Split tolerance to a viral antigen expressed in thymic epithelium and keratinocytes

0014-2980/98/0909-2791$17.50+.50/0© WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

Split tolerance to a viral antigen expressed inthymic epithelium and keratinocytes

Ian H. Frazer1, Germain J. P. Fernando1, Nina Fowler1, Graham R. Leggatt1, Paul F.Lambert2, Amy Liem2, Karen Malcolm1 and Robert W. Tindle3

1 Center for Immunology and Cancer Research, University of Queensland, Department ofMedicine, Princess Alexandra Hospital, Woolloongabba, Queensland, Australia

2 McArdle Institute for Cancer Research, University Avenue, Madison, USA3 Sir Albert Sakzewski Virus Research Laboratories, Royal Children’s Hospital, Herston,

Queensland, Australia

When expressed as a transgene from the keratin 14 (K14) promoter in an MHC class II-deficient mouse, I-Ab expressed in thymic cortical epithelium promotes positive but not neg-ative selection of I-Ab-restricted CD4+ T cells (Laufer, T. M. et al., Nature 1996. 383:81–85).Transgenic mice expressing the E7 protein of human papilloma virus 16 from the K14promoter were studied to determine the consequence of expression of a cytoplasmic/nuclear protein from the K14 promoter. K14E7-transgenic mice express E7 in the thymusand skin without evidence for autoimmunity to E7. Repeated immunization of FVB(H-2q) orF1(C57BL/6J×FVB) mice with E7 elicited similar antibody responses to the defined B cellepitopes of E7 in K14E7-transgenic and non-transgenic animals. In contrast, for each gen-etic background, a single immunization with E7 elicited demonstrable T cell proliferativeresponses to the major promiscuous T helper epitope of E7 in the transgenic but not thenon-transgenic animals. Further, E7-immunized non-transgenic F1 (FVB×C57BL/6J) ani-mals developed strong E7-specific cytotoxic T lymphocyte (CTL) responses and were pro-tected against challenge with E7+ tumors, whereas similarly immunized K14E7-transgenicanimals had a markedly reduced CTL response to E7 and no E7-specific tumor protectionwas observed, although the antibody and CTL response to ovalbumin was normal. Expres-sion of E7 protein as a transgene from the K14 promoter in the skin and thymus thus inducesE7-specific tolerance in the cytotoxic T effector repertoire, together with expansion of theE7-specific T helper repertoire. These findings demonstrate that limited tissue distribution ofan autoantigen may result in “split” tolerance to that autoantigen.

Key words: Thymus / Skin / Repertoire development / Autoimmunity / T lymphocyte

Received 23/3/98Revised 15/6/98Accepted 18/6/98

[I 18264]

Abbreviations: GST: Glutathione S-transferase HPV:Human papilloma virus K14: Keratin 14

1 Introduction

The T cell repertoire is determined by contact of devel-oping T cells with antigen/MHC complexes in the thymus(“central” regulation), and by the subsequent exposureof cells surviving thymic selection to antigen presentedon somatic cells without specialist antigen presentationfunctions (“peripheral” regulation). For MHC class I-restricted CD8+ T cells, general expression of antigeninduces centrally determined deletional tolerance ofhigh-affinity T cells [1], probably as a consequence ofexpression on thymic medullary dendritic cells [2]. Thisprocess leaves a population of cells with low affinity forself protein [3] that can be activated to promote transient

autoimmune reactivity [4]. Peripheral expression of anti-gen can result in deletion or anergy of the correspondingCTL repertoire, or can leave the repertoire unaltered(“immunological ignorance”) [5–8]. For CD8+ T cells,ignorance or reversible anergy seems the most commonoutcome of strictly peripheral self antigen presentation,suggesting that control of potential autoimmune cyto-toxic T cell function may rest largely with the induction orotherwise of autoantigen-specific T help [9]. For MHCclass II-restricted Th cells, the consequence of general-ized expression of antigen depends on the level of anti-gen expression in the thymus. Lower levels of antigenexpression are ignored, intermediate levels produceclonal anergy of antigen-specific CD4+ cells, and higherlevels produce deletional tolerance [10, 11]. Peripheralexpression of antigen in the context of MHC class II canhave a similar range of outcomes [11]. A special situationarises when alloantigen expression is restricted to a sub-

Eur. J. Immunol. 1998. 28: 2791–2800 Split tolerance to thymic self-antigens 2791

set of thymic epithelial cells by use of the Keratin 14(K14) promoter. This pattern of expression results in pos-itive selection of a CD4+ population specific for thealloantigen [12]. Expression of a cytoplasmic antigen inthymic cortical epithelial cells might therefore also beexpected to have a significant effect on the MHC class II-restricted antigenic repertoire, as the nature of the pep-tide within the MHC class II molecule plays a role inshaping the MHC class II-restricted T cell repertoire [13].

