Succession of bacterial and fungal communities within biofilms of a chlorinated drinking water distribution system

Accepted Manuscript Succession of bacterial and fungal communities within biofilms of a chlorinated drinking water distribution system I. Douterelo, K.E. Fish, J.B. Boxall PII:

S0043-1354(18)30353-1

DOI:

10.1016/j.watres.2018.04.058

Reference:

WR 13755

To appear in:

Water Research

Received Date: 14 December 2017 Revised Date:

5 April 2018

Accepted Date: 25 April 2018

Please cite this article as: Douterelo, I., Fish, K.E., Boxall, J.B., Succession of bacterial and fungal communities within biofilms of a chlorinated drinking water distribution system, Water Research (2018), doi: 10.1016/j.watres.2018.04.058. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

RI PT

% relative abundance

100

90

SC 0

20

M AN U

Fungal ITS genes mm-2

Fungal-Bacterial Interactions in Drinking Water Biofilms

40

60

80

80 70 60

100

30 20

TE D

AC C

0

50 40

10 0 9 31 37 10 D14

11 12

D34

EP

Bacterial 16S rRNA genes mm-2

Days

13 15

33 16

D42

17 18 19 20 34 22 24 32 D70 D84

D52

20

40

60

Days

80

100

ACCEPTED MANUSCRIPT Succession of bacterial and fungal communities within

2

biofilms of a chlorinated drinking water distribution system

3

I. Douterelo*, K.E. Fish and J.B. Boxall

4

Authors Affiliation

5

Pennine Water Group, Department of Civil and Structural

6

Engineering, Mappin Street, University of Sheffield, Sheffield,

7

S1 3JD, UK.

8

∗Corresponding Author:

9

Isabel Douterelo

M AN U

SC

RI PT

1

E-mail: [email protected]

11

Abstract

12

Understanding the temporal dynamics of multi-species biofilms

13

in Drinking Water Distribution Systems (DWDS) is essential to

14

ensure safe, high quality water reaches consumers after it

15

passes through these high surface area reactors. This research

AC C

EP

TE D

10

16

studied the succession characteristics of fungal and bacterial

17

communities under controlled environmental conditions fully

18

representative of operational DWDS.

19

Microbial

20

complexity after one month of biofilm development but they

21

did not reach stability after three months. Changes in cell

22

numbers were faster at the start of biofilm formation and tended

communities

were

observed

to

increase

in

1

ACCEPTED MANUSCRIPT to decrease over time, despite the continuing changes in

24

bacterial community composition. Fungal diversity was

25

markedly less than bacterial diversity and had a lag in

26

responding to temporal dynamics. A core-mixed community of

27

bacteria including Pseudomonas, Massillia and Sphingomonas

28

and the fungi Acremonium and Neocosmopora were present

29

constantly and consistently in the biofilms over time and

30

conditions studied.

31

Monitoring and managing biofilms and such ubiquitous core

32

microbial communities are key control strategies to ensuring

33

the delivery of safe drinking water via the current ageing

34

DWDS infrastructure.

35

Keywords: fungal-bacterial interactions, biofilm, drinking

36

water, selection, succession

37

Highlights

SC

M AN U

TE D

Succession of multi-species biofilms was studied under

representative conditions

AC C

39

•

EP

38

RI PT

23

40

•

continuously over time

41 42

•

45

Bacteria responded rapidly while fungi had a lag in

responding to biofilm dynamics

43 44

A core fungal-bacterial community was observed

•

Core microorganisms are a potential tool to develop new control strategies

2

ACCEPTED MANUSCRIPT 1. Introduction

47

The delivery of high-quality, safe drinking water depends, to a

48

large extent, on the optimal operation of the transportation

49

infrastructure known as Drinking Water Distribution Systems

50

(DWDS). DWDS are complex pipe networks which function as

51

dynamic ecosystems where microorganisms, dominated by

52

those attached within biofilms on the inner pipe surfaces, are

53

involved in a range of processes that ultimately determine the

54

quality of the delivered water. Microbial dynamics in any

55

ecosystem

56

microorganisms and environmental factors. However, within

57

DWDS these microbial dynamics and their temporal variation

58

regarding inter-taxa biofilms remain largely unexplored.

59

Most DWDS management strategies consider only planktonic

60

microorganisms, free organisms transported in the bulk water.

61

Reinforced by legislation that only obliges water utilities to

62

analyse bulk water samples from service reservoirs and

63

customer taps and using plate counting techniques and faecal

determined

by

interactions

between

AC C

EP

TE D

are

M AN U

SC

RI PT

46

64

indicators as a measurement of failures in the system. However,

65

the majority of the microbial biomass in DWDS is found

66

attached to the inner surfaces of pipes forming microbial

67

consortiums known as biofilms, rather than in the water column

68

(Flemming 1998). Biofilms provide advantages to embedded

69

microorganisms, including sharing of nutrients and metabolic

70

products and facilitation of resistance to environmental stress 3

ACCEPTED MANUSCRIPT (Chao et al. 2015). Microorganisms within biofilms can be

72

associated with several processes occurring in DWDS. Biofilms

73

can affect directly the infrastructure via biocorrosion of metal

74

pipes (Wang et al. 2012) and/or modify general water

75

characteristics (Lehtola et al. 2004) and trigger discolouration

76

events through the mobilisation of the attached biofilm into the

77

bulk water (Husband et al. 2016). Another major concern is the

78

potential for biofilms to act as reservoirs of opportunistic

79

pathogens (Wingender & Flemming 2011; Douterelo et al.

80

2014a). The most widespread strategies used to control

81

microorganisms in the bulk water of DWDS is the quality of

82

the water as it leave the treatment works and / or the use of a

83

disinfectant residual in the bulk water, mainly chlorine or

84

chloramine. The efficiency of these strategies to control biofilm

85

formation and properties is limited (Schwering et al. 2013).

86

There are no standard methods to monitor or control the growth

87

of biofilms in DWDS and, despite their importance, biofilms

88

are normally neglected when management and control

AC C

EP

TE D

M AN U

SC

RI PT

71

89

strategies are designed.

90

Biofilm formation is a successional process (Martiny et al.

91

2003; Prest et al. 2016) beginning when free-living

92

microorganisms colonize and attach to pipe surfaces, growing

93

whilst generating extracellular polymeric substances (EPS) and

94

modifying the pipe environment in a way that supports or

95

excludes the incorporation of other microorganisms into the

4

ACCEPTED MANUSCRIPT 96

biofilm. However, most of the studies reporting data on the

97

microbiology of DWDS occur over relatively short time frames

98

(Simoes et al. 2010; Douterelo et al. 2013)

99

mainly the temporal-spatial dynamics in bulk water samples

100

(Hoefel et al. 2005; van der Wielen & van der Kooij 2010;

101

Roeselers et al. 2015) or biofilm growth in reactors (Gomes et

102

al. 2014; Revetta et al. 2016). Improving our understanding of

103

biofilm behaviour in these systems requires long-term series

104

data of the attached microbial phase.

105

Commonly overlooked in DWDS studies is the fact that

106

biofilms

107

microorganisms. Microbiologists working in DWDS often

108

focus on one taxonomic group, predominately bacteria (Revetta

109

et al. 2010; Liu et al. 2014a; Vaz-Moreira et al. 2017) or a

110

specific group of bacteria (Gomez-Alvarez et al. 2016) despite

111

evidence that other microbial taxa such as fungi, viruses and

112

amoeba are present in biofilms (Wingender & Flemming 2011;

113

Buse et al. 2013; Ashbolt 2015; Fish et al. 2015; Douterelo et

M AN U

formed

by

a

complex

amalgamate

of

AC C

EP

TE D

are

SC

RI PT

and evaluate

114

al. 2016a). Knowledge on the interactions between different

115

microbial strains in actual biofilms is still at its infancy,

116

understanding how the full microbial consortium forms, how

117

different microbial strains interact, when these interactions start

118

and who the main drivers of change are remains largely

119

unknown. Advances in microscopy and molecular methods are

5

ACCEPTED MANUSCRIPT now available that allow a better understanding of microbial

121

communities dynamics including new sequencing platforms.

122

The aim of the research reported here was to study the

123

dynamics and temporal patterns of fully representative DWDS

124

biofilm community development and succession over a three-

125

month period. We particularly aimed to study correlations

126

between bacteria and fungi and biofilm cellular developmental

127

rates that have been traditionally overlooked in drinking water

128

studies. Such bacterial-fungal community characteristics and

129

their impact upon biofilm development/formation needs to be

130

understood in order to find patterns of microbial behaviour that

131

can inform modelling, and management strategies such as

132

disinfectant residual.

