The identification of nontuberculous mycobacteria (NTM) by proteomic procedures has improved in the last few years. Strains from different geographical locations differ in their proteomic patterns, limiting the applicability of general databases for accurate identifications.
MethodsWe have optimized an alternative extraction protocol to Myco-Ex, and MBT, the protocols recommended by Bruker Daltonics, the manufacturer of MALDI Biotyper, a Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS) system. 120 clinical isolates, one environmental strain (Mycobacterium canariasense) and 7 quality controls were tested from the Spanish Society of Clinical Microbiology and Infectious Diseases (SEIMC). All strains were also sent to the Laboratory of Mycobacteria belonging to the National Spanish Reference Center (Instituto de Salud Carlos III), that uses several molecular methods for mycobacterial identification. The proteomic extraction method includes a single extraction/inactivation formic acid step. 30 NTM strains were analyzed to certify the biosafety of the procedure, testing their viability after this procedure.
ResultsThe score ranges of the isolates were 1.61–2.16 (mean 1, 84) for the 88 isolates of the slow growing mycobacteria group (SGM) and 1.66–2.33 (mean 2.06) for the 40 isolates of the rapid growing mycobacteria group (RGM). Isolates exposed to formic acid were incubated for 8 weeks and growth was observed in neither the clinical isolates nor the controls.
ConclusionsWe have developed a simpler and faster procedure for the proteomic identification of nontuberculous mycobacteria (MALDI-TOF) retaining significant scores without compromising biosafety.
La identificación de micobacterias no tuberculosas (NTM) por técnicas proteómicas ha avanzado considerablemente en los últimos años. Aislamientos procedentes de diferentes localizaciones geográficas difieren en sus patrones proteómicos, limitando la aplicabilidad de las bases de datos y la exactitud de las identificaciones.
MétodosHemos optimizado un sistema de extracción alternativo a los recomendados por el fabricante de MALDI Biotyper, Bruker Daltonics, que son Myco-Ex and MBT. Se han probado 120 aislamientos clínicos de NTM, una cepa ambiental (M. canariasense) y 7 controles de calidad de la Sociedad Española de Enfermedades Infecciosas y Microbiología Clínica (SEIMC). Todos los aislamientos se enviaron al Laboratorio español de Referencia (Instituto de Salud Carlos III), que utiliza distintos métodos moleculares para la identificación de micobacterias. El método de extracción proteómico consiste en único paso basado en extracción/inactivación con ácido fórmico. Se probó la viabilidad de 30 aislamientos de NTM sometidos a dicho procedimiento de extracción.
ResultadosLos rangos de scores obtenidos para los 88 aislamientos de micobacterias de crecimiento lento (SGM) fueron 1,61 – 2,16 (con una media de 1,84). Para los 40 aislamientos de micobacterias de crecimiento rápido (RGM) los scores obtenidos fueron 1,66 – 2,33 (con una media de 2,06). Respecto a los 30 controles de este procedimento y expuestos a ácido fórmico, a las 8 semanas de incubación no se obtuvo crecimiento en ninguno de ellos.
ConclusionesHemos desarrollado un método de extracción sencillo y más rápido para la identificación proteómica de micobacterias no tuberculosas (MALDI-TOF), manteniendo scores significativos sin comprometer la bioseguridad.
