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) [1–6]. 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 [9–11]. 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, 12–16]. Brucellosis in humans is mostly caused by B. abortus, B. melitensis, B. suis, and sometimes B. canis [14, 17–19], 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 [21–26]. 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-year-old 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 (BO1T) associated with a breast implant wound of a 71-year-old patient from Portland, Oregon with clinical signs of brucellosis [7, 8].

Results

Phenotypic characterization

BO2 cells grown on SBA or RBA at 35-37°C with or without 5% CO2 for 24 to 48 h were circular, convex, entire, smooth and opaque. The organisms were gram-negative, 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 BO1T 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). BO1T 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 BO1T strain of the B. inopinata sp. than the other classical Brucella spp.

Table 1 MIC results for 5 antimicrobial agents tested against BO1T, BO2 strains and 93 Brucella strains
Full size table

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 BO1T 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 BO1T and other Brucella strains with low intensity as reported earlier [8].

Figure 1
figure 1

IS 711 profiles of PCR amplified products analyzed by gel electrophoresis on a 2% E-Gel displaying the following: molecular weight marker (lane 1), no template control (lane 2), B. abortus ATCC 23448 (lane 3), B. melitensis 16 M (lane 4), B. suis ATCC 23444 (lane 5), B. ovis ATCC 25840 (lane 6), BO1T (lane 7), and BO2 (lane 8).

Full size image

Real-Time PCR for BO1T/BO2

A TaqMan PCR assay targeting conserved regions of the BO1T 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 BO1T 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 BO1T/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. inopinata- like 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 BO1T, 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 BO1T and 99.6% identity with other Brucella spp. (Table 2).

Table 2 Comparative percent identity based on pair-wise analysis of five genes of BO2 with BO1T and classical Brucella spp. using MEGA4.
Full size table

Omp2a/2b genes sequence analysis

We also analyzed two highly homologous outer membrane porin genes (omp2a and omp2b) of the BO2 and BO1Tstrains 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 BO1T and an atypical B. suis 83-210 strain [32]. The omp2a gene of BO2 is only 1.0% divergent from that of BO1T. 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 BO1T and B. suis 83-210 strains, respectively (Figure 2, Table 2). Clustering analysis demonstrates that BO1T, BO2 and the B. suis 83-210 strains form consistent sub-groups based on their omp2a and omp2b gene homology [32].

Figure 2
figure 2

Phylogenetic tree reconstructed with omp2a (1093 bp) and omp2b (~1211 bp) sequences using MEGA v.4.0 neighbor joining analysis. The bootstrap consensus tree inferred from 1000 replicates is taken to represent the evolutionary history of the taxa analyzed. The significance of each branch is indicated by a bootstrap percentage calculated from 1000 replicates.

Full size image

RecA gene sequence analysis

The recA gene (948 bp) of strain BO2 was compared to those of BO1T, 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 BO1T recA sequence differing by 17 nucleotides. Phylogenetic analysis of BO1T and BO2 strains with other Brucella and Ochrobactrum spp. shows that the Brucella spp. clade including BO2 and BO1T, are distantly similar to the Ochrobactrum spp. with approximately 85% sequence identity (Figure 3).

Figure 3
figure 3

Phylogenetic tree reconstructed with recA (948 bp) sequences using MEGA v.4.0 neighbor joining analysis. The bootstrap consensus tree inferred from 1000 replicates is taken to represent the evolutionary history of the taxa analyzed. The significance of each branch is indicated by a bootstrap percentage calculated from 1000 replicates.

Full size image

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). BO1T 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 BO1T and BO2 species are from the classical Brucella sequence types (Figure 4). Throughout the 4,396-bp sequence examined, the BO1T and BO2 genomes have 32 common SNPs while there are 30 BO1T and 26 BO2 specific nucleotide changes that further characterize the divergence of these two strains at these highly conserved loci in the Brucella genus.

Figure 4
figure 4

Unrooted phylogenetic reconstruction of the concatenated sequences of nine house-keeping genes (4,396 bp) using the neighbor-joining approach. Represented are the 27 known Brucella sequence types along with BO1T and their relation to BO2.

Full size image

Multiple-Locus Variable-Number Tandem Repeat Analyses

Both BO2 and BO1T 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]. Our MLVA-15 typing analysis of both BO2 and BO1T strains demonstrated unique VNTR profiles in which both strains have six Brucella-loci with the same alleles (VNTR 2, -3, -14, -20, -21 and -25); and seven loci with variable VNTR amplicons (VNTR1, -7, -27, -29, -30, -31 and -33). All VNTRs successfully amplified in both BO1 and BO2 with the exception of VNTR16 and -28 in BO1T. MLVA-15 analysis revealed that both BO2 and BO1T had distinct VNTR profiles in comparison to each other and other Brucella strains (Figure 5).

Figure 5
figure 5

Condensed unweighted pair group method analysis (UPGMA) dendogram of multiple-locus variable number tandem repeat analysis (MLVA) genotypes of BO1T, BO2 strains along with 209 characterized Brucella strains.

Full size image

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 (BO1T) 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 H2S production, inability or very weak agglutination with Brucella specific antisera for the lipopolysaccharide-O-antigens or acriflavin. Neither the BO1T 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 BO1T 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 [36–38]. The presence of multiple copies of BO1T-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 BO1T 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 BO1T 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 was 100% identical to that of BO1T and 99.6% identical to the Brucella spp. consensus 16S rRNA gene sequence. The high sequence identity of the BO2 16S rRNA sequence to the recently described B. inopinata sp. is remarkable and represents the first recognized Brucella species to have a divergent 16S rRNA sequence [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 BO1T. Sequence analysis revealed that the recA nucleotide sequence of the BO2 strain shared greater similarity with the Brucella spp. recA consensus sequence than to BO1T. Both BO2 and BO1T 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 BO1T, 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 [42–45]. 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 BO1T 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 BO1T 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 BO1T 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 BO1T 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 BO1T 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 BO1T, 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 BO1T 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.

Methods

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.

Phenotypic analysis

The BO2 strain was routinely maintained on Trypticase soy agar with 5% defribinated sheep blood agar (SBA) or rabbit blood agar (RBA) (BBL Microbiology Systems, Cockeysville, MD). Phenotypic identification of the BO2 strain was performed according to the laboratory techniques in brucellosis described by Alton et. al. in the World Health Organization monogram [7, 8, 28].

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]

Molecular analysis

Detection of IS711

To detect the Brucella- specific insertion sequence IS711 element (842 bp) [37], cell lysate DNA templates from strains BO2, BO1T, B. abortus (ATCC 23448), B. suis 1330 (ATCC 23444), B. ovis (ATCC 25840) and B. melitensis 16 M (ATCC 23456) were amplified and the amplicons were analyzed by 2% E-Gel agarose gel electrophoresis as mentioned previously [8].

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 BO1T (positions 145 to 148; GenBank accession no. EU053207). We designed two PCR primers16SF (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 MgCl2, 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 BO1T 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 BO1T and other Brucella and Ochrobactrum spp. available in GenBank. 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 BO1T, 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.

MLVA typing

Molecular typing of the BO2 strain based on multiple-locus variable-number tandem repeat (VNTR) analysis (MLVA) was investigated by examining fifteen Brucella spp. VNTR genetic markers (MLVA-15) [48, 49], and a distance tree was generated in BioNumerics v.5.1 (Applied Maths, Saint-Martens-Latem, Belgium) by clustering analysis using the unweighted-pair group method with arithmetic averages (UPGMA) and saved in newick format. Tree manipulations and labeling were done in MEGA 4.0 [46].