CRY–BARs: Versatile light-gated molecular tools for the remodeling of membrane architectures

BAR (Bin, Amphiphysin, and Rvs) protein domains are responsible for the generation of membrane curvature and represent a critical mechanical component of cellular functions. Thus, BAR domains have great potential as components of membrane-remodeling tools for cell biologists. In this work, we describe the design and implementation of a family of versatile light-gated I-BAR (inverse BAR) domain containing tools derived from the fusion of the Arabidopsis thaliana cryptochrome 2 photoreceptor and I-BAR protein domains (“CRY–BARs”) with applications in the remodeling of membrane architectures and the control of cellular dynamics. By taking advantage of the intrinsic membrane-binding propensity of the I-BAR domain, CRY–BARs can be used for spatial and temporal control of cellular processes that require induction of membrane protrusions. Using cell lines and primary neuron cultures, we demonstrate here that the CRY–BAR optogenetic tool evokes membrane dynamic changes associated with cellular activity. Moreover, we provide evidence that ezrin, an actin and phosphatidylinositol 4,5-bisphosphate–binding protein, acts as a relay between the plasma membrane and the actin cytoskeleton and therefore is an important mediator of switch function. Overall, we propose that CRY–BARs hold promise as a useful addition to the optogenetic toolkit to study membrane remodeling in live cells.

BAR (Bin, Amphiphysin, and Rvs) protein domains are responsible for the generation of membrane curvature and represent a critical mechanical component of cellular functions. Thus, BAR domains have great potential as components of membrane-remodeling tools for cell biologists. In this work, we describe the design and implementation of a family of versatile light-gated I-BAR (inverse BAR) domain containing tools derived from the fusion of the Arabidopsis thaliana cryptochrome 2 photoreceptor and I-BAR protein domains ("CRY-BARs") with applications in the remodeling of membrane architectures and the control of cellular dynamics. By taking advantage of the intrinsic membrane-binding propensity of the I-BAR domain, CRY-BARs can be used for spatial and temporal control of cellular processes that require induction of membrane protrusions. Using cell lines and primary neuron cultures, we demonstrate here that the CRY-BAR optogenetic tool evokes membrane dynamic changes associated with cellular activity. Moreover, we provide evidence that ezrin, an actin and phosphatidylinositol 4,5-bisphosphate-binding protein, acts as a relay between the plasma membrane and the actin cytoskeleton and therefore is an important mediator of switch function. Overall, we propose that CRY-BARs hold promise as a useful addition to the optogenetic toolkit to study membrane remodeling in live cells.
Membrane-bound architectures, including filopodia, lamellipodia, and dendritic spines in neurons, are critical for a cell's ability to transmit and respond to extracellular cues. As such, chemists and biologists have developed numerous optogenetic and chemo-optogenetic tools to control cellular architectures via manipulation of plasma membrane dynamics (1)(2)(3). However, these tools have to date incorporated a relatively small fraction of the numerous proteins involved in the building and dismantling of these architectural features. In particular, nonenzymatic proteins have been overlooked: while numerous optogenetic strategies exist for enzyme-mediated control of membrane dynamics (4)(5)(6)(7)(8)(9), far fewer have incorporated nonenzymatic proteins as the basis of optogenetic switch action (10). As mechanical control of membrane architecture by nonenzymatic proteins is a critical component of cellular signaling (11), addressing this oversight will be a bridge to a more comprehensive understanding of cellular dynamics and function.
In recent years, the critical role of proteins involved in membrane curvature inducing and sensing has come to light (12). In particular, BAR (Bin, Amphiphysin, and Rvs) domaincontaining proteins possess diverse activities at the plasma membrane (13)(14)(15)(16)(17)(18)(19); as such, a heightened understanding of these roles has invited recent investigations of their suitability for control with optogenetic tools. In a recent work, Jones et al. (20) used a two-component approach (iLID/SspB) to design a light-activated switch enabling recruitment of the extended Fes-CIP4 homology BAR domain from FBP17 to the plasma membrane and a hybrid two-component approach (iLID/ SspB/pdDronpa1) enabling recruitment of the inverse BAR (I-BAR) domain from IRSp53 to the plasma membrane. These switches, designed to have minimal interaction with the cell membrane in the absence of light, promoted either positive (Fes-CIP4 homology BAR; FBP17) or negative (I-BAR; IRSp53) membrane curvature in response to light activation.
The MTSS1 (Missing in Metastasis 1) protein is considered the prototype of I-BAR proteins (21), but, to the best of our knowledge, has not previously been applied in an optogenetic context. In this work, we investigated the potential of the MTSS1 I-BAR domain to serve as an actuator of membrane architecture and plasma membrane dynamics. We demonstrate that the I-BAR domain from MTSS1, in conjunction with the Cry2 photoreceptor protein (22)(23)(24)(25), forms the basis of a versatile optogenetic approach ("CRY-BAR") for controlling membrane dynamics and cellular architecture. In contrast to previous approaches, we have not sought to minimize the interaction between our constructs and the plasma membrane. Instead, CRY-BAR combines the intrinsic membrane-binding affinity of the I-BAR domain (26) with the homo-oligomerizing capability of Cry2 (22), resulting in a membrane-localized switch that induces membrane remodeling in response to blue light. We also provide insight into the mode of action of CRY-BAR by showing that ezrin, a membrane-and cytoskeletal-relay protein, is linked to the ability of CRY-BAR to induce light-activated membrane remodeling and restriction of cellular dynamics by acting as a relay between CRY-BAR activation and the actin cytoskeleton.

