bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

1

2 AFF3 and BACH2 are master regulators of metabolic inflexibility, b/a-cell transition, and

3 dedifferentiation in type 2 diabetes

4

5

1,2* 3,* 1,2# 3,# 6 Jinsook Son , Hongxu Ding , Domenico Accii , Andrea Califano

7

8

1 2 3 9 Naomi Berrie Diabetes Center and Departments of Medicine and Systems Biology

10 Vagelos College of Physicians & Surgeons of Columbia University

11 New York, New York 10032, USA

12

13

14

15

16

17

18 * These authors contributed equally

19 # To whom correspondence should be addressed: ([email protected]) or

20 ([email protected])

21 Conflict of Interest Statement: A.C. is founder, director, SAB member and shareholder of

22 DarwinHealth Inc., a Company that has licensed the VIPER software. Other authors have

23 declared that no conflict of interest exists. Columbia University is also a shareholder of

24 DarwinHealth Inc.

1 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

25 ABSTRACT

26

27 Type 2 Diabetes is associated with defective insulin secretion, reduced b-cell mass, and

28 increased glucagon production. Cell lineage-tracing in rodents and human autopsy surveys

29 support the notion of b-cell dedifferentiation as a unifying mechanism for these abnormalities.

30 Yet, mechanistic determinants of human b-cell failure remain elusive. Using regulatory-network-

31 based single-cell analysis of human islets, we identify aberrant, diabetes-enriched transitional

32 states characterized by metabolic inflexibility, a/b-transition, and endocrine progenitor/stem cell

33 features. A coordinated transcription factor hierarchy mediating cell state transition emerged

34 and was validated using barcoded guide-based, single-cell gene transfer and calcium flux

35 measures in primary human islet cells. Specifically, two master regulators and associated

36 epigenetic drivers emerged, one (AFF3) controlling b- to a-like-cell reprogramming, the other

37 (BACH2) transition to a dedifferentiated endocrine progenitor-like cell. The findings provide

38 mechanistic insight into diabetic islet cell dysfunction and suggest actionable pathways for

39 pharmacological intervention.

40

2 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

41 INTRODUCTION

42

1 43 Type 2 diabetes (T2D) is characterized by defects of insulin action and production . The latter

44 include pancreatic b-cell function and mass abnormalities, which in turn likely impinge on the

2 45 function of other pancreatic hormones, primarily glucagon . Unveiling mechanisms of islet cell

46 dysfunction is important for our ability to design mechanism-based, durable, and safe disease-

3 47 modifying interventions. Theories on b-cell failure abound . Among them, dedifferentiation of

4 48 mature b-cells into endocrine progenitor-like cells , with an attendant increase in cell

5,6 49 heterogeneity, has been demonstrated by cell lineage tracing in rodents . Human autopsy

7-9 50 studies are consistent with this hypothesis but, in view of morphological and functional

51 differences between the two species, cannot be considered dispositive.

52 That b-cell undergo dedifferentiation is teleologically attractive, in view of the clinical

10,11 53 features associated with b-cell failure: limited adaptive potential of b-cell mass , increased

2 12 54 glucagon “tone” , rapid progression in response to incipient hyperglycemia , and prompt

13 55 reversibility upon treatment, at least in the initial phases of the disease . In this regard, recent

14 56 clinical trials on diabetes reversal can be reconciled with the idea of b-cell dedifferentiation .

57 Unfortunately, the limited ability to access the endocrine pancreas in vivo, let alone in a

58 prospective fashion, hinders our ability to functionally interrogate islet cells. Several groups have

59 thus sought to address this problem using computational cell state clustering at the single-cell

15-21 60 level . Yet, while these studies have demonstrated islet cell gene expression

18,20,22 61 heterogeneity , a clear diabetic islet cell signature remains elusive.

23 62 To circumvent these limitations , we employed a systems biology approach to leverage

24 63 accurate regulatory networks—as inferred by the ARACNe algorithm —for the identification of

64 master regulator (MR) proteins representing mechanistic determinants of gene expression

3 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

25 65 signatures associated with diabetic vs. non-diabetic islet cell state at the single-cell level.

66 Algorithmic predictions were then experimentally validated in a biologically relevant human

67 primary cell context. This approach, while extensively and successfully tested to elucidate

26 68 mechanistic determinants of a variety of cancer- and neurodegenerative-related phenotypes ,

69 had not yet been applied to the study of metabolic diseases.

70 Using single-cell RNA-seq (scRNA-seq) from human islets—harvested from either

71 normal (ND) or T2D donors—we first generated islet-cell-specific transcriptional regulatory

72 networks that were then used to measure the transcriptional activity of ~5600 transcriptional

73 regulator proteins—including transcription factors, co-factors, and signal-transduction proteins—

25 27 74 using an extension of the VIPER algorithm for single-cell analysis (metaVIPER ). Basically,

75 the activity of each protein is measured based on the expression of its transcriptional targets,

76 akin to a multiplexed gene-reporter assay.

77 Protein-activity-based cluster analysis identified multiple, transcriptionally-distinct

78 signatures, representing physiologic, b- and a-cells related states, as well as aberrant,

79 intermediate transcriptional states that were highly enriched in T2D-derived cells. The latter

80 were characterized by: (a) metabolic inflexibility/stress response, (b) mixed a/b-cell identity, and

81 (c) endocrine progenitor/stem cell features. The analysis also identified MR proteins

82 representing putative mechanistic determinants of the aberrant, T2D-related transitional states.

83 These analytical predictions were validated by gain-of-function studies in primary b-cells, using

84 a single-cell, ectopic gene expression methodology, where TFs were screened for their ability to

85 convert ND into T2D-enriched cells. Finally, to interrogate b-cell function of fate-converted cells,

86 we employed single-cell measurements of intracellular calcium influx in response to glucose. By

87 so doing, we identified TFs drivers of different cellular phenotypes, and obtained a

88 comprehensive functional map of cellular abnormalities in diabetic islets.

89

4 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

90 RESULTS

91

92 Heterogeneity of ND and T2D islet cells

93 We analyzed single cell populations from four non-diabetic controls (ND) and six T2D islet

94 donors of different ages, body mass index (BMI), known duration and control (HbA1c) of disease

95 (Supplementary Table 1). ScRNA-seq profiles were generated using the Fluidigm C1 high-

3 96 throughput integrated fluidic circuits (HT IFC). Since b-cell number is reduced in T2D islets , we

97 enriched samples for b-cells via FAD-mediated autofluorescence-activated cell sorting

98 (Supplementary Fig. 1), and independently analyzed scRNA-seq profiles from either b-cell-

99 enriched samples or whole islets. This approach supported a comprehensive assessment of the

100 b-cell repertoire in T2D patients, while also providing an unbiased assessment of islet cell

101 composition at the single-cell level. After loading cells onto C1 HT IFCs, we visually inspected

102 individual chambers to exclude multiple cells captured into the same chamber. A total of 6,137

103 cells met quality control criteria and were analyzed further.

104 T-SNE plotting of raw single-cell mRNA data revealed that individual sample variation

105 and batch effects (color-coded, Fig. 1A) trumped cellular identities, as indicated by the

106 emergence of single-cell clusters that were largely independent of disease status. To buffer

107 donor-to-donor variation and batch effect, we thus relied on the network-based metaVIPER

108 algorithm, which has been shown to be highly effective in removing systematic sample-to-

27 109 sample bias and batch effects that do not reflect the biology of the cellular state .

110 Transcription factors (TF) are key to understand cellular identity, but tend to be

111 expressed at much lower levels compared to most cell surface receptors or enzymes in

28 112 terminally differentiated cells . This poses a challenge when trying to reliably detect subtle

113 differences in TF levels due to the very low sequencing depth of scRNA-seq profiles. Indeed,

114 most genes (in some cases >80%) are not detected by even one transcript (i.e., gene dropout)

5 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

29 115 and most detected genes are supported by a single read . As a result, reliable differential gene

27 116 expression assessment is challenging. As previously shown , metaVIPER is especially well

117 suited to overcome this limitation because the activity of a protein is measured from tens to

118 hundreds of its transcriptional targets, thus allowing reliable protein activity assessment from

119 profiles with only 20K to 50K reads. Indeed, Spearman correlation between a 30M and a 50K

120 read, VIPER-inferred protein activity profile is ρ = 0.85, while for gene expression it is only

25 121 ρ = 0.3 . In particular, metaVIPER allows quantitative activity assessment of proteins whose

122 encoding mRNA is not detected by even a single read, thus allowing characterization of key

123 lineage marker proteins in virtually every single cell.

124 These algorithms rely on accurate mappings of proteins’ transcriptional targets

24 125 (regulons). To address this, we used the ARACNe algorithm to assemble islet cell-specific

126 regulons using 1,813 transcription factor (TF), 969 cofactor (co-TF), and 3,370 signal

127 transduction (ST) proteins. We then performed metaVIPER analysis, using these regulons, to

128 measure TF/co-TF activity in normal and T2D islet cells. As expected, protein-activity-based t-

129 SNE analyses largely removed the batch effect of individual samples and were effective in

130 identifying clusters associated with two major populations (clusters A and B) (Fig. 2B), without

27 131 noticeable donor-to-donor variation. Consistent with prior results , this shows that protein-

132 activity-based cluster analyses can buffer technical bias and batch effects typical of scRNA-seq

133 profiles to reveal hidden, biologically-relevant population structures.

134 Clusters A and B were largely independent of disease status, as evidenced by the

135 distribution of color-coded donors (Fig. 1B and Supplementary Fig. 2). Surprisingly, however,

136 they did not recapitulate the b- and a-cell populations of pancreatic islets, as shown by

137 integrative assessment of INS and GCG mRNA expression levels (Fig. 1C and 1D) and MAFA

138 and IRX2 protein activity. The latter represent well-established markers of b- and a-cells that

6 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

139 show mutually exclusive activity in single cells (Fig. 1E and 1F). Indeed, INS and GCG mRNA

140 matched perfectly with MAFA and IRX2 activity, respectively, showing that both clusters could

141 be further divided into b- and a-cell sub-clusters—thus suggesting a deeper, hierarchical single-

142 cell structure—and confirming metaVIPER’s ability to effectively measure established marker

143 activity (Fig. 1C and 1D). As shown in Supplementary Fig. 3, activity of these proteins would

144 have been virtually undetectable in a majority of cells based on single-cell expression of their

145 encoding genes.

146

147 Biological signatures of islet cell clusters

148 To assess the biology of clusters A and B (Fig. 1B), we further surveyed their differentially active

149 proteins (Supplementary Data 1). Gene Ontology (GO) analyses of cluster A vs. B (illustrated in

24,25,27 150 a REVIGO summary in Supplementary Fig. 4) , shows highly significant enrichment for GO

151 terms such as metabolism, development, and differentiation in cluster A. Among the proteins

152 subsumed under metabolism in GO, we detected a striking activation of PPARa and PPARg

153 (Fig. 2A and 2B), two markers of metabolic inflexibility–a stage in the progression of b-cell

6 154 failure (Fig. 2A and Supplementary Fig. 5) . In addition, FOXO1, HIF1a, HSF1 and TP53 were

30 155 activated in cluster A, consistent with a metabolic stress response (Supplementary Fig. 5) . We

156 also found activation of RFX6 and RFX7 (Fig. 2C and 2D). RFX6 promotes islet cell

31 32 157 differentiation, and its mutations are associated with islet agenesis and diabetes in humans.

158 Furthermore, TF proteins associated with cell stemness (NANOG, MYCL, and POU5F1) were

159 differentially active in a tightly-clustered subset of RFX6/7-positive cells (Fig. 2E and 2F),

160 especially in T2D islets. Thus, cluster A appears to be comprised of endocrine progenitor-like

4,33-35 161 cells, displaying stem cell features, a prominent signature of b-cell dedifferentiation . These

162 results suggest that the two main functional clusters in human ND and T2D islets are defined by

7 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

163 drivers of metabolic inflexibility/stress response, and endocrine-progenitor/stem-cell-like

164 features.

165

166 Sub-clustering analysis

167 To explore the fine-grain substructure of clusters A and B, we performed hierarchical clustering

36 168 using the iterCluster algorithm . The resulting sub-cluster architecture is visualized as a

169 hierarchical-clustering heatmap (Fig. 3A and Supplementary Fig. 6), and as t-SNE plots (Fig.

170 3B-3D). To refine functional characterization of individual sub-clusters, we mapped single-cell-

171 specific features, including (a) hormone expression (INS, GCG, and SST), as well as activity of

172 TFs associated with (b) either b- (MAFA, PDX1, NKX2.2, NEUROD1) or a-cell identity (IRX2,

37 173 ARX, GLI,3, IGFBP2, ITGB8, HSPB1, F10, SPOCK3, MYO10, CLU) , (c) metabolic

6 174 inflexibility/stress response (PPARa/g, FOXO1, RB1, FOXM1) , and (d) endocrine progenitor-

38 175 (PAX4, PAX6, ISL1, RFX6/7, HES1) or stem-like cell identity (POUF5F1, MYCL, NANOG).

176 This analysis identified six molecularly-distinct sub-clusters of cluster A (A1-A6) and six

177 of cluster B (B1-B6) (Fig. 3A). Four clusters displayed established b-cell features: B1, B2, A3,

178 and A5. Specifically, B1 contained the healthiest b-cells, as indicated by the highest level of INS

179 expression and b-cell-specific TF activity, as well as by inactivity of proteins associated with

180 metabolic inflexibility/stress response, as well as progenitor/stem cell and a-cell identity; B2 had

181 similar, yet less pronounced b-cell-specific TF features. In contrast, A3 and A5, while retaining

182 b-cell features, also showed activation of proteins associated with metabolic inflexibility/stress

183 response and endocrine progenitor-like cells, which were more pronounced in A3 than A5. The

184 latter phenotype strikingly recapitulates the response documented in diabetic rodent islets, as

6 185 discussed below (Fig. 3A) .

8 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

186 In contrast, eight clusters showed clear a-cell-like features: A6 showed the strongest a-

187 cell-like identity, with high GCG expression and IRX2 activity, modest activation of proteins

188 associated with metabolic inflexibility/stress response, and virtually no b-cell-like features. While

189 retaining a-cell-like features, A4 and A2 presented strong activation of proteins associated with

190 metabolic inflexibility/stress response and endocrine-progenitor identity and, in cluster A2, TFs

191 associated with stem-cell-like identity. A1 was similar to A2, but showed a mixed pattern of INS

192 and GCG expression (Fig. 3A). Finally, B3-B6 showed weaker a-cell identity, characterized by

193 GCG expression but low IRX2 and metabolic inflexibility/stress response activity. When the two

194 main a-cell-like sub-clusters are compared (A2, A4, A6 vs. B3-6), strong IRX2 activity was

195 significantly associated with activation of metabolic inflexibility/stress response proteins

196 (p = 1.5E-4, by Chi-square test).

197

198 T2D-enriched sub-clusters

199 Next, we asked how disease status co-segregated with the 12 molecularly-distinct clusters

200 identified by our analysis. t-SNE plots revealed highly distinct cell distribution patterns between

201 ND and T2D islets, with additional cell sub-populations and cluster representation differences in

202 the latter (Fig. 3C and 3D). We excluded batch effects as potential confounding factors, since all

203 T2D donors share consistent t-SNE patterns that are distinct from t-SNE patterns of individual

204 ND donors (Supplementary Fig. 2C).

205 We thus proceeded to quantitatively assess differential representation and associated

206 statistical significance of ND- and T2D-derived cells across the 12 clusters (Fig. 3E and 3F).

207 Four clusters were significantly enriched in ND islets: B1, B3, A2, and A4. B1 (p = 3.4E-8)

208 corresponds to the healthiest b-cells. B3 (p = 8.2E-10) shows an unusual pattern, characterized

209 by both INS and GCG expression, without activation of either b- or a-like transcriptional

9 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

210 programs. We surmise that they are comprised of functionally quiescent cells. A2 (p = 1.9E-3)

211 and A4 (p = 1.0E-2) also displayed a-like features and activation of the metabolic

212 inflexibility/stress response. A2 shows progenitor/stem-cell-like features. Overall, ND islets

213 display a greater share of healthy b-cells, as well as cells that combine transitional b/a,

214 metabolic inflexibility/stress response and progenitor/stem cell-like features. Total cell counts for

215 each cluster in each donor are shown in Fig. 3F.

216 Four clusters were significantly enriched in T2D islets: A3 and B4-6. A3 (p = 9.6E-13) is

217 characterized by b-cell, metabolic inflexibility/stress response, and progenitor/stem-cell-like

218 features. It is interesting to compare this cluster with the “less healthy” b-cell cluster, B2, which

219 doesn’t show disease-state enrichment. In the latter, MAFA activity levels are low but proteins

220 associated with metabolic inflexibility/ stress response are not activated. In contrast, A3 shows

221 higher MAFA activity and activation of metabolic inflexibility/stress response proteins. This is

222 reminiscent of a model in which, in response to declining b-cell function (i.e., lower MAFA

30 223 activity), FOXO1 is activated to reboot MAFA . However, this also heralds the onset of

6 4 224 metabolic inflexibility (PPARa/g) , as well as a drift towards dedifferentiation , as indicated by

31 225 the activation of RFX6 .

226 Clusters B4-6 (p = 7.9E-71, 2.2E-20, and 1.0E-80) are the most enriched in T2D-related

227 cells, and are rarely found in ND islet cells (Fig. 3E and 3F). Unlike the neighboring, ND-

228 enriched cluster B3, these clusters display significant IRX2 and (in B5 and B6) HES1 activity, an

3 229 inhibitor of the endocrine differentiation program . These findings suggest that diabetic islets are

230 enriched in cells that combine weak a-like features with a muted stress response. These may

231 correspond to converted a-like cells, as identified by NKX6.1/ALDH1A3 immunohistochemistry

7 232 of pancreatic islets in diabetic patients .

10 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

233

234 Different driver proteins elicit distinct T2D cell state transitions

235 To identify MR proteins that mechanistically modulate the transcriptional identity of T2D-

236 enriched sub-clusters, we assessed transcriptional regulators whose targets (including activated

237 and suppressed genes) were over- or under-expressed in cluster B6 (the most representative

238 T2D cluster), compared to well-defined b- and a-cells (clusters B1 and A6, respectively). These

239 proteins have the highest metaVIPER-measured differential activity in the two comparisons and

240 are thus inferred as mechanistic determinants of the differential expression signature between

241 these clusters (Supplementary Data 2).

242 Interestingly, we detected several TFs with significantly increased activity in T2D-

243 enriched clusters but no TFs with decreased activity, suggesting that transition to a diabetic

244 phenotype is associated with gain-of-function (Supplementary Data 2). We selected 10

245 candidate MRs for experimental validation based on their activity in T2D-enriched clusters and

40 246 their association with diabetes susceptibility loci in GWAS studies. They included: AFF3 ,

41 42 43 44 45 46 47 48 247 BACH2 , BNC2 , GAS7 , MYT1L , NFATC3 , RFX7 , TSHZ2 , ZRANB3 , and

49 248 ZNF385D . Critically, their mRNA expression was not significantly different based on scRNA-

249 seq analysis, suggesting that these proteins could not have been identified using more

250 traditional approaches (Supplementary Data 3).

251 To functionally interrogate these candidates, we performed gain-of-function studies in

252 primary human ND islets using a modification of Perturb-seq, a CRISPR-Cas9-mediated

50,51 253 scRNA-seq screen , which we termed single-cell gain-of-function sequencing (scGOF-seq)

254 (Fig. 4B). This approach allowed us to evaluate the consequences of the increased activity of a

255 candidate gene at the single-cell level in a physiologically relevant context. We generated

256 bicistronic adenoviral vectors to express candidate genes and a BFP reporter followed by a

257 unique 18-nt barcode with a PolyA signal (Supplementary Fig. 7). The transcribed DNA barcode

11 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

258 maps the targeted gene to an individual cell, allowing its identity and the effect of the gene

259 manipulation to be read out by scRNA-seq and analyzed by metaVIPER. To streamline the

260 experiments, we transduced islets with pooled adenoviruses, each pool containing three

261 constructs (See methods). This allowed us to not only simultaneously screen several candidates,

262 but also to test potential epistatic interactions in cells that incorporated more than one barcode.

263 We interrogated candidates for their ability to convert the transcriptional identity of ND

264 islet cells to that of the two most characteristic T2D states: A3 (b-like with metabolic

265 inflexibility/stress response and progenitor/stem-like features) and B6 (putatively converted, a-

266 like) (B6-like signature in Fig. 4). As an integrated measure of the A3 sub-cluster, we used the

267 combined activities of RFX6, RFX7, FOXO1, PPARa, PPARg, RB1, POUF51, NANOG and

268 MYCL, referred to as DeDi signature. As a measure of B6, we used the combined activities of

269 IRX2, ZNHIT1, ZFPL1, PAX6 and DRAP1, referred to as B6-like signature (Fig. 4). We

270 monitored preservation of b-cell identity throughout the procedure by comparing with non-

271 transduced and BFP-transduced cells (black and gray plots, respectively). The data are

272 presented as violin plots illustrating the shift from the basal (non-transduced or BFP-transduced)

273 to the scGOF-seq cell population, along with quantitative bar graphs (Fig. 4 and Supplementary

274 Fig. 7). In some experiments, the number of cells transduced with individual TFs was too low to

275 allow independent evaluation of each individual factor, due to the stringent cutoff imposed for

276 barcode calling required to remove cells with low candidate-gene expression.

277 The strongest effects to drive the DeDi signature were seen in cells transduced with the

278 BACH2/TSZH2 or RARB/GAS7/ZNF385D combinations, as well as ZRANB3 (Fig. 4A and 4B

279 and Supplementary Fig. 8). Next, we asked which TFs can drive conversion to a-like-cells (B6-

280 like). AFF3 showed the strongest effect, followed by the BACH2/TSHZ2 combination and

281 ZRANB3 (Fig. 4C and 4D). To understand this latter effect in more detail, we did a sub-analysis

12 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

282 asking whether these TFs affected directly the IRX2 network, and found that indeed AFF3

283 directly induced IRX2 activity (Fig. 4E and 4F and Supplementary Fig. 8). These data suggest

284 that an active transition to an a-like cell state is characteristic of T2D.

285 To determine the mechanism of the observed fate transition to a DeDi signature, we

286 analyzed TFs whose activity changed upon conversion. We found that BACH2, FOXO1, MYTL1,

287 NFATC3, RFX7 and TCF4 were co-regulated in all conditions conducive to increasing the DeDi

288 signature (Supplementary Fig. 9). While FOXO1 and TCF4 ectopic expression didn’t drive the

289 DeDi signature, their activity significantly increased upon BACH2 or ZRANB3-dependent

290 conversion, suggesting that these six TFs form a hierarchical module driving conversion. This

291 observation is consistent with RNA-seq analyses of b-cell-specific FoxO1 knockout mice,

5,6 292 showing impaired levels of Bach2 mRNA .

293 In a replication experiment using islets from a different donor, we confirmed that gain-of-

294 function of BACH2/NFATC3/TSHZ2, or CUX2/RFX7 increased DeDi signature cells, as did co-

295 expression of ZRANB3/BNC2/MYT1L (Supplementary Fig. 10).

296

297 Functional effect of master regulators on fate-converted cells

298 Finally, we tested whether converting ND b-cells to a T2D-enriched signature impairs b-cell

2+ 299 function, as assessed by measuring intracellular Ca flux in response to glucose at the single-

52 300 cell level. We selected BACH2 (due to its link to diabetes susceptibility ), AFF3 (due to the

301 strong effect on a-like-cell conversion), and TCF4 (as a negative control) for these experiments.

302 To gate b-cells, we co-transduced ND islets with RIP-zsGreen together with adenovirus

53 303 expressing each candidate TF . After three days, we partially dissociated and plated islets for

2+ 304 Ca imaging. We individually gated zsGreen/BFP-double-positive cells to monitor intracellular

2+ 2+ 305 Ca concentration based on the Ca indicator, Rhod-2 signal (Fig. 5A). b-cells transduced with

13 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

2+ 306 Tag-BFP showed a brisk Ca response to glucose or KCl (Fig. 5B). The glucose response was

307 blunted in cells transduced with AFF3 or BACH2, while the KCl response was intact, indicating

308 that AFF3 or BACH2 gain-of-function impairs b-cell glucose sensing but does not alter

309 membrane depolarization (Fig. 5C and D). This is consistent with the loss of b-cell features

310 observed in scGOF-seq experiments with AFF3 or BACH2 (Fig. 4D). In contrast, TCF4 gain-of-

2+ 311 function had no statistical effect on Ca influx compared to Tag-BFP (Fig. 5E), consistent with

312 the scGOF-seq results, showing that TCF4 failed to affect b-cell identity. These data further

313 validate our approach to functional testing and strengthen the observations on AFF3 and

314 BACH2.

315

14 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

316 DISCUSSION

12 317 The clinical progression of T2D is characterized by a rapid degradation of b-cell function that

318 becomes refractory to treatment over time, requiring combination therapy and resulting in life-

319 threatening complications. Glp1-based agents have b-cell protective effects, but don’t appear to

54 320 significantly delay monotherapy failure, or reverse established b-cell failure . Our work aims to

321 identify mechanistic networks capable to reverse b-cell dedifferentiation as a disease-modifying

14 322 approach .

323 Unbiased analysis of cell transition states identified three salient features of islet cell

324 populations: (a) metabolic inflexibility/stress response (FOXO1 and PPARa/g); (b) endocrine

325 progenitor (RFX6/7) and stem-cell-like features (NANOG, L-MYC expression); and (c) a-cell-like

326 features (GCG expression, activation of IRX2 and related regulators). Within this distribution,

327 healthy b-cells (sub-cluster B1) are enriched in ND subjects, while b-cells with activated

328 metabolic inflexibility/stress response, as well as progenitor/stem cell features are enriched in

329 T2D (sub-cluster A3). In the latter sub-cluster, activation of the two nuclear hormone receptors

330 PPARa and g may appear counterintuitive, since they oversee different lipid metabolic functions,

331 but is in fact consistent with the observation that, as islets fail, synthetic and oxidative branches

6 332 of lipid metabolism are coactivated . These data provide a potential explanation for the

333 longstanding clinical observation that PPARg agonists have beneficial effects on diabetes

55 334 prevention and b-cell function . While the latter have traditionally been ascribed to improved

335 insulin sensitivity, the present data are suggestive of direct effects on b-cells.

336 A striking feature of the MR analysis is the existence of intermediate states providing

337 continuity between the main b- and a-cell identities. This can be reconciled with the observation

56 338 that the latter can be reprogrammed to serve as surrogate b-cells , as well as with the overall

339 plasticity of endocrine islet cell fate. It is striking that T2D islets are enriched in “stressed” b-cells

15 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

340 (i.e., b-cells with markers of metabolic inflexibility/stress response/progenitor/stemness),

341 whereas ND islets are enriched in “stressed” a-like-cells. In this regard, there appears to be a

342 direct correlation between GCG mRNA, activation of the IRX2 network, and metabolic

343 inflexibility/stress response. Cells in which these features are more pronounced prevail in ND

344 islets, whereas cells with more muted a-like characteristics are enriched in T2D, somewhat

7 345 reminiscent of mixed-feature cells identified previously . What can this mean? One potential

346 interpretation is that a-like cells displaying a more robust stress response are more resilient,

347 hence capable of reversal to b-cells (if indeed this is their origin), whereas those lacking a stress

348 response are transitioning toward a functionally quiescent state.

349 The biological relevance of the hierarchical clustering is supported by functional studies

350 in primary islets, showing that imputed regulatory networks can drive different aspects of the

351 cellular pathology. In this regard, our data favor a network model in which a small subset of TFs

352 can drive different aspects of the diabetic islet pathology. Thus, BACH2/TSHZ2 strongly affect

353 dedifferentiation, whereas AFF3 has the strongest effect on driving the a-like-cell phenotype.

354 This analysis is strengthened by the functional data on calcium fluxes, indicating that gain-of-

355 function of BACH2 or AFF3 does indeed impair the features of a differentiated b-cell.

356 More is known about the function of BACH2 than that of AFF3 in islets. Traditionally

57 357 viewed as a repressor of lymphocytes lineages , BACH2 has been linked to genetic

52 358 susceptibility to type 1 diabetes , and changes to its chromatin structure have been reported in

58 359 human diabetic islets . In addition, it’s regulated by the Akt/mTOR pathway, has been shown to

57 360 dimerize with MAF proteins , and has emerged as a candidate from a previous analysis of b-

5 361 cell dedifferentiation , a role confirmed by the present findings. AFF3 has been known to serve

362 as a scaffold for the transcription elongation factor (P-TEFb)-containing super elongation

59 363 complex (SEC), which regulates rapid gene activation upon exposure to environmental stimuli .

16 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

364 It has also been linked to enhancer activation and expression of long noncoding RNA and

60 61 365 miRNA genes from the active allele , as well as to neural development .

366 Previous studies using scRNA-seq have been only partly successful to characterize

15-21 367 distinct cell sub-groups in T2D , potentially due to low sample size and sequencing depth, as

368 well as untested functional consequences of altered gene expression profiles among different

369 cell sub-populations. In this work, we leveraged the scRNA-seq procedure in two ways. First, we

370 generated and interrogated islet-specific transcriptome regulatory networks using ARACNe and

371 metaVIPER. Moreover, we experimentally validated the systems-level approach by a contextual

372 screen in b-cells. As next steps, we plan to test whether T2D-enriched sub-populations can be

373 reconverted to healthy b-cells, and use the OncoTreat method to identify chemical modulators

62 374 of specific MR activity signatures . Ultimately, this work will expand our knowledge of T2D

375 pathophysiology and pave the way to the identification of treatments to benefit T2D patients.

376

377

17 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

378 METHODS

379 Human islet cell culture

380 Human ND and T2D islets were obtained from the National Institutes of Health’s Integrated Islet

381 Distribution Program (IIDP). Upon arrival, human islets were plated at a density of 10,000 IEQ

382 per 10-cm non-treated tissue culture dish (Corning, Corning, NY; cat. no. 430591) into 10 ml of

383 islet culture medium (Prodo Labs, PIM(S)®, cat#PIM-CS001GMP), supplemented with 5 ml

384 PIM(G)® Glutamine/Glutathione (Prodo Labs, cat#PIM-G001GMP), and 5% PIM(ABS)® Human

385 AB Serum (Prodo Labs, cat#PIM-ABS001GMP), along with triple antibiotics, PIM(3X) ®, which

386 includes Ciprofloxacin (Ref 61-277-RF, 10mg/1000ml), Gentamycin (Sigma, G1272,

387 10mg/1000ml), and Amphotericin B (Omega, FG-70, 2500mcgm/1000ml). Islets were cultured

388 for no longer than one week after arrival and medium was replaced every 2 days.

389

390 Fluorescence-Activated b-Cell Sorting

391 On the day of scRNA sequencing, islets were collected, washed in phosphate-buffered saline

392 (PBS) once, and dispersed into single cells by mechanical shaking at 37°C using 0.05% trypsin

393 (Gibco, 25300054). Dispersed islet cells were incubated for 1 hr. at 37°C in MEM (Gibco,

394 11090081) containing 1% BSA, then collected into 5 mL round-bottom polystyrene tubes with a

395 cell strainer (BD Falcon, 352235) at a final SYTOX Red concentration 5 nM (ThermoFisher,

396 S34859). High- and low-fluorescence cells were sorted on a fluorescence-activated Influx cell

397 sorter (BD Influx system). Flavin adenine dinucleotide (FAD) content of cells was analyzed at an

398 excitation wave length of 488 nm, and collected at 525 nm. For scGOF-seq, adenovirus-

399 transduced cells were further gated for PacBlue-positive cells at an excitation wave length of

400 401 nm, and collected at 452 nm.

401

402 RNA-seq library preparation using the Fluidigm C1 800 platform

18 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

403 Sorted cells with high or low/medium FAD content were suspended in C1 Cell Suspension

404 Reagent (Fluidigm) and loaded onto each inlet of the C1 high-throughput integrated fluidic

405 circuit (HT IFC). The number of cells captured in each chamber was visualized and noted using

406 a phase contrast microscope. Only chambers with single-cell capture were used for analysis.

407 Cells were lysed, and mRNA reverse transcribed and PCR-amplified using C1 Single-cell Auto

408 Prep IFC (Fluidigm, protocol 101–4964). The quality and yield of cDNA were determined by

409 Agilent Bioanalyzer using Agilent High Sensitivity DNA Chip. Libraries for sequencing were

410 prepared using Nextera XT DNA library preparation kit (Illumina FC-131-1096) and sequenced

411 with paired-end 50 cycles on Illumina HiSeq2500. Each library pool was subjected to 2 lanes of

412 Illumina HiSeq2500.

413

414 10x Genomics platform RNA-seq library preparation

415 Sorted cells were treated with a Chromium Single Cell 3' Library and Gel Bead Kit V2 (PN-

416 120237), Chromium Single Cell 3' Chip Kit V2 (PN-120236) and Chromium i7 Multiplex Kit (PN-

417 120262) and analyzed with a 10x Genomics Chromium for Single-Cell Library Preparation

418 Instrument, per the manufacturer's specifications and sequenced paired-end 150 bp on HiSeq

419 4,000 to a depth of 90,000 UMI per cell. UMI counts for each cellular barcode were quantified

420 and used to estimate the number of cells successfully captured and sequenced. The Cell

421 Ranger Single-Cell Software suite was used for demultiplexing, barcode processing, alignment,

422 and initial clustering of the raw scRNA-seq profiles.

