Lymphotoxin-α-Deficient Mice Make Delayed, But Effective, T and B Cell Responses to Influenza¹

Influenza type A viruses are cytolytic viruses (1) that cause acute respiratory infections and are a significant source of morbidity and mortality in human populations (2). Influenza infection is typically limited to respiratory epithelium (3) and is cleared by a combination of cellular and humoral immune mechanisms (4, 5). CD8 T cells clear infection primarily through their ability to kill virally infected cells (6), while CD4 T cells provide B cell help and promote the production of neutralizing Ab (7, 8). In normal individuals protection from an initial infection is mediated primarily by the cytolytic activity of activated CD8 T cells (9), while neutralizing Ab provides protection from secondary encounters with the same virus (10).

In addition to the cytotoxic activity of CD8 cells and the B cell helper activity of CD4 cells, both CD4 and CD8 T cells secrete a variety of cytokines that may interfere with viral replication and stimulate local immune mechanisms (11, 12, 13). One such cytokine, lymphotoxin-α (LTα)³ is a member of the TNF family of cytokines (14) and is active as either a soluble homotrimer (15) or in combination with LTβ as a membrane-bound heterotrimer (16). Soluble LTα can signal through the TNF receptor type 1 (TNFR1) (p55) as well as TNFR2 (p75), while membrane-bound LTαβ signals through the LTβR (17). Like TNF-α, LTα can trigger apoptosis, differentiation, proliferation, and the expression of a variety of inflammatory cytokines and chemokines depending on the responding cell type and the receptor that it binds (18). Although LTα is produced by a variety of lymphocytes (19), it is produced at high levels by activated type 1 CD4 (Th1) and CD8 (Tc1) T cells (20) and is thought to play an important role in the effector functions of these cells.

Another important activity of LTα is to induce the development of organized lymphoid tissues (21). LTα^−/− mice entirely lack lymph nodes and Peyer’s patches and have a disrupted splenic architecture (22). The pathways behind these defects have been studied in detail, and it is now known that loss of LTβ (23), LTβR (24), or NF-κB-inducing kinase (a molecule in the LTβR signaling pathway) (25) leads to defects in lymphoid organ development, the disruption of splenic organization, and the loss of follicular dendritic cells (FDCs) and germinal centers.

Given the absence of lymph nodes and Peyer’s patches in LTα^−/− mice and the lack of FDCs, organized B cell follicles and germinal centers in the spleens of these mice, it is not surprising that LTα^−/− mice are defective in isotype switching (26) and affinity maturation (27) and generally make poor Ab responses (28). Furthermore, given that LTα facilitates dendritic cell (DC) maturation (29) and induces the expression of chemokines that coordinate lymphocyte migration (30), it is also not surprising that mice defective in the LT signaling pathway make defective cellular immune responses. For example, mice that have a loss of function mutation in NF-κB-inducing kinase (aly/aly mice) do not make detectable primary CTL responses to vesicular stomatitis virus or lymphocytic choriomeningitis virus (LCMV) and make a reduced CTL response to vaccinia virus (31, 32). Similarly, LTα^−/− mice have severely impaired cellular immune responses to LCMV (33) and cannot clear HSV (34). However, LTα^−/− mice are not completely unable to mount anti-viral responses, as LTα^−/− mice can clear murine gammaherpes virus 68 (MHV-68) infection, albeit with delayed kinetics (35). Likewise, LTα-TNF^−/− mice can clear vaccinia virus and LCMV after a significant delay (36). Together, these results demonstrate that mice lacking an intact LT signaling pathway have clear defects in cellular immunity to viral infections. However, it remains unclear whether cellular immunity is inefficient in these mice due to lack of CTL priming, an inability to expand Ag-specific T cells, or alterations in the differentiation of Ag-specific T cells to functional effector cells.

