Skip to main content
PMC home page
eLife logoLink to eLife
. 2020 Jul 21;9:e56038. doi: 10.7554/eLife.56038

Emergence and diversification of a host-parasite RNA ecosystem through Darwinian evolution

Taro Furubayashi 1, Kensuke Ueda 2, Yohsuke Bansho 3, Daisuke Motooka 4, Shota Nakamura 4, Ryo Mizuuchi 5,6, Norikazu Ichihashi 2,3,5,7,✉
Editors: Detlef Weigel8, Detlef Weigel9
PMCID: PMC7378860  PMID: 32690137

Abstract

In prebiotic evolution, molecular self-replicators are considered to develop into diverse, complex living organisms. The appearance of parasitic replicators is believed inevitable in this process. However, the role of parasitic replicators in prebiotic evolution remains elusive. Here, we demonstrated experimental coevolution of RNA self-replicators (host RNAs) and emerging parasitic replicators (parasitic RNAs) using an RNA-protein replication system we developed. During a long-term replication experiment, a clonal population of the host RNA turned into an evolving host-parasite ecosystem through the continuous emergence of new types of host and parasitic RNAs produced by replication errors. The host and parasitic RNAs diversified into at least two and three different lineages, respectively, and they exhibited evolutionary arms-race dynamics. The parasitic RNA accumulated unique mutations, thus adding a new genetic variation to the whole replicator ensemble. These results provide the first experimental evidence that the coevolutionary interplay between host-parasite molecules plays a key role in generating diversity and complexity in prebiotic molecular evolution.

Research organism: None

Introduction

Host-parasite coevolution is at the center of the entire course of biological evolution (Claverie, 2006; Forterre and Prangishvili, 2009; Koonin and Dolja, 2013; Koskella and Brockhurst, 2014). Parasitic replicators, such as viruses, are the most prosperous biological entities (Bergh et al., 1989; Suttle, 2007) that offer ever-changing selection pressure and genetic reservoirs in the global biosphere. The development of the sophisticated adaptive immunity (Müller et al., 2018) that prevails in all domains of life is a hallmark of the power of host-parasite coevolution, and accumulating evidence highlights the potential key roles of parasites in the development of the basic biological architectures and functions (Claverie, 2006; Deininger et al., 2003; Elbarbary et al., 2016; Forterre, 2013; Forterre and Prangishvili, 2009; Iranzo et al., 2014; Koonin and Dolja, 2013; Koskella and Brockhurst, 2014).

Parasitic replicators have probably worked as evolutionary drivers since the prebiological era of molecular replication (Higgs and Lehman, 2015; Joyce and Szostak, 2018; Orgel, 1992; Szathmáry and Maynard Smith, 1997; Wochner et al., 2011). Even in a simplest form of replication systems, parasites inevitably appear through a functional loss of self-replicating molecules and threaten the sustainability of the replication system (Bansho et al., 2012; Koonin et al., 2017). Theoretical studies suggested that spatial structures, such as cell-like compartments, allow self-replicators (i.e. hosts) to survive by limiting the propagation of parasitic replicators (Bresch et al., 1980; Furubayashi and Ichihashi, 2018; Szathmáry and Demeter, 1987; Takeuchi and Hogeweg, 2009). Subsequent experimental studies demonstrated that the compartmentalization strategy effectively support the replication of host replicators in the presence of parasitic replicators (Bansho et al., 2016; Bansho et al., 2012; Matsumura et al., 2016; Mizuuchi and Ichihashi, 2018).

In a previous study (Ichihashi et al., 2013), we constructed an RNA replication system consisting of an artificial genomic RNA and a reconstituted translation system of Escherichia coli (Shimizu et al., 2001) encapsulated in water-in-oil droplets, to study how a simple molecular system develops through Darwinian evolution. In this system, the artificial genomic RNA (host RNA) replicates through the translation of the self-encoded replicase subunit. During replication, a deletion mutant of the host RNA (parasitic RNA), which lost the encoded replicase subunit gene, spontaneously appears and replicates by freeriding the replicase provided by the host RNA. Through serial nutrient supply and dilution, the host and parasitic RNAs in water-in-oil droplets undergo repeated error-prone replication and natural selection processes, that is Darwinian evolution.

In a subsequent study (Bansho et al., 2016), we performed a serial transfer replication experiment of the aforementioned RNA replication system to study the evolutionary process of the host and parasitic RNA replicators. We reported that the host and parasitic RNAs showed oscillating population dynamics and that the host RNA acquired a certain level of parasite-resistance in the final rounds of the replication experiment (43 rounds, 215 hr). However, we did not observe counter-adaptative evolution of the parasitic RNA to the host RNA, and the coevolutionary process of the host and parasitic RNAs remains unclear.

In this study, we reasoned that a much longer time may be necessary for coevolution of the host and parasitic RNA replicators; hence, we extended the replication experiment by an additional 77 rounds (385 hr). To understand their evolutionary dynamics during the replication experiment, we performed sequence analysis of the host and parasitic RNAs. We also conducted competitive replication assays using evolved host and parasitic RNA clones to confirm the coevolution of the host and parasitic RNAs. Moreover, we fully reanalyzed the host-parasite RNA population (up to 43 rounds) partially reported earlier (Bansho et al., 2016). In this paper, we present an analysis of 120 rounds (600 hr) of a longer term replication experiment, incorporating new data.

Results

RNA replication system

The RNA replication system used in this study consists of two classes of single-stranded RNAs (host and parasitic RNAs) and a reconstituted translation system of E. coli (Shimizu et al., 2001; Figure 1A). A distinctive feature of the host and parasitic RNAs is the capability of providing an RNA replicase (Qβ replicase). The host RNA provides the catalytic β-subunit of the replicase (via translation), which forms active replicase by associating with EF-Tu and EF-Ts subunits in the translation system, whereas the parasitic RNA lacks the intact gene. The host RNA replicates using the self-provided replicase, whereas the parasitic RNA relies on the host-provided replicase. We used a clone from round 128 in our previous study (Ichihashi et al., 2013) as the original host RNA because it replicates fast and had been characterized in detail.

Figure 1. Host and parasitic RNA replication system.

Figure 1.

Open in a new tab

(A) Replication scheme of the host and parasitic RNAs. The host RNA encodes the Qβ replicase subunit, whereas the parasitic RNA does not. Both RNAs are replicated by the translated Qβ replicase in the reconstituted translation system of E. coli. (B) Replication-dilution cycle for a long-term replication experiment. The host RNA is encapsulated in water-in-oil droplets with ~ 2 μm diameter. The parasitic RNA spontaneously appears. (1) The droplets are incubated at 37°C for 5 hr for translation and replication. (2) Eighty percent of the droplets are removed and (3) diluted with new droplets containing the translation system (i.e. five-fold dilution). (4) Diluted droplets are vigorously mixed to induce fusion and division among the droplets. We repeated this cycle for 120 rounds. The reaction volume was 1 mL, with 1% aqueous phase, corresponding to ~ 108 droplets.

In this system, parasitic RNAs spontaneously emerge from the host RNA by deleting the internal replicase gene plausibly through nonhomologous recombination (Bansho et al., 2012; Chetverin et al., 1997). The parasitic RNAs reported previously have similar sizes (~200 nt). We refer to parasitic RNA of this size as ‘parasite-α'. Parasite-α replicates much faster than the original host RNA (~2040 nt), owing to its smaller size, and thus inhibits the host replication through competition for the replicase. The replication with Qβ replicase is error-prone, approximately 1.0 × 10−5 per base (García-Villada and Drake, 2012), and mutations are randomly introduced into the host and parasitic RNAs during the replication reaction.

Long-term replication experiment

We performed a long-term replication experiment of the host and parasitic RNAs. The replication reaction was performed in a water-in-oil emulsion (~2 × 109 droplets in each round) by repeating a fusion-division cycle with the supply of new droplets containing the translation system (Figure 1B). A single round of the experiment consisted of four steps: 1) incubation, 2) partial removal, 3) dilution, and 4) mixing. In the incubation process, the water-in-oil droplets were incubated at 37°C for 5 hr to induce internal translation and RNA replication reactions. We started with a clonal population of the host RNA (1 nM, ~6 × 109 molecules) without parasite-α, which was, however, detected within two rounds. In the partial removal process, we removed 80% of the water-in-oil droplets. In the dilution process, we substituted them with new water-in-oil droplets containing the cell-free translation system (i.e. five-fold dilution). In the mixing process, droplets were vigorously mixed with a homogenizer to induce fusion and division among the droplets and allow the mixing of RNAs and other components. This replication-dilution cycle does not require manual mutagenesis, selection procedures, and control of the RNA copy number in the droplets, allowing easy implementation of long-term in vitro molecular evolution. We repeated this cycle for 120 rounds (600 hr) in total. All the following results were derived from this single long-term replication experiment.

Population dynamics of host and parasitic RNAs

We measured the concentrations of the host and parasitic RNAs after every incubation process (Figure 2A). The host RNA was measured using quantitative PCR after reverse transcription (RT-qPCR). The parasitic RNA was measured using the band intensity after polyacrylamide gel electrophoresis (Figure 2—figure supplement 1) because these parasites were deletion mutants of the host RNA and could not be uniquely targeted by RT-qPCR. In some rounds (7–12, 16–22, and 75–84), the parasitic RNAs were under the detection limit (less than ~30 nM) and not visible due to the lower sensitivity of gel analysis compared to that of RT-qPCR.

Figure 2. Coevolutionary dynamics of host and parasitic RNAs.

(A) Population dynamics of the host and parasitic RNAs during a long-term replication experiment. In the regions without points, parasitic RNA concentrations were under the detection limits (<30 nM) of the gel analysis. Three different parasitic species (α, β, and γ) are classified based on their sizes. (B) Schematic representation of the sequence alignments of the host and parasitic RNA species. The terminal regions (red) of all the RNA species are derived by the replicase MDV-1 (Mills et al., 1973), from a small replicable RNA. The β-subunit encoding regions are shown in blue, and the branched stem-loop of the M-site, one of the binding sites for Qβ replicase, is also indicated. Deleted regions are shown using black lines. (C, D, E) 2D maps of the dominant RNA genotypes for the host RNA (C), parasite-α (D), and parasite-β and parasite-γ (E). The top 90 dominant genotypes were plotted for each round. A point represents each genotype. The color depths are consistent with those in (A). Black lines connect pairs of points one Hamming distance apart in the same RNA species. A broken line connects a pair of points zero Hamming distance apart (perfect match) in the different RNA species, ignoring the large deletion between host and parasitic RNAs. Stars represent the genotypes of the evolved RNAs used for the competitive replication assay shown in Figure 4A. The original host RNA is Host-0. Round-by-round data are shown in Figure 3.

