Melioidosis is a severe community-acquired infectious disease caused by the Gram negative bacillus Burkholderia pseudomallei. The bacterium is commonly found in soil and water in Southeast Asia and northern Australia; however, increasing cases of melioidosis have been reported in other tropical regions and the bacterium is now believed to present a serious global threat [1, 2]. Clinical manifestations of melioidosis are extremely diverse and vary from acute sepsis to chronic localised pathology to latent infections which can reactivate decades later [3, 4]. In all series, the lung was the most commonly affected organ, either presenting with cough and fever resulting from a primary lung abscess or pneumonia, or secondary to septicaemic spread [5]. The overall mortality rate in individuals infected with B. pseudomallei range from 30-70%, depending on whether the patient is septicaemic or not [5]. In northeast Thailand the mortality rate is reported to be 50% (35% in children) and 19% in Australia [3, 6].

Melioidosis is treatable if detected early, however, early medical intervention is hindered by two major factors; firstly, the definitive diagnosis involves bacterial culture and identification which is time consuming and often leads to a critical delay in antibiotic therapy and successful management of the acute form of the disease [7]. Secondly, B. pseudomallei is resistant to diverse groups of antimicrobials including third generation cephalosporins whilst quinolones and aminoglycosides have no reliable effect [8, 9]. Hence, a definitive and rapid diagnosis is critical and vital prior to the administration of ceftazidime or carbapenems as these antibiotics are generally not used as empirical treatment for septicaemia in endemic regions. Culturing of bacteria from clinical samples represents the diagnostic gold standard for melioidosis [3, 10]. Whilst it is simple, reliable and economical, the long duration required to reach a definitive diagnosis is a major drawback. The most commonly used serological method, the indirect hemagglutination test (IHA), is poorly standardised worldwide. Moreover, IHA has limited clinical value in regions of endemicity due to the high background antibody titres in healthy individuals, most likely the result of repeated environmental exposure to B. pseudomallei [11]. In addition, IHA uses crude whole-cell preparation or bacterial lysates which may potentially lead to high false-positive results. A critical limitation of this assay is the lack of standardisation between laboratories with respect to the antigens used, the antigens remain poorly characterised and are likely to be variable between isolates [12]. The indirect immune-fluorescence antibody test (IFAT) using whole B. pseudomallei cells as antigen was found to be sensitive and superior to IHA and requires only a day to obtain the results [13]. The only drawback is that IFAT requires a fluorescence microscope and skilled personnel which might not be readily available in rural endemic regions of Southeast Asia. Enzyme-linked immunosorbent assay (ELISA) and related serodiagnostic strategies are being considered more favourably as rapid and reliable tools for definitive diagnosis of melioidosis [14–16]. Various antigen preparations such as crude and purified exopolysaccharide (EPS) and lipopolysaccharide (LPS) as well as recombinant flagellin and Bip proteins have been reported as potential diagnostic antigens in an ELISA format, however, sensitive and reliable antigens are yet to be identified.

We have previously reported the characterisation of several immunogenic B. pseudomallei recombinant proteins including outer membrane protein A (Omp3), Omp85 and serine protease MprA (smBpF4) [17–19]. Recently, Chieng et al. [20] described the elevated induction levels of B. pseudomallei type VI secretion system HCP protein (TssD-5) over a 6 hr infection period in human macrophages and preliminary analysis of the antigenicity of this protein has indicated strong binding to antibodies in a small cohort of melioidosis-confirmed patient sera (Chieng et al. in preparation). All four recombinant proteins are unique to Burkholderia spp. and interestingly, TssD-5 is highly conserved in B. pseudomallei only whilst the other recombinant proteins are conserved in B. pseudomallei as well as in Burkholderia spp. such as B. mallei and B. thailandensis. Thus, all the selected proteins should demonstrate minimal cross reactivity with sera from non-melioidosis individuals presenting with other bacterial infections. Concomitantly, these 4 recombinant proteins were strongly reactive with melioidosis patients’ or B. pseudomallei infected animal sera [17–19]. As these B. pseudomallei recombinant proteins were patient sero-reactive proteins, we attempt to evaluate their potential as suitable antigen(s) for diagnosis of melioidosis by an ELISA-based screen, a rapid diagnostic method to replace the conventional gold standard which involves bacterial culture and is time consuming.

A recent study reported the true sensitivity of culture as a diagnostic tool as 60% [27]. Thus, there is critical need for a fast, easy and accurate diagnostic assay for melioidosis as well as to differentiate individuals with active infection from those previously exposed to the bacteria, particularly in endemic countries [28]. Recombinant proteins as diagnostic antigens offer many advantages over the use of bacterial cells including eliminating health risks to laboratory personnel and reducing inconsistent antigen preparation [28]. Therefore, the use of recombinant proteins as potential diagnostic antigens via an ELISA detection-based screen were evaluated but with varying results. Recombinant-truncated flagellin was tested by ELISA as a diagnostic antigen and demonstrated high sensitivity and specificity (93.8% and 96.3%, respectively) [14], but only a limited number of samples from non-endemic regions was used. BipB and BipD were also evaluated as antigens by screening samples from 2 endemic regions, Thailand and Australia. However, neither of them demonstrated better diagnostic accuracy than the IHA [16].

