Regulation of endogenous transmembrane receptors through optogenetic Cry2 clustering

Transmembrane receptors are the predominant conduit through which cells sense and transduce extracellular information into intracellular biochemical signals. Current methods to control and study receptor function, however, suffer from poor resolution in space and time and often employ receptor overexpression, which can introduce experimental artifacts. We report a genetically-encoded approach, termed Clustering Indirectly using Cryptochrome 2 (CLICR), for spatiotemporal control over endogenous transmembrane receptor activation, enabled through the optical regulation of target receptor clustering and downstream signaling using non-covalent interactions with engineered Arabidopsis Cryptochrome 2 (Cry2). CLICR offers a modular platform to enable photocontrol of the clustering of diverse transmembrane receptors including FGFR, PDGFR, and integrins in multiple cell types including neural stem cells. Furthermore, light-inducible manipulation of endogenous receptor tyrosine kinase (RTK) activity can modulate cell polarity and establish phototaxis in fibroblasts. The resulting spatiotemporal control over cellular signaling represents a powerful new optogenetic framework for investigating and controlling cell function and fate.


Introduction
Cells receive dynamic information via transmembrane receptor signals and integrate these cues to regulate diverse biological processes including cell proliferation, motility, differentiation, and death. Currently utilized approaches to study transmembrane receptor signaling 1,2 often employ overexpression of receptors and receptor fusions, which fundamentally alter how cells respond to extracellular signals and frequently induce hypersensitive or constitutive signal activation [3][4][5] . Indeed, receptor gene amplification and overexpression is a common driving factor of cancer cell transformation 6,7 . Moreover, current approaches do not provide a high degree of temporal and spatial control that would be desirable to study receptor signaling dynamics and spatially heterogeneous behavior, respectively.
In recent years, the emerging field of optogenetics has offered several methods to optically control and study numerous signaling phenomena with spatiotemporal precision [8][9][10][11][12][13][14][15][16][17] , though each currently relies on the ectopic expression of the signaling protein of interest. In a recent example, we co-opted the inherent blue-light-dependent oligomerization of Arabidopsis Cry2 to regulate clustering and activation of fusion proteins in the cytosol of mammalian cells 16 . In this report, we demonstrate the optogenetic control of endogenous transmembrane receptor activity. Our approach makes use of receptor clustering, which is a natural mechanism of activation for many important transmembrane receptors, including receptor tyrosine kinases, immune receptors (e.g. TCR, BCR, FcR), Wnt receptors, and integrins [18][19][20][21][22][23][24][25][26] . Specifically, we re-designed and implemented Cry2 clustering to optically target, cluster, and regulate endogenous transmembrane proteins in a modular strategy called Clustering Indirectly using Cryptochrome 2, or CLICR.

CLICR design
In designing CLICR (Fig. 1a), we hypothesized that Cry2 fused to a binding domain (Cry2-BD) possessing limited affinity for a target receptor would remain largely in the cytoplasm in the absence of light. Upon blue light illumination, clusters of Cry2-BD would assemble and thereby effectively create high local concentrations of the BD that may enable simultaneous binding to multiple receptor endodomains, resulting in receptor clustering and signal activation.
As Cry2 optogenetic clustering had been well characterized using cytoplasmic proteins 16 , we first assessed the feasibility of CLICR by investigating whether we could optically cluster a Cry2 binding target in the cytoplasm and whether indirect clustering could induce signal activation (Fig. 1b, Supplementary Fig. 1). Using a heterodimeric pair of leucine zippers 27 , we grafted one of the LZ helices (LZa) onto the Cry2 module and its partner helix (LZb) onto a GFP-LRP6c fusion, whose clustering has been shown to robustly stimulate βcatenin pathway signaling 16,28 . A list of fusion architectures examined can be found in Supplementary Table 1, and a complete list of constructs generated for this work can be found in Supplementary Data 1. Upon co-transfection of the LZ-appended Cry2 and LRP6c fusions into HEK 293T cells, we observed light-induced cluster formation in both the mCherry and GFP channels, and these clusters colocalized, suggesting that Cry2 clusters were successfully clustering GFP-LRP6c fusions via the LZa-LZb interaction (Fig. 1c,  Supplementary Fig. 1b).

