Improved Transgenic Mouse Model for Studying HLA Class I Antigen Presentation

HLA class I (HLA-I) transgenic mice have proven to be useful models for studying human MHC-related immune responses over the last two decades. However, differences in the processing and presentation machinery between humans and mice may have profound effects on HLA-I restricted antigen presentation. In this study, we generated a novel human TAP-LMP (hTAP-LMP) gene cluster transgenic mouse model carrying an intact human TAP complex and two human immunoproteasome LMP subunits, PSMB8/PSMB9. By crossing the hTAP-LMP strain with different HLA-I transgenic mice, we found that the expression levels of human HLA-I molecules, especially the A3 supertype members (e.g., A11 and A33), were remarkably enhanced in corresponding HLA-I/hTAP-LMP transgenic mice. Moreover, we found that humanized processing and presentation machinery increased antigen presentation of HLA-A11-restricted epitopes and promoted the rapid reduction of hepatitis B virus (HBV) infection in HLA-A11/hTAP-LMP mice. Together, our study highlights that HLA-I/hTAP-LMP mice are an improved model for studying antigen presentation of HLA-I molecules and their related CTL responses.

HLA class I (HLA-I) transgenic mice have proven to be useful models for studying human MHCrelated immune responses over the last two decades. However, differences in the processing and presentation machinery between humans and mice may have profound effects on HLA-I restricted antigen presentation. In this study, we generated a novel human TAP-LMP (hTAP-LMP) gene cluster transgenic mouse model carrying an intact human TAP complex and two human immunoproteasome LMP subunits, PSMB8/PSMB9. By crossing the hTAP-LMP strain with different HLA-I transgenic mice, we found that the expression levels of human HLA-I molecules, especially the A3 supertype members (e.g., A11 and A33), were remarkably enhanced in corresponding HLA-I/hTAP-LMP transgenic mice. Moreover, we found that humanized processing and presentation machinery increased antigen presentation of HLA-A11-restricted epitopes and promoted the rapid reduction of hepatitis B virus (HBV) infection in HLA-A11/hTAP-LMP mice. Together, our study highlights that HLA-I/hTAP-LMP mice are an improved model for studying antigen presentation of HLA-I molecules and their related CTL responses.
Human leukocyte antigen class I (HLA-I) molecules that present antigenic peptides to CD8 + T cells and trigger the cytotoxic T lymphocyte (CTL) response are essential for the human immune system to combat viral infections and clear transformed tumor cells 1,2 . A better understanding of HLA-I-related antigen processing and presentation would improve the rational design of preventive or therapeutic vaccines against viral infection and cancer. Early studies using HLA-I transfected into murine cells 3,4 , as well as later efforts utilizing HLA-I transgenic mice [5][6][7] , have provided a wealth of knowledge about the antigen presentation of different HLA-I molecules. However, existing differences between humans and HLA-I transgenic mice make it difficult to determine whether responses in the transgenic mice exactly reflect responses in humans 8 . Distinctions between the mouse and human antigen processing and presentation machinery may account for some of these differences 8 .
In both humans and mice, the MHC class I antigen presentation pathway proceeds through several stages 9,10 : i) endogenous proteins in the cytosol are degraded into short peptides of 3-22 residues by the proteasome or immunoproteasome; ii) peptide products of 8-12 amino acids are transported into the endoplasmic reticulum (ER) by a dimer complex, the transporter associated with antigen processing (TAP); and iii) empty MHC class I molecules in the ER are stabilized by binding to suitable peptides to form peptide-MHC complexes, which are then exported to the cell surface for presentation to CD8 + T cells. The TAP complex, which is composed of TAP1 and TAP2, is essential for peptide transportation into the ER, where the peptides bind to MHC class I molecules. Deficiency of TAP1 and/or TAP2 in mice and humans results in a severe defects in MHC class I antigen presentation and a substantial reduction in CD8 + T-cell numbers [11][12][13][14][15] .
