Mo, 31.8.2026 | Original article from: Anal. Chem. (2026) 98 (14): 10572–10583
Propionic acid increases peptide identifications by 39% versus formic acid in LC-MS proteomics while maintaining chromatographic performance and compatibility.Propionic Acid Outperforms Formic and Acetic Acid in MS Sensitivity for High-Flow Reversed-Phase LC-MS Bottom-Up Proteomics
Anal. Chem. (2026) 98 (14): 10572–10583: Figure 2. (A) Base peak chromatograms of 1 μg of bevacizumab peptides separated within a 20 min gradient using the 2.1 × 150 mm Acquity Premier CSH C18 column maintained at 60 °C and mobile phase containing FA, AcA, or PrA. (B) Total identified peptides in analyses of five sample inputs using three additives. Database search was performed with semitryptic specificity, allowing up to two missed cleavages. (C) Distribution of relative change of peak area (AUC), tR, and peak width at half height (w0.5) of 44 representative peptides normalized to those observed using FA-containing mobile phase. The peak areas of all identified precursors were summed. The mean and standard deviation from duplicates are illustrated. (D) Dependence of tR change on peptide isoelectric point (pI) when switching to AcA and PrA from FA with linear regressions. The equations of the linear regression, determination coefficients, and Pearson correlation coefficients are shown below. The retention times of 38 unmodified peptides were evaluated. Colored dots illustrate 90% prediction bands.
This study evaluates propionic acid as an alternative mobile-phase additive to formic and acetic acid in reversed-phase LC-MS bottom-up proteomics. By reducing mobile-phase ionic strength and surface tension, propionic acid improved electrospray ionization efficiency, producing 39% more peptide identifications than formic acid and 12% more than acetic acid.
The improvement was reproduced across laboratories, LC flow regimes, column chemistries, and sample complexities, while chromatographic performance remained largely unchanged. With good stability, instrument compatibility, and negligible background signal, propionic acid represents a practical drop-in alternative for LC-MS workflows requiring greater sensitivity and proteome depth.
The original article
Propionic Acid Outperforms Formic and Acetic Acid in MS Sensitivity for High-Flow Reversed-Phase LC-MS Bottom-Up Proteomics
Mykyta R. Starovoit; Siddharth Jadeja; Rudolf Kupčík; Saša Vatić; Jan Rasl; Derya Demir; Petr Novák; Cameron Braswell; Benjamin C. Orsburn; Juraj Lenčo *
Anal. Chem. (2026) 98 (14): 10572–10583
licensed under CC-BY 4.0
Selected sections from the article follow. Formats and hyperlinks were adapted from the original.
Reversed-phase liquid chromatography coupled with mass spectrometry (RPLC-MS) has become the gold standard in bottom-up proteomics. (1, 2) Several decades ago, a pivotal shift occurred in selecting acidic agents: formic acid (FA) replaced trifluoroacetic acid (TFA) as the default mobile phase additive for RPLC-MS workflows. (3−5) Trifluoroacetic acid, a strong acid with a pKa of 0.23, was initially favored for its dual benefits in RPLC. (6, 7) It maintains most residual silanol groups on silica-based stationary phases in the undissociated state, thereby minimizing unwanted electrostatic interactions. Additionally, its conjugate base, the trifluoroacetate anion, readily couples with protonated peptides and forms stable ion pairs with a reduced net charge, which have lower affinity to dissociated silanol groups and increased retention. These phenomena result in excellent chromatographic performance. However, while ideal for LC-UV analyses, TFA proved to be poorly compatible with electrospray ionization. (8, 9) The ion-pairing mechanism effectively “neutralizes” protonated peptides, shielding them from the electric fields transmitting ions into the ion optics, thus dramatically reducing MS signal intensity. (10) In contrast, formic acid, with a pKa of 3.75 and typically used at a concentration of 0.1%, generates almost 6-fold lower ionic strength in the mobile phase. While this results in weaker ion pairing and less efficient prevention of silanol interactions, it significantly enhances ESI efficiency, leading to higher MS signal intensity. Although the lower ion-pairing capacity and slightly higher pH of FA should theoretically compromise chromatographic performance compared to TFA, this effect is strongly dependent on stationary phase chemistry and peptide properties, and modern RPLC stationary phases have largely mitigated this issue. (11, 12) Innovations such as end-capping, steric shielding, or introducing positively charged groups into a stationary phase surface minimize silanol-related interactions, providing high separation performance even using mobile phases with reduced ionic strength. Our recent study demonstrated that the latter technology allows columns to maintain the separation performance even using mere 0.01% FA, further increasing MS sensitivity through reduced ionic strength. (13, 14) This mechanism is supposedly applicable to a structurally similar acid, acetic acid (AcA), with a pKa of 4.76. At higher concentrations, typically around 0.5%, AcA provides acidity comparable to that of 0.1% FA while maintaining about half the ionic strength. It also reduces the surface tension of the mobile phase and concentrates in ESI droplets for a longer period due to its lower volatility, which further enhances the surface tension-reducing effect. Recent works by Lenčo et al. and Battellino et al. have demonstrated that AcA can significantly improve the number of peptide identifications by increasing MS signal intensity more than 2-fold on average, (15, 16) despite contradictory early reports. (5, 17, 18) These studies indicate that AcA can offer a compelling alternative to FA in bottom-up proteomic workflows, particularly when prioritizing MS sensitivity.