423 We used the Chromium instrument and the Single Cell 3’ Reagent kit (V1) to prepare

424 individually barcoded single-cell RNA-Seq libraries following the manufacturer’s protocol (10X

425 Genomics). For QC and to quantify the library concentration we used both the BioAnalyzer

426 (Agilent BioAnalyzer High Sensitivity Kit) and qPCR (Kapa Quantification kit for Illumina

427 Libraries). Sequencing with dual indexing was conducted on an Illumina NextSeq machine using

19 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

428 the 150 cycles High Output kit. Sample demultiplexing, barcode processing, and single-cell 3’

429 gene counting was performed with the Cell Ranger Single Cell Software Suite CR2.0.1. Each

430 droplet partition’s contents were tagged with a unique molecule identifier (UMI) – a barcode

431 encoded as the 2nd read of each sequenced read-pair. We followed the 10X Single Cell 3’

432 Reagent Kits v2 protocol as written, using 12 cycles for cDNA amplification and 12 cycles for

433 sample index PCR. Samples were sequenced to a depth of ~400M reads per sample on a

434 NovaSeq 6000 (R1 = 26bp, R(i) = 8bp, R2 = 91bp).

435

436 Plasmids

437 We synthesized open reading frames of each scGOF-seq candidate with a Tag-BFP and an 18-

438 nt unique barcode (Supplemental Table 2) (Qinglan Biotech) and cloned them into the

439 pENTR2b vector using KpnI and EcoRV (AFF3, CUX2, FOXO1, GAS7, TSHZ2 and ZFN385D),

440 BamHI and NotI (BACH2, BNC2, EBF1, RARB, RFX7 and TCF4), SalI and NotI (MYT1L and

441 NFATC3), or KasI and NotI (ZRANB3).

442

443 Adenovirus generation

444 Recombinant adenoviruses were generated using the pAd/CMV/V5-DEST Gateway

445 recombination system (Life Technologies) after cloning the full-length cDNA into the pENTR

446 vector. Individual adenoviruses were packaged and amplified in HEK-293A cells, then pools of

447 three P1 adenoviruses as detailed below were expanded into high-titer virus. Pool1: ZNF385D,

448 RARB, GAS7; Pool2: EBF1, FOXO1, TCF4; Pool3: BACH2, TSHZ2, NFATC3; Pool4: ZRANB3,

449 BNYC2, MYT1L, and Pool5: AFF3, RFX7, CUX2. Adenoviruses were purified by PD-10 column

450 (17085101, GE Healthcare). Titers were determined by plaque assay (PFU). Each virus pool

451 was transduced into HCT116 cells and expression was analyzed by qPCR. FOXO1, TCF4,

452 NFACT3, RFX7 and AFF3 were amplified individually from P1 adenovirus.

20 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

453

454 Adenovirus transduction

455 200-300 human ND islets were handpicked for each condition of transduction and placed on 5

456 mL round-bottom polystyrene test tube. Thereafter, islets were washed and incubated with 100

457 ul serum-free islet culture medium containing 1mM EGTA, and transduced at an MOI 20. After

458 5- to 6-hr transduction, 1ml of complete islet culture medium with 5% PIM(ABS)® Human AB

459 Serum (Prodo Labs, cat#PIM-ABS001GMP) was added overnight. Islets were then transferred

460 to 60 mm non-treated tissue culture dishes (Fisher Scientific FB0875713A), and medium was

461 replaced with fresh islet culture medium every 2 days for seven days for scGOF-seq

462 experiments, and three days for single-cell intracellular calcium microfluimetry.

463

464 Single-Cell Intracellular Calcium Microfluimetry

465 Similar-size human islets from non-diabetic donors were handpicked and transduced with

466 adenovirus expressing each candidate cDNA. The day after transduction, islets were partially

467 dispersed using 0.05% trypsin for 5 min at 37°C, then plated on 35mm glass bottom dishes with

468 10mm microwells (In vitro Scientific D35-10-0-N) pre-coated with fibronectin (Sigma, F1141).

469 Cells were washed with islet media and allowed to rest for two additional days. On the third day,

470 each plate was washed with KRBH buffer and incubated with 2.8 Mm glucose containing KRBH

471 buffer for 30 min, then loaded in the dark with 5 µM Rhod-2, AM (Thermo Fisher R1244) in

472 KRBH buffer. Cells were washed and transferred into a perifusion chamber placed in the light

473 path of a Zeiss Axiovert fluorescence microscope (Zeiss, USA), and perifused with low glucose

474 (2.8 mM), high glucose (16.8 mM), or KCl (40 mM) in KRBH buffer. b-cells were excited by a

475 Lambda DG-4 150 Watt xenon light source (Sutter, Novato, USA), using alternating

476 wavelengths of 340 and 380 nm at 0.5 s intervals, and imaged at 510 nm. For each data set,

21 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

477 regions of interest corresponding to the locations of 10-20 individual cells were selected and

478 images were recorded using an AxioCam camera controlled by Stallion SB.4.1.0 PC software

479 (Intelligent Imaging Innovations, USA). Single-cell intracellular Ca2+ mobilization data are

480 plotted as a function of time.

481

482 Quantitative analyses of single-cell gene expression

483 For single-cell RNA-Seq using Fluidigm C1 system, after demultiplexing, the resulting raw reads

484 were aligned to hg19 reference index by Bowtie2-2.2.6. Aligned reads were sorted and indexed

485 by samtools-1.2. Counts matrices were quantified with R package GenomicFeatures-

486 1.24.5GenomicAlignments-1.8.4 and TxDb.Hsapiens. UCSC. hg19.knownGene-3.2.2 from

487 Bioconductor. Counts matrices were then converted to log2(reads per million reads + 1)

27 488 matrices as the final gene expression quantification .

489 For ScGOF-seq using 10x Genomics Chromium, UMI (Unique Molecular Identifier,

490 barcodes each mRNA transcript) matrices were quantified using scPLATE-Seq pipeline, using

491 parameters –umiLen 10 and –barcodeLen 16 considering the UMI and cell barcode length of

492 Chromium, with customized STAR 2.5.1a reference index with parameter --

493 genomeSAindexNbases 7 considering the length of cDNA and sgRNA sequences, including

494 regular UCSC hg38 reference index and cDNA+sgRNA reference index (one cDNA/sgRNA

495 sequence per artificial chromosome). UMI matrices were then converted to log2(transcripts per

27 496 million transcripts + 1) matrices as the final gene expression quantification .

497

498 Regulatory networks and transcriptional regulator activity analysis

499 Islet-specific regulatory network generation. Donor-specific regulatory networks were reverse

24 500 engineered by ARACNe . ARACNe was run with 200 bootstrap iterations using 1,813

501 transcription factors (genes annotated in Gene Ontology molecular function database, as

22 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

502 GO:0003700, 'transcription factor activity', or as GO:0003677, 'DNA binding', and GO:0030528,

503 'transcription regulator activity', or as GO:00034677 and GO: 0045449, 'regulation of

504 transcription'), 969 transcriptional cofactors (a manually curated list, not overlapping with the

505 transcription factor list, built upon genes annotated as GO:0003712, 'transcription cofactor

506 activity', or GO:0030528 or GO:0045449) and 3,370 signaling pathway related genes

507 (annotated in GO Biological Process database as GO:0007165 'signal transduction' and in GO

508 cellular component database as GO:0005622, 'intracellular', or GO:0005886, 'plasma

509 membrane'). Parameters were set to 0 DPI (Data Processing Inequality) tolerance and MI

510 (Mutual Information) p-value (using MI computed by permuting the original dataset as null model)

511 threshold of 10-8. Transcriptional regulator activity profiles were inferred from combined

512 expression matrix (from all 11 donor-specific matrices) by integrating all 11 donor-specific

27 513 regulatory networks with the metaVIPER algorithm .

514 Dimension reduction, clustering and pseudo-lineage analysis. Dimension reduction was done

515 using both gene expression and metaVIPER-inferred activity profiles. For gene expression, cells

516 were first projected to principal component space using PCA (Principal Component Analysis),

517 further projected to 2-D t-SNE space. T-SNE function in CRAN R package t-SNE-0.1.3 using 1-

518 r(cell-wise Pearson on principal component space) as distance matrix. For metaVIPER-inferred

519 activity, cells were projected to 2-D t-SNE space. T-SNE function in CRAN R package t-SNE-

25 520 0.1.3 using cell-wise activity dissimilarity as distance matrix .

36 521 Clustering analysis was done using iterClust iterative clustering analysis framework .

522 iterClust function in Bioconductor R package iterClust-1.0.2 at metaVIPER-inferred activity level.

523 In total three iterations were done, separating cell populations at different metabolic stress

524 states and cell types sequentially. Using scRNA-seq data from cells with more than 0.1 million

525 mapped reads, we first projected single cells onto on 2D-space with t-SNE.

526

23 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

527 Data availability

528 Single-cell RNA-Seq data using Fluidigm C1 system for all donors have been deposited at the

529 Gene Expression Omnibus (GEO) under accession number GSE98887. ScGOF-seq data using

530 10x Genomics Chromium platform have been submitted at the GEO.

531

24 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

532 Author contributions

533 J.S and H.D. designed and performed experiments, analyzed data, and wrote the manuscript.

534 A.C and D.A. designed experiments, oversaw research, and wrote the manuscript.

535

536 ACKNOWLEDGEMENTS

537 We thank the Genome Technology Center at NYU and Sulzberger Genome Center at Columbia

538 for help with scRNA-Sequencing, the Human Islet and Adenovirus Core of the AECOM/MSSM

539 DRC for help in generating the adenoviruses. FACS experiments were performed in the DRC

540 Flow Cytometry Core (S10OD020056). The content is solely the responsibility of the authors

541 and does not necessarily represent the official views of the National Institutes of Health. This

542 work was supported by NIH grants DK64819, DK63608 (Columbia University Diabetes

543 Research Center). J.S. was supported by Kirschstein-NRSA postdoctoral fellowship

544 F32DK117574. We are grateful to members of the Accili and Califano laboratories for insightful

545 discussions of the data. We thank Travis Morgenstern and Dr. Henry Colecraft (Columbia

546 University) for training and equipment for calcium flux imaging. We also thank to Drs. Andrew

547 Stewart, Adolfo Garcia-Ocaña and Peng Wang for providing RIP-zsGreen adenovirus.

548

549 CONFLICT OF INTEREST

550 A.C. is founder, director, SAB member and shareholder of DarwinHealth Inc. a Company that

551 has licensed the VIPER software. Other authors have declared that no conflict of interest exists.

552 Columbia University is also a shareholder of DarwinHealth Inc.

553

554

555

556

25 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

557 REFERENCES

558 1 Accili, D. Insulin Action Research and the Future of Diabetes Treatment: The 2017

559 Banting Medal for Scientific Achievement Lecture. Diabetes 67, 1701-1709,

560 doi:10.2337/dbi18-0025 (2018).

561 2 Gromada, J., Chabosseau, P. & Rutter, G. A. The alpha-cell in diabetes mellitus. Nature

562 reviews. Endocrinology 14, 694-704, doi:10.1038/s41574-018-0097-y (2018).

563 3 Halban, P. A. et al. beta-cell failure in type 2 diabetes: postulated mechanisms and

564 prospects for prevention and treatment. Diabetes Care 37, 1751-1758,

565 doi:10.2337/dc14-0396 (2014).

566 4 Talchai, C., Xuan, S., Lin, H. V., Sussel, L. & Accili, D. Pancreatic beta cell

567 dedifferentiation as a mechanism of diabetic beta cell failure. Cell 150, 1223-1234,

568 doi:10.1016/j.cell.2012.07.029 (2012).

569 5 Kim-Muller, J. Y. et al. Aldehyde dehydrogenase 1a3 defines a subset of failing

570 pancreatic beta cells in diabetic mice. Nature communications 7, 12631,

571 doi:10.1038/ncomms12631 (2016).

572 6 Kim-Muller, J. Y. et al. Metabolic inflexibility impairs insulin secretion and results in

573 MODY-like diabetes in triple FoxO-deficient mice. Cell Metab 20, 593-602,

574 doi:10.1016/j.cmet.2014.08.012 (2014).

575 7 Cinti, F. et al. Evidence of beta-Cell Dedifferentiation in Human Type 2 Diabetes. J Clin

576 Endocrinol Metab 101, 1044-1054, doi:10.1210/jc.2015-2860 (2016).

577 8 Sun, J. et al. Beta cell dedifferentiation in T2D patients with adequate glucose control

578 and non-diabetic chronic pancreatitis. J Clin Endocrinol Metab, doi:10.1210/jc.2018-

579 00968 (2018).

580 9 Butler, A. E. et al. beta-Cell Deficit in Obese Type 2 Diabetes, a Minor Role of beta-Cell

581 Dedifferentiation and Degranulation. J Clin Endocrinol Metab 101, 523-532,

582 doi:10.1210/jc.2015-3566 (2016).

26 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

583 10 Rudenski, A. S. et al. Natural history of pancreatic islet B-cell function in type 2 diabetes

584 mellitus studied over six years by homeostasis model assessment. Diabetic medicine : a

585 journal of the British Diabetic Association 5, 36-41 (1988).

586 11 Dor, Y., Brown, J., Martinez, O. I. & Melton, D. A. Adult pancreatic beta-cells are formed

587 by self-duplication rather than stem-cell differentiation. Nature 429, 41-46 (2004).

588 12 Weyer, C., Bogardus, C., Mott, D. M. & Pratley, R. E. The natural history of insulin

589 secretory dysfunction and insulin resistance in the pathogenesis of type 2 diabetes

590 mellitus. J Clin Invest 104, 787-794 (1999).

591 13 Savage, P. J. et al. Diet-induced improvement of abnormalities in insulin and glucagon

592 secretion and in insulin receptor binding in diabetes mellitus. J Clin Endocrinol Metab 48,

593 999-1007 (1979).

594 14 Taylor, R., Al-Mrabeh, A. & Sattar, N. Understanding the mechanisms of reversal of type

595 2 diabetes. Lancet Diabetes Endocrinol, doi:10.1016/S2213-8587(19)30076-2 (2019).

596 15 Xin, Y. et al. RNA Sequencing of Single Human Islet Cells Reveals Type 2 Diabetes

597 Genes. Cell Metab, doi:10.1016/j.cmet.2016.08.018 (2016).

598 16 Dorrell, C. et al. Human islets contain four distinct subtypes of beta cells. Nature

599 communications 7, 11756, doi:10.1038/ncomms11756 (2016).

600 17 Dorajoo, R. et al. Single-cell transcriptomics of East-Asian pancreatic islets cells.

601 Scientific reports 7, 5024, doi:10.1038/s41598-017-05266-4 (2017).

602 18 Baron, M. et al. A Single-Cell Transcriptomic Map of the Human and Mouse Pancreas

603 Reveals Inter- and Intra-cell Population Structure. Cell Syst 3, 346-360 e344,

604 doi:10.1016/j.cels.2016.08.011 (2016).

605 19 Li, J. et al. Single-cell transcriptomes reveal characteristic features of human pancreatic

606 islet cell types. EMBO Rep 17, 178-187, doi:10.15252/embr.201540946 (2016).

27 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

607 20 Lawlor, N. et al. Single-cell transcriptomes identify human islet cell signatures and reveal

608 cell-type-specific expression changes in type 2 diabetes. Genome Res 27, 208-222,

609 doi:10.1101/gr.212720.116 (2017).

610 21 Segerstolpe, A. et al. Single-Cell Transcriptome Profiling of Human Pancreatic Islets in

611 Health and Type 2 Diabetes. Cell Metab 24, 593-607, doi:10.1016/j.cmet.2016.08.020

612 (2016).

613 22 Wang, Y. J. et al. Single cell transcriptomics of the human endocrine pancreas.

614 Diabetes, doi:10.2337/db16-0405 (2016).

615 23 Wang, Y. J. & Kaestner, K. H. Single-Cell RNA-Seq of the Pancreatic Islets--a Promise

616 Not yet Fulfilled? Cell Metab 29, 539-544, doi:10.1016/j.cmet.2018.11.016 (2019).

617 24 Basso, K. et al. Reverse engineering of regulatory networks in human B cells. Nat Genet

618 37, 382-390, doi:10.1038/ng1532 (2005).

619 25 Alvarez, M. J. et al. Functional characterization of somatic mutations in cancer using

620 network-based inference of protein activity. Nat Genet 48, 838-847, doi:10.1038/ng.3593

621 (2016).

622 26 Califano, A. & Alvarez, M. J. The recurrent architecture of tumour initiation, progression

623 and drug sensitivity. Nat Rev Cancer 17, 116-130, doi:10.1038/nrc.2016.124 (2017).

624 27 Ding, H. et al. Quantitative assessment of protein activity in orphan tissues and single

625 cells using the metaVIPER algorithm. Nature communications 9, 1471,

626 doi:10.1038/s41467-018-03843-3 (2018).

627 28 Vaquerizas, J. M., Kummerfeld, S. K., Teichmann, S. A. & Luscombe, N. M. A census of

628 human transcription factors: function, expression and evolution. Nat Rev Genet 10, 252-

629 263, doi:10.1038/nrg2538 (2009).

630 29 Kharchenko, P. V., Silberstein, L. & Scadden, D. T. Bayesian approach to single-cell

631 differential expression analysis. Nat Methods 11, 740-742, doi:10.1038/nmeth.2967

632 (2014).

28 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

633 30 Kitamura, Y. I. et al. FoxO1 protects against pancreatic beta cell failure through NeuroD

634 and MafA induction. Cell Metab 2, 153-163, doi:10.1016/j.cmet.2005.08.004 (2005).

635 31 Smith, S. B. et al. Rfx6 directs islet formation and insulin production in mice and

636 humans. Nature 463, 775-780, doi:10.1038/nature08748 (2010).

637 32 Akiba, K., Ushijima, K., Fukami, M. & Hasegawa, Y. A heterozygous protein-truncating

638 RFX6 variant in a family with childhood-onset, pregnancy-associated and adult-onset

639 diabetes. Diabetic medicine : a journal of the British Diabetic Association,

640 doi:10.1111/dme.13970 (2019).

641 33 Wang, Z., York, N. W., Nichols, C. G. & Remedi, M. S. Pancreatic beta cell

642 dedifferentiation in diabetes and redifferentiation following insulin therapy. Cell Metab 19,

643 872-882, doi:10.1016/j.cmet.2014.03.010 (2014).

644 34 Taylor, B. L., Liu, F. F. & Sander, M. Nkx6.1 is essential for maintaining the functional

645 state of pancreatic beta cells. Cell reports 4, 1262-1275,

646 doi:10.1016/j.celrep.2013.08.010 (2013).

647 35 Puri, S., Akiyama, H. & Hebrok, M. VHL-mediated disruption of Sox9 activity

648 compromises beta-cell identity and results in diabetes mellitus. Genes Dev 27, 2563-

649 2575, doi:10.1101/gad.227785.113 (2013).

650 36 Ding, H., Wang, W. & Califano, A. iterClust: a statistical framework for iterative clustering

651 analysis. Bioinformatics 34, 2865-2866, doi:10.1093/bioinformatics/bty176 (2018).

652 37 Nica, A. C. et al. Cell-type, allelic, and genetic signatures in the human pancreatic beta

653 cell transcriptome. Genome Res 23, 1554-1562, doi:10.1101/gr.150706.112 (2013).

654 38 Benner, C. et al. The transcriptional landscape of mouse beta cells compared to human

655 beta cells reveals notable species differences in long non-coding RNA and protein-

656 coding gene expression. BMC Genomics 15, 620, doi:10.1186/1471-2164-15-620

657 (2014).

29 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

658 39 Guo, S. et al. Inactivation of specific beta cell transcription factors in type 2 diabetes. J

659 Clin Invest 123, 3305–3316, doi:10.1172/JCI65390 (2013).

660 40 Sandholm, N. et al. New susceptibility loci associated with kidney disease in type 1

661 diabetes. PLoS Genet 8, e1002921, doi:10.1371/journal.pgen.1002921 (2012).

662 41 Morris, D. L. et al. Genome-wide association meta-analysis in Chinese and European

663 individuals identifies ten new loci associated with systemic lupus erythematosus. Nat

664 Genet 48, 940-946, doi:10.1038/ng.3603 (2016).

665 42 Paterson, A. D. et al. A genome-wide association study identifies a novel major locus for

666 glycemic control in type 1 diabetes, as measured by both A1C and glucose. Diabetes 59,

667 539-549, doi:10.2337/db09-0653 (2010).

668 43 Yang, X. et al. Validation of candidate causal genes for obesity that affect shared

669 metabolic pathways and networks. Nat Genet 41, 415-423, doi:10.1038/ng.325 (2009).

670 44 Hiramoto, M. et al. A type 2 diabetes-associated SNP in KCNQ1 (rs163184) modulates

671 the binding activity of the locus for Sp3 and Lsd1/Kdm1a, potentially affecting CDKN1C

672 expression. Int J Mol Med 41, 717-728, doi:10.3892/ijmm.2017.3273 (2018).

673 45 Keller, M. P. et al. The Transcription Factor Nfatc2 Regulates beta-Cell Proliferation and

674 Genes Associated with Type 2 Diabetes in Mouse and Human Islets. PLoS Genet 12,

675 e1006466, doi:10.1371/journal.pgen.1006466 (2016).

676 46 Osman, W., Tay, G. K. & Alsafar, H. Multiple genetic variations confer risks for obesity

677 and type 2 diabetes mellitus in arab descendants from UAE. Int J Obes (Lond) 42, 1345-

678 1353, doi:10.1038/s41366-018-0057-6 (2018).

679 47 Huang, Y. C. et al. JPH2 is a novel susceptibility gene on chromosome 20q associated

680 with diabetic retinopathy in a Taiwanese population. Scienceasia 39, 167-173,

681 doi:10.2306/scienceasia1513-1874.2013.39.167 (2013).

682 48 Adeyemo, A. A. et al. ZRANB3 is an African-specific type 2 diabetes locus associated

683 with beta-cell mass and insulin response. Nature Communications 10, doi:ARTN 3195

30 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

684 10.1038/s41467-019-10967-7 (2019).

685 49 Awata, T. et al. A genome-wide association study for diabetic retinopathy in a Japanese

686 population: potential association with a long intergenic non-coding RNA. PLoS One 9,

687 e111715, doi:10.1371/journal.pone.0111715 (2014).

688 50 Adamson, B. et al. A Multiplexed Single-Cell CRISPR Screening Platform Enables

689 Systematic Dissection of the Unfolded Protein Response. Cell 167, 1867-1882 e1821,

690 doi:10.1016/j.cell.2016.11.048 (2016).

691 51 Dixit, A. et al. Perturb-Seq: Dissecting Molecular Circuits with Scalable Single-Cell RNA

692 Profiling of Pooled Genetic Screens. Cell 167, 1853-1866 e1817,

693 doi:10.1016/j.cell.2016.11.038 (2016).

694 52 Onuma, H. et al. Variants in the BACH2 and CLEC16A gene might be associated with

695 susceptibility to insulin-triggered type 1 diabetes. J Diabetes Investig,

696 doi:10.1111/jdi.13057 (2019).

697 53 Kenty, J. H. & Melton, D. A. Testing pancreatic islet function at the single cell level by

698 calcium influx with associated marker expression. PLoS One 10, e0122044,

699 doi:10.1371/journal.pone.0122044 (2015).

700 54 Holst, J. J. From the Incretin Concept and the Discovery of GLP-1 to Today's Diabetes

701 Therapy. Front Endocrinol (Lausanne) 10, 260, doi:10.3389/fendo.2019.00260 (2019).

702 55 Kernan, W. N. et al. Pioglitazone after Ischemic Stroke or Transient Ischemic Attack. N

703 Engl J Med 374, 1321-1331, doi:10.1056/NEJMoa1506930 (2016).

704 56 Furuyama, K. et al. Diabetes relief in mice by glucose-sensing insulin-secreting human

705 alpha-cells. Nature 567, 43-48, doi:10.1038/s41586-019-0942-8 (2019).

706 57 Zhou, Y., Wu, H., Zhao, M., Chang, C. & Lu, Q. The Bach Family of Transcription

707 Factors: A Comprehensive Review. Clin Rev Allergy Immunol 50, 345-356,

708 doi:10.1007/s12016-016-8538-7 (2016).

31 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

709 58 Bysani, M. et al. ATAC-seq reveals alterations in open chromatin in pancreatic islets

710 from subjects with type 2 diabetes. Scientific reports 9, 7785, doi:10.1038/s41598-019-

711 44076-8 (2019).

712 59 Luo, Z., Lin, C. & Shilatifard, A. The super elongation complex (SEC) family in

713 transcriptional control. Nat Rev Mol Cell Biol 13, 543-547, doi:10.1038/nrm3417 (2012).

714 60 Luo, Z. et al. Regulation of the imprinted Dlk1-Dio3 locus by allele-specific enhancer

715 activity. Genes Dev 30, 92-101, doi:10.1101/gad.270413.115 (2016).

716 61 Moore, J. M. et al. Laf4/Aff3, a gene involved in intellectual disability, is required for

717 cellular migration in the mouse cerebral cortex. PLoS One 9, e105933,

718 doi:10.1371/journal.pone.0105933 (2014).

719 62 Alvarez, M. J. et al. A precision oncology approach to the pharmacological targeting of

720 mechanistic dependencies in neuroendocrine tumors. Nat Genet 50, 979-989,

721 doi:10.1038/s41588-018-0138-4 (2018).

722

723

32 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

724 Figure Legends

725

726 Figure 1. metaVIPER analysis reduces donor-to-donor variability (A) Single cells from

727 human ND and T2D islets were projected onto 2-D t-SNE space based on whole mRNA

728 expression. Each dot represents a single cell, color-coded according to donor. (B) t-SNE

729 clustering as in (A) but based on protein activity inferred from islet specific regulatory networks.

730 (C) INSULIN and (D) GLUCAGON mRNA expression plotted in t-SNE at the single cell level. (E)

731 Computationally-inferred MAFA and (F) IRX protein activity plotted in t-SNE at a single-cell level.

732

733 Figure 2. Metabolic inflexibility and stemness markers define two major clusters in ND

734 and T2D islet cells. (A, B) Selected markers of metabolic inflexibility PPARa (A) and PPARg (B)

735 protein activity plotted in t-SNE. (C, D) Endocrine progenitor marker, RFX6 (C) and its cognate

736 factor RFX7 (D) protein activity presented in t-SNE space. (E, F) Stemness markers, NANOG (E)

737 and MYCL (F) protein activity plotted in t-SNE at a single-cell level.

738

739 Figure 3. iTerClust classifies ND and T2D islet cells in different biological states. (A)

740 hclustering using the iterCluster algorithm structures cell positioning and sub-groups. Each sub-

741 group is color-coded. Each bar denotes a single cell. Black bars represent T2D cells and white

742 bars ND cells. SST, GCG and INS mRNA expression is plotted at a single-cell level. (B-D) For

743 illustration purpose, sub-clusters were projected onto 2D t-SNE space according to metaVIPER

744 inference as ND and T2D combined (B), or as ND only (C) and T2D only (D). (E) Bar-plots

745 presenting the percentage of ND or T2D cells in each sub-cluster. A dashed line represents the

746 proportion of cells from ND or T2D islets subjected to scRNA-seq analyses. (F) Bar-plots

747 showing the number of cells from individual donors in each subcluster.

748

33 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

749 Figure 4. scGOF-seq experiments in human islets. (A) Violin plots showing the distribution of

750 cells following transduction with each individual candidate or combination thereof analyzed

751 according to DeDi signature, an intergrated value of RFX6, RFX7, FOXO1, PPARa, PPARg,

752 RB1, POUF51, NANOG and MYCL protein activities. Non-transduced and BFP-transduced ND

753 islets serve as negative controls. (B) Bar-plots showing the proportion of islet cells with a

754 positive DeDi signature (> activity 0) in each scGOF-seq condition. A red dashed line indicates

755 the percentage of islets cells with a positive DeDi signature in non-transduced negative controls.

756 (C) Violin plots showing cells with a B6-like signature, which is an intergrated value of IRX2,

757 ZNHIT1, ZFPL1, PAX6 and DRAP1. (F) Bar-plots showing the proportion of islet cells with a B6-

758 like signature as in (B). (E) Violin plots showing cells with IRX2 activity as in (A). (F) Bar-plots

759 showing the proportion of islet cells with a positive IRX2 activity as in (B).

760

761 Figure 5. Single-b-cell Calcium microfluorimetry. (A) Schematic drawing of the single-cell

2+ 2+ 762 Ca imaging procedure. (B)-(E) Representative traces of Ca flux measured by Rhod2 loading

763 in Ad-BFP (B), Ad-AFF3 (C), Ad-BACH2 or Ad-TCF4 transduced primary human b-cells. Red

764 arrows indicate the timing of addition of 16.8mM glucose, and black arrows indicate addition of

765 40mM KCl.

766

34 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

767 Supplementary Figure Legends

768

769 Supplementary Figure 1. Islet cells sorted by FAD content-based autofluorescence. b-

770 cells are enriched by sorting for increased auto-fluorescence. In non-diabetic samples (ND, left),

771 ~40% islets cells are putative b-cells, whereas in diabetic human islets (T2D, right) only ~20 %

772 are putative b-cells.

773

774 Supplementary Figure 2. T-SNE Density Clustering of human islet cells. T-SNE Density

775 Clustering of human non-diabetic (A) or T2D islets (B) based on metaVIPER-inferred

776 transcriptional regulator activity profiles. (C) Single cells from individual donors were projected

777 onto 2-D t-SNE space based on metaVIPER-inferred transcriptional regulator activity profiles.

778

779 Supplementary Figure 3. mRNA expression pattern in tSNE plots. mRNA expression of

780 MafA (A), Irx2 (B), Rfx6 (C), Rfx7 (D), PPARa (E) or PPARg (F) was visualized as t-SNE plots.

781

782 Supplementary Figure 4. Functional enrichment of GO terms. Data were analyzed with GO

783 term finder using the list of proteins significantly activated in cluster A compared to cluster B in

784 Figure 1. Functional enrichment was summarized using REVIGO. Enriched terms remaining

785 after the redundancy reduction are represented as scatterplots in a two-dimensional space,

786 which summarizes GO terms’ semantic similarities.

787

788 Supplementary Figure 5. Characterization of cells with metabolic inflexibility/stress

789 response and stem-like cell features. MetaVIPER-inferred activity of metabolic inflexibility

790 regulators (A) RB1 and (B) FOXO1, stemness markers (C) FOXM1 and (D) POU5F1 plotted as

35 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

791 t-SNE. Metabolic stress-related transcriptional regulators, including (E) hypoxic stress-related

792 HIF1A, (F) oxidative stress-related TP53, (G) JNK family, (H) NFKB1, (I) HSF1, and (J) ER

793 stress-related XBP1 plotted as t-SNE.

794

795 Supplementary Figure 6. iTerClust classifies human ND or T2D islet cells in different

796 biological states. (A) hclustering using the iterCluster algorithm structures cell positioning and

797 sub-groups of islet cells only from ND donors. Each sub-group was color-coded accordingly.

798 SST, GCG and INS mRNA expression is plotted at a single-cell level. (B) Same as (A) but from

799 T2D donors. (C) Sub-clusters were projected onto 2D t-SNE space according to metaVIPER-

800 inference for ND islets. Gray dots were a overlay of T2D cells (D) Same as (C) but for T2D islets.

801

802 Supplementary Figure 7. Heatmap to denote cell identity after scGOF-seq. (A) Schematic

803 drawing of scGOF-seq plasmids. (B) Cell annotation of scGOF-seq is presented for each group,

804 color-coded for single candidate or combinatorial candidate transduction.

805

806 Supplementary Figure 8. scGOF-seq analyses using ND islets. (A) Bar-plots showing the

807 proportion of islet cells with positive MAFA activity in each condition. A red dashed line indicates

808 the percentage of islets cells with positive MAFA activity in non-transduced negative controls. (B)

809 Same as (A) but for IRX2 activity. (C) Bar-plots showing the proportion of islet cells with a

810 positive DeDi signature (> activity 0) in each condition. (D) Bar-plots showing the proportion of

811 islet cells with a positive B6-like signature (> activity 0) in each condition.

812

813 Supplementary Figure 9. Core driver network of ND cell conversion into DeDi signature

814 cells. Violin plots of core driver network, BACH2, NFATC3, MYT1L and TCF4 for each single

815 candidate or in combination according to DeDiSig score.

36 bioRxiv preprint doi: https://doi.org/10.1101/768135; this version posted September 12, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Son, Ding et al. b-cell dedifferentiation in diabetes

816

817 Supplementary Figure 10. Biological scGOF-seq replicate. (A) Bar-plots showing the

818 proportion of islet cells with positive MAFA activity in each gain-of-function condition. A red

819 dashed line indicates the percentage of islets cells with positive MAFA activity in non-

820 transduced negative control. (B) Same as (A) but for IRX2 activity. (C) Bar-plots showing the

821 proportion of islet cells with a positive DeDi signature (> activity 0) in each condition.

822

823 Supplementary Data 1. List of differentially activated proteins between Cluster A and Cluster B.

824

825 Supplementary Data 2. (A) List of differentially activated TFs between Cluster B1 vs. A3, or A6

826 vs. A3. (B) List of differentially activated TFs between Cluster B1 vs. B6, or A6 vs. B6.

827

828 Supplementary Data 3. List of differentially expressed genes between Cluster B1 vs. A3, A6

829 vs. A3, B1 vs. B6, or A6 vs. B6.

830

831

37 normal.20151026 normal.20160224 normal.20160310 normal.20160427 patient.20160712 patient.20160714 patient.20160202 patient.20160303 patient.20160121 patient.20160314

Exp. Fig.1 A mRNA expression B Protein ActivityAct.

