Transgenic mice expressing membrane-bound OVA under the rat insulin promoter, RIP-mOVA, has previously been suggested to display deletional tolerance toward the dominant CTL epitope, SIINFEKL, and provide an elegant model system to test the hypothesis that the lack of T cell help contributes to the tolerance. To understand how the CD8 tolerance is maintained in these mice, a set of neo-self-Ags, OVA, modified to contain a foreign Th peptide, were constructed and tested for their ability to induce CTL responses in RIP-mOVA mice. Immunization with these Th peptide-modified OVA molecules and not with the wild-type OVA induced self-reactive CTLs recognizing dominant CTL peptide, SIINFEKL. Importantly, immunization with the modified OVA constructs also prevented the growth of OVA-expressing tumors in transgenic mice. Since endogenous OVA Th peptides did not contribute toward breaking self CTL tolerance, these results also highlighted a very robust CD4 T cell tolerance toward OVA in RIP-mOVA mice that has not been previously described. These results therefore provide direct evidence that it is the tolerance in the CD4 Th cell compartment that helps maintain the CTL tolerance against self-Ag in these mice. Since the CTL tolerance can be broken or bypassed by foreign Th peptides inserted into a self Ag, potential of using this approach in generating effective therapeutic cancer vaccines is discussed.
The importance of T cell help with regard to CTL induction via activation of APCs, has gained a substantial interest in the last few years. It is generally believed that in most cases, the first event in the activation of CTL precursors is the interaction of an APC and a Th cell. This leads to activation of the APC, and in turn to expression of multiple costimulatory molecules and cytokines, which is critical for the subsequent activation of CTL precursors (1, 2, 3). This three-cell model relies on cognate help, and therefore the Ag has to be processed and enter both the MHC class I and the MHC class II pathway in the same APC.
Studies with both non-transgenic and transgenic mice expressing model Ags have shown that T cells specific for self-Ags exists in the peripheral pool of T cells, even in the case of thymic expression of the Ag (4, 5, 6). Thus, thymic expression of Ags does not result in complete deletion of self-specific T cells and, therefore, recruitment of these cells by immunization could potentially generate CTLs with cytolytic functions that are, for example, capable of destroying tumor cells expressing the particular self-Ag. Activation of CD8 cells is central for the development of effective tumor vaccines. It is known that CD4+ Th cells play crucial roles in priming, expansion, and memory and in survival of CD8+ CTLs (1, 2, 3, 7, 8, 9, 10, 11, 12). Studies in which reversal of a non-responsive state in CD8 cells has been achieved by providing cytokines and co-stimulatory signals suggest indirectly that T cell help is crucial in maintaining and breaking CD8 CTL tolerance (13, 14, 15).
A large number of tumors express certain self-Ags (HER-2, epidermal growth factor receptor, prostate-specific membrane Ag, etc.) that serve as targets for prostate-specific membrane Ag immunotherapy (16, 17, 18). Both Abs as well as CTLs recognizing these self-Ags could be effective tools in the treatment of tumors. A number of studies within the area of cancer immunotherapy have relied on the approach that intrinsic MHC class II peptides from the wild-type (wt)3 tumor Ag would provide the help needed for CTL induction. However, help provided by peptides from the tumor Ags will often be weak or absent due to the tolerance toward these self-proteins, which could hamper the therapeutic effect of a vaccine.
Our aim in this study was to determine whether cognate T cell help provided by an inserted foreign Th epitope inserted into a self-Ag, can lead to activation of self-reactive CTLs and whether these CTLs can protect mice from tumor growth. We chose a well studied transgenic mouse model, rat insulin promoter (RIP)-mOVA, to address this issue. RIP-mOVA expresses a membrane bound truncated OVA sequence under the control of RIP in pancreatic islets as well as in the kidney proximal tubules, in thymus and in testis of male mice (19). This mouse strain has previously been suggested to exhibit deletional tolerance toward the SIINFEKL epitope (20). However, the natural state of tolerance in these mice is unclear, since most experiments were based on studies using a high affinity SIINFEKL-specific clone (OT-I) (20, 21). A set of model vaccine molecules were constructed consisting of the OVA gene with an inserted sequence comprising a promiscuous immunodominant Th peptide, P30, derived from tetanus toxoid. P30 modified OVA cDNA molecules were then used for immunization and CTL responses were measured against the dominant OVA epitope, SIINFEKL, in RIP-mOVA transgenic mice. Our findings demonstrate that insertion of a foreign Th epitope in the neo-self-Ag (OVA) is sufficient to break the CD8 tolerance in RIP-mOVA mice. This induction of self-reactive CTLs is dependent on cognate T cell help provided by the inserted foreign Th epitope in OVA. The fact that insertion of a single foreign Th epitope in the self-Ag could break the CD8+ tolerance, argues that the tolerance in the CD4 compartment is a major regulator of the CD8+ T cells tolerance against self-Ags. Furthermore, insertion of a foreign Th peptide in OVA was able to change the vaccine molecule from being an inefficient molecule to a potent vaccine that also protected mice from tumor growth.