Results and discussion
Increased MTSS1 activity has been associated with exerciseinduced enhancement of synaptic function (27). This prosynaptic plasticity effect of MTSS1 is attributed to the presence of an I-BAR domain within its structure. Moreover, I-BAR domains have also been shown to promote formation and stabilization of dendritic spines (Fig. 1A), leading to improved synaptic function and resilience against neurodegeneration (16). To gain insights into the molecular mechanisms that control membrane expansion associated with cell morphology changes, we fused the I-BAR domain of MTSS1 with a blue light-sensing domain (Cry2PHR; Fig. 1, B and C), creating a light-activatable I-BAR protein (CRY-BAR). For this investigation, we engineered four permutations of the CRY-BAR switch: I-BAR-Cry2-mCh, I-BAR-Cry2-mCh-WH2, Cry2-mCh-I-BAR, and Cry2-mCh-MTSS1 (Fig. 1C), where I-BAR = residues 1 to 250 of MTSS1, WH2 = residues 601 to 759 of MTSS1, and MTSS1 = the intact 759 residue protein. Each construct was confirmed by Sanger and NextGen sequencing and expressed predominantly at the expected molecular weight (Fig. S1).
CRY-BAR constructs were initially probed for their response to blue light in the presence of CIBN-CAAX, a membrane-localized binding partner to Cry2, in human embryonic kidney 293T (HEK293T) cells. As anticipated, light activation of the Cry2-mCh control resulted in rapid recruitment to the plasma membrane ( Figs. 2A and 3A). Lightactivated recruitment to CIBN-CAAX was also observed for all the I-BAR-containing constructs (Figs. 2 and 3, B and C). Interestingly, the I-BAR domain-containing constructs I-BAR-Cry2-mCh and Cry2-mCh-I-BAR, while exhibiting light-activated recruitment to the plasma membrane (Fig. 2, A and C), had significant prelocalization to the plasma membrane in the dark. We attributed this to the phosphatidylinositol 4,5-bisphosphate (PIP2)-binding propensity of the I-BAR domain (readily observed in the preillumination images in Fig. 2, A and C). Because of the significant prelocalization of these constructs to the plasma membrane, we then investigated whether they might exhibit light-activated recruitment, via Cry2 homo-oligomerization, in the absence of CIBN-CAAX. These experiments revealed significant light-activated clustering/membrane recruitment (Fig. 2, B and D) in the absence of CIBN-CAAX. By contrast, Cry2-mCh-MTSS1, which also accumulated in nonlight responsive clusters, did not exhibit light-activated membrane recruitment in the absence of CIBN-CAAX (50 of 50 cells pooled from six replicates; Fig. 2D). As a result, this construct was not pursued further.
Homo-oligomerization of the CRY-BAR constructs I-BAR-Cry2-mCh and Cry2-mCh-I-BAR results in membrane deformation and restriction of cellular dynamics (Movie S1, A and B). These effects are reversible in the dark. To demonstrate this using Cry2-mCh-I-BAR, we conducted a 10 min light activation sequence, followed by a 30 min observation period in the absence of blue light. Rapid restriction of membrane dynamics was observed during the 10 min blue light activation period, with full reversibility of membrane rounding achieved after 30 min in the absence of blue light ( Fig. 4 and Movie S2). We subsequently demonstrated that this effect can be selectively induced using localized illumination of the cell on a confocal microscope ( Fig. 5 and Movie S3). We also investigated whether the presence of a WH2 domain (present in I-BAR-Cry2-mCh-WH2 but not in I-BAR-Cry2-mCh and Cry2-mCh-I-BAR) impacted the membrane remodeling capabilities of the optogenetic switch. Light activation of an N-terminal I-BAR domain without a C-terminal WH2 domain (I-BAR-Cry2-mCh) promoted its accumulation in numerous filopodia-like protrusions that rapidly coalesced with continued light exposure (Fig. 6A). By contrast, a construct that paired an N-terminal I-BAR domain with a Cterminal WH2-containing domain (I-BAR-Cry2-mCh-WH2) did not accumulate in filopodia-like protrusions (Fig. 6). The interaction between WH2 and monomeric actin and associated complexes (28), which would disfavor the preassociation of I-BAR and PIP2 required for CRY-BAR function, is likely responsible for the observed failure to accumulate at the plasma membrane. Analogously, a recent investigation of the I-BAR protein IRSp53 revealed that the I-BAR domain of IRSp53, expressed as a fusion to GFP, localized to filopodia, whereas the C terminus of IRSp53 was entirely cytosolic (29). Taken together, these results illustrate the modular nature of I-BAR proteins, where the N-terminal I-BAR-containing Light-activated membrane remodeling domain is required for membrane binding and localization, and the WH2 domain is required for recruitment of actinpolymerizing components that promote protrusion formation and elongation (17). Only upon expression as isolated entities are the competing contributions of these domains to subcellular localization apparent. Finally, a side-by-side comparison of light activation of I-BAR-Cry2-mCh, I-BAR-Cry2-mCh-WH2, and Cry2-mCh-I-BAR demonstrated that while robust membrane recruitment is associated with light activation of Cry2-mCh-I-BAR and I-BAR-Cry2-mCh (Movie S4), Cry2-mCh-I-BAR induces a more robust effect on cell membranes, as evidenced by impact on total cell area (Fig. 4B). We attribute this to enhanced PIP2 binding and clustering capability of the I-BAR domain when in the C-terminal position of the optogenetic fusion protein.
We postulated that the dynamic membrane-remodeling activity observed with CRY-BARs might be due to its interaction with proteins that link the plasma membrane with the cytoskeleton. Ezrin is one such protein that has both lipidbinding and cytoskeleton-binding domains (30). We hypothesized that clustering of the PIP2-bound CRY-BARs might impact dynamics by also clustering ezrin, resulting in restriction of cytoskeletal-associated phenomenon such as membrane ruffling. To investigate this, we cotransfected Cry2-mCh-I-BAR with an ezrin-GFP fusion. In the presence of light, we observed colocalization of ezrin-GFP with Cry2-mCh-I-BAR clusters at the plasma membrane (Fig. 7). This effect was pronounced in the light, and absent in the dark, indicating that CRY-BAR activation actively restricts ezrin dynamics. CRY-BAR activation accompanied by ezrin sequestration also resulted in increased cell thickness (Fig. 8), with no such effect being observed in cells expressing the Cry2-mCh control. Accompanying live-cell experiments demonstrate that, while not exclusively so, the accumulation of ezrin near the cell membrane largely followed that of CRY-BAR during light activation and exhibits a close temporal relationship with CRY-BAR activation (Fig. S2). The CRY-BAR-ezrin link implies that CRY-BAR impacts actin dynamics as well. To explore this, we performed live-cell imaging of CRY-BAR light activation in the presence of a GFP-labeled actin (Fig. 9). Actin-rich protrusions were observed to coalesce upon CRY-BAR activation (Fig. 9A) but were not generally colocalized with CRY-BAR. Interestingly, in the dark, CRY-BAR is present in filopodia tips and at lamellipodial edges and interspersed with actin (Fig. 9A). However, upon light activation, their spatial relationship changes, as rounded CRY-BAR clusters and actin-rich regions at the cell membrane become distinct (Fig. 9A, inset and Movie S5).

