Fine-Tuning of Label-Free Single-Cell Proteomics Workflows
More
Advanced Search
  • Free to Read
  • Editors Choice
Technical Note

Fine-Tuning of Label-Free Single-Cell Proteomics Workflows
Click to copy article linkArticle link copied!

  • Pauline Perdu-Alloy
    Pauline Perdu-Alloy
    Laboratoire de Spectrométrie de Masse BioOrganique (LSMBO), IPHC UMR7178, CNRS, Université de Strasbourg, 25 Rue Becquerel, Strasbourg, Grand Est 67087, France
    Infrastructure Nationale de Protéomique ProFI, UAR2048, Strasbourg 67087, France
    More by Pauline Perdu-Alloy
  • Charline Keller
    Charline Keller
    Laboratoire de Spectrométrie de Masse BioOrganique (LSMBO), IPHC UMR7178, CNRS, Université de Strasbourg, 25 Rue Becquerel, Strasbourg, Grand Est 67087, France
    Infrastructure Nationale de Protéomique ProFI, UAR2048, Strasbourg 67087, France
    More by Charline Keller
  • Anjali Seth
    Anjali Seth
    Cellenion SASU, 60 Avenue Rockefeller, Bioserra2, Lyon, Auvergne-Rhône-Alpes 69008, France
    More by Anjali Seth
  • Christine Carapito*
    Christine Carapito
    Laboratoire de Spectrométrie de Masse BioOrganique (LSMBO), IPHC UMR7178, CNRS, Université de Strasbourg, 25 Rue Becquerel, Strasbourg, Grand Est 67087, France
    Infrastructure Nationale de Protéomique ProFI, UAR2048, Strasbourg 67087, France
    *Email: [email protected]
    More by Christine Carapito
    Orcidhttps://orcid.org/0000-0002-0079-319X
Open PDFSupporting Information (1)

Journal of Proteome Research

Cite this: J. Proteome Res. 2026, 25, 4, 2200–2208
Click to copy citationCitation copied!
https://doi.org/10.1021/acs.jproteome.5c01075
Published March 4, 2026

Copyright © 2026 The Authors. Published by American Chemical Society. This publication is licensed under these Terms of Use.

Request reuse permissions

Abstract

Click to copy section linkSection link copied!
High Resolution Image
Download MS PowerPoint Slide

Mass spectrometry-based single-cell proteomics emerges as the most promising method for studying cellular heterogeneity at the global proteome level with unprecedented depth and coverage. Its widespread application remains limited due to robustness, reproducibility, and throughput requirements, still difficult to meet as analyzing large cohorts of single cells is necessary to ensure statistical confidence. In this context, we conducted method optimizations at three levels. First, we benchmarked three distinct workflows compatible with the nanoElute2 platform using different sample collection/preparation plate supports (EVO96 oil-free, LF48 oil-based, and LF48 oil-free, a streamlined automated sample resuspension, and direct injection protocol). Then, we compared the optimized EVO96 workflow on nanoElute2 with Evosep-based separations operating at two analytical throughputs (80 and 120 samples per day). Subsequently, we evaluated digestion efficiency using a range of enzyme/protein ratios (1:1; 10:1; 20:1; 50:1) to maximize peptide recovery. Finally, the chromatographic setup was refined to determine the best compromise between throughput and robustness. Altogether, these optimizations allowed to establish a robust workflow quantifying up to 5000 proteins in 10 min gradient time per single HeLa cell at a 55 samples-per-day throughput.

This publication is licensed for personal use by The American Chemical Society.

Copyright © 2026 The Authors. Published by American Chemical Society

Subjects

what are subjects

Keywords

what are keywords

1. Introduction

Click to copy section linkSection link copied!
Qualitative and quantitative analyses of the whole proteome highlight the response of an organ or population of cells to diseases or treatments. However, under varying conditions and overtime, multicellular organisms exhibit great heterogeneity in their protein expression patterns. (1−3) In this complexity, bulk protein analysis is unable to distinguish healthy from diseased/infected cells (4−6) and identify subclonal populations that behave differently from the rest of the cell population (e.g., in cancer). (7−10) Thanks to constant technological progress, the investigation of the proteome of single cells has now become feasible but still faces analytical challenges, mainly due to its need for sample miniaturization (picoliter volume dispense), highly sensitive instrumentation (enabling detection of 80–400 pg of protein input material), and sophisticated computational tools (for decomplexification of high-throughput data sets (11,12)).
Tailored sample preparation protocols are a crucial step of the workflow, and since its introduction in 2017, the cellenONE robot has transformed single-cell sorting using an image-based technology to select cells based on their morphological parameters, without the need for labeling. Combined with recent breakthroughs in ion mobility-mass spectrometry using ultrafast and ultrasensitive mass analyzers, latest generation couplings have allowed improvement of the resolution and depth of single-cell proteome analysis. In particular, the creation of new precursor scan designs and their fragmentation management in tandem with their accumulation, enabled by the dia-PASEF (Data-Independent Acquisition-Parallel Accumulation Serial Fragmentation) strategy (13−15) enhance selectivity and efficiency.
In this context, we have undertaken a series of method optimizations at all steps of the workflow to further increase the single-cell proteome depth and coverage.

2. Experimental Procedures

Click to copy section linkSection link copied!

2.1. Single Cells’ Preparation on the CellenONE System

In-house cultured HeLa cells were resuspended to a final concentration of approximately 200 cells/μL in DPBS and then loaded into a piezoelectric dispensing M-size capillary in the cellenONE system (Cellenion, Lyon France), with droplet generation settings of 97 V, 48 μs, and 500 Hz.
A 300 nL solution of Mastermix (16) containing lysis and digestion reagents (100 mM TEAB, ref: 15658320 (ThermoFisher Scientific); 0.2% DDM, 850520P-1G (Sigma-Aldrich); and 10 ng/μL trypsinLysC, ref: V5072 (Progema Corporation)) was dispensed into each well of the substrate of choice (either an EVO96, an LF-48 with oil, or an oil-free LF-48 proteoCHIP) at 10 °C with 50% humidity. Single-cell sorting followed fixed parameters for optimal selection: intensity min 1 and max 255; elongation max 2; and diameter min 16 μm and max 40 μm. The detection was set for diameters of min 2 μm and max 40 μm and for elongation of max 4. This isolation step was performed at 10 °C with 50% humidity.
After dispensing, single cells were incubated at 50 °C for 1.5 h from 10 °C to 20 °C at 75% humidity and from 20 °C to 50 °C at 85% humidity, with rehydration cycles (260 nL/well, 500 Hz) to prevent evaporation of the lysis buffer. Samples were then cooled to 20 °C with 85% humidity, maintaining rehydration cycles throughout. Digestion was then halted by manually adding 3.2 μL of 2% acetonitrile (ACN) and 0.1% formic acid (FA) per well at room temperature, except for the LF48 oil-free protocol, for which the LF48 chip was transferred from the cellenONE to the nanoElute2 autosampler right after incubation.
For EVO96 support injected on the nanoElute system, thanks to a centrifugation step at 700g for 30 s, the peptide mixture was then transferred and collected to a 96-well injection plate, which was stored at −80 °C before mass spectrometry analysis. For EVO96 support injected on the Evosep system, after EVOTIPs cleanup procedure, peptides were loaded on the EVOTIPs and cleaned up by centrifugation at 800g for 60 s. The peptide mixture was then analyzed by mass spectrometry. For the LF48 oil-based support, after a cooling step at 4 °C in a fridge, the liquid peptide mixture was collected and transferred to a 96-well injection plate, which was stored at −80 °C before mass spectrometry analysis.
For LF48 oil-free support, the plate was collected directly after the incubation step and transferred from the cellenONE to the nanoElute autosampler, without storage or freezing. Each single cell was automatically resuspended in 1.5 μL of 2% ACN, 0.015% DDM, and 0.1% FA (final concentrations) by the nanoElute system right before injection and sequential analysis.
For column length and gradient optimizations, the LF48 oil-based workflow was applied with an enzyme-to-protein ratio of 10:1, while for enzyme-to-protein ratio optimizations, the EVO96 support was used.

