Detection of CD4⁺- and CD8⁺-T-Cell Responses to Human Papillomavirus Type 1 Antigens Expressed at Various Stages of the Virus Life Cycle by Using an Enzyme-Linked Immunospot Assay of Gamma Interferon Release

Abstract

Human papillomavirus (HPV) antigens are expressed in epithelial cells at different stages of differentiation, and this may affect how they are handled by the immune system. We assessed the relative immunogenicities of four different HPV type 1 proteins: E6 and E7, which are made early in basal or parabasal cells; E4, which is made suprabasally in differentiating cells; and L1, a late protein which appears in the highly differentiated upper spinous layers. Pools of 15-mer peptides covering the primary sequences of all four proteins were used to screen 15 normal donors in enzyme-linked immunospot assays of gamma interferon release for both CD4⁺- and CD8⁺-T-cell reactivities. CD8⁺-T-cell responses were detected to the L1 protein in 7 of the 15 samples examined. No responses to E6, E7, or E4 were detected. CD4⁺-T-cell reactivities were again detected in 7 of the 15 donors. A broader spectrum of responses to E6 (three of seven), E4 (six of seven), and L1 (three of seven) was apparent, but there was no reactivity to E7. The predominant CD4⁺ response was to E4. Reactivities were seen in some cases to corresponding regions on other common HPV types but were probably due to a multiple infection rather than to a cross-reaction. Antibodies to HPV1 virus-like particles were detected in 12 of the 15 (80%) donors, but antibody status did not correlate with T-cell reactivity. The differences in the relative immunogenicities of the four proteins revealed in this study are discussed in relation to how they may be processed and presented to the immune system by differentiating epithelial cells.

The link between certain types of human papillomavirus (HPV) and malignant disease emphasizes the clinical importance of these viruses and the need to understand how they are normally handled by the immune system. From that understanding, one might be able to design immunotherapies based on T-cell intervention at one stage or another of the disease process. Evidence for increased papilloma incidence in T-cell-immunosuppressed patients strongly suggests that CD4⁺- and/or CD8⁺-T-cell responses play a vital role in controlling infection with these agents (6). This is supported by histological evidence of T-cell infiltration into both cutaneous (7, 26, 37) and mucosal (15) lesions during the spontaneous regression of papillomas. The nature of these immune responses and the mechanism of their initiation are not fully understood.

Epithelial keratinocytes, the natural targets of HPV infection, are nonprofessional antigen-presenting cells (APCs). Under normal noninflammatory conditions they do not express major histocompatibility complex class II or important costimulatory and adhesion molecules such as B7.1 (CD80), B7.2 (CD86), and intercellular adhesion molecule 1 (ICAM-1; CD54). Although they may be capable of delivering antigen-specific signals to T cells, it is difficult to understand how they can provide the costimulatory signals required for full T-cell activation, and they are unlikely to be able to prime either CD4⁺- or CD8⁺-T-cell responses themselves. Primary responses to HPV antigens are more likely to be initiated by Langerhans cells (LCs), the professional APCs within epithelial surfaces which are equipped to capture antigens by macropinocytosis and receptor-mediated endocytosis (34). Humans have ∼10⁹ epidermal LCs which are located above the basal layer of proliferating keratinocytes (3). Their presence in the skin ensures early contact with viruses during infection, and they play a central role in triggering primary antiviral immune reactions (3, 4). How and where LCs access HPV antigens is not obvious since infection with these viruses does not cause cell lysis. During cutaneous infections, virion assembly occurs in the uppermost differentiated cells of the epidermis and, in order to infect a new host, virus particles must be released from cornified cells. This requires the cornified cell envelope, a normally very durable structure, to break apart. In HPV type 1 (HPV1) infections the proteins that comprise the cornified envelope are downregulated or even absent (8), and there is evidence that HPV11-infected differentiating keratinocytes are also morphologically abnormal, being thinner and more fragile than cell envelopes derived from healthy epithelium (9). HPV infections therefore seem to result in epithelial tissue which is more vulnerable and more likely to “leak” viral proteins, and this may be compounded by treatment or trauma. LCs could then access exogenous viral proteins, and after undergoing a maturation step, migrate to regional lymph nodes where the presentation of major histocompatibility complex-antigen complexes, together with costimulatory molecules, leads to T-cell activation (12, 27).

It is relatively easy to understand how CD4⁺-T-cell immunity to HPV could be initiated in this way through the class II processing of exogenous viral antigens by LCs. It is less clear, however, how HPV is able to prime specific CD8⁺ cytotoxic-T-cell (CTL) responses. Since the virus does not infect the APCs themselves, there are presumably no endogenously synthesized viral antigens within the LCs available for the usual class I processing pathway. It is therefore likely that the LCs are capable of processing exogenous proteins to produce class I-binding peptides and cross-prime CTLs. Precisely what form of exogenous antigen is processed in this way is unclear, but it is known that entry into the class I-restricted pathway of antigen presentation other than through the endogenous route is possible. There is evidence that dendritic cells (DCs) can ingest materials, including host cells that express viral antigens, and have them processed and presented by class I products (33). They can also process antigens from dying infected cells by means of class I as efficiently as if they were infected themselves (1, 34). Antigen-specific CD4⁺ T cells are crucial for the cross-priming of these CTL responses. They appear to help by modifying the antigen-loaded APCs, which allows it to become fully competent for CTL priming (5, 14, 32, 35).

