WO2020048996A1 - Self-assembling globular proteins and uses thereof - Google Patents

Abstract

A fusion system, in particular a protein, is proposed exhibiting a phase transition, the fusion protein comprising: (a) one or more core system, e.g. a biologically and/or chemically active, globular, water-soluble molecules of interest selected from at least one of the group consisting of proteins and peptides or derivatives thereof; (b) at least two proteins exhibiting a phase transition joined to the core e.g. at the N as well as the C terminus, respectively, to the biologically active molecule of interest; and (c) optionally, a spacer or intermolecular interaction sequence separating any of the proteins of (b) from any of the biological molecule(s) of interest of (a), wherein the proteins of (b), taken individually, together or both individually and together, have a length of at least 40 amino acids, are intrinsically disordered, and do not show a repetitive amino acid pattern.

Description

TITLE

SELF-ASSEMBLING GLOBULAR PROTEINS AND USES THEREOF

TECHNICAL FIELD

The present invention relates to fusion proteins allowing tailored liquid/liquid phase separation, and for example to uses thereof by maturing them into an irreversible stable state and e.g. recruiting further molecules into the corresponding globular structures.

PRIOR ART

Proteins are intricate macromolecules that are capable of performing specific tasks by interacting with other proteins and biomolecules. In many cases, the activity of proteins requires the self-assembly into complexes and aggregates. Examples in nature include actin filaments, which provide mechanical support in eukaryotic cells, and amyloid structures, which have been recently associated with a variety of functions including hormone storage and biofilm formation.

As a result of their high specificity, biocompatibility, and safety, self-assembling protein materials have increasingly found many applications in biotechnology. In this context, the ability to encode specific self-assembly properties by engineering the protein’s primary amino acid sequence represents a flexible approach to generate a variety of functional materials. Moreover, the possibility to accurately control the size and the surface chemistry of the protein materials confers tailored properties in terms of mass transport and activity. Examples include peptides that self-assemble into filaments and gels, protein nanocages, proteinosomes and functional amyloids. These materials have found applications in a variety of fields, including food science, water treatment, tissue engineering, vaccine delivery, biosensors and enzyme technology.

US-A-2017355977 discloses a protein construct including a gene encoding a light-sensitive protein fused to at least one of either a low complexity sequence (LCS), an intrinsically disordered protein region (IDR), or a repeating sequence of a linker and another gene encoding a light-sensitive protein. The protein construct may also include cleavage tags. The 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 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 mentioned, the kit including the protein construct and a light source capable of producing a wavelength that the light-sensitive protein will respond to.

US-A-2014106399 relates to a method for production and purification of polypeptides. In particular, it relates to a fusion protein comprising a solubility-enhancing peptide tag moiety, a moiety of target peptide, and a self-aggregating peptide moiety fused in this order and to a method for production and purification of target peptides through expressing said fusion protein.

WO-A-2012045822 proposes a thermo-responsive polymer covalently bound with a peptide, wherein the peptide comprises a peptide moiety that is able to self-assemble and a functional peptide moiety comprising a bioactive sequence.

EP-A-2407479 discloses a peptide gel with practically sufficient mechanical strength and a self-assembling peptide capable of forming the peptide gel with a particular amino acid sequence.

US-A-2001034050, US-A-2005255554 as well as US- A-20151 12022 describe environmentally responsive polypeptides capable of displaying stimuli-triggered conformational changes in a reversible or irreversible manner that may be accompanied by aggregation. Polypeptides include a number of repeated motifs and may be elastomeric or non-elastomeric, inspired by elastin in the form of elastin like peptides (ELPs), e.g. using a block-copolymeric structure with VHPGVG blocks.

WO- A-2016196249 discloses recombinant ELPs comprising one or more homologous amino acid repeats; and, non-immunogenic bioconjugates comprising recombinant polypeptides comprising one or more homologous amino acid repeats and one or more therapeutic agents. Also, disclosed are pharmaceutical compositions including the recombinant polypeptides; and methods of administering the recombinant polypeptides to patients for the treatment of cancer or infections.

US 2012/0088268 discloses repetitive precursor proteins with the aim of using them for producing antimicrobial peptides.

US 6,818,61 1 proposes to attach stabilizing groups in the form of small peptides or small stable proteins to a bioactive peptide for screening purposes. The stabilizing group can take the form of a small stable protein, such as the Rop protein, glutathione sulfotransferase, thioredoxin, maltose binding protein, or glutathione reductase, or one or more proline residues. An intracellular selection system allows concurrent screening for peptide bioactivity and stability. Randomized recombinant peptides are screened for bioactivity in a tightly regulated expression system, preferably derived from the wild-type lac operon. Bioactive peptides thus identified are inherently protease- and peptidase-resistant.

EP-A-2664340 relates to a method for delivering a drug depot of a compound of interest to a selected region in a subject. The method comprises administering a composition directly to said region of interest, the composition comprising the compound of interest to be delivered (such as an anti-inflammatory agent or a chemotherapeutic agent) and a polymer (such as an elastin-like peptide or ELP) that undergoes an inverse temperature phase transition, so that a sustained release of the compound of interest at the selected region is provided. Compositions useful for carrying out the invention are also described.

US-A-2008032400 discloses ELP fusion proteins, multimeric ELP spider complexes formed of ELP fusion proteins, and methods of using the same. The construct may be in the form of an ELP spider structure complex including multi-leg moieties comprising ELP fusion proteins capable of forming covalent disulfide bonds. The multimeric fusion constructs may be employed in peptide production and purification and/or to enhance proteolytic resistance of a protein or peptide moiety in a fusion construct, by provision of the fusion protein in an ELP spider complex.

Schuster et al (Nature Communications volume 9, Article number: 2985 (2018) report on many intrinsically disordered proteins self-assembling into liquid droplets that function as membrane-less organelles. They manipulated the intrinsically disordered, arginine/glycine- rich RGG domain from the P granule protein LAF-l to generate synthetic membrane-less organelles with controllable phase separation and cargo recruitment. They demonstrated enzymatically triggered droplet assembly and disassembly, whereby miscibility and RGG domain valency were tuned by protease activity. Second, they controlled droplet composition by selectively recruiting cargo molecules via protein interaction motifs. Further they demonstrated protease-triggered controlled release of cargo. Droplet assembly and cargo recruitment were reported to be robust, occurring in cytoplasmic extracts and in living mammalian cells.

Protter et al (Cell Rep. 2018 Feb 6;22(6): 1401-1412), Martin et al (Biochemistry. 2018 May 1 ;57(l 7):2478-2487), as well as Aumiller et al (Adv Colloid Interface Sci. 2017 Jan;239:75- 87) report on phase transition and compartment formation of LCS systems and their interactions with other proteins. Cheng et al (RNA. 2005 Aug; 1 1(8): 1258-70) report on Crystal structure and functional analysis of DEAD-box protein Dhhlp. Faltova et al (ACS Nano 2018 1210 9991-9999) report on multifunctional protein materials and microreactors using low complexity domains as molecular adhesives.

SUMMARY OF THE INVENTION

Despite these successes, there is still a need to develop strategies to introduce functionalities into self-organized structures in a simple and flexible manner, for example if multiple functions are desired within one same structure. Current approaches to generate multifunctional peptide-based materials rely on the post-functionalization of pre-formed scaffolds, such as nanoparticles and vesicles, with different proteins. This strategy requires the careful control of the conjugation reactions to avoid modifications of protein structure and loss of the material. The development of functional proteins with intrinsic selfassociation properties would represent an attractive alternative to simplify the aforementioned strategy and convey varying functionalities into a single structure.

Recent findings indicate that a class of disordered amino acid sequences promotes functional phase transition of biomolecules in nature. Such sequences consist of low complexity domains (LCDs) of a few amino acids. In this document, these sequences are exploited by conjugating them to soluble globular domains to develop molecular adhesives that enable sensitive, controlled self-assembly of these proteins into supramolecular architectures. In particular, the enzyme adenylate kinase and the green fluorescent protein were used as soluble domains, and it is shown that the addition of low complexity regions induces the formation of protein particles via a multi-step process. This multi-step pathway involves an initial liquid-liquid phase transition, which creates protein-rich droplets that can be matured into protein aggregates over time. These protein aggregates consist of porous structures that maintain activity and can release active soluble proteins. It is shown that the LCDs dictate specific non-covalent intermolecular interactions and phase properties that for many systems can be largely independent of the given globular domain. It is further demonstrated that this feature, together with the dynamic state of the initial dense liquid phase, allows one to directly assemble different globular domains within the same architecture, thereby, inter alia, enabling the generation of multifunctional biomaterials and microscale bioreactors.

Chimera proteins are made that contain a functional, globular domain, responsible for the desired bioactivity, and an intrinsically disordered domain, which acts as a molecular adhesive and enables highly sensitive, controlled self-assembly. Here this is done by exploiting sequences from proteins that undergo in vivo phase transition processes associated with the formation of cellular compartments. These motifs are intrinsically disordered and are commonly referred to as low complexity domains (LCD) or low complexity regions (LCR). These sequences typically contain a high fraction of charged and polar amino acids, such as glutamine, asparagine, serine, arginine and lysine, and a low fraction of hydrophobic residues. Furthermore, these motifs are also enriched in residues which disrupt secondary structure, such as proline and glycine. In these sequences, multivalent attractive interactions between side chains are mediated by poorly soluble polar residues and the highly extended nature of these proteins in aqueous environments. This positive energetic contribution counteracts the entropic loss associated with de-mixing, and consequently leads to liquid-liquid phase separation of protein solutions.

The advantages of these LCDs are two-fold. First, these domains offer the unique possibility to accurately induce a variety of specific intermolecular interactions using sequences of simple composition. Secondly, the attractive interactions mediated by such molecular adhesives can induce the formation of a variety of protein-rich phases, ranging from liquid to gel to solid. This high degree of flexibility presents an opportunity to design materials with desired biophysical properties, which are highly sensitive to external stimuli. It is becoming increasingly apparent that nature exploits this ability to induce controlled phase separation of proteins and nucleic acids in order to generate specific membraneless bodies which coordinate reactions in space and time.

Here, chimeric proteins are created in which LCDs are attached to a globular domain to create a hybrid molecule which maintains the functionality of the globular domain, while mimicking the colocalization or rather self-assembly behaviour of the disordered regions that are found in nature. It is shown that these molecules undergo liquid-liquid phase transitions to form droplets which mature into protein aggregates and solid particles over time. Interestingly, it can be shown that the dynamic state of the initial dense liquid phase allows one to manipulate structural properties that are maintained in the mature solid state. It is demonstrated that this strategy enables the development of protein particles for prolonged release of active proteins as well as microreactors and multifunctional biomaterials, in which different proteins can be directly recruited within the same structure. More generally speaking, the present invention relates to a fusion system, preferably a fusion protein, exhibiting a phase transition, the fusion system / fusion protein comprising:

(a) one or more biologically and/or chemically active core system, which core system is e.g. a globular, water-soluble or water insoluble molecule or particle of interest selected from at least one of the group consisting of proteins and peptides or derivatives thereof;

(b) at least two proteins exhibiting a phase transition joined (via a chemical bond or via a strong intermolecular interaction, in particular an antigen/antibody interaction, such as interaction between a His tag of or attached to a protein exhibiting a phase transition and a His tag antibody attached to the core system) to the core system, in case the core system is a protein or a peptide joined at the N as well as the C terminus, respectively; and

(c) optionally, at least one or serveral spacer or intermolecular interaction sequences (e.g. tag and antibody) separating any of the proteins of (b) from any of the core system or (biological) molecule(s) of interest of (a),

wherein the proteins of (b), taken individually, together or both individually and together, have a length of at least 40 amino acids, are intrinsically disordered, and preferably do not show a repetitive amino acid pattern.

