Appl Microbiol Biotechnol (2014) 98:3317–3326 DOI 10.1007/s00253-013-5402-z

ENVIRONMENTAL BIOTECHNOLOGY

Tracking human sewage microbiome in a municipal wastewater treatment plant Lin Cai & Feng Ju & Tong Zhang

Received: 2 October 2013 / Revised: 8 November 2013 / Accepted: 11 November 2013 / Published online: 5 December 2013 # Springer-Verlag Berlin Heidelberg 2013

Abstract Human sewage pollution is a major threat to public health because sewage always comes with pathogens. Human sewage is usually received and treated by wastewater treatment plants (WWTPs) to control pathogenic risks and ameliorate environmental health. However, untreated sewage that flows into water environments may cause serious waterborne diseases, as reported in India and Bangladesh. To examine the fate of the human sewage microbiome in a local municipal WWTP of Hong Kong, we used massively parallel sequencing of 16S rRNA gene to systematically profile microbial communities in samples from three sections (i.e., influent, activated sludge, and effluent) obtained monthly throughout 1 year. The results indicated that: (1) influent sewage bacterial profile reflected the human microbiome; (2) human gut bacterial community was the dominant force shaping influent sewage bacterial profile; (3) most human sewage bacteria could be effectively removed by the WWTP; (4) a total of 75 genera were profiled as potentially pathogenic bacteria, most of which were still present in the effluent although at a very low level; (5) a grouped pattern of bacterial community was observed among the same section samples but a dispersed pattern was found among the different section samples; and (6) activated sludge was less affected by the influent sewage bacteria, but it showed a significant impact on the effluent bacteria. All of these findings provide novel insights toward a mechanistic understanding of the fate of human sewage microbiome in the WWTP.

Electronic supplementary material The online version of this article (doi:10.1007/s00253-013-5402-z) contains supplementary material, which is available to authorized users. L. Cai : F. Ju : T. Zhang (*) Environmental Biotechnology Laboratory, Department of Civil Engineering, The University of Hong Kong, Hong Kong, SAR, China e-mail: [email protected]

Keywords Sewage . Human sewage microbiome . Pathogens . Wastewater treatment plant . Activated sludge . 454 pyrosequencing

Introduction The human sewage microbiome, here for the first time, is referred to as the collective microbes in sewage from human domestic waste such as feces, urine, sweat, washing, bathing, etc. These microbes are mainly derived from the human body including skin, respiratory tract, oral cavity, gastrointestinal tract, and urogenital tract. Most of them are harmless, and even beneficial to human health, e.g., human gut bacteria play important roles in food digestion, nutrient intake, vitamin synthesis, and immune response (LeBlanc et al. 2013; Prakash et al. 2011). However, their occurrence in environments indicates that potential pathogens may be co-existing (Ferguson et al. 2012; Sidhu et al. 2012; Staley et al. 2012). Sewage pollution is one of the major sources responsible for water quality impairment because sewage contains both rich nutrients and abundant microbes. A recent report from the Centre for Science and Environment (New Delhi, India) states that 80 % of sewage in India is untreated and flows into water systems, thereby creating a serious pollution and health risk of drinking water. A case study revealed that groundwater from shallow tube wells in rural Bangladesh was contaminated by human fecal bacteria, where both bacterial and viral pathogens were detected at very high levels (Ferguson et al. 2012). The finding of human sewage bacteria in aquatic environments represents a public health threat to drinking water, water entertainment, and aquatic food consumption. As a result, there is an increasing interest in the detection and monitoring of sewage pollution in various aquatic environments by determining a small subset of fecal indicator bacteria such as Escherichia coli , Enterococci , and fecal Streptococci

3318

Appl Microbiol Biotechnol (2014) 98:3317–3326

(Ahmed et al. 2012; Chen and Walker 2012; Jeanneau et al. 2012; Pickering et al. 2012; Shanks et al. 2012). Methods most commonly used include culture-based technique (Ahmed et al. 2012; Sidhu et al. 2012; Staley et al. 2012), PCR (Gordon et al. 2013; Murugan et al. 2012), quantitative PCR (Ahmed et al. 2012; Shanks et al. 2012), microarray (Dubinsky et al. 2012), and 454 pyrosequencing (Lee et al. 2011; McLellan et al. 2010; McLellan et al. 2013; Newton et al. 2011; Shanks et al. 2013; Unno et al. 2010). Municipal wastewater treatment plants (WWTPs) are fullscale bioreactors mainly designed to treat domestic sewage released by urban residents. Generally, all sewage should be received and treated by WWTPs before being discharged into environments. Hence, a crucial task for sewage treatment is to monitor and track the human sewage microbiome to document its fate and removal. However, for the past decades, most WWTPs employed traditional colony count of selected indicator bacteria as the detection technique, which may not reflect the real situation. Newly established high-throughput sequencing technique and analysis can instead accurately and efficiently analyze microbial community from global WWTPs in a single test (Zhang et al. 2012). Importantly, there is a lack in systematic monitoring human sewage microbiome in a WWTP using this technique. This study is aimed to detect and track human sewage microbiome and pathogens contained in influent, activated sludge, and effluent from the municipal WWTP of Hong Kong monthly throughout 1 year using barcoded pyrosequencing of 16S rRNA gene. It is also aimed to confirm the hypothesis that influent sewage bacterial profile reflects human microbiome and the bacteria involved can be effectively removed by wastewater treatment. In this study, a comprehensive profile of the human sewage microbiome in three different section samples of the WWTP will be made in an effort to better understand its fate and environmental impact. Moreover, it is anticipated that the influent sewage bacterial profile of this study will serve as a valuable reference for other sewage pollution investigations through a comparison of 16S pyrosequencing results.

nitrogen≤20 mg/L. The effluent was finally directed into Hong Kong Victoria Harbor via discharge tunnel.

