Proc. Natl. Acad. Sci. USA
Vol. 88, pp. 2283-2287, March 1991
Protection against lethal Sendai virus infection by in vivo priming
of virus-specific cytotoxic T lymphocytes with a free
W. MARTIN KAST*t¶, LAURENT Rouxt, JOSEPH CURREN*, HENDRIKA J. J. BLOM*, ARIE C. VOORDOUW*,
ROB H. MELOEN§, DANIEL KOLAKOFSKYt, AND CORNELIS J. M. MELIEF*
*Division of Immunology, The Netherlands Cancer Institute, Antoni van Leeuwenhoek Huis, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands;
*UniversityofGeneva, 9 Avenue deChampel, 1211 Geneva, Switzerland; and §CentralVeterinary Institute, P.O. Box 65, 8200 AB Lelystad, The Netherlands
Communicated by Stanley G. Nathenson, December 17, 1990
recognized by cytotoxic T lymphocytes (CTL) in B6 mice was
found with (g) the use ofrecombinant vaccinia virus constructs
containing separate genes of Sendai virus and (d) a set of
overlapping peptides completely spanning theidentifiednudeo-
protein (NP) gene product. This immunodominant NP peptide
is recognized by Sendai virus-specific CTL that are known to
have therapeutic effects in vivo. By subcutaneous i'mmuniza-
tion, this peptide induced Sendai virus and NP peptide-specific
CTL memory responses in vivo. Most importantly, mice that
had been immunizd with this peptide were protected against
a lethal virus dose, indicating that viral peptides can be used as
antiviral T-cell vaccines. The induction of T-cell memory by
free peptide immunization potentially has wide applicability in
biology and medicine, including protection against infectious
The only peptide of Sendai virus that is
Antiviral cytotoxic T lymphocytes (CTL) recognize short
peptides derived from viral proteins when bound to major
histocompatibility complex class I molecules (1).
Such virus-specific CTL are crucially important in the
resistance against many virus infections, including lympho-
cytic choriomeningitis virus (2, 3), influenza virus (4-6),
herpes simplex virus type 1 (7), and Sendai virus (8). The
elicitation ofvirus-specific CTL responses is best achieved in
vivo by immunization with attenuated virus, but this bears a
certain risk of causing disease. Therefore, means are sought
to induce virus-specific CTL responses in vivo with synthetic
viral antigens. Recently it was demonstrated that immuniza-
tion with short synthetic peptides of the influenza virus
nucleoprotein (NP) can induce virus-specific CTL to the
priming peptide and to the virus when these peptides are
chemically modified and linked to a lipid component. Free
peptides, however, failed to do so (9). Furthermore, immu-
nization with anunbound lymphocytic choriomeningitis virus
peptide also induced a virus-specific CTL response, provided
the peptide was dissolved in incomplete Freund's adjuvant
(IFA) (10). However, these studies did not show whether
peptide vaccination induced protection against subsequent
virus challenge. Since we have demonstrated that cloned
Sendai virus-specific CTL can cure mice that are lethally
infected with this virus (8), we evaluated whether protection
from lethal virus infection can be achieved by vaccination
with the peptide recognized by these cloned CTL.
The Sendai virus NP was identified as the protein contain-
ing the peptide recognized by these CTL clones with the use
of vaccinia virus recombinants expressing different genes of
Sendai virus. Furthermore, a single peptide recognized by
these clones was identified by screening with a set of over-
lapping peptides covering the complete amino acid sequence
of the NP. Vaccination with this peptide, known to be
recognized by CTL clones that are effective in vivo, led to
successful induction ofCTL memory associated with in vivo
protection against a lethal dose of virulent Sendai virus that
was introduced by the natural respiratory route of infection.
MATERIALS AND METHODS
Mice. B6 mice were bred at The Netherlands Cancer
Institute under specific pathogen-free conditions. They were
used for in vivo experiments when about 8 wk old. In vivo
experiments were carried out at the Central Laboratory ofthe
Netherlands Red Cross Blood Transfusion Service (Amster-
dam, The Netherlands).
