Atmospheric air plasma induces increased cell aggregation during the formation of Escherichia coli biofilms

Atmospheric air plasma has previously been shown to be a novel and effective method for biofilm eradication. Here we study the effects of plasma on both microbial inactivation and induced structural modification for forming biofilms. New structures are created from aggregates of extracellular polysaccharides and dead bacterial cells, forming a protective and resilient matrix in which the remaining living cells grow and reproduce under proper growth conditions. The new colonies are found to be more resilient in this state, reducing the efficacy of subsequent plasma treatment. We verify that the observed effect is not caused by chemicals produced by plasma reactive species, but instead by the physical processes of drying and convection caused by the plasma discharge.


Introduction
Biofilms are colonies of microorganisms surrounded by a complex fluid matrix made 15 predominantly of extracellular polysaccharide polymers (EPS). The EPS provides a protective 16 barrier for bacterial colonies in a biofilm, [1] increasing the resistance of bacteria to chemical 17 and antibiotic treatments and also reducing the efficacy of physical treatment. Biofilms can 18 thus survive most conventional methods of eradicating more freely dispersed, or planktonic, 19 bacteria [2] . Biofilms can form on many surfaces, including the skin of fresh fruits and 20 vegetables, industrial pipe surfaces, in between teeth, and on medical devices. [3,4] Due to their 21 widespread existence and resilience, biofilms are known to be the main cause of persistent 22 bacterial infections in hospitals, [5] contamination of foods in process environments, [6] and 23 reduced process cleaning efficiency in manufacturing. Biofilm physical and flow properties 24 have recently been studied as a means of understanding molecular transport through the 25 matrix and to better enable destruction. [7,8] New approaches are being developed to more 26 aggressively treat biofilms during formation, for example to interfere with the attachment of 27 these bacteria to surfaces and disturb their structure. [9] 28 One novel treatment currently being investigated for this purpose is atmospheric plasma, 29 which is essentially an ionized gas that is generated at ambient temperatures and under 30 atmospheric conditions that allows treatment of sensitive biological matter. [10,11] Numerous 31 recent studies have demonstrated the anti-microbial efficacy of atmospheric plasma for 32 planktonic bacteria or cells embedded in biofilms. [12] Plasma species are reported to be 33 capable of penetrating into the biofilm structure. [13] Plasma can inactivate biofilms with 34 treatment times of less than 60 seconds [14] and cause a 5 log reduction in biofilm viability, [15] 35 while longer treatments can decrease viable cells to undetected levels. [15][16][17] This ability of 36 plasma to inactivate bacteria is thought to be an effect of its production of short-and long-37 lived reactive species [18] such as ozone and other radicals. [19] Long-lived species have been 38 F o r P e e r R e v i e w -4 -shown to be effective to treat Escherichia coli suspensions even after a 7-day period, 39 following plasma liquid generation. [20] 40 Apart from its ability to inactivate bacteria in a biofilm, atmospheric air plasma has been 41 shown to change the overall biofilm structure by disrupting and degrading the EPS biofilm 42 components. [21] For example, separation of initially aggregated bacteria has been observed 43 during EPS degradation due to plasma treatment. [22] Plasma-induced EPS degradation causes 44 a decrease in biofilm thickness [21,23] and volume [21] as well as an increase in its roughness and 45 porosity. [21] Plasma-treated biofilms are also known to have reduced adhesion to surfaces. [23, 46 24]

47
In model systems, monolayers of surface-deposited Listeria innocua responded to plasma 48 treatment by forming cell aggregates of damaged cells, into which viable cells were then 49 moved, affecting plasma inactivation kinetics. [28] Bayliss et al [28] suggested such sheltering of 50 cells extends the treatment time needed for bacterial inactivation and is driven by plasma gas 51 flow-induced drying and the resultant fluid shear stresses. Although the work was carried out 52 on a manually-deposited layer of cells, it likely has relevance for more developed biofilm 53 community environments as well.

