The CD8 T cell response to vaccinia virus exhibits site-dependent heterogeneity of functional responses

International Immunology, Vol. 19, No. 6, pp. 733–743doi:10.1093/intimm/dxm039

ª The Japanese Society for Immunology. 2007. All rights reserved.For permissions, please e-mail: [email protected]

The CD8 T cell response to vaccinia virus exhibitssite-dependent heterogeneity of functional responses

Zhengguo Xiao, Julie M. Curtsinger, Martin Prlic, Stephen C. Jameson and Matthew F. Mescher

Center for Immunology and Department of Laboratory Medicine and Pathology, University of Minnesota, 420 DelawareStreet S.E., Minneapolis, MN, USA

Keywords: cell activation, cytotoxic T cells, viral infection

Abstract

CD8 T cell responses to vaccinia virus (VV) and a virus-encoded ovalbumin peptide (OVAP) epitopewere examined using adoptively transferred OT-I T cells. The results demonstrate that upon intra-peritoneal challenge with ovalbumin-expressing VV (VV-OVAP), OT-I T cell proliferation occursinitially in lymph nodes and spleens followed by migration of the divided cells to the peritoneal cavity.Massive clonal expansion occurs in response to both the virus and the virus-encoded ovalbumin(OVA) epitope, as demonstrated using low numbers of adoptively transferred cells, and theresponding OT-I cells display marked site-dependent functional heterogeneity with respect to IFN-gand tumor necrosis factor-a (TNF-a) production and granzyme B expression. OT-I cells responding toVV-OVAP develop the capacity to produce IFN-g in response to antigen as they proliferate anddifferentiate. In marked contrast, naive OT-I cells rapidly produce TNF-a upon antigen recognition,and this capacity declines as the cells proliferate in response to the virus, suggesting that this potentinflammatory cytokine may be important primarily during initiation of the response. At the peak ofclonal expansion, a large fraction (30–60%) of the OT-I cells responding to the virus express highIL-7Ra levels, and the majority of these cells is subsequently lost. While high IL-7Ra expression maybe necessary for a CD8 T cell to transition to memory, it is clearly not sufficient. Thus, OT-I cellsresponding to VV infection exhibit a high degree of heterogeneity within the responding populationthat differs depending on their anatomical location, despite the specificity and affinity of the TCRbeing identical on all of the cells.

Introduction

Vaccinia virus (VV), a large DNA virus, is closely related tothe other orthopoxviruses including variola virus, the causeof smallpox. Immunization with VV protects against variolavirus infection (1), and use of VV as a vaccine has playeda key role in the eradication of smallpox. Infection with VVinduces strong CD8 T cell responses in Balb/c (2) andC57BL/6 mice (3), but the CD8 T cell response is not re-quired for protection in normal mice. The absence ofperforin- or Fas-dependent cytotoxicity (4), or the almostcomplete absence of CD8 T cells in B2m�/� mice (5), donot affect viral clearance or survival. In a more detailedstudy of the basis for protective immunity, Xu et al. (3) re-cently showed that CD4 T cell-dependent antibody produc-tion plays the major role in viral clearance following acuteinfection, but that the CD8 T cell response could mediateprotection in the absence of an antibody response. Thisstudy also demonstrated that VV-specific memory CD8T cells could protect against secondary infection.

In addition to providing an important vaccine for protec-tion against smallpox, VV has been widely used as an ex-pression vector for foreign genes and as a vaccine for otherinfectious diseases and cancers (6). Despite this, CD8 T cellresponses to VV and epitopes expressed by recombinantVVs have not been extensively studied. Harrington et al. (2)characterized the time course for the CD8 T cell response inthe spleen to a recombinant VV and showed that there wasa coordinate response to the VV vector and the foreign epi-tope, and that the response was comparable to that inducedby lymphocyte choriomeningitis virus (LCMV) infection. Theyfurther showed that the cells responding to the foreign epi-tope developed potent function, as assessed by productionof effector cytokines, and established a long-term, respon-sive memory population.Virus-specific CD8 T cells respond to viral infection by un-

dergoing multiple rounds of cell division leading to clonal ex-pansion, and during this time the naive cells differentiate to

Correspondence to: M. F. Mescher; E-mail: [email protected] Received 3 November 2006, accepted 7 March 2007

Transmitting editor: M. J. Bevan Advance Access publication 1 June 2007

by guest on January 11, 2016http://intim

m.oxfordjournals.org/

Dow

nloaded from

http://intimm.oxfordjournals.org/

develop effector functions, including cytolytic activity andthe ability to rapidly produce effector cytokines includingIFN-c and tumor necrosis factor-a (TNF-a) upon encounterwith antigen. A number of studies have shown that there isconsiderable heterogeneity in the effector functionsexpressed by CD8 T cells responding to various viruses(7–9) both within the population from a given anatomical siteand between populations from different sites. A number offactors can potentially influence development of function, in-cluding levels of antigen and co-stimulation and affinities ofTCRs on the responding cells. In addition, there is now con-siderable evidence that antigen and co-stimulation are notsufficient to drive differentiation and acquisition of effectorfunctions, and that a ‘third’ signal is required that can beprovided by either IL-12 (10, 11) or Type I IFNs (12). Thus,a number of factors may play a role in determining the effec-tor competence of developing CTL.In order to further investigate the functional heterogeneity

that arises during virus-specific CTL responses and charac-terize in more detail the parameters of CD8 T cell responsesto VV as an expression vector, we have examined theresponses of endogenous CD8 T cells and adoptively trans-ferred TCR transgenic OT-I CD8 Tcells to ovalbumin-expressingVV (VV-OVAP). Use of an adoptive transfer approachmakes it possible to examine the earliest phase of the re-sponse, before substantial clonal expansion has occurred,and assures that variations in functional status do not resultfrom differences in TCR affinities, as might be the case whenpolyclonal responders are examined. The results demon-strate that despite identical TCRs, OT-I cells responding toVV-OVAP exhibit marked site-dependent heterogeneity withrespect to their expression of granzyme B (grzB) and theirability to produce IFN-c and TNF-a. In addition, a large frac-tion of the OT-I cells present at the peak of the response ex-press high levels of the IL-7Ra chain, but this is not sufficientto insure that they survive to become long-lived memory cells.

