Can 5HP2O linkers build stable multifunctional protein-oligonucleotide conjugates?

A linker that must survive storage and assembly

The paper asks whether a single chemical hub can join a protein to an oligonucleotide while also carrying extra cargo, without falling apart during the handling steps in between. The authors position 5HP2O (5-hydroxy-1,5-dihydro-2H-pyrrol-2-ones) as that hub. The molecule combines a Michael acceptor, which reacts with a cysteine thiol on the protein, an NHS active ester, which reacts with a primary amine on the oligonucleotide, a tunable spacer, and optionally a secondary cargo.

The design matters because the two ends are used at different times. The oligonucleotide is functionalized first, the intermediate may be stored, and the protein is attached last. Any hydrolysis of the Michael acceptor during that interval destroys the reactive site before the protein ever arrives. The paper therefore treats hydrolytic stability of the oligonucleotide intermediate as a prerequisite, not a side property.

One-pot photooxidation builds the linker library

The linkers are made from 2-substituted furans. Methylene blue acts as a photo- and redox catalyst while air is bubbled through the solution under light, converting the furan in a one-pot sequence; a spacer amine is then introduced, and the carboxylic acid is converted to the NHS ester. Because the spacer is supplied as an amine, aliphatic, ethylene glycol, and peptide-based spacers can be swapped in without redesigning the core.

This modularity produced the panel the paper tests: heterobifunctional linkers joining protein and oligonucleotide, trifunctional versions carrying a fluorophore (NBD) or a click handle (azide, alkyne, ketone), and a tetrafunctional version containing a valine-citrulline sequence intended as an enzymatic cleavage site. Isolated yields for the trifunctional one-pot photoreactions were 9–20%, below the 20–30% of the bifunctional linkers, which the authors attribute to altered photooxidation efficiency of substituted furans and to partial isolation of carboxylic-acid 5HP2Os as triethylammonium salts that impede DCC-mediated NHS coupling.

Oligonucleotide intermediates stay fully intact

The central stability comparison uses the same amino-oligonucleotide, D1, functionalized either with 5HP2O-C5-NHS or with the maleimide analogue Mal-C5-NHS. The isolated 5HP2O-C5-D1 construct remained 100% intact after 28 days at pH 6 to 9. The maleimide construct hydrolyzed quickly under the same conditions: about 60% remained reactive after 3 days at pH 7, hydrolysis was complete within 7 days at pH 8, and within 3 days at pH 9.

Conversion during functionalization follows the same logic. Raising the linker excess to 40 equivalents relative to the amino-oligonucleotide was needed for near-complete 5HP2O functionalization within 2 hours; dropping to 10 or 2 equivalents stalled conversion at roughly 20% and 5%. Raising the pH to 9 accelerated 5HP2O functionalization, whereas the same basic condition accelerated maleimide ring-opening and lowered maleimide conversion. On a 100 nmol preparative scale, isolated yield was 34% for 5HP2O-C5-D1 and 12% for Mal-C5-D1.

These are two different measurements and should not be merged: the 100% figure describes how much isolated construct survives storage, while the 34% and 12% figures describe how much product was recovered from the functionalization reaction.

Protein conjugation is slower and needs excess protein

The first protein model was an anti-interleukin-23 Alphabody, a triple-helical coiled-coil protein with a calculated isoelectric point of 4.2, so it is negatively charged at the pH 7–10 conjugation conditions and electrostatically repelled by the polyanionic oligonucleotide. With a 10-fold excess of reduced protein, 5 hours at pH 10 gave 70% conjugate versus only 3% at pH 7. Overnight incubation narrowed the gap to 93% at pH 10 and 61% at pH 7, and reactions between pH 7 and 9 consistently plateaued near 60%.