Plasma setup 73
The power supply used to drive the plasma discharge was an HV half bridge resonant inverter 74 circuit (PVM2000, Information Unlimited, New Hampshire, USA). The power source has a 75 maximum output voltage of 50 kV with a variable frequency of 20 kHz to 100 kHz, 76 depending on the plasma load capacitance. The plasma setup consists of a FluoroDish™ used 77 to grow the biofilm (see section 2.1) that is placed in between the electrodes of the Dielectric 78 Barrier Discharge, or DBD, consisting of a 2 mm thick poly(methyl methacrylate) dielectric 79 and a top electrode that is partially recessed within the imaging dish to reduce the discharge 80 gap to 6 mm ( Figure 1a). The discharges were induced in open atmospheric air conditions. 81 82

Direct treatment 84
The growing biofilms were exposed to direct plasma treatment, Figure 1a  included, the reader is referred to Lu et al [25] for characterisation of discharges with this power 89 source. The DBD design incorporating the dish used to grow the biofilm allows for non-90 invasive sample preparation, which is critical for later imaging of a biofilm's structure. The 91 design also offers the added benefit of a relatively controlled discharge in terms of spatial 92 homogeneity and treatment time when compared to plasma jets. Precise control of treatment 93 time (~1s) allows the effects of short plasma treatment times on biofilm behaviour to be 94 investigated. 95 For time-dependent studies, biofilms aged 48 h were exposed to direct plasma for times 96 ranging from 0 to 60 s. The biofilm was kept wet by adding 200 µL of PBS into the dish. For 97 liquid coverage studies, different amounts of PBS were added to the cell culture dish, from 98 200 µL to 1000 µL, and biofilms aged 48 h were used. In the regrowth study, biofilms aged 99 24 h and 48 h were used and exposed to plasma for 30 s. On each day, biofilms were 100 compared to untreated controls (Table 1). After exposure to plasma, biofilms were incubated 101 again with fresh nutrient broth at 37° C. All nutrients were changed every 24 h until the final 102 day (72 h). 103 104

Indirect (liquid) treatment 105
Plasma-treated liquid was generated by treating 1 mL of PBS in the same setup as direct 106 treatment, as indicated in Figure 1B

Image analysis 122
All images were analysed using Image-J. [26] Green and red channels from CLSM data were 123 separated and then analysed individually to calculate biofilm coverage area. From the 124 literature it is known the approximate size of one E. coli cell is 1μm x 3μm. [27] Assuming the 125 cells are perfectly oval, the area of one E. coli cell is 2.35 μm 2 . Hence, any number that is less 126 than this value is disregarded in the calculation. The percentage of red cells was calculated 127 from total area covered by red cells divided by the total area covered by both green and red 128 cells. Each data set contains at least six fields of view that are used for data quantification. 129 130

Hydrogen peroxide (H 2 O 2 ) measurement 131
Quantification of H 2 O 2 concentration in the plasma liquid was performed following the 132 protocol of Pick and Keisari. [28] Briefly, 5 g of horseradish peroxidase Type II (Sigma Aldrich, 133 Sydney, Australia) powder was dissolved in 0.05 M phosphate buffer. Phenol red dye is used 134 to detect colour change due to the presence of H 2 O 2 , using a concentration of 0.28 mM. 135 Standard curves were then prepared by measuring spectra of milli-Q water containing various 136 concentrations of H 2 O 2 from 0-60 µM. The solution was taken out of the dish, transferred into 137 F o r P e e r R e v i e w -8 -a small glass vial, and incubated for 1 hour before spectra measurement. Just before spectra 138 measurement, 10 µL of the horseradish peroxidase solution and 10 µL of the phenol red 139 solution were added into the standard samples and plasma-treated liquid. These vials were 140 then incubated again at 25° C for 5 mins. After incubation, NaOH was added to the solution 141 to change its color from orange to purple and keep the colour stable. [28] Spectra of samples at 142 610 nm were then recorded using a UV-VIS spectrophotometer (Shimadzu Corporation, 143 Kyoto, Japan). 144 145 146