To determine the basis for the defective cellular immune responses observed in LTα^−/− mice, we compared LTα^−/− and normal wild type (WT) mice for their ability to generate Ag-specific CD8 T cells to influenza and to clear virus. We found that although LTα^−/− mice were somewhat more susceptible to infection with high doses of influenza virus, they were competent to generate influenza-specific CD8 T cells that produced IFN-γ and exhibited killing activity. Furthermore, influenza-infected LTα^−/− mice could generate B cells that produced isotype-switched, influenza-specific Abs. However, unlike the immune response in WT mice, the immune response to influenza in LTα^−/− mice was delayed by several days. Thus, when LTα^−/− mice were infected with larger doses of virus, they succumbed to infection before specific adaptive immunity could be generated. Together, these results suggest that neither the LT signaling pathway nor the presence of organized lymphoid tissues is absolutely required for T or B cell effector generation or function. However, an intact LT signaling pathway and the presence of organized lymphoid organs do facilitate the rapid induction of immune responses.

C57BL/6 (WT mice) and C57BL/6.129Lta^tm1Dch (LTα^−/− mice) were obtained from The Jackson Laboratory (Bar Harbor, ME). Both strains were bred and maintained in the Animal Breeding Facility at Trudeau Institute. All procedures using animals were approved by the Trudeau Institute institutional animal care and use committee and were conducted according to the principles outlined by the National Research Council.

Mice were infected intranasally with 100 egg infectious units (EIU) of influenza A/PR8/34 in 100 μl unless otherwise indicated. Viral titers in the lungs of infected mice were quantified in embryonated eggs. Briefly, lungs were homogenized in 2 ml of PBS, and 500 μl of this stock was used to make 10-fold serial dilutions. One hundred microliters of each dilution was inoculated into each of three eggs. Allantoic fluid was harvested from inoculated eggs 4 days later, and infected eggs were scored by hemagglutination of chicken RBC. The viral end-point titer was defined as the highest dilution in which two or more eggs positively scored in the hemagglutination assay.

Mice were sacrificed at the indicated time points, and the spleens and lungs (without external bronchi and trachea) were removed and mechanically disrupted by passage through a wire mesh. Live leukocytes were obtained by density gradient centrifugation using Lympholyte-Poly as a cushion (Cedarlane, Hornby, Canada). Cells were incubated in 3% FCS in PBS containing 10 μg/ml 2.4G2 to block Fc receptor binding, followed by the addition of fluorochrome-conjugated Abs or MHC class 1 tetramers. All fluorochrome-conjugated Abs were obtained from BD PharMingen (San Diego, CA). The MHC class I tetramer H-2D^b containing nucleoprotein (NP)_366–374 peptide used to identify influenza-specific T cells was generated by the Trudeau Institute Molecular Biology Core Facility. Flow cytometry was performed on a dual laser FACSCalibur available through the Flow Cytometry Core Facility at Trudeau Institute.

Live leukocytes obtained from the spleens and lungs of infected mice were cultured at 5 × 10⁶/ml in 200 μl of complete RPMI containing 10 μg/ml brefeldin A and 10 U/ml IL-2. Cells were stimulated with influenza NP_366–374 peptide or influenza polymerase-2 (PA)_224–233 peptide at 1 μg/ml or with 5 ng/ml PMA and 650 ng/ml of A23187 for 5 h. Simulated cells were then washed, blocked with 2.4G2, and probed with anti-CD8 before fixation in 4% paraformaldehyde in PBS. Fixed cells were washed with 0.1% Triton X-100/3% FCS in PBS and probed with fluorochrome-labeled anti-IFN-γ in the same buffer.

EL4 cells were cultured with no peptide, influenza NP_366–374 peptide or influenza PA_224–233 peptide at 1 μg/ml and were labeled overnight with ⁵¹Cr. After labeling, the EL4 targets were washed and cultured at 10⁴ cells/well. Live leukocytes from the lungs of infected mice were added to the labeled EL4 cells at E:T cell ratios between 0.1:1 and 100:1 (in triplicate), and the cells were cultured for 5 h. Supernatants were then collected, and the released radioactivity was counted in a gamma counter. Spontaneous ⁵¹Cr release was determined from labeled EL4 cells in the absence of effectors, and maximum possible release was determined from EL4 cells cultured in 1% Triton X-100.

