Writing in Nature, Bill Jia, Sean Megason, Adam Cohen, and colleagues ask the question: how does the heart go from silent to beating during embryonic development? Using genetically encoded sensors and optogenetic tools, they captured the very first heartbeat of a zebrafish embryo and dissected the biophysical basis for its timing and spatial structure. We chatted with first author Bill Jia to hear the story behind the paper.
How did you come to join the labs of Adam Cohen and Sean Megason?
Early in my PhD, I developed a fascination with how cells compute as a collective. Embryonic development jumped out to me as an incredible example of this phenomenon – cells must talk to each other and then make their own decisions based on the inputs they receive, yet the outcomes of these decisions are often patterns and functions at a level of organisation higher than the individual. The Megason lab’s goal is to identify the conceptual rules by which this comes to be, so it was a perfect fit. However, to understand the complex messages that cells send to each other, it is critical to find the channels through which this information is passed – what exactly are cells measuring? In developmental biology, a major emphasis has traditionally been placed on signal transduction of chemical morphogens. But this is just one of many processes that constitute cellular physiology, each with its own time and length scales over which it can transmit or process information. The Cohen lab has pioneered optical approaches to measure and perturb different aspects of cellular physiology. It also focuses on neuroscience, another discipline concerned with complex and collective computation which happens to exemplify the idea that patterns with different spatiotemporal scales can carry information in biology. I wanted to bring together the distinct strengths and shared interests of these labs. This has put me in a unique position to study developmental cell-cell communication and pattern formation in media that have historically been relatively inaccessible to biological inquiry – bioelectricity, second messenger signalling, and others.
Movie 1: Calcium dynamics across zebrafish gastrulation, convergence, and extension, recorded using the sensor jGCaMP6s.
How did the project get started?
We started out with a broad survey of different physiological dynamics in zebrafish development, using biosensors to look for patterns that might be correlated between cells across time and space. Our efforts imaging cellular calcium were particularly successful (Movie 1)! Consistent with previous work from the Miller lab and others (1), we found an enormous diversity of calcium transients in different cell types over early development. We struggled for a while to decide which of these patterns would be worth following up on because there were simply so many. Adam proposed looking at the first heartbeats because the purpose of the calcium dynamics in this system was known, so it would just be a matter of figuring out how they appeared. At first, I was concerned that too much would have been previously studied, but the literature suggested that there were still some major mysteries to be solved.
What was known about the emergence of the first heartbeat in vertebrate embryos before your work?
The question of how the heartbeat emerges can be broken down into a few pieces. At what time in development does it happen? What are the roles of different cells in the process? How does the heart first gain its regular timing, and how is its function coordinated across the entire tissue? It was already known that periodic electrical and calcium activity emerges in the heart much earlier in development than it is needed (e.g. for blood flow) (2). It was also previously suggested, based on surgical separation, dissociation, or genetic perturbation of heart morphogenesis, that most early cardiomyocytes had the capacity to beat on their own and that the fastest cell would synchronise the tissue (3–5). But reports differed on the exact details of these observations, which we think mainly owes to two reasons.
First, individual heartbeats are very fast compared to the progression of development – it is difficult to ascertain from a short recording of a few beats how far along the process any given animal is. As we saw in our measurements, there could be very large changes in the dynamics over very short developmental periods. Second, disruption of the tissue geometry by the methods described is also highly disruptive to electrophysiology, making any observed activity preparation-dependent. One simple and instructive example is to imagine a cell that expects to be coupled to its neighbour by gap junctions – after dissection of that interface, it now has hemichannels exposed to the extracellular fluid instead, which might completely change its electrical behaviour. Notwithstanding these potentially contradictory previous results, there was also no proposed mechanism for how the heart gains its timing, whose quantitative properties are integral to the organ’s function.
Can you summarize the findings of the paper?
First, we used long, multiplexed recordings of genetically encoded calcium sensors to efficiently capture the transition from silent to beating (including the very first heartbeat!) in many zebrafish hearts (Movie 2). This was a dramatic and sudden switch, which in combination with subsequent experiments to probe biophysical and molecular mechanism gave us confidence that something important had happened at that very moment. We then looked at the statistics of the beats after the first one and found that the trajectories of average beat rate and beat rate variability were quite stereotyped. This lent itself to a simple mathematical abstraction describing how oscillations can emerge from an apparently silent system, called a stochastic saddle-node on invariant-circle (SNIC) bifurcation.
This model has two important predictions – first, highly periodic oscillations can be driven by random noise; second, big changes in fast dynamics (beats) can emerge from very small changes in an underlying control parameter describing a decreasing distance between a resting state and a “threshold” for a single oscillation. This second idea, of small changes in one quantity resulting in big changes in another, is analogous to a miniscule change in temperature across the freezing point turning water into ice – which is why we describe it as a phase transition. Similar phenomena have been observed in tissue mechanical properties in development (7, 8), and perhaps more await discovery.
