Introduction

Humans and chimpanzees share a common ancestor that lived approximately five million years ago (mya) (Fujiyama et al. 2002). Two species of chimpanzees—the common chimpanzee and the bonobo—are officially recognised. Both have their natural habitat in Africa. Based on geographic distribution and mitochondrial DNA (mtDNA) variation, the common chimpanzee is subdivided into four different subspecies: Pan troglodytes verus (West Africa), Pan troglodytes troglodytes (Central Africa), Pan troglodytes schweinfurthii (East Africa) and Pan troglodytes ellioti (Nigeria-Cameroon) respectively (Bowden et al. 2012; Gagneux et al. 1999). The bonobo species comprises several populations inhabiting different sections of forest, lying south of the Congo River (Kawamoto et al. 2013; Zsurka et al. 2010). Due to deforestation and hunting in the context of the bush-meat trade, both chimpanzee species have become highly endangered (IUCN&ICCN 2012; Junker et al. 2012; Laporte et al. 2007). As a consequence, zoos around the world have initiated breeding programmes dedicated to conservation, such as the European Endangered Species Programme (EEP), providing support, for instance, to sustaining the (sub)species P. t. verus as well as the bonobo.

Common chimpanzees can be infected naturally with a simian immunodeficiency virus SIVcpz that is considered to be the ancestral progenitor of the virus that causes the acquired immunodeficiency syndrome (AIDS) pandemic in humans (Hahn et al. 2000). Until now, natural infections with SIVcpz have only been documented for P. t. troglodytes and P. t. schweinfurthii (Huet et al. 1990; Peeters et al. 1989; Peeters et al. 1992), whereas for the other two subspecies but also for the bonobo such natural infections have not been recorded (Leendertz et al. 2011; Li et al. 2012; Locatelli et al. 2016; Van Heuverswyn et al. 2007). Worldwide, approximately 150 common chimpanzees, mainly P. t. verus, have been infected experimentally with various human immunodeficiency virus type 1 (HIV-1) strains, but most of these animals appeared to control the infection, and progression towards AIDS was rarely observed (Heeney et al. 2006; Novembre et al. 1997; Novembre et al. 2001). Similarly, AIDS-like symptoms in free-ranging animals are documented anecdotally for naturally SIVcpz infected animals of the P. t. troglodytes and P. t. schweinfurthii subspecies (Etienne et al. 2011; Keele et al. 2009). Noteworthy, however, is that some zoos and primate centres have housed naturally SIVcpz infected common chimpanzees for decades without any sign of AIDS-like disease (Heeney et al. 2006). Hence, common chimpanzees seem to be relatively resistant to the development of AIDS after an HIV-1/SIVcpz infection.

The classical major histocompatibility complex (MHC) class I genes, in humans designated HLA-A, HLA-B and HLA-C, encode molecules that play a prominent role in the immune response, by presenting degraded antigen particles (peptides) of intracellular origin to cytotoxic T cells (CTL). When the peptide originates from a foreign structure—such as a virus—the subsequent recognition by the T cell receptor complex on the CTL may then result in elimination of the infected cell. One hallmark feature of the HLA system is that its class I molecules display abundant levels of polymorphism. As most of the polymorphic residues map to key positions forming the scaffolding of the peptide-binding site, each allotype has the tendency to bind one peptide selected from its own unique gradient of peptides. In HIV-1 infected humans, resistance to the developing of AIDS is also observed and control of the infection correlates with the presence of particular MHC class I molecules such as HLA-B*27 and -B*57 (Carrington and O'Brien 2003; Dalmasso et al. 2008; Fellay et al. 2007; Goulder et al. 1996; International et al. 2010; Kaslow et al. 1996; Kiepiela et al. 2004). Equivalents of the HLA-A, HLA-B and HLA-C genes are also present in common chimpanzees and bonobos and are designated Patr-A, Patr-B, and Patr-C, and Papa-A, Papa-B and Papa-C, respectively (Maccari et al. 2017). One would expect that species such as humans and chimpanzees—that shared a common ancestor rather recently—would possess highly similar MHC repertoires. However, a comparative genetic study demonstrated that common chimpanzees, in the distant past, experienced a selective sweep affecting the MHC class I region, and a hypothesis was formulated that the sweep was possibly caused by an HIV-1-like retrovirus (de Groot et al. 2002). A subsequent functional study provided evidence that the skewing of the common chimpanzee MHC class I repertoire seems to have favoured the survival of animals whose allotypes are able to recognise conserved areas of the SIVcpz/HIV-1 Gag proteome, similar to HLA-B*27/B*57 (de Groot et al. 2010). As a consequence, the contemporary living common chimpanzee populations have modified MHC repertoires, partly reduced, but they are able to cope with their natural habitat and with retroviral infections such as SIVcpz/HIV-1 (de Groot and Bontrop 2013).

