6.1 Introduction

Plasmids are circular- or linear-extrachromosomal replicons found in the microorganisms of Bacteria, Archaea, and Eukaryota [1]. Plasmids are not only vertically inherited from parent cells to daughter cells but can also be horizontally transferred between cells by conjugation and natural transformation [2]. Conjugation is one of the most effective mechanisms to spread genetic elements among bacteria, and plasmids are thus important “vehicles” for facilitating rapid evolution without mutations in the host genome and for adapting to new environments. While plasmids are important genetic tools for microbial engineering, as cloning vectors for biotechnology, some plasmids can mediate horizontal gene transfer (HGT), which spreads antibiotic resistance, virulence, and other traits among different bacteria in microbial communities. The World Health Organization (WHO) has stated that “antibiotic resistance is one of the biggest threats to global health, food security, and development today” [3].

One of the most important aspects of plasmids is “host range,” i.e., which plasmid can be hosted by which microbe. The host range of plasmids can be defined by either plasmid replication or conjugation. The so-called broad-host-range (BHR) plasmids can be hosted (replicated and transferred) by phylogenetically distant organisms, while narrow-host-range (NHR) plasmids can be hosted by closely related organisms (e.g., those belonging to the same species), although the definition of broadness or narrowness is still controversial. For example, BHR plasmid pB10 can be replicated in, and transferred to, bacteria belonging to different classes of Proteobacteria. Other examples of BHR plasmids and their host ranges are recently summarized in [4]. Various factors have been found to determine or affect the host range of a plasmid, including factors affecting the replication, maintenance, and/or conjugation of the plasmid, as well as factors affecting the host chromosome. Recently, several other factors have been reported to regulate host fitness during plasmid maintenance in the host cells. In this chapter, we focus on bacterial plasmids and features that could affect the fate of plasmids in the host candidates.

6.2 Plasmid Function

In this section, we summarize factors that determine plasmid host ranges, including plasmid genes for replication, maintenance, and conjugative transfer, and nucleoid-associated proteins (NAPs).

6.2.1 Replication and Maintenance

In plasmids, DNA replication initiates at a specific site known as the origin of vegetative replication (oriV). The well-known replication systems of circular plasmids include theta-type replication, strand displacement-type replication, and rolling-circle replication. Many theta-type replicating plasmids, whose lagging strand is synthesized discontinuously producing replication intermediates that look like the Greek letter “theta,” contain repeated DNA sequences, or iterons, to which the replication initiation protein binds [reviewed in [5]]. The protein bound to the iteron sequences opens the double-stranded DNA with host factors (DnaA or PriA) and recruits the DNA polymerase of the host cell [6, 7]. ColE1-family plasmids also use theta-type replication systems strictly controlled by an antisense RNA [8, 9]. Representative plasmids with strand displacement-type replication are incompatibility (discussed later) group Q (IncQ) plasmids, which encode a helicase RepA, a specific primase RepB, and replication initiation protein RepC [10]. Plasmids with this system can be continuously replicated, including the lagging strand [10]. This system is independent of host factors for its replication initiation, enabling the host range of the plasmid to be broad [10]. The other major plasmid replication system is the rolling-circle replication (RCR) mechanism, present in many small multi-copy plasmids [11, 12]. In any system, the replication of plasmids is dependent on several molecules in the host cell including DnaA, DNA polymerase, RNA polymerase, RNase, ribosomes, helicase, nucleotides, and ATP. Because the chromosome and plasmid(s) in the same host share the host’s replication system, plasmids with similar replication initiation systems should have similar host ranges.

Two major maintenance systems of plasmids in the host cells are partition (par) systems and toxin-antitoxin (TA) systems. The former involves actively delivering low-copy-number plasmids from parental cells to daughter cells [recently reviewed in [13]]. In brief, the par system is composed of two proteins, ATPase or GTPase and DNA-binding protein, which requires one cis site for its binding. For the TA systems encoded on the plasmid, one of the two gene products (stable toxin) can kill or stop growth of cells without the other gene product (unstable antitoxin) [14] (see Chap. 3). These systems contribute to the fate of plasmids in the host cells.