Materials and Methods
Synthetic peptides representing the dominant Kb binding motif from OVA, OVA257–264 (SIINFEKL) (22), the promiscuous Th epitope P30 from tetanus toxoid, TT947–967 (FNNFTVSFWLRVPKVSASHLE) (23), and the Ab epitope from OVA, OVA265–280 (TEWTSSNVMEERKIKV) (24) were synthesized using conventional F-moc chemistry on a NovaSyn peptide synthesizer (Novabiochem). Peptides were purified by reverse phase HPLC on a C-18 column (Phenomenex) and were subsequently sequenced on a 476A peptide sequencer (Applied Biosystems) to confirm the correct sequences. Peptides were >95% pure.
Chicken OVA cDNA was kindly provided by Dr. Francis Carbone (University of Melbourne, Australia). OVA-P30 constructs were prepared by introducing the oligonucleotide encoding the promiscuous Th epitope P30 from tetanus toxoid into the DNA sequences using standard PCR mutagenesis procedures. In OVA-P30.1, amino acid residues 271–291 from OVA were substituted with the P30 sequence (FNNFTVSFWLRVPKVSASHLE). In OVA-P30.2 amino acid residues 320–340 were replaced with P30. These constructs were truncated at the amino terminus corresponding to amino acid residue 139 from OVA to give the same OVA amino acid sequence that was used to generate the corresponding OVA transgenic mice (RIP-mOVA). Non-modified and modified OVA coding sequences were inserted into the eukaryotic expression vector pVAX1 (Invitrogen Life Technologies) under the control of the CMV promoter. All DNA batches were purified, and endotoxin was removed using EndoFree plasmid kit (Qiagen). DNA batches were monitored for contaminating protein by measuring (OD 260 nm/OD 280 nm) and were only used if the ratio was within the interval of 1.8–1.9.
The Kb-expressing LB27.4 cell line (American Type Culture Collection) was retrovirally transduced to express either the dominant Kb binding motif from OVA (SIINFEKL) or the scrambled sequence (FILKSINE) (25). These two transduced cell lines, LB27.4-SIINFEKL and LB27.4-FILKSINE were used in CTL assays as target cells and control cells, respectively. EL-4 and E.g7-OVA were obtained from American Type Culture Collection. E.g7-OVA originates from EL-4 and is transfected with full-length OVA. All cells were maintained in complete RPMI 1640 containing 400 μg/ml geneticin (Invitrogen Life Technologies), except for EL4 that was maintained in complete RPMI 1640 only.
C57BL/6 mice (6–8 wk old) were purchased from Mollegaard and Bomholdtgaard (M&B). RIP-mOVA mice that are on C57BL/6 background were obtained from The Walter and Eliza Hall Institute of Medical Research (Melbourne, Australia). They were bred in the animal facilities of Pharmexa and M&B. RIP-mOVA mice express OVA under the RIP and have, as such, OVA as a neo-self-Ag. They are transgenic for truncated OVA (OVA139–386) that is expressed as membrane bound molecule in pancreatic islets, kidney proximal tubules, and testis of male mice. The OVA gene expression is heterozygous and transgenic and non-transgenic littermates were identified by PCR on isolated DNA from blood samples as previously described (19).
Synthetic peptides were dissolved in MilliQ water, adjusted to 1 mg/ml, and emulsified 1:1 (v/v) in CFA. Then 200 μl of each emulsion were injected s.c. at the lower back for CTL assays. For proliferation assays, 200 μl were injected s.c. at the lower back and in the hind footpads. Immunizations were performed once.
