Association between Lactobacillus species and bacterial vaginosis-related bacteria, and bacterial vaginosis scores in pregnant Japanese women

Background Bacterial vaginosis (BV), the etiology of which is still uncertain, increases the risk of preterm birth. Recent PCR-based studies suggested that BV is associated with complex vaginal bacterial communities, including many newly recognized bacterial species in non-pregnant women. Methods To examine whether these bacteria are also involved in BV in pregnant Japanese women, vaginal fluid samples were taken from 132 women, classified as normal (n = 98), intermediate (n = 21), or BV (n = 13) using the Nugent gram stain criteria, and studied. DNA extracted from these samples was analyzed for bacterial sequences of any Lactobacillus, four Lactobacillus species, and four BV-related bacteria by PCR with primers for 16S ribosomal DNA including a universal Lactobacillus primer, Lactobacillus species-specific primers for L. crispatus, L. jensenii, L. gasseri, and L. iners, and BV-related bacterium-specific primers for BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium. Results The prevalences of L. crispatus, L. jensenii, and L. gasseri were significantly higher, while those of BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium were significantly lower in the normal group than in the BV group. Unlike other Lactobacillus species, the prevalence of L. iners did not differ between the three groups and women with L. iners were significantly more likely to have BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium. Linear regression analysis revealed associations of BVAB2 and Megasphaera with Nugent score, and multivariate regression analyses suggested a close relationship between Eggerthella-like bacterium and BV. Conclusion The BV-related bacteria, including BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium, are common in the vagina of pregnant Japanese women with BV. The presence of L. iners may be correlated with vaginal colonization by these BV-related bacteria.


Background
Bacterial vaginosis (BV) is the disturbed vaginal flora, in which normal lactobacilli are replaced by an overgrowth of various anaerobic bacteria [1]. This condition is common in women of reproductive age [1,2] and may cause malodorous vaginal discharge, although in many women it is asymptomatic [3]. In pregnant women, bacterial vaginosis has been suggested to be a risk factor of perinatal complications, including preterm birth [1,[4][5][6][7][8][9][10][11][12] and chorioamnionitis [4,13]. These complications are closely associated with neonatal morbidity and mortality worldwide.
Bacteria detected in BV flora include Gardnerella vaginalis, Mycoplasma hominis, Mobiluncus species (sp.), and other anaerobic bacteria, i.e., Peptostreptococcus sp., Prevotella sp., and Bacteroides sp. [1,[14][15][16]. Recently, bacteria such as Atopobium vaginae, Megasphaera sp., Leptotrichia sp., and Eggerthella-like bacterium have been reported as microorganisms related to this condition by molecular analyses [17][18][19]. Fredricks et al. identified three phylogenetically distinct bacterial DNA sequences in human vaginal samples highly specific for this condition and designated them BV-associated bacteria (BVAB) 1~3 [19]. They showed that BVABs, Megasphaera, Leptotrichia, and Eggerthella-like bacterium are more specific for BV than Gardnerella and Atopobium [19]. Among BVABs, BVAB2 was shown to be more sensitive for BV than BVAB1 and BVAB3, while the specificities of three BVABs were similar [19]. We use the term "BV-related bacteria" for bacteria including BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium in this manuscript. However, it is important to note that these organisms have not been proven to be causative agents of BV.
The healthy human vaginal flora in reproductive age is usually predominated by Lactobacillus species. Their metabolic products, such as hydrogen peroxide (H 2 O 2 ), lactic acid, and bacteriocin are believed to play an important role in maintenance of the normal vaginal flora by inhibiting colonization by other pathogens [20][21][22][23]. The predominant Lactobacillus species in the normal lactobacillary flora were shown by molecular biological analyses to be L. crispatus, L. gasseri, and L. jensenii [23][24][25][26][27][28]. In recent studies, L. iners described as L. 1086V by Antonio et al. [24] was identified as one of the common Lactobacillus species colonizing the human vagina [18,[28][29][30][31]. Only 9% of the strains of this species produce H 2 O 2 , whereas almost all strains of L. crispatus and L. jensenii produce H 2 O 2 [24].
To date, there have been few studies regarding the frequencies of the BV-related bacteria described above and Lactobacillus species in healthy and abnormal vaginal flora in pregnant women. The present study was performed to evaluate the prevalence of the BV-related bacteria and the common Lactobacillus species in normal and BV flora in pregnant Japanese women. We used a specific PCR method targeting the bacterial 16S ribosomal DNA (rDNA) region for this purpose.

Patients
A total of 163 pregnant Japanese women were enrolled in this study during routine prenatal visits at Hokkaido University Hospital from May 2005 to February 2006. Informed consent was obtained from all participants in verbal form. Vaginal fluid samples were collected at a mean of 23 weeks of gestation. Estimated date of delivery was determined from the last menstrual period and early gestational fetal ultrasonographic measurements.

