Predicting Plasmid Promiscuity Based on Genomic Signature

Despite the important contribution of self-transmissible plasmids to bacterial evolution, little is understood about the range of hosts in which these plasmids have evolved. Our goal was to infer this so-called evolutionary host range. The nucleotide composition, or genomic signature, of plasmids is often similar to that of the chromosome of their current host, suggesting that plasmids acquire their hosts’ signature over time. Therefore, we examined whether the evolutionary host range of plasmids could be inferred by comparing their trinucleotide composition to that of all completely sequenced bacterial chromosomes. The diversity of candidate hosts was determined using taxonomic classification and genetic distance. The method was first tested using plasmids from six incompatibility (Inc) groups whose host ranges are generally thought to be narrow (IncF, IncH, and IncI) or broad (IncN, IncP, and IncW) and then applied to other plasmid groups. The evolutionary host range was found to be broad for IncP plasmids, narrow for IncF and IncI plasmids, and intermediate for IncH and IncN plasmids, which corresponds with their known host range. The IncW plasmids as well as several plasmids from the IncA/C, IncP, IncQ, IncU, and PromA groups have signatures that were not similar to any of the chromosomal signatures, raising the hypothesis that these plasmids have not been ameliorated in any host due to their promiscuous nature. The inferred evolutionary host range of IncA/C, IncP-9, and IncL/M plasmids requires further investigation. In this era of high-throughput sequencing, this genomic signature method is a useful tool for predicting the host range of novel mobile elements.Comparative genomics has clearly shown that bacterial evolution occurs not only through genetic changes that are vertically inherited but also by extensive horizontal gene transfer between closely and distantly related bacteria (9). Mobile genetic elements such as plasmids and phages serve as important agents of horizontal gene transfer that can exchange genetic material between chromosomes (26). Plasmids also play a critical role in rapid bacterial adaptation to local environmental changes, as best exemplified by the alarmingly rapid spread of plasmid-encoded multidrug resistance in human pathogens (44, 66). In spite of this, very little is understood about the range of bacterial hosts in which these plasmids may have resided and evolved in natural or clinical environments over time, i.e., their potential “evolutionary host range.” Understanding the evolutionary history of virulence, catabolic, and other plasmids may help us to reconstruct the plasmid transfer network among microorganisms and track the pathways of gene dissemination.A plasmid''s host range can be defined in different ways, but it is typically understood as the range of hosts in which a plasmid can replicate (replication host range, or from here on simply called “host range”). This host range is often narrower than the range of hosts to which the plasmid can transfer by conjugation (transfer host range) (32, 72) but wider than the range in which it can be stably maintained (long-term host range) (16). The host range of a plasmid is often determined by mating assays, wherein that plasmid is transferred into a set of recipient strains followed by selection for transconjugant clones that can express one of the traits encoded by the plasmid (40, 47). Ideally, the physical presence of the plasmids is then verified to confirm independent replication. Sometimes the host range is also inferred from the observed natural range of hosts in which a plasmid is found in various habitats (24, 72). The plasmid host range is known to be highly variable among plasmids, and the terms “narrow host range” and “broad host range” are used as qualitative indicators (18, 49, 62). For example, it has been generally considered that incompatibility (Inc) groups IncF, IncH, and IncI contain self-transmissible narrow-host-range plasmids, while IncN, IncP, and IncW plasmids transfer and replicate in a broad range of hosts (13, 49, 62). This oldest system of plasmid classification into Inc groups is based on the inability of plasmids from the same group to be maintained in the same host due to similarity in replication or partitioning systems (11, 53). We note that IncP plasmids are also called IncP-1 in the Pseudomonas classification system, but they are here referred to as IncP. The entire range of hosts, including ancestral forms and extant bacteria, in which a plasmid has replicated at some point during its evolutionary history is of course unknown but expected to be narrower than its replication range. Here, we designate this range the “evolutionary host range.”To understand the contributions of plasmids to horizontal gene transfer and bacterial evolution, it is not sufficient to know the hosts in which plasmids can potentially replicate and be maintained when tested in the laboratory or the field. While very valid, such experiments (13, 17, 40, 47, 56, 72) do not allow us to evaluate which plasmids have in fact spread among the widest range of hosts in the past and therefore contributed most so far to horizontal gene transfer across distantly related bacteria. We also need to gain insight into the range of hosts in which they have actually resided over evolutionary time—their evolutionary host range. This insight into the evolutionary history of plasmids will also shed light on the reservoirs of the many unwanted drug resistance and virulence plasmids (65). Previous studies have shown that the dinucleotide composition (2-mer genomic signatures) of plasmids tend to be similar to those of the chromosomes of their known host, suggesting that the plasmids acquire the host''s genomic signature (7, 67). It has previously been suggested that host-specific mutational biases homogenize the nucleotide compositions of genetic elements that are being replicated in the same host (plasmids, phages, and DNA fragments inserted in the chromosome); this phenomenon has been designated “genome amelioration” (7, 43). In addition, due to the potential DNA exchange between chromosomes and plasmids by recombination and transposition (8, 42), acquisition of large sections of chromosomal DNA by plasmids may also result in similar signatures between plasmids and their evolutionary hosts. It thus follows that a similar genomic signature between a plasmid and a host''s chromosome may indicate residence of the plasmid in that or a closely related host during its evolutionary history. Therefore, it should be possible to infer the evolutionary host range for plasmids whose genome sequences have been determined, based on the similarity in genomic signature with that of completely sequenced bacterial chromosomes.The goal of this study was to infer the evolutionary host range of various plasmids based on their genomic signatures. Specifically, we postulate (i) that known broad-host-range plasmids from Proteobacteria have evolved in a wider range of hosts than narrow-host-range plasmids and (ii) that our genomic signature approach can be used to assess the promiscuity of sequenced but uncharacterized plasmids and other mobile elements. To develop our approach, we chose self-transmissible plasmids belonging to six incompatibility groups, whose host ranges have been studied intensively and are thought to be narrow (IncF, IncH, and IncI) or broad (IncN, IncP, and IncW). To propose candidate evolutionary hosts of these plasmids, we compared the genomic signature of each plasmid with those of 817 chromosomes of prokaryotes for which complete sequences were available. Our results suggest that the evolutionary host range is broad for IncP plasmids, narrow for IncF and IncI plasmids, and intermediate for IncH and IncN plasmids. The lack of hosts with signatures similar to the IncW plasmids raises the hypothesis that they have not been ameliorated for any host due to their promiscuity. We then used the same method to infer the evolutionary host range of additional plasmid groups, such as IncA/C (also called IncP-3), IncL/M, IncP-9, IncQ (IncP-4), IncU, and PromA and plasmids Ri and Ti from Agrobacterium sp. (designated Ri/Ti). The similarities and discrepancies between our findings and previous knowledge on plasmid host range are discussed.