Diphtheria is caused by toxigenic strains of Corynebacterium diphtheriae harboring integrated bacteriophage DNA containing the toxin gene. The infection spreads from person to person via airborne droplets and develops into classic pharyngeal or nasal diphtheria.
Over the past century, mass immunization programs have dramatically cut down the incidence of diphtheria [1]. In Russia, the incidence rate of the disease has stabilized due to good vaccination coverage (≥ 95%) [2]. In 2017, no incident cases of diphtheria and only 2 asymptomatic carriers were reported in Russia. In 2018, 4 incident cases of the disease and 3 carriers were reported, whereas in the first 9 months of 2019, there were 3 new cases of diphtheria and 2 carriers [3]. In the past few years, there have been no reports of the secondary cases or lethal infection. Most clinical forms of diphtheria are mild localized forms.
Today, diphtheria is a rare disease; therefore, it can pose a diagnostic difficulty to the clinician. This, as well as the existence of latent carriers, who aсt as a reservoir for the infection, and the fact that the epidemic process unfolds in the vaccinated population, still renders diphtheria a clinically important problem [3].
Recently, infections caused by nontoxigenic C. diphtheriae strains have been on the rise. They manifest atypically as pharyngitis, respiratory tract infections, endocarditis, osteomyelitis, septic arthritis or skin infections [4–8].

Historically, C. diphtheriae were classified into 4 biotypes based on their biochemical phenotypes: gravis, mitis, intermedius and belfanti [9, 10]. Representatives of the same biovar, though, can be genetically distant [11, 12]. This is why genomics does not support the use of biovars as a reliable classification tool for C. diphtheriae [13]. Besides, there is no correlation between the biovar and pathogenicity [14]. Multilocus sequence typing (MLST) based on the determination of allelic profiles of 7 housekeeping genes has made it possible to cluster the entire diversity of C. diphtheriae strains into 2 evolutionary lineages: lineage-1 (the majority of the strains) and lineage-2 (only strains of the belfanti biotype) [14].

In 2018, a paper was published describing 3 nontoxigenic strains of C. diphtheriae [15]. One of them had been isolated from a Swiss patient with tracheobronchitis and multiple lesions on the distal trachea and the mainstem bronchi; the 2 other strains had been isolated from nasal swabs in the UK and India. Genome comparison that used publicly available C. diphtheriae genomes demonstrated that average nucleotide identity between the isolated strains and the NCTC 11397 C. diphtheriae reference genome was lower (95.24 to 95.39%) than between the reference genome and other previously published C. diphtheriae genomes (> 98.15%). Phylogeny reconstruction based on whole genome sequencing data confirmed the existence of two monophyletic clusters of C. diphtheriae corresponding to lineage-1 and lineage-2. Consequently, it was proposed to classify C. diphtheriae into two subspecies: C. diphtheriae subsp. diphtheriae and C. diphtheriae subsp. lausannense.

The aim of this study was to characterize the genotype and the phenotype of nontoxigenic C. diphtheriae strains isolated in 2017–2018 that can be identified as C. diphtheriae subsp. lausannense based on the aggregated data yielded by a variety of different methods.

METHODS

In the experimental part of the study, we analyzed 2 nontoxigenic C. diphtheriae strains isolated at the bacteriological laboratory of Khabarovsk Regional Psychiatric Hospital in 2017–2018, the control toxigenic strain of C. diphtheriае (gravis biotype, accession number 665) from the State collection of pathogenic microorganisms (SCPM-Obolensk), freshly isolated toxigenic strains of C. diphtheriае (gravis biotype, strain numbers 66- 19, 98-19 and mitis biotype, strain numbers 55-19, 56-19), nontoxigenic strains of C. diphtheriае (gravis biotype, strain numbers 57-19, 67-19 and mitis biotype, strain numbers 60- 19, 91-19) that had been delivered to the Reference Center for the Surveillance of Measles, Rubella, Mumps, Pertussis, and Diphtheria (G.N. Gabrichevsky Research Institute for Epidemiology and Microbiology) from different Russian regions.
The analysis of C. diphtheriae strains was conducted following the guidelines № 4.2.3065-13 for laboratory diagnostics of diphtheria. The isolates were plated onto tellurite blood agar (2% fishmeal hydrolysate agar base, State Research Center for Applied Microbiology & Biotechnology; Obolensk; Russia) supplemented with 7% bovine blood (Leitran; Russia) and potassium tellurite (State Research Center for Applied Microbiology & Biotechnology; Obolensk, Russia) and kept in a temperature-controlled chamber at 37 °С for 24–48 hours. Grown colonies of C. diphtheriae were evaluated for their morphological, toxigenic and biochemical properties. The toxigenicity of C. diphtheriae strains was evaluated in a precipitation test using a Corynetoxagar medium (State Research Center for Applied Microbiology & Biotechnology; Obolensk, Russia) supplemented with 20% bovine serum (Leitran; Russia) and filter discs soaked in diphtheria antitoxin (Diagnostic Systems; Nizhny Novgorod, Russia). Each antitoxin-impregnated disc contained 5 ± 1 IU of diphtheria antitoxin (as suggested by the guidelines 4.2.3065-13). Biochemical properties of the cultures were determined from their cysteinase, urease, saccharolytic and nitrate reductase activity using the media prepared at our laboratory and a commercial DS-DIPH-CORYNE kit (Diagnostic Systems; Nizhny Novgorod, Russia).