Incompatibility (Inc) is one of the classical methods of plasmid classification, based on the Inc test, to assess whether two different plasmids can be propagated stably in the same host cell line. If the two plasmids share similar replication and/or par systems, either of the plasmids will be unstable in the host cell line [15]. Inc groups have been independently classified in hosts of three different taxonomic groups; there are 27 Inc groups in the family Enterobacteriaceae, 14 Inc groups in the genus Pseudomonas, and approximately 18 Inc groups in the genus Staphylococcus (Table 6.1) [73, 74,75,77, 82, 83]. Several Inc groups of Pseudomonas are identical to those in Enterobacteriaceae, such as IncP-1 (equivalent to IncP), IncP-3 (equivalent to IncA and/or IncC) [84], IncP-4 (equivalent to IncQ) [85], and IncP-6 (equivalent to IncG/U) [86] (Table 6.1). Recently, it was reported that parMRC partitioning systems in plasmids of Clostridium perfringens determined their compatibility, even though they had almost identical replication initiation protein genes [87, 88]. The Inc test can yield biologically relevant information, but recent classification of Inc groups is usually based on their similarity to genes involved in replication and/or partition. Rozwandowicz et al. showed that IncK plasmids formed two distinct subclusters, IncK1 and IncK2, based on the presence or absence of accessory genes, and that the plasmids were incompatible with each other within the subcluster, but were compatible between subclusters, using a traditional Inc test [89]. It was shown that IncA and IncC plasmids are compatible and thus they proposed that they should not be referred to as “IncA/C” [90]. These findings showed that even plasmids with homologous replication initiation proteins could be compatible in Inc tests. In-depth comparisons of similar plasmids based on their nucleotide sequences and experimental Inc tests will provide more information about co-occurrence of two different plasmids in the same cell and their host ranges.

Table 6.1 List of plasmids in different incompatibility groups
Full size table

6.2.2 Conjugation

Conjugation is an important mechanism for horizontally transferring plasmid DNA between different organisms. Below, we focus on the conjugative transfer of plasmids found in both Gram-negative and Gram-positive bacteria.

Self-transmissible plasmids in Gram-negative bacteria generally carry complete sets of the genes required for transfer, i.e., the origin of transfer (oriT), relaxase protein, type IV coupling protein (T4CP), and type IV secretion system (T4SS). Garcillán-Barcia et al. [91, 92] and Smillie et al. [78] classified the self-transmissible, or mobilizable, plasmids in the GenBank database into six mobility (MOB) types (MOBC, MOBF, MOBH, MOBP, MOBQ, and MOBV) and four classes of mating pair formation (MPF; MPFF, MPFG, MPFI, and MPFT) based on all-against-all BLASTP analysis followed by Markov clustering (MCL), to identify and classify homologous proteins of relaxases (for MOB), T4CPs, and T4SSs (for MPF). Mobilizable plasmids are non-self-transmissible because they have only MOB, or MOB and T4SS, but could be transferred by other self-transmissible plasmids, such as helper plasmids with T4SS and MPF. Conjugation is also affected by the type of sex pili, one of the features of MPF (rigid or flexible), and whether the preference in mating conditions (solid surface or liquid environment) between donor and recipient cells is different [93, 94]. It should be noted that the combination of Inc groups, MOB types, and MPF classes could be important for the host range of plasmids (Table 6.1).

