CRISPR-based genome editing has revolutionized functional genomics, enabling thousands of perturbations to be concurrently assayed in single experiments. However, for methods such as saturation genome editing (SGE), which aims to generate and assay libraries of point mutations, a challenge is that only one region (e.g., one exon) is studied per experiment. Here, we describe prime-SGE, a prime editing-based framework in which libraries of specific point mutations are installed into genes throughout the genome and then functionally assessed by sequencing of prime editing guide RNAs (pegRNAs) rather than the mutations themselves. We apply prime-SGE in two cell lines to assay thousands of point mutations in eight oncogenes for their ability to confer drug resistance to four tyrosine kinase inhibitors. Our prime-SGE strategy, combined with ongoing improvements in prime editing efficiency, opens the door to efficient positive selection screens of large numbers of point mutations at locations throughout the genome.

We developed pgMAP, an analysis pipeline to map gRNA sequencing reads from dual-targeting CRISPR screens. pgMAP output includes a dual gRNA read counts table and quality control metrics including the proportion of correctly-paired reads and CRISPR library sequencing coverage across all time points and samples. pgMAP is implemented using Snakemake and is available open-source under the MIT license at https://github.com/fredhutch/pgmap_pipeline.

Standard care of lung cancer treatment has shifted away from non-specific, cytotoxic chemotherapy in favor of targeted therapies based on genetic mutations within tumors. In 2014, somatic mutations in the small GTPase RIT1 (Ras-like in all tissues) were discovered as oncogenic drivers of lung adenocarcinoma. Thousands of patients per year are diagnosed with RIT1-driven cancer, but treatment options are limited. A targeted therapy for RIT1-driven tumors would address a major unmet clinical need. Little is known about how RIT1 drives cellular transformation. To genetically dissect RIT1 function, we performed a genome-wide CRISPR/Cas9 screen in isogenic PC9 lung adenocarcinoma cells. This screen took advantage of the observation that RIT1-mutant cells are resistant to EGFR inhibition. We leveraged this drug resistance phenotype to identify genetic dependencies (gene knockouts that are detrimental to cell growth) and cooperating factors (gene knockouts that are beneficial to cell growth) in RIT1-mutant cells. From this screen, we found that one of the top essential genes was the deubiquitinase USP9X. This is intriguing given that the protein abundance of RIT1 is known to be important for its function. Therefore, we sought out to test the hypothesis that USP9X regulates RIT1 abundance and that inhibition of USP9X could be an effective therapeutic strategy for abrogating RIT1-driven tumor growth. Our model suggests that USP9X promotes proteasome-mediated degradation of RIT1. To test this, we assessed RIT1 abundance in the context of USP9X knockout (KO). We found that RIT1 protein abundance was decreased in USP9X KO PC9 cells compared to parental cells. Furthermore, cycloheximide (CHX)-chase experiments revealed that RIT1 stability was decreased in USP9X KO cells, and RIT1 degraded faster than in parental cells. The average half-life of RIT1 in USP9X KO cells was 3.4 hrs while the average half-life in parental PC9 cells was 12.3 hrs (95% CI = -12.5 to -5.4 hrs). Treatment with the proteasome inhibitor bortezomib (BTZ) rescued RIT1 degradation by 99% in parental cells and 190% in USP9X KO cells (95% CI = 57 to 126%). In addition to assessing protein abundance and stability, we performed co-immunoprecipitation experiments in RIT1-expressing HEK293T cells and found that RIT1 and USP9X physically interact. Taken together, these data support the hypothesis that RIT1 is a substrate of USP9X. In addition to providing better insight on the protein regulation of RIT1, this work has crucial therapeutic implications. The protein abundance of RIT1 is important for its function, and our model suggests that USP9X inhibition could be an effective means of reducing RIT1 protein abundance and abrogating tumor growth. Overall, this work is poised to significantly impact the field of RIT1 biology and address a major unmet clinical need for the treatment of RIT1-driven diseases. Citation Format: Amanda K. Riley, Athea Vichas, Naomi T. Nkinsi, Phoebe C. Parrish, Shriya Kamlapurkar, Alice H. Berger. Identification of USP9X as a novel regulator of RIT1 protein abundance and as a potential therapeutic target in RIT1-driven lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2972.

