From spikes to intercellular waves: Tuning intercellular calcium signaling dynamics modulates organ size control

Information flow within and between cells depends significantly on calcium (Ca2+) signaling dynamics. However, the biophysical mechanisms that govern emergent patterns of Ca2+ signaling dynamics at the organ level remain elusive. Recent experimental studies in developing Drosophila wing imaginal discs demonstrate the emergence of four distinct patterns of Ca2+ activity: Ca2+ spikes, intercellular Ca2+ transients, tissue-level Ca2+ waves, and a global “fluttering” state. Here, we used a combination of computational modeling and experimental approaches to identify two different populations of cells within tissues that are connected by gap junction proteins. We term these two subpopulations “initiator cells,” defined by elevated levels of Phospholipase C (PLC) activity, and “standby cells,” which exhibit baseline activity. We found that the type and strength of hormonal stimulation and extent of gap junctional communication jointly determine the predominate class of Ca2+ signaling activity. Further, single-cell Ca2+ spikes are stimulated by insulin, while intercellular Ca2+ waves depend on Gαq activity. Our computational model successfully reproduces how the dynamics of Ca2+ transients varies during organ growth. Phenotypic analysis of perturbations to Gαq and insulin signaling support an integrated model of cytoplasmic Ca2+ as a dynamic reporter of overall tissue growth. Further, we show that perturbations to Ca2+ signaling tune the final size of organs. This work provides a platform to further study how organ size regulation emerges from the crosstalk between biochemical growth signals and heterogeneous cell signaling states.

1. Although details of the computational model seem to be explained in the supplement, it would be nice to have a brief discussion in the main text to help readers understand the model better. And many clarifications are needed: a. Fig 1B-C: It is unclear what do the spatial axes X, Y and L represent in the wing disc. b. Fig 1D: The meaning of each annotation should be further explained. Why GPCRs and RTKs use the same annotation, and same issue for PLCϒ and PLCβ. Both GJ and IP3R are represented by white box, are they the same thing? c. Fig 1D and   2. Line 314-343: the authors proposed two hypotheses to explain how wing disc growth regulates Ca2+ pattern. Then they used both computational models and experiments to eliminate one hypothesis. However, I found this argument lacks of evidence. a. What is the difference observed in these two hypotheses in computational model? As I can tell, the only difference is that they have different numbers of initiator cells at initial point. I don't understand how experiment helps eliminate one of them? How can experiment identify initiator cells? Experimental observations are shown in Fig 4C-4D, and where is the consistence stated in Line 337-339 being found comparing to the computational model? b. How parameters or any other conditions are perturbed in the two cases for computational model? The relation between number of initiator cells and disc size, and the relation between permeability and disc size in two cases should be given. c. In the first scenario, how to get decreasing fraction of initiator cells? If it is manipulated by decreasing VPLC, different decreasing rate of VPLC may lead to different observations. The authors should give a more systematic analysis to show how generally decreased number of initiator cells leads to decreased Ca2+ activity. d. All conclusions from simulations should be made from multiple independent repeats and their statistics, instead of a single simulation.
3. Line 381-383: The authors only showed insulin affect Ca2+ functions. It is completely unknown how insulin interacts with the system the authors modeled. The perturbations on two parameters for mimicking experimental manipulation on insulin make no sense. c. Information for Tissue model is not given in Supporting Information (Line 575). d. Information for multicellular model is not given in Supporting Information as stated in Line 579-580. e. Line 552: how many discs were collected and quantified for simulations? In simulation, how many real discs were used to calculate one computational domain? f. ctot value is not given in Table S1. Fig 5H. Arrows point to everywhere without any specific meaning.

I don't understand what was summarized in
6. Many aspects of the model formation need to be clarified: a. In the computational model, only randomness in PLC production is considered while other components are in deterministic description. Any rationale? The Ca2+ concentration may also be largely perturbed in different cells. b. In the single-cell model, only IP3 has degradation rate. Any explanations why degradation is not considered for IP3R and Ca2+? From the value of k5,p, The half-life for IP3 is 0.46 sec, which is extremely fast, explanation? The degradation term for p in Fig 1E should Fig 1E, it is not common to have a cube (second term) on the Michaelis-Menten term. Any interpretation? d. In Table S1, authors should point out clearly the source for each parameter. If the parameter is from this work, how they were selected should be stated. A systematic sensitivity analysis for all parameter is necessary. e. In one simulation, is the value of VPLC randomly perturbed in each cell, and is it dynamically perturbed during the simulation or only at the initial state?
The ranges of VPLC in each phenotype are not pointed out in Table S1 (as stated in line 198).
7. Explanations on functions of each experimental manipulation would be helpful to readers, such as expression region for each GAL4/UAS systems and function of GCaMP6f sensor.
8. When the authors point to Supporting Information, it would be nice to point out the specific sections.