The design of effective immunotherapies for HPV should therefore consider both the CD4⁺ and CD8⁺ arms of the T-cell response. Most of the work thus far, however, has focused on CTL responses to the E6 and E7 transforming proteins, particularly of HPV16. This virus is causally associated with cervical carcinoma (40), and these are the only viral antigens constitutively expressed in HPV-positive tumor cells. In contrast, there has been no concerted effort to look at the totality of responses to a single virus type, including responses to viral antigens that are expressed later in the replicative cycle. These latter responses could be therapeutically valuable in targeting the virus replicative lesions from which tumors arise since these express a range of HPV proteins.

In this study we focused on the cutaneous HPV1, which is predominantly associated with benign papillomas occurring on the palmar and plantar surfaces of the hands and feet. These lesions can be remarkably persistent, but most are likely to undergo spontaneous regression eventually (10). Serological evidence suggests that most individuals have been infected (36). By using an enzyme-linked immunospot (ELISPOT) assay of gamma interferon (IFN-γ) release (24), which allows the detection and quantitation of antigen-specific T cells, we assessed the relative immunogenicities of HPV1 antigens to both CD4⁺ and CD8⁺ T cells. We focused on four HPV1 proteins made at different stages of the virus life cycle: the early proteins E6 and E7, the E4 protein, and the late structural antigen L1. These four proteins are expressed in keratinocytes at different points in the movement of cells from the basal layer to the epithelial surface and therefore in cells at different stages of differentiation (13). E6 and E7 are first expressed in undifferentiated basal cells, E4 is expressed in differentiating parabasal or suprabasal layers, and L1 is expressed higher up in the lesion in more differentiated epithelial cells (Fig. 1). During this study the L1 and E4 proteins were found to be particularly immunogenic to CD8⁺ T cells and CD4⁺ T cells, respectively. CD4⁺-T-cell reactivity to E6 was identified in a few cases, but no responses to E7 were apparent. The results from the ELISPOT assays were also related to the serological status of each individual and, where possible, T-cell responses were checked for cross-reactivity against other common HPV types. Information such as this about the relative immunogenicities of different viral proteins will provide a fuller understanding of the spectrum of T-cell responses induced by HPV antigens and of the mechanisms whereby viral antigens expressed within a localized epithelial lesion are accessed by the immune response.

FIG. 1.

Schematic representation of the normal life cycle of HPV in normal differentiating squamous epithelium and the expression patterns of the various viral proteins. After infection of the basal layer, viral DNA is released and maintained as an episome. The early proteins (E1, E2, E5, E6, and E7) are expressed in the basal and suprabasal layers. Viral replication takes place in the more superficial layers where E4 is made. Virion assembly takes place in the most highly differentiated layers with the expression of L1 and L2, the viral capsid proteins.

MATERIALS AND METHODS

Preparation of CD4-enriched and CD8-enriched T-cell populations.

To determine whether responding IFN-γ-secreting T cells were CD8⁺ or CD4⁺, peripheral blood lymphocytes were negatively depleted by using magnetic beads. Leukocyte-enriched buffy coat samples were obtained from normal donors attending the National Blood Transfusion Centre in Birmingham and then diluted in an equal volume of RPMI 1640 medium (Gibco-BRL) containing heparin (20 U/ml). Unfractionated mononuclear (UM) cells were separated by isopycnic centrifugation on lymphocyte separation medium (Lymphoprep; Nycomed). Purified UM cells were washed three times in RPMI 1640 and were used as required or were cryopreserved in liquid nitrogen. CD4⁺- and CD8⁺-T-cell depletions were carried out separately by using antibody-coated magnetic beads (Dynal, Oslo, Norway). Briefly, UM cells were combined with the magnetic beads at a ratio of 4 to 10 beads/cell in a small volume (1 × 10⁷ to 2 × 10⁷ cells/ml) of RPMI 1640 in a fetal calf serum (FCS)-coated tube and rotated at 4°C for 30 min. Depleted cells were removed by magnetic separation, and the remaining enriched populations were washed twice in RPMI 1640 containing 10% FCS and used directly as responder cells in the ELISPOT assay. Fluorescence-activated cell sorting analysis with the dual-labeled reagent anti-human CD8:FITC/anti-CD4:RPE or the isotyped-matched negative control IgG1:FITC/IgG1:RPE (Serotec) was carried out every time to verify the depletions.

ELISPOT assay of IFN-γ release.

Synthetic peptides (15-mers; overlapping by 10 amino acid residues; Alta Bioscience, Birmingham, United Kingdom) covering the entire primary sequences of all four HPV1 proteins (E6, E7, E4, and L1) were used to screen for CD4⁺- and CD8⁺-T-cell responses by using an ELISPOT assay of IFN-γ release (ELISPOT assay for human IFN-γ; Mabtech). The peptides were used initially in pools of four, each at a concentration of 10 μg/ml. Any positive reactivities were checked by using individual peptides from that pool, at 10 μg/ml, in order to identify the peptide sequence(s) recognized. In some cases, peptides covering the corresponding sequences from other common HPV types were also tested to check for cross-reactive responses. ELISPOT wells were always set up in duplicate, and all buffers, including the wash buffer, were passed through a 0.2-μm-pore-size filter.

