biofilm formation potential of heat resistant escherichia coli dairy isolates and the complete genome of multidrug resistant heat resistant strain fam21845

Some strains, like 205 FAM22996, were consistent in their lack of biofilm formation over almost all media 206 and temperatures tested overall score = 4, SD = 0.85.. As ABTCAA was found t

coli Dairy Isolates and Complete Genome of MDR Heat

greatly between strains, media and assay Our results highlight the importance of the

depends on the bacteria comprising the biofilm, but the major constituents are

cannot be treated with high temperatures and other decontamination procedures are

a very recent qPCR study found the LHR in only 0.5% of clinical E coli isolates (n =

98

613, (19)), which further supports the speculation of a selective pressure in the dairy

99 environment For the purpose of this study, heat resistance refers to that mediated by

107 and biofilms can be found on many contact surfaces in the food industry (23) Should

108 such strains be heat resistant as well, increased contamination of raw milk products

116 gene transfer (26), it is also of concern in this context

adhesion rates and Bioflux flow cells (FC)) Macrocolony assays were employed to

126 sequenced and assembled the genome of this isolate It was analyzed with respect

127

to its LHR, biofilm formation genes and was also found to carry multiple antimicrobial

128 resistance (AMR) and heavy metal resistance genes We detected the presence of

129 five plasmids, including one harboring several AMR genes, a heavy metal resistance

130 operon and a disinfectant resistance gene, and one containing a TEM-1 β-lactamase

134 the food industry The combination of these many resistance and persistence factors

135

is problematic, as co-selection can lead to retaining all of them, even if selection

136 pressure were only applied to one Strains like FAM21845 give rise to another

137 concern: the possible spread of resistance and persistence factors in the food

138 industry Also, transfer of antimicrobial resistance genes in the gastrointestinal tract

139

of humans and in animal models have been observed (reviewed in (27, 28)) and a

140 spread via this route cannot be excluded

146

sensitive strains isolated from raw milk and raw milk cheese, as well as E coli K-12

147 MG1655 as a further heat sensitive strain and reference Strains were considered

148 heat resistant if their reduction in CFU after 30 min incubation at 55°C was less than

157 morphologies dependent on the production of these two matrix constituents

158 Production of both curli fibers (curli) and cellulose results in highly structured colonies

159 with network-like appearance, while high amounts of curli without cellulose result in

160 colonies with concentric wrinkled rings (33) Congo red (staining both curli and

161 cellulose (34)) and calcofluor (staining cellulose (35)) further aid in the qualitative

162

evaluation of production of these two matrix components In most but not all E coli

163 strains, the regulator CsgD, mediating expression of both curli and cellulose is

164 expressed below 30°C (35) It was found that for K-12 strains, CsgD production is

165 higher, resulting in more curli and cellulose production, when they are grown on salt-

166 free LB plates (36) In general, the combination of temperature below 30°C and salt-

167 free media is expected to result in the strongest production of both curli and cellulose

168

(37) We grew macrocolonies on regular and salt-free LB agar plates (in addition to

169 ABTCAA and RPSMdil agar) at both 28 and 37°C, to assess the impact of salt and

170 temperature on curli and cellulose production in the present strains collective

177

to five at 37°C For LB it was eight at 28°C and two at 37°C and in ABTCAA seven at

178 28°C and one at 37°C Neither curli nor cellulose was produced on RPSMdil agar at

179

either temperature by any strain tested (Table S1) The switch from 28 to 37°C

180 reduced production of curli and/or cellulose in seven isolates on LBnoS, 15 in LB and

188 dependency of the extent of biofilm formation on both medium and temperature was

189

observed (Table 2) For each medium except RPSMdil, 28°C led to the greatest

190 overall score of biofilm formation RPSMdil also showed by far the lowest overall

191 score at 12°C In contrast to the other three media, there was no strain which did not

192 form biofilm in RPSMdil It is also interesting to note that 37°C, while being the

193 temperature with the lowest overall biofilm formation for the rich media LB and

194

LBnoS, was the best and second best temperature for RPSMdil and ABTCAA

195 respectively The best overall medium for biofilm formation at each individual

196 temperature tested was ABTCAA

197

As with media and temperatures, there were large variations in biofilm

198 formation between strains Overall scores per strain ranged from four to 42, with K-12