133

3. Methods

134

In order to meet the specific aims, a holistic approach was

135

adopted combining environmental physico-chemical analysis,

136

microscopy and molecular methods to study the process of

137

biofilm formation and maturation. These were applied to

AC C

EP

TE D

M AN U

SC

RI PT

120

138

samples obtained from the pipe wall of a full scale, temperature

139

controlled experimental pipe facility that fully replicates the

140

conditions of an operational DWDS. This study comprises a

141

range of methods that allow characterization at physiological

142

(cell volume and spread) and phylogenetic levels (gene

143

quantification and taxonomic characterization) yielding a more

144

comprehensive understanding of microbial dynamics over time. 6

ACCEPTED MANUSCRIPT 3.1 Experimental facility

146

The experiment was carried out using a DWDS test facility

147

which comprised a pipe loop fed with water from the local

148

water supply via an independent tank and pump (Figure 1),

149

which recirculated water around the pipe loop. An independent

150

system residence time of 24 h was set using a trickle-feed and

151

drain to provide representative water quality. The pipe loop

152

consists of 9.5 × 21.4 m long coils of 79.3 mm internal

153

diameter High-Density Polyethylene (HDPE) pipe, thus has a

154

total length of 203 m such that pipe surface area is dominant

155

over ancillaries (Figure 1B). HDPE was selected as it is a

156

prevalent and representative material used in DWDS world-

157

wide (WHO, 2006). The room temperature of the experimental

158

facility was set to 16 °C for all results reported here; this is

159

representative

160

temperatures in UK DWDS.

161

Before experiments commenced, the pipe loop was set to an

TE D

M AN U

SC

RI PT

145

average

spring

and

summer

water

AC C

EP

of

162

initial state by disinfection with 20 mg/l of RODOLITE H

163

(RODOL Ltd, Liverpool, UK); a solution of sodium

164

hypochlorite with less than 16% free chlorine. Then the system

165

was run at maximum flow rate (4.2 l/s) for 24 h and flushed

166

afterwards at the maximum flow rate with fresh water until the

167

levels of chlorine were similar to those of the local tap water.

168

After disinfecting the system, sterile PWG coupons (Deines et

7

ACCEPTED MANUSCRIPT al. 2010) were fitted along and around the sample length of

170

each pipe loop. The PWG coupons have an insert suitable for

171

direct microscopy observations and an outer part that allows for

172

obtaining biofilm biomass for subsequent molecular analysis

173

(Figure 1A).

174

For the experiments reported here a varied flow hydraulic

175

profile was applied based on daily patterns observed in real

176

DWDS in the UK. The regime follows a typical domestic

177

dominated diurnal pattern with night time low flow of 0.2l/s

178

and morning peak flow of 0.5l/s. This is the low varied flow

179

regime originally reported and used in (Husband et al. 2008).

180

The daily regime was repeated for a growth phase of 84 days.

181

3.2 Physico-chemical analysis of water quality

182

Triplicate bulk water samples were collected on days 0, 14, 34,

183

42, 52, 70 and 84 during the growth phase. Free chlorine was

184

measured on site at the time of sampling using a HACH

185

DR/2010 spectrophotometer. Measurements of temperature and

AC C

EP

TE D

M AN U

SC

RI PT

169

186

pH were also made on site at the time of sampling using a

187

Hanna H1991003 meter. Water samples for Total Organic

188

Carbon (TOC), iron and manganese were sent to an

189

independent accredited laboratory; AlControl Laboratories

190

(Rotherham, UK) for analysis. Turbidity was constantly

191

measured by an ATI A15/76 turbidity monitor (ATI, Delph,

192

UK) installed in the experimental facility. 8

ACCEPTED MANUSCRIPT 3.3 Confocal Laser Scanning Microscopy (CLSM)

194

CLSM was used to obtain images of biofilm and enable the

195

quantification of the cell coverage on the coupon surfaces.

196

Three coupons were studied for each sampling day from day 0

197

(used as a control) to days 14, 34, 42, 52, 70 and 84. For this

198

analysis the insert of the PWG coupon was removed and fixed

199

in 5 % formaldehyde for 24 h and then transferred to phosphate

200

buffer solution (PBS) and stored at 4 °C until analysed. After

201

fixing, the inserts were stained with 20 µmol l−1 Syto® 63, a

202

cell-permeative

203

Invitrogen, UK) for 30 min at room temperature. Imaging was

204

performed using a Zeiss LSM 510 Meta Confocal Florescent

205

Microscope at the Kroto Imaging Facility at The University of

206

Sheffield and the LSM 510 Image Examiner Software (Zeiss,

207

UK). Each insert was imaged for seven random fields of view

208

(FOV) to generate lambda z-stacks from which the cell stain

209

signal was isolated and quantified, as described in detail in Fish

210

et al (2015). The images were then processed to obtain a

M AN U

SC

RI PT

193

acid

stain,

(Molecular

Probes,

AC C

EP

TE D

nucleic

211

relative quantification of the cell coverage (i.e. volume of cells

212

and their spread) at each layer (Fish et al. 2015). The resultant

213

dataset was analysed using Python and R to calculate the

214

volume and spread of the cells (see Fish et al. 2015 for detailed

215

methods). For each sample day replication was n=21, apart

216

from Day 0 (n=20) and Day 82 (n=19) where FOV were

217

corrupted or removed as an outlier. 9

ACCEPTED MANUSCRIPT 218

Differences among days in biofilm cell volume and spread

219

were tested statistically using

220

Kruskall-Wallis test for comparison of all time points and a

221

Tukey and Kramer (Nemenyi) test for pairwise comparisons

222

between two specific time points.

223

3.4 DNA extraction

224

To extract biofilm DNA from the coupons, the outer area of

225

each coupon was brushed to remove biofilm following the

226

procedure used by Deines et al. (2010). After brushing, biofilm

227

suspensions were concentrated by filtering through 0.22-µm

228

nitrocellulose membrane filters (Millipore, Corp.) as previously

229

explained (Douterelo et al. 2013). Biofilm samples taken over

230

the course of the experiment (n=18) were then preserved in the

231

dark at −80 °C until DNA was extracted. To extract DNA, a

232

method based on proteinase K digestion followed by a standard

233

phenol/chloroform/isoamyl

234

(Neufeld et al. 2007).

alcohol

extraction

was

used

AC C

EP

TE D

M AN U

SC

RI PT

non-parametric tests; the

235

3.5 Q-PCR

236

Real-time PCR was used to quantify changes in the number of

237

bacterial 16S and fungal ITS gene copies over time (see Table

238

1 for primer details). Quantification of samples involved the

239

use of internal standard curves prepared by using a serial

240

dilution of each targeted gene amplified from biofilm samples

10

ACCEPTED MANUSCRIPT obtained from a previous experiment in the same test loop

242

facility. For each gene, a standard was generated by running

243

PCRs using the primers in Table 1 then purifying via gel-

244

purification (Qiagen Gel-Extraction Kit) and combining the

245

purified amplicons into one “DWDS biofilm” standard per

246

gene. The number of gene copies in each standard was

247

determined by quantifying the DNA content via a Nanodrop

248

8000

249

calculating gene copies using the equation:

250

Gene number = 6.023 × 1023 (copies mol-1) × concentration of

251

standard (g µl-1) MW (g mol-1).

252

Where:

253

molecular weight of the targeted gene.

254

Standards, no template controls and samples were amplified (in

255

triplicate) via qPCR reactions according to the QuantiFast

256

SYBR Green PCR kit (Qiagen, UK). Briefly, qPCR reactions

257

were 25 µl in total volume, which contained 12.5 µl of

(Thermo

Scientific, UK)

and

M AN U

SC

Spectrophotometer

RI PT

241

AC C

EP

TE D

6.023 × 1023 is Avogadro’s constant, MW is the

258

QuantiFast® SYBR® Green PCR MasterMix, 9 µl of nuclease-

259

free water (Ambion, Warrington, UK), 1.25 µl of each primer

260

(10 µM; Table 1) and 1 µl of DNA template. Real-time PCR

261

was carried out using an Applied Biosystems StepOne qPCR

262

machine and the cycling conditions advised in the Qiagen kit

263

(95 °C for 5 min, then 35 cycles of: 95 °C for 10 seconds, 60

264

°C for 30 seconds, melting curve analysis was also run for 11

ACCEPTED MANUSCRIPT bacterial gene qPCRs only), the number of gene copies was

266

determined using the StepOne software.

267

3.6 Illumina sequencing

268

Sequencing was performed by Illumina MiSeq technology with

269

the paired-end protocol by Mr DNA Laboratory (TX, USA)

270

using primers 28F (GAGTTTGATCNTGGCTCAG) and 519

271

(RGTNTTACNGCGGCKGCTG) and the fungal 18S rRNA

272

gene

273

SSUfungiF(TGGAGGGCAAGTCTGGTG)/SSUFungiR

274

(TCGGCATAGTTTATGGTTAAG) (Hume et al. 2012).

275

Paired-end reads were merged and de-noised to remove short

276

sequences, singletons, and noisy reads. Chimeras were detected

277

using UCHIME (Edgar et al. 2011) and removed from further

278

analysis. Sequences were clustered in Operational Taxonomic

279

Units (OTUs) and selected using UPARSE (Edgar 2013).

280

Taxonomic assignments were made with USEARCH global

281

alignment program (Edgar, 2013). The software PAST v3.12

282

(Hammer et al. 2008) was used to estimate Alpha-diversity at

RI PT

265

amplified

using

AC C

EP

TE D

M AN U

SC

was

283

97% sequence similarity and the Shannon diversity index,

284

Chao-I and Dominance-H were calculated. Briefly, the

285

Shannon index (H) measures diversity and indicates the

286

proportion of OTUs abundance to the whole community, this

287

index varies from 0 for communities with only a single taxon to

288

higher values (max < 5 in this study). Chao 1, is an estimate of

289

total richness and estimates total number of OTUS in a 12

ACCEPTED MANUSCRIPT community based upon the number of rare OTUs found in a

291

sample (Chao 1984). In a sample with many singletons, the

292

probability of having more undetected OTUs is higher, and the

293

Chao 1 index will estimate greater richness than for a sample

294

without rare OTUs. The evenness of a sample is a measure of

295

the distribution of the samples in a community, and

296

communities dominated by one or few species have low

297

evenness. The Dominance index (1-Simpson index) ranges

298

from 0 (all taxa are equally present) to 1 (one taxon dominates

299

the community completely) (Harper 1999). Approximate

300

confidence intervals for these indexes were computed with a

301

bootstrap procedure (default 9999) and a 95% confidence

302

interval was then calculated.

