The field of single-cell proteomics is shaped by a single constraint with no equivalent in transcriptomics — proteins cannot be amplified. There is no polymerase chain reaction for a protein, so a peptide lost to a tube wall is not recoverable, and no downstream cleverness restores it. Every methodological development in this area is a response to that fact.
Key Takeaways
|
Why Is There No Amplification Step?
Because proteins have no template. Nucleic acids can be copied enzymatically, which is what makes single-cell transcriptomics tractable: a vanishingly small amount of starting material becomes a comfortable amount of sequencing library. Proteins offer nothing analogous. What you collect from the cell is all you will ever have.
A 2025 review of mass spectrometry-based solutions for single-cell proteomics states the consequence directly: since protein amplification is still not possible, technological improvements focus on minimising sample loss while increasing throughput, resolution, and sensitivity, as well as achieving measurement depth. That is the field’s organising principle, and it explains design decisions that would otherwise look like fussiness.
Working in analytical science?
Register for a FREE Separation Science account to subscribe to the Separation Science Newsletter.
Three consequences worth holding in mind while reading anything about this field.
- Every transfer is a loss. Protein adsorption to surfaces, chemical modification, and losses during ion manipulation are all cited as persistent challenges. At single-cell amounts they are proportionally severe rather than negligible.
- Sensitivity gains are the only route to depth. Where transcriptomics could amplify, proteomics has had to improve instrument sensitivity, chromatography, and handling. Progress has therefore been incremental and instrument-driven.
- Multiplexing serves sensitivity, not just throughput. As the next sections show, the reason for pooling cells is partly to run more of them and partly to give the instrument more material to work with.
Sample Handling at Single-Cell Scale
The handling problem is the same one described for microdissected tissue in Laser Capture Microdissection + LC-MS Proteomics, intensified by roughly an order of magnitude. The responses are correspondingly more aggressive.
Four principles recur across published protocols.
- Isolate directly into the reaction vessel. Cells are sorted by flow cytometry or dispensed by a robotic system straight into multiwell plates or low-volume chips, so the cell arrives where it will be lysed and digested with no intervening transfer.
- Use minimal-volume, clean lysis. The SCoPE2 protocol explicitly seeks to avoid the losses inherent in sample cleanup procedures by using a clean lysis approach, so that no cleanup step is needed at all. Removing a step is more effective than optimising it.
- Keep volumes small and surfaces few. Digestion is performed in around a microlitre in published protocols, which limits both dilution and the surface area available for adsorption.
- Automate the whole sequence. Manual handling at these volumes is variable and slow. Published workflows note that the approach can be fully automated using widely available equipment and scaled to thousands of single cells.
One detail from the protocol literature illustrates how far the attention to detail goes: digestion buffers may include a nuclease to degrade DNA that would otherwise interfere with downstream processing, and trypsin amounts are scaled separately for single-cell wells and for the much larger carrier wells. These are not incidental refinements. At this scale, the difference between a working and a failing protocol is usually a handling detail rather than an instrument setting.
What Does the Isobaric Carrier Actually Do?
It lends the single cells enough signal to be identified — and understanding the mechanism matters because it also explains the method’s main weakness.
The approach was introduced in 2018, when the isobaric tagging protocol designed for 100 micrograms of protein per channel was revised to work with single cells and a carrier sample of 200 cells was added. The SCoPE2 protocol, published in Nature Protocols, describes the logic: an isobaric carrier minimises losses during chromatography while clean lysis avoids cleanup losses, and because samples are multiplexed, protein identification can be performed on material pooled from multiple cells rather than from a single cell. With sixteen-plex reagents, twelve to fourteen single cells can be analysed per experiment.
The mechanism is worth stating precisely. The higher-abundance peptides contributed by the carrier provide more ions for the multiplexed sample, boosting both precursor and fragment ion signals, which yields better sequencing quality for the single-cell peptides. Identification is effectively performed on the carrier while quantification is read from the individual reporter channels.
The Carrier Has a Documented Ceiling That division of labour is also the weakness. Because the carrier dominates the ion population, it can bias quantification in the single-cell channels, and the size of the carrier relative to the single cells has been evaluated by several groups. The reported guidance is specific: the suggested carrier proteome upper limit is 200 times in multiplexed quantification. That is a designable constraint rather than a vague caution. If your carrier exceeds roughly 200 cell equivalents relative to a single-cell channel, you are trading quantitative accuracy for identification depth, and the trade should be deliberate. It is also why the field has pursued multiplexing approaches that do not rely on isobaric reporter ions at all, which is the subject of the next section. |
The Carrier May Be Becoming Optional Worth knowing before designing around the carrier: it may not be needed for much longer. Work published in Nature Communications using an automated chip-based preparation with a current-generation high-sensitivity instrument reports up to 4,000 protein groups, averaging 3,500 per single cell, without a carrier and without match-between-runs. The workflow spans four orders of magnitude and identified over 50 E3 ubiquitin-protein ligases. That matters for the trade-off above. The carrier exists to lend identification signal that single cells could not supply on their own, and it costs quantitative accuracy to do so. If instrument sensitivity and lossless handling together reach several thousand proteins per cell unaided, the carrier’s bias becomes a cost without a corresponding benefit. Check whether your platform still needs one before accepting its constraints. |
Multiplexing Without Ratio Compression
Isobaric labelling has a second known limitation independent of the carrier, namely ratio compression, in which measured differences between channels are systematically understated. Non-isobaric multiplexed data-independent acquisition was developed to avoid it.