Cytoplasmic and membrane-targeted signal activation through CLICR
To determine whether the light-stimulated LRP6c clusters were sufficient to induce βcatenin signaling, we transfected combinations of Cry2-and LRP6c-containing fusion pairs into 293T cells harboring a β-catenin responsive luciferase reporter 29 (Fig. 1c). Significant increases in β-catenin signaling in illuminated vs. unilluminated samples were observed with all Cry2/LRP6c fusion pairs tested, though this signal was markedly diminished compared to that obtained with a single-chain Cry2-mCh-LRP6c fusion ( Supplementary Fig. 1d) 16 , potentially due to LRP6c cluster geometries, expression level differences, or differences in efficiencies of 2-component vs. 1-component clustering. We further demonstrated the feasibility and modularity of the CLICR approach by indirectly clustering the Rac1 GTPase with the same LZ binding adapters ( Supplementary Fig. 1e). Consistent with our previous report of clustering the Rac1 protein 16 , indirect Rac1 clustering induced rapid membrane translocation of the Rac1 (and associated Cry2) fusion ( Supplementary Fig. 1f). Together, the LRP6c and Rac1 results demonstrate that CLICR can serve as a modular strategy for clustering binding partners via non-covalent interactions.
We next investigated whether a CLICR-induced avidity increase was sufficient to elicit translocation of a cytoplasmic Cry2 fusion to the plasma membrane, which would be necessary for CLICR activation of membrane receptors (Fig. 1a). We co-transfected cells with a membrane-localized LZb-GFP-CaaX construct, which is lipid-modified at the Cterminal CaaX box, and Cry2 fused to mCherry with the LZa helix at either the C-or Nterminus (Fig. 1d, Supplementary Fig. 2a). In the dark state, LZb-GFP-CaaX appeared membrane bound, whereas Cry2/LZa fusions were largely cytoplasmic with few regions of visible membrane association, suggesting weak basal affinity of the LZa-LZb binding partners. Upon light illumination, both Cry2 fusions exhibited translocation to the cell periphery and to immobilized puncta within the cell body, suggesting strong light-induced binding to cell membranes (Fig. 1e, Supplementary Fig. 2b). To analyze whether CLICRmediated membrane translocation could induce signaling upon binding and clustering of a transmembrane receptor, we replaced the LZb-GFP-CaaX fusion with a GFP-LZb fused to the C-terminus of the full-length LRP6 receptor ( Supplementary Fig. 3a), whose clustering induces β-catenin signaling 28 . LRP6-GFP-LZb was localized to multiple cellular compartments ( Supplementary Fig. 3b), and illumination of Cry2 with N-or C-terminally fused LZa ( Supplementary Fig. 3c, d), but not with Cry2 lacking the LZa ( Supplementary  Fig. 3e, f), successfully induced Cry2 translocation and colocalization with LRP6-GFP-LZ as well as activation of the β-catenin luciferase reporter, though high background signal was observed in the dark ( Supplementary Fig. 3g) due to receptor overexpression ( Supplementary Fig. 3h). By comparison, a direct LRP6-Cry2 fusion resulted in a constitutive signal that could not be further enhanced by the light, further representative of artifact that can accompany overexpression of transmembrane receptor constructs ( Supplementary Fig. 4). These results both provided support for the CLICR approach and motivated adapting it to activate endogenous receptors and thus overcome overexpression artifact.

