Hydrolytically Stable Maleimide-end Functionalized Polymers for Site-Specific Protein Conjugation

Abstract

Site-specific conjugation to cysteines of proteins often utilizes ester groups to link maleimide or alkene groups to polymers. However, the ester group is susceptible to hydrolysis, potentially losing the benefits gained through bioconjugation. Here we present a simple conjugation strategy that utilizes the amide bond stability of traditional 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide/N-hydroxysuccinimide coupling while introducing site-specificity. Hydrolytically stable maleimide-end functionalized polymers for site-specific conjugation to free cysteines of proteins were synthesized using reversible addition-fragmentation chain transfer (RAFT) polymerization. The polymers’ alpha terminus was amidated with a furan-protected aminoethyl maleimide using carbodiimide based chemistry Finally, the maleimide was exposed by a retro Diels-Alder reaction to yield the maleimide group allowing for thiol-maleimide click chemistry for bioconjugation. A thermophilic cellulase from Fervidobacterium nodosum (FnCel5a) was conjugated utilizing various strategies including random EDC/NHS coupling, site-specific hydroxyethyl maleimide (HEMI) end functionalized coupling, hydroxyethyl acrylate (HEA) end functionalized coupling, and amidoethyl maleimide (AEMI) end functionalized coupling. Only the polymers conjugated by EDC and AEMI remained conjugated a week after attachment. This indicates that hydrolytically stable amide based maleimides are an important bioconjugation strategy for conjugates that require long term stability, while esters are better suited to systems which require the debonding of the polymer over time.

Graphical Abstract

INTRODUCTION.

Biological macromolecules, such as proteins, have been utilized for industrial and biomedical applications such as gene delivery,¹ commercial detergents,² disease treatment,^{3, 4} and biofuel synthesis.⁵ Synthetic polymer conjugation to these target proteins have been shown to tune protein function and stability. Protein-polymer conjugation is traditionally accomplished by attaching polymers to the amino acid side chains using simple organic chemistry coupling. This was first introduced in 1977 by Abuchowski and coworkers, where the group covalently attached polyethylene glycol (PEG) to bovine serum albumin (BSA), work that laid the foundation for a field that has revolutionized the use of polymers.⁶ Many researchers have used pegylation to increase protein activity, proteolytic resistance, pH stability, and thermal stability.^7-12 Though pegylation is still utilized in protein-polymer bioconjugation, rationally designed synthetic polymers with functional groups able to electrostatically interact with substrates have gained a lot of attention.

The choice of bioconjugation methods is dependent on the amino acids available on the surface of the protein of interest and whether site-specificity is required.¹³ Protein-polymer conjugates can be synthesized using grafting-to or grafting-from approaches. Grafting-to refers to the attachment of a synthesized polymer to the protein while grafting from refers to attaching an initiator or chain transfer agent to the protein and subsequently growing a polymer directly from the protein in the presence of monomers.¹⁴ Coupling chemistries often utilize lysine, cysteine, tyrosine, and arginine residues creating amide,^15-17 thiol-Michael,¹⁸ ester,^19-21 or disulfide bonds^{22, 23} linking the polymer to the protein. Site specific conjugation allows researchers to work with homogeneous samples,²⁴ have more accuracy with conjugation,²⁵ and create therapeutic agents for biological and therapeutic applications.²⁶

Although protein-polymer bioconjugation has many benefits for certain proteins, for other proteins the impacts are unacceptable – resulting in significant loss of activity or function, especially if the conjugated residues are close to the active site of the protein.²⁷ Optimizing pharmacological properties, such as circulation half-time in the body, while maintaining protein activity and function is a serious challenge faced in therapeutics and biocatalysis.^{28, 29} Rational design of therapeutic and catalytic bioconjugates that could prolong the life-time in vivo while allowing for polymer dissociation as the enzyme reaches its target would be an ideal approach to this challenge.^{30, 31} In particular, two of the most common conjugation linkers, esters and amides, offer access to control over this timescale of conjugation, with esters undergoing hydrolysis 1-2 orders of magnitude faster than amides.³² Esters and amides often appear in polymer backbones, or in linking fragments between attachment sites and the conjugated molecule.¹³ Therefore, choice of the linking groups could afford control over the lifetime of the conjugated polymer. Prior work has shown that linker choice and subsequent environments can impact bioconjugate stability, for instance the maleimide attachment group can be stabilized against retro-Michael reactions by ring-opening the maleimide,^{33, 34} while polyester backbones and crosslinks can hydrolyze to facilitate degradation of the polymer.^35-37 In contrast, amide bonds are substantially more hydrolytically stable, with estimates of hydrolytic half-lifetimes of the amide bond used in protein backbones estimated to be over 100 years for uncatalyzed systems.^{38, 39} However, in the context of bioconjugates, to the best of our knowledge there is no systematic study of how different linkers impact the attachment lifetime of the bioconjugate. If the linker degrades over a timeframe of days under storage conditions, then bioconjugates must be used almost immediately. Therefore, it is necessary to evaluate bioconjugate stability against decoupling of the synthetic and biological components of the hybrid molecule.

