First Report of Bartonella spp. in Marsupials from Brazil, with a Description of Bartonella harrusi sp. nov. and a New Proposal for the Taxonomic Reclassification of Species of the Genus Bartonella

The genus Bartonella (Rhizobiales: Bartonellaceae) encompasses facultative intracellular Gram-negative alphaproteobacteria that parasitize mainly erythrocytes and endothelial cells, as well as macrophages, monocytes and dendritic cells. Although they can infect numerous mammal species and arthropod vectors worldwide, reports of Bartonella infections in marsupials are scarce. In fact, such agents have only been detected in marsupials and/or associated ectoparasites in Australia and the United States of America until the present moment. The present study aimed to isolate and characterize molecularly, morphologically and phenotypically Bartonella infecting free-living marsupials sampled in the Brazilian Pantanal, the largest wetland in South America. Two marsupials were captured in December 2018 and six marsupials in February 2019, totaling eight small mammals sampled: five (62.5%) Thylamys macrurus and three (37.5%) Monodelphis domestica. All blood samples were submitted to qPCR for Bartonella spp. based on the nuoG gene, a pre-enrichment liquid culture and a chocolate agar solid culture. Bartonella sp. was isolated from 3 T. macrurus and one M. domestica. One Bartonella isolate obtained from a T. macrurus blood sample (strain 117A) that showed to be closely related to the Bartonella vinsonii complex and Bartonella machadoae was selected for whole genome sequencing using a hybrid approach based on Illumina NovaSeq and Nanopore sequencing platforms. This strain showed a genome of 2.35 Mbp, with an average C + G content of 38.8%, coding for 2013 genes, and a 29 kb plasmid with an average C + G content of 34.5%. In addition, this strain exhibited an average nucleotide identity (ANI) of 85% with Bartonella species belonging to the B. vinsonii group and 91% with B. machadoae. Phylogenomic analysis based on 291 protein coding genes shared by the genomes of 53 Bartonella species positioned this strain closely to B. machadoae. This new isolated species was named Bartonella harrusi sp. nov., which was characterized as having small capnophilic, microaerophilic and aerobic rods with an absence of pili and flagella. In conclusion, the present work describes the biochemical, phenotypic and genomic characteristics of Bartonella harrusi, a new species isolated from the T. macrurus blood samples of the Brazilian Pantanal. Finally, a review of the taxonomic classification of members of the genus Bartonella is proposed, based on the ANI values accessed by whole genome sequencing analyses.


Introduction
The Bartonella genus comprise Gram-negative, intracellular facultative and fastidious α-2-Proteobacteria [1,2]. These bacteria are known to infect a wide range of vertebrates worldwide, mostly mammals, and may cause different clinical manifestations depending on the host and species of Bartonella involved [3]. These agents are transmitted by hematophagous arthropods (fleas, ticks, mosquitoes, flies and lice) [4,5].
Currently, this genus comprises 37 species, among which some are represented by more than one subspecies, validated according to the "List of Prokaryotic names with Standing in Nomenclature database" (https://lpsn.dsmz.de/genus/bartonella-accessed on 1 June 2022) [6][7][8][9][10]. The vast majority of the validated Bartonella species were submitted to biochemical, phenotypic and molecular characterization by DNA-DNA hybridization techniques or whole genome sequencing (List of Prokaryotic names with Standing in Nomenclature database). In view of the availability of collections of prokaryotic genomes, bioinformatics approaches have become robust tools to trace taxonomic and phylogenetic relationships between organisms.
Marsupials are viviparous mammals belonging to the Infraclass Marsupialia with the completion of development in the marsupial [11]. In Brazil, 58 species of marsupials have been described, most of them are small in size and with mainly arboreal habits and can be found in all Brazilian biomes (https://www.icmbio.gov.br/portal/faunabrasileira/ estado-de-conservacao/2802-mamiferos-marsupiais-accessed on 1 June 2022). Although Bartonella spp. is widely studied in felids, rodents and bats, there are few studies on the detection, molecular characterization and/or isolation of Bartonella in marsupials. In fact, to date, only one species of Bartonella (Bartonella australis) has been isolated from kangaroos (Macropus giganteus) in Australia [12]. Additionally, a Bartonella sp. phylogenetically related to B. tamiae was detected in Ixodes tasmani ticks collected from koalas (Phascolarctos cinereus) in Australia [13]. Additionally, 'Candidatus Bartonella bandicootii' and 'Candidatus Bartonella woyliei' were detected in fleas (Pygiopsylla tunneyi) of "bandicoots" (Perameles bougainville) and fleas (Pygiopsylla hilli) and ticks (Ixodes australiensis) collected from woylies (BIettongia penicillata), respectively [7]. In the Americas, Bartonella spp., closely related to B. henselae and B. clarridgeiae, were molecularly detected in Ctenocephalides felis fleas collected from opossums (Didelphis virginiana) in the USA [14,15]. In Brazil, scarce are the studies on the occurrence of Bartonella spp. in marsupials. In fact, only three studies were found in the literature regarding the molecular detection of Bartonella spp. in this group of mammals. Based on conventional and real-time PCR assays, Bartonella DNA was not found in marsupials' blood samples in the states of Mato Grosso do Sul and Rio de Janeiro [16][17][18].
The present study aimed to detect and characterize, using microbiological, molecular and bioinformatics analyses, Bartonella sp. in blood samples from wild marsupials sampled in the Pantanal biome, a flooded area in central-western Brazil. Additionally, a new isolate of Bartonella from Thylamys macrurus was characterized, using morphological, biochemical, phenotypic and molecular characters. Finally, by using bioinformatics analyses of the whole genomes of the Bartonella species deposited in international databases, the present study aimed to contribute to the re-assessment of the separation of the Bartonella species using ANI values, as well as to infer differences in the metabolic profile of this group of agents.

