Multilocus Sequence Analysis of Clinical “Candidatus Neoehrlichia mikurensis” Strains from Europe

Abstract

“Candidatus Neoehrlichia mikurensis” is the tick-borne agent of neoehrlichiosis, an infectious disease that primarily affects immunocompromised patients. So far, the genetic variability of “Ca. Neoehrlichia” has been studied only by comparing 16S rRNA genes and groEL operon sequences. We describe the development and use of a multilocus sequence analysis (MLSA) protocol to characterize the genetic diversity of clinical “Ca. Neoehrlichia” strains in Europe and their relatedness to other species within the Anaplasmataceae family. Six genes were selected: ftsZ, clpB, gatB, lipA, groEL, and 16S rRNA. Each MLSA locus was amplified by real-time PCR, and the PCR products were sequenced. Phylogenetic trees of MLSA locus relatedness were constructed from aligned sequences. Blood samples from 12 patients with confirmed “Ca. Neoehrlichia” infection from Sweden (n = 9), the Czech Republic (n = 2), and Germany (n = 1) were analyzed with the MLSA protocol. Three of the Swedish strains exhibited identical lipA sequences, while the lipA sequences of the strains from the other nine patients were identical to each other. One of the Czech strains had one differing nucleotide in the clpB sequence from the sequences of the other 11 strains. All 12 strains had identical sequences for the genes 16S rRNA, ftsZ, gatB, and groEL. According to the MLSA, among the Anaplasmataceae, “Ca. Neoehrlichia” is most closely related to Ehrlichia ruminantium, less so to Anaplasma phagocytophilum, and least to Wolbachia endosymbionts. To conclude, three sequence types of infectious “Ca. Neoehrlichia” were identified: one in the west of Sweden, one in the Czech Republic, and one spread throughout Europe.

INTRODUCTION

The strict intracellular bacterium “Candidatus Neoehrlichia mikurensis” is placed phylogenetically within the family Anaplasmataceae (1). It was first detected in Ixodes ricinus in the Netherlands in 1999 and referred to as an Ehrlichia-like bacterium (2). This agent has subsequently been reported to be widely spread among ticks and rodents in Europe (3–9). The first human case of “Ca. Neoehrlichia mikurensis” infection was published in 2010 and diagnosed by PCR-amplification and subsequent sequencing of the 16S rRNA genes detected in several blood samples from an immunocompromised patient (10). Neoehrlichiosis, the human infectious disease caused by “Ca. Neoehrlichia mikurensis,” is believed to be transmitted via tick bites (11) and may present as a severe febrile illness with thromboembolic events in immunocompromised patients (12). “Ca. Neoehrlichia mikurensis” infection may also pass unnoticed in healthy persons (13) or give rise to fever with additional symptoms (14).

The aims of this study were to investigate the genetic diversity of clinical strains of “Ca. Neoehrlichia mikurensis” from various parts of Europe and to determine their degree of relatedness to other species within the family of Anaplasmataceae. Bacterial genotypic analyses can be performed only on the strains in the blood samples of patients, not on isolated bacterial strains, since “Ca. Neoehrlichia mikurensis” has not been cultivated yet. Because of the current inability to obtain the bacterium in culture, strains are prescribed within the “Candidatus” taxonomic category, which was established for the purpose of cultivation-independent, genotypic-based descriptive and comparative analyses of uncultivable microorganisms (15, 16). Previous studies of the genetic diversity of “Ca. Neoehrlichia mikurensis” have focused on the 16S rRNA gene and groEL sequences (17–20). However, as the 16S rRNA gene is highly conserved, it lacks the power to define the degree of relatedness between organisms below the species level (21, 22). Here, we have developed a multilocus sequence analysis (MLSA) protocol based on five housekeeping genes in addition to the 16S rRNA gene. These housekeeping genes encode essential proteins that are less conserved and have higher mutation frequencies compared to those displayed by genes encoding ribosomal components. MLSA can map the genetic changes in these housekeeping genes and can serve as a reliable method for the study of epidemiological relationships (23).We have analyzed clinical “Ca. Neoehrlichia mikurensis” strains derived from 12 immunocompromised neoehrlichiosis patients from Sweden, Germany, and the Czech Republic; nine of these cases have been published already (10, 12, 20, 24, 25).We also report on three new cases of neoehrlichiosis diagnosed in Sweden during 2014 and the associated strains that were typed with this new assay.

