Articial Phylloplanes Resembling Physicochemical Characteristics of Selected Fresh Produce and Their Use in Bacteria Attachment/Removal Studies

The recurrence of food-borne illness outbreaks caused by consumption of fresh produce highlights the importance of developing a good understanding of the bacteria-leaf-surfaces interactions. In this study, we proposed and developed a new method to fabricate articial phylloplanes that mimic the topographical and epicuticular characteristics of fresh produce, to be used as a platform for the development of food safety interventions for fresh produce. Romaine lettuce and spinach were selected to create phylloplane replicas using a double-cast procedure. The surface hydrophobicity of the articial phylloplanes made from polydimethylsiloxane (PDMS) was modied by adding a non-ionic surfactant with different hydrophilic-lipophilic balance (HLB) values to match the hydrophobicity of produce leaves. Key epicuticular wax compounds identied from the natural spinach and lettuce leaves were coated on the leaf replica to mimic the chemical composition of natural leaf surfaces. These surrogate surfaces were used to study the attachment Escherichia coli O157:H7 and Listeria innocua. In addition, these surfaces are reusable, and have surface hydrophobicity, surface roughness values and epicuticular wax compositions similar to fresh produce. The articial phylloplanes of fresh produce can be used as a platform for studying the interactions between human pathogens with produce surfaces and for developing new sanitation strategies.


Introduction
Fresh and fresh-cut produce continue to be associated with outbreaks of foodborne illness. This has promoted researchers to develop understanding of the interactions between human pathogens with produce phylloplanes, with the purpose of providing insight into pathogens colonization and persistence on produce surfaces, route of contamination, and means to remove pathogens more effectively or decontaminate produce surfaces [1][2] . Adaxial and abaxial surface of plant leaves provide habitats for a diverse assemblage of microorganisms including fungi, yeast, viruses and bacteria [3][4] . Leafy greens can become contaminated with microorganisms at multiple stages in the farm-to-fork continuum. For instance, during growth, produce surfaces can be contaminated with soil, improperly processed manure, or contaminated irrigation water [5][6][7] . During processing of fresh produce, such as cutting, microorganisms present on outer leaves may come into contact with inner leaves causing cross-contamination of the nal product [8][9][10] . In most cases, bacteria must come into contact with produce surfaces to initiate adhesion and colonization.
The variation in chemistry and topography of plant surfaces oftentimes results in inconsistence in attachment studies [11] . Research regarding interaction between food-borne pathogens and fresh produce leaves is limited compared to that on phytopathogens [12][13] . Most previous studies regarding bacteriafresh produce interaction were focused on the effect of bacterial species, inoculation method, and produce surface characteristics on the attachment of bacteria to produce [14] . whereas most studies with fresh produce are aim at determining the best combination of sanitizer concentration and processing time to remove bacteria [15][16][17][18] .
Studies have shown that the properties of fresh produce phyllosphere such as produce surface roughness, surface hydrophobicity, and epicuticular composition play an important role in bacteria attachment and removal [19][20][21] . Noticeably, these properties undergo constant changes during produce pre-harvest growth and post-harvest transportation and storage, as fresh produce are living plant tissues with active metabolic activities. Produce variety and growth conditions (weather, soil, water, fertilization, etc.) also play a key role determining the produce surface properties. Consequently, the produce surface conditions in sanitation experiments performed by different research groups would not be the same for the same produce type and same sanitizer. As a result, most previous sanitation tests were performed, in a sense, under uncontrolled conditions with regard to produce surface properties. This may be the reason for often observed inconsistent sanitation data reported by different research groups. There is a need to develop a platform for studying bacteria and fresh produce interactions, as well as produce sanitation with controllable and constant surface property phylloplanes that resemble produce surface physical, biochemical, and biological properties.
The requirements for an arti cial phylloplane that can be considered as a replica of a natural produce leaf surface should thus include 1) resembling the 3D topological features of natural produce leaf surfaces, 2) having a similar surface hydrophobicity, 3) having a similar epicuticular chemical composition, mainly epicuticular wax composition, 4) producing a similar bacterial attachment pattern, and 5) reproducible and reusable, including autoclave-able and compatible with stomacher.
In the past decade, some research groups have attempted to develop man-made microstructures that could be used as a replica of natural produce surfaces, with varying degrees of success. The early work of University of California, Davis had developed plant surface structure analogs with a microfabrication method (photolithography) [11,[22][23][24][25] . To study the effects of plant surface microstructure on attachment of Escherichia coli O137:H41, they fabricated uniformly patterned vertical micro-pillars, pyramids, or grooves on polydimethylsiloxane (PDMS) pieces to mimic trichomes, stomata, and ridges between plant cells on produce leaf surfaces, respectively. They used the silanization method to produce hydrophobic surfaces on the silicon, resulting in hydrophobic microstructures similar to those on natural plant surfaces. As the rst documented effort, their goal was to use analogs of trichomes, stomata, and intercellular grooves "with controlled shapes, sizes, and distributions to avoid uncontrolled variables that occur on natural plant surfaces." Therefore, their method only produced man-made arrays of vertical trichomes, stomata, and grooves with uniform shape and size, not the 3D topology of the produce surfaces. The initial work at The University of Illinois was evolved from making simple patterns on PDMS lms [26] to fabrication of PDMS surfaces with 2D patterns of natural produce leaves produced from the SEM image of spinach leaf surface [18] . They also developed a method to modify surface hydrophobicity of PDMS by mixing it with different ratios of surfactant with different hydrophile-lipophile balance (HLB) values and pouring it onto a silicon wafer mold with features resembling the surface of spinach leaves. Their method provided a means to mimic hydrophobicity of any leaf surfaces. But it cannot reproduce a 3D surface topology from produce leaves. A simple two-stage double-casting method to transfer of 3D natural patterns on Trembling Aspen leaf surfaces was developed by McDonald et al. [27] . They used PDMS to produce negative mold of leaf surface. Then a self-assembled monolayer of 2H-per uorodecyltrichlorosilane (FDTS) was used as an anti-adhesion agent to facilitate transfer of micro-patterns to PDMS positive replica. Slightly later, a USDA ARS group reported a method to reproduce 3D fresh produce surface topology on a PDMS lm [28] . After obtaining the negative 3D image of the produce surface on PDMS, they utilized a relatively complex chemical surface modi cation method to coat the negative PDMS surface with a layer of Pd nanoparticles to make a PDMS negative mold. Then the positive PDMS leaf surface analog was produced using a thermal molding method (120 °C, 20 minutes) from the negative mold. The leaf surface replica after the thermal molding process happened to have a water contact angle (WCA) similar to that of spinach. Recently, Doan et al. [3] reported a two-step casting process to generate topomimetic "replicasts" in PDMS that resembled leaf surface topography at submicrometer scale and used it to study microbial colonization of the phyllosphere with P. agglomerans, a plant pathogen to study the removal of food-borne pathogen E. coli from spinach leaves.
Noticeably, almost all previous studies exploring biological surface replica have focused on topological or physical reproduction of the surfaces. Much less efforts have been placed on developing arti cial plant surfaces with similar chemical and biological properties with natural leaves. To ll in this gap, we conducted a comprehensive investigation to develop a PDMS-based arti cial phylloplane surface to resemble the topographical, chemical, and epicuticular characteristics of romaine lettuce and 'Carmel' spinach to high delity. This method enables us to modify and control produce leaf surface properties such as surface roughness, surface hydrophobicity, and epicuticular composition so that to provide an insight into the role played by leaf surface property on sanitation treatment. The PDMS leaf replica was used as a substrate to evaluate attachment and removal of E. coli O157:H7 EDL933 and Listeria innocua. In addition, we examined the reusability of these phylloplane surfaces by exposing them to commonly used disinfection practices in laboratory settings.

Results And Discussion
2.1 Bactericidal effect of PDMS utilized for the development of arti cial phylloplanes.
As seen in Figure 1, the process to develop arti cial phylloplanes involved a double casting procedure utilizing PDMS as the base polymer. The PDMS-double-casting technique has been widely utilized in other applications such as replication of "high-aspect-ratio microstructures", development of components of nanophotonic devices, as well as fabrication of micro uidic devices [29][30][31][32] ., One of the advantages of using PDMS as a base for double casting process is the low cost, low labor and time involved with the procedure. However, one of the obstacles of utilizing PDMS is the high hydrophobicity of the material (WCA=120°), which is different from most fresh plant phylloplanes [33] . Studies have shown that mixing non-ionic surfactants directly with PDMS can lower the hydrophobicity of PDMS [34][35] and thus in our case improve the wettability of the arti cial phylloplanes. Nonetheless, in the food industry a common practice is to utilize surfactant as a component of chemical sanitizers for wash of fresh produce [8,[36][37] . It is therefore necessary to understand if the PDMS surfaces with surface hydrophobicity modi ed by a surfactant is toxic to bacteria. As seen in Figure 2, after up to 24 hours of growth in the non-ionic surfactant solutions with HLB of 7 and 11, there is no signi cant change (or reduction) in E. coli O157:H7 population at each sampling time (0, 2, 12, 24 hrs.) for each of the surfactants tested. Thus, the arti cial phylloplanes with addition of surfactants do not possess a bactericidal to E. coli cells. To further examine the delity of the arti cial phylloplanes to the fresh leaves, the surface hydrophobicity and surface roughness of the natural leaves and PDMS replicas were measured. As shown in Table 1, no signi cant differences (P > 0.05) were observed in the hydrophobicity (water contact angle) values between the fresh leaf of 'Romaine' lettuce and the lettuce arti cial phylloplane made with 10% surfactant. Similarly, no signi cant differences (P > 0.05) in the hydrophobicity values of the 'Carmel' spinach fresh leaf and the spinach arti cial phylloplane made with 10% surfactant (values). Also, we were able to determine that a slightly hydrophobic sample could be achieved by mixing PDMS and surfactants at concentrations <5%. Moreover, as seen in Table 1 no signi cant differences were observed between the surface roughness (mm) of the fresh 'Romaine' lettuce and its PDMS replica, as well as between the fresh leaf of 'Carmel' spinach and the spinach arti cial phylloplane (P > 0.05).

Epicuticular composition of developed arti cial phylloplanes
One important component of fresh leaves is their epicuticular wax, which acts as a barrier that prevents loss of water from the surface of the plant and as a barrier from abiotic stresses [38][39] . Epicuticular wax composition varies depending on the species, cultivar, age and environmental factors. On leaf surfaces, the wax usually exists in the form of a mixture of smooth amorphous layer and hierarchical structures (crystals) [33,[40][41] .
To faithfