We analyzed a total of 216 clinical isolates of SMG from the collection of isolates identified in our microbiology laboratory between 2014 and 2016. The isolates consisted of 101 S. pneumoniae from blood (68), bronchial aspirates (14), conjunctival and otic exudates (5), normally sterile fluids (5), bronchoalveolar lavage (3), sputum (3), and others (3) and 115 non-pneumococcal SMG isolates from blood (52), normally sterile fluids (16), abscesses (14), urine (13), different types of biopsies (11), surgical wounds (5), and bronchial or tracheal aspirates (4). Besides, 8 reference strains were also analyzed: S. anginosus (ATCC 12395), S. equinus (ATCC 33317), S. gordonii (ATCC 33399), S. intermedius (ATCC 27335), S. mutans (ATCC 25175), S. oralis (ATCC 35037), S. pneumoniae (ATCC 27336), and S. salivarius (ATCC 7073). All the isolates were subcultured on Columbia agar + 5% sheep blood (Biomérieux, Marcy L'Étoile, France) and incubated at 37°C in 5% CO₂ for 48 h.

All the strains were identified as VGS by Gram staining, catalase test and alpha-haemolysis. Besides, the optochin susceptibility and bile solubility tests confirmed the identification of S. pneumoniae. On the other hand, the non-pneumococcal isolates were all confirmed by their resistance to optochin (Kellogg et al., 2001; Facklam, 2003).

DNA sequence analysis of the superoxide dismutase (sodA) and the pneumolysin (plys) genes was performed in order to provide species-level identification to non-pneumococcal isolates. A 480-bp fragment within the sodA gene was amplified using the degenerated primers d1 (5′'-CCITAYICITAYGAYGCIYTIGARCC-3′') and d2 (5′'-ARRTARTAIGCRTGYTCCCAIACRTC-3; Poyart et al., 1998). For the amplification of the plys gene, the forward primer (5′'-TGCAGAGCGTCCTTTGGTCTAT-3′') and reverse primers (5′'-TGCAGAGCGTCCTTTGGTCTAT-3′') were used in combination with a labeled probe (VIC–5′'-TGGCGCCCATAAGCAACACTCGAA-3′') as described by Corless et al. (2001). The amplified fragments were purified using the GFX PCR DNA and Gel Band Purification kit (GE Healthcare, UK) and sequenced with an AbiPrism 3,130 × l Genetic Analyzer equipment (Applied Biosystems, CA, USA). Sequences were compared with the GeneBank library using BLAST alignment software (http://www.ncbi.nlm.nih.gov/blast) and BIBI (bioInformatic bacterial identification tool: https://umr5558-bibiserv.univ-lyon1.fr/lebibi/lebibi.cgi). And the identification of the strains was carried out following the criteria established by the CLSI guidelines –(CLSI, 2008)-, with species level identification only when the homology was >99%.

Conventional methods and DNA sequencing analysis were considered as the reference methods for SMG identification.

All SMG isolates were analyzed by MALDI-TOF MS, using a Microflex LT bench top mass spectrometer (Bruker Daltonics, Bremen, Germany) as described previously –(Rodríguez-Sánchez et al., 2016)- Briefly, FlexControl 3.3 and MALDI Biotyper 3.1 software (Bruker Daltonics, Germany) was used for the acquisition of the spectra and for spectra analysis and subsequent comparison with the database (updated with 6,903 MSP entries), respectively. The available database contains 31 S. pneumoniae, 39 S. mitis and 38 S. oralis references. A commercial standard Bacterial Test Standard (Bruker Daltonics, Bremen, Germany) was used for calibration of the equipment and acquisition of mass spectra was performed using default settings.

Sample preparation prior to MALDI-TOF identification has been described before –(Rodríguez-Sánchez et al., 2014)-. A small amount of bacteria was transferred to a polished steel MALDI target plate using a wooden toothpick. The spots were covered with 1 μl of 100% formic acid and allowed to dry at room temperature prior to the addition of 1 μl of matrix (α-cyano-4-hydroxy-cinnamic acid solution in 50% acetonitrile and 2.5% trifluoroacetic acid) -prepared following the manufacturer's instructions-. The spots were allowed to air dry before spectra acquisition and comparison with the database. All the isolates were analyzed by MALDI-TOF MS in duplicates and the higher score value was recorded as well as the identification provided by MALDI-TOF MS. Score values for species level (≥2.0) and genus-level identification (≥1.7) used in this study are those proposed by the manufacturer. Score values below 1.7 were, accordingly, considered as not reliable. Identification agreement at the species level was considered when MALDI-TOF provided the same result as the sodA gene sequencing and the conventional methods. However, the identification was considered a major error when MALDI-TOF MS identified an isolate as S. pneumoniae whilst the conventional methods and sodA gene sequencing indicated that the isolate was a non-pneumococcal isolate and viceversa. When the discrepancy was between two non-pneumococcal SMG species it was considered as a minor error.