Activation of receptor tyrosine kinases
The receptor tyrosine kinase (RTK) class of receptors is broadly important in regulating proliferation, growth, differentiation, survival, and motility, and activation of many RTKs involves lateral interactions including clustering 23 . Furthermore, signaling artifacts are commonly observed upon RTK overexpression within cells 3,5 . To target a cell's native RTKs with CLICR, we used the N-terminal src-homology 2 (SH2) domain from PLC-γ as our binding domain adapter (Fig 2a). Since the SH2 domain typically does not bind unphosphorylated peptide sequences, we hypothesized that SH2 domain clustering may enhance its avidity for unactivated or weakly activated receptors 30 and subsequently cluster and activate the RTKs. To test RTK-targeting ability, we expressed mCherry-Cry2 with an N-terminal SH2 domain (SH2-N) in several cell types, including HEK 293Ts, 3T3 fibroblasts, and adult neural stem cells (NSCs). In 293Ts, illumination of SH2-N yielded membrane-anchored puncta (Fig. 2b, top row; Supplementary Movie 1). Analogously, 3T3 fibroblasts exhibited rapid localization of SH2-N to membrane anchored puncta as well as to focal adhesion structures (Fig. 2c, top row, Supplementary Movie 2), and NSCs exhibited light-induced punctate formation in neurite extensions as well as in the cell body, collectively suggesting that SH2-N may target endogenous membrane proteins. (Fig. 2d, top row; Supplementary Movie 3). A C-terminal fusion of SH2 to Cry2-mCh (SH2-C) was also tested and yielded similar phenotypes, though with slower kinetics, and the SH2-N construct was thus used for all further experiments. Treatment with the RTK inhibitor PD-089828 (RTKi) -which inhibits phosphorylation of the RTKs FGFR1, PDGFRβ, and EGFRmarkedly reduced SH2-N focal enrichment at the membrane in all three cell types (Fig. 2b, c, d bottom rows). These data indicate that strong SH2-N localization is dependent on RTK phosphorylation, and thus that SH2-N clusters are indeed targeting endogenous RTK receptors.

Optogenetic activation of endogenous transmembrane receptors
We next investigated whether SH2-based CLICR targeting of endogenous RTKs is capable of signal activation. Illumination of 3T3 fibroblasts expressing SH2-N yielded a broad increase in tyrosine phosphorylation, compared to unilluminated and RTKi-inhibited controls (Fig. 3a). Furthermore, analysis of canonical RTK signaling pathways showed an increase in levels of phospho-Akt and phospho-Erk1/2 upon 15 minutes of blue light illumination, a difference abrogated in the presence of the RTKi inhibitor. A central benefit of employing CLICR for RTK activation is the lack of receptor overexpression, which may introduce artifacts including hypersensitity to extracellular ligand. Indeed, in contrast to cells overexpressing a PDGFRβ-Cry2 fusion, SH2-N expressing cells retained native sensitivity to extracellular PDGF-BB ( Supplementary Fig. 6).
We further investigated whether CLICR clustering of the PLCγ SH2 domain demonstrated selectivity towards individual receptors. Since activated PDGFRβ has been shown to colocalize with focal adhesions through interaction with focal adhesion kinase (FAK) 33 , we hypothesized that endogenous PDGFRβ activation was largely responsible for the CLICR induced signal activation and focal adhesion translocation of SH2-N in our study (Fig. 2c).
To address this possibility, we examined the light-induced translocation of SH2-N in the presence of pathway specific inhibitors. In 3T3 fibroblasts, SH2-N exhibited a rapid, lightinduced localization to the plasma membrane and focal adhesion structures. Moreover, 10 μM PDGFRi treatment largely inhibited focal adhesion translocation of SH2-N ( Supplementary Fig. 5), whereas 10 μM FGFRi had minimal observed effect on focal SH2-N localization, suggesting that translocation was indeed PDGFRβ dependent and that CLICR clustering was modulating PDGFRβ activity.
