Migration of activated CD8⁺ T lymphocytes to sites of viral infection does not require endothelial selectins

E/P-selectin deficient (E/P-sel −/−) and E-selectin deficient (E-sel −/−) mice14 were the progeny of breeder pairs kindly provided by Daniel Bullard, University of Alabama, whereas P-selectin (P-sel −/−) and P-selectin/ICAM-1 (P-sel/ICAM-1 −/−) deficient mice were obtained directly from the Jackson Laboratory, Bar Harbor, ME. All knockout strains were on a C57BL/6 background, and wild-type C57BL/6 mice were purchased from Bomholtgaard Ltd, Ry, Denmark. Seven- to 10-week-old mice were used in all experiments, and animals from outside sources were always allowed to acclimatize to the local environment for at least 1 week. All animals were housed under specific pathogen-free (SPF) conditions as validated by testing of sentinels for unwanted infections, according to the Federation of European Laboratory Animal Science Association guidelines.

LCMV of the Traub strain, produced and stored as previously described,41 was used in all experiments. Mice to be infected intracerebrally (ic) received a dose of 10³LD₅₀ in a volume of 0.03 mL. This route of inoculation results in a fatal T-cell–mediated meningitis from which the animals succumb 6 to 8 days after infection (pi).30,38 Mice to undergo systemic infection were given the same virus dose intravenously (iv) in a volume of 0.3 mL. This type of inoculation results in a transient, immunizing infection.42 

Virus titrations were carried out by ic inoculation of 10-fold dilutions of 10% organ suspension into young adult Swiss mice. Titration endpoints were calculated by the Kärber method and expressed as mean lethal doses (LD₅₀).

Mortality was used to evaluate the clinical severity of acute LCMV-induced meningitis. Mice were checked twice daily for a period of 14 days or until 100% mortality was reached.

Three different approaches were used to assess LCMV-specific DTH. (1) Mice were infected locally in the right hind footpad (fp) with 103 LD₅₀ of LCMV, and the local swelling reaction was followed between day 6 and 17 pi. (2) Mice were infected iv with the same dose of virus and challenged in the right hind footpad with 30 μL of an immunodominant class I restricted peptide (LCMV GP33-41, 50 μg/mL) on day 8 pi; footpad swelling was measured 16, 24, 48, and 72 hours later. (3) Mice were inoculated in their right hind fP with 30 μL of virus containing 10³ LD₅₀ of LCMV. Four hours later they were injected iv with 5 × 10⁷ splenic lymphocytes from wild-type donors, infected 8 days previously with the same dose of virus iv. Donor cells were routinely treated with mitomycin C (25 μg/mL, to prevent secondary expansion in the recipients) and depleted of plastic adherent cells by incubation for 1 hour at 37°C in tissue culture flasks. Footpad thickness was measured 16, 24, 48, and 72 hours after cell transfer, and to compensate for interexperimental variation wild-type recipients given the same donor cells were always included as a positive standard. Footpad thickness was measured with a dial caliper (Mitutoyo 7309, Mitutoyo Co, Tokyo, Japan), and virus-specific swelling was determined as the difference in thickness of the infected/challenged right and the uninfected/unchallenged left foot.42 43 

LCMV-primed donor splenocytes treated as described previously were incubated with sVCAM-1⁴⁴(kindly provided by L. Burkly, Biogen Inc, Cambridge, MA) or human IgG (Jackson ImmunoResearch) for control (0.5 mg/10⁸ cells/mL) for 30 minutes before injection into preinfected recipients (see above).

