microbial communities of aerobic granules- granulation mechanisms

8
Microbial communities of aerobic granules: Granulation mechanisms Yi Lv a , Chunli Wan a , Duu-Jong Lee a,b,c,, Xiang Liu a , Joo-Hwa Tay d a Department of Environmental Science and Engineering, Fudan University, 220 Handan Road, Shanghai 200433, China b Department of Chemical Engineering, National Taiwan University of Science and Technology, Taipei 106, Taiwan c Department of Chemical Engineering, National Taiwan University, Taipei 106, Taiwan d Department of Civil Engineering, University of Calgary, Calgary, Canada highlights Microbial communities of mature and growing aerobic granules were probed. Sliced sample study showed a spherical core with anaerobic Rhodocyclaceae. High-throughput sequencing showed that flocs were transited to young granules. Microbial community data supported deterministic mechanism for granulation. article info Article history: Received 10 June 2014 Received in revised form 30 June 2014 Accepted 1 July 2014 Available online 8 July 2014 Keywords: Slicing PCR–DGGE Spatial distribution Microbial community abstract Aerobic granulation is an advanced biological wastewater treatment technology. This study for the first time identified the microbial communities of sliced samples of mature granules by polymerase chain reaction (PCR) amplification and denaturing gradient gel electrophoresis (DGGE) technique and those of whole growing granules by high-throughput sequencing technique. The sliced sample study revealed that mature granules have a spherical core with anaerobic Rhodocyclaceae covered by an outer spherical shell with both aerobic and anaerobic strains. The growing granule study showed that the flocculated flocs were first transited to young granules with increased abundances of Flavobacteriaceae, Xanthomo- nadaceae, Rhodobacteraceae and Microbacteriaceae, then the abundances of anaerobic strains were increased owing to the formation of anaerobic core. Since the present granules were cultivated from floc- culated flocs, the microbial community data suggested that granules were formed via a deterministic rather than via a random aggregation–disintegration mechanism. Ó 2014 Elsevier Ltd. All rights reserved. 1. Introduction Aerobic granulation is an emerging biotechnology for wastewa- ter treatment (Adav et al., 2008a). Biological cells can be aggre- gated into a compact form without a carrier in sequencing batch reactor (SBR) so the yielded granules have high biomass content, excellent settling rate, and supreme resistance to toxicity in feed wastewaters (Adav et al., 2008b). Owing to the high retention capability of slow-growing strains, the aerobic granules were applied in various industrial wastewater treatments that need syn- ergetic growths of different groups of bacteria (Rosman et al., 2014; Jemaat et al., 2014; Li et al., 2014; Wan et al., 2014a,b). Aerobic granules were generally produced by cultivating floccu- lated sludge flocs under starvation and hydrodynamic stress (Lee et al., 2010). Specific functional strains in the seed sludge would secret excess extracellular polymeric substances (EPS) for assisting granulation (Adav et al., 2009). Li et al. (2008) noted that high organic loading rates reduced microbial diversities in seed sludge and identified that b- and c-Proteobacteria and Flavobacterium were the dominating species in their glucose granules. Zhao et al. (2013) cultivated aerobic granules with piggery wastewaters at high chemical oxygen demand (COD) and nitrogen (N) loadings. These authors noted that strains Thauera, Zoogloea and heterotro- phic and autotrophic denitrifiers were present in their granules. At very high organic loading rates, Adav et al. (2009) noted strains Zoogloea resiniphila, Acinetobacter sp. clone JT2 and bacterium clone P1D1-516 retained in their aerobic granules. Winkler et al. (2013) noted from their nitrifying granules that Nitrosomonas was the dominant ammonia-oxidizing bacterium (AOB) in seed sludge while in the formed granules both Nitrosomonas and Nitrosospira http://dx.doi.org/10.1016/j.biortech.2014.07.005 0960-8524/Ó 2014 Elsevier Ltd. All rights reserved. Corresponding author at: Department of Environmental Science and Engineering, Fudan University, Shanghai, 200433, China. Tel.: +86 21 65642018; fax: +86 21 65643597. E-mail address: [email protected] (D.-J. Lee). Bioresource Technology 169 (2014) 344–351 Contents lists available at ScienceDirect Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Upload: anonymous-5l8pabb

Post on 18-Feb-2016

231 views

Category:

Documents


2 download

DESCRIPTION

Granulation Mechanisms

TRANSCRIPT

Bioresource Technology 169 (2014) 344–351

Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Microbial communities of aerobic granules: Granulation mechanisms

http://dx.doi.org/10.1016/j.biortech.2014.07.0050960-8524/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: Department of Environmental Science andEngineering, Fudan University, Shanghai, 200433, China. Tel.: +86 21 65642018;fax: +86 21 65643597.

E-mail address: [email protected] (D.-J. Lee).

