Development of MHC-Linked Microsatellite Markers in the Domestic Cat and Their Use to Evaluate MHC Diversity in Domestic Cats, Cheetahs, and Gir Lions

Abstract

Diversity within the major histocompatibility complex (MHC) reflects the immunological fitness of a population. MHC-linked microsatellite markers provide a simple and an inexpensive method for studying MHC diversity in large-scale studies. We have developed 6 MHC-linked microsatellite markers in the domestic cat and used these, in conjunction with 5 neutral microsatellites, to assess MHC diversity in domestic mixed breed (n = 129) and purebred Burmese (n = 61) cat populations in Australia. The MHC of outbred Australian cats is polymorphic (average allelic richness = 8.52), whereas the Burmese population has significantly lower MHC diversity (average allelic richness = 6.81; P < 0.01). The MHC-linked microsatellites along with MHC cloning and sequencing demonstrated moderate MHC diversity in cheetahs (n = 13) and extremely low diversity in Gir lions (n = 13). Our MHC-linked microsatellite markers have potential future use in diversity and disease studies in other populations and breeds of cats as well as in wild felid species.

The major histocompatibility complex (MHC) is a genomic region found in all jawed vertebrates that encodes proteins important for disease resistance (Klein 1986). These MHC proteins bind foreign peptides and present them to T cells, allowing for self/non-self recognition (Brown et al. 1988). MHC genes are the most polymorphic genes in the vertebrate genome (Krausa and Browning 1996) with more than 1500 allele variants identified for a single Class II locus in humans (Marsh et al. 2010). The diversity of MHC genes is likely to have arisen through pathogen-driven selection and is maintained by balancing selection (Klein et al. 1993). Class I MHC proteins present endogenous peptides to CD8+ T cells and serve as recognition elements for natural killer cells. These proteins play a key role in the recognition of virally infected cells and neoplastic cells (Williams et al. 2002). The expression of class II MHC proteins is restricted to lymphocytes (Hunt et al. 1995; Delves et al. 2006). These molecules present peptides generated in the vesicles of antigen-presenting cells, which are recognized by CD4+ T cells, activating B cells to produce antibodies (Villadangos 2001).

The feline MHC, also known as the feline leukocyte antigen (FLA) region, has been recently sequenced and assembled using bacterial artificial chromosomes (Yuhki et al. 2003, 2007, 2008). The class I region of the FLA region encodes 12 functional class I genes (Yuhki et al. 2007). Based on promoter structure, sequence polymorphism, and tissue expression, 3 class I genes (FLAI-E, FLAI-H, and FLAI-K) are proposed to be classical class I genes (Yuhki et al. 2008; Holmes et al. 2013). Although many eutherian mammals have 3 families of classical class II genes within the class II region (DP, DQ, and DR), only the DR family contains functional genes in the FLA region (Yuhki et al. 2003). This family has been expanded in the FLA region to contain 4 or 5 DRB and 3 DRA genes.

MHC gene diversity is often determined by sequence-based typing. However, as this method is labor intensive and expensive, it can be impractical for large diversity studies. A technique for inferring MHC polymorphism, which is becoming increasingly popular, is the use of MHC-linked microsatellites. Microsatellites closely linked to a gene under selection may be selected through genetic hitchhiking. This principle has been extensively evaluated in the Rhesus macaque (Macaca mulatta), where MHC-linked microsatellites were shown to have strong linkage disequilibrium (Penedo et al. 2005; Doxiadis et al. 2007; de Groot et al. 2008). Similarly, research has confirmed hitchhiking of microsatellites within the MHC region of mice (Mus musculus; Meagher and Potts 1997) and humans (Doxiadis et al. 2009).

MHC-linked microsatellites have been successfully correlated to MHC polymorphism in several studies. Aguilar et al. (2004) demonstrated that while the critically endangered San Nicolas Island fox (Urocyon littoralis dickeyi) had monomorphic neutral markers, the MHC-linked microsatellite markers were highly polymorphic. This suggests that intense selection at the MHC loci has selected for diverse MHC alleles and, by proxy, diverse microsatellite alleles. In a number of sheep populations, diversity in MHC-linked microsatellites was more varied than in neutral markers (Santucci et al. 2007). MHC-linked microsatellites have been used in conjunction with sequencing in MHC diversity studies (Lillie et al. 2012). MHC-linked microsatellites can be used with neutral microsatellites for the inference of population genetic structure and detection of signatures of selection (Hansen et al. 2007; Santucci et al. 2007). Studies have also used MHC-linked microsatellites to investigate the correlation between disease and MHC homozygosity in horses (Andersson et al. 2012) and plague resistance in rats (Tollenaere et al. 2012). Microsatellites may be used in species that do not have a well-characterized MHC.

To date, only 4 studies have looked at MHC diversity in domestic cats (Yuhki and O’Brien 1997; Kuwahara et al. 2000; Kennedy et al. 2002; Yuhki et al. 2008). These have shown that the FLA region demonstrates comparable allelic variation to the HLA complex. Currently, for the FLA-DRB locus, 75 feline DRB alleles have been characterized in 140 domestic cats (Yuhki and O’Brien 1997; Kuwahara et al. 2000; Kennedy et al. 2002; Yuhki et al. 2008). However, the FLA-DRA locus appears to have very limited diversity with only 1 allele currently characterized for exon 2 (Yuhki and O’Brien 1997).

