OPTOGENETIC TOOL FOR RAPID AND REVERSIBLE CLUSTERING OF PROTEINS

Abstract

A protein construct including a gene encoding a light-sensitive protein fused to at least one of either a low complexity sequence, an intrinsically disordered protein region (IDR), or a repeating sequence of a linker and another gene encoding a light-sensitive protein. Among the many different possibilities contemplated, the protein construct may also advantageously include cleavage tags. This protein construct may be utilized for a variety of functions, including a method for protein purification, which requires introducing the protein construct into a living cell, and inducing the formation of clusters by irradiating the construct with light. The method may also advantageously include cleaving a target protein from an IDR, and separating the clusters via centrifuge. A kit for practicing in vivo aggregation or liquid-liquid phase separation is also included, the kit including the protein construct and a light source capable of producing a wavelength that the light-sensitive protein will respond to.

Description

BACKGROUND OF THE INVENTION

Cellular function relies on coordinating the thousands of reactions that simultaneously take place within the cell. Cells accomplish this task in large part by spatio-temporally controlling these reactions using diverse intracellular organelles. In addition to classic membrane-bound organelles such as secretory vesicles, mitochondria and the endoplasmic reticulum, cells harbor a variety of membrane-less organelles. Most of these are abundant in both RNA and protein, and are referred to as ribonucleoprotein (RNP) bodies. Among dozens of examples include nuclear bodies such as nucleoli, Cajal bodies, and PML bodies, and cytoplasmic germ granules, stress granules and processing bodies ((Mao et al., 2011), (Anderson and Kedersha, 2009), (Buchan and Parker, 2009), (Handwerger and Gall, 2006)). By impacting a number of RNA processing reactions within cells, these structures appear to play a central role in controlling the overall flow of genetic information, and are also increasingly implicated as crucibles for protein aggregation pathologies ((Li et al., 2013), (Ramaswami et al., 2013)).

From a biophysical standpoint, these structures are remarkable in that they have no enclosing membrane and yet their overall size and shape may be stable over long periods (hours or longer), even while their constituent molecules exhibit dynamic exchange over timescales of tens of seconds (Phair and Misteli, 2000). Moreover, many of these structures have recently been shown to exhibit additional behaviors typical of condensed liquid phases. For example, P granules, nucleoli, and a number of other membrane-less bodies will fuse into a single larger sphere when brought into contact with one another ((Brangwynne et al., 2009), (Brangwynne et al., 2011), (Feric and Brangwynne, 2013)), in addition to wetting surfaces and dripping in response to shear stresses. These observations have led to the hypothesis that membrane-less organelles represent condensed liquid states of RNA and protein that assemble through intracellular phase separation, analogous to the phase transitions of purified proteins long observed in vitro by structural biologists ((Ishimoto and Tanaka, 1977), (Vekilov, 2010)). Consistent with this view, RNP bodies and other membrane-less organelles appear to form in a concentration-dependent manner, as expected for liquid-liquid phase separation ((Brangwynne et al., 2009), (Weber and Brangwynne, 2015), (Nott et al., 2015), (Wippich et al., 2013), (Molliex et al., 2015)). These studies suggest that cells can regulate membrane-less organelle formation by altering proximity to a phase boundary. Movement through such an intracellular phase diagram could be accomplished by tuning concentration or intermolecular affinity, using mechanisms such as posttranslational modification (PTM) and nucleocytoplasmic shuttling.

Recent work has begun to elucidate the molecular driving forces and biophysical nature of intracellular phases. Weak multivalent interactions between molecules containing tandem repeat protein domains appear to play a central role ((Li et al., 2012), (Banjade and Rosen, 2014)). A related driving force is the promiscuous interactions (e.g. electrostatic, dipole-dipole) between segments of conformationally heterogeneous proteins, known as intrinsically disordered protein or intrinsically disordered regions (IDP/IDR, which are typically low complexity sequences, LCS). Hereinafter, the terms intrinsically disordered protein, intrinsically disordered region, and intrinsically disordered protein region are used interchangeably. RNA binding proteins often contain IDRs with the sequence composition biased toward amino acids including R, G, S, and Y, which comprise sequences that have been shown to be necessary and sufficient for driving condensation into liquid-like protein droplets ((Elbaum-Garfinkle et al., 2015), (Nott et al., 2015), (Lin et al., 2015)). The properties of such in vitro droplets have recently been found to be malleable and time-dependent ((Elbaum-Garfinkle et al., 2015), (Zhang et al., 2015), (Weber and Brangwynne, 2012), (Molliex et al., 2015), (Lin et al., 2015), (Xiang et al., 2015), (Patel et al., 2015)), underscoring the role of IDR/LCSs in both liquid-like physiological assemblies and pathological protein aggregates.

Despite these advances, almost all recent studies rely primarily on in vitro reconstitution, due to a lack of tools for probing protein phase behavior within the living cellular context. However, a growing suite of optogenetic tools have been developed to control protein interactions in living cells. The field has primarily focused on precise control over homo- or hetero-climerization ((Toettcher et al., 2011), (Kennedy et al., 2010), (Levskaya et al., 2009)). But recent work suggests the potential of optogenetics for studying intracellular phases, demonstrating that light-induced protein clustering can be used to activate cell surface receptors (Bugaj et al., 2013), as well as to trap proteins into inactive complexes ((Lee et al., 2014), (Taslimi et al., 2014)).

To date, no platform has been provided which can be used to dynamically modulate intracellular protein interactions, enabling the spatiotemporal control of phase transitions within living cells. Such a platform would be highly desirable.

