Identification of an unusual Brucella strain (BO2) from a lung biopsy in a 52 year-old patient with chronic destructive pneumonia

Background Brucellosis is primarily a zoonotic disease caused by Brucella species. There are currently ten Brucella spp. including the recently identified novel B. inopinata sp. isolated from a wound associated with a breast implant infection. In this study we report on the identification of an unusual Brucella-like strain (BO2) isolated from a lung biopsy in a 52-year-old patient in Australia with a clinical history of chronic destructive pneumonia. Results Standard biochemical profiles confirmed that the unusual strain was a member of the Brucella genus and the full-length 16S rRNA gene sequence was 100% identical to the recently identified B. inopinata sp. nov. (type strain BO1T). Additional sequence analysis of the recA, omp2a and 2b genes; and multiple locus sequence analysis (MLSA) demonstrated that strain BO2 exhibited significant similarity to the B. inopinata sp. compared to any of the other Brucella or Ochrobactrum species. Genotyping based on multiple-locus variable-number tandem repeat analysis (MLVA) established that the BO2 and BO1Tstrains form a distinct phylogenetic cluster separate from the other Brucella spp. Conclusion Based on these molecular and microbiological characterizations, we propose that the BO2 strain is a novel lineage of the newly described B. inopinata species.


Background
Brucellosis is primarily a zoonotic disease, caused by members of the genus Brucella, which currently constitutes several species based on pathogenicity, host preferences and phenotypic characteristics: B. abortus (cattle), B. canis (dogs), B. melitensis (goats), B. suis (pigs), B. ovis (rams), B. neotomae (desert rats), B. ceti and B. pinnipedialis (marine mammals), and B. microti (common vole) [1][2][3][4][5][6]. Recently, a novel species, Brucella inopinata, associated with a human infection has been recognized as the newest member of the genus Brucella [7,8]. In early 1985, whole genome hybridization analysis studies revealed a high degree of genetic homology among the Brucella species, which led to the proposal that the genus Brucella was a mono-specific species with B. melitensis being the primary species and all others as sub-species and biovars [9][10][11]. However, due to the limited acceptability of the one-species concept, the traditional classification of Brucella spp. based on phenotypic characteristics has been re-instated by the Brucella Taxonomy Subcommittee in 2006 [3].
Brucella are facultative intracellular pathogens that infect many organs and soft tissues, including mammary glands. Infection frequently results in abortion, low milk production and fetal death in animals [2,[12][13][14][15][16]. Brucellosis in humans is mostly caused by B. abortus, B. melitensis, B. suis, and sometimes B. canis [14,[17][18][19], and is commonly associated with the consumption of unpasteurized dairy products, meat from infected animals and exposure to infected animal tissues or laboratory transmission [1,2,20]. Human brucellosis is a chronic debilitating infection with a very broad clinical picture potentially affecting any major organ, including the lung, causing varying respiratory symptoms [20]. Respiratory infections in humans caused by Brucella spp. is a rare manifestation with reports describing multifocal abscesses or nodules, hilar adenopathy and hemorrhagic pleural effusion with resolution by antimicrobial therapy and lung decortications [21][22][23][24][25][26]. Most pulmonary brucellosis cases were found in farmers handling infected meat or travelers who consumed raw infected animal meat or unpasteurized milk products while visiting countries endemic for brucellosis [26,27].
We report the isolation and identification of an unusual gram-negative, non-motile Brucella-like coccoid bacillus (BO2) isolated from a lung biopsy in a 52-yearold male in Australia with a history of chronic destructive pneumonia. The patient traveled worldwide but denied any common risk factors associated with brucellosis. Both biochemical and molecular characteristics of the BO2 strain have demonstrated unique similarity with a recently described B. inopinata strain (BO1 T ) associated with a breast implant wound of a 71-year-old patient from Portland, Oregon with clinical signs of brucellosis [7,8].

Phenotypic characterization
BO2 cells grown on SBA or RBA at 35-37°C with or without 5% CO 2 for 24 to 48 h were circular, convex, entire, smooth and opaque. The organisms were gramnegative, generally stained uniformly; and appeared coccoid to short coryneform rods. Colonies of the BO2 strain ranged in size from punctuate to 1.5 mm in diameter and they were non-motile, mucoid colonies on MacConkey agar; positive for oxidase and catalase, exhibited nitrate reduction with production of gas and rapid urease production (< 5 min). Hydrogen sulfide production by the BO2 strain was observed by the development of a dark gray color on lead acetate paper suspended above the heart infusion agar slant.
Subculture of individual colony types produced similar profiles and no hemolytic reaction was observed on SBA plates after overnight incubation at 37°C. The BO2 cells grew in the presence of thionine (1:25,000, 1:50,000 and 1:100,000 dilutions) and basic fuchsin (1:50,000 and 1:100,000 dilutions) dyes within 24 to 48 h. Both the acriflavin and gel formation tests were negative. However, lysis by Tbilisi phage specific for detection of Brucella spp. in two routine test dilutions (1× and 4× RTD) appeared incomplete [7,8,28] and agglutination of the BO2 cells with either monospecific anti-M or anti-A antisera were very weak.

