Antibiotic dose and nutrient availability differentially drive the evolution of antibiotic resistance and persistence.

Effective treatment of bacterial infections proves increasingly challenging due to the emergence of bacterial variants that endure antibiotic exposure. Antibiotic resistance and persistence have been identified as two major bacterial survival mechanisms, and several studies have shown a rapid and strong selection of resistance or persistence mutants under repeated drug treatment. Yet, little is known about the impact of the environmental conditions on resistance and persistence evolution and the potential interplay between both phenotypes. Based on the distinct growth and survival characteristics of resistance and persistence mutants, we hypothesized that the antibiotic dose and availability of nutrients during treatment might play a key role in the evolutionary adaptation to antibiotic stress. To test this hypothesis, we combined high-throughput experimental evolution with a mathematical model of bacterial evolution under intermittent antibiotic exposure. We show that high nutrient levels during antibiotic treatment promote selection of high-level resistance, but that resistance mainly emerges independently of persistence when the antibiotic concentration is sufficiently low. At higher doses, resistance evolution is facilitated by the preceding or concurrent selection of persistence mutants, which ensures survival of populations in harsh conditions. Collectively, our experimental data and mathematical model elucidate the evolutionary routes toward increased bacterial survival under different antibiotic treatment schedules, which is key to designing effective antibiotic therapies.

Toward greater realism in inclusive fitness models: the case of caste fate conflict in insect societies.

In the field of social evolution, inclusive fitness theory has been successful in making a wide range of qualitative predictions on expected patterns of cooperation and conflict. Nevertheless, outside of sex ratio theory, inclusive fitness models that make accurate quantitative predictions remain relatively rare. Past models dealing with caste fate conflict in insect societies, for example, successfully predicted that if female larvae can control their own caste fate, an excess should opt to selfishly develop as queens. Available models, however, were unable to accurately predict levels of queen production observed in Melipona bees-a genus of stingless bees where caste is self-determined-as empirically observed levels of queen production are approximately two times lower than the theoretically predicted ones. Here, we show that this discrepancy can be resolved by explicitly deriving the colony-level cost of queen overproduction from a dynamic model of colony growth, requiring the incorporation of parameters of colony growth and demography, such as the per-capita rate at which new brood cells are built and provisioned, the percentage of the queen's eggs that are female, costs linked with worker reproduction and worker mortality. Our revised model predicts queen overproduction to more severely impact colony productivity, resulting in an evolutionarily stable strategy that is approximately half that of the original model, and is shown to accurately predict actual levels of queen overproduction observed in different Melipona species. Altogether, this shows how inclusive fitness models can provide accurate quantitative predictions, provided that costs and benefits are modeled in sufficient detail and are measured precisely.

Predicting and improving complex beer flavor through machine learning.

The perception and appreciation of food flavor depends on many interacting chemical compounds and external factors, and therefore proves challenging to understand and predict. Here, we combine extensive chemical and sensory analyses of 250 different beers to train machine learning models that allow predicting flavor and consumer appreciation. For each beer, we measure over 200 chemical properties, perform quantitative descriptive sensory analysis with a trained tasting panel and map data from over 180,000 consumer reviews to train 10 different machine learning models. The best-performing algorithm, Gradient Boosting, yields models that significantly outperform predictions based on conventional statistics and accurately predict complex food features and consumer appreciation from chemical profiles. Model dissection allows identifying specific and unexpected compounds as drivers of beer flavor and appreciation. Adding these compounds results in variants of commercial alcoholic and non-alcoholic beers with improved consumer appreciation. Together, our study reveals how big data and machine learning uncover complex links between food chemistry, flavor and consumer perception, and lays the foundation to develop novel, tailored foods with superior flavors.

Orthogonal LoxPsym sites allow multiplexed site-specific recombination in prokaryotic and eukaryotic hosts.

Site-specific recombinases such as the Cre-LoxP system are routinely used for genome engineering in both prokaryotes and eukaryotes. Importantly, recombinases complement the CRISPR-Cas toolbox and provide the additional benefit of high-efficiency DNA editing without generating toxic DNA double-strand breaks, allowing multiple recombination events at the same time. However, only a handful of independent, orthogonal recombination systems are available, limiting their use in more complex applications that require multiple specific recombination events, such as metabolic engineering and genetic circuits. To address this shortcoming, we develop 63 symmetrical LoxP variants and test 1192 pairwise combinations to determine their cross-reactivity and specificity upon Cre activation. Ultimately, we establish a set of 16 orthogonal LoxPsym variants and demonstrate their use for multiplexed genome engineering in both prokaryotes (E. coli) and eukaryotes (S. cerevisiae and Z. mays). Together, this work yields a significant expansion of the Cre-LoxP toolbox for genome editing, metabolic engineering and other controlled recombination events, and provides insights into the Cre-LoxP recombination process.

Metabolism: How a eukaryote adapted to life without respiration.

A new study finds that Schizosaccharomyces japonicus, a eukaryote that lost the ability to respire, modified its central carbon metabolism to maintain efficient ATP production, cofactor regeneration, and amino-acid production. This remarkable metabolic flexibility opens new avenues towards applications.

Do microbes have a memory? History-dependent behavior in the adaptation to variable environments.

Microbes are constantly confronted with changes and challenges in their environment. A proper response to these environmental cues is needed for optimal cellular functioning and fitness. Interestingly, past exposure to environmental cues can accelerate or boost the response when this condition returns, even in daughter cells that have not directly encountered the initial cue. Moreover, this behavior is mostly epigenetic and often goes hand in hand with strong heterogeneity in the strength and speed of the response between isogenic cells of the same population, which might function as a bet-hedging strategy. In this review, we discuss examples of history-dependent behavior (HDB) or "memory," with a specific focus on HDB in fluctuating environments. In most examples discussed, the lag time before the response to an environmental change is used as an experimentally measurable proxy for HDB. We highlight different mechanisms already implicated in HDB, and by using HDB in fluctuating carbon conditions as a case study, we showcase how the metabolic state of a cell can be a key determining factor for HDB. Finally, we consider possible evolutionary causes and consequences of such HDB.