Microbe Lynx System for Automatic Bacterial Identification

TheMirobeLynx rapid bacterial identification system has been developed in collaboration between Manchester Metropolitan University (MMU; Manchester, UK), the Molecular Identification Service Unit of the Health Protection Agency (MISU; London, UK), and the Waters Corporation (Manchester, UK). The system has

Table 7.1. Comparison of MALDI-TOF MS techniques for bacterial identification.

Target molecules

Sample preparation from pure culture for transfer to MALDI target plate

Identification based on

General reference

Comments

Proteins

SELDI

Cell surface proteins

Selection, amplification, PCR, transcription, cleanup.

Add lysate, mix, wash, possibly separate.

Add lysate, mix, separate by inoculation on selected SELDI target plate, wash & apply MALDI matrix Transfer of whole cell to target plate & application of MALDI matrix

DNA sequences

Selective biomarkers or selective protein identification Selective biomarkers

Fingerprints

  • Nordhoff et al., 1996)
  • Demirev et al., 1999; Jarman et al., 2000)
  • Lancashire et al., 2005)
  • Keys et al., 2004)

Lengthy and specialized sample preparation and analysis requiring a high level of expertise.

High level of expertise required for analysis of proteins.

Useful comparison of very closely related organisms, employing specialized separation technology.

Rapid, simple, automated analysis with high throughput, suitable for routine analysis.

at its core a curated database of fingerprint spectra over the mass range 500 to 10,000 Da derived from quality-controlled BS EN ISO 9001:2000 freeze-dried bacterial strains supplied from the National Collection of Type Cultures (NCTC; London, UK). The database spectra are prepared using strict protocols, to ensure reproducibility. Initially, the freeze-dried quality-controlled ampoule is rehy-drated and inoculated onto Columbia blood agar (CBA) (Oxoid Ltd, Basingstoke, UK) then incubated aerobically for 24 h at 37° C. In some cases, however, these conditions are altered to facilitate growth (e.g., for strict anaerobes). To ensure the organism has recovered fully from the stressed dehydrated state, two further subcultures are undertaken on CBA as above, prior to analysis. For analysis of "real" samples against the database, however, where the sample is unlikely to be stressed, one subculture is sufficient prior to analysis. Single colonies of the bacterium are then used to directly inoculate a minimum of 4 MALDI target wells, using a 1|L sample loop. For database preparation, however, a total or 12 target wells are used in order to (i) assess the reproducibility of the fingerprints prior to database addition and (ii) produce a statistical estimation of variance. The use of a single colony also has the advantage that different colonies on a mixed culture plate can be distinguished and identified separately. The time taken to inoculate

Sample wells

Calibration wells

Sample wells

Calibration wells

Figure 7.1. Ninety-six-well MALDI target plate, detailing sample rows A to H and the 24 external calibration wells between the sample rows.

the 96-well MALDI target plate is very short, of the order 5 to 10 min. After allowing the bacterial samples to dry for approximately 1 h, the MALDI target wells are overlaid with 1 ^L of the MALDI matrix solution to aid the ionization process. For Gram-positive organisms, this is a saturated solution of 3 Mg/mL of 5-chloro-2-mercaptobenzothiazole (CMBT) dissolved in acetonitrile, methanol, and water in the ratio 1:1:1 containing 0.1% formic acid and 0.01 M 18-crown-6-ether. For Gram-negative organisms, the CMBT is replaced by 14 Mg/mL a-cyano-4-hydroxycinnamic acid (aCHCA). The saturated CMBT and aCHCA solutions are freshly prepared prior to use; the acetonitrile, methanol, formic acid, and 18-crown-6-ether solvent can, however, be prepared and stored in a cool, dark glass bottle for up to 6 months. In order to calibrate the time of flight tube and correct for any variation in the flight length across the MALDI target plate, lock mass wells, positioned between the sample wells are inoculated with aCHCA matrix solution containing seven peptides of known mass (Fig. 7.1). Upon application of the appropriate MALDI matrix solution to each target well, the samples are left at room temperature for approximately 5 min to allow co-crystallization of the bacterial sample in the MADLI matrix before inserting the plate into the MALDI-TOF MS for automatic analysis.

Automatic acquisition of the mass spectral fingerprints is then achieved using the sample list, which details the well numbers of the bacterial samples, the corresponding data files, the wells containing replicate samples together with the experimental parameters used to collect the data and details of the database search. The experimental details are preselected by the operator after initially setting up the instrument to perform (i) a spatial calibration to automatically locate the well positions; (ii) optimization of the rennin substrate peak resolution, by adjustment of the pulse voltage to produce a sharp narrow peak at 1760 Da (i.e., <3 mass units

View Cwtaba« VrtrxJow

Microbcl ynx Or

Search results

S.J Genus

SubSpeow Retaovc P»ofeab*ty ftbsotut* Prgbab*y | PMSMatch Mcda [ Watrr.

Escherichia Escherichia Enter ctoacter GschertíHa aa.ia t>.& 4.79 1.60 I.158 0 76 0.72 0-24

401.26

142.08

115.91

6.42

5.15

-182-78

ClED CIED

CLED CLED CLED

Alpha-eyano Alphs-íyíftO

Alpha-c y ano Alpha-cyano Alpha-cyano Alpha-cyano Atpha-cy-ano Alcha-cyano

3S.00 16.00 OxyQA

37,00 16.00 0*ygm

  1. 00 16.00 Oxygei
  2. 00 16.00 Oiyyei
  3. 00 16.00 Oxygei
  4. 00 16.00 O'vofr

36 00 16 00 O'ygoi

35,00 16,00 Oxygei

Escherichia Escherichia Enter ctoacter GschertíHa

CLED CLED CLED

Figure 7.2. The browser details the wells searched (row A), the number of spectra collected for each replicate well (not shown), the fingerprint spectrum for each replicate well, the combined spectrum for all the replicate wells, and comparison of the test sample spectrum with up to 8 top spectral matches (only first match shown; second to eighth matches not presented). A list of the search results, together with information on database entry with respect to the probability of the match, the RMS value, basonyms or previous names (not shown), and the culture conditions used for the database spectra are also given.