The time point of the selective sweep lies deep in evolutionary history and may have predated the speciation of common chimpanzees and bonobos (de Groot et al. 2002). Hence, we were keen to investigate whether signatures of the selective sweep could be traced back in the MHC repertoire of bonobos. For this purpose, we selected two panels of animals to study their MHC class I repertoire. Part of the mitochondrial DNA D-loop was also sequenced in order to determine the potential relationship between the bonobos within the panels, as well as to rule out any possibility that we were sampling animals from an inbred population.

Materials and methods

Genomic DNA samples

Genomic DNA (gDNA), isolated either from blood or hair samples of 25 bonobos participating in the European Endangered Species Programme (EEP), was selected for this report (Supplementary Fig. 1). Fifteen of these animals are founder animals of the population currently housed in different European zoos, and then animals represent their offspring. This means that we were also able to analyse, to a limited extent, the segregation of polymorphic MHC markers in a family. To expand the panel, gDNA of 20 bonobos from our DNA repository was incorporated into the study (Supplementary Fig. 1).

mtDNA analysis

To identify part of the mitochondrial D-loop, primers were used that had been described previously (de Groot et al. 2005), which amplify the first 380 bp of the mtDNA D-loop that represents the part with the most polymorphic sites. Depending on the DNA sample, we were able to sequence between 305 and 326 of the 380 bp product. The polymerase chain reaction (PCR) mixture (50 μl) contained gDNA (50 ng), 0.5 μM of each primer, 1× PCR buffer, 2 mM MgCl2, 0.1 mM of each deoxiribonucleoside triphosphate (dNTP), and 0.2 μl platinum Taq polymerase (Invitrogen). A touch-down PCR was performed as reported earlier (de Groot et al. 2005). The PCR products were purified using the GeneJet Gel Extraction Kit (Thermo Fisher Scientific) and sequenced directly on a 3500XL automated capillary system using BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific). The data were analysed using the software Lasergene SeqMan Pro version 12.0.0 (DNAstar) and MacVector version 14.5.3 (MacVector). Sequences reported in this publication have been deposited in the GenBank database (for accession numbers see Supplementary Fig. 1).

Papa-A, Papa-B and Papa-C gene sequencing and nomenclature

As well as exons 2 and 3, intron 2 is informative with regard to understanding lineage evolution and assignment (de Groot et al. 2002). Locus-specific primers were therefore selected, annealing in intron 1 and intron 3. For Papa-A: (5′AIN) 5′GAA ACS GCC TCT GYG GGG AGA AGC AA3′ and (3′AIN) 5′TGT TGG TCC CAA TTG TCT CCC CTC3′. For Papa-B: (5′BIN) 5′GGG AGG AGC GAG GGG ACC GCA G3′ and (3′BIN) 5′GGC CAT CCC CGG CGA CCT AT3′. For Papa-C: (5′CIN) 5′GCG GAG AGG AGC GAG GGG CC3′ and (3′BCIN) 5′GGA GAT GGG GAA GGC TCC CCA CT3′. The PCR was performed on gDNA (50 ng) in a 25-μl reaction mixture containing 1× HF buffer, 0.1 mM of each dNTP (Papa-A), or 0.3 mM of each dNTP (Papa-B and Papa-C), 0.5 μM of each locus specific primer, 3% DMSO and 0.5 U Phusion DNA hot start II polymerase (Thermo Fisher Scientific). Amplification started with a hot start of 30 s at 98 °C, followed by 33 cycles each consisting of 10 s at 98 °C, 50 s at 65 °C and elongation for 50 s at 72 °C and a final extension of 5 min at 72 °C.

