Effects of Solid-Medium Type on Routine Identification of Bacterial Isolates by Use of Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry

Abstract

Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) is a rapid method for the identification of bacteria. Factors that may alter protein profiles, including growth conditions and presence of exogenous substances, could hinder identification. Bacterial isolates identified by conventional methods were grown on various media and identified using the MALDI Biotyper (Bruker Daltonics, Billerica, MA) using a direct smear method and an acid extraction method. Specimens included 23 Pseudomonas isolates grown on blood agar, Pseudocel (CET), and MacConkey agar (MAC); 20 Staphylococcus isolates grown on blood agar, colistin-nalidixic acid agar (CNA), and mannitol salt agar (MSA); and 25 enteric isolates grown on blood agar, xylose lysine deoxycholate agar (XLD), Hektoen enteric agar (HE), salmonella-shigella agar (SS), and MAC. For Pseudomonas spp., the identification rate to genus using the direct method was 83% from blood, 78% from MAC, and 94% from CET. For Staphylococcus isolates, the identification rate to genus using the direct method was 95% from blood, 75% from CNA, and 95% from MSA. For enteric isolates, the identification rate to genus using the direct method was 100% from blood, 100% from MAC, 100% from XLD, 92% from HE, and 87% from SS. Extraction enhanced identification rates. The direct method of MALDI-TOF analysis of bacteria from selective and differential media yields identifications of varied confidence. Notably, Staphylococci spp. from CNA exhibit low identification rates. Extraction enhances identification rates and is recommended for colonies from this medium.

INTRODUCTION

Prompt and accurate identification of bacterial isolates is an objective of the clinical microbiology laboratory and is paramount for patient care. Conventionally this has been achieved using macro- and microscopic observation of morphology and biochemical analysis. These methods often require isolation of individual colonies from polymicrobial cultures and subculture prior to isolate identification. These steps can add significant time to identification of isolates. Molecular methods provide reliable results, though many are expensive to perform, time-consuming, and technically demanding (6). Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has recently emerged as a rapid, accurate, and cost-effective method of identification, applicable to a wide range of bacterial isolates (9, 12, 13, 16, 17, 19).

A major benefit of MALDI-TOF technology is the ability to obtain identifications using a single colony, eliminating the need for subculture and further incubation. To identify most isolates, a single colony can be picked from solid culture media using a swab or toothpick and smeared directly onto a polished steel target plate for identification (direct method) (1, 9, 14, 17, 18). Alternatively, isolates with formidable cell walls or those that produce excess exopolysaccharide matrix (i.e., mucoid pseudomonads) may require processing through a short (approximately 5-min) formic acid-acetonitrile extraction and centrifugation step prior to application on the target (extracted method) (1, 7, 9, 13, 15, 16, 18). Though both preparation methods are rapid, the ability to directly smear colonies to test plates is particularly attractive to a laboratory with high throughput and limited resources. The extracted method, however, helps to release proteins of interest, allowing more-efficient ionization and thus higher-quality identifications (1, 9).

Limited study has been conducted on the effect of media on MALDI-TOF identification using both direct and extraction methods. Culture media contain a variety of differential and selective components, including antibiotics, salts, and pH indicators. Some components, such as salt, are well known inhibitors of mass spectrometry, and different media can induce changes in bacterial protein expression (1, 9, 10). To address these concerns, we examined the effect of commonly used culture media on the rate and confidence of bacterial identification using MALDI-TOF.

MATERIALS AND METHODS

Collection of isolates.

