Background

Anaplasma phagocytophilum is a Gram-negative obligate intracellular bacterium that replicates in neutrophil granulocytes [1]. It is transmitted to humans and animals by ticks of the Ixodes ricinus complex [2]. The main vector in Europe is I. ricinus [3]. Anaplasma phagocytophilum causes febrile illness primarily in humans [2], domestic animals such as horses [4], dogs [5] and cats [6], and farm animals such as sheep, cattle and goats [7]. The disease is called granulocytic anaplasmosis in humans and domestic animals and tick-borne fever in ruminants.

Anaplasma phagocytophilum is widely distributed globally, having been detected in the Americas, Europe, Asia and Africa [8]. However in cats, clinically apparent disease together with a consistent molecular identification of the infecting agent has been rarely described. Well-documented feline cases have been reported from Austria [9], Finland [10], Germany [9, 11, 12], Poland [13, 14], Sweden [15], Switzerland [9], the UK [16] and the USA [17, 18].

The clinical signs of feline granulocytic anaplasmosis are unspecific and most often comprise fever, lethargy and anorexia [19]. Typical laboratory findings are thrombocytopenia, lymphopenia and anemia [19]. The diagnosis is made best by amplification of pathogen-specific DNA from EDTA-anticoagulated blood [20] followed by sequencing of the amplicon to ensure specificity [6]. Usage of the msp2 gene as a molecular target is discouraged because a significant proportion of msp2 PCR results are false positives for currently unknown reasons [21]. The microscopic detection of bacterial inclusions in granulocytes, so-called morulae, in Giemsa-stained blood smears is possible but less sensitive and observer-dependent [22]. Although widely performed, serology is useless in patients presenting with an acute A. phagocytophilum infection because most are seronegative at this time point [20, 23].

The genetic characterization of A. phagocytophilum strains by various single and multilocus sequence typing schemes has revealed an association of specific types with the host species, vector species and geographic origin [24, 25]. However to date, it is not possible to clearly determine which of these three interdependent variables has the biggest impact due to huge sampling biases, the variety of typing methods applied and sequence length variation.

Presently, information on the A. phagocytophilum types infecting cats is limited. Strains from four European cats suffering from granulocytic anaplasmosis were characterized using multilocus sequence typing (MLST) and ankA-based typing [26, 27]. The A. phagocytophilum strains detected in three asymptomatic cats from Italy were subjected to groEL-based typing [28]. Furthermore, genetic characterization of strains identified in two asymptomatic cats from Korea relied on the sequencing of groEL and msp2 gene fragments [29]. Additionally, one groEL sequence from a feline host originating from the USA was part of the study by Jaarsma et al., but information on the disease status of the animal was not provided [30].

Thus, A. phagocytophilum strains eliciting clinical disease in cats have been characterized only rarely. Furthermore, several target genes and target gene sequences of differing length have been used in the past, which hinders the ability to compare results. Therefore, we attempted to genetically characterize A. phagocytophilum strains from seven cats suffering from granulocytic anaplasmosis using typing schemes that have been previously applied in cats (ankA and groEL gene-based typing, MLST) and that are highly standardized and curated (MLST).

Methods

Clinical signs of feline granulocytic anaplasmosis

The clinical data and laboratory findings of seven cats suffering from granulocytic anaplasmosis were part of a previously published study [9] but are reported here in more detail. Breed, sex, age, country of origin, clinical signs, rectal temperature, complete blood cell counts and year of detection were recorded. In some cats, electrolytes (sodium, potassium, phosphorus), transaminases (alanine aminotransferase, aspartate aminotransferase), alkaline phosphatase, bilirubin, creatinine and urea were also measured. Complete blood cell counts were determined using the ADVIA 2120i (Siemens Healthineers, Erlangen, Germany) or the Sysmex XT-2000i (Sysmex Deutschland, Norderstedt, Germany) instruments. Platelet counts < 90 × 109/l were confirmed by manual counting using a hemocytometer. Giemsa-stained blood smears were investigated microscopically for morulae at a magnification of 10 × 100. Electrolytes, transaminases, alkaline phosphatase, bilirubin, creatinine and urea were determined using the Cobas 8000 (Roche Diagnostics, Mannheim, Germany).