303

Bray Curtis dissimilarity matrixes at 97% sequence similarity

304

cut off were calculated and visualised using non-metric multi-

305

dimensional scaling (MDS) diagrams generated using the

306

software PAST v3.12 (Hammer et al. 2008). Stress of the non-

307

metric MDS was determined to estimate the statistical fit; stress

AC C

EP

TE D

M AN U

SC

RI PT

290

308

varies between 0 and 1, with values near 0 indicating better fit.

309

The analysis of group similarities (ANOSIM) was performed

310

based on Bray–Curtis dissimilarity distance matrices to test the

311

differences in community composition among groups of

312

samples using the software Primer6 (Clarke & Warwick 2005).

313

From this analysis gobal-R statistic values were calculated

314

showing the strength of the impact that the factors analysed had

13

ACCEPTED MANUSCRIPT on the samples, in this case time (days). Global-R values vary

316

between 0 and 1, where 1 indicates high separation of the

317

samples between levels of the factor (time) and 0 indicates no

318

separation.

319

3.7 Statistical analysis of microbiological parameters and

320

key taxonomic groups

321

Correlations

322

explored by Spearman's rank non parametric correlations using

323

SPSS Statistics 24 (IBM, USA). Alpha-diversity metrics, the

324

relative abundance of the most representative taxonomic class

325

and the measurement of cells and genes for bacteria and fungi

326

respectively were used as biological parameters in the

327

establishment of correlations.

328

4 Results

329

4.1 Water physico-chemical analysis

330

As shown in Table 2 pH values were near neutral (7.08-7.99)

331

for all the samples over time. Temperature ranged between

microbiological

parameters

were

AC C

EP

TE D

M AN U

SC

between

RI PT

315

332

14.6-15.9 °C for all samples, within an average of 15 +/- 0.4

333

°C. Free chlorine levels were slighted elevated on Day 0. This

334

was due to initial filling of the system with fresh water that has

335

a slightly higher residual than is ultimately established with the

336

24 hour residence time from the trickle drain and feed, and the

337

use of an elevated residual over night from cleaning to the start

338

of the experiment. This elevated value was less than the final 14

ACCEPTED MANUSCRIPT water quality leaving the local water treatment works, and

340

dropped to the experimental level within 24 hours. The average

341

chlorine residual in the system was then maintained at 0.42 +/-

342

0.2 mg/l.

343

Higher turbidity levels were observed on Day 0 (1.51 NTU) but

344

subsequently were consistently at an average of 0.05 +/- 0.03

345

NTU, this slight elevation is again associated with the filling of

346

the system and upstream effects from the external network

347

supplying the laboratories. TOC values were stable over time

348

ranging from 1.13 to 1.83 mg/l. Iron was the physico-chemical

349

factor that fluctuated the most, with minimum values of 9.9

350

µg/l on Day 34 to a maximum of 37.0 µg/l on Day 84.

351

Manganese levels were very low over time, generally below

352

detection limits (< 3.6 µg/l) except for Day 14 which was 0.1

353

µg/l higher.

354

4.2 Volume and spread of cells

355

Figure 2 shows the volume and spread of cells within the

AC C

EP

TE D

M AN U

SC

RI PT

339

356

DWDS biofilms over the 84-day period. Note that Figure 2A

357

and Figure 2B are the same data plotted on different vertical

358

scales in order to visualise the variability of all the samples

359

over the 3 months studied. As anticipated, the cell volume

360

within the DWDS biofilms increased from Day 0 throughout

361

the experiment as the biofilm developed (Figure 2A) and

362

significant difference were observed among data time points 15

ACCEPTED MANUSCRIPT (Kruskall Wallis, χ2 = 80.80, P-value< 0.01). This increase in

364

the cells volume was not linear, rather the volume oscillates

365

with cell volume dropping at Day 52 to volumes comparable to

366

those seen at Day 14 (Figure 2B). However, no significant

367

differences were detected using the pairwise comparison

368

Nemeyi test among Day 52 and any other time points.

369

Following this reduction in cells, the final two sample points

370

(Day 70 and 84) have the greatest median cell volume (Figure

371

2A and B) although Day 70 biofilms had a greater cell volume

372

than Day 84.

373

Figure 2C shows the 'spread' metric calculated from the biofilm

374

CLSM data. Spread was calculated as volume divided by area

375

covered (Fish et al 2015), to provide a measure of spread or

376

compactness of the biofilm, irrespective of the total volume and

377

uncertainty in defining the edges of the distribution across the

378

z-stack. While the trends, when all time points are compared,

379

were statistically significant (Kruskal-Wallis, χ2=26.411, p-

380

value < 0.05) it is notable that the data shows again an

AC C

EP

TE D

M AN U

SC

RI PT

363

381

oscillating pattern. This pattern appears to be inverse and

382

slightly lagged behind the volume data. When time points were

383

compared pairwise, statistically significant differences (Nemeyi

384

test, p-value < 0.05) were observed only between Day 70

385

samples and the other sampling days with the exception of Day

386

0 and Day 14.

16

ACCEPTED MANUSCRIPT 387

4.3 Bacterial and fungal gene quantification

388

The number of bacterial 16S rRNA gene copies (Figure 3A)

389

increased from Day 0 (no genes were quantified) to Day 42

390

reaching a maximum of 1.16 x107 16S rRNA gene copies mm-

391

2

392

over the rest of the experiment, despite this decline the Day 84

393

samples had a greater number of 16S rRNA genes than the Day

394

14 samples.

395

Throughout the 84 day period the fungal ITS gene (Figure 3B),

396

was present at least two magnitudes less than the bacterial gene

397

was. ITS genes mm-2 were <10000 in biofilms from Days 0 to

398

42, by Day 52 gene copies had increased reaching a peak at

399

Day 70 (max. 30000 ITS genes mm-2). This pattern highlights a

400

lag of 20-30 days in the fungi population reaching the

401

maximum quantification in comparison to the bacteria where a

402

peak was reached at Day 42. Similarly, to the bacteria, fungal

403

gene copies decreased after this peak but the Day 84 biofilms

404

had similar ITS gene copies as the Day 14 samples.

AC C

EP

TE D

M AN U

SC

RI PT

. After Day 42 the number of gene copies decreases steadily

405

If the lagged growth and decline trends in bacteria and fungi

406

(Figures 3 A and B) gene copies are combined, with some

407

weighting for relative size of organism, it is possible that the

408

cyclic trend in DNA cellular volume (Figures 2 A and B) could

409

be explained.

410

4.4 Microbial Community Structure

17

ACCEPTED MANUSCRIPT 4.4.1 Rarefaction Analysis

412

The rarefaction analysis (Figure 4) yields insights into the

413

sequencing effort and compares the diversity of the observed

414

number of OTUs at 97% sequence similarity cut off in all the

415

samples over time. Most of the samples, independently of the

416

time of sampling, showed rarefaction curves that did not reach

417

a plateau. This trend in the curves was particularly marked for

418

fungal communities (Figure 4B), where the samples analysed

419

showed a steeper slope when compared with bacteria,

420

suggesting that further sequencing and time series would be

421

needed to reach a full taxonomic representation of the

422

microbial communities. This data can also be associated to the

423

fact that biofilm communities did not reach a mature or stable

424

state after three months of development in the system.

425

4.4.2 Taxonomic Analysis

426

The dominant bacterial class (Figure 5A) over time in the

427

biofilm

EP

TE D

M AN U

SC

RI PT

411

was

Proteobacteria

and

within

it

AC C

samples

428

Betaproteobacteria (representing a 70% of the total community

429

on Day 34 samples), Gammaproteobacteria (5-67 %) and

430

Alphaproteobacteria (6-28 %) were the most abundant classes.

431

However, the percentages of each of these bacterial classes

432

clearly varied over time and there was high variability between

433

replicates for each day, with Day 14 having the most extreme

434

example of this. The fungal community (Figure 5B) was

18

ACCEPTED MANUSCRIPT dominated by Ascomycota (10- 75%) and Sordaryomicetes (14-

436

75 %) but Day 14 samples from two coupons (31 and 37) were

437

dissimilar to the other samples and were dominated by

438

Dothideomycetes (22-63 %) and Saccharomyctes (yeast and

439

yeast like –fungi, 7-9%). Saccharomycetes were dominant in

440

the first stages of biofilm development until Day 42, and then

441

this fungal group was not substantially represented in the

442

community.

443

At genus level (Figure 6A), Massilia (max. 62%) and

444

Pseudomonas (max. 64%) were highly abundant on the first

445

days of biofilm development. It is notable that on Day 14, one

446

of the replicates (coupon 9) showed a different microbial

447

community when compared with the other two samples despite

448

of being biological replicates. These two samples from Day 14

449

had a higher relative abundance of Pseudomonas (38-64%),

450

this genus was dominant in the community composition and

451

showed percentages similar to those on Day 42.

