Optimizing the strain engineering process for industrial-scale production of bio-based molecules.

Biomanufacturing could contribute as much as ${\$}$30 trillion to the global economy by 2030. However, the success of the growing bioeconomy depends on our ability to manufacture high-performing strains in a time- and cost-effective manner. The Design-Build-Test-Learn (DBTL) framework has proven to be an effective strain engineering approach. Significant improvements have been made in genome engineering, genotyping, and phenotyping throughput over the last couple of decades that have greatly accelerated the DBTL cycles. However, to achieve a radical reduction in strain development time and cost, we need to look at the strain engineering process through a lens of optimizing the whole cycle, as opposed to simply increasing throughput at each stage. We propose an approach that integrates all 4 stages of the DBTL cycle and takes advantage of the advances in computational design, high-throughput genome engineering, and phenotyping methods, as well as machine learning tools for making predictions about strain scale-up performance. In this perspective, we discuss the challenges of industrial strain engineering, outline the best approaches to overcoming these challenges, and showcase examples of successful strain engineering projects for production of heterologous proteins, amino acids, and small molecules, as well as improving tolerance, fitness, and de-risking the scale-up of industrial strains.

Homologue engagement controls meiotic DNA break number and distribution.

Meiotic recombination promotes genetic diversification as well as pairing and segregation of homologous chromosomes, but the double-strand breaks (DSBs) that initiate recombination are dangerous lesions that can cause mutation or meiotic failure. How cells control DSBs to balance between beneficial and deleterious outcomes is not well understood. Here we test the hypothesis that DSB control involves a network of intersecting negative regulatory circuits. Using multiple complementary methods, we show that DSBs form in greater numbers in Saccharomyces cerevisiae cells lacking ZMM proteins, a suite of recombination-promoting factors traditionally regarded as acting strictly downstream of DSB formation. ZMM-dependent DSB control is genetically distinct from a pathway tying break formation to meiotic progression through the Ndt80 transcription factor. These counterintuitive findings suggest that homologous chromosomes that have successfully engaged one another stop making breaks. Genome-wide DSB maps uncover distinct responses by different subchromosomal domains to the ZMM mutation zip3 (also known as cst9), and show that Zip3 is required for the previously unexplained tendency of DSB density to vary with chromosome size. Thus, feedback tied to ZMM function contributes in unexpected ways to spatial patterning of recombination.

Meiotic crossover control by concerted action of Rad51-Dmc1 in homolog template bias and robust homeostatic regulation.

During meiosis, repair of programmed DNA double-strand breaks (DSBs) by recombination promotes pairing of homologous chromosomes and their connection by crossovers. Two DNA strand-exchange proteins, Rad51 and Dmc1, are required for meiotic recombination in many organisms. Studies in budding yeast imply that Rad51 acts to regulate Dmc1's strand exchange activity, while its own exchange activity is inhibited. However, in a dmc1 mutant, elimination of inhibitory factor, Hed1, activates Rad51's strand exchange activity and results in high levels of recombination without participation of Dmc1. Here we show that Rad51-mediated meiotic recombination is not subject to regulatory processes associated with high-fidelity chromosome segregation. These include homolog bias, a process that directs strand exchange between homologs rather than sister chromatids. Furthermore, activation of Rad51 does not effectively substitute for Dmc1's chromosome pairing activity, nor does it ensure formation of the obligate crossovers required for accurate homolog segregation. We further show that Dmc1's dominance in promoting strand exchange between homologs involves repression of Rad51's strand-exchange activity. This function of Dmc1 is independent of Hed1, but requires the meiotic kinase, Mek1. Hed1 makes a relatively minor contribution to homolog bias, but nonetheless this is important for normal morphogenesis of synaptonemal complexes and efficient crossing-over especially when DSB numbers are decreased. Super-resolution microscopy shows that Dmc1 also acts to organize discrete complexes of a Mek1 partner protein, Red1, into clusters along lateral elements of synaptonemal complexes; this activity may also contribute to homolog bias. Finally, we show that when interhomolog bias is defective, recombination is buffered by two feedback processes, one that increases the fraction of events that yields crossovers, and a second that we propose involves additional DSB formation in response to defective homolog interactions. Thus, robust crossover homeostasis is conferred by integrated regulation at initiation, strand-exchange and maturation steps of meiotic recombination.

Exploiting spore-autonomous fluorescent protein expression to quantify meiotic chromosome behaviors in Saccharomyces cerevisiae.

The budding yeast Saccharomyces cerevisiae has proven to be a rich source of information about the mechanisms and regulation of homologous recombination during meiosis. A common technique for studying this process involves microdissecting the four products (ascospores) of a single meiosis and analyzing the configuration of genetic markers in the spores that are viable. Although this type of analysis is powerful, it can be laborious and time-consuming to characterize the large numbers of meioses needed to generate statistically robust data sets. Moreover, the reliance on viable (euploid) spores has the potential to introduce selection bias, especially when analyzing mutants with elevated frequencies of meiotic chromosome missegregation. To overcome these limitations, we developed a versatile, portable set of reporter constructs that drive fluorescent protein expression specifically in only those spores that inherit the reporter. These spore-autonomous fluorescence constructs allow direct visualization of inheritance patterns in intact tetrads, eliminating the need for microdissection and permitting meiotic segregation patterns to be ascertained even in aneuploid spores. As proof of principle, we demonstrate how different arrangements of reporters can be used to quantify crossover frequency, crossover interference, gene conversion, crossover/noncrossover ratios, and chromosome missegregation.

Chloroquine susceptibility and reversibility in a Plasmodium falciparum genetic cross.

Mutations in the Plasmodium falciparum chloroquine (CQ) resistance transporter (PfCRT) are major determinants of verapamil (VP)-reversible CQ resistance (CQR). In the presence of mutant PfCRT, additional genes contribute to the wide range of CQ susceptibilities observed. It is not known if these genes influence mechanisms of chemosensitization by CQR reversal agents. Using quantitative trait locus (QTL) mapping of progeny clones from the HB3 × Dd2 cross, we show that the P. falciparum multidrug resistance gene 1 (pfmdr1) interacts with the South-East Asia-derived mutant pfcrt haplotype to modulate CQR levels. A novel chromosome 7 locus is predicted to contribute with the pfcrt and pfmdr1 loci to influence CQR levels. Chemoreversal via a wide range of chemical structures operates through a direct pfcrt-based mechanism. Direct inhibition of parasite growth by these reversal agents is influenced by pfcrt mutations and additional loci. Direct labelling of purified recombinant PfMDR1 protein with a highly specific photoaffinity CQ analogue, and lack of competition for photolabelling by VP, supports our QTL predictions. We find no evidence that pfmdr1 copy number affects CQ response in the progeny; however, inheritance patterns indicate that an allele-specific interaction between pfmdr1 and pfcrt is part of the complex genetic background of CQR.

Dual kinase-mediated regulation of PITK by CaMKII and GSK3.

Phosphatase Interactor Targeting K protein (PITK) was previously identified as a novel PP1 targeting subunit implicated in modulating the phosphorylation of the transcriptional regulator heterogeneous nuclear ribonucleoprotein K (hnRNP K) [Kwiek NC, Thacker DF, Datto MB, Megosh HB, Haystead TA. Cell Signal 18 (10) (2006) 1769.]. Through the phosphorylation of PITK at S1013 and S1017 (residues that flank or reside within a PP1C-binding motif), the binding of the PP1 catalytic subunit to PITK, and subsequently the activity of the holoenzyme, are discretely controlled. Herein, we demonstrate that PITK phosphorylation at S1013 and S1017 also dictates the subcellular localization of the holoenzyme. Whereas both wildtype-and an S1013,1017D-PITK mutant displayed a speckled nuclear localization, a constitutively dephosphorylated form of PITK (S1013,1017A-PITK) resulted in a diffuse localization throughout the cell including the cytoplasm. Additionally, through the use of unbiased proteomics techniques, we provide evidence for a dual kinase-mediated regulation of the PITK holoenzyme whereby PITK phosphorylation at S1017 is catalyzed by calcium/calmodulin-dependent kinase II-delta (CaMKIIdelta), promoting the subsequent phosphorylation of S1013 by glycogen synthase kinase-3 (GSK3) in vitro. Taken together, our findings provide further insight into the regulation of PITK, PP1, and hnRNP K by reversible phosphorylation.

PITK, a PP1 targeting subunit that modulates the phosphorylation of the transcriptional regulator hnRNP K.

Protein phosphatase-1 (PP1), through interactions with substrate targeting subunits, plays critical roles in the regulation of numerous cellular processes. Herein, we describe a newly identified regulatory subunit (PITK; Phosphatase Interactor Targeting K protein) that specifically targets the catalytic subunit of PP1 to nuclear foci to selectively bind and dephosphorylate the transcriptional regulator heterogeneous nuclear ribonucleoprotein K (hnRNP K) at a regulatory S284 site. Additionally, PITK is phosphorylated in vivo at S1013 and S1017, residues that flank or reside within the PP1C-binding motif, and this phosphorylation negatively regulates the binding of the phosphatase to PITK. A mutant variant, S1013,1017A-PITK, when expressed in intact cells, exhibited an increase in native PP1 binding and elicited a more profound dephosphorylation of hnRNPK at S284. A global analysis of transcription by Affymetrix microarray revealed that the expression of PITK resulted in the altered expression of 47 genes, including a marked induction of MEK5 (>14-fold, p<0.007). Additionally, the effects of PITK and S1013,1017A-PITK on transcription could be modulated by the co-expression of hnRNP K. Taken together, our findings provide a putative mechanism by which transcriptional activity of hnRNP K can be discretely controlled through the regulation of PP1 activity.