To evaluate the ability of the analyzed strains to ferment N-acetylglucosamine, a phenol red broth was ex tempore supplemented with N-acetylglucosamine (Sigma-Adlrich; USA). Then, a loop full of overnight C. diphtheriae cultures grown on serum agar was added to 3 ml of the solution. The cultures were incubated at 37 °С for 24–48 h. Fermentation was evaluated based on the change in the color of the solution. Two toxigenic and two nontoxigenic gravis strains, as well as two toxigenic and two nontoxigenic mitis strains, were used as controls.
The sample of the analyzed published genomic sequences comprised 204 C. diphtheriae genomes representing diphtheriae and lausannense subspecies deposited in the NCBI Refseq database, 3 genomes of C. diphtheriae subsp. lausannense from the NCBI Genbank and one genome of Corynebacterium ulcerans BR-AD22, which served as an outgroup for phylogenetic reconstruction. In total, 208 genomes were included in the analyzed dataset.

Coding sequences retrieved from the genome annotations in the corresponding databases were clustered into ortholog groups using OrthoMCL [16] with standard settings (inflation index of 1.5; protein sequence similarity threshold of 50%; e-value of 10–5). For phylogeny reconstruction, we used the groups of orthologs that were made up of the genes present in every genome in the amount of 1 copy. Nucleotide sequences were aligned in MUSCLE software [17] and then concatenated. Phylogeny reconstruction was performed following the Maximum Likelihood algorithm implemented in FastTree software [18] using the GTR+CAT model. MLST types of the published sequences were predicted based on the data retrieved from PubMLST. Clonal clusters were formed in Phyloviz 2 using the goeBURST algorithm at the SLV level [19].
Total DNA was isolated from overnight C. diphtheriae cultures grown on fishmeal hydrolysate agar (State Research Center for Applied Microbiology & Biotechnology; Obolensk, Russia) supplemented with 10% bovine serum (Leitran; Moscow) using a standard boiling extraction method with subsequent centrifugation.

Detection of tox gene fragments in nontoxigenic C. diphtheriae strains was performed in accordance with the protocol described in [20]. The PCR reaction mix contained 1.5 mM MgCl₂, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 0.1 μM of forward and reverse primers, 200 mM of each dNTP, and 1 unit of Taq polymerase (Thermo Fisher Scientific; USA). DNA of the control toxigenic C. diphtheriae strain (gravis biovar, accession number 665) was used as a positive amplification control. MLST types of С. diphtheriae strains were determined following the international protocol [14]; fragments of 7 housekeeping genes were Sanger-sequenced, including atpA, dnaE, dnaK, fusA, leuA, odhA, and rpoB. Sequencing was carried out by Evrogen JSC (Moscow). Allele identification was done using the PubMLST database.
To identify dtxR fragments in the sequences of C. diphtheriae strains, PCR was carried out with one pair of primers for the entire region of the dtxR gene: GGGACTACAACGCAACAAGAA and TCATCTAATTTCGCCGCCTTTA as described in [20, 21]. The following primers were used for subspecies-specific PCR: ptsI_F: ACTTTCCGAACCTGCCATCC and ptsI_R:GTGTACTCCTTCGTCTGCTC; narG_F: CTGACCACTGGGGCGAGG and narG_R: GAGTTGTCATAACGCCACTG.

RESULTS

The C. diphtheriae strains В-8759 and В-8760 had been isolated from the pharynx of two patients (26 and 77 years) admitted to 2 different units of a psychiatric hospital; the patients had undergone a standard preadmission test for diphtheria (see Paragraph 3.4. of the Guidelines 3.1.3018-12 on the epidemiological surveillance of diphtheria infection). The isolated strains were identified from their morphological, toxigenic and biochemical properties as recommended by the Guidelines 4.2.3065-13 on the laboratory diagnostics of diphtheria. On tellurite blood agar, the grown colonies appeared grayish-black, fuzzy, crumbly, with slightly irregular margins and a raised center. Toxigenicity of the grown colonies was determined using discs impregnated with diphtheria antitoxin (the Feldman method). After 24 and 48 hours of incubation, the isolated cultures still had not produced specific precipitation lines, unlike the control toxigenic C. diphtheriae strain (gravis biotype, accession number 665), suggesting the absence of toxigenicity. PCR performed to detect the fragment of the tox gene demonstrated that the analyzed strains did not carry the diphtheria toxin gene.

