Poxvirus-Encoded Serpins Do Not Prevent Cytolytic T
Cell-Mediated Recovery from Primary Infections1
Arno Mu ¨llbacher,2* Reinhard Wallich,†Richard W. Moyer,‡and Markus M. Simon§
Previous observations that the highly conserved poxvirus-encoded serpins inhibit cytotoxic activities of alloreactive CTL via
granule and/or Fas-mediated pathways was taken to indicate their involvement in immune evasion by poxviruses. We now show
that interference with51Cr release from target cells by ectromelia and cowpoxvirus is limited to alloreactive but not MHC-
restricted CTL. The data are in support of the paramount importance of CTL and its effector molecule perforin in the recovery
from primary ectromelia virus infection and question the role of serpins in the evasion of poxviruses from killing by CTL. Further
analysis of poxvirus interference with target cell lysis by alloreactive CTL revealed that suppression primarily affects the Fas-
mediated, and to a lesser extent, the granule exocytosis pathway. Serpin-2 is the main contributor to suppression for both killing
pathways. In addition, inhibition of lysis was shown to be both target cell type- and MHC allotype-dependent. We hypothesize that
differences in TCR affinities and/or state of activation between alloreactive and MHC-restricted CTL as well as the quality (origin)
of target cells are responsible for the observed phenomenon. The Journal of Immunology, 1999, 162: 7315–7321.
tions of a biological meaningful function for CTL (1–3). Until
than, CTL were primarily generated and analyzed as alloreactive
CTL. That both CTL populations come from the same precursor
pool and have overlapping repertoires could be deduced from pre-
cursor frequency analysis (4), cross-reactivity of CTL clones on
allogeneic and self-modified targets (5, 6), and cross-suppression
leading to holes in the CTL repertoire (7). For these reasons, it is
generally assumed that findings from studies on cytolytic mecha-
nisms using alloreactive CTL are universally applicable for all
To date, two pathways of target cell killing by cytolytic leuko-
cytes have been described. One is the granule exocytosis pathway
mediated by perforin and serine proteases or granzymes (gzm) (8,
9). This is generally believed to be the dominant mechanism by
which CTL eliminate virus-infected cells (10). The second mech-
anism, called Fas-mediated pathway, requires the interaction of the
Fas receptor on the target cell with the Fas ligand on the killer cell
(11) and is supposedly involved in immunoregulation (12).
he demonstration that CTL are of fundamental impor-
tance in the recovery from primary ectromelia (ECT)3
virus infections in mice was one of the first demonstra-
Over twenty years ago, Gardner et al. (13) reported on poxvirus-
mediated suppression of alloreactive CTL-mediated target cell ly-
sis using mouse poxvirus, ECT. They observed severe inhibition of
lysis of ECT-infected target cells by alloreactive CTL, while the
same targets were highly susceptible to lysis by MHC class I-re-
stricted ECT-immune CTL. It was proposed then that the inhibi-
tion of alloreactive CTL-mediated lysis was due to a decrease in
normal MHC class I cell surface expression and a replacement by
virus-modified MHC, a consequence of poxvirus-mediated host
protein synthesis inhibition (14).
More recently, it was found that poxviruses encode proteins
related to the serpin family of proteinase inhibitors, termed SPI-1,
-2, and -3 (15, 16). SPI-2 (or cytokine response modifier (crmA))
was shown to inhibit both Asp-specific serine and cysteine pro-
teases, including components involved in CTL-mediated cytotoxic
and inflammatory responses, such as gzmB (17) and caspases 1
(IL-1 converting enzyme (ICE)) (18), 3 (CPP32) (19), and 8 (Fas-
associated death domain-like ICE (FLICE)) (20), respectively. Al-
though the rates of inhibition of the various proteases by crmA
were shown to differ by several orders of magnitude, they seem
fast enough to be of biological significance (17, 18). Serpin-like
genes have been found in all poxviruses analyzed so far (15), and
the high homology observed between the serpins of vaccinia virus
Western reserve (VV-WR), cowpoxvirus (CPV), rabbitpoxvirus
(RPV), Variola, and ECT (R.W. et al., unpublished observations),
indicate conserved function and evolutionary benefit. The fact that
crmA is able to prevent target cell apoptosis, mediated largely via
the Ca2?-independent (Fas-mediated) pathway, by alloreactive
CTL (21), suggested that serpins evolved as an immune escape
mechanism to avoid immune destruction of infected cells before
viral replication and viral-induced cytolysis. This was also empha-
sised by findings of Macen et al. (22) that showed that target cell
lysis by an alloreactive CTL line, as measured by51Cr release, was
greatly reduced for both cytolytic pathways, upon infection with
either CPV or RPV, but restored by infection with virus deletions
of the serpin SPI-1 and SPI-2 genes.
