Genetic modifiers as relevant biological variables of eye disorders

Abstract

From early in the study of mammalian genetics, it was clear that modifiers can have a striking influence on phenotypes. Today, several modifiers have now been studied in enough detail to allow a glimpse of how they function and influence our perspective of disease. With respect to diseases of the eye, some modifiers are an important source of phenotypic variation that can elucidate how genes function in networks to collectively shape ocular anatomy and physiology, thus influencing our understanding of basic biology. Other modifiers represent an opportunity for new therapeutic targets, whose manipulation could be used to mitigate ophthalmic disease. Here, we review progress in the study of genetic modifiers of eye disorders, with examples from mice and humans that together illustrate the ubiquitous nature of genetic modifiers and why they are relevant biological variables in experimental design. Special emphasis is given to ophthalmic modifiers in mice, especially those relevant to selection of genetic background and those that might inadvertently be a source of experimental variability. These modifiers are capable of influencing interpretations of many experiments using targeted genome manipulations such as knockouts or transgenics. Whereas there are fewer examples of modifiers of eye disorders in humans with a molecular identification, there is ample evidence that they exist and should be considered as a relevant biological variable in human genetic studies as well.

Introduction

Modifier genes (‘modifiers’) are non-allelic genetic variations that shape genotype–phenotype relationships. Implicit in discussions of modifiers is a long-standing central theme that most, probably all, phenotypes are the outcome of multi-genic networks (1–3). Several ophthalmic modifiers have now been described in enough detail to allow a glimpse of how these networks function and influence our perspective of eye disease. Here, we review progress in the study of modifiers, with an emphasis on examples from mice and humans that together illustrate the ubiquitous nature of genetic modifiers and why they are relevant biological variables in experimental design.

Modifiers have previously been discussed using multiple frameworks; some focusing on binary systems in which modifiers alter the phenotype of a target gene (4) while others separately consider secondary modifiers (directly linked molecular functions) versus tertiary modifiers (altering phenotype by any other process) (5). Several authors point out that categorization is problematic for modifier systems, which tend to display continuous, not hierarchical, organization (6,7). Cognizant of these challenges, we settle on a broad definition of genetic modifiers that includes traditional modifiers, whose functions are primarily evident by their influence on a second-site mutation, and modifiers that act as quantitative trait loci (QTL) and that may have independent phenotypes.

Brief History

From early in the study of mammalian genetics, it was clear that modifiers can have a striking influence on phenotypes (8,9), including those of the eye (10). Likewise, the action of modifiers in human ophthalmic genetics has also long been appreciated, likely accounting for differences in clinical presentation, age of onset, and penetrance for many familial eye disorders (11–15). However, it wasn’t until later, when tools for mouse genetics were developed, that our knowledge of the molecular underpinnings of modifiers began to take shape.

C.C. Little is credited with the idea of inbreeding mice to genetic homogeneity so that modifiers are no longer segregating, yielding increased phenotypic predictability (16). Although mice within an inbred strain are isogenic and tend to have very similar phenotypes between individuals, inbred mice are still sensitive to environmental and stochastic influences that can cause variability. For example, C57BL-related mouse strains have a propensity to develop sporadic ocular abnormalities at a rate of 4–10%, including microphthalmia, anophthalmia, cataracts and corneal opacities (17). Nonetheless, the advantages of inbred mice led to a large proliferation of mouse strains that continues today. Inbred strains are not a perfect solution to every conceivable approach, and as we describe below, choosing an appropriate genetic background requires many considerations involving genetic modifiers that will differ according to the experimental goals.

Pathways to Discovery

As with all forms of phenotype-driven genetics, identification of modifiers represents an opportunity to discover new molecular pathways contributing to disease, including those that may not have been considered otherwise. Many projects seeking to identify modifiers in mice stem from an intentional interbreeding of strains and the observation of background-dependent variable expressivity. In these instances, traditional genetic mapping approaches can allow identification and characterization of modifiers (18).

