| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
*
Department of Microbiology and Immunology, Kimmel Cancer Institute, Thomas Jefferson University, Philadelphia, PA 19107; and
Department of Microbiology and Immunology, Faculty Health Sciences, University of Ben-Gurion, Beer-Sheva, Israel
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. Upon appropriate stimulation,
naive CD8+ T cell precursors undergo extensive
proliferation and clonal expansion. A re-evaluation of this initial T
cell expansion has indicated that it is much larger and more
Ag-specific than previously determined (3, 4, 5, 6, 7). Further,
the development of the primary effector T cell population directly
affects both the magnitude and repertoire of the resulting memory
population (5, 8, 9, 10, 11, 12, 13). Numerous parameters can influence the character of the primary CD8+ T cell response. Genetic factors, such as TCR repertoire and MHC-encoded gene products, may significantly bias the T cell response to different epitopes (14, 15, 16). In addition, cytokines such as type I IFN and IL-15 can skew the type or magnitude of a CD8+ T cell response (17, 18) and may vary significantly during different infections. Further, the level of costimulation can modulate the level of Ag necessary to stimulate CD4+ (19) or CD8+ (20, 21) T cells.
Another variable that may be a determinant of T cell function is the density of epitope presented at the surface of APCs. This issue has been addressed by many laboratories using both in vitro and in vivo models, with no clear consensus emerging. Many in vitro studies have shown that the concentration of peptide pulsed onto target cells has a dose-dependent effect on the functions of activated T cells such as cytotoxicity, proliferation, or IL-2 production. Reports disagree as to whether the same is true for naive CD8+ T cells stimulated in vitro. The concentration of peptide available during priming of naive T cells in vitro has been shown to affect the lytic capacity of resulting populations in a graded fashion (22). However, studies using cells pulsed with different concentrations of peptide or beads coated with different densities of MHC/peptide complexes have suggested a threshold effect for the priming of CTL in vitro (23, 24).
Evaluating the impact of modulating the level of epitope expressed in vivo has been more challenging as additional variables complicate the interpretation of results. Altering the quantity of input Ag by varying the dose of virus or transfected cells has been shown to affect the size of the CTL population generated or the duration and efficacy of CTL memory (25, 26, 27). However, this also alters the dose of Ags available to other arms of the immune system. Indeed, the amount of Ag presented to CD4+ T cells can modulate the choice of differentiation into Th1 or Th2 (28, 29), altering the overall immune response and potentially the relevant CD8+ T cell population. Additional studies comparing the responses to different epitopes have inferred a relationship between epitope density and T cell responses in vivo by: 1) correlating the expression level of different epitopes in vitro with the number of epitope-specific clones generated or lysis of target cells by ex vivo CTL (30, 31); 2) determining that less immunogenic epitopes are released quickly by MHC class I molecules, perhaps reducing the effective epitope density in vivo (31, 32, 33, 34); and 3) showing that the number of CTL induced in vivo to subdominant epitopes is lower than that induced to dominant epitopes (5, 35). Data from these systems suggest that the level of epitope presented influences the magnitude of the CTL response; however, the distinct populations of T cells involved limit direct comparisons. In addition, peptides that dissociate too quickly from class I molecules may not be capable of providing the necessary stimulus to fully activate a naive T cell, further hindering comparison of responses to distinct epitopes. In fact, when CTL specific for three different viral epitopes were adoptively transferred, their protective capacity did not correlate with epitope densities measured in vitro (31). Therefore, though suggestive, the link between epitope expression levels and the magnitude and effectiveness of the response in these studies is indirect.
Ideally, this question should be addressed by measuring in vivo T cell
responses to different levels of the same epitope in the context of
equivalent overall immune responses. Two studies have accomplished this
goal and yet arrive at different conclusions. The expression of
minigenes by recombinant vaccinia virus
(vac),3 encoding only
the minimal epitope, leads to higher levels of epitope expression than
expression of full-length protein (36, 37). Priming with
minigene-expressing vectors was shown to generate a more potent CTL
population, as assessed in vitro by bulk lytic capacity, presumably due
to higher epitope density (38). However, CTL populations
were not quantitated in this study, and it is unclear whether the
increase in lytic capacity was a result of an increase in numbers or
overall avidity of the in vitro effectors. In addition, minigene
expression systems may be somewhat unique as it is unlikely that T
cells encounter this density of epitope during any natural infection. A
more thorough examination of the role of epitope expression levels on
the priming of naive precursors in vivo was performed by systematically
varying the density of a single epitope and quantitating resulting
epitope-specific CTL populations. By altering Ag processing efficiency,
epitope levels were modulated in the context of a constant dose of
Listeria monocytogenes (39). A density of
roughly 200 epitopes/cell elicited no detectable CTLs, whereas epitope
densities from 
1000 to 4000 copies/cell elicited the same number of
T cells in vivo. We concluded that the activation of naive
CD8+ in vivo is binary and that after an Ag
presentation threshold is reached, further increase in epitope
expression does not result in an increase in the size of the responding
CTL population. These results would appear to substantiate concerns
regarding the alternative approaches described above and support the in
vitro studies suggesting a threshold for stimulation of naive CTL
precursors (CTLp) (23, 24). An extension of this
conclusion is that strategies designed to optimize epitope expression
would have little impact on the CTL response in vivo. This is a crucial
issue considering the effort that has been expended in determining how
to raise and maintain effective CTL populations against a number of
human pathogens including HIV.