Yi Lv a, Chunli Wan a, Duu-Jong Lee a,b,c,⇑, Xiang Liu a, Joo-Hwa Tay d

a Department of Environmental Science and Engineering, Fudan University, 220 Handan Road, Shanghai 200433, Chinab Department of Chemical Engineering, National Taiwan University of Science and Technology, Taipei 106, Taiwanc Department of Chemical Engineering, National Taiwan University, Taipei 106, Taiwand Department of Civil Engineering, University of Calgary, Calgary, Canada

h i g h l i g h t s

�Microbial communities of mature and growing aerobic granules were probed.� Sliced sample study showed a spherical core with anaerobic Rhodocyclaceae.� High-throughput sequencing showed that flocs were transited to young granules.� Microbial community data supported deterministic mechanism for granulation.

a r t i c l e i n f o

Article history:Received 10 June 2014Received in revised form 30 June 2014Accepted 1 July 2014Available online 8 July 2014

Keywords:SlicingPCR–DGGESpatial distributionMicrobial community

a b s t r a c t

Aerobic granulation is an advanced biological wastewater treatment technology. This study for the firsttime identified the microbial communities of sliced samples of mature granules by polymerase chainreaction (PCR) amplification and denaturing gradient gel electrophoresis (DGGE) technique and thoseof whole growing granules by high-throughput sequencing technique. The sliced sample study revealedthat mature granules have a spherical core with anaerobic Rhodocyclaceae covered by an outer sphericalshell with both aerobic and anaerobic strains. The growing granule study showed that the flocculatedflocs were first transited to young granules with increased abundances of Flavobacteriaceae, Xanthomo-nadaceae, Rhodobacteraceae and Microbacteriaceae, then the abundances of anaerobic strains wereincreased owing to the formation of anaerobic core. Since the present granules were cultivated from floc-culated flocs, the microbial community data suggested that granules were formed via a deterministicrather than via a random aggregation–disintegration mechanism.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Aerobic granulation is an emerging biotechnology for wastewa-ter treatment (Adav et al., 2008a). Biological cells can be aggre-gated into a compact form without a carrier in sequencing batchreactor (SBR) so the yielded granules have high biomass content,excellent settling rate, and supreme resistance to toxicity in feedwastewaters (Adav et al., 2008b). Owing to the high retentioncapability of slow-growing strains, the aerobic granules wereapplied in various industrial wastewater treatments that need syn-ergetic growths of different groups of bacteria (Rosman et al.,2014; Jemaat et al., 2014; Li et al., 2014; Wan et al., 2014a,b).

Aerobic granules were generally produced by cultivating floccu-lated sludge flocs under starvation and hydrodynamic stress (Leeet al., 2010). Specific functional strains in the seed sludge wouldsecret excess extracellular polymeric substances (EPS) for assistinggranulation (Adav et al., 2009). Li et al. (2008) noted that highorganic loading rates reduced microbial diversities in seed sludgeand identified that b- and c-Proteobacteria and Flavobacteriumwere the dominating species in their glucose granules. Zhao et al.(2013) cultivated aerobic granules with piggery wastewaters athigh chemical oxygen demand (COD) and nitrogen (N) loadings.These authors noted that strains Thauera, Zoogloea and heterotro-phic and autotrophic denitrifiers were present in their granules.At very high organic loading rates, Adav et al. (2009) noted strainsZoogloea resiniphila, Acinetobacter sp. clone JT2 and bacterium cloneP1D1-516 retained in their aerobic granules. Winkler et al. (2013)noted from their nitrifying granules that Nitrosomonas was thedominant ammonia-oxidizing bacterium (AOB) in seed sludgewhile in the formed granules both Nitrosomonas and Nitrosospira

Table 1Slicing of collected granules on 70 d in operation.

No. Diameter (mm) Layers (B) Distance from surface (lm)

2 7 12 0–583583–11671167–17501750–23332333–29172917–3500

4 7 10 0–700700–14001400–21002100–28002800–3500

Y. Lv et al. / Bioresource Technology 169 (2014) 344–351 345

were present in equal amounts. Liu et al. (2010) noted that the floc-culated flocs and the aerobic granules collected from the samereactor did not show microbial community difference, suggestingthat granulation posed low microbial selection pressure on theseed strains. However, Song et al. (2010) noted that the seed sludgefrom beer wastewater treatment plant with dominant strains Para-coccus sp., Devosia hwasunensi, Pseudoxanthomonas sp. can be usedto cultivate mature granules. Conversely, the seed sludge frommunicipal treatment plant with dominant strains Lactococcus raff-inolactis and Pseudomonas sp. needed longer time to cultivatemature granules. Li et al. (2010) noted that high organic loadingrate favors growth of bacterial granules over filamentous granules.Therefore, if the function strains were enriched in seed sludge, themicrobial community would experience little change over granula-tion; otherwise, significant microbial community change isexpected.

