Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Letter
  • Published:

Multiplex digital spatial profiling of proteins and RNA in fixed tissue

Nature Biotechnology volume 38, pages 586–599 (2020)Cite this article

Subjects

Abstract

Digital Spatial Profiling (DSP) is a method for highly multiplex spatial profiling of proteins or RNAs suitable for use on formalin-fixed, paraffin-embedded (FFPE) samples. The approach relies on (1) multiplexed readout of proteins or RNAs using oligonucleotide tags; (2) oligonucleotide tags attached to affinity reagents (antibodies or RNA probes) through a photocleavable (PC) linker; and (3) photocleaving light projected onto the tissue sample to release PC oligonucleotides in any spatial pattern across a region of interest (ROI) covering 1 to ~5,000 cells. DSP is capable of single-cell sensitivity within an ROI using the antibody readout, with RNA detection feasible down to ~600 individual mRNA transcripts. We show spatial profiling of up to 44 proteins and 96 genes (928 RNA probes) in lymphoid, colorectal tumor and autoimmune tissues by using the nCounter system and 1,412 genes (4,998 RNA probes) by using next-generation sequencing (NGS). DSP may be used to profile not only proteins and RNAs in biobanked samples but also immune markers in patient samples, with potential prognostic and predictive potential for clinical decision-making.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Overview of the DSP workflow and validation of DSP with oligonucleotide-conjugated antibodies.
Fig. 2: Validation of oligonucleotide-conjugated RNA probes for DSP.
Fig. 3: Spatially resolved, multiplexed protein characterization of geometric regions on FFPE tissue.
Fig. 4: Protein and RNA profiling with gridded ROIs.
Fig. 5: Segmented, rare cell and contour profiling of different compartments of colorectal tumor and IBD tissues.
Fig. 6: Next-generation sequencing readout allows high-plex analyses of each transcript.

Similar content being viewed by others

Data availability

The data that support the findings of this study are available from the authors on reasonable request (see the author contributions for details on specific datasets). All unique materials used are readily available from NanoString Technologies.

References

  1. Garon, E. B. et al. Pembrolizumab for the treatment of non-small-cell lung cancer. N. Engl. J. Med. 372, 2018–2028 (2015).

    Article  Google Scholar 

  2. Yu, H. et al. PD-L1 expression by two complementary diagnostic assays and mRNA in situ hybridization in small cell lung cancer. J. Thorac. Oncol. 12, 110–120 (2017).

    Article  Google Scholar 

  3. Yu, H., Boyle, T. A., Zhou, C., Rimm, D. L. & Hirsch, F. R. PD-L1 expression in lung cancer. J. Thorac. Oncol. 11, 964–975 (2016).

    Article  Google Scholar 

  4. Ting, D. T. et al. Aberrant overexpression of satellite repeats in pancreatic and other epithelial cancers. Science 331, 593–596 (2011).

    Article  CAS  Google Scholar 

  5. Garber, K. Oncologists await historic first: a pan-tumor predictive marker, for immunotherapy. Nat. Biotechnol. 35, 297–298 (2017).

    Article  CAS  Google Scholar 

  6. Sokolenko, A. P. & Imyanitov, E. N. Molecular tests for the choice of cancer therapy. Curr. Pharm. Des. 23, 4794–4806 (2017).

    Article  CAS  Google Scholar 

  7. Stack, E. C., Wang, C., Roman, K. A. & Hoyt, C. C. Multiplexed immunohistochemistry, imaging, and quantitation: a review, with an assessment of tyramide signal amplification, multispectral imaging and multiplex analysis. Methods 70, 46–58 (2014).

    Article  CAS  Google Scholar 

  8. Wang, F. et al. RNAscope: a novel in situ RNA analysis platform for formalin-fixed, paraffin-embedded tissues. J. Mol. Diagn. 14, 22–29 (2012).

    Article  CAS  Google Scholar 

  9. Goltsev, Y. et al. Deep profiling of mouse splenic architecture with CODEX multiplexed imaging. Cell 174, 968–981 (2018).

    Article  CAS  Google Scholar 

  10. Jungmann, R. et al. Multiplexed 3D cellular super-resolution imaging with DNA-PAINT and Exchange-PAINT. Nat. Methods 11, 313–318 (2014).