An anti-human IFN-γ capture monoclonal antibody (1-DIK) was diluted to 15 μg/ml in phosphate-buffered saline (PBS) and used to coat (at 50 μl/well) flat-bottom 96-well nitrocellulose plates (Multiscreen HA plates; Millipore) overnight at 4°C. The plates were washed six times with RPMI 1640 before 100 μl of 10% FCS in RPMI 1640 (FCS medium) was added to each well for 1 h at 37°C to block nonspecific binding. CD4-enriched or CD8-enriched responder cell populations were added at 4 × 10⁵ cells per well in a volume of 100 μl of FCS medium containing the appropriate peptide or peptide pool. After 16 to 20 h of incubation at 37°C, plates were washed six times with PBS-0.001% Tween (wash buffer) and incubated for 2 to 4 h at room temperature with 1 μg of a biotinylated anti-IFN-γ detection monoclonal antibody (7-B6-1)/ml. After six additional washes, streptavidin-alkaline phosphatase (diluted 1/1,000) was added to the wells, and the plates were incubated for 1 to 2 h at room temperature. After six more washes, the assay was performed with the alkaline phosphatase substrate BCIP (5-bromo-4-chloro-3-indolylphosphate)-nitroblue tetrazolium (Bio-Rad, Richmond, Calif.). The wells were washed out with tap water to stop the reaction after ca. 15 min or when spots were clearly visible. Once the plates were dry, only spots with a fuzzy border and a purple color were counted. A negative control well with no peptide and a positive control well containing cells and 0.2% PHA-P (Difco Laboratories) instead of peptide were included for every sample.

HLA typing.

HLA class I and class II tissue typing was carried out on all buffy coat samples by using standard PCR methodology at the National Blood Transfusion Centre, Birmingham, United Kingdom.

ELISA to determine HPV1 antibody levels.

Purified baculovirus-expressed HPV1 virus-like particles (VLPs) were used in an enzyme-linked immunosorbent assay (ELISA) to determine HPV1-specific antibody levels in plasma samples obtained from each donor. Unless otherwise stated, 100 μl of reagent was used per well. VLP preparations were diluted in carbonate buffer (pH 9.0) to 25 μg/ml and used to coat 96-well ELISA plates. Carbonate buffer alone or containing 25 μg of bovine serum albumin/ml was used to coat the background control wells. After overnight incubation at 4°C, all wells were washed out with freshly distilled water and then blocked by the addition of PBS containing 5% dried skimmed milk (200 μl/well) for 60 min at 37°C. Plates were emptied, and human plasma samples were diluted 1/10 in milk buffer (PBS containing 0.05% Tween 20 and 5% milk) added to the appropriate test and control wells in duplicate. After incubation for 60 min at 37°C, wells were washed several times with PBS containing 0.05% Tween 20, and peroxidase-labeled anti-human immunoglobulin G (Sigma) diluted 1/1,000 in milk buffer was added for a further 60 min at 37°C. After a thorough washing as described above, the wells were incubated with peroxidase substrate solution (Sigma) consisting of o-phenylenediamine dihydrochloride (0.4 mg/ml) and urea hydrogen peroxide (0.4 mg/ml) in 50 mM phosphate citrate buffer for 10 to 20 min. The reaction was terminated by the addition of 12.5% sulfuric acid (25 μl) to each well, and the absorbance was read spectrophotometrically at 492 nm. Positive and negative serum controls were included on every plate. Background levels, where no antigen had been bound to the plate, were subtracted from experimental readings for every sample. The antibody levels were quantitated by relating the absorbancies to that obtained from the positive control run under standard conditions on every assay plate. The negative control provided the negative cutoff point.

RESULTS

We examined here the capacity of T cells taken directly from the peripheral blood of 15 healthy donors to recognize peptides covering four HPV1 proteins. We used the ELISPOT assay to detect and quantify HPV-specific T cells that secrete cytokines, in this case IFN-γ, in response to antigenic stimulation. Lymphocytes were incubated with HPV peptide antigens for 16 to 20 h in nitrocellulose microtiter wells which had been coated with anti-IFN-γ antibodies. Secreted IFN-γ was captured by the coated antibody and then revealed with a biotinylated second antibody-streptavidin complex coupled to a chromogenic substrate. Thus, locally secreted cytokine molecules formed spots, with each spot corresponding to one IFN-γ-secreting cell. The number of spots enables the quantitation of IFN-γ-secreting cells specific for a given antigen or peptide.

CD4⁺- and CD8⁺-T-cell depletions.

Fluorescence-activated cell sorting analysis was carried out to check that the depletion step had been successful and in order to quantitate how many responder T cells of the enriched phenotype were present in the well. In nearly all cases there was a depletion of >99% of the appropriate T-cell subset.

T-cell reactivities.

Encouragingly, 8 of the 15 buffy coat samples demonstrated some kind of T-cell reactivity, with 6 of 8 demonstrating both CD4⁺ and CD8⁺ responses. The results with buffy coat sample 13 are shown in Fig. 2 and demonstrate both the CD4⁺ and CD8⁺ reactivities noted during the initial screens with peptide pools (Fig. 2a and b) and the results of subsequent assays with individual peptides from those pools (Fig. 2c). CD4⁺-T-cell responses to pools 5 (within E4), 12 (within E6), 29 (within L1), and 32 (within L1) and CD8⁺ reactivity to pool 27 (within L1) were detected. The CD4⁺ responses to E4 and E6 were narrowed down to two overlapping peptides in each case, but CD4⁺ reactivity to the L1 pools could not be confirmed. The CD8⁺ response was restricted to a single peptide within pool 27 from the L1 protein.

FIG. 2.

