MyD88-Dependent Immunity to a Natural Model of Vaccinia Virus Infection Does Not Involve Toll-Like Receptor 2

Animals.

C57BL/6 mice were purchased from Charles River Laboratories. TLR2^−/− (catalog no. 004650), MyD88^−/− (catalog no. 009088), IL-6^−/− (catalog no. 002650), LysM:cre (catalog no. 004781), and MyD88:flox (catalog no. 008888) mice were purchased from Jackson Laboratory and subsequently bred at the Hershey Medical Center. LysM:wt/cre MyD88:flox/flox mice were generated by first breeding LysM:cre and MyD88:flox mice to create LysM:cre MyD88:flox double transgenics and then breeding LysM:cre MyD88:flox mice to MyD88:flox mice. All knockout mouse strains were on the C57BL/6 background after a minimum of 12 backcrosses to this strain. All animals were maintained in the specific-pathogen-free facility of the Hershey Medical Center and treated in accordance with the National Institutes of Health and AAALAC International regulations. Food and water were provided ad libitum, with MyD88^−/− mice receiving Baytril in chow and acidified water (pH 2.75) as antibacterial measures. All animal experiments and procedures were approved by the Penn State Hershey IACUC.

Viruses and infections.

VACV (strain WR) stocks were produced in 143B.TK⁻ cell monolayers and ultracentrifuged through a 45% sucrose cushion to purify viral particles from other cellular debris. For intradermal infection, mice were sedated using ketamine-xylazine and injected in each ear pinna with 10⁴ PFU of VACV in a volume of 10 μl. For i.v. and i.p. infection, mice were injected with 10⁶ or 10⁷ PFU of VACV in a volume of 500 μl. For assaying neutralizing antibodies in serum, VACV-NP-SIINFEKL-GFP (34) was used to infect B6/WT-3 cells. Wild-type VACV (Western Reserve strain) was used in all other experiments.

To assess dermal pathogenesis, ear thickness was measured using a 0.0001-in. dial micrometer (Mitutoyo), and lesion progression was measured using a ruler. To analyze the presence of replicating virus, organs were harvested, subjected to three freeze-thaw cycles in Hanks balanced salt solution (HBSS), ground in a 7-ml Dounce homogenizer, and sonicated. Lysate was placed on a monolayer of 143B.TK⁻ cells, and plaques were counted 2 days later. 143B.TK⁻ and B6/WT-3 cells were maintained in Dulbecco modified Eagle medium (DMEM) with 5% fetal bovine serum (FBS) (HyClone).

Flow cytometry of cells at the site of infection.

Ear pinnae were cut into strips and digested in 1 mg/ml collagenase XI (Sigma) for 60 min at 37°C. Live cells were blocked and stained on ice in 2.4G2 cell supernatant containing 10% normal mouse serum (Sigma). Antibodies included CD11b (M1/70), CD11c (N418), CD8 (53-6.7), B220 (RA3-6B2), and Ly6G (1A8) from eBioscience and Ly6C (clone AL-21), CD45.2 (104), CD19 (1D3), CD90.2 (53-2.1), NK1.1 (PK136), IL12p40/p70 (C15.6), and IL-6 (MP5-20F3) from BD Pharmingen. Phycoerythrin (PE)-Cy7-streptavidin (BD) was used to label biotin-conjugated antibodies. Dendritic cell (DC) subsets were identified as follows: B220⁺ “plasmacytoid” DCs (CD19⁻ CD90⁻ NK1.1⁻ CD11c⁺ CD11b⁻ B220⁺), CD11b⁺ “conventional” DCs (CD19⁻ CD90⁻ NK1.1⁻ CD11c⁺ CD11b⁺ B220⁻ CD8⁻), and CD8α⁺ “lymphoid resident” DCs (CD19⁻ CD90⁻ NK1.1⁻ CD11c⁺ CD11b⁻ B220⁻ CD8⁺). Sample acquisition was performed with an LSRII flow cytometer (BD), and data were analyzed with FlowJo software (TreeStar).