Figure 2—source data 1. Read numbers of deep sequencing.
Note that the coverage is 100% for all the reads.
Figure 2—source data 2. Sequence data file after the alignment with the original host sequence, used to identify the 74 dominant mutations.

Figure 2.

Figure 2—figure supplement 1. The native polyacrylamide gel electrophoresis of the RNA mixture during the long-term replication experiment.

Figure 2—figure supplement 1.

The numbers above the gels indicate the sampled rounds. The parasitic RNAs exhibit multiple bands due to structural heterogeneity.
Figure 2—figure supplement 2. Replication of Parasite-β99 and Parasite-γ115 without host species.

Figure 2—figure supplement 2.

The RNA replication reactions were performed with 10 nM of each RNA for 5 hr, and RNA concentration was measured by sequence-specific RT-qPCR. Error bars represent standard errors of three independent competition assays. Host-0 was also replicated for comparison.
Figure 2—figure supplement 3. Dominant mutations and fixation dynamics among host and parasitic RNAs.

Figure 2—figure supplement 3.

The base numbers are determined based on the single reference sequence (shown in Supplementary file 1), which contains all dominant insertions (129A, 296U, 1985T, and 2002T) in the original host sequence (Host-0). α, β, and γ in the ‘Appearance in different species’ are shorthand expressions for the parasite-α, -β, and -γ. The numbers above color boxes represent rounds of the long-term replication experiment. The host, parasite-α, -β, and -γ RNAs each corresponds to blue, red, green, and purple color. The intensity of colors represents fixation frequency. Gray boxes represent the rounds where parasitic RNAs were not recovered and sequenced.
Figure 2—figure supplement 4. Phylogenic analysis of the host and parasite RNAs.

Figure 2—figure supplement 4.

Top three most frequent of host and parasite sequences in all the sequenced rounds are shown. Αlpha, Βeta, and Gamma each represents parasite-α, parasite-β, and parasite-γ. The shapes and colors of markers correspond to those in Figure 2. The ancestral host and parasite-α of the earliest sequenced round (round 13) are marked with blue and red asterisks, respectively. A green tick indicates parasite-α that probably appeared by the deletion event of a host RNA in a later round. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site.
Figure 2—figure supplement 4—source data 1. Alignment data used for Figure 2—figure supplement 4.
Open in a new tab

The population dynamics of the host and parasitic RNAs gradually changed throughout the rounds (Figure 2A). In the early stage (rounds one to ~35), the host RNA and parasite-α showed a relatively regular oscillation pattern caused by competition between the host and parasitic RNAs in compartments. In this regime, the concentrations of parasite-α were higher than those of the host RNA. In the middle stage (rounds ~ 35 to~75), the concentrations of the host RNA increased, and the oscillation pattern became irregular. The elevation of the host RNA concentration can be attributed to less replication inhibition by parasite-α. We have previously reported that some nonsynonymous mutations in Qβ replicase encoded by the host RNA in round 43 selectively reduce the replication efficiency of parasite-α (Bansho et al., 2016). The prevalence of these mutations probably allows the host RNA population to maintain higher concentrations than that in the early stage. In the later stage (rounds ~ 95 to~116), the concentrations of the host and parasite-α further increased, and the oscillation pattern became more unclear. In this regime, we observed the appearance of new parasitic RNA species of different sizes and classified them as parasite-β (~1000 nt, green squares) and parasite-γ (~500 nt, purple diamonds) according to their sizes. We termed these new RNAs ‘parasites’ because each clone of these RNAs did not replicate alone (Figure 2—figure supplement 2). Such continuously changing population dynamics can be caused by successive adaptation processes between host and parasitic RNAs.

Sequence analysis

To investigate the evolutionary dynamics of host-parasite RNA populations at the sequence level, we recovered RNA mixtures from 17 points (rounds 13, 24, 33, 39, 43, 50, 53, 60, 65, 72, 86, 91, 94, 99, 104, 110, and 115), and subjected them to reverse transcription followed by deep sequencing with PacBio RS II for the host, parasite-β, and parasite-γ and MiSeq for parasite-α. With PacBio RS II sequencing, we obtained 365–4143 reads for each class of RNA in the sequenced rounds. With MiSeq sequencing, we obtained ~5000 reads for each round (Figure 2—source data 1).

Sequence analysis revealed that four major RNA classes with different sizes existed in the long-term replication experiment, consistent with the band positions observed in the polyacrylamide gels:~2040 nt (the host),~220 nt (parasite-α),~1070 nt (parasite-β), and ~510 nt (parasite-γ). The sequences of all the classes of parasitic RNA shared a high degree of similarity with those of the host RNAs but lacked a large part of the replicase subunit gene (Figure 2B). The parasite-α sequence class lacks the entire gene. The parasite-β sequence class lacks approximately the 3’ half of the gene, and parasite-γ sequence class further lacks ~25% of the remaining 5’ region of the gene. Both parasite-β and parasite-γ retain a part of the M-site sequence, one of the recognition sites for Qβ replicase (Meyer et al., 1981; Schuppli et al., 1998), in the middle of the gene.

We then determined the dominant genotypes of all the classes of RNA (host, parasite-α, parasite-β, and parasite-γ). Although the RNA replication by Qβ replicase is error-prone and introduces many random mutations that produced quasi-species for each genotype, we focused on the consensus sequences that consist of mutations commonly found in the RNA population. We first identified 74 dominant mutations that were present in more than 10% of the population of each class of RNA in a sequenced round. The dominant mutations consisted of 60 base substitutions, four insertions, and 10 deletions in total (Figure 2—figure supplement 3). Then, we measured the frequencies of all the genotypes composed of the combination of these 74 dominant mutations in every sequenced round for each class of RNA. All the genotypes and their frequencies are shown in the Supplementary file 1.

We then investigated the relationships of the detected genotypes. To visualize evolutionary trajectories, we calculated Hamming distances between all combinations of the top 90 genotypes of all the classes of RNA species in the sequenced rounds and then plotted them in a single two-dimensional (2D) map, using Principal Coordinate Analysis. RNA species-wise data are shown in Figure 2C–E, and round-wise data of all the RNA species are plotted together in Figure 3 to recapitulate the evolutionary dynamics of the entire RNA population throughout the replication experiment (animation of these snapshots is provided in Figure 3—animation 1). A point represents each genotype, and the colors of points represent the rounds they appeared consistent with the colors of the markers in Figure 2A. A black line connects a pair of genotypes one Hamming distance apart in the same RNA class. We assigned zero distance for the large deletions between the host and parasitic RNAs. The host RNA genotypes gradually became distant from the original genotype (Host-0) as the rounds proceeded (Figure 2C and Figure 3). From round 0 to round 43, sequences diversified around the original genotype. Then, until round 72, most of the genotypes moved toward the upper-right branches. However, in round 86, a certain fraction of the genotypes shifted to the left branch and dominated until round 99. In round 104, most of the genotypes moved back to the right branch again and stayed there until round 115. These frequent changes in dominant lineages imply that the fittest genotype changes frequently during the long-term replication experiment.

Figure 3. Series of snapshots of dominant RNA genotypes on 2D maps for the host RNA, parasite-α, -β and -γ.

Figure 3.

Open in a new tab

The upper-left numbers indicate the round. The top 90 dominant genotypes of each RNA species were plotted for each round. A point represents each genotype. The colors of points are consistent with Figure 2, the host (blue), parasite-α (red), -β (green), and -γ (purple). A star in each figure represents the most frequent genotype of the host (blue), parasite-α (red), -β (green), and -γ (purple). Black lines connect pairs of points one Hamming distance apart in the same RNA species. A broken line connects a pair of points zero Hamming distance apart in the different RNA species, ignoring the large deletion, which represent a plausible generation route of each parasite. Colors of the broken lines correspond to the host and parasite-α (red), the host and parasite-β (green), and the host and parasite-γ (purple). Parasite-α is not shown in the round-39, 50, and 65 because they could not have been recovered and sequenced.

The population of parasite-α represented a cluster distinct from host RNA populations (Figure 2D), and most of the genotypes were connected (i.e. one Hamming distance apart). Parasite-α did not show clear directionality throughout the long-term replication experiment (Figure 3). Interestingly, we identified 18 unique mutations specific to parasite-α (Figure 2—figure supplement 3), which were not found in the corresponding region of the host sequence. The persistence of the unique mutations of parasite-α and the differences in the mutational patterns of parasite-α and the coevolving hosts indicate that many of the new parasite-α genotypes were not newly generated from evolving host RNAs, and that parasite-α maintained its own lineage and evolved independently of the host RNA. We also observed the appearance of new parasite-α species from the evolved host RNAs owing to deletion. For example, a parasite-α genotype that appeared in round 94 perfectly matched with a host genotype in round 94 (connected with a red broken line in Figures 2D and 3), except for a large internal deletion, suggesting that this parasite was generated from the evolving host through a deletion event. Note that we could not obtain the sequence data of parasite-α in rounds 39, 50, and 65 because the cDNA could not be recovered.

The populations of parasite-β and parasite-γ formed distinct clusters (Figure 2E), and most of the genotypes were closely related within each class (connected with one Hamming distance lines). Sequences of some parasite-β and parasite-γ perfectly matched with some dominant host RNAs coexisting in the same rounds as those connected with green or purple broken lines each, suggesting that these parasitic RNAs originated from the host RNAs. Unlike parasite-α, we found only 2 and 1 unique mutations for parasite-β and parasite-γ, respectively (Figure 2—figure supplement 3).

To understand the relationship between the host and parasite lineages, we performed phylogenic analysis of the top three most frequent genotypes of the host and parasite RNAs in all the sequenced rounds (Figure 2—figure supplement 4). The phylogenic tree contains two large branches: branch P (colored in red) contains most parasite-α and branch H (colored in blue) contains all the other RNAs. This result confirmed that parasite-α evolved independently. Branch H further contains two sub-branches: branch H1 contains all parasite-β and host RNAs in rounds 60–99, and branch H2 contains all parasite-γ and host RNAs during the early (until 65–86) and later (104-115) rounds. This result support that there are two lineages in the host RNAs, corresponding to the population rounds, Host-99 and Host-115, as shown in Figure 2C, and that parasite-β and parasite-γ are their respective descendants (Figure 2E). We could not find a clear trend in transition for parasite-α (i.e. in branch P). The earliest parasite-α in round 13 (indicated with red asterisks) were already distributed into different sub-branches, and those at later rounds existed within or around the sub-branches. This result indicates that many of the mutations that characterized parasite-α appeared and were fixed by round 13, and then parasite-α wandered around in the sequence space. It is also notable that a parasite-α genotype (Alpha 094R Rank2 with a green tick) are located within host clusters, indicating that it emerged from the evolved host in a later round.

Competitive replication assay of host and parasitic RNAs

The diversification of host genotypes and the appearance of novel parasite classes can be a consequence of the coevolution between the hosts and parasites to adapt to each other. To test this possibility, we performed a series of competitive replication assays using three representative host and parasitic RNAs. We chose the most dominant host genotypes in rounds 0, 99, and 115 (Host-0, Host-99, and Host-115, respectively). For parasite-α, parasite-β, and parasite-γ, we chose the most dominant genotypes in rounds 13, 99, and 115 (Parasite-α13, Parasite-β99, and Parasite-γ115), respectively (sequences are available in Supplementary file 1). We mixed a pair of these host and parasitic RNA clones, according to their order of appearance, at an equivalent molarity, and performed competitive replication. The concentrations of replicated RNAs were measured by sequence-specific RT-qPCR (Figure 4A). RT-qPCR of parasites was possible in this experiment because we designed primers very specific to each parasite clone, which was not possible for the evolving RNA mixture containing various mutations. In the first pair (Host-0 vs Parasite-α13), Host-0 hardly replicated (less than 2-fold) and Parasite-α13 predominantly replicated (~200 fold), indicating that Parasite-α13 severely inhibits the original host replication, whereas in the second pair (Host-99 vs Parasite-α13), Host-99 efficiently replicated (~700 fold) with negligible replication of Parasite-α13, indicating that Host-99 acquired resistance to Parasite-α13. In the third pair (Host-99 vs Parasite-β99), Host-99 replicated efficiently (~1000 fold), but Parasite-β99 also replicated up to ~20 fold, indicating that Parasite-β99 acquired the ability to co-replicate with Host-99. In the fourth pair (Host-115 vs Parasite-β99), Host-115 repressed the replication of Parasite-β99 to less than twofold, indicating that Host-115 acquired the ability to evade co-replication of Parasite-β99. In the final pair (Host-115 vs Parasite-γ115), Parasite-γ115 acquired the ability to replicate up to ~20 fold with Host-115. These results demonstrated that successive counter-adaptive evolution (i.e. evolutionary arms races) occurred among the host and parasitic RNAs, as schematically illustrated in Figure 4B. We also examined the Host-99 vs Parasite-γ115 relationship and found that Parasite-γ115 was hardly replicated by Host-99 (Figure 4—figure supplement 1), indicating that parasite-β and parasite-γ are specifically parasitic to Host-99 and Host-115, respectively.

Figure 4. Evolutionary arms races between host and parasitic RNAs.

(A) Competitive replication assays of each pair of the evolved host and parasitic RNA clones. The RNA replication reactions were performed with 10 nM of the host and parasitic RNAs for 3 hr, and each concentration was measured by sequence-specific RT-qPCR. Error bars represent standard errors of three independent competition assays. (B) Schematic representation of the host-parasite relationships among the RNA clones.

Figure 4—source data 1. Dominant mutations in Host-99 and Host-115.
Host-99 and Host-115, which exhibited distinct parasite resistance (Figure 4), have very different mutation sets with only three redundant dominant mutations each other. Mutation indexes correspond to those in Figure 2—figure supplement 3.

Figure 4.

Figure 4—figure supplement 1. Combinatorial competitive replication assay of Hosts-99 and −115 with Parasites-β99 and -γ115.

Figure 4—figure supplement 1.

The RNA replication reactions were performed with 10 nM of the host and parasitic RNAs for 3 hr and each concentration was measured by sequence-specific RT-qPCR. The three results (Host-99 vs parasite-β99, Host-115 vs parasite-β99, and Host-115 vs parasite-γ115) are the same as those in Figure 4. Error bars represent standard errors of three independent competition assays.
Open in a new tab

Discussion

Coevolution of host and parasitic replicators is a major driver in the evolution of life. In this study, we investigated the Darwinian evolution process of an RNA replication system and demonstrated the emergence of a host-parasite ecosystem in which new types of host and parasitic RNAs appeared successively and exhibited antagonistic coevolutionary dynamics. Notably, all the host and parasite RNAs that appeared in the long-term replication experiment are descendants of the single host RNA. Throughout the replication experiment, the host RNA continued to evolve and diverge into distinct evolutionary branches in a sequence space (Figures 2C and 3, and Figure 2—figure supplement 4), which stands in sharp contrast to the previously reported unidirectional and rapidly slowing evolution of the host RNA in the absence of frequent interactions with parasitic RNAs (Ichihashi et al., 2013). The diversification of parasitic RNAs into three distinct parasite classes is also a new phenomenon that was not observed in our previous study (Bansho et al., 2016). The dynamic change of the host-parasite genotypes (Figures 2C–E and 3) and phenotypes (Figure 4 and Figure 4—figure supplement 1) indicates that evolving parasites could have driven the diversification of the host RNA by providing varying selection pressure. In fact, the diverged host RNAs (Host-99 and Host-115) had very different mutational patterns, with only a few shared mutations (Figure 4—source data 1), supporting the possibility that parasites with different phenotypes promoted the evolution of different strategies of host RNAs, as discussed below. This coevolution-driven diversification is consistent with the consequence of natural host-parasite coevolution (Bohannan and Lenski, 2000; Buckling and Rainey, 2002; Ebert, 2008; Thompson, 1999; Woolhouse et al., 2002) and simulated in silico evolution (Takeuchi and Hogeweg, 2008; Zaman et al., 2011). Therefore, evolutionary arms races between host-parasite molecules may have been an important mechanism to generate and maintain diversity in molecular ecosystems before the origin of life.

High-resolution sequence tracking of RNA populations allowed the observation of reciprocal host-parasite mutational dynamics underlying evolutionary arms-race history. In the first sequenced round (round 13), parasite-α had already accumulated as many as seven mutations, whereas the host did not have fixed beneficial mutations (Figure 2—figure supplement 3). Six out of these seven mutations are unique mutations of parasite-α and five of them continued to exist until the final round (round 115), implying that parasite-α, which appeared in the early stage of evolution, maintained continuous lineage and persisted throughout the long-term replication experiment. A clear sign of adaptive evolution of the host first appeared in round 39, consistent with the elevation of host RNA concentration (Figure 2A). Nonsynonymous mutations (Lys208Asp, Leu448Arg, and Gln459Arg) in Qβ replicase that occurred in this round were found responsible for the resistance against parasite-α in our previous study (Bansho et al., 2016). Interestingly, between rounds 50 and 72, many mutations appeared and disappeared in both the host and parasitic RNAs (Figure 2—figure supplement 3), and both genotypes wandered around the sequence space (Figure 3), suggesting that a co-evolutionary event had occurred. For example, the parasite-α RNA concentration suddenly recovered in round 53 (Figure 2A), and quick accumulation of the C1986U mutation might have been beneficial. Thereafter, the host accumulated two non-synonymous mutations (A452G and A626G), and the C1986U mutation disappeared from parasite-α in round 65. From round 86, the host accumulated as many as 11 mutations simultaneously, and the population rapidly converged toward the left branch, including Host-99, in the sequence space (Figures 2C and 3). In round 104, coincident with the rise of parasite-β, all the 11 mutations that characterizes the left-branch hosts almost disappeared from the population. Instead, eight new mutations accumulated in the host, and the population quickly moved toward the right branch, including Host-115. The genotype of parasite-α also drastically changed in round 104, suggesting its adaptive evolution to the hosts in the right branch. Upon the invasion of parasite-γ in round 115, some mutations (e.g. C72U, C259U, U501C, and A851G) appeared and disappeared in the host population. These host-parasite mutational dynamics exhibit how coevolution progressed during the replication experiment. Finally, we mention that we searched for possible recombination events in the host and parasite sequences throughout the replication experiment, using the RDP4 program (Martin et al., 2015), but did not detect a recombination signal.

The mutational patterns of the host and parasitic RNAs in this study suggest an interesting possibility that parasites could bring about new information in a molecular population. Parasite-α accumulated nine dominant mutations in the 3’-UTR, whereas the host RNA never accumulated dominant mutations during long-term evolution in the region (Figure 2—figure supplement 3). This result suggests that mutations in the 3’-UTR of the host RNA are severely limited (constraint imposed by translation efficiency). Evolving and persistent parasitic molecules with less mutational constraints may add genetic novelties to the whole molecular ensemble and play a role in the evolution of complexity (Adami et al., 2000).

It is generally believed that evolution progresses toward more complexity in nature (Petrov, 2001; Sharov, 2006); however, genome reduction is also a popular mode of evolution (Albalat and Cañestro, 2016; Morris et al., 2012; Wolf and Koonin, 2013). Therefore, the condition in which genomic information expands and reduces is a fundamental question. Especially in the prebiotic molecular evolution context, the benefit of genome reduction is obvious because shorter molecules can replicate faster. In fact, in previous in vitro Darwinian evolution experiments (Ichihashi et al., 2013; Mills et al., 1967), evolution favored shorter genomes for faster replication; selection for longer genomes has not been reported. A remarkable phenomenon in our study is that longer parasites with a long RNA genome appeared after long-term evolution (after 94 rounds). The new parasites, parasite-β and parasite-γ, became longer because they retained a part of the M-site sequence, a recognition site for Qβ replicase (Meyer et al., 1981; Schuppli et al., 1998), which did not exist in parasite-α. A plausible scenario for the appearance of these parasites is as follows: (1) parasite-α first invaded the system, taking advantage of its short genome for faster replication; (2) the host RNA evolved the specificity of Qβ replicase to host-specific sequences (including the M-site) to circumvent parasite-α; and (3) the new parasites invaded the system because they retained evolved M-sites that were recognized by evolved Qβ replicases when they appeared from the evolved hosts. According to this scenario, the new parasites appeared to be expanding the genomic information to cope with the evolved strategy of the host RNA, which may be consistent with recent theoretical studies that suggest that host-parasite antagonistic coevolution is an effective mechanism to increase the complexity of individuals (Seoane and Solé, 2019; Zaman et al., 2014). The next important question would be whether further long-term coevolution can lead to genome expansion of the host RNA.

A typical phenomenon in host-parasite coevolution is Red Queen dynamics (Rabajante et al., 2015; Van Valen, 1973), in which host and parasite populations oscillate due to persistent replacement of dominant hosts and parasites. The host-parasite RNA population in our replication experiment exhibited Red Queen dynamics with a remarkable feature of damping fluctuations. One possible reason for the damped oscillation is simply the elevation of the average parasite resistance against parasite-α in the evolved host RNA population, which may be partly supported by the weakened inhibition of the host replication by the parasitic RNAs in later rounds (Figure 4). Another possibility is that increased diversity (Figures 2C–E and 3) allows competition among various types of host and parasitic RNAs to average the population dynamics. A study on Daphnia and its parasite also reported damped long-term host-parasite Red Queen coevolutionary dynamics and suggested that the increased host diversity as a consequence of coevolution could decrease fluctuations in host-parasite Red Queen dynamics (Decaestecker et al., 2013). Theoretical studies also suggest that intra-species phenotypic divergence (Van der Laan and Hogeweg, 1995) and mutation rate elevation (Kaneko and Ikegami, 1992) can lead to stable host-parasite (or prey-predator) coexistence with small-amplitude oscillation. Our simple and fast-evolving host-parasite RNA replication system may offer a useful platform to investigate these tendencies of ecological and evolutionary dynamics of hosts and parasites and further pursue an exciting evolution scenario, such as the emergence of cooperation between host-parasite replicators.