2.1.1. Nanoliquid Chromatography

nanoElute2 system. Peptide mixtures were separated on a 5 cm (or 25 cm for the LC–MS/MS method optimization experiments) C18-RP analytical column (75 μm inner diameter, 1.7 μm particle size, AuroraRapid, IonOpticks) heated to 50 °C using the nanoElute2 nanoliquid chromatography system (Bruker Daltonics, Bremen, Germany). Separation was performed using a linear gradient from 5 to 35% of solvent B (solvent A: 0.1% FA in H2O and solvent B: 0.1% FA in ACN) over a 10 min gradient at 0.25 μL/min flow rate for the 5 cm column and over a 22 min gradient for the 25 cm column. Column washing and regeneration were performed in 6 min by ramping the percentage of mobile phase B from 35% to 90% and then back to 5%.
Evosep system. Peptides mixtures were separated with the same column as for the nanoElute system on the Evosep One system (Evosep Biosystems, Odense, Denmark). Peptides’ separation was performed using Whisper Zoom methods 80 SPD and 120 SPD. Separation was performed using a gradient from 0 to 40% of solvent B (solvent A: 0.1% FA in H2O and solvent B: 0.1% AF in ACN over a 16.3 min and a 10.3 min gradient at 0.2 μL/min flow rate, respectively).

2.1.2. Tandem Mass Spectrometry

Separated peptides were analyzed on a timsTOF Ultra 2 mass spectrometer (Bruker Daltonics, Bremen, Germany) operated in the dia-PASEF mode. The instrument settings included a capillary voltage of 4.5 kV, with a gas flow rate maintained at 3.0 L/min at 200 °C. Fragmentation windows, with a width of 25 m/z, were established from 0.64 to 1.37 V s/cm2 along the ion mobility range, covering a mass-to-charge ratio (m/z) range from 400 to 1000. The accumulation time for each ion mobility scan was set at 100 ms with a ramp time of 100 ms. Collision energies were applied across the ion mobility range, starting from 0.60 V·s/cm2 at 20 eV and increasing to 1.60 V·s/cm2 at 59 eV to optimize ion fragmentation. Each cycle, lasting 0.96 s, comprised a full MS1 scan, followed by 24 MS/MS windows. Within each of these windows, ions were fragmented across 8 MS/MS ramps. Complete data set has been deposited in the ProteomeXchange Consortium via the PRIDE partner (17) repository with the data set identifier PXD067019.

2.2. Data Treatment

Raw instrument files collected from dia-PASEF were analyzed through DIA-NN (18) (v. 1.8.2beta27). The processing was performed using a library-free mode. In silico digestion using Trypsin/LysC was performed on a human proteome fasta (20,409 reviewed Swiss-Prot entries), allowing a maximum of 1 missed cleavage. The theoretical peptide spectra were predicted according to their retention time and ion mobility thanks to a deep learning-based algorithm provided by DIA-NN. (19) For peptide spectrum matches (PSMs), precursors were filtered at 1% FDR. Variable modifications on peptides were set to methionine oxidation and N-term acetylation with a maximum of 5 variable modifications. For data processing, the peptide length ranged from 7 to 30 amino acids; precursor charge ranged from 1 to 4 whereas their m/z ranged from 300 to 1800 and finally, fragment ions m/z ranged from 200 to 1800. All mass accuracies, notably the precursor masses (MS1) and fragment masses (MS2), were defined at a tolerance of 10 ppm. Match between runs (MBR) was checked for cross-run analysis considering single-cell replicates only.

3. Results and Discussion

Click to copy section linkSection link copied!

3.1. Sample Preparation Plate Support Optimization

Single-cell proteome preparation workflows consist of five main steps: (1) dispensing lysis and digestion mixture (Mastermix), (2) single-cell isolation, (3) incubation, (4) peptide dilution, and (5) most often peptide transfer into an injection plate. Steps 1 to 3 are common to all workflows. In our case, the Mastermix contains Trypsin-LysC at 10 ng/μL, 0.2% n-dodecyl β-d-maltoside (DDM), and 100 mM triethylammonium bicarbonate (TEAB) in initial concentrations. Single cells are isolated into the collection plate wells, followed by an incubation step of 1.5 h at 50 °C, allowing for cell lysis, protein extraction, and digestion into peptides. To assess potential peptide losses and sources of human-introduced variability─primarily caused by evaporation, solvent dilution (i.e., a small final peptide volume of 0.3 μL diluted in a relatively large recovery solvent volume of 3.2 μL), and manual pipetting─four workflows based on various sample collection plate supports were benchmarked as illustrated in Figure 1:

  • Option 1─The LF48 oil-based support, the original cellenOne protocol, for the collection of 48 single cells per chip, integrating a thin hexadecane layer requiring a final manual sample transfer.

  • Options 2 and 3─The EVO96 support, an oil-free proteoCHIP for the single-cell collection (of up to 96), with hands-free peptide transfer via a short centrifugation step. Although originally developed for coupling with the Evosep system (Option 2), we also attempted its adaptation for nanoElute injections (Option 3).

  • Option 4─The LF48 oil-free support enabling automated resuspension and injection, using a custom-designed proteoCHIP holder compatible with the nanoElute2 autosampler. In this workflow, the proteoCHIP was transferred from the cellenONE to the nanoElute autosampler as soon as the cell isolation/digestion was completed (no storage, no freezing), with each single-cell digest in approximately 300 nL (corresponding to the Mastermix volume). Each single-cell sample was automatically resuspended and rehydrated by the nanoElute system in 1.5 μL of 2% ACN, 0.015% DDM, and 0.1% FA (final concentrations) right before injection.

Figure 1

Figure 1. Four single-cell sample preparation protocols benchmarked using the cellenONE sorter/liquid dispenser. Lysis and digestion, followed by incubation with automated rehydration cycles are common steps to all three workflows. Then, option 1 relies on a LF48 oil-based plate support involving a manual dilution step in liquid oil, followed by a manual peptide transfer after oil solidification (Tf (hexadecane) = 18 °C); option 2 uses an EVO96 oil-free plate support, involving manual dilution followed by peptide loading onto preconditioned EVOTIPs after centrifugation for injection on the Evosep system; option 3 employs the same EVO96 oil-free support, repurposed for nanoElute injection, where peptides are transferred into a 96-well injection plate via centrifugation; and option 4 relies on a LF48 oil-free plate support suited for automated resuspension and direct injection on a nanoElute 2 system.

High Resolution Image
Download MS PowerPoint Slide
Figure 2 summarizes the results obtained with the three evaluated workflows compatible with the nanoElute 2 platform (Options 1, 3, and 4). In terms of proteome coverage, among the three evaluated workflows, the LF48 oil-free direct injection protocol enables the highest coverage with a median and average 4682/4437 protein groups and 30,995/29,411 peptides identified.

Figure 2

Figure 2. Benchmarking of three single-cell proteomics workflows (EVO96 (blue), LF48 oil-based (green), and LF48 oil-free (orange)) run on nanoElute 2 on 96 single-cell replicates per workflow. (A) Violin plots representing the number of protein groups identified per single cell for each tested workflow using DIA-NN with MBR (left) and without MBR (right). (B) Violin plots representing the number of unique peptides identified per single cell for each tested workflow using DIA-NN with MBR (left) and without MBR (right). (C) Venn diagram showing proteins’ overlap across the three workflows. (D) Coefficients of variation (CVs) density distributions across the three sample preparation workflows. (E) Protein abundance ranking plot comparing the dynamic range of protein groups across the three workflows. Black dots correspond to the shared proteins across the conditions, whereas the colored ones correspond to unique proteins per condition.