We have also established a system that allows modulation of epitope expression with a constant overall antigenic challenge. Using a system that can generate cell surface epitope densities from 200 to 60,000 epitopes/cell in vitro, we have investigated the impact of epitope density on CTL priming in vivo. Our results show that the level of epitope expressed has a clear titrating effect on the size of the T cell population induced in vivo that is more dramatic than the effect on CTL recognition in vitro. Interestingly, this relationship breaks down at the highest level of epitope expression resulting in a decrease in the size of the epitope-specific T cell population generated in vivo.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Six- to 8-wk-old female C3H (H-2k), BALB/c (H-2d), or C57BL/6 (H-2b) mice were purchased from Taconic Laboratories (Albany, NY) and maintained in the Thomas Jefferson University Animal Facilities (Philadelphia, PA). The murine L929 (H-2k; American Type Culture Collection (ATCC), Manassas, VA) cells, L929 transfected with the Kb gene (L-Kb cells, kindly provided by Dr. Yvonne Paterson, University of Pennsylvania, Philadelphia), MC57g (H-2b) cells, B8 cells (H-2d, kindly provided by Dr. Peter Kloetzel, Humboldt University, Berlin, Germany), and CV-1 (CCL70; ATCC) cells for recombinant vaccinia virus generation and 143B HuTK- cells (CRL 8303; ATCC) for vac titrations were maintained in DMEM supplemented with 5% FCS at 9% CO2. EL-4.G7-OVA (a kind gift of Drs. J. W. Yewdell and J. R. Bennink, National Institutes of Health, Bethesda, MD), and EL-4 were maintained in RPMI 1640 supplemented with 10% FCS, 10 µg/ml gentamicin, and 5 x 10-5 M 2-ME at 6% CO2. The OVA257–264/Kb-specific, LacZ-transfected T cell hybridoma, B3Z, and the fusion partner, BWZ.36 (kindly provided by Dr. Nilabh Shastri, University of California, Berkeley, CA) were maintained in RPMI 1640 supplemented with 10% FCS, 10 µg/ml gentamicin, and 5 x 10-5 M 2-ME at 6% CO2. All spleen cultures and 51Cr release assays were performed in RPMI 1640 supplemented with 10% FCS, 10 µg/ml gentamicin, and 5 x 10-5 M 2-ME (assay medium). All chemicals were purchased from Sigma (St. Louis, MO) unless otherwise noted.
Hairpin (HP) generation and molecular manipulations
To generate the NP/SIINFEKL construct, a silent NheI site was encoded in the NP gene at position 1098. Synthetic oligonucleotides encoding the OVA257–264 epitope (SIINFEKL) plus two additional amino acids, alanine and lysine, were inserted in frame into the NheI-cut NP gene in pBluescript II SK+/- (Stratagene, La Jolla, CA). This insertion maintained the integrity of the H-2Db-restricted epitope NP366–374 and contained an internal MscI site for screening. The HP20 construct has been previously described (40). This construct contains a 20-bp duplex structure between the XbaI and SalI sites of the pSC11 plasmid and also contains a KpnI site at the top of the duplex structure. Generation of HPs of other sizes using this original construct has also been previously described (41). Briefly, synthetic oligonucleotides were inserted between the KpnI and SalI sites of the HP20 construct encoding the desired number of mismatches, one disrupting the KpnI site at the top of the HP and the other(s) at the bottom of the stem loop structure. Confirmation of constructs was performed by screening for the loss of KpnI and by sequencing. The HP series containing the firefly luciferase gene was generated by excising the luciferase gene from the pGL3-Basic Vector (Promega, Madison, WI) with BamHI and HindIII and ligating into BamHI, HindIII-cut pBluescript. Luciferase was excised from pBluescript with SalI and NotI and cloned into the SalI/NotI cut HP vectors described above. All enzymes were purchased from New England Biolabs (Beverly, MA).
Viruses
The recombinant vaccinia viruses encoding NP, NP(M)50–57, and NP(M)147–155 have been previously described (40, 42). The OVA(M)257–264 vac (designated (M)SIINFEKL in the figures) was a kind gift of Drs. Yewdell and Bennink. Recombinant vaccinia viruses were made as previously described (42). All constructs were inserted into the multiple cloning site of the modified pSC11 plasmid for expression from the vaccinia P7.5 promoter. Plasmids were introduced into the vaccinia genome by homologous recombination in CV-1 cells and triple plaque purified in 143B cells in the presence of 5 mg/ml 5-bromo-2'-deoxyuridine (Boehringer Mannheim, Indianapolis, IN) and then expanded and titered on 143B HuTK- cells.
Western blot and immunoprecipitation
Western blot was performed by infecting 106 L-Kb cells with 5 pfu/cell of HP vacs for 1 h in BSS/BSA and for an additional 7 h in DMEM supplemented with 5% FCS. Cells were lysed in 0.01 M Tris-HCl, 0.14 M NaCl, 0.5% Nonidet P-40 containing protease inhibitors (1 mM PMSF, 2 mM EDTA, 20 µM pepstatin A, 20 µM leupeptin, and 100 µM 4-(2-aminoethyl) benzenesulfonyl fluoride (AEBSF)). Samples were boiled with 2x reducing SDS buffer (0.125 M Tris-HCl, 4% SDS, 20% glycerol, and 10% 2-ME) and run on a 10% polyacrylamide gel containing 0.1% SDS. Protein was transferred to a nitrocellulose membrane overnight and blotted with culture supernatant from the anti-NP hybridoma, H19-S24 (a kind gift of Dr. Walter Gerhard, Wistar Institute, Philadelphia, PA) and then 5 µg/ml peroxidase-labeled goat anti-mouse IgG (Vector Laboratories, Burlingame, CA). The blot was developed using 3,3' diaminobenzidine and ß-chloronaphthol dissolved in methanol and added to 10 ml PBS containing 20 µl H2O2 (30%). Radioimmunoprecipitation was performed by infecting 106 L-Kb cells with the various HP constructs in BSS/BSA. After 1 h of infection, DMEM with 5% FCS and 50 µCi/106 cells [35S]Met/Cys (Amersham, Arlington Heights, IL) was added for and additional 7 h. Cells were centrifuged, washed, and lysed as for Western blot. Lysates were separated on a 10% SDS-PAGE gel and analyzed by autoradiography.
