Combinatorial transcription factor profiles predict mature and functional

human islet α and β cells

Shristi Shrestha*, Diane C. Saunders*, John T. Walker*, Joan Camunas-Soler, Xiao-Qing Dai,

Rachana Haliyur, Radhika Aramandla, Greg Poffenberger, Nripesh Prasad, Rita Bottino, Roland Stein,

Jean-Philippe Cartailler, Stephen C. J. Parker, Patrick E. MacDonald, Shawn E. Levy, Alvin C.

Powers#, Marcela Brissova#

* These authors contributed equally, # Corresponding authors

Supplemental Information A Arda et al. 2016 B Blodgett et al. 2015 30 α 250 α 25 β β 200 20 * 15 150 * 10 100 * 5 50 Normalized expression (TPM)

Normalized expression (RPKM) 0 0 ARX MAFA MAFB ARX MAFA MAFB

ARX MAFA MAFB C (Tissue donor 6) donor (Tissue 20y

D (Tissue donor 7) donor (Tissue 35y

GCG (α cells) CPEP (β cells)

Supplemental Figure 1. ARX is expressed specifically in human α cells and MAFA in β cells, while MAFB is expressed in both α and β cells. (A–B) Normalized expression of ARX, MAFA, and MAFB in previously published bulk RNA-sequencing (RNA-seq) datasets Arda et al. 2016 (10) (A) and Blodgett et al. 2015 (26) (B) from α cells (green) and β cells (blue). Bars in both panels show mean + SEM; symbols represent individual donors. Asterisks indicate significantly different (adjusted p-value <0.05) fold change (α vs. β). See also Figure 1B. (C–D) Immunohistochemical staining of pancreatic sections from nondiabetic adults (Table S4), showing specificity of ARX, MAFA, and MAFB (red) in α cells (GCG; green) and β cells (CPEP; blue). Arrowheads indicate cells negative (white) or positive (purple) for transcription factors; scale bars, 50 μm. See also Figure 1C.

S1 See also also See RNA bulk to unique expressed highly most 1,000 (grey). stress markers for α ( α for markers (D 2G. Figure 2A Figure (B) endocrine marker) and additional positivity for HPa3 ( HPa3 for (pan HPi1 positivity for additional and marker) endocrine positivity on based isolated then were subpopulations cell Endocrine (PI). iodide propidium using excluded were scatter (FSC) and side scatter (SSC), single cells were identified by voltage pulse geometry(FSC pulse voltage by identified were cells single (SSC), scatter andside (FSC) scatter (A) clustering. unsupervised by identified those and markers surface cell by analyzed cells β and α Purified 2. Figure Supplemental B A D Number of genes detected was similar among α and β cells identified by cell surface markers and unsupervised clustering. See clustering. unsupervised and markers surface cell by identified cells β and α among similar was detected genes of Number F

Figure 2D Figure Log (expression) WI-SC-α Number of genes for schematic of experimental design. DPP4

1000 2000 3000 4000 14y (islet prep 2) 39y (Islet prep 1) (F – D ARX TM4SF4 E) IRX2

Gating strategy for sorted α and β cells identified by cell surface markers. Cell debris were excluded by forward forward by excluded were debris Cell markers. surface cell by identified cells β and α sorted for strategy Gating SSC-A –

) and) β cells (E SSC-A WI

G) G) 11,512 cells o epeso) FACS Log (expression)

Comparison of average log expression of genes across cells identified by unsupervised clustering or cell surface cell or clustering unsupervised by identified cells across genes of expression log average of Comparison -SC ALDH1A1 KEGG enrichment performed on genes common between between common genes on performed enrichment KEGG and -

DDIT3 α Islet cells Islet FSC 2F

WI -A 6,247 cells MAFB -

. SC - HERPUD1 β ATF4 ). Genes ). highlighted are α cell FACS FSC-H FSC-H

XBP1 7,193 cells -SC -α Single cells - SC

FACS FSC GCG -α

3,129 cells -A -SC -β (C)

Count Count Heatmap depicts expression of genes contributing to variability in PCA of PCA in variability to contributing genes of expression depicts Heatmap α Propidium Iodide cells) or NTPDase3 or ( cells) - seq (F, (F, seq Live cells - enriched (yellow), β cell β (yellow), enriched α cells; G, cells; G C E