Light-activated membrane remodeling
Having investigated the CRY-BAR response in immortalized cell culture, we next evaluated the capability of the sensor to report membrane expansion dynamics in neurons. For this, dissociated postnatal cortical neuron cultures prepared from newborn mice were transfected with Cry2-mCh-I-BAR, or cotransfected with CIBN-GFP-CAAX, according to protocol we developed for expressing optogenetic sensors in primary cultures (31). Light activation of Cry2-mCh-I-BAR, but not Cry2-mCh, resulted in elongation of neuronal processes that might indicate early stage spinogenesis (Fig. 10, Movies S6 and S7). There was no apparent benefit to activation of Cry2-mCh-I-BAR in the presence of CIBN-GFP-CAAX, providing additional evidence that activation of Cry2-mCh-I-BAR alone is sufficient for effecting membrane dynamics (Fig. 10C). Accordingly, whole-cell illumination of neurons transfected with only Cry2-mCh-I-BAR results in abundant membranous cluster formation throughout neuronal processes (Fig. S3A). Notably, oligomerization of I-BAR domain proteins is essential for their membrane-remodeling activities (32). In the context of CRY-BAR, optimal membrane remodeling is achieved via the homo-oligomerization of Cry2. By contrast, recruitment of CRY-BAR to CIBN-GFP-CAAX, while not completely detrimental to CRY-BAR activity, results in less membrane activity than activation of CRY-BAR alone (Fig. 10C). This is presumably because of a combination of reduced CRY2 homooligomerization in the presence of CIB, a phenomenon that has been previously described (22), and possible orientation effects of the Cry2-CIB interaction on the I-BAR domain. CIBN-GFP-CAAX has a similar membrane distribution to CRY-BAR and could interfere with CRY-BAR activity at the cell membrane (Fig. S4).
Finally, using localized illumination conditions, we demonstrated that Cry2-mCh-I-BAR can be selectively activated in neuronal processes (Fig. 11 and Movie S8). Taken together, this work provides evidence that the Cry2-mCh-I-BAR molecular optogenetic tool is suitable for the targeted manipulation of cytoskeletal structures and dynamics at the plasma membrane. To our knowledge, this switch represents the first application of the I-BAR domain from MTSS1 in an optogenetic switch. In comparison to prior work in this area, CRY-BAR has the advantage of being a single-component switch, which can reduce experimental complexity and facilitate its implementation into transgenic model organisms. We note that the mechanisms of I-BAR domain proteins are not completely understood, and there are structural and mechanistic differences in the I-BAR domains across different proteins (26). As a result, one can reasonably anticipate that the incorporation of I-BAR domains from different I-BAR-containing proteins and their orthologs could produce different experimental phenotypes within the same optogenetic background. This intriguing possibility illustrates the value of pursuing multiple optogenetic variations of BAR domain proteins. Potential applications of CRY-BARs include sitedirected mutagenesis of the I-BAR domain, which could provide valuable structure-function information regarding the mechanism of the I-BAR-plasma membrane interaction, and investigation of membrane deformation-dependent cell signaling pathways, such as calcium signaling (33).

