Molecular Characterization of Penicillin-Binding Protein2x, 2b and 1a of Streptococcus pneumoniae Causing Invasive Pneumococcal Diseases in China: A Multicenter Study

Streptococcus pneumoniae is a common human pathogen that can cause severe invasive pneumococcal diseases (IPDs). Penicillin-binding proteins (PBPs) are the targets for β-lactam antibiotics (BLAs), which are the common empirical drugs for treatment of pneumococcal infection. This study investigated the serotype distribution and antibiotic resistance patterns of S. pneumoniae strains causing IPD in China, including exploring the association between penicillin (PEN) susceptibility and PBPs variations. A total of 300 invasive S. pneumoniae isolates were collected from 27 teaching hospitals in China (2010-2015). Serotypes were determined by Quellung reaction. Serotypes 23F and 19F were the commonest serotypes in isolates from cerebrospinal fluid (CSF), whilst serotypes 19A and 23F were most commonly seen in non-CSF specimens. Among the 300 invasive S. pneumoniae strains, only one strain (serotype 6A, MIC = 0.25 μg/ml) with PEN MIC value ≤ 0.25 μg/ml did not have any substitutions in the PBPs active sites. All the strains with PEN MIC value ≥ 0.5 μg/ml had different substitutions within PBPs active sites. Substitutions in PBP2b and PBP2x active sites were common in low-level penicillin-resistant S. pneumoniae (PRSP) strains (MIC = 0.5 μg/ml), with or without PBP1a substitution, while all strains with PEN MIC ≥ 1 μg/ml had substitutions in PBP1a active sites, accompanied by PBP2b and PBP2x active site substitutions. Based on the three PBPs substitution combinations, a high degree of diversity was observed amongst the isolates. This study provides some new insights for understanding the serology and antibiotic resistance dynamics of S. pneumoniae causing IPD in China. However, further genomic studies are needed to facilitate a comprehensive understanding of antibiotic resistance mechanisms of S. pneumoniae.

Streptococcus pneumoniae is a common human pathogen that can cause severe invasive pneumococcal diseases (IPDs). Penicillin-binding proteins (PBPs) are the targets for β-lactam antibiotics (BLAs), which are the common empirical drugs for treatment of pneumococcal infection. This study investigated the serotype distribution and antibiotic resistance patterns of S. pneumoniae strains causing IPD in China, including exploring the association between penicillin (PEN) susceptibility and PBPs variations. A total of 300 invasive S. pneumoniae isolates were collected from 27 teaching hospitals in China (2010China ( -2015. Serotypes were determined by Quellung reaction. Serotypes 23F and 19F were the commonest serotypes in isolates from cerebrospinal fluid (CSF), whilst serotypes 19A and 23F were most commonly seen in non-CSF specimens. Among the 300 invasive S. pneumoniae strains, only one strain (serotype 6A, MIC = 0.25 µg/ml) with PEN MIC value ≤ 0.25 µg/ml did not have any substitutions in the PBPs active sites. All the strains with PEN MIC value ≥ 0.5 µg/ml had different substitutions within PBPs active sites. Substitutions in PBP2b and PBP2x active sites were common in low-level penicillin-resistant S. pneumoniae (PRSP) strains (MIC = 0.5 µg/ml), with or without PBP1a substitution, while all strains with PEN MIC ≥ 1 µg/ml had substitutions in PBP1a active sites, accompanied by PBP2b and PBP2x active site substitutions. Based on the three PBPs substitution combinations,

INTRODUCTION
Streptococcus pneumoniae is one of the most common Grampositive cocci that is mainly transmitted through the respiratory tract. The organism usually colonizes the human nasopharynx and can migrate to the middle ear and lungs causing local non-invasive pneumococcal disease (NIPD) such as otitis media and pneumonia in immune-deficient people (Lynch and Zhanel, 2009;Henriques-Normark and Tuomanen, 2013). Data from the World Health Organization (WHO) shows that pneumonia killed 808, 694 children under five years old in 2017, accounting for 15% of all deaths in children. S. pneumoniae is the most common pneumonia pathogen in children worldwide, with a mortality rate in children much higher than other diseases such as AIDS, malaria, and measles (Lynch and Zhanel, 2009). In Europe and America, S. pneumoniae is also the most common cause of community-acquired pneumonia in adults (World Health Organization [WHO], 2007). In addition to respiratory tract infections, S. pneumoniae can also migrate to the blood and brain and cause severe invasive pneumococcal disease (IPD), such as bacteremia, meningitis etc., causing a huge economic and medical burden on both developed and developing countries (Mehr and Wood, 2012).
