Rapid detection of Mycobacterium tuberculosis cultures by direct ambient corona discharge ionization mass spectrometry of volatile metabolites

Abstract

The effective control of tuberculosis (TB) heavily relies on global population screening, but the current diagnostic routines cannot fully meet the demanded specificity and speed of TB recognition. Here we report on the detection Mycobacterium tuberculosis (MTB), which is the primary cause of TB infection, using ambient corona discharge ionization mass spectrometry analysis of volatile organic compounds (VOCs) emitted by bacterial cultures. For the initial MTB inoculum of 10⁴ colony-forming units (CFU) per mL, highly specific differential volatile metabolites of MTB, such as methyl nicotinate and benzoic acid, 4-methoxy-, methyl ester, could be detected after 15–20 days of incubation (Löwenstein–Jensen medium). The throughput of analysis was ca. 3 samples per minute, which is much faster than in traditional techniques. Overall, our results indicate the high suitability of ambient corona discharge ionization mass spectrometry for the rapid and cost-efficient TB screening using commonly available MS instrumentation.

---

Tuberculosis (TB) kills more than 4000 people each day and now ranks alongside HIV as a leading cause of death worldwide.¹ In the absence of an effective vaccine, TB control largely relies on global population screening. Still, the primary diagnostic technique is acid-alcohol fast stain complemented by culture and PCR techniques in the developed world. The current diagnostic procedure of TB worldwide is primarily based on Ziehl–Neelsen staining and microscopic inspection of sputum cultures. Unfortunately, these approaches lack sufficient specificity and cannot reliably differentiate between the causative agent of TB, Mycobacterium tuberculosis (MTB), and nontuberculous mycobacteria (NTM).^2,3 Additional tests are required to differentiate between MTB and NTM based on the growth rate at different temperatures, colony morphology pigment production and drug susceptibility, which are resourceful, time-consuming and require manual manipulation by qualified personnel. The outstanding challenge is to reduce the operational time and costs associated with the diagnosis of TB worldwide.⁴

A number of powerful approaches have been developed for the MTB detection based on nucleic acid amplification,⁵ gas-chromatography mass spectrometry (GC-MS)⁶ and liquid chromatography mass spectrometry (LC-MS).⁷ GC-MS and LC-MS can provide a result within just a few hours of analysis, obviating bacterial culture routine and thus allowing point-of-care diagnosis. Unfortunately, the integration of these approaches into resource-limited settings is restrained due to the high cost and lack of infrastructure and expertise.^8,9 Another limitation of these approaches is associated with a relatively long instrument time per sample (1–2 h) and the requirement of sample pretreatment, which hinders the use of these approaches for the screening of large sample arrays. Out of the recently introduced approaches, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) has by far been the most commercially successful approach for the identification of TB and other bacteria.^10–12 Several dedicated MALDI-TOF platforms for bacterial identification are already available on the market from various MS vendors. For TB detection by MALDI-MS samples usually need to be cultured for ca. 2 weeks, but the MS analysis on average requires only ca. 6 min per sample at 3–5 times reduced cost of identification compared to traditional methods.¹²

Over recent years, identification of bacteria via the direct analysis of emitted volatile organic compounds (VOCs) has received attention, owing to the non-invasiveness, practical simplicity, cost-efficiency, high speed and toxicological safety of analysis.^13–18 A number of VOCs specific to MTB have been identified in earlier GC-MS studies.^19,20 Direct VOC detection of these metabolites without sample collection and chromatographic separation may considerably improve the cost and time efficiency of TB detection. In our previous studies we have demonstrated the ambient corona discharge ionization MS as a practical platform for the rapid analysis of bacterial VOCs.^21–23 The method is a variation of the classical atmospheric pressure chemical ionization (APCI)²⁴ in which bacterial VOCs are transported to the tip of the discharge needle using room-temperature nitrogen gas without carrier solvent and accessory heating. Here we extend the investigation of ambient corona discharge ionization MS toward TB detection.

