Comparison of API 50 CH Strips to Whole-Chromosomal DNA Probes for Identification of Lactobacillus Species

Abstract

The API 50 CH identification system was evaluated for the identification of 97 strains of commensal lactobacilli. This system agreed with the species-level identifications for none of the 7 reference strains and only 4 of 90 vaginal isolates identified using whole-chromosomal DNA probes.

Lactobacillus species maintain a healthy vaginal ecosystem and are associated with decreased acquisition of bacterial vaginosis (10), human immunodeficiency virus (13), and gonorrhea (13). There are over 50 species recognized in the genus Lactobacillus (7), and several of these species have been proposed as probiotics (4, 14). In some instances, Lactobacillus strains found in food or probiotics have caused peritonitis (12) or endocarditis (9). Correct species identification of lactobacilli is essential to guide appropriate antibiotic therapy (12). Correct identification of lactobacilli is also relevant to studies of epidemiology (2, 3) and innate immunity (1).

Phenotypic tests have been used as the primary methods for classifying Lactobacillus species. Klein et al. have reported that none of the commercial kits can be recommended and that some species of lactobacilli are incorrectly identified as Lactobacillus acidophilus (12). Using phenotypic techniques, L. acidophilus has been reported to be the most frequently isolated Lactobacillus species recovered from the vagina (1, 6, 16). However, using genotypic methods, studies done in the United States (2), Japan (18), and Europe (8, 19) have reported that L. crispatus, L. jensenii, L. iners, L. gasseri, and L. vaginalis are the most common species of lactobacilli found in the vagina. This suggests that the phenotypic methods most commonly used for the identification of Lactobacillus species have poor concordance with genomics-based tests.

Some investigators have relied on API 50 CH carbohydrate fermentation strips (bioMérieux, Inc., Marcy l'Etoile, France) to identify vaginal strains of Lactobacillus to the species level (1, 15). In this study, we evaluate the use of API 50 CH carbohydrate fermentation strips for the identification of five common commensal Lactobacillus species isolates.

Lactobacillus isolates (n = 97) stored at −80°C in litmus milk (Becton Dickinson Microbiology Systems, Cockeysville, MD) were tested for species identification using the API 50 CH carbohydrate fermentation strips. Seven of these were reference strains: L. crispatus ATCC 33197 and ATCC 202225 (also known as CTV-05), L. jensenii ATCC 25258, L. gasseri ATCC 9857 and ATCC 4963, L. iners CCUG 28746, and L. vaginalis ATCC 49540. The remaining 90 isolates were recovered from vaginal specimens and identified to the species level using whole-chromosomal DNA probes (3). These included isolates of L. crispatus (n = 16), L. jensenii (n = 18), L. gasseri (n = 18), L. iners (n = 19), and L. vaginalis (n = 19). These isolates were obtained from women in Seattle (n = 26), Pittsburgh (n = 28), and Uganda (n = 36) from 1995 to 2002.

Each isolate was inoculated onto a Columbia agar with 5% sheep blood (BA) (PML Microbiologicals, Portland, OR) plate, incubated at 37°C in 6.1% CO₂ for 48 h, and checked for purity before being tested. The package insert for the API 50 CH test recommends subculturing the strains onto MRS agar (bioMérieux, Inc.); however, MRS agar does not support the growth of some vaginal Lactobacillus species, such as L. iners (7, 17). Therefore, both MRS (Oxoid LTD., Basingstoke, Hampshire, England) and BA were used to evaluate two ATCC L. crispatus strains. Both media yielded the same API patterns (data not shown). Therefore, all the lactobacilli tested were modified with growth on BA, according to the API 50 manual (bioMérieux, Inc.). All of the isolates were evaluated using the same kit reference and lot numbers in order to limit test variability. The results at 48 h were transmitted to a bio-Mérieux representative, who entered each biochemical profile into the identification software database (APILab Plus; bioMérieux, Inc.).

The API 50 CH profiles for the 97 commensal strains of lactobacilli demonstrated phenotypic diversity. None of the 18 L. crispatus isolates shared the same phenotypic pattern, and L. gasseri yielded 18 different patterns for the 20 isolates tested. The 20 L. vaginalis isolates yielded 15 different reaction patterns. In addition, the 19 L. jensenii isolates produced 11 different patterns. However, all of the L. iners isolates and three of the L. vaginalis isolates were nonreactive for all of the tests in the API 50 CH system. The profiles for each species are summarized in Table 1.

TABLE 1.