Both granules and flocculated flocs co-existed in large-scalereactor with real wastewaters (Liu et al., 2010). Short settling timeis commonly applied to facilitate washout of flocculated sludgeflocs and fine aggregates from the reactor, which is often referredto hydrodynamic selection pressure for granulation. Weissbrodret al. (2012) claimed that washout is a microbial selection processwhile Zoogloea was enriched in their dense granules and filamen-tous Burkholderiales dominated in their loose granules. Verawatyet al. (2012) added fluorescent-labeled crushed granules intolabeled flocculated flocs and noted that the flocs would attach ontothe crushed granules to minimize their washout, so shorten thegranulation time. Zhou et al. (2014) labeled the flocculated bio-mass with fluorescent microspheres in a continuous-flow bioreac-tor and noted that bioflocs detach, collide and aggregate randomlyin granulation process.

Although the layered structure of mature granules were pro-posed in literature, few reports revealed the stereological distribu-tion data of dominating strains in mature granules and in growinggranules. The mechanisms for aerobic granulation were proposedas a sequential process including intra-, inter- and multi-genericcell-to-cell attachments (Palmer et al., 2001). Liu and Tay (2002)proposed a four-step mechanistic granulation mechanism as fol-lows: (1) cell-to-cell contact; (2) initial attachment to form aggre-gates; (3) enhancement in attachment by secreted extracellularpolymeric substances (EPS); (4) hydrodynamic shear force to stabi-lize the three dimensional structure of the granule. Conversely,Zhou et al. (2014) negated the mechanistic approach and claimedthat the granulation is a random coagulation process. This studycultivated aerobic granules in acetate + propionate wastewaterusing waste activated sludge as seed sludge. Then the cultivatedgranules, slices on mature granules or whole young and maturegranules, were probed on their microbial communities by poly-merase chain reaction (PCR) amplification and denaturing gradientgel electrophoresis (DGGE) technique or by high-throughputsequencing technique. The mechanisms for granulation were dis-cussed based on the community data.

5 6 12 0–500500–10001000–15001500–20002000–25002500–3000

6 8 10 0–800800–16001600–24002400–32003200–4000

8 8 10 0–800800–16001600–24002400–32003200–4000

2. Methods

2.1. Granule cultivation

The seed sludge was collected from the reflux sludge stream of awastewater treatment plant in Shanghai, China.

A 6 cm SBR of total volume of 5 l and working volume of 2.2 lwas used. Air bubbles were supplied through an air sparger atthe bottom of the reactor with a superficial air velocity of2.95 cm s�1. The volume exchange was 0.75. The reactor was oper-ated in successive cycle and each cycle included 5 min feeding,195 min aerating, gradually decreasing settling time from 30 min

to 1 min, 5 min effluent withdrawal and 5 min idle time. The initialdissolved oxygen and pH of aeration phase were 7.52 mg l�1 and7.05, respectively. The composition of the synthetic wastewaterwas: peptone 400 mg l�1, yeast exact 250 mg l�1, NH4Cl 3.74 mM,KH2PO4 4.85 mM, CaCl2 0.27 mM, MgSO4 0.21 mM, FeSO4

0.13 mM, NaHCO3 0.15 mM. The carbon source in municipalwastewater was acetate and propionate at molar ration 3:1, givingan organic loading rate (OLR) of 1.5 kg COD m�3d�1. The test wasperformed at 29 ± 1 �C.

2.2. Granules sampling

Two batches of samples were collected. Five mature granuleswere collected on 70 d in operation. Using OCT compound toembed single granule and then placed the embedded granules at�20 �C for 30 min (Adav et al., 2010a,b), the granules were slicedlayer by layer with their size and thickness being listed in Table 1.The volume of each slice was determined by:

V ¼Z h2

h1

pðR2 � X2Þdx

where h2 denotes the distance between the distal edge of the layerto equator and h1 denotes the distance between the proximal edgeof the layer to equator. R denotes the radius of sphere and X denotesthe distance from equator of the sphere.

In the startup phase of the reactor, the seed sludge, and thegranules from early stage of size 1.5 mm to later stage of size6.5 mm were collected. The whole microbial communities of thesecollected granules were probed using high-throughput sequencingtechnology.

