Protein phase separation and determinants of in cell crystallization

Liquid‐liquid phase separation (LLPS) in cells is known as a complex physicochemical process causing the formation of membrane‐less organelles (MLOs). Cells have well‐defined different membrane‐surrounded organelles like mitochondria, endoplasmic reticulum, lysosomes, peroxisomes, etc., however, on demand they can create MLOs as stress granules, nucleoli and P bodies to cover vital functions and regulatory activities. However, the mechanism of intracellular molecule assembly into functional compartments within a living cell remains till now not fully understood. in vitro and in vivo investigations unveiled that MLOs emerge after preceding liquid‐liquid, liquid‐gel, liquid‐semi‐crystalline, or liquid‐crystalline phase separations. Liquid‐liquid and liquid‐gel MLOs form the majority of cellular phase separation events, while the occurrence of micro‐sized crystals in cells was only rarely observed, however can be considered as a result of a preceding protein phase separation event. In vivo, also known and termed as in cellulo crystals, are reported since 1853. In some cases, they have been linked to vital cellular functions, such as storage and detoxification. However, the occurrence of in cellulo crystals is also associated to diseases like cataract, hemoglobin C diseases, etc. Therefore, better knowledge about the involved molecular processes will support drug discovery investigations to cure diseases related to in cellulo crystallization. We summarize physical and chemical determinants known today required for phase separation initiation and formation and in cellulo crystal growth. In recent years it has been demonstrated that LLPS plays a crucial role in cell compartmentalization and formation of MLOs. Here we discuss potential mechanisms and potential crowding agents involved in protein phase separation and in cellulo crystallization.

