Genomic Epidemiology of Streptococcus pneumoniae Isolated in a Tertiary Hospital in Beijing, China, from 2018 to 2022

Streptococcus pneumoniae is one of the most common bacterial pathogens of a wide range of community-acquired infections. It has been more and more recognized that this bacterium could also play a role as a cause of nosocomial infections. In this study, by retrospective analysis of the phenotypic resistance characteristics and genomic characteristics of 52 S. pneumoniae isolates in a hospital in Beijing, China, from 2018 to 2022, we explored the carriage of resistance genes and mutations in penicillin-binding proteins corresponding to the resistances, and identified the population diversity based on the prediction of serotypes and identification of sequence types (STs). The isolates displayed resistances to erythromycin (98%), tetracycline (96%), sulfonamide (72%) and penicillin G (42%). Among the 52 isolates, 41 displayed multiple-drug resistance. In total, 37 STs and 21 serotypes were identified, and the clonal complex 271 serogroup 19 was the most prevalent subtype. Only 24 isolates (46.2%) of 7 serotypes were covered by the 13-valent pneumococcal conjugate vaccination. The isolates showed high carriages of resistance genes, including tet(M) (100%) and erm(B) (98.1%); additionally, 32 isolates (61.5%) had mutations in penicillin-binding proteins. We also observed 11 healthcare-associated infections and 3 cases infected by different subtypes of isolates. We did not find nosocomial transmissions between the patients, and these cases might be associated with the asymptomatic colonization of S. pneumoniae in the human population. Our results called for further active surveillance of these subtypes, as well as the continuous optimization of the treatment protocols.


Introduction
Streptococcus pneumoniae is one of the most common bacterial pathogens of a wide range of community-acquired infections, including respiratory infections, such as pneumonia, bronchitis and rhinitis, and invasive infections, such as meningitis, bacteremia and septicemia [1]. It is considered to be a major cause of death of children under 5 years old worldwide [2]. Moreover, it has been more and more recognized that this bacterium could also play a role as a cause of nosocomial infections [3]. The prevalence of S. pneumoniae is partially due to its wide carriage by healthy people on the mucosal surface of the upper airways, e.g., 20-60% of school-aged children and 5-10% of adults without children are healthy carriers [4].
S. pneumoniae are of large diversity. There are more than one hundred known serotypes and more than seventeen thousand sequence types (STs) [5]. The most prevalent serotypes include 19F, 23F, 14, 6B, 1, 19A and 3, and the most prevalent STs include ST199, ST320, ST81, ST271 and ST876 [6][7][8]. Currently, up to 13 serotypes are included in pneumococcal conjugate vaccines (PCVs). However, researchers have observed serotype replacement after the use of PCVs. The increased use of PCVs led to decreases in the serotypes covered by PCVs, and increases in the serotypes not in PCVs [9][10][11]. In addition, the recombination occurring in the gene cluster required for the capsular synthesis between different subtypes led to serotype switching [12], and the new strains might have advantages under the immune selective pressure constructed by the wide use of PCVs [13].
In 2017, the WHO included penicillin-non-susceptible S. pneumoniae as one of the 12 most critical, priority pathogens to guide the research, discovery and development of new antibiotics. The resistance of S. pneumoniae against ß-lactams and macrolides is a major concern worldwide [14]. In China, the proportion of penicillin-resistant S. pneumoniae (PRSP) is 86.9% in non-meningitis isolates in children [15]. The amino acid mutations occurring in penicillin-binding proteins (PBPs) result in decreased affinity for penicillin to inhibit the final steps of peptidoglycan synthesis by binding to the PBPs [14]. The macrolide resistance is most associated with two mechanisms: one is erm(B), encoding an enzyme that methylates 23S rRNA to induce target-site alteration; the other is mef (A) and mef (E), encoding active efflux pumps [16]. The increasing resistance is causing S. pneumoniae to become more dangerous in both community and healthcare settings.
In this study, by retrospective analysis of the phenotypic resistance characteristics and genomic characteristics of S. pneumoniae isolates in a hospital in Beijing, China, we explored the carriage of resistance genes and mutations corresponding to the resistances and identified the population diversity based on the prediction of serotypes and identification of STs. The clonal complex 271 serogroup 19 was the most prevalent subtype in our investigation, with high carriage of resistance genes and mutations in PBPs. Our results called for further active surveillance of these PRSP subtypes, as well as the continuous optimization of the treatment protocols.