Amplification success was verified using gel electrophoresis, and the PCR products were subsequently purified using the GeneJet Gel Extraction Kit and ligated into the pJET1.2/blunt-cloning vector using the CloneJet PCR cloning kit (Thermo Fisher Scientific). Transformation was carried out using competent XL-1 blue Escherichia coli and the TransformAid Bacterial Transformation Kit (Thermo Fisher Scientific). Per ligated PCR product, 24 colonies were picked for plasmid isolation. Sequencing reaction was performed using the BigDye Terminator v3.1 Cycle Sequencing Kit, and samples were run on an automated capillary system ABI 3130XL or 3500XL. The data were analysed using the same software as described above. All alleles referred to in this communication were recovered in at least two independent PCR reactions. Sequences reported in this publication have been deposited in the Genbank database (accession numbers KY652679-KY652717, KY688125-KY688126 and KY964060-KY964064). The nomenclature rules for alleles of the MHC systems of nonhuman primates have been reported in detail (de Groot et al. 2012; Ellis et al. 2006). The combination of annotated allele sequences and relevant names has been deposited in the IPD-Immunopolymorphism Database (Maccari et al. 2017).

Phylogenetic analyses of mtDNA and MHC class I sequences

Maximum likelihood (mtDNA, partial sequences), Neighbour-Joining (MHC class I exon 2 and 3 sequences) and UPGMA (MHC class I intron 2 sequences) trees were constructed with the MEGA 7.0.18 application (Kumar et al. 2016) using the Jukes-Cantor or Kimura-2 parameter method for calculating the pairwise distances. Bootstrap values were calculated based on 1000 replicates.

Results

Mitochondrial DNA variation in the bonobo panels covers five out of six known clades

mtDNA is inherited through the maternal lineage and thus can provide insight into potential family relations, but it may also shed light on the diversity probed within a given sample. Part of the mtDNA D-loop in 12 animals of the European Endangered Species Programme (EEP) cohort and 13 animals of the additional panel could be amplified (Supplementary Fig. 1) and produced high-quality genetic material. In the EEP cohort, we found that Bono, Simon (child of Kombote) and Daniela and Mato (children of Margrit) shared identical partial mtDNA D-loop sequences, which suggest that the EEP founder animals Bono, Komboto and Margrit are related to each other. We also found evidence that animals grouping within the two different panels may be related. For example, Djanoa and PpA08 share the same partial mtDNA D-loop part with Bono, Simon (child of Kombote) and the children of Margrit (Daniella and Mato). Based on the bonobo studbook (Pereboom et al. 2016), we were able to confirm that Djanoa is the granddaughter of Margrit. Further sharing of the partial mtDNA sequences was observed for two other examples: Chipita and PpA01 and PpA02, PpA17 and PpA20 respectively.

For several of the animals included in the EEP cohort, the whole mitochondrial genome was published recently (Lobon et al. 2016) (Supplementary Fig. 1). This provides the opportunity to use the entire mtDNA D-loop (1124 bp) sequences for comparison and further analysis. These extended data illustrated, however, that Bono, Kombote and Margrit all have unique mtDNA D-loop sequences (Supplementary Fig. 1, see column seq. id.) and thus are not related via the maternal lineage. In addition, the partial mtDNA D-loop sequence of Chimba matches with the first part of the complete mtDNA D-loop sequence of Natalie, which may indicate an above-average relatedness between these animals.