Microsatellites developed in 1 species are frequently successfully utilized in closely related species. Ten microsatellites that were developed in the domestic cat were used to examine diversity in different wild cat species including the puma, tiger, and lion (Menotti-Raymond and O’Brien 1995). Two felid populations that are believed to have restricted MHC diversity are the cheetah (Acinonyx jubatus) and the Gir lion (Panthera leo persica). The cheetah is listed as a vulnerable species with only 15 000 individuals left in the wild (Freeman et al. 2001). For decades, cheetahs have been considered to be a model of a wild population with low MHC diversity. Low MHC polymorphism has been inferred through skin graft studies, where it was demonstrated that cheetahs would not reject skin grafts from unrelated individuals (O’Brien et al. 1985). A lack of MHC diversity was supported by other studies (Yuhki and O’Brien 1990, 1994; Drake et al. 2004), but these studies were limited by the use of low-resolution methods or small sample sizes. A recent study using cloning and sequencing of MHC along with single-strand conformation polymorphism in a large wild cheetah population (Castro-Prieto et al. 2011) contradicted previous studies of MHC class I, showing a higher level of MHC class I diversity in wild cheetahs then was previously recognized.

The Gir lion is a subspecies of lion restricted to the Gir sanctuary in India. Gir lions are the only remaining population of the once widespread Asiatic lion, which had a range across southwestern Asia (Joslin 1984). This population has been isolated since the 1880s and has undergone genetic bottlenecking with numbers reduced to as low as 20 individuals in the 1930s (Caldwell 1938). Today the population consists of approximately 300 individuals (Johnsingh et al. 2007). Studies in allozyme (Yuhki and O’Brien 1990) and minisatellite DNA fingerprints (Gilbert et al. 1991) have suggested genetic monomorphism in Gir lions outside the MHC region. Screening of class I MHC loci using restriction fragment length polymorphism (RFLP) revealed extreme genetic uniformity with no polymorphism across 15 individuals (Yuhki and O’Brien 1990). However, a recent study that studied sequence polymorphism of class I loci has suggested that there are substantial levels of MHC class I diversity in the Gir lion (Sachdev et al. 2005). As these studies are in conflict, further research is required to accurately determine the level of polymorphism of MHC loci in the Gir lion.

The domestic cat in Australia was founded by an unknown number of individuals around 200 years ago (Abbott 2002, 2008). In this study, we have developed MHC-linked microsatellite markers to be used as a proxy measure for MHC diversity in cats. Using these markers, we have investigated levels of class I and class II MHC polymorphism in the Australian domestic cat and have compared this to diversity at neutral genetic markers. We have then compared the diversity in the outbred population to purebred Burmese cats. Finally, the markers were used along with MHC sequencing to investigate MHC diversity in the cheetah and Gir lion.

Methods

Feline DNA Sampling

Blood samples were collected from 129 domestic mixed breed and 61 purebred Burmese cats presented to the Valentine Charlton Cat Centre, University of Sydney. These cats were presented to the centre with a variety of problems. The study was approved by the University of Sydney’s Animal Ethics Committee, N00/6-2009/1/4985. Cheetah (13) and Gir lion (13) DNA samples were obtained from the National Cancer Institute in Fredrick, Maryland, United States. Cheetah samples were collected from various zoos and wildlife reserves in the United States and Africa. Two subspecies were represented in our sample: Acinonyx jubatus jubatus found in Namibia, Botswana, Zimbabwe and South Africa, and Acinonyx jubatus raineyi found in Kenya, Uganda, and Tanzania. Gir lion samples were collected from lions at the Sakkarbaug Zoo, Junagadh, India. These were captive-born animals, which originated from the Gir forest in India. Genomic DNA was extracted using the MoBio UltraClean BloodSpin Kit (California) according to the manufacturer’s protocol.

MHC-Linked Microsatellites

The complete MHC sequence produced by Yuhki et al. (2008) (GenBank EU153401, EU153402) was analyzed by Repeat Masker (UCSC Genome Bioinformatics) in order to identify simple repeats. About 557 simple repeats were identified in the MHC region and unbroken dinucleotide repeats with a repeat length of 15 or greater and a distance of less than 10kb from a classical MHC were selected. Positions of the MHC genes within the sequence are described in Yuhki et al. (2008) and the exon/intron structure of the MHC genes was identified using EMBOSS GenScan. Uninterrupted dinucleotide repeats within the intron of a MHC gene or less than 10kb from a MHC gene were selected. Primers were designed using the Oligo 6 Primer Analysis Software (Molecular Biology Insights). One primer from each pair was 5ʹ-labeled with a CAG tag (Schable et al. 2002). Primers were analyzed with BLAST against the cat genome and cat MHC sequence to ensure high specificity to the target sequence. Four class I and two class II primer sets were ordered from Sigma-Aldrich (New South Wales, Australia).

PCR was carried out in 25 µL PCR reactions containing 10–60ng genomic DNA, 200 µM dNTP (Sigma-Aldrich), 1.5 units Taq polymerase (Invitrogen, CA), 1× PCR buffer (Invitrogen), 0.6mM unlabeled primer, 0.06mM CAG labeled primer, 0.6mM 5ʹ-fluorescent CAG primer (Applied Biosystems) and either 1.6mM or 2.4mM MgCl₂ (Invitrogen) determined by prior optimization. The following PCR conditions were used: denaturation at 90 °C for 3min, followed by 33 cycles of 94 °C for 30 s, annealing temperature between 52.0 and 60.4 °C (determined by prior primer optimization) for 30 s and extension at 72 °C for 30 s, with a final extension step at 72 °C for 10min. PCR products were submitted to the Ramaciotti Centre for Gene Function Analysis, UNSW for fragment analysis. The software program, PeakScanner (Applied Biosystems, Victoria, Australia) was used to assign genotypes.

