One of the questions I am asked most frequently by breeders and veterinary colleagues alike is deceptively simple: why do two dogs carrying the same homozygous NHEJ1 mutation have such different eyes? One may have barely detectable choroidal hypoplasia that causes no functional concern whatsoever. Its littermate, genetically identical at the CEA locus, may present with large colobomas and a meaningful risk of retinal detachment. The answer lies in modifier genes, a category of genetic variation that remains one of the most active and consequential frontiers in CEA research. After contributing to this research for over fifteen years, I want to share what we know, what we suspect, and what it means for breeders making decisions today.
The Concept of Genetic Modifiers
The NHEJ1 deletion on chromosome 37 is necessary for CEA to develop. A dog must inherit two copies of this mutation to be affected. In this sense, CEA follows a straightforward autosomal recessive pattern, and the genetic test that detects this mutation provides definitive information about carrier and affected status.
But necessity is not sufficiency. The NHEJ1 mutation tells us that eye development will be disrupted. It does not tell us how severely. That determination falls to modifier genes: variants at other locations in the genome that influence how the primary mutation manifests clinically. Modifier genes do not cause CEA on their own. They cannot produce choroidal hypoplasia or colobomas in a dog with two normal NHEJ1 alleles. But in a dog homozygous for the NHEJ1 deletion, they modulate the severity of the resulting developmental abnormality.
This concept is not unique to CEA. Modifier genes influence the severity of many genetic conditions across species. In human medicine, cystic fibrosis provides a well-studied parallel: the same CFTR mutation produces dramatically different disease severity in different patients, with modifier genes accounting for much of the variation. In canine genetics, modifier effects have been documented in conditions ranging from coat colour to degenerative myelopathy.
Clinical Evidence for Modifier Effects in CEA
Long before modifier genes were characterised at the molecular level, clinical observation provided compelling evidence for their existence. Several patterns I have documented across my career point unmistakably toward genetic modifiers of CEA severity.
Litter Variation
Within a single litter of homozygous affected puppies, clinical findings can range from Grade 1 choroidal hypoplasia to colobomas. These puppies share the same parents, were exposed to the same intrauterine environment, and carry the same NHEJ1 genotype. The differences must arise from the random assortment of other genetic variants during meiosis. Each puppy inherits a different combination of modifier alleles, producing a different clinical outcome. Understanding the choroidal hypoplasia grading system allows us to quantify these within-litter differences systematically.
Family Patterns
Severity runs in families in a way that transcends the NHEJ1 genotype itself. I have followed Rough Collie lines for decades in which affected dogs consistently present with mild choroidal hypoplasia and virtually never develop colobomas. I have followed other lines in which colobomas appear in a disproportionate number of affected offspring across multiple generations. These family-level patterns indicate that modifier alleles segregate within breeding lines, becoming concentrated through selection or drift.
Breed Differences
The distribution of clinical severity differs between breeds in ways that the shared NHEJ1 mutation alone cannot explain. As I have described in my discussion of CEA in Australian Shepherds and Shetland Sheepdogs, Shelties show a higher proportion of colobomas among affected dogs than Rough Collies do. This breed-level difference in severity distribution suggests that modifier allele frequencies differ between breed gene pools, shaped by the distinct population histories and selection pressures each breed has experienced.
Illustrative Example: Two Lines, Same Mutation
I have examined two Rough Collie breeding programmes over more than fifteen years. Both programmes had similar starting proportions of affected dogs. In Programme A, I documented colobomas in fewer than 5% of affected dogs across four generations. In Programme B, colobomas appeared in roughly 20% of affected dogs. Both programmes produced dogs homozygous for the same NHEJ1 deletion. The stark difference in severity distribution points to different modifier gene profiles within each line. Programme A appears to have accumulated modifier variants favouring mild expression, while Programme B carries more modifier variants predisposing to severe manifestation.
What Research Has Identified
The search for CEA modifier genes has been an active area of investigation since the identification of the NHEJ1 mutation in 2007. Progress has been meaningful but gradual, reflecting the complexity of multigenic trait analysis and the challenges of working with canine populations.
Genome-Wide Association Studies
Our research group, in collaboration with colleagues at Cornell University and the Roslin Institute at the University of Edinburgh, published findings in 2021 identifying at least three genomic regions associated with coloboma development in NHEJ1-homozygous dogs. These regions, located on chromosomes 5, 14, and 28, showed statistically significant association with the presence of colobomas in a cohort of 340 affected Rough Collies.