Based on the mtDNA variation present in several different bonobo populations, six different clades were recognised and designated A1, A2, B1, B2, C and D (Kawamoto et al. 2013; Zsurka et al. 2010). To identify the clade of the animals in the present panel, the 25 partial mtDNA D-loop sequences (12 EEP cohort and 13 additional panel) in combination with a selection of sequences representing all previously described clades were subjected to phylogenetic analysis (Fig. 1). Additionally, mtDNA sequences identified for Desmond, Hermien, Hortense, Kombote and Natalie were added to the phylogenetic analysis (Lobon et al. 2016). MtDNA D-loop information of human as well as all different subspecies of common chimpanzees was implemented as outgroup to root the resulting tree. As can be seen, the EEP cohort comprises animals originating from clades A2, B1, B2, C and D, whereas animals from the additional panel cluster in clades A2, B2 and C (Fig. 1). This would suggest that, based on mtDNA variation, the bonobo samples included in our study reflect the diversity seen in bonobo populations living in the wild.

Fig. 1
figure 1

Phylogenetic analysis of partial mtDNA D-loop bonobos sequences (289 bp in length). The sequences of common chimpanzees and humans are taken as out-groups (de Groot et al. 2008; Gagneux et al. 1999; Ingman et al. 2000). Relevant sequences—for clade identification—for bonobos are incorporated in the analysis with their accession numbers in brackets (Kawamoto et al. 2013), and the different clades, A1, A2, B1, B2, C and D, are indicated. Orange boxes indicate EEP cohort animals; aqua boxes indicate animals in the additional panel. Relevant bootstrap values are indicated

Full size image

Allelic variation of MHC class I genes in bonobos

The most polymorphic part of the MHC class I gene, encoding the peptide-binding region (exons 2 and 3), was sequenced and revealed the presence of 14 Papa-A, 20 Papa-B and 11 Papa-C alleles in the two panels. None of the bonobo alleles was identical to a reported common chimpanzee allele. Eight out of 14 Papa-A alleles had not been reported previously (Fig. 2, boldface). Most of the Papa-A alleles were shared between members of the two different panels, except for Papa-A*01:01, *03:04, *08:02, *09:01 and *09:02, which were only detected in the EEP cohort. For the Papa-B locus, five previously unreported alleles were detected (Fig. 2, boldface), whereas 15 others were confirmed (Wroblewski et al. 2017). Nine of the 20 B alleles were shared between the two panels; these were Papa-B*01:02, *02:02, *07:01, *09:01, *12:02, *14:01, *17:01, *19:02 and *21:01. In addition to the three already catalogued, eight previously unreported Papa-C alleles were detected (Fig. 2, boldface). Four newly identified alleles were detected only in animals in the additional panel, of which three showed variations in their introns as reflected by the eight-digit designation. Six alleles were shared, respectively: Papa-C*03:02, *03:01:01:01, *03:03:01:01, *04:01, *04:02 and *05:01. The deduced protein alignments of the different Papa-A, Papa-B and Papa-C allotypes are provided (Fig. 3). These alignments suggest that the Papa-C alleles seem to show the highest level of conservation, and most polymorphism seems to have been generated by means of simple point mutations. This also appears to be the case for the Papa-A locus alleles, but within this locus, more variation is observed as compared to the -C locus. For the Papa-B alleles, an exchange of sequence motifs has played an important role in generating allelic polymorphism (Fig. 3). All this is in concordance with the picture painted earlier for common chimpanzees (de Groot et al. 2000).