The major empirical antimicrobial drugs used in the treatment of S. pneumoniae infections are β-lactam antibiotics (BLAs), which act on the bacterial cell wall. Penicillin-binding proteins (PBPs) are crucial enzymes in the biosynthesis of peptidoglycan (PG), a major cell wall component that surrounds the cytoplasmic membrane and is required to maintain the shape and osmotic stability of bacteria (Hakenbeck et al., 2012). The target of BLAs are PBPs, and function by covalently binding to the active site serine of PBPs through the β-lactam ring, thereby interfering with the synthesis of bacterial cell walls and eventually leading to bacterial cell death. With the widespread use of antibiotics, penicillin-intermediate S. pneumoniae (PISP) and penicillin-resistant S. pneumoniae (PRSP), commonly referred to as penicillin-non-susceptible S. pneumoniae (PNSP), have emerged and are detected continually worldwide. Data from the Asian Network for Surveillance of Resistant Pathogens (ANSORP) shows that the isolation rate of PNSP from 2012 to 2017 (9.0%) was significantly higher than that from 2008 to 2009 (4.9%), and the detection rates of PNSP from patients in China was 1.9% (2012-2017, oral breakpoint) (Kim et al., 2012(Kim et al., , 2020.
The main mechanism of BLAs resistance by S. pneumoniae is through PBPs substitutions. Alterations in PBPs via substitutions reduce their reactivity for β-lactam attachment to the binding site and thereby reduce their effectiveness .
S. pneumoniae has six PBPs, but only three PBPs, PBP2x, PBP2b and PBP1a play a main role in BLAs resistance. Alterations in all other PBPs have been described occasionally . Mosaicity is the product of homologous recombination that causes the sequence diversity in S. pneumoniae. In most resistant clinical S. pneumoniae isolates, the sequencing revealed that the mosaic genes encode PBP2x, PBP2b, and PBP1a (Laible et al., 1991;Martin et al., 1992;Sibold et al., 1994).The active sites of PBPs comprises three conserved sequences SXXK, SXN and KT(S)G. The serine of the SXXK motif is the active site residue that reacts with BLAs. They are located in PBP2x: 337STMK340, 395SSN397, 547KTG549; in PBP2b: 386STMK389, 443SSN445, and 615KTG617; and in PBP1a: 370STMK373, 428SRN430, 557KTG559 (Hakenbeck et al., 2012). Changes in these active site motifs and their adjacent sequences lead to a decrease in the affinity of PBPs to penicillin resulting in antibiotic resistance (Zhanel et al., 2006;Hakenbeck et al., 2012). Few studies have been carried out to understand the association between penicillin susceptibility and PBPs variations in S. pneumoniae isolates from patients with IPD in mainland China.
Due to the widespread use of antimicrobial drugs, and the fastidious nature of S. pneumoniae, the number of isolates from IPD specimens is very low, and hence a scarcity of relevant research studies (Xue et al., 2010;Yao et al., 2011;Quan-Cheng et al., 2016). This study aimed to analyze the serotype distribution and antibiotic resistance pattern of S. pneumoniae strains causing IPD in China, and to explore the association between penicillin susceptibility and PBPs variations.

Bacterial Isolates
A total of 300 non-duplicate invasive S. pneumoniae isolates from 27 teaching hospitals in 13 provinces of China (2010China ( -2015 were studied (Supplementary Table 1). The isolates were transported to the Department of Clinical Laboratory in Peking Union Medical College Hospital for re-identification and further analysis. The most common specimen type was blood, accounting for 72.7% (218/300) of the isolates, followed by cerebrospinal fluid (CSF) (19.0%, 57/300) and pleural effusion (5.7%, 17/300). Other specimen types included ascites, joint drainages, pleural drainage and lung tissue, each accounting for ≤ 2.0% of the isolates. The majority of the patients were males, accounting for 65.7% (197/300) of the isolates. The average age of patients was 46 ± 26.69 years old.

Serotyping
All the S. pneumoniae isolates were serotyped by Quellung reaction (Maha et al., 2014). The serotype was first determined by latex agglutination test using the checkerboard typing system. Then specific type antiserum was mixed with the bacterial suspension to determine the final serotype. Capsular swelling was observed under oil immersion microscope. If the test strain was negative with all antisera, it was classified as nontypeable (NT).

