A life in science, ending soon. David J. Linden
The Solomon H. Snyder Department of Neuroscience and the Kavli Neuroscience Discovery Institute. Johns Hopkins University School of Medicine.
Abstract
A diagnosis of terminal cancer has caused me to think about my life in science and the delight and surprise it has brought me and the experiments I would pursue if I had just a bit more time.
Main text
Seventeen months ago, after surgery to remove a large mass within my pericardial sac, the pathology report came back and the news was not good. I was diagnosed with synovial sarcoma of the heart and was given a lifespan estimate of 6 – 18 months. Because the tumor was embedded in my heart wall, it was impossible for the surgeon to remove all of the cancer cells without compromising cardiac function. My tumor, previously the size of a soda can, was reduced to approximately the size of an apricot. After the surgery, I received radiation and chemotherapy with all the unpleasantness that brings. However, the last dose of chemotherapy was delivered 9 months ago and so the side-effects have worn off. A recent CT scan detected no further growth of the inoperable portion of the tumor left behind in my heart and no metastases.
At the age of 60, I find myself in the odd and cognitively dissonant situation of having a terminal illness but feeling just fine and having no immediate threat to my health (for more thoughts on this see https://www.theatlantic.com/ideas/archive/2021/12/terminal-cancer-neuroscientist- prepares-death/621114/). From the outside, you wouldn’t know anything was wrong. Last week, I went to a jazz club with my wife. Today, I’m in the lab, chatting with students about their experiments. Next week, I’ll take a vacation on a lake in Canada with family. For now, life is good and I’m savoring its sweetness- trying to make the most of it in all domains- family, friends and work.
This situation is cognitively dissonant for my colleagues too. I get odd looks when passing in the hallways. Everyone is unfailingly polite but I can tell that they’re thinking, in the nicest possible way, “Wasn’t he supposed to be dead by now?”
For the last 30 years, I’ve run a neuroscience lab at the Johns Hopkins University School of Medicine. We use the techniques of cellular electrophysiology and live tissue imaging to reveal the cellular processes underlying the brain’s persistent response to experience. Such experience ranges from sensory-motor signals, to brain injury, to the long-lasting effects or drugs or hormones. This neural plasticity includes fine-scale changes in the structure of neurons (and other cells in the brain like glia and vascular endothelial cells) as well as functional changes in synapses and the ion channels that make neurons electrically excitable. Neural plasticity seems to underlie phenomena like learning/memory, addiction and the recovery of function after brain injury.
This work has been, and continues to be, a huge load of fun, mostly due to the smart, creative and kind people that surround me- my students and postdocs, colleagues and collaborators. I feel enormous gratitude for having been given a job where I can follow my own curiosity and do it in such a stimulating and supportive environment. Such intellectual freedom is a gift like no other. And the fact that this job allows me to have a parallel career, writing books about brain function for a general audience, only sweetens the pot. By any reasonable measure, I’ve had a great life in science. And I fully recognize that as a cis white man with educated parents, I’ve had a smoother ride on this path than many others.
These days, my lab still runs, but we’re winding down. We don’t take on any new projects or people or write any new grant proposals. I’m hoping to live long enough to help launch my current crop of splendid trainees onto the next stages of their careers. We have six ongoing projects that we hope to conclude and publish, and then we’ll be done. After 42 years in the lab (going back to my first undergraduate lab job in 1980) that’s a very difficult thing to say. There are always exciting new experiments to do and the idea that I will have to stop is challenging to my deepest sense of self. Who am I, really, if not an experimental scientist?
As scientists, we all struggle with a fundamental contradiction. We must be deeply present in the moment, to truly engage with people and data and ideas. At the same time, we must always be mentally projecting into the future- what happens if we get result X? What insights might emerge from applying technique Y to our question? How does this new finding reported from another lab change our outlook or our experimental design? We can never entirely “be here now.” We’re obligated to always be thinking about the next thing, and that’s OK.
I wish I could tell you that my impending demise has imparted some new or surprising revelation about how best to have a happy, fulfilling and productive life in science, but it hasn’t. The keys are very simple and well-known: to stay curious and embrace the uncertainty inherent to science, to enjoy and respect your colleagues (particularly your trainees) and to remember that our ultimate goal, to understand the natural world, transcends journals and grants and awards and will outlive all of us.
While it’s not “one weird trick,” there is a small and specific piece of advice that I’ll offer, or rather repeat, because others have said it before me. That is to periodically step back from the day-to- day challenges of experimental science, pull your head up, look around and think deeply about what scientific questions you’d really most like to answer. This is an exercise I’ve tried my best to do every few years, with varying success. And I’m doing it now at the end of my life, even though I won’t be able to translate these ideas into experiments.
So, if you’ll indulge me, I’d like to tell you about a question that would be at the top of my list if I were able to run my lab for a few more years.