Each mouse was injected once, unless otherwise stated, with 100 μg of endotoxin-free DNA in 100 μl of 0.9% saline. All injections were performed intradermally.
Lymph nodes were taken 14 days after the boost in mice injected with DNA and after 10 days in mice immunized with peptide in CFA. Draining lymph nodes were removed, and single cell suspensions were prepared. Serial dilutions of Ag were mixed with 4 × 105 cells per well in microtiter plates in total volumes of 200 μl/well of RPMI 1640 containing 1% syngeneic mouse serum. Four days later, cells were pulsed with 0.5 μCi of [3H]thymidine (Amersham Biosciences) and harvested 8 h later.
Spleens were removed and pooled within the groups, and single cell suspensions were prepared. For each group of mice, 9 × 107 cells were cultured in 30 ml of complete RPMI 1640 supplemented with 1 mM pyruvate and 1× MEM nonessential amino acids in a 200-ml culture flask (standing) and restimulated with 1 μM SIINFEKL peptide. Cultures were incubated for 5 days followed by measurement of lytic ability using standard 51Cr-release assay. LB27.4-SIINFEKL and LB27.4-FILKSINE were used as target cells. Cells were incubated for 4 h before harvest. Maximum release was determined by the addition of 1% Nonidet P-40. Percentage of specific lysis in 51Cr-release assays was calculated as: [(experimental release − spontaneous release)/(maximum release − spontaneous release) × 100].
For comparison of the CTL avidity obtained from RIP-mOVA and C57BL/6 mice, the procedure was essentially the same as described above, except that target cells were EL-4 cells pulsed with SIINFEKL peptide for 45 min in the presence of 51Cr and SIINFEKL peptide in serial dilutions of 10−5, 10−6, 10−7, 10−8, and 10−9 M. Afterwards, target cells were washed and used in a standard 51Cr-release assay using non-pulsed EL-4 cells as control cells.
Groups of five C57BL/6 or RIP-mOVA mice were immunized with pVAX or OVA-P30.2. Each mouse was injected with 100 μg of DNA intradermally. Mononuclear cells from the spleens were purified on Nycoprep density media 14 days later (Axis-shield, Oslo, Norway). Cells (106) were resuspended in 20 μl of rabbit serum followed by the addition of 2 μl of anti-mCD8α-FITC (BD Biosciences) and 25 μl of PE-coupled SIINFEKL-MHC(Kb)-tetramer solution for a final dilution of 1/200 (Proimmune). They were labeled at 18°C for 1.5 h, washed twice, and analyzed on a FACSCalibur (BD Biosciences).
Tumor protection experiments
Groups of either five C57BL/6 or RIP-mOVA mice were immunized with pVAX, OVA-wt(trunc), OVA-P30.1, or OVA-P30.2. Each mouse was injected three times with 100 μg of DNA (intradermally) in 0.9% saline at 14-day intervals. Mice were injected s.c. with 7 × 104 E.g7-OVA cells (C57BL/6 derived) 13 days after the last boost. A third boost with DNA was performed 5 days later. Tumor sizes were measured as the product of two perpendicular tumor diameters. Mice were sacrificed when tumors reached a size of approximately 100 mm2 corresponding to the maximal allowed size as specified by Danish laws.
Construction of modified OVA-DNA molecules
To investigate whether insertion of a foreign Th epitope in the chicken OVA gene (OVA) was able to induce self-reactive CTLs in OVA transgenic mice, a set of different OVA plasmids, where the sequence for the promiscuous Th epitope P30 was substituted into the OVA cDNA, were constructed (Fig. 1). Three of the OVA constructs, OVA-wt(trunc), OVA-P30.1, and OVA-P30.2 comprised OVA truncated at the amino terminus at the amino acid 139 in OVA for use in RIP-mOVA mice. It is important to note that RIP-mOVA mice are transgenic for OVA139–386. In OVA-P30.1, the P30 sequence was substituted for a known MHC class II (Ab) Th epitope from OVA, encompassing amino acid 265–280. In OVA-P30.2, the sequence OVA320–340 was substituted by the P30 Th epitope. The different OVA sequences were then cloned into the eucaryotic expression vector pVAX1, and used for DNA vaccinations.
