This figure is a modification of the advertisement used for the seminar club event. The portrait photograph was kindly provided by Prof. Kaufman. The MHC figure is from Kaufman et al., Cell 1984 and was also used by Prof. Jack Strominger in a 2006 article describing his MHC journey.

Summary of ICMS Seminar Club presentation on Friday, October 29, 2021.

Title: The MHC and disease: a new understanding how it works

Speaker: Prof. Dr. Jim Kaufman, Professor of Immunology, Institute for Immunology and Infection Research, University of Edinburgh, UK.

On Friday, October 29, by Zoom, Prof. Kaufman gave a presentation to the ICMS Seminar Club of Fujita Health University. There were >40 participants. Prof. Kaufman has been involved in major histocompatibility complex (MHC) research at the top level from almost the very molecular beginning more than 40 years ago. This was reflected in the depth of his story and the answers to the questions of the audience. He is also a very kind person with a lot of humor, which made the event very enjoyable. We had a mixed audience, including both MHC experts and researchers who are not so familiar with MHC. Although Prof. Kaufman’s introduction of MHC was well-spoken, it may have been too fast for many of the people working in very different fields. I heard very positive reactions from all attendees who work with MHC, whereas several others said they had struggled to understand. The most positive reaction came from Prof. Uwe Fischer, who does study MHC, as he mentioned that Prof. Kaufman’s presentation was one of the best he ever attended. Luckily, several young researchers with a general background in immunology also said that they could understand the story, and one highlighted that she had enjoyed it very much.

For me, personally, it was a great presentation as, more or less, it was the first time that I felt rather convinced that an association between MHC polymorphism and differences in resistance against infectious diseases had been found. Of course, since this is my field of expertise, I am aware of many studies that have made claims of this nature previously, but most of them are (arguably) unconvincing when one starts analyzing the details. Prof. Kaufman explained how several lines of evidence had led to his new model, and this personal recollection gave a layer of insight that only an oral presentation can give.

Because some attendees had trouble understanding the content, I will write this summary a bit different from usual. The sections below are: (1) Summary of Content, which is a short general description of the presentation contents; (2) The Generalists versus Specialists model, which tries to describe Prof. Kaufman’s main message in a detailed but simple form; (3) The questions by the audience.

Summary of Content

All proteins in our body, our own as well as those from pathogens, are continuously broken down with small peptides as an intermediate form. To monitor conditions and to search for possible dangers, the immune system binds these peptides to major histocompatibility complex (MHC) molecules that at the cell surface offer these peptides to T lymphocytes for screening. There are MHC class I (MHC-I) and class II (MHC-II) molecules, which have similar structures but different functions. MHC-I molecules offer peptides to CD8+ cytotoxic T lymphocytes and MHC-II molecules offer peptides to CD4+ helper or regulatory T lymphocytes. MHC alleles (allelic molecules are variants encoded from the same gene in the same species) differ in their peptide binding grooves in which a single peptide (e.g., derived from a virus) can be bound. Each MHC allele can bind a variety of peptides but not all peptides, because at some of the peptide positions its amino acids need to be agreeable with the groove properties of the specific MHC allele.

Prof. Kaufman introduced major histocompatibility complex (MHC) molecules and the genomic region where their genes are located. This region is also called MHC but can, to avoid confusion, be written as “Mhc.” The Mhc region encodes several polymorphic (means variable between individuals of the same species) MHC molecules and, in some but not in other species, can also be polymorphic for other proteins involved in MHC presentation (e.g., the TAP peptide transporter proteins and the tapasin chaperone protein as found in chicken). Arguably, despite compelling theories which have some (but limited) empirical support, we probably aren’t sure why MHC molecules are so variable.

After giving a general introduction into MHC/Mhc, Prof. Kaufman discussed what may be considered the strongest cases indicating that MHC polymorphism can be associated with differences in resistance against infectious diseases: HIV (AIDS) in humans and several diseases in chicken. Chicken is Prof. Kaufman’s model species of choice. Compared to most mammals, chicken have only a few MHC genes in their Mhc region which may result in a bigger impact if one of those MHC genes is different. That may explain why in chicken, unlike in humans, associations between Mhc haplotypes (a haplotype is a set of linked genes that in one individual of the same species can be different from that in another individual) and disease resistance against several pathogens were readily found.

The standard theory for explaining MHC-allele associated differences in infectious disease resistance is that the alleles differ in their ability to present an “immunodominant” peptide (epitope) of the relevant pathogen. Empirical evidence for this theory, however, can be considered scarce at best. 

Prof. Kaufman introduced a new concept of MHC polymorphism, one he automatically stumbled upon when investigating chicken, but which also seems to partially explain HIV resistance in humans. Rather than looking at the preferred peptide ligand features of each MHC-I allele, he looked at two MHC allelic properties that he found to be correlated: Selectivity (for peptide ligands; the width of the peptide ligand repertoire) and Expression levels (the amount of an MHC-I allele found at the cell surface). Based on these properties, he distinguished:

Generalists: These alleles bind a wide variety of peptides (they are “promiscuous”) and are expressed at lower levels on the cell surface.

Specialists: These alleles are more selective and only bind a narrow repertoire of peptides (they are “fastidious”) and are expressed at higher levels on the cell surface.

The specialist alleles appear to form more stable complexes with their bound peptides than the generalist alleles. This results in a difference in MHC allele longevity (because without bound peptide MHC proteins are degraded) and can importantly explain why the specialist alleles are found at higher levels at the cell surface. At least in humans, an important mechanical explanation for why the specialist alleles produce surface complexes of a higher stability is that they rely stronger on the chaperone tapasin, which assures that they only bind high affinity peptides.

In chicken, it was readily found that against several of the common infectious diseases plaguing the poultry industry, the generalist MHC-I alleles confer a higher resistance. This may explain the dominant frequencies of these generalist MHC-I alleles in modern commercial chicken breeds.

