Current culture-based methods for measuring Escherichia coli (E. coli) diversity are slow and laborious and often fail to identify E. coli types present at low abundance in communities dominated by one or two other distinct clones. Routinely the 16S rRNA gene is targeted using amplicon sequencing methods to provide information on the overall bacterial community associated with a sample, but this method fails to provide any intra-species resolution. Thus, 16S rRNA community profiling can provide information on the relative abundance of E. coli but provides no information on the diversity of E. coli present in the sample, nor whether it is E. coli or Shigella spp. present as neither can be separated using the 16S rRNA gene.
The E. coli genome is a highly impacted by horizontal gene transfer leading to a mosaic structure of prophage insertions, acquisition and loss of genetic traits, and genome rearrangement through recombination events. A region of significant recombination in the E. coli genome is the O-antigen biosynthesis gene cluster (O-AGC) that encodes the O-antigen component of lipopolysaccharide, an antigenic extracellular component anchored within the outer membrane of the E. coli cell wall. The immunogenic nature of the O-antigen causes purifying selection of various E. coli in the gut environment resulting in extensive antigenic variability of the O-antigen due to recombination. At least 184 antigenically distinct E. coli O-groups have been identified, many of which are also genetically distinct through recombination events where O-group gene cluster exchange has taken place. gnd encodes 6-phophogluconate dehydrogenase, an enzyme that catalyses the third stage of the pentose phosphate pathway and is found in all E. coli and many Enterobacteriaceae. Although playing no role in O-antigen biosynthesis, gnd is often found close by to the O-AGC and has a hypervariable nucleotide sequence brought about by recombination events impacting the O-antigen. As a result, gnd has been described as a 'hitch hiker' of recombination.
We have targeted the gnd gene as a suitable E. coli barcode gene that provides enhanced resolution to identify E. coli community diversity using culture-independent methods (Cookson et al. 2017; https://www.nature.com/articles/s41598-017-00890-6), akin to targeting 16S rRNA for bacterial community analysis . Like the 16S rRNA gene, the complete gnd gene at 1407bp is unable to be used as a target for routine amplicon sequencing methods using the Illumina platform. Therefore, by using whole genome sequence data to compare gnd sequences from hundreds of E. coli isolates, we have been able to identify unique PCR primer sites that permit the amplification of a 284bp region of gnd allowing the differentiation of many E. coli. Furthermore, the use of WGS data available on public repositories and described in studies to evaluate the genomic diversity of E. coli, has provided valuable resources from which we have been able to assess the hypervariability of the gnd gene and develop a database of unique gnd partial amplicons. Previous studies employing sequencing of the full O-AGC region have targeted the wzx/wzy or wzm/wzt gene pairs located within the O-AGC to provide targets for O-group-specific PCR amplification and the identification of novel O-groups. However as additional WGS data from E. coli becomes available, subsequent analysis of O-AGC regions has provided evidence of genetically distinct O-AGC, wzx/wzy or wzm/wzt gene pairs and unique gnd sequence variants. For example, to extend our preliminary analysis of gnd allelic variation obtained from WGS data submitted to GenBank, IMG and PATRIC databases, we also examined genomes from 2244 E. coli isolates consisting of a species-wide collection assembled for the purposes of examining E. coli genomic diversity. This collection included multiple animal and environmental isolates and isolates from six different phylogroups (A, B1, B2, D, E, F). To date, 534 unique 284bp partial gnd sequences identified from E. coli have been included in the gnd database together with corresponding E. coli O-group/serotype data and a representative genome or nucleotide accession number. O and H-groups were identified from WGS in silico using SerotypeFinder (https://cge.cbs.dtu.dk/services/SerotypeFinder/). OXY (wzx/wzy) and OMT (wzm/wzt) are provisional O-group designations given to the O-AGC region of untyped (<95% similar at the nucleotide level) E. coli not in the serotyping panel.
Differentiation of gnd SNP variants
Unlike 16S rRNA amplicon sequencing where operational taxonomic units are arbitrarily clustered at the 97% homology level to represent distinct ‘species’, and to negate the effect of sequencing errors associated with the Illumina platform, 284bp partial gnd sequences representative of E. coli isolates sometimes differ by only one nucleotide/single nucleotide polymorphism. Thus, the resolution required for E. coli community analysis to distinguish genuine gnd sequence types that differ by only one nucleotide from those generated from sequencing error has necessitated the development of an ‘error correction’ model where a probability threshold can be established that provides a mechanism where the relative abundance of two gnd sequence types that differ by a single nucleotide, can be compared to determine whether they are likely to be genuine or not. For example, two gnd sequence types that are equally abundant in a single sample and that differ by only a single nucleotide are more likely to be genuine than two gnd sequence types that differ by only a single nucleotide where one is represented by a high number of reads, and the other is represented by only a small number of reads. Application of the error correction method in parallel with the inclusion of ‘mock’ E. coli communities with distinct gnd sequence types, allows many of the less abundant gnd sequences types to be clustered with a ‘parent’ type.
Enhanced resolution of E. coli populations using gnd amplicon sequencing methods
This E. coli community analysis technique is widely applicable to understand the impact of various interventions on E. coli populations associated with animal or environmental samples. DNA sample extractions or sample enrichments provide a source of genetic material from which the 284bp partial gnd sequence region from all E. coli can be amplified with indexed PCR primers to generate a unique amplicon. By using unique combinations of indexed forward and reverse primers with Illumina tags, unique gnd amplicons from separate samples can then be pooled and sequenced using the Illumina MiSeq platform. Subsequent demultiplexing, pairing of forward and reverse sequence reads and comparison with the gnd database provides information on the overall community diversity and read numbers associated with each gnd sequence type. The development of such a metabarcoding technique has provided a much greater resolution of E. coli community diversity to be established from complex matrices such as faecal material where frequently, an animal may be colonised by >10 distinct E. coli, where one or two gnd sequence types are present in a sample together with a larger number of sub-dominant gnd sequence types.
Target primer sites
Our observations suggest that the hypervariability of the gnd locus and lack of conserved regions precludes the opportunity to develop PCR primers that do not contain degeneracies. Current primers 2gndF 5′-TCYATYATGCCWGGYGGVCAGAAAGAAG (gnd coordinates 415 to 442) and 2gndR 5′-CATCAACCARGTAKTTACCSTCTTCATC (gnd coordinates 754 to 726) generate a 284bp amplicon, excluding primer sequences. Allelic variability outside of this 340bp total amplicon results in some redundancy where E. coli with distinct O-AGC cannot be distinguished. However, other E. coli O-groups (O37, O111, O113) or even serotypes (O157:H7), in the current iteration of the database, possess unique gnd sequence types.