Efficient and multiplexed somatic genome editing with Cas12a mice

34 35 Somatic genome editing in mouse models has increased our understanding of the in vivo effects of 36 genetic alterations in areas ranging from neuroscience to cancer biology and beyond . However, 37 existing models are limited in their ability to create multiple targeted edits . Thus, our 38 understanding of the complex genetic interactions that underlie development, homeostasis, and 39 disease remains incomplete. Cas12a is an RNA-guided endonuclease with unique attributes that 40 enable simple targeting of multiple genes with crRNA arrays containing tandem guides . To 41 accelerate and expand the generation of complex genotypes in somatic cells, we generated 42 transgenic mice with Cre-regulated and constitutive expression of enhanced Acidaminococcus sp. 43 Cas12a ( enAsCas12a). In these mice , enAsCas12a -mediated somatic genome editing robustly 44 generated compound genotypes, as exemplified by the initiation of diverse cancer types driven by 45 homozygous inactivation of trios of tumor suppressor genes. We further integrated these modular 46 crRNA arrays with clonal barcoding to quantify the size and number of tumors with each array, as 47 well as the efficiency of each crRNA . These Cas12a alleles will enable the rapid generation of 48 disease models and broadly facilitate the high -throughput investigation of coincident genomic 49 alterations in somatic cells in vivo. 50 51

52 Genetically engineered mouse models have been widely used to uncover phenotypes 53 resulting from defined genetic alterations 1. However, the creation of new mouse alleles has 54 remained a barrier to the study of more complex genotypes, due to the expense and time required 55 to generate a new allele and cross it with other alleles of interest 2. Somatic genome editing with 56 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Cas9 has greatly increased the rate at which single genes can be studied both ex vivo and in vivo, 57 but the ability of Cas9 to model complex genotypes has been limited by the need for each guide to 58 have its own promoter and tracrRNA, thereby requiring the laborious cloning of guides in 59 sequence3. 60 Cas12a (previously known as Cpf1) is a Class 2 Type V CRISPR -Cas system that has 61 unique value in genome editing due to its ability to easily target multiple genomic loci4–6. Unlike 62 Cas9, Cas12a has both RNase and DNase activity, and can process a single pre-crRNA transcript 63 (crRNA array) containing numerous spacers (guides) into its constituent crRNAs to direct Cas12a-64 mediated cleavage of their target regions4. Cas12a is also distinct from Cas9 in recognizing a T-65 rich 4 bp protospacer adjacent motif (PAM) and generating staggered cuts that are distal to the 66 PAM5. Cas12a from multiple species has been employed for genome editing in vitro, including 67 Cas12a from Acidaminococcus sp. (As) and Lachnospiraceae bacterium (Lb)4,6–11. Large-scale 68 screens have identified Cas12a variants with increased efficiency, broadened PAM recognition 69 ranges, and reduce d off-target cleavage 12–15. In particular , an enhanced version of AsCas12a 70 (enAsCas12a) with several substitutions (E174R/N282A/S542R/K548R) has superior genome 71 editing relative to wild type AsCas12a, in addition to a substantially expanded targeting range13. 72 The impact of complex genotypes is highly relevant in human cancer, which is defined by 73 its genetic complexity. Individual tumors can have tens to hundreds of non-synonymous mutations, 74 in addition to aberrant epigenetic modifications, gene duplications, chromosomal rearrangements, 75 and gains or losses of entire chromosomes16–18. Any or all of these alterations could contribute to 76 tumorigenesis, tumor progression, and resistance to therapy. Deconvoluting functional drivers 77 from human cancer sequencing data and understanding how these genes interact to generate cancer 78 phenotypes is complicated by the limited availability of samples, differences among patients and 79 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint which therapies they have received, and technical variability of the approaches used to analyze 80 them. Functional interrogation of the phenotypes resulting from defined combinations of 81 alterations is thus critical for isolating the key drivers of these complex disease states. 82 To accelerate and expand our ability to generate complex genotypes in somatic cells, we 83 generated Cre-regulated and constitutive en AsCas12a transgenic mice. Efficiently multiplexed 84 somatic genome editing using Cas12a transgenic mice enables the rapid generation of complex 85 models and should facilitate the high-throughput investigation of coincident genomic alterations 86 in vivo. 87 88

89 90 Generation and validation of Cre-regulated and constitutive enAsCas12a alleles 91 To enable Cas12a -mediated somatic genomic editing and thus take advantage of the 92 features that distinguish Cas12a from Cas9 (Supplementary Figure 1a), we generated transgenic 93 mice expressing enAsCas12a with an optimized nuclear localization sequence (NLS) 94 configuration that increases cutting efficiency and an HA tag to facilitate identification 95 (enAsCas12a; Figure 1a)13,19. We generated transgenic mice through integration of a CAGGS 96 promoter-driven Lox-Stop-Lox (LSL)-enAsCas12a-PolyA cassette into the H11 locus (H11LSL-97 Cas12a) ( Figure 1a ,b). Expression of Cre in fibroblasts from H11LSL-Cas12a mice induced 98 recombination of the LSL cassette and expression of Cas12a protein (Figure 1c). We also crossed 99 H11LSL-Cas12a mice to CMV-Cre “deleter” mice to generate a constitutive H11Cas12a allele (Figure 100 1d and Supplementary Figure 1b ). H11Cas12a mice were viable and fertile and had widespread 101 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint expression of Cas12a (Figure 1e), with prominently nuclear protein localization ( Figure 1f and 102 Supplementary Figure 1c). 103 104 Cas12a-induced inactivation of Nf1, Rasa1 and Pten generate lung adenocarcinoma 105 Oncogene-negative lung adenocarcinoma represents ~30% of all lung adenocarcinomas 106 and has been modeled in mice through combinatorial inactivation of the tumor suppressor genes 107 Nf1, Rasa1 and Pten in lung epithelial cells 20. This was initially accomplished using Cas9 -108 mediated somatic genome editing and multi-step sgRNA cloning20. As an initial test of the ability 109 of the H11LSL-Cas12a allele to generat e tumors with complex genotypes in vivo , we generated 110 lentiviral vectors expressing Cre and a pre-crRNA array targeting Nf1, Rasa1, and Pten. Given the 111 ability to synthesize these three-crRNA arrays and their ease of cloning, we generated a pool of 27 112 lentiviral Cre vectors with every combination of three guides targeting each tumor suppressor gene 113 (Figure 2a). We cloned these pre-crRNA arrays into a vector that contained a diverse 16-114 nucleotide barcode (BC) at the 3’ end of the U6 promoter and Cre recombinase (Lenti -U6BC-115 crNf1/Rasa1/Pten-Cre; Figure 2b). Cre in these vectors induce s Cas12a expression in somatic 116 cells in H11LSL-Cas12a mice, and amplification of the BC-crRNA region from bulk tumor-bearing 117 lungs followed by tumor barcoding and high -throughput barcode sequencing (Tuba -seqUltra; U6 118 barcode Labeling with per -Tumor Resolution Analysis) can quantify the size of each clonal 119 tumor21–23. 120 We transduced the lungs of H11LSL-Cas12a mice (some of which also contain ed a R26LSL-121 Tomato Cre-reporter allele), wild -type negative control mice, and KrasLSL-G12D;R26LSL-Tomato (KT) 122 positive control mice with Lenti-U6BC-crNf1/Rasa1/Pten-Cre (Figure 2c). Tumors in KT mice 123 form due to Cre-mediated expression of oncogenic KRAS in the absence of any Cas12a-mediated 124 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint gene targeting. We anticipated that Cas12a-mediated homozygous inactivation of Nf1, Rasa1, and 125 Pten might be limited; therefore, we used a 5-fold higher titer in H11LSL-Cas12a mice relative to KT 126 mice in which every transduced cell will express oncogenic KRAS. Unexpectedly, only 9 weeks 127 after tumor initiation, the H11LSL-Cas12a mice developed extremely high tumor burden as assess ed 128 by lung weight (>15-folder higher than KT mice after subtracting normal lung weight) , direct 129 imaging of the lung lobes, and histology (Figure 2 d-f). Histology, positive immunohistochemical 130 staining for NKX2-1/TTF-1, and negative staining for UCHL1 confirmed that these tumors were 131 lung adenomas and adenocarcinomas (Figure 2f)20. Neoplastic cells in these tumors expressed 132 nuclear Cas12a (Supplementary Figure 2g). PCR amplification and Sanger sequencing of each 133 of the 9 target ed regions from genomic DNA from bulk tumor -bearing lungs from H11LSL-Cas12a 134 mice confirmed indels at each target site (Figure 2h). 