Methods

Mice, virus and reagents

OT-I mice having a transgenic TCR specific for H-2Kb andOVA257–264 (ovalbumin) (13) were a gift from F. Carbone(University of Melbourne, Melbourne, Australia). OT-I micewere also crossed with Thy1-congenic B6.PL-Thy1a/Cy(Thy1.1) mice (Jackson ImmunoResearch Laboratories) andbred to homozygosity. The OT-I and OT-I/PL breeding colo-nies were maintained under specific pathogen-free condi-tions at the University of Minnesota. C57BL/6NCr mice werepurchased from the National Cancer Institute. All directlyconjugated fluorescent antibodies were purchased from BDBiosciences or eBioscience. Recombinant VV-GFP-JAW-OVA (VV-OVAP) was provided by J. Yewdell, National Insti-tutes of Health, Bethesda, MD, USA. This recombinant virusincludes the OVA257–264 epitope fused C-terminally to GFPand the transmembrane region of JAW-1. The Western Re-serve vaccinia strain (VV-WR) was from M. Bevan (Universityof Washington, Seattle, WA, USA). Mice were infected byintra-peritoneal (i.p.) injection of 5 3 106 PFU of the indicatedvirus, with the viral titer determined by plaque assays per-

formed using 143B cells. These studies were reviewed andapproved by the Institutional Animal Care and Use Commit-tee of the University of Minnesota.

Naive T cell purification

Inguinal, axillary, brachial, cervical and mesenteric lymphnodes were harvested from OT-I/PL mice, pooled and disrup-ted to obtain a single-cell suspension. Cells were enrichedfor CD8+ CD44low cells by negative selection using MACSmagnetic cells sorting (Miltenyi Biotec). In brief, cells werecoated with FITC-labeled antibodies specific for CD4, B220,I-Ab, CD11c and CD44. Anti-FITC magnetic MicroBeads(Miltenyi Biotech) were added to the cells, which were thenpassed over separation columns attached to the MACSmagnet. The cells that did not bind to the column werecollected and were >95% CD8+ and <0.5% CD44high.

Adoptive transfer of OT-I/PL transgenic cells

A total of either 1.5 3 106 or 5 3 103 purified naive CD8+cells from OT-1/PL mice in 0.3 ml PBS were transferred viatail vein injection into age- and sex-matched naive 6- to 8-week old C57BL/6 recipients. Recipient mice were restedfor 24 h before virus inoculation. In some experiments, OT-I/PL cells were labeled with CFSE prior to adoptive transfer.Purified cells were re-suspended to 20 3 106 ml�1 in HBSSand warmed in a 37�C water bath for 10 min before addingan equal volume of 6 lM CFSE in HBSS and incubating foran additional 5 min at 37�C. Cells were washed once withice-cold RP-10 and twice with ice-cold PBS before adoptivetransfer.

Flow cytometric analysis of OT-I cells

Mice were sacrificed at the indicated times after adoptivetransfer and viral infection. Spleen cells, peritoneal cells andlymph node (LN) cells (pooled from axillary, brachial, cervi-cal, inguinal and mesenteric nodes) were harvested,counted by trypan blue dye exclusion to determine total via-ble cell counts and stained with antibodies to CD8 and Thy1.1 to detect the transferred OT-I/PL cells. Stained cells wereanalyzed on a FACSCaliburTM flow cytometer using CELL-QuestTM software (BD Biosciences) to determine the percentand total OT-I/PL cells in the mice.

Intracellular staining for cytokine production and grzBexpression

Spleen cells, peritoneal cells and LN cells were incubated at2 3 106 cells ml�1 in RP-10 with 0.2 lM OVA257–264 and 1 llGolgiPlug (BD Biosciences) for 3.5 h at 37�C, and thenwashed and stained with the antibodies to CD8 and Thy 1.1to mark the OT-I/PL cells. The cells were then fixed in Cytofixbuffer (BD Biosciences) for 15 min at 4�C, permeabilized insaponin-containing Perm/Wash buffer (BD Biosciences) for15 min at 4�C and stained with FITC–anti-IFN-c mAb andPE–anti-TNF-a mAb for 30 min at 4�C. Following staining,cells were washed once with Perm/Wash buffer and oncewith PBS containing 2% FBS and analyzed on a FACSCali-burTM flow cytometer using CELLQuestTM software. GrzB in-tracellular staining was done directly ex vivo without peptidere-stimulation.

734 CD8 T cell response to vaccinia virus



Dow

nloaded from


Results

Clonal expansion in response to infection

When naive TCR transgenic OT-I T cells specific for H-2Kb/OVA257–264 (13) are adoptively transferred into C57BL/6mice, they persist for weeks as resting naive cells in theLNs and spleens. Upon infection of the mice by i.p. inocula-tion with recombinant VV that expresses the OVA257–264 epi-tope (VV-OVAP), the OT-I cells undergo rapid clonalexpansion (Fig. 1A). Expansion is greatest in the spleen,with OT-I numbers peaking by day 5. Numbers then rapidlydecline over the next few days, and a stable population ofcells having a memory phenotype (data not shown) is pres-ent by day 30. Clonal expansion is seen in the LN with a sim-ilar time course, but in much smaller numbers than in thespleen. Few, if any, naive cells are detectable in the perito-neal cavity (PC) prior to infection, but activated OT-I cellsbecome detectable at this site by day 3, peak in number atday 5 and then decline (Fig. 1A).Concomitant with expansion of the OT-I cells, there is