Analysis of neutralizing antibodies in serum.

Mice were deeply sedated with ketamine-xylazine, and a 25-gauge 5/8-in. needle was inserted below the xiphoid process to draw 400 to 800 μl blood from the heart. Serum was obtained by repeated centrifugation to remove all cells, followed by at least one freeze-thaw cycle. Serum was serially diluted in normal mouse serum and incubated at 37°C with VACV-NP-SIINFEKL-GFP. After 1 h, virus-serum mixtures were added to WT-3 cells at a multiplicity of infection (MOI) of 10:1 and incubated on a rotating rack at 37°C. Infectivity was measured as the percentage of green fluorescent protein-positive (GFP⁺) cells after 6 h of incubation with virus. Sample acquisition was performed with a FACSCanto flow cytometer (BD), and data were analyzed with FlowJo.

Intracellular cytokine staining.

Spleens were digested using 1 mg/ml collagenase D (Roche) for 30 min at 37°C and a 10-min ammonium-chloride-potassium (ACK) lysis to lyse red blood cells. The resulting splenocytes were stimulated with infectious VACV at an MOI of 10 in DMEM with 5% FBS for a total of 6 h. During the last 4 h of incubation, 10 μg/ml brefeldin A (Sigma) was added to the medium. To detect intracellular cytokines, cells were fixed in 1% formaldehyde (Acros Organics) and stained and washed in 2.4G2 cell supernatant containing 10% normal mouse serum and 0.5% saponin (Sigma). Sample acquisition was performed with an LSRII flow cytometer, and data were analyzed using FlowJo.

ELISpots.

Previously defined VACV major histocompatibility complex (MHC) class I and II epitopes (35) were screened for reactivity against splenocytes from WT C57BL6, TLR2^−/−, and MyD88^−/− mice at 10 days post-VACV ear infection. Splenocytes from naive mice were used as antigen-presenting cells (APCs). Naive splenocytes were incubated with the indicated peptide (final concentration, 2 μg/ml) in 37°C, 5% CO₂. VACV-primed and peptide-pulsed splenocytes were then coincubated overnight, and IFN-γ-positive T cell responses were assayed by IFN-γ enzyme-linked immunosorbent spot assay (ELISpot) (BD). Spots were counted using ImmunoSpot software (CTL). Peptides were synthesized by New England Peptide (Gardner, MA).

Real-time PCR.

The extraction of total RNA from tissue and reverse transcription of RNA were carried out as described previously (36), and the PCR was done as before (37). The following dye combinations were used for detection and data normalization: 6-carboxyfluorescein (FAM) (reporter for IFNs, MX-1, and IL-6) and hexachlorofluorescein (HEX) (reporter for glyceraldehyde-3-phosphate dehydrogenase [GAPDH]).

TLR2 is superfluous in controlling intradermal VACV infection.

To examine the role of TLR2 in poxvirus recognition, we infected mice via the natural intradermal (i.d.) route in the ear pinnae with 10⁴ PFU of VACV and monitored pathology and viral replication. TLR2^−/− mice were indistinguishable from wild-type mice in the degree and rate of ear swelling (Fig. 1A). Following the initial period of skin infection in which uncontrolled viral replication spurs recruitment of inflammatory monocytes, a lesion develops and is observed on both the inoculated convex surface and the smoother concave surface of the ear. Viral replication decreases past day 5 of infection and is no longer detectable once the lesion is resolved by loss of necrotic tissue (22). Tissue damage develops more aggressively in many immunodeficient mouse strains (38–40), but in TLR2^−/− mice, we observed that the pathology developed similarly to that in wild-type (WT) mice (Fig. 1B). At no stage of intradermal infection was there significantly more VACV replication in TLR2^−/− mice than in WT mice (Fig. 1C), although 7 of 12 TLR2^−/− mice, compared to 4 of 12 WT mice, had some detectable virus remaining on day 16. VACV normally remains restricted to the ear pinnae following i.d. infection, but components of the immune system are required to prevent systemic spread (38), and these components may be distinct from those that control virus replication locally. Therefore, we also looked for evidence that TLR2 restricted the spread of virus from the ear pinnae both by quantification of virus titers in the primary site of replication following systemic infection, the ovaries, and by evaluation of morbidity following VACV infection. In multiple experiments, TLR2^−/− mice had no detectable VACV in ovaries above that detected in WT mice (Fig. 1D), did not lose weight (Fig. 1E), and had no visible signs of systemic infection such as tail lesions. These data indicate that a successful innate response to a natural route of VACV infection did not depend on TLR2.

