Infection, Genetics and Evolution 29 (2015) 35–41

Contents lists available at ScienceDirect

Infection, Genetics and Evolution journal homepage: www.elsevier.com/locate/meegid

Sequence types diversity of Legionella pneumophila isolates from environmental water sources in Guangzhou and Jiangmen, China Jingyu Guo a,b, Ting Liang a,c, Chaohui Hu b, Ruichen Lv a, Xianwei Yang a, Yujun Cui a, Youtao Song c, Ruifu Yang a, Qingyi Zhu b,⇑, Yajun Song a,⇑ a b c

State Key Laboratory of Pathogen and Biosecurity, Beijing Institute of Microbiology and Epidemiology, Beijing, China Guanzhou Kingmed Centers for Clinical Laboratory, Guangzhou, China Department of Microbiology, Liaoning University, Shenyang, China

a r t i c l e

i n f o

Article history: Received 8 July 2014 Received in revised form 17 October 2014 Accepted 25 October 2014 Available online 31 October 2014 Keywords: Legionella pneumophila Sequence-based typing Recombination

a b s t r a c t In this study, 159 Legionella pneumophila strains isolated from various natural and artificial water sources in Guangzhou and Jiangmen, China, were subjected to genotyping by the sequence-based typing (SBT) scheme. These isolates were assigned into 53 sequence types (STs) (50 STs with seven loci data and three unidentified STs with incomplete loci profiles) with ST1 as the dominant one (14.5%), and the index of diversity (IOD) was 0.950. Eight new alleles and 34 new STs were reported here. Notably, most of the newly identified STs with seven loci data (24/34) contained no new allele, implying frequent recombination events in L. pneumophila. Five intragenic recombination events were identified in the concatenated sequences of seven loci. The diversity of STs in natural environmental isolates (41 STs, IOD = 0.956) is higher than that of artificial environmental ones (17 STs, IOD = 0.824). The ST patterns varied in isolates from these two sources: the most common STs from artificial water sources, ST1 and ST752 (39.2% and 13.7%), were only occasionally isolated from natural water sources (2.9% and 3.8%, respectively); while the predominant STs from natural water sources, ST1048, ST739 and ST1267 (15.2%, 6.7% and 6.7%), were less frequently seen in artificial environments (2.0%, 0% and 0%, respectively). We also found out that Legionnaires’ disease associated STs might be more frequently isolated in artificial environments than in natural ones. Our data revealed remarkable genetic diversity of L. pneumophila isolates from environmental water systems of Guangzhou and Jiangmen, and the different ST distribution patterns between natural water and artificial water sources as well. Ó 2014 Published by Elsevier B.V.

1. Introduction Legionella pneumophila is the causative agent of Legionnaires’ disease, an atypical pneumonia (Fraser et al., 1977), and Pontiac fever, a self-limited flu-like illness (Glick et al., 1978). Although there are 57 valid species in the Legionella genus already, L. pneumophila causes approximately 90% of all reported Legionnaires’ disease cases, 84% of which are due to L. pneumophila serogroup 1 (Fields et al., 2002; Yu et al., 2002). As reviewed in a recent paper (Luck et al., 2013), several molecular typing methods have been applied to molecular epidemiological investigations of L. pneumophila, such as pulsed-field gel electrophoresis (PFGE), amplified fragment length polymorphism (AFLP) and sequence-based typing (SBT). The SBT scheme, proposed by the European Working Group ⇑ Corresponding authors. E-mail addresses: [email protected] (Q. Zhu), [email protected] (Y. Song). http://dx.doi.org/10.1016/j.meegid.2014.10.023 1567-1348/Ó 2014 Published by Elsevier B.V.

for Legionella Infections (EWGLI, now the ESCMID Study Group for Legionella Infections, ESGLI), is performed by sequencing and comparing seven loci (flaA, pilE, asd, mip, mompS, proA and neuA) (Gaia et al., 2005; Ratzow et al., 2007). The results are reported as the combination of alleles (seven-digit allelic profiles) and serial numbered sequence types (STs). With the advantages of data robustness and ease for inter-laboratory comparisons, SBT is now globally accepted and widely used in genotyping of L. pneumophila. Since human-to-human transmission has not been reported for Legionnaires’ disease (LD) yet, human LD is termed as an environmental disease, which is always transmitted via the inhalation of Legionella-containing aerosols (Blatt et al., 1993). L. pneumophila is ubiquitous in aquatic environments, including cooling towers, water towers, ponds, lakes and rivers. Theoretically, these aquatic environments could all serve as potential sources of Legionella infection. Previous studies reported that almost all cases of Legionnaires’ disease could be traced to artificial water environments instead of natural aquatic environments (Coetzee et al., 2012).

