Understanding Polygenic Inheritance

by C,A, Sharp

Originally published in the Fall 2004 issue of Double Helix Network News  Rev. Sept 2013


Everyone knows this old game:  You place beans or other markers on numbered squares arranged in a grid with columns labeled with the letters B-I-N-G-O.  Someone announces number and letter combinations pulled at random from a container.  The first person to form a line of markers across the grid in any direction yells, “Bingo!”

Breeding dogs is like playing Bingo, but instead of arranging beans on a card you are shuffling genes.  When the right combination lines up, one or more of the puppies in your litter may exhibit a polygenic trait:  Bingo!

Polygenic traits are probably the most difficult for breeders to understand.  They are certainly the most difficult to control in a breeding program, whether you want the trait or you don’t.  The phenotype of the dog—what you see or the behavior you get—will spring from the combined actions of multiple genes.  Sometimes, as with hip dysplasia (HD), the genetic potential may be swayed by environmental effects.

Unfortunately, many inherited diseases, like HD, are polygenic.  It follows an irregular inheritance pattern:  Some dogs will never produce HD, others may produce it occasionally, and some will produce it frequently.  Some families will have it every generation, others may go for several generations with no cases at all.  Excellent hips may produce dysplastic and affected parents may produce sound offspring.  Whether a dog will produce HD or not depends on exactly what combination of genes it has and who it is bred to.  Even though the environment plays a role, a dog won’t have HD unless it has the genes to do so.

Thinking of the different genes that contribute to polygenic traits as letters in the word BINGO may help you visualize what happens as they are passed from one generation of dogs to the next.  For the sake of this discussion we will say that to produce a polygenic trait, the dog must have particular alleles (versions) of five different genes before it will have that trait.  We will call them B I N G and O.  If a dog has any combination of alleles at those genes other than BINGO, it will not exhibit the trait.  BINGo won’t do it and neither will BInGO.  Dogs like these with an incomplete set are carriers of the trait, but they will only produce it if mated to a dog with the missing alleles.

You can breed BiNGO to BiNGO and never see the trait, but cross once to a dog that has I rather than “i” and there may be BINGO puppies.   Obviously a dog with only one BINGO allele will produce the trait far less often than one that has all but one of them.  In fact, a dog that just has just O might never be bred to one that has BING and therefore never produce the trait even though it is a carrier.

Though we don’t think of it as such, coat color results from action of multiple genes.  Some determine color (black, blue, yellow.)  Others provide pattern (mask, piebald, merle).  A particular combination of color and pattern alleles determines what color  your dog is.  For example, an acceptably colored blue merle Australian Shepherd with tan points would have a genotype that could be described thus:  at– B- C- D- E- gg  kk Mm  si-.   Some genes must be dominant, others need two recessives.  The ones with one allele and a dash may have two different alleles, provided tat least one is the allele noted and the other is the same or recessive to it.  Change a single allele and you can have an entirely different color dog:  A pair of “m”s makes the dog a black tricolor; add a dominant K and it becomes a black bicolor.

Maintaining desirable polygenic traits is easier than getting rid of them, but even so you can breed a BINGO pair and wind up with a dud litter.  Diligent selection for BINGO over many generations may ultimately lead to the trait becoming a very common.  Even so, the occasional non-BINGO pup will occur because of a chance combination of non-BINGO alleles floating around in the gene pool in low frequency.

Completely eliminating a polygenic trait can be extremely difficult if not impossible.  If ancestors of your dog have produced an unwanted BINGO, some or all of the genes may come down to your dog.  If Lucky is heavily linebred on the Old Granddad/Grandma cross and Gramps and Granny were known to have produced BINGO, there’s a fair shot Lucky will carry at least some of the genes even if he doesn’t have the trait himself.  The more ancestors that have had or produced BINGO and the closer up they are in the pedigree, the more likely Lucky will at least be a carrier.

Unfortunately, there is much we do not know about polygenic diseases.  Some don’t start until the dog is an adult and may already have been bred.  Sometimes diagnosis is not clear-cut.  We may not be able to predict which cases will be controllable and which crippling or lethal.  And in most breeds we can only guess at the specifics of inheritance.  While it is possible with careful, diligent selection over generations, to clear a line of a trait, its absence doesn’t mean that you have eliminated BINGO.  Maybe you have only eliminated G.