The authors attribute that plateau to competitive reoxidation of Alphabody monomers back to the unreactive disulfide dimer, which limits how much reactive protein remains available. Reducing the protein excess from 10 to 5 equivalents cost only 8 percentage points of conversion, but 2 equivalents gave insufficient product. The isolated conjugate was confirmed by mass spectrometry at 18,345.76 Da, consistent with a potassium adduct of a 1:1 protein-oligonucleotide construct.

The same chemistry behaved differently at higher concentration: peptide-oligonucleotide conjugation at 500 µM oligonucleotide proceeded with only a 4-fold peptide excess in 3 hours, which the authors link to the low 40 µM oligonucleotide concentration and electrostatic repulsion in the Alphabody case.

Retro-Michael stability beats aliphatic maleimide

Stability of the finished conjugate was tested by incubating purified constructs with a 100-fold excess of glutathione at basic pH, which drives retro-Michael exchange and strips the cargo off the protein. After 6 days at pH 8, the 5HP2O conjugate remained 80% intact, while the aliphatic maleimide conjugate left only 40% intact. Against the cyclohexyl maleimide reference made with the commercial SMCC linker, the 5HP2O conjugate showed a directly comparable stability profile.

The comparison also exposes a practical asymmetry. Synthesizing the maleimide reference conjugates gave complex mixtures with prominent side products, restricting conversion to 39% for the aliphatic variant and 77% for the cyclohexyl variant, whereas 5HP2O-mediated conjugation cleanly exceeded 90% conversion. So the 5HP2O route wins on both the stability of the assembled construct relative to the aliphatic maleimide and the cleanliness of the conjugation step, but it does not beat the cyclohexyl maleimide on stability.

Antibody and fluorophore conjugates work in cells

The platform was extended to rituximab, where reduction of interchain disulfides followed by 5HP2O-C5-D1 conjugation gave the LHH-D1 species as the primary reaction product on SDS-PAGE, regardless of residual TCEP. A trifunctional linker carrying NBD allowed the oligonucleotide to be tracked: the fluorescent Alphabody conjugate was taken up into HeLa cells after Oligofectamine-mediated transfection, with fluorescent foci localized inside cell boundaries by confocal microscopy.

The most demanding demonstration used antisense oligonucleotides against MALAT1 RNA. Amino-ASOs M4 and M5 were functionalized with an alkyne-5HP2O linker in 51% and 76% isolated yields, then reacted with a pre-reduced pTag peptide in 14–23% isolated yields. Residual TCEP from the peptide reduction cross-reacted with the 5HP2O handle and formed a side product, which the authors note limited final conjugation efficiency and could be minimized by in situ quenching, a step not investigated here. Copper-catalyzed azide-alkyne cycloaddition then installed either NBD or a second M4 ASO, and the fluorescent pTag(NBD)-M5 construct loaded onto the bispecific anti-CD40 antibody STRIKE2001 gave fluorescent signal in 80% of CD40-overexpressing Daudi cells by flow cytometry, while the construct alone produced no fluorescent cells.

Where the 5HP2O advantage ends

The paper's own comparison sets the priority: choose 5HP2O when the oligonucleotide intermediate must be stored in aqueous buffer or handled without strict pH control, and when a clean single-site conjugation matters more than reaction speed. The 28-day, 100% intact result at pH 6–9 and the 80% versus 40% glutathione retention at pH 8 are the numbers that support that choice.

The advantage does not extend to every dimension. 5HP2O reacts more slowly than standard maleimides, needs pH 9 for near-quantitative oligonucleotide functionalization, and required a 10-fold protein excess to reach 93% conversion with the Alphabody at pH 10, with a plateau near 60% between pH 7 and 9 caused by protein reoxidation. Against a cyclohexyl maleimide the stability gain disappears. And the multifunctional constructs were demonstrated at low isolated yields, with residual TCEP cross-reactivity and unconsumed starting material in the CuAAC step flagged by the authors as room for optimization. The practical threshold is therefore stability-critical, storage-heavy workflows, not a blanket replacement for maleimide chemistry.

Sources used in this answer