Strand by strand
How a single quadrupole mass spectrometer confirms what a synthetic oligonucleotide is: by its charges, its mass, and where it elutes.
Oligonucleotides are short synthetic strands of DNA or RNA, increasingly used as medicines. The LCMS-2050's DUIS ion source ionizes them by electrospray (ESI) and APCI at the same time.
Scroll, or press D for a guided demoOne instrument,
four questions.
- iClean spectra under ion pairingCan source settings strip HFIP adducts from the charge states? SSI 02-SSI-LCMS-173 · Nov 2024
- iiMethod development, ion-pair RPHow quickly can a 20-mer be separated from five related impurities? 01-00766-EN · Sep 2024
- iiisiRNA without ion pairsCan both strands of a duplex be confirmed with no ion-pair reagent? 01-01180-EN · May 2026
- ivHILIC for impuritiesCan HILIC replace ion pairing for impurity analysis? 01-01238-EN · Jul 2026
Every linkage
is a charge.
Oligonucleotides carry a phosphate (or phosphorothioate) linkage between neighbouring nucleotides, so they are usually detected as negative ions. One molecule picks up several charges at once and appears at a whole ladder of m/z values.
Shake off
the HFIP.
Ion-pair LC puts HFIP and triethylamine in the mobile phase (here 95 mM HFIP and 4.3 mM TEA). Some HFIP reaches the mass spectrometer still attached, so each charge state splits into [M]z−, [M+HFIP]z− and [M+2HFIP]z−. That clutter makes deconvolution harder.
Of the seven source settings, Qarray voltage had the most substantial effect on signal and adducts. At the default −20 V and below, adducts were high and total signal low. Signal rose up to about −40 V for the 15-mer and about −60 V for the 30- and 60-mer, then fell past −70 V. −40 V was chosen.
| Setting | Range tested | Chosen |
|---|---|---|
| Qarray voltage | −10 to −90 V | −40 V |
| Interface voltage | −1 to −4 kV | −3 kV |
| Desolvation line temp. | 150–250 °C | 250 °C |
| Desolvation temp. | 350–500 °C | 500 °C |
| Nebulizing gas | 0.5–3 L/min | 2 L/min |
| Drying gas | 3–5 L/min | 5 L/min |
| Heating gas | 3–7 L/min | 7 L/min |
| DNA | Theoretical | Observed | Error |
|---|---|---|---|
| 15-mer | 4528.8 | 4528.3 | 0.5 Da |
| 30-mer | 9111.6 | 9110.9 | 0.7 Da |
| 60-mer | 18292.2 | 18290.4 | 1.8 Da |
Most abundant masses after deconvolution in Insight Biologics. Hotter desolvation also cut adducts; drying gas mattered little.
Six strands
that look alike.
The sample mixes a 20-mer full-length product (FLP) with five impurities: two 19-mers each missing one end nucleotide (one at the 3′ end, one at the 5′ end), an n-3 17-mer, an n+1 21-mer, and "PO", the FLP with one phosphorothioate at the 5′ end turned into a phosphate diester. Their UV spectra are almost identical (similarity 0.99 or higher).
P is the number of peaks found and Rs the resolution between neighbours. LabSolutions MD ranks every run by this value; only the published top 10 carry a number here. Dashed cells showed baseline fluctuation.
Chasing
the last 0.44.
Starting from the best screen (100 mM HFIP, 10 mM TEA), three more settings were varied: acetonitrile share of the organic solvent (40, 50, 60%), oven temperature (55, 60, 65 °C) and initial gradient concentration (6, 7, 8%). Resolution generally improved as each went up, though the design spaces showed the best acetonitrile share depended on the pair.
Overlaid design spaces then found the point that maximizes the hardest pair, n-1(5′)/n-1(3′), while keeping n-1(3′)/PO above Rs 0.7 and the last peak before 16 min: ACN 54% / MeOH 46%, 65 °C, initial 8%.
| Pair | Rs | Criterion |
|---|---|---|
| n-3 / n-1(5′) | 7.4 | |
| n-1(5′) / n-1(3′) | 0.44 | maximized |
| n-1(3′) / PO | 0.81 | > 0.7 ✓ |
| PO / FLP | 1.4 | |
| FLP / n+1 | 3.5 | n+1 < 16 min ✓ |
| Peak | Length | Estimated MW | Theoretical |
|---|---|---|---|
| n-3 | 17 | 5985 | 5986 |
| n-1(5′) | 19 | 6772 | 6774 |
| n-1(3′) | 19 | 6773 | 6775 |
| PO | 20 | 7151 | 7152 |
| FLP | 20 | 7167 | 7168 |
| n+1 | 21 | 7561 | 7563 |
Retention times shift whenever a setting changes, and the UV spectra are too similar to track peaks by, so LabSolutions MD tracked them through all runs by molecular weight from the LCMS-2050. The two n-1 strands differ by only 1 Da; the note attributes their poor separation to their very similar structure.
A duplex
comes apart.
Ion-pair reagents such as hexylamine or triethylamine raise concerns that instrument contamination may affect other analyses. This note drops them: a C18 column (Shim-pack Scepter Claris C18-300) with 10 mM ammonium bicarbonate in water and a methanol gradient.
The sample is a double-stranded siRNA based on the patisiran sequence, two 21-nt strands.
At both 25 °C and 60 °C the UV trace showed two peaks, not one, so the duplex had dissociated into single strands. Insight Biologics identified them as the sense and antisense strands, each within 1 Da of its theoretical mass.
| Sense strand, 60 °C | m/z |
|---|---|
| [M+4H]4+ | 1691.85 |
| [M+5H]5+ | 1353.81 |
| [M+6H]6+ | 1128.25 |
| Deconvoluted mass | 6763.75 |
Detected in positive mode with charges 4+ to 6+, which shows the mass can be confirmed without an ion-pair reagent. The two strands are drawn here as separating for illustration.
A water layer
instead.
Hydrophilic interaction chromatography needs no ion-pair reagent. On a Shodex HILICpak VN-50 2D column, with 100 mmol/L ammonium acetate mixed 30:70 into water (A) or acetonitrile (B) and a gradient from 92% B downward, a mipomersen-based 20-mer and its n-1(5′) and n-3(5′) variants eluted at 8.432, 8.956 and 9.300 min: shortest strand first.
When peaks overlap, Insight Biologics splits the UV area by each component's share of the MS signal:
| Check | Result |
|---|---|
| FLP 1–20 µmol/L | r² ≥ 0.999 |
| n-1(5′), n-3(5′) at 2–50% of a 10 µmol/L FLP | all detected and identified |
| n-3(5′) linearity | y = 15705x + 2102.7, r² 0.9995 |
| n-1(5′) linearity | y = 15484x + 53054, r² 0.9996 |
| Mass errors (FLP and variants) | within 0.3 Da |
Sensitivity: MS signal was about one-fifth to one-tenth of that in ion-pair reversed phase, so low concentrations remain a challenge.
Buffer: ammonium bicarbonate (pH 8) also worked, but retention times shortened over consecutive runs as dissolved CO₂ from the air changed the mobile phase, and MS signal was about half that with ammonium acetate. Ammonia at pH 10 gave no gain. Ammonium acetate was preferred.
Upside: with no ion-pair reagent, the system need not be dedicated to this one analysis.