When we see a schematic diagram of a neuron, we typically find a cell body, from which emerges multiple branching dendrites, the main information-receiving structures of neurons, where synapses are formed, and a single straight axon, the main information-sending structure, along which electrical signals called action potentials are conveyed. The axon is most often drawn a single unbranched line with a lone swelling on the end, the terminal bouton, from which neurotransmitter is released.
However, it’s been known since the days of the pioneering neuroanatomists like Camillo Golgi and Santiago Ramon y Cajal that axons can branch, sometimes ramifying repeatedly, allowing them to innervate many different targets. More recently, modern genetic and imaging techniques have converged to allow neuroanatomists to reconstruct the entire axonal and dendritic arbors of many neurons in the adult mouse brain (Winnubst et al., 2019). To me, one of the most interesting general findings from these efforts is just how common it is for single neurons in the brain to have massively branching axonal arbors that distribute over widely divergent brain regions. In many cases the many targets of single axons are not just anatomically distant but also functionally distinct, at least at our present level of understanding.
For some neurons, you might expect this. For example, single neurons of the locus coeruleus have axons that project widely in the brain. And these axons primarily release the neurotransmitter noradrenaline which, due to its action at slow, G-protein coupled receptors has correspondingly slow neuromodulatory effects reflecting states like wakefulness, vigilance and mood. It’s not surprising that these signals would be widely distributed across brain regions as they seem to convey more slowly-changing information that is useful for setting the general tone of computation across wide swathes of the brain. By contrast, single neurons that convey sensory or motor signals are more often working through fast receptors directly coupled to ion channels (like certain of those for the neurotransmitters glutamate and GABA and glycine) and are conveying rapidly changing information (on the scale of milliseconds to tens of milliseconds). One would not necessarily expect it to be useful for such rapidly-changing information to be widely distributed in the brain. Yet, modern neuroanatomy suggests that it is.
For many years, the dominant idea about axons has been that they function like slow electrical wires- you put an electrical signal in one end and, a bit later, it comes out at the other end largely unchanged. If the wire is branched, the signal will propagate through the branchpoints and thereby will be distributed throughout the axonal arbor.
In recent years, enormous effort has been expended to create connectomic maps of the brain’s axonal wiring in various organisms. The assumption underlying this work is that electrical signals are reliably propagated though these neuronal wires and so, like the schematic diagram of an electrical circuit, a wiring diagram of the brain will tell us where neural information flows. But is this assumption always true?
This is the question: In the adult, intact, unanesthetized brain, do action potentials, reliably propagate neural signals through highly branched axons, or, rather, is action potential failure at branchpoints common?
Surely, you’re thinking, this is something we already know. But we really don’t.
Axons can come in many forms. They have varying diameters that correlate with their conduction speeds. Axons that are wrapped in myelin protein from specialized glial cells called oligodendrocytes also conduct action potentials more quickly. Typically, a single axon emerges from the cell body of a neuron and its initial segment is specialized in a number of ways (most notably a high density of voltage-sensitive Na channels in the plasma membrane) to generate a brief, stereotyped electrical signal called the action potential (or spike). Once generated at the axon hillock, the action potential then propagates away from the cell body, towards the end of the axon (it also back-propagates towards the soma and dendrites using different mechanisms but that’s a story for another day). As action potentials propagate, they don’t decrement in amplitude, like ripples from a stone dropped in still pond, but rather regenerate their peak amplitude as they move. This is due to a clever positive feedback loop in which depolarization- opened Na channels allow for the influx of Na ions, thereby producing more depolarization and the spread of this depolarization to adjacent segments of axonal membrane. When action potentials reach specialized regions called active zones, they trigger a series of biochemical events that ultimately result in the probabilistic fusion of neurotransmitter-laden vesicles with the plasma membrane, thereby releasing neurotransmitter molecules into the extracellular space, conveying signals to other neurons (and some additional non-neuronal cells types as well). These active zones can either be localized at the very tip of an axon, in structures called terminal boutons, or at vesicle laden swellings distributed along the length of an axon called axonal varicosities. A single axon can have hundreds of these active zones.
So, what have experiments shown us about action potentials in real branching axons? In invertebrate preparations like the leech (Yau, 1976) or crayfish (Smith, 1980), the answers are mixed. Sometimes axonal branchpoint failure occurs and sometimes it doesn’t. It’s more likely to occur during high-frequency bursts of action potentials, which are common neuronal firing patterns in the neurons studied.
In the mammalian brain, there are some experiments in brain slices of juvenile rodent neocortex (Cox et al., 2000; Koester and Sakmann, 2000) or cerebellum (Foust et al, 2010) or using dispersed cultures of young neurons (Mackenzie and Murphy, 1998) showing reliable propagation of action potentials through axonal branchpoints. Other studies, using mammalian hippocampal organotypic cultures, have employed paired somatic recordings to infer axonal branchpoint failure, although the axons weren’t recorded directly (Debanne et al., 1997). But to my knowledge, the experiments we really need to settle this issue, involving measurements in the adult, unanesthetized brain of a mammal, with anatomy and neuromodulation intact, have not been done.