Genetic characterization

DNA samples positive for A. phagocytophilum were isolated from EDTA-blood drawn between the years 2018 and 2020 from seven cats suffering from granulocytic anaplasmosis. 523 bp (without primers) of the ankA gene [26] and 530 bp (without primers) of the groEL gene [31] were amplified and sequenced bi-directionally as described. For the purpose of this publication, ankA allele numbers were assigned to all applicable sequences published previously. The ankA cluster was determined as defined by Huhn et al. [26]. groEL haplotypes of the long fragment and groEL clusters were assigned according to the nomenclature defined by Jaarsma et al. [30]. The PubMLST database (https://pubmlst.org/) was updated to contain not only the MLST allele definitions but also the ankA allele and the groEL long-fragment haplotype nomenclature. For MLST, seven housekeeping genes (pheS, glyA, fumC, mdh, sucA, dnaN and atpA) were amplified and sequenced bi-directionally as reported previously [26]. Different sequences of a given locus were ascribed a unique but arbitrary allele number, and each unique combination of alleles was assigned a sequence type (ST). Clonal complexes (CC) were defined by sharing identical alleles at five of the seven loci with at least one other member of the group. MLST clusters were defined as described previously [26, 27].

Sequences were aligned by ClustalW applying the IUB matrix. Pairwise distances were calculated using MEGA 11.0.13 [32]. The phylogenetic tree was constructed using the neighbor-joining method with the Jukes-Cantor model in the program MEGA 11.0.13. Bootstrap analysis was conducted with 1000 replicates. To test for the concordance between different typing methods, adjusted Wallace coefficients [33] were calculated using the online tool accessible at: http://www.comparingpartitions.info/index.php?link=Tool.

The nucleotide sequences are available at GenBank under the accession numbers OQ435068–OQ435074 (ankA), OQ435150–OQ435152 (groEL), OQ435080–OQ435084 (pheS), OQ435090–OQ435094 (glyA), OQ435100–OQ435104 (fumC), OQ435110–OQ435114 (mdh), OQ435120–OQ435124 (sucA), OQ435130—OQ435134 (dnaN) and OQ435140–OQ435144 (atpA).

Results

Clinical signs of feline granulocytic anaplasmosis

Breed, sex, age, country of origin, clinical signs and rectal temperature of the seven cats are shown in Table 1. The median age of the cats was 10.0 years. Six cats lived in Germany and one in Switzerland. Information on clinical signs and rectal temperature was available in six out of seven cats. The most common clinical signs were lethargy and anorexia. All six cats had an increased body temperature.

Table 1 Breed, sex, age, country of origin, clinical signs and rectal temperature of the seven cats
Full size table

Complete blood cell counts of the seven cats are shown in Table 2. The most common laboratory finding was thrombocytopenia, which was present in five cats. The thrombocytopenia was mild in three and moderate in two animals. Leukopenia was observed in two, lymphopenia in four and anemia in one cat. Three animals showed a mild to moderate leukocytosis. Morulae inside neutrophils were present in the blood smears of five cats.

Table 2 Complete blood cell counts of the seven cats
Full size table

When tested, electrolytes, transaminases, alkaline phosphatase, bilirubin, creatinine and urea were within normal ranges (data not shown) with the exception of a moderate hyperbilirubinemia of 15.2 µmol/l (normal range < 3.4 µmol/l) in cat A480.

ankA gene

The ankA gene could be amplified from the blood of all seven cats. The obtained sequences were compared to four sequences from cats (GenBank accession numbers GU236864, FJ515309, MH987707, MH987708) published previously [26, 27] and to one unpublished sequence from a cat available at GenBank (GenBank accession number OQ435075). All 12 feline sequences originated from Europe and were part of ankA cluster 1 [26, 27]. They were 99.2–100% identical to each other and belonged to the ankA alleles 12 (9/12), 13 (1/12), 14 (1/12) and 15 (1/12) as shown in Table 3.