Materials and methods

PCR

WWTP plant description

To target bacterial 16S rRNA gene hypervariable V3-V4 region, a primer set of 338 F (5′-ACTCCTACGGGAGGCA GCAG-3′, E. coli 16S position: 338–357) and 802R (5′TACNVGGGTATCTAATCC-3′, E. coli 16S position: 785– 802) was employed for PCR amplification (Baker et al. 2003). This primer set was strongly recommended for use in PCRbased 16S pyrosequencing because it showed a better performance than the other primer sets (Cai et al. 2013). Unique barcodes were fused between 454 A adaptor and forward primer of 338 F to allow multiplex sequencing. The PCR

All samples used in this study were collected from Hong Kong Shatin WWTP (114°12′50″ E, 22°24′25″ N). Set up in 1982, it was the largest biological WWTP in Hong Kong with a total capacity of treating 340,000 ton domestic wastewater per day. The raw sewage had a chemical oxygen demand of 200–500 mg/L. The discharge standards of treated effluent were total suspended solids≤30 mg/L, 5-day biochemical oxygen demand≤20 mg/L, ammonia nitrogen≤5 mg/L, total

Sampling We sampled influent, activated sludge, and effluent at the same positions every month from August 2011 to July 2012. The sample IDs were designated as IN2011-08 to IN2012-07, AS2011-08 to AS2012-07, and EF2011-08 to EF2012-07 (Table S1). For example, IN2011-08, AS2011-08, and EF2011-08 represented influent, activated sludge, and effluent collected in August 2011, respectively. Additionally, three samples of IN2012-07R, AS2012-07R, and EF2012-07R collected in July 2012 were designed as technical replicates. At sampling, suspended activated sludge containing both flocs and free microbial cells was taken into 50-mL sterilized tubes and immediately fixed with an equal volume ratio of ethanol. Influent and effluent samples were transferred to sterilized containers and incubated using an ice bath. After reaching the laboratory, the influent sample was centrifuged to collect the cell pellet which was suspended with 50 % ethanol. The effluent sample was filtered through a 0.22-μm membrane to catch microbial cells which were fixed with 50 % ethanol. After pretreatment, the samples were all stored at −20 °C before total DNA extraction. DNA extraction FastDNA® Spin Kit for Soil (MP Biomedicals, France) was used for DNA extraction because it has been evaluated as the most suitable kit to isolate metagenomic DNA from WWTP samples by comparing with several commercially available kits (Guo and Zhang 2013). Total DNA was extracted from the stored influent, activated sludge, and effluent samples following the manufacturer’s protocol. Each sample was extracted in duplicate to minimize the potential bias during DNA extraction. Duplicate DNA extracts were then pooled to determine the DNA purity and yield using a Thermo NanoDrop 1000 Spectrophotometer. Finally, the prepared DNA extracts were used as PCR templates.

Appl Microbiol Biotechnol (2014) 98:3317–3326

was conducted by a BioRad i-Cycler under a program of 5 min at 94 °C for initial denaturation, followed by 30 cycles of 94 °C for 0.5 min, 50 °C for 0.5 min, 72 °C for 1 min, and followed by 5 min at 72 °C for final extension. Each PCR reaction contained 50 ng DNA, 200 nM of each primer (Integrated DNA Technologies, US), Premix Ex Taq 25 μL (TaKaRa, Japan), and ddH2O up to 50 μL. Triplicate amplification was performed for every DNA sample. Each triplicate PCR product was pooled for purification (~150 μL) using PCRquick-spin™ PCR Product Purification Kit (iNtRON Biotechnology, Korea). The purified products were quantified by the Thermo NanoDrop 1000 Spectrophotometer and pooled at equivalent ratios for 454 massively parallel sequencing. 454 pyrosequencing The pooled PCR sample was submitted to Centre for Genomic Sciences at The University of Hong Kong to conduct emulsion PCR and sequencing on a Roche 454 FLX Titanium sequencer (Roche, US). A relatively deep sequencing of two 454 runs was carried out because a previous study demonstrated that bacterial communities in 14 global WWTPs are highly complex (Zhang et al. 2012). For public use, the generated data of two runs have been deposited at MGRAST server (http://metagenomics.anl.gov/) under accession numbers 4525937.3, 4525938.3, 4525939.3, and 4525940.3. Data quality filtration The raw data was firstly trimmed using QIIME platform to assign multiplexed 16S pyrotags into each sample identified by a unique barcode (Caporaso et al. 2010). The denoising and chimera checking were further performed by QIIME’s Denoiser and ChimeraSlayer wrappers, respectively. After quality filtration under QIIME pipeline, the clean pyrotags of each sample was recovered for downstream bioinformatic analysis.

3319

(PCoA). The OriginPro 8 and MATLAB 7.12.0.635 tools were employed for graph plotting.

Results Deep sequencing and unusual classification profile for the influent pyrotags The numbers of clean pyrotags obtained after filtering lowquality, noise, and chimera reads by the QIIME platform are listed in Table S1. A total of 1,017,026 clean pyrotags were generated in the two 454 runs, revealing an average of 26,078 clean pyrotags assigned to each sample. As shown in Table S1, the numbers of genera identified by the RDP Classifier in the influent, activated sludge, and effluent samples were 225–280, 149–194, and 222–306, respectively. The genus coverage (number of pyrotags to cover per genus averagely) for each sample was in the range of 78–159, with an average value of 116. The rarefaction curves shown in Fig. S1 revealed that sufficient sequencing depth was obtained. The sequencing depth, genus coverage, and rarefaction analysis all suggested that a relatively deep sequencing was conducted in this study. Using the RDP Classifier at an 80 % confidence threshold, pyrotags were classified at six taxonomic levels. As profiled in Fig. 1, most of the influent pyrotags (averagely 84 %) could be assigned down to the genus level, but much less for the activated sludge (averagely 33 %) and effluent (averagely 46 %) pyrotags. The activated sludge and effluent pyrotags shared similar classification profile but significantly different from the influent pyrotags (Fig. 1), hinting a particular bacterial community in the influent sewage. This high percentage of influent pyrotags assigned at genus level suggested a similar property with the human gut microbiome (Claesson et al. 2010). Influent sewage bacterial profile reflected the human microbiome