CD4 tolerance in RIP-mOVA mice
Since CD4 Th cells play such a crucial role in immunity and tolerance, initial experiments were designed to determine the level of CD4 tolerance in RIP-mOVA mice. Data for CD4 tolerance in RIP-mOVA has not previously been shown, although the strain has been suggested to be tolerant (21). To determine whether RIP-mOVA mice make Th responses to P30 and a dominant I-Ab-restricted peptide, OVA265–280, these mice as well as normal C57BL/6 mice were immunized with these peptides. Both RIP-mOVA and C57BL/6 mice responded equally well to the P30 peptide from tetanus toxoid (Fig. 2,A). Lymph node cells from RIP-mOVA mice immunized with OVA265–280 did not proliferate. In contrast, lymph node cells from C57BL/6 immunized with this peptide responded normally to OVA265–280 peptide (Fig. 2 B). Another I-Ad/I-Ab-restricted OVA peptide, OVA323–339 (26), also failed to generate T cell proliferation in RIP-mOVA mice following immunization, whereas wt C57BL/6 mice responded to OVA323–339 as has been described previously (data not shown). These experiments show that RIP-mOVA transgenic mice display robust CD4 tolerance toward OVA. These results support previous experiments in a hemagglutinin (HA) system where induction of peripheral tolerance for CD4 T cells was demonstrated in a transfer system upon transferring of HA-specific transgenic CD4 T cells in HA transgenic mice (27). It is important to note that proliferation to a full-length OVA was not tested in RIP-mOVA mice because these mice are transgenic for only a truncated portion of OVA, encoding OVA139–389 protein.
Breaking CD8+ tolerance with modified OVA constructs
We next examined CTL reactivity in RIP-mOVA transgenic mice toward OVA CTL peptide SIINFEKL. Mice were immunized with SIINFEKL peptide (dominant OVA-derived Kb epitope) in CFA, which is known to give a Th-independent activation of CTLs (28). Although wt mice generated strong SIINFEKL-specific CTL responses, the CTL response in RIP-mOVA against SIINFEKL was not detectable (Fig. 3). These data suggest that RIP-mOVA mice are tolerant to the SIINFEKL epitope and are in accordance with previous evidence of deletional tolerance in these mice toward the SIINFEKL epitope (22).
If deletion of SIINFEKL-specific CD8+ T cells occurs in the thymus of RIP-mOVA mice, we reasoned that SIINFEKL-specific T cells of low avidity might be present in the periphery but their low avidity for SIINFEKL may require T cell help for priming of proliferation of cytolytic activity. Since OVA-specific T help was lacking in RIP-mOVA, we tested whether P30-containing OVA DNA constructs could generate self-reactive CTL’s in these mice. Four groups of RIP-mOVA mice and normal C57BL/6 mice were DNA immunized intradermally with either OVA-wt(trunc), OVA-P30.1, OVA-P30.2, or vector alone (pVAX1). CTL assays were performed using LB27.4 cells that were retrovirally transduced to express either SIINFEKL or the scrambled FILKSINE sequence as target cells as described in Materials and Methods. In the transgenic RIP-mOVA mice, only immunization with P30-modified OVA constructs but not wt OVA DNA construct induced SIINFEKL-specific CTLs (Fig. 4,A). All three OVA DNA vaccines elicited CTL responses in C57BL/6 mice (Fig. 4 B). This CTL response coincided with a P30-specific T cell proliferation in RIP-mOVA mice whereas no proliferation was seen to I-Ab-restricted OVA peptide (data not shown). The ability of two P30-modified OVA constructs to elicit slightly different CTL responses in RIP-mOVA is conceivably due to the location of P30 peptide in OVA and its subsequent Ag presentation. This could also be due to differential expression levels of these vaccine molecules following immunization. Overall, these results demonstrate that cognate T cell help can overcome the CD8 tolerance in this model.