Fig. 2
figure 2

a Overview of the MHC class I variation observed in EEP cohort animals. The relevant mtDNA information, clade and Seq. id (sequence identification), is provided (see also supplementary Fig. 1). Previously unreported alleles are indicated in boldface. Animals that are family members are indicated by ^ or &. The deduced MHC haplotypes are colour coded. *Zorba is a potential founder, as it has not generated offspring. Nd means not determined. b Overview of the MHC class I variation observed in the additional bonobo samples. The relevant mtDNA information, clade and Seq. id, is provided. The newly identified alleles are shown in boldface. Nd means not determined

Full size image
Fig. 3
figure 3

Deduced amino acid alignments (alpha 1 and 2 domains) of the different Papa-A, Papa-B and Papa-C alleles identified in the two panels. The human sequences HLA-A*01:01:01, HLA-B*07:02:01 and HLA-C*01:02:01 are taken as consensus (Robinson et al. 2015). A dash indicates identity to the consensus, and an amino acid replacement is represented by the conventional one-letter code. Amino acid residues constituting the B and F pockets are indicated in grey in the consensus sequences. The newly identified alleles are shown in boldface. The coloured boxes highlight the polymorphic sequence motifs that are shared between Papa-B molecules

Full size image

The bonobo alleles identified in the two panels were subjected to phylogenetic analysis together with representative human and common chimpanzee alleles (Supplementary Fig. 2). All Papa-A alleles cluster together with HLA-A1/A3/A11 family members, but orthologues of the other five HLA-A family members are absent. The same finding has been made for the A-locus alleles of the common chimpanzee (de Groot et al. 2000). For the B locus, common chimpanzee and bonobo alleles tend to cluster together into lineages. Intermingling with HLA-B alleles has not been observed, and potential evolutionary relatedness is probably obscured due to the frequent recombination of polymorphic motifs (Belich et al. 1992; Watkins et al. 1992). The C alleles of the common chimpanzee and bonobo tend to cluster tightly together, illustrating the relatedness of both species.

In conclusion, the presence of orthologues of only one out of five HLA-A family members in both common chimpanzees and bonobos would fit with the notion that the ancestor of these species, in the evolutionary past, experienced the same selective sweep targeting the MHC repertoire. It is tempting to speculate that also the -C locus shows signs of a sweep (low allelic heterogeneity), but it was preferred to make a thorough analysis of the MHC class I intron 2 sequences in the bonobo, as these turned out to be highly informative for studying the evolutionary past of common chimpanzees (de Groot et al. 2002).

The bonobo MHC class I intron 2-repertoire is even more reduced than in common chimpanzees

It is generally accepted that introns are not subject to strong evolutionary forces selecting for variation, whereas this is obvious for MHC class I exons 2 and 3, encoding the peptide binding side (Hughes and Nei 1988). As a consequence, intron sequences can be valid markers to study the origin or relatedness of genes in species. Hence, the genetic variation for the MHC class I intron 2, mapping between exons 2 and 3, was analysed for our bonobo panels. These data were combined with the available human and common chimpanzee intron 2 sequences and subjected to phylogenetic analysis (Supplementary Fig. 3). Based on phylogeny, human and common chimpanzee intron 2 sequences were previously classified into different lineages (de Groot et al. 2002). For the A, B and C intron 2 sequences, ten, six and nine different lineages have been characterised, respectively, and named in alphabetical order. For example, the A locus lineages are named a1 to a10 (Supplementary Fig. 3). This format was taken as a lead to classify the intron 2 sequences observed in bonobos. The analysis of two panels recovered only two Papa-A, three Papa-B and six Papa-C intron 2 sequences (Supplementary Fig. 3). More in detail, both Papa-A intron 2 sequences appeared to be identical to those detected earlier in common chimpanzees, suggesting that all allelic variation at the A locus in bonobos stems from only two ancestral structures. They cluster either into the a2 lineage, shared between humans and common chimpanzees, or in the a9 lineage, which appears to be present in common chimpanzees and bonobos (Fig. 4). Humans and common chimpanzees both harbour a5-lineage members, which seem to be absent in bonobos. Till now, members of the a10 lineage have only been recovered from common chimpanzees (Fig. 4). Overall, the intron 2 A-repertoire of bonobos shows less variation as compared to common chimpanzees. For the B gene, similar observations have been made. The three sequences detected in bonobos group into lineage b3, which appears to have evolutionary cousins in common chimpanzees and humans (Fig. 4). Members of the b5 lineage, apparently indicative for common chimpanzees, have not been detected. Six sequences were detected in the case of the MHC class I C locus in bonobos, and they cluster in the c1, c7 and c9 lineages (Fig. 4). One might have expected to find members of the c8 lineage in bonobos, but these appear to be absent. Three intron 2 C sequences are identical to equivalents in common chimpanzees.