Herein, a thermophilic cellulase, FnCel5a, is modified with synthetic polymers using grafting-to and chain extension approaches to better understand the effect of four conjugation strategies on enzymatic activity and polymer hydrolytic stability. The focus of stability studies is at neutral pH of 7 and at 4 °C as this represents typical storage conditions for many bioconjugates, evaluating how different linkers impact the time over which the conjugate can be stored or must be used. In particular, a linking strategy based on amidoethyl maleimide (AEMI), where the amide group is bonded to the polymer, can be used to site specifically modify proteins. Maleimide conjugation using ester free linkers creates a simple conjugation strategy that combines amide bond stability with bioconjugate site-specificity.

RESULTS AND DISCUSSION.

A series of biocompatible, hydrophilic polymers were synthesized for conjugation to FnCel5a. These polymers were used as models to explore various conjugation chemistries and their hydrolytic stability. These well-defined polymers were synthesized by reversible addition-fragmentation chain-transfer (RAFT) polymerization as shown in Scheme 1.

Scheme 1:

(A) Synthesis of poly(Dimethylacrylamide) using RAFT polymerization; (B) Structure of all monomers and co-monomers used in this study

A homopolymer targeting 30 units of dimethyl acrylamide (DMAm), a co-polymer targeting 30 total units with ratios of 4:1 DMAm/DMAEMA (dimethyl acrylamide/ 2-(N,N-dimethylamino)ethyl methacrylate), and a homopolymer targeting 8 units of OEOA (poly(ethylene glycol) methyl ether acrylate) were created. In this study, we compare the four conjugation strategies where poly(DMAm), poly(DMAm/DMAEMA), and poly(OEOA) were linked to hydroxyethyl acrylate (HEA), hydroxyethyl maleimide (HEMI), and amidoethyl maleimide (AEMI) for thiol-Michael bioconjugation to FnCel5a. Modified and unmodified poly(DMAm), poly(DMAm/DMAEMA), and poly(OEOA) were characterized using NMR and GPC, data can be found in Figure S1 and Table S1.

Synthesis of HEA-poly(DMAm), HEA-poly(DMAm/DMAEMA), and HEA-poly(OEOA) only required a one-step reaction due to the availability of HEA as an reagent and acrylates being less reactive Michael acceptors when compared to maleimides.⁴⁰ HEA was coupled to the carboxylic acid end groups of the targeted polymer using 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and 4-dimethylaminopyridine (DMAP) through Steglich esterification in which an ester bond was formed between HEA and the targeted polymer. This is demonstrated in Scheme 2A. Furan-protected hydroxyethyl maleimide (Fp-HEMI) and furan-protected aminoethyl maleimide (Fp-AEMI) were employed to synthesize maleimide chain end polymers for thiol-maleimide conjugation. Fp-HEMI and Fp-AEMI were coupled to the carboxylic acid of the targeted polymer using EDC and DMAP to create ester and amide bonds, respectively, between the protected maleimides and polymers. The maleimide end groups in HEMI-capped polymers and AEMI-capped polymers were protected by furan during synthesis but easily deprotected using retro Diels-Alder chemistry at elevated temperature prior to conjugation to FnCel5a.⁴¹ The furan protecting group minimizes the likelihood of the electrophilic maleimide reacting with nucleophiles during the synthetic processes. This is demonstrated in Scheme 2B and 2C, respectively. Synthesis of Fp-HEMI and Fp-AEMI are demonstrated in Scheme S1.

Scheme 2:

Synthesis of (A) HEA-poly(DMAm), (B) AEMI-poly(DMAm), and (C) HEMI-poly(DMAm). Similar approaches were used for OEOA and DMAm/DMAEMA polymers.

Each polymer was conjugated to FnCel5a using two conjugation strategies. Poly(DMAm), poly(DMAm/DMAEMA), and poly(OEOA) were conjugated to FnCel5a using EDC-N-hydroxysuccinimide (NHS) coupling as demonstrated in Scheme 3A. FnCel5a has 26 naturally occurring lysine residues and an amine on the N-terminus, allowing for amidation at any of these 27 amine positions. The ratio of amine to polymer used in all conjugation reactions was 1:20 (reactive side chain : polymer).

Scheme 3:

Attachment of (A) poly(DMAm), (B) HEA-poly(DMAm), (C) HEMI-poly(DMAm), and (D) AEMI-poly(DMAm) to FnCel5a (blue) cellulase enzyme. Similar approaches were used for OEOA and DMAm/DMAEMA polymers.

