Glaucoma in Border Collies

Report on Current Status of Research into Link Between G590A / R197Q Mutations in Olfactomedin-like Protein 3 and Closed-Angle Glaucoma in Border Collies

Adrian L Smith, M.A. Ph.D. (Cantab)

Submitted: July 11, 2018

This report and accompanying recommendations is prompted by requests the BOD has received to promote a new commercial test to Members for a loss-of-function mutation in the Olfml3 gene that has recently been identified as a contributing factor to severe goniodysgenesis in Border Collies. This report contains a summary of the findings reported by researchers at the Roslin Institute working in collaboration with Animal Genetics, Inc. and partnered with the Pastoral Breeds Health Foundation (UK).1 See:

The genetic research at the Roslin Institute was prompted by the observation of sudden onset primary glaucoma cases in young Border Collies, initially in Australia starting in the late 1990s and with more recent observations in the UK and elsewhere around the world. The affected dogs were found almost exclusively to have severe pectinate ligament dysplasia (PLD; reported as severe goniodysgenesis during gonioscopy examination), a developmental abnormality of the eye characterized by narrowing or closure of the iridocorneal angle (ICA) through which aqueous humor drains to relieve pressure in the eye. It is noteworthy that an unrelated previous study in Border Collies reported that 11 / 102 (10.8%) of Border Collies examined were moderately or severely affected by PLD.2

 The Pastoral Breeds Health Foundation (PBHF) started collecting data on affected dogs, and ancestral analysis of affected dogs led back through a small number of popular sires to 3 common progenitors from the late 1970s / early 1980s in Australia. This data enabled researchers to conduct research into underlying genetic causes. The researchers used industry-standard genome sequencing techniques, and the quality of the data generated appears to be good.

Identification of a causative genetic mutation was performed by initially carrying out genome wide single-nucleotide polymorphism (SNP) analysis on 59 dogs (17 severely affected and 42 unaffected by goniodysgenesis) in order to identify genetic regions associated with goniodysgenesis. This identified a strong association between severe goniodysgenesis and a region of chromosome 17, with a window of 40 SNPs (covering a region of 106 base pairs) having the lowest p-value (3 × 10-7). [Statistically highly significant].

Whole genome sequencing on this region was then performed on 9 dogs from 4 nuclear families (3 affected by glaucoma, 3 affected by severe goniodysgenisis but not glaucoma, and 3 unaffected controls) in order to identify specific genetic mutations that differentiate the phenotypes in closely related dogs. This resulted in the identification of a mis-sense mutation in the Olfml3 gene that codes for Olfactomedin-like protein 3 (OLFML3). It is stated that the identified mutation (G590A) predicts an R197Q mutation in the olfactomedin domain of the protein – this Arg-197 residue is highly conserved throughout mammals, birds, and reptiles making it likely that this mutation will lead to loss-of-function in OLFML3. [Note: While there is no reason to doubt the claimed link between the G590A and R197Q mutations, I could not see the data to support this claim in the manuscript and am not sure where this comes from.] It is noted that the allele frequency of this mutation in Border Collies from the Dog Biomedical Variant Database Consortium is 0.06 (4 / 35 dogs), but is absent in other breeds reported to be associated with goniodysgenesis. In addition, the manuscript reports that Animal Genetics, Inc. genotyped 350 dogs recorded as Border Collies (unknown clinical status) identifying 2 homozygotes and 29 heterozygotes, giving an allele frequency of 0.05 for the breed.


Clinical Observation







Glaucoma 9 0 0
Severe Goniodysgenesis 10 2 0
Unaffected 1 27 30

Table 1. Genotyping for G590A mutation in Olfactomedin-like protein 3 in Border Collies included in study. AA represents homozygous for mutation, AG is heterozygous for the mutation, and GG is wt normal.

Validation of the G590A mutation finding was performed by genotyping for this mutation in the original SNP group plus 33 additional Border Collies (2 with glaucoma, 5 severely affected by goniodysgenesis, and 26 unaffected). A summary of the results is presented in Table 1 showing a clear correlation between the glaucoma / severe gonioscopy affected dogs and the homozygous mutant genotype as would be anticipated from a loss-of-function recessive mutation being responsible for severe goniodysgenesis.

DNA Genotype n   Severe Goniodysgenesis
AA 10 3.1% 3 / 5
AG 77 24.1% 4 / 52
GG 232 72.7% 2 / 133

Table 2. Genotyping for G590A mutation in Olfactomedin-like protein 3 in Border Collies submitted to PBHF database (as of 07/10/2018). AA represents homozygous for mutation, AG is heterozygous for the mutation, and GG is wt normal.