Fig. 4
figure 4

Pie charts showing the presence/absence of MHC class I intron 2 lineages in humans, common chimpanzee and bonobo. A coloured section in a pie indicates the presence of a particular intron 2 lineage in that species: red for MHC-A, blue for MHC-B and orange for MHC-C

Full size image

Taken together, as compared to humans, only a limited number of intron 2 lineages was observed in bonobos (Fig. 4), which is comparable to the situation observed in common chimpanzees. Bonobos may have lost two, one and one lineage members for the A, B and C locus, respectively, as compared to common chimpanzees. Both chimpanzee species share a considerable number of intron 2 sequences, and six out of 11 are identical, thus reflecting the close relatedness of both species. In contrast, none of the exon 2 and 3 sequences identified during this study are identical to common chimpanzee alleles, suggesting that diversifying selection remained active on the gene products encoded by the exons of the MHC class I molecules in common chimpanzees and bonobos.

Definition of five bonobo MHC class I haplotypes by segregation analyses

Due to the availability of genetic material of family members, we had the opportunity to define haplotypes within the EEP cohort. The first family investigated comprised mother Margrit, children Daniela and Mato and grandchildren Djanoa and Lomela. Moreover, Bono is an informative individual, as he fathered Lomela (Supplementary Fig. 1). Daughter Daniela inherited the following combination of genes/alleles: Papa-A*06:02, Papa-B*01:02 and Papa-C*04:02 from mother Margrit (Fig. 2a, red haplotype). Mato inherited the Papa-A*06:01, Papa-B*01:01 and Papa-C*04:01 haplotype from mother Margrit (Fig. 2a, green haplotype). Both Daniela and Mato are positive for the Papa-A*05:02, Papa-B*07:01 and Papa-C*03:02 haplotype, which segregates from the paternal line via Camillo (Fig. 2a, blue haplotype). Both grandchildren Djanoa and Lomela inherited the red haplotype, which can be traced back to Margrit. Lomela inherited Papa-A*09:01 and -B*09:01 from her father Bono. For Lomela, only one C locus allele could be detected, Papa-C*04:02, suggesting that this animal may be homozygous for this marker. Bono is heterozygous, as we recovered Papa-C*03:02 and Papa-C*04:02. Within the bonobos studied, Papa-C*03:02 and Papa-B*07:01 are assumed to segregate together, suggesting a strong linkage between these two entities. Hence, the most likely deduction is that Papa-C*04:02 segregates with Papa-A*09:01 and Papa-B*09:01, and Bono’s second haplotype comprises Papa-A*03:02, Papa-B*07:01 and Papa-C*03:02 (Fig. 2a, orange and purple haplotypes, respectively). A second family investigated comprised mother Catherine and child Mofana, a son. Mofana most likely inherited the alleles Papa-A*05:02, Papa-B*07:01 and Papa-C*03:02 via the maternal line and Papa-A*06:02, Papa-B*09:01 and Papa-C*04:02 from the paternal line via Vernon (Fig. 2a, blue and brown haplotypes, respectively).

Thus, five MHC class I haplotypes could be defined with certainty in the EEP cohort. Within the additional bonobo samples analysed, these haplotypes are most likely also present. For example, the blue haplotype may be present in PpA09 and PpA17, and the red haplotype seems to be present in Ppa16, although we were not able to link the C*0402 allele to this haplotype in this animal. As the family history was unfortunately not available to us, the factual presence of these haplotypes in these animals could not be substantiated.

Furthermore, within the studied bonobos, a strong linkage was observed between Papa-B*07:01 and Papa-C*03:02. The combination was present in four different mtDNA clades and was observed in combination with different A gene alleles. The dispersed presence of Papa-B*07:01/C*03:02 suggests that this combination of alleles may fulfil an important immunological function in the evolutionary history of bonobos.