Table 3 ankA cluster, ankA allele, groEL cluster, groEL haplotype, MLST cluster, clonal complex (CC), sequence type (ST) and housekeeping gene allele numbers of the Anaplasma phagocytophilum strains from the seven cats that were part of the present study and from eight cats with sequences available at GenBank
Full size table

groEL gene

The groEL gene could be amplified only from three of the seven cats as the extracted DNA from the other four had been previously depleted. The obtained sequences were compared to three unpublished sequences from cats available at GenBank (OQ435153, OQ656703 and OQ656704) and to three sequences from cats (GenBank accession numbers KU519284, KU519285 and DQ680012) published previously [29, 30]. The groEL sequence with the GenBank accession number KU712086 (isolate 971) probably originates from the same animal as the sequence with the GenBank accession number OQ656704 (cat_971). However, the information concerning the country of origin (Finland versus Germany) and the disease state (asymptomatic carrier versus granulocytic anaplasmosis) is conflicting [35]. Here, only the sequences from cat_971 were considered for analysis. Unfortunately, three feline sequences reported by Balboni et al. [28] were too short to be included (232 bp–520 bp). Thus, nine groEL sequences of feline origin were analyzed.

Six sequences originated from Europe, two were from Asia and one from North America. Seven sequences belonged to the groEL cluster 1 [30], six of which were from Europe and one from North America. The two sequences from Asia were part of groEL cluster 4 [30]. The nine groEL sequences were 95.8–100% identical to each other and belonged to the groEL long-fragment haplotypes 5 (1/9), 6 (5/9), 113 (1/9), 178 (1/9) and 179 (1/9) as shown in Table 3.

The seven sequences from Europe and North America (groEL cluster 1) had an identity of 98.9–100% to each other and the two sequences from Asia (groEL cluster 4) of 99.8%, respectively. Thus, the feline groEL sequences from Europe and North America were more similar to each other than those from Asia, as shown in Fig. 1.

Fig. 1
figure 1

Phylogenetic tree calculated from the nine feline groEL sequences. The data set contained 530 positions. Tree construction was achieved by the neighbor-joining method using the Jukes-Cantor matrix. Bootstrap values are shown next to the branches. The scale bar indicates the number of nucleotide substitutions per site. Animals with identical groEL haplotype are shown in the same color

Full size image

MLST

All seven housekeeping genes could be amplified only from five of the seven cats as the extracted DNA from the other two had been previously depleted. After sequencing, the sequence type (ST) was ascribed as outlined in the Methods section. The respective STs were compared to four STs from feline A. phagocytophilum strains published previously [26, 27] and using unpublished sequences from one cat available at GenBank (OQ435085, OQ435095, OQ435105, OQ435115, OQ435125, OQ435135 and OQ435145). All ten STs of feline origin belonged to the MLST cluster 1 [26, 27]. The ten concatenated housekeeping gene sequences (2877 bp) were 99.5–100% identical to each other. ST 25 was found in six, ST 55 in two and ST 188 and ST 242 in one animal, respectively (Table 3). ST 25, 55 and 188 belong to clonal complex (CC) 1 [26, 27].

Concordance between typing methods

The concordance between ankA, groEL and MLST cluster was 100% (Table 3). Adjusted Wallace coefficients [33] were calculated to compare the partitioning of ankA alleles, groEL haplotypes and ST using the six A. phagocytophilum strains with complete information (Table 4). The concordance between ankA allele and groEL haplotype and ankA allele and ST was 100%. The concordance between ST and ankA allele and between ST and groEL haplotype was 100% as well. The concordance between groEL haplotype and ankA allele and between groEL haplotype and ST was only 33.3% as one of the two haplotypes found (haplotype 6) was present in 83% of the samples.