These findings, together with the evidence that functionally
active ECT-immune CTL, and in particular the granule protein
perforin (23), are required for recovery from primary ECT virus
infections, led us to study the role of poxvirus-encoded serpins in
*Division of Immunology and Cell Biology, John Curtin School of Medical Research,
Australian National University, Canberra, Australia;†Institut fu ¨r Immunologie der
Universita ¨t Heidelberg, Heidelberg, Germany;
Medical Microbiology, University of Florida, Gainesville, FL; and§Max-Planck-In-
stitut fu ¨r Immunbiologie, Freiburg, Germany
‡Department of Immunology and
Received for publication November 30, 1998. Accepted for publication March
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
1M.M.S. was supported in part by a grant from the Deutsche Forschungsgemein-
schaft, Si 214/7-1.
2Address correspondence and reprint requests to Dr. Arno Mu ¨llbacher, Division of
Immunology and Cell Biology, John Curtin School of Medical Research, Australian
National University, P.O. Box 334, Canberra, ACT 2601, Australia. E-mail address:
3Abbreviations used in this paper: ECT, ectromelia; VV, vaccinia virus; WR, West-
ern reserve; CPV, cowpox virus; RPV, rabbitpox virus; SPI, serine proteinase inhib-
itor (serpin); crmA, cytokine response modifier; gzm, granzyme; ICE, IL-1 converting
enzyme; Perf, perforin; NPP, nuclear protein peptide from influenza virus; HAP,
hemagglutinin protein from influenza virus.
Copyright © 1999 by The American Association of Immunologists0022-1767/99/$02.00
the inhibition of target cell lysis by alloreactive and MHC-
restricted CTL in vitro.
Materials and Methods
C57BL/6 (KbDb) (B6), CBA/H (KkDk) (CBA), BALB/c (KdDd) (B/c),
C3H.H2o(KdDk) (OH), B10.HTG (KdDb) (HTG), B10.A(2R) (KkDb) (2R),
B10.A(5R) (KbDd) (5R), and the perforin-deficient mutant (Perf?/?) (24)
and the granzyme A and B deficient mutant (A ? B?/?) (25) were bred
under pathogen-free conditions at the Animal Breeding Facility of the John
Curtin School of Medical Research (Canberra City, Australia). The Fas
ligand mutant mice (gld) bred on B6 background were obtained from the
Centenary Institute (Sydney, Australia). Only female animals ?12 wk of
age were used.
Viruses and synthetic peptides
The cowpoxviruses wild type (CPV), the mutants with a defect in serpin 1
(CSPI-1) and serpin 2 (CSPI-2) (22), and the vaccinia virus (VV) WR
strain (VV) were grown on CV-1 cell monolayers. The ECT virus Moscow
strain was grown in mouse spleen. All poxviruses were titrated as described
(26). The influenza virus strains A/WSN (H1N1) and A/JAP (H2N2) were
prepared and titrated as described (26).
The synthetic peptide derived from the nucleoprotein of influenza A
virus specific for Kd, TYQRTRALV, and specific for Kk, SDYEGRLI
(NPP) (27), was synthesized at the Biomolecular Resource Facility (Aus-
tralian National University, Canberra, Australia). The synthetic peptide
derived from the hemagglutinin of A/JAP (H2N2) virus specific for
Kd(LYQNVGTYV) (HAP) was obtained from Chiron (Melbourne,
The mouse cell lines L929 (H-2k), L929-Fas (kindly provided by P. Kram-
mer, Heidelberg, Germany), MC57 (H-2b), RMA (H-2b), HTG (KdDb), 2R
(KkDb), and 5R (KbDd) were grown in Eagle’s minimal essential medium
supplemented with 10% FCS. The cells were infected with poxviruses at a
multiplicity of infection (MOI) of 10–20 PFU per cell for 16 h before being
labeled with51Cr for 1 h and used for analysis. Target cells were peptide-
treated with NPP at the same time as labeled with51Cr as previously de-
scribed (26). Modification of targets with HAP was as for NPP.
Animals were immunized with 106PFU of ECT into hind footpad, 107
PFU of VV-WR, 1 ? 106PFU CPV, or 104hemagglutination units of
A/WSN (H1N1) i.p. For ECT virus, mice were infected with 1 ? 106PFU
ECT into the hind footpads.
L929 (KkDk) or HTG (KdDb) cells were infected with 20 PFU of ECT, as
described for target cells. At 16 h after infection, the cells were washed,
resuspended in growth medium at 107cells per ml, and labeled at 4°C for
45 min with mAb HB-160 (American Type Culture Collection (ATCC),
Manassas, VA) specific for Kk, mAb 15-5-5S (a gift from F. Momburg,
Heidelberg, Germany) specific for Dkfor L929 cells, mAb HB-159
(ATCC) specific for Kd, or mAb HB-27 (ATCC) specific for Db, followed
by FITC-conjugated sheep anti-mouse Ig (Silenus, Hawthorn, Australia)
staining. Cells were examined with a FACScan flow cytometer (Becton
Dickinson, Mountain View, CA).
Generation of CTL
For primary poxvirus, immune CTL splenocytes of 6-day immunized an-
imals were used ex vivo. For the generation of alloreactive CTL, 8 ? 107
responder splenocytes were cocultured with 4 ? 107irradiated (2000 rad)
allogeneic stimulator cells for 5 days in 40 ml Eagle’s minimal essential
medium, 10% FCS plus 10?5M 2-ME. The generation of secondary NPP-
immune CTL has been described (26).