135 136 Quantification of Cas12a-mediated tumorigenesis 137 We PCR-amplified and high-throughput sequenced the BC-crRNA region of the integrated 138 Lenti-U6BC-crNf1/Rasa1/Pten-Cre vectors from bulk tumor-bearing lungs (Tuba-seqUltra)21,23. This 139 approach enabled the precise quantification of the number of cancer cells in each clonal tumor, the 140 number of tumors in each mouse , and the relative efficiency of each guide (Figure 3a). Overall 141 tumor burden normalized to viral titer was more than 10-fold greater in H11LSL-Cas12a than KT mice 142 (Figure 3b). We also assessed the number of clonal barcoded tumors in each mouse and found 143 that H11LSL-Cas12a mice had 2-fold more tumors on average than KrasLSL-G12D mice when corrected 144 for viral titer (Figure 3c). These results are consistent with the high tumorigenic potential of lung 145 epithelial cells with combined inactivation of Nf1, Rasa1, and Pten20,24. 146 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Tumor sizes in H11LSL-Cas12a mice were dramatically larger than in KT mice, with the log-147 normal mean tumor size around 4 -fold greater in H11LSL-Cas12a mice ( Figure 3 d), suggesting 148 increased tumor growth of crNf1/Rasa1/Pten tumors compared to those driven by oncogenic 149 KRAS. Finally, each of the 9 crRNAs (Figure 3e-f) and 27 crRNA arrays (Supplementary Figure 150 2a-b) generated relatively similar tumor numbers and sizes. These results indicate efficient gene 151 inactivation and multiplexed somatic genome editing in H11LSL-Cas12a mice. 152 153 Induction of small-cell lung cancer through the Cas12a-mediated inactivation of Rb1, Trp53, 154 and Rbl2 155 Small-cell lung cancer (SCLC) is a neuroendocrine cancer that has been modeled in mice 156 through Cre/lox -mediated, and more recently Cas9 -mediated, inactivation of Rb1, Trp53 and 157 Rbl225,26. We generated a lentiviral vector expressing Cre and a pre-crRNA array with all 158 combinations of three guides targeting Rb1, Trp53, and Rbl2 to create 27 vectors for combinatorial 159 gene inactivation (Lenti -U6BC-crRb1/Trp53/Rbl2-Cre; Figure 4a). We transduced H11LSL-Cas12a 160 mice (some of which also contained an R26LSL-Tomato allele) and wild-type negative control mice 161 with Lenti-U6BC-crRb1/Trp53/Rbl2-Cre (Figure 4b). After 18 weeks, transduced H11LSL-Cas12a 162 mice had high lung tumor burden ( Figure 4c-d). Tumors in these mice included UCHL1positive 163 bronchiolar early-stage SCLC, along with more poorly differentiated regions (Figure 4e). Cancer 164 cells stained positive for the Cas12a HA tag (Figure 4e). Amplification of the BC-crRNA region 165 from bulk tumor-bearing lungs, high-throughput sequencing, and Tuba -seqUltra analysis revealed 166 that H11LSL-Cas12a mice transduced with Lenti -U6BC-crRb1/Trp53/Rbl2-Cre had large numbers of 167 clonal tumors and high tumor b urden, consistent with robust tumor growth (Figure 4f -g). 168 Microdissected tumors had indels at the targeted genomic sites (Figure 4h). Quantitative analysis 169 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint of the crRNA arrays in tumors showed that different crRNAs generated similar tumor numbers 170 and sizes ( Figure 4i-j). Thus, H11LSL-Cas12a mice enabled rapid and efficient generation of this 171 recalcitrant cancer type. 172 173 Generation of autochthonous PDAC with inactivation of Trp53, Cdkn2a, and Smad4 174 TP53, CDKN2A, and SMAD4 are the three most frequently mutated tumor suppressor 175 genes in human pancreatic ductal adenocarcinom a (PDAC), and many diagrams of the genomic 176 progression of this cancer type show the acquisition of mutations /alterations in all three of these 177 genes27,28. However, despite extensive work to investigate these tumor suppressor genes in 178 pancreatic carcinogenesis in vivo 29–35, a model in which all three tumor suppressor genes are 179 inactivated has yet to be published. To determine whether Cas12a -mediated somatic genome 180 editing would also be efficient in adult pancreatic epithelial cells, we generated a lentiviral vector 181 expressing Cre and pre-crRNA arrays with three gRNAs targeting Trp53, Cdkn2a, and Smad4 182 (Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre; Figure 5a). 183 We delivered this pool to the pancrea ta of KT;H11LSL-Cas12a, Cas12a-negative KT, and 184 H11LSL-Cas12a mice through retrograde pancreatic ductal injection ( Figure 5b)36. Several 185 KT;H11LSL-Cas12a mice developed large multifocal pancreatic tumors as early as 6 weeks after 186 transduction, with some also showing metastatic dissemination to the liver ( Figure 5c-e). While 187 KT mice only developed PanIN lesions, KT;H11LSL-Cas12a mice developed large, Cas12a-expressing 188 primary tumors , including poorly-differentiated PDAC with cytological atypia and smaller 189 Cytokeratin-19positive mucinous regions with desmoplastic stroma and lobular structures (Figure 5f 190 and Supplementary Figure 3a -b). The pancreata of transduced H11LSL-Cas12a mice appeared 191 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint entirely normal, suggesting that even the combined inactivation of these three tumor suppressors 192 has minimal effect in the absence of oncogenic KRAS (Figure 4f). 193 Tuba-seqUltra analysis indicated that KT;H11LSL-Cas12a and KT mice developed similar 194 numbers of clonal expansions/ tumors, consistent with the importance of oncogenic KRAS in 195 tumorigenesis (Figure 5g). However, KT;H11LSL-Cas12a mice had much larger tumors and thus 196 higher tumor burden ( Figure 5h and Supplementary Figure 3c-d). Analysis of bulk tissue 197 uncovered indels at all target sites (Supplementary Figure 3e). Tumors with each of the 9 crRNAs 198 had similar tumor number and size, further underscoring the broad efficiency of Cas12a-mediated 199 gene inactivation (Figure 5i and Supplementary Figure 3f). 200 Lentiviral vectors generated as a pool suffer from lentiviral template switching during 201 reverse transcription , which can shuffle different elements 37. We observed minimal 202 recombination-mediated shuffling of guides between different crRNA arrays (Supplementary 203 Figure 4a-f), likely due to the short distance between guides in Cas12a -based crRNA arrays, in 204 contrast to Cas9 -based approaches37,38. These data highlight the versatility of somatic Cas12a-205 mediated gene inactivation to generate novel and clinically relevant complex cancer genotypes. 206 207

208 In this study, we used Cas12a mice and 3-guide crRNA arrays to rapidly generate complex 209 genotypes in somatic cells. By synthesizing and cloning arrays in a simple pooled manner, this 210 approach should enable the rapid generation of numerous complex genotypes in parallel, 211 dramatically increasing both the scale and the rate at which new disease models with complex 212 genotypes can be studied. Previous in vitro studies have employed as many as 25 guides together7, 213 suggesting that many loci could be targeted together in vivo with this transgenic model. 214 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint We applied somatic CRISPR/Cas12a genome editing initially to study cancer, as cancer 215 has tremendous genomic complexity, yet the genetically engineered mouse models that have been 216 used to study these diseases generally have only a few designed genetic changes. While this has 217 enabled a reductionist approach to uncover key phenotypes controlled by different genes in cancer, 218 it has also limited our ability to model and understand key aspects of cancer that are driven by 219 combinations of genomic alterations. For instance, engineered mouse models often fail to develop 220 spontaneous metastases and those that do rarely metastasize to all organs commonly observed in 221 humans2. Indeed, metastases tend to have a greater degree of genomic complexity than primary 222 tumors39, so modeling metasta tic cancer in mice may be facilitate d by the generation of tumors 223 with additional engineered alterations. 224 While Cre/lox- and CRISPR/Cas9-mediated somatic genome editing have been used to 225 generate cancer models driven by complex combinations of loss of function alleles, the generation 226 of new models remains time-consuming2. Furthermore, given the complexity of genotypes within 227 human tumors, the generation of a single model of a given cancer type (e.g. our model of oncogene-228 negative lung adenocarcinoma) 20 should not be seen as a model for all patients with th at broad 229 class of tumors. Here, we have developed enAsCas12a mice to close the gap between the 230 complexity of human tumors and the relative simplicity of genetically engineered mouse models. 231 While our study focused on the use of Cas12a for multi ple gene inactivation in cancer 232 models, this system is versatile and should have numerous applications within and beyond cancer 233 modeling. Cas12a-generated staggered DNA breaks could make it more suitable for the 234

