Photo: Amanda Goodman-Lee
Sick honey bee colonies can be treated with antibiotics and miticides, but pathogens are evolving resistance to fight back. Now, researchers have developed a new counter-attack.

Living in a honey bee hive is like living in a house with 40,000  siblings. It’s a pathogen’s dream. Left unchecked, contagious diseases can bring a colony to its knees, but honey bees – as well as other social insects – have evolved a way to fight back. Over millions of years, they have developed a collection of behaviors called ‘social immunity traits’ that help combat disease and parasite outbreaks.

Leonard Foster, who is at the University of British Columbia in Vancouver, saw an opportunity to use social immunity to fight diseases and parasites instead of using synthetic treatments. With the critical help of collaborating scientists and beekeepers across the country, he and his team identified molecules produced in the bees’ antennae that accurately predict how good the bees are at doing social immunity tasks. By measuring these molecules (biomarkers) in the lab, they can now find A+ colonies without needing meticulous behavioral observations, and without challenging colonies with real diseases to see who survives. Importantly, they aren’t breeding a disease-resistant lineage to disseminate – instead, they have created the tools to allow breeders to produce diverse stock themselves. “These aren’t GMO bees,” Foster clarifies. “What we’re doing is enriching the natural disease resistance that already exists in the population.” The lab testing will be offered as a service to bee breeders for a fee, with the hope that breeders will use it to choose which colonies to propagate in their own breeding programs.

The social immunity traits the researchers are focusing on are grooming, hygienic behavior, and varroa-sensitive hygiene (VSH), each of which contributes to colony resistance in complementary ways. Adult bees will bite and comb each other to rid themselves of parasites (grooming). Nurses will detect and remove developing brood that’s dead or diseased and drag them out of the colony (hygienic behavior). Finally, self-sacrificing brood and hygienic workers contribute to suppressing mite reproduction (VSH).

An “A+” scoring hygienic colony – all the brood in the circular freeze-killed patches were removed after 24 hours. Photo credit: Amanda Goodman-Lee.

This is an incredibly complex suite of behaviors, and it’s remarkable that molecular markers (thirteen of them, to be exact) can be used to selectively breed for any one of them, let alone all three at once. Adding to the innovation, most molecular markers for selective breeding are parts of the DNA sequence, which is normally the easiest kind of life-molecule to measure, but a peculiarity of honey bee biology makes that approach impractical.

DNA markers are found using ‘genome-wide association studies,’ in which the DNA from hundreds or thousands of individuals is sequenced, then variations in the sequences are correlated with the trait of interest. This approach has been used in other livestock industries to identify markers that can predict characteristics that are hard to see, for example, if a cow will produce high-quality meat or milk enriched with protein. The trait may be hard to see, but with modern technology, reading the DNA sequences (represented as a four-letter code of A, T, G and C) is surprisingly easy. Here’s a hypothetical example: if every dairy cow with the sequence AAT at a specific spot in their DNA is a high-protein milk-producer, but cows with the sequence ATT are not, then that sequence could be used to indicate protein content. The DNA marker helps predict which cows are most likely to yield offspring that are high-protein milk-producers without having to measure protein amounts first hand. And, well, how would you do this for a steer, anyway?

One important feature of DNA markers (and, as it turns out, why they don’t work well for honey bees) is that the marker sequence doesn’t need to cause the good milk production for it to be a good predictor. In fact, most DNA markers don’t even occur in genes; rather, they occupy parts of DNA that don’t code for anything at all. But the DNA marker is still a good predictor because it’s joined – that is, physically linked – to the part of the DNA that does have the causal genetic difference. In our example, this means that the AAT sequence is physically close to the genetic change that causes those cows to be good milk producers. Since sequences that are physically close on a chromosome tend to travel together from parent to offspring, we can read the simple AAT/ATT marker in generation after generation and assume that all the AAT cows continue to carry the causal genetic change.