In general, activated PDGFR colocalization with FAK within focal adhesions has been shown to modulate FAK activation, focal adhesion turnover, and cell migration 34 . Additionally, PDGFRβ induced membrane recruitment of phosphatidylinositol-3-kinase (PI3K) leads to conversion of phosphatidylinositol-4,5-bisphosphate (PIP2) to phosphatidylinositol-3,4,5-trisphosphate (PIP3) 35 , whose localization at the leading edge of cells is a key factor in cellular polarity establishment and PDGFR-dependent fibroblast migration. We found that in fibroblasts, SH2-N translocation was accompanied by filopodial and lamellipodial extension, often followed by cellular polarization and movement (Fig. 3b,  Supplementary Movie 4). This polarization was dependent on PI3K activity, since addition of the PI3K inhibitor wortmannin led to only transient lamellipodial extensions (Supplementary Fig. 6a, Supplementary Movie 5), which then rapidly stopped protruding or collapsed. PI3K activity was thus required to sustain membrane extensions and establish morphological polarity, consistent with CLICR-mediated activation of PDGFRβ and PI3K.
We next quantified these PDGFR and PI3K dependent effects. Lamellipodial extension induced by whole field illumination was observed in 78% of cells expressing SH2-N, and RTKi inhibition significantly reduced this response to 17% of cells (Fig. 3c, with significance determined through non-overlapping 95% confidence intervals). Likewise, PDGFRi significantly decreased the number of cells extending light-induced lamellipodia from 78% to 37%. In contrast, FGFRi had no effect on cell morphological responses. In cells that did exhibit lamellipodial response, we quantified the time interval from the start of blue light exposure to the first apparent lamellipodial extension. Untreated cells extended initial lamellipodia with an average interval of 123 s, while RTKi or PDGFRi significantly delayed such lamellipodial extension in responding cells to 481 s and 300 s, respectively (Fig. 3d, ANOVA with Tukey post-hoc analysis, p < 0.0001). In contrast, FGFRi treatment had no apparent effect on lamellipodial extension timing. We then examined the fraction of cells that established polarity and initiated migration upon light treatment, as observed by inspecting lamellipodial extension coupled with trailing edge retraction and morphological reorientation. 65% of uninhibited fibroblasts exhibited polarity establishment, and this response was significantly suppressed to 17% and 12% of cells treated with RTKi and PDGFRi, respectively (Fig. 3e, significance determined through non-overlapping 95% confidence intervals). By comparison, FGFRi treatment only moderately reduced lightdependent polarity induction (47% of cells). These results were further corroborated through Western blotting, as light-induced Akt activity was observed in uninhibited and FGFRi inhibited cells, but absent in RTKi and PDGFRi inhibited cells (Fig. 3f). Collectively, though not demonstrating unique specificity of receptor targeting, these data demonstrate the ability of SH2-N to modulate endogenous PDGFRβ activation within 3T3 fibroblasts.
Since PDGFRβ is a key mediator of fibroblast chemotaxis 36 , we next sought to exploit the spatiotemporal precision of CLICR and rewire a cell's chemotactic circuitry to regulate migration in response to a light signal instead of a chemical one. We hypothesized that focal blue light illumination would establish and define a local PIP3 gradient, and that this would allow directed polarization, actin polymerization, and motility of the cell. In fibroblasts coexpressing SH2-N and PH-Venus, which comprises Venus fused to the pleckstrin homology (PH) domain of Akt and is an established biosensor of (PIP3) production 37 , the PH-Venus was initially inhomogeneously distributed, indicating initial polarization of PIP3 within the cell. Illuminating the cell in a region not exhibiting initially high levels of PIP3 induced membrane ruffling and lamellipodial extension toward the illumination region, coupled with a redistribution of PH-Venus (Fig. 3g, Supplementary Movie 6). In particular, the extending lamellipodium exhibited high levels of PH-Venus recruitment, and within minutes the cell began moving towards the light, establishing a PIP3 gradient and a new leading edge as defined by the illumination region (Fig. 3g, Supplementary Movie 6). In the presence of PI3K inhibitor wortmannin, however, focal illumination was unable to induce the sustained extension of the illuminated region and establishment of a PIP3 gradient, confirming that local PIP3 production, polarization, and motility were PI3K dependent. (Supplementary Fig.  6b, Supplementary Movie 7). Additionally, focal blue light illumination of SH2-N cells coexpressing a Lifeact-Venus probe for F-actin stimulated initial lamellipodial actin polymerization, membrane ruffling, and extension within 30s along with a coordinated initiation of trailing edge retraction (Fig. 3h, Supplementary Movie 8). The cell proceeded to phototax in the direction of the initial illumination region until the light stimulus was reoriented, at which point the established lamellipodium collapsed, and the cells repolarized to migrate in the newly defined direction (Fig. 3h, Supplementary Movie 8).