Strain Collection and Associated Clinical Characteristics
We performed a retrospective study from July 2018 to July 2022 in Beijing Tsinghua Changgung Hospital, a Class A tertiary hospital in Beijing, China. Our hospital possesses 1000 beds and mainly provides services to the population of Beijing, having 24.1-39.1 thousand admissions each year in the last 5 years. S. pneumoniae isolates were cultured on a Columbia blood agar plate. The monoclonal isolates selected were first identified by MALDI-TOF mass spectrometry and then confirmed by the Vitek compact 2 system. All of the S. pneumoniae isolates were stored at −80 • C. The clinical characteristics, including the age, sex, underlying diseases, infection type, use of invasive devices and outcomes, were obtained from electronic medical records. We defined community-acquired infections (CAIs) as those that occurred before or within 48 h of admission, and healthcare-associated infections (HAIs) as those that occurred after 48 h following admission. This study was approved by the Beijing Tsinghua Changgung Hospital Ethics Committee (18,116-0-01), and the Guidelines for Human Experimentation (PRC) were followed throughout. Informed consent was not obtained for this study since clinical data were de-identified.

Antimicrobial Susceptibility Testing
The antimicrobial susceptibility testing was performed according to the 2020 Clinical and Laboratory Standards Institute (CLSI) guidelines, using the following agents: penicillin G, amoxicillin, cefotaxime, ceftriaxone, ertapenem, meropenem, levofloxacin, moxifloxacin, ofloxacin, erythromycin, telithromycin, linezolid, vancomycin, tetracycline, chloramphenicol and sulfonamide. The results were interpreted following the Clinical and Laboratory Standards Institute (CLSI) guidelines [17]. The multiple-drug resistance (MDR) phenotype was defined as being resistant to three or more different classes of antimicrobial agents.

Whole-Genome Sequencing and De Novo Assembly
The genomic DNA of all isolates was extracted with the TIANamp Bacteria DNA Kit (Cat. no. DP302, Tiangen, China), and then sequenced using the NovaSeq 6000 platform (Illumina, San Diego, CA, USA) by constructing paired-end libraries with an average insert size of 500 bp to obtain 150 bp reads.
The multi-locus sequence types of the isolates were analyzed by comparing them against the reference database (https://pubmlst.org/organisms/streptococcus-pneumoniae, accessed on 18 November 2022) by BLAST [20]. The serotypes of the isolates were predicted based on the clean read data using PneumoCaT (https://github.com/phe-bioinformatics/ PneumoCaT, accessed on 18 November 2022). Antimicrobial resistance genes, resistanceassociated mutations, virulence genes and plasmid replicon types were annotated by comparison with relevant databases, including ResFinder, Virulence Factor Database, ISfinder and plasmidFinder [21][22][23], using the BLAST software [20]. We selected thresholds of 90% identity and minimum length coverage of 80% to identify antimicrobial resistance and virulence genes.

Phylogenetic Analysis
As part of our phylogenetic analysis, the sequencing reads were mapped to the reference sequence of S. pneumoniae ATCC700669 (accession no.: NC_011900) using Bowtie 2 v2.2.8 software, and the single-nucleotide polymorphisms (SNPs) were analyzed with Samtools v1.9 and combined using the iSNV-calling pipeline (https://github.com/generality/ iSNV-calling, accessed on 18 November 2022) we previously constructed [24,25]. Highquality SNPs supported by more than five reads of mapping quality >20 were retained. The recombination sites were subsequently detected by Gubbins v2.4.1 software [26]. The concatenated sequences of filtered polymorphic sites conserved in all the isolates were used to perform phylogenetic analysis using the maximum likelihood method by IQ-Tree v2.1.2 software [27]. The best-fitting substitution model (GTR + F + ASC + R3) was identified using ModelFinder [28]. The consensus tree was constructed from 1000 bootstrap trees using UFBoot2 [29].

