When I tell breeders that their dog's CEA results from a 7.8 kilobase deletion in the NHEJ1 gene on canine chromosome 37, I watch their eyes glaze. The biological detail feels distant from the practical reality of puppy examinations and breeding decisions. But understanding the molecular basis of this condition — even in broad terms — helps explain puzzles that otherwise seem inexplicable: why affected dogs vary so much in severity, why the same test works across so many breeds, and what limits our current ability to predict outcomes. This article is for anyone who wants to understand the "why" behind the condition they are managing.
What Is the NHEJ1 Gene?
NHEJ1 stands for Non-Homologous End Joining Factor 1. The protein encoded by this gene participates in DNA repair, specifically in the pathway that repairs double-strand DNA breaks. When DNA is damaged in both strands simultaneously — which happens through radiation exposure, certain chemical damage, and normal cellular stress — the cell must rejoin the broken ends. NHEJ (non-homologous end joining) is one of the primary mechanisms for accomplishing this repair, and NHEJ1 is a component protein in this machinery.
At first glance, a DNA repair protein seems an unlikely cause of an eye condition. We might expect mutations in genes encoding structural eye proteins or developmental regulators. The connection between NHEJ1 and eye development reflects how extraordinarily sensitive developing tissue is to any disruption in the intricate cellular processes that create normal anatomy from undifferentiated cells.
During foetal development, the choroid — the vascular layer that nourishes the retina — forms through a precisely choreographed sequence of cell division, migration, and differentiation. These rapidly dividing cells are particularly vulnerable to inadequate DNA repair capacity. When NHEJ1 function is compromised, subtle accumulation of unrepaired DNA damage in the dividing choroidal progenitor cells disrupts normal development, resulting in the hypoplastic choroid that characterises CEA.
The Specific Deletion
The CEA-causing mutation is not a single nucleotide change but a large deletion: approximately 7.8 kilobases of DNA are missing from the NHEJ1 locus on chromosome 37. This deletion disrupts normal NHEJ1 protein production in affected dogs, though the precise mechanism — whether through complete loss of protein production, production of a non-functional truncated protein, or disruption of regulatory elements — has not been fully characterised in published literature.
What makes this mutation particularly interesting from a population genetics perspective is its consistency. The same deletion, identifiable by the same molecular markers, causes CEA in Rough Collies, Smooth Collies, Shetland Sheepdogs, Border Collies, Australian Shepherds, Lancashire Heelers, Nova Scotia Duck Tolling Retrievers, and every other affected breed. This indicates that the mutation arose once in a common ancestral dog population and spread through the various breeds as they diverged from shared herding dog stock — probably several centuries ago, long before any formal breeding records existed.
This single common origin is why one genetic test works across all affected breeds, which I discuss in the context of CEA in non-Collie breeds. The laboratory does not need breed-specific assays; the deletion is the same regardless of whether the sample comes from a Rough Collie or an Australian Shepherd.
Autosomal Recessive Inheritance
CEA follows classic autosomal recessive inheritance. Each dog carries two copies of every gene, one inherited from each parent. For the NHEJ1 gene, dogs can carry:
- Two normal copies (N/N): Genetically clear. Cannot be affected and cannot produce affected puppies unless bred to an affected dog.
- One normal copy and one mutant copy (N/CEA): Carrier. Clinically normal or nearly normal. Passes the mutation to approximately 50% of offspring.
- Two mutant copies (CEA/CEA): Homozygous affected. Will show some degree of clinical CEA, ranging from minimal choroidal hypoplasia to severe colobomas.
The autosomal designation means the gene is located on an autosome (non-sex chromosome), so the condition affects males and females equally. There is no sex-linked modulation of expression.
Why Carriers Are Usually Normal
Carriers have one functional copy of NHEJ1. In most contexts, one functional copy produces sufficient protein for normal cellular function — this is called haploinsufficiency tolerance. The developing choroid in a carrier dog has enough NHEJ1 function to proceed normally. Only when both copies are disrupted does the shortage of functional protein fall below the threshold needed for normal choroidal development.
In rare cases, carriers may show very subtle clinical changes — minor choroidal irregularities that fall below the threshold for formal diagnosis. I have observed what I would call "sub-threshold" findings in a small number of obligate carriers (dogs whose parentage and offspring genotypes confirm carrier status), but this is uncommon and clinically insignificant. For breeding purposes, carriers are treated as clinically normal. The genetic test distinguishes them from clear dogs despite their normal appearance.