Gram-positive bacteria transfer plasmids by two methods. First, a single strand of plasmid DNA is transported via a T4SS-like plasmid in Gram-negative bacteria, which seems to be widely used as means for transferring plasmids in Gram-positive bacteria [95]. pIP501 (found in Streptococcus agalactiae), the broadest transfer host range plasmid in Gram-positive bacteria, contains 15 genes for T4SS [96, 97]. Second, plasmids found in the order Actinomycetales have conjugative systems that function in a similar manner to the segregation of chromosomal DNA during bacterial cell division and sporulation [reviewed in [98]]. The translocation of double-stranded DNA to the recipient cell is mediated by an FtsK-homologous protein, which is known as ATP-dependent DNA translocase [95, 98, 99].

Multiple regulation systems for conjugation have been found both on plasmids and host chromosomes [reviewed in [100]]. Several plasmids including F, R100, R27, and RA3 encode their own transcriptional regulators (repressors) [75, 99,100,101,102,103,106], probably because conjugation is an energetically costly process. Several elements encoded on host chromosomes also participate in the conjugation of plasmids, for example, the transfer of Rhizobium leguminosarum plasmid pSym is induced by homoserine lactone [107]; plasmid F is influenced by the extracellular response element CpxA [108]; and plasmid R100 is regulated by host-encoded regulators, such as Dam methylase and Lrp protein [109, 110].

All conjugative plasmids contain at least one entry exclusion systems; when the cells already contain conjugative plasmids, they become inefficient recipients. This system could limit an excess transfer of plasmids, which can kill the recipients in a process known as lethal zygosis, and is almost essential for conjugative plasmids, though the physiological importance of this is still unclear [reviewed in detail by [111]]. These systems are negative factors for conjugations, at least in laboratory conditions, and are important for the host range of plasmids.

Plasmids have been classified based on the similarity between the nucleotide sequences of genes (or the amino-acid sequences of proteins) involved in replication and conjugation, including PCR-based replicon typing systems (PBRT) [112] and MOB typing [109,110,115]. Plasmids of IncA and/or IncC, IncHI1, IncHI2, IncI, and IncN groups have been subtyped by plasmid multilocus sequence typing (pMLST) using specific genes in plasmids as targets for PCR [116]. These classifications are effective but not sufficient in predicting their host ranges. This is because plasmid conjugation and its efficiency can vary, depending on many other factors including cell density, growth rate, nutrient availability, temperature, and high-salt stress [93, 94, 113,114,119]. Recently, divalent cations (Ca2+ and Mg2+) have been found to increase the conjugation efficiency of several Inc plasmids, especially the IncP-7 plasmid, pCAR1 [120, 121]. It is also affected by combinations of donor and recipient strains [122]. Sakuda et al. [120] found that conjugation frequencies of plasmids pCAR1 and NAH7 were similar in mating with one donor strain (Pseudomonas putida) and one of the two recipient strains (P. putida and Pseudomonas resinovorans) (mating with one donor and one recipient). In contrast, these plasmids were transferred more frequently to P. putida than to P. resinovorans when the two recipient strains were mixed (mating with one donor and two recipients, Sakuda et al., unpublished). The results suggest that a host-specific factor(s) could affect the host range of plasmids. There are several systems to prevent conjugation of plasmids. Restriction-modification and CRISPR-Cas systems inhibit the conjugation of plasmids by cleaving DNA sequences [123] (see Chap. 3). These physiological or environmental conditions of hosts and/or recipients could also affect the host range of plasmids.