Propionic acid (PrA), a homologous carboxylic acid, has been largely overlooked as a potential additive to the mobile phase. It has mainly attracted attention as a postcolumn additive that enhances MS signal intensity by modifying the composition of eluent droplets containing TFA. (9, 10) In our laboratory, PrA is frequently used as a dopant in the desolvation gas to improve ESI efficiency. (19−21) Its performance inspired us to investigate the direct use of PrA in the mobile phase for LC-MS proteomics. To our knowledge, PrA has been applied only in a single RPLC-UV study investigating peptide retention and as an additive to a TFA-containing mobile phase in HILIC-MS for the analysis of basic drugs. (22, 23) With a pKa of 4.88, 0.5% PrA produces a similar pH to that achieved with FA or AcA, but generates a lower anion concentration, suggesting additional potential for minimizing signal suppression. Moreover, its longer alkyl chain confers lower surface tension and volatility, (24) properties expected to promote droplet formation and enhance ionization efficiency. For further mechanistic discussion, we refer readers to the section “Theoretical Considerations”.
In this study, we hypothesized that 0.5% PrA can outperform 0.5% AcA and 0.1% FA in MS sensitivity as an acidic additive to the mobile phase for RPLC-MS bottom-up proteomics. We compared its impact on ionization efficiency, separation performance, retention, and performance in peptide identifications to established setups using FA and AcA. The experiments were conducted independently at four research facilities, following local expertise and without constraints imposed by the principal investigators. We examined different column chemistries, including positively charged C18-, traditional C18-, and polyphenyl-bonded stationary phases, and evaluated performance for the samples of various complexity using a range of peptide sample loads. Recognizing the diversity of experimental setups in proteomics, our study encompassed analytical-, micro- (collectively referred to as high-flow), and nanoflow regimes, as well as MS instruments from two leading vendors, employing both standard and nanoESI sources, and both DDA and DIA acquisition modes. Additionally, we addressed the practical aspects of routine PrA use, including mobile phase stability and instrument compatibility, which was evaluated by GC-MS and ICP-MS analysis of leachables from the LC system. The findings presented here explore the utility of PrA as an alternative eluent additive for proteomic analyses requiring maximum sensitivity and extend prior investigations into AcA, (15, 25) particularly concerning separation performance and in-column artificial modifications.