Comparison of peptide-binding pockets

The biological function of MHC class I allotypes is the binding of degraded antigen segments (peptides) of pathogens and the subsequent activation of immune cells. Most polymorphism is confined to the amino acid residues that form the scaffolding of the peptide-binding site (Bjorkman et al. 1987; Saper et al. 1991). Hence, different allotypes have the tendency to bind a distinct range of peptides. At this stage, we do not have peptide-binding data for the different MHC class I allotypes of the bonobo. However, experience has taught us that the actual composition of the prominent binding pockets B and F may provide some insight into their peptide-binding potential (de Groot et al. 2010). Thus, we compared the composition of these pockets in common chimpanzees and bonobos. As can be seen, there is a high level of similarity—though they are not identical—with regard to the combination of B and F pockets for the A locus allotypes in both species (Fig. 5). Some of the common chimpanzee A locus allotypes were found to be promiscuous binders (de Groot et al. 2010; McKinney et al. 2000; Sidney et al. 2006), and because of the high similarity, this may also be the case for bonobos. Based on the B pocket composition, the B locus allotypes of the bonobo can be divided into two major groups (Fig. 5). One of those groups (Fig. 5, green) seems to be shared with common chimpanzees, whereas the allotypes in the other group seem to have evolved in bonobos (Fig. 5, orange). Allotypes with identical or highly similar pockets that bind conserved elements of the Gag protein of retroviruses in common chimpanzees (Patr-B*01:01, Patr-B*03:01, Patr-B*05:01 (de Groot et al. 2010)) are not present in bonobos (Fig. 5), although some allotypes in the orange group and Patr-B*05:01 have amino acid residues in common. The sequence comparisons indicated that the C allotypes of common chimpanzees and bonobos are highly similar, as is reflected by their B and F pockets (Fig. 5).

Fig. 5
figure 5

B- and F-pocket combinations for the bonobo MHC-A, MHC-B and MHC-C allotypes. For comparison, representative B- and F-pocket combinations from common chimpanzee (Patr) and human (HLA) are given. The amino acid residues determining the B-pocket are 7, 9, 24, 25, 34, 45, 63, 66, 67, 70 and 99 and for the F-pocket are 74, 77, 80, 81, 84, 95, 97, 114, 116, 123, 133, 143, 146 and 147. The B- and F-pocket residues of HLA-B*57:01 are taken as a consensus. A dash indicates identity to the consensus, and an amino acid replacement is represented by the conventional one-letter code. The names of the newly identified alleles are indicated in boldface. Underlined Patr names are the allotypes that were shown to bind conserved elements of the Gag protein of retroviruses (de Groot et al. 2010)

Full size image

Discussion

The formation of the Congo River is considered to be one of the important geographical drivers that was instrumental in the speciation process that led eventually to the existence of bonobos. The speciation process started about 1.5 to 2 mya (Becquet et al. 2007; Prado-Martinez et al. 2013), which—on a geological and biological time scale—is relatively recent. Hence, common chimpanzees and bonobos are related as is reflected by their genetic material, and there is even evidence of several introgression events (de Manuel et al. 2016).