6.2.3 Nucleoid-Associated Protein (NAP)

Nucleoid-associated proteins (NAPs) are DNA-binding proteins encoded on plasmids, as well as on chromosomes, and can function as transcriptional regulators by binding to several regions of the DNA. NAPs aid chromosomal DNA compaction in bacterial cells [124] (see Chap. 1). One member of the NAPs, a histone-like nucleoid-structuring (H-NS) protein, can bind to horizontally acquired elements and repress their transcription (silencing), which can reduce the deleterious effects of harboring foreign DNAs [125]. The NAP genes are also found in plasmids, which are proposed to have “stealth” effects that minimize the fitness reduction caused by carriage of plasmids [126], because deletion (or disruption) of the NAP genes in plasmids had greater effects on the host transcriptome than did the plasmid carriage [127, 128]. Therefore, NAPs encoded both on host chromosome and plasmids can affect host fitness. Because the numbers and combinations of NAPs alter in different plasmids and host chromosomes [129], they are important to determine the host range of plasmids. In fact, NAPs could affect the stability and conjugation of plasmids. For example, the IncHI1 plasmid R27 has a gene encoding H-NS, involved in the modulation of R27 transfer by interacting with a Hha/Ymo family protein [130]. Another example is the IncP-7 plasmid pCAR1, which has three NAP genes, pmr, phu, and pnd [131]. Double deletion mutants of pmr and phu or pnd reduced its stability and lost transferability [132]. Since the binding sites for Pmr, both on pCAR1 and its host (P. putida) chromosome, were found to be similar to those of chromosomally encoded H-NS-like proteins, TurA and TurB [133], it could be cooperatively regulated by the H-NS proteins encoded by both the plasmid and host chromosome, though their detailed molecular mechanisms are unclear. Similarly, another H-NS-like protein, Acr2, encoded on the IncA and/or IncC plasmid, was found to negatively regulate its conjugative transfer [134, 135]. Therefore, NAPs encoded on plasmids are important factors to determine their maintenance and conjugation in different hosts.

6.3 Fitness Cost

A plasmid can bring its host potential benefits through its accessory genes, but sometimes a burden (fitness cost), reducing the host growth rate and competitiveness under no selective pressure (e.g., that exerted by antibiotics). One of the experimental approaches to investigate the molecular mechanisms of how plasmid carriage affects the host fitness is to obtain a compensated mutant from the parental plasmid-bearing host [reviewed in [136]]. San Millan et al. showed that compensating mutations on chromosomal genes encoding helicase, kinase, or the global regulator GacA could reduce the fitness cost of plasmid carriage [137, 138]. Sota et al. [139] showed that mutation to the trfA1 gene encoding the replication initiation protein TrfA1 (TrfA1 variant) could reduce the fitness cost of the IncP-1 plasmid in an “inappropriate” or naïve host, Shewanella oneidensis. This reduction was not observed in Escherichia coli or Cupriavidus pinatubonensis, but in P. putida [140]. The tight binding of the wild-type TrfA to the host helicase DnaB caused fitness cost of host, and the TrfA1 variant had a lower affinity to the DnaB resulting in the reduction of fitness cost [140]. Stalder et al. [141] reported that three distinct patterns of evolution exist to reduce the fitness cost of plasmids: (1) mutations in trfA1 gene, (2) acquisition of a putative toxin-antitoxin system on a transposon from a co-existing plasmid, and (3) a mutation in the fur gene encoding one of the global regulators in its host. These facts indicate that the co-evolution of plasmid and host increases the persistence of plasmids in their hosts.

Effects on three different host Pseudomonas (P. putida, P. aeruginosa, and P. fluorescens) by carriage of the IncP-7 plasmid pCAR1 were compared based on their phenotypes and transcriptomes [138,139,144]. Changes in fitness varied between different hosts [144], although detailed molecular mechanisms have not yet been elucidated. One of the most striking responses in two of the three hosts was the induction of genes on prophages by pCAR1 carriage [143, 144]. Notably, Martinez-Garcia et al. [145, 146] showed that deletions of the four prophages from P. putida, which harbored no plasmids, increased the growth rate, transformation efficiency, and protein expression from the plasmid vector, suggesting that prophages themselves affect host fitness. Recently, Shintani et al. (unpublished) found that the deletion of these prophages of P. putida could affect the fitness of hosts with the IncP-1 and IncP-7 plasmids (Shintani et al. unpublished). Thus, there could be cross talk between the two mobile genetic elements, prophages (which integrate into the host chromosome) and plasmid, although their mechanisms remain unclear.

