Locally Produced IL-10 Limits Cutaneous Vaccinia Virus Spread

Skin infection with the poxvirus vaccinia (VV) elicits a powerful, inflammatory cellular response that clears virus infection in a coordinated, spatially organized manner. Given the high concentration of pro-inflammatory effectors at areas of viral infection, it is unclear how tissue pathology is limited while virus-infected cells are being eliminated. To better understand the spatial dynamics of the anti-inflammatory response to a cutaneous viral infection, we first screened cytokine mRNA expression levels after epicutaneous (ec.) VV infection and found a large increase the anti-inflammatory cytokine IL-10. Ex vivo analyses revealed that T cells in the skin were the primary IL-10-producing cells. To understand the distribution of IL-10-producing T cells in vivo, we performed multiphoton intravital microscopy (MPM) of VV-infected mice, assessing the location and dynamic behavior of IL-10 producing cells. Although virus-specific T cells were distributed throughout areas of the inflamed skin lacking overt virus-infection, IL-10+ cells closely associated with large keratinocytic foci of virus replication where they exhibited similar motility patterns to bulk antigen-specific CD8+ T cells. Paradoxically, neutralizing secreted IL-10 in vivo with an anti-IL-10 antibody increased viral lesion size and viral replication. Additional analyses demonstrated that IL-10 antibody administration decreased recruitment of CCR2+ inflammatory monocytes, which were important for reducing viral burden in the infected skin. Based upon these findings, we conclude that spatially concentrated IL-10 production limits cutaneous viral replication and dissemination, likely through modulation of the innate immune repertoire at the site of viral growth.


Author Summary
While ineffective antiviral immune responses can result in illness or even death, excessive host responses can also cause substantial injury. Anti-inflammatory proteins play an important regulatory role in limiting immune-mediated damage, but it is unknown where the cells making these modulators need to be for the greatest effect. The best-described Introduction Ideally, the antiviral immune response eliminates actively replicating virus and any viral reservoirs without undue host damage. For many viruses, however, the immune response extends beyond that necessary for viral clearance and creates disease symptoms. For example, infection with certain influenza virus (IAV) strains results in lung recruitment of high numbers of neutrophils (and other leukocytes) leading to a fast-progressing viral pneumonia and extensive lung damage [1,2]. Neutrophil recruitment at late stages of infection can also lead to CNS pathology during coronavirus infection [3]. Virus-specific CD4 + T cells substantially contribute to respiratory syncytial virus (RSV)-induced bronchiolitis in children [4]. Further, early RSV vaccination strategies actually enhanced disease due to excessive cellular infiltration of the lungs and subsequent pulmonary injury [5]. Thus, the immune response, essential for eliminating pathogens, also produces disease if not appropriately modulated.
Paradoxically, during some acute viral infections highly activated, pro-inflammatory CD8 + T cells can also produce IL-10 and even represent a major source of IL-10 in infected organs [21][22][23][24]. The role of such poly-secretory CD8 + T cells appears to be dictated by both viral tropism and the site of infection, but the principle function ascribed to these cells is suppression of inflammatory tissue damage rather than direct alteration of viral replication. For example: antibody (Ab) blockade of the IL-10 receptor (IL-10R) during IAV infection increases lung inflammation and mortality without reducing viral burden [21]. Similarly, RSV-activated IL-10 + CD8 + effector T cells prevent RSV-induced inflammatory lung pathology [23]. During coronavirus-infection, IL-10 + CD8 + T cells limit CNS demyelination and slightly increase viral burden as a necessary cost [12]. The extent to which localized IL-10 production by T cells occurs in other acute viral infections in organs more resilient to inflammatory damage is uncertain.
Multiple aspects of IL-10 production have been proposed to shape disease outcome, including IL-10 production kinetics and shifts in the anatomical location of IL-10 depots [9,25]. The local distribution of IL-10 (or any other inflammatory cytokine) production in situ or the spatiotemporal contribution of IL-10-producing cells to viral clearance is unknown. Previously, we reported that during vaccinia virus (VV) infection of the skin, myeloid innate immune cells and CD8 + effector T cells exhibited strikingly different distributions [26]. Thus, the precise positioning and behavior of IL-10-producing cells could dramatically influence virus clearance in localized areas.
Here, we analyze the dynamics of IL-10-producing cells in situ after epicutaneous (ec.) VV infection. VV skin infection induces a large population of IL-10 + CD4 + and IL-10 + CD8 + T cells in the infected tissue. Intriguingly, IL-10 + T cells more closely localize to viral replicative foci than do the bulk population of antiviral CD8 + T cells in the tissue. Further, IL-10 + T cells move in paths adjacent to areas of virus infection, and neutralization of IL-10 enhances virus spread and shapes the innate immune population located adjacent to these foci. Together, our data demonstrate that spatially directed anti-inflammatory cytokine production inhibits viral dissemination within the infected tissue and illuminate a novel aspect of antiviral immunity.