Fascinating about the finding by Prof. Kaufman and others is that both in chicken and humans the distinction between specialist and generalist alleles can be made. However, in humans, possessing three MHC class I genes HLA-A, -B, and -C, as for the width of the presented peptide repertoire there is already a “generalist” tendency by having a higher gene number compared to the single dominant MHC-I gene in chicken (see Fig. 23 below). This may explain why initially only a few specialist HLA-B alleles were found to confer a higher protection against HIV, as these specific alleles present immunodominant peptides of which the sequences the virus cannot mutate without loss of its fitness. However, in a very recent study, it was found that if a few outlier HLA-B alleles were neglected, among the other HLA-B alleles a higher resistance against HIV was conferred by the generalists than by the specialists. The results suggest that the generalists are better in conferring protection, unless the specialists match well with the relevant infectious pathogen. That makes sense because the generalists present a wider variety of peptides, whereas the specialist more efficiently present a more restricted set of peptides.

The Generalists-Specialists model by Prof. Kaufman is still not fully understood, but it provides an additional viewpoint for explaining MHC polymorphism. For this new model—which does not deny but has incorporated the traditional explanation model saying that different MHC-I alleles differ in their efficiency to bind immunodominant epitopes of various pathogens—there appears to be more empirical evidence than for the traditional model alone. For MHC experts, Prof. Kaufman’s new model is an exciting development in MHC research, which is why several MHC experts joined this seminar.

The Generalists versus Specialists model

This section is divided in paragraphs explaining:

  1. The structure of MHC molecules
  2. The chicken Mhc region has fewer MHC genes than in humans; chicken MHC-I allelic variation appears to be amplified by allelic variation in linked genes for tapasin, TAP1, and TAP2
  3. Different MHC alleles have different peptide binding grooves and therefore bind sets of peptides with different consensus motifs
  4. Some MHC-I alleles are lowly expressed, promiscuous “generalists” whereas other MHC-I alleles are highly expressed, selective “specialists”
  5. In chicken, the generalist MHC-I alleles give a higher protection against diseases common in the poultry industry; this explains why there has been a selection for these alleles
  6. In HIV disease in humans, a few outlier HLA-B alleles give a clearly higher or lower protection, but among the other human MHC-I alleles on average the generalists give a higher protection

1. The structure of MHC molecules

The structure of a human MHC-I allele (HLA-A2) as described already by Pamela Bjorkman in 1987 (Fig. 1) is similar to both MHC-I and MHC-II structures that since then have been elucidated (Fig. 2A). However, MHC-I complexes include a long heavy chain and a single domain β2-m (beta-2 microglobulin) molecule, whereas MHC-II complexes include an alpha and beta chain of similar size (Fig. 2A). In the groove formed by the membrane-distal domains (the a1 and a2 domains in MHC-I; the a1 and b1 domains in MHC-II) a peptide ligand is bound (the peptide ligands are in red in Fig. 2A). All classes of jawed vertebrate species have both MHC-I and MHC-II (Fig. 2B).

Figure 1. This figure is a slight modification of a slide used in Prof. Kaufman’s presentation. The structure figures are HLA-A2 in which the bound peptides are not shown and are from Bjorkman et al. 1987.
Figure 2. This figure is from Wu et al. 2021. (A) Representative pMHC-I and pMHC-II structures of different species. PDB accession numbers are between brackets. The top row shows side views, with the MHC molecules in cartoon format and the peptide in spheres format. The bottom row shows top views, in surface format. (B) Cladogram showing the phylogeny of the species shown in (A). Because in Agnatha no MHC genes are found and all classes of Gnathostomata possess both classical MHC-I and MHC-II, it can be assumed that MHC genes emerged and differentiated into classical MHC-I and MHC-II in the period highlighted by the blue bar. MYA, million years ago.

2. The chicken Mhc region has fewer MHC genes than in humans; chicken MHC-I allelic variation appears to be amplified by allelic variation in linked genes for tapasin, TAP1, and TAP2

As shown in Fig. 3, compared to human, in chicken the Mhc is highly condensed, with fewer MHC-I and MHC-II genes. Of the two MHC-I genes in chicken (named BF1 and BF2), only one (BF2) is highly expressed. In the chicken Mhc, genes for tapasin, TAP1, and TAP2, which all three play a role in peptide loading of MHC-I, are also polymorphic and closely linked with the MHC-I genes. The tapasin, TAP1, and TAP2 alleles appear to form functional haplotypes together with the BF2 allele which they help provide with matching peptides (Figs. 4 and 5; for discussion of an effect of polymorphic tapasin in chicken see van Hateren et al. 2013). Having only one dominantly expressed MHC-I allele, plus having the MHC-I allelic groove variation amplified by polymorphism in peptide loading pathway molecules, may explain why chicken Mhc variation is more strongly linked with differences in infectious disease resistance than in humans.

Figure 3. This figure is a slight modification of a slide used in Prof. Kaufman’s presentation. The condensed nature of the chicken Mhc was already described by Prof. Kaufman in 1999. The genomic region depictions are from Kaufman 2018. Wallny et al. 2006 showed how chicken have only one dominant MHC-I allele.
Figure 4. A nice drawing about the MHC-I peptide presentation pathway by the Neefjes group (Rock et al. 2016). Intracellular proteins are degraded by the proteasome (or immunoproteasome, which chicken don’t have) and some of these peptides are selected by the TAP1/TAP2 complex for transport into the endoplasmic reticulum. There, with help of a chaperone (tapasin or TAPBPR) that helps stabilize the MHC-I/β2m complex, peptides are loaded onto the MHC-I molecules. The MHC-I/β2m/peptide complexes are then transported to the cell surface for screening by CD8+ cytotoxic T lymphocytes (CTL).
Figure 5. This figure was used as a slide in Prof. Kaufman’s presentation. It schematically shows how electrical charges of the peptide transportation pore formed by polymorphic TAP1 and TAP2 can affect the selection of transported peptides by favoring certain charges. For details on this selection see Kaufman et al. 1999 and Walker et al. 2011.