Bioinformatic analysis In this study, all cleaned 16S pyrotags were assigned at each taxonomic level by the online tool of Ribosomal Database Project (RDP) Classifier at an 80 % confidence threshold (Cole et al. 2009; Wang et al. 2007). The resulting taxonomic data were copied into an Excel table individually to form a matrix, which were then sorted and extracted at phylum and genus levels for statistical analysis. Potentially pathogenic bacteria were profiled using a heat map at genus level based on the taxonomic list of bacterial pathogens in Table S1 (Woolhouse et al. 2006). The PAST statistical software package was used to conduct principal coordinates analysis

As shown in Figs. 2 and S2, phylum abundance analysis revealed that Firmicutes was the most dominant phylum in the influent samples at an average percentage of 54 %. This high abundance of Firmicutes found in the influent sewage was consistent with the human microbiome (Huttenhower et al. 2012). Proteobacteria , Actinobactria , and Bacteroidetes were detected as the other less dominant phyla for the influent sewage, averagely accounting for 34, 7, and 2 %, respectively. TM7 , Fusobacteria , Synergistetes , Verrucomicrobia, and Chloroflexi were profiled as the rare phyla with an average abundance of 0.6, 0.5, 0.4, 0.1, and 0.1 %, respectively. All of these phyla and abundance of the

3320

Appl Microbiol Biotechnol (2014) 98:3317–3326

influent sewage bacteria showed a similar feature with the human microbiome. As shown in Figs. 3, S3, and S4, genus abundance analysis further revealed that the influent sewage bacterial profile shared a high similarity with the human microbiome. For example, the genera of Streptococcus , Moraxella and Corynebacterium, and Lactobacillus represented the most dominant bacteria in the oral cavity, respiratory tract, and vagina, respectively (Huttenhower et al. 2012; Ravel et al. 2011). These four genera were also detected with high abundances in this study, referring to no. 1, 25, 61, and 35 genera in Fig. 3. Moreover, many typical gut bacteria were also found at very high levels in the influent sewage such as Blautia , Lachnospiracea incertae sedis , Acinetobacter , Bifidobacterium , Enterococcus , Faecalibacterium , Ruminococcus, Dorea, and others, referring to no. 4, 9, 10, 11, 12, 13, 16, and 21 genera in Fig. 3. The total abundance of these genera was nearly 50 %, indicating that the largest bacterial collection in the influent sewage was derived from the human gut. Therefore, the human gut bacterial community was the dominant force shaping the influent sewage bacterial profile. Fate of human sewage microbiome in the Shatin WWTP As shown in Fig. 1, a very high percentage of the influent pyrotags could be classified at genus level but with an extremely decreased level for both pyrotags of the activated sludge and effluent, suggesting a significant shift of bacterial community from the influent sewage to the activated sludge and effluent. At phylum level (Fig. 2), the most abundant phylum in the influent sewage was Firmicutes (54 %), but its abundance in the activated sludge and effluent decreased significantly, only accounting for 3 and 6 %, respectively.

Besides, two rare phyla including Fusobacteria and Synergistetes also showed a decreased abundance from the influent sewage to the activated sludge and effluent. At genus level (Figs. 3 and S3), the disappearance of the human sewage bacteria was even more obvious. It is evident that most of the human sewage bacteria found with a high abundance in the influent were present at an extremely low abundance in the activated sludge and effluent, e.g., those genera marked with an arrow in Fig. 3. The removal percentage for most human sewage bacteria in the influent sewage was estimated at over 95 %. These findings indicated that human sewage bacteria were effectively removed by wastewater treatment. Occurrence and fate of potential bacterial pathogens in the Shatin WWTP A total of 140 genera were identified as potential bacterial pathogens and listed in Table S2 (Woolhouse et al. 2006). The detected genera and abundance for all samples of this study were illustrated using a heat map (Fig. 4). Among the 140 genera, 75 of them (54 %) were found in the Shatin WWTP, which suggested that the WWTP was a large reservoir of bacterial pathogens and a potential threat to environmental health. As shown in Fig. 4, the overall diversity and abundance of bacterial pathogens in the influent was much higher than that in the activated sludge, indicating a decreased level of bacterial pathogens in the total bacterial population of the activated sludge. Compared to the activated sludge, a slightly higher level of bacterial pathogens was observed in the effluent, revealing an elevated percentage of bacterial pathogens in the total effluent bacterial population. However, the overall abundance of bacterial pathogens in the effluent was much lower than that in the influent except several exceptions, such as Clostridium and Mycobacterium . These results

100

Fig. 1 Percentage of pyrotags assigned at each taxonomic level

90

70 60 50 40

domain phylum class order family genus

30 20

IN2011-08 IN2011-09 IN2011-10 IN2011-11 IN2011-12 IN2012-01 IN2012-02 IN2012-03 IN2012-04 IN2012-05 IN2012-06 IN2012-07 IN2012-07R AS2011-08 AS2011-09 AS2011-10 AS2011-11 AS2011-12 AS2012-01 AS2012-02 AS2012-03 AS2012-04 AS2012-05 AS2012-06 AS2012-07 AS2012-07R EF2011-08 EF2011-09 EF2011-10 EF2011-11 EF2011-12 EF2012-01 EF2012-02 EF2012-03 EF2012-04 EF2012-05 EF2012-06 EF2012-07 EF2012-07R