The E7 protein of human papilloma virus (HPV) 16, whenexpressed ectopically in skin as a transgene from the§ A-crystallin promoter [14], has little effect on the E7-

specific Th or CTL repertoire of the FVB and F1(FVB×C57BL/6J) mouse. This transgene leaves the ani-mal “naive” to E7, and able to respond immunologicallyto increased endogenous levels of expression of E7, aswell as to exogenously administered E7 protein [15]. Thisobservation contrasts with the antigen-specific toleranceinduced by the Kb antigen expressed as a transgene inthe skin [7], and suggests that the consequences for theimmune repertoire may be different for alloantigen andcytoplasmic antigen expressed by keratinocytes. Toinvestigate the consequences for the immune repertoireof expression of a conventional cytosolic antigen inthymic and skin epithelium, we studied a line of micetransgenic for the E7 protein of HPV16 driven from theK14 promoter [16]. Expression of E7 in skin and thymicepithelium was found to result in functional E7-specificCTL tolerance and enhanced E7 Th responsiveness.These results suggest that endogenous expression ofantigen in thymic and peripheral epithelial cells can resultin “split” tolerance to a self antigen. These findings haveimplications for immunization against vertically transmit-ted epitheliotropic viruses, and for the pathogenesis ofimmune complex-mediated autoimmunity.

2 Results

Mice transgenic for the E7 protein of HPV16 driven fromthe K14 promoter (K14E7.FVB) express E7 protein in thethymus and skin keratinocytes from birth, with maximalexpression at 2–4 weeks [16]. They develop a skin phe-notype characterized by hypertrophy and tumor devel-opment, but without the inflammatory skin disease seenin #19 ( § A-crystallin. HPV16E6E7.FVB) mice [17]. Theselatter mice express low levels of E7 protein in skin, anddevelop humoral and cellular immunity to E7 protein inassociation with the onset of skin inflammation [15, 18].To assess the immunological consequences of theexpression of the E7 transgene at higher levels inskin and thymus, K14E7-transgenic FVB and F1(FVB×C57BL/6J) mice were sham immunized or immu-nized with a range of E7 proteins and peptides, and the

humoral and cellular immune response to the E7 proteinassessed in vivo and in vitro in contrast with non-transgenic littermate controls.

Groups of K14E7-transgenic mice and FVB littermatecontrols were immunized with E7 as a glutathione S-transferase (GST) fusion protein, with pooled peptidescomprising the whole length of E7 [18] or alternativelywith a polynucleotide vaccine encoding the E7 protein.Antibody responses to the E7 protein and to peptidesincluding the known B cell epitopes of E7 were assessedby ELISA (Table 1). FVB mice developed antibodies to E7protein and to the major characterized B cell epitopes ofE7. Antibodies to E7 were efficiently elicited by immu-nization with GSTE7 protein. K14E7-transgenic micedeveloped antibody responses similar to those of FVBmice, and antibodies to the same major B cell epitopeswere seen following immunization with BT12D, a syn-thetic peptide comprising the promiscuous Th epitope ofE7 and the three major B cell epitopes [19, 20], confirm-ing that there was no demonstrable tolerance to thedefined B cell epitopes in the E7 protein. Antibodies toE7 were also elicited by an E7 polynucleotide vaccine(FVB mice: A=0.26±SEM 0.18; K14E7 mice,A=0.66±0.39), suggesting that the E7-specific Th com-partment was also functionally intact. In similar experi-ments, F1(C57×K14E7) transgenic mice immunized withGSTE7 developed antibody titers to E7 (A=1.52±0.22SEM) which were comparable to those of F1 (C57×FVB)mice (A=1.48±0.31) and demonstrated antibodies to thesame E7 peptides whose sequence included the domi-nant B cell epitopes (data not shown). To examine forimmune deviation resulting from expression of the trans-gene, the quantity of E7-specific IgG2a was comparedfollowing immunization in E7-transgenic and non-transgenic mouse lines, and no significant differenceswere observed (Table 2). To establish whether expressionof E7 protein in the skin elicited spontaneous humoralimmunity to E7 in K14E7 mice, as seen in the older #19( § A-crystallinE7) transgenic mice after the onset of skindisease, unimmunized K14E7-transgenic mice wereobserved for development of antibodies to E7 over an11-month period, and none were observed. Thus the Bcell compartment seems ignorant of E7 proteinexpressed in the skin and thymus from the K14 promot-er, and there is functional evidence for at least some E7-specific Th cells in the K14E7-transgenic mice.

The above results suggested that the immune repertoirespecific for the major promiscuous Th epitope (...DRA-HYNIVTF...) of E7 was likely to be at least partially pre-served in K14E7-transgenic mice. To confirm this,K14E7-transgenic FVB and F1 (FVB×C57BL/6J) micewere immunized with E7, and draining lymph node andspleen cells assessed for proliferation in response to E7

2792 I. H. Frazer et al. Eur. J. Immunol. 1998. 28: 2791–2800

Table 1. Antibodies to B cell epitopes of HPV16 E7 in K14E7.FVB and FVB mice immunized with various preparations ofHPV16E7a)

Reactivity by ELISA with antigen (absorbance)

Recallantigen

E7 GF101-103b) GF104–106 GF107–109 Blank

Mousetype

FVB K14 FVB K14 FVB K14 FVB K14 FVB K14

Immuno-gen

GSTE7 1.41±0.03 1.52±0.07 1.62±0.09 1.67±0.15 1.23±0.42 0.87±0.22 1.19±0.19 0.62±0.34 0.03±0.07 0.04±0.01

BT12Dc) 0.95±0.41 1.72±0.08 0.70±0.31 1.54±0.11 0.01±0.01 0.01±0.00 0.01±0.01 0.00±0.00 0.01±0.01 0.00±0.00

BSA 0.30±0.14 0.30±0.14 0.24±0.06 0.24±0.06 0.03±0.01 0.03±0.12 0.15±0.02 0.15±0.02 0.14±0.05 0.14±0.05

a) All immunogens were given three times in Freund’s adjuvant, and blood collected 14 days after the last immunization.Results are from pooled sera from groups of five mice. Similar results were obtained with E7 given with Quil A adjuvant.

b) 101–109 are a series of 20mer peptides scanning the length of the E7 ORF.c) BT12D is a peptide comprising the promiscuous Th epitope of HPV16 E7 and three defined linear B cell epitopes.

