Role of Macrophage Inflammatory Protein-1α in T-Cell-Mediated Immunity to Viral Infection

Abstract

The immune response to lymphocytic choriomeningitis virus in mice lacking macrophage inflammatory protein-1α (MIP-1α) was evaluated. Generation of virus-specific effector T cells is unimpaired in MIP-1α-deficient mice. Furthermore, MIP-1α is not required for T-cell-mediated virus control or virus-induced T-cell-dependent inflammation. Thus, MIP-1α is not mandatory for T-cell-mediated antiviral immunity.

CD8⁺ T cells play an important role in antiviral immunity (11, 12). The main effector function of this cell subset is contact-dependent lysis (16, 35), but production of cytokines such as gamma interferon (IFN-γ) is also important (23, 26). Both of these effector mechanisms have a short action range, and cell-cell contact is therefore essential for virus-specific CD8⁺ T cells to exert their antiviral effector function (19). It has previously been found that viral infections are often associated with local production of inflammatory chemokines known to function as chemoattractants of activated lymphocytes (4, 24). Therefore, chemokines are likely to play a pivotal role in directing antiviral effector T cells to sites of viral replication in vivo.

Macrophage inflammatory protein-1α (MIP-1α) is one of the chemokines known to be produced in the context of many viral infections (8), and a number of studies have provided strong evidence indicating that MIP-1α plays an important role in virus-induced inflammation and in some cases also in antiviral immunity (8, 13, 15, 34). However, in the previous studies the cell population(s) responsible for reduction of viral titers has not been clearly defined, and the mechanisms underlying the observed inflammatory reactions are complex. Thus, to our knowledge no studies have previously addressed the role of MIP-1α in the generation and recruitment of CD8⁺ T cytotoxic type 1 (Tc1) effector cells to sites of viral infection. There is, however, evidence in the literature indicating that MIP-1α might be important for other types of CD8⁺ T-cell-mediated immunity (9, 30).

To study the role of MIP-1α in T-cell-mediated antiviral immunity, we used the murine lymphocytic choriomeningitis virus (LCMV) model (32). LCMV is a noncytopathic virus that causes little or no inflammation in the absence of a virus-specific T-cell response (2, 20). In immunocompetent mice, however, the appearance of effector T cells is associated with substantial inflammation in infected organs, and intracerebral (i.c.) infection leads to a fatal T-cell-mediated meningitis (10). Previous studies have revealed that the inflammatory exudate consists mainly of activated (L-selectin^low LFA-1^high very late antigen-4^high) (3) CD8⁺ Tc1 cells and monocytes and macrophages, whereas it contains very few CD4⁺ T lymphocytes (5, 10). By use of this model, our group has previously shown that MIP-1α is expressed in LCMV-infected organs on both the mRNA and protein levels (24). Furthermore, inflammatory exudate cells express chemokine receptors, which may bind MIP-1α. Thus, virus-activated CD8⁺ T cells express CCR5, and activated monocytes and macrophages express both CCR1 and CCR5 (24). It therefore appeared likely that MIP-1α might play a significant role in the recruitment of virus-specific Tc1 cells to sites of virus-induced inflammation. Consequently, the primary aim of this study was to define the importance of MIP-1α in directing Tc1 cells to sites of viral infection. In addition, we wanted to elucidate the contribution of Tc1 cells to MIP-1α production in situ during viral inflammation. For those purposes, we compared various parameters of the LCMV-specific T-cell response in wild-type (C57BL/6) and MIP-1α-deficient (MIP-1α^−/−) mice (B6.129P2-Scya3^tm1Unc backcrossed seven times to a C57BL/6 background and obtained as breeder pairs from the Jackson Laboratory [Bar Harbor, Maine]).

Migration of activated T cells toward MIP-1α in vitro.

To elucidate the potential role of MIP-1α as a chemoattractant for virus-activated CD8⁺ T cells, recruitment of splenocytes from wild-type as well as MIP-1α^−/− mice was evaluated in an in vitro chemotaxis assay (18). MIP-1α was found to induce the migration of splenocytes from virus-infected mice (Fig. 1A), and flow cytometric analysis confirmed that the recruited cells were preferentially activated (very late antigen-4^high) CD8⁺ T cells (Fig. 1B through D). Similar results were obtained when the knockout mice were used (Fig. 1A and data not shown). The capacity of MIP-1α to attract virus-activated CD8⁺ T cells together with the fact that MIP-1α is known to be produced in LCMV-infected tissues (4, 24) prompted us to study the biological importance of this chemokine in LCMV-induced inflammation.