of exogenous DNA sequences through homology -directed repair 9. Furthermore, 235 while we used delivery of lentiviral Cre driven by a ubiquitous promoter to initiate Cas12 a 236 expression and editing, use of Cre driven by more specific promoters would enable control of 237 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Cas12a expression in particular cell types of interest. Somatic Cas12a-mediated genome editing 238 may also be well -suited for generating chromosomal rearrangements, such as oncogenic 239 translocations that require the simultaneous targeting of pairs of loci, as well as the generation of 240 panels of larger deletions. The multiplicity of Cas12a targeting could make it suitable for modeling 241 even more complex mutational states , such as chromothripsis or aneuploidy (through 242 chromosomal destruction). Finally, transgenic Cas12a in our model s obviates the need for 243 exogenous delivery of the large (4 kb) Cas12a gene and negates concerns of pre-existing immunity, 244 which has been shown to dramatically affect cells with exogenous Cas9 expression40. 245 Compared to Cas9, Cas12a creates more diverse indels and continues cutting its target loci 246 for longer due to the greater distance between its PAM and cleavage site s41. As a result, we 247 previously used Cas12a to generate an evolving barcode system to map cellular phylogen ies41. 248 Incorporation of this system into in vivo models of development and disease could complement 249 Cas9-based approaches, and the multiplexed ability of Cas12a crRNA expression could allow 250 lineage tracing in combination with gene inactivation. This could dramatically increase our ability 251 to both generate and study complex genotypes42–45. 252 The generation of somatic cells with complex genotype s could be of broad value, both in 253 cancer modeling and beyond. Somatic genome editing with Cas9 transgenic mice has enabled 254 rapid analysis of the in vivo effects of genetic alterations in neuroscience 46–49, cancer 22,50–53, 255 immunology54, and other fields55,56. These Cas12a transgenic mice will enable the generation of 256 pairwise and higher -order combinations of genetic alterations to be generated in cell types of 257 interest to map complex phenotypes to complex genotypes and improve our understanding of 258 human disease. 259 260 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint FIGURE LEGENDS 261 262 Figure 1. Generation of Cre-regulated and constitutive Cas12a mice 263 a. Schematic of the H11LSL-Cas12a transgene. Enhanced AsCas12a containing several substitutions 264 to increase on-target efficiency and PAM binding sequence range (Kleinstiver et al. (2019). Nature 265 Biotechnology), has two nuclear localization sequences (NLS), and three HA tags at the C -266 terminus (Liu et al. (2019), Nucleic Acids Research), controlled by Cre-mediated removal of the 267 LoxP-Stop-LoxP (LSL) cassette. 268 b. PCR genotyping of mice of the indicated H11LSL-Cas12a and wild-type (WT) genotypes. 269 c. Western blots on tail tip fibroblasts from H11LSL-Cas12a and wild-type control mice 3 days after 270 Adeno-Cre (Cre) treatment. Cas12a was detected by both anti -HA and anti -Cas12a antibodies. 271 αTubulin shows loading. 272 d. Schematic of the H11Cas12a transgene, following Cre-mediated removal of the LSL cassette. 273 e. Western blots on tissue lysates from H11Cas12a/WT or H11LSL-Cas12a/WT mice. αTubulin shows 274 loading. 275 f. Immunohistochemical staining for the HA tag on Cas12a on liver sections from the indicated 276 genotypes of mice. Scale bars, 25 μm. 277 278 Figure 2. Rapid and efficient generation of oncogene-negative lung tumors through Cas12a-279 mediated coincident inactivation of three tumor suppressor genes 280 a. Lenti-U6BC-crNf1/Rasa1/Pten-Cre pool has guides targeting Nf1, Rasa1, and Pten. Each gene 281 is targeted by three crRNAs in all combinations. 282 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint b. Design of a lentiviral vector that expresses Cre recombinase and has a pre -crRNA array with 283 three spacers (guides) downstream of a bovine U6 promoter with an integrated barcode region. 284 c. Intratracheal delivery of Lenti -U6BC-crNf1/Rasa1/Pten-Cre to H11LSL-Cas12a, KrasLSL-285 G12D;R26LSL-Tom (KT), and wild-type ( WT) mice. Viral titer (infectious units, ifu) and mouse 286 numbers are indicated. 287 d. Lung weights from mice of the indicated genotypes 9 weeks after transduction with Lenti-U6BC-288 crNf1/Rasa1/Pten-Cre. Each dot represents a mouse. Means +/- standard deviation are indicated. 289 e. Light (upper) and Tomato fluorescent (lower) images of lung lobes from the indicated genotypes 290 of mice 9 weeks after transduction with Lenti -U6BC-crNf1/Rasa1/Pten-Cre. Dashed lines outline 291 tissues. Tumors are Tomato-positive due to recombination of the R26LSL-Tomato allele. Scale bars, 5 292 mm. 293 f. Hematoxylin & eosin (upper), NKX2-1 (middle), and UCHL1 (lower) staining of lung sections 294 from the indicated genotypes. Scale bars, 100 μm. 295 g. Immunohistochemical staining for the Cas12a HA tag on lung tissue from representative H11LSL-296 Cas12a and WT mice transduced with Lenti-U6BC-crNf1/Rasa1/Pten-Cre. Scale bars, 100 μm. 297 h. Indel frequencies for genomic regions targeted by each crRNA within bulk tumor-bearing lung 298 tissue from H11LSL-Cas12a mice. Means +/ - standard deviation of three tumor -bearing lungs are 299 shown. 300 301 Figure 3. Integration of crRNA arrays with tumor barcoding enables quantification of tumor 302 initiation and tumor size 303 a. Schematic of tumor barcoding with high -throughput BC-crRNA sequencing to determine the 304 size of each Cas12a-induced clonal tumor. 305 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint b. Total number of neoplastic cells (Total tumor burden) in each mouse normalized to viral titer. 306 Each dot represents a mouse, and the bars are means. P-value calculated with Wilcoxon rank-sum 307 test. 308 c. Number of tumors in each mouse normalized to viral titer. Each dot represents a mouse, and the 309 bars are means. P-value calculated with Wilcoxon rank-sum test. 310 d. Mean tumor size given a log-normal tumor size distribution. Each dot represents a mouse, and 311 bars are means. P-value calculated with Wilcoxon rank-sum test. 312 e. Number of tumors for each crRNA (guide) relative to the median value for each gene. 95% 313 confidence intervals are shown. 314 f. Mean tumor size assuming a log -normal distribution for each crRNA (guide) relative to the 315 median value for each gene. 95% confidence intervals are shown. 316 317 Figure 4. Induction of small -cell lung cancer through the simultaneous Cas12a -mediated 318 inactivation of Trp53, Rb1 and Rbl2 319 a. Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre pool has guides targeting Trp53, Rb1, and Rbl2. Each gene 320 is targeted by three crRNAs in all 27 combinations. 321 b. Intratracheal delivery of Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre to H11LSL-Cas12a and wild-type (WT) 322 mice. Viral titer (infectious units, ifu) and mouse numbers are indicated. 323 c. Lung weights from mice of the indicated genotypes 18 weeks after transduction with Lenti -324 U6BC-crTrp53/Rb1/Rbl2-Cre. Each dot represents a mouse. Mean +/ - standard deviation is 325 indicated. 326 d. Light (upper) and Tomato fluorescent (lower) images of lung lobes from the indicated genotypes 327 of mice 18 weeks after transduction with Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre. Dashed line outlines 328 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint tissue. Tumors are Tomato-positive due to recombination of the R26LSL-Tomato allele. Scale bars, 5 329 mm. 330 e. Hematoxylin & eosin (upper), UCHL1 (middle) and HA tag (lower) staining of lung sections 331 from an H11LSL-Cas12a mouse. Scale bars, 100 μm. 332 f. Total number of neoplastic cells (Total tumor burden) in each mouse normalized to viral titer. 333 Each dot represents a mouse, and the bar is the mean. 334 g. The number of tumors in each mouse normalized to viral titer. Each dot represents a mouse, and 335 the bar is the mean. 336 h. Indel frequencies for genomic regions targeted by each crRNA within micro -dissected tumor 337 tissue from H11LSL-Cas12a mice. Symbol shapes and colors identify values from the same sample. 338 Bars represent means. 339 i,j. Number of tumors ( i) and mean tumor size assuming a log -normal distribution ( j) for each 340 crRNA (guide) relative to the median value for each gene. 95% confidence intervals are shown. 341 342 Figure 5. Rapid generation of PDAC through somatic Cas12a-mediated inactivation of 343 commonly mutated tumor suppressor genes 344 a. Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre pool has guides targeting Trp53, Cdkn2a, and Smad4. 345 Each gene is targeted by three crRNAs in all 27 combinations. 346 b. Intrapancreatic delivery of Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre to KrasLSL-G12D;R26LSL-Tom 347 (KT), KT;H11LSL-Cas12a, and H11LSL-Cas12a mice. Viral titer (infectious units, ifu) and mouse numbers 348 are indicated. 349 c. Survival curve of mice of the indicated genotypes transduced with Lenti-U6BC-350 crTrp53/Cdkn2a/Smad4-Cre. 351 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint d. Light (upper) and fluorescent Tomato (lower) pancreas images from the indicated genotypes of 352 mice 6 weeks ( KT;H11LSL-Cas12a) or 11 weeks ( KT) after transduction with Lenti-U6BC-353 crTrp53/Cdkn2a/Smad4-Cre. Dashed lines outline tissues. Scale bars, 5 mm. 354 e. Light and fluorescent Tomato images of a liver lobe from a KT;H11LSL-Cas12a mouse. Dashed 355 line outlines tissue. Scale bars, 5 mm. 356 f. Immunohistochemistry images of pancreas sections from the indicated genotypes, showing 357 hematoxylin & eosin (H&E; upper) and Alcian Blue (AB; lower) staining. KT;H11LSL-Cas12a panels 358 show representative regions with poorly differentiated sarcomatoid carcinoma (left; approximately 359 80-90% of tumor area) and more differentiated areas with mucinous cells and some lobular 360 structure (right). Scale bars, 100 μm. 361 g. The number of tumors in each mouse normalized to viral titer. Each dot represents a mouse, and 362 bars are means. Note that two H11LSL-Cas12a mice did not have detectable tumor burden and are not 363 plotted. P-value calculated with Wilcoxon rank-sum test. 364 h. Mean tumor size given a log-normal tumor size distribution. Each dot represents a mouse, and 365 bars are means. Note that two H11LSL-Cas12a mice did not have detectable tumor burden and are not 366 plotted. P-value calculated with Wilcoxon rank-sum test. 367 i. Number of tumors for each crRNA (guide) relative to the median value for each gene. 95% 368 confidence intervals are shown. 369 370 371 372 373 374 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint

375 376 Generation of the H11LSL-Cas12a transgenic allele 377 A targeting vector containing 5’ and 3’ homology arms flanking the chicken beta -actin 378 /CMV enhancer/gamma globulin splice acceptor (CAGGS) promoter, loxP-STOP(6xSV40 379 polyA)-loxP cassette (LSL), enhanced Acidaminococcus sp. Cas12a 380 (E174R/N282A/S542R/K548R; enAsCas12a) 13, a nucleoplasmin nuclear localization signal 381 (NLS), a 3xHA epitope sequence, an SV40 NLS 19, and a rabbit β -globin polyadenylation signal 382 (RBG pA) was used to generate the H11LSL-Cas12a knock-in mice. An sgRNA targeting the mouse 383 Hipp11 (H11) locus (GAACACTAGTGCACTTATCCTGG), the targeting vector, and Cas9 384 mRNA were co -injected into fertilized C57BL/6J mouse oocytes (Cyagen Biosciences). F0 385 founder animals were identified by PCR followed by sequence analysis. A founder mouse was 386 bred with C57BL/6J mice to establish the H11LSL-Cas12a line. H11LSL-Cas12a mice were crossed to 387 CMV-Cre “deleter” mice to generate the constitutive H11Cas12a mice. PCR genotyping for the 388 H11LSL-Cas12a transgene was performed with the following primers: forward, 5’ 389 ATGCCATCATGCTCTCACTGC 3’; reverse, 5’ GGCTATGAACTAATGACCCCGTAATTG 390 3’; alternative reverse, 5’ CTTGTGGGTCTTCCACCTTTCTT 3’. PCR genotyping to distinguish 391 H11LSL-Cas12a from H11Cas12a was performed with the following primers: forward, 3’ 392 AGGTCGAGGGACCTAATAACTTCG 5’; alternative forward, 5’ 393 ATCTGTGCGGAGCCGAAATC 3’; reverse, 5’ TGCGTGCTTTGTCTTCCTCG 3’. 394 395 396 397 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Design, generation, barcoding, and production of lentiviral vectors 398 Pre-crRNA arrays with three guides were designed with diverse direct repeat sequences 399 flanking each guide 11, with BsmBI cut sites on both ends of the array (all guide and array 400 sequences in Supplementary Table 1). Guides for enAsCas12a were designed using CRISPick 401 for the Mouse GRCm38 reference genome (NCBI RefSeq v.108.20200622) 11,57. Arrays were 402 ordered as single -stranded DNA pools (Twist Biosciences) containing all 27 combinations of 3 403 guides for each of the 3 genes targeted in an array. Barcoded lentiviral vector backbones were 404 created by cloning a 98 bp oligo with a 16 -nucleotide diverse barcode (BC) and two BsmBI 405 restriction sites into the 3’ end of the bovine U6 promoter with Gibson Assembly (NEBuilder HiFi, 406 New England Biosciences) in a vector also containing PGK -Cre recombinase23. After low-cycle 407 PCR amplification, pre-crRNA arrays were ligated via Golden Gate Assembly with BsmBI into 408 the barcoded vector backbone. The cloning product was then electroporated into competent E. coli 409 (C3020K, New England Biosciences) and plated onto LB -ampicillin plates. To ensure sufficient 410 barcode diversity for Tuba -seqUltra sequencing and delineation of tumors, ~10 6 colonies were 411 collected for each Lenti-U6BC-crRNA-Cre pool. 412 Lentiviral vectors were produced using polyethylenimine (PEI)-based transfection of 293T 413 cells with delta8.2 and VSV-G packaging plasmids in 15 cm cell culture plates. Sodium butyrate 414 (B5887, Sigma Aldrich) was added 8 hours after transfection to achieve a final concentration of 415 20 mM. Media was refreshed 24 hours after transfection. 20 mL of virus -containing supernatant 416 was collected 36, 48, and 60 hours after transfection. The three collections were then pooled and 417 concentrated by ultracentrifugation (112,000 g for 1.5 hours), resuspended overnight in 120 µL 418 PBS, and then frozen at -80°C. Viruses were titered against a lab standard of known titer. 419 420 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Mice and tumor initiation 421 The use of mice for this study was approved by the Institutional Animal Care and Use 422 Committee at Stanford University, protocol number 26696. CMV-Cre 423 (RRID:IMSR_JAX:006054)58, KrasLSL-G12D/+ (RRID:IMSR_JAX:008179)59 and R26LSL-tdTomato 424 (RRID:IMSR_JAX:007914)60 mice were on a C57BL/6 J background. H11LSL-Cas12a mice 425 (JAX:038388) and H11Cas12a mice (JAX:038389) are available through the Jackson Laboratory. 426 Lung tumors were initiated by intratracheal delivery of 60 μ L of lentiviral vectors in PBS 427 to isoflurane-anesthetized mice. Pancreatic tumors were initiated by retrograde ductal delivery of 428 150 μL of lentiviral vectors in PBS to isoflurane -anesthetized mice as previous described , and 429 mice were injected intraperitoneally with 100 µg/kg cerulein (C9026, Sigma-Aldrich) every hour 430 for 8 hours over two consecutive days, two weeks after lentiviral delivery 36. For small-cell lung 431 cancer tumors, mice were pre-treated with naphthalene53. Briefly, corn oil (C8267, Sigma-Aldrich) 432 was filter-sterilized with a 0.22 µm filter and aliquoted into 5 mL portions. Naphthalene was then 433 dissolved into the corn oil vehicle at a concentration of 50 mg/mL before it was administered into 434 mice via intraperitoneal injections at a dosage of 200 mg/kg. Roughly 46-48 hours later, mice were 435 administered virus intratracheally and supplemented with a bowl of DietGel76A (72 -07-5022, 436 ClearH2O) food to promote recovery. Infectious units (ifu) of lentivirus used for each experiment 437 are indicated in each respective figure. 438 439 Histology and immunohistochemistry 440 Tissues were fixed in 4% formalin for 24 hours, stored in 70% ethanol, and paraffin -441 embedded. 4 µm thick sections were used for Hematoxylin and Eosin staining and 442 immunohistochemistry. Immunohistochemistry was performed using an Avidin/Biotin Blocking 443 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Kit (SP-2001, Vector Laboratories), Avidin-Biotin Complex kit (PK-4001, Vector Laboratories), 444 and DAB Peroxidase Substrate Kit (SK-4100, Vector Laboratories) following standard protocols. 445 Alcian Blue and Trichrome stains were performed following standard protocols (Histo -Tec 446 Laboratory, Inc.). The following primary antibodies were used: anti -HA (3724, Cell Signaling 447 Technology), anti -TTF1 (NKX2 -1; ab76013, Abcam), anti -CK19 (Ab2133570, TROMA -III, 448 Developmental Study Hybridoma Bank), and anti-UCHL1 (HPA005993, Sigma-Aldrich). 449 450 Western blot analyses 451 For immunoblotting, cells and homogenized tissues were lysed in Cell Lysis Buffer (Cell 452 Signaling Technology) containing a complete mini-protease inhibitor cocktail (Roche), then equal 453 quantities of protein lysate (3 -10 µg) were separated by SDS -PAGE using 4 –12% gradient gels 454 (Invitrogen) and transferred to PVDF membranes. Membranes were blocked with 5% milk in PBS 455 with 0.1% Tween 20 (PBST) for 1 hour at room temperature, followed by incubation with primary 456 antibodies diluted in PBST with 5% milk overnight at 4oC. After 4x15 minute washes with PBST, 457 membranes were incubated with an HRP-conjugated goat anti-rabbit secondary antibody (12-348, 458 Sigma-Aldrich) diluted 1:5000 in PBST with 5% milk. After 4x15 minute washes with PBST, 459 protein expression was then visualized with enhanced chemiluminescence reagents (PI80196, 460 Fisher Scientific). Primary antibodies were used at the following dilutions: rabbit anti-HA, 1:1000 461 (3724, Cell Signaling); rabbit anti-Cas12a, 1:1000 (19984, Cell Signaling); and rabbit anti-alpha-462 Tubulin, 1:2000 (2144, Cell Signaling). 463 464 465 466 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Analysis of indels at target sites 467 The target sites for each crRNA were PCR -amplified from genomic DNA extracted from 468 bulk tumor-bearing tissue (Figure 2h and Supplementary Figure 3e) or microdissected tumors 469 (Figure 4h ) using GoTaq Green® Mastermix (Promega) and the primer pairs listed in 470 Supplementary Table 2 . Microdissected tumors were isolated under a dissecting microscope; 471 however, due to the high tumor burden, samples may have contained more than one tumor. 