Ethics Statement
The procedures and management protocols involving the capture and blood sampling of wild marsupials were approved by the "Chico Mendes Institute for Biodiversity Conservation" (SISBIO number 64204) and the "Ethics Committee on Animal Use of the Faculty of Agrarian and Veterinary Sciences" (FCAV/UNESP) (protocol number #11794).

Area of Study and Collection of Samples
Wild marsupials were captured during two expeditions in the Pantanal of the state of Mato Grosso do Sul, central-western Brazil, more specifically in the "Alegria" farm, Nhecolândia region (−19.050399; −56.675623): one in December 2018 and the second in February 2019. The procedures performed to obtain biological samples of the small mammals were described in a previous work [10].
The blood samples collected from the marsupials were stored in DNAse/RNAse-free microtubes containing ethylenediaminetetraacetic acid (EDTA) and were kept in liquid nitrogen cylinders. Once in the laboratory they were stored in a freezer −80 • C (Figure 1).

Bartonella Isolation
A pre-enrichment liquid medium previously described as Bartonella-Alphaproteobacteria Growth Medium (BAPGM) was used in the present study. Briefly, 500 mL of IPL-41 insect medium (Sigma-Aldrich, St. Louis, MO, USA) were supplemented as previously described [10,[19][20][21]. Two vials of negative controls were also added: one containing only the pre-enrichment medium and the other one containing both the preenrichment medium and sheep defibrinated blood [10,21].
After seven days, 200 µL of liquid culture were submitted to DNA extraction with the Illustra Tissue and Cells genomic Prep Mini Spin kit, following the manufacturer's recommendations. Additionally, 200 µL of the contents of each bottle were seeded onto an enriched chocolate agar (Laborclin, Pinhais, Paraná, Brazil) and kept at 35 • C with 5% CO 2 for up to 60 days in a CO 2 /O 2 Water Jacketed incubator (NuAire, Plymouth, MA, USA) and observed daily. Three negative controls were used: two corresponding to the negative controls previously used in the liquid culture of pre-enrichment and the other represented by a chocolate agar plate without any biological samples. Smooth and circular colonies suggestive of Bartonella were collected and submitted to DNA extraction by the boiling method [22] and qPCR assay for Bartonella sp. based on the nuoG gene [23]. Colonies molecularly confirmed as belonging to the genus Bartonella were then submitted to five passages until pure isolates were obtained.

Quantitative
Real-Time PCR Assay (qPCR) for Bartonella sp. Based on the nuoG Gene (Nicotinamide Adenine Dinucleotide Dehydrogenase Gamma Subunit) and Conventional PCR (PCR) for the Mammalian Endogenous Gene DNA from the marsupial blood samples was extracted using the InstaGene™ Matrix kit (Bio-Rad Laboratories, São Paulo, Brazil), according to the manufacturer's recommenda-tions. A PCR assay based on the mammalian endogenous glyceraldehyde gene 3-phosphate dehydrogenase (gapdh) [24] was used to verify the absence of inhibitors in the DNA samples extracted. A quantitative PCR protocol (qPCR) based on the nuoG gene [23] was used to detect and quantify Bartonella sp. DNA (the number of copies of a 83 bp fragment of nuoG gene/µL) in the marsupial blood samples, marsupial blood samples added to the BAPGM-based liquid culture and colonies isolated from the chocolate agar plates. PCR assays were conducted on low profile multiplate PCR plates (BioRad™, Hercules, CA, USA) using CFX96 thermocycler (BioRad™, Hercules, CA, USA). The standard curves were constructed with serial dilutions of plasmid DNA (pIDTSMART-Integrated DNA Technologies) (1.0 × 10 7 to 1.0 × 10 0 copies/µL), which encoded an 83 pb fragment of the Bartonella henselae nuoG gene. The number of copies of the plasmid was determined by (Xg/µL DNA/[plasmid length in pb × 660]) × 6.022 × 10 23 × plasmid/µL copies [23] All the DNA samples were initially tested in duplicates. All duplicates whose difference in Cq (the cycle of quantification) values was higher than 0.5 were retested in triplicates.
2.6. Concatenated Phylogenetic Analysis Based on the 16S rRNA, gltA, groEL, ftsZ, rpoB Genes and ITS Intergenic Region Products amplified in conventional PCR assays were purified using ExoSAP-IT (Thermo Fisher Scientific, Waltham, MA, USA) and sequenced using the BigDye™ Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific™, Waltham, MA, USA) and the ABI PRISM 310 DNA Analyzer (Applied Biosystems™, Foster City, CA, USA) [28]. The primers used in the sequencing reactions were the same used in the conventional PCR assays for Bartonella sp. The sequences obtained were first submitted to a screening test using the Phred-Phrap software version 23 [29,30] to assess the quality of the electropherogram and to obtain consensus sequences from the alignment of the forward and reverse sequences.
The BLAST program (BLASTn) [31] was used to compare the obtained nucleotide sequences with those from the international database (GenBank) [32]. The consensus sequences obtained in this study and those retrieved from GenBank were aligned using the Clustal/W software [33] via Bioedit v. 7.0.5.3 [34]. Phylogenetic analysis was based on the Maximum Likelihood (ML) method and was inferred with the W-IQ-Tree tool available online (http://iqtree.cibiv.univie.ac.at/-accessed on 12 December 2021) [35] using 1000 bootstrap replicates. The best evolutionary model was selected by the jModel-Test2 program (version 2.1.6) in XSEDE [36]. The tree was edited in Treegraph 2.0.56-381 beta [37].
In addition, nucleotide sequences were aligned with sequences previously deposited in GenBank and were subjected to a Split-Network-based distance analysis using SplitsTree v4.11.3 software [38].