Approach | How Cells Are Distinguished | Main Advantage | Main Limitation |
Label-free quantification | Each cell analysed in its own run | Simplest workflow, no labelling chemistry, no compression | Lowest throughput, and no carrier to boost identification |
Isobaric labelling with carrier | Mass-encoded reporter ions read in the fragmentation scan | High throughput, and the carrier substantially aids identification | Ratio compression, and carrier-induced quantification bias above roughly 200 times |
Non-isobaric multiplexed DIA | Mass-shifted labels distinguished at the precursor level | Improves quantification rates and accuracy without isobaric ratio compression | Fewer channels than isobaric tagging, so lower multiplexing factor |
Table 1. Three multiplexing strategies for single-cell proteomics. The third column of the second row is why isobaric labelling became dominant; the fourth column is why alternatives were developed.
The non-isobaric approach works by labelling peptides with reagents that shift precursor mass rather than producing identical precursors with distinguishable fragments. Published in Nature Biotechnology, the approach uses three-plex labelling to triple throughput for low sample amounts, and one channel can still serve as a carrier to boost sensitivity by limiting adsorption losses, so the sensitivity benefit is not entirely forfeited. The trade is a lower multiplexing factor in exchange for quantification that does not compress.
Which approach to choose follows from what limits your study. If you need many cells and can tolerate compressed ratios, isobaric labelling with a carrier remains the throughput leader. If quantitative accuracy between cell states is the point of the experiment, the non-isobaric route addresses the specific artefact that would undermine it. And if depth on a small number of cells matters most, recent label-free work suggests that avoiding labelling altogether has become competitive again, which was not true a few years ago.
How Deep Can It Go Now?
Deeper than most published overviews suggest, and the trajectory is steep enough that any figure needs a date attached to be meaningful.
Approach | Reported Depth | Throughput | Status of Source |
Carrier-assisted isobaric labelling | 1,000 to 1,500 proteins per cell | More than 250 cells per day | Peer-reviewed review, 2025 |
Multiplexed isobaric, characterised generally | Quantifying more than 1,000 proteins per cell | Twelve to fourteen cells per experiment at sixteen-plex | Peer-reviewed protocol |
Label-free, microfluidic pick-up workflow | Up to 3,000 protein groups; 2,278 to 3,257 in one cell line across 44 cells | Not directly comparable | Nature Communications, 2024 |
Automated chip workflow, no carrier | Up to 4,000 protein groups, averaging 3,500 per single cell | Automated, no manual handling | Nature Communications, 2024 |
Chip-and-trap workflow, label-free | Over 5,000 proteins and 40,000 peptides from single cells | Scalable | Nature Methods, 2024 |
High-throughput analyser, companion study | Up to 5,300 proteins per single cell | Not directly comparable | Nature Methods, January 2025 |
Table 2. Reported single-cell proteome depth by approach, with the publication status of each figure. Depth figures are for cultured cells, generally an immortalised line, and are not transferable to tissue-derived cells without validation.
The upper end of that range is now peer-reviewed rather than provisional. A chip-and-trap workflow published in Nature Methods reports over 5,000 proteins and 40,000 peptides from single cells, and a companion study in the same journal challenging a high-throughput analyser reports up to 5,300 proteins per single cell. A separate label-free microfluidic approach published in Nature Communications quantified up to 3,000 protein groups, reporting 2,278 to 3,257 protein groups across 44 cells of one line, and noting that the field’s previous identification level of around 1,000 proteins per cell was insufficient for practical applications.
One qualification still matters when quoting any of these figures. They come from cultured cell lines, which are larger and more uniform than most primary cells and therefore contain more protein per cell. A tissue-derived cell will generally give less, and a slice through a cell less again.
The pattern across the table is nonetheless clear. What was a thousand proteins per cell has become several thousand within a few years, driven by better handling, faster and more sensitive instruments, and acquisition schemes designed for low load rather than adapted from bulk methods. Anyone deciding whether single-cell proteomics can answer their question should check the current literature rather than a review from two years ago.
Toward Spatially Resolved Single-Cell Proteomics
Everything above concerns cells in suspension, which means the spatial context has been discarded before the measurement begins. Recovering it is the current frontier and the reason this article sits in a spatial cluster.
The limitation is acknowledged in the field’s own framing: proteome depth, throughput, and the lack of spatial context have all been identified as limiting biological usefulness. Dissociating a tissue to obtain single cells destroys exactly the information a spatial study needs.
The route around it is to excise the cell rather than dissociate it. Single-cell Deep Visual Proteomics, published in Nature Methods, integrates high-content imaging, laser microdissection, and multiplexed mass spectrometry, and reported a depth of 1,700 proteins from a hepatocyte cell slice while retaining its position in the tissue. Half of the measured proteome was differentially regulated spatially, with protein levels changing markedly near the central vein.
Two observations about that result are worth carrying forward.
- Depth in tissue is competitive with depth in suspension. 1,700 proteins from a cell slice sit within the range reported for suspension single cells, which suggests retaining spatial context need not cost an order of magnitude in coverage.
- The unit is a cell slice, not a whole cell. A section through a cell contains less material than the whole cell, so this is arguably a harder measurement than suspension single-cell proteomics rather than an easier one.
The imaging-guided workflow that makes this possible, including its dependence on prior marker knowledge and the pooling strategy used to gain depth, is covered in Deep Visual Proteomics: Imaging-Guided Mass Spectrometry. For how single-cell measurement sits alongside region-level microdissection and direct imaging on the depth-against-resolution curve, see Spatial Proteomics by Mass Spectrometry: LCM, Single-Cell, and Imaging Approaches, and for the wider spatial landscape, Spatial Analysis in Analytical Science: Mass Spectrometry Imaging and Spatial Omics.
This article was produced under Separation Science's AI Editorial Guidelines.