as stress granules, nucleoli and P bodies to cover vital functions and regulatory activities. However, the mechanism of intracellular molecule assembly into functional compartments within a living cell remains till now not fully understood. in vitro and in vivo investigations unveiled that MLOs emerge after preceding liquid-liquid, liquidgel, liquid-semi-crystalline, or liquid-crystalline phase separations. Liquid-liquid and liquid-gel MLOs form the majority of cellular phase separation events, while the occurrence of micro-sized crystals in cells was only rarely observed, however can be considered as a result of a preceding protein phase separation event. In vivo, also known and termed as in cellulo crystals, are reported since 1853. In some cases, they have been linked to vital cellular functions, such as storage and detoxification. However, the occurrence of in cellulo crystals is also associated to diseases like cataract, hemoglobin C diseases, etc. Therefore, better knowledge about the involved molecular processes will support drug discovery investigations to cure diseases related to in cellulo crystallization. We summarize physical and chemical determinants known today required for phase separation initiation and formation and in cellulo crystal growth. In recent years it has been demonstrated that LLPS plays a crucial role in cell compartmentalization and formation of MLOs. Here we discuss potential mechanisms and potential crowding agents involved in protein phase separation and in cellulo crystallization. In context of liquid-liquid phase separation (LLPS) of biomolecules into liquid condensates in cells occasionally protein crystals were observed. 1 The observation of crystals in living cells, reported in the literature as in cellulo or in vivo crystals is known for decades and was repeatedly described as a natural phenomenon. [1][2][3][4][5][6] Examples are crystals of storage proteins in seeds, [7][8][9] insulin crystals within secretory granules, 10 solid state catalysts, 11,12 biocrystallization of the DNA-binding protein (Dps) and DNA as a response to cellular damage and stress, [13][14][15] or β-hematin crystals produced as a detoxification strategy by malaria parasites. [16][17][18][19] Although in cellulo crystallization still is a new exciting area in cell biology, many natively crystallizing proteins in living cells first function as storage such as vitellin yolk protein crystals from bullfrog oocytes, 20 from leopard frogs, 21 35 Moreover, crystals occurrence is related to wound sealing in case of Hex-1 from Neurospora crassa, fungal crystals seal the septal core, 36,37 and P-protein from V. faba. 38 Those abovementioned natively occurring crystals have been detected either by powder diffraction or electron microscopy. In cellulo/in vivo crystallization or crystalline matter in cells has been observed, however the individual precise function remains widely speculative. It may be harmless or harmful and even useful like storage of proteins or peptides, or removal of toxins.
Further, in cellulo crystallization has also been associated with several diseases like cataract, [39][40][41][42] hemoglobin C diseases, 43 formation of Charcot-Leyden crystals (CLCs), [44][45][46][47] Reinke's crystals [48][49][50]  diseases. 45 A recent study solved the Galectin 10 structure using CLCs from patients with rhinosinusitis and asthma. Further, CLCs bind and are dissolved by antibodies. 46 This opens new avenues to explore disease related in vivo crystals as drug target. 47 In cellulo protein crystallization has also gained attention as a new and alternative method to produce high amounts of micro-or nano-sized crystals which can be used to determine the 3D structure of the crystallized protein using either high brilliant X-ray free electron laser 67,68 or highly brilliant micro-focused synchrotron radiation applying serial diffraction data collection. 69,70 However, despite an increasing number of publications reporting in vivo crystallization, the physicochemical parameters required and the molecular mechanism of in vivo crystallization guiding crystallization in cells are up to date only poorly understood, considering that even conventional, in vitro, protein crystallization till now remains a challenge. 71,72 In a first assumption, it seems reasonable to consider the process to be analogous to crystal formation in vitro, where a purified protein at relatively high concentration is used for crystallization screening experiments. By mixing the protein solution with different precipitant agents, including salts or organic polymers, the phase diagram of crystallization is systematically screened to achieve supersaturation of the protein and to reach conditions required for nucleation. Additionally, parameters such as temperature, pH, and others are typically optimized in in vitro protein crystallization experiments. 73,74 However, the identification of in vitro protein crystal nucleation conditions is till now a trial and error process, and mostly unpredictable. Therefore, today in vitro protein crystallization experiments are routinely performed by screening many conditions applying either vapor diffusion, dialysis, counter diffusion, or batch crystallization techniques. 71,74 During in vitro crystallization, a protein solution is brought to supersaturation to first induce liquid dense cluster (LDC) formation as a precursor followed by nucleation and further crystal growth, which thermodynamically is a second-order transition. [75][76][77] In this context, several in vitro investigations are focused to obtain insights about the LDC formation and nucleation pathways in order to understand protein crystallization in more detail. [78][79][80][81][82][83][84] This data can in principle also support understanding of in vivo crystallization. Likewise, in vivo crystallization at first requires a high local concentration of the protein to render crystallization thermodynamically favorable. Therefore, protein sorting into organelles with limited space seemed originally a prerequisite for in vivo crystallization. However, occurrence of in vivo crystals within the cytosol 3,42,57,59-63 may also reflect preceding LLPS events. To reach the required protein supersaturation for nucleation and crystal growth, phase separation is supported by crowding agents in the cell, which promote attractive protein-protein interactions and act similar to precipitants in in vitro experiments. 76,85,86 in vivo crystallization was observed in different cells and specific organelles.
However, the in vivo nucleation mechanism and the molecular components, like crowding agents, essential for phase separation and crystallization are still challenging to address experimentally. 42,87 LLPS is a phenomenon denoting the demixing of structurally different molecules in aqueous solution above a certain concentration, considering a distinct physicochemical environment. 88 LLPS is known to be the primary process underlying, for example, the formation of stress granules, the nucleolus, or P bodies. 86 85,[92][93][94][95] LDCs are densely assembled molecules in aqueous solution, appearing in context of LLPS, which can be considered as a mandatory precursor of a nucleation process preceding the in vitro protein crystallization process. 76 As an example, lysozyme undergoes LLPS, gelation, and crystallization depending on certain conditions of temperature, precipitants and protein concentration. 85,96 More recently it was demonstrated that oligomeric peptides can undergo LLPS when stimulated by low temperature, crowding agents such as polyethylene glycol (PEG), or a pH sometimes close to their isoelectric points. 97 Crowding agents are used for conventional protein crystallization, 98 but also used to investigate the formation of the nucleolus, 99,100 protein stabilization and folding, 101,102 and formation of other MLOs 103 in vitro. Further, molecular crowding was also noticed to promote amyloid formation. 104 Crowding agents, such as PEGs, dextrans, and even low molecular compounds like trimethylamine N-oxide (TMAO) have been used for in vitro investigations of LLPS. [105][106][107][108] From these in vitro experiments, it can be concluded that LLPS is obviously the process that governs the formation of membrane-less compartments in cells, which can occur in all different cell organelles or in the cytosol. This assumption is supported by the fact that in vivo grown crystals were observed in different organelles, such as rough endoplasmic reticulum (rER), mitochondria, lysosomes, peroxisomes or the nucleus, as indicated in Figure 1. These facts led us to consider that LLPS might not only be linked to cell compartmentalization or disease-related protein aggregation but may also be a prerequisite for in vivo crystallization.