Luciferase assay
For in vitro assays, infection of 4 x 105 L-Kb cells with 5 pfu/cell of the various luciferase vacs was performed for various time intervals. After 1 h in 500 µl BSS/BSA in 6-well plates, BSS/BSA was removed and 2 ml DMEM supplemented with 5% FCS was added for the remainder of the infection. Luciferase was measured using the Luciferase Assay System with Reporter Lysis Buffer (Promega) kit according to manufacturer’s directions. Light intensity was measured in an EG&G Berthold Lumat LB9507 Luminometer (Bad Wildbad, Germany).
For measurement of luciferase expression in vivo, mice were infected i.p. with 107 pfu/mouse. Approximately 48 h later spleens were removed, homogenized, and washed, and luciferase activity was measured from 107 spleen cells/mouse as above.
Flow cytometry
For the assessment of epitope expression in vitro, 1 x 106 L-Kb cells were infected with 5 pfu/cell of the indicated vacs for 1 h in 200 µl BSS/BSA after which 5 ml of DMEM supplemented with 5% FCS was added and the infection continued for 5, 7, 9, or 13 h. Cells were centrifuged and transferred to 96-well round-bottom plates, washed three times with PBS/azide, and incubated for 2 h with a 1:10 dilution of 25.D1.16 culture supernatant (anti-OVA257–264/Kb) (37) or an isotype-matched control (3G8 kindly provided by Dr. Bice Perussia, Thomas Jefferson University, Philadelphia, PA) at 4°C. Cells were washed three times in PBS/azide and incubated for 2 h at 4°C with 3 µg/ml fluorescein-conjugated horse anti-mouse IgG (Vector Laboratories). Cells were washed three times with PBS/azide and incubated for 15 min in 0.5% paraformaldehyde (PFA) at room temperature, washed once, resuspended in 100 µl PBS/azide, and analyzed using an Epics Profile flow cytometer (Coulter, Miami, FL) in the Kimmel Cancer Center Flow Cytometry Facility (Thomas Jefferson University). Determination of cell surface complexes was performed using Quantum Simply Cellular Microbeads (Flow Cytometry Standards, San Juan, PR). Beads were stained under the same conditions as infected cells according to manufactures directions and cell surface complexes calculated by linear regression analysis.
For the detection of epitope on cells infected in vivo, C57BL/6 mice were infected i.p. with 107 pfu of various HP vacs. Approximately 24 h later spleens were removed and homogenized, RBC were lysed, and the remaining cells were washed two times. A total of 2 x 106 cells were stained using the following procedure: Fc receptors were blocked for 15 min at 4°C using a mixture of anti-Fc receptor Ab (hybridoma 2.4G2 (HB197 ATCC) supernatant), isotype matched hybridoma supernatant (hybridoma MCP21 specific for human proteasome (European Collection of Animal Cell Cultures, Salisbury, U.K.)), and normal goat serum (Vector Laboratories) at a ratio of 2.5:2.5:1, washed three times in PBS/azide + 1% BSA, incubated with biotinylated 25.D1.16 at 20 µg/ml or a biotinylated isotype control (biotin mouse IgG1, PharMingen, San Diego, CA) for 30–60 min at 4°C, washed three times, incubated with FITC-labeled avidin D (Vector Laboratories) at 10 µg/ml for 30–60 min at 4°C, washed three times, and resuspended in 100 µl of 2% PFA before examining by flow cytometry as described above. In all cases, mean fluorescence of the experimental group was determined based on the fluorescence intensity of a control group of cells for each condition stained with only the secondary reagent (FITC-labeled avidin D).
T cell hybridoma assays
A total of 5 x 104 L-Kb cells were infected with 5 pfu/cell of the HP vacs for 1 h in 200 µl BSS/BSA with 375 µg/ml Ara-C. Cells were centrifuged, washed twice with PBS, and transferred to flat-bottom 96-well plates in DMEM supplemented with 5% FCS and 375 µg/ml Ara-C. A total of 5 x 104 B3Z or BWZ.36 cells were added to the wells, and cells were cultured overnight at 37°C with 6% CO2. Cultures were developed by soluble or insoluble substrate methods as previously described (43). For the insoluble assay, the substrate X-Gal was used at 1 mg/ml. For the soluble assays, chlorophenol red ß-galactoside (CPRG) was used at 0.15 mM and 4-methylumbelliferyl ß-D-glucuronide (MUG) at 33 µg/ml.
CTL generation and 51Cr release assays
NP50–57-, NP147–155- NP366–374-, and OVA257–264-, or vac-specific CTL populations were generated by immunization of C3H, BALB/c, or C57BL/6 mice as previously described (42, 44). Briefly, mice were immunized i.p. with 107 pfu of vac expressing the appropriate minigene in 250 µl BSS/BSA. At least 2 wk later, spleens were harvested and homogenized. Of the total splenocytes, 2/9 were infected with influenza A PR/8/34 (for NP-specific CTL) and 1/9 were infected with vac (for vac-specific CTL) for 1 h at room temperature (PR8) or 37°C (vac) and washed, and the PR8-infected cells were mixed with 4/9 of the total spleen (2/3 of the remaining uninfected cells) and the vac-infected cells mixed with 2/9 of the total spleen (1/3 of the remaining uninfected cells) and incubated in assay medium (see above). Cultures were harvested after 6–7 days and used for CTL assays. For OVA257–264-specific CTL, 2/3 of the spleen was incubated with 10-9 M SIINFEKL peptide and 3.3 µg/ml ß2-microglobulin (Scripps Laboratories, La Jolla, CA) in 1.5 ml BSS/BSA for 1 h, washed, and incubated in RPMI with 20U/ml IL-2. 51Cr release assays were performed essentially as previously described (44). Briefly, L-Kb, MC57g, or B8 cells were harvested by trypsinization, pelleted, and resuspended at 2.5 x 106 cells/ml and infected for 1 h with 10 pfu/cell vac at 37°C. Two to 10 ml of DMEM supplemented with 5% FCS was added, and the infection was continued for 3 h at 37°C. Infected cells were pelleted and resuspended at 2 x 107 cells/ml in DMEM with 100 µCi/106 cells and incubated for 1 h at 37°C. Cells were washed three times with PBS, suspended in medium, combined with CTL, and coincubated for 4 h at 37°C. Half the supernatant (100 µl) was collected and counted on a gamma counter and percent lysis determined.