Fig. 1a shows VOC fingerprint of MTB from intradermal injection of Bacillus Calmette–Guérin (BCG) vaccine (Chengdu Institute of Biological Products Co. Ltd., Sichuan, China) grown in a Mccartney bottle on solid Löwenstein–Jensen medium (Baso Biological Co. Ltd., Beijing, China) at 35 °C with 5% CO₂ over 20 days starting from the inoculum concentration 10⁴ CFU per mL. VOCs from the headspace of culture bottles were continuously transferred into ionization region via plastic tubing (ID 1.0 mm) assisted by nitrogen gas flow (0.1 MPa, 1 L min⁻¹) and directly ionized in front of a mass spectrometer (LTQ-XL and Orbitrap-XL, Thermo Scientific, San Jose, CA, USA). Ambient corona discharge was created by applying +4 kV to a stainless steel needle (OD 150 μm) with a sharp end (curvature radius ∼7.5 μm). The angle between the discharge needle and the outlet of the sample tubing was 30°. The distance from the tip of the needle to the end of the outlet tubing was 2 mm. The distance from the tip of the needle to the MS inlet capillary was 6 mm. The scheme of the ion source is available free of charge in the online ESI material of our previous publication.²¹ Mass spectrum was averaged in the m/z 60–200 range from the 20 scans obtained over the sampling period of 10 s. The mass spectrum was converted into MTB-specific VOC fingerprint by the subtraction of signals from non-inoculated growth medium incubated under the same conditions. MTB specific signals, i.e. those only present in the MTB spectrum but not in the pure medium spectrum are listed in Table 1. The most intense signal at m/z 138.055 corresponds to methyl nicotinate. The structural assignment was confirmed by reference MS/MS analysis of authentic compound (Fig. 2). Methyl nicotinate signal was also the earliest detected MTB specific signal (Table 1, Fig. 3). Methyl nicotinate signal became visible in all the tested replicate samples (at least 4) after 15 days of incubation at the initial inoculum concentration of 10⁴ CFU per mL and after 12 days of incubation at the initial inoculum concentration of 10⁵ CFU per mL (Fig. 3). Note that for each incubation time and initial MTB concentration in Fig. 3 the replicate samples were prepared and analyzed independently. Importantly, methyl nicotinate signal was not observed in the headspace of tested NTMs, M. smegmatis and M. turtle, incubated under identical conditions (Fig. 1b, Table 1), suggesting that methyl nicotinate could be a useful differential biomarker of MTB.

Fig. 1 VOC fingerprints of MTB (a) and M. smegmatis (b) cultures produced by ambient corona discharge ionization mass spectrometry.

Table 1 MTB VOCs observed by ambient corona discharge ionization mass spectrometry

m/z^a	Formula	Fragments^b	Structure^c	Visible after^d (Days)	Specificity^e
a Measured by Orbitrap-MS with 5 ppm accuracy.b Produced by collision-induced dissociation.c The structures for m/z 138 and for m/z 167 were confirmed by reference analysis of authentic compounds.d Data for the initial MTB inoculum concentration 10^4 CFU per mL.e NTM = also visible in M. smegmatis and/or M. turtle cultures.
138.0549	C7H8NO2	123, 108, 106, 96, 94, 78		15	MTB
140.1115	C8H14NO	122, 112, 98, 96, 84, 70, 69		18	MTB/NTM
167.0704	C9H11O3	149, 135, 123, 108		20	MTB
79.0211	C2H7SO	64		25	MTB/NTM
74.0599	C3H8NO	N/A		25	MTB
96.0483	C5H6NO	N/A	N/A	25	MTB
106.0508	C3H8NO3	91, 88, 79, 58	N/A	25	MTB
114.0429	C4H6N2O2	N/A	N/A	25	MTB
129.1027	C6H8NO2	N/A	N/A	25	MTB/NTM
118.099	C6H14O2	101, 100, 72	N/A	25	MTB/NTM
161.111	C7H17N2S	129, 97, 69	N/A	25	MTB

---

Fig. 2 MS/MS spectra of m/z 138 signal produced from MTB culture (a) and from authentic methyl nicotinate (b) under identical experimental parameters (collision energy 40%).

Fig. 3 (a) The intensity of most abundant MTB-specific signals as a function of culture time (the concentration of original MTB inoculum is 10⁴ CFU per mL). (b) The intensity of methyl nicotinate signal (m/z 138) produced by MTB cultures at different initial MTB concentrations. For each data point at least four independently grown replicate samples were analyzed.