Epicutaneous VV infection elicits a robust IL-10 + T cell response
Multiple inflammatory immune effectors, including cytotoxic CD8 + T cells and innate Ly6G + cells, synergize to control recombinant VV (rVV) infection of the skin [26]. Despite a highly inflammatory Th 1 -polarized cellular infiltrate, viral lesions after epicutaneous (ec.) infection of C57Bl/6 mice resolve in 14 days with limited, rapidly repaired tissue damage.
To understand the potential mechanisms of immune self-restraint during ec. VV infection, we performed qPCR arrays, analyzing expression of cytokine mRNAs isolated from infected tissue at 6.5 days post-infection (d.p.i.) (Fig 1A). At this time, VV titers begin to wane in the skin concomitant with the entry of substantial numbers of virus-specific T cells [26]. qPCR revealed that VV-infection induces multiple pro-inflammatory cytokine mRNAs, including IL-1β, IL-6, and IFN-γ (black dots), but also surprisingly, IL-10 (magenta dots), which increased > 10-fold. Temporal mRNA analysis showed maximal IL-10 mRNA induction peaked around day 5 p.i. followed by a gradual decline ( Fig 1B).
Using flow cytometry to monitor IL-10 production, we examined single cell suspensions derived from the ears of ec. VV-infected IL-10gfp reporter mice ("TIGER" (interleukin-ten IRES GFP-enhanced reporter mice)) [27] (Fig 1C-1H). Kinetic analysis identified GFP + IL-10-producing cells in the skin by 5 d.p.i., peaking on day 6 at 2.2% of total cells, and then declining to a near constant level of~1% of total cells until 10 d.p.i. (Fig 1C). Consistent with previous reports [27], we detected few (0.3%) GFP + cells in uninfected reporter mice, or in infected wild-type mice (0.05% GFP + ) ( Fig 1D). We found similar frequencies of GFP + cells (1.84%) using an alternative IL-10 reporter mouse (Fig 1D, far right panel) (Vert-X, also expressing IL-10-IRES-eGFP, but with a chimeric poly A tail sequence on transcripts [28] (see [29] for discussion of the properties of IL-10 reporter strains).
Surprisingly, the majority (> 65%) of IL-10gfp + skin cells on day 6 post-recombinant vaccinia virus (rVV)-infection of TIGER mice were CD8 + T cells (Fig 1E), as defined by the cellsurface markers CD45, CD8α, and CD8β. CD4 + T cells comprised the second largest GFP + population (~15%) on day 6, with minimal contributions from dendritic cells (DCs, CD11c + , CD11b -) or macrophages (CD11b + CD11c -). GFP + populations were virtually identical in Vert-X reporter mice (shown 6 d.p.i., Fig 1F), confirming CD8 + T cells as the most prevalent IL-10-producing cells recovered from VV-infected skin on day 6 p.i. On a per-cell basis on day 6 p.i., CD8 + T cells expressed slightly higher levels of the GFP than CD4 + T cells, which were, in turn, higher than the very small number of GFP + macrophages or DCs (Fig 1G). Over time, the mean fluorescence intensity of T cells was roughly equivalent, until CD4 + T cells acquired higher MFI at late times in infection ( Fig 1H). We confirmed that both CD4 + and CD8 + T cells were producing IL-10 protein by intracellular cytokine staining (S1 Fig).
Together, these data show that ec. VV infection elicits a sizeable population of IL-10-expressing T cells in the skin over the course of acute infection, with IL-10 + CD8 + T cells being more numerous.

IL-10gfp + T cells accumulate in VV-infected tissue
We next examined VV replication after ec. infection of the skin (Fig 2A). Viral titers peaked at 6 d.p.i. at~5 x 10 6 plaque forming units (PFU)/ear and dropped to close to the limit of detection by 10 d.p.i. Using flow cytometry, we enumerated total CD4 + and CD8 + T cells in ear single cell suspensions over the course of infection (Fig 2B). Few T cells were present in uninfected skin. Recombinant VV infection recruited T cells to the skin beginning at 5 d.p.i., with 2 x 10 5 cells/ear, peaking 4 days later at > 10 6 each of CD4 + and CD8 + T cells per ear. Throughout the course of infection, CD8 + T cells were more prevalent than CD4 + T cells. Unlike total T cell numbers, IL-10gfp + CD8 + T cells peaked soon after initial migration into infected tissue at 6 days p.i. (1.5 x 10 5 cells/ear (Fig 2C)). Notably, IL-10gfp + CD8 + T cell numbers declined within a day to a lower constant level, while IL-10gfp + CD4 + T cells reached peak values at day 6 and remained relatively constant (average 5.8 x10 4 /ear daily).
To determine whether IL-10 + T cells were restricted to the infected tissue, we analyzed the spleen and draining lymph nodes (dLNs) from VV-infected animals (Fig 2D and 2E). We detected almost no IL-10gfp + CD8 + T cells outside of the ear, despite the presence of highly activated CD8 + T cells in the spleen or node, demonstrating that IL-10gfp production is not inextricably linked to VV-induced T cell activation. A small and steady population of IL-10gfp + CD4 + T cells (1 x 10 5 cells) was present in DLNs, likely representing regulatory T cells [30] (Fig 2E).
Thus, at peak, between 10 and 15% of T cells express IL-10 in the skin during VV infection.

VV-elicited IL-10gfp + T cells are highly activated effectors
Remarkably, IL-10 + CD8 + T cells responding to IAV, RSV, or coronavirus infection co-produce a number of pro-inflammatory effector cytokines [12,21]. We examined skin T cells present at 6 days p.i. for expression of gene products associated with activated effector cells. All skin-localized CD8 + T cells expressed T-bet (a transcription factor required for effector T cell differentiation [31,32]), regardless of IL-10 expression (determined by intracellular staining, gray shaded histogram = IL-10cells, black line = IL-10 + cells) (Fig 3A, far left, top panel). In contrast, while IL-10 + CD8 + T cells nearly uniformly expressed high levels of T-bet, total CD4 T cells contained both T-bet high and T-bet low cells ( Fig 3A, far left, bottom panel). While there were clear populations of CD69 high and CD69 low IL-10 -T cells, all IL-10 + T cells expressed high levels of the activation marker CD69 (Fig 3A). In both CD4 + and CD8 + T cells, effector molecules IFN-γ and Granzyme B were markedly higher in IL-10 + cells ( Fig 3B). We did not detect regulatory CD8 + T cells in the ear (S2 Fig). Thus, both IL-10 + CD4 + and CD8 + T cells in VV-infected skin are activated effectors.