TMEM176A TMEM176B FSC-A FSC-A ADCYAP1 SAMD11 PRSS23 Log (expression) WI-SC-β SLC7A2 PCSK1 FXYD3 NPTX2 FABP5 Endocrine cells TIMP2 MEG3 HADH RGS4 GCG GSN FAP β MAFA F10 INS GC β cells) using using cells) HPi1 cells). Antibody information can be found in in foundbe can information Antibody cells). PDX1 scRNA

o epeso) FACS Log (expression) WI

NKX6

- SC - - seq and bulk RNA andbulk seq enriched (blue), or selected markers of cell of markers selected or (blue), enriched

DDIT3

- α -1

Metascape NTPDase3 NTPDase3 ATF4 XBP1 HERPUD1 α

and WI HPa3 IAPP

β

- SC - cells

A v. FSC v. A -

. β - SC FACS - -β seq as well as on the onas well asseq INS - - SC H), and non and H),

SC

FACS

-

β - Expression Scaled - -1 viable ce viable 0 1 2 Table S3 Table - S2 lls . G 16.7 G 1.7 G 16.7 G 1.7 G 16.7 + IBMX 100 + Epi 1 KCl 20 G 16.7 + IBMX 100 + Epi 1 KCl 20 A 6 250 G 5.6 G 5.6 G 5.6 G 5.6 G 5.6 G 5.6 G 5.6 G 5.6 G 5.6 G 5.6

5 200

4 150

3 100 2 Insulin secretion (pg/100 IEQs/min) (ng/100IEQs/min) Glucagon secretion 50 1

0 0 0 30 60 90 120 150 0 30 60 90 120 150 Time (min) Time (min)

B 66 C Cell Type α cells β cells 59 α Acinar

50 β Ductal δ Endothelial 39 γ Stellate 14y Donor age (years) 39y 14 ε Immune 50y 59y 0 10 20 30 40 50 60 70 80 90 100 66y Frequency (%)

D E

F

Supplemental Figure 3. Detailed characterization of endocrine cells from five nondiabetic human islet donors. (A) Insulin and glucagon secretion were assessed in islets isolated from n=5 donors (age range 14–66 years) stimulated with 5.6 mM glucose (G 5.6), 16.7 mM glucose (G 16.7), 16.7 mM glucose + 100 mM isobutylmethylxanthine (IBMX) (G 16.7 + IBMX 100), 1.7 mM glucose + 1 mM epinephrine (G 1.7 + Epi 1), and 20 mM potassium chloride (KCl 20). Insulin and glucagon secretion is normalized to overall islet cell volume (expressed as islet equivalents; IEQs). (B) Bar graph illustrating cell type distribution within each islet preparation as per cell types annotated in Figure 3A. (C) UMAP of only α or β cells, showing clustering by islet preparation. See also Figures 4A and 5A. (D) User-friendly web portal application for searching and viewing pancreatic cell types and their expression; available at https://powersbrissovalab.shinyapps.io/scRNAseq-Islets/. (E) Number of genes detected in cells expressing low (ln[UMI per 10,000 +1] <0.5) or high (ln[UMI per 10,000 +1] >0.5) levels of transcription factors ARX and MAFB in α cells (pink) and MAFA and MAFB in β cells (green); dotted line represents quality threshold of 500 genes. (F) PCA showing cell cycle genes among the four MAFA/MAFB β cell populations. S3 9 6 3

2

1 Log expression 0 RFX6 PDX1 PAX6 NKX6-1 NKX2-2 NEUROD1 MAFB MAFA ISL1 IRX2 # ^ # ^ $ ^ # ^ $ # ^ $ # ^

9 6 3

2

1 Log expression 0 FOXA2 ARX TTR SLC38A4 PTPRT POU6F2 PLCE1 PCSK2 PAPPA2 LOXL4 ^ # ^ $ ^ $ # ^ $ ^ $ # ^ $

9 6 3

2

1 Log expression 0 KLHL41 IGFBP2 GC FXYD3 DPP4 TMCO1 SCN3B KCTD12 KCNMA1 KCNK3 $ # $ # ^ ^

9 6 3

2

1 Log expression 0 KCNK16 KCNJ8 KCNJ6 CACNB2 CACNA2D1 CACNA1D CACNA1C CACNA1A ABCC8 ACLY # ^ # ^ $ ^ # ^ # ^

9 6 3

2

1 Log expression 0 PKM PDHA1 MDH2 GSTK1 GSTA4 GPX3 GPI G6PC2 ENO2 VTI1B # ^ # ^ $ # ^ ^ # ^ $ #

9 6 3

2

1 Log expression 0 VAMP2 SYT7 SYT5 SYT13 STXBP1 STX1A SNAP25 RIMS2 PCLO TSPAN7 # ^ ^ ^

9 6 3 None ARX MAFB Both 2 # None differs from 1 all other groups ^ Both differs from Log expression 0 all other groups TSPAN1 RGS9 GYG1 FAM159B BMP5 HSPA5 HERPUD1 DDIT3 ATF4 $ ARX and # ^ $ # $ MAFB differ