Sources of the Isolates and Their Antimicrobial Susceptibility Profiles
We identified 52 S. pneumoniae isolates from 47 patients in total ( Table 1). The samples from which the isolates were cultured were blood (8 isolates) and respiratory samples (44 isolates), including sputum (26 isolates), throat swab (10 isolates), bronchoalveolar lavage fluid (3 isolates) and bronchial aspirate (2 isolates). The information of 2 patients was missing, and the remainder included 37 males and 8 females, with a median age of 67 years (IQR: 6.5-74 year). Of these, 12 patients (26.7%) were 4-7-year-old children, and 27 patients (60%) were older than 60 years. The underlying diseases included sleep apnea, lung cancer, lung infection and cerebral infarction (Table 1). A total of 34 patients were diagnosed as CAIs, and 11 patients were diagnosed as HAIs. Among the seven cases with systematic infections of which the isolates were cultured from blood, three cases were CAIs and four were HAIs. Subjected to antimicrobial susceptibility testing, the isolates displayed resistance to multiple classes of antimicrobials, including erythromycin (98%), tetracycline (96%), sulfonamide (72%) and penicillin G (42%) (Figure 1). Among the isolates, 41 (78.8%) were identified as MDR and 21 (40.4%) were PRSP. We observed a decrease in resistance rate and minimum inhibitory concentration over the years, particularly in 2022.
Phylogenetic analysis revealed the isolates were of large diversity, corresponding with the relationships of the STs that the isolates belonged to. Those of the same STs and CCs were clustering together, and the others of different STs were phylogenetically distinct from each other. However, the serotypes of the isolates were not always in accordance with the STs and phylogenetic relationships. The serogroup 19 (serotypes 19A and 19F) was only identified in CC271, and isolates belonging to CC271 and serogroup 19 were the most common, indicating that this group might be the most prevalent. In contrast, serogroup 6 (serotypes 6A, 6C and 6D), 15 (serotypes 15A and 15C) and 23 (serotypes 23A and 23F) were all identified in two or more phylogenetically distinct STs.

Carriage of Resistance Genes and Mutations
In line with the phenotypic characteristics of the antimicrobial susceptibility, multiple resistance genes were identified from the genome sequences. The most common were tet(M)-positive in all of the isolates conferring resistance to tetracyclines, and erm(B)-positive in 51 (98.1%) isolates conferring resistance to macrolides, lincosamides and streptogramin B (MLSB phenotype). Gene cat was positive in 12 (23.1%) isolates of several STs conferring resistance to chloramphenicols, and mef (A) was positive in 10 of the 11 CC271 isolates, with the other 2 isolates conferring resistance to macrolides. In addition, resistance-associated point mutations were identified in 32 isolates located in 2 penicillin-binding proteins, PBP1a and PBP2×, conferring non-susceptibility to penicillin. The most common amino acid mutations included N605T in PBP2× in 30 isolates (57.2%), I371T in PBP2× in 28 isolates (53.8%) and N609D in PBP1a in 18 isolates (34.6%). A total of 19 isolates carried mutations in both of the PBPs, including all of the 11 CC271 isolates. Among the 21 PRSP isolates, 19 (90.5%) carried mutations in PBPs, indicating that the mutations were significantly associated with the non-susceptibility to penicillin (p < 0.05).   Phylogenetic analysis revealed the isolates were of large diversity, corresponding with the relationships of the STs that the isolates belonged to. Those of the same STs and CCs were clustering together, and the others of different STs were phylogenetically distinct from each other. However, the serotypes of the isolates were not always in accordance with the STs and phylogenetic relationships. The serogroup 19 (serotypes 19A and 19F) was only identified in CC271, and isolates belonging to CC271 and serogroup 19 were the most common, indicating that this group might be the most prevalent. In contrast, serogroup 6 (serotypes 6A, 6C and 6D), 15 (serotypes 15A and 15C) and 23 (serotypes 23A and 23F) were all identified in two or more phylogenetically distinct STs. . Phylogenetic analysis of S. pneumoniae isolates in our hospital correlated with the distribution of resistance genes (red-white heatmap) and resistance mutations (blue-white heatmap). The serotypes and STs of the isolates are also displayed. The patient numbers are attached after the isolate numbers, and the patients with more than one isolate are in red. N1 to N6 stand for the six new STs we identified.

