Abstract
Animal-microbe facultative symbioses play a fundamental role in ecosystem and organismal health (1–3). Yet, due to the flexible nature of their association, the selection pressures acting on animals and their facultative symbionts remain elusive (4, 5). Here, by applying experimental evolution to a well-established model of facultative symbiosis: Drosophila melanogaster associated with Lactobacillus plantarum, one of its growth promoting symbiont (6, 7), we show that the diet, instead of the host, is a predominant driving force in the evolution of this symbiosis and identify the mechanism resulting from the bacterial adaptation to the diet, which confers host growth benefits. Our study reveals that adaptation to the diet can be the foremost step in the determination of the evolutionary course of a facultative symbiosis.
Main Text
In facultative symbioses, microbes do not persistently colonize the host; nevertheless, they confer essential benefits to their animal partners (8, 9). The flexible nature of these relationships suggests that there are reciprocal costs and benefits associated with maintaining such symbiosis (3, 9, 10). However, the ecological and evolutionary forces that drive the emergence and evolution of the benefits that facultative symbionts confer to their animal hosts remain largely elusive. To address this question, we experimentally tested microbial evolution using Drosophila melanogaster associated with one of its most abundant facultative symbionts, Lactobacillus plantarum, with whom it establishes nutritional mutualism (9, 11–14). As growth promotion during undernutrition is one of the major advantages conferred by L. plantarum to its animal host (11, 15), we asked if and how this bacterium can increase its potential to support animal growth while evolving with its host. To this end, we performed experimental evolution of NIZO2877 (LpNIZO2877), a strain of L. plantarum isolated from processed human food (16), which was previously shown to moderately promote growth both in Drosophila and mice (11, 15). We mono-associated germ-free (GF) Drosophila eggs with a fully sequenced clonal population of LpNIZO2877 on a low-nutritional diet and studied the partners for 20 Drosophila generations (i.e 313 days, which correspond to about 2000 bacterial generations; see Methods and Fig. S1-2). At each generation, we selected the first emerging pupae carrying a subpopulation of L. plantarum strains, and transferred them to a new sterile diet (Methods; Fig. S1). The adults rapidly emerged from the pupae and deposited the new embryos and their associated L. plantarum strains that subsequently colonized and propagated in the new environment. We then isolated the LpNIZO2877-evolved strains associated with the adult flies eclosed from the transferred pupae, selected a representative set of isolates and measured individually their growth promoting capacity on an independent set of GF fly larvae. After only two fly generations (i.e. after about 124 bacterial generations, Fig. 1A,B), we identified a few evolved LpNIZO2877 strains that significantly improved larval growth and accelerated pupariation compared to the ancestor strain. Specifically, the evolved strains exhibited the same effect as LpWJL, a potent L. plantarum growth promoting strain (15) (Fig. 1A,B). These results show that the evolution of LpNIZO2877 in the context of its symbiosis with Drosophila leads to the rapid improvement of L. plantarum animal growth promotion (Fig. S3).
To identify the genetic changes underlying the rapid microbial adaptation responsible for the improved growth of the host, we sequenced the genomes of 11 evolved LpNIZO2877 strains (Table S1, replicate 1) with increased host growth promoting potential across the 20 Drosophila generations. We identified a total of 11 mutations, including nine single-nucleotide polymorphisms (SNPs) and two small deletions (Fig. 1C; Table S2). In particular, in the strain isolated from the second fly generation (FlyG2.1.8), we found a single change in the genome within one of the three acetate kinase genes (ackA). Remarkably, this first mutation was subsequently fixed and strictly correlated with the improved animal growth phenotype (Fig. 1C).
