Summary. MHC class II molecules are essential for shaping the CD4+ T-cell repertoire in the thymus and for selecting antigenic peptides that are presented to CD4+ T cells in the periphery. A range of different mouse models humanized for HLA class II antigens have been developed to study the regulation of MHC-class II restricted immune responses. These mouse models have been used to identify immunodominant peptides that trigger diseases and to characterize the interactions of T-cell receptors with disease-associated peptides and MHC class II molecules. Peptides presented to CD4+ T cells in these mouse models were shown to be similar to peptides presented to CD4+ T cells in patients who carry the same MHC class II haplotype. Opportunities and limitations associated with these mouse models will be discussed and the potential application of these models for understanding the regulation of antibody responses against factor VIII in hemophilia A will be indicated.
Significance of MHC class II molecules for the regulation of immune responses against proteins
MHC class II molecules play an important role in the control of adaptive immune responses against protein antigens. They are essential for shaping the CD4+ T-cell repertoire in the thymus and they select the antigenic peptides that are presented to CD4+ T cells in the periphery . Today, it is generally accepted that B cells need the help of activated CD4+ T cells to develop high-affinity antibody responses against protein antigens [2,3]. CD4+ T cells express T-cell receptors that recognize antigen-derived peptides which are presented by MHC class II molecules expressed on specialized antigen-presenting cells in the periphery . Structural features of both the MHC class II molecule and the antigenic peptide determine the specificity of CD4+ T cells that can bind to the complex formed between MHC class II molecules and peptides [4–6]. The conditions under which CD4+ T cells interact with these complexes determine whether the immune system is nonresponsive, is activated to develop specific antibodies or is tolerized to suppress antibody development [6,7].
MHC-class II molecules are heterodimeric proteins. They consist of two non-covalently associated polypeptide chains, the alpha chain and the beta chain (Fig. 1) [for review see 8]. The amino-terminal alpha 1 (α1) and beta 1 (β1) regions of the polypeptide chains form an open binding groove for peptides. The binding groove presents peptides to CD4+ T cells [for review see 8]. Crystal structures of complexes between human MHC class II molecules and peptides have indicated that the peptide binding grooves of the different MHC class II isotypes are superimposable and that the backbone of peptides bound to the binding grooves is highly conserved [for review see 8]. The binding grooves are mainly characterized by the properties of the so-called P1, P4, P6 and P9 pockets (Fig. 1), which confer the specificity to the anchor residues of the peptides bound to the groove [for review see 8]. The conformation of the peptide (residues 1–9, see Fig. 1) that is bound to the groove is determined by hydrogen bonding by MHC class II residues to the peptide backbone. Therefore, the conformation adopted by the peptide is independent of its sequence and is different from the conformation of the peptide sequence in the context of its native protein [8,9].
MHC class II molecules bind peptides in endocytic vesicles of antigen-presenting cells. Peptides bound to MHC class II molecules are heterogenous in size (usually 12–26 mers). They are selected as families consisting of a common binding core (peptides 1–9, see Fig. 1) and varying flanking regions . The plasticity of peptide- MHC class II interactions permits the formation of multiple conformational isomers by the same peptide and MHC class II molecule . These conformational isomers are determined by the flanking regions of the peptide . Different conformational isomers of the same peptide-MHC class II complex might be recognized by different CD4+ T cells .
The identification of peptides selected by MHC class II molecules during natural processing of factor VIII (FVIII) will be of key importance in understanding the repertoire of FVIII-specific T cells and how these T cells modulate anti-FVIII antibody responses in patients with hemophilia A. Most interesting will be the answer to the question whether or not the same peptides are generated during processing of different biologic forms of FVIII, e.g. FVIII bound to von Willebrand factor as present in the circulation, thrombin-activated FVIII as present at the site of coagulation, and FVIII fragments as present after FVIII inactivation.