FnCel5a has 2 naturally occurring cysteine residues, at positions 102 and 296. To determine whether conjugation would occur site-specifically or randomly, the cysteine residues were removed individually by mutation of the cysteine to a serine. For the C102S or the C296S mutants, conjugation was attempted, and polyacrylamide gel electrophoresis (PAGE) was used to identify if conjugation was successful. We used the crystal structure of FnCel5A from F. nodosum Rt17-B1(Protein Data Bank Identification: 3NCO) to calculate the solvent accessible surface area using GETAREA.^35,42 It was found that Cys296 has 1.75 Ų solvent exposed surface area for the side chain while Cys102 is completely buried in the protein with no solvent accessible surface area. Calculated sidechain solvent exposed surface area for these residues can be found in Table S2. Conjugation tests of HEA-poly(DMAm), HEMI-poly(DMAm), and AEMI-poly(DMAm) to FnCel5A_C102S and FnCel5A_C296S (Figure S2) found successful conjugation to FnCel5A_C102S but not to FnCel5A_C296S indicating that, as predicted by GETAREA solvent accessible surface area calculations, cysteine coupling to FnCel5A occurs specifically at Cys296. HEA-poly(DMAm) was conjugated to FnCel5a using thiol-Michael click chemistry as demonstrated in Scheme 3B. HEMI-poly(DMAm) and AEMI-poly(DMAm) were conjugated using thiol-Maleimide click chemistry as demonstrated in Scheme 3C and 3D.

PAGE was conducted for samples of all FnCel5a bioconjugates taken initially following conjugation and taken daily over the course of a week, where samples were incubated at 4 °C and pH 7. The choice of 4 °C and pH 7 (20mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 50mM sodium chloride) for incubation was due to this being typical conditions used to store bioconjugates in a refrigerator. Studying the stability of the protein-polymer connection at 4 °C and pH 7 determines the bioconjugate’s lifetime under common storage conditions. Initially, conjugation is confirmed using this method through an increase in the apparent molecular weight evidenced by smearing caused by dispersity of polymers conjugated to the protein and hydrodynamic radius of polymers. Dissociation of the polymer from the protein is observed through the return of bands at the molecular weight of unmodified, wild-type FnCel5a and a decrease in high apparent molecular weight streaking that is commonly associated with bioconjugation, as demonstrated in Figure 1. The return to the molecular weight of the unmodified wild-type FnCel5a, as determined by PAGE, is most likely due to the fact that hydrolysis of the polymer only leaves a small organic fragment bound to the protein at the attachment site. This is consistent with the idea that certain types of ester, such as the thioether ester from thiol-acrylate conjugation, are particularly susceptible to hydrolysis.⁴³ Degradation of the carbon-carbon bonds in the polymer backbone are unlikely, since only the systems containing the ester links showed the reduction in bioconjugate apparent molecular weight. In contrast, backbone degradation would show a reduction of bioconjugate apparent molecular weight, regardless of the chosen linking group. The similarity of the wild type and HEA and HEMI systems after full degradation is due to any remaining organic fragment having a minimal impact on the overall protein’s molecular weight and size in solution and this small change in molecular weight. Therefore, such small organic fragments would not yield a difference in apparent molecular weight that is resolvable by PAGE.

Figure 1:

PAGE gel of FnCel5a and poly(DMAm)₃₀-FnCel5a bioconjugates made using (A) EDC/NHS coupling of poly(DMAm), (B) thiol-Michael coupling of HEA-poly(DMAm), (C) thiol-maleimide coupling of HEMI-poly(DMAm), and (D) thiol-maleimide coupling of AEMI-poly(DMAm). Lanes: 1, ladder; 2, wild-type FnCel5a; 3, Initial Conjugation; 4, Conjugation after 1 day; 5, Conjugation after 2 days; 6, Conjugation after 3 days; 7, Conjugation after 4 days; 8, Conjugation after 5 days; 9, Conjugation after 6 days; 10, Conjugation after 7 days. Samples were incubated at 4 °C and pH 7.

Although in principle retro-Michael reactions could disconnect the polymer from the protein, the comparison of the AEMI-based conjugates and the HEMI-based conjugates suggests that disconnections between the protein and the polymer are unlikely to be occurring through retro-Michael reactions. Both HEMI and AEMI-based conjugates have essentially the same thiol-maleimide linkage connecting the protein and the maleimide. However, HEMI uses the more hydrolytically sensitive ester group to connect the polymer to the maleimide, while AEMI uses the more hydrolytically stable amide linkage. Since decoupling of the polymer and the protein is only observed for the HEMI not the AEMI based conjugates, this suggests disconnection of the polymer and protein occurs predominantly through the hydrolysis of the ester not the retro-Michael reaction.

In all cases, essentially no unmodified enzyme was present immediately after each conjugation reaction; suggesting that free enzyme is unlikely to have a significant effect on interpreting bioconjugate activity. Bioconjugates synthesized using EDC/NHS coupling showed no significant hydrolysis or differentiation in apparent molecular weight over the course of a week. Thiol-Michael coupling using HEA-poly(DMAm) exhibited essentially complete hydrolysis over the course of one week while thiol-maleimide conjugation using HEMI-poly(DMAm) exhibited very significant hydrolysis over the course of one week. This is likely a result of ester hydrolysis, allowing the small molecule HEA and HEMI fragments to remain linked to Cys296 while the polymer dissociates. Bioconjugates synthesized using AEMI-poly(DMAm) performed similarly to EDC/NHS coupling, with no significant hydrolysis observed over the course of one week. Similar results are shown using poly(DMAm/DMAEMA) and poly(OEOA)-based bioconjugates in Figure S3 and Figure S4, respectively. Additionally, a shorter chain length of 10 units of DMAm was used, and as seen in Figure S5 the result of the poly(DMAm)₁₀ were similar results to the longer poly(DMAM)₃₀ system. In all cases loss of the bioconjugate was observed over the course of 1 week for the HEMI and HEA based linkers, with faster decoupling of the polymer for HEA. There was essentially no loss of the bioconjugate for the amide-based EDC coupled system and the AEMI based systems regardless of the polymer structure or the chain length. This indicates that the dissociation of the polymer from the protein is not caused functional groups and chain length of the polymer but rather due to the linkage between the polymer and the protein.

