Reply to: Do actin isoforms have unique functionalities at the protein level?

replying to J. Ervasti et al. Nature Communications https://doi.org/10.1038/s41467-025-57103-2 (2025)

The driving force behind our study¹ was to gain molecular and mechanistic insight into observations that different actin isoforms, in particular β- and γ-actin, exhibit markedly different localizations within a single cell², and to interrogate the functions of each isoform in a cellular context where both β- and γ-actin are expressed at endogenous levels. Many labs have postulated that different actin isoform networks interact with a distinct subset of actin-binding proteins, reviewed in ref. ³. Moreover, recent structural studies suggest that different actin isoforms can interact with the same partners in slightly different ways⁴. As such, the biological roles of each isoform are far from settled. If the β- and γ-actin isoforms perform specialized functions in cells that express both, i.e., the “normal” context, then re-positioning isoform networks by targeting formins to different regions of the cell would also cause actin interactors with an isoform preference to relocalize (mislocalize) along with their preferred network, as seen in Shah et al. ¹. Depending on the actin-binding protein and its precise function, spatiotemporal relocalization would be expected to have consequences for the cell. This is indeed what we found.

Our studies uncovered a large effect on cytokinesis in tissue culture cells, a powerful model in which to distinguish the roles of the β- versus γ-actin isoforms as the localization and function of each actin isoform network has been extensively characterized^5,6. Strikingly, while cytokinesis is strongly inhibited when the β- and/or γ-actin isoform networks are perturbed, some cells—even after 72 h—are still mononucleate and therefore were able to divide. Thus, a population of cells in culture either tolerate or adapt to the repositioning of the actin isoform networks, possibly due to an inherent redundancy in cytokinetic mechanisms. In an oft-cited keynote address, Ray Rappaport famously said “When I began working on cytokinesis, I thought I was tinkering with a beautifully made Swiss watch, but what I was really working on was an old Maine fishing boat engine: overbuilt, inefficient, never-failed, and repaired by simple measures.” ⁷. We couldn’t agree more.

Numerous examples of apparent discrepancies in results between tissue culture cell lines and whole animal studies have been previously noted. For example, several studies found that regulators (for example PI-3K-C2α and CEP55) of the ESCRT machinery, which localize to the site of abscission in the intercellular bridge, are required for successful cell division in tissue culture cells whilst humans or mice lacking expression develop to adulthood with only neurological phenotypes^8,9,10. We do not take the view that anyone is wrong, but that biology is complex and that each approach shines a different lens on the problem providing greater molecular and mechanistic insight that must ultimately be reconciled in the bigger picture.

Multiple studies have reported actin isoform-specific interactions through the variable N-terminal amino acids, including a recent structural study showing that the different N-termini of actin isoforms interact differently with non-muscle myosin II⁴. Based on this, we envisage that differences in interaction affinities with one actin isoform over another would allow for the preferential association of specific actin interactors (for instance non-muscle myosin II and DIAPH3) with one versus another actin isoform. However, this does not preclude the possibility that associations with the non-preferred isoform could occur, especially when it is present at higher concentrations and/or in the absence of the preferred substrate. Indeed, work examining the relationship between different myosins and actin isoforms identified subtle but not absolute differences¹¹. There are many such examples of this type of genetic compensation that can be found in the literature, particularly when gene families are concerned. It is possible that in the actb^c-g knockout mice, the increased levels of γ-actin protein allow it to functionally substitute for β-actin in many cellular contexts. How efficiently it does so is unclear, but the robust viability of the actb^c-g mice argues that in the absence of β-actin, increased expression of the γ-actin isoform can functionally substitute and provide genetic redundancy^12,13. This is consistent with observations whereby decreased expression of one actin isoform induces transcriptional response leading to a compensatory increase in the expression of another isoform. The extent to which this transcriptional response extends throughout a cell or organism is not fully understood and could exert considerable effect. In contrast, our experimental approach did not directly alter the levels of the different actin isoforms, rather the nucleator preferences were altered and this may consequently exert different pressures on the cell. Whilst is it possible that the chimeric formins we constructed alter the absolute levels of actin polymers at distinct sub-cellular locations, the crucial variable is that the actin polymers produced are of a biochemically different nature, being made from different actin isoforms. As the short timescale of the experiment provides little opportunity for the cells to adapt genetically, this system is likely to be less prone to compensation and more sensitive to changes in localization.

One obvious difference between the mice studies and that of Shah et al. is that the latter was performed in a cancer cell line, HeLa cells. Numerous studies have implicated changes in actin isoform levels, both β and γ-actin, with cancer^14,15. Whether this or a potential change in expression levels of actin cytoskeleton modifiers is the direct cause or consequence of transformation remains unknown. Regardless, the use of HeLa cells has shed light on the potential specialized functions of β and γ actin.

We do not interpret our results demonstrating specific functions for actin isoforms as a competing model to studies indicating that the coding sequences of the actin genes play an important role in actin gene function. Rather, taken together, the models describe multi-faceted regulatory mechanisms that speak to the complexity and exquisite attention the cell pays to the expression level and positioning of actin isoforms. If ever there was a biological example where the simplest case didn’t apply, actin regulation is it (at least in mammalian cells—models with only one actin gene notwithstanding).

We respectfully disagree with criticisms that claim using actin purified from chicken gizzards is inappropriate to study the γ-actin isoform and negates the specificity findings of Shah et al. Firstly, with respect to γ-actin: the γ-actin antibodies used in the study by Shah et al., were developed and extensively characterized in the Charpentier lab². The data in Fig. 1 of that study convincingly demonstrate that the γ-actin antibody specifically recognized γ-actin and does not recognize smooth muscle-γ-actin. Until evidence is provided to the contrary, we are confident that the γ-actin we detect in our experiments is cytoplasmic γ-actin and not smooth muscle-γ-actin. Secondly, with respect to concerns about the use of actin purified from chicken gizzards for some in vitro experiments, Shah et al. also demonstrated the same findings in parallel experiments carried out using actin purified from platelets, as well as using human cell lines that only express the β and γ cytoplasmic actin isoforms. All of these approaches yielded consistent complimentary results and therefore assertions that conclusions are based on the use of actin from gizzard are incorrect.

We hope that the study of Shah et al. will drive further experiments to resolve the fundamental question of why metazoans express multiple actin isoforms that exhibit dramatically different localizations within a cell. The simplest interpretation is that distinct actin isoforms perform specialized functions that can be independently regulated through their formins. If in the future further work concludes that these events are limited to cancer cells, a possibility raised, this would be a significant finding in furthering our understanding of mechanisms that underpin tumorigenesis.