Protein-Protein Cross-Coupling via Palladium-Protein Oxidative Addition Complexes from Cysteine Residues

Abstract

Few chemical methods exist for the covalent conjugation of two proteins. We report the preparation of site-specific protein-protein conjugates that arise from the sequential cross-coupling of cysteine residues on two different proteins. The method involves the synthesis of stable palladium-protein oxidative addition complexes (Pd-protein OACs), a process that converts nucleophilic cysteine residues into an electrophilic S-aryl-Pd-X unit by taking advantage of an intramolecular oxidative addition strategy. This process is demonstrated on proteins up to 83 kDa in size and can be conveniently carried out in water and open to air. The resulting Pd-protein OACs can cross-couple with other thiol-containing proteins to arrive at homogeneous protein-protein bioconjugates.

Graphical Abstract

Bioconjugation reactions that can operate on proteins are recognized as important tools for probing structure-function relationships,^1,2 developing new therapeutics,^3,4 and providing total chemical synthesis routes.⁵ When modifying proteins, homogeneous products are often envisaged for application. Outside of enzymatic techniques, two methods for site-specific chemical modification of proteins are commonly employed that rely on: (1) installation of a non-canonical amino acid possessing a side chain with unique reactivity⁶ or (2) especially reactive natural amino acids that are present in low abundance. In the former strategy, click chemistry provides a means for the site-specific conjugation of both small and large molecules onto proteins.^7,8 Click reactions can be used to produce homogeneous products by exploiting orthogonal reactivity,⁹ but require the installation of the non-natural functional groups into each piece to enable the conjugation. The latter approach obviates the need to install two orthogonal, non-canonical reactive partners on separate components by chemoselectively targeting particularly reactive native amino acid residues. In this regard, cysteine side chains, owing to their low natural abundance and the unique reactivity of the thiol functional group, provide an attractive opportunity for site-specific modification of proteins.¹⁰ Within our ongoing program focused on Pd-mediated bioconjugation, we set out to devise a practical strategy for chemical bioconjugation of native proteins, allowing for both coupling partners to be comprised of only canonical amino acids while maintaining site-specificity.

In a Pd-catalyzed transformation, a suitable (pseudo)arylhalide substrate (III, Figure 1A) undergoes oxidative addition in the presence of Pd(0) (II) to form an oxidative addition complex (OAC) (IV). This electrophilic intermediate can be treated with a nucleophile, which, after reductive elimination, results in the formation of C–C or C–heteroatom bonds.¹¹ The synthesis of Pd OACs appended onto proteins would allow for their diverse functionalization. However, approaches to access electrophilic Pd-protein intermediates often rely on the use of a large excess of Pd reagent and nucleophilic coupling partner (e.g., 50 and 500 equivalents of Pd and nucleophile, respectively) to maximize the probability of the nucleophile capturing the transiently generated electrophilic Pd complex.^{7,12–13,14,15,16}

Figure 1.

An intramolecular approach to generate Pd-protein oxidative addition complexes from cysteine residues and their use toward protein-protein cross-coupling.

Recently, our groups have reported the use of pre-synthesized bench-stable electrophilic Pd-OACs derived from small molecules (<500 Da) for use in cross-coupling with cysteine residues of peptides and proteins.^17,18 By pre-generating the OAC reagents, we avoid the need to carry out an oxidative addition in the presence of the diverse functional groups presented at protein surfaces that may interfere with the desired reactivity. However, the inert atmosphere and organic solvents used to prepare these OACs are incompatible with the aqueous conditions required to manipulate protein substrates. Previously, we employed a sequential cross-coupling strategy for the direct ¹¹CN-labelling of unprotected peptides (cf. Figure 1B).¹⁹ These conditions generated a Pd-peptide OAC in situ and only tolerated up to 10% water with a maximal yield of 64% for the cyano-labelled compound. The success of this approach stemmed from a “nucleophile-nucleophile” coupling strategy based on an OAC derived from a dihaloarene.

We reasoned this strategy coupled with a judicious choice of ligand might enable facile oxidative addition on densely functionalized substrates in aqueous conditions with near-quantitative conversion to provide a new class of stable reagents: electrophilic Pd-protein OACs. By preparing an OAC of a 1,4-dihaloarene (VI, Figure 1B), a coupling reaction of a cysteine residue (V) with a Pd-OAC complex would generate a hypothetical intermediate π-complex (VII). A subsequent intramolecular oxidative addition into the Ar–X bond would form the Pd-protein OAC VIII. The Pd-protein OAC can then be isolated and used as a reagent for subsequent coupling reactions with various nucleophiles, including other cysteine-containing proteins (Figure 1C). The intramolecular oxidative addition process is commonly taken advantage of in catalyst transfer polymerization reactions (CTP);^20–,21,22 however, unlike polymerization chemistry, we would need only sufficient yield from a single Pd transfer event to obtain Pd-protein OACs for use in protein-protein cross-coupling.