Viruses. Virulent Sendai virus strain p3193 was a gift from
J. C. Parker (Microbiological Associates). This virus was
propagated in the lungs of129/J mice. The titerwas 3.2 x 106
tissue culture median infectious dose (TCID50) per ml. Non-
virulent Sendai virus, lot 40340087, was obtained from Flow
Laboratories. This virus has been propagated in pathogen-
free eggs and does not cause cytopathic effects on rhesus
monkey kidney cells (8). The titer was 104 hemagglutination
Virus Challenge. Mice were inoculated intranasally with 20
,ulof diluted virulent Sendai virus at a median lethal dose of
2LD50(= 300 TCID50) (8, 11). Infected mice were observed
for disease or death up to day 40. They were housed in a
Immunization. Mice were primed by one i.v. injection of
100,ugoffree synthetic peptide orby one s.c. injection of100
pg offree synthetic peptide dissolved in phosphate-buffered
saline or dissolved in IFA or by one i.p. injection of 102
hemagglutination units of nonvirulent Sendai virus and used
between 4 and 6 wk after the immunization.
In Viro Generation ofSendai Virus-Specific Bulk CTL. The
method to generate Sendai virus-specific CTL has been
described (12) and was performed with slight modifications.
In brief, responder spleen cells (5 x 106) from in vivo primed
mice were cocultured with irradiated (25 Gy) spleen cells (5
x 106) in the presence of 10 uM Sendai virus NP-(321-336)
peptide (fragment from amino acids 321 to amino acid 336) in
2 ml of culture medium for 5 days at 37°C in humidified air
containing 5% CO2. For the preparation of Sendai virus-
Abbreviations: TCID50, tissue culture median infectious dose; NP,
nucleoprotein; MEC, mouse embryo cell; IFA, incomplete Freund's
adjuvant; CTL, cytotoxic T lymphocytes.
tPresent address: Department of Immunohematology and Blood
Bank, Academic Hospital Leiden Rijnsburgerweg 10, 2333 AA
Leiden, The Netherlands.
tTo whom reprint requests should be sent at the present address.
The publication costs ofthis article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Medical Sciences: Kast et al.
infected stimulator cells, 108 spleen cells in 2 ml of culture
medium were incubated with 6 x 102 hemagglutination units
of nonvirulent Sendai virus for 1 hr at 370C and then were
washed three times. The culture medium consisted of Is-
cove's modified Dulbecco's medium (Flow Laboratories)
supplemented with 10% fetal calf serum, penicillin (100
international units/ml), kanamycin (100 jug/ml), and 2-mer-
captoethanol (20AM).When indicated, 2 units of recombi-
nant interleukin 2 (rIL-2) (= 10 international'units) (Euroce-
tus, Amsterdam) was added to the culture.
Cloned CTL. The Sendai virus-specific CTL clone Tc5 was
used in this study. The establishment, culture, function, major
histocompatibility complex restriction specificity (H-2Kb), and
phenotype (CD4- CD8+) ofthis clone have been described (8).
Cell-Mediated Cytotoxicity Assay. Various numbers of ef-
fector cells were added to 2 x 103 Na25lCrO4 (51Cr)-labeled
target cells in 0.2 ml of culture medium in 96-well U-bottom
plates and were incubated for 6 hr at 370C in humidified air
containing 5% CO2. After incubation, the supernatant was
collected. The percentage of specific 51Cr release was calcu-
lated by the formula:
% specific lysis =
cpm experimental well - cpm background 51Cr release
cpm 2% Triton X-100 release - cpm background 51Cr release
Background (medium) release was always less than 30% of
maximal release. The SD of triplicate wells was always less
than 5% of the specific 51Cr release. Virus-infected target
cells were prepared as described above for stimulator cells,
except that 107 cells were incubated with 10 hemagglutination
units of virus in 100 ,ul ofmedium. When peptides were used
to label uninfected target cells, they were present during the
whole assay at a concentration of 50 ,M when the peptide
was made in bulk amounts, or0.4jigofpeptide was added per
well when the peptide was made in the "pepscan" method
(see below). When targets were infected with vaccinia virus
recombinant, 5 x 106 cells were incubated with 10 plaque-
forming units per cell in 5 ml of culture medium for 14 hr at
37°C in humidified air containing 5% CO2 and were washed
Synthesis of Peptides. Peptides were synthesized by two
methods. The first method (pepscan) has been described (13,
14) and was used with modifications as described (15). In
brief, synthesis took place on polyethylene rods onto which
polymers of polyacrylic acid had been formed by radiation
grafting. To permit the specific detachment of peptides after
synthesis, a tripeptide sequence (Asp-Pro-Gly) was cosyn-
thesized between the peptide and the activated polyethylene
rods. After deprotection and removal on noncovalently
bound impurities, acid cleavage of Asp-Pro bonds was per-
formed by heating in 0.2 ml of 70% formic acid for 20 hr at
37°C. The solvent was removed by lyophilization. The sec-
ond method to generate bulk amounts of peptides has been
described (16) and was performed with slight modifications.