A collection of 68 bacterial isolates was obtained from Dynacare Laboratories (Milwaukee, WI). These isolates were collected from various patient populations, including both outpatient clinic and hospitalized patients throughout the Milwaukee area. Isolates included 23 Pseudomonas spp., 20 Staphylococcus spp., and 25 Enterobacteriaceae isolates. Pure cultures of each organism were cultured on Columbia agar with 5% sheep blood (blood) (Remel, Lenexa, KS) at 5% carbon dioxide concentration and at 35°C. Isolates were subsequently identified using a combination of manual (Gram stain, catalase, coagulase, oxidase, etc.) and automated (Phoenix [BD Diagnostics, Sparks, MD], Vitek 2 [bioMérieux, Marcy-l'Étoile, France], Rapid NH [Remel], and API 20E [bioMérieux]) biochemical methods according to the manufacturer's instructions. Prior to testing, Pseudomonas isolates were then subcultured to Pseudocel (CET; BD) and MacConkey agar (MAC; Remel). Staphylococcus isolates were subcultured to colistin-nalidixic acid agar (CNA; Remel) and mannitol salt agar (MSA; Remel). Enteric isolates were subcultured to xylose lysine deoxycholate agar (XLD; Remel), Hektoen enteric agar (HE; Remel), salmonella-shigella agar (SS; Remel), and MAC. All isolates grown on selective media were incubated for 24 h at 35°C with 5% carbon dioxide. MALDI-TOF identification was performed on each isolate from each type of medium in addition to blood, using both direct and extracted methods.

Smear method.

Using a cotton swab, one to five colonies of each isolate were picked. The cotton swab was then smeared over an individual spot on the MALDI-TOF target plate, creating a confluent layer of bacteria. The bacteria were dried for 5 min, after which 1 μl of matrix solution (α-cyano-4-hydroxycinnamic acid solution; HCCA), was applied to the spot and allowed to dry. This method was performed in duplicate for each isolate.

Extraction method.

Using a cotton swab, one to five colonies were picked from solid media and inoculated into a tube containing 300 μl of deionized water. To this mixture, 600 μl of 100% ethanol was added (final concentration, 75% ethanol). Each tube was then centrifuged at 16,000 × g for 2 min. Following centrifugation, the supernatant was discarded and the resultant cell pellet was allowed to dry. Once dry, the cell pellets were resuspended in 50 μl of 70% formic acid and 50 μl acetonitrile. The suspension was again centrifuged at 16,000 × g (14,000 rpm) for 2 min. Following centrifugation, 2 μl of the resultant supernatant containing the extracted proteins was applied to each well and dried for 5 m. Matrix solution (1 μl) was then applied to each well and allowed to dry. This method was performed in duplicate for each isolate.

MALDI-TOF analysis.

MALDI-TOF analysis was performed using the MALDI Microflex LT instrument and Biotyper 3.0 software (Bruker). Each 96-well target plate was tested using the automated analysis feature of the Biotyper software. A compilation of no less than 240 individual measurements was obtained and used to compile a consensus profile. This profile was then compared to reference spectra in the Biotyper database, yielding identification and a degree of confidence expressed as a score from 0.00 to 3.00. Confidence scores of less than 1.70 corresponded to “unreliable identification,” scores of 1.70 to 1.99 corresponded to “genus-only” identification, and scores of 2.00 and higher corresponded to “genus and species” identification (7, 14, 16). Isolates which failed to generate 240 useable measurements or confidence score values below 2.00 were analyzed using a manual analysis method in an effort to obtain better spectra. This entailed generating a compilation of 240 individual measurements utilizing user-targeted laser shots rather than those randomly generated by the automated analysis software.

RESULTS

The direct method and smear method were performed from each combination of isolate/medium. The level of identification was determined by the 0.00-to-3.00 score provided by the Biotyper database. Since isolates were analyzed in duplicate, only the isolate with the highest confidence score was reported. Results for each strain of Pseudomonas spp. are shown in Table 1. Analysis using the direct method from blood agar resulted in identification of 83% of isolates to the genus level and 65% of isolates to the species level. Using the direct method from MAC, 78% of isolates were identified to the genus level and 52% to the species level, while the direct method from CET yielded 94% identification to the genus level and 47% to the species level. Extraction slightly increased identification of Pseudomonas spp. to genus (87%) but significantly increased identification to the species level (70%) for colonies cultured on MAC. Extraction of colonies cultured on CET also increased identification rates, resulting in 100% identification to the genus level and 88% to the species level. Extraction from blood agar yielded 83% identification to the genus level and 61% identification to the species level, results similar to those obtained with the direct method. Manual analysis (performed when the confidence score was <2.00) was required for 48% of isolates cultured on MAC when the direct method was used, which reduced to 35% when extraction was performed. Using the direct method, 53% of isolates cultured on CET required further manual analysis, which reduced to 12% when extraction was performed. Of isolates cultured on blood, 35% required manual with the direct method, which reduced to 39% when extracted.