Avidity of self-reactive CTLs
The avidity of SIINFEKL-specific CTLs from RIP-mOVA mice and C57BL/6 were investigated by incubating generated CTLs with EL-4 cells that were loaded with SIINFEKL peptide in varying concentrations (Fig. 5). As seen, the CTLs from RIP-mOVA were clearly of lower overall avidity than those obtained from C57BL/6. At a SIINFEKL concentration of 10−8 M a pronounced difference appeared between C57BL/6 and RIP-mOVA mice and at 10−9 M of SIINFEKL the response for RIP-mOVA cells was not detectable while C57BL/6 CTLs still had a clearly detectable response. To demonstrate the frequency of SIINFEKL-specific CTLs in C57BL mice and RIP-mOVA mice following immunization with P30-modified OVA constructs, we performed a SIINFEKL tetramer staining on freshly isolated spleen cells from both mouse strains (Fig. 6). Mononuclear cells from the spleens were isolated and co-stained with PE-labeled SIINFEKL tetramer and FITC-labeled anti-CD8. There are approximately nine times less SIINFEKL-specific CD8 cells in RIP-mOVA than in C57BL/6 mice and the mean fluorescence intensity using tetramers are also lower in RIP-mOVA mice. Together these results show that both the avidity (Fig. 5) and the numbers of SIINFEKL-specific precursors (Fig. 6) are lower in RIP-mOVA compared with C57BL/6 mice.
Modified OVA construct protects mice from tumor growth
To further test if the CTL responses generated in RIP-mOVA mice were functional in their ability to suppress growth of OVA-expressing tumors, a tumor model was established. OVA-expressing E.g7 (Eg7-OVA) tumor cell line of C57BL/6 origin grow progressively in both normal C57BL/6 mice and in OVA transgenic mice. Wild-type mice (C57BL/6) as well as OVA-transgenic RIP-mOVA mice were injected three times with OVA-wt(trunc), OVA-P30.1, OVA-P30.2, or vector alone (pVAX). E.g7-OVA cells were injected s.c. 10 days after the third DNA injection. A fourth boost was administered 5 days after the inoculation of OVA-expressing tumor cells. Not surprisingly, the tumors were rejected in wt control C57BL/6 mice when immunized with OVA-wt(trunc), OVA-P30.1, and OVA-P30.2 (Fig. 7,A) since OVA acts as a foreign Ag in these mice. In RIP-mOVA transgenic mice, only immunization with the two Th peptide-modified constructs (OVA-P30.1 and OVA-P30.2) but not the OVA-wt(trunc) construct reduced the tumor growth (Fig. 7 B).
Elucidation of the mechanisms behind tolerance and breaking of tolerance is of paramount interest, not only to understand the fundamental concepts in immunology but also for vaccine development. In pursuit of this, we have developed a system to investigate the role of cognate T cell help in CD8 self-tolerance in RIP-mOVA mice. We show cognate foreign T cell help breaks CD8 T cell tolerance in RIP-mOVA mice, previously suggested to be exhibiting central tolerance in the CD8 compartment toward OVA. Importantly, these CTLs were able to recognize the dominant CTL epitope, SIINFEKL from OVA and prevent growth of tumors expressing OVA.
It is generally believed that CD4+ and CD8+ cells recognizing self-peptides with high affinity in the thymus are deleted (29, 30). However, tolerance to peripheral Ags appears to be dependent on the dose of Ag. It has been reported that high dose of Ag cause tolerance, whereas progressively lower doses cause ignorance (21, 31). Our results demonstrate that some CTL precursors recognizing SIINFEKL must have escaped thymic deletion in RIP-mOVA mice. These cells are not activated by the OVA-wt(trunc) because the mice lack CD4 Th cells specific for OVA. Thus, the mice neither make OVA-specific CD4 T cell responses nor anti-OVA Ab following immunization with OVA-wt DNA. However, immunization of RIP-mOVA mice with foreign Th peptide-modified OVA, OVA-P30, does induce functional CTLs recognizing the dominant OVA peptide SIINFEKL. Therefore, unlike in wt mice, induction of SIINFEKL-specific CD8+ T cell responses requires cognate CD4+ T help in mice expressing thymic and pancreatic self protein at a level similar to what is present in RIP-mOVA mice. Moreover, SIINFEKL-specific CTLs in RIP-mOVA transgenic mice are fewer in numbers and are of lower avidity than in non-transgenic mice. Taken together, the data support the idea that low avidity CD8-positive T cells in RIP-mOVA transgenic mice can be activated but only if cognate T cell help is provided.