Percentage (%)

80

Influent

Activated sludge

Effluent

Appl Microbiol Biotechnol (2014) 98:3317–3326

3321 80

demonstrated that most bacterial pathogens were effectively treated by the Shatin WWTP. Percentage (%)

At phylum level, minor monthly variations of bacterial community were observed in the influent, activated sludge, or effluent of the Shatin WWTP (Fig. S2). At genus level, less monthly variations were found for the dominant genera within the same section samples, but more variations were observed for the non-abundant genera (Fig. S4). Therefore, the major bacterial community in the influent, activated sludge, and effluent was relatively stable in the Shatin WWTP throughout the year investigated. The PCoA profile, based on the most abundant 100 genera, showed a grouped pattern for the same section samples but a dispersed pattern for the different section samples (Fig. 5). The bacterial community of the activated sludge shared a close distance with the effluent, but a long distance with the influent, suggesting that the bacterial population of activated sludge was less affected by the influent bacteria, but it had a major impact on the effluent bacteria.

60 50 40 30

Max 75% quantile Mean 50% quantile 25% quantile Min

20 10 0 Firmi Proteo Bacter Actino TM7 cutes bacteria oidetes bacteria

Nitro spira

5 4

Percentage (%)

Bacterial community monthly shift in three sections of the Shatin WWTP

Influent Activated sludge Effluent

70

3 2 1

Discussion

Fuso Synerg Cyano OD1 bacteria istetes bacteria

Chlam ydiae

SR1

1.6 1.4

Percentage (%)

In Hong Kong, human sewage is the major source of WWTP influent since more than 7 million people live in this small region with no industry or agriculture and where wastewater and stormwater are collected separately by sewer systems. As shown in the results, the influent sewage bacterial profile reflected the human microbiome, even though the raw sewage had a retention time of several hours in the sanitary sewers before reaching the Shatin WWTP. This finding is consistent with our hypothesis. Moreover, the observation that the human gut bacterial community is the major component of influent sewage bacteria also conforms to the fact. Hence, domestic sewage in Hong Kong represents a composite sampling of a large number of urban residents, which can serve as a representative sample for studying the human microbiome. Further investigation showed that most of the influent sewage bacteria disappeared in the activated sludge and effluent. This is understandable since the gut environment is anaerobic but the activated sludge environment is completely aerobic. An unpublished communication, which employed propidium monoazide to discriminate alive and dead bacterial cells during DNA preparation, also demonstrated that most dead bacterial cells observed in activated sludge were derived from human gut. In the past few years, the NIH Human Microbiome Project has increasingly enlarged our knowledge of highly complex structure and function of the human microbiome due to the development and improvement of high-throughput sequencing technologies. Related studies have provided new insights

0

1.2 1.0 0.8 0.6 0.4 0.2 0.0 Acido Verruco Chlor Plancto Spiro Chlo bacteria microbia oflexi mycetes chaetes robi

Fig. 2 Abundance at phylum level. Pyrotags were compared in three groups of influent, activated sludge, and effluent and visualized using a box chart statistically. The detailed data were plotted using stacked column in Fig. S2

into the microbial composition, function, and variation among different individuals (Arumugam et al. 2011; Huttenhower et al. 2012; Schloissnig et al. 2013; Yatsunenko et al. 2012), of which the findings were compared with this study. As shown in the results, the human sewage microbiome revealed some common features with the human microbiome. Many efforts have been made to look for novel representative

Si m ka Ca n T Ac te hi ia( o l an lib th 76) th ac rix o te (7 Hy ple riu 7) dr uri m og ba (78 en cte ) op r( ha 79) ga Ps (8 yc Ca h Gp 0) r te ob 4(8 ni ba act 1) e M cte r(8 eg ri 2 as um ) ph (8 3 a SR G era ) 1_ alli (84 ge co ) n la Al era (85 ka _I ) S( La lifl 8 ct exu 6) oc s Pa oc (87 rv cu ) ul a s(8 So rcu 8) l l An imo a(89 Ig ae na ) na ro s( vi si 90 ba nu ) ct s( Pa eri 91) st um eu (9 2 r M Vei ella ) es llo ( or ne 93) hi zo lla( bi 94 ) Di um al (9 5 i s ) K t Ca leb er(9 s 6 Br ste iell ) ev lla a(9 un nie 7) di lla Fr mo (98 an na ) s ci se (99 lla ) (1 00 )

Ac et oa na er ob iu Ilu Th m( m au 51) at er a o Ch ba (52 lo cte ) ro r ph (53 ) Sh yta Ae ine (54 Er ) ro lla ys m (5 ip el Do ona 5) o Ps tr kdo s( eu ich ne 56) do ac lla e (5 c Cl hro ae_ 7) os ba IS ( t c Co rid tru 58) ry ium m( ne X 59 Fe bac VII ) rr te I(6 u 0 r Th gin ium ) i al as bac (61 Ba so ter ) ci mo (62 lla n ) rio as ph (63 ) y E ta( Te ns 64) rr ife im r( Ac on 65) Fi ido as ( lo m vo 66 Ac icr rax ) tin ob (67 ob ium ) a Ae cul (68 qu um ) Tu ori (69 v ) r G icib ita( em ac 70 ) in ic ter( oc 71 Ch ry La cus ) se br ( ob en 72) ac zia te (7 riu 3) De m( vo 74 si ) a( 75 )