Antimicrobial susceptibility test
The antimicrobial susceptibility profile of the BO2 strain was compared with a set of 93 other Brucella spp. strains (74 B. melitensis, 14 B. suis and 5 B. abortus) along with BO1 T based on CLSI interpretive requirements for Brucella spp. [8,29,30]. Both strains had very similar MIC patterns to all Brucella reference strains tested previously [8,30] (Table 1). BO1 T and BO2 strains grew well in cation-adjusted Mueller-Hinton broth (CAMHB) after just 20 hours of incubation, unlike other Brucella spp. (e.g., B. abortus, B. melitensis, and B. suis) which do not routinely grow very well in CAMHB and require 48 hours of incubation in Brucella broth for MIC testing [30]. Our standard phenotypic characterization, including the antimicrobial susceptibility profiles, suggested that the BO2 strain more closely resembled the BO1 T strain of the B. inopinata sp. than the other classical Brucella spp.

Molecular characterization Detection of IS711 element by PCR
The Brucella specific insertion sequence (IS711) PCR was performed amplifying an 842-bp repetitive element using BO2 genomic DNA. The IS711 profile observed in strain BO2 was approximately the same size as that of the BO1 T strain and the classical Brucella spp. including B. ovis (ATCC 25840) ( Figure 1). The BO2 strain also generated several large amplicons (>1000 bp) similar to BO1 T and other Brucella strains with low intensity as reported earlier [8].

Real-Time PCR for BO1 T /BO2
A TaqMan PCR assay targeting conserved regions of the BO1 T and Brucella spp.16S rRNA gene sequence was designed for rapid differentiation of potential B. inopinata-like strains from all other classical Brucella and Ochrobactrum spp. This real-time PCR assay, using two hybridization probes: BI-P specific for B. inopinata spp. and BRU-P specific for Brucella/Ochrobactrum spp., gave average crossing threshold (Ct) values in the range of 15 to 20 (strong positive). The BI-P probe demonstrated perfect agreement for both BO1 T and BO2 strains as did the BRU-P probe for all other Brucella or Ochrobactrum spp. respectively. Both probes showed no cross reactivity against the other non-Brucella strains tested to date [31] demonstrating very high specificity of the target sequences in the PCR assay. Both the BO1 T / BO2 and the Brucella/Ochrobactrum specific probes were capable of optimal detection of template down to 10 fg/μl concentration of genomic DNA template (data not shown).

16S rRNA gene sequence analysis
Rapid identification of the BO2 strain as B. inopinatalike by the BO1 PCR assay led to sequence analysis of the full-length 16S rRNA gene (1,412 bp) of the BO2 strain. Full sequence alignment with the 16S rRNA gene sequences of BO1 T , reference Orchrobactrum spp. strains, and the Brucella spp. consensus sequence confirmed that the BO2 strain shared 100% 16S rRNA gene sequence identity to that of BO1 T and 99.6% identity with other Brucella spp. (Table 2).

Omp2a/2b genes sequence analysis
We also analyzed two highly homologous outer membrane porin genes (omp2a and omp2b) of the BO2 and BO1 T strains and compared their full-length sequences with that of other Brucella species available in GenBank. The phylogenetic relationships derived by neighbor-joining clustering analysis of the BO2 omp2a (1093 bp) and omp2b (~1212 bp) genes with the NCBI sequences of other Brucella strains and the Ochrobactrum anthropi LMG 3331 reference strain demonstrated considerable intra-and inter-species variability ( Figure 2). The BO2 omp2a and omp2b genes are 84.6% homologous to each other. Neighbor-joining clustering analysis of both omp2a and omp2b nucleotide sequences shows that BO2 clusters closest to BO1 T and an atypical B. suis 83-210 strain [32]. The omp2a gene of BO2 is only 1.0% divergent from that of BO1 T . The omp2b gene is characteristically more diverse within the Brucella spp. and is also evident with the BO2 omp2b gene which was 95.3% and 94.1% identical to the BO1 T and B. suis 83-210 strains, respectively ( Figure 2, Table 2). Clustering analysis demonstrates that BO1 T , BO2 and the B. suis 83-210 strains form consistent sub-groups based on their omp2a and omp2b gene homology [32].

