CD1d Mediates T-Cell-Dependent Resistance to Secondary Infection with Encephalomyocarditis Virus (EMCV) In Vitro and Immune Response to EMCV Infection In Vivo

Animals.

Mice deficient in both CD1d genes were prepared as described previously (22). CD1d-KO mice were further backcrossed for a total of 12 generations to the C57BL/6J background. Age-matched male (4- to 5-week-old) wild-type C57BL/6 (Jackson Labs. or Taconic) controls were used.

Cells and cell lines.

Splenocyte cultures were prepared from spleens that were minced, strained through a 70-μM filter, treated with red cell lysis buffer (Sigma), and seeded at 10⁶ cells/ml in 24-well plates. Cultures were grown in complete RPMI, with 10% fetal calf serum and IL-2 (20 IU/ml; NBRMP, NCI) added at initiation. The mouse L929 cell line was obtained from ATCC and was grown in Dulbecco's modified Eagle medium (DMEM) with 10% fetal calf serum. Mixed L929/splenocyte cultures were incubated in a RPMI-DMEM mixture (1:1 ratio) during the first 24 h of passage and then transferred to complete RPMI.

EMCV virus preparation, virus titration, and infection in vitro and in vivo.

EMCV virus was grown and titrated on L929 cultures. Viral stocks were prepared on day 1 postinfection (later time points resulted in a lower titer), cleared of cell debris, filtered (0.45 μM), tested in plaque-forming assay, and stored at −80°C. For plaque-forming assay, 24- to 48-h-old L929 cultures were treated with various (usually 10⁻³ to 10⁻⁶) viral dilutions for 1 h in phosphate-buffered saline, covered by complete DMEM with 1% agarose, and upon solidification incubated at 37°C. Plaques were counted at 3 days by staining viable cells with 1% Neutral Red. The same assay was used to determine the viral titer in infected splenocyte culture supernatants. For the infection of fresh splenocytes in vitro, 1 × 10⁶ cells per well in 24-well plates were infected with EMCV-D (multiplicity of infection [MOI] = 0.1 to 3 PFU per cell) as previously described (18, 19, 22).

For the in vivo experiments, test animals (4 to 6 weeks) were infected intraperitoneally with 500 PFU of diabetogenic EMCV strain (kindly provided by N. Bigley, Dayton, OH) as described previously (22) and observed for 2 to 7 days. Mice were bled for systemic cytokine and virus titer measurements as noted above, and their spleen cell populations were analyzed via fluorescence-activated cell sorting (FACS).

Both in vitro and in vivo experiments were repeated at least three times in every series, with the exception of those involving anti-mouse CD1d antibody, which were performed twice. The soluble CD1d antibody 1B1 and isotype control (BD-Biosciences, La Jolla, CA) were used at a concentration of 50 μg/ml. Representative data from individual replicate wells or animals or means (± standard deviation [SD]) for each experimental group are shown, and Student's t test was used as indicated.

Measurement of cytokine expression in vitro and in vivo.

IFN-α and IFN-γ levels in splenocyte cultures and sera of experimentally infected animals were determined by standard enzyme-linked immunosorbent assay using commercial kits (R & D, Minneapolis, MN, and PBL Biomedical Laboratories, Piscataway, NJ, respectively) according to manufacturers' recommendations. Culture medium samples or serum samples were taken at different time points and if not tested immediately were stored at −80°C. Limit of detection in splenocyte supernatants, which were assayed undiluted, was 1 pg/ml for IFN-γ and 5 pg/ml for IFN-α. Animal serum samples were assayed at several dilutions. Results were individual representative or means (± SD).

FACS analysis.

Splenocytes were stained with different combinations of fluorescein isothiocyanate-conjugated anti-mouse CD69, phycoerythrin (PE)-conjugated anti-mouse NK1.1, CD8, and CD19, and CyChrome (Cy7)-conjugated anti-mouse CD3 (all reagents from BD-Pharmingen, La Jolla, CA). Matched isotype antibodies from rat and hamster were used as controls. Cells were stained, washed, and resuspended in Hanks balanced salt solution-0.1% mouse serum buffer containing 0.1% sodium azide and analyzed by using an FC500 flow cytometer (Coulter). Gating was on viable leukocytes. Means (± SD) and/or individual representative results were shown or tabulated.

FACS sorting.

Splenocytes were stained, washed, resuspended in sterile medium containing 0.1% mouse serum, and sorted by using a MoFlo flow cytometer (Cytomation, Boulder, CO). Antimouse monoclonal antibodies (MAb) used were PE-conjugated anti-DC 33D1 (61) and CD11c or NK1.1 and Ly49b (“NK/NKT”) (BD-Pharmingen). Gating was done on viable leukocytes using PE-labeled isotype control.