In vivo priming assays
C3H (NP50–57-specific response) or C57BL/6 (OVA257–264-specific response) were infected i.p. with 107 pfu (C3H) or 106 pfu (C57BL/6) of the various HP vacs, minigene, and control vacs in 250 µl BSS/BSA. Spleens were removed after 14 days and restimulated essentially as above, except that spleen populations were adjusted to the same cell density for restimulation in a given experiment. Restimulation of C57BL/6 spleens for some assays was performed with peptide at 10-6 M or irradiated (10,000 cGy) EL-4.OVA cells at a ratio of 1:25 to spleen cells with 20 U/ml IL-2. For the NP50–57 response, C3H mice were immunized with an older generation HP series expressing NP without the OVA257–264 epitope and restimulated with flu as for epitope-specific CTL populations above. CTL activity was assessed after 6–7 days in a standard 51Cr release assay (as above) against L929 or L-Kb cells infected with the appropriate minigene or control virus.
Limiting dilution analysis (LDA)
Mice were immunized as above and 14 days later spleens removed
and homogenized. Cells were titrated in V-bottom 96-well plates
(Costar, Cambridge, MA) with 24 replicates at each titration in a
volume of 100 µl in assay medium. Splenocytes from uninfected mice,
used as stimulators, were irradiated (2000 cGy) and infected with flu
(1 h in BSS/BSA for the NP50–57 response) pulsed
with peptide (10-6 M + 3.3 µg/ml
ß2m in 1.5 ml BSS/BSA for the
OVA257–264 response), or infected with wild-type
modified vaccinia ankara (MVA) at 0.5–2 pfu/cell. MVA is a version
of vac that is noncytopathic in most mammalian cells (45)
and provides a useful method to restimulate for vac-specific CTLp.
Peptide-pulsed or MVA-infected stimulators were spun and resuspended in
assay medium supplemented with 40 U/ml IL-2 (20 U/ml final culture
concentration) to a final density of 2.5–5 x
106 cells/ml depending upon the experiment and
100 µl stimulators added to the plates with responders and plates
were incubated at 37°C with 6% CO2. After 7
days, wells were split two ways and tested for the ability to lyse
L-Kb cells pulsed with the appropriate or an
irrelevant peptide (for peptide restimulated) or uninfected
L-Kb cells or L-Kb cells
infected with MVA (for MVA restimulated). Using MVA-infected targets
greatly reduces spontaneous lysis of target cells during the long assay
compared with wild-type vac, making interpretation of results more
straightforward. Target cells were pulsed with the appropriate or an
irrelevant peptide (10-6 M + 3.3 µg/ml
ß2m), infected with MVA at 10 pfu/cell or left
uninfected while being labeled with 51Cr at 100
µCi/106 cells in 25
µl/106 cells for 1–1.5 h. Targets cells were
washed three times in PBS and 100 µl targets coincubated with 100
µl effectors for 7–8 h in round-bottom 96-well plates at 37°C with
6% CO2. One-half the supernatant of each well
was harvested and 51Cr release measured. Positive
wells were identified as those which caused lysis of the appropriate
target cells (appropriate peptide pulsed or MVA infected) 3 SD above
lysis of the 24 wells of cells incubated with the control target cells
(irrelevant peptide pulsed or uninfected). Any wells that caused lysis
of control targets above 3 SD were eliminated from the analysis.
Precursor frequencies were calculated by 
2
analysis as previously described (46) using a computer
program by Dr. Richard Miller (University of Michigan, Ann Arbor, MI)
kindly provided by Dr. Ralph Tripp (Centers for Disease Control,
Atlanta, GA).
ELISPOT
ELISPOT analysis was performed essentially as previously
described (47). Mice were immunized as above, and either 7
or 14 days later spleens were removed and homogenized. Spleen cells
were plated at various numbers of cells/well in 50 µl assay
media into 96-well ELISPOT plates coated 1 day previously with 20
µg/ml HB170 (ATCC) anti-IFN-. For the
NP50–57-specific response, wells also received
2 x 105 L929 cells irradiated with 10,000 cGy and
pulsed with peptide (as above), infected with irrelevant vac (as
above), or left untreated. For the
OVA257–264-specific response, wells also
received 2 x 105 EL-4 cells irradiated with
10,000 cGy, EL-4 cells irradiated with 10,000 cGy and infected with
irrelevant vac (as above), or EL-4.OVA cells irradiated with 10,000
cGy. Plates were incubated at 37°C with 6% CO2
for 18–24 h, washed extensively, and incubated for 2 h to
overnight with a second biotinylated anti-IFN- Ab at 4 µg/ml
(PharMingen). After washing, 10 µg/ml HRP-conjugated avidin D
(Vector Laboratories) was added to each well and the plate was
incubated at room temperature for 2 h. Following further washing,
spots were developed using 3,3' diaminobenzidine and ß-chloronaphthol
dissolved in methanol and added to 10 ml PBS containing 20 µl
H2O2 (30%). Spots were
counted using a dissecting microscope by two separate investigators,
one of whom was blinded to the experimental design.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The model Ag used for most studies is depicted in Fig. 1
. It is composed of the full-length
influenza A PR/8/34 nucleoprotein (NP) with aa 257–264 of chicken OVA
inserted as shown (single letter code SIINFEKL). NP contains three
well-defined murine CTL epitopes (NP50–57
restricted to H-2Kk,
NP147–155 restricted to
H-2Kd, and NP366–374
restricted to H-2Db) that we have reported on
previously (40, 41, 44, 48, 49). Responses to the
H-2Kb-restricted
OVA257–264 epitope have been studied
extensively, leading to the generation of reagents exploited here. The
resulting protein, encoding all four CTL epitopes is termed NP/SIINFEKL
(NP/S).