● ● ● ● ● ● ● ●● ● ●●●●● ND1 ● ● ●● ● ● ● ● ●●● ● ● ●● ● ●● ● ● ● ● ●●●●●● ● ● ●● ● ●● ● ●●●●●●● ●●● ● ● ●● ● ● ● ● ● ● ●●● ●● ● ●●●●●●●●● ● ● ● ● ● ● ● ●● ●●●●● ●●● ●● ● ●●●● ● ● ● ● ● ● ●● ●● ● ● ● ● ●●●● ● ●●● ● ● ● ●● ● ●● ● ●● ●● ●● ● ● ● ● ● ● ● ● ●●●● ●● ● ● ● ● ● ●●● ●●●●●● ●●● ● ●● ND2 ● ●● ●● ●● ●● ● ● ● ● ●●●●●●● ● ● ●●● ● ● ●● ●● ● ●● ● ● Cluster B ● ● ●●●●●● ●● ● ● ●● ● ● ● ● ● ● ● ●● ●● ● ● ●●●●●● ● ● ● ● ● ● ● ● ●●●●●● ●●● ● ● ●●●●●●● ● ● ●● ● ●●●●● ●●●●●●● ●● ● ● ● ● ● ● ●●●●●●●●●● ● ● ● ●● ● ● ●● ●●●●●●●● ● ●● ●●● ● ● ● ● ●● ●●●●● ●●●● ●● ●● ● ●● ● ● ●●●●●●●● ●● ●● ●●●●●●●●●● ●● ●● ●●● ● ● ●● ● ●●●● ●● ●●●●●●●●● ● ● ● ● ●● ● ●● ●●●●●● ●●●●●●●● ●●● ● ●● ● ● ● ● ● ●●●●●●● ●●●●●●●●●● ● ● ● ● ● ● ● ● ● ●●●●●●●●●●● ●●●●●● ●●●●●●●●● ● ● ● ●●● ● ● ● ● ●●● ●●●●●●● ●● ● ● ● ● ●● ● ●●●●●●●●●●● ● ●● ND3 ● ●● ● ● ● ●●● ● ●● ● ● ● ● ● ●●● ●● ●●●●●● ● ● ● ● ● ● ●● ● ●●●●● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ●● ●●●●●●●●●●●●●● ● ● ● ● ● ●● ●● ● ●●●● ●● ●●● ● ● ● ● ● ● ●● ● ● ●● ●●● ●●●●● ● ●● ●●●●●●● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ●● ● ●● ●● ●●● ●●●●● ●●●● ●●● ●●● ●●● ● ●● ● ● ● ●● ● ● ● ● ● ●●● ●● ● ●● ● ● ● ●● ● ●●● ● ●●● ●●● ●● ●●●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●●●● ● ●●●●●● ●● ●● ● ●●●● ● ● ● ● ● ● ● ● ●●●● ●●● ● ● ● ● ● ● ● ● ●● ●●● ●● ● ●●● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●●●● ● ● ●●● ● ●● ● ●●●● ●●●●●●●●● ●●● ● ● ●●● ● ● ● ●● ●●● ● ● ● ● ●● ● ●● ● ● ● ● ● ●●●● ●●●●● ● ●●●●● ●●●●●●●● ● ●● ● ● ● ● ND4 ● ● ● ●●● ● ● ●● ●● ● ●●● ● ● ●●● ●●●●●●● ●● ● ● ●● ●● ●●●● ● ●●● ● ● ● ●● ● ● ● ● ●● ● ● ●●● ●●● ● ● ● ● ● ●●●● ●● ●● ●●●●●●●●● ●● ● ● ●● ● ● ● ● ●●● ● ●●● ● ● ● ● ● ●● ●● ●●● ● ● ●●● ●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ● ● ●● ● ●●● ●●● ●● ● ● ● ● ●● ● ● ● ● ●● ●● ●●● ● ●● ●● ● ● ● ●●● ● ●● ●●● ●●●●●● ● ● ● ●● ● ● ●●● ● ● ● ● ●●●●●●●● ● ● ● ●●● ● ● ●● ● ●●●●● ●●●●● ● ● ●●●●● ● ●● ● ●●● ●● ● ●●●● ● ● ● ● ●●● ● ● ● ●●●●●●●● ● ●● ●●● ●● ● ●● ● ● ●●● ●●● ●●●●●●●● ●●● ●● ● ● ● ●●●● ●●●●● ● ●● ●●●●●●●●●●●● ● ● ●●●● ● ● ● ● ● ●●● ●● ●● ● ● ● ● ●●● ●● ● ● ●●●● ●●●●● ● ● ●●●● ●● ● ●● ● ●●●● ●● ● ●● ●●●●● ●● ● ●●●●●●●●●●●● ●●●● ●● ● ● ● ● ● ● ● ●● ●●●●●● ●● ● ● ●●●●●●● ● ● ●● ● ● ●● ● ● ● ● ● ●●● ●●●●●●●●●●● ● ●● ● ●●● ● ●● ●● ●●● ●● ● ● ●●●●● ●● ●● ● ●●● ● ● T2D1 ● ● ●● ●● ● ● ●●● ●● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ●●● ● ●● ●●●●●●● ●●● ●● ● ● ● ●● ● ● ● ●●● ● ● ● ● ● ●●● ●●●●●● ●● ● ●●● ● ● ● ● ● ● ● ● ● ●●●● ● ●●●●●●●● ● ●●●●●● ● ● ● ● ● ●● ● ●● ● ●● ●●●● ● ● ●● ● ● ● ● ●●● ●● ● ● ● ●●● ●● ● ● ● ●●●● ●● ●●●●●●● ●●● ● ● ● ●● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ●● ●●●●●● ●●●● ●●●●●●● ● ● ● ●●●● ●● ●● ●● ● ● ● ●● ●● ● ● ●● ● ● ● ●● ● ● ● ● ● ●● ● ● ●●● ●●● ●●●●●●●● ●●● ● ● ●● ●● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ●●● ● ● ● ● ● ● ● ● ●●●●● ●● ● ●● ●● ● ●●●●●●● ● ●●● ● ● ● ● ● ● ●● ● ● ●● ●● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ●●●● ●● ● ● ●●●● ● ●●● ●●●●●●●●●●●●●●●●●●●● ●●● ●●●●● ●●● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ●● ●● ● ● ●●● ●●●●●●●●● ● ● ● ● ● ●● ●●● ●●●●●●●●● ● ● ●●●●● ● ● ● ● ● ● ●●●● ● ● ● ● ●●● ● ● ● ● ●●●●● ● ●● ● ●●● ●●● ● ● ●● ● ●● ●● ● ●●●●●●● ●●●●●● ●●●●●●●●●● ●● ● ●● ● ●● T2D2 ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ●● ● ● ● ●● ● ●● ●● ●● ●● ●●●●●● ●● ●●●●●● ●●● ●● ●●●● ● ● ● ● ●●● ●● ● ●● ● ●●● ●● ● ●● ● ● ● ●● ● ●●●●● ●●●● ● ● ●●● ●● ●●●●●● ●●●●●●●●●●●●●● ● ●●●● ●●●●●● ●● ●● ● ●●● ●●● ● ● ● ● ●●● ● ● ●● ● ● ● ●●●●●●●● ● ● ●● ● ● ●●●● ●●●●●●●●●●●●●●●●● ●●● ●●●●●● ●●● ●● ● ●●● ● ● ● ● ● ● ● ● ● ●●● ● ● ●● ● ● ● ● ●● ● ● ● ●● ● ● ● ●●●●●●●●●●●● ●●●●● ●● ●●●●●●● ● ● ●●● ● ● ● ● ●●● ● ● ● ● ● ● ●● ●●●● ● ●● ● ● ●● ● ●● ●●●●●●●●● ● ● ●● ●●● ●●● ●●●●●●●●●●●●●● ● ●● ●●●●●●●●●●● ●● ● ●● ●●● ● ● ● ● ● ● ●● ●●● ● ● ●● ●●● ●● ●● ● ● ● ● ●●● ●● ●●●●●●●●●●●●●●●●● ●● ● ●●●●●●●●●●● ● ●● ●●● ●● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ●● ●●●●●●●● ●● ●● ● ● ● ● ●●● ●●● ● ●●● ● ●●●●●● ●●●●●●●● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●●●●●●● ●● ●● ●●●● ●●●● ● ● ● ● ● ●● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●● ● ●● ● ● ● ● ●● ●●● ●●● ● ● ●● ●● ● ●● ●●● ● ●● ● ●● ● ● ● ● ● ●●● ●●●●●●●● ●●●●●●●●●●● ●● ● ●●● ● ● ● T2D3 ● ● ● ● ● ●● ● ● ●● ● ● ●●●● ● ●● ● ●● ●●●●● ● ● ●● ● ● ● ●● ●● ●● ●●●●●●● ●● ●●●●● ●●●●● ● ●● ● ● ● ● ● ●● ● ● ● ●●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ●●●● ●● ●●●●●● ●●●●●●●●●●● ●●●●●●●●● ●●● ●● ● ● ● ● ●● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ●●●● ● ●● ●● ● ● ● ●● ● ●●● ● ● ● ● ●● ● ● ● ● ● ● ●●●●●●●●●●●●●●●● ●●●●●● ●●●●●● ●●●●●●●●● ● ●● ● ●● ● ●● ● ● ● ● ● ● ●●● ● ●●● ● ●● ● ●●● ●●●● ● ● ● ●●●● ● ●●●●●●●●●●●●● ●●●●●●●●● ●● ●● ●● ● ● ●●●●● ● ● ● ●● ● ● ● ● ●● ● ●● ●● ● ●● ● ● ● ● ● ●● ● ● ● ●● ●● ●●●●●●●● ●●●●●●●●●●●●●●●● ●●●●●●● ●● ●● ● ● ● ● ● ●● ● ●● ● ● ●● ● ● ● ● ●●● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●● ● ●●●● ● ● ● ●●●●●●● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ●● ● ● ●●●●● ●● ● ●●●● ● ●●● ●● ●● ● ● ● ●● ●●●● ●● ●● ● ● ● ● ● ● ● ● ● ●●● ● ●●●● ● ●● ●● ●● ● ● ● ● ● ●● ●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●● ● ●● ● ●●●● ● ●●● ●●●● ●● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ● ● ● ●● ●●● ●●● ●● ●●● ●●●●● ●●●●●●● ● ●●● ● T2D4 ● ● ● ● ● ● ● ● ●● ● ●●● ● ●● ● ●● ● ● ●●●●●●●● ● ●●● ●●● ● ● ● ● ● ●●●●●● ● ●● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ●●●● ●●●●● ●●● ●●●●● ● ● ●●●●●●●●● ●● ● ● ●● ● ● ● ● ●●● ● ●● ● ● ● ●●●●● ● ● ● ● ●● ● ●●●● ● ●● ●● ● ● ● ● ● ● ● ● ● ●● ●● ●●●● ●● ●● ●●● ●● ●●●●● ● ●●●● ● ●● ●● ● ●● ●● ● ●●●●●● ● ●● ●●● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ●●●●● ●● ●●● ●●●● ● ●●● ●● ● ● ● ● ●●● ● ● ●● ●● ● ● ● ●● ●●●● ● ● ●● ● ●●● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ●● ● ●● ●●●● ●● ●●●●● ●●●● ● ● ●● 2 ●●●● ●●●●● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ●● ● ● ●●● ●● ●●● ●● ●●●● ●●● ●●● ● ● ● ●●●● ●● ● ● ● ●● ● ●●●● ●● ● ● ● ●● ● ● ●●● ● ● ●●● ● ● ● ● ● ●● ● ●●●● ● ●● ● ●● ●●● ●● ● ●●● ● ● ● ● ●●●● ● ●●● ● ●● ● ● ● ● ● ● ● ● ● ●●● 2 ● ● ● ● ● ●● ●●●● ● ● ●●●●●●● ● ●● ●●● ●●● ●● ●●●● ● ● ●●●● ● ●● ●● ● ●● ● ● ● ●●● ● ● ● ● ● ●● ● ● ● ●●● ● ●●● ●● ●●●●● ● ●●● - ●● ● ●● ●●●● ● ●● ●● ●● ● ● ● ●●● ● ● ●● ●● ●●● ● ● ●● ●● ●● ●●● ●●● ● ● ● ● T2D5 ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●●● ● ●● ● ● ●● ●● ●● ●● ●● ● ● ● ● ●●●●● ●● ●●● ●● ● ● ●● ● ● ● ● ●● ●●●● ●●●● ●●● ●●●● ●●●● ● ●●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● - ● ● ● ● ● ●● ● ●●●●● ●● ● ● ● ● ●●● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●● ●●● ● ● ● ●●●● ● ● ●● ●●●●● ● ● ● ●●●●●●● ● ●●● ● ● ● ● ●●● ●● ●● ● ●● ● ● ● ● ●●● ● ● ●● ● ●● ●●● ● ● ●●● ● ●●●● ● ●● ● ●● ●●●●● ● ●● ● ●●● ● ●●●● ● ●● ● ● ●● ●● ● ● ● ● ●●● ● ●● ●● ●● ●●●● ●● ●● ● ● ● ● ●●● ● ● ● ●●●● ● ● ●● ● ● ●●● ● ●●●● ● ● ● ●●● ● ● ● ● ●● ● ● ●●● ● ●●● ●●●● ●●●● ●●●●● ●●● ●●● ● ● ●● ● ● ●● ● ● ●● ● ●● ●●●● ● ● ●●●●● ●● ● ● ● ● ● ●● ●● ● ● ● ●●● ● ● ● ●● ● ● ● ● ●● ● ● ●●●● ● ● ●● ●●● ●● ● ●● ●●● ●●●● ● ● ● ● ● ●●● ● ●● ● ● ●●● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●●●● ● ● ● ● ●●●●●●●● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●●● ●● ● ● ● ● ● ●●●●● ●●●●●●●● ● ● ●● ● ● ● ● ● ● ● 2 ● ● ●● ● ●●● ● ●● ● ● ● ● ● ● ● ● ● ● ●●● ●● ● ● ● T2D6 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ●● ● ● ● ● ● ●● ● ● ●●● ● ● ● ● ●●● ● ●● ● ●●●● ● ● ● ●● ● ● ● ● ● ● ● ● ●●●●● ● ● ●●●●●●●●●●●●● ● ●● ● ● ●● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ●● ● ●●●●●●●● ● ●● ●● ● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ●● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ●●●● ● ● ● ● ● ●●●●●● ● ● ● ●● ● 2 ●● ● ● ● ●●●● ●●● ● ●●●●●● ● ●●● ● ● ● ●● ● − ● ●● ● ● ● ● ● ●● ● ● ● ● ● ●●●●●●● ● ● ● ● ● ● ● ●● ● ● ●●●●●●●●●● ●●● ● ● ●● ●●●●● ● ●●● ●● ● ●● ●●● ● ● ● ● ● ● ● ●●●●●●● ● ● ● ● ●●●●●● ●● ● ● ●● ● ● ● ● ●● ●● ● ●● ● ● ● ● ●●● ●●●●●●●● ●●● ● ●● ● ● ● ● ●● ●● ● ● ● ● ● ●● ● ● ● ●●●●● ● ●● ● ●● ●●●●●●●● ● ● ● ●● ● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ● ● ●● ● ● ● ● ●● ● ●● ● ● ● ● ●● ●●● ●● ●● ● ●● ● − ● ● ●●● ● ● ● ● ●● ● ●●●●●●● ● ●●● ● ● ● ● ●● ● ● ● ● ●● ● ●● ●●● ●●● ● ● ●●●● ● ● ● ●● ● ● ● ●●●●●●●● ●●●●●●●●●●●● ●●●●●●●● ● ●●● ● ● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●●●● ● ● ● ●●●● ●● ●●●● ● ●●●●●● ●●● ● ●● ● ● ● ●● ●●●● ●●●● ●●● ●●●●●● ●●●●●●● ●●●●●●●● ● ● ● ● ●● ● ● ●●● ● ●●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ●● ●●●●●● ● ●●●● ●●●●●●●●●●● ●●●●● ●●● ●● ●●●●●● ● ● ● ●● ● ● ● ● ● ● ● ● ●●● ● ●●●● ● ●● ●●● ●● ● ●● ● ●● ● ●●●●●● ●● ●● ●●●● ●●●● ●●●●● ● ● ● ● ●●● ● ● ● ●● ● ●● ● ● ●●● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ●● ●●● ● ●●●●● ● ● ●●● ● ●●●●●●● ●●●● ● ●●●●●●●●● ●● ● ● ●●● ● ● ● ● ●● ● ● ●●● ●● ●● ● ● ●●● ●● ●● ● ●●●● ●● ● ● ●●● ● ● ● ●●●●●●● ● ● ●●●● ●●●● ●●●● ●●●●●●●●● ●● ● ● ● ● ● tSNE ● ● ● ●●● ●● ● ● ● ● ●● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ●●●● ●●●●● ● ● ●●●●●● ●●●●● ●● ● ●●●●●● ● ●● ● ● ●● ●● ● ●●● ●● ● ● ● ●●● ● ●● ● ●● ● ● ●●● ● ●● ● ●● ●● ● ● ● ●● ● ●●● ●●●● ● ● ●● ●● ●● ●●●● ● ●●●● ● ●●● ● ● ● ● ● ● ● tSNE ● ● ● ● ● ● ● ● ● ●● ● ●●● ● ● ● ●● ● ● ●● ● ● ● ● ●●●● ●●●● ● ● ● ●●● ●● ● ● ● ●● ●●● ●●●●●●●●●● ● ●●● ● ●● ● ● ●● ● ● ● ●●●● ● ● ● ● ● ● ●● ● ●●●● ● ● ● ● ●● ● ● ●● ●● ●●●● ● ● ●●●●●●●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ●● ● ●● ●●●●●●● ● ●● ●● ● ●● ● ●●●● ● ● tSNE ●● ● ● ● ● ● ● ●●● ● ●●● ● ●●● ● ● ● ●● ● ● ● ● ● ●● ●●●●●● ● ●● ● ● ●● ●●● ● ● ● ● ● ●●●● ●●● ● ● ● ●● ●● ●● ●● ● ● ● ●● ● ●● ● ● ● ● ● ● ●● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ● ●●●●● ●●●● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ●● ● ● ● ● ●●● ●● ● ●●● ● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ●●●● tSNE ●● ● ● ●● ●● ●●● ● ● ●● ●● ● ● ●● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●● ●● ●●● ●● ● ● ● ●●●●●● ●● ● ●● ●●● ●● ● ● ● ●●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ●● ●●●● ● ●● ●●● ● ● ● ● ●●●● ●●● ●●● ● ●● ● ● ●● ●●● ●●● ● ●● ● ● ● ● ● ● ● ● ● ●●● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ●● ●● ●● ● ●● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ●● ● ●● ●●● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ●● ● ● ●● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●●●●● ● ●● ● ●● ● ●●● ● ● ●● ● ● ● ●● ● ●●●●● ● ● ●● ●● ● ● ● ● ● ●● ● ● ● ●●● ●●● ● ● Cluster A ●● ● ●● ● ●● ●● ●● ● ●● ●●●● ●● ●● ●● ● ● ●● ● ●● ● ●● ● ●● ● ●● ●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ●●●● ●● ● ● ● ● ● ●●●●● ●●●● ●● ● ● ● ●● ● ●● ● ● ● ● ● ● ●●●● ● ● ● ●● ● ● ● ●● ● ●●● ● ●●● ● ●●●● ● ● ● ● ● ● ● ● ●● ● ●●● ● ●● ● ●● ●● ●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●● ● ● ● ● ●●●● ●● ● ●●●●●●● ●●●●● ●● ● ● ● ● ● ● ● ● ●●●●● ● ● ●● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ●●● ● ●●●● ● ● ● ● ● ●● ● ●● ● ●●● ● ● ●●●●● ●● ●● ● ●●● ●● ●●●●●●●●●●● ● ●●●● ● ● ● ● ● ●●● ●●● ● ● ● ● ● ●● ● ●●●● ●●●● ●● ● ● ●●● ●●●● ●● ●●●● ● ● ● ● ● ● ● ●●● ● ● ●●●●●●●●●● ● ●● ●●●● ●● ●● ●●● ● ● ● ● ● ● ● ●● ● ●● ● ● ●●● ●●●● ● ●● ●●● ●● ● ●●●●●● ● ●● ● ● ●●●● ●●● ● ●●● ● ●● ●● ●● ● ● ●●● ●●●●●● ●●●●●● ● ● ●●●● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ●● ● ● ●●●● ● ● ●●● ●● ● ● ● ●● ●● ● ●●● ●● ● ●● ● ● ●● ●● ● ●●●● ● ● ●●●●●●●● ● ● ●● ●●● ● ● ● ●●● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ●● ● ●● ●●●●● ●●● ● ●● ● ● ● ● ●●● ● ● ● ●●● ●●●●●● ●● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ●●●●●● ● ● ● ● ● ● ●● ● ● ● ●●● ●● ●●● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●●● ●● ●●● ● ● ● ●● ● ●● ● ● ● ●●●●●● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ●● ● ● ● ●●● ● ●●● ● ● ● ● ●●● ● ● ● ● ●●● ●●●● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●●● ● ● ●● ● ● ● ● ● ● ● ● ●●●● ●● ●●● ●●●●● ● ●●● ● ● ●●● ● ● ● ● ● ● ● ●● ● ●● ●● ●●● ● ● ● ●● ●● ● ●● ●● ●● ● ● ● ● ●●● ● ●●●● ● ● ● ● ● ●●● ●●● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ●● ●● ● ●● ● ● ● ● ● ●●● ● ● ●●●●●●● ●●●● ●● ●●● ● ● ● ●●● ● ● ● ● ● ● ●●● ●● ●● ● ● ● ● ●● ● ● ●●● ● ●● ● ●● ●●●●● ● ●●●● ● ●●● ● ● ● ● ● ●●● ●●●● ● ● ● ● ●● ●●●●● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ●● ● ●● ● ●●●●●● ● ● ● ●●● ●● ● ● ● ●● ● ●● ● ●● ●●● ●● ● ● ● ● ● ● ●●● ●● ● ● ●● ●●● ● ●● ● ●● ● ●● ●● ● ● ●●● ●● ●● ● ● ● ● ● ● ●●●●●● ●●●●●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●●● ●●●●●● ●● ● ●● ● ● ● ●● ● ● ● ●● ● ●●● ●● ●●●● ●●● ● ● ● ● ● ●● ●● ● ● ● ● ●●●● ● ●●●● ● ● ●● ●●● ● ● ● ● ●●● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ●● ●●●●●●●● ● ● ●● ●● ● ● ● ● ●● ●●●●●●●●● ● ● ●● ● ● ● ● ●● ● ● ● ●● ●● ●●● ●● ●●●●● ●● ● ● ● ● ● ● ● ●● ●●●●●●●●●●●●● ● ●● ●● ● ●● ●● ● ●● ● ●●●●●● ● ●●●● ● ●● ●●● ● ●● ●●●●●●●●● ● ●● ● ●●● ● ●●●● ● ● ● ● ●●●●●●●●●●● ●● ● ● ●●●●● ● ●●●●● ● ● ● ●●● ●● ● ●● ● ●●●● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ●●●● ●● ●●● ●● ● ● ● ● ● ●●●● ●●●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ●● ●●●●●●●● ●● ●● ● ● ●● ● ● ● ●●● ● ●●● ● ●●● ● ● ● ● ● ● ● ●● ●●●●●●●●● ● ● ● ● ● ● ● ●●●●● ●● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ●●● ●●●● ●● ● ●● ● ● ● ● ● ● ● ● ●● ●● ● ●● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ●● ● ● ● ● ● ● ●●● ● ●●● ● ● ● ● ● ● ● ●●●● ● ● ● ● ●●●● ● ●● ●● ● ● ● ● ●●● ● ● ●●●●●●● ● ●● ● ● ● ●●●● ● ●● ● ●● ● ● ● ●●● ●● ●● ● ●●●●●● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ●● ●●● ●● ● ●●●● ●● ● ● ● ● ● ●● ●●● ●●● ● ● ● ●●● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ●●● ●● ●● ● ● ●● ● ●● ●●●● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ●●●● ● ● ● ● ● ●● ● ● ●●●●● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ●●● ●●●●●●● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ●●● ●●● ● ● ●● ●●● ●● ●●● ●●●●● ● ●●●● ● ● ● ●● ● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● ●● ●●●● ●●● ●●●●●●●●●●●● ● ● ● ● ● ●●●●●●●● ● ●●● ● ● ● ●● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ●● ●●●●●●●●●●●●●●● ● ● ●●● ●●● ● ●●●●● ●●● ● ●●●● ● ● ● ●● ● ●● ● ● ●● ● ● ●●● ● ●● ● ●●●● ● ● ●● ●● ●● ● ● ●● ●●●● ● ●●● ●●●● ● ●●● ●●●● ● ●●●● ● ●● ● ● ●● ●● ● ● ● ●● ● ● ●● ●●● ● ● ● ● ●● ●● ●● ●● ●● ● ● ●● ●●●●● ● ●● ● ●● ●●●●● ● ● ● ●●● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ●●●●● ●●●●● ●●●●●● ● ●● ● ●● ●●●●●● ●●● ●●●●●●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●●●●●●● ● ●● ●●●● ●●● ●● ● ● ● ● ● ●●●● ●●● ●●● ●● ● ●●●●●●●●● ●●●● ●●● ● ● ●● ●● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ●●●●●●● ●●● ●●● ●●●●● ● ●●●●●●● ●● ●● ● ●● ●●●●●●● ● ●●●●● ● ● ● ● ●● ●● ● ● ● ●● ● ●●●●●● ●● ● ● ● ●● ● ● ● ●●●● ●●●●● ●●● ●● ●● ● ●●●● ●●●●●●●●●●●●● ●● ●●●●●●●●●●●● ●●● ● ●●●●● ● ● ● ● ● ● ● ● ● ●●●●● ● ● ● ● ●●● ● ● ● ● ●●●● ● ●●●●●● ● ●● ●● ● ●●● ●●●● ●●●●●●●●●●●●●●● ●●● ● ●● ● ● ● ● ●● ●● ●● ●●●● ● ●● ●●●● ●●● ● ● ● ● ●● ●●●●● ●● ● ●●●●● ●●●● ●●●● ● ●●●●●●●●●●●● ●●●● ●●●● ●●● ●●●●● ● ●●● ●● ● ● ● ● ● ● ●● ●● ● ● ● ● ●●● ●●● ●●● ●● ● ● ● ●● ● ● ● ●●●●●● ●●●● ●● ●●● ● ●● ● ● ● ●●● ●●● ●●● ●● ●●●●●●●●●●● ● ●●● ● ●● ● ● ● ● ● ● ●● ●● ●● ●●● ● ●●● ●● ●●●●●●● ●●●●●●●● ● ●● ●●● ●● ● ●●● ● ●● ●● ● ●●●●● ●●●●●● ●●●●● ●●● ●●● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●●● ●●●●●● ● ● ●● ● ● ●● ● ●●●●●● ●●● ●● ● ● ●●●●●● ●● ●● ●●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ●●● ●● ● ● ● ● ●●● ●● ●● ● ● ● ● ● ●●● ●●● ● ●●●● ●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●● ● ●● ●●●● ●●●●●●●●● ●●●●●●●● ●●●● ●●●●●● ● ● ● ● ●● ● ● ● ●●● ● ● ● ●●●●●● ●●●●●● ● ● ●●●●●●●●●●●● ●● ●●●●●●● ●●● ● ●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●● ●●● ●●●●●●●●● ● ● ● ●● ● ● ● ●● ● ●●● ●● ●●●●● ●● ● ● ● ●● ● ●●●●●●●●●●●●●●●●●●●●● ● ● ●●● ●●●●●●●●●●●●●●●●●●●●● ●● ● ●●●● ● ● ●●● ● ● ● ● ●● ● ● ● ● ●●● ● ● ●● ● ●●● ● ● ● ●●●●●●●●●●●●●●●●●● ●● ●● ●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●● ●●● ● ●● ● ● ● ● ●●● ●●● ● ● ● ● ●● ●● ●●●●● ●●●● ● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●● ●● ● ● ● ●● ● ●● ●● ●● ● ● ● ● ● ● ●●●●● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ●●● ● ●● ● ● ● ● ● ● ●●● ● ● ●● ● ●● ● ●● ● ● ● ● ● ● ● ● ●●●●●●●●●●● ●●● ● ●● ● ●● ●● ● ●● ● ● ● ● ● ● ● ● ● ●● ●●●● ●●●●●●●●●●● ● ● ●●● ●●● ● ● ●● ● ● ●● ● ● ● ●●●●●●●●● ●●●●●●●● ●● ●● ●● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ●● ●● ● ● ●●● ●●●● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ●● ● ●● ● ● ●● ●● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ●● ●● ●● ● ●●●●●● ●●●●●●● ●● ● ● ● ● ● ● ● ● ●●●●● ● ●● ●● ● ●● ● ● ●● ● ● ●● ● ●● ●●● ● ● ● ● ● ● ● ●● ●●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ●●● ● ● ●●● ●● ●●●● ●● ● ● ●● ●● ● ●● ● ● ● ● ●● ● ● ●