Characterization of CD4⁺- and CD8⁺-T-cell reactivities for buffy coat sample 13. Lymphocytes from buffy coat sample 13 were screened in an ELISPOT assay of IFN-γ release against 15-mer peptides covering the entire primary sequences of the HPV1 E6, E7, E4, and L1 proteins. Pools containing four peptides (each at 10 μg/ml) were tested with both CD8-depleted (a) and CD4-depleted (b) populations of cells initially. (c) Positive reactivities were confirmed and characterized by using the same assay with individual peptides from the positive pools (10 μg/ml).

Overall, seven donors showed positive CD4⁺-T-cell reactivities (Table 1). A broad spectrum of responses to the E4 (six of seven), E6 (three of seven), and L1 (three of seven) proteins was apparent, but no reactivity to E7 was detected. Four of the seven samples showed more than one peptide reactive to more than one protein, and there was a predominant CD4⁺ response to E4, with several samples recognizing the same E4 peptide sequences (Table 1). Three of the six samples that recognized E4 detected this antigen only.

TABLE 1.

Summary of CD8⁺- and CD4⁺-T-cell reactivities to HPV1 E6, E7, E4, and L1 proteins^a

Response type and buffy coat sample no.	Protein	Peptides recognized	aa residues	No. of spots/106 CD4+ or CD8+ T cells	HLA class I and II type of donor
CD4 responses
6	E4	GLTDGEDPEVPEVED	61-75	88	DR0401-27, DR1501-06, DQ0302, DQ0602/10/11/13
		EDPEVPEVEDEEKEN	66-80	119	DR0401-27, DR1501-06, DQ0302, DQ0602/10/11/13
		PEVEDEEKENQRPLG	71-85	143	DR0401-27, DR1501-06, DQ0302, DQ0602/10/11/13
		EEKENQRPLGHPDLS	76-90	174	DR0401-27, DR1501-06, DQ0302, DQ0602/10/11/13
8	E4	TTPNSQDRGRPRRSD	36-50	27	DR1101-24, DQ0301/04
		QDRGRPRRSDKDSRK	41-55	9	DR1101-24, DQ0301/04
10	E4	MADNKAPQGLLGLLQ	1-15	264	NA
		APQGLLGLLQYTPTT	6-20	19	NA
	E4	EEKENQRPLGHPDLS	76-90	33	NA
	E4	LEVYTQRLKRDILQD	96-110	25	NA
	E6	ELRCVTCIKKLSVAE	76-90	124	NA
	L1	ATCKYPDYIRMNHEA	231-245	69	NA
		PDYIRMNHEAYGNSM	236-250	17	NA
11	E4	HLYADGLTDGEDPEV	56-70	38	DR0301/04, DR1302, DQ0201, DQ0604-09/12
		GLTDGEDPEVPEVED	61-75	80	DR0301/04, DR1302, DQ0201, DQ0604-09/12
	E4	LLRETLEVYTQRLKR	91-105	43	DR0301/04, DR1302, DQ0201, DQ0604-09/12
		LEVYTQRLKRDILQD	96-110	22	DR0301/04, DR1302, DQ0201, DQ0604-09/12
12	E6	VPEIEEILDRPLLQI	81-95	16	DR0401-27, DQ0301, DQ0302
		EILDRPLLQIELRCV	86-100	11	DR0401-27, DQ0301, DQ0302
		PLLQIELRCVTCIK	91-105	22	DR0401-27, DQ0301, DQ0302
	E6	NRLKAKCSLCRLYAI	126-140	36	DR0401-27, DQ0301, DQ0302
	L1	TNYVGTPSGSMVSSD	296-310	64	DR0401-27, DQ0301, DQ0302
13	E4	LLRETLEVYTQRLKR	91-105	110	DR1501-06, DR1601-05, DQ050104, DQ0602/10/11/3
		LEVYTQRLKRDILQD	96-110	50	DR1501-06, DR1601-05, DQ050104, DQ0602/10/11/3
	E6	KLEVVSNGERVHRVR	111-125	26	DR1501-06, DR1601-05, DQ050104, DQ0602/10/11/3
		SNGERVHRVRNRLKA	116-130	35	DR1501-06, DR1601-05, DQ050104, DQ0602/10/11/3
15	E4	LLRETLEVYTQRLKR	91-105	47	DR0102/04, DR1101-24, DQ0301, DQ0501-04
		LEVYTQRLKRDILQD	96-110	43	DR0102/04, DR1101-24, DQ0301, DQ0501-04
		QRLKRDILQDLDDFC	101-115	34	DR0102/04, DR1101-24, DQ0301, DQ0501-04
		DILQDLDDFCRKLGI	106-120	34	DR0102/04, DR1101-24, DQ0301, DQ0501-04
		LDDFCRKLGIHPWSV	111-125	45	DR0102/04, DR1101-24, DQ0301, DQ0501-04
	L1	VFRVRFADPNRFAFG	76-90	31	DR0102/04, DR1101-24, DQ0301, DQ0501-04
	L1	EDQYRFLGSSLAAKC	421-435	25	DR0102/04, DR1101-24, DQ0301, DQ0501-04
		FLGSSLAAKCPEQAP	426-440	28	DR0102/04, DR1101-24, DQ0301, DQ0501-04
		LAAKCPEQAPPEPQT	431-445	31	DR0102/04, DR1101-24, DQ0301, DQ0501-04
CD8 responses
4	L1	CTPASGEHWTSRRCP	166-180	47	A0101, A2301, B1501-35, B4403/04/07, Cw0401
6	L1	IPKVSPNAFRVFRVR	66-80	206	A0201-20, A2402-14, B1501-35, B2703-10, Cw0202, Cw0304
		PNAFRVFRVRFADPN	71-85	245	A0201-20, A2402-14, B1501-35, B2703-10, Cw0202, Cw0304
8	L1	LRGIEIGRGQPLGIG	106-120	77	A0301/02, A2402-10, B3501-21, B4002/06/09, Cw0202, Cw0401
		IGRGQPLGIGITGHP	111-125	34	A0301/02, A2402-10, B3501-21, B4002/06/09, Cw0202, Cw0401
10	L1	RILSTDEYVTRTNLF	26-40	22	NA
	L1	FEISSNQTVTIPKVS	56-70	16	NA
		NQTVTIPKVSPNAFR	61-75	22	NA
		IPKVSPNAFRVFRVR	66-80	29	NA
12	L1	ISMKNNASTTYSNAN	351-365	101	A1101-03, A6801/02, B4001, B5101-08, Cw0303, Cw0304
		NASTTYSNANFNDFL	356-370	37	A1101-03, A6801/02, B4001, B5101-08, Cw0303, Cw0304
	L1	MTQRTATSSTTKRKT	481-495	62	A1101-03, A6801/02, B4001, B5101-08, Cw0303, Cw0304
13	L1	CTPASGEHWTSRRCP	166-180	277	A0101, A2901/02, B3501-21, B4002/06/09, Cw0202, Cw1203
15	L1	YHATSERLLLVGHPL	41-55	126	A1101-03, A3001-04, B3501-21, B5001, Cw0401, Cw0602
		VGHPLFEISSNQTVT	51-65	159	A1101-03, A3001-04, B3501-21, B5001, Cw0401, Cw0602

