SGM Journals
DIAGNOSTICS, TYPING AND IDENTIFICATION

Phenotypes and genotypes of macrolide-resistant Streptococcus pyogenes isolated in Seoul, Korea

Sook Young Bae, Jang Su Kim, Jung-Ah Kwon, Soo-Young Yoon, Chae Seung Lim, Kap No Lee, Yunjung Cho, Young Kee Kim, Chang Kyu Lee

  • Department of Laboratory Medicine, Korea University College of Medicine, Seoul, Republic of Korea
  • Correspondence
    Chang Kyu Lee
    cklee{at}korea.ac.kr

    • Journal of Medical Microbiology 2007; 56(2):229–235 · https://doi.org/10.1099/jmm.0.46825-0

    View at publisher PubMed

    Abstract

    The mechanisms of resistance to macrolides in 51 erythromycin-resistant clinical isolates of Streptococcus pyogenes collected from 1997 through 2003 in Seoul, Korea were evaluated. They were characterized by their antimicrobial susceptibility, phenotype (using triple-disc and induction tests), resistance genotype, emm genotyping (M typing) and phylogenetic analysis. Erythromycin resistance was observed in 23 % of isolates. Inducible phenotype was the most common (iMLS, 51 %, 26 strains), followed by the constitutive phenotype (cMLS, 31 %, 16 strains) and the M phenotype (18 %, 9 strains). Eight of twenty-six iMLS isolates exhibited the iMLS-C phenotype. The remaining 18 isolates gave small inhibition zones (<12 mm) around all three discs, and mild blunting of the spiramycin and clindamycin zones of inhibition proximal to the erythromycin disc. They showed remarkable inducibility in erythromycin and clindamycin resistance. The MIC90 of erythromycin and clindamycin rose from 8 to >128 μg ml−1 and from 0.5 to >128 μg ml−1, respectively. Their resistance characteristics did not fit into any known iMLS subtype reported so far in the literature. So, it was named as an iMLS-D, new subtype. All of these iMLS-D strains harboured the erm(B) gene, demonstrated the emm12 genotype, except one, and formed a tight cluster in a phylogenetic tree, with 89.2 to 100 % sequence homology, suggesting that they are closely related. Nine of sixteen cMLS strains had the emm28 genotype, which had been reported to be associated with multiple drug resistance.

    INTRODUCTION

    Streptococcus pyogenes causes a wide array of infections in humans (Bisno, 1991). Macrolides have been effective in the treatment of individuals with S. pyogenes infection and a good alternative in patients who are allergic to other agents (e.g. penicillin). Unfortunately, because of the general use of these agents, macrolide-resistant S. pyogenes have been isolated in many countries. Resistance is caused by two different mechanisms: target-site modification (Weisblum, 1995) and active drug efflux (Sutcliffe et al., 1996). In these organisms, target-site modification is mediated by a methylase enzyme that reduces the binding for macrolides, lincosamides and streptogramin B (MLS) antibiotics with their target site in the bacterial ribosome. MLS resistance can be expressed either constitutively (cMLS phenotype) or inducibly (iMLS phenotype) in Gram-positive cocci. Three subtypes of the iMLS macrolide resistance phenotype have been distinguished: iMLS-A, iMLS-B and iMLS-C (Giovanetti et al., 2002). Active efflux is associated with the M phenotype. The emm gene of S. pyogenes encodes the cell-surface M virulence protein that is putatively responsible for at least 100 known M serospecificities of S. pyogenes (Beall et al., 1998, 2000; Jones & Fischetti, 1988; Whatmore et al., 1994). A new typing system, which is based on a sequence analysis of the portion of the emm gene, which encodes M serospecificity, was introduced to overcome many problems with the traditional M protein serotyping system, such as limited availability of M typing sera, newly encountered M types and difficulty in interpretation. Several studies have shown that increased macrolide consumption is associated with the occurrence of macrolide resistance (Granizo et al., 2000; Seppala et al., 1992). Various resistance rates have been reported up to approximately 50 %, depending on the country and geographical region (Beall et al., 1998; Bisno, 1991; De Azavedo et al., 1999; Sungho et al., 2001; Giovanetti et al., 2002). However, data on macrolide resistance of S. pyogenes in Korea are relatively scarce. The prevalence of erythromycin resistance in Korea increased rapidly from only 2.0 % in 1994 to 44.8 % in 2002 (Sungho et al., 2001; Kim & Lee, 2004). The aim of this study was to determine the antimicrobial resistance, phenotypic and genetic characteristics of erythromycin-resistant S. pyogenes in Seoul, Korea.