Primary infection of splenocyte cultures with EMCV.

Previous results have shown that optimal resistance to EMCV infection (21, 22), as well as that by certain other pathogens, depends upon CD1d-restricted T cells. Therefore, we further investigated mechanisms of resistance to EMCV infection in vitro. Primary infection of mouse splenocytes with EMCV resulted in a rapid decrease of viral titer in infected cultures (Fig. 1A), as previously reported (19, 42). This was inversely correlated with increased levels of IFN-γ production, which in turn was viral dose dependent (Fig. 1B). Control pharmacological and mitogenic stimuli induced more-potent IFN-γ release. Both lipopolysaccharide (LPS) and poly(I:C) induced robust IFN-γ levels on days 1 and 2 from WT (Fig. 1B) and CD1-KO (not shown) splenocytes. iNKT-cell-specific synthetic lipid antigen α-GalCer stimulated delayed IFN-γ specifically only in WT cultures, which reached very high levels comparable to those of LPS and poly(I:C) by day 4 (Fig. 1B).

Noticeably, a higher level of virus titer was maintained through several days in splenocytes from CD1d-KO mice at both MOI shown (Fig. 1A). This difference, however, was not reflected in the levels of NK cells, B cells, or T cells expressing activation marker CD69 after infection (Fig. 1C; Table 1), nor in IFN-γ release, which did not differ between WT (Fig. 1A) and CD1d-KO mice (not shown). In all cases, robust day-1 NK-cell (NK1.1⁺ CD3⁻) and B-cell (CD19⁺) activation in response to EMCV infection was consistently observed, coupled with very modest but reproducible T-cell (CD3⁺) activation (Fig. 1C; Table 1). Both T-cell and B-cell activation declined with time, starting with day 2, while NK-cell response consistently dropped on day 2 but then rose somewhat by day 3 (Table 1). The late NK-cell activation response at least partly reflected CD69 expression by NK cells in mock-infected cultures, which increased progressively, although staying below levels in virus-infected cultures. Elevated levels of NK-cell CD69 expression in virus-infected and control cultures (Table 1) also reflected generally increasing IFN-γ levels (Fig. 1B).

Increased susceptibility of CD1d-KO splenocytes to reinfection with EMCV.

It was clear that EMCV infection of C57BL/6 mouse splenocytes in vitro is not productive, and it is likely that the number of primary splenocyte cells susceptible to productive EMCV infection in such cultures is relatively low. However, it is also possible that infection may be partially aborted by the immediate immune response, which could be related to NK-cell activation. In either case, the effects observed during the previous experiments using a single infection cycle in vitro may less-fully reflect infection defense mechanisms existing in vivo. Moreover, it has been shown that resolution of EMCV infection in vivo is substantially delayed in CD1d-KO mice (21, 22). To mimic the situation that exists during continuous viral infection, we first subjected EMCV-treated splenocyte cultures to reinfection with EMCV 4 to 6 days after the initial infection. This resulted in decreased EMCV recovery in the wild-type cultures relative to that in CD1d-KO splenocytes (Fig. 2A). Inhibition of viral replication was reflected in differential IFN-γ production in these cultures. Wild-type cultures exhibited increased production of IFN-γ in response to reinfection with EMCV, with hardly any additional IFN-γ production observed in CD1d-KO cultures (Fig. 2B).

As shown in Fig. 2C, before reinfection on day 5, both wild-type and CD1d-KO cultures exhibit considerable NK-cell activation, substantial T-cell activation (including both CD4⁺ and CD8⁺), and little if any activation of B cells (absolute B-cell numbers in the cultures declined as expected due to lack of B-cell growth factor, whereas both NK cells and T cells persisted beyond 7 days in the presence of IL-2). Reinfection led to additional stimulation of NK cells in both wild-type and CD1d-KO cultures. Interestingly, however, stronger additional T-cell stimulation was observed in wild-type cultures, especially among CD8⁺ T cells (Fig. 2C and D). In fact, minimal stimulation was seen in T cells from CD1d-KO mice, whereas ∼50% of wild-type cells were activated (and ∼25% of total CD8⁺ cells). No additional activation of B cells was seen in either type of culture (Fig. 2C and D).

Increased susceptibility of CD1d-KO splenocyte cultures to infection with EMCV in presence of susceptible L929 cells.

Another system which may provide for an additional round of infection with EMCV in vitro is coculture of splenocytes with susceptible L929 cells. L929 is a mouse fibroblast cell line capable of rapid productive infection with EMCV. Thus, splenocytes infected with EMCV in the presence of L929 cells will be exposed to multiple rounds of virus infection—by the original virus stock and by virus released from EMCV-infected L929 cells. L929 cells are lysed by EMCV after overnight incubation; therefore, one may assume that such a system will provide for no more than one or two additional viral infectious rounds.