Preferably, the core system is at least one organic or inorganic nanoparticle, preferably a magnetic and/or metallic and/or catalytic nanoparticle, and/or at least one water-soluble molecule of interest with low intermolecular interactions selected from at least one of the group consisting of: therapeutically active organic molecule, bio macromolecule, in particular protein, peptide, DNA, RNA, oligonucleotide, carbohydrate or derivatives thereof. The properties of the LCDs can thus also be transferred to inorganic nanoparticles, inducing controlled phase separation that exhibits dynamicity and stimulus-responsiveness to ionic strength and pH. We illustrate this concept with magnetic nanoparticles, generating dynamic protein-composite magnetosomes of controlled size and morphology and microreactors that can localize enzymatic reactions and sense the concentration of biomolecules in the surrounding environment. The dynamicity and adaptability of this class of proteins or protein segments can be transferred to inorganic nanoparticles, generating composite materials with controlled self-assembly properties and sensitivity to external cues, thereby creating stimulus-responsiveness. This concept is demonstrated with enzymes and magnetic nanoparticles, generating magnetosomes of controlled size and morphology in the micron range that can act as enzymatic microreactors and that can sense the concentration of biomolecules in the surrounding environment. Overall, this platform represents a convenient tool to generate protein-inorganic materials with highly controlled self-assembly behavior, stimulus-responsiveness, and multiple functionalities.

In particular in case of using inorganic nanoparticles several approaches are possible. One approach is that the fusion system consists of the above (a) and (b), optionally with (c), and where the biologically and/or chemically active core system of (a) is an inorganic nanoparticle.

Also possible is a situation where the system comprises two core systems (a), one of them being at least one water-soluble molecule of interest with low intermolecular interactions selected from at least one of the group consisting of: therapeutically active organic molecule, bio macromolecule, in particular protein, peptide, DNA, RNA, oligonucleotide, carbohydrate or derivatives thereof, e.g. adenylate kinase as described below, and one of them being an inorganic nanoparticle, ln this case it is preferred that the at least one water- soluble molecule of interest with low intermolecular interactions selected from at least one of the group consisting of: therapeutically active organic molecule, bio macromolecule is attached at both ends by way of a chemical bond each to (b) two proteins exhibiting a phase transition, and the inorganic nanoparticle is then joined, preferably not by way of a chemical bond but by way of a strong intermolecular interaction, e.g. an antibody/antigen interaction, to one of these proteins exhibiting a phase transition. To allow for that joining the inorganic nanoparticle can be surface modified with an element allowing for the strong intermolecular interaction, e.g. an antibody, and the protein exhibiting a phase transition comprises a sequence to selectively bind to that antibody, which sequence can be engineered into that protein if required.

Preferably the solubility at room temperature in aqueous buffer solutions at ionic strength lower than 150 mM is at least 0.5 g/L .

Preferably the fusion system can be reversibly switched by a change in the environment between the desired phases without any disassociation of the individual components, in particular without the requirement of enzymatic cleavage for phase transition. The change in the environment can be a change in temperature, pH and/or ionic strength. Preferably therefore the fusion system is free from cleavage sequences adapted for cleaving the fusion system, in particular adapted for cleaving the LCD domain from the core. Most preferably the fusion system is free from cleavage sequences for the tobacco etch virus (TEV) NIa protease and/or for thrombin and/or for the human rhinovirus 3C protease (HRV3C).

The core system is e.g. a protein or a peptide, preferably in globular form, and the at least two proteins exhibiting a phase transition are joined at the N as well as the C terminus, respectively, to the biologically/chemically active molecule of interest. If the core system is not a protein or peptide, the (b) at least two proteins exhibiting a phase transition can be joined by corresponding functionalization of the core.

Possible protein reactive groups which can be employed in covalent attachment are the amine group (-NH2) the thiol group (-SH), but conjugation reactions can also involve the hydroxyl group (-OH), the carboxyl group and carbonyl groups such as aldehydes and ketones. Examples of functionalization of polymeric and e.g. gold nanoparticles with proteins include covalent attachment via click-chemistry reactions via N- hydroxysuccinimide (NHS) ester, carbonyl diimidazole, organic sulfonyl chlorides, epoxides or maleimide groups present on the surface of the nanoparticles. For carbohydrates, examples of such functionalization with proteins include reductive amination, active ester and cyanylation, RNA molecules can be also labeled with proteins, for instance by generating RNA via solid-phase synthesis and converting the nucleobase amino or ketone functional groups to a thiol, thereby allowing for post-synthetic modifications.

When talking about a repetitive amino acid pattern, this means that the corresponding proteins of (b) preferably do not contain blocks of 2-10 or 3-8 amino acid which are repeated more than three times, or more than five times in an immediate sequence. In particular these proteins are not ELP type proteins.

The sequences of the proteins of (b), these are so-called low complexity sequences or low complexity domains (LCD) which preferentially encode cation-pi interactions (at least 10 residues) and/or positive-negative charge attraction (at least 10 residues).

The strategy to engineer the sequences is based on the set of acquired experimental data, investigating the interactions encoded by the aminoacid sequences via in silico approaches, and based on this analysis optimized sequences are designed. These sequences are then tested experimentally, thus providing feedback information to the model. This operation can be repeated in loop.

According to a first preferred embodiment the proteins of (b), taken individually, together or both individually and together, have the following composition by number of aminoacids: polar amino acids: in the range of 20-60%, preferably 25 and 50 %;

apolar amino acids: in the range of 20-60%, preferably 25 and 50 %;

negatively charged amino acids: in the range of 3-30%, preferably 5 and 20 %;

positively charged amino acids: in the range of 3-30%, preferably 5 and 20%;

hydrophobic amino acids: in the range of 5-30%, preferably 10 and 25 %;

aromatic amino acids: in the range of 3-45%, preferably 5 and 40 %;

glycince/proline: in the range of 5-40%, preferably 10 and 30 %; wherein the polar amino acids are selected from the group consisting of S, T, N, Q, R, H, K, D, E, C and combinations thereof;

wherein the apolar amino acids are selected from the group consisting of A, U, L, M, F, W, Y, V, G, P and combinations thereof;

wherein the negatively charged amino acids are selected from the group consisting of C, D, E and combinations thereof;

wherein the positively charged amino acids are selected from the group consisting of R, H, K, Y and combinations thereof;

wherein the hydrophobic amino acids are selected from the group consisting of W, Y, K, D, E and combinations thereof;

wherein the aromatic amino acids are selected from the group consisting of F, Y W, H and wherein glycince/proline are selected from the group consisting of G, P.

The proteins of (b), can be selected from natural low complexity domain stretches. For example they can be selected from LCD stretches from DEAD-box proteins, in particular of Dhhl , Dbpl , Lafl , Ddx4, hNRNPAl, FUS, hnRNPNA2 (see Seq-ID 7-15). The sequences can be partially modified by engineering, to optimise them or to avoid undesired interactions with particular molecules of interest or surroundings. Typically these modified systems are polypeptides comprising an amino acid sequence characterized by at least (>)85%, >87.5%, >90%, >92%, >94%, >95%, 96%, >97%, >98%, >99% sequence identity and/or amino-acid composition identity to these naturally occurring LCD stretches. Also possible are variants of these naturally occurring LCD stretches in which not more than 5, preferably not more than 3, or 2 or 1 amino acid is removed from the naturally occurring sequences at one or both ends (N and/or C terminus) thereof.

The proteins of (b) can for example be selected for the N-terminus of the molecule of interest from SEQ-ID 7 and SEQ-ID 9 and for the C-Terminus of the molecule of interest from SEQ- ID 8 and SEQ-ID 10 or, again for example after optimisation, from polypeptides comprising an amino acid sequence characterized by at least (>)85%, >87.5%, >90%, >92%, >94%, >95%, 96%, >97%, >98%, >99% sequence identity and/or amino-acid composition identity to these LCD stretches. Again also possible are variants of these naturally occurring LCD stretches in which not more than 5, preferably not more than 3, or 2 or 1 amino acid is removed from the naturally occurring sequences at one or both ends thereof.

Such a protein may further comprise a purification tag such as His Tag, glutathione transferase Tag, Strep tag, flag tag etc. - attached either to the N- or C- terminus of the protein of (b). Also, such a protein may comprise a cleavage tag.

The biologically/chemically active molecule of interest (a) of the proposed protein can be selected from the group consisting of enzymes, receptors, toxins, antibodies, collagenes, ion channels, transport proteins, hormones, fluorescent proteins and other non-biological objects such as metal-organic species, organic pharmaceuticals, nanoparticles .

The biologically/chemically active molecule of interest, or the mentioned derivative thereof, can also be a post-translationally modified protein, in particular a glycosylated, phosphorylated, acetylated, acylated or sulphated protein.

The protein typically exhibits phase separation when exposed to a temperature that is (i) above a lower critical solution temperature of the protein, and/or (ii) below an upper critical solution temperature of the protein.

Preferably the protein exhibits heat-irreversible phase separation when exposed to a temperature that is above a lower critical solution temperature of the protein, and exhibits reversible phase separation when exposed to a temperature below the upper critical solution temperature.

For example it is possible that the protein exhibits a reversible phase separation in response to a first stimulus and an irreversible phase separation in response to a second different stimulus.

Furthermore the present invention relates to the use, preferably the in vitro use, of a protein as detailed above various purposes, so for example for liquid-liquid phase separation, in particular for the purification of biologically/chemically active molecules of interest, for the preparation of water-in-water emulsions, including those with recruitment of further components in the disperse phase, for surface coating, for the provision of enzymes in organic-like phases.

Also possible is the use, preferably the in vitro use, for the tailored mixing of different biologically/chemically active molecules of interest in liquid-liquid phase separated globular structures. This can be achieved either by attaching the same proteins exhibiting a phase transition to the different biologically/chemically active molecules to mix them within the same globular structures, or by attaching different proteins exhibiting a phase transition to the different biologically/chemically active molecules to provide for distinct globular structures with either of the biologically/chemically active molecule.

Furthermore the present invention proposes the use, preferably the in vitro use, of a protein as described above for the generation of particles and/or gels, for providing multifunctional nanomaterials in biotechnology; for targeted drug delivery; for controlled release depots of pharmaceuticals; immobilized enzymes; for tissue engineering scaffolds; for components for biosensing and bioanalysis.

Furthermore, the present invention proposes a method for the generation of stable reversible globular structures in a liquid/liquid phase separated system, said globular structures comprising at least one protein as described above or consisting of at least one or more of the proteins as described above, wherein a protein as described above is immersed in water, and wherein preferably the size of the globular structure is adapted by tuning the pH, the ionic strength, the temperature, the shear, or a combination thereof, to show a number average globular structure diameter in the range of 1-100 pm, preferably in the range of 5- 50 pm.