Table 2. E7-specific IgG subclass antibodies in K14E7 and FVB mice after immunization with E7

Immunogen Mouse IgG2a IgG1 Total IgGa)

E7GST/QA K14 (n=5) 1.35±0.26 0.38±0.06 0.48±0.08

E7GST/QA FVB (n=5) 1.09±0.31 0.61±0.02 0.58±0.07

OVA/QA K14 pooled sera 0.00 0.01±0.01 0.00

OVA/QA FVB pooled sera 0.01±0.01 0.01±0.01 0.01±0.00

a) K14×C57 and FVB×C57 mice were immunized at day 0 and 21 and bled at day 35. Immunogens used were E7GST/QuilA(QA) or as control OVA QA. Sera diluted 1:400 and ELISA assays were performed using plates coated with E7MS2 fusionprotein. The absorbance at 415 nm±SEM for triplicate samples for each individual mouse is shown. Total IgG was determinedon a separate ELISA plate at a serum dilution of 1:4000, and therefore cannot be directly compared with the IgG2a and IgG1data.

protein and peptides. Strong proliferative responseswere induced to the E7 protein by immunization withGSTE7 protein, with similar responses in K14E7-transgenic and FVB littermate control mice (Table 3). Lin-ear peptides from E7 were used as immunogens andproliferative responses could be recalled with peptides(GF105, GF106) incorporating the promiscuous Th epi-tope of E7, confirming that this epitope is also functionalin H-2q mice. Similar responses to the GF106 peptidewere observed in GF106-immunized transgenic and theparent mice. Proliferative responses to pooled peptidesnot incorporating the DRAHYNIVTF epitope were gener-ally weak and inconsistent in E7 peptide-immunized FVB

and K14E7.FVB mice, suggesting that this epitope is themajor dominant source of T help for E7 in this, as inother, genetic backgrounds of mice. As in previousexperiments with C57BL/6J and other mouse strains, asingle immunization of FVB mice with E7 protein or an E7polynucleotide vaccine failed to elicit a T cell proliferativeresponse that could be directly recalled with the pep-tides incorporating the promiscuous epitope. Unexpec-tedly, however, a consistent response to GF106 wasobserved in K14E7.FVB but not FVB mice immunizedwith E7 protein or polynucleotide vaccine (Table 3). Asimilar discrepancy was observed between the T pro-liferative response to E7 in F1(K14E7×C57) mice immu-


Table 3. T cell proliferation to HPV16E7 and constituent peptides in K14E7.FVB and FVB mice immunized with various prepara-tions of HPV16E7

Recallantigen

E7 101–103a) 104–106 107–109 Medium (cpm)

Mousetype

FVB K14 FVB K14 FVB K14 FVB K14 FVB K14

Immunogen

GSTE7 126b) 57.6 0.7 0.7 1.0 17.7 0.6 0.9 72 60

E7 DNA 52 14.5 0.8 1.0 0.8 5.5 0.9 2.0 43 67

GF101–103 5.5 22.7 1.3 0.8 1.3 1.1 3.7 0.8 43 56

GF104–106 69.3 46.0 3.6 10.6 21.7 28.7 5.6 12.1 55 49

GF107–109 15 37.1 1.0 0.6 1.0 0.6 3.9 0.6 64 57

QA 0.9 0.9 0.9 1.1 0.8 0.9 0.9 1.1 54 66

a) Pools of three linear peptides from HPV16 were used for immunization or for recall.b) Results, except for the medium control, are expressed as stimulation indices calculated from quadruplicate replicates as cpm

(antigen)/cpm (medium). Pools of LN cells from groups of five mice were assayed. One of two experiments with similar resultsis shown.

nized with E7GST (stimulation index 6.0±1.2), whereasF1(FVB×C57) mice failed to respond (stimulation index0.8±0.2). These data suggested that the transgenic ani-mals, in contrast to their littermate controls, had agreater precursor frequency of Th cells specific for thisepitope of E7. Immunization with peptide generallyselects for both higher and lower affinity Th cells,whereas immunization with protein selects for the Thcells with higher affinity for the peptide epitope. The pro-liferative response to GF106 induced by GF106 in boththe transgenic and non-transgenic mice, and by E7GSTin transgenic but not in non-transgenic mice, suggeststhat the expansion in the GF106-specific precursor poolin the transgenic animals may be limited to the Th cellswith higher affinity for this epitope.

The E7 protein of HPV16 lacks an MHC class I-restrictedepitope on the FVB (H-2q) background [18]. To establishcytotoxic T effector function in mice carrying the K14E7transgene, F1 (C57BL/6J×K14E7.FVB) mice were there-fore used, as there is a well-characterized cytotoxic Tcell (Tc) epitope (RAHYNIVTF) restricted by H-2 Db [21].Mice were immunized with E7/Quil A, previously shownto elicit strong CTL responses in a variety of mousestrains, or the minimal CTL epitope of E7 which also in-duces good CTL precursor responses, and cytotoxic Tcell precursors were assessed after in vitro restimulationof cells from the draining lymph node and spleen, from

5–35 days after immunization. F1(C57BL/6J×FVB) micedeveloped the expected strong RAHYNIVTF-restrictedCTL response (Fig. 1), whereas their transgenic littermatecontrols developed a much reduced but measurable E7-specific CTL activity, more obvious with RAHYNIVTF-pulsed EL4 targets than with EL4 targets expressing E7endogenously (C2 cells). Antibody and CTL responses toOVA were equally strong in both the E7-transgenic andnon-transgenic F1 hybrid mice (data not shown), exclud-ing a generalized immunosuppression or impairment ofCTL responses in mice expressing the K14E7 transgeneas an explanation for the lack of E7 CTL activity in thesemice.

