UCSF

From disease diagnostics to precision dosing of medication, genome sequencing has the potential to revolutionize healthcare. A key challenge in translating genetic information into clinical action is understanding the phenotypic consequences of genetic variants in clinically important genes. Transporters encoded by genes in the Solute Carrier (SLC) family 22 have strong clinical relevance in rare genetic disease and pharmacogenetics. For example, loss-of-function variants in SLC22A5, encoding the carnitine transporter OCTN2, cause the rare metabolic disorder Carnitine Transporter Deficiency (CTD), and variants with functional consequences in SLC22A1, encoding the hepatic uptake transporter OCT1, contribute to interindividual differences in exposure and response for many commonly used medications. Experimental studies to uncover phenotypic consequences of coding region variants in transporters and other genes have struggled to match pace with the rate at which variants are identified by next-generation sequencing, slowing translation into clinically actionable information. The goal of this dissertation research is to experimentally and computationally address the key challenge of understanding the phenotypic effects of genetic variants in SLC22 transporters, with a primary focus on OCTN2 and OCT1.

The dissertation begins with an overview of current practices and limitations in the interpretation of variation in genes underlying inborn errors of metabolism and drug response, detailing experimental approaches to validating a variant as causative or pathogenic and summarizing advances in computational methods aiming to predict variant effect on protein function. We propose a vision for a genomic learning healthcare system (GLHS) that facilitates the translation of a patient’s genome into clinically actionable information for diagnostic and therapeutic purposes. After the overview, we present a rich set of experimental and computational approaches, which were developed and used to improve the functional prediction of genetic variants in OCTN2 for diagnosis of CTD. We functionally characterized 150 OCTN2 missense variants and found that 71% of variants had a significant effect on the uptake of carnitine. 25% of variants reduced transporter function to less than 20% of the wild-type OCTN2, a clinically meaningful threshold for CTD. We asked what was causing reduced function, and identified improper subcellular localization to be a major loss-of-function mechanism affecting 62% of variants. These data were then used in machine learning to build a protein-specific variant effect prediction model that accurately classified variants of OCTN2 as functional (>20%) or LOF (<20%) (area under the receiver operating characteristics curve 0.895±0.025). The machine learning models outperformed current models in terms of functional predictions of genetic variants in OCTN2. Limitations, however, included the number of variants experimentally tested to inform the models, which were limited by experimental methodologies. Therefore, we asked how we can increase throughput of SLC transporter variant phenotyping and transfer predictive models to other SLC22 family members. To this end, we developed a platform for deep mutational scanning (DMS) of a homolog of OCTN2 in the SLC22 family, OCT1 (SLC22A1) to increase the scale and diversify the phenotypes with which we can investigate functional genomics of transporters. We generated a landing-pad based cell system for expression of OCT1 and validated the system with multiple assays involving diverse substrates and phenotypes: uptake of the fluorescent substrate ASP+; uptake of the radiolabeled substrates MPP+ and metformin, and cytotoxicity of the OCT1 substrates, oxaliplatin and platinum analogs SM73 and SM85. We confirmed that OCT1 variants exhibit substrate-specific functional effects with variants p.R61C, p.P117L, and p.G401S. Then, we constructed a library of all 11,572 possible missense and single amino acid deletion variants to undergo functional and spatial characterization by the established DMS system. Data generated with this platform will be useful in the interpretation of OCT1 variants and clinically actionable for drug dosing purposes in precision medicine.

In summary, this dissertation research led to a top performing model for predicting the functional effects of variants in OCTN2, which may be causal for CTD. Importantly, we addressed limitations in our model and developed experimental methodologies that extended both the scale of the genetic variants under investigation and the functional phenotypes assessed. Collectively, these methods together with artificial intelligence including machine and transfer learning methods should lead to comprehensive models for accurately predicting the function of coding region variants of all genes in the SLC22 family. These studies will pave the way to a new understanding of the effects of genetic variants in SLC transporters that are causal for human disease and diverse pharmacogenetic phenotypes.