Candidatus Mycoplasma haemohominis in Human, Japan

Hemotropic mycoplasmas are common pathogens in animals, but it remains unclear what role these pathogens play in human infections. We report clinical and biologic characterization of Candidatus Mycoplasma haemohominis infection in a 42-year-old man in Japan. The patient had severe hemophagocytic syndrome 1 month after an accidental needlestick injury. Metagenomic deep sequencing identified Candidatus M. haemohominis and determined its draft genome for an isolate from serum of the patient. A high copy number of the Candidatus M. haemohominis genome was detected in serum and bone marrow samples. Electron microscopy examination showed morphologic characteristics of Candidatus M. haemohominis. Levofloxacin monotherapy induced resistance caused by a gyrase A gene mutation in the quinolone resistance–determining region, but a combination treatment with moxifloxacin and minocycline was effective. We identified Candidatus M. haemohominis in a patient who had life-threatening symptoms related to multiple organ infection. Human infection with this mycoplasma might occur more frequently than has been generally recognized.

admitted to the hospital because lymphadenopathy, hepatosplenomegaly, and pyrexia developed.
Test results were negative for hepatitis A, B, C, and E viruses; measles virus; rubella virus; parvovirus; and HIV. Results of antinuclear antibody (titer 1:80) and smooth muscle antibody (titer 1:40) tests were positive, but test results for antimitochondrial M2 antibody, mitochondrial antibody, double-stranded DNA antibody, and lupus anticoagulant were negative. Epstein-Barr virus (EBV) DNA load determined by PCR was 3.0 × 10 2 copies/mL. However, Southern blot hybridization did not detect clonality of EBV-infected cells. Levels of herpes simplex virus, human herpesvirus 6 and 8, varicella zoster virus, and cytomegalovirus were below reference values, as determined by PCR. The serum IgG level was 3,967 mg/dL (reference range 800-1,750 mg/dL). However, the result of a serum-free light chain test was within the reference limit.
Our initial diagnosis was hemophagocytic syndrome (HPS) related to undetermined disease, and the patient was given steroid pulse therapy (1,000 mg) on day 1. Because his symptoms persisted, he was given etoposide (100 mg/m 2 ), cyclosporine (5 mg/kg), and dexamethasone (20 mg) on day 7. After this treatment was started, whole-genome sequencing of a peripheral blood sample detected a novel hemotropic Mycoplasma sp.
On day 11, we found a prolonged activated partial thromboplastin time (50.6 s, reference range 25-45 s), normal prothrombin time (12.6 s), low coagulation factor VIII activity (1.8%, reference range 78%-165%), and low von Willebrand factor (VWF) activity (ristocetin cofactor <10%, reference range 50%-150%). Moreover, serum levels of fibrinogen, antithrombin III, and disintegrin and metalloproteinase with thrombospondin type 1 motif, member 13 were within reference ranges. Microscopic examination of a Giemsa-stained blood smear showed coccoid forms on the erythrocyte surface ( Figure 3, panel A). A bone marrow aspirate showed hemophagocytosis and increased levels of reactive plasma cells (19.5%) ( Figure 3, panels B, C). On day 12, hypoglycemia (glucose level <5 mg/dL, reference range, 61-139 mg/dL) was noted in blood from the collection tube used for serologic tests, but the glucose level was within the reference range for blood in a container that contained citric acid with NaF.
On day 13, the case-patient received a diagnosis of Candidatus M. haemohominis infection-associated HPS and was given levofloxacin. Hypoglycemia improved after administration of levofloxacin. However, pyrexia and anemia developed, and coccoid forms on erythrocytes in peripheral blood and bone marrow were again observed on day 37. On day 46, he was given moxifloxacin and minocycline because of suspected bacterial resistance to levofloxacin. After this treatment was initiated, his symptoms, such as pyrexia, anemia, and hypoglycemia, promptly resolved, and the patient was discharged on day 61. One year after discontinuation of treatment, the patient remained well, and no Candidatus M. haemohominis DNA was detected in his serum.

High-Throughput Unbiased RNA Sequencing
We purified total RNA from a patient serum sample by using the RNeasy Mini Kit (QIAGEN, https://www. qiagen.com). We prepared RNA-Seq libraries by using the ScriptSeq version 2 RNA-Seq Library Preparation Kit (Illumina, https://www.illumina.com) and sequenced these libraries as single-end 151-mers by using the NextSeq 500 sequencer (Illumina) (13,14).

Whole-Genome Analysis
We performed metagenomic short-read DNA sequencing by using NextSeq 500 (Illumina) with DNA extracted from serum specimens. We excluded human-related short reads by using a Burrows-Wheeler Mapping Program with default parameters against human genome sequences (GRCh38.p13) (15).  We obtained a draft genome sequence of Candidatus M. haemohominis by de novo assembly using metagenomic DNA-Seq short reads. First, we excluded human genomic DNA sequences (≈58.3%) by using readmapping analysis, followed by de novo assembly with the remaining short reads. The total number of contigs was 703, and total length was ≈1.42 Mb, suggesting that whole contigs included bacterial and human-related DNA sequences. Therefore, to extract the bacterial sequences, we considered read depth and coverage, % GC of each contig, and a blastn search (https://blast.ncbi. nlm.nih.gov). A total of 23 contigs showed a marked read depth coverage (average × 5,500) and 30% GC content, although blastn search analysis showed that 4 contigs were assigned to the Mycoplasma genome and 19 contigs were assigned to unknown sequences.
We fixed tissue samples with 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 mol/L phosphate buffer (pH 7.4) for 2 h at room temperature, postfixed these samples in 1% osmium tetroxide, and embedded them in Epon resin. We stained ultrathin sections with uranyl acetate and lead citrate and observed them under a transmission electron microscope (HT7700; Hitachi) at 80 kV.

