The circadian clock is entrained to daily environmental cues. Integrin-linked signaling via actin cytoskeleton dynamics transduces physical niche cues from the extracellular matrix to myocardin-related transcription factor (MRTF)/serum response factor (SRF)-mediated transcription. The actin cytoskeleton organization and SRF-MRTF activity display diurnal oscillations. By interrogating disparate upstream events in the actin cytoskeleton-MRTF-A/SRF signaling cascade, we show that this pathway transduces extracellular niche cues to modulate circadian clock function. Pharmacological inhibition of MRTF-A/SRF by disrupting actin polymerization or blocking the ROCK kinase induced period lengthening with augmented clock amplitude, and genetic loss of function of Srf or Mrtfa mimicked the effects of treatment with actin-depolymerizing agents. In contrast, actin polymerization shortened circadian clock period and attenuated clock amplitude. Moreover, interfering with the cell–matrix interaction through blockade of integrin, inhibition of focal adhesion kinase (FAK, encoded by Ptk2) or attenuating matrix rigidity reduced the period length while enhancing amplitude. Mechanistically, we identified that the core clock repressors Per2, Nr1d1 and Nfil3 are direct transcriptional targets of MRTF-A/SRF in mediating actin dynamics-induced clock response. Collectively, our findings defined an integrin-actin cytoskeleton-MRTF/SRF pathway in linking clock entrainment with extracellular cues that might facilitate cellular adaptation to the physical niche environment.
The circadian clock has evolved as a time-keeping mechanism to anticipate and adapt to daily environmental changes essential for organismal fitness and survival (Finger et al., 2020; Takahashi, 2017). Under physiological conditions, a central clock pacemaker residing in suprachiasmatic nuclei (SCN) of the hypothalamus is entrained by daily light input and synchronizes clock circuits within peripheral tissues. In additional to SCN-driven signals for synchronization, cell-autonomous peripheral oscillators in tissues outside SCN respond to humoral signals and distinct tissue-specific cues (Dibner et al., 2010; Finger et al., 2020; Schibler and Sassone-Corsi, 2002). Synchronization of clock circuits in the body ensures coordinated temporal orchestration of daily physiological processes. Disruption of clock synchrony, such as shiftwork or jet lag, predispose to various disease risks, including the development of cancer (Kettner et al., 2014; Lin and Farkas, 2018; Mocellin et al., 2018) and metabolic disorders (Buxton et al., 2012; Karatsoreos et al., 2011; Scheer et al., 2009; Shi et al., 2013; Turek et al., 2005). Better understanding of the external or endogenous stimuli that entrain the circadian clock provides fundamental knowledge of the mechanisms that underlie environmental adaptation of the circadian clock (Finger et al., 2020).
The molecular circuit driving the ∼24-h rhythm of circadian oscillators is composed of a transcription-translation feedback loop (Takahashi, 2017). Circadian locomotor output kaput (CLOCK) and brain and muscle Arnt-like 1 (Bmal1, encoded by Arntl), the key transcriptional activators of this molecular-clock feedback loop, heterodimerize and activate the transcription of clock repressor proteins, the period proteins (Per1, Per2 and Per3) and cryptochromes (Cry1 and Cry2). Ensuing cytosolic accumulation, phosphorylation and nuclear translocation of the period and cryptochrome proteins inhibit Bmal1/CLOCK-dependent transcription via direct interaction with the heterodimer. This negative transcriptional feedback cycle coupled with translational control constitutes the core clock regulatory loop. An additional Rev-erbα/ROR-mediated Bmal1 transcriptional oscillation re-enforces the robustness of this core clock mechanism (Preitner et al., 2002).
Daily entrainment to environmental cues is essential for internal synchronization of the body clock system and adaptation to cyclic changes (Finger et al., 2020). Despite our current knowledge of the intricate molecular network driving clock oscillation and entrainment, how the clock responds and adapts to its immediate niche environment is yet to be addressed. Various extracellular physical or chemical cues, including cell adhesion to extracellular matrix (ECM) via integrin or growth factor activation by cell-surface receptors, are transduced intracellularly by a signaling cascade that involves actin cytoskeleton remodeling and a downstream transcriptional response mediated by serum response factor (SRF) in response to myocardin-related transcription factor (MRTF) activation (Olson and Nordheim, 2010; Posern and Treisman, 2006). Integrins, through direct interactions with specific ECM components, form focal adhesion complexes that connect the cellular physical environment with the intracellular actin cytoskeleton network through a myriad of signaling pathways (Romero et al., 2020). Activation of integrin-mediated intracellular signaling transduction, including focal adhesion-associated kinase (FAK, encoded by Ptk2), Rho-GTPases, ROCK kinase, and their associated effector molecules, transmits extracellular microenvironment cues to modulate actin polymerization via the formation of filamentous actin (F-actin) from monomeric globular actin (G-actin). Various growth factors or cytokines, including PDGF, TGF-β or Wnt, also elicit Rho-GTPase/ROCK signaling through cell-surface receptors to regulate actin cytoskeleton dynamics. Actin polymerization leads to the release of MRTFs (MRTF-A or MRTF-B, collectively MRTF-A/B) from sequestration by G-actin monomers, with subsequent nuclear translocation and activation of SRF-mediated transcription (Posern and Treisman, 2006). Nuclear shuttling of MRTF, in response to regulation of actin dynamics, promotes interaction with SRF on its cognate DNA-binding motif, the CArG box, to control target gene expression (Miano, 2003). MRTF/SRF-controlled genes participate in various physiological processes during tissue development, growth and remodeling involving cytoskeleton organization, such as matrix adhesion, migration, proliferation and differentiation (Gualdrini et al., 2016; Long et al., 2007; Olson and Nordheim, 2010; Wang et al., 2002).
