bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder.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.

1 Deletion of encoding PU.1 and Spi-B leads to B cell acute lymphoblastic leukemia associated

2 with driver mutations in Janus Kinases

3

4 Carolina R. Batista*†, Michelle Lim*†, Anne-Sophie Laramée*†, Faisal Abu-Sardanah*†, Li S. Xu *‡,

5 Rajon Hossain*†, and Rodney P. DeKoter*†

6

7 *Department of Microbiology & Immunology and the Centre for Human Immunology, Schulich School

8 of Medicine & Dentistry, Western University, London, Ontario, Canada.

9 †Division of Genetics and Development, Children’s Health Research Institute, Lawson Research Institute,

10 London, Ontario, Canada.

11

12 This work was supported by the Canadian Institutes of Health Research Grants MOP-10651 and MOP-

13 137414 (to R.P.D.), a grant from the Leukemia and Lymphoma Society of Canada (to R. P. D.), and an

14 Ontario Trillium Scholarship (to C.R.B.)

15

16 Corresponding author: Rodney P. DeKoter, Department of Microbiology & Immunology, Schulich School

17 of Medicine & Dentistry, Western University, London, Ontario, Canada N6A 5C1. Phone: (519) 661-

18 2084; Fax: (519) 661-3499; E-mail address: [email protected]

19

1 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder.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.

20 Abstract (200 words)

21 Precursor B-cell acute lymphoblastic leukemia (B-ALL) is associated with recurrent mutations that occur

22 in cancer-initiating cells. There is a need to understand how spontaneous driver mutations influence clonal

23 evolution in leukemia. The ETS-transcription factors PU.1 and Spi-B (encoded by Spi1 and Spib) execute

24 a critical role in B cell development and serve as complementary tumour suppressors by opposing the

25 proliferative events mediated by IL-7R signaling. Here, we used a mouse model to conditionally delete

26 Spi1 and Spib genes in developing B cells. These mice developed B-ALL with a median time to euthanasia

27 of 18 weeks. We performed RNA and whole-exome sequencing (WES) on leukemias isolated from Mb1-

28 CreΔPB mice and identified single-nucleotide variants (SNVs) in Jak1, Jak3 and Ikzf3 genes, resulting in

29 amino acid changes and in the gain of early stop-codons. JAK3 mutations resulted in amino acid

30 substitutions located in the pseudo-kinase (R653H, V670A) and in the kinase (T844M) domains.

31 Introduction of these mutations into wild-type pro-B cells conferred survival and proliferation advantages.

32 We conclude that mutations in Janus kinases represent secondary drivers of leukemogenesis in the absence

33 of Spi-B and PU.1 transcription factors. This mouse model represents an useful tool to study clonal

34 evolution and tumour heterogeneity in B-ALL.

2 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

35 Introduction

36 Acute lymphoblastic leukemia is the most common type of childhood cancer, with approximately

37 6000 new cases diagnosed in the United States each year1. Most leukemias originate within the B cell

38 rather than the T cell lineage2,3. Precursor B cell acute lymphoblastic leukemia (pre-B-ALL) is a disease

39 that is revealed by the presence of transformed precursor B cells in the blood, bone marrow, and tissues;

40 and is most common in 1-5 year old patients4. Most pre-B-ALL cases are associated with genetic

41 abnormalities that include chromosomal translocations or point mutations. In pre-B-ALL, up to two thirds

42 of genes with point mutations encode transcriptional regulators such as Pax-5, Ikaros, or EBF13. Pre-B-

43 ALL cells are frequently arrested at an early stage of development, express interleukin-7 (IL7R),

44 and have high levels of Janus Kinase (JAK)-STAT signaling to sustain survival and proliferation5–7.

45 Activating mutations of the IL7R have been described in human pre-B-ALL8. JAK and IL7R

46 mutations are frequent in several subtypes of pre-B-ALL including the recently described disease Ph-like

47 leukemia9,10. In summary, mutations that activate cytokine signaling, and impair differentiation, function

48 as driver mutations in pre-B-ALL.

49 PU.1 (encoded by SPI1) and Spi-B (encoded SPIB in mice) are transcription factors of the E26-

50 transformation-specific (ETS) family11. These two share a conserved DNA binding domain and

51 interact with an overlapping set of DNA binding sites within the genome12. PU.1 and Spi-B complement

52 one anothers function, and activate multiple genes involved in B cell receptor signaling12–15. Lack of these

53 factors in developing B cells results in a block to B development at the small pre-B cell stage associated

54 with impaired Ig light chain rearrangement15,16. Importantly, conditional deletion of Spi-B and PU.1 in

55 developing B-cells leads to high incidence of B-ALL in mice, but the mechanisms of leukemogenesis in

56 the absence of these transcription factors are still undetermined17. In conclusion, PU.1 and Spi-B are

57 required for B-cell development, and function as complementary tumour suppressors in the B cell lineage.

3 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

58 B cell neoplasms, like all cancers, are thought to be diseases in which there is clonal evolution

59 from a common precursor, in which acquired gene mutations drive an evolutionary natural selection

60 process18,19. The mechanisms by which cancer-initiating cells respond to selection pressures during clonal

61 evolution have been classified into a number of common hallmarks20. In response to selection pressure,

62 the genetic makeup of cancer-initiating cells changes during the course of disease due to acquired

63 mutation. Mutations can be broadly classified as drivers or passengers18,21. Driver mutations are those that

64 provide a growth advantage to a cancer clone, whereas passenger mutations do not provide a growth

65 advantage. Pediatric B-ALL is less curable upon relapse due to clonal evolution of the leukemia, resulting

66 in driver mutations inducing a more aggressive disease22. High levels of intratumoral heterogeneity of

67 mutations is a poor prognostic marker for leukemia23. Whole-exome or whole-genome sequencing of pre-

68 B-ALL cases is expected to lead to a deeper understanding of the genetic causes of this disease, ultimately

69 permitting molecular targeted therapy for individual patients2.

70 In this study, we investigated the molecular features of leukemogenesis in a spontaneous model of

71 B-ALL induced by deletion of genes encoding PU.1 and Spi-B. We generated Mb1+/CreSpi1lox/loxSpib-/-

72 mice, called here Mb1-CreDPB mice16. We found that Mb1-CreDPB mice developed pre-B-ALL

73 characterized by the high expression of IL-7R, with a median time to euthanasia of 18 weeks. Using

74 whole-exome sequencing (WES) and RNA-seq, we identified single-nucleotide variants (SNVs), most of

75 which were predicted to have a role in the control of cell proliferation, communication and metabolism.

76 Strikingly, we identified recurrent SNVs in genes encoding Aiolos, Jak1, and Jak3 in mouse leukemias.

77 Further analysis revealed that SNVs located in Jak3 resulted in three different types of amino acid

78 substitutions within the pseudo-kinase domain (R653H, V670A) and kinase domain (T844M). We

79 confirmed the ability of these mutations to provide survival and proliferation advantages to normal pro-B

80 cells. In summary, this study shows that Jak3 mutations are secondary drivers of leukemogenesis in the

4 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

81 absence of Spi-B and PU.1. This mouse model may be useful to determine the effects of molecular targeted

82 therapies on intratumoral heterogeneity and clonal evolution in B-ALL.

5 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

83 Materials and Methods

84 Mice and breeding

85 Mb1-Cre mice have been described previously24. For this study, Mb1-Cre mice were crossed with

86 Spi1lox/lox Spib-/- to generate Mb1+/Cre Spi1lox/lox Spib-/- mice (also referred to as Mb1-CreDPB mice).

87 C57BL/6 mice were purchased from Charles River Laboratories (Saint-Constant, QC, Canada).

88 Mb1+/CreSpi1lox/lox Spib+/+ (referred to as Mb1-CreDP) and Mb1+/CreSpi1+/+ Spib-/- or Mb1+/+Spi1+/+ Spib-/-

89 (referred to as Mb1-CreDB) mice were used as experimental controls. Mice were fed with regular chow

90 and tap water ad libitum and housed with a 12-h light–dark cycle. Mice showing signs of illness were

91 euthanized and examined for spleen, thymus and lymph node enlargement and analyzed as described

92 below. All experiments were performed in compliance with the Western University Council on Animal

93 Care.

94

95 Histology and microscopic analysis

96 Mb1-CreDPB mice showing signs of disease were euthanized by CO2 and spleen and thymus were

97 removed for histological analysis. Spleen and thymus removed from Mb1-CreDB mice were used as

98 control. Organs were fixed in 10% buffered formalin. Tissues were paraffin embedded, sectioned, and

99 stained with hematoxylin and eosin. High-resolution micrographs were captured using a Q-Color3 digital

100 camera (Olympus, Markham, ON, Canada).