RecA gene sequence analysis
The recA gene (948 bp) of strain BO2 was compared to those of BO1 T , the classical Brucella spp.(n = 8) and several representative Ochrobactrum spp. [31,33]. Within the genus Brucella, the recA gene is highly conserved with 100% nucleotide sequence identity among the different species. Interestingly, the BO2 recA nucleotide sequence reveals 99.2% identity to the Brucella consensus recA sequence due to 8 nucleotide substitutions. However, the BO2 recA gene has a lower identity (98.2%) when compared to the BO1 T recA sequence differing by 17 nucleotides. Phylogenetic analysis of BO1 T and BO2 strains with other Brucella and Ochrobactrum spp. shows that the Brucella spp. clade including BO2 and BO1 T , are distantly similar to the Ochrobactrum spp. with approximately 85% sequence identity ( Figure  3).

Multiple Locus Sequence Analysis
Multiple locus sequence analysis (MLSA) of nine Brucella spp. house-keeping genes has been used to differentiate Brucella spp. into distinct sequence types (ST). BO1 T was determined to be 1.67% divergent from ST1 and to possess novel alleles at all nine loci [8]. BO2 has shown similar divergence (1.5%) from ST1 by MLSA also with novel alleles in all nine loci. Neighbor-joining phylogenetic analysis clearly shows how divergent the BO1 T and BO2 species are from the classical Brucella sequence types ( Figure 4). Throughout the 4,396-bp sequence examined, the BO1 T and BO2 genomes have 32 common SNPs while there are 30 BO1 T and 26 BO2 specific nucleotide changes that further characterize the divergence of these two strains at these highly conserved loci in the Brucella genus.

Multiple-Locus Variable-Number Tandem Repeat Analyses
Both BO2 and BO1 T strains were also investigated by multiple-locus variable-number tandem repeat (VNTR) analysis (MLVA) using fifteen VNTR loci by capillary electrophoresis. Results were compared with a panel of well-characterized Brucella strains (n = 209) representing known species from our collection [31].  Figure 5).