Conclusion
In this report, we describe the development of a family of optogenetic switches, collectively named CRY-BAR, that comprise a versatile platform for controlling membrane dynamics in live cells with high spatial and temporal resolution. These switches combine homo-and heterooligomerization of the Cry2-CIB photoreceptor system with the innate PIP2-binding affinity of the I-BAR domain. As their function varies depending on the presence of various functional elements, such as WH2-binding domains, we anticipate that a modular approach can be undertaken to adapt these optogenetic switches for other applications, including the recruitment of enzyme and receptor-activating and -inhibitory domains. Finally, CRY-BARs are suitable tools to study membrane dynamics not only in immortalized cells but also in sensitive and difficult to transfect primary cultures, such as Light-activated membrane remodeling neurons. Therefore, the CRY-BAR optogenetic switches we report here are expected to have wide applicability for investigating cellular processes associated with membrane dynamics in a variety of experimental paradigms.

Cell lines and transfection
HEK293T cells (passage 8) used for these experiments were purchased from American Type Culture Collection and were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS) and 1% penicillinstreptomycin. Cultures were transfected at 70% confluency with the Lipofectamine 3000 reagent (Invitrogen) following the manufacturer's suggested protocols. Briefly, for dual transfections in 35 mm glass bottom dishes for cell imaging or 6-well culture plates for lysis, plasmid DNA was combined in a 1:1 ratio (1250 ng per plasmid) in 125 μl of Opti-MEM, followed by the addition of 5 μl of P3000 reagent. For single transfections, 2500 ng of plasmid DNA was used per transfection. In a separate vial, 3.75 μl of Lipofectamine 3000 were added to 125 μl of Opti-MEM. The two 125 μl solutions were combined and allowed to incubate at room temperature for 10 min, followed by dropwise addition to cell culture. For ezrin-GFP and actin-GFP cotransfection with CRY-BAR, 2000 ng of CRY-BAR DNA and 1000 ng of GFP DNA were added to 100 μl of DMEM along with subsequent addition of 3 μl of CalFectin (SignaGen Laboratories). After 10 min of incubation at room temperature, the solution was added dropwise to cell cultures. Each transfection solution remained on cells overnight. Transfected cells were maintained at 37 C and 5% CO 2 in a humidified tissue culture incubator, in culture medium consisting of DMEM supplemented with 10% FBS and 1% penicillin-streptomycin.

