Population genetics and natural selection in rheumatic disease

Synopsis

Human genetic diversity is the result of population genetic forces. This genetic variation influences disease risk and contributes to health disparities. Natural selection is an important influence on human genetic variation. Since immune and inflammatory function genes are enriched for signals of positive selection, the prevalence of rheumatic disease risk alleles seen in different populations is partially the result of differing selective pressures (e.g., due to pathogens). This review summarizes the genetic regions associated with susceptibility to different rheumatic diseases and concomitant evidence for natural selection, including known agents of selection exerting selective pressure in these regions. Integrating rheumatic disease susceptibility studies with population genetics to investigate how natural selection has contributed to genetic variation that influences disease risk will help identify functional variants and elucidate biological mechanisms.

Introduction

Rheumatic diseases are a family of more than 100 chronic, and often disabling, illnesses characterized by inflammation and loss of function, especially in the joints, tendons, ligaments, bones, and muscles. They collectively affect over 20% of US adults, with osteoarthritis, rheumatoid arthritis, spondylarthritides, gout and fibromyalgia being the most prevalent.^1,2 Patients often endure lifelong debilitating symptoms, reduced productivity at work, and high medical expenses. Arthritis and related illnesses, as well as back or spine problems are major causes of disability.³ Importantly, since many rheumatic diseases present before or during a woman's reproductive years, they can have effects on fetal and maternal outcomes,⁴ such as pregnancy loss in women with systemic lupus erythematosus^4,5 and vasculitis,⁴ and infertility in women with rheumatoid arthritis.⁵

Most rheumatic diseases exhibit marked gender and ethnic disparities. Most predominately afflict women (e.g. rheumatoid arthritis, systemic lupus erythematosus, systemic sclerosis, fibromyalgia), but spondyloarthropathies and gout are more common in men.⁶ African Americans are at higher risk than European Americans for systemic lupus erythematosus and systemic sclerosis, which they tend to develop earlier in life and experience more severe disease.⁷ Despite the variation in prevalence, incidence and disease severity that are known to vary among ethnic groups, little is known about the genetic etiology of these diseases in the different populations and the reasons for the ethnic disparities remain elusive.

Left untreated, most rheumatic diseases can affect the ability to raise offspring that successfully reproduce and result in reduced reproductive fitness. Thus, alternative forces must exist that permit the relative high frequency of risk alleles. Since immune and inflammatory responses can be highly sensitive to environmental change,⁸ evolutionary adaptation to specific environments might have driven selection on immune-related genetic variants, impacting variant frequencies and leaving signatures of selection in the genome. Given that infectious organisms are strong agents of natural selection,^9,10 it is plausible that alleles selected for protection against infection confer increased risk of autoimmune and inflammatory diseases, as the “hygiene hypothesis”¹¹ postulates. It is thought that the adaptation to pathogen pressure through functional variation in immune-related genes conferred a specific selective advantage for host survival, including protection from pathogens and tolerance to microbiota.¹² However, the emergence of such variation conferring resistance to pathogens is also influencing immune and inflammatory disease risk in specific populations.

In the past decade, multiple genome scans for signatures of selection on common variation have identified many immune-related loci.^13-17 Similarly, 90 genome-wide association studies (GWAS) (Table 1) have established rheumatic disease-associated alleles. There is also growing evidence that autoimmune and inflammatory disease-associated variants are under selection.^17-21 This review expands on our previous work²² and summarizes the evidence for rheumatic disease-associated loci under selection and the candidate selective pressures. Given that genomic variation can have clinically important consequences,²³ elucidating the patterns of variation and the functional role of the selective pressure might contribute to a better understanding of disease etiology and the development of new therapies for improved disease management.

Table 1.

Rheumatic diseases with published GWAS and respective number of associated loci.