Neutral Microsatellites

Five dinucleotide microsatellite primers (Fca045, Fca058, Fca078, Fca247, and Fca294) were selected from Menotti-Raymond et al. (1999) due to their location on different chromosomes, variety in length, and compatibility with the CAG tag. These markers are assumed to be neutral and are not known to be near a gene under selection. One primer from each pair was 5ʹ-labeled with a CAG tag (Schable et al. 2002). PCR reactions, in a total volume of 25 µL, were performed containing 10–100ng sample DNA, 1× PCR Buffer (Invitrogen), 200 µM dNTP (Sigma-Aldrich), 0.66mM tagged primer, 44 µM untagged primer (Sigma-Aldrich), 0.66mM fluorescent dye (Sigma-Aldrich), 1.5 units of Taq (Invitrogen), and the optimum MgCl₂ (Invitrogen) concentration determined by previous optimization tests. A touchdown PCR protocol was used where samples underwent a 3min 95 °C denaturation, followed by 12 cycles of 95 °C denaturation for 30 s, 30 s of annealing starting at 60 °C decreasing by 2 °C every 2 cycles, with an extension of 72 °C for 45 s. On completion of this touchdown phase, each sample went through 28 cycles of 95 °C denaturation for 30 s, 50 °C annealing for 30 s, and 72 °C extension for 45 s, with a terminal extension of 30min at 72 °C. Fragment analyses of PCR products were carried out as above and genotypes were assigned using Peak Scanner.

Statistical Analysis

Arlequin 3.5 (Excoffier and Lischer 2010) was used to determine the expected and observed heterozygosities, conformation to Hardy–Weinberg equilibrium and linkage disequilibrium between markers. GenePop 4 (Raymond and Rousset 1995) established the significance of the heterozygous changes. FSTAT 2.9.3 (Goudet 1995) determined the number of alleles, allelic richness, and F statistics including F _ST and F _IS. Allelic richness was used to correct for differences in pooled sample sizes through a rarefaction method, thereby producing the number of alleles independent of sample size. STRUCTURE (Pritchard et al. 2000) was used to identify subpopulations with the Evanno et al. (2005) calculation of K applied. Estimation of null allele frequency was performed in Micro-Checker (van Oosterhout et al. 2004). Linkage disequilibrium between the MHC-linked microsatellite markers was examined through the test of nonrandom association of alleles using Arlequin with 10 000 permutations and 10 initial conditions for the Expectation–Maximization algorithm.

Class I Exon 2 MHC Cloning and Sequencing

MHC class I alpha 1 domain alleles were amplified in 8 cheetah samples (subspecies jubatus) and 8 Gir lion samples using multilocus primers designed by Sachdev et al. (2005) (a1F:5ʹ–CCACTCCCTGAGGTATTTCTACACC–3ʹ; a1R: 5ʹ–GGAC TCGCTCTGGTTGTAGTAGCG–3ʹ). Additionally, MHC alleles from cheetah were amplified using the same reverse primer as previous but with an alternate forward primer designed by Castro-Prieto et al. (2011) (Acju_Ex2MhcI_CF: 5ʹ–GCTCCCACTCCTGAGGTAT–3ʹ) in order to obtain missing alleles (AcjuIaI*01, *02, and *03). Although both exon 2 and exon 3 are polymorphic, exon 2 (alpha I domain) was selected for sequencing as it was the most polymorphic peptide-binding domain based on previous studies. PCR reactions, in a total volume of 25 µL, were performed containing 10–100ng sample DNA, 1× High Fidelity PCR Buffer (Invitrogen), 200 µM dNTP (Sigma-Aldrich), 10 pm of each primer (Sigma-Aldrich), 1.5 units of Platinum Taq DNA Polymerase High Fidelity (Invitrogen), and 2mM MgSO4 (Invitrogen). The following PCR conditions were used: denaturation at 94 °C for 2min, followed by 33 cycles of 94 °C for 30 s, annealing of 60 °C for 30 s, and extension at 68 °C for 30 s, with a final extension step at 68 °C for 10min. Samples were run on 2% agarose gel and bands were excised. Gel bands were purified using the QIAquick Gel Extraction kit (QIAGEN). The purified fragments were cloned into plasmids using the pGEM®-T Easy vector system (Promega, Madison, WI). Plasmids were transformed into JM109 Escherichia coli bacterial cells (Promega), and clones were individually picked and cultured overnight at 37 °C. Plasmids were purified using the QIAprep Minispin Kit (QIAGEN). The plasmid DNA was sequenced at the Australian Genome Research Facility (AGRF, Westmead, NSW, Australia). Sequences were edited and quality checked using Sequencher 4.1.4 (Gene Codes). Sequences had to be present in two separate PCRs, either in the same or different individual, to be considered valid alleles. Sequence alignments were produced in BioEdit using the ClustalW alignment tool (Thompson et al. 1994; Hall 1999). MEGA 4 (Tamura et al. 2007) was used to construct phylogenies. In fulfillment of data archiving guidelines (Baker 2013), we have deposited the primary data underlying the analyses of this study with Dryad.