Also the present invention relates to a method for the generation of stable essentially irreversible globular or fibrillar structures comprising at least one protein as described above or consisting of at least one or more of the proteins as described above, wherein a protein as described above is immersed in water, and wherein

in a first step the proteins are allowed to form liquid droplets in the water environment, wherein preferably the size of the globular or fibrillar structure is adapted by tuning the pH, the ionic strength, the temperature, the shear, or a combination thereof, more preferably to show a number average globular structure diameter in the range of at least 200 nm, or in the range of 1-100 pm, preferably in the range of 5-50 pm and

wherein in a second step the globular or fibrillar structures are incubated for at least an hour, optionally followed by washing, to form said stable globular or fibrillar structures, further optionally followed by recruitment of further components into the globular or fibrillar structure.

According to a further aspect of the present invention it relates to globular or fibrillar structure comprising at least one protein as described above or consisting of at least one or more of the proteins as described above, preferably obtained or obtainable using a method as detailed above, wherein it is preferably a porous globular structure, more preferably recruited with further components for functional use thereof.

The globular or fibrillar structure can preferably be embedded in a cross-linked water- soluble polymer structure, preferably a hydrogel, in particular

for slowing down release rates of the fusion system and/or of further components compared to the free fusion system and/or of further component, and/or for switching of rate limiting steps from diffusion of the free fusion system to dissolution rate of the droplets, and/or

for controllable release rates tailored for the respective application, and/or

to increase protein concentration.

According to yet another aspect of the present invention it relates to a polynucleotide comprising a gene or cell line expressing a protein as detailed above.

According to a further aspect of the present invention it relates to a pharmaceutical composition comprising a preferably non-immunogenic protein as described above or a globular or fibrillar structure as detailed above and optionally a pharmaceutically acceptable carrier.

Also the present invention relates to a method of producing one or more proteins as given above comprising:

(a) transforming a host cell with an expression vector comprising a polynucleotide comprising a nucleotide sequence encoding a as given above; and

(b) causing the host cell to express the protein.

Further embodiments of the invention are laid down in the dependent claims.

BRIEF DESCRIPTION OF THE DRAWINGS

Preferred embodiments of the invention are described in the following with reference to the drawings, which are for the purpose of illustrating the present preferred embodiments of the invention and not for the purpose of limiting the same. In the drawings,

Fig. 1 shows the conjugation of soluble proteins with LCD sequences (top) and enzymatic activity of solubilized AK chimera proteins (bottom);

Fig. 2 shows the amino acid distributions of the considered systems;

Fig. 3 shows the Liquid-liquid phase separation of chimera proteins, in a) the determination of upper critical solution temperature (UPST) for DbplN-AK- DbplC (AK-LCD2), in b) the determination of saturation concentration for DbplN-AK-DbplC, in c) the reversibility of phase transition of DbplN-AK- DbplN and in d) the coalescence of liquid-liquid phase separated DbplN- AK-DbplN droplets;

Fig. 4 shows the trigger possibilities for phase separation as a function of pH and ionic strength, x stands for no droplet formation, o stands for droplets;

Fig 5 shows the Liquid-Liquid Phase separation of chimera proteins - Variability in interactions;

Fig. 6 shows how to control of size distribution of liquid-liquid phase separated droplets;

Fig. 7 shows the maturation of liquid droplets into different morphologies, in a)

Bright field microscopy image of DbplN-AK-DbplC droplets after 15 h ripening in 50 mM Tris, 100 mM KC1 and pH 7.5, in b) Bright field microscopy image of DbplN-AK-DbplC droplets after 15 h ripening in 50 mM Tris, 100 mM NaCl and pH 8.0, in c) Bright field microscopy image of DhhlN-AK-DhhlC (AK-LCD1) droplets after 15 h ripening in 50 mM Citrate Buffer, 100 mM and pH 5.0;

Fig. 8 shows the maturation of liquid droplets into different morphologies -

Handability, in a) ThT stained confocal image of DbplN-AK-DbplC droplets formed under 100 rpm shaking over night, subsequent pipetting and settling of 2 days, top down view, side-length of box ~450 pm, in b) Same condition as a) Epi-fluorescence image in wide focal microscope; c) Same condition as a) bottom up, zoomed in view; Bright field microscopy image of DbplN-AK- DbplC solidified droplets after lOx washing;

Fig. 9 shows multiple functional protein assemblies - Recruitment and partitioning, in a) Firefly nanparticles (*a 200 nm) recruited into droplets of DbplN-AK- DbplC, in b) GFP recruited into droplets of DbplN-AK-DbplN, in c) DCVJ recruited into droplets of DbplN-AK-DbplN, in d) Nile Red recruited into droplets of DbplN-AK-DbplN;

Fig. 10 shows how LCDs on their own are capable of material formation, wherein left: DbplN liquid-liquid phase separated droplets at pH 8.0 and 33 mM Tris; right: Saturation concentration estimation of DbplN at pH8.5 and 62 mM Tris;

Fig. 1 1 shows the synthesis and characterization of active chimera proteins, in A)

Illustration of the chimera proteins developed in this work in which globular proteins are conjugated with low complexity domains which act as molecular adhesives. The enzyme adenylate kinase (AK) and the green fluorescent protein (GFP) have been conjugated with low complexity domains (LCD) encoding self-assembling behaviour (see Table Sl). B) Amino acid composition and main physiochemical properties of the LCDs including the fraction of charged and non-polar residues (FCR and FNPR, respectively) pi is the isoelectric point of the proteins. (C-D) Characterization of the molecular weight of the expressed chimera proteins by size exclusion chromatography coupled with multi-angle light scattering: (C) representative chromatograms of AK (black line) and AK conjugated with LCDs originating from Dhhl (blue line); (D) Theoretical versus measured molecular weights of AK (·), AK conjugated with LCDs originating from elastin (·), Dhhl (¨) and Dpbl (_■), and GFP conjugated with LCDs originating from Dhhl (A) and Dpbl (▼). 95% two-sided confidence intervals are smaller than the markers. (E) Activity of AK and AK chimera proteins evaluated with a fluorometric adenylate kinase activity assay. The conjugation of molecular adhesives has no significant effect on the enzyme activity;

Fig. 12 shows how molecular adhesives encode specific self-assembly of the chimera proteins. Phase diagrams of AK conjugated with LCDs originating from Dhhl (LCD1) (A) and Dpbl (LCD2) (B), and of GFP conjugated with LCDs originating from Dhhl (C) and Dpbl (D). Circles and red crosses indicate the presence and the absence of phase separation, respectively. The analysis has been performed at room temperature. Protein concentrations were: AK- LCD1, 25 mM; AK-LCD2, 20 mM; GFP-LCD1, 25 mM; GFP-LCD2, 5 mM. Representative microscope images of solutions of AK (E), AK conjugated with LCD1 (F, pH 5.0, no salt), and conjugated with LCD2 (G, pH 8.5, no salt);

Fig. 13 shows the maturation of reversible protein droplets into aggregates. A)

Coalescence of the liquid protein-rich droplets of 20 mM AK-LCD2 in 50 mM Tris buffer at pH 8.0. B) Reversibility of the phase separation of 20 mM AK- LCD2 in 50 mM Tris buffer at pH 8.5. The ionic strength was initially decreased from 500 mM to 30 mM and then increased from 30 mM to 500 mM. C) During incubation time, the droplets shown in panel A) evolve into aggregates that are irreversible and maintain a spherical shape;

Fig. 14 shows how protein particles are active porous structures and release soluble proteins over time. A) Bright field (left) and fluorescence microscopy (right) images of liquid droplets and solid particles of the chimera protein AK-LCD2. The formation of the product of the enzymatic reaction is monitored by recording fluorescence emission at 587 nm after excitation at 535 nm. B-C) Diffusion of the fluorescent probe Thioflavin T (ThT) into and out of the solid aggregates of the chimera protein AK-LCD2. (B) Increase of the fluorescent intensity inside the aggregates over 40 minutes after introduction of ThT into the solution. From bottom to top, the curves represent data acquired at 0, 5, 10 and 20 minutes after the addition of ThT. (C) Decrease of the fluorescence intensity inside protein aggregates that have been pre-equilibrated for two hours in a ThT solution and then introduced into a ThT-free buffer. From top to bottom, curves represent data acquired at 0, 10, 20, 40, 60 and 120 minutes after buffer exchange. D) Enzymatic activity of the supernatant from a solution containing particles formed of protein AK-LCD1 over time. Error bars indicate the two-sided 95% confidences interval for the measured rate. Two independent sets of data are indicated in blue and green colours. In all the experiments AK-LCD2 particles have been generated from a 20 mM protein solution at pH 8.5 containing no salt. E) Schematic drawing of the two-step mechanism leading to the formation of protein aggregates from monomeric solutions via the formation of dense protein liquid state.

Fig. 15 shows multifunctional materials comprised of multiple chimeric proteins. A)

Microscope images of droplets formed from 20 mM AK-LCD2 (left) and 8 mM GFP-LCD2 (right) solutions in 50 mM Tris buffer at pH 8.5; B) Bright field (left) and epi-fluorescence (right) images of the mixture of the two solutions in A), showing the incorporation of the fluorescent GFP-LCD2 droplets into the larger AK-LCD2 droplets; C) After incorporation, the GFP- LCD2 droplets merge into the AK-LCD2 structures and lose their identity, thereby generating a homogeneous phase; D) Time evolution of 3-D fluorescence profiles of GFP-LCD2 chimera proteins diffusing into the AK- LCD2 matrix. The profiles have been extracted from images acquired at 0, 10 and 20 seconds after incorporation of the GFP-LCD2 droplet into the AK- LCD2 structures. The right-most plot shows the decrease of the fluorescence intensity along the cross section of the GFP-LCD2 droplet at 5, 10, 15, 20 and 25 seconds after the incorporation of one GFP-LCD2 droplet into one AK- LCD2 droplet. The dotted lines represent simulations based on Fick’s diffusion law. E) Fluorescent images showing the activity of the bifunctional AK-LCD2/GFP-LCD2 structures: fluorescence emission of GFP (left) and of the product of the enzymatic reactions catalysed by AK (right). F) Bright field (left) and epi-fluorescence (right) images of the mixture of 25 mM AK-LCD1 and 8 mM GFP-LCD2 show the absence of incorporation of the fluorescent GFP-LCD2 droplets into the AK-LCD1 droplets in in 50 mM citrate buffer at pH 6.0 and 10% PEG.

Fig. 16 shows the hydropathy of the LCD sequences calculated according to the Kyte and Dolittle scale. N and C denote the N- and C- termini of the LCD sequences, respectively. The abscissa represents the number of amino acids counted from the N-terminus.

Fig. 17 shows SDS-gel analysis of the chimera proteins. The 4-12% Bis-Tris gels

(Invitrogen) were run at 200V, l50mA for 35 min. The left lane contains the Markl2 (Invitrogen) marker and the right lane contains the respective protein band.

Fig. 18 shows SEC-MALS analysis of the chimera proteins. The SEC chromatograms measured by UV absorbance (left axis) and the molecular weight calculated by multi angle light scattering (MALS) (right axis) for all the chimera proteins investigated in this work.

Fig. 19 shows an Activity assay of AK and AK chimera proteins. We evaluated the activity of AK and AK chimera proteins by measuring an enzymatic reaction that converts the substrate adenosine diphosphate into the fluorescent product of the assay. We monitored the generation of the product over time by recording the fluorescence intensity at different enzyme amounts. The steepest slope of these curves represents the maximum enzymatic rate (vmax) which is shown in Figure 1 1E in the main text.