36

J. Guo et al. / Infection, Genetics and Evolution 29 (2015) 35–41

To determine whether there are different STs distributions patterns of L. pneumophila isolates between artificial environments and natural environments, we performed SBT assays for L. pneumophila strains isolated from two water sources from Guangzhou and Jiangmen in Guangdong Province, China. 2. Materials and methods 2.1. Sampling and processing of water samples During Oct 2003 to Dec 2008, 249 water samples were collected from natural (163 samples) and artificial water environments (86 samples) in Guangzhou City and Jiangmen City, Guangdong Province, China. Samples from lakes, rivers, harbors, ponds, and springs were termed as natural water sources, while those from cooling towers and water towers as artificial ones. Isolation of Legionellalike bacteria were performed as previously described (Zhan et al., 2014). Briefly, 500 ml water was concentrated by filtration through MicroFunnel 4800 with 0.45 lm Metricel membrane filters (Pall Inc., USA). Filter membranes were aseptically transferred into 5 ml of sterile water and vortexed for 5 min to release the microorganisms. 2.2. Isolation of L. pneumophila Fifty microliters of the processed samples were plated onto GVPC plate (Oxoid Inc., USA) after heat treatments (50 °C for 30 min) and acid treatments (0.2 mol/L HCl–KCl, pH 2.0 for 5 min) to eliminate non-Legionella organisms. Plates were incubated at 37 °C enriched with 5% CO2 (v/v) for 3–7 days. Colonies exhibiting Legionella-like morphology were transferred to BCYEa agar (with L-cysteine), BCYEa-cys agar (without L-cysteine), and Columbia blood agar (Oxoid Inc., USA). Colonies able to grow on BCYEa but not on BCYEa-cys- agar and Columbia blood agar plates were possibly Legionella species, which were then identified to the species level by sequencing partial 16S rRNA gene (Wilson et al., 2007) and mip gene as previously reported (Ratcliff et al., 1998).

2.5. Population genetics analysis Hunter and Gaston’s modification of Simpson’s index of diversity (IOD) was calculated by using the allele profiles of all the isolates to evaluate the diversity of STs (Hunter and Gaston, 1988). The proportion of each ST was compared between the natural and artificial water isolates using Fisher exact test (PASW Statistics 17.0, SPSS Inc., USA). A maximum likelihood phylogenetic tree was obtained by using the concatenated sequences of seven loci (2501/ 2498 bp) and 1000 bootstrap replicates using MEGA5 package (Tamura et al., 2011). The population structure of the strain collections was reflected by a minimum spanning tree (MSTree) based on allele profiles by using Bionumerics 6.6 (Applied Maths Inc., Belgium). To compare the diversity of concatenated SBT sequences between the isolates from natural and artificial water sources, hierarchical AMOVA (Analyses of Molecular Variance) was performed with program Arlequin 3.5 (Excoffier et al., 2005). 2.6. Recombination analysis The standardized index of association (ISA) was calculated with START2 software (Jolley et al., 2001) to evaluate the linkage disequilibrium by using allele profiles of the isolates. The intragenic recombination was determined with the Sawyer’s test implemented in START2 (Jolley et al., 2001) by using individual single locus sequences. The intergenic recombination was screened by RDP4 software (Martin et al., 2010) using concatenated sequences of seven loci (the highest P-value was set at P 6 0.01). To visualize possible recombination events in the population, a reticulate network tree was prepared by using the Neighbor-net algorithm of SplitsTree4 software (Huson and Bryant, 2006). ClonalFrame (Didelot and Falush, 2007) was used to estimate two key parameters: the ratio of recombination and substitution rates (q/h), and the ratio of probabilities that a given site is altered through recombination or substitution (r/m). 3. Results and discussion

2.3. DNA extraction from L. pneumophila isolates

3.1. Isolation of L. pneumophila

The identified L. pneumophila isolates were grown on BCYEa agar plates at 37 °C for two days, and then the bacteria cultures were harvested from the plates. Genomic DNAs were extracted by using the conventional SDS lysis and phenol–chloroform method. DNAs were dissolved in TE buffer (10 mmol/L Tris/HCl, 1 mmol/L EDTA, pH 8.0) and stored at 20 °C after concentration measurements.