AC C

EP

TE D

M AN U

SC

RI PT

435

452

On Day 42, samples showed a more diverse community

453

enriched (coincident in time with a peak in the cell volume,

454

Figure 2 A and B) with genera such as Nevskia, Methylophilus,

455

Sphingomonas, Variovorax and Limnohabitants. The most

456

abundant genera in fungal communities (Figure 6B) at all

457

sampling times were Acremonium (8-70%) and Neocosmospora

458

(17-77%) but in two of the coupons from Day 14 (31 and 37)

19

ACCEPTED MANUSCRIPT there was high abundance of Letendraea (36%) and

460

Cladosporium (19-24%). The fugal composition of these two

461

samples was markedly different when compared with the most

462

even structure observed for the samples at other time points.

463

This has been also observed, as has been previously pointed,

464

for the bacterial communities of these two samples. The studied

465

succession finished with samples from Day 84 showing a high

466

abundance of the genus Metacordyceps in the fungal

467

community (2-12%).

468

MDS analysis of bacteria (based on the Bray-Curtis similarity

469

matrix of bacterial sequences with 97% similarity cut off) is

470

presented in Figure 7A. Variability in bacterial community

471

distribution between sample points was observed but these

472

differences were not significant (ANOSIM: global-R= 0.143 p-

473

value = 0.074) and there was no clear temporal pattern. For

474

fungi (Figure 7B) the MDS analysis revealed a tight clustering

475

of the biofilm samples, which corresponds to very little

476

variation in community composition over time as shown in the

AC C

EP

TE D

M AN U

SC

RI PT

459

477

ANOSIM analysis (R=0.023, p-value = 0.399).

478

4.4.3 Diversity and richness indicators

479

Figure 8 shows the richness (i.e. number of different OTUs)

480

and diversity (i.e. number of different OTUs taking into

481

account their relative abundance) of bacterial and fungal

482

communities at 97% sequence similarity cut off. When 20

ACCEPTED MANUSCRIPT compared, bacterial communities were more dynamic and

484

diverse than fungi; fungal communities showed limited

485

diversity, dominated by few taxonomic groups with little

486

change over time. When analysed separately, the dominance

487

indicator for bacterial communities decreased over time,

488

increasing slightly by Day 70 and decreasing again by Day 84,

489

reaching a minimum value. This trend suggests that initially

490

there are a few bacteria that attach and colonize the pipe

491

surfaces, from which the biofilm develops towards a more

492

diverse community, with less dominating taxa. This is

493

supported by the patterns in diversity, evenness and richness

494

values. However, bacterial richness indicator did not change

495

considerably over time. Fungal communities showed a clear

496

dominance of certain OTUs, which was less marked at the

497

beginning of the data series studied and increased over time.

498

Similarly, evenness was higher for the first 42 days of biofilm

499

development and subsequently fluctuated at lower level after

500

that day. The fungal communities were not very diverse with

AC C

EP

TE D

M AN U

SC

RI PT

483

501

all samples having Shannon index values below 2. Richness (<

502

100 OTUs) fluctuated only slightly among samples. Clearly,

503

few fungal OTUs dominated the composition of biofilms over

504

time.

505

4.5 Correlations between microbiological data

506

Table 3 shows a matrix of the non-parametric correlations

507

between microbiological data and indicates statistically

21

ACCEPTED MANUSCRIPT significant correlations at different p-levels. The diversity and

509

richness indicators showed a significant positive correlation for

510

bacteria but not for fungi. Only the number of fungal ITS genes

511

correlated significantly with richness and diversity for that

512

taxonomic group. The cell volume did not significantly

513

correlate with any of the other parameters. When associations

514

between different microbial groups were taken into account, the

515

different Proteobacteria class tended to be correlated positively

516

among them. The microbial groups that correlated the most

517

with other microorganisms were the bacteria Actinobacteria,

518

Cytophagia and the fungi Dothideomycetes, Basiodioycota,

519

Leotiomycetes and Ascomycota. This result shows that certain

520

microbial groups are more or less likely to co-exist within the

521

studied biofilm samples.

522

5 Discussion

523

Biofilm development was monitored over a three-month period

524

and the characteristics of inter-taxa interactions were studied to

AC C

EP

TE D

M AN U

SC

RI PT

508

525

evaluate and detect patterns of temporal change and drivers of

526

biofilm behaviour. Sequencing results indicated that biofilm

527

developmental patterns were initially driven by bacteria. Fungi

528

showed a lag response to temporal changes in terms of ITS

529

gene copy numbers and the community structure displayed

530

limited changes over time. Generally, it was observed that an

531

initial, diverse community (dominated by bacteria) was able to

22

ACCEPTED MANUSCRIPT attach and was embedded in the biofilm. After one month of

533

development, biofilm diversity was temporally reduced,

534

coincident with a decrease in cell volume from Days 42 to 52,

535

likely due to a selective mechanism that ultimately yielded a

536

higher diversity after two months development. Confirming the

537

leading role of bacteria in the process of succession it was

538

observed that the rise in diversity from 34 to 42 days,

539

corresponded with a peak in the number of 16s rRNA gene

540

copy numbers. The observed diversity pattern could have been

541

a result of competition, with some bacteria/fungi being out-

542

competed (hence diversity decreasing) but the resultant more

543

mature biofilm then being a suitable niche for other cells to

544

subsequently colonise (hence diversity increasing again). We

545

hypothesis that the daily fluctuating hydraulic conditions were

546

acting as a selective mechanism to shape microbial diversity

547

and structure, as has been previously observed in previous

548

research carried out in the same system (Douterelo et al. 2013;

549

Fish et al. 2017). If daily patterns help to maintain natural

AC C

EP

TE D

M AN U

SC

RI PT

532

550

diverse communities and more compact and stable biofilms,

551

this can have positive implications for the management of

552

DWDS, safeguarding these systems from the incursion of non-

553

desirable microorganisms into biofilms and avoiding their

554

survival and later mobilisation into bulk water.

555

The ability of microorganisms to respond to changing

556

environmental conditions has been previously correlated with 23

ACCEPTED MANUSCRIPT the bacterial genome size and with their cell volume (Yooseph

558

et al. 2010). Here the dominant peak in cell volume (Day 70 on

559

Figure 2A) corresponded to the peak in ITS gene numbers

560

(Figure 3B) which can be explained by the generally larger cell

561

size of fungi, particularly of filamentous fungi when forming

562

hyphae. If this hypothesis is correct, fungi may be more

563

resilient to change and persist over time, explaining the limited

564

change in diversity observed during the succession process and

565

the observed lag between the peak in bacteria and fungi genes

566

apparent when comparing Figures 4 A and B. Fungi can

567

survive and grow in DWDS within biofilms, particularly at

568

warmer temperature or in systems where the flow is low or

569

stagnant (Siqueira et al. 2012; Douterelo et al. 2016b; Oliveira

570

et al. 2016). The night time flow rate of the hydraulic regime

571

selected for these experiments was at the upper end for those

572

observed in 75-100mm pipes in real systems, thus fungi are

573

possibly even more prevalent and important in real operational

574

systems. The challenge for water utilities is how to

AC C

EP

TE D

M AN U

SC

RI PT

557

575

manage/control these microorganisms since fungal walls

576

protect cells against mechanical damage and stop the ingress of

577

toxic molecules, making these organisms even more resistant to

578

environmental changes (Cole 1996). In addition, fungi can form

579

spores which increase their resistance to disinfection (Mamane-

580

Gravetz & Linden 2005; Sonigo 2011). This study reinforces

24

ACCEPTED MANUSCRIPT that fungi must be included in monitoring campaigns in DWDS

582

when assessing microbial risks.

583

It has been suggested that, generally, microbial communities

584

tend to develop towards a stable community (Faust et al. 2015)

585

or plateau phase. In this study, no mature/stable biofilm

586

communities were established on the pipe walls after three

587

months of development. This result suggests that longer time

588

periods may be needed to fully develop mature microbial

589

communities in DWDS. However, determining the scale of

590

change to study succession in an ecosystem is challenging,

591

given that depending on the habitat, community changes can be

592

triggered by fluctuations in environmental conditions or by a

593

transient perturbation that moves the system into an alternative

594

steady state (Fierer et al. 2010). The experimental DWDS used

595

in this study minimised any such transient perturbations, by

596

controlling hydraulics and offering a relatively stable physico-

597

chemical environment over time. While the local source water,

598

treatment works and trunk main system prior to the laboratory

AC C

EP

TE D

M AN U

SC

RI PT

581

599

facility will have had some variations over the 3 months, Table

600

2 indicates these where minimal and they would have been

601

gradual fluctuations rather than rapid dynamic changes. Hence

602

the changes observed such as those on Day 42 are most

603

probably due to intrinsic biofilm processes, related to microbial

604

processes during development. Martiny et al. (2003) studied the

605

process of biofilm development over a 3-year period in a model 25

ACCEPTED MANUSCRIPT distribution

using

fingerprinting

and

Sanger

607

sequencing to monitor bacterial communities. Using these

608

methods, Martiny et al. (2003) established that after 500 days,

609

the biofilm entered a stable state, which was characterised by a

610

greater richness of bacteria. We have observed higher bacterial

611

diversity at the final stages of biofilm formation but after three

612

months of biofilm development our communities were not

613

mature or stable.

