Mild Radical Chemistry for Late-Stage Peptide Modification

What “Milder” Actually Means in This Context

Mild, in peptide chemistry, is not a marketing adjective. It encodes specific operating parameters: pH 6–8, temperature ≤ 37 °C, aqueous-compatible solvent systems, and tolerance for hydroxyl, carboxyl, amino, thioether, imidazole, and guanidinium side chains without preprotection. Published criteria for bioconjugation compatibility generally define this range explicitly, and the recent radical chemistry literature increasingly reports experiments performed squarely inside it.

That said, “mild” still varies across the transformation classes described below. Photoredox methods using organic photocatalysts and blue or green LEDs tend to be the most biocompatible by default. HAT-based cascade methods and electrochemical C–H/S–H coupling methods can require co-solvents (typically DMSO/water mixtures) at ratios that affect solubility of longer or more aggregation-prone sequences. Directed C(sp³)–H arylation methods — particularly the palladium-catalyzed variants — impose additional constraints around directing group installation and removal, which adds steps and potential impurity sources. The question for any development team is not whether a method is “mild in principle” but whether it is mild enough for the specific sequence and downstream application.

Photoredox Conjugation: What the Evidence Supports

Photoredox catalysis has generated the largest and most reproducible body of late-stage peptide functionalization data. The core mechanism — exciting a photocatalyst with visible light, generating reactive radical intermediates at defined positions, and coupling them to radical acceptors or partners — offers chemoselectivity through the electronic selectivity of the excited-state catalyst rather than through protecting-group maneuvers.

Tryptophan as the primary handle. C2-alkylation and C2-sulfenylation of tryptophan-containing peptides have both been demonstrated under metal-free and photocatalyst-free conditions, the latter using arylthianthrenium salts. A 2023 report in Organic Letters (ACS, chemoselective late-stage functionalization via photocatalytic C2-alkylation of tryptophan) described functional group tolerance compatible with unprotected aspartate, glutamate, lysine, and serine side chains — which is the minimum bar for credible peptide late-stage functionalization. A 2024 clickable tryptophan modification described in Science Advances reported catalyst-free, late-stage C2-sulfenylation tolerant of free cysteine residues, a particularly demanding test of chemoselectivity.

Histidine alkylation. A visible-light Minisci-type C–H alkylation of histidine-containing peptides using 4-alkyl-1,4-dihydropyridines as radical precursors has been reported. Histidine is ordinarily a problematic residue in photochemical contexts because of its oxidation susceptibility, so validated methods targeting the imidazole C2 or C5 position add a chemically distinct handle.

Dehydroalanine (Dha) as a versatile radical acceptor. Dha installed enzymatically or through chemical conversion from serine or cysteine acts as a Michael-type radical acceptor under a broad range of conditions. A 2025 report on site-selective construction of N-linked glycopeptides through photoredox catalysis (PubMed) used Dha as the acceptor for N-glycosyl oxamic acid-derived radicals under visible light, demonstrating Giese-type addition without protecting the remaining glycine, alanine, or serine residues present. The yield and diastereoselectivity figures in that study — and the absence of a stereoselective solution for Dha radical additions in many earlier reports — highlight a persistent gap: radical addition to Dha is often not stereoretentive, which matters when the α-carbon configuration affects receptor binding or protease stability.

Decarboxylative conjugation. MacMillan’s group showed that the C-terminal carboxylate of peptides can be selectively oxidized and coupled through radical decarboxylation under blue light, exploiting the difference in oxidation potential between internal and terminal carboxylates. This approach is mechanistically clean and avoids any side-chain preprotection, but it consumes the C-terminus — which limits its applicability unless a free C-terminus is expendable or the goal is terminal labeling.

Development note: Photoredox methods have demonstrated the broadest functional-group tolerance across the literature. The open questions for development use are reagent and photocatalyst removal, oxygen sensitivity during reaction and workup, and whether solvent systems (often DMSO/water) are compatible with downstream purification and lyophilization without introducing persistent residuals.