Previous studies of human or animal TAP transportation demonstrate that the TAP complex selects peptides with preferential sequences, and TAP binding affinity has a significant impact on epitope presentation [16][17][18][19][20][21] . A minimal TAP affinity is required for peptide presentation. Epitopes with high TAP affinity can easily to be selected and recognized by CTLs. Interestingly, the peptide binding specificities between human and mouse TAP are quite different. The murine TAP displays strong specificity for binding to peptides with hydrophobic C-termini, and the

Results
Generation of hTAP-LMP transgenic mice. Differences between human and mouse antigen processing and presentation machinery raise the possibility that HLA-I-restricted antigen presentation is not intact in HLA-I transgenic mice 8 . Thus, we generated a novel humanized TAP-LMP transgenic mouse by microinjecting pronuclei with a human BAC clone (RP11-10A19) encoding the intact hTAP-LMP gene cluster (Fig. 1A). Sequence alignment of BAC RP11-10A19 revealed that this BAC clone carries TAP1*0101 and TAP2*0201, both of which occur at high frequencies in different human populations 31,32 . By using PCR screening, two founder mice (F2 and F14) carrying all six human genes were established (Fig. 1B). mRNA and protein expression of human TAP1, TAP2, PSMB8, and PSMB9 in both founder mice were further confirmed by using real-time PCR and western blotting, respectively (Fig. 1C,D). Additionally, the human TAP1 mRNA expression level in the transgenic mice was approximately equal to that in human PBMC ( Supplementary Fig. S1). Transgenic expression of this hTAP-LMP gene cluster had little effect on mouse T cell homeostasis, as indicated by the normal percentage of CD4 + and CD8 + T cells found in the spleens and thymus of hTAP-LMP transgenic mice ( Fig. 1E and Supplementary  Fig. S2). The transgenic mice of the F14 founder were kept and used in further experiments.
Enhancement of HLA class I expression in HLA-I/hTAP-LMP transgenic mice. We next set out to test whether expression of the hTAP-LMP gene cluster promoted HLA-I-restricted antigen presentation. HLA-A3 supertypes, including A11 and A33, which prefer to bind peptides with small or aliphatic residues at position 2 and basic residues at the C-terminus (R or K) 27 , are more likely to rely on human TAP than other HLA supertypes. In addition, an HLA-A11 transgenic mouse that is a representative mouse model for the A3 supertype displays a defect in processing natural A11-restricted epitopes 5,29 . Therefore, hTAP-LMP mice were crossed with HLA-A11 transgenic mice to generate HLA-A11/hTAP-LMP double transgenic mice, and endogenous antigen presentation was evaluated by surface staining of HLA-A11. We found that the expression of surface mouse class I (H2-K b ) protein and mRNA was not significantly affected by transgenic expression of hTAP-LMP ( Fig. 2A,B). In contrast, surface HLA-A11 molecules but not their mRNA levels were strikingly increased in the HLA-A11/ hTAP-LMP transgenic animals (Fig. 2C,D). Indeed, approximately four-fold higher HLA-A11 levels were found in the HLA-A11/hTAP-LMP animals than the control HLA-A11 mice (Fig. 2D). This result suggests that reconstitution of humanized TAP-LMP enabled more HLA-A11-matched peptides to be transported into the ER to stabilize HLA-A11 molecules and increase their restricted antigen display.
The promotion of HLA-I antigen presentation by hTAP-LMP was highly selective, as we also found dramatic enhancement of another member of the A3 subtype, HLA-A33 molecules, in the HLA-A33/hTAP-LMP double transgenic mice (Fig. 2E), whereas the surface presentation of HLA-A2 only showed a slight up-regulation in HLA-A2/hTAP-LMP mice (Fig. 2F). This is consistent with their different peptide binding preferences (i.e.,   binding to peptides with positive charges at their C-termini). The HLA-A2 molecule also has its own peptide in the signal sequence that does not require cytosolic processing or TAP transport 33,34 and, thus, is less affected by the transgenic hTAP-LMP genes.

Increased CTL responses against HLA-A11-restricted epitopes in the HLA-A11/hTAP-LMP mice.
To test whether the introduction of the human TAP-LMP gene cluster would have an impact on HLA-A11-restricted CTL responses and to better establish a link between TAP affinity, HLA-I expression, and CTL responses, we utilized DNA vaccination to allow antigens to be processed and presented in the intracellular pathway. HLA-A11/hTAP-LMP mice were prime-boost immunized via intramuscular injection of plasmid pcDNA3.1(+ )/HBcAg, which encodes the full-length hepatitis B virus core antigen (HBcAg) (Fig. 3A). HLA-A11-restricted CTLs were evaluated by Elispot assays and intracellular IFN-γ cytokine staining (ICS). Two known HLA-A11-restricted epitopes, HBc 141-151 (STLPETTVVRR) 5 with high affinity to human TAP (TAP score 0.67, IEDB Analysis Resource, http://tools.iedb.org/processing/) and HBc 88-96 (YVNTNMGLK) 35 with low TAP affinity (TAP score 0.15), as well as 19 peptides (Table 1) with K/R C-termini that potentially elicit HLA-A11-restricted CTL responses, were synthesized and used in the Elispot assays (Fig. 3B). Initially, the 19 peptides were divided into three pools for ex vivo stimulation (Table 1). Because mouse class I molecules prefer to bind peptides with hydrophobic C-termini, the studied peptides were unlikely to trigger mouse class I-restricted CTLs.
