Of mice and humans: how good are HLA transgenic mice as a model of human immune responses?

Background Previous studies have defined vaccinia virus (VACV)-derived T cell epitopes in VACV-infected human leukocyte antigen-A*0201 (HLA-A2.1) transgenic (Tg) mice and A2.1-positive human Dryvax vaccinees. A total of 14 epitopes were detected in humans and 16 epitopes in A2.1 Tg mice; however, only two epitopes were independently reported in both systems. This limited overlap raised questions about the suitability of using HLA Tg mice as a model system to map human T cell responses to a complex viral pathogen. The present study was designed to investigate this issue in more detail. Results Re-screening the panel of 28 A2.1-restricted epitopes in additional human vaccinees and in A2.1 Tg mice revealed that out of the 28 identified epitopes, 13 were detectable in both systems, corresponding to a 46% concordance rate. Interestingly, the magnitude of responses in Tg mice against epitopes originally identified in humans is lower than for epitopes originally detected in mice. Likewise, responses in humans against epitopes originally detected in Tg mice are of lower magnitude. Conclusion These data suggest that differences in immunodominance patterns might explain the incomplete response overlap, and that with limitations; HLA Tg mice represent a relevant and suitable model system to study immune responses against complex pathogens.


Background
Model systems to study human responses to potential pathogens in general, and to complex viruses such as poxviruses in particular, are of considerable practical interest. They allow for characterization of immune responses directed against epitopes potentially capable of being recognized in humans in a rigorously controlled experimental situation, without having to rely on samples derived from exposed individuals. These issues are particularly rel-evant in the case of pathogens of bioterrorism concern, as well as emerging and re-emerging pathogens.
Several alternative approaches have been described to circumvent the need to use samples from infected humans. These include using mice re-colonized with human hematopoietic stem cells, primary in vitro restimulation utilizing peripheral blood mononuclear cells (PBMC) from healthy volunteers, and human leukocyte antigen (HLA) transgenic (Tg) mice. Murine/human chimeric systems appear to be associated with significant experimental variability, and responses of low magnitude and diversity compared to those observed in humans have been reported [1]. Primary in vitro restimulation systems are an attractive alternative, but are associated with limited throughput, high cost, labor intensiveness, and the lack of testing the impact of variables associated with host-pathogen interactions in a living organism.
Several groups have reported the development and validation of HLA Tg mice as a model system to study T cell responses restricted by human major histocompatibility complex (MHC) molecules [2][3][4]. It has been established that HLA Tg mice represent a fairly accurate model of human immune responses based on peptide immunizations. For example, Wenthworth and co-workers analyzed the repertoire of influenza-specific T cell responses obtained in HLA Tg mice with that obtained following primary induction of human cytotoxic T lymphocyte (CTL), and observed concordant results for approximately 85% of the peptides assayed [4].
The question of how suitable HLA Tg mice are for defining responses to complex pathogens and antigens has, however, not been addressed in detail. This issue is relevant because significant differences between humans and HLA Tg mice could exist at several levels. It is possible that coexpression of different MHC molecules might impact the TCR repertoire, in that certain specificities are lost while others are gained, as a function of the MHC alleles present [5]. Differences also exist in the cellular processing machinery operating in mice and humans at the level of the proteasome and TAP transport [6][7][8]. Discrepancies might also exist in the pattern of expression of viral antigens in infected cells of human and murine origin. Furthermore, it is currently unknown to what degree these differences might influence the global outcome of immune responses.