    METHODS

    Bacterial strains.

    A total of 51 erythromycin-resistant isolates were evaluated among 222 S. pyogenes isolates from clinical specimens from January 1997 to December 2003 in Seoul, Korea. The specimens were collected from two university hospitals and were from the culture of 26 throat swabs, 16 pus samples, 5 sputa samples and 4 blood samples. Of these, 25 isolates were recovered from children (1–16 years old, median age 5 years) and 26 isolates from adults. Multiple isolates from the same patient were avoided. Isolates were identified as S. pyogenes by colony morphology, β-haemolysis on sheep blood agar, and Lancefield grouping by using a commercially available agglutination technique (Seroiden Strepto kit) and bacitracin disc test (0.04 U, Taxo A, BBL Microbiology system). The strains were maintained in glycerol at −70 °C until all of the isolates were collected and then subcultured twice on blood agar before undergoing susceptibility testing.

    Antimicrobial susceptibility testing.

    Penicillin, erythromycin, spiramycin, clindamycin, tetracycline, chloramphenicol, azithromycin and clarithromycin were obtained in the form of standard reference powders of known potency from Sigma or from their respective manufacturers. The antibiotics were incorporated into the medium (Mueller–Hinton agar medium with 5 % sheep blood) in a log2 dilution series of 0.03 to 128 μg ml−1. Inocula were prepared by diluting bacterial suspensions equivalent in turbidity to a McFarland 0.5 standard, resulting in 104 c.f.u. in each spot when applied by a Steer's replicator (Craft Machine). The plates were incubated overnight at 35 °C in an atmosphere containing 5 % carbon dioxide. The interpretative categories for each antibiotic were those recommended by Clinical and Laboratory Standards Institute (CLSI) (National Committee for Clinical Laboratory Standards, 2005). MIC breakpoints suggested by the CLSI were used for erythromycin, clarithromycin and clindamycin (susceptible at ⩽0.25 μg ml−1, resistant at ⩾1 μg ml−1), azithromycin (susceptible at ⩽0.5 μg ml−1, resistant at ⩾2 μg ml−1), tetracycline (susceptible at ⩽2 μg ml−1, resistant at ⩾8 μg ml−1), chloramphenicol (susceptible at ⩽4 μg ml−1, resistant at ⩾16 μg ml−1) and penicillin (susceptible at ⩽0.12 μg ml−1). Given the lack of CLSI recommendations for spiramycin, we used the Comité de l'Antibiogramme de la Société Française de Microbiologie recommendation of a susceptible MIC of ⩽1 μg ml−1 and a resistant MIC of ⩾4 μg ml−1 (Soussy et al., 2000). Streptococcus pneumoniae ATCC 49619 was used for quality control.

    Triple-disc test for the phenotypes of macrolide resistance.

    A triple-disc test was set up by adding spiramycin (a 16-ring macrolide) to the erythromycin and clindamycin double-discs, thereby enabling us to distinguish the subtypes of inducible-resistance phenotypes. Commercial discs (Oxoid) of erythromycin (15 μg), clindamycin (2 μg) and spiramycin (100 μg) were used. Three discs were placed 15 to 20 mm apart from each other in a triangular formation. Three different patterns (clindamycin-susceptible, erythromycin-resistant; clindamycin-inducible, erythromycin-resistant; clindamycin-resistant, erythromycin-resistant) were recognized as being related to the well-recognized phenotypes (M, inducible, constitutive) of MLS resistance. Subtypes of the inducible macrolide-resistance phenotype were determined as described by Giovanetti et al. (1999).

    Induction of MLS resistance.

    Induction of MLS resistance was evaluated under pre-growth conditions (3 h at 35 °C) in erythromycin at a subinhibitory concentration (0.05 μg ml−1), as described by Giovanetti et al. (1999). The culture was then washed, and the cells were used to prepare the inoculum for MIC testing using the same agar dilution method. The post-induction MICs for those antibiotics were evaluated.