Materials and methods

Long-term replication experiment

In this study, we performed an additional 77 rounds of replication using the RNA population of round 43 of a previous experiment, using the same method (Bansho et al., 2016). In this method, initially, 10 μL of the reconstituted E. coli translation system (Shimizu et al., 2001) containing 1 nM of the original host RNA, Host-0, and the round 128 clone in a previous study (Ichihashi et al., 2013), was mixed with 1 mL of a buffer-saturated oil prepared as described previously (Ichihashi et al., 2013), using a homogenizer (POLYTRON PT-1300D; KINEMATICA), at 16,000 rpm for 1 min on ice. The water-in-oil droplets were incubated at 37°C for 5 hr to induce protein translation and RNA replication reactions. For the next round of RNA replication, a fraction of the water-in-oil droplets (200 μL) was transferred and mixed with the new buffer-saturated oil (800 μL) and translation system (10 μL), using the homogenizer, at 16,000 rpm for 1 min on ice, and then incubated at 37°C for 5 hr. The average diameter of the water-in-oil droplets was ~2 μm (Bansho et al., 2016), and the number of droplets was ~2 × 109. After the incubation step in each round, RNA concentrations were measured as described below. The composition of the translation system has been described previously (Bansho et al., 2016).

Measurement of host RNA concentrations

After the incubation step, the water-in-oil droplets were diluted 10,000-fold with 1 mM EDTA (pH 8.0) and subjected to RT-qPCR (PrimeScript One Step RT-PCR Kit (TaKaRa)) with primers 1 and 2 after heating at 95°C for 5 min. These primers specifically bind to the host RNA. To draw a standard curve in RT-qPCR, dilution series of the water-in-oil droplets containing the original host RNA diluted 10,000-fold with 1 mM EDTA were used.

Measurement of parasitic RNA concentrations

To determine the concentrations of the parasitic RNAs that appeared during the long-term replication experiment (Figure 2A), polyacrylamide gel electrophoresis was performed, followed by quantification of the fluorescence intensities of the parasitic RNA bands using ImageJ. The water phases were collected from the water-in-oil droplets after the incubation step at each round, and RNAs were purified with spin columns (RNeasy, QIAGEN). The purified RNA samples and dilution series of the standard parasitic RNA (S222 RNA [Hosoda et al., 2007]) were subjected to 8% polyacrylamide gel electrophoresis with 0.1% SDS in TBE buffer (pH 8.4) containing tris(hydroxymethyl)aminomethane (100 mM), boric acid (90 mM), and EDTA (1 mM), followed by staining with SYBR Green II (Takara). The fluorescence intensities of the parasitic RNA bands were quantified, and the concentrations were determined based on the standard curve drawn with the dilution series of the standard parasitic RNA bands.

In a previous study (Bansho et al., 2016), we determined the parasitic RNA concentration from the replication kinetics using a purified Qβ replicase, and the detection limit was lower than that of this study. This method could not be employed in this study because it was unable to distinguish the different classes of parasitic RNAs that appeared.

Sequence analysis

The RNA mixtures of rounds 13, 24, 33, 39, 43, 50, 53, 60, 65, 72, 86, 91, 94, 99, 104, 110, and 115 in the long-term replication experiment were purified with spin columns (RNeasy, QIAGEN). The purified RNAs were reverse-transcribed using PrimeScript reverse transcriptase (Takara) and primer three and then PCR-amplified using primers 3 and 4. The PCR products were subjected to agarose gel electrophoresis, and the bands corresponding to the host and parasitic cDNA were separately extracted using E-gel CloneWell (Thermo Fisher Scientific). The host, parasite-β, and parasite-γ were sequenced using PacBio RS II with C4/P6 chemistry (Pacific Biosciences), and parasite-α was sequenced using MiSeq (Illumina). To reduce read errors in the PacBio RS II sequencing, we used circular consensus sequencing (CCS) reads comprising at least five and ten reads for the host and parasites, respectively, to eliminate sequence errors. The read numbers in the Supplementary file 1 indicates those of CCS. All the sequence reads were subjected to sequence alignment with a reference sequence (the original host sequence) for each molecular species (i.e. the host, parasite-α, parasite-β, and parasite-γ), using MAFFT v7.294b with the FFT-NS-2 algorithm (Katoh et al., 2002). The sequence data after alignment was provided as Figure 2—source data 2. Frequencies of mutations were calculated for each sample, and 74 dominant mutations that were present in more than 10% of the population of each class of RNA in a sequenced round were identified (Figure 2—figure supplement 3). These mutations were located in 72 sites (i.e. a few mutations were introduced in the same sites). In the subsequent analysis, we focused on only the genotypes associated with these 72 mutation sites. Focusing only on these dominant mutation sites minimizes the influence of remaining sequencing errors and non-dominant mutations in the other sites.

Mapping dominant genotypes in two-dimensional space

Among the genotypes associated with the 72 mutation sites, the top 90 most dominant genotypes were identified for each host and parasitic species in each round. Hamming distances between all the pairs of genotypes were calculated, and a square distance matrix D, whose i,j-th component dij represented the square of the Hamming distance between the i-th and j-th genotypes, was constructed. Using principal coordinate analysis on the square distance matrix D, the positions of each genotype were determined. Matrix D was transformed into a kernel matrix K = −1/2CDC, where C is the centering matrix. λk and ek ≡ (ek1, ek2, …, ekM) denote the k-th eigenvalue and the k-th normalized eigenvector, where λ1 > λ2 > … > λM and | ek |=1 for all k and M is the dimension of the kernel matrix K. The eigenvalues and eigenvectors of the kernel matrix K were calculated, and the i-th genotype was plotted in two-dimensional space with a coordinate described as follows:

(X(i),y(i))=(λ1e1i,−λ2e2i)

Phylogenic analysis of parasite RNA species by the maximum likelihood method

We extracted the top three most frequent sequences of the host, parasite-α, parasite-β, and parasite-γ from every sequenced round and conducted evolutionary analyses using MEGA X (Kumar et al., 2018). The evolutionary history was inferred using the maximum likelihood method and Tamura-Nei model (Tamura and Nei, 1993). Initial tree(s) for the heuristic search were obtained automatically by applying the Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the Tamura-Nei model and then selecting the topology with a superior log likelihood value. The gap/missing dataset treatment option was set as ‘complete deletion’.

Recombination scan of host and parasite RNAs using RDP4

We extracted the top 50 most frequent sequences of the host, parasite-α, parasite-β, and parasite-γ from every sequenced round and created a FASTA file. Using the RDP4 program (Martin et al., 2015), we performed a full exploratory recombination scan of the FASTA file with the RDP, Chimaera, GENECONV, 3Seq, and MaxChi algorithms.

Competitive replication assay of host and parasitic RNAs

Six plasmids, each containing the T7 promoter and cDNA sequences of Host-0, Host-99, Host-115, Parasite-α13, Parasite-β99, and Parasite-γ115, were constructed using the gene synthesis service of Eurofins Genomics. Each RNA was synthesized from the plasmids digested with SmaI by in vitro transcription with T7 RNA polymerase (TaKaRa), in accordance with a previous study (Yumura et al., 2017). We mixed 10 nM each of host and parasitic RNAs in the cell-free translation system and incubated them at 37°C for 3 hr. The concentrations of the host and the parasitic RNAs were measured by RT-qPCR (PrimeScript One Step RT-PCR Kit (TaKaRa)) with sequence-specific primers (Supplementary file 1).

Acknowledgements

We thank Nobuto Takeuchi, Kunihiko Kaneko, Yoshihiro Sakatani, Yannick Rondelez, and Tetsuya Yomo for the useful discussions and comments. This work was supported by JSPS KAKENHI grant numbers JP15KT0080, JP15H04407, and JP17J01023; the ‘Innovation inspired by Nature’ Research Support Program; SEKISUI CHEMICAL CO., LTD.; and the Astrobiology Center Program of the National Institutes of Natural Sciences (NINS) (Grant Number AB021005).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Norikazu Ichihashi, Email: ichihashi@bio.c.u-tokyo.ac.jp.

Detlef Weigel, Max Planck Institute for Developmental Biology, Germany.

Detlef Weigel, Max Planck Institute for Developmental Biology, Germany.

Funding Information

This paper was supported by the following grants:

  • Japan Society for the Promotion of Science JP15KT0080 to Norikazu Ichihashi.

  • Japan Society for the Promotion of Science JP15H04407 to Norikazu Ichihashi.

  • Japan Society for the Promotion of Science JP17J01023 to Taro Furubayashi.

  • Sekisui Chemical Innovations Inspired by Nature Research Support Program to Norikazu Ichihashi.

  • National Institutes of Natural Sciences Astrobiology Center Program AB021005 to Norikazu Ichihashi.

Additional information

Competing interests

No competing interests declared.

Author contributions

Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft.

Investigation.

Data curation, Investigation, Methodology.

Resources, Data curation, Methodology.

Resources, Data curation, Methodology.

Investigation, Writing - review and editing.

Conceptualization, Supervision, Funding acquisition, Visualization, Project administration, Writing - review and editing.

Additional files

Supplementary file 1. Clone sequence, primer list, the dominant genotypes and their frequencies.
elife-56038-supp1.xlsx (440KB, xlsx)
Transparent reporting form

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 2 and 4.