BRIEF SUMMARY OF THE INVENTION

The present invention is directed to protein constructs with the ability to induce and control reversible liquid-liquid phase separation in living cells, both globally and at specific subcellular locations. This system reveals that the location within the phase diagram can be used to dictate the material state of phase-separated IDR clusters, ranging from dynamic liquid droplets to arrested but reversible gels, which can over time mature into irreversible aggregates. In the present invention, systems and methods are provided that utilize protein constructs containing genes encoding light sensitive proteins fused to a low complexity sequence (LCS), intrinsically disordered protein region (IDR), or one or more repeatable sequences.

Among the many different possibilities contemplated, a protein construct may advantageously contain a desired protein to purify, a fluorophore, an IDR that is a portion of FUS, Ddx4, or hnRNPA1, contain a light sensitive protein that is sensitive to visible light, or could contain a light sensitive protein that includes Cry2, Cry2olig, PhyB, PIF, light-oxygen-voltage sensing (LOV) domains, or Dronpa. It is contemplated that these protein constructs will be configured such that after being introduced into a living cell, exposing the living cell with the protein construct to certain wavelengths of light will induce molecules within the living cell to cluster. It is further contemplated that if these protein constructs contain cleavage tags, such as self-cleaving tags, Human Rhinovirus 3C Protease (3C/PreScission), Enterokinase (EKT), Factor Xa (FXa), Tobacco Etch Virus Protease (TEV), or Thrombin (Thr), then after a first induction, it may be advantageous to cleave and induce clustering again.

BRIEF DESCRIPTION OF DRAWINGS

FIGS. 1A and 1B illustrate two embodiments of protein constructs.

FIG. 2A is a graph of assembly rates for various protein constructs.

FIG. 2B is a graph of disassembly rates for various protein constructs.

FIGS. 3A and 3B are graphs of the temporal evolution of background concentrations outside clusters, Cbg, for certain cells, where the activation interval is T=30s (FIG. 3A) or T=180s (FIG. 3B). Dots denote time points when cell images were taken and dashed lines represent time points for blue light activation.

FIG. 3C is a graph indicating the steady-state background fluorescence intensities of individual cells (open circles) under the given activation interval increase linearly (solid lines) with total concentration of molecules.

FIG. 3D is a graph indicating the impact of activation interval or weaker light intensity on the steady-state fractions of inactivated molecules; blue light powers (488 nm, in μW) used in the measurements are specified, and solid lines denote a global fit to data using the kinetic model.

FIG. 3E is a depiction of the calculated steady-state concentrations of activated molecules, showing a clear concentration threshold for light-mediated clustering, and a dotted horizontal line indicating the saturation concentration.

FIG. 4 is a schematic diagram illustrating the broadening effects of slower inactivation kinetics and lower Csat on the localized phase transition.

FIGS. 5A and 5B are images of the distinct morphology of phase separated optoFUS clusters for shallow (5A) and deep (5B) supersaturation.

FIG. 6 is a graph of FRAP recovery curves of optoFUS clusters formed with varying supersaturation depths.

FIG. 7 illustrates an embodiment of protein construct using repeatable units.

FIG. 8A is an illustration of a two-component system leading to light-activatable repeats.

FIG. 8B illustrates the self-assembled construct shown in FIG. 8A in the absence of activating light.

FIG. 8C is an illustration showing the resulting clusters upon light activation.

FIG. 8D and 8E are micrographs illustrating clustering upon light activation.

FIG. 9A is an illustration of an example 3D phase diagram as a function of the concentration of a particular repeat number for illustration purposes showing PhyB_N+PIF_M, and apparent binding affinity, controlled by the 650/750 nm light ratio.

FIG. 9B is another illustration of a phase diagram.

FIG. 10 is a graph of a normalized number of clusters during disassembly cycles for four constructs.

FIG. 11 is a flowchart depicting a method for protein purification.

DETAILED DESCRIPTION OF THE INVENTION

Before the present invention is described in further detail, it is to be understood that the invention is not limited to the particular embodiments described, as such may, of course, vary. It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only, and is not intended to be limiting, since the scope of the present invention will be limited only by the appended claims.

Where a range of values is provided, it is understood that each intervening value, to the tenth of the unit of the lower limit unless the context clearly dictates otherwise, between the upper and lower limit of that range and any other stated or intervening value in that stated range is encompassed within the invention. The upper and lower limits of these smaller ranges may independently be included in the smaller ranges is also encompassed within the invention, subject to any specifically excluded limit in the stated range. Where the stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included in the invention.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although any methods and materials similar or equivalent to those described herein can also be used in the practice or testing of the present invention, a limited number of the exemplary methods and materials are described herein.

It must be noted that as used herein and in the appended claims, the singular forms “a”, “an”, and “the” include plural referents unless the context clearly dictates otherwise.

FIG. 1A depicts a generalized embodiment of the present invention. A protein construct (10) will typically comprise a light sensitive region (20) and a functional region (30). The construct may also optionally comprise a first cleavage tag (40), a target protein (50), or a second cleavage tag (60). Typically, the second cleavage tag (60) will not be needed unless the construct is followed by another domain. The specific arrangement shown in FIG. 1A just serves as an example and other permutated arrangements can be employed.

The light sensitive region (20) typically includes a first segment comprising at least one gene encoding at least one protein sensitive to at least one wavelength of light. In preferred embodiments, this segment includes at least one gene encoding for Cry2, Cry2olig, PhyB, PIF, light-oxygen-voltage sensing (LOV) domains, or Dronpa. In other embodiments, the segment includes at least one gene encoding for a protein sensitive to a visible wavelength of light, typically including wavelengths from about 400 nm to about 800 nm.

The functional region (30) may include a second segment fused to the first segment, the second segment comprising one or more low complexity sequences, one or more intrinsically disordered protein regions (IDRs), one or more synthetic or natural nucleic acid binding domains, or at least one repeatable sequence, the repeatable sequence comprising a linker fused to at least one additional gene encoding at least one protein sensitive to at least one wavelength of light. In preferred embodiments, the protein construct comprises an IDR, where the IDR is a portion of a first protein selected from the group consisting of FUS, Ddx4, and hnRNPA1. In some embodiments, the IDR comprises amino acids 1-214 of FUS, 1-236 of DDX4, or 186-320 of HNRNPA1.