Serum is known as a universal synchronizing signal of cellular clocks (Balsalobre et al., 1998). Serum-stimulated intracellular actin turnover entrains the liver clock (Gerber et al., 2013), whereas MRTF/SRF mediate the transcriptional response to actin dynamics-elicited clock entrainment (Esnault et al., 2014). Despite the role of intracellular actin dynamics and MRTF/SRF activity in mediating cellular interactions with the ECM, it remains unclear whether the extracellular niche impacts circadian clock function. In the current study, we employed pharmacological and genetic approaches targeting disparate steps of the integrin-actin cytoskeleton-MRTF/SRF signaling transduction to establish its role in modulating clock oscillation.
Pharmacological perturbations of actin polymerization modulate clock oscillation
Polymerization of monomeric actin, in response to upstream signals, is the key regulatory step that releases MRTF from G-actin sequestration to activate SRF-mediated transcription (Olson and Nordheim, 2010). To test whether signaling associated with actin dynamics modulates the circadian clock, we first determined the effects of pharmacological agents that disrupt or promote actin polymerization using mouse fibroblasts containing a Period-2 promoter-driven luciferase reporter knock-in (Per2::Luc) (Yoo et al., 2004). Continuous monitoring of bioluminescence activity of Per2::Luc fibroblasts revealed that cytochalasin D (Cyt D), a known actin-depolymerization compound (Hill et al., 1995; Wei et al., 2001), significantly altered clock cycling properties, as shown by average tracing for 5 days (Fig. 1A,B). Treatment with 1–5 μM Cyt D induced longer period length (Fig. 1C) and a dose-dependent increase in oscillation amplitude (Fig. 1D). The effect of Cyt D on actin organization was demonstrated using phalloidin immunofluorescence staining of F-actin stress fibers. Cyt D-treated fibroblasts displayed marked reductions of F-actin that persisted until 24 h after treatment, with cells viable at the two concentrations tested (Fig. 1E). The loss of actin polymerization is consistent with the rounded-up cell morphology observed (Fig. S1). As expected of Cyt D action in interfering with actin polymerization and consequent inhibition of MRTF/SRF activity, the expression of known MRTF/SRF target genes, connective tissue growth factor (Ctgf or Ccn2) and four and a half LIM-domain protein 1 (Fhl1), was markedly downregulated by Cyt D (Fig. 1F). Mrtfa transcript levels were also reduced, although Srf expression was not altered. Analysis of core clock genes revealed significant effects of Cyt D on inhibiting Clock, Per2 and Cry2 expression, without altering other components (Fig. 1G). Downregulation of clock genes by Cyt D suggests that the potential transcriptional regulation via actin cytoskeleton-related signaling and the dampening of positive clock components with an augmented negative regulatory loop might contribute to the period-lengthening effect of Cyt D. Using U2OS cells with the Per2::Luc reporter, we further examined the effects of Cyt D on clock modulation and found that it induced similar dose-dependent period lengthening (Fig. 1H,I), although with a reduced cycling amplitude at 5 μM (Fig. 1J).
We next determined whether additional molecules that inhibit actin polymerization, such as latrunculin B (Lat B), displayed shared clock-modulatory activity with Cyt D (Allingham et al., 2006). Indeed, treatment with 2 µM Lat B led to significant period lengthening (Fig. 2A–C), with a dose-dependent effect on increasing amplitude (Fig. 2D). Lat B appear to induce phase advance compared to Cyt D (Fig. 2B), although this was not directly assessed with re-synchronization. The loss of F-actin organization induced by Lat B was more rapid and robust than that induced by Cyt D, with nearly abolished F-actin staining at 30 min of treatment that persisted until 24 h (Fig. 2E). The effect of Lat B on the actin cytoskeleton was also reflected in the rapid change in cell morphology (Fig. S1). Lat B treatment resulted in a marked inhibition of SRF target genes at the concentrations tested, with near complete inhibition of Ctgf and strong suppression of Srf, vinculin (Vcl), Fhl1 and Fhl2 (Fig. 2F). Lat B treatment led to significant downregulation of Bmal1, whereas Per1, Dbp and Nr1d2 were induced at 5 μM concentration (Fig. 2G). In U2OS cells, Lat B induced comparable effects on increasing the period and amplitude (Fig. 2H–J). Additionally, we tested another compound that disrupts actin cytoskeleton, cytochalasin B, and found comparable effects in prolonging the period with augmentation of cycling amplitude in Per2::Luc mouse fibroblasts (Fig. S2).
Furthermore, we tested whether promoting actin polymerization via jasplakinolide (Jas) (Allingham et al., 2006), thereby stimulating MRTF/SRF-mediated transcription, affects clock function. In contrast to Cyt D, Jas resulted in dose-dependent period shortening upon treatment from 0.1 to 0.5 μM (Fig. 3A–C), with significantly decreased oscillation amplitude at 0.5 μM (Fig. 3D). Despite the limited effect of Jas on augmenting F-actin (Fig. 3E), potentially due to robust F-actin formation in fibroblasts under normal culture conditions, gene expression analysis revealed activation of SRF-mediated transcription, as shown by marked inductions of Vcl, Ctgf, Fhl1, Fhl2 and Srf itself (Fig. 3F). Compared to the suppression of clock repressors by Cyt D, Jas elicited upregulations of Bmal1, Per2, Cry2, Nr1d1 and Nr1d2 in a dose-dependent manner (Fig. 3G), although Per1 levels were reduced. Notably, Jas had a stronger effect on reducing the period length and clock amplitude in U2OS cells containing Per2::Luc starting at 0.5 μM (Fig. 3H–J). Thus, promoting actin polymerization exerts opposing effects on clock-cycling properties compared to actin depolymerization.
Genetic inhibition of SRF or MRTF function modulates clock properties
Intracellular signaling induced by actin dynamics ultimately results in a MRTF/SRF-mediated transcriptional activation response. Using siRNA-mediated silencing of these effectors of actin dynamics, we tested whether SRF or MRTF-A mediates the effects of actin cytoskeleton-modifying molecules on clock modulation. One out of three Srf siRNAs tested was effective in abolishing SRF protein expression (Fig. 4A). Srf silencing altered clock oscillatory properties similarly to treatment with actin-depolymerizing agents (Fig. 4B), with increased period length (Fig. 4C) and a tendency toward higher amplitude (Fig. 4D). We synchronized the cells by serum to determine the effect of SRF inhibition on clock genes at two circadian times, circadian time (CT) 12 and CT24. Mrtfa, Fhl1 and Fhl2 displayed circadian time-dependent regulation with higher expression at CT24 than that at CT12 (Fig. 4E). Knockdown of Srf markedly reduced Ctgf and Fhl1 expression at both time points, compared to that of the scrambled controls (SC), whereas Mrtfa and Fhl2 levels were significantly lower only at CT24. Analysis of clock genes revealed induction of Per1 and Cry1 expression at CT24 compared to their levels at CT12, whereas both were significantly downregulated by siSrf at CT24 (Fig. 4F). Furthermore, loss of MRTF-A by siRNA knockdown (siMrtf) resulted in similar effects as SRF inhibition. Two siRNAs against Mrtfa/b that were examined largely abolished MRTF-A protein expression (Fig. 4G), and both resulted in increased period length (Fig. 4H,I) with augmented amplitude (Fig. 4J), similar to Srf inhibition. These findings indicate that SRF and MRTF are involved in clock function and transcriptional regulation of the core clock circuit.
Inhibition of upstream signaling pathways that induce actin cytoskeleton remodeling modulates clock function
The Rho GTPases (RhoA, Rac1 and Cdc42) and Rho-associated kinase (ROCK) are upstream signaling molecules transmitting cell surface stimuli to modulate actin dynamics (Hill et al., 1995; Wei et al., 2001). Cell-surface ECM or biochemical cues, via integrin or receptor tyrosine kinases, activate Rho guanine nucleotide exchange factors (GEFs) and the Rho/ROCK-mediated kinase cascade to alter the G-actin:F-actin ratio (Esnault et al., 2014). We thus tested whether perturbing Rho GTPase signaling upstream of actin dynamics by a specific ROCK kinase inhibitor, Y27632, impacts clock function. As shown in Fig. 5A–D, Y27632 exerted similar effects as actin-disrupting molecules, with dose-dependent lengthening of period (Fig. 5C) accompanied by higher amplitude (Fig. 5D). Y27632 induced a robust but delayed effect on F-actin compared to actin-depolymerizing agents, as shown by the change in cell shape and attenuated phalloidin staining, which was most evident at 2 h following treatment (Fig. S3). As expected with the inhibition of MRTF/SRF activity, Y27632 markedly downregulated SRF targets with ∼80% lower mRNA expression of Srf, Vcl, Ctgf and Mrtfa, compared to that in DMSO-treated controls (Fig. 5E). Y27632 treatment also suppressed Per2, Nr1d1 and Nr1d2 expression, similarly to Cyt D treatment (Fig. 5F), whereas its induction of Cry1 and Cry2 were distinct. Thus, perturbing Rho GTPase signaling upstream of actin remodeling largely recapitulated the effects of actin depolymerization on clock modulation. In comparison, Y27632 effect on period length was moderate in U2OS cells, with significantly prolonged period and increased amplitude only observed at 50 μM (Fig. S4).
