LC-MS and HPLC Methods for DNA-Encoded Library Building Block Conjugates

Method 1: reverse-phase HPLC for purification and quantification

Both crude reaction mixtures and purified conjugates are run on an XBridge BEH C18 column (2.5 µm, 4.6 × 50 mm), eluted with 0.1 M TEAA pH 7 / 5% MeCN (Eluent A) against neat MeCN (Eluent B), on a 1%→30% B gradient over 20 minutes at 1.5 mL/min. This is the same method used both analytically (confirming a reaction worked before committing a plate to preparative purification) and preparatively (the actual purification run, with fractions collected on an 8-window, 5-minute-wide schedule keyed to the expected product’s retention time).

Quantification comes directly off this chromatogram. Oligonucleotide concentration is calculated from the 260 nm peak area using:

n (pmol) = 19500 × Area / extinction coefficient

The extinction coefficient at 260 nm is calculated per-sequence (the building block and linker’s own contribution is small enough to neglect), and the constant — 19,500 — is empirically calibrated against a DNA standard of known concentration. This lets every well on a purification plate be quantified from the same chromatographic run that purified it, with no separate UV measurement step.

Method 2: intact-mass LC-MS

Final identity confirmation uses a shorter, different column and a different eluent system, optimized for oligonucleotide ionization rather than preparative resolution: a Gemini C18 column (3 µm, 110 Å, reverse-phase), eluted with 200 mM HFIP + 8 mM triethylamine in water against methanol, analyzed on a Shimadzu LCMS-2020 with mass obtained after deconvolution. HFIP/TEA is a standard ion-pairing system for negative-mode oligonucleotide LC-MS — it gives cleaner, more interpretable deconvoluted spectra for a highly charged, ~16,500 Da species than a straight TEAA system would.

What the data actually looks like

Across a representative block of 18 building blocks conjugated to the same linker-oligonucleotide scaffold (calculated masses spanning roughly 16,525–16,651 Da), deconvoluted LC-MS gave mass deviations between calculated and found of no more than about ±2.4 Da — under 0.02% of the intact mass. That level of agreement is what makes intact-mass LC-MS a reliable pass/fail gate at this scale: a conjugate that’s missing its building block, still carrying its protecting group, or has picked up a side-reaction adduct shows up as a distinct, resolvable mass shift rather than ambiguous noise.

That resolving power is also what makes LC-MS the right tool for diagnosing a failed reaction, not just confirming a successful one. When a building block is cleaved from the oligonucleotide during deprotection (see our note on Boc self-cleavage), the LC-MS trace shows three distinguishable species side by side — protected conjugate, deprotected-but-intact conjugate, and bare linker with no building block at all — each with its own calculated and found mass. Being able to name and quantify all three from one deconvoluted spectrum is what turns “the yield was lower than expected” into an actual mechanistic explanation.

Practical notes

• Run both methods on the same plate workflow. The TEAA/MeCN method drives purification and gives you concentration; the HFIP/TEA/MeOH method confirms identity. Neither substitutes for the other — a well-quantified peak on the purification column can still be the wrong mass.
• dT-modified oligonucleotides can show a minor doubled peak on HPLC. Both peaks carry the same mass and the same functional amino-linker; the doubling is attributed to isomerism in the base modification rather than a separate species, and it’s a normal feature of this chemistry rather than a purification failure.