Spleens were removed from mice killed by ether anesthetization. Single-cell suspensions were obtained by pressing the organs through a fine steel mesh, and erythrocytes were lysed by 0.83% NH₄Cl treatment (Gey's solution). CSF exudate cells were obtained from the fourth ventricle of mice that had been ether anesthetized and exsanguinated; background level in uninfected mice is < 100 cells/μL.29 43 

Virus-specific cytotoxic T lymphocyte (T_C) activity was assayed in a standard ⁵¹Cr-release assay using histocompatible MC57G cells infected with LCMV for 48 hours as specific targets; uninfected MC57G cells served as control targets. Assay time was 5 hours, and percent specific release was calculated as described previously.41 42 

The following mAb were all purchased from PharMingen (San Diego, CA) as rat antimouse antibodies: FITC conjugated anti-CD49d (common α-chain of LPAM-1 and VLA-4, R1-2), phycoerythrin (PE) conjugated anti-CD8a,^53–67 biotin conjugated anti-CD62L (L-selectin, MEL-14), PE conjugated anti-CD162 (P-selectin glycoprotein-1, PSGL-1, 2PH1), FITC conjugated anti-Mac-1 (M1/70), and PE conjugated anti-IFN-γ (XMG1.2).

One × 10⁶ cells were stained with labeled mAbs in staining buffer (1% rat serum, 1% bovine-serum albumin [BSA], 0,1% NaN₃ in phosphate-buffered saline [PBS]) for 20 minutes in the dark at 4°C and subsequently washed. Cells incubated with biotin-conjugated Ab were further incubated with steptavidin-Tri-color (Caltag Laboratories, San Francisco, CA) and washed. After washing twice, cells were fixed with 1% paraformaldehyd. For visualization of LCMV-specific CD8⁺ T cells, splenocytes (10⁷ cells/mL) were incubated with an immunodominant MHC class I–resticted epitope (LCMV GP 33-41, 0.1 μg/mL), monensin (3 μmol) and IL-2 (100 IU/mL) for 5 hours, after which cells were surface labeled and fixed. Then the cells were permeabilized using PBS/saponin (0.05%), and anti-IFN-γ was added. After incubation for 20 minutes in the dark, cells were washed twice in PBS/saponin.31 Samples were analyzed using a FACSCalibur (Becton Dickinson, San Jose, CA), and 1 to 2 × 10⁴viable mononuclear cells were gated using a combination of forward angle and side scatter to exclude dead cells and debris. Data analysis was carried out using the Cell Quest program, and results are presented as contour plots.45 

We have previously shown that LCMV infection induces the generation of a large subset of CD8⁺ T cells with an activated phenotype (CD44^high LFA-1^highVLA-4^high L-sel^low) and capacity to exert effector functions (cytolysis and IFN-γ release) (reviewed in Thomsen et al34). Most, if not all, of the expanded T-cell population represents antigen-specific cells as recently disclosed through staining with specific peptide/MHC class I tetramers.46 To evaluate these cells with regard to their ability to participate in selectin-dependent binding, splenocytes from day 8–infected mice were analyzed for expression of L-selectin and PSGL-1, the best described receptor for P-selectin.47 In addition, the surface phenotype of LCMV-specific T cells was visualized through brief stimulation with an immunodominant LCMV-derived peptide (LCMV GP 33-41), followed by staining for IFN-γ intracellularly. Our results (Figure 1) revealed, as expected, that all LCMV-specific CD8⁺ T cells are of the VLA-4^high phenotype. In addition, we confirm that these cells express low levels of L-selectin. In contrast, the level of PSGL-1 was found to be increased on a large proportion of activated CD8⁺ T cells. As CD8⁺ T cells found at sites of inflammation are recruited exclusively from this activated subset (data not shown), we considered it pertinent to study the role of endothelial selectins in targeting of effector T cells to sites of infection.

To validate the use of E-sel −/− and E/P-sel −/− mice in the analysis of CD8⁺T-cell–mediated inflammatory reactions, initial experiments were carried out to ascertain that there was no impairment in the ability of these mice to generate effector T cells belonging to this subset. Functional analysis of the effector cell capacity in mutant mice compared with wild-type mice revealed an unimpaired virus-specific cytotoxic T-cell response, as there was no significant difference in the ability of splenic T cells to lyse virus-infected MC57G target cells 8 days after iv infection with LCMV in either E/P-sel −/− or E-sel −/− compared with wild-type mice (Figure 2). Direct visualization of antigen-specific T cells through detection of IFN-γ⁺cells gave similar results (data not shown). Furthermore, because the inflammatory response in LCMV-infected mice has previously been found to directly correlate with the number of activated CD8⁺ T cells generated in the spleen,35 45 the frequency of splenic CD8⁺ T cell with this phenotype (VLA-4^high) was analyzed in ic-infected mice on day 7 pi. The number of activated CD8⁺ T cells tended to be higher in E-sel −/− mice and slightly lower in E/P-sel −/− mice compared with wild-type mice (21% [15%-21%], 30% [27%-33%], and 26% [14%-30%] for E/P-sel −/−, E-sel −/− and wild type mice, respectively, medians (and ranges) of 4 mice), but none of these differences were statistically significant. Thus, the generation of CD8⁺ effector T cells seems to be unimpaired in infected mice lacking expression of endothelial selectins, allowing direct comparison of leukocyte migration to infected sites between these animals and wild-type mice.

To study the role of E- and P-selectin in the recruitment of lymphocytes to the LCMV-infected meninges, E/P-sel −/−, E-sel −/−, and wild-type mice were infected ic with 10³ LD₅₀ of LCMV and clinical susceptibility to meningitis, measured as mortality, was determined. Surprisingly, both groups of mutant mice developed LCMV-induced meningitis and succumbed to the infection to the same degree as did wild-type mice (Figure3); if anything, E/P-sel −/− mice tended to die earlier than wild-type mice. Quantitative and qualitative analyses of the cellular exudate were performed in parallel. No significant differences in the number of mononuclear cells contained in the CSF were revealed either on day 6 or 7 pi, and the composition of the inflammatory exudate with regard to CD8⁺T cells and Mac-1⁺ monocytes/macrophages was found to be similar in all 3 strains of mice (Table 1). Altogether, the above findings indicate that recruitment of virus-activated CD8⁺ T cells to the LCMV-infected meninges do not require local expression of endothelial selectins.

Another classical model of LCMV-induced CD8⁺T-cell–mediated inflammation is the primary footpad swelling reaction that is the response to subdermal viral challenge.43,48,49To study the role of endothelial selectins in this DTH-like reaction, E/P-sel −/− mice and wild-type B6 mice were infected in the right hind footpad with 10³ LD₅₀ of LCMV, and footpad swelling was measured between day 6 and 17 pi. No significant difference in the response pattern of mutants and wild-type mice was observed (Figure 4), indicating that, also, this subdermal inflammatory response could proceed in the absence of endothelial selectins. However, a relatively high degree of interindividual variation is intrinsic to this type of assay and this may mask minor differences in the kinetics. Some of the interindividual variation is likely to result from small variations in local virus replication and the time point of subsequent generalized virus spread. Therefore, we additionally measured the response of iv-infected mice to local challenge (on day 8 pi) with an immunodominant viral MHC class I–restricted peptide (LCMV GP33-41); previous results have clearly established that this is a valid way of assessing CD8⁺ T-cell dependent responses.33 50 Under these conditions, a slightly reduced response was observed in E/P-sel −/− mice, but not in E-sel −/− mice (Figure5). However, in all mice, a very substantial inflammatory reaction was induced, and similar results were obtained in mice primed 2 months earlier (data not shown), demonstrating that the redundancy of endothelial selectins did not simply result from the high number of recently activated effector cells present in mice undergoing acute infection.

A criticism often raised against the use of gene-targeted mice is that functional redundancy not relevant in wild-type mice may conceal the true importance of the targeted molecule. To minimize the risk that this argument might be valid in the case of selectin knockout mice, adoptive transfer experiments, using infected wild-type mice as cell donors and selectin-deficient mice as recipients, were performed. Mice were challenged with 103 LD₅₀ in the right hind footpad and subsequently transplanted 4 hours later with primed lymphocytes from wild-type donor mice infected iv 8 days previously with the same dose of virus. Footpad swelling was measured for the next 3 days; previous results have clearly established that this response is dependent on transfer of virus-specific CD8⁺ T cells.38 49 As seen from Figure6A, significant footpad swelling was induced in all recipients, and no major difference in the swelling reaction was observed between mutant and wild-type recipients, although in some experiments marginal reductions in the initial responses of especially E-sel −/− recipients were noted. These findings further suggest that virus-induced lymphocyte homing to infected areas can proceed without local expression of selectins. For comparison, the same reaction was also followed in recipients deficient of P-selectin (P-sel −/−) alone and of P-selectin together with ICAM-1 (P-sel/ICAM-1 −/−). A significantly reduced footpad swelling reaction was observed in P-sel/ICAM-1 −/− mice when compared with wild-type recipients. In contrast, little if any reduction was observed using P-sel −/− recipients (Figure 6B). Together, this indicates a dominant role for endothelial ligands of the Ig superfamily in the formation of this inflammatory response. To confirm this view, additional experiments were carried out in which the primed donor cells preincubated with soluble VCAM-1-Ig chimeric protein. Under these conditions, footpad swelling in ICAM-1 −/− recipients was virtually absent (Figure 6C).

As a final experiment, lymphocyte recruitment in the absence of endothelial selectin expression was indirectly analyzed by evaluating the kinetics of virus clearance in nonlymphoid organs (liver and lungs). Virus clearance in infected organs is a relevant measure of the capacity of CD8⁺ T cells to extravasate and exert their function locally.39 40 Because identical cytotoxic T-cell responses in mutant mice and wild-type controls have been documented above (Figure 1), a direct comparison between the capacity to clear virus from liver and lungs is valid. Mice were infected with 10³ LD₅₀ iv and, on the indicated days, 3 to 4 mice per group were killed, and organs (spleen, liver, and lungs) were removed and assayed for virus content (Figure7). Virus content in the spleen was used as an internal control indicative of T-cell effector capacity in the mice. Compared with wild-type mice, similar levels of virus was found in organs from E-sel −/− mice and E/P-sel −/− mice both before, during, and after the peak of the effector CD8⁺ T-cell response (∼day 8 pi). This pattern further suggests that effector T-cell migration to infected sites can proceed in the absence of local selectin expression and that, if there is a minor delay under these conditions, it is of marginal biologic relevance.

Selectins are generally assumed to play an essential role in the recruitment of leukocytes to inflamed tissue by mediating the initial tethering and rolling required before firm adhesion can be established.1 Several studies that use monoclonal antibody against selectins, as well as studies from knockout mice, have evidenced the general importance of these molecules in inflammatory reactions (recently reviewed in 2). Moreover, recent evidence indicates that endothelial selectins (E- and P-selectin) are needed for CD4⁺ T cells to migrate into inflamed skin.18 19 However, no studies have previously systematically addressed the role of these selectins in the migration of CD8⁺ effector T cells, which are central to clearance of many viral infections. Consequently, the current study was undertaken to investigate the requirement for local expression of selectins in extravasation of T_C1 cells at sites of viral infection.

As our primary experimental model, we used LCMV-induced meningitis, which is the direct result of antiviral T_C1–mediated inflammation of the meninges and ependyma in ic-infected mice.26,29,30,51 Additionally, localized subdermal inflammation (footpad swelling), as well as virus clearance in internal organs, was analyzed; both these reaction types are known also to be critically dependent on the recruitment of T_C1 cells.26,39,40,43 On the basis of previous observations concerning T_H1 cells,18,19 we had expected E- and P-selectins to be important for extravasation and the subsequent biologic effects of LCMV-specific T_C1 cells. However, careful quantitative and qualitative analyses of the inflammatory cells present in CSF of ic-infected mice failed to reveal any significant differences between knockouts and wild-type mice. Therefore, our results strongly indicate that migration of T_C1 cells to the infected meninges do not require local expression of endothelial selectins, and that other molecules must suffice in mediating the initial contact between CD8⁺ effector T cell and local endothelium, at least in this organ site. These findings are in variance with studies of cytokine-induced meningitis, in which it was demonstrated that the development of meningitis is inhibited in E/P-sel −/− mice.24 However, differences in the involved cytokines and/or the composition of the inflammatory exudate may be the cause of this difference. Expression of endothelial selectins was additionally found to be redundant regarding the ability of effector cells to eliminate virus in nonlymphoid organs. Virus clearance in infected organs is an direct measure of the capacity of T_C1 cells to function in situ,26,42 and the fact that we do not see any substantial delay in this parameter (day 8 pi values represent virus levels found at the steep slope of the virus elimination curve42) in selectin double knockout mice, further support the suggestion that T_C1 cells can be recruited to sites of infection in the absence of selectin-mediated rolling.

Using another model of LCMV-induced T-cell–mediated inflammation, virus-induced footpad swelling, we found that this subdermally induced inflammatory reaction was at most slightly impaired in E/P-sel −/− mice compared with wild-type mice. Similar results were obtained in LCMV-immune mice (mice primed 2 months earlier), demonstrating that redundancy of endothelial selectins could not be explained simply by the high numbers of effector cells or their activation state (effector vs memory cells). After adoptive transfer of wild-type effector cells into E/P-sel −/− recipients, the difference in the magnitude of footpad swelling was marginal or absent. Since it is our general experience that the adoptive approach is a very sensitive method of establishing the functional importance of a given adhesion molecule,34 the latter finding strongly indicates that endothelial selectins play, at most, a marginal role in the extravasation of T_C1 cells. The adoptive transfer experiments using primed wild-type donor cells should also partially counter the objection, that leukocytes in mutant mice could have adapted alternative ways to migrate to infected sites. In contrast to the intact responsiveness of E- and P-selectin–deficient mice, markedly impaired footpad swelling was observed in recipients deficient in expression of ICAM-1. Furthermore, the DTH response in the latter mice was completely inhibited by preincubation of the donor cells with soluble VCAM-1-Ig chimeric protein. Thus, in contrast to lack of endothelial selectins, preventing interactions with ligands of the Ig superfamily may completely inhibit the DTH response. This clearly confirms a pivotal role for endothelial ligands of the Ig superfamily,36 and may be taken to support recent findings suggesting that ICAM-1 and VCAM-1 may serve an auxiliary function in leukocyte rolling.52-55 Previous immunohistochemical analysis has clearly demonstrated the up-regulation of these molecules on the endothelium.29 35 

In other experimental models, eg, oxalone-induced DTH, absence of E- and P-selectin expression has been found to markedly inhibit the subdermal inflammatory response,22 and homing of T_H1 cells to inflamed skin has recently been found to be critically dependent on expression of these selectins.18-20Given that we have also investigated the capacity of E/P-sel −/− mice to support inflammation in this localization, our findings indicate that T_H1 and T_C1 differ in their requirements for extravasation at sites of subdermal inflammation. More important, using the present model system, we demonstrate that clinically relevant parameters such as virus elimination and susceptibility to ic infection as well as the formation, magnitude, and composition of an inflammatory exudate in the CNS during a viral infection is unaffected by deficiency of endothelial selectins. Additionally, our results suggest that other adhesion molecules suffice for rolling of T_C1 cells. As the level of L-selectin expression is down-regulated to a similar degree on activated CD8⁺ T cells in LCMV-infected E/P-sel −/− mice (data not shown), L-selectin would not appear to be the most obvious candidate. In contrast, α₄ integrin is likely to play a role.52 53 

We wish to thank Grethe Thørner Andersen and Lone Malte for expert technical assistance. Daniel Bullard is gratefully acknowlged for providing us with breeder pairs of E- and E/P-selectin knockout mice and Linda Burkly for providing soluble VCAM-1-Ig chimeric protein.