101

102 Flow Cytometry

103 Single cell suspensions were prepared from enlarged spleen and thymus from Mb1-CreDPB mice.

104 Red blood cells were removed from single-cell suspensions using hypotonic lysis. Flow cytometric

105 analyses were performed using a FACSCanto or LSRII instrument (BD Immunocytometry Systems, San

6 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

106 Jose, CA). Antibodies were purchased from eBioscience (San Diego, CA), BioLegend (San Diego, CA),

107 or BD Biosciences (Mississauga, ON, Canada) and included PE-anti-CD19 (1D3), FITC-anti-BP-1 (6C3),

108 APC-anti-IgM (II/ 41), PE-anti-Igk (187.1), FITC-anti-IL-7Ra (A7R34), BV421–anti-B220 (RA3-6B2),

109 PE–anti-BP-1 (6C3). Data were analyzed using FlowJo 9.1 software (Tree Star, Ashland, OR).

110

111 Whole-Exome Sequencing

112 Three matched-tumour and tail mouse samples (853, 854 and 857) had genomic DNA isolated

113 using Wizard Genomic DNA Purification Kit (Promega Corporation, Madison WI). Whole-Exome

114 Sequencing was performed by McGill University and Génome Québec Innovation Centre (Montreal,

115 Canada). The SureSelectXT Mouse All Exon kit (Agilent Technologies, Mississauga, Canada) was used

116 to perform exome target capture. Paired-end DNA library preparation was performed using SureSelectXT

117 Target Enrichment System. An Illumina HiSeq4000 instrument was used to perform the whole-exome

118 sequencing. BAM files were converted to Fastaq format on Galaxy suite25 using “SAMtoFastaq” tool.

119 Fastaq reads had Illumina Adaptors were removed using TrimGalore!. Trimmed reads were aligned to the

120 mouse reference genome (mm10) using Bowtie. BAM aligned files were used as input for single somatic

121 variant callers. Somatic variants of three matched-tumour mouse samples were called by using two

122 independent methods, Strelka and Mutect26,27. For both methods, variants were called using the standard

123 settings recommend by the package documentation. VCF files containing the somatic nucleotide variants

124 (SNVs) and small insertions and deletions (INDELs) generated by Strelka and Mutect were further filtered

125 following the criteria ‘passed’ or ‘keep’ for the Strelka and Mutect outputs, respectively. Filtered VCF

126 files outputted by Strelka and Mutect were intersected using VCFtools and annotated using the SnpEff28

127 package using the mouse genomic annotation (mm10). Annotated variants were classified according the

128 variant impact in HIGH, MODERATE, MODIFIER and LOW, and overlapping variants among the three

7 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

129 samples were determined. Independently, outputs generated by Strelka and Mutect were also annotated

130 using the SnpEff tool for further analysis.

131

132 RNA Sequencing

133 RNA was extracted from thymic tumours 853, 854 and 857 using RNeasy Kit (QIAGEN, ON,

134 Canada). RNA-sequencing was performed by McGill University and Génome Québec Innovation Centre

135 (Montreal, Canada). Paired-end (mRNA-sequencing stranded) libraries were prepared using Truseq

136 stranded total RNA library prep kit. Libraries were sequenced using an Illumina HiSeq4000 PE100

137 instrument. Data analysis was performed as described previously16. In summary, Standard Illumina

138 sequencing adaptors were removed using Trim Galore!. Trimmed Fastaq reads were aligned to the mouse

139 reference genome version mm10 using TopHat229 in mate-paired mode using Galaxy suite. Assembled

140 transcript abundance was determined as previously using Cufflinks29.

141

142 Sanger Sequencing

143 RNA extracted from 20 thymic leukemias was reversed transcribed to complementary DNA using

144 iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). Primers flanking the Ikzf3, Jak3 and Jak1 mutations

145 identified by WES were designed using Primer-BLAST. cDNA was used as a template for PCR reactions

146 using specific primers (Sequences described in Supplementary Information). PCR reactions were run

147 in a 1.5% agarose gel and the PCR products were excised and purified from the gel using QIAquick Gel

148 Extraction Kit (QIAGEN, ON, Canada). Purified DNA was submitted for Sanger sequencing at London

149 Regional Genomics Centre, Western University (London, ON, Canada). Sequencing chromatograms and

150 DNA sequences were analyzed in the software 4Peaks (Mekentosj, Amsterdam).

151

8 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

152 DNA Constructs and Site-Directed Mutagenesis

153 The mouse MSCV-IRES-GFP-JAK3 plasmid was kindly provided by Dr. Kevin D. Bunting.

154 T844M, R653H, and V670A mutations were generated by site-directed mutagenesis using Q5® Site-

155 Directed Mutagenesis Kit (New England BioLabs, Ipswich, MA). Site-directed mutagenesis primers for

156 each one of the specific mutations were designed using NEBaseChanger tool. Mutations were confirmed

157 by Sanger Sequencing before retroviral production and spinoculation.

158

159 Retroviral production and infection

160 Retroviral supernatants for wild-type Jak3, and T844M, R653H, V670A mutants were generated

161 Platinum-E retroviral packaging cells30 using PEIPro transfection reagent (PolyPlus, New York, NY). 10

162 µg of MSCV plasmid were cotransfected with 4 µg pECO plasmid (Clontech, Mountain View, CA).

163 Plasmid DNA was mixed with polyethyleneimine (PEI) at a 2:1 PEI/DNA ratio and applied to 3 × 106

164 cells in a 10-cm plate. The medium was changed after overnight incubation, and virus-containing

165 supernatant was collected 48 hours after medium change. Wild-type pro-B cells were infected by

166 spinoculation with 1 mL of viral supernatant containing 10ul of polybrene at the concentration of

167 10mg/ml. Cells were centrifuged at 3000 × g for 2 hours at 30°C and allowed to recover for 24 hours

168 before medium change and infection was confirmed by flow cytometric analysis for GFP.

169

170 Cell Culture

171 Wild-type and Jak3 mutant-infected pro-B cell lines used in this study were cultured in IMDM

172 (Wisent, QC, Canada) containing 10% FBS (Wisent), 1X penicillin/streptomycin/L-glutamine (Lonza,

173 Shawinigan, QC, Canada), and 5 x 10-5 M b-mercaptoethanol (Sigma-Aldrich St. Louis, MO). Media also

9 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

174 contained 5% or 0.5% conditioned medium from the IL-7 producing cell line J558-IL-731. Cell lines were

175 maintained in 5% CO2 atmosphere at 37˚C.

176

177 Availability of data

178 WES data is available from the Sequence Read Archive, accession SRP136503

179 (https://www.ncbi.nlm.nih.gov/sra/SRP136503). RNA-seq data is available from the

180 Omnibus, accession GSE112506 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE112506).

181

182 Statistical Analysis

183 All data reported in this study were graphed as mean ± SEM. Statistical analysis was performed

184 using Prism 5.0 (Graphpad Software, La Jolla, CA) using ANOVA or Student’s t test. Values with p ≤

185 0.05 were considered significant, where p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***), p≤ 0.0001 (****).

10 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

186 Results

187 Deletion of genes encoding PU.1 and Spi-B leads to B-cell acute lymphoblastic leukemia

188 We recently reported that deletion of the genes encoding both PU.1 and Spi-B in B cells under control

189 of the Cd79a (Mb1) promoter (Mb1+/Cre Spi1lox/lox Spib-/- mice, abbreviated to Mb1-CreDPB) resulted in a

190 severe impairment to B cell development at the large pre-B to small pre-B cell transition in the bone marrow

191 of 6-10 week-old mice16. Mb1-CreDPB mice that express Cre recombinase and are deleted for Spi-B but are

192 homozygous for the wild type Spi1 allele were fertile and healthy. In contrast, Mb1-CreDPB mice had a

193 median survival of 18 weeks, at which point they required euthanasia due to signs of illness, including

194 lethargy and laboured breathing (Fig. 1A, 1B). Dissection of euthanized mice revealed enlargement of the

195 spleen and thymus (Fig. 1A, 1C, 1D). Histological analysis revealed that normal spleen and thymus

196 organization was completely effaced in moribund Mb1-CreDPB mice compared to the controls (Fig. 1E).

197 Flow cytometry analysis was used to determine the stage in which leukemic cells begin to infiltrate

198 the spleen and thymus of pre-leukemic Mb1-CreDPB mice. We observed the presence of B220+CD19+ cells

199 in the spleen and thymus of Mb1-Cre DPB mice aged 11 weeks, but not in either C57BL/6 or Mb1DB mice

200 (Fig. 2A and 2B). At the time of euthanasia of leukemic Mb1-Cre DPB mice aged >15 weeks there were

201 high frequencies of B220+CD19+ cells in both organs (Fig. 2A and 2B). Further analysis seeking to

202 determine the phenotype of cells infiltrating the spleen and thymus showed that these organs contained high

203 frequencies of B220+ and CD19+ cells that also expressed c-Kit, CD43, BP-1, and IL-7Ra (Fig. 2D and data

204 not shown). This analysis suggested that these cells were a pre-B cell-like acute lymphoblastic leukemia

205 similar to that previously reported in CD19-CreDPB mice17 or mice deleted for genes encoding PU.1 and

206 IRF4/832. Next, leukemic cells obtained from the spleen or thymus of Mb1-CreDPB mice were characterized

207 to determine expression of cell surface IgM and Igk. Analysis of leukemias from Mb1-CreDPB mice revealed

208 that at least 74% of individual mouse leukemias did not express IgM or Igk at the cell surface (Fig. 2E,

11 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

209 bottom; 2F, left panel). 26% of leukemias from Mb1-CreDPB mouse spleen or thymus had cell surface

210 expression of both IgM and Igk (Fig. 2E, top; 2F, left panel). In contrast, analysis performed on CD19-

211 CreDPB mice showed that 78% of leukemias had IgM and Igk on the cell surface (Fig. 2F, right panel).

212 Most leukemias isolated from Mb1-CreDPB mice had IgH rearrangements, although fewer leukemias

213 expressed IgM at the cell surface (Fig. 2C). In summary, these results show that leukemia cells begin to

214 appear in thymus and spleen of Mb1-CreDPB mice after 11 weeks of age. All leukemias in Mb1-CreDPB

215 mice at 11-26 weeks of age expressed IL-7R, but most did not express Ig on their cell surface, suggesting

216 that these cells resembled pro-B or large pre-B cells.