High Resolution Image
Download MS PowerPoint Slide
The lowest protein groups/peptides (2114/2064; 10,581/10,792 median and average) numbers are reached with the EVO96 collection plate support, while the LF48 oil-based protocol shows intermediate coverage performances (3114/3109; 17,774/18,065 median and average protein groups/peptides, respectively) (Figure 2A,B). Importantly, although MBR increases the number of identifications by reducing the number of missing values with elution profile recovery from other single cells analyzed in parallel, the relative performance trend between workflows remained unchanged. It confirms that the differences observed are robust and not partially due to the MBR processing. Furthermore, the EVO96 workflow offers an efficient and user-friendly format thanks to the centrifugation step, which eliminates manual pipetting and reduces hands-on time allowing for a total preparation time of ∼3 h for 192 cells (96 × 2 plates) but the trade-off is a slight reduction in sensitivity when compared to the LF48 plate supports.
The higher coverage achieved with the oil-based LF48 plate support reflects the benefit of the hexadecane oil layer, which, in addition to the regular and automatic water dispensing aimed at minimizing evaporation, provides an additional protective layer that reduces evaporation effects. However, the highest coverage is reached with the oil-free direct injection LF48 support, which highlights the strong negative effects of sample dilution and sample transfer (whether done manually or by centrifugation). Indeed, thanks to the adapter plate support compatible with the nanoElute2 automater, single-cell peptide extracts can be directly resolubilized and injected from the well without any intermediate transfer or dilution step. Of note is that with this protocol, the automated injector resuspension volume could be reduced to 1.5 μL (compared to the 3.2 μL manually added in options 1, 2 and 3) and thus the lower dilution factor may have limited adsorption phenomena. This streamlined workflow minimizes sample handling and eliminates peptide losses associated with adsorption to surfaces, evaporation, or pipetting. However, a noticeable drawback of this workflow is that it shows more outliers than the other two protocols. This remains true despite good, reproducible and stable instrument performances over time as attested by the seven QC samples (250 pg HeLa digest) intermittently injected over the 96 single-cell replicates injection sequence (Supporting Information Figure 1A). This increase in the number of outliers is possibly attributable to time-dependent effects (Supporting Information Figure 1B) which contribute to the separation of two distinct cell populations in the PCA─suggesting that further optimizations─particularly in plate preservation and injection throughput─could help limit this drawback.
In terms of proteome overlap, the oil-free LF48 resulted in the highest proportion of unique protein identifications (22%), while 2365 protein groups (41%) were shared among all three conditions (Figure 2C). Proteins uniquely identified with oil-based LF48 and EVO96 are marginal and represent less than 5% of the total proteins.
In addition, in order to further evaluate the consistency of the proteome across different workflows, we calculated the missing values’ percentages (Supporting Information Figure 2A) for the three workflows. LF48 oil-free shows the lowest fraction of missing values (15% and 36% at protein and precursor levels, respectively) when compared to EVO96 (32% and 64% at protein and precursor levels, respectively) and LF48 Oil-based (30% and 48% at protein and precursor levels, respectively). This is in line with the higher coverage achieved with the LF48 oil-free protocol. In addition, we have also applied a data completeness threshold of 75% to all single-cell samples (Supporting Information Figure 2B). A highly comparable overall completeness was obtained across workflows, with percentages of cells with more than 75% of identified proteins being all above 80%. The EVO96 workflow shows the highest consistency, with 98% of the cells satisfying the 75% completeness filter. Furthermore, the analysis of single-peptide hits (Supporting Information Figure 2C) revealed that EVO96 contained a higher proportion of unique peptide identifications (∼20%) compared to the LF48 oil-free and oil-based workflows (11% and 14%, respectively), which is again consistent with the higher coverage achieved with the LF48 workflows. It should be noted that single-peptide hits rarely contained modifications (6% in EVO96, 1% in oil-free LF48, and 2.5% in oil-based LF48).
In terms of quantitative reproducibility, the coefficients of variation (CVs) density distributions revealed lower variability under the LF48 oil-free condition, followed by the EVO96 workflow (Figure 2D). The LF48 oil-based workflow showed the highest variability, suggesting that despite its evaporation-protective role, the manual peptide transfer introduces additional and unavoidable variability. Oil-free and EVO96 plate supports show good and equivalent performances in terms of quantification reproducibility. The LF48 oil-free condition also achieved the widest dynamic range, enabling the detection of less abundant proteins (Figure 2E). Protein abundance correlation curves for all common proteins identified across all three workflows show that overall protein abundances correlate quite well with 98% of proteins correlating with less than 20% difference in abundance between EVO96 and oil-based LF48 protocols. Slightly weaker correlations are observed between oil-free LF48 and both oil-based LF48 and EVO96, which can be explained by the overall higher abundances measured with the oil-free LF48 protocol (Supporting Information Figure 3A).

Figure 3

Figure 3. Comparison of SCP EVO96-based workflows across two chromatographic systems. EVO96_nE was analyzed on a nanoElute 2 system (N = 96 single-cells) compared to two Evosep workflows that were acquired using two sample throughputs (80 and 120SPD with N = 40 single-cell replicates each). (A) Violin plots representing the number of protein groups identified per single cell for each workflow (EVO96_nE in blue, EVOSEP80SPD in gray, and EVOSEP120SPD in purple). (B) Venn diagram showing the overlap of identified protein groups across the workflows. (C) Violin plots representing the number of unique peptides identified per single cell for each workflow (EVO96_nE in blue, EVOSEP_80SPD in gray, and EVOSEP_120SPD in purple). (D) Density distributions of protein abundance coefficients of variation (CVs) across the three workflows.

High Resolution Image
Download MS PowerPoint Slide
To further assess the nature of identified peptides, the hydrophobicity of each detected peptide sequence was evaluated using the GRAVY (Grand Average of Hydropathy) score across the three workflows. The GRAVY score is calculated as the average of the hydrophobicity values of all amino acids in a peptide sequence according to the Kyte and Doolittle scale. (20) Thus, a positive GRAVY score indicates the presence of more hydrophobic peptides; while a negative score indicates the presence of more hydrophilic peptides. Supporting Information Figure 4 shows the GRAVY score average for each single cell across the three workflows. Although the resulting scores are within the same range (from −0.6 to 0.25), it is interesting to note that the EVO96_nE workflow tends to lead to more hydrophobic peptides’ recovery (with an approximate median GRAVY of −0.1), while the LF48 workflows cover a higher proportion of hydrophilic peptides (with an approximate median GRAVY of −0.4). These differences may reflect the composition of the injection plates: as the Teflon plates used in LF48 could inherently favor the loss of hydrophobic peptides, while the polypropylene plates used in EVO96 could favor the loss of hydrophilic ones.
Moreover, we also considered the number of peptides carrying methionine oxidations (UniMod: 35). The distribution of oxidized peptides differs depending on the workflow (Supporting Information Figure 5). Indeed, the EVO96_nE and LF48 direct injection workflows show lowest median numbers of oxidized peptides (80 ± 39; 35 ± 12 respectively), in contrast to the LF48 oil-based workflow (using hexadecane) that shows a wider distribution and greater heterogeneity (119 ± 80) concerning this oxidation phenomena. This suggests that the presence of oil during the preparation process may increase the variability of methionine oxidations between single cells, potentially underlying differences in peptide exposure to oxidative conditions.
As a conclusion, we demonstrate hereby that every single step of the sample preparation workflow can drastically alter identification results and drive a potential loss of proteome coverage and robustness. Sample preparation needs to be finely tuned. Furthermore, we are aware that the use of the EVO96 well plate coupled with nanoElute is diverted from its original purpose, which is generally to be integrated into downstream cleanup analysis on the Evosep system. Therefore, we compared the EVO96 workflow on the two chromatographic systems in the following section.

3.2. Comparison of the EVO96 Workflow on Two Chromatographic Platforms

Then, we compared the EVO96 plate adapted to nanoElute (EVO96_nE) with the original Evosep-based workflow at two different analysis throughputs, specifically 80 and 120 samples per day (SPD). To ensure a fair comparison, the EVO96_nE workflow was performed at a “80-SPD-like” throughput, which closely matches the analysis gradient of the 80SPD on Evosep.
Our results, presented in Figure 3A,C, shows an increase in the number of protein and peptide identifications across all Evosep workflows. In contrast to the EVO96_nE workflow that delivers 2064 protein groups and 10,792 peptides, the Evosep workflow identified 3754/3696 and 25,109/24,065 protein groups/peptides at 80SPD and 120SPD, respectively.
An increase of approximately 80% in protein identifications and up to 130% at the peptide level was observed using Evosep when compared to the EVO96_nE workflow. Its tips (called EVOTIPs) work as a SPE-like purification and preconcentration step, protecting the nanoLC system and ensuring that unlike nanoElute injections, where the entire cell content is loaded on a column in direct injection mode and only the eluted peptide fraction is predominantly transferred to the mass spectrometer. This coupling allows for reaching a high robustness and sensitivity with peptide recovery enhancement. In addition, thanks to this functioning, the Evosep platform does not require blank injections, while they are crucial on the nanoElute to avoid overpressurization of the system by periodic washing (with on average one 16 min blank injection every three cells injected, included in the SPD calculation). Thereby, the analysis throughput can be increased.
In terms of proteome overlap, 51% (2957 protein groups) is common across all three workflows, with only 4% (214 proteins) uniquely identified with the EVO96_nE workflow. An additional 1249 protein groups (24%) are common between the two Evosep workflows, with 693 protein groups (13%) uniquely identified at 120SPD (Figure 3B).
In terms of quantitative reproducibility, the coefficients of variation (CVs) density distributions revealed the lowest variability with the EVO96_nE workflow (Figure 3D). The slightly higher CV distributions with both Evosep workflows may be imputed to the variability introduced by EVOTIPs usage.
Protein abundance correlation curves for all common proteins identified across all three workflows show that overall protein abundances correlate quite well, with 95% of proteins correlating with less than 20% difference in abundance between the two Evosep protocols. Slightly weaker correlations are observed between the EVO96_nE and both Evosep protocols, which is explained by the overall higher abundances measured with the Evosep protocols (Supporting Information Figure 3B).

3.3. Enzyme-To-Protein Ratio Optimization

We then focused on the lysis and digestion protocol, evaluating four enzyme-to-protein ratios (specifically 1:1, 10:1, 20:1, and 50:1). These ratios were intentionally chosen to not be in line with the lower trypsin concentrations commonly used in bulk proteomics (typically ranging from 1:100 to 1:10). (21) Given the need to work at the nanoliter scale, the lysis and digestion steps must be meticulously refined to accommodate the extremely low protein amounts. Maintaining a high enzyme concentration is essential to ensure efficient digestion kinetics, which in single-cell proteomics occurs within 1–2 h, much faster than the overnight digestion commonly used in bulk workflows. (22−24) In this context, optimizing trypsin concentration involves finding the optimal concentration, meaning the right balance between minimizing trypsin autolysis, which can lead to signal suppression, and maximizing peptide generation to achieve better proteome coverage. Thus, while higher trypsin concentrations can enhance peptide generation, they also carry the risk of autolysis, which could compromise the sensitivity of the analysis.
Both extreme ratios, namely, 1:1 and 50:1, led to important technical issues, notably early and numerous column blockages during chromatographic separation. These cloggings were attributed to either an excess of protease (in a 50:1 ratio) or an insufficient amount of protease (in a 1:1 ratio) to properly digest cellular proteins. Consequently, these two ratios were excluded from further analysis.
The results obtained for the two other evaluated enzyme-to-protein ratios, namely, 10:1 and 20:1, are presented in Figure 4. A higher trypsin concentration resulted in higher proteome coverage with a 27% increase in peptide identifications (8503 vs 10,792 peptides), a 10% increase in protein identifications (1879 vs 2064 protein groups) (Figure 4A,B), and a higher average number of peptides per protein (Supporting Information Figure 6B). One of the main challenges of single-cell proteomic studies is to ensure consistent sampling and quantification of thousands of proteins, while minimizing missing data. (25) Here, we observed not only fewer protein missed cleavages (1% less at 20:1) but also fewer overall missing values (36% at 10:1 compared with 32% at 20:1), indicating better digestion efficiency with more consistent quantification results with the higher enzyme-to-protein ratio (Supporting Information Figure 6B,C).

Figure 4

Figure 4. Evaluation of Trypsin-LysC digestion ratios. This figure presents a comprehensive analysis of HeLa single cells (N = 96 single-cell replicates per condition), comparing two digestion ratios (10:1 and 20:1). (A) Violin plots showing the distributions of protein groups identified per single-cell replicates under each digestion ratio, with or without MBR processing. (B) Violin plots showing the distributions of unique peptides identified per single-cell replicates under each digestion ratio, with or without MBR processing. (C) Venn diagrams illustrate the overlap in identified precursors (left), protein groups (right), and single-peptide hit proteins (down) between the two digestion ratios. (D) Density plots represent the distribution of coefficient of variation (CV) values across single cells.

High Resolution Image
Download MS PowerPoint Slide
Furthermore, to evaluate the digestion reproducibility, we were also interested in the variability of the results across the hundred isolated cells per condition as a function of sample preparation condition. The distribution of the coefficients of variation (CVs) (Figure 4D) demonstrates that refining the workflow by adapting the trypsin ratio is crucial for consistency of the results. Indeed, the 20:1 ratio leads to a reduction in variability, as shown by the lower CV compared with the 10:1 ratio (maximum of CV density of 6.1 vs 11.2, respectively). In addition to this reduction in variability, which is critical for data interpretation, the 20:1 ratio also presents higher numbers of unique identifications at the protein and precursor levels with slightly less single-peptide hits (Figure 4C).
To conclude, our results highlight that extremely low protein amounts require also fine-tuning of the lysis and digestion conditions (i.e., enzyme to substrate ratio) to ensure the maximum peptide generation with the lowest possible trypsin autolysis. It appears that the 20:1 enzyme-to-protein ratio not only enhances peptide and protein identifications but also reduces variability, evidenced by lower coefficients of variation and better identification consistency.

3.4. LC–MS/MS Method Optimization

Finally, after sample collection and digestion conditions optimizations, we optimized the chromatographic separation with the objective to reduce sample analysis time and thus increase sample throughput. To this end, we tested two separation column lengths: 25 cm (75 μm, 1.7 μm, IonOpticks) with a 22 min gradient and 5 cm (75 μm, 1.7 μm, IonOpticks) with a reduced 10 min gradient.
Comparative experiments showed that reducing the gradient duration from 22 min (25 cm column) to 10 min (5 cm column) maintained, even slightly improved, performances for single-cell injections (Figure 5A). More precisely, across single-cell replicates (N = 10), an average of 3782 protein groups were identified using the 5 cm column, compared to 3688 with the 25 cm column. To benchmark this result against more classical higher load samples (in the nanogram range), we have added two replicates of a pool of 10 sorted cells (approximately 2.5 ng). As expected, the proteome coverage was slightly higher with the longer column/gradient for those high load samples (on average 5898 protein groups versus 4542 protein groups for the short gradient). When injecting single-cell amounts, column capacity was not limiting in both configurations and diffusion were thus reduced with the shorter column. In addition, the shorter column configuration also demonstrates good consistency with a protein overlap of 87% (6074 shared protein groups with the 25 cm column). In addition, the coefficients of variation distributions were equivalent (Figure 5C), confirming that the shorter column not only maintains sensitivity but also offers comparable quantitative performances at single-cell equivalent quantities.

Figure 5

Figure 5. Gradient time and column length optimizations. (A) Average and standard deviation of the number of proteins identified per single-cell replicate (pink, N = 10) and per ten cells’ replicates (gray, N = 2), analyzed using 5 and 25 cm IonOpticks separation columns. (B) Union of all proteins identified across all 5 cm column replicates (blue) versus the union of all proteins identified across all 25 cm column replicates (orange). (C) The blue and orange dashed lines indicate the mean CV values calculated for 250 pg HeLa QC lysates (N = 3) run on a 5 cm (average 3753 protein group IDs) versus 25 cm (average 3585 protein group IDs) column, respectively.

High Resolution Image
Download MS PowerPoint Slide
These optimizations, validated on single-cell replicates, enabled a final throughput of up to 55 single cells per day (SPD), compared to 30SPD with the 22 min gradient time configuration, with blanks and QCs included. While maintaining consistent coefficients of variation values ranging from 10% to 35%, this workflow offers in-depth coverage, reproducibility, and improved analytical throughput, making it well-suited for large-scale single-cell proteomics studies. Moreover, this new chromatography configuration enables us to maintain high proteome coverage without performance loss, highlighting that the chromatographic retention miniaturization did not compromise data quality or quantification accuracy.

4. Conclusion

Click to copy section linkSection link copied!
The rigorous evaluation of the different sample preparation options available on cellenONE allowed us to identify the strengths and limitations of each workflow. When coupled with nanoElute injections, we have demonstrated a clear advantage of using LF48 chips and a major benefit of using an oil-free direct injection protocol with automated resuspension in the nanoElute autosampler, avoiding sample transfer. We have also confirmed the clear gains of coupling EVO96 chips with their optimized EVOTIPs and injecting captured samples on the adapted Evosep LC system. In parallel, we have undertaken to optimize the lysis and digestion step and concluded that a 20:1 enzyme-to-protein ratio enhances peptide and protein coverage, reduces missed cleavages and missing values, and generates results with reduced variability. Finally, we have optimized the nanoLC setup and demonstrated that the chromatography time can be reduced and thus the sample throughput improved without performance loss by using short 5 cm columns. Our optimized workflows allow achieving a coverage of up to 5000 protein groups on a single HeLa cell at 55SPD, while up to 3700 protein groups are identified at 120SPD.