Fig. 20 shows the solubility of globular proteins. Phase diagrams of AK (A), GFP (B) and AK conjugated with LCDs originating from elastin (C). Red crosses indicate the absence of phase separation. The analysis has been performed at room temperature. Protein concentrations were: AK, 40 mM; GFP, 25 mM; AK-elastin, 25 mM.

Fig. 21 shows the Amyloidogenic propensity of the individual LCD sequences. The amyloidogenicity was calculated along the N and C-terminal LCD sequences of the investigated constructs via the packages TANGO and AmylPred2. Typical values of aggregation propensity of aggregation prone regions in A- b 42 are between 80 to 98 using TANGO. The abscissa represents the amino acid count starting from the N-terminus of the respective sequence.

Fig. 22 shows activity measurements by epi-fluorescent microscopy. The activity of the phase separated droplets and protein aggregates was assessed by performing the activity assay (See Materials and Methods) on an epi- fluorescent microscope. A-B) Liquid droplets formed in a solution of 20 mM AK-LCD2 in 50 mM Tris at pH 8.5 with 0 mM NaCl with (A) and without (B) the addition of the enzymatic reaction mixture. C-D) Protein aggregates formed in a solution of 20 mM AK-LCD2 in 50 mM Tris at pH 8.5 with 0 mM NaCl were washed ten times and imaged with (C) and without (D) the addition of the enzymatic reaction mixture.

Fig. 23 shows the activity of the aggregates of the AK chimera proteins. The activity of AK-LCD1 and AK-LCD2 arrested structures were measured after three washes of the supernatant. Blue circles and squares denote independent measurements. Black stars represent the measurements of control solutions without the substrate. Error bars indicate the 95% confidence intervals of the maximum slope. The different conditions are: AK-LCD1 Cl : 8 pM protein at pH 5.0 and 100 mM MgC12; AK-LCD1 C2: 4 pM protein in the same solution of Cl . AK-LCD2 Cl : 20 pM protein at pH 7.5 and 100 mM KC1; C2: 20 pM protein at pH 7.5 and 100 mM NaCl; C3: 20 pM protein at pH 8.5 and no salt; C4: 20 pM protein at pH 9.0 and no salt.

Fig. 24 shows the removal of soluble proteins from the supernatant. Before analysing the release of soluble proteins over time, the supernatant of the protein particles was exchanged multiple times with the buffer required by the activity assay. A) We verified that this operation did not destroy the structure of the protein-rich phase by bright field microscopy after several washing steps. Bright field image of a AK-LCD2 droplets at pH 8.0 and 50 mM Tris after ten washing steps. B) Measurements of the residual activity of the supernatant indicate that ten washes are sufficient to eliminate the soluble proteins.

Fig. 25 shows that the ability to self-assemble of the LCDs is transferred to magnetic nanoparticles. A) Schematic illustration of the chimera protein used, consisting of a globular enzyme (AK) conjugated with two low complexity domains derived from the biological protein LAF-l. B) Representative brightfield microscopy images of the droplets formed by chimera proteins after lh incubation at 10 mM NaCl concentration and at pH = 7.5. Scale bar is 30 pm. The concentration of protein was 0.5 pM. C) Schematic representation of the nanoparticle functionalized with the LAF1-AK-LAF1 chimera protein via non-covalent interactions between the His-Tag sequence of the LAF1-AK-LAF1 and the anti His-Tag antibody attached to the surface of the magnetic NP. The conjugation with the chimera protein confers to the magnetic nanoparticle the ability to phase separate. D) Size distribution evaluated via dynamic light scattering and TEM image (insert) of the NP composite clusters formed after lh incubation at 10 mM NaCl concentration and pH=7.5. The protein/NP molar ratio was 3000, and the concentrations of LAF1-AK-LAF1 and NPs were 0.5 pM and 0.16 nM, respectively.The chimera protein and the protein/NP composite phase separate at the same protein concentration, pH, and ionic strength.

Fig. 26 shows that the self-assembly behavior of the LCDs is transferred to magnetic nanoparticles. A-B) Representative brightfield microscopy images of solutions of the functionalized nanoparticles at 500 (A) and 10 mM (B) NaCl concentration. Scale bar is 30 pm. C-D) Distribution of hydrodynamic diameters of solutions of the functionalized nanoparticles at 500 (C) and 10 mM (D) NaCl concentration with a protein/NP molar ratio of 3000 after lh incubation. 0.5 pM LAF1-AK-LAF1 and 0.16 nM NPs at pH=7.5. E-F) TEM images of the magnetic nanoparticles (E) and of the protein/NP clusters (F) at 10 mM NaCl concentration, respectively. Scale bar is 200 nm (E) and 500 nm (F). Experiments were performed at a protein/NP molar ratio of 3000 after 1 h incubation. 0.5 pM LAF1-AK-LAF1 and 0.16 nM NPs at pH=7.5.

Fig. 27 shows the deletion of the LCD domains suppressing the propensity of the functionalized nanoparticles to self-assemble. A) Representative brightfield microscopy image of solutions of the magnetic nanoparticles functionalized with the AK lacking LCD domains at 10 mM NaCl concentration with a protein/NP molar ratio of 3000 after lh incubation. 0.5 pM AK and 0.16 nM NPs at pH=7.5. Scale bar is 30 pm. B) Time evolution of the average hydrodynamic diameter of the same solution of functionalized NPs. The functionalized NPs lacking LCDs are stable for several days and do not phase separate.

Fig. 28 shows the functionalized nanoparticles do not aggregate at high ionic strength for several days. Time evolution of the average hydrodynamic diameter of the functionalized nanoparticles at 500 mM NaCl concentration with a protein/NP molar ratio ~ 3000. 0.5 mM LAF1-AK-LAF1 and 0.16 nM NPs in 0.5 mM Tris buffer at pH=7.5.

Fig. 29 shows the self-assembly of the protein/NP composite at different ionic strength and pH values. A) (top) Brightfield microscopy images of solutions of 0.5 mM LAF1-AK-LAF1 at pH = 7.5 at increasing ionic strength. Scale bar is 10 pm. (bottom) Average hydrodynamic diameter of the clusters formed by the phase separation of the protein-NP composites with a protein/NP molar ratio of 3000 as a function of ionic strength under the same conditions of the top panel. B) Brightfield microscopy images of solutions of 0.5 pM LAF1- AK-LAF1 (top) and protein/NP composites (bottom) with a protein/NP molar ratio of at 10 mM NaCl concentration at pH of 4, 7.5 and 10 (from left to right). Scale bar is 10 pm. Red cross and black circle indicate absence and presence of phase separation, respectively.

Fig. 30 shows the pH clock of the protein/NP composites. Time evolution of the pH value (top) and of the average hydrodynamic diameter (bottom) of a solution of protein/NP composite. 0.5 pM LAF1-AK-LAF1 and 0.16 nM NPs 10 mM NaCl concentration. The increase and decrease of the average diameter over time indicates the reversible assembly and disassembly of the protein/NPs into clusters.

Fig. 31 shows the ionic strength clock. A) Representative brightfield microscopy images of the NP/protein composite at protein/NP molar ratio of 3000 before (left) and after (right) addition of 5 M NaCl. 1 pM LAF1-AK-LAF1 and 0.16 nM NPs at pH=7.5. Scale bar is 30 pm. B) Time evolution of the average hydrodynamic diameter of the same functionalized nanoparticles before and after 5 M NaCl addition.

Fig. 32 shows the morphology control over the protein/NP composite. A) Schematic representation of the solid linear fibrils obtained via applying a magnetic field to the dispersion of functionalized nanoparticles with a protein/NP molar ratio of 3000 and below corresponding representative bright field microscopy image of the solid-like linear fibrils. B) Schematic representation of the droplet chains obtained via applying a magnetic field to the dispersion of functionalized nanoparticles with a protein/NP molar ratio of 12000 and below corresponding representative bright field microscopy image of the droplet chains. 0.5 mM (A) or 2 mM (B) LAF1 -AK-LAF1 was mixed with 0.16 nM NPs at 10 mM NaCl and pH=7.5. Scale bar is 30 mih.

Fig. 33 shows the pH clock of the fibrils and droplet chains. A) Representative brightfield microscopy images of the NP/protein composite at protein/NP molar ratio of 3000 at increasing pH (from left to right) under magnetic field. 0.5 mM LAF1-AK-LAF1 and 0.16 nM NPs at 10 mM NaCl concentration. Scale bar is 30 pm. B) Representative brightfield microscopy images of the NP/protein composite at protein/NP molar ratio of 12000 at increasing pH (from left to right) under magnetic field. 2 mM LAF1-AK-LAF1 and 0.16 nM NPs at 10 mM NaCl concentration. Scale bar is 30 pm. The pH was changed via titration with a 10 mM NaOH solution or 10 mM HC1 solutions starting from pH=7.5.

Fig. 34 shows the ionic strength clock of the fibrils and droplet chains. A)

Representative brightfield microscopy images of the NP/protein composite at protein/NP molar ratio of 3000 before (left) and after (right) addition of 5M NaCl. 0.5 mM LAF1-AK-LAF1 and 0.16 nM NPs at pH=7.5. Scale bar is 30 pm. B) Representative brightfield microscopy images of the NP/protein composite at protein/NP molar ratio of 12000 at 10 mM (left) and 500 mM (right) NaCl concentration under magnetic field. 2 pM LAF1-AK-LAF1 and 0.16 nM NPs at pH=7.5. Scale bar is 30 pm.

Fig. 35 shows the behaviour of the bare nanoparticles under a magnetic field A)

Representative brightfield microscopy image of 0.16 nM magnetic NPs solution after lh incubation at pH=7.5 and 10 mM NaCl under a magnetic field. Scale bar is 30 pm. B) Time evolution of the average hydrodynamic diameter of the same solution of bare NPs kept constantly under the same magnetic field. The NPs alone are not able to phase separate

Fig. 36 shows multifunctional microreactors. A) Schematic illustration of the enzymatic reaction mediated by the AK globular domain that occurs into the synthetic compartments. B-C) Representative bright field and fluorescence microscopy images of the droplet chains generated at a protein/NP molar ratio of 12000, with 2 mM LAF1-AK-LAF1 and 0.16 nM NPs at 10 mM NaCl and pH=7.5. The fluorescence signal corresponds to the product of the enzymatic reaction, stained by a commercial kit. Scale bar is 30 pm. D) Schematic illustration of the cascade reaction catalyzed by the iron nanoparticles that it is expected to occur into the synthetic compartments. This reaction can be applied for glucose sensing.

DESCRIPTION OF PREFERRED EMBODIMENTS

Fig. 1 shows the general scheme how the globular soluble proteins, in this particular case adenylate kinase (AK) and green fluorescent protein (GFP) are modified by attaching corresponding low complexity domain (LCD) sequences to the N- and C-terminus, respectively. Using recombinant expression the corresponding AK/GFP-LCD chimera proteins are generated, forming multifunctional particles.

As illustrated in the lower portion of Fig. 1, and as will be detailed further below, those chimera systems retain their enzymatic activity if solubilised. So basically the attachment of the LCD sequences to both terminal ends does not alter the activity of the central molecular structure of interest, however provides for the possibility of tailored liquid/liquid phase separation and further aspects detailed below. Notably for the AK-elastin hybrid, as will be detailed further below, the enzymatic activity is there however no liquid/liquid phase separation is available. The same of course holds true for the AK alone.

The LCD sequences which are used in this experimental section, are derived from naturally occurring systems comprising these low complexity domains. In particular the present LCD domains are based on DEAD-box proteins derived from yeast (Dbpl, Dhhl) from C. elegans (Laf-l) and from human sources (Ddx4).