IL-10gfp + T cells accumulate around VV keratinocytic foci
We next used intravital MPM to directly visualize the location and movement of IL-10gfp + cells in VV-infected reporter (TIGER) mice. Consistent with ex vivo experiments, we detected  Consistent with flow cytometric results, by 6 d.p.i., numerous IL-10gfp + cells were recruited around VV replicative foci ( Fig 4C, left panels). Outlying areas contained only slightly more IL-10gfp + cells on day 6 p.i. than day 5 p.i. (Fig 4B and 4C). Computer generated maximum intensity projection (MIP) reconstructions clearly locate the bulk of IL-10gfp + cells in the dermis below VV lesions, with a few IL-10gfp + cells managing to enter lesions and interact with BFP + infected cells (Fig 4D and S1 Movie).
To confirm that peri-lesional GFP + cells represent IL-10gfp + T cells detected ex vivo in dissociated skin suspensions, we stained frozen tissue sections taken 6 d.p.i. with anti-CD8 Ab (CD8α, white) ( Fig 4E). Near lesions, we detected high numbers of CD8 + T cells, consistent with our previous description of CD8 + T cell accumulation near viral lesions [34]. Intriguingly, nearly all GFP + cells surrounding viral foci expressed CD8, often in a punctate pattern reminiscent of Ag-engagement-induced TCR clustering [35] (for example, see Fig 4F, yellow arrow).
Taken together, these data show that IL-10gfp + T cells strongly localize at the perimeter of VV-infected keratinocytes.

IL-10gfp + T cells closely associate with viral lesions in the skin
To further understand the localized delivery of IL-10 near viral replicative foci in vivo, we MPM imaged IL-10gfp + cells in reporter mice over time (Fig 5). Using MPM imaging, we previously described the location and mobility of endogenously activated virus-specific T cells in the skin after ec. VV-infected mice [26]. Although all CD8 + T cells that immigrated into infected skin were Ag-specific effectors, many cells did not accrue at the border of lesions but instead occupied areas of the tissue that were not overtly infected as determined by virusdriven fluorescence expression [26]. Since we did not observe this pattern in IL-10gfp + reporter mice (in which GFP-producing cells appeared to predominantly localize to lesions), we next used MPM imaging to compare the distribution of IL-10gfp + cells to adoptively transferred virus-specific T cells endogenously activated in IL-10gfp-reporter mice (Fig 5). We first transferred non-IL-10gfp + dsRed + OT-I CD8 + T cells (antigen specificity = K b -SIINFEKL) into naïve IL-10gfp reporter mice, then infected ec. On day 6 p.i. (rVV expression of BFP is pseudocolored magenta), we detected both IL-10gfp + cells (green) and IL-10gfp -/adoptively transferred OT-I CD8 + T cells (red) at the border of viral lesions (Fig 5A and S2 Movie). We also consistently observed areas of the skin distal from viral keratinocytic lesions with heavy infiltration of dsRed + OT-I CD8 + T cells, but no or few IL-10gfp + cells (Fig 5B and S3 Movie).
Calculating the distance of either cell type to the closest micrograph voxels containing viral fluorescence revealed that although both cell populations accrued near VV-infected foci, the population of IL-10gfp + cells was significantly closer (81.6 ± 9.8 μm vs. 169.6 ± 8.1 μm from lesion border) (Fig 5C and 5D). Further, almost half of IL-10gfp + cells were located within 50 μms of a lesion, whereas only 20% of dsRed + OT-I CD8 + T cells (IL-10gfp -/ -) were present in the same area (Fig 5E and 5F). Thus, unlike the majority of virus-specific CD8 + T cells recruited to the skin post-VV infection, on day 6 post-infection IL-10gfp + cells are largely restricted to areas adjacent to viral lesions.