Consistent with a previous influenza virus infection model 5 , only weak HLA-A11-restricted CTL responses were detected by Elispot assays in the A11 transgenic mice following DNA vaccination (Fig. 3B above). However, much stronger CTL responses against HBc 141-151 (but not HBc 88-96 ) and pool 2 peptides were found in HLA-A11/ hTAP-LMP mice (Fig. 3B above). Interestingly, further analysis revealed that the only peptide from pool 2 that was capable of stimulating the IFN-γ response was HBc 142-152 (Fig. 3B below). The HBc 142-152 peptide was derived from the same region as HBc 141-151 , with the differences being that it lacks a serine at its N-terminus and has an additional arginine at its C-terminus. This suggests that HLA-A11-restricted epitopes in the HBcAg DNA vaccination model dominantly reside between residues 141 and 152. Moreover, a much higher number of HBc 141-151 epitope-specific CTLs detected via ICS was also found in HLA-A11/hTAP-LMP mice, highlighting the importance of humanized TAP-LMP in the HLA-A11-dependent CTL response (Fig. 3C).
Similarly, by DNA vaccination of another plasmid encoding a minigene that contains a known HLA-A11-restricted epitope, NP 91-100 (RTGGPIYRR) 36 with high TAP affinity (TAP score 0.62), a stronger NP 91-100 -specific CTL response was also found in HLA-A11/hTAP-LMP mice as analyzed by ICS (Fig. 3D, left and middle panel) and IFN-γ Elispot assays (Fig. 3D, right panel). We then detected the CTL responses to the HLA-A2 restricted epitope HBc 18-27 (FLPSDFFPSV) with low TAP affinity (TAP score 0.07) 37 . There was no significant difference between HLA-A2/hTAP-LMP mice and HLA-A2 mice ( Supplementary Fig. S3). Overall, our results indicated that the introduction of hTAP-LMP prompts better intracellular antigen presentation of HLA-A11 molecules and notably improved HLA-A11-restricted CTL responses to epitopes with high affinity for human TAP.
Antigen presentation of HLA-A11-restricted epitopes in a long peptide vaccine elicited anti-viral immunity. HBc 141-151 is a CTL epitope shared by other class I alleles in the HLA-A3 supertype 5,38 . In addition, this epitope elicits a specific CTL response that is correlated with HBV clearance 39,40 . To explore the possibility of boosting the HBc 141-151 -specific CTL response as an intervention method to inhibit hepatitis B virus infection, we designed a long peptide vaccine containing residues 123-157 of HBcAg (HBc 123-157 , Fig. 4A), which encompassed the HBc 141-151 CTL epitope. Unlike short peptides, long peptide vaccines (which are capable of forming tertiary structure to protect the peptides from exopeptidase degradation 41 ) are predicted to be internalized and cross-presented by professional APCs. To test potential advantages of the long peptide HBc 123-157 , HLA-A11/ hTAP-LMP mice and a control strain were subcutaneously immunized with the long peptide HBc 123-157 . Two weeks later, hydrodynamic injection (HDI) of pAAV/HBV1.2 was used to mimic HBV infection in vivo 42 (Fig. 4A). The plasmid pAAV/HBV1.2 contains a replication-complement HBV DNA sequence that can mimic HBV infection after liver-targeting HDI into mice. The mouse models will help to further explore new treatment of HBV infections 42 .

Discussion
In the last two decades, HLA-I transgenic mice have proven to be a unique in vivo model to study human class I-restricted CTL responses in various infectious diseases 6,29 , as well as cancer immunotherapy 43 . However, the murine antigen processing and presentation machinery is not capable of completely mimicking human HLA-I-restricted antigen presentation. In this study, a novel BAC transgenic mouse carrying the human TAP1, TAP2, PSMB8, and PSMB9 genes (hTAP-LMP mice) was generated, and we found that reconstitution of the hTAP-LMP gene cluster notably improved human HLA-I antigen presentation and restricted CTL responses. This effect was especially evident in the A3 supertype. Our data support the notion that particular HLA-I molecules co-evolved with TAP-LMP for efficient peptide processing and presentation. This research also highlights the potential for HLA-I/hTAP-LMP mice as an improved experimental model for studying antigen presentation of HLA-I molecules and their related immune responses.