472 Amplicons were run on 1% agarose gels, gel-extracted (QIAquick PCR & Gel Cleanup Kit, 473 Qiagen), and Sanger sequenced (Elim Biopharm, Inc). The Sanger sequencing traces were 474 analyzed with TIDE61 to estimate the percent of the DNA with indels at each site. As these lungs 475 each contain many clonal tumors, the DNA should be a mix of that from non -neoplastic cells 476 (normal lung and stromal cells), tumors with indels at that site, and tumor with indels at the target 477 sites of the other crRNAs targeting that gene. 478 479 Tuba-seqUltra library preparation and analysis 480 Benchmark control cell lines containing unique barcodes ( 50,000 cells each) were added 481 to each lung sample prior to lysis to enable the calculation of the absolute number of neoplastic 482 cells in each tumor23. Bulk tumor-bearing tissues were homogenized in 6 m L of cell lysis buffer 483 (Puregene, 158063, Qiagen) using a FastPrep -24 5G tissue homogenizer (116005500, MP 484 Biomedicals) and digested overnight with proteinase K (AM2544, Life Technologies). To remove 485 RNA, samples were incubated with RNase A prior to genomic DNA extraction using a Puregene 486 kit according to the manufacturer’s recommended protocol. Q5 High -Fidelity 2x Master Mix 487 (M0494X, New England Biolabs) was used to amplify the U6-BC-crRNA region from 32 μg of 488 genomic DNA in a total reaction volume of 800 μL per sample using unique dual-indexed primers 489 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint as described23. The conc entration of the amplified barcode product in each PCR was measured 490 using D1000 screentape reagents ( 5067-5582, Agilent Technologies) and the Tapestation 491 instrument (Agilent Technologies). Amplicons were pooled at equal molar ratios of barcode 492 product, normalized to the estimated burden of tumors in each mouse lung sample (measured lung 493 mass minus an estimated normal lung weight of 0.15 g). Pooled PCR products were purified using 494 Agencourt AMPure XP beads (A63881, Beckman Coulter) using a double size selection protocol 495 and sequenced on the Illumina NextSeq 6000 platform (read length 2x150bp) for barcode analysis 496 and MiSeq (2x300) for crRNA array analysis. 497 Paired-end reads were processed with regular expressions to identify the three gRNA 498 sequences (henceforth referred to as the tumor -genotype) and clonal barcodes. When identifying 499 gRNA sequences, we strictly required a perfect match with the designed sequences. The 12 -500 nucleotide random clonal barcode sequence possesses a high theoretical diversity of approximately 501 412 (> 107). Given the virus titer we used (2x10 5 or 1x106 infectious units per mouse), there are 502 typically fewer than 10,000 clonal tumors for each tumor -genotype per mouse. As such, the 503 probability of two clonal tumors with the same tumor-genotype possessing two barcodes within a 504 1-hamming distance of each other is extremely low. Consequently, when we encountered low -505 frequency clonal barcodes within a 1 -hamming distance of high -frequency clonal barcodes, we 506 attributed them to sequencing or PCR errors. These low -frequency barcodes were merged with 507 barcodes of higher frequencies. After extracting barcode and crRNA information, we converted 508 the read counts associated with clonal tumors into absolute neoplastic cell numbers. This 509 conversion was accomplished by normalizing the reads of the clonal tumor to the number of reads 510 of the "spike-in" benchmark cell lines added to each sample prior to lung lysis and DNA extraction. 511 We imposed a minimum tumor size cutoff of 300 cells for downstream analysis. 512 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint 513 Characterization of overall tumor growth and initiation 514 To quantify tumor growth, we used the log-normal mean (LN mean) tumor size, which is 515 the maximum-likelihood estimate of mean tumor size , assuming a log -normal tumor size 516 distribution. In addition to the tumor size metric, we characterized the effects of gene inactivation 517 on tumorigenesis by accounting for both the number of tumors (“tumor number”) and the 518 cumulative number of neoplastic cell s (“tumor burden”). When comparing tumor growth and 519 initiation across different genotypes of mice tumor number and tumor burden are normalized by 520 the amount of virus delivered. 521 522 Estimation of tumor growth and initiation effects of individual crRNA arrays 523 To estimate the relative effect of individual crRNAs on tumor growth, we calculated the 524 relative LN mean tumor size by normalizing the LN mean of tumors with a specific crRNA to the 525 average LN mean size of the three crRNAs targeting the same gene. This normalization process 526 extended to the analysis of tumors with specific crRNA arrays, where the relative LN mean tumor 527 size for a tumor with an array was normalized to the mean of all 27 distinct crRNA arrays. Unlike 528 the LN mean tumor size, tumor number is linearly affected by lentiviral titer and is thus sensitive 529 to underlying differences in the representation of each Lenti-U6BC-crRNA-Cre vector in the viral 530 pool. Relative tumor number was computed by initially standardizing the tumor number in Cas12a-531 expressive mice against that in control mice lacking Cas12a. Subsequently, this standardized tumor 532 number was then normalized using the same methodology applied to the relative LN mean. 533 534 535 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Tumor exclusion based on PCR template shuffling and unexpected crRNA arrays 536 Ideally, each clonal barcode would be uniquely paired with one crRNA array. However, 537 instances where the same clonal barcode is associated with multiple crRNA arrays have been 538 observed, which may be attributed to several factors: (1) multiple copies of the same barcode 539 molecule ligated to distinct arrays during vector cloning; (2) template switching during lentiviral 540 reverse transcription; and (3) PCR uncoupling throughout library preparation37,38. The first two 541 scenarios would yield tumors that accurately reflect the genotype information conveyed by the 542 crRNA array, while the latter scenario would generate spurious tumors (i.e. BC-crRNA reads 543 that do not originate from a genuine tumor in the sample). Such spurious tumors, resulting from 544 PCR uncoupling (a relatively rare event) will typically be markedly smaller than their genuine 545 counterparts. To discern spurious tumors from genuine tumors with the same clonal barcode, we 546 created a reference null distribution of size ratios between large and small tumors by randomly 547 sampling pairs of tumors within the same mouse. When a clonal barcode was found associated 548 with multiple crRNA arrays, we designated the crRNA array with the largest tumor as genuine 549 and others as potentially aberrant. By calculating the size ratio of the genuine to the potential 550 spurious tumor and comparing it against the upper 95th percentile of the reference distribution, 551 we can identify and eliminate most, if not all, spurious tumors. 552 In cases when tumors were found to be associated with crRNA arrays that were not from 553 the “correct” lentiviral libraries, these tumors could have arisen from the coincident transduction 554 of a cell with the “correct” virus and a “contaminating” virus. To eliminate these tumors from the 555 analysis in the most conservative manner, we identified each clonal barcode linked to the 556 “contaminating” vector, a corresponding clonal barcode associated with the “correct” vector that 557 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint exhibited the most similar tumor size within the same mouse, and excluded both clonal barcodes 558 from further analysis. 559 560 Cell line generation 561 Tail tip fibroblasts were generated by mechanically dissociating depilated tail tips, 562 digesting them in 0.25% trypsin (Life Technologies) for 30 minutes, collecting tissue fragments 563 with centrifugation at 500 rcf for 5 minutes, then resuspending and plating in high glucose 564 Dulbecco’s Modified Eagle Medium (Life Technologies) containing 20% fetal bovine serum and 565 2% penicillin-streptomycin (Thermo Fisher Scientific). 566 567

568 569 We thank the Stanford Veterinary Animal Care Staff for expert animal care, Greg Charville 570 for advice on histologic characterization of tumors, and members of the Winslow laboratory for 571 helpful comments. J.D.H was supported by an American Cancer Society Fellowship (PF-21-112-572 01-MM) and a TRDRP Postdoctoral fellowship (T31FT1619). Y.J.T was partly supported by the 573 Canadian Institute of Health Research (CIHR) postdoctoral fellowship (MFE-176568). P.A.R. was 574 supported by the National Science Foundation Graduate Research Fellowship (DGE-2146755) and 575 the Lucille P. Markey Stanford Graduate Fellowship. This work was supported by NIH R01 -576 CA230025 (to M.M.W), NIH P01-CA244114 (to M.M.W.), NIH R01-CA231253 (to M.M.W and 577 D.A.P), NIH R01-CA234349 (to M.M.W and D.A.P.), NIH R35 -CA231997 (to J.S.), NIH R35 -578 HG011316 and R01 -GM141627 (to L.C.), and in part by the Stanford Cancer Institute support 579 grant (NIH P30-CA124435). 580 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint 581

582 1. Kersten, K., de Visser, K. E., van Miltenburg, M. H. & Jonkers, J. Genetically engineered 583 mouse models in oncology research and cancer medicine. EMBO Mol. Med. 9, 137–153 584 (2017). 585 2. Hebert, J. D., Neal, J. W. & Winslow, M. M. Dissecting metastasis using preclinical models 586 and methods. Nat. Rev. Cancer 23, 391–407 (2023). 587 3. Murray, C. W. et al. An LKB1-SIK Axis Suppresses Lung Tumor Growth and Controls 588 Differentiation. Cancer Discov. 9, 1590–1605 (2019). 589 4. Paul, B. & Montoya, G. CRISPR-Cas12a: Functional overview and applications. Biomed. J. 590 43, 8–17 (2020). 591 5. Khan, S. & Sallard, E. Current and Prospective Applications of CRISPR-Cas12a in 592 Pluricellular Organisms. Mol. Biotechnol. 65, 196–205 (2023). 593 6. Zetsche, B. et al. Multiplex gene editing by CRISPR–Cpf1 using a single crRNA array. Nat. 594 Biotechnol. 35, 31–34 (2017). 595 7. Campa, C. C., Weisbach, N. R., Santinha, A. J., Incarnato, D. & Platt, R. J. Multiplexed 596 genome engineering by Cas12a and CRISPR arrays encoded on single transcripts. Nat. 597