217

218 Whole-exome sequencing (WES) identifies somatic nucleotide variants in leukemias generated in the

219 lack of Spi-B and PU.1

220 As described above, Mb1-CreDPB mice had a variable time to requirement of euthanasia as well

221 as heterogeneity in Ig expression status of leukemias (Fig. 1 and 2). This variability suggested that

222 secondary driver mutations are required to induce leukemia in addition to the initiating lesion of Spi1/Spib

223 deletion. In order to discover potential driver mutations in Mb1-CreDPB mice leukemias, we performed

224 whole-exome sequencing (WES) analysis of three mouse tumours, 853, 854, and 857, as well as on

225 matched genomic tail DNA as a control (Supplementary Fig. 1, data available in Supplementary

226 Information). Single Nucleotide Variants (SNVs) for each one of the samples were identified by

227 comparing leukemia and control tail DNA exome sequences using the Strelka somatic variant caller26.

228 The number of SNVs identified by Strelka differed for each one of the tumours, where sample 854 showed

229 the highest number of SNVs (3,887) (Fig. 3A-C). Samples 853 and 857 presented a very similar number

230 of SNVs, totaling 2,558 and 2,465, respectively. Although the tumours had a variable number of SNVs,

231 the distribution of the SNVs by followed a similar pattern in which chromosome 2 showed

12 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

232 the highest frequency of SNVs, followed by chromosome 11, chromosome 1 and chromosome 7

233 (Supplementary Fig. 2A-C). The majority of the nucleotide alterations in the three tumours analyzed

234 were C•G -> A•T transversions (Supplementary Fig. 2D). In order to gain insight into the nature of the

235 mutational processes in Mb1-CreDPB mice, we determined the mutational signature of leukemias using

236 DeconstructSigs33.This package compares the similarity of each tumour sample to a list of previous

237 published signatures generated by the analysis of 40 distinct types of human cancer34. This analysis

238 confirmed enrichment of C•G -> A•T transversions in leukemias 853, 854 and 857 (Supplementary Fig.

239 3A-C). Mutational signature analysis performed using DeconstructSigs indicated that the signatures in

240 leukemias 853, 854, and 857 was most similar to COSMIC Signatures 18, 24, 4, and 9 (Supplementary

241 Fig. 3A-C). Signature 18, which showed the highest inferred weight for all leukemia samples analyzed,

242 has been also commonly observed in human childhood cancers including B-ALL and neuroblastoma, but

243 does not have a known mechanism34. We also observed an enrichment of C->A transversions flanked

244 5’base by either adenine (A), cytosine (C), guanine (G) or thymine (T) nucleotides. These transversions

245 were consistently flanked at the 3’ base by an adenine (A) nucleotide (Supplementary Fig. 3D).

246 In order to increase the confidence in the SNVs identified by Strelka and to reduce the number of

247 SNVs without a potential biological function, somatic variants were called with a second variant discovery

248 method, Mutect27. Mutect detected fewer SNVs than Strelka (Fig. 3A-C). The overlap of SNVs identified

249 by both methods was used to determine high confidence SNVs. We found that by using this strategy the

250 number of SNVs was reduced, resulting in 183 SNVs for sample 853, 293 SNVs for sample 854, and 199

251 SNVs for sample 857. High confidence SNVs were annotated using the mouse genome (mm10) using the

252 SnpEff tool, which also enables the prediction of the biological effect of the variants28. We found that the

253 majority of the high confidence SNVs were classified by as having or a MODIFIER or MODERATE

254 deleterious impact while a few SNVs were classified as HIGH (Fig. 3D). We next looked at the effect of

13 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

255 SNVs classified as MODERATE and HIGH, and found that most of the SNVs classified as MODERATE

256 resulted in missense variants; while the few SNVs predicted as having HIGH impact caused the gain or loss

257 of stop codons, or alterations in splice acceptor or donor sites, defined by the two bases before exon start or

258 ends, respectively (Fig. 3E). We next investigated whether there were common SNVs among the three mouse

259 tumours sequenced. Samples 854 and 857 showed common SNVs in the genes Sntg1, Brd8, Lrp5 and

260 Slco1b2 (Fig. 3F). Samples 853 and 854 showed common SNVs in the genes Ryr2 and Skint6. Finally,

261 samples 853 and 857 showed common SNVs in the genes Ikzf3 and Jak3 (Fig. 3F). Interestingly, Ikzf3 and

262 Jak3 are well-characterized driver mutations in human pre-B cell acute lymphoblastic leukemia35. The

263 identification of common Ikzf3 and Jak3 SNVs suggests that this approach is effective at identifying potential

264 driver mutations.

265

266 Mutations in Janus Kinase 1 and 3 (Jak1/Jak3) and Aiolos (Ikzf3) genes are potential secondary drivers

267 of leukemogenesis in Mb1-CreDPB mice

268 As cancer driver genes would be expected to contain mutations and be expressed21, we also performed

269 RNA-sequencing analysis on leukemias 853, 854, and 857 in order to assess gene expression levels. To filter

270 data based on levels of gene expression, Fragments Per Kilobase of transcript per Million mapped reads

271 (FPKM) was determined from RNA-seq data using Cufflinks. FPKM was plotted against VAF determined

272 using Strelka26 for three mouse leukemias 853, 854, and 857. In each of the three samples there were highly

273 expressed genes that also had high VAF (Fig. 4A-C). Importantly, sample 853 had high FPKM and VAF for

274 variants in Jak3 and Ikzf3 (Fig 4A). Sample 854 had high FPKM and VAF for a variant in Jak1 (Fig 4B).

275 Finally, sample 857 had high FPKM and VAF for variants in Jak3 and Ikzf3 (Fig 4C). The elevated

276 expression levels added to the high VAF of Jak1, Jak3 and Ikzf3 genes in Mb1-CreDPB mice leukemias

277 supports the hypothesis that these variants represent secondary driver mutations.

14 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

278 Next, we sought to define the biological processes in which mutated genes could be involved by

279 performing analysis in genes with VAF greater than 20%, using PANTHER Gene List

280 Analysis36. We observed a pattern for all three samples where most of the genes were categorized in two

281 major biological processes, “Cellular Process” and “Metabolic Process”. Specifically, genes within “Cellular

282 Process” were subcategorized in sub processes as cell communication, cell proliferation and cell cycle (Fig.

283 4D-F). Genes as Cit, Cdk9 and Stag1 are related to the control of cell cycle by regulating cytokinesis,

284 transcription and cohesion of sister chromatids after DNA replication, respectively. Importantly, Smarcb1

285 encodes a core subunit of the ATP-dependent SWI/SNF chromatin remodeling complex, and was

286 previously identified as a tumor suppressor37. Nap1l4, encodes a member of the nucleosome assembly protein

287 and has a role as a histone chaperone. Genes involved in cell communication and adhesion, as Cntnap5b,

288 Fras1, Sdk1 and Magi, were also enriched in our pathway analysis. In summary, Jak3 and Ikzf3 mutations

289 were identified as potential drivers of leukemogenesis in Mb1-CreDPB mice, based on the observation that

290 these genes are mutated in two of three leukemias, have high VAF, and have high levels of expression.

291

292 Recurrent mutations in Janus Kinase 3 (Jak3) and Janus Kinase 1 (Jak1) in leukemias from Mb1-

293 CreDPB mice

294 Analysis of the impact of variants on protein expression showed that SNVs in leukemias 853 and 857

295 resulted in coding changes in Jak3 (R653H, V670A, and/or T844M) while the SNV in Jak1 in sample 854

296 resulted in a V657F substitution (Fig. 5A, 5B). These genes encode for Jak1 and Jak3 that are critically

297 required for signaling through cytokine receptors IL7R and CRLF238. Comparison of Jak3 and Jak1 protein

298 sequences among six different vertebrates showed that amino acids that underwent substitution in

299 consequence of a SNV were highly conserved between these species (Fig. 5A, 5B). Sanger sequencing of

300 samples 853, 854 and 857 confirmed the presence of SNVs identified by WES in the Jak3, Jak1 and Ikzf3

15 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

301 genes (Supplementary Fig. 3). To determine whether these variants represent recurrent leukemia driver

302 mutations, we performed Sanger sequencing on a total of 19 Mb1-CreDPB leukemias using primers flanking

303 the regions containing the SNVs identified by WES. Interestingly, this examination revealed that Jak3 R653H

304 mutation was detectable in 5/19 leukemias analyzed. Jak1 mutations V657F or V655L showed even higher

305 recurrence than Jak3 mutation, being detectable in 10/19 leukemias (Fig. 5D). Ikzf3 mutations were detected

306 only in samples 853 and 857, confirming the exome sequencing (Fig. 5C, 5D). In summary, these data show

307 that leukemia in Mb1-CreDPB mice is accompanied by recurrent secondary driver mutations in Jak1 and

308 Jak3.

309

310 R653H, V670A, and T844M amino acid substitutions in Jak3 confers survival and proliferation

311 advantages

312 We next tested whether R653H, V670A, and T844M amino acid substitutions in Jak3 were able to

313 confer growth advantage to normal cells. Site-directed mutagenesis was used to introduce these mutations to

314 the wild-type mouse Jak3 coding region in an MSCV-IRES-GFP vector (a kind gift from Dr. Kevin D.