HAT and MHAT: Remote Activation Without Prefunctionalization

Hydrogen atom transfer (HAT) chemistry occupies a different niche from photoredox. Rather than targeting a predefined redox-active handle, HAT-based methods abstract hydrogen atoms from C(sp³)–H bonds to generate carbon-centered radicals at positions that would otherwise require directing groups or prefunctionalization. In the context of peptide modification, this primarily means two things: remote functionalization via 1,5-HAT relay processes and sequential cascade reactions combining radical generation with a second catalytic bond-forming step.

Synteza peptydów A 2025 paper in Organic Chemistry Frontiers (RSC, three-component alkylation of glycine derivatives and peptides via a site-selective 1,5-HAT cascade) demonstrated visible-light-mediated N-centered radical generation followed by 1,5-hydrogen atom transfer to the α-carbon of glycine residues, enabling a three-component alkylation. The selectivity arises from the geometry of the 1,5-shift rather than from electronic preference for any particular side chain, which has both advantages and limitations: the approach is broadly applicable to glycine-containing sequences but does not inherently distinguish between similarly positioned residues in longer chains.

Metal-assisted HAT, sometimes referred to in its cobalt- or iron-mediated forms as MHAT, brings additional control. The metal hydride generates a carbon radical under mild conditions — typically room temperature, hydrogen or silane as hydrogen source — and the resulting radical can be trapped by an alkene or radical acceptor in the same flask. For peptide application, the relevant question is whether the metal catalyst and hydrogen-source reagent are compatible with histidine, methionine, and cysteine side chains, all of which are potential competing reaction partners. Published precedent exists for simple dipeptide and tripeptide substrates; validation at longer chain lengths with multiple reactive side chains remains limited in the primary literature as of mid-2026.

Development note: HAT and MHAT methods are attractive for modifying sequences where a natural radical handle (Cys, Trp, Dha) is absent or must be preserved for biological activity. The site-selectivity challenge is real and sequence-dependent — a short peptide where selectivity looks excellent may behave differently in a 20-residue sequence with multiple competing abstractable positions.

C(sp³)–H Functionalization: Backbone-Adjacent Editing

Directed C(sp³)–H functionalization represents the most ambitious arm of this chemistry for peptide applications, targeting internal backbone-adjacent positions without a traditional reactive handle. Two approaches have demonstrated preparative-scale potential on peptide substrates.

Side-chain-directed arylation. A 2024 study from Hou et al. — summarized in a 2026 RSC review (recent advances in site-specific modifications of peptides and proteins) — reported β-C(sp³)–H arylation of internal positions in both linear and cyclic peptides using S-alkyl cysteine as the directing group and 2-pyridone as an auxiliary ligand, under palladium catalysis. The method achieved arylation at the β-carbon of the residue adjacent to the cysteine-derived directing group under exceedingly mild conditions for a palladium-catalyzed C–H activation.

Backbone-inserted directing groups. A 2025 report in JACS (PubMed, Pd(II)-catalyzed C(sp³)–H functionalization directed by a removable, backbone-inserted amidoxime ether) described amidoxime ether insertion at any amide bond in the backbone, using it as a removable directing group for subsequent C(sp³)–H arylation. The amidoxime ether is installed and removed in separate synthetic steps, adding two transformations to the route — but the positional flexibility it provides has no equivalent in current standard modification approaches.

For sequence-agnostic positioning, this class is chemically significant. The practical burden — palladium removal, directing group installation and removal, SPPS-compatible protecting group compatibility, and potential epimerization at the α-carbon under palladium catalysis — is also the heaviest among the three transformation families discussed here.

Sulfur-Based Radical Methods: Chemoselectivity Backed by Data

Cysteine remains the most analytically tractable residue for site-selective peptide modification because its sulfur nucleophilicity and redox chemistry are sufficiently distinct from other side chains to allow clean reactions with appropriate electrophilic radical sources. Several 2024–2025 reports extend this toolkit in directions relevant to conjugation programs.