Table 1.

Pseudomonas isolates

Pseudomonas species	Score for indicated medium witha:
	Direct method			Extraction method
	MAC	CET	Blood	MAC	CET	Blood
Pseudomonas aeruginosa	2.19	1.93	2.15	2.34	2.36	2.34
Pseudomonas aeruginosa	1.94	NI	NI	2.12	2.32	2.31
Pseudomonas aeruginosa	2.38	2.29	2.31	2.40	2.36	2.41
Pseudomonas aeruginosa	NI	1.98	2.31	NI	2.31	2.21
Pseudomonas aeruginosa	2.43	2.18	2.36	2.46	2.41	2.20
Pseudomonas aeruginosa	2.31	2.19	2.22	2.40	2.46	2.18
Pseudomonas aeruginosa	1.99	2.25	2.25	2.55	2.42	2.40
Pseudomonas aeruginosa	2.25	1.97	2.34	2.41	2.48	2.39
Pseudomonas aeruginosa	2.25	2.06	2.29	2.35	2.41	2.39
Pseudomonas aeruginosa	2.36	2.24	2.44	2.49	2.54	2.33
Pseudomonas fluorescens group	1.78	1.92	1.84	1.80	1.96	1.75
Pseudomonas fluorescensb	NI	1.71	NI	1.82	1.73	NI
Pseudomonas oryzihabitans	NI	NG	1.85	2.00	NG	1.87
Pseudomonas oryzihabitans	2.16	NG	2.08	2.11	NG	2.10
Pseudomonas oryzihabitans	1.75	NG	1.73	1.86	NG	1.74
Pseudomonas oryzihabitans	1.85	NG	1.87	2.07	NG	NI
Pseudomonas putida	2.08	1.82	2.27	2.47	2.35	2.42
Pseudomonas putida	2.31	1.85	2.12	2.40	2.34	2.38
Pseudomonas putida	1.99	2.02	2.14	1.96	2.01	1.79
Pseudomonas putidac	2.07	2.13	2.19	2.02	2.21	1.86
Pseudomonas putidad	2.21	1.95	2.07	2.28	2.21	2.17
Pseudomonas species	NI	NG	NI	NI	NG	NI
Pseudomonas stutzeri	NI	NG	NI	NI	NG	NI
Average (SD)e	2.13 (0.21)	2.03 (0.17)	2.15 (0.20)	2.22 (0.24)	2.29 (0.21)	2.17 (0.25)
% Isolates identified to:
Genus level	78%	94%	83%	87%	100%	83%
Species level	52%	47%	65%	70%	88%	61%

^aMALDI-TOF top identification matched conventional identification with only the discrepancies listed in footnotes b to d. NI, results not interpretable; NG, no growth.

^bMALDI-TOF top identification was Pseudomonas species on MAC (extracted) and CET (extracted).

^cMALDI-TOF top identification was Pseudomonas monteilii on MAC (direct), CET (extracted), MAC (extracted), CET (extracted), and blood (direct).

^dMALDI-TOF top identification was always Pseudomonas plecoglossida.

^eValues obtained from isolates with a >1.70 confidence value.

Results for each tested strain of Staphylococcus spp. are shown in Table 2. When performing the direct method from blood agar, 95% of isolates were identified to genus and 65% were identified to the species level. Extraction from blood agar increased identification rates to 100% genus and 90% species identification. Performing the direct method from CNA yielded 75% identification to the genus level and 55% identification to the species level. Extraction markedly improved identification from CNA to 100% genus identification and 80% species identification. The direct method performed on isolates cultured on from MSA yielded 95% identification to genus and 75% identification to the species level. Extraction from MSA improved results to 100% genus identification and 80% species identification. Of the isolates analyzed using the direct method from CNA, 45% required further manual analysis, which reduced to 20% when extracted. Direct analysis from MSA required further manual analysis for 25% of isolates, which reduced to 20% if extracted. Of the colonies cultured on blood agar, 35% required manual analysis when the direct method was used, whereas 10% required manual analysis if extracted.

Table 2.

