Manipulation of the dually thermoresponsive behavior of peptide‐based vesicles through modification of collagen‐like peptide domains

Abstract Materials that respond to temporally defined exogenous cues continue to be an active pursuit of research toward on‐demand nanoparticle drug delivery applications, and using one or more exogenous temperature stimuli could significantly expand the application of nanoparticle‐based drug delivery formulations under both hyperthermal and hypothermal conditions. Previously we have reported the development of a biocompatible and thermoresponsive elastin‐b‐collagen‐like polypeptide (ELP‐CLP) conjugate that is capable of self‐assembling into vesicles and encapsulating small molecule therapeutics that can be delivered at different rates via a single temperature stimulus. Herein we report the evaluation of multiple ELP‐CLP conjugates, demonstrating that the inverse transition temperature (T t) of the ELP‐CLPs can be manipulated by modifying the melting temperature (T m) of the CLP domain, and that the overall hydrophilicity of the ELP‐CLP conjugate also may alter the T t. Based on these design parameters, we demonstrate that the ELP‐CLP sequence (VPGFG)6‐(GPO)7GG can self‐assemble into stable vesicles at 25°C and dissociate at elevated temperatures by means of the unfolding of the CLP domain above its T m. We also demonstrate here for the first time the ability of this ELP‐CLP vesicle to dissociate via a hypothermic temperature stimulus by means of exploiting the inverse transition temperature (T t) phenomena found in ELPs. The development of design rules for manipulating the thermal properties of these bioconjugates will enable future modifications to either the ELP or CLP sequences to more finely tune the transitions of the conjugates for specific biomedical applications.