Nontuberculous mycobacteria (NTM) are a diverse group of microorganisms that are gaining attention due to many factors. An increasing incidence has been reported in different countries of the world and there was an increase in the number of NTM isolated from clinical samples from patients. The isolation of the most common species is constantly changing in most geographical areas, and newer species are emerging due to better diagnostic techniques to detect and identify NTM.1,2
Classically, mycobacterial identification relied on biochemical tests, but new molecular techniques have largely replaced them, and now there are several commercially available PCR tests for NTM detection in clinical samples.3 Normally, NTM identification requires previous growth on solid or automated liquid media. A three-step procedure is commonly followed, including DNA extraction, multiplex amplification with biotinylated primers and reverse hybridization.4 As an alternative based on proteomics, many clinical laboratories have implemented Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS) in the last decade for bacterial identification with excellent results. Mass spectrometry allows the analysis of a portion of mycobacterial proteome including ribosomal proteins ranging from 20 to 20,000Da. This successful approach is suitable for general bacteria, but the special configuration of mycobacteria complicates the procedure. Structural analysis reveals that, within eubacteria, mycobacterial ribosome composition and structure exhibits marked differences with that of Escherichia coli5 because mycobacterial ribosomal proteins are more abundant, bigger in size (like r-proteins) and more complex. Detailed information about their structures and function is outlined by Hentschel et al.6 Mycobacterial identification depends on factors such as the extraction method, the protein profile or the number and type of species in the library. The extraction method is one of the most controversial due to the complexity of the mycobacterial cell wall. For this reason, several approaches have been proposed by manufacturers for protein release from these structures.7 Current methods require a long processing time involving inactivation, ethanol extraction, zirconia beads shaking, optional sonication step and acetonitrile and formic acid extraction. The score cut-offs recommended by Bruker Daltonics, manufacturers of MALDI Biotyper, are: <1.60 as no reliable identification, 1.60–1.79 as low confidence identification and > or =1.80 as high confidence identification. In a multicenter study a total of 98 NTM clinical isolates were identified in 15 laboratories from 9 european countries with the MALDI-TOF Biotyper. A score>1.6 was obtained for 100% of isolates in 5 laboratories following the standard Myco-Ex extraction method recommended by the manufacturer.8 The purpose of this work is to reduce the extraction time for NTM, simplifying the sample processing to the level of the applied to other common bacteria. The accuracy of the resulting scores of isolates is checked on a parallel identification by molecular procedures (PRA-HSP 65,9 Genotype CM or AS Hain) at the Spanish National Reference Center (ISC-III). After the formic acid extraction step, a viability assay has been performed to ensure the safety of the technique.
Materials and methodsBacterial strainsA total of 128 isolates were tested. Most of the strains (120) were clinical isolates, one was from environmental origin and the remaining seven were quality control isolates belonging to the Spanish Society of Clinical Microbiology and Infectious Diseases (SEIMC, Table 1). A loop of a fresh culture was streaked on Middlebrook 7H11 agar plates and incubated at 37°C. Plates were inspected weekly for colony growth. When the bacterial mass was appropriate, colonies were picked-up and the extraction system for common bacteria proposed for MALDI-TOF MS (Bruker®) was followed
Proteomic analysis of nontuberculous mycobacteria strains (MALDI-TOF Brucker®) showing the strain number and type of MNT, quality control strains, median score (by strain or type), median score repetitions and species identification conflicts (PRA-HSP-65).