MATERIALS AND METHODS

Clinical samples.

The origins of the clinical “Ca. Neoehrlichia mikurensis” strains are presented in Table 1, and their geographic distribution is shown in Fig. 1. All the clinical samples were from infected, immunocompromised patients. The Swedish patients participated in the “Neo-VÄST” study, which was approved by the Local Ethics Committee in Gothenburg, Sweden (394-12); details on the Czech and German patients have been published previously (20, 25). The study was performed in accordance with the declaration of Helsinki. Neoehrlichiosis was diagnosed by the analysis of EDTA-plasma samples using a specific real-time PCR targeting the groEL sequence and confirmed by panbacterial PCR assays with subsequent sequencing of the genes (10, 12, 20, 25).

TABLE 1.

Origins of the clinical “Candidatus Neoehrlichia mikurensis” strains and EBI^a accession numbers

Patient	Country	City	Date of diagnosis	Reference	16S rRNA	groEL	lipA	gatB	ftsZ	clpB
SE01	Sweden	Kungälv	July 2009	10	LN831013	LN830965	LN830953	LN830977	LN830989	LN831001
SE02	Sweden	Kungälv	July 2011	12	LN831014	LN830966	LN830954	LN830978	LN830990	LN831002
SE03	Sweden	Kungälv	July 2011	12	LN831015	LN830967	LN830955	LN830979	LN830991	LN831003
SE04	Sweden	Arvika	January 2013	12	LN831016	LN830968	LN830956	LN830980	LN830992	LN831004
SE06	Sweden	Gothenburg	June 2013	12	LN831017	LN830969	LN830957	LN830981	LN830993	LN831005
SE09	Sweden	Kungälv	February 2014	This study	LN831018	LN830970	LN830958	LN830982	LN830994	LN831006
SE10	Sweden	Skövde	March 2014	This study	LN831019	LN830971	LN830959	LN830983	LN830995	LN831007
SE11	Sweden	Lund	April 2014	24	LN831020	LN830972	LN830960	LN830984	LN830996	LN831008
SE12	Sweden	Uddevalla	June 2014	This study	LN831021	LN830973	LN830961	LN830985	LN830997	LN831009
CZ01	Czech Republic	Prague	March 2008	25	NDb	LN830974	LN830962	LN830986	LN830998	LN831010
CZ02	Czech Republic	Prague	July 2009	25	ND	LN830975	LN830963	LN830987	LN830999	LN831011
DE01	Germany	Franconia	June 2007	20	LN831022	LN830976	LN830964	LN830988	LN831000	LN831012

^aEuropean Bioinformatics Institute database (29).

^bND, not determined.

FIG 1.

The geographic distribution of the 12 immunocompromised patients that were infected with “Ca. Neoehrlichia mikurensis” and whose strains were characterized with the MLSA assay. (A) SE01, SE02, SE03, SE06, and SE09; (B) SE04; (C) SE10; (D) SE12; (E) SE11; (F) DE01; (G) CZ01 and CZ02.

Bacterial DNA preparation.

DNA from the Swedish samples was prepared from EDTA-plasma using a MagNA Pure Compact extraction robot (Roche, Basel, Switzerland) and nucleic acid isolation kit I (Roche) according to the manufacturer's protocols. The “Ca. Neoehrlichia mikurensis” DNA from the Czech and German patients was also extracted from EDTA-plasma but with different methods described previously (20, 25); these samples were sent frozen to Sweden for the MLSA.

MLSA target loci and primer design.