    Detection of erythromycin-resistance gene.

    The erm(A), erm(B) and mef(A) genes were detected by PCR, using previously published primers (Sutcliffe et al., 1996; Claire et al., 2003). DNA preparation, amplification and detection, as well as visualization of the product, were carried out using established procedures (Sutcliffe et al., 1996). The expected sizes were 459 bp for erm(A), 638 bp for erm(B) and 348 bp for mef(A); their specificity was confirmed with an appropriate restriction enzyme (XmnI, XmnI and PvuII, respectively).

    emm genotyping (M typing) and phylogenetic analysis.

    For emm genotyping, the 5′ terminal ends of emm gene were amplified with M1 (5′-ATAAGGAGCATAAAAATGGCT-3′) and M2 (5′-AGCTTAATTTTCTTCTTTGCG-3′) (Saunders et al., 1997) and were partially sequenced (129 bp). The emm sequences analysis of all erythromycin-resistant S. pyogenes was performed with BioEdit v5.0.9 (Hall, 1999). Multiple sequence alignment was conducted with culstal x, version 1.81 (Thompson et al., 1997), with a gap-opening penalty of 15 and a gap extension penalty of 6.66. The phylogenetic tree was prepared in mega 3 (Kumar et al., 2004) using the Kimura two-parameter neighbour-joining method. Nodal confidence values indicate the results of bootstrap resampling (n=1000).

    RESULTS AND DISCUSSION

    In this study, a total of 222 S. pyogenes isolates from clinical specimens from patients referred to the university hospitals in Seoul, Korea over 6-year period, from 1997 through 2003, were included. The resistance rates of erythromycin during the study period were 20.0 % in 1997, 27.8 % in 1998, 24.3 % in 1999, 17.9 % in 2000, 27.3 % in 2001, 22.2 % in 2002 and 21.5 % in 2003. Table 1⇓ shows the results of susceptibility testing of 51 erythromycin-resistant isolates in vitro, as well as the MIC ranges, and the calculated MIC50 and MIC90 values of antibiotics tested. Erythromycin-resistant strains were resistant to the other macrolides, clarithromycin and azithromycin, as was reported elsewhere (Hsueh et al., 2002; Kim & Lee, 2004). The azithromycin MIC50 was higher than that of clarithromycin, because isolates with the iMLS-D phenotype had high azithromycin MICs (>128 μg ml−1). Fourteen isolates with a high level of erythromycin resistance (MIC >128 μg ml−1) were all resistant to clindamycin and spiramycin.

    Table 1.

    In vitro susceptibilities to 9 antimicrobial agents of 51 erythromycin-resistant isolates of S. pyogenes

    The patterns of susceptibility to MLS antibiotics according to the phenotypes of macrolide resistance are presented in Table 2⇓. Out of a total of 51 erythromycin-resistant strains, 26 (51 %) had the iMLS phenotype, 16 (31 %) had the cMLS phenotype and 9 (18 %) expressed the M phenotype. These features differed from those observed in European countries and Canada, where the M phenotype was most common (De Azavedo et al., 1999; Giovanetti et al., 1999; Grivea et al., 2006; Kataja et al., 2000). Among the inducible phenotypes, neither iMLS-A nor iMLS-B was found, but the iMLS-C and iMLS-D phenotypes were identified on the basis of the diameter of the inhibition zone. The iMLS-C phenotype had the same pattern as that described in another report (Giovanetti et al., 1999), with blunted spiramycin and clindamycin zones of inhibition proximal to the erythromycin disc, and a visible zone of inhibition around the erythromycin disc. All except one had the erm(A) resistance gene. The iMLS-D phenotype had smaller inhibition zones (<12 mm) around each disc and a lesser degree of blunting in the spiramycin and clindamycin discs proximal to the erythromycin disc than iMLS-C (Fig. 1⇓). The ability to induce MLS resistance was confirmed by the increase in MIC after incubation with erythromycin. The MICs of erythromycin before induction were in range of 1–8 μg ml−1, which was compatible with known iMLS-C, and were >128 μg ml−1 after induction, which was compatible with known iMLS-A or iMLS-B. This remarkable inducibility of erythromycin resistance was characteristic of iMLS-D. The MICs of clindamycin and spiramycin were closer to those for known iMLS-A isolates while the MICs of clarithromycin were closer to those for known iMLS-C isolates (Giovanetti et al., 1999). All of these iMLS-D strains had the erm(B) gene. These strains did not fit into any known subtype reported in the literature so far.

