Detection of Huanglongbing Disease Using Differential Mobility Spectrometry

Abstract

The viability of the multibillion dollar global citrus industry is threatened by the "green menace", citrus greening disease (Huanglongbing, HLB), caused by the bacterial pathogen Candidatus Liberibacter. The long asymptomatic stage of HLB makes it challenging to detect emerging regional infections early to limit disease spread. We have established a novel method of disease detection based on chemical analysis of released volatile organic compounds (VOCs) that emanate from infected trees. We found that the biomarkers "fingerprint" is specific to the causal pathogen and could be interpreted using analytical methods such as gas chromatography/mass spectrometry (GC/MS) and gas chromatography/differential mobility spectrometry (GC/DMS). This VOC-based disease detection method has a high accuracy of ∼90% throughout the year, approaching 100% under optimal testing conditions, even at very early stages of infection where other methods are not adequate. Detecting early infection based on VOCs precedes visual symptoms and DNA-based detection techniques (real-time polymerase chain reaction, RT-PCR) and can be performed at a substantially lower cost and with rapid field deployment.

Evaluating plant immunity using mass spectrometry-based metabolomics workflows

“…More than 40 compounds were identified that can be used as biomarkers of disease, potentially before visual or non-volatile chemical traits can be observed. A similar volatile biomarker-discovery study was described for Citrus sinensis infection with Candidatus liberibacter (Aksenov et al, 2014), highlighting the potential use of chemical biomarkers of disease with non-destructive sampling during the growing season.…”

A review of techniques for detecting Huanglongbing (greening) in citrus

“…The results showed that the sensor clearly highlighted the starch accumulation in the HLB-infected leaf and differentiated it from visually analogous symptoms of zinc deficiency (Pourreza et al 2015). Aksenov et al (2014) developed a method of disease detection based on chemical analysis of released volatile organic compounds (VOCs) that emanate from infected trees. They found that the biomarkers' "fingerprint" is specific to the causal pathogen and could be interpreted using analytical methods, such as gas chromatographymass spectrometry (GC-MS) and gas chromatographydifferential mobility spectrometry (GC-DMS).…”

Supervised semi-automated data analysis software for gas chromatography / differential mobility spectrometry (GC/DMS) metabolomics applications

“…Pyrolysis helps to provide biomarker liberation from bacteria/endospores (Cheung et al 2009; Eiceman et al 2006; Krebs et al 2006b; Prasad et al 2008), which can then be separated and detected by DMS. Gas chromatography allows for a modest pre-separation of a very complex mixture before additional DMS separation and detection (Aksenov et al 2014; Arasaradnam et al 2014a; Arasaradnam et al 2014b; Basanta et al 2010; Covington et al 2013; Rutolo et al 2014; Schivo et al 2013). The data sets that result are usually extremely complex and multi-dimensional, with some data features being very abundant but perhaps unimportant, and other data features having small abundance but high significance.…”

“…Linear discriminant analysis (Covington et al 2013; Rutolo et al 2014) and Fisher discriminant analysis (Arasaradnam et al 2014b) have been used as classification methods for biological data as well. Partial least squares (Aksenov et al 2014; Cheung et al 2009) is well established and a valuable method to classify groups of GC/DMS data, and our software system has initially included this analytical tool for model building. A description of our software applying this technique (PCA, PLS, background correction and smoothing) to data that have been previously published (Aksenov et al 2014) is presented as an illustrative example throughout this paper.…”

“…Partial least squares (Aksenov et al 2014; Cheung et al 2009) is well established and a valuable method to classify groups of GC/DMS data, and our software system has initially included this analytical tool for model building. A description of our software applying this technique (PCA, PLS, background correction and smoothing) to data that have been previously published (Aksenov et al 2014) is presented as an illustrative example throughout this paper. The software is designed so that additional capabilities and applications can be added in future versions.…”