51Cr release cytotoxicity assay
The methods used for target cell lines have been described in detail else-
where (26). The duration of the assays was 6 h. Percentage of specific lysis
was calculated by the formula: % specific lysis ? [(experimental release ?
medium release)/(maximum release ? medium release)] ? 100. Data
given are the means of triplicate determinations. SEM values were always
Cross-reactivity of poxvirus-immune CTL and susceptibility of
poxvirus-infected target cells to lysis by poxvirus-immune CTL
To confirm the earlier results of Gardner et al. (13) and to extend
them to other poxviruses, we used primary ex vivo-derived splenic
effector cells immune to VV, CPV, or ECT and tested them on
target cells infected with either the homologous or heterologous
poxviruses. Fig. 1 shows two representative experiments using two
mouse strains, CBA (Fig. 1A) and B6 (Fig. 1B), as donors of pox-
virus-immune CTL that were tested for their cytolytic potential to
lyse H-2-matched target cells, L929 and MC57, respectively. The
results clearly demonstrate that poxvirus-immune CTL are broadly
cross-reactive within the poxvirus family, indicating conservation
of immunodominant peptides among different viruses.
The lysis of target cells in Fig. 1, A and B, reached plateau levels
of 60–70% at E:T ratios of 30–60:1. Such levels of lysis are
similar to those found for other primary ex vivo-derived virus-
immune CTL (28), thus indicating that inhibition by poxvirus-
encoded serpins, if it occurs at all, may only be marginal in af-
fecting poxvirus-infected cell lysis by poxvirus-immune CTL.
Differential inhibition of lysis of poxvirus-infected target cells by
alloreactive and MHC-restricted CTL
To investigate further the role of serpins in the down-regulation of
target lysis by MHC-restricted vs alloreactive responses, we made
use of peptide modification of target cells. It is obvious that lysis
of target cells by poxvirus-immune CTL could not be used to eval-
uate possible interference of lysis by serpins. In Fig. 2A, L929
target cells were infected with ECT for 16 h or left uninfected.
ex vivo primary poxvirus-immune CTL. A, Lysis of L929 target cells by
CBA VV- and CPV-immune splenocytes. B, Lysis of MC57 target cells by
B6 ECT-immune and B6 VV-immune splenocytes. Targets were mock-
infected (f) or infected with VV (Œ), ECT (F), or CPV (E) for 16 h.
Cytotoxic assay time was 6 h. Each point constitutes the mean of percent
specific lysis of three separate wells. Spontaneous release was always
Cross-reactivity of lysis of poxvirus-infected target cells by
7316POXVIRUS SERPINS FAIL TO INHIBIT MHC-RESTRICTED KILLERS
Infection of target cells for 1 or 3 h before assay did not affect
alloreactive CTL lysis, but did sensitize for poxvirus-immune CTL
lysis (data not shown). The cells were tested for lysis by two al-
Dk) and two MHC-restricted CTL populations, namely primary ex
vivo-derived ECT-immune CTL and secondary influenza NPP-im-
mune CTL. For the latter, target cells were incubated for 1 h before
assay with 10?4M NPP peptide with the motif for Kk. Fig. 2A
(first panel) shows the lysis by ECT-immune effectors. Levels of
lysis of the ECT-infected targets reaches 70–80%, demonstrating
successful target cell infection. When the mock and ECT-infected
target cells were treated with NPP and tested for lysis by NPP-
immune effector cells (second panel), mock and ECT-infected
NPP-treated target cells were lysed to the same extent and signif-
icantly exceeded that of untreated targets. In contrast, using allo-
reactive CTL, target cell lysis was greatly suppressed by ECT in-
fection compared with mock-infected targets. This suppression
was more pronounced with CTL directed against the D end than K
end (third and fourth panels). The presence of NPP did not to any
significant amount affect lysis of alloreactive or poxvirus-immune
CTL (data not shown).
The data shown in Fig. 2B extend the findings using ECT and an
additional poxvirus CPV and a similar experimental design, but
different effector and target cells. The two alloreactive CTL pop-
ulations, B6 anti-HTG (anti-Kd) and 5R anti-B6 (anti-Db) lysed
HTG (KdDb) mock-infected targets to a similar extent. Infection
with either ECT or CPV reduced the release of51Cr 9- to 20-fold,
with CPV being the stronger inhibitor (second and third panel). On
the other hand, B/c secondary influenza HAP-immune CTL lysed
HAP-modified HTG target cells to a similar extent, irrespective of
poxvirus infection (first panel). Although small inhibition is ap-
parent, the slopes of the killing curves are the same with HAP-
immune CTL, inferring the same rate of killing, but are different
with alloreactive CTL.
Gardner et al. (13) have shown that H-2kalloreactive CTL lysed
80% of mock-infected L929 target cells but only 7% of ECT-
infected targets, which, however, were lysed to 100% by ECT-
immune effectors. These experiments, together with the present
data, clearly establish that inhibition of target cell lysis by poxvirus
infection is predominantly observed with alloreactive but not
Cell surface MHC class I expression after poxvirus infection
To test whether the results obtained in Fig. 2 with alloreactive CTL
can be explained by a decrease in MHC class I cell surface ex-
pression as originally proposed by Gardner et al. (13), we under-
took FACS analysis of mock- and virus-infected target cells. Fig.