Sample collection, Nugent's scoring, and bacterial culture
A sterile speculum was inserted into the vagina and a specimen of vaginal fluid was obtained by brushing the posterior vaginal fornix with a swab. A vaginal smear was prepared by rolling a swab onto a glass slide, which was then air-dried, heat-fixed, and Gram-stained. The smears were then assessed according to Nugent criteria [32]. The other swab was spread onto Columbia blood agar plates, and incubated at 35°C under aerobic conditions in 5% CO 2 and anaerobic conditions for 48 h. Lactobacilli were identified to the genus level by Gram staining of colonies and from colony morphology on blood agar plates.

DNA extraction and PCR
Another swab was placed in 1 ml of PBS with subsequent vigorous vortexing to dislodge cells. The cells were centrifuged at 14,000 rpm for 5 min. The pellet was digested with proteinase K at 56°C for 30-60 min and the DNA was extracted and purified with a QIAmp DNA Mini Kit Sequences and annealing temperatures for the various primer sets are listed in Table 1 [19,33]. All primers were located in the 16S rDNA region. PCR was carried out for 40 cycles. For the Lactobacillus genus and its four species, the denaturation was performed at 95°C for 15 s followed by a 1-min annealing and extension step. For four BVrelated bacteria, the denaturation step was set at 94°C for 30 s, followed by the annealing step for 30 s for BVAB2, Megasphaera, and Leptotrichia and for 40 s for Eggerthellalike bacterium, with extension at 72°C for 1 min for all reactions. A final extension step at 72°C for 7 min was added for all reactions. Aliquots of 7 µl of the PCR products were electrophoresed in agarose gels and visualized by ultraviolet transillumination after ethidium bromide staining.
The specificity of the Lactobacillus species-specific PCR for 14 common intestinal Lactobacillus species was evaluated and confirmed using 10 6 copies of one Lactobacillus species to each reaction as template DNA ( Table 2). The universal Lactobacillus primer amplified all of these Lactobacillus species. The specific primers for L. crispatus, L. jensenii, and L. gasseri only amplified the corresponding species and did not amplify 13 other species (Table 2). They also did not amplify a cloned fragment of 16S rDNA region of L. iners. The specific primers for L. iners did not amplify any of 14 Lactobacillus species (Table 2). We ana-lyzed PCR products from several vaginal samples amplified by the specific primers for L. iners and confirmed that the sequences of PCR products were completely consistent with L. iners (GenBank AY526083).
The sensitivities of the species-specific PCR for L. crispatus, L. jensenii, and L. gasseri were measured using serial dilutions of DNA solution of the reference strain. Similarly, for L. iners, serial dilutions of a cloned fragment of 16S rDNA region of L. iners were used instead. The sensitivity of the species-specific PCR for L. crispatus, L. jensenii, and L. gasseri and that of L. iners PCR were 10 2 to 10 3 copies and 10 2 copies per reaction, respectively.

Statistical analysis
Fisher's exact probability test was used for statistical analysis. Multivariate logistic-regression analysis using SPSS™ for Windows was performed to evaluate the independent risk factors, and P < 0.05 was considered statistically significant. The clinical characteristics of the pregnant women are summarized in Table 3. There were no statistically significant differences in the mean maternal age, number of nulliparous women, gestational week at sampling, gestational week at delivery, or birth weight of the neonate among the three groups. Of the total of 132 women, 35 (26.5%) delivered at <37 weeks, 9 (6.8%) at <33 weeks, and 4 (3.0%) at <30 weeks of gestation.

Detection rate of lactobacilli and BV-related bacteria in three groups by PCR
Genus Lactobacillus (any Lactobacillus) was detected in almost all women irrespective of Nugent score (Table 4). The detection rates of L. crispatus, L. jensenii, and L. gasseri were significantly higher in the normal group than in the BV group, while that of L. iners did not differ between the three groups. In contrast, the detection rates of BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium were significantly lower in the normal group than in the BV group.

Independent risk factors for abnormal Nugent score
Multivariate logistic regression analysis was performed to evaluate the independent contributions of the various bacteria to the abnormal vaginal flora (Tables 5 and 6). Seven bacteria, i.e., L. crispatus,L jensenii,L. gasseri, BVAB2, Megasphaera,Leptotrichia and, Eggerthella-like bacterium, were entered as variates to be analyzed (P < 0.1, Fisher's exact probability test). The absence of L. crispatus and the presence of Megasphaera were selected as two independent risk factors of Nugent score ≥ 4, giving Odds ratios of 0.2 and 13.3, respectively (Table 5). Likewise, the presence of Eggerthella-like bacterium was selected as an independent risk factor of Nugent score ≥ 7, giving an Odds ratio of 6.2 ( Table 6). Linear regression analyses revealed that BVAB2 and Megasphaera were associated with Nugent score.

Coexistence of BV-related bacteria with L. iners
L. iners was detected by PCR in 55 of 132 (41.7%) women, and its prevalence did not differ between the groups classified according to Nugent score (Table 4). However, the presence of L. iners appeared to be positively associated with colonization by BV-related bacteria ( Table 7). The detection rates of all BV-related bacteria were significantly higher in samples harboring L. iners. No such association was seen between the presence or absence of L. iners and the detection rate of any other Lactobacillus species.