Results and discussion
Additional testing in A2.1 Tg mice reveals a more extensive repertoire overlap We previously detected 14 different epitopes utilizing PBMC from A2.1-positive human volunteers vaccinated with Dryvax ( Figure 1A) [9]. Of these epitopes, only two were also independently identified in VACV-infected A2.1 Tg mice [5,[13][14][15][16]. Based on these results, Terajima and Ennis hypothesized that the large number of potentially immunogenic peptides encoded by a complex pathogen such as VACV might explain the incomplete overlap observed [17]. Herein, we addressed this issue in more detail.
First, A2.1 Tg mice were inoculated with VACV and the reactivity against each of the 28 individual A2.1-restricted epitopes was evaluated. Responses were detected for all epitopes previously reported in HLA Tg mice ( Figure 1A; white and grey bars). In addition, responses were detected for 6 peptides that had been originally described in A2.1positive humans [9,14,18], but not reported in studies utilizing Tg mice [5,15] (Figure 1A; black bars). Altogether, responses directed against 8 of 14 (57%) epitopes reported in humans were now also detected in A2.1 Tg mice.
The present epitope screen differs in two experimental settings from the original study by Pasquetto et al. [5]. First, individual peptides were used here instead of peptide pools. Second, the initial epitope identification used splenocytes as effector cells [5], whereas the present study utilized purified CD8 + T cells. Herein we examined the impact of these two experimental differences. When three representative epitopes (A14L 51-59 , B14R 327-335 , and M1L 374-383 ) were tested either as individual peptides or in pools of 10 peptides, similar response magnitudes were observed ( Figure 1B). We also tested the response to the same 3 epitopes using either purified CD8 + T cells or splenocytes, and found that the higher non-specific interferon (IFN)-γ production obtained when using splenocytes led to a reduced stimulation index, and thereby decreased sensitivity ( Figure 1C).
We observed that the average responses against these epitopes not initially identified in Tg mice were ~3 times lower than those against the epitopes identified in the original screen (P = 0.0098; Figure 1D). Although the 3 dominant epitopes (A14L 51-59 , A17L 81-90 , and M1L 347-383 ) contribute to a significant portion of the higher response identified in mice, exclusion of these 3 epitopes still results in a significant difference in the magnitude of the response for the epitopes identified in humans versus mouse (P = 0.0012). Taken together, these results suggest that while 8 of 14 (57%) peptides reported in humans can also be detected in A2.1 Tg mice, differences in the pat-terns of immunodominance might render them less prominent and consequently make their detection more difficult.
The mice utilized by Pasquetto et al. were A2.1 bxd mice [5]. To examine whether expression of endogenous mouse MHC molecules might also contribute to the incomplete overlap, we tested the reactivity of the 14 epitopes identified in humans using CD8 + T cells from VACV-infected HHD A2.1 mice, which do not express endogenous mouse MHC [19]. However, no additional epitopes were recognized in the HHD mice, and the response magnitude and immunodominance pattern was similar to that in A2.1 bxd mice (data not shown).

Testing of a broader panel of human vaccinees reveals a more extensive overlap
As described above, a total of 16 epitopes were initially identified in HLA Tg mice [5,13,15]. Of these, only two (13%) were also independently identified in human vaccinees ( Figure 1A). Most of the candidate peptides were tested in only 6 donors, with each individual donor yield-Testing A2.1-restricted VACV T cell epitopes in A2.1 Tg mice using purified CD8 + T cells and individual peptides Figure 1 Testing A2.1-restricted VACV T cell epitopes in A2.1 Tg mice using purified CD8 + T cells and individual peptides. (A) Splenic CD8 + cells were purified from VACV-infected A2.1 Tg mice and incubated with peptide-pulsed Jurkat-A2.1/ K b cells, after which number of IFN-γ-producing cells was enumerated in an ELISPOT assay. Shown are the average net SFC/10 6 CD8 + T cells in response to each peptide; black bars, epitopes identified by Oseroff et al. [9] in A2.1 positive donors and recognized in Tg mice upon re-screening; white bars, peptides identified by Pasquetto et al. [5] in Tg mice; grey bars, epitopes identified in both studies. (B) The specific T cell response to three different epitopes (B14R 327-335 , M1L 374-383 , and A14L 51-59 ) was assessed 7 days after VACV infection in A2.1 Tg mice. Shown is the stimulation index for the epitopes tested either as individual peptides (black bars) or in pools of ten peptides (white bars). (C) Shown is the stimulation index for the epitope-specific responses by using either CD8 + T cells (black bars) or splenocytes (white bars). (D) Shown are the average net SFC/10 6 CD8 + T cells in response to groups of epitopes recognized in Tg mice upon re-screening (black bar) and identified in Tg mice in the original study by Pasquetto et al. [  ing a rather distinct pattern of epitope recognition. While some epitopes were identified in multiple donors, most of the epitopes were only detected in 1 out of the 6 donors investigated [9].
To test the hypothesis that the epitopes not initially detected in humans were associated with lower and/or less frequent responses, we examined the responses to the set of 16 A2.1-restricted epitopes identified in Tg mice using PBMC from 8 different A2-positive vaccinees. The samples tested included frozen cell aliquots from 5 of the donors tested in the original study of Oseroff et al. [5], as well as PBMC from 3 additional A2-positive vaccinees. PBMC were cultured in the presence of individual peptides, and the numbers of IFN-γ-producing cells were enumerated in an enzyme-linked immunosorbent spot (ELISPOT) assay. The same two peptides previously identified as positive in these donors were also re-identified in the course of the present experiments, and responses to an additional 5 peptides originally identified in A2.1 Tg mice, but not identified in the human donor screen of Oseroff et al. [9], were detected. Similar to what we observed in the mouse experiments, responses to the newly detected epitopes were in general lower than to epitopes identified in the initial screen (see Additional file 1). Overall, responses directed against 7 of 16 (44%) peptides initially identified in HLA Tg mice, were also detected in humans. Considering the large donor-todonor variation observed, it is likely that testing of additional donors would further increase the overlap between the two systems.