FIG. 1B shows an alternate arrangement of the protein construct, comprising an optional fluorophore (70). In FIG. 1B, the fluorophore (70) is fused between the light sensitive region (20) and the functional region (30), although other locations may be advantageous. In preferred embodiments, the fluorophore is mCherry, although the use of other fluorescent proteins is also envisioned, including but not limited to GFP variants.

An example of the protein construct was produced by fusing the “sticky” IDR from various proteins to the photolyase homology region (PHR) of Arabidopsis thaliana Cry2, a light-sensitive protein which is known to self-associate upon blue light exposure. This IDR-Cry2 fusion protein would recapitulate the modular domain architecture of many phase separating proteins, but confer tunable light-dependence to its multivalent interactions.

The approach of replacing the multi-valent interaction domains of FUS and other IDR-containing RNA binding domains with a light-activatable Cry2(WT) motif is inspired by native mechanisms utilized by cells to control intracellular phase transitions. Phase separation in cells appears to be regulated in two distinct but complementary ways: 1) changing the concentration of molecules, for example by protein translation, altered degradation, or nucleocytoplasmic shuttling, and 2) changing their intermolecular interaction strengths, for example through PTMs, particularly phosphorylation which deposits a negative charge on S, T, or Y residues, which are commonly found in IDRs driving phase separation. Indeed, FUS is found in stress granules, one type of membrane-less body whose assembly depends on PTMs and protein concentration, and which has been suggested to assemble by regulated intracellular phase separation.

DNA fragments encoding IDRs of human FUS (residues 1-214) and human hnRNPA1 (residues 186-320) were amplified by PCR using FUS cDNA (GeneCopoeia, GC-F0952) and pET9d-hnRNP-A1 (Addgene plasmid #23026), respectively. A gene for the IDR of human DDX4 (residues 1-236) was synthesized (Integrated DNA Technologies). Sequences for mCherry and Cry2olig (Addgene plasmid #60032) were cloned into the pHR lentiviral backbone to generate the pHR-mCh-Cry2olig plasmid. A site-directed mutagenesis was then performed to produce the Cry2WT version. For IDR-fusion Cry2 plasmids, DNA fragments encoding the IDRs were inserted into the linearized pHR-mCh-Cry2WT (or Cry2olig) backbone using In-Fusion Cloning Kit (Clontech). The resulting constructs were fully sequenced to confirm the absence of unwanted substitutions.

These constructs were introduced into living cells. mCherry-labeled Cry2 PHR (hereafter: Cry2WT) was first expressed in NIH 3T3 cells along with a few other variants. NIH 3T3 cells were cultured in 10% FBS (Atlanta Biological) in DMEM (Gibco) supplemented with penicillin, streptomycin and GlutaMAX (Gibco) at 37° C. with 5% CO₂ in a humidified incubator. To produce stable cell lines expressing cry2 fusion constructs, lentiviral constructs were transfected with FuGENE (Promega), following the manufacturer's recommended protocol, into 293T cells that had been plated in the 6-well dishes 1 day prior to the transfection. Viral supernatants were collected 2 d after transfection and passed through a 0.45-μm filter to remove cell debris. NIH 3T3 cells plated at ˜70% confluency in the 6-well dishes were infected by adding 0.4-1 ml of filtered viral supernatant directly to the cell medium. Viral medium was replaced with normal growth medium 24 h after infection.

The cells were then induced to cluster with blue light. Consistent with previous reports, Cry2WT alone showed little clustering upon blue light activation. Strikingly, fusing the N-terminal IDR of FUS (FUS_N) to Cry2WT (hereafter optoFUS) leads to rapid blue-light dependent cluster assembly in most cells. Similar results were seen upon fusing the C-terminal IDR of the ALS-related RNA binding protein hnNRNPA1 (optoHNRNPA1), or the N-terminal IDR of Ddx4 (optoDDX4), both of which have been reported to drive liquid-liquid phase separation. 100421 As shown in FIG. 2A, the optoFUS construct (103) exhibits ˜9-fold faster assembly than Cry2olig (102) assemblies, under similar expression level and activation conditions. Referring to FIG. 2B, it is also noted that the optoFUS (114) assemblies disassemble ˜3-fold faster than Cry2olig assemblies (112). Interestingly, the assembly of Cry2olig is also dramatically enhanced when it is fused to FUS_N (FIG. 2A, element 104). However, the disassembly of these FUS_N-Cry2olig clusters (113) is now much slower than for Cry2olig alone (112). Moreover, these assemblies (113) do not completely dissolve, even after >1 hour without light activation.

Liquid droplets tend to adopt round shapes due to surface tension. Consistent with this feature of liquids, optoFUS clusters have round morphologies. A second feature common to liquid phase droplets is that the molecules within undergo dynamic exchange with the surrounding solution. Fluorescence recovery after photobleaching (FRAP) experiments, involving bleaching the mCherry signal, shows a nearly complete recovery of the fluorescence signal, with a recovery time scale of 140±10 s. Finally, in non-biological systems, small phase separated droplets can dissolve at the expense of larger droplets, an effect known as Ostwald ripening. Ostwald ripening is frequently observed in the optoFUS clusters, particularly within those that assemble within the cell nucleus. Together, these data strongly suggest that optoFUS clusters formed upon blue light exposure are liquid phase droplets.

These liquid-like behaviors suggest that optodroplet assembly may represent light-inducible liquid-liquid phase separation within the cell. A light-induced increase in Cry2 self-association affinity could represent a controllable change to the effective valency of the constructs. In the presence of light, each FUS-Cry2 fusion protein can associate with other monomers through Cry2-Cry2 or FUS-FUS interactions, whereas only FUS-FUS interactions would be possible in the dark. In this physical picture, the light-increased avidity would result in the crossing of a phase boundary and consequent initiation of liquid-liquid phase separation.