The abstract mathematical resting state and threshold described have a biological basis in the membrane voltage dynamics of many types of electrically excitable cells, including cardiomyocytes. We tested whether this was relevant in the earliest heartbeats by performing voltage imaging and perturbing various ion transporters. In a panel of several genes important in the adult spontaneous cardiac action potential, we found that only the L-type voltage-gated calcium channel (VGCC) was required for both voltage and calcium oscillations. This is consistent with the model – the voltage threshold for activation of the VGCC could be the threshold described in the equations.
We then used light-activated ion channels (channelrhodopsins) to see if we could optogenetically cross the predicted threshold before any endogenous activity could be observed. As the model predicted, the heart became increasingly easy to excite as the first spontaneous beat approached. Furthermore, local excitations could propagate across the tissue, explaining how many cells seemed to turn on all at once from the first beat (Movie 3) – the cells are primed and ready to respond to their neighbours.
Our observations on how the heart’s earliest temporal patterns were driven by noise and how the spatial patterns were always synchronised demanded a more careful analysis of functional specialisation in individual cells. If the threshold crossing is random, what permits a particular cell to do it? We know that the adult heart has a defined pacemaker region, and we knew from previous work where these cells ought to be at the time of the first heartbeat. Does the beat start in the same place every time, and is this where the pacemaker will be? How does the previously observed widespread spontaneity play into all of this? We performed imaging of the calcium activity together with markers defining cardiomyocyte subpopulations, showing that the spatial origin was indeed variable, but tended to start away from the future pacemakers. Using optogenetics we were able to establish that a fast intrinsic beat rate was in fact the biophysical mechanism that set the spatial origin.
Together our results provide a picture of how many aspects of heartbeat initiation fit together. Random noise allows the heartbeats to emerge from variable positions, but the cells are synchronised by frequency-based competition. This is likely implemented molecularly by a gradual increase in activation sensitivity of VGCCs. This process does not seem to be tightly coupled to the earliest steps of pacemaker development, but the biophysical mechanisms are sufficient to ensure early coordination and periodicity of the heartbeat.
Were you surprised to find that the individual heart cells abruptly start beating all at once?
Yes! Before we did the experiments, the analogy I had in my head was a swarm of fireflies. I was expecting to see individual cells blinking separately, with some sort of synchronisation process occurring gradually over time. How the heartbeat initiation could hypothetically play out depends on the relative ordering of the development of excitability, electrical coupling, and spontaneous oscillations – in my picture, the coupling would have had to emerge last. There is evidence that connexins are expressed very early on in heart development (9), so maybe the tissue-scale initiation ought to have been less surprising. But the dependence of the spatiotemporal dynamics on each of these features is quantitative, so someone who had started with that prior assumption could have just as likely been wrong – we had to do the measurements to find out!
Do you think this bioelectrical phase transition model is found across all developing vertebrate hearts?
It is very possible, given that many features of the transition have been suggested by other observations in different vertebrate models. The same L-type calcium channels are known to be required for early cardiac activity in mice (10), and similar observations of lability in the spatial origin of the heartbeat have been seen in chicks and mice (3, 10). One major way other species might differ is in the length scale of synchronisation. In animals with larger hearts, it may be possible to observe multiple regions of synchronisation with their own distinct origin points.
The paper involves optogenetic manipulations and live functional imaging, producing a lot of beautiful videos. Can you tell us more about the techniques and tools you used?
Many of our conclusions rested on measuring how the dynamics of the heartbeat changed in their response to fast perturbations over the slow timescales of development. To do this, it was essential to simultaneously manipulate and image the electrical activity without disrupting the heart’s developmental trajectory. A key enabler of these experiments was FR-GECO1c, a novel far-red calcium sensor developed by the labs of Robert Campbell and Yi Shen (11). This sensor offered spectral compatibility with several blue light-activated ion channels (channelrhodopsins), which allowed us to target the heart as accurately as light can be patterned. Using these tools, we reversibly stimulated or silenced the heart at scales of milliseconds and micrometres, while watching its response in real time. There has been a recent explosion in new optogenetic tools for manipulating and visualising cellular physiology beyond membrane potential, including activation of canonical signalling pathways (12). I think that these will push the field of developmental biology forward tremendously, as they open up causative studies of the spatial and temporal couplings that ultimately underlie the progression of organ patterning.
Can you postulate the function of early electrical activity even before the heart is connected to the circulatory system?
It could be a type of developmental checkpoint for later steps in heart development – now that the heart has established its basic electrical function, it is ready to perform later steps like specification of cell subtypes or the complex steps of morphogenesis involved in chamber formation. The distribution of spontaneity could be a mechanism of robustness in this process – if one cell that was supposed to drive the beat fails to develop, another one can take its place. One particularly interesting hypothesis is that calcium signalling downstream of the beats supports development (10, 13). Given that calcium is an important second messenger molecule for many cellular processes, it is unclear whether it is the calcium fluxes downstream of the beat that matter and whether these must be deconvolved from some other cellular calcium dynamics. This is a problem that applies to all second messenger signalling pathways, so the developing heart might be a useful model to teach us more about the general mechanisms of cellular computation.