This traditional phenotype-driven approach is exemplified by studies of the Nr2e3^rd7 mutation. The rd7 mutation was first recognized by its retinal phenotype, presumably due to a spontaneous mutation that became fixed in the ‘77-2C2A-special’ strain of inbred mice at The Jackson Laboratory. The mutation, later mapped to the Nr2e3 gene, causes progressive photoreceptor degeneration and electroretinogram deficits (19) through a mechanism of abnormal cone cell differentiation (20). In humans, mutations in NR2E3 cause enhanced S-cone syndrome (21) and retinitis pigmentosa. In an approach deliberately designed to identify genetic modifiers of Nr2e3^rd7, a C57BL/6J (B6) congenic strain was generated and utilized in intercrosses with AKR/J and CAST/EiJ mice. The use of indirect ophthalmoscopy to score disease severity in resulting progeny, coupled with a genome-wide scan and QTL analysis, identified two significant suppressor loci on chromosomes 8 and 19 derived from CAST/EiJ and two suggestive loci on chromosomes 7 and 11 from AKR/J (22). Initially focusing on an AKR/J-derived locus, physical mapping refined the chromosome 11 modifier locus to a 3.3 cM region containing ∼200 genes, approximately half of which are expressed in the retina. Among these, candidates were prioritized by their known functions, leading to a focused consideration of Nrd1d1 (23). Sequencing identified two putative Nrd1d1 mutations between B6 and AKR/J, and importantly, in vivo electroporation of Nrd1d1 transcripts into the retina of neonatal Nr2e3^rd7/rd7 mice rescued the rd7 phenotype (24). Thus, Nrd1d1 is likely the chromosome 11 modifier.

These studies of Nr2e3 and one of its modifiers, Nrd1d1, have both clinical and basic science implications. First, over-expression of Nrd1d1 suppresses Nr2e3-mediated retinal pathology, suggesting that similar manipulations could be used as a form of gene therapy to complement, or be used in place of, therapeutic manipulations to Nr2e3 itself. From a more basic perspective, Nr2e3 and Nrd1d1 both encode nuclear hormone receptors that directly bind one another, an interaction hypothesized to be important for transcriptional regulation of genes involved in photoreceptor homeostasis and function, such as Nrl and Crx (23). Together, these studies of Nr2e3^rd7 illustrate the power of phenotype-driven approaches to identify genetic modifiers.

The influence of modifiers in mice can also become apparent ‘accidentally’ through routine husbandry. For example, MERTK mutations cause retinitis pigmentosa in humans. To model this disease, knock-out mice (Mertk^tm1Gkm) were generated on a B6/129 mixed genetic background. On the original mixed background, mice were described as having near-complete photoreceptor degeneration by post-natal day 45 (25). However, backcrossing onto the B6 background for six generations suppressed the retinal degeneration phenotype. Genotyping suggested a modifier locus on chromosome 2 and a backcross narrowed the critical region to 2.1 Mb containing 53 known genes, including Tyro3, a paralog of Mertk. To initially assess Tyro3 as a candidate, qPCR of Tyro3 in the retinal pigment epithelium was performed, finding that comparative transcript levels were Tyro3^B6/B6 > Tyro3^B6/129 > Tyro3^129/129. Thus, a hypothesis that high Tyro3 expression levels in B6 mice could compensate for the absence of Mertk function was suggested. To functionally test the ability of these genes to compensate for one another, Mertk knock-out mice were crossed with an independent strain of Tyro3 knock-out mice. Accelerated photoreceptor degeneration observed in Mertk/Tyro3 double knock-out mice is consistent with a conclusion that Tyro3 is a modifier of Mertk. Overall, this work again illustrates the utility of modifier studies to generate novel insight into the multi-genic pathways associated with eye disease.