Percentage (%) An a Cl ero os s tr tip id e iu s(2 Ro m S 6) se S( 2 G bur 7) em ia Ba m (28 ct ige ) er r oi (29 de ) s M et Ha (30 hy li ) lo ea( cy 31 st ) H is M oe (32 eg fl ) e La am a(3 ct on 3) ob a ac s(3 Hy 4 il ph Go lus ) om rdo (35 ic nia ) ro bi (36 ) u Ba m( 3 Ca uld 7) Ps ld ia eu ili (38 ) d n Es N om ea( o ch on 39 c er ar a ) ic di s(4 o h Cl ia/ ide 0) oa Sh s( 4 ci i ba gel 1) Hy cte la(4 r 2 Fl pho ium ) av m ob on (43 a a ) Co cte s(4 pr riu 4) o Am co m(4 ar ccu 5) Er ico s( 4 yt hr ccu 6) ob s( 4 a Le c 7 pt ter ) on (4 Le em 8) w a(4 in el 9) la (5 0)

Percentage (%) ep to c Pa oc r TM a cus c 7_ oc (1) ge cu ne s( ra 2) _ M yc B IS( ob lau 3) a Cl c tia os ter (4) tr ium id ( i Ar um 5) co X La ba I(6 ch A c ) no zo ter( sp ar 7) Ac ira cus c ( i Bi net ea_ 8) fid ob IS ob ac (9 ) t En act er( Fa ter eriu 10) ec oc m ( al o ib cc 11) ac us Rh te (1 od riu 2) ob m( ac 13 Ru Nitr ter ) m os (14 Ph ino pi ) yl c ra lo oc (15 ba cu ) ct s er (16 iu m ) Te V (1 Br tra ibri 7) ad sp o( yr ha 18) hi er zo a( bi 19 um ) U Do (20 O rub re ) D1 u a( _g rue 21) e l Co ne la(2 ne ra_ 2) xi IS ba (2 M cte 3) or ax r(24 el ) la (2 5)

St r

Percentage (%)

Fig. 3 Abundance for the most dominant 100 genera (no. 1–100 bacterial genera). Pyrotags were compared in three groups of influent, activated sludge, and effluent. The mean abundance of each genus was plotted using drop line. The detailed data were displayed using a heat map in Fig. S4

Percentage (%)

3322 Appl Microbiol Biotechnol (2014) 98:3317–3326

14

12

10

0.5

Influent

Activated sludge

No. 1-25 bacterial genera

8

Effluent

6

4

2

0

1.6

1.2

No. 26-50 bacterial genera

0.8

0.4

0.0

No. 51-75 bacterial genera

0.4

0.3

0.2

0.1

0.0

0.3

No. 76-100 bacterial genera

0.2

0.1

0.0

Appl Microbiol Biotechnol (2014) 98:3317–3326

3323

indicator bacteria to monitor sewage pollution and assess water quality (McLellan et al. 2010; McLellan et al. 2013; Newton et al. 2011; Wery et al. 2010). However, it is impossible to find one or several ideal indicator bacteria to represent the actual composition of sewage bacteria. Nevertheless, the influent sewage bacterial profile of this study may serve as a composite indicator for environmental detection and evaluation of sewage pollution because it can truly reflect the human microbiome. As shown in Fig. 4, the overall diversity and abundance of bacterial pathogens decreased significantly from the influent to the activated sludge. The most possible explanation is that these pathogens are dead or unable to grow in a highly aerobic environment; thereby, their abundance is becoming lower and lower in the activated sludge. However, compared to activated sludge, a slightly elevated level of bacterial pathogens was

Fig. 4 Abundance for potentially pathogenic bacteria profiled at genus level. All values were log10-transformed for plotting. Scales of ND, −4, −3, −2, and −1 indicated the abundance of 0, 0.01, 0.1, 1, and 10 %, respectively. Genera with pink fonts showed medically significant pathogens. Genera with yellow highlights represented highly abundant genera including pathogenic species

observed in the effluent. This is not surprising because most bacterial pathogens may live in free cells and cannot be settled effectively during activated sludge sedimentation; thus, they are relatively concentrated in the effluent. However, the use of different methods for sample pretreatment (i.e., centrifugation vs filtration) may also contribute to this result. As currently practiced, it is difficult to obtain species-level identification using 16S rRNA gene pyrosequencing and analysis largely because of the potential mis-annotation. In most cases, bacterial pathogens should be specifically identified at species level; however, they were profiled at genus level in this study because the genus is the lowest taxonomic level given by the RDP Classifier. Also, it is not feasible to make species classification relying only on the 16S partial sequence such as the V3–V4 region used in this study. 16S

Abiotrophia Achromobacter Acidaminococcus Acinetobacter Actinobacillus Actinomyces Aerococcus Aeromonas Anaerococcus Arcobacter Bacillus Bacteroides Bifidobacterium Bilophila Bordetella Brevundimonas Brucella Burkholderia Chryseobacterium Citrobacter Clostridium Collinsella Comamonas Corynebacterium Coxiella Eggerthella Enterococcus Escherichia Eubacterium Fibrobacter Finegoldia Francisella Fusobacterium Gemella Gordonia Granulicatella Haemophilus Klebsiella Lactobacillus Legionella Leifsonia Leptotrichia Megamonas Megasphaera Mogibacterium Moraxella Mycobacterium Neisseria Ochrobactrum Orientia Pantoea Pasteurella Peptococcus Peptostreptococcus Prevotella Pseudomonas Pseudonocardia Psychrobacter Rhodococcus Rickettsia Rothia Ruminococcus Salmonella Sebaldella Selenomonas Shigella Sphingomonas Staphylococcus Stenotrophomonas Streptobacillus Streptococcus Suttonella Tsukamurella Veillonella Vibrio

-0.5

-1

-2

-3

-4 ND

Detected genera

Influent

Activated sludge

Effluent

3324

Appl Microbiol Biotechnol (2014) 98:3317–3326 0.3

PCo2 (15.9% of total variation)