Data Availability

Click to copy section linkSection link copied!

Complete data set has been deposited in the ProteomeXchange Consortium via the PRIDE partner17 repository with the data set identifier PXD067019. This research did not involve human or animal participants.

Supporting Information

Click to copy section linkSection link copied!

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jproteome.5c01075.

  • Outliers representation for the oil-free LF48 direct injection workflow on the 96 single-cell replicates; missing values and single-peptide hits’ distributions for EVO96, LF48 oil-free, and LF48 oil-based workflows;pairwise protein abundance correlations for common proteins identified across EVO96/LF48 oil-free/LF48 oil-based and nanoElute_EVO96/EVOSEP_80SPD/EVOSEP_120SPD workflows; repartition and distribution of GRAVY scores for EVO96, LF48 oil-free, and LF48 oil-based workflows; distribution of methionine oxidations identified across workflows; and impact of enzyme-to-protein ratios (10:1 and 20:1) on peptide and protein identifications (PDF)

Terms & Conditions

Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.

Author Information

Click to copy section linkSection link copied!
  • Corresponding Author
    • Christine Carapito - Laboratoire de Spectrométrie de Masse BioOrganique (LSMBO), IPHC UMR7178, CNRS, Université de Strasbourg, 25 Rue Becquerel, Strasbourg, Grand Est 67087, France;  Infrastructure Nationale de Protéomique ProFI, UAR2048, Strasbourg 67087, France;  Orcidhttps://orcid.org/0000-0002-0079-319X;  Email: [email protected]
  • Authors
    • Pauline Perdu-Alloy - Laboratoire de Spectrométrie de Masse BioOrganique (LSMBO), IPHC UMR7178, CNRS, Université de Strasbourg, 25 Rue Becquerel, Strasbourg, Grand Est 67087, France;  Infrastructure Nationale de Protéomique ProFI, UAR2048, Strasbourg 67087, France
    • Charline Keller - Laboratoire de Spectrométrie de Masse BioOrganique (LSMBO), IPHC UMR7178, CNRS, Université de Strasbourg, 25 Rue Becquerel, Strasbourg, Grand Est 67087, France;  Infrastructure Nationale de Protéomique ProFI, UAR2048, Strasbourg 67087, France
    • Anjali Seth - Cellenion SASU, 60 Avenue Rockefeller, Bioserra2, Lyon, Auvergne-Rhône-Alpes 69008, France
  • Funding

    This work was supported by the Agence Nationale de la Recherche via the French Proteomic Infrastructure (ProFI UAR2048; ANR-10-INBS08-03 and ANR-24-INBS-0015), by the Region Grand-Est (SC-Proteomics project) and by the ITMO Cancer of Aviesan within the framework of the 2021–2030 Cancer Control Strategy, on funds administered by Inserm (ProteomiSC project), for equipment funds. It was also supported by the French Ministry of Higher Education and Research for the PhD fellowship of P.P.A and by the Interdisciplinary Thematic Institute IMS, the drug discovery and development institute, as part of the ITI 2021-2028 program of the University of Strasbourg, CNRS and Inserm, supported by IdEx Unistra (ANR-10-IDEX-0002), and by SFRI-STRAT’US project (ANR-20-SFRI-0012) for the PhD fellowship of C.K. C.C. is further supported by the European Union’s Horizon Europe MSCA PROHITS project under grant agreement (no. 101119980) and the CHIST-ERA project ODEEP-EU (ANR-23-CHRO-0005).

  • Notes
    The authors declare no competing financial interest.

Acknowledgments

Click to copy section linkSection link copied!

The authors thank Dr. Christoph Krisp and Pierre-Olivier Schmit from Bruker Daltonics for their support and insightful advices on the instrumental settings.

References

Click to copy section linkSection link copied!

This article references 25 other publications.