    Article  CAS  Google Scholar 

  11. Chen, K. H., Boettiger, A. N., Moffitt, J. R., Wang, S. & Zhuang, X. Spatially resolved, highly multiplexed RNA profiling in single cells. Science 348, aaa6090 (2015).

    Article  Google Scholar 

  12. Lee, J. H. et al. Highly multiplexed subcellular RNA sequencing in situ. Science 343, 1360–1363 (2014).

    Article  CAS  Google Scholar 

  13. Stoeckli, M., Chaurand, P., Hallahan, D. E. & Caprioli, R. M. Imaging mass spectrometry: a new technology for the analysis of protein expression in mammalian tissues. Nat. Med. 7, 493–496 (2001).

    Article  CAS  Google Scholar 

  14. Cornett, D. S., Reyzer, M. L., Chaurand, P. & Caprioli, R. M. MALDI imaging mass spectrometry: molecular snapshots of biochemical systems. Nat. Methods 4, 828–833 (2007).

    Article  CAS  Google Scholar 

  15. Giesen, C. et al. Highly multiplexed imaging of tumor tissues with subcellular resolution by mass cytometry. Nat. Methods 11, 417–422 (2014).

    Article  CAS  Google Scholar 

  16. Angelo, M. et al. Multiplexed ion beam imaging of human breast tumors. Nat. Med. 20, 436–442 (2014).

    Article  CAS  Google Scholar 

  17. Stoeckius, M. et al. Simultaneous epitope and transcriptome measurement in single cells. Nat. Methods 14, 865–868 (2017).

    Article  CAS  Google Scholar 

  18. Shahi, P. et al. Abseq: ultrahigh-throughput single cell protein profiling with droplet microfluidic barcoding. Sci. Rep. 7, 44447 (2017).

    Article  CAS  Google Scholar 

  19. Stahl, P. L. et al. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science 353, 78–82 (2016).

    Article  CAS  Google Scholar 

  20. Rodriques, S. G. et al. Slide-seq: a scalable technology for measuring genome-wide expression at high spatial resolution. Science 363, 1463–1467 (2019).

    Article  CAS  Google Scholar 

  21. Vickovic, S. et al. High-density spatial transcriptomics arrays for in situ tissue profiling. Nat. Methods 16, 987–990 (2019).

    Article  CAS  Google Scholar 

  22. Keren, L. et al. A structured tumor-immune microenvironment in triple negative breast cancer revealed by multiplexed ion beam imaging. Cell 174, 1373–1387 (2018).

    Article  CAS  Google Scholar 

  23. Feng, Z. et al. Multispectral imaging of formalin-fixed tissue predicts ability to generate tumor-infiltrating lymphocytes from melanoma. J. Immunother. Cancer 3, 47 (2015).

    Article  Google Scholar 

  24. Tsujikawa, T. et al. Quantitative multiplex immunohistochemistry reveals myeloid-inflamed tumor-immune complexity associated with poor prognosis. Cell Rep. 19, 203–217 (2017).

    Article  CAS  Google Scholar 

  25. Remark, R. et al. In-depth tissue profiling using multiplexed immunohistochemical consecutive staining on single slide. Science Immunol. 1, aaf6925 (2016).

    Article  Google Scholar 

  26. Ayers, M. et al. IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade. J. Clin. Invest. 127, 2930–2940 (2017).

    Article  Google Scholar 

  27. Mlecnik, B. et al. Integrative analyses of colorectal cancer show Immunoscore is a stronger predictor of patient survival than microsatellite instability. Immunity 44, 698–711 (2016).

    Article  CAS  Google Scholar 

  28. Geiss, G. K. et al. Direct multiplexed measurement of gene expression with color-coded probe pairs. Nat. Biotechnol. 26, 317–325 (2008).

    Article  CAS  Google Scholar 

  29. Lee, J. et al. Implementation of a multiplex and quantitative proteomics platform for assessing protein lysates using DNA-barcoded antibodies. Mol. Cell Proteomics 17, 1245–1258 (2018).

    Article  CAS  Google Scholar 

  30. Falcon, B. L. et al. High-content multiplexed tissue imaging and quantification for cancer drug discovery. Drug Discov. Today 18, 510–522 (2013).