pCAR1 is the plasmid endowing carbazole-degrading ability (converting carbazole to catechol) to its host, but the growth rates of different hosts can change in minimal medium with carbazole as the sole carbon source. The growth of P. fluorescens Pf0-1 with pCAR1 was significantly slower than that of P. aeruginosa PAO1, or P. putida KT2440 [48]. This is because of the toxicity of accumulated catechol, an intermediate compound of carbazole degradation, and of differences in catechol metabolism in these hosts, whose catabolism is mediated by enzymes encoded on the host chromosomes [48, 147]. Notably, DNA rearrangements were found on pCAR1 and the chromosome of its host Pf0-1 cultured in a minimal medium with carbazole as the sole carbon source, which could allow the host to avoid the accumulation of catechol [48, 147]. This indicates that fitness could vary between different host bacteria or growth conditions. The differences in fitness costs in different host cells could influence the stability of plasmids in the host cells and their host ranges.

6.4 Prediction of Plasmid Host Range

6.4.1 Prediction of Host Range Based on Bioinformatics

Novel plasmids have been found in host genomes and metagenomes, thanks to the recent revolution in nucleotide sequencing technology and in bioinformatic tools. There are now 12,015 complete sequences of plasmids in the NCBI database: 11,710 are from Bacteria, 192 are from Archaea, and the remaining 113 are from Eukaryota (based on the NCBI database, ftp://ftp.ncbi.nlm.nih.gov/genomes/GENOME_REPORTS/plasmids.txt, downloaded on Mar. 2018). As more than 10,000 complete sequences of plasmids are available in public databases, we can now use bioinformatic approaches for predicting plasmid host ranges based on sequence features, including plasmid size, nucleotide composition (G+C content, oligonucleotide frequencies, and codon usage), and replication strand asymmetry (asymmetric nucleotide compositions and gene proportion between leading and lagging strands of DNA replication). Intragenomic variation in nucleotide composition has been used to detect putative alien genes acquired by horizontal transfer [148, 149].

Plasmid sizes (Kb) varied widely among sequenced plasmids, ranging from 0.537 (Xanthomonas campestris pv. campestris str. CN14) to 5836.680 (Pseudomonas monteilii). There are clear differences in sizes between non-transmissible and conjugative plasmids (including mobilizable- and self-transmissible plasmids) [78, 83]. Notably, larger plasmids and conjugative plasmids frequently carried multiple NAP genes [129]. Similar NAP genes were found on both plasmids and host chromosomes and thus, their combinations might be important for the stable maintenance of plasmids within their hosts.

The relative frequency of guanine and cytosine (G+C content), calculated by the formula (G+C)/(A+T+G+C), varies widely among bacterial genomes, especially at synonymously variable third positions within codons [150, 151]. G+C content is correlated with a number of variables including genome size, aerobiosis, lifestyle, and environments [152], and can be shaped by mutation and selection [153]. The G+C content variability reflects differences in the DNA polymerase III alpha subunit [154, 155]. G+C contents also varied widely among different plasmids, ranging from 19.3% (Eukaryota Moniliophthora roreri plasmid pMR2) to 87.5% (Actinobacteria Streptomyces autolyticus plasmid). Previous studies revealed that G+C content is lower in plasmids than in hosts [156, 157], and that there is a strong correlation between the G+C content of plasmids and host chromosomes [158]. For a correlation of G+C contents between 2296 plasmids and their corresponding hosts analyzed here, the Pearson correlation coefficient was 0.97, and in 1704 (74.2%) cases, the plasmids had lower G+C contents than their hosts (Fig. 6.1). Therefore, the G+C content of plasmids could be an important indicator for predicting their host ranges.