In brief, peptides were synthesized with COOH-terminal
ends on a Biosearch 9500 peptide synthesizer with the use of
Construction of the Recombinant Vacciniia Viruses. The
assembly ofgenes encoding the P/C, HN, and F Sendai virus
proteins (see Results) has been described (17, 18). The gene
encoding the matrix protein M was pieced together from two
overlapping clones: (i) clone S1-2 (19), containing most ofthe
5' end sequence oftheM gene and extending into the adjacent
P gene; and (ii) clone 3'/2' [obtained with the same cloning
procedure as described in ref. 18], containing the exact 3' end
flanked by a Kpn I site. S1-2 was cleaved at its 5' end at the
Sma I site (nucleotide 1815 of the P gene) and recombined
with 3'/2' at the unique HindIII site oftheMgene (nucleotide
854). A full-length clone of the gene encoding NP was
produced by the cloning procedure described by Vidal et al.
(18). The genes encoding M, HN, F, and NP were cloned in
plasmid pSP64 and that encoding P/C was cloned in plasmid
pSP65. The individual genes were then transferred to the
insertion vector pGS62 (kindly provided by G. L. Smith,
Cambridge University), and the vaccinia recombinants were
prepared exactly as described (17) for the P/C gene.
Protein Specificity of Sendai Vius-Specific B6 Bulk and
Cloned CTL. Bulk and cloned Sendai virus-specific CTL
were assayed on a panel of target cells expressing different
viral proteins. The target cells were mouse embryo cells
(MEC) infected with different recombinant vaccinia virus
constructs containing individual Sendai virus genes. Both the
Sendai virus-specific bulk and cloned CTL recognize Sendai
virus-infected MEC, whereas uninfected or vaccinia virus-
infected MEC were not recognized, indicating Sendai virus
specificity (Table 1). On assaying MEC that were infected
with the different recombinant vaccinia virus constructs, we
noted that the infected cells expressing the Sendai virus
fusion (F) protein, hemagglutinin neuraminidase (HN), ma-
trix (M) protein, and polymerase (P/C) protein were not
recognized. Only MEC that expressed the Sendai virus NP
were recognized by both the bulk and cloned CTL. This
indicates thatthe majorreaction ofSendai virus-specific CTL
of B6 mice is directed against the NP of this virus.
Specificity of Sendai Virus-Specific Bulk and Cloned CTL
Assayed with Overlapping Peptides. To identify the NP pep-
Protein specificity of Sendai virus-specific B6 bulk and
51Cr-labeled target cells
51Cr release by
effector cells, %t
B6 target cells were 51Cr-labeled MEC (2 x 103). Infection was
carried out as described in Materials and Methods. Target cells
infected by each separate vaccinia virus construct have been shown
to express the appropriate Sendai virus (SV) protein as tested by
[3S]methionine labeling followed by specific immunoprecipitation
(L.R., J.C., and D.K., data not shown).
*Effector-totarget cell ratio.
tMean percentage of specific 51Cr release of three experiments.
Proc. Natl. Acad Sci. USA 88(1991)
Proc. Natl. Acad. Sci. USA 88 (1991)
CTL assayed with overlapping peptides
Specificity of Sendai virus-specific bulk and cloned
Sendai NP fragment
B6 target cells were 51Cr-labeled Sendai virus-infected MEC (2 x
103). The Sendai NP fragment added (0.4Ag)is specified by the
amino acid (AA) sequence numbered as in ref. 20. n, All 12-amino-
acid-long peptides running from amino acid sequence nos. 3-319; n',
all 12-amino-acid-long peptides running from amino acid sequence
*Mean percentage ofspecific 51Cr release at an effector-to-target-cell
ratio of 8:1.
AA sequence nos.