Discussion
In this paper we present the identification of an atypical Brucella-like strain (BO2) isolated from the lung biopsy of a 52-year-old patient. As a young adult he lived in Oregon on two occasions (1981 and 1985-1987), and experienced an unexplained 'liver failure' and then severe pneumonia (with pleurisy) from which he recovered with multiple courses of antimicrobial therapy as reported by the patient to his physicians in Australia. This patient was originally misdiagnosed because of the misidentification of the BO2 strain as O. anthropi on an AP1 20NE system. It is a common practice for clinical labs to attempt rapid identification of gram-negative coccobacillus organisms like Brucella spp. from blood culture using automated systems. However, the Brucella spp. are often misidentified due to their similar phenotypic characteristics to closely related organisms such as Ochrobactrum spp. [34,35]. Though the patient was initially treated for both Ochrobactrum and Brucella infections due to the difficulties in diagnosis, he recovered with an extended course of combination oral antimicrobial therapy. This BO2 strain is phenotypically and molecularly similar to the recently identified B. inopinata type strain (BO1 T ) recovered from a patient from Oregon, which was also originally misidentified as O. anthropi by the API 20E and API 20NE [7,8]. Both these strains share common colony morphology and biochemical characteristics including rapid urease and positive H 2 S production, inability or very weak agglutination with Brucella specific antisera for the lipopolysaccharide-O-antigens or acriflavin. Neither the BO1 T or BO2 strains supports gel formation or exhibits growth inhibition to the dye media as shown by common members of the genus Brucella. BO2 also exhibited incomplete lysis by Tbilisi phage and had very similar antimicrobial susceptibility profiles to BO1 T in comparison to other Brucella reference strains.
Insertion sequence (IS) fingerprinting in the Brucella species has shown that the genomic localization and copy number of the IS711 insertion element (also called IS6501) is species-specific and could have an association with specific pathogenicity for a preferred host [36][37][38]. The presence of multiple copies of BO1 T -like IS711 insertion sequences suggest not only that BO2 is a member of the Brucella genus ( Figure 1) but that the BO2-IS711 amplification pattern specifically resembles that of the newly described B. inopinata species [8]. Positive identification of the BO2 strain as a member of B. inopinata by our real-time BO1 PCR assay was significant. Both BO1 T and BO2 strains were the cause of distinct and unusual forms of human brucellosis. Atypical clinical isolates of this nature can often be misdiagnosed by automated systems as was the case with BO1 T and the BO2 strain described here [8,35]. The availability of the real-time TaqMan assay served as a reliable first-line tool for determining B. inopinata-like species.
These initial findings led to further characterization and sequence-based typing which provided additional supporting evidence that this new BO2 strain most resembles the B. inopinata sp. within the Brucella genus. Using broad-range eubacterial primers, Gee et. al. effectively demonstrated the advantage of 16S rRNA gene sequencing to identify Brucella isolates reporting 100% identity in all the strains examined [31]. Interestingly, the full-length 16S rRNA gene sequence of BO2    [8].
The recA gene has been investigated as an alternative phylogenetic marker for several bacterial genera due to its highly conserved nature and ubiquity in prokaryotes [33,39,40]. Unlike the high sequence homology of the recA gene within the Brucella genus [41], we identified unique variability in the recA gene sequences of BO2 and BO1 T . Sequence analysis revealed that the recA nucleotide sequence of the BO2 strain shared greater similarity with the Brucella spp. recA consensus sequence than to BO1 T . Both BO2 and BO1 T recA sequences are distanced by 8 and 11 unique SNPs, respectively, from the Brucella spp. recA consensus sequence, and share only one common transversion at the 517 nucleotide position. Translation of the recA gene sequences of BO1 T , BO2 and the Brucella spp. consensus sequence shows that all base pair changes were synonymous substitutions having no effect on protein structure or function. The Brucella outer membrane proteins have been studied extensively for their function in virulence, pathogenicity, bacteriophage reception, antigenic factors and antibacterial evasion [42][43][44][45]. The genetic variability among the omp genes within the Brucella spp. has proven effective at characterizing Brucella spp. and strain types and is often used for higher resolution molecular typing [4,32,43,45]. The omp2a/2b genetic analysis we report here is very interesting in that BO2 consistently associates with not only BO1 T but the atypical B. suis 83-210 strain that was isolated from a rodent in Australia [32]; and thus further investigation may be warranted into rodents as a possible natural reservoir for these novel Brucella species.
Investigation of the nine housekeeping genes by multi locus sequencing analysis demonstrates that BO2 is genetically distinct from BO1 T yet exhibits remarkably similar divergence (1.5%) from the classical Brucella sequence types as shown in Figure 4. The relative similarity of the nucleotide sequences of BO1 T and BO2 by MLSA demonstrates uniquely distant sequence types within the currently characterized Brucella spp. and should be considered as a new group of STs within the Brucella genus. They also exhibit distinct allelic profiles by MLVA although all alleles in both the BO1 T and BO2 allelic profiles have been observed in other Brucella spp. Furthermore, the phylogenetic analysis shown in Figure 5 demonstrates that these strains form a single separate cluster from the classical Brucella spp. [8].
The molecular and microbiological characteristics presented here provide supporting evidence that strain BO2 is most closely associated with the BO1 T strain and should be considered as a novel lineage of B. inopinata sp. Attempting to understand the evolutionary origin of these two strains is somewhat confounded by the interesting and disparate medical histories of the case patients (who both happened to have lived in Portland, Oregon) from whom these strains were isolated and suggests that there are new and emerging Brucella strains capable of causing unusual presentation of human brucellosis.

Conclusion
Phenotypic and genomic analysis of the unusual Brucella strain (BO2) from a lung biopsy have established it as a lineage of the recently identified novel B. inopinata sp. type strain BO1 T , which was isolated from a wound associated with a breast implant. This is the first report of a human brucellosis case associated with chronic destructive pneumonia caused by an atypical Brucella strain. An interesting finding from our molecular analysis reveals that both strains BO1 T and BO2 appeared to be closely related to a less-characterized B. suis strain 83-210 (isolated from a rodent in Australia) by their omp2a/2b genes, which may suggest a common ancestor and may also provide insight into the ecological niche, and host reservoir for these novel Brucella strains causing unusual human infections.