Penicillin-Binding Protein Gene Amplification and Sequencing
The isolates were cultured on blood agar plates and incubated overnight at 35 • C in a 5% CO 2 atmosphere. DNA was extracted using the AxyGen amp DNA Mini Extraction Kit (Axygen, United States) according to the manufacturer's instructions. The final pure DNA was stored at −20 • C until use. The nucleotide sequence of an around1-kb region encoding the penicillinbinding domain of pbp2x, pbp2b and pbp1a genes were amplified and sequenced based on published primers (Table 1; Zhanel et al., 2006). PCR products were sent to Beijing RuiBiotech Co., Ltd., for sequencing. CLC Sequence Viewer software (CLC, Denmark) was used to manually correct the sequencing peaks and sequences to ensure the sequencing quality, and a two-way splicing was carried out. The spliced sequences were translated to simulated protein sequences and compared with PBP2x, PBP2b, and PBP1a corresponding to S. pneumoniae reference strain R6 (PSSP, GenBank accession No. NC003098) for variation analysis.

Data Analysis and Statistical Analysis
Differences in antimicrobial susceptibility were analyzed by MIC range, MIC 50 and MIC 90 , and statistical analysis was performed by chi-square test or Fisher's exact probability test using SPSS software (version 22.0, SPSS Inc., Chicago, IL, United States). P value < 0.05 was considered statistically significant.

Antimicrobial Susceptibility
Concerning non-BLAs drugs, all the S. pneumoniae isolates were susceptible to LEV, VA and LZD, with MIC 90 values of 1 µg/ml, 0.5 µg/ml, and 1 µg/ml. The prevalence of resistance of the isolates to SXT was 65.3%. Resistance rates to CLA, E and DA were extremely high, all above 90%. PISP accounted for 4.3% of the isolates based on the nonmeningitis (R ≥ 8 µg/ml) breakpoint while none, 67.7% and 44.7% of the isolates were classified as PRSP based on nonmeningitis, meningitis (R ≥ 0.12 µg/ml) and oral administration (R ≥ 2 µg/ml) breakpoints, respectively. Susceptibility to AMC was about 97.7%. The prevalence of resistance of the isolates to the second-generation cephalosporins CXM was about ≥60%.  In the PBP2b conserved motifs or nearby sites, 190 isolates had T446A substitution (threonine → alanine), 3 isolates had T446S substitution (threonine → serine), and 72 isolates had A619G substitution (alanine → glycine), with substitution rates of 63.3, 1, and 24%, respectively. In the PBP1a conserved motifs or nearby sites, 53 isolates had T371A substitution (threonine → alanine), 122 isolates had T371S substitution (threonine → serine), and 175 isolates had P432T substitution (proline → threonine), with substitution rates of 17.7, 40.7, and 58.3%, respectively (Table 4). Based on the serotype and PEN MIC distribution among the isolates, ten isolates had only one PBP gene active site or nearby sites substitution, including two isolates each of serotype 6A and 6B, one isolate each of serotypes 19F, 22 F and 23A, and three isolates of serotype 23F. Seven of these isolates had PBP2b active sites or nearby sites substitutions; two had only PBP2x active sites or nearby sites substitutions, while none had any substitution in single PBP1a active sites or nearby sites. Save for one isolate of serotype 6A with PEN MIC of 0.25 µg/ml, all these strains had an MIC of 0.5 µg/ml. Thirteen isolates had substitutions in two of the PBPs gene active sites or nearby sites, including four isolates of serotype 20, two isolates of serotype 23A, and one isolate each for serotypes 6B, 13, 14, 29, 15B, 19A, and 23F. Among them, three isolates had substitutions in both PBP2b and PBP1a genes (serotypes 19A, 23F, and 23A), and the rest had substitution in PBP2x and PBP2b. The PEN MIC of most isolates in the group with two substitutions in the PBPs gene active sites or nearby sites was 0.5 µg/ml. The remaining 171 isolates had substitutions in the active sites or nearby sites of all the three PBPs genes, mainly distributed in serotypes 14 (n = 26), 19A (n = 39), 19F (n = 40) and 23F (n = 38).

Association Between Specimen Types and Variations in Penicillin-Binding Proteins Active Sites
Among the 300 invasive S. pneumoniae isolates, 57 (19.0%) were isolated from CSF, and the rest (81.0%) from other sterile body fluids, mostly in blood. Antimicrobial susceptibilities were interpreted according to meningitis and non-meningitis break points. PRSP and PSSP from CSF accounted for 70.2 and 29.8% of the isolates, respectively. PISP and PSSP from non-CSF accounted for 4.5 and 95.5% of the isolates, respectively. Although the proportion of PRSP derived from CSF was significantly higher than that from non-CSF sources (P < 0.0001), the proportion of CSF derived isolates with PEN MIC ≥ 1µg/ml (59.6%) was similar to that from non-CSF sources (54.3%) (P = 0.5644) (Figure 1). Analysis of PBPs active sites of strains from CSF showed that all PSSP, and one strain of PRSP, with PEN MIC of 0.25 µg/ml, had no substitution in the PBPs active sites. PBP2x and PBP2b substitutions were common in low-level PRSP strains (MIC = 0.5µg/ml), with or without PBP1a substitution. The strains with PEN MIC ≥ 1 µg/ml all had substitutions in the PBP1a active site or nearby sites, accompanied by substitutions in the PBP2x and PBP2b regions ( Table 5).
Analysis of PBPs active sites or nearby sites of strains derived from non-CSF sources revealed that all PSSP with PEN MIC ≤ 0.25 µg/ml had no substitutions in PBPs active sites or nearby sites, except one serotype 6A strain, in which a T446A substitution nearby the active site of PBP2b was detected. Various substitutions in the PBPs active sites or nearby sites of PSSP strains with PEN MIC ≥ 0.25 µg/ml were detected. Similar to CSF derived strains, the substitutions in active sites or nearby sites of PBP2x and PBP2b were common in strains with low PEN MIC level, with or without PBP1a active site substitution. In contrast, strains with PEN MIC ≥ 1 µg/ml all had substitutions in PBP1a active sites or nearby sites, accompanied by PBP2x or PBP2b substitutions ( Table 6).

Variations of the pbp2x Gene
Compared with the reference strain R6 (GenBank accession No. NC003098), a total of 338 amino acids from positions 259 to 596 of the PBP2x protein were analyzed. A total of 98 different substitutions at 73 amino acid positions were detected in the PBP2x sequence, among which D567N (Asp → Asn) was the most common substitution, accounting for 69.3% (208/300), followed by D488N (Asp → Asn) and S576N (Ser → Asn), each accounting for 64.0% (192/300). Strains with L565S (Leu → Ser) and R384G (Arg → Gly) substitutions each accounted for 62.7% (188/300) and 61.3% (184/300), respectively. According to the distribution of various substitutions in PBP2x, all strains could be divided into 43 groups. The number of substitution sites in each group ranged from 1 to 42, accounting for 0-12.4% of the total number of amino acids analyzed. Group 2X37 was the commonest, accounting for 16.3% of all strains (49/300). There were 38 substitutions detected in all strains within the group, and the substitution rate was 11.2%. The second common group was 2X01, accounting for 13.3% of all strains (40/300), and the PBP2x sequence of strains in this group was exactly the same as that of R6 strain. There were 33 strains in the 2X41 group, accounting for 11.0% of the strains. Forty substitutions were detected in all strains of this group and the substitution rate was 11.8% (Table 7). Compared with PBP2b and PBP1a sequences, PBP2x had the least number of total substitutions (98 vs. 112 vs. 105). According to the distribution of PEN MIC, the number of detectable substitution sites in strains with PEN MIC ≤ 0.25 µg/ml ranged from 10-32, while for strains with PEN MIC = 0.5, 1, 2, and 4 µg/ml the number of detectable substitutions were 86, 46, 57, and 48, respectively (Supplementary Table 2).

Variations of the pbp2b Gene
Compared with the reference strain R6 (GenBank accession No. NC003098), a total of 372 amino acids from positions 305 to    676 of the PBP2b protein were analyzed for the isolates. A total of 111 different substitutions and two insertion variations at 88 amino acid positions were detected in the PBP2b sequence. The two insertion variations were only found in two strains, both located between amino acids 424-425, and the insertion sequences were YIW (Tyr-Ile-Try) and YTW (Tyr-Thr-Try). The most common substitution was E476G (Glu → Gly), accounting for 64.7% (194/300) of the isolates, followed by T446A (Thr → Ala), Q438E (Gln → Glu), L455I (Leu → Ile), which accounted for 63.3% (190/300), 61.7% (185/300) and 61.0% (183/300), respectively. According to the distribution of various substitutions in PBP2b, all strains could be divided into 46 groups. Except for group 2B45 and 2B46, each containing one insertion variation, the other groups all had amino acid substitutions in PBP2b protein. The number of substituted amino acid sites ranged from 0 to 45, accounting for 0-12.1% of the total number of amino acids (n = 372) analyzed. Group 2B01 was the most common, accounting for 17.3% of the strains (52/300), and had exactly the same sequence of PBP2b as reference strain R6. The second common group was 2B20, accounting for 14.3% of the strains (43/300). A total of 13 amino acid substitutions were detected in the PBP2b sequence of the strains in this group, and the substitution rate was 3.5% (Table 8). According to the distribution of PEN MICs, the number of substitution sites in strains with MIC ≤ 0.12µg/ml were less than ten, while the number of detectable substitution sites in strains with MIC = 0.25, 0.5, 1, 2, and 4µg/ml were 12, 75, 61, 52, and 54, respectively (Supplementary Table 3).