^aCD4 and CD8 enriched responder cell populations were screened against pools of 15-mer peptides (overlapping by 10 aa) by using ELISPOT assays of IFN-γ release. Positive reactivities were checked against individual peptides within the pools to confirm the response and to narrow down the peptide sequences recognized. The columns from left to right indicate the protein(s) to which responses were detected, the overlapping peptide sequences recognized and their coordinates, the number of spots obtained in each case, and the HLA type of the donor. These are the results of duplicate wells, and any background reactivity in the absence of peptide has been subtracted. NA, not available.

CD8⁺-T-cell responses were detected in 7 of the 15 donors; 2 of these 7 donors recognized more than one peptide sequence. In every case, the reactivity was seen in response to a peptide derived from the L1 protein, and no responses were seen to the other three antigens, E6, E7, or E4. (Table 1). Two samples (samples 4 and 13), which share the HLA-A0101 genotype, both recognized the same peptide (CTPASGEHWTSRRCP).

Frequency of T-cell responses.

The numbers of spots obtained were generally low (9 to 264 spots/10⁶ CD4⁺ T cells; 16 to 277 spots/10⁶ CD8⁺ T cells), but the results are similar to those reported in similar studies with peptides derived from subdominant antigens of Epstein-Barr virus (EBV) (38). Presumably, in instances where no responses were detected there were no HPV-specific T cells or else their detection was below the sensitivity of the assay.

Cross-reaction between common HPV types.

Given the degree of sequence homology between related HPV types, it is possible that the HPV1 peptide screening panel may have picked up cross-reactive responses to other HPV types. This is most likely to be a problem with L1, which is the most highly conserved of the four proteins tested, showing 60 to 70% sequence homology to L1 proteins from related cutaneous HPVs. The E6 and E7 proteins only exhibit ca. 40% sequence homology between related virus types, and E4 sequences are highly type specific. We aligned the peptide sequences identified by either CD4⁺ or CD8⁺ T cells within the HPV1 L1 molecule in this study to L1 sequences from other common HPV types, both cutaneous and mucosal, in order to determine whether the responses we obtained are directed at highly conserved regions (Fig. 3). It can be seen that several of the sequences recognized did appear to lie within homologous regions of the L1 molecule with several conserved amino acid residues or blocks of conserved sequence, whereas in other cases very little sequence homology was observed. To investigate the possibility that the ELISPOT assay was picking up cross-reactive T-cell responses, samples were checked against peptide sequences derived from the corresponding regions on other common HPV types (types 2, 6, 16, and 18). These sequences and the results to these experiments are shown in Table 2. For buffy coat 12, CD4 responses to HPV1 L1 (amino acids [aa] 296 to 310) and HPV1 E6 (aa 126 to 140) and the CD8 response to HPV1 L1 (aa 351 to 370) were investigated. The CD4 L1 response to the corresponding peptide from HPV6, the CD4 E6 response to that from HPV16, and the CD8 L1 response to that from HPV2 were determined. For buffy coat 13, the CD4 response to HPV1 E6 (aa 111 to 130) and the CD8 response to HPV1 L1 (aa 166 to 180) were checked and, with this donor again, peptides from other virus types were recognized. CD4⁺ reactivity was detected with the corresponding HPV1 E6 sequence from HPV2, and CD8⁺ T cells recognized peptides derived from HPV types 6, 16, and 18. In the case of buffy coat 15, the CD4 response to HPV L1 (aa 76 to 90) was also seen with peptides derived from HPV types 2, 6, and 18.

FIG. 3.