199 MG1655 in the mid-range (score 24) Only seven strains had a score ≥ 1 under all

200 conditions tested (FAM21845, FAM22954, FAM22961, FAM22963, FAM23016,

201

FAM23030, and FAM23109, Table 2) Variability of biofilm formation (measured by

202 standard deviation of all category values; ‘-‘ = 0), was twice as high for FAM19195

203 (1.60) than for FAM21845 (0.65), demonstrating clear differences of the extent of

204 biofilm formation regulation under the conditions tested Some strains, like

205 FAM22996, were consistent in their lack of biofilm formation over almost all media

206 and temperatures tested (overall score = 4, SD = 0.85)

207

Strains with overall low biofilm formation scores tended to form very little to no

208 biofilm in LB and LBnoS Some weak biofilm formers (FAM22996, FAM21808,

209

FAM22321) also formed little or no biofilm in ABTCAA (Table 2) FAM21845 had the

210 highest overall score of 42 (2nd: FAM22954 with 34), being the most consistent

211

biofilm former (Table 2) although it did not always have the highest absolute OD /

212 ODc ratio At 28°C, we found positive correlations between cellulose production and

217 which is then followed by maturation if conditions allow (38) To assess this critical

than a factor of ten, from 1.2 × 103 for FAM22321 to 1.7 × 104 cells / (min × cm2) for

220 FAM21845 One-Way ANOVA found significant differences between three groups of

224 Dynamism in the surrounding media changes conditions for biofilm formation by

225 introducing shear forces and delivering fresh media to adherent cells (no nutrient

226 depletion) (39) To assess biofilm formation under dynamic conditions, we used the

227 Fluxion Bioflux system As ABTCAA was found to be the overall best media for

228 biofilm formation under static conditions, this media was chosen for dynamic biofilm

229 formation assays For technical reasons, experiments were run for 24 instead of 48 h,

230 and 37°C was used to allow for sufficient growth, even though overall scores were

231 slightly higher at 28°C under static conditions (overall score 114 vs 97) Due to

232 variation between replicates, strains had to be categorized into two sets with overall

233 either good or poor reproducibility Good reproducibility was assigned to strains that

234 consistently either did or did not form biofilm on the sides of the channel and/or within

235 the FC channel between replicates Poor reproducibility was defined as biofilm

236

formation in some, but not all replicates (Table S1) All strains were able to produce

237 biofilm on the walls of the channel (poor reproducibility in FAM21808, FAM22954,

238 and FAM22996) A bacterial lawn, defined as light gray coverage clearly darker than

239 the negative control, but much less dense than true biofilm covering the bottom of the

240 channel, was formed by all strains (poor reproducibility in FAM21808, FAM23078,

241

FAM23106, and FAM22996, Fig 3A) A total of 11 strains showed good

242 reproducibility and consistently produced biofilm within the FC channel Average area

243

coverage percentage graphs of these are given (Fig 3C&D) Importantly, area

244 coverage percentage varied greatly between replicates for most strains, even if they

245

produced biofilm within the FC channel in every replicate For six strains consistently

246 forming biofilm within the FC channel, a sudden decrease of area coverage, between

247 two time points was observed (FAM21805, FAM21843, FAM22871, FAM22936,

248 FAM23012, FAM23093 and K-12 MG1655) This sloughing off of biofilm material was

249 observed in individual replicates only It is illustrated for strain FAM21843 at 18 vs

250

18.5 h of incubation (Fig 3B) The fastest strains reached 5% average area

251 coverage within 4.5 h (K-12 MG1655) and 5.5 h (FAM21805 and FAM22942), while

252 FAM23101 exceeded 5% average area coverage only after 14 h The highest

253 average area coverage, 76.3%, was reached by K-12 MG1655 after 16 h incubation

254 FAM22962 was the weakest of the consistent biofilm formers and its average area

255 coverage was increasing until the end of the experiment (13.3% after 24 h) We note

256 that FAM21845, the strongest, most consistent biofilm former on PS surface, did not

257 form biofilm in the Bioflux FC in any replicate

261 yielded an OD / ODc ratio of greater than one, and did therefore score ‘1’ in our

262

categorization scheme (Table 2) The average ratios of the top five strains were 1.30,

263 0.92, 0.91, 0.87 and 0.86 for FAM21845, FAM23113, FAM23092, FAM21805, and

264 FAM23014, respectively The ratio of FAM21845 was significantly higher than that of

265 all other strains taken together (Mann-Whitney rank sum test of individual replicates,