Neuron cultures and transfection
Postnatal-dissociated cortical neuron cultures were prepared as previously described (31,36), from newborn B6 mice. Neurons were plated into 35 mm glass bottom Petri dishes at a 1 million/ml density in culture medium consisting of Basal Medium Eagle supplemented with 10% bovine calf serum and 1% penicillin-streptomycin. On day in vitro 2 (DIV2), culture medium was changed to Neurobasal A medium, supplemented with B27-plus reagent (Invitrogen), Glutamax, and 1% penicillin-streptomycin. Neurons were transfected with the CryBAR optogenetic system (6 μg plasmid/plate) on DIV5 using Lipofectamine LTX reagent. On DIV7, culture medium was removed and neurons were placed in imaging solution (Mg-free HEPES-buffered artificial cerebrospinal fluid (37)) Live-cell imaging was performed before and after illumination using a Leica DMi8 Live Cell Imaging System. Animal use protocols were approved by East Carolina University Institutional Animal Care and Use Committee.

Fixed cell experiments
Immediately prior to fixation, transfected HEK293T cells were either kept in dark conditions or continually illuminated with LED blue light (Sunlite LED Par30 Reflector, item #80021, 4 W, 120 V, 66.21 μmol/s/m 2 ; placed 10 cm from cell culture dishes) for 5 min. Cells were washed with Dulbecco's PBS (DPBS) (with calcium and magnesium; 1× 1 ml), then fixed for 10 min at room temperature with prewarmed 4% paraformaldehyde solution in DPBS (37 C; prepared from 16% paraformaldehyde [Electron Microscopy Sciences]). Following fixation, cells were washed with DPBS and then stored in DPBS at 4 C.

Live-cell experiments
Transfected HEK293T cell media were replaced with 10% FBS in Leibovitz's L-15 medium. Cells were allowed to equilibrate in the live-cell incubation system (OKOLab) for 10 min prior to beginning the illumination sequence.

Confocal microscopy
Confocal images of fixed cells were collected with a Zeiss LSM 700 laser scanning microscope using ZEN Black 2012 software. Ezrin-GFP and actin-GFP live-cell images were collected with a Zeiss LSM 700 laser scanning microscope equipped with a stage-top incubator and Light-activated membrane remodeling ZEN Black 2012 software. Fluorescence images were colorized and overlaid using FIJI software (38) equipped with Bio Formats (39).

Widefield microscopy
A Leica DMi8 Live Cell Imaging System, equipped with an OKOLab stage-top live cell incubation system, LASX software, Leica HCX PL APO 63×/1.40 to 0.60 numerical objective oil objective, Lumencor LED light engine, CTRadvanced+ power supply, and a Leica DFC900 GT camera, was used to acquire images. Exposure times were set at 200 ms (mCherry, 553 nm) and 50 ms (GFP, 480 nm), with LED light sources at 50% power, and images were acquired every 30 s over specified time course.

Western blotting
Transfected HEK293T cells were lysed with 200 μl of M-PER lysis buffer (Thermo Fisher Scientific) containing 1× Halt protease-phosphatase inhibitor cocktail (Thermo Fisher Scientific). After 10 min on ice, lysates were collected and centrifuged for 15 min (94 rcf; 4 C). The supernatants were combined with Laemmli SDS sample buffer (Alfa Aesar) and incubated at 65 C for 10 min. The resulting lysates were subjected to electrophoresis on a 10% SDS-PAGE gel and then transferred onto polyvinylidene fluoride membranes (20 V, 800 min). Membranes were then blocked for 1 h with 5% bovine serum albumin (BSA) in 1× Tris-buffered saline (TBS) with 1% Tween (TBST), followed by incubation with primary antibody (anti-mCherry antibody [Cell Signaling] 1:1000 dilution in 5% BSA-TBST; anti-GAPDH antibody [Invitrogen] 1:1000 dilution in 5% BSA-TBST) overnight at 4 C on a platform rocker. The membranes were then washed 3× for 5 min each with TBST and incubated with the appropriate secondary antibody in 5% BSA-TBST (1 h; room temperature). After washing 3× for 5 min with TBST, the membranes were exposed to a chemiluminescent substrate for 5 min and imaged with an Azure cSeries imaging station.  Light-activated membrane remodeling

Data availability
Experimental data are available upon request. All CRY-BAR plasmids will be available through Addgene, a nonprofit DNA repository.
Supporting information-This article contains supporting information.
Acknowledgments-We thank Mr Collin T. O'Bryant for assistance with confocal microscopy.