Discovery of CD80 and CD86 as recent activation markers on regulatory T cells by protein-RNA single-cell analysis

  1. Dominik Trzupek
  2. Melanie Dunstan
  3. Antony J. Cutler
  4. Mercede Lee
  5. Leila Godfrey
  6. Lorna Jarvis
  7. Daniel B. Rainbow
  8. Dominik Aschenbrenner
  9. Joanne L. Jones
  10. Holm H. Uhlig
  11. Linda S. Wicker
  12. John A. Todd
  13. Ricardo C. Ferreira

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Background

Traditionally, the transcriptomic and proteomic characterisation of CD4 + T cells at the single-cell level has been performed by two largely exclusive types of technologies: single-cell RNA sequencing (scRNA-seq) and antibody-based cytometry. Here, we present a multi-omics approach allowing the simultaneous targeted quantification of mRNA and protein expression in single cells and investigate its performance to dissect the heterogeneity of human immune cell populations.

Methods

We have quantified the single-cell expression of 397 genes at the mRNA level and up to 68 proteins using oligo-conjugated antibodies (AbSeq) in 43,656 primary CD4 + T cells isolated from the blood and 31,907 CD45 + cells isolated from the blood and matched duodenal biopsies. We explored the sensitivity of this targeted scRNA-seq approach to dissect the heterogeneity of human immune cell populations and identify trajectories of functional T cell differentiation.

Results

We provide a high-resolution map of human primary CD4 + T cells and identify precise trajectories of Th1, Th17 and regulatory T cell (Treg) differentiation in the blood and tissue. The sensitivity provided by this multi-omics approach identified the expression of the B7 molecules CD80 and CD86 on the surface of CD4 + Tregs, and we further demonstrated that B7 expression has the potential to identify recently activated T cells in circulation. Moreover, we identified a rare subset of CCR9 + T cells in the blood with tissue-homing properties and expression of several immune checkpoint molecules, suggestive of a regulatory function.

Conclusions

The transcriptomic and proteomic hybrid technology described in this study provides a cost-effective solution to dissect the heterogeneity of immune cell populations at extremely high resolution. Unexpectedly, CD80 and CD86, normally expressed on antigen-presenting cells, were detected on a subset of activated Tregs, indicating a role for these co-stimulatory molecules in regulating the dynamics of CD4 + T cell responses.

Article activity feed

  1. Version published to 10.1186/s13073-020-00756-z
  2. Version published to 10.1101/706275v3 on bioRxiv
  3. PREreview

    PREreview of "Simultaneous mRNA and protein quantification at the single-cell level delineates trajectories of CD4+ T-cell differentiation"

    Journal Club PREreview by: Hector Hernandez-Vargas, Apostol Apostolov, Olivier Fesneau, Ossama Labiad, Valentin Thevin, Vincent Flacher, Saidi Soudja, Julien Marie.

    Cancer Research Center of Lyon (CRCL), TGF-beta and Immune Evasion Team, Lyon, France.

    This is a review of the preprint by Dominik Trzupek, Melanie Dunstan, Antony J. Cutler, Mercede Lee, Leila Godfrey, Dominik Aschenbrenner, Holm H. Uhlig, Linda S. Wicker, John A. Todd, and Ricardo C. Ferreir, posted on bioRxiv on October 03, 2019 (DOI: https://doi.org/10.1101/706275). 

     

    Overview

    The preprint by Trzupek et al. shows how the combined quantification of transcripts and protein at the single cell level may provide useful data, in particular in the characterization of immune cell subpopulations. The authors use a technically sound approach to illustrate how transcript and protein provide non-overlapping information about cell identities. To this end, they multiplex an impressive number of protein targets, combined with a more limited list of targeted transcripts. While such methodology will interest single cell researchers in general, the preliminary identification of certain CD4+ T cell subtypes is of particular interest for immunologists. We felt that the manuscript could still be improved to fulfill the expectations of such diverse audiences.

     

    Major concerns:

    • The choice of sample type (i.e. a SLE patient) was somehow surprising. We thought a blood sample from a healthy donor would have been more meaningful in a first stage. Such sample would have been easier to contrast with established knowledge on CD4+ T cell subpopulations. An immune condition or a controlled perturbation of CD4+ subtypes or proportions would be a good choice for a second phase of analyses, once the technique is validated.
    • Moreover, a single SLE sample was used for this initial test. Again, this design does not account for inter-individual variation and is therefore more difficult to contrast with established cell type proportions.
    • The authors discuss how protein and mRNA provide complementary, non-redundant information about cell identities. Unfortunately, they do not perform a separate analysis of the data with each of these two types of information. We expected a more rigorous comparison: how many clusters are obtained when only protein/mRNA are considered? are the same cell types identified? to which level of resolution? From a practical point of view, based on the current description of the data, the reader cannot infer how much information is lost (if any) if only one type of data (mRNA or protein) is considered.
    • A similar concern is valid for pseudotime analyses. Are the same or similar differentiation trajectories obtained with mRNA and protein?

     

    Minor concerns:

    • The title is somehow misleading, there is no real comparison between protein and mRNA. It may be that the same CD4 differentiation trajectories are obtained by only protein or mRNA profiling, as discussed above.
    • Which genes were profiled for gene expression and how were they selected?
    • There is a discussion about cluster 8 being an intermediate between Th17 and Treg differentiation. Can something similar be said about cluster 10?
    • The Methods section would benefit from a more detailed description of the analysis: how was the 35 threshold selected for filtering based on number of features? did the authors used expression of mitochondrial genes (if any was included in their selected targets) to exclude dead cells? similarly, did the authors control for cell cycle effects (if any expected)?
    • The data should be uploaded into a public repository.

     

    Finally, we would like to thank the authors for posting this interesting work as a preprint.

  4. Version published to 10.1101/706275v2 on bioRxiv
  5. Version published to 10.1101/706275v1 on bioRxiv