51Cr release by
effector cells, %*
tide recognized by Sendai virus-specific bulk and cloned
CTL, these cells were assayed against a set of 513 overlap-
ping 12-amino-acid-long peptides. This set covered the com-
plete sequence of the Sendai virus NP (20). The individual
peptides were added to B6 MEC, and these cells were used
without preincubation as targets for the CTL in a standard
cytotoxicity assay. Table 2 shows that both the bulk and
clonedCTL recognized Sendai virus-infected MEC, whereas
uninfected MEC were not recognized. Added to B6 MEC,
only five peptides encompassing the amino acid sequence
Ser-Tyr-Ala (positions 321-336 ofthe Sendai NP) resulted in
recognition by the CTL. Two independently derived Sendai
virus-specific CTL clones of B6 mice recognized the same
peptide (not shown). Since both bulk and cloned CTL rec-
ognized the same peptide, this epitope constitutes the immu-
nodominant epitope of Sendai virus recognized by B6 CTL.
Induction of Sendai Virus- and Viral Peptide-Specifc CTL
byPeptideImmunization. B6mice were immunized in various
ways with the peptide that represents the immunodominant
epitope of Sendai virus recognized by CTL. Table 3 shows
that Sendai virus-immunized mice mounted a CTL response
against both Sendai virus-infected cells and NP-(321-336)
peptide-coated cells (experiment 1). NP-(321-336) peptide-
immunized mice likewise generated a CTL response of
similar magnitude against both Sendai virus-infected cells
and NP-(321-336)-coated cells. The inducibility ofthis mem-
ory response depended on the mode of immunization. Best
results were achieved when the animals were immunized
once s.c. with the NP-(321-336) peptide dissolved in IFA
(experiment 2). Omission of the IFA resulted in a CTL
response of intermediate strength after s.c. immunization
(experiment 3), but injection ofthe NP-(321-336) peptide i.v.
did not result in a detectable CTL response (experiment 4),
as was also the case with unimmunized mice (experiment 5).
The combined results clearly indicate that the mode of
peptide immunization is of crucial importance in the induc-
Induction of virus-specific CTL memory by peptide immunization
B6 spleen cells
51Cr release from B6 MEC target cells"
Incubated with NP-(321-336)
Incubated with NP-(321-336)
Incubated with NP-(321-336)
Incubated with NP-(321-336)
Incubated with NP-(321-336)
The percentage of specific 51Cr release is the mean of three independent experiments. SV, Sendai virus.
*Spleen cells from unprimed, Sendai virus-primed, or NP-(321-336)-primed B6 mice were cultured in the presence of 10
units of interleukin 2 per ml. Priming was performed 4-6 wk earlier.
tSendai virus-infected B6 spleen cells orB6 spleen cells incubated with NP-(321-336) peptide. The peptide remained present
during the whole culture.
§Effector-to-target cell ratio.
'Target cells were LPS-induced 51Cr-labeled B6 blasts (2 x 103) that were untreated or infected with Sendai virus or
incubated with NP-(321-336) peptide. The peptide remained present during the CML assay.
Medical Sciences: Kast et al.
Medical Sciences: Kast et al.
tion of virus-specific CTL memory and that best results are
obtained by injecting the peptide s.c. in IFA, although the use
of IFA is not an absolute prerequisite (Table 3).
Protection Against Lethal Sendai Virus Infection by Peptide
Vaccination. Because s.c immunization with NP-(321-336)
peptide dissolved in IFA resulted in optimal induction of
Sendai virus-specific CTL in vivo that could be measured 4-6
wk later in vitro (Table 3), mice were vaccinated in this way
and challenged 4-6 wk later with a lethal dose of virulent
Sendai virus. The results (Fig. 1) indicate that almost all
control nonimmunized mice, mice treated with IFA, and mice
treated with IFA and unrelated adenovirus peptide (21) died
between days 7 and 16 (only 2 of 18 mice survived). But
almost all mice that had been immunized once s.c. with the
Sendai NP-(321-336) peptide in IFA survived the infection
(16 of 18 mice survived). This shows that mice can be
protectedfrom a lethal virus infection by vaccination with the
immunodominant viral peptide recognized by CTL.
Two major points emerge from these studies. First, by
assaying CTL against targets infected with vaccinia con-
structs containing single Sendai virus genes and by screening
with a set of overlapping peptides, we identified the immu-
nodominant Sendai virus NP peptide recognized by B6 CTL.