HBV infection is the most common liver disease in the world. More than 350 million individuals are infected with HBV, and the estimated number in China alone is close to 100 million 44,45 . HLA A*1101, a member of the A3 supertype, is the major HLA-I allele in chronic hepatitis B patients from China 46 . Thus, identification of HLA-A*11:01-restricted HBV epitopes that can boost protective CTL responses are important for the treatment of chronic HBV infection. Here, using our novel HLA-A11/hTAP-LMP mice, we demonstrated that a long peptide vaccine containing residues 123-157 of HBcAg (HBc 123-157 ) could be efficiently presented to APCs and elicit protective HBc 141-151 -specific CTL responses. Thus, HBc 123-157 could have important therapeutic potential in preventing HBV infection. This result also suggests that the long peptide was more efficiently cross-presented in HLA-A11/hTAP-LMP mice, which is consistent with previous research showing that cross antigen presentation of long peptides is dependent on proteasome and TAP function 47 . However, a recent study by Ma et al. 48 found that cross-presentation of long peptides requires a vacuolar pathway that depends on newly synthesized MHC class I molecules but not the proteasome or TAP molecules 48 . Thus, further studies are needed to clarify whether enhanced cross antigen presentation of long peptides in HLA-A11/hTAP-LMP mice is associated with transgenic human TAP and LMP molecules.
Because the TAP-LMP gene cluster plays important roles in HLA-I antigen presentation 30 , future studies on the effect of hTAP-LMP on HLA-I-related human diseases are particularly interesting. HLA-A33 molecules are related to susceptibility to persistent infection by HBV 49,50 and Enterovirus 71 infection 51 . It will be of great interest to use both HLA-A33/hTAP-LMP and HLA-A33 transgenic mice to study the contributions of HLA-A33 molecules to virus infection.
Though the TAP and LMP molecules are tightly linked as a gene cluster in HLA-I/hTAP-LMP mice, it is interesting to clarify their different contributions to antigen presentation of HLA-I molecules. One of the most obvious effects from the transgenes is that the HLA-A11 expression levels were dramatically elevated in HLA-A11/ hTAP-LMP mice. To determine whether this effect is due to the human TAP transgene, splenocytes of HLA-A11/ hTAP-LMP mice were infected with a retrovirus expressing a HSV-2 protein ICP47-2 which was reported to specifically inhibit human TAP 52 . Interestingly, HLA-A11 expression was reduced robustly and sensitively in splenocytes that were successfully infected by ICP47-2 exressing virus, while the H2-K b expression displayed a less sensitive manner (Supplementary Fig. S5). The reduction of H2-K b expression was expected because ICP47-2 also inhibits murine TAP transport 52 . Importantly, in a previous study, inhibition of human TAP by ICP47 molecules cound downregulate HLA class I in human cells 53 . Thus, it is quite possible that transgenic human TAP molecules make a substantial contribution to the elevation of HLA-A11 molecules and the induction of strong CTL responses. Regardless, whether the human LMP molecules have a similar effect is uncertain and will be tested in further studies.
In conclusion, we demonstrated that HLA-I/hTAP-LMP transgenic mice are an efficient in vivo model for studying HLA-I antigen presentation and CTL responses. We expect that these mice will be useful tools for future vaccine development and cancer immunotherapy. restricted CTL responses were synthesized and used in the Elispot assays. Initially, the 19 peptides were divided into three pools for ex vivo stimulation (Table 1). Single or overlapping peptides from pool 2 were further used to screen for HLA-A11-restricted epitopes. (C) Splenocytes as in (B) were used to test for HBc 141-151 -specific CD8 + T cells by ICS. The left plot shows the representative FACS diagram, indicating the percentage of CD8 + IFN-γ + cells in the total CD8 + T cells; the right chart depicts the cumulative data. (D) A minigene encoding a known HLA-A11-restricted peptide, NP 91-100 (RTGGPIYRR), was used to vaccinate HLA-A11 and HLA-A11/ hTAP-LMP mice. The NP 91-100 -specific CTL response was tested by IFN-γ ICS (left) and IFN-γ Elispot assays (right). The results are representative of at least three independent experiments with n ≥ 3. Data represent the mean ± SD. *p < 0.05 and **p < 0.01, unpaired t-test.