The concentration of light-activated optoFUS can be changed using at least two independent methods: 1) by changing the total concentration of optoFUS molecules within the cell, and 2) by changing blue light intensity. If the assembly mechanism is liquid-liquid phase separation, then droplet formation should depend on both optoFUS concentration and light activation level. Consistent with this, droplet formation shows a strong dependence on blue light activation intensity. For an activation protocol which begins at a very weak power, initially no cells exhibit droplets, even after continuous weak blue light activation for 16 min. However, when the blue light power was tripled, those cells which express high levels of the optoFUS construct now assemble droplets. Distinct and spatially separated droplets slowly nucleate and then grow in size; qualitatively, this behavior is very similar to the well-known nucleation and growth regime observed in shallow-quench phase transitions—i.e., in systems which are only moderately supersaturated.

The dependence of droplet assembly kinetics on the total concentration of optoFUS was also tested. Blue light power was fixed and assembly was examined in cells with different expression levels; intracellular optoFUS concentrations ranging from about 0.2 to about 13 μM were used, comparable to the estimated intracellular concentration of endogenous FUS: ˜1-10 μM. Consistent with the results obtained for varying blue-light activation, the lowest expressing cells do not form droplets at all. Interestingly, for cells that do form droplets, the higher the expression level, the faster the assembly kinetics. Thus, both the total optoFUS concentration and blue light intensity collectively affect light-induced droplet formation.

Finally, it was determined if the opposing effects from these two parameters can compensate each other to give rise to similar assembly dynamics. Indeed, higher expressing cells exposed to weaker blue light show similar clustering kinetics as lower expressing cells exposed to stronger light. Taken together, these data suggest that the concentration of light activated optoFUS can be used for controlling droplet formation.

A simple kinetic framework for measuring the concentration of activated molecules, and its relationship to the onset of droplet condensation, was developed to quantitatively test whether this system reflects light-controllable phase separation. We assume that the inactivated state undergoes a first-order reaction to the activated state, with a reaction rate proportional to light intensity, according to k₁=k_act*[blue], where k_act is an activation rate constant and [blue] is the intensity of activating blue light. The activated molecules can also convert back to the inactivated state, at a rate given by k₂. In this model, blue light exposure increases the concentration of activated, self-associating molecules, which drives global phase separation upon exceeding the saturation concentration, i.e., when C_act>C_sat.

A series of activation protocols were employed with different activation intervals and blue light power. Cycles of light followed by dark were utilized, since this provides the ability to probe both the activation rate constant, k_act, and the inactivation rate constant, k₂. When optoFUS cells are exposed to a pulse train of activating light stimuli with high enough intensity, cells form droplets typically after a short lag phase. As molecules are recruited into droplets, the background concentration (fluorescence intensity) outside of droplets decreases. As shown in FIGS. 3A and 3B, when the interval between consecutive pulses is relatively long (e.g., T=180s, shown in FIG. 3B) cycles of partial assembly and disassembly are observed, but for intervals shorter than 1 min (e.g., T=30s, shown in FIG. 3A) the background intensity exhibits a monotonic decay to a steady state, Cbg,st.

In the simplest phase transition model, the steady-state background concentration is equal to a sum of the concentration of inactivated molecules, C_inact,st, and the activated molecules outside clusters, C_sat. Expressing C_inact,st as a fraction F_inact,st of the total concentration: C_inact,st=F_inact,stC_tot, the steady-state background concentration is thus C_bg,st=F_inact,stC_tot+C_sat. Consistent with this model prediction, the steady-state background concentration of optoFUS cells increases linearly with total concentration. Moreover, as shown in FIG. 3C, varying activation intervals yield different slopes (F_inact,st), but converge to a similar y-intercept (C_sat, corresponding to ˜1.4 μM), consistent with the saturation concentration representing an intrinsic property of the optoFUS construct. Indeed, the identical activation protocol, when applied to optoDDX4 cells, yields 2-fold lower C_sat, implying stronger intermolecular interaction between DDX4 IDRs.

This kinetic framework can be utilized to quantify the rate constants for activation. We first computed the steady-state fraction of inactivated molecules for each cell using the relationship, F_inact,st=(C_bg,st−C_sat)/C_tot, and the measured saturation concentration (FIG. 3D). In agreement with the model predictions, this illustrates that the fraction of inactivated molecules increases with either longer activation intervals or weaker blue light intensity. This can be well-fit to the functional dependence predicted by the model, yielding values for the rate constants, k_act=7.4±4.7 μW⁻¹s⁻¹ and k₂=0.011±0.005 s⁻¹. Moreover, this also agree with the model prediction that at high enough power, the inactivated fraction becomes independent of blue light intensity, since all molecules already populate the activated state. Finally, phase separation should only occur if the total concentration of activated molecules exceeds the saturation concentration, Csat. This prediction is in good agreement with data for this embodiment, which show a sharp concentration threshold for the activated molecules, below which no cytoplasmic clusters were observed (See FIG. 3E).

The preceding experiments and theoretical analysis show that fusing, for example, self-associating IDRs to the light activation domain of Cry2WT enables light-activated phase separation. However, it is also possible to modulate the assembly dynamics by changing the light activation domain. Previously, a point mutant version of Cry2 (E490G), known as Cry2olig, was shown to exhibit significant clustering. The assembly of Cry2olig is also dramatically enhanced when it is fused to FUS_N, exhibiting ˜9-fold faster assembly under similar expression level and activation conditions, comparable to the rapid assembly of the optoFUS construct (i.e. FUS_N-Cry2WT).