Second, by immunization with this viral peptide, which is the
immunodominant epitope for in vivo therapeutic CTL, pro-
tection against lethal virus infection can be obtained that is
associated with induction of antiviral CTL memory. Al-
though survival curves indicated that about 90% of the mice
that had been immunized survived the infection (Fig. 1), we
observed that all surviving mice exhibited minor symptoms of
infection afterbeing challenged with a lethal virus dose. Since
in this haplotype of mice there appears to be only one major
peptide recognized by CTL (Tables 1 and 2), lack of activa-
tion of Sendai virus-specific CTL directed against other
epitopes ofthis virus can be ruled out. However, since we do
not know yet which epitope of the virus is recognized by
helper T cells and therefore have not immunized the mice
with the corresponding peptides, one may argue that after
infection the peptide-primed animals initially have to mount
days after virus inoculation
vaccination. Eight-week-old B6 mice weighing about 20 g were
immunized s.c. once with 100 .&gofSendai virusNP4321-336) peptide
dissolved in IFA. They were challenged4-6 wk later with 300TCID50
(= 2 xLD50; ref. 11) of virulent Sendai virus, after which the
percentage of surviving animals was scored. -, Survival curve of
control nonimmunized mice (n =12);
treated mice (n=6); -.-., survival curve of mice immunized with the
adenovirus typeS EIA peptide (21) (n=6); ---,survival curve ofmice
immunized with the Sendai virus NP-(321-336) peptide (n=18).
Protection against lethal Sendai virus infection by peptide
, survival curve of IFA-
a naive helper T-cell response against the virus before the
helperT cell-dependent secondary CTL response against the
virus (8) is fully functional. We surmise that interleukin 2
production by virus-specific helper T cells will facilitate in
vivo activity ofvirus-specific CTL, as demonstrated earlier in
experiments with Sendai virus-specific helper T cells and
CTL clones (8). In these experiments, the helperT-cell clone
by itselfhad no protective effect, and the cooperative activity
of the helper T-cell clone in the protection shown by a CTL
clone could bereplaced by interleukin 2. The unprimedstatus
of helper T cells in peptide-immunized mice in this study
would leave some time for the virus to initiate signs of
disease. When the peptide(s) of Sendai virus recognized by
helper T cells in B6 mice are known, one could test whether
immunization with peptides recognized by both CTL and
helper T cells protects mice also from the initial signs of
sickness. Complete absence ofdisease is also observed upon
challenge of mice with virulent virus following preimmuni-
zation with nonlethal doses of live virus (11).
The data ofTable 3 indicate that the manner in which mice
were immunized with peptides is of crucial importance; s.c.
injection with peptides dissolved in IFA resulted in optimal
induction of memory CTL responses as measured in vitro.
This confirms the in vitro data in the lymphocyticchoriomen-
ingitis virus model (10). But it is also apparent that interme-
diate CTL memory can begenerated without the use ofIFA.
Intravenous injection failed to prime the mice, however,
resulting in undetectable CTL memory. This last observation
is in agreement with the data published in the influenza virus
model (9) in which it was shown that lipid-bound peptides
could be used when injected i.v., whereas unbound peptides
could not prime when injected i.v. However, the conclusion
drawn in that report (9)-i.e., that free synthetic peptides
cannot be used for immunization-holds only for i.v. routes
and notforother ones like the morecommonlyused s.c. route
that clearly is effective, as shown in this study.
While we have shown that it is possible to protect mice
from a lethal virus infectionby peptideimmunization and that
this protection is associated withequallevels ofpeptide-and
virus-specific CTL memory, improvements in the method of
vaccination for optimal application of this principlein animal
and human vaccination programs are still needed. For ex-
ample the use of IFA is not possible for clinicalapplication.
It is noteworthy in this regardthat our in vitro data(Table 3)
already indicate that the use of IFA is not an absolute
prerequisite. Indeed IFAprobably onlyfunctions as adepot
to retain peptides locally and prevent their rapid dissipation
and disappearance through, for instance, urine. Arachis oil
could be used in man as adepotinstead of IFA.