Fig. 2 shows that these low complexity domains show a very particular and characteristic amino acid distribution, which, without being bound to any theoretical explanation at this stage, seems to be largely responsible for the ability thereof to form liquid/liquid phase separation states under corresponding conditions. In the naturally occurring membraneless compartments formed by the proteins containing these LCD terminal ends it was found that these compartments are liquid -like, have a high concentration of proteins and chemicals, the formation is governed by physical processes, and is dependent on physical-chemical properties of the respective proteins.

The main characteristics of the liquid/liquid droplets structure as available for the AK- chimera systems are illustrated in Fig. 3. As given by a) the average radius of the corresponding droplets is increasing as a function of temperature until reaching an upper critical solution temperature (UPST), in this case at around 47°C. Above that temperature no droplets are observed anymore. The corresponding values depend on various parameters such as pH, ionic strength, et cetera. In b) the linear behaviour of the saturation concentration is illustrated, there seems to be no upper limit. In c) the reversibility of the droplet formation is shown, in this case not as a function of temperature but as a function of ionic strength, by cycling from a high ionic strength (left) without droplets to a lower ionic strength (middle) showing droplets, and then increasing the ionic strength again (right) destroying the droplet structure. Last but not least in d) it is illustrated how the liquid droplets dynamically coalesce over time.

As mentioned above, tailored phase separation is available for the proposed systems. The influencing factors are temperature, pH, and ionic strength. The latter two are illustrated in Fig. 4: it is shown how different LCDs encode different amounts of different interactions and hence the proposed systems allow tuning the phase diagram. It seems that unexpectedly the liquid/liquid phase separation behaviour is essentially exclusively determined by the LCD stretches, and the functionality of the central molecule of interest is unaffected by the LCD stretches and vice versa. Furthermore the graphics in Fig. 4 show that there is a strong tunability via length at constant interaction density, and that there is a strong dependence on accessibility.

Fig. 5 shows the variability in the interaction in the liquid-liquid phase separation of the chimera proteins. In the top row on the left side droplet system LCD2-AK is shown and on the right side droplet system LCD2-GFP is shown. In the middle row one can see that if these two liquid/liquid dispersions are mixed, globular structures are formed which contain both, since the LCD structures are compatible. In the lower row a mixture of LCD2-GFP and LCD1-AK is shown. In this case due to the incompatibility of the LCD tails in many cases no common globular structures are formed, by contrast, individual and clearly separate globular structures are formed, one group with exclusively with LCD2-GFP and another one with exclusively with LCD1-AK. So as one can see, different LCDs encode different amounts of different interactions and hence allow for tuning the phase diagram. Furthermore different chimera proteins with sufficiently different LCDs do not merge, while different chimera proteins with the same LCDs do merge.

Fig 6 shows how the size distribution of the liquid-liquid phase separated droplets can be controlled. The phase separated droplet size can be tuned with shear, via pumping the solution back and forth through alternating smaller and bigger channels.

Fig. 7 illustrates how the liquid droplets can be matured into different morphologies. This is a behaviour which is not observed in the naturally occurring systems comprising the corresponding LCD tails. The corresponding matured globular structures are stable and the maturation is irreversible. It seems the maturing process is associated with an increase of order and a decrease of hydration. The phase separated protein rich droplets are capable of maturing into solidified morphologies, for example fibrils, solidified droplets and heavily viscous droplets.

The maturation of liquid droplets into different morphologies as a function of pH, salt and concentration is summarised in the following table:

As for the porosity of the liquid droplets obtained after maturation reference is made to the discussion of figure 14 below, diffusion of the thioflavin T into and out of the solidified droplets shows the porosity and accessibility of the droplets for small chemicals.

As for the handability of the liquid droplets after maturation Fig. 8 shows that the solidified droplets can be washed at least 10 x with a different buffer, without destroying the architecture. Shaking over night prevents the sticking to the well glass bottom. After shaking droplets can be pipetted into another well.

As for the activity of the matured droplets reference is made to figure 14A and the corresponding discussion below as well as to Fig. 23 detailed below. Liquid - liquid phase separated droplets are the major source of AK enzymatic activity determined by the fluorometric assay observed via epi-fluorescent microscopy. Matured and 10 x washed droplets still exhibit activity as determined by epi-fluorescent microscopy and matured droplet activity is also observed via fluorescence emission spectroscopy.

As for the procedure to obtain multiple functional protein assemblies reference is made to Fig. 15 and the corresponding detailed discussion below. Chimera proteins with different globular domains but same LCDs are forming homogeneously dispersed liquid liquid phase separated droplets. GFP and AK are used as model globular domains, but in theory any protein is possible. The homogeneity is conserved over the course of solidification.

Activity of GFP and AK is preserved in solidified mixed matured architectures.

Fig. 9 illustrates recruitment and partitioning in multiple functional protein assemblies. Liquid - liquid phase separated chimera proteins are capable of recruiting non-LCD components such as other proteins (GFP); hydrophobic dyes (DCVJ, Nile Red, Rhodamine); nanoparticles (PS). Recruitment is possible via non-covalent interactions including hydrophobicity.

The release of active enzymatic species over time & via mechanical washing shows the following: Incubation of solidified protein droplets in a different buffer releases enzymatically active species over time. No initial burst is visible, but slow release over time. Repeated mechanical washing with a different buffer can release active species into the washing supernatant.

Fig. 10 shows that the LCDs on their own are also capable of material formation. This can be exploited as material/matrix for other substances.

To summarise this first part the following:

Conjugation of soluble proteins with low complexity sequences (LCDs) at both the N and C -terminus is possible. LCDs are derived from proteins discovered in living organisms. The LCDs trigger the formation of protein-rich droplets via liquid-liquid phase transition at suitable conditions of salt and pH. The liquid droplets are converted into solid aggregates via incubation at room temperature over several hours. The solid particles are active and porous structures, and release active monomers over time. This simple strategy is particularly attractive to assemble different functional proteins within the same architecture, and opens up many applications in pharmaceutical, food and enzyme industry. Synthesis, characterization and activity of the chimera proteins

We considered the green fluorescence protein (GFP) and the enzyme adenylate kinase (AK) as model globular proteins. We conjugated these two proteins at both the N- and the C- termini with different low complexity domains (LCDs), each with different compositions and physicochemical properties (Fig. 1 1 A). Indicated as LCD1 and LCD2, these molecular adhesives are derived from the sequences of the DEAD-box proteins Dhhl and Dpbl , respectively. In their native biological context, Dhhl and Dpbl have been shown to undergo liquid-liquid phase separation (LLPS) associated with the formation of processing bodies (P- bodies) in yeast.

The full sequences of the expressed chimera proteins are summarized in Table 1 and the main physiochemical properties, such as hydropathy, amino acid content, fraction of charged and non-polar residues, are shown in Fig. 1 1B and Fig. 16.

Table 1 : Amino acid sequences of the protein variants analysed in this work bold, italics underlined , underlined and normal indicate the His-tag, the AK protein, the elastin LCD and the GFP protein, respectively. LCD1 and LCD2 are represented with bold underlined italics.

The N terminal LCD1 (DhhlN) chain is thus given by:

MGSINNNFNTNNNSNTDLDRDWKTALNIPKKDTRPQTDDVLNTKGNT (SEQ-ID 7) The C-terminal LCD1 (DhhlC) chain is given by:

VPVPFPIEQQSYHQQAIPQQQLPSQQQFAIPPQQHHPQFMVPPSHQQQQAYPPPQM PSQQGYPPQQEHFMAMPPGQSQPQY (SEQ-ID 8)

The N terminal LCD2 (DbplN) chain is thus given by:

MADLPQKVSNLSINNKENGGGGGKSSYVPPHLRSRGKPSFERRSPKQKDKVTGGD FFRRAGRQTGNNGGFFGFSKERNGGTSANYNRRGSSNYKSSGNRWVNGKHIPGP KNAKLQKAELFGVHDDPDYHSSGIKFDNYDNIPVDASGKDVPEP1L (SEQ-ID 9)

The C-terminal LCD2 (DbplC) chain is given by:

GGRTRGGGGFFNSRNNGSRD YRKHGGNGSFGSTRPRNT GTSN WGSIG GGFRN DN EKNGYGNSNASWW (SEQ-ID 10)

Further possible LCD systems are as follows:

Lafl from C.elegans (DDX3 RNA helicase, 10.1073/pnas.15048221 12, AA 1 -168) is given by:

MESNQSNNGGSGNAALNRGGRYVPPHLRGGDGGAAAAASAGGDDRRGGAGGG GYRRGGGNSGGGGGGGYDRGYNDNRDDRDNRGGSGGYGRDRNYEDRGYNGGG GGGGNRGYNNNRGGGGGGYNRQDRGDGGSSNFSRGGYNNRDEGSDNRGSGRSY NNDRRDNGGDG (Seq-ID-l 1)

Ddx4 (human, DEADbox RNA helicase, l0.1016/j.tnolcel.2015.0l .013, AA 1-236) is given by:

MGDEDWEAEINPHMSSYVPIFEKDRYSGENGDNFNRTPASSSEMDDGPSRRDHFM KSGFASGRNFGNRDAGECNKRDNTSTMGGFGVGKSFGNRGFSNSRFEDGDSSGF WRESSNDCEDNPTRNRGFSKRGGYRDGNNSEASGPYRRGGRGSFRGCRGGFGLG SPNNDLDPDECMQRTGGLFGSRRPVLSGTGNGDTSQSRSGSGSERGGYKGLNEEV ITGSGKNSWKSEAEGGES (Seq-ID-l 2)

FUS (human, RNA-binding protein, 10.1016/j. cell.2015.07.047, AA 1-214) is given by: MASNDYTQQATQSYGAYPTQPGQGYSQQSSQPYGQQSYSGYSQSTDTSGYGQSS YSSYGQSQNTGYGTQSTPQGYGSTGGYGSSQSSQSSYGQQSSYPGYGQQPAPSSTS

GSYGSSSQSSSYGQPQSGSYSQQPSYGGQQQSYGQQQSYNPPQGYGQQNQYNSSS

GGGGGGGGGGNYGQDQSSMSSGGGSGGGYGNQDQSGGGGSGGYGQQDRG

(Seq-ID-l3)

hNRNPAl (human, Single-strand RNA-binding protein, 10.1016/j.cell.20l 5.09.015, AA 187-372) is given by:

ASASSSQRGRSGSGNFGGGRGGGFGGNDNFGRGGNFSGRGGFGGSRGGGGYGGS GDGYNGFGNDGGYGGGGPGYSGGSRGYGSGGQGYGNQGSGYGGSGSYDSYNN GGGGGFGGGSGSNFGGGGSYNDFGNYNNQSSNFGPMKGGNFGGRSSGPYGGGG QYFAKPRNQGGYGGSSSSSSYGSGRRF (Seq-ID 14)

hnRNPNA2 (human, Heterogeneous nuclear ribonucleoproteins, 10.1016/j.molcel.2017.12.022, AA 190-341) is given by:

GRGGNFGFGDSRGGGGNFGPGPGSNFRGGSDGYGSGRGFGDGYNGYGGGPGGG NFGGSPG Y GGGRGG Y GGGGPG Y GNQGGGY GGG YDN Y GGGN Y GSGN YNDFGN Y NQQPSNYGPMKSGNFGGSRNMGGPYGGGNYGPGGSGGSGGYGGRSRY (Seq-ID 15)

Both LCD1 and LCD2 contain a high fraction (40%) of non-polar residues, particularly glycine and proline, although several charged and polar residues are common throughout the sequences. In particular, both LCDs share polar residues such as asparagine and serine, and LCD2 contains a high number of the positively charged residues arginine and lysine.