315 Bunting). Pseudovirus was generated to spin-infect wild-type pro-B cells that grow in cultures containing

316 IL-7 and ST2 stromal cells. Pro-B cells were also infected with MSCV-Jak3 and a MSCV- empty vectors as

317 controls. Following infection, the cell lines were cultured at high (5% conditioned medium) and low IL-7

318 (0.5%). Flow cytometry was performed every two days for 14 days to determine the frequency of GFP+ cells.

319 Pro-B cells infected with Jak3 mutations R653H, V670A, or T844M outgrew uninfected or control infected

320 pro-B cells at low and high IL-7 concentration (Fig. 6A, 6B). Next, in order to verify whether Jak3 mutations

321 resulted in greater proliferation, cells infected with mutant Jak3 were plated at the initial concentration of

322 1x105 cells/ml in conditions of low and high IL-7. Cell numbers and the percentage of viable GFP+ cells were

323 determined after cells were kept in culture for the period of four days. At low IL-7 concentration, pro-B cells

16 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

324 infected with Jak3 R653H, V670A, and T844M mutations showed a significant increase in the number of

325 GFP+ cells after four days in culture (Fig. 6C, 6D). Therefore, we conclude that T844M, R653H and V670A

326 amino acids substitutions were able to increase the proliferation of pro-B cells in conditions of low IL-7.

327 Taken together, these data indicate that Jak3 mutations identified in this analysis confer survival and growth

328 advantages to pro-B cells, consistent with a role as driver mutations.

17 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

329 Discussion

330 In this study, we showed that Mb1-CreDPB mice develop precursor B-ALL that is detectable in

331 mice by 11 weeks of age and results in a requirement for euthanasia in 100% of mice by median 18 weeks.

332 The latency of 11-18 weeks, and the variable frequency of leukemias expressing Ig, show that there is

333 clonal variability, suggesting that secondary driver mutations are required for disease progression. By

334 characterizing the mutation landscape of Mb1-CreDPB leukemias using whole exome sequencing coupled

335 with RNA-sequencing, we discovered that mutations in Jak1 and Jak3 represent recurrent secondary

336 driver mutations. Introduction of mutant Jak3 into wild type cells was sufficient to promote growth in

337 response to low IL-7 concentrations. Taken together, these results show that deletion of genes encoding

338 PU.1 and Spi-B result in leukemia by predisposition to additional de novo driver mutations in genes

339 encoding Janus kinases.

340 Dysregulation of PU.1 and/or Spi-B expression are known to be involved in human leukemia and

341 have been established as tumor suppressors in mouse models17,39. Minimal reductions in PU.1 expression

342 are sufficient to induce a preleukemic condition in hematopoietic stem cells40. Whole exome sequencing

343 and whole genome sequencing studies have revealed that inactivating mutations in genes encoding PU.1

344 and Spi-B are detectable and infrequent in human leukemia41,42. In contrast, PU.1 levels are repressed by

345 frequently occurring chromosomal abnormalities such as FLT3 internal tandem duplication (FLT3-ITD)43

346 and RUNX1-ETO fusion44. Spi-B levels are repressed by ETV6-RUNX1 chromosomal fusion39. It is not

347 known how reduced PU.1 and/or Spi-B might lead to acquisition of secondary driver mutations. However,

348 PU.1 and Spi-B are involved in attenuating IL-7R signaling in developing pro- and pre-B cells. Target

349 genes of PU.1 and Spi-B involved in regulating IL-7-induced proliferation include Btk encoding the

350 tumour suppressor Bruton tyrosine kinase15, and Blnk encoding B Cell Linker Protein45. Btk and Blnk

351 work together to attenuate IL-7R signaling in developing B cells46. We speculate that mutagenesis in Mb1-

18 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

352 Cre-DPB is associated with genomic instability due to uncontrolled cell proliferation. Interestingly, a

353 similar disease is induced by combined deletion of PU.1 and IRF4, suggesting a common axis of gene

354 regulation by PU.1/Spi-B and PU.1/IRF432.

355 We identified a number of single nucleotide variants (SNVs) in our analysis, which suggest that

356 several rounds of mutational events contributed to leukemogenesis. To select the potential leukemia

357 initiator lesions, we focused our attention in genes that were highly expressed and had high variant allelic

358 frequencies (VAF), reasoning that genes within this category are likely to have originated earlier in

359 leukemia progression. This analysis identified Aiolos (Ikzf3), Jak1 (Jak1) and Jak3 (Jak3) as potential

360 secondary driver mutations. These genes are involved in signaling downstream of cytokine receptors

361 during B cell development. Cytokine receptor genes are commonly mutated in human B-cell acute

362 lymphoblastic leukemias47, and mutations in Jak3 and Ikzf3 have recently emerged as novel mutated genes

363 in high-risk B-ALL48. In particular, our analysis focused on R653H, V670A, and T844M mutations in

364 Jak3. Jak3 R653H and V670A mutations were recently observed in B-ALL generated in activation-

365 induced cytidine deaminase (AID) deficient mice49. Human equivalents of Jak3 R653H and V670A have

366 been described in human ALL (R657Q, V674F), and furthermore have been shown to function by

367 activation of proliferation in response to interleukin-750. Importantly, human JAK3 V674F is sufficient to

368 induce T-cell acute lymphoblastic leukemia when retrovirally delivered to hematopoietic precursors50.

369 Jak1 missense mutations V655L or V657F were detectable in 10 of 19 leukemias in our analysis. Jak1

370 V657F also corresponds to a human JAK1 mutation that is a frequent driver in human B-ALL (V658F)51.

371 Interestingly JAK1 V658F mutation has been previously observed in T-cell acute lymphoblastic and acute

372 myeloid leukemias52. V658F is also thought to be paralogous to the JAK2 V617F mutation that functions

373 as a driver in more than 95% of cases of polycythemia vera53. In summary, the mutations identified in this

374 study in Jak1 and Jak3 are relevant to human leukemia.

19 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

375 Taken together, this study confirms that leukemia in Mb1-CreDPB mice is accompanied by de

376 novo recurrent secondary driver mutations in the Janus Kinase signaling pathway. Mutations in the JAK

377 signaling pathway are recurrent in human leukemia including the recently discovered Ph-like

378 classification that represents a high priority for discovering new therapies9,10. In order to study the genetic

379 clonal evolution that underlies diseases such as Ph-like leukemia, there is still a need for mouse models.

380 In order to advance understanding, mouse models should have the following characteristics: 1) They

381 should develop leukemias with high penetrance and reproducibility, 2) they should replicate the genetic

382 and molecular heterogeneity of tumors and involve de novo mutations 3) they should occur in immune

383 competent mice, and 4) they should mimic that clinical behaviour of human disease54,55. The Mb1-CreDPB

384 mouse model develops B-ALL with 100% penetrance by 18 weeks of age that is driven by heterogeneous

385 de novo driver mutations. This mouse model may be useful to determine the effects of molecular targeted

386 therapies on intratumoral heterogeneity and clonal evolution in B-ALL.

387

20 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

388 Acknowledgements

389 We thank Michelle Ho for assistance with genotyping of mice. We thank Michael Reth (Freiburg,

390 Germany) for providing Mb1-Cre mice. We thank Dr. Kevin D. Bunting (Winship Cancer Institute,

391 Atlanta, USA) for providing the MSCV-IRES-GFP-JAK3 vector. We also thank scientists and staff of the

392 McGill University and Genome Quebec Innovation Centre for performing library construction and next-

393 generation sequencing. We also thank Kristin Chadwick and the London Regional Flow Cytometry Core

394 Facility for assistance with flow cytometric analysis.

395

396 Conflict of Interest

397 The authors have no financial conflicts of interest.

21 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

398 References

399

400 1 Siegel R, Ma J, Zou Z, Jemal A. Cancer statistics, 2014. CA Cancer J Clin 2014; 64: 9–29.

401 2 Hunger SP, Mullighan CG. Redefining ALL classification: toward detecting high-risk ALL and

402 implementing precision medicine. Blood 2015; 125: 3977–3987.

403 3 Inaba H, Greaves M, Mullighan CG. Acute lymphoblastic leukaemia. Lancet 2013; 381: 1943–

404 1955.

405 4 Pui CH, Robison LL, Look AT. Acute lymphoblastic leukaemia. Lancet 2008; 371: 1030–1043.

406 5 Buchner M, Swaminathan S, Chen Z, Muschen M. Mechanisms of pre-B-cell receptor checkpoint

407 control and its oncogenic subversion in acute lymphoblastic leukemia. Immunol Rev 2015; 263:

408 192–209.