3 shows the expression of MHC class I Kkand Dkon L929 target
cells 16 h after ECT infection. The same cells were used for the
experiment shown in Fig. 2A. Fluorescence intensity increased af-
ter infection in regard to anti-KkAbs and decreased when infected
cells were labeled with anti-DkAb. Note that there is no correla-
tion between the level of MHC class I expression and inhibition of
In a separate experiment from Fig. 2B, HTG cells were infected
with ECT or left uninfected and examined to determine to what
or infected with ECT (F) for 16 h, and treated with NPP (SDYEGRLI) (broken line) or left untreated (solid line) for 1 h were tested for lysis by CBA
ECT-immune (first panel), CBA NPP-immune (second panel) MHC-restricted CTL, and B6 anti-2R (anti-Kk), and B/c anti-OH (anti-Dk) alloreactive CTL
(third and fourth panels). B, HTG target cells mock-infected (f), infected with either CPV (E), or ECT (F) for 16 h, and treated with HAP (LYQNVGTYV)
(broken line) or left untreated (solid line) for 1 h, were tested for lysis by B/c HAP-immune (first panel) MHC-restricted CTL, and B6 anti-HTG (anti-Kd)
and 5R anti-B6 (anti-Db)alloreactive CTL (second and third panels). Cytotoxic assay time was 6 h. Each point constitutes the mean of percent specific
lysis of three separate wells. Spontaneous release was always ?20%.
Differential inhibition of lysis of poxvirus-infected target by MHC-restricted and alloreactive CTL. A, L929 target cells mock-infected (f)
7317The Journal of Immunology
extent poxvirus infection alters cell surface expression of Kdand
Db. Fig. 3 (lower panels) shows that both class I molecules are
marginally reduced on the cell surface as compared with mock-
Poxvirus inhibition of target cell lysis is cell type-dependent and
varies with different CTL effector populations
The variability between target cells in poxvirus inhibition was fur-
ther explored within one experiment using two alloreactive effector
populations, 5R anti-B6 (anti-Db) and 2R anti-B6 (anti-Kb). These
effector populations were tested on various Kb- and/or Db-express-
ing target cell lines, RMA (KbDb), MC57 (KbDb), 5R (KbDd), and
2R (KkDb) (Fig. 4). The targets were either mock-infected or in-
fected with ECT, CPV, or the selective CPV-derived serpin mu-
tants CSPI-1 or CSPI-2 (15, 22). First, the level of lysis of RMA
target cells by either of the two alloreactive CTL effectors was, if
at all, only marginally affected after infection with any of the four
virus preparations, although the targets were susceptible to poxvi-
rus-immune CTL (data not shown). Similar results were found
with other target cell populations of hematopoietic origin, such as
El-4, L1210, or P815 (data not shown). Lysis of the other three
targets (all of which are of connective tissue origin and fibroblast-
like) were affected by poxvirus infection, however, to different
degrees. Anti-Kballoreactive CTL lysed mock-infected MC57 and
5R targets to a similar extent. Infection with ECT, CPV, and
CSPI-1 completely inhibited lysis of 5R targets, but only partially
inhibited lysis of MC57 targets. The mutation in CSPI-2 partially
relieved the suppression in both targets but substantially more in
MC57 than 5R targets. Anti-Dbeffectors lysed mock-infected
MC57 targets to a lesser extent than 2R. Suppression by CPV was
30- and 5-fold for 2R and MC57, respectively, and by ECT, 5- and
3-fold, respectively. The mutation in SPI-1 did not affect suppres-
sion as compared with wild-type CPV in any of the assays. CSPI-2
virus completely abrogated suppression on MC57 targets and only
partially released inhibition on 2R targets. The data are consistent
with the interpretation that SPI-1 is not involved in suppression.
SPI-2, on the other hand, either completely or partially reduces
target cell lysis, depending on the effector cells employed. Thus,
additional virus molecules, such as SPI-3 (15), may be responsible
for some inhibition of CTL effector function in particular circum-
The role of poxvirus in inhibiting Fas- and granule exocytosis-
mediated cell death
To evaluate the influence of poxvirus infection on Fas-mediated
cytolysis vs the granule exocytosis pathway, we made use of three
mutant mouse strains. One was deficient in perforin (Perf?/?) and
as presently understood, CTL derived from such mice rely on the
Fas pathway for cell cytolysis (24). The second strain was gld
mice, which are defective in Fas ligand expression (29) and thus
kill primarily via the exocytosis pathway (24). The third strain was
A ? B?/?mice; the CTL of which exert cytolytic but not nucleo-
lytic activity via the Fas-independent pathway (25). Splenocytes
from these mutant and wild-type B6 mice stimulated in vitro with
cells after being infected for 16 h with ECT (broken line) or mock-infected
(solid line) and labeled with mAb specific for Kkor Dkon L929 cells (top
panels) and mAb specific for Kdor Dbon HTG cells (bottom panels).