Alignment of HPV L1 protein sequences. HPV1 L1 sequences recognized in ELISPOT assays of IFN-γ release were aligned with those of other commonly found HPV types. Related cutaneous viruses (HPV types 2, 4, 5, and 8) and common mucosal types (HPV types 6, 11, 16, 31, 18, 33, and 45) were also included. Conserved amino acids or blocks of conserved sequence are shown in boxes.

TABLE 2.

ELISPOT responses obtained with peptides covering regions on HPV types 2, 6, 16, and 18 that correspond to the HPV1 sequence recognized^a

Buffy coat sample no.	Type of T-cell response	HPV1 sequence recognized (aa)	Aligned sequences from corresponding regions on other common HPV types		Response
			HPV	Sequence
12	CD4	L1 (296-310)	HPV1	TNYVGTPSGSMVSSD	Yes
			HPV2	HVYTSTPSGSMVSSE	No
			HPV6	SIYVNTPSGSLVSS	Yes
			HPV16	SNYFPTPSGSMVTSD	No
			HPV18	SAFFPTPSGSMVTSE	No
	CD4	E6 (126-140)	HPV1	NRLKAKCSLCRLYAI	Yes
			HPV2	NISGRWTGHCMNCGS	No
			HPV6	FIKLNCTWKGRCLHC	No
			HPV16	HNIRGRWTGRCMSCC	Yes
			HPV18	RFHNIAGHYRGQCHS	No
	CD8	L1 (351-370)	HPV1	IS--MKNNASTTYSNANFNDFL	Yes
			HPV2	LC-ATEA-SDTNYKATNFKEYL	Yes
			HPV6	LC-ASVT-TSSTYTNSDYKEYM	No
			HPV16	LC-AAISTSETTYKNTNFKEYL	No
			HPV18	ICASTQSPVPGQYDATKFKQYS	No
13	CD4	E6 (111-130)	HPV1	KLEVVSNGERVHRVRNRLKA	Yes
			HPV2	WEEKEALLVGNKRFHNISGR	Yes
			HPV6	LCEVEKVKHILTKARFIKLN	No
			HPV16	CPEEKQRHLDKKQRFHNIRG	No
			HPV18	PLNPAEKLRHLNEKRRFHNI	No
	CD8	L1 (166-180)	HPV1	CTPASGEHW-TSRRCP	Yes
			HPV2	CKPPIGEHWSKGTTCN	No
			HPV6	CAPPLGEHWGKGTQCT	Yes
			HPV16	CKPPIGEHWGKGSPCT	Yes
			HPV18	CAPAIGEHWAKGTACK	Yes
15	CD4	L1 (176-90)	HPV1	VFRVRFADPNRFAFG	Yes
			HPV2	VFHVKLPDPNKFGLP	Yes
			HPV6	VFKVVLPDPNKFALP	Yes
			HPV16	VFRIHLPDPNKFGFP	No
			HPV18	VFRVQLPDPNKFGLP	Yes

^aThe possibility that the ELISPOT assays may be picking up cross-reactive responses to other HPV types was investigated with cells from donors 12, 13, and 15. Samples were checked against peptides covering regions on HPV types 2, 6, 16, and 18 that corresponded to the HPV1 sequences recognized. These peptide sequences are shown in the fourth column of the table, and the last column indicates whether the peptide was recognized in each case.

HPV1-specific antibody levels.

Antibodies to HPV1 VLPs were detected in 12 of the 15 (80%) plasma samples (Fig. 4). These results, along with previous findings in this laboratory (36), suggest that the vast majority of normal individuals have been infected with HPV1. Of the 12 seropositive samples obtained in this study, 7 showed either CD4⁺- or CD8⁺-T-cell reactivity to HPV1 peptides in the ELISPOT assays. Only one sample out of the three with negative serology gave a T-cell response.

FIG. 4.

Relative HPV1-specific antibody levels obtained by using an ELISA that employed HPV1 VLPs. The antibody levels were quantitated by relating the absorbancies obtained for each sample to that obtained from the positive control run under standard conditions on every assay plate. The negative control provided the negative cutoff point (—). Background readings, where no antigen had been bound to the plate, were subtracted from experimental readings in every case. Both the CD4⁺- and CD8⁺-T-cell reactivities detected by the ELISPOT assays are indicated for each sample.

DISCUSSION

By using lymphocytes isolated from normal buffy coat samples in ELISPOT assays of IFN-γ release against peptide panels, we established that there are differences in the relative immunogenicities of the four HPV1 antigens E6, E7, E4, and L1 (Table 1). These proteins are expressed at various stages of the virus life cycle and therefore in epithelial cells at different stages of differentiation (Fig. 1). The L1 protein was found to be particularly immunogenic to CD8⁺ T cells, and no CD8⁺ responses to E6, E7, or E4 were observed. The E4 protein appeared to be particularly immunogenic to CD4⁺ T cells. CD4⁺-T-cell reactivity to L1 and E6 was also identifiable in a few cases, but no responses to E7 were demonstrated.