3. Different MHC alleles have different peptide binding grooves and therefore bind sets of peptides with different consensus motifs

Although there is some selection by the proteasome and the TAP1/TAP2 transporter complex for the peptides available for loading onto MHC-I molecules, there is a still a large pool of different peptides available for MHC-I binding. The grooves of MHC-I molecules are (in most cases) closed causing a restriction of peptide length between 8 and 14 amino acids (aa); most commonly, they bind peptides of 9 aa. The grooves of the different MHC-I alleles differ in the properties of their pockets in which the peptide ligand sidechains can be inserted (example in Fig. 6), and they tend to be (but are not always) most selective for the amino acid residues at the 2nd and C-terminal position of the peptide ligand (strongly selected residues are called “anchor” residues). For many other positions in the peptide ligand, often hardly any preferences are observed, which in traditional experiments that performed pool sequencing (by Edman degradation) on the mix of peptides presented by an MHC-I allele had these residues often denoted by an X as representing any amino acid. The anchor residues, and even the anchor positions, can differ between alleles of the same MHC-I gene (e.g., compare HLA-B8 and HLA-B37 in Fig. 7). In later years, improved sensitivity of the Edman degradation analysis revealed that also at non-anchor positions of the peptide ligand the MHC-I alleles can have slight preferences for certain amino acids (Fig. 8). Moreover, in contrast to MHC-II, in the grooves of most MHC-I alleles the peptide ligand backbone is not in a relaxed fashion (it is slightly contorted) causing a larger effect of sidechains on each other; these inter-sidechain interactions also limit the repertoire of peptides that can be bound with sufficient affinity, but are not captured by the techniques used for results as shown in Figs. 7 and 8.

Figure 6. An example of different allelic molecules encoded from the same gene (in this case human HLA-B) exhibiting substantial differences in the pockets A-to-F for binding a peptide ligand. The figures show the electrostatic potential mapped to the surface of the structures of HLA-B*57:01, HLA-B*57:03 and HLA-B*58:01 respectively (red—electronegative, blue—electropositive). The α1 and α2 helices and the positions of peptide-binding pockets A–F are shown. The figure is from Illing et al 2018.
Figure 7. This figure was used as a slide in Prof. Kaufman’s presentation. Anchor residues of peptides bound to chicken, mouse, and human MHC-I alleles (for chicken it is more correct to say “haplotypes” as an effect of the minor BF1 locus can’t be excluded) as determined by Edman degradation pool sequencing in the early 1990s. At the top, a schematic sideview of a peptide bound into an MHC-I groove is shown: the peptide conformation can differ per peptide and per MHC-I allele, but commonly the sidechain of the N-terminal residue (“P1”) of the peptide points upwards, the sidechain of the P2 residue inserts sideways/downwards into a pocket (the B pocket), and the C-terminal residue sidechain inserts downwards into the F pocket (for pocket locations see Fig. 6). This figure is modified from Kaufman et al. 1995.
Figure 8. Examples of consensus motifs of MHC-I loaded peptides as determined by more sensitive than those used for Fig. 7 (so that also weak preferences could be detected). In six different alleles (more correctly “haplotypes”) of chicken MHC-I different preferences for certain residues are found at different positions (“Anchor” means a very strong preference). For the Specialists B4, B12, and B15 data, this figure is slightly modified from Wallny et al. 2006. For the Generalists B2, B14, and 21 data, this figure is slightly modified from Chappell et al. 2015.

4. Some MHC-I alleles are lowly expressed, promiscuous “generalists” whereas other MHC-I alleles are highly expressed, selective “specialists”

In chicken, long before the peptide binding features of the dominant MHC-I molecules that they encoded were known, Mhc haplotypes were already found strongly associated with resistance against Marek’s disease virus (MDV; see the paragraph below). The MHC-I expression on the surface of erythrocytes of chicken with the resistance conferring haplotype B21 was approximately 10-fold lower than on erythrocytes of the (relative) MDV-sensitivity conferring haplotypes B4, B12, B15, and B19 (Fig. 9). When spleen cells of chicken with the MDV-resistant haplotypes B2, B14, and B21 were compared with those of the sensitive haplotypes B4, B12, B15, and B19, also a several-fold lower MHC-I surface expression was observed (Fig. 10).

It turned out that at the MHC-I synthesis levels there were no significant differences between the MDV-resistant and -sensitive haplotypes, but in case of the resistant haplotypes the MHC-I/β2-m/peptide complexes were less stable as indicated by their thermostability in Fig. 11 for the resistant haplotypes B2, B14, and B21 versus the sensitive haplotypes B12, B15, and B19 (Tregaskes et al. 2016). Outside MHC-I/β2-m/peptide complexes, MHC-I molecules are rapidly degraded, which can explain how a lower complex stability results in lower amounts of MHC-I protein.

As already shown in Fig. 8, the peptides bound to chicken generalist MHC-I alleles display far less pronounced anchor motifs than found for the chicken specialist MHC-I alleles. This suggests that they can bind a larger repertoire of peptides. Supportive evidence for that assumption was provided by showing that the number of different peptides loaded onto MHC-I in a chicken cell line with the B21 haplotype was three-fold higher than in a chicken cell line with the B19 haplotype, despite that the MHC-I surface expression was 2-fold lower (Fig. 12).

At the structural level, the width of the peptide repertoire appears to correlate with the width of the peptide binding groove, with the wide groove of the B21 MHC-I allele (BF2*2101) being unusually wide and allowing many peptide conformations (Figs. 13 and 14; see Chappell et al. 2015 for more alleles).

In summary, the chicken MHC-I alleles with the more restrictive anchor motifs (Fig. 8) seem to have a narrower peptide repertoire indeed (Fig. 12), in correlation with a narrower groove (Fig. 13), and they are expressed at higher levels at the cell surface (Figs. 9 and 10) which can be explained by their higher stability (Fig. 11).