CLICR can target diverse transmembrane receptors
Finally, to demonstrate that CLICR may be a highly modular method for regulating diverse endogenous pathways, we devised a method to target β-integrin receptors, which participate in cellular adhesion complexes in their active state and can diffuse freely in the membrane in an inactive state. The important role of clustering in integrin affinity and activation [38][39][40] renders endogenous integrins attractive CLICR targets. For the binding adapter, we used the phosphotyrosine-binding (PTB)-like F3 domain from talin, which binds β3-integrin cytoplasmic tails 41 and whose expression in a Cry2 fusion yielded no observed impact on basal cell signaling (Supplementary Fig. 8). Co-immunoprecipitation of F3-mCh-Cry2 (F3-N) fusion with a co-expressed integrin-β3-GFP fusion revealed a strong light-dependent association, suggesting that F3 may be used as a binding adapter for targeting integrin-β3 clustering (Figure 4a). Illumination of F3-N in fibroblasts yielded rapid punctate formation and notable motion of the induced puncta along defined intracellular trajectories (Fig. 4b,  Supplementary Movie 9). These phenotypic results were distinct from those described for SH2-N and were absent in cells expressing mCherry-Cry2 without the F3 domain ( Figure  4b). 3T3 cells co-expressing both F3-N and integrin-β3 revealed strong co-clustering of the two constructs under light illumination, with a reversion to diffuse localization in the dark (Figure 4c). This co-clustering was observed in all 116 co-expressing cells in which the F3 construct exhibited cluster formation (Figure 4d). Furthermore, analysis of over 400 clusters within cytoplasmic regions of these cells revealed that every F3-N cluster co-localized with an integrin-β3 cluster (Figure 4e, f), suggesting a strong ability for the F3 domain to enable integrin-β3 clustering via CLICR and overall demonstrating the generality of the CLICR approach for targeting diverse membrane targets 42 .

Discussion
Spatially defined and rapidly time-varying signals are key regulatory factors in diverse biological processes including cell migration 43 , asymmetric cell division 44 , differentiation 5 , and apoptosis 45 . The difficulties of perturbing biological systems with such spatiotemporal precision, however, often preclude our understanding of how dynamic signals regulate cellular function. Optogenetic tools address this need. Although methods for optical protein homodimerization 15,46 , heterodimerization [11][12][13][14] , and homo-oligomerization 16 of ectopically expressed proteins have been developed, ectopic expression of many signaling proteinsparticularly transmembrane receptors -can fundamentally alter how the system of study behaves, potentially confounding conclusions reached with even the most sophisticated perturbative techniques. We devised the CLICR method to address this dilemma.