Percentages of positive tests after 48 hours

Lactobacillus species (no. of isolates)	% of API 50 CH tests that were positive fora:
	GAL	GLU	FRU	MNE	MAN	NAG	AMY	ARB	ESC	SAL	CEL	MAL	LAC	MEL	SAC	TRE	RAF	AMD	GLYG	GEN	TUR	TAG
L. crispatus (18)	72	100	94	56	44	100	33	33	89	78	72	100	44	6	100	67	39	61	44	17	0	0
L. jensenii (19)	0	100	89	95	0	5	74	84	100	95	95	95	0	21	100	79	21	16	0	47	0	0
L. iners (20)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
L. gasseri (20)	70	95	95	80	0	80	55	45	100	70	85	80	35	5	85	80	5	30	0	75	20	50
L. vaginalis (20)	30	50	25	0	0	5	0	0	0	0	0	80	35	50	50	5	55	5	5	0	0	0

^aOnly the tests with a positive result were included. GAL, d-galactose; GLU, d-glucose; FRU, d-fructose; MNE, d-mannose; MAN, d-mannitol; NAG, N-acetylglucosamine; AMY, amygdalin; ARB, arbutin; ESC, esculin ferric citrate; SAL, salicin; CEL, d-cellobiose; MAL, d-maltose; LAC, d-lactose (bovine origin); SAC, d-saccharose (sucrose); TRE, d-trehalose; RAF, d-raffinose; AMD, amidon (starch); GLYG, glycogen; GEN, gentiobiose; TUR, d-turanose; TAG, d-tagatose.

Currently, the API identification profile for L. crispatus suggests that all isolates of this species will be positive by the d-fructose, esculin ferric citrate, salicin, d-cellobiose, and amidon (starch) tests (bioMérieux, Inc.). However, in Table 1, the L. crispatus profile reveals that all the isolates were positive by the d-glucose, N-acetylglucosamine, d-maltose, and d-saccharose (sucrose) tests.

The heterogeneity of biochemical profiles and the lack of database profiles for L. gasseri, L. jensenii, L. iners, and L. vaginalis may account for the fact that only 4 (4%) of the 97 vaginal Lactobacillus isolates were identified to the species level, in agreement with the results of genomic methods (Table 2). The API 50 CH system misidentified 33 of the 97 isolates as either L. acidophilus or L. fermentum. Of the L. jensenii and L. gasseri isolates, 59% were identified as L. acidophilus through the API system. The API 50 CH database also identified 7 out of 20 (35%) L. vaginalis isolates as L. fermentum. Over half of the 97 isolates yielded an uninterpretable or doubtful API profile.

TABLE 2.

Lactobacillus species identified through the API identification software database

API identification	No. (%) of isolates of indicated species (total no.) whose identification was based on DNA homology with a type strain
	L. crispatus (18)	L. jensenii (19)	L. gasseri (20)	L. vaginalis (20)	L. iners (20)
L. crispatus	4	0	0	0	0
L. acidophilus	3	12	11	0	0
Lactococcus lactis subsp. cremoris	1	0	0	0	0
Lactococcus raffinolactis	1	0	0	0	0
L. fermentum	0	0	0	7	0
L. delbrueckii subsp. delbrueckii	1	0	0	2	0
Lactococcus lactis subsp. hordniae	0	1	0	0	0
Pediococcus damnosus	0	0	1	0	0
Uninterpretable or doubtful identification	8 (44)	6 (32)	8 (40)	11 (55)	20 (100)

These data suggest that the use of the current API 50 CH database for identification of commensal Lactobacillus species will lead to misidentification or uninterpretable results. In previous studies done by Nagy et al. (16) and Alvarez-Olmos et al. (1), L. acidophilus was identified in 36 to 49% of their subjects and was the most frequently isolated vaginal Lactobacillus species based on identification by the API 50 CH strips. Using this same system, we found that over half of the L. jensenii and L. gasseri isolates were erroneously identified as L. acidophilus.

Before Embley et al. classified L. vaginalis as a new species, it was grouped with L. fermentum (5). Our results indicated that almost half of the L. vaginalis isolates would have been identified as L. fermentum if the API 50 CH system was the only identification technique used. L. crispatus was once thought to be the same species as L. acidophilus until genotypic methods proved it to be a separate DNA homology group (11). The two L. crispatus reference strains that we tested could potentially have been identified to the correct species level because the biochemical profile for L. crispatus is included in the API 50 CH database. However, ATCC 33197 and ATCC 202225 were not identified as L. crispatus but as L. acidophilus, suggesting that phenotypic methods may not be adequate to reliably distinguish between closely related Lactobacillus species.

In conclusion, there is an inherent high level of phenotypic variability observed among commensal Lactobacillus species. This variability, in combination with a limited database for these species, limits the usefulness of this and other phenotypic identification methods. When lactobacilli are recovered from clinically significant specimens, use of genomic methods such as 16S rRNA sequencing may be preferable (19).