    Figure image not available in archive
    Fig. 1.

    Phenotypes of erythromycin-resistant S. pyogenes isolates determined by the triple-disc test: (A) cMLS phenotype, (B) iMLS phenotype (subtype iMLS-C), (C) iMLS phenotype (subtype iMLS-D), (D) M phenotype. Clockwise from left on each plate are the erythromycin disc, the spiramycin disc and the clindamycin disc.

    Table 2.

    MIC (μg ml−1) ranges of 48 isolates of among 51 erythromycin-resistant S. pyogenes to MLS antibiotics according to phenotypes of macrolide resistance

    Conventionally, the inducible clindamycin resistance has been detected using a double-disc test with erythromycin and clindamycin discs. The triple-disc test was set up by adding a spiramycin disc to the double-disc test and it was useful to classify heterogeneous MLS resistance phenotypes, especially when the discs were arranged in triangular formation. The inducible nature of iMLS-D phenotype was distinct by the triple-disc test while it was not by the double-disc test.

    The prevalence of the recognized phenotypes of erythromycin-resistant S. pyogenes seemed to vary considerably from area to area. Empirical erythromycin administration to patients with infection by the iMLS-D phenotype S. pyogenes might result in the induction of a high level of erythromycin and clindamycin resistance and, consequently, lead to treatment failure of erythromycin-resistant and clindamycin-susceptible isolates. There was a case report describing clindamycin treatment failure in a patient with meticillin-resistant, erythromycin-resistant and clindamycin-susceptible Staphylococcus aureus (Siberry et al., 2003). More specific information about resistance phenotype using the triple-disc test could help physicians choose appropriate antibiotics for treating infections by S. pyogenes.

    Most of the cMLS phenotypes (13 out of 16) showed a high degree of resistance (MIC90 >128 μg ml−1) to MLS antibiotics (erythromycin, clarithromycin, azithromycin, spiramycin and clindamycin). They had no inhibition zone around the triple discs with the erm(B) gene. There were three isolates showing atypical subtypes of cMLS strains, having a small inhibition zone around the erythromycin and spiramycin discs and no inhibition zone around the clindamycin disc. Their MIC50 and MIC90 of erythromycin were 32 and >128 μg ml−1, respectively, while the MIC50 and MIC90 of clarithromycin, azithromycin, spiramycin and clindamycin were all >128 μg ml−1. Their resistance genes were erm(A), erm(B) and erm(A) plus erm(B), respectively.

    Nine strains of the M phenotype were all susceptible to tetracycline, spiramycin and clindamycin. The MIC90 values of erythromycin and clarithromycin were higher than those for the isolates with the iMLS phenotype (8–16 and 8–32 μg ml−1 vs 1–8 and 1–8 μg ml−1). Macrolide resistance induction was not observed in any of the M phenotype strains, all of which had the mef(A) gene (Table 3⇓).

    Table 3.

    Distribution of erythromycin-resistant S. pyogenes isolates according to their phenotype, genotype and emm genotype

    Two isolates had two resistance genes at the same time – erm(A) and erm(B). However, their phenotypic characteristics – e.g. their MICs or inducibility – were similar to those of other isolates within each group.

    In this study, emm genotyping was used instead of M protein serotyping for epidemiological purposes. PFGE and multilocus sequence typing were not available. The hypervariable portion (129 bp) of the 5′ terminal end of the emm gene in all erythromycin-resistant strains (except one) was sequenced and the phylogenetic tree was analysed. The emm12, 28 and 2 genotypes were the most common, comprising 39, 18 and 10 %, respectively, of all genotypes. Seventeen out of twenty emm12 strains had the iMLS-D phenotype and erm(B) gene. There were some reports of an association between the emm type and the pattern of antibiotic resistance (Alberti et al., 2003; Grivea et al., 2006). The emm28 genotype has been reported to be associated with multiple drug resistance (Mihaila-Amrouche et al., 2004; Perez-Trallero et al., 1999). Our nine emm28 genotype strains had the cMLS phenotype with multiple resistances, which is consistent with the findings in previous reports. All of the emm2 genotype strains expressed the iMLS-C phenotype (Table 3⇑).