The Variable Expression Problem
One of the most scientifically interesting and practically frustrating aspects of CEA is the enormous variability in clinical severity among homozygous affected dogs. Two CEA/CEA dogs with identical genotypes at the NHEJ1 locus may have dramatically different clinical presentations — one barely diagnosable on examination, the other with colobomas threatening vision. How can the same mutation produce such different outcomes?
Several mechanisms contribute to this variability. My article specifically exploring modifier genes and current research into CEA severity addresses this in depth, but the core concept is worth introducing here.
Modifier Genes
The NHEJ1 deletion creates a vulnerability in choroidal development, but the eventual clinical outcome depends on interactions with many other genes across the genome. These "modifier genes" influence processes like cellular stress response, alternative DNA repair pathways, inflammatory signalling during development, and vascular formation. Dogs whose modifier gene profile compensates effectively for the NHEJ1 deficiency develop mild lesions; dogs with less effective compensation develop more severe disease.
Research published by my collaborators at Cornell University in 2021 identified three chromosomal regions associated with coloboma development in CEA-affected dogs. These are candidate modifier loci, though the specific genes within these regions have not yet been definitively identified. Characterising these modifiers is an active research priority because tests for modifier alleles could eventually allow us to predict likely severity from a blood sample — a genuinely transformative capability for breeding decisions.
Stochastic Developmental Variation
Beyond modifier genes, some variability in CEA severity likely reflects inherent randomness in developmental processes. Foetal development is not perfectly deterministic; small variations in cell positioning, timing of signal gradients, and microenvironmental conditions create biological "noise." For CEA, this might explain why some affected dogs have asymmetric lesions — different severity in the two eyes — even though both eyes share the same genetic constitution.
Why the Deletion Persisted in Populations
An evolutionary question worth addressing: why has a deleterious mutation reached such high frequency in affected breeds? If CEA causes eye problems, shouldn't affected dogs have been at a disadvantage and the mutation gradually eliminated?
The answer has several components. First, most affected dogs have only mild choroidal hypoplasia and are entirely functional for any working or companion purpose. There was no selection pressure against affected dogs in traditional herding contexts. Second, herding breeds were selected intensively for working ability, temperament, and herding instinct — characteristics entirely unrelated to CEA status. Dogs carrying the mutation were selected as breeding stock because they excelled at the traits breeders valued. Third, the high prevalence of the mutation means that excluding affected dogs would have severely restricted the breeding pool, preventing breeders from maintaining other desirable traits.
Now that genetic testing exists, breeders have the information needed to gradually reduce mutation frequency without sacrificing other breed qualities. The key is using testing data intelligently rather than imposing blanket exclusions. Breeding strategies that thoughtfully reduce CEA incidence while preserving genetic diversity represent the optimal approach for long-term breed health.
Testing Accuracy and Limitations
Because the NHEJ1 deletion is large and consistent across breeds, PCR-based detection of the deletion is highly accurate. Reputable laboratories report sensitivity and specificity approaching 100% for the mutation itself. The test reliably distinguishes N/N, N/CEA, and CEA/CEA dogs based on the presence or absence of the deletion in one or both copies.
The primary limitation is not in the test's ability to detect the mutation but in translating genotype to phenotype prediction. Knowing a dog is CEA/CEA tells us they are affected; it does not tell us whether their lesions will be barely detectable choroidal hypoplasia or significant colobomas. This prediction gap is the frontier that current modifier gene research aims to close.
Additionally, the test assumes CEA in your dog's breed results from the known NHEJ1 deletion. There is a theoretical possibility that other mutations could cause similar-appearing eye changes, but in all breeds where CEA has been studied, the same NHEJ1 deletion is responsible. Diagnostically, if a dog tests genetically clear but shows clinical findings resembling CEA, I would recommend specialist re-examination and possibly a second genetic test from a different laboratory before concluding that an alternative mutation is present.
Future Genetic Research Directions
The next decade in CEA genetics is likely to bring several advances. Whole genome sequencing costs continue to fall, making comprehensive variant analysis economically feasible for more researchers and eventually for commercial testing programmes. This may reveal the modifier genes that determine severity, leading to tests that provide not just "affected or not" but "likely mild versus likely severe" predictions.
Additionally, as population genomics in dogs advances, we will better understand the spread of the CEA deletion through breed history, the relationship between population bottlenecks and current mutation frequencies, and how different national populations compare at the genomic level. This information will guide future breeding programme design. The international data already available from countries with long-standing mandatory testing programmes hints at what properly coordinated, science-based breeding management can achieve over generations.