Rheumatic diseases	Number of
	GWAS	loci
ANCA-associated vasculitis (AAV)	1	18
Ankylosing spondylitis (AS)	3	21
Behçet's disease (BD)	5	9
Dermatomyositis (DM)	1	1
Gout	4	14
Granulomatosis with polyangiitis (GPA)	1	6
Juvenile idiopathic arthritis (JIA)	3	6
Kawasaki disease (KD)	6	16
Osteoarthritis (OA)	9	16
Osteoporosis (OP)	3	3
Paget's disease (PD)	2	9
Psoriasis (PS)	11	60
Psoriatic arthritis (PsA)	2	4
Rheumatoid arthritis (RA)	19	129
Sjögren's syndrome (SS)	1	4
Systemic lupus erythematosus (SLE)	16	124
Systemic sclerosis (SScl)	3	10

Numbers compiled from the NHGRI-EBI Catalog of Published Genome-Wide Association Studies (https://www.ebi.ac.uk/gwas) accessed on October 24^th, 2016.⁴⁶

Shared genetic etiology in rheumatic diseases

The family of rheumatic diseases is remarkable for its heterogeneity and similar underlying mechanisms. The genetic heritability of rheumatic diseases is extremely variable, ranging from very high in ankylosing spondylitis to almost negligible in systemic sclerosis.²⁴ GWAS have proved particularly powerful for autoimmune diseases,²⁵ including many autoimmune rheumatic diseases, which might be due to their immune and inflammatory genetic etiology. Table 1 summarizes the rheumatic diseases with published GWAS and the number of disease-associated loci uncovered from these GWAS. The common genetic etiology is exemplified by the sharing of associated loci among rheumatic diseases, such as the Human Leukocyte Antigen (HLA), STAT4, TNIP1, TNFAIP3, and BLK.²⁶ This sharing of risk loci is greater among the groups of diseases characterized by the presence of particular serum autoantibodies (seropositive; such as rheumatoid arthritis, systemic lupus erythematosus, etc.) than it is between the seropositive and seronegative diseases (those typically characterized as not having associated serum autoantibodies.²⁶ This supports the consensus that there is a common genetic background predisposing to autoimmunity and inflammation, and that further combinations of more serologically defined and disease specific variation at HLA and non-HLA genes, in interaction with epigenetic and environmental factors, contribute to disease and its clinical manifestations. It has been suggested that different population genetic factors (e.g., natural selection with coevolution with pathogens, random mutation, isolations, migrations and interbreeding) in similar or distinct environments led to the establishment of the current plethora of loci that predispose to autoimmunity.²⁷ It is thus plausible that population-level phenomena are a reason behind the complexity of gene effects in different autoimmune and rheumatic diseases.

Population genetics, natural selection and adaptation

The genetic basis of disease is influenced by individual and population variation. Population- level phenomena such as mutation, migration, genetic drift and natural selection, have left an imprint on genetic variation that is likely to influence phenotypic expression in specific populations.²³ Given its role in driving genetic variation, population genetics can help elucidate human genetic diversity and, consequently, disease etiology.

Natural selection is the process by which a trait becomes either more or less common in a population depending on the differential reproductive success of those with the trait. Natural selection drives adaptation, the evolutionary process whereby over generations the members of a population become better suited to survive and reproduce in that environment. Negative (or purifying) selection is the most common mechanism of selection, usually associated with rare Mendelian disorders. Positive selection increases the prevalence of adaptive traits by increasing the frequency of favorable alleles and is often associated with common complex traits.²⁸ The enrichment for signals of positive selection among genes associated with complex traits is well documented.^14,29-31 Balancing selection favors genetic diversity by retaining variation in the population as a result of heterozygote advantage and frequency-dependent advantage. Despite rarer, a pertinent example is the HLA (also known as major histocompatibility complex (MHC)) region,^32,33 where highly polymorphic loci play a central role in the recognition and presentation of antigens to the immune system. The high levels of polymorphism are the results of pathogen-driven balancing selection.³⁴ The heterozygote advantage against multiple pathogens contributes to the evolution of HLA diversity, which in turn confers resistance against multiple pathogens and explains the persistence of alleles conferring susceptibility to disease.³⁵ Nevertheless, there is also recent evidence that positive selection might be acting on specific HLA alleles in a local population due to unique environmental pressures.³⁶