Cell surface expression of MHC class I of L929 and HTG
pression of lysis of poxvirus serpin de-
letion mutants-infected target cells by
alloreactive CTL. MC57, 5R, 2R, and
RMA target cells mock-infected (f),
infected with CPV (E), ECT (F), C
SPI-1 (C), or CSPI-2 (?) for 16 h
Kb) (top panels) and 5R anti B6 (anti-
Db) alloreactive CTL. Cytotoxic assay
time was 6 h. Each point constitutes
the mean of percent specific lysis of
three separate wells. Spontaneous re-
lease was always ?20%.
Partial release of sup-
7318 POXVIRUS SERPINS FAIL TO INHIBIT MHC-RESTRICTED KILLERS
Kk(2R) allogeneic splenocytes were tested on L929 and the Fas-
transfected variant L929.Fas for
measured after assay times of 6 and 12 h (Fig. 5). Effectors from
B6 wild-type mice lysed ECT- and CPV-infected L929 target cells
three to times less efficiently than mock-infected cells, at both time
points. No differences in lysis were observed between mock and
CSPI-2-infected L929 cells. The differential in susceptibility of
mock vs ECT- and CPV-infected targets was even greater on
L929.Fas targets. In contrast to L929, L929.Fas target lysis was
partially suppressed by CSPI-2. Gld-derived effectors lysed ECT-
and CPV-infected L929 and L929.Fas target cells substantially less
efficient than mock-infected targets. However, infection of both
targets with CSPI-2 did not effect their lysability. Thus, in the
absence of the Fas-pathway, inhibition of lysis is due exclusively
to SPI-2 (crmA). Perf?/?-derived CTL did not lyse L929 target
cells in a 6-h assay, but lysed L929.Fas mock-infected targets ef-
ficiently. Lysis of the latter targets was completely abrogated upon
infection with ECT or CPV. Infection with CSPI-2 only partially
restored lysability. In 12-h assays, the same effectors did lyse
mock-infected L929 targets to a significant extent and only slightly
less CSPI-2-infected targets. However, again, ECT- or CPV-in-
fected targets were not lysed to any significant extent. The effects
of CPV, ECT, and CSPI-2 on lysis of L929.Fas by perf?/?-derived
CTL in the 12-h assay were similar to those seen at 6 h. In this
51Cr release was
experiment, the lysis profiles with A ? B?/?-derived alloreactive
effector cells are lower on all target cells independent on the state
of infection, but in essence similar to that obtained with B6-de-
rived effectors. This argues against the possibility that the inhibi-
tion of the exocytosis pathway by ECT and CPV is due to inacti-
vation of granzymes, in particular of gzmB by SPI-2.
The main points from the data shown here are as follows. First,
poxvirus-immune CTL lyse target cells infected with homologous
and heterologous poxviruses equally well. Second, lysis by allo-
reactive, but not by biologically relevant MHC-restricted, CTL is
inhibited when target cells are infected with the poxviruses ECT or
CPV. Third, the suppression of target cell lysis is haplotype-de-
pendent and more pronounced with cells of connective tissue (fi-
broblast-like) than hemopoietic origin. Fourth, poxvirus infection
completely abrogated the Fas-mediated51Cr release and partially
inhibited the granule exocytosis pathway. Fifth, SPI-2 (crmA) is
the main inhibitor of the Fas pathway, but also interferes with
granule exocytosis-mediated lysis by processes independent of
gzmA and or gzmB.
The data described here and previous results (13) are incompat-
ible with recent interpretations (21, 22) that inhibition of CTL-
mediated target cell lysis by poxvirus-encoded serpins, SPI-1 and
SPI-2, is a means for virus to escape from cytotoxic T cell-medi-
ated cytolysis. This is emphasized by 1) the original finding that
virus-immune CTL cells are critical for recovery from infection
with ECT (1, 2, 30), 2) the fact that Perf?/?mice are highly sus-
ceptible to ectromelia infection as compared with wild-type B6
mice (Ref. 23; and G. Karupiah, unpublished observations), and 3)
the present demonstration that inhibition of target cell lysis, as
observed with CPV and ECT, is predominantly seen with biolog-
ically irrelevant alloreactive, but not with relevant MHC class I-
restricted, CTL. Together with the finding that poxvirus mainly
interferes, most probably via crmA (SPI-2), with Fas-mediated cy-
tolytic processes (Ref. 21, and data shown here), these data clearly
establish that perforin-mediated control of ECT infections, a cy-
topathic virus, is of primary importance. The converse was in-
ferred from experiments using the nonmouse pathogen VV (31), an
infection that was shown before to be controlled in the absence of
CTL (32). Similarly, studies presented here and elsewhere (22),
using nonrelated poxviruses, CPV and RPV, in a mouse model,
may also be inadequate to uncover strategies of either host or virus,
which are of evolutionary significance. This is, in fact, exemplified
by a recent comparison of the course of infection in perforin
knockout mice upon inoculation with either ECT or CPV (23).