Integrin-mediated adhesion signaling and ECM stiffness modulate clock function
The ECM constitutes the immediate physical micro-environment in which cells reside, and the integrin-mediated focal adhesion complex links ECM components with intracellular actin cytoskeleton (Olson and Nordheim, 2010). Focal adhesion kinase (FAK), together with additional components of the focal adhesion complex, transduces ECM cues to modulate F-actin stress-fiber formation with activation of the MRTF/SRF transcriptional response (Romero et al., 2020). Based on the finding that actin dynamics-induced MRTF/SRF activity modulates the circadian clock, we postulated that blocking the integrin–ECM interaction and associated focal adhesion signaling might impact clock function and, thus, tested the effect of integrin blockade using an integrin αV-targeting cyclic peptide (Cyclo RGD). Compared to treatment with a control peptide with a scrambled sequence, RGD-treated cells displayed a tendency toward longer period (Fig. 6A,B) with significantly higher oscillation amplitude at 2 μM (Fig. 6C). Downregulation of SRF transcriptional targets, Srf, Vcl and Ctgf, revealed inhibition of MRTF/SRF signaling by Cyclo RGD (Fig. 6D), together with suppressed Bmal1 and Per2 expression, compared to control peptide-treated cells (Fig. 6E). To further test integrin-induced intracellular signaling through FAK, we used an siRNA-mediated knockdown of FAK (siFAK), with demonstrated efficiency in attenuating FAK protein levels (Fig. 6F). All three FAK siRNAs tested led to consistently prolonged clock periods (Fig. 6G,H), although siFAK treatment did not affect cycling amplitude (Fig. 6I). Activation of ERK mediates integrin-FAK downstream signaling (Schlaepfer et al., 1994), whereas the MEK/ERK inhibitor U0126 blocks 12-0-tetradecanoylphorbol-13-acetate (TPA) induction of clock gene expression (Akashi and Nishida, 2000). We thus determined whether blockade of FAK downstream signaling by U0126 modulated clock function in fibroblasts and U2OS cells (Fig. S5). In agreement with findings from FAK inhibition, U0126 increased clock period length, although this effect was significant in fibroblasts only at 0.5 μM (Fig. S5A,B) with a dose-dependent period lengthening observed in U2OS cells (Fig. S5D,E). In contrast, the cycling amplitude was attenuated by U0126 in fibroblasts (Fig. S5F).
To determine whether physical properties of the ECM have direct impact on clock function, we applied two engineered electrospun scaffolds with differing stiffness to Per2::Luc fibroblast cultures. A soft polycaprolactone (PCL) matrix with 19 kPa elastic modulus (EM) and a stiff polyether ketone ketone (PEKK) with EM of ∼300 kPa were selected to test ECM rigidities, as they are analogous to soft adipose tissue and bone-like tissue, respectively, as previously described (Maldonado et al., 2016, 2017, 2015). In comparison to cultures using the standard tissue-culture polystyrene plastic that is much more rigid with EM>1000 kPa as a control, cultures using both engineered scaffolds prolonged clock period length, with PCL inducing a longer period than PEKK (Fig. 7A,B). The clock-cycling amplitudes on these matrices were significantly augmented (Fig. 7C). Similar clock responses to matrix stiffness were obtained using U2OS cells containing a Bmal1 promoter-driven luciferase reporter, with the soft PCL resulting in period lengthening and PEKK augmenting clock amplitude (Fig. 7D–F). These effects on the clock were in line with findings from actin-disrupting chemicals, suggesting that reduced ECM tension could be transduced, potentially via actin cytoskeleton dynamics, to influence clock activity.