Opportunities for mouse models humanized for HLA class II antigens as models for human diseases
A range of different mouse models humanized for HLA-class II antigens (human MHC class II) have been developed to study the regulation of HLA class II-restricted immune responses, in particular the immunopathology of HLA class II-associated autoimmune diseases such as rheumatoid arthritis, multiple sclerosis, insulin-dependent diabetes mellitus or celiac disease [for review see 12, 13].
Initially, mouse models transgenic for human HLA class II antigens expressed the human HLA class II molecules in addition to the murine MHC class II . Results obtained with these models were difficult to evaluate because it was hard to differentiate between immune responses that were initiated by the murine MHC class II molecules and immune responses that were initiated by the human MHC class II molecules. In 1999, Madsen et al.  generated a knockout mouse that lacked all four of the classic murine MHC class II genes (MHCII Δ/Δ mice) via a large (80-kb) deletion of the entire class II region that was engineered by homologous recombination and Cre recombinase-mediated excision. This knockout mouse model has subsequently been used for developing mouse models that expressed only human MHC class II molecules on the background of a complete knockout of all murine MHC-class II molecules. The human MHC class II molecules in these new models have been shown to be functionally active. They participate in shaping the CD4+ T-cell repertoire in the thymus and mediate CD4+ T-cell responses in the periphery [for review see 12]. Furthermore, human MHC class II molecules determine the same disease susceptibility in the mouse models as they do in humans [for review see 12]. Even though these models might not be perfect (see following section), they mimic important aspects of human diseases. Thus, they provide a suitable in vivo testing ground for human diseases that are associated with the activation of CD4+ T cells. Murine T-cell clones and hybridomas are more easily maintained than human clones and mice permit manipulations that are not possible in humans.
In recent years, mouse models humanized for HLA class II antigens have been used to identify immunodominant peptides that trigger diseases, to characterize the interactions of T-cell receptors with disease-associated peptides and MHC class II molecules [for review see 12] and to search for new antigen-specific immunotherapies [12,16,17]. Importantly, peptides presented to CD4+ T cells in these mouse models were shown to be similar to peptides presented to CD4+ T cells in patients who carry the same MHC class II haplotype [for review see 12, 18].
Limitations of mouse models humanized for HLA class II antigens
The MHC class II genes are the most polymorphic genes in the mammalian genome. There are over 300 alleles of the DR beta genes, and over 100 alleles of the DQ alpha and beta genes. DR alpha is a nonpolymorphic gene [for review see 19]. Thus, every human has a unique combination of MHC class II genes (MHC class II haplotype) and we inherit one set of alleles from each parent. It is simply not possible to have animal models that cover all the potential human MHC class II haplotypes that are found in the human population. Consequently, any mouse model humanized for HLA class II antigens will always cover only a portion of the human population. However, despite the highly polymorphic nature of HLA class II genes, a majority of autoimmune diseases are linked to a limited set of HLA class II DR or HLA class II DQ alleles [for review see 19]. Three HLA class II haplotypes have been found to be the most autoimmune prone genes: HLA-DQ2/DR3, HLA-DQ6/DR15 and HLA-DQ8/DR4. These three haplotypes are associated with about 90% of all human autoimmune diseases [for review see 19]. Currently, it is not clear what makes these three haplotypes so unique among the potentially thousands of MHC class II haplotypes in the human population. Importantly, one of the three MHC class II haplotypes (HLA-DQ6/DR15) that are associated with autoimmune diseases was also shown to be associated with an increased risk for developing neutralizing anti-FVIII antibodies in patients with severe hemophilia A [20–22]. In several studies, both HLA-DQ6 and HLA-DR15 (the major serotype of HLA-DR2) were shown to be associated with an increased risk for the development of neutralizing anti-FVIII antibodies [20–22]. Therefore, it would be particularly interesting to develop hemophilic mouse models that are transgenic for HLA-DQ6 and/or HLA-DR15. It could be that the HLA-DQ6/DR15 haplotype is associated with a particular efficient presentation of peptides to CD4+ T cells. This could be an advantage for the defense against viral or bacterial infections, but a disadvantage if it comes to susceptibility for developing autoimmune diseases and developing antibodies against therapeutic proteins.