16, 887–893 (2019). 598 8. Dede, M., McLaughlin, M., Kim, E. & Hart, T. Multiplex enCas12a screens detect functional 599 buffering among paralogs otherwise masked in monogenic Cas9 knockout screens. Genome 600 Biol. 21, 1–19 (2020). 601 9. Dai, X. et al. Massively parallel knock-in engineering of human T cells. Nat. Biotechnol. 41, 602 1239–1255 (2023). 603 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint 10. Magnusson, J. P., Rios, A. R., Wu, L. & Qi, L. S. Enhanced Cas12a multi-gene 604 regulation using a CRISPR array separator. eLife 10, e66406 (2021). 605 11. DeWeirdt, P. C. et al. Optimization of AsCas12a for combinatorial genetic screens in 606 human cells. Nat. Biotechnol. 39, 94–104 (2021). 607 12. Ma, E. et al. Improved genome editing by an engineered CRISPR-Cas12a. Nucleic Acids 608 Res. 50, 12689–12701 (2022). 609 13. Kleinstiver, B. P. et al. Engineered CRISPR–Cas12a variants with increased activities 610 and improved targeting ranges for gene, epigenetic and base editing. Nat. Biotechnol. 37, 276–611 282 (2019). 612 14. Gier, R. A. et al. High-performance CRISPR-Cas12a genome editing for combinatorial 613 genetic screening. Nat. Commun. 11, 3455 (2020). 614 15. Zhang, L. et al. AsCas12a ultra nuclease facilitates the rapid generation of therapeutic 615 cell medicines. Nat. Commun. 12, 3908 (2021). 616 16. Vogelstein, B. et al. Cancer Genome Landscapes. Science 339, 1546–1558 (2013). 617 17. Aaltonen, L. A. et al. Pan-cancer analysis of whole genomes. Nature 578, 82–93 (2020). 618 18. Cerami, E. et al. The cBio cancer genomics portal: an open platform for exploring 619 multidimensional cancer genomics data. Cancer Discov. 2, 401–404 (2012). 620 19. Liu, P. et al. Enhanced Cas12a editing in mammalian cells and zebrafish. Nucleic Acids 621 Res. 47, 4169–4180 (2019). 622 20. Yousefi, M. et al. Combinatorial Inactivation of Tumor Suppressors Efficiently Initiates 623 Lung Adenocarcinoma with Therapeutic Vulnerabilities. Cancer Res. 82, 1589–1602 (2022). 624 21. Rogers, Z. N. et al. A quantitative and multiplexed approach to uncover the fitness 625 landscape of tumor suppression in vivo. Nat. Methods 14, 737–742 (2017). 626 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint 22. Cai, H. et al. A functional taxonomy of tumor suppression in oncogenic KRAS-driven 627 lung cancer. Cancer Discov. (2021) doi:10.1158/2159-8290.CD-20-1325. 628 23. Tang, Y. J. et al. A functional map of epigenetic regulators in lung carcinogenesis. 629 Manuscr. Prep. 630 24. Yousefi, M. et al. Fully accessible fitness landscape of oncogene-negative lung 631 adenocarcinoma. Proc. Natl. Acad. Sci. U. S. A. 120, e2303224120 (2023). 632 25. Schaffer, B. E. et al. Loss of p130 Accelerates Tumor Development in a Mouse Model 633 for Human Small-Cell Lung Carcinoma. Cancer Res. 70, 3877–3883 (2010). 634 26. Duplaquet, L. et al. KDM6A epigenetically regulates subtype plasticity in small cell lung 635 cancer. Nat. Cell Biol. 25, 1346–1358 (2023). 636 27. Iacobuzio-Donahue, C. A., Velculescu, V. E., Wolfgang, C. L. & Hruban, R. H. Genetic 637 basis of pancreas cancer development and progression: insights from whole-exome and 638 whole-genome sequencing. Clin. Cancer Res. Off. J. Am. Assoc. Cancer Res. 18, 4257–4265 639 (2012). 640 28. Halbrook, C. J., Lyssiotis, C. A., Pasca di Magliano, M. & Maitra, A. Pancreatic cancer: 641 Advances and challenges. Cell 186, 1729–1754 (2023). 642 29. Izeradjene, K. et al. Kras(G12D) and Smad4/Dpc4 haploinsufficiency cooperate to 643 induce mucinous cystic neoplasms and invasive adenocarcinoma of the pancreas. Cancer Cell 644 11, 229–243 (2007). 645 30. Hingorani, S. R. et al. Trp53R172H and KrasG12D cooperate to promote chromosomal 646 instability and widely metastatic pancreatic ductal adenocarcinoma in mice. Cancer Cell 7, 647 469–483 (2005). 648 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint 31. Bardeesy, N. et al. Smad4 is dispensable for normal pancreas development yet critical in 649 progression and tumor biology of pancreas cancer. Genes Dev. 20, 3130–3146 (2006). 650 32. Aguirre, A. J. et al. Activated Kras and Ink4a/Arf deficiency cooperate to produce 651 metastatic pancreatic ductal adenocarcinoma. Genes Dev. 17, 3112–3126 (2003). 652 33. Hingorani, S. R. et al. Preinvasive and invasive ductal pancreatic cancer and its early 653 detection in the mouse. Cancer Cell 4, 437–450 (2003). 654 34. Whittle, M. C. et al. RUNX3 Controls a Metastatic Switch in Pancreatic Ductal 655 Adenocarcinoma. Cell 161, 1345–1360 (2015). 656 35. Bardeesy, N. et al. Both p16(Ink4a) and the p19(Arf)-p53 pathway constrain progression 657 of pancreatic adenocarcinoma in the mouse. Proc. Natl. Acad. Sci. U. S. A. 103, 5947–5952 658 (2006). 659 36. Chiou, S.-H. et al. Pancreatic cancer modeling using retrograde viral vector delivery and 660 in vivo CRISPR/Cas9-mediated somatic genome editing. Genes Dev. 29, 1576–1585 (2015). 661 37. Hill, A. J. et al. On the design of CRISPR-based single-cell molecular screens. Nat. 662

15, 271–274 (2018). 663 38. Hegde, M., Strand, C., Hanna, R. E. & Doench, J. G. Uncoupling of sgRNAs from their 664 associated barcodes during PCR amplification of combinatorial CRISPR screens. PloS One 665 13, e0197547 (2018). 666 39. Nguyen, B. et al. Genomic characterization of metastatic patterns from prospective 667 clinical sequencing of 25,000 patients. Cell 185, 563-575.e11 (2022). 668 40. Li, A. et al. AAV-CRISPR Gene Editing Is Negated by Pre-existing Immunity to Cas9. 669 Mol. Ther. J. Am. Soc. Gene Ther. 28, 1432–1441 (2020). 670 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint 41. Hughes, N. W. et al. Machine-learning-optimized Cas12a barcoding enables the recovery 671 of single-cell lineages and transcriptional profiles. Mol. Cell 82, 3103-3118.e8 (2022). 672 42. Yang, D. et al. Lineage tracing reveals the phylodynamics, plasticity, and paths of tumor 673 evolution. Cell 185, 1905-1923.e25 (2022). 674 43. Li, L. et al. A mouse model with high clonal barcode diversity for joint lineage, 675 transcriptomic, and epigenomic profiling in single cells. Cell 186, 5183-5199.e22 (2023). 676 44. Bowling, S. et al. An Engineered CRISPR-Cas9 Mouse Line for Simultaneous Readout 677 of Lineage Histories and Gene Expression Profiles in Single Cells. Cell 181, 1410-1422.e27 678 (2020). 679 45. Chan, M. M. et al. Molecular recording of mammalian embryogenesis. Nature 570, 77–680 82 (2019). 681 46. Santinha, A. J. et al. Transcriptional linkage analysis with in vivo AAV-Perturb-seq. 682 Nature 622, 367–375 (2023). 683 47. Jin, X. et al. In vivo Perturb-Seq reveals neuronal and glial abnormalities associated with 684 autism risk genes. Science 370, eaaz6063 (2020). 685 48. Zheng, X. et al. Massively parallel in vivo Perturb-seq reveals cell type-specific 686 transcriptional networks in cortical development. BioRxiv Prepr. Serv. Biol. 687 2023.09.18.558077 (2023) doi:10.1101/2023.09.18.558077. 688 49. Lin, R. et al. Directed evolution of adeno-associated virus for efficient gene delivery to 689 microglia. Nat. Methods 19, 976–985 (2022). 690 50. Li, B. E. et al. In vivo CRISPR/Cas9 screening identifies Pbrm1 as a regulator of myeloid 691 leukemia development in mice. Blood Adv. 7, 5281–5293 (2023). 692 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint 51. Romero, R. et al. Keap1 loss promotes Kras -driven lung cancer and results in 693 dependence on glutaminolysis. Nat. Med. 23, 1362–1368 (2017). 694 52. Ng, S. R. et al. CRISPR-mediated modeling and functional validation of candidate tumor 695 suppressor genes in small cell lung cancer. Proc. Natl. Acad. Sci. U. S. A. 117, 513–521 696 (2020). 697 53. Lee, M. C. et al. A multiplexed in vivo approach to identify driver genes in small cell 698 lung cancer. Cell Rep. 42, 111990 (2023). 699 54. Belk, J. A. et al. Genome-wide CRISPR screens of T cell exhaustion identify chromatin 700 remodeling factors that limit T cell persistence. Cancer Cell 40, 768-786.e7 (2022). 701 55. Song, X. et al. Genome Editing with AAV-BR1-CRISPR in Postnatal Mouse Brain 702 Endothelial Cells. Int. J. Biol. Sci. 18, 652–660 (2022). 703 56. Thürkauf, M. et al. Fast, multiplexable and efficient somatic gene deletions in adult 704 mouse skeletal muscle fibers using AAV-CRISPR/Cas9. Nat. Commun. 14, 6116 (2023). 705 57. Kim, H. K. et al. Deep learning improves prediction of CRISPR-Cpf1 guide RNA 706 activity. Nat. Biotechnol. 36, 239–241 (2018). 707 58. Schwenk, F., Baron, U. & Rajewsky, K. A cre-transgenic mouse strain for the ubiquitous 708 deletion of loxP-flanked gene segments including deletion in germ cells. Nucleic Acids Res. 709 23, 5080–5081 (1995). 710 59. Jackson, E. L. et al. Analysis of lung tumor initiation and progression using conditional 711 expression of oncogenic K-ras. Genes Dev. 15, 3243–3248 (2001). 