Massively parallel simultaneous profiling of the transcriptomic and epigenomic landscape at single cell resolution

Shamoni Maheshwari, Sharmila Chatterjee, Jerald Sapida, Joe Shuga, Li Wang, Yiming Kang, Brett Olsen, Vijay Kumar, Corey Nemec, Martin Sauzade, Maengseok Song, Catie McConnell, Jason Bell, Jean Wang & Kamila Belhocine 10x Genomics, Inc., Pleasanton, CA

Introduction Parsing cellular heterogeneity within a lymphoma • We present here a single cell high-throughput assay to simultaneously profile 3’ expression (GEX) and open chromatin (via ATAC-seq) We generated joint GEX and ATAC data from 9158 single cells from a CD20+ diffuse small cell lymphoma sample. Using from the same nucleus. markers, we annotated the immune cell types present and could classify this tumor as a gastric MALT lymphoma (data not shown). We used • Such simultaneous profiling allows for finer resolution of cell types both in terminally differentiated cells and cells on a developmental trajectory. two orthogonal lines of evidence to differentiate between tumor and normal B cells: i) mapping mutation load using SNPs from a TCGA B cell • This assay has the potential to reveal the complex interplay between regulatory elements, open chromatin and gene expression. lymphoma project and ii) the BANK1-CD40 axis to identify the hyper-activated tumor B cells. Workflow A. GEX-derived tsne

• The assay input consists of a single nuclei suspension that undergoes transposition in bulk. CD20 SNVS alternate alelle We used publicly replicating tumor B cells reference alelle Weavailable used publicly mutation • Transposed nuclei are partitioned using a microfluidic chip along with Gel Beads that contain barcoded oligonucleotides and enzymes that B cells no data availabledata from mutation the dataTCGA-DLBC from the project enable the simultaneous conversion of polyadenylated mRNA and open chromatin into barcoded cDNA and ATAC fragments, respectively. TCGA-DLBCon Diuse projectLarge onB Diffuse cell Lymphoma Large B normal B cells • After partitioning a standard NGS workflow yields two libraries: a barcoded 3’ GEX library and a barcoded ATAC library. replicating celland Lymphoma filtered it and to T cells filteredretain it onlyto retain 279 SNVs • These libraries can be sequenced on a variety of Illumina sequencers. onlythat 279 were SNVs predicted that wereto have predicted a deleterious to • The sequencing reads are processed via Cell Ranger, a turnkey software solution that performs barcoded read alignment, cell barcode havephenotype a deleterious and phenotypepresent in and the dbSNP database. identification, and secondary analyses like dimensionality reduction, clustering, and differential analyses. monocytes present in the dbSNP database. • Furthermore, the GEX and ATAC data can be analyzed using Loupe Browser, allowing the user to explore the data in an intuitive manner. T cells B. ATAC-derived tsne

BANK1 (B cell scaold Collect GEM Post GEM-RT ATAC Single Cell BANK1 CD40 monocytes with ankyrin Incubation Cleanup SI-PCR Pre-amplification ATAC Library BANK1repeats (B cell 1) modulatesscaffold protein with ankyrin 10x Barcoded normal B cells B-cell antigen Gel Beads repeatsreceptor 1) modulates (BCR)- cDNA Amplification B-cellinduced antigen calcium  Gene receptormobilization (BCR)- and Oil in Well Expression inducedweakens calcium CD40- mobilizationmediated andAkt Fragmentation weakensactivation CD40- to End Repair & Ligation mediatedprevent Akt B cell activationhyper-activation to  [1]. Transposed tumor B cells prevent B cell  Nuclei hyper-activation [1]. T cells Enzymes SI-PCR

Transposition of Single Nuclei 10x Barcoded Accessible 10x Barcoded Nuclei in bulk GEMs DNA Fragments + DNA + RNA Single Cell Barcoded mRNA Gene Expression Library Identifying a tumor-specific signaling pathway Dysregulation of the JAK-STAT pathway is frequently observed in many primary human tumors. Differential gene expression analysis Sample prep GEM generation Library construction Sequencing Data processing Data visualization revealed over-expression of the IL4 receptor (IL4R) in the tumor B cells (A), and furthermore, we found that the STAT6 binding motif is accessible specifically in the tumor B cells (B).

Prepare nuclei A B IL4 Stat3 Stat6 suspension SignalSignal transducer transducer

normal B tumor B P andand activator activator of of P Jak3 Expression transcriptiontranscription (STAT) (STAT) FCRL5 Jak1 P proteinsproteins are are critical critical mediators of cytokine GRHPR P Stat6 mediators of cytokine signaling. However, P Stat3 signaling. However, MBD4 STATsSTATs are are latent latent cytoplasmiccytoplasmic proteins A sensitive and scalable solution MIR155HG makingmaking expression expression

a poor proxy for

P a poor proxy for P Stat3

• Shown in panel A are the results of a 5000 cell PBMC experiment where the ATAC and GEX libraries were sequenced to a depth of ~65k and FCRL3 Stat6 function. Stat3 Stat6 function. 57k raw read pairs per cell, respectively. The partitions that contain nuclei (in yellow) have strong ATAC signal (measured as high mapping P P IL4R IL4 quality fragments on X axis) and GEX signal (measured as high mapping quality transcriptomic UMI counts on Y axis) and are well separate UponUpon Jak-mediated Jak-mediated RASGRF1 Stat3 MOTIF Stat6 MOTIF phosphorylationphosphorylation an an activated STAT from the noise (in black) due to empty partitions. P activated STAT 1.0 P Jak3 translocates to the IGF2BP3 Accessibility translocates to the Jak1 nucleus and binds to • The assay allows for profiling between 500 to 10,000 single nuclei in a single chip well. Shown in panel B are the results of loading ~1.5k cells P nucleus and binds to QSOX2 itsits DNA-recognition DNA-recognition 0.5 P Stat6 motifs,motifs, in inthe the that are a mixture of human GM12878 and mouse 3T3 nuclei. Note that most barcodes either contain exclusively human reads (yellow, 567 cells) P Stat3 PDGFD promoterspromoters of of  or exclusively mouse reads (green, 935 cells) and 13 observed doublets (blue). Based on the number of observed human-mouse doublets, 0 cytokine-induciblecytokine-inducible IGLC2 .genes. Among Among the the

we estimate the total doublet rate at approximately 1% in this experiment. sevenseven STAT STAT proteins, proteins,

P P TIMP1 Stat3

−0.5 Stat6 STAT6STAT6 is isactivated activated by by Stat3 Stat6 • Shown in panel C is a comparison of nuclei from an embryonic mouse brain (E18) sample split and run on Chromium Single Cell Gene P P FN1 IL-4IL-4 and and IL-13 IL-13 [2]. [2]. −1.0 Expression v3, Chromium Single Cell ATAC and Chromium Single Cell ATAC + Gene Expression. Here we show that the sensitivity of the joint assay IL4 GBP5 Bcl-2 IgE is comparable to the single assays (on transposed nuclei) as a function of average sequencing depth per cell. STAT6STAT6 activation activation is is UTRN P P Jak3

Expression knownknown to to promote promote Jak1 immunoglobulin GBP1 P immunoglobulin A.5k PBMC B. 1.5k barnyard C. 1k embryonic mouse brain (E18) classclass switching switching to to IgE IgE P Stat6 andand prevention prevention of of LMO4 P Stat3 apoptosisapoptosis through through

ATAC BANK1 thethe induction induction of of  60,000 anti-apoptoticanti-apoptotic genes genes

15,000 (e.g. Bcl-2) [2,3]. IFI44L (e.g. Bcl-2) [2,3].

P

P

Stat3 Stat6

4 Stat3 Stat6 10 40,000 P P XAF1 10,000 mm10 UMI counts GRCh38 cell 20,000 Chromium Single Cell Unique HQ fragments mm10 cell ATAC + Gene Expression 5,000 3 10 doublet Chromium Single Cell ATAC 0 Linking cis-regulatory elements to target genes 0 0 20,000 40,000 60,000 0 5,000 10,000 15,000 20,000 25,000 30,000 Cell Barcode GRCh38 UMI counts Total reads Chromatin accessibility tracks around the IL4R highlight an upstream peak specific to the tumor B cells (A). The UCSC browser plot of that 102 Non-Cell Barcode UMIs RNA UMI counts 1000 same region shows an overlap with H3K4me1, a mark often found near regulatory elements, and an interaction in the curated GeneHancer

200,000 800 database (B). Accessibility at this linked regulatory peak is a much better predictor of IL4R expression than its promoter accessibility (C). 101 600

A C 100,000 ansposition events in peaks Chromium Single Cell 400 1.00 GRCh38 cell Gene Expression v3 Median Genes per Cell 3.47 mm10 cell IL4R expression Chromium Single Cell mm10 tr 0 doublet ATAC + Gene Expression tumor B cells 10 200 0.75 1 2 3 4 5 0 10 10 10 10 10 10 0 3.47 0 ATAC transposition events 0 100,000 200,000 300,000 400,000 0 10,000 20,000 30,000 40,000 0.50 GRCh38 transposition events in peaks Raw reads per cell normal B cells

3.47 0.25 Conclusions monocytes 3.47 1.00 chr16:27311587-27320059 T cells

• We demonstrate the ability to simultaneously measure single cell gene expression and chromatin accessibility from the same nucleus. promoter peak 0.75 • Combined GEX and ATAC-seq profiling enables deeper cell phenotyping and association of regulatory elements to gene expression. IL4R 0.50 • Interrogation of a diffuse small cell lymphoma sample linked a tumor-specific regulatory element to higher IL4R expression in tumor B cells. B chr16: 27,300,000 27,310,000 27,320,000 27,330,000 27,340,000 27,350,000 27,360,000 27,370,000 27,380,000 27,390,000 • Scalable workflow enables experiments ranging from 500-10,000 nuclei per sample, with low multiplet rates. 0.25 GENCODE v32 Comprehensive Transcript Set AC106739.1 IL4R IL4R IL4R • End-to-end solution from nuclei sample to turnkey data analysis and visualization software IL4R 1.00 50 _ chr16:27297936-27300205 H3K4Me1 Mark on 7 cell lines from ENCODE Layered H3K4Me1 0 _ • Compatible with cell suspensions, fresh and frozen samples. 100 _ 0.75 linked peak Layered H3K27Ac H3K27Ac Mark on 7 cell lines from ENCODE 0 _ GeneHancer Regulatory Elements and Gene Interactions 0.50 GH Reg Elems (DE) References IL4R GeneCards genes TSS (Double Elite) Interactions between GeneHancer regulatory elements and genes (Double Elite) 0.25 1. Aiba, Y. et al. BANK negatively regulates Akt activation and subsequent B cell responses. Immunity 24, 259–268 (2006). 2. Wurster, A. L., Tanaka, T. & Grusby, M. J. The biology of Stat4 and Stat6. Oncogene 19, 2577–2584 (2000). T cells monocytes normal B tumor B 3. Scott, D. W. & Gascoyne, R. D. The tumour microenvironment in B cell lymphomas. Nat. Rev. Cancer 14, 517–534 (2014)