tSNEtSNE−1-1 tSNEtSNE-1 −1

C INS Expression GCG Expression INS expression D GCG expression

● ● ● ● ● ● ●● ● ● ●● ●●●●● ●●●● ● ● ● ●● ● ● ● ●●●● ● ● ●●● ● ●●● ● ● ● ● ●●●●●● ● ● ● ● ● ●●● ●● ● ●● ● ●●●●●●● ●●● ● ●● ● ●●●●●●● ●●● ● ● ●●● ●● ● ●●●●●●●●● ●●● ●● ● ●●●●●●●● ●● ●●●● ●●● ●● ● ●●●● ●● ●●●● ●●●● ●● ● ●●●● ● ● ● ●●●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ●● ●●●● ● ● ● ● ● ●● ● ●●●● ● ● ● ●●● ●●●●●●●●●● ● ● ● ● ●● ●●●●●●●●● ● ● ●●● ● ●●●●●●● ●● ●● ● ● ●●●●●●●●● ●●● ● ● ● ● ●●●●●●● ●●● ● ● ● ● ● ● ●●●●●● ● ● ● ●● ● ● ● ● ● ● ●●●●●● ●●●● ● ● ● ●● ● ●●●●●●● ●●●● ● ●● ● ●● ●● ●●●●●●● ●● ● ● ●● ● ●● ●● ●●●●●● ●● ● ● ● ●● ● ● ●●● ●●●●●●●●● ● ● ●●● ● ● ●● ●● ● ●●● ●●●●●●●●● ● ● ●●● ● ● ●●●●●●●● ●● ●● ●●●●●●●● ●● ●●● ●● ● ● ● ●●●●●●●●● ●● ●● ●●●●●●●● ●● ●●● ●● ● ● ● ●● ● ●● ●●●●●● ●●●●●●● ●●● ● ●● ● ● ● ●● ● ●● ● ●●● ●●●●●●● ●●● ● ●● ● ● ● ● ● ● ● ●●●●●●●●●●●● ●●●●●● ●●●●●●●●● ● ● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ●● ● ●●●●●●●●●● ● ●● ● ● ●● ● ● ●●●●●●●●● ● ●● ● ● ● ●●● ● ●● ●●●●●● ● ●● ● ● ● ● ●●●●● ●● ●●●●●● ● ●● ● ● ● ● ●● ● ● ●● ●●●●●●●●●●●●●● ● ● ● ● ● ●● ● ● ●●● ●●●●●●●●●●●●●● ● ● ● ● ● ● ●● ● ●● ●●● ●●●●● ● ●● ●●●●●●● ● ● ●● ● ● ● ●● ● ● ●● ●●●● ●●●●● ● ●● ●●●●●●● ● ● ●● ● ●● ●● ●●● ●●●●● ●●●● ●●● ●●● ●●● ● ●● ●● ●●● ●●●●●● ●●●● ●●● ●●● ●●● ● ●● ● ● ● ●● ● ●●● ● ●●● ●●● ●● ●●●● ● ● ● ● ●● ● ●●● ● ●●●● ●●● ●● ●●●● ● ● ● ●●●● ● ●●●●●● ●● ●● ● ●●●● ● ● ● ●●●● ● ●●●●●● ●● ●● ● ●●●● ● ● ●● ●●● ●● ● ●●● ●●●● ● ● ● ● ●● ●●● ●●● ● ●●● ●●●● ● ● ● ● ●●● ● ●● ● ●●●● ●●●●●●●●● ●●● ● ● ● ● ●●● ● ●● ● ●●●● ●●●●●●●●● ●●● ● ● ● ● ● ● ● ●●●● ●●●●● ● ●●●●● ●●●●●●●● ● ●● ● ● ● ● ● ● ● ● ●●●● ●●●●● ● ●●●●● ●●●●●●●● ● ●● ● ● ● ● ● ● ●●● ●●●●●●● ●● ● ● ●● ●● ●●●● ● ●●● ● ● ● ●●● ●●●●●●● ●● ● ● ●● ●● ●●● ● ●●● ● ● ● ●●●● ●● ●● ●●●●●●●●● ●● ● ● ●● ● ● ●●●● ●● ●● ●●●●●●●●● ●●● ● ● ●● ● ●●● ●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ● ● ● ●●● ●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ● ● ●● ●● ● ●● ● ●●● ● ●● ●●●● ●●●●●●●● ● ● ●● ● ● ●● ●● ● ●● ● ●●● ● ●● ●●●● ●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●● ●●●● ● ●●●●●●● ● ●● ●● ●●● ●●● ● ● ● ● ●●●●● ●●●● ● ●●●●●●● ● ●● ●● ●●● ●●● ● ● ● ● ●●●● ●●● ●● ●●●● ●●●● ● ● ●● ●● ● ● ●●●● ●●● ●● ●●●● ●●●● ● ● ●● ● ● ● ●●● ●● ● ● ●● ●●●●● ● ● ●●● ●● ● ● ● ● ● ●● ●● ● ● ●●● ●●●●● ●● ● ●●● ●● ● ● ● ● ● ●● ●● ●●● ●● ● ● ●●●●●●●● ● ● ● ●● ● ● ● ● ●● ●● ● ● ●● ● ●●●●●●●●● ● ● ● ●● ● ● ● ● ●● ●●● ●●●● ●● ●●● ●● ●● ●●● ● ● ● ● ● ●● ●●● ●●●● ●● ●●● ●● ●● ●●● ● ● ● ● ● ●●● ● ●● ●●●●●●● ●●●●●● ●● ● ● ●● ● ● ● ●●● ● ● ● ●●● ● ●● ●●●●●●● ●●●●●●● ●● ● ● ●● ● ● ● ●●● ● ● ● ● ●● ●●●● ● ●●●●●● ●● ●● ●●●●● ● ● ● ● ● ● ● ● ●● ●●●●●● ●●●●● ●● ●●●●●●●● ●● ● ● ● ●● ● ● ●● ●● ● ● ● ●●●●● ● ●●●●●●● ● ●●● ● ●● ●●● ●● ●● ● ● ●●●●● ● ●●●●●● ● ●●● ● ● ● ●●●● ●●● ●●●●●●● ● ● ● ●●●● ●● ●●● ●● ●● ● ● ● ● ●●●●● ●●● ●●●●●●● ● ●●● ●●●● ●● ●●● ●● ●● ● ● ●●● ●●●● ●●●● ●●●● ●●● ● ●● ●● ●● ●● ●● ● ●● ● ●● ●●●● ●●●● ●● ● ● ● ●● ●● ●● ● ● ● ● ● ●● ●●●●● ●● ● ●●● ● ●●●●●●●●● ● ●●● ●● ● ● ● ● ● ● ●● ●●●●● ●● ● ●●● ● ●●●●●●●●● ● ●●● ●● ● ● ● ●●●● ●●● ●●●●●●●●●●●●●●●●●●● ● ●●● ● ● ●●●● ● ● ● ● ● ●●● ●●● ●●●●●●●●●●●●●●●●●●●● ●● ●●●● ● ● ●●●● ● ● ● ● ● ● ●● ● ●● ●●● ●●●●●●●●●● ● ●●●●●● ●●●● ● ● ● ● ●● ● ● ●● ●●● ●●●●●●●●●●● ● ●●●● ●●●● ● ● ● ● ● ●● ● ●●●●● ●●●●●●● ●●●●●●●●●●●●●●●●● ●● ● ●● ● ●● ● ● ●● ● ●●●●● ●●●●●● ●●●●●●●●●●●●●●●●● ●● ● ●● ● ●● ● ● ●●●● ● ●● ●●●●●●●●●●● ●●●● ●● ●● ● ●● ●●●● ● ● ●●●● ● ●●●●●●●●●●●●●●● ●●●● ●● ●● ● ●● ●●●● ● ●● ● ● ●● ●●●●●●●● ●●●●●●●●●●●●●● ●● ●●●● ●● ● ● ● ● ●● ●● ●● ● ● ●● ●●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●●● ● ● ● ●● ● ●● ● ● ● ● ●● ● ●●●●●●●●●●●●●●● ●●●● ●●●●●● ●●● ● ●● ● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●●●● ●●● ●●●●●●● ●●●● ● ● ● ●● ● ●● ● ●● ● ●●●●●●●●●●● ●●●●● ●● ●●●● ●● ● ● ●● ● ●● ● ● ●● ● ●● ● ●●●●●●●●●●●●●●●● ●●● ●●●● ● ● ●● ● ● ●● ●● ●●● ●●● ●●● ●●●●●●● ●●●●●● ●●●●●●●●●●●●●● ●● ●● ●● ●● ●●● ●●● ●●● ●●●●●●●● ●●●●●●●●●● ●●●●●●●●●● ● ●● ●● ●●● ● ●●●●●● ●● ●●●●●●●●●●● ●●●●●●●●●●●●●●● ●● ●●● ● ●●● ● ●●●●●● ●● ●●●●●●●●● ●● ●●●●●●●●●●●●●●● ●● ●●● ● ● ● ●●● ●● ●●●●●●●●● ●●●● ● ●●●●●●● ●● ●● ● ● ● ● ●●● ● ●●● ●●● ●●●●●●●●● ●●●● ●● ●●●●● ●● ●● ● ● ● ●● ● ● ●● ●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●● ● ●● ● ● ●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●● ● ● ● ● ● ●● ● ● ●●● ●●●●●●●●● ●●●●●●●●●●●● ●● ●● ● ● ● ● ● ● ● ● ●● ● ● ●● ●● ●●●●●● ●●●●●●●●●●●● ●●● ●● ● ● ● ● ● ● ● ●● ●●●●● ●●●●● ●● ●●●●●●●●●●● ● ●● ● ●●● ● ● ●●●● ● ● ●●● ●●●●●● ●●●●● ●●●●● ●●●●●●●●●●● ● ●● ● ●●● ● ● ● ●●●● ● ●●● ●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●● ●●● ● ● ● ● ● ● ● ● ● ●●● ●● ●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ● ●●● ●● ● ● ● ● ● ●●● ●●●●●●●●●●● ●●●● ●●●●●● ●●●●●●● ● ● ●● ● ●●●● ● ● ● ● ●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●●● ● ● ● ●●● ● ● ●●●● ● ●●●●●●●●●●●●●●●● ●●●● ● ●● ●● ●● ● ● ●● ●● ● ● ●●●● ● ●●●●●●●●● ●●●●●●●● ●●●● ●●●● ●● ●● ●● ● ● ●● ● ● ● ●● ●● ●●●● ●●●● ●●●●●●●●●●●●●● ●●●●●●●● ●● ●● ● ●● ● ● ● ● ●● ●●●●●● ●●●● ●●●● ●●●●●●●●● ●●●●●●● ●●●●●●●● ●● ●● ● ●● ● ● ● ●● ● ● ●●●●● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●● ● ●●●● ● ● ● ● ●● ● ●●● ●●●●●●●●● ●●●●●●●●●●●●●●●●●●●● ●● ●● ●●● ● ● ●● ● ●●● ●●● ●● ●● ●●●●●●●● ● ●● ● ● ● ● ● ●●●●● ●● ●●●● ●●●●●●●●●● ● ●● ● ● ● ● ●● ●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●● ●● ● ● ● ● ●● ●●●●●●●● ●●● ● ●● ●●●●●●●●●●●●●●● ● ●● ● ● ● ● ● ● ●●●● ●●●●● ●●● ●●● ●●● ●●●●● ● ●●● ● ● ● ●●●●●●●●●●● ●● ●●● ●●●● ●●●● ● ●●● ● ● ● ● ● ●● ●●●● ● ●●●●●●●●●● ● ● ● ●●● ●●● ●● ●●●● ● ● ● ● ● ● ● ● ●●● ●● ●●● ● ●●●●●●●● ● ● ● ●●●●● ●●● ●●● ● ●● ● ● ● 2 ●● ● ●●● ●●● ● ●● ●●● ● ●●●●●●● ● ● ● ● ●● ●●● ● ●● ●●● ●● ●● ●● ●●● ● ●● ● ● ● ●● ●● ●●●● ●● ●●● ●●● ●● ●●●●● ● ●●●● ● ● ● ●● ●● ●● ● ●●●●●●●●●● ●●●●● ● ● ●● ● ● ● ● ●● ●● ●●●●● ●● ●●● ●●●● ●● ●●● ●● ● ● ●● ●● ● ●● ●●●●● ● ●●● ●●●● ●● ●●● ● ● ● ● ● ● ● ● ●● ●●●● ●● ●●●●● ●●● ● ● ●● ● ● ● ● ● ●● ●●●● ● ●● ●●●●● ●●● ● ● ● - ● ● ● ● ● ● ●● ● ● ● ●●● ●●● ●● ●● ●●●● ●●●●● ●● ● ● ●● ● ●●● ●●● ● ●● ● ● ● ●●● ●●●● ●● ●● ●● ● ●●●●● ●●● ●●●● ●● ● ●●● ●●● ● ● ● ● ●●●● ●●●● ●●●● ●● ●● ● ● ● ● ● ● ● ● ●●● ● ● ● ●● ● ● ● ● 2 ● ● ● ● ●● ● ●●● ●● ● ●● ● ● ● ● ● ● ●●● ●● ● ●● ● ●●● ● ●● ● ● ● ●● ● ●●● ●●●●●● ●●● ● ● ● ● ● ●● ● ● ●●●●● ●● ● ●●● ●●● ●●●● ● ● ●● ● ● ● ●● ●●●● ●●●● ●● ●●● ●●● ● ●●● ●● ● ●●●●●●● ●● ●● ●● ●● ●●● ●●●●● ●● ●● ● ● ● ●●●●●●●● ●● ● ●● ●● ●●●●● ●●● ● ● ●● ● ● ● ● ● ●●● ● ● ●●● ● ●● ● ● ● - ● ● ●● ● ● ● ● ●● ● ●● ●● ● ● ●●● ●● ●●●● ● ●● ● ●● ● ●● ●●● ● ● ●●● ● ●●●●● ●● ●● ● ●● ● ●● ● ●● ●● ● ● ●● ●● ●● ● ● ● ●● ● ●● ●● ●● ●●●● ●● ●● ● ● ● ● ● ●●● ●●● ● ●●●● ●●●● ● ●●● ●●● ● ● ● ●● ● ● ● ● ●●● ● ●●● ●●●● ●●●● ●●●●● ●●● ●●● ● ● ● ● ●● ● ● ●●● ●● ●●● ●● ●● ●● ●● ● ● ●●●● ● ● ● ●●● ●● ● ●● ●● ● ●●● ● ●● ● ● ● ● ●● ●● ●● ●●● ● ● ● ●● ● ●●●●●● ●● ●● ●● ● ● ● ●● ● ●●●● ● ● ● ● ●●●●●●●●● ● ●● ● ● ●● ●●●●●● ●●●●● ● ●●● ● ● ● ●●● ● ● ● ● ● ● ●●●●●●● ●●●●● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●●●●●●● ● ● ● ● ● ● ●● ● ● ●●●●●●●●●●●●● ●●●● ● ● ● ● ● ● ● ●● ●● ● ●●● ●●●●●●●●●●●● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●●● ● ● ●●●●●●●●●●●●● ● ●● ● ● ●● ● ●● ●● ● ●● ● ●●●●●● ●● ●●●●●● ●●● ●● ●●●●● ●●● ● ● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ● ● ● ● ●● ●● ● ●● ●● 2 ●● ● ●● ● ●●●●●● ● ● ● ● 2 ●● ● ● ● ● ●● ● ●●●●●● ●●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ●● ●●●●● ● ●●●●●● ● ● ● ●● ●●●●● ● ●●●●● ● ● ● ●● ●●● ●●●●●●●●●● ●●● ● ● ●● ●●●●● ● ●●● ● ● ● ● ●● ● ●● ●●●● ●●● ● ●●● ●●●●●●●● ●●●●● ●● ● ● ● ●● ●● ● ●● ● ● ● ● ●●● ●●●●●●●● ●●● ● ●● ● ● ●● ●●●●● ●●● ● ●● ● ●●● ●●● ●●●●●●●● ●●●●● ● ●●●●●●● ● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ●● ● ● ●●● ● ● ●● ● ● ● − ●● ● ● ●● ●●●●● ● − ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ●●●●●● ● ● ●● ●●●● ●●●●● ●●● ● ● ● ●● ● ● ● ● ● ● ●●●●● ● ●●● ●●● ●●● ●●● ● ●●● ● ●● ● ●● ● ●●●●● ●●●●●●●● ● ● ●●●● ●●●●● ●●●●●●●● ● ●●● ●●● ● ● ● ●●● ● ●●●●●●●●●● ● ●●●● ●●●●●●● ●●●●●●● ●● ● ● ●●● ●● ● ●●● ●● ●●●●●●●●●● ●●●●●●●● ●●●●●● ● ● ● ● ● ● ● ● ●●●●● ●● ● ●●●●●●●●●●●●●●●●● ● ●●●● ●● ● ●●●●● ● ● tSNE ● ●● ●● ●●● ● ● ●● ●●●●●●●● ●●●● ●●● ● ●●● ● ● ● ● ● ●●●● ● ●● ● ● ● ●●● ● ●●● ● ●● ● ● ● ●● ●● ●●●●●●● ● ●● ●●●● ●● ● ●●●●● ● ● ● ● ● ●●● ● ●●● ●● ●●● ● ●●●●●●● ● ●●●●●●●● ● ● ● ●●●● ●●●●● ● ● ●●● ● ●●●●●●●● ●●●●●●● ●●●●●●● ●●● ●● ● ● ● ●●●●●● ●●●●●● ● ● ● ● ●●●●●●● ●●●●●●●●●●● ●● ●●● ●● ● ● ● ● ●●●● ●● ● ● ●●●● ●●●●●●●● ●●●●●●● ●● ● ● ●●● ●●●● ● ● ●●●● ●●●●●●●●● ●●● ●● ●● ● ●●●●●●● ●●●● ●● ● ●●●●●● ● ●●● ● ● ●●● ●● ● ●●●●●● ●●●● ●● ● ●●●●● ● ● ●● ● ● ●● ●●● ●● ●●● ●●●●● ● ● ●● ●●●●● ●●●● ●●● ● ● ● ● ● ● ●●●●● ● ●● ●● ●●●● ●● ●● ●●●●●● ●● ● ● ● ●● ● ● ●● ●●●●● ●● tSNE ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●●●● ●● ● ● ● ● ● ● ●●● ● ● ●● ● ● ●●●●●●●● ● ● ● ● ● ●●●● ● ● ●● ● ●● ● ● ● ●●●●●●● ● ●● ●● ●● ●● ●●●● ● ●● ● ● ● ● ●●●●●● ●●●● ● ● ●● ●●●●●● ● ● ● ● ● ●● ●●●● ● ● ● ● ● ● ●●●●●● ● ●●● ● ● ● ●● ●●● ● ● ● ● ●●●●● ● ●● ● ●● ● ● ●●● ● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ● ● ● ●●● ● ●● ●●● ●● ● ●●● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ● ● ● ● ● ● ● tSNE ●● ● ● ●● tSNE ●● ● ● ● ●● ●● ●●● ● ●●● ● ●● ● ● ●● ● ●● ●● ● ● ●●●●●● ●● ●● ● ●●● ●● ● ● ● ●●● ● ●● ● ●●● ● ●● ●● ●●● ●● ●●●●● ●●●● ●●● ●● ● ●● ●●● ●●● ●●● ●●● ● ●●● ● ●●● ● ● ●●● ● ●●● ●●● ● ● ● ● ● ● ● ● ● ● ●●●● ●● ●● ● ● ● ● ● ● ● ●● ●●● ●● ●●● ● ●● ● ● ● ● ● ● ● ● ●●● ●●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●●● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●● ● ●● ● ● ●●●●● ● ●●● ●●● ●● ●●● ●●●●●●● ● ●●● ●● ● ●● ● ● ● ●● ● ● ●● ●●●● ●● ●● ● ●● ● ● ●● ● ● ●● ● ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ●●●● ●●●●● ● ● ●● ●● ●● ● ●●●● ●●●●● ● ● ● ●● ● ● ● ● ● ● ●●● ●● ●●● ● ●●●● ● ● ●● ●● ● ● ● ● ● ● ●● ● ● ● ●●●●● ●●● ● ●●● ● ● ● ●●●● ● ● ●●●●●●● ●●●●● ●● ● ●● ●●● ● ●●●● ●●●●●●●●●●● ●●● ●●● ● ● ●● ● ● ● ● ●● ● ●●● ● ● ●●● ● ●● ●●● ●● ●● ● ● ●●●●●● ●● ● ●● ● ●●●● ●●●● ● ● ● ●●● ●●●● ●● ● ●● ●●●●● ●● ●● ● ● ● ● ● ●●●●●●●● ● ●●● ● ● ● ●●● ●●●● ● ●● ●●● ●● ● ●●●●●● ● ●● ● ● ● ●● ● ● ● ●●●●●● ●● ● ●● ●●● ● ● ●●●● ● ●● ●●● ●● ● ● ● ● ●● ●● ● ● ● ● ●● ● ● ●● ● ● ● ●● ● ●●●● ● ● ● ●●●● ●● ● ●● ●●●●● ●●● ● ●● ● ● ● ● ●● ● ●● ●●●●●●● ●● ● ●●●●●●●● ● ● ● ● ●● ● ●● ● ● ● ●●● ●● ● ●●● ●● ● ● ● ●●●● ● ● ● ● ● ●●● ● ● ●● ● ● ● ● ●● ●● ● ● ● ●● ●● ● ●●● ●● ● ● ● ●●●● ● ● ●● ●●● ● ● ● ● ● ●● ●● ● ●● ●● ● ●●● ●● ● ●●● ●● ●● ●● ● ● ●● ● ● ●● ●●● ●● ● ●●●● ● ● ●● ●● ●● ● ● ● ●● ●● ●● ● ● ● ●● ● ● ● ● ●● ● ● ● ●●● ● ●● ●● ● ● ● ● ● ● ● ● ●●●● ● ●●● ● ● ●● ●● ● ● ● ● ● ●●●●● ● ●●●● ● ●●● ● ● ● ● ● ● ● ● ●● ● ● ●● ●●● ● ● ● ●●●●●● ● ● ● ● ● ● ● ● ● ●● ● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ●● ● ● ● ● ●● ● ●● ●●● ● ●● ● ● ● ● ● ● ● ● ●● ●●●●●●● ●● ●●● ● ● ● ● ● ● ● ● ● ● ● ●●●●●● ●● ●●●● ● ● ● ● ● ● ● ● ● ● ● ●● ●●●● ● ● ●●● ● ●● ● ● Activity (Z−Score) ● ●● ● ●●● ●● ●●●●●●● ●●● ● ● ● ● ● ● ●● ●●● ●● ●●●●●● ●●●●● ● ● ● ● ● ● ● ● ●● ● ● ●●●●●●●●● ● ● ● ●● ●●● ● ●●●●● ●●●● ● ● ● ●● ● ●●●●● ● ● ● ● ● ● ●●● ● ●●●● ● ● ●●●●●●●●● ●●●●● ● ●● ●●●●●●●●●●●● ● ● ● ● ●● ●●●●●●●●●●● ● ● ●●● ● ● ● ●● ● ●●●●●●●● ● ● ●●● ● ●● ● ●●●● ● ●●● ●● ● ● ● ●● ●●● ● ●●●●● ●●● ●● ● ● ● ● ●● ● ●●●●● ●●●● ● ●● ● ● ● ● ● ●●●●●●●●● ●● ● ● ● ● ●●● ●● ●●● ●●●●●● ● ● ● ● ● ●● ●●● ● ●●● ●● ●●●● ● ● ● ● ● ●● ● ●●●● ●●● ● ●● ●● ● ● ● ● ● ● ●●● ● ●● ● ● ● ● ●●● ●● ●●● ●● ● ●●●● ● ● ● ● ● ●● ●● ●●●●● ● ●● ●● ● ●● ●● ●●●● ●● ● ●● ● ●● ● ● ● ●●● ●●●●● ● ●●●●●● ● ● ● ● ● ● ● ●●●●● ●●●● ●●●●●●●●● ●● ● ● ●● ●● ● ● ● ●● ●●● ●● ● ●●●●●●● ● ● ●●● ● ● ●● ●● ●● ● ● ● ● ●●● ● ● ● ● ● ● ●●●● ●●● ●● ●● ● ●● ● ●● ● ●● ●● ● ● ●●● ●● ●● ●● ● ●●● ● ● ● ● ● ●●● ● ● ● ● ● ● ●● ●● ● ●●● ● ●● ● ●●●● ●● ● ● ●● ● ●● ●● ●● ●● ●●●● ● ● ● ●●● ● ●● ● ●●●● ●● ●●●● ● ● ● ● ● ●● ● ●● ●●● ●● ● ● ●●● ● ● ●●● ●●● ● ● ●● ● ● ●● ●●● ● ●●● ●● ●●●●● ● ●● ●●●● ●● ●●●● ●●●● ● ● ●● ● ●● ● ●●●●●●●●●● ● ● ●● ● ●●●●●●●●●●● ● ● ●●● ● ●● ●●●●●●●●● ●●●●●●●●●●●●●● ● ●●● ● ●●● ● ●●●●●●●●● ● ●● ● ● ● ●● ● ●●●●●●●●●●●●●●●●●●●●●● ●● ● ●●● ●●● ● ●●●●●● ●●●● ● ●●●● ●● ● −2 0 2 ● ● ●●● ● ●●●● ●●● ●●●● ● ● ●●● ● ●●●● ●● ●●● ● ●●●●● ● ● ●● ●● ● ●● ● ● ●●●●● ● ●●●● ● ●●● ● ●●● ● ●● ● ●●●● ● ●● ●● ●● ● ● ●● ● ●●●● ●●● ●● ● ●●●● ●●●●● ● ●●●●● ●● ● ● ●●●●● ●● ● ●● ● ● ●●● ●●●●● ● ● ●● ●●● ● ● ● ● ● ● ●●●●● ●● ●● ● ● ● ●●●● ●●●●● ● ●● ● ●● ●●●●● ● ●●●● ●● ●●●● ● ●●●● ● ●● ●● ● ●● ● ●●●● ●●●●●●●●● ● ● ● ●●●●●● ● ●● ● ●●●● ● ●● ●● ● ● ● ●●●●●● ●● ●●●●●●● ●● ●● ● ● ●●●● ●● ●●●●●● ●●● ●● ● ● ● ● ● ● ●●●● ● ●● ●●● ● ●●●● ●●●●● ● ● ●● ● ●●●●●● ● ●● ●●● ●●●● ●●● ● ● ● ●●● ●●●● ●● ●●● ●●●● ● ●●●● ●● ●●● ●● ●●●●●●●●●● ●●● ●●●● ●●●●● ●●●●●● ● ● ●● ●● ●●●●● ●●●● ● ●●●●●● ● ● ●● ●●●●●●● ● ●●●●●●● ●● ●● ● ●●●●●●● ●●● ●● ●●●●● ●● ●●●●● ●●●●●●● ● ● ● ●●●● ●●●●● ●● ●● ●● ● ●●●●● ●●●●●●●●●●●●● ● ●● ●●●●●●●●●● ● ●●●● ● ● ● ●●● ● ●●● ●●●● ●● ●● ●●●● ●●● ●●●● ●● ●● ● ●● ●●●●●●●● ●● ● ●● ●● ● ●●● ● ●● ●● ● ●● ● ● ●●●●●●●●●● ●●●●●●●●●●● ●●●● ● ● ● ●●●●● ●●●●● ●● ● ● ●●● ●● ● ● ●●●●●●●●●● ●●●●●●●●●●●●● ●●● ●● ● ● ● ● ●● ●● ●●● ●● ● ●●●●● ●●●● ● ● ● ●●●●●●●●●●● ●●●● ● ●●●●●● ●● ● ● ●●●● ●● ● ● ● Expression ●● ● ●● ● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ●●● ● ●●●●● ● ● ●●● ● ●● ●● ●●●● ●● ●●●●● ●●●● ●●●● ● ●●●●●●●●●●● ●● ●●● ● ● ●●●●●● ● ●● ● ● ●● ● ●● ● ● ● ●● ●● ●●●● ●● ●●●●●●●●●●●●●● ●●●●●● ● ●● ● ● ●●●●●●●●●● ●● ● ●●● ● ● ● ● ● ●● ●● ●●●●● ●●●●●●●●●●●●● ●●●● ●● ●●● ● ● ●●●●●●●● ● ●●● ●● ●● ● ●● ●● ●●● ●●●●●●●●●●● ●●●● ●●● ● ● ● ●●●●●●● ●● ● ● ●●●● ● ● ● ● ● ● ●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ●●●●●●● ● ●● ●● ● ●●●●●●●● ●● ●● ●●●● ●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ●●●●● ●● ● ● ● ● ●●●●●●●●●●●● ●● ●● ●● ●●●● ●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●● ●●●●●●●● ● ● ●●●●● ●●●●●●●●●●●●●●● ●● ●●●●●● ● ●● ●● ●●●●●●●●●● ●●●●●●●●● ● ● ●●● ●● ● ● ●●●● ●●●●●●●●●●●●●● ●●● ●●●●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●● ●●● ● ●●●●● ● ● ●●●●●●●●●●●●●●●● ●●●●●●●● ●●● ●● ●●●●●●●●●●●●●●●●●●●● ●● ●●●● ●●● ●●●●●●●●● ● ●● ● ●●●●●●●●●●●●●● ●●●●●●● ●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●● ●●● ●● ●●●●●●●● ● ● ● ●●●●●●●●●●●●●●●●●● ●●● ●● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●● ●● ● ●●●● ●● ● ●●● ● Expression (log2(RPM+1)) ● ● ● ●●●●●●●●●●●●●●●●●●● ● ●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●● ●●●●●●● ●●●● ● ● ● ●●● ● ●●●●●●●●●●●● ●●● ●● ●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ●●● ● ● ●● ● ●●●●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●● ●●● ● ●● ●● ●●● ●●●● ● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●● ●● ●●● ● ● (log2(RPM+1) ● ● ●● ●● ● ●● ● ●● ● ● ● ● ●● ● ●● ● ●●●●● ●●● ● ●●●●●● ●●●●●●●●●●●●●●●●●●●● ● ● ●● ●●● ● ● ● ●● ● ●● ●●● ●●● ●●●●●●●●●●●●● ●●●●●●●●●●●●● ● ● ●●●●● ● ● ● ●● ● ● ●●● ● ●●●●●●●●●●●●●● ●●●●●●●● ●●● ●●●● ● ● ● ● ●● ●● ●● ● ●●●●●●●●●● ● ● ●●● ●● ● ● ●● ● ●● ● ●●●●●●●●●●● ● ●●● ●●● ●●● ● ● ●●●● ●●●●●●●●●●●●●● ● ●●●● ● ● ● ●● ●●●●● ●●●●●●●●●●●●●●● ●●●●●●●● ● ● ●● ● ● ●● ●●●●●● ●●●●●●● ●● ●● ●● ●● ●● ●● ●●● ●●● ●●●●●● ●●● ●● ●● ● ● ● ● ● ● ● ●● ●● ●●● ●●● ● ●●● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ● ●●● ● ● ● ● ●● ●● ●● ● ● ●● ● ● ●● ●● ●● ● ● ●● ● ● ●● ●● ● ● ●●● ●● ● ●● ● ● ●●● ●● ●● ● ●●● ●●●●● ●●●●● ● ● ● ● ● ●● ●● ●●● ●●● ● ●●●●● ●●● ● ● ● ● ●● ●●●● ●●●● ●● ●● ● ● ● ● ● ●●● ● ● ●● ●● ● ● ● ●● ● ● ● ● ●● ●●● ● ● ● ● ●● ● ● ●● ● ●● ●●● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ●●● ●●●● ● ● ● ●●●● ● ● ● ●● ● ● ●● ●●● ● ● ● ●● ● ●●● ● ● ● ● ● ●●●●● ● ●● ● ● ●● ● ●● ●●●●● ● ● ● ●● ● ●● ● ● ● ●● ●● ● ● ●● ● 0 9 18 ● ● ● 0 9 18