Blood was obtained from euthanized mice by severing the renal artery and pipetting into a 1.5-ml tube. After clotting for 2 h at 37°C, the precipitate was pelleted in a microcentrifuge, and the serum was removed. Influenza-specific ELISAs were performed by coating plates with purified virus at 1 μg/ml. Serum samples were diluted in 3-fold serial dilutions in PBS with 10 mg/ml BSA and 0.1% Tween 20 before incubation on coated plates. Bound Ig was detected with HRP-conjugated goat anti-mouse IgM, goat anti-mouse IgG, or goat anti-mouse IgA.

To determine whether the lack of lymph nodes and the disorganized splenic structure in LTα^−/− mice conferred increased susceptibility to influenza infection, we intranasally inoculated groups of LTα^−/− and WT mice with 50, 240, 1200, or 6000 EIU of influenza A/PR8/34 (PR8) and monitored the survival and morbidity (weight) of the infected mice over the next 30 days. As shown in Fig. 1, both WT and LTα^−/− mice succumbed to 6000 EIU of PR8 with similar kinetics. However, there were small, but consistent, differences in the survival of LTα^−/− and WT mice at the intermediate doses of virus. For example, while all LTα^−/− mice succumbed to infection with 1200 EIU, only three of five of the WT mice died at this dose of virus. Furthermore, while two of five LTα^−/− mice succumbed to 240 EIU, none of the WT mice died. All mice in both groups survived 50 EIU of influenza. The survival of the LTα^−/− mice infected with lower doses of virus was not due to a nonproductive infection, as all mice in each group exhibited signs of morbidity, including temporary weight loss (Fig. 1), hunched posture, and ruffled fur. Together, these results suggest that although the LD₅₀ of PR8 for LTα^−/− mice is ∼5-fold lower than that for WT mice, LTα^−/− mice do make an immune response that is protective against lower doses of virus.

To determine whether the increased susceptibility of LTα^−/− mice to influenza was due to an inability to mount a protective CD8 T cell response, we examined CD8 T cell activation in the spleen and lungs of flu-infected WT and LTα^−/− mice. Groups of WT and LTα^−/− mice were infected with a dose of PR8 sufficient to induce infection, but not death, in either group of mice (100 EIU). Five mice from each group were sacrificed at 0, 6, 9, 12, 15, and 21 days postinfection, and the cells from lung and spleen were pooled. The accumulation of activated influenza NP_366–374 specific CD8 T cells was determined by flow cytometry using an H2-D^bNP_366–374 tetramer. As shown in Fig. 2,A, influenza-specific CD8 T cells first appeared in the spleens of WT mice on day 9, increased in frequency on day 12, and declined on days 15 and 21. The influenza-specific CD8 T cells appeared to have an activated phenotype, as assessed by reduced expression of CD62L (Fig. 2). Unlike what we observed in the spleens of flu-infected WT animals, activated influenza-specific CD8 T cells did not appear in the spleens of LTα^−/− mice until day 12. However, despite the delayed appearance of these cells, the frequency of activated, flu-specific CD8 T cells on days 12, 15, and 21 was increased in LTα^−/− mice compared with that in WT mice. When the total number of activated, Ag-specific cells in the spleen was calculated, we found that a similar number of activated, flu-specific CD8 T cells accumulated in the spleens of LTα^−/− and WT mice, even though the appearance of these cells was delayed by 3 days in LTα^−/− mice (Fig. 2 B).

Similar results were observed in the lungs of infected mice. Activated, influenza-specific CD8 T cells appeared in the lungs of WT mice on day 9, increased on day 12, and decreased on days 15and 21 (Fig. 2,C). Again, the appearance of activated Ag-specific CD8 T cells in the lungs of LTα^−/− mice was delayed until day 12. However, unlike what we observed in the spleen, the frequency of activated, influenza-specific CD8 T cells was lower in the lungs of LTα^−/− mice than in the lungs of WT mice at all time points (Fig. 2,C). Despite the dramatic reduction in the frequency of activated NP-specific CD8 T cells in the lungs of LTα^−/− mice, however, the total number of these cells was actually higher in the lungs LTα^−/− mice than in WT mice at later time points (Fig. 2 D). This is due to the substantially higher inflammatory response that occurs in the LTα^−/− lung. Although we recovered similar numbers of leukocytes from the lungs of uninfected WT and LTα^−/− mice (an average of 1 × 10⁶ cells/animal), the numbers of leukocytes recovered from the lungs of influenza-infected WT mice peaked on day 9 (an average of 4 × 10⁶ cells/animal), while the numbers of leukocytes recovered from lungs of influenza-infected LTα^−/− mice peaked between days 12 and 15 (an average of 13 × 10⁶ cells/animal).