Experimental Section
Instruments
Analytical- and microflow LC-MS analyses at the Faculty of Pharmacy in Hradec Králové (FPh) were performed using a Vanquish Horizon UHPLC system coupled to a Q Exactive HF-X mass spectrometer (Thermo Fisher Scientific) operating in positive ion mode with electrospray ionization via a HESI-II probe. At the Biomedical Research Centre at University Hospital Hradec Králové (BRC), micro- and nanoflow analyses were performed using a Dionex Ultimate 3000 UHPLC system hyphenated to a Q Exactive Plus mass spectrometer with a HESI-II probe and a Dionex Ultimate 3000 RSLCnano system coupled to an Orbitrap Exploris 480 mass spectrometer equipped with a NanoSpray Flex NG ion source and a FAIMS Pro Duo interface, respectively (Thermo Fisher Scientific). Experiments at the Organ Pathobiology and Therapeutics Institute, University of Pittsburgh (OPTIn) employed a nanoflow Evosep One LC system (Evosep Biosystems) interfaced with a timsTOF Ultra 2 mass spectrometer (Bruker Daltonics). Analyses at the Institute of Microbiology of the Czech Academy of Sciences in Prague (IMB) were conducted using a nanoflow Evosep One LC system coupled to a timsTOF SCP mass spectrometer (Bruker Daltonics). Detailed ion source and mass analyzer settings are specified in the Supporting Information (Tables S1 and S2). Unless otherwise stated, analyses were performed in triplicate. The LC-MS files from FPh, BRC, and IMB were deposited in the ProteomeXchange repositories via PRIDE with the identifiers PXD069554 and PXD070747 , and from OPTIn via MassIVE with the identifier MSV000099496. (26)
LC-UV determination of acidic additives in incubated samples of mobile phase was performed on a UltiMate 3000 RSLC system (Dionex) equipped with a diode array detector DAD-3000 RS and a 2.5 μL flow cell. The 54 elements, including heavy metals, released during eluent circulation were quantified using an Agilent 7900 ICP-MS system equipped with a collision cell ORS4. GC-MS profiling of volatile residues was conducted using an Agilent 7890 A system interfaced with an Agilent 5975 inert mass spectrometer operating in EI mode at 70 eV (Agilent Technologies).
Results and Discussion
Effects of Propionic Acid on Peptide Mapping of Monoclonal Antibody
Tryptic digests of monoclonal antibodies typically yield a few dozen unique peptides, enabling statistically robust analysis of strong population-level dependencies. At the same time, they provide high individual peptide concentrations, ensuring consistent surpassing of DDA intensity thresholds, which facilitates the detection of low-abundance chemical modifications and multiple peptide precursors. Therefore, bevacizumab peptides were separated at multiple injected mass loads using a column with the same analytical i.d. and positively charged C18 chemistry as those used for iRT and Alberta peptides. AcA and PrA increased the peak area of individual peptides by 53% and 113% on average, resulting in a greater number of peptide identifications in analyses of all sample quantities (Figure 2).
Anal. Chem. (2026) 98 (14): 10572–10583: Figure 2. (A) Base peak chromatograms of 1 μg of bevacizumab peptides separated within a 20 min gradient using the 2.1 × 150 mm Acquity Premier CSH C18 column maintained at 60 °C and mobile phase containing FA, AcA, or PrA. (B) Total identified peptides in analyses of five sample inputs using three additives. Database search was performed with semitryptic specificity, allowing up to two missed cleavages. (C) Distribution of relative change of peak area (AUC), tR, and peak width at half height (w0.5) of 44 representative peptides normalized to those observed using FA-containing mobile phase. The peak areas of all identified precursors were summed. The mean and standard deviation from duplicates are illustrated. (D) Dependence of tR change on peptide isoelectric point (pI) when switching to AcA and PrA from FA with linear regressions. The equations of the linear regression, determination coefficients, and Pearson correlation coefficients are shown below. The retention times of 38 unmodified peptides were evaluated. Colored dots illustrate 90% prediction bands.
With a higher sample complexity, we observed an average tR decrease of 0.8% using AcA and 0.1% using PrA, compared to FA (Figure 2). Deviations in tR were primarily driven by peptide acid–base properties, as indicated by Pearson correlation coefficients of r = −0.70 (p < 0.0001) and r = −0.53 (p = 0.0007) for AcA and PrA, respectively, showing a tR decrease with increasing pI. Switching to PrA led to an average w0.5 increase of 4.1%, while AcA broadened peaks by only 0.9% (Figure 2). Both results are generally negligible when considering the advantages of alternative acidic additives in ESI enhancement. The more efficient generation of multiply charged precursors followed the same trend as for the model Alberta peptides (Figure S2).
Monitoring the common modification sites in the bevacizumab structure, we found that the relative quantities of modified peptides increased slightly upon replacing FA with AcA and PrA, even with a short 20 min separation method and a moderately elevated column temperature of 60 °C (Figure 3). An extended 90 min gradient separation at 80 °C corroborated our observations. Together with the previously observed increase in artificial modifications at a lower FA concentration of 0.01%, (13) these results indicate that elevating the mobile phase pH adversely affects the abundance of commonly monitored modifications. The increase in artifact levels cannot be attributed to improved MS sensitivity, as it would proportionally increase the signals of both modified and unmodified peptides, leaving their AUC ratio unchanged. Therefore, maintaining pH at 2.7 using 0.1% FA and avoiding elevated column temperature appears to be the most efficient ways to prevent artificial modifications.
Anal. Chem. (2026) 98 (14): 10572–10583: Figure 3. Relative abundance of the modified peptide forms in the 20 and 90 min separations of bevacizumab peptides using the 2.1 × 150 mm Acquity Premier CSH C18 column maintained at 60 and 80 °C. The abundance was calculated as the peak area of all the precursors of the modified peptide divided by the summed area of both peptide forms. The most abundant modified peptide containing the modified amino acid was used. Abbreviations: Lc – light chain, Hc – heavy chain. The superscripted numbers correspond to the position of the modified amino acid in the chain sequence.
Instrument Compatibility and Mobile Phase Stability
In contrast to the analysis of small molecules, the composition of mobile phases in proteomics workflows is seldom reoptimized once established. This conservatism likely stems from the field’s reliance on empirically validated formulations and from concerns regarding unexplored effects on instrument performance or data quality. To address such concerns, we deemed it essential to assess the compatibility of a new additive with respect to instrument safety, mobile phase stability, and MS background noise. Formic acid, a default additive, is widely considered safe and fully compatible with routine operation. To our knowledge, no study has explicitly assessed the safety profile of AcA. Nevertheless, several proteomics groups have used AcA without reported complications, (15, 63) and its application is also widespread outside the proteomics field. (64, 65) Since PrA is a weaker acid than both FA and AcA, and a 0.5% solution yields a pH within the operational range of standard instrumentation, we did not expect its safety profile to be inferior.
In the LC-MS system, the mobile phase comes into contact with various components, including storage containers, chromatographic columns, tubing, capillaries, pump head metal components, and seals. These are typically made of glass, stainless steel, titanium or titanium-based alloys, fused silica, PEEK, and other organic polymers. Trace amounts of these materials may leach into the mobile phase and be carried into the mass spectrometer. Nonvolatile substances can accumulate in the MS front end, potentially causing contamination, while others may persist in mass spectra, elevating background noise. Quantifying the spectrum of compounds in the mobile phase that has passed through the entire LC system offers a practical way to evaluate leaching rates, thereby providing insight into the safety of the mobile phase.
To compare PrA to FA, we allowed the acidified 50% ACN to circulate incessantly through the LC instrument at a high flow rate for 7 days, while another portion of the sample was stored in a glass container. The exploited LC instrument had no polymeric lining that prevents the mobile phase from contacting metal surfaces. The flow path did not include the chromatographic column because of its ability to retain trace amounts of metals. (66) We believe that reducing mobile phase acidity would only improve column lifetime, and searching for the opposite effect is unnecessary. The samples were quantified for 54 elements by ICP-MS, nontargeted direct infusion ESI-MS, and GC-MS (Table S3).
ICP-MS analysis revealed increased concentrations of Fe, Cr, Ni, Cu, Mo, and Mn, indicating the release of these elements from stainless steel components (Figure S9). With an increased iron concentration, we detected positively charged ions in ESI-MS spectra at m/z 548.94, 621.97, 696.01, and 770.04, corresponding to carboxylate oxygen-centered triangular complexes formed between iron and PrA. (67) Analogous complex formation has been reported for AcA at m/z 538.96, which may suggest similar metal-leaching properties. A minor increase in abundance of (2ACN+Cu)+ ions with m/z of 144.98 and 146.98 was also observed along with the increased copper levels. The boron concentration remained unchanged, confirming that the mobile phase additives had no effect on the leaching of glass. The quantities of trace elements typical for glass, such as Na and K, fell below the lower limit of quantitation of 10 and 55 ppb, respectively, in all the samples. The limits of quantification for these metals are close to the common maximum concentrations allowed by LC-MS solvent manufacturers in their products (Table S3), so any increase in their concentrations due to switching to PrA would still not exceed these limits. An increased concentration of Co in samples stored in glass containers remained unexplained, as laboratory glass may contain only trace amounts of this metal. An aggregate mass of all the quantified elements leached by FA and PrA within the experiment was 48.6 ± 0.1 μg and 86.8 ± 1.3 μg, respectively, representing a 1.8-fold increase. Given that the samples completed almost 81 full circulation cycles through the LC instrument, an extrapolated amount of elements leached under normal operational conditions is insignificant. GC-MS profiling detected no anticipated contaminants. The concentrations of all mobile phase additives remained constant throughout the 7-day stability examination, with RSDs of <1% and <3% for the mobile phase additives and pH, respectively.