A total of 159 L. pneumophila strains were isolated from environmental water sources from Guangzhou City (112, 70.4%) and Jiangmen City (47, 29.6%) from Oct 2003 to Dec 2008 (Table S1). Amongst the 159 L. pneumophila isolates, 106 were isolated from natural water sources, including lakes, rivers, harbors, ponds, and springs (163 water samples). The other 53 L. pneumophila strains were isolated from artificial water sources, including cooling towers and water towers (86 water samples). The isolation rates of L. pneumophila from natural and artificial water sources were similar (65.0% vs. 61.6%), which was close to that of cooling tower waters in Shanghai (58.9%, 189/321) and hot spring waters in Beijing (Lin et al., 2009; Qin et al., 2013). These high isolation rates of Legionella from environmental water sources in China deserved more investigations.

2.4. Sequence-based typing The ST of each L. pneumophila isolate was determined by using the standard protocol from ESGLI with seven gene fragments (flaA, pilE, asd, mip, mompS, proA and neuA) (Gaia et al., 2005; Ratzow et al., 2007). For the strains that failed to get neuA amplification, primers targeting neuAh were used as suggested by ESGLI (Farhat et al., 2011). The PCR products were sent to BGI Beijing Inc. for purification and sequencing. The sequence trace files were submitted to the L. pneumophila SBT website to get the trimmed sequences, and then queried against existing alleles in the SBT database by using the ‘‘Sequence Quality Tool’’ module (http://www.hpa-bioinformatics.org.uk/legionella/legionella_sbt/ php/sbt_homepage.php). Strains with known allele profiles were assigned with the corresponding ST in the database, while those with new alleles or allele profiles were assigned as new STs by the database curator.

3.2. L. pneumophila sequence-based typing Of these 159 isolates, 156 yielded complete SBT data (all seven loci) and were identified as 50 STs, while the remaining three isolates with only six allele profiles were designated as unidentified STs (named as STGDXX) and excluded for further analysis (Table S1). For the three strains with incomplete loci data, two failed flaA amplification and one failed proA gene, which might be caused by genetic variations within the primer-binding regions of the corresponding isolates.

J. Guo et al. / Infection, Genetics and Evolution 29 (2015) 35–41

Eight new alleles (asd48, 49 and 50; mip52, 73; mompS67, 68 and 69) were reported in this study (Table S1), and mip52 and mompS67 occurred exclusively in the same isolate (km067). Of the 50 identified STs, thirty-four were newly reported in this study (Table S1). Notably, only 10 of these new STs had new alleles and all the other 24 STs resulted from the reassortment of the existing alleles, which implied recombinations events in the population. Taken together, 50 STs in 156 isolates implied a high discriminative power in our strain collection (IOD = 0.950, 95% CI, 0.935– 0.965). The IOD of environmental isolates in our investigation (0.950) is higher than that previously reported in China (0.749) (Qin et al., 2014), Japan (0.886) (Amemura-Maekawa et al., 2012), Canada (0.888) (Reimer et al., 2010) and United States (0.822) (Kozak et al., 2009), and slightly lower than United Kingdom (0.954) (Harrison et al., 2009) and Singapore (0.970) (Lim et al., 2011). Taking the small sampling area of this study (two cities) into account, we speculated a higher genetic diversity of L. pneumophila isolates in our study. The most dominant ST was ST1, which accounted for 14.7% (23/ 156) of all L. pneumophila isolates. ST1 is the most prevalent ST around the world (Harrison et al., 2009; Kozak et al., 2009; Reimer et al., 2010) and the most common ST reported to the ESGLI SBT database (ca. 14%). While ST1048, a novel ST identified in this study, constituted 10.9% (17/156) of all isolates. Other major STs with more than 5 isolates were ST752 (n = 11), ST630 (n = 9), ST739 (n = 7), ST1267 (n = 7), ST1054 (n = 6), ST1056 (n = 6), ST1417 (n = 6) and ST1777 (n = 5), and 33 STs included only one isolate (Table S1). ST59, ST630 and ST752 were also found in water samples from other cities in China in a recent study (Qin et al., 2014). The wide distribution of these STs around China is worth of further investigation. 3.3. Population structure of L pneumophila isolates Fig. 1 shows the minimum spanning tree (MSTree) of the 50 STs with seven loci based on allele profiles, which revealed 6 clonal complexes (CCs, connected with single locus variants) and 30 singletons. Among the 6 CCs, CC1 (including ST1, ST172, ST630, ST752, ST1040, ST1046, ST1262 and ST1777) was the dominant CC including 52 isolates. CC2 included 3 isolates belonging to ST1785 and ST1786. CC3 included 3 isolates which belonged to ST160, ST1778 and ST1779. While CC4 included 7 isolates belonging to ST1054 and ST1265, CC5 included 7 isolates (ST1056 and ST1060) and CC6 included 8 isolates (ST1059, ST1413 and ST1417). Of the 30 singletons, ST1048 was the prevalent ST, which was mainly isolated from natural sources (16/17, 94.1%). The other