2.3. DNA extraction, PCR, DGGE and high-throughput sequencing

The genomic DNA of granule samples was extracted usingPower SoilTM DNA Isolation Kit (Mobio, USA). The nanodrop ND-3300 fluorescence spectrophotometer (Thermo, USA) was used toquantify extracted DNA with picogreen (SIGMA, USA) as the dye.The extracted DNA was stored at �20 �C. Subsequently, the

346 Y. Lv et al. / Bioresource Technology 169 (2014) 344–351

variable V3 region of the bacterial 16S rDNA gene sequence wasamplified by PCR with forward primer 8F and reverse primer518R. PCR amplification was performed with a BIO-RAD (USA)using a 50 ll (total volume) mixture containing 1.5U of Taq DNApolymerase, 5 ll of 10�PCR buffer minus Mg2+, 1 ll of 10 mMdNTP mixture, 1.5 ll of 50 mM MgCl2, 0.5 ll of the forward primer(10 lM) and 0.5 ll of the reverse primer (10 lM), 1 ll of the DNAextraction solution (50 mgl�1). The PCR temperature programstarted with 10 min of activation of the polymerase at 94 �C.Twenty PCR cycles were then conducted, with the first two cyclesconsisting of 1 min denaturation at 94 �C, 1 min annealing at 55 �Cand 1 min synthesis at 72 �C. With the same temperatures fordenaturation and synthesis, the annealing temperature was subse-quently decreased by 0.1 �C for each cycle since cycle 3 until reach-ing 52 �C. Finally, after 30 PCR cycles, a 10 min extension step at72 �C was performed.

The PCR-amplified DNA products were separated by DGGE on8% polyacrylamide gels with a linear gradient of 30–55% denatur-ant (100% denaturant 1=4 40% v/v formamide plus 42% w/v urea)using the DCodeTM Universal Mutation Detection System (Bio-Rad,USA). Gels were run for 9 h at 140 V in 1�TAE buffer maintainedat 60 C. Selected DNA bands from the DGGE gels were asepticallyexcised, purified by Axygen DNA extraction kit (USA) and re-amplifiedby the same PCR procedure described previously. After cloning andsequencing, the sequences were analyzed in comparison with the16S rDNA sequences in the GenBank by BLAST search (NationalCentre for Biotechnology Information, US National Library of

1 2 3 4 5 6 6' 5' 4' 3' 2' 1'0

1

2

3

4

5

6

7

8

9

Gra

y R

atio

Rhodocyclaceae Microbacteriaceae Moraxellaceae Brucellaceae Flavobacteriaceae Rhodobacteraceae Cytophagaceae

1 2 3 4 5 6 6' 5' 4' 3' 2' 1'0

5

10

15

20

25

30

Gra

y R

atio

Rhodocyclaceae Comamonadaceae Microbacteriaceae Xanthomonadaceae Moraxellaceae Flavobacteriaceae Rhodobacteraceae Sphingobacteriaceae

1 2 3 4 5 5'0

1

2

3

4

5

6

7

8

Gra

y R

atio

Rhodocyclaceae Rh Brucellaceae Morax Flavobacteriaceae M Sphingobacteriaceae

(c)

(a)

(e)

Fig. 1. Gray ratio (microbial density) of the eac

Medicine) for species identification. The relative intensity of eachDGGE band was determined by Gel-pro analyzer (MediaCybernetics,Rockville, MD, US). From the band intensity data, the gray ratio ofacquired bands can be obtained by dividing the gray value by thesample volume.

The Roche Emulsion-PCR technology was adopted to preparesingle molecular PCR product. The amplified genomic sampleswere taken for high-throughput sequencing by Ion Torrent PGM(Life Technology, New York, US).

3. Results and discussion

3.1. Image and microbial communities of sliced samples

The collected aerobic granules were smooth on surface appear-ance and intact in structure (images not shown). The equatorialslice of granule No. 8 is shown in Fig. S1. The internal core (A)was white with crowded spots. Moving outward the color turnedyellow, suggesting the presence of excess biomass. The interior ofthe granule was compact with layered structure, correlating withthe fluorescence results presented elsewhere (Adav et al.,2010a,b). One essential feature of granules is the compact interior,inconsistent with numerous granule studies which in fact handledsome forms of aggregates with loose structure.

The DGGE patterns for granules No. 2, 4, 5, 6, 8 were shown inFig. S2 with 1 and 10 corresponding to two surface slices andincreased numbers in the figure denoted the inner slices. Based

1 2 3 4 5 5' 4' 3' 2' 1'0

1

2

3

4

5

6

Gra

y R

atio

Band 1 Band 2 Band 3 Band 4 Band 5 Band 6 Band 7 Band 8 Band 9 Band 10 Band 11 Band 12 Band 13 Band 14 Band 15 Band 16

1 2 3 4 5 5' 4' 3' 2' 1'0

1

2

3

4

5

6

7

8

Gra

y R

atio

Rhodocyclaceae Microbacteriaceae Xanthomonadaceae Rhodobacteraceae Moraxellaceae Sphingobacteriaceae Flavobacteriaceae Brucellaceae

4' 3' 2' 1'

odobacteraceae ellaceae icrobacteriaceae

(b)

(d)

h layer at family level in the five granules.