Variations of the pbp1a Gene
Compared with the reference strain R6 (GenBank accession No. NC003098), a total of 344 amino acids from positions 312 to 655 of the PBP1a protein were analyzed for the studied isolates. A total of 105 different substitutions at 85 amino acid positions were detected in PBP1a. The E388D substitutions (Glu → Asp) was detected in all the 300 isolates. Apart from this, the most common substitution was S540T (Ser → Thr), accounting for 66.7% (200/300), followed by N546G (Asn → Gly), A550P (Gly → Pro), T574N (Thr → Asn), S575T (Ser → Thr), Q576G (Gln → Gly), F577Y (Phe → Tyr), N609D (Asn → Asp), each being detected in 60.0% (180/300) of the strains. According to the distribution of various substitutions in PBP1a, all strains could be divided into 25 groups. The number of substituted amino acids in each group ranged from 1 to 55, accounting for 0.3 −16.0% of the total number of amino acids analyzed. Based on the above substitution classification, group 1A15 was the most common, accounting for 28.7% (86/300) of the strains. All strains in this group had 43 amino acid substitution sites, with the substitution rate of 12.5% of the total number of amino acids analyzed, followed by group 1A04, accounting for 14.7% (44/300) of all the strains ( Table 9). Based on the distribution of PEN MIC, the number of substitution sites in strains with MIC ≤ 0.12 µg/ml were less than ten. The number of substitution sites in strains with MIC = 0.25, 0.5, 1, 2, and 4 µg/ml were 17, 97, 88, 71, and 68, respectively (Supplementary Table 4).

DISCUSSION
In this study, 300 S. pneumoniae strains isolated from IPD were analyzed for serotype distribution, antimicrobial susceptibility and PBPs substitutions. To our knowledge, this is the largest multicenter study involving invasive S. pneumoniae strains in China, in terms of number and diversity of origin of the isolates studied. Based on the Quellung reaction, 40 different serotypes were detected amongst 299 typeable S. pneumoniae strains, with the five most common serotypes being 23F, 19A, 19F, 3, and 14. The serotype distribution in the current study is consistent with those from other studies in China, although the isolation rates varied (Zhao et al., 2013(Zhao et al., , 2020. All strains were susceptible to ETP, LEV, LZD, and VA, which is consistent with most studies in our region and abroad (Zhanel et al., 2006;Zhao et al., 2013), (Zhou et al., 2016;Cai et al., 2018;Fu et al., 2019;Golden et al., 2019;Shi et al., 2019). AMC was the second most active antimicrobial drug with a non-susceptibility rate of 2.3%, which is significantly lower than that reported in the national data for 2006 -2008 in which the non-susceptibility rate was 5.3%, and the one from Beijing for the period 2012 and 2017 (10.2%) (Zhanel et al., 2006;Shi et al., 2019). A meta-analysis of invasive S. pneumoniae strains from Chinese children (2000 to 2016) also showed a very high proportion of AMC resistance among the isolates at 16.1%, which may be due to the higher proportion of PCV7-related serotype strains in these studies of 60.8%, compared to 42.3% in the present study (Fu et al., 2019). It has been previously reported that invasive S. pneumoniae strains of PCV7-related serotypes are more resistant to AMC than non-PCV7 serotype strains (15.7% vs. 1.7, P = 0.013) (Liu et al., 2010).