References

  1. Adami C, Ofria C, Collier TC. Evolution of biological complexity. PNAS. 2000;97:4463–4468. doi: 10.1073/pnas.97.9.4463. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Albalat R, Cañestro C. Evolution by gene loss. Nature Reviews Genetics. 2016;17:379–391. doi: 10.1038/nrg.2016.39. [DOI] [PubMed] [Google Scholar]
  3. Bansho Y, Ichihashi N, Kazuta Y, Matsuura T, Suzuki H, Yomo T. Importance of parasite RNA species repression for prolonged translation-coupled RNA self-replication. Chemistry & Biology. 2012;19:478–487. doi: 10.1016/j.chembiol.2012.01.019. [DOI] [PubMed] [Google Scholar]
  4. Bansho Y, Furubayashi T, Ichihashi N, Yomo T. Host-parasite oscillation dynamics and evolution in a compartmentalized RNA replication system. PNAS. 2016;113:4045–4050. doi: 10.1073/pnas.1524404113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bergh O, Børsheim KY, Bratbak G, Heldal M. High abundance of viruses found in aquatic environments. Nature. 1989;340:467–468. doi: 10.1038/340467a0. [DOI] [PubMed] [Google Scholar]
  6. Bohannan BJM, Lenski RE. Linking genetic change to community evolution: insights from studies of Bacteria and bacteriophage. Ecology Letters. 2000;3:362–377. doi: 10.1046/j.1461-0248.2000.00161.x. [DOI] [Google Scholar]
  7. Bresch C, Niesert U, Harnasch D. Hypercycles, parasites and packages. Journal of Theoretical Biology. 1980;85:399–405. doi: 10.1016/0022-5193(80)90314-8. [DOI] [PubMed] [Google Scholar]
  8. Buckling A, Rainey PB. Antagonistic coevolution between a bacterium and a bacteriophage. Proceedings of the Royal Society of London. Series B: Biological Sciences. 2002;269:931–936. doi: 10.1098/rspb.2001.1945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chetverin AB, Chetverina HV, Demidenko AA, Ugarov VI. Nonhomologous RNA recombination in a Cell-Free system: evidence for a transesterification mechanism guided by secondary structure. Cell. 1997;88:503–513. doi: 10.1016/S0092-8674(00)81890-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Claverie JM. Viruses take center stage in cellular evolution. Genome Biology. 2006;7:110. doi: 10.1186/gb-2006-7-6-110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Decaestecker E, De Gersem H, Michalakis Y, Raeymaekers JA. Damped long-term host-parasite red queen coevolutionary dynamics: a reflection of dilution effects? Ecology Letters. 2013;16:1455–1462. doi: 10.1111/ele.12186. [DOI] [PubMed] [Google Scholar]
  12. Deininger PL, Moran JV, Batzer MA, Kazazian HH. Mobile elements and mammalian genome evolution. Current Opinion in Genetics & Development. 2003;13:651–658. doi: 10.1016/j.gde.2003.10.013. [DOI] [PubMed] [Google Scholar]
  13. Ebert D. Host-parasite coevolution: insights from the Daphnia-parasite model system. Current Opinion in Microbiology. 2008;11:290–301. doi: 10.1016/j.mib.2008.05.012. [DOI] [PubMed] [Google Scholar]
  14. Elbarbary RA, Lucas BA, Maquat LE. Retrotransposons as regulators of gene expression. Science. 2016;351:aac7247. doi: 10.1126/science.aac7247. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Forterre P. Why are there so many diverse replication machineries? Journal of Molecular Biology. 2013;425:4714–4726. doi: 10.1016/j.jmb.2013.09.032. [DOI] [PubMed] [Google Scholar]
  16. Forterre P, Prangishvili D. The great billion-year war between ribosome- and capsid-encoding organisms (cells and viruses) as the major source of evolutionary novelties. Annals of the New York Academy of Sciences. 2009;1178:65–77. doi: 10.1111/j.1749-6632.2009.04993.x. [DOI] [PubMed] [Google Scholar]
  17. Furubayashi T, Ichihashi N. Sustainability of a compartmentalized Host-Parasite replicator system under periodic Washout-Mixing cycles. Life. 2018;8:3. doi: 10.3390/life8010003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. García-Villada L, Drake JW. The three faces of riboviral spontaneous mutation: spectrum, mode of genome replication, and mutation rate. PLOS Genetics. 2012;8:e1002832. doi: 10.1371/journal.pgen.1002832. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Higgs PG, Lehman N. The RNA world: molecular cooperation at the origins of life. Nature Reviews Genetics. 2015;16:7–17. doi: 10.1038/nrg3841. [DOI] [PubMed] [Google Scholar]
  20. Hosoda K, Matsuura T, Kita H, Ichihashi N, Tsukada K, Yomo T. Kinetic Analysis of the Entire RNA Amplification Process by Qβ Replicase. Journal of Biological Chemistry. 2007;282:15516–15527. doi: 10.1074/jbc.M700307200. [DOI] [PubMed] [Google Scholar]
  21. Ichihashi N, Usui K, Kazuta Y, Sunami T, Matsuura T, Yomo T. Darwinian evolution in a translation-coupled RNA replication system within a cell-like compartment. Nature Communications. 2013;4:2494. doi: 10.1038/ncomms3494. [DOI] [PubMed] [Google Scholar]
  22. Iranzo J, Lobkovsky AE, Wolf YI, Koonin EV. Virus-host arms race at the joint origin of multicellularity and programmed cell death. Cell Cycle. 2014;13:3083–3088. doi: 10.4161/15384101.2014.949496. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Joyce GF, Szostak JW. Protocells and RNA Self-Replication. Cold Spring Harbor Perspectives in Biology. 2018;10:a034801. doi: 10.1101/cshperspect.a034801. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kaneko K, Ikegami T. Homeochaos: dynamics stability of a symbiotic network with population dynamics and evolving mutation rates. Physica D: Nonlinear Phenomena. 1992;56:406–429. doi: 10.1016/0167-2789(92)90179-Q. [DOI] [Google Scholar]
  25. Katoh K, Misawa K, Kuma K, Miyata T. MAFFT: a novel method for rapid multiple sequence alignment based on fast fourier transform. Nucleic Acids Research. 2002;30:3059–3066. doi: 10.1093/nar/gkf436. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Koonin EV, Wolf YI, Katsnelson MI. Inevitability of the emergence and persistence of genetic parasites caused by evolutionary instability of parasite-free states. Biology Direct. 2017;12:31. doi: 10.1186/s13062-017-0202-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Koonin EV, Dolja VV. A virocentric perspective on the evolution of life. Current Opinion in Virology. 2013;3:546–557. doi: 10.1016/j.coviro.2013.06.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Koskella B, Brockhurst MA. Bacteria-phage coevolution as a driver of ecological and evolutionary processes in microbial communities. FEMS Microbiology Reviews. 2014;38:916–931. doi: 10.1111/1574-6976.12072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kumar S, Stecher G, Li M, Knyaz C, Tamura K. MEGA X: molecular evolutionary genetics analysis across computing platforms. Molecular Biology and Evolution. 2018;35:1547–1549. doi: 10.1093/molbev/msy096. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Martin DP, Murrell B, Golden M, Khoosal A, Muhire B. RDP4: detection and analysis of recombination patterns in virus genomes. Virus Evolution. 2015;1:vev003. doi: 10.1093/ve/vev003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Matsumura S, Kun Á, Ryckelynck M, Coldren F, Szilágyi A, Jossinet F, Rick C, Nghe P, Szathmáry E, Griffiths AD. Transient compartmentalization of RNA replicators prevents extinction due to parasites. Science. 2016;354:1293–1296. doi: 10.1126/science.aag1582. [DOI] [PubMed] [Google Scholar]
  32. Meyer F, Weber H, Weissmann C. Interactions of qβ replicase with qβ RNA. Journal of Molecular Biology. 1981;153:631–660. doi: 10.1016/0022-2836(81)90411-3. [DOI] [PubMed] [Google Scholar]
  33. Mills DR, Peterson RL, Spiegelman S. An extracellular darwinian experiment with a self-duplicating nucleic acid molecule. PNAS. 1967;58:217–224. doi: 10.1073/pnas.58.1.217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Mills DR, Kramer FR, Spiegelman S. Complete nucleotide sequence of a replicating RNA molecule. Science. 1973;180:916–927. doi: 10.1126/science.180.4089.916. [DOI] [PubMed] [Google Scholar]
  35. Mizuuchi R, Ichihashi N. Sustainable replication and coevolution of cooperative RNAs in an artificial cell-like system. Nature Ecology & Evolution. 2018;2:1654–1660. doi: 10.1038/s41559-018-0650-z. [DOI] [PubMed] [Google Scholar]
  36. Morris JJ, Lenski RE, Zinser ER. The black queen hypothesis: evolution of dependencies through adaptive gene loss. mBio. 2012;3:00036-12. doi: 10.1128/mBio.00036-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Müller V, de Boer RJ, Bonhoeffer S, Szathmáry E. An evolutionary perspective on the systems of adaptive immunity. Biological Reviews. 2018;93:505–528. doi: 10.1111/brv.12355. [DOI] [PubMed] [Google Scholar]
  38. Orgel LE. Molecular replication. Nature. 1992;358:203–209. doi: 10.1038/358203a0. [DOI] [PubMed] [Google Scholar]
  39. Petrov DA. Evolution of genome size: new approaches to an old problem. Trends in Genetics. 2001;17:23–28. doi: 10.1016/S0168-9525(00)02157-0. [DOI] [PubMed] [Google Scholar]
  40. Rabajante JF, Tubay JM, Uehara T, Morita S, Ebert D, Yoshimura J. Red queen dynamics in multi-host and multi-parasite interaction system. Scientific Reports. 2015;5:10004. doi: 10.1038/srep10004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Schuppli D, Miranda G, Qiu S, Weber H. A branched stem-loop structure in the M-site of bacteriophage qbeta RNA is important for template recognition by qbeta replicase holoenzyme. Journal of Molecular Biology. 1998;283:585–593. doi: 10.1006/jmbi.1998.2123. [DOI] [PubMed] [Google Scholar]
  42. Seoane LF, Solé R. How parasites expand the computational landscape of life. arXiv. 2019 doi: 10.1103/PhysRevE.108.044407. https://arxiv.org/abs/1910.14339 [DOI] [PubMed]
  43. Sharov AA. Genome increase as a clock for the origin and evolution of life. Biology Direct. 2006;1:17. doi: 10.1186/1745-6150-1-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Shimizu Y, Inoue A, Tomari Y, Suzuki T, Yokogawa T, Nishikawa K, Ueda T. Cell-free translation reconstituted with purified components. Nature Biotechnology. 2001;19:751–755. doi: 10.1038/90802. [DOI] [PubMed] [Google Scholar]
  45. Suttle CA. Marine viruses--major players in the global ecosystem. Nature Reviews Microbiology. 2007;5:801–812. doi: 10.1038/nrmicro1750. [DOI] [PubMed] [Google Scholar]
  46. Szathmáry E, Demeter L. Group selection of early replicators and the origin of life. Journal of Theoretical Biology. 1987;128:463–486. doi: 10.1016/S0022-5193(87)80191-1. [DOI] [PubMed] [Google Scholar]
  47. Szathmáry E, Maynard Smith J. From replicators to reproducers: the first major transitions leading to life. Journal of Theoretical Biology. 1997;187:555–571. doi: 10.1006/jtbi.1996.0389. [DOI] [PubMed] [Google Scholar]
  48. Takeuchi N, Hogeweg P. Evolution of complexity in RNA-like replicator systems. Biology Direct. 2008;3:11. doi: 10.1186/1745-6150-3-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Takeuchi N, Hogeweg P. Multilevel selection in models of prebiotic evolution II: a direct comparison of compartmentalization and spatial self-organization. PLOS Computational Biology. 2009;5:e1000542. doi: 10.1371/journal.pcbi.1000542. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Tamura K, Nei M. Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Molecular Biology and Evolution. 1993;10:512–526. doi: 10.1093/oxfordjournals.molbev.a040023. [DOI] [PubMed] [Google Scholar]
  51. Thompson JN. Specific hypotheses on the geographic mosaic of coevolution. The American Naturalist. 1999;153:S1–S14. doi: 10.1086/303208. [DOI] [Google Scholar]
  52. Van der Laan JD, Hogeweg P. Predator—prey coevolution: interactions across different timescales. Proceedings of the Royal Society of London. Series B, Biological Sciences. 1995;259:35–42. doi: 10.1098/rspb.1995.0006. [DOI] [Google Scholar]
  53. Van Valen L. A new evolutionary low. Evolutionary Theory. 1973;1:1–30. [Google Scholar]
  54. Wochner A, Attwater J, Coulson A, Holliger P. Ribozyme-catalyzed transcription of an active ribozyme. Science. 2011;332:209–212. doi: 10.1126/science.1200752. [DOI] [PubMed] [Google Scholar]
  55. Wolf YI, Koonin EV. Genome reduction as the dominant mode of evolution. BioEssays. 2013;35:829–837. doi: 10.1002/bies.201300037. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Woolhouse ME, Webster JP, Domingo E, Charlesworth B, Levin BR. Biological and biomedical implications of the co-evolution of pathogens and their hosts. Nature Genetics. 2002;32:569–577. doi: 10.1038/ng1202-569. [DOI] [PubMed] [Google Scholar]
  57. Yumura M, Yamamoto N, Yokoyama K, Mori H, Yomo T, Ichihashi N. Combinatorial selection for replicable RNA by qβ replicase while maintaining encoded gene function. PLOS ONE. 2017;12:e0174130. doi: 10.1371/journal.pone.0174130. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Zaman L, Devangam S, Ofria C. Rapid host-parasite coevolution drives the production and maintenance of diversity in digital organisms. 13th Annual Genetic and Evolutionary Computation Conference, GECCO; 2011. [DOI] [Google Scholar]
  59. Zaman L, Meyer JR, Devangam S, Bryson DM, Lenski RE, Ofria C. Coevolution drives the emergence of complex traits and promotes evolvability. PLOS Biology. 2014;12:e1002023. doi: 10.1371/journal.pbio.1002023. [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision letter

