Joining a peptide to an oligonucleotide is difficult not because the bond is hard to form but because the two halves are finished under conditions that damage each other. Fmoc/t-Bu peptide synthesis ends in concentrated trifluoroacetic acid to remove t-butyl-type side-chain protection. Phosphoramidite oligonucleotide synthesis ends in strong aqueous base to remove acyl protection from the nucleobases. Each of those terminal steps is a hazard for the other polymer, and every synthetic strategy in this field is an answer to that one conflict.
The two failure modes
On the acid side the danger is depurination. Klabenkova and co-workers, reviewing the chemistry of peptide-oligonucleotide conjugates, state that it is impractical to use t-butyl-family protecting groups such as Boc on the peptide fragment of a purine-containing conjugate because of the danger of depurination during standard concentrated TFA deprotection. On the base side, the same review notes that the peptide fragment can undergo side reactions under the strongly basic conditions of the final deprotection of the acyl-type N-protecting groups used in conventional oligonucleotide synthesis [1]. Both statements come from the review's open-access full text rather than its abstract. Neither is a rate or a yield; both are constraints on which protecting-group scheme is admissible at all.
Two routes follow. Either you build one polymer on top of the other on a single solid support and find a protection scheme both can survive, or you build them separately, deprotect each under its own optimal conditions, and join the finished pieces by a chemoselective reaction. The review literature frames these as the stepwise solid-phase approach and post-synthetic conjugation in solution or on the solid phase, the latter often by click chemistry [1]. Malinowska and colleagues put the trade plainly: the stepwise method holds advantages in purification and in yield, while its disadvantages arise from the non-compatibility of the peptide and oligonucleotide synthetic methods [2].
Stepwise assembly on one support
Grandas and colleagues describe the classical stepwise procedure for phosphodiester-linked conjugates: the peptide is assembled first on a suitably derivatised solid matrix, and the oligonucleotide is then elongated at the free hydroxyl group of the linking amino acid. Temporary acid-labile and permanent base-labile protecting groups are combined, and they are explicit that careful choice of the protection scheme is required to prevent and minimise side reactions that may degrade the target molecule [3]. Stetsenko and Gait present total stepwise solid-phase synthesis on a single support as one of two general methods for conjugate preparation [4].
The consequence of building the peptide first is structural rather than empirical: every amino acid side chain then has to sit through the entire oligonucleotide assembly that follows, not merely through one final deprotection. Residues that are marginal under repeated exposure to the reagents of a phosphoramidite cycle are a liability for the whole synthesis, which is why the protection scheme, rather than the coupling efficiency, is usually what determines whether a stepwise conjugate is feasible.
Aviñó and colleagues addressed the acid problem directly. Using solid-phase methods to prepare oligonucleotides carrying peptides, they found that a stepwise protocol using acid treatment for the final deprotection allowed the preparation of polypyrimidine oligonucleotides carrying longer and arginine-rich peptides, whereas post-synthetic conjugation protocols were what succeeded for oligonucleotides carrying small peptides [5]. Read the qualifier carefully. The acid-final route was demonstrated on polypyrimidine sequences; restricting the sequence to C and T removes the purines that concentrated TFA attacks. That is a genuine solution and simultaneously a hard sequence constraint, unavailable if the sequence you need contains A or G.
Fragment conjugation, and the solvent it needs
The alternative is to give each polymer the deprotection it wants. The native-ligation route of Stetsenko and Gait is the clearest statement of that logic. A peptide carrying an N-terminal S-benzylthiosuccinyl moiety was generated in solution by trifluoroacetic acid deprotection of a standard Fmoc-based assembly; a 5'-S-tert-butylsulfenyl-L-cysteinyl oligonucleotide was generated by aqueous ammonia deprotection of a standard phosphoramidite assembly. Each fragment saw only its own chemistry. The two functionalised fragments were then used without purification in native ligation conjugation reactions in aqueous/organic solution, with tris(2-carboxyethyl)phosphine removing the tert-butylsulfenyl group in situ and thiophenol acting as a conjugation enhancer; the conjugates were purified by reversed-phase HPLC [6].