Table 4 Adjusted Wallace coefficients and 95% confidence intervals (in parentheses) in percent indicating the concordance between the partitions ankA allele, groEL haplotype and ST for the six Anaplasma phagocytophilum strains with complete information
Full size table

The association between disease state and typing method could not be calculated because there was a strong bias regarding geographic origin as the two asymptomatic carriers were both from Asia (Table 5).

Table 5 groEL cluster, groEL haplotype, disease state, country and continent of origin, year of detection and source of detection of the Anaplasma phagocytophilum strains from the seven cats that were part of the present study and from eight cats with sequences available at GenBank
Full size table

The concordance between groEL cluster, groEL haplotype and continent was analyzed for the nine strains with known groEL haplotype. The concordance between groEL cluster and continent was 50.9% as groEL cluster 1 was found in cats from Europe and North America whereas groEL cluster 4 was restricted to the two animals from Asia (Tables 5, 6). Vice versa, the concordance between continent and groEL cluster was 100% (Table 6).

Table 6 Adjusted Wallace coefficients and 95% confidence intervals (in parentheses) in percent indicating the concordance between the partitions groEL cluster, groEL haplotype and continent of origin for the nine Anaplasma phagocytophilum strains with complete information
Full size table

The concordance between groEL haplotype and continent was 100% because the respective haplotypes found were continent-specific. However, only one sample from North America and two samples from Asia were part of the analysis, whereas all others were from Europe.

Discussion

In the present study, fever, lethargy and anorexia were the most common clinical signs in cats suffering from granulocytic anaplasmosis. Furthermore, the most frequent laboratory finding was thrombocytopenia. Thus, clinical and laboratory observations were in line with the literature [19].

Detailed medical records were available only in seven cats. For six cats, the disease state could be extracted from GenBank or from the literature. In two cats, this information was conflicting or missing. Only two asymptomatic carriers were part of the study, and both were from Asia. Thus, due to the small number of cases analyzed here and the substantial sampling bias, it was impossible to correlate distinct clinical or laboratory observations or the disease state with certain A. phagocytophilum types.

The 12 European feline ankA sequences analyzed here belonged to ankA cluster 1. ankA cluster 1 has been reported previously [26, 27] to contain, among others, A. phagocytophilum strains from humans, domestic animals (horses, dogs), farm animals (sheep, cattle, goats), wildlife (European bison, red deer, red foxes, wild boar) and small mammals (European hedgehog). The ankA cluster 1 was initially described to harbor sequences from Europe and North America [26, 27] but is now restricted to strains from Europe since North American ankA clusters 11 and 12 were separated from cluster 1 [27]. The ankA alleles 12, 13, 14 and 15 found here in cats were detected previously in samples from humans, dogs, horses and other hosts [26, 27].

The nine feline groEL sequences analyzed here belonged to groEL cluster 1. groEL cluster 1 has been previously described [30] to contain A. phagocytophilum strains from humans, domestic animals (horses, dogs), farm animals (sheep, cattle, goats), wildlife (moose, red deer, roe deer, red foxes, wild boar) and small mammals (European hedgehog, Northern white-breasted hedgehog), among others. Sequences belonging to groEL cluster 1 were reported to be restricted to Europe and the Americas [30].

The two sequences from Asia belonged to groEL cluster 4. groEL cluster 4 has been described previously [30] to harbor A. phagocytophilum strains from a human, two dogs, two goats, rodents and ticks [30]. groEL cluster 4 is restricted to strains from Asia. However, it was previously reported to harbor strains from Europe as well [30], because strains from the Asian part of Russia were erroneously ascribed to the European continent.

The groEL long-fragment haplotypes 5, 6, 113, 178 and 179 found here in cats were detected previously, among others, in samples from humans, dogs and horses [30].