Fig. 6.1
figure 1

Plot of G+C contents (GC%) between 2296 plasmids and their corresponding hosts. Each point represents a plasmid-host pair from 920 taxa. The y-axis indicates the G+C content of plasmids (plasmid GC%) and the x-axis indicates the G+C content of hosts (host GC%), retrieved from the National Center for Biotechnology Information (NCBI) genome list (ftp://ftp.ncbi.nlm.nih.gov/genomes/GENOME_REPORTS/)

Full size image

Karlin et al. proposed that each genome has a characteristic “signature,” consisting of the relative abundance of vectors of oligonucleotides such as di-, tri-, and tetra-nucleotides (i.e., k-mers) [159, 160]. The oligonucleotide compositions of DNA sequence segments are relatively constant along the genome, and those from closely related taxa tend to be more similar than those from distantly related taxa. The oligonucleotide compositions of plasmids tend to be more similar to those of their known host chromosomes than to those of other bacterial chromosomes [161, 162]. Thus, bacteria with the most similar oligonucleotide compositions may be the most probable hosts in which plasmids have evolved [163, 164].

Most amino acids can be encoded by more than one codon, with codons encoding the same amino acid called synonymous codons. Synonymous codon usage varies among genes, between different organisms, and even within a single genome. The codon usage of genes can reflect a balance between translational selection, mutational biases, and other factors [165, 166]. For example, the strength of translational selection for synonymous codon usage varies among the three replicons of Sinorhizobium meliloti (the class Alphaproteobacteria), i.e., it is much weaker in the plasmid pSymA, than in the plasmid pSymB and the chromosome [167]. In Agrobacterium tumefaciens (the class Alphaproteobacteria), the differences in codon usage between chromosomes (circular and linear) and plasmids (pAt and pTi) are larger than the differences between two chromosomes or two plasmids [168]. In Borrelia burgdorferi (the class Spirochaetes), there is a high similarity in codon usage among the cp32 family plasmids, and between the chromosomal leading strand and linear plasmid lp38 [168], as well as a significant difference in codon usage between the leading and lagging strands due to strand-specific mutational biases [169, 170]. The codon usage of genes in the largest plasmid, as well as the chromosome of Lawsonia intracellularis (the class Deltaproteobacteria), can be affected by strand-specific mutational biases [171]. Codon usage of any replicon (plasmids and chromosomes) can reflect a complex balance between host-specific mutational biases and selective pressures, resulting in varying degrees of codon-usage similarity between replicons.

Another important feature of bacterial genomes is replication strand asymmetry. In bacterial genomes, essential and highly expressed genes are preferentially located on the leading strands of DNA replication [168,169,170,175]. GC skew, defined as (C−G)/(C+G), has been used for measuring strand compositional asymmetry and for predicting the origin and terminus of replication on bacterial chromosomes and plasmids [171, 176, 177]. A measure of the strength of GC skew, quantified by the GC skew index (GCSI), detected a difference in GCSI between replicons with different types of replication machinery (e.g., GCSI between eubacteria and archaea chromosomes, and GCSI between RCR and non-RCR plasmids), and a correlation between GCSI of plasmids and their host chromosomes [178], suggesting that any replicon (plasmids and chromosomes) replicated and repaired in the same cell, has been subject to host-specific mutational biases and selective pressures, resulting in similar degrees of GC skew.

There is evidence of plasmid-mediated horizontal gene transfer, and some plasmid genes integrate into host chromosomes [179]. For example, IncP (IncP-1) plasmid sequences were detected in the chromosomes of bacteria such as Pseudomonas [180] and Brucella [181], and genes encoding the replication initiator protein TrfA were also found in bacterial chromosomes [182]. Fondi et al. [183] identified genes shared between plasmids and chromosomes as possible indications of gene transfer between them in the genus Acinetobacter. Network analyses of homologous DNA families shared among chromosomes and mobile elements showed that betweenness centralities are higher in plasmids than in phages, indicating that plasmids (e.g., promiscuous IncP-1 plasmid pB10), rather than viruses, are key vectors of DNA exchange between bacterial chromosomes [184]. The presence of DNA sequences shared between plasmids and chromosomes suggests that these replicons co-resided in the same hosts at some point in their history, although we cannot rule out the possibility of natural transformation, i.e., the uptake of extracellular plasmid DNA.

Bioinformatic tools have been developed to find novel plasmids in genomic and metagenomic data, which can be divided into two categories: those that reconstruct plasmids via assembly of sequencing reads (PLACNET, PlasmidSPAdes, and Recycler) [181,182,187], and those to identify plasmids in assembled contigs (PlasmidFinder, cBar, and PlasFlow) [184,185,190]. Arredondo-Alonso et al. compared the performance of four tools (PlasmidSPAdes, Recycler, cBar, and PlasmidFinder) for detecting plasmids from short read sequencing data [191]. Krawczyk et al. demonstrated that PlasFlow outperformed cBar on test data [189]. These bioinformatic tools allow us to update the range of hosts in which plasmids are found.

6.4.2 Prediction of Host Range Based on Experiments

There have been several reports for experimental detection and separation of unknown hosts of plasmids from environmental samples. Since most bacteria in natural environments have not been isolated yet [192, 193], there must be a large number of unidentified hosts of plasmids. To obtain these (potential) hosts from environmental samples, cultivation-independent methods have been developed and adopted. One of the most efficient ways is to use fluorescent protein (FP) to visualize and detect plasmid-bearing cells, as part of a method developed by Molin et al. [190,191,196]. They use a lac-like promoter, whose expression is repressed in the presence of the LacI repressor, upstream of the FP gene. The lacI gene is then introduced into the donor chromosome. If the plasmid is transferred from the donor cell to a recipient cell without lacI, then FP can be expressed, and the transconjugant cell can be observed by fluorescence microscopy. This system could be used with fluorescence microcopy and/or a fluorescence activated cell sorter (FACS) to obtain the transconjugant cells [197, 198]. After sorting a single transconjugant cell, multiple displacement amplification (MDA) can be used to amplify genomic DNA from a single bacterial cell without cultivation processes [199]. Shintani and Klümper discovered previously unknown transconjugants using these systems, FACS or MDA, and sequencing of the 16S rRNA genes of sorted transconjugants [200, 201]. These systems could be used for the detection of unidentified plasmid hosts, but still show bias in the transconjugants detected and separated from environmental samples. The biases can be caused by two major factors: (1) the fluorescence intensity of FP could be drastically different from host to host [201], probably because the expression levels of FP vary between different hosts due to its promoter or codon usage; (2) MDA may not necessarily amplify all the genomic DNA from a single cell.

Another potential method for overcoming the bias of GFP expression in different hosts in environmental samples might be the use of fluorescence in situ hybridization (FISH) for detecting plasmid DNA [201, 202], although it is still difficult to apply this method as bacterial cells are so small and also contaminated with debris and particles in environmental samples. Comparisons by Raman spectrum of cells might be efficient in detecting hosts with catabolic genes obtained via plasmids, as the host cells could incorporate a specific substrate (the target of the catabolic gene products) labeled with a stable isotope. Indeed, Huang et al. reported that they successfully identified hosts of a naphthalene-degradative plasmid by Raman microscopy [203]. Introduction of new methods to detect differences between plasmid-free and plasmid-bearing cells will enable us to identify transconjugants in natural environments and expand our knowledge of plasmid host range.

6.5 Conclusions and Remarks

The determinants for a plasmid’s hosts are diverse and it is still challenging to clearly understand which can possess the plasmid. Bioinformatics will be instrumental in predicting the candidate hosts of plasmids, based on their nucleotide sequence features, with experimental data showing the prerequisite conditions required for the host cell to acquire the plasmid. Recent studies at the single-cell level have identified plasmid hosts in different environments that have the potential to provide us with important information on how plasmids spread between different bacteria in nature, including “transient” hosts that can be mediators and reservoirs of accessory genes. Nevertheless, experimental data, including from these recent studies, are still insufficient as predictors, and additional data will be required. The analyses will shed light on the mechanisms of plasmid spreading in natural environments, as well as bacterial adaptation and evolution.