Editor: Detlef Weigel1
Reviewed by: Eörs Szathmáry2, Paulien Hogeweg3, Erik Hom4

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

During evolution, the emergence of parasites or cheaters is inevitable, but in the end, co-evolution results in a stalemate between the parasite and its hosts. This exciting paper uses the simplest possible system to investigate the emergence of a parasite and subsequent co-evolution with the host: A self-replicating RNA. This work and the elegance of this particular system set the stage for obtaining mechanistic insights into how mutations increase fitness of the parasite, followed by corresponding mutations that increase host fitness, followed by mutations that increase parasite fitness, and so on in perpetuity.

Decision letter after peer review:

Thank you for submitting your article "Emergence and diversification of a host-parasite RNA ecosystem through Darwinian evolution" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Detlef Weigel as the Senior and Reviewing Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Eörs Szathmáry (Reviewer #1); Paulien Hogeweg (Reviewer #2); Erik Hom (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This is an exciting and very relevant study about molecular in vitro co-evolution of host and parasite RNA species. The generated data are interesting and provide a nice avenue for understanding coevolutionary dynamics in an experimentally tractable way. Although, as such experiments tend to do, they stop at the moment that tantalizing new dynamics starts, the authors clearly demonstrate co-evolution of the RNA coding for the replicase-β subunit (called host) and the immediately arising parasites (lacking this) in a few competition experiments. This work (and indeed, the elegance of this particular system) sets the stage for obtaining mechanistic insights into how mutations lead to compensatory fitness increases in host and parasite along Red Queen trajectories.

The agreement was that additional data analysis would enhance the work. Specific suggestions are for a deeper analysis of lineage tracking and diversification emergence (as suggested by the title), analysis of potential recombination events, clarifications of methods, and broader contextualization of the findings with ideas in the literature about host-parasite coevolution.

There are also a couple of suggestions for additional sequencing data, but we recognize that it might currently be difficult to obtain such data. If that is the case, please provide other explanations and/or adjust your claims.

Essential revisions:

1) The idea of an arms race in this system is attractive, but it raises questions about the role of homologous copy-choice recombination in the host RNA population. Under such conditions of directional selection (exerted by the parasites) recombination should be advantageous. Is there a signal of this process in the sequences of the host population?

2) It is confusing that different measurements of density for host (replicase) as well as the different parasites are used. Please explain.

3) The population dynamics first shows strong oscillation, which later dampen till t~75. Competition is only shown for parasite t-13 and host t=0 and parasite t=13 and host t=99 (after one more deep oscillation). What happens in between? How can the replicase sustain such high densities of host and parasite RNA, while that is initially not the case?

4) Is the α-parasite after t=90 offspring of the older α parasite or a newly evolved from the host by a large deletion?

5) The β and γ parasites are clearly offspring of the later hosts, each of one of the evolved 'subspecies' of the host. The β parasites appear to strongly diminish the ancestor subspecies, but not the other host subspecies. If possible, it would be good to further elucidate that by sequencing the competition experiments; also with respect to the γ parasite. If additional sequencing experiments are not possible, can you provide other explanations?

6) While more clarity in these processes could be obtained by more competition experiments between α parasites and hosts at different timepoints, but better representation of the data should help.

7) The evolutionary arms race experiments of Figure 3 are a nice demonstration that there are "alternating" fitness improvements in subsequent host/parasites when challenged with a prior parasite/host partner. However, in light of the complete sequencing record outlined in Figure 2, one would want to know and see more specifically how distinct parasite/host lineages arose over the course of the coevolution experiment, especially since this seems to be a beautiful advantage of this RNA ecosystem. For example, can one draw a lineage map for parasite lineages over time based on the specific rounds that the authors focused on for RNA sequencing (13, 24, 33, 39,.…99, 104, 110)? It would seem that clusters with common mutations could be derived at each of these snapshots. We would like to see phylogenetic analyses of sequences as a function of time (especially at particularly key time points of population transition, e.g., between rounds 99 and 115).

You highlight 3 dominant groups of parasites, α, β, and γ, but what can you say about finer-grained lineage clustering and "transitions" that occur within these 3 dominant groups (e.g., parasite-α13 and parasite-α24 seem to be different transitional forms (Figure 2A)-what exactly distinguishes these sub-strains at the genetic level? Instead of the mutation index tables in Supplementary Figure 4 and Figure 2—figure supplement 43, please report how strains are delineated by sets of mutations (instead of focusing on summarizing each individual mutation and which strains had them). The Hamming distance metric/analysis by itself is not very satisfying for displaying/characterizing strain clusters or distinguishing genotype centroids in Figure 2; including time/round information may help resolve lineages.

8) What specific reciprocal mutational changes in host and parasite occur over the course of the evolution experiment? Figure 3 demonstrates clear fitness benefit changes of host and parasite, but it would be good to highlight the underlying adaptive genetic changes responsible for the evolutionary arms race. This could provide fertile ground for follow-on mechanistic studies for how host or parasite fitness improves in response to the other.

9) You suggest, as alluded to in the title, that coevolution drives diversification (Discussion, first paragraph), but it was not clear what you mean by this. There was little discussion of strain or mutational diversity: what do the authors precisely mean by "diversity" as it relates to the results of the present study. I imagine one would need at least one (if not several) metrics of diversity, and apply it either to the diversity of lineages in the population and/or the distribution/spectra of mutations that are accrued in host or parasite (or ecosystem) as a function of time. The emergence of diversity/diversification is suggested in the title, but this theme does not seem to be adequately addressed in the discussion of results. In the third paragraph of the Discussion, it is pointed out that antagonistic host-parasite coevolution could increase the complexity of individuals, but this idea also does not seem to be addressed in this study. The results from this work seem to recapitulate a well-known and accepted idea that parasites undergo genome reduction, but I feel the authors need to discuss more of the implications of their findings in relation to the literature on gene loss/genome reduction (e.g., Wolf and Koonin, 2013 and Albalat and Cañestro, 2016. What new insight(s) on this topic is revealed by the authors' new results? I found it intriguing that after drastic genome reduction early on, late parasite lineages had genome expansion, which goes counter to a naïve view that the genomes of parasites "just get smaller"-perhaps the authors' results tell us something deeper about the conditions for genome reduction in parasites?

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Emergence and diversification of a host-parasite RNA ecosystem through Darwinian evolution" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Detlef Weigel as Reviewing and Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Eörs Szathmáry (Reviewer #1); Erik Hom (Reviewer #3).

As you will see, the reviewers greatly appreciated your efforts to revise the work, and based on their advice, I am happy in principle to accept the submission for publication. I would like, however, to ask you to accommodate the suggestions by reviewer #2, and consider more fully meeting the specific concern of reviewer #3, who said that while Figure 2—figure supplement 5 is appreciated, this analysis does not quite reveal how the different parasite lineages arise or are related to one another in time. Please consider adding a joint temporal-sequence correlation analysis beyond a single phylogenetic tree of lineages (Figure 2—figure supplement 5).

Reviewer #1:

The paper has been duly revised and it will be a valuable contribution to the field of experimental molecular coevolution.

Reviewer #2:

I like to congratulate the authors on their improved manuscript, which is a really nice and valuable addition to our knowledge on RNA evolution.

I just have 2 suggestions:

1) Rooting of phylogenetic tree. In this case a "real" root is known: the initial host sequence. I would suggest to reroot the tree accordingly (and indeed add the initial host)

2) In the discussion about complexity generation (Discussion) the wording suggest that the parasite evolved more complexity by adding, whereas, as is now very clear from the rest of the manuscript is arose as close mutant of the current dominant host, by deletion, but retaining mor of the host genome. Please make that clear also in this discussion, also emphasizing its co-occurence with the α parasite lineage.

Finally, I missed what the mutation index means in the mutation table, which, by the way is very nicely improved.

Reviewer #3:

This manuscript is much improved. I am satisfied with most of the responses and the changes that the authors' made in addressing the reviewer comments. I feel the key points of novelty and insight for the current work are much better presented, including incorporating nuances raised in responses to Essential revision comments #5 and #9, making this an exciting contribution.