| INTRODUCTION
Thermoresponsive materials have been of prevailing interest for the last decade and continue to be widely studied for purposes ranging from biomaterials engineering 1 to the textiles industry. 2 Specifically, externally-triggered, stimuli-responsive materials have been developed for long-term application in coated surfaces for anti-fouling films, 3 cellular adhesion and removal devices, 4 anti-opsonization materials, 5 and on-demand drug and gene-releasing nanoparticles. [6][7][8][9] Drug delivery applications would benefit in particular from the spatiotemporal control that is provided by thermoresponsive materials since a locally altered temperature profile allows for site-specific release of therapeutics and thereby is ideal to enable improved drug targeting. 10 Thermoresponsive materials also would allow selection of when a drug is to be released (i.e., based upon when the external temperature stimulus is applied) relative to a patient's condition, and this temporal control would in turn allow the local concentration of the drug to be maintained above a therapeutically relevant threshold at any given time. 11 Dual stimuli-responsive polymeric nanoparticles have garnered particular research interest in recent years. These dual stimuliresponsive nanoparticles often employ thermal responsivity as one of the two stimuli, with designs employing pH/temperature-, magnetic/ temperature-, light/redox-, to pH/redox-responsive particles among others 6,[12][13][14] ; moreover, the behavior of many dual-responsive particles is tailored for a specific disease. For instance, Zhang et al. produced nanoparticles from diblock copolymers that were both pH and thermoresponsive, and they showed increased drug release in the presence of both a higher temperature (e.g., >20 C) and a lower pH (e.g., <6.8) stimuli that are both commonly found in cancerous tumor microenvironments. 6 Similarly, Sun et al. reported the synthesis of pH and reduction sensitive polymersomes for the delivery of apoptotic proteins to cancer cells that typically possess an imbalance in reduction potential. 15,16 One general challenge in applying such approaches in a clinical setting is that the conditions in endogenous tumor environments are often difficult to predict and commonly vary from person to person. 17 While the specificity offered by a mixed dual stimuli-responsive (e.g., two stimuli that differ from one another in type) system is desired in certain cases, dual temperature-responsive particles would have broad and diverse applications for drug delivery in multiple disease models. Specifically, a dual thermoresponsive particle able to dissociate upon application of either hyperthermal or hypothermal temperatures would be relevant for therapeutic regimens that include a co-operative applied temperature stimulus. For instance, hyperthermal treatments have been used in conjunction with small molecule drugs for the treatment of cancer. 18 Similarly, hypothermal therapies have been shown to aid the delivery of small molecules in diseases such as osteo-and rheumatoid arthritis, 19 embolic stroke, 20 and spinal cord injuries. 21 In both cases, the same nanoparticle formulation could be used for treatment with only the drug cargo being altered for different diseases.
Additionally, dual thermoresponsive particles may offer unique drug release kinetics from each of the two different temperatureresponsive domains, 22,23 enable unique alterations in particle morphology to increase targeting ligand valency/presentation on the particle surface, 24,25 or enable nanoparticle aggregation 6,26 to sitespecifically trap particles and their cargo in specific locations.
Interestingly, there are very few reports that focus on dual thermoresponsive polymeric nanoparticle systems. [27][28][29] Of the different types of dual thermoresponsive nanoparticles that have been studied, many employ copolymers comprising blocks of poly(Nisopropylacrylamide) (PNIPAAM), 27,30 poly(ethylene oxide), 31 and poly(diethylene glycol methyl ether methacrylate), 32 which possess a lower critical solution temperature (LCST). Such dual thermoresponsive particles typically include two distinct polymer domains, each with a distinct LCST, so that particle size or morphology can be changed at each of the LCSTs. 28 The transition above or below either LCST can also result in the complete dissociation of the nanoparticle to its constituent assembly units, and thereby act as a spontaneous drug release mechanism. 17 While many of these dual thermoresponsive polymer nanoparticle systems have been foundational in the field, there are a limited number of polymers with LCSTs in the physiologically-relevant range, and the few reports that mention manipulation of the LCST require complex chemical modifications that range from the attachment/replacement of specific chemical handles or conjugation with peptides, proteins, or other polymers. 10,33,34 The synthesis routes for many of these polymers are facilitated by techniques such as reversible addition-fragmentation chain transfer 29 and quasi-living radical polymerization via Ce(IV) ion reduction, 31 which can require careful selection of specialized monomer and chain-transfer agents. 35,36 Nature-inspired thermoresponsive biomaterials such as recombinantly engineered or chemically synthesized polypeptides serve as good alternatives to synthetic polymers, owing to the defined peptide sequence and chain length, which in turn provide precise control of structural, chemical, and thermodynamic properties. 37,38 The most common thermoresponsive polypeptides that have been studied in recent years include collagen-like peptides (CLPs) 39 and elastin-like polypeptides (ELPs). 40 Briefly, CLPs consist of a characteristic (G-X AA -Y AA ) n amino acid repeat sequence where X AA and Y AA can be any Lamino acid but are typically the imino acids proline and hydroxyproline (P and O, respectively). Similarly to native collagen protein, individual chains of CLPs with repeat lengths of n ≥ 6 form a left-handed polyproline type II helix, and these strands in turn form a right-handed triple helix upon alignment of the three individual chains with a one residue stagger between each chain. 41 The resulting triple helix possesses a characteristic melting temperature (T m ) that is dependent on both the number of G-P-O repeats and the amino acid sequence; the effects of such parameters on T m have been rigorously investigated by others. [42][43][44][45] CLPs have been tested for their potential in diagnostic applications, 46 their ability to form supramolecular structures, [47][48][49][50] and their properties as components in hydrogels 39,51 and cell adhesion scaffolds, 52,53 but CLPs have infrequently been studied as thermoresponsive elements in materials.
Similar to CLPs, ELPs are derived from a mammalian protein, in this case the elastin precursor protein tropoelastin. Most sequences that have been studied possess the characteristic repeat of (V-P-G-X AA -G) n , where X AA can be any amino acid with the exception of proline. 37 Like the polymers PNIPAAM and PEG, ELPs possess an LCST-like transition, also known as the inverse transition temperature (T t ), above which they form a coacervate phase in which they are not soluble and aggregate; upon re-cooling below the T t they resolubilize. 37 The ELP transition temperature phenomenon is rooted in the dehydration and hydrophobic association of the ELP chains, but like CLP folding/unfolding, is fully reversible. 54 ELPs have been investigated for their ability to form nanoparticles with promising properties in drug delivery applications. 55,56 In such applications, most ELP nanoparticles drive the formation of micelles by a variety of mechanisms including electrospraying, 57 amphiphilic selfassembly (e.g., using di-block ELP fusion constructs of differing hydrophobicities), 58 or conjugation with hydrophobic drug loaded cysteine ligands. 25 While such assemblies are perhaps more attractive over synthetic polymers due to their innate biocompatibility, they still rely largely on a singular T t for assembly and do not fully utilize this thermodynamic parameter for time-activated release of encapsulated drug molecules. 59 A recent report has highlighted the recombinant fusion constructs of block ELPs and resilin-like polypeptides to form a variety of self-assembled morphologies, though despite the presence of two thermoresponsive blocks, no dual thermoresponsiveness was reported. 60 We previously reported that the short ELP peptide, (VPGFG) 6 , possesses an experimentally inaccessible T t (>80 C in pure aqueous conditions) 61 ; conjugating the CLP (GPO) 4 GFOGER(GPO) 4 GG to the ELP enables the ELP to undergo an inverse phase transition and collapse, but only when the CLP domain is in triple helical form. The resulting folded ELP-CLP trimers, upon coacervation of the ELP domain, were observed to spontaneously self-assemble into vesicles with a bilayer comprising ELP-CLP ternary subunits. These vesicles were shown to be stable over a significant temperature range but could dissociate at~70 C due to the unfolding of the CLP triple helical domain. 61 Herein, we sought to not only lower this CLP-mediated hyperthermal dissociation temperature of vesicles to a more clinically approachable value but also determine an underlying set of design rules for the ELP-CLP vesicles and the role that CLPs have in their self-assembly. To accomplish this, we synthesized a small library of CLP sequences that have previously been shown to possess T m values lower than that of the (GPO) 4 GFOGER(GPO) 4 GG CLP originally studied. 41