Fig. 5 Principal coordinate analysis based on abundance for the 100 most dominant genera

Activated sludge

0.2

2011-08 2011-09 2011-10

0.1

2011-11 2011-12

0.0

2012-01 Influent

-0.1

2012-02 2012-03

Effluent

2012-04 2012-05

-0.2

2012-06 2012-07

-0.3

2012-07R -0.4 -0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

PCo1 (46.0% of total variation)

rRNA gene coupled with other fingerprints such as virulence factor and clade-specific marker gene can make an accurate species identification (Cai and Zhang 2013). However, such analysis is only suitable to metagenomic data sets and not applicable to 16S pyrotags. Although it was not possible to identify pathogens at species level in this study, the data provided a full-scale evaluation of bacterial pathogens in the Shatin WWTP. In this study, there is a limitation in that we only profiled the bacterial abundance in the WWTP based on the sequenced 16S rRNA gene but not cell number. However, there is currently no accurate way for their conversion because the 16S rRNA gene copy number from different bacterial species varies significantly (from single copy to 15 copies). There is another limitation in that we did not conduct canonical correlation analysis between bacterial community and environmental factors, but we believe that some environmental factors such as temperature and salinity should contribute variations to certain bacteria in activated sludge and effluent, but not to influent bacteria since most of them were derived from human gut as demonstrated in this study. Streptococcus (no. 1 bacterial genus in Fig. 3, averagely accounting for 13.1 %) was detected as the most dominant genus in the influent sewage, which was also found with a very high abundance in human oral cavity but at an extremely low level in the human gut (Arumugam et al. 2011; Huttenhower et al. 2012). This low level of Streptococcus found in the human gut is because members of the fecal Streptococcus have been reclassified and assigned to the genus Enterococcus (Facklam 2002; Kohler 2007). For example, Streptococcus faecalis , Streptococcus faecium , Streptococcus durans , and Streptococcus avium are now Enterococcus faecalis, Enterococcus faecium , Enterococcus durans, and Enterococcus avium, respectively. As shown in Fig. 3, Enterococcus was detected as a dominant genus in the

influent sewage, with an average abundance of 3.5 %, all of which were previously known as fecal Streptococcus . Therefore, this high level of Streptococcus identified in this study was derived from the oral cavity but not from the gut. Most Streptococcus species are non-pathogens or opportunistic pathogens, but some of them can be seriously pathogenic, such as Streptococcus agalactiae , Streptococcus pneumoniae, Streptococcus pyogenes, etc. Fortunately, the abundance of Streptococcus was decreased dramatically in the activated sludge and effluent, indicating that sewage treatment reduced the pathogenic risks of Streptococcus greatly. Blautia (no. 4 bacterial genus in Fig. 3) was detected as the second dominant genus, with 8.9 % average abundance in the influent sewage. A recent report showed that sewage Blautia was proposed to be specifically derived from the human feces but not from the animal feces (McLellan et al. 2013). Hence, the occurrence and high abundance of Blautia detected in this study suggested that human gut bacteria were present abundantly in the influent sewage. This observation is also consistent with the finding discussed earlier. Paracoccus (no. 2 bacterial genus in Fig. 3) was detected as the third dominant genus in the influent sewage with an average abundance of 7.3 %. It should be noted that this genus was not prominent in the human microbiome except as found in several case studies (Funke et al. 2004; Machado-Ferreira et al. 2012; Pandey et al. 2012). Most identified Paracoccus species are instead environmental isolates. For instance, several novel species are identified from South China Sea such as Paracoccus halophilus and Paracoccus beibuensis (Liu et al. 2008; Zheng et al. 2011). The finding of abundant Paracoccus in the influent sewage may be derived from the marine environments because seawater is used to flush the toilet instead of freshwater in Hong Kong. Further investigation is needed to clarify this issue. It is worth noting that Paracoccus was detected as a very stable genus in the

Appl Microbiol Biotechnol (2014) 98:3317–3326

activated sludge system (Zhang et al. 2012), of which some species were involved in denitrification and sulfur oxidation (Baumann et al. 1996; Friedrich et al. 2000). Mycobacterium tuberculosis is a medically significant pathogen causing human tuberculosis. The genus of Mycobacterium (no. 5 bacterial genus in Fig. 3) was detected with a low abundance in the influent, but at a much higher abundance in both of the activated sludge and effluent, which is consistent with the result obtained from metagenomic analysis (Cai and Zhang 2013). Although the species M. tuberculosis has been concluded not to be present in the Shatin WWTP, the concern on this issue still remains because M. tuberculosis-like species may be predictive of potential pathogenic risks (Cai and Zhang 2013). In this study, we employed massively parallel pyrosequencing of the 16S rRNA gene to detect and monitor the human sewage microbiome in the influent, activated sludge, and effluent from the largest WWTP of Hong Kong. To our knowledge, this is the first effort to use this technique to conduct this type of investigation in a systematic manner. The findings of this study are comprehensive and provide the basis for understanding the fate of the human sewage microbiome in WWTPs. Moreover, the pyrosequencing technique is a powerful tool applicable for the detection of sewage pollution in similar investigations where comparisons with the results obtained from this study are readily made.

Acknowledgments Dr. Lin Cai thanks The University of Hong Kong for the Postdoctoral Fellowship. The authors wish to thank Agnes Chan and Wilson Chan for their technical support and service on 454 pyrosequencing. The authors also acknowledge the Research Grants Council of Hong Kong for the financial support of this study (HKU7201/11E).