IL-10gfp + cells are mobile within a localized area
While static images revealed that IL-10-producing cells were located near areas of virus-infected cells, they could be 1) engaged and motionless, 2) motile at the perimeter of infection, or 3) only transiently associated with the borders of lesions. To examine these possibilities, we analyzed the movement of peri-lesional IL-10gfp + cells in relation to dsRed + OT-I CD8 + T cells within the same imaging field (Fig 6). Intriguingly, we noted both sessile and motile IL-10gfp + cells at lesional borders during two-hour imaging sessions (Fig 6A and S4-S6 Movies). Calculating the tracks of each cell's movements during this period revealed that mobile IL-10gfp + cells followed paths near the outer border of viral lesions, remaining proximal to areas of viral infection ( Fig  6B). We did not visualize IL-10gfp + cells migrating away from infected areas during the course of  imaging. An overlay of the tracks of all of the IL-10gfp + cells clearly showed that IL-10gfp + cells exhibited less overall displacement than did dsRed + OT-I CD8 + T cells (Fig 6B, right panels).
We next tracked cellular speeds over the course of MPM imaging. Overall, peri-lesional IL-10gfp + cells moved at slightly lower mean speeds than dsRed + OT-I CD8 + T cells (mean speed of~2.3 vs. 2.8 μm/min) (Fig 6C), though individual cells ranged in speed from nearly sessile to > 4 μm/min. A majority (~70%) of IL-10gfp + cells were motile (defined by a speed > 2.5 μm/min) (Fig 5D). The bi-phasic behavior of cellular movement was confirmed by performing 7 independent experiments (S3 Fig).
Together, these data show that most IL-10gfp + cells are mobile in confined areas around keratinocytic foci of infection.
Localized IL-10 production affects tissue monocyte accumulation at infection sites Monocytes are recruited to and limit viral titers in VV-infected skin [26]. To analyze whether localized IL-10 production could affect the recruitment of these innate immune effectors, we administered anti-IL-10 neutralizing Ab (NAb) to wild-type mice on 5 and 6 d.p.i. This allowed IL-10 neutralization only after T cells had entered the tissue and eliminated possible confounding early effects on innate immunity (as could occur in IL-10 -/animals). On day 7 p.i., we detected a significantly decreased number and percentage of CCR2 + tissue monocytes in the ears of IL-10-neutralized animals (Fig 7A).
We next MPM imaged CCR2 + tissue monocytes in heterozygous knock-in animals expressing RFP (in place of CCR2 protein) [36] (Fig 7B). We chose heterozygote mice to avoid complete deletion of CCR2 + cells (S4 Fig). In uninfected animals, we detected low levels of CCR2rfp + cells in the ear (red, left panels). In contrast, VV infection resulted in dramatic accumulation of CCR2rfp + cells in areas with viral-driven GFP expression (right panels and Fig  7C). We also detected a number of CCR2rfp + cells that were directly infected with VV, as determined by GFP expression (Fig 7D and 7E). Many of these infected monocytes were motile (Fig 7E and S7 Movie). Further, CCR2rfp + /mice had dramatically increased VV titers in the ear compared to wild-type controls (Fig 7F).
Together, these data suggest that locally produced IL-10 enhances the recruitment of CCR2 + monocytes, which localize to areas of viral infection and reduce viral titers in the skin.

Localized IL-10 production inhibits virus spread
To determine whether localized IL-10 production could shape tissue monocyte recruitment and affect VV clearance in the skin, we again administered anti-IL-10 NAb to wild-type mice on 5 and 6 d.p.i. and assessed the effect on virus replication and dissemination in the skin. Confocal microscopic montages of complete ears indicated an increase in VV-encoded fluorescence intensity and larger areas of fluorescence in anti-IL-10 NAb-treated mice (Fig 8A). Closer examination revealed that many of the VV lesions in IL-10-neutralized mice exhibited "satellites" indicative of secondary virus spread [37] (shown in higher magnification in Fig 8B). Quantitating total montage fluorescence revealed a 3-fold increase in VV-expressed GFP signal in IL-10 neutralized animals (Fig 8C). Likewise, the average number of VV lesions per (7x7 mm 2 mosaic) image increased 2.3-fold (4.2 to 9.6) in IL-10-neutralized ears (Fig 8D). Importantly, real time PCR analysis of VV genomes present in the skin showed a concomitant 3-fold increase in virus (1.7 x 10 5 to 5.7 x 10 5 VV genomes/μg DNA) after IL-10 neutralization (Fig 8E).
Despite increased viral load, IL-10 neutralization did not dramatically alter the overall number of CD8 + T cells in the skin (Fig 8F). Intriguingly, neutralization trended toward an increased frequency and number of IL-10gfp + CD8 + T cells in the skin (Fig 8G and 8H). In IL-10-neutralized animals, the enhanced number of IL-10gfp + cells again localized to areas of virus replication (Fig 8I).
Based on these findings, we conclude that locally produced IL-10 limits skin VV-replication and dissemination. This extends the overly simplistic characterization of IL-10 as an "antiinflammatory" cytokine, to include a direct antiviral role in immunity to pathogens.  To begin to address these issues crucial to understanding antiviral immunity, we examined the production of the well-known anti-inflammatory cytokine IL-10 during rVV infectionincluding real time visualization of IL-10-producing cells in infected skin. We uncovered an unlikely primary cellular source of this cytokine: highly activated effector T cells (Figs 1-3) localizing to the immediate vicinity of intense virus replication (Fig 4). Remarkably, although IL-10 production should limit the ability of T cells to kill virus-infected cells, Ab-mediated neutralization of IL-10 actually increased viral titers in the skin.