tSNE-tSNE1 −1 tSNEtSNE-1 −1

E F IRX2 Activity MAFA activityMAFA Activity IRX2 activity

● ● ● ● ● ● ● ●● ● ●● ● ● ●●●●● ● ● ●●●●● ● ●●●● ● ●●● ● ● ● ● ●●● ● ● ●● ●●●● ● ●● ● ●●●● ●● ● ● ● ● ● ● ●●●● ●● ● ●●● ●●●● ●●●●● ●● ● ●●●●● ●●●●●●●●●● ●● ● ●●●● ●●● ●● ●● ●● ●●●● ●●● ● ● ● ●● ● ● ● ●●●●●● ● ● ●●● ● ● ●●●● ●●● ●● ● ●●●● ● ● ● ●● ● ●●●●● ● ● ● ● ● ●●●● ● ●●● ● ● ● ● ● ●● ●●●●●●●● ● ● ● ● ●● ●●● ●●●● ● ● ● ●●●●●● ●●●● ●●● ● ● ● ●●● ●●●● ●●● ●● ● ● ●●●●● ● ● ● ●●● ● ● ● ●●●●●●●● ●●● ● ● ●●● ●● ●●●●● ●●● ● ● ● ●●●●●●● ●●●● ● ●● ● ●● ● ● ●● ●● ●●●●●●●●●●● ● ● ● ●● ● ●●●●●●●●●● ● ●● ● ● ●● ● ● ● ● ●●●●●●●● ● ●● ● ● ●● ● ● ●●● ●●●● ●●●● ● ● ● ● ●●●●●●●● ● ●● ● ●●●●●●●● ● ●●● ●● ●● ● ●●●●●●● ● ● ● ● ●●● ●●●● ● ●●●● ●●●● ● ●●● ● ●● ● ●●● ●●●●●●● ●● ● ●● ● ●● ● ●●●● ●● ● ● ●●● ●●●●●●● ●● ● ●●●● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●● ●● ● ● ● ● ●● ●● ●●●●●●●●●●●●●●● ●● ● ●● ● ● ● ●● ●● ●●●● ●●●●●● ●● ● ● ● ● ● ● ●●●●●● ●●●●●● ●●●●●● ● ● ● ● ● ● ● ●● ●● ●●●●●●●● ● ● ● ●● ● ● ●●●● ●●●●●●●● ● ●● ● ● ● ● ●●● ●●●●●●●●●●●●● ● ●●● ● ● ● ●● ●●●●●●●●●●●●●● ● ●●● ● ● ● ●● ● ● ● ●● ●●●●● ●●● ●●●●● ●●●●●●●●● ●●● ●● ● ● ● ● ● ● ●● ●●●● ●● ●●●●●●●●●●●● ●● ●● ● ●● ●● ●●●●● ●●● ●●●●● ●●● ●● ● ●● ● ● ●● ● ● ●●●●● ●● ●●●●●● ●● ● ●● ● ● ● ● ●● ● ●●●●●●●●●●● ●●● ● ●●● ● ● ● ●● ●●●●●●●●●●●●●● ● ●● ●● ● ●● ● ● ●●● ● ●●● ●● ● ●●● ● ● ●●●● ●● ● ● ● ●●●● ● ●● ●● ● ●●●● ● ●● ●●● ● ● ●● ●●●● ●● ● ●●●● ●●●● ● ● ● ● ● ●●● ● ● ●● ●●●● ●●●● ● ●● ● ● ● ● ●● ● ●●●●● ●●●●●●●● ●●● ● ● ● ● ●●●● ●●●●● ●●●●●●● ●●● ● ● ● ● ● ● ●●● ● ●● ●● ●● ●●●● ●●●●●●●● ● ●● ● ● ● ● ● ● ●●● ● ●●●● ●● ●● ●●● ●●●● ● ● ●● ● ● ● ● ● ●● ● ●●● ●●●●● ●● ●● ●●●● ●●● ● ●●● ● ● ●●●●● ●●●●● ●●●●● ●● ●●● ●●● ● ●● ● ● ●●●● ●● ●●● ● ●●●● ● ● ● ●● ●● ● ● ● ●●●●●●● ●●●● ●●● ●●●● ● ● ● ● ●● ● ●● ●●●●●●● ●●●●●●●●●●●● ●●●●●●●● ● ● ● ● ● ●●● ● ● ●●●●●●●●●●● ●● ●● ● ●● ● ● ● ● ● ● ●● ● ●● ● ●●●●●●●●●●● ● ●● ● ● ● ● ● ●●●●●●● ●●●●●●●●●●●● ● ●● ● ● ● ● ●● ● ● ●●●●●● ●● ●●● ● ●●●● ● ● ● ● ●● ●● ● ●● ● ● ●● ●● ●●● ● ●●●● ● ●●●●● ● ● ● ●● ● ● ● ● ● ●● ●●●●●● ●●●●●●● ●●●●●● ● ●● ● ● ● ●●●●● ●●●● ●● ●●●●●●● ●●●● ●●● ● ●● ● ● ● ● ● ● ●●●● ● ● ● ●● ● ● ●●● ●● ● ●●●● ●● ● ●● ● ● ● ●●● ● ● ●● ● ●●● ● ● ●●●●●● ●●● ●● ● ●● ● ●●●●● ● ●●●●● ●●● ● ● ●●● ●● ● ● ● ● ●●● ● ● ●●●● ●●● ●● ● ● ●●●●● ● ● ● ● ●● ● ●● ● ●● ● ●●●●●●● ● ●● ● ●●●● ● ● ● ● ● ● ●● ●● ●● ● ● ● ●●●●●●●●● ● ● ● ●● ● ● ● ●● ● ●●● ●● ●●●●●●● ●●●●●● ●● ●● ●● ● ●● ●●● ● ● ●●●● ●●● ●●●●● ● ●●● ●● ●● ●●● ● ●●● ● ●● ● ● ● ●●●●●● ●●●●●●●●● ● ●●●●●●●●● ● ● ● ● ● ●●●●●● ●●●●●●●●●● ● ●●●●●●●●● ● ● ●● ●● ● ● ● ● ●●● ●● ●● ●●● ●● ● ●● ●●●● ●● ●●● ● ● ● ● ● ● ● ● ● ●● ●● ●● ● ● ● ● ●●●● ● ●● ● ● ●●●●●● ●●●● ●●●●●●●●● ● ● ●●●● ●● ●● ●● ● ● ● ●● ●●● ●●●●● ● ●●●●●●● ●● ● ●●●●● ● ●● ● ● ● ● ●●● ●●● ●●●●●●●●● ●●●●● ● ●● ●● ● ●● ●● ● ● ●● ●● ●●● ●●●●●●●● ●●●●● ● ●●● ●● ● ● ●● ● ● ●●●●● ●● ● ●● ●● ● ●●●●●●● ● ●● ● ● ● ● ● ● ● ●●●● ● ●● ● ● ● ●●● ●● ● ● ● ●● ● ● ● ●●●● ● ●●●● ●●●●●●●●●●●●●●●●●● ●●● ●●●●● ●●●● ● ● ● ● ● ● ● ● ●● ● ● ●● ●●●●●●●● ●●●●● ● ●●●● ●●●●● ● ●●● ●● ● ●● ● ● ●● ●●● ● ●● ●●●●● ●● ● ●●●●● ● ● ● ● ● ●● ●●●●●●●● ●●●●●●●●●●● ● ●●● ● ●● ● ●● ● ● ● ● ●● ● ●● ●●●●●● ●●● ●●●●●● ●●●●●●●●● ●● ●● ● ● ● ● ●● ● ● ● ● ●●●●●●● ●●● ● ●● ●● ●●●●●●● ● ● ● ●● ● ●● ●● ●● ●● ●●●●●●● ●● ●●●●●● ●●● ● ●● ●●●● ● ● ● ●● ●● ●● ●●●●●●●●●●● ● ●●●●●●●●●●●●●● ●●● ● ●● ● ● ●●● ●● ●●● ●●●●●●●●●●●●●●●● ● ●●● ●●●●●● ●● ●● ● ● ●●●●● ●●●●●●●●● ● ●●●●●● ●● ●●● ●●●●● ●● ●●● ● ● ●● ● ● ●●●●● ●●●●●●●●●●●●●●● ●●● ●●●●●● ●●● ●● ● ●● ● ●●●●●● ●●●●●●●●●●●●● ● ●●●●●● ●● ●● ●● ● ● ●● ● ● ● ● ● ● ● ●●●●●●●●●●● ●●●● ●● ●●●●●●● ● ● ●●● ● ● ● ● ●● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●● ●●● ● ● ● ●● ●●●● ●●● ●●●●●●●●●●●●●●● ● ●● ●●●●●●●●●●● ●● ● ●● ●●● ●● ● ● ●● ● ●●●●● ●●●●●● ● ● ●●● ●●●● ● ●● ●● ● ● ●●●● ●● ● ●●● ●● ●●●●●●●●●●●●●●●●●● ●● ● ●●●●●●●●●●● ● ●● ●●● ●● ●● ●●●●● ●●● ● ●●●●●● ●●●● ● ● ●● ●●●●●●●●●●●●● ●●● ● ● ● ●●● ●● ● ●●● ● ●●●●●● ●●●●●●● ●●● ● ● ● ● ● ● ●● ●●●●●● ●●●●●●●●● ●● ●●●●● ● ●●●● ●● ● ● ● ● ● ● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ● ● ● ● ● ●● ●●● ● ● ●●●● ●●●●● ●●●● ●●●●●● ● ●●●● ● ● ● ● ● ● ● ● ● ●●● ●●●●●●●● ●●●●●●●●●●● ●● ●● ●●● ● ● ● ● ● ● ●● ● ●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●● ●●● ● ● ● ● ● ● ●● ●● ●● ●●●●●●●● ●● ●●●●●● ●●●●● ● ●● ● ● ● ● ●● ● ●● ● ●● ● ● ●● ●● ●●●● ●● ●●●●●●●●● ●● ●●● ● ●● ● ● ● ● ● ● ●●●● ●● ●●●●●● ●●●●●●●●●●● ●●●●●●●● ●● ●● ● ● ● ● ● ●● ●● ● ● ● ● ● ●●●●●● ●●●●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ●●●●● ● ● ● ● ● ●●●●●●●●●●●●●●●● ●●●●●● ●●●●●● ●●●●●●●●● ● ●● ● ●● ● ● ● ●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ●●● ●● ● ● ● ●●●● ● ●●●●●●●●●●●●● ●●●●●●●●●● ● ●● ●● ● ● ●●●●●● ● ● ●●●● ● ● ●●● ●●●●●●●●●● ●●●●●●●●●● ● ●●●● ●● ● ●●●●● ● ●● ●● ●●●●●●●● ●●●● ●●●●●●●●●● ●●●●● ●● ●● ●● ● ● ● ● ● ● ●● ●●●●●●●●●●●● ●●●●●●●●● ●●●● ●●●●●● ● ●● ● ● ● ●● ● ● ●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●● ● ●● ● ●● ● ● ● ●● ● ●●●●●● ●●●●●●●● ●●●●●●●●● ●●●●●● ●●●●●●●●● ● ●● ● ●● ● ● ●● ● ● ●● ●●●●● ●● ● ●●●● ● ●●●● ●● ●● ● ●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●● ●● ●●● ● ● ●● ●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●● ● ●● ● ● ● ● ●●●● ● ● ●●●●● ●●●●●●●●●●● ● ● ●● ● ● ● ● ●● ●●● ●●● ●● ●●● ●●●●● ●●●●●●●● ● ● ● ● ● ●● ● ● ●●●●●●● ●●● ● ●● ●●● ●●●●●●●●●●●●● ● ● ● ● ● ● ●● ●●● ● ●●●●●●●●●●●● ●●●●● ●●●●● ● ●● ●● ●●● ● ● ●●●●●●●●● ● ● ●● ●●●● ●●● ● ●● ● 2 ● ● ● ●●● ●●●● ● ●●● ●●●●● ●●●●●●●● ●● ● ●● ● ● ● ● ● ●●● ●● ●●● ● ●●●●●●●● ● ● ● ●●●●● ●●● ●●● ● ●● ● ● ● ● ● ● ● ●● ●● ●●●● ●● ●● ●●● ●● ●●●●● ● ●●●● ● ● ● ●●● ●●● ●● ●●●●●●●●●●● ●● ●● ●●●●● ● ●●●● ● ● ● ● ● ●● ●●●●● ●● ●●● ●●●● ●● ●●● ●● ● 2 ● ●● ● ● ●●●●●● ● ●●● ● ● ● ● ● ● ●● ● ●● ●●●● ●● ●●●● ●●● ● ● ●● ●● ● ● ●● ●●●● ● ●●● ●●● ●● ●●● ● ● ● ● - ● ● ●●● ● ●●● ● ●●● ● ● ● ● ● ● ● ● ●● ●●●● ● ●● ●●●●● ●●● ● ● ● ●●● ●●● ● ●● ● ● ● ●●● ●●●● ●● ●● ●● ● ●●● ●●● ●● ●● ●●●● ●●●●● ●● ● ● ● ●● ● ● ● ● ● ●●●● ●●●● ●●●● ●● ●● ● ● - ●●●●● ●●● ●●●● ●● ● ●●● ●●● ● ● ● ● ● ● ● ●●● ● ●●● ●● ●●●●● ● ●●● ● ● ● ● ●● ● ● ●●●● ● ●●● ●● ●●●●● ● ● ● ● ● ● ● ●● ● ●●● ●● ●●● ●●●● ●● ● ●● ● ● ● ● ●● ● ●● ●● ●●● ●●● ● ● ● ● ● ● ●● ●●●●● ●●●● ●● ●● ●●●● ●●●● ● ●● ● ● ● ● ● ●●● ●●●● ●●●●● ●● ●●●● ●●●● ● ●●● ●● ● ● ● ● ●● ●●●● ●● ● ● ● ●●● ● ●● ●● ● ●●● ● ● ● ●● ●●●●● ●● ●● ●● ●●●●● ● ●● ●● ● ● ● ● ●● ● ●● ●●● ● ● ●●● ● ●●●●● ●● ●● ● ●● ●● ● ●● ● ●● ● ● ●●● ●● ● ● ●●●● ●● ● ●● ●● ●● ●●●● ●● ●● ● ● ● ● ●● ● ●● ●● ● ● ●● ● ●●● ● ● ●● ● ● ● ●●● ● ●●● ●●●● ●●●● ●●●●● ●●● ●●● ● ● ● ● ● ●●● ●● ● ●●● ●●●● ● ●●● ● ●●● ● ●● ● ● ●●●● ● ● ●● ●●● ●● ● ●● ●● ● ●●● ● ●● ● ● ●● ● ● ● ●●● ●●● ●●● ● ●● ●● ●● ● ● ● ● ●● ●●●● ● ● ● ● ●●●●●●●●● ● ●● ● ● ● ● ●● ●● ●●● ● ● ● ● ●● ●●●●●● ●●● ●● ●● ● ● ● ●●●●●●● ●●●●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ●● ●● ●● ● ●● ● ●●● ●● ● ● ●●● ● ● ●● ● ● ●●●●●●●●●●●●● ●●● ● ● ●● ● ● ● ●●●●●●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●●● ● ● ●●●●●●●●●●●●● ● ●● ● ● ●● ● ● ● ● ● ● ●● ●● ● ●●● ●●●●●●●●●●●● ● ●● ● ● ● ● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ● ● ● ● ●● ● ● ● ● 2 ●● ● ● ● ●● ● ●●●●●● ●●● ● ● ● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ● ● ●● ●●●●● ● ●●●●● ● ● ●●●● ●●●●● ● ●● ●● 2 ●● ● ● ● ●● ● ●●●●●● ●●● ● ● ● ● ● ● ● ●● ●● ● ●● ● ● ●● ● ● ● ●● ●●●●● ● ●●●●● ● ● ●●●● ●●●●● ● ●● ●● ● ● ●● ●● ● ●● ● ● ● ● ●●● ●●●●●●●● ●●● ● ●● ● ● ● ● ● ● ●● ● ●● ●●●● ●●● ● ●●● ●●●●●●●● ●●●●● ●● ●● ● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ●● ● ● ● ● ● ● ● − ●● ● ● ●● ● ●●●●● ● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ● ● ● ●●●●●● ● ● ●●● ●●● ●●●●●●● ● ● ● ● ●● ● ● − ● ●●● ● ● ● ● ●● ● ●●●●●● ● ●● ● ● ● ● ● ●●●●●● ● ● ●●●● ● ●● ●●●● ● ●●● ●● ●● ● ● ● ● ●●●●●●● ●●●●●●●● ●●● ●●●● ●●● ● ●● ● ●● ● ●●● ●●●● ●● ●●●●●●●●● ●●●●●●● ●●●●●● ● ● ● ● ● ●● ● ●● ●●● ● ● ●● ●●●● ●●●●●● ● ● ●●● tSNE ● ●● ●● ●●●●● ● ●● ●●●●●●●●● ●●●● ●●● ●● ●●●●●● ● ● ●●●●● ●●● ●● ●●●●●●●●●●●●●●●●●●● ● ●●●● ●●● ● ●●● ● ●● ● ●●●●●● ●● ● ● ● ●●●● ●●●●●●● ● ● ● ● ● ●●● ● ●●●● ● ●● ● ● ● ●●● ●●●●● ● ●●● ● ● ● ● ● ●●● ● ●●●● ● ● ●●●●● ●● ●●●●●●●● ● ●●●●●●●●● ●● ● tSNE ● ● ● ●●● ●● ●●● ● ●●●●●●● ●●●●● ● ● ● ●● ●● ● ●● ● ● ●●● ●●●●●● ●●●●●●●●●●●● ● ● ● ● ● ● ● ●●●●● ●●●●●● ● ● ● ● ●●●●●●● ●●●●●●●●●●● ●●● ●●● ●● ● ● ● ● ●●●●●●● ●●●●● ●● ● ●●●●●● ● ●● ●●●● ●● ● ●●●●●●●● ● ●●●● ●●●●●●●●●● ●●●● ●● ●● ● ●● ●●●● ●● ●● ● ● ● ● ●●● ● ●●●● ● ●● ●● ● ● ●●●●●● ●●●● ●● ●● ●●●●● ● ● ●● ● ● ●● ●●● ●● ● ●●● ● ●● ●● ●● ●●● ●● ●●●●●● ●● ● ● ●●● ● ●●● ●●●●● ● ●● ●● ●●●● ●● ●● ●●●●●●● ●● ● ● ● ● ● ●●● ● ● ●● ● ● ●●●●●●●●● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●●●● ●● ● ● ● ● ●● ●● ●●●● ● ●● ● ● ● ● ●●●●●● ●●●● ● ● ● ● ●●●● ● ● ●● ● ●● ● ● ● ●●●●●●● ● ●● ●● ● ●●●●●● ● ●●● ● ● ● ●● ●●● ● ● ● ● ●● ●●●●●● ● ● ● ● ● ● ●● ●●●● ● ● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ●●●●● ● ●● ● ●● ● ● ●●● ● ● ● ● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ● ● tSNE ●● ● ● ●● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ●●●● ● ●● ● ● ●● ● ●● ●● ● ● ● tSNE ●● ● ● ●● ●● ●●● ● ●● ● ●● ● ●● ●● ●●● ● ●●●●● ●●●●●● ●● ●● ●● ●●● ●● ● ● ● ●●● ● ●●● ● ●●● ● ● ●●● ● ●●● ●●● ● ● ●●●● ●●● ●● ● ●● ●●● ●●● ●●● ●●● ● ● ● ● ● ● ●● ●●● ●● ●●● ● ● ● ● ● ● ● ●●●● ●● ●● ● ● ● ●● ● ● ● ● ● ● ●●● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ●●● ●●● ● ● ● ● ●●● ● ● ●●● ● ● ●●● ●● ●●● ●●●●●●● ● ●●● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●● ● ●● ● ●● ● ● ● ●● ● ●● ● ●● ● ● ●● ● ●● ● ● ● ●● ● ● ●● ●●●● ●● ●● ●●●●● ● ● ●● ●● ●●● ● ●●●● ●●●●● ● ●●● ● ● ● ● ● ● ● ● ● ● ●●●● ●● ●● ● ● ● ● ● ● ●● ● ● ● ●●●●● ●●● ●●● ●● ● ● ● ● ● ● ●●● ●● ●● ● ●●●● ● ● ● ● ● ●●● ● ●●●● ●●●●●●●●●●● ●●● ●●● ● ● ● ●●●● ● ● ●●●●●●● ●●●●● ● ●●●● ●● ●●● ●● ●● ● ● ●●● ●● ●● ●●● ● ● ●● ● ● ● ● ●● ● ●●● ● ● ●● ●●●●● ●● ●● ● ● ● ● ● ●●●●●●●●●● ●●● ● ● ● ●● ●●●● ●●●●● ● ● ● ● ●●●● ●●●● ●● ● ● ● ● ●● ● ● ● ●●●●●● ●● ● ●● ●●● ● ● ● ●● ●●●● ● ●● ●●● ● ● ●●●●●● ● ●● ● ● ● ● ●● ● ● ●● ● ● ● ●● ● ●●●● ● ●●●● ● ●● ●●● ●●● ● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ●● ●●●● ●● ● ● ● ●●●● ● ● ● ●● ●●●●● ●●● ● ●●● ●● ● ● ● ●●●● ● ● ● ● ● ● ●●● ● ●●●●●●●● ● ● ● ● ●● ● ● ● ● ● ●●●●●● ● ● ● ●● ●● ● ●●● ● ● ●● ● ● ● ● ●● ● ●●● ● ● ● ● ● ●● ●● ● ●● ●● ● ●● ● ● ● ●●●● ● ● ● ●● ● ●● ●● ● ●●●● ● ● ●● ●● ●● ● ● ● ●●● ●● ● ●●● ●● ●● ●● ● ●● ● ● ● ● ● ● ●● ● ● ●●● ● ● ●● ● ● ● ● ●● ●● ●● ● ● ●● ● ● ● ● ● ● ●●●●●● ● ●●●● ● ● ●●● ● ● ● ● ● ● ● ●●●● ● ●●● ● ● ●● ●● ● ● ● ● ● ●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ●●● ● ● ● ● ● ●● ●● ● ●●●● ● ●● ● ● ● ● ● ●● ● ●●●●● ● ● ● ● ● ● ● ● ● ● ●● ●● ●●● ●● ●● ● ● ● ● ● ●● ● ●● ●●● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ●●●● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●●●● ●● ●●●● ● ● ● ● ● ● ● ● ● ● ●●● ●● ● ●●●●●● ●●●●● ● ● ● ●● ● ●●● ●● ●●●●●●● ●●● ● ● ● ● ● ●● ● ●● ● ● ●●●● ●●●● ● ● ● ● ● ● ●● ● ●●●●●●●●● ● ● ●● ● ● ●●●● ●●●● ●● ● ●● ●● ●●●●●● ● ● ● ●● ●●●●●●●●●● ● ● ● ● ●●●●● ●● ●●●●● ● ● ● ●● ● ●●●●●●●● ●● ● ● ●●● ● ● ● ● ●●●●●●●● ●●● ● ● ●●● ● ● ●●● ● ●●●●● ●●● ●● ● ● ● ●● ●●●● ● ●●● ●● ● ● ●● ● ●●●●● ● ●● ●● ● ● ● ● ● ●● ●● ●●●●●● ●●● ●● ●●● ● ● ● ●● ●●●● ● ●●● ● ●● ● ● ●●● ●● ●●●●●●●●● ● ● ● ● ●●●● ●● ● ●● ●●●● ● ● ● ● ● ●● ● ●●●● ●● ● ●●● ●● ● ● ●● ● ● ● ● ● ●●● ●● ●●● ● ● ● ● ●● ● ●●● ●● ●● ●● ● ●● ●● ●●●● ● ●●●● ● ●●● ● ● ● ●● ● ● ● ● ● ●●●●● ● ●● ● ●● ● ● ● ● ● ●●●●●●●●● ●●●●●● ● ● ● ● ● ● ● ●●●● ●●●●● ●●●●●●●● ● ● ● ●● ● ●● ●● ● ●●● ● ●●● ● ● ●●● ● ●●● ●● ● ● ●● ●● ● ● ●●●● ●● ● ● ●● ●● ● ●● ●●● ●● ● ● ● ●● ● ●●●● ● ● ● ● ● ● ● ●●●● ●●● ●● ●● ● ● ●● ● ●● ● ● ● ●● ●● ● ●● ● ●● ●● ●● ●● ●● ● ●●●● ● ●● ● ● ● ● ●● ● ●● ●●●●● ● ● ●● ● ● ●● ●● ●●● ●● ● ● ●●● ● ● ● ● ● ● ●● ●●● ●●●●●●● ●● ● ● ● ●● ●●●●●●● ● ●●●● ●● ●●●● ●●●●● ●● ● ●●● ●●● ●●● ● ●● ●●● ●● ●●● ●●●●● ● ●●●● ● ●● ●●●●●●●● ●●●●●●●●●●●●●●●● ● ● ● ● ●●● ● ● ●●●●●●●● ●● ●● ● ●● ●●●● ●●● ●●●●●●●●●●● ● ● ● ● ● ●●●●●●●● ● ●●● ● ●● ● ●●● ● ●●● ●●● ●●●●● ● ● ●●● ● ● ●●● ●● ● ● ●●●●● ● ● ● ● ●●● ● ●● ●●●●●●●●●●●●●●● ● ● ●●● ●● ● ●●● ● ●●● ● ●●●● ● ● ● ●● ● ● ● ● ●●● ● ●●● ●● ●●● ● ● ●● ●●●● ● ●● ●● ● ●● ●● ● ● ●● ●●●● ● ●●● ●●●● ● ●●● ●●●● ● ●●●● ● ●● ●● ● ● ●●●●●● ●● ●● ● ● ●●● ●● ● ● ● ● ●●● ●●●●●● ● ● ● ● ●● ●● ●● ●● ●● ● ● ●● ●●●●● ● ●● ● ●● ●●●●● ● ● ● ●● ●●●●● ●●● ●● ●●●●●● ● ●●● ●●● ● ●● ● ●●●●●● ●●● ●●●● ●●●●● ● ●●●● Activity ● ●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ●● ● ● ●● ●●●●● ●●● ●● ●●●●●● ● ●●● ●●● ● ●● ● ●●●●●● ●● ●●●● ●●●●● ● ●●●● ●●●●●●●●●● ●●● ●●●● ●●●●● ●●●●●● ● ● ●● ●● ●●●●● ●●●● ● ●●●●●● ● ● ● ●●●●●● ● ● ● ● ●●●● ●●● ● ● ●●●● ●●● ●●● ●●● ●●●●●●●● ●●● ●●●●● ● ● ●● ●●●●● ●● ●● ●● ● ●●●●● ●●●●●●●●●●●●● ● ●● ●●●●●●●●●● ● ●●●●● ● ● ● ●●● ●●●●●●●● ● ●●●●● ● ●● ●● ● ●●● ●●● ●●● ● ●●●●●●●●●●●●●● ●●●●●●●● ● ● ● ●● ●●● ● ●● ●● ● ●● ● ● ●●●●●●●●● ●●●●●●●●●●● ●●● ● ● ●● ● ● ●●●● ● ●● ●●●● ●●● ●●●● ●● ●●● ●● ●●●●●● ●●● ● ● ●● ●● ●● ●●● ●● ● ●●●●● ● ●● ● ● ● ●●●●●●●●● ● ●● ● ●●●●● ●●●● ●●●● ●● ● ● ● ●● ●●● ●● ●●●● ●● ● ● ●●● ● ● ● ●●●●●●●● ●●●●●●● ●●●● ●●● ●● ● ● ● ● ●●●●● ● ● ●● ●●● ● ●● ●● ● ●● ●●● ●●●●● ●●●● ●●● ● ●●●●●●●●●● ● ●●● ● ● Activity (Z−Score) ●●●●● ● ●● ●●●●● ●●● ●●● ●● ●●●●●●●● ●●●● ●●●●●●●●●●●●●●● ● ●● ● ● ●●●●●●●●●●● ●● ● ●● ● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●● ●●● ●●●● ●● ●●● ● ● ●●●●●●●● ● ●● ● ● ●● ● ●● ● ● ● ●● ●● ●●●●●● ●●●●●●●●●●●●●●● ●●●●●● ● ●●● ● ● ● ●●●●●● ●● ●● ● ●●●●● ●● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●● ● ● ● ● ●●●●●●●● ● ●●●● ●●● ●● ● ●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●● ● ● ● ● ●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●● ●●● ●● ● ●●●●●●●●●●●●●●●● ●●●●●●● ●●●● ●●●●●●●● ● ● (Z-Score) ●●● ● ● ● ●● ●● ● ● ●● ●●● ● ●●● ● ● ● ● ● ● ●● ● ●● ● ●● ● ● ●●● ●●●● ● ● ● ●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●● ●●● ●● ●●●●●●●●●●●●●●●●● ●●●●●● ●●●● ●●●●●●●● ● ● ● ●●●●●●●●●●●●● ●● ●●●●●●●● ●●● ● ●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●● ●●● ●●●●●●●●● ● ● ● ●●●●●●● ●●●● ●●● ●●●●●● ● ●● ●●●● ●●●●●●●●●●●●●●●●●●●● ●●● ● ●●●●● ● ● ● ● ● ●●●●●●●●●●●●●●● ●●● ● ● ● ●●● ●●●●●●●●●●●●●●●●●●●●●● ●● ● ●●●● ● ● ●●●● ● ●● ● ●●●●●●●●●●●●●● ●●●●●● ●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●● ●● ●● ● ●●●●● ● ● ● ●●●●●●●●●●●●●●●●●● ●●● ●● ●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●● ●●● ● ● ●● ● ● ●● ●●●●●●●●●●●●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●● ●● ●●●●●●● ●●●● ● ●● ●● ●●●● ●●●● ● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●● ●● ● ● ● ● ● ●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ●●●● ● ● ● ●●●●● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ●●● ● ●● ● ● ● ●● ● ●●●●● ●●●● ● ●●●●●● ●●●●●●●●●●●●●●●●●●●● ● ● ● ●●● ● ● ● ● ● ● ● ● ●●●●●●●●●●● ●●● ● ●●● ● ●● ●● ●● ● ● ●●● ● ●●●●●●●●●●●●●● ●●●●●●●● ●●● ●●●●● ● ● ● ● ● ● ●● ●●●● ●●●●●●●●●●● ● ● ●● ●●●● ● ● ●● ● ●● ● ●● ● ●●●●●●●●●●● ● ●●● ●●● ●●● ● ● ●●●●●●●●●●●● ●●●●●●● ●● ●● ●● ●● ●●●●● ●●●●●●●●●●●●●●● ●●●●●●●● ● ● ●● ● ● ● ●●● ●● ●●● ● ● ● ●● ● ●● ●●● ●●● ●●●●●● ●●● ●● ●● ● ● ● ● ● ● ● ●● ● ●● ● ●● ● ● ●● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ● ●●● ● ● ●●● ●● ● ● ● ●● ● ● ●● ●● ●● ● ● ●● ● ●●● ●● ●● ● ●● ● ●●●●●● ●●●●●● ●● ● ●●● ●● ●● ● ●●● ●●●●● ●●●●● ● ● ● ● ● ● ●●● ●●●● ● ●● ●● ● ●● ● ● ● ● ●● ●●●● ●●●● ●● ●● ●● ● ● ●● ●● ●● ●●● ● −2 0 2 ●● ● ● ● ● ●● ●●● ● ● ● ●● ●● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ●● ● ●●● ● ● ● ● ● ● ●●● ●●●● ● ● ●●● ● ●●● ●● ● ●●● ● ●● ●●● ● ● ● ●● ● ● ● ●● ● ●● ● ●● ● ● ● ● ●● ●●●● ● ● ● ●● ●● -2 0 2 ● ● ● ● ●● ●● ● ● ● ●

Expression (log2(RPM+1))

tSNE-1tSNE−1 tSNE-tSNE1 −1 0 9 18 Figure 1. metaVIPER analysis reduces donor-to-donor variability (A) Single cells from human ND and T2D islets were projected onto 2-D t-SNE space based on whole mRNA expression.Each dot represents a single cell, color-coded according to donor. (B) t-SNE clustering as in (A) but based on protein activity inferred from islet specific regulatory networks. (C) INSULIN and (D) GLUCAGON mRNA expression plotted in t-SNE at the single cell level. (E) Computationally-inferred MAFA and (F) IRX protein activity plotted in t-SNE at a single-cell level. PPARA Activity PPARG Activity A PPARA activity B PPARG activity Fig.2