The differential susceptibility of the same poxvirus-infected tar-
get cells to alloreactive and MHC class I-restricted CTL can be the
result of a number of different mechanisms. One possibility may be
that the three different effector CTL employed here, primary ex
vivo-derived poxvirus immune, in vitro alloreactive, and in vitro
secondary influenza immune, may greatly differ in their activation
state, which may influence their cytolytic potential in the presence
of serpins. Another explanation may be fundamental differences in
the mechanisms by which the two effector populations are engaged
to deliver their lethal hit by yet unknown molecular basis. The
possibility that the two CTL populations express distinctly differ-
ent cytotoxic potentials is doubtful in light of their inherent ca-
pacity to specifically lyse their respective target cells and their
overlapping repertoire. It is more likely that they possess TCRs
with different affinities, and, consequently, require differing num-
bers of receptor/ligand interactions to achieve an avidity sufficient
and Fas?target cells by alloreactive CTL from perforin, granzyme A plus
B, and Fas receptor-deficient mice. L929 and L929.Fas target cells mock-
infected (f), infected with CPV (E), ECT (F), or CSPI-2 (?) for 16 h
were tested for lysis by 2R anti-B6 (anti-Kb) alloreactive CTL from wild-
type B6, Perf?/?, A ? B?/?or gld mice. Cytotoxic assay time was 6 and
12 h. Each point constitutes the mean of percent specific lysis of three
separate wells. Spontaneous release was always ?20%.
Lysis of poxvirus and SPI-2 deletion mutant infected Fas?
7319The Journal of Immunology
for triggering effector function. Quantitative consideration in TCR
engagement to achieve triggering of Tc-mediated cytotoxic pro-
cesses, such as exocytosis or Fas ligand-Fas ligation, may well be
reflected in qualitatively different signals received by target cells
Two distinct possibilities by which poxvirus infection may af-
fect alloreactive, but not MHC restricted, CTL lysis can be envis-
aged. First, alloreactive CTL may induce a qualitatively different
death pathway than MHC-restricted CTL due to their requirement
for multiple receptor/ligand interactions. Such possibilities have
been proposed recently (34). Second, poxviruses may be able to
alter the target cell ligands recognized by CTL. This may occur in
two ways; first, by a possible down-regulation of MHC class I, as
has originally been proposed by Gardner et al. (13) due to poxvirus
inhibition of host protein synthesis (14). However, the data shown
in Fig. 3 indicate that the changes in cell surface expression of
MHC class I after poxvirus infection do not correlate with the CTL
lysis inhibition results, as lysis to both Kkand Dkis reduced but
only Dkcell surface expression is lower than that on mock-infected
targets. The more pronounced inhibition seen in the case of Dk- vs
Kk-reactive CTL (Fig. 2) may be due to the decrease of Dkcell
surface expression. Alternatively, one could postulate that poxvi-
rus infection affects MHC class I cell surface motility, preventing
aggregation of class I molecules and TCRs bound to them, neces-
sary for low-affinity alloreactive CTL to kill, but which would
leave CTL with high-affinity TCRs unaffected. The different
strength of inhibition seen with different targets and different MHC
class I molecules on one and the same target is consistent with this
hypothesis and may reflect differential MHC class I cell surface
concentrations of K and D Ags and varying cell membrane fluidity.
It is known that poxvirus infection alters the cytoskeletal structures
within the cell (35), and such alterations may be responsible for
changes in MHC class I cell surface motility. We are at present
investigating this possibility.
In addition, the observation that fibroblast-like target cells in-
fected with poxvirus were prevented from lysis but not, or only
marginally, cells of hematopoietic origin may be for the same rea-
son, namely differential fluidity in the cell membrane. Alterna-
tively, these two groups of target cells derived from different tis-
sues, though similarly infected as indicated by their susceptibility
to lysis by poxvirus-immune CTL, may express the poxvirus en-
coded “suppresser” molecules in different quantities.
Another possible explanation for the ability of CPV and ECT to
inhibit target cell lysis by alloreactive CTL has been proposed by
Macen et al. (22) and others (21, 17–19). Poxvirus-encoded serpins
may interfere with cytolysis and/or nucleolysis by inhibiting pro-
teases involved in the death pathways, such as caspases 1, 3, and
8, as well as gzmB. In favor of this interpretation is the fact that
CPV with a mutation in serpin-2 restores, at least partially, target
cell lysis by alloreactive CTL (Figs. 4 and 5). In light of previous
findings that peptide caspase inhibitors blocked both nucleolysis
and cytolysis by the CTL-mediated Fas pathway, but only nucle-
olysis and not cytolysis induced via granule exocytosis (36), the
present data indicate that target cell lysis elicited by virus-immune
CTL is exclusively mediated by perforin. They also suggest that
the reduced capacity of alloreactive CTL to exert their full cyto-
lytic potential on poxvirus-infected target cells is due to incom-
plete granule exocytosis as a consequence of suboptimal TCR en-
As to the effect of poxvirus infection on the proteolytic activity
on gzm, it was found that the cowpox serpin inhibitor SPI-2 was
able to inhibit gzmB in vitro (17). Although the rate of inhibition
is fast enough to be of physiological significance, its implication
for the survival from natural poxvirus infection is unclear. We are
currently investigating the role of gzmB in poxvirus infection and
should be able to provide definitive evidence if serpin/gzmB in-
teractions are biologically significant in determining the survival
from natural poxvirus infections. What is clear is that, in the ab-
sence of gzmA and gzmB, lack of SPI-2 restores lysability to target
cells in extended assays. As this is also true with effectors from Fas
ligand-defective mice (gld), it suggests that SPI-2 interferes with
yet an undefined pathway leading to51Cr release. Inhibition of
gzmA by poxvirus is rather unlikely because normal B6 mice are
able to control infection, whereas gzmA?/?mice are highly sus-
ceptible under similar conditions with increased mortality and
morbidity (37). Since alloreactive and MHC class I-restricted CTL
from gzmA knockout mice express normal cytolytic potential, this
was attributed to a direct effect of gzmA on virus replication rather
than interference with CTL cytotoxicity.