712 60. Madisen, L. et al. A robust and high-throughput Cre reporting and characterization 713 system for the whole mouse brain. Nat. Neurosci. 13, 133–140 (2010). 714 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint 61. Brinkman, E. K., Chen, T., Amendola, M. & van Steensel, B. Easy quantitative 715 assessment of genome editing by sequence trace decomposition. Nucleic Acids Res. 42, e168 716 (2014). 717 718 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint a H11LSL-Cas12a enAsCas12a CAGGS NLS NLS rBG PolyA loxP loxP 3xHA (E174R/N282A/S542R/K548R) 6x SV40 PolyA Stop Hebert et al. H11Cas12a enAsCas12a CAGGS NLS NLS rBG PolyA loxP 3xHA (E174R/N282A/S542R/K548R) Control Cre - + - + Cas12a αTubulin 140 kDa- 50 kDa- αTubulin HA H11LSL-Cas12a 140 kDa- 50 kDa- 140 kDa- 50 kDa- Brain Heart Kidney Lung Muscle Pancreas Spleen Thymus HA αTubulin H11Cas12a/WT Brain Heart Kidney Liver Lung Muscle Pancreas Spleen Thymus H11LSL-Cas12a/WT Liver HA H11LSL-Cas12a/WT H11Cas12a/WTH11WT/WT H11LSL-Cas12a/LSL-Cas12a H11LSL-Cas12a/WT H11WT/WT H11WT H11LSL-Cas12a 500 bp- 200 bp- d b c e f Figure 1. Generation of Cre-regulated and constitutive Cas12a mice a. Schematic of the H11 LSL-Cas12a transgene. Enhanced AsCas12a containing several substitutions to increase on-target efficiency and PAM binding sequence range (Kleinstiver et al. (2019). Nature Biotechnology), has two nuclear localization sequences (NLS), and three HA tags at the C-terminus (Liu et al. (2019), Nucleic Acids Research), controlled by Cre-mediated removal of the LoxP-Stop-LoxP (LSL) cassette. b. PCR genotyping of mice of the indicated H11 LSL-Cas12a and wild-type (WT) genotypes. c. Western blots on tail tip fibroblasts from H11LSL-Cas12a and wild-type control mice 3 days after Adeno-Cre (Cre) treatment. Cas12a was detected by both anti-HA and anti-Cas12a antibodies. αTubulin shows loading. d. Schematic of the H11 Cas12a transgene, following Cre-mediated removal of the LSL cassette. e. Western blots on tissue lysates from H11Cas12a/WT or H11LSL-Cas12a/WT mice. α−Tubulin shows loading. f. Immunohistochemical staining for the HA tag on Cas12a on liver sections from the indicated genotypes of mice. Scale bars, 25 μm. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Hebert et al. Lenti-U6BC-crNf1/Rasa1/Pten-Cre Wild-type (WT; 106 ifu/mouse; n=5) 9 weeks KrasLSL-G12D ;R26LSL-Tom (2x105 ifu (1/5th titer)/mouse;KT; n=4) H11LSL-Cas12a (+/- R26LSL-Tom(T);106 ifu/mouse; n=6) Lenti-U6BC-crNf1/Rasa1/Pten-Cre crNf1#2 crNf1#3 crNf1#1 crRasa1#2 crRasa1#3 crRasa1#1 crPten#2 crPten#3 crPten#1 X X 27 combinations bU6 pre-crRNA array Guides Repeats U6-integrated diverse barcode (BC) Cre PGK LTRLTR 0.00 0.25 0.50 0.75 1.00 1.25 1.50Lung weight (g) H11LSL-Cas12a WT KT(1/5 th titer) normal lung Light H11LSL-Cas12a; R26LSL-Tom WT KT(1/5th titer) Tomato H&ENKX2-1 H11LSL-Cas12a WT KT(1/5th titer) UCHL1 H11LSL-Cas12a HA WT crNf1#2crNf1#3crNf1#1 crRasa1#2crRasa1#3crRasa1#1 crPten#2crPten#3crPten#1 Indel Percent 0 10 20 30 Figure 2. Rapid and efficient generation of oncogene-negative lung tumors through Cas12a-mediated coincident inactivation of three tumor suppressor genes a. Lenti-U6 BC-crNf1/Rasa1/Pten-Cre pool has guides targeting Nf1, Rasa1, and Pten. Each gene is targeted by three crRNAs in all combinations. b. Design of a lentiviral vector that expresses Cre recombinase and has a pre-crRNA array with three spacers (guides) downstream of a bovine U6 promoter with an integrated barcode region. c. Intratracheal delivery of Lenti-U6 BC-crNf1/Rasa1/Pten-Cre to H11LSL-Cas12a, KrasLSL-G12D;R26LSL-Tom (KT), and wild-type (WT) mice. Viral titer (infectious units, ifu) and mouse numbers are indicated. d. Lung weights from mice of the indicated genotypes 9 weeks after transduction with Lenti-U6 BCcrNf1/Rasa1/Pten-Cre. Each dot represents a mouse. Means +/- standard deviation are indicated. e. Light (upper) and Tomato fluorescent (lower) images of lung lobes from the indicated genotypes of mice 9 weeks after transduction with Lenti-U6 BC-crNf1/Rasa1/Pten-Cre. Dashed lines outline tissues. Tumors are Tomato-positive due to recombination of the R26LSL-Tomato allele. Scale bars, 5 mm. f. Hematoxylin & eosin (upper), NKX2-1 (middle), and UCHL1 (lower) staining of lung sections from the indicated genotypes. Scale bars, 100 μm. g. Immunohistochemical staining for the Cas12a HA tag on lung tissue from representative H11LSL-Cas12a and WT mice transduced with Lenti-U6BC-crNf1/Rasa1/Pten-Cre. Scale bars, 100 μm. h. Indel frequencies for genomic regions targeted by each crRNA within bulk tumor-bearing lung tissue from H11LSL-Cas12a mice. Means +/- standard deviation of three tumor-bearing lungs are shown. a b c d e f g h was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Figure 3. Integration of crRNA arrays with tumor barcoding enables quantification of tumor initiation and tumor size a. Schematic of tumor barcoding with high-throughput BC-crRNA sequencing to determine the size of each Cas12a-induced clonal tumor. b. Total number of neoplastic cells (Total tumor burden) in each mouse normalized to viral titer. Each dot represents a mouse, and bars are means. P-value calculated with Wilcoxon rank-sum test. c. Number of tumors in each mouse normalized to viral titer. Each dot represents a mouse, and bars are means. P-value calculated with Wilcoxon rank-sum test. d. Mean tumor size given a log-normal tumor size distribution. Each dot represents a mouse, and bars are means. P-value calculat- ed with Wilcoxon rank-sum test. e. Number of tumors for each crRNA (guide) relative to the median value for each gene. 95% confidence intervals are shown. f. Mean tumor size assuming a log-normal distribution for each crRNA (guide) relative to the median value for each gene. 95% confidence intervals are shown. Extract DNA from tumor-bearing lungs Determine the number of tumors with each crRNA array and the size of each clonal tumor PCR amplify BC-crRNA region High-throughput sequence Hebert et al. Mean tumor size (# of cells) 0.011 0 1,000 2,000 3,000 4,000 5,000 H11LSL-Cas12a KT Tumor number (# of tumors/105 ifu) 0 10,000 20,000 30,000 ns (0.14) H11LSL-Cas12a KTH11LSL-Cas12a KT 106 107 108 Total tumor burden (# of neoplasatic cells/105 ifu) 0.011a b c d e f 1.2 0.8 1.0 0.6 0.4 Relative tumor number crNf1#2crNf1#3crNf1#1 crRasa1#2crRasa1#3crRasa1#1 crPten#2crPten#3crPten#1 0.2 0.0 1.2 1.0 0.8 0.6 0.2 crNf1#2crNf1#3crNf1#1 crRasa1#2crRasa1#3crRasa1#1 crPten#2crPten#3crPten#1 Relative log-normalmean tumor size 0.4 0.0 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Figure 4. Induction of small-cell lung cancer through the simultaneous Cas12a-mediated inactivation of Trp53, Rb1 and Rbl2 a. Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre pool has guides targeting Trp53, Rb1, and Rbl2. Each gene is targeted by three crRNAs in all 27 combinations. b. Intratracheal delivery of Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre to H11LSL-Cas12a and wild-type (WT) mice. Viral titer (infectious units, ifu) and mouse numbers are indicated. c. Lung weights from mice of the indicated genotypes 18 weeks after transduction with Lenti-U6 BC-crTrp53/Rb1/Rbl2-Cre. Each dot represents a mouse. Mean +/- standard deviation is indicated. d. Light (upper) and Tomato fluorescent (lower) images of lung lobes from the indicated genotypes of mice 18 weeks after transduc- tion with Lenti-U6 BC-crTrp53/Rb1/Rbl2-Cre. Dashed line outlines tissue. Tumors are Tomato-positive due to recombination of the R26LSL-Tomato allele. Scale bars, 5 mm. e. Hematoxylin & eosin (upper), UCHL1 (middle) and HA tag (lower) staining of lung sections from an H11LSL-Cas12a mouse. Scale bars, 100 μm. f. Total number of neoplastic cells (Total tumor burden) in each mouse normalized to viral titer. Each dot represents a mouse, and the bar is the mean. g. The number of tumors in each mouse normalized to viral titer. Each dot represents a mouse, and the bar is the mean. h. Indel frequencies for genomic regions targeted by each crRNA within micro-dissected tumor tissue from H11LSL-Cas12a mice. Symbol shapes and colors identify values from the same sample. Bars represent means. i,j. Number of tumors (i) and mean tumor size assuming a log-normal distribution (j) for each crRNA (guide) relative to the median value for each gene. 95% confidence intervals are shown. 