Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry Analysis of Gram-Positive, Catalase-Negative Cocci Not Belonging to the Streptococcus or Enterococcus Genus and Benefits of Database Extension

Abstract

Matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry with a Bruker Daltonics microflex LT system was applied to 90 well-characterized catalase-negative, Gram-positive cocci not belonging to the streptococci or enterococci. Biotyper version 2.0.43.1 software was used singly or in combination with a database extension generated in this study with 51 collection strains from 16 genera. Most strains were identified by using both databases individually, and some were identified only by applying the combined database. Thus, the methodology is very useful and the generated database extension was helpful.

TEXT

A new revolution in the identification of bacteria and fungi is ongoing with the introduction of mass spectrometry (MS) in the form of matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) MS (1, 2, 3, 6, 7). Identification depends on recognition of peak patterns characteristic of and mostly constant for different taxa (4, 5, 10). It is within the last decade and especially within the last 5 years that MALDI-TOF MS has been a major contribution to clinical microbiology (8, 10, 11). A series of developments, including the emergence of robust and easy-to-use software and hardware, drastic shortening of the time to identification of a positive culture, and even considerable lowering of costs, has put the methodology in a central position (2, 5). Three databases are commercially available at present: (i) the MALDI Biotyper database (for MS instruments from Bruker Daltonics, used in this study), (ii) Saramis (bioMérieux; for MS instruments from Shimadzu), and (iii) Andromas (compatible with either Bruker Daltonics or Shimadzu hardware) (5).

Catalase-negative, Gram-positive cocci that are not streptococci or enterococci represent a group of bacteria which, over the last decades, has become increasingly well characterized, and the number of taxonomic entities has steadily grown, thereby complicating the identification of these organisms (9). Strains most often recognized belong to the genera Gemella, Granulicatella, Abiotrophia, and Aerococcus. More seldom, strains from at least nine other genera may also be encountered in human samples (9). Many of the genera and species cause bloodstream infections and infective endocarditis. We applied MALDI-TOF MS to 51 collection strains (representing 16 genera and 51 species) and 90 well-characterized strains from the Danish national reference laboratory in order to evaluate usability and look for benefits of supplementing the existing database with new entries. The multiple genera and species investigated in this study have been only scarcely included in other studies.

Information on collection and challenge strains is given in Table 1 and Table S1 in the supplemental material. Of the 90 challenge strains, 78 strains were from positive blood cultures, 9 strains were from other types of specimens, and 3 strains had no data available. Preparation of strains for MALDI-TOF MS and identification (using MALDI Biotyper automation control 2.0.43.1 software) was performed as recommended by the software manufacturer. Briefly, a small amount of an overnight culture was added to a spot, followed by 70% formic acid to obtain extraction, before the addition of the matrix solution (a saturated solution of α-cyano-4-hydroxycinnamic acid [HCCA]). For each collection strain, a mass spectrum profile (MSP) based on 24 separate determinations (using MALDI Biotyper 2.0SR1 [build 223.8]) was created and stored in a separate library that could be combined with the standard database.

Table 1.

Numbers of strains of different genera examined in this study^a

Genus	No. of type strains (n = 51)	No. of species in Biotyper database (n = 29)	No. of challenge strains (n = 90)
Aerococcus	7	5	35
Gemella	7	3	23
Granulicatella	3	3	8
Abiotrophia	1	1	2
Lactococcus	4	3	5
Globicatella	2	0	5
Leuconostoc	4	3	5
Rothia	6	5	3
Facklamia	6	2	1
Vagococcus	1	1	1
Helcococcus	3	1	1
Alloiococcus	1	1	1
Pediococcus	3	2	0
Ignavigranum	1	0	0
Dolosicoccus	1	0	0
Dolosigranulum	1	0	0

^aType strains of different genera were obtained from the Culture Collection of the University of Gothenburg (CCUG), and the numbers of these species included in the Biotyper version 2.0.43.1 database are shown. Challenge strains are grouped according to their genus allocation based on combined results of phenotypic and partial 16S rRNA gene analyses.