|
). The HP constructs are
termed according to the number of base pairs in the duplex. For
example, HP17 contains a 17-bp thermostable duplex barrier before the
start codon of the NP gene (see Materials and Methods). In
addition, the wild-type construct without a HP, termed HP0, and
minigene constructs for each epitope were utilized. The minigene
constructs, encoding the minimal epitope preceded by only an initiating
methionine, induce very high numbers of epitope/class I complexes at
the cell surface (36, 37), likely due to the minimal
processing required before transport and loading onto class I
molecules. Thus, equivalent infection by a panel of vacs expressing the
different constructs is expected to result in a wide range of cell
surface epitope/MHC class I complexes.
The effect of the HP series on protein expression was assessed by
measuring steady state levels of NP/S. L cells transfected with the
H-2Kb gene (L-Kb) were
infected with 5 pfu/cell of the various vac HP constructs for 8 h,
lysed, and subjected to Western blot analysis with an Ab specific for
wild-type NP. Fig. 2A shows a
Western blot of NP/S from the HP series and illustrates the
dose-dependent effect of HP size on the amount of the NP/S protein
detectable. With this relatively insensitive technique, expression of
NP/S from behind HP20, HP19, and HP18 is undetectable. A band is
detectable from the HP17 construct, the intensity of which is 30%
of the band from HP0. The expression level of protein from HP16 and
HP14 falls between that of HP17 and HP0; however, they are marginally
different from each other. When L-Kb cells were
metabolically labeled for 7 h during infection with the HP series,
lysed, and subjected to immunoprecipitation, similar results were
observed (Fig. 2B). It is noteworthy that the expression
level of wild-type NP detected by this technique is significantly
higher than NP/S, likely because the insertion of the
OVA257–264 epitope into NP causes some degree of
misfolding and destabilization. Because neither Western blot nor
immunoprecipitation detected protein from constructs with larger HPs, a
more sensitive method was employed. The gene encoding firefly
luciferase was cloned behind the HP series, and enzymatic activity from
cell lysates was examined. Fig. 2C shows luciferase activity
in L-Kb cell lysates measured after an 8-h
infection. Activity is detectable from all HP constructs including HP19
and HP20. In all cases, the amount of enzyme activity correlates with
the size of the HP. Measurement at other time points after infection
including 2, 4, 6, and 24 h showed similar titration curves and
enzyme activity was detectable above background for all constructs at
all time points (data not shown). The amount of enzyme activity
generated by a given HP vac appeared to plateau at later time points,
indicating that distinct steady-state levels of protein were reached
for each construct (data not shown). Fig. 2D shows the
levels of luciferase activity detected from spleen cells of mice
infected 2 days previously with 107 pfu of the
indicated luciferase HP vacs. The relationship between HP size and Ag
expression is maintained in vivo. Thus, trace protein expression occurs
even from HP19 (both in vitro and in vivo) and HP20 (in vitro), though
not detectable by Western blot or immunoprecipitation.
|
Cell surface expression of
OVA257–264/H-2Kb complexes
can be detected by a recently described Ab specific for this complex
(37). L-Kb cells were infected with
5 pfu/cell of the various HP constructs and examined by flow cytometry
for expression of cell surface
OVA257–264/Kb complexes in
parallel with microbeads coated with a known amount of anti-mouse
IgG Ab (see Materials and Methods). Fig. 3A shows the number of
OVA257–264/H-2Kb complexes
detected by this Ab 8 h after infection of
L-Kb cells with the indicated HP construct. Cell
surface epitope/class I complexes can be detected from all constructs
and a clear titrating effect of epitope complexes expressed at the cell
surface is seen which mirrors the expression of intracellular Ag as
detected by the techniques described above. The relationships hold at
all time points tested including 6, 10, and 14 h postinfection and
the level of cell surface epitope expression appears to reach steady
state by 10 h postinfection as no construct generates a
significant increase between 10 and 14 h postinfection (data not
shown). As can be seen, nearly 60,000 cell surface epitope complexes
are produced after infection with minigene, in good agreement with
previous estimates (36, 37), whereas only half that number
is produced from the full-length protein without a HP and HP20 allows
just a few hundred complexes to reach the cell surface. To test whether
this relationship holds in vivo, mice were infected i.p. with selected
HP vacs at a dose of 107 pfu/mouse. Approximately
24 h later spleens were removed, homogenized, and stained for
OVA257–264/Kb. Fig. 3B shows the mean fluorescence of spleen cells stained for
expression of
OVA257–264/Kb complexes.