We expressed the chimera proteins in E. coli and characterized their purity and molecular weight by SDS gel electrophoresis (Fig. 17) and size exclusion chromatography coupled with a multi-angle light scattering detector (MALS). A representative chromatogram and the measured molecular weights are shown in Fig. 11 C and 1 1D, respectively, and the full set of chromatograms is reported in Fig. 18. The experimental molecular weights measured by MALS are in excellent agreement with the theoretical values (Fig. 1 1D). Moreover, to verify that the activity of AK is not significantly altered by the conjugation with the LCDs we performed a fluorometric adenylate kinase assay which measures the conversion of the substrate adenosine diphosphate into the fluorescent product (Fig. 1 1E and Fig. 19).

Molecular adhesives encode specific self-assembling behaviour

We next characterized the phase behaviour of the expressed proteins by optical microscopy. We investigated phase separation at varying salt concentrations and pH values at room temperature (Fig. 12). As expected, the globular proteins AK and GFP without LCDs remained soluble under each condition (Fig. 20). In contrast, the fusion proteins containing the LCD1 and the LCD2 domains exhibited phase separation under varying conditions of pH and salt concentration (Fig. 12A-D).

Notably, the same globular protein conjugated with different LCDs showed a varying solubility as a function of ionic strength and pFI, while different proteins (GFP and AK) conjugated with the same LCD shared similar phase behaviours. These results not only demonstrate that LCDs can induce phase separation, but also that different LCDs can be used to tune specific interactions. Despite the dominant effect of LCD conjugation, different proteins (AK and GFP) conjugated with the same sequence exhibited subtle differences in the phase diagram (Fig.l2B and D), indicating that the globular domain can further modulate intermolecular interactions.

It is interesting to note that the critical pH value at which phase transition was observed correlates with the net charge of the chimera proteins (Fig. 2H), which, in turn, is strongly affected by the isoelectric point (pi) of the LCDs. LCD1 has a pi of 6.2, and the pi of the chimera proteins AK-LCD1 and GFP-LCD1 (6.3 and 6.1 , respectively) are close to the pi of the non-conjugated proteins AK and GFP (6.4 and 6.1 , respectively). LCD2 has a pi of 10.3, and the pi of the chimera proteins AK-LCD2 and GFP-LCD2 (9.6 and 9.3, respectively) is higher than the pi of AK and GFP. We con-elated the measured phase diagrams of the chimera proteins (Fig. 12A-D) with theoretical calculations of the protein net charge as a function of the pH (Fig. 12H). The comparison indicates that a net positive charge of around + 8e is required to induce phase separation. This net positive charge likely promotes the random coil conformations of the LCDs, leading to the exposure of multiple short linear motifs that are capable of inducing the attractive multidomain interactions required for phase separation. Moreover, this result indicates that this liquid-liquid phase separation is different from precipitation processes that are typically promoted by a decrease of a given protein’s net charge close to its pi.

To confirm the absence of hydrophobic effects we generated chimera proteins with a hydrophobic LCD derived from elastin, following a strategy that has been recently proposed in the literature (Fig. 1 ID). The composition of these sequences is markedly different from LCD1 and LCD2 and contains a high fraction of hydrophobic residues (see Fig. 16). The chimera proteins containing the LCD derived from elastin did not undergo phase separation under any investigated condition at room temperature, despite the high number of hydrophobic residues (Fig. 20). Moreover, we analysed the propensity of the molecular adhesives to form intermolecular b- sheet structures. In particular, we calculated the propensity of LCD 1 and LCD2 to convert into amyloid aggregates by using two common predictors of amyloid formation as a function of primary sequences, namely TANGO and AmylPred2. According to both algorithms, the sequences considered in this work have essentially no propensity to form b-sheet structures (Fig. 21).

Overall, these results show that the molecular adhesives are capable of inducing attractive protein-protein interactions that in turn promote phase separation and the formation of protein-rich droplets. Such attractive interactions involve non-polar, polar, and electrostatic forces, which can be tuned by pH and salt concentration, and are not based on a generic hydrophobic effect.

From reversible liquid droplets to irreversible aggregates

The phase separation of our chimera proteins generates protein-rich liquid droplets that are highly dynamic and undergo coalescence (Fig. 13 A). We tested the reversibility of this phase separation by varying salt concentration at pH 6.0 for AK-LCD1 and pH 8.5 for AK-LCD2. We observed that the phase transition can be induced at low salt concentration and can be reverted by increasing the ionic strength, thereby indicating that this process is reversible (Fig. 13B). Notably, during incubation at room temperature over several hours we observed the maturation of these liquid droplets into irreversible structures that do not coalesce and maintain the shape of the droplets (Fig. 13C).

Similar observations have been made with several other peptides and proteins which have been shown to form protein crystals and other supramolecular structures via generation of metastable clusters consisting of dense liquid. Examples include both short peptides such as diphenylalanine and globular proteins like sickle-cell haemoglobin. It is becoming increasingly evident that proteins which undergo liquid-liquid phase separation have a high propensity to form aggregates via this two-step nucleation. For instance, disordered regions of key ribonucleoprotein (RNP) granule components and the full-length granule protein hnRNPAl have been shown to produce dynamic liquid droplets which mature to more stable states and form fibrous structures. Moreover, single point mutations of the fusion sarcoma protein (FUS) induce the maturation of dense liquid droplets into amyloid structures which have been associated with amyotrophic lateral sclerosis (ALS) disease. Our results, together with these previous observations, indicate that molecular adhesives based on LCDs have the generic potential to trigger controlled phase transition of soluble proteins into crystal and solid protein particles via formation of a dense liquid phase that matures over time. This understanding creates an attractive possibility to control the composition, size, and biophysical properties of the final structures by modulating the self-assembling process during early stages, when the protein-rich phase is dynamic.

Protein particles are active porous structures and release soluble proteins over time

Before exploring the possibility of exploiting the initial dynamic state of the protein-rich phase to tune the composition of the aggregates, we first verified that the chimera proteins maintain their activity in the aggregated state. We observed that the aggregates of the GFP chimera proteins exhibit fluorescence, thereby indicating that GFP maintains its native activity. We then measured the activity of AK in the protein-rich phase by adding the corresponding substrate into a solution of AK protein particles and monitoring the formation of the fluorescent product by epi-fluorescence microscopy (Fig. 14A). We tested both liquid droplets generated during the early stages of the process and mature solid aggregates formed over longer incubation times. For both structures, the images show the co-localization of the fluorescence intensity within the protein-rich phases, thereby indicating that the chimera- proteins self-assemble into aggregates that retain their native activity (Fig. 14A and Fig. 22). We note that the protein particles exhibit similar activities to those of the soluble proteins at concentrations that are around 200-fold lower (Fig. 23). Such decrease of the enzymatic activity in the aggregate state can be likely explained by the reduced amount of accessible active sites of the enzyme.

Despite this decrease in activity, these results demonstrate the possibility to generate active protein particles that can potentially work as micro-bioreactors. To do so, these protein aggregates should exhibit a high level of porosity, thereby allowing the diffusion of soluble reagents and products into and out of these structures. We investigated this feature by monitoring the diffusion of a fluorescent probe, namely the dye Thioflavin T (ThT) in the presence of such aggregates. We observed that this molecular rotor enhances its fluorescence inside the protein-rich phase, either because of the increased viscosity or because of interactions with proteins at high concentration, and can therefore be used to stain both the liquid droplets and the protein particles. We exploited this feature to monitor the diffusion of ThT into and out of the particles of AK-LCD2 by epi-fluorescence microscopy. To this aim, ThT was introduced into a solution of the protein particles and the fluorescence profiles along the cross-section of protein aggregates were monitored over time (Fig. 14B). In a second experiment, protein aggregates were pre-equilibrated with a solution containing ThT and incubated in a buffer without ThT, and the release of ThT into the supernatant was recorded over time by monitoring the fluorescence profile inside the protein particles (Fig. 14C). The two experiments revealed that the dye is capable of diffusing both into and out of the protein structures, thereby indicating that the protein particles are porous structures. Next, we tested whether or not the aggregates of AK-LCD2 release active soluble proteins over time. We washed the droplets ten times with the assay buffer to remove any soluble proteins and exchanged the supernatant with a solution containing an excess of substrate of AK (see Materials and Methods and Suppl. Fig. 24). We then measured the activity of the supernatant by fluorescence spectroscopy over several days. The fluorescence intensity is proportional to the concentration of the product of the enzymatic reaction, and therefore to the concentration of the soluble enzyme in solution. We observed an increase of the fluorescence intensity over time (Fig. 14D), indicating the release of soluble AK chimera proteins. Overall, these results demonstrate that the chimera protein maintains activity in the aggregate state.

Multi-functional biomaterials and micro-bioreactors

Lastly, we demonstrated the possibility of incorporating multiple functions within a given particle by exploiting the dynamic behaviour of the dense liquid phase (Fig. 14E). We predicted that mixtures of different globular proteins conjugated with the same molecular adhesives would induce not only self-assembly of a given chimera (Fig. 12) but also attractive interactions between chimera proteins with different globular domains. To test this, we generated liquid droplets from solutions containing either AK-LCD2 or GFP-LCD2 (Fig. 15 A). Interestingly, when we mixed these solutions containing the preformed liquid droplets, we initially observed similar particles that were generated from the solutions containing the individual proteins alone (Fig. 15B). However, we observed the GFP-LCD2 droplets in the mixture becoming incorporated into the larger AK-LCD2 structures over time (Fig. 15C). After incorporation, the GFP chimera droplets appeared to diffuse into the dynamic matrix of AK-LCD2 proteins, thereby losing their identity as shown by the time-dependent fluorescence profiles of GFP-LCD2 droplets after their incorporation within the AK-LCD2 structures (Fig. 15C and 15D). In order to interpret our data in a more quantitative way we described the time evolution of the cross-sectional fluorescence by Fick’s diffusion law (Fig. 15D). The comparison between model simulations and experimental data provided a diffusion coefficient of the GFP-LCD2 protein of l.lxlO^-12 m²/s, which is about one order of magnitude smaller compared to the value measured in solution by dynamic light scattering (8.9xl0^~12 m²/s). This result indicates that the viscosity of the protein-rich phase is indeed higher as compared to aqueous solutions, yet still sufficiently low to guarantee high molecular diffusion. After one hour, we observed an arrest of the diffusion process and the presence of a homogenous structure containing two different proteins, which is preserved during the maturation into the solid-like aggregates. We verified that the function of both proteins is preserved within this structure by adding the reagents of the enzymatic reaction catalysed by AK and simultaneously acquiring fluorescence images corresponding to both the GFP protein and the fluorescent product of the reaction (Fig. 15E).

In contrast to the mixture of AK-LCD2 and GFP-LCD2, we did not observe any fusion event between the droplets generated in a mixture of AK-LCD1 and GFP-LCD2 (Fig. 15F), supporting our conclusion that specific interactions can be modulated via LCD sequence. Taken together, our results demonstrate that the molecular adhesives described in this work represent a promising strategy to bring together different proteins within the same architecture, thereby creating multi-functional biomaterials and microscale bioreactors exhibiting high concentrations of homogeneously dispersed reagents.