Figure 7.2. The browser details the wells searched (row A), the number of spectra collected for each replicate well (not shown), the fingerprint spectrum for each replicate well, the combined spectrum for all the replicate wells, and comparison of the test sample spectrum with up to 8 top spectral matches (only first match shown; second to eighth matches not presented). A list of the search results, together with information on database entry with respect to the probability of the match, the RMS value, basonyms or previous names (not shown), and the culture conditions used for the database spectra are also given.

at half peak height); and (iii) calibration of the time of flight tube from the known masses of the 7 peptide calibrants. These parameters are then automatically used in the experimental file together with criteria for rejecting any spectra that are either too intense or too noisy. This ensures that only quality data from good spots on the target well are selected. Furthermore, the sample well is sampled from a minimum of 3 different sites to produce a maximum of 15 spectral profiles, with each profile produced from the sum of 10 individual shots to maximize the signal to noise ratio and further optimize reproducibility. The experimental file also has the potential to ramp through a series of laser energies in order to acquire the optimized spectra, should this be required. The quality-controlled reproducible spectra from the replicate bacterial sample wells are then automatically combined and searched against the chosen database and the results presented in a browser format (Fig. 7.2). The browser details the wells searched, the number of spectra collected for each replicate well, the spectrum for each individual well, the combined spectrum for all the replicate wells, and comparison of the test sample spectrum with up to 8 top spectral matches. A list of the search results, together with information on the database entry with respect to the probability of the match, the root mean square (RMS) value, basonyms, and the culture conditions used for the database spectra are also given. This information together with the comparison of the spectral profiles produces information as to the classification and identification of the bacterial sample. The mass spectral analysis requires approximately 1.5 h to acquire and analyze the data for a 96-well target plate. This means that in a normal working day (9 a.m. to 5 p.m.), the first MALDI target plate containing a maximum of 24 samples can be run approximately 1 h 15 min after culture, (10 min to inoculate the bacteria onto the plate + 1 h drying + 5 min for the addition of matrix and peptide solution and co-crystallization). The preparations of subsequent plates are then concurrent with analysis of the previous plate and result in the comfortable analysis of 5 MALDI target plates containing 120 samples during one working day. Because the sample preparation is simple and rapid and the MALDI-TOF MS analysis is automatic, minimal operator time is required for the instrument, leaving sufficient time for preparing further overnight cultures following the appropriate protocols. The high sample throughput, rapid analysis, and ease of acquiring the skills to prepare the target plates and run the instrument make this an attractive method for routine bacterial identification, where current microbiology staff can easily adapt to this new technique. Furthermore, because the cost of consumables is negligible due to the low concentrations of matrix and peptide solutions used, together with ability to reuse the MALDI target plates, the main cost factor is the instrument and database. Depending on the sample throughput, however, this cost can be offset against the high consumable costs currently associated with other identification techniques. This together with the speed of analysis now makes MALDI-TOF MS either an attractive complementary or an alternative to currently used identification techniques.

The system also allows for the production of an "in-house" database, which can be searched alone or together with the proprietary MMU database. Addition of spectral fingerprints to a database requires comparison of data for each replicate well using a root mean square (RMS) (Storms et al., 2004 ) function. Each replicate spectrum is compared in turn with the average of the other combined replicate spectra, and any spectrum found to be significantly different is automatically rejected. The total combined spectra for the acceptable replicate data is then added to the database, along with a statistical estimate of variance. The proprietary database generally uses 12 replicate wells to obtain a statistically representatative fingerprint spectrum for the database. Significant numbers of the spectral patterns in the proprietary MMU database are also checked for reproducibility using different operators and instruments prior to release.

Currently, the MMU database contains spectral fingerprints for the NCTC type strains, together with other representative strains of the same species and at present includes more than 4000 spectral fingerprints, covering more than 500 different species. The database is updated yearly with a minimum of 500 new spectral fingerprints (Fig. 7.3a). It is currently separated into 4 databases, which can be searched simultaneously, or separately; the core aerobic database, a database for Urinary Tract Infection (UTI) (Hofstadler et al., 2005) employing a more specialized media, a database of anaerobes, and latterly a database of clinical strains from well-recognized sources (Fig. 7.3b). Further details on the compilation of the MMU fingerprint databases are given by Keys et al. (2004).

  • a) □ No. genera
  • No. species
  • Total no. fingerprints
  • a) □ No. genera
  • No. species
  • Total no. fingerprints

2001

2002

2003

2004

2005

  • Core database
  • UTI database
  • Anaerobe database
  • Clinical database
  • Total spectra
  • Total organisms

2001

2002

2003

2004

2005

4500 4000 3500 3000 2500 2000 1500 1000 500 0

  • Core database
  • UTI database
  • Anaerobe database
  • Clinical database
  • Total spectra
  • Total organisms

4500 4000 3500 3000 2500 2000 1500 1000 500 0

2001

2002

2003

2004

2005

Figure 7.3. Yearly growth (a) and separation (b) of the proprietary MMU databases.

2001

2002

2003

2004

2005

Figure 7.3. Yearly growth (a) and separation (b) of the proprietary MMU databases.

Was this article helpful?

0 0

Post a comment