References Ahmed W, Richardson K, Sidhu JPS, Toze S (2012) Escherichia coli and Enterococcus spp. in rainwater tank samples: comparison of culturebased methods and 23S rRNA gene quantitative PCR assays. Environ Sci Technol 46(20):11370–11376 Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR, Fernandes GR, Tap J, Bruls T, Batto JM, Bertalan M, Borruel N, Casellas F, Fernandez L, Gautier L, Hansen T, Hattori M, Hayashi T, Kleerebezem M, Kurokawa K, Leclerc M, Levenez F, Manichanh C, Nielsen HB, Nielsen T, Pons N, Poulain J, Qin JJ, Sicheritz-Ponten T, Tims S, Torrents D, Ugarte E, Zoetendal EG, Wang J, Guarner F, Pedersen O, de Vos WM, Brunak S, Dore J, Weissenbach J, Ehrlich SD, Bork P, Consortium M (2011) Enterotypes of the human gut microbiome. Nature 473(7346):174–180 Baker GC, Smith JJ, Cowan DA (2003) Review and re-analysis of domain-specific 16S primers. J Microbiol Meth 55(3):541–555 Baumann B, Snozzi M, Zehnder AJ, Van Der Meer JR (1996) Dynamics of denitrification activity of Paracoccus denitrificans in continuous culture during aerobic–anaerobic changes. J Bacteriol 178(15): 4367–4374

3325 Cai L, Ye L, Tong AH, Lok S, Zhang T (2013) Biased diversity metrics revealed by bacterial 16S pyrotags derived from different primer sets. PLoS One 8(1):e53649 Cai L, Zhang T (2013) Detecting human bacterial pathogens in wastewater treatment plants by a high-throughput shotgun sequencing technique. Environ Sci Technol 47(10):5433–5441 Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Pena AG, Goodrich JK, Gordon JI, Huttley GA, Kelley ST, Knights D, Koenig JE, Ley RE, Lozupone CA, McDonald D, Muegge BD, Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ, Walters WA, Widmann J, Yatsunenko T, Zaneveld J, Knight R (2010) QIIME allows analysis of high-throughput community sequencing data. Nat Methods 7(5):335–336 Chen GX, Walker SL (2012) Fecal indicator bacteria transport and deposition in saturated and unsaturated porous media. Environ Sci Technol 46(16):8782–8790 Claesson MJ, Wang QO, O'Sullivan O, Greene-Diniz R, Cole JR, Ross RP, O'Toole PW (2010) Comparison of two next-generation sequencing technologies for resolving highly complex microbiota composition using tandem variable 16S rRNA gene regions. Nucleic Acids Res 38(22):e200 Cole JR, Wang Q, Cardenas E, Fish J, Chai B, Farris RJ, Kulam-SyedMohideen AS, McGarrell DM, Marsh T, Garrity GM, Tiedje JM (2009) The Ribosomal Database Project: improved alignments and new tools for rRNA analysis. Nucleic Acids Res 37(Database issue): D141–D145 Dubinsky EA, Esmaili L, Hulls JR, Cao YP, Griffith JF, Andersen GL (2012) Application of phylogenetic microarray analysis to discriminate sources of fecal pollution. Environ Sci Technol 46(8):4340–4347 Facklam R (2002) What happened to the Streptococci: overview of taxonomic and nomenclature changes. Clin Microbiol Rev 15(4): 613–630 Ferguson AS, Layton AC, Mailloux BJ, Culligan PJ, Williams DE, Smartt AE, Sayler GS, Feighery J, McKay LD, Knappett PS, Alexandrova E, Arbit T, Emch M, Escamilla V, Ahmed KM, Alam MJ, Streatfield PK, Yunus M, van Geen A (2012) Comparison of fecal indicators with pathogenic bacteria and rotavirus in groundwater. Sci Total Environ 431:314–322 Friedrich CG, Quentmeier A, Bardischewsky F, Rother D, Kraft R, Kostka S, Prinz H (2000) Novel genes coding for lithotrophic sulfur oxidation of Paracoccus pantotrophus GB17. J Bacteriol 182(17): 4677–4687 Funke G, Frodl R, Sommer H (2004) First comprehensively documented case of Paracoccus yeei infection in a human. J Clin Microbiol 42(7):3366–3368 Gordon KV, Brownell M, Wang SY, Lepo JE, Mott J, Nathaniel R, Kilgen M, Hellein KN, Kennedy E, Harwood VJ (2013) Relationship of human-associated microbial source tracking markers with Enterococci in Gulf of Mexico waters. Water Res 47(3):996–1004 Guo F, Zhang T (2013) Biases during DNA extraction of activated sludge samples revealed by high throughput sequencing. Appl Microbiol Biotechnol 97(10):4607–4616 Huttenhower C, Gevers D, Knight R, Abubucker S, Badger JH et al (2012) Structure, function and diversity of the healthy human microbiome. Nature 486(7402):207–214 Jeanneau L, Solecki O, Wery N, Jarde E, Gourmelon M, Communal PY, Jadas-Hecart A, Caprais MP, Gruau G, Pourcher AM (2012) Relative decay of fecal indicator bacteria and human-associated markers: a microcosm study simulating wastewater input into seawater and freshwater. Environ Sci Technol 46(4):2375–2382 Kohler W (2007) The present state of species within the genera Streptococcus and Enterococcus . Int J Med Microbiol 297(3): 133–150 LeBlanc JG, Milani C, de Giori GS, Sesma F, van Sinderen D, Ventura M (2013) Bacteria as vitamin suppliers to their host: a gut microbiota perspective. Curr Opin Biotech 24(2):160–168