Results

MHC-Linked Microsatellite Primers

Six MHC-linked microsatellite primers were developed and optimized (Table 1). Four of these flanked FLA class I genes and 2 were in close proximity to FLA class II genes with the distance from their associated gene ranging from 7kb to within an intron of the gene (see Figure 1). The product lengths ranged from 174 to 300bp. All the class I and II microsatellite primers amplified a single polymorphic locus in the domestic cat. The MHC-linked microsatellites markers were designated MHCI-A, MHCI-B, MHCI-C, MHCI-D, MHCII-A, and MHCII-B. Linkage disequilibrium was tested between the MHC-linked microsatellite markers in the domestic mixed breed population. All associations between marker pairs were highly significant (P < 0.01).

Table 1.

MHC microsatellite primer sequences and characteristics

Marker ID	Flanking Gene	Position in relation to gene	Forward sequence	Reverse sequence	Product length (bp)	Optimum MgCl2 (µL)	Optimum annealing temperature (°C)
MHCII-A	DRB4	Between exon 1 and 2	GGTGATCTGAGTTTCCTTAG	CAGTCGGGCGTCATCATGTGTTCTCCTTTCCTAATC	292	0.8	60
MHCII-B	DRB2	<1kb from DRB2	ATCCCCTGAAAGAACTGT	CAGTCGGGCGTCATCATGTGCAACTCCTTAAATAAC	236	1.2	58
MHCI-A	FLAI-E	5kb from FLAI-E	AGCCATCCACAGAATACAC	CAGTCGGGCGTCATCAGCAAATGGTACAAAAATACA	288	0.8	55
MHCI-B	FLAI-K	7kb from FLAI-K	AAGAAGTAAACGATGGAGAG	CAGTCGGGCGTCATCACCAATAGTACAAATCTGACAT	243	0.8	55
MHCI-C	FLAI-K	3kb from FLAI-K	TGTCCTACTTCCAAAATCTT	CAGTCGGGCGTCATCAGACCAGTAGGGAAAGTGTA	300	1.2	60
MHCI-D	FLAI-O	3kb from FLAI-O	TATGAGGAGGGAAAATAATA	CAGTCGGGCGTCATCAGTTTCCATATAAGGCACTC	207	0.8	51

Reverse sequences include 5ʹ-CAG tag.

Figure 1.

Diagram of domestic cat MHC class II and class I classical regions showing position of MHC genes and microsatellite markers.

Genetic Variation in the MHC-Linked and Neutral Microsatellites

One hundred and twenty-nine domestic mixed breed and 61 Burmese cats were genotyped for MHCI-A, MHCI-B, MHCI-C, MHCI-D, MHCII-A, and MHCII-B MHC-linked microsatellite markers and Fca045, Fca058, Fca708, Fca247, and Fca294 neutral microsatellite markers. Genetic diversity was established for both the MHC-linked and neutral markers through allelic and heterozygosity traits. Although the allelic richness was similar between the MHC and neutral microsatellites, there were a greater number of private alleles within the neutral, compared to the MHC, markers (Table 2). Greater diversity was seen in the domestic mixed breed than in the Burmese group, with the domestic mixed breed displaying a significantly higher allelic richness than the Burmese (P < 0.01). The domestic mixed breed cats also had significantly higher expected heterozygosity (P < 0.01) and observed heterozygosity (P = 0.02) than the Burmese.

Table 2.

Genetic variation for each microsatellite marker for the domestic mixed breed and Burmese groups

	Locus	N	A	P A	A R	H E	H O	HWE	Null
Domestic mixed bred	MHCII-A	128	8	1	7.79	0.69	0.46	0.000	Yes
	MHCII-B	128	6	1	5.99	0.75	0.59	0.000	Yes
	MHCI-A	126	9	4	7.31	0.81	0.78	0.000	No
	MHCI-B	129	8	2	7.84	0.80	0.71	0.000	No
	MHCI-C	129	12	4	11.02	0.80	0.75	0.000	No
	MHCI-D	127	12	4	11.18	0.76	0.36	0.000	Yes
	Average MHC	128	9.00	2.67	8.52	0.77	0.61
	Fca045	129	11	4	10.10	0.85	0.80	0.204	No
	Fca058	129	7	3	6.39	0.71	0.73	0.534	No
	Fca078	129	8	3	7.42	0.79	0.67	0.000	No
	Fca294	129	12	6	10.35	0.85	0.81	0.468	No
	Fca247	129	10	5	7.25	0.64	0.49	0.000	Yes
	Average neutral	129	9.60	4.20	8.30	0.77	0.70
	Overall average	128	9.38	3.43	8.41	0.77	0.65
Burmese	MHCII-A	61	7	0	6.94	0.32	0.13	0.000	Yes
	MHCII-B	59	5	0	4.98	0.43	0.29	0.000	Yes
	MHCI-A	61	6	1	6.00	0.78	0.70	0.847	No
	MHCI-B	58	7	1	7.00	0.63	0.60	0.608	No
	MHCI-C	59	8	0	7.97	0.56	0.56	0.033	No
	MHCI-D	60	8	0	7.97	0.74	0.35	0.000	Yes
	Average MHC	60	6.83	0.33	6.81	0.58	0.44
	Fca045	59	7	0	6.97	0.29	0.20	0.000	Yes
	Fca058	61	4	0	4.00	0.50	0.49	0.928	No
	Fca078	61	7	2	5.90	0.57	0.49	0.028	No
	Fca294	60	7	1	6.97	0.73	0.63	0.036	No
	Fca247	59	6	1	5.98	0.73	0.61	0.043	No
	Average neutral	60	6.00	0.80	5.96	0.56	0.49
	Overall average	60	6.55	0.55	6.42	0.57	0.46

N, sample size; A, number alleles; PA, number of private alleles; A _R, allelic richness; H _E, expected heterozygosity; H _O, observed heterozygosity; HWE, P value for conformation to Hardy–Weinberg equilibrium; Null, significant null allele frequency. P < 0.05 are considered to deviate from Hardy–Weinberg equilibrium (these values are shaded).