Y. Lv et al. / Bioresource Technology 169 (2014) 344–351 347

on the DGGE patterns, the gray ratios were the highest at surfaceslices and were the lowest at the equatorial slices (Fig. 1). Con-versely, the Shannon indices were the highest at equatorial slicesand were the lowest at surface slices (Fig. 2).

1 2 3 4 5 6 6' 5' 4' 3' 2' 1'2.10

2.15

2.20

2.25

2.30

2.35

2.40

2.45

2.50

Shan

non

inde

x (H

')

1 2 3 4 5 6 6' 5' 4' 3' 2' 1'2.05

2.10

2.15

2.20

2.25

2.30

Shan

non

inde

x (H

')

1 2 3 4 52.40

2.45

2.50

2.55

2.60

2.65

Shan

non

inde

x (H

')

(a)

(c)

(e

Fig. 2. Shannon–Wiener diversity index of 16S

Table 2Phylogenetic sequence affiliation and similarity to the closest relative of amplified 16S rR

Band No. Family Genus

1 Sphingobacteriaceae Sphingobacterium sp. A2 Microbacteriaceae Leucobacter komagata3 Microbacteriaceae Leucobacter4 Flavobacteriaceae Flavobacterium5 Moraxellaceae Acinetobacter6 Microbacteriaceae Leucobacter7 Brucellaceae Pseudochrobactrum8 Moraxellaceae Acinetobacter9 Moraxellaceae Acinetobacter

10 Rhodobacteraceae Rhodobacter sphaeroid11 Microbacteriaceae Leucobacter komagata12 Moraxellaceae Acinetobacter13 Microbacteriaceae Leucobacter komagata14 Microbacteriaceae Leucobacter15 Microbacteriaceae Leucobacter komagata16 Rhodocyclaceae Azoarcus indigens stra17 Rhodocyclaceae Azoarcus indigens stra18 Rhodocyclaceae Thauera

The bands 2, 3, 6, 11, 13, 14, 15 in Fig. S2 belong to family Micro-bacteriaceae, bands 5, 8, 9, 12 belong to family Moraxellaceae, andband 16, 17, 18 belong to different genera of family Rhodocyclaceae.Restated, the microbial community of different granules was

1 2 3 4 5 5' 4' 3' 2' 1'

2.55

2.60

2.65

Shan

non

inde

x (H

')

1 2 3 4 5 5' 4' 3' 2' 1'1.90

1.95

2.00

2.05

2.10

2.15

2.20

Shan

non

inde

x (H

')

5' 4' 3' 2' 1'

(b)

(d)

)

rRNA gene sequences in the five granules.

NA gene sequences excised from DGGE gels of the No. 8 granule.

Similarity (%) Phylum

g8 93 Bacteroidetese strain 96 Actinobacteria

98 Actinobacteria94 Bacteroidetes99 Proteobacteria

100 Actinobacteria100 Proteobacteria100 Proteobacteria100 Proteobacteria

es strain DB803 97 Proteobacteriae 97 Actinobacteria

100 Proteobacteriae strain Z3_S_TSA4 95 Actinobacteria

99 Actinobacteriae 97 Actinobacteriain VB32 95 Proteobacteriain VB32 95 Proteobacteria

100 Proteobacteria

Fig. 3. Distributions of bacterial strains in L1–L7 at level of phylum.

348 Y. Lv et al. / Bioresource Technology 169 (2014) 344–351

mainly distributed in three phyla: Proteobacteria, Bacteroidetes andActinobacteria (Table 2).

3.2. Microbial communities in growing granules

The high-throughput sequencing yielded 7446, 208,512, 10,724,5229, 4367, 4980, 8515 high quality and effective reads for sample

Fig. 4. Distributions of bacterial strains i

L1–L7, respectively, all with >82% coverage. After quality check, theread length of majority reads were 450–600 with an average lengthabove 450 and average read number above 500. The number ofoperational taxonomic unit (OTU) identified at 97% similarity insample L1–L7 were respectively 1933, 11,400, 2083, 1231, 1197,1065 and 1710. The Chao1 indices in all samples were higher thanthe numbers of OTU, justifying the adopted 97% similarity for OTUidentification.

In L1–L7 communities, the number of bacterial phylum were 9,13, 11, 11, 10, 10 and 13, respectively, giving a total of 16 phyla(Fig. 3). All samples were dominated by Proteobacteria, Bacteroide-tes and Firmicutes, with Proteobacteria phylum being at the highestabundance followed by Bacteroidetes and then Firmicutes. Restated,the most mature L1 (with the largest diameter) was dominated byBacteroidetes and Firmicutes; while the seed sludge and younggranules L2 (with the smallest diameter) were dominated byProteobacteria.