The choice of the housekeeping genes used for the MLSA assay was based on reports of orthologous genes and published protocols used for multilocus sequence type (MLST) analysis of other species belonging to the family Anaplasmataceae, i.e., Ehrlichia ruminantium (26) and Wolbachia pipientis (27). For each gene, primers were designed manually using sequence alignments of genes from 10 whole-genome sequenced from strains of the family Anaplasmataceae. Several potential loci were identified, and the selected genes and primer sequences that were tested and used are listed in Table 2.

TABLE 2.

Housekeeping genes and primers used in the “Ca. Neoehrlichia mikurensis” MLSA

Gene	Product	Function	Primers	Sequence (5′–3′)
groEL	Heat shock protein	Chaperone	Gro1 Fa	GAA GCA TAG TCT AGT ATT TTT G
			Gro1 Rb	GTC TAC CTG ATC TTG CTA TAT T
			Gro2 F	AAG GAA TTA GTA TTA GAA TCT T
			Gro2 R	TTA ACT TCT ACT TCA CTT GAA
lipA	Lipase A	Lysosomal acid lipase	lipA F	GTA GGY TGY AAA TAY TGA CCA AT
			lipA R	AAT ATT GGT GAG TGT TGG AAT AA
clpB	Caseinolytic peptidase B protein homolog	Chaperone, ATP-dependent protease	clpB F	TTC MGC YTG CCA YTT ACT A
			clpB R	GGW GCA ATG GAT GCA TCY AA
ftsZ	Filamenting temp-sensitive mutant Z	Cell division protein	ftsZ F	GCW GTN AAY AAY ATG ATA CAG TC
			ftsZ R	GCT TTG CCC ATR TCG CTC
gatB	Glutamyl-tRNA amidotransferase (subunit B)	Translation	gatB F	GAT GTA GCA ATG CCW GGT ATG
			gatB R	CWG AAA CAT TTG CAT CAC AAC G
16S rRNA	rRNA 16 S subunit	Protein synthesis	SSU1mod F	CGG CGT GCC TAA TAC ATG CAA GTC G
			806 R	GGA CTA CCA GGG TAT CTA AT

^aF, forward primer.

^bR, reverse primer

PCR and sequencing.

PCR amplifications of the housekeeping genes were performed in 20-μl reaction volumes using 10 μl Platinum SYBR green qPCR SuperMix (Invitrogen, CA, USA), 500 nM of each primer, and 4 μl of template DNA. The PCR conditions were as follows: an initial denaturation at 95°C for 2 min followed by 45 cycles of denaturation at 95°C for 15 s, annealing at 52°C for 45 s for the genes gltA, ftsZ, lipA, clpB or annealing at 54°C for 45 s for groEL, and extension at 72°C for 1 min. PCR amplification of the 16S rRNA gene of the Swedish “Ca. Neoehrlichia mikurensis” strains was performed as previously described (28). The resulting PCR products were loaded on a 2% agarose gel, stained with ethidium bromide, and visualized using UV fluorescence. The DNA bands of the expected size were cut out from the gel, purified with QIAquick gel extraction kit (Qiagen, Hilden, Germany) and subjected to cycle sequencing in both directions as previously described (28). The DNA sequences of the PCR products were purified by ethanol precipitation and then analyzed using the BigDye Terminator v. 3.1 kit and an ABI Prism 3130 genetic analyzer (Applied Biosystems, CA, USA). The 16S rRNA gene sequences from the Czech and German samples, and the groEL sequence of the German isolate were already published (20, 25).

Sequence data analysis.

For each of the genes, the sequences were edited using Sequencing Analysis software v. 5.2 (Applied Biosystems) and aligned using BioNumerics v. 6.6 (Applied Maths, Sint-Martens-Latem, Belgium). Housekeeping gene and 16S rRNA gene sequences from other species of Anaplasmataceae were retrieved using BLAST and aligned in the same way. The unweighted pair group method with arithmetic mean (UPGMA) dendrograms of the sequence similarities were constructed for each gene based on aligned sequence data from the 12 clinical “Ca. Neoehrlichia mikurensis” strains and three strains of E. ruminantium (GenBank accession number CR925678.1), Anaplasma phagocytophilum (GenBank accession numbers CP000235.1), and a Wolbachia endosymbiont (GenBank accession number AM999887.1) using BioNumerics v. 6.6.

Nucleotide sequence accession numbers.

All MLSA sequences generated in this study have been deposited in the EBI database (29) under accession numbers LN830953 to LN831022 (Table 1) (http://www.ebi.ac.uk).

RESULTS

Patients.

The 12 neoehrlichiosis patients were middle aged or elderly (median age, 67 years; range, 55 to 78), mostly male (8/12), and immunocompromised due to an underlying hematologic or rheumatologic disease and immune suppressive therapy. Nine of the 12 patients were from different parts of southern or central Sweden, one patient was from Germany, and two patients were from the Czech Republic (Table 1).

Three of the Swedish cases have not been published before (SE09, SE10, and SE12). Two of these patients had a systemic rheumatic disease, i.e., rheumatoid arthritis (SE09) and granulomatosis with polyangiitis, previously known as Wegener's granulomatosis (SE10). The third patient had pre-B acute lymphatic leukemia (SE12). All patients had received immunosuppressive therapy during the preceding 3 to 6 months: SE09 had been given methotrexate and rituximab, SE10 had also been treated with rituximab, and SE12 had received mercaptopurine, methotrexate, and systemic corticosteroids. All patients developed a fever of >39°C, patient SE09 had myalgia and arthralgia, and patient SE10 had duodenal ulcers, red and white blood cells in urinary sediment in the absence of raised antineutrophilic cytoplasmic antibody (ANCA) titers, weight loss, a punctate rash around the ankles, cough, and nausea. None were splenectomized. Two patients developed deep vein thrombosis encompassing the left thigh, knee, and lower leg (SE10) or solely the lower leg (SE12). SE09 and SE10 recovered completely after oral treatment with doxycycline (100 mg orally (p.o.) twice daily for 2 weeks). SE12, who was hypersensitive to doxycycline, received oral rifampin (300 mg p.o. twice daily for 2 weeks) and recovered completely within 1 week. Details of the clinical pictures and outcome of the remainder of the patients have been reported previously (10, 12, 20, 24, 25).

Primer performance.

All MLSA loci were successfully amplified from each of the 12 clinical samples. The sequencing analyses of the amplified PCR products generated sequences in both directions (Table 2). The sequences were trimmed at the 5′ and 3′ ends such that only high-quality sequence data were kept for alignments and cluster analyses.

Phylogeny of individual MLSA loci of “Ca. Neoehrlichia mikurensis” strains.

Sequence alignments and cluster analyses were conducted for each of the six loci separately, and individual UPGMA trees were constructed for each gene. The “Ca. Neoehrlichia mikurensis” strains derived from three of the Swedish patients (SE01, SE02, and SE09) differed from the other strains by one nucleotide exchange at position 114 (T instead of C) of the lipA locus (Table 3 and Fig. 2). This nucleotide exchange was a so called synonymous one, which did not change the corresponding amino acid. These three patients all came from the same part of Sweden (region A in Fig. 1) and were diagnosed between the years 2009 and 2014. However, two other clinical strains that were isolated from the same area and during the same time period (SE03 and SE06) did not harbor this nucleotide substitution (Fig. 2). A single nucleotide substitution was also observed in the clpB locus in one strain derived from one of the Czech patients, CZ01 (Fig. 3), with a nucleotide exchange at position 62 (G instead of C) (Table 3). This was a nonsynonymous nucleotide exchange that resulted in an amino acid change of valine for leucine in position 421; E. ruminantium also has leucine in this position (GenBank accession number CR925678.1). Finally, the ftsZ, gatB, groEL, and 16S rRNA loci were totally conserved, such that no nucleotide variation was noted for any of the 12 “Ca. Neoehrlichia mikurensis” strains (Table 3). Thus, three genotypes were apparent: one identified in the western part of Sweden and characterized by a variable nucleotide in the lipA gene, one from central Europe (Czech Republic) with a variable nucleotide in the clpB gene, and one pan-European type with conserved lipA and clpB gene sequences, respectively.