MRTF and SRF exert direct transcriptional control of core clock regulators
In response to an altered G-actin:F-actin ratio, MRTF translocates to the nucleus and binds SRF to activate gene transcription via a consensus CArG box motif response element (Sun et al., 2006). Although SRF association with cognate DNA-binding sites is largely constitutive, MRTF is responsive to extracellular stimulus-elicited Rho-ROCK-actin signaling (Esnault et al., 2014; Gualdrini et al., 2016). To test whether actin dynamics-induced MRTF/SRF activity exerts direct transcriptional control of the molecular clock circuit, we determined MRTF or SRF chromatin occupancy on core clock genes. MRTF/SRF DNA-binding elements (CArG boxes) were identified within gene regulatory regions containing proximal promoters (±2 kb of the transcription start site) of core clock genes using TRANSFAC (https://genexplain.com/transfac/). Using chromatin immunoprecipitation-quantitative PCR (ChIP-qPCR), ∼8- to 10-fold SRF enrichment on known CArG sites within α-actin (Acta1) and Vcl promoters was detected over the IgG control, as expected (Fig. 8A). Similar degrees of ∼6- to10-fold enrichment of SRF occupancy were found on clock gene regulatory regions within Per1, Per2, Nr1d1 and Nfil3, demonstrating that these clock genes are direct targets of SRF. We next performed immunoprecipitation with an anti-MRTF-A antibody to further determine MRTF/SRF-mediated transcriptional control of these genes. As shown in Fig. 8B, we observed a similar degree of MRTF-A association with chromatin for the clock gene regulatory regions examined. Interestingly, MRTF-A occupancy of the identified CArG site on the Per2 promoter was markedly higher than that of SRF. To explore whether loss of SRF or MRTF affects their chromatin occupancy on clock target genes, we generated stable cell lines containing SRF or MRTF-A/B shRNA knockdown (Liu et al., 2020). Loss of Srf by stable knockdown in Per2::Luc fibroblasts largely abolished binding to identified regulatory regions of clock gene targets, with the detected enrichment comparable to that of IgG control (Fig. 8C). Similar loss of MRTF-A binding to target promoters was observed in cells with stable expression of MRTF-A/B shRNA (Fig. 8D). Notably, the extent of MRTF-A enrichment was mostly attenuated in cells with scrambled control expression compared to normal Per2::Luc fibroblasts. To determine whether MRTF-A chromatin occupancy on identified clock gene promoters was circadian time-dependent, we performed ChIP-qPCR analysis at Zeitgeber time (ZT) 4 and ZT16 following serum-shock synchronization (Fig. 8E). Enrichment of MRTF-A binding to its known target, the α-actin promoter, was nearly twofold higher at ZT16 than at ZT4. Interestingly, its binding to Nr1d1 and Per2 promoters also displayed circadian dependency, albeit the enrichment was increased at ZT16 compared to ZT4, suggesting potential distinct kinetics of MRTF/SRF modulation of their gene targets. In fibroblasts with MRTF-A silencing, analysis of Nr1d1 and Per2 protein expression revealed their marked downregulation, providing support for the functional transcription control by MRTF-A (Fig. 8F). These findings thus identified direct MRTF/SRF transcriptional control of specific core clock components. Collectively, our results revealed a MRTF/SRF-mediated regulatory mechanism in controlling clock gene transcription, and this mechanism is responsive to extracellular niche signals transmitted via the Rho-ROCK-actin remodeling signaling cascade (Fig. 8G).
Through systematic interrogation of the distinct steps involved in integrin-actin cytoskeleton-MRTF/SRF signaling on circadian clock modulation, our current study established the role of this signal transduction cascade in linking the circadian clock response to its extracellular physical niche. The actin cytoskeleton-MRTF/SRF pathway is a key mechanism driving cellular developmental processes involving adhesion, migration, proliferation and differentiation (Gualdrini et al., 2016; Long et al., 2007; Olson and Nordheim, 2010). Our findings thus implicate the involvement of the circadian clock in fundamental aspects of cellular behavior in tissue development, growth and remodeling processes.
Modulation of MRTF/SRF activity, by interfering with actin cytoskeleton organization and its upstream signaling events, yielded largely consistent effect on key clock properties. Chemicals disrupting actin polymerization, including Cyt D, Lat B and Cyt B, augmented the cycling amplitude with longer period, whereas enhancing F-actin polymerization by Jas led to the opposite effects of a reduced amplitude with shortened period. Blocking events upstream of actin remodeling by inhibition of ROCK or FAK, or integrin blockade also led to increased period length and oscillation amplitude. However, the regulation of clock genes, particularly by pharmacological agents, was more complex. These pharmacological approaches might have non-specific effects beyond actin modulation, which could contribute to certain distinct responses of the clock gene modulation we observed in the study, such as Per2 expression. In addition, the transcriptional feedback mechanisms coupled with translational control govern molecular clock oscillation, rendering it challenging to distinguish direct transcriptional output versus indirect feedback regulation (Takahashi, 2017). The observed changes in clock period length or amplitude are likely the combined effect of the molecular network on the clock and not solely attributed to modulation of direct SRF/MRTF targets. Although identified as a direct gene target of SRF/MRTF, Per2 regulation in response to actin-depolymerizing agents was not uniform. Upon Cyt D treatment, the Clock gene was downregulated in addition to attenuated Per2 expression, suggesting the possibility that the clock modulatory activity involves Per2, whereas the exact properties could be influenced by the collective output of clock regulators. Nonetheless, genetic inhibition of FAK and transcriptional transducers of the ECM–actin dynamics-induced response, SRF and MRTF-A/B, corroborated the role of actin dynamics-related signaling events in clock modulation. The results further corroborated findings from pharmacological approaches demonstrating the direct effects of FAK, Srf and Mrtfa in modulating clock function. These interventions robustly influenced clock gene transcription, consistent with MRTF/SRF signaling activity, and we identified the clock repressors Per2, Cry2 and Nfil3 as direct transcriptional targets of SRF and MRTF.