a large increase in the number of endogenous CD8 T cellsin the spleen during days 5 through 7 with numbers thenrapidly declining to normal resting levels (Fig. 1B). Endoge-nous CD8 T cells also increase in number in the PC duringdays 5 through 7, but no significant increase is seen in LN.Further characterization of these endogenous cells is de-scribed below. When mice received OT-I cells by transferand were then challenged with the Western Reserve vacciniastrain (VV-WR) that does not express the OVA epitope, theOT-I cells did not increase in number and retained a naivephenotype (data not shown). In contrast, endogenous CD8Tcells in the spleen and PC increased in number comparablyto when infection was with VV-OVAP (data not shown). Virusis rapidly cleared in the male mice used in these experi-ments, and is undetectable in the PC or other sites when pla-que-forming assays are done on day 3 or 5 (data not shown).OT-I cells labeled with CFSE were found to be largely un-

divided in the LN and spleen the first day after infection(Fig. 2). By day 2, however, the cells had undergone severaldivisions and by day 3 the CFSE was fully diluted, indicatingthat the cells had divided at least seven to eight times. Theextent of CFSE dilution on day 2 was consistently greater inthe spleen than in the LN. When OT-I cells appear in the PCby day 5, they have fully diluted CFSE (data not shown), in-dicating that they migrate to this site only after undergoingmultiple rounds of division. The rapid onset of cell divisionin lymphoid tissue, together with the fact that division ofCD8 T cells does not commence until ;24 h after initial rec-ognition of antigen (14), suggests that antigen rapidlybecomes available to most of the naive CD8 T cells followingonset of infection.

Functional heterogeneity of responding CD8 T cells

Major effector functions of activated CTL include direct lysisof virus-infected cells and production of cytokines includingTNF-a and IFN-c. The capacity of OT-I cells to produceTNF-a and IFN-c can be assessed by intracellular cytokinestaining following stimulation for 3.5 h with OVA257–264 pep-tide, and levels of intracellular grzB expression correlate

strongly with cytolytic activity of the cells (15). When effectorcells responding to VV-OVAP infection were examined onday 5, there was substantial heterogeneity in expression ofthese proteins depending on the location of the cells. OT-Icells from the PC, the primary site of infection, expressedthe highest levels of grzB and produced the highest levelsof IFN-c and TNF-a upon in vitro re-stimulation, as assessedby both the percent of cells expressing the protein and thelevel of expression on a per cell basis (Fig. 3A). Expressionof these proteins was substantially lower in OT-I cells in thespleen and lowest in cells from the LN (Fig. 3A).By day 32, when numbers have declined and a memory

population has been established, there was less heterogeneity

Fig. 1. Clonal expansion of OT-I and endogenous CD8 T cells toinfection with VV-OVAP. B6 mice received 1.5 3 106 naive OT-1 byadoptive transfer, were rested for one day and were then inoculatedi.p. with 53 106 PFU VV-OVAP (day 0). At each time point, the PC waswashed with 10 ml RPMI medium, and peripheral lymph nodes andspleen (SP) were then harvested. Transferred OT-I and endogenousCD8 T cells were identified by CD8 and Thy 1.1 as described inMethods. The total number of cells recovered from each location isshown. Each of the points represents the average of three or moremice and the bars indicate the standard error. (A) OT-I CD8 T cells.Mice that received OT-I cells by adoptive transfer and were notinfected (i.e. equivalent to day 0) had;1.23 104–1.53 104 OT-I cellsin the LN, 5 3 104 OT-I cells in the spleen and an undetectablenumber of OT-I cells in the PC, and there was a small decline in thesenumbers over the course of 15 days. (B) Endogenous CD8 T cells.

CD8 T cell response to vaccinia virus 735



Dow

nloaded from


of function for OT-I cells from the different sites (Fig. 3B). Thesmall number of cells that could be recovered from the PC,however, continued to express somewhat higher levels ofgrzB and produce more TNF-a than those from other sites.Memory cells at all sites had substantially higher capacity toproduce IFN-c and TNF-a than did naive cells or day 5effector cells, while grzB expression was decreased. Thus,long-term memory cells all have the capacity to rapidly pro-duce high levels of effector cytokines upon reencounter withantigen.Examination of the time course of expression of the pro-

teins important for function showed that differences weregreatest at day 5, the peak of clonal expansion, when con-sidered as a percent of the OT-I cells that produced IFN-c(Fig. 4A) and TNF-a (Fig. 4B) or expressed grzB (Fig. 4C).Measured in this way, OT-I cells in the PC clearly have thegreatest functional capacity during the effector phase of theresponse. Total functional capacity, however, presumablydepends on both the capacity per cell and the number ofcells at the site. When examined in this way, a different pic-ture emerges. Thus, by far the greatest number of cells withthe capacity to produce IFN-c (Fig. 4A) and TNF-a (Fig. 4B)and express grzB (Fig. 4C) are present in the spleen duringthe effector phase of the response.In the experiments described above, adoptive transfer

was done using 1.5 3 106 OT-I cells per mouse so that therewere sufficient cells to allow detection and characterizationat the earliest times in the response, before clonal expansionhad occurred. This clearly represents a much higher fre-quency of antigen-specific cells than is normally present. Todetermine if functional heterogeneity also occurs with pre-cursor frequencies in a more physiological range, experi-ments were done that employed transfer of 5 3 103 OT-IT cells. At this level, naive OT-I cells cannot be detected inthe recipient mice prior to infection. Despite this, the numberof effector OT-I cells present at day 5, the peak of theresponse, was comparable to that seen when 1.5 3 106

