Human formin FHOD3-mediated actin elongation is required for sarcomere integrity in cardiomyocytes

Reviewer #1 (Public review):

Summary:

Formins are complex proteins with multiple effects on actin filament assembly, including nucleation, capping with processive elongation, and bundling. Determining which of these activities is important for a given biological process and normal cellular function is a major challenge.

Here, the authors study the formin FHOD3L, which is essential for normal sarcomere assembly in muscle cells. They identify point mutants of FHOD3L in which formin nucleation and elongation/bundling activities are functionally separated. Expression of these mutants in neonatal rat ventricular myocytes shows that the control of actin filament elongation by formin is the major activity required for the normal assembly of functional sarcomeres.

Strengths:

The strength of this work is to combine sensitive biochemical assays with excellent work in neonatal rat ventricular myocytes. This combination of approaches is highly effective for analyzing the function of proteins with multiple activities in vitro.

Weaknesses:

FHOD3L does not seem to be the easiest formin to study because of its relatively weak nucleation activity and the short duration of capping events. This difficulty imposes rigorous biochemical analysis and careful interpretation of the data, which should be improved in this work.

Reviewer #2 (Public review):

This article elucidates the biochemical and cellular mechanisms by which the FHOD-family of formins, particularly FHOD3, contributes to sarcomere formation and contractility in cardiomyocytes. Formins are mainly known to nucleate and elongate actin filaments, with certain family members also exhibiting capping, severing, and bundling activities. Although FHOD3 has been well-established as essential for sarcomere assembly in cardiomyocytes, its precise biochemical functions and contributions to actin dynamics remain poorly understood.

In this study, the authors combine in vitro biochemical assays with cellular experiments to dissect FHOD3's roles in actin assembly and sarcomere formation. They demonstrate that FHOD3 nucleates actin filaments and acts as a transient elongator, pausing elongation after an initial burst of filament growth. Using separation-of-function mutants, they show that FHOD3's elongation activity - rather than its nucleation, capping, or bundling capabilities - is key for its sarcomeric function.

The experiments have been conducted rigorously and well-analyzed, and the paper is clearly written. The data presented support the authors' conclusions. I appreciate the detailed description and rationale behind the FHOD3 constructs used in this study.

However, I was somewhat surprised and a bit disappointed that while the authors conducted single-color TIRF experiments to observe the effects of FHOD3 on single filaments, they did not use fluorescently labeled FHOD3 to directly visualize its behavior. Incorporating such experiments would significantly strengthen their conclusions regarding FHOD3's bursts of elongation interspersed with capping activity. While I understand this might require a few additional weeks of experiments, these data would add considerable value by directly testing the proposed mechanism.

There is a typo in the word "required" in line number 30. The authors also use fit data to extract parameters in several panels (e.g., Figures 2b, 2d, 3a, and 3b). While these fit functions may be intuitive to actin experts, explicitly describing the fit functions in the figure legends or methods would greatly benefit the broader readership.

Reviewer #3 (Public review):

Valencia et al. aim to elucidate the biochemical and cellular mechanisms through which the human formin FHOD3 drives sarcomere assembly in cardiomyocytes. To do so, they combined rigorous in vitro biochemical assays with comprehensive in vivo characterizations, evaluating two wild-type FHOD3 isoforms and two function-separating mutants. Surprisingly, they found that both wild-type FHOD3 isoforms can nucleate new actin filaments, as well as elongate existing actin filaments in conjunction with profilin following barbed-end capping. This is in addition to FHOD3's proposed role as an actin bundler. Next, the authors asked whether FHOD3L promotes sarcomere assembly in cardiomyocytes through its activity in actin nucleation or rather elongation. With two function-separating mutants, the authors evaluated the numbers and morphology of sarcomeres, as well as their ability to beat and generate cardiac rhythm. The authors found that while the wild-type FHOD3L and the K1193L mutant can rescue sarcomere morphology and physiology, the GS-FH1 mutant fails to do so. Given that in GS-FH1 mainly elongation activity is compromised, the authors concluded that the elongation activity of FHOD3 is essential for its role in sarcomere assembly in cardiomyocytes, while its nucleator activity is dispensable. Overall, this important study provided a broadened view on the biochemical activities of FHOD3, and a pioneering view on a possible cellular mechanism of how FHOD3L drives sarcomere assembly. If further validated, this can lead to new mechanistic models of sarcomere assembly and potentially new therapeutic targets of cardiomyopathy.