| IN VIVO CRYSTALS
For more than a century it has been observed that protein crystallization occurs within living cells. 4 In Table 1  but till now no X-ray diffraction study has been reported. 59  crystals is a highly dynamic process and that these crystals were located either inside of peroxisomes or within the cytosol, respectively. So far, characterization of firefly luciferase using scanning electron microscopy (SEM) and GFP-μNS using X-ray powder diffraction studies have been reported. 63 Tsutsui et al reported expression of an Xpa Coral protein in HEK293 cells, were crystals were encapsulated by autophagosome/lysosomal membranes (although some crystal-like structures were also found in the nucleus). The authors describe that selective autophagy engulfs the crystals into a cargo within the cells. 56 Baskaran et al described in vivo crystallization and X-ray structure analysis of human PAK4 in complex with its inhibitor Inka1. no structure or X-ray diffraction data were published for these proteins. 42 The author also speculates about potential crowding agents, such as cellular proteins and especially organelle resident proteins that might support in vivo crystallization of neuraminidase, IgG, γ-crystallin D or CLC. They also emphasize the urgent need to identify intracellular crowding agents or external factors to enhance or to predict the possibility to obtain protein crystals in a cell organelle. 42 In terms of our own investigations, we obtained in vivo grown crystals for Trypanosoma brucei (Tb) CatB and IMPDH using the baculovirus expression system (Sf9 insect cells and High five cells). 2 Initially, we performed a bioanalytical characterization that revealed the identity of the crystallized material and showed the homogeneity of the intracellular crystal lattices by TEM and SEM (Figure 2). Analysis of the TEM micrographs revealed that TbCatB crystals were exclusively located within the rER and TbIMPDH crystals in the cytosol ( Figure 2). We applied those crystals for serial diffraction data collection and could solve and refine the structures to 2.4 and 2.8 Å resolution, respectively. 2,3,53 T A B L E 2 Summary of potential in cellulo crowding agents

Crowding agents Chemico-physical properties and functions References
Ribonucleotide: ATP, GTP, UTP, CTP; Deoxynucleotides: dATP; dGTP, dCTP, dTTP Source of energy, affecting protein solubility and preventing macromolecular aggregation (ATP hydrotropic activity). Dissolving LLPS droplets and amyloid fibers Traut, 127 Rice and Rosen, 128  Weber and Brangwynne, 130 Banani et al, 91 Langdon and Gladfelter, 131 Faya and Anderson, 132  Poudyal et al, 133 Hasegawa 42 Chemical chaperones/osmolytes, and other low-molecular-weight metabolites (carbohydrates, fatty acids and sterols, etc.), inorganic ions (Mg 2+ , H + /OH − , etc.) Osmolytes or molecular chaperones are known to enhance or reduce the stability of protein molecules. They can trigger non-covalent protein-protein interactions to initiate LLPS. They are used as precipitant agents in vitro protein crystallization. Metabolites and ions are required for folding and catalytic activity of many enzymes. The metabolic enzyme can be regulated through phase separation by interaction with some metabolite Diamant et al, 134 Papp and Csermely, 135 Marshall et al, 136 Frankel et al, 129 Poudyal et al, 133 Hasegawa, 42 Prouteau et al 140

| Crowding agents triggering LLPS
To date it remains challenging to investigate in cell phase separation phenomena on a molecular level. 123,124 Therefore, several investigations on LLPS and biomolecular condensates are currently performed applying well-defined in vitro systems using selected proteins and distinct crowding agents. 106,108 Synthetic macromolecular crowders, such as PEG, dextran or ficoll are mainly in use to investigate their single or synergy effects to induce or enhance LLPS in vitro. 124 The "self-assembling" of proteins, such as FUS/TDP43, α-synuclein, tau, Aβ and the huntingtin into insoluble fibers that can even further aggregate is under investigation for some years, also in context to identify crowding agents initiating or supporting the molecular assembly. 125,126 In Table 2 we summarize information about crowding agents in use for in vitro LLPS assays. Low molecular weight nucleotides such as ATP, which is not only acting as an energy source in cells, was recognized to prevent aggregation to amyloid fibers of FUS, however also can dissolve LLPS droplets and amyloid fibers. 128 values. 129 Hence, RNAs, especially RNAs related to the ribosomal translation machinery, may stabilize and/or surround a homogeneous protein phase separation state. 130 Further, RNAs were already identified to be involved in LLPS in vitro. 137,138 Compared to nucleic acids, the structural diversity of proteins in cells is much higher and many site-specific proteins are available in different quantities in different cellular organelles. Individual proteins were shown to initiate fiber and amyloid formation and surface properties or catalytic activities of a protein may be required to support LLPS formation. 139 This might be either directly or by regulating the water content, the pH value or the abundancy of another crowding agent in close proximity. The involvement of proteins in such a process was initially discussed by Hasegawa. Further experimental studies in the field, for example, to identify differences in the proteome during crystal formation, remain to be performed.
Small molecules with diverse chemical and structural properties are known to influence (enhance or reduce) the stability of protein molecules. 135 Resulting changes of the tertiary structure of a protein can trigger and essentially influence noncovalent protein-protein interactions to initiate LLPS and even crystalline lattice order.
Thereby, these molecules might also act as in cellulo crowding agents and may be considered as a precipitating agent for in vitro protein crystallization. 42,136 Typically those compounds are called osmolytes acting as molecular chaperones. 134,135 This group of structurally diverse molecules includes sugars, amines like betaine, urea and peptides. 140,141 Prouteau et al also reviewed that metabolic enzymes can even be regulated through phase separation, which could further regulate the availability of metabolites, acting as crowding agents. 137,142 Overall, next to other reasons, like local stress-response, pH value or local water content, the availability of certain metabolites most probably influences the location that is suitable to initiate clustering of a protein within a cell.