Conclusions and outlook

We have developed molecular adhesives based on disordered sequences of proteins found in nature associated with phase separation. We have demonstrated that the conjugation of globular proteins to these sequences induces a controlled self-assembly into supramolecular structures via formation of dense liquid phases that mature over time into solid-like aggregates. The molecular adhesives enable one to induce specific attractive interactions with sequences of low complexity. Such attractive interactions include non-polar, polar, and electrostatic forces and are not based on general hydrophobicity. In particular, we found that a positive net charge is required to promote phase separation. The globular domains within our fusion proteins maintain their activity in the aggregated state. We have shown that this strategy can be applied to develop porous protein materials that release active proteins over time. Moreover, we have demonstrated that the dynamic nature of the protein-rich phase during the early stages of the maturation process, together with the specific interactions induced by the LCDs, can be exploited to control the composition and the size of the final aggregates. Furthermore, we have shown that the molecular adhesives can bring together different functionalities within the same architecture, thereby opening novel opportunities for the generation of multifunctional biomaterials and micro-scale bioreactors. Adaptive multifunctional composites via supramoleeular co-assembly of inorganic nanoparticles and bio-inspired intrinsically disordered proteins:

In this work, for the first time, we also transfer the dynamicity and adaptability of this class of proteins to inorganic nanoparticles, generating composite materials with controlled self- assembly properties and sensitivity to external cues, thereby creating stimulus- responsiveness.

We demonstrate this concept with enzymes and magnetic nanoparticles, generating magnetosomes of controlled size and morphology in the micron range that can act as enzymatic microreactors and that can sense the concentration of biomolecules in the surrounding environment. Overall, this platform represents a convenient tool to generate protein-inorganic materials with highly controlled self-assembly behavior, stimulus- responsiveness, and multiple functionalities.

We functionalized magnetic nanoparticles (NPs) with LCDs derived from the helicase DEAD-box protein LAF-1, which is the first protein identified in association with functional liquid-liquid phase separation in biology and one of the most characterized both in vivo and in vitro. These LCDs are enriched in polar aminoacids and mediate pi-cation, hydrophobic, and electrostatic interactions. We have above demonstrated that the conjugation of LCDs derived from DEAD-box proteins to globular domains generate chimera proteins that can phase separate into dynamic liquid droplets. For this reason, we conjugated the LAF-l with the globular enzyme adenylate kinase (AK) to combine the self-assembly behaviour of the LCD with an enzymatic activity (Figure 25 A-B). We functionalized the NPs with this chimera protein by non-covalent interactions between an anti-His antibody conjugated on the surface of the nanoparticles and the conventional His-Tag engineered into recombinant proteins to promote their purification (Figure 25 C).

The functionalization of the nanoparticles with LCDs confers to the protein-nanoparticle composite the same ability to self-assemble of the individual chimera proteins. We verified this prediction by a combination of light scattering, transmission electron microscopy, and optical microscopy (Figure 25 D, Figure 26). The composite is able to assemble into solid- like micron-sized clusters that are structurally composed of inorganic nanoparticles embedded in a protein-rich phase (insert Figure 25 D, Figure 26 E-F). Control experiments with the poly His-Tag antibody nanoparticles and the enzyme AK lacking the LCDs confirmed that the formation of these protein/NP clusters was specifically due to interactions mediated by the LCDs (Figure 27). In analogy with the individual chimera proteins, we observed that at higher ionic strength the functionalized nanoparticles remained stable over several days (Figure 28). In contrast, a decrease in ionic strength induced the formation of clusters of proteins and nanoparticles in the micrometer size range (Figure 29A). In analogy with the droplets formed by the bare protein, the size of the composite clusters increased with decreasing the ionic strength, indicating the presence of electrostatic attractive interactions that are screened at low salt concentrations.

The importance of the electrostatic interactions is confirmed by the observed dependence of the self-assembly on the pH value (Figure 29B), which, in analogy with ionic strength, is similar for the protein-NP composites and the protein chimera. Phase separation occurred in the pH interval ranging from 5 to 9 while homogenous solutions were observed at more acidic and basic pH values (Figure 29B). This behavior is likely due to the charged and polar residues of the LCDs, in particular to negatively charged aminoacids and tyrosines, which modulate electrostatic and cation-pi interactions, respectively.

In analogy with the biological LCDs, the different self-assembly behavior in response to pH and ionic strength confers adaptability to these composites, allowing them to rapidly react to changes in the environment. We demonstrated this concept by developing a pH clock and inducing the reversible assembly and disassembly of the NP/protein composite upon variations in the pH value over time (Figure 30). The pH was changed via titration with 10 mM NaOH and 10 mM HC1 solutions. Following the phase diagram shown in Figure 29B, the NP/protein composites remained stable at acidic pH values and formed clusters in the interval of pH ranging from 5 to 9. These clusters reversibly dissociated at pH higher than 9. Notably, we observed the same behavior by reversing the titration and starting with pH above 9 and then reducing the pH below 5. We repeated this cycle sequentially over time, demonstrating the possibility to achieve dynamic spatiotemporal control on the self- assembly of the composite material (Figure 30). Moreover, we generated a similar clock by changing ionic strength over time (Figure 31).

Next, we demonstrated the possibility to control the morphology and the rheological properties of our adaptive composite materials. To this aim, we took advantage of the response of the inorganic nanoparticles to magnetic fields. Specifically, the alignment of the magnetic dipoles caused by the magnetic field, together with the phase separation promoted by the interactions encoded by the LCDs, induced the formation of supramolecular linear magnetosomes (Figure 32A). Moreover, by increasing the stoichiometry of proteins with respect to NPs, we could generate not only solid clusters but also round-shaped droplets that incorporate the functionalized magnetic nanoparticles and, at the same time, that are enough viscous to suppress droplet- droplet coalescence events . Application of the magnetic field induced the formation of droplet chains of these protein-rich droplets (Figure 32B). In addition, in sharp contrast with the systems reported in literature so far, our solid fibrils and droplet chains are dynamic and reversible upon pH and ionic strength variation. In fact, they retain the same behavior of the NP/protein clusters (Figure 33 and 34). In contrast, the bare nanoparticles remain stable under the same magnetic field for several days demonstrating that the phase separation induced by the chimera protein is strictly required to generate these supramolecular structures (Figure 35).

Finally, we proved the possibility to use our responsive composite materials to achieve control of activities in time and space (Figure 36). By applying fluorescence microscopy, we monitored the activity of the phosphotransferase enzyme AK, which catalyzes the interconversion of adenine nucleotides, and observed that the droplet chains locally concentrate the activity of the enzyme in space and, at the same time, provide a supramolecular micro-environment where reactants and/or products can be up-concentrated (Figure 36A-C). In analogy, we expect also the localization of the peroxidase-like activity of the inorganic nanoparticles, which can catalyze the reduction of the hydrogen peroxide generated by the glucose oxidation (Figure 36D). The glucose oxidase/peroxidase cascade reaction is a widely established method to quantify glucose concentration in diagnostics and it has been widely adopted in literature to produce functional biosensors. Here we expect that our synthetic membrane-less composite compartments will be able to sense the glucose presence in the surrounding environment by localizing the fluorescent signal generated by the iron nanoparticles-mediated conversion of a fluorogenic reporter (Amplex Red) (Figure 36D).

In this part we have reported on a strategy to transfer the adaptability of biological biopolymers to inorganic materials. Specifically, we have functionalized magnetic nanoparticles with intrinsically disordered sequences derived from biological proteins associated with the formation of membrane-less compartments in cells. This strategy allowed us to reproduce the controlled biological phase separation in synthetic systems, generating protein- nanoparticle materials with controlled self-assembly properties and stimulus responsiveness to pH and ionic strength. We have further proved that our approach allows controlling multiple functionalities in space and time, generating both multifunctional materials and open micro-reactors capable to locally confine chemical reactions and to sense the concentration of biomolecules in the surrounding environment. As a proof of principle, we have generated reactive colloidal linear magnetosomes characterized by enzymatic activity.

Overall, our findings represent the synthesis of advanced adaptive organic-inorganic systems capable to perform multiple tasks and to respond to multiple stimuli. Applications are in several fields including heterogeneous biocatalysis, biosensing, and nanomedicine.

Experimental Section

Protein Synthesis and Purification. AK and LAF1-AK-LAF1 were produced according a previously published protocol with the constructs reported in Table S l . Briefly, the chimera proteins were expressed in E.coli BL21-GOLD (DE3) cells. Protein production was induced at optical density (OD) equal to 0.7 with 0.5 mM isopropyl d-thiogalactopyranoside (99%, PanReac AppliChem). After 16 h at 37 °C, the recombinant proteins were purified using His-tag immobilized metal ion affinity chromatography (Chelating Sepharose, GE Flealtchare). The proteins were further purified via size exclusion chromatography using a Superdex 75 26/600 on an AKTA Prime system (GE Healthcare) in 50 mM Tris (pH =7.5) with 500 mM NaCl. The quality of the purified proteins was confirmed by SDS-PAGE electrophoresis and SEC-MALS analysis. The proteins were concentrated at 600-700 mM and aliquots were frozen and stored at -20 °C.

Nanoparticles Characterization. Magnetic nanoparticles decorated with antibody selective for the His-Tag sequence were bought from Miltenyi Biotec (Anti-His MicroBeads). NP size, NP Pdl and NP concentration were evaluated via nanoparticle tracking analysis (NTA) (ZetaView, Particle Matrix) and dynamic light scattering (DLS) (Zetasizer Nano, Malvern Panalytical). CNP_S= 9.9 10¹² #/ml.

Transmission Electron Microscopy (TEM). TEM images were acquired on a HITACHI HT77000 EXALENS (ScopeM, ETH Zurich).

Solid Linear Fibrils Synthesis and Characterization. The solid linear fibrils were produced at 0.5 mM LAF1-AK-LAF1 with a Protein/NP molar ratio of 3000. Briefly, 1 mΐ of 9.9 10¹² #/ml NPs were incubated for 30 minutes with 1 mΐ of 50 mM labeled LAF1-AK- LAF1 in 50 mM Tris (pH=7.5) with 500 mM NaCl. The NP composite was then pipetted in 98 mΐ deionized water in a 384- well plate (Glass Bottom, Brooks) with a common commercial magnet positioned horizontally in order to generate a magnetic field. Representative images were collected on an epi-fluorescence microscope (Eclipse Ti-E, Nikon)

Droplet chains synthesis and characterization. The protein-nanoparticle droplet chains were produced at 2 mM LAF1 -AK-LAF1 with a Protein/NP molar ratio of 12 000. Briefly, 1 mΐ of 9.9 10¹² #/ml NPs were incubated for 30 minutes with 1 mΐ of 200 mM labeled LAF1 - AK-LAF1 in 50 mM Tris (pH=7.5) with 500 mM NaCl. The NP composite was then pipetted in 98 mΐ deionized water in a 384-well plate (Glass Bottom, Brooks) with a common commercial magnet positioned horizontally in order to generate a magnetic field. Representative images were collected after 1 h incubation on an epi-fluorescence microscope (Eclipse Ti-E, Nikon) equipped with a CCD Camera (Zyla sCMOS, Andor) using a 60x oil objective (CFI Plan Apo Lambda 60x Oil, Nikon). Fluorescence was measured using a 455 nm high power LED light source (ledHUB light engine, Omicron) and a EGFP ET Filter set (Chroma Technology Corporation).