Natural selection leaves a distinctive molecular signature in the targeted genomic region, and different statistical methods have been developed to detect signatures of selection.¹² It has been hypothesized by Klironomos and colleagues that, in addition to genetic (sequence) variation, heritable epigenetic modifications can affect rates of fitness increase, as well as patterns of genotypic and phenotypic change during adaptation.³⁷ However, the role of epigenetic variation in the response to natural selection has not been formally assessed, as the methodology to test signatures of natural selection on epigenetic variation is just emerging.³⁸

Natural selection in rheumatic disease

Given that, if untreated, rheumatic diseases can diminish reproductive potential and impair the ability to raise offspring that successfully reproduce, some evolutionary process must sustain the relative high frequency of risk alleles seen in current populations around the world. Since the human genome is shaped by adaptation to environmental pressures at the population level, one plausible reason for the higher frequency of disease-risk alleles may be the direct effect of population-specific natural selection. This hypothesis is supported by the experimental evidence for MHC heterozygote superiority against multiple pathogens, a mechanism that would contribute to the evolution of HLA diversity and explain the persistence of alleles conferring susceptibility to disease.³⁵

There is compelling evidence that natural selection is acting on a significant fraction of the human genome.^15,39-43 Immune function genes and pathways are consistently reported in tests for natural selection. As a result of several genome-wide scans, over 300 immune-related genes have been suggested as putative targets of positive selection.^13-17 Although the challenge in validating the true signals remains,⁴⁴ several genes involved in immune-related functions have been shown to be under selection.^20,45

A total of 61 regions with evidence for selection and association with at least one rheumatic disease are shown in Table 2. This table includes 35 regions previously reported as being under selection in the literature,²² plus rheumatic disease-associated loci from current GWAS (in Table 1) and evidence of recent positive selection from Hapmap phase II data.¹⁵ Specifically, a region published in the GWAS Catalog⁴⁶ as associated with a rheumatic disease was considered as exhibiting evidence for natural selection if it contained at least two SNPs within 200 kb with an absolute integrated Haplotype Score (iHS) value in the top 0.1% of the genome-wide distribution in one population (Asian, European or African). A total of 39 regions that met these criteria are included in Table 2, 14 of which were previously reported. These 39 regions with evidence for selection represent about 10% of all regions associated with a rheumatic disease in a GWAS: 13% for systemic lupus erythematosus (SLE), 9% for rheumatoid arthritis (RA), 7% for psoriasis (PS), and 5% for ankylosing spondylitis (AS). This fraction of disease-associated loci with concomitant evidence for selection is higher than previous reports focusing on SNPs instead of regions. Notably, when using the top 1% of iHS variants, Raj and colleagues²¹ reported that inflammatory diseases (which included AS, RA and SLE) have 5% of SNPs targeted by positive selection. Limiting comparisons to SNPs instead of regions might miss regions with both evidence for disease association and selection at different SNPs. The numbers of GWAS-associated loci, including those with and without concomitant evidence for recent positive selection, are illustrated in Figure 1. Among all regions in Table 2, a higher number of signals of selection were found in European (36%), followed by Asian (32%) and African (32%) populations. This is consistent with previous reports of enrichment of inflammatory-disease SNPs targeted by positive selection in subjects of European ancestry.^21,47

Table 2.

Rheumatic disease regions with evidence for selection and implicated agents of selection.