Supravalvar aortic stenosis (SVAS) is a characteristic feature of Williams-Beuren syndrome (WBS). SVAS is present in 67% of those with WBS, but severity varies; 21% have clinically significant SVAS requiring surgical intervention while 33% have no appreciable aortic disease. Little is known about genetic modifiers outside the 7q11.23 region that might contribute to SVAS severity. To investigate, we collaboratively phenotyped 473 individuals with WBS and performed the largest whole-genome- sequencing study to date. We developed a set of strategies for modifier discovery including extreme phenotyping (surgical SVAS vs. no SVAS) and prioritization of non-synonymous variants with increased predicted functional impact along with an allele frequency difference between the extreme phenotype groups. We identified pathways enriched in common or less frequent variants, followed by association testing of SVAS severity with the enriched pathways. The common variant analysis identified pathways including the extracellular matrix and the innate immune system, while pathways encompassing adaptive immunity, ciliary function, lipid metabolism and PI3KAKT were captured by both the common and less frequent variant analyses. Cell cycle and estrogen responsive pathways were among those identified through the less frequent variant analysis. Among the 69 genes reported in other large genome wide association studies assessing aortic traits, 11 genes, including PCSK9 and ILR6, were found in our study, suggesting overlapping disease mechanisms. In summary, this study presents novel strategies for identification of disease modifiers in rare conditions like WBS. ![Figure][1] ### Competing Interest Statement All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: BPR, CAM, and CBM have received funding from the WSA in the past. As a parent advocacy group, the WSA does have an interest in the submitted work but does not stand to financially profit from the findings. Additional government support noted above. No financial relationships with any other organizations that might have an interest in the submitted work occurred in the previous three years; no other relationships or activities that could appear to have influenced the submitted work are noted. ### Funding Statement The NIH effort was supported by the NHLBI Division of Intramural Research (BAK). BPR and CAM were supported by grants from the Williams Syndrome Association (WSA) and LRO received funding from the Canadian Institutes for Health Research (MOP77720. CBM received support from the National Institute of Neurological Disorders and Stroke (R01 NS35102) and the WSA (WSA 0104 and WSA 0111). CAM and LRO were also partially supported by subcontracts from the grant of R01 NS35102. The Genomic Disorder Biobank of the Telethon Network of Genetic Biobanks was supported by Telethon Italy grant GTB12001G, GM) ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: Institutional Review Board of the National Institutes of Health, the University of Nevada School of Medicine Internal Review Board, the University of Toronto Health Sciences Research Ethics Board, the Boston Children Hospital Internal Review Board, and Fondazione IRCCS Casa Sollievo della Sofferenza Ethics Board gave ethical approval for this work. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines and uploaded the relevant EQUATOR Network research reporting checklist(s) and other pertinent material as supplementary files, if applicable. Yes The summary statistics of the non-synonymous variants in no SVAS and surgical SVAS groups will be available upon reasonable request to the corresponding author, Dr. Beth Kozel via beth.kozel@nih.gov. [1]: pending:yes

Synthetic lethal therapies are a promising approach to expand therapeutic options for cancer patients. Since synthetic lethal therapies target tumor cells specifically, they have fewer off-target effects than oncogene targeted approaches. To identify new cancer drug targets, we focused on paralogs, ancestrally-duplicated genes that frequently retain redundant or overlapping functions. To find synthetic lethal human paralogs, we developed paired guide RNAs for Paralog gENetic interaction mapping (pgPEN), a pooled CRISPR-Cas9 single and double knockout approach targeting over 2,000 paralogs. We applied pgPEN to two cancer cell lines and found that 12% of human paralogs exhibit synthetic lethality in at least one context. To our knowledge, pgPEN represents the largest experimental assessment of human paralog synthetic lethality to date. We next identified paralog pairs to prioritize for follow-up study. A key drawback to synthetic lethal therapies is that many genetic interactions are context-dependent. To identify likely penetrant interactions, we compared our data to other published paralog screens and computational predictions of paralog synthetic lethality. We found that over 75% (n=96) of pgPEN hits were predicted to be broadly synthetic lethal, and nearly 10% (n=10) of pairs were synthetic lethal in multiple paralog screens. Finally, we prioritized paralogs targeted by existing small molecule therapies. Of the 122 synthetic lethal pairs identified by pgPEN, 16% (n=20) are currently druggable. We mined drug repurposing data from DepMap to find pairs where a paralog-targeting drug showed a stronger effect in cell lines with low target gene copy number or expression relative to cell lines with normal target gene copy number or expression. These drugs could selectively target cancer cells in cases where one or both paralogs is lost in tumors but retained in normal tissue. We also leveraged The Cancer Genome Atlas tumor sequencing data to find paralog pairs where one member is recurrently lost in cancer. Taken together, these studies identify druggable, highly penetrant synthetic lethal paralog interactions. In combination with tumor sequencing data, we identify high-priority paralog drug targets that can be further tested and translated to the clinic. Paralog synthetic lethal therapies could provide a relatively low-toxicity therapeutic approach to improve the efficacy of cancer treatments and prevent the emergence of acquired resistance. Citation Format: Phoebe C. Parrish, James D. Thomas, Austin M. Gabel, Shriya Kamlapurkar, Robert K. Bradley, Alice H. Berger. Expanding cancer therapy options by leveraging synthetic lethal interactions between druggable paralogs [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2278.

Abstract Standard care of lung cancer is moving away from chemotherapy in favor of personalized approaches based on specific mutations in each tumor. Genome-sequencing studies have identified somatic mutations in the small GTPase RIT1 (Ras-like in all tissues) in lung adenocarcinoma patients. Of the identified mutations, M90I is the most recurrent. Thousands of patients per year are diagnosed with RIT1-driven cancer, but treatment options are limited. A targeted therapy for RIT1-driven disease could greatly improve patient outcomes. Little is known about how RIT1 drives cellular transformation. To genetically dissect signaling pathways downstream of RIT1, we performed a genome-wide CRISPR/Cas9 screen in isogenic PC9 lung adenocarcinoma cells in which cell survival is dependent on expression of RIT1M90I. We found that RIT1-mutant cells were highly dependent on components of the Spindle Assembly Checkpoint (SAC), including the Aurora kinases (A and B). The SAC is a surveillance mechanism that ensures proper chromosome segregation during mitosis. We hypothesized that RIT1M90I weakened the SAC and rendered cells vulnerable to loss of mitotic regulators. To explore this, we performed time-lapse imaging in HeLa H2B-GFP cells stably expressing RIT1M90I. In parental cells, the median duration of mitosis was 70.5 min (95% CI = 63 – 82 min), while in RIT1M90I-mutant cells, this was reduced to 48 min (95% CI = 45 – 51 min). Mitotic index was unaffected, suggesting that RIT1M90I does not regulate mitotic entry. This difference in mitotic timing was eliminated by treatment with reversine, an inhibitor that abolishes the SAC, demonstrating that RIT1M90I perturbs mitotic timing at the level of the SAC. If RIT1M90I weakens the SAC, we would expect higher prevalence of chromosomal abnormalities (such as chromosome bridges) in RIT1M90I-mutant cells. Indeed, analysis of fixed-cell populations indicated that RIT1M90I-mutant cells showed higher prevalence of mitotic abnormalities (79% of RIT1M90I-mutant cells compared to 46% of parental cells, 95% CI= 15 – 53). These data support the hypothesis that RIT1M90I weakens the SAC and imply that further SAC perturbation could be lethal. To explore this, we performed a small molecule screen of 160 clinically-relevant inhibitors in PC9-RIT1M90I and PC9-KRASG12V cells. Intriguingly, RIT1M90I-mutant cells were more sensitive than RAS-mutant cells to alisertib and barasertib, inhibitors of Aurora kinase A and B, respectively. Furthermore, alisertib or barasertib treatment abrogated soft agar colony formation in RIT1M90I-mutant cells but not in RAS-mutant cells. Together with our functional genomic analysis, we propose a model whereby expression of RIT1M90I weakens the SAC, thereby accelerating mitosis, increasing the abundance of mitotic abnormalities, and rendering cells vulnerable to genetic knockdown of SAC genes or small molecule inhibition of Aurora kinases A/B. Citation Format: Amanda Riley, Athea Vichas, Naomi T. Nkinsi, Phoebe C. Parrish, Shriya Kamlapurkar, Alice H. Berger. The spindle assembly checkpoint as a therapeutic vulnerability in RIT1-mutant lung adenocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2269.

Abstract Genome-scale CRISPR-Cas9 screens have enabled the identification of essential genes that can serve as cancer drug targets. However, single-gene knockout screens frequently underestimate the essentiality of paralogs, ancestrally duplicated genes that often functionally compensate for one another's loss; we have observed this directly in single-gene CRISPR screen data from a lung cancer cell line (P < 1E-16). Since the human genome is highly duplicated, our ability to identify novel drug targets is greatly limited. To identify paralog pairs essential for cancer cell survival, we developed a multiplexed CRISPR approach that uses paired guide RNAs to knock out human paralogs individually and in pairs. Our library includes 2,060 paralogs (1,030 pairs); the largest human paralog CRISPR library to date. We screened lung adenocarcinoma (PC9) and cervical carcinoma (HeLa) cell lines to identify synthetic lethal paralogs which have minimal growth effects when targeted individually but whose simultaneous loss leads to severely decreased growth. We found that 128 (16%) of the paralog pairs in our study were synthetic lethal and essential in at least one cell line. Gene set enrichment analysis revealed that our top paralogs are overrepresented in pathways related to cell cycle regulation, protein secretion, DNA repair, and PI3K-AKT signaling (FDR q-value < 2.52E-4). Importantly, 15 (18%) synthetic lethal paralog pairs have at least one member that is currently druggable. We validated our screen results using in vitro competition assays and DepMap CRISPR data analysis. For known synthetic lethal paralogs MAGOH and MAGOHB, competition assay data confirmed significantly reduced growth in the dual gene knockout condition versus single-gene knockouts (P < 0.05) and analysis of DepMap CRISPR score correlation with paralog expression showed that MAGOHB is more essential in cell lines with low MAGOH expression (P < 4.6E-41). For novel synthetic lethal paralogs CCNL1/CCNL2, PSMB5/PSMB8, and OXSR1/STK39, DepMap CRISPR score vs. expression correlation analysis again confirmed our findings as cell lines with low expression of paralog 1 showed lower CRISPR scores when paralog 2 was knocked out (P < 1.2E-8 for each pair), suggesting that our novel synthetic lethal paralogs are essential across a range of cancer cell lines. These interactions are also being validated via competition assay, and CRISPR knockout efficiency will be confirmed via Western blot and genomic DNA sequencing. Our studies point to a number of novel synthetic lethal paralogs that could serve as lung cancer drug targets. These paralogs could be targeted alone or in combination with existing therapies to suppress lung cancer growth and prevent acquired drug resistance. Paralog synthetic lethal therapies could make a major impact on clinical care by improving patient outcomes. Citation Format: Phoebe C. Parrish, James D. Thomas, Shriya Kamlapurkar, Robert K. Bradley, Alice H. Berger. Enabling cancer drug target discovery through genome-scale identification of synthetic lethal paralog pairs [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2139.

CRISPR screens have accelerated the discovery of important cancer vulnerabilities. However, single-gene knockout phenotypes can be masked by redundancy among related genes. Paralogs constitute two-thirds of the human protein-coding genome, so existing methods are likely inadequate for assaying a large portion of gene function. Here, we develop paired guide RNAs for paralog genetic interaction mapping (pgPEN), a pooled CRISPR-Cas9 single- and double-knockout approach targeting more than 2,000 human paralogs. We apply pgPEN to two cell types and discover that 12% of human paralogs exhibit synthetic lethality in at least one context. We recover known synthetic lethal paralogs MEK1/MEK2, important drug targets CDK4/CDK6, and other synthetic lethal pairs including CCNL1/CCNL2. Additionally, we identify ten tumor suppressor paralog pairs whose compound loss promotes cell proliferation. These findings nominate drug targets and suggest that paralog genetic interactions could shape the landscape of positive and negative selection in cancer.

CRISPR-based cancer dependency maps are accelerating advances in cancer precision medicine, but adequate functional maps are limited to the most common oncogenes. To identify opportunities for therapeutic intervention in other rarer subsets of cancer, we investigate the oncogene-specific dependencies conferred by the lung cancer oncogene, RIT1. Here, genome-wide CRISPR screening in KRAS, EGFR, and RIT1-mutant isogenic lung cancer cells identifies shared and unique vulnerabilities of each oncogene. Combining this genetic data with small-molecule sensitivity profiling, we identify a unique vulnerability of RIT1-mutant cells to loss of spindle assembly checkpoint regulators. Oncogenic RIT1M90I weakens the spindle assembly checkpoint and perturbs mitotic timing, resulting in sensitivity to Aurora A inhibition. In addition, we observe synergy between mutant RIT1 and activation of YAP1 in multiple models and frequent nuclear overexpression of YAP1 in human primary RIT1-mutant lung tumors. These results provide a genome-wide atlas of oncogenic RIT1 functional interactions and identify components of the RAS pathway, spindle assembly checkpoint, and Hippo/YAP1 network as candidate therapeutic targets in RIT1-mutant lung cancer.

Cuella-Martin et al. (2021) and Hanna et al. (2021) showcase CRISPR base editing in large-scale pooled screens in human cells to discover both loss- and gain-of-function variants, enabling protein structure/function insights and clinical variant interpretation.

ABSTRACTAdvances in precision oncology have transformed cancer therapy from broadly-applied cytotoxic therapy to personalized treatments based on each tumor’s unique molecular alterations. Here we investigate the oncogene-specific dependencies conferred by lung cancer driver variants ofKRAS, EGFR, andRIT1. Integrative analysis of genome-wide CRISPR screens in isogenic cell lines identified shared and unique vulnerabilities of each oncogene. The non-identical landscape of dependencies underscores the importance of genotype-guided therapies to maximize tumor responses. Combining genetic screening data with small molecule sensitivity profiling, we identify a unique vulnerability ofRIT1-mutant cells to loss of spindle assembly checkpoint regulators. This sensitivity may be related to a novel role of RIT1 in mitosis; we find that oncogenic RIT1M90Ialters mitotic timing via weakening of the spindle assembly checkpoint. In addition, we uncovered a specific cooperation of mutantRIT1with loss of Hippo pathway genes. In human lung cancer,RIT1mutations and amplifications frequently co-occur with loss of Hippo pathway gene expression. These results provide the first genome-wide atlas of oncogenicRIT1-cooperating factors and genetic dependencies and identify components of the RAS pathway, spindle assembly checkpoint, and Hippo/YAP1 network as candidate therapeutic targets inRIT1-mutant lung cancer.

Williams-Beuren syndrome (WBS) is a multisystem disorder caused by a hemizygous deletion on 7q11.23 encompassing 26-28 genes. An estimated 2-5% of patients have "atypical" deletions, which extend in the centromeric and/or telomeric direction from the WBS critical region. To elucidate clinical differentiators among these deletion types, we evaluated 10 individuals with atypical deletions in our cohort and 17 individuals with similarly classified deletions previously described in the literature. Larger deletions in either direction often led to more severe developmental delays, while deletions containing MAGI2 were associated with infantile spasms and seizures in patients. In addition, head size was notably smaller in those with centromeric deletions including AUTS2. Because children with atypical deletions were noted to be less socially engaged, we additionally sought to determine how atypical deletions relate to social phenotypes. Using the Social Responsiveness Scale-2, raters scored individuals with atypical deletions as having different social characteristics to those with typical WBS deletions (p = .001), with higher (more impaired) scores for social motivation (p = .005) in the atypical deletion group. In recognizing these distinctions, physicians can better identify patients, including those who may already carry a clinical or FISH WBS diagnosis, who may benefit from additional molecular evaluation, screening, and therapy. In addition to the clinical findings, we note mild endocrine findings distinct from those typically seen in WBS in several patients with telomeric deletions that included POR. Further study in additional telomeric deletion cases will be needed to confirm this observation.

Supravalvular aortic stenosis (SVAS) is a narrowing of the aorta caused by elastin (ELN) haploinsufficiency. SVAS severity varies among patients with Williams-Beuren syndrome (WBS), a rare disorder that removes one copy of ELN and 25-27 other genes. Twenty percent of children with WBS require one or more invasive and often risky procedures to correct the defect while 30% have no appreciable stenosis, despite sharing the same basic genetic lesion. There is no known medical therapy. Consequently, identifying genes that modify SVAS offers the potential for novel modifier-based therapeutics. To improve statistical power in our rare-disease cohort (N = 104 exomes), we utilized extreme-phenotype cohorting, functional variant filtration and pathway-based analysis. Gene set enrichment analysis of exome-wide association data identified increased adaptive immune system variant burden among genes associated with SVAS severity. Additional enrichment, using only potentially pathogenic variants known to differ in frequency between the extreme phenotype subsets, identified significant association of SVAS severity with not only immune pathway genes, but also genes involved with the extracellular matrix, G protein-coupled receptor signaling and lipid metabolism using both SKAT-O and RQTest. Complementary studies in Eln+/-; Rag1-/- mice, which lack a functional adaptive immune system, showed improvement in cardiovascular features of ELN insufficiency. Similarly, studies in mixed background Eln+/- mice confirmed that variations in genes that increase elastic fiber deposition also had positive impact on aortic caliber. By using tools to improve statistical power in combination with orthogonal analyses in mice, we detected four main pathways that contribute to SVAS risk.

BACKGROUND:Large, multigenic deletions at chromosome 7q11.23 result in a highly penetrant constellation of physical and behavioral symptoms known as Williams-Beuren syndrome (WS). Of particular interest is the unusual social-cognitive profile evidenced by deficits in social cognition and communication reminiscent of autism spectrum disorders (ASD) that are juxtaposed with normal or even relatively enhanced social motivation. Interestingly, duplications in the same region also result in ASD-like phenotypes as well as social phobias. Thus, the region clearly regulates human social motivation and behavior, yet the relevant gene(s) have not been definitively identified. METHOD:Here, we deeply phenotyped 85 individuals with WS and used exome sequencing to analyze common and rare variation for association with the remaining variance in social behavior as assessed by the Social Responsiveness Scale. RESULTS:We replicated the previously reported unusual juxtaposition of behavioral symptoms in this new patient collection, but we did not find any new alleles of large effect in the targeted analysis of the remaining copy of genes in the Williams syndrome critical region. However, we report on two nominally significant SNPs in two genes that have been implicated in the cognitive and social phenotypes of Williams syndrome, BAZ1B and GTF2IRD1. Secondary discovery driven explorations focusing on known ASD genes and an exome wide scan do not highlight any variants of a large effect. CONCLUSIONS:Whole exome sequencing of 85 individuals with WS did not support the hypothesis that there are variants of large effect within the remaining Williams syndrome critical region that contribute to the social phenotype. This deeply phenotyped and genotyped patient cohort with a defined mutation provides the opportunity for similar analyses focusing on noncoding variation and/or other phenotypic domains.

Current use of microbes for metabolic engineering suffers from loss of metabolic output due to natural selection. Rather than combat the evolution of bacterial populations, we chose to embrace what makes biological engineering unique among engineering fields - evolving materials. We harnessed bacteria to compute solutions to the biological problem of metabolic pathway optimization. Our approach is called Programmed Evolution to capture two concepts. First, a population of cells is programmed with DNA code to enable it to compute solutions to a chosen optimization problem. As analog computers, bacteria process known and unknown inputs and direct the output of their biochemical hardware. Second, the system employs the evolution of bacteria toward an optimal metabolic solution by imposing fitness defined by metabolic output. The current study is a proof-of-concept for Programmed Evolution applied to the optimization of a metabolic pathway for the conversion of caffeine to theophylline in E. coli. Introduced genotype variations included strength of the promoter and ribosome binding site, plasmid copy number, and chaperone proteins. We constructed 24 strains using all combinations of the genetic variables. We used a theophylline riboswitch and a tetracycline resistance gene to link theophylline production to fitness. After subjecting the mixed population to selection, we measured a change in the distribution of genotypes in the population and an increased conversion of caffeine to theophylline among the most fit strains, demonstrating Programmed Evolution. Programmed Evolution inverts the standard paradigm in metabolic engineering by harnessing evolution instead of fighting it. Our modular system enables researchers to program bacteria and use evolution to determine the combination of genetic control elements that optimizes catabolic or anabolic output and to maintain it in a population of cells. Programmed Evolution could be used for applications in energy, pharmaceuticals, chemical commodities, biomining, and bioremediation.

We are developing a process called Programmed Evolution to optimize orthogonal metabolism in E. coli. Applications of the process include the production of pharmaceuticals, biofuels, chemical commodities, and bioremediation. The system contains a gene expression device that controls production of one or more enzymes in a metabolic pathway. One Programmed Evolution cycle includes introducing genetic variation into the gene expression device using a Combinatorics Module, selecting optimal metabolism with a Fitness Module, using a Stress Response Module to mitigate damage, and measuring metabolic output with a Biosensor Module. We developed and tested Fitness Modules based on the adhE gene (for use of ethanol as an energy source) and the tetA gene (resistance to the antibiotic tetracycline). Both Fitness Modules rely on a riboswitch that is activated by binding to the product of the optimized metabolic pathway. We also incorporated the riboswitch into a Biosensor Module to measure metabolic output.