| Nontuberculous mycobacteria species | Number of isolatesN=128 | Clinical strainsN=121 | Quality Control StrainsN=7 | Mean score | Mean score repeat | Identification conflicts |
|---|---|---|---|---|---|---|
| Slow Growing Mycobacteria (SGM) | 88 | 83 | 5 | |||
| M. avium | 23 | 23 | 1.86 | 9.70 | None | |
| M. chimaera/IC | 11 | 10 | 1 | 1.95 | 9.50 | None |
| M. lentiflavum | 10 | 10 | 2.16 | 9.90 | None | |
| M. gordonae | 5 | 5 | 1.81 | 9.20 | None | |
| M. celatum | 3 | 3 | 1.98 | 10 | None | |
| M. paragordonae | 3 | 2 | 1 | 1.73 | 6.50 | None |
| M. kumamotonense | 1 | 1 | 1.65 | 10 | None | |
| M. malmoense | 1 | 1 | 1.95 | 9 | None | |
| M. xenopi | 9 | 9 | 1.66 | 7.70 | None | |
| M. kansasii | 8 | 7 | 1 | 1.95 | 8.30 | None |
| M. marinum | 2 | 1 | 1 | 1.87 | 10 | None |
| M. paraffinicum | 1 | 1 | 1.61 | 3 | None | |
| M. mucogenicum | 2 | 2 | 1.96 | 10 | None | |
| M. interjectum | 7 | 7 | 1.58 | 4.71 | None | |
| M. terrae | 1 | 1 | 1.90 | 2 | None | |
| M. parascrofulaceum | 1 | 1 | 1.79 | 10 | None | |
| Median score SGM | 1.84 | 8.09 | ||||
| Rapidly Growing Mycobacteria (RGM) | 40 | 38 | 2 | |||
| M. chelonae | 8 | 8 | 1.80 | 6.50 | M. stephanolepidis | |
| M. abscessus | 13 | 13 | 1.83 | 6.70 | None | |
| M. mageritense | 3 | 3 | 2.29 | 9 | None | |
| M. peregrinum | 2 | 2 | 2.10 | 9.50 | None | |
| M. elephantis | 4 | 4 | 2.13 | 10 | None | |
| M. porcinum | 3 | 3 | 2.11 | 7.60 | None | |
| M. sépticum | 2 | 1 | 1 | 2.02 | 7.50 | None |
| M. holsaticum | 1 | 1 | 1.66 | 3 | None | |
| M. fortuitum | 2 | 2 | 2.33 | 10 | None | |
| M. canariasense | 2 | 1 | 1 | 2.32 | 8 | None |
| Median score RGM | 2.10 | 7.80 | ||||
| N° total strains | 128 | 121 | 7 | |||
The standard guidelines for common bacteria recommended by the manufacturer (MALDI Biotyper MBT Smart apparatus from Bruker) were followed. A full loop of a colony was inoculated in an Eppendorf containing 2μl of 70% formic acid and incubated for 10min. The pellet was then placed on the measuring cell of the plate and allowed to dry. Finally, a drop of matrix was added and let to dry. Measures were performed with the latest version of data base (v6.0) containing 1.143 Main Spectra (MSPs) that correspond to mycobacterial strains. The reading method used was MALDITOF-MS MBT AutoX with a mass range between 0 and 20,000Da. Two readings were executed, the first one was automatic (AutoXecute) and the second one manual. Each strain was analyzed twice, one loop for the standard mode and another for the mycobacterial mode. Only the highest score obtained from the best mode of reading and the repetition rate (number of the times matching the identification that reaches the highest score) was taken into account. The cut-off scores used were those recommended by the manufacturer, <1.60 as not reliable identification, 1.60–1.79 as low confidence identification and 1.80 and higher as high confidence identification.
Identification accuracy
Strains were also sent to the Laboratory of Mycobacteria belonging to the National Spanish Reference Center (Instituto de Salud Carlos III), that uses several molecular methods for mycobacterial identification, including PRA HSP65, GenoType CM or AS, Hain).
Measurement of mycobacterial viability after protein extractionAn amount of 30 isolates belonging to 17 different MNTs species were selected as follows in Table 2. The slow growing MNTs were a total of 18 isolates belonging to 9 different species. The rapidly growing MNTs were 12 isolates belonging to 8 species. A loop of a fresh Middelbrook 7H11 agar grown strain was mixed in a 10μl 70% formic acid in an ependorf for 10min. After processing, a loop of each pellet was striked in a fresh Middlebrook 7H11 agar plates and incubate at 37°C for eight weeks. The plates were inspected weekly for colony growth.
