Intravenous (i.v.) inoculation of C57BL/6 mice with a low (5×104
PFU), medium (2×105
PFU), or high (2×106
PFU) dose of LCMV, strain Clone 13, resulted in different degrees of pathology, as indicated by weight loss () and by histological analysis of lung sections at day 15 p.i. (). The high dose caused a precipitous drop in body weight during the first week of infection (, right
), but, thereafter, clonal exhaustion and deletion of LCMV-specific T cells resulted in a persistent infection9–10
associated with minimal lung pathology (, Right
) and 100% (77 of 77) survival (, Top
). Selective depletion of NK cells using 25 µg of anti-NK1.1 mAb (Supplemental Fig. 1
) one day prior to high dose infection resulted in 58% (35 of 65) mortality between days 9 and 13 of infection (, Top
) associated with severe pulmonary edema (data not shown) and reduced viral titers by day 7 p.i. (, Right
). Under these high dose conditions, therefore, the presence of NK cells promoted persistence and prevented mortality.
NK cells influence T cell-dependent pathology and viral persistence during LCMV infection
In contrast to the beneficial role of NK cells during high dose infection, NK cell-depletion prevented the severe weight loss (, Middle) and tissue pathology (, Middle) associated with the medium dose of LCMV. Twenty-three percent (7 of 31) of control-treated mice succumbed to the medium dose during the second week of infection, and the lungs of surviving mice exhibited bronchus associated lymphoid tissue, pulmonary edema, and interstitial mononuclear infiltration. Lung pathology was absent in NK cell-depleted mice, which uniformly survived medium dose challenge (, Bottom). Moreover, while high levels of replicating virus persisted in surviving control mice at day 15 p.i., NK cell-depletion resulted in complete viral clearance (, Middle). In this case the presence of NK cells was detrimental for the host, as they promoted immune pathology and death.
Irrespective of the presence of NK cells, inoculation with a low dose of virus was uniformly non-lethal in 18 of 18 (100%) control and 18 of 18 (100%) of NK cell-depleted mice by >50 days p.i., with minimal weight loss (, Left) and minimal lung pathology (, Left). Virus was completely cleared in both groups of mice by day 15 of low dose infection (data not shown), but NK cell depletion resulted in more rapid elimination of LCMV in liver by day 7 p.i. (, Left).
The weight loss, lung pathology, and mortality observed in medium dose-infected wild-type mice () did not occur after infection of αβ T cell-deficient (TCRβ−/−
) mice, and NK cell depletion of TCRβ−/−
mice did not alter weight loss or viral burden (Supplemental Fig. 2
). Thus, NK cells regulate viral clearance and immunopathology during LCMV infection through a T cell-dependent mechanism.
As early as day 6 after medium dose infection, the proportion and number of IFN-γ+
LCMV-specific CD8 T cells was increased 2- to 6-fold in mice depleted of NK cells ( and Supplemental Fig. 3
), and anti-viral T cells from these mice displayed an enhanced ability to co-produce TNF (Supplemental Fig. 3
). The number of NP396–404
tetramer-binding CD8 T cells in the spleen on day 5 p.i. was increased 4- to 20-fold in NK cell-depleted mice relative to non-depleted control mice after infection with all doses of virus (). The number of virus-specific IFN-γ+
CD4 T cells was also amplified 7-to 20-fold by NK cell depletion compared to control mice on different days after medium dose infection (). Moreover, co-production of TNF and IL-2 by anti-viral CD4 T cells was augmented by NK cell depletion ( and Supplemental Fig. 3
). The increased magnitude of the LCMV-specific T cell response in the absence of NK cells during medium dose infection correlated with rapid viral clearance (). Depletion of NK cells using a carefully titrated dose of anti-asialo GM1 antibody, which eliminates NK cells but not CD8 T cells11
, also enhanced anti-viral CD4 and CD8 T cell responses during medium dose infection (Supplemental Fig. 4
LCMV-specific T-cell responses enhanced in NK cell-depleted mice
The enhanced anti-viral T cell responses suggested that NK cell depletion may augment proliferation of LCMV-specific T cells. Transfer of CFSE-labeled Thy1.1+
T cells revealed a larger population of CFSElow
donor CD4 () and CD8 (data not shown) T cells in multiple host tissues at day 6 p.i. of high dose infection in the absence of NK cells. There also was greater specific lysis of viral peptide-coated target cells as detected by a conventional in vivo
cytotoxicity assay at day 4 of infection (Supplemental Fig. 5
). Moreover, LCMV-specific Ly5.1+
TCR transgenic (P14) CD8 T cells (transfer 104
) were recovered from tissues of NK cell-depleted recipient (Ly5.2+
) mice at 2- to 9-fold greater numbers than control-treated mice 6 days after low dose infection (Supplemental Fig. 5
). Together these results indicate that NK1.1+ cells repress the size of the anti-viral T cell response during LCMV infection.
The activities of CD4 T cells are important for maintaining CD8 T cell function during LCMV infection12–14
. To assess whether CD4 T cells were involved in the NK cell suppression of LCMV-specific CD8 T cells, mice were treated with antibodies to concurrently deplete both NK and CD4 T cells. Whereas depletion of NK cells prior to medium dose LCMV infection resulted in a >200-fold reduction in splenic viral titers at day 7 p.i. relative to control and CD4-depleted (ΔCD4) mice (), depletion of both NK and CD4 T cells (ΔNKΔCD4) had no effect on viral titers. The increased numbers and enhanced multiple cytokine production by anti-viral CD8 T cells caused by NK cell depletion were also prevented by co-depletion of CD4 T cells (). In contrast, co-depletion of NK and CD8 T cells did not prevent an increase in IFN-γ+
-specific CD4 T cells (Control: 3.8±0.5 % vs. ΔNK: 9.8±0.7 % vs ΔNK/ΔCD8: 10.7±1.6 %, n=3, p
<0.05 vs control) at day 12 of medium dose infection.
Role of CD4 T cells in NK suppression of anti-viral CD8 T cells, viral control, and immunopathology
Paradoxically, at the high virus dose, co-depletion of NK and CD4 T cells prevented the severe pulmonary edema () and heightened mortality () associated with depletion of NK cells alone. In this experiment, mice were harvested on day 12 p.i., when 3 surviving NK cell-depleted mice were moribund and required euthanasia, while all double-depleted mice displayed relatively normal vigor. The livers of NK/CD4 double-depleted mice contained 25-fold more PFU than livers from mice depleted of NK cells alone (NK: 5.7±0.2 PFU vs ΔNK/ΔCD4: 7.1±0.1 PFU, n=5, p
<0.0001). Enhancement of LCMV-specific CD8 T cells in the absence of NK cells was also abrogated by concurrent depletion of CD4 T cells (Supplemental Fig. 6
). Together these data indicate that CD4 T cells are needed for NK cell modulation of antiviral CD8 T cell responses associated with viral clearance, persistence, and immunopathology.
We utilized a modified in vivo cytotoxicity assay by injecting splenocytes from medium dose-infected NK cell-depleted mice (Ly5.1+, day 4 p.i.) into medium dose-infected NK cell-depleted (ΔNK) or isotype IgG2a-treated (Control) recipient mice (Ly5.2+, day 3 p.i.). After 5 hours, similar proportions of total donor T (Control: 0.16±0.03 % vs. ΔNK: 0.15±0.02 %, n=21, p=0.80) and B cells (Control: 1.8±0.2 % vs. ΔNK: 1.7±0.2 %, n=21, p=0.88) were recovered from infected recipients, regardless of NK cell depletion. Likewise, recovery of activated (CD44hi CD43(1B11)+) donor CD8 T cells was similar from spleens of Control and ΔNK mice, with minimal loss relative to uninfected control mice (). In contrast, there was a substantial loss of activated donor CD4 T cells in infected relative to uninfected recipients, and this loss was prevented by depletion of NK cells (). The magnitude of NK cell-dependent loss of activated donor CD4 T cells was similar in low-, medium-, and high dose-infected recipients (). More activated CD4 T cells, both donor- and host-derived, in infected (Control) mice stained positively for the apoptosis indicator, AnnexinV, in comparison to naive donor CD4 T cells or to activated donor CD4 T cells in medium dose-infected ΔNK recipient mice (). In contrast to activated donor CD4 T cells, the recoveries of naïve (CD44low) phenotype CD4 and CD8 donor T cells were not altered by NK cell-depletion (data not shown). These data indicate that NK cells in WT mice selectively and rapidly target activated CD4 T cells for elimination during LCMV infection.
NK cells rapidly eliminate activated CD4 T cells
We next examined the involvement of NK cell cytolytic mediators FasL, TNF, and perforin (Prf1) in this process. The loss of activated WT donor CD4 T cells in infected WT recipient mice () was seen when activated lpr (Fas mutant) mouse donor cells were transferred into WT recipient mice or when WT donor cells were transferred into TNF−/− recipient mice (). In contrast, there was relatively little loss of activated WT donor CD4 T cells in Prf1−/− hosts (), whose retention of activated donor CD4 cells was not significantly different (p > 0.1) from that in NK cell-depleted WT or Prf1−/− hosts. Thus, NK cell elimination of activated CD4 T cells is mediated through a perforin-dependent pathway that does not require Fas or TNF.
Previous work has implicated NKG2D in targeting of activated T cells by murine NK cells in vitro15–17
, but we observed no differences in activated WT donor CD4 T cell survival in WT vs. NKG2D−/−
recipients () or in WT mice treated with a blocking mAb to NKG2D18
(data not shown). Of note is that we did not observe the expression of ligands for activating NK cell receptors including NKG2D, NKp46, DNAM-1, and TRAIL on these early activated CD4 T cells, and NK cell-mediated elimination of activated donor CD4 T cells also occurred in antibody-deficient (µMT−/−
) mice (data not shown), precluding a role for antibody-dependent mechanisms. Activated CD4 T cells did, however, express much higher levels of adhesion molecules than naïve cells, and these molecules have previously been shown to trigger NK cell cytotoxicity via LFA-119–20
. Somewhat surprising was the observation that the activated CD4 T cells were far more susceptible than activated CD8 T cells to direct killing by the NK cells, even though both expressed high levels of adhesion molecules. We previously had shown that the presence of the negatively signaling receptor CD244 (2B4) on NK cells prevented NK cell-mediated lysis of activated CD8 T cells21
. We found here that while expression of the CD244 ligand, CD48, was up-regulated on T cells after medium dose LCMV infection, expression levels of CD48 were much higher on activated CD8 than on activated CD4 cells (MFI of CD48: activated CD4, 3,423±147; activated CD8, 6,180±166; n=9, p
<0.0001) (Supplemental Fig. 7
In order to assess whether NK cell-mediated lysis of activated CD4 T cells is a general principle of virus infections, we examined the loss of LCMV-activated CD4 T cells following transfer into mice inoculated with an unrelated Arenavirus, Pichinde virus (PV), the Coronavirus mouse hepatitis virus (MHV), or the interferon inducer and NK cell activator poly I:C (pI:C). All three stimuli induced measureable loss of activated donor CD4 T cells that was dependent upon the presence of NK cells (). In reciprocal experiments, CD4 T cells activated during infection with PV, MHV, vaccinia virus (VV), and MCMV were lost upon transfer into mice infected with medium dose LCMV when NK cells were present ().
