Massively Parallel Combinatorial Genetics to Probe and Treat Human Diseases

Combinatorial gene sets are important in coordinately regulating complex biological phenotypes and human diseases. For example, multiple biological factors are needed to differentiate stem cells into distinct cell types and to reset somatic cells into induced pluripotent stem cells. Targeting multiple pathways via combinatorial drug therapies can yield enhanced efficacy over standard monotherapies. Moreover, even though genome-wide association studies have identified multiple individual loci involved in multifactorial phenotypes, such loci can only account for a minor fraction of disease heritability on their own. Genetic interactions may be significant in explaining this missing heritability but existing strategies for systematically mapping the functions of high-order gene combinations are limited in the order of genetic complexity that can be studied and the throughput and scale that can be achieved.

To overcome these bottlenecks and enable massively parallel characterization of genetic combinations, we created a broadly applicable technology for the rapid and scalable assembly of high-order barcoded combinatorial genetic libraries that can be readily quantified in pooled screens with next-generation sequencing. We have applied this platform, Combinatorial Genetics En Masse (CombiGEM), to bacterial and human systems to discover novel combinatorial gene sets that underlie important diseases.

For example, new therapies are urgently needed to drug-resistant bacterial infections, which constitute a significant growing threat to global health. Thus, we sought to identify novel combinatorial genetic perturbations that were effective against drug-resistant bacteria. Using CombiGEM, we built ~34,000 overexpression constructs containing nearly all pairwise combinations of Escherichia coli transcription factors (TFs). By leveraging high-throughput pooled screens together with Illumina sequencing, we discovered diverse genetic combinations that modulated antibiotic-resistance phenotypes in carbapenem-resistant Enterobacteriaceae (CRE). Specifically, we found multiple TF pairs that enhanced antibiotic killing by up to 1,000,000-fold. By delivering these specific genetic combinations into target bacteria via phagemids and co-treating with antibiotics, we significantly increased the killing of highly drug-resistant E. coli harboring New Delhi metallo-betalactamase-1 (NDM-1). Furthermore, we assembled libraries of three-wise TF combinations with over four million unique members and validated that these could be quantified via high-throughput sequencing, thus demonstrating the scalability of CombiGEM. 

Furthermore, we have established CombiGEM for high-throughput combinatorial screening in human cells. For example, we constructed high-coverage barcoded libraries containing 1,521 two-wise and 51,770 three-wise combinations of 39 human microRNA (miRNA) precursors. We carried out systematic screens with these libraries to discover combinatorial miRNA effectors that sensitized drug-resistant ovarian cancer cells to chemotherapy and/or inhibited cancer cell proliferation. Highly ranked hits from these pooled screens were validated for their ability to suppress drug resistance and block cancer cell growth. Thus, this effort uncovered novel insights into miRNA networks underlying drug-resistance and cell-proliferation phenotypes in cancer. Furthermore, this work identified new miRNA combinations that may be useful as effective anti-cancer therapeutics.

In summary, our work establishes a powerful technology for massively parallel combinatorial genetics using the tools of synthetic biology. We envision that CombiGEM will be useful for the high-throughput profiling of multifactorial genetic combinations that regulate a myriad of phenotypes relevant to biomedical, biotechnology, and basic science applications.