    Article  CAS  Google Scholar 

  31. Pages, F. et al. International validation of the consensus Immunoscore for the classification of colon cancer: a prognostic and accuracy study. Lancet 391, 2128–2139 (2018).

    Article  Google Scholar 

  32. Bodenmiller, B. Multiplexed epitope-based tissue imaging for discovery and healthcare applications. Cell Syst. 2, 225–238 (2016).

    Article  CAS  Google Scholar 

  33. Amaria, R. N. et al. Neoadjuvant immune checkpoint blockade in high-risk resectable melanoma. Nat. Med. 24, 1649–1654 (2018).

    Article  CAS  Google Scholar 

  34. Blank, C. U. et al. Neoadjuvant versus adjuvant ipilimumab plus nivolumab in macroscopic stage III melanoma. Nat. Med. 24, 1655–1661 (2018).

    Article  CAS  Google Scholar 

  35. Toki, M. et al. High-plex predictive marker discovery for melanoma immunotherapy treated patients using Digital Spatial Profiling. Clin. Cancer Res. 25, 5503–5512 (2019).

    Article  CAS  Google Scholar 

  36. Johnson, D. B. et al. A case report of clonal EBV-like memory CD4+ T cell activation in fatal checkpoint inhibitor-induced encephalitis. Nat. Med. 25, 1243–1250 (2019).

    Article  CAS  Google Scholar 

  37. Stellwagen, N. C., Magnusdottir, S., Gelfi, C. & Righetti, P. G. Measuring the translational diffusion coefficients of small DNA molecules by capillary electrophoresis. Biopolymers 58, 390–397 (2001).

    Article  CAS  Google Scholar 

  38. Schindelin, J., Rueden, C. T., Hiner, M. C. & Eliceiri, K. W. The ImageJ ecosystem: an open platform for biomedical image analysis. Mol. Reprod. Dev. 82, 518–529 (2015).

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank A. VanSchoiack for editing the manuscript and H. Metz for performing IHC.

Author information

Authors and Affiliations

  1. NanoString Technologies, Inc., Seattle, WA, USA

    Christopher R. Merritt, Giang T. Ong, Sarah E. Church, Kristi Barker, Patrick Danaher, Gary Geiss, Margaret Hoang, Jaemyeong Jung, Yan Liang, Jill McKay-Fleisch, Karen Nguyen, Zach Norgaard, Kristina Sorg, Isaac Sprague, Charles Warren, Sarah Warren, Philippa J. Webster, Zoey Zhou, Daniel R. Zollinger, Dwayne L. Dunaway & Joseph M. Beechem

  2. Precision Oncology, Oregon Health & Science University Knight Cancer Institute, Portland, OR, USA

    Gordon B. Mills

Authors
  1. Christopher R. Merritt
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Giang T. Ong
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Sarah E. Church
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Kristi Barker
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Patrick Danaher
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Gary Geiss
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Margaret Hoang
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Jaemyeong Jung
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Yan Liang
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Jill McKay-Fleisch
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Karen Nguyen
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Zach Norgaard
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Kristina Sorg
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Isaac Sprague
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. Charles Warren
    View author publications