  1. 1
    Laisné, M.; Lupien, M.; Vallot, C. Epigenomic Heterogeneity as a Source of Tumour Evolution. Nat. Rev. Cancer 2025, 25 (1), 7– 26,  DOI: 10.1038/s41568-024-00757-9
    Google Scholar
    There is no corresponding record for this reference.
  2. 2
    Budnik, B.; Levy, E.; Harmange, G.; Slavov, N. SCoPE-MS: Mass Spectrometry of Single Mammalian Cells Quantifies Proteome Heterogeneity during Cell Differentiation. Genome Biol. 2018, 19 (1), 161,  DOI: 10.1186/s13059-018-1547-5
    Google Scholar
    There is no corresponding record for this reference.
  3. 3
    Minakshi, P.; Kumar, R.; Ghosh, M.; Saini, H. M.; Ranjan, K.; Brar, B.; Prasad, G. Single-Cell Proteomics: Technology and Applications. In Single-Cell Omics; Barh, D., Azevedo, V., Eds.; Elsevier, 2019; pp 283– 318.
    Google Scholar
    There is no corresponding record for this reference.
  4. 4
    Tian, Y.; Li, Q.; Yang, Z.; Zhang, S.; Xu, J.; Wang, Z.; Bai, H.; Duan, J.; Zheng, B.; Li, W.; Cui, Y.; Wang, X.; Wan, R.; Fei, K.; Zhong, J.; Gao, S.; He, J.; Gay, C. M.; Zhang, J.; Wang, J.; Tang, F. Single-Cell Transcriptomic Profiling Reveals the Tumor Heterogeneity of Small-Cell Lung Cancer. Signal Transduct. Target Ther. 2022, 7 (1), 346,  DOI: 10.1038/s41392-022-01150-4
    Google Scholar
    There is no corresponding record for this reference.
  5. 5
    Nalla, L. V.; Kanukolanu, A.; Yeduvaka, M.; Gajula, S. N. R. Advancements in Single-Cell Proteomics and Mass Spectrometry-Based Techniques for Unmasking Cellular Diversity in Triple Negative Breast Cancer. Proteomics Clin. Appl. 2025, 19 (1), e202400101  DOI: 10.1002/prca.202400101
    Google Scholar
    There is no corresponding record for this reference.
  6. 6
    Marino, F. Z.; Bianco, R.; Accardo, M.; Ronchi, A.; Cozzolino, I.; Morgillo, F.; Rossi, G.; Franco, R. Molecular Heterogeneity in Lung Cancer: From Mechanisms of Origin to Clinical Implications. Int. J. Med. Sci. 2019, 16 (7), 981– 989,  DOI: 10.7150/ijms.34739
    Google Scholar
    There is no corresponding record for this reference.
  7. 7
    Black, G. S.; Huang, X.; Qiao, Y.; Moos, P.; Sampath, D.; Stephens, D. M.; Woyach, J. A.; Marth, G. T. Long-Read Single-Cell RNA Sequencing Enables the Study of Cancer Subclone-Specific Genotypes and Phenotypes in Chronic Lymphocytic Leukemia. Genome Res. 2025, 35 (4), 686– 697,  DOI: 10.1101/gr.279049.124
    Google Scholar
    There is no corresponding record for this reference.
  8. 8
    Mund, A.; Coscia, F.; Kriston, A.; Hollandi, R.; Kovács, F.; Brunner, A.-D.; Migh, E.; Schweizer, L.; Santos, A.; Bzorek, M.; Naimy, S.; Rahbek-Gjerdrum, L. M.; Dyring-Andersen, B.; Bulkescher, J.; Lukas, C.; Eckert, M. A.; Lengyel, E.; Gnann, C.; Lundberg, E.; Horvath, P.; Mann, M. Deep Visual Proteomics Defines Single-Cell Identity and Heterogeneity. Nat. Biotechnol. 2022, 40 (8), 1231– 1240,  DOI: 10.1038/s41587-022-01302-5
    Google Scholar
    There is no corresponding record for this reference.
  9. 9
    Gerniers, A.; Bricard, O.; Dupont, P. MicroCellClust: Mining Rare and Highly Specific Subpopulations from Single-Cell Expression Data. Bioinformatics 2021, 37 (19), 3220– 3227,  DOI: 10.1093/bioinformatics/btab239
    Google Scholar
    There is no corresponding record for this reference.
  10. 10
    Heide, T.; Househam, J.; Cresswell, G. D.; Spiteri, I.; Lynn, C.; Kimberley, C.; Mossner, M.; Zapata, L.; Gabbutt, C.; Ramazzotti, D.; Chen, B.; Fernandez-Mateos, J.; James, C.; Vinceti, A.; Berner, A.; Schmidt, M.; Lakatos, E.; Baker, A.-M.; Nichol, D.; Costa, H.; Mitchinson, M.; Werner, B.; Iorio, F.; Jansen, M.; Barnes, C.; Caravagna, G.; Shibata, D.; Bridgewater, J.; Rodriguez-Justo, M.; Magnani, L.; Graham, T. A.; Sottoriva, A. Assessment of the Evolutionary Consequence of Putative Driver Mutations in Colorectal Cancer with Spatial Multiomic Data. bioRxiv 2021, 451265,  DOI: 10.1101/2021.07.14.451265
    Google Scholar
    There is no corresponding record for this reference.
  11. 11
    Zhang, Z.; Mathew, D.; Lim, T.; Mason, K.; Martinez, C. M.; Huang, S.; Wherry, E. J.; Susztak, K.; Minn, A. J.; Ma, Z.; Zhang, N. R. Signal Recovery in Single Cell Batch Integration. bioRxiv 2023, 539614,  DOI: 10.1101/2023.05.05.539614
    Google Scholar
    There is no corresponding record for this reference.
  12. 12
    Fritzsch, F. S. O.; Dusny, C.; Frick, O.; Schmid, A. Single-Cell Analysis in Biotechnology, Systems Biology, and Biocatalysis. Annu. Rev. Chem. Biomol. Eng. 2012, 3, 129– 155,  DOI: 10.1146/annurev-chembioeng-062011-081056
    Google Scholar
    There is no corresponding record for this reference.
  13. 13
    Dodds, J. N.; Baker, E. S.; Lubman, D. M. Ion Mobility Spectrometry: Fundamental Concepts, Instrumentation, Applications, and the Road Ahead. J. Am. Soc. Mass Spectrom. 2019, 30 (11), 2185– 2195,  DOI: 10.1007/s13361-019-02288-2
    Google Scholar
    There is no corresponding record for this reference.
  14. 14
    Fernandez-Lima, F. A.; Kaplan, D. A.; Park, M. A. Note: Integration of Trapped Ion Mobility Spectrometry with Mass Spectrometry. Rev. Sci. Instrum. 2011, 82 (12), 126106,  DOI: 10.1063/1.3665933
    Google Scholar
    There is no corresponding record for this reference.
  15. 15
    Meier, F.; Brunner, A.-D.; Koch, S.; Koch, H.; Lubeck, M.; Krause, M.; Goedecke, N.; Decker, J.; Kosinski, T.; Park, M. A.; Bache, N.; Hoerning, O.; Cox, J.; Räther, O.; Mann, M. Online Parallel Accumulation–Serial Fragmentation (PASEF) with a Novel Trapped Ion Mobility Mass Spectrometer. Mol. Cell. Proteomics 2018, 17 (12), 2534– 2545,  DOI: 10.1074/mcp.TIR118.000900
    Google Scholar
    There is no corresponding record for this reference.
  16. 16
    Ctortecka, C.; Hartlmayr, D.; Seth, A.; Mendjan, S.; Tourniaire, G.; Udeshi, N. D.; Carr, S. A.; Mechtler, K. An Automated Nanowell-Array Workflow for Quantitative Multiplexed Single-Cell Proteomics Sample Preparation at High Sensitivity. Mol. Cell. Proteomics 2023, 22 (12), 100665,  DOI: 10.1016/j.mcpro.2023.100665
    Google Scholar
    There is no corresponding record for this reference.
  17. 17
    Perez-Riverol, Y.; Bandla, C.; Kundu, D. J.; Kamatchinathan, S.; Bai, J.; Hewapathirana, S.; John, N. S.; Prakash, A.; Walzer, M.; Wang, S.; Vizcaíno, J. A. The PRIDE Database at 20 Years: 2025 Update. Nucleic Acids Res. 2025, 53 (D1), D543– D553,  DOI: 10.1093/nar/gkae1011
    Google Scholar
    There is no corresponding record for this reference.
  18. 18
    Demichev, V.; Messner, C. B.; Vernardis, S. I.; Lilley, K. S.; Ralser, M. DIA-NN: Neural Networks and Interference Correction Enable Deep Proteome Coverage in High Throughput. Nat. Methods 2020, 17, 41– 44,  DOI: 10.1038/s41592-019-0638-x
    Google Scholar
    There is no corresponding record for this reference.
  19. 19
    Demichev, V.; Messner, C. B.; Vernardis, S. I.; Lilley, K. S.; Ralser, M. DIA-NN: Neural Networks and Interference Correction Enable Deep Proteome Coverage in High Throughput. Nat. Methods 2020, 17, 41– 44,  DOI: 10.1038/s41592-019-0638-x
    Google Scholar
    There is no corresponding record for this reference.
  20. 20
    Kyte, J.; Doolittle, R. F. A Simple Method for Displaying the Hydropathic Character of a Protein. J. Mol. Biol. 1982, 157 (1), 105– 132,  DOI: 10.1016/0022-2836(82)90515-0
    Google Scholar
    There is no corresponding record for this reference.
  21. 21
    Wang, Y.; Guan, Z.-Y.; Shi, S.-W.; Jiang, Y.-R.; Zhang, J.; Yang, Y.; Wu, Q.; Wu, J.; Chen, J.-B.; Ying, W.-X.; Xu, Q.-Q.; Fan, Q.-X.; Wang, H.-F.; Zhou, L.; Wang, L.; Fang, J.; Pan, J.-Z.; Fang, Q. Pick-up Single-Cell Proteomic Analysis for Quantifying up to 3000 Proteins in a Mammalian Cell. Nat. Commun. 2024, 15 (1), 1279,  DOI: 10.1038/s41467-024-45659-4
    Google Scholar
    There is no corresponding record for this reference.
  22. 22
    Ctortecka, C.; Mechtler, K. The Rise of Single-Cell Proteomics. Anal. Sci. Adv. 2021, 2 (3–4), 84– 94,  DOI: 10.1002/ansa.202000152
    Google Scholar
    There is no corresponding record for this reference.
  23. 23
    Mansuri, M. S.; Bathla, S.; Lam, T. T.; Nairn, A. C.; Williams, K. R. Optimal Conditions for Carrying out Trypsin Digestions on Complex Proteomes: From Bulk Samples to Single Cells. J. Proteomics 2024, 297, 105109,  DOI: 10.1016/j.jprot.2024.105109
    Google Scholar
    There is no corresponding record for this reference.
  24. 24
    Woessmann, J.; Petrosius, V.; Üresin, N.; Kotol, D.; Aragon-Fernandez, P.; Hober, A.; Auf Dem Keller, U.; Edfors, F.; Schoof, E. M. Assessing the Role of Trypsin in Quantitative Plasma and Single-Cell Proteomics toward Clinical Application. Anal. Chem. 2023, 95 (36), 13649– 13658,  DOI: 10.1021/acs.analchem.3c02543
    Google Scholar
    There is no corresponding record for this reference.
  25. 25
    Slavov, N. Single-Cell Protein Analysis by Mass Spectrometry. Curr. Opin. Chem. Biol. 2021, 60, 1– 9,  DOI: 10.1016/j.cbpa.2020.04.018
    Google Scholar
    There is no corresponding record for this reference.

Cited By

Click to copy section linkSection link copied!

This article has not yet been cited by other publications.

Go to
Get e-Alerts
Get e-Alerts

Journal of Proteome Research

Cite this: J. Proteome Res. 2026, 25, 4, 2200–2208
Click to copy citationCitation copied!
https://doi.org/10.1021/acs.jproteome.5c01075
Published March 4, 2026

Copyright © 2026 The Authors. Published by American Chemical Society. This publication is licensed under these Terms of Use.

Request reuse permissions

Article Views

455

Altmetric

-

Citations

-
Learn about these metrics

Article Views are the COUNTER-compliant sum of full text article downloads since November 2008 (both PDF and HTML) across all institutions and individuals. These metrics are regularly updated to reflect usage leading up to the last few days.

Citations are the number of other articles citing this article, calculated by Crossref and updated daily. Find more information about Crossref citation counts.

The Altmetric Attention Score is a quantitative measure of the attention that a research article has received online. Clicking on the donut icon will load a page at altmetric.com with additional details about the score and the social media presence for the given article. Find more information on the Altmetric Attention Score and how the score is calculated.

  • Abstract

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 1

    Figure 1. Four single-cell sample preparation protocols benchmarked using the cellenONE sorter/liquid dispenser. Lysis and digestion, followed by incubation with automated rehydration cycles are common steps to all three workflows. Then, option 1 relies on a LF48 oil-based plate support involving a manual dilution step in liquid oil, followed by a manual peptide transfer after oil solidification (Tf (hexadecane) = 18 °C); option 2 uses an EVO96 oil-free plate support, involving manual dilution followed by peptide loading onto preconditioned EVOTIPs after centrifugation for injection on the Evosep system; option 3 employs the same EVO96 oil-free support, repurposed for nanoElute injection, where peptides are transferred into a 96-well injection plate via centrifugation; and option 4 relies on a LF48 oil-free plate support suited for automated resuspension and direct injection on a nanoElute 2 system.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 2

    Figure 2. Benchmarking of three single-cell proteomics workflows (EVO96 (blue), LF48 oil-based (green), and LF48 oil-free (orange)) run on nanoElute 2 on 96 single-cell replicates per workflow. (A) Violin plots representing the number of protein groups identified per single cell for each tested workflow using DIA-NN with MBR (left) and without MBR (right). (B) Violin plots representing the number of unique peptides identified per single cell for each tested workflow using DIA-NN with MBR (left) and without MBR (right). (C) Venn diagram showing proteins’ overlap across the three workflows. (D) Coefficients of variation (CVs) density distributions across the three sample preparation workflows. (E) Protein abundance ranking plot comparing the dynamic range of protein groups across the three workflows. Black dots correspond to the shared proteins across the conditions, whereas the colored ones correspond to unique proteins per condition.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 3

    Figure 3. Comparison of SCP EVO96-based workflows across two chromatographic systems. EVO96_nE was analyzed on a nanoElute 2 system (N = 96 single-cells) compared to two Evosep workflows that were acquired using two sample throughputs (80 and 120SPD with N = 40 single-cell replicates each). (A) Violin plots representing the number of protein groups identified per single cell for each workflow (EVO96_nE in blue, EVOSEP80SPD in gray, and EVOSEP120SPD in purple). (B) Venn diagram showing the overlap of identified protein groups across the workflows. (C) Violin plots representing the number of unique peptides identified per single cell for each workflow (EVO96_nE in blue, EVOSEP_80SPD in gray, and EVOSEP_120SPD in purple). (D) Density distributions of protein abundance coefficients of variation (CVs) across the three workflows.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 4

    Figure 4. Evaluation of Trypsin-LysC digestion ratios. This figure presents a comprehensive analysis of HeLa single cells (N = 96 single-cell replicates per condition), comparing two digestion ratios (10:1 and 20:1). (A) Violin plots showing the distributions of protein groups identified per single-cell replicates under each digestion ratio, with or without MBR processing. (B) Violin plots showing the distributions of unique peptides identified per single-cell replicates under each digestion ratio, with or without MBR processing. (C) Venn diagrams illustrate the overlap in identified precursors (left), protein groups (right), and single-peptide hit proteins (down) between the two digestion ratios. (D) Density plots represent the distribution of coefficient of variation (CV) values across single cells.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 5

    Figure 5. Gradient time and column length optimizations. (A) Average and standard deviation of the number of proteins identified per single-cell replicate (pink, N = 10) and per ten cells’ replicates (gray, N = 2), analyzed using 5 and 25 cm IonOpticks separation columns. (B) Union of all proteins identified across all 5 cm column replicates (blue) versus the union of all proteins identified across all 25 cm column replicates (orange). (C) The blue and orange dashed lines indicate the mean CV values calculated for 250 pg HeLa QC lysates (N = 3) run on a 5 cm (average 3753 protein group IDs) versus 25 cm (average 3585 protein group IDs) column, respectively.

    High Resolution Image
    Download MS PowerPoint Slide

  • References

    This article references 25 other publications.