To enable protein-protein cross-coupling, we demonstrate here the site-specific formation of electrophilic Pd-protein OACs from cysteine residues by leveraging an intramolecular Pd-transfer strategy. These cross-coupled products ultimately derive from a “nucleophile-nucleophile” based coupling approach, wherein two cysteine-containing proteins are cross-coupled. Mild conditions, stable linkages, and rapid rates are attributes of Pd-mediated transformations that satisfy a number of criteria considered to be critical for enabling protein-protein conjugation.²³ We show that several proteins are selectively coupled under our new reaction conditions.

We chose to optimize our method using Z33 (Figure 2), the minimized Z-domain of protein A that is capable of binding immunoglobulin G (IgG) with nanomolar affinity.²⁴ Z33 consists of 33 amino acids and was selected due to its protein-like structure and the ease of incorporating a single cysteine into its primary sequence in a straightforward manner through solid-phase synthesis. We anticipated two main possible pathways following the initial reaction of 2 with the Z33 cysteine mutant: (1) intramolecular oxidative addition to provide the desired Pd-transfer product 4 via intermediate 3 or (2) Pd dissociation to give arylhalide 5 (Figure 2).

Figure 2.

Reagent L3-Pd-I enables near-quantitative Pd-reinsertion for Pd-protein OAC formation in air from the cysteine residues of model substrate Z33-N17C with 5% DMF additive in aqueous buffer. ^aReactions were incubated for 30 minutes after which they were quenched by the addition of 3-mercaptopropionic acid (10 equivalents relative to Pd-reagent used). ^bConversion calculated as 100×(4+5)/(1+4+5) (see Supporting Information) ^cFinal concentration was 5% in DMSO and 5% in DMF. The ratio of 4:5 reflects the total ion count detected by LC-MS analysis followed by deconvolution. Indirect evidence for the formation of 4 was obtained by quenching the reaction mixture with 3-mercaptopropionic acid (see Supporting Information). Structure of Z33-N17C adapted from PDB 1ZDA.

In carrying out the optimization for the preparation of 4, it was unclear if an aerobic aqueous setting would permit the desired reaction pathway, or if the resulting protein OAC would be stable. We examined a set of dialkylbiaryl phosphine ligands to examine if any trends in Pd transfer could be observed. We began by employing conditions adapted from our work on the radiolabeled cyanation of unprotected peptides,¹⁹ using BrettPhos as a ligand, 1,4-diiodobenzene as the dielectrophile, and DMSO as a co-solvent. With 5% DMSO co-solvent, we initially observed an 81:19 ratio of oxidative addition product:aryl halide (4:5, entry 1, Figure 3) alongside conversions in the single digit range (<5%). Given our prior success using sulfonated ligands in conjugation reactions involving aryl bromides,²⁵ we utilized 1,4-dibromobenzene to compare the effect of a water-soluble sulfonated ligand system (sSPhos) with the non-sulfonated form of the same ligand (SPhos). We saw a modest increase in product derived from Pd transfer complex 4 when using the sulfonated ligand system sSPhos (entries 2–3). To further facilitate oxidative addition, the aryldiiodide was reexamined with sSPhos, resulting in a reversed selectivity to give 4:5 in a 76:24 ratio (entry 4). To minimize homodimerization (e.g., reaction of 4 with 1), 10 equivalents of L3-Pd-I were employed (entry 5), however resulting in formation of a precipitate. Hypothesizing that solvent composition was a key parameter that would influence intramolecular oxidative addition, we examined the impact of small quantities of additives on the reaction. The reaction performed with a 1:1 DMSO:DMF (v/v) mixture gave small improvements in both reinsertion efficiency and overall conversion compared to DMSO alone (entries 5–6). To avoid precipitation, the reaction was carried out at 10-fold dilution, wherein the use of 5% DMF was found to be superior to DMSO additive, producing a 95:5 ratio of products with >85% conversion (entries 7–9). Increasing the percentage of DMF gave comparable efficiencies in both re-insertion and conversion (entry 10). We note that throughout the course of this work, no solvents were degassed nor was any effort made to exclude air except for the initial formation of small molecule Pd OACs 2.

Figure 3.

Chemoselective Pd-protein-OAC formation from single cysteine-containing proteins. (A) Synthesis of mPA-OAC on 1.0 mg scale of mPA and purification by size-exclusion chromatography. Conditions: DMF:H₂O = 5:95 v/v, 20 mM Tris, 150 mM NaCl, 10 μM of protein and 100 μM of L3-Pd-I. (B) Deconvoluted mass spectra of three single cysteine-containing proteins before (top) and after (bottom) treatment with L3-Pd-I. Left mPA, middle T4 lysozyme, right SUMO-DARPin. Reaction conditions: 10 μM of protein, 100 μM of reagent L3-Pd-I, H₂O, 20 mM Tris, 150 mM NaCl, pH 7.5, 30 min reaction time.

To probe the generality of our approach, we generated Pd-protein OACs from several cysteine-containing proteins of varying size. Site-specific Pd-protein OAC formation with reagent L3-Pd-I was successful in each case. Included in this subset of proteins were cysteine variants of anthrax protective antigen (mPA, 83 kDa), T4 lysozyme (19 kDa) and SUMO-DARPin (27 kDa) (Figure 3) all of which were obtained via literature protocols via genetic expression (See Supporting Information).^17,26

Utilizing L3-Pd-I in 10-fold excess to minimize substrate dimerization, we were able to obtain the corresponding Pd-protein OACs with high conversions as determined by LC-MS analysis. The Pd-protein OAC of mPA was chosen because of its high molecular weight (83 kDa) to investigate the reactivity of a large protein (Figure 3A). After derivatization of mPA into the corresponding OAC, we probed ways to remove the excess small molecule reagent. Even though the OAC of the mPA could be recovered using a 7 kDa molecular weight cut-off desalting column, the product consistently co-eluted with the small molecule OAC L3-Pd-I.²⁷ After storage of this Pd-mPA OAC on the bench in aqueous buffer (20 mM Tris, 150 mM NaCl, pH 7.5) for one day, no loss of activity in reaction with the small molecule thiol 3-mercaptopropionic acid was observed as determined by LC-MS and mass deconvolution. Knowing that we could synthesize Pd-protein OACs with high efficiency, and that the resulting OACs were stable on the bench for at least 24 h, we next probed competence in cross-coupling chemistry.

Guided by the affinity of thiols for the soft Pd-metal center, we chose to engage these nucleophilic cysteine side chains due to their presence in native proteins. In addition, we anticipated protein-protein cross-coupling would proceed without necessitating any additives.

Treatment of the Pd-mPA OAC (83 kDa) (1.0 mg, 1 μM final concentration) reagent with 5 equivalents of SUMO-DARpin (27 kDa, 5 μM final concentration) reached 65% conversion within 12 hours and produced only minimal byproducts as determined by SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis, Figure 4A, also see supporting information Figure S12). After nickel-affinity and size-exclusion chromatography, the protein-protein conjugate was isolated in 28% yield and no longer contained any of the L3-Pd-I reagent.²⁸ Inductively coupled plasma mass spectrometry (ICP-MS) analysis indicated that 90% of the Pd was removed after conjugation.²⁹ This final conjugate, as well as the intermediate Pd-mPA-OAC, showed activity in line with wild-type PA (see Supporting Information for metal content analysis and bioactivity data).

Figure 4.

Protein-protein cross-coupling occurs within 12 hours at single digit micromolar concentrations. (A) Conjugation with 1.0 mg of Pd-mPA-OAC (1 μM) and SUMO-DARpin (5 μM). The conjugate was isolated via HisTrap affinity chromatography followed by size-exclusion chromatography to give a 28% isolated yield of the conjugate. Deconvoluted mass spectra are depicted for the starting SUMO-DARpin (top), Pd-mPA-OAC (middle), and final conjugate (bottom). (B) Cross-coupling with various thiol-containing proteins. ^aConversion was determined by SDS-PAGE followed by densitometry analysis. ^bThe band that elutes with the same retention factor as mPA is comprised predominantly of mPA conjugated with 3-mercaptopropionic acid.

This single Pd-mPA-OAC was cross-coupled with a variety of proteins bearing surface-exposed cysteine residues to exemplify the advantage of conjugating individually expressed proteins (Figure 4B). The reaction conversion efficiency for each reaction was again evaluated by SDS-PAGE with observed conversions ranging from 57% to 79%.

An alternative means of accessing protein-protein conjugates is via genetic expression, wherein the two proteins are solely ligated from the N-terminus of one protein and the C-terminus of the other.³⁰ Unlike genetic expression, individually folded proteins could be mixed and matched for access to a variety of conjugates without necessitating reengineering of the parent plasmid of a co-expressed protein. In addition, this conjugation strategy permits fusion of proteins from amino acid side-chains.

In conclusion, we report the first aqueous synthesis of air-stable Pd-protein OACs via a Pd-transfer strategy on protein substrates comprised solely of canonical residues, which achieves protein-protein cross-coupling from cysteine side chains. In forming an electrophilic OAC from a low abundance nucleophilic thiol, we generated a protein with umpolung reactivity. Thus, the products depicted in Figure 4 arise from a “nucleophile-nucleophile” based coupling approach. We envision this chemistry can be employed to generate stably linked protein therapeutics including new classes of antibody-drug conjugates and antibody-protein conjugates, alongside accommodating the use of a variety of nucleophilic functional groups.