I collect weird scientific objects. This post is about some truly unique material, which will likely never be made again:

Between the 1960’s and the 1990’s, a father-son team of Anthony B. Bubenik and George A. Bubenik made and explored a remarkable discovery. They studied deer antlers – huge structures which drop off every year and re-grow. This process is amazing in its own right, because it shows that large, adult mammals are capable of massive regeneration, growing bone, vasculature, innervation, and velvet (skin) at a rate of up to 1-1.5cm per day. Note that antlers are not horns, which are a much simpler structure and does not regenerate. This has massive implications for regenerative medicine and puts to bed the common idea that mammals simply can’t regenerate complex structures in adulthood.

But what the Bubeniks discovered is even more profound: trophic memory. What they found was that if an injury occurs at a particular point in the branched structure of the antler, it makes a small callus and heals; the rack will be shed as normal, and next year, a new rack will grow, with an ectopic tine (branch) at the location where the damage occurred in the previous year. This is one of my favorite examples when I teach developmental biology students, on the topic of “here are some things not in your developmental biology textbook”. Using the tools we normally use in the field – chemical gradients, gene-regulatory networks, molecular pathways – try to come up with a model of how the point of damage is sensed at one location of a complex structure, then the whole thing falls off, and the memory is somehow kept – where – in the growth plate on the scalp? And then months later, a new structure appears, with a pattern dictated not just by the emergent result of genetically-encoded protein production (hard enough to explain) but also by the previous physiological experience of cells that are no longer here, which tells bone growth dynamics to take an extra turn and grow out in a very specific place. The effect disappears after a few years and they go back to normal. Here’s an image from Daniel Lobo‘s and my paper on this topic which also discusses some other examples like crab claws etc.

This kind of pattern memory – what my group studies as a kind of learning process in the collective intelligence of cells operating in physiological and morphogenetic spaces, is a fascinating and highly important result because it reveals the dynamic, physiological plasticity of genetically-encoded hardware. Amazingly, almost no one talks, teaches, or writes about it.

In 2005, I emailed George A. Bubenik, then at the University of Guelph in Ontario, Canada, to discuss this phenomenon. Eventually he emailed me saying he needed to get rid of his collection of antlers – would I be interested in inheriting it? You bet I would! We received 13 boxes of meticulously labeled antlers. Here is a post-doc in my group, organizing the first batch:

Just think of how difficult and time-consuming each experiment was: you need to track each animal (and these are deer, not Drosophila!) and get a baseline for a few years, then make the notch in the bone and document the next 5-6 years of the trophic memory, and then a few years of normal growth after that. And it has to be done for many animals, to get statistical significance. The boxes were full of sets labeled “Lenny 1986” and the like. Imagine trying to get funding for this kind of study now – given the modern emphasis on rapid results, whose career could possibly support such a dataset? It will likely never be able to be duplicated. We had all the antlers CT-scanned by the Tufts Veterinary School. Some of them are hanging in the front hall of my lab. At that time, we had many discussions about the bioelectric, symmetry, hormonal, behavioral, and other aspects of deer antlers. He was an encyclopedic, enthusiastic, profound scholar, always happy to share his wisdom and extend into new directions. Sadly, George passed in 2018, but his legacy lives on, and I predict, will have a lot of impact.

We created a more tractable model system for the study of trophic memory, in planarian flatworms, which regenerate their entire body. In 2008, I asked Laryssa Wozniak, a Boston University student doing research in my lab, to re-cut in plain water the two-headed animals we created by a brief modulation of their bioelectric pattern memory. To my knowledge this hadn’t been done before (even though Thomas Hunt Morgan and others saw 2-headed planaria as early as 1903 or so), likely because it seemed so obvious that with a normal genome, if the ectopic head was removed, surely it would just go back to normal. By taking the notion of morphogenetic memory seriously we were able to find that their fragments once again regenerate as two-headed, despite their un-edited genetics, and Nestor Oviedo and Junji Morokuma in my lab subsequently studied this phenomenon, across many rounds of cutting. The 2-headedness is persistent across the animals’ normal reproductive mode (fission + regeneration) which means that it is stable across generations – a kind of unconventional inheritance. The reviewers made us take out discussions of the implications of this for evolution out of the primary paper, but I discussed it a bit here and more broadly, the importance for evolution of the competencies of living, agential medium (cell collectives) here.

That work was followed up by Fallon Durant in 2019 who actually discovered a third type of worm we can make, besides 1-headed and 2-headed: destabilized cryptic worms that can’t make up their mind and, like a bistable visual illusion, randomly make 1 or 2 heads when cut into pieces (in perpetuity); in fact, multiple pieces cut from the same worm flip a coin and make 1 or 2 heads: the cells within each organ share the same story of what they are building, but cells across multiple worms do not, and can disagree about whether they are part of a 1- or 2-headed animal.

Like in the deer, the large-scale target morphology can be revised – the pattern memory re-written – by transient physiological experience. The genetics sets the hardware with a default pattern outcome, but like any good cognitive system, it has a re-writable memory that learns from experience.

The memory for “how many heads should I have?” is stored in a bioelectric circuit (see here and here, characterized by Wendy Beane, Fallon Durant, and others in my group). Using voltage-sensitive dye imaging, we can now literally see the bioelectric pattern encoded in the tissue for what future regeneration must build – a basic kind of counterfactual memory that may serve as the evolutionary basis of advanced brains’ capacity for mental time travel – the ability to think about and remember things that are not true right now. Just like neuroscientists try to read out and decode the memories inside a living brain, we can now read and write (a little bit…) the anatomical goals and memories of the collective intelligence of morphogenesis.

The first time I presented this at a conference – genetically wild-type worms with a drastically different, rewritten, permanent, target morphology – someone stood up and said that this was impossible and “those animals can’t exist”. Here’s a video taken by Junji Morokuma, of them hanging out. Sometimes the 2 heads cooperate, sometimes they don’t; many behavioral neuroscience research programs can be envisioned about what information they share and what it’s like to be a creature with multiple brains.

While the 2-headed animals were permanently changed, our other example – planaria that can be made to grow heads belonging to a different species 100-150 million years in evolutionary distance – do go back to normal after a few months, like the deer antlers. See this paper for a discussion of this topic.

There is also a vertebrate model of this, found in our work with Jessica Whited: axolotl limbs, amputated multiple times, eventually stop regenerating. We don’t yet know if this has a canonical explanation (they run out of some kind of rate-limiting and non-renewing cell population) or a more interesting one based on learning mechanisms and revision of the target morphology in light of experience.

This kind of work, starting from observations (and meticulous follow-up) made of a random deer in a field who injured their antlers on a metal fence, and moving toward the molecular biophysics of pattern memory in cellular collectives, has huge implications for understanding the relationship between the genomic hardware and the (reprogrammable) physiological software that determines policies for the collective intelligence navigating anatomical spaces. Future regenerative medicine approaches will surely benefit from exploiting this capacity and targeting its many disorders.