Though the overall staining was notably lower than for
L-Kb cells infected in vitro, the relationship
between HP size and epitope expression levels was similar. The
mechanisms of CTL priming in vivo during vac infection are unknown, but
may include epitope expression on dendritic cells as a result of
cross-priming (50) or direct infection because dendritic
cells are susceptible to vac infection in vitro (Ref. 50 ;
and M. Hsieh and L. C. Eisenlohr, unpublished observations). We
attempted to assess epitope expression on dendritic cells ex vivo by
costaining DEC-205+ cells with the
OVA257–264/Kb-specific Ab;
however, measurement of any epitope expression on the small population
of splenic DEC-205+ cells was below the detection
limits of this approach (data not shown). Nonspecific staining of
uninfected spleen cells in H-2b mice (see Fig. 3B, Unprimed) has been previously noted for this Ab
(37). Staining at time points later than 24 h
postinfection yielded no signal and may reflect nonspecific effects of
vac infection as has been described previously (37). Thus,
it is possible to generate a range of cell surface epitope densities in
vitro that exceeds 200-fold. Importantly, the relationship between
epitope expression levels achieved using different HP vacs observed in
vitro is maintained in vivo.
|
As a first assessment of the impact that varying epitope density
has on T cell recognition, the T cell hybridoma B3Z was employed. This
hybridoma is specific for the OVA257–264 epitope
in the context of H-2Kb and its activation leads
to expression of ß-galactosidase (ß-gal), which has been
transfected under the control of the NF-AT enhancer element
(43). L-Kb cells were infected with
5 pfu/cell of selected HP vacs for 1 h, mixed with B3Z cells, and
incubated overnight. The cells were fixed and assessed for the levels
of ß-gal produced. Fig. 4A
shows that cell surface epitope complexes can be detected by this
hybridoma from all constructs and that overall activation correlates
with the level of epitope expression observed in Fig. 3 up to HP16.
Levels of cell surface epitope density higher than that achieved with
HP16 do not result in further T cell activation. Activation of the
hybridoma occurred at low, but detectable, levels using HP19 and HP20,
consistent with the luciferase expression data from Fig. 2 and cell
surface epitope density shown in Fig. 3. Thus, the impact of the HPs on
cell surface epitope density observed for
OVA257–264 (see Fig. 3) correlates well with the
influence of the HPs on T cell recognition of infected target
cells.
|
, B–E
(upper panels), shows CTL recognition of target cells
infected with the HP series at 10 pfu/cell. Fig. 4, B and
C, shows recognition of the
OVA257–264 and NP50–57
epitopes, respectively, from the NP/S HP series. These curves indicate
a clear effect of cell surface epitope density on in vitro CTL
recognition. As noted for the hybridoma, maximal detection is reached
at a point when cell surface epitope density (for
OVA257–264; see Fig. 3) or intracellular Ag
expression (see Fig. 2) is submaximal. Epitope presentation from HP20
can be observed consistently for the OVA257–264
epitope. Production of the NP50–57 epitope from
HP20 can be detected in some assays and has been previously noted for
wild type NP behind this HP (Ref. 40 , and data not shown).
Recognition of the same target cells by a vac-specific population of
CTL (lower panels) demonstrates that all targets were
infected with the same dose of virus (Fig. 4, B and
C, lower panels). Two additional epitopes were
examined using an older generation NP without the
OVA257–264 epitope. Fig. 4, D and
E, shows CTL recognition the
NP366–374 and the
NP147–155 epitopes from a panel of NP HP vacs.
Of note, the titration curves are not identical for all epitopes. In
particular, NP147–155 is not recognized
maximally until the weakest HP, HP12, is used. This observation is in
good agreement with previous indications that this epitope is
inefficiently processed (36, 44). Thus, in vitro
recognition of target cells infected with the HP series by
heterogeneous populations of CTL specific for
OVA257–264, NP50–57,
NP366–374, and NP147–155,
follows a similar titration curve to the T cell hybridoma B3Z. Graded CTL activity from mice primed with HP vacs
The bulk CTL populations used for the in vitro HP titration are
highly activated and may require significantly fewer cell surface
epitope/MHC complexes to be triggered (22, 23). To test
the influence of epitope density on naive CTL populations in vivo, mice
were immunized with selected viruses at 107
pfu/mouse for C3H mice and 106 pfu/mouse for
C56BL/6 mice. Spleen populations from primed mice were restimulated in
vitro under identical conditions and cytolytic potentials were assessed
by co-incubation with L-Kb cells infected with
the appropriate minigene vac or EL-4 cells transfected with full-length
OVA (EL-4.G7-OVA). Fig. 5, A
and B, shows the NP50–57- and
OVA257–264-specific CTL responses generated
after priming with three HP vacs as well as NP/S (for the
OVA257–264 response) or NP (for the
NP50–57 response) and the appropriate minigene
and minigene control vacs. For both epitopes a clear gradation of CTL
activity is detected after in vivo infection with the HP series. Vacs
expressing the HP20 construct primed for a weak response just above the
control virus, with other viruses priming for successively greater
responses up to and including the minigene for
NP50–57. Unexpectedly, the
OVA(M)257–264 vac generated weaker CTL activity
than HP0, a trend that is repeated in data shown below. To confirm that
the graded responses observed were not due to variation in the dose of
immunizing virus, vac-specific CTL populations were generated for all
primings. The bottom panels of Fig. 5 show that the level of
vac-specific CTL activity was equivalent for all immunizations. These
data support the notion that naive T cells in vivo are more sensitive
to changes in epitope expression, at least at high densities, than
activated cytotoxic T cells in vitro (22).
|
As has been noted previously (1), bulk populations provide only a crude measurement of the relative numbers of CTL induced in vivo, and the impact of epitope expression levels on T cell priming, to this point, could still be attributed to T cell avidity differences. To obtain a more quantitative assessment of the ability of different epitope levels to prime T cells in vivo, LDA and ELISPOT assays were performed.