The FnCel5a-poly(DMAm) bioconjugates synthesized in this study were evaluated for enzymatic activity, as shown in Figure 2. Soluble carboxymethylcellulose (CMC) was used as a substrate in place of insoluble cellulose for the enzymatic assays at optimal conditions for FnCel5a, 81 °C and pH 5.^{44, 45} The colorimetric assay used in this study measures the rate at which wild-type FnCel5a, FnCel5a-poly(DMAm₃₀) and FnCel5a-poly(DMAm₁₀) bioconjugates, FnCel5a-poly(DMAm₂₄/DMAEMA₆) bioconjugates, and FnCel5a-poly(OEOA₈) bioconjugates degrade CMC. The resulting free reducing ends of carboxymethyl glucose reduced 3,5-dinitrosalicyclic acid (DNSA) to 3-amino-5-nitrosalicylic acid (ANSA), enabling quantification of ANSA concentration due to ANSA’s absorbance at 530nm. As shown in earlier work, conjugation of poly(DMAm) and poly(DMAm/DMAEMA) to FnCel5a led to a statistically significant increase in enzymatic activity.⁴⁵

Figure 2:

Relative activity of (A) poly(DMAm₃₀), (B) poly(DMAm₁₀), (C) poly(DMAm₂₄/DMAEMA₆), and (D) poly(OEOA₈) bioconjugates over the course of 7 days of incubation at pH 7 and 4°C compared to wild-type, unmodified FnCel5a. Relative activity of bioconjugates made using EDC/NHS coupling (orange circle), thiol-Michael coupling of HEA-polymers (green square), thiol-maleimide coupling of HEMI-polymers (blue triangle), and thiol-maleimide coupling of AEMI-polymers (purple triangle). Each bar represents the standard deviation from the triplicate experiment.

Bioconjugates synthesized using EDC/NHS coupling showed no significant change in activity over the course of one week after conjugation. Initially, as shown in Figure 2A and B, the FnCel5a-poly(DMAm₃₀) and FnCel5a-poly(DMAm₁₀) bioconjugates has an activity of 161 ± 8% and 112 ± 12% respectively, relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 165 ± 10% and 112 ± 8%, respectively. As shown in Figure 2B, the FnCel5a-poly(DMAm₂₄/DMAEMA₆) bioconjugate has an activity of 149 ± 3% relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 143 ± 5%. As shown in Figure 2C, the FnCel5a-poly(OEOA₈) bioconjugate has an activity of 75 ± 6% relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 78 ± 8%. Thiol-Michael coupling using HEA-poly(DMAm) and HEA-poly(DMAm/DMAEMA) showed a very significant loss in activity within 4 days. Initially, the FnCel5a-HEA-poly(DMAm₃₀) and FnCel5a-HEA-poly(DMAm₁₀) bioconjugates exhibit an activity of 156 ± 11% and 111 ± 5%, respectively, relative to the wild-type sample, while on the seventh day after the bioconjugation the activity was 102 ± 10% and 99 ± 9%, respectively, which is statistically equivalent to the activity of wild-type FnCel5a. Additionally, FnCel5a-HEA-poly(DMAm₂₄/DMAEMA₆) bioconjugates showed an activity of 151 ± 2% relative to the wild-type sample upon initial conjugation, while by the seventh day after bioconjugation the activity was 105 ± 7%. Interestingly, FnCel5a-HEA-poly(OEOA₈) bioconjugates showed restoration of lost activity upon detachment of the polymer. Initially, the FnCel5a-HEA-poly(OEOA₈) bioconjugates exhibit an activity of 76 ± 4% relative to the wild-type sample, while on the seventh day after the bioconjugation the activity was 99 ± 6% .

Further, bioconjugates synthesized by thiol-maleimide coupling using HEMI-poly(DMAm₃₀) and HEMI-poly(DMAm₂₄/DMAEMA₆) exhibited substantial loss in activity over the course of a week. Initially, the FnCel5a-HEMI-poly(DMAm₃₀) and FnCel5a-HEMI-poly(DMAm₁₀) bioconjugates has an activity of 164 ± 9% and 111 ± 10%, respectively, relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 116 ± 8% and 99 ± 7%, respectively. Similarly, the FnCel5a-HEMI-poly(DMAm₂₄/DMAEMA₆) bioconjugates has an activity of 146 ± 2% relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 101 ± 7%. FnCel5a-HEMI-poly(OEOA₈) bioconjugates has an activity of 80 ± 5% relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 97 ± 4%.