409 6 Muschen M. Rationale for targeting the pre-B-cell receptor signaling pathway in acute

410 lymphoblastic leukemia. Blood 2015; 125: 3688–3693.

411 7 Mangolini M, de Boer J, Walf-Vorderwulbecke V, Pieters R, den Boer ML, Williams O. STAT3

412 mediates oncogenic addiction to TEL-AML1 in t(12;21) acute lymphoblastic leukemia. Blood

413 2013; 122: 542–549.

414 8 Shochat C, Tal N, Bandapalli OR, Palmi C, Ganmore I, te Kronnie G et al. Gain-of-function

415 mutations in interleukin-7 receptor-alpha (IL7R) in childhood acute lymphoblastic leukemias. J

416 Exp Med 2011; 208: 901–908.

417 9 Jain N, Roberts KG, Jabbour E, Patel K, Eterovic AK, Chen K et al. Ph-like acute lymphoblastic

418 leukemia: a high-risk subtype in adults. Blood 2017; 129: 572–581.

419 10 Roberts KG, Yang Y-L, Payne-Turner D, Lin W, Files JK, Dickerson K et al. Oncogenic role and

420 therapeutic targeting of ABL-class and JAK-STAT activating kinase alterations in Ph-like ALL.

22 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

421 Blood Adv 2017; 1: 1657–1671.

422 11 Ray D, Bosselut R, Ghysdael J, Mattei MG, Tavitian A, Moreau-Gachelin F. Characterization of

423 Spi-B, a related to the putative oncoprotein Spi-1/PU.1. Mol Cell Biol 1992; 12:

424 4297–4304.

425 12 Solomon LA, Li SKH, Piskorz J, Xu LS, DeKoter RP. Genome-wide comparison of PU.1 and Spi-

426 B binding sites in a mouse B lymphoma cell line. BMC Genomics 2015; 16: 76.

427 13 Garrett-Sinha LA, Su GH, Rao S, Kabak S, Hao Z, Clark MR et al. PU.1 and Spi-B are required

428 for normal B cell receptor-mediated signal transduction. Immunity 1999; 10: 399–408.

429 14 Li SK, Abbas AK, Solomon LA, Groux GM, DeKoter RP. Nfkb1 activation by the E26

430 transformation-specific transcription factors PU.1 and Spi-B promotes Toll-like receptor-mediated

431 splenic B cell proliferation. Mol Cell Biol 2015; 35: 1619–1632.

432 15 Christie DA, Xu LS, Turkistany SA, Solomon LA, Li SK, Yim E et al. PU.1 Opposes IL-7-

433 Dependent Proliferation of Developing B Cells with Involvement of the Direct Target Gene Bruton

434 Tyrosine Kinase. J Immunol 2015; 194: 595–605.

435 16 Batista CR, Li SKH, Xu LS, Solomon LA, DeKoter RP. PU.1 Regulates Ig Light Chain

436 Transcription and Rearrangement in Pre-B Cells during B Cell Development. J Immunol 2017; 198:

437 1565–1574.

438 17 Sokalski KM, Li SK, Welch I, Cadieux-Pitre HA, Gruca MR, DeKoter RP. Deletion of genes

439 encoding PU.1 and Spi-B in B cells impairs differentiation and induces pre-B cell acute

440 lymphoblastic leukemia. Blood 2011; 118: 2801–2808.

441 18 Greaves M, Maley CC. Clonal evolution in cancer. Nature 2012; 481: 306–313.

442 19 Nowell PC. The clonal evolution of tumor cell populations. Science (80- ) 1976; 194: 23–28.

443 20 Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011; 144: 646–674.

23 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

444 21 Vogelstein B, Papadopoulos N, Velculescu VE, Zhou S, Diaz LAJ, Kinzler KW. Cancer genome

445 landscapes. Science 2013; 339: 1546–1558.

446 22 Ferrando AA, Lopez-Otin C. Clonal evolution in leukemia. Nat Med 2017; 23: 1135–1145.

447 23 Hu Z, Sun R, Curtis C. A population genetics perspective on the determinants of intra-tumor

448 heterogeneity. Biochim Biophys Acta 2017; 1867: 109–126.

449 24 Hobeika E, Thiemann S, Storch B, Jumaa H, Nielsen PJ, Pelanda R et al. Testing gene function

450 early in the B cell lineage in mb1-cre mice. Proc Natl Acad Sci U S A 2006; 103: 13789–13794.

451 25 Afgan E, Baker D, van den Beek M, Blankenberg D, Bouvier D, Čech M et al. The Galaxy platform

452 for accessible, reproducible and collaborative biomedical analyses: 2016 update. Nucleic Acids Res

453 2016; 44: W3–W10.

454 26 Saunders CT, Wong WSW, Swamy S, Becq J, Murray LJ, Cheetham RK. Strelka: Accurate somatic

455 small-variant calling from sequenced tumor-normal sample pairs. Bioinformatics 2012; 28: 1811–

456 1817.

457 27 Cibulskis K, Lawrence MS, Carter SL, Sivachenko A, Jaffe D, Sougnez C et al. Sensitive detection

458 of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotechnol 2013; 31:

459 213–219.

460 28 Cingolani P, Platts A, Wang LL, Coon M, Nguyen T, Wang L et al. A program for annotating and

461 predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of

462 Drosophila melanogaster strain w1118; iso-2; iso-3. Fly (Austin) 2012; 6: 80–92.

463 29 Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR et al. Differential gene and transcript

464 expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat Protoc 2012; 7: 562–

465 578.

466 30 Morita S, Kojima T, Kitamura T. Plat-E: an efficient and stable system for transient packaging of

24 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

467 retroviruses. Gene Ther 2000; 7: 1063–1066.

468 31 Winkler TH, Rolink A, Melchers F, Karasuyama H. Precursor B cells of mouse bone marrow

469 express two different complexes with the surrogate light chain on the surface. Eur J Immunol 1995;

470 25: 446–450.

471 32 Pang SHM, Minnich M, Gangatirkar P, Zheng Z, Ebert A, Song G et al. PU.1 cooperates with IRF4

472 and IRF8 to suppress pre-B-cell leukemia. Leukemia 2016; 30: 1375–1387.

473 33 Rosenthal R, McGranahan N, Herrero J, Taylor BS, Swanton C. deconstructSigs: delineating

474 mutational processes in single tumors distinguishes DNA repair deficiencies and patterns of

475 carcinoma evolution. Genome Biol 2016; 17: 31.

476 34 Alexandrov LB, Nik-Zainal S, Wedge DC, Aparicio SA, Behjati S, Biankin A V et al. Signatures

477 of mutational processes in human cancer. Nature 2013; 500: 415–421.

478 35 Iacobucci I, Mullighan CG. Genetic Basis of Acute Lymphoblastic Leukemia. J Clin Oncol 2017;

479 35: 975–983.

480 36 Mi H, Muruganujan A, Casagrande JT, Thomas PD. Large-scale gene function analysis with the

481 PANTHER classification system. Nat Protoc 2013; 8: 1551–1566.

482 37 Tokheim CJ, Papadopoulos N, Kinzler KW, Vogelstein B, Karchin R. Evaluating the evaluation of

483 cancer driver genes. Proc Natl Acad Sci U S A 2016; 113: 14330–14335.

484 38 Babon JJ, Lucet IS, Murphy JM, Nicola NA, Varghese LN. The molecular regulation of Janus

485 kinase (JAK) activation. Biochem J 2014; 462: 1–13.

486 39 Pang SH, Minnich M, Gangatirkar P, Zheng Z, Ebert A, Song G et al. PU.1 cooperates with IRF4

487 and IRF8 to suppress pre-B-cell leukemia. Leukemia 2016; 30: 1375–1387.

488 40 Will B, Vogler TO, Narayanagari S, Bartholdy B, Todorova TI, da Silva Ferreira M et al. Minimal

489 PU.1 reduction induces a preleukemic state and promotes development of acute myeloid leukemia.

25 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

490 Nat Med 2015; 21: 1172–1181.

491 41 Mullighan CG, Goorha S, Radtke I, Miller CB, Coustan-Smith E, Dalton JD et al. Genome-wide

492 analysis of genetic alterations in acute lymphoblastic leukaemia. Nature 2007; 446: 758–764.

493 42 Mullighan CG, Zhang J, Kasper LH, Lerach S, Payne-Turner D, Phillips LA et al. CREBBP

494 mutations in relapsed acute lymphoblastic leukaemia. Nature 2011; 471: 235–239.

495 43 Mizuki M, Schwable J, Steur C, Choudhary C, Agrawal S, Sargin B et al. Suppression of myeloid

496 transcription factors and induction of STAT response genes by AML-specific Flt3 mutations. Blood

497 2003; 101: 3164–3173.

498 44 Vangala RK, Heiss-Neumann MS, Rangatia JS, Singh SM, Schoch C, Tenen DG et al. The myeloid

499 master regulator transcription factor PU.1 is inactivated by AML1-ETO in t(8;21) myeloid

500 leukemia. Blood 2003; 101: 270–277.