    You can also search for this author in PubMed Google Scholar

  16. Sarah Warren
    View author publications

    You can also search for this author in PubMed Google Scholar

  17. Philippa J. Webster
    View author publications

    You can also search for this author in PubMed Google Scholar

  18. Zoey Zhou
    View author publications

    You can also search for this author in PubMed Google Scholar

  19. Daniel R. Zollinger
    View author publications

    You can also search for this author in PubMed Google Scholar

  20. Dwayne L. Dunaway
    View author publications

    You can also search for this author in PubMed Google Scholar

  21. Gordon B. Mills
    View author publications

    You can also search for this author in PubMed Google Scholar

  22. Joseph M. Beechem
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

C.R.M., S.E.C., P.J.W. and J.M.B. wrote the manuscript, and C.M. prepared the figures. J.M.B. and G.M. conceived the project. J.M.B., D.L.D. and C.R.M. supervised the project. C.R.M. supervised experimental studies. G.G. supervised antibody labeling. C.R.M., G.T.O., S.E.C., D.Z., K.S., K.B., M.H. and J.M.B. designed experiments. K.B. performed antibody validation experiments. G.T.O. and K.B. performed protein profiling experiments. Y.L. and S.E.C. performed IHC interpretation. C.M. and P.W. developed the IHC protocol. C.M., P.W. and K.S. developed the RNA profiling protocol. D.Z. and J.M.-F. performed RNA profiling experiments. K.N. and K.S. performed RNAscope experiments. C.R.M., G.T.O., D.Z., K.S., K.B., M.H. and P.D. performed data analysis. Z.N. established the NGS data analysis pipeline. J.J., I.B.S. and D.L.D. developed initial instrumentation setup. J.J. configured the microscope and performed preliminary spatial protein profiling experiments. C.W. and I.B.S. developed the final instrumentation and fluidics setup. I.B.S. developed the instrumentation automation.

Corresponding authors

Correspondence to Christopher R. Merritt or Joseph M. Beechem.

Ethics declarations

Competing interests

Patent applications have been filed related to the subject matter of this publication. G.B.M. is a co-inventor of the technology and receives research support from NanoString. All NanoString Technologies employees (C.R.M., G.T.O., S.E.C., K.B., P.D., G.G., M.H., J.J., Y.L., J.M.-F., K.N., Z.N., K.S., I.S., C.W., S.W., P.J.W., Z.Z., D.R.Z., D.L.D. and J.M.B.) declare that they are employees and shareholders of NanoString Technologies.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Integrated supplementary information

Supplementary Figure 1 UV bleedover characterization and optimization.

a, Masking strategy on cell pellet model system to mimic the dimensions of immune and tumor compartments found in tumor samples. 10 µm gaps (approximate width of a T cell) between tumor cell-enriched regions are frequently found. A cross with 10 µm thickness and surrounding regions were profiled on cell pellet sections. b, Two masks were sequentially profiled for each region-of-interest, with Mask 1 representing “tumor” and Mask 2 representing “immune”. The gap between Mask 1 and Mask 2 was titrated across ROIs and the amount of signal remaining in Mask 2 was characterized. c, A control Mask 1 was included to characterize the maximum possible impact of UV bleedover between masks (“Control UV+ entire region”, left). Mask 2 from this control is expected to have the minimum signal. A second control Mask 2 was included to characterize the minimum possible impact of UV bleedover between masks (“Control UV- entire region, right). Mask 2 from this control is expected to have the maximum signal. Illustrated between these two controls are the distance titration of Mask 1. The smallest gaps are expected to give the largest loss of signal. The minimal gap distance leading to signal equal to the UV- control mask is the approximate distance where no bleedover is seen with this readout. d, Distance titrations performed on CCRF-CEM cell pellets across three different UV exposure times (1 second, 2 second, or 4 second). Box plots indicate the median, upper and lower quartiles with whiskers extending to the minimum and maximum values within 1.5x the distance between the first and third quartiles. Data outside of the whiskers are plotted as points. These outliers are also highlighted with a * sign. Six antibodies giving robust signal in CCRF-CEM are shown. Per mask distance, the total number of ROIs profiled was n=5, n=6, and n=6 for 1 second, 2 second, and 4 second UV, respectively. A total of 136 ROIs were profiled. The replicates were averaged and normalized to 4 second UV- Control (arrow) and shown for each condition. This normalization highlights the decrease in UV cleavage efficiency with the decrease in exposure time. Per UV exposure time, to determine significant change versus UV- Control a two sample, one-sided, unpaired t-test was calculated for each mask distance versus the UV-Control. P values of each condition compared to the UV- control are shown. Largest impact of UV bleedover is seen with the 4 sec exposure. Loss of signal in Mask 2 is not significant with the 0 µm gap, or larger gaps, with the 2 sec UV exposure time. The Mask 1 UV+ control for each UV exposure time (left) shows that UV cleavage is less than 100%, with signal still present in Mask 2. The remaining signal decreases with increasing exposure time in Mask 1. The optimal exposure time for this system was ~2 sec, since maximum counts with minimal bleedover were seen with this exposure time. This data shows cross talk between neighboring compartments can be minimized to undetectable amounts when UV exposure time was optimized at levels leading to less than 100% release of the UV-cleavable oligos. With higher UV exposure, cross-talk can be reduced by increasing the distance between the two compartments.

Supplementary Figure 2 Validation of oligo-conjugated antibodies based on fluorescence intensity.