In similar fashion to the studies of retinal disease modifiers described above, other studies of eye diseases in mice have detected genetic modifiers that are active across a broad range of ocular tissues (26–29) and ages (30,31). Some known modifiers (summarized in Table 1) are limited to rarely utilized strains or specific molecular pathways; others are widely present and broad acting (32). Although there is undoubtedly an inclination for reporting only positive results, the literature largely supports the old presumption that modifiers are relevant biological variables for essentially every gene. A statement made by Dr Sewall Wright in 1931 remains difficult to refute: ‘As genetic studies continued, ever smaller differences were found to mendelize, and any character, sufficiently investigated, turned out to be affected by many factors.’ (3)

Table 1.

Modifiers of ocular phenotypes in mice

Mouse allelea	Human ocular diseaseb	Modifierc	Modifier effect on phenotype
1. Retinal phenotypes
A;Cg-Rp1tm1Eap	Retinitis pigmentosa 1	Genetic background: C57BL/6J; DBA/1J	Suppressed photoreceptor degeneration (118)
B6.Cg-Nr2e3rd7	Enhanced S cone syndrome; Retinitis pigmentosa 37	Genetic background: AKR/J; CAST/EiJ; NOD.NON-H2nb1/LtJ	Reduced penetrance of retinal spotting (22)
		Chr8 QTL; Chr19 QTL	CAST/EiJ alleles suppress retinal spotting (22)
		Mor7 Chr11 QTL; Nr1d1	AKR/J alleles suppress retinal spotting (24)
B6;C3H-Rs1tmgc1	X-linked juvenile retinoschisis	Genetic background: AKR/J	Variability in retinoschisis severity (119)
		Mor1 Chr7 QTL; Tyrc-2J	AKR/J alleles suppress retinoschisis (48,119)
B6;129-Mertktm1Gkm	Retinitis pigmentosa 38	Genetic background: C57BL/6J	Suppressed photoreceptor degeneration; more severe phenotype in females (25)
		Tyro3b6	Suppressed photoreceptor degeneration; more severe phenotype in females (25)
B6;Cg-Prom1tm1.1(DTA)T°k°	Stargardt disease 4; Retinitis pigmentosa 41; Bull's eye macular dystrophy; Cone-rod dystrophy 12	Genetic background: CBA/NSlc	Suppressed photoreceptor degeneration (120)
B6.C3Ga-Mfrprd6	Microphthalmia, isolated 5; Nanophthalmos 2	Genetic background: CAST/EiJ	Variability in outer nuclear layer cell count (121)
		Rd6m1 Chr1 QTL	CAST/EiJ alleles suppress retinal degeneration (121)
B6;129-Tulp1tm1Pjn	Leber Congenital Amaurosis 15; Retinitis pigmentosa 14	Genetic background: AKR/J	Variability in outer nuclear layer thickness (122)
		Chr2 QTL; Mtap1a	AKR/J alleles suppress retinal degeneration (122)
B6-Tubtub	Retinal dystrophy and obesity	Genetic background: AKR/J	Variability in outer nuclear layer thickness (123)
		Motr1 Chr11 QTL	AKR/J alleles suppress photoreceptor degeneration (123)
RBF.Cg-Rd3rd3	Leber Congenital Amaurosis 12	Genetic background: Rb(11.13)4Bnr; In(5)30Rk	Suppressed photoreceptor degeneration (124)
BALB.Cg-Rd3rd3		Genetic background: B6.Cg-Tyrc-2J/J	Suppressed photoreceptor degeneration (125)
		Chr5 QTL	B6.Cg-Tyrc-2J/J alleles enhance photoreceptor degeneration (125)
		Chr6 QTL; Chr8 QTL; Chr14 QTL; Chr17 QTL	B6.Cg-Tyrc-2J/J alleles suppress photoreceptor degeneration (125)