Figure 9. This figure is a modification of a slide used in Prof. Kaufman’s presentation. MHC-I/β2-m expression on generalist B21 erythrocytes is ~10-fold lower than on   Flow cytometry of erythrocytes from 5 strains of chicken stained with mAbs to chicken class I, to chicken β2-m, or to antigens not present in chicken (negative control). Left, dot-plot with fluorescence of anti-chicken class I on the abscissa with fluorescence of negative control antibody compared to anti-class I and anti-β2-m antibodies. Relative fluorescence was measured on a log scale using a FACscan. This figure is modified from Kaufman et al. 1995.
Figure 10. On the surface of chicken spleen cells, the amounts of specialist MHC-I alleles are also higher than that of generalists. Cell surface expression levels of class I molecules vary markedly between chicken haplotypes, as determined by a quantitative flow cytometric assay. Spleen cells from various inbred experimental chicken lines (with MHC haplotypes indicated) were stained with the monoclonal antibody F21-2 against chicken major histocompatibility complex (MHC) class I heavy chain and the specific antigen binding capacity (SABC, which reflects number of epitopes per cell calculated in reference to specific antibody-binding calibration beads). Results are means of triplicate stains, with error bars indicating standard deviation. This figure is from Chappell et al. 2015.
Figure 11. High-expressing haplotypes produce class I molecules that are more thermostable than low-expressing haplotypes. Detergent lysates of chicken erythrocytes or blood PBLs were incubated at the indicated temperatures, IPs with the mAb to β2-m were analyzed by SDS gel electrophoresis, followed by WB using the mAb to HC, and the amount of HC was quantified by densitometry after fluorography. Results from four experiments were normalized and averaged, with SEM indicated by error bars. This figure is from Tregaskes et al. 2016.

Figure 12. There is an inverse correlation between the cell surface expression levels of class I molecules and the variety of peptides isolated from class I molecules. (A) The chicken liver lymphoma B19 cell line 265L and the chicken transformed spleen cell B21 cell line AVOL-1 were analyzed by flow cytometry by staining with the mAb F21-2 to chicken class I molecules. AVOL-1 had slightly more autofluorescence, so the settings on the FACScan were adjusted so that the mean fluorescence intensity of the isotype control sample was the same as for 265L. The histogram shows the fluorescence intensity in the FL1 channel on the x-axis and the number of events on the y-axis. (B) In the same flow cytometry experiment, the calibration beads from the QIFIKIT were stained separately with the secondary antibody for calibration curves to calculate the SABC, which reflects the absolute numbers of epitopes on the cell surface. As a separate experiment, the class I molecules were isolated from each cell line by affinity chromatography with F21-2 and analyzed by LC-MS/MS. Table shows the SABC and the number of different peptides found for each cell line. This figure is from Chappell et al. 2015.

Figure 13. This figure was used as a slide in Prof. Kaufman’s presentation. The peptide binding groove of the chicken specialist MHC-I allele BF2*0401 (of the B4 haplotype) is narrow whereas that of the chicken generalist MHC-I allele BF2*2101 (of the B21 haplotype) is unusually wide. The data are from Koch et al. 2007 and Zhang et al. 2012.

Figure 14. This figure was used as a slide in Prof. Kaufman’s presentation. In the unusually wide groove of the chicken generalist MHC-I allele BF2*2101 (of the B21 haplotype) the peptides can be bound in a variety of conformations, which may explain the wide repertoire of the peptides bound by this allele. The data were also shown in Chappell et al. 2015.

5. In chicken, the generalist MHC-I alleles give a higher protection against diseases common in the poultry industry; this explains why there has been a selection for these alleles

In chicken, MHC-I alleles were readily found associated with resistance against Marek’s disease virus (MDV), an oncogenic herpesvirus that causes T cell lymphoma (Figs. 15 and 16). Kaufman et al. 2015 stated that in many studies over decades, a hierarchy of MHC haplotypes was found, with B21 generally conferring the greatest resistance and B19 generally conferring the greatest susceptibility: B21 > B2 > B6 > B14 > B4 > B15 > B12 > B19. This order was also generally consistent with the resistant haplotypes encoding MHC-I alleles that are less stable and have broader peptide repertoires and lower surface expression (see above) and also TAP molecules that are less restrictive in their selection of transported peptides (summarized in Fig. 17).

Interestingly, some of the same haplotypes conferring resistance and susceptibility to Marek’s disease do the same upon a very different infection, namely avian influenza virus (AIV). MHC-I haplotype investigation among 390 dead and 340 surviving Thai indigenous chickens from smallholder farms after avian influenza virus disease outbreaks revealed that B21 conferred a superior resistance compared to B4, B12, B15, and B19 (Fig. 17; Boonyanuwat et al. 2006).

Associations in chicken between Mhc and differences in disease resistance were also found for other infectious diseases besides MDV and AIV. In his 2018 article, Prof. Kaufman used a nice figure (here Fig. 19) to summarize the above-mentioned MDV study (Fig. 16) and the above-mentioned AIV study (Fig. 18) together with studies on Rous sarcoma virus (RSV; a retrovirus inducing sarcoma) and infectious bronchitis virus (IBV; a coronavirus) (Fig. 19). In each case, the Mhc haplotypes with the generalist MHC-I alleles (rainbow color in Fig. 19) conferred the highest resistance.

Thus, against a number of diseases common in the poultry industry, generalist MHC-I alleles appear to confer a better protection. This seems to explain why those generalist alleles became dominant in the chicken breeds kept in the modern poultry industry whereas in African village chicken no such trend and a much larger variety of MHC-I alleles is observed (unpublished data by the Kaufman group).

Figure 15. The B21 Mhc proper region is associated with protection against Marek’s disease. In this study, B19 homozygous sires were crossed with a dam heterozygous for B19 or B21 and a recombined haplotype with the Mhc proper (BF and BL) region of B21 and the BG region of B19. The Mhc proper region of B21 appeared to confer its resistance to Marek’s disease as compared to B19 with 61% of the B19/B19 offspring and only 12% of the B(F21-G19)/B19 offspring getting the disease. The table is from Briles et al. 1983.
Figure 16. This figure is a slight modification of a slide used in Prof. Kaufman’s presentation. The graph with its legend below is from Simonsen 1987.
Figure 17. This figure was used as a slide in Prof. Kaufman’s presentation. A hierarchy of chicken class I molecules that vary in a suite of properties. Major histocompatibility complex (MHC) haplotypes have historically had a very strong association with resistance and susceptibility to Marek’s disease, which correlates with various properties of the dominantly expressed class I molecules and TAPs. TAP, transporter associated with antigen presentation. This summary is from Kaufman et al. 2015.
Figure 18. This figure was used as a slide in Prof. Kaufman’s presentation. In chicken on small farms in Thailand, analysis of dead and surviving chicken after an avian influenza outbreak revealed superior resistance conferred by the B21 haplotype and sensitivity by the B4 (also known as “B13”) haplotype. The data are from Boonyanuwat et al. 2006 and are discussed by Tregaskes and Kaufman 2021.
Figure 19. Chicken Mhc haplotypes encoding generalist MHC-I (Rainbow) can confer protection against a variety of viral infections under experimental and field conditions, whereas Mhc haplotypes encoding specialist MHC-I (Red) generally confer susceptibility. (A) Percentage of Mhc genotypes in a flock before and after experimental infection with Marek’s disease virus (MDV), with the B2 and B21 haplotypes conferring protection (Simonsen 1987). (B) Percentage of Rous sarcoma virus (RSV) strains that progress to give lethal tumors after experimental infection, with the B6 haplotype conferring survival (McBride et al. 1981). (C) Percentage survival after natural infection with avian influenza virus (AIV) under field conditions in rural Thailand, with the presence of a single promiscuous haplotype conferring protection, except in one combination (B2/B13) for reasons that are not understood Boonyanuwat et al. 2006. (D) Percentage of chickens ill from infectious bronchitis virus (IBV) on day 10 after experimental infection, with the B2 haplotype conferring protection (Banat et al. 2013). This figure is from Kaufman 2018.