Light Microscopy and Scanning Electron Microscopy (SEM)
The Bartonella strain 117A was submitted to light microscopy and scanning electron microscopy (SEM). For light microscopy, the isolate was stained with Gram (Hy-labs) and visualized using 100× objectives. For SEM, the Bartonella isolate (strain 117A) was collected from the chocolate agar plate and fixed in a solution containing 2% paraformaldehyde, 2.5% glutaraldehyde and 0.1 M cacodylate buffer for 1 h and then washed three times in a phosphate buffer. Then, 1% osmium tetroxide was added and the solution was incubated for 30 min. The bacteria were then dehydrated through a graded ethanol series (50-100%) and dried with a critical point dryer (EMS 850). After the critical point, the sample was fixed in metal stubs, and coated with gold (5 nm) in a Danton Vacuum Desk II. Finally, the bacterial cells were visualized by SEM. The samples were analyzed and photographed using a Zeiss Evo 10 scanning electron microscope (10-20 kV).

Biochemical and Metabolic Characterization
For biochemical and metabolic characterization, the Bartonella strain 117A isolated from T. macrurus was tested with the identification systems BBL Crystal Enteric/ Nonfermenter (BD Becton Dickinson, Franklin Lakes, NJ, USA), API 20E and NF III (Probac), following the manufacturers' instructions. The enzymatic catalase reaction was carried out with hydrogen peroxide (3%).

Growth Curve
To perform the growth curve of the Bartonella isolated from the marsupial (strain 117A), 10 colony-forming units (CFU) were removed from the chocolate agar plates and deposited into 500 mL of a supplemented BAPGM medium. The inoculum was then maintained at 35 • C with 5% CO 2 in a CO 2 /O 2 culture incubator (Water Jacketed, NuAire, Plymouth, USA). Every 6 h, 100 µL were collected for 8 days, totaling 32 collections. This volume was subjected to a 9:1 serial dilution in a supplemented BAPGM medium (10 −1 to 10 −6 ). One hundred microliters of each of the six dilutions were plated onto chocolate agar plates with a Drigalski loop. Bacterial quantification was performed using the standard plate counting method, determining the number of colony-forming units (CFU/mL). Each plate was stored at 35 • C with 5% CO 2 in a CO 2 /O 2 culture incubator (Water Jacketed, NuAire). The incubation period of the plates was 72 h and, after this period, colony-forming units per mL (CFUs/mL) were counted.

Whole Genome Sequencing of Bartonella Isolated from Blood Samples of Thylamys Macrurus (Strain 117A) and Genome Annotation
The genomic DNA extraction from the Bartonella isolate of T. macrurus (strain 117A) previously grown on chocolate agar plates was performed using the commercial kit DNeasy ® Blood and Tissue (QIAGEN, Hilden, Germany), following the manufacturers' instructions. For genome assembly, the genomic DNA sample was sequenced by Illumina Novaseq 6000 (PE 2 × 150 bp). A total of 5,871,200 paired-end reads were obtained. These reads were trimmed with Trimmomatic [39]. In addition, FASTQ files were also obtained from a MinION run (kit SQK-LSK109, Oxford Nanopore, Oxford, UK), totaling 118,191 reads, with an average read length of 5471 bp. The readings were clipped to a low quality and to adapters using NanoFilt and Porechop, respectively [40]. All the trimmed Illumina and Nanopore reads were then used for hybrid assembly using the MaSuRCa v4.0.4 assembler with default parameters [41]. The assembled genome was automatically annotated with the NCBI Prokaryotic Genome Annotation Pipeline [42].