Despite the largely consistent modulatory effects observed among distinct interventions of the integrin-actin cytoskeleton-MRTF/SRF cascade on clock activity, the transcriptional response observed among specific clock components, including Clock, Per1, Per2 and Nr1d1, differ. This could be due to the complexity of the interlocked transcriptional-translational feedback loop that drives the molecular clock circuitry, whereas the precise regulation of clock function by these interventions might depend on the distinct combined effects of the transcriptional controls involved. Notably, the period lengthening and amplitude-enhancing effects of actin-depolymerizing agents were comparable among the established Per2::Luc reporter lines tested, mouse fibroblasts and the U2OS osteosarcoma cell line. In contrast, jasplakinolide induced a stronger response at low concentrations in U2OS than in Per2::Luc mouse fibroblasts, potentially due to differing intrinsic sensitivity to actin polymerization between these cell types. The effect of blocking integrin-mediated adhesion signaling on the clock, by RGD treatment or FAK knockdown, are relatively moderate compared to actin-disrupting molecules. As a multitude of intracellular molecules and pathways are involved in transducing ECM cues to modulate actin dynamics-driven MRTF/SRF transactivation (Olson and Nordheim, 2010), it is conceivable that interference with cell–matrix interactions through RGD treatment or FAK knockdown alone might not be as robust as intervention of downstream actin remodeling or direct inhibition of MRTF/SRF activity.
As a timekeeping mechanism to anticipate and adapt to cyclic environmental stimuli, the circadian clock is entrained to extrinsic and endogenous timing cues (Finger et al., 2020). Serum stimulation is a strong entrainment signal for peripheral clocks (Balsalobre et al., 1998). Gerber et al. (2013) demonstrated that serum-induced actin remodeling and modulation of MRTF/SRF activity mediates clock entrainment in liver, and this pathway is responsive to upstream signaling events involving Rho kinase activation in fibroblasts (Esnault et al., 2014). Our new findings further extend previous observations by elucidating how the ECM-integrin-mediated signaling cascade transduces cellular physical niche cues to control clock gene expression, via the actin cytoskeleton dynamics-induced MRTF/SRF transcriptional response. Distinct integrins form linkages with the intracellular actin cytoskeleton through interactions with specific ECM components (Maartens and Brown, 2015; Romero et al., 2020), thereby transducing ECM cues through actin-MRTF/SRF-mediated signaling (Posern and Treisman, 2006). Our findings linking integrin-associated signaling with clock modulation suggest that integrin might transduce specific physical niche cues from the ECM to induce clock synchronization. Additionally, by culturing cells on matrices with defined physical properties, we demonstrated that softening of the ECM led to progressive period lengthening with an augmented cycling amplitude. Collectively, these findings suggest a potential circadian clock-sensing mechanism of the immediate microenvironment involving integrin-mediated interactions with the ECM and mechanical tension of matrix stiffness. These results are in line with recent reports from Yang et al. (2017) demonstrating that matrix stiffness modulates clock function in mammary epithelial cells, and that a softer three-dimensional matrix induced stronger clock gene oscillations than stiff two-dimensional matrices (Williams et al., 2018). It is notable that in mammary epithelial cells, ROCK kinase inhibition by Y27632 was found to augment clock amplitude as we observed in both Per2::Luc fibroblasts and U2OS cells (Yang et al., 2017). Interestingly, fibroblasts from the lung and mammary mesenchyme were reported to display inverse regulation by matrix rigidity (Williams et al., 2018), suggesting that the physical properties of ECM, including stiffness or composition, might have a distinct influence on clock functions in distinct cell types that might provide tissue-specific adaptations to unique extracellular environments (Streuli and Meng, 2019). Based on our findings, it will be intriguing to explore whether these cell type-dependent clock responses to matrix cues could be mediated by actin cytoskeleton-mediated MRTF/SRF transcriptional events, and they could be further interrogated in diverse cellular or in vivo models.