6.  In HIV disease in humans, a few outlier HLA-B alleles give a clearly higher or lower protection, but among the other human MHC-I alleles on average the generalists give a higher protection

Among people infected with HIV, those with certain MHC-I alleles such as HLA-B*5701 or HLA-B*2705 have a higher disease resistance (Kaslow et al. 1996; Migueles et al. 2000; Goulder and Walker 2012).

Initially, there were several studies that found that (1) also in humans, specialist and generalist MHC-I alleles exist with the specialists presenting a narrower peptide repertoire and being expressed at higher levels at the cell surface, but that, (2) in contrast to the observations in chicken, it was the specialist MHC-I alleles that conferred a higher resistance against HIV disease (AIDS):

Košmrlj et al. 2010 found narrower peptide binding repertoires for the HIV-resistance conferring alleles HLA-B*5701 and HLA-B*2705 than for the HIV-sensitivity conferring alleles HLA-B*0702 and HLA-B*3501 (Fig. 20A). The group of Prof. Kaufman then found that the HIV-resistance alleles HLA-B*5701 and HLA-B*2705 were expressed at higher levels at the cell surface than HLA-B*0702 and HLA-B*3501 (Fig. 20B). Thus, the generalist MHC-I alleles (HLA-B*0702 and -*3501) are expressed at lower levels than the specialist MHC-I alleles (HLA-B*5701 and -*2705) as described above for chicken, but in contrast to the above-described situations in chicken, among these HLA-B alleles it is the specialists that confer a higher protection against HIV.

At a note, the explanation by Košmrlj et al. (2010) for explaining the differences in HIV resistance conferred by generalist versus specialist MHC-I alleles was entirely different from the one proposed by Prof. Kaufman (for the latter, see the next paragraph). (Only for the experts among the readers here: The model by Košmrlj et al. proposed that the MHC-I specialist alleles conferred a higher resistance because their narrower peptide repertoire leads to a lesser negative selection of T lymphocytes in the thymus and therefore to an increased T lymphocyte cross-reactivity against mutating viral peptides).

Rizvi et al. 2014 also thought that besides the matching of the groove for immunodominant epitopes other characteristics of the HLA-B molecules might play a role in HIV resistance. Therefore, they investigated the level of surface expression of various HLA-B alleles after their transfection into a tapasin-lacking cell line (Fig. 21). Tapasin is a chaperone that in the endoplasmic reticulum holds the MHC-I/β2-m complex and inserts a loop into the MHC-I peptide binding groove (Jiang et al. 2017; Thomas and Tampé 2017); by doing this, it allows only peptides to bind that have a high affinity for the groove because they have to outcompete tapasin (this is part of a process called “peptide editing”). The higher the dependence of an MHC-I allele on tapasin, the higher the affinity of a peptide has to be for being able to outcompete tapasin. Rizvi et al. found that the two above-mentioned HLA-B alleles with a narrow peptide repertoire that confer resistance to HIV, HLA-B*5701 and -*2705, exhibited a lower expression in their system (revealing a higher dependence on tapasin) than the two above-mentioned HLA-B alleles with a broad peptide repertoire that confer sensitivity to HIV, HLA-B*0701 and -*3501 (Fig. 21). These data are opposite from the data obtained in cells that do express tapasin (Fig. 20B). Thus, what seems to happen under normal conditions (in cells where tapasin is present) is that the generalists “do not really want to wait” for tapasin-mediated peptide editing and move to the cell surface with a wide variety of low affinity peptides and therefore there are less stable than the specialist alleles that waited in the endoplasmic reticulum to be loaded with more optimal peptides. (At a note, although Rizvi et al. 2014 speculated on stable and unstable binding of peptides, it was Prof. Kaufman who was the first to interpret their data in regard to the width of peptide repertoires and the generalists-specialists model; the summary in this blog-post is written from Prof. Kaufman’s point of view)

Rizvi et al. 2014 also found that after expression of the HLA-B alleles and β2-m in E.coli, followed by their denaturation and renaturation, in the absence of peptides the generalist HLA-B alleles were better in forming properly formed complexes with only β2-m (and no peptide) than the specialists (not shown here). Thus, the dependency of the specialist MHC-I alleles on tapasin and a high affinity peptide appears to be correlated with these alleles being more unstable compared to the generalists if only bound to β2-m.

It was puzzling how in chicken the generalist MHC-I alleles were more readily found to be associated with a higher resistance against infectious diseases, whereas in human HIV disease the specialists performed better. Then, just one year ago, Bashirova et al. 2020 published an article titled “HLA tapasin independence: broader peptide repertoire and HIV control” that seems able to unify both findings.

Bashirova et al. 2020 found after measuring the tapasin dependence levels of nearly 100 HLA variants that the level of tapasin dependence negatively correlates with the peptide ligand repertoire width, which is consistent with the findings in the studies mentioned above. However, they also found that if they neglected three of the above-mentioned HLA-B alleles that confer an unusually high resistance or susceptibility (HLA-B*5701, -*2705, and B*3501), on average the generalist HLA-B alleles were associated with a slower progression of HIV disease. Thus, after all, also in humans the generalists seem to confer a higher protection, unless a specialist allele makes a good match with the particular disease. Some of the results by Bashirova et al. 2020 are shown in Fig. 22.