614

We have observed that specific microbial taxa were dominant

615

throughout the succession process with a core microbial

616

population present in the biofilms over time but at different

617

relative abundances. Thus, it can be suggested these particular

618

microbial taxa are fundamental in directing the process of

619

succession. The taxa in this core community include the

620

bacteria Pseudomonas, Massillia and Sphingomonas and the

621

fungi Acremonium and Neocosmopora. In a previous study, the

622

role of Pseudomonas as primary and initial colonizer of the

623

same experimental facility was established in the first 28 days

AC C

EP

TE D

M AN U

SC

systems

RI PT

606

624

of biofilm development (Douterelo et al. 2014b) and here its

625

continual key role in shaping biofilm structure over a longer

626

period of time has been confirmed. Species belonging to the

627

genus Pseudomonas might benefit from what is known as the

628

founder effect, as an initial member of biofilms in the pipe

629

surface it will have the advantage over secondary colonizers

630

and will remain dominant over time (Kelly et al. 2014). The 26

ACCEPTED MANUSCRIPT 631

presence of core populations have been observed before in

632

water meters of DWDS (Ling et al. 2016),

633

chloraminated DWDS (Kelly et al., 2014) and in DWDS of

634

chlorinated systems (Douterelo et al., 2017). Kelly et al., 2014

635

showed in chloraminated systems that the core community was

636

enriched

637

Methylobacteria and Methylophylus were also important

638

representatives of bacterial communities over time in this study

639

and the relative abundance of Methylophylus increased after

640

two months of biofilm development. Methylotrophic bacteria

641

have been found in several drinking water-related ecosystems

642

(McCoy & VanBriesen 2012; Douterelo et al. 2014b; Liu et al.

643

2014b) but the reasons behind their abundance in these systems

644

are unclear. However, methylotrophic bacteria

645

related with the generation of holoacetic acid which is a

646

common disinfectant by-product in chlorinated systems (Kelly

647

et al. 2014), therefore it is possible that this group is promoted

648

due to the decay of residuals within DWDS.

and

RI PT

methanotrophs

methylotrophs.

have been

AC C

EP

TE D

M AN U

SC

with

biofilms of a

649

The fungi Acremonium and Neocosmopora were also forming

650

part of the fungal-bacteria core community. Acremonium is a

651

filamentous fungus which has been previously isolated in water

652

supply reservoirs, drinking water treatment plants and in bulk

653

water of drinking water distribution systems (Zacheus &

654

Martikainen 1995; Pereira et al. 2010; Oliveira et al. 2016; Bai

655

et al. 2017; Douterelo et al. 2017) and has been associated with 27

ACCEPTED MANUSCRIPT the occurrence of taste and odour issues in drinking water (Bai

657

et al. 2017). Neocosmopora is commonly found in soils and

658

some species are phytophatogenic (Aoki et al. 2014). It has

659

been found that Neocomospora, under laboratory conditions is

660

able to uptake chloride (Miller & Budd 1975) and can produce

661

cyclosporine a fungal metabolite with antifungal activity; these

662

abilities can explain its supremacy over other microorganisms

663

in mixed-species microbial communities in chlorinated DWDS.

664

This study confirms the presence of Acremonium and

665

Neocomospora from the early stages of biofilm formation to a

666

more developed biofilm in chlorinated systems forming core

667

communities with bacteria. Despite of fungi involvement in the

668

organoleptic deterioration of water, act as potential pathogens

669

and toxin producers, fungi are usually omitted from drinking

670

water regulation and there are no limits or standards regarding

671

their presence in drinking water. This study shows the ubiquity

672

of fungi in forming mix-species consortium with the most

673

common bacteria in DWDS biofilms. Taking into account these

AC C

EP

TE D

M AN U

SC

RI PT

656

674

results, further consideration should be given to these

675

microorganisms in order to guarantee the delivery of good

676

quality water and to control biofilm formation at the initial

677

stages of development in DWDS.

678

Environmental stressors such as chlorine, can influence

679

successional trajectories in the early stages of succession

680

(Fierer et al. 2010) by increasing the abundance of taxa that are 28

ACCEPTED MANUSCRIPT resistant to the particular stress and can subsequently modify

682

the habitat by producing EPS (extracellular polymeric

683

substances) to protect the cells from oxidative stress (Besemer

684

et al. 2007). This study indicates the capability of

685

microorganisms to proliferate despite the presence of chlorine

686

residual throughout biofilm development, from initial stages

687

with low cell numbers and thinner and more environmentally

688

exposed biofilms, to more developed biofilms exhibiting larger

689

biomass and diversity. From these results, it can be

690

hypothesised that the microbial communities involved in initial

691

attachment to the pipes are more resistant to chlorine, then once

692

more EPS is produced, other species which are less resistant

693

can join the biofilm and survive within it protected against the

694

action of chlorine by the EPS, hence the biofilm environment

695

may favour more diverse communities over time. This suggests

696

that, despite its widespread use to control planktonic

697

microorgansims, the application of chlorine to control the

698

attached microbial life within DWDS is questionable. In

AC C

EP

TE D

M AN U

SC

RI PT

681

699

agreement with our observations, Simões et al. (2016) under

700

controlled laboratory conditions, studied the role of inter-

701

kingdom interactions in chlorine resistance and showed that

702

associations between fungi and bacteria were ecologically

703

beneficial and promoted resistance to disinfection.

704

The stable state of a biofilm is not entirely determined by

705

external factors and interactions among microbial community 29

ACCEPTED MANUSCRIPT 706

members might play a key role in determining biofilm status

707

(Faust et al. 2015). These complex interactions between

708

microorganisms

709

contributors to biofilm dynamics and are relatively unexplored

710

in the context of DWDS. Here the relative abundance of certain

711

bacterial and fungal taxa correlated together including

712

Proteobacteria and Basidiomycota (Table 3), both members of

713

a core microbial community present continuously over the

714

successional process in this study. Fungal-bacteria associations

715

can lead to changes in nutrient availability, even in the case of

716

DWDS where the level of nutrients such as TOC in the system

717

are controlled to a certain extent. The presence of certain taxa

718

can ease nutrient limitation and it is well known that in other

719

ecosystems bacterial and fungal consortium are able to increase

720

nutrient bioavailability and mobilization of key nutrients

721

(Rashid et al. 2016) and this can support the process of

722

succession. However, if biofilms are to a certain extent self-

723

sufficient, independent of the concentration of available

with

their

environment

are

key

AC C

EP

TE D

M AN U

SC

RI PT

and

724

nutrients in the bulk-water, manipulating or changing external

725

nutrient parameters is not going to stop biofilm formation and

726

development in DWDS.

727

6. Conclusions

728

Diverse bacterial communities cohabiting with more stable

729

fungal communities, with the more common taxa

30

ACCEPTED MANUSCRIPT ubiquitous and constantly present over time and a temporal

731

variability that does not follow a specific pattern driven by

732

bacteria.

733

To deal with the heterogeneity of the process of biofilm

734

development in DWDS and to model these systems, the

735

focus needs to be on targeting dominant and ubiquitous

736

microorganisms.

737

In order to manage microbial risks in DWDS we need to

738

better

739

microorganisms, to design monitoring strategies and assess

740

their use as bioindicators of overall biofilm status in the

741

system.

SC

the

behaviour

of

these

key

M AN U

understand

RI PT

730

Acknowledgements

743

The research reported here was supported by the UK

744

Engineering and Physical Sciences Research Council (EPSRC).

745

EPSRC-LWEC Challenge Fellowship EP/N02950X/1, the

746

Challenging Engineering Grant EP/G029946/1 and the EPSRC

AC C

EP

TE D

742

747

Platform Grant Funding EP/1029346/1.

748

References

749

Aoki T., O’Donnell K. & Geiser D.M. (2014). Systematics of

750

key phytopathogenic Fusarium species: current status and

751

future challenges. Journal of General Plant Pathology, 80,

752

189-201.

31

ACCEPTED MANUSCRIPT Ashbolt N.J. (2015). Microbial contamination of drinking water

754

and human health from community water systems. Current

755

environmental health reports, 2, 95-106.

756

Bai X., Zhang T., Qu Z., Li H. & Yang Z. (2017). Contribution

757

of filamentous fungi to the musty odorant 2,4,6-trichloroanisole

758

in water supply reservoirs and associated drinking water

759

treatment plants. Chemosphere, 182, 223-230.

760

Besemer

761

Hochedlinger G., Hodl I., Baranyi C. & Battin T.J. (2007).

762

Biophysical controls on community succession in stream

763

biofilms. Appl Environ Microb, 73, 4966-4974.

764

Buse H.Y., Lu J., Struewing I.T. & Ashbolt N.J. (2013).

765

Eukaryotic diversity in premise drinking water using 18S

766

rDNA sequencing: implications for health risks. Environ Sci

767

Pollut R, 20, 6351-6366.

768

Chao A. (1984). Non parametric estimation of the number of

769

classes in a population. Scand. J. Stat., 11, 265-270.

770

Chao Y., Mao Y., Wang Z. & Zhang T. (2015). Diversity and

Singer

G.,

Limberger

R.,

Chlup

A.K.,

AC C

EP

TE D

M AN U

SC

K.,

RI PT

753

771

functions of bacterial community in drinking water biofilms

772

revealed by high-throughput sequencing. Scientific Reports, 5,

773

10044.

774

Clarke K. & Warwick R. (2005). Primer-6 computer program.

775

Natural Environment Research Council, Plymouth.