We thank Ron Tha Hla, Seow Chin, and Thao Tran for excellent technical
assistance; and Dr R. V. Blanden for helpful discussions.
1. Blanden, R. V. 1971. Mechanisms of recovery from a generalized viral infection:
mousepox. II. Passive transfer of recovery mechanisms with immune lymphoid
cells. J. Exp. Med. 133:1074.
2. Blanden, R. V. 1974. Mechanisms of cell-mediated immunity in viral infection.
In Progress in Immunology II. Clinical Aspects I, Vol. 4. L. Brent, and
J. Holborow, eds. North Holland, Amsterdam, p. 117–125.
3. Blanden, R. V., and I. D. Gardner. 1976. The cell-mediated immune response to
ectromelia virus infection. I. Kinetics and characteristics of the primary effector
T cell response in vivo. Cell. Immunol. 22:271.
4. Skinner, M. A., and J. Marbrook. 1976. An estimation of the frequency of pre-
cursor cells which generate cytotoxic lymphocytes. J. Exp. Med. 143:1562.
5. von Boehmer, H., H. Hengartner, W. Nabholz, W. Lernhardt, M. H. Schreier, and
W. Haas. 1979. Fine specificity of a continuously growing killer cell clone spe-
cific for H-Y antigen. Eur. J. Immunol. 9:592.
6. Pfizenmaier, K., H. Jung, R. Kurrle, M. Ro ¨llinghoff, and H. Wagner. 1980. Anti
H-2Ddalloreactivity mediated by herpes-simplex-virus specific cytotoxic H-2kT
lymphocytes is associated with H-2Dk. Immunogenetics 10:395.
7. Mu ¨llbacher, A. 1981. Neonatal tolerance to alloantigens alters MHC-restricted
response patterns. Proc. Natl. Acad. Sci. USA 78:7689.
8. Henkart, P. A. 1994. Lymphocyte-mediated cytotoxicity: two pathways and mul-
tiple effector molecules. Immunity 1:343.
9. Podack, E. R., H. Hengartner, and M. G. Lichtenheld. 1991. A central role of
perforin in cytolysis? Annu. Rev.Immunol. 9:129.
10. Ka ¨gi, D., B. Ledermann, K. Bu ¨rki, R. M. Zinkernagel, and H. Hengartner. 1995.
Lymphocyte-mediated cytotoxicity in vitro and in vivo: mechanisms and signif-
icance. Immunol. Rev. 146:95.
11. Rouvier, R., M.-F. Luciani, and P. Golstein. 1993. Fas involvement in Ca2?-
independent T-cell-mediated cytotoxicity. J. Exp. Med. 177:195.
12. Nagata, S. 1997. Apoptosis by death factor. Cell 88:355.
13. Gardner, I., N. A. Bowern, and R. V. Blanden. 1975. Cell-mediated cytotoxicity
against ectromelia virus-infected target cells. III. Role of the H-2 gene complex.
Eur. J. Immunol. 5:122.
14. Moss, B. 1968. Inhibition of HeLa cell protein synthesis by the vaccinia virion.
J. Virol. 2:1028.
15. Turner, P. C., P. Y. Musy, and R. W. Moyer. 1995. Poxvirus serpins. Springer
16. Buller, R. M. L., and G. J. Palumbo. 1991. Poxvirus pathogenesis. Microbiol.
17. Quan, L. T., A. Caputo, R. C. Bleackley, D. J. Pickup, and G. S. Salvesen. 1995.
Granzyme B is inhibited by the cowpox virus serpin cytokine response modifier
A. J. Biol. Chem. 270:10377.
18. Ray, C. A., R. A. Black, S. R. Cronheim, T. A. Greenstreet, P. R. Sleath,
G. S. Salwesen, and D. J. Pickup. 1992. Viral inhibition of inflamation: Cowpox
virus encodes an inhibitor of the interleukin-1? converting encyme. Cell 69:597.
19. Tewari, M., L. T. Quan, K. O’Rourke, S. Desnoyers, Z. Zeng, D. R. Beidler,
G. G. Poirier, G. S. Salvesen, and V. M. Dixit. 1995. Yama/CPP32?, a mam-
malian homolog of CED-3, is a CrmA-inhibitable protease that cleaves the death
substrate poly(ADP-ribose) polymerase. Cell 81:801.
20. Medema, J., C. Scaffidi, F. Kischkel, A. Shevchenko, M. Mann, P. Krammer, and
M. Peter. 1997. FLICE is activated by association with the CD95 death-inducing
signaling complex (DISC). EMBO J. 16:2794.