23 core cell pellet tissue microarray used for antibody validation stained with SYTO13 nuclear dye to highlight different cores. Two replicate ROIs across three profiled sizes, 50 µm, 100 µm, and 300 µm diameter circles. Fluorescently labeled antibodies for CD3, CD45, and PanCK were included with a mix of 44 oligo-conjugated antibodies for reference. Average nCounter counts from DSP profiling for the 100 µm ROIs displayed in a heatmap next to the single channel fluorescent images for each target. Scale bar, 2 mm.

Supplementary Figure 3 Validation of DSP with oligo-conjugated antibodies based bulk RNA expression and correlation across replicate ROIs.

a, Additional data from experiment presented in Fig. 1c. 23 core FFPE cell pellet array stained with CD3, CD45, and PanCK fluorescently-conjugated antibodies and SYTO13 nuclear dye, along with a mix of 41 oligo-conjugated antibodies. N=2 replicate ROIs across n=3 ROI areas (50 µm, 100 µm, and 300 µm diameter circles) were profiled (n=6 total technical replicates). Y-axes are scaled according the expression range for each antibody and each ROI size. nCounter counts from 100 ng purified RNA (see methods) were used to estimate which cell lines are positive (> 100 counts) or negative (<10 counts). Targets with nCounter counts between 10 and 100, targets expected to be expressed in all cell lines (housekeeper), or antibodies with no specificity (IgG controls) were labeled as “Unknown” expression. Box plots indicate the median, upper and lower quartiles with whiskers extending to the minimum and maximum values. Box plot were used for these n=2 studies to clearly show the replicate datapoints. b, Unsupervised clustering of scaled data across all ROIs. Data was scaled to characterize general expression patterns for each ROI, while considering variable ROI sizes. Average R2 correlations across the 6 ROIs for each cell line were determined by comparing the first replicate of the 100 µm ROI against the 5 other ROIs. These 5 independent R2 values were average to obtain the values shown across the heatmap annotation row [Avg R2]. These average R2 values were all greater than 0.96, with a range from 0.963 (A549) to 0.999 (MOLT-4). Expected clustering of similar cell line types was also seen with clustering of the leukemia and lymphoma cell lines (Jurkat, CCRF-CEM, HuT 78, MOLT-4, K-562, SU-DHL-1, SU-DHL-4, Daudi, Raji) and the breast cancer cell lines (MDA-MB-468, SK-BR-3). Targets also tended to cluster by protein target type. For example, clustering was observed for: the housekeeping proteins Histone H3 and Ribosomal Protein S6 (RPS6); the T cell markers CD3, CD4, and CD8, and the negative control IgGs. PD1, a target that is not expressed in any of the analyzed cell lines, also clustered with the negative control IgGs, showing antibodies with no specificity resemble the behavior of antibodies on samples lacking the expression of target epitopes.

Supplementary Figure 4 Protein DSP results compared to previously published RNA-seq data.

Results for the 100 µm diameter ROIs shown in Suppl. Fig. 3: were also compared to the Broad’s Cancer Cell Line Encyclopedia (https://portals.broadinstitute.org/ccle/). DSP data represents the average of two replicate ROIs displayed next to the corresponding CCLE RNAseq RSEM data. Cell lines and targets that were not found in the Cancer Cell Line Encyclopedia were excluded from this figure.

Supplementary Figure 5 Additional cell line titration analyses characterizing the limit of detection of the protein DSP assay.

a, Analysis of varying mixtures of CCRF-CEM (CD3+ cells) in the background of HEK293 cells (primary embryonic human kidney cells). nCounter counts plotted against the number of CCRF-CEM cells in each 100 µm ROI for 5 targets expressed exclusively in CCRF-CEM (CD3, CD4, CD45, CD45RO, VISTA), 3 targets expressed in both cell lines (Histone H3, RPS6, AKT), and 1 negative control antibody (Mouse IgG2a). ROIs containing a variable number of CCRF-CEM cells were profiled (n=24 ROIs). The LOD of each target was calculated as the mean + 3x standard deviation of all ROIs containing no CCRF-CEM cells (n=4). Arrows and labels indicate the estimated number of cells needed for detection of the antibody. R2 (from Pearson’s R) values are shown. b, Additional targets were analyzed in a similar manner with a different mixture of cell lines that included SU-DHL-1 (CD20-positive, B cell lymphoma-derived) in the background of NIH-H596 (CD20 negative, lung adenosquamous carcinoma-derived). ROIs containing a variable number of SU-DHL-1 cells were profiled (n=24 ROIs). 7 differentially-expressed targets were analyzed for SU-DHL-1 (top 7 plots). Control targets that are not differential expressed are shown below (bottom 4 plots). The LOD of each target was calculated as the mean + 3x standard deviation of all ROIs containing no CCRF-CEM cells (n=4). Arrows and labels indicate the estimated number of cells needed for detection of the antibody. R2 (from Pearson’s R) values are shown.