Phylogenomic Analyses
The assembled genome of the new Bartonella isolate (strain 117A) and 53 complete Bartonella genomes retrieved from the NCBI RefSeq database [43] were used for the comparative analysis. For this purpose, 291 shared genes were retrieved using the roary tool [44]. These genes were aligned with MAFFT [45] and were later used for phylogenomic analyses.
The average nucleotide identities (ANI) between Bartonella harrusi sp. Nov., Bartonella machadoae and the species of the Bartonella vinsonii complex (B. vinsonii subsp. Vinsonii, B. vinsonii subsp. Arupensis and B. vinsonii subsp. Berkhoffii) were calculated with OrthoANIu 1.2 [46]. In addition, an ANI analysis was performed between the nucleotide sequence of the plasmid found in Bartonella harrusi compared with eight other plasmid sequences from the Bartonella species deposited in GenBank.

Taxonomic Reclassification of Bartonella Species Based on ANI
The whole genome consensus sequence of the new Bartonella isolate (strain 117A) and the 216 whole genomes of the Bartonella species retrieved from the NCBI RefSeq database [43] were submitted to the dRep pipeline [47]. After filtering by length, the completeness and contamination of the sequences were analyzed using CheckM [48]. Primary and secondary clustering was performed using pairwise MASH, in order to calculate the ANI. As ANI cut-off parameters, 95% and 97% were used for primary and secondary clustering, respectively, aiming at separating the species and assigning objective taxonomic classifications to be applied to all the available genomes of the Bartonella species.

"In Silico" Analysis of the Metabolic Profile and Ribosomal and Polymerase Phylogenetic Markers for Bartonella Species
The assembled whole genome of the new Bartonella isolate (strain 117A) and 53 whole genomes of the Bartonella species retrieved from the NCBI RefSeq database [43] were aligned by the RASTtk (Rapid Annotations using Subsystems Technology) pipeline [49]. After annotation, a GenomeSet was generated, which was used to analyze the metabolic profile and search for phylogenetic markers via the HMMER tool available online (https: //www.kbase.us/-accessed on 12 December 2021) [50].

Search for Active Carbohydrate Enzymes (CAZy) Families
A search for the occurrence of carbohydrate active enzymes (CAZy) families was performed. For this purpose, the GenomeSet described in the section above was submitted to the dbCAN database for the automated annotation of the active carbohydrate enzymes [51] using HMMER 3 via the HMMER tool available online (https://www.kbase.us/-accessed on 12 December 2021) [50].

Occurrence of Bartonella sp. in Marsupial Blood Samples
Two marsupials were captured in December 2018 and six marsupials in February 2019. Among the eight marsupials sampled, five (62.5%) animals belonged to the species Thylamys macrurus and three (37.5%) to the species Monodephis domestica. All eight marsupial blood DNA samples were positive in the PCR for the endogenous mammalian gapdh gene. None of the eight DNA samples extracted from marsupial blood were positive in the qPCR for Bartonella sp. based on the nuoG gene. The efficiency, R 2 , slope and intercept on the Y axis of the reactions ranged from 91.4% to 103.9%, 0.995 to 0.999, −3.547 to −3252 and 38.571 to 40, respectively (Table 1).

Isolation of Bartonella sp. in Chocolate Agar
Out of the eight marsupial blood samples submitted to the pre-enrichment liquid BAPGM culture, four (50%; two T. macrurus blood samples and two M. domestica blood samples) were positive in the qPCR for Bartonella sp. The number of copies of the nuoG gene fragment per microliter ranged from 5.9 × 10 2 to 2.93 × 10 6 . The efficiency, R 2 , slope and intercept on the Y axis of the reactions ranged from 91.4% to 103.9%, 0.995 to 0.999, −3.547 to −3252 and 38.571 to 40, respectively.
Out of the four qPCR-positive liquid culture samples, three (45.16%) (two T. macrurus blood samples and one M. domestica blood sample) showed the presence of Bartonella colonies on chocolate agar plates, whose identity was confirmed by qPCR for Bartonella spp. based on the nuoG gene (Table 1).

Multilocus and Phylogenetic Analyses for Bartonella sp.
A BLASTn analysis demonstrated that the different target genes of the Bartonella colonies isolated from marsupials showed high identities to B. machadoae and Bartonella genotypes previously reported in ectoparasites collected from rodents from Brazil (Table 2). The phylogenetic tree based on the sequences of the 16S rRNA, gltA, groEL, ftsZ, rpoB genes and ITS intergenic region of three Bartonella isolates from one T. macrurus individual (strains 117A, 117B and 117C) and one from M. domestica (strain 78A) inferred by the Maximum Likelihood method (ML) and evolutionary model TVM + F + I formed a single and monophyletic cluster. The sequences of Bartonella isolated from marsupials were positioned together with sequences previously detected in Polygenis

Diversity Analyses
The nucleotide sequences (16S rRNA, gltA, groEL, ftsZ, rpoB and ITS) obtained from four Bartonella isolates from two marsupials, one from the T. macrurus species (strains 117A, 117B and 117C) and one from the M. domestica species (strain 78A), were identical to each other. The genealogical analysis of the nucleotide sequences by Network, using the program Splitstree v4.11.3 [38], showed a similar result to that obtained by the phylogenetic analysis, forming three relevant groups. Cluster #1, consisting of sequences from Bartonella isolates of T. macrurus and M. domestica, was related to Bartonella genotypes previously detected in rodent-associated fleas in the Pantanal (MKY304482; KF021604). Cluster #2 consisted of Bartonella genotypes previously detected in rodents from the Brazilian states of Mato Grosso do Sul, Maranhão, Ceará and Rio de Janeiro (KX578719, KX578720, KX086721, KX270253, MN613442, KX270232, MN613441, KX270248), rodents from Peru (KF021604) and B. machadoae. Cluster #3 contained Bartonella genotypes previously detected in rodents from Peru and the USA (Figure 3).