Number and type of nontuberculous mycobacterial strains tested for viability after exposition to 70% formic acid protein extraction procedure.
| Nontuberculous mycobacteria species | Number of isolatesN=30 | Clinical strainsN=24 | Quality control strainsN=6 |
|---|---|---|---|
| Slow Growing Mycobacteria (SGM) | 18 | 15 | 3 |
| M. avium | 3 | 3 | |
| M. chimaera/IC | 2 | 1 | 1 |
| M. lentiflavum | 3 | 3 | |
| M. xenopi | 2 | 2 | |
| M. kansasii | 2 | 1 | 1 |
| M. marinum | 2 | 1 | 1 |
| M. interjectum | 2 | 2 | |
| M. malmoense | 1 | 1 | |
| M. parascrofulaceum | 1 | 1 | |
| Rapidly Growing Mycobacteria (RGM) | 12 | 9 | 3 |
| M. chelonae | 1 | 1 | |
| M. abscessus | 3 | 3 | |
| M. mageritense | 2 | 2 | |
| M. elephantis | 1 | 1 | |
| M. sépticum | 2 | 1 | 1 |
| M. terrae | 1 | 1 | |
| M. fortuitum | 1 | 1 | |
| M. canariasense | 1 | 1 | |
A detailed description is available in Table 1. The scores of the SGM group (n=88) ranged between 1.58 and 2.16 with a mean of 1.80. The average number of repetitions for this SGM group is 8.10. The scores of RGM (n=40) ranged between 1.66 and 2.33 with a mean of 2.10. The average number of repetitions for this RGM group was 7.80. The scores of the quality control strains (n=7) were as follows: in the SGM group (n=5) ranged between 1.74 and 2.12 with a mean of 1.93 and the average number of repetitions was 8.40. The scores of the RGM group (n=2) ranged between 1.86 and 2.44, with a mean of 2.15. The average number of repetitions was 8. A single discrepancy was observed, because one isolate identified by MALDI-TOF MS as Mycobacterium stephanolepidis (score=2.10) and two repetitions) was characterized as Mycobacterium chelonae by molecular biology (PRA-HSP65) in the Spanish Reference Center (ISC-III). It should be noted seven of the identification scores in the repetition list of this isolate were M. chelonae, the first of them possessing a score of 1.87. In summary, in 10 of the 16 SGM species tested (62.5%) including the most common SGM species (Mycobacterium avium, Mycobacterium chimaera, Mycobacterium lentiflavum or Mycobacterium kansasii) and in 9 of the 10 RGM species (90%) as common as Mycobacterium abscessus or M. chelonae, the resulting mean scores were higher than 1.80.
A selection of isolates (Table 2) was exposed to formic acid for protein extraction and striked on 7H11 Middlebrook agar plates. No growth could be observed for any of the cultures after 8 weeks of incubation.
DiscussionMALDI-TOF MS is a technology based on the comparison of proteomic profile of microorganisms, implemented by an increasing number of laboratories. MALDI-TOF MS procedure is fast, easy to perform, cheap and a large number of bacteria and other microorganisms may be accurately identified. However, the special characteristics of mycobacteria complicate the application of this technology. In some reports the identification accuracy by MALDI-TOF is equivalent or higher than that of other molecular approaches such as the commercially available systems such as GenoType CM or AS.10,11 In a novel commercial approach, Fluorotype Mycobacteria assay VER1.0, Giraud et al.12 have tested 119 NTM isolates (85 SGM and 34 RGM), reporting a reduction of handling time to 45min (extraction time is not included) compared to GenoType assays and reducing the accuracy level.
The identification precision depends on the extraction methods, such as the Myco-EX protocol recommended by de manufacturer Bruker, that has proved to be reproducible in a multicenter study.8 However, this procedure is cumbersome because requires numerous extraction steps, including inactivation at 95°C 30min, an ethanol extraction, centrifugation, zirconia and shaking with silica beads/acetonitrile, formic acid and matrix. In a more recent approach, the manufacturer Bruker has developed a new extraction method, MBT Mycobacteria kit IVD protocol based on formic acid and acetonitrile extraction step. The processing time announced is 45min, but the 30min extraction time is not included, finally resulting in 75min.