Not surprisingly, wild-type mouse splenocytes responded strongly to the EMCV infection in coculture with L929 cells, while coincubation with mock-infected L929 did not have any effect. In Fig. 3 it can be seen that the major subsets of immune cells (mostly B but also T cells) were rapidly and relatively strongly activated by EMCV infection in the presence of L929 cells beyond the activation observed during EMCV infection of pure splenocyte cultures. This was especially true for activation of B cells, which, as shown in Fig. 1C and 2C, served as an early marker of primary splenocyte EMCV infection in vitro. A similar phenomenon has been reported for in vitro splenocyte infection by another unrelated virus, murine herpesvirus 68 (52).

Titration of EMCV showed that although no clear difference in viral replication between L929-mixed CD1d-KO and wild-type cultures could be observed early in the infection, markedly higher levels of EMCV were maintained at the later time points in L929-mixed CD1d-deficient cultures (Fig. 4A). However, no significant differences in IFN-γ production between knockout- and wild-type-derived splenocytes were detected after primary infection of mixed L929/splenocyte cultures (not shown).

Immune cell activation observed at day 1 after initiation of EMCV infection in such mixed cultures also showed no difference between wild-type and CD1d-KO cultures (not shown). However, on day 2 the T-cell response in CD1d-KO culture was somewhat less, while B-cell activation was markedly greater (Fig. 4B and C), unlike in primary infection, potentially reflecting subsequently higher levels of viral persistence (Fig. 4A).

More marked differences in the T-cell activation profile of wild-type and CD1d-KO splenocytes emerged when these L929-mixed cultures were subjected to an additional round of infection with exogenous EMCV, which was also seen with both major T-cell populations. Differences in CD8⁺ cells activation in wild-type and CD1d-KO splenocytes were most pronounced (Fig. 5A and B), although CD4⁺ cells were also differently activated (not shown). This was closely matched by levels of IFN-γ production, which were again markedly higher in wild-type-derived cultures after an additional round of infection, while decreasing in CD1d-KO-derived cultures (Fig. 5C). Even more-pronounced differences after reinfection of wild-type and CD1d-deficient L929/splenocyte mixed cultures were observed in relation to IFN-α production (Fig. 5D). Therefore, stronger activation of T cells and higher levels of cytokine production in these reinfected wild-type cultures were reciprocally related to levels of viral growth in vitro (Fig. 4 and 5).

CD1d MAb increased susceptibility of WT splenocyte cultures to EMCV infection.

L929-mixed CD1d-KO splenocyte cultures consistently showed higher levels of EMCV replication compared to wild-type-derived cultures at the later time points (Fig. 6A), which was manifested especially clearly after reinfection of such cultures (Fig. 6A and B). Notably, wild-type-derived mixed cultures treated with soluble anti-CD1d antibody maintained markedly higher levels of viral replication (200 to 600% after MAb addition), while anti-CD1d had little if any specific effect on CD1d-KO cultures (Fig. 6A and B). When anti-CD1d antibody was added to wild-type splenocytes concurrently with primary infection, EMCV titers were higher than the levels of viral replication in CD1d-KO-derived cultures during primary infection and still higher than those in wild-type cultures after reinfection of previously MAb-treated cultures (Fig. 6A). When anti-CD1d antibody was freshly added only upon reinfection (to the previously untreated wild-type and CD1d-deficient cultures), the increase of viral titers in reinfected wild-type cultures was most pronounced (Fig. 6B).

Increased susceptibility of immune-subset-depleted splenocyte cultures to EMCV infection.

Since CD1d-KO splenocytes demonstrated a role for CD1d in resistance to EMCV replication in vitro (Fig. 1 to 5), we next determined the contribution of distinct immune subsets. We used high-speed FACS sorting, thereby eliminating concern of any residual undepleted cells and of detection of positively purified cells (blocking by depleting MAb) when using magnetic beads. Even with soluble fluorescence-conjugated MAb, it was also important to avoid stimulating residual cells by antibody cross-linking through multiparameter sorting (e.g., using CD3/NK1.1, leaving single positive NK cells and conventional T cells both activated by MAb cross-linking). Therefore, positive depletion of the two subsets tested was employed (i.e., FACS sorting only negative cells) (Fig. 7A), without use of either multiparameter staining (overlapping with other retained subsets) or negative depletion (leaving lineage-specific MAbs on most cells). Therefore, for DC we employed CD11c and the DC-specific marker 33D1 (61), thereby determining whether loss of all DC subsets (Fig. 7A and B) was critical in the antiviral response.