All ten STs of feline origin belonged to the MLST cluster 1. MLST cluster 1 has been previously reported [26, 27] to contain A. phagocytophilum strains from different hosts which included humans, domestic animals (horses, dogs), farm animals (sheep, cattle, goats), wildlife (European bison, red deer, red foxes, wild boar) and small mammals (European hedgehog). The MLST cluster 1 harbors strains from Europe and North America [26, 27]. ST 25 and ST 55 found here in cats were previously detected in samples from humans, dogs and horses as well as from other hosts [26, 27].

Thus, ankA-based typing, groEL-based typing and MLST showed consistent results. Specifically, the A. phagocytophilum strains found to infect cats were the same that caused disease in humans, dogs and horses. This was reported before for four cats (cat_1, cat_2, cat_971 and cat_596400) regarding ankA-based typing and MLST [26, 27] and for three cats (cat_S-DD-20, cat_S-DD-21, cat_DQ680012) regarding groEL-based typing [30] but is confirmed here on a broader data basis including a direct comparison of the three typing methods. Thus, strain divergence of feline strains is unlikely to explain the fact that granulocytic anaplasmosis is much less frequently diagnosed in cats than in dogs and horses. Therefore, due to the unspecific clinical signs, it should be considered that granulocytic anaplasmosis might be under-diagnosed in cats. However, it is also possible that cats are less susceptible to the same strains than dogs and horses. Differences in animal behavior between dogs and cats such as efficient removal of ticks by self-grooming cats might also serve as an explanation [36].

The observation that the same A. phagocytophilum strains infect cats, humans, dogs and horses was true for Europe, North America and Asia. However, feline strains from Europe and North America were more similar to each other than to those from Asia according to the groEL-based typing, although the overall identity was high with 95.8–100%. This result has to be confirmed by further studies with higher sample numbers as there was a substantial sampling bias regarding the continent of origin. Considering only A. phagocytophilum strains from non-vector species, the groEL data base contains 669 strains from Europe, 25 strains from the Americas and 25 strains from Asia [30]. The same is true for the ankA-based typing with 469 strains from Europe, 25 strains from North America and 7 strains from Asia and for MLST with 386 strains from Europe, 18 strains from North America and 7 strains from Asia [26, 27].

Unfortunately, three feline groEL sequences reported by Balboni et al. [28] were too short to be included. This underlines the importance of standardizing the typing of A. phagocytophilum strains to ensure the comparability of results across studies. Furthermore, during our analysis, it was noticed that samples from the Asian part of Russia were erroneously ascribed to the European continent by Jaarsma et al. [30] and that conflicting information concerning country of origin and disease state was reported for cat_971 [26, 35]. Therefore, we further emphasize the importance of collecting and reporting epidemiological data as completely and correctly as possible. Aliases of sample names should be avoided or correctly documented to prevent the same strain being considered multiple times within an analysis.

In our opinion, the typing of A. phagocytophilum strains should concentrate on target genes that already have substantial information available concerning sample numbers, different host species and diverse geographic origin. The same fragment length and position must also be used. Curated databases available via the internet should ensure quality control, correctness and completeness of the data and a universal nomenclature. At least, host species, country of origin, year, site of detection and disease state should be reported. To facilitate the accessibility of the data, the PubMLST database (https://pubmlst.org/) was updated to contain not only the MLST allele definitions, but also the ankA allele and groEL long-fragment haplotype nomenclature. Scientists are invited to submit their typing and isolates data to this database.

Conclusions

Although our analysis was limited by small sample numbers, we nonetheless provide important information on the clinical signs of cats suffering from granulocytic anaplasmosis. The genetic characterization using ankA-based typing, groEL-based typing and MLST showed that cats are infected by the same A. phagocytophilum strains as humans, dogs and horses. Given the sparse reports of granulocytic anaplasmosis in cats, feline infection by A. phagocytophilum might be under-diagnosed. However, the possibility remains that cats might be less susceptible to the same strains than dogs and horses are. Increased disease awareness in feline hosts should help to answer this question in the future.