That phrase, aqueous/organic solution, is where the practical difficulty of fragment conjugation lives. An oligonucleotide is a polyanion and is comfortable in water and salt; many peptides are not. Hayashi and colleagues state the problem from the peptide side: ligation is generally conducted at high aqueous concentration to increase the reaction rate, and this often causes unpredictable aggregation and precipitation of starting or resulting peptides because of their hydrophobicities. Their answer exploits the same asymmetry that causes the trouble. They used the hydrophilicity of oligodeoxynucleotides to increase the water solubility of the peptides and the hybridisation of two complementary 15-mer conjugates to supply a proximity effect, which enabled carbodiimide-mediated ligation of a hydrophobic peptide at micromolar concentration [7]. Note what was in the tube: that experiment ligated two peptide fragments to each other on a DNA scaffold, not a peptide to an oligonucleotide. The solubility lesson still transfers — the nucleic acid half is the more forgiving partner, and attaching it early can carry a peptide that will not otherwise stay in solution.
Where a peptide will not tolerate water at all, the coupling can be moved onto the support and out of aqueous buffer entirely. Gras and colleagues report a catalytic approach to peptide-oligonucleotide conjugation that can be applied either on solid support in anhydrous media or post-synthetically in aqueous buffer. Single amino acids, peptides and cell-penetrating peptides were conjugated to various oligo(ribo)nucleotides with high conversions and good isolated yields, and the applicability of the method was demonstrated on more than 35 examples [8]. Having one bond-forming chemistry available in both an anhydrous and an aqueous format is worth more than a marginal yield gain, because it lets the solvent be dictated by whichever fragment is the more fragile.
Choosing the coupling chemistry
A cysteine thiol on one fragment and a maleimide on the other is a common pairing. Maleimide-type linkages proceed without a catalyst in aqueous buffers and give stable covalent linkages [2]. Kye and Lim combined solid-phase peptide fragment condensation with either thiol-maleimide or copper-catalysed azide-alkyne cycloaddition to build self-assembling conjugates, and the substance of their paper is guidelines for the selection of reactive functional groups and their placement during the conjugation reaction [9]. Placement, not identity alone, decides whether the reaction runs.
Linkage stability is the next question, and it separates the maleimide and disulfide options sharply. Youngblood and colleagues compared linkage types in cell-penetrating peptide-phosphorodiamidate morpholino oligomer conjugates in human serum and in cells, reading the products by MALDI-TOF mass spectrometry: an amide or a maleimide linkage was stable in both serum and cells, whereas an unhindered disulfide linkage was not stable in either [10]. The same study found the peptide half to be the labile component — D-configuration peptides were completely stable while L-peptides were degraded in both serum and HeLa cells, and the morpholino portion of the conjugate was completely stable in both. Mind the analyte class before transferring any of this: those constructs carried a phosphorodiamidate morpholino oligomer with an uncharged backbone, not a phosphodiester or phosphorothioate oligonucleotide. Where a cleavable linkage is the design goal rather than a defect, the disulfide's lability is the reason to choose it; disulfide linkages are described as designed to be cleaved in a reducing environment [2].
Carbonyl-based ligation avoids thiols altogether. Forget and colleagues developed a convergent route in which an oxyamine-containing peptide is treated with an aldehyde-containing oligonucleotide, or the reverse, to form an oxime, and in which a peptide acylated with a cysteine residue couples to an oligonucleotide derivatised by an aldehyde function to form a thiazolidine. For both approaches the conjugates were obtained in good yield without the need for a protection strategy and under mild aqueous conditions, and the oxime ligation also proved useful for directly conjugating duplex oligonucleotides [11]. The second of the two general methods set out by Stetsenko and Gait belongs to the same family: oligonucleotides containing 2'-aldehydes are ligated chemoselectively with functionalised peptides to form oxime or thiazolidine linkages, among others [4]. What chemoselectivity buys here is specific and worth stating plainly — Forget's couplings required no protection strategy at all, so the deprotection conflict that governs the stepwise route simply does not arise.
Strain-promoted cycloaddition removes the copper catalyst from the picture. Jawalekar and colleagues demonstrated that bicyclo[6.1.0]nonyne can be introduced into synthetic oligonucleotides by standard solid-phase phosphoramidite chemistry, and presented a range of phosphoramidite building blocks for incorporating either BCN or azide, on-support or in solution; conjugation of the resulting oligonucleotides was straightforward and high-yielding, demonstrated for biotinylation, fluorescent labelling, dimerisation and attachment to polymer rather than for peptides [12]. Their stated motive for BCN is the one that matters here: the reported cyclooctynes are large and hydrophobic, which is the wrong property to append to a molecule whose solubility is already a compromise. Flory and colleagues used a solution-phase strain-promoted copper-free click reaction to conjugate a fluorescent dye to two bifunctional PNA strands, and note that PNA can be more challenging to work with in aqueous solvents because of its hydrophobic nature [13].