B6.SJL-Tg(Rho)1Wbae	Autosomal dominant congenital stationary night blindness; Retinitis pigmentosa	Genetic background: B6.FVB	Albino offspring have more severe retinal degeneration than pigmented offspring (49)
B6.Cg-Crb1rd8	Cone-rod dystrophy; Leber congenital amaurosis; Retinitis pigmentosa	Genetic background: C57BL/6JOlaHsd	Variability in retinal autofluorescent lesion number and size (126)
		Chr15 QTL	C57BL/6 alleles suppress the number of autofluorescent retinal lesions (126)
B6N;129-Ccl2tm1R°l Cx3cr1tm1Zm Crb1rd8	CX3CR1: Age-related macular degeneration 12; CCL2: N/A	Crb1b6°lad	Suppressed autofluorescent retinal lesions initially attributed to Ccl2tm1R°l and Cx3cr1tm1Zm (127)
B6N;CBA-Tg(CD11c-eYFP) Crb1rd8		Crb1b6d	Altered reporter labeling and expression (128,129)
B6N;129-Grk1tm1Citb Crb1rd8	GRK1: Oguchi disease 2	Crb1b6d	Suppressed retinal disorganization and degeneration initially attributed to Grk1tm1Citb (130)
B6N;129-Hdctm1Nagy Crb1rd8	HDC: N/A	Crb1b6d	Rescued the retinal phenotype initially attributed to Hdctm1Nagy (131)
2. Lens phenotypes
C3H-Gpr161vl	N/A	Genetic background: MOLF/EiJ	Reduced penetrance of congenital cataract (132)
		Modvl3 Chr4 QTL	MOLF/EiJ alleles enhance penetrance of congenital cataract (132)
		Modvl4M°LF	Reduced penetrance of congenital cataract (31)
129.Cg-Gja8tm1Paul	Cataract 1	Genetic background: C57BL/6	Suppressed cataract (133)
BALB.Cg-Cpoxnct	Porphyria	Genetic background: MSM/M	Variable progression and density of cataract opacity (134)
		Chr3 QTL	MSM/M alleles are associated with a shift from diffuse to pin-head type cataract (134)
3. Anterior segment phenotypes
A.BY-Dstnc°rn1	N/A	Genetic background: C57BL/6J	Suppressed corneal neovascularization, epithelial hyperproliferation, and inflammation (29)
		Modn1 Chr3 QTL	A.BY-H2bcH2-T18f/SnJ alleles enhance corneal vascularization (135)
B6;Cg-Jag1tm1Grid	Alagille Syndrome	Genetic background: C3H/HeSnJ	Reduced penetrance of ocular abnormalities: pupil dysmorphologies, corneal opacity, microphthalmia (136,137)
B6-Lystbg-J	Chediak-Higashi Syndrome	Genetic background: DBA/2J	Enhanced iris disease (28)
D2.B6-Lystbg-J Tyrp1b		Tyrp1b6	Suppressed iris disease (28)
B6;Cg-Col4a1deltaex40	COL4A1-related brain small-vessel disease; Hereditary angiopathy with nephropathy, aneurysms, and muscle cramps (HANAC) syndrome	Genetic background: CAST/EiJ; 129S6/SvEvTac	Suppressed anterior segment dysgenesis and optic nerve hypoplasia (138)
D2-Tyrp1bGpnmbR150X	TYRP1: Oculocutaneous albinism; GPNMB: N/A	Genetic background: C57BL/6J	Suppressed high IOP and glaucoma; no apparent effect on pigment dispersing iris disease (50)
B6;129-Cyp1b1tm1G°nz	Early onset glaucoma; Peter’s anomaly	Tyrc	Enhanced ocular drainage structure abnormalities (30)
B6.129-Cyp1b1tm1G°nz		Tyrc-2J	Enhanced ocular drainage structure abnormalities (30)

N/A – not applicable; Chr – chromosome; QTL – quantitative trait locus; IOP – intraocular pressure.

^aPrimary mutation causing an ocular phenotype; genetic background was inferred from the cited publication and was not always clearly described.

^bDerived from Genetics Home Reference (139).