501 45 Xu LS, Sokalski KM, Hotke K, Christie D a, Zarnett O, Piskorz J et al. Regulation of B cell linker

502 protein transcription by PU.1 and Spi-B in murine B cell acute lymphoblastic leukemia. J Immunol

503 2012; 189: 3347–54.

504 46 Clark MR, Mandal M, Ochiai K, Singh H. Orchestrating B cell lymphopoiesis through interplay of

505 IL-7 receptor and pre-B cell receptor signalling. Nat Rev Immunol 2014; 14: 69–80.

506 47 Ma X, Liu Y, Liu Y, Alexandrov LB, Edmonson MN, Gawad C et al. Pan-cancer genome and

507 transcriptome analyses of 1,699 paediatric leukaemias and solid tumours. Nature 2018.

508 doi:10.1038/nature25795.

509 48 Zhang J, Mullighan CG, Harvey RC, Wu G, Chen X, Edmonson M et al. Key pathways are

510 frequently mutated in high-risk childhood acute lymphoblastic leukemia : a report from the

511 Children ’ s Oncology Group. Blood 2011; 118: 3080–3087.

512 49 Auer F, Ingenhag D, Pinkert S, Kracker S, Hacein-Bey-Abina S, Cavazzana M et al. Activation-

26 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

513 induced cytidine deaminase prevents pro-B cell acute lymphoblastic leukemia by functioning as a

514 negative regulator in Rag1 deficient pro-B cells. Oncotarget 2017; 8: 75797–75807.

515 50 Losdyck E, Hornakova T, Springuel L, Degryse S, Gielen O, Cools J et al. Distinct Acute

516 Lymphoblastic Leukemia (ALL)-associated Janus Kinase 3 (JAK3) Mutants Exhibit Different

517 Cytokine-Receptor Requirements and JAK Inhibitor Specificities. J Biol Chem 2015; 290: 29022–

518 29034.

519 51 Hornakova T, Staerk J, Royer Y, Flex E, Tartaglia M, Constantinescu SN et al. Acute lymphoblastic

520 leukemia-associated JAK1 mutants activate the Janus kinase/STAT pathway via interleukin-9

521 receptor alpha homodimers. J Biol Chem 2009; 284: 6773–6781.

522 52 Kralovics R, Passamonti F, Buser AS, Teo S-S, Tiedt R, Passweg JR et al. A Gain-of-Function

523 Mutation of JAK2 in Myeloproliferative Disorders. N Engl J Med 2005; 352: 1779–1790.

524 53 Bandaranayake RM, Ungureanu D, Shan Y, Shaw DE, Silvennoinen O, Hubbard SR. Crystal

525 structures of the JAK2 pseudokinase domain and the pathogenic mutant V617F. Nat Struct Mol

526 Biol 2012; 19: 754–759.

527 54 Jacoby E, Chien CD, Fry TJ. Murine models of acute leukemia: important tools in current pediatric

528 leukemia research. Front Oncol 2014; 4: 95.

529 55 Kohnken R, Porcu P, Mishra A. Overview of the Use of Murine Models in Leukemia and

530 Lymphoma Research. Front Oncol 2017; 7: 22.

531

27 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

532 Figure legends

533

534 Fig. 1. Mb1-CreDPB mice develop B cell acute lymphoblastic leukemia (B-ALL). (A) Mb1-CreDPB

535 mice (DPB) mice developed B-ALL characterized by splenic and thymic enlargement (indicated by

536 arrows). (B) Percentage survival of mice of the indicated genotypes: Mb1+/CreSpi1lox/lox Spib-/- (Mb1-

537 CreDPB, n= 43); Mb1+/CreSpi1+/+ Spib-/- mice (Mb1-CreDB, n=36) and Mb1+/+Spi1lox/loxSpib-/- (Mb1-

538 CreDB, n=14). (C). Comparisons of enlarged spleens and thymuses extracted from Mb1-CreDPB mice

539 compared to control DB mice (Mb1+/+ Spi1lox/lox Spib-/-). (D) Spleen (left) and thymus (right) weight in

540 grams (g) relative to the body weight in WT and Mb1-CreDPB mice. WT, n=10 (spleen), 5 (thymus);

541 Mb1-CreDPB, n=11 (spleen), 8 (thymus). p≤ 0.05 (*), mean ± SEM. (E). Histologic sections

542 (Hematoxylin & Eosin staining) of spleen and thymus illustrating the lymphocytic infiltration and loss of

543 organs normal structure in Mb1-CreDPB compared to controls Mb1DB (4x).

544

545 Fig. 2. Most leukemias from Mb1-CreDPB mice resemble pro-B cells and do not express either IgM

546 or Igk at the cell surface. (A) Representative flow cytometric analysis for the presence of CD19+ B220+

547 B cells in the spleen of C57BL/6 mice; 11 week old Mb1-CreDB mice and Mb1-CreDPB mice (left

548 panels); and >15 week old Mb1-CreDPB mice (right panel). (B) Representative flow cytometric analysis

549 for the presence of CD19+ B220+ B cells in the thymus of C57BL/6 mice; 11 week old Mb1-CreDB mice

550 and Mb1-CreDPB mice (left panels); and >15 week old Mb1-CreDPB mice (right panel). (C) PCR for

551 detection of heavy chain rearrangements (J558-JH4) in leukemia B cells prepared from the thymus of

552 Mb1-CreDPB mice (575-411). B cells prepared from WT mouse (C57BL/6) were used as control. The

553 Cd79 gene was used as control for DNA quality. (D) Representative flow cytometric analysis of leukemic

554 cells from Mb1-CreDPB mice gated on CD19+ B220+ cells showed that leukemias expressed IL-7R on the

28 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

555 cell surface (top and bottom). (E) Representative IL-7R+ leukemias expressing IgM and Igk, (top, Ig+) and

556 those not expressing IgM and Igk (bottom, Ig-). (F) Percentage of leukemias expressing IgM and Igk (Ig+)

557 or not expressing IgM and Igk (Ig-). Left panel, Mb1-CreDPB mice, n=14 (spleen) and n=8 (thymus).

558 Right panel, CD19-CreDPB mice, n=9 (thymus).

559

560 Fig. 3 Identification of High Confidence Somatic Nucleotide Variants (SNVs) in Mb1-CreDPB

561 mouse leukemias. (A, B and C) SNVs identified by two different variant caller methods, Strelka and

562 Mutect, were combined and overlapping SNVs were classified as High Confidence SNVs. (D)

563 Classification of High Confidence SNVs for impact. Graph shows the number of SNVs classified

564 according to impact. (E) Predicted biological effect of the High Confidence SNVs classified as having

565 MODERATE and HIGH impact. (F). Venn diagram showing the overlap among High Confidence SNVs

566 identified in the three individual mouse leukemias.

567

568 Fig. 4 Integration of whole-exome sequencing (WES) and RNA-seq. (A, B and C). Scatter plot

569 correlating the levels of gene expression in Fragments Per Kilobase of Transcript per Million of reads

570 mapped (FPKM log10) and the Variant Allele Frequency (VAF) for genes in which FPKM was greater

571 than zero. Leukemias 853, 854 and 857 are shown, respectively. (D, E and F). Biological Pathway

572 Analysis in genes with VAF equal or greater than 20% was performed using Panther – Gene List Analysis.

573 Diagram shows the number of genes enriched according the biological process. Enrichment for genes

574 related to “Cellular Process” and “Metabolic Process” is shown for the three samples analyzed.

575

576 Fig. 5 Identified mutations in conserved regions of Jak3, Jak1, and Ikzf3 genes. (A, B and C; left)

577 Schematic shows the protein domains of Jak3, Jak1 and Aiolos (Ikzf3). Amino acids substitutions caused

29 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

578 by single nucleotide variants identified in the whole-exome sequencing in samples 853, 854 and 857 are

579 also indicated. (A, B and C; right) BLAST analysis comparing Human (Homo sapiens), Mouse (Mus

580 musculus), Rat (Rattus norvergicus), African Elephant (Loxodonta africana), Giant Panda (Ailuropoda

581 melanoleuca), Atlantic Salmon (Salmo salar), Zebrafish (Danio rerio), Tropical Frog (Xenopus tropicalis)

582 protein sequences for Jak1, Jak3 and Aiolos shows that amino acids which undergo substitution in

583 consequence of a single nucleotide variation are highly conserved between the two species. (D) Summary

584 of Sanger sequencing screening for the presence of amino acids substitutions in Jak1 (V657F and V655L),

585 Jak3 (T844M and R653H), and Aiolos (R137* and H195T) in a panel of nineteen leukemias prepared

586 from Mb1-CreDPB mice. Filled boxes indicate samples in which mutations were identified by Sanger

587 Sequencing.

588

589 Fig 6. Jak3 mutations confer growth advantage and increased proliferative potential in normal pro-

590 B cells. (A and B) Percentage of GFP+ cells over the course of 14 days after infection with MSCV-empty,

591 MSCV-JAK3, MSCV-T844M, MSCV-R653H, MSCV-V670A or MSCV-V670A/T844M in conditions

592 of low (0.5%) and high (5%) IL-7 conditioned medium (CM), respectively. (C and D) Absolute number

593 of viable GFP+ cells after four days in culture under conditions of low (0.5%) and high (5%) IL-7. Cells

594 were infected with either MSCV-empty, MSCV-JAK3, MSCV-T844M, MSCV-R653H, MSCV-V670A

595 or MSCV-V670A/T844M. Data is presented as number of cells/ml; n=3.

