MALDI-TOF MS and genomic analysis can make the difference in the clarification of canine brucellosis outbreaks

Brucellosis is a widespread zoonosis of livestock (cattle, sheep/goats, pigs) caused by Brucella abortus, B. melitensis, and B. suis, with human infection typically via direct animal contact or contaminated animal products. Underreporting due to limited surveillance of classical species and lack of awareness of atypical species likely underestimates global prevalence, particularly in developing regions. Canine brucellosis, first described in the late 1960s, is most commonly due to Brucella canis and is associated with reproductive failure; transmission occurs via mating, ingestion of aborted material or reproductive secretions, and contaminated urine. Management in kennels facilitates spread. Beyond B. canis, B. suis (notably biovar 1 from feral pigs) can infect dogs, especially hunting dogs or those fed raw boar meat; B. suis bv 4 has been detected in carnivores in Arctic regions. Although B. canis is considered less pathogenic to humans than B. suis, it can cause both mild and severe/prolonged human disease. Diagnostic challenges exist because B. canis lacks the O-polysaccharide of smooth Brucella, rendering standard smooth-antigen serologic tests ineffective for B. canis. Consequently, human B. canis infection is difficult to diagnose. The objective of this study was to compare traditional phenotypic methods with newer omics-based approaches (MALDI-TOF MS and whole genome sequencing) to identify the infectious agent in Brazilian kennel dogs with brucellosis-like manifestations and to clarify the outbreak and its epidemiological context.

Prior work establishes that canine brucellosis is globally prevalent in kennels and is primarily caused by Brucella canis, with B. suis (notably biovar 1 and 4) also implicated in dogs, particularly those exposed to wild boar or raw meat. Human infections with B. canis, though reported as rare, may be underestimated; B. suis biovar 1 is highly pathogenic to humans. Diagnostic limitations include asymptomatic infections in dogs, intermittent bacteremia leading to culture false negatives, and serologic challenges: standard smooth LPS-based tests fail to detect B. canis antibodies due to its rough LPS, and cross-reactivity or variable sensitivity/specificity can yield false positives/negatives. MALDI-TOF MS is sensitive and specific for genus-level Brucella identification, but species-level discrimination (e.g., B. canis vs B. suis biovars) has been difficult with public libraries; prior studies emphasize the need for customized reference spectra and biomarkers. Genotyping methods (ribotyping, PFGE, MLST, MLVA) can subtype Brucella, but high interspecies homology often necessitates whole-genome SNP-based analyses for precise phylogeny and outbreak resolution.