Phenotypic and Morphological Characterization of Bartonella sp.
The Bartonella 117A strain isolated from a blood sample of T. macrurus was characterized by Gram-negative rods 1-2.0 µm long and 0.3-0.6 µm wide, without flagella and pili. The bacteria proved to be capnophilic, microaerophilic and aerobic, since there was no growth under anaerobic conditions. In SEM, the bacteria appeared as small rods without polar pili and flagella, forming compact aggregates of bacteria (Figures 4 and S1).

Biochemical and Metabolic Characterization
The Bartonella strain 117A obtained from the blood sample of T. macrurus was negative for oxidase, catalase, urease, indole production, nitrate reduction and acetoin production (Voges-Proskauer reaction). It was inert for all the tested carbohydrates (glucose, mannitol, inositol, sorbitol, rhamnose, sucrose, melibiose, amygdalin and arabinose) and negative for arginine, lysine and tryptophan arylamidase activities. Furthermore, it was negative for esculin, DNAse activity, β-D-galactosidase nitrate and McConkey growth. It showed sensitivity to Polymyxin B. 3.6.

Growth Curve
The B. harrusi growth curve showed moderately slow growth in the supplemented BAPGM medium. The bacteria entered the logarithmic phase 30 h after the inoculum preparation and maintained an average quantification of 10 7 CFU/mL until 90 h of the experiment. After this period, the bacteria entered the decline phase ( Figure 5).

Phylogenomic Analysis
The genome was assembled into a 2.23 Mbp circular chromosome and a 29 Kbp circular plasmid. The genomic characteristics of the Bartonella strain 117A are described in Table 3. The whole genome ANI of strain 117A was calculated in relation to phylogenetically related species: ANI values of 85% and 91.98% were found when compared to Bartonella species belonging to the B. vinsonii complex (B. vinsonii subsp. vinsonii, B. vinsonii subsp. arupensis and B. vinsonii subsp. berkhoffii) and B. machadoae, respectively ( Figure 6). In the ANI analysis of the plasmid sequences of the Bartonella species available in GenBank, the Bartonella strain 117A showed 81.45% when compared to Bartonella krasnovii, which was the highest value found. Plasmids of the Bartonella sp. strain HY038 and the pMS plasmid from B. schoenbuchensis did not show any phylogenetic relationship with the plasmids found in other Bartonella species (Figure 7).     The phylogenomic analysis was based on 291 protein-coding genes shared by 53 whole genomes of Bartonella species with validly published names and some Bartonella species not yet named. Based on this analysis, the Bartonella strain 117A was positioned in a clade closely related to B. machadoae and Bartonella species belonging to the B. vinsonii complex, albeit separated with 100% bootstrap (Figure 8).

Use of Average Nucleotide Identity (ANI) to Separate Species of the Genus Bartonella
The Bartonella genomic sequences available in Genbank were analyzed for completeness and contamination in order to determine the quality of the genomes for use in ANIbased taxonomic classification. Among the 217 genomes available, only the genome of 'Candidatus Bartonella saheliensis' (LR607310) presented a low quality both for completeness (below 70%) and percentages of contamination (above 15%) ( Figure 9A,B), and it was therefore removed from subsequent analyses. Regarding the mean nucleotide identity (ANI) in general, most Bartonella species showed ANI values between 75 and 90%. However, ANI values lower than 70% were found for the B. tamiae, B. apis and Bartonella sp. strain HY038, species considered more diversified within this genus when compared to the other Bartonella species analyzed ( Figure S2). Furthermore, it was possible to observe ANI values greater than 95% among the available Bartonella genomes, which were divided into 50 different groups ( Figure S2 and Table S1). Among these groups, 20 were represented by a single sequence and 30 had more than one sequence with a variation between 95 and 100%. Genomes with more than one sequence were subdivided using cut-off points 95 < ANI < 97%. Eighteen groups presented ANI values above 97% when compared to the other genomes; 12 genomes had one to four sequences with variations between 95 to 97% of ANI, and they were reclassified as subspecies (Table S1).

"In Silico" Analysis of Metabolic Profiles, Ribosomal and Polymerase Phylogenetic Markers for Species of the Genus Bartonella
The analysis of the metabolic profile for species of the genus Bartonella using the genome available in Genbank showed similar profiles: all species analyzed (n = 53) had the capacity to oxidize hydrogen and oxygen, as well as to fix carbon. Some Bartonella species (strain 117A, B. machadoae, B. tribocorum, B. schoenbuchensis, B. massiliensis, B. capreoli, B. ancashensis, B. bovis, B. bacilliformis, B. rattimassiliensis and B. melophagi) were not able to oxidize sulfur because they did not have related genes. The most diverse species found were the Bartonella sp. Strain HY038 and Bartonella apis, which presented in their genomes genes related to nitrogen fixation, the oxidation of urea, selenium, arsenic and halogenated compounds, a profile that was not observed in any other Bartonella species. Interestingly, the Bartonella sp. strain HY038, Bartonella apis and Bartonella tamiae showed genes related to the use of C1 compounds ( Figure S3).
Regarding the profile of ribosomal phylogenomic markers and polymerases, only punctual differences were found. In fact, none of the analyzed species showed considerable differences ( Figure S4).