Gene region	Position	Rheumatic disease association	References for evidence of natural selection	Population	Selective pressure	References for pathogen-driven selection
TNFRSF14, MMEL1*	1p36.32	RA		YRI
IL23R	1p31.3	AS	18		protozoa	53
MAGI3, PTPN22*	1p13.2	RA, SLE	19,20	YRI	protozoa	53
FCGR2B	1q23.3	SLE	57		Plasmodium falciparum	57
TNFSF4	1q25.1	RA, SS, SLE	19
NCF2, RGL1*	1q25.3	SLE		ASI
CR1	1q32	SLE	65		Plasmodium falciparum	65
TLR5	1q41-q42	SLE	20	YRI	Salmonella enterica ser. Typhimurium and other exposures	20
PELI1*	2p14	KD		ASI
ALMS1P, DGUOK*	2p13.1	SLE	19	CEU
PARD3B*	2q33.3	OA	20	CEU
CNTN6*	3p26.3	SLE		ASI
XCR1, CCR3*	3p21.31	BD		YRI
CCDC66, ARHGEF3	3p14.3	RA	20	YRI
BTLA	3q13.2	RA		ASI
ARHGAP31, CD80	3q13.33	JIA, SLE	21	YRI
MRPS22*	3q23	KD	20	ASI
SLC2A9*	4p16.1	Gout		YRI
KCNIP4*	4p15.2	RA	20	CEU, YRI
TECRL*	4q13.1	KD		CEU
ANTRX2	4q21	AS	21
IL2, IL21*	4q27	RA	21,47	YRI
Intergenic*	4q28.3	SLE		ASI
PTGER4	5p13.1	AS	21,53		protozoa	53
COMMD10, SEMA6A*	5q23.1	GPA		ASI, CEU
ALDH7A1*	5q23.2	OP		CEU
TNIP1	5q33.1	SLE, SScl, PsA	19
PTTG1	5q33.3	SLE	19
IRF4*	6p25.3	RA		CEU
ITPR3	6p21.31	SLE	20	YRI
HLA*	6p22.1-6p21.31	AAV, AS, BD, GPA, JIA, KD, OA, PS, PsA, RA, SS, SLE, SScl	20,21,66-69	ASI, CEU, YRI	bacterial infection	34,54,55
SNRPC, UHRF1BP1*	6p21.31	SLE	19-21	CEU	Mycobacterium tuberculosis	70
VARS, LSM2	6p21	SLE	21
CCDC167, MIR4462*	6p21.2	SLE		YRI
PRDM1, ATG5*	6q21	RA, SLE		YRI
TRAF3IP2*	6q21	PS, PsA		YRI
IKZF1	7p12.2	SLE	19
GTF2I*	7q11.23	SS	20	ASI
HIP1*	7q11.23	SLE	21	YRI
LSMEM1, NPM1P14*	7q31.1	OA		ASI
XKR6, BLK*	8p23.1	KD, RA, SS, SLE, SSc	19,20	ASI
GRHL2*	8q22.3	RA		CEU
KDM4C*	9p24.1	SLE		ASI
NTNG2, SETX*	9q34.13	SLE	20	CEU
FAM171A1*	10p13	SLE		ASI
CTNNA3*	10q21.3	PsA		CEU
CD5	11q12.2	RA	71
GRM5*	11q14.3	RA		CEU
OSBPL8*	12q21.2	SLE		CEU
SH2B3, NAA25	12q24.12-q24.13	RA	21,56		bacterial infection	56
KIAA0391*	14q23.1	PS		ASI
PRKCH, HIF1A*	14q23.1	RA	20	CEU, YRI
CLEC16A, CIITA	16p13.13	RA, SLE	19,21,72
ITGAM, ITGAX	16p11.2	SLE	19,20
PRSS54*	16q21	SLE		YRI
WWOX	16q23.2	OA	20	CEU
IRF8	16q24.1	RA, SScl	73
RABEP1, NUP88*	17p13.2	RA		CEU
BCAS3, NACA2*	17q23.2	Gout, OA	20	ASI
TYK2	19p13.2	RA, SLE	53		protozoa	53
PAK7*	20p12.2	PS		YRI