Study setting and sampling: A breeding kennel in São Paulo, Brazil (2013–2014) with 17 adult pugs (2 males, 15 females) experienced multiple abortions. All dogs underwent clinical examination and three rounds of sampling (March, August, November 2014). Ethical approval: CEUA 3113091015/2015 (FMVZ-USP), procedures per CONCEA. Serology: Approximately 5 ml of blood obtained via cephalic/jugular venipuncture; half anticoagulated with sodium citrate; serum from clotted blood stored at −20 °C. Tests for anti–B. canis antibodies included immunochromatographic test (ICT; Bionote), ELISA (Novateinbio), and rapid slide agglutination test (RSAT; D-TEC CB) with or without 2-mercaptoethanol (2ME-RSAT). 2ME-RSAT was performed only on RSAT-positive sera. Culture and phenotypic characterization: Citrated blood underwent enrichment in tryptose phosphate broth with 5% fetal calf serum (FCS) at 37 °C for 30 days, with subculture every 4 days onto tryptose agar + 5% FCS. Colonies screened as Brucella by genus-specific PCRs targeting IS711 and bcsp31. Phenotypic tests included Gram/Stamp/crystal violet staining; growth (CO2 requirement; ambient air growth; blood agar and Brucella agar ± serum); H2S production; oxidase, catalase, urease activities; agglutination with anti-A, anti-M, and anti-R monospecific sera; dye sensitivity (basic fuchsin, thionine); motility (TTC agar); and phage lysis (F1, F25, Tb, BK2, Iz, Wb, R/C) for species/biovar differentiation. Direct PCR from blood: For culture-negative samples, IS711 PCR was performed on DNA extracted from whole blood using mechanical (zirconia/silica bead beating) and enzymatic lysis (lysozyme, then overnight proteinase K with SDS), followed by phenol/chloroform purification and alcohol precipitation. MALDI-TOF MS: Isolates were inactivated in 75% ethanol (≥90 min, −20 °C storage), processed by ethanol–formic acid extraction, spotted on 96-spot steel targets with HCCA matrix. Spectra acquired on Bruker Microflex LT. Initial identification used Bruker Biotyper 3.1 with MBT 7311 MSP library and Security-Relevant Database; databases lacked B. canis, so an in-house spectral library of B. canis and B. suis strains (from prior work and this study) was added. For each sample, 24 spectra were acquired; a main spectrum (MSP) from the 20 best technical replicates generated identification scores interpreted per manufacturer. Spectral processing and visualization were also performed in R (MALDIquant, Tidyverse). A weighted pattern-matching approach with modified scoring emphasized discriminant m/z biomarkers enabling discrimination among B. canis, B. suis bv 1, and B. suis bv 4. Whole-genome sequencing (WGS): DNA extracted (PureLink Genomic DNA kit) from Brucella canis grown on Brucella Agar. Libraries prepared with Nextera XT; sequencing on Illumina MiSeq (2×300 bp paired-end). Quality control with FastQC. Assemblies via SPAdes v3.10.0 (BayesHammer, careful, cov-cutoff auto). Assembly quality assessed with QUAST versus B. canis ATCC 23365 reference. Read mapping with Bowtie2 to ATCC 23365 or to a consensus genome assembled from concatenated high-quality outbreak read sets (isolates 19, 113, 114, 115), termed B. canis BfR-SPBR-consensus. Mapping stats via QualiMap; ANI via Gegenees; IS elements via BLASTn against ISfinder; assembly graph inspection with Bandage. In silico MLVA-16 (Python script) and MLST-21 (mlst tool) were performed. Phylogenomics and SNP analysis: Genetic relatedness to global strains assessed using ParSNP v1.0 with available B. canis genomes (PATRIC/NCBI/SRA) plus B. suis bv 4 as outgroup; trees by FastTree2; visualization/metadata integration via iTOL. Whole-genome SNPs were called using two strategies: (1) read-based in BioNumerics v7.6 with rigorous trimming and mapping; filters required ≥5× coverage and representation on forward and reverse reads; manual curation resolved low coverage or ambiguous calls; clustering by Neighbor Joining. (2) assembly-based in Mauve, with SNP lists extracted also via ParSNP and annotated with SnpEff using RefSeq annotation. Consensus genome features and SNPs relative to ATCC 23365 were cataloged. Raw reads deposited in ENA (ERX4130724–ERX4130727; BioProject ERP121782).

- Outbreak context and clinical signs: In a São Paulo kennel of 17 pugs, an initial abortion in 2013 was followed by five abortions over four months. On clinical exam, 14/17 dogs had brucellosis-like manifestations (vaginal discharge, lymphadenopathy, abortion/stillbirth, infertility); 3 were asymptomatic. - Serology: Across three sampling dates, all 17 dogs had antibodies against rough Brucella antigens by at least one test (ICT, RSAT, ELISA). Results were often inconsistent across tests (e.g., RSAT/ELISA positive with ICT negative; 2ME-RSAT frequently negative despite other positives), highlighting sensitivity/specificity limitations. - Culture/PCR: Brucella was isolated from 15/17 dogs at least once, yielding 29 isolates from 51 blood samples. Among 22 culture-negative blood samples, 3 (13.6%) were IS711 PCR-positive, indicating infection despite culture failure. - Phenotypic identification: Isolates were Gram-negative, short rods, non-motile; oxidase+, catalase+, urease+, no H2S; growth in ambient air on blood and Brucella agars; growth on thionine and basic fuchsin; phage lysis pattern matched B. canis RM 6/66 (constant lysis by phage R/C only). Agglutinated with anti-R but not anti-A/M sera; crystal violet staining confirmed rough LPS—consistent with B. canis. - MALDI-TOF MS: Commercial libraries initially misidentified isolates as B. melitensis (scores 2.0–2.3). Using a validated in-house database and weighted pattern matching, isolates matched B. canis and were differentiable from B. suis bv 1 and bv 4. Discriminant biomarkers included: m/z 9075 (and its double/triple-charged ions at m/z 4539, 3024) distinguishing B. canis/B. suis bv 4 from B. suis bv 1; m/z 9109 and 4555 prominent in B. suis bv 1; a shared B. canis/B. suis bv 4 biomarker at m/z 7073; intensity differences at m/z 7661 (single) and 3830 (double) separating these from B. suis bv 1; peaks at m/z 5900 and 3926 largely specific to B. canis versus B. suis bv 4. - Genomics: All isolates identified as B. canis by WGS; average draft genome 3,295,525 bp; mean G+C 57.27%; 24–57 contigs per assembly due to repetitive elements (IS711, ISBm2/ISBm3, IS1953, IS2020, rRNA operons). High clonality among kennel isolates with only 7 SNPs detected; five isolates differed by a single SNP; isolate 116 had two SNPs. Some SNPs at later sampling dates may reflect lab mutations or low coverage artifacts. - Consensus genome: B. canis BfR-SPBR-consensus comprised 23 contigs ≥500 bp totaling 3,293,143 bp (≈19.6 kb shorter than ATCC 23365). Comparative SNP analyses (BioNumerics and Mauve/ParSNP) identified 180 common SNPs distinguishing the outbreak consensus from ATCC 23365 (57 coding SNPs on chromosome 1; 47 on chromosome 2; plus noncoding variants). - Phylogeny: Core-genome alignment (ParSNP) clustered the outbreak strain with Brazilian and neighboring South American B. canis isolates. Closest relatives were São Paulo kennel strains from 2005 and 1998; strains from Chile and Colombia were more divergent. A human Argentine B. canis (CNGB 1324) clustered within the Brazilian dog group, supporting zoonotic linkage and regional endemicity. - Functional SNP patterns: South American B. canis strains shared conserved missense/synonymous SNPs and frequent stop-loss mutations in coding genes (e.g., autotransporter proteins), while stop-gain mutations were less common and more prevalent in Colombia/Chile strains, suggesting potential phenotypic variability within a genetically close sub-lineage.