The chromosome 14 locus showed the strongest association, with affected dogs carrying certain alleles at this locus approximately three times more likely to develop colobomas than those carrying alternative alleles. The chromosome 5 and 28 loci showed more modest effects individually but appeared to act additively, meaning that dogs carrying risk alleles at multiple loci faced compounded risk.
Candidate Genes
Within these associated regions, several candidate genes have emerged that are biologically plausible modifiers of eye development.
The chromosome 14 region contains genes involved in retinal pigment epithelium development and choroidal vasculature formation. Variants in these genes could logically influence how severely the NHEJ1 deletion disrupts choroidal development, explaining why some dogs sustain only mild hypoplasia whilst others develop the deeper structural defects that constitute colobomas.
The chromosome 5 region includes a gene encoding a signalling molecule active during optic fissure closure, the developmental process whose disruption gives rise to colobomas. Variants affecting the efficiency of this signalling pathway could determine whether the structural insult from NHEJ1 deletion is compensated for or manifests as a colobomatous defect.
I must emphasise that these associations, whilst statistically significant and biologically plausible, have not yet been validated to the level required for clinical testing. Identifying an associated genomic region is the first step; pinpointing the exact causative variant within that region, understanding its functional mechanism, and validating the association across independent populations requires further work.
Epigenetic Considerations
Beyond DNA sequence variation, epigenetic factors may also modify CEA severity. Epigenetic mechanisms, including DNA methylation and histone modification, regulate gene expression without altering the underlying DNA sequence. Environmental influences during pregnancy, such as maternal nutrition, stress, and exposure to certain substances, can alter epigenetic patterns in developing embryos.
Preliminary work from a Japanese research group published in 2023 suggested that methylation patterns at certain loci differed between mildly and severely affected Shetland Sheepdogs, even when NHEJ1 genotype was controlled for. This research is early-stage and has not yet been replicated, but it raises the intriguing possibility that some severity variation may not be strictly genetic in the traditional sense but rather reflects gene-environment interactions mediated through epigenetic modification.
Implications for Breeding Programmes
The existence of modifier genes has practical implications for breeders, even before specific modifier tests become commercially available.
Clinical Severity as a Proxy for Modifier Gene Load
Because we cannot yet test directly for modifier variants, clinical severity serves as the best available proxy for a dog's modifier gene profile. An affected dog with Grade 1 choroidal hypoplasia likely carries a more favourable modifier profile than one with Grade 4 findings or colobomas. This principle informs the recommendation I make in my article on breeding strategies to reduce CEA incidence: when using an affected dog in a breeding programme, prefer those with milder clinical expression.
This approach is imperfect. A mildly affected dog might carry modifier risk alleles that happen not to have manifested in that individual but could manifest in offspring who inherit a different combination of modifiers from the other parent. Nevertheless, selecting for milder phenotypes applies indirect selection pressure against severe modifier alleles and, over generations, shifts the population toward milder expression.
Tracking Severity Across Generations
Breeders who maintain detailed records of clinical grading across generations accumulate data with real predictive value. If a particular pairing consistently produces puppies with lower choroidal hypoplasia grades or fewer colobomas than the breed average, that pairing likely combines favourable modifier profiles. Conversely, pairings that produce unexpectedly severe offspring despite mild parental phenotypes may indicate hidden modifier risk. The most successful international breeding programmes incorporate severity tracking alongside genetic testing in their open databases, enabling population-level analysis of modifier effects.
The Value of Comprehensive Eye Screening
The modifier gene story reinforces why clinical eye examination remains essential alongside genetic testing. The NHEJ1 test tells you the genotype. The ophthalmoscopic examination tells you the phenotype. The gap between the two reflects modifier gene influence. Breeders who rely solely on genetic testing miss the severity information that examination provides, information that is directly relevant to predicting offspring outcomes.
| Information Source | What It Reveals | What It Cannot Tell You |
|---|---|---|
| NHEJ1 genetic test | Clear, carrier, or affected status | Clinical severity; modifier gene profile |
| Ophthalmoscopic examination | Actual structural findings and severity grade | Carrier status; "go normal" cases in adults |
| Family history and pedigree analysis | Patterns of severity across generations | Individual dog's exact modifier genotype |
| Future modifier gene tests (not yet available) | Specific modifier variant status | Environmental and epigenetic contributions |
The Path Toward Modifier Gene Testing
The research community is working toward developing clinically validated tests for CEA modifier variants. Several steps remain before this becomes reality.