Staphylococcus isolates

Staphylococcus species	Score for indicated medium witha:
	Direct method			Extraction method
	CNA	MSA	Blood	CNA	MSA	Blood
Staphylococcus aureus	2.04	2.25	2.14	2.36	2.16	2.20
Staphylococcus aureus	2.01	2.28	2.26	2.33	2.38	2.42
Staphylococcus aureus	2.23	2.22	2.21	2.42	2.26	2.22
Staphylococcus aureus	2.25	2.31	2.02	2.41	2.43	2.33
Staphylococcus aureus	2.20	2.24	2.28	2.44	2.26	2.29
Staphylococcus aureus	2.17	2.21	1.77	2.32	2.26	2.28
Staphylococcus aureus	1.76	2.14	2.05	2.47	2.42	2.33
Staphylococcus aureus	2.19	2.18	2.06	2.38	2.26	2.37
Staphylococcus aureus	2.06	2.27	2.26	2.44	2.37	2.35
Staphylococcus epidermidis	2.06	2.22	1.97	2.25	2.03	2.15
Staphylococcus epidermidis	1.87	2.20	1.90	2.21	2.08	1.98
Staphylococcus epidermidis	2.12	2.14	2.08	2.33	1.97	1.96
Staphylococcus epidermidis	2.11	2.07	2.15	2.25	2.04	2.15
Staphylococcus intermediusb	NI	NI	NI	2.07	1.79	2.13
Staphylococcus lugdunensis	1.99	2.15	1.83	2.16	2.14	2.25
Staphylococcus lugdunensisc	1.71	2.16	1.93	2.21	2.09	2.16
Staphylococcus saprophyticus	NI	1.99	2.19	1.98	1.97	1.95
Staphylococcus saprophyticus	NI	1.78	2.17	1.77	1.99	2.19
Staphylococcus saprophyticus	NI	1.95	1.86	1.72	2.17	2.16
Staphylococcus warnerid	NI	1.90	2.17	1.95	2.01	2.15
Average (SD)e	2.05 (0.16)	2.14 (0.14)	2.07 (0.16)	2.22 (0.22)	2.15 (0.17)	2.20 (0.13)
% Isolates identified to:
Genus level	75%	95%	95%	100%	100%	100%
Species level	55%	75%	65%	80%	80%	90%

^aMALDI-TOF top identification matched conventional with only the discrepancies listed in footnotes b to d. NI, results not interpretable; NG, no growth.

^bMALDI-TOF top identification was Staphylococcus pseudointermedius on MSA (extracted).

^cMALDI-TOF top identification was always Staphylococcus epidermidis.

^dMALDI-TOF top identification was always Staphylococcus saprophyticus.

^eValues obtained from isolates with a >1.70 confidence value.

Results for each tested enteric isolate are shown in Table 3. When analyzed using the direct method from blood agar, 100% of the isolates identified to the species level. Similar results were obtained when these organisms were extracted from blood agar, with 100% genus identification and 88% species identification. Direct analysis of isolates cultured on MAC resulted in 100% identification to the genus level and 88% to the species level. These results were similar following extraction with 100% genus identification and 96% species identification. All isolates cultured on XLD and analyzed with the direct method were identified to the genus level, and 84% were identified to the species level. Extraction of isolates from XLD yielded 92% identification to the genus level and 88% identification to the species level, rates similar to those obtained with the direct method. Direct-method analysis of isolates cultured on HE resulted in 92% identification to the genus level and 68% to the species level. Similar results were obtained for extraction from HE, with 84% genus and 72% species identification. Direct-method analysis of isolates cultured on SS medium resulted in 87% identification to the genus level and 74% identification to the species level. These values improved when the extraction method was used, with 96% identification to the genus level and 87% identification to the species level. These findings are summarized in Table 3. No manual analysis had to be performed on isolates smeared from blood agar, though it was required for 12% of those extracted from blood agar. Isolates cultured on MAC required manual spotting at a rate of 12% when the direct method was used and at a rate of 4% with the extracted method. Isolates cultured on XLD required manual analysis at a rate of 16% of the time when the direct method was used, but at a rate of only 12% when extracted. Isolates cultured on HE agar required manual analysis at a rate of 32% when the direct method was used, which reduced to a rate of 28% when extracted. Those isolates cultured on SS medium required manual analysis at a rate of 26% when the direct method was used and at a rate of 13% when extracted.

Table 3.