Thiol-maleimide coupling using AEMI-poly(DMAm₃₀) demonstrated retention of activity over the course of a week after conjugation. Initially, the FnCel5a-AEMI-poly(DMAm₃₀) and FnCel5a-AEMI-poly(DMAm₁₀)bioconjugates has an activity of 160 ± 7% and 113 ± 9%, respectively, relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 165 ± 8% and 112 ± 7%, respectively. FnCel5a-AEMI-poly(DMAm₂₄/DMAEMA₆) bioconjugates has an activity of 149 ± 4% relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 144 ± 12%. FnCel5a-AEMI-poly(OEOA₈) bioconjugates has an activity of 75 ± 5% relative to the wild-type sample, while on the seventh day after the bioconjugation has an activity of 77 ± 6%.

To be more relevant to human physiology FnCel5A bioconjugates (FnCel5A-poly(DMAm₃₀), FnCel5A-HEA-poly(DMAm₃₀) and FnCel5A-AEMI-poly(DMAm₃₀)) were also incubated at 37 °C and pH 7 (20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 50 mM sodium chloride) over the course of 1 week. Similar results were also shown here, where bioconjugation was lost for HEA based linker after 1 week of incubation with no essential loss of bioconjugation for the amide-based EDC-coupled system and AEMI based system (Figure S6). The activity assay data also exhibited a similar pattern, where HEA-based bioconjugate was potentially losing activity after 3 days, amide-based EDC-coupled system and AEMI based system were able to retain activity for 5 days (Figure S7).

The polymers used for all FnCel5a bioconjugates in this study offer variation in molecular weight, polymer geometry, charge, and hydrogen bonding ability. The decrease in activity coinciding with ester hydrolysis of bioconjugates synthesized through thiol-Michael and thiol-maleimide attachment, using HEA and HEMI respectively, is an interesting trend that suggests polymer attachment is necessary to significantly impact enzymatic activity. Polymer detachment after conjugation returns the enzyme to the properties of the wild-type FnCel5a, suggesting the polymer conjugation is responsible for the increase in measured catalytic activity for poly(DMAm) and poly(DMAm/DMAEMA) bioconjugates. This may be due to poly(DMAm) and poly(DMAm/DMAEMA) recruiting the CMC substrate towards the cellulase, increasing the local concentration of substrate and thereby increasing the CMC hydrolysis rate.⁴⁵ This newly introduced method for bioconjugation, thiol-maleimide coupling using AEMI-capped polymers takes advantage of the amide bond linking AEMI to the targeted polymer, similar to that observed in EDC/NHS coupling, while the maleimide chain end allows for the same level of site-specific conjugation that is found in thiol-Michael coupling of HEA-capped polymers.

To evaluate the scope and limits of this method, we attempted to extend the polymer chain of the existing bioconjugate as shown in Scheme 4. RAFT polymerization is a reversible deactivation radical polymerization technique, in which chain growth can be reinitiated after a reaction has ended using the a reactive monomer, including new monomers to synthesize block copolymers.^46-48 In this experiment, FnCel5a-AEMI-poly(DMAm) served as a macro-CTA and underwent polymerization in the presence of DMAm monomer and VA-044 initiator. Initial [Macro-CTA]/[VA-044]/[DMAm] ratios of 1/0.3/300 were used. Chain extension was determined by comparing the apparent molecular weight of the sample to the initial FnCel5a-AEMI-poly(DMAm) bioconjugate as measured by PAGE (Figure 3A). Additionally, Figure 3B shows this chain extension resulted in a significant increase in enzymatic activity of the bioconjugate (Table S8).

Scheme 4:

Synthetic strategy for chain extension of FnCel5a-AEMI-poly(DMAm)

Figure 3:

(A) PAGE of chain extension with FnCel5a-AEMI-poly(DMAm). Lanes:1, ladder; 2, wild-type FnCel5a; 3, FnCel5a-AEMI-poly(DMAm); 4, FnCel5a-AEMI-Chain Extension, (B) relative activity of AEMI-poly(DMAm) and chain extension. Each bar represents the standard deviation from triplicate experiments.

CONCLUSION.

In summary, well-defined polymers of DMAm, DMAm/DMAEMA and OEOA were conjugated using various methods to FnCel5a. The resulting bioconjugates showed similarly significant increases in enzymatic activity for all conjugation methods. For bioconjugates in which the polymer was attached through an ester-based linker, hydrolysis over the course of a week resulted in significant polymer dissociation and the loss of enzymatic benefits gained by polymer conjugation. As these polymers dissociated, the enzymatic activity returned to activity ranges similar to the unmodified FnCel5a suggesting that polymer dissociation has no significant impact of the dissociated enzyme’s activity. For bioconjugates in which the polymer was attached to a linker or directly to the protein through an amide bond or imide and amide, the bond remained stable and there was no significant impact on enzymatic activity over 7 days. These results suggest that the bioconjugation strategy used when synthesizing protein-polymer conjugates is important and that polymer dissociation kinetics can be tuned through the choice of linker. Furthermore, we have developed a straightforward method to synthesize AEMI-poly(DMAm) with a maleimide end group for conjugating to proteins in a hydrolytically-stable, site-specific manner.