    The cMLS, M and iMLS-C phenotype strains were clustered relatively loosely in the phylogenetic tree based on emm sequences (Fig. 2⇓), the sequence homologies were 58.9–100 %, 56.1–99.2 % and 76.1–98.4 %, respectively. The iMLS-D phenotype strains formed tighter clusters than did strains with other phenotypes, with a nucleotide sequence identity of 89.2–100 %. These findings suggest that they were genetically closely related.

    Figure image not available in archive
    Fig. 2.

    Phylogenetic tree of emm sequences of 50 erythromycin-resistant S. pyogenes isolates using mega 3 with the Kimura two-parameter neighbour-joining method. Nodal confidence values indicate the results of bootstrap resampling (n=1000). Strain number, phenotype, emm genotype and year of isolation are presented. aCMLS, atypical cMLS phenotype.

    Acknowledgments

    We are indebted to Jin Heo, Yoon Ho Kim and the staff at the Clinical Microbiology Laboratory of KUMC Guro Hospital for their expert technical assistance.

    References

    1. Alberti, S., Garcia-Rey, C., Dominguez, M. A., Aguilar, L., Cercenado, E., Gobernado, M. & Garcia-Perea, A. (2003). Survey of emm gene sequences from pharyngeal Streptococcus pyogenes isolates collected in Spain and their relationship with erythromycin susceptibility. J Clin Microbiol 41, 2385–2390.
    2. Beall, B., Facklam, R. R., Elliott, J. A., Franklin, A. R., Hoenes, T., Jackson, D., Laclaire, L., Thompson, T. & Viswanathan, R. (1998). Streptococcal emm types associated with T-agglutination types and the use of conserved emm gene restriction fragment patterns for subtyping group A streptococci. J Med Microbiol 47, 893–898.
    3. Beall, B., Gherardi, G., Lovgren, M., Facklam, R. R., Forwick, B. A. & Tyrrell, G. J. (2000). emm and sof gene sequence variation in relation to serological typing of opacity-factor-positive group A streptococci. Microbiology 146, 1195–1209.
    4. Bisno, A. L. (1991). Group A streptococcal infections and acute rheumatic fever. N Engl J Med 325, 783–793.
    5. Claire, P., Laurence, J., Gilles, Q., Patrick, B. & Patrick, T. C. (2003). Genetic basis of antibiotic resistance in Streptococcus agalactiae strains isolated in a French hospital. Antimicrob Agents Chemother 47, 794–797.
    6. De Azavedo, J. C., Yeung, R. H., Bast, D. J., Duncan, C. L., Borgia, S. B. & Low, D. E. (1999). Prevalence and mechanisms of macrolide resistance in clinical isolates of group A streptococci from Ontario, Canada. Antimicrob Agents Chemother 43, 2144–2147.
    7. Giovanetti, E., Montanari, M. P., Mingoia, M. & Varaldo, P. E. (1999). Phenotypes and genotypes of erythromycin-resistant Streptococcus pyogenes strains in Italy and heterogeneity of inducibly resistant strains. Antimicrob Agents Chemother 43, 1935–1940.
    8. Giovanetti, E., Brenciani, A., Burioni, R. & Varaldo, P. E. (2002). A novel efflux system in inducibly erythromycin-resistant strains of Streptococcus pyogenes. Antimicrob Agents Chemother 46, 3750–3755.
    9. Granizo, J. J., Aguilar, L., Casal, J., Dal-Re, R. & Baquero, F. (2000). Streptococcus pyogenes resistance to erythromycin in relation to macrolide consumption in Spain (1986–1997). J Antimicrob Chemother 46, 959–964.
    10. Grivea, I. N., Al-Lahham, A., Katopodis, G. D., Syrogiannopoulos, G. A. & Reinert, R. R. (2006). Resistance to erythromycin and telithromycin in Streptococcus pyogenes isolates obtained between 1999 and 2002 from Greek children with tonsillopharyngitis: phenotypic and genotypic analysis. Antimicrob Agents Chemother 50, 256–261.
    11. Hall, T. A. (1999). Bioedit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symp Ser 41, 95–98.