The response to Essential revision comment #7, however, remains somewhat wanting in my opinion. The authors did improve their argument for "reciprocal host-parasite mutational dynamics underlying armed-race history" through refinements in text and Figure 2—figure supplement 2, and the inclusion of the phylogenetic tree for top 3 host and parasite lineages in Figure 2—figure supplement 5. However, it still feels like there is a missed opportunity to really describe the co-evolutionary dynamics of this system in (1) time, and (2) that the takes into account the specific sequence relationships between the different host and parasite quasi-species. The uniqueness of this RNA system and the described study is that (nearly) all the lineages of host and parasite have been sequenced over the course of coevolution. Thus, it would seem like a lineage map (in time) could be constructed (even if there were only snapshots of specific rounds from the experiment); but it doesn't seem like the sequence data time series is being mined for the fullest insights. While Figure 2—figure supplement 5 is an improvement in showing the phylogenetic relatedness of parasite vs. host lineages, this analysis doesn't really reveal how the different parasite lineages arise or are related to one another in time (E.g., how does parasite-β or parasite-γ actually arise in response to the sequences (both host and parasite) that exist leading up to their emergence in the population (e.g., in round 104)?) Although I've not spent enough time thinking about how best to do such an analysis, I imagine it might involve some joint temporal-sequence correlation analysis beyond a single phylogenetic tree of lineages (Figure 2—figure supplement 5). The color-coding of Figure 2—figure supplement 2 is definitely an improvement from the previous version of the manuscript and does capture some of this temporal-sequence information-but I still find the figure hard to parse towards understanding how mutations in different parasite size classes (α/β/γ) relate to one another and hosts over the course of coevolution. The second paragraph of the Discussion does a good job explaining an example of apparent reciprocal sequence changes between parasite-α and host, but one cannot distil this from the main figures (Figure 2A shows oscillations in concentration, but this is unrelated to reciprocal mutation changes) and it is quite cumbersome to distil this from Figure 2—figure supplement 2. Figure 2—figure supplement 2 is a sensible mapping of mutations to gene body sequence position but this is only one view of the data: e.g., grouping this data according to parasite class (α/β/γ) (or even similarity of persistence profile over time) might be alternatively illuminating to get at or synthesize what I feel is missing in the analysis presented (as mentioned above). The Hamming distance plots as a function of time in Figure 2—figure supplement 4 get at this, but might be better presented as an animation rather than a frame series; it also only focuses on a subset of host and parasite-α sequences. I recommend the authors work to reduce salient points from Figure 2—figure supplement 2 into an additional main figure that highlights the reciprocal sequence changes that occur and supports the claims of reciprocal coevolution broadly, not focused merely on parasite-α but showing the relevant coevolutionary dynamic of all parasite size classes.

The other main suggested edit is to include a discussion in the Sequence Analysis section in either the Materials and methods or Results for how sequence errors were handled in relation to singleton reads. In the Supplementary file 1 (which is very helpful), many reads are listed with a genotype frequency of 1. In the Materials and methods section (subsection “Sequence analysis”), it is stated that circular consensus sequencing for at least 5-10 reads was performed. It is not clear how these two things relate, nor how the singleton sequences of Supplementary file 1 were incorporated into the main analyses described: were singletons included in all analyses? If so, should one really include them? What thresholds were applied to reject sequences that might be erroneous? Clarification of these points in the manuscript would be very helpful.

eLife. 2020 Jul 21;9:e56038. doi: 10.7554/eLife.56038.sa2

Author response

Essential revisions:

1) The idea of an arms race in this system is attractive, but it raises questions about the role of homologous copy-choice recombination in the host RNA population. Under such conditions of directional selection (exerted by the parasites) recombination should be advantageous. Is there a signal of this process in the sequences of the host population?

We agreed that the recombination among the host RNA population can be advantageous and very interesting. We examined this possibility by using a recombination detection program RDP4 (Martin et al., 2015). We extracted top-50 most frequent host and also parasite sequences for every sequenced round from the aligned cDNA data set (we provide this data upon the request from reviewer #2) and exerted exploratory recombination scan with all the available algorithms, but unfortunately, no recombination signal was detected. In fact, as we report in Figure 3—source data 1, two diverged host lineages (Host-99 and Host-115) share only a few dominant mutations that accumulated before the divergence and then underwent very distinct evolutionary pathways without any “jumping in” of shared mutations into each other. Therefore, while further long-term evolution might lead to beneficial recombination events in the future, we have no clear evidence that it has happened so far. We briefly mention that we did not detect a signal of recombination in the second paragraph of Discussion as follows.

“Finally, we would like to mention that we searched for possible recombination events in the host and parasite sequences throughout the replication experiment, using the RDP4 program (Martin et al., 2015), but did not detect a recombination signal.”

2) It is confusing that different measurements of density for host (replicase) as well as the different parasites are used. Please explain.

We are sorry for this confusion. In Figure 2A, we measured the host and parasitic RNAs by different methods, quantitative PCR after reverse transcription (RT-qPCR) and by quantification of band intensity after gel electrophoresis, respectively. That is because RT-qPCR was not applicable to the parasitic RNA population, the mixture of deletion mutants sharing the highly similar sequences as those of host RNAs. Therefore, we could not design a primer set for RT-qPCR that specifically detects the parasitic RNAs that appeared during evolution. In Figure 3A, on the other hand, we knew the specific sequence of parasite RNA “clones”, thus we were able to design primers including unique mutations that are not shared among different RNA species. To clarify this point, we changed manuscripts in Results section as following:

In the subsection “Population dynamics of host and parasitic RNAs”

“The parasitic RNA was measured using the band intensity after polyacrylamide gel electrophoresis (Figure 2—figure supplement 1) because these parasites were deletion mutants of the host RNA and could not be uniquely targeted by RT-qPCR. In some rounds (7-12, 16-22, and 75-84), the parasitic RNAs were under the detection limit (less than ~30 nM) and not visible due to the lower sensitivity of gel analysis compared to that of RT-qPCR.”

In the subsection “Competitive replication assay of host and parasitic RNAs”

“RT-qPCR of parasites was possible in this experiment because we designed primers very specific to each parasite clone, which was not possible for the evolving RNA mixture (Figure 2A) containing various mutations.”

3) The population dynamics first shows strong oscillation, which later dampen till t~75. Competition is only shown for parasite t-13 and host t=0 and parasite t=13 and host t=99 (after one more deep oscillation). What happens in between? How can the replicase sustain such high densities of host and parasite RNA, while that is initially not the case?

In the initial stage (t = 0 – 25), the host RNA was not able to replicate in the same compartment with the parasite-α and thus a clear oscillation pattern appeared, as reported in our previous study (Bansho et al., 2016). In the paper, we also showed that the host RNA at t=43 acquired partial resistance against parasite-α, which probably allowed the co-replication of hosts and parasites at higher concentrations exhibiting clear contrast to the initial clear oscillation. We clarified this point by adding the following statements in the Results section (in the subsection “Population dynamics of host and parasitic RNAs”).

“The elevation of the host RNA concentration can be attributed to less replication inhibition by parasite-α. […] The prevalence of these mutations probably allows the host RNA population to maintain higher concentrations than that in the early stage.”

We further discussed population dynamics between t=13~99 in the new second paragraph in the Discussion section (see also Essential revision comment #6).

“High-resolution sequence tracking of RNA populations allowed the observation of reciprocal host-parasite mutational dynamics underlying evolutionary arms-race history. […] Upon the invasion of parasite-γ in round 115, some mutations (e.g., C72U, C259U, U501C, and A851G) appeared and disappeared in the host population. These host-parasite mutational dynamics exhibit how coevolution had progressed during the replication experiment.”

4) Is the α-parasite after t=90 offspring of the older α parasite or a newly evolved from the host by a large deletion?

Based on the sequence similarity, the parasite-α after t=90 is the mixture of offsprings of the older ones at t=13 and deletion mutants of the evolved hosts in later rounds. We added the explanation of this point in the Results section (in the subsection “Sequence analysis”):

“We also observed the appearance of new parasite-α species from the evolved host RNAs owing to deletion. […] We could not obtain the sequence data of para”.

5) The β and γ parasites are clearly offspring of the later hosts, each of one of the evolved 'subspecies' of the host. The β parasites appear to strongly diminish the ancestor subspecies, but not the other host subspecies. If possible, it would be good to further elucidate that by sequencing the competition experiments; also with respect to the γ parasite. If additional sequencing experiments are not possible, can you provide other explanations?

If our understanding is correct, the reviewer is asking whether parasite-β or -γ specifically diminish each ancestor host RNA, Host-99 or Host-115, respectively. That is an interesting question and such specificity would be reasonable if each of the parasites appeared from each of the host RNA through adaptive evolution. To answer this question, competition experiments all four combinations of Hosts-99 and -115 with parasite-β and -γ. Since three of the experiments (Host-99 vs. parasite-β99, Host-115 vs. parasite-β99, and Host-115 vs. parasite-γ115) has been already performed (Figure 3), we newly conducted the remaining competition experiment (Host-99 vs. Parasite-γ115) and confirmed that Parasite-γ115 is hardly replicated by Host-99 (new Figure 3—figure supplement 1). This result indicates that the parasite-β and -γ selectively “infect” each of their ancestors, Host-99 and Host-115, respectively. We added the explanation of this result in the Results section as follows.

“We also examined the Host-99 vs. Parasite-γ115 relationship and found that Parasite-γ115 was hardly replicated by Host-99 (Figure 3—figure supplement 1), indicating that parasite-β and parasite-γ are specifically parasitic to Host-99 and Host-115, respectively.”

6) While more clarity in these processes could be obtained by more competition experiments between α parasites and hosts at different timepoints, but better representation of the data should help.

We are thankful for the useful suggestion. We agreed that the evolution of parasite-α is another interesting theme, and thus we would like to address it in the next study. Instead of the additional experiments, we added a new paragraph in the Discussion section as shown in the response to Essential revision comment #3 above to describe host-parasite evolutionary processes in detail, including the arms race between the host and the parasite-α (indicated with underlines).

7) The evolutionary arms race experiments of Figure 3 are a nice demonstration that there are "alternating" fitness improvements in subsequent host/parasites when challenged with a prior parasite/host partner. However, in light of the complete sequencing record outlined in Figure 2, one would want to know and see more specifically how distinct parasite/host lineages arose over the course of the coevolution experiment, especially since this seems to be a beautiful advantage of this RNA ecosystem. For example, can one draw a lineage map for parasite lineages over time based on the specific rounds that the authors focused on for RNA sequencing (13, 24, 33, 39,.…99, 104, 110)? It would seem that clusters with common mutations could be derived at each of these snapshots. We would like to see phylogenetic analyses of sequences as a function of time (especially at particularly key time points of population transition, e.g., between rounds 99 and 115).

You highlight 3 dominant groups of parasites, α, β, and γ, but what can you say about finer-grained lineage clustering and "transitions" that occur within these 3 dominant groups (e.g., parasite-α13 and parasite-α24 seem to be different transitional forms (Figure 2A)-what exactly distinguishes these sub-strains at the genetic level? Instead of the mutation index tables in Supplementary Figure 4 and Figure 3, please report how strains are delineated by sets of mutations (instead of focusing on summarizing each individual mutation and which strains had them). The Hamming distance metric/analysis by itself is not very satisfying for displaying/characterizing strain clusters or distinguishing genotype centroids in Figure 2; including time/round information may help resolve lineages.