Rheumatic disease associations were reported in the literature (column “references for evidence of natural selection”), and/or in the NHGRI-EBI GWAS Catalog accessed on October 24^th, 2016. In addition to the disease-associated regions with evidence for selection reported in the literature, rheumatic disease-associated loci from the GWAS catalog with evidence of recent positive selection from HapMap phase II data (assessed by the presence of at least two SNPs within approximately 200 kb with an absolute iHS value in the top 0.1% of the genome-wide distribution in one population) are also included and denoted by the asterisk. See Table 1 for disease abbreviations. ASI: Asian, CEU: European, YRI: African populations.

Figure 1.

Number of rheumatic disease GWAS-associated loci, including those with concomitant evidence for recent positive selection (dark shaded area). See Table 1 for disease abbreviations.

Agents of selection

The wide variety of environments inhabited by human populations is likely exerting different selective pressures that lead to adaptation through natural selection. Climatic factors such as altitude, latitude, ultra-violet radiation levels, temperature, as well as diet and pathogens have been reported as agents of selection driving adaptations to these environments and lifestyles. As recently reviewed,²² some relevant examples include signals of natural selection driven by annual photoperiod variation reported for restless leg syndrome risk variants,⁴⁸ correlation between climate variables and SNPs involved in immune response, as well as pathways related to UV radiation, infection and immunity, and cancer,⁴⁹ and correlations between worldwide migration trajectories and variants associated with, among other, systemic lupus erythematosus and systemic sclerosis.⁵⁰ Interestingly, expression QTLs (eQTLs) (see article by Laufer et al elsewhere in this issue for definition) from immune function and metabolism genes are enriched in signals of environmental adaptation,⁵¹ which highlights the importance of regulatory variations in local adaptation.

Nevertheless, the strongest effect of climate is in shaping the spatial pattern and species diversity of human pathogens,⁹ which is directly relevant to immune and inflammatory disease predisposition. As recently reviewed,⁵² in the constant co-evolutionary battle between host and pathogen, pathogens that diminish reproductive potential, either through death or poor health, drive selection on genetic variants that affect pathogen resistance. As Hancock suggested,⁴⁹ it is likely that selection signals in immune-related loci may implicate variants evolving under a model of antagonistic pleiotropy, where the selective pressure was pathogen resistance, and the inflammatory disorder is a pleiotropic consequence of the resistance allele. This could hence be a mechanism explaining the prevalence of immune risk alleles that are common in the population.

Indeed, pathogens have been the main selective pressure through human evolution.¹⁰ In an analysis that included climate, diet regimes, and pathogen loads, Fumagalli and colleagues¹⁰ showed that the diversity of the local pathogenic environment is the predominant driver of local adaptation, and that climate conditions only played a relatively minor role. In addition, they reported an enrichment of genes associated to SLE, RA and AS, which supports the hypothesis that some susceptibility alleles for rheumatic diseases may be maintained in human population due to past selective processes.¹⁰ The enrichment for signals of positive selection in inflammatory-disease susceptibility loci has been recently corroborated.²¹ Reviews of selection signatures left by pathogen-exerted pressure, including immune-related genes, can be found elsewhere.^52,53