For LDA mice were primed with HP viruses, spleens removed 14 days
later, and dilutions of splenocytes were restimulated in 96-well plates
with Ag. After 7 days each well was split two ways and tested for
cytotoxicity against peptide-pulsed/unpulsed or vac-infected/uninfected
target cells. Wells were scored as positive or negative for
epitope-specific cytotoxic activity and CTLp frequencies calculated
(see Materials and Methods). Fig. 6 shows the frequency of CTLp determined
by LDA. In either C57BL/6 mice (for OVA257–264)
or C3H mice (for NP50–57), the hierarchy of
responses seen by in vitro T cell recognition and in vivo priming
assays is borne out in the frequency of CTL induced by these
constructs. Fig. 6A shows that the number of
CTLp/106 splenocytes specific for
NP50–57 detected after immunization with control
vac, HP20, HP0, or minigene decreases with the predicted decrease in
epitope expression. CTLp are not detectable after priming with control
vac or HP20, and the minigene induces roughly 6-fold more CTLp than
does HP0. Fig. 6B shows the results for
OVA257–264. In this case, HP20 primes for a
small, but detectable, epitope-specific response and HP17 and HP14
prime for successively larger epitope-specific populations, in
agreement with level of epitope expressed by these constructs. However,
the OVA minigene primes for a slightly smaller epitope-specific
population than HP14 despite expressing a significantly higher cell
surface epitope density in vitro and generating the highest level of ex
vivo epitope-specific staining (Fig. 3), consistent with the less
efficient generation of bulk populations noted above for this
construct. To determine that all mice were primed equivalently, CTLp to
vac were measured. The technical difficulties in using the cytolytic
vac in LDA were overcome by using the nonlytic MVA (45).
The bottom panels of Fig. 6 show the number of vac-specific
CTLp/105 splenocytes from the mice in the
top panels. The numbers are essentially equivalent for all
primings, indicating that differences in
epitope-specific CTLp were not a result of a disparity in the dose of
priming virus or overall anti-vac response.
|
production
after incubation with peptide-pulsed or -infected target cells. Mice
were immunized with HP vacs and spleens removed and epitope- or
virus-specific IFN- ELISPOTs performed. Fig. 7 shows the number of
NP50–57-specific (A and C)
and OVA257–264-specific (B and
D) IFN- spots/106 splenocytes
detected from C3H or C57BL/6 spleen populations 7 days (A
and B) and 14 days (C and D) after
priming with the indicated HP vacs. In either strain of mice the number
of spots/well correlates inversely with the size of the HP, confirming
the results from the bulk priming experiments and LDA. Priming with the
OVA(M)257–264 minigene vac elicits a similar
level of T cell induction as HP0 at day 7, but by day 14 the size of
the population elicited by OVA(M)257–264 is
again smaller than that induced by HP0 and HP14. The lower numbers of
epitope-specific cells detected for the
OVA257–264-specific response compared with the
NP50–57-specific response at day 7 may reflect
the 10 fold lower dose of virus used to prime for
OVA257–264specific responses. Interestingly,
however, the responses are equivalent by day 14. A similar hierarchy of
responses was detected for the
OVA257–264-specific response following priming
with 5 x 106 pfu/mouse (data not shown).
Assessment of equivalent priming was performed by quantitating the
number of vac-specific IFN--secreting cells. The bottom
panels of Fig. 7 show the number of vac-specific IFN-
spots/105 splenocytes is essentially equivalent
for all primings. All results were confirmed in multiple independent
experiments. Thus, in the context of similar anti-viral immune
responses, the level of epitope expressed has a graded effect on the
epitope-specific T cell population until very high levels of epitope
are achieved.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
, 6, and 7). A recent report also achieved the goal of varying
epitope levels in vivo with similar infectious doses of L.
monocytogenes, modulating the amount of epitope generated by
altering processing efficiency through flanking sequence mutations
(39). The authors demonstrated that while 200
epitopes/cell elicited no T cell response, once a threshold for T cell
activation of 1000 epitopes/cell was reached, further increase in
epitope expression to 2000 or 4000 epitopes/cell did not result in
additional recruitment of T cells. This lead to the conclusion that the
effect of varying epitope density on in vivo CTL generation is binary
rather than graded. This would imply that, in many cases, efforts to
alter epitope expression through increased Ag degradation rates
(53, 54, 55, 56) or construction of minigenes (38, 42, 57, 58) with minimal processing requirements would be ineffective at
increasing immunogenicity. Differences between responses to viral and
bacterial infections offer many possible explanations for the
conflicting results presented here. However, because both vaccinia and
Listeria are complex intracellular pathogens that elicit a
significant CD8+ T cell response (59, 60), it seems more likely that the discrepancy is due to the
range of epitope expression achieved. In the present study, we attained
in vitro cell surface epitope densities from 300 to 60,000 copies/cell
for OVA257–264, suggesting that greater T cell
responses would have been observed in the Listeria system
had higher epitope densities been achieved. Alternatively, the
disparity could be a result of the distinct Ags examined which may not
access, or be handled, equivalently by the Ag processing machinery or
may prime for CTL responses by distinct mechanisms. Without
manipulation, NP/S generates 30,000 epitopes/cell in vitro whereas
wild-type p60 results in only 2000 epitopes/cell in vivo.
Alternatively, a dose-dependent effect may exist in the bacterial
system between 200 and 1000 epitopes/cell. Notably, the lower limits of
detectable T cell responses are similar in the two systems, on the
order of a few hundred epitopes per cell, and correspond well with
inferred limits for in vivo naive T cell activation from other systems
(30, 61).
Analysis of epitope-specific T cell frequencies by two separate
functional criteria, LDA and ELISPOT analysis, leads to a similar
conclusion regarding the influence of epitope density on in vivo T cell
population size. The frequency of epitope-specific cells as determined
by ELISPOT in other systems is equivalent to those obtained using MHC
class I/peptide tetramers and intracellular IFN- staining (5, 7) and is 10- to 100-fold higher than frequencies obtained by
LDA (5). In the present system, ELISPOT analysis generates
frequencies 5- to 10-fold higher than LDA with overall numbers in
good agreement with ELISPOT data following minigene vac priming
reported elsewhere (25). These ELISPOT frequencies predict
that epitope-specific CTL would be at the limits of detection of the
flow cytometric techniques using the present HP system. Preliminary
experiments staining for intracellular IFN- confirmed that the
epitope-specific T cell frequencies were low, but detectable, by this
technique and followed a trend similar to the data presented in Figs. 5, 6, and 7 (data not shown). The use of MHC class I tetramers may
allow similar experiments to be performed in the future
(62).