776

Cole G.T. (1996). Basic biology of fungi.

32

ACCEPTED MANUSCRIPT Deines P., Sekar R., Husband P.S., Boxall J.B., Osborn A.M. &

778

Biggs C.A. (2010). A new coupon design for simultaneous

779

analysis of in situ microbial biofilm formation and community

780

structure in drinking water distribution systems. Appl Microbiol

781

Biot, 87, 749-756.

782

Douterelo I., Boxall J.B., Deines P., Sekar R., Fish K.E. &

783

Biggs C.A. (2014a). Methodological approaches for studying

784

the microbial ecology of drinking water distribution systems.

785

Water Res, 65, 134-156.

786

Douterelo I., Jackson M., Solomon C. & Boxall J. (2016a).

787

Microbial analysis of in situ biofilm formation in drinking

788

water distribution systems: implications for monitoring and

789

control of drinking water quality. Appl Microbiol Biot, 100,

790

3301-3311.

791

Douterelo I., Jackson M., Solomon C. & Boxall J. (2016b).

792

Microbial analysis of in situ biofilm formation in drinking

793

water distribution systems: implications for monitoring and

794

control of drinking water quality. Appl Microbiol Biotechnol,

AC C

EP

TE D

M AN U

SC

RI PT

777

795

100, 3301-11.

796

Douterelo I., Jackson M., Solomon C. & Boxall J. (2017).

797

Spatial and temporal analogies in microbial communities in

798

natural drinking water biofilms. Sci Total Environ, 581, 277-

799

288.

800

Douterelo I., Sharpe R. & Boxall J. (2014b). Bacterial

801

community dynamics during the early stages of biofilm

33

ACCEPTED MANUSCRIPT 802

formation in a chlorinated experimental drinking water

803

distribution

804

discolouration. J Appl Microbiol.

805

Douterelo I., Sharpe R.L. & Boxall J.B. (2013). Influence of

806

hydraulic regimes on bacterial community structure and

807

composition in an experimental drinking water distribution

808

system. Water Res, 47, 503-516.

809

Edgar R.C. (2013). UPARSE: highly accurate OTU sequences

810

from microbial amplicon reads. Nat Methods, 10, 996-8.

811

Edgar R.C., Haas B.J., Clemente J.C., Quince C. & Knight R.

812

(2011). UCHIME improves sensitivity and speed of chimera

813

detection. Bioinformatics, 27, 2194-2200.

814

Faust K., Lahti L., Gonze D., de Vos W.M. & Raes J. (2015).

815

Metagenomics meets time series analysis: unraveling microbial

816

community dynamics. Curr Opin Microbiol, 25, 56-66.

817

Fierer N., Nemergut D., Knight R. & Craine J.M. (2010).

818

Changes through time: integrating microorganisms into the

819

study of succession. Res Microbiol, 161, 635-642.

for

drinking

water

RI PT

implications

AC C

EP

TE D

M AN U

SC

system:

820

Fish K., Osborn A. & Boxall J. (2017). Biofilm structures (EPS

821

and bacterial communities) in drinking water distribution

822

systems

823

discolouration. Sci Total Environ, 593, 571-580.

824

Fish K.E., Collins R., Green N.H., Sharpe R.L., Douterelo I.,

825

Osborn A.M. & Boxall J.B. (2015). Characterisation of the

826

Physical Composition and Microbial Community Structure of

are

conditioned

by

hydraulics

and

influence

34

ACCEPTED MANUSCRIPT Biofilms

a

Model

Full-Scale

Drinking

Water

828

Distribution System. PLoS ONE, 10, e0115824.

829

Flemming H.C. (1998). Relevance of biofilms for the

830

biodeterioration of surfaces of polymeric materials. Polym

831

Degrad Stabil, 59, 309-315.

832

Gardes M. & Bruns T.D. (1993). ITS primers with enhanced

833

specificity for basidiomycetes--application to the identification

834

of mycorrhizae and rusts. Mol Ecol, 2, 113-8.

835

Gomes I.B., Simões M. & Simões L.C. (2014). An overview on

836

the reactors to study drinking water biofilms. Water Res, 62,

837

63-87.

838

Gomez-Alvarez V., Pfaller S. & Revetta R.P. (2016). Draft

839

Genome Sequence of Two Sphingopyxis sp. Strains, Dominant

840

Members of the Bacterial Community Associated with a

841

Drinking Water Distribution System Simulator. Genome

842

Announcements, 4.

843

Hammer Ø., Harper D. & Ryan P. (2008). PAST-

844

palaeontological statistics, ver. 1.89. Paleontological Museum,

AC C

EP

TE D

M AN U

SC

within

RI PT

827

845

University of Oslo, Noruega.(También disponible en línea:

846

http://folk. uio. no/ohammer/past/index. html).

847

Harper D.A.T.e. (1999). Numerical Palaeobiology. John Wiley

848

& Sons.

849

Hoefel D., Monis P.T., Grooby W.L., Andrews S. & Saint C.P.

850

(2005). Profiling bacterial survival through a water treatment

35

ACCEPTED MANUSCRIPT process and subsequent distribution system. J Appl Microbiol,

852

99, 175-186.

853

Hume M.E., Hernandez C.A., Barbosa N.A., Sakomura N.K.,

854

Dowd S.E. & Oviedo-Rondón E.O. (2012). Molecular

855

identification and characterization of ileal and cecal fungus

856

communities in broilers given probiotics, specific essential oil

857

blends, and under mixed Eimeria infection. Foodborne Pathog

858

Dis, 9, 853-860.

859

Husband P.S., Boxall J.B. & Saul A.J. (2008). Laboratory

860

studies investigating the processes leading to discolouration in

861

water distribution networks. Water Res, 42, 4309-4318.

862

Husband S., Fish K.E., Douterelo I. & Boxall J. (2016).

863

Linking discolouration modelling and biofilm behaviour within

864

drinking water distribution systems. Water Science and

865

Technology: Water Supply.

866

Kelly J.J., Minalt N., Culotti A., Pryor M. & Packman A.

867

(2014). Temporal variations in the abundance and composition

868

of biofilm communities colonizing drinking water distribution

AC C

EP

TE D

M AN U

SC

RI PT

851

869

pipes.

870

Lane, D.J. (1991) 16S/23S rRNA Sequencing. In: Stackebrandt,

871

E. and Goodfellow, M., Eds., Nucleic Acid Techniques in

872

Bacterial Systematics, John Wiley and Sons, New York, 115-

873

175.

874

Lehtola M.J., Miettinen I.T., Keinänen M.M., Kekki T.K.,

875

Laine O., Hirvonen A., Vartiainen T. & Martikainen P.J.

36

ACCEPTED MANUSCRIPT (2004). Microbiology, chemistry and biofilm development in a

877

pilot drinking water distribution system with copper and plastic

878

pipes. Water Res, 38, 3769-3779.

879

Ling F., Hwang C., LeChevallier M.W., Andersen G.L. & Liu

880

W.-T. (2016). Core-satellite populations and seasonality of

881

water meter biofilms in a metropolitan drinking water

882

distribution system. ISME J, 10, 582-595.

883

Liu G., Bakker G., Li S., Vreeburg J., Verberk J., Medema G.,

884

Liu W. & Van Dijk J. (2014a). Pyrosequencing reveals

885

bacterial

886

distribution system: an integral study of bulk water, suspended

887

solids, loose deposits, and pipe wall biofilm. Environ Sci

888

Technol, 48, 5467-5476.

889

Liu R., Zhu J., Yu Z., Joshi D., Zhang H., Lin W. & Yang M.

890

(2014b). Molecular analysis of long-term biofilm formation on

891

PVC and cast iron surfaces in drinking water distribution

892

system. Journal of Environmental Sciences, 26, 865-874.

893

Mamane-Gravetz H. & Linden K.G. (2005). Relationship

SC

unchlorinated

drinking

M AN U

in

water

AC C

EP

TE D

communities

RI PT

876

894

between physiochemical properties, aggregation and u.v.

895

inactivation of isolated indigenous spores in water. J Appl

896

Microbiol, 98, 351-63.

897

Martiny A.C., Jorgensen T.M., Albrechtsen H.J., Arvin E. &

898

Molin S. (2003). Long-term succession of structure and

899

diversity of a biofilm formed in a model drinking water

900

distribution system. Appl Environ Microb, 69, 6899-907.

37

ACCEPTED MANUSCRIPT McCoy S.T. & VanBriesen J.M. (2012). Temporal variability

902

of bacterial diversity in a chlorinated drinking water

903

distribution system. Journal of Environmental Engineering,

904

138, 786-795.

905

Miller A.G. & Budd K. (1975). The development of an

906

increased rate of Cl− uptake in the ascomycete Neocosmospora

907

vasinfecta. Can J Microbiol, 21, 1211-1216.

908

Muyzer G., Dewaal E.C. & Uitterlinden A.G. (1993). Profiling

909

of complex microbial populations by denaturing gradient gel

910

electrophoresis analysis of polymerase chain reaction amplified

911

genes coding for the 16S rRNA. Appl Environ Microbiol, 59,

912

695-700.

913

Oliveira H.M.B., Santos C., Paterson R.R.M., Gusmão N.B. &

914

Lima N. (2016). Fungi from a Groundwater-Fed Drinking

915

Water Supply System in Brazil. International journal of

916

environmental research and public health, 13, 304.