The intensity of the MS background noise of the mobile phases was comparable across additives and even decreased for AcA and PrA at lower flow rates (Figure S10). Metal-associated ion clusters in the range of 530–630 m/z were observed only in the AcA spectra, while the increased noise in the PrA spectra was primarily caused by ions below 350 m/z. In contrast to DMSO, the use of alternative additives had no long-term effects on the MS background. (52) In addition, PrA did not exhibit any unpleasant odor during LC-MS operation. When mobile phases were prepared in a fume hood, the handling of PrA was odor-neutral. During the study, the LC-MS grade PrA from Honeywell was discontinued; therefore, starting with the analyses of bevacizumab peptides, we used the p.a. grade PrA (≥99.5% GC purity) from Merck/Sigma-Aldrich. Surprisingly, the p.a. grade product produced lower background noise (TIC 2.2 × 106 vs 2.7 × 106), which dispelled the purity-related concerns.
Conclusion
Small organic acids are traditionally favored as acidic additives to the mobile phase in RPLC-MS bottom-up proteomics analyses, with formic acid long regarded as the gold standard. Recently, acetic acid has been revisited as a superior alternative that was largely abandoned decades ago, following early reports showing its advantages over FA. In this study, we broadened the scope of applicable acidic additives within the homologous series of carboxylic acids by introducing propionic acid. Acknowledging the general conservatism within the proteomics community, evidenced by the slow adoption of even clearly beneficial methodological advances, we systematically investigated a broad range of aspects related to PrA utilization.
Due to its lower ionic strength, surface tension, and volatility, 0.5% PrA significantly enhanced electrospray ionization efficiency and outperformed 0.5% AcA, 0.01% FA, and 0.1% FA in terms of MS sensitivity using analytical- and microflow configurations. This resulted in an average 12% increase in peptide identifications under PrA conditions compared to AcA, with the greatest benefit in AUC for acidic and hydrophilic peptides. In contrast, no improvement was observed in the nanoLC configuration, consistent with the comparable ionization efficiency of nanoESI sources using both additives. Similarly to AcA, an additional increase in identifications was achieved when it was combined with DMSO.
Chromatographic performance under PrA conditions remained comparable to that with FA, with peptide peak widths increasing by no more than 5% on average. Due to its relatively weaker ion-pairing properties, PrA induced a minor reduction in peptide retention, most notably affecting the hydrophilic species, in contrast to the greatest signal increase. The most pronounced retention decrease, previously observed for AcA in high-pI and weakly retained peptides, was also confirmed for PrA, with slightly higher significance. Therefore, PrA may not be the ideal additive for workflows focused on hydrophilic peptides, especially when using trap-elute configurations. Nevertheless, for most standard proteomics applications, PrA consistently outperformed other additives in terms of peptide identifications, even when used with columns that exhibited the most pronounced peak broadening and retention decrease.
The use of PrA in combination with elevated column temperatures should also be used with caution in studies of post-translational modifications that may also occur in the column and interfere with the modifications of interest formed before the analytical phase. Although we initially hypothesized that reducing mobile phase acidity would lower the abundance of artificial modifications, our results indicated that 0.1% FA remains the safest known additive in this context. Still, we believe that the observed increase in modifications under PrA conditions is unlikely to negatively impact standard proteomics experiments that are not explicitly focused on post-translational modifications.
Furthermore, mobile phases containing PrA were fully compatible with LC-MS instrumentation, did not increase MS background noise, and remained stable over standard storage durations. Taken together with its functional advantages, we conclude that adopting propionic acid represents a simple, low-cost, and powerful strategy to substantially enhance proteomic performance in high-flow LC-MS analyses, offering an attractive step before pursuing more extensive and expensive instrumental optimizations.