A major finding was a dominant T-cell response, both CD4⁺ (three of seven) and particularly CD8⁺ (seven of seven), to peptides covering the HPV1 L1 protein. Since the L1 protein, which is 508 aa, contains more unique sequence than E6 (140 residues), E7 (98 residues), or E4 (125 residues), it may be predicted to contain more epitopes. The evidence suggests, however, that L1 is a highly immunogenic protein. It has recently been shown that fully assembled papillomavirus VLPs alone, without the addition of exogenous cytokines, can induce the phenotypic and functional maturation of DCs, resulting in the induction of both CD4⁺ and CD8⁺ primary T-cell responses (30). We propose that the ability of VLPs to activate DCs in this way is likely to be related to the ordered repetitive array of virion surface proteins and their interaction with as-yet-unidentified DC surface receptors. Receptor-mediated antigen internalization by immature APCs in this way is probably much more efficient for the uptake and accumulation of proteins than macropinocytosis, which is less specific. The differences in the immunogenicity observed between L1 and the other HPV1 proteins in this study may therefore be due to the fact that this antigen probably contains more epitopes due to its size and also that DCs are particularly efficient at L1 processing and presentation.

The L1 protein is highly conserved, and HPV1 L1 shows 60 to 70% sequence homology to other L1 proteins from related cutaneous HPVs (16). It is therefore possible that the peptide panel employed in the ELISPOT assays may pick up cross-reactive responses to other common HPV types. T-cell cross-reactivity usually requires conserved blocks of sequence identity, and this would be expected to restrict the number of epitopes that would be shared between virus types. The example of EBV is instructive here in that the immunodominant EBNA 3A, 3B, and 3C proteins are 80 to 90% homologous between EBV types 1 and 2, and yet more than half of the defined CD8 epitopes to date are type specific (31). Nevertheless, a comparison of the HPV1 L1 sequences recognized in this study with analogous regions on the L1 molecules of other common HPV types (Fig. 3) has revealed that some of the peptide sequences did cover homologous regions of the protein and contained conserved residues and blocks of sequence. The question of cross-reactivity required further investigation and so T-cell responses obtained from buffy coat samples 12, 13, and 15 were checked in ELISPOT assays against peptides covering corresponding sequences from other common HPV types (types 2, 6, 16, and 18). In all cases, responses to peptides derived from the other virus types were observed (Table 2). It is difficult to determine whether this represents cross-reaction or simply reflects the fact that the donors may have been infected with more than one of these common virus types. Some of the responses observed were to regions on the protein with very little sequence homology and so are highly unlikely to represent cross-reaction (buffy coat 12, E6 [aa 126 to 140] and L1 [aa 351 to 370]; buffy coat 13, E6 [aa 111 to 130]). The CD4 response to HPV1 L1 (aa 296 to 310) also shown by buffy coat 12 was to a fairly homologous region, but an additional response was observed only against the corresponding sequence from HPV6. Again, we feel that this is unlikely to be a cross-reaction since such responses against the very similar peptides derived from the other virus types would be expected. However, when responses to multiple virus types against regions which share blocks of sequence homology (buffy coat 13, L1 [aa 166 to 180]; buffy coat 15, L1 [aa 76 to 90]) are detected, then the possibility of cross-reaction cannot be ruled out. In conclusion, our results demonstrate that in most cases we were unlikely to detect cross-reactive T-cell responses with peptide panels in ELISPOT assays, and reactivities to corresponding regions on other HPV types were probably due to multiple infections. However, it is not possible to rule out cross-reaction in some cases without knowing precisely which types of HPV have been encountered by those donors. From the point of view of designing effective HPV-specific peptide vaccines, establishing the degree of T-cell cross-reaction is important. It would be an advantage to have a formulation that would provide protection against multiple virus types.

It is not clear what determines whether a CD4⁺ or CD8⁺ response to HPV proteins is preferentially induced. There is little doubt that tissue destruction effectively increases the amount of antigen available for cross-presentation. Both apoptotic and necrotic cells appear to be favorite targets for APCs and cross-priming. However, in some experiments DCs have been shown to cross-present both apoptotic and necrotic cells to class II-restricted T cells, but only apoptotic cells provide a source of class I-restricted determinants (1, 21, 25). The reason for the discrepancy between the antigenic requirements for the class I versus class II pathways of cross-presentation is unclear, but it is probably reasonable to conclude that the extent of cross-priming and cross-presentation of virally derived proteins will depend on the health of the cells expressing them. This could well have relevance for the priming of responses to HPV1 antigens being expressed in epidermal keratinocytes undergoing processes akin to apoptosis during terminal differentiation or undergoing necrotic changes or tissue destruction after trauma or treatment.

This study has revealed major differences in T-cell reactivities to the HPV1 L1 and E4 proteins. L1 appears to induce a very dominant CD8⁺-T-cell response, whereas E4 evokes a stronger CD4⁺-T-cell response. We also demonstrated CD4⁺-T-cell reactivity to HPV1 E4 in a previous study (36). Both L1 and E4 are expressed in large amounts by suprabasally differentiating keratinocytes (8, 18, 19), but they appear to be processed very differently, L1 preferentially through the class I processing pathway and E4 through the class II processing pathway. This may be due to differences in how the protein is seen by the immune system upon initial exposure. E4 is predominantly cytoplasmic, whereas L1 is a nuclear protein and comprises part of the virus particle. There may well be substantial differences in the accessibility of these proteins to LCs, possibly in how they would be internalized, and therefore in how T-cell responses would be primed initially. The HPV1 E4 protein is interesting in that it exists as multiple species that are expressed in different layers of the infected epithelium. The primary product is a 17K polypeptide, and the other species are derived from this by progressive proteolytic cleavage at the N terminus to give 16K, 11K, and 10K species and/or by dimerization to give 34K, 32K, 23K, and 21K species (17). Only the 17K species is expressed in the lower layers of an HPV1-induced lesion, with the smaller products and multimers thereof, becoming apparent during the latter stages of differentiation. In fact, all but two of the E4 peptide sequences identified lie at the highly conserved C terminus of the protein and would be present in all eight of the E4 species. It is interesting to consider how the proteolytic processing of E4 relates to its antigenic processing and presentation.