The CLICR strategy provides a modular, single construct system to optically cluster and activate transmembrane receptors from the cytosol and is a powerful method for reversibly modulating endogenous transmembrane receptor activity. We accomplished this by pairing Cry2 clustering with a binding adapter for our target receptors. Though we envisioned the ideal CLICR adapter to have weak target affinity in the monomeric state and strong avidity upon clustering, it is possible that the CLICR strategy would also work for adapters with strong target affinities. In this case, Cry2-BD would bind the target in the monomeric dark state, and light would induce clustering of these Cry2-BD:target complexes. Indeed, this would be analogous to activation of direct Cry2-receptor fusions 47,48 with the added benefit of targeting endogenous pools of signaling proteins. We note that we do anticipate some level of basal (light-independent) interaction of our LZa-LZb binding adapters in the dark state in Fig. 1 and Supplementary Figs. 1-3, and our ability to enhance these interactions and observe light dependent phenotypes further suggests that CLICR can work for binding adapters with a range of binding affinities.
We present CLICR here as a proof of concept strategy demonstrating the feasibility of targeting and activating endogenous receptor activity with light. Since the SH2-N and F3-N constructs bind endogenous phospho-tyrosine sites and can therefore impact and be subject to the native regulation of those phosphorylation sites, we stress that the orthogonality and appropriateness of these and similar tools must be empirically validated for each system prior to use. Development of alternative, "orthogonal" binding domains, for example an intracellular antibody domain specifically evolved to target a receptor endodomain 49 , may in the future offer additional versatility, specificity, and reduction of basal signal perturbation. Studies should also include controls for Cry2 expression. Though it is unclear to what extent this plant protein domain interacts with mammalian cell machinery, at least one interaction has already been identified 50 .
While our report was under revision, several studies that also further developed optogenetic Cry2 clustering were published. One group applied Cry2 homo-oligomerization for optogenetic activation of overexpressed receptors 47,48 . The demonstrated ability to apply this strategy to multiple RTKs makes this a powerful approach, though we again caution that receptor overexpression often leads to enhanced signaling artifacts (Supplementary Figs. 3h,  3g, 4, and 6), potentially altering the system of study. An additional study engineered Cry2 (Cry2-olig) 51 to form larger clusters than reported with wild-type Cry2 16 and used this mutant to sequester and interfere with the function of endogenous proteins. This expansion of Cry2 clustering variants presents a choice for researchers based on their optogenetic needs. For studies seeking to sequester and inhibit protein function, large clusters with slow OFF-kinetics, such as those provided by Cry2-olig, may be ideal. For studies requiring temporally precise activation of endogenous proteins with low background activation, the weaker-associating wild-type Cry2 with faster OFF faster kinetics may be more appropriate.
In sum, receptor overexpression can often elevate pathway signaling above the physiologically basal state and/or alter signal response dynamics. As an alternative, our demonstrated activity, modularity, and versatility of CLICR across numerous cell types suggests its utility as an optogenetic tool to manipulate endogenous protein activity across biological systems.

DNA Vector Assembly
All vectors were constructed using the CPEC method 52 and through standard restriction enzyme cloning 16 . Full sequences and descriptions of all constructs used are provided in Supplementary Data 1. The LZa and LZb heterodimeric leucine zipper pair corresponds to leucine zipper pair (vii), as described 27 , and was synthesized through primer overlap extension. We received the full length LRP6 construct from Dr. Xi He, the Rac1 construct from Dr. G. Steven Martin, the Akt construct from Dr. S. Ferguson, the PDGFRβ, and 7x TCF constructs from Addgene (#23893 and #24308, respectively), and the PLCγ-1 construct from the DNASU plasmid repository (clone IDs HsCD00321979 and HsCD00435250, respectively). The FGFR1 construct was cloned from a rat NSC cDNA library by Dr. M. Varedi.

Luciferase Assay
HEK 293T cells stably transduced with a 7x Tcf luciferase reporter were seeded in blackwalled 96 well plates. When ~50% confluent, the cells were transfected with 5 ng of each expression vector diluted in 150 ng pBluescript. Cells were given 24 hrs to express protein and were then illuminated overnight for 16 hours with one 500 ms pulse of LED light every 10 s as administered by a custom LED illumination device 16 . After illumination, cells were lysed, and luciferase was measured with the Luc-screen Firefly Luciferase Gene Reporter System (Life Technologies).