Applying the same method of cycled light activation described above, it was found that there is also a saturation concentration of FUS_N-Cry2olig. However, the saturation concentration of FUS_N-Cry2olig is 5-fold lower than optoFUS, consistent with the point mutation (E490G) in Cry2olig increasing homo-interaction strength. Moreover, the inactivation rate of FUS_N-Cry2olig is 5-fold slower than optoFUS, consistent with the previous findings. Thus, utilizing IDR fusions with various other self-associating optogenetic proteins can be used to tune the dynamics of light-induced intracellular phase separation. 100541 While the above example utilizes IDRs, the functional region may also utilize other proteins, such as synthetic or natural nucleic acid binding domains. Many RNA binding proteins contain self-associating IDRs or LCSs that can drive phase separation. However, additional RNA binding domains can enhance phase separation via multivalent interactions with RNA. For example, FUS is an ALS-related RNA binding protein involved in diverse nucleic acid processing including DNA repair, transcription and pre-mRNA splicing. While the self-associating N-terminal IDR of FUS has been shown to be necessary and sufficient for liquid-liquid phase separation, C-terminal RNA binding domains appear to further promote phase separation. In preferred embodiments, the synthetic or natural nucleic acid binding domains utilizes RNA recognition motifs (RRM), double-stranded RNA binding domains (dsRBD), S1, zinc finger binding domains, YT521-B homologies (YTH), DNA and RNA helicase domains, Pumilio, or S-adenosylmethionine (SAM) structures.

Rapid growth and fast inactivation lead to localized phase separation. Local changes in molecular interaction strength can induce intracellular phase separation at specific subcellular locations, as in the case of P granule condensation during C.elegans embryo development. By controlling the spatial distribution of blue light, analogous local phase separation is achievable. When the corners of individual optoFUS cells were locally illuminated, droplets rapidly assembled near the activation zone, with a wave of droplet assembly propagating outward, but only over a short range near the activation zone. This was verified with single line activation, localized in time and space. When a line pulse was applied to optoFUS cells, droplets immediately form along the activation line. The width of cluster distribution was maintained over a narrow band, before all droplets began disassembling within a few minutes.

To quantitatively elucidate the dynamics of phase separation upon localized activation, a simplified coarse-grained model was developed that is consistent with a mesoscale model. This model describes the concentration of activated molecules, c(x), as well as the droplets they nucleate, which are characterized by the single field variable θ_d(x,t) that represents the volume fraction of droplets within a given spatial volume. The model predicts that the steady-state droplet profile width for continuous localized activation is given by: x₀^SS˜√{square root over (D/k)}₂ln[k₁E/(c_sat√{square root over (D(k₁+k₂))})], indicating that the primary factor is the reaction-diffusion length scale, √{square root over (D/k₂)}, where D is the molecular diffusion coefficient in cytoplasm. Thus, diffusion of activated monomers will tend to oppose localized droplet formation, while rapid reversion to the dark state would sharpen droplet localization patterns. Numerical simulations of the model support this physical picture by reproducing the evolution time and extent of experimentally-observed droplet profiles, provided heterogeneous (seeded) nucleation kinetics are employed; interestingly, the observed behaviors are not consistent with homogeneous nucleation.

This coarse-grained model predicts that the 5-fold slower inactivation rate (k₂) and 5-fold lower c_sat exhibited by FUS_N-Cry2olig relative to optoFUS would limit the ability to localize droplet assembly (See FIG. 4, compare diameter of localized phase transition 260 vs localized phase transition 262). Consistent with the model prediction, in FUS_N-Cry2olig cells, clusters first rapidly appeared at the localized activation zone, but a wave of cluster formation then propagated slowly across the entire cell; a single line pulse activation also displayed a broader cluster distribution than for optoFUS. Moreover, cells expressing Cry2olig alone exhibited a long lag time, followed by the concomitant appearance of clusters even far away from the activation zone. These data demonstrate that localized phase separation seen in optoFUS depends on the rapid growth conferred by the IDR, combined with the relatively fast inactivation kinetics of Cry2WT.

The location within the phase diagram provides a degree of control over material properties and aging potential of clusters that are induced. In simple non-biological systems, quenching deep into the two phase region, corresponding to a high degree of supersaturation, can lead to condensation of assemblies with arrested dynamics, typically referred to as gels or glasses. Similar arrested dynamics can be observed in living cells, by exposing cells with similar expression levels to varying blue light intensity, thus moving into different depths beyond the phase boundary. For shallow quenching, cells typically showed no clustering during a long lag period of ˜100 seconds, followed by slow phase separation. As the quenching depth increases, the lag period shortens; for sufficiently high blue light activation, phase separation is initiated immediately after activation. Notably, as shown in FIGS. 5A and 5B, while shallow quenching (FIG. 5A) tends to give rise to the relatively round droplet-like assemblies such as those described above, deep quenching (FIG. 5B) leads to the formation of structures with highly irregular shapes. Small diffraction-limited puncta that appeared immediately upon blue light exposure grew in size over time, in large part due to sticking to one another, forming highly branched elongated structures. Consistent with the apparent gel-like nature of these assemblies, FRAP measurements reveal that the major fraction of molecules within these clusters do not exchange with the surrounding cytoplasm. Indeed, as shown in FIG. 6, as the quenching depth increases from shallow (310) to intermediate (320) to deep (330), the fraction of recovery decreases, implying an increase in the solid fraction. Thus, the material state of light-activated assemblies can be tuned by controlling the cytoplasmic location within the phase diagram.

The assembly of structures such as P granules, DDX4 puncta, and nucleoli also appear to be controlled through a combination of PTMs and protein concentration levels, which would similarly allow cells to move their cytoplasm into different regions of a high-dimensional phase diagram.