Upon this comment, we newly added phylogenic analysis of the top 3 most frequent of host and parasite sequences in all the sequenced rounds (Figure 2—figure supplement 5). The explanation of this analysis is included in the Results section in the revised manuscript as follows.

“To understand the relationship between the host and parasite lineages, we performed phylogenic analysis of the top 3 most frequent genotypes of the host and parasite RNAs in all the sequenced rounds (Figure 2—figure supplement 5). […] It is also notable that some parasite-α genotypes are located within host clusters (Α 072R Rank2, Α 094R Rank2, and Α 099 Rank1 in Figure 2—figure supplement 5 with green ticks), indicating that they emerged from the evolved hosts in later rounds.”

We also improved Figure 2—figure supplement 3 with heatmaps of mutation fixation dynamics for all the RNA species to easily grasp transitions of major sequences and existing mutations.

8) What specific reciprocal mutational changes in host and parasite occur over the course of the evolution experiment? Figure 3 demonstrates clear fitness benefit changes of host and parasite, but it would be good to highlight the underlying adaptive genetic changes responsible for the evolutionary arms race. This could provide fertile ground for follow-on mechanistic studies for how host or parasite fitness improves in response to the other.

We detailed how host-parasite coevolution has progressed with reciprocal mutational changes in a newly added paragraph in Discussion (please see response to Essential revisions comment #3).

9) You suggest, as alluded to in the title, that coevolution drives diversification (Discussion, first paragraph), but it was not clear what you mean by this. There was little discussion of strain or mutational diversity: what do the authors precisely mean by "diversity" as it relates to the results of the present study. I imagine one would need at least one (if not several) metrics of diversity, and apply it either to the diversity of lineages in the population and/or the distribution/spectra of mutations that are accrued in host or parasite (or ecosystem) as a function of time. The emergence of diversity/diversification is suggested in the title, but this theme does not seem to be adequately addressed in the discussion of results.

As you and reviewer #3 point out, we agree that we should more clearly state what is diversity in the Discussion section. By the word “diversity”, we mean that there are two or three distinct lineages (or branches) in both host and parasite RNAs. To clarify this point, we changed a sentence in the Abstract and also largely modified the first paragraph of the Discussion section as shown below.

Abstract

“In prebiotic evolution, molecular self-replicators are considered to develop into diverse, complex living organisms. […] These results provide the first experimental evidence that the coevolutionary interplay between host-parasite molecules plays a key role in generating diversity and complexity in prebiotic molecular evolution.”

Discussion section

“Coevolution of host and parasitic replicators is a major driver in the evolution of life. In this study, we investigated the Darwinian evolution process of an RNA replication system and demonstrated the emergence of a host-parasite ecosystem in which new types of host and parasitic RNAs appeared successively and exhibited antagonistic coevolutionary dynamics. […] Therefore, evolutionary arms races between host-parasite molecules may have been an important mechanism to generate and maintain diversity in molecular ecosystems before the origin of life.”

In the third paragraph of the Discussion, it is pointed out that antagonistic host-parasite coevolution could increase the complexity of individuals, but this idea also does not seem to be addressed in this study. The results from this work seem to recapitulate a well-known and accepted idea that parasites undergo genome reduction, but I feel the authors need to discuss more of the implications of their findings in relation to the literature on gene loss/genome reduction (e.g., Wolf and Koonin, 2013 and Albalat and Cañestro, 2016. What new insight(s) on this topic is revealed by the authors' new results? I found it intriguing that after drastic genome reduction early on, late parasite lineages had genome expansion, which goes counter to a naïve view that the genomes of parasites "just get smaller" – perhaps the authors' results tell us something deeper about the conditions for genome reduction in parasites?

We thank the reviewer for the important suggestion. Regarding the complexity of individuals, as the reviewer pointed out, our message was that late parasites (-β and -γ) had expanded genome compared to the early parasite (-α), which may have been caused by coevolution. This genome expansion of parasites has never been observed at least in Qβ RNA-based molecular evolution studies (there are not many other artificial replication systems available yet) and also seems to be against the trend of reductive evolution observed in nature (Wolf and Koonin, 2013, Morris, Lenski and Zinser, 2012). To clarify the implication of our results in a reductive evolution context, we largely modified the fourth paragraph of the Discussion section as follows.

“It is generally believed that evolution progresses toward more complexity in nature (Petrov, 2001; Sharov, 2006); however, genome reduction is also a popular mode of evolution (Albalat and Cañestro, 2016; Morris et al., 2012; Wolf and Koonin, 2013). […] The next important question would be whether further long-term coevolution can lead to genome expansion of the host RNA.”

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

As you will see, the reviewers greatly appreciated your efforts to revise the work, and based on their advice, I am happy in principle to accept the submission for publication. I would like, however, to ask you to accommodate the suggestions by reviewer #2, and consider more fully meeting the specific concern of reviewer #3, who said that while Figure 2—figure supplement 5 is appreciated, this analysis does not quite reveal how the different parasite lineages arise or are related to one another in time. Please consider adding a joint temporal-sequence correlation analysis beyond a single phylogenetic tree of lineages (Figure 2—figure supplement 5).

We are grateful to the reviewers for their effort to read and comments on our manuscript, which improved our manuscript significantly. In accordance with the reviewers’ advice, we have changed some figures and revised our manuscript. We believe that the reviewers’ comments have now been addressed in the revision.

Reviewer #2:

I like to congratulate the authors on their improved manuscript, which is a really nice and valuable addition to our knowledge on RNA evolution.

I just have 2 suggestions:

1) Rooting of phylogenetic tree. In this case a "real" root is known: the initial host sequence. I would suggest to reroot the tree accordingly (and indeed add the initial host)

Following the reviewer’s suggestion, we added the initial host (Ancester host, indicated with the blue asterisk) in the phylogenic three (Figure 2—figure supplement 4).

2) In the discussion about complexity generation (Discussion) the wording suggest that the parasite evolved more complexity by adding, whereas, as is now very clear from the rest of the manuscript is arose as close mutant of the current dominant host, by deletion, but retaining mor of the host genome. Please make that clear also in this discussion, also emphasizing its co-occurence with the α parasite lineage.

We agreed that the previous description was misleading. We rewrote the corresponding part as follows to clarify that the new parasites appeared from the evolved host through deletion.

“The new parasites, parasite-β and parasite-γ, became longer because they retained a part of the M-site sequence, a recognition site for Qβ replicase (Meyer et al., 1981; Schuppli et al., 1998), which did not exist in parasite-α. […] According to this scenario, the new parasites appear to be expanding the genomic information to cope with the evolved strategy of the host RNA, which may be consistent with recent theoretical studies that suggest that host-parasite antagonistic coevolution is an effective mechanism to increase the complexity of individuals (Seoane and Solé, 2019; Zaman et al., 2014).”

Reviewer #3:

This manuscript is much improved. I am satisfied with most of the responses and the changes that the authors' made in addressing the reviewer comments. I feel the key points of novelty and insight for the current work are much better presented, including incorporating nuances raised in responses to Essential revision comments #5 and #9, making this an exciting contribution.

The response to Essential revision comment #7, however, remains somewhat wanting in my opinion. […] I recommend the authors work to reduce salient points from Figure 2—figure supplement 2 into an additional main figure that highlights the reciprocal sequence changes that occur and supports the claims of reciprocal coevolution broadly, not focused merely on parasite-α but showing the relevant coevolutionary dynamic of all parasite size classes.

We agreed that the reciprocal evolutionary progress of the host and the parasites was difficult to grasp from our previous figures. According to the reviewer’s recommendation, we refined Figure 2—figure supplement 4 (Round-by-round 2D mapping of top genotypes) to include parasites-β and -γ and moved to a main figure (new Figure 3). We also included an animation of the same data (new Figure 3—figure supplement 1). We hope that these figures help readers to grasp the co-evolutionary dynamics.

The other main suggested edit is to include a discussion in the Sequence Analysis section in either the Materials and methods or Results for how sequence errors were handled in relation to singleton reads. In the Supplementary file 1 (which is very helpful), many reads are listed with a genotype frequency of 1. In the Materials and methods section (subsection “Sequence analysis”), it is stated that circular consensus sequencing for at least 5-10 reads was performed. It is not clear how these two things relate, nor how the singleton sequences of Supplementary file 1 were incorporated into the main analyses described: were singletons included in all analyses? If so, should one really include them? What thresholds were applied to reject sequences that might be erroneous? Clarification of these points in the manuscript would be very helpful.

In this study, all sequences were read more than 5 times to minimize the effect of sequence error and the resultant circular consensus sequences (CCS) were used for analyses. The read number in the Supplementary file 1 is those of the CCS that are obtained from at least 5-10 reads. This point was not clearly written in the previous manuscript. Then, we rewrote the corresponding parts and added a short explanation about the method to minimize sequencing errors in the Materials and methods section.

“Sequence analysis

The RNA mixtures of rounds 13, 24, 33, 39, 43, 50, 53, 60, 65, 72, 86, 91, 94, 99, 104, 110, and 115 in the long-term replication experiment were purified with spin columns (RNeasy, QIAGEN). […] In the subsequent analysis, we focused on only the genotypes associated with these 72 mutation sites. Focusing only on these dominant mutation sites minimizes the influence of remaining sequencing errors and non-dominant mutations in the other sites.”

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Figure 2—source data 1. Read numbers of deep sequencing.

    Note that the coverage is 100% for all the reads.

    elife-56038-fig2-data1.xlsx (10.7KB, xlsx)
    Figure 2—source data 2. Sequence data file after the alignment with the original host sequence, used to identify the 74 dominant mutations.
    elife-56038-fig2-data2.zip (2.2MB, zip)
    Figure 2—figure supplement 4—source data 1. Alignment data used for Figure 2—figure supplement 4.
    elife-56038-fig2-figsupp4-data1.txt (193.8KB, txt)
    Figure 4—source data 1. Dominant mutations in Host-99 and Host-115.

    Host-99 and Host-115, which exhibited distinct parasite resistance (Figure 4), have very different mutation sets with only three redundant dominant mutations each other. Mutation indexes correspond to those in Figure 2—figure supplement 3.

    elife-56038-fig4-data1.xlsx (9.7KB, xlsx)
    Supplementary file 1. Clone sequence, primer list, the dominant genotypes and their frequencies.
    elife-56038-supp1.xlsx (440KB, xlsx)
    Transparent reporting form
    elife-56038-transrepform.docx (246.3KB, docx)

    Data Availability Statement

    All data generated or analyzed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 2 and 4.

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES