Photo: Alison McAfee
HYGIENIC WORKER BEES are key players in the hive’s social immune system, removing sick and dying brood before the diseases they harbor can spread. 

Evolutionarily, hygienic behavior – or worker bees’ ability to detect, uncap, and removed sick brood – makes sense. It probably evolved because in the face of brood diseases, the hygienic colonies were more likely to fight off the disease and survive. But how exactly does it work?

Scientists have been trying to answer these questions since at least the 1960’s, when WC Rothenbuhler, at Iowa State University, started working out the genetic basis of this trait. By observing how different inbred lines of honey bees responded to dead brood, Rothenbuhler deduced that there were probably two genes controlling the trait – one gene for uncapping, and one for removing. About 40 years later, Marla Spivak showed that this model was too simplistic. In reality, there are at least seven locations on the genome that control the trait – probably many more – suggesting that hygienic behavior is more complex than Rothenbuhler ever imagined. Spivak’s foundational work also showed that bees’ sense of smell is indispensable for completing hygienic tasks.

In the March issue of American Bee JournalI described two odorants emitted from dead brood which we think are involved in hygienic behavior – β-ocimene and oleic acid. At the time, I had not actually done any behavioral experiments – we just knew that these were strong signals coming from dead pupae, and the roles these chemicals play in other contexts lends them probable cause. Hungry larvae emit β-ocimene like a volatile flag, waving to attract the attention of workers and alert them that the larva is starving. Oleic acid, on the other hand, induces necrophoretic behavior (i.e., transport of corpses away from the colony) in ants and termites, and avoidance behavior in cockroaches and crickets. Together, this led me to think that β-ocimene probably attracts hygienic workers, while oleic acid is the determinant death cue. Now, I have behavioral data to back that up.

To be clear, these odorants are not likely responsible for all hygienic behavior – just that which is induced by freeze-killed brood. Other diseases and parasites (chalkbrood, American foulbrood, and Varroa destructor) could easily stimulate hygienic behavior with different odorants, and some work by Spivak and Olav Rueppell suggests that is the case. Studying freeze-killed brood, though, is a useful model to begin with, because it’s a simpler system than one that involves real pathogens. It’s also much safer than cyanide.

We started with a quick and dirty behavioral test. We simply uncapped brood cells and added small amounts of the different odorants with a pipette (including a control treatment of hexane), replaced the frame and came back 3 hours later to see how many were removed. The hexane treatment acted as a reference point to compare β-ocimene and oleic acid to, since we know that it’s bad at inducing hygienic behaviour. It’s important to include treatments like this because even the simple act of uncapping a cell and adding something – anything – can induce low levels of hygienic behaviour even if it’s not a death or disease odor. In total, we uncapped about 3,000 brood cells spread over 10 different colonies, with each cell cap individually picked away with tweezers (I’m very grateful to have had the diligent help of another student that summer!). The results? Hygienic colonies indeed removed β-ocimene- and oleic acid-treated brood more often than hexane-treated brood, and they were better at removing brood than non-hygienic colonies. These results told us we were on the right track, but it’s arguably not a very realistic experiment. In reality, workers need to sense the odor through the wax cap, without being in direct contact with the brood. What’s more, since oleic acid is so oily, I doubted that it could become sufficiently airborne for its odor to penetrate the cap in the first place.

I wanted a better behavioral test. One that allowed me to add odorants to a brood cell without breaking its integrity. Without uncapping, without puncturing the wax, and without harming the pupa inside. I was stumped. And when I’m stumped, I talk to Heather.

Heather Higo has been in the business of bee science since before my own supervisor – now a full professor at the University of British Columbia – had even finished his undergrad. She’s also one of the leaders for the selective breeding project I wrote about in the article “Breeding a better bee: Three social immunity traits, one massive experiment.” In 2016, she received the Fred Rathje Memorial Award – a prestigious recognition for her contributions to improving the Canadian beekeeping industry. Not surprisingly, Heather instantly came up with a solution to my odor-introduction problem. “Can’t you just use a Jenter set?” she asked, as if wondering why I hadn’t thought of that before.

In case you’re not familiar, a Jenter set is a contraption normally meant for queen rearing. It includes square queen cages, complete with artificial comb cells. What’s special about them is that they have removable plugs that make up the comb cell bottoms. This set-up is meant to allow bee breeders to do graftless queen rearing – that is, the queen is caged in the artificial comb, she populates it with eggs, then the hatched larvae are popped out (still stuck to the plugs) and inserted into a queen cup. No grafting tools necessary. However, if the larvae are allowed to develop into pupae, then odorants can be added through the removable plugs and sealed again like a trap door, maintaining perfect integrity of the wax cell cap on the other side. It was a simple, brilliant idea.

After testing five colonies by this method, I concluded that oleic acid was indeed detectable even through the brood cell cap. This was a surprising result, because at the same time, we also know that oleic acid is bad at stimulating the nerves in bees’ antennae – hygienic or otherwise – when applied as a puff of air over an odor-soaked paper strip. Even at warmer, hive-realistic temperatures, the antennae are stimulated just barely above baseline. To me, this suggests that the workers must be extremely close to oleic acid in order to sense it. Intriguingly, the treatment that induced hygienic behavior most consistently was not oleic acid, but the odorant blend, supporting the notion that the two chemicals are working together as an attractor and death cue team.