266

p = 0.004) When comparing strains pairwise, biofilm formation of FAM21845 on SSC

267 was significantly stronger than that of all other strains except FAM23113 and

268 FAM23092 (one-tailed t-test, p < 0.05)

269

FAM21845 encodes antimicrobial and heavy metal resistance genes

270 FAM21845 showed the highest IAR and was the best biofilm former of the strains

271 tested in this study under static conditions on PS and SSC Assays on SSC were

272 performed at 12°C in milk like medium, an assay mimicking conditions encountered

273

in the dairy/food industry This strain is also heat resistant and MDR (Table 1) For

274 these reasons, and to be able to link some of the observed phenotypes with the

275 genotype, and as a basis for future functional genomics or systems biology studies

276 (40, 41), we decided to sequence the entire genome of FAM21845

282

resistance gene database (BacMet (42), Table S3) Among these are

resistance-283 nodulation-cell division (RND) efflux pumps (AcrAB, AcrAD, AcrEF, MdtABC and

284 MdtEF each with TolC and CusCFBA) with broad substrate ranges including

285 antimicrobials (43) and these three operons of interest: 1) the arsenic resistance

286

operon ars composed of arsRBC as well as arsRDABC, an extended version of the

287 operon associated with further increased arsenic resistance (44, 45), 2) the silver

293 The first major plasmid is pFAM21845_1 It is a 147.2kb conjugative IncFII plasmid

pLV501 (all E-values < 10-13) It encodes antibiotic resistance genes against

302 all strains) The 54.2 kb conjugative IncX1 plasmid pFAM21845_2 features a TEM-1

303

β-lactamase (bla TEM-1) and we found close homologs for all conjugative transfer

304 proteins (apart from hypothetical ones) of the IncX reference plasmid R6K (all E-

305 value: < 10-39)

306

The FAM21845 locus of heat resistance

307 The locus of het resistance (LHR) is delineated by 5’ and 3’ mobile elements and

308 encodes a total of 16 open reading frames (ORF), with a high degree of

315

slaughter plant in Canada, respectively (49) They do differ between orf4 and 7,

316 where FAM21845 encodes the cell division protein FtsH (same length and only one

974 bp only have one single nt mismatch between the two strains (99.9% identity)

322

The LHR of FAM21845 is 15,080 bp in length from orf1 to 16 homologs with a GC

323 content of 62.3% (AW1.7: 14,087 bp, GC content: 62.0%) Another difference

324 between the two strains is the presence of three 5’ mobile elements in FAM21845

325 (one in AW1.7) Both strains feature one 3’ mobile element We found an additional

342 2,847 nt) More detailed investigation revealed that this high-scoring segment pair

343 was so short due to six deletions in the putative Flu protein of FAM21845 (FluFAM21845,

347

proteins feature an N-terminal ESPR domain (extended signal peptide of type V

348 secretion system, PF13018.3) and a C-terminal autotransporter β-domain

349 (PF03797.16) FluFAM21845 features four and FluK-12 three AIDA domains (adhesin of

350 bacterial autotransporter system, probable stalk, PF16168.2) FAM21845 encodes an

356 major differences: 1) CsgD of FAM21845 is 203 aa in length, while the one of K-12

357 MG1655 is 216 aa long (13 aa deletion in FAM21845 at positions 2 to 14 in CsgDK-12

358 otherwise 100% identical) and 2) FAM21845 encodes an insertion element IS1

362

major constituent of the E coli extracellular matrix involved in binding to abiotic

363 surfaces and adhesion between cells (52) FAM21845 encodes the full operon

372

pFAM21845_2 increases biofilm formation in K-12 MG1655 transconjugants

377 recipient for pFAM21845_1 and _2, respectively (p = 0.024, 2-tailed, paired t-test) As

378 pFAM21845_1 encodes both AMP and TET resistances, we re-streaked LBNAL,RIF,AMP

379 selected transconjugants (putatively pFAM21845_2 only) onto LBNAL,RIF,AMP,TET plates

380

to assess possible co-transfer of pFAM21845_1 None of 120 re-streaked colonies

381 grew, demonstrating that the vast majority of transconjugants selected in this way are

382 indeed only positive for pFAM21845_2, which is in line with the much higher transfer

383 rate observed for this smaller plasmid Transconjugants selected on LBNAL,RIF,AMP,TET