cells were transferred (Fig. 5A). For both input levels, thenumbers declined comparably by day 7. The OT-I cells thatexpanded from a low precursor frequency also showedfunctional heterogeneity, with the cells from the PC havingthe greatest capacity to produce IFN-c and TNF-a inresponse to antigen and the highest grzB expression(Fig. 5B).Thus, development of functional heterogeneity occurred at

both high and low transfer numbers. However, clonal expan-sion was clearly very limited at high precursor numbers. Inthe experiment shown in Fig. 5A, ;105 naive cells werepresent in the spleen prior to infection, and 17 3 106 at thepeak of the response, for an expansion of 170-fold. For lowtransfer numbers, OT-I cells could not be reliably detectedprior to infection, while 13 3 106 OT-I cells were present atthe peak of the response. Even if all the transferred cells(5 3 103) were in the spleen prior to infection, this repre-sents an expansion of at least 2600-fold. Thus, while transferof high numbers allows assessment of the earliest activationevents (Fig. 2) and exhibits comparable functional heteroge-neity, it results in a large underestimate of the potential forclonal expansion. The extent of clonal expansion seen atlow transfer numbers for VV-OVAP is similar to that reported

Fig. 2. Antigen-specific proliferation of OT-I Tcells in different tissues.Purified naive OT-1 cells (1.5 3 106) were labeled with CFSE andtransferred into B6 mice. One day later the recipient mice wereinoculated with VV-OVAP (day 0). On days 1, 2 and 3, cells wereisolated from peripheral lymph nodes, spleen and PC, and the CFSEcontent of the OT-I T cells determined by flow cytometry. Each of thepanel is representative of the results obtained for three miceexamined on each day. (A) CFSE profile for OT-I cells from LNs ofuninfected control mice. (B) CFSE dilution for OT-I cells from LNs ofinfected mice. (C) CFSE dilution for OT-I cells from spleens of infectedmice.




Dow

nloaded from


by Blattman et al. (16) for antigen-specific CD8 T cellsresponding to LCMV.

Response of endogenous CD8 T cells to VV-OVAP

As shown in Fig. 1(B), the number of endogenous CD8T cells increases greatly in the spleen, and to a lesser ex-tent, in the PC, upon VV-OVAP infection. Total CD8 T cells in-crease from the normal level of ;15 3 106–20 3 106 per

spleen to ;50 3 106–60 3 106 on days 5 through 7 of infec-tion. A large fraction of these appear to be cells specificallyresponding to VV-OVAP based on expression levels of grzB.Endogenous CD8 T cells from control mice have almost nodetectable grzB-positive cells, while grzB high populationscan be readily detected in the spleen, PC and LN of VV-OVAP-infected mice on day 5 after infection (Fig. 6A). Byfar, the greatest numbers of grzB high cells are present inthe spleen at the peak of the response on days 5 and 7, with

Fig. 3. Heterogeneity of IFN-c and TNF-a production and grzB expression by OT-I cells in different tissues. B6 mice received naive OT-1 cells byadoptive transfer (1.5 3 106 per mouse), were rested 1 day and were then infected with VV-OVAP. Cells from PC, LN and spleen (SP) wereisolated from groups of mice on days 5 (A) and 32 (B). Aliquots of cells were stained directly ex vivo for grzB expression, and following in vitrostimulation for 3.5 h with SIINFEKL peptide for IFN-c and TNF-a. Histograms shown are for OT-I Tcells, and are representative of the results for thethree mice per group for day 5 and four mice per group for day 32. The experiment was repeated twice with essentially the same results.




Dow

nloaded from


fewer numbers in the PC and LN (Fig. 6B). On day 5, ;40%of the CD8 T cells in the spleen are grzB positive (Fig. 6Aand C), indicating that the increase in total CD8 T cells inthe spleen at this time is largely accounted for by cellsresponding to the virus. The PC, however, has the highestpercent of grzB high cells as a percent of the total CD8T cells present at the site, and the majority is grzB high byday 3 (Fig. 6C). As is the case for OT-I cells (Fig. 4C), grzBhigh cells have declined precipitously at all sites by day14 and beyond, but a small population of grzB-positive cellsremains in the PC even at day 32 (Fig. 6C).

Expression of IL-7Ra chain

IL-7Ra chain expression is rapidly down-regulated on CD8T cells when they respond to antigen (17–19). When IL-7Raexpression was examined on OT-I cells responding to VV-OVAP infection, levels were found to be slightly reduced onday 3 in comparison to naive cells, significantly down-regulated on day 5 and had returned to naive levels by day7, and this was the case for cells in the LN, spleen and PC(Fig. 7A). Consistent with the evidence from grzB expressionindicating that a large fraction of the endogenous CD8 cellsin the spleen on day 5 are responding to the virus, a largefraction of these cells had also down-regulated expressionof IL-7Ra in comparison to levels on CD8 T cells in unin-fected mice (data not shown).It was previously shown that at the peak of the response

to LCMV, the majority of virus-specific CD8 T cells hasdown-regulated IL-7Ra expression, with only ;5–15% of thecells expressing high levels (19). It was further shown thatthese IL-7Ra high cells survived better than the IL-7Ra lowcells upon transfer into uninfected recipients, leading to thesuggestion that expression of IL-7Ra marks the respondingcells that are destined to become memory cells. In contrastto the LCMV model (19), >50% of the OT-I cells that haveclonally expanded in response to VV-OVAP have highIL-7Ra expression at the peak of the response, and the pro-portion of IL-7Ra high cells increases at longer times (Fig. 7B).This is also the case for cells in the LN and PC, although thenumbers of cells present at these sites is much lower (Fig.1A). When the total number of IL-7Ra high cells present dur-ing the course of the response is determined (Fig. 7B), it isclear that high expression of this receptor does not insure sur-vival of the cells to become memory cells. Thus, while highIL-7Ra expression may be required for survival and transitionto memory, it is clearly not sufficient.