Enteric isolates

Species	Score for indicated medium witha:
	Direct method					Extraction method
	MAC	XLD	HE	SS	Blood	MAC	XLD	HE	SS	Blood
Aeromonas caviae	2.06	1.86	1.70	2.11	2.26	2.14	1.88	1.90	2.25	2.16
Aeromonas hydrophilab	2.22	2.31	2.33	2.34	2.21	2.22	2.22	2.23	2.22	2.25
Alcaligenes faecalis	2.28	2.01	2.18	NG	2.35	2.29	2.00	1.96	NG	2.41
Citrobacter freundii	1.83	2.18	1.74	1.88	2.40	2.25	NI	1.89	2.05	2.34
Citrobacter koseri	2.11	2.29	2.07	1.94	2.15	2.34	2.27	NI	NI	2.43
Enterobacter aerogenes	2.36	2.38	2.20	2.38	2.46	2.48	2.33	2.35	2.40	2.49
Enterobacter cloacae	2.43	2.14	1.99	2.25	2.92	2.29	2.25	2.37	2.36	2.00
Escherichia coli	2.33	2.38	2.05	NI	2.42	2.42	2.33	2.30	2.02	2.41
Escherichia coli	2.08	2.04	2.05	NI	2.37	2.39	NI	NI	1.91	2.47
Klebsiella oxytoca	2.25	2.30	2.34	2.06	2.07	2.38	2.38	2.36	2.30	2.43
Klebsiella oxytocac	1.77	1.89	NI	1.86	2.24	2.03	2.27	NI	2.05	1.97
Klebsiella pneumoniae	2.13	2.19	2.16	NG	2.36	2.37	2.28	NI	NG	2.51
Klebsiella pneumoniaed	1.82	1.88	NI	NI	2.26	2.27	2.17	2.15	2.12	2.16
Morganella morganii	2.44	2.17	1.90	2.23	2.45	2.52	2.32	2.30	2.41	2.59
Proteus mirabilis	2.11	2.14	2.12	2.19	2.40	2.39	2.38	2.39	2.36	2.32
Providencia rettgeri	2.00	2.41	2.17	2.29	2.41	2.50	2.58	2.60	2.34	2.43
Providencia rettgerie	2.16	1.89	1.89	2.14	2.08	1.99	2.01	2.03	1.98	1.98
Providencia stuartii	2.21	2.31	2.12	2.11	2.11	2.20	2.62	2.24	2.18	2.26
Salmonella species	2.55	2.44	2.35	2.37	2.40	2.38	2.28	2.24	2.44	2.33
Salmonella species	2.40	2.31	2.36	2.41	2.16	2.38	2.30	2.17	2.37	2.39
Salmonella species	2.21	2.27	2.13	2.13	2.28	2.44	2.35	2.51	2.42	2.51
Salmonella species	2.28	2.31	2.21	2.19	2.19	2.43	2.41	2.43	2.41	2.40
Salmonella species	2.29	2.35	1.93	2.08	2.37	2.43	2.47	2.46	2.44	2.44
Serratia marcescens	2.32	2.25	2.27	2.37	2.32	2.13	2.25	2.21	2.21	2.28
Shigella sonneif	2.37	2.36	2.27	2.37	2.35	2.19	2.26	2.34	2.34	2.24
Average (SD)g	2.20 (0.20)	2.20 (0.18)	2.11 (0.19)	2.19 (0.17)	2.32 (0.17)	2.31 (0.14)	2.29 (0.17)	2.26 (0.19)	2.25 (0.17)	2.33 (0.17)
% Isolates identified to:
Genus level	100%	100%	92%	87%	100%	100%	92%	84%	96%	100%
Species level	88%	84%	68%	74%	100%	96%	88%	72%	87%	88%

^aMALDI-TOF top identification matched conventional with only the discrepancies listed in footnotes b to f. NI, results not interpretable; NG, no growth.

^bMALDI-TOF top identification was Aeromonas caviae on MAC (extracted).

^cMALDI-TOF top identification was Raoultella ornitholytica on MAC (extracted), SS (extracted), and blood (extracted).

^dMALDI-TOF top identification was Klebsiella variicola on XLD (extracted) and blood (extracted).