EXPERIMENTAL SECTION

Materials

N,N-Dimethylacrylamide (DMAm), furan, ethylenediamine, and 2-hydroxyethyl acrylate (HEA) were obtained from TCI. Azobis(isobutyronitrile) (AIBN), anhydrous methanol, and ethanethiol were purchased from Sigma-Aldrich. Potassium hydroxide, carbon disulfide, and ethanolamine were obtained from Fisher Scientific. 2-Bromopropionic acid and 4-dimethylaminopyridine (DMAP) were purchased from Acros Organic. Maleic anhydride was purchased from Alfa Aesar and 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) was purchased from Carbosynth.

Lithium dodecyl sulfate (LDS), terrific broth (TB), lennox broth (LB), magnesium chloride (MgCl₂), 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), sodium chloride (NaCl), glucose, sodium acetate, calcium chloride hexahydrate, Tween 20, and GelCode Blue were purchased from Thermo-Fisher Scientific. Antifoam was purchased from Sigma-Aldrich, carboxymethylcellulose (CMC, average molecular weight of 200 000) was purchased from Millipore Corp, and HisTrap™ nickel affinity column was purchased from GE Healthcare. Carbenicillin, DNAse I, 3-(N-morpholino)propanesulfonic acid (MOPS), Lysozyme, 4-benzenesulfonyl fluoride hydrochloride (AEBSF) were purchased from Gold Biotechnology. FnCel5a plasmid was purchased from GenScript and BL21 (DE3) Escherichia coli cells were obtained from New England Biolabs. Polyacrylamide gel electrophoresis (PAGE) was done using Bio-Rad Laboratories TGX Fast Cast 12% Acrylamide Kit. All materials were used without further purification unless otherwise specified.

Methods

Cloning, Expression, and Purification of FnCel5a

A plasmid encoding for a N-terminal His-tag, a protease cleavage site for TEV, followed by residues 11-320 of FnCel5a was produced by cloning FnCel5a into a pET-15b vector. FnCel5a plasmid was then cultured using BL21(DE3) Escherichia coli cells, inoculated in LB, and grown in TB. The cells were then harvested and resuspended in buffer (20mM HEPES, 150 mM NaCl, 1 mg/mL lysozyme, 20 μg/mL DNAseI, and 100 μg/mL AEBSF), frozen in liquid nitrogen, and thawed overnight. Cell lysate was centrifuged and supernatant was purified using filtration, heating, and Fast Protein Liquid Chromatography (HisTrap nickel affinity column), FnCel5a was eluted in 20mM HEPES, pH 7.5, 50mM NaCl and 500mM imidazole. Eluted enzyme was dialyzed in the presence of TEV to remove the N-terminal His-tag and excess imidazole, followed by Fast Protein Liquid Chromatography prior to use for polymer conjugation. FnCel5a purity was confirmed by polyacrylamide gel electrophoresis.

Site-directed Mutagenesis

Mutagenic primers, purchased from IDT DNA, was used with a QuikChange II XL Site-directed mutagenesis kit, purchased from Agilent Technologies. A forward primer, reverse primer, dNTP, QuikSolution reagent, molecular grade water, and PfuUltra HF DNA polymerase were mixed with FnCel5a and underwent three PCR cycles, was treated with DpnI restriction enzyme, immediately incubated, and transformed into NEB5α competent cells. The resulting plasmid was isolated and shipped to Eurofins Genomics for sequence confirmation. The confirmed plasmid was cultured using BL21(DE3) Escherichia coli cells and protein was expressed and purified as mentioned above.

Polyacrylamide gel electrophoresis (PAGE) Analysis

Purified FnCel5a samples from FPLC were denatured using LDS and heat. After heating, the sample was separated according to apparent molecular weight by electrophoresis (100V for 50 minutes) using a 12.5% polyacrylamide gel. The apparent molecular weight of purified protein samples were determined by comparison of migration distance as compared to a Precision Plus Protein Kaleidoscope molecular weight marker, purchased from Bio-Rad. Following electrophoresis, the gel was submerged in GelCode Blue Stain, heated in the microwave for 30 seconds, stained overnight, and washed in water for 3 hours to remove excess stain.

Synthesis of PAETC

Reversible addition-fragmentation chain-transfer (RAFT) polymerization utilizes the chain transfer agent (CTA) 2-(((ethyl-thio)-carbonothioyl)thio)propanoic acid (PAETC) to synthesize polymers.⁴⁶ The synthesis of PAETC followed the method reported in the literature. ⁴⁶

RAFT Polymerization of dimethylacrylamide polymer of chain length 30.

RAFT polymerization was utilized to synthesize poly(dimethylacrylamide) (pDMAm) with a targeted chain length of thirty repeating units and proceeded as shown in Scheme 1. DMAm (21.2277g, 214.14mmol), PAETC (1.5622g, 7.43mmol), and AIBN (0.1165g, 0.71mmol) were dissolved in ethanol (25mL), deoxygenated using nitrogen gas, and the reaction was stirred in a 65 °C oil bath for 15 hours. Conversion of monomer to polymer was determined using NMR. After polymerization was confirmed, the polymer was precipitated using ice-cold hexane. Excess solvent was removed using centrifugation and the polymer was dried in a vacuum oven overnight. The polymer was characterized by NMR and Gel Permeation Chromatography (GPC). ⁴⁵

RAFT Polymerization of dimethylacrylamide/2-(dimethylamino)ethyl methacrylate polymer of chain length 30.