Mice. BAC RP11-10A19 was microinjected into pronuclei of fertilized B6 × DBAF1 mouse oocytes to generate human TAP transgenic mice (hTAP-LMP mice). The hTAP-LMP mice were further back crossed at least seven generations to C57BL/6. HLA-A2 transgenic mice in the C57BL/6 background were kindly provided by Songdong Meng's lab. HLA-A11 transgenic mice were purchased from Taconic (Model# 9660). HLA-A33 transgenic mice were generated in our lab by microinjection of the Rosa26 BAC (RP24-85L15 CHORI, Oakland, CA, USA) inserted with a HLA-A*3303/K b fused gene into pronuclei of fertilized B6 × DBAF1 mouse oocytes, and the mice were back crossed at least seven generations to C57BL/6. HLA-A2/hTAP-LMP, HLA-A11/hTAP-LMP, and HLA-A33/hTAP-LMP mice were obtained by crossing HLA-A2, HLA-A11, and HLA-A33 mice with hTAP-LMP mice, respectively. All transgenic mice were maintained as heterozygotes. All mice were housed under specific pathogen-free conditions at the Institute of Microbiology, Chinese Academy of Sciences in accordance with the guidelines for care and use of laboratory animals established by the Beijing Association for Laboratory Animal Science. All mouse experiments were performed in accordance with the "Regulation of the Institute of Microbiology, Chinese Academy of Sciences of Research Ethics Committee". The protocol was approved by the Research Ethics Committee of the Institute of Microbiology, Chinese Academy of Sciences (permit number PZIMCAS2012003). Lysates of splenocytes from hTAP-LMP mice were used to detect human TAP1, TAP2, PSMB8, and PSMB9 expression by western blotting. An anti-human TAP1 mAb (clone 148.3, Cat MABF125, Merck Millipore), anti-human TAP2 mAb (clone TAP2.17, MBL), anti-human PSMB8 mAb (clone 1A5, CST), and anti-human PSMB9 mAb (clone 792520, R&D) were used as primary antibodies, and anti-mouse IgG/HRP (ZDR-5307, ZSGB-BIO, Beijing) was used as the secondary antibody. The mouse anti-β -actin antibody (DKM9001, Tianjin Sungene Biotech) was used to detect β -actin as an internal reference.
Immunization and HDI. For intramuscular immunization, 100 μ g of DNA vaccine (pcDNA3.1(+ )/HBcAg or pcDNA3.1(+ )/minigene) dissolved in 100 μ L of PBS was injected into the tibialis anterior muscle (50 μ L per leg), followed with electroporation at the injection site 42 . The injection was performed twice within a 2-week interval. Eleven days after the last injection, the mice were sacrificed, and splenocytes were used to analyze peptide-specific CTLs by IFN-γ Elispot or IFN-γ ICS. Subcutaneous immunization of peptides was performed as previously described 5 with some modifications. Briefly, HBc 123-157 (100 μ g/mouse) and the helper IA b -restricted epitope HBc 128-140 (100 μ g/mouse) in PBS/5% DMSO were emulsified with IFA and subcutaneously injected (s.c.)

Position
Sequence Pool into the base of the tail. Two weeks after vaccination, the mice were hydrodynamically injected with pAAV/ HBV1.2 plasmids (10 μ g/mouse) in PBS with a volume (mL) equivalent to 8% of the mouse body weight (g), as previously described 42 . The mice were sacrificed 7 days later, and splenocytes were used to analyze CTL responses were subcutaneously (s.c.) injected with 100 μ g of HBc 123-157 . Two weeks later, 10 μ g of pAAV/HBV1.2 was hydrodynamically injected to mimic HBV infection. HBc 141-151 -specific CD8 + T cells were determined using IFN-γ ICS. The left plot shows the representative FACS diagram, indicating the percentage of CD8 + IFN-γ + cells in the total CD8 + T cells; the right chart depicts the cumulative data. (C) Relative serum HBsAg levels of the mice in (B) were determined by using ELISAs at day 7 after HDI. (D) Correlation plot of relative serum HBsAg and the percent of HBc 141-151 -specific CD8 + T cells in HLA-A11 or HLA-A11/hTAP-LMP mice of (B,C). Each symbol represents data from one animal. Data represent the mean ± SD. **p < 0.01, unpaired t-test.