Supplemental Figure 4. Raw expression values for transcription factor, α cell-enriched, ion channel, glucose metabolism, vesicle trafficking, exocytosis, and stress genes in ARX/MAFB populations. Violin plots depict gene expression in α cell populations lacking ARX and MAFB (grey) and those expressing MAFB only (red), ARX only (blue), or co-expressing both MAFB and ARX (purple); n=24,248 total α cells. Data corresponds to dot plot in Figure 4C. Symbols underneath gene names indicate significance (p<0.05) from Tukey’s multiple comparisons test following 2-way ANOVA. S4 A Baron et al. 2,326 α cells, 4 donors (InDrop) Transcription factors α-enriched Ion channels Glucose metabolism Vesicle trafficking Exocytosis Stress

Segerstolpe et al. 415 α cells, 5 donors (Smart-Seq2)

Transcription factors α-enriched Ion channels Glucose metabolism Vesicle trafficking Exocytosis Stress

Camunas et al. 511 α cells, 18 donors (Smart-Seq2) Transcription factors α-enriched Ion channels Glucose metabolism Vesicle trafficking Exocytosis Stress

Current study 24,248 α cells, 5 donors (10x) Transcription factors α-enriched Ion channels Glucose metabolism Vesicle trafficking Exocytosis Stress

1.0 0 Expression 0.5 25 Cells expressing z-score 0.0 50 (% total cells) -0.5 75 100 -1.0

B DAPI ARX MAFB ARX MAFB GCG DAPI ARX MAFB ARX MAFB GCG 3 3 3 3

4 4 4 4 4 4 4 4 3 3 3 3

1 1 1 1

1 1 1 1 20y (Tissue donor 6) donor (Tissue 20y 35y (Tissue donor 7) donor (Tissue 35y 2 2 2 2 2 2 2 2

Supplemental Figure 5. Validation of α cell populations based on ARX and MAFB expression, as determined by previous scRNA-seq studies. (A) Dot plots showing the expression patterns of selected genes related to cell identify, ion flux, glucose metabolism, vesicle trafficking, and exocytotic machinery. α cell populations (rows) are identified by expression of neither ARX nor MAFB (None), ARX only (ARX), MAFB only (MAFB), and co-expression of ARX and MAFB (Both). Dot size indicates the percentage of cells with detectable transcripts; color indicates gene’s average scaled expression. Headers list study details from previously published datasets (Segerstolpe et al., 2016 [29]; Baron et al., 2016 [28]; Camunas-Soler et al., 2020 [18]) and final dot plot is as shown in Figure 4C for comparison. (B) Immunohistochemical staining of ARX (blue) and MAFB (red) in glucagon (GCG)- expressing α cells (green) of two nondiabetic adults (Table S4). Numbered arrowheads indicate the presence of α cell populations: 1, ARXlo MAFBlo; 2, ARXhi MAFBlo; 3, ARXlo MAFBhi; 4, ARXhi MAFBhi. See also Figure 4E. S5 8 6 4 3 2 1 Log expression 0 RFX6 PDX1 PAX6 NKX6-1 NKX2-2 NEUROD1 MAFB MAFA ISL1 IRX2 # ^ # ^ $ # # ^ $ # ^

8 6 4 3 2 1 Log expression 0 FOXA2 ARX SIX3 SAMD11 RGS16 RASD1 PCSK1 NPTX2 MEG3 IAPP $ # ^ # # ^ # # $ # ^

8 6 4 3 2 1 Log expression 0 GSN ENTPD3 DLK1 CASR ADCYAP1 TMCO1 SCN3B KCTD12 KCNMA1 KCNK3 # # ^ $

8 6 4 3 2 1 Log expression 0 KCNK16 KCNJ8 KCNJ6 CACNB2 CACNA2D1 CACNA1D CACNA1C CACNA1A ABCC8 ACLY # $ ^ # ^ # ^

8 6 4 3 2 1 Log expression 0 PKM PDHA1 MDH2 GSTK1 GSTA4 GPX3 GPI G6PC2 ENO2 VTI1B # $ ^ # ^ $