RAFT polymerization was utilized to synthesize poly(dimethylacrylamide/2-(dimethylamino)ethyl methacrylate) (pDMAm/DMAEMA) with a targeted chain length of thirty repeating units (twenty four units of DMAm and six units of DMAEMA) and proceeded as shown in Scheme 1. DMAm (11.3125g, 114.12mmol), DMAEMA (4.4932g, 28.58mmol), PAETC (1.0333g, 4.91mmol), and AIBN (0.0742g, 0.45mmol) were dissolved in ethanol (20mL), deoxygenated using nitrogen gas, and the reaction was stirred in a 65 °C oil bath for 15 hours. Conversion of monomer to polymer was determined using NMR. After polymerization was confirmed, the polymer was precipitated using a (4:1) hexane/ diethyl ether mixture. Excess solvent was removed using centrifugation and the polymer was dried in a vacuum oven overnight. The polymer was characterized by NMR and Gel Permeation Chromatography (GPC). ⁴⁵

RAFT Polymerization of dimethylacrylamide polymer of chain length 10.

RAFT polymerization was utilized to synthesize poly(dimethylacrylamide) (pDMAm) with a targeted chain length of ten repeating units and proceeded as shown in Scheme 1. DMAm (4.72g, 47.63mmol), PAETC (0.996g, 4.74mmol), and AIBN (0.0788g, 0.48mmol) were dissolved in ethanol (7.5mL), deoxygenated using nitrogen gas, and the reaction was stirred in a 65 °C oil bath for 15 hours. Conversion of monomer to polymer was determined using NMR. After polymerization was confirmed, the polymer was precipitated using ice-cold hexane. Excess solvent was removed using centrifugation and the polymer was dried in a vacuum oven overnight. The polymer was characterized by NMR and Gel Permeation Chromatography (GPC). ⁴⁵

RAFT Polymerization of oligoethyleneoxide acrylate polymer of chain length 8.

RAFT polymerization was utilized to synthesize poly(oligoethyleneoxide) (pOEOA) with a targeted chain length of eight repeating units and proceeded as shown in Scheme 1. OEOA (11.429g, 19.05mmol), PAETC (0.508g, 2.42mmol), and AIBN (0.0392g, 0.24mmol) were dissolved in ethanol (15mL), deoxygenated using nitrogen gas, and the reaction was stirred in a 65 °C oil bath for 15 hours. Conversion of monomer to polymer was determined using NMR. After polymerization was confirmed, the polymer was precipitated using ice-cold hexane. Excess solvent was removed using centrifugation and the polymer was dried in a vacuum oven overnight. The polymer was characterized by NMR and Gel Permeation Chromatography (GPC). ⁴⁵

Synthesis of Fp-MAn

Furan-protected maleic anhydride (Fp-MAn) was synthesized through a Diels-Alder reaction between maleic anhydride and furan in solvent. The synthesis of Fp-Man was performed following protocols outlined in the literature.⁴¹

Synthesis of Fp-HEMI

The synthesis of furan-protected HEMI (Fp-HEMI) followed protocols established in the literautre.⁴¹

Synthesis of Fp-AEMI

The synthesis of furan-protected AEMI (Fp-AEMI) followed protocols established in the literature. ⁴⁹

Synthesis of HEA-poly(DMAm)

Poly(dimethylacrylamide), HEA, and DMAP were dissolved in methylene chloride and EDC was subsequently added to the solution. The reaction occurred at room temperature under consistent stirring for 2 hours. HEA-linked poly(DMAm) (HEA-poly(DMAm) was precipitated dropwise using an ice-cold hexane/ethyl ether mixture. HEA-poly(DMAm) was resuspended in water and dialyzed for 16 hours using 2 kDa MWCO dialysis cassettes to remove excess HEA. After dialysis, water was evaporated, and the resulting polymer was confirmed through NMR and utilized for protein bioconjugation. A similar approach was used for poly(OEOA) and poly(DMAm/DMAEMA). [Polymer]/[HEA]/[EDC]/[DMAP] ratios of 1/1.7/2/0.2 were used to synthesize HEA-polymers.

Formation of Maleimide Chain End poly(DMAm)

HEMI-poly(DMAm) and AEMI-poly(DMAm) were synthesized through a retro Diels-Alder reaction, allowing furan to dissociate and evaporate from the flask. Fp-HEMI-poly(DMAm) and Fp-AEMI-poly(DMAm) were dissolved in toluene, stirred, and heated to 110 °C in an oil bath for 8 - 10 hours. After reacting, the solution was allowed to cool to room temperature and the solvent was removed through rotary evaporation. The resulting polymers were confirmed through NMR and was utilized for protein bioconjugation.¹⁹ A similar approach was used for poly(OEOA) and poly(DMAm/DMAEMA). [Polymer]/[HEMI]/[EDC]/[DMAP] ratios of 1/1.3/1.3/0.1 were used to synthesize HEMI-polymers. [Polymer]/[AEMI]/[EDC]/[DMAP] ratios of 1/1.3/1.3/0.1 were used to synthesize AEMI-polymers.

Gel Permeation Chromatography

Poly(DMAm), HEA-poly(DMAm), HEMI-poly(DMAm), and AEMI-poly(DMAm) were characterized using gel permeation chromatography (GPC). The polymers were dissolved in N,N-dimethylformamide (DMF) with a drop of toluene and filtered through 0.2 μm PTFE filters before analysis. Size-exclusion chromatography (SEC) analysis utilized an Agilent 1260 gel permeation chromatography system equipped an isocratic pump, a degasser, an autosampler, a guard, 2 × Polargel-M columns, and a refractive index detector. To analyze the polymers, a flow rate of 1mL/min at 50 °C was used with DMF with 0.1 wt% LiBr as an eluent. This instrument was calibrated using poly(methyl methacrylate) standards.

Conjugation of Polymers to FnCel5a

Polymers were conjugated to FnCel5a using two methods which allowed for site-specific attachment to cysteine residues or random attachment to lysine residues. When attaching to lysine residues we utilized a NHS-EDC coupling reaction where the reactive carboxylic acid end group of the polymer were attached to the amine groups of FnCel5a. Coupling reactions were completed with FnCel5a, excess polymers, NHS, and EDC at pH 9 and were rotated at room temperature. [FnCel5a]/[Polymer]/[EDC]/[NHS] ratios of 1/540/54/5.4 were used for conjugation. The conjugation was quenched after 2 hours with glycine buffer and allowed to rotate to mix. The resulting protein-polymer bioconjugate was diluted with buffer in a 30 kDa MWCO PES Centricon centrifugal filtration devices and centrifuged to remove unreacted polymers and unwanted reaction products. Filtration was repeated until all excess polymer was removed and PAGE gel was used to confirm polymer attachment.

When attaching to cysteine residues we utilized thiol-Michael addition and thiol-maleimide addition reactions where the reactive vinyl group linked to the polymer attaching to the reactive thiol groups on FnCel5a. HEA-poly(DMAm), HEMI-poly(DMAm), and AEMI-poly(DMAm) were dissolved in pH 9 buffer, introduced to FnCel5a, and rotated at room temperature. After 2 hours, the resulting protein-polymer bioconjugate was diluted with buffer in a 30 kDa MWCO PES Centricon tubes and centrifuged to remove unreacted polymers. Filtration was repeated until all excess polymer was removed and PAGE gel was used to confirm polymer attachment.

Chain Extension

RAFT polymerization is a living polymerization technique that was utilized to extend the polymer chain of the existing protein-polymer bioconjugate.⁴⁶ VA-044 (0.0022g, 0.0068mmol) and DMAm (0.0075g, 0.075mmol) were dissolved in water and deoxygenated. FnCel5a-AEMI-poly(DMAm) (1mg, 2.44x10⁻⁵mmol), which is used as a macro-CTA during this reaction, was subsequently concentrated and added to the solution via syringe. The reaction was stirred in a 40 °C oil bath for 4 hours and the resulting protein-polymer bioconjugate was diluted with buffer in a 30 000 MWCO PES Centricon tubes and centrifuged to remove unreacted DMAm monomer and unwanted reaction products. Chain extension was determined using PAGE.

Hydrolytic Stability

Conjugations were performed at the same time under the same conditions every day for 7 days after the initial conjugation. Following conjugation, the bioconjugates were cleaned and the buffer was exchanged. The conjugates were stored at 4 °C, pH 7 over the course of a week and hydrolysis was determined using PAGE. As hydrolysis occurred and polymers dissociated from the bioconjugates, the apparent molecular weight of the conjugate decreases and the appearance of the wild-type protein begins to appear.

Cellulase Activity Assays

Activity of wild-type FnCel5a and resulting bioconjugates were determined using a colorimetric assay that measured the rate of degradation of carboxymethyl cellulose (CMC).⁵⁰ Glucose standards, wild-type FnCel5a samples, bioconjugate samples, and CMC samples were prepared in pH 5 buffer (100mM sodium acetate, 20mM calcium chloride, 10% Tween 20). Glucose standards, wild-type FnCel5a samples, bioconjugate samples, and CMC samples were preincubated at 81 °C in a Life Technologies Proflex 3 × 32-Well PCR System. CMC was then added to the glucose standards, enzyme samples, and bioconjugate samples after incubation. At each 2.5 min time point, the reaction was stopped using a DNSA solution (0.4 M 3,5-dinitrosalicylic acid, 0.4 M sodium hydroxide, 0.4 M potassium hydroxide, and 1 M potassium sodium tartrate tetrahydrate) and incubated at 90 °C for 10 minutes. Samples were then placed on ice for 5 minutes, plated on a 384-well plate, and absorbance at 530 was measured on a BioTek Synergy H1 microplate reader.