37

frequently identified singleton-STs were, ST739 (7 isolates), ST1267 (7 isolates), ST45, ST242 and ST1049 (4 isolates). Eighteen singleton STs included only one isolate in this study (Table S1). Notably, ESGLI SBT database now contains more than 1800 STs, and the singletons described here were only based on the 50 STs in this study. Some of the ‘‘singletons’’ will be identified as members of certain CCs if they were analyzed with all the STs in the database. Clonal complexes are based on allele profile comparison. However, two strains can differ in one locus by more nucleotides differences than the sum of the SNPs introduced by a double loci variant or triple variant. Fig. 2A showed the maximum likelihood (ML) tree of the concatenated sequences of the seven loci of the 50 STs by using MEGA 5 software (Tamura et al., 2011). Notably, some strains belonging to the same CC are far away from each other in the ML tree. For instance, ST1777 is a member of CC1, but it does not cluster with ST1 in the ML tree. This is also the case for ST1785 (CC2), ST1413 and ST1417 (CC6). All these STs failed to get neuA amplification, and get successful amplification with neuAh primers. Significant nucleotide variations had been reported in neuA homologues (Farhat et al., 2011; Mentasti et al., 2014), which accounted for the inconsistent results between allele profiles based CC analysis and nucleotides based phylogenetic analysis. 3.4. ST distributions in different environments Table 1 shows the ST compositions of L. pneumophila isolated from the natural water and artificial water sources. One hundred and five isolates from natural water were grouped into 41 STs, while 51 artificial environmental isolates were grouped into 17 STs. The diversity of isolates from natural environments (IOD = 0.956, 95% CI, 0.938–0.974) is higher than that of strains from artificial environments (IOD = 0.824, 95% CI, 0.734–0.914), which implied a higher genetic diversity in the natural water isolates. It had been reported that the diversity of strains from natural environments (12 STs for 13 strains, IOD = 0.974) is higher than the ones from artificial water environment (33 STs for 73 strains, IOD = 0.866) (Reimer et al., 2010). A recent study involving 6 cities in China also revealed a similar pattern: IOD of isolates from hotsprings (0.934, 19 STs for 42 strains) were higher than that from cooling towers (0.711, 25 STs for 96 strains) (Qin et al., 2014). Besides IOD comparison, we also compared the genetic diversity of the concatenated sequences of natural and artificial water isolates. In the hierarchical AMOVA analysis, all isolates were assigned into two groups (Guangzhou and Jiangmen), and isolates in each group were assigned into two populations (natural and artificial water systems). The fixation index among groups (FCT)

Fig. 1. Minimum spanning tree (MSTree) of 50 L. pneumophila STs with seven alleles in this study based on SBT allele profiles. STs are shown as nodes scaled to isolates numbers. The thick lines links single locus variants (SLV) and the thin lines links double loci variants (DLV). Eight clonal complexes (CC) composed of SLVs are also shown.

38

J. Guo et al. / Infection, Genetics and Evolution 29 (2015) 35–41

A

ST-1050 ST-1057 ST-1056 93 ST-1060 ST-1052 ST-763 ST-1048 ST-1058 59 93 ST-1061 ST-1062 64 ST-1047 ST-242 ST-93 69

ST-1060 ST-1061 ST-1052 ST-1048 ST-1047 ST-1050 ST-1056 ST-763 ST-1062 ST-1057 ST-1058

B

ST-59 ST-1054 ST-93 ST-114

ST-1778

ST-160

ST-1265 ST-242

ST-1788

ST-1779 ST-1266 ST-1049 ST-739 ST-624 ST-1059

ST-114 ST-1778 ST-160 100 84 ST-1779

ST-1053 ST-1055 ST-1051 ST-1264

ST-45

ST-752 ST-1 ST-172 ST-630 ST-1046 ST-1040 ST-1262 ST-1786

ST-1263 ST-1267

ST-59 89 ST-1054 96 ST-1265 ST-45 ST-624 98 ST-1049 75 ST-739 53 ST-1266 ST-1059 90 ST-1053 86 ST-1055 ST-1051 ST-1264 ST-1788 ST-1786 ST-1040 53 ST-1262 ST-1046 ST-630 77 63 ST-172 55 ST-1 93 ST-752 ST-1263 95 ST-1267