Optimally resolving acute viral infections is a highly choreographed dance between effector cells that eliminate virus infection and anti-inflammatory cells that limit effector behavior to
Despite intense study, knowledge of all of the cells and immune factors necessary for VV clearance from the skin is still incomplete. After epicutaneous infection, a complex, spatially organized immune response resolves virus infection [34]. While T cells clear infected inflammatory monocytes from the skin, Ly6G + monocytes also eliminate infectious virus. It is likely that many other immune cells types contribute synergistically to viral clearance. Here, we show that CCR2 + monocytes associate closely with virus-infected cells in the skin (even becoming infected (S7 Movie)), and also clearly aid in viral control (Fig 7). Interestingly, nitric oxide has been shown to prevent productive replication of VV in some macrophages [41]. Thus, CCR2 + monocytes could limit virus spread by "sopping up" infectious virus and preventing infection of cell types capable of propagating new virions.
Because IL-10 shapes macrophage populations in the tissues [42][43][44], we neutralized IL-10 and examined the effect on CCR2 + monocytes. Administration of IL-10 neutralizing Ab resulted in an approximately 25% reduction in the number of CCR2 + monocytes in the skin (Fig 7A). Altogether, our data strongly suggest that IL-10 reduces recruitment or maintenance of CCR2 + monocytes around sites of VV replication in the skin, which in turn results in failure to adequately control VV replication and dissemination.
IL-10 could participate in restriction of VV replication in the skin via other mechanisms besides CCR2 + monocyte recruitment. M2-polarized (also known as "alternatively activated" or "wound healing") macrophages promote tissue restoration post-trauma, infection, or allergic inflammation [45][46][47][48]. Because VV has a proclivity for replicating/disseminating in inflamed skin (e.g. in individuals with atopic dermatitis [49,50]), IL-10 could also influence VV replication by reducing VV-induced inflammation. Darling et al. recently used tape stripping and ovalbumin-sensitization to create atopic dermatitis-like skin lesions and then infected mice with VV [51]. IL-10 knockout mice had enhanced viral replication and skin lesions in an IL-17-dependent manner. Thus, IL-10 may also limit VV-replication by suppressing Th 17 responses.
Additionally, IL-10 may directly inhibit VV replication by inducing innate anti-viral pathways in infected cells. An early study characterizing an IL-10-overexpressing rVV revealed no effect on viral replication, however IL-10 may already be saturating in wild-type mice [52]. Intraperitoneal infection of IL-10 -/mice or Ab-mediated IL-10R blockade decreased ovarian VV titers slightly at 7 d.p.i. [53,54]. Byrd and colleagues recently reexamined VV replication in human macrophages (previously described as non-permissive for productive VV replication [55,56]) differentiating them into M1 and M2 polarized phenotypes [57]. VV replicated in both cell types, and importantly, the addition of exogenous IL-10 dramatically decreased VV replication in M2 macrophages in a STAT3-dependent manner. These findings suggest that IL-10 may regulate VV replication by inducing an anti-viral state in selective cell types.
Though IL-10 may influence the antiviral immune response via multiple pathways including via CCR2 + monocytes, our findings demonstrate that the effects of IL-10 are likely to be highly dependent on the location of IL-10 secreting T cells. Interestingly, nearly all IL-10-producing cells in infected reporter mice were amassed at the borders of viral lesions, rather than being more distributed throughout the inflamed, edematous tissue as were many virus-specific T cells (Fig 5). Likewise, most VV-specific IL-10gfp + cells moved in restricted paths around keratinocytic foci of virus replication (Fig 6). This motility may play an important role in evenly distributing IL-10 around active viral lesions.
Due to the limitations of the IL10gfp + reporter system used, we cannot, however, determine the extent to which IL-10 is secreted by circling vs. sessile IL-10gfp + CD8 + T cells. Indeed, prior imaging studies of CD4 + T cells show that effector cytokine production is anti-correlative with movement [58,59]. This uncertainty should be resolved with improved transgenic reporter systems that afford more precise measurement of ongoing IL-10 synthesis. Additionally, further study will be needed to carefully delineate the contributions of IL-10 generated by each T cell subset, as well as by those cells that are only minimally fluorescent in IL-10 reporter mice.
In toto, our findings uncover a unique aspect of local antiviral immunity with clear practical implications. With a critical role for IL-10-producing CD8 + T cells becoming evident for a number of widely different viruses, vaccines aimed at eliciting high numbers of multifunctional CD8 + T cells should consider adding IL-10 to the desired T cell-cytokine repertoire. Likewise, antiviral therapeutics that temporarily boost IL-10 levels in the tissue during acute infections may reduce viral burdens and hasten recovery, particularly if IL-10 can be delivered to the precise anatomical location where it is most effective.
For adoptive T cell transfer, spleens and nodes were removed from OT-I dsRed mice (as described in [26,60]). Cells were purified to > 90% purity using an Automacs (Miltenyi Biotech) and using negative selection for CD8 + T cells. At least one day prior to infection, 1 x 10 5 cells were transferred into IL-10gfp animals.