Discussion

Use of adoptively transferred Kb/OVA-specific OT-I T cellshas allowed us to determine the time course for CD8 T cellresponses to i.p. infection with VV-OVAP with respect to sites

Fig. 4. Time course for heterogeneity of grzB expression andcapacity for production of IFN-c and TNF-a in response to VV-OVAPinfection. B6 mice received naive OT-I Tcells by adoptive transfer (1.53 106 per mouse), were rested 1 day and were then inoculated withVV-OVAP. Aliquots of cells recovered at the indicated sites and timeswere stained directly ex vivo for grzB expression (C), and following

in vitro stimulation for 3.5 h with SIINFEKL peptide for IFN-c (A) andTNF-a (B). Each of the values is an average for the OT-I cells from threemice per group, and bars indicate standard error. Histograms (left)show the gating used to determine positive cells. Upper right panelsshow the positive cells as a percent of total OT-I cells and lower rightpanels show the absolute numbers of positive cells recovered.




Dow

nloaded from


of proliferation and clonal expansion, extent of expansion,development of effector functions at various sites and ex-pression levels of IL-7Ra on the cells. Despite the fact thatthe primary site of infection is the PC, the OT-I cells first pro-liferate in response to antigen in the spleens and draininglymph nodes (DLNs), with several rounds of cell division al-ready occurring by day 2 after infection (Fig. 2). By day 3,CFSE is fully diluted, indicating that the cells have under-gone at least seven to eight rounds of division, and signifi-cant clonal expansion is detectable at this time (Fig. 1A).Although not detectable with CFSE, cell division must con-tinue since clonal expansion continues with the numbers ofcells peaking at day 5 and then declining (Fig. 1A). OT-I

cells only become detectable in the PC by day 4 or 5(Fig. 1A), when they have fully diluted their CFSE (data notshown). Thus, it appears that the OT-I cells only migrateto the primary site of infection after undergoing multiplerounds of division in the DLN and spleen. This is in markedcontrast to OT-I adoptively transferred mice challenged byi.p. injection of OVA-expressing E.G7 tumor, where the initial

Fig. 6. GrzB expression in endogenous CD8 T cells during VV-OVAPinfection. B6 mice received naive OT-I Tcells by adoptive transfer (1.53 106 per mouse), were rested 1 day and were then inoculated withVV-OVAP (day 0). Cells were stained directly ex vivo for grzBexpression, and endogenous CD8 T cells were defined by CD8 andThy 1.1 staining on live-gated cells. (A) Representative dot plotsshowing grzB expression by CD8 T cells from control and day5 infected mice. (B) Total numbers of grzB expressing endogenousCD8 Tcells. (C) Percent of endogenous CD8 Tcells expressing grzB.Values in (B) and (C) are averages from three or more mice, and barsindicated standard error.

Fig. 5. Clonal expansion and expression of IFN-c, TNF-a and grzB atdiffering numbers of transferred OT-I cells. B6 mice received naiveOT-I T cells by adoptive transfer at either 1.5 3 106 per mouse or 5 3103 per mouse, were rested 1 day and were then inoculated with VV-OVAP (day 0). (A) The number of OT-I cells recovered in spleens onthe indicated day after infection. Values shown are averages for twomice per group at 1.5 3 106 OT-I input and three mice per group at5 3 103 OT-I input. (B) Aliquots of cells recovered from spleens onday 5 were stained directly ex vivo for grzB expression, and followingin vitro stimulation for 3.5 h with SIINFEKL peptide for IFN-c and TNF-a.Histograms shown are representative of results obtained from the allmice in each group.




Dow

nloaded from


clonal expansion occurs in the PC, and not in the spleen orDLN (20).Transfer of high numbers of OT-I cells (1.5 3 106) is neces-

sary to examine the earliest events that occur during the re-sponse, including sites of proliferation (Fig. 2) and initialclonal expansion (Fig. 1), but underestimates the extent ofclonal expansion that can occur. When a low number of OT-Icells is transferred, 5 3 103 cells per mouse, the transferredcells cannot be reliably detected prior to infection, but chal-lenge with VV-OVAP yields almost as many OT-I cells in thespleen, DLN and PC at the peak of the response as when1.5 3 106 cells are transferred (Fig. 5A). If it is assumed thatall the 5 3 103 transferred cells survive and can respond,then the cells expand ;2500-fold in response to the virus.This is almost certainly an underestimate of the expansion,since at higher transfer numbers only ;10% of the input OT-Icells can be found in the spleens and LNs a day or two laterat the time of infection. The much lower expansion that occursat high transfer numbers (10- to 20-fold) is probably due tocompetition for antigen or ‘space’ limiting the response.The response of endogenous CD8 T cells to the virus par-

allels that seen for OT-I cells, with expansion peaking at day5–7 and occurring predominantly in the spleen (Fig. 1B)

and with CD8 T cells increasing in number in the PC at thistime. There is a large increase in the number of CD8 T cellsin the spleen at the peak of the response, with 2- to 3-foldgreater numbers than in spleens of uninfected mice(Fig. 1B). About 40% of the CD8 T cells present at the peakof the response express high grzB levels (Fig. 6), and aneven higher percentage of the CD8 T cells in the PC expresshigh grzB (Fig. 6C). GrzB-expressing cells were also presentin the DLN at the peak of the response, but in much lowernumbers than in the spleen. Thus, there is rapid, massiveexpansion of both the virus-specific cells [Fig. 1 and (2, 3)]and the cells responding to the foreign ovalbumin peptide(OVAP) encoded by the virus. These results agree well withthose of Harrington et al. (2) who found that ;30% of theCD8 cells in the spleens of i.p. infected mice on day 7, thepeak of the response, made IFN-c in response to stimulationwith VV-infected cells. Similarly, Xu et al. (3) found that 22%of the cells in the spleens of infected mice produced IFN-cwhen re-stimulated on day 7 with an infected cell line. Thesestudies did not examine responding CD8 T cells at othersites or levels of grzB expression. The use of adoptivelytransferred OT-I T cells made it possible to examine site-dependent functional heterogeneity in our studies.

