PaperlayerView exact revision on arXiv

Source-linked AI summary

Routinely quantifying single cell proteomes: A new age in quantitative biology and medicine

Harrison Specht, Nikolai Slavov

arXiv:1804.01069v1q-bio.GNq-bio.QMq-bio.TO

TL;DR

Population-average proteomes can obscure protein patterns within heterogeneous cell populations, limiting measurements relevant to disease and biological modeling. The paper outlines mass-spectrometry strategies to improve single-cell proteomics sensitivity and throughput, which the authors believe can increase by orders of magnitude.

Abstract

from arXiv · show

Many pressing medical challenges - such as diagnosing disease, enhancing directed stem cell differentiation, and classifying cancers - have long been hindered by limitations in our ability to quantify proteins in single cells. Mass-spectrometry (MS) is poised to transcend these limitations by developing powerful methods to routinely quantify thousands of proteins and proteoforms across many thousands of single cells. We outline specific technological developments and ideas that can increase the sensitivity and throughput of single cell MS by orders of magnitude and usher in this new age. These advances will transform medicine and ultimately contribute to understanding biological systems on an entirely new level.

Limitations of population-average measurements

Population averages can conceal protein patterns in heterogeneous cell populations, including trends that disappear or reverse when cell types are combined. The authors argue that single-cell protein measurements are needed to reveal these relationships and support tissue baselines and biomarker discovery.

Limitations of population-average measurements

  • Protein trends within cell types can disappear or reverse in population averages, making heterogeneous samples difficult to interpret.The authors identify this aggregation effect as Simpson’s paradox.Source: Limitations of population-average… · p. 2
  • Proteins that appear positively correlated across cells can be inversely related within each cell type, and marker-based sorting may not reveal that relationship.The authors say it can be observed only by measuring proteins in single cells.Source: Limitations of population-average… · p. 2
  • Single-cell protein measurements are proposed for establishing healthy tissue baselines and diagnosing disease, with biomarker discovery and diagnostic use as immediate applications.The authors also describe longer-term opportunities for modeling and rationally controlling biological systems.Source: Limitations of population-average… · p. 2

Single-cell proteomics opens exciting frontiers

The authors describe single-cell proteomics as a route to studying protein relationships across individual cells, while identifying sensitivity and throughput as technical challenges. They outline opportunities for scaling these measurements and using the resulting distributions to model protein dependencies and signaling.

Single-cell proteomics opens exciting frontiers

  • Single-cell proteomics is presented as a basis for causal protein inference and unbiased models of direct and indirect protein interactions.The authors say estimates of protein joint and conditional distributions across many single cells are needed; such estimates had been obtained for only a few proteins using antibodies or fluorescent proteins.Source: Single-cell proteomics opens exciting… · p. 2
  • Reliable joint distributions of proteoforms could support models of protein dependencies and causal relationships without assuming a specific functional form, but only among measured proteins.The authors connect mutual information estimates to identifying upstream and downstream kinases in signaling cascades.Source: Single-cell proteomics opens exciting… · pp. 2–3
  • The authors believe sensitivity and throughput could each increase by orders of magnitude through the opportunities discussed in the paper.They characterize the challenges as technical and tractable, while noting that top-down proteomics faces lower sensitivity, harder multiplexing, and more challenging protein identification.Source: Single-cell proteomics opens exciting… · p. 4, Transformative opportunities for… · p. 5
  • Antibody-based assays have measured up to a few dozen proteins per cell, whereas early single-cell MS efforts had quantified hundreds per cell and thousands across many cells.The authors focus on MS for its potential specificity, measurement accuracy, proteome depth, and experimental flexibility.Source: Transformative opportunities for… · p. 4
  • The authors describe the field as ready to scale the numbers of accurately quantified proteins and assayed single cells by orders of magnitude from an existing single-cell MS method.They report a recent method quantifying over a thousand proteins across many single mouse stem cells.Source: Transformative opportunities for… · p. 4
  • High protein copy numbers per cell are described as supporting deep proteome quantification and potentially lower sampling error than single-cell RNA-seq.The supplied passage contrasts protein copy number with RNA-seq counting noise and missing values.Source: Transformative opportunities for… · p. 4

Sample preparation

The authors propose minimizing protein loss and reagents while enabling robust, high-throughput sample preparation for large numbers of single cells.