A total of 92 families were detected in L1–L7, with 9 of whichwere the dominating families (Fig. 4a). Microbial communities ofL1–L7 were distributed at 182 genera of bacteria (Fig. 4b). The heatmap (Fig. 5) suggests that most species in L7 were higher in abun-dance than the granule samples. Hence, granulation limits thegrowth of specific bacterial strains. The clustering in Fig. 5 demon-strates that L6 differed more than did L7 from L1–L5. The betadiversity in Fig. 6 also revealed that the microbial diversity of

n L1–L7 at family (a) and genus (b).

Fig. 5. Richness heat map of microbial community at family level.

Y. Lv et al. / Bioresource Technology 169 (2014) 344–351 349

L1–L5 had high similarity, while L6 was the one departing from allthe other samples including the seed sludge.

3.3. Microbial community and granule formation mechanism

In sliced samples of mature granules, family Microbacteriaceaewas the principal species on the surface slice, which was fast-growing, obligate aerobic nitrifying bacterium (Kim and Lee,2011). Other microbes locating at surface slice were familySphingobacteriaceae (band 1), family Moraxellaceae genus Acineto-bacter (band 8), and family Rhodobacteraceae genus Rhodobacter(band 10). Family Sphingobacteriaceae is nitrification bacteriumwith degradation ability of ammonia and organic matters (Askeret al., 2008). The genus Acinetobacter has highly hydrophobic cellmembrane and can secret excess EPS, so having high self-aggrega-tion ability and adhesion capability to solid surface (Wan et al.,2014a,b). The genus Rhodobacter is an anaerobic denitrifying bacte-rium that is also able to secret EPS for assisting solid attachment.These predominant species should contribute to the formation ofaerobic granules with their secreted EPS. The family Rhodocycla-ceae had a high abundance on the equatorial slice. This observationsuggested that family Rhodocyclaceae was enriched at granule core.The b-Proteobacteria family Rhodocyclaceae genus Thauera (band 16and 18) belong to the facultative anaerobic denitrifying bacteria(Shinoda et al., 2004), which can express denitrification enzymein anoxic environment. This strain can produce excess extracellularpolysaccharides and proteins (Jiang, 2011) so should contribute togranule core strength.

On growing granule studies, correlating with the DGGE resultsin Section 3.1, although Rhodocyclaceae belongs to facultativeanaerobic denitrifying bacteria, it was present in both seed sludgeand young/mature granules. At the family level, Flavobacteriaceae,Xanthomonadaceae, Rhodobacteraceae, Microbacteriaceae were

increased significantly in abundance from L7 to L6. Growing fromL6 to L5 led to increase in abundance of Rhodobacteraceae. FromL5 to L4, the abundance of Microbacteriaceae was increased. FromL4 to L3, abundance of both Xanthomonadaceae and Microbacteria-ceae were increased. Growing from L3 to L2, the abundance of Rho-dobacteraceae was increased. From L2 to L1, the abundances ofFlavobacteriaceae and Cytophagaceae were significantly increased.Significant variation in richness diversity occurred from L6 to L5;while sample L3 and L4 had almost identical microbial communitystructures. Thauera was at the least abundance in L6 (young gran-ule) while Flavobacterium was at the most abundant in L1 (maturegranule). Thauera is a facultative anaerobic denitrifying bacterium,which increased in abundance with granule growth. Also, Rhodob-acter with facultative anaerobic denitrifying function emerged inL6. The abundances of Flavobacteriaceae and Xanthomonadaceaewere high in L6 and L7 which should be supported by the efficientoxygen transport to surface of large granules.

In the four granulation steps by Liu and Tay ((1) cell-to-cell con-tact; (2) initial attachment to form aggregates; (3) enhancement byEPS; (4) hydrodynamic packing), when the activated sludge wasthe seed sludge, granulation can start from step (3). Processes toenhance cell hydrophobicity such as periodic feast-famine condi-tions in SBR operation can stimulate step (3); high bubbling thatcompacts the granular interior can help step (4). Then the spheri-cal-shape granules should have layered structures with differentgroups of bacteria of different ecological niches residing at differ-ent depths from the surface. Conversely, if the granules wereformed by a random coagulation-detachment mechanism, withfine biological aggregates attaching and detaching on a large gran-ule matrix in a dynamic manner (Zhao et al., 2014), there should beno clear layered architecture in the granules. The present observa-tion together with Fig. S1 suggested a concentric spherical struc-ture of the granules: a spherical core with anaerobic strains

Fig. 6. Beta diversity for L1–L7. (a) Distance heat map; (b) 2-D PAC analysis; (c) 3-D PAC analysis; (d) phylogenetic tree.

350 Y. Lv et al. / Bioresource Technology 169 (2014) 344–351

(Rhodocyclaceae) covered by an outer spherical shell with both aer-obic strains (Microbacteriaceae, Sphingobacteriaceae, Moraxellaceae)and anaerobic strains (Rhodocyclaceae). Restated, the microbialcommunity data herein obtained supported the deterministicmechanism.