To determine whether the relative lack of CTL precursorsin vitro in the F1 (C57BL/6J×K14E7.FVB) mice had afunctional correlate in vivo, these mice were immunizedwith HPV16E7 protein and challenged with one of twoE7+ H-2b tumors, i.e. C2, an EL4 derivative, and C3, afibroblast line, protection against both of which is CD8mediated and requires no E7-specific CD4 response[22]. Immunization of F1 (C57BL/6J×FVB) mice effect-ively protected these mice from challenge with an other-wise lethal inoculum of either tumor, whereas E7-immunized F1 (C57BL/6J×K14E7.FVB) mice wereunprotected and developed tumors as rapidly as unim-munized controls (Fig. 2).


Figure 1. E7-specific CTL responses in K14E7-transgenicand control mice. E7-specific CTL activity in LN lympho-cytes from F1 (K14E7.FVB×C57BL/6J) transgenic mice ( 1 )or F1 (FVB×C57BL/6J) mice ( Æ ) immunized with HPV16 E7with Quil A adjuvant, or F1 (FVB×C57BL/6J) mice immun-ized with OVA ( | ). CTL activity in draining lymph nodes wasassessed against targets as shown. C2 cells are EL4 cellsexpressing E7; EG7 are EL4 cells expressing OVA. PeptideGF001 is RAHYNIVTF, the minimal CTL epitope of E7. SIIN-FEKL is the minimal CTL epitope of OVA. Results are fromLN cells pooled from groups of five mice. GL001 ( ! ) is aRAHYNIVTF-specific CTL line. One of two experiments withsimilar results is shown.

3 Discussion

In this study we have demonstrated that expression ofthe E7 protein of HPV16 as a transgene driven from theK14 promoter results in loss of E7-specific cytotoxic Tcell function, together with enhancement of the E7-specific Th repertoire. Expression of a cytoplasmic pro-tein from the K14 promoter in thymic epithelial cellstherefore allows positive but not negative selection in the

MHC class II-restricted T cell compartment, in keepingwith recent similar observations for alloantigenexpressed from the K14 promoter [12].

Enhancement of the antigen-specific Th repertoire as aconsequence of expression of the antigen in thymic epi-thelium was suggested in our model by the T cell prolifer-ation studies. It was confirmed by studies in whichinduction of E7-specific IgG was enhanced followingimmunization with a polynucleotide vaccine, as in theseexperiments the contribution of an exogenous source ofT help to the E7-specific B cell response could be dis-counted. In previous studies, it has been demonstratedthat restricting the material presented by MHC class II inthe thymus to a single peptide results in a limited but stillremarkably diverse MHC class II-restricted repertoire[10]. The current study complements this finding bydemonstrating that a single protein expressed in thymiccortical epithelial cells in the context of MHC class II canhave a measurable effect on the MHC class II-restrictedrepertoire specific for that protein. E7 expression in skinis unlikely to contribute to the observed alteration of theTh repertoire in K14E7 mice, as resting skin keratino-cytes are MHC class II negative, and expression of anti-gen in the context of MHC class II on keratinocytes istolerogenic rather than immunogenic [23].

The normal B cell repertoire in the K14.E7-transgenicanimals when immunized with E7 would be predictedfrom studies with hen egg lysozyme (HEL)-transgenicmice, in which B cell tolerance has been associated onlywith high levels of secreted or membrane-associatedtransgenic protein [24]. The loss of CTL functionobserved in the K14E7-transgenic mice is in contrast tothe observation that Kb expressed in the skin from thekeratin 4 (K4) promoter does not alter the survival orfunction in vitro of Kb-specific CTL [7]. Our resultsresemble the deletional tolerance associated withexpression of a transgene peripherally at sites other thanskin [6, 7, 25]. The residual CTL activity observed inF1(K14E7×C57BL/6) mice immunized with E7, which isapparent particularly against targets pulsed with highconcentrations of the minimal CTL epitope RAHYNIVTF,may reflect induction by immunization of E7-specificCTL of lower avidity, not deleted by the naturally pre-sented E7 peptide in E7-transgenic animals. Deletionaltolerance has been attributed to T cell cross-priming byprofessional APC in the absence of CD4 priming and co-stimulatory activity [25]. The contrast between the cur-rent study and the study employing the K4 promoter toproduce alloantigen expression on keratinocytes mightrelate to the concurrent thymic expression of the trans-genic antigen in our model. Such a conclusion wouldinfer that presentation of antigen by thymic epithelialcells enhanced tolerance induction. However, in a previ-


Figure 2. Tumor protection induced by E7 immunization in K14E7-transgenic and control mice. Protection ofK14E7.FVB×C57BL/6J and FVB×C57BL/6J mice immunized with various preparations of HPV16 E7 against H-2b tumors expre-ssing HPV16E7 (C2 and C3) is shown. Mice were immunized twice with E7 or OVA and Quil A as indicated and challenged withtumor cells 14 days after the second immunization. Tumor mass was assessed at 12 days after grafting. One of two experimentswith similar results is shown.