Real-Time PCR
We used a DNA fragment of the 16S rRNA gene of Candidatus M. haemohominis in a Taqman real-time PCR (16).

Sequence Data
We deposited metagenomic short-read sequences for DNA-Seq in the DDBJ (BioProject PRJDB7871; BioSample SAMD00156464; DRR accession no. DRR164892). We deposited the draft, annotated genome sequence of Candidatus M. haemohominis SWG34-3 in the DNA Data Bank of Japan (accession nos. SAMD00156495 and BIMN01000001-23). The 16S rRNA sequence was deposited under accession no. MHSWG343_r0010. The predicted gyrase A gene nucleotide sequence was deposited under accession no. MHSWG343_02220, and the coding sequence was deposited under accession no. GCE63237.1.

Identification and Whole-Genome Sequencing of Candidatus M. haemohominis
Metagenomic deep RNA sequence analysis strongly suggested that rather than other pathogens or virus infections, the Mycoplasma spp. could be associated with signs and symptoms of the case-patient. RNA-Seq short reads related to the Mycoplasma spp. were increasingly detected in the serum of the case-patient (Figure 4). A total of 41.7% of human-unmapped reads were extracted and then subjected to de novo assembly to identify the mycoplasma draft genome (23 contigs, 967,846 bp) ( Figure 5, panel A). The draft genome sequence identified a potential pathogen that could be Candidatus M. haemohominis on the basis of 16S rRNA sequence homology. Also, this pathogen showed similarity with closely related species, such as M. haemofelis and M. haemocanis ( Figure 5, panel B).
Negative-staining and electron microscopy of Candidatus M. haemohominis particles identified ribosomes, DNA, and soluble RNA in the cytoplasm ( Figure 6, panel A). In situ hybridization identified Candidatus M. haemohominis on the surface of erythrocytes ( Figure 6, panel B) and in the cytoplasm of macrophages in bone marrow ( Figure 6, panel C).

Monitoring and Treatment for Candidatus M. haemohominis Infection
We determined the level of Candidatus M. haemohominis DNA in serum by using real-time PCR (Figure 7). A high copy number for Candidatus M. haemohominis DNA was detected 18 days after the accidental needlestick injury. The bacterial load in serum decreased below the detection limit (<10 copies/reaction) 14 days after the patient was given levofloxacin, but again increased. Metagenomic analysis also indicated a high detection rate for Mycoplasma spp. reads in serum before and after treatment with levofloxacin. Read-mapping analysis showed that only 2-nt mutations were identified in samples after treatment with levofloxacin. Moreover, 1 of 2 mutations was a nonsynonymous mutation in the quinolone resistance-determining regions (QRDR) of the DNA gyrase subunit A GyrA (i.e., Gly95Cys, corresponding to aa 81 by numbering for Escherichia coli) (Figure 8). The bacterial load promptly decreased after combination therapy with moxifloxacin and minomycin. We also identified Candidatus M. haemohominis by using realtime PCR and in situ hybridization with samples from a liver biopsy of the patient.

Discussion
Infection with Candidatus M. haemohominis needs to be distinguished from other hemoplasmas in 16 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 26, No. 1, January 2020  terms of clinical symptoms. As described previously, infection with possible hemoplasmas results in hemolytic anemia and pyrexia (6)(7)(8)(9)(10)(11) We suggest that our results will help in diagnoses of Candidatus M. haemohominis infections, enabling early therapeutic intervention that might cure patients with these infections. Candidatus M. haemohominis patients are highly susceptible to misdiagnosis with other diseases, such as collagen diseases (e.g., systemic lupus erythematosus) and virus infections (e.g., severe fever with thrombocytopenia syndrome and EBV-associated HPS), because of the wide range of complications.
Generally, antimicrobial drug therapy with fluoroquinolones or tetracyclines is effective against hemoplasmas in animals. Combination therapy (9) or sequential treatment (19) with these drugs can be more effective against hemoplasmas because a single agent is often insufficient for consistent elimination of bacteremia (19). Moreover, detection of mutations in QRDR of gyrase A (Gly95Cys) suggested that Candidatus M. haemohominis is resistant to older fluoroquinolones (e.g., levofloxacin) (20,21). In contrast, newer fluoroquinolones, such as moxifloxacin or sitafloxacin, and tetracyclines, are effective against mycoplasmas harboring this mutation (21).
Laboratory tests for our case-patient showed low blood glucose levels (<5 mg/dL, reference range 61-139 mg/dL) but related no clinical signs. Hemoplasma-associated hypoglycemia in the absence of any associated clinical signs has been described in various animal species (22). This phenomenon is proportional to the severity of bacteremia and depends on the hemoplasma species (22). Because it is possible that glucose would be useful only for carbohydrate metabolism in the hemoplasma species (23), this finding might cause attachment in the epierythrocytic environment, which is rich in glucose, and result in alternative energy source pathways that become redundant.
We observed moderate to severe hemorrhagic episodes for this case-patient. Although the pathogenesis of bleeding in patients infected with Candidatus M. haemohominis is not completely understood, a dramatic decrease in coagulation factor Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 26, No. 1, January 2020 VIII and VWF activities was observed in this casepatient, which is similar to that observed in patients with acquired von Willebrand syndrome (24,25). Given the hypergammaglobulinemia and increased plasma cell levels observed in this case-patient and other patients. the possible mechanisms of infection are adsorption of vWF onto plasma cells or activated platelets apart from the presence of vWF-specific antibodies and inhibitors (24,25).
In conclusion, we identified and characterized Candidatus M. haemohominis infection in a human.