    1. 1
      Laisné, M.; Lupien, M.; Vallot, C. Epigenomic Heterogeneity as a Source of Tumour Evolution. Nat. Rev. Cancer 2025, 25 (1), 7– 26,  DOI: 10.1038/s41568-024-00757-9
      There is no corresponding record for this reference.
    2. 2
      Budnik, B.; Levy, E.; Harmange, G.; Slavov, N. SCoPE-MS: Mass Spectrometry of Single Mammalian Cells Quantifies Proteome Heterogeneity during Cell Differentiation. Genome Biol. 2018, 19 (1), 161,  DOI: 10.1186/s13059-018-1547-5
      There is no corresponding record for this reference.
    3. 3
      Minakshi, P.; Kumar, R.; Ghosh, M.; Saini, H. M.; Ranjan, K.; Brar, B.; Prasad, G. Single-Cell Proteomics: Technology and Applications. In Single-Cell Omics; Barh, D., Azevedo, V., Eds.; Elsevier, 2019; pp 283– 318.
      There is no corresponding record for this reference.
    4. 4
      Tian, Y.; Li, Q.; Yang, Z.; Zhang, S.; Xu, J.; Wang, Z.; Bai, H.; Duan, J.; Zheng, B.; Li, W.; Cui, Y.; Wang, X.; Wan, R.; Fei, K.; Zhong, J.; Gao, S.; He, J.; Gay, C. M.; Zhang, J.; Wang, J.; Tang, F. Single-Cell Transcriptomic Profiling Reveals the Tumor Heterogeneity of Small-Cell Lung Cancer. Signal Transduct. Target Ther. 2022, 7 (1), 346,  DOI: 10.1038/s41392-022-01150-4
      There is no corresponding record for this reference.
    5. 5
      Nalla, L. V.; Kanukolanu, A.; Yeduvaka, M.; Gajula, S. N. R. Advancements in Single-Cell Proteomics and Mass Spectrometry-Based Techniques for Unmasking Cellular Diversity in Triple Negative Breast Cancer. Proteomics Clin. Appl. 2025, 19 (1), e202400101  DOI: 10.1002/prca.202400101
      There is no corresponding record for this reference.
    6. 6
      Marino, F. Z.; Bianco, R.; Accardo, M.; Ronchi, A.; Cozzolino, I.; Morgillo, F.; Rossi, G.; Franco, R. Molecular Heterogeneity in Lung Cancer: From Mechanisms of Origin to Clinical Implications. Int. J. Med. Sci. 2019, 16 (7), 981– 989,  DOI: 10.7150/ijms.34739
      There is no corresponding record for this reference.
    7. 7
      Black, G. S.; Huang, X.; Qiao, Y.; Moos, P.; Sampath, D.; Stephens, D. M.; Woyach, J. A.; Marth, G. T. Long-Read Single-Cell RNA Sequencing Enables the Study of Cancer Subclone-Specific Genotypes and Phenotypes in Chronic Lymphocytic Leukemia. Genome Res. 2025, 35 (4), 686– 697,  DOI: 10.1101/gr.279049.124
      There is no corresponding record for this reference.
    8. 8
      Mund, A.; Coscia, F.; Kriston, A.; Hollandi, R.; Kovács, F.; Brunner, A.-D.; Migh, E.; Schweizer, L.; Santos, A.; Bzorek, M.; Naimy, S.; Rahbek-Gjerdrum, L. M.; Dyring-Andersen, B.; Bulkescher, J.; Lukas, C.; Eckert, M. A.; Lengyel, E.; Gnann, C.; Lundberg, E.; Horvath, P.; Mann, M. Deep Visual Proteomics Defines Single-Cell Identity and Heterogeneity. Nat. Biotechnol. 2022, 40 (8), 1231– 1240,  DOI: 10.1038/s41587-022-01302-5
      There is no corresponding record for this reference.
    9. 9
      Gerniers, A.; Bricard, O.; Dupont, P. MicroCellClust: Mining Rare and Highly Specific Subpopulations from Single-Cell Expression Data. Bioinformatics 2021, 37 (19), 3220– 3227,  DOI: 10.1093/bioinformatics/btab239
      There is no corresponding record for this reference.
    10. 10
      Heide, T.; Househam, J.; Cresswell, G. D.; Spiteri, I.; Lynn, C.; Kimberley, C.; Mossner, M.; Zapata, L.; Gabbutt, C.; Ramazzotti, D.; Chen, B.; Fernandez-Mateos, J.; James, C.; Vinceti, A.; Berner, A.; Schmidt, M.; Lakatos, E.; Baker, A.-M.; Nichol, D.; Costa, H.; Mitchinson, M.; Werner, B.; Iorio, F.; Jansen, M.; Barnes, C.; Caravagna, G.; Shibata, D.; Bridgewater, J.; Rodriguez-Justo, M.; Magnani, L.; Graham, T. A.; Sottoriva, A. Assessment of the Evolutionary Consequence of Putative Driver Mutations in Colorectal Cancer with Spatial Multiomic Data. bioRxiv 2021, 451265,  DOI: 10.1101/2021.07.14.451265
      There is no corresponding record for this reference.
    11. 11
      Zhang, Z.; Mathew, D.; Lim, T.; Mason, K.; Martinez, C. M.; Huang, S.; Wherry, E. J.; Susztak, K.; Minn, A. J.; Ma, Z.; Zhang, N. R. Signal Recovery in Single Cell Batch Integration. bioRxiv 2023, 539614,  DOI: 10.1101/2023.05.05.539614
      There is no corresponding record for this reference.
    12. 12
      Fritzsch, F. S. O.; Dusny, C.; Frick, O.; Schmid, A. Single-Cell Analysis in Biotechnology, Systems Biology, and Biocatalysis. Annu. Rev. Chem. Biomol. Eng. 2012, 3, 129– 155,  DOI: 10.1146/annurev-chembioeng-062011-081056
      There is no corresponding record for this reference.
    13. 13
      Dodds, J. N.; Baker, E. S.; Lubman, D. M. Ion Mobility Spectrometry: Fundamental Concepts, Instrumentation, Applications, and the Road Ahead. J. Am. Soc. Mass Spectrom. 2019, 30 (11), 2185– 2195,  DOI: 10.1007/s13361-019-02288-2
      There is no corresponding record for this reference.
    14. 14
      Fernandez-Lima, F. A.; Kaplan, D. A.; Park, M. A. Note: Integration of Trapped Ion Mobility Spectrometry with Mass Spectrometry. Rev. Sci. Instrum. 2011, 82 (12), 126106,  DOI: 10.1063/1.3665933
      There is no corresponding record for this reference.
    15. 15
      Meier, F.; Brunner, A.-D.; Koch, S.; Koch, H.; Lubeck, M.; Krause, M.; Goedecke, N.; Decker, J.; Kosinski, T.; Park, M. A.; Bache, N.; Hoerning, O.; Cox, J.; Räther, O.; Mann, M. Online Parallel Accumulation–Serial Fragmentation (PASEF) with a Novel Trapped Ion Mobility Mass Spectrometer. Mol. Cell. Proteomics 2018, 17 (12), 2534– 2545,  DOI: 10.1074/mcp.TIR118.000900
      There is no corresponding record for this reference.
    16. 16
      Ctortecka, C.; Hartlmayr, D.; Seth, A.; Mendjan, S.; Tourniaire, G.; Udeshi, N. D.; Carr, S. A.; Mechtler, K. An Automated Nanowell-Array Workflow for Quantitative Multiplexed Single-Cell Proteomics Sample Preparation at High Sensitivity. Mol. Cell. Proteomics 2023, 22 (12), 100665,  DOI: 10.1016/j.mcpro.2023.100665
      There is no corresponding record for this reference.
    17. 17
      Perez-Riverol, Y.; Bandla, C.; Kundu, D. J.; Kamatchinathan, S.; Bai, J.; Hewapathirana, S.; John, N. S.; Prakash, A.; Walzer, M.; Wang, S.; Vizcaíno, J. A. The PRIDE Database at 20 Years: 2025 Update. Nucleic Acids Res. 2025, 53 (D1), D543– D553,  DOI: 10.1093/nar/gkae1011
      There is no corresponding record for this reference.
    18. 18
      Demichev, V.; Messner, C. B.; Vernardis, S. I.; Lilley, K. S.; Ralser, M. DIA-NN: Neural Networks and Interference Correction Enable Deep Proteome Coverage in High Throughput. Nat. Methods 2020, 17, 41– 44,  DOI: 10.1038/s41592-019-0638-x
      There is no corresponding record for this reference.
    19. 19
      Demichev, V.; Messner, C. B.; Vernardis, S. I.; Lilley, K. S.; Ralser, M. DIA-NN: Neural Networks and Interference Correction Enable Deep Proteome Coverage in High Throughput. Nat. Methods 2020, 17, 41– 44,  DOI: 10.1038/s41592-019-0638-x
      There is no corresponding record for this reference.
    20. 20
      Kyte, J.; Doolittle, R. F. A Simple Method for Displaying the Hydropathic Character of a Protein. J. Mol. Biol. 1982, 157 (1), 105– 132,  DOI: 10.1016/0022-2836(82)90515-0
      There is no corresponding record for this reference.
    21. 21
      Wang, Y.; Guan, Z.-Y.; Shi, S.-W.; Jiang, Y.-R.; Zhang, J.; Yang, Y.; Wu, Q.; Wu, J.; Chen, J.-B.; Ying, W.-X.; Xu, Q.-Q.; Fan, Q.-X.; Wang, H.-F.; Zhou, L.; Wang, L.; Fang, J.; Pan, J.-Z.; Fang, Q. Pick-up Single-Cell Proteomic Analysis for Quantifying up to 3000 Proteins in a Mammalian Cell. Nat. Commun. 2024, 15 (1), 1279,  DOI: 10.1038/s41467-024-45659-4
      There is no corresponding record for this reference.
    22. 22
      Ctortecka, C.; Mechtler, K. The Rise of Single-Cell Proteomics. Anal. Sci. Adv. 2021, 2 (3–4), 84– 94,  DOI: 10.1002/ansa.202000152
      There is no corresponding record for this reference.
    23. 23
      Mansuri, M. S.; Bathla, S.; Lam, T. T.; Nairn, A. C.; Williams, K. R. Optimal Conditions for Carrying out Trypsin Digestions on Complex Proteomes: From Bulk Samples to Single Cells. J. Proteomics 2024, 297, 105109,  DOI: 10.1016/j.jprot.2024.105109
      There is no corresponding record for this reference.
    24. 24
      Woessmann, J.; Petrosius, V.; Üresin, N.; Kotol, D.; Aragon-Fernandez, P.; Hober, A.; Auf Dem Keller, U.; Edfors, F.; Schoof, E. M. Assessing the Role of Trypsin in Quantitative Plasma and Single-Cell Proteomics toward Clinical Application. Anal. Chem. 2023, 95 (36), 13649– 13658,  DOI: 10.1021/acs.analchem.3c02543
      There is no corresponding record for this reference.
    25. 25
      Slavov, N. Single-Cell Protein Analysis by Mass Spectrometry. Curr. Opin. Chem. Biol. 2021, 60, 1– 9,  DOI: 10.1016/j.cbpa.2020.04.018
      There is no corresponding record for this reference.

  • Supporting Information

    Supporting Information

    The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jproteome.5c01075.

    • Outliers representation for the oil-free LF48 direct injection workflow on the 96 single-cell replicates; missing values and single-peptide hits’ distributions for EVO96, LF48 oil-free, and LF48 oil-based workflows;pairwise protein abundance correlations for common proteins identified across EVO96/LF48 oil-free/LF48 oil-based and nanoElute_EVO96/EVOSEP_80SPD/EVOSEP_120SPD workflows; repartition and distribution of GRAVY scores for EVO96, LF48 oil-free, and LF48 oil-based workflows; distribution of methionine oxidations identified across workflows; and impact of enzyme-to-protein ratios (10:1 and 20:1) on peptide and protein identifications (PDF)


    Terms & Conditions

    Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.