384 were sure to have received pFAM21845_1 Since transfer of this larger plasmid was

385 approximately 3.6 × 102 less frequent than that of pFAM21845_2, we tested 49 of

386 these transconjugants for presence of both plasmids by specific PCRs All 49 were

387 positive by pFAM21845_1 specific PCR (as expected), while 20 were also positive for

388 pFAM21845_2 Transfer of both plasmids, either separate or together, was thus

389 observed Both pFAM21845_1 & _2 feature plasmid maintenance systems, and are

393 phenotype Transfer of pFAM21845_2, but not pFAM21845_1, resulted in statistically

394 significantly increased biofilm formation of the K-12 MG1655 transconjugant in 48 h

DISCUSSION

397

In this study, we analyzed 36 E coli dairy isolates for their biofilm formation

398 potential We found very strong strain specific differences with regard to all aspects of

399 biofilm formation tested, even though all strains were isolated from raw milk cheeses

400 (except FAM22321 and FAM22871 from raw milk) Curli and cellulose production in

401 macrocolonies, for instance, ranged from no production under any condition (i.e

402 FAM22936) to production of both under all conditions tested apart from RPSMdil agar

403

(FAM21843, Fig 1, Table S1) We have found fewer strains switching off curli and

404 cellulose production in LBnoS than in LB when increasing incubation temperature

405 from 28 to 37°C (five and 11 clear cases, respectively) This is consistent with both

406

the temperature below 30°C and the absence of salt increasing csgD transcription

407 and resulting in greater production of these two matrix constituents (37) A (largely)

408 temperature independent production of cellulose and curli (as is the case for

409 FAM21843) has previously been observed (35) The overall biofilm formation score of

410 individual strains in CV assays on PS over all media and temperatures ranged from

411 four (FAM22996) to 42 (FAM21845) We found ABTCAA to be the best medium in

412 terms of maximizing biofilm formation, which is in agreement with previous

413 observations (56) For each media except RPSMdil, 28°C lead to the greatest overall

417 same amount of biofilm within 48 h, it appears that 28°C is a good compromise

418 between these two effects, generally leading to highest biofilm formation In RPSMdil,

419 part of the retained CV may be due to precipitated milk proteins caused by bacterial

420 growth, rather than actual biofilm matrix or cells, as CV binds to proteins (57) as well

was the best overall temperature in RPSMdil in 96well PS plates and why every strain

423 produced at least some biofilm in this media as determined by CV assays, even

424 though there was no visible production of either curli or cellulose on RPSMdil agar for

428 categorize strains into those either consistently or inconsistently producing biofilm

429 within the FC channel The absolute area coverage percentage varied greatly

430 between replicates, even in strains which consistently produced biofilm within the FC

431 channel Some examples to illustrate this point: average area coverage of FAM22936

432 peaked starting at 13.5 h (19.9%) to a maximum of 29.7% at 15 h, after which it

433

decreased to 18.3% at 17 h (Fig 3D) This peak is mainly due to one single replicate

434 that peaks beginning at 13.5 h (28.8%) to the maximum of 85.5% at 15 h and then

435 plummets down to 5.4% at 17 h The other replicates of this strain are much more

436 uniform in this time period, one ranging between 42.2 and 52.5%, the others at 30%

437 and below, without pronounced peaks FAM21843 exhibits a similarly extreme peak

438 (22.6% at 18 h to 67.2% at 18.5 h and back down to 11.7% at 19 h) and another

439 replicate shows a gradual increase up to 49.1% at 18 h, with a drop to 11.2% at 18.5

440

h (Fig 3B) FAM21805 averages 44.5% area coverage after 24 h (Fig 3C) This

441 average is the result of replicates with area coverages as diverse as 92.9% and 5.4%

442

at the 24 h time point This second, lower replicate mentioned had gradually peaked

443 earlier in the run, clearly having formed biofilm at that time (5.0% at 4 h, 26.9% at 8 h

444 and 10.3% at 13 h) The sudden drops in area coverage observed for several strains

445 are likely the result of sloughing events, which are one mechanism of cell dispersion

446 during the late stages of biofilm formation (59) As they did not occur in every

447 replicate, and at different times, they are certainly a contributor to the poor

448

quantitative reproducibility of biofilm formation in FC (even in consistent biofilm