^cThe genetic background, significant QTL, or gene/allele with functional evidence for modifying the phenotype conferred by the allele in column one.

^dThe presence of this confounding allele was unknown at the outset of the experiment.

Source of Experimental Variability

The example of Mertk/Tyro3 also points to a potential confounder in many experiments—modifier alleles can easily enter into the background of mouse colonies. Although the group that performed the Mertk modifier study was purposely attentive to the possibility of background modifiers, it is a striking example of what can potentially happen if genetic background is overlooked. Many genetic manipulations in mice are performed with methods yielding hybrid backgrounds. For example, knockout strains are often generated using embryonic stem cells from 129-derived strains and mice are subsequently crossed to B6. Similarly, many transgenic strains are generated via pronuclear injection into fertilized B6SJLF1 eggs (33–35). Unless deliberate action is taken to create a congenic strain of mice, these strains are particularly prone to confounding effects of modifiers. The literature is replete with examples of studies in which the nature of the background is omitted or vague (36). In many such cases, the likelihood that genetic background could be skewing, or completely misleading, experimental interpretations is high. Technical advice on generating congenic strains is available in the literature (37,38); proper nomenclature guidelines for describing genetic background are on the MGI website (28).

Among the many potential modifiers that could unintentionally confound an experiment, several are particularly important to note (39). 1) Crb1^rd8. This mutation is present in C57BL/6N mice and all derivatives, including mice from the KOMP project (40). Mice homozygous for the rd8 mutation classically exhibit large white retinal spots and a slow progressive retinal degeneration. However, fundus examination is a relatively poor method to detect the mutation because the degree of spotting is highly variable. Among 66 Crb1^rd8 mutant strains identified at The Jackson Laboratory, only six had white retinal spots. 2) Pde6b^rd1. At least 99 strains available from The Jackson Laboratory and other distributors carry the Pde6b^rd1 mutation, including the popular inbred strains C3H/HeJ, CBA/J, FVB/NJ and SJL/J. Mice homozygous for the rd1 mutation typically exhibit severe early onset retinal degeneration. However, the age of onset varies according to background (41). 3) Tyr^c. Many mouse strains carry mutations influencing coat colour, including the albinism-causing c allele of the tyrosinase gene (42). Among many others, the Tyr^c mutation is present in the commonly used 129, A/J, AKR, BALB/c, FVB and SJL strains. Often overlooked as a mere cosmetic feature, albinism has profound influences on many aspects of the mouse eye (43–47) and is a known modifier of eye disease in several models (30,48–50). Similarly, other mutations affecting coat colour can cause (51–53) and/or modify ocular phenotypes (28,54,55). 4) Genes on the X-chromosome. Gender-specific effects are observed in various eye diseases; for most it is unclear whether this reflects sporadic chance related to small cohort sizes, secondary consequences of gender, or the influence of modifier genes located on the X-chromosome. Adding further possible phenotypic variability, the expressivity of any gene on the X-chromosome can be influenced by lyonization. For these reasons, it is indeed good practice to maintain gender balance between experimental and control cohorts of mice.