ous study [12] expression of class II alloantigen from theK14 promoter in the thymus did not result in central dele-tional tolerance, but rather enhanced the generation oftransgene-specific CTL activity. Further, studies onthymic grafting in an F1 (K14E7×HLA A2.1) model con-firm that E7-specific CTL responses are lacking in theK14E7-transgenic animals, whereas other CTL respons-es are preserved, and suggest that full reconstitution ofthe E7-specific CTL response is not achieved by thymicgrafts [26], arguing for at least some peripheral contribu-tion to the development of CTL tolerance following anti-gen expression in the skin and thymus from the K14 pro-moter. The limited sites of expression of E7 in the K4.Kb-transgenic mouse must therefore be insufficient toinvoke peripheral deletional tolerance. These findings areparalleled by the lack of deletional tolerance observed inanimals expressing low levels of transgene in the liver [5],and also by the lack of CTL tolerance in F1(FVB×C57)recipients of skin grafts from F1( § A-crystallinE7×C57BL/6) mice [27] and F1(K14E7×C57BL/6) mice (unpublishedobservation). The total level of expression of E7 in theskin may therefore be a major determinant of whetherthe immune system responds to or ignores the autoanti-gen. This has been observed as a dose-dependentinduction of E7-specific DTH following grafting of E7-transfected keratinocytes [28]. Despite apparently nor-mal Kb-specific CTL responses in vitro in the K14.Kb

model referred to above, Kb-positive skin allografts werenot rejected from the K4.Kb-transgenic animals [7],although they were rejected from non-transgenic litter-mate controls. These data suggest that there was somefunctional impairment of the Kb-specific response in theK4.Kb-transgenic animals. The discrepancy betweeninduction of CTL activity in vitro and CTL function in this

model in vivo has a parallel in the lack of rejection of E7-expressing grafts by mice with demonstrableimmunization-induced E7-specific CTL activity [27].Taken together, these data suggest that under some cir-cumstances peripheral CTL anergy can be reversedwhen cells are removed from the tolerogenic environ-ment. In a further study in which HPV16E7 wasexpressed from the K14 promoter in the thymus and inthe skin, no difference was seen between transgenicmice and littermate controls in the cytotoxic immuneresponses to E7, and E7-specific Tc precursor fre-quency, avidity, and tumor protection was maintained inE7-immunized animals [29]. The observed differencebetween this model and our apparently similar modelcould reflect differing levels or timing of expression of E7antigen in the two transgenic lines. The E7-transgenicline used in our work was one of several available lines[16], and was selected for high-level expression. Alterna-tively, although the transgene was presented by H-2b inboth models, the differing genetic backgrounds of thetwo animals might allow for a non-MHC-linked geneticdifference in immune responsiveness to the K14E7transgene.

The K14E7-transgenic mice were evaluated immunolog-ically as potential models in which to test immunother-apy for papillomavirus-associated epithelial cancer. Thecurrent results in a mouse with split tolerance to E7 con-firm findings from depletion and peptide immunizationstudies in conventional mice that protection against thetransplantable tumors C2 and C3 is entirely CD8 medi-ated [22, 30]. Induction of E7-specific tumor tolerance asa consequence of expression of E7 in the skin and thy-mus suggests that high-level expression of a tumor anti-


gen in a restricted range of sites can result in cross-priming and tolerance induction as has been describedfor OVA expressed as a soluble or membrane-associated protein in the kidney and pancreas [25]. Thelack of E7-specific CTL observed in the K14E7-transgenic mice with a normal Th repertoire implies thatE7 antigen is not entering the professional APC pathwayfrom skin to prime for T help in these mice, as this shouldprevent induction of peripheral CTL tolerance [31]. Alter-natively, there may be a window in the ontogeny of theimmune system in which presentation of antigen to Thprecursors is not able to rescue CTL from cross-priming-induced deletional tolerance.

The term “split” tolerance has been used to refer to dele-tional Tc tolerance to a self antigen, together with partialTh and no B cell tolerance. It has been observed previ-ously as a lack of B cell response to autoantigen whichcan be overcome by appropriate immunization [32]. Thispresent study suggests another type of split tolerancewhereby an autoantigen with limited tissue distribution inepithelial cells might positively select for autoreactive Thcells in the thymus, whereas Tc cells specific for theautoantigen are deleted or otherwise tolerized in theperiphery. A subsequent immunogenic exposure to theautoantigen could then give rise to a strong Th and anti-body response to the autoantigen. One consequence ofbreaking self tolerance through immunization that weobserved in this study (Table 3) was an apparent spread-ing of the immune response to the autoantigen from theimmunodominant autoepitope to other epitopes withinthe same autoantigen [33, 34]. Given split tolerance to anautoantigen, a lack of regulatory CD8+ antigen-specificCTL [35] might allow persistence of the autoreactive Band Th cells. On an appropriate “autoimmune” geneticbackground, continued presentation of autoantigen bythe B cells to the enhanced Th cell repertoire in the pres-ence of split tolerance might allow epitope spreading inthe autoimmune response, and therefore enhance therisk of development of autoimmune disease to an auto-antigen whose tissue distribution favored developmentof split tolerance.

4 Materials and methods

4.1 Source of mice

C57BL/6J (H-2b), FVB (H-2q) and K14E7-transgenic FVBmice [16] were from the Animal Resources Center, Perth,WA, and were 6–12 weeks old when used. Mice were heldunder specific pathogen-free conditions for the duration ofthe experiments, which were approved by the institutionalAnimal Ethics Experimentation Committee.