An analysis of the window of time at which NK cell regulation of T cells occurred in the LCMV medium dose model showed reduced frequencies of activated donor CD4 T cells relative to uninfected recipients by in vivo
cytotoxicity assays after transfer into NK cell-sufficient mice one (43%), two (32%), three (26%), four (4%), and five (8%) days after infection (Supplemental Fig. 8
). The frequencies of activated donor CD4 T cells were increased by NK cell depletion in recipient mice only at day 2 and day 3 p.i. These results suggest that NK cells target activated CD4 T cells mainly on the second and third day of infection, when the cytolytic activity of NK cells is at its peak22
These results show that NK cells can play a crucial role in controlling virus-associated morbidity, mortality, and persistence in the absence of direct NK cell-mediated control of virus replication, and they do so by altering the numbers and polyfunctionality of virus-specific T cells. Their effect on activated CD4 T cells was presumably due to direct cytotoxicity, as demonstrated by rapid perforin-dependent elimination of activated CD4 T cells by NK cells in short term in vivo
cytotoxicity assays and the observation of enhanced annexin reactivity of CD4 T cells in the presence of NK cells. The effect of activated NK cells on CD8 T cells and virus clearance appeared to be indirect, and depended on the presence of CD4 T cells, which are known to produce factors that preserve CD8 T-cell viability and functionality12–13,23–25
. This suggests a three-way interaction, whereby NK cells suppress the CD4 T cell response, thereby preventing augmentation of the CD8 T cell response, which, in turn, directly regulates viral clearance and immunopathology in this system ().
Our observation of direct NK cell-mediated lysis of T cells during virus infection is distinct from published accounts of NK cell regulation of anti-viral T cells during MCMV infection, in which NK cell-mediated lysis of virus-infected cells contributes to control of viral burden and persistence of MCMV-infected dendritic cells (DC) that in turn regulate activity of anti-viral T cells6–7,26
. Moreover, we found no NK cell-dependent changes in the number and antigen-presenting function of splenic DCs during LCMV infection (Supplemental Fig. 9
), consistent with our finding that NK cells directly regulated the T cells. A possible concern was that in vivo
cellular depletion with antibodies against NK1.1 and CD4 have the potential to affect frequencies of NKT, γδ T, and regulatory T cells, but the frequencies of these lymphocytes in the spleen of LCMV-infected mice were not altered after 3 to 4 days of infection by the low concentration of anti-NK1.1 used in these studies (Supplemental Figs. 1 & 10
). Moreover, depletion of NK cells in γδ T cell-deficient (TCRδ−/−
) and NK T cell-deficient (CD1d−/−
) mice during medium dose infection enhanced LCMV-specific T cell responses and reduced viral loads (Supplemental Fig. 9
), similar to that in WT mice. Therefore, these lymphocyte lineages appear dispensable for NK cell immunoregulatory function during LCMV infection.
By adjusting the dose of the LCMV inocula, one can generate diverse patterns of CD8 T cell-regulated pathogenesis similar to the variety of pathogenic patterns a human HCV infection can take, including rapid viral clearance, severe T cell-dependent immunopathology, and long-term persistence. We show here that at a high dose of LCMV NK cells act beneficially by suppressing T cell responses, thereby preventing severe pathology and mortality while enabling the development of a persistent infection from which mice eventually recover and clear virus27
. At the medium dose inoculum, NK cell suppression of T cells is detrimental to the host, as virus clearance is impaired due to the limited number and functionality of T cells. However, a medium dose of virus is not sufficient for complete clonal exhaustion of T cells, ultimately resulting in severe T cell-dependent immunopathology that can lead to death of the host.
These results suggest that NK cells can serve as rheostats or master regulators of anti-viral T cell responses. Consistent with the fact that many virus infections induce cytokines that potently activate NK cells28
, we found that NK cell lysis of activated CD4 T cells was triggered by several viruses as well as after inoculation with pI:C, which induces interferon and activates NK cells. Although a previous study found that NK cell depletion did not alter the magnitude of antiviral T cell responses during infection with the Armstrong strain of LCMV29
, we have observed enhanced anti-viral T cell responses and improved viral control at early time points after infection of NK cell-depleted mice with both LCMV Armstrong and Pichinde virus (SNW, unpublished observations). Thus, the timing and the type of evaluation may be important to detect detrimental effects of NK cells on T cells during more benign viral infections.