● ● ● ● ● ● ●● ● ● ●● ● ●●●●● ●●● ● ●●● ● ● ● ●●● ● ●● ●●●●● ● ● ●●●● ●● ● ● ● ●● ●● ● ● ● ● ● ●●●● ●● ● ●● ● ●●●●● ●●●●●●●●●● ● ● ●●●●●●● ●● ● ●● ● ●●● ● ● ●●●●● ●● ●● ●●●● ●●●●● ● ● ●●●● ●●● ●● ●●● ●● ● ●●● ●●● ●● ●● ●●●● ● ● ● ●●●● ● ●●● ● ● ● ● ● ● ●●●●● ● ● ●● ● ● ●● ●●● ●●●● ● ● ● ● ● ● ●● ● ●●●● ● ● ● ● ●●● ●●●● ●●● ●● ● ●● ●●●●●●●●●● ● ● ● ● ●●●●●●●● ●● ● ●●●●●●●● ●● ●●● ● ● ● ●●●●●●● ●●●● ● ●● ● ●● ● ● ●●●● ● ● ● ●●● ● ● ●● ● ●●●●●●●●●● ● ●● ● ● ●●● ●● ●●●●● ●●● ● ●● ● ● ●●● ●●●● ●●●● ● ● ● ● ● ●● ●● ●●●●●●●●●●● ● ● ● ●●●●●●● ● ● ● ● ●●●●●●●● ● ●●●● ●●●● ● ● ●● ● ● ● ● ●●●●●●●● ● ●● ● ● ● ●●●● ●● ● ● ●●● ●●●●●●● ●● ● ●●●● ● ● ●●●●●●●● ● ●● ● ●●●●●●●● ● ●●● ●● ●● ●● ● ● ● ● ●● ●● ●●●●●●●●●●●●●●● ●● ● ●● ● ●●● ● ●● ● ●●● ●●●●●●● ●● ● ●● ● ●● ● ● ● ● ●●●●●● ●●●●●● ●●●●●● ● ● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●●●●●●●●●● ●● ●● ● ● ●●●● ●●●●●●●● ● ●● ● ● ● ● ●● ●● ●●●●●●●●●●●● ●● ● ● ● ● ●● ●●●●●●●●●●●●●● ● ●●● ● ● ● ● ● ●● ●● ●●●●●●●● ● ● ● ● ● ● ● ● ●● ●●●● ●● ●●●●●●●●●●●● ●● ●● ● ● ●●● ●●●●●●●●●●●●● ● ●●● ● ● ●● ● ● ●● ● ● ●●●●● ●● ●●●●●● ●● ● ●● ● ● ● ●● ● ● ● ●● ●●●●● ●●● ●●●●● ●●●●●●●●● ●●● ●● ● ● ● ●● ●●●●●●●●●●●●●● ● ●● ●● ● ● ●● ●● ●●●●●● ●●● ●●●●● ●●● ●● ●● ● ● ● ●●●● ● ●● ●● ● ●●●●● ●● ●●● ● ● ● ●● ● ●●●●●●●●●● ●●● ● ●●● ● ● ● ●●● ● ● ●● ●●● ●●●● ● ●● ● ●● ● ● ●●● ● ●●● ●● ● ●●● ● ● ●●●● ● ● ●●●● ●●●●● ● ● ●● ●● ● ● ● ● ● ●● ●●●● ●● ● ●●●● ●●●● ● ● ● ● ● ● ● ●●● ● ●●●● ●● ●●●●●●● ●●●●● ● ● ●● ● ● ● ● ● ●● ● ●●●●● ●●●●●●●● ●●● ● ● ● ● ● ●●●●● ●●●●● ●●●●● ●● ●●● ●●● ● ●● ● ● ● ●●● ● ●● ●● ●● ●●●● ●●●●●● ● ●● ● ● ● ● ●●●●●●● ●●●● ●●● ●●●● ● ● ● ● ●● ● ● ● ●● ● ●●● ●●●●● ● ●● ● ●●●● ●●● ● ●●● ● ● ●●● ● ● ●●●●●●●●●●● ●● ●● ● ●● ● ● ● ● ●●●● ●● ●●● ●●● ●●●● ● ● ● ●● ●● ● ● ● ● ●●●●●●● ●●●●●●●●●●●● ● ●● ● ● ●● ●●●●●● ●●●●●●●●●●●● ●●●●●●● ● ● ● ●● ● ● ●● ●● ●●● ● ●●●● ● ●●●●● ● ● ● ●● ● ● ● ● ● ● ●● ● ●● ● ●●●●●●●●●● ● ●● ● ● ● ● ●●●●● ●●●● ●● ●●●●●●● ●●●● ●●● ● ●● ● ● ● ● ● ●● ● ●● ●●●●●● ●● ●●●● ●●●●●● ● ● ● ● ●● ● ● ● ● ● ●●● ● ● ●● ● ●●● ● ● ●●●●●● ●●● ●● ● ●● ● ● ● ●● ●● ●●● ●●●●●●●● ●●●●●● ● ●● ● ● ● ● ●●● ● ● ●●●● ●●● ●● ● ● ●●●●● ● ● ● ● ● ● ●●●● ● ● ● ●● ● ●● ●●● ●● ●● ●●●● ●● ● ●● ● ● ● ●● ●● ●● ● ● ● ●●●●●●●●● ● ● ● ●● ● ● ●●●●● ● ●●●●● ●●● ● ● ●●● ●● ● ● ● ● ●●●● ●●● ●●●●● ● ●●● ●● ●● ●●● ● ●●● ● ●● ● ●● ● ●● ●● ●● ● ● ●●●●●●● ● ●● ● ●● ● ● ● ● ● ● ●●●●●● ●●●●●●●●●● ● ●●●●●●●● ● ● ●● ● ●● ● ●●● ●● ●●●●●●● ●●●●● ●● ●● ●● ● ●●● ●●● ● ● ● ● ● ● ●● ●● ●● ● ● ● ● ●●●● ● ● ●● ● ● ● ● ●●●●●● ●●●●●●●●● ● ●●●●●●●●● ● ● ● ● ●● ●●● ●●●●● ● ●●●●●●● ●● ● ●●●●● ● ●● ● ● ● ● ● ● ● ● ● ● ●●● ●● ●●● ●● ● ●● ●●●● ●● ●●● ● ● ● ● ●● ●● ●●● ●●●●●●●● ●●●● ●● ●● ● ● ●● ●● ● ●● ●● ●●●● ● ●●●●●● ● ● ●●●●● ● ●● ● ● ● ● ● ● ●●●● ● ●● ●●● ● ● ● ●● ● ●● ● ●● ● ● ●●● ●● ●●● ●●●●●●●●● ●●●●● ● ●●● ●● ● ● ●● ● ● ● ● ●● ● ● ●● ●●●●●●●● ● ●● ● ●●●● ●●●●●● ● ●●● ●● ● ●● ● ● ●●●● ●● ● ●● ● ● ● ●●●●●● ● ● ● ● ● ●● ●●●●●●●● ●●●●●●●●●●●● ● ●●● ● ●● ● ●● ● ● ● ● ●● ● ● ●●● ●●●●●●●●● ●●●●●●●● ●●● ●●●●● ●●●●●●● ●● ● ● ●● ● ● ●● ● ● ● ●● ●●●●●●● ●●● ● ●● ●● ●●●●● ● ● ●● ● ●● ●●●●●● ● ●●●●●●●●● ●● ● ●●● ● ● ● ● ● ●● ●● ● ●●●●● ●●●●● ● ●●●●●●●●●●●●●●●● ●●● ● ●● ● ● ● ● ●● ● ● ● ●●●●●●● ●●● ●●●●● ●●●●●●●●● ●● ● ● ● ● ●●●●● ●●●●●●●●● ●●●●●●●● ●●● ●●● ●●●●● ●● ●●● ● ● ●● ●● ●● ●● ●●●●●●●●●● ● ●●●●●●●● ●● ●● ● ● ●● ● ●● ● ●●●●●● ●●●●●●●●●●●●●● ● ●●●●●●● ●● ●● ●● ● ● ● ●●●●● ●● ●●● ●●● ●●●●●●●● ●● ●●● ●●●●●● ●● ●● ● ● ●● ● ● ● ● ●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●● ●●● ● ● ● ●● ● ●●●●●● ●●●●●●●●●●●●●●●● ●●● ● ●●● ●●● ●● ● ● ●● ● ● ●● ● ●●●●● ●●●●●● ●● ●●● ●●●● ● ●● ●● ● ● ●●●● ●● ● ● ● ● ● ● ● ●●●●●●●●● ●●●●●●●● ●●● ●●●●● ● ● ● ● ●● ●● ●●●●● ●●● ● ●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●●● ●●● ●● ● ●● ●●●●● ●●●●●●● ● ● ●●● ●●●● ●●●● ●● ● ● ●●●● ● ● ● ●●● ●●●●●● ●●●●●●●●● ●● ●●●●● ●●●●●● ●● ● ● ●● ● ●●●●● ●●●●●●●●●●●● ●●●● ● ● ●● ●●●●●●●●●●●●● ● ● ●●●● ● ● ● ●● ●●● ● ● ●●●● ●●●● ●●●● ●●●●●● ● ● ●● ● ● ● ● ● ● ● ●● ●●● ● ●●●●●●●● ●●● ● ●●●●● ●●● ● ● ● ● ● ●● ● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●● ●●● ● ● ● ● ● ●● ●●● ●●● ●●●●● ●●●●●●●●●●●●●●●●●●●● ● ●●●●● ● ● ● ● ● ● ●● ● ● ●● ●● ●●●● ●● ●●●●●●●● ●● ●●● ● ●● ● ● ● ● ● ● ● ●●● ●● ●●●●●●● ●●●●●●●●●● ●● ●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ●●●●●● ●●●●●●●●●●● ●●●●●●●●●● ● ● ●● ● ● ● ●●●●● ● ●● ● ●● ● ●●● ●●●●●●● ●●● ●●●●●●● ● ●● ●●● ● ●● ● ● ● ● ●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ●●● ●● ● ● ●●●● ●●●●●●●●● ●●●●●●●●●● ●●●●●●●●● ● ●● ●● ● ● ● ● ●●●●● ● ● ●●●● ● ● ●●● ●●●●●●●●●● ●●●●●●●●●● ● ●●●● ●● ● ●●●●● ● ● ● ● ● ●●●●●●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●● ● ● ●● ● ●● ● ● ● ●● ●●●●●●●●●●●● ●●●●●●●●● ●●●● ●●●●●● ● ●● ● ● ● ●●●● ● ●●● ●● ●●●●● ●●●●●●●●●● ● ●●●● ●●● ● ●●●●● ● ● ● ●● ● ●●●●●● ●●●●●●●● ●●●●●●●●● ●●●●●● ●●●●●●●●● ● ●● ● ●● ● ● ●● ● ● ●●●●●●●●●●●●● ●●●●●●●●● ●●●●● ●●●●●● ● ● ● ● ● ●●●● ●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●● ●● ●●● ● ● ● ● ●● ● ● ●●● ●●●● ●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ●● ●●● ●● ● ● ● ● ●●●● ● ● ●●●●● ●●●●●●●●●●● ● ● ●● ● ● ●● ●● ●●●● ●●●●● ●●● ●●●●●● ● ●●●●●●● ●● ● ●● ● ● ●● ● ● ●●●●●●● ●●● ● ●● ●●● ●●●●●●●●●●●●● ● ● ● ● ● ●● ●●●●●●●●●●● ●●●●● ●●●●●●●●●●●● ● ●● ● ● ● ● ●●●●●●●●●● ● ● ●● ●●●● ●●● ● ● ●●● ● ● ● ● ●●● ●●● ● ●● ●●● ●●●●●● ●●●●●●● ● ● ● ● ● ● ● ●●●● ●● ●●● ●●● ●●●●● ● ● ●●●●●●●●● ●● ● ● ●● ● ● ●● ● ● ●●●●●●●●●● ● ● ●●● ●●●● ● ● ● ●●●● ● ● ● ● ● ●●● ●● ●●●● ●● ●● ●●● ●● ●●●●● ● ●●● ● ● ●● ● ● ●●●● ●●●●●●● ●●● ●●●●●● ●●● ●●●●●●●●● ●● ● ● ●● ● ● ● ● ● ● ●● ●●●●● ●●●●● ●●●● ● ●●● ●● ● ● ● ●● ●● ●●●● ●● ●●● ●●● ●● ●●●●● ● ●●●● ● ● ● ●● ● ●● ●●●● ●● ●●●●● ●●●● ● ● ●● ● ● ●● ●● ●●●●● ●● ●●● ●●●● ●● ●●● ●● ● ● ● ●●● ● ●●● ●● ●●●● ●●● ●●●● ● ● ● 2 ● ● ● ● ●● ●●●● ●● ●●●●● ●●● ● ● ●● ● ●● ●● ● ●●●● ● ●● ● ●● ●●● ●● ● ● ●●● ●● ●●● ●● ●●●●● ●●● ●●● ● ● ● ● ● ● ● ● ● ●● ●● ●●●● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ●● ● ● ●● ● ● ●● ● ● 2 ● ● ● ● ● ●●● ● ●● ● ● ● ● ● ●● ●●● ●●●● ●● ●● ● ● ● ● ● ● ● ●● ● ● ●● ●● ●●● ●●●● ● ● ●● ● - ● ● ● ● ● ● ●●● ● ●●● ● ●●●● ● ●●● ● ● ●● ●●●● ●●●● ●●● ●●●● ●●●● ● ●●● ● ● ● ● ● ● ● ●● ● ●●● ●● ●●● ●●●● ●● ● ●● ● ● ● ●● ●●●●● ●● ● ●● ●●●●● ● ●● ●● ● ●●● ● ● ● ● ● ● ● - ● ● ●●●● ●●● ●● ●●● ●● ● ●●● ●●●●● ●● ● ● ● ●● ● ●● ●●● ● ● ●●● ● ●●●●● ●● ●● ● ●● ● ● ● ●●●● ●● ● ● ● ●●●● ● ● ●● ●●●● ● ●● ● ●● ●● ●● ●●●● ●● ●● ● ● ● ●● ● ●● ●●● ● ● ●●● ● ●●●● ● ●● ● ●● ● ● ● ●●● ● ●●● ●●●● ●●●● ●●●●● ●●● ●●● ● ● ● ●●● ● ●● ●● ●● ●●●● ●● ●● ● ● ● ● ●●●● ● ● ●● ●●● ●● ● ●● ●● ● ●●● ● ●● ● ● ● ●●● ● ●●● ●●●● ●●●● ●●●●● ●●● ●●● ● ● ●● ● ● ● ● ●● ●●●● ● ● ● ● ●●●●●●●●● ● ●● ● ●●●● ● ● ●● ●●● ●● ● ●● ●●● ●●●● ● ● ● ● ● ●●●●●●● ●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●●●● ● ● ● ● ●●●●●●●● ● ● ●● ●● ● ● ●●●●●●●●●●●●● ●●●● ● ● ● ● ● ● ● ●●●●● ●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●● ● ● ●●●●●●●●●●●●● ● ●● ● ● ●● ● ● ● ● ●●●●● ●●● ●●●● ●●● ● ● ● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ● ● ● ● ● ● ●●●●● ● ● ●●●●●●●●● ●●● ●● ● ● ●● ● 2 ●● ● ● ● ●●●● ●●●● ● ●●●●●● ● ●●● ● ● ●● ●● ● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ● ● ● ●● ● ●● ●●●●●●●●●● ●●● ●●●●●●● ●●●● ●●● ●●● 2 ●● ● ● ● ●●●● ●●● ● ●●●●●● ● ●●● ● ● ● ●● ● ● ●● ●● ● ● ● ● ● ●●● ●●● ●● ●● ● ● ● ● ● ● ●● ● ● ●●●●●●●●●● ●●● ● ● ●● ●●●●●● ● ●●● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ● ●● ●● ● ● ● ● ● ● ●●● ●●●●●●●● ●●● ● ● ● ● − ● ● ●●● ● ● ● ● ●● ● ●●●●●●● ● ●●● ● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ●● ● ● ● ●●●●●●●● ●●●●●●●●●●●● ●●●●●●●● ● ●●● ● ● ●● − ● ● ●●● ● ● ● ● ●● ● ●●●●●● ● ●●● ● ● ● ●● ●●●● ●●●● ●●● ●●●●●● ●●●●●●● ●●●●●●●● ● ● ● ●● ● ● ● ●●●●●●●● ●●●●●●●●●●●● ●●●●●●●● ● ●●● ● ● ● ● ●● ●● ●●●●●● ● ● ●●● ●●●●●●●●●●● ● ●●●● ●●● ●● ●●●●●● ● ● ● ● ●● ●●●● ●●●● ●●● ●●●●●● ●●●●●●● ●●●●●●●● ● ● ● ● ●● ● ●●●●●● ●● ● ● ●●●● ●●●●●●● ● ● ● ● ●● ●● ●●●●●● ● ● ●●●● ●●●●●●●●●●● ●●●●● ●●● ●● ●●●●●● ● ● ●● ● ●●●● ● ● ●●●●● ●●●●●●●●●●● ● ●●●●●●●●● ●● ● ●● ● ●●●●●● ●● ●● ●●● ●●●● ●●●●● ● ● ● ● ●●●●● ● ●● ● ● ●●●● ●●●●●● ●●●●●●●●●●●●●● ● ● ● ● ● ● ● ●●● ● ●●●●● ● ● ●●● ● ●● ●●● ●●● ● ●●●●●●●●● ●● ● ● ● ●●●●●●● ●●●●● ● ● ●●●●●● ● ●● ●● ● ●● ●● tSNE ●●●●● ● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ● ● ● ● ● ● ●● ●●●● ● ● ●● ● ● ● ●●● ● ●●●● ● ●● ● ●●●●●● ●●●●● ● ● ●●●●●● ● ●● ●● ● ●● ●● ● ●●● ● ● ●● ●● ●●● ●●● ●●●●●●●●● ● ● ●●● ● ●● ●●●● ● ● ●● ●●● ● ●●●● ● ●●●● ● ●● tSNE ● ● ●●● ● ● ●● ● ● ●●●●●●●● ● ● ● ●●● ● ● ● ●● ●●● ●●● ●●●●●●●●● ● ● ●●● ● ●● ●● ●●●● ● ●● ● ● ● ● ●●●●●● ●●●● ● ● ● ● ●● ● ●●●● ● ● ●●●●●●●●●●● ● ● ● ●●●●●● ● ●●● ● ● ● ●● ●●● ● ● ● ● ●●● ●●● ●●●● ● ●● ●● ● ●●●● ●●●● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ●●●●●● ● ● ●● ● ● ●● ●●● ● ● ● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ●●● ●● ● ● ● ●● ● ● ● tSNE ●●● ●● ● ● ● ●● ●● ●●● ● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ●● ● ●● ●● ● ●●● ●●● ● ● ●●● ● tSNE ●● ● ●● ●● ● ●● ● ●● ● ● ● ●●●● ● ●● ● ● ●● ●● ●● ● ● ● ● ●●● ● ●●● ● ● ●● ● ●●●● ●●● ● ● ● ●●● ● ●●● ● ●● ●●● ●●●● ● ●●●●● ● ● ● ● ● ●● ● ●● ●●● ●●● ● ● ●● ● ● ●● ● ● ●● ● ●● ● ●●● ● ● ● ● ● ●●● ● ●● ● ● ● ● ● ●● ●● ● ●● ● ● ● ● ● ● ●● ● ●●● ●●● ●●● ● ● ●● ●●● ●●● ●● ●●● ●● ●●● ●●●● ●● ●●● ● ● ●●● ● ●● ●● ● ● ● ● ● ●● ● ● ●● ● ●● ● ● ● ● ● ●● ● ● ●●●● ● ● ●● ●●● ●●● ●● ●●● ●● ●●● ●●●● ●● ●● ●●●●● ● ● ●● ●● ●● ● ●●●● ●●● ● ●● ● ● ● ● ●● ● ●●●● ● ●● ●●● ● ● ● ● ● ●●●●● ● ●●●●●●●●●●●●● ●●●●● ● ● ●● ●● ●● ● ●●●● ●●● ● ●●● ●● ● ● ● ● ●● ●●●●●● ●●●● ● ●● ●●● ● ● ● ● ● ●●●●● ●● ● ●●●●●●●●●●●●● ● ●●● ● ●●● ●●● ● ● ● ●●● ●●●●● ●● ●●● ● ● ● ● ● ●● ● ●●● ●●● ● ● ●●● ●●●●●● ●●● ● ● ● ● ● ● ●●●●●●● ● ●● ● ● ●●● ● ●●● ●●● ●● ● ● ● ●●●● ●●●● ●● ● ●●●● ● ● ●●●● ●● ● ● ● ●●● ● ●●● ●●●●●● ●●● ● ● ● ● ●●●●●●● ● ●● ● ● ● ● ● ●● ●● ● ●● ● ● ●●●●● ●●●● ● ●●●● ● ● ●●●● ●● ● ● ● ●● ● ●●●●●● ● ● ● ● ● ●● ●●● ●●●● ●● ● ● ● ● ●● ●● ● ●● ● ● ● ●●●●● ●●●● ● ● ●● ● ● ●●●● ● ● ● ● ●●● ● ●●●●●● ● ● ● ● ● ● ●●● ●●●● ●● ● ● ● ●● ● ●●● ● ● ●● ● ● ●●● ● ● ● ●● ●●● ●●● ● ● ● ● ●● ●● ● ●● ●● ● ● ● ● ●● ● ●● ●●● ●● ●● ●●●● ● ● ● ●● ●● ● ●● ●●● ● ● ●● ● ●● ●● ● ● ●● ●● ● ● ● ●● ● ● ● ● ●●●● ● ● ●● ●● ● ● ●● ●● ● ●● ●● ●●●● ● ● ● ● ●● ● ●● ● ● ● ● ● ●● ● ●●●● ● ●●● ● ● ● ● ● ● ●● ● ● ● ●● ●● ●● ● ●● ● ● ● ●●●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ●● ●●●● ● ● ● ● ● ● ● ● ●●●● ●● ● ● ● ●● ● ● ● ● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ●● ●●●●●●● ●●●●●● ● ● ● ● ● ● ● ●●●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●●● ● ●●● ● ● ● ● ● ● ● ● ● ●● ●●●●●●● ●●●●●● ● ● ● ● ● ● ●● ●● ●● ●●●●●● ●●●●● ● ● ● ● ● ● ● ● ●● ●●●● ● ●●● ● ● ● ● ● ● ●● ●●● ● ●●●●● ●●●● ● ● ● ● ●● ●● ●● ●●●●● ●●●●● ● ● ● ● ● ● ●●● ● ●●●● ● ● ●● ●●● ● ●●●●● ●●●● ● ● ●● ●●●●●●●●●●●● ● ● ● ● ● ●●● ● ●●●● ● ● ● ●● ● ●●●●●●●● ● ● ●●● ● ● ●●●●●●●●●●●● ● ● ● ●● ●●● ● ●●●●● ●●● ●● ● ● ● ● ●● ● ●●●●●●●● ● ● ●●● ● ● ● ● ●●●●●● ●● ●● ● ● ● ● ●● ●●● ● ●●●●● ●●● ●● ● ● ● ● ●● ●●● ●● ●● ●●●● ● ● ● ● ● ●●●●●●●●● ●● ● ● ● ● ●●●● ● ● ●●● ● ● ● ● ● ●● ●●● ● ●●● ●● ●●●● ● ● ●●●● ● ● ● ● ● ●● ●● ●●●●● ● ● ● ● ● ●●● ● ● ●● ● ● ● ●●● ●●●●● ● ●●●●●● ● ● ● ●●●● ● ● ● ● ● ●● ●● ●●●●● ●● ● ● ● ● ●● ●●●●●● ● ●●●●●●● ● ● ● ●● ● ● ● ●●● ●●●●● ● ●●●●●● ● ● ● ● ●● ● ● ● ●●● ●●● ● ●● ●●● ● ● ●● ●● ● ● ● ● ●● ●●●●●● ● ●●●●●●● ● ● ●● ● ● ● ● ● ● ● ●●●● ●●● ● ● ● ● ● ●● ● ● ● ●●● ●●● ● ●● ●●● ● ● ●● ● ● ● ●●● ● ●● ● ●● ●● ●●●● ● ● ●● ● ● ● ● ● ● ● ●●●● ●●● ● ● ● ●● ● ● ● ●● ● ●●● ●● ● ●●●●● ● ● ● ● ●●● ● ●● ● ●● ●● ●●●● ● ● ● ● ● ●●●●●●●●● ● ●●●● ● ●●● ●●●●●●●●● ● ●●● ●● ● ● ● ●● ● ●●● ●● ● ●●●●● ● ● ●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ●●● ●●● ● ●● ●●●●●●● ● ●● ●● ● ● ● ● ●●●●●●●●● ● ●●●● ● ●●● ●●●●●●●●●● ● ●●● ●● ●● ● ●● ● ● ● ●●●● ● ●●●● ● ● ●●● ● ● ● ● ●●●● ●●●●● ●● ● ●● ●●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ●●● ●●● ● ●● ●●●●●●● ● ●● ●● ● ● ● ●●●●●● ● ●● ● ● ●●● ●●●●● ● ● ●●●● ●●● ● ● ●● ●● ●● ● ●● ● ● ● ●●●● ● ●●●● ● ● ●●● ● ● ● ● ●●●● ●●●●● ●● ●● ● ●●●● ●● ●●●● ●● ●● ●● ● ●● ● ● ●● ●●●●●●●●● ● ● ● ●●●●●● ● ●● ● ● ●●● ●●●●● ● ● ●●●● ●● ● ● ●● ● ●●●●●●●● ●●●●●● ●●● ●● ● ●●● ● ● ● ●● ● ●●●●●● ●● ●●●● ●●●●● ● ●●●● ●● ● ●●●● ●● ●●●● ●● ●● ●● ● ●● ● ● ●● ●●●●●●●●● ● ● ●●●●●● ●●● ●●● ●●●● ●●●●● ● ●●●● ●●● ●● ●● ●●●●●● ●● ● ●●●●●● ● ● ● ●●●●●●●● ●●● ●● ●●● ●● ● ●●● ● ● ● ●● ● ●●●●●● ●● ●●●● ●●●●● ● ●●●● ● ●● ●●●●●● ●● ●● ●● ● ●●●●●●●●●●●●●●●●●● ● ●● ●●●●●●●●●●●● ● ●●●●● ● ● ● ● ●●●●●● ●●● ●●●● ●●●● ●●●●● ● ●●●● ●●● ●● ●● ●●●●●● ●● ● ●●●●●● ● ● ●● ●●● ● ●● ●● ● ●● ●● ● ●●●●●●●●● ●●●●●●●● ●● ● ● ● ●● ●●●●●● ●● ●● ●● ● ●●●●●●●●●●●●●●●●●● ● ●● ●●●●●●●●●●●● ● ●●●●● ● ● ● ●● ●● ●●● ●● ● ●●● ●● ● ● ●● ● ● ● ●●●●●●●●● ●●●● ●●●●●●● ●●●●● ●●●● ●● ● ● ● ●● ●●● ● ●● ●● ● ●● ●● ● ●●● ●●●●● ●●●●●●●● ●● ● ● ●●● ● ● ● ●● ●●● ● ●● ●● ●●●● ●●● ●●●●● ●●●● ●●● ● ●●●●●●●●●● ● ●●● ● ● ● ●●● ●●● ●● ● ●●● ●● ● ● ●● ● ● ●● ●●●●●●●●●● ●●●● ●●●●●●● ●●●● ●●●● ●● ● ● ● ●●●●●●●●●● ●● ● ●● ● ● ● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●●●●● ●●●● ●● ●●● ● ● ● ●●● ● ● ● ●● ●●● ● ●● ●● ●●●● ●●● ●●●●● ●●●● ●●●● ● ●●●●●●●●●● ●●●●● ● ● ●●●●●●● ●● ● ● ●●●●● ●● ● ● ● ● ●●●●●●● ●●●●●●●●●●●●●●●●● ●● ● ● ● ●●●●●●●●●● ●● ● ●● ● ● ● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●●●● ●●●● ●● ●●● ● ● ●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●● ●●● ●● ●●●●●●●●●●●●●●●●●● ●●●●●● ●●●● ●●●●●●●● ● ● ●●●●●●● ●● ● ● ●●● ● ●● ● ● ● ● ● ●●●●● ●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ●●●●●●● ●●●●● ●●● ●●●●●● ● ● ●●●●● ●●●●●●●●●●●●●●●●●●●● ●●● ● ●●●●● ● ● ● ●●●●●●●●●●●●●● ●● ●● ●●●●●●● ●●● ●● ●●●●●●●●●●●●●●●●●●● ●●●●●●● ●●●● ●●●●●●●● ● ● ●● ● ●●●●●●●●●●●●●● ●●●●●● ●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●● ●● ●● ● ●●●●● ●●●● ●●●●●●●●●●●●●● ●●● ●●●●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●●● ●● ● ●●●● ● ● ●● ● ● ●● ●●●●●●●●●●●●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●●●●●●● ●●● ●● ●●●●●●● ●●●● ● ●● ● ●●●●●●●●●●●●●● ●●●●●● ●●●● ●● ●●●●●●●●●●●●●●●●●●●● ●●● ●●● ●● ●●●●●●●● ● ● ●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ●●●● ● ●● ● ● ●● ●●●●●●●●●●●●●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●● ●● ●●●●●●● ●●●● ● ●● ● ●●●●● ●●●● ● ●●●●●● ●●●●●●●●●●●●●●●●●●●● ● ● ● ●●● ● ● ● ● ●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ●●●● ● ●● ● ● ●●● ● ●●●●●●●●●●●●●● ●●●●●●●● ●●● ●●●●● ● ● ● ● ● ●●●● ●●●● ● ●●●●●● ●●●●●● ●●●●●●●●●●●●●●●● ● ● ●●● ● ● ● ●● ● ●● ● ●●●●●●●●●●● ● ●●● ●●● ●●● ● ● ●● ● ●● ●●● ● ●●●●●●●●●●●●●● ●●●●●●●●●● ●●● ●●●●● ● ● ● ●●●●● ●●●●●●●●●●●●●●● ●●●●●●●● ● ● ●● ● ● ●● ● ●● ● ●●●●●●●●●●● ● ●●● ●●● ●●● ● ●● ●●● ●●● ●●●●●● ●●● ●● ●● ● ●●●●● ●●●●●●●●●●●●●●● ●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ● ●●● ●● ●●● ●●● ●●●●●● ●●● ●● ●● ● ● ● ●● ●● ●● ● ● ●● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ●●● ● ●●● ●● ●● ● ●●● ●●●●● ●●●●● ● ● ● ● ● ●● ●● ●● ● ● ●● ● ● ● ● ●● ●●●● ●●●● ●● ●● ● ●●● ●● ●● ● ●●● ●●●● ● ●●●●● ● ● ●● ● ● ● ● ●● ●●● ● ● ● ● ● ● ● ●● ●●●● ●●●●● ●● ●● ●● ● ● ● ●● ● ● ●● ● ●● ● ● ● ● ●● ●●● ● ● ● ● ● ● ● ● ●●● ●●●● ● ●● ● ● ● ●● ● ● ●● ● ●● ●● ●●● ● ● ● ●● ● ● ● ● ● ●●● ●●●● ● ● ● ● ● ●● ● ●● ●●●●● ● ●● ●●● ● ● ● ●● ● ● ● ●● ●● ● ●● ● ● ●● ● ●● ●●●●● ● ● ● ● ●● ●● ● ●