Fig. 22A compares the levels of cell surface expression after transfection of various HLA-A, -B, and -C alleles to the human tapasin-deficient B cell line “.220” versus tapasin-reconstituted .220 cells. The figure shows that the ratio could be 100-fold higher in the latter for some alleles, proving their tapasin dependence, whereas for other alleles there hardly was a difference. In line with the above-mentioned reports, the specialist HLA-B alleles *B5701 and *B2705 showed a higher tapasin-dependence than the generalist HLA-B allele *3501 (Fig. 22A). For analysis if the specialist or generalist HLA alleles conferred higher resistance to HIV, Bashirova et al. removed three HLA-B alleles that are known to confer especially high protection (two) or susceptibility (one) from the equation, namely HLA-B*57, -*27, and B*35-Px alleles. Doing this, they were able to pick up a significant signal indicating that the generalist MHC-I alleles conferred a slower progression of AIDS and were associated with lower viral loads (Figs. 22B and -C; see Table 2 in the Bashirova et al. paper for the separate analysis of the sets of HLA-A, -B, and -C alleles).

Figure 20. This figure is a modification of a slide used by Prof. Kaufman. (A) Human HLA-B alleles with a wider peptide repertoire confer a higher sensitivity to HIV. This figure was slightly modified from Fig. 3 in Košmrlj et al 2010; for an explanation see the figure legend in that publication. (B) Human class I molecules show an inverse correlation between cell surface expression level and peptide binding promiscuity. The Y-axes show the levels of specific antibody binding capacity (SABC) for the mAb Tu149 for ex vivo lymphoctyes and monocytes. Each point represents the sample from a particular donor; bars indicate the mean for each HLA-B allele. The graphs are from Chappell et al. 2015.
Figure 21. This figure is a modification of a slide used by Prof. Kaufman and shows the expression levels of various transfected HLA-B alleles in a tapasin-deficient melanoma cell line M553. The y-axes show cell surface expression ratios (MFI ratios) relative to mock transfected control cells. M553 cells that were infected with a virus lacking MHC class I (vector). MHC class I surface expression was analyzed by flow cytometry using the W6/32 Ab. Statistical analyses were done using a one-way ANOVA test, followed by a Tukey’s multiple comparisons procedure for all pairwise differences of means. Significant differences are indicated (with an asterisk) on the graph (p < 0.05). Data represent averaged MFI ratios derived from 10–15 independent flow cytometric analyses from five independent infections (infections 1–5) of M553 cells. (HLA-B alleles can be divided into the serotypes HLA-Bw4 and HLA-Bw6.) The data and graph figure are from Rizvi et al. 2014.
Figure 22. This figure is a modification of a slide used by Prof. Kaufman. (A) Tapasin influences HLA class I surface expression. HLA expression levels in .220 cells without and with tapasin reconstitution were measured by flow cytometry. MFI, median fluorescent intensity. Tapasin depence (TD) for each HLA allotype, defined as the ratio of MFI of tapasin-positive over tapasin-negative cells, is shown in log10 scale. Error bars correspond to SDs calculated from multiple measurements. (B and C) Tapasin dependence impacts HIV-1 disease as revealed if data for patients with HLA-B*57, -*27, and B*35-Px alleles were excluded from the analysis. (B) Kaplan–Meyer curves for time to AIDS-1987 are shown for a cohort of ART-naïve HIV-1 seroconverters equally divided based on their global TD (high, medium, and low, n = 318 in each group). (C) Mean log10VL (Virus Load) plotted against log10TD level is shown for a cohort of ART-naïve HIV-1 patients. Each dot represents the mean log10VL of patient groups divided into increasing bins of 0.1 log10TD. Estimate (est), and P value were derived by regression analysis adjusted by race. The individual figures are modified from Bashirova et al. 2020.

7. Speculations as to why the generalists and specialists confer differences in disease resistance

To explain why there are both specialist and generalist MHC alleles, Prof. Kaufman presented a model (Kaufman 2018) which is shown in Fig. 23 and Fig. 24.

Fig. 23 is a schematic comparison between the MHC-I situation in humans and chicken. In chicken, the variation between specialist and generalist MHC-I alleles is more extreme than in humans. For example, compared to mammals, the groove differences between chicken BF*0401 and BF*2101 are very extreme and MHC-I molecules with such a wide groove as chicken BF*2101 have, to my knowledge, not been described so far in mammals (Figs. 7, 8, 13, 14). Combined with the fact that in chicken, MHC-I allelic differences cannot be compensated by MHC-I molecules encoded by other loci (in contrast to HLA-A, -B, and -C in humans), this probably can explain why the generalist-specialist divide in MHC-I features was first observed in chicken. The reason that chicken has some “extremely generalist” alleles such as BF*2101 may be to present a large enough repertoire of peptides despite the absence of other MHC-I genes.

But human MHC-I alleles can also be distinguished along a generalist-specialist axis, and this may be true in all species with MHC. So why are there both types of alleles?

The emerging picture is that “generally” (against the majority of infectious diseases) the “generalists” confer a higher protection but that in “special” cases (for diseases where they efficiently present a relevant immunodominant epitope) the “specialists” can confer a higher protection. At least superficially, this makes sense as the generalists present a wider variety of peptides (thus have a better chance for presenting some epitopes of any pathogen) whereas the specialists present more of the same peptide (which provokes a stronger immune response if this is an immunodominant epitope).

In Fig. 24, Prof. Kaufman proposes that under standard conditions it is the generalist alleles that confer the best protection and therefore are dominant in a population of species. However, under special conditions such as the exposure to a new or very virulent pathogen, the specialists may confer a better protection and increase in population frequency.

The generalist-specialist model is only being developed and still needs integration of other functions in which MHC is involved besides resistance against infectious diseases, such as autoimmunity, cancer, reproduction, and brain development (Tregaskes and Kaufman 2021). Eventually, the critical biology of MHC polymorphism may turn out to be even more complicated than currently speculated on, but for now the generalist-specialist hypothesis that evolved from work by Prof. Kaufman and others and was first presented as a coherent model by Prof. Kaufman, feels like a natural enrichment and a breath of fresh air in the research of MHC polymorphism.