Owing to the compact interior of granules that limited masstransfer of substances (Tsai et al., 2008), anaerobic strains existedat the granule core that may lead to deterioration of the granulesin long-term operation (Chen et al., 2007). Adav et al. (2010a,b)showed that active strains mainly distributed at 200–250 lm depthfrom their granules’ surface. Chiu et al. (2006, 2007) noted that dis-solved oxygen was completely exhausted by active surface layer sothe interior of granules was at anaerobic condition. However,Verawaty et al. (2012) and Zhou et al. (2014) noted the occurrenceof attachment and detachment of bioflocs from the granule bio-mass. We believe under high hydrodynamic shear coagulationand disintegration of fine particles from granules occurred overthe entire granulation process, just the coagulation mechanismsshould dominate in step (1) and step (2) in Liu and Tay’s scheme.The layered structure occurs in transition from L6 to L5 should beattributable to the formation of anaerobic core in the granules,

contributed by the step (3) and step (4) in Liu and Tay’s scheme.Conversely, washout is a physical process to screen off fine aggre-gates, hence the less-flocculating strains, from the reactor. Applyingwashout to retain large aggregates (and granules) depends onwhether the operators dislike the co-existence of flocculated flocsand granules in their reactors, not as the proposal by Zhao et al.(2014) that microbial selection pressure is not a prerequisite forgranulation.

4. Conclusions

Microbial communities of aerobic granules were identified fromslices of mature granules and from growing granules. During gran-ulation, the microbial community of seed flocs decreased in diver-sity, which first rapidly shifted to a community for young granulesthat deviated from seed flocs. Then owing to formation of an intra-granular anaerobic regime, a layered structure with anaerobicRhodocyclaceae at core covered by an outer spherical shell withFlavobacteriaceae, Xanthomonadaceae, Rhodobacteraceae and Micro-bacteriaceae was formed. Granules were formed via a deterministicrather than via a random aggregation–disintegration mechanism.

Y. Lv et al. / Bioresource Technology 169 (2014) 344–351 351

Acknowledgements

This project is partially supported by NSFC Project 51176037and by National Science Council.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.biortech.2014.07.005.

References

Adav, S.S., Lee, D.J., Show, K.Y., Tay, J.H., 2008a. Aerobic granular sludge; recentadvances. Biotechnol. Adv. 26, 411–423.

Adav, S.S., Lee, D.J., Tay, J.H., 2008b. Extracellular polymeric substances andstructural stability of aerobic granule. Water Res. 42, 1644–1650.

Adav, S.S., Lee, D.J., Lai, J.Y., 2009. Proteolytic activity in stored aerobic granularsludge and structural integrity. Bioresour. Technol. 100, 68–73.

Adav, S.S., Lin, J.C.T., Yang, Z., Whiteley, C.G., Lee, D.J., Peng, X.F., Zhang, Z.P., 2010a.Stereological assessment of extracellular polymeric substances, exo-enzymes,and specific bacterial strains in bioaggregates using fluorescence experiments.Biotechnol. Adv. 28, 255–280.

Adav, S.S., Lee, D.J., Lai, J.Y., 2010b. Microbial community of acetate utilizingdenitrifiers in aerobic granules. Appl. Microbiol. Biotechnol. 85, 753–762.

Asker, D., Beppu, T., Ueda, K., 2008. Nubsella zeaxanthinifaciens gen. nov., sp. nov., azeaxanthin-producing bacterium of the family Sphingobacteriaceae isolatedfrom freshwater. Int. J. Syst. Evol. Microbiol. 58, 601–606.

Chen, M.Y., Lee, D.J., Tay, J.H., 2007. Distribution of extracellular polymericsubstances in aerobic granules. Appl. Microbiol. Biotechnol. 75, 467–474.

Chiu, Z.C., Chen, M.Y., Lee, D.J., Tay, S.T.L., Tay, J.H., Show, K.Y., 2006. Diffusivity ofoxygen in aerobic granules. Biotechnol. Bioeng. 94, 505–513.

Chiu, Z.C., Chen, M.Y., Lee, D.J., Wang, C.H., Lai, J.Y., 2007. Oxygen diffusion andconsumption in active aerobic granules of heterogeneous structure. Appl.Microbiol. Biotechnol. 75, 685–691.

Jemaat, Z., Suarez-Ojeda, M.E., Perez, J., Carrera, J., 2014. Sequentially alternatingpollutant scenarios of phenolic compounds in a continuous aerobic granularsludge reactor performing simultaneous partial nitrification and o-cresolbiodegradation. Bioresour. Technol. 161, 354–361.

Jiang, K., 2011. Genomic and molecular analysis of the exopolysaccharideproduction in the bacterium Thauera aminoaromatica MZ1T (PhDdissertation). University of Tennessee.