4.2 Immunogens

HPV16 GSTE7 recombinant fusion protein was purified bypreparative PAGE of bacterial inclusion bodies solubilizedusing SDS (Fernando et al., manuscript in preparation), andendotoxin was removed by extraction with Triton X-114 [36].Purified GSTE7 was administered as a sterile non-pyrogenicpreparation substantially free of endotoxin as shown byLimulus assay. A set of 8–23mer peptides comprising anoverlapping set spanning the E7 protein of HPV16 and asynthetic peptide BT12D comprising the three linear B epi-topes of E7 together with the promiscuous Th epitope [19]were synthesized as previously described [37] or purchasedas HPLC-purified synthetic peptides from Peptide Express(Fort Collins, CO). Quil-A (Spikoside®) was obtained fromISCOTEC AB, Sweden. MS2E7 fusion protein was preparedas previously described [38]. OVA was purchased fromSigma (Raritan, NJ). A polynucleotide vaccine comprisingthe full-length sequence of the HPV16E7 open readingframe (ORF) expressed from the CMV promoter was con-structed by cloning the E7ORF, excised from pJ4 Y E7 [39]using a HindIII/BglIII digestion and Klenow blunt ending, intothe pCMV g (Clontech, Palo Alto, CA) vector followingremoval of the g -galactosidase gene using a NotI/MluIdigestion. These excisions were designed to position theE7ORF between the CMV immediate early gene promoter/enhancer and SV40 polyadenylation site and the splicedonor/splice acceptor sites.

4.3 Tumor protection assays

The EL4.E7(C2) and C3 tumor cell lines and their growthconditions have been described [21, 40]. Mice (5–8 pergroup) were immunized s.c. at the base of the tail with 50 ? gE7GST protein, or 50 ? g OVA as control, and 10 ? g Quil-A asadjuvant. Mice were challenged with 3×106 C2 or 2×106 C3tumor cells 14 days after the last immunization. Fourteendays after tumor challenge, the mice were killed and tumortissue weighed. In some experiments mice were checked forpalpable tumor every other day and the data recorded. Theexperiments were repeated at least twice.

4.4 In vitro assays of E7-specific immunity

E7-specific antibody and IgG subclass responses weremeasured using ELISA as previously described [41]. CTLactivity was measured using a standard 4-h 51Cr-releaseassay, with EL4, 10–6M peptide-pulsed EL4, EL4E7 (C2) andEL4.OVA(EG7) cells as targets, and results expressed aspercentage specific lysis from the formula 100×(specificrelease – background release)/(total release – backgroundrelease). Cytotoxic effector cells were obtained by culturingdraining lymph node cells with 20 U/ml rIL-2 (Gibco-BRL,Gaithersburg, MD) for 4 days in complete RPMI medium, orby culturing splenocytes with irradiated E7/EL4 cells for 4days in complete RPMI medium, and then another 4 days


with complete medium containing 20 U/ml rIL-2. Controleffector cells were prepared from mice immunized with OVAand the appropriate adjuvant, and were similarly restimu-lated in vitro. An E7.H-2 Db-specific CTL line (GL001,manuscript in preparation) was also used as a positive con-trol. T cell proliferation assays were performed using drain-ing lymph nodes 8–10 days after immunization, or usingsplenocytes 35 days after immunization, essentially as pre-viously described [19].

Acknowledgements: The authors thank Barbara Murray,Margaret Passmore, Donna West, and Trina Stewart fortechnical assistance. This work was funded in part by NIHgrant ROI-CA-67366, and by grants from the QueenslandCancer Fund, the National Health and Medical ResearchCouncil of Australia, the Mayne Bequest, and the PA Hospi-tal Foundation.

5 References

1 Kappler, J. W., Boehm, N. and Marrack, P., T cell toler-ance by clonal elimination in the thymus. Cell 1997. 49:273–280.

2 Brocker, T., Riedinger, M. and Karjalainen, K., Tar-geted expression of major histocompatibility complex(MHC) class II molecules demonstrates that dendriticcells can induce negative but not positive selection ofthymocytes in vivo. J. Exp. Med. 1997. 185: 541–550.

3 Heath, W. R. and Miller, J. F. A. P., Tolerance to H-2Kb

expressed under the control of the rat insulin promoter.Transgenics 1994. 1: 349–355.

4 Heath, W. R., Karamalis, F., Donoghue, J. and Miller,J. F. A. P., Autoimmunity caused by ignorant CD8+ Tcells is transient and depends on avidity. J. Immunol.1995. 155: 2339–2349.

5 Ferber, I., Schönrich, G., Schenkel, J., Mellor, A. L.,Hämmerling, G. J. and Arnold, B., Levels of peripheralT cell tolerance induced by different doses of tolerogen.Science 1994. 263: 674–676.

6 Bertolino, P., Heath, W. R., Hardy, C. L., Morahan, G.and Miller, J. F. A. P., Peripheral deletion of autoreactiveCD8+ T cells in transgenic mice expressing H-2Kb in theliver. Eur. J. Immunol. 1995. 25: 1932–1942.

7 Schönrich, G., Momburg, F., Malissen, M., Schmitt-Verhulst, A.-M., Malissen, B. and Hämmerling, G. J.,Distinct mechanisms of extrathymic T cell tolerance dueto differential expression of self antigen. Int. Immunol.1992. 4: 581–590.