FIG. 1.

Activated T cells migrate toward a MIP-1α gradient. Mice were infected with 200 PFU of LCMV Traub. (A) Spleen cells (1.5 × 10⁶) from mice infected (Inf.) 7 days earlier or from naive animals were allowed to migrate for 3 h using Transwell chambers (Corning Costar Corp.) with culture inserts of 6.5 mm and a 5-μm pore size; lower chambers contained MIP-1α (50 μg/ml; R&D Systems) or medium for control. The total number of migrated cells was determined, and results are presented as percentage of migrated cells; columns represent average ± standard deviation (SD) (n = 3). Representative fluorescence-activated cell sorter plots of input cells (B), transmigrated cells (MIP-1α) (C), and transmigrated cells (medium) from an infected mouse (D). Results of one of two similar experiments are presented.

CD8⁺ T-cell effector function ex vivo and virus clearance.

LCMV infection is primarily controlled through perforin-mediated killing of virus-infected cells (16, 35). To investigate whether the absence of MIP-1α would influence the generation of CTL effector cells, splenocytes from infected MIP-1α^−/− and wild-type mice were assayed for ex vivo lysis of ⁵¹Cr-labeled target cells coated with the immunodominant class I restricted peptides GP33-41 and nucleoprotein (NP) 396-404; peptide-specific lysis was evaluated on days 7, 9, and 14 after intravenous (i.v.) infection. As can be seen in Fig. 2A, splenocytes from MIP-1α^−/− and wild-type mice were equally cytotoxic against GP33-41-coated targets, and similar results were obtained for NP396-404-coated targets (data not shown). Enumerating virus-specific CD8⁺ T cells through intracellular staining for peptide-induced IFN-γ (22) confirmed that the absence of MIP-1α did not impair the afferent phase of the LCMV-induced immune response (data not shown). Together, these findings validate the subsequent use of these knockout mice to investigate the role of MIP-1α in directing the migration of effector T cells to sites of viral infection.

FIG. 2.

LCMV-specific Tc activity and organ virus titers in MIP-1α^−/− and wild-type mice. (A) Mice were infected i.v. with 200 PFU of LCMV, and cytotoxic activity was assayed in a ⁵¹Cr release assay by use of GP33-41 peptide-pulsed EL-4 cells as target cells. Unpulsed EL-4 cells served as control target cells. Results from individual mice are depicted; results of one of two similar experiments are presented. E/T, effector/target ratio. (B) Organs (spleen, lungs, and liver) were harvested on the days indicated, and organ virus titers were determined. Points represent individual mice. ID₅₀, 50% infective dose; D.L., detection limit.

A critical measure of the ability of CD8⁺ effector T cells to function in vivo is virus clearance from infected organs. Elimination of virus occurs most rapidly between day 7 and day 10 postinfection (p.i.), except in the lungs, where clearance is slightly delayed compared to the kinetics in organs like the spleen and liver (23, 31). To study the role of MIP-1α in CD8⁺ T-cell-mediated clearance of virus, virus levels in MIP-1α^−/− mice and wild-type controls were compared. Mice were infected i.v. with 200 PFU of LCMV Traub, and on day 9 p.i., groups of three mice were sacrificed, and the spleen, lungs, and liver were removed for determination of virus content. Since virus clearance is well under way but not complete at this point, virus levels measured at this time point would appropriately reveal any delay in virus clearance (7). As seen in Fig. 2B, similar levels of virus were found in MIP-1α^−/− and wild-type mice regardless of the organ site investigated. To further evaluate the influence of MIP-1α on the ability to clear the infection in nonlymphoid organs, lung and liver virus titers were also determined on day 14 p.i., and again similar (and compared to day 9 p.i., lower) levels of virus were observed, regardless of the genotype of the host.

Virus-induced delayed-type hypersensitivity in absence of MIP-1α.