596

597 Supplementary Fig. 1. Experimental design for RNA-seq and WES-seq experiments. (A) DNA and

598 RNA were isolated from leukemic cells extracted from thymus of leukemic mice. (B) Weeks of survival

599 and weight of the organs extracted from representative leukemic Mb1-CreDPB mice. Samples 853, 854,

600 and 857 were used for sequencing experiments. (C) Flow cytometric characterization (IgM, x-axis; Igk,

30 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

601 y-axis; gated on CD19+ B220+ cells) of leukemic mice in which DNA and RNA were isolated for

602 sequencing. C57BL/6, control; 853, 854, 857, leukemic Mb1-CreDPB; FMO = fluorescence minus one.

603

604 Supplementary Fig. 2. Characterization of the nucleotide somatic nucleotide variants (SNVs)

605 identified by Strelka somatic variant caller in three Mb1-CreDPB mice leukemias. (A, B and C)

606 SNV distribution according to chromosome number in leukemias 853, 854 and 857, respectively. (D)

607 Allelic change frequency for 853, 854 and 857 leukemias showing an enrichment for G>T and C>A

608 substitutions.

609

610 Supplementary Fig. 3. Mutational signature of Mb1-CreDPB mouse leukemias shows strong bias to

611 C•G -> A•T transversions. (A-C) DeconstructSigs reconstruction analysis of mouse leukemias 853, 854

612 and 857. COSMIC mutation signatures were used to reconstruct the mutational profile of mouse leukemias

613 and to predict similarity. (D) Heat map illustrating the mutational context of 5’ and 3’ nucleotides

614 surrounding the transitions identified in samples 853, 854 and 857.

615

616 Supplementary Fig. 4. Sanger sequencing confirms the presence of SNVs identified by WES. PCR

617 for amplification of the region containing the SNVs were performed on cDNA synthesized from RNA

618 prepared from mouse tumours. PCR products were submitted for Sanger sequencing using primers

619 targeting the region of interest. (A); Sanger sequencing results of region containing Ikzf3 SNV identified

620 in the sample 857, transition C-T; (B) Ikzf3 SNV identified in the sample 853, transition C-T; (C) Jak1

621 SNV identified in the sample 854; transition G-T; (D) Jak3 SNV identified in the sample 853; transition

622 G-A. (E) Jak3 SNV identified in the sample 857; transition C-T. Chromatograms are shown for each SNV

623 investigated. Note that most SNVs are detectable at 50% or lower allelic frequency.

31 A Figure 1 D C bioRxiv preprint

Relative weight (g) Mb1-CreΔB Mb1-CreΔPB Spleen doi: certified bypeerreview)istheauthor/funder.Allrightsreserved.Noreuseallowedwithoutpermission. WT https://doi.org/10.1101/291054 Mb1-Cre Thymus ΔPB Spleen * B ; Relative weight (g) this versionpostedApril2,2018. Mb1-CreΔPB Thymus WT Mb1-Cre ΔPB * The copyrightholderforthispreprint(whichwasnot weeks 18.1 E

Thymus Spleen Mb1-CreΔPB Mb1-CreΔB Mb1-CreΔPB Mb1ΔB bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

Figure 2

A 11 weeks old Leukemic >15wo C C57BL/6 Mb1ΔB Mb1-CreΔPB Mb1-CreΔPB Mb1-Cre ΔPB

WT 575 577 578 351 345 347 411 Water VHJ558 -JH4 53.5% 51.1% 43.1% 11.7% B220 B220 B220 B220

44% 43.4% 54.5% 87.9% JH1 JH2 CD19 CD19 CD19 CD19 JH3 C57BL/6 Mb1ΔB Mb1ΔPB Mb1-CreΔPB B JH4

99.4% 99.5% 7.6% 3.11% B220 B220 B220

B220 CD79a

0.56% 0.2% 84.3% 96.6%

CD19 CD19 CD19 CD19

Gated on Gated on CD19+ B220+ IL-7R+ 4 D 1000 E 10 F Mb1-CreΔPB CD19-CreΔPB 89% 800 81% 103 mice mice

600 102 IgM+ Igk+ 400 SSC-H: Side Scatter 101 200

81.2 0 100 0 1 2 3 4 10 10 10 10 10 IgM 100 101 102 103 104 SSC FL1H IL7R Igκ

4 1000 10 9.92 0%0.163 800 103

600 102 IgM- Igk- Frequency of Leukemias Frequency of Leukemias 400 SSC-H: Side Scatter 101 200 89.8 91%90.8 0.158 0 100 Spleen 0 1 2 3 4 10 10 10 10 10 100 101 102 103 104 Thymus Thymus SSC FL1H IgM IL7R Igκ bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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. Figure 3

D A SNVs calling methods

Leuk_853 164

Strelka Mutect (2558) (1967) 106 104 79 2375 183 1784 59 62 37 25 14 4 13 8

B Leuk_854 E Strelka Mutect (3887) (2735)

3594 293 2442 79 59 62

11 3 7 1 1 11 C Leuk_857

Strelka Mutect (2465) (1706) 2266 199 1507

F Overlapping annotated SNVs 4 common elements in "854" and "857": Sntg1 Leuk_854 Brd8 Leuk_857 Lrp5 283 4 Slco1b2 172 0 2 common elements in "853" and "857": 2 2 Ikzf3 Jak3 174 2 common elements in "853" and "854": Ryr2 Leuk_853 Skint6 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

Figure 4

A Leuk_853 BCLeuk_854 Leuk_857 Rplp0 Rpl4 Hnrnpf Actg1 Rps20 Sf3b3 Uqcr11 Smarcb1 2 2 Eif2a 2 Ikzf3 Ikzf3 Jak3 C1qc Gorasp Jak1 Wnk1 Srsf6 Scaf11 Bfar Jak3 Ifi44 Ssr1 Cul4b Ranbp2 Pacs1 Clip1 Znf236 Itpr1 Apbb2 Myo19 0 0 Thada 0

Map6 FPKM FPKM FPKM Rhbdf1 H2-Bl Gucy2g Notch4 Neurl1b Rarb Apol10b Dnaic2 Sv2a Disc1 Sox12 Tbc1d30 Magi1 Itih4 Ros1 Agbl1 Rbm20 -2 -2 Luzp4 -2 Corin Fras1 Sdk1 Cntnap5b Lrp1b Kcnma1 Mpdz Rab27b Shank1 Obscn Trpm3 Prdm8 Ptprt -4 -4 -4 0 10 20 30 40 50 0 20 40 60 0 10 20 30 40 50 VAF VAF VAF

D 6 Leuk_853 E Leuk_854 F 10 5 6 5 Leuk_857 5 6 4 8 Cellular Process: 7 4 Cellular Process: Cellular Process: 14 Cell Communication Cell Proliferation Cell Communication Cell Proliferation Cell Communication Cell Proliferation Casp9 Jak3 Sv2a Jak1 20 Ssr1 Ssr1 Cntnap5b Tgfb2 16 Itpr1 24 Tll1 Jak3 11 22 28 Fras1 Cell Cycle Gng7 Cell Cycle Agbl1 Cell Cycle 9 Rhbdf1 Ranbp2 Jak1 Stag1 Lama2 Cdk9 5 Jak3 Dnaic2 4 Sdk1 Als2 6 Jak3 Cit Myo19 Smarcb1 Tmtc3 Nap1l4 Dgkk Biological Adhesion Zfyve28 Myo19 Fndc7 Cit Tgfb2 Biological Adhesion Lrp5 Biological Adhesion Gucy2g Biological Regulation Magi1 Biological Regulation Wnk1 Biological Regulation Biogenesis Biogenesis Stab2 Biogenesis Cellular Process Cellular Process Cellular Process Metabolic Process: Metabolic Process: Metabolic Process: Developmental Process Developmental Process Developmental Process Apbb2 Hk2 Jak1 Pign Adck2 Pus7 Agpat2 Immune System Process Immune System Process Localization Eif2a Ranbp2 Rps20 Thoc2 Leng8 Agbl1 Ssr1 Localization Zfp236 Dnaic2 Localization Bfar Nap1l4 Metabolic Process Cdk9 Apol10b Metabolic Process Ikzf3 Smarcb1 Metabolic Process Sdk1 Wnk1 Multicellular Organismal Process Tbc1d8b Jak3 Multicellular Organismal Process Nop2 Rhbdf1 Multicellular Organismal Process Zmym1 Slc3a2 Response to Stimulus Entpd1 Dgkk Reproduction Tbc1d2 Jak3 Reproduction Phf6 Pbx4 Cul4b Response to Stimulus Tgfb2 Response to Stimulus Csl Ikzf3 Papolb Clip1 Als2 Csad Gucy2g Sox12 Tc1d30 Itih4 Malsu1 Sf3b3 Slco1b2 Tep1 Dennd4a Cit bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