Supplementary Figure 6 Linearity of DSP detection down to a single cell in various model systems.

a, Representative section of FFPE cell pellet (CCRF-CEM cell line) showing of ROIs (UV-cleavage areas) ranging from 331,000 µm2 (a circle with a 650 µm diameter) to 1,960 µm2 (a circle with a 50 µm diameter). Concentric areas are displayed for illustration purposes only; in the experiment, independent regions of the cell pellet were illuminated for each ROI. b, Average counts from n=3 replicate measurements for each ROI size. Best fit linear trend line shown. R2 calculated from Pearson’s R. Error bars represent one standard deviation. c, Representative CCRF-CEM cell pellet illustrating ROIs illuminated for 1 to 8 cells. For each ROI, the designated number of cells were UV illuminated simultaneously and collected with a single aspiration step. A fluorescently conjugated secondary antibody targeting primary antibodies against CD3 and CD4 was used to visualize cell membranes. This experiment was performed once. d, Counts of replicate ROIs (n=12 for 1 cell, n=9 for 2 cells and 4 cells, and n=6 for 8 cells) for each ROI size plotted versus size of UV-exposed area. Best fit linear trend line shown. R2 calculated from Pearson’s R. Limit-of-detection (LOD) value shown as dotted line: the average plus two standard deviations of replicate counts of samples with no illumination, representing assay background. This data shows detectability of CD3 down to a single cell. e, Tonsil stained with CD3 and CD4 rabbit antibodies (T cell markers; green) and CD19 and CD20 mouse antibodies (B cell markers; magenta) and visualized with goat anti-mouse and goat anti-rabbit antibodies labeled with different fluorophores. Average cell diameters for the two cell types differ, and UV-illumination areas, here outlined with circles, were adjusted accordingly. This experiment was performed once. f, CD3 or CD20 counts from single cell illumination of T cells (n=12) and B cells (n=12). Middle black bar represents the average counts for all replicates, and error bars represent one standard deviation. LOD value shown is the average plus two standard deviations of replicate counts of no-UV samples (representing background counts). The expected detectability was seen for single T cells and B cells in this tissue. g, Various 1, 2, or 4 cell ROIs of a PBMC cell pellet were UV illuminated; representative images shown A fluorescently conjugated secondary antibody targeting antibodies against CD3 and CD4 was used to visualize T cells. This experiment was performed once. h, Counts of replicate ROIs for each ROI size plotted versus number of cells illuminated (n=6 for 1 cell, n=6 for 2 cells, and n=5 for 4 cells). Error bars are one standard deviation. Best fit linear trend line shown. R2 calculated rom Pearson’s R (three outlier points, gray, outside of one standard deviation were not used for this calculation). Limit-of-detection (LOD) value shown is the average plus two standard deviations of replicate counts of no-UV samples (representing background counts). This data shows detectability of CD3 down to a single cell.

Supplementary Figure 7 Representative images of CD40 RNAscope quantification and additional RNA correlation data between DSP and bulk RNA profiling.

a, Representative 40x FOV max intensity Z projection images of CD40 RNAscope quantification. CD40 RNA (magenta) and DAPI (gray). For each cell line, estimated cell boundaries from the nuclear stain and RNAscope spot boundaries were defined with the QuPath quantification process. Closeup images of representative quantification boundaries across all cell lines is shown. Similar experiments were performed independently 2 times (Supplementary Figures 8, 9). b, Additional data from experiment presented in Fig. 2d. Summary of R2 values for all targets meeting criteria in d and the range of expression (highest nCounter counts / lowest nCounter counts across all cell pellets). Samples in blue have R2 >0.75 and samples in red have R2 <0.75.