When the collection strains were identified with the Biotyper version 2.0.43.1 software (Table 2), more than half of the collection strains obtained low score values (the score value is the logarithmic value of the weighted sum of matching peaks between the MSP for the unknown strain and MSPs existing in the database), the reason being that the taxa were not included in the database. The protocol given by the manufacturer for creating one's own MSPs in a separate library, though somewhat labor-intensive, makes it possible to extend the database for one's own wishes and needs. Not surprisingly, this extension made all the collection strains obtain score values of ≥2.000. The many relatively new species and the fast taxonomic developments among the Gram-positive, catalase-negative cocci emphasize the importance of being familiar with the content of the database used. Assessment of the obtained top 10 result matches is essential for confident interpretation, taking into consideration the mean score for the best taxon match, the number of identical matches, and the distance between the mean scores for the best and next best taxon matches. As shown in Table 3, results differ considerably among genera/species with respect to distances among taxa, intraspecies diversities, and certainty in the level of identification. All challenge strains were allocated to the presumed genera. For all 35 Aerococcus species strains, of which Aerococcus urinae strains dominated in number, best taxon matches were in agreement with the presumed species identifications. Use of the combined database resulted in score values exceeding 2.000 for all strains, except the two Aerococcus viridans strains. Among the 23 Gemella strains, Gemella morbillorum and Gemella haemolysans strains dominated in number. All G. morbillorum strains and two of six G. haemolysans strains were identified only with the use of the combined database; the median score value just exceeded 2.000 with the use of the combined database.

Table 2.

Allocation of 51 collection strains and 90 challenge strains to interpretation groups based on obtained score values^a

Identification database and group (no. of strains)	No. of strains with results of:
	2.300-3.000 (+++; green)	2.000-2.299 (++; green)	1.700-1.999 (+; yellow)	0.000-1.699 (−; red)
Biotyper (version 2.0) database
Collection strains (51)	6	18	7	20
Challenge strains (90):	15	35	18	22
Aerococcus species (35)	7	22	5	1
Gemella species (23)	0	5	5	13
Abiotrophia/Granulicatella species (10)	1	1	7	1
Mixed group of strains (22)b	7	8	0	7c
Biotyper database with database extensiond
Collection strains (51)	38	13	0	0
Challenge strains (90):	39	39	12	0
Aerococcus species (35)	15	18	2	0
Gemella species (23)	6	13	4	0
Abiotrophia/Granulicatella species (10)	7	2	1	0
Mixed group of strains (22)b	11	6	5e	0

^aThe meanings of score values obtained were as follows: 2.300 to 3.000, highly probable species identification (marked +++ and green in reports); 2.000 to 2.299, secure genus identification, probable species identification (marked ++ and green in reports); 1.700 to 1.999, probable genus identification (marked + and yellow in reports); and 0.000 to 1.699, no reliable identification (marked − and red in reports).

^bGenera: Globicatella (5 strains), Lactococcus (5 strains), Leuconostoc (5 strains), Rothia (3 strains), Facklamia (1 strain), Vagococcus (1 strain), Helcococcus (1 strain), and Alloiococcus (1 strain).

^cIdentified as Globicatella (5 strains), Alloiococcus (1 strain), and Rothia mucilaginosa (1 strain).

^dMSPs created in this study.

^eIdentified as Globicatella (2 strains), Lactococcus (1 strain), Leuconostoc (1 strain), and Rothia mucilaginosa (1 strain).

Table 3.

Identification data for 68 challenge strains^a

Database and species (no. of strains)	No. of strains with unreliable identification	Score value(s)		No. of identical matches	Difference(s) between score values for best and next best taxon matches
		Range	Median		Range	Median
Biotyper version 2.0 database
Aerococcus species (35)
A. urinae (27)		1.859–2.431	2.211	4	0.498–1.160	0.896
A. sanguinicola (5)		1.945–2.169	2.046	2	0.612–0.902	0.833
A. viridans (2)	1	1.954	1.954	6	0.671	0.671
A. christensenii (1)		1.835	1.835	1	0.476	0.476
Gemella species (23)
G. morbillorum (12)	12
G. haemolysans (6)	2	1.723–2.040	1.870	1	0.181–0.541	0.437
G. bergeri (4)		1.962–2.245	2.178	1	0.443–0.952	0.738
G. sanguinis (1)		2.003	2.003	1	0.464	0.464
Granulicatella species (8)
G. adiacens (7)	1	1.720–1.907	1.826	1 to 2	0.314–0.691	0.500
G. elegans (1)		1.800	1.800	1	0.527	0.527
Abiotrophia defectiva (2)		2.063–2.337	2.200	3	0.951–1.068	1.009
Biotyper database with database extension
Aerococcus species (35)
A. urinae (27)		2.068–2.489	2.249	5	0.697–1.182	0.965
A. sanguinicola (5)		2.352–2.556	2.462	3	1.066–1.213	1.213
A. viridans (2)		1.724–1.917	1.821	4	0.058–0.166	0.112
A. christensenii (1)		2.219	2.219	2	0.875	0.875
Gemella species (23)
G. morbillorum (12)		1.906–2.164	2.003	1 to 2	0.089–0.462	0.323
G. haemolysans (6)		2.191–2.431	2.208	1	0.102–0.315	0.160
G. bergeri (4)		2.174–2.395	2.260	1	0.596–0.947	0.851
G. sanguinis (1)		2.348	2.348	1	0.276	0.276
Granulicatella species (8)
G. adiacens (7)		1.971–2.412	2.353	2 to 3	0.603–1.105	0.883
G. elegans (1)		2.093	2.093	2	0.941	0.941
Abiotrophia defectiva (2)		2.400–2.244	2.322	4	1.070–1.244	1.157