tSNEtSNE-1 −1 tSNEtSNE-1−1

C RFX6 activityRFX6 Activity D RFX7 activityRFX7 Activity

● ● ● ● ●● ● ●●●● ● ● ● ● ●●●● ● ● ● ● ●●●● ● ●●●● ● ● ● ● ●●●● ●● ● ● ●●● ● ● ● ●●●●●●● ●● ● ●● ●●● ● ●● ●● ● ●●●● ●●●●● ● ● ● ● ●●●●● ●● ● ●●● ●●● ●● ●● ●●●● ●● ●● ●●●●● ● ●● ● ●● ● ● ● ● ●●●●● ● ● ●● ●●● ● ● ●●●● ●●● ● ● ● ● ●● ● ●●●● ● ● ●● ●●●●● ●●● ●● ● ●●●● ● ●● ●●●●●●●●● ● ● ● ● ● ● ●●● ● ●● ● ● ● ● ●●●●●●●● ●● ●●● ● ● ● ● ●● ● ●● ● ● ● ●●●● ● ● ● ●●● ● ● ● ●●● ●●●●●●●●●● ● ●● ● ● ●●● ●● ●●●●● ●●● ● ●●● ●●●●●●● ● ●● ● ●● ● ●●●●●●●●●●● ● ● ● ● ● ●●●●●●●● ●●● ● ● ● ●● ● ●● ●● ● ● ●●● ●●●●●●●● ● ●● ● ● ● ● ● ●●●●●● ●●● ● ●●●●●●●● ● ●● ● ●●●●●●●● ● ●●● ●● ●● ●● ● ●●●●● ●●●●●●● ●● ● ● ● ● ●●● ● ●● ● ●●● ●●●●●●● ●● ● ●● ● ●● ● ●● ● ● ●● ●●●●●●●● ● ●● ●●● ● ● ● ● ●● ●● ●●●●●●●●●●●●●●●●●●●●● ●● ● ●●●●●●●● ●● ●● ●●●●●●●●●● ●● ●● ●●● ● ● ● ● ● ●● ●● ●●● ●●●●●●●● ●● ● ● ● ●● ● ●● ●●●●●● ●●●●●●●● ●●● ● ●● ● ● ● ● ● ●●●●● ●●●●●●●● ● ● ● ● ● ● ● ● ●●●●●●●●●●● ●●●●●● ●●●●●●●●● ● ● ● ● ● ●●● ●●●●●●●●●●●●● ● ●●● ● ●● ● ●●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ●● ●●●●● ●●● ●●●●●●●●●●●●●● ●●● ●● ● ● ● ●●● ● ●● ●●●●●●●● ● ● ● ● ● ●● ●● ●●●● ● ● ●●●●●● ●●● ●● ● ● ● ● ●● ● ● ●● ●●●●●●●●●● ●●●● ● ● ● ● ● ● ●● ●●●●●●●●●●●●● ●●● ● ●●● ● ● ●● ● ●● ●●● ●●●●●● ● ●● ●●●●●●● ● ● ●● ● ●● ● ● ● ●●● ● ●●● ●● ● ●●● ● ● ●●●● ●● ●● ●●● ●●●●●●●●●● ●●● ●●● ●● ● ● ● ●●● ● ●● ●● ● ●●●● ●●●● ● ● ● ●● ● ● ● ● ● ● ●● ● ●●● ●●● ●● ●●●● ● ● ● ● ●● ●●●● ●●●●●●●● ●●● ● ● ● ● ● ●●●● ● ●●●●●●●●● ●● ● ●●●● ● ● ● ●● ● ● ●● ● ●● ●●●● ●●●● ● ● ●● ● ● ● ● ●● ●●● ●● ● ●● ●●●●● ● ● ● ● ● ●●●●●● ●●●●● ● ●● ● ●●●●● ●●● ● ●●● ● ● ●●● ● ●● ● ●●●● ●●●●●●●●● ●● ● ● ● ● ● ●●●● ●●● ●●● ●●● ●●●● ● ● ● ● ●● ● ● ● ● ● ●●●● ●●●●● ● ●●●●● ●●●●●●●● ● ●● ● ● ● ● ●● ●●●● ●●●●●●●●●●●● ●●● ●●● ● ● ● ● ● ● ●●● ●●●●●●● ●● ● ● ●● ●● ●●● ● ●●● ● ● ● ● ● ●● ●●●●●● ●●●●●●●●●●● ● ●● ● ● ● ● ●●●● ●● ●● ●●●●●●●●● ●●● ● ● ●● ● ● ●● ● ●● ●●● ● ●●●● ●●●●● ● ● ● ● ●● ● ● ● ● ●●● ●●●●●●● ●●●●●●●●●●●●●●●●● ● ● ● ● ● ●●●● ●●●● ●●● ●●●●●●● ●●● ●●● ● ●● ● ● ●● ●● ● ● ● ●●● ● ●● ●●● ●●●●●● ● ● ● ●● ● ● ● ● ● ●●●● ● ● ● ●● ● ●● ●●● ●● ●●● ●●● ●● ● ●● ● ●● ● ●●●●● ●●●●● ● ● ●●●●● ● ●● ● ●●● ●● ● ● ●●●●●● ● ●●●● ●●● ● ● ●●●●●● ● ● ● ● ● ●●● ●●● ●●●●●●●● ●●● ●● ● ● ● ● ●● ●● ●● ●● ● ● ● ●●●●●●● ● ●● ● ●● ● ● ● ● ● ●●● ●● ● ●●●●● ●●●●● ● ● ●●●● ●● ● ●● ● ● ●●● ●●● ●●●●●●● ●●●●● ●● ●● ●● ● ●●● ●● ● ● ● ●● ●●●●●● ●● ● ● ●●●●●●● ● ● ●● ● ● ● ● ● ●●●●● ●●●●●●●●● ● ●●●●●●●●● ● ● ●● ● ●● ●● ●●● ●●●● ● ●●●●● ●● ●● ● ●●● ● ● ● ● ● ● ● ● ●●● ●●●●●● ● ● ●●● ●●●● ● ●●● ● ● ● ● ● ● ● ●●● ● ●● ●●●●●●● ●●●●●● ● ● ● ●● ● ● ● ●●● ● ●● ●●● ●●●● ● ●●●●●●● ● ● ●●●●● ● ●● ● ● ● ● ● ● ● ●●●● ● ●●●●●● ● ●●●●●●●● ● ● ● ● ●● ●● ●●● ●●●●●●●●● ●●●●● ● ●●● ●● ● ● ●● ●● ● ● ● ●● ●●● ● ● ● ●●●●● ●● ●●●●●●● ●●● ● ● ● ● ● ● ●● ● ●● ● ● ● ●●● ●● ● ● ● ●● ● ●●●●●● ●●● ●●●●●●●● ● ● ● ●●● ●●● ●● ●● ●● ● ● ● ● ● ●● ● ● ●●● ●●●●●●●●● ●●●●●●● ●●●● ●●●●● ● ●●● ●● ● ● ●● ● ●●● ●●● ●●●●●●●●● ●●● ● ●● ●● ●● ● ●● ● ●● ●●●●●● ●●●●●●●● ●● ● ●●● ● ●● ● ●● ● ● ● ● ● ●●●●● ●● ● ● ●● ● ●●●●●● ● ● ●●● ● ● ● ● ● ● ●● ● ● ● ●●●●●●● ●●● ●● ●● ●●●●●●●● ●● ● ● ● ● ● ●●●● ● ●●● ●●●●●●●●●●●●●●●●●●●● ● ● ●●●●● ●●●● ● ● ● ● ● ● ●● ●● ●● ● ●●●●●●●●● ● ●●●●●●●●●● ●●● ●● ● ●● ● ● ● ●● ●●● ●●●●●●●●●● ● ● ●●●●● ● ● ● ● ● ● ●●●●● ●●●●●●●●●● ●●●●●●● ●● ●●● ●●●●●● ●● ●●● ● ● ● ● ●● ● ●● ●● ● ●●●●●●● ●●●●●● ●●●●●●●●●● ●● ● ●● ● ●● ●● ● ●●●●●● ●●●●●●●●●●●●●●● ●●●●● ●●● ●●● ●● ● ● ●● ● ●●●●● ●●● ●● ●●●●●● ●● ●●●●●● ●●● ●● ●●●● ● ●● ● ● ● ● ● ●●●●●●●●● ●●●●●●●●● ●●●● ●●●●● ● ● ● ● ● ● ● ●● ●●●●●● ●●●●●●●●●●●●●●●● ● ●●●● ●●●●●●● ●● ● ●● ● ●●● ● ●● ● ●●●●● ●●●●●● ● ● ●●● ●●●● ●●●● ●● ● ● ●●●● ● ● ● ● ● ●●●● ●●● ●● ●●●●●●●●●● ●●● ●●●●●● ●●● ● ● ● ●● ● ●●●●● ●●●●● ●●●●● ●●●● ● ● ●● ●●●●●●●●●●●●● ● ● ●●●● ●● ● ●● ● ● ● ●●●●●●●●●●●● ●●●●● ●● ●●●●●●● ● ● ●● ● ● ● ● ● ● ● ●● ●●●●●● ●●●●●●●● ●● ●●●●●● ● ●●● ● ● ● ●● ●●● ●●● ●●●●●●●●●●●●●● ● ●● ●●●●●●●●●●● ●● ●● ●●● ● ● ● ●● ●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●● ● ● ● ● ● ● ●●● ●● ●●●●●●●●●●●●●●●●● ●● ● ●●●●●●●●●●● ● ●● ●●● ● ● ●● ●● ●●●●●●●●●●●●●●●●●●●● ●● ●●●● ●● ●● ● ● ● ● ● ● ●●● ●●● ● ●●● ● ●●●●●● ●●●●●●●● ●● ● ● ● ● ● ●● ● ●● ● ● ●● ●●●●●●● ●● ●●●●●●●●● ●● ●●● ● ●● ● ● ● ● ● ● ● ●● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●● ● ● ● ● ● ●●●●●● ●●●●●●●●●●● ●●●●●●●●● ● ● ●● ● ● ● ● ●●●●● ● ● ● ● ● ●●● ●●●●●●●● ●●●●●●●●●●● ●● ● ●●● ● ● ● ● ●●●● ●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●● ● ● ●● ●● ● ● ● ● ●● ●● ●● ●●●●●●● ●● ●●●●● ●●●●● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ●●● ●●●●●●●●●● ●●●●●●●●● ● ●●●● ●● ● ●●●●● ● ●●●● ●● ●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●● ●● ● ● ● ● ● ●● ●● ● ● ● ●● ●●●●●●●●●●●● ●●●●●●●●● ●●●● ●●●●●● ● ●● ● ● ● ● ● ●●● ●●●●●●●●●●●● ●●●●●● ●●●●●● ●●●●●●● ● ●● ● ●● ● ●● ● ● ● ●● ● ●●●●●● ●●●●●●●● ●●●●●●●●● ●●●●●● ●●●●●●●●● ● ●● ● ●● ● ● ● ● ●●●● ● ●●●●●●●●●●●● ●●●●●●● ● ●● ●● ●● ● ● ●●●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●● ●●●●●●●●●● ●● ●● ●●● ● ● ●● ●● ●●●●●●●● ●●●●● ●●●●●●●●●● ●●●●●●● ●● ●● ● ● ● ● ● ● ● ●●●● ● ● ●●●●● ●●●●●●●●●●● ● ● ●● ● ● ● ● ●● ● ● ●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●● ● ●●● ● ● ● ●● ● ● ●●●●●●● ●●● ● ●● ●●● ●●●●●●●●●●●●● ● ● ● ● ● ●● ● ●● ●●●●● ●● ● ●●●● ● ●●● ●● ●● ● ● ● ● ● ●●●●●●●●●● ● ● ●● ●●●● ●●● ● ● ●●● ● ● ●● ●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●● ● ●● ● ● ● ● ● ●●●● ●● ●●● ●●● ●●●●● ● ● ●●●●●●●●● ●● ● ● ●● ● ● ● ● ● ●● ●●● ●●● ●● ●●● ●●●●● ●●●●●●● ● ●●● ● ● ● ● ● ●●● ●● ●●●● ●● ●● ●●● ●● ●●●●● ● ●●● ● ● ● ● ●● ●●● ● ●●●●●●●●●●●● ● ●●●● ●●●●● ● ●● ●● ● ● ● ● ● ● ● ● ●● ●●●●● ●●●●● ●●●● ● ●●● ●● ● ●● ● ●●● ●●●● ● ●● ●●●●●●●●● ●● ●●●●●●● ●● ●● ●● ● 2 ● ● ● ●● ● ●● ●● ● ● ● ● ●● ● ● ● ●● ●● ● ●● ● ●● ●●● ● ●● ● ● ● ● ● ●●●● ●●●●●● ●● ●●●●● ●●●● ● ● ● ● ● ● ●● ●●●●●●●●● ●●● ● ● ●● ●● ● ●●● ●●● ● ●●● ● ● ● ●●● ●●●● ●● ● ●● ● ● ● ●● ● ●● ●●●● ● ●●● ●●●●●● ●●●● ●● ● ●● ● ● ● ●●●● ●●●● ●●●● ●● ●●● ● ● ● ●●● ● ●●●● ●● ●●●● ●●● ●●●● ● ● ● - ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ●●●●● ●● ● ●●●● ● ●● ● ●●●●● ●●●● ● ●● ●● ● ●● ●●● ● ● ● ● ●●●● ● ●● ● ●● ●●●● ●● ● ●● ● ● ● ● ● ●● ● ●●●● ● ●●● ● ●● ● ●● ● ● ● ●● ●● ● ●●●● ● ●●● ●●● ● ● ● ● ● 2 ● ● ● ● ● ● ● ● ●● ● ● ●● ●●●●● ●● ●● ●● ●●●●● ● ●● ●● ● ● ● ● ● ● ● ●●●● ● ●●● ●● ●●● ●●●● ●● ● ●● ● ●● ● ●● ● ●● ● ● ●●● ●● ●● ● ●●●● ● ● ●● ●●● ● ●●●● ●● ●● ●●● ●●●● ● ●● ● ● ● ● ●● ● ●● ●● ● ● ●● ● ●● ● ● ● ● ● ●● ●●●● ●●● ● ● ● ●●●● ● ●● ●● ●●●●● ● ●● ● ● ● ● ● - ● ● ● ● ● ● ●●● ● ●●● ●●●● ●●●● ●●●●● ●●● ●●● ● ● ●● ● ●● ● ●● ● ●●● ● ●● ●●● ● ● ●●● ● ●● ● ●● ● ●●●● ● ● ●● ●●● ●● ● ●● ●● ● ●●●● ● ● ● ●●● ● ● ● ● ●● ●●● ●●●●●●● ●●● ● ● ● ● ● ● ●● ●●●● ● ● ● ● ●●●●●●●●● ● ●● ●● ●● ●●● ● ●●● ●●● ● ●●● ● ●● ●● ● ●● ● ● ● ● ● ●●●●●●● ●●●●● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ●● ●●●●● ● ● ● ●● ●●●● ●●● ●● ●● ● ● ● ●●●●●●●● ●●●● ●●● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ●●●●●●● ● ● ● ● ● ● ● ●●●●● ● ● ●●●●●●●●● ●●● ● ●● ● ● ●● ● ● ● ● ● ● ●●●●●●●●●●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ● ● ● ● ● ●●● ●● ●●● ●● ●●●● ●●●●●● ● ● ● ● ● 2 ●● ● ● ● ● ●●● ● ●●●●●● ● ●●● ● ● ● ●● ● ● ● ● ●●● ● ● ● ●●●● ● ● ● ● ● ● ● ● ●● ●●●●● ●●●●●●●●●● ●●● ● ●● ●●●●● ● ●●● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ● ● 2 ● ●● ● ● ● ● ● ● ●● ●● ● ●● ● ● ● ● ●●● ●●●●●●●● ●●● ● ●● ● ● ●● ● ● ● ● ●●●●● ●●● ● ●●● ● ●●●● ● ● ●● ●● ● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ● ● ● ● ●● ●● ●●●● ●●●●●● ●●● ●●●●●●●● ●●●● ●●● ●● ● − ● ●●● ● ● ● ● ●● ● ●●●●●● ● ●● ● ●● ● ● ● ● ● ● ●●● ● ● ●● ●● ● ● ●● ● ● ● ●●●●●● ● ●●●●● ●●● ●●● ●●● ● ●● ● ●● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ● ●● ●● ● ● − ● ●● ● ● ● ●● ●●●●● ●●●● ●●● ●●●●●● ●●●●●●● ●●●●●●●● ● ● ● ●●● ● ●●●●●● ● ● ● ●● ●● ●●●●● ●●●● ● ● ● ●● ● ● ● ●● ●● ●●●●● ● ● ●●● ●●●●●●●●●●● ● ●●●● ●●● ●● ●●●●●● ● ● ●● ● ●●●● ●●●●●●●● ● ● ●●●●●●●●●●●● ●●●●●●● ● ●●●● ●● ●● ● ●●●●●● ●● ● ● ●●● ●●●●●●● ● ● ● ● ● ● ●●● ●● ●●●●●●●●● ●●●●●●● ●●●●●● ● ● ● ● ●● ● ●●●● ● ● ●●●●● ●●●●●●●●●●● ● ●●●●●●●● ●● ● ● ●● ●● ●●●●● ● ● ●● ●●●●●●●●●● ●●●●●● ●●● ●● ●●●● ● ● tSNE ●●●●●● ●● ● ● ●●●● ●●●● ●● ●●●●●●●●●● ● ● ● ● ● ● ● ●● ● ●●●●●● ● ●● ●●●● ●● ● ●●●● ● ● ● ● ● ●●●●●●● ●●● ● ●● ● ●●●●●●●●●● ● ●● ●● ●● ● ●●● ● ●●●●● ● ● ●●● ● ●●●●●●● ●●●● ● ●●●●●●●●● ●● ● ● ● ● ●●●●●● ●● ●● ● ● ● ●●● ● ● ●● ●● ● ●● ●● ● ●●●●●●● ● ● ●●●● ●●●● ●●●● ●●●●●●●●● ●● ● ● ● ● ● ● ● ●●● ● ●● ●●●● ●●● ●● ●●●●●●● ●● ● ● ●●● ● ● ●●●●●● ●●●●● ●● ● ●●●●●● ● ●● ● ● ●● ●● ● ● ●●● ● ● ●● ● ●●●●●●●●● ● ● ●●● ●●●●● ● ● ●● ●● ●● ●●●● ● ●●●● ● ●● ● ●● ●● ●●●● ● ●● ● ● ● ● ● ●●●●●● ●●●● ● ●● ● ● ●●● ●● ● ● ● ●● ●●● ●●●●●●●●●● ● ●●● tSNE ● ● ●●●●●● ● ●●● ● ● ● ●● ●●● ● ● ● ● ● ●●● ●● ● ●●●● ● ● ● ●●●●●●●● ● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ●● ●●●●●●●● ●● ●● ● ● ● ●●●● ●●●●●● ● ● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ●●●● ● ● ●● ● ● ● ● ●● ● ●● ● tSNE ●● ● ● ●● ● ●●● ● ●● ●● ● ● ●● ●●●● ● ●● ● ● ●● ●● ●● ● ● ●● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ●● ● ●● ● ●● ● ● ● ● ●●● tSNE ●● ● ● ●● ● ●●● ● ●●● ● ●●● ● ● ●●● ● ●●● ●●● ● ● ●●●● ● ●● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ●●● ●● ●●● ●●● ● ●●● ● ●● ●●● ●●●● ●● ●●●● ● ● ●● ● ● ● ● ● ● ● ●●● ● ●● ● ● ●● ● ● ●● ● ● ●● ● ●● ● ●●● ● ● ● ● ● ●●● ● ● ●●● ● ● ●●● ●● ●●● ●●●●●●● ● ●●● ● ● ● ● ● ● ●● ● ●●● ●●● ●●● ● ● ●● ● ● ● ●● ● ●● ● ●● ● ● ● ● ●●● ● ● ●● ● ●● ● ● ● ● ●●●● ● ● ●● ● ●●●●● ● ● ●● ●● ●●● ● ●●●● ●●●●● ● ● ●● ●● ● ●● ●● ●●● ● ●●● ●●●● ●● ●● ●● ●● ● ● ● ● ● ● ●● ● ● ● ●●●●● ●●● ●●● ● ● ● ● ●● ● ●●●● ● ● ● ●●● ● ●●●● ●●●●●●●●●●● ●●● ●●●● ● ● ●● ●● ●● ● ●●●● ●●● ● ● ●●●● ●● ●●● ●● ●● ● ● ●●● ●● ●● ●● ●●●● ● ● ● ● ● ●●●●● ●● ● ●●●●●●●●●●●●● ●● ●●●●● ●● ●● ● ● ● ● ● ●●●●●●●●●● ●●● ● ●●●● ● ● ● ● ● ●● ● ●●● ●●● ● ● ● ● ●● ● ● ● ●●●●●● ●● ● ●● ●●● ● ● ●● ● ●●● ●●● ●● ● ● ● ●●●● ●●●● ●● ● ● ● ● ●● ● ● ●● ● ● ● ●● ●●●● ● ●●● ●●●●●● ●●● ● ●● ● ● ●●●●●●● ● ●● ● ● ●● ● ● ● ● ●● ● ● ●● ●●●●●●● ● ●●●● ● ● ●●●● ●● ● ● ● ● ●● ● ●●● ●● ● ● ● ●●●● ● ● ● ● ● ● ●●● ● ● ● ● ●● ●● ●● ● ● ●●●●● ●●●●● ● ● ● ●● ●● ● ●●● ● ●●●●●● ● ● ● ● ● ● ● ● ●●● ●●●● ●● ●●● ● ● ● ● ● ●● ●● ● ●● ●● ● ● ● ●● ● ●●● ● ●● ●●● ●● ●● ● ●●●● ● ●● ●● ●● ●● ● ● ● ● ● ●● ● ●● ● ● ●● ● ●● ● ● ●● ● ● ●●● ● ● ●●● ● ●● ●● ● ● ●● ●● ● ● ● ● ● ●●●●●●● ● ●●●● ● ● ●●● ● ● ●● ● ●● ●● ●●● ● ● ● ● ●● ● ●● ● ● ● ●●●● ● ●● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ●● ●● ●● ● ●● ● ● ● ●● ●● ● ●●●● ● ●● ● ● ● ● ● ● ● ● ●●● ●● ●●●● ● ● ● ● ● ● ● ●● ● ●● ●● ●● ●● ● ● ● ● ● ● ● ● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ●●●●●● ●● ●●●● ● ●● ● ● ● ● ● ●●●● ●● ● ● ● ●● ● ● ● ● ● ● ● ●●● ●● ● ●●●● ● ●●● ● ● ● ● ● ● ● ● ●● ●●●●●●● ●●●●●● ● ● ● ● ● ● ● ●● ● ● ●●●●●●●●●● ● ● ● ● ● ● ● ●● ●●●● ● ●●● ● ● ● ● ● ● ●● ● ● ●●●● ●●● ● ● ● ●● ●● ●● ●●●●● ●●●●● ● ● ● ●● ●●●●●●●●● ● ●● ● ● ●● ●●● ● ●●●●● ●●●● ● ● ● ●● ● ●●●●●●●● ●● ● ●●● ● ● ● ● ●●● ● ●●●● ● ● ●● ● ●●●●● ●●● ●● ● ● ● ●●●●●●●●●●●● ● ● ● ●● ● ● ●●●●● ●● ●● ● ●● ● ● ● ● ●● ● ●●●●●●●● ● ● ●●● ● ● ●● ●●● ●● ●●● ● ●●● ● ● ●● ●●● ● ●●●●● ●●● ●● ● ● ● ● ● ●●●● ●●● ● ●● ●●● ● ● ● ● ● ●●●●●●●●● ●● ● ● ● ●● ● ● ● ● ● ● ●●● ●● ●●● ● ● ● ● ● ●● ●●● ● ●●● ●● ●●●● ● ● ●● ●● ● ●● ● ●●●● ● ●● ● ● ●●● ● ● ● ● ● ● ●●● ●● ● ● ● ● ● ●●●●●●●●●● ●●●●●●●● ● ● ● ●●●● ● ● ● ● ● ●● ●● ●●●●● ●● ● ●● ●● ● ●●● ● ● ● ● ● ●●● ● ●● ● ● ● ●●● ●●●●● ● ●●●●●● ● ● ● ●● ●● ● ● ●●● ●● ● ●● ● ●●● ● ●●●● ● ● ● ●● ●● ● ● ● ●● ●●●●●● ● ●●●●●●● ● ● ●●●● ●● ● ● ●● ● ●● ●● ●● ●● ●● ● ● ● ● ● ● ●●●● ●●● ●● ●● ● ●● ● ●● ● ●● ● ● ●● ●● ●●● ●● ● ● ●●● ● ● ●● ● ● ● ● ● ● ● ●● ● ●●● ● ●● ● ● ●● ●●●●●●● ●●●● ●● ●●●● ●●●●● ● ● ●●● ● ● ● ●●● ● ●● ● ●●●● ●● ●●●● ● ● ● ●● ●●●●●●●● ●●●●●●●●●●●●●●● ● ●●● ● ●●● ● ● ●●●●●●●● ●●● ● ●● ● ● ● ●● ● ●● ●● ● ●●●●● ● ●● ● ●●● ● ●●●● ●●● ●●●●● ● ● ●●● ● ● ●●● ●● ● ● ● ●●●●● ● ● ● ●● ● ●●●●●●●●● ● ●●●● ● ●● ●●●●●●●●●● ● ●●● ● ●● ● ● ● ● ●●● ● ●● ●● ●● ● ● ●● ●●●● ● ●● ●● ● ● ●● ● ●●●●●● ●●●●●●●●●●●●●●●●● ● ● ●●● ●●● ● ●● ●●●●●●● ● ●●● ●● ● ●● ● ● ●●●●●● ●● ●● ● ● ●●●● ●●● ● ●●● ● ●●● ●●●●●● ●● ●● ● ●● ● ● ● ●●●● ● ●●●● ● ● ●●● ● ● ● ● ●●●● ●●●● ●● ● ●● ●●●●● ●●● ● ●●●●●● ● ●●● ●● ● ●● ● ●●●●●● ●● ●●●●●●●●●● ● ●● ●● ● ● ●●●●●● ● ●● ● ● ●●● ●●●●● ● ● ●●●● ●●● ● ● ●● ● ●●●●●● ● ● ●●● ●●●● ●●● ● ● ●●●● ●●● ●●● ●●● ●●●●●● ● ●●● ●●●●● ● ●● ● ●●●● ●● ●●●● ●● ●● ●● ● ●● ● ● ●● ●●●●●●●●● ● ●●● ●●●●●●●● ● ●●●●● ● ●● ●● ● ●●● ●●● ●●● ● ●●●●●●●●●●●●● ●●●●●●●● ● ● ● ● ●●●●●●●● ●●●●●● ●●● ●● ● ●●● ● ● ● ●● ● ●●●●●● ●● ●●●● ●●●●● ● ●●●● ●● ● ● ●●●● ● ●● ●●●● ●●● ●●●● ●● ●●● ● ●● ●●●●●● ●● ● ● ●● ● ●●●●●● ●●● ●●● ●●●● ●●●●● ● ●●●● ●●● ●● ●● ●●●●●● ●● ● ●●●●●●● ● ● ●● ●●● ●● ●●●● ●● ● ● ●●● ● ● ● ●●●●●●●● ●●●●●●●● ●●●● ●●● ●● ● ● ● ● ● ●● ●●●●●● ●● ●● ●● ● ●●●●●●●●●●●●●●●●●● ● ●● ●●●●●●●●●●●● ● ●●●●● ● ● ● ●●●●● ● ●● ●●●●● ●●● ●●● ●● ●●●●●●●● ●●●● ●●●●●●●●●●●●●●● ● ●● ● ● ●● ●●● ● ●● ●● ● ●● ●● ● ●●●●●●●●● ●●●●●●●● ●● ● ● ●●●●●●●● ● ●● ● ● ●● ● ●● ● ● ● ●● ●● ●●●●●● ●●●●●●●●●●●●●●● ●●●●●● ● ●●● ● ● ● ●● ●●● ●● ● ●●● ●● ● ● ●● ● ● ● ●●●●●●●●● ●●●● ●●●●●●● ●●●● ●●●● ●● ● ● ● ● ●●●●●●●● ● ●●●● ●●● ●● ● ●●● ●●●●●●● ●●●●●●●●●●●●●●●● ●●● ● ● ● ● ● ●●● ● ● ● ●● ●●● ● ●● ●● ●●●● ●●● ●●●●● ●●●● ●●● ● ●●●●●●●●●● ●●●●● ● ● ● ●●●●●●●●● ●● ●● ●● ●●●●● ●● ●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●● ● ●●●●●●●●●● ●● ● ●● ● ● ● ● ● ● ● ● ●● ●●●●●●●●●●●●●●●●●●● ●●●● ●● ●●● ● ● ●●●●● ●●●●●●●●●●●● ●●● ●●●●●● ● ●● ●●● ●●●●●●●●●● ●●●●●●●●● ●●● ●● ●● ●●●●● ● ●●●●●●● ●● ● ● ●●● ● ●● ● ● ● ● ●●●●●●● ●●●●●●●●●●●●●●●●●● ●● ● ● ● ●● ●●● ●●●●●●●●●●●●●● ●●●●●●● ●●● ●● ●●●●●●●●●●●●●●●●●●●● ●● ●●● ●● ●●●●●●●● ● ●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●● ●●● ●● ●●●●●●●●●●●●●●●●●● ●●●●●● ●●●● ●●●●●●●● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●●● ●● ● ● ● ● ●●●●●●●●●●●●●●●●●●●●●●●● ●● ● ●●●● ●● ● ●●● ● ●● ● ●●●●●●●● ●●●●● ●●● ●●●●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●●● ●●● ● ●●●●● ● ● ● ● ●● ●●●●●●●●●●●●● ●●● ●●● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●● ● ●●●● ● ● ●● ● ●●●●●●●●●●●●●● ●●●●●● ●●●● ●● ●●●●●●●●●●●●●●●●●●●● ●●● ●● ●● ● ●●●●● ●● ● ●● ●●●● ● ●● ●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●● ● ●● ●●● ● ● ●● ● ● ●● ●●●●●●●●●●●●●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●●●●●● ●●● ●● ●●●●●●● ●●●● ● ● ●● ●●●● ●●● ●●● ●●●●●●●●●●●●●●●●●●●●●●● ● ● ●●●●● ● ● ● ● ● ●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ●●●● ● ● ●● ●● ●● ● ●●●●●●●●●●● ● ● ● ●●● ●● ● ● ● ●●●● ●●●● ● ●●●●●● ●●●●●● ●●●●●●●●●●●●●●●● ● ● ●●● ● ● ● ●●●● ●●●●●●●●●●●●●● ● ●●●● ● ● ● ●● ● ●● ● ●● ●●● ● ●●●●●●●●●●●●●● ●●●●●●●●●● ●●● ●●●●● ● ● ● ● ●● ●●●●●● ●●●●●●● ●● ●● ●● ●● ●● ● ●● ● ●● ● ●●●●●●●●●●● ● ●●● ●●● ●●● ● ● ● ● ● ● ● ●● ●● ●●● ●●● ● ●●● ●●●●● ●●●●●●●●●●●●●●● ●●●●●●●● ● ● ●● ● ● ● ● ● ●● ●● ●● ● ● ●● ●● ●●● ●●● ●●●●●● ●●● ●● ●● ● ● ● ●● ●● ● ● ●●● ●● ● ●● ● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ● ●●● ● ● ● ●● ●● ●●● ●●● ● ●●●●● ●●● ● ● ●● ●● ●● ● ● ●● ● ● ● ● ● ●●● ● ● ●● ●● ● ● ● ●●● ●● ●● ● ●●● ●●●●● ●●●●● ● ● ● ●● ● ● ●● ● ●● ●●● ● ● ● ● ● ●● ●●●● ●●●● ●● ●● ● ● ● ●●● ● ● ● ●● ● ● ● ● ●● ●●● ● ● ● ● ● ● ●●● ● ● ● ●● ● ●● ● ● ● ●● ● ● ●● ● ● ●●● ● ● ●● ● ● ●●●●● ● ● ● ● ● ●●● ●●●● ● ● ● ●● ● ●● ● ● ●● ●●● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ●● ● ●● ●●●●● ● ● ● ●● ●● ● ●

tSNEtSNE-1 −1 tSNEtSNE-1−1 E F NANOG activityNANOG Activity MYCL activityMYCL Activity