449 formers) Even though K-12 MG1655 is not among the strongest biofilm formers

450 under static conditions (overall CV score 24), and it shows statistically significantly

454 tested (FAM22936, FAM23012, FAM23092, FAM23093 and FAM23101)

458 However, strain FAM21845 stood out in that it was the only isolate we tested that

459 exceeded this background level of CV staining enough to reach an average OD /

460 ODc ratio of greater than one Due to its interesting traits (highest biofilm formation

461

on PS and SSC, highest IAR, heat resistance and MDR phenotype), FAM21845 was

462 fully sequenced

466 for the transcription of these two operons (33)), as we indirectly ascertained its

467

functionality in all E coli strains used in this study via activity of catalase (30, 31) It is

468 important to note that RpoS, depending on the exact background, can have positive

469 (29) or negative (60) effects on biofilm formation CsgD is required for transcription of

470

csgBAC and thus curli synthesis and also positively affects transcription of the bcs

471 operon and thus cellulose synthesis (61) The 13 aa deletion at the N-terminus of

WP_001119291.1) between csgD and csgB likely negatively affect CsgD mediated

478 clearly produced less biofilm and were unable to produce three-dimensional

479 structures of mature biofilm (62) FAM21845 does feature many other biofilm related

480

genes, however The pga operon synthesizing PGA, involved in binding to abiotic

481 surfaces and adhesion between cells (52) and the complete colanic acid synthesis

482

operon (wza to wcaL) (53) are present on the chromosome It is important to note

483 that colanic acid can negatively affect biofilm formation by masking adhesins like

492 _2 are conjugative plasmids, the strain can produce conjugative pili (experimentally

493 proven by HGT experiments), which are associated with increased biofilm formation

494 (65-67) Reisner and colleagues also showed that expression of the F conjugative

495 pilus could functionally substitute for other known adhesion factors such as type I pili,

496 Ag43 (shortened version in FAM21845) or curli (66) Another study showed that even

497

an engineered reduced-genome E coli lacking curli, type I fimbriae,

498 exopolysaccharide polymers and the autoinducer-2 signaling molecule can produce

499

mature biofilm (68) Taking these findings into account, it is not surprising that

500 FAM21845 can be a strong biofilm former, even without curli and cellulose

501 production

505 biofilm former under static conditions (overall CV score 33), which does not produce

506 curli or cellulose and forms no biofilm within FC channels In contrast, FAM23101

507 produces as much biofilm under static conditions as FAM23016, produces no curli or

508 cellulose but consistently forms biofilm within the channel A consistent former of

509 biofilm in FC channels must not necessarily be strong under static conditions either,

510

as exemplified by FAM22962, which had an overall CV score of 11 (rank 34) The

511 different biofilm formation assays correlate well for FAM22996, which is the weakest

512 biofilm former in CV assays (overall score 4), produces no curli or cellulose and also

513 forms no biofilm in FC channels Good correlation between assays is also seen for

514 FAM21843 (CV score 30) This strain is among the stronger biofilm formers under

515 static conditions (rank 9), especially at 28°C in LB, LBnoS and ABTCAA (Ranks 1, 2

516 and 1 in absolute OD / ODc ratios) It produces curli and cellulose on LB, LBnoS and

517 ABTCAA agar at both 28 and 37°C, and also consistently forms biofilm in FC

521 and a pan-genome of more than 16,000 gene clusters (69) Also, strong phenotypic

formation (71, 72) and cell hydrophobicity and resulting surface attachment behavior

525 (73) have also previously been described This helps to explain the imperfect

526 correlation between macrocolony assays and biofilm formation We only found a

527 consistent correlation between cellulose production and CV score at 28°C (all media

528 except RPSMdil, Table S2) Another source of the large differences observed

529 between assays in this study could be the different materials used: PS and SSC for

530 biofilm formation under static conditions, PVC for IAR, and the FC channels used for

531 biofilm formation under dynamic conditions are made from polydimethylsiloxane

532 (PDMS, sides and top) and glass (bottom) It is of note that inconsistent biofilm

533 formation between static and dynamic conditions has been observed before (66)

534 Expanding the Bioflux FC assays to include other media and temperatures, or the

535 use of another FC system entirely, would likely yield different results under dynamic

536 conditions, which may correlate better with the static assays in some cases

537

We can conclude that our strain collective tested in this study is extremely

538 diverse with regard to biofilm formation, and no single assay can adequately predict