Another point to bear in mind is that the Sendai NP
fragment identified in this study is the immunodominant
peptide for this strain of mice. In an outbredpopulation,T
cells from individuals with different MHCtypesarelikelyto
recognize different peptides of viralproteins (22), indicating
that a universal peptide vaccine suitable to vaccinate an
outbred population should contain a mixture of different
peptides. The identification of thepeptidesneeded in such a
mixture maybe facilitatedbytherecentlydescribed method
to measure binding ofpeptides to MHC class I molecules in
the lysate of cells (23). By screeningwith this method, all
peptides that bind to a particular MHC molecule are candi-
date peptides for apeptide-basedvaccine.
We thank the animal technicians ofthe CentralLaboratoryofThe
Netherlands Red Cross Blood Transfusion Service and The Neth-
erlands Cancer Institute formaintainingthemice,M. A. van Halem
fortypingthemanuscript,and S. G. Nathenson forhelpfuldiscus-
sions. This work wassupported bythe World HealthOrganization.
Proc. Natl. Acad Sci. USA 88(1991)
Medical Sciences: Kast et al. Download full-text
Townsend, A. R. M., Rothbard, J., Gotch, F. M., Bahadur,
G., Wraith, D. & McMichael, A. J. (1986) Cell 44, 959-968.
Byrne, J. A. & Oldstone, M. B. A. (1984) J. Virol. 51, 682-686.
Lehmann-Grube, F., Moskophidis, D. & Lohler, J. (1988)Ann.
N. Y. Acad. Sci. 532, 238-256.
Lin, Y. L. & Askonas, B. A. (1981) J. Exp. Med. 154, 225-234.
Taylor, P. M. & Askonas, B. A. (1983) Eur. J. Immunol. 13,
Lukacher, A. E., Braciale, V. L. & Braciale, T. J. (1984) J.
Exp. Med. 160, 814-826.
Sethi, K. K., Omata, Y. & Schneweis, K. E. (1983) J. Gen.
Virol. 64, 443-447.
Kast, W. M., Bronkhorst, A. M., De Waal, L. P. & Melief,
C. J. M. (1986) J. Exp. Med. 164, 723-738.
Deres, K., Schild, H., Wiesmuller, K.-H., Jung, G. & Ram-
mensee, H.-G. (1989) Nature (London) 342, 561-564.
Aichele, P., Hengartner, H., Zinkernagel, R. M. & Schulz, M.
(1990) J. Exp. Med. 171, 1815-1820.
Kast, W. M., Bluestone, J. A., Heemskerk, M. H. M., Spaar-
garen, J., Voordouw, A. C., Ellenhorn, J. D. I. & Melief,
C. J. M. (1990) J. Immunol. 145, 2254-2259.
De Waal, L. P., Kast, W. M., Melvold, R. W. & Melief,
C. J. M. (1983) J. Immunol. 130, 1090-1096.
Proc. Natl. Acad. Sci. USA 88 (1991)2287
Geysen, H. M., Meloen, R. H. & Barteling, S. J. (1984) Proc.
Nati. Acad. Sci. USA 81, 3998-4002.
Geysen, H. M., Barteling, S. J. & Meloen, R. H. (1985) Proc.
Nati. Acad. Sci. USA 82, 178-182.
Van der Zee, R., Van Eden, W., Meloen, R. H., Noordzij,
A. & Van Embden, J. D. A. (1989) Eur. J. Immunol. 19, 43-
Hodges, R. S. & Merrifield, R. B. (1975) Anal. Biochem. 65,
Curran, J. & Kolakofsky, D. (1989) EMBO J. 8, 521-526.
Vidal, S., Mottet, G., Kolakofsky, D. & Roux, L. (1989) J.
Virol. 63, 892-900.
Blumberg, B. M., Rose, K., Simona, M. G., Roux, L., Giorgi,
C. & Kolakofsky, D. (1984) J. Virol. 52, 656-663.
Shioda, T., Hidaka, Y., Kanda, T., Shibuta, H., Nomoto, A. &
Iwasaki, K. (1983) Nucleic Acids Res. 11, 7317-7330.
Kast, W. M., Offringa, R., Peters, P. J., Voordouw, A. C.,
Meloen, R. H., Van der Eb, A. J. & Melief, C. J. M. (1989)
Cell 59, 603-614.
Taylor, P. M., Davey, J., Howland, K., Rothbard, J. B. &
Askonas, B. A. (1987) Immunogenetics 26, 267-272.
Schumacher, T. N. M., Heemels, M. T., Neefjes, J. J., Kast,
W. M., Melief, C. J. M. & Ploegh, H. L. (1990) Cell 62,