The ECM-actin dynamics-MRTF/SRF cascade in modulating clock oscillation we elucidated might function to coordinate stem-cell properties in tissue development or remodeling processes known to involve ECM–stem-cell interactions. Stem-cell quiescence and self-renewal properties are intimately linked with their physical niche environment (McBeath et al., 2004), whereas alterations in ECM stiffness and composition in development, tissue remodeling and aging directly impacts stem-cell behavior (Rayagiri et al., 2018; Stearns-Reider et al., 2017). Accumulating studies established the critical role of the circadian clock in conferring temporal control in stem-cell quiescence, activation, proliferation and differentiation in distinct tissue compartments involving epidermal or intestinal stem cells, or myogenic progenitors (Chatterjee et al., 2013, 2019, 2014; Janich et al., 2011; Karpowicz et al., 2013; Plikus et al., 2013). Our findings implicate an ECM-actin dynamics-clock regulatory axis that might coordinate stem-cell behaviors with ECM remodeling involved in tissue injury, regeneration and potentially aging-related stem-cell dysfunction. Clock function is required for coordinating the epidermal stem-cell activation response with environmental stimuli, which drives a dormant versus poised state, with disruption of clock oscillation resulting in dysregulation of tissue homeostasis and susceptibility to tumorigenesis (Janich et al., 2011). In skeletal muscle, the clock components Bmal1 and Rev-erbα exert antagonistic roles in muscle stem-cell proliferation and differentiation during muscle regeneration (Chatterjee et al., 2013, 2019, 2014; Gao et al., 2020). Intriguingly, in Duchene muscular dystrophy with repeated cycles of damage-induced degeneration and regeneration, failed regeneration was found to be driven by an asynchronously regenerating microenvironment (Dadgar et al., 2014). Notably, steroid treatment, known to synchronize clock oscillation (Balsalobre et al., 2000), ameliorated this phenotype, raising the intriguing possibility of a role of a connection between the ECM and clock oscillation in the regenerative behavior of muscle stem cells in this disease condition. An interplay between cellular physical niche signals and the clock circuit, with dynamic changes during development, regenerative growth or aging, might temporally orchestrate stem-cell function or coordinate distinct cell-type responses involved in morphogenesis and tissue homeostasis through clock modulation. Future studies are warranted to explore how an ECM-actin cytoskeleton-MRTF/SRF regulatory axis in modulating clock activity might apply to stem-cell biology in distinct tissue compartments in vivo, particularly in disease processes.
Our study revealed that small-molecule modulators of actin dynamics and related signaling pathways significantly impact circadian clock function, mediated by MRTF/SRF transcriptional activity. It is conceivable that agents such as RhoA/ROCK inhibitors or SRF modulators might have unintended biological activities on clock modulation. However, these compounds or chemical derivatives might also have potential applications in metabolic regulation or cancer therapy. With the recent effort in targeting clock for cancer and metabolic diseases (Chen et al., 2012; Cho et al., 2012; Dong et al., 2019; Oshima et al., 2019; Solt et al., 2012; Sulli et al., 2018), identification of the actin cytoskeleton-MRTF/SRF pathway in regulating the circadian clock might provide a novel mechanistic basis for future studies.
MATERIALS AND METHODS
Mouse primary fibroblasts that overcame replicative senescence were originally obtained from enzymatic digestion of the tail of mPER2::LUC-SV40 knock-in mice containing the mPer2 Luciferase-SV40 reporter (Welsh et al., 2004). These cells and a stable U2OS cell line containing the Per2::Luc reporter transgene were kind gifts from Dr Steve Kay, University of Southern California (Liu et al., 2008). Cells were maintained in Dulbecco's modified Eagle medium (Gibco DMEM; Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (Cytiva) and antibiotics in a standard tissue culture incubator at 37°C with 5% CO2, as described previously (Liu et al., 2008).
Chemicals and reagents
Cytochalasin D, cytochalasin B, latrunculin B, jasplakinolide and the ROCK inhibitor Y27632 were purchased from Cayman Chemicals. The integrin-blocking peptide Cyclo RGD (Arg-Gly-Asp-D-Phe-Val) and the control Cyclo (Arg-Ala-Asp-D-Phe-Val) were obtained from Enzo Life Sciences. The electrospun scaffolds PCL and PEKK were synthesized as described previously (Maldonado et al., 2015). The substrates were air plasma-treated and collagen-conjugated for cell adhesion, and their mechanical properties assessed by Young's modulus using atomic force microscopy (MFP-3D AFM; Asylum Research, Santa Barbara, CA). The scaffolds were sterilized and inserted in 24-well plates for cell culture.
Luminometry, bioluminescence recording and data analysis
Mouse fibroblasts or U2OS cells were seeded in 24-well plates and grown to confluency. Cells were synchronized for 2 h treatment with 0.1 μM dexamethasone (Sigma-Aldrich), followed by switching to Lumicycle bioluminescence-recording medium for continuous luminescence recording for 7 days, as described previously (Liu et al., 2008). The recording medium contains HEPES-buffered, air-equilibrated DMEM with 10 mM HEPES, 1.2 g/l NaHCO3, 25 U/ml penicillin, 25 g/ml streptomycin, 2% B-27 and 1 mM luciferin. The measurement of bioluminescence rhythms from Per2::Luc knock-in mouse fibroblasts or Per2::Luc stable transgenic U2OS cell cultures were conducted using the luminometer LumiCycle 96 (Actimetrics), as described previously (Chen et al., 2012; Liu et al., 2008; Ramanathan et al., 2018). Briefly, 24-well tissue culture plates were sealed with plastic film and placed inside a bacterial incubator maintained at 36°C and 0% CO2. Chemicals were added at the time of bioluminescence recording-medium change. Luminescence from each well was measured for ∼70 s at intervals of 10 min and recorded as counts/second. Seven days of real-time bioluminescence recording data were analyzed using the LumiCycle Analysis Program (Actimetrics) to determine the clock oscillation period, length amplitude and phase. Briefly, raw data following the first cycle from days 2 to 5 were fitted to a linear baseline, and the baseline-subtracted data (polynomial number=1) were fitted to a sine wave, from which period length, goodness-of-fit and damping constant were determined. For samples that showed persistent rhythms, a goodness-of-fit of >80% was usually achieved.