Honey bees, however, have some of the highest rates of genetic recombination – the fancy term for swapping around DNA sequences during sperm and egg production – that we have ever seen. This means that even if DNA-based molecular markers are identified (and some have been), the physical link to the desired trait is quickly broken. In just a few generations, the marker’s predictive power is compromised. Complicating things further, social immunity is a colony-level trait, and each colony contains not one but tens of different genomes (due to the queen’s multiple matings with many drone fathers), each with their own sequence variations. The cost and complexity of using genetic markers for breeding escalates, while the payoff decays. Luckily, there is an alternative approach.

Gene sequences ultimately exist to produce proteins, and these are the molecules that Foster and his team decided to analyze. Proteins work together to do different jobs in cells, which are themselves specialized to fulfill different needs of the organism. Cells in the antennae, for example, produce proteins that enable bees to sense different odors. Theoretically, protein abundance – rather than DNA sequences – could also be measured and correlated to traits to find protein markers for selective breeding. Traditionally, this is seldom done because while DNA markers are stable (i.e., they generally don’t change with age, diet, season, etc.), protein abundance can fluctuate dramatically in an organism’s lifetime. Finding proteins that are reliably produced and happen to correlate with your favorite trait is a dismal task. It’s probably for that reason that protein markers have never been used for selective breeding in any agricultural plant or animal – until now.

Foster’s research team started off by finding protein markers for hygienic behavior, since this trait is relatively well-understood and has a robust test to evaluate it in the field. After surveying several hundred colonies that originated from four Canadian provinces, as well as California and Chile, Foster and his colleagues identified a handful of proteins that met their strict requirements to be included in the marker panel. They accomplished this by doing conventional hygienic testing in the field in parallel with measuring thousands of the proteins produced in the bees’ antennae. Around fifty honey bees were dissected and analyzed for each colony, and colonies were examined over multiple years, so this quickly approaches 100,000 individual antennae. “How did they dissect so many antennae,” you ask? Well, as a lowly undergrad student at the time, I was one of the antenna-snippers. We spent weeks on end using tiny scissors to snip pairs of antennae from thousands of fuzzy honey bee heads. It’s a wonder that we didn’t develop permanent cross-eyes or carpal-tunnel syndrome – life has been much improved since I became a graduate student in the lab.

After rigorous statistical analyses correlating protein abundances with colonies’ hygienic scores, just nine proteins out of the thousands that were measured made the cut for the marker panel team. Considering what we know about these proteins’ functions, each of them seems to belong there. For example, to do hygienic behavior, honey bees must be able to smell the difference between diseased and healthy brood. It’s no coincidence that two of the markers are ‘odorant binding proteins,’ which are necessary to smell certain odors. The other markers are related to olfaction in different ways, like aiding nerve signal transmission, which is necessary to send the ‘smell’ signal to the brain.

Using a similar approach, the researchers also identified protein markers for grooming and VSH. On its own, hygienic behavior is only partially effective at combatting varroa mites, since hygienic honey bees are not always targeting mite removal specifically. However, mites are the bane of beekeepers and honey bees alike, so the VSH and grooming traits were added to the panel to create a pathogen-and-parasite-resistance trifecta.

Grooming might seem like a lowly resistance mechanism, but when done frequently and ferociously, it helps keep varroa mite populations in check. Mites rear their families in the protective recesses of capped brood cells, but they also spend time piggybacking on adult bees (during a period known as the ‘phoretic’ stage) to find new infestable brood cells or new colonies to decimate. While hitchhiking, the mites are vulnerable to the bite and claw (more properly, mandibles and tarsi) of grooming sisters, which can render the mites as amputees, or worse. Good groomers will also signal to their sisters that they need a deep clean by doing a jitter dance, which even to the untrained eye looks a lot like a distress signal. Despite the dance, grooming is a troublesome trait to score because unlike hygienic behavior, it doesn’t have a good field test. Instead, the grooming score is calculated by sifting through mites that drop on a sticky board and ranking them based on how battered they appear, presumably from enduring rigorous grooming.