The effect of plasma treatment on biofilm structure 148
Plasma treatment has been reported previously to destabilize biofilm structures. [21] Here we 149 use an Escherichia coli biofilm that is in a younger state than the previously studied biofilms 150 of Pseudomonas aeruginosa or Staphylococcus aureus. [21] During this early stage of biofilm 151 development, no microcolonies have been formed. Figure  These effects are contrary to those reported by Ferrell et al, [21] with plasma treatment 157 inducing aggregation and forming a new structure rather than structure breakdown. This 158 plasma-induced structural re-arrangement has been observed previously in surface-deposited 159 planktonic bacteria. [29]  bacterial cells in the biofilm. This behaviour has been observed in many studies that study the 169 effect of treatment on bacterial viability. [15][16][17] However, for the current system, it is found that 170 after 40 s the number of dead cells reaches a plateau of 40%, Figure 2b. This indicates that 171 there is a limit to the number of bacteria that can be killed with plasma treatment, perhaps 172 because aggregation offers some form of protection. 173 Of particular interest is that the aggregation of the cells and the mortality effects of the plasma 174 appear to both plateau, although on different time scales, after 40 s for cell viability and after 175 10 s for cell aggregation (Figure 2b). 176 The biofilms used in this study are considered mature once they are 48 h old, but we also 177 examined the effects of biofilm age on aggregation and mortality response to plasma 178 treatment. This is because the amount of EPS increases with biofilm age, and it may play a 179 role in protecting cells from plasma and aggregation induced by plasma. colonies as longer growth time increases cell cluster size. [30] 216 In addition, as seen from Figure 4a, a plasma-treated biofilm consists of only living cells. 217 Analysis shows that despite 30 s of plasma treatment causing cell death of a significant 218 proportion of cells (Figure 4c), only a very small number (< 10 %) of dead cells could be 219 detected after biofilm re-growth. However, it is likely that some dead cells are hidden within 220 the new structure. However, the percentage of these red cells is still quite low, less than 10%, 221 which is not significant. 222 In section 3.2, it was found that after plasma treatment, bacteria in a biofilm can utilize the 227 new structure to reproduce and grow. In previous work by Ferrell et al, [21] a mature biofilm 228 with large aggregates was shown to change structure by increasing the porosity of the biofilm 229 structure. In this kind of mature biofilm, the high amount of EPS should prevent the 230 aggregation of bacteria as this EPS provides elastic resistance to deformation by flow. The 231 plasma-treated biofilm has a structure more similar to the mature biofilm used by Ferrell et 232 al. [21] It is interesting to know if this plasma-mediated structure has a similar behaviour to a 233 mature biofilm. 234 To answer this, biofilms were exposed to plasma after 24 h of growth. This sample is 235 incubated again for another 24 h before exposing this to the second plasma treatment. Figure  236 5a shows that clumping is still apparent in this system. However, quantitative analysis shows 237  [21] work. This also indicates that after 239 a certain point, aggregation is not possible anymore as biofilms might produce enough EPS to 240 resist deformation by plasma. Another explanation is that subsequent plasma treatments can 241 destroy structures formed by previous treatments. 242 Interestingly, Figure 5a also indicates that biofilms that have been previously treated mainly 243 consist of live cells. This result is unexpected as when the sample is treated twice, it is likely 244 that the percentage of red cells should be higher compared to 24 h or 48 h old biofilms. As 245 can be seen from Figure 5c favourable locations with high concentrations of attractants or to avoid repellents, [32] such as 260 chemicals produced by plasma. Although chemotaxis traditionally is known only for motile 261 cells, recent finding shows that chemotaxis might also occur in surface-attached cells. [33] 262 [ 19,34] For this 264 work only H 2 O 2 is measured, for a more comprehensive species diagnostic of PAW using this 265 power source, the reader is referred to our recent publications. [25,35] Figure 6b indicates that 266 the concentration of H 2 O 2 in the liquid increases with increasing treatment time. This 267 behaviour has been seen in plasma-treated water previously, where initially the concentration 268 of peroxide increases linearly before reaching a plateau. [25] 269 If the aggregation observed previously is related to the presence of chemicals produced by 270 plasma reactive species, we should be able to induce such aggregation by adding commercial 271 control. Figure 7b also shows that compared to peroxide only, plasma water increases the 279 extent of clumping by 2 times (from 3% to 6%), which might suggest that presence of other 280 chemicals that also give rise to cell clumping. However, the change in clumping caused by 281 chemicals (~6%) is not as much as the clumping caused by direct treatment (~20%). This 282 suggests that aggregate formation might be slightly affected by chemicals present in plasma-283 treated water, but it is not the main mechanism. Movement of bacteria is also required for 284 aggregation and is likely controlled by plasma discharge-induced flow. [28] 285 Additionally, the use of hydrogen peroxide and plasma liquid here does not cause significant 286 cell death. As shown in Figure 7b, the percentage of cells killed by treatment is very small, 287 less than 2%. These values are similar to the levels in untreated biofilms. This means there is 288 very little effect of plasma-treated water, which is not in agreement with literature as plasma-289 treated liquid has been shown to inactivate bacteria in biofilms. [36,37] But, literature [38,39] has 290 indicated that in order for plasma-treated liquid to be effective in inactivating bacteria, 291 acidified conditions are required. Naїtali et al [38] showed that in plasma-treated water, a 292 bacterial population was reduced from 8 log CFU to 2 log CFU. However, the effect was 293 diminished for buffered plasma liquid where only a minimal reduction was observed. As all 294 experiments here use a buffer solution, PBS, the pH of the solution is not expected to change 295 and become acidified. 296 297

Dilution effect on biofilm structure 298 299
As mentioned before, the formation of ring structure has been observed in surface deposited 300 bacteria, which is said due to drying by plasma jet. [29] This means that there is high possibility 301 that the structure here is also caused by drying. To understand better the drying by our plasma 302 system, we measured how much water removed when exposed to plasma. 303 Table 2 shows that for 30s treatment time, plasma treatment removes between 0.04-0.06 g 304 water from the system by evaporation regardless of the starting amount of water. From this 305 result, it appears that there is a maximum amount of water that can be removed by plasma for 306 the same treatment time. On the other hand, Table 2 also indicates that the percentage of 307 water removed changes depending on the amount of initial liquid covering biofilm. In this 308 case, the maximum of water removed is 32.9% for a biofilm covered with 200 μL of water 309 (Table 2). Additionally, this suggests that after plasma treatment for 30s, biofilms will not 310 completely dry out. Thus, from this observation it is therefore likely that larger volumes of 311 water could reduce the drying and convective effects of plasma treatment in a specified 312 F o r P e e r R e v i e w -15 -treatment time. Interestingly, we have observed that biofilms that were completely dried in an 313 oven overnight have a similar structure to these plasma-treated samples (data not shown). 314 The above experiments were repeated with biofilms present in varying amounts of water and 315 a constant plasma exposure time of 30s. Figure 8

Explanation of structure formation 332
Our results from the previous section indicate that the structure generated by plasma treatment 333 is mainly due to a drying effect. There is a difference in the convection produced by plasma 334 and standard oven, as Figure 10a  The circular pattern observed in Figure 9a resembles Benard cells, hexagonally-ordered 340 structures that spontaneously form in fluids with a convection flow during heating or 341 evaporation. [40] The length scale of this structure is on the order of µm and is similar to 342 structures formed by surface-deposited bacteria, [29] as depicted in Figure 9c. 343 Deegan et al [41] showed that various patterns can be created by changing the conditions of 344 evaporation. Apart from the formation of Benard cells where the deposit forms a ring, Deegan 345 et al [41] also observed the formation of compact structures as we observed in our biofilm 346 ( Figure 9b). As biofilms are known to have a heterogeneous spatial structure, the plasma jets 347 are also generally heterogeneous in their effects on targets, resulting in the two distinct 348 structures observed. Fischer [42] reported the formation of such ring structures only occurs 349 when there is outward flow to replenish liquid evaporating from the edges. 350 The fact that there is a limit of maximum liquid coverage of biofilms for significant 351 convective effects may be related to the conditions required for Benard cell formation in thin 352 films, namely that the thickness be less than 1 mm. [43] In our experiments, water mainly 353 covered the inner area of the FluoroDish™, which has an overall diameter of 23.5mm. 354 Assuming that liquid covers the inner area uniformly and the area is in cylindrical shape, the 355 volume of liquid added to each system allows us to calculate the height of liquid covering the 356 biofilm. It was found that only biofilm containing 200 μL and 400 μL liquid is covered by 357 water layer which is less than 1 mm thick. This agrees with the finding that aggregation of 358 cells is more apparent in those samples. 359 Drying of 200 μL water for 30 s by oven only removed 1.6 ±0.25% water, which is around 20 360 times lower than drying the same amount of water by plasma (Table 2). Probstein [43]

368
Plasma can be an effective treatment for biofilm eradication. However, this study found that 369 plasma can also induce new structures within the biofilm, which can persist after treatment 370 during regrowth. This phenomenon was evident for both young and more mature biofilms. 371 Once such structures form, subsequent treatments are less effective in terms of efficacy, likely 372 due to the surviving bacteria becoming increasingly resistant to plasma. The structures 373 induced for the biofilms tested are similar to those observed previously for plasma-treated 374 surface-deposited bacteria. [29] The observed structures are reminiscent of Benard cells, whose 375 main mechanism of formation is convection. Secondary plasma species formed in the liquid 376 phase were not found to induce the formation of such structures. 377