Sample preparation

  • Nanoliter reaction volumes can reduce protein loss from surface adsorption, given the roughly 500 pg of protein in a typical mammalian cell.The authors also propose automated precision liquid handling and MS-compatible lysis reagents, such as sonication in water, to improve robustness and avoid cleanup.Source: Sample preparation · p. 5

Peptide separation

Single-cell MS samples only a small fraction of peptide ions under typical bulk-proteomics settings. The authors propose sharper separations and greater peak sampling, with targeted measurements as an additional strategy, to raise sensitivity.

Peptide separation

  • Less than 1% of the ions for each peptide are sampled under typical bulk-proteomics settings, motivating approaches that deliver more ions to MS detectors.The figure caption describes peak sharpening and increased sampling time as routes to analyzing a larger fraction of ions.Source: Peptide separation · p. 5, Figure 2 · p. 6
  • Sharper, taller separation peaks can increase the fraction of ions sampled for a fixed injection time, using specialized LC columns or capillary electrophoresis.The authors identify capillary electrophoresis as particularly promising for top-down MS.Source: Peptide separation · p. 6, Figure 2 · p. 6
  • 10–100-fold greater sensitivity is the potential of combining improved peptide separation and sampling more of each elution peak, alone or with targeted measurements.The authors note that sampling more of a peak can reduce throughput, while improved separation or targeting can partly compensate.Source: Peptide separation · pp. 6–7, Figure 2 · p. 6

Parallel ion accumulation

Single-cell MS needs longer ion accumulation to sample a larger fraction of ions, but serial accumulation trades sensitivity for throughput. Parallel accumulation is proposed to improve sensitivity without sacrificing throughput.

Parallel ion accumulation

  • Longer serial accumulation improves sensitivity but reduces the number of quantified peptides per unit time.Bulk samples can be reliably quantified with short accumulation because their ions are abundant.Source: Parallel ion accumulation · p. 7
  • Parallel accumulation is proposed to increase sensitivity without losing throughput, using DIA or parallel accumulation with serial injection.DIA can be adapted to use complement ions bound to peptide fragments; serial injection can use additional accumulation traps or trapped ion mobility spectrometry.Source: Parallel ion accumulation · p. 7
  • For long accumulation times, parallel accumulation is expected to increase throughput 2–10 fold, proportional to the number of ions accumulated in parallel.Source: Parallel ion accumulation · p. 7

Higher-multiplexing for increased throughput

Because high chromatographic resolution and quantification of thousands of proteins can require an hour or more of LC-MS/MS, measuring thousands of cells within hours requires multiplexing. Isobaric barcoding could expand multiplexing, with redesigned chemistry also addressing coisolation.

Higher-multiplexing for increased throughput

  • Isobaric chemical barcoding could multiplex single cells in one LC-MS/MS run, where high-resolution measurement of thousands of proteins can take an hour or more.Source: Higher-multiplexing for increased… · p. 7
  • The largest commercially available barcode set allows simultaneous analysis of 11 samples, while proposed larger sets could measure 30–100 individual cells in parallel.Source: Higher-multiplexing for increased… · pp. 7–8
  • Current barcode fragments can be polluted by fragments from other peptides because they are not peptide-specific; stronger barcode–peptide bonds or lower-energy fragmentation could enable peptide-specific complement-ion quantification.Source: Higher-multiplexing for increased… · p. 8

Protein identification and data analysis

Single-cell MS data analysis must address low-abundance peptide identification, integrate information across repeated runs, and handle proteins missing from some cells. Proposed approaches include carrier channels, peptide characteristics, and careful merging of quantitative datasets.

Protein identification and data analysis

  • At low abundance, peptides may lack enough fragment ions for confident identification, so a carrier channel can pool peptides to support identification.Increasing distinct tandem-mass-tags could further facilitate identification and decrease the need for carrier cells.Source: Protein identification and data analysis · p. 8
  • Retention time and empirical spectra can supplement peptide identification, using similar repeated runs to build spectral libraries.Source: Protein identification and data analysis · p. 8
  • Merging separately acquired quantitative datasets requires new computational approaches because shotgun runs partially overlap in quantified proteins, leaving some proteins unquantified in some cells.Such missing data must be handled carefully to avoid artifacts.Source: Protein identification and data analysis · p. 8
Loading 1804.01069v1…