3326 Lee JE, Lee S, Sung J, Ko G (2011) Analysis of human and animal fecal microbiota for microbial source tracking. ISME J 5(2):362–365 Liu ZP, Wang BJ, Liu XY, Dai X, Liu YH, Liu SJ (2008) Paracoccus halophilus sp. nov., isolated from marine sediment of the South China Sea, China, and emended description of genus Paracoccus Davis 1969. Int J Syst Evol Microbiol 58(Pt 1):257–261 Machado-Ferreira E, Piesman J, Zeidner NS, Soares CA (2012) A prevalent alpha-proteobacterium Paracoccus sp. in a population of the Cayenne ticks (Amblyomma cajennense ) from Rio de Janeiro, Brazil. Genet Mol Biol 35(4):862–867 McLellan SL, Huse SM, Mueller-Spitz SR, Andreishcheva EN, Sogin ML (2010) Diversity and population structure of sewage-derived microorganisms in wastewater treatment plant influent. Environ Microbiol 12(2):378–392 McLellan SL, Newton RJ, Vandewalle JL, Shanks OC, Huse SM, Eren AM, Sogin ML (2013) Sewage reflects the distribution of human faecal Lachnospiraceae. Environ Microbiol 15(8):2213–2227 Murugan K, Prabhakaran P, Al-Sohaibani S, Sekar K (2012) Identification of source of faecal pollution of Tirumanimuttar River, Tamilnadu, India using microbial source tracking. Environ Monit Assess 184(10):6001–6012 Newton RJ, VandeWalle JL, Borchardt MA, Gorelick MH, McLellan SL (2011) Lachnospiraceae and Bacteroidales alternative fecal indicators reveal chronic human sewage contamination in an urban harbor. Appl Environ Microbiol 77(19):6972–6981 Pandey PK, Verma P, Kumar H, Bavdekar A, Patole MS, Shouche YS (2012) Comparative analysis of fecal microflora of healthy full-term Indian infants born with different methods of delivery (vaginal vs cesarean): Acinetobacter sp. prevalence in vaginally born infants. J Biosci 37(6):989–998 Pickering AJ, Julian TR, Marks SJ, Mattioli MC, Boehm AB, Schwab KJ, Davis J (2012) Fecal contamination and diarrheal pathogens on surfaces and in soils among Tanzanian households with and without improved sanitation. Environ Sci Technol 46(11):5736–5743 Prakash S, Tomaro-Duchesneau C, Saha S, Cantor A (2011) The gut microbiota and human health with an emphasis on the use of microencapsulated bacterial cells. J Biomed Biotechnol 2011: 981214 Ravel J, Gajer P, Abdo Z, Schneider GM, Koenig SS, McCulle SL, Karlebach S, Gorle R, Russell J, Tacket CO, Brotman RM, Davis CC, Ault K, Peralta L, Forney LJ (2011) Vaginal microbiome of reproductive-age women. Proc Natl Acad Sci U S A 108(Suppl 1): 4680–4687 Schloissnig S, Arumugam M, Sunagawa S, Mitreva M, Tap J, Zhu A, Waller A, Mende DR, Kultima JR, Martin J, Kota K, Sunyaev SR,

Appl Microbiol Biotechnol (2014) 98:3317–3326 Weinstock GM, Bork P (2013) Genomic variation landscape of the human gut microbiome. Nature 493(7430):45–50 Shanks OC, Newton RJ, Kelty CA, Huse SM, Sogin ML, McLellan SL (2013) Comparison of the microbial community structures of untreated wastewaters from different geographic locales. Appl Environ Microbiol 79(9):2906–2913 Shanks OC, Sivaganesan M, Peed L, Kelty CA, Blackwood AD, Greene MR, Noble RT, Bushon RN, Stelzer EA, Kinzelman J, Anan'eva T, Sinigalliano C, Wanless D, Griffith J, Cao YP, Weisberg S, Harwood VJ, Staley C, Oshima KH, Varma M, Haugland RA (2012) Interlaboratory comparison of real-time PCR protocols for quantification of general fecal indicator bacteria. Environ Sci Technol 46(2): 945–953 Sidhu JP, Hodgers L, Ahmed W, Chong MN, Toze S (2012) Prevalence of human pathogens and indicators in stormwater runoff in Brisbane, Australia. Water Res 46(20):6652–6660 Staley C, Reckhow KH, Lukasik J, Harwood VJ (2012) Assessment of sources of human pathogens and fecal contamination in a Florida freshwater lake. Water Res 46(17):5799–5812 Unno T, Jang J, Han D, Kim JH, Sadowsky MJ, Kim OS, Chun J, Hur HG (2010) Use of barcoded pyrosequencing and shared OTUs to determine sources of fecal bacteria in watersheds. Environ Sci Technol 44(20):7777–7782 Wang Q, Garrity GM, Tiedje JM, Cole JR (2007) Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol 73(16): 5261–5267 Wery N, Monteil C, Pourcher AM, Godon JJ (2010) Human-specific fecal bacteria in wastewater treatment plant effluents. Water Res 44(6):1873–1883 Woolhouse M, Gowtage-Sequeria S, Evans B (2006) T16: quantitative analysis of the characteristics of emerging and re-emerging human pathogens. Centre for Infectious Diseases, University of Edinburgh Yatsunenko T, Rey FE, Manary MJ, Trehan I, Dominguez-Bello MG, Contreras M, Magris M, Hidalgo G, Baldassano RN, Anokhin AP, Heath AC, Warner B, Reeder J, Kuczynski J, Caporaso JG, Lozupone CA, Lauber C, Clemente JC, Knights D, Knight R, Gordon JI (2012) Human gut microbiome viewed across age and geography. Nature 486(7402):222–227 Zhang T, Shao MF, Ye L (2012) 454 pyrosequencing reveals bacterial diversity of activated sludge from 14 sewage treatment plants. ISME J 6(6):1137–1147 Zheng Q, Wang Y, Chen C, Xia X, Fu Y, Zhang R, Jiao N (2011) Paracoccus beibuensis sp. nov., isolated from the South China Sea. Curr Microbiol 62(3):710–714