Fig. 7. IL-7Ra expression on OT-I Tcells responding to VV-OVAP infection. B6 mice received naive OT-I Tcells by adoptive transfer (1.53 106 permouse), were rested 1 day and were then inoculated with VV-OVAP (day 0). Cells were stained directly ex vivo for IL-7Ra expression. (A)Representative histograms showing IL-7Ra expression on OT-I cells from LN (top), spleen (middle) and PC (bottom) on days 3, 5 and 7. Ascontrols, naive OT-I cells from adoptively transferred, uninfected mice are shown. (B) Histograms (left) show the gating used to determine IL-7Ra-positive cells. The upper right panel shows the absolute numbers of positive cells recovered and the lower right panel the positive cells asa percent of total OT-I cells. Each of the values is an average for the OT-I cells from three mice per group, and bars indicate standard error.




Dow

nloaded from


The OT-I cells responding to VV-OVAP exhibited consider-able heterogeneity with respect to functional capacity, bothat the level of the individual cells within the population fromeach anatomical site and between the populations at differ-ent sites. Few, if any, naive OT-I cells from mice that wereadoptively transferred but not infected produced IFN-c uponstimulation with OVAP (Fig. 4A), but they developed this ca-pacity within 3 days of infection. Despite all of the cells hav-ing proliferated in response to antigen (Fig. 1), however,only a fraction of the cells made IFN-c (Figs 3, 4A and 5B).At all times, the population recovered from the PC, the pri-mary site of infection, had the largest fraction of IFN-c-producing cells. However, although a smaller fraction of OT-Iin the spleen made IFN-c, the much larger number of cellsin the spleen made this the site of most of the IFN-c-produc-ing cells. The memory cells remaining 32 days after infectionhad the highest capacity for rapid IFN-c production upon re-stimulation (Figs 3B and 4), and ;75% of the cells from allsites produced IFN-c. Even at this stage, however, the smallnumber of memory cells present in the PC had greater func-tion as measured by the amount of IFN-c they produced ona per cell basis.GrzB expression was similar to that of IFN-c in that naive

cells expressed little or none, while a large fraction of thecells expressed high levels by day 3. Here too, expressionwas highest in cells from the PC, slightly lower in spleenand significantly lower in LN (Figs 3A, 4C and 5B). In con-trast to IFN-c production, however, the fraction of cells thatexpressed grzB declined steadily at all sites beyond day3 and expression was very low in memory cells by day 32.However, a significant fraction of the small number of cellspresent in the PC by day 32 still expressed some grzB(Figs 3B and 4C). GrzB expression correlates strongly withthe ability to kill antigen-expressing cells (15), suggestingthat cytolytic activity is greatest on a per cell basis early inthe response and declines rapidly as antigen is cleared. Ini-tial proliferation of OT-I cells occurs in both LN and spleen(Fig. 2), where cells may be responding to either antigen-bearing dendritic cell (DC) that have migrated from the PCor antigen presented by resident DC as a result of virus dis-seminating from the PC. The site of initial expansion may in-fluence the functional heterogeneity of the resulting effectorcells, but interpretation of this is complicated by the fact thatthe activated cells will migrate to other sites. Thus, it is likelythat at later times, OT-I cells present in the spleen will in-clude cells that initially expanded in the spleen, as well ascells that initially expanded in LN, and vice versa.Development of most CD8 T cell effector functions, includ-

ing cytolytic activity and capacity to produce IFN-c, requiresproliferation and differentiation subsequent to antigen en-counter. However, this is not the case for production of TNF-a,a potent inflammatory cytokine (21, 22). Brehm et al. (23) haverecently shown that naive CD8 T cells can produce TNF-awithin a few hours of TCR engagement, and obtained evi-dence to suggest that this may influence maturation of anti-gen-presenting DC. We similarly found that a substantialfraction of naive OT-I cells from mice that were adoptivelytransferred but not infected were able to produce TNF-awithin hours of stimulation with OVAP (Fig. 4B). In addition,the capacity to produce TNF-a decreased significantly in the

effector cells at the peak of response to VV-OVAP, particularlyin the cells in the spleen and LN (Figs 3 and 4B). This is inmarked contrast to grzB expression and capacity to produceIFN-c, which are low or absent in naive cells and high in ef-fector cells (Fig. 4A and C). The memory OT-I cells present32 days after virus infection again show rapid and high TNF-a production upon stimulation with antigen (Figs 3A and 4B).These results suggest that TNF-a production may have an im-portant role early in a response, possibly involved in modify-ing antigen-presenting cell functions as suggested by Brehmet al. (23), but that this capacity is reduced in effector cells toavoid the immunopathology that can be induced by TNF-a(24, 25). Although the pattern of TNF-a expression in re-sponse to VV differs from that of IFN-c and grzB, it also exhib-its heterogeneity depending on the site examined, with thecapacity to produce TNF-a being highest in cells recoveredfrom the PC in comparison to those from spleen and LN(Fig. 3).Site-dependent heterogeneity of cytokine production by