The conclusions of this paper are mostly well supported by the comprehensive biochemical analyses performed by the authors. However, the sarcomere assembly defect phenotype in the GS-FH1 rescue condition requires further investigation, as the extremely low level of GS-FH1 signal in transfected cells in Figure 6A may reflect a failure of actin-binding by this construct in vivo, rather than its inability to drive elongation. Though the authors do show in Figure 6 that GS-FH1 can bind to normal-looking sarcomeres when they are present, this may be due to a lack of siRNA activity in these cells, such that endogenous FHOD3L is still present. In this possible scenario, GS-FH1 may dimerize with endogenous FHOD3L. The authors should demonstrate that GS-FH1 alone can indeed interact with existing actin filaments in vivo. While this has been clearly demonstrated in vitro, given the more complex biochemical environment in vivo where additional unknown binding partners may present, cautions should be made when extrapolating findings from the former to the latter.

Author response:

We thank the reviewers for their careful readings of our paper and their very positive assessment. Here we address the two major concerns they raised, referring to the revised version of the manuscript that will be submitted:

(1) Important points were raised regarding the brief elongation events we reported. The time resolution and noise in our system reduce the accuracy of the burst velocity measurements. To address this, we have reached out to a colleague who is set up to repeat these measurements with microfluidics-assisted TIRF. The noise should be greatly reduced and the system is also optimal for directly visualizing labeled FHOD3, as suggested. We hope this experimental approach will provide new insights.

In the meantime, we analyzed our data more closely. We were asked about the pauses we observe before bursts of elongation and how we know they are functionally relevant. The short answer is that we do not know. We reported them because they were so common: in three independent experiments with wild type FHOD3L-CT we analyzed a total of 20 filaments. We detected 112 dim regions and 97 of these were pause/burst events (~87%). Among the cases lacking a pause we include instances of apparent "double bursts" with no time for capping in between (which may be a time resolution issue) and some cases where the burst was in progress when data collection started. In the latter case, we cannot determine whether or not a pause was missed. We cannot rule out that this pause reflects an interaction with the surface but might expect the frequency to be lower if it were. In fact, we did detect pauses in the profilin-actin negative control but only 4 pauses were detected across 21 filaments analyzed compared to 97 pauses observed in the presence of wild type FHOD3L across 20 filaments analyzed. We will revise the text to make our conclusions about pauses more circumspect.

For comparison to our current data, we further analyzed the filaments in TIRF assays with no formin present. As the reviewers point out, inhomogeneities in filament intensity are normal. Thus, we examined any dim spots for pauses and/or bursts. We will report (future Figure 2G) that the velocity of growth of these dim spots was the same as the velocity of the rest of the filament. While our numbers may not be perfectly accurate due to the noise in our system, the difference of 3-4 fold increase versus no detectable change in rate is substantial and statistically different. In addition, we determined the number of dim spots per length of filament. We found a higher frequency of dim spots when FHOD3L-CT or FHOD3S-CT was present vs no formin, as will be shown in Figure 2 – figure supplement 1G and 2D.

We are convinced that the brief dim events we observed in the presence of FHOD3L-CT do, in fact, reflect formin-mediated elongation and hope that the reviewers concur. This does not preclude our interest in the microfluidics and two-color assays, which we will pursue in the future.