Supplementary Figure 8 DSP RNA, RNAscope, and nCounter expression data cross-platform comparison.

a, Additional RNAscope analysis was performed, as described in the text, but with 4 replicate 200 µm diameter ROIs. Additional targets were also analyzed. In general, high concordance was seen across all platforms. Target expression that was not concordant is highlighted in orange. b, CD40 and CD274 (PD-L1) data highlighted. Both of these targets showed cell lines just above background for the DSP RNA assay (arrows). This data was used to estimate the LOD of the system in Supplementary Fig. 9.

Supplementary Figure 9 DSP RNA detection LOD estimation based on RNAscope.

a, Calculations performed to estimate the LOD of the DSP RNA assay. The average number of cells (nuclei counted) in the 4 replicate ROIs analyzed in Suppl. Fig. 8 are shown on the right. These cell counts were multiplied by the total number of RNAscope spots (transcripts) counted to estimate the number of transcripts present in the ROIs profiled with DSP. Three targets detected just above background are highlighted in yellow. The average of these RNA transcripts predicted to be present in these ROIs was 606 transcripts (or ~2 copies per cell across a total of ~300 cells). b, Because our LOD appeared to be around 600 transcripts, we estimate that DSP should be able to detect ~30 transcripts per cell across ~20 cells. To test this, we profiled smaller, 50 µm diameter circle ROIs. We found that expression was readily detectable down to 21 cells, thus confirming our LOD for the higher expressors examined in this study was approximately 20 cells.

Supplementary Figure 10 Reproducibility of the DSP system.

a, Serial sections of tonsil tissues analyzed in Fig. 3a. ROIs were selected across each of the tissues (n=36), and the overlapping ROIs (n=25) were used for this analysis. Sections were stained with CD3, CD20, and PanCK antibodies directly conjugated to fluorescent dyes. b, Correlation of CD3, CD20, and PanCK on both serial sections across the 25 overlapping ROIs. c, Four ROI types selected, based on visualization marker expression, for correlation analysis of the 44 antibodies in the cocktail. d, Correlation of antibody counts across overlapping ROIs (n=25) on adjacent sections.

Supplementary Figure 11 RNA DSP slide reuse.

a, Fluorescent images of one FFPE tonsil slide stained for CD3 (cyan), CD45 (magenta), PanCK (green), and nuclei (gray). 300 μm diameter circular ROIs were profiled in various regions of the sample (Initial). Slides were then washed in 2x SSC, incubated in Tris EDTA pH 9.0 for 25 minutes at 80-85°C in a pressure cooker to remove bound detection probes. Slides were washed again in 2x SSC, placed on a transparent tray, and exposed to UV light for 3 minutes using a UV gel box to cleave any residual probes. The protocol was repeated starting at probe application and slides were profiled using the same ROI size. Half of the ROIs were placed in the same regions as initial and half were placed in adjacent regions. This experiment was performed independently 2 times with similar results. b, Counts from each ROI and detectable targets were log2 transformed and plotted. Pearson R2 concordance after re-probing was high. Furthermore, absolute counts were similar between initial and reprobe (mean ratio reprobe profiling to initial profiling = 0.982). This experiment was performed independently 2 times with similar results. c, Log2 counts were compared between ROI where the same region was profiled in both initial and reprobe show high concordance. This experiment was performed independently 2 times with similar results. d, Log2 counts were compared between ROI where adjacent regions were profiled in both initial and reprobe show concordance. This experiment was performed independently 2 times with similar results. e, RNA was extracted from sections after performing DSP and from non-processed serial section controls and quantified with the nCounter IO360 PanCancer Pathways panel. Quantification shows similar levels of counts are recovered from both sections (mean ratio DSP sample to control sample = 1.05) and the two samples show high concordance. This experiment was performed once. f, Serial sections were cut from a FFPE tumor microarray consisting of multiple tissue types. One serial section was stained with H&E and another was profiled via DSP and then stained with H&E. Representative cores are shown. With the exception of a small portion of the colorectal cancer core, overall morphology is intact following DSP. This experiment was performed once.