CD8 T cells responding to viral infections has been reportedfor other viruses, including LCMV and vesicular stomatitis vi-rus (VSV) (7) as well as influenza virus (8) and hepatitis Bvirus (9). The basis for this heterogeneity has remained un-clear, and is likely to be influenced by a number of factors.Differences in TCR affinities in a responding polyclonal pop-ulation could potentially contribute to the heterogeneity indevelopment of effector functions but this is clearly not theonly factor, as demonstrated by the marked heterogeneity ofthe responses of TCR transgenic OT-I T cells responding toVV-OVAP (Figs 3 and 4). Kristensen et al. (7) reached a simi-lar conclusion upon examining responses of TCR transgenicCD8 T cells to LCMV and VSV infections.Differentiation of naive CD8 T cells to acquire effector

functions, including grzB expression and IFN-c production,requires, in addition to antigen and co-stimulation, a thirdsignal that can be provided by IL-12 (10, 11) or Type I IFN(12), and possibly other cytokines. In the case of LCMV in-fection, the CD8 T cell response to the virus is almost com-pletely eliminated if the CD8 T cells lack the Type I IFNreceptor (26, 27). In contrast, the response to VV was con-siderably less dependent on the Type I IFN receptor, sug-gesting that an alternative third signal was also available tosupport the response (27). Optimal development of effectorfunctions requires prolonged exposure of CD8 T cells to thesignal three cytokine during their interaction with antigen(28). Thus, functional heterogeneity may arise in vivo as a re-sult of some cells receiving adequate levels and duration ofboth antigen and signal three cytokine, while other cells re-ceive sub-optimal signals, which may be particularly thecase for CD8 T cells recruited into the response at latertimes when antigen levels and inflammation are waning.Site-dependent heterogeneity of the response could arise ifdiffering antigen and cytokine levels are available at thesites during differentiation of the cells. Alternatively, thosecells that receive optimal signals for full differentiation mayalso be the cells that traffic most effectively from the DLNand spleen to peripheral sites. When optimal signals arepresent, CD8 Tcells differentiate to become grzB-expressing,IFN-c-producing effector cells within ;3 days. Theymay retain plasticity with respect to expression of these




Dow

nloaded from


functions, however, so that signals received at sites they mi-grate into can change their functional phenotype. There is ev-idence for both CD4 (29) and CD8 (30) memory T cells thatthe tissue microenvironment can alter the phenotype of thecells.Naive CD8 T cells express high levels of IL-7 receptor on

their surface and rapidly lose IL-7Ra chain expression asthey respond to antigen. During a response to LCMV, only5–15% of the cells express IL-7Ra at high levels, and Kaechet al. (19) demonstrated that these are the cells capable offorming a long-lived memory population. These results sug-gested that the subset of cells that either never lost IL-7Raexpression or regained expression identified the memoryprecursors. Lacombe et al. (31) subsequently showed ina peptide immunization model that a large fraction of the ef-fector CD8 T cells at the peak of the response expressedhigh IL-7Ra, but the majority did not survive to become mem-ory cells. The same is true for the OT-I response to VV-OVAP.At the peak of the response ;40% of the cells express highIL-7Ra levels, but the numbers decline rapidly over the nextweek and only a small fraction of persist long term as mem-ory cells (Fig. 7). Thus, while expression of IL-7Ra may berequired for a cell to transition to memory, it is not sufficient.Although a CD8 T cell response is not essential for clear-

ance of a VV infection (3, 5), our results demonstrate thata rapid and massive response to both viral epitopes anda foreign protein encoded by the virus does occur, and thatthe effector cells at various sites exhibit a high degree offunctional heterogeneity. A long-lived population of memorycells results from the response, and primed CD8 T cells canmediate protection against the re-infection (3). This informa-tion should contribute to the development of immunizationstrategies to optimize generation of potent effector cells andhigh numbers of long-lived memory cells.

Acknowledgements

We would like to thank Debra Lins for expert technical assistance andSarah Hamilton for advice. This work was supported by NationalInstitutes of Health grants RO1 AI34824 (M.F.M.), PO1 AI35296(M.F.M.) and Center for Disease Control grant RO1 CI00100 (S.C.J.).M.P. was supported by National Institutes of Health Training GrantT32 AI07313.

Abbreviations

DC dendritic cellDLN draining lymph nodegrzB granzyme Bi.p. intra-peritonealLCMV lymphocyte choriomeningitis virusLN lymph nodeOVA ovalbuminOVAP ovalbumin peptideVV-OVAP ovalbumin-expressing VVPC peritoneal cavityTNF-a tumor necrosis factor-aVSV vesicular stomatitis virusVV vaccinia virus

References

1 Downie, A. W. 1939. Immunological relationship of the virus ofspontaneous cowpox to vaccinia virus. Br. J. Exp. Pathol. 20:158.

2 Harrington, L. E., van der Most, R., Whitton, J. L. and Ahmed, R.2002. Recombinant vaccinia virus-induced T-cell immunity:quantitation of the response to the virus vector and the foreignepitope. J. Virol. 76:3329.

3 Xu, R., Johnson, A. J., Liggitt, D. and Bevan, M. J. 2004. Cellularand humoral immunity against vaccinia virus infection of mice.J. Immunol. 172:6265.

4 Kagi, D., Seiler, P., Pavlovic, J. et al. 1995. The roles ofperforin- and Fas-dependent cytotoxicity in protectionagainst cytopathic and noncytopathic viruses. Eur. J. Immunol. 25:3256.

5 Spriggs, M. K., Koller, B. H., Sato, T. et al. 1992. Beta-2-microglobulin-, CD8 T-cell-deficient mice survive inoculation withhigh doses of vaccinia virus and exhibit altered IgG responses.Proc. Natl Acad. Sci. USA 89:6070.

6 Carroll, M. W. and Moss, B. 1997. Poxviruses as expressionvectors. Curr. Opin. Biotechnol. 8:573.

7 Kristensen, N. N., Madsen, A. N., Thomsen, A. R. and Christen-sen, J. P. 2004. Cytokine production by virus-specific CD8 T cellsvaries with activation state and localization, but not with TCRavidity. J. Gen. Virol. 85:1703.

8 Mayer, K. D., Mohrs, K., Crowe, S. R. et al. 2005. The functionalheterogeneity of Type I effector T cells in response to infection isrelated to the potential for IFN-gamma production. J. Immunol.174:7732.

9 Isogawa, M., Furuichi, Y. and Chisari, F. V. 2005. Oscillating CD8T cell effector functions after antigen recognition in the liver.Immunity 23:53.

10 Curtsinger, J. M., Schmidt, C. S., Mondino, A. et al. 1999.Inflammatory cytokines provide third signals for activation of naiveCD4+ and CD8+ T cells. J. Immunol. 162:3256.

11 Curtsinger, J., Lins, D. C. and Mescher, M. F. 2003. Signal threedetermines tolerance versus full activation of naive CD8 T cells:dissociating proliferation and development of effector function.J. Exp. Med. 197:1141.

12 Curtsinger, J. M., Valenzuela, J. O., Agarwal, P., Lins, D. andMescher, M. F. 2005. Cutting edge: type I IFNs provide a thirdsignal to CD8 T cells to stimulate clonal expansion anddifferentiation. J. Immunol. 174:4465.

13 Hogquist, K., Jameson, S., Heath, W., Howard, J., Bevan, M. andCarbone, F. 1994. T cell receptor antagonist peptides inducepositive selection. Cell 76:17.

14 van Stipdonk, M. J., Lemmens, E. E. and Schoenberger, S. P.2001. Naive CTLs require a single brief period of antigenicstimulation for clonal expansion and differentiation. Nat. Immunol.2:423.

15 Curtsinger, J. M., Lins, D. C., Johnson, C. M. and Mescher, M. F.2005. Signal 3 tolerant CD8 T cells degranulate in response toantigen but lack granzyme B to mediate cytolysis. J. Immunol.175:4392.

16 Blattman, J. N., Antia, R., Sourdive, D. J.D. et al. 2002. Estimatingprecursor frequency of naive antigen-specific CD8 T cells. J. Exp.Med. 195:657.

17 Foxwell, B. M., Taylor-Fishwick, D. A., Simon, J. L., Page, T. H. andLondei, M. 1992. Activation induced changes in expression andstructure of the IL-7 receptor on human T cells. Int. Immunol.4:277.

18 Xue, H. H., Kovanen, P. E., Pise-Masison, C. A. et al. 2002.IL-2 negatively regulates IL-7 receptor alpha chain expres-sion in activated T lymphocytes. Proc. Natl Acad. Sci. USA99:13759.

19 Kaech, S. M., Tan, J. T., Wherry, E. J., Konieczny, B. T., Surh, C. D.and Ahmed, R. 2003. Selective expression of the interleukin7 receptor identifies effector CD8 Tcells that give rise to long-livedmemory cells. Nat. Immunol. 4:1191.

20 Shrikant, P. and Mescher, M. F. 1999. Control of syngeneic tumorgrowth by activation of CD8+ T cells: efficacy is limited bymigration away from the site and induction of nonresponsiveness.J. Immunol. 162:2858.

21 Ruddle, N. H. 1992. Tumor necrosis factor (TNF-alpha) andlymphotoxin (TNF-beta). Curr. Opin. Immunol. 4:327.

22 Aggarwal, B. B. 2003. Signalling pathways of the TNF superfamily:a double-edged sword. Nat. Rev. Immunol. 3:745.




Dow

nloaded from


23 Brehm, M. A., Daniels, K. A. and Welsh, R. M. 2005. Rapidproduction of TNF-alpha following TCR engagement of naive CD8T cells. J. Immunol. 2005:5043.

24 Grivennikov, S. I., Tumanov, A. V., Liepinsh, D. J. et al. 2005.Distinct and non-redundant in vivo functions of TNF produced byT cells and macrophages/neutrophils: protective and deleteriouseffects. Immunity 22:93.

25 Xu, L., Yoon, H., Zhao, M. Q., Liu, J., Ramana, C. V. and Enelow, R.I. 2004. Cutting edge: pulmonary immunopathology mediated byantigen-specific expression of TNF-alpha by antiviral CD8 T cells.J. Immunol. 173:721.

26 Kolumam, G. A., Thomas, S., Thompson, L. J., Sprent, J. andMurali-Krishna, K. 2005. Type I interferons act directly on CD8T cells to allow clonal expansion and memory formation inresponse to viral infection. J. Exp. Med. 202:637.

27 Aichele, P., Unsoeld, H., Koshella, M., Schweier, O., Kalinke, U.and Vucikuja, S. 2006. Cutting edge: CD8 T cells specific for

lymphocytic choriomeningitis virus require Type I IFN receptor forclonal expansion. J. Immunol. 176:4525.

28 Curtsinger, J. M., Johnson, C. M. and Mescher, M. F. 2003. CD8 Tcell clonal expansion and development of effector function requireprolonged exposure to antigen, costimulation, and signal 3 cyto-kine. J. Immunol. 171:5165.

29 Kassiotis, G. and Stockinger, B. 2004. Anatomical heterogeneity ofmemory CD4 T cells due to reversible adaptation to themicroenvironment. J. Immunol. 173:7292.

30 Masopust, D., Vezys, V., Wherry, E. J., Barber, D. L. and Ahmed,R. 2006. Cutting edge: gut microenvironment promotes differen-tiation of a unique memory CD8 T cell population. J. Immunol.176:2079.

31 Lacombe, M. H., Hardy, M. P., Rooney, J. and Labrecque, N.2005. IL-7 receptor expression levels do not identify CD8 memoryT lymphocyte precursors following peptide immunization.J. Immunol. 175:4400.




Dow

nloaded from


The CD8 T cell response to vaccinia virus exhibits site-dependent heterogeneity of functional responses

Documents