An abundant protein such as HPV1 L1 or E4 has a higher chance of yielding antigenic peptides, and it is likely that the level of antigen expressed by a cell may have to be above a certain threshold for cross-priming to occur. This may explain the finding here that T-cell responses to the E6 protein were less abundant and those to E7 were nonexistent, since their levels of expression are so low. Cross-priming to these proteins may be inefficient because the low antigen concentration makes transfer of antigen from the antigen-expressing cell to the APC inefficient. There is experimental evidence for this in animal models, in which immunization with tumor cells expressing the HPV16 E7 protein was not sufficient to induce a CTL response, whereas peptide vaccination did induce a CTL response (20). It has also been observed that transgenic mice expressing HPV16 E6 or E7 epidermally remain immunologically naive to these proteins at the B-cell and T-helper-cell levels. This was explained not only by the fact that they are expressed at low levels but also because they are expressed peripherally in the epidermis, away from the central immune system (2, 23). Antigen expression levels are likely to be one of the major factors determining several types of immunity (22). With particular relevance to HPV, it is probable that the site, amount, and timing of expression may all play a role in dictating whether these peripherally expressed viral antigens induce a positive immune response, a tolerizing response, or whether the immune system remains ignorant of the antigen.

We have detected antibodies to HPV1 in 80% of the individuals involved in this study, thus enabling the identification of donors likely to have encountered the virus and presumably more likely to have T-cell reactivity in the ELISPOT assays (Fig. 4). However, the antibody status of an individual did not seem to be a good predictor of T-cell reactivity, since only 7 of the 12 samples with positive serology gave a T-cell response. It is difficult to draw conclusions about the seronegative donors due to the lag phase of several months between infection with the virus and the appearance of antibodies in the blood (11) and the fact that antibody levels eventually diminish. Thus, negative serology does not mean that the individual has never been infected or is unlikely to possess HPV1-specific memory T cells. Reassuringly, of the eight donors who did give T-cell responses to peptides in the ELISPOT, seven were seropositive. However, of the seven samples that showed no T-cell reactivity, five possessed HPV1-specific antibodies. This discrepancy underlines the fact that it is difficult to assess the interrelationship between T-cell reactivity and antibody status to HPV because we know surprisingly little about the natural history of HPV infection and the relative timescales of the T- and B-cell responses.

There has been a revolution in our understanding of virus-specific T-cell responses due to the development of more accurate methods for measuring them. The 24-h ELISPOT assay has distinct advantages over more traditional methods, such as limiting-dilution analysis (LDA) and chromium release cytotoxicity assays. Chromium release assays are rarely sensitive enough to look directly at ex vivo blood samples and, like LDA, generally require an in vitro stimulation step in order to detect antigen-specific CTLs. The ELISPOT assay is technically much less demanding and is considered more sensitive than both LDA and chromium release assays for studying populations of T cells present at low frequency (28, 29). One disadvantage, however, is that the frequency of antigen-specific T cells can be underestimated if some cells are nonfunctional or secrete different cytokines. The numbers obtained here are relatively low, but this was not unexpected since HPV-specific T cells are only present in the peripheral blood in low numbers, and it has been notoriously difficult to demonstrate their presence ex vivo. It also has to be noted that if cross-reactive responses are being detected, particularly in the case of the L1 protein, then the numbers of spots obtained may be cumulative and may reflect the fact that the individual has been infected with more than one virus type. This is the first report of the ELISPOT assay being used to characterize and quantitate human T-cell responses to HPV directly ex vivo, and the results are encouraging since the protocol employed here is obviously capable of their detection. This assay used no restimulation or in vitro manipulations, so the results are likely to reflect the in vivo repertoire. The assay would be well adapted for monitoring immune responses to HPV-specific vaccines and immunotherapies, since it is sensitive and versatile, can be performed directly ex vivo, and uses relatively small numbers of T cells. It could also be applied to the design of peptide-based vaccines for the identification of appropriate CD4⁺- and CD8⁺-T-cell epitopes.

The ultimate goal of effective immunotherapies for HPV would be the stimulation or reactivation of HPV-specific CTLs. Whether or not these T cells can recognize and kill their targets is largely dependent upon the ability of the infected keratinocyte to process and present the relevant antigen appropriately on the cell surface. In the case of HPV, there is evidence that there are problems with antigen presentation for a variety of reasons, and we obviously need to circumvent these problems in order to design effective T-cell immunotherapies. One aspect we sought to investigate further was whether the differences in relative immunogenicities revealed in this study reflect changes in the antigen-processing and presenting function of epithelial cells at different stages of differentiation.

This study has contributed to our understanding of how these epitheliotropic viruses are seen and controlled by the immune system by providing information about the relative immunogenicities of HPV antigens and particular peptide sequences. Of the many thousands of peptides encoded by a complex foreign antigen that can potentially be presented to T cells, only a small fraction of them induce measurable responses. Typically, only one (or a few) potential epitopes elicit strong immunodominant CTL responses. A few others may elicit weak (subdominant) responses, while the majority elicit no responses (cryptic epitopes) (39). A thorough understanding of this immunodominance, as well as establishing the precise role that immune responses play in the natural history of HPV infection, is imperative for the rational design of vaccines meant to elicit CTL responses.