Patient
The patient was born in Malta in 1956 and immigrated to Australia at age two, where he would continually return and eventually settle throughout extensive worldwide travel including the Western region of the United States. Between 2003 and 2007, the patient was hospitalized multiple times in different hospitals in Australia for abnormal liver function, community acquired pneumonia, anterior chest wall abscess and sinus infection. In September 2007 a percutaneous lung biopsy was performed and a gram-negative organism was isolated from a broth culture of the fine needle aspirate of the patient's lung and identified as Ochrobactrum anthropi on an API20NE system. The testing laboratory was aware of the possibility of Brucella sp. being misidentified as Ochrobactrum anthropi [35] and the isolate was referred for further testing. The patient was treated with combination therapy of doxycycline and rifampicin for twelve months and ciprofloxacin for three months (the latter was ceased after molecular testing confirmed Brucella species). The culture was initially tested according to standard microbiological and molecular procedures and then forwarded to the Centers for Disease Control and Prevention (CDC), Atlanta, GA, for further characterization. This gram-negative organism was designated as BO2 and stored at -70°C in defibrinated rabbit blood until further evaluation.

Antimicrobial susceptibility analysis
The antimicrobial susceptibility testing of the BO2 strain was performed by the broth microdilution method in CAMHB and Brucella broth in accordance with the Clinical and Laboratory Standards Institute (CLSI) protocol as described previously [8,29]

Real-Time PCR assay
A real-time TaqMan PCR assay was developed targeting a four base nucleotide substitution within the 16S rRNA gene sequence of BO1 T (positions 145 to 148; GenBank accession no. EU053207). We designed two PCR pri-mers16SF (5'-CGGGCCGATCATTTGC-3') and16SR (5'-AACTCAGGGAAACTTGTGCTAATACC-3') to amplify a 72-bp region of the 16S rRNA Brucella consensus sequence and two hybridization probes, BI-P (5'-AAATCTTTCCCCTTTCGGGCAC-3') and BRU-P (5'-AAATCTTTCCCCCGAAGGGCAC-3'), targeting a 4-bp polymorphic region within the 72-bp amplicon. Both probes were synthesized with a 6-carboxyfluorescein reporter molecule attached at the 5' end and Black Hole Quencher 1 on the 3' end. Each final PCR reaction mix contained 2 μl of DNA template and 18 μl of PCR master mixture containing 1 × LightCycler Faststart DNA Master HybProbe mix (Roche Applied Sciences, Indianapolis, IN), 4 mM MgCl 2 , 0.4 μM of each primer and 0.2 μM of probe. The LightCycler thermal cycling conditions were 95°C for 8 min followed by 45 cycles of 95°C for 5 sec and 60°C for 5 sec ending in a 45°C hold for 1 min 15 sec. A panel of 54 well characterized Brucella strains and 28 near-neighbors, including 5 Ochrobactrum strains [31] were evaluated by the assay. Positive results are expressed in log scale as crossing threshold values (Ct) of fluorescence released above the no-template control baseline of 0.01 following each amplification as described by the manufacturer.

16S rRNA gene analysis
The full length amplicon of 16S rRNA gene was generated using the BO2 cell-lysate DNA and sequenced using the BigDye terminator cycle 3.1 sequencing kit (ABI, Foster City, CA) as described previously [31]. A comparative full-length sequence analysis of BO2 was performed with the consensus 16S rRNA gene sequence of Brucella spp. [31], and the Ochrobactrum intermedium type strain (GeneBank accession no. AM114411T) along with that of the B. inopinata BO1 T strain (GeneBank accession no. EU053207) using the GCG Wisconsin software package (version 10.2; Accelrys, San Diego, CA) and MEGA 4.0 [31,46].

Omp2a/2b and recA genes analysis
The full-length outer membrane porin genes omp2a and omp2b, and also the recA gene of BO2 were sequenced [33,45], and compared with sequences of BO1 T and other Brucella and Ochrobactrum spp. available in Gen-Bank. Contigs were assembled and edited before multiple sequence alignments were constructed in the DNASTAR Lasergene 8 genetic analysis software suite (DNASTAR Inc., Madison, WI). Neighbor-joining consensus trees inferred from 1000 bootstrap replicates were constructed using MEGA version 4.0 [46].

MLSA typing
To assess the relation of BO2 with other classical Brucella spp. and BO1 T , the multi locus sequence analysis (MLSA) primer sets were used to amplify and sequence nine discrete house-keeping genes as described previously [47]. Multiple sequences were aligned and neighbor-joining phylogenetic trees were constructed as described above.

Sequence identities
Similarity values reported throughout the text and in Table 2 were calculated from estimates of evolutionary divergence between the sequences represented in the corresponding dendograms. All results are based on the pairwise analysis of inclusive sequences using the Maximum Composite Likelihood method in MEGA 4.0 [46]. All positions containing gaps and missing data were eliminated from the dataset.