Healthcare-Associated Infections and Cases Infected by Multiple Isolates
Although S. pneumoniae is the most common bacterial pathogen for CAIs, we found 15 infection cases occurring in the hospital, which were defined as HAIs (Table 1). However, by clinical investigation of the related cases in the same ward, we did not find any epidemiological relationships among the cases infected by isolates of the same STs. Therefore, we cannot confirm any nosocomial transmissions of the HAI cases.
In addition, from three patients we obtained more than one isolate. From patient P02, we obtained four isolates, including one ST271 serotype 19F isolate cultured from blood, and three serotype 16F isolates of a novel ST cultured from bronchial aspirate, blood in conduit or bronchoalveolar lavage fluid. The systematic infection occurred three months after admission, and the isolates were obtained within five days. From patient P14 (suffering from acute respiratory distress syndrome and septic shock), we also obtained two ST11958 serotype 23F isolates from peripheral blood and sputum within three days, respectively. From patient P09, we obtained a ST1937-1LV serotype 19F isolate and a ST7752 serotype 35A isolate, both from sputum samples, but the second isolation occurred two months later during a subsequent infection. These results indicated that infections might be associated with different isolates of different STs or serotypes at the same time, or successively.

Discussion
S. pneumoniae is the most common respiratory pathogen in children and older people. The diversity of this bacterium is extremely large, revealed by the large number of STs and serotypes [30]. Particularly, it can colonize on the mucosal surface of the upper airways amongst heathy people of all age groups, which are the sources of both CAIs and HAIs. In this study, we investigated the phenotypic resistance characteristics and molecular characteristics of S. pneumoniae in our hospital over a period of four years by whole-genome sequencing and analysis. Our results provided epidemiological information, which is meaningful for adjusting vaccine use and treatment strategies.
Although 37 STs were identified in our isolates, the CC271 was the dominant subtype, which is in line with previous studies [31,32]. The serotype prediction and analysis indicated that the PCV13 and PPV23, the two vaccines currently used in China, only covered a fraction of the serotypes we identified. Some less common serotypes are noteworthy, including serotype 6D, 15A, 15C and 16F, which are not covered by PCV13 and PPV23. Serotype replacement and the emergence of new serotypes could reduce the effectiveness of the vaccine. We also observed the high potential of serogroups 6, 15 and 23 to recombine amongst phylogenetically distinct STs, which indicated that we should pay more attention to the serotype variation of these serogroups.
Our results also demonstrated the serious antimicrobial resistances of S. pneumoniae, with high proportions of MDR and PRSP isolates. The carriages of erm(B), tet(M) and mutations in PBPs, including PBP1a and PBP2×, were remarkably high, corresponding to the high resistance rates to erythromycin (98%), tetracycline (96%) and penicillin G (42%). Fortunately, the isolates showed high susceptibility to fluoroquinolones. The resistance rates of levofloxacin (2%), ofloxacin (2%) and moxifloxacin (0%) were all quite low. These results indicated that fluoroquinolones would be effective. In contrast, the isolates showed certain resistance levels to cefotaxime (10%), amoxicillin (14%), ceftriaxone (14%) and chloramphenicol (18.8%), indicating the risks of treatment failure by these antimicrobials.
We also observed HAIs and cases infected by different subtypes of isolates. Since we did not find other cases infected by the same subtypes as those that infected the HAI cases, these cases might be infected by the isolates colonized in themselves or other patients not infected in the same wards, or even healthcare workers. Considering the high colonization rate of S. pneumoniae in healthy people, and the rapid transmission through the respiratory tract, the nosocomial transmissions and infections might be more common.
In this study, there were several limitations. Since it was a retrospective study and the number of isolates was limited, we could not reveal more about the changes in the molecular subtypes and resistance characteristics of the isolates in different patient groups or over time. We therefore focused more on the molecular epidemiology of the isolates and less on the diseases of the patients, since the information was limited. In addition, as this was a single-center study, the results might be more representative of the local population, meaning more studies and results from other institutions are necessary to reveal an overall epidemiology of S. pneumoniae infection in the region.

Conclusions
In summary, we performed a retrospective analysis of the characteristics of the resistance phenotypes, genes and molecular subtypes of S. pneumoniae isolates in our hospital over a period of four years. The high carriages of resistance genes erm(B) and tet(M), and mutations in PBPs, highlight the importance of systematic surveillance. The clonal complex 271 serogroup 19 was the most prevalent subtype, exhibiting MDR phenotypes with a high carriage of resistance genes and mutations in PBPs. Our results called for further active surveillance of these subtypes, as well as the continuous optimization of the treatment protocols based on the surveillance.