^eMALDI-TOF top identification was Pseudomonas vermicolor on MAC (direct), XLD (direct), HE (direct), SS (direct), HE (extracted), and blood (extracted).

^fMALDI-TOF top identification was always Escherichia coli.

^gValues obtained from isolates with a >1.70 confidence value.

DISCUSSION

The results of our study suggest that media can affect MALDI-TOF confidence scores when testing the same bacterial isolate. This effect was most apparent when the direct method was used on Pseudomonas species cultured on MAC, Pseudomonas species cultured on CET, and Staphylococcus species cultured on CNA. These isolates yielded identifications to the species level 52%, 47%, and 55% of the time, respectively, and required significantly more manual analysis. These effects appeared to resolve after extraction, which increased the number of Pseudomonas (69.6% from MAC and 88% from CET) and Staphylococcus (80% from CNA) isolates identified to the species level. Interestingly, while culture on MAC appeared to negatively affect the identification of Pseudomonas isolates when the direct method was used, enterics cultured on MAC appeared unaffected, identifying to the species level 88% of the time using the direct method. This may be due to changes in the cell wall and mucoid phenotype observed with Pseudomonas isolates grown on MAC. Alternatively it may be due to the better overall identification rates seen with enteric isolates in general. Importantly, while there were occasional discrepancies between conventional and MALDI-TOF identification, no specific medium was ever consistently associated with such discrepancies.

The exact reasons for the differences in identification rates between media is unclear; however, similar findings have been reported. Walker et al. (18) demonstrated that Staphylococcus aureus produced different expression profiles when cultured on different media. It was observed that colonies from Columbia blood agar produced more spectral peaks than colonies obtained on mannitol salt agar, with extra peaks being attributable to blood components. While the extra peaks did not affect organism identification in this study, the findings suggest that solid-medium components can affect protein spectral profiles. A study by Buskirk et al. (4) found that fungal pigments suppress the desorption/ionization process. It could be reasoned that pigmented media such as MacConkey agar could have a similar effect. Another factor shown to affect mass spectrometry, including MALDI-TOF analysis, is ion suppression (2, 10, 11). Ion suppression is caused by the presence of ions within the test matrix, which will decrease the efficiency at which the analyte is ionized by blocking the ionization process. This effect is typically observed when testing from matrices with high salt content (2). In our study, results from MSA were similar to those from blood agar for Staphylococcus isolates, implying that testing colonies directly from MSA is not hindered by ion suppression. However, ion suppression may account for lower confidence scores obtained from Pseudomonas colonies grown on MAC compared to those grown on blood agar, given that MAC contains bile salts as a primary ingredient. Data from enteric isolates show slightly lower species identification rates from isolates grown on HE and SS media, which contain higher levels of bile salt than MAC and XLD. Though these findings may be the result of ion suppression, additional studies are required.

A significant weakness in this study lies in the small sample size tested in each category. The purpose of this study was to identify potential relationships between identification rates and a large variety of solid media. While sample sizes may be sufficient to identify such relationships, larger sample sizes and more-specific studies should be designed to confirm and characterize such findings. Another potential weakness is the lack of comparison to a gold standard for discrepant results. Isolates, such as one strain of Pseudomonas spp., were classified as such by conventional means but failed to be identified using MALDI-TOF. This could reflect a lack of the organism in the Biotyper database, an error in conventional identification, or a mixed culture.

The efficacy and utility of MALDI-TOF analysis for the identification of clinical isolates have been thoroughly demonstrated in recent literature (3, 5–9, 13–19). Some studies suggest that MALDI-TOF analysis, in addition to being a rapid method of initial bacterial identification, could serve as a cost-effective substitute to sequencing as a reference method (3).

In our study, direct-method analysis of isolates cultured on most of the tested specialized media gave similar identification rates as those smeared from blood. For those media that gave inferior results, notably MAC and CET for Pseudomonas spp. and CNA for Staphylococcus spp., extraction enhanced identification rates and decreased the amount of time needed for manual analysis. An important point, though, is that while identification rates were affected, these media were not associated with consistent incorrect identification. Therefore, performing the direct method is an acceptable practice, though with the trade-off of lower identification rates. The data of this study demonstrate that such effects resolve following extraction. Therefore, we would recommend that extraction be performed on any isolate which fails to identify from these media.