Supplementary Figure 12 High throughput TMA analysis with geometric ROIs.

a, Scan of 384 cores analyzed from a high-density colorectal cancer TMA (See Supplementary Tables 3 and 4 for additional information). These 384 cores contained n=2 replicate cores from 192 donor samples. 665 µm diameter circle ROIs were profiled for each core. b, Correlation of selected targets across 384 cores. CD3/CD8 (T cell markers) and PMS2/MLH1 (mismatch repair proteins) show high correlation (R2 > 0.75), while comparisons of unrelated markers (PanCK/CD3 or CD45/MLH1) show no correlation (R2 < 0.05). R2 calculated rom Pearson’s R. c, Correlation matrix of all target pairs analyzed. R2 calculated rom Pearson’s R. d, Hierarchically clustered heatmap of region-specific digital data across selected protein targets (n=32) across all cores (n=384). e, Heatmap of PD-L1 across the 384-cores shows high PD-L1 expression for both replicate cores from a single donor sample. f, IHC analysis of PD-L1 on a serial section to confirm the expression of PD-L1 in these two cores. This experiment was performed once.

Supplementary Figure 13 Gridded profiling heatmaps for all protein targets in tonsil.

Heatmaps for each antibody analyzed in the gridded analysis presented in Fig. 4a-f.

Supplementary Figure 14 Correlation between nCounter and NGS RNA DSP probes.

a, ROI selection on the 23-core cell pellet array for NGS readout. Three 300 µm diameter circles were selected as replicates from each cell pellet. Three additional ROIs on the glass surface only were included as a negative control. b, nCounter IO360 expression data was used to determine target with expression confidently above background (> 50 counts). These target/cell lines (n=659 total data points) were used to analyze the correlation between the nCounter and NGS RNA DSP probes. Average SNRs were calculated and log transformed. R2 calculated rom Pearson’s R. c, Individual correlation plots for targets highlighted in Supplementary Fig. 7b. Representative targets shown with a minimum of n=10 data points significantly above background for nCounter and covering at least one order of magnitude of expression. Individual R2 calculated from Pearson’s R are shown.

Supplementary Figure 15 NGS readout allows high-plex analyses of each transcript.

a, Graphs of NGS readout counts from multiplexing 928 RNA in situ hybridization probes onto the 23-core cell pellet array. Glass negative control graph placed on the top right. Y-axis indicates NGS digital counts, with each point representing the average of the three ROIs replicates for each cell pellet. X-axis indicates each tile grouped by gene. 80 negative control probes are grouped into sets of 10. b, Reproducibility and scalability with the NGS readout of RNA probes as plex is increased from 96 gene/ 928 probes to 1412 gene/4998 probes. DSP high plex RNA was performed on 5 um FFPE sections of tissue microarrays (TMA) consisting of 23 cell lines. Data from ROIs obtained for four cell pellets is shown. Counts for overlapping content between the 96 gene and 1412 gene panels had an R2 = 0.93. Two independent experiments (serial sections) with the 1412 gene/4998 probe panels had an R2 = 0.95. c, Counts from three cell lines and a negative control ROI (glass surface). Selected targets are shown with each of the ten tiles plotted from left to right to highlight similar behavior of each tile. Negative control probes are shown on the far right for each graph.

Supplementary information

Supplementary Information

Supplementary Figs. 1–15

Reporting Summary

Supplementary Table 1

Oligonucleotide-conjugated antibodies used in these studies. Forty-four antibodies conjugated to unique PC-oligos were mixed into a single cocktail for these studies. NanoString provides a NanoString Protein Probe ID for each unique antibody conjugation clone in lieu of specific identifying information about the antibodies as NanoString considers this confidential information. Conjugated IHC validation (as shown in Fig. 1b) was performed for most antibodies in these studies. Antibody validation was performed on positive-control FFPE tissue samples and FFPE cell pellets.

Supplementary Table 2

RNA probe target sequences used in these studies. The unique RNA probe ID, target transcript and exon targeted are shown.

Supplementary Table 3

FFPE samples used in these studies.

Supplementary Table 4

High-density FFPE tissue microarray samples used in these studies.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Merritt, C.R., Ong, G.T., Church, S.E. et al. Multiplex digital spatial profiling of proteins and RNA in fixed tissue. Nat Biotechnol 38, 586–599 (2020). https://doi.org/10.1038/s41587-020-0472-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41587-020-0472-9

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing