Strain design optimization using reinforcement learning

Engineered microbial cells present a sustainable alternative to fossil-based synthesis of chemicals and fuels. Cellular synthesis routes are readily assembled and introduced into microbial strains using state-of-the-art synthetic biology tools. However, the optimization of the strains required to reach industrially feasible production levels is far less efficient. It typically relies on trial-and-error leading into high uncertainty in total duration and cost. New techniques that can cope with the complexity and limited mechanistic knowledge of the cellular regulation are called for guiding the strain optimization.In this paper, we put forward a multi-agent reinforcement learning (MARL) approach that learns from experiments to tune the metabolic enzyme levels so that the production is improved. Our method is model-free and does not assume prior knowledge of the microbe’s metabolic network or its regulation. The multi-agent approach is well-suited to make use of parallel experiments such as multi-well plates commonly used for screening microbial strains.We demonstrate the method’s capabilities using the genome-scale kinetic model of Escherichia coli, k-ecoli457, as a surrogate for an in vivo cell behaviour in cultivation experiments. We investigate the method’s performance relevant for practical applicability in strain engineering i.e. the speed of convergence towards the optimum response, noise tolerance, and the statistical stability of the solutions found. We further evaluate the proposed MARL approach in improving L-tryptophan production by yeast Saccharomyces cerevisiae, using publicly available experimental data on the performance of a combinatorial strain library.Overall, our results show that multi-agent reinforcement learning is a promising approach for guiding the strain optimization beyond mechanistic knowledge, with the goal of faster and more reliably obtaining industrially attractive production levels.     

A mouse model for Li-Fraumeni-Like Syndrome with cardiac angiosarcomas associated to POT1 mutations

The shelterin protein POT1 has been found mutated in many different familial and sporadic cancers, however, no mouse models to understand the pathobiology of these mutations have been developed so far. To address the molecular mechanisms underlying the tumorigenic effects of POT1 mutant proteins in humans, we have generated a mouse model for the human POT1^R117C mutation found in Li-Fraumeni-Like families with cases of cardiac angiosarcoma by introducing this mutation in the Pot1a endogenous locus, knock-in for Pot1a^R117C. We find here that both mouse embryonic fibroblasts (MEFs) and tissues from Pot1a^+/ki mice show longer telomeres than wild-type controls. Longer telomeres in Pot1a^+/ki MEFs are dependent on telomerase activity as they are not found in double mutant Pot1a^+/ki Tert^-/- telomerase-deficient MEFs. By using complementation assays we further show that POT1a pR117C exerts dominant-negative effects at telomeres. As in human Li-Fraumeni patients, heterozygous Pot1a^+/ki mice spontaneously develop a high incidence of angiosarcomas, including cardiac angiosarcomas, and this is associated to the presence of abnormally long telomeres in endothelial cells as well as in the tumors. The Pot1a^+/R117C mouse model constitutes a useful tool to understand human cancers initiated by POT1 mutations.     

ATM requirement in gene expression responses to ionizing radiation in human lymphoblasts and fibroblasts

The heritable disorder ataxia telangiectasia (AT) is caused by mutations in the AT-mutated (ATM) gene with manifestations that include predisposition to lymphoproliferative cancers and hypersensitivity to ionizing radiation (IR). We investigated gene expression changes in response to IR in human lymphoblasts and fibroblasts from seven normal and seven AT-affected individuals. Both cell types displayed ATM-dependent gene expression changes after IR, with some responses shared and some responses varying with cell type and dose. Interestingly, after 5 Gy IR, lymphoblasts displayed ATM-independent responses not seen in the fibroblasts at this dose, which likely reflect signaling through ATM-related kinases, e.g., ATR, in the absence of ATM function.     

Growth and leaf production in the tropical palm Euterpe edulis: Light conditions versus developmental constraints

We studied developmental and environmental constraints on leaf dynamics, morphology and physiology in the monopodial tropical palm of the Atlantic Forest biome, Euterpe edulis. Plastic responses to light environments in terms of photosynthesis, leaf size, leaf life span, patterns of biomass allocation and growth were analysed. Plants were grown during 14 months in a shade house under four different growth irradiances. Plants of Euterpe edulis were able to adjust leaf demography and biomass allocation in the different light treatments. Leaf life span increased by 100 days with decreasing light levels while the rate of leaf production decreased, consistent with lower electron transport rates. At low light levels, adjustments in biomass allocation to leaf components allowed E. edulis to reduce self-shading and increase light interception. At high light plants allocated more biomass to roots, and the plants exhibited small leaf sizes when leaves were compared using an explicit ontogenetic analysis. Ontogeny constrained the maximum size that each consecutive leaf could achieve, while growth irradiance determined the rate of leaf production and other leaf traits. Consequently, there were both, developmental constraints and environmental determinants influencing leaf demography and morphology in E. edulis. The findings of this ecophysiological and demographic study are relevant to palms growing under natural conditions and help to explain the success of E. edulis in the forest understory and its absence from large gap openings. Our results not only confirm that E. edulis is a shade tolerant species, but also show that palms are able to acclimate to different growing condition as well as trees.     