Did you have any particular result or eureka moment that has stuck with you?
The success of the quantitative fit of our data by the SNIC bifurcation model was very gratifying. It’s really satisfying to me when important features of a complex biological process can be distilled down into a model that is simple, yet precise enough to be written as a mathematical formula. But one of my favourite things about this story was how interdisciplinary the findings are – I really believe there is a little something for everyone in there. It’s super fun to see feedback from developmental biologists, cardiologists, and physicists expressing excitement about completely different parts of the results.
And the flipside: were there any moments of frustration or despair?
At times I thought that our findings were too simple or obvious given what was already known about the developing heart. But something I’ve come to appreciate is that this is very subjective and depends strongly on how much one is willing to make assumptions based on what is already known.
What’s next for this story? And what’s next for you personally?
I think there are two major future directions for this story. The first is to understand the implications of the early self-organising activity for the heart’s development, as we discussed. The second is to understand in more depth the molecular basis of the predicted biophysical mechanism. How are the VGCCs being sensitised? Is it because of changes in channel expression levels, relative intracellular and extracellular calcium levels, changes in resting potential by modulation of other ion pumps and channels, or something else? What is the source of the noise that seems to drive the timing? Personally, I am nearing the end of my PhD. I hope to continue along the lines of the first direction – how do properties of biological electricity instruct cell decision making and tissue patterning in development? I want to do so by opening my own lab but am also thinking about postdoctoral positions that might allow me to pursue these questions.
1. S. E. Webb, R. A. Fluck, A. L. Miller, Calcium signaling during the early development of medaka and zebrafish. Biochimie. 93, 2112–2125 (2011).
2. W. Rottbauer, K. Baker, Z. G. Wo, M.-A. P. K. Mohideen, H. F. Cantiello, M. C. Fishman, Growth and Function of the Embryonic Heart Depend upon the Cardiac-Specific L-Type Calcium Channel α1 Subunit. Dev. Cell. 1, 265–275 (2001).
3. T. Sakai, T. Yada, A. Hirota, H. Komuro, K. Kamino, A regional gradient of cardiac intrinsic rhythmicity depicted in embryonic cultured multiple hearts. Pflüg. Arch. 437, 61–69 (1998).
4. J. F. Reiter, J. Alexander, A. Rodaway, D. Yelon, R. Patient, N. Holder, D. Y. R. Stainier, Gata5 is required for the development of the heart and endoderm in zebrafish. Genes Dev. 13, 2983–2995 (1999).
5. R. L. DeHaan, Synchronization of pulsation rates in isolated cardiac myocytes. Exp. Cell Res. 70, 214–220 (1972).
6. B. Z. Jia, Y. Qi, J. D. Wong-Campos, S. G. Megason, A. E. Cohen, A bioelectrical phase transition patterns the first vertebrate heartbeats. Nature, 1–7 (2023).
7. A. Mongera, P. Rowghanian, H. J. Gustafson, E. Shelton, D. A. Kealhofer, E. K. Carn, F. Serwane, A. A. Lucio, J. Giammona, O. Campàs, A fluid-to-solid jamming transition underlies vertebrate body axis elongation. Nature. 561, 401–405 (2018).
8. N. I. Petridou, B. Corominas-Murtra, C.-P. Heisenberg, E. Hannezo, Rigidity percolation uncovers a structural basis for embryonic tissue phase transitions. Cell. 184, 1914-1928.e19 (2021).
9. N. C. Chi, R. M. Shaw, B. Jungblut, J. Huisken, T. Ferrer, R. Arnaout, I. Scott, D. Beis, T. Xiao, H. Baier, L. Y. Jan, M. Tristani-Firouzi, D. Y. R. Stainier, Genetic and Physiologic Dissection of the Vertebrate Cardiac Conduction System. PLOS Biol. 6, e109 (2008).
10. R. C. Tyser, A. M. Miranda, C. Chen, S. M. Davidson, S. Srinivas, P. R. Riley, Calcium handling precedes cardiac differentiation to initiate the first heartbeat. eLife. 5, e17113 (2016).
11. R. Dalangin, M. Drobizhev, R. S. Molina, A. Aggarwal, R. Patel, A. S. Abdelfattah, Y. Zhao, J. Wu, K. Podgorski, E. R. Schreiter, T. E. Hughes, R. E. Campbell, Y. Shen, bioRxiv, in press, doi:10.1101/2020.11.12.380089.
12. H. M. McNamara, B. Ramm, J. E. Toettcher, Synthetic developmental biology: New tools to deconstruct and rebuild developmental systems. Semin. Cell Dev. Biol. 141, 33–42 (2023).
13. N. D. Andersen, K. V. Ramachandran, M. M. Bao, M. L. Kirby, G. S. Pitt, M. R. Hutson, Calcium signaling regulates ventricular hypertrophy during development independent of contraction or blood flow. J. Mol. Cell. Cardiol. 80, 1–9 (2015).