siRNA and shRNA transfection, lentiviral plasmid construction and infection
siRNAs were purchased from Integrated DNA Technologies (Table S4). Transfection was conducted using Lipofectamine RNAiMAX (Invitrogen). About 48–72 h post transfection, cells were collected for immunoblotting or used for Lumicycle analysis. Lentiviral vectors expressing SRF or MRTF-A/B shRNA were obtained from Open Biosystems, as previously described (Liu et al., 2020). Three siRNA or shRNAs for each target were tested for knockdown efficiency. To generate shRNA knockdown lines, recombinant lentiviruses were produced by transient transfection in 293T cells using the calcium-phosphate method. Infectious lentiviral particles were harvested at 48 h post transfection to infect Per2::Luc fibroblasts. Two days post infection, cells were selected with 2 μg/ml puromycin to obtain stable expression lines used for luminometry analyses. Lentiviral transduction efficiency was tested using a lentiviral-expressing GFP construct for GFP expression efficiency at nearly 95%.
Approximately 20–40 µg of total protein was resolved on SDS-PAGE gels and transferred to PVDF membranes for immunoblotting (Liu et al., 2020). Immunoblots were developed using a chemiluminescence kit (Pierce Biotechnology). The sources and dilutions used for primary and secondary antibodies are listed in Table S1.
Real-time quantitative PCR analysis
RNeasy miniprep kits (QIAGEN) were used to isolate total RNA from cells. cDNA was generated using qScript cDNA kit (Quanta Biosciences) and real-time quantitative PCR (RT-qPCR) was performed on an ABI Light Cycler with SYBR Green (Quanta Biosciences). Relative mRNA expression was determined using the comparative Ct method to normalize target genes with 36B4 (Rplp0) as an internal control. Primers were designed using experimentally validated sequences from PrimerBank. Primer sequences are listed in Table S2.
Phalloidin staining of F-actin
Fluorescence staining of actin was performed using Alexa Fluor 488-conjugated phalloidin, similarly as described previously (Liu et al., 2020). Briefly, cells were fixed with 3.7% formaldehyde and permeabilized using 0.1% Triton X-100. Alexa Fluor 488 phalloidin (10 µg/ml, Thermo Fisher Scientific) at a 1:200 dilution with DAPI (1:2000) was incubated for 30 min at room temperature. Fluorescence images were taken using an ECHO microscope.
Immunoprecipitation was performed using anti-SRF or anti-MKL1 (MRTF-A) antibodies or control rabbit IgG with Magnetic Protein A/G beads (Magna ChIP A/G kit, Millipore), as described previously (Chatterjee et al., 2013). Mouse myofibroblast chromatin was sonicated and purified following formaldehyde fixation. Real-time PCR was performed in triplicates using purified chromatin with specific primers for predicted Bmal1 binding E or E’-box elements identified within the gene regulatory regions. Primers flanking the known Bmal1 E-box in the Rev-erbα promoter were used as positive controls and TBP first exon primers as negative controls. Fold enrichment was expressed normalized to 1% of input over IgG. ChIP primer sequences are listed in Table S3.
Data are expressed as mean±s.d. Differences between groups were examined for statistical significance using unpaired two-tailed Student's t-test or one-way ANOVA using Tukey's post hoc test for multiple group comparisons as indicated using Prism by GraphPad. P<0.05 was considered statistically significant.
Conceptualization: K.M.; Methodology: K.M., X.X., W.L., J.N., M.Q., S.A.K.; Software: X.X.; Validation: X.X., W.L.; Formal analysis: K.M., X.X., W.L.; Investigation: X.X., W.L.; Resources: K.M., J.N., M.Q., S.A.K.; Data curation: X.X., W.L.; Writing - original draft: K.M.; Writing - review & editing: K.M., X.X., J.N., M.Q., S.A.K.; Supervision: K.M.; Project administration: K.M.; Funding acquisition: K.M.
K.M. is a faculty member supported by the National Cancer Institute-designated Comprehensive Cancer Center, City of Hope, National Cancer Center. This project was supported by the National Institutes of Health grant 1R01DK112794 to K.M. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Deposited in PMC for release after 12 months.
Peer review history
The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260094.reviewer-comments.pdf.
The authors declare no competing or financial interests.