| Synthesis and characterization of elastin-like and collagen-like peptides
The CLP peptides N 3 -(GPO) 6

| CD spectroscopy of peptides and conjugates
In order to characterize triple helix formation, CD spectroscopy was performed on all peptides and conjugates. All CD experiments were performed using a Jasco 810 CD spectropolarimeter (Jasco, Easton, and F6-GFOGER conjugates, as these sequences did not selfassemble under the experimental conditions. Conversely, a melting temperature scan could not be performed for F6-GPP10 due to significant aggregation and settling. All wavelength scan data (at 4 and 80 C) are provided in Figures S13, S14a, S15, and S16 (CLP, ELP, conjugates excluding F6-GPP10, and F6-GPP10 respectively). The ELP melting curve scan is provided in Figure S14b and all CLP melting curves are provided in Figure S17. The melting temperature for all CLPs and ELP-CLP conjugates was defined as the minima of the first derivative of a Boltzmann function that was fit to the melting curve transition between 4 and 80 C. The fitting of the melting curves using the sigmoidal Boltzmann function was completed using Origin 2017 software (Originlab Corporation, Northampton, MA).

| Vesicle inverse transition temperature determination
In order to determine the diameter of vesicles through the inverse transition temperature process for a given conjugate, the solution of vesicles was slowly cooled in 5 C intervals from 25 to 5 C, with each temperature setting employing a 5-min equilibration step followed by three light scattering measurements. As before, automatic attenuation and measurement position were employed. The scattering angle and refractive indices of the sample and water employed were the same as those noted in Section 2.5.1. The autocorrelation function was recorded with a 10 s correlation delay time with collection of 20 subrun correlation functions for each measurement. Each set of autocorrelation data was analyzed using the nonlinear least squares Malvern algorithm and the Stokes-Einstein equation as described above.

| TEM of ELP-CLP vesicles
All of the vesicles were prepared for TEM in a similar fashion with either a single staining protocol or a triple staining protocol that was used to enable observation of the vesicle bilayer. Carbon-coated copper grids were ionized with a PELCO easiGlow ® (Ted Pella Inc., Redding, CA) glow discharge unit to make the grids more hydrophilic prior to grid spotting. After overnight incubation of the sample at 25 C,  To determine the inverse transition temperature, the turbidity of the F6-GPO7 vesicle solution was continuously monitored with a 600 nm laser over the course of cooling from 25 to 0 C with a cooling rate of 1 C/min. The inverse transition temperature was defined as the temperature at which the turbidity was equal to that of 50% maximal turbidity, as in previous studies. 37 To assess the reversibility of the dissolution and re-assembly of vesicles, F6-GPO7 vesicles were repeatedly thermally cycled from In between heating and cooling steps, the 25 or 0 C temperature was held on average for 3 min while the turbidity or lack thereof was monitored.

| Turbidity measurements of F6-GPP10 aggregates
The attempted particle formation procedure for F6-GPP10 resulted in bulk aggregates. To assess the thermal responsiveness of these aggregates, turbidity measurements were performed with heating or cooling temperature gradients. To assess the hyperthermal mediated re-dissolution of the F6-GPP10 aggregates, a 1 mg/ml suspension of aggregates was heated in cuvette with a stir bar in a Cary 60 UV-Vis instrument with a 400 rpm stirring rate and a 2.5 C/min heating rate while being simultaneously monitored for turbidity with a 600 nm laser. For determining the inverse transition temperature, the F6-GPP10 aggregates were again suspended with stirring at 400 rpm and cooled from 25 to 0 C using a 1 C/min cooling rate. The bulk hyperthermal dissolution temperature and the hypothermal inverse transition temperatures for this conjugate were defined as the temperature where a 50% reduction in the maximal turbidity occurred.