The overall expected heterozygosity for both marker types was very similar. However, there was a deviation from Hardy–Weinberg equilibrium across many of the markers, both MHC linked and neutral. The deviation from Hardy–Weinberg equilibrium was more prominent in the MHC than in the neutral markers. All 6 and 4 out of 6 MHC microsatellites, for the domestic mixed breed and Burmese groups, respectively, did not conform to Hardy–Weinberg equilibrium (Table 2). Two out of 5 domestic mixed breed and 4 out of 5 Burmese neutral markers displayed significant deviation from expectations (Table 2). A heterozygote deficiency was demonstrated in all the markers that deviated from Hardy–Weinberg equilibrium. Null alleles were likely present in some of the markers that showed deviation from Hardy–Weinberg equilibrium, with 4 of the 8 markers that deviated from Hardy–Weinberg equilibrium in both the domestic mixed breed and Burmese cats showing evidence of null alleles (Table 2). This accounts for some, but not all, of the markers that had a heterozygote deficiency in both the neutral and MHC-linked markers. To confirm that the finding that these markers showed heterozygote deficiency was not the result of a technical artifact, our results were replicated with a subset of our samples in a separate laboratory. This replication confirmed a heterozygote deficiency.

Inbreeding, estimated by the inbreeding coefficient and determined using the neutral markers, was demonstrated to be low among domestic mixed breed (F _IS = 0.089), whereas the inbreeding in the Burmese group was much higher (F _IS = 0.14). This was very similar to the value calculated from microsatellite typing in American Burmese cats (F _IS = 0.16) in the recent study by Kurushima et al. (2012).

Genetic Differentiation Between the Domestic Mixed Breed and Burmese Groups

Significant genetic differentiation, estimated by fixation index, was observed between the domestic mixed breed and Burmese for both the MHC and neutral microsatellites. The neutral markers (F _ST = 0.156, P = 0.01) showed slightly more differentiation than the MHC-linked microsatellites (F _ST = 0.129, P = 0.01), although this difference was not significant (P = 0.68). The variation in samples was accounted for with 15.6% and 12.9% occurring between the domestic mixed breed and Burmese for neutral and MHC-linked markers, respectively, whereas 84.4% and 87.0% was found within the populations themselves. Genetic differentiation was also indicated by division of the entire sample population into 2 distinct subpopulations, domestic mixed breed and Burmese, when analyzed with STRUCTURE using combined neutral and MHC markers. When analyzed with the MHC-linked microsatellites, but not the neutral microsatellites, the domestic mixed breed population was further divided into 2 subpopulations with an F _ST of 0.27.

MHC-Linked Microsatellites in Cheetahs and Gir Lions

MHCII-A did not amplify in either the cheetah or lion samples. In the Gir lion, primers MHCII-B and MHCI-D were monomorphic. Marker MHCI-C amplified 2 fragments, whereas MHCI-A and MHCI-B amplified 3 fragments. These fragments were identical in every sample, so we conclude that these are duplicated, monomorphic loci being amplified by the primers. Therefore, all the primer sets amplified monomorphic loci in the Gir lion.

In the cheetah, primer MHCI-C amplified multiple loci of similar length so the results could not be interpreted. Microsatellite MHCI-D was monomorphic, whereas the remaining 3 microsatellites were polymorphic. The allelic and heterozygosity statistics for the polymorphic alleles are summarized in Table 3. The statistics were similar for both the jubatus and raineye subspecies. There was an average of 3.66 alleles per locus in each of the 2 cheetah subspecies.

Table 3.

Genetic variation for each microsatellite marker in the cheetah

	Locus	N	A	A R	H E	H O	HWE
Jubatus	MHCII-B	8	4	3.92	0.74	0.38	0.721
	MHCI-B	8	3	2.93	0.62	0.29	0.461
	MHCI-A	8	4	3.63	0.60	0.63	0.529
	Average	8	3.66	3.49	0.65	0.43
Raineye	MHCII-B	5	5	5	0.80	0.60	0.725
	MHCI-B	5	4	4	0.78	0.60	0.321
	MHCI-A	5	2	2	0.47	0.20	0.531
	Average	5	3.66	3.66	0.68	0.47
Pooled	Average	13	4.33	4.33	0.66	0.45

N, sample size; A, number alleles; PA, number of private alleles; A _R, allelic richness; H _E, expected heterozygosity; H _O, observed heterozygosity; HWE, P value for conformation to Hardy–Weinberg equilibrium.