Confocal Microscopy
Time-lapse microscopy of activated Cry2 fusions was performed on a Zeiss LSM 710 AxioObserver confocal microscope with full incubation chamber in conjunction with the Zeiss ZEN software. Cry2 translocation experiments were carried out at 25°C, while experiments examining cell motility and polarity were carried out at 37°C and in 5% CO 2 . EGFP, Venus, and mCherry were visualized with 488, 514, and 561 nm laser excitation, respectively, through either a 40x or 63x oil immersion objective. Whole field Cry2 activation was achieved using 450 or 488 nm illumination. Spatially defined activation was carried out under whole cell imaging of Venus with a 514 nm laser in conjuction with focal illumination using the FRAP application with 458 or 488 nm laser light at 1-5% power and 10 μs dwell time per pixel in a 10 μm diameter region. For whole field observation of lamellipodial induction and polarity establishment, cells were excited every 13 seconds with 488 nm light and mCherry was imaged simultaneously.

Image Analysis
For whole field observation of lamellipodial induction and polarity establishment, cells were simultaneously excited with 488 nm light and imaged for mCherry and Lifeact-Venus localization. Time-lapse image stacks were assembled in ImageJ, and each cell was examined for the appearance of new lamellipodia and for the time elapsed from the start of blue light excitation to initial lamellipodial appearance. To assess polarity establishment, cells were scored for the presence of lamellipodial formation coupled with visible trailing edge retraction and morphological reorientation.
Immunoprecipitation 293Ts stably expressing full-length PDFGRβ-GFP or FGFR1-GFP were seeded in 10 cm plates. When ~70% confluent, each plate was transfected with 1 μg 6xHis-SH2-mCh-Cry plasmid DNA diluted in 9 μg pBluescript by the calcium phosphate method. 24 h after transfection, cells were serum starved overnight. Light-treated plates were illuminated with a 19-LED array for 10 minutes at 25°C, after which cells were lysed in 1 mL RIPA buffer. After incubation on ice for 30 minutes, the cell lysate was spun at 4°C for 10 minutes, and the supernatant from each sample was incubated with 5 μg of anti-GFP antibody (Life Technologies, A-11122). Antibody-lysate solutions were incubated overnight at 4°C with gentle shaking and were then incubated in 12.5 μL immobilized protein A/G agarose beads (Thermo Scientific #20421) for 1 hour as per manufacturer instructions. Beads were spun down and washed with lysis buffer three times, followed by one final wash with water. Beads were then incubated in 2X SDS sample buffer, boiled for 10 minutes, spun down one final time, separated via SDS-PAGE, and blotted onto a nitrocellulose membrane. Blots were then probed for GFP or mCherry (Clontech # 632543, 1:1000 dilution).