| Synthesis and characterization of peptides
In this report, we seek to provide insight regarding the role of the CD spectroscopy wavelength and temperature scans were performed on all pure ELP and CLP products. The ELP wavelength scans and temperature scan are provided in Figure S14. It should briefly be noted that there is no significant positive contribution at 225 nm from the ELP at 4 C and only a minor negative ellipticity at this wavelength at 80 C ( Figure S14a). Additionally, the ellipticity change of the ELP at 225 nm through the course of heating from 4 to 80 C is linear and is not representative of a secondary structure transition ( Figure S14b). 68 This result was anticipated given that this short ELP has previously been shown to possess a T t greater than 80 C. 61 The CLPs were shown to form triple helices that in turn could subsequently be unfolded upon heating (Figures S13 and S17). The experimentally determined melting temperatures for each CLP are delineated in Table 1 Table 1 suggest that self-assembly of these ELP-CLP conjugates should be possible over a temperature range (27-45 C) that is physiologically and therapeutically relevant. 70 Furthermore, two conjugates exhibiting nearly the same T m (e.g., ((GPO) 7 and (GPP) 10 )) facilitate comparisons between the self-assembly of conjugates with different CLP chain lengths and/or hydrophobicities.
The ELP-CLP conjugates were subsequently purified via RP-HPLC and analyzed via UPLC-MS methods that generated elution profiles and corresponding mass spectra that confirmed the purity of each conjugate ( Figures S7-S11). Confirmation of successful purification of the ELP-CLP conjugates was further provided from FTIR absorbance spectra which verified the disappearance of the CLP azide chemical functionality of the ELP-CLP conjugates ( Figure S12).
As we have previously reported, the formation of the CLP triple helix is a critical step that enables the self-assembly of the ELP-CLP vesicles. 61 CD spectroscopy wavelength scans were therefore conducted on the purified conjugates to ensure triple helix formation (- Figures S15 and S16). It should briefly be noted here that the spectra of the ELP-CLP conjugates in Figures S15 and S16 correspond distinctly to triple helices. Though the existence of β-turn secondary structures for ELPs in the collapsed state has been reported previously, we show no conclusive evidence for their presence in the CD spectra of the conjugates reported here, most likely due to the considerably weaker absorption of β-turns relative to the absorption of triple helices, β-sheets, and α-helices. 71 No transitions were observed in the ELP domains alone ( Figure S14), as the ELP does not undergo a thermally-induced coacervation as previously reported. 61 The melting temperature of each of the ELP-CLP conjugates (with the exception of F6-GPP10) was determined by temperature scans that are shown in Figure 1. The T m of the conjugates showed the same stability trends as the corresponding CLP domains alone, with values of 31, 32, and 50 C for F6-GPO6, F6-GFOGER, and F6-GPO7, respectively ( Table 2). The melting temperature of F6-GPP10 could not be determined quantitatively given its precipitation at lower temperatures, but was estimated to be between 50 and 62.5 C given the N 3 -(GPP) 10 GG melting temperature (Table 1) and turbidity data (see