To test the repeatability of this finding, we conducted an independent replicate of L. plantarum experimental evolution while in symbiosis with Drosophila. Both the phenotypic and genomic evolution of L. plantarum were again obtained: LpNIZO2877 improved its animal growth promoting potential by rapidly acquiring and fixing mutations, including variants in the ackA gene (Fig. S4, Table S2). In the first experiment, the evolved LpNIZO2877 strains with improved animal growth potential all carried a three-nucleotide deletion in the ackA gene that removed one proline residue. From the second replicate, the evolved strains carried a SNP that resulted in a premature stop-codon leading to protein truncation (Fig. S5). These independently isolated mutations likely generate an inactive ackA protein. Following the fixation of ackA variants, additional mutations appeared in both replicates of L. plantarum experimental evolution, which seem to further improve its symbiotic benefit (Fig. 1A, Fig. S4A). Nevertheless, the two evolved strains each bearing only one mutation in ackA (FlyG2.1.8 and FlyG3.1.8) already showed a statistically significant Drosophila growth improvement compared to their ancestor (Fig. 1A,B). Based on these observations, we propose that the de novo appearance of the ackA mutation is the first fundamental step in shaping the evolutionary trajectory in the LpNIZO2877/Drosophila symbiosis model.
To fully establish that ackA mutation is responsible for the evolution of LpNIZO2877/Drosophila symbiosis, we employed CRISPR-Cas9 to re-insert the deleted CCT triplet in the FlyG2.1.8 ackA locus (Methods; Fig. S6), so that we genetically revert the ackA allele in the FlyG2.1.8 isolate back to its ancestral form (17). The reverted strain (FlyG2.1.8Rev) bearing the ancestral ackA allele lost its increased capacity to promote animal growth when compared to the ancestor strain (Fig. 1D,E). These results therefore demonstrate that the ackA mutation in LpNIZO2877 is a causative change resulting in faster and increased Drosophila growth.
To investigate the complete L. plantarum population dynamics during Drosophila symbiosis evolution, we sequenced the metagenome of whole bacterial population samples across the 20 Drosophila generations of the first replicate experiment. We identified both segregating and fixed mutations and tracked their frequencies through time (Methods). We found that the ackA mutation was the first variant to appear in the population. Remarkably, the ackA variant showed a rapid selective sweep and became fixed as early as after three Drosophila generations (Fig. 2A). This observation suggests a competitive advantage of the evolved LpNIZO2877 strains bearing this variant. To test this hypothesis, we performed a competition assay between the ancestral LpNIZO2877 strain and the derived FlyG2.1.8 isolate in symbiosis with Drosophila (Methods, Fig.2B, Fig. S7). We find that the evolved strain bearing only the ackA mutation starts outcompeting the ancestor strain as early as after one day, demonstrating that the ackA mutation confers a strong competitive advantage in symbiosis with Drosophila. To test whether such advantage requires the host’s presence, we performed the same competition assay by inoculating only the bacterial strains on the Drosophila nutritional environment (i.e. the diet). Surprisingly, we observed that FlyG2.1.8 outcompeted the ancestral strain even when the Drosophila host is absent (Fig. 2C). Therefore, the competitive advantage of L. plantarum isolates bearing the ackA variant is likely independent of the animal host.
Intrigued by this result, we questioned whether the animal host has an influence on the evolution of its symbiotic bacteria. To test this, we experimentally evolved LpNIZO2877 in the same low-yeast fly diet, but without Drosophila (Methods; Fig. S8) and tested their capacity to promote fly growth throughout the course of the experimental evolution. Strikingly, in two parallel experiments, the LpNIZO2877 strains evolved in the absence of the host also increased their ability to promote Drosophila growth (Fig. 3A,B). Furthermore, genome sequencing of single evolved isolates from both experiments again revealed the acquisition of novel mutations in the ackA gene (Fig. 3C; Fig. S9). Taken together, these findings show that the genomic evolution of L. plantarum is driven by the adaptation to host nutritional environment, rather than to its host per se; the acquisition of the ackA variant is sufficient to drive the adaptive process to the nutrition, which ultimately results in the improvement of L. plantarum symbiotic effect on Drosophila.