To determine whether the tetramer-binding CD8 T cells observed in LTα^−/− mice could make IFN-γ upon restimulation, we isolated cells from the lungs and spleens of influenza-infected WT and LTα^−/− mice at various times after infection and cultured the cells for 5 h with influenza NP_366–374 peptide, influenza PA_224–233 peptide, or no peptide. The stimulated cells were subsequently analyzed for IFN-γ production by intracellular cytokine staining. IFN-γ was produced by CD8 T cells from the lungs of both WT and LTα^−/− mice in response to NP peptide (Fig. 3,A), as well as PA peptide (Fig. 3,B), but not in the absence of peptide (Fig. 3,C). While NP- and PA-specific IFN-γ-producing CD8 T cells appeared by day 9 in WT mice, their appearance was delayed in LTα^−/− mice until day 12. Similar results were observed in the spleen (not shown). When the total number of Ag-specific IFN-γ-producing CD8 T cells was determined, we found that these cells were present at much lower levels in the lungs and spleens of LTα^−/− mice compared with WT mice on day 9 postinfection. However, by days 15–21 the number of Ag-specific, IFN-γ-producing CD8 T cells in the lungs and spleens of LTα^−/− mice exceeded that in WT mice (Fig. 3 D).

To next determine whether the LTα^−/− mice could generate effective CTLs, we isolated cells from the bronchial alveolar lavage (BAL) and the lung tissue of influenza-infected WT and LTα^−/− mice at various times after infection and cultured the cells for 5 h with ⁵¹Cr-labeled EL4 target cells in the presence of influenza NP_366–374 peptide, influenza PA_224–233 peptide, or no peptide. As shown in Fig. 4, robust CTL activity in response to both NP and PA peptides was observed in WT cells on day 9. WT CTL activity was reduced on day 12 (particularly to the PA peptide) and was absent by day 15. No NP-specific CTL activity was observed in LTα^−/− cells on day 9, but, by days 12–15, NP-specific CTL activity could be observed in BAL and lung tissue isolated from LTα^−/− mice. However, a higher E:T cell ratio was needed to observe significant killing of the targets. Furthermore, the CTL activity directed toward the PA peptide in LTα^−/− mice was very weak and was observed only on day 15. Together these data indicate that influenza-specific CTLs can be generated in LTα^−/− mice. However, the frequency of these cells is reduced due to the increased lung inflammatory response in the infected LTα^−/− mice.

Although CTL activity is thought to be the principal form of immunity responsible for resistance to influenza in a primary response (9), the production of high affinity isotype switched Abs is thought to provide the major source of protection against secondary infection with homotypic virus (10). Since high affinity isotype-switched B cells are generated in germinal centers (37), we first examined whether the production of germinal center B cells was altered in virus-infected LTα^−/− mice. To determine the frequency of germinal center B cells at various times after infection we isolated cells from the lungs and spleens of influenza-infected WT and LTα^−/− mice and analyzed the expression of Fas and peanut agglutinin (PNA) on CD19⁺ B cells by flow cytometry. As expected, we found that the frequency (Fig. 5,A) and total number (Fig. 5,B) of Fas^highPNA^highCD19⁺ germinal center B cells increased rapidly in the spleens of influenza-infected WT mice. Additionally, in agreement with numerous reports demonstrating the inability of LTα^−/− mice to make germinal centers in the spleen, we observed that the frequency (Fig. 5,A) and total number (Fig. 5,B) of germinal center B cells remained near background levels in the spleens of influenza-infected LTα^−/− mice. Surprisingly, however, we observed B cells with a germinal center phenotype in the lungs of both WT and LTα^−/− mice after influenza infection (Fig. 5,C). The frequency of germinal center B cells in the lung was slightly higher in LTα^−/− mice than in WT mice, particularly on days 12 and 15 postinfection (Fig. 5,C), and the absolute number of these cells was much higher in LTα^−/− mice at these times (Fig. 5,D). Although we did observe a transient population of CD19⁺PNA⁺Fas^int cells on day 6 in both WT and LTα^−/− mice (Fig. 5,C, circled population), these cells did not have the classic phenotype of germinal center cells (T. D. Randall, unpublished observations); thus, we did not include them in our calculation of the absolute number of germinal center B cells on day 6 (Fig. 5 D).