Aryl radicals generated from aryl pinacol boronates under mild aqueous conditions have been shown to perform selective S-arylation of peptides and proteins (aryl radicals from aryl pinacol boronates for peptide and protein modification, TU Darmstadt/Wiley). The selectivity for sulfur over aromatic amino acid side chains under aqueous, near-neutral conditions represents a meaningful advance over aryl diazonium salt chemistry, which is far less selective.

An electrochemical approach to C–H/S–H cross-coupling for cysteine-containing peptides, including intramolecular macrocyclization, was described in Nature Communications (electro-induced C–H/S–H cross-coupling for functionalization/macrocyclization of cysteine-containing peptides, 2025). The umpolung strategy — generating a thiyl radical electrochemically and coupling it to an adjacent C–H bond — operates without external oxidant or photocatalyst and tolerates a range of side chains.

For programs involving stapled peptides, ADC-like conjugation, or fluorescence labeling at defined positions, these sulfur-selective methods represent the lowest-risk entry point because chemoselectivity data exists across the longest validated chain lengths and the most diverse side-chain environments.

Compatibility, Impurity, and Analytical Questions That Must Be Answered

The chemistry described above is real and reproducible at discovery scale. Translation to preparative and development scale requires honest answers to four categories of questions before any of these methods enter a formal development workflow.

Compatibility Assessment

Before committing to a late-stage radical modification route, development teams should map the following:

Impurity Profile Prediction

Late-stage radical modification introduces a class of impurities distinct from SPPS-derived impurities. Expect and plan analytical methods for:

• Unreacted starting material: unconjugated peptide co-eluting with or near the product

• Over-modified species: if more than one reactive site exists (e.g., two tryptophan residues), multiple-modification products will appear as higher-mass signals on ESI-MS or MALDI-TOF

• Regioisomeric products: for Dha radical addition or histidine alkylation, positional isomers that differ only in the site of bond formation may be analytically indistinguishable by RP-HPLC alone

• Oxidation products: radical conditions can oxidize methionine to the sulfoxide or tryptophan to the oxindole — peaks that are structurally close to the starting material on reverse-phase chromatography

• Catalyst-derived adducts: organic photocatalyst fragments, palladium residues (for C–H functionalization), or reagent by-products from HAT co-reagents that form adducts with the peptide

The regulatory and CMC implications of this impurity landscape align with the EMA’s Guideline on the Development and Manufacture of Synthetic Peptides, which explicitly addresses process reagents, elemental impurities, and the control strategy for synthetic intermediates.

Analytical Requirements

Standard RP-HPLC at 214 nm, which works well for SPPS impurity profiling, is insufficient for radical late-stage modification outputs. Teams should plan for:

• LC-MS or LC-HRMS as the primary characterization tool — MS is required to distinguish unconjugated starting material from mono-modified product when they have similar chromatographic retention

• Orthogonal chromatography modes (e.g., HILIC or IEX in addition to RP) when regioisomers are anticipated; regioisomers with identical mass will not resolve by mass detection alone

• Amino acid analysis (AAA) for confirming that the amino acid composition after modification is consistent with a single addition event, not backbone cleavage or unusual side-chain loss

• Stability-indicating methods that can distinguish synthesis impurities (present in the initial batch) from degradation products that appear over time — critical when a radical modification step introduces new chromophores or functional groups that alter the oxidation or hydrolysis stability profile of the peptide

Scale-Up Engineering

Photoredox methods present the most technically specific scale-up challenges because photon delivery does not scale linearly. The incident light intensity at the center of a 1-liter flask is a fraction of that at the vessel surface, and lamp-to-reactor geometry determines conversion in a way that has no analogue in thermally driven chemistry. Flow chemistry reactors designed for photocatalysis address this through thin-film or flow-through irradiation, and several validated reactor designs now exist for pharmaceutical process development — but they require upfront capital and method re-optimization at each scale step.

HAT and electrochemical methods have different scale-up profiles. Electrode surface area scales more predictably than photon delivery, but electrochemical methods introduce questions about electrode materials compatibility, electrolyte removal, and current density uniformity at preparative scale.