Kim, H.J., Lee, S.S., 2011. Leucobacter kyeonggiensis sp. nov., a new species isolatedfrom dye waste water. J. Microbiol. 49, 1044–1049.

Lee, D.J., Chen, Y.Y., Show, K.Y., Whiteley, C.G., Tay, J.H., 2010. Advances in aerobicgranule formation and granule stability in the course of storage and reactoroperation. Biotechnol. Adv. 28, 919–934.

Li, A.J., Yang, S.F., Li, X.Y., Gu, J.D., 2008. Microbial population dynamics duringaerobic sludge granulation at different organic loading rates. Water Res. 42,3552–3560.

Li, A.J., Zhang, T., Li, X.Y., 2010. Fate of aerobic bacterial granules with fungalcontamination under different organic loading conditions. Chemosphere 78,500–509.

Li, Y.M., Zou, J., Zhang, L., Sun, J., 2014. Aerobic granular sludge for simultaneousaccumulation of mineral phosphorus and removal of nitrogen via nitrite inwastewater. Bioresour. Technol. 154, 178–184.

Liu, Y., Tay, J.H., 2002. The essential role of hydrodynamic shear force in theformation of biofilm and granular sludge. Water Res. 36, 1653–1665.

Liu, Y.Q., Moy, B., Kong, Y.H., Tay, J.H., 2010. Formation, physical characteristics andmicrobial community structure of aerobic granules in a pilot-scale sequencingbatch reactor for real wastewater treatment. Enzyme Microb. Technol. 46, 520–525.

Palmer, R.J., Kazmerzak, K., Hansen, M.C., Kolenbrander, P.E., 2001. Mutualismversus independence: strategies of mixed-species oral biofilms in vitro usingsaliva as the sole nutrient source. Infect Immun. 69, 5794–5804.

Rosman, N.H., Anuar, A.N., Chelliapan, S., Md Din, M.F., Ujang, Z., 2014.Characteristics and performance of aerobic granular sludge treating rubberwastewater at different hydraulic retention time. Bioresour. Technol. 161, 155–161.

Shinoda, Y., Sakai, Y., Uenishi, H., Uchihashi, Y., Hiraishi, A., Yukawa, H., Yurimoto,H., Kato, N., 2004. Aerobic and anaerobic toluene degradation by a newlyisolated denitrifying bacterium, Thauera sp. strain DNT-1. Appl. Environ.Microbiol. 70, 1385–1392.

Song, Z.W., Pan, Y.J., Zhang, K., Ren, N.Q., Wang, A.J., 2010. Effect of seed sludge oncharacteristics and microbial community of aerobic granular sludge. J. Environ.Sci. 22, 1312–1318.

Tsai, M.W., Lee, D.J., Lai, J.Y., 2008. Mass transfer limit of fluorescent dyes duringmulticolor staining of aerobic granules. Appl. Microbiol. Biotechnol. 78, 907–913.

Verawaty, M., Pijuan, M., Yuan, Z., Bond, P.L., 2012. Determining the mechanisms foraerobic granulation from mixed seed of floccular and crushed granules inactivated sludge wastewater treatment. Water Res. 46, 761–771.

Wan, C.L., Yang, X., Lee, D.J., Liu, X., Sun, S.P., Chen, C., 2014a. Partial nitrification ofwastewaters with high NaCl concentrations by aerobic granules. Bioresour.Technol. 152, 1–6.

Wan, C.L., Yang, X., Lee, D.J., Wang, X.Y., Yang, Q.L., Pan, X.L., 2014b. Aerobicgranulation of aggregating consortium X9 isolated from aerobic granules androle of cyclic di-GMP. Bioresour. Technol. 152, 557–561.

Weissbrodr, D.G., Lochmatter, S., Ebrahimi, S., Rossi, P., Maillard, J., Holliger, C.,2012. Bacterial selection during the formation of early-stage aerobic granules inwastewater treatment systems operated under wash-out dynamics. Front.Microbiol. 3, 332–354, art.

Winkler, M.K.H., Kleerebezem, R., de Bruin, L.M.M., Verheijen, P.J.T., Abbas, B.,Habermacher, J., van Loosdrecht, M.C.M., 2013. Microbial diversity differenceswithin aerobic granular sludge and activated sludge flocs. Appl. Microbiol.Biotechnol. 97, 7447–7458.

Zhao, Y.G., Huang, J., Zhao, H., Yang, H., 2013. Microbial community and N removalof aerobic granular sludge at high COD and N loading rates. Bioresour. Technol.143, 439–446.

Zhou, D.D., Niu, S., Xiong, Y.J., Yang, Y., Dong, S.S., 2014. Microbial selection pressureis not a prerequisite for granulation: dynamic granulation and microbialcommunity study in a complete mixing bioreactor. Bioresour. Technol. 161,102–108.