21. Tewari, M., W. G. Telford, R. A. Miller, and V. M. Dixit. 1995. CrmA, a pox-
virus-encoded serpin, inhibits cytotoxic T-lymphocyte-mediated apoptosis.
J. Biol. Chem. 270:22705.
7320 POXVIRUS SERPINS FAIL TO INHIBIT MHC-RESTRICTED KILLERS
22. Macen, J. L., R. S. Garner, P. Y. Musy, M. A. Brooks, P. C. Turner, Download full-text
R. W. Moyer, G. McFadden, and R. C. Bleackley. 1996. Differential inhibition
of Fas- and granule-mediated cytolysis pathways by the orthopoxvirus cytokine
response modifier A/SPI-2 and SPI-1 protein. Proc. Natl. Acad. Sci. USA 93:
23. Mu ¨llbacher, A., R. Tha Hla, C. Museteanu, and M. M. Simon. 1999. Perforin is
essential for the control of ectromelia virus but not related poxviruses in mice.
J. Virol. 73:1665.
24. Ka ¨gi, D., B. Ledermann, K. Bu ¨rki, P. Seiler, B. Odermatt, K. J. Olsen,
E. R. Podack, R. M. Zinkernagel, and H. Hengartner. 1994. Cytotoxicity medi-
ated by T cells and natural killer cells is greatly impaired in perforin-deficient
mice. Nature 369:31.
25. Simon, M. M., M. Hausmann, T. Tran, K. Ebnet, J. Tschopp, R. Tha Hla, and A.
Mu ¨llbacher. 1997. In vitro and ex vivo-derived cytolytic leukocytes from gran-
zyme A ? B double knockout mice are defective in granule-mediated apoptosis
but not lysis of target cells. J. Exp. Med. 186:1781.
26. Mu ¨llbacher, A., A. Hill, R. Blanden, W. Cowden, N. King, and R. Tha Hla. 1991.
Alloreactive cytotoxic T cells recognize MHC class I antigen without peptide
specificity. J. Immunol. 147:1765.
27. Rammensee, H.-G., T. Friede, and S. Stevanovic. 1995. MHC ligands and peptide
motifs: first listing. Immunogenetics 41:178.
28. Zinkernagel, R. M., and P. C. Doherty. 1979. MHC-restricted cytotoxic T cells:
studies on the biological role of polymorphic major transplantation antigens de-
termining T-cell restriction-specificity, function, and responsiveness. Adv. Immu-
29. Takahashi, T., M. Tanaka, C. I. Brannan, N. A. Jenkins, N. G. Copeland, T. Suda,
and S. Nagata. 1994. Generalized lymphoproliferative disease in mice, caused by
a point mutation in the Fas ligand. Cell 76:969.
30. Blanden, R. V., and I. D. Gardner. 1976. The cell-mediated immune responses to
extromelia virus infection. I. Kinetics and characteristics if the primary effector T
cell response in vivo. Cell. Immunol. 22:271.
31. Ka ¨gi, D., P. Seiler, P. Pavlovic, B. Ledermann, K. Bu ¨rki, R. M. Zinkernagel, and
H. Hengartner. 1995. The roles of perforin- and fas-dependent cytotoxicity in
protection against cytopathic and noncytopathic viruses. Eur. J. Immunol. 25:
32. Spriggs, M. K., B. H. Koller, T. Sato, P. J. Morrissey, W. C. Fanslow,
O. Smithies, R. F. Voice, M. B. Widmer, and C. R. Maliszewski. 1992. ?2-
microglobulin-, CD8?T-cell-deficient mice survive inoculation with high doses
of vaccinia virus and exhibit altered IgG responses. Proc. Natl. Acad. Sci. USA
33. Alberola-Ila, J., S. Takaki, J. D. Kerner, and R.M. Perlmutter. 1997. Differential
signaling by lymphocyte antigen receptors. Annu. Rev. Immunol. 15:125.
34. Combadiere, B., C. Reis e Sousa, R. N. Germain, and M. J. Lenardo. 1998.
Selective induction of apoptosis in mature T lymphocytes by variant T cell re-
ceptor ligands. J. Exp. Med. 187:349.
35. Wolffe, E. J., E. Katz, A. Weisberg, and B. Moss. 1997. The A34R glycoprotein
gene is required for induction of specialized actin-containing microvilli and ef-
ficient cell-to-cell transmission of vaccinia virus. J. Virol. 71:3904.
36. Sarin, A., M. S. Williams, M. A. Alexander-Miller, J. A. Berzofsky,
C. M. Zacharchuk, and P. A. Henkart. 1997. Target cell lysis by CTL granule
exocytosis is independent of ICE/Ced-3 family proteases. Immunity 6:209.
37. Mu ¨llbacher, A., K. Ebnet, R. V. Blanden, T. Stehle, C. Museteanu, and
M. M. Simon. 1996. Granzyme A is essential for recovery of mice from infection
with the natural cytopathic viral pathogen, ectromelia. Proc. Natl. Acad. Sci.
7321The Journal of Immunology