With the commercial launch of the test at Animal Genetics, Inc. and voluntary submission of data to the PBHF database (, it is now possible to start collating public data on the incidence of the mutation within the breed. The data as of July 10, 2018 is presented in Table 2. This data indicates an allele frequency of 0.152 for the mutation in the dogs being tested, which is higher than the allele frequency previously reported in the wider Border Collie community by the Dog Biomedical Variant Database Consortium. Statistically, this new data predicts an “at-risk” AA genotype frequency of 0.023 in line with the data that is emerging. It is possible that the higher frequency of the mutation in dogs submitted to the PBHF database compared with the previous data (allele frequency 0.152 vs 0.05) may be a result of the data submitted to the database being primarily from show lines and therefore more likely to be descended from the dogs where the mutation originated.

A plausible scientific rationale for the role of OLFML3 in development of the eye is given by the researchers. Expression of OLFML3 is associated with microglia, which are contained in the retina and are important for retinal health. Opening of the ICA during development involves considerable remodeling of the mesenchymal tissue within the angle, and microglia may contribute to this process. Loss-of-function in OLFML3 would then disrupt opening of the ICA and proper formation of drainage channels.

With an “at-risk” frequency of 0.023, this suggests that around 2.3% of Border Collies will likely be homozygous for the G590A OLFML3 mutation. There were 4162 Border Collies registered with AKC in 2016 (latest registration data available); statistically this suggests that there will be around 96 AKC-registered puppies born in the US each year that are homozygous for the mutation. While there is a strong correlation between severe goniodysgenesis and the AA genotype in the study described, there is a disconnect between dogs identified as having severe goniodysgenesis and those suffering from glaucoma. The American College of Veterinary Ophthalmologists database records 1 case of glaucoma in Border Collies in the US between 1991 and 2015 out of 25,577 Border Collies examined, and I am aware of a further 2 Border Collie pups born more recently in the US that have glaucoma (from the same dam). It should be noted that the authors of the current study also report that there are too few clinical cases of glaucoma in their study to allow the identification of other loci involved in the progression from goniodysgenesis to glaucoma in the Border Collie.

The authors of the current study attempt to rationalize the disconnect between OLFML3 status and the occurrence of glaucoma in the Border Collie. OLFML3 is a member of a large family of proteins characterized by the olfactomedin (OLF) domain, with several members being expressed in the eye. The authors suggest that these other OLF family members may compensate for OLFML3 abnormalities and that genetic variation in these family members may modify the impact of the OLFML3 mutation in the Border Collie, thereby explaining the variable clinical phenotype in homozygotes for the mutation. They postulate that inadequate compensation by OLF family members for the putative effects of the R197Q OLFML3 mutation may be responsible for the progression from severe goniodysgenesis to glaucoma.

The authors conclude by stating that although the G590A OLFML3 genetic test will allow breeders to select against homozygotes and heterozygotes in order to decrease the prevalence of the risk for severe goniodysgenesis in the breed, they caution against reducing the breeding pool too much because of the risk of other recessive conditions being increased. This raises the question of what breeding decisions should be made based upon this test, and it is a matter for individual breeders to decide upon given the low prevalence of glaucoma in the USA. I would suggest that if a breeder is planning a breeding in the near future and they believe that they may have a problem with glaucoma in the lines being bred, then certainly go ahead and test for the mutation to reduce the risk. Meanwhile, I would encourage the collection of more data. Animal Genetics, Inc. are currently the main commercial provider of this DNA test in the US, but there are no restrictions on other testing companies including the test in their panels and costs will likely drop as it becomes more widely available. I have been in discussion with the scientists at Embark about including this test in their panel since my own dog has already been tested with them in their broad DNA test panel. Their focus is on collecting big data in order to learn about the genetic underpinnings of disorders (they currently test a DNA sample for 250,000 markers). They indicated that they should be able to include the test in their panel once validated, and it would be added to a dog’s DNA Test Report at minimal additional cost if they already have DNA on file for a dog. This will be driving my personal decision on when and how to test.

  1. Pugh CA, Farrell LL, Carlisle AJ, et al. Arginine to glutamine mutation in olfactomedin-like 3 (OLFML3) is a candidate for severe goniodysgenesis and glaucoma in the Border Collie dog breed. Unpublished 2018.
  2. Oliver JA, Ekiri AB, Mellersch CS. Pectinate ligament dysplasia in the Border Collie, Hungarian Vizsla and Golden Retriever. Vet Rec 2017;180:279.