The above results indicate that CD8⁺ effector T cells are capable of targeting sites of viral infection independently of whether or not MIP-1α is produced locally. However, to more directly study the kinetics of CD8⁺ effector T-cell migration into a solid tissue, the primary LCMV-induced footpad swelling reaction was used. This is a classical model of T-cell-mediated inflammation that represents the CD8⁺ T-cell-mediated response to subdermal virus challenge (6). To study the role of MIP-1α in this delayed-type hypersensitivity (DTH)-like reaction, MIP-1α^−/− mice and wild-type mice were infected in the right hind footpad, and the difference in thickness of the infected right and the uninfected left foot was determined between days 6 and 13 p.i. We observed no significant difference in the response pattern of MIP-1α^−/− and wild-type mice (Fig. 3A), indicating that a subdermal inflammatory response can also proceed normally in the absence of MIP-1α expression.

FIG. 3.

Kinetics of LCMV-induced inflammation in MIP-1α^−/− and wild-type mice. (A) A primary footpad swelling was elicited by infecting mice with 200 PFU of LCMV Armstrong in the right hind footpad. Points represent individual mice, and group medians are connected; results of one of two similar experiments are presented. (B) LCMV-induced meningitis was studied in MIP-1α^−/− and wild-type mice infected i.c. with 200 PFU of LCMV Traub. For some of the mice, mortality was registered (curves). For the remainder, CSF was harvested from the fourth ventricle of mice that had been ether anesthetized and exsanguinated, and the total number of cells present was determined (columns); background level in uninfected mice is <100 cells/μl. The number of mice per group is indicated in the figure, and results are presented as averages ± SD. Flow cytometric analysis (not depicted) revealed an identical composition of CSF cells (CD8⁺ T cells and Mac-1⁺ mononuclear phagocytes) regardless of genotype.

Role of MIP-1α in LCMV-induced meningitis.

When mice are infected i.c. with LCMV, a fatal T-cell-mediated meningitis is induced (2, 6, 10, 20). To investigate whether MIP-1α is necessary to recruit activated T cells to the brain, we infected MIP-1α^−/− mice and wild-type mice i.c., and for a portion of the mice we recorded the resulting mortality. In parallel experiments, we determined the number and composition of cells recovered from the cerebrospinal fluid (CSF) of infected mice. Regardless of the genotype, infected animals died about 8 to 10 days after i.c. challenge, and there was no difference in cell numbers or composition of inflammatory cells in the CSF (Fig. 3B). Thus, the recruitment of mononuclear effector cells to the LCMV-infected meninges is unimpaired in the absence of MIP-1α.

Taken together, the above results provide strong evidence that effector T-cell homing to virus-infected organs does not require MIP-1α. However, the redundancy of MIP-1α could be explained by compensatory mechanisms, e.g., the upregulation of other chemokines or cytokines at the inflammatory sites in LCMV-infected knockout mice. To investigate this possibility, expression of chemokine and cytokine genes was evaluated by RNase protection assays on total RNA (22, 24). RNA was extracted by use of the RNeasy kit (Qiagen, Hilden, Germany) from brains of MIP-1α^−/− and wild-type mice that were infected i.c. with LCMV 3 and 7 days earlier; these time points reflect the innate and T-cell-dependent phases of the inflammatory response, respectively (4, 24). Control animals of either strain were included to show basal transcript levels (Fig. 4). MIP-1α^−/− mice constitutively expressed mRNA for MIP-1β and TCA-3 in the brain, while there was no detectable expression of chemokines in the brains of uninfected wild-type mice. This difference in the basal expression of two chemokines, and of course the lack of MIP-1α, were the only differences that we observed regarding the composition and kinetics of mRNA for chemokines (Fig. 4) and cytokines (data not shown). To explore whether TCA-3 could be involved in some form of compensatory circuit, a separate RNase protection assay was run in which we looked for expression of CCR8, the receptor for TCA-3. No expression was observed in any group (data not shown). This finding, together with the fact that the onset of inflammation was associated with little or no increase in TCA-3 expression, led us to conclude that TCA-3 expression does not explain the redundancy of MIP-1α.

FIG. 4.

Cerebral chemokine transcripts in MIP-1α^−/− and wild-type mice following i.c. infection with LCMV. Mice were infected i.c. with 200 PFU of LCMV Traub. On the indicated days (0, 3, or 7 days p.i.), total RNA was isolated from the brain, and 20 μg of total RNA was subjected to RNase protection assay using the following kits obtained from PharMingen (San Diego, Calif.): RiboQuant in vitro transcription kit, mouse multiprobe template set mCK-5 (including the template for IP-10), and the RiboQuant RNase protection assay kit. Results from individual mice are presented.