| Formation and characterization of ELP-CLP vesicles
In order to form vesicles, purified ELP-CLP conjugates were dissolved in pure deionized water, heated to 80 C, and equilibrated for 2 hr so that the CLP domains of each conjugate would be sufficiently Note: All peptides were measured in a 0.2 cm cuvette, in HPLC grade water (pH 6.5) at a concentration of 0.35 mM, with a scanning rate of 10 C/hr. process is significantly more facile than other vesicle formation techniques that require specific pH aqueous environments, 6 organic to aqueous solvent exchanges, 8 or specific instrumentation (e.g., spray drying). 73 The hydrodynamic diameter distributions derived from raw autocorrelation decay functions for each conjugate sequence were recorded at each temperature throughout the cooling steps (Figures 2 and S18). The average hydrodynamic diameters for the vesicles formed from F6-GPO7 and F6-GPP10 at each temperature are provided in Table S1.
The data in Figure 2 demonstrate that the hydrodynamic diameter (D h ) distributions at 80 C for all the conjugates were approximately within the expected range of the estimated length of a single ELP-CLP subunit. This suggests that all of the conjugates adopted their monomeric form as nontriple helical sub-units at 80 C, which is consistent with our previous reports. 61,62 In the case of the F6-GPO7 and F6-GPP10 conjugates, the correlation delay times increased as cooling proceeded, indicating distributions with higher D h values, and suggesting the formation of vesicles (Figure 2c,d). The temperature at which the assembly began to occur correlated well with the measured T m values determined by CD ( Figure 1 and Table 2), corroborating that the triple helical folding of the CLP domain in the ELP-CLP conjugates is necessary for the self-assembly of these conjugates into vesicles.
The lack of any significant increases in the hydrodynamic diameter of the F6-GPO6 and F6-GFOGER conjugates as temperature decreased showed that these conjugates did not form vesicles (Figure 2a,b). We briefly note that the size distributions of the F6-GPO6 and F6-GFOGER at all temperatures appear to be modestly different from each other, though these subtle differences are likely not significant given the limitations of the characterization method.
These F6-GPO6 and F6-GFOGER conjugates possess significantly lower melting temperatures (by~20 C) than F6-GPO7 and F6-GPP10 but were still shown to be capable of forming triple helices (Table 2, Figures 1, and S15). As such, it was expected that some degree of assembly would occur around the melting temperatures for these conjugates, similar to what was observed at 50 C for F6-GPO7 and F6-GPP10 (Figure 2c,d). It was postulated that the self-assembly of the previously reported (VPGFG) 6 -(GPO) 4 GFOGER(GPO) 4 GG conjugate occurred due to the reduction in the inverse transition temperature and collapse of the ELP domain that was facilitated by the increase in the local concentration of ELPs upon formation of the CLP triple helix. 61 This rationale was made based on ample ELP literature, which has shown that the T t is a function of ELP concentration (e.g., a reduction in T t is a result of increased ELP concentration and vice versa). 54,74 However, based on our current observations of the F6-GPO6 and F6-GFOGER conjugates, which clearly fold into triple helices (Figures 1, S15, and S16) but yet do not form assembled structures at temperatures near or below their T m, we suggest that changes in the local concentration alone are not sufficient to drive assembly.
The data in this work indicate that the differing triple helical gate. 76 We find via DLS and TEM that the F6-GFOGER conjugate assembles into vesicles when cooled to approximately its T m (Table 2) in the presence of 100 mM NaCl ( Figures S19 and S20) shifts sufficiently such that self-assembly can occur. In the context of this work, we speculate that this critical T m is between 32 and 50 C.

| Dual thermoresponsiveness and morphological characterization of F6-GPO7 vesicles
The two conjugates (F6-GPO7 and F6-GPP10) which formed vesicles in water were monitored for an extended period of time at 25 C via DLS to ensure that the formed aggregates were colloidally stable over the timescales of the reported experiments. Continuous DLS analysis of the F6-GPO7 vesicle solution was carried out over a 34-hr period at 25 C, and throughout that time the hydrodynamic diameter, as well as the attenuation-corrected photon count rate, remained relatively constant and the vesicles in solution appeared homogenous, well dispersed, and stable ( Figure S21).
To ensure that there was no significant hysteresis in assembly and disassembly, the F6-GPO7 vesicles were heated over the same temperature range at which they were formed. Figure 3a shows the corresponding D h distributions throughout the heating process. Similar to the cooling distributions in Figure 2c, the heating distributions showed intact vesicles present at 25 and 35 C, but that there is a significant decrease in D h at the CLP melting temperature of 50 C. With  Figure S22b) was similar to that of the observed dissociation of vesicles at temperatures greater than 50 C (Figure 3a and Table S1), suggesting an ELP-mediated dissociation event was taking place since: (a) the CLP triple helix is still folded at these temperatures ( Figure 1c) and (b) the change was rapid (within a 5 C window), which is consistent with the rapid nature of ELP coacervation and resolubilization. 37 The reversibility of this transition was assessed via UV-Vis analysis of the turbidity of the vesicle dispersion over a minimum of 7 cycles; the results shown in Figure 3c indicated that upon cooling to 0 C, the normalized turbidity of the vesicles diminishes to that of water (bottom image in Figure 3c); meanwhile, when the solution was reheated to 25 C, the solubilized triple helical F6-GPO7 conjugates readily reassembled to their original turbidity (top image in Figure 3c). These data highlight the reversibility and speed of the vesicle dissolution/formation event, consistent, as noted above, with the expected phase transition of the ELP domain. 37,74 The apparent T t of the F6-GPO7 conjugate was also assessed via these turbidity measurements, 37 vesicles as the temperature was cooled from 25 to 0 C and then subsequently reheated to 25 C after a short holding period of~3 min at each end-point temperature. Seven cycles in total were performed confirming the robust reversibility of the apparent F6-GPO7 T t phenomena.
(d) Normalized turbidity of F6-GPO7 vesicles that was measured as a function of a 1 C/min cooling rate in order to determine the T t which was defined as the temperature in which 50% of maximal turbidity is reached (marked by dashed gray lines) was slightly lower than all other temperatures (Figure 3b). Given the sensitivity of DLS measurements and the shift in the size distribution, it is possible that vesicles are present at 10 C, but at low concentrations. 78 Table 2). At 80 C, the data in Figure 4d suggest a lack of vesicles likely due to complete unfolding of the triple helix and subsequent vesicle dissociation. The apparent differences in morphology between the solubilized conjugates in Figure 4a and Figure 4d likely  86,87 and can occur owing to differences in population sample sizes 88 and particle hydration. 89 In contrast to the F6-GPO7 vesicles that showed colloidal stability over a 24-hr period ( Figure S21), DLS of the F6-GPP10 nanoparticles indicated instability of the particle solution, showing distinct features of micron-scale aggregation after 24-hr incubation at 25 C, and ultimately precipitation ( Figure S25).

| Turbidity analysis of F6-GPP10 aggregates
Despite the fact that the F6-GPP10 conjugate precipitated with extended incubation, the temperature-responsive properties of the F6-GPP10 aggregates were assessed through turbidity measurements as a function of temperature. Starting with F6-GPP10 conjugates in the aggregated state ( Figure S25), the F6-GPP10 aggregate solution was stirred (to suspend the aggregates) and heated from 25 to 80 C, crossing the T m threshold of the (GPP) 10 CLP (Table 1 and Figure S17). The turbidity of the suspended aggregates remained relatively constant throughout the heating (Figure 5a), until the temperature reached~60 C, at which point the turbidity began to decrease rapidly until the solution was clear at temperatures greater thañ 75 C due to dissolution of the aggregates. Notably, the onset of the dissociation (at~60 C) of the F6-GPP10 conjugate occurred at temperatures that were significantly higher than the T m of the N 3 -(GPP) 10 GG CLP (Table 1 and Figure S17), in contrast to the relatively small difference between the conjugate and CLP T m values observed for the other ELP-CLP conjugates in this study (Tables 1 and 2). This likely occurred both because the T m of the F6-GPP10 conjugate was higher than the T m of the N 3 -GPP10 by itself (following the trend of the other conjugates), and also because of the more rapid heating rate used in this turbidity experiment relative to the slow heating rates used in the CD experiment. Although the turbidity data do not provide a precise T m for the F6-GPP10 conjugate aggregates, they do confirm that folding of the CLP is required for aggregation/assembly to occur.
At the end of the hyperthermal dissolution experiment, the F6-GPP10 was then cooled and monitored for aggregate formation via turbidity measurements. Detailed analyses of the re-aggregation of the F6-GPP10 conjugate and the kinetics of aggregation as well as the morphological characterization of these aggregates is discussed in Supporting Information Section 1.8 and presented in Figure S26. Following the kinetics study and reprecipitation, the F6-GPP10 aggregates were again suspended with stirring and continuously monitored for turbidity while being cooled from 25 to 0 C to determine if there was a T t -mediated dissolution for the F6-GPP10 conjugate. The data for this experiment are presented in Figure 5b, which shows a continuous unperturbed turbidity signal across the entire temperature window, indicating that the T t of the F6-GPP10 conjugate was less than 0 C.
The highly different assembly/aggregation behavior, despite nearly identical estimated T m values, between the F6-GPO7 and F6-GPP10 conjugates, suggests that contributions from the final hydration states of the trimerized CLPs can also play an important role in controlling the coacervation behavior of the ELP domain, consistent with the fact that (GPO)-containing CLPs are moderately more hydrated relative to (GPP)-containing CLPs. 90 It previously has been reported that an ELP conjugated with a more hydrophilic moiety (such as N-methyl-1,6-OH 1,4,5,6-tetrahydronicontinamide) will exhibit a higher T t than the same ELP conjugated to a less hydrophilic moiety (such as N-methyl-1,6-OH 1,4,5,6-tetrahydronicontinamide). 74  ORCID Lucas C. Dunshee https://orcid.org/0000-0001-6025-5222