TABLE 3.

Results of multilocus sequence analysis of 12 “Ca. Neoehrlichia mikurensis” strains and three related bacterial species within the Anaplasmataceae family

“Ca. Neoehrlichia mikurensis” strain or Anaplasmataceae reference strain (accession no.)	No. of differing nucleotides/total no. of sequenced nucleotides (%a)
	lipA	clpB	ftsZ	gatB	groEL	16S rRNA
SE01 Sweden	1/455b	0/356	0/473	0/346	0/532	0/470
SE02 Sweden	1/455b	0/356	0/473	0/346	0/532	0/470
SE03 Sweden	0/455	0/356	0/473	0/346	0/532	0/470
SE04 Sweden	0/455	0/356	0/473	0/346	0/532	0/470
SE06 Sweden	0/455	0/356	0/473	0/346	0/532	0/470
SE09 Sweden	1/455b	0/356	0/473	0/346	0/532	0/470
SE10 Sweden	0/455	0/356	0/473	0/346	0/532	0/470
SE11 Sweden	0/455	0/356	0/473	0/346	0/532	0/470
SE12 Sweden	0/455	0/356	0/473	0/346	0/532	0/470
DE01 Germany	0/455	0/356	0/473	0/346	0/532	0/470
CZ01 Czech Republic	0/455	1/356c	0/473	0/346	0/532	Ad
CZ02 Czech Republic	0/455	0/356	0/473	0/346	0/532	A
Ehrlichia ruminantium (CR925678)	102/455 (22)	93/356 (26)	97/473 (20)	76/346 (22)	93/532 (17)	33/470 (7)
Anaplasma phagocytophilum (AM999887)	141/455 (31)	100/356 (28)	122/473 (26)	93/346 (27)	139/532 (26)	287/470 (61)
Wolbachia endosymbiont (CP000235)	152/455 (33)	98/356 (27)	130/473 (27)	102/346 (29)	153/532 (29)

^aPercent genetic variability compared with gene sequence of “Ca. Neoehrlichia mikurensis.”

^bNucleotide diversity in position 114 of the lipA gene (T instead of C).

^cNucleotide diversity in position 62 of the clpB gene (G instead of C).

^dA, absence of nucleotide sequence.

FIG 2.

Dendrogram from cluster analysis of the lipA locus. An UPGMA tree was constructed based on the lipA sequences from the 12 clinical “Ca. Neoehrlichia mikurensis” isolates. The strains from SE01, SE02, and SE09 have one nucleotide that differed from the other strains (99.8%).

FIG 3.

Dendrogram of the clpB locus. An UPGMA tree was constructed based on the clpB sequences from the 12 clinical “Ca. Neoehrlichia mikurensis” strains. The strain CZ01 has one nucleotide difference from the other strains (99.7%).

Genetic relatedness of “Ca. Neoehrlichia mikurensis” to other species in the family Anaplasmataceae.

Construction of the UPGMA trees using sequence data from the species E. ruminantium, A. phagocytophilum, and Wolbachia endosymbiont revealed a high degree of genetic diversity between the “Ca. Neoehrlichia mikurensis” strains and the other species within the Anaplasmataceae family (Fig. 4). E. ruminantium was observed to be most closely related to “Ca. Neoehrlichia mikurensis,” with high sequence dissimilarity in the loci ftsZ, gatB, groEL, lipA, and clpB, ranging from 17% to 26% (Table 3). This resulted in an approximate 80% degree of genetic similarity to “Ca. Neoehrlichia mikurensis” for each of the analyzed loci. A. phagocytophilum and Wolbachia endosymbiont differed even more from “Ca. Neoehrlichia mikurensis” with respect to the analyzed genes, having nucleotide dissimilarity rates of 26% to 33% in all MLSA genes (Table 3).

FIG 4.