Figure 5

A JAK3 FERM SH2 Pseudokinase Kinase (Jak3) Homo KVLLAREG--ADGSPPFIKLSDPG V SPAVL VELCRYDPLGDNTGALVAVKQLQ Mus KVLLAREG--GDGNPPFIKLSDPG V SPTVL VELCRYDPLGDNTGPLVAVKQLQ 1 1,100aa Arg653HisVal670Ala (853) (857)Thr844Met (857) Rattus KVLLAREG--VDGNPPFIKLSDPG V SPTVL VELCRYDPLGDNTGPLVAVKQLQ Elephant KVLLAREG--AEGSPPFIKLSDPG V SPTVL VELCRYDPLGDNTGALVAVKQLQ Panda KVLLAREG--ADGNLPFIKLSDPG V SPTVL VELCRYDPLGDNTGALVAVKQLQ Salmo NLLLAREGDASQGSSPFIKLSDPG I NVAML VELCRYDPLGDNTGELIAVKKLQ Danio NLLLVREGD----S-PFIKLSDPG V SMSLL VELCRYDPWGDNTGELVAVKELQ Xenopus KILLSREG--DKGNPPFIKLSDRG V SIKVL VELCRYDPLGDNTGELVAVKKLQ ::** *** . ******* * : . :* ******** ***** *:***:**

B JAK1 FERM SH2 Pseudokinase Kinase (Jak1) Homo AFFEAASMMRQVSHKHIVYLYGVCVRDVENIMVEEFVEGGPLDLFMHRKSDVLTT Mus AFFEAASMMRQVSHKHIVYLYGVCVRDVENIMVEEFVEGGPLDLFMHRKSDALTT 1 1,153aa Val657Phe (854) Rattus AFFEAASMMRQVSHKHIVYLYGVCVRDVENIMVEEFVEGGPLDLFMHRKSDALTT Elephant AFFEAASMMRQVSHKHIVYLYGVCVRDVENIMVEEFVEGGPLDLFMHRRSDALTT Panda AFFEAASMMRQVSHKHIVYLYGVCVRDVENIMVEEFVEGGPLDLFMHRKSDVLTT Salmo AFFETASMMRQVSHKHIALLYGVCVRHLENIMVEEFVQLGPLDVFMRRQRSPLST Danio AFFETASMMRQISHKHIALLYGVCVRHQENIMVEEFVQYGPLDLFMRRQTTPLST Xenopus AFFETASMMRQVSHKHIVLLHGVCVRDVENIMVEEFVDFGPLDLFMHRKSEVLTT ****:******:*****. *:*****. *********: ****:**:*: *:*

N-terminal C-terminal ZF ZF ZF ZF C AIOLOS 1 Homo GLSCISFNVLMVHKRSHTGERPFQCNQ QRRDALTGHLRT (Ikzf3) HSVEKPYKCEFCGR Mus GLSCISFNVLMVHKRSHTGERPFQCNQ QRRDALTGHLRTHSVEKPYKCEFCGR Arg137*His195Tyr (857) (853) 507aa Rattus GLSCISFNVLMVHKRSHTGERPFQCNQ QRRDALTGHLRTHSVEKPYKCEFCGR Elephant GLSCISFNVLMVHKRSHTGERPFQCNQ QRRDALTGHLRTHSVEKPYKCEFCGR Panda GLSCISFNVLMVHKRSHTGERPFQCNQ QRRDALTGHLRTHSVEKPYKCEFCGR Xenopus GLACNSLNVLLVHKRSHTGERPFQCNQ QRRDALTGHLRTHSVEKPYKCEFCGR **:* *:***:**************** **************************

D Mb1-CreΔPB mouse tumours 853 854 856 857 996 405 427 429 851 861 908 932 933 952 956 966 968 973 406 JAK3 T844M JAK3 R653H JAK1 V657F JAK1 V655L AIOLOS H195T AIOLOS R137* bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

Figure 6

A B

5% CM 0.5% CM

%

%

+ + +

GFP GFP

CD 5% CM 0.5% CM

Day 0 (1x105 cells/ml) Day 0 (1x105 cells/ml)

Day 4 Day 4

cells (cells/ml) cells

cells (cells/ml) cells

+ +

+ +

Number of GFP of Number Number of GFP of Number bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

Supplementary Fig.1

A B

Thymic Tumour Thymic (Spi1lox/loxSpib-/-) B-ALL cells

RNA/DNA *RNA-Seq isolation *Whole Exome-Seq

C B cells C57BL/6 853 IgM-Igκ- 854 IgM+Igκ+ 857 IgM+Igκ+ FMO IgM 5 10 46.7% 2.61% 1.07% 13.2% 86.6% 31.7% 66.0% 17.5% 0.2%

104

103

0

B220 0 3 104 105 10 90.6% 5.74% 0.18% 0.018% 1.98% 0.30% CD19 Igκ 82.3% 0.030% IgM C57BL/6 406 IgM+Igκ- 968 IgM+Igκ+ 973 IgM+Igκ+ FMO Igκ 9.61% 86.8% 0.12% 5.45% 7.79% 90.7% 0.38% 58.8% 0% 0.081%

Igκ 0.30% 3.28% 0.59% 93.8% 0.17% 1.36% 0.025% 40.8% 3.34% 96.6%

IgM bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

Supplementary Fig.2

A C Mutation Load (Leuk_853) 2,558 SNVs Mutation Load (Leuk_857) 3,887 SNVs 250

200 200

150 150

100 100

50 50 Number of Mutations Number of Mutations

0 0

chr1 chr10 chr11 chr12 chr13 chr14chr15 chr16chr17 chr18chr19 chr2 chr3 chr4 chr5 chr6 chr7 chr8 chr9 chrX chrY chr1 chr10chr11 chr12chr13 chr14chr15 chr16chr17 chr18chr19 chr2 chr3 chr4 chr5 chr6 chr7 chr8 chr9 chrX chrY Chromosome Chromosome B D Mutation Load (Leuk_854) 2,465 SNVs

300

200

100 Number of Mutations

0

chr1 chr10 chr11 chr12 chr13 chr14chr15 chr16chr17 chr18chr19 chr2 chr3 chr4 chr5 chr6 chr7 chr8 chr9 chrM chrX chrY Chromosome bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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.

Supplementary Fig.3 A

Leukemia_853 Signature 4: 0.137 & Signature 18: 0.445 & Signature 24: 0.23 & Signature 29: 0.187 0.10 C>A 0.08 C>G C>T 0.06 T>A T>C Fraction 0.04 T>G 0.02 0.00 D heatmap scale (log10) B C>AC>G C>T T>AT>C T>G Leukemia_854 Signature 4: 0.156 & Signature 18: 0.451 & Signature 24: 0.295 & Signature 29: 0.091 -1.5 -1 -0.5 0 0.5 1 1.5 0.10 A C 0.08 C>A 853 C>G G T 0.06 C>T T>A A 3’base 0.04 Fraction T>C C 854 G T>G 0.02 T

0.00 A C C 857 G T

Leukemia_857 Signature 4: 0.192 & Signature 18: 0.436 & Signature 24: 0.297 & Signature 29: 0.075 A C G TA C G T A C G TA C G TA C G TA C G T 0.10 5’base 0.08 C>A C>G 0.06 C>T T>A

Fraction 0.04 T>C 0.02 T>G

0.00 bioRxiv preprint doi: https://doi.org/10.1101/291054; this version posted April 2, 2018. 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. Supplementary Fig.4

A. Ikzf3 SNV identified by WES in the sample 857

Mouse Gene Chrom Position Reference Mutation Sanger Sequencing

857 Ikzf3 chr8 98,490,344 CTTTCATAAGCGAAGCCATAC 854 Ikzf3 chr8 98,490,344 CTTTCATAAGCGAAGCCATAC 853 Ikzf3 chr8 98,490,344 CTTTCATAAGCGAAGCCATAC

857 854 853

B. Ikzf3 SNV identified by WES in the sample 853

Mouse Gene Chrom Position Reference Mutation Sanger Sequencing

857 Ikzf3 chr8 98,488,858 C T TTAGGACACATTCT 854 Ikzf3 chr8 98,488,858 CTTTAGGACACATTCT 853 Ikzf3 chr8 98,488,858 CTTTAGGACATATTCT

857 854 853

C.Jak1 SNV identified by WES in the sample 854

Mouse Gene Chrom Position Reference Mutation Sanger Sequencing

857 Jak1 chr4 101,163,673 GTGCGTGTGTGTCCGAGAT 854 Jak1 chr4 101,163,673 GTGCGTGTGTGTCCGAGAT 853 Jak1 chr4 101,163,673 GTGCGTGTGTGTCCGAGAT

857 854 853

D.Jak3 SNV identified by WES in the sample 853

Mouse Gene Chrom Position Reference Mutation Sanger Sequencing

857 Jak3 chr8 71,684,008 GACTGGCTCGTGAGG 854 Jak3 chr8 71,684,008 GACTGGCTCGTGAGG 853 Jak3 chr8 71,684,008 GACTGGCTCNTGAGG

857 854 853

E. Jak3 SNV identified by WES in the sample 857

Mouse Gene Chrom Position Reference Mutation Sanger Sequencing

857 Jak3 chr8 71,685,437 CTACAATACGGGAC 854 Jak3 chr8 71,685,437 CTACAATACGGGAC 853 Jak3 chr8 71,685,437C T ACAATACGGGAC

857 854 853