Referring again to FIG. 1, when cleavage tags are utilized, at least one cleavage tag (40) is typically inserted between the functional region and a protein that has been targeted for, e.g., purification. A wide variety of cleavage tags are envisioned, including but not limited to: self-cleaving tags, Human Rhinovirus 3C Protease (3C/PreScission), Enterokinase (EKT), Factor Xa (FXa), Tobacco Etch Virus Protease (TEV), and Thrombin (Thr).

The functional region may also utilize a repeatable element. As shown in FIG. 7, the functional region can comprise repeatable units of a linker fused to a gene encoding a protein sensitive to light, where the number of repeats is typically, although not limited to, 1 to 20. Preferably the number of repeats is from 2 to 9, and more preferably the number of repeats is about 4.

Although many variants are envisioned, FIGS. 8A-C depict one example involving a first protein (810) construct which may have, for example, 4 tandem copies (812) of the GCN4 peptide (SEQ ID NO. 0001: EELLSKNYHLENEVARLKK), each separated by linkers (814), which is then fused to a fluorophore (816). A second protein construct (820) may comprise scFV-GCN4-sfGFP-Cry2, where the functional region comprises scFV-GCN4 (822), sfGFP (824) acts as a fluorophore, and the light sensitive region comprises Cry2 (826). As depicted in FIG. 8B, without light, the second protein construct binds to the peptide binding sites on the first protein construct. Upon exposure to light, Cry2-Cry2 binding occurs, creating large clusters of the first and second protein constructs, shown in FIG. 8C.

FIG. 8D illustrates several NIH3T3 cells, some of which only express mCh-10XGCN4 but not scFV-GNC4-sfGFP-Cry2 (red cells), while others express both constructs (yellow cells). FIG. 8E illustrates that upon light activation, yellow cells expressing both constructs exhibit light-activated clustering, while red cells expressing only mCh-10XGCN4 do not show any clustering, consistent with the fact that clustering for the system in the example above would require Cry2 multimerization.

At least three protein construct system configurations are also envisioned that utilize multiple, different protein constructs comprising repeatable units. In the first configuration, a system is envisioned wherein the light sensitive protein regions are identical, but the number of repeating units is different. An example of this is a Cry2(-linker-Cry2)_n arrangement, where the system contains three types of constructs, where n=2, 5, and 12. Although preferred embodiments utilize between about 2 and about 12 repeatable units, it is envisioned that a protein construct can utilize any number of repeatable units.

In the second configuration, at least two types of constructs are used, each having the same number of repeating units, but having different light sensitive regions. In a preferred embodiment, the system uses two types of constructs, each comprising at least a portion of one of a pair of proteins, such as Cry2-CIB, PhyB-PIF, or iLID-SspB.

This second configuration is based on the recognition that by changing the affinity and valency of protein-protein interactions we can control both the phase behavior and properties of the resulting droplets. The PhyB/PIF optogenetic system is able to change interaction affinities by varying the ratio of 650 nm and 750 nm light applied to the PhyB protein. The range of achievable interaction affinities is broadly tunable—individual PhyB-PIF interactions are very weak under pure 750 nm light (>100 μM), but very strong under pure 650 nm light (<100 nM). By changing the 650/750 nm ratio, any intermediate affinity can be attained. By using PhyB and PIF constructs with different number of repeats, the multivalency can be further tuned to induce phase separation under even modest concentrations (<1 μM). Other light-activatable proteins may also be used, such as the PHR domain of the protein Cry2. When activated with blue light (488 nm), these multimerized Cry2 constructs will phase separate into droplets or gel-like structures. This enables building light induced clusters, either within living cells or in vitro, with potential applications from perturbing intracellular organization, to purifying proteins through fusion with affinity purification tags and centrifuging the phase separated droplets.

In the case where the light activatable domains are PhyB/PIF pairs, one can express and purify in E. coli (BL21) polymers (e.g., 5-mers) of poly PhyB (PhyB₅), and 5-mers of poly PIF (PIF₅). It is also advantageous to include, for example, a TEV-cleavable His-tag, although other known methods for cleaving are envisioned.

1 These constructs can then be mixed, and illuminated with defined ratios of 650/750 nm light from computer-controlled LED sources. Lower repeat numbers, e.g. PhyB₃ and PIF₃, as well as higher repeat numbers, e.g. PhyB ₁₀ and PIF ₁₀ may also be used. Moreover, it is possible to use mixed repeat number solutions, i.e., PhyB_M+PIF_N, where M≠N under different light activation settings (650/750 nm), to optimize for maximum optical control of assembly under different physiological protein concentrations ([PhyB_M], [PIF_N]≦μM); using where M≠N allows for selective sequestration. The precise concentrations and repeat number of the various constructs, together with the degree of light activation, allows for control over the phase behavior and properties of the resulting assembles, which may be liquid like or more solid like, as shown in the example schematic in FIG. 9A.

Conversion of molecular species from weak self-association state to high self-association one, for example through post-translational modification or exposure of RNA in RNP complexes, leads to liquid-liquid phase separation. As shown in FIG. 9B, when the depth is shallow, this process follows the green path (250) to produce liquid droplets (251). Deep supersaturation along the red path (252) results in the formation of solid-like gels (253), with arrested molecular dynamics. Gels are initially reversible, but slow dynamics within promote the formation of irreversible aggregates over time (254).

Disassembly dynamics of these constructs, upon turning off blue light, has also been tested. As shown in FIG. 10, Without FUS_N, Cry2WT (410) only forms noticeable clusters in a small subset of cells, but these disassemble relatively quickly and show no irreversibility even after multiple activation cycles. By contrast, under the same conditions (shallow quench), the optoFUS construct (430) forms clusters which disassemble at a rate roughly 3 fold slower than clusters seen in Cry2WT alone (410); this indicates that the self-associating FUS_N chains interact with sufficient strength to delay dissolution. Gel-like optoFUS clusters assembled from deep quenching (440) also shrink in size upon turning off blue light, while maintaining their overall irregular morphology, and appear to be completely dissolved by ˜20 min. Interestingly, despite their reversiblity after the first activation cycle, these gel-like clusters disassemble ˜1.4 times slower than liquid-like optoFUS droplets (shallow quench) (430), suggesting the material state of clusters impacts the disassembly rate.