539 any given strains behavior in another Biofilm formation must therefore be assessed

544 disinfectants, antibiotics and heavy metals, resulting in ample opportunity for co-

545 selection This can occur by means of co-resistance (resistance factors located on

546 the same genetic element) or cross-resistance (one mechanism providing resistance

547 against more than one agent (75)) Co-resistance in FAM21845 occurs on both the

due to resistance-nodulation-cell division (RND) efflux pumps (AcrAB, AcrAD, AcrEF,

551 MdtABC, and MdtEF each in combination with TolC and CusCFBA), all of which are

552 present in FAM21845 and are known to have several substrates each (43) We note

553 that copper vats, rather than stainless steel ones, are used in many traditional Swiss

554 dairies and may result in some degree of selection pressure and increased

555

transcription of the cus or pco operons The possibilities for co- and

cross-556 resistances, combined with the drastically increased heat resistance and strong

557 biofilm formation make strains like FAM21845 a serious concern for the spread of

558 resistance and persistence factors in the food industry, and conceivably to

559 commensal or pathogenic bacteria in the gastrointestinal tract upon consumption of

560 contaminated foods (27, 28)

561

MATERIALS AND METHODS

562

Media

563 The following media were used throughout this study: Luria-Bertani Lennox broth

564 (LB, 10 g/l peptone, 5 g/l yeast extract, 5 g/l NaCl, pH 7.0), LBnoS (LB without

565 addition of NaCl), tryptic soy broth (TSB, Oxoid, Pratteln, Switzerland), AB minimal

566 media with 0.5% casamino acids as carbon source (ABTCAA, (56)) and reconstituted

567 powdered skim milk (RPSM) full strength (10.5%, wt/vol) and a diluted version

568 (RPSMdil, 0.2%, wt/vol) All media except TSB and full strength RPSM were also

569 used as agar plates (1.2%, wt/vol agar) Overnight (ON) cultures were grown at 37°C

576 slide Visible bubble formation indicated presence of catalase (O2 production) The

577 phylogenetic groups of strains were determined by quadruplex and group C and E

578 specific PCRs (76) and the sequence type by7 allele multi locus sequence typing

579 scheme (MLST Database at UoW (77)) Heat resistance of strains was determined

580

by PCR clpK and orfI and was phenotypically confirmed (< 1log reduction in CFU

581 after 30 min incubation at 55°C) as previously described (18) Antimicrobial

582 resistance profiles were determined according to CLSI guidelines (78) The following

583 antimicrobials were tested: gentamicin (GEN), kanamycin (KAN), streptomycin

584 (STR), chloramphenicol (CHL), tetracycline (TET), nalidixic acid (NAL), ciprofloxacin

585 (CIP), trimethoprim (TMP), sulfamethoxazol/trimethoprim (19/1, SXT), ampicillin

586 (AMP), cefoxitin (FOX), cephalothin (CEF), cefuroxime (CXM), cefotaxime (CTX),

587

cefepime (FEP), aztreonam (ATM), amoxicillin/clavulanic acid (20/10, AMC) and

588 ertapenem (ETP) All strains used in this study and their relevant characteristics are

592

Crystal violet assays on polystyrene surfaces

593 Biofilm formation was assessed by CV assay in 96well plates (untreated PS surface,

594 CytoOne, StarLab, Hamburg, Germany) ON cultures of strains were diluted 1:100

595 into fresh media and 150 µl were added per well (eight wells per strain and biological

596 replicate) Plates were incubated at 12, 28 and 37°C for 48 h After incubation, plates

597 were washed three times with 200 µl dilution solution (8 g/l NaCl, 1 g/l peptone) per

598 well and subsequently stained with 200 µl 0.1% CV solution (Sigma-Aldrich, Buchs,

599 Switzerland) per well for 20 min Staining was followed by three washes with ddH2O

600 and biofilms were dissolved in 200 µl 96% ethanol per well Biofilm formation was

601 assessed by measurement of optical density at λ = 600 nm (OD600) The assay was

602

performed in biological triplicate Biofilm formation in Table 2 is reported in categories

603 defined by OD to ODc ratios (79) Cut-off value: ODc = average ODnegative control + 3 ×

604 SD(ODnegative control) The categories are: OD ≤ ODc: -; ODc < OD ≤ 2 × ODc: 1; 2 ×

605 ODc < OD ≤ 4 × ODc: 2; 4 × ODc< OD ≤ 8 × ODc: 3; 8 × ODc < OD: 4 Overall

606 scores for media and strains are the sums of category numbers in columns and rows

607 respectively For comparison of K-12 MG1655 wt and its pFAM21845_1 and _2

608 transconjugants, the triplicate of the values ODstrain – ODnegative control were used (Table

612 Goodfellow Cambridge Ltd, Huntingdon, England) were treated with professional

613

cleaning in place products for 30 min at 55°C in an ultrasonic bath Treatment with

614 alkaline cleaner (with booster) was followed by acid cleaner treatment

615 (Pasteurreiniger 405, Halaplus and Halacid sauer respectively, Halag Chemie AG,

616 Aadorf, Switzerland) ON cultures of strains to be tested were diluted 1:100 into full

617 strength RPSM Coupons were placed in 6well plates (CytoOne, StarLab, Hamburg,

618 Germany) and submerged in the inoculated media Plates were incubated for 48 h at

619 12°C, coupons removed and washed three times by immersion and slight agitation in

620 dilution solution Coupons were stained by immersion in 0.1% CV solution (Sigma-

621 Aldrich, Buchs, Switzerland) for 20 min and subsequently washed three times in

622 ddH2O The CV was removed by adding the coupons to 10 ml of modified biofilm

623 dissolving solution (MBDS: 10% SDS dissolved in 80% ethanol (80)) in a 50 ml tube

624 and vortexing Two times 200 µl of the resulting stained MBDS were added to a

625 96well microtiter plate for measurement of OD600 The assay was done in technical

626 duplicate (two coupons per strain) and biological triplicate and results are reported as

627 categories of OD to ODc ratios as for the assays done on PS surface in 96well plates

631 (81) A CoverWell perfusion chamber (19 × 6 × 0.5 mm, C18128, Invitrogen) was

632 placed on top of a dry, uncoated PVC microscopy slide (treated in 70% EtOH / 1%

633 HCl ON and washed with ddH2O) and sealed with silicone lubricant ON cultures

634 (TSB, 37°C for 22 ± 2h, 225 rpm) were diluted into citric acid-Na2HPO4 buffer (pH

635 6.6) to OD600 of 0.100 ± 0.005 and pumped through the chamber at a pressure of

636 0.0505 Pa Three separate vistas in the middle of the channel (along the x-axis) were

637 taken every 5 min for 30 min using a 40 × objective (Zeiss Fluar 40 × / NA 1.3, oil

638 immersion) and an inverted microscope with automated stage (Zeiss Axiovert 135

639

TV) Cells were counted for each vista over time and the median number of cells

640 attached for each time point determined The initial adhesion rate was defined as the

641 slope of the linear regression through median number of attached cells over time

642 (with y-intercept 0, unit: cells / (min × cm2)) Measurements were done at least in

643 biological duplicate

644

Biofilm assays under dynamic conditions

645 Biofilm formation under flow conditions was assessed using the Fluxion Bioflux

646 system (Fluxion Biosciences Inc., San Francisco, CA, USA) ON cultures of strains

647 were centrifuged (2 min, 12,000 × g), the supernatant removed and resuspended in

648 the same volume of ABTCAA minimal media We used 48 well low shear plates (0 to

649

20 dyne/cm2 product No.: 910-0047) The cross-sections of FC channels for biofilm

650 formation are 350 µm wide and 70 µm tall The channel roof and sides consist of

651 PDMS, while the bottom is standard 180 µm cover slip glass Channels were first

652 wetted by adding 100 µl ABTCAA to the inlet well and applying 2 dyne/cm2 until small

653 drops formed in the outlet well Inoculation was done by adding 20 µl of resuspended

654 cells to the outlet well and applying 2 dyne/cm2 for 2 s Cells were allowed to adhere

655 for 1 h at room temperature, before 1 ml ABTCAA was added to each inlet well and

656 the experiment was started Runs were done at 37°C for 24 h applying 0.15

657 dyne/cm2, with image acquisition every 30 min using a 5 × objective with bright field

658

A technical duplicate (neighboring channels) and a minimum of three biological

659 replicates were performed for each strain Percent area coverage of biofilm

660 compared to the entire visible part of the channel (based on between four and seven

661 technical replicates) was determined for consistent biofilm formers using Fiji (ImageJ

662 1.51g (82))

663