Dendrograms of each MLSA locus. Individual UPGMA trees for the ftsZ, lipA, clpB, gatB, groEL, and 16S rRNA genes were constructed based on the gene sequences from the 12 clinical “Ca. Neoehrlichia mikurensis” strains and sequence data obtained for the reference strains of E. ruminantium, A. phagocytophilum, and Wolbachia endosymbiont, all members of the Anaplasmataceae family.

DISCUSSION

The MLSA scheme was clearly capable of discriminating between “Ca. Neoehrlichia mikurensis” and other species in the family of Anaplasmataceae. Our data show that “Ca. Neoehrlichia mikurensis” is easily distinguishable from E. ruminantium, A. phagocytophilum, and Wolbachia endosymbiont. This supports earlier results based on sequencing of the 16S rRNA, gltA, and the groEL genes (1). Unexpectedly, E. ruminantium, the agent of heartwater in cattle and sheep, was more closely related to “Ca. Neoehrlichia mikurensis” than to A. phagocytophilum, the cause of anaplasmosis in humans.

Overall, there was low genetic diversity in the five analyzed MLSA loci among the clinical “Ca. Neoehrlichia mikurensis” strains. Despite the low diversity, three sequence types became apparent, one identified in the western part of Sweden and characterized by one variable nucleotide in the lipA gene, one from central Europe (the Czech Republic) with one nucleotide exchange in the clpB gene, and one pan-European genotype with identical sequences in the lipA and clpB genes. These findings suggest that there are at least three potential clusters of “Ca. Neoehrlichia mikurensis” in Europe. von Loewenich et al. detected three different sequence types when analyzing “Ca. Neoehrlichia mikurensis” groEL and 16S rRNA gene sequences derived from ticks, rodents, dogs, and humans from Japan, Siberia, Germany, Switzerland, and the Netherlands (20). Li et al. reported four distinct clusters of “Ca. Neoehrlichia mikurensis” based on groEL and 16S rRNA phylogenetic analyses of “Ca. Neoehrlichia mikurensis” DNA derived from rodents, ticks, and infected humans in China (14, 18). A comparative alignment of the Chinese sequences with the deposited European “Ca. Neoehrlichia mikurensis” sequences (16S rRNA and groEL) did not reveal any sequence types in common, which suggests that the strains that infect humans in Europe differ from those in Asia. Our data also indicate that the “Ca. Neoehrlichia mikurensis” sequence variants may be stable over time since one of the genotypes was identified over a period of 5 years.

One reason for the low genetic diversity among the “Ca. Neoehrlichia mikurensis” strains observed in this study may be that the samples were all derived from humans, with all of them being immunocompromised. These patients were selected because they exhibited very high loads of bacteria in their blood (12), which allowed for the DNA sequence-based analyses of multiple genes. One of the challenges of this study was to select accurate genotyping markers for “Ca. Neoehrlichia mikurensis” since its genome is virtually unmapped, and the knowledge of this new bacterial species is very limited. The selected genes were based on published MLST protocols for related species within the family of Anaplasmataceae (26, 27). The potential for establishing an MLST database for evolutionary and functional patterns will increase as more allelic profiles become available for “Ca. Neoehrlichia mikurensis” isolates from infected humans with different clinical histories and from different geographic regions.

MLSA of selected housekeeping genes estimates relationships between closely related organisms and is particularly applicable when species boundaries are not well known; the MLSA data can be exploited to improve species description. In MLSA, the DNA sequences are used in downstream analyses, whereas in multilocus sequence typing (MLST), the downstream analyses are based on allele numbers and sequence types. MLST is a tool that is more easily communicable between different laboratories and is usually applied for investigating strains that belong to a well-defined species (30). Further, MLST involves analyses of a larger number of housekeeping genes (typically 7 genes excluding the 16S rRNA gene), which is preferable for epidemiologic and phylogenetic purposes (22, 31, 32). To sum up, this multilocus sequence analysis for “Ca. Neoehrlichia mikurensis” is a first step forward for the more precise genotyping of this emerging pathogen. Future MLSA should be done on strains after cultivation since a critical point in developing this MLSA was the limited availability of bacterial DNA from infected patients.