For all transformation classes, the transition from analytical-scale HPLC purification to preparative HPLC introduces additional resolution challenges. A late-stage modification product co-eluting with mono-modified regioisomers at analytical scale will require careful re-optimization of gradient slope, mobile phase composition, and column chemistry at preparative loading — and preparative HPLC recovery from a complex mixture is the rate-limiting step in peptide scale-up more often than the synthesis itself.

A Framework for Evaluating Adoption Feasibility

Rather than adopting any of these methods wholesale, development teams should run a structured feasibility assessment tied to the specific modification target. The following decision matrix reflects the compatibility and risk dimensions discussed above.

Passing the full checklist does not guarantee success — but it identifies which experiments to prioritize before committing development resources. A failed compatibility screen at the 5-milligram scale saves far more time than the same failure at 100 milligrams during a late CRO run.

Where Conjugation Platforms Already Available Can Help

Programs evaluating these newer methods alongside existing, validated modification approaches will naturally reference what their synthesis partner can execute without a custom process development phase. Peptide modification platforms that routinely handle click chemistry handles (azides, maleimides, alkynes), glycopeptide installation, I fluorescent or isotope label conjugation already operate within the functional group tolerance envelope that radical late-stage methods are working to expand.

The value of the newer radical methods is not to replace these established conjugation routes — it is to access modification positions that classical chemistries cannot reach, particularly internal backbone positions and residues that lack convenient nucleophilic handles. When that capability is the bottleneck, a targeted radical functionalization method is worth the analytical and process development investment.

For everything else, established niestandardowa synteza peptydów workflows with defined impurity acceptance criteria and validated analytical packages remain the faster path to a QC-released, development-ready batch.

Questions the Field Has Not Yet Answered

To close with an honest inventory: the 2024–2025 literature has established that mild radical chemistry works on peptides under discovery conditions. It has not yet established, for most methods:

1. Scalability: how conversion and selectivity behave at 100 milligrams versus 1 gram versus 100 grams under process-realistic conditions, with proper oxygen exclusion and temperature control

2. Stereoselectivity at Dha: radical addition to dehydroalanine generates a new α-center with variable diastereomeric ratio; this is an open problem, and the recent ChemBioChem review (advances in dehydroalanine-specific modification, 2025) explicitly flags it as unresolved for most substrate classes

3. Regulatory precedent: none of these routes has, to public knowledge, been the subject of a CMC IND section covering impurity qualification, genotoxic impurity assessment of photocatalyst residues, or stability-indicating method validation for radical-modified API — the groundwork remains to be laid

4. Long-term stability of the modification bond: radical additions to Dha, Trp, and His create C–C, C–S, or C–N bonds with varying stability profiles; stability study data is sparse for most conjugates

These are not reasons to dismiss the chemistry. They are the experimental and regulatory program that must be designed alongside any decision to incorporate these methods into a clinical development workflow.

Practical Next Steps for Development Teams

For teams actively assessing whether mild radical late-stage functionalization belongs in their modification strategy, the most productive near-term actions are:

• Commission a small-scale feasibility run (5–10 mg) with the exact sequence, targeting the specific residue type, under the proposed radical conditions — with LC-HRMS characterization of the crude mixture, not just HPLC purity

• Request stability data on the radical adduct bond under pH extremes, oxidative stress, and thermal stress before designing a purification method

• Map the impurity profile of the crude against what your preparative HPLC system can resolve, using spiked reference standards if available

• Identify whether regulatory precedent for the catalyst or reagent class exists in published IND CMC summaries, FDA guidance documents, or EMA assessment reports

The chemistry is moving fast. The analytical and process development infrastructure to use it responsibly in a development program takes longer to build. Starting that construction now, on a well-defined target sequence, is the right approach.

Zmiany MOL-a supports development teams navigating complex modification requirements across more than 300 functional group transformation types, including click chemistry conjugation, glycopeptide synthesis, and isotope-labeled peptide production within a Class 100 sterile manufacturing environment. If your program involves modification positions or conjugation targets that fall outside standard SPPS workflows, a technical feasibility discussion is a practical starting point.