Microreactor Activity. The activity of the droplet chains was evaluated using a standard AK fluorimetric assay (Abeam, ab2l 1095). The AK reaction mix was prepared according to the manufacturer protocol and then 1 mΐ of this solution was added to the heteropolymer solution. Fluorescence was measured by an epi-fliuorescence microscope (Ti-E, Nikon) equipped with a 550 nm high power LED light source (ledHUB light engine, Omicron) and a Cy3 AT Filter set (Chroma Technology Corporation).

Table Sl . Amino acid sequences of the protein variants used in this part of the work. Dark grey, bright grey backgrounds and bright grey underlined indicate the His-tag, the AK protein, and the LAF-l low complexity domains, respectively.

GKQAKDIMDAGKLVTDELVIALVKERIAQEDCRNGFLLDGFPRTIPQADAMKEAGINVDYVLEFDV PDELIVDRIVGRRVHAPSGRVYHVKFNPPKVEGKDDVTGEELTTRKDDQEETVRKRLVEYHQMTAP LIGYYSKEAEAGNTKYAKVDGTKPVAEVRADLEKILG

LAF_AK_LAF (SEQ-ID 16)

M· iSM II m m lUVst .l \ pm .Sl IMESNOSNNGGSGNAALMRGGRYVPPHLRGGDGGAAAAASAGGDD RRGGAGGGGYRRGGGNSGGGGGGGYDRGYNDNRDDRDNRGGSGGYGRDRNYEDRGYNGGGGG GGNRGYNNN RGGGGGGYNRQDRGDGGSSNFSRGGYNNRDEGSDNRGSGRSYNNDRRDNGGDGM

RIILLGAPGAGKGTQAQFI EKYGIPQISTGDMLRAAVKSGSELGKQAKDIMDAGKLVTDELVIALV

KERIAQEDCRNGFLLDGFPRTIPQADAMKEAGINVDYVLEFDVPDELIVDRIVGRRVHAPSGRVYHV KFNPPKVEGKDDVTGEELTTRKDDQEETVRKRLVEYHQMTAPL1GYYSKEAEAGNTKYAKVDGTK

PVAEVRADLEKILGMESNOSNNGGSGNAALNRGGRYVPPHLRGGDGGAAAAASAGGDDRRGGAG

GGGYRRGGGNSGGGGGGGYDRGYNDNRDDRDNRGGSGGYGRDRNYEDRGYNGGGGGGGNRGY

NNNRGGGGGGYNRODRGDGGSSNFSRGGYKNRDEGSDNRGSGRSYNNDRRDNGGDG

LIST OF REFERENCE SIGNS AND ABBREVIATIONS

AK Adenylate Kinase LCD2 DEAD-box protein Dbpl

GFP Green fluorescent Protein LCD

LCD Low complexity domain UPST upper critical solution

LCD1 DEAD-box protein Dhhl temperature

LCD

Claims

1. A fusion system exhibiting a phase transition, the fusion system comprising:

(a) one or more biologically and/or chemically active core systems;

(b) at least two proteins exhibiting a phase transition joined to the core system; and

(c) optionally, a spacer or intermolecular interaction sequence separating any of the proteins of (b) from any of the core system(s) of interest of (a),

wherein the proteins of (b), taken individually, together or both individually and together, have a length of at least 40 amino acids, are intrinsically disordered, and do not show a repetitive amino acid pattern.

2. A fusion system according to claim 1, wherein the core system is at least one organic or inorganic nanoparticle, preferably a magnetic nanoparticle, and/or at least one water-soluble molecule of interest with low intermolecular interactions selected from at least one of the group consisting of therapeutically active organic molecule, bio macromolecule, in particular protein, peptide, DNA, RNA, oligonucleotide, carbohydrate or derivatives thereof, wherein preferably the solubility at room temperature in aqueous buffer solutions with ionic strength lower than 150 mM is at least 0.5 g/L.

3. A fusion system according to claim 1 or 2, wherein the core system is a protein or a peptide, preferably in globular form, and the at least two proteins exhibiting a phase transition are joined at the N as well as the C terminus, respectively, to the biologically and/or chemically active molecule of interest

and/or wherein the core system is an inorganic nanoparticle, preferably a magnetic and/or metallic and/or catalytic nanoparticle.

4. A fusion system according to any of the preceding claims, wherein the proteins of

(b), taken individually, together or both individually and together, have the following composition by number of amino acids:

polar amino acids: in the range of 20-60%, preferably 25 and 50 %;

apolar amino acids: in the range of 20-60%, preferably 25 and 50 %; negatively charged amino acids: in the range of 3-30%, preferably 5 and 20 %; positively charged amino acids: in the range of 3-30%, preferably 5 and 20%; hydrophobic amino acids: in the range of 5-30%, preferably 10 and 25 %;

aromatic amino acids: in the range of 3-45%, preferably 5 and 40 %;

glycince/proline: in the range of 5-40%, preferably 10 and 30 %;

wherein the polar amino acids are selected from the group consisting of S, T, N, Q, R, H, K, D, E, C and combinations thereof;

wherein the apolar amino acids are selected from the group consisting of A, U, L, M, F, W, Y, V, G, P and combinations thereof;

wherein the negatively charged amino acids are selected from the group consisting of C, D, E and combinations thereof;

wherein the positively charged amino acids are selected from the group consisting of R, H, K, Y and combinations thereof;

wherein the hydrophobic amino acids are selected from the group consisting of W, Y, K, D, E and combinations thereof;

wherein the aromatic amino acids are selected from the group consisting of F, Y W, H

and wherein glycince/proline are selected from the group consisting of G, P.

5. A fusion system according to any of the preceding claims, wherein the proteins of (b), are selected from LCD stretches from naturally occurring, in particular RNA binding proteins, in particular of Dhhl, Dbpl, Lafl, Ddx4, hNRNPAl, FUS, hnRNPNA2 as well as polypeptides comprising an amino acid sequence characterized by at least (>)85%, >87.5%, >90%, >92%, >94%, >95%, 96%, >97%, >98%, >99% sequence identity and/or amino-acid composition identity to these LCD stretches and/or in which not more than 5, preferably not more than 3, or 2 or 1 amino acid is removed from these sequences at one or both ends thereof, wherein preferably the core system is a protein or a peptide, and the proteins of (b) are selected for the N-terminus from SEQ-ID 7 and SEQ-ID 9 and for the C- Terminus from SEQ-ID 8 and SEQ-ID 10, or in each case a fraction of at least 50% in length thereof, or from polypeptides comprising an amino acid sequence characterized by at least (>)85%, >87.5%, >90%, >92%, >94%, >95%, 96%, >97%, >98%, >99% sequence identity and/or amino-acid composition identity to these LCD stretches and/or in which not more than 5, preferably not more than 3. or 2 or 1 amino acid is removed from these sequences at one or both ends thereof.

6. A fusion system according to any of the preceding claims, wherein the core system is a protein or a peptide, and is selected from the group consisting of enzymes, including kinases, receptors, toxins, antibodies, collagenes, ion channels, transport proteins, hormones, fluorescent proteins, or is a post- translational ly modified protein, in particular a glycosylated, phosphorylated, acetylated, acylated or sulphated protein.

7. A fusion system according to any of the preceding claims, wherein it exhibits phase separation when exposed to a temperature that is (i) above a lower critical solution temperature of the fusion system, and/or (ii) below an upper critical solution temperature of the fusion system, wherein the phase separation can be reversible or irreversible, and wherein preferably the fusion system exhibits heat-irreversible phase separation when exposed to a temperature that is above a lower critical solution temperature of the fusion system, and exhibits reversible phase separation when exposed to a temperature below the upper critical solution temperature, wherein it preferably exhibits a reversible phase separation in response to a first stimulus and an irreversible phase separation in response to a second different stimulus.

8. Use, in particular in vitro use, of a fusion system according to any of the preceding claims for liquid-liquid phase separation, in particular for the purification of biologically and/or chemically active molecules of interest, water-in-water emulsions with recruitment of further components in the disperse phase, surface coating, provision of enzymes in organic-like phases, tailored mixing of different biologically and/or chemically active molecules of interest in liquid-liquid phase separated globular structures, either by attaching the same proteins exhibiting a phase transition to the different biologically and/or chemically active molecules to mix them within the same globular structures, or by attaching different proteins exhibiting a phase transition to the different biologically and/or chemically active molecules to provide for distinct globular structures with either of the biologically and/or chemically active molecule.

9. Use, in particular in vitro use, of a fusion system according to any of the preceding claims for the generation of particles and/or gels, for providing multifunctional nanomaterials in biotechnology; targeted drug delivery; controlled release depots of pharmaceuticals; immobilized enzymes; tissue engineering scaffolds; components for (bio)sensing and (bio)analysis.

10. Method for the generation of stable reversible globular structures in a liquid/liquid phase separated system, said globular structures comprising at least one fusion system according to any of the preceding claims or consisting of at least one or more of the fusion systems according to any of the preceding claims, wherein a fusion system according to any of the preceding claims is immersed in water, and wherein the size of the preferably globular structure is adapted by tuning the pH, the ionic strength, the temperature, the shear, or a combination thereof, to show a number average globular structure diameter in the range of at least 200 nm, or in the range of 1-100 pm, preferably in the range of 5-50 pm.

11. Method for the generation of stable essentially irreversible globular or fibrillar structures comprising at least one fusion system according to any of the preceding claims or consisting of at least one or more of the fusion systems according to any of the preceding claims, wherein a fusion system according to any of the preceding claims is immersed in water, and wherein

in a first step the fusion systems are allowed to form liquid droplets in the water environment, wherein preferably the size of the globular or fibrillar structure is adapted by tuning the pH, the ionic strength, the temperature, the shear, or a combination thereof, more preferably to show a number average globular structure diameter in the range of 1-100 pm, preferably in the range of 5-50 pm and wherein in a second step the globular or fibrillar structures are incubated for at least an hour, optionally followed by washing, to form said stable globular or fibrillar structures, further optionally followed by recruitment of further components into the globular or fibrillar structure.

12. Globular or fibrillar structure comprising at least one fusion system according to any of the preceding claims or consisting of at least one or more of the fusion systems according to any of the preceding claims, preferably obtained or obtainable using a method according to claim 10 or 1 1, wherein it is preferably a porous globular structure, more preferably recruited with further components for functional use thereof,

wherein the globular or fibrillar structure can preferably be embedded in a cross- linked water-soluble polymer structure, preferably a hydrogel, in particular

for slowing down release rates of the fusion system and/or of further components compared to the free fusion system and/or of further component, and/or

for switching of rate limiting steps from diffusion of the free fusion system to dissolution rate of the droplets, and/or

for controllable release rates tailored for the respective application, and/or to increase protein concentration.

13. A polynucleotide comprising a gene or cell line expressing a fusion system in the form of a protein according to any of the preceding claims.

14. A pharmaceutical composition comprising a non-immunogenic fusion system, preferably a protein according to any of the preceding claims or a globular or fibrillar structure according to claim 1 1 and a pharmaceutically acceptable carrier.

15. A method of producing one or more fusion system in the form of a protein according to any of the preceding claims comprising:

(a) transforming a host cell with an expression vector comprising a polynucleotide comprising a nucleotide sequence encoding a protein according to any of the preceding claims; and

(b) causing the host cell to express the protein.