● ● ● ● ● ● ●● ● ● ●● ● ●●●● ● ● ● ●● ● ● ● ●●●●● ● ● ●●●● ● ● ●●●● ● ● ● ●● ●●●●● ● ● ● ●●●● ●● ● ●● ●● ●●●●● ● ●●● ● ● ●● ● ●●●● ●● ● ●●● ● ● ●●●● ●●●● ●● ● ●●● ●●●● ●●●●● ●● ●●●●● ●●● ●● ● ●●●● ●● ● ●●●● ●●● ●● ●● ●● ● ● ● ● ●●●● ● ●● ● ● ● ● ● ● ●●●●●● ● ● ●●● ● ● ● ● ● ●● ● ● ● ● ● ● ●●●●● ● ● ● ● ●●● ●●●●●●●●●● ● ●● ● ● ● ●● ●●● ●●● ● ● ●●● ●●●●●●● ● ●● ● ●●●● ● ●●●● ●●● ● ● ● ● ●●●●●●●● ●●● ● ● ● ●● ● ● ●●●●●● ●● ● ● ●●● ● ● ● ● ●●●●●● ●●● ● ● ●●● ●●● ●●●●● ● ● ● ●● ● ●●●●● ●●●●●●● ●● ● ● ● ● ●● ● ●●●●●●●●●● ● ●● ● ●● ● ● ●● ●●●●●●●● ● ●● ●●● ●● ● ● ● ●●● ●●●● ●●●● ● ●● ● ● ● ●●●●●●●● ●● ●● ●●●●●●●●●● ●● ●● ●●● ● ● ● ●●●●●●●● ● ●● ● ●●●●●●●● ● ●●● ●● ●● ● ●● ● ●● ●●●●●● ●●●●●●●● ●●● ● ●● ● ● ● ●●● ●● ●● ● ●●● ●●●●●● ●● ● ●● ● ● ● ● ● ● ● ●●●●●●●●●●● ●●●●●● ●●●●●●●●● ● ● ● ●● ● ● ● ●● ●● ●●●●●●●●●●●●●●● ●●●●● ● ●● ● ●●● ● ●●●●●●●●●●● ● ●● ● ● ● ●● ●● ●●● ●●●●●●●●● ● ●● ● ● ● ● ●●● ●● ●●●●●●●● ●● ● ● ● ● ● ● ●●●●● ●●●●●●●● ● ● ● ● ● ● ● ● ● ●● ●● ●●●●●●●●●●●●●●● ● ● ● ● ● ●●● ●●●●●●●●●●●●●● ● ●●● ● ●● ● ●● ●● ●●●●●● ●● ● ●●●●●● ● ● ●● ● ● ● ●● ● ● ● ●● ●●●●● ●●● ●●●●● ●●●●●●●● ●●● ●● ● ● ●● ●● ●●● ●●●●●●●●●●●●●● ●●● ●● ● ● ●● ●● ●●●● ● ● ●●●●●● ●●● ●● ●● ● ● ● ● ● ● ● ● ●● ●●● ●● ●●●● ● ● ●● ●●●●●●●●●●●●● ●●● ● ●●● ● ● ●● ●●●● ● ●●●●●●●● ●● ● ●●● ●● ● ● ● ●●● ● ●●● ●● ● ●●● ● ● ●●●● ● ● ●● ●●● ● ●● ●● ●●●●● ● ● ● ● ●●● ● ●● ●● ● ●●●● ●●●● ● ● ● ● ● ● ● ●●● ●● ● ●●● ●●●●●●●●●●●● ● ● ● ● ● ● ●● ●●●● ●●●●●●●● ●●● ● ● ● ● ● ●●●● ●●●●● ● ●●●●●● ●●● ●●●● ● ●● ● ● ● ● ● ● ● ●● ● ● ●● ● ●● ●●●● ●●●● ● ● ●● ● ● ● ● ●●● ●●●●●●● ●● ● ● ●● ●● ●●●● ● ●●● ● ● ● ● ●●●●●● ●●●●● ● ●● ● ● ●●● ●●● ● ●●● ● ● ●●●●●●● ● ●● ●● ●●●●●●●●● ●● ● ● ●● ● ● ● ●●●● ●●● ●●● ●●● ●●●●● ● ● ● ● ●● ● ● ●● ●●●●●● ●●●●●●●●●●●●●● ● ● ● ● ●● ●●●● ●●●●●●●●●●●● ●●● ●●● ● ● ● ● ●● ●● ● ●● ● ●●● ● ●● ●●●● ● ●●●●●● ● ● ●● ● ● ● ● ● ● ●● ●●●●●● ●●●●●●●●●●● ● ●● ● ● ● ● ● ●●●●● ●●●● ● ●●●●●●● ● ●● ●● ●●● ●●● ● ● ● ●● ● ●● ●●● ● ●● ●●●● ●●●●● ● ● ● ● ●● ● ● ● ●● ● ● ●●●● ●●● ●● ●●●● ●●●● ● ● ●● ● ●●●● ● ●● ●●● ●●●●●●● ●●●●●● ● ●● ● ● ● ● ● ●● ●● ● ● ●●● ●●●●● ●● ● ●●● ●● ● ● ● ● ● ●●●● ● ● ● ●● ● ●● ●●● ●● ●● ●●●● ●● ● ●● ● ● ● ●● ●● ● ● ●● ● ●●●●●●●●● ● ● ● ●● ● ● ●●●●●● ● ●●●● ●●● ● ● ●●● ●● ● ● ● ● ● ●● ●●● ●●●● ●● ●●● ●● ●● ●●● ● ● ● ●● ●● ●● ●● ● ● ● ●●●●●●● ● ●● ● ●● ● ● ● ● ● ●●● ● ●● ●●●●●●● ●●●●●● ●● ● ● ●● ● ● ● ●●● ● ● ●●● ●●● ●●●●●●● ●●●●● ●● ●● ●● ● ●●● ●● ● ● ● ● ●● ●●●● ● ●●●●●● ●● ●● ●●●●● ● ● ● ● ● ● ● ● ●●●●● ●●●●●●●●● ● ●●●●●●●●● ● ● ●● ●● ● ● ●● ●● ● ● ● ●●●●● ● ●●●●●●● ● ●●● ● ● ● ● ● ● ● ●●● ●●●●●● ● ● ●●● ●●●● ● ●●● ● ● ● ● ●●●●●● ●●● ●●●●●●● ● ● ● ●●●● ●● ●●● ●● ●● ● ● ● ●● ●●● ●●●● ● ●●●●●●● ● ● ●●●●● ● ●● ● ● ● ●●● ●●●● ●●●● ●●●● ●●● ● ●● ●● ●● ●● ●● ● ● ●● ●● ●●● ●●●●●●●●● ●●●●● ● ●●● ●● ● ● ●● ● ● ● ● ● ●● ●●●●● ●● ● ●●● ● ●●●●●●●●● ● ●●● ●● ● ● ● ● ● ● ●● ● ●● ● ● ● ●●●●●● ● ● ● ● ● ● ●●●● ●●● ●●●●●●●●●●●●●●●●●●● ● ●●● ● ● ●●●● ● ● ● ● ● ● ● ●● ● ● ●●● ●●●●●●●●● ●●●●●●●● ●●●● ●●●●● ●●●●●● ●● ● ● ●● ● ● ●● ● ●● ●●● ●●●●●●●●●● ● ●●●●●● ●●●● ● ● ●● ●●●●●● ● ●●●●●●●●● ●● ● ●●● ● ● ● ● ● ● ● ● ●● ● ●●●●● ●●●●●●● ●●●●●●●●●●●●●●●●● ●● ● ●● ● ●● ● ● ●● ● ● ● ●●●●●●● ●●● ●●●●● ●●●●●●● ●● ● ● ● ● ● ●●●● ● ●● ●●●●●●●●●●● ●●●● ●● ●● ● ●● ●●●● ● ●● ●● ●● ●● ●●●●●●●●● ● ●●●●●●●●● ●●● ●● ● ● ●● ● ●● ● ● ●● ●●●●●●●● ●●●●●●●●●●●●●● ●● ●●●● ●● ● ● ● ● ●● ●● ● ●●●●● ●● ●●● ●●● ●●●●●●● ●● ●●● ●●●●●● ●● ●● ● ● ● ● ● ● ●● ● ●●●●●●●●●●●●●●● ●●●● ●●●●●● ●●● ● ●● ● ●● ● ●●●●●● ●●●●●●●●●●●●●●●● ●●●●● ●●● ●●● ●● ● ● ●● ● ●● ● ●● ● ●●●●●●●●●●● ●●●●● ●● ●●●● ●● ● ● ●● ● ●● ●● ● ● ● ● ● ● ● ●●●●●●●●● ●●●●●●●● ●●●● ●●●●● ● ● ● ● ●● ●●● ●●● ●●● ●●●●●●● ●●●● ● ●●●●●●●●●●●●●● ●● ●● ●● ●● ● ●● ●●●●● ●●●●●● ● ● ●●● ●●●● ●●●● ●● ● ● ●●●● ●●● ● ●●●●●● ●● ●●●●●●●● ●● ● ●●●●●●●●●●●●●●● ●● ●●● ●● ● ●●●●● ●●●●●●●●●●● ●●●● ● ● ●● ●●●●●●●●●●●●● ● ● ●●●● ● ● ● ●●● ●● ●● ●●●●●● ●●●●● ●●●●●●● ●● ● ● ● ● ● ● ● ● ●● ●●●●● ●●●●●●●● ●● ●●●●●● ● ●●● ● ● ● ● ● ●● ●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●● ● ● ● ●● ●●● ● ● ●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●● ● ● ● ● ● ●● ● ● ●●● ●●●●●●●●● ●●●●●●●●●●● ●● ●● ● ● ● ● ● ● ● ●●● ●● ●●●●●●●●●●●●●●●●●●●●●● ●●●● ●● ● ● ● ● ● ● ● ● ●● ●●●●● ●●●●●● ●● ●●●●●●●●●●● ● ●● ● ●●● ● ● ●●●● ● ●● ● ●● ● ●● ●●●●●●● ●● ●●●●●●● ● ●● ●●● ● ●● ● ● ● ● ●●● ●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●●● ●●● ● ● ● ● ● ● ● ● ● ●● ● ●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●● ● ●● ●● ● ● ● ● ●●●●● ● ● ● ● ● ●●● ●●●●●●●●●●● ●●●●● ●●●●●● ●●●●●●● ● ● ●● ● ●●●● ● ● ●● ●● ●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●● ● ● ●● ●● ● ● ●●●● ● ●●●●●●●●●●●●●●●● ●●●● ● ●● ●● ●● ● ● ●● ●● ● ● ● ● ●●●● ●●● ● ●●●●● ●●●●●●●●● ● ●●●● ●●● ● ●●●●● ● ● ● ●● ●● ●●●●●●●●● ●●●●●●●●●●●●●● ●●●●●●●● ●● ●● ● ● ● ● ● ● ● ●● ●●●●●●●●●●●● ●●●●●●●●● ●●●●● ●●●●●● ● ● ● ●● ● ● ●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●● ● ●●●● ● ● ● ● ● ●● ● ●● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ● ● ● ●● ● ● ●● ● ●●●●●●● ●● ●● ●●●●●●●●● ● ●● ● ● ● ● ● ●●●● ●●●●● ●●● ●●●●●● ● ●●●●●●●● ●● ●● ●●● ● ● ●● ●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●● ●● ● ● ● ●● ●●● ●●●● ●●●●● ●●●●● ●●●●●●●●●●● ● ●● ● ● ● ● ● ●●●● ●●●●● ●●● ●●● ●●● ●●●●● ● ●●● ● ● ● ●●●● ●●●● ● ●● ●●● ●●●●●●●●●●●●●● ● ● ● ● ● ● ● ●● ●●●● ● ●●●●●●●●●● ● ● ● ●●●● ●●● ●● ●●●● ● ● ● ● ●● ● ● ●●●●●●●●●● ● ● ●●● ●●●● ● ● ● ●●●● ● ●● ● ● ●●● ●●● ● ●●●●●●●●●●●● ●● ●●● ●●●●● ● ●●●● ● ● ●● ● ● ●●●● ●●●●●●● ●●● ●●●●●● ●●● ●●●●●●●●● ●● ● ● ●● ● ● ●● ● ● ●●●●●● ● ●●● ● ● ● ● ● ● ● ●● ●● ●●●● ●● ●●● ●●● ●● ●●●●● ● ●●●● ● ●● ● ● ●● ●●●● ● ●●● ●●● ●● ●●● ● ● ● ● 2 ● ● ● ●● ●●● ●● ● ●●● ● ●● ● ● ● ● ● ● ● ●● ●●●● ● ●● ●●●●● ●●● ● ● ● ● ● ●● ● ●● ●●● ● ●●● ●●●●●● ●●● ●● ● ● ●● ●●● ●●● ●● ●● ●●●● ●●●●● ●● ● ● ● ●● ● ● ●●● ●●●●● ●● ●●●● ●●● ●●●● ●● ● ● 2 ●●● ● ●●● ●● ●● ● ●● ●●● ● ● ●● ● ●● ● ●● ● ● ●● ● ●● ●● ●● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ●● ● ● ● ● - ● ● ● ●● ●●● ●●●● ●● ●● ● ● ● ● ● ● ● ●● ●● ● ●● ● ●●● ● ●● ● ● ● ● ● ●●● ● ●●● ●● ●●●●● ● ●●● ● ● ●●●●● ●● ● ●●●● ●●● ●●●● ● ● ●● ● ● ● ● ●● ● ●● ●● ●●● ●●● ● ● ● ● - ● ●● ● ● ●● ● ● ● ●●● ● ● ● ● ● ● ● ●● ●●● ●●●● ●● ●●● ●● ● ● ● ● ● ● ● ●● ●●●●● ●● ● ●● ●●●●● ● ●● ●● ● ●●● ● ●●●●●●●● ●●● ● ●● ●● ●●● ● ●●● ●● ●● ● ● ●● ● ●● ●● ● ● ●●●● ● ●●●● ● ●● ● ●● ● ● ● ● ● ● ●●●● ● ● ●●●●●● ● ● ●● ● ●● ●● ●●●● ●● ●● ●● ●● ● ●● ● ●● ● ●●● ● ●● ●●● ● ● ●●● ● ●● ● ●● ● ● ● ● ●●● ● ●● ●●●● ●●●● ● ●●● ●●● ●●● ● ● ● ●●● ● ● ● ● ●● ●●● ●●●●●●● ●●● ● ●● ● ● ●●●● ● ● ●● ●●● ●● ● ●● ●● ● ●●● ● ●● ● ●● ●● ●●● ● ●●● ●●● ● ●●● ● ●● ●● ●● ● ● ● ● ●● ● ●●●● ● ● ● ● ●●●●●●●●● ● ●● ● ●● ● ● ●● ●●● ● ● ● ● ●● ●●●●● ●●● ●● ●● ● ● ● ●●●●●●● ●●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ●● ● ● ●● ● ● ●●●●● ● ●● ● ● ●●●●●●●●●●●●● ●●●● ● ● ●● ● ● ● ● ●●●●●●●●●●●●●●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ●●●●● ● ● ●●●●●●●●●●●●● ● ●● ● ● ●● ● ● ● ● ● ● ●● ●●● ●● ●● ●●●● ●●●●●●● ● ● ● ● ● ● ● ● ● ● ●● ● ●●●●●●● ●●●●●●●● ●● ●● ●●●●● ●●●● ●● ●● ●● ● ●● ● ●●●●●● ●● ●●●●● ●●● ● ●●●● ● ● ● ● 2 ●● ● ● ● ● ●● ● ●●●●●● ●●● ● ● ● ● ● ● ● ●●● ●● ● ● ● ●●● ●● ● ● ● ● ● ● ● ● 2 ●● ● ●● ● ●●●●●● ● ● ● ● ● ● ● ●● ●●● ●●●●●●●●●● ●●● ● ● ●● ●●●●● ● ●●● ● ● ● ● ●● ●●●●● ● ●●●● ● ● ● ●●● ● ●●●●● ● ●● ●● ● ● ●● ●● ● ●● ● ● ● ● ●●● ●●●●●●●● ●●● ● ●● ● ● ● ● ● ● ●● ●● ●●●● ●●●●● ●●● ●●●●●●●●●●● ●● ●●●●● ● ●● ●● ● ●●● ● ● ● ●●●● ●●●●● ●●●●●●●●●●●●●● ●● ●●●●●●● ● ● ●●●● ● ● ● ● ● ● ● ●●●●●●●● ●●● ● ● ●●●●●●● ● − ●● ● ● ● ● ●● ● ●●●●● ● ● ● ●●● ● ● ●●● ●●●●● ● ● ●●●● ● ● ● ● ● ● ●●●●● ● ●●● ●●● ●●●●●●● ● ● ● ● ●● ● ● − ● ●● ●●●● ● ● ● ●● ●● ●●●●●●●●●● ● ● ●●●● ● ● ● ● ● ●●● ●●●●●●●●●● ● ●●●● ●●●● ●●●●●● ● ●●● ●●● ●● ● ● ● ●●●●●● ● ● ●●●●●●●●●●●● ●●●●●●● ● ●●● ●● ● ●● ● ●●● ●● ●●●●●●●●●●●●●●●●●●● ●●●●● ●●● ● ●●● ● ● ●●●● ●●●● ●●● ●●●●●●●●● ●●●●●● ●●●●●●●● ● ● ● ● ● ●● ● ●●●●● ● ●● ● ● ● ●●● ● ●●● ● ●●● ● ● ● ● ● ● ●● ●● ●●●●● ● ● ●● ●●●●●●●●●●● ●●●●● ●●● ●● ●●●●●● ● ● ●●● ● ●●● ●● ●●● ● ●●●●●●● ● ●●●●●●●●●● ● ● ●● ●● ●●●●●●● ● ●● ●●●● ●●●● ●●●●● ● ● ● ● ● ●●●● ●●●●●● ● ● ● ● ●●●●●●● ●●●●●●●●●● ● ●●● ●● ● ● ● tSNE ●● ●●●● ● ●●● ● ●● ●●● ●●● ● ●●●●● ●● ● ● ●●●●●●● ● ●●●● ●●●●●●●● ●●●●●●● ●● ● ●●●●●● ● ●● ● ● ●●●● ●●●●●● ●●●●●●●●●●●●● ●● ● ●● ● ● ● ● ● ●●●●●● ●●●● ●● ●● ●●●●●●● ● ●● ● ● ●●● ●● ●●●●●●● ●●●●● ●● ● ●●●●●● ● ●● ●●● ● ●● ● tSNE ●● ●●● ● ●● ●● ●● ●●● ● ● ● ● ●● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ●●● ● ● ● ●● ●●● ●●● ●●●●●●●●● ● ● ●●● ● ● ● ●●●●● ● ●● ●● ●●●● ●● ●● ●●●●●●● ●● ● ● ● ● ● ●● ● ●●●● ● ●●●●●●●●●●● ● ● ● ● ● ●● ● ● ● ●● ● ● ●●●● ●● ● ● ● ● ●● ●●● ●●●● ● ●● ● ● ● ●●●● ●●●● ● ● ● ● ●●●● ● ● ●● ● ●● ● ● ● ●●●●●●● ● ●● ●● ● ●●●●●● ● ●●● ● ● ●● ●●● ● ● ● ●● ●●●●●● ● ● ● ● ● ● ●● ●●●● ● ● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ●●●●● ● ●● ● ●● ● ● ● ● ● ● ● ● ●● ●● ● ● ●●● ● ●●● ●● ●● ● ●●● ● ● ● ● ●●● ● ● ●●● ●● ● ●●● tSNE ●● ● ●● ●● ● ● ● ●● ● ● ● ●●● ● ●● ● ● ●● ●● ●● ● ● tSNE ●● ● ● ●● ●● ●●● ● ● ●● ● ●●● ● ●● ●● ●●● ●● ●●●●● ●●●●●● ●● ● ●● ●●● ●● ● ● ● ●●● ● ●●● ● ●●● ● ● ●●● ● ●●● ●●● ● ● ● ●●●● ●●● ●●● ● ●● ● ● ●● ●●● ●●● ● ● ● ● ● ● ●● ●●● ●● ●●● ● ● ● ● ● ● ● ●● ●● ●● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ● ● ●● ● ●● ●●● ●●● ● ● ● ● ●●●●● ● ●●● ●●● ●● ●●● ●●●●●●● ● ●●● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ●●●●●● ● ●● ● ●● ● ● ●● ● ● ●● ● ●● ● ● ● ●● ● ●● ● ● ● ●● ● ● ●● ●●●● ●● ●● ●●●●● ● ● ●● ●● ●● ● ●●●● ●●●● ● ●●● ● ● ● ● ● ● ● ● ● ● ●●●● ●● ●● ● ● ● ● ● ● ●● ● ● ● ●●●●● ●●● ● ● ●● ● ● ● ● ● ● ●●● ●● ●●● ● ●●●● ● ● ●● ● ●● ●●● ● ●●●● ●●●●●●●●●●● ●●● ● ●●● ● ● ● ●●●● ● ● ●●●●●●● ●●●●● ● ●●● ● ●● ●●● ●● ●● ● ● ●●●●●● ●● ●●● ● ● ●● ● ● ● ● ●● ● ●●● ● ● ●● ●●●●● ●● ●● ● ● ● ● ● ●●●●●●●● ●●● ● ● ● ● ●●●● ●●●●● ● ● ● ●●●● ●●●● ●● ● ● ● ● ●● ● ● ● ●●●●●● ●● ● ●● ●●● ● ● ● ●●● ●●●● ● ●● ●●● ●● ● ●●●●●● ● ●● ● ● ● ● ●● ● ● ●● ● ● ● ●● ● ●●●● ● ●●●● ● ●● ●●● ●● ● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ●● ●●●● ●● ● ● ● ●●●● ●● ● ●● ●●●●● ●●● ● ●●● ●● ● ● ● ●●●● ● ● ● ● ● ● ●●● ● ●●●●●●●● ● ● ● ● ●● ● ●● ● ● ● ●●● ●● ● ● ● ●● ●● ● ●●● ● ● ●● ● ● ● ● ●● ●● ●●● ● ● ● ● ● ●● ●● ● ●● ●● ● ●● ● ● ● ●●●● ● ● ●● ●● ●● ● ●●●● ● ●● ●● ●● ●● ● ● ● ●●● ●● ● ●●● ●● ●● ●● ● ● ●● ● ● ●● ● ● ●● ● ● ●●● ● ● ●● ● ● ● ● ●● ●● ●● ● ● ● ●● ● ● ● ● ● ● ●●●●●● ● ●●●● ● ● ●●● ● ● ● ● ● ● ● ●●●● ● ●●● ● ● ●● ●● ● ● ● ● ● ●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ●●● ● ● ● ● ● ●● ●● ● ●●●● ● ●● ● ● ● ● ● ●● ● ●●●●● ● ● ● ● ● ● ● ● ● ● ●● ●● ●●● ●● ●●● ● ● ● ● ● ●● ● ●● ●●● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ●●● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●●●● ●● ●●●● ● ● ● ● ● ● ● ● ● ● ●●● ●● ● ●●●●●● ●●●●● ● ● ● ●● ● ●●● ●● ●●●●●●● ●●● ● ● ● ● ● ●● ● ●● ● ● ●●●●● ●●●● ● ● ● ● ● ● ●● ● ●●●●●●●●● ● ● ●● ● ● ●●●● ●●●● ●● ● ●● ●● ●●●●●● ● ● ● ●● ●●●●●●●●●● ● ● ● ● ●●●●● ●● ●●●●● ● ● ● ●● ● ●●●●●●●● ●● ● ● ●●● ● ● ● ● ●●●●●●●● ●●● ● ● ●●● ● ● ●●● ● ●●●●● ●●● ●● ● ● ● ●● ●●●● ● ●●● ●● ● ● ●● ● ●●●●● ● ●● ●● ● ● ● ● ● ●● ●● ●●●●●● ●●● ●● ●●● ● ● ● ●● ●●●● ● ●●● ● ●● ● ● ●●● ●● ●●●●●●●●●● ● ● ● ● ●●●● ●● ● ● ●●●● ● ● ● ● ● ●● ● ●●●● ●●● ● ●● ●● ● ● ●● ● ● ● ● ● ●● ●● ●●● ● ● ● ● ● ●● ●●●● ●● ●● ●● ● ●● ●● ●●●● ● ●●●● ● ●●● ● ● ● ●● ● ● ● ●● ● ●●●● ●● ● ●● ● ● ● ● ● ●●●●●●●●● ●●●●●● ● ● ● ● ● ● ● ● ●●●●● ●●●●● ●●●●●●●● ● ● ● ●● ● ● ●●● ●●● ● ●●● ● ●●● ● ● ● ●●● ● ●●● ● ● ● ● ●● ● ● ●●●●●●● ● ● ●● ●● ● ● ● ● ● ● ● ●● ●●●● ● ● ● ●● ● ● ● ●●●● ●●● ●● ●● ● ● ●● ● ● ● ● ● ● ●● ●● ● ●● ● ●● ●● ●● ●● ●● ● ●●●● ●● ● ● ● ● ●● ● ●● ●●●● ● ● ●● ● ● ●● ●● ● ●● ●● ● ● ●●● ● ● ● ●● ● ● ●● ●●● ●●●●●●● ●● ● ● ● ● ●● ●●●●●●● ● ●●●● ● ●●●● ●●●●● ●● ● ●●● ●● ●●●● ● ●● ●●● ●● ●●●● ●●●● ● ●●●● ● ●● ●●●●●●●● ●●●●●●●●●●●●●●●● ● ● ● ● ●●● ● ●● ●●●●●●●●●●●● ●● ● ●● ●●●●●●●● ●●●●●●●●●●● ● ●●● ● ●●● ●●●●●●●●● ● ●●● ● ● ●● ● ●●● ● ●●●● ●● ●●●●● ● ● ●●● ● ● ● ● ●● ●● ●●●●● ● ● ● ● ●●● ● ● ●●●●●●●●●●●● ●●● ● ●● ● ● ●●●● ●● ●●● ● ●●●● ● ● ● ●● ● ●● ● ● ●● ● ●●● ●● ●●● ● ● ●● ●●● ● ●● ●● ● ●● ●● ● ● ●● ● ●●●● ●●● ●●●● ● ●●● ●●●●● ● ●●●● ● ●● ●● ● ● ●●●●● ●● ●● ● ●● ●●● ●●●● ● ● ● ●●● ●●●●●● ● ● ● ● ●● ●● ●● ●● ● ● ● ●●●● ●●●●● ● ●● ● ●● ●●●●● ● ● ●● ●●●●●● ●●● ●● ●●●●●● ● ●●● ● ● ● ●● ● ● ●●●● ●●● ●●●● ●●●●● ● ●●●● ● ●●●●● ● ●●● ● ●●●● ● ●● ●● ● ● ● ●●●●●● ●●● ●●●●●●● ●● ● ● ●●●●●●● ●●● ●●●● ●●● ● ● ● ●●●●● ●●● ●●● ●●● ●●●●●●● ●●● ●●●●●● ● ● ●●●●●●● ● ●● ●●● ●●●● ●●● ● ● ● ● ● ●●●● ●● ●●● ●● ● ● ●●●● ●● ●●●● ●●● ●●●● ●●●●●● ● ●●● ● ●● ●● ●●●●●●●●●●● ●●●● ● ●●●●●●●●●●●●●● ●●●●●●●● ● ● ● ●●●●●●●●●● ●●●●●●● ●● ●● ● ●●●●●●● ●●● ●● ●●●● ●●●●●●● ●●●●●●● ● ● ● ●● ●●● ●● ●● ● ●● ●●● ● ●●●● ●●●●●● ●● ●●●●●● ●●● ● ●● ●●● ●●●●● ●●● ● ●● ● ●●●● ●●●● ●●●● ●● ●● ●●●● ●●●●●●●● ●● ● ●●●● ●● ●●● ●●● ●●●●●●●●●●● ● ● ●●● ● ● ●● ●●●●●●●●● ●●●● ●● ●●●●●● ●●●● ●●● ● ● ● ● ● ●●●● ●●●●●●●● ● ●●● ●● ● ● ●●●●●●●●● ●●●●●●●●●●●●● ●●● ● ● ● ● ●●●●● ● ● ●●● ● ●● ●● ●●●● ●● ●●●●● ●●●● ●●●● ● ●●●●●●●●●●● ●● ● ● ● ● ● ●●●●●●●● ● ●●● ● ●●●●● ●●● ●●● ●●●●●●●●●● ●●●● ●●●● ●●● ●●●●●● ● ●●● ● ● ● ●●●●●●●● ● ●● ● ●● ● ● ● ● ● ● ●● ● ●●●●● ●●●●●●●●●●●●● ●●●●●●●● ●●● ● ● ●●●●●● ● ●● ● ● ●● ● ●● ● ● ● ●● ●●● ●●●● ●● ●●●●●●●●●●●● ●●●●●● ● ●● ● ● ●●●●●●● ● ●● ● ●●●●●● ● ● ● ● ● ●●●●●●●● ●●●●●●●●●●●●●●● ●●●● ● ● ● ● ●●●●●●●● ● ●●● ●● ●●● ● ●● ●● ●●● ●●●●●●●●●●●● ●●●●● ●●● ● ● ● ● Activity (Z−Score) ● ●●●●●●●●●● ●● ●● ●● ●● ●● ●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●● ● ● ● ●●●●●●● ● ●● ●● ● ● ●●●●●● ●● ●● ●●●● ●●●●●● ●●●●●● ●●●●●●●●●●●●●●●● ●●●●● ● ● ● ● ●●●●● ●●●●●●●●●●●●●● ●●● ●●●●●● ● ●● ●●● ●●●●●●●●●●●●●●●●● ●●● ● ●●●●● ● ●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●● ● ●● ●●●● ●●●●●●●●●●●●●●●●●● ●●● ●●● ●● ● ● ●● ● ●●●●●●●●●●●●●●● ●●●●●●●● ●●●● ●● ●●●●●●●●●●●●●●● ●●●● ●● ●●● ●● ●●●●●●●●● ● ●●●●●●●●●●●●● ●● ●●●●●●●● ●●● ● ●●●●●●●●●●●●●●●●●●●●●● ●●● ●●●● ●●● ●●●●●●●●● ● Activity ● ● ●●●●●●●●●●●● ● ● ●●●●●●●●●●●●●●●●●●● ●● ● ●● ● ●● ● ● ●●●●●●●●●●● ●●● ● ● ● ●● ●●● ●●●●●●● ● ●● ● ● ● ● ● ●●●●●●●●●●●●●●●●● ●● ●● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●● ●●●● ● ●● ● ● ● ●● ●●●●●●●●●●●●●●●● ● ●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●●● ●● ●●●●●●● ●●●● ● ●● ● ●●●●●●●●●● ● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ●● ● ● ● ●● ● ●●●●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●● ●●●● ● ● ●● ●●●●● ●●● ●●●●●●● ●●●●●●● ●●●●●●●●●● ● ● ● ●● ● ● ● ● ● ● ● ●●●● ●●●● ● ●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●● ● ● ●●● ● ● ● ● ● ● ● ● ●●●●●●●●●●●●●● ●● ● ●●● ●● ●● ●●● ● ●● ● ●● ●●● ● ●●●●●●●●●●●●●● ●●●●●●●●●● ●●● ●●●●● ● ● ● ● ●● ● ● ●●●● ●●●●●●●●●●● ●●● ●●● ●●●● ●● ● ●● ● ●● ● ●● ● ●●●●●●●●●●● ● ●●● ●●● ●● ● ●● ●●●●●●●●●●●● ●●●●●● ●● ●● ● ●● ●●●●● ●●●●●●●●●●●●●● ● ●●●●●●● ● ● ●● ● ● ● ●●● ●● ●●●● ● ● ● ●● ●● ●●●●●●● ●●●●●●● ●●● ●● ●● ● ● ● ● ● ● ● ●● ●● ●● ●● ●● ● ● ●●● ● ● ● ● ● ● ● ●● ● ●● ●●● ● ●●● ● ● ●● ● ● ● ●● ● ● ● ●● ●● ●● ● ● ●● ● ●●● ●● ●● ● ●●● ● ●●●●●● ●●●●●●● ● ● ●●● ●● ●● ● ●●● ●●●● ● ●●●●● ● ● ● ● ● ● ●●● ●●●● ● ●● ● ●●● ● ● ● ● ●● ●●●● ●●●●● ●● ●● ●● ● ● ●● ●● ●● ●●● ● ● ●● ● ● ● ● ●● ●●● ● ● ● ●● ●● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ●● ● ●●● ● ● ● ● ● ● ●●● ●●●● ● −2 0 2 ● ●●● ● ●●● ●● ● ●●● ● ●● ●●● ● ● ● ●● ● ● ●● ●● ●● ●● ● ●● ● ● ●● ● ●● ●●●●● ● ● ● ●● ●● ● ● ●● ●● ● ● ● ● -2.5 0 2.5

Expression (log2(RPM+1)) tSNEtSNE-1 −1 tSNEtSNE-1−1 0 9 18

Figure 2. Metabolic inflexibility and stemness markers define two major clusters in ND and T2D islet cells. (A, B) Selected markers of metabolic inflexibility PPARa (A) and PPARg (B) protein activity plotted in t-SNE. (C, D) Endocrine progenitor marker, RFX6 (C) and its cognate factor RFX7 (D) protein activity presented in t-SNE space. (E, F) Stemness markers, NANOG (E) and MYCL (F) protein activity plotted in t-SNE at a single-cell level. A1 A2 A3 A A4 A5 A6 Fig.3 B1 B2 B3 B4 B5 B6

A1 A2 A3 A4 A5 A6 B1 B2 B3 B4 B5 B6 2 18 SST Score) −

GCG 0 9

INS Activity (Z Expression Expression (log2(RPM+1)) (log2(RPM+1) 0 9 18 MAFA 2 0 PDX1 − NKX2-2 4 1 6 18 1.5 NEUROD1 14552 MAFB Islet 9 0 0 0 0

PAX6 #Genes Activity Activity Activity Activity Ac tivity PAX4 Expression (log2(RPM+1)) differentiation ISL1 4 1 6 0 4 0 4 1.5 − − − − 1901 IRX2 - ARX GLI3 4 1 6 18 1.5 IGFBP2 14552 ITGB8 9 0 0 0 0 HSPB1 #Genes Activity Activity Activity Activity Expression (log2(RPM+1)) cell factors F10 Ac tivity - 4 1 6 0 1.5 6 0 6 − − − a SPOCK3 − 1901 MYO10 - CLU RFX7 4 1 6 18 1.5 RFX6 14552 FOXO1

PPARA 9 0 0 0 0 #Genes Activity Activity Activity Activity Ac tivity

PPARG Expression (log2(RPM+1)) RB1 4 1 6 4 0 4 0 1.5 FOXM1 − − − − 1901 - progenitor/ inflexibility HES1 4 1 6

NEUROG3 18 1.5 14552 POU5F1 9 0 0 0 0 MYCL #Genes Activity Activity Activity Activity Expression (log2(RPM+1))

NANOG Ac tivity 4 1 6 0 1.5 − − − − 1901 1 0 1 - Stemness B Combined C ND D T2D 2 2 2 - - - tSNE tSNE tSNE

tSNE-1 tSNE-1 tSNE-1 E F A1 A2 A3 A4 03 02 13 08 10 71 22 80 ------# Cells p=1.9e p=1.0e p=9.6e p=3.4e p=8.2e p=7.9e p=2.2e p=1.0e T2D6 ND1 ND2 ND3 ND4 T2D2 T2D6 T2D6 T2D6 T2D1 T2D5 ND1 ND2 ND3 ND4 T2D2 ND1 ND2 ND3 ND4 T2D2 ND1 ND2 ND3 ND4 T2D2 T2D4 T2D3 T2D1 T2D1 T2D1 T2D5 T2D5 T2D5 T2D4 T2D4 T2D4 1.0 T2D3 T2D3 T2D3

T2D A5 A6 B1 B2 0.8

0.6 # Cells

0.4 T2D6 T2D6 T2D6 T2D6 ND1 ND2 ND3 ND4 T2D2 ND1 ND2 ND3 ND4 T2D2 ND1 ND2 ND3 ND4 T2D2 ND1 ND2 ND3 ND4 T2D2 T2D1 T2D1 T2D1 T2D1 T2D5 T2D5 T2D5 T2D5 T2D4 T2D4 T2D4 T2D4 T2D3 T2D3 T2D3 T2D3

0.2 B3 B4 B5 B6 ND 0.0 1 2 3 4 5 6 All B1 B2 B3 B4 B5 B6 All 71 22 A A A A A A 13 08 10 80 − − − − − − # Cells C1, p=0.64 C4, p=0.01 C5, p=0.46 C6, p=0.46 C8, p=0.039 C2, p=0.0019 C12, p=1e C3, p=9.6e C7, p=3.4e C9, p=8.2e T2D6 T2D6 T2D6 T2D6 ND1 ND2 ND3 ND4 T2D2 ND1 ND2 ND3 ND4 T2D2 ND1 ND2 ND3 ND4 T2D2 ND1 ND2 ND3 ND4 T2D2 T2D1 T2D1 T2D1 T2D1 T2D5 T2D5 T2D5 T2D5 T2D4 T2D4 T2D4 T2D4 T2D3 T2D3 T2D3 T2D3 C10, p=7.9e C11, p=2.2e Figure 3. iTerClust classifies ND and T2D islet cells in different biological states. (A) hclustering using the iterCluster algorithm structures cell positioning and sub-groups. Each sub-group is color-coded. Each bar denotes a single cell. Black bars represent T2D cells and white bars ND cells. SST, GCG and INS mRNA expression is plotted at a single-cell level. (B-D) For illustration purpose, sub-clusters were projected onto 2D t-SNE space according to metaVIPER inference as ND and T2D combined (B), or as ND only (C) and T2D only (D). (E) Bar-plots presenting the percentage of ND or T2D cells in each sub-cluster. A dashed line represents the proportion of cells from ND or T2D islets subjected to scRNA-seq analyses. (F) Bar-plots showing the number of cells from individual donors in each subcluster. Fig.4

A ND cell conversion to DeDi state B ND cell conversion to DeDi state infection infection - - NFATC2 NFATC2 ZNF385D DeDi Signature ZNF385D BACH2/TSHZ2/ BACH2/TSHZ2 Non BACH2/TSHZ2/ RARB/GAS7/ BACH2/TSHZ2 Non BFP CTRL NFATC3 ZRANB3 BFP CTRL NFATC3 TCF4 ZRANB3 AFF3 RARB/GAS7/ TCF4 AFF3 1.0 1.5 + 0.8 +

0.6

0 ● Signature ● ● ● ● ● ● 0.4 ● ●

DeDi Signature 0.2 - Cell Percentage DeDi −

1.5 0.0 − 19 36 - - All 19 36 30 19 30 19 - -

Ctrl C12 C1 C2 C3 C8 C9 C10 C17 − − − − p=0.5 p=0.41 p=0.002 p=0.0013 p=0.002 p=0.3.3e p=0.3.6e p=0.001 p=2.1e P=0.5 P=0.41 p=3.3e p=3.3e IRX2_ZNHIT1_ZFPL1_PAX6_DRAP1p=3.6e p=2.1e p=3.3e 4 2 C D IRX2 a 1.0 a

ND cell conversion to B6-like cell state 1 ND cell conversion to B6-like cell state ● + ● 0.8

0 ● ● ● ● ● ● ● Irx2 ● ● ● ● 0.6 0 ● ● ● ● 0.4 ● 1 4 − − 0.2

infection − infection - Ctrl C12 C1 C2 C3 C8 C9 C10 C17 2 - NFATC2 NFATC2 IRX2_ZNHIT1_ZFPL1_PAX6_DRAP1 0.0 − ZNF385D ZNF385D All 11 05 12 05 34 BACH2/TSHZ2/ BACH2/TSHZ2 Non RARB/GAS7/ BFP CTRL NFATC3 ZRANB3 BACH2/TSHZ2/ TCF4 AFF3 − − − − − BACH2/TSHZ2 Non RARB/GAS7/ BFP CTRL NFATC3 ZRANB3 TCF4 AFF3

2 1.0

p=0.0041 + p=0.00021 p=0.00062 0.8 p=4.1e p=4.6e p=1.9e p=1.6e p=8.7e + 1

● 0.6 ● ● ● 0 ● ● ● ● ● 0.4 1 − like Signature 0.2 - - Cell Percentage − 2 − B6 0.0 22 25 12 37 17 05 - - - - - DeDi Signature- All 1.0 37 17 05 22 25 12 + 1.0 − − − − − − 0.8 p=0.008 p=2.2e p=3.9e p=1.1e p=4.6 e p=0.005 p=3.5 e p=6.9 e +

0.6 0.8 p=0.0052 p=0.00081 p=2.2e p=3.9e p=1.1e p=4.6e p=3.5e p=6.9e

0.4 1.5 0.6 0.2 − 0.4 E 0.0 F

0 ● 0.2 ● All ● 37 17 ●05 ●22 ●25 ● 12 − ND cell conversion to IRX2 active cell state● − − − − ● − − ND cell conversion to IRX2 active cell state 0.0 DeDi Signature p=0.0052 All 19 36 30 19 p=0.00081 p=2.2e p=3.9e p=1.1e p=4.6e p=3.5e p=6.9e − − − − 1.5 p=0.5 p=0.41 − p=0.002 p=0.0013

Ctrl C12 C1 C2 C3 C8 C9 C10 C17 p=3.3e p=3.6e p=2.1e p=3.3e infection - infection NFATC2 - NFATC2 ZNF385D ZNF385D BACH2/TSHZ2/ BACH2/TSHZ2 Non

IRX2 RARB/GAS7/ BFP CTRL NFATC3 ZRANB3 TCF4 AFF3 BACH2/TSHZ2/ BACH2/TSHZ2 Non RARB/GAS7/ BFP CTRL NFATC3 ZRANB3 TCF4 AFF3 1.0 4 + 0.8 + ● ● 0.6

0 ● ● ● ● ● ● Irx2 ● 0.4

IRX2 activity 0.2 Cell Percentage - − 4

− 0.0 11 05 04 12 05 34 04 ------All Ctrl C12 C1 C2 C3 C8 C9 C10 C17 11 05 12 05 34 − − − − − p=4.1e p=4.6e p=2.1e P=0.004 P=1.9e P=1.6e P=8.7e P=6.2e p=0.0041 p=0.00021 p=0.00062 p=4.1e p=4.6e p=1.9e p=1.6e p=8.7e Figure 4. scGOF-seq experiments in human islets. (A) Violin plots showing the distribution of cells following transduction with each individual candidate or combination thereof analyzed according to DeDi signature, an intergrated value of RFX6, RFX7, FOXO1, PPARa, PPARg, RB1, POUF51, NANOG and MYCL protein activities. Non-transduced and BFP-transduced ND islets serve as negative controls. (B) Bar-plots showing the proportion of islet cells with a positive DeDi signature (> activity 0) in each scGOF-seq condition. A red dashed line indicates the percentage of islets cells with a positive DeDi signature in non-transduced negative controls. (C) Violin plots showing cells with a B6-like signature, which is an intergrated value of IRX2, ZNHIT1, ZFPL1, PAX6 and DRAP1. (F) Bar-plots showing the proportion of islet cells with a B6-like signature as in (B). (E) Violin plots showing cells with IRX2 activity as in (A). (F) Bar-plots showing the proportion of islet cells with a positive IRX2 activity as in (B). A Fig.5

Human ND islets

Virus Infection Partial Digestion & Plating Dye Loading Calcium imaging Candidate_BFP Singe Cell Analysis RIP-zsGreen

2.8 mM Glucose AFF3 B C 16.8 mM Glucose BFP_CTRL AFF3AFF3 40 mM KCl BFP CTRLBFPBFP CTRLCTRL BFP CTRLBFPBFP CTRL CTRL BFP CTRL BFP CTRL

1.5 1.5 1.5 2 1.5 1.5 - 2

1.5 -

1.0 1.0 1.0 1.0 1.0 AFF3 AFF3 BFP CTRL BFP CTRL BFP CTRL BFP CTRL 0.5 0.5 0.5 0.50.5 Normalized Rhod Normalized Rhod 0.0 0.0 0.0 0 0.00.0200000 400000 600000 800000 1000000 1200000 0 2000000 200000200000400000 400000400000600000 600000600000800000 8000001000000800000 100000010000001200000 12000001200000 0 200000200000Time400000 (ms)400000600000 600000800000 1000000800000 12000001000000 1200000 Time (ms)TimeTime (ms)(ms) Time (ms) Time (Time (ms)msTime) (ms)

AFF3 AFF3 D E BACH2BFP CTRL BFP CTRL TCF4BFPTCF4 CTRL BFP CTRL BACH2

1.5 1.5 1.5 1.5 1.5 1.5 2 2 - - TCF4 1.0 Bach2 1.0 1.0 1.0 AFF3 0.5 AFF3 BFP CTRL BFP CTRL TCF4 0.5 0.5

Bach2 0.0 0.5 0 500000 1000000 1500000

Time (ms) Normalized Rhod Normalized Rhod 0.0 0.0 0 200000200000 400000400000600000 600000800000 1000000800000 12000001000000 1200000 0 200000200000 400000400000600000 600000800000 1000000800000 12000001000000 1200000 0.0 Time (Timems (ms)Time) (ms) Time (Time ms(ms)Time) (ms) 0 500000 1000000 1500000 Time (ms) Figure 5. Single-b-cell Calcium microfluorimetry. (A) Schematic drawing of the single-cell Ca2+ imaging procedure. (B)-(E) Representative traces of Ca2+ flux measured by Rhod2 loading in Ad-BFP (B), Ad-AFF3 (C), Ad-BACH2 or Ad-TCF4 transduced primary human b-cells. Red arrows indicate the timing of addition of 16.8mM glucose, and black arrows indicate addition of 40mM KCl.