Ethics statement
All animal experiments were conducted in accordance with the Animal Welfare Act and the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. NIAID animal facilities have full accreditation from the Association for Assessment and Accreditation of Laboratory Animal Care and are PHS-assured (Assurance Number: # A4149-01). All animal procedures were approved by the NIAID Animal Care and Use Committee.

Viruses and infections
Mice were infected ec. in the ear pinnae by using a drop of 1x10 8 PFU/mL rVV stock placed on the ear skin and gently poked a total of 10 times per ear with a bifurcated needle (similar to the human vaccination protocol). VVs used in this study were VV-Ova (full length ovalbumin protein containing SIINFEKL), VV-NP-S-GFP (with the influenza nucleoprotein (NP), SIINFEKL peptide (S), and green fluorescent protein; fluorescence is nuclear), VV-NP-S-BFP, and VV-BFP-Ub-SIINFEKL (cellular ubiquitin hydrolase liberates SIINFEKL peptide from the fusion protein). All have been previously described [26] and virus stocks were grown and titered in house.
For intracellular cytokine staining, brefeldin A (10 μg/ml, Sigma-Aldrich) was added during collagenase digestion, and liberated cells were incubated an additional 3 hours in BFA-containing RPMI (Life Technologies) at 37°C. Cells were stained with 5 μg/mL of ethidium monoazide (EMA) for 15 min in the dark, exposed to light for 10 min to mark dead and dying cells. Cells were subsequently stained for the cell surface markers CD4, CD8, CD69, and CD45, fixed at room temperature for 30 min with eBiosciences FoxP3 Fix/Perm buffer, washed, then stained with anti-IFN-γ, anti-TNF-α, anti-Granzyme B, anti-T-bet and/or anti-FoxP3 in Perm buffer overnight at 4°C. Cells were analyzed on an LSR II flow cytometer (BD Biosciences) and resultant data analyzed using FlowJo (Treestar).