MHC Class I Sequencing in Cheetahs and Gir Lions

In cheetahs, 16 unique alleles were amplified. Amino acid alignment of these alleles is shown in Figure 2 with the previously characterized MHC alleles AJUMHCAJUI1, AJUMHCAJUI3 (Yuhki and O’Brien 1994), and Acju-MHCI*02-12 (Castro-Prieto et al. 2011). Three of the alleles identified in this study were pseudogenes with frameshift mutations. An additional allele AcjuIαI*13 is likely to be a pseudogene as Castro-Prieto et al. (2011) identified a frameshift mutation downstream of the alpha I domain sequenced in this study. Seven of the alleles are believed to be functional, expressed classical class I alleles based on homology to classical alleles identified by Castro-Prieto et al. (2011). Five remaining alleles appear to be nonclassical alleles based on phylogenetic analyses (Figure 3). Of the 6 functional alleles identified by Castro-Prieto et al. (2011), which were unique in the alpha I domain, 3 were present in our sample set, whereas 4 novel classical alpha I domain sequences were identified. Four clusters of classical class I alleles corresponding to putative class I loci in the cheetah were identified by Castro-Prieto et al. (2011) based on phylogenetic analyses. Alleles from each of these clusters were identified in the present study; these alleles have been assigned to the clusters identified by Castro-Prieto et al. (2011) in Table 4. Only one classical allele (AcjuIαI*07) could not be assigned to a cluster. Between 1 and 4 classical alleles were found in each cheetah.

Figure 2.

Amino acid alignment of MHC class I alpha domain alleles in the cheetah. Dots indicate identity to the top sequence. Included are previously characterized MHC alleles AJUMHCAJUI1, AJUMHCAJUI3 (Yuhki and O’Brien 1994), and Acju-MHCI*02,04,05, 07,12 (Castro-Prieto et al. 2011) (only those unique at the alpha 1 domain are shown). Predicted classical alleles, nonclassical alleles, and pseudogenes are indicated by red, blue, and green dots, respectively. Asterisks above the sequence indicate putative peptide-binding sites based on alignment with human MHC I.

Figure 3.

Phylogenetic analysis of 43 MHC class I alpha I domain nucleotide sequences, including cheetah and Gir lion alleles from this study, previously characterized cheetah alleles (Yuhki and O’Brien 1994; Castro-Prieto et al. 2011), and domestic cat MHC class I alleles (Yuhki et al. 2008). FLAI-E, FLAI-H, and FLAI-K are domestic cat classical class I alleles, whereas FLAI-C, FLAI-F FLAI-J, FLAI-M, FLAI-O, and FLAI-Q are nonclassical alleles. The phylogenetic relationship was inferred using the neighbor-joining method (Saitou and Nei 1987). The percentage of replicate trees in which the associated sequences clustered together in the bootstrap test (1000 replicates) are displayed next to the branches, indicating the level of reliability of the phylogeny (Felsenstein 1985). Predicted classical alleles, nonclassical alleles, and pseudogenes are indicated by red, blue, and green dots, respectively. Domestic cat, cheetah, and lion alleles are indicated by black squares, stars, and diamonds, respectively.

Table 4.

Class I exon 2 genotypes in 8 cheetahs

Cheetah	Classical							Nonclassical					Pseudogenes
	01	02	03	04	05	06	07	08	09	10	11	12	13	14	15	16
	D	D	D	B	A	A							C
Acju1	x				x		x	x					x	x	x
Acju2		x		x	x		x		x			x	x			x
Acju3			x			x	x			x		x	x			x
Acju4	x			x	x					x		x	x	x	x	x
Acju5					x			x			x		x	x	x	x
Acju6				x	x							x	x			x
Acju7				x	x							x	x	x	x	x
Acju8				x	x					x		x	x	x	x	x

Abbreviated allele names are shown in row 2 followed by cluster in row 3 (see Castro-Prieto et al. 2011). Of the classical alleles only AcjuIαI*07 could not be confidently assigned to a cluster due to its divergence from the other alleles. Labels in column 1 (Acju1-8) are animal ID numbers. An x indicates the presence of the allele in the sample.

Nine unique alleles were identified in the Gir lion (Figure 4). Three of these alleles are pseudogenes with frameshift mutations (PaleIαI*07-09). Two alleles appear to be nonclassical alleles based on phylogenetic analysis (PaleIαI*05-06). The remaining 4 alleles (PaleIαI*01-04) are believed to be classical class I alleles. Of these alleles, PaleIαI*01 and PaleIαI*02 differ by only 2 nucleotide substitutions, whereas PaleIαI*03 and PaleIαI*04 differ by only a single-nucleotide substitution. Allele Pa6leIαI*01 was present in all individuals, whereas PaleIαI*03 was present in 7 of the 8 individuals. PaleIαI*02 and PaleIαI*04 were only present in 1 individual each. As the lion MHC has not been extensively characterized, we cannot rule out that the primers used are not missing some alleles, however, 6 of the 8 Gir lions sequenced were identical in their MHC class I classical alpha I domain complement suggesting extremely low MHC diversity. Individual lions had either 2 or 3 classical class I alleles (Table 5). This along with phylogenetic clustering of the alleles into 2 clades (Figure 3) suggest that 2 classical class I loci have been amplified.

Figure 4.

Amino acid alignment of partial MHC class I alpha domain alleles in the Gir lion. Dots indicate identity to the top sequence. Predicted classical alleles, nonclassical alleles, and pseudogenes are indicated by red, blue, and green dots, respectively. Asterisks above the sequence indicate putative peptide-binding sites based on alignment with human MHC I.

Table 5.