8 Ohashi, P. S., Oehen, S., Buerki, K., Pircher, H., Oha-shi, C. T., Odermatt, B., Malissen, B., Zinkernagel, R.M. and Hengartner, H., Ablation of “tolerance” andinduction of diabetes by virus infection in viral antigentransgenic mice. Cell 1991. 65: 305–317.

9 Miller, J. F. A. P. and Basten, A., Mechanisms of toler-ance to self. Curr. Opin. Immunol. 1996. 8: 815–821.

10 Bevan, M. J., In thymic selection, peptide diversity givesand takes away. Immunity 1997. 7: 175–178.

11 Akkaraju, S., Ho, W. Y., Leong, D., Canaan, K., Davis,M. M. and Goodnow, C. C., A range of CD4 T cell toler-ance: Partial inactivation to organ-specific antigenallows nondestructive thyroiditis or insulitis. Immunity1997. 7: 255–271.

12 Laufer, T. M., DeKoning, J., Markowitz, J. S., Lo, D.and Glimcher, L. H., Unopposed positive selection andautoreactivity in mice expressing class II MHC only onthymic cortex. Nature 1996. 383: 81–85.

13 Sant’Angelo, D. B., Waterbury, P. G., Cohen, B. E.,Martin, W. D., Van Kaer, L., Hayday, A. C. and Jane-way, C. A. Jr., The imprint of intrathymic self-peptideson the mature T cell receptor repertoire. Immunity 1997.7: 517–524.

14 Griep, A. E., Herber, R., Jeon, S., Lohse, J. K., Dubiel-zig, R. R. and Lambert, P. F., Tumorigenicity by humanpapillomavirus type 16 E6 and E7 in transgenic mice cor-relates with alterations in epithelial cell growth and differ-entiation. J. Virol. 1993. 67: 1373–1384.

15 Frazer, I. H., Leippe, D. M., Dunn, L. A., Liem, A.,Tindle, R. W., Fernando, G. J., Phelps, W. C. and Lam-bert, P. F., Immunological responses in human papillo-mavirus 16 E6/E7-transgenic mice to E7 protein corre-late with the presence of skin disease. Cancer Res.1995. 55: 2635–2639.

16 Herber, R., Liem, A., Pitot, H. and Lambert, P. F.,Squamous epithelial hyperplasia and carcinoma in micetransgenic for the human papillomavirus type 16 E7oncogene. J. Virol. 1996. 70: 1873–1881.

17 Lambert, P. F., Pan, H., Pitot, H., Liem, A., Jackson, M.and Griep, A. E., Epidermal cancer associated withexpression of human papillomavirus type 16 E6 and E7oncogenes in the skin of transgenic mice. Proc. Natl.Acad. Sci. USA 1993. 90: 5583–5587.

18 Herd, K., Fernando, G. J. P., Dunn, L. A., Frazer, I. H.,Lambert, P. and Tindle, R. W., E7 oncoprotein of humanpapillomavirus type 16 expressed constitutively in theepidermis has no effect on E7-specific B- or Th-repertoires or on the immune response induced or sus-tained after immunization with E7 protein. Virology 1997.231: 155–165.

19 Tindle, R. W., Fernando, G. J., Sterling, J. C. and Fra-zer, I. H., A “public” T-helper epitope of the E7 trans-forming protein of human papillomavirus 16 providescognate help for several E7 B-cell epitopes from cervicalcancer-associated human papillomavirus genotypes.Proc. Natl. Acad. Sci. USA 1991. 88: 5887–5891.


20 Tindle, R. W., Smith, J. A., Geysen, H. M., Selvey, L. A.and Frazer, I. H., Identification of B epitopes in humanpapillomavirus type 16 E7 open reading frame protein. J.Gen. Virol. 1990. 71: 1347–1354.

21 Feltkamp, M. C. W., Smits, H. L., Vierboom, M. P. M.,Minnaar, R. P., De Jongh, B. M., Wouter Drijfhout, J.,Ter Schegget, J., Melief, C. J. M. and Kast, W. M., Vac-cination with cytotoxic T lymphocyte epitope-containingpeptide protects against a tumor induced by humanpapillomavirus type 16-transformed cells. Eur. J. Immu-nol. 1993. 23: 2242–2249.

22 Fernando, G. J., Stewart, T., Tindle, R. W. and Frazer,I. H., Induction of tumor antigen specific CTL mediatedtumour protection, is not altered by prior induction of aTH2 type response specific for the same tumour antigen.J. Immunol. 1998, in press.

23 Bal, V., McIndoe, A., Denton, G., Hudson, D., Lom-bardi, G., Lamb, J. and Lechler, R., Antigen presenta-tion by keratinocytes induces tolerance in human T cells.Eur. J. Immunol. 1990. 20: 1893–1897.

24 Goodnow, C. C., Crosbie, J., Adelstein, S., Lavoie, T.B., Smith-Gill, S. J., Brink, R. A., Pritchard-Briscoe,H., Wotherspoon, J. S., Loblay, R. H., Raphael, K.,Trent, R. J. and Basten, A., Altered immunoglobulinexpression and functional silencing of self-reactive Blymphocytes in transgenic mice. Nature 1988. 334:676–682.

25 Kurts, C., Kosaka, H., Carbone, F. R., Miller, J. F. andHeath, W. R., Class I-restricted cross-presentation ofexogenous self-antigens leads to deletion of autoreac-tive CD8+ T cells. J. Exp. Med. 1997. 186: 239–245.