    12. Hsueh, P. R., Teng, L. J., Lee, L. N., Yang, P. C., Ho, S. W., Lue, H. C. & Luh, K. T. (2002). Increased prevalence of erythromycin resistance in streptococci: substantial upsurge in erythromycin-resistant M phenotype in Streptococcus pyogenes (1979-1998) but not in Streptococcus pneumoniae (1985-1999) in Taiwan. Microb Drug Resist 8, 27–33.
    13. Jones, K. F. & Fischetti, V. A. (1988). The importance of the location of antibody binding on the M6 protein for opsonization and phagocytosis of group A M6 streptococci. J Exp Med 167, 1114–1123.
    14. Kataja, J., Huovinen, P. & Seppala, H. (2000). Erythromycin resistance genes in group A streptococci of different geographical origins. J Antimicrob Chemother 46, 789–792.
    15. Kim, S. & Lee, N. Y. (2004). Antibiotic resistance and genotypic characteristics of group A streptococci associated with acute pharyngitis in Korea. Microb Drug Resist 10, 300–305.
    16. Kumar, S., Tamura, K. & Nei, M. (2004). mega3: integrated software for molecular evolutionary genetics analysis and sequence alignment. Brief Bioinform 5, 150–163.
    17. Mihaila-Amrouche, L., Bouvet, A. & Loubinoux, J. (2004). Clonal spread of emm type 28 isolates of Streptococcus pyogenes that are multiresistant to antibiotics. J Clin Microbiol 42, 3844–3846.
    18. National Committee for Clinical Laboratory Standards (2005). Performance Standards for Antimicrobial Susceptibility Testing, fifteenth informational supplement, M100-S15. Wayne, PA: National Committee for Clinical Laboratory Standards.
    19. Perez-Trallero, E., Marimon, J. M., Montes, M., Orden, B. & de Pablos, M. (1999). Clonal differences among erythromycin-resistant Streptococcus pyogenes in Spain. Emerg Infect Dis 5, 235–240.
    20. Saunders, N. A., Hallas, G., Gaworzewska, E. T., Metherell, L., Efstratiou, A., Hookey, J. V. & George, R. C. (1997). PCR-enzyme-linked immunosorbent assay and sequencing as an alternative to serology for M-antigen typing of Streptococcus pyogenes. J Clin Microbiol 35, 2689–2691.
    21. Seppala, H., Nissinen, A., Jarvinen, H., Huovinen, S., Henriksson, T., Herva, E., Holm, S. E., Jahkola, M., Katila, M. L. & other authors (1992). Resistance to erythromycin in group A streptococci. N Engl J Med 326, 292–297.
    22. Siberry, G. K., Tekle, T., Carroll, K. & Dick, J. (2003). Failure of clindamycin treatment of methicillin-resistant Staphylococcus aureus expressing inducible clindamycin resistance in vitro. Clin Infect Dis 37, 1257–1260.
    23. Soussy, C. J., Carret, G., Cavallo, J. D., Chardon, H., Chidiac, C., Choutet, P., Courvalin, P., Dabernat, H., Drugeon, H. & other authors (2000). Comité de l'Antibiogramme de la Société Française de Microbiologie Communiqué 2000–2001. Pathol Biol 48, 832–871.
    24. Sungho, C., Hoanjong, L., Kyungwon, L., Kyujam, H., Songmee, B. & Younghee, L. (2001). The emergence of erythromycin-resistant Streptococcus pyogenes in Seoul, Korea. J Infect Chemother 7, 81–86.
    25. Sutcliffe, J., Grebe, T., Tait-Kamradt, A. & Wondrack, L. (1996). Detection of erythromycin-resistant determinants by PCR. Antimicrob Agents Chemother 40, 2562–2566.
    26. Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. (1997). The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res 24, 4876–4882.
    27. Weisblum, B. (1995). Erythromycin resistance by ribosome modification. Antimicrob Agents Chemother 39, 577–585.
    28. Whatmore, A. M., Kapur, V., Sullivan, D. J., Musser, J. M. & Kehoe, M. A. (1994). Non-congruent relationships between variation in emm gene sequences and the population genetic structure of group A streptococci. Mol Microbiol 14, 619–631.