Class I exon 2 genotypes in 8 Gir lions

Lion	Classical				Nonclassical		Pseudogenes
	01	02	03	04	05	06	07	08	09
Ple1	x		x		x	x	x	x	x
Ple2	x		x		x		x		x
Ple3	x		x			x	x	x	x
Ple4	x				x	x	x
Ple5	x		x		x	x	x		x
Ple6	x		x
Ple7	x	x	x		x	x	x	x	x
Ple8	x		x	x	x	x		x

Abbreviated allele names are shown in row 2. Labels in column 1 (Ple1-8) are animal ID numbers. An x indicates the presence of the allele in the sample.

Discussion

MHC Diversity in Domestic Cat

Microsatellite markers linked to both classical class I and class II domestic cat MHC genes were designed and optimized and used to investigate MHC diversity in the Australian domestic cat. The 6 MHC-linked and 5 neutral microsatellites were used to genotype 129 domestic mixed breed and 61 Burmese cats. The microsatellite markers were polymorphic in domestic cats with an average of 9 alleles and average expected heterozygosity of 0.77 in the domestic mixed breeds. It has been suggested that MHC-linked microsatellites should be more diverse than neutral microsatellites as they are impacted by the balancing selection that maintains MHC diversity (Aguilar et al. 2004). Our finding that the average allelic richness was greater in the MHC microsatellites than in the neutral microsatellites for both the domestic mixed breeds (8.52 and 8.30, respectively) and the Burmese (6.81 and 5.96, respectively) supports this hypothesis, although this difference was not significantly different.

There was a significantly greater allelic richness (P < 0.01) and expected heterozygosity (P < 0.01) in the domestic mixed breed than that in the Burmese cats. The lower genetic diversity in the Australian Burmese is likely a result of a smaller gene pool and controlled breeding, resulting in inbreeding in the population. This hypothesis is supported by the high inbreeding coefficient for the Burmese neutral microsatellites (F _IS = 0.14). The inbreeding coefficient was higher in the Burmese than in the domestic mixed breed. This reflected the trend identified in Lipinski et al. (2008), where American purebred cats were more inbred than feral felines.

The average expected heterozygosity for the neutral microsatellites used in this study was previously determined to be 0.772 in a population of unrelated, outbred cats (Menotti-Raymond et al. 1999). In this current study, the average expected heterozygosities for the domestic mixed breed and Burmese were H _E = 0.767 and H _E = 0.564, respectively. Although the domestic mixed breed heterozygosity was very similar to that measured previously, that of the Burmese was much lower. Comparisons between our data and other species of Felidae highlighted that the Burmese group had a similar heterozygosity to mountain lions (Puma concolor) and tigers (Panthera tigris; Ernest et al. 2003; Mondol et al. 2009). This low heterozygosity in the Australian Burmese is of concern. It may be beneficial to introduce new genetic diversity into the Burmese breed to prevent and decrease the incidence of heritable diseases, which are significantly higher in the Burmese than in the outbred cats, such as diabetes (Lederer et al. 2009).

Population Structure

Analysis of both the neutral and MHC-linked microsatellites with STRUCTURE separated the samples into 2 subpopulations: Burmese and domestic mixed breed. However, when analyzing the MHC-linked microsatellites, the domestic mixed breed cats were further divided into 2 subpopulations. These domestic mixed breed subpopulations were moderately divergent with an F _ST of 0.27, and this divergence was seen at both class I and class II microsatellites. It is possible that this population division is related to different genetic backgrounds at the MHC in response to selective pressures from different diseases in the cat population. The samples used in this study were obtained from cats presenting a variety of problems. No distinction could be made between the 2 groups of cats based on clinical data, so further investigation will be required to determine the cause of this population structuring.

Deviation from Hardy–Weinberg

Many of the microsatellite markers used in this study deviated from Hardy–Weinberg equilibrium in both the Burmese and domestic mixed breed populations. While the presence of null alleles may account for some of these, for the MHC-linked microsatellites, 3 of the 6 in the domestic mixed breed and 1 of the 6 in the Burmese markers did not conform to equilibrium and did not show evidence for null alleles. Out of the neutral markers, 1 domestic mixed breed and 3 Burmese markers also displayed a significant deviation from Hardy–Weinberg equilibrium without evidence of null alleles being present. This deviation was the result of a heterozygote deficiency. There are a number of reasons why these markers may display a heterozygous deficiency.

Firstly, as discussed above, the MHC-linked microsatellites appear to be affected by population substructure. This can cause deviation from Hardy–Weinberg equilibrium due to the Wahlund effect. Additionally, the MHC-linked microsatellites are likely to be under selection due to hitchhiking of the microsatellite with the MHC genes. MHC loci are generally under positive balancing selection (Hughes and Yeager 1998), which leads to a heterozygote excess. However, we saw the opposite result in the cats tested. Therefore, selection at the MHC alone is unlikely to explain the heterozygote deficiency.

Among Burmese cats, inbreeding, which was demonstrated in this population, is likely to be contributing to the deviation from Hardy–Weinberg. While the domestic mixed breed population does not appear to be inbred, this population may have been affected by the founder effect early in their history. The Australian cat population is likely to have been established by a small number of individuals, which may have lead to a heterozygote-deficient population. Interestingly, there is a high prevalence of blood type B among Australian cats (Malik et al. 2005). This finding has been documented in only a few areas of the world, including the south east of England (Forcada et al. 2007). This area was the origin of the First Fleet, which likely brought the first domestic cats to Australia. The cause of the deviation from Hardy–Weinberg seen in these populations cannot be accurately determined at this point, but it is likely that it is multifactorial with inbreeding, selection, and population structure all contributing to this effect.