In a previous work, we detailed an extraction method faster than Myco-Ex, testing 106 strains from 22 different NTM species, obtaining a time processing reduction from 83min (Myco-Ex) to 59min (tested method), keeping the mean score higher than 1.80 in 67.92% of isolates.13 In other study Pastrane et al.14 tested 55 NTM strains comparing two proteomic extraction procedures recomended by the manufacturer Bruker (Myco-Ex and the new MBT) obtaining an identification accuracy of 95% and 97%. These authors reported a processing time reduction from 80min (Myco-Ex) to 40min. Taking into account the inactivation time (30min), the total processing time is 70min. In our work, the total processing time is about 15min because inactivation and extraction steps are unified, but keeping biosafety conditions controlled. The resulting protocol is similar to the used for other common bacteria, but providing an accurate identification for most of the isolates.
It has been reported that the final score depends on the growth media, solid or liquid, the number of species and isolates tested, and the laboratory procedures.14,15 This variability represents an obstacle to establish the best procedure that improves the scores. We can increase the accuracy of the system developing a combination of the tools that MALDI Biotyper offer, such as the score level, the number of repetitions, the standard or mycobacteria reading modes, the occasional manual reading and the added local NTM libraries. This approach allows the reduction of the processing time, offering a faster response to the clinician. In our experience, some species, such as M. chelonae, reach a better score when testing fresh colonies from blood agar when the presence of mycobacteria is not suspected, using the processing guidelines for common bacteria.
The number and type of species and strains of NTM in the MALDI Biotyper database increases each new edition, accounting for the current total of 1143 Main Spectra (MSP). Alcaide et al.,16 in a multicentre study, highlighted the absence of 2–17 uncommon NTM species depending on the database edition of MALDI Biotyper. Geographic differences may be the cause for misidentification of some species poorly represented in the databases. In a recent example, following the extraction method recommended by the manufacturer, the species-level accuracy was 100% and more than 99% of the isolates obtained a satisfactory log>1.60.8 Following the simplified procedure that we propose, our results indicate that some of the most common RGM (M. abscessus and M. chelonae) and SGM species (M. avium, M. chimaera, M. lentiflavum or M. kansasii) display scores higher than 1.80. Similar results were reported by Toney et al.17 for 314 NTM isolates, obtaining a 99% accuracy for RGM and 87% for SGM. Because our study analyzed a limited number of species and isolates, the suitability of our procedure for other NTM species needs to be further assessed. Following the proposed rapid method, the resulting scores are as high as 1.80, similar to other authors.14 Misidentifications by MALDI-TOF MS has been reported, but they are uncommon.18,19 In this report there was only one misidentification, a M. chelonae strain classified as Mycobacterium stepahanolepidis, a novel species closely related.20 In this case the number of the identification scores in the repetition list as M. chelonae, may suggest a potential misidentification. The manipulation procedures for mycobacteria must follow biosafety guidelines and comply with current regulations. For this reason, a serious concern is the viability of mycobacteria after the extraction procedures and the risk of transmission to the laboratory personnel. We have shown that the extraction step with formic acid inactivates a wide array of mycobacteria, suggesting that it will be effective against most, if not all, mycobacteria. This feature of the protocol avoids the addition of previous inactivation steps and the mandatory Biological Safety Cabinet (BSC) will not be needed after the extraction.
In summary, we propose a simpler and faster procedure as compared with the previous Myco-Ex and MBT extraction methods, keeping significant score levels and/or repetition rates and operating in a biosafety controlled environment. Further studies will be required to validate these results, processing more NTM species to accomplish a good reproducibility between laboratories.
FundingThis work has not been supported by public or private funds.
Conflicts of interestThe authors declare that they have no conflicts of interest.
To the Mycobacteria Laboratory of the Carlos III Health Institute for their collaboration in the molecular identification of the mycobacteria used in this study.