TLR2 contributes to activation of dendritic cells by VACV in vitro but not recruitment of myeloid cells in vivo.

The results above led us to inquire as to whether our stock of VACV was capable of activating dendritic cells through TLR2 in vitro. We harvested splenocytes from naive TLR2^−/− or WT mice and stimulated them with VACV before adding brefeldin A to block cytokine secretion. Cells were stained with antibodies for CD11c and other markers of dendritic cell subsets to divide DCs (CD19⁻ CD90⁻ NK1.1⁻ CD11c⁺) into “plasmacytoid” DCs (CD11b⁻ B220⁺), “conventional” DCs (CD11b⁺ B220⁻ CD8⁻), and CD8α⁺ “lymphoid resident” DCs (CD11b⁻ B220⁻ CD8⁺). The cells were also stained to reveal production of the proinflammatory cytokines IL-6 and IL-12. Though VACV is not as potent a stimulus as ECTV (not shown), more IL-6 and IL-12 were detected in VACV-stimulated cells than in unstimulated cells. TLR2^−/− DCs expressed lower levels of IL-6 (Fig. 2A) and IL-12 (Fig. 2B) than did WT DCs. Although it has been suggested that many reported TLR2 ligands are artifacts of contamination (13), we observed that the cytokine production triggered by sucrose cushion-purified VACV was at least as TLR2 dependent as that produced following exposure of splenocytes to crude lysate from VACV-infected cells (not shown).

Our in vitro results concur with other investigators' findings that TLR2 is important early in infection for production of proinflammatory cytokines, particularly IL-6 (25, 33). We investigated whether this deficit had practical importance in the recruitment of cells to the site of acute infection. The normal course of cell recruitment in the intradermal VACV model involves minimal swelling until day 3; an influx of classical inflammatory monocytes which peaks at day 5; then development of lesions and entry of tissue-protective CD11b⁺ Ly6G⁺ cells, which are attracted by unknown mechanisms and outnumber inflammatory monocytes by day 7; and a large number of lymphocytes (especially CD8⁺ T cells), along with continued influx of Ly6G⁺ cells, on days 9 to 11 (38). We used flow cytometry to characterize cells in the infected ears as infiltrating myeloid cells (CD45⁺ FSC^HI CD11b⁺ CD19⁻ CD90⁻ NK1.1⁻) and then as dendritic cells (CD11c⁺), inflammatory monocytes (CD11c⁻ Ly6C^HI Ly6G⁻), or tissue-protective Ly6G⁺ cells (CD11c⁻ Ly6C^MED Ly6G⁺) (Fig. 3A). For each of these subsets, WT and TLR2^−/− mice showed similar adherence to the predicted kinetics of cell recruitment (Fig. 3B to D). These data indicate that any role for TLR2 in stimulating the secretion of chemoattractants is not essential for attracting the cells most important for controlling VACV infection.

One of the most obvious reported deficiencies in TLR2^−/− mice challenged i.v. with VACV is their impaired secretion of IL-6 (33) or production of type I interferons (17). Therefore, to expand our investigations, we measured induction of transcripts of type I interferons, the interferon-inducible gene MX-1, or IL-6, on day 1 or day 7 after infection with VACV in the ear of TLR2^−/− mice versus wild-type mice by quantitative PCR. We found that induction of IL-6 and MX-1 transcripts was not reduced in TLR2^−/− mice versus those in WT mice on either day 1 or day 7 postinfection. Similarly, induction of type I interferons in WT or TLR2^−/− mice was not reduced (and may have been slightly enhanced) on day 1 postinfection and was in a similar range on day 7 postinfection. To expand our results, we examined the induction of transcripts in mice lacking MyD88 or in LysM:wt/cre MyD88:flox/flox mice, which have the MyD88 gene deleted specifically in myeloid cells, including those that infiltrate the site of infection. Surprisingly, we found little difference in any of the transcripts examined in either knockout strain (Fig. 4), indicating a MyD88-independent redundant pathway in the induction of these genes following i.d. VACV infection. Furthermore, IL-6 was not important for control of VACV following i.d. infection as virus titers and recruitment of myeloid cells were similar in IL-6 knockout and wild-type mice (not shown).

TLR2 may play a limited role in control of virus dissemination following systemic VACV infection.

To attempt to reconcile our above findings with the published work that describes a role for TLR2 in control of VACV infection, we systemically infected TLR2^−/− mice and WT mice with much larger inocula (up to 3 logs higher) than those used intradermally and monitored mice for 32 days for morbidity or mortality. During systemic VACV infection (i.p. or i.v.), the initial stage of immune mobilization is spurred by viremia and viral replication in internal organs and may be followed by skin lesions visible on the tails of mice (34, 41, 42)—akin to the progression of ECTV infection or smallpox but markedly less virulent (43, 44). In our system, i.v. infection with 10⁷ PFU is followed by weight loss that reaches a peak on day 5 postinfection. By day 5, the tails of infected mice become swollen and appear limp, and individual pocks are not evident until day 6, when they become distinguishable as discolored or ulcerated hairless spots. By day 9, the swelling recedes, and pocks begin to heal and appear as pale patches.

Using 12 mice in each group, we saw that VACV-infected TLR2^−/− and WT mice displayed similar kinetics of weight loss and recovery (Fig. 5A). All mice survived infection, but TLR2 knockouts had a statistically greater number of pocks on the tails (Fig. 5B) and were more prone to losing skin altogether near the tail tip (Fig. 5C). MyD88^−/− mice were also more susceptible to i.v. infection, suffering severe weight loss and mortality (Fig. 5D), which was notably increased in comparison to both WT and TLR2^−/− mice.

VACV is normally cleared from the liver and other internal organs after about a week (45, 46) but is particularly tropic for ovarian follicles and can be detected there after it has been cleared from all other organs (47). We saw that TLR2^−/− and WT mice infected with VACV i.v. had similar titers during the peak of viral replication, and there was no evidence that TLR2^−/− mice were slower to clear the infection (Fig. 5E). In addition to i.v. infected mice, we looked at VACV replication at an early/peak time point (72 h) in mice infected with 10⁷ PFU or 10⁶ PFU i.p. In none of these infection models did TLR2^−/− mice have more viral replication in either the liver or the ovaries (Fig. 5F). Other studies have examined VACV levels in the liver following infection (17, 33). The liver is one of the major sites of replication for ECTV (48), but following infection with VACV, titers in the liver typically do not approach those observed in the ovaries. As expected, we found that titers in the liver were 3 to 5 logs lower than those found in the ovaries, and there was no discernible difference between titers found in WT and TLR2^−/− mice. Our data suggest that TLR2^−/− mice are capable of controlling VACV infection in the internal organs, but the observation of increased lesions on TLR2^−/− infected mice supports a model in which TLR2 signaling, probably in phagocytes, limits the dissemination of the virus.

TLR2-independent MyD88 signaling is important for adaptive anti-VACV immunity.

Although it appears that TLR2 is superfluous for viral control and an innate immune response, we examined whether it may play a role in establishing the adaptive immunity that makes VACV such a valuable tool for immunization. CD8 T cell responses have been described to be dependent upon cell-intrinsic MyD88 following intranasal infection (28), a model in which the T cell response is required for protection. Following i.d. infection, T cells alone are not required for protection or to restrict VACV spread from the ear (38, 49), but we examined the role of TLR2 and MyD88 in regulation of CD4 responses (which do have a role following i.v. infection [42]) following i.d. infection. We used a library of peptides containing class II-restricted epitopes that are targeted by a significant number of T cells after VACV infection of mice (35). The number of functional splenic T cells targeting each peptide, 10 days postinfection, was assayed using gamma interferon ELISpot. Similar numbers of T cells targeting each of 17 class II-restricted epitopes detected at levels above background (Fig. 6) were found in in TLR2^−/− mice and WT mice. However, in MyD88^−/− mice, all 17 epitopes stimulated fewer T cells than in WT mice, with a significant difference for 7 of the 17.

CD4 T cell-dependent production of antibody is required for effective protection following i.v. challenge with VACV (42), so we examined whether the lack of MyD88 (or TLR2) that had an effect upon CD4 responses impacted the production of neutralizing antibody. We isolated serum either during acute i.d. infection (10 days postinfection) or after mice had successfully resolved the infection (32 days postinfection). To assay for the presence of neutralizing antibodies, we made serial dilutions of cell-free serum and then mixed serum with a strain of VACV engineered to express GFP upon infection of a target cell. Virus-serum mixtures were incubated with the murine WT-3 cell line, and the percentage of GFP⁺ cells was determined by flow cytometry. At both postinoculation time points (Fig. 7A and B), sera from WT and TLR2^−/− mice were equally capable of neutralizing VACV infectivity. However, MyD88^−/− sera had significantly lower levels of neutralizing antibody both during acute infection (Fig. 7C) and during the early stages of humoral memory (Fig. 7D). Therefore, our conclusions on the significance of MyD88 signaling for adaptive immunity are limited to B cells and CD4⁺ T cells. These data indicate that TLR2 alone does not play a role in establishing adaptive immunity to VACV, but at least one receptor upstream of MyD88 does play a major role.

TLR2-independent MyD88 signaling is needed for optimal control of intradermal VACV infection.

If redundancy in the innate immune system means that TLR2 alone is not required, there may still be a requirement for TLR2 in combination with other MyD88-dependent pathways. Synergy between two or more TLRs has been found to be important in other viral infections (10, 50). Therefore, we looked for evidence that other MyD88-dependent receptors are involved in controlling VACV replication and pathogenesis in this natural route of infection.

When observing the pathogenesis of intradermal VACV, we saw that the kinetics of ear swelling were the same in MyD88^−/− and WT mice (Fig. 8A), suggesting that there was no deficit in the immediate inflammatory mediators responsible for vascular permeability. Ear lesions spread at similar rates in the two mouse strains, but tissue loss proceeded at a slightly higher rate in MyD88-deficient mice, suggesting more tissue necrosis (Fig. 8B). When virus titers in the ear pinnae were measured on day 3, 5, or 7 postinfection, MyD88^−/− mice showed significantly more viral replication (Fig. 8C), a particularly profound finding as virus replication occurs similarly despite large differences in the inoculating dose (22). This difference was not detected on day 10, due partially to MyD88^−/− mice losing more of the tissue in which the virus replicates. The enhanced viral replication in MyD88 knockouts was not observed in LysM:wt/cre MyD88:flox/flox mice, which have the MyD88 gene deleted specifically in myeloid cells (Fig. 8D). Finally, we looked for viral replication in the ovaries on day 5 postinfection and saw no VACV in either MyD88^−/− or wild-type mice (data not shown), further suggesting that there is no impairment in the function of the myeloid phagocytes that block viral spread from the site of infection. However, these data show a clear role for TLR2-independent MyD88 signaling in the timely control of VACV replication at the site of active infection, as well as in the mobilization of lymphocytes to protect against future poxvirus challenges.