(2) The reviewers were concerned about the low protein levels in the GS-FH1 rescue experiments as reflected in the HA fluorescence intensity distributions shown in Fig. 5 – figure supplement 2A. While the scenario proposed could explain our observations with the GSFH1 rescues, it is quite complex and does not preclude the conclusion that the FH1 domain is critical. One limit of this scenario would be that the protein levels in the GS-FH1 cells reflect completely inactive protein, as opposed to FHOD3L that cannot elongate (by design). Given that the C-terminal half of the protein folds and functions and that the changes are made within an intrinsically disordered region, we do not favor this model. The reviewers suggest that the mutant protein detected in the few cells with (probably residual) sarcomeres could be stabilized, in part or entirely, by heterodimerization with residual endogenous wild type protein. We agree that heterodimerization is possible. The question becomes, how active is a heterodimer? If heterodimers have any activity, it seems far from sufficient to rescue sarcomere formation, suggesting that two functional FH1 domains are critical. To confirm this possibility, we would have to be able to determine whether the few sarcomeres present in these cases are residual and/or the new sarcomeres the low level of heterodimers could make. That said, we do not see evidence of correlation between protein levels and rescue at the level present in these cells (addressed below). Unfortunately, the proposed IP to test whether FHOD3L binds actin in vivo would only potentially report on filament side binding (both direct and indirect). It would not address whether the GS-FH1 mutant functions as a nucleator, elongator, bundler and/or capping protein in vivo.

If we assume that the protein present is active, the critical question that we can address is whether the phenotype is due to low protein levels or if the phenotype is due to loss of elongation activity by FHOD3L. To address this question, we revisited our data.

First, we plotted the distributions of the intensities of the cells we analyzed further, in addition to the automated readout of all the cells in the dish we originally presented (e.g. Fig. 4 – figure supplement 2A,B). These cells were selected randomly and, as should be the case, the distributions of their intensities agree well with the original distributions for the three different rescue constructs: FHOD3L, K1193L, and GS-FH1 (Fig. 6 – figure supplement 1A,B). We then asked whether there was any correlation between HA intensities with the sarcomere metrics. Consistent with in our pilot data, no correlation is evident in any of the three cases across the range of intensities we collected (400 – 2700 a.u.) (Fig. 6 – figure supplement 1C,D,E). We were originally satisfied with the GS-FH1 data, despite the low average intensity levels, because the intensities were well within the range that we established in pilot studies. These data reconfirm that the intensity levels are reasonable in a larger study.

To more specifically address the question of whether low HA fluorescence intensity is likely to reflect sufficient protein levels to build sarcomeres, we re-examined two data sets from the FHOD3L WT rescue data. We found that, by chance, the first replicate of data from the wild type rescue has a comparable intensity distribution to that of the GSFH1 rescues (580 +/- 261 / cell vs. 548 +/- 105 / cell). In addition, we collected all of the data from cells with intensity levels <720, selected to mimic the distribution of the GS-FH1 cells (Fig. 6 – figure supplement 3A). We then compared the sarcomere metrics (sarcomere number, sarcomere length, sarcomere width) between the full data set and the two low intensity subsets using statistical tests as reported for the rest of the cell biology data set:

· Sarcomere number is the only non-normal metric. We therefore used the Mann Whitney U test for each pairwise comparison, which shows no difference between all 3 WT distributions.

· We compared Z-line lengths by Student’s two-sample, unpaired t-test for each pairwise comparison, again finding no significant difference for all distributions.

· Sarcomere length shows a weakly significant difference (p=0.017 (compared to 0.033 for 3 treatment groups based on Bonferroni correction)) between the whole WT data set and bio rep 1, but no difference between the whole WT data set and the HA<720 group via Student’s two-sample, unpaired t-test.

An alternate statistical analysis approach, one-way ANOVA and Tukey post hoc tests, gave similar results. Thus, cells expressing wild type FHOD3L at levels comparable to levels detected in GS-FH1 mutant rescues, are fully rescued. Based on these findings we conclude that the expression levels in the GS-FH1 are high enough to rescue the FHOD3 knock down, supporting our conclusion that the defect is due to loss of elongation activity. We will add this analysis and discussion to the revised manuscript.

In future studies we will design less severe mutations to the FH1 domain. We hope to identify one with a strong effect on elongation and another with an intermediate effect. Once the best candidates are characterized in vitro, we will test them in our rescue experiments. If the strong mutant mimics the GS-FH1 rescue and the intermediate mutant is less severe, we will have strengthened our conclusion that elongation is a critical FHOD3L activity in sarcomere formation.

Additional improvements will be made to the manuscript based on recommendations we received from the reviewers.