26 Doan, T., Chambers, M., Street, M., Fernando, G. J. P.,Herd, K., Lambert, P. and Tindle, R. W., Mice express-ing the E7 oncogene of HPV16 in epithelium show cent-ral tolerance, and evidence of peripheral anergisingtolerance to E7-encoded cytotoxic T-lymphocytes (CTL)epitopes. Virology 1998. 244: 352–364.

27 Dunn, L. A., Evander, M., Tindle, R. W., Bulloch, A. L.,De Kluyver, R. L., Fernando, G. J., Lambert, P. F. andFrazer, I. H., Presentation of the HPV16E7 protein byskin grafts is insufficient to allow graft rejection in an E7-primed animal. Virology 1997. 235: 94–103.

28 Chambers, M. A., Wei, Z., Coleman, N., Nash, A. A.and Stanley, M. A., “Natural” presentation of humanpapillomavirus type-16 E7 protein to immunocompetentmice results in antigen-specific sensitization or sustainedunresponsiveness. Eur. J. Immunol. 1994. 24: 738–745.

29 Melero, I., Singhal, M. C., McGowan, P., Haugen, H.S., Blake, J., Hellstrom, K. E., Yang, G., Clegg, C. H.and Chen, L., Immunological ignorance of an E7encoded cytotoxic T lymphocyte in transgenic miceexpressing the E7 and E6 oncogenes of human papillo-mavirus type. 16. J. Virol. 1997. 71: 3998–4004.

30 Fernando, G. J. P., Stewart, T., Tindle, R. W. and Fra-zer, I. H., Vaccine induced Th1 type responses are domi-nant over Th2 type responses in the short term whereaspre-existing Th2 responses are dominant in the longerterm. Scand. J. Immunol. 1998. 47: 459–465.

31 Bennett, S. R., Carbone, F. R., Karamalis, F., Miller, J.F. and Heath, W. R., Induction of a CD8+ cytotoxic Tlymphocyte response by cross-priming requires cognateCD4+ T cell help. J. Exp. Med. 1997. 186: 65–70.

32 Zinkernagel, R. M., Pircher, H. P., Ohashi, P., Oehen,S., Odermatt, B., Mak, T., Arnheiter, H., Burki, K. andHengartner, H., T and B cell tolerance and responses toviral antigens in transgenic mice: implications for thepathogenesis of autoimmune versus immunopatholog-ical disease. Immunol. Rev. 1991. 122: 133–171.

33 Reynolds, P., Gordon, T. P., Purcell, A. W., Jackson, D.C. and McCluskey, J., Hierarchical self-tolerance to Tcell determinants within the ubiquitous nuclear self-antigen La (SS-B) permits induction of systemic autoim-munity in normal mice. J. Exp. Med. 1996. 184:1857–1870.

34 Topfer, F., Gordon, T. and McCluskey, J., Intra- andintermolecular spreading of autoimmunity involving thenuclear self-antigens La (SS-B) and Ro (SS-A). Proc.Natl. Acad. Sci. USA 1995. 92: 875–879.

35 Zhang, L., Martin, D. R., Fung-Leung, W.-P., Teh, H.-S.and Miller, R. G., Peripheral deletion of mature CD8+

antigen-specific T cells after in vivo exposure to maleantigen. J. Immunol. 1992. 148: 3740–3745.

36 Aida, Y. and Pabst, M. J., Removal of endotoxin fromprotein solutions by phase separation using Triton X-114. J. Immunol. Methods 1990. 132: 191–195.

37 Fernando, G. J., Tindle, R. W. and Frazer, I. H., T-helper epitopes of the E7 transforming protein of cervicalcancer associated human papillomavirus type 18(HPV18). Virus Res. 1995. 36: 1–13.

38 Seedorf, K., Oltersdorf, T., Krammer, G. and Rowen-kamp, W., Identification of early proteins of the humanpapillomaviruses type 16 (HPV16) and type 18 (HPV18)in cervical carcinoma cells. EMBO J. 1987. 6: 139–144.

39 Crook, T., Storey, A., Almond, N., Osborn, K. andCrawford, L., Human papillomavirus type 16 cooper-ates with activated ras and fos oncogenes in thehormone-dependent transformation of primary mousecells. Proc. Natl. Acad. Sci. USA 1988. 85: 8820–8824.

40 Tindle, R. W., Croft, S., Herd, K., Malcolm, K., Geczy,A. F., Stewart, T. and Fernando, G. J. P., A vaccine con-jugate of ‘ISCAR’ immunocarrier and peptide epitopes ofthe E7 cervical cancer-associated protein of humanpapillomarvirus type 16 elicits specific Th1- and Th2-typeresponses in immunized mice in the absence of oil-basedadjuvants. Clin. Exp. Immunol. 1995. 101: 265–271.


41 Fernando, G. J., Stenzel, D. J., Tindle, R. W., Merza,M. S., Morein, B. and Frazer, I. H., Peptide polymerisa-tion facilitates incorporation into ISCOMs and increasesantigen-specific IgG2a production. Vaccine 1995. 13:1460–1467.

Correspondence: Ian H. Frazer, Center for Immunology andCancer Research, University of Queensland, Department ofMedicine, Princess Alexandra Hospital, Woolloongabba,Queensland, AustraliaFax: +61-7 32 40 20 48e-mail: ifrazer — medicine.pa.uq.edu.au


Documents

Split tolerance to a viral antigen expressed in thymic epithelium and keratinocytes