It will be very exciting to follow the further developments of this model.

Figure 23. This figure is a slight modification of a slide used in the presentation by Prof. Kaufman. Compared with mammals, the Mhc of chickens has strong genetic associations with resistance and susceptibility to infectious diseases. (A) A multigene family in the human Mhc can encode multiple fastidious class I molecules each of which has a chance to find a protective peptide. Altogether the typical human Mhc haplotype confers more or less resistance to most pathogens, a situation that reads out as a weak genetic association (since there is not much difference between haplotypes). (B) By contrast, the single dominantly expressed class I molecule encoded by the chicken MHC can have a fastidious peptide motif that may or may not find a protective peptide from any given pathogen, a situation that reads out as strong genetic associations (since there can be enormous differences between haplotypes). (C) However, the single dominantly expressed class I molecule encoded by the chicken Mhc can have a promiscuous peptide motif capable of binding a wide variety of peptides (much like the multigene family of human class I molecules acting together). Comparison of two promiscuous alleles may read out as a weak genetic association (since there is not much difference between them) but comparison of a fastidious allele with a promiscuous allele in chickens may give strong genetic associations. This figure and its legend are from Kaufman et al. 2018.
Figure 24. This figure was used as a slide in the presentation by Prof. Kaufman. A model Illustrates the shift in gene frequencies from a few predominant generalist MHC alleles on selection by new and/or particularly virulent pathogens. The diameter of each circle indicates the frequency of a particular MHC allele in a population before and after selection by a pathogen. The rainbow colors indicate promiscuous molecules that act as generalists, conferring protection against most pathogens including those regularly found in the environment. The single colors indicate fastidious molecules encoded by genes that arise by mutation and are present at low frequency but with the possibility of presenting a protective peptide from a particular pathogen. Scenarios for three different pathogens are shown: one of the generalist molecules confers protection against the first pathogen (top); one of the specialist molecules confers protection against the second pathogen (middle); and another of the specialist molecules confers protection against the third pathogen (bottom). This figure and its legend are from Kaufman et al. 2018.
 

Questions by the audience

(These are the questions that I remember [I forgot a few]. They are modified for clarity, and Prof. Kaufman helped intensively with the writing of the answers)

Question 1.

You worked in two places which are very famous to immunologists. Namely, the Strominger group at Harvard University and the Basel Institute for Immunology. The famous work by Pamela Bjorkman (elucidating the first structure of an MHC molecule), who was a member of the Strominger group, was done in intensive collaboration with the Don Wiley group of the same department. In Basel, Susumu Tonegawa, who as a geneticist did not know anything about immunology when he entered the Basel institute but would get the Nobel prize for his work on the principle of antibody generation, said that he learned immunology from his Basel colleagues. It is clear these two institutes had many smart people and were well-funded. But how did the communication happen, as that communication was apparently very productive. Were there very regular lab meetings, or did you guys have your main discussions in the canteen and in bars?

Answer:

To be honest, I was a bit naïve when I was young, expecting great science done to be done by great scientists who worked together to figure out some objective truth. So, when I arrived at Harvard as a PhD student, I was shocked by the people there and the way things worked. It was as though the scientists were in competition for success at any cost, both within and between groups. Of course, there were friendships between the PhD students which transcended this sense of competition, and the discussions between Pamela Bjorkman and myself that led to the joint project between the labs of Jack Strominger and Don Wiley was an example of that.

In Basel the situation was very different, with some 50 scientists of more-or-less equal status, each with “a bench, a budget, a technician and total freedom”, as was the saying there. At least when I arrived, there were only a few groups (and even they were small), and often people came up with ideas by talking together in the cafeteria (which was located in the center of the building) or in cafes and “Stuebli” (bars) around town, and then worked together to get things done. Certainly this collegiality was most important for my own work, which involved many people in and out of Institute.

Question 2.

Which alleles are evolutionary older, the generalists or the specialists?

Answer:

I guess we aren’t in the position to know this yet, since we are looking at species which have had a large collection of MHC alleles for a long time. Personally, I like the idea that generalists came first since they are less restricted, but the ways that many peptides are bound by the three promiscuous chicken class I molecules that we understand best are very different, as though they arose independently. From work with human class I alleles, it looks like single amino acid differences between alleles can strongly affect their dependence on tapasin and thus their peptide repertoires, which suggests that alleles can change in either direction.

(A comment that Prof. Kaufman made elsewhere which is somewhat related to this question:) When comparing MHC-I and MHC-II, and assuming that MHC-II is the older type, the evolutionary emergence of MHC-I resulted in a more restricted selection of peptides because of the closed groove in MHC-I versus the open groove in MHC-II.

Question 3.

Extension of peptides from the F pocket as you found for several chicken MHC-I alleles looks like the situation in MHC-II. Can you elaborate on that?

Answer:

It is true that long ago I noticed that a so-called “invariant tyrosine” in mammalian class I molecules that is important for binding the carboxy-terminus of the peptide and closing off the groove is replaced by an arginine in chickens and all non-mammalian vertebrates examined, and that this arginine is in common with human class II molecules. However, it was only years later, in collaboration with some crystallographer friends that we showed that a chicken class I allele would allow the bound peptide to hang out of the groove at the C-terminal end. Most of the alleles that we had studied up to that time did this rarely, but as we looked at alleles in commercial chickens, we found various alleles for which many or most of the peptides hang out of the groove. The mystery is why many chicken alleles do not seem to have peptides hanging out of the groove, since they all have the arginine rather than a tyrosine as in mammals. One possibility is that this restriction occurs at the level of different TAP alleles pumping peptides from the cytoplasm to the lumen of the endoplasmic reticulum.

Question 4.

In the past, we found that in Parkinson’s disease β2-m (the soluble single domain component of the heterodimer MHC-I structures) is upregulated in dopaminergic striatal regions [JMD: see Mogi et al. 1995]. What is your interpretation of that?