Antigen-specific CD8⁺ T cells are the main source of MIP-1α in the LCMV-infected central nervous system.

MIP-1α may be produced by a variety of cell types (21), including CD8⁺ T cells (9). To investigate whether LCMV-specific T cells produce MIP-1α, splenocytes from naive mice and mice infected 9 days earlier were stimulated in vitro with a mixture of the immunodominant LCMV class I restricted peptides GP33-41 and NP396-404. Large amounts of MIP-1α (as determined by a sandwich enzyme-linked immunosorbent assay [R&D Systems, Minneapolis, Minn.]) were produced by LCMV-primed cells when stimulated with these class I restricted peptides for 6 h (data not shown). Virus-primed spleen cells left alone or stimulated with an irrelevant peptide did not produce detectable amounts of MIP-1α, nor did naive spleen cells stimulated with the same peptides. Thus, LCMV-primed cells produce MIP-1α upon recognition of cognate antigen.

To investigate the role of CD8⁺ T cells as producers of MIP-1α in vivo, we used an adoptive transfer setup, in which MIP-1α^−/− recipients received a small number of donor cells from TCR318 mice, which express a T-cell receptor (TCR) directed against GP33-41 on 50 to 60% of their CD8⁺ T cells (25). One day prior to i.c. infection with LCMV, MIP-1α^−/− mice received either 3 × 10⁶ whole spleen cells, the same number of splenocytes from which CD8⁺ T cells were depleted (by using anti-CD8 and rabbit complement [Cederlane, Hornby, Ontario, Canada]) or an equivalent number of purified (>90% pure through depletion of major histocompatibility class II⁺, Ig⁺, and CD4⁺ cells using monoclonal rat antibodies and magnetic beads coupled with sheep anti-rat and anti-mouse immunoglobulin G [Dynal, Oslo, Norway]) CD8⁺ T cells from naive TCR318 mice. Five days later (these mice developed meningitis with accelerated kinetics and died around day 6 p.i.), total RNA from the brains of the infected animals was extracted and analyzed with regard to chemokine transcripts (Fig. 5). Control mice and untransplanted mice infected i.c. 7 days earlier were included as references with regard to constitutive and maximal expression of chemokine transcripts in the brains of the two strains. As shown in Fig. 5, it is evident that both groups of recipients which received donor cells that included TCR⁺ CD8⁺ T cells have completely regained the capacity to express MIP-1α mRNA. This is documented by visual inspection as well as by quantification of band intensities (Fig. 5B); in the latter case, quantifications of mRNA for IP-10 and RANTES have been included as positive controls for an inflammatory response. Recipients that received splenocytes from which CD8⁺ T cells were depleted before cell transfer had no MIP-1α mRNA expression in the brain. Thus, our results indicate that CD8⁺ T cells are potent producers of MIP-1α and are the main source of this chemokine in the brains of LCMV-infected mice, although a minor contribution of parenchymal MIP-1α production cannot be ruled out.

FIG. 5.

Antigen-specific T cells are the main source of MIP-1α production in the brains of LCMV-infected mice. (A) Mice were infected i.c. with 200 PFU of LCMV Traub, total RNA was isolated from the brain, and 20 μg of total RNA was subjected to RNase protection assay. Groups 1 to 4, brains assayed on the indicated days; groups 5 to 7, MIP-1α^−/− mice infected i.c. with LCMV Traub 24 h after they had been injected i.v. with splenocytes from TCR 318 mice (group 5, 3 × 10⁵ splenocytes enriched for CD8⁺ T cells; group 6, 3 × 10⁶ total splenocytes; and group 7, 3 × 10⁶ splenocytes from which CD8⁺ T cells were depleted using anti-CD8 and complement). (B) Quantitative analysis of chemokine mRNA expression. Columns represent averages ± SD of the results from two to four mice.

Redundancy of MIP-1α does not require infectious antigen.