VSH is an even more complex trait with new facets still arising to this day. It is itself an intricate weave of behavior and physiology which culminate in suppressing the mite’s ability to reproduce. VSH can involve hygienic behavior against varroa specifically, which is strongest if the mites also carry a heavy virus load, but it also includes less well-understood characteristics. For example, even if workers in VSH colonies are not allowed to uncap infested brood cells, they still yield fewer progeny mites. For years, this was just a mysterious ‘brood effect’ contributing to mite resistance (although recent work by Paul Page has been illuminating). The most well-established method of scoring colonies for VSH is to calculate the infestation frequency in a donor comb – meaning researchers must uncap patches of brood and count the number of infested cells – then see how much the infestation frequency changes after incubation in the potentially-VSH colony. This scoring method is somewhat incomplete, since it does not capture the brood effect, but it’s the most feasible method we have. In total, these grooming and VSH measurements allowed the researchers to correlate a further four proteins with these immune behaviors, bringing the total to thirteen protein markers.

With the complete marker panel, it was time to test how well they performed in a real-world scenario. Foster and his colleagues measured these thirteen protein markers in an independent population of honey bee colonies and used this information like a molecular report card, selecting the best colonies to breed over three generations. They found that even using just the protein markers to choose the colonies – and not the conventional field test – hygienic behavior was significantly enriched in each subsequent generation. This proved for the first time that protein markers can be effective tools for selective breeding in honey bees. But the real test lies in how well these colonies withstand diseases.

This is where the researchers committed unthinkable (but necessary) acts on their colonies. In a controlled infection yard, they inoculated colonies with normally-detrimental loads of varroa mites and American foulbrood (mercifully, not at the same time). They left the colonies untreated until the following spring, when they assessed which ones survived. Three different kinds of colonies were tested: Canadian unselected stock, New Zealand imports (also unselected), and the colonies bred using marker-assisted selection. About 70-75% of the New Zealand imports and the Canadian unselected stock perished after mite inoculation, while – amazingly – less than 25% of the selected stock met the same fate. About 87% of imports and 70% of unselected stock succumbed to the American foulbrood challenge, compared with just 40% of selected colonies. Foster confides, “The observation that varroa resistance seemed to be better with MAS [the protein markers] than with FAS [the hygienic field test] was incredible. I felt vindication after having a lot of people tell me it wasn’t going to work when we first started.” Importantly, Foster’s team also showed that their selectively bred colonies still produced ample honey, raking in the gold (~70 kg/year, or ~150 lbs) like the best of them.

Critics say selective breeding is a short-term solution because over time, if rigorous selection is maintained, then the population will become inbred. Alternatively, if selection wanes, the population quickly returns to less-desireable baseline characteristics. However, the breeding model encouraged by Foster – where bee breeders use the markers to select colonies from their own diverse, locally-adapted stock, rather than having one central genetic lineage – should circumvent widespread inbreeding. “What is really under-appreciated in selective breeding is that you need to start with a large number (at least 50 to 100 hives, in the case of bees) to have enough diversity to hope to be able to make good selections,” Foster explains. “The same thing will apply with MAS – a breeder (or, possibly, a group of breeders) – will need to start with many colonies. They will also need to have some means of either closed breeding or instrumental insemination.”

Another criticism is that after all this work, protein markers still don’t provide stronger selection than the conventional methods of field testing. In fact, marker-assisted selection performs marginally worse, as far as hygienic score is concerned (although apparently, this didn’t hurt their ability to clear mite infestations). The major benefit, though, is that it’s far less time and effort (think: sending a sample of bees in the mail) than the laborious field test. Of course, the sample of bees would also be mailed with a check, but at a target price of about $30/colony, the hope is that the sacrifice will be worth it. In addition, no one can slack off. Foster explains, “As with any selective breeding in bees, the selection will need to be done every generation or every second generation [to maintain the desired qualities]”

Over the years, this project has involved many hundreds of colonies from across Canada and many collaborating beekeepers. The accomplishments to date would not have been possible without this backing from the beekeeping community. Founded on this teamwork, Foster and his colleagues have created a robust breeding technique that will let honey bee breeders create their own disease-and parasite-resistant stock, if desired. Boosted by their success, they’re now expanding the marker panel to cover traits like honey production and gentleness. Now more than ever, science and industry need to continue to work together to generate sustainable solutions to our modern problems.

This article appeared in the November 2017 issue of American bee journal.