Diversity of an ectomycorrhizal fungal community studied by a root tip and total soil DNA approach

Molecular methods based on soil DNA extracts are increasingly being used to study the fungal diversity of ectomycorrhizal (EM) fungal communities in soil. Contrary to EM root tip identification, the use of molecular methods enables identification of extramatrical mycelia in soil. To compare fungal diversity as determined by root tip identification and mycelial identification, six soil samples were analysed. Root tips were extracted from the six samples and after amplification, the basidiomycete diversity on the root tips was analysed by denaturing gradient gel electrophoresis (DGGE). The soil from the six samples was sieved, total soil DNA was extracted and after amplification, the basidiomycete diversity in the soil fractions was analysed by DGGE. Fourteen different bands were excised from the DGGE gel and sequenced; fungal taxon names could be assigned to eight bands. Out of a total of 14 fungal taxa detected in soil, 11 fungal taxa were found on root tips, of which seven were EM fungal taxa. To examine whether the sieving treatment would affect EM species diversity, two different sieve mesh sizes were used and in addition, the organic soil fraction was analysed separately. DGGE analysis showed no differences in banding pattern for the different soil fractions. The organic fraction gave the highest DGGE band intensities. This work demonstrates that there is a high correspondence between basidiomycete diversity detected by molecular analysis of root tips and soil samples, irrespective of the soil fraction being analysed.     

Effects of the emerald ash borer invasion on the community composition of arthropods associated with ash tree boles in Maryland,U.S.A.

相似文献   

Loss of Starch Granule Initiation Has a Deleterious Effect on the Growth of Arabidopsis Plants Due to an Accumulation of ADP-Glucose

STARCH SYNTHASE4 (SS4) is required for proper starch granule initiation in Arabidopsis (Arabidopsis thaliana), although SS3 can partially replace its function. Unlike other starch-deficient mutants, ss4 and ss3/ss4 mutants grow poorly even under long-day conditions. They have less chlorophyll and carotenoids than the wild type and lower maximal rates of photosynthesis. There is evidence of photooxidative damage of the photosynthetic apparatus in the mutants from chlorophyll a fluorescence parameters and their high levels of malondialdehyde. Metabolite profiling revealed that ss3/ss4 accumulates over 170 times more ADP-glucose (Glc) than wild-type plants. Restricting ADP-Glc synthesis, by introducing mutations in the plastidial phosphoglucomutase (pgm1) or the small subunit of ADP-Glc pyrophosphorylase (aps1), largely restored photosynthetic capacity and growth in pgm1/ss3/ss4 and aps1/ss3/ss4 triple mutants. It is proposed that the accumulation of ADP-Glc in the ss3/ss4 mutant sequesters a large part of the plastidial pools of adenine nucleotides, which limits photophosphorylation, leading to photooxidative stress, causing the chlorotic and stunted growth phenotypes of the plants.The metabolism of starch plays an essential role in the physiology of plants. Starch breakdown provides the plant with carbon skeletons and energy when the photosynthetic machinery is inactive (transitory starch) or in the processes of germination and sprouting (storage starch). Deficiencies in the accumulation of transitory starch in Arabidopsis (Arabidopsis thaliana) have been described previously, specifically in mutants affected in the plastidial phosphoglucomutase (PGM1) or the small subunit (APS1) of the ADP-Glc pyrophosphorylase (AGPase). While they are described as “starchless,” they actually contain small amounts of starch (1%–2% of the wild-type levels; Streb et al., 2009) and share similar phenotypic alterations, such as growth retardation when cultivated under a short-day photoregime and increased levels of soluble sugars during the light phase and reduced levels during the night (Caspar et al., 1985; Lin et al., 1988b; Schulze et al., 1991). Carbon partitioning is altered in these plants. As photosynthate cannot be accumulated as starch, it is diverted via hexose phosphates in the cytosol to the synthesis of Suc, which accumulates together with the hexose sugars, Glc and Fru (Caspar et al., 1985). In Arabidopsis, there are five starch synthase isoforms: one granule-bound starch synthase and four soluble starch synthases: SS1, SS2, SS3, and SS4. We have described previously an Arabidopsis mutant plant lacking SS3 and SS4 that is also severely affected in the accumulation of starch (Szydlowski et al., 2009). SS4 is involved in the initiation of the starch granule and controls the number of granules per chloroplast (Roldán et al., 2007). The elimination of SS3 in an ss4 background leads to an absence of starch in most of the chloroplasts, despite the fact that SS1 and SS2 are still present and total starch synthase activity is only reduced by 35% (Szydlowski et al., 2009). However, a very small proportion of chloroplasts of this mutant plant contain a single huge starch granule, which is also a characteristic of chloroplasts in the ss4 single mutant (D’Hulst and Mérida, 2012). Thus, like aps1 and pgm1, ss3/ss4 plants contain only small amounts of starch. However, unlike aps1 or pgm1 plants, most of the cells of this mutant have empty chloroplasts, without starch (Szydlowski et al., 2009).In this work, we have analyzed the phenotypic effects of the impaired starch accumulation of ss3/ss4 plants. We show that this mutant displays phenotypic changes that are not found in other mutants with very low levels of starch, such as aps1 or pgm1 plants. We provide evidence that extremely high levels of ADP-Glc accumulate in the ss3/ss4 plants. Using reverse genetics to block the pathway of starch synthesis upstream of the starch synthases reduced the level of ADP-Glc in ss3/ss4 plants and reverted the other phenotypic traits. This suggests that ADP-Glc accumulation is the causal factor behind the chlorotic and stunted growth phenotypes of the ss3/ss4 mutant.