Genetic regions associated with susceptibility to different rheumatic diseases and evidence of selection that has been attributed to host-pathogen coevolution are shown in Table 2. In a fraction of the regions with evidence for selection and disease-association, known pathogens have been implicated as the selective pressure. Variation in the HLA and SH2B3 has been reported as a protective factor against bacterial infection.^34,54-56 Resistance to protozoa and tuberculosis infection have been implicated as the selective pressures for PTPN22 and UHRF1BP1, respectively. Interestingly, the SLE susceptibility allele in UHRF1BP1 is associated with decreased UHRF1BP1 RNA expression in different cell subsets, suggesting that the disease risk allele under selection has a regulatory effect.²¹ In the context of SLE predisposing loci, Clatworthy et al.⁵⁷ has shown that FCGR2B is important in controlling the immune response to Plasmodium falciparum, the parasite responsible for the most severe form of malaria, and suggests that the higher frequency of human FCGR2B polymorphisms predisposing to SLE in Asians and Africans may be maintained because these variants reduce susceptibility to malaria. Grossman et al.²⁰ implicated Salmonella typhimurium and other exposures that directionally drive selection of the toll-like receptor 5 (TLR5) gene,⁵⁸ which is involved in recognition of flagellated bacteria. Unlike endosomal TLRs, such as TLR8 and TLR8, that have been subject to purifying selection, cell-surface TLRs involved in pathogen recognition experienced more relaxed constraints.⁵⁹ The non-synonymous variant in PTPN22 shows complex signatures of selection, increasing the risk of SLE, RA, and other autoimmune diseases, but being protective against Crohn's disease.⁶⁰ Karlsson et al.⁶¹ have recently reported that cholera has exerted strong selective pressure on pro-inflammatory pathways. Despite the modest number of examples that offer clear functional hypotheses (e.g. SH2B3, TRL5), collectively this list supports the hypothesis that the increased prevalence of rheumatic disease may result, at least partially, from past events of selection that increased host resistance to infection.⁶²

Discussion

This review summarizes the genetic regions associated with susceptibility to different rheumatic diseases and concomitant evidence for selection, including known agents of selection exerting selective pressure in these regions. Uncovering these rheumatic disease-associated loci under selection underscores the importance of population genetics and how the understanding of human genetic diversity is crucial to understanding disease etiology or treatment response at both the population and individual levels.

A combination of population-level phenomena, including possibly bottlenecks, migration, admixture, natural selection, and random genetic drift, are likely contributors to this complexity of gene effects in different rheumatic diseases. Given the complex history of selective pressures acting on humans, unequal selective pressures and a diverse spectrum of plausible evolutionary models are expected to be exerted on susceptibility loci for rheumatic diseases.²⁸ It is likely that several pathogens have exerted pressure on the same loci and that selection can vary in form, intensity, time and space, which is consistent with the observation that both risk and protective alleles for rheumatic diseases increased in frequency due to selection.¹⁷ For most regions, the exact selective pressure leaving the signature of selection is unclear. Clearly, these signatures are not necessarily the result of adaptation, but might be a consequence of random genetic drift. In any case, regardless of the population phenomenon shaping current human genetic diversity, this genetic variation is the basis clinically relevant traits at both the individual and population levels.²³

An important next step to delineate the selective advantage conferred by these rheumatic disease risk variants are functional studies using in vitro experiments and model organisms to identify the underlying functional variants and quantify the phenotypic consequences of the candidate adaptive alleles. Human- pathogen coevolution is ongoing and, despite the emergence of new pathogens (e.g. HIV), potential pathogens driving these host-specific adaptations are expected to have long-standing relationships with humans, including those that cause malaria, smallpox, cholera, tuberculosis and leprosy,⁶³ as well as the human microbiome.⁶⁴ Regardless of the agent of selection and the reasons for the emergence of both common and rare rheumatic disease-causing alleles, incorporating population genetics to understand human genetic diversity will lead to a better understanding of the causes of health disparities, identification of functional variants and discovery of cellular mechanisms, and contribute to the development of new therapies.

Key Points.

• If untreated, rheumatic diseases can diminish reproductive potential and impair the ability to raise offspring that successfully reproduce. Thus, it is likely that the frequency of disease-risk alleles seen in populations around the world is influenced by population-specific natural selection.

• Both autoimmune and non-autoimmune rheumatic disorders show genetic associations in regions with signatures of selection.

• The prevalence of rheumatic disease may result, at least partially, from past events of selection that increased host resistance to infection.

• Many of the complexities of gene effects in different rheumatic diseases can be explained by population genetics phenomena.