Answer:

I don’t have any special insights for this interesting finding. There are papers that suggest the MHC (and by extension classical MHC alleles) has effects on human brain development. Also, β2-m is not only a heterodimer partner of the classical MHC-I molecules (the functions of which were the topic of my presentation), but also a number of so-called “nonclassical” MHC-I molecules which derived from gene duplications of the classical ones and have a wide variety of functions. In mice, some of these nonclassical class I molecules from the M region are known to be expressed in and affect the development of the central nervous system. Although it wasn’t noticed at first, β2-m knockout mice have affected behavior and brain morphology [JMD: e.g., see Huh et al. 2000; Loconto et al. 2003; and Smith et al. 2015).

Question 5.

You mention “high” versus “low” expressing alleles, but what level of differences are we talking about?

Answer:

In the erythrocyte example [in chicken, erythrocytes are nucleated cells] that I showed, the difference was up to 10-fold. However, in other cell types the differences are up to about 3-4 fold. We found that MHC-I protein stability (which depends on the stability of the MHC-I/β2-m/peptide complex) is lower in the generalist MHC-I alleles and assume that this lower stability is an important reason for their lower amounts on the cell surface. We found little or no synthesis of protein in erythrocytes from chicken blood, and so all the class I molecules present on the cell surface have been there for months. Comparing the cell surface level of class I molecules between leukocytes and erythrocytes, we found only a two-fold difference for typical fastidious (“high expressing”) class I molecules but a ten-fold difference for typical promiscuous (“low expressing”) molecules, consistent with the promiscuous molecules falling apart on the erythrocytes but being constantly replaced on the surface of leukocytes.

Question 6.

How common is it that peptides protrude out of the groove?

Answer:

It is more common than originally thought, both in chickens and in mammals. Some researchers looking at human class I peptides by immunopeptidomics (mass spectroscopy) believe that peptides frequently hang out of the groove, but the actual examples are a few (some at the N-terminus and some at the C-terminus). For the C-terminal protrusions, there is significant movement of the alpha helices to accommodate the overhang. In chickens, there is no deformation of the groove to allow the overhangs, but as just mentioned, many chicken class I alleles do not show a lot of overhanging peptides. For both humans and chickens, overhangs may be more common among MHC-I/β2-m/peptide complexes during their early stage of formation in the endoplasmic reticulum, where bound peptides can still be exchanged for others (“peptide editing”) than in the complexes presented at the cell surface.

Question 7.

Is it of concern that commercial chicken have limited MHC diversity?

Answer:

Initially, we were enormously surprised at the relatively low diversity of MHC haplotypes found in most commercial chickens. Once we realized that these haplotypes expressed class I alleles that are mostly promiscuous generalists, it became clear that the breeding companies had enriched for those haplotypes that conferred disease resistance (and we have some limited historical evidence that this is what happened). Typically, commercial chickens live under controlled conditions with as much vaccination and as much biosecurity as can be afforded, so they may not experience the same risk of infection as they might in the wild; if there is a disease outbreak, the typical response is to cull all the birds. On the other hand, chicken kept under semi-wild conditions in African villages do not display such restricted set of MHC haplotypes, which suggests that those other haplotypes confer resistance to other types of diseases. If those diseases would enter chicken farms, the chicken might have little resistance.

Question 8.

Is there a difference between the T cells that interact with “generalist” versus “specialist” MHC-I alleles?

Answer:

We have begun to look at T cells in a limited way, collaborating with our friends to examine T cell receptor beta repertoire in (pre-)neoplastic lesions formed by Marek’s disease virus. So far, we see that the chicken line with a promiscuous class I allele has a few large clones of CD8 T cells that infiltrate the (pre-)neoplastic lesions, whereas the line with a fastidious allele has a wide range of T cells with no large clones. However, the large clones of T cells in different individuals with the same promiscuous class I allele have different T cell receptor sequences (that is, no “public clones”). Another approach we are trying is to make peptide/MHC tetramers (multimers) to look at ex vivo T cells by flow cytometry, using the pathogen peptides that we have found by immunopeptidomics. Obviously, there is still a lot of work to do done to understand T cell responses in any detail.

Question 9.

Do I understand correctly that, in chicken, MHC-I haplotype variation involves TAP2 allelic variation?

Answer:

In chickens, both the TAP1 and TAP2 genes are polymorphic, with different alleles for each MHC haplotype. In several examples we have shown that the TAP molecule of a particular haplotype pumps peptides that are tailored for the dominantly-expressed class I molecule of that haplotype (encoded by the BF2 gene) (Walker et al 2011; Tregaskes et al 2016). Many mammals have (nearly) monomorphic TAPs that pump a wide range of peptides (what we call an “average best fit”) for any class I molecule in that species. However, there are differences between species: mouse TAPs pump primarily peptides with a hydrophobic (aliphatic) amino acid at the C-terminus, while human TAPs pump a wider range of peptides including those with hydrophobic and basic (but not acidic) amino acids at the C-terminus of the peptide. Some mammalian species have limited TAP polymorphism. For example, two allelic lineages of rat TAPs pump peptides with different requirements for amino acids at the C-terminus, one more like mice and the other more like humans. These two kinds of TAP specificities come from rat MHC haplotypes that have class I molecules with similar specificities, indicating that the TAP and class I genes in rats have co-evolved (as originally proposed by the lab of Jonathan Howard) [JMD: see Joly et al. 1998 and Tuncel et al. 2014].

Question 10.

Since the early work in the Strominger and Wiley labs, have insights been changed and/or new things been established on MHC structures?

Answer:

As the techniques for characterization of bound peptides and for X-ray crystallography improved, many more MHC structures were solved including for class II molecules, and these data led to many new ideas. Moreover, a combination of biochemistry, cell biology and new technologies like cryo-electron microscopy (cryo-EM) have permitted us to understand the interactions between different molecules in the MHC pathways, both in loading the peptides and in the recognition by T cell and natural killer (NK) cell receptors (and co-receptors). Also, methods like molecular dynamics (MD) and nuclear magnetic resonance (NMR) have begun to allow us to appreciate the dynamic nature of MHC molecules, bending, twisting and breathing.  However, the work from Wiley and Strominger labs are at the foundation of these molecular understandings, and those structures have stood the test of time in a way that should make us all proud to be scientists.

Categories: Posts