In this study, TssD-5 demonstrated the highest sensitivity (71%) and good specificity (96%) within the cohort of sera samples. Furthermore, TssD-5 demonstrated minimal antibody response in healthy controls as compared to the other 3 antigens with only one weakly positive response observed in disease controls. This species-specific feature is valuable in endemic regions where high sero-positivity in healthy individuals poses a challenge to accurate diagnosis of melioidosis [29]. Omp3, on the other hand, demonstrated only moderate sensitivity (59%) although an earlier study on the same protein reported higher sensitivity (95%) [30], which may be attributed to different cut-off values used. In our study, control sera were obtained from a cohort of Malaysian individuals and therefore classified as from an endemic region. As such, it is not surprising that baseline sero-positivity (cut-off value) was high in the healthy population indicating possible previous exposure to B. pseudomallei. Of the previously reported immunogenic proteins, smBpF4 and Omp85 exhibited moderate to poor sensitivity of 41% and 19%, respectively, but showed good specificity (89-96%). Nonetheless, a PCR assay targeting the mprA gene of B. pseudomallei demonstrated 100% sensitivity and specificity in detecting the gene in clinical samples [31, 32]. This suggests that whilst the mprA (smBpF4) gene may be a good bioindicator of B. pseudomallei by PCR, the protein may be unsuitable for serodiagnosis of melioidosis.

The evaluation of a cocktail of 4 proteins gave an improvement in specificity (99%) but the sensitivity (65%) was somewhat lower than TssD-5 alone (Table 3 and Figure 1). This finding was unexpected as the combination of antigens was predicted to improve the sensitivity or be equally sensitive as TssD-5. A similar finding was reported by Lyashchenko et al. [33] where combinatorial cocktails of 4 or 8 antigens resulted in significant loss of sensitivity for detection of Mycobacterium tuberculosis compared to the use of a single antigen. This reduced sensitivity may be explained by the limited protein binding capacity of the polystyrene solid phase in the microtitre plate, resulting in the reduced display of epitopes or proteins competing with each other for binding to polystyrene which may result in larger proteins swathing their smaller counterparts [33].

Zhang et al. [25] demonstrated that the heterogeneous antibody response to different antigens could be utilised to increase the sensitivity of diagnosis. This laid the foundation for the development of a multiple-antigen ELISA system for screening of tuberculosis with improved sensitivity and specificity over conventional diagnostic methods [25]. We used a similar approach by combining the 4 antigens, to achieve an improved sensitivity from 71% using a single antigen to 88.2% whilst retaining good specificity (96%).

Using a B. pseudomallei protein microarray, Felgner et al. [34] identified 49 potential serodiagnostic antigens for melioidosis including Omp3 confirming the potential of Omp3 as a candidate antigen. However, to date, none of these antigens have been proposed as a sensitive or specific diagnostic tool for melioidosis. A large number of these serodiagnostic proteins were classified as extracellular or outer membrane proteins as predicted by the computational annotation PSORTb and most of them are encoded on Chromosome 2 of B. pseudomallei K96243. TssD-5 satisfies both criteria and in addition, the up-regulation of the TssD-5 gene transcript during the bacteria’s adaptation to the intracellular environment proposes that TssD-5 is a potentially better biomarker candidate for diagnosis of melioidosis. Furthermore, our study suggests that protein size may be an important factor to consider when identifying sensitive sero-diagnostic antigens. TssD-5 (22.9 kDa) and Omp3 (27 kDa) demonstrated better diagnostic values compared to the larger proteins, smBpF4 (55 kDa) and Omp85 (85 kDa). Indeed, among 49 potential diagnostic proteins for melioidosis reported by Felger et al., more than a half were < 60 kDa [34]. Smaller molecular weight proteins have also been reported as sensitive diagnostic antigens for tuberculosis [25], acute leptospirosis [35] and Brucella abortus infection [36]. In summary, the present study demonstrated that TssD-5 is a potentially superior diagnostic candidate for melioidosis.

We screened 4 recombinant B. pseudomallei proteins as potential diagnostic markers for melioidosis in an ELISA system. The recombinant Type VI secretion system protein TssD-5, achieved 71% sensitivity and 96% specificity, which is superior to the culture method with a recently reported sensitivity of 60%. The use of a multiple-antigen ELISA demonstrated better sensitivity (88.2%) while retaining good specificity (96%). Thus, TssD-5 in combination with the other 3 recombinant proteins tested demonstrated good potential as safe and sensitive serodiagnostic agents for melioidosis.