917

Pereira V.J., Fernandes D., Carvalho G., Benoliel M.J., San

918

Romão M.V. & Barreto Crespo M.T. (2010). Assessment of the

AC C

EP

TE D

M AN U

SC

RI PT

901

919

presence and dynamics of fungi in drinking water sources using

920

cultural and molecular methods. Water Res, 44, 4850-4859.

921

Prest E.I., Weissbrodt D.G., Hammes F., van Loosdrecht

922

M.C.M. & Vrouwenvelder J.S. (2016). Long-Term Bacterial

923

Dynamics in a Full-Scale Drinking Water Distribution System.

924

PLoS ONE, 11, e0164445.

38

ACCEPTED MANUSCRIPT Rashid M.I., Mujawar L.H., Shahzad T., Almeelbi T., Ismail

926

I.M.I. & Oves M. (2016). Bacteria and fungi can contribute to

927

nutrients bioavailability and aggregate formation in degraded

928

soils. Microbiol Res, 183, 26-41.

929

Revetta R.P., Gomez-Alvarez V., Gerke T.L., Santo Domingo

930

J.W. & Ashbolt N.J. (2016). Changes in bacterial composition

931

of biofilm in a metropolitan drinking water distribution system.

932

J Appl Microbiol, 121, 294-305.

933

Revetta R.P., Pemberton A., Lamendella R., Iker B. & Santo

934

Domingo J.W. (2010). Identification of bacterial populations in

935

drinking water using 16S rRNA-based sequence analyses.

936

Water Res, 44, 1353-1360.

937

Roeselers G., Coolen J., van der Wielen P.W.J.J., Jaspers M.C.,

938

Atsma A., de Graaf B. & Schuren F. (2015). Microbial

939

biogeography of drinking water: patterns in phylogenetic

940

diversity across space and time. Environmental Microbiology,

941

17, 2505-2514.

942

Schwering M., Song J., Louie M., Turner R.J. & Ceri H.

AC C

EP

TE D

M AN U

SC

RI PT

925

943

(2013). Multi-species biofilms defined from drinking water

944

microorganisms provide increased protection against chlorine

945

disinfection. Biofouling, 29, 917-928.

946

Simões L.C., Chaves A.F., Simões M. & Lima N. (2016). The

947

role of inter-kingdom interactions in chlorine resistance of

948

biofilms formed by drinking water-isolated microorganisms.

949

Biofilms 7, 165-165.

39

ACCEPTED MANUSCRIPT Simoes L.C., Simoes M. & Vieira M.J. (2010). Adhesion and

951

biofilm formation on polystyrene by drinking water-isolated

952

bacteria. Antonie van Leeuwenhoek, 98, 317-29.

953

Siqueira V.M., Oliveira H.M.B., Santos C., Paterson R.R.M.,

954

Gusmão N.B. & Lima N. (2012). Biofilms from a Brazilian

955

water distribution system include filamentous fungi. Can J

956

Microbiol, 59, 183-188.

957

Sonigo P., De Toni, A., and Reilly, K. (2011). A review of

958

fungi in drinking water and the implications for human health.

959

Report, Bio Intelligent Service, France.

960

van der Wielen P.W. & van der Kooij D. (2010). Effect of

961

water composition, distance and season on the adenosine

962

triphosphate concentration in unchlorinated drinking water in

963

the Netherlands. Water Res, 44, 4860-7.

964

Vaz-Moreira I., Nunes O.C. & Manaia C.M. (2017).

965

Ubiquitous and persistent Proteobacteria and other Gram-

966

negative bacteria in drinking water. Sci Total Environ, 586,

967

1141-1149.

AC C

EP

TE D

M AN U

SC

RI PT

950

968

Vilgalys R. & Hester M. (1990). Rapid genetic identification

969

and mapping of enzymatically amplified ribosomal DNA from

970

several Cryptococcus species. J Bacteriol, 172, 4238-4246.

971

Wang H., Hu C., Hu X., Yang M. & Qu J. (2012). Effects of

972

disinfectant and biofilm on the corrosion of cast iron pipes in a

973

reclaimed water distribution system. Water Res, 46, 1070-1078.

40

ACCEPTED MANUSCRIPT Wingender J. & Flemming H.C. (2011). Biofilms in drinking

975

water and their role as reservoir for pathogens. Int J Hyg Envir

976

Heal, 214, 417-423.

977

World Health Organization. 2006. Health Aspects of Plumbing

978

Yooseph S., Nealson K.H., Rusch D.B., McCrow J.P., Dupont

979

C.L., Kim M., Johnson J., Montgomery R., Ferriera S., Beeson

980

K., Williamson S.J., Tovchigrechko A., Allen A.E., Zeigler

981

L.A., Sutton G., Eisenstadt E., Rogers Y.H., Friedman R.,

982

Frazier M. & Venter J.C. (2010). Genomic and functional

983

adaptation in surface ocean planktonic prokaryotes. Nature,

984

468, 60-6.

985

Zacheus O.M. & Martikainen P.J. (1995). Occurrence of

986

heterotrophic bacteria and fungi in cold and hot water

987

distribution systems using water of different quality. Can J

988

Microbiol, 41, 1088-1094.

AC C

EP

TE D

M AN U

SC

RI PT

974

41

ACCEPTED MANUSCRIPT Gene target (organisms) 16S rRNA

Primer Pair Eub338

Primer sequences (5’ – 3’)

Primer reference

ACTCCTACGGGAGGCAGCAG

Lane, 1991

Eub518

ATTACCGCGGCTGCTGG

Muyzer et al. 1993

ITS (fungi)

ITS1F 5.8S

TCCGTAGGTGAACCTGCGG CGCTGCGTTCTTCATCG

Gardes & Bruns, 1993 Vilgalys & Hester 1990

AC C

EP

TE D

M AN U

SC

RI PT

(bacteria)

ACCEPTED MANUSCRIPT Free Chlorine (mg/l)

Turbidity (NTU)

0.98 ± 0.06 0.43 ± 0.01 0.43 ± 0.02 0.44 ± 0.03 0.36 ± 0.01 0.42 ± 0.00 0.39 ± 0.05

1.51 ± 0.05 0.08 ± 0.00 0.05 ± 0.00 0.04 ± 0.00 0.05 ± 0.00 0.05 ± 0.00 0.06 ± 0.00

TOC (mg/l)

Fe (µg/l)

1.13 ± 0.04 18 ± 0.81 1.30 ± 0.08 21 ± 1.41 1.37 ± 0.09 9.90 ± 0.00 * 18.33 ± 2.62 1.17 ± 0.04 14.33 ± 0.93 1.33 ± 0.00 17 ± 0.00 1.83 ± 0.11 37 ± 0.00

EP

TE D

M AN U

SC

7.46 ± 0.04 14.60 ± 0.00 7.42 ± 0.37 14.70 ± 0.10 7.61 ± 0.17 14.83 ± 0.00 7.12 ± 0.15 15.03 ± 0.20 7.99 ± 0.31 15.20 ± 0.20 7.07 ± 0.20 15.9 ± 0.10 7.49 ± 0.22 15.03 ± 0.40

AC C

Day 0 Day 14 Day 34 Day 42 Day 52 Day 70 Day 84

T (°C)

Mn (µg/l) < 3.6 ± 0.00 3.7 ± 0.00 < 3.6 ± 0.00 < 3.6 ± 0.00 < 3.6 ± 0.00 < 3.6 ± 0.00 < 3.6 ± 0.00