Conclusion
In summary, of the 28 epitopes identified in either humans or mice, in this study we showed that 13 were detected in both systems, with an overall concordance rate of 46% (Table 1). This degree of overlap is, albeit incomplete, much greater than previously detected. In a similar study, a 68% concordance rate was observed between human immunodeficiency virus type 1 (HIV-1)-specific CD8 + T cell epitopes recognized in both HLA-A2.1 Tg mice and HLA-A2 HIV-1 positive patients [20]. A higher degree of overlap was likely detected because HIV-1 epitope recognition in A2.1 Tg mice was evaluated following peptide immunization and not natural infection, a larger cohort of donors were tested, and HIV-1 has a significantly smaller genome than VACV. These data suggest that while the Tg mouse system is suitable to model immune responses to complex pathogens, caution should be taken in interpreting the results obtained.
Despite the more extensive and sensitive system used in the current study, approximately one half of the epitopes recognized specifically in one system were not recognized in the other. Differences in TCR repertoire between mouse and human might account for the different epitope recognition. However, peptide immunizations of the human epitopes in A2.1 Tg mice only failed in 2 of 14 instances to generate a T cell response. This is similar to the 85% success rate previously reported by Wentworth et al. [4], and suggests that repertoire differences account for a relatively minor percentage of the discrepancies. Differences in the processing apparatus and antigen presentation in murine versus human dendritic cell (DC) subsets and other antigen-presenting cells (APCs) involved in priming CTL immunity might also explain the differential recognition of some epitopes [7]. In mice, multiple DC subsets have been described, however, CD8 + DCs are considered the primary DC subset involved in directly activating naïve CD8 + T cells during VACV infection [21]. Recently, the CD8 -DC subset has also shown to play an important role [22]. In humans, various cell types are susceptible to VACV infection, including dermal DCs, Langerhans cells, and macrophages [23,24], but the primary APC responsible for naïve CD8 + T cell priming is unknown. As the peptides identified in the Tg mice were effectively processed in VACV-infected human APCs [5], differences in processing appear an unlikely explanation. Furthermore, while mice were immunized i.p. and humans were vaccinated by dermal scarification, we recently reported that either route generates similar T cell responses, and only minor differences in magnitude of responses were observed in mice [25].
Overall, our data suggest that while about half of the epitopes are recognized both in humans and Tg mice, the actual magnitudes vary, and thus differences in the immunodominance pattern contribute to the degree of overlap in the responses observed. Discrepancies in the kinetics of viral antigen expression in infected cells of human and murine origin might impact immunodominance [26]. Interestingly, a recent study mapping HLA-A2.1-restricted epitopes derived from modified VACV Ankara-infected human B cells by differential HPLC-coupled mass spectrometry found epitopes predominantly derived from early gene products [27]. However, we identified A2.1restricted epitopes in HLA Tg mice and human vaccinees from gene products expressed both early and late during the viral life cycle [28]. Despite having comparable T cell repertoires, the differences in immunodominance might be due to distinct naïve CD8 + T cell precursor frequencies in the two systems. We have recently demonstrated that precursor frequencies shape the CD8 + T cell immunodominance hierarchy following LCMV infection [29], and therefore this might also apply to a more complex virus, such as VACV. These conclusions are in agreement with those drawn from a recent study mapping the T cell responses in VACV-immune individuals, where it was pointed out that an epitope might be considered immunodominant if recognized by both humans and HLA Tg mice [30]. In summary, the present study suggests that, with limitations, HLA Tg mice represent a relevant and suitable model system for identification and validation of T cell epitopes recognized during the course of complex viral infection in humans.

Peptides
Peptides were synthesized by A and A Labs and purified to >95% homogeneity by reverse-phase high-performance liquid chromatography (HPLC). Purity was determined using analytical reverse-phase HPLC and amino acid analysis, sequencing, and/or mass spectrometry. Peptides were radio-labeled with the chloramine T method as described [31].

Characteristics of study population
Healthy males and females between 18 and 59 years of age were used in this study. Exclusion criteria were body weight of < 45.4 kg and established pregnancy. Recruited donors were vaccinated by arm scarification with a VACV (Dryvax) vaccination as a prophylactic measure either because of their potential exposure to VACV in a laboratory or hospital setting or because of their enrollment into military and health worker vaccination program. The blood draw was performed within 1 year of the Dryvax vaccination. All experiments carried out with human subjects were in compliance with the Helsinki Declaration. Institutional Review Board approval (FWA# 00000032) and appropriate informed consent were obtained for this study.

Viruses and infection
The Western Reserve strain of VACV was obtained from Dr. Bernard Moss (NIAID). Mice were infected i.p. with 2 × 10 6 PFU of VACV. On day 7 post-infection, the mice were sacrificed and splenic CD8 + T cells were used in mouse IFN-γ ELISPOT assays.