Search for Active Carbohydrate Enzymes (CAZy) Families
When analyzing genes related to groups of enzymes, Carbohydrate Esterases (CAZy), enzymes classified into families 24, 94 and 103 of the group of glycoside hydrolases enzymes, families 19, 29, 30, 51 and 84 of the group of glycoside transferases, and family 11 of the group of carbohydrate transferases were found in all the analyzed genomes ( Figure S5). However, genes related to the production of family 3 of enzymes of the glycoside hydrolases group were found only in some Bartonella species belonging to lineages 1 (B. tamiae, B. apis, B. ancachensis and the Bartonella sp. strain HY038), 2 (B. bovis, B. schoenbuchensis and B. chomelii) and 3 (B. clarridgeiae, B. rochalimae and the Bartonella spp. strains Coyote22, stunk WD16.2 and 1-1C JB63). Family 4 of the group of glycoside transferases was found only in a few species, including B. machadoae recently isolated from rodents in the same region where the present study was performed, but not in the Bartonella sp. strain 117A described in the present study ( Figure S5). Interestingly, although family 24 of glycoside hydrolases was found in the genome of the 53 Bartonella species analyzed, there was a difference in the number of occurrences depending on the analyzed genome, with variations from 1 to 12 genic regions ( Figure S5).

Discussion
The present work aimed to contribute to the diagnosis, molecular, morphological and phenotypic characterization of Bartonella spp. in wild marsupials from the Pantanal, state of Mato Grosso do Sul, central-western Brazil.
There are few works in the literature that have investigated the role of marsupials as reservoirs of Bartonella spp. For instance, Fournier et al. [12] isolated and described Bartonella australis in a kangaroo (Macropus giganteus) in Australia. Bartonella DNA was also detected in Ixodes tasmani ticks collected from koalas (Phascolarctos cinereus) [13] in Australia. Additionally, 'Candidatus Bartonella bandicootii' and 'Candidatus Bartonella woyliei' were molecularly detected in fleas (Pygiopsylla tunneyi) from bandicoots (Perameles bougainville) and in fleas (Pygiopsylla hilli) and ticks (Ixodes australiensis) collected from woylies (BIettongia penicillata), respectively [7]. When it comes to Brazil, Bartonella DNA was not detected in marsupials (Thylamys macrurus, Gracilinanus agilis, Monodelphis domestica and Didelphis albiventris) in the Pantanal of Mato Grosso do Sul [16]. This finding might be associated with the low Bartonella bacteremia, which might have hampered the detection of this agent by qPCR directly from the blood. The Bartonella bacteremia in the marsupials sampled by [16] was probably below the limit of detection of the qPCR technique based on the nuoG gene. Similar results were found herein when marsupial blood DNA samples were submitted to qPCR assays. In the present study, Bartonella DNA was found in marsupial blood samples only after passage in a BAPGM liquid medium (a pre-enriched liquid culture) and the isolation of Bartonella spp. on a chocolate agar. The present work brings the first detection and isolation of Bartonella in marsupials in the Americas.
The use of a BAPGM pre-enrichment medium and isolation on a chocolate agar brings greater sensitivity in the diagnosis of Bartonella spp., as previously reported in [19] where, when using the abovementioned diagnostic platform for Bartonella spp. in dog blood samples, they found a positivity of 8.1%; on the other hand, when they performed a molecular diagnosis directly from the blood DNA samples, a positivity of 4.7% was found. Our research group has been using this diagnostic platform and showing the advantages of using the "liquid medium" for the successful diagnosis and isolation of Bartonella henselae in blood samples from cats [21]. More recently, using the same diagnostic platform used in the present study, we reported the diagnosis and isolation of Bartonella machadoae from rodent blood samples sampled in the Pantanal of Mato Grosso do Sul [10].
Regarding the phenotypic characteristics, the strain 117A of Bartonella sp. isolated from a blood sample of T. macrurus showed microbiological and morphological characteris-tics identical to the B. machadoae recently described by our research group [10], showing excellent growth under conditions of increased CO 2 . Additionally, this strain showed aerobic, capnophilic and microaerophilic characteristics. It was characterized by rounded and smooth colonies (1-2 mm) after 4-5 days of plating on a chocolate agar. Similarly to B. machadoae, the strain 117A was not able to grow under anaerobic conditions. Such growth characteristics are similar to those reported in other species of the genus Bartonella, such as in B. elizabethae, B. henselae (the Houston-3 strain), B. clarridgeiae, B. quintana and B. vinsonii subsp. Vinsonii [52]. In contrast, B. kosoyi and B. krasnovii isolated from Rattus rattus and Synosternus cleopatrae fleas collected from Gerbillus andersoni, respectively, did not grow under strictly aerobic and anaerobic conditions [53].
The Bartonella strain 117A was negative for oxidase, catalase, urease, indole production, acetoin production and nitrate reduction. These characteristics were similar to those found in the biochemical profiles of the Bartonella species of the B. vinsonii complex [54] and B. machadoae [10]. However, differences were found with other Bartonella species: while B. krasnovii was able to ferment sucrose, citrate and arabinose, Bartonella kosoyi was able to ferment sorbitol, melibiose, rhamnose and glucose, in addition to showing a positive reaction to arginine [2]. On the other hand, but similarly to B. machadoae, Bartonella harrusi sp. nov. was not able to ferment the abovementioned carbohydrates [10].
Regarding the bacterial growth curve, the results of the present study showed discrepancies between the growth curves of the two isolated species: while B. machadoae (unpublished data) required a shorter time to enter into the logarithmic phase and presented a higher peak of CFU/mL, remaining longer in the stationary phase, the Bartonella strain 117A needed more time to reach the highest CFU/mL count, in addition to remaining in the stationary phase for a shorter time. When compared with other species of the genus, B. henselae (using a supplemented RPMI 1640 medium) and B. quintana (using a DS2 medium formulated to favor the cell growth of insect cells) showed curves similar to that presented by the Bartonella strain 117A, both requiring a longer time to enter into the logarithmic phase and, similarly to the present work, they did not remain in a stationary phase for a long time [55].