0.00 0.25 0.50 0.75 1.00 1.25 1.50Lung weight (g) H11LSL-Cas12a WT Hebert et al. cba e fd crTrp53#2 crTrp53#3 crTrp53#1 crRb1#2 crRb1#3 crRb1#1 crRbl2#2 crRbl2#3 crRbl2#1 X X 27 combinations Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre g H11LSL-Cas12a/LSL-Cas12a (+/- R26LSL-Tom; n=6) Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre Wild type (WT; n=4) 18 weeks (106 ifu/mouse for all) H11LSL-Cas12a 0 8,000 2,000 4,000 6,000 Tumor number (# of tumors/105 ifu) H11LSL-Cas12a Total tumor burden (# of neoplasatic cells/105 ifu) 106 107 108 H&E H11LSL-Cas12a UCHL1 HA H11LSL-Cas12a;R26LSL-Tom WT Tomato Light h i 0 20 40 60 80 100Indel Percent crTrp53#2crTrp53#3crTrp53#1 crRb1#2crRb1#3crRb1#1 crRbl2#2crRbl2#3crRbl2#1 Relative log-normalmean tumor size 1.2 0.6 0.4 0.2 0.0 crTrp53#2crTrp53#3crTrp53#1 crRb1#2crRb1#3crRb1#1 crRbl2#2crRbl2#3crRbl2#1 j Relative tumor number 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 crTrp53#2crTrp53#3crTrp53#1 crRb1#2crRb1#3crRb1#1 crRbl2#2crRbl2#3crRbl2#1 0.8 1.0 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Hebert et al. cba Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre crTrp53#2 crTrp53#3 crTrp53#1 crCdkn2a#2 crCdkn2a#3 crCdkn2a#1 crSmad4#2 crSmad4#3 crSmad4#1 X X 27 combinations fed KT;H11LSL-Cas12a (n=4) Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre KrasLSL-G12D ;R26LSL-Tom (KT; n=4) H11LSL-Cas12a (n=4) 0 5 10 0 50 100 Weeks post-transduction Percent survival KT;H11LSL-Cas12a H11LSL-Cas12a KT KT;H11LSL-Cas12a KT;H11LSL-Cas12a KT Tomato Light H&E KT;H11LSL-Cas12a KT H11LSL-Cas12a AB (106 ifu/mouse for all) KTKT; H11LSL-Cas12a H11LSL-Cas12a Tumor number (# of tumors/105 ifu) ns (0.39) 102 103 104 105 100 101 Mean tumor size (# of cells) KTKT; H11LSL-Cas12a H11LSL-Cas12a 0.043 1,000 10,000 g h i Figure 5. Rapid generation of PDAC through somatic Cas12a-mediated inactivation of commonly mutated tumor suppressor genes a. Lenti-U6 BC-crTrp53/Cdkn2a/Smad4-Cre pool has guides targeting Trp53, Cdkn2a, and Smad4. Each gene is targeted by three crRNAs in all 27 combinations. b. Intrapancreatic delivery of Lenti-U6 BC-crTrp53/Cdkn2a/Smad4-Cre to KrasLSL-G12D;R26LSL-Tom (KT), KT;H11LSL-Cas12a, and H11LSL-Cas12a mice. Viral titer (infectious units, ifu) and mouse numbers are indicated. c. Survival curve of mice of the indicated genotypes transduced with Lenti-U6 BC-crTrp53/Cdkn2a/Smad4-Cre. d. Light (upper) and fluorescent Tomato (lower) pancreas images from the indicated genotypes of mice 6 weeks (KT;H11LSL-Cas12a) or 11 weeks (KT) after transduction with Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre. Dashed lines outline tissues. Scale bars, 5 mm. e. Light and fluorescent Tomato images of a liver lobe from a KT;H11LSL-Cas12a mouse. Dashed line outlines tissue. Scale bars, 5 mm. f. Immunohistochemistry images of pancreas sections from the indicated genotypes, showing hematoxylin & eosin (H&E; upper) and Alcian Blue (AB; lower) staining. KT;H11LSL-Cas12a panels show representative regions with poorly differentiated sarcomatoid carcinoma (left; approximately 80-90% of tumor area) and more differentiated areas with mucinous cells and some lobular structure (right). Scale bars, 100 μm. g. The number of tumors in each mouse normalized to viral titer. Each dot represents a mouse, and bars are means. Note that two H11LSL-Cas12a mice did not have detectable tumor burden and are not plotted. P-value calculated with Wilcoxon rank-sum test. h. Mean tumor size given a log-normal tumor size distribution. Each dot represents a mouse, and bars are means. Note that two H11LSL-Cas12a mice did not have detectable tumor burden and are not plotted. P-value calculated with Wilcoxon rank-sum test. i. Number of tumors for each crRNA (guide) relative to the median value for each gene. 95% confidence intervals are shown. crTrp53#2crTrp53#3crTrp53#1 crCdkn2a#2crCdkn2a#3crCdkn2a#1 crSmad4#2crSmad4#3crSmad4#1 2.00 1.25 0.50 0.25 0.00 Relative tumor number 1.75 1.50 1.00 0.75 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Hebert et al. Ability to sequence across 3 guides difficult easy Cas9 Cas12a RNA scaffold tracrRNA (~100 nt/guide) Direct repeats (DR;~20 nt each) Promoter one per sgRNA one per crRNA array Cut site relative to PAM 3 bp from PAM 18 bp from the PAM Cuts blunt-end 5 nt staggered Cloning one step per sgRNA one step per crRNA array Structural features Cutting Guide uncoupling due to lentiviral template switching high low a H11LSL-Cas12a/WT H11Cas12a/WT H11LSL-Cas12a H11Cas12a500 bp- 200 bp- LungLiverKidney H11LSL-Cas12a/WT H11Cas12a/WTH11WT/WT HA Supplementary Figure 1. Comparison of salient features of Cas9 and Cas12a, and broad nuclear expression of Cas12a in H11Cas12a mice a. Summary of features of Cas9 and Cas12a relevant to multiplexed genome editing. Note that the potential impact of off-target effects is increased when targeting more genes. b. PCR genotyping of mice of the indicated H11 LSL-Cas12a and H11Cas12a genotypes. c. Immunohistochemical staining for the Cas12a HA tag in lung, liver and kidney sections from the indicated genotypes of mice. Scale bars, 50 µm. Higher magnification images of these liver sections is shown in Figure 1f. b c was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Supplementary Figure 2. Incorporation of Cas12a-mediated genome editing with Tuba-seq Ultra enables quanitification of the effects of each crRNA array a,b. Number of tumors (a) and mean tumor size assuming a log-normal distribution (b) for each crRNA array (27 combinations in total) relative to the median value for all arrays in H11LSL-Cas12a mice. 95% confidence intervals are shown. Hebert et al. a b Pten crRNA # 1 2 1 1 1 1 1 1 1 12 2 2 2 2 2 2 23 3 3 3 3 3 3 3 Rasa1 crRNA # 1 1 2 3 1 2 3 1 2 32 3 1 2 3 1 2 31 2 3 1 2 3 2 3 Nf1 crRNA # 1 1 1 1 2 2 2 3 3 31 1 2 2 2 3 3 31 1 1 2 2 2 3 1 3 3 3 Relative tumor number 1.6 0.8 0.6 0.4 0.2 0.0 1.0 1.2 1.4 Pten crRNA # 1 2 1 1 1 1 1 1 1 12 2 2 2 2 2 2 23 3 3 3 3 3 3 3 Rasa1 crRNA # 1 1 2 3 1 2 3 1 2 32 3 1 2 3 1 2 31 2 3 1 2 3 2 3 Nf1 crRNA # 1 1 1 1 2 2 2 3 3 31 1 2 2 2 3 3 31 1 1 2 2 2 3 1 3 3 3 Relative log-normal mean tumor size 1.2 0.6 0.8 0.4 0.0 0.2 1.0 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Hebert et al. a c Supplementary Figure 3. Inactivation of Trp53, Cdkn2a, and Smad4 greatly increases tumor size and leads to the development of PDAC with differentiated areas with CK19+ cancer cells and desmoplastic stroma as well as more poorly differentiated areas a. Immunohistochemistry images of pancreas sections from the indicated genotypes, showing Trichrome (upper) and Cytokeratin-19 (CK19; lower) staining. KT;H11LSL-Cas12a panels show representative poorly differentiated (left) and more differentiated (right) regions. Scale bars, 100 μm. b. Immunohistochemical staining for the Cas12a HA tag on pancreas tissue from representative KT; H11LSL-Cas12a, KT and H11LSL-Cas12a mice transduced with Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre. Scale bars, 100 μm. c. Total number of neoplastic cells (Total tumor burden) in each mouse normalized to viral titer. Each dot represents a mouse and the bar is the mean. Note that two H11LSL-Cas12a mice did not have detectable tumor burden and are not plotted. P-value calculated with Wilcoxon rank-sum test. d. Probability distribution of the size of tumors in the indicated genotypes of mice. e. Indel frequencies for genomic regions targeted by each crRNA within bulk tumor-bearing pancreas tissue from KT; H11LSL-Cas12a mice. Tissue samples had variable tumor burden, likely explaining observed variability in indel frequencies. Symbol shapes and colors identify values from the same mouse. Bars represent means. f. Mean tumor size assuming a log-normal distribution for each crRNA (guide) relative to the median value for each gene. 95% confidence intervals are shown. TricrhomeCK19 KT;H11LSL-Cas12a KT H11LSL-Cas12a d KTKT; H11LSL-Cas12a H11LSL-Cas12a 0.021 106 107 108 Total tumor burden (# of neoplasatic cells/105 ifu) 103 104 105 KT KT;H11LSL-Cas12a H11LSL-Cas12a Tumor size (# of neoplastic cells) Probability 106 107103 104 105 10-3 10-2 10-1 100 10-5 10-4 b e HA KT;H11LSL-Cas12a KT H11LSL-Cas12a crTrp53#2crTrp53#3crTrp53#1 crCdkn2a#2crCdkn2a#3crCdkn2a#1 crSmad4#2crSmad4#3crSmad4#1 Indel Percent 0 10 20 30 40 f crTrp53#2crTrp53#3crTrp53#1 crCdkn2a#2crCdkn2a#3crCdkn2a#1 crSmad4#2crSmad4#3crSmad4#1 3.0 1.5 1.0 Relative log-normalmean tumor size 2.0 2.5 0.5 0.0 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre Hebert et al. Fraction of clonal barcodes associated with only one crRNA array H11LSL-Cas12a KT 1.04 1.02 1.00 0.98 0.96 0.94 0.92 0.90 Fraction of total tumor burden after filtering 1.01 1.00 0.99 0.98 0.97 0.96 0.95 H11LSL-Cas12a KT 1.04 1.02 1.00 0.98 0.96 0.94 0.92 0.90 Fraction of clonal barcodes associated with only one crRNA array H11LSL-Cas12a Fraction of total tumor burden after filtering 1.01 1.00 0.99 0.98 0.97 0.96 0.95 H11LSL-Cas12a Lenti-U6BC-crNf1/Rasa1/Pten-Cre a Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre b c d e f Supplementary Figure 4. Generation of Cas12a Lenti-U6BC-crRNA-Cre vector libraries is associated with minimal shuffling of crRNA sequences a,c,e. Fraction of all U6-integrated barcode reads that are associated with only a single crRNA array within each vector pool in the indicated mouse genotypes: Lenti-U6 BC-crNf1/Rasa1/Pten-Cre (a), Lenti-U6 BC-crTrp53/Rb1/Rbl2-Cre (c), and Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre (e). Bars represent medians. b,d,f. Fraction of total neoplastic cells (total tumor burden) remaining after filtering out spurious “tumor” reads arising during library preparation and sequencing of each pool in the indicated mouse genotypes: Lenti-U6 BC-crNf1/Rasa1/Pten-Cre (b), Lenti-U6BC-crTrp53/Rb1/Rbl2-Cre (d), and Lenti-U6BC-crTrp53/Cdkn2a/Smad4-Cre (f). Bars represent medians. 1.05 1.00 0.95 0.90 0.85 0.80 Fraction of clonal barcodes associated with only one crRNA array KTKT; H11LSL-Cas12a H11LSL-Cas12a Fraction of total tumor burden after filtering 1.0 0.9 0.8 0.7 0.6 0.5 0.4 KTKT; H11LSL-Cas12a H11LSL-Cas12a was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted May 25, 2024. ; https://doi.org/10.1101/2024.03.07.583774doi: bioRxiv preprint