The appearance of B cells with a germinal center phenotype in the lungs of influenza-infected LTα^−/− mice suggested that the production of isotype-switched B cells and Abs may also be intact in these mice. To test this hypothesis, cells were isolated from the lungs of WT and LTα^−/− mice 21 days postinfection, and the frequency of CD19⁺syndecan⁺ plasma cells that expressed IgM, IgG1, IgG2a/b, IgG3, or IgA was determined. As shown in Fig. 6,A, similar frequencies of IgM⁺, IgG1⁺, IgG2a/b⁺, and IgA⁺ plasma cells were found in WT lungs. In contrast, in the lungs of LTα^−/− mice the frequencies of IgM⁺ and IgA⁺ plasma cells were increased relative to the frequencies of the IgG isotypes. These differences were maintained when the absolute numbers of IgM- and IgA-expressing plasma cells present in the lungs of WT and LTα^−/− mice were calculated (Fig. 6,B). Finally, the frequencies of IgG1⁺-, IgG2a/b⁺-, and IgG3⁺-expressing B cells that were not plasma cells (isotype switched and syndecan⁻) were higher in WT mice than in LTα^−/− mice. However, the frequency of isotype-switched IgA⁺syndecan⁻ cells was similar in WT and LTα^−/− mice (Fig. 6 A).

Although the FACS analysis suggested that B cells from influenza-infected LTα^−/− mice were differentiating into class-switched, Ab-secreting cells, this analysis did not distinguish between Ag-specific and nonspecific B cells. Therefore, to confirm that influenza-specific, isotype-switched Ab was being produced, we performed ELISAs to determine the relative concentration of influenza-specific Ab in the serum of animals infected for various times. As shown in Fig. 6,C, we found that the concentration of influenza-specific IgM rapidly increased in WT mice, peaked between days 9 and 12, and then gradually declined on days 15 and 21. Influenza-specific IgM appeared with slightly delayed kinetics in the serum of LTα^−/− mice, but peaked between days 12 and 21 at a level higher than that observed in WT mice. Although influenza-specific IgG was barely detectable in WT mice on day 6, it rapidly climbed on days 9 and 12 and continued to increase on days 15 and 21 (Fig. 6,C). Influenza-specific IgG was not detected until day 9 in LTα^−/− mice; however, by day 15 the titer of flu-specific IgG climbed >1000-fold. Despite the large burst of influenza-specific IgG produced by the LTα^−/− mice, the titer of this Ab remained ∼10-fold less than that observed in WT mice. Finally, although the number of IgA-producing plasma cells was increased in the lung tissue of LTα^−/− mice compared with WT mice (Fig. 6,B), the titers of influenza-specific IgA were similar in the serum of these animals (Fig. 6 C).

The appearance of CTLs and influenza-specific Abs in LTα^−/− mice coupled with the survival of these mice at low doses of virus suggested that adaptive immune mechanisms could clear virus in these mice. To test this directly, we assayed viral burden in the lungs of influenza-infected WT and LTα^−/− mice. As shown in Fig. 7, the viral burden in WT and LTα^−/− was similarly high on day 6, a time when there was little evidence of adaptive immune responses in either strain of mice. While the viral burden rapidly declined on days 8 and 10 in WT mice, the viral burden remained high on these days in LTα^−/− mice. Finally, however, virus was cleared in both WT and LTα^−/− mice by days 15–20.