Even though in this work we only investigated the VOCs of two NTM strains, the high specificity of methyl nicotinate release to MTB is evidenced by earlier studies done on larger selection of NTM strains using different detection approaches.^19,20 Syhre and Chambers used the combination of solid phase microextraction (SPME) with GC/MS analysis to systematically study volatiles in the headspace of various mycobacteria, including MTB, M. bovis, M. avium, M. fortuitum, M. chelonae and M. abcsessus.²⁰ Methyl nicotinate was only detected in MTB and M. bovis cultures but not in the rest of mycobacteria for all the growth media tested, including Löwenstein–Jensen/glycerol, sheep blood agar and the liquid BacT/Alert™ MP. In another GC/MS study Mgode et al. identified methyl nicotinate in the volatile headspace of various MTB strains but not in the headspace of tested NTM species, including M. smegmatis, M. avium, M. scrofulaceum, M. vaccae, M. aichiense, M. aurum, M. neoaurum, grown on media with various chemical composition.¹⁹ Further, earlier studies reported no trace of methyl nicotinate in VOC headspace of most common respiratory pathogens, including A. fumigatus, A. flavus, A. niger, A. terreus, Fusarium spp., R. arrhizus, S. apiospermum, C. albicans, P. aeruginosa, B. cepacia, P. fluorescens, S. aureus, E. coli, S. pneumoniae, M. catarrhalis and H. influenza.^19,22,25 Altogether, these data indicate that methyl nicotinate can be considered as a highly specific differential VOC biomarker of MTB.

The entire list of MTB-specific VOC signals observed in our study is shown in Table 1. Tentative chemical assignment of MS signals was done based on hi-res mass measurements of parent ions in full MS mode and fragment ions in MS/MS mode. Compounds with high proton affinity (PA) are mostly observed, e.g., those with secondary or tertiary amine functionality. Ambient ionization of VOCs by corona discharge occurs in a ladder-like fashion in which protons are gradually transferred from compounds with lower PA values to compounds with higher PA values.²⁴ This greatly favors the observation of high-PA VOCs even at a trace level. In a model experiment methyl nicotinate could be detected in the volatile headspace of aqueous solution down to ca. 100 ppb concentration (LTQ-MS). The high sensitivity of detection to methyl nicotinate is directly related to the high PA value of this compound (925.6 kJ mol⁻¹).²⁶ This high chemical sensitivity allows the observation of methyl nicotinate in the volatile headspace of TB cultures without any vapor collection. Indeed, the ultimate sensitivity of analysis also depends on the type of a mass spectrometer used for detection. In contrast to the high-PA VOCs, ambient corona discharge ionization of low-PA VOCs, such as acids, alcohols, aldehydes, hydrocarbons, etc., is extremely inefficient. As a result, a number of MTB VOCs discovered by GC/MS and other ionization approaches^6,19,20,27 are not visible using ambient ionization by corona discharge. Out of the VOCs tentatively identified in this study, methyl nicotinate and benzoic acid, 4-methoxy-, methyl ester were reported in earlier studies as highly specific MTB metabolites.^19,20

An important merit of ambient corona discharge MS as far as the problem of global TB screening is concerned is the high speed and throughput of sampling. At present we can routinely achieve scanning rate of ca. 3 samples per min, i.e. ca. 200 samples per h (Fig. 4). The high throughput of sampling is related to the technical simplicity of experimental procedure and the efficient obviation of sample carry-over effects in VOC analysis. We estimate that the sampling throughput can be further increased using robotized sampling procedure, similar to that in commercial GC instruments. For comparison, the throughput of MALDI-MS analysis is ca. 10 samples per h (ref. 12) and the throughput of GC-MS/LC-MS analysis is on the order of 1 sample per h. Therefore, even though TB identification in ambient corona discharge MS requires longer culture time than in GC-MS or LC-MS, the reported method is better suited for the purpose of high-throughput population screening, when large arrays of cultures need to be processed. In contrast, chromatographic approaches can be recommended when an urgent diagnosis is required for a certain individual, because those approaches require shorter culture time.

Fig. 4 Single ion chromatogram of methyl nicotinate signal (m/z 138) showing six independent samplings of MTB cultures.

Overall, our results indicate that ambient corona discharge MS has a potential to significantly increase the current throughput of population screening and enhance the safeguard against TB threat. In addition, ambient corona discharge MS also features the low cost and high stability of operation owing to the total obviation of sample preparation, solvents and chemical reagents.²³ Importantly, the method can be directly implemented on any type of a mass spectrometer with atmospheric interface and is extremely easy to be learned by people without deep MS knowledge, promising broad usability. In a follow-up work we will systematically examine the accuracy of the method for clinical application using samples from clinical patients diagnosed with TB using conventional methods.

Acknowledgements

We gratefully acknowledge financial support from the National Natural Science Foundation of China (NSFC) (no. 21305012, no. 81460327), Program for Changjiang Scholars and Innovative Research Team in Universities (PCSIRT) (no. IRT13054), Science and Technology Planning Project at the Ministry of Science and Technology of Jiangxi Province, China (no. 20152ACB21013, no. 2015ACG70014).