Purification
Two contaminants have to be removed: unreacted oligonucleotide and unreacted peptide. They differ from the product in opposite directions, one in charge and one in hydrophobicity, which is why a single chromatographic mode rarely suffices.
Anion exchange sorts on the phosphate backbone. Lloyd and colleagues, using a strong anion-exchange functionality on a pH-stable polymeric particle and a poly-T oligonucleotide size standard, resolved full-length oligonucleotide from the n−1 and n−2 truncated species due to coupling failure up to at least the 30mer, and separated a phosphodiester contaminant from a phosphorothioate oligonucleotide by anion-exchange HPLC at high pH [14]. Those analytes were deprotected, trityl-off oligonucleotides, not conjugates. The retention mechanism transfers; the selectivity does not transfer automatically, because the peptide contributes charge of its own and a basic peptide reduces the net anionic charge the column is reading.
Denaturing gel electrophoresis is the fallback when chromatography fails on aggregation. Kye and Lim found polyacrylamide gel electrophoresis under denaturing conditions most suitable for recovering self-assembling conjugates in high yield [9]; the qualifier is load-bearing, since their conjugates were designed to self-assemble and that is precisely what defeats a chromatographic separation. Flory and colleagues reached for denaturing PAGE for a different reason: it separated the charged, dye-labelled PNA conjugate from large excesses of unreacted dye and of unreacted, neutral PNA [13]. In each case the separation runs on the property the conjugation reaction actually changed.
Reversed phase then does two jobs, resolving on the peptide's hydrophobicity and stripping the non-volatile components an ion-exchange step leaves behind. Li and colleagues built exactly this as a two-dimensional method for synthetic oligonucleotides: a reversed-phase/weak anion-exchange stationary phase in the first dimension provided selectivity for structurally closely related sequences and deletion products under a mixed pH and triethylammonium phosphate buffer gradient, and heart-cut peaks were transferred to a reversed-phase second dimension used for desalting, so that the analyte reached the source entirely free of non-volatile buffer components and ion-pair agents [15]. Stetsenko and Gait's conjugates went straight to reversed-phase HPLC [6], which is sufficient when the ligation is clean and the fragments are short.
Mass spectrometry, and what the spectrum licenses
The charge envelope of a conjugate belongs to the nucleic acid. In negative-ion electrospray the phosphodiester backbone supplies an ionisable site per internucleotide linkage, so the molecule appears across a spread of charge states, and the reported mass is only as good as the charge assignment used to deconvolute it.
That envelope is not a fixed property of the molecule; it is set by the solution being sprayed. Chen and colleagues examined organic modifiers in oligonucleotide electrospray and found that the Henry's Law constant of the alkylamine, which reflects its volatility, had a prominent effect on both charge-state distribution and desorption efficiency. Alkylamines with higher values such as hexylamine, piperidine and imidazole reduced the charge-state distribution by forming complexes with the oligonucleotide that dissociate from it in the gas phase; those with extremely low values reduced it by facilitating ion emission at an earlier stage of the desorption process; ion-pairing agents with moderate values did not alter the distribution. Ionisation enhancement was observed with hexafluoroisopropanol, attributed to its low Henry's Law constant and moderate acidity [16]. Those measurements were made on oligonucleotides, not on conjugates. The operational point survives the transfer: change the ion-pairing agent between runs and the envelope moves, so a shifted envelope is not by itself evidence that the sample changed.
Two consequences for characterisation. First, the desalting dimension described above is not a convenience — non-volatile buffer and ion-pair components have to be off the analyte before electrospray if the mass is to be measured under MS-compatible conditions [15]. Second, an intact mass matching the calculated conjugate establishes that the two pieces are joined and that no gross deletion occurred. It does not establish where they are joined when more than one nucleophile was available on either fragment, and it does not by itself confirm that the linkage formed is the one drawn. Site assignment needs either a handle installed by design, so that only one position can react, or a fragmentation experiment. A conversion read from an analytical trace is also not an isolated yield, and it is the self-assembling conjugates that Kye and Lim describe as harder to synthesise and purify than the rest [9].
Related methods in this library
The guide to peptide nucleic acids covers the case where the nucleic acid half carries no charge at all [13]; that removes the charge asymmetry the anion-exchange step here depends on, but leaves the solubility problem intact and, if anything, worse. The guide to click chemistry for peptide conjugation treats the copper-catalysed and strain-promoted variants in general terms; the constraint this article adds is the solvent one set out under coupling chemistry above, which is what governs the choice of cyclooctyne [12].