PRSP accounted for 70.2% of CSF derived strains in this study, which is lower than in previous studies based on invasive S. pneumoniae strains from children only, ranging from 76.6 to 95.7%, but higher than that reported in isolates from IPD cases in both adults and children, at 51.5% (Xue et al., 2010;Zhou et al., 2016;Shi et al., 2019). No PRSP was detected in strains from non-CSF specimens, and PISP accounted for 4.5% of the isolates, which was higher than that reported in IPD isolates from both adults and children at 3.8% (Zhao et al., 2013), but significantly lower than reported in children isolates from Shenyang (2010 to 2014), in which the resistance rate was very high at 32.3% (Fu et al., 2019). According to the oral breakpoint interpretation for all strains studied, PISP and PRSP accounted for 23.0% and 44.7% of the isolates, respectively, which were higher than reported in the comprehensive meta-analysis study of invasive S. pneumoniae strains from children (oral break points PISP: 42.6%, PRSP: 32%) (Fu et al., 2019). Data from the SENTRY program of the global multi-center surveillance study (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) showed that the Asia-Pacific region had the lowest S. pneumoniae penicillin susceptibility rate (52.4%, oral breakpoint) but the highest multi-drug (49.8%) and pan-drug resistance (17.3%) rates, compared with North America, Europe, and Latin America. However, the SENTRY program did not include data from mainland China (Sader et al., 2019). Previous data based on 881 S. pneumoniae isolates from 23 teaching hospitals across China from 2011 to 2016, showed that PRSP accounted for 51.6% of the isolates (oral breakpoint), a rate which is slightly higher than in the present study, but only 11.6% of the isolates were considered invasive in that previous study .
The resistance mechanism of S. pneumoniae to penicillin and other BLAs is mainly affected through the modification of PBPs (Hakenbeck et al., 2012). There are 6 different PBP types described in S. pneumoniae, including PBP1a, PBP1b, PBP2a, PBP2b, PBP2x and PBP3. They can be divided into three categories according to their molecular weights and functions; type A: high molecular weight PBP containing PBP1a, PBP1b and PBP2a with transglycosidase activity and transpeptidase activity; type B: high molecular weight PBPs containing PBP2b and PBP2x with transpeptidase activity; type C: low molecular weight PBPs containing PBP3 with carboxypeptidase activity Hakenbeck et al., 2012).
Three types of PBPs play a major role in BLAs resistance, namely PBP2x, PBP2b, and PBP1a . Normally, BLAs can combine with serine at the PBPs active site to form serine esters, thereby inhibiting the synthesis of bacterial cell walls, and leading to bacterial death. When there is PBPs substitution, the reactivity for BLAs drugs is reduced resulting in limited effectiveness of the drugs and eventual development of drug resistance . Studies have shown that substitutions in the active sites of PBP2x and PBP2b, and their adjacent sites only cause low levels of BLAs resistance (Laible et al., 1991;Grebe and Hakenbeck, 1996;Sifaoui et al., 1996). However, combination substitutions involving PBP1a and those of PBP2b and PBP2x, can cause high levels of BLAs resistance, but PBP1a substitution alone cannot cause an increase in BLAs resistance levels (Zerfass et al., 2009). Substitutions in the PBP2b active sites mainly cause bacteria to be resistant to penicillin, but have nothing to do with cephalosporin resistance. On the contrary, substitutions in the PBP2x active sites are mainly related to cephalosporin resistance, and can result in high-level resistance to third-generation cephalosporin when combined with PBP1a substitutions Hakenbeck et al., 2012).
Although many studies have been performed on PBPs, both domestically in Asia and abroad, there are limited studies on S. pneumoniae isolates from IPD in mainland China. Our study mainly focused on the gene substitutions of pbp2x, pbp2b, and pbp1a among 300 invasive S. pneumoniae strains and their relationship with PEN MIC and serotype. In this study, the variations in the PBPs gene active sites were highly diverse, and the number of substitutions were higher than those in previous studies (Zhou et al., 2016;Chu et al., 2018).
Among the invasive S. pneumoniae isolates studied, strains with PEN MIC ≤ 0.25 µg/ml had no PBPs active site substitutions, except for one strain of serotype 6A (MIC = 0.25 µg/ml). On the other hand, strains with PEN MIC ≥ 0.5 µg/ml all had PBPs active site substitutions, while strains with high PEN MIC levels (MIC ≥ 1 µg/ml) all had PBP1a substitutions, which were accompanied by PBP2x and PBP2b substitutions. Even if the antimicrobial susceptibility phenotype of invasive S. pneumoniae derived from non-CSF was interpreted as susceptible by the meningitis breakpoint, the strains still had substitutions in different PBPs active sites. Thus the PBPs substitution of the strains was mainly related to PEN MICs, and less influenced by specimen type and antibiotic breakpoints.
We analyzed a total of 338 amino acids from positions 259-596 of PBP2x, and 98 different substitutions at 73 amino acid positions were detected. According to the distribution of PEN MIC, the number of substitution sites of strains with PEN MIC ≤ 0.12 µg/ml were more than that of PBP1a and PBP2b, ranging from 10-32. With the increase in PEN MIC, the number of substitution sites increased also, and the strains with MIC = 0.5 µg/ml had the maximum number of substitutions. In the study of Zhou et al., the average number of substitutions in PSSP strains was much higher for PBP2x than for PBP2b (29.26 vs. 8.22). Interestingly, they also found a significantly higher number of substitution sites in PRSP (65.55 ± 2.93) and PISP (63.37 ± 2.51) compared to PSSP (29.26 ± 27.88) (Zhou et al., 2016). Thus, it could be presumed that the MIC value of PEN was associated with the number of substitution sites in PBP2x. Specifically, common substitutions in PBP2x active sites included T338A, M339F, H394L, and L546V, which have also been reported in previous studies (Smith and Klugman, 1995;Zhou et al., 2016;Cai et al., 2018). In the present study, 181 strains had the T338A substitution and 39 strains also contained the M339F substitution. The PEN MIC of most strains was 2-4 µg/ml. The M339F substitution was first discovered in a highly resistant clinical strain in France, and then successively detected in the USA and Japan (Coffey et al., 1995;Asahi et al., 1999;Nagai et al., 2002). Previous crystal structure studies showed that the active center 337STMK of PBP2x which had the T338A and M339F double-site substitution was deformed. Furthermore, the serine active site Ser337 also presented another conformational change, resulting in reduction in the acylation efficiency of penicillin and cefotaxime by more than 20 times (Lu et al., 2001;Chesnel et al., 2003).
T338A and M339F substitutions have also been described in research reports from Taiwan, Japan, South Africa and other places, and were related to resistance in penicillin, amoxicillin and third-generation cephalosporins (Smith and Klugman, 1995;Du Plessis et al., 2002;Sanbongi et al., 2004;Davies et al., 2012;Liu et al., 2016;Zhou et al., 2016). Previous studies have shown that the L546V substitution is associated with high level resistance in BLAs, especially cephalosporins, but this substitution has also been reported in PSSP and cephalosporinsusceptible strains, suggesting that a single L546V substitution is not sufficient enough to cause high level resistance of BLAs (Nichol et al., 2002;Davies et al., 2012;Liu et al., 2016). Both crystal structure studies in vitro and transformation experiments in vivo have confirmed that R384G could affect the susceptibility of bacterial strains to penicillin and cefotaxime (Smith and Klugman, 2005;Maurer et al., 2008;Liu et al., 2016). A crystal structure study of PBP2x has found that the amino acid substitutions at position 371 and 384 affected the mobility of loop between amino acids 365-394 and were important for BLAs resistance (Carapito et al., 2006). In our study, strains with PBP2b sequences belonging to Group 2X27−2X43 all harbor the T338A substitution along with some other relevant substitutions, such as I371T, R384G, and M400T, but it is unknown which substitutions contribute to the resistance of the strains. Further experiments are needed to figure out the inner association between resistance and various substitutions. The Q552E substitution was located near the third catalytic motif 547KTG549 at the end of strand β3 loop, thus substitutions can led to an increase of the acylation efficiency for BLAs (Pernot et al., 2004;Zapun et al., 2008). In previous studies, Q552E was widely reported while in this study Q552V substitution was frequently detected in strains with PEN MIC ≤ 0.5 µg/ml. D567N substitution was previously detected in S. pneumoniae strains with high penicillin resistance in Taiwan, but has also been described in PSSP, which is consistent with the findings of this study .
A total of 372 amino acids from positions 305-676 of PBP2b were analyzed, and a total of 111 substitutions and 2 insertion variations at 88 amino acid sites were detected. Similar to PBP1a, the number of substitution sites in strains with PEN MIC ≤ 0.12 µg/ml was no more than 10. Furthermore, the number of substitution sites increased with MIC increase, but the most substitution sites were detected in strains with PEN MIC = 0.5 µg/ml. The two insertion variations YIW and YTW were found in two strains, located between 386SVVK and 443SSN in the active site of PBP2b, which have never been reported elsewhere. Previously, Japanese researchers reported that there was a duplication of the amino acid sequence WYT at the positions 429-431 between the amino acid positions 431-432 of PBP2b (Yamane et al., 1996). The insertion sequence was detected in 13 strains, of which 10 were of serotype 19, and three were serotype 6. In that study, the MIC of PEN and cefotaxime of S. pneumoniae strains carrying the above insertion variation (WYT amino acid sequence) were 0.125-2 µg/ml and 0.063-1 µg/ml, respectively (Yamane et al., 1996). In the present study, the two strains were of serotype 6B and NT and the PEN MIC were 0.5 µg/ml and 1 µg/ml respectively.
Previous studies have shown that the insertion of amino acid sequence between the conserved sequences of PBP2b in Neisseria gonorrhoeae reduced its affinity to PEN, further studies are needed to confirm its function in S. pneumoniae (Brannigan et al., 1990). Substitutions in the PBP2b active sites included T446A, T446S, and A619G, which have been reported in previous studies (Yamane et al., 1996;Sanbongi et al., 2004;Granger et al., 2005;Izdebski et al., 2008;Tian et al., 2008).
In vitro experiments have shown that the affinity of T446A substitution strain to penicillin was 60% lower than that of the wild strain (Pagliero et al., 2004). However, researchers in the United States (USA), South Korea and Canada also found this substitution site in PSSP strains (Nagai et al., 2002;Baek et al., 2004;Granger et al., 2006), a finding which was also confirmed in one serotype 6A PSSP strain derived from blood (MIC = 0.25 µg/ml) in the present study. The A619G substitution has also been reported in strains in United States, Spain, Mexico, and other regions, and has been shown to be associated with high level amoxicillin resistance (Kosowska et al., 2004;Cafini et al., 2006). Other common substitutions identified in this study included E476G, Q438E, and L455I. Among them, Q438E and E476G have been reported in studies in Taiwan, France and Japan, and these substitutions were related increased levels of bacterial resistance to BLAs (Sanbongi et al., 2004;Chesnel et al., 2005;Liu et al., 2016). To the best of our knowledge, the L455I substitution is reported for the first time in this study.
A total of 344 amino acids from positions 312-655 of PBP1a were analyzed, and 105 different substitutions at 85 amino acid positions were detected. Based on PEN MIC distribution, isolates with MIC ≤ 0.12 µg/ml had no more than 10 substitution sites, and the substitution sites increased with increasing MIC. However, the maximum number of substitution sites was found in strains with MIC = 0.5 µg/ml. Substitutions in the PBP1a active sites, including T371A, T371S, and P432T, have been reported in multinational studies, and are related to high level resistance in penicillin and cephalosporin (Smith and Klugman, 1998;Liu et al., 2016;Zhou et al., 2016;Chu et al., 2018). In vitro site-directed mutagenesis experiments showed that T371A substitution reduces the acylation efficiency of PBP1a to cefotaxime and penicillin by 2.4 and 26 times, respectively, and TSQF574-577NTGY substitution reduces the acylation efficiency of PBP1a to cefotaxime and penicillin by 5.5 and 49 times, respectively (Job et al., 2008). In vivo transformation studies showed that only when the two substitutions exist at the same time, the resistance level of substitution strains to BLAs would increase significantly (Smith and Klugman, 1998;Job et al., 2008). In addition, the sites with high substitution rates in this study included S540T, N546G, A550P, T574N, S575T, Q576G, F577Y, and N609D, and part of these sites have been reported in some strains in Shenyang (Zhou et al., 2016).
Classification of isolates based on combination of substitution patterns in the three PBPs revealed a high degree of diversity among the isolates. Strains with the same serotype and PEN MIC exhibited different PBPs substitution combinations due to differences in the three PBPs substitution sites. For example, most strains of serotype 3 had a PEN MIC ≤ 0.015 µg/ml, but there were as many as 13 corresponding PBPs substitution combinations, although the difference between the different PBPs sites was small. Likewise, isolates with similar PBPs substitution combinations did not necessarily have the same PEN MIC levels, and this was reflected in serotypes 14, 19A, 19F, and 23F. The substitution combinations P100, P072, P070, and P085 were detected in isolates with PEN MICs between 2-4 µg/ml. A similar finding has also been reported previously, where five strains of serotype 23F with exactly the same PBPs substitutions and murM had PEN MICs ranging between 0.25-2 µg/ml (Chesnel et al., 2005).
In summary, we investigated the variations in pbp2x, pbp2b, and pbp1a genes, and serotype distribution of IPD S. pneumoniae isolates collected between 2010 and 2015 in China. We analyzed the serotype distribution, resistance to PEN and PBPs substitutions amongst these strains. There was a great diversity detected in PBPs substitutions patterns among the strains, suggesting that the PEN MIC level of S. pneumoniae may be affected by several other factors. Therefore, a comprehensive understanding of antibiotic resistance mechanism of S. pneumoniae needs to be further examined at the genomic level.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.