When cells expressing optoFUS undergo a sequence of repeated cycles of deep quenching (440), some clusters appear to remain as early as the end of the second cycle (442). By the third cycle, roughly 20% of clusters were not fully dissolved. Concomitantly, the disassembly rate of gel-like clusters gradually slowed down over subsequent cycles. When tested up to five cycles, the number of remaining clusters increases progressively for each cycle. These aggregates are truly irreversible: after the cessation of light activation cycles, they remain assembled for at least 6 hours. Remarkably, no irreversible clustering is observed in liquid-like optoFUS clusters, formed through cycles of shallow quenching (430). One interpretation for these results is that deep quenches develop irreversible aggregates simply because more material has assembled into each cluster. However, even when the total amount of phase separated material is smaller than cells with liquid droplets, cycles through the gel state robustly accumulate irreversible aggregates, confirming that the gel state specifically promotes irreversible aggregate formation.

The gel state provides a crucible for promoting irreversible aggregate formation. This irreversibility is reminiscent of observations of clusters remaining after just a single round of assembly in cells expressing FUS_N-Cry2olig (FIG. 10, 420). Since FUS_N-Cry2olig clusters form gels regardless of quenching depth, prolonged incubation of molecules in the gel state due to the slow inactivation rate of Cry2olig may be enough to induce irreversible aggregate formation even from a single round of quenching.

Dynamically tuning protein interactions with light achieves high degree of control over intracellular phase space, which can be exploited to study the phase diagram of FUS-mediated assemblies within living cells. Varying the degree of quenching depth leads to clusters spanning different material states, ranging from liquid droplets to gels. Shallow quenching leads to liquid droplets, similar to those observed with FUS and other proteins both in vitro and in vivo. However, deep quenching results in the formation of gels, which exhibit minimal molecular dynamics and highly irregular aggregate-like morphologies. These assemblies are reminiscent of gel-like structures previously observed in vitro for a variety of globular proteins. Notably, lysozyme, a well-folded protein whose phase behavior has been extensively studied in vitro, exhibits liquid-liquid phase separation at modest supersaturation, but upon deep quenching exhibits phase separation whose progress is arrested, with the condensed material forming a solid-like gel network. The gel state appears to represent a kinetically trapped state arising from the slow relaxation between strongly interacting molecules, rather than a thermodynamically favored state. Over time, such gels can develop into crystals and fibers.

This suggests that increasing the strength or effective valency of molecular self-association (e.g., through light activation or endogenously through PTMs) can lead to liquid-liquid phase separation, or for higher supersaturation can result in gelation. It is known that membrane-less organelles can exhibit at least partially solid-like properties (i.e., viscoelasticity). Indeed, large variations in the immobile fraction of stress granule proteins are often measured in FRAP experiments, and in some cases stress granules begin to resemble irregularly shaped gels. These apparent differences in material state reflect different depths into the cytoplasmic phase diagram. This ability to tune material states by moving within the phase diagram could be exploited by cells, since highly dynamic liquid-like states may be useful as microreactors, while gel-like structures would provide an ideal storage environment. However, assembling such arrested, gel-like structures deep within the phase diagram comes with the danger of producing potentially toxic species, due to irreversible aggregation and fibrillization.

A method for protein purification, utilizing these constructs, is illustrated in FIG. 11. As described above, the method (500) generally comprises at least seven steps: providing the protein construct (510) as described above, where the protein construct also comprises a target protein (see, e.g., FIG. 1, element 50). The construct is expressed (520) in cells. Cells are then lysed (532) and cellular debris is removed. The induction step (534) is followed by centrifugation to remove molecules other than the clustered protein construct. A cleaving step (536) follows, where the target protein is cleaved from, for example, an intrinsically disordered protein. After cleaving, a second induction step (538) is utilized. The second induction is typically for separating the light-activatable constructs from the desired protein to be purified. After the second induction step (538), the clusters are separated (539), typically via centrifuge. When a goal is, for example, to purify molecules interacting with the target proteins, the method may be modified slightly. The living cell is exposed to at least one wavelength of light that the light sensitive proteins are responsive to, which induces molecules within the living cell to cluster (530). The induction step can modify at least one of the transport or reactivity of enzymes and other molecules within the living cell, and/or cause intermolecular interactions, protein activation or inactivation, manipulation of signaling pathways, or gene expression through the induction of membrane-less bodies. Cells are gently lysed (540) and the induced clusters are then separated (550), typically via centrifuge, and the separated molecules are then purified (560) using typical protein purification methodologies.

In some embodiments, an induction step may also lead to nucleating droplets of tunable viscoelasticity at defined genomic loci, using at least one of LacO arrays or dCas9. The engineered dCas9 with peptide repeats, for example GCN4 peptide (SEQ ID NO. 0001: EELLSKNYHLENEVARLKK) or GFP11 (SEQ ID NO. 0002: RDHMVLHEYVNAAGIT), is co-expressed with a construct comprising the first segment of peptide-binding protein, either scFV-GCN4 or GFP1-10, and the second segment of FUS IDR. Coexpressing sgRNAs programmed for targeting specific genomic loci delivers dCas9 complexes with FUS IDR to the loci, which serves as a seed for droplet assembly. The viscoelasticity of droplets is tuned using the similar strategy described above, a varying degree of supersaturation.

The platform can also facilitate catalytic activity upon photo-stimulation by locally concentrating enzymes inside or outside cells, for instance for intracellular production of natural products, biofuels etc. This may be accomplished by, for example, recruiting enzymes into the phase separated environment generated by the clustering of intrinsically disordered protein regions.