8 6 4 3 2 1 Log expression 0 VAMP2 SYT7 SYT5 SYT13 STXBP1 STX1A SNAP25 RIMS2 PCLO TSPAN7 # ^ ^ # # # ^ #

8 6 None MAFB 4 MAFA Both 3 # None differs from 2 all other groups 1 ^ Both differs from

Log expression all other groups 0 $ MAFA and TSPAN1 RGS9 GYG1 FAM159B BMP5 HSPA5 HERPUD1 DDIT3 ATF4 MAFB differ # # $ # # ^ # #

Supplemental Figure 6. Raw expression values for transcription factor, β-enriched, ion channel, glucose metabolism, vesicle trafficking, exocytosis, and stress genes in MAFA/MAFB populations. Violin plots depict gene expression in β cell populations (n=11,034 total β cells) lacking MAFA and MAFB (grey) and those expressing MAFA only (red), MAFB only (blue), or co-expressing both MAFA and MAFB (purple); n=11,034 total β cells. Data corresponds to dot plot in Figure 5C. Symbols underneath gene names indicate significance (p<0.05) from Tukey’s multiple comparisons test following 2-way ANOVA. S6 A Baron et al. 2,525 β cells, 4 donors (InDrop) Transcription factors β-enriched Ion channels Glucose metabolism Vesicle trafficking Exocytosis Stress

Segerstolpe et al. 157 β cells, 5 donors (Smart-Seq2) Transcription factors β-enriched Ion channels Glucose metabolism Vesicle trafficking Exocytosis Stress

Camunas et al. 194 β cells, 18 donors (Smart-Seq2) Transcription factors β-enriched Ion channels Glucose metabolism Vesicle trafficking Exocytosis Stress

Current study 11,034 β cells, 5 donors (10x) Transcription factors β-enriched Ion channels Glucose metabolism Vesicle trafficking Exocytosis Stress

1.0 0 Expression 0.5 25 Cells expressing z-score 0.0 50 (% total cells) -0.5 75 100 -1.0

B DAPI MAFA MAFB MAFA MAFB CPEP DAPI MAFA MAFB MAFA MAFB CPEP

1 1 1 1 1 1 1 1

2 2 2 2 2 2 2 2

3 3 3 3 3 3 3 3 20y (Tissue donor 6) donor (Tissue 20y 7) donor (Tissue 35y 4 4 4 4 4 4 4 4

Supplemental Figure 7. Validation of β cell populations based on MAFA and MAFB expression, as determined by previous scRNA-seq studies. (A) Dot plots showing the expression patterns of selected genes related to cell identify, ion flux, glucose metabolism, vesicle trafficking, and exocytotic machinery. β cell populations (rows) are identified by expression of neither MAFA nor MAFB (None), MAFA only (MAFA), MAFB only (MAFB), and co-expression of MAFA and MAFB (Both). Dot size indicates the percentage of cells with detectable transcripts; color indicates gene’s average scaled expression. Headers list study details from previously published datasets (Segerstolpe et al., 2016 [29]; Baron et al., 2016 [28]; Camunas-Soler et al., 2020 [18]) and final dot plot is as shown in Figure 5C for comparison. (B) Immunohistochemical staining of MAFA (red) and MAFB (blue) in C-peptide (CPEP)-expressing β cells (green) of two nondiabetic adults (Table S4). Numbered arrowheads indicate the presence of β cell populations: 1, MAFAlo MAFBlo; 2, MAFAhi MAFBlo; 3, MAFAlo MAFBhi; 4, MAFAhi MAFBhi. See also Figure 5E. S7 Table S1. Human Islet Donor Information

Checklist for Reporting Human Islet Preparations Used in Research Adapted from Hart NJ, Powers AC (2018) Progress, challenges, and suggestions for using human islets to understand islet biology and human diabetes. Diabetologia https://doi.org/10.1007/s00125-018-4772-2.