Discussion
In the present study, we confirmed that L. crispatus, L. gasseri, and L. jensenii were common species in pregnant Japanese women with normal vaginal flora by speciesspecific PCR of the 16S rDNA region. These three species were less prevalent in women with BV. In contrast, four BV-related bacteria, i.e., BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium, were detected at higher prevalence in women with BV. As all these results were in accordance with those of Fredricks et al. [19] who analyzed the vaginal fluid of non-pregnant women with and without BV using the broad-range 16S rDNA PCR and cloning methods, BV is suggested to have remarkably similar microbiological profiles among women with different demographic characteristics, including race and pregnancy, as suggested by the conventional cultivation method.
L. crispatus, L. gasseri, and L. jensenii are common Lactobacillus species found in the vagina [24][25][26][27][28]31,34]. L. iners, described recently as a new Lactobacillus species [29], is one of the common Lactobacillus species of the vaginal microbiota [18,19,23,28,30,31], which was also confirmed in the present study. The results showed that L. iners was present in 40% to 50% of women irrespective of Nugent score, as observed in an earlier study [19]. We examined twelve samples positive for L. iners (6 from normal flora and 6 from BV flora) to determine whether the abundance of L. iners was different in the two groups. The   species-specific PCR for L. iners using serial dilutions of each sample revealed that both normal and BV flora contained 10 3 to 10 5 copies/µl of L. iners and the median concentration was 10 4 copies/µl for both.
As the presence of H 2 O 2 -producing lactobacilli in the vaginal fluid is associated with a reduced risk of BV [15,24] and because the concentration of H 2 O 2 in the vaginal fluid is low in women with BV as compared with those with normal vaginal flora [21], the H 2 O 2 -producing ability of lactobacilli is thought to play a significant role in protecting the vaginal ecosystem from BV infection, although direct evidence to support this notion is lacking. Nearly all strains of L. crispatus and L. jensenii have been reported to produce H 2 O 2 , whereas only 9% of the strains of L. iners produce H 2 O 2 [24]. The prevalences of L. crispatus and L. jensenii were significantly higher in the normal group than in the BV group and the detection rates of all BV-related bacteria were significantly higher in women with than in those without L. iners in this study. Although this observation is consistent with the notion that H 2 O 2producing ability of lactobacilli is important in protecting the vaginal ecosystem from BV infection, it remains to be determined whether these observations resulted from differences in H 2 O 2 -producing ability of these lactobacilli.
The newly proposed "BV-related bacteria," including BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium, were all shown to be associated with BV in the present study, confirming the results of a recent study by Fredricks et al. [19]. However, the detection rates of these bacteria in women with BV were lower, while those in women with normal flora were similar to their results [19]. BVAB2 is cultivation-resistant, one of three bacteria (provisionally named BV-associated bacteria: BVAB1, BVAB2, and BVAB3) newly found to be highly specific for BV in the vagina of non-pregnant women [19], and not closely related to other bacteria as shown by comparison of 16S rDNA. In the present study, BVAB2 was present in 38.5% (5/13) and 3.1% (3/98) of women with BV and with normal vaginal flora, respectively, while Fredricks et al. reported these rates to be 88.9% (24/27) and 4.3% (2/ 46), respectively [19]. Similarly, detection rates of Megasphaera (69.2%),Eggerthella-like bacterium (53.8%), and Leptotrichia (53.8%) in women with BV in the present study were lower than those of 96.3%, 92.6%, and 85.2% reported by Fredricks et al. [19], while detection rates of Megasphaera (11.2%),Eggerthella-like bacterium (7.1%), and Leptotrichia (14.3%) in women with normal vaginal flora were comparable to their values of 8.7%, 8.7%, and 4.3%, respectively [19].
The results of the present study raised the possibility that the four BV-related bacteria were less prevalent in pregnant Japanese women with BV as compared with nonpregnant American women. However, the number of subjects with BV in the present study was too low to draw definitive conclusions about the prevalence of bacteria in different populations. Further studies using different demographic populations are needed to determine the roles of these BV-related bacteria in the pathogenesis of BV.  Twelve (92%) of 13 women with BV were positive for genus Lactobacillus by 16S rDNA PCR using the universal Lactobacillus primer, including 5 women with L. iners, one with both L. iners and L. crispatus, one with L. crispatus, and 5 with unspecified Lactobacillus. Of these 13 women, only one with L. iners was positive for Lactobacillus by general cultivation methods and positive for Gram-positive rods on Gram staining. These results suggested that many women with BV harbor genus Lactobacillus in the vagina and that the number of these lactobacilli colonizing the vagina is small. Further, as L. iners has been reported to require specialized blood agar media for isolation [29], the conventional culture method used in this study may have failed to reveal its colonization in the vagina.

Conclusion
Our results suggested that BV-related bacteria, including BVAB2, Megasphaera, Leptotrichia, and Eggerthella-like bacterium, were associated with BV in pregnant Japanese women. The presence of L. iners, one of the common Lactobacillus species in the vagina, may be correlated with vaginal colonization by these BV-related bacteria. It remains to be determined whether BV-related bacteria cause BV or are common and abundant as a consequence of BV.