RI PT

pH

Saccharomycetes

Basidiomycota

Taphrinomycetes

Dothideomycetes

Sordariomycetes

Ascomycota

Cytophagia

Actinobacteria

Clostridia

Sphingobacteria

Bacilli

Actinobacteria

Planctomycetacia

Gammaproteobacteria

Alphaproteobacteria

Mediam cell volume

Genes ITS/mm2

Genes 16S /mm2

Shannon F

Betaproteobacteria

RI PT

Shannon B

Chao 1 F

Chao 1 B

Shannon B

ACCEPTED MANUSCRIPT

.513*

1.000

-0.012

-0.045

-0.302

-0.181

0.084

1.000

0.047

-0.078

-0.216

-0.254

1.000

0.292

0.266

.455*

-.442*

-0.257

1.000

-0.066

0.223

0.311

-0.254

-0.029

0.086

Betaproteobacteria

.728**

0.224

-0.193

-.472*

0.129

-0.028

0.185

1.000

Alphaproteobacteria

.635**

0.271

0.060

-.728**

0.034

0.335

0.185

.744**

1.000

-0.307

-0.259

*

.520

0.257

0.009

*

.461

-0.047

-.641**

-.401*

1.000

.546**

.428*

-0.326

-0.349

0.094

0.251

-.489*

0.353

.475*

-0.261

1.000

-0.238

-0.018

-0.027

0.288

-0.323

0.103

-0.154

-.420*

-0.284

0.375

-0.065

1.000

.453*

.470*

-0.081

-0.042

-0.317

0.367

-0.034

0.170

0.098

0.096

0.344

.525*

1.000

*

**

-0.070

-0.290

0.147

0.229

*

.505

0.368

0.106

-0.170

0.080

-0.123

.455*

1.000

0.076

**

.627

-0.023

-0.039

-0.201

0.216

0.194

-0.169

0.077

0.049

0.088

0.390

0.354

0.305

1.000

-0.203

0.150

0.123

.428*

-0.386

0.135

-0.254

-.531*

-.441*

.422*

-0.006

.715**

.422*

-0.018

0.292

.610**

.449*

0.258

0.015

-0.060

0.342

0.072

0.399

0.245

0.158

0.269

0.034

.593**

.470*

0.051

0.255

1.000

0.230

-0.148

0.185

-0.185

0.129

0.160

0.354

0.344

0.379

-0.079

-0.144

-0.176

-0.220

-0.011

-0.266

-.432*

-0.141

1.000

0.038

0.075

.477*

-.558**

0.129

0.335

0.223

0.038

.437*

0.100

0.074

-.414*

-0.273

-0.046

0.116

-0.179

-0.065

-0.022

1.000

0.220

0.257

0.017

.449*

-.430*

-0.116

-0.223

0.117

-0.220

-0.156

0.086

0.092

0.354

0.226

-0.070

.420*

.575**

-.482*

-.418*

1.000

-0.115

-0.011

.614**

-0.321

0.160

0.285

0.348

-0.123

0.311

0.249

-0.145

-0.383

-.484*

-0.187

0.139

-0.327

-0.245

0.271

.835**

-.587**

1.000

.466*

0.089

-0.050

-0.219

0.329

-0.342

0.000

.677**

.512*

-.630**

0.273

-.466*

-0.265

0.089

-0.336

-.411*

0.222

0.159

0.069

0.229

-0.032

1.000

0.019

0.183

-0.028

.480*

-.408*

-0.223

-0.220

-0.018

-0.197

-0.145

0.166

0.058

0.281

-0.147

-0.149

.403*

0.382

-0.320

-0.227

.592**

-0.377

0.032

1.000

0.082

-0.036

.514*

-0.116

0.189

0.186

0.396

0.274

.405*

0.096

-0.141

-0.372

-.460*

-0.076

-0.147

-.410*

0.145

.607**

0.283

-0.236

.589**

0.326

-0.191

Gammaproteobacteria Planctomycetacia Actinobacteria Bacilli Sphingobacteria Clostridia Actinobacteria Cytophagia Ascomycota Sordariomycetes Dothideomycetes Taphrinomycetes Basidiomycota Saccharomycetes Leotiomycetes

.418

.639

1.000

SC

Mediam cell volume

M AN U

Genes ITS/mm2

TE D

Genes 16S /mm2

1.000

EP

Shannon F

AC C

Chao 1 F

1.000

ACCEPTED MANUSCRIPT

M AN U

SC

RI PT

A)

AC C

EP

TE D

B)

Figure 1: A) Temperature controlled pipe-test facility at the University of Sheffield. PWG coupons were inserted along the length of each loop to allow for subsequent biofilm removal and examination. B) Schematic showing the characteristics of the loop used in this study to report biofilm development.

ACCEPTED MANUSCRIPT

C)

M AN U

SC

RI PT

A)

AC C

EP

TE D

B)

Figure 2: Confocal analysis of biofilm cell development over time (n=21). A) showing cell volume without data outliers, B) reduced Y-axis

scale to appreciate differences in cell volume at first stages of biofilm development C) graph showing results from cell spread on coupons. (n=21, in all cases apart from Day 0 where n=20 and Day 82 where n=19). In all cases box and whisker plots show the entire range of the data, the interquartile range and the median.

ACCEPTED MANUSCRIPT 1.80E+07 1.60E+07 1.40E+07

RI PT

1.20E+07 1.00E+07

8.00E+06 6.00E+06

SC

4.00E+06 2.00E+06 0.00E+00 0

10

20

M AN U

Bacterial 16S rRNA genes mm-2

A)

30

40

50

60

70

80

90

50

60

70

80

90

Day

B)

TE D

8.00E+04 7.00E+04

EP

5.00E+04 4.00E+04 3.00E+04

AC C

Fungal ITS genes mm-2

6.00E+04

2.00E+04 1.00E+04 0.00E+00 -1.00E+04 -2.00E+04

0

10

20

30

40

Day

Figure 3: Quantitative-PCR showing the number of genes per mm2 of biofilm over time for the A) 16S rRNA gene for bacteria and B) ITS gene for fungi.

ACCEPTED MANUSCRIPT

B)

AC C

EP

TE D

M AN U

SC

RI PT

A)

Figure 4: Rarefaction analysis of microbial communities at 97% sequence similarity cut off , consider as species level for ; A) bacteria and B) fungi. Numbers indicate the day of sampling ( D14, D34, D42, D52, D70, D84) and the biological replicate (1, 2 , 3).

ACCEPTED MANUSCRIPT

B)

80

60

sordariomycetes 90

dothideomycetes eurotiomycetes

80

70

60

50

taphrinomycetes fungi unclassified basidiomycota

exobasidiomycetes saccharomycetes tremellomycetes

M AN U

50

40

30

20

10

0 9 31 37 10 11 12 13 15 33 16 17 18 19 20 34 22 24 32 D34

D42

D52

D70

D84

40

agaricomycetes

30

ustilaginomycetes ascomycota

20

leotiomycetes 10

neocallimastigomycetes pezizomycotina

0 9 31 37 10 11 12 13 15 33 16 17 18 19 20 34 22 24 32 D14

D34

D42

D52

D70

D84

basidiomycota unspecified class

AC C

D14

TE D

% relative abundance

70

ascomycota unspecified class

100

RI PT

90

acidimicrobiia sphingobacteria cytophagia op8 (candidate division) alphaproteobacteria bacilli nostocophycideae gammaproteobacteria verrucomicrobiae actinobacteria (class) fusobacteria (class) acidobacteria flavobacteria spartobacteria thermoleophilia thermomicrobia nitrospira (class) betaproteobacteria deltaproteobacteria anaerolineae synechococcophycideae deinococci subsectionii subsectioniv epsilonproteobacteria planctomycetacia clostridia erysipelotrichi class unspecified bacteroidia actinobacteria tm7 (candidate division) gemmatimonadetes opitutae

SC

100

EP

A)

Figure 5: Temporal changes in the microbial community structure over time at class level for A) bacteria and B) fungi. Numbers within each sampling day (D14, D34, D42, D52, D70 and D84) indicates the number of the coupon sampled.

ACCEPTED MANUSCRIPT

A)

brevundimonas

100

B) 100

acremonium

pseudomonas

neocosmospora

massilia 90

sphingomonas

didymella

90

aspergillus mesorhizobium

RI PT

80

metacordyceps

80

herbaspirillum sphingobium achromobacter

70

70

bradyrhizobium

60

bosea

flavobacterium 50

60

fungi unclassified scolecobasidium letendraea

M AN U

% relative abundance

methylobacterium

ascomycota

SC

methylophilus

taphrina

cladophialophora

50

cladosporium

sediminibacterium nevskia

40

TE D

variovorax

basidiomycota

40

pyrenochaeta pyricularia

limnohabitans arenimonas

30

malassezia

30

EP

hydrocarboniphaga

chaetomium

blastobacter

20

AC C

opitutus

saccharomyces

20

eurotium

mucilaginibacter gemmata

10

0 9 31 37 10 11 12 13 15 33 16 17 18 19 20 34 22 24 32 D14

D34

D42

D52

D70

D84

physalospora 10

sporotalea

hypocrea

rhodococcus

fusarium

escherichia_shigella op8 (candidate division)

0 9 31 37 10 11 12 13 15 33 16 17 18 19 20 34 22 24 32 D14

D34

D42

D52

D70

D84

capronia acantholichen

Figure 6: Analysis of microbial community data at genus level , showing the relative abundance of the most abundant genera inhabiting biofilms A) bacteria B) fungi. Numbers within each sampling day (D14, D34, D42, D52, D70 and D84) indicates the number of the coupon sampled.

ACCEPTED MANUSCRIPT

Stress= 0.1106

Stress= 0.1208

AC C

EP

TE D

B)

M AN U

SC

RI PT

A)

Day 14

Day 34

Day 42

Day 52

Day 70

Day 84

Figure 7: Nonmetric multidimensional scaling plot of biofilm samples (n=18 ) from various sampling times (days). The analysis was based on Bray–Curtis similarity matrix calculated from the relative abundance of OTUs at 97% sequence similarity cut off. A)bacteria and B) fungi.

ACCEPTED MANUSCRIPT

Dominance

A)

Shannon

B) 4

0.6

0.5

3

2

0.3

0.2

1

SC

0.1

0

0.0 D34

D42

D52

D70

D84

D14

Eveness

D)

0.14

TE D

0.12 0.10

EP

0.08 0.06

0.02 0.00 D14

AC C

0.04

D34

D42

D52

D34

M AN U

D14

C)

RI PT

0.4

D70

D42

D52

D70

D84

D70

D84

Chao I

700 600 500 400 300 200 100

0 D84

Bacteria

D14

D34

D42

D52

Fungi

Figure 8: Result for A) dominance, B) diversity (Shannon), C) evenness and D) Chao I indexes for bacteria and fungi at 97% sequence similarity cut off over time. Bars indicate standard deviation (n=3).

ACCEPTED MANUSCRIPT Highlights

Succession of multi-species biofilms was studied under representative conditions

•

A core fungal-bacterial community was observed continuously over time

•

Bacteria responded rapidly while fungi had a lag in responding to biofilm dynamics

•

Core microorganisms are a potential tool to develop new control strategies

AC C

EP

TE D

M AN U

SC

RI PT

•