We have shown that unlike the case with CD1 KO mice (22), loss of invariant NKT cells alone (Jα18 KO mice) is not sufficient to impair the antiviral response in vivo (21), and there is currently no way to only deplete both invariant and “noninvariant” CD1d-dependent T cells. Nearest approaches either also deplete NK cells or activate NKT cells first, thereby changing the immune cell environment (4, 11, 15, 25, 29, 47, 58). Hence, we employed two different MAb (NK1.1 and Ly49b) to efficiently deplete both NK and NKT populations (Fig. 7A and B).

Effective depletion of either DC or NK/NKT cells was further documented in the corresponding cultures post-EMCV infection (Fig. 7B). Following infection there was a small increase in CD11c and NK1.1 expression (to 0.6% and 0.5%, respectively, from undetectable), as expected for these activation markers. More interestingly, in the NK/NKT-cell-depleted cultures, CD11c⁺ cells markedly increased (Fig. 7B) (to 12.0% versus 1.5% for undepleted and 0.6% for DC depleted). Since very few of these cells coexpressed CD4, most of these appear to be DC. CD4⁺ cells also increased by twofold in both depleted culture types. Whether each cell type proliferated in response to EMCV or was selectively spared from virus-mediated cell death in the context of the corresponding depletions is unclear. Significantly, however, the large numbers of DC-like cells were apparently unable to compensate for the profound effect of lack of NK/NKT cells on viral responses in these cultures.

Notably, splenocyte-L929 mixed cultures depleted of DC or NK and NKT cells maintained higher levels of viral replication in primary and secondary infection (Fig. 7C), most profoundly in secondary infection (2-fold for DC; 10-fold for NK/NKT cells), compared to mock-depleted cultures (Fig. 7C). Furthermore, systemic early IFN-γ levels were low (both with DC or NK/NKT-cell depletion) whereas IFN-α (DC only) was only modestly reduced (Fig. 7D and E), indicating that the majority of IFN-α in this infection was coming from other cell types.

T cells were found to be actually most active without both NK and NKT cells (Fig. 7F and G), rather than after DC depletion (modest increase, particularly in CD8⁺ cells but only at 2 days reinfection) (Fig. 7G). The responses were most pronounced for CD8⁺ T cells (Fig. 7F and G). Apparently, loss of NK/NKT cells so affected the response (notably, more than CD1d KO alone) that T cells actually became more activated. Therefore, loss of both NK and NKT cells, but not of all DC, severely impaired the antiviral response in vitro.

Increased susceptibility of CD1d-KO mice to infection with EMCV in vivo coupled with lower levels of lymphocyte activation and IFN-α production.

To determine whether viral replication and the immune response of CD1d-KO mice were similarly impacted by infection in vivo, mice were infected with EMCV-D as described previously (21, 22). Levels of virus and IFN-α and IFN-γ production in the sera were evaluated. Splenocytes were isolated from infected animals and the level of lymphocyte activation assessed ex vivo as in vitro. Representative results from three separate in vivo experiments are summarized in Fig. 8 and 9. The levels of EMCV-D in animal sera were found to be markedly higher in CD1d-KO mice (Fig. 8A). Interestingly, highest levels of CD8⁺ T-cell activation were found with least systemic viral load and vise versa, and viremia was inversely correlated with the levels of CD8⁺ T-cell activation with each individual animal (Fig. 8A, B, and C). Therefore, similarly to the above in vitro data, CD8⁺ T lymphocytes isolated from EMCV-infected wild-type animals showed uniformly higher activation than the equivalent cells isolated from EMCV-infected CD1d-KO mice (Fig. 8B and C). These differences were most pronounced at early points of infection when CD69 levels were highest (Fig. 8B and C) but were maintained until later points (not shown).

The lower level of activation of immune cells in CD1d-KO EMCV-infected animals was also observed for both T cells and B cells (Fig. 9A and B). The overall level of T-cell activation was higher in EMCV-infected wild-type animals, both for CD8⁺ and CD4⁺ populations, and was consistently observed in different experimental groups (Fig. 8B and C and A and B). However, no difference of activation levels was seen for cells stained with DC marker CD11c or monocyte marker F4/80 (not shown). In addition, increased viremia did not correspond to any detectable changes in the levels of systemic IFN-γ (not shown), as previously found in vivo (22). In contrast, the levels of IFN-α were notably elevated in the sera of wild-type mice upon EMCV infection, compared to those of CD1d-KO animals, which displayed little if any systemic IFN-α (Fig. 9C), as with mock-infected control animals (not shown).