A common breeders’ dilemma is the discovery that one of your dogs is half-sibling to a dog with an undesirable polygenic trait.  With half-sibs, you know that the common parent had some part of BINGO.  Therefore, your half-sib probably inherited something.  Depending on what genes the half-sib’s other parent has, it may have picked up a few more—possibly enough to have BINGO itself.  Even without a complete BINGO it is a carrier.  If you are going to breed a half-sib, you need to consider how clear of the trait the other side of its pedigree is.  Prospective mates will need to be from families clear of the trait in order to minimize the likelihood that it will be produced.

While inbreeding and linebreeding in and of themselves do not perpetuate unwanted polygenic traits, the risk of doing so increases if you linebreed on dogs that carry the necessary alleles.  When you do this, you can unintentionally increase the frequency of those genes in the group of dogs you use for breeding.  Coefficients of inbreeding can tell you which crosses are less linebred, but COIs are only part of the picture.  If the particular dogs in a pedigree that raise the COI are not problematic for the trait, the COI will have no bearing on whether or not you might get the trait.

Close inbreeding and heavy linebreeding on a BInGO dog will never produce an affected pup unless the N is brought in with another dog.  It is possible to have an entire line that is free of the trait, but if you outcross to a line that happens to have that missing N, you may “suddenly” have a genetic problem for which your line was “clear.”  This has been called “outcrossing surprise.”  Outcrossing will only help if the line you outcross to a dog or line that lacks BINGO alleles.

Eliminating  BINGO alleles may not be merely a matter of heavy and consistent selection against the trait.  It may be impossible.  Geneticists have recently discovered that genes multi-task.  Where we once thought dogs had 80-100 thousand genes, today we know the number is actually more like 25-30 thousand.  Different parts of genes may perform different tasks.  They may have different functions at different points in life, or do different jobs in different tissues.  It is possible that N is beneficial when present alone, and only becomes a problem when the whole BINGO is there.  Efforts to eliminate N along with BIGO might create more trouble than the trait you are trying to get rid of.

When the BINGO trait doesn’t show up until later in life, the affected dog may have been bred.  It will have passed along those alleles to its offspring but is unlikely to  pass the entire set.  A dog can give only half of its genes to its offspring.  If it is affected, those offspring will certainly receive some part of BINGO, but the pups won’t have the trait unless the other parent provides whatever alleles are missing.

The individual genes that make up BINGO may be dominant, recessive, codoimnant, incomplete penetrance (whether you see the trait or not) or incomplete expressivity (how pronounced the trait is or when it manifests).  The different genes may have only different numbers of alleles.  Only one or several of the possible alleles may contribute to BINGO.  Some genes may override the action of others.  Genes may work in concert to produce the BINGO phenotype.  If even one BINGO gene is on the X chromosome, sex ratios for the trait may be uneven.

Let’s take another look at coat color:  Pretend that all other colors are possible in your breed but blue merle is unacceptable.  Since all colors can happen, the actions of other color genes may confuse the issue or make it impossible to know if a dog is genetically blue merle.

When a dog is totally white,  you won’t be able to tell whether it is genetically merle.  Piebald pattern dogs will have patches of color, but they may be very small.  This kind of trait expression tempts breeders to conclude that perhaps it isn’t important because the dog only has “a little bit” of the trait.  However, the genes are there and it will reproduce like a blue merle even if you only see a tiny patch of the color or, as with the yellow dog, nothing at all.

This is an example of how a trait might be termed dominant with variable expressivity.  The color requires two dominant genes, however the variety of trait expression—big color patches, tiny ones, or none, depends on another gene.  These traits are not single gene; they are polygenic.  Therefore both parents of an offspring with the trait will have contributed alleles necessary to produce that trait.  They are both carriers.

Yellow color results from having two recessive copies of a gene called MC1r, or “E.”  If a dog with a blue merle genotype has two copies of the recessive “e,” you will not be able to tell that it is merle but if it has even one copy of the dominant “E” allele the dog would be merle.  This gene also has an allele that causes the mask pattern “Em” which could result in a yellow dog with a merle mask.  This might be termed dominant with incomplete penetrance, meaning sometimes you get the trait (EmEm or Eme) and sometimes you don’t (EEm).  This also is a type of polygenic inheritance and both parents contribute.