Another major limitation of HLA class II transgenic mice that needs to be considered is that the HLA class II molecule is only one of many proteins that are involved in the modulation of immune responses. Consequently, the HLA class II haplotype is only one of many potential genetic risk factors that contribute to the development of immunological diseases or the development of antibodies against therapeutic proteins. HLA class II molecules are essential for presenting peptides to CD4+ T cells. Therefore, HLA class II molecules are involved in shaping the repertoire of CD4+ T cells in the thymus and in regulating adaptive immune responses that involve CD4+ T cells in the periphery. Whenever these aspects of immune responses are of interest, mice transgenic for HLA class II antigens present a major advantage. However, when other aspects of the immune regulation are to be investigated, different types of mouse models should be used.
New strategies to improve humanized mouse models further are actively pursued. The combination of human HLA class II with other human transgens such as CD4 or disease-associated human T-cell receptors are examples [23,24]. More recent approaches focus on the reconstitution of immunodeficient mice with human hematopoietic stem cells that are expected to develop into a full human immune system [25–30]. Three such mouse strains have been described. All three strains have a knockout of the gamma-chain common to the IL-2 cytokine receptor family together with SCID [25–28] or Rag-2 [29,30] deficiencies. These deficiencies are bred on NOD, NOD/Shi or Balb/c backgrounds. All three lines of mice develop human myeloid and lymphoid cells when inoculated with human hematopoietic stem cells. Furthermore, they develop mature human B and T cells and express antigen-specific MHC-restricted human T-cell responses [25–30]. Future studies will show whether these models are suitable for studying the regulation of antibody responses against protein antigens.
Hemophilic mice that express human MHC-class II molecules
Until recently, there have been no animal models for hemophilia A that expressed human MHC class II molecules. To overcome this limitation, we developed a humanized mouse model for hemophilia A in which the regulation of anti-FVIII immune responses is driven by FVIII-derived peptides that are presented by the human MHC class II haplotype HLA-DRB1*1501 . The rationale for choosing this particular haplotype is the reported connection of HLA-DRB1*1501 with major immunologic diseases (see previous section) and the reported association of HLA-DRB1*1501 with an increased risk that patients with severe hemophilia A have for developing FVIII inhibitors [20,21]. Although a study by Astermark et al.  published in 2006 did not confirm the association of the HLA-DRB1*1501 haplotype with an increased risk for inhibitor development, a recent report by Oldenburg`s group presenting data obtained from 260 well characterized patients with severe hemophilia A reconfirmed this association . In any case, we would like to emphasis that our new hemophilic mouse model can only represent a certain proportion of patients, namely, patients who express HLA-DRB1*1501. Further models expressing other major human MHC class II haplotypes should be useful to supplement our current model.
The parent line that we used to develop our new hemophilic mouse model was the HLA-DR2 Kim mouse line [33,34] that was kindly provided by Lars Fugger. This mouse line expresses chimeric human MHC class II molecules that consist of the α1 and β1 sequences of HLA-DRA1*0101 and HLA-DRB1*1501 and the α2 and β2 domains of the murine IEα and IEβ, respectively, on the background of a knockout of all normal mouse MHC class II molecules. This chimeric human–mouse MHC class II protein contains the human sequences for the binding of peptides and the murine sequence for the binding of murine CD4. The hybrid protein was shown to allow an optimal human MHC class II-restricted T-cell response . We further backcrossed the mice to the MHC class II knockout mice MHCIIΔ/Δ  resulting in offsprings that were either heterozygous for the chimeric human MHC class II transgene or had a complete knockout of all MHC class II molecules. Both heterozygous and knockout mice were then backcrossed several times to hemophilic C57BL/6 E17 mice [35–37] to generate mice that had a knockout of the murine FVIII gene (E17) and a knockout of all murine MHC class II genes, and were either heterozygous for the chimeric human MHC class II molecule or had a complete knockout of all MHC class II molecules.