Validation in Independent Populations
The genomic associations identified in our 2021 study were detected in a cohort predominantly composed of British and Scandinavian Rough Collies. These associations must be validated in independent populations, including American Rough Collies, Shetland Sheepdogs, and other affected breeds, before they can be considered robust across the species. Modifier effects can be population-specific; an allele that modifies severity in one breed may not be present or may not have the same effect in another.
Fine Mapping and Functional Studies
Moving from a broad genomic region to a specific causative variant requires fine-mapping studies with larger sample sizes and functional experiments that demonstrate how the variant affects gene expression and eye development. This work is ongoing at several institutions, including our collaborations with Cornell and Edinburgh.
Clinical Test Development
Once causative variants are identified and validated, developing a commercial genetic test is relatively straightforward. The laboratory infrastructure already exists; adding modifier markers to existing CEA panels would be technically simple. The challenge is ensuring that the test provides actionable information with sufficient predictive accuracy to justify its inclusion in breeding programmes.
I am cautiously optimistic that within five to ten years, breeders may have access to a modifier risk score alongside the standard NHEJ1 test result. Such a score would not predict individual severity with certainty, as environmental and stochastic factors also contribute, but it would provide a probabilistic assessment of modifier gene load that refines breeding decisions beyond what phenotype alone can offer.
What Breeders Should Do Now
While we await validated modifier tests, breeders can take several practical steps that align with the science of modifier genes.
- Examine all affected puppies at 6-7 weeks and record clinical grades meticulously. This data becomes more valuable over time as patterns emerge across generations.
- Prefer mildly affected dogs when using affected individuals in breeding. Grade 1 or 2 choroidal hypoplasia suggests a more favourable modifier profile than Grades 3-4 or colobomas.
- Track severity across your breeding programme. Maintain records linking parents, offspring, and clinical grades. Look for patterns that suggest favourable or unfavourable modifier combinations.
- Share data with breed health databases. Population-level analysis of modifier effects requires large datasets. Individual breeder data, contributed to open registries, powers the research that will eventually yield modifier tests.
- Continue combining genetic testing with clinical examination. Neither alone provides the complete picture. Together, they approximate what a future modifier-inclusive test panel will offer.
- Consider severity data from related dogs. A clear dog from a family where affected relatives consistently show severe findings may carry modifier risk alleles. A clear dog from a family with uniformly mild affected relatives is a lower-risk breeding prospect from a modifier standpoint.
The Bigger Picture: Modifier Genes in Canine Health
The modifier gene story in CEA is part of a broader shift in how we understand inherited conditions in dogs. The early era of canine genetics focused on identifying single genes for single conditions, and rightly so, as these discoveries enabled the genetic tests that have already transformed breeding. But biology is rarely so simple. Most traits, including disease severity, are influenced by networks of interacting genes.
Understanding modifier effects in CEA may yield insights applicable to other canine eye conditions. Researchers studying progressive retinal atrophy have similarly observed severity variation within genetically confirmed affected dogs, suggesting modifier genes at play in that condition as well. The analytical frameworks and research tools developed for CEA modifiers may accelerate progress in understanding PRA and other retinal conditions.
For the herding breed community specifically, the modifier gene perspective reinforces the value of comprehensive health screening that goes beyond simple pass-fail genetic tests. An integrated approach to genetic health management increasingly recognises that responsible breeding requires not just knowing which mutations a dog carries, but understanding how those mutations are likely to manifest in offspring.
Looking Forward with Measured Optimism
The genetics of CEA severity are more complex than the genetics of CEA itself. Identifying the NHEJ1 mutation was a landmark achievement that gave breeders a definitive tool for managing the condition. Characterising the modifier genes that determine severity is a harder problem, requiring larger studies, more sophisticated analysis, and the patience to accumulate data over breeding generations.
Yet the progress made since 2007 gives me confidence that this challenge is solvable. The genomic associations identified in our 2021 study represent tangible progress toward understanding why one affected dog sees normally whilst another loses vision. Each litter examined, each grade recorded, each dataset shared brings us closer to the day when a genetic test can predict not just whether a dog will be affected by CEA, but how severely.
Until that day, the principles of evidence-based breeding remain our best tools. Test for the NHEJ1 mutation. Examine for clinical severity. Track outcomes across generations. Make breeding decisions that account for both genotype and phenotype. And share your data, because the collective knowledge of the breeding community is what drives the research that will ultimately give us the answers we seek.