To put a BINGO spin on incomplete and variable penetrance, consider this scenario: B is dominant and INGO are not. Both B and O are vital to exhibiting the trait while IN and G are additive, causing variations in the presentation or progress of the disease.   Some cases will be worse than others (variable expressivity.)  Most will appear to be inherited in a dominant fashion, but every once in a while bINGO will be bred to BINGo, both of whom are normal, and BINGO will result (variable penetrance.)

Additive genes, like  ING in the example above, are often referred to as “modifiers.”  These are genes that tweak a quantitative phenotype, like coat length, how tall a dog is, or how heavy its bone.  Additive genes may determine things like seizure threshold in primary epilepsy, or age of seizure onset of cataracts.  Other modifying genes are qualitative, determining things like whether a dog has a full collar, white legs, white up the stifle, or a blaze.  Genes that contribute to qualitative  and quantitative traits are likely regulatory – directing other genes to do more or less of something, or stop now rather than later.

Individual genes may have any mode of inheritance.  With color, we know the major genes and we know what they do to the dog.  Most of the primary ones have been identified in the DNA.  We recognize color phenotypes and have a good understanding about how they are inherited.  For the most part, color is there to see before the puppies leave the whelping box.  But coat color is the exception; with most polygenic traits we know little of this.

It’s only natural for people to seek the simplest answer to a question.  It isn’t uncommon to hear assertions made about polygenic disease, based on nothing but hope, that one or most of the genes involved are dominant.  This is a dangerous mind game to play.  The more genes are involved in a polygenic trait, the less likely that all will have the same mode.  Most mutations for diseases tend not to be simple dominant because both Nature and breeders select vigorously against them so they tend not to persist in any great frequency.

We have no way to evaluate how many genes are involved in polygenetic traits or by what mode of inheritance they transmit their individual contributions.  We like easy answers and would prefer to be able to say that whatever went wrong is the fault of the other parent rather than the one you own, but it is extremely shortsighted  to assume that dominant genes are responsible without compelling evidence in support of that supposition.

There is hope that someday science will come up with a way to determine genotype on polygenic traits.  Early research into canine genetic disease focused on single-gene inheritance.  These traits were the easiest to track.  Now that scientists can pick apart the very structure of the dog’s DNA and untangle developmental pathways and networks of gene interaction there will come a day when we’ll have solid information of the genetics of these diseases.

Once genetic tests are developed, we can work toward eliminating them even though we may not totally remove all the BINGO alleles from the breed.  With a screening test, if you know the dog as B and G and the bitch has only I, you are safe to breed.  But if the bitch has N and O as well as I, she isn’t a good match for your dog.  Even if you know the genotypes of the parents, you won’t know how the BINGO alleles fell in the pups, so you would then screen any that were to be bred so the results could be compared with those for the puppies’ prospective mates.  This way you will be able to avoid crosses that produce polygenic disease.  The point is not to totally eradicate all BINGO alleles, but to produce healthy pups.   With screening tests, we will even be able to view genes for very serious diseases as a sort of fault rather than an automatic reason to take an animal out of the breeding pool.

Lacking any kind of screening test, the best breeders can do is evaluate risk and try to lower it via careful mating choices.  That may mean never breeding an otherwise excellent individual because the risk is too high.  Risk analysis of pedigrees will not totally prevent unwanted traits, but it is the best tool available at present.  However, it is dependent on the open exchange of information between breeders about what dogs have had or produced disease.

Until genetic tests become available, breeders need to amass as much information as possible about both positive and negative traits., not only in their own dogs, but their relatives, too.  Careful study of pedigrees, including the vertical pedigrees which include siblings of the dog and its progenitors, noting which dogs had or produced a trait, how many of them there are in the pedigree and how close they are to the subject dog will provide the breeder with an idea of how likely a particular dog is to have or throw the trait.

With careful record keeping, diligent study of pedigrees, and—as they become available—genetic screening tests, a breeder can make progress toward desirable polygenic traits and away from the undesirable.  With a high level of honesty and cooperation between breeders, progress will come even faster.  Someday nobody will have to yell “BINGO!” because of a bad line of beans.

The author would like to acknowledge Michele Betit for suggesting the BINGO concept on the EpiGenes discussion list.