Heath and colleagues (20) had previously concluded that there were no CTL precursors toward OVA peptide (SIINFEKL) in RIP-mOVA mice. In agreement with these workers, we also did not observe SIINFEKL-specific reactivity when these mice were immunized with SIINFEKL-CFA. We believe that the cognate T cell help provided by the OVA-P30 constructs and not by SIINFEKL in CFA is absolutely necessary to break peripheral CD8 tolerance. Although a strong Th response is generated toward mycobacterial components in a mixture with SIINFEKL, this help is non-cognate. Exactly how non-cognate T cell help leads to CTL induction to SIINFEKL in normal mice is not explicitly clear but there are various possibilities of indirect activation of dendritic cells via cytokines or other inflammatory stimuli from adjuvants (32, 33). And since there are more Ag-specific CTL precursors in a normal mouse compared with the transgenic mouse, small triggers could generate effective CTL responses in a normal situation. In any event, to generate self-reactive CTL response or for breaking self CTL tolerance, where the frequency and the avidity of CTL precursor is believed to be considerably low, the same APC needs to present both the helper as well as the CTL peptide.
In our system, Th cells generated against the inserted foreign Th peptide stimulate/activate professional APCs via costimulation. Once activated, these APC are now fully equipped to effectively present MHC class I-restricted peptides to circulating CTL precursors leading to their activation. Therefore, RIP-mOVA mice reflect a tolerant state where CTL precursors remain ignorant/unresponsive to SIINFEKL due to their low avidity for the Ag and they only respond if cognate T cell help is provided. These results therefore show that 1) a proportion of CD8+ cells reactive to a dominant self peptide, SIINFEKL, do escape deletion even though the Ag is expressed in thymus; 2) the SIINFEKL-reactive CTL precursors are fewer in numbers and of lower avidity in RIP-mOVA mice than in wt mice; 3) Th cells regulate the activity of SIINFEKL-specific CD8+ T cells in RIP-mOVA mice; and 4) cognate T help is critical for activation of self-specific CTL precursors in RIP-mOVA mice.
Generating a CTL response against a tumor-associated Ag for destruction of tumor cells remains a crucial parameter of success in the field of Ag-specific cancer immunotherapy. The current studies model clinical scenarios in which tumor self-Ags are expressed in thymus and peripheral tissues such as in pancreas and kidney, although other distributions of self-Ags may lead to similar precursor avidity/numbers since other factors, e.g., the level of expression, is important, too. Therefore, due to deletion of self-specific high avidity CD4 and CD8+ T cells, remaining tumor-specific CTL precursors are ignorant/unresponsive of tumor Ags due to their lower avidity. We provide here a model that demonstrates that this unresponsiveness can be overcome by providing cognate T cells help. Although clear inhibition of tumor growth was observed when mice received Th peptide modified OVA DNA constructs, the suppression of tumor growth was not complete. This is likely to be due to the fact that the activated CTLs in RIP-mOVA mice are fewer in numbers and are of lower avidity as seen from the avidity/tetramer experiments. In this context, Sherman and colleagues (34) have shown that provided low avidity CTLs could be activated in their HA transgenic system, these cells were capable of rejecting tumors.
Although CD8 T cells are known to be important in combat of cancer cells, other immunological mechanisms such as tumor inhibitory Abs are also of major importance if surface Ags are targeted (35, 36, 37). We have previously found that the concept of inserting a foreign Th epitope in a self-Ag is an efficient way to induce autoantibodies (38, 39, 40, 41). The present study suggests that insertion of a foreign Th epitope in a self-Ag may also be a suitable way to induce self-reactive CTLs. Combining the induction of self-reactive CTLs, strong T cell help, and induction of autoantibodies, as this vaccine design does, may be an advantage especially for vaccines based on membrane bound tumor associated self-Ags but also intracellular tumor Ags are potential targets for this technology.
We thank Anett Ravn for excellent technical and Anne Tolstrup for the target cells LB27.4-SIINFEKL and LB27.4-FILKSINE. We also thank Ranjeny Thomas and Graham Leggatt at Centre for Immunology and Cancer Research for valuable feedback to the manuscript.
L. Steinaa, P. B. Rasmussen, A. M. Wegener, L. Sonderbye, D. Leach, S. Mouritsen, and A. M. Gautam are either current or past employees of Pharmexa, and hold stock or equity in the company.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Abbreviations used in this paper: wt, wild type; RIP, rat insulin promoter; HA, hemagglutinin.