Although we lack space to broadly review studies of targeted mutations as genetic modifiers, there are several caveats to their use that are relevant to the topic of unintentional genetic modifications. Similar in nature to the introduction of modifiers via genetic background, targeted genetic manipulations can likewise introduce unintended effects, including: 1) Gene targeting can influence expression of non-targeted neighbouring genes (56). 2) Transgenes can be a source of insertional mutagenesis and may cause ocular phenotypes independent from the insert sequence (57,58). 3) Mini-gene inserts used to tag or regulate expression of constructs used in genome manipulations can cause independent effects (59–61). 4) Among multiple caveats of Cre systems (62), expression of Cre-drivers in the germ line can cause conditional alleles to become constitutive knockouts, resulting in phenotypic modification (63). In addition, some Cre-drivers can independently cause background-dependent ocular phenotypes (64). 5) Husbandry practices for strains with targeted mutations can promote modifying effects via genetic drift. For example, targeted mutations are sometimes maintained as a homozygous stock, eliminating the need to genotype offspring. A separately maintained inbred strain with a corresponding wild-type allele is typically used as a control in this scheme. Over time, independent spontaneous genetic variations are likely to occur that cause the experimental and control strains to genetically drift from one another. New mutations conferring a selective advantage to the homozygous stock are of particular concern. Thus, using crosses between heterozygotes or occasionally refreshing the mutant and corresponding control strains with newly obtained mice from an outside foundation colony is recommended to mitigate this possibility. 6) Each method of genome manipulation is likely prone to unique unintentional modifications. This point is particularly relevant to the emerging question of whether alleles generated by traditional targeted disruptions will yield the same phenotypes as alleles generated via CRISPR-Cas9—which has already caused some controversy (65,66).

In addition to modifier systems where the majority of observed phenotypic variability is attributed to a single gene, there are several disease relevant traits in mice that are polygenically influenced by QTL and should be presumed capable of manifesting as modifiers. For example, retinal neuron number (67), optic nerve axon number (68), corneal thickness (69), lens weight (70), retinal ganglion cell susceptibility to optic nerve crush (71), retinal susceptibility to light damage (72) and cataract severity (73) are all influenced by QTL. These loci are distributed throughout the genome and even small regions of heterogeneity segregating in a strain, including those in linkage disequilibrium with targeted mutations, are a potential source of variability. Importantly, similarly named mouse strains can have large amounts of genetic variability capable of harbouring such QTL. The similar sounding ‘C57BL/6J’ and ‘C57BL/6N’ strains could be, and have been (74), mispaired with one another. In fact, the two strains are not identical, having at least 236 genetic variants between them, including the Crb1^rd8 mutation and others causing key metabolic differences (36). Likewise, similar eye-relevant genetic differences exist in comparisons of C57BL/6NJ versus C57BL/6NHsd (75), and DBA/2J versus DBA/2NTac mice (76).

Human Modifiers

There is robust evidence that modifiers are also important in human eye diseases. Augmenting a wide range of reports of phenotypic variability within a single pedigree suggestive of modifiers, there are now similar observations among individuals sharing the same molecular lesion. As with mice, genetic modifiers of eye disease in humans are active across a broad range of tissues and ages (77–82), including retinal disease caused by mutations in NR2E3 (83) and MERTK (84). In the vast majority of human modifier studies, the molecular identity of the modifier remains unknown, though there are a few exceptions (85–90).

One example of a successful modifier study in human eye disease involves CNOT3 as a modifier of PRPF31. PRPF31 mutation causes autosomal dominant retinitis pigmentosa and, as observed for many ocular diseases (77–81,91–93), exhibits intrafamilial variable expressivity and non-penetrance, indicating the presence of modifiers (94–96). Linkage analysis of 56 individuals with a PRPF31 mutation from three pedigrees exhibiting reduced penetrance identified a modifier locus on chromosome 19q13.4 in close proximity and in trans to the PRPF31 mutant allele (97). A significant advantage for the identification of a modifier in this study came from the use of patient-derived cell lines. qPCR of ten candidate genes in immortalized lymphoblastoid cell lines (LCLs) from four non-penetrant and six penetrant individuals from a single family identified a statistically significant difference in CNOT3 expression, with higher expression in penetrant individuals. This pattern is inversely correlated to that observed for PRPF31, whose expression from the non-mutant allele in LCLs is significantly reduced in penetrant versus non-penetrant individuals (98,99). Sequencing of CNOT3 in a sib-pair with a discordant penetrance phenotype identified five genetic variants. Analysis of these variants in a second discordant sib-pair from the same pedigree identified only one SNP (rs4805718) segregating with the penetrance phenotype. This SNP was sequenced in additional members of the same pedigree and other individuals from a second unrelated family, and the C-allele was significantly associated with penetrance. Chromatin immunoprecipitation in LCLs showed that CNOT3 binds directly to the PRPF31 promoter, providing a feasible biological mechanism for the genetic observations. In an independent test, siRNA-mediated knockdown of CNOT3 in ARPE-19 cells significantly increased PRPF31 mRNA and protein expression, consistent with the increased PRPF31 expression level observed in non-penetrant individuals. Overall, these results indicate that CNOT3 is a modifier of the penetrance of PRPF31-mediated retinitis pigmentosa through direct regulation of PRPF31 expression (85). The large pedigree size and synergistic use of cell biology both clearly contributed to the success of this study—advantages that will likely not be feasible for many others.

