Vampire Worms

Illustration of the fabled Mongolian death worm, via Neatorama

What comes to mind when I say the word "worm"? If you're not acquainted with invertebrate neurobiology, it's probably that squiggly, segmented creature with five hearts that you accidentally cut in half with a spade once while digging around in your garden. If you're familiar with the nematode Caenorhabditis elegans, you may instead think of a tiny roundworm with exactly 959 somatic cells that develop in the same way every time in every worm.

Nope. What you should really be thinking is 'fangs.' Or, more precisely, 'tooth-like denticles,' though unfortunately, 'worms with tooth-like denticles' doesn't quite conjure up the same imagery as 'worms with fangs.'

Dorsal tooth outlined in blue, from Bento et al. 2010

To be fair, not every worm has a fang. But we can take advantage of the similarity between nematodes with and without fangs to learn how variations in their neural circuitry may contribute to differences in their feeding behavior.

C. elegans is a bacterivore. It crawls around in soil, foraging for bacteria and gulping them down into its bicameral pharynx. When the bacteria get to the posterior chamber, they run into a grinder, a hard structure that breaks them down mechanically, like the stones in the gizzards of chickens and herbivorous saurians.

Pristionchus pacificus, a predatory cousin of C. elegans, instead develops one or two teeth akin to the fangs of a snake. It can then use its dorsal tooth to puncture another worm and suck out its viscera. To help digest its prey in the absence of a grinder, P. pacificus plays host to a set of gut bacteria that do the work for it.

Now here’s the fun stuff: in the first 20-30 seconds of the video below, you can see the back-and-forth pumping motion as the pharynx moves food down the gut. In the last 30 seconds, when the C. elegans stops moving, the tooth and movement of the mouth muscles are more clear. Warning: graphic nematode-on-nematode violence.

If I were a nematode, a cannibalistic vampire worm would certainly be the stuff of nightmares. Luckily, I am 1600 times taller than P. pacificus is long.

Since these two types of feeding behaviors require different sets of muscle movements, it stands to reason that the neurons of the pharynx might be wired somewhat differently. Daniel Bumbarger, a postdoctoral fellow in the Sommer lab at the Max Planck Institute in Tübingen, borrowed from Google and graph theory to compare the anatomical circuitry of these two species. He took advantage of the fact that the pharyngeal nervous system in both of these species is almost a closed system, connected to the rest of the worm’s sensory and motor functions by only a single neuron. In a marvelous stroke of luck that probably made up for the tedious work of creating a 3000-slice EM reconstruction of the P. pacificus pharynx, it turned out that the identities and locations of the 20 C. elegans pharyngeal neurons are preserved almost perfectly in P. pacificus – it’s only their connectivity that differs (although we can argue about how connectivity may define identity, neuron identity in C. elegans is defined by lineage as well as function). The muscle cells pm1 and pm3, which contract rhythmically in predatory feeding as P. pacificus punctures its prey, are more highly innervated in P. pacificus. Conversely, pm7, the cell that normally drives the grinder in C. elegans, exists but receives no neural input at all in P. pacificus.

Comparison of pharynx cells in C. elegan and P. pacificus from Bumbarger et al. 2013

It's a classical comparative neuroanatomy study, but on a much smaller scale. The qualitative information is interesting, but what else can we do with a map of anatomical connections between neurons? The authors decided that they might be able to tease something more out of this connectivity map by integrating mathematical algorithms from other disciplines. In this case, the field they drew from was that of the search engine – specifically, Google. The PageRank system, named not for the fact that it ranks webpages, but for Google co-founder Larry Page, is meant to rank the importance of any given node in the network, based recursively on the number and weight of other nodes that link to it. To put it simply, if important people think you’re important, then you’re considered more important, which in turn affects the rank of people you think are important.

In comparing the two worm neural networks, the authors found that neurons controlling the behavior of the anterior pharynx (where the tooth is) were more important in P. pacificus, while neurons controlling the posterior pharynx (where the grinder is) were more important in C. elegans. What does this actually mean? Well, it suggests that more information is flowing through that particular part of the circuit, and that the physical behavior of the worm is most dominated by what those particular neurons say.

To look at where these specific interneurons were getting information from, the researchers then used a measure of focused centrality from graph theory (a branch of mathematics that focuses on the links between pairs of objects in a network). This revealed that neurons I1 and I2 receive a lot of indirect input from the motor neuron M4, and send out a whole lot of indirect input to the muscle cell pm4. The authors suggest that this higher proportion of indirect information flow in P. pacificus as opposed to C. elegans may correlate with more complex functions and the ability to switch between different behaviors.

It is a little difficult to understand what some of these methods could actually teach us about the system – do we really need to do a closeness centrality analysis to find that muscle cells that move the tooth receive more inputs in P. pacificus? We could just compare the number of synapses onto each of those cells between the two species. There is still a lot of work to be done here in finding what kinds of analyses might actually yield biologically relevant insights, but once we've identified the best methods in a small, isolated system such as this, we could expand their use to understanding indirect information flow through larger networks of neurons. We might also find something interesting if we take a look at the characteristics of the information being passed to these important-looking neurons. We can integrate this kind of information flow analysis with knowledge of whether the synapses are excitatory or inhibitory, the strength of each synapse, the changes that might arise with learning, and the effect of neuromodulators like dopamine and serotonin (which are especially important in regulating worm feeding behavior and which play large but poorly understood roles in the human alimentary system as well).

I'm hopeful that we will soon see functional data to correlate with the behavior, from ablation studies, optogenetic inhibition, in vivo electrophysiology, or even imaging with voltage indicators. In addition to testing the predictions made about the importance of certain cells, such data could shed light on various unanswered observations, such as why pm4, a muscle in the middle of the pharynx, and the gland cell, whose function is unknown, seem to be so central and important in the network analyses. And maybe we can even solve the mystery of why worms with fangs are so cool.


Bumbarger, D. J., Riebesell, M., Rödelsperger, C. & Sommer, R. J. System-wide Rewiring Underlies Behavioral Differences in Predatory and Bacterial-Feeding Nematodes. Cell 152, 109–119 (2013).

Bento, G., Ogawa, A. & Sommer, R. J. Co-option of the hormone-signalling module dafachronic acid-DAF-12 in nematode evolution. Nature 466, 494–497 (2010).