^aIdentification data for 68 challenge strains belonging to the genera Aerococcus, Gemella, Granulicatella, and Abiotrophia were obtained by MALDI-TOF MS by using the MALDI Biotyper version 2.0.43.1 software database and also by extending this database with new mass spectrum profiles. The automatic evaluation of the obtained identification in addition to the score, the number of identical hits, and the distance to the next best taxon match were taken into consideration.

Strains from the following genera were convincingly identified using both databases, although the number of examined strains was limited: Lactococcus, Leuconostoc, Rothia, Facklamia, Helcococcus, and Vagococcus. The genera Globicatella and Alloiococcus were not included in the Biotyper version 2.0.43.1 database, but strains were convincingly identified by using the combined database.

In no cases did misidentification occur at the genus level. Three strains had divergent identifications at the species level. When strains belonging to genera or species not included in the Biotyper database were examined, identifications were given as not reliable. Differences in score values between best taxon matches for strains from the same species varied among species and could be close to 1.000 for some species, illustrating the species diversity and probably also examination conditions for created MSPs; thus, for the 27 strains of A. urinae a median distance of 0.575 and a range of 0.375 to 0.998 were seen (data not presented). Next best taxon matches for 27 A. urinae strains showed great variation in suggested taxa, with the four most often obtained species being Arthrobacter ardleyensis (7 strains), Staphylococcus vitulinus (4 strains), Aerococcus christensenii (3 strains), and Staphylococcus aureus (2 strains). The number of identical matches depends on the number of strains of the taxon included in the database, and when an identical match is present, it makes the identification more confident (Table 3). The score distance to the next best taxon match is an important parameter (Table 3). The difference illustrates the closeness of the relationship between different genera and species, which, as well known, varies considerably; when present, a major distance in mean score supports a confident identification. The size of the difference (value or percentage) varies according to the taxon examined. A minimum difference has not been established, but this has to have a magnitude that also takes into consideration the uncertainty of the MS system and the best taxon match mean score. Often, top 10 result matches included closely related taxa, although results including more distantly related genera/species also occurred. To establish the intraspecies diversity among different taxa and validate the MS method, collections of strains have to be examined.

Dendrograms are useful for demonstrating the ability of MALDI-TOF MS to differentiate among the included species and to visualize intraspecies similarity or variation when more strains are examined. Dendrograms for the species most often recognized among the challenge strains were created using the Biotyper MSP dendrogram creation standard method. The MSP dendrograms for Aerococcus strains (Fig. 1a) and the genera Gemella, Granulicatella, and Abiotrophia (Fig. 1b) convincingly delineated the included species.

Fig 1.

MSP dendrograms for included challenge strains of Aerococcus species (a) and for challenge strains of the genera Gemella, Granulicatella, and Abiotrophia (b). The consensus mass spectra generated in this study for type strains of identified species are also included.

In conclusion, MALDI-TOF MS seems very promising for the identification of strains of catalase-negative, Gram-positive cocci not belonging to the streptococci or the enterococci. Such strains are frequently hard to identify both at the genus and at the species levels. Some of the genera include very closely related species, as illustrated with the Gemella genus, which is also challenging for this methodology. For routine use, MALDI-TOF MS has been shown to be robust. However, standardization of examination procedures for the identification of closely related species and for taxonomic purposes and classification is necessary. In the analysis of Biotyper reports, the identification may be strengthened by first considering the automatic evaluation of the obtained identification based on the score value but then also evaluating the number of identical hits and the distance to the next best taxon match. The possibility of creating one's own consensus mass spectra and adding them, in separate libraries, opens up the potential for using this methodology in many specialized settings.