References

1. W. H. Organization, in Global tuberculosis report 2015, ed. W. H. O. Global TB Programme, Global TB Programme, World Health Organization, Geneva, Switzerland, 20th edn, 2015.
2. S. E. Dorman, Clin. Infect. Dis., 2010, 50, S173–S177.
3. A. H. J. Kolk, J. J. B. N. van Berkel, M. M. Claassens, E. Walters, S. Kuijper, J. W. Dallinga and F. J. van Schooten, Int. J. Tubercul. Lung Dis., 2012, 16, 777–782.
4. T. H. M. Ottenhoff, J. J. Ellner and S. H. E. Kaufmann, Tuberculosis, 2012, 92, S17–S20.
5. J. L. Davis, L. Huang, W. Worodria, H. Masur, A. Cattamanchi, C. Huber, C. Miller, P. S. Conville, P. Murray and J. A. Kovacs, PLoS One, 2011, 6, e16321.
6. N. A. Dang, H.-G. Janssen and A. H. J. Kolk, Bioanalysis, 2013, 5, 3079–3097.
7. V. J. LaBombardi, R. Katariwala and G. Pipia, Clin. Microbiol. Infect., 2006, 12, 478–481.
8. F. A. Drobniewski, M. Caws, A. Gibson and D. Young, Lancet Infect. Dis., 2003, 3, 141–147.
9. N. W. Schluger, Am. J. Respir. Crit. Care Med., 2001, 164, 2020–2024.
10. S. Biswas and J.-M. Rolain, J. Microbiol. Methods, 2013, 92, 14–24.
11. S. K. P. Lau, B. S. F. Tang, J. L. L. Teng, T.-M. Chan, S. O. T. Curreem, R. Y. Y. Fan, R. H. Y. Ng, J. F. W. Chan, K.-Y. Yuen and P. C. Y. Woo, J. Clin. Pathol., 2014, 67, 361–366.
12. P. Seng, M. Drancourt, F. Gouriet, B. La Scola, P.-E. Fournier, J. M. Rolain and D. Raoult, Clin. Infect. Dis., 2009, 49, 543–551.
13. M. Shirasu and K. Touhara, J. Biochem., 2011, 150, 257–266.
14. L. D. J. Bos, P. J. Sterk and M. J. Schultz, PLoS Pathog., 2013, 9, e1003311.
15. E. Tait, J. D. Perry, S. P. Stanforth and J. R. Dean, TrAC, Trends Anal. Chem., 2014, 53, 117–125.
16. D. Smith and P. Španěl, TrAC, Trends Anal. Chem., 2011, 30, 945–959.
17. C. L. Lonsdale, B. Taba, N. Queralto, R. A. Lukaszewski, R. A. Martino, P. A. Rhodes and S. H. Lim, PLoS One, 2013, 8, e62726.
18. K. Chingin, J. C. Liang and H. W. Chen, RSC Adv., 2014, 4, 5768–5781.
19. G. F. Mgode, B. J. Weetjens, T. Nawrath, D. Lazar, C. Cox, M. Jubitana, A. Mahoney, D. Kuipers, R. S. Machang'u, J. Weiner, S. Schulz and S. H. E. Kaufmann, Tuberculosis, 2012, 92, 535–542.
20. M. Syhre and S. T. Chambers, Tuberculosis, 2008, 88, 317–323.
21. J. Liang, Y. Hang, K. Chingin, L. Hu and H. W. Chen, RSC Adv., 2014, 4, 25326–25329.
22. K. Chingin, J. Liang, Y. Hang, L. Hu and H. W. Chen, RSC Adv., 2015, 5, 13952–13957.
23. L. Hu, J. Liang, K. Chingin, Y. Hang, X. Wu and H. W. Chen, RSC Adv., 2016, 6, 8449–8455.
24. T. R. Covey, B. A. Thomson and B. B. Schneider, Mass Spectrom. Rev., 2009, 28, 870–897.
25. M. Syhre, J. M. Scotter and S. T. Chambers, Med. Mycol., 2008, 46, 209–215.
26. E. P. L. Hunter and S. G. Lias, J. Phys. Chem. Ref. Data, 1998, 27, 413–656.
27. N. A. Dang, M. Mourao, S. Kuijper, E. Walters, H. G. Janssen and A. H. J. Kolk, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2015, 986–987, 115–122.

---