There is ample genetic and functional evidence of a selective sweep that targeted the MHC class I region of common chimpanzees (de Groot et al. 2002, 2008, 2010). The selective sweep resulted in a reduction of MHC class I lineages, and many of the contemporary common chimpanzee MHC class I allotypes are able to bind conserved segments of retroviruses, such as HIV-1/SIVcpz. This work is based mainly on studies done in the West-African subspecies (P. t. verus). However, all the other MHC class I alleles that have been extracted from the other common chimpanzee subspecies fit within the picture (Maccari et al. 2017). We wished to investigate whether the signatures of this selective sweep are also detectable in contemporary bonobos. To this end, we selected a panel of subjects that is a relatively thorough snapshot of the wild-ranging animals. Indeed, our MHC class I repertoire analyses showed that common chimpanzees and bonobos have highly similar profiles. Notably, however, the intron 2 lineage analysis seems to highlight the most relevant message, as bonobos seem to have an even more limited MHC class I intron 2 lineage repertoire than common chimpanzees. The first implication would be that—as our initial calculations suggested (de Groot et al. 2002)—the selective sweep in the MHC class I repertoire predates the speciation of bonobos and common chimpanzees. The second question that remains to be answered is why do bonobos have an even more reduced intron 2 repertoire. One possibility is that there exists a gradient of MHC class I diversity, which diminishes from the West African to the Central African subspecies. This is unlikely, as there are reports that the West African chimpanzees display the lowest level of genetic variability (MacFie et al. 2009; Wegmann and Excoffier 2010). The other possibility is that we are dealing with a founding effect (the bonobo population started with a limited number of individuals) or that the MHC class I repertoire of bonobos was further reduced by subsequent selective processes. At this stage, we cannot exclude any one of these factors or a combination of them. However, other studies also have suggested that bonobos may have experienced strong genetic bottle necks (Prado-Martinez et al. 2013)

Another marker substantiates the fact that the MHC region in common chimpanzees was targeted by a selective sweep, namely the MIC gene, which maps in the MHC region next to the B gene. Most nonhuman primate species have haplotypes with a tandem of genes named MICA and MICB. However, in common chimpanzees, only one MIC gene is present, which arose from a deletion event that fused the tail of an MICB and the head of an MICA gene (Kulski et al. 2002). This single gene moiety, produced by a fusion event, is also present in bonobos (de Groot et al. 2005). A population study in common chimpanzees showed that haplotypes carrying this single hybrid MIC gene became fixed (de Groot et al. 2005). This indicates that haplotypes with the original tandem of MIC genes were negatively selected. As the hybrid MIC gene originated next to the MHC class I B gene, the fixation of the hybrid MIC gene on all haplotypes also indicates that due to linkage the positive selection on the hybrid gene may have coincided with the rescue of a limited MHC class I repertoire. Of course, it may have been the other way around, and even a strong positive selection on both factors cannot be excluded.

Although we observed substantial sharing of intron 2 sequences between common chimpanzees and bonobos, the MHC class I alleles themselves detected in our panels appear to be unique for these species. There is one report, however, claiming the sharing of alleles for the C locus between common chimpanzees and bonobos (Cooper et al. 1998). The sharing of identical MHC class I alleles between nonhuman primate species seems to be a rare event and is only reported for those species where there is introgression (Otting et al. 2017; Yan et al. 2011). This may explain the above-mentioned example as well. The majority of common chimpanzee and bonobo alleles appear to have acquired further mutations during their evolution, resulting in species-specific allotypes. Thus, the obvious question to ask is have contemporary bonobos also maintained MHC class I allotypes that are able to preferentially bind peptides from SIVcpz. One needs to be cautious in answering this question just by looking at the genetics. In the human population, the HLA-B*27 and HLA-B*57 allotypes are associated with elite control of HIV-1 infections. Therefore, the simple answer would involve looking for the presence of HLA-B*27- and HLA-B*57-like alleles in common chimpanzees and bonobos. However, experience has taught us that humans and common chimpanzees have evolutionarily distant MHC class I molecules that are able, however, to bind the same set of peptides (de Groot et al. 2010): for example, the common chimpanzee molecule that was evolutionarily more related to the HLA-B*27 molecule bound peptide that was favoured by the HLA-B*57 molecule and vice versa (Balla-Jhagjhoorsingh et al. 1999). Thus, as it can be tricky to compare distant species—as it is all about deciphering the function of these molecules—elucidation of the bonobo MHC class I peptide-binding motifs is essential. Nevertheless, with regard to the number of MHC class I intron 2-lineages and sequences, the contemporary bonobo population displays limited variation, but apparently the animals are able to cope with their pathogenic environment.