The recent discovery of bacteria resistant to a last-resort antibiotic called colistin prompted headlines warning of a ‘post-antibiotic era’. Colistin resistance had in fact been found before – what alarmed scientists this time was the location of the colistin resistance gene. Rather than being found on the main chunk of bacterial DNA (the chromosome), the resistance gene was located on a smaller chunk of DNA called a plasmid. A bacterium’s genetic content is replicated and passed on through the generations. In addition to this, many plasmids can pass a copy of themselves on to a neighbouring bacterium – even a bacterium of a different species! Therefore, a troublesome resistance gene located on a plasmid is more likely to spread and threaten lives in future, compared with the same gene located on the chromosome.
Bacteria engaged in ‘conjugation’, in which a plasmid can be transferred between bacterial cells.
Picture credit: AJC1, flickr
Classifying plasmids using ‘typing’ schemes
To understand the spread of ‘resistance plasmids’ (plasmids carrying resistance genes), researchers often classify plasmids into different types, based on their DNA sequence. Plasmids sometimes exchange bits of their DNA with other DNA molecules occupying the same bacterial cell (other plasmids, the chromosome). Consequently, plasmids that are closely related, through shared evolutionary history, may be quite different in terms of their overall sequences. So, to type plasmids, instead of looking at entire plasmid sequences, scientists focus on stretches of sequence which are less likely to be lost or exchanged; e.g. those performing key plasmid functions whose loss could spell doom for the plasmid. In this way, the typing schemes should be able to type many different plasmids, and types should reflect evolutionary relationships. Plasmids of the same type are likely to be evolutionary related, and follow-up analysis can reveal the degree of relatedness; highly related plasmids may belong to the same chain of transmission. This line of reasoning was used to show that migratory birds may be responsible for spreading antibiotic resistance across continents!
The main plasmid typing schemes are ‘replicon typing’ and ‘MOB typing’. They target different stretches of plasmid sequence with different functions: replicon sequence (plasmid replication) and relaxase sequence (plasmid mobility). To type a plasmid, its sequence is compared with a set of ‘reference’ replicon or relaxase sequences, each representing a type; following comparison, sequence matches will determine plasmid type.
In our study, published in the journal Plasmid, we assessed the performance of replicon and MOB typing schemes. For example, we wanted to find out what proportion of known plasmids could be typed*. We first needed to get hold of a large database of complete plasmid sequences. This turned out to be tricky. Although scientists deposit their sequence data in a public database called NCBI, sequences are not labelled in a consistent way, so distinguishing a partial plasmid sequence (say, a plasmid gene) from a complete plasmid sequence is not trivial. Also, a sequence that is actually a chromosome may occasionally be mislabelled as a plasmid. We developed a protocol for curating the sequences, which enabled us to compile over 2000 verified complete plasmid sequences. Looking at the number of plasmids added to the NCBI database over time (based on the date they were first submitted), as well as the method used for sequencing, we found that new sequencing technology has massively increased availability of complete plasmid sequences in recent years…all the more reason to investigate replicon and MOB typing schemes, which previously, had only been assessed using considerably smaller plasmid datasets.
Analysis of our curated plasmid dataset showed that 85% of plasmids could be replicon typed and 65% could be MOB typed. Why did some plasmids remain untyped? In the case of MOB typing, a plausible explanation, based on existing knowledge of plasmid biology, is that the untyped plasmids simply lacked the relaxase sequence required for typing. Further explanation can be found in this blog. On the other hand, controlling replication is an essential plasmid function, so it seems unlikely that plasmids could get away without any replicon sequence of their own. Perhaps the untyped plasmids did contain replicon sequences, but they were different from any of the reference sequences, meaning no match was found.
The study investigated various other avenues of research, such as whether particular plasmid types tend to carry certain resistance genes more than others (for details, see the original paper). Overall, the study shows that current typing schemes fail to classify the complete diversity of plasmids. The curated plasmid dataset compiled for this study is publicly available, and could be useful for plasmid researchers, including for applying novel methods of assessing plasmid diversity, that depend on large plasmid datasets.
This blog was written by Alex Orlek, a DPhil student at Modernising Medical Microbiology.
Ordering the mob: Insights into replicon and MOB typing schemes from analysis of a curated dataset of publicly available plasmids. Orlek, A., Phan, H., Sheppard, A.E., Doumith, M., Ellington, M., Peto, T., Crook, D., Walker, A.S., Woodford, N., Anjum, M.J., Stoesser, N.
*We focused only on plasmids from the Enterobacteriaceae family of bacteria, since replicon typing is most well-developed for plasmids from this group of bacteria.