We next investigated how L. plantarum adaptation to the nutritional environment enhances Drosophila growth. We postulated that L. plantarum adaptation to the specific nutritional environment of Drosophila would lead to the production of metabolites that are beneficial for Drosophila growth. To test this hypothesis, we analyzed the metabolome of Drosophila diets colonized with either LpNIZO2877 or the evolved FlyG2.1.8 strain that bears only the ackA variant. Among all of the metabolites differentially detected in the substrate (Table S6), we observed a significant and robust increase in the levels of N-acetyl-amino-acids in the diet processed by the evolved strain (Fig. 4A). Specifically, N-acetyl-glutamine is one of the most differentially represented compounds between the two conditions. We therefore tested whether N-acetylglutamine is sufficient to improve the animal growth promoting capacity of LpNIZO2877. Remarkably, we find that, when N-acetyl-glutamine is added in a dose-dependent manner in the diet, the ancestor strain LpNIZO2877 is able to recapitulate the beneficial effect conferred by FlyG2.1.8 on Drosophila growth (Fig. 4B). We then asked whether N-acetyl-glutamine enhances fly growth by improving LpNIZO2877 fitness. To test this, we performed a competition assay between LpNIZO2877 and FlyG2.1.8 strains in the host diet supplemented with 0.1g/L of N-acetyl-glutamine. We find that FlyG2.1.8 outcompetes the ancestor strain even in presence of N-acetyl-glutamine (Fig. S10). This result indicates that N-acetyl-glutamine does not confer a competitive advantage to LpNIZO2877 over FlyG2.1.8 while growing on the diet; nevertheless it benefits the host physiology. Taken together, these findings establish N-acetyl-amino-acids, and in particular N-acetyl-glutamine, as molecules produced by the evolved L. plantarum strains during growth on the Drosophila diet, which enhance Drosophila growth but not LpNIZO2877 fitness.
Our results uncover the nature of an adaptive process of L. plantarum while in symbiosis with its fly host. To our knowledge, this is the first direct experimental evidence showing that the host nutritional environment, and not the host per se, drives microbial adaptation and metabolic changes that alter the functional outputs of a facultative nutritional symbiosis. In our experimental context, the dietary substrate asserts the predominant selective pressure dictating the evolutionary change of facultative symbiotic bacteria and their consequent benefits to host physiology. Rapid adaptation of L. plantarum to the host nutritional environment occurred in multiple independent experimental lineages through the parallel fixations of different variants of a single gene, the acetate kinase ackA. This is a spectacular case of parallel evolution, indicating that the ackA mutation is the preferred or possibly the unique means for L. plantarumNIZO2877 to adapt to its host nutritional environment. These harsh nutritional conditions of our experimental setting affect L. plantarum physiology by delaying its growth (Fig. S2). It was shown that the expression of L. plantarum ackA (ack2 in the L. plantarum reference strain WCFS1) is down-regulated at low growth rates suggesting that silencing ackA would be required to cope with poor growth condition (18). This observation may explain the observed strong selection pressure on ackA in our experimental settings, which led to the rapid de novo emergence of different variants in the population (Fig. 2A). As a consequence, the strong competitive advantage given by these mutations led to their fixation (Fig. 2). Indeed, the ackA mutations found in the independent lineages of adaptive evolution improve the fitness of L. plantarum cells on the fly diet (Fig. S11), and leads to the accumulation of bacterial products, such as N-acetyl-glutamine, that enhance host growth. However, N-acetyl-glutamine does not per se improve bacterial fitness so it remains elusive how ackA variants confers competitive advantage to L. plantarum cells on the fly diet. Our results indicate that these mutations possibly cause a shift in the metabolism of L. plantarum by modifying the usage of cellular acetyl groups, which would confer benefits to Drosophila larvae growth. ackA participates in the reversible conversion of acetate to acetyl-phosphate; ackA variants might impede this reaction, and therefore shunt the pools of cellular acetyl groups into different metabolic routes leading to the accumulation of other acetylated compounds, such as N-acetyl-amino-acids, which, once secreted, are consumed and beneficial to the host. Our results identify ackA as the first target of selection exerted by the nutritional environment on LpNIZO2877. Due to the high genetic variability of L. plantarum species (19), we posit that such target hinges upon the genomic background of LpNIZO2877. According to their network of genetic polymorphisms, other non-beneficial isolates might mutate different genes in order to adapt to the host environment and improve their symbiotic benefit. Regardless of the specificity of selection target, our findings determine that the host nutritional environment is the first driving force of such evolution.