Viral titering via plaque assay
Ears were removed at various times p.i., collagenase digested for 1 hour at 37°C, and disrupted by vigorous pipetting. This suspension was freeze-thawed 3X, sonicated 3X, serially diluted and plated on TK -/cells. Cells were incubated for 2 days before counting the resulting plaques.

qPCR cytokine array
Total RNA was isolated from whole ears using Trizol reagent (Invitrogen) according to the manufacturer's instructions. cDNA was generated using RT 2 first strand kit (Qiagen). Common cytokine PCR array (Qiagen) was used to quantitate cytokine mRNA fold increase over uninfected controls, according to the manufacturer's instructions.

VV genome quantification by qPCR
DNA was isolated from whole ears using the ArchivePure DNA Tissue Kit, (5 Prime, Gaithersburg MD) according to the manufacturer's guidelines. Viral genomes were quantified by realtime PCR using primers specific for VV E3L gene. Primer sequences: 5 0 -GCAGAGATTGTGT GTGCGGCTATT-3 0 and 5 0 -GGTGACAGGGTTAGCATCTTTCCA-3 0 as previously reported [61]. Amplification of E3L and GAPDH were done in parallel using RT 2 SYBR green qPCR Master Mix (SABiosciences). To quantify VV genome copies, a standard curve was generated using DNA from purified VV stock with a known PFU (determined by plaque assay). Viral copies were determined using a standard curve and were normalized to the amount of total input DNA.

Intravital MPM imaging
MPM imaging was performed essentially as described previously [26,62,63]. Briefly, images were acquired on an upright Leica SP5 confocal microscope (Leica Microsystems) equipped with two Mai Tai Ti:Sapphire lasers (Spectra Physics) with 10-Watt pumps. Mice were anesthetized with avertin or isoflurane. Ears were immobilized on an imaging platform and bathed in warm saline; a 20x dipping objective (NA 1.00) was dipped in saline overlying ears for image acquisition. For imaging BFP/collagen/eGFP, imaging was performed in sequential mode with the first laser tuned to 900 nm and the second tuned to 800 nm for BFP excitation. Emitted fluorescence was collected with a four-channel non-descanned detector. For blue/green channels, wavelength separation was accomplished with a dichroic mirror at 495 nm followed by emission filters of 460/50 nm bandpass and 525/50 nm bandpass. For most movies (e.g. when time was critical), images were acquired using 2x zoom, using a 3-μm z-step for a total depth of 150 μm for 1 min intervals between series (needed to switch lasers). For 3D reconstructions, images were acquired at 1x zoom using 1024x1024 resolution and 1 μm z-steps for a total depth of 80-300 μms.

Confocal imaging of frozen sections
Ears were fixed in PLP fixative (periodate-lysine-paraformaldehyde) overnight as reported in [64], cryoprotected in 15% sucrose, embedded in OCT medium (Electron Microscopy Sciences) and frozen in dry-ice cooled isopentane. Sixteen-micron sections were cut on a Leica cryostat (Leica Microsystems), blocked with 5% goat or donkey serum then stained with CD8 (53-6.7), anti-GFP (clone 5F12.4). Sections were incubated with secondary antibodies only as controls, and images were acquired using identical PMT (photomultiplier tube) and laser settings.

Confocal tile scans
Scans were taken of entire infected ears equaling a 7 mm 2 imaged area and individual fields (tiles) were merged into one image. Total image fluorescence was calculated using ImageJ software.

MPM and confocal image analyses
For MPM analysis, maximum intensity projections (MIPs) were processed from z-stacks using Imaris (Bitplane). Because both collagen (second harmonic generation) and BFP appear in the blue channel, BFP expression was pseudocolored magenta. For tracking cellular movement, images were processed using a Gaussian filter, tracks were calculated using the "spot" function of Imaris. Average speeds were calculated using the spot detection function and the following parameters: autoregressive motion, gapclose 1, 7.5 μm object diameter, 20 μm maximum distance. Following automated analyses, tracks were analyzed individually for erroneous connections. Track straightness was calculated according to the standard convention of displacement/ track length.

Statistical analyses
Significances were calculated using GraphPad (Prism) using unpaired two-tailed Student's t test.