Population Differentiation

The domestic mixed breed and Burmese populations were significantly differentiated. This was seen in both the MHC-linked (F _ST = 0.129, P < 0.01) and neutral (F _ST = 0.156, P < 0.01) markers. The genetic differentiation between the Burmese and domestic mixed breed (F _ST = 0.156) is comparable to the F _ST value for Burmese and outbred domestic cats in the United States (F _ST = 0.16; Menotti-Raymond et al. 2008).

MHC Diversity in the Gir Lion

Previous research into Gir lion polymorphism at the MHC has yielded conflicting results. We detected no variability in either class I- or class II-linked microsatellite markers. Three of the primers amplified duplicated loci, all of which were monomorphic. Sequencing of class I exon 2 alleles in 8 Gir lions revealed only 4 classical class I exon 2 alleles. Six of the 8 lions were identical in their allele complement and the remaining 2 lions’ alleles differed by only 1 or 2 amino acid substitutions. Although we cannot rule out that there is additional diversity in exon 3 of Gir lion MHC alleles, these data strongly support that there is critically low variability in the MHC of captive Gir lions. This is consistent with the findings of several previous studies including those into allozyme diversity (Yuhki and O’Brien 1990), minisatellite DNA fingerprints (Gilbert et al. 1991), and RFLP of MHC class I loci (Yuhki and O’Brien 1990), all of which revealed no genetic polymorphism. In contrast to these results, Sachdev et al. (2005) used cloning and sequencing to compare class I sequence polymorphism and concluded that there was a high level of MHC class I diversity in wild Gir lions. However, this study did not use a conservative approach in their methods. A requirement used in most MHC sequence diversity studies is that alleles must appear in 2 independent PCRs from the same or different individuals in order to be considered a unique allele (e.g., Pokorny et al. 2010; Castro-Prieto et al. 2011). Although we used this approach in the current study, Sachdev and colleagues did not and this possibly resulted in an overestimation of allele diversity. Additionally, these authors did not use a proofreading polymerase, which may have resulted in a greater number of PCR-induced substitutions. These issues could account for the discrepancy between their results and ours, though as the current study only investigated diversity in small sample set, additional testing with a larger sample set will be needed to definitively determine MHC diversity in the Gir lion population.

While previous studies have shown that MHC diversity can be maintained despite the loss of overall genetic diversity in a bottlenecked population (Aguilar et al. 2004), this does not appear to be the case in the Gir lion. Our findings suggest that captive Gir lions have critically low MHC diversity. Alternative sites, outside the Gir sanctuary, are currently being sought for reintroduction of the Gir lion into the wild (Johnsingh et al. 2007). This project should take into account the MHC diversity of the population to be reintroduced in order to maximize their immunological fitness. These results highlight the importance of accurately determining the degree of variation in the MHC of captive and wild Gir lions and maintaining polymorphism in the captive breeding program in order to ensure long-term viability.

MHC Diversity in the Cheetah

In the past, the cheetah has been used as a classic example of a population with low MHC diversity (O’Brien et al. 1985; Drake et al. 2004). However, the results of the recent sequencing of 150 wild Namibian cheetahs has challenged this paradigm (Castro-Prieto et al. 2011), demonstrating a higher MHC class I diversity than previously recognized. Our findings also support greater MHC polymorphism in cheetahs with an average of 4.33 alleles per microsatellite marker in just 13 cheetahs and with an expected heterozygosity of 0.66. MHC sequencing in 8 cheetah samples revealed 7 classical MHC class I alpha I domain alleles, which are likely to be expressed and functional. This was equal to the 7 alleles that were unique in the alpha I domain identified by Castro-Prieto et al. (2011). Our sequencing results correlate well with the findings of Castro-Prieto et al. (2011), and this suggests that even in captivity, a moderate MHC class I diversity has been maintained.

The results of the MHC microsatellite typing in the cheetah and lion highlight the utility of these MHC-linked markers in exploring MHC polymorphism in different Felidae species. Most of the markers developed in the domestic cat were conserved in the cheetah and lion. The results of the microsatellite typing correlated well with that of the MHC cloning and sequencing method. This demonstrates that MHC-linked microsatellites may be a viable alternative to cloning and sequencing as a more rapid and cost-effective method of determining MHC diversity in a range of species.

Future Directions

The MHC-linked markers developed in this study will be useful in examining and comparing MHC diversity in different breeds and populations of domestic cats. Furthermore, as most of the microsatellites developed in this study are conserved in distantly related Felidae species, these markers may be used to study MHC diversity in other wild Felidae species.

The markers developed in this study may enable future research into associations between MHC alleles and diseases of domestic cats. To date, a single study investigating the association between MHC loci and disease in cats has been published. Addie et al. (2004) found no association between DRB alleles and feline infectious peritonitis using reference strand–mediated conformation analysis. Although the MHC-linked microsatellite developed in this study cannot be used to identify individual haplotypes due to the low number of markers and the extreme diversity of the MHC region, they may offer a low cost method of screening multiple MHC loci to identify disease associations and may be used alone or in conjunction with MHC sequencing to identify susceptible or resistant MHC types as has been done previously (Paterson et al. 1998; Andersson et al. 2012).

Funding

Australian Companion Animal Health Foundation ; Intramural Research Program of the National Institutes of Health; National Cancer Institute, Center for Cancer Research (contract N01-CO-12400). Russian Ministry of Science (11.G34.31.0068 to S.J.O., Principal Investigator).