Kaira Wagoner (at Olav Rueppell’s lab in North Carolina) recently found that a different, highly viscous compound (even more so than oleic acid) emitted from varroa-infested brood also induces hygienic behavior. As far as I know, they haven’t tested it in combination with other, more volatile odors, so whether my attractor and death cue hypothesis holds true for detecting varroa remains unknown. I haven’t come across any volatile odors that are emitted by varroa-infested brood in the first place, actually – the only other candidates I know of are Z-6-pentadecane, brood ester pheromone (a blend of many different compounds that normally acts as a contact pheromone), and possibly oleic acid itself – all of which are non-volatile chemicals. Perhaps varroa detection involves a different mechanism altogether, which would in part explain why the varroa-sensitive hygiene trait only partially overlaps with hygienic behavior.

Interestingly, Wagoner and Rueppell are currently developing a modified-hygienic test that involves spritzing their varroa compound on brood comb with an airbrush device (although this information is still not published, to my knowledge). Until now, we’ve always thought that the hygienic behaviour-stimulating compounds must be volatile in order to permeate through the brood cap. The fact that at least two oily compounds are definitively involved in hygienic behavior at all is changing the way we think about this trait. The most exciting part of our research, though, is seeing how the odorants interact with the bees’ odorant binding proteins.

Odorant binding proteins are produced in the bees’ antennae, and one of their jobs is to grab odorant molecules out of the air at the antennal pores and transport them to receptors on the olfactory nerves. This initiates the cascade of events that leads to the perception of smell. You might remember that odorant binding proteins (OBP16 and OBP18) are two of the nine biomarkers that we have determined can accurately predict a colony’s hygienic score – the more of the OBPs in the antennae, the more hygienic the workers tend to be. For years, I’ve been trying to manipulate how much of these OBPs are produced in the antennae, with my pinnacle goal being to see if this changes the bees’ sensitivity to certain odorants. Thus far, I have been unsuccessful; however, with the help of researchers in Florence, Italy, we did the next best thing.

Immacolata Iovinella and Paolo Pelosi specialize in producing, purifying, and characterizing OBPs. They trick bacteria into producing large amounts of these proteins, then they isolate the OBPs using specialized analytical techniques. After that, Iovinella and Pelosi can test, for example, how strongly a particular OBP binds a particular odorant. The rationale for these experiments is that the better an OBP binds an odorant, the better it probably is at getting small amounts of the odorant to stimulate the olfactory nerve and elicit a perceivable smell. When we tested how well OBP16 and OBP18 bound β-ocimene, oleic acid, and hexane, we found that their affinities closely match what we observe behaviorally. That is, OBP16 binds β-ocimene, OBP18 binds oleic acid, and neither OBP binds hexane. So, the odorants that strongly interact with at least one of the OBPs induce hygienic behavior, but the one that doesn’t bind to either OBP doesn’t induce hygienic behavior. Coincidence? Maybe, but the results of one more odorant are convincing me otherwise.

The last odorant we tested was phenethyl acetate. This molecule comes from chalkbrood mummies, and it’s important because years ago, Spivak found that it was a very good hygienic behavior inducer for their bees. I included it in my behavioral experiments, thinking it would be a good positive control (that is, a reference point for what a ‘good’ hygienic inducer looks like). What I found, though, was the exact opposite: it wasn’t a good inducer at all. In fact, it induced similar levels of hygienic behavior as hexane (the negative control), which was much lower than the other two odorants. Based on my enthusiasm, you might predict what the outcomes of the binding assays were: phenethyl acetate was a bad binder to the two OBPs we tested – again, very similar to hexane, and consistent with the idea that OBP16 and OBP18 help bees sense odorants that induce hygienic behavior. So why does our behavioral data not agree with previous studies? Our contradictory results don’t mean that either study was wrong; rather, it suggests that hygienic bees in different places, with different genetic lineages, are sensitive to different odors – adding to the complexity of an already complicated trait.

I am not the first one to postulate that there are probably many routes for bees to become hygienic. Reflecting on this, for such a multi-faceted trait it would be surprising (although possible) for there to be a single universal mechanism. Martin Beye – a researcher at the Heinrich-Heine University of Duesseldorf, and discoverer of the honey bee sex-determination locus – was among the first to bring to my attention how peculiar it is that there is so little agreement between the many differential expression studies that have been done on hygienic and non-hygienic bees. That is, many research groups have looked for differences in genetics and gene expression between hygienic and non-hygienic colonies, but very few of the differences they identified overlapped from study to study. This lack of overlap suggests that each selected population may have arrived at the same solution – hygienic behavior – through modifying different genes.

Considering OBPs alone, while we found that OBP16 and OBP18 were abundantly produced in hygienic workers, Mondet found OBP3 and OBP14, and Dell’Orco found OBP4 . . . and that’s just the first page of Google results. Of all these studies, ours has the biggest sample size by an order of magnitude (or two) and we were careful to include a wide variety of genetic lineages, so I still think OBP16 and OBP18 are (on average) the most robust hygienic behavior predictors. But this disagreement between studies suggests that OBP16 and OBP18 alone are probably not going to be applicable markers in honey bee populations around the world, and there could even be local populations that have taken a different evolutionary trajectory to become hygienic.

It could simply be that all those OBPs have some affinity to different death or disease odors and all are appropriate avenues to enable hygienic behavior, given a specific set of conditions. Answering this question, however, would likely require another PhD (or two). For now, I accept that while we may be one step closer to identifying the mechanism behind this hygienic behavior, there may be many other hygienic pathways to investigate in the future.

The full article appears in the February 2018 issue of American bee journal.

This research was also covered by New Scientist (“Smell of death tells undertaker bees it’s time to remove the corpses”) and Scientific American (“Scent of death: Honeybees use odors to detect deceased broods”).