With regard to multilocus analysis (MLSA), do Amaral et al. [10] demonstrated that, due to the high genotypic diversity of the Bartonella species described in different hosts, previously used criteria, such as those proposed by la Scola et al. [62], have less accuracy and precision for the discrimination of strictly related Bartonella species. In the present work, it was possible to observe a lack of resolution of the phylogeny based on six molecular markers, which were able to separate the Bartonella sp. isolated from marsupials in the present work from the Bartonella species belonging to the B. vinsonii complex, but were not able to clearly separate them from B. machadoae, a rodent-associated Bartonella species recently described in the same region where the present study was performed [10] (Figure 2; Table 4). Phenotypic and biochemical analyses are shown to be inefficient in the discrimination of new species of Bartonella due to very similar biochemical profiles, with rare exceptions in more diversified species, as shown in the present study by "in silico" analyses. A Multilocus analysis based on 6-8 molecular markers was replaced by a WGS analysis, due to greater ease and more accessible values of WGS of the Bartonella isolates. In fact, phylogenomic analyses based on hundreds of genes provide higher reliability in the definition of species, in addition to allowing the inference of biochemical profiles without the use of timeconsuming and laborious techniques [10,63].
In the present study, a bioinformatics analysis applied to whole genomes of the Bartonella species deposited in databases allowed us to find a relative diversity in genes related to CAZy families, such as the presence of family 3 in some Bartonella species and the variations in the amount of occurrences of glycosyl hydrolases in family 24, enzymes known as β-N-acetylglucosaminidase and muramidase, respectively, and related to the degradation of chitin-based carbohydrates [64]. More studies related to CAZy families should be conducted, since chitinases may play a role in the interaction of bartonellae and the peritrophic membrane of ectoparasites, enabling bartonella mobility and favoring transmission. Furthermore, it was possible to identify genes related to family 4 glycosyltransferase enzymes in some species of Bartonella. Considering that such enzymes are associated with the accumulation of glycogen for the survival of the agent in the environment, they may be involved in the maintenance of some Bartonella species in flea and lice feces [65]. Interestingly, although no genes related to this family of enzymes were found in the genome of the Bartonella strain 117A isolated herein, such genes were found in the genome of B. machadoae, which was recently isolated from Trichomys fosteri' blood samples in the same region where the present study was performed [10]. Thus, the role of this enzyme family in the survival and transmission of Bartonella species between their vertebrate and invertebrate hosts should be further investigated.
Although the number of species and Candidatus for the genus Bartonella has increased in the last years, especially those associated with small mammals, many of them have not been fully characterized and, quite often, with insufficient information for the reliable determination of new species and subspecies [63]. In addition to demonstrating the inability to distinguish phylogenetically close species by the criteria previously used, the present work proposes a review of the taxonomic classification of species of the genus Bartonella by using ANI values obtained from whole genomes. ANI values below 95% are used as a cut-off point for determining new species [2,66,67]. Based on this criterion, it is possible to separate the genus Bartonella into 50 different species (Table S1). In addition, we were able to solve taxonomic issues in several subgroups of the genus Bartonella. For instance, the taxa Bartonella apis and Bartonella tribocorum are composed of two and three different species, respectively. Bartonella subspecies belonging to the Bartonella vinsonii complex actually comprise three different species (Bartonella vinsonii, Bartonella arupensis and Bartonella berkhofhii), since they have an ANI value below 95% among themselves. On the other hand, B. schoenbuchensis, Bartonella chomelii, B. capreoli and B. melophagi should be grouped as a subspecies of B. schoenbuchensis (the species described first among the four cited), since they show ANI values above 95% among themselves. Finally, based on the Bartonella species (B. henselae, B. quintana and B. bacilliformis) with the highest number of genomes deposited in the database, the variation of the ANI values within the same species ranged from 97 to 100%. Therefore, we propose a cut-off point of 95 < ANI < 97% for the classification of new Bartonella subspecies (Table S1).
The present work brings the first description of Bartonella in marsupials outside the Australian continent. We propose that it be named Bartonella harrusi sp. nov., in honor of Israeli researcher Shimon Harrus, professor at The Hebrew University of Jerusalem, Koret School of Veterinary Medicine, for his extensive contribution to the study of members of the Bartonella genus.
Despite the vectors involved in the transmission of this new Bartonella species are still unkown, siphonapterans may act as possible vectors of this agent. Indeed, gltA sequences from B. harrusi sp. nov. showed 100% identity with Bartonella genotypes previously detected in rodent-associated fleas (Polygenis (Polygenis) bohlsi bolshi, Polygenis occidentalis occidentalis, Polygenis platensis and Craneopsylla minerva minerva) in the biomes of Pantanal (the state of Mato Grosso do Sul), Pampa and the Atlantic Forest, in the state of Rio Grande do Sul [16,68]. Interestingly, the Bartonella genotypes detected in fleas from centralwestern and southern Brazil were found in regions where the circulation of B. machadoae, a species recently described in rodents, was also reported [10]. Recently, Santana et al. [69] detected B. machadoae in Amblyomma sculptum ticks collected from wild boars (Sus scrofa) in southeastern Brazil. Thus, it is necessary to evaluate the diversity of hosts, vectors and the zoonotic potential of B. machadoae and B. harrusi, considering that both are closely related to Bartonella species belonging to the B. vinsonii complex, which have already been shown to have zoonotic potential in several regions in the world [6,10,54,[70][71][72].
Description of Bartonella harrusi sp. nov.
The bacterial cells of this new proposed species are rods (1-2.0 µm long and 0.3-0.6 µm wide) without flagella and pili, showing excellent growth under capnophilic conditions (5% CO 2 ) at 35 • C. Colonies grown on chocolate agar are smooth and circular. The time required for growth to reach colonies of 1.0-2.0 mm is 4 days at 35 • C. They are negative for oxidase, catalase, urease, indole production, acetoin production and nitrate reduction. They are unable to ferment the carbohydrates sorbitol, melibiose, rhamnose and glucose, albeit they showed a positive reaction to arginine. The type strain (117A) carried a single circular chromosome of 2.23 MB, with an average G + C DNA content of 38.8% by mol. and a 29,892 bp circular plasmid with an average G + C content of 34.5%. The closest species with a valid published name is B. machadoae (91.98% ANI). Bartonella harrusi sp. nov. (strain 117A) is stored at the Laboratory of Immunoparasitology, Department of Pathology, Reproduction and Single Health, Faculty of Agrarian and Veterinary Sciences/Universidade Estadual Paulista, UNESP, Jaboticabal, SP, Brazil.