The data presented here clearly demonstrate that LTα^−/− mice can generate Ag-specific CD8 T cells in response to viral infection. Furthermore, the influenza-specific T cells from LTα^−/− mice expanded (Fig. 2) and differentiated into competent effector cells that could make IFN-γ upon restimulation (Fig. 3), could kill peptide-loaded target cells (Fig. 4), and could clear virus in vivo (Fig. 7). In fact, the only major difference between the CD8 T cell response of the LTα^−/− mice and WT mice was that there was a 2- to 3-day delay in the appearance of Ag-specific CD8 T cells in LTα^−/− mice ( Figs. 2–4). These results are in apparent contrast with other studies (31, 32, 33, 34, 36) in which mice deficient in the LT signaling pathway were unable to make normal CTL T cell responses to viruses such as LCMV, HSV, and vesicular stomatitis virus. However, the conclusions from some of these earlier studies were based on data from early time points after virus infection (6–8 days); thus, it is likely that CTL responses were not observed in LTα^−/− (33, 34), aly/aly (31, 32), or TNF^−/−LTα^−/− (36) mice because CTLs had not yet developed at the time points that were analyzed. We would have arrived at similar conclusions had we only looked for influenza-specific CTL responses on days 9–10.

The delayed accumulation of virus-specific CD8 T cells is not the only reason that reduced CTL activity was observed in previous studies using LT signaling-deficient mice. Since CTL activity is a function of CTL frequency, rather than total CTL number, CTL activity can be reduced (as we observed in Fig. 4, day 12) even though the total number of virus-specific cells may actually be similar between WT and LTα^−/− mice (as we observed in Fig. 2, day 12). This difference between frequency and total number of Ag-specific CD8 T cells is due to the more severe inflammatory response that occurs in influenza-infected LTα^−/− mice. This increased inflammatory response is most likely a result of prolonged virus production in the LTα^−/− mice (see Fig. 7).

Other studies also reported that IFN-γ production by virus-specific CD8 T cells is reduced in mice defective in LT signaling (34, 35, 38). Again, this could be an issue of timing. We found that IFN-γ production by virus-specific CD8 T cells was delayed in LTα^−/− mice (Fig. 3), but this delay corresponded with the delayed appearance of virus-specific CD8 T cells (Fig. 2). Furthermore, although one study demonstrated that similar numbers of Ag-specific CD8 T cells could be generated in WT and LTα^−/− mice after infection or immunization, those cells exhibited reduced killing activity and were unable to produce IFN-γ (34). Unlike our studies, however, in which we examined T cell function immediately ex vivo, that study expanded T cells in vitro for 5 days before assay. Thus, it is unclear whether the observed defects in the in vitro cultured LTα^−/− CD8 T cells are representative of how these cells behave in vivo.

Although we feel that timing is one critical difference between our results and those of others, the type of infection or its location within the body may also significantly impact how LTα^−/− mice respond to the infection. For example, it is reported that LTα^−/− mice make poor CD8 T cells responses to LCMV (31, 32, 36, 38), a systemic virus, and HSV (34), a neurotropic virus, while we found that LTα^−/− mice could generate CD8-mediated immunity to influenza, a respiratory virus. Interestingly, our data are most consistent with the reported resistance of LTα^−/− mice to another respiratory virus, MHV-68 (35). In that study CTL activity was intact, IFN-γ was produced (although at lower levels), and virus was cleared with delayed kinetics from the lungs of MHV-68-infected LTα^−/− mice. This raises the possibility that lymphotoxin and/or lymph nodes may not be absolutely required to generate immunity to respiratory viruses, but may be necessary to generate CD8 T cell responses to viruses with other tropisms.

In previous studies we found that nasal-associated lymphoid tissue (NALT) is present in LTα^−/− mice (39), even though these mice are completely devoid of lymph nodes and Peyer’s patches. These studies also showed that the presence of NALT in LTα^−/− mice is not sufficient to generate Ag-specific T or B cells or to clear influenza virus from the lungs by day 10 (39). This is entirely consistent with our current results, which show that the generation of influenza-specific T cells ( Figs. 2–4), the production of influenza-specific IgG (Fig. 6), and the clearance of virus (Fig. 7) are all severely impaired before day 10 in LTα^−/− mice. However, our current results also demonstrate that immune responses are eventually generated in LTα^−/− mice by day 12. These results suggest that the spleen and/or NALT of LTα^−/− mice contribute to immune responses even though the lymphoid areas of these tissues are not organized in a normal fashion. However, we cannot distinguish the relative importance of the spleen or NALT in the generation of immunity to influenza in LTα^−/− mice at this time.