For peak analysis, 20 S. pneumoniae, 20 S. mitis and 20 S. oralis isolates were submitted to a standard protein extraction procedure: a 1 μl-loopful of bacteria was resuspended in 300 μl of distilled water and 900 μl ethanol, vortexed and centrifuged at maximal speed for 1 min; the pellet was then resuspended in 20 μl of 70% formic acid and the same amount of acetonitrile, vortexed again and centrifuged again for another minute at maximal speed. 1 μl of supernatant was pipetted onto the MALDI plate, allowed to dry and covered with 1 μl of matrix. The supernatant from every strain was spotted onto 6 positions. In a different spot, 1 μl of BTS was spotted on top of the dried supernatant, allowed to dry and covered with 1 μl of matrix. This position was used for calibration purposes.

FlexAnalysis 3.3 (Bruker Daltonics, Bremen, Germany) was applied for the comparison of mass spectra from the different SMG species. Spectra were smoothed, the baseline was subtracted and BTS was used as references for calibration purposes. Peaks from individual spectra were compared for strain-specific variations and spectra from all 20 isolates from the same species were considered together as a single MSP entry in order to find species-specific peaks.

The 60 isolates submitted to protein extraction were used for the characterization of species-specific peaks present in each Streptococcus species. Afterwards, the presence of species-specific peaks was analyzed in all S. pneumoniae-101-, S. mitis-50-, and S. oralis-55- isolates included in this study.

The hospital Ethics Committee approved this study and gave consent for its development. Since this study has been performed on microbiological samples, not human products, all the conditions to waive the informed consent have been met.

During the study period, 216 SMG isolates were identified with MALDI-TOF MS as part of the clinical microbiology laboratory routine. MALDI Biotyper with 6,903 MSPs correctly discriminated S. pneumoniae from non-pneumococcal SMG isolates. Therefore, no major errors were detected when this database was implemented. Concordantly, all non-pneumococcal isolates were correctly identified at group level, although at the species level 14 S. mitis/oralis isolates were misidentified according to sodA sequencing analysis results (Table 1).

Interestingly, one of the isolates selected for this study and wrongly listed as a non-pneumococcal SMG isolate was consistently identified by MALDI-TOF MS as S. pneumoniae (score value = 2.524). When its identification by the optochin test, the bile solubility test and sodA gene sequencing was double-checked it turned out that the isolate was actually a S. pneumoniae one, thus, confirming MALDI-TOF MS identification.

Despite MALDI-TOF MS high level of accuracy in the identification of SMG isolates, in this study we have observed that challenging strains may still be found, leading to misidentifications. Thus, the protein spectra from the 101 analyzed S. pneumoniae isolates were compared with the spectra from those belonging to the S. mitis group, in order to single out S. pneumoniae specific peaks. One peak was found in at least 95.0% of the S. pneumoniae isolates -4,964.32 m/z- and other two peaks in 90.5% of the isolates: 6,888.90 and 9,516.46 m/z –Table 2. Their presence was also confirmed in the reference strain ATCC 27336 (Figure 1).

On the other hand, 92.0% of the S. oralis strains showed two specific peaks of 5,297.61 and 5,822.53 m/z also present in the reference strain ATCC 35037. These peaks were not detected in S. pneumoniae isolates neither in the other species of non-pneumococcal VGS analyzed. Besides, S. mitis and S. oralis isolates shared a peak of 6,839.07 m/z. This peak was the only species-specific feature of S. mitis isolates, whilst S. oralis displayed it together with the 5,297.61 and 5,822.53 m/z peaks (Figure 1).

Finally, although MALDI-TOF MS accurately identifies the remaining VGS groups (S. anginosus, S. bovis, S. mitis, S. mutans, S. salivarius, and S. sanguinis) and no misidentifications with S. pneumoniae have been reported as far as the authors know, peak comparison was performed between S. pneumoniae isolates and the rest of non-pneumococcal VGS, including the reference strains (data not shown). None of the S. pneumoniae specific peaks were found alone or in combination in the remaining non-pneumococcal VGS included in this study.

In our study, out of the 14 non-pneumococcal SMG isolates misidentified by the Biotyper system with 6,903 MSPs, all but two isolates could be correctly identified by peak analysis. The discrepant identifications corresponded to two S. mitis isolates identified by sodA sequencing analysis that were identified as S. oralis by the updated database and by peak analysis –Table 3. Further genomic targets should be analyzed in order to resolve these discrepancies.

In summary, the updated database allowed the correct species-level identification of 100% of the S. pneumoniae isolates and 100% of the non-pneumococcal isolates at the group-level. Besides, the performance of peak analysis confirmed the identification of all S. pneumoniae isolates and provided a correct species-level discrimination of 103/105 S. mitis and S. oralis isolates (98.1%).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.