Flow-cytometric analysis demonstrated that immune cells obtained from the spleen, lymph nodes and peripheral blood of humanized E17 HLA-DRB1*1501 mice did not express any murine MHC class II but did express the human HLA-DR protein. For comparison, conventional E17 hemophilic mice only express the murine MHC class II proteins, whereas E17 mice that have a complete knockout of all MHC class II genes (negative control) do not express any MHC class II proteins (Fig. 2).
When we studied the distribution of CD4+ and CD8+ T cells in all major lymphoid organs of the humanized E17 HLA-DRB1*1501 mice, we found a normal distribution of these cells, which indicates that the humanized MHC-class II protein is functionally active and is able to select positively CD4+ T cells in the thymus.
To test the sensitivity of E17 HLA-DRB1*1501 mice to develop antibodies against human FVIII, we treated mice with weekly i.v. doses of human FVIII and studied the development of anti-FVIII immune responses. About 80% of all humanized hemophilic mice developed anti-FVIII antibodies that were detectable after eight i.v. doses of FVIII and had neutralizing activity (Fig. 3). In comparison, E17 hemophilic mice with a complete knockout of all MHC class II molecules did not develop any anti-FVIII antibodies (Fig. 3). When we compared E17 HLA-DRB1*1501 with conventional E17 C57BL/6 mice, the anti-FVIII antibody response in E17 HLA-DRB1*1501 was weaker and showed higher variability between individual animals (Figs 3 and 4). One of the reasons for the higher variability in the humanized mice could be the mixed background (about 60% C57BL/6 and 40% S129) that is currently being backcrossed to obtain a pure C57BL/6 background. Furthermore, some of the E17 HLA-DRB1*1501 mice might develop immune tolerance rather than antibodies after treatment with FVIII. The IgG subclass distribution of the anti-FVIII immune responses was similar in E17 HLA-DRB1*1501 and in conventional E17 C57BL/6 mice (Fig. 4).
Currently, we are using the new E17 HLA-DRB1*1501 mouse model to generate large libraries of FVIII-specific HLA-DR15 restricted CD4+ T-cell hybridomas that are needed to address a variety of important questions relating to the regulation of anti-FVIII antibody responses by FVIII-specific CD4+ T cells. Furthermore, we include these mice in the preclinical development of new FVIII products and use them for the design of new immunotherapeutic strategies to induce FVIII-specific immune tolerance.
Despite many concerns around the use of mouse models to mimic human immunological diseases, there are currently no real alternatives to murine models for studying important questions of in vivo immune regulation. Therefore, tremendous efforts have been invested in the development of improved mouse models that express important elements of the human immune system. MHC class II molecules play an important role in the control of adaptive immune responses against protein antigens. They are vital for shaping the CD4+ T cell repertoire in the thymus and they select the antigenic peptides that are presented to CD4+ T cells in the periphery. Whenever these aspects of immune responses are of interest, mice transgenic for HLA-class II present a major advantage. However, when other aspects of the immune regulation are to be investigated, different types of mouse models should be used.
Currently, new mouse models are in development that express a full human immune system derived from human haemopoietic stem cells. Future studies will show whether these models are suitable for studying the regulation of antibody responses against protein antigens.
We would like to thank Nicole Pfeffer, Fatima Al-Awadi, Monika Grewal, Nida Mohammad, Erika Gehringer, Christian Lubich and Elisabeth Hopfner for technical assistance. Furthermore, we are grateful to Elise Langdon-Neuner for editing the manuscript. This work was supported by Baxter BioScience.
Disclosure of Conflict of Interests
All authors are employees of Baxter BioScience, Vienna, Austria.