An optimal CTL response is probably determined, in large part, by the size and overall avidity of the resulting population. Adoptive transfer of high avidity clones has been shown to clear virus more efficiently than low avidity clones to the same epitope as has transferring higher numbers of cells of the same epitope-specific CTL line (31, 63). Studies by others suggest that the average avidity of the response is influenced by epitope density. Berzofsky and coworkers (63) have shown that, in vitro, high concentrations of peptide result in low avidity CTL populations and low concentrations, in high avidity CTL. In another study, adoptive transfer experiments led Butz and Bevan (6) to propose that T cells compete for stimulation by cognate peptide on APC in vivo similar to a model proposing competitive exclusion of low avidity T cells during an immune response (64). In the adoptive transfer model the number of professional APC presenting peptide may be small compared with the number of Ag-specific T cells allowing competition between T cells. If this holds under normal conditions (perhaps after several rounds of T cell replication), a prediction of the model is that lower epitope expression levels will lead to stimulation of predominantly high avidity T cells while high expression levels will generate less competition and allow inclusion of lower avidity clones in the response. Together, these data and ours suggest a reciprocal relationship between population size and average avidity with changing epitope levels. A critical question is whether the overall effectiveness of the CTL response remains constant with changing population sizes and avidities or whether an epitope density exists that elicits an optimal combination of the two. However, if such an optimum exists it is likely to depend upon the specific pathogen to which the overall immune response will be directed and the individual epitopes being examined. In addition, variations in the sequences of the TCRs utilized in individual mice may be an important parameter to investigate when populations responding to different levels of epitope are examined at the clonal level. Mouse to mouse variation appears not to impact the overall magnitude, diversity, or cross reactivity of the response (65), but such differences can, in some cases, significantly affect the recruitment of individual clones involved in primary, as well as future, responses (12, 13).
Our results demonstrate that the relationship between epitope density
and CTL population size does not hold at high epitope levels.
Immunization with the OVA minigene vac vs vac expressing full-length
protein generates: 1) bulk restimulated populations that are less
efficient at lysing infected targets (see Fig. 5, HP0), 2) a smaller
OVA-specific LDA frequency (see Fig. 6), and 3) lower ELISPOT
frequencies (see Fig. 7). We speculate that under the conditions
generated by the OVA minigene, the high avidity clones in the
responding population are rendered undetectable by the assays used
here. Infection of mice with high doses of an aggressive strain of
virus can lead to similar decreases in CTL function (66).
Recent reports suggest that this exhaustion may arise as a result of
either functional tolerance or physical deletion. High levels of the
male specific HY Ag have been shown to tolerize Ag-specific CTL in
vivo, leaving the cells present but with reduced ability to proliferate
and altered cytokine secretion profiles in response to TCR stimulation
(67). In contrast, stimulation of CTL lines in vitro with
high doses of peptide results in decreased Bcl-2 expression and cell
death via a TNF-
/TNF-RII apoptosis pathway (68). Future
work is required to understand the mechanism of the decrease in CTL
detected here following priming with virus expressing high levels of
epitope in vivo. This high dose effect was not observed for the
NP50–57 minigene, indicating that it is not a
general effect of minigene constructs. The
NP50–57 minigene vac has been reported to
generate an epitope expression level of 24,000 copies/cell in vitro
(36). Although the epitope density corresponding to the
point of diminishing returns for T cell priming is likely influenced by
the epitope itself, an endoplasmic reticulum-targeted version of the
NP50–57 minigene, perhaps expressing a higher
epitope density than the cytosolic version, does not increase the
epitope specific response at day 7 (data not shown). It will be
interesting to see how this response is regulated at other time points.
As seen in Fig. 7, the kinetics of the response may be regulated by the
level of epitope expression as the difference between the number of
IFN--secreting cells detected following priming with the
OVA(M)257–264 and full-length protein is minimal
at day 7 but dramatic at day 14. Further, in the context of the
extraordinarily high density of a single epitope achieved with
minigenes in vitro (which for OVA257–264 may
represent 50–85% of the total cell surface Kb
molecules (Fig. 3 and Ref. 37), the anti-vac response
remains relatively constant in magnitude at least as detected by the
techniques employed here. Future work utilizing the vac HP system may
provide key insights into issues concerning precursor recruitment,
immunodominance, and the establishment and maintenance of
CD8+ T cell memory. Certainly, the interplay of
all these factors is central to understanding how intracellular
infections are controlled and to optimizing T cell vaccine
strategies.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Laurence C. Eisenlohr, Department of Microbiology and Immunology, Kimmel Cancer Institute, 233 South 10th Street, BLSB 726, Thomas Jefferson University, Philadelphia, PA 19107. E-mail address:
3 Abbreviations used in this paper: vac, vaccinia virus; NP, influenza A PR/8/34 nucleoprotein; NP/S, influenza A PR/8/34 nucleoprotein with the OVA257–264 epitope inserted; HP, hairpin; ß-gal, ß galactosidase; pfu, plaque-forming unit; LDA, limiting dilution analysis; CTLp, CTL precursor; MVA, modified vaccinia ankara; BSS/BSA, balanced salt solution/BSA; MUG, 4-methylumbelliferyl ß-D-glucuronide; PFA, paraformaldehyde; Ara-C, cytosine ß-D-arabinofuranoside; X-Gal, 5-bromo-4-chloro-3-indolyl ß-D-galactopyranoside; CPRG, chlorophenol red ß-gal; ELISPOT, enzyme-linked immunospot; ß2m, ß2-microglobulin.
Received for publication March 30, 1999. Accepted for publication July 22, 1999.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