Statistical methodology
Statistical power in luciferase assays was determined through the non-parametric Mann-Whitney Wilcoxon test. For comparing lamellipodial extension delays, varying dosages for RTKi and PDGFRi treatment were compared using one-way ANOVA followed by Tukey post-hoc pairwise analysis, with a significance threshold adjusted using the Bonferroni correction. FGFRi inhibition was compared to untreated cells using a Student's t-test. All statistical analysis was performed using the JMP statistical software suite.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. The CLICR strategy enables Cry2 activation of transmembrane receptors. (a) CLICR allows modular Cry2 clustering and activation of membrane receptors via non-covalent interactions, avoiding complications associated with overexpression of receptor fusions. With CLICR, Cry2 fused to a receptor-targeting binding domain (BD) is expressed in the cytoplasm. In the dark, unclustered state, BD affinity for the receptor is weak and imparts no or little membrane localization. Upon light induced clustering, BD-Cry2 oligomers increase local BD concentration, enabling membrane translocation, binding, and nucleation of a receptor cluster. (b) CLICR clustering allows activation of cytoplasmic protein targets. Upon light activation, the Cry2-mCh-LZa (red) and LZb-LRP6c-GFP (green) components co-cluster (c), suggesting successful clustering of non-covalently bound proteins via homooligomerization of Cry2. (d,e) CLICR clustering also enables membrane translocation of the initially cytoplasmic Cry2-mCh-LZa. Scale bars = 20 μm. CLICR can target and cluster endogenous transmembrane receptors. (a) To target endogenous receptors with CLICR, an adapter domain with affinity for native receptors was used in place of the engineered LZa-LZb interaction used previously. The SH2 domain from PLCγ was used as a binding domain fused to the N-terminus of mCh-Cry2 (SH2-N) to target endogenous receptor tyrosine kinases. Illuminated SH2-N expressed in 293Ts (b, top panels) localized to membrane-anchored foci. In 3T3 fibroblasts (c, top panels), illumination of SH2-N induced translocation to the membrane periphery and to structures resembling focal adhesions. In neural stem cells (NSC) (d, top panels), illuminated SH2-N localized to immobilized membrane clusters both in the cell body and in neurites. In all 3 cell types, inhibition with an FGFR1/PDGFRβ/EGFR-specific inhibitor (RTKi) largely abrogated visible SH2-N translocation. (b, c, d, lower panels). In 293Ts, upon blue light illumination SH2-N colocalized (arrows) to exogenously expressed, full-length FGFR1 (e) and PDGFRβ (g) fused to GFP, and this colocalization was abrogated in the presence of inhibitors specific to these receptors (f, h). Scale bars = 20 μm. Light-stimulated association of SH2-N with either FGFR1-GFP or PDGFRβ-GFP was confirmed through co-immunoprecipitation of SH2-N with the receptors (i). Bugaj   CLICR clustering allows photoactivation of endogenous RTKs in fibroblasts and enables PDGFRβ-dependent phototaxis. Western blot analysis (a) of serum-starved fibroblasts treated for 10 minutes with blue light shows broad upregulation of tyrosine phosphorylation in illuminated vs unilluminated samples, and this difference is abrogated upon 10 μm RTK inhibition. Samples grown in 2% serum were used as a positive control. In particular, lighttreated samples demonstrated an RTK-dependent increase in Akt and Erk1/2 phosphorylation. (b) Time lapse imaging of SH2-N expressing fibroblasts shows lamellipodial extension and cell repolarization upon whole field blue light exposure. (c) The proportion of cells extending lamellipodia in response to whole field illumination was reduced by broad RTK or PDGFR inhibition, but not FGFR inhibition. Error bars represent 95% confidence intervals, n = 24-83 cells per condition. (d) Among cells extending lightinduced lamellipodia under broad RTK inhibition or PDGFR inhibition, a longer delay before initial lamellipodial extension was evident, while FGFR inhibition did not induce a delay in extension. Graph shows means ± 1 s.d, n = 4-58 cells per condition. ***p < 0.001 by one way ANOVA with Tukey pairwise analysis. (e) The proportion of cells exhibiting light-induced polarization was reduced by RTKi or PDGFRi treatment, and slightly reduced by FGFRi treatment. Error bars represent 95% confidence intervals, n = 24-83 cells per condition. (f) Focal illumination enabled the spatial definition of the lamellipodial extension region coupled with re-establishment of a PIP3 gradient at the site of illumination (arrows), as observed through a PH(Akt)-Venus biosensor. (g) CLICR-enabled focal activation of lamellipodial extension and polarity establishment allows the effective rewiring of PDGFRdependent chemotaxis to respond to light. Fibroblasts co-expressing SH2-N and Lifeact-Venus were illuminated focally, and actin polymerization, lamellipodial extension, and cell motility were induced in the direction of the illuminated region. Illuminating the trailing edge induced re-polarization and a reversal of migration direction. Scale bars = 20 μm.