Fortunately, contemporary tools allow for this type of investigation, at least in those parts of the brain that lie within about 500 microns of the skull (Broussard and Petreanu, 2021). The experiments to address this question involve a technique for visualizing neuronal structure and function in the brains of intact mammals called in vivo two-photon microscopy and involve removing a section of the skull to implant a glass window and thereby provide optical access to the brain. It is now possible using genetic tools, to sparsely express fluorophores that dynamically report transmembrane voltage or internal free Ca concentration (as a slower proxy for transmembrane voltage, with certain important caveats) in genetically-defined types of neuronal axon and then make fast time-lapse movies in the awake, intact brains of mice harboring these reporters. Line-scanning 2-photon imaging at 500-2,000 Hz is sufficient to resolve actions potentials as they propagate through axonal branchpoints in order to determine if they are extinguished at the branchpoint, pass into one daughter branch or into both. Yet another interesting possibility is that the shape of action potentials are modified at branchpoints (as they are known to be in axon terminals during high-frequency bursts (Geiger and Jonas, 2000) and in response to certain neurotransmitters.
In my view, this is a crucial experiment to understand how information flows in neural circuits. When we observe that a neuron in area A contacts a neuron in area B can we assume that information reliably flows to the presynaptic side of the synaptic junction between A and B or not? My strong suspicion is that, like most things in biology the answer will be “it depends.” Maybe action potentials only reliably propagate though branchpoints in larger diameter or myelinated axons but fail in smaller, unmyelinated ones. Perhaps the probability or pattern of branchpoint failure is dependent upon the dynamics of information conveyed. One could imagine, for example, that during a high-frequency bursts, only the first few action potentials would propagate though the branchpoints while the later ones would fail in various ways. It is not unreasonable to hypothesize that the recent activity history of the axon (say, during the last 200 msec) might influence branchpoint failure. And finally, it would not be in the least surprising if axonal branchpoint failure were modified by the overall internal state of the animal (sleepy, hungry, anxious, pregnant, etc.) as well as by psychoactive drugs or brain temperature or anesthesia.
So, there you have it. It’s not an entirely novel idea and I’m sure some labs are pursuing aspects of it already. And of course, it builds upon the experimental and theoretical work of many others. I think it’s a fundamental question for understanding neural circuits and I hope you agree. I’d certainly walk down that path of inquiry if I had time on my side, as I hope you do.
Acknowledgments
Thanks to Indira Raman and Marion Winik for their insightful comments on an earlier draft of this manuscript.
References
Broussard, G.J. and Petreanu, L. (2021). Eavesdropping wires: recording activity in axons using
genetically encoded calcium indicators. J. Neurosci. Methods 360, 109251.
Cox, C.L., Denk, W., Tank, D.W. and Svoboda, K. (2000). Action potentials reliably invade axonal
arbors of rat neocortical neurons. Proc. Natl. Acad. Sci. USA 97, 9724-9728.
Debanne, D., Guerineau, N.C., Gahwiler, B. and Thompson, S.M. (1997). Action potential
propagation gated by an axonal IA-like K+ conductance in hippocampus. Nature 389, 286-289.
Foust, A., Popovic, M., Zecevic, D. and McCormick, D.A. (2010). Action potentials initiate in the axon initial segment and propagate through axon collaterals reliably in cerebellar Purkinje neurons. J. Neurosci. 30, 6891-6902.
Geiger, J.R.P. and Jonas, P. (2000). Dynamic control of presynaptic Ca2+ inflow by fast-inactivating K+ channels in hippocampal mossy fiber boutons. Neuron 28, 927-939.
Koester, H. and Sakmann, B. (2000). Calcium dynamics associated with action potentials in single nerve terminals of pyramidal cells in layer 2/3 of the young rat neocortex. J. Physiol. 529, 625-646.
Mackenzie, P.J. and Murphy, T.H. (1998). High safety factor for action potential conduction along axons but nto dendrites of cultured hippocampal and cortical neurons. J. Neurophysiol. 80, 2089- 2101.
Smith, D.O. (1980). Mechanisms of action potential propagation failure at sites of axon branching in the crayfish. J. Physiol. 301, 243-259.
Winnubst, J et al. (2019). Reconstruction of 1,000 projection neurons reveals new cell types and organization of long-range connectivity in the mouse brain. Cell 179, 268-281.
Yau, K.W. (1976). Receptive fields, geometry and conduction block of sensory neurons in the central nervous system of the leech. J. Physiol. 263, 513-538.