Understanding how evolutionary forces shape host-microbe symbiosis is essential to comprehend the mechanisms of their functional influence. Using the facultative nutritional mutualism between Drosophila and Lactobacillus plantarum as a model, our results reveal that the primary selection pressure acting on Lactobacillus plantarum originates from the nutritional substrate alone, which is strong enough to drive the rapid fixation of a de novo mutation. The resulting genetic change confers a fitness advantage to the evolved bacteria and triggers a metabolic adaptation in bacterial cells, which is quickly capitalized by Drosophila as a physiological growth advantage, and symbiosis can henceforth be perpetuated. Our results do not rule out the possibility that the animal host might exert additional selection pressure on its bacterial partners. Indeed, Drosophila is also known to directly impact the fitness of its own microbiota through the activity of innate immune effectors (20, 21) or the secretion of bacterial maintenance factors (22). Nevertheless, our findings demonstrate the utmost importance of the shared nutritional substrate in the evolution of Drosophila-L. plantarum symbiosis.
Symbiosis is an evolutionary imperative and facultative symbioses are widespread in nature. Despite their unequivocal diversity, animal-microbe symbioses share striking similarities (4) and nutrition often plays a major role in shaping the composition of symbiotic microbial communities (23–28). Our results provide the first direct experimental evidence that nutrition drives the evolution of a bacterial symbiont and, given that other animal and microbe partners have likely faced nutritional challenges over time, common evolutionary trajectories might have occurred. We therefore posit that bacterial adaptation to the diet can be the first step in the emergence and perpetuation of facultative animal-microbe symbioses. Our work provides another angle to unravel the complex adaptive processes in the context of evolving symbiosis.
Funding
This work was funded by an ERC starting grant (FP7/2007-2013-N°309704). M.E.M. was funded by the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement N°659510. The lab of F.L. is supported by the FINOVI foundation and the EMBO Young Investigator Program. The CRISPR/Cas9 work was supported through funding from the National Science Foundation (MCB-1452902 to C.L.B.).
Author Contributions
M.E.M. and F.L. designed the project; M.E.M. and H.G. conducted the experiments; M.E.M. and P.J. conducted the bioinformatics analyses; R.L., M.S. and C.L.B. designed and performed the CRISPR-Cas9 engineering experiments; S.H. and B.G. generated the sequencing data; M.E.M. and F.L. analyzed the data and wrote the paper.
Competing interests
The authors declare no competing financial interests.
Data and materials availability
All data is available in the main text or the supplementary materials.
Acknowledgements
We thank B. Prud’homme and colleagues at the Institut de Génomique Fonctionnelle de Lyon for critical reading of the manuscript. WCFS1 was a kind gift from Dr. Nikhil U. Nair. Plasmids pJP042 and pJP005 were both provided by the van-Pijkeren lab. pMSP3545 (CN#46888) and pCas9 (CN# 42876) were both obtained from Addgene. EC135 was provided to us by Tingyi Wen. We gratefully acknowledge support from the PSMN (Pôle Scientifique de Modélisation Numérique) of the ENS de Lyon for the computing resources. We thank University of Padua and Dr. Barbara Cardazzo for hosting M.E.M. during the last stages of this research.