Conclusions
The present work presents the first description of Bartonella in marsupials from the Americas. Bartonella harrusi sp. nov. was isolated from blood samples of T. macrurus in the Pantanal of Mato Grosso do Sul. The diagnostic platform combining qPCR results based on the nuoG gene from blood samples in a pre-enrichment liquid culture and chocolate agar isolation proved to be efficient in detecting Bartonella in marsupials. A phylogenomic analysis based on 291 genes and an ANI value of 91.98% obtained by the WGS of Bartonella 117A allowed the separation of Bartonella harrusi sp. nov. from Bartonella machadoae. A bioinformatics analysis applied to complete genomes should be used for the classification and taxonomic review of members of the genus Bartonella based on ANI values.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/microorganisms10081609/s1. Figure S1. Gram-staining of Bartonella harrusi sp. nov. (strain 117A). Figure S2. Dendrogram created from the dRep analysis of all genomes deposited in GenBank using the 95% ANI value (average nucleotide identity) as a cut-off point by obtaining 50 different strains. Figure S3. Heat map generated in the search for genes related to metabolic profiles using the genomes of Bartonella species available in GenBank. Figure S4. Heat map generated in the search for genes related to ribosomal and polymerase phylogenomic markers for species of the genus Bartonella. Figure S5. Heat map generated in the search for families of active carbohydrate enzymes (CAZy) for species of the genus Bartonella. Table S1. Taxonomic reclassification of the Bartonella species based on the 95% cut-off point assessed by the Average Nucleotide Identity (ANI) value. Bartonella species with a new proposed taxonomic status are red marked. n/a: species that did not have ANI variation below the 97% cut-off.

Institutional Review Board Statement:
The procedures and management protocols involving the capture and blood sampling of wild marsupials were approved by the "Chico Mendes Institute for Biodiversity Conservation" (SISBIO number 64204) and the "Ethics Committee on Animal Use of the Faculty of Agrarian and Veterinary Sciences" (FCAV/UNESP) (protocol number #11794).

Data Availability Statement:
The data that support the findings of this study are openly available at the National Center for Biotechnology Information at https://www.ncbi.nlm.nih.gov/-accessed on 1 June 2022, reference number genome BioProject: PRJNA854809, genome accession numbers: CP101113-CP101114.