To test whether the redundancy of MIP-1α reflected the use of live antigen as a local trigger of inflammation, mice were infected i.v. and challenged in the footpad with LCMV-specific peptide (GP33-41); this method has previously been established as a valid way to assess CD8⁺ T-cell-mediated inflammation (17, 22). As can be seen in Fig. 6A, a substantial inflammatory response was induced in all MIP-1α^−/− mice, and no major difference was observed between MIP-1α^−/− and wild-type mice, although marginal reductions were noted in some experiments. Finally, to see if live, replicating antigen was required at any point during the immune response, mice were immunized subcutaneously with inert antigen (sheep red blood cells [SRBC]) and challenged 5 days later with this antigen in the footpad (Fig. 6B). In the context of this CD4⁺ T-cell-mediated DTH reaction, we also found MIP-1α to be redundant, suggesting that the redundancy we observed for LCMV-specific CD8⁺ T cells is not a unique situation.

FIG. 6.

Outcome of peptide-induced and SRBC-induced DTH. (A) Peptide-induced footpad swelling was elicited in mice infected with 200 PFU of LCMV Traub 8 days earlier. Mice were challenged in their right hind footpads with 30 μl of LCMV GP33-41 peptide (10 μg/ml), and the induced footpad swelling was measured at 16, 24, 48, and 72 h after peptide injection. (B) Mice were immunized subcutaneously with 0.1 ml of a freshly washed suspension of SRBC (2% in saline). Five days later, mice were challenged in their right hind footpads with 30 μl of 1.25% SRBC, and footpad swelling was measured at 16, 24, and 48 h after challenge. Averages ± SD of four to five mice are presented. Results of one of two or three similar experiments are presented.

At first glance, this conclusion appears to be at odds with findings from other viral model systems. For example, it has previously been found that after influenza (8) or paramyxovirus infection (13), MIP-1α^−/− mice have increased lung virus titers compared to wild-type mice. Furthermore, it has been documented that MIP-1α^−/− mice are resistant to coxsackievirus-induced myocarditis (8) and that such mice have a reduction in lung inflammation when infected with respiratory syncytial virus (15). However, the cellular and molecular events involved in the pathogenesis of those infections tend to be more complex, and the mechanisms underlying virus clearance are not as strongly related to the activity of a single effector cell population, namely effector Tc1 cells, as is true in the case of LCMV. This is clearly exemplified by analysis of murine cytomegalovirus-induced hepatitis. In this model, a complex series of cellular interactions is triggered by the viral infection, which at one particular step critically involves MIP-1α (27, 28): NK cells are crucial for early IFN-γ production, which again is required for local T-cell accumulation via a regulatory effect on Mig production and is critical for the initiation of this cascade early MIP-1α production at the site of viral replication. However, in the LCMV model, NK cells are not essential for inflammation (1), and the elicitation of an inflammatory response is entirely dependent on the generation and local accumulation of virus-specific CD8⁺ T cells (10, 20), which in this case can apparently function without involving MIP-1α at any level. Τhis comparison immediately raises questions as to the generality of our findings: is the chemokine profile unique to this infection explaining the redundancy, or can our results be extended to other cases of T-cell-mediated inflammation? At least two observations support the assumption that effector T-cell migration may often bypass a need for MIP-1α. First, the chemokine profile that is found in the context of the LCMV infection seems to be shared with several other viral infections (14, 24, 29, 33), indicating that there is nothing unique about this infection. Furthermore, even when inert antigen, namely SRBC, was used for both immunization and local challenge, we did not see any major impairment of T-cell-mediated inflammation, and the same was true when i.v. infected MIP-1α^−/− mice were challenged locally with GP33-41 peptide. Taken together, these findings strongly indicate that the redundancy of MIP-1α is not a reflection of a special chemokine profile related to this particular infection or even to the use of live virus. In conclusion, our results strongly indicate that although Tc1 effector cells are a relevant local source of MIP-1α and may migrate in response to this chemokine, the presence of MIP-1α is not mandatory for T-cell-mediated inflammation. For this reason, we propose that prior studies indicating a role for this chemokine in virus-induced inflammation be reinterpreted. Thus, it is possible that MIP-1α may serve a more critical role in attracting effector cells other than T cells, and the critical involvement of this chemokine in the context of certain viral infections may therefore reflect a pivotal role played by non-T-effector cells, e.g., NK cells, in the murine cytomegalovirus infection model (27, 28).