurs requires further investigation. These data are relevant to understanding immunity to enteric attaching and effacing bacterial pathogens of humans.     

Influenza Virus-Specific Neutralizing IgM Antibodies Persist for a Lifetime

Detection of immunoglobulin M (IgM) antibodies has long been used as an important diagnostic tool for identifying active viral infections, but their relevance in later stages has not been clearly defined in vivo. In this study, we followed the kinetics, longevity, and function of influenza virus-specific IgM antibodies for 2 years following sublethal infection of mice with live mouse-adapted A/PR/8/34 virus or immunization with formalin-inactivated virus. These groups mounted robust protective immune responses and survived lethal challenges with 50× 50% lethal dose (LD₅₀) mouse-adapted A/PR/8/34 virus 600 days after the primary exposure. Surprisingly, the virus-specific IgM antibodies persisted along with IgG antibodies, and we found a significantly higher number of IgM-positive (IgM⁺) virus-specific plasma cells than IgG⁺ plasma cells that persisted for at least 9 months postexposure. The IgM antibodies were functional as they neutralized influenza virus in the presence of complement just as well as IgG antibodies did.     

Carbohydrate biopolymers enhance antibody responses to mucosally delivered vaccine antigens

We have evaluated the ability of two carbohydrate biopolymers, chitosan and gellan, to enhance antibody responses to subunit influenza virus vaccines delivered to the respiratory tracts of mice. Groups of mice were vaccinated three times intranasally (i.n.) with 10 microg of purified influenza B/Panama virus surface antigens (PSAs), which consist of hemagglutinin (HA) and neuraminidase (NA), either alone or admixed with chitosan or gellan solutions. Separate groups were vaccinated subcutaneously (s.c.) with PSAs adsorbed to Alhydrogel or chitosan or gellan alone i.n. Serum antibody responses were determined by enzyme-linked immunosorbent assay (ELISA) for influenza virus-specific immunoglobulin G (IgG) and by HA inhibition (HAI) and NA inhibition (NAI) assays. The local respiratory immune response was measured by assaying for influenza virus-specific IgA antibody in nasal secretions and by enumerating nasal and pulmonary lymphocytes secreting IgA, IgG, and IgM anti-influenza virus-specific antibodies by enzyme-linked immunospotting (ELISPOT). When administered alone i.n., B/Panama PSA was poorly immunogenic. Parenteral immunization with B/Panama PSA with Alhydrogel elicited high titers of anti-B/Panama antibodies in serum but a very poor respiratory anti-B/Panama IgA response. In contrast, i.n. immunization with PSA plus chitosan stimulated very strong local and systemic anti-B/Panama responses. Gellan also enhanced the local and serum antibody responses to i.n. PSA but not to the same extent as chitosan. The ability of chitosan to augment the immunogenicity of influenza vaccines given i.n. was confirmed using PSA prepared from an influenza A virus (A/Texas H1N1).     

T cell-independent and T cell-dependent immunoglobulin G responses to polyomavirus infection are impaired in complement receptor 2-deficient mice

Polyomavirus (PyV) infection induces protective T cell-independent (TI) IgM and IgG antibody responses in T cell-deficient mice, but these responses are not generated by immunization with viral proteins or virus like particles. We hypothesized that innate signals contribute to the generation of isotype-switched antiviral antibody responses. We studied the role of complement receptor (CR2) engagement in TI and T cell-dependent (TD) antibody responses to PyV using CR2-deficient mice. Antiviral IgG responses were reduced by 80-40% in CR2-/- mice compared to wild type. Adoptive transfer experiments demonstrated the need for CR2 not only in TD, but also in TI IgG responses to PyV. Transfer of CR2-/- B lymphocytes to SCID mice resulted in TI antiviral IgG responses that corresponded to 10% of that seen in wild-type B cell-reconstituted mice. Thus, our studies revealed a profound dependence of TI and TD antiviral antibody responses on CR2-mediated signals in PyV-infected mice, where the viral antigen is abundant and persistent.     

Immunoglobulin A-deficient mice exhibit altered T helper 1-type immune responses but retain mucosal immunity to influenza virus

We have previously demonstrated that immunoglobulin A (IgA)(-/-) knockout (KO) mice exhibit levels of susceptibility to influenza virus infection that are similar to those of their normal IgA(+/+) littermates. To understand the mechanism of this apparent mucosal immunity without IgA, immunoglobulin isotype and T helper 1 (Th1)-type [interferon-gamma (IFN-gamma)] and Th2-type [interleukin (IL)-4, IL-5)] cytokine responses to influenza vaccine were evaluated. Intranasal immunization with influenza virus subunit vaccine plus cholera toxin/cholera toxin B subunit (CT/CTB) induced significant influenza virus-specific immunoglobulin G (IgG) antibody in the serum and nasal passages of both IgA(-/-) and IgA(+/+) mice, while IgA antibodies were induced only in IgA(+/+) mice. IgA KO mice exhibited an IgG1 subclass haemagglutinin (HA)-specific response but no detectable IgG2a and IgG2b responses. In contrast, IgA(+/+) mice exhibited significant IgG1 as well as IgG2a responses. This indicates a predominant Th2-type response in IgA KO mice compared to normal mice. Following stimulation with influenza virus in vitro, splenic lymphocytes from immunized IgA(-/-) mice produced significantly lower levels of IFN-gamma than IgA(+/+) mice (P < 0.001), but elaborated similar levels of IL-4 and IL-5. This was true at both protein and mRNA levels. Immunized mice were challenged intranasally with a small inoculum of influenza virus to allow deposition of virus in the nasal mucosal passages. Compared to non-immunized mice, immunized IgA(-/-) and IgA(+/+) mice exhibited significant, but similar levels of reduction in virus titres in the nose and lung. These results demonstrate that in addition to IgA deficiency, IgA gene deletion also resulted in down-regulated Th1-type immune responses and confirm our previous data that IgA antibody is not indispensable for the prevention of influenza virus infection.     

Local IL-4 expression in the lung reduces pulmonary influenza-virus-specific secondary cytotoxic T cell responses

We studied the effect of lung-specific IL-4 expression on the T cell response during primary and secondary heterologous infection with influenza virus by using transgenic mice that express IL-4 under a lung-specific promoter. Subsequent to primary infection with a type A/H1N1 influenza virus these transgenic mice exhibited similar local recruitment of CD4(+) and CD8(+) T cells and only slightly decreased virus-specific CTL activity. However, during secondary challenge with a heterologous influenza virus, the local infiltration with virus-specific, MHC class I-restricted CD8(+) T cells was significantly decreased compared to that of nontransgenic littermates. The ability of IL-4 transgenic mice to clear the heterologous infection was delayed but not abrogated. This was associated with a faster virus-neutralizing antibody response in IL-4 transgenic mice and with their ability to mount significant Th1 responses even in the presence of increased local IL-4 expression. Our observations demonstrate a negative regulatory effect of IL-4 on memory Tc1/CD8(+) T cells, but are also consistent with complementary mechanisms important for virus clearance such as virus-neutralizing antibodies. The reduction of memory CTL in the presence of IL-4 might have consequences for understanding the course of influenza infection in situations where T(H)2 immunity is increased.     

Granzyme K expressing cytotoxic T lymphocytes protects against influenza virus in granzyme AB-/- mice

Granzyme (grz) AB(-/-) H2(b) mice generate numerically normal cytotoxic T lymphocyte (CTL) responses to the prominent influenza A virus D(b) NP(366) and D(b) PA(224) epitopes and terminate the infectious process in the pneumonic lung with the same kinetics as the WT controls. Though grz B protein expression is fully compromised, there is only a partial effect on the level of CTL activity measured in a classical, short-term (51)Cr release assay. Single-cell polymerase chain reaction (PCR) analysis of both highly activated effector and "resting" memory CD8(+) T cells from influenza A virus-infected grzAB(-/-) mice showed a high prevalence of grzK mRNA(+) expression in tetramer (tet)(+) CTLs as was found in WT mice. However, a marked reduction in cytotoxicity present in the primary splenic CTLs of grzAB(-/-) mice correlated with decreased grzK expression, as measured by real-time PCR. This suggests that grzK plays an important role in CD8(+) T-cell cytotoxicity both in the presence and absence of grzA and B.     

Mucosal immunization with attenuated Shigella flexneri harboring an influenza hemagglutinin DNA vaccine protects mice against a lethal influenza challenge

Mucosal surfaces are important for the induction of immunity against influenza virus. In a murine intranasal immunization model, we demonstrated that the attenuated Shigella flexneri Deltaasd strain 15D, carrying a DNA construct encoding the influenza virus hemagglutinin (HA), induces protective immunity against a lethal respiratory challenge with influenza A/WSN/33. Influenza virus-specific IFN-gamma T cells were detected among splenocytes, and anti-HA IgG was detected in serum post-immunization, albeit at low levels. Following influenza virus challenge, an accelerated anti-HA IgA antibody response was detected in bronchoalveolar lavage (BAL) washings from mice vaccinated with attenuated shigella containing the HA construct. These results suggest that S. flexneri Deltaasd strain 15D is a promising vector for mucosal DNA vaccine immunization against influenza virus and other mucosal pathogens.     

Mucosal immunization of CD4+ T cell-deficient mice with an inactivated virus induces IgG and IgA responses in serum and mucosal secretions

Immunoglobulin (Ig) class switching can occur in the absence of alphabeta+ or gammadelta+ T cells when mice are infected with certain live viruses, although CD4 T helper cells are believed to be essential for induction of a high-affinity antibody response and for efficient isotype switching from IgM to IgG and IgA production. However, little information is available about the immune responses after mucosal immunization of CD4+ T cell-deficient mice with inactivated virus. In this study, we show that intranasal immunization with formalin-inactivated influenza A/PR8/34 virus induces IgG and IgA responses in serum and IgA responses in mucosal secretions in CD4+ T cell-deficient mice. All four subclasses of IgG were produced. IgG1/IgG2a ratios were found to be from 1 to 1.75, indicating that both Th1 and Th2 immune responses are induced by the inactivated influenza virus. The sera and mucosal secretions were found to have neutralizing activity against influenza virus in vitro. In addition, the mucosally immunized CD4+ T cell-deficient mice were protected completely from challenge with a lethal dose of live, pathogenic influenza virus. To our knowledge, this is the first demonstration that mucosal immunization with an inactivated virus induces immune responses in serum and mucosal secretions in CD4+ T cell-deficient mice.     

B-cell-deficient mice show an exacerbated inflammatory response in a model of Chlamydophila abortus infection

The resolution of Chlamydophila abortus (Chlamydia psittaci serotype 1) infection is dependent on gamma interferon and CD8(+) T cells, and classically, B cells have been considered to play a minimal role in host defense. The role of B cells in the immune response was studied by using a model of infection in mice with genetically modified immunoglobulin M transmembrane domains ( micro MT). In the absence of B cells, infection with C. abortus leads to an acute severe fatal disease that involves a disseminated intravascular coagulation syndrome. micro MT mice displayed an increased level of proinflammatory cytokines in serum, and an increased number of neutrophils was observed in the lesions. The possible deleterious role of neutrophils in the pathogenesis of disease in micro MT mice was determined by depletion of the neutrophils with the monoclonal antibody RB6-8C5. This led to an enhancement of the bacterial burden and early mortality in both micro MT and wild-type mice, while necrotic lesions remained. Analysis of the presence of immunoregulatory cytokines showed significantly lower levels of transforming growth factor beta in the sera of micro MT mice. However, mice lacking mature B cells were able to establish a specific immune response that protected them from a secondary challenge. Taken together, these data suggest an immunomodulatory role for B cells in the early events of C. abortus primary infection that can protect mice against an exaggerated inflammatory response.     

CD59a deficiency exacerbates influenza-induced lung inflammation through complement-dependent and -independent mechanisms

Influenza-specific immune activity not only promotes virus clearance but also causes immunopathology, thereby underlining the importance of mounting a measured anti-viral immune response. Since complement bridges both the innate and adaptive immune systems and has been implicated in defence against influenza, the role of the complement regulator CD59a in modulating the response to influenza was explored. For this purpose, immune responses to influenza virus, strain E61-13-H17, in mice deficient in the complement regulator protein CD59a (Cd59a(-/-) mice) were compared to those in wild-type mice. The severity of lung inflammation was significantly enhanced in the lungs of Cd59a(-/-) mice with increased numbers of infiltrating neutrophils and CD4(+) T cells. When complement was inhibited using soluble complement receptor 1, the frequency of lung-infiltrating neutrophils in influenza-infected Cd59a(-/-) mice was much reduced whilst numbers of CD4(+) T cells remained unchanged. These results demonstrate that CD59a, previously defined as a complement regulator, modulates both the innate and adaptive immune response to influenza virus by both complement-dependent and -independent mechanisms.     

Enhanced viral clearance in interleukin-18 gene-deficient mice after pulmonary infection with influenza A virus

T helper 1 driven immune responses facilitate host defence during viral infections. Because interleukin-18 (IL-18) mediates T helper 1 driven immune responses, and since mature IL-18 is up-regulated in human macrophages after influenza virus infection in vitro, it has been suggested that IL-18 plays an important role in the immune response to influenza. To determine the role of IL-18 in respiratory tract infection with influenza, IL-18 gene-deficient (IL-18(-/-)) and normal wildtype mice were intranasally inoculated with influenza A virus. Influenza resulted in an increase in constitutively expressed IL-18 in the lungs of wildtype mice. The clearance of influenza A was inhibited by IL-18, as indicated by reduced viral loads on day 8 and day 12 after infection in IL-18(-/-) mice. This enhanced viral clearance correlated with increased CD4(+) T-cell activation in the lungs as reflected by CD69 expression on the cell surface. Surprisingly, interferon-gamma (IFN-gamma) levels were similar in the lungs of IL-18(-/-) mice and wildtype mice. Intracellular IFN-gamma staining revealed similar expression levels in lung-derived natural killer cells, CD4(+) and CD8(+) T cells, indicating that IFN-gamma production is IL-18-independent during influenza virus infection. Tumour necrosis factor-alpha production by CD4(+) T cells was significantly lower in IL-18(-/-) mice than in wildtype mice. Our data indicate that endogenous IL-18 impairs viral clearance during influenza A infection.     

Anti-viral immunity induced by recombinant nucleoprotein of influenza A virus. II. Protection from influenza infection and mechanism of protection

Protection against influenza A virus infection in mice immunized with recombinant nucleoprotein (rNP) was studied. Nucleoprotein-immune mice were protected against a lethal challenge with A/Puerto Rico/8/34 (A/PR8) virus but showed considerable morbidity before recovery. Local boosting of the immune system with rNP by intranasal immunization improved the protection in NP-immune mice, and the decrease in morbidity after infection was reflected in accelerated clearance of virus from lungs. However, immune, boosted mice also rapidly cleared an antigenically unrelated influenza B virus from their lungs. Mice immunized with rNP precipitated with alhydrogel, that subsequently developed significant resistance to virus infection, failed to generate detectable levels of class I major histocompatibility complex (MHC)-restricted killer cells. Furthermore, B10.A(5R) mice that are non-responders for NP-specific class I killer cells could also be protected by immunization with rNP. In contrast, rNP-immunized mice developed strong proliferative T-cell responses to rNP. These data argue for an important role for helper T cells rather than virus-specific class I cytotoxic T cells in protection against influenza virus infection induced by rNP.     

Cathelin-related antimicrobial peptide differentially regulates T- and B-cell function

Mammalian antimicrobial peptides (AMPs) play an important role in host defense via direct antimicrobial activity as well as immune regulation. The mouse cathelin-related antimicrobial peptide (mCRAMP), produced from the mouse gene Camp, is the only mouse cathelicidin identified and the ortholog of the human gene encoding the peptide LL-37. This study tested the hypothesis that mouse B and T cells produce and respond to mCRAMP. We show that all mature mouse B-cell subsets, including follicular (FO), marginal zone (MZ), B1a, and B1b cells, as well as CD4(+) and CD8(+) T cells produce Camp mRNA and mCRAMP protein. Camp(-/-) B cells produced equivalent levels of IgM, IgG3, and IgG2c but less IgG1 and IgE, while Camp(-/-) CD4(+) T cells cultured in Th2-inducing conditions produced more IL-4-expressing cells when compared with WT cells, effects that were reversed upon addition of mCRAMP. In vivo, Camp(-/-) mice immunized with TNP-OVA absorbed in alum produced an enhanced TNP-specific IgG1 response when compared with WT mice. ELISpot analysis revealed increased numbers of TNP-specific IgG1-secreting splenic B cells and FACS analysis revealed increased CD4(+) T-cell IL-4 expression. Our results suggest that mCRAMP differentially regulates B- and T-cell function and implicate mCRAMP in the regulation of adaptive immune responses.     

CD4(+) T cell-mediated protection against a lethal outcome of systemic infection with vesicular stomatitis virus requires CD40 ligand expression, but not IFN-gamma or IL-4

To investigate the mechanism(s) whereby T cells protect against a lethal outcome of systemic infection with vesicular stomatitis virus, mice with targeted defects in genes central to T cell function were tested for resistance to i.v. infection with this virus. Our results show that mice lacking the capacity to secrete both IFN-gamma and perforin completely resisted disease. Similar results were obtained using IL-4 knockout mice, indicating that neither cell-mediated nor T(h)2-dependent effector systems were required. In contrast, mice deficient in expression of CD40 ligand were more susceptible than wild-type mice, and residual resistance in these mice was almost completely abrogated by depletion of CD8(+) T cells. In keeping with this, mice lacking both MHC class I and class II expression succumbed to the infection, whereas most class II-deficient mice normally survive. Adoptive transfer experiments using B cell- and T cell-deficient recipients revealed that no protection could be obtained in the absence of B cells, whereas treatment with virus-specific immune (IgG) serum controlled viral spreading to the central nervous system (CNS), but did not necessarily accomplish virus elimination. Taken together, these results underscore that B cells are essential in preventing early infection of the CNS, but T cells are required for long-term survival. CD4(+) T cells are most efficient in this context and the key function is to provide cognate help to B cells. However, if CD4(+) cell function is compromised, CD8(+) T cells become critical and may suffice for survival.     

B cells modulate T cells so as to favour T helper type 1 and CD8+ T-cell responses in the acute phase of Trypanosoma cruzi infection

In this study, we have evaluated the production of pro- and anti-inflammatory cytokines and the formation of central and effector memory T cells in mice lacking mature B cells (mu MT KO). The results show that Trypanosoma cruzi infection in C57Bl/6m mu MT KO mice is intensified in relation to control mice and this exacerbation is related to low levels of inflammatory cytokines produced during the acute infection and the lower numbers of central and effector memory CD4(+) and CD8(+) T cells generated during the acute phase of the infection. In addition, a marked reduction in the CD8(+) T-cell subpopulation was observed in mu MT KO infected mice. In agreement to this, the degree of tissue parasitism was increased in mu MT mice and the tissue inflammatory response was much less intense in the acute phase of the infection, consistent with a deficit in the generation of effector T cells. Flow cytometry analysis of the skeletal muscle inflammatory infiltrate showed a predominance of CD8(+) CD45Rb low in B-cell-sufficient C57Bl/6 mice, whereas the preponderant cell type in mu MT KO skeletal muscle inflammatory infiltrate was CD4(+) T cells. In addition, CD8(+) T cells found in skeletal muscle from mu MT KO infected mice were less activated than in control B-cell sufficient infected mice. These results suggest that B cells may participate in the generation of effector/memory T cells. In addition and more importantly, B cells were crucial in the maintenance of central and effector memory CD8(+) T cell, as well as the determination of the T cell cytokine functional pattern, and they may therefore account for critical aspects of the resistance to intracellular pathogens, such as T. cruzi.     

Echinococcus multilocularis proliferation in mice and respective parasite 14-3-3 gene expression is mainly controlled by an alphabeta CD4 T-cell-mediated immune response

The role of specific B lymphocytes and T-cell populations in the control of experimental Echinococus multilocularis infection was studied in micro MT, nude, T-cell receptor (TCR)-beta(-/-), major histocompatibility complex (MHC)-I(-/-) and MHC-II(-/-) mice. At 2 months postinfection, the parasite mass was more than 10 times higher in nude, TCR-beta(-/-) and MHC-II(-/-) mice than in infected C57BL/6 wild-type (WT) mice, and these T-cell-deficient mice started to die of the high parasite load at this time-point. In contrast, MHC-I(-/-) and micro MT mice exhibited parasite growth rates similar to those found in WT controls. These findings clearly point to the major role that CD4(+) alphabeta(+) T cells play in limiting the E. multilocularis proliferation, while CD8(+) T and B cells appeared to play a minor role in the control of parasite growth. In the absence of T cells, especially CD4(+) or alphabeta(+) T cells, the cellular immune response to infection was impaired, as documented by the lack of hepatic granuloma formation around the parasite and by a decreased splenocyte responsiveness to concanavalin A (Con A) and parasite antigen stimulation. Surprisingly, in T-cell-deficient mice, the ex vivo expression of interferon-gamma (IFN-gamma) and other inflammatory cytokines (except for interleukin-6) were increased in association with a high parasite load. Thus, the relative protection mediated by CD4(+) alphabeta(+) T cells against E. multilocularis infection seemed not be IFN-gamma dependent, but rather to rely on the effector's function of CD4(+) alphabeta(+) T cells. The local restriction of parasite germinal cell proliferation was reflected by a regulatory effect on the expression of 14-3-3 protein within the parasite tissue in T-cell-deficient mice. These results provide a strong indication that the CD4(+) alphabeta(+) T-cell-mediated immune response contributes to the control of the parasite growth and to the regulation of production of the parasite 14-3-3 protein in metacestode tissues.     

Viral neuraminidase treatment of dendritic cells enhances antigen-specific CD8(+) T cell proliferation, but does not account for the CD4(+) T cell independence of the CD8(+) T cell response during influenza virus infection

In vitro studies demonstrate that the increased alloreactive T cell response to dendritic cells (DC) that are treated with either live or inactivated influenza virus A/PR/8/34 is due to viral neuraminidase (NA) activity. Since virus-specific cytotoxic T lymphocytes (CTL) play an important role in immunity to heterologous influenza strains, we compared the activation of CD8(+) T cells by untreated and NA-treated DC. Increased CTL activity was induced by NA-treated DC both in vitro and in vivo. Since the generation of CTL in response to influenza virus infection does not require prior "activation" of DC by CD4(+) T cells (as is the case for many antigens), we asked whether NA activity contributed to this unconditional CD8(+) T cell response. This was not the case. Future studies will determine the factors that are responsible for the CD4(+) T-cell-independent influenza virus-specific CTL response.     

Role of ICOS versus CD28 in antiviral immunity

The costimulatory protein ICOS is inducibly expressed on activated T cells. Previous results have shown that ICOS(-/-) mice are defective in germinal center formation, antibody (Ab) production and class switch as well as Th1 and Th2 cytokine production in response to protein or parasite antigens. However, ICOS-Ig failed to block antiviral Ab responses. To date the immune response to viruses has not been examined in ICOS(-/-) mice. In this report we compared antiviral Ab responses to LCMV, VSV and influenza virus in ICOS(-/-) versus wild-type mice. Our results show that ICOS is important in the Ab response to all three viruses, with greater effects on primary as compared to secondary responses. Although ICOS(-/-) mice are impaired in some immune responses following influenza infection, the effects were less severe than for CD28(-/-) mice. There was no defect in initial influenza-specific CD8 T cell expansion in ICOS(-/-) mice or in cytotoxic effector function. However, ICOS was important in maintaining CD4 cytokine production and CD8 T cell numbers late in the primary response. Upon secondary infection, ICOS(-/-) mice show wild-type levels of influenza-specific CD8 T cells, whereas CD28(-/-) mice show greatly impaired secondary CD8 T cell expansion. Overall, our results show that ICOS plays a clear role in the primary response to viruses at the level of Ab production, germinal center formation and Th cytokine production, but has diminished effects following secondary viral challenge.