The study addressed the challenge of rapid, accurate identification and outbreak clarification of canine brucellosis by combining optimized MALDI-TOF MS and whole-genome sequencing. Traditional culture and phenotypic assays confirmed Brucella infection and supported B. canis identification but required several days and were hampered by intermittent bacteremia and the limitations of serology (variable sensitivity/specificity, cross-reactivity). The MALDI-TOF MS approach, enhanced by an in-house reference database and weighted biomarker scoring, enabled rapid species-level identification of B. canis and discrimination from B. suis biovars 1 and 4—critical for clinical management and public health risk assessment given differing zoonotic potentials. WGS provided high-resolution evidence of a clonal, single-strain outbreak within the kennel and placed the outbreak strain within a South American sub-lineage that includes canine and human isolates, indicating regional endemicity and potential cross-border transmission linked to dog trade. The detection of conserved coding-region SNPs, including stop-loss and occasional stop-gain mutations in genes related to nutrient utilization and virulence-associated autotransporters, suggests microdiversity that could translate to phenotypic differences despite overall genetic homogeneity. Together, these findings demonstrate that MALDI-TOF MS can substantially accelerate initial diagnostics, while WGS clarifies transmission dynamics and epidemiology, thereby informing control strategies to mitigate spread among dogs and reduce zoonotic risk to humans.

This work demonstrates that MALDI-TOF MS, when supported by a curated in-house Brucella reference library and weighted biomarker analysis, allows rapid and reliable species-level identification of Brucella canis and discrimination from closely related B. suis biovars relevant to canine infections. Complementary whole-genome sequencing resolved the kennel event as a clonal single-strain outbreak and positioned the strain within a South American sub-lineage encompassing canine and human isolates, underscoring regional endemicity and zoonotic implications. The combined proteomic-genomic approach shortens time to actionable diagnosis and enriches epidemiological understanding of canine brucellosis. Future research should: (1) validate and disseminate expanded MALDI-TOF MS libraries and scoring schemes across laboratories; (2) integrate rapid WGS-based surveillance to monitor transmission and cross-border spread; (3) investigate functional consequences of conserved SNPs (e.g., stop-loss/gain in autotransporters) on virulence, host range, and clinical outcomes; and (4) improve serologic tools specific for B. canis to enhance screening and case confirmation.

- Culture sensitivity is limited, particularly at later stages with intermittent or ceased bacteremia, leading to false negatives; several dogs were culture-negative at later time points but PCR-positive. - Serologic tests (ICT, RSAT, ELISA, 2ME-RSAT) produced inconsistent and sometimes contradictory results due to sensitivity/specificity constraints and potential cross-reactivity with other pathogens; timing relative to seroconversion and fluctuating titers can cause false negatives. - Short-read Illumina sequencing impeded generation of complete circular chromosomes due to repetitive elements (e.g., IS711, rRNA operons); some SNP calls had low coverage and may include lab-acquired mutations or artifacts. - Initial MALDI-TOF MS identification using commercial libraries misassigned isolates (e.g., as B. melitensis) because public databases lacked B. canis reference spectra, necessitating an in-house library. - The study focuses on a single kennel outbreak; broader generalizability would benefit from larger, multi-center datasets and inclusion of more B. suis canine cases for comparative diagnostics.