Perhaps the greatest challenge to modifier studies with human families is that pedigree size limits statistical power. In the discovery phase of an experiment, results seldom survive statistical correction for multiple tests and conclusions are sometimes reported as anecdotal associations. One approach for addressing this challenge is to limit the number of tests, either by using separate patient samples in discovery and replication stages, or by focusing on candidates previously identified in mouse studies. The latter strategy was used to successfully identify AHI1 as a modifier of retinal degeneration penetrance in humans with nephronophthisis caused by NPHP1 mutations. On the basis of studies in mice showing that mutations in Ahi1 can modify Nphp1-mediated retinal degeneration, 153 independent individuals with nephronophthisis were genotyped for a hypomorphic, but not disease-causing, non-synonymous variant (C2488T) in AHI1. The T-allele was significantly more common in individuals with nephronophthisis and retinal degeneration (i.e. Senior Loken Syndrome) versus those with nephronophthisis only, with a relative risk of 7.5 for retinal degeneration (89).

Additional modifiers of eye disease will potentially be identified through larger genome-wide association studies, some of which already hint at the action of modifiers (100). Human genome-wide modifier studies have had some successes (101,102), including those for a few ophthalmic traits or diseases (103–105). Though computationally burdensome and challenged by multiple test corrections, some models predict that genome-wide modifier detection in humans is feasible (106) and new approaches for detecting gene–gene interactions continue to be proposed (107).

Conclusions and Outlook

Genetic modifiers will be of continuing importance in the study of eye disorders. Intentional phenotype-driven studies of modifiers are inherently innovative as they can identify genes of functional influence in the context of a whole animal, regardless of whether the gene is currently thought to be important to ophthalmology or to any field of biology. Therefore, this approach has the potential to make novel observations and possible breakthroughs (108–113). New tools for studying genomes and the increasing appreciation of gene networks (114) will enhance existing efforts.

Conversely, the unintentional influence of genetic modifiers as a source of biological variability in ophthalmic experiments will also continue to be a confounding problem. To prevent genetic modifiers from unintentionally skewing results in experiments with mice, several tactics are warranted. First, control genetic background by reducing heterogeneity when feasible and attempting to equalize the amount of genetic variability in experimental and control cohorts. Second, use formal nomenclature for describing the genetic background of mice in publications, including the backcross generation (if congenic) or the acknowledgment of uncertainty (if segregating). Third, test mice in new colonies for common mutations such as Crb1^rd8 and Pde6b^rd1[for genotyping protocols, see: rd8 (40,115), rd1 (116,117)] and be cautious with any colonies in which coat colour is varying. Best practices for human genetic studies are difficult to distill into a few bullet points, though it is promising that new methods for identification of gene–gene interactions continue to be developed and that the recent NIH emphasis on rigor and transparency is keeping the ophthalmic genetics community attentive to the possibility of genetic modifiers as relevant biological variables in both mouse and human studies.

Conflict of Interest statement. None declared.

Funding

Merit Review Award (#I01 RX001481) from the United States (U.S.) Department of Veterans Affairs Rehabilitation Research and Development Service; National Institutes of Health, NEI (R01EY017673) grants to M.G.A. The contents do not represent the views of the U.S. Department of Veterans Affairs or the U.S. Government.