ST-1785 ST-1777 ST-1417 ST-1413

100 ST-1413 ST-1417 ST-1777 ST-1785 100 ST-1782 ST-1781 ST-1418 100 ST-1784 81 65 ST-1780 94

99

100

0.01

ST-1418 ST-1782 ST-1780 ST-1781 ST-1784

0.01

Fig. 2. (A) Maximum likelihood tree from the concatenated alignments of the concatenated sequences of seven loci (2501/2498 bp) of 50 L. pneumophila STs in this study. Bootstrap values (1000 replicates) greater than 50% are labeled on the branches. (B) Reticulate network tree by using Neighbor-net algorithm of SplitsTree4 using the sevenloci concatenated alignments of the 50 STs. All internal nodes represent hypothetical ancestral STs and edges correspond to reticulate events such as recombinations.

was 0.082, and the variation components did not vary significantly among the groups (P = 0.66), implying no difference of genetic diversities of isolates from these two cities. On the other hand, fixation index among populations (FSC) was 0.178, and genetic diversity varied significantly among populations (P < 0.001). Our results, together with previously reports, support the notion that isolates from natural water systems have more genetic diversities than those from artificial ones. Considerable differences can be found in the ST distributions in these two water sources. There were more natural water specific STs (33 STs) than artificial specific STs (9 STs). Only eight STs were shared by isolates from both water sources (ST1, ST242, ST630, S752, ST1048, ST1054, ST1417 and ST1777), and they had different patterns in these two sources. For instance, ST1 and ST752 were frequently isolated from artificial environments (39.2% and 13.7%, respectively) but less frequently seen in natural environments

(2.9% and 3.8%, respectively); while ST1048 was common in natural environments (15.2%) but infrequently isolated in artificial environments (2.0%). The other two STs predominant in natural environments, ST739 and ST1267, were not seen in artificial water systems. The frequencies for other shared STs from these two water sources also differed from each other. Table 1 summarized ST distributions L. pneumophila in these two water sources. The proportion of ST1 strains was significantly higher in artificial environments (Fisher exact test, P < 0.001), and the proportion of ST1048 isolates was higher in natural environments (Fisher exact test, P = 0.012). Our findings reinforce the previous findings that the distribution of STs between natural water sources and artificial environment was significantly different (Amemura-Maekawa et al., 2012; Qin et al., 2014; Reimer et al., 2010), with more L. pneumophila strains from natural water sources included in our study to support the conclusion.

39

J. Guo et al. / Infection, Genetics and Evolution 29 (2015) 35–41 Table 1 ST distributions in natural and artificial environments isolates.

a b c

ST

Allelic profile

ST1 ST1048 ST752 ST630 ST739 ST1267 ST1054 ST1056 ST1417 ST1777 ST45 ST242 ST1049 ST1052 ST1053 ST1055 ST59 ST1785 ST1266 ST1780 Other STsa Totalc

1,4,3,1,1,1,1 6,10,17,3,4,14,11 22,4,3,1,1,1,1 1,4,3,1,1,1,10 12,8,11,2,10,12,2 2,6,48,6,48,5,40 32,12,50,6,48,11,9 2,10,14,28,21,14,13 8,6,34,9,2,8,209 1,4,3,1,1,1,215 5,1,22,26,6,10,12 3,10,1,28,1,9,3 12,8,11,2,11,12,4 2,10,15,28,21,3,2 6,16,14,28,4,14,3 6,16,15,28,4,4,3 7,6,17,3,13,11,11 2,15,3,73,29,1,201 12,15,11,56,29,12,34 6,10,20,28,13,14,207

Natural

Artificial

Fisher’s exact test (P-value)

n

%

n

%

3 16 4 7 7 7 2 6 3 4 4 1 4 3 3 3 0 2 2 2 22 105

2.9 15.2 3.8 6.7 6.7 6.7 1.9 5.7 2.9 3.8 3.8 1.0 3.8 2.9 2.9 2.9 0.0 1.9 1.9 1.9 21.0 100.0

20 1 7 2 0 0 4 0 3 1 0 3 0 0 0 0 2 0 0 0 8 51

39.2 2.0 13.7 3.9 0.0 0.0 7.8 0.0 5.9 2.0 0.0 5.9 0.0 0.0 0.0 0.0 3.9 0.0 0.0 0.0 15.7 100.0