Islet preparation 1 2 3 4 5

RRID: RRID: Unique identifier DON184 DON185 R232 SAMN08768781 SAMN08768783

Donor age (years) 14 39 50 59 66

Donor sex Male Female Male Female Female

Donor BMI (kg/m2) 24.13 34.76 22.4 32.3 18.5

Donor HbA1c 5.4 4.7 N/A N/A 6.1 (ng/mL)

Origin/source of OPO OPO IIDP IIDP ADI islets

University of University of University of University of University of Islet isolation center Pennsylvania Pennsylvania Pennsylvania Wisconsin Alberta

Donor history of No No No No No diabetes? Yes/No

Donor cause of Anoxia/ Anoxia/ Cerebrovascular/ Cerebrovascular/ Cerebrovascular/ death Cardiovascular Drug intoxication Stroke Stroke Stroke Glucose-stimulated insulin secretion or Perifusion Perifusion Perifusion Perifusion Perifusion other functional measurement Handpicked to Yes Yes Yes Yes Yes purity? Yes/No

Warm ischaemia N/A N/A No No N/A time (hours)

Cold ischaemia time 12 8.5 9.9 N/A N/A (hours)

Estimated purity (%) 75 95 95 95 90

Total culture time 36 96 19 44 144 (hours)

ADI, Alberta Diabetes Institute; IIDP, Integrated Islet Distribution Program; N/A, not available; OPO, Organ Procurement Organization

S8 Table S2. Markers used for cell type annotation

Cell type Gene marker(s) α cell GCG β cell INS δ cell SST γ cell PPY ε cell GHRL Acinar PRSS1 Ductal KRT19 Stellate PDGFRB, COL1A1 Endothelial PECAM1 Immune HLA-DRA

S9 Table S3. Antibodies used for immunohistochemistry and flow cytometry

Antigen/Conjugate Species Source Catalog # Application Dilution

ARX Sheep R&D Systems AF7068 IHC (1º) 1:2000 Developmental Studies C-peptide Rat GN-ID4 IHC (1º) 1:200 Hybridoma Bank NTPDase3 (CD39L3) Mouse J. Sévigny N/A FC (1º) 1:50

Glucagon Mouse Abcam ab10988 IHC (1º) 1:100

HPa3 (HIC3-2D12) Mouse P. Streeter/M. Grompe N/A FC (1º) 1:200

HPi1 (HIC0-4F9) – biotin Mouse Novus NBP1-18872B FC (1º) 1:100

MAFA Rabbit Novus NBP1-00121 IHC (1º) 1:250

MAFB Mouse R&D Systems MAB3810 IHC (1º) 1:1000

Mouse Ig – APC Goat BD Biosciences 550826 FC (2º) 1:500

Mouse IgG – Cy3 Donkey Jackson Immunoresearch 715-165-150 IHC (2º) 1:500

Mouse IgG – Cy5 Donkey Jackson Immunoresearch 715-175-150 IHC (2º) 1:300

Mouse IgM – PE Goat Jackson Immunoresearch 115-116-075 FC (2º) 1:1000

Rabbit IgG – Cy3 Donkey Jackson Immunoresearch 711-165-152 IHC (2º) 1:500

Rabbit IgG – Cy5 Donkey Jackson Immunoresearch 711-175-152 IHC (2º) 1:300

Rat IgG – Cy2 Donkey Jackson Immunoresearch 712-225-153 IHC (2º) 1:500

Sheep IgG – Cy2 Donkey Jackson Immunoresearch 713-225-147 IHC (2º) 1:500

Streptavadin – BV420 N/A BD Biosciences 563259 FC (2º) 1:500

1º, primary antibody; 2º, secondary antibody; FC, flow cytometry; IHC, immunohistochemistry

S10 Table S4. Human Pancreatic Tissue Donor Information

Checklist for Reporting Human Islet Preparations Used in Research Adapted from Hart NJ, Powers AC (2018) Progress, challenges, and suggestions for using human islets to understand islet biology and human diabetes. Diabetologia https://doi.org/10.1007/s00125-018-4772-2.

Pancreatic tissue 6 7 8

Unique identifier DON54 DON42 DON61

Donor age (years) 20 35 55

Donor sex Male Male Male

Donor BMI (kg/m2) 19.442 26.852 35.565

Donor HbA1c (ng/mL) 5.6 5.1 Not recorded

Origin/source of tissue IIAM IIAM IIAM

Donor history of No No No diabetes? Yes/No

Donor cause of death Head trauma Head trauma Cerebrovascular/Stroke

Glucose-stimulated insulin secretion or Not done Not done Not done other functional measurement Warm ischemia time DBD DBD No (hours)

Cold ischemia time 6.2 15.1 26.6 (hours)

ADI, Alberta Diabetes Institute; DBD, Donation after Brain Death (minimal warm ischemia time); IIAM, International Institute for the Advancement of Medicine

S11