Initial experiments with LTα^−/− mice immunized with experimental Ags such as SRBC or haptenated proteins demonstrated that the formation of germinal centers and the production of isotype-switched Abs to T-dependent Ags were severely impaired in these mice (22, 27, 28, 40). The defective class-switched Ab response was attributed primarily to the inability of LTα^−/− mice to organize either primary or secondary lymphoid follicles in the spleen, rather than to the lack of lymph nodes, since the transfer of WT cells to LTα^−/− mice enabled the mice to form germinal centers and produce isotype-switched Abs (26, 40, 41). Interestingly, there are a few reports that LTα^−/− B cells are able to isotype-switch in the apparent absence of germinal center formation, particularly in response to viral infections (31, 35). In agreement with these reports we found that Ab production in response to influenza infection is not impaired per se, as the total numbers of Ab-secreting cells in the lungs of LTα^−/− mice is actually greater than that in WT mice (Fig. 6). However, while the production of IgM and IgA was increased in LTα^−/− mice (Fig. 6, B and C), the production of IgG was decreased (Fig. 6 C), suggesting that switching to some, but not all, isotypes was impaired. This might be due to the loss of LT signaling in B cells or to the lack of FDCs and an inability to form germinal centers in the spleen.

As previously reported we found no evidence for germinal center formation in the spleens of LTα^−/− mice (Fig. 5,A), suggesting that isotype switching in response to virus infection must have occurred independently of splenic germinal centers. Interestingly, we observed B cells with a germinal center phenotype in the lungs of LTα^−/− as well as WT mice (Fig. 5 C). Although the lung is not a site in which germinal centers are typically believed to form, there are reports that organized lymphoid tissue and even germinal centers can be found in the lung after immunization (42). Unlike bronchus-associated lymphoid tissue, which is constitutively present in some species, these organized lymphoid areas are not found in naive mice and must be induced by infection or inflammation (43, 44, 45). Once formed, however, they may serve as sites that can initiate primary immune responses (46, 47). If germinal centers are formed in the lungs of LTα^−/− mice after infection, this would suggest that the mechanism controlling their formation in the lung is LT independent and must be fundamentally different from the LT-dependent mechanism that controls germinal center formation in lymphoid tissues such as the spleen. These questions are currently being addressed.

Since LTα^−/− T cells are not thought to have intrinsic defects in activation (22, 38), the observed delay in CD8 T cell responses in LTα^−/− mice is most likely due to a defect in Ag presentation. This could be manifest at the level of DC activation, DC migration from the lung to the spleen, or the ability of T cells to find APCs. Any or all of these mechanisms could be operating in LTα^−/− mice. For example, LT can induce DC maturation and activation. Furthermore, the ability of T cells to interact with DCs is impaired in LTα^−/− mice at several levels. First, there are no lymph nodes in LTα^−/− mice (22, 28), the traditional site of T cell priming. Second, the expression of the chemokines B lymphocyte chemoattractant, secondary lymphoid tissue chemokine, and EBV-induced molecule-1 ligand chemokine is dependent on LTα (30); thus, the ability of lymphocytes and DCs to home from peripheral tissues to secondary lymphoid organs is disrupted. This is dramatically demonstrated by the paucity of DCs in the spleens of LTα^−/− mice (29), even though the number of DCs in the lungs of uninfected LTα^−/− mice is normal (T. D. Randall, unpublished observations) or elevated (29). Together, these defects could easily explain the observed delay in T cell activation observed in the LTα^−/− mice.

In summary, we show that LTα^−/− mice are capable of mediating all aspects of a primary cellular and humoral anti-influenza response. Although the generation of influenza-specific CD8 T cells was delayed, once formed, those cells could kill Ag-loaded target cells, produce IFN-γ, and clear virus in vivo. Furthermore, LTα^−/− mice produced increased levels of influenza-specific IgM, normal levels of IgA, and only slightly decreased levels of IgG, demonstrating that B cell activity was largely intact. Together, these results suggest that the presence of lymph nodes and an organized spleen are not absolutely required for the generation of immunity, but that they facilitate the rapid induction of immune responses.