Kits may also be provided to simplify the use of these methods. The kits will generally include a protein construct as described above, as well as at least one light emitting device that can be used to activate the light sensitive proteins of the protein construct.

Thus, specific constructs and methods which can be used for, e.g., rapid and reversible clustering of proteins, have been disclosed. It should be apparent, however, to those skilled in the art that many more modifications besides those already described are possible without departing from the inventive concepts herein. The inventive subject matter, therefore, is not to be restricted except in the spirit of the disclosure. Moreover, in interpreting the disclosure, all terms should be interpreted in the broadest possible manner consistent with the context. In particular, the terms “comprises” and “comprising” should be interpreted as referring to elements, components, or steps in a non-exclusive manner, indicating that the referenced elements, components, or steps may be present, or utilized, or combined with other elements, components, or steps that are not expressly referenced.

Claims

1. A protein construct comprising: a first segment comprising at least one gene encoding at least one protein sensitive to at least one wavelength of light; anda second segment fused to the first segment, the second segment comprising at least one of: a low complexity sequence (LCS); intrinsically disordered protein region (IDR); a synthetic or natural nucleic acid binding domain; or at least one repeatable sequence, the repeatable sequence comprising a linker fused to at least one additional gene encoding at least one protein sensitive to at least one wavelength of light.

2. The protein construct according to claim 1, wherein the IDR is a portion of a first protein selected from the group consisting of FUS, Ddx4, and hnRNPA1.

3. The protein construct according to claim 1, wherein the at least one protein is selected from the group consisting of Cry2, Cry2olig, PhyB, PIF, light-oxygen-voltage sensing (LOV) domains, or Dronpa.

4. The protein construct according to claim 1, wherein the at least one wavelength of light is between and including 400 and 800 nm.

5. The protein construct according to claim 1, wherein the IDR is a portion of FUS, the at least one protein is Cry2, and the N-terminal IDR of FUS is fused to the photolyase homology region of Cry2.

6. The protein construct according to claim 1, further comprising at least one fluorophore fused to the at least one protein or fused between the IDR and the at least one protein.

7. The protein construct according to claim 6, wherein the fluorophore is mCherry.

8. The protein construct according to claim 6, wherein the IDR is a portion of FUS and the at least one protein is Cry2, and wherein the IDR is fused to a mCherry-labeled Cry2 photolyase homology region.

9. The protein construct according to claim 1, wherein the protein construct is comprised of between 2 and 9 repeatable sequences.

10. The protein construct according to claim 1, wherein the synthetic or natural nucleic acid binding domain is selected from the group consisting of: RNA recognition motif (RRM), double-stranded RNA binding domain (dsRBD), S1, zinc finger binding domains, YT521-B homology (YTH), DNA and RNA helicase domains, Pumilio, or S-adenosylmethionine (SAM) structures.

11. A cell line that expresses the protein construct according to claim 1.

12. A protein construct system, comprising a first protein construct of claim 1 comprising a first number of repeatable sequences; anda second protein construct of claim 1 comprising a second number of repeatable sequences.

13. The protein construct according to claim 12, wherein the at least one protein pair is selected from the group consisting of Cry2-CIB, PhyB-PIF or iLID-SspB.

14. A method for protein purification, comprising the steps of: providing a protein construct comprising:a first segment comprising at least one gene encoding at least one protein sensitive to at least one wavelength of light;a second segment fused to the first segment, the second segment comprising at least one of: a low complexity sequence (LCS); intrinsically disordered protein region (IDR); a synthetic or natural nucleic acid binding domain; or at least one repeatable sequence, the repeatable sequence comprising a linker fused to at least one gene encoding at least one protein sensitive to at least one wavelength of light; anda third segment comprising at least one target protein to be purified. introducing the protein construct into a living cell;inducing molecules within the living cell to cluster by exposing the protein construct to at least one predetermined wavelength of light at a predetermined intensity for a predetermined period of time.

15. The method according to claim 14, wherein inducing the molecules to cluster modifies at least one of the transport or reactivity of molecules within the living cell.

16. The method according to claim 14, wherein inducing the molecules to cluster causes intermolecular interactions, protein activation or inactivation, manipulation of signaling pathways, or gene expression through the induction of membrane-less bodies.

17. The method according to claim 14, further comprising the steps of: separating the clustered molecules via centrifuge; andpurifying the separated molecules.

18. The method according to claim 14, wherein the protein construct further comprises a cleavage tag between the target protein and the intrinsically disordered region.

19. The method according to claim 18, wherein the cleavage tag is selected from the group consisting of: Human Rhinovirus 3C Protease (3C/PreScission), Enterokinase (EKT), Factor Xa (FXa), Tobacco Etch Virus Protease (TEV), and Thrombin (Thr).

20. The method according to claim 18, wherein the cleavage tag is a self cleaving tag.

21. The method according to claim 14, further comprising the steps of: cleaving the target protein from the intrinsically-disordered protein;inducing the formation of clusters or nucleation of pathological protein aggregates by exposing the light-sensitive protein to at least one predetermined wavelength of light; andseparating the clusters via centrifuge.

22. The method according to claim 14, further comprising nucleating droplets of tunable viscoelasticity at defined genomic loci, using at least one of LacO arrays or dCas9.

23. The method according to claim 14, further comprising locally concentrating a plurality of enzymes.

24. A kit for inducing in vivo aggregation or liquid-liquid phase separation, comprising: one or more aliquots of composition comprising:a protein construct comprising:a first segment comprising at least one gene encoding at least one protein sensitive to at least one wavelength of light; anda second segment fused to the first segment, the second segment comprising at least one of an intrinsically disordered protein region (IDR) or at least one repeatable sequence, the repeatable sequence comprising a linker fused to at least one gene encoding at least one protein sensitive to at least one wavelength of light; andat least one light emitting device configured to activate at least a portion of the protein construct.