Biochemical properties of the cultures were determined from their cysteinase, urease, saccharolytic and nitrate reductase activity. The cultures exhibited cysteinase activity and formed a brown halo following inoculation into the Pisu medium; the cultures fermented glucose, maltose, fructose and galactose, did not ferment saccharose and starch, and exhibited no urease or nitrate reductase activity (table). The tests allowed us to provisionally assign the analyzed C. diphtheriae strains to the belfanti biotype typically seen in lausannense subspecies.

In the next step, we analyzed the previously published genomes of C. diphtheriae, which was necessary to verify that the studied species can be distinctly divided into subspecies and to conduct a search for species-specific protein-encoding genes.

The constructed phylogenetic tree (fig. 1) confirmed the results previously obtained on a smaller sample indicating that representatives of C. diphtheriae constituted two clades corresponding to the subspecies diphtheriae and lausannense. The tree also showed that the representatives of these subspecies belonged to non-overlapping groups of sequence types. The goeBURST clustering analysis of MLST types described in PubMLST (fig. 2) revealed that all representatives of the lausannense subspecies whose genomes had been previously sequenced belonged to the sequence types ST106, ST360, or ST409, and to one previously undescribed type that differed from ST359 in just one allele. All these sequence types formed one clonal complex.
It could be hypothesized that other sequence types (such as ST35, ST37, ST69, or ST81) constituting the same clonal complex also belong to the lausannense subspecies. An additional argument in favor of our hypothesis is that almost all isolates representing the sequence types from this clonal complex have been described in the PubMLST database as representing the belfanti biotype typical to the lausannense subspecies.

The analysis of ortholog groups revealed the existence of loci specific to C. diphtheriae subspecies. For example, all strains of the lausannense subspecies had a region (presumably, an operon) harboring genes of the phosphotransferase system, for which N-acetylglucosamine is a hypothesized substrate. The following primers were selected for the gene coding for phosphoenolpyruvate-protein phosphotransferase: ptsI_F: ACTTTCCGAACCTGCCATCC and ptsI_R: GTGTACTCCTTCGTCTGCTC (the expected product length was 489 bp). At the same time, a locus encoding the synthesis of nitrate reductase subunits was detected only in the genomes of the diphtheriae subspecies. The following primers were selected for the gene encoding its α-subunit (the gene was present in the sequences of 201 out of 202 strains representing this subspecies in the analyzed sample): narG_F: CTGACCACTGGGGCGAGG and narG_R: GAGTTGTCATAACGCCACTG (the expected product length was 691 bp).

PCR with primers for the amplification of ptsI and narG fragments (fig. 3) showed that the samples containing DNA of В-8759 and В-8760 strains carried the ptsI gene and did not carry the narG gene, whereas “classic” C. diphtheriae strains had the narG gene and did not have ptsI. There were no samples that carried either both of these genomic loci or none of them.
These findings and the results of biochemical identification allowed us to conclude that the analyzed C. diphtheriae strains belonged to C. diphtheriae subsp. lausannense. The conclusion was corroborated by the fact that the isolated strains represented the sequence type ST199 included in the clonal complex presumably typical to the representatives of this subspecies (fig. 2). Another piece of evidence confirming our conclusion was the sequence of the dtxR gene that coincided with the sequences found in the genomes of lausannense subspecies.
Considering that strains of C. diphtheriae subsp. lausannense carried the gene coding for phosphoenolpyruvate-protein phosphotransferase, which is part of the phosphotransferase system for N-acetylglucosamine, we conducted a few experiments to investigate the phenotypic manifestations of this gene. The experiments showed that unlike C. diphtheriae subsp. diphtheriae, both analyzed strains, which we classified as C. diphtheriae subsp. lausannense, fermented N-acetylglucosamine (fig. 4).
The analyzed strains, which we classified as C. diphtheriae subsp. lausannense, were deposited in the State collection of pathogenic microorganisms (SCPM-Obolensk).

DISCUSSION

We were able to identify the two analyzed C. diphtheriae strains isolated from the samples of Russian residents as nontoxigenic representatives of subsp. lausannense. Our findings along with the reports of foreign researchers [14, 22] suggest that these strains are ubiquitous. They belong to the sequence type ST199, which is part of the lineage-2 cluster typical to the lausannense subspecies, and carry the sequence of the dtxR gene characteristic of lausannense representatives. The analysis of ortholog groups established the existence of loci specific to the subspecies of С. diphtheriae: the region containing the genes of the N-acetylglucosamine-phosphotransferase system (specific to the lausannense subspecies) and the region encoding the synthesis of nitrate reductase subunits (specific to the diphtheriae subspecies). Our findings are consistent with the results of earlier genomic studies of the lausannense subspecies [15] and the studies of the biochemical properties of the belfanti biotype [9, 10]. Primers designed for these genes and the subsequent PCR allowed us to classify the two analyzed strains as C. diphtheriae subsp. lausannense.

CONCLUSIONS

We have identified the 2 analyzed strains collected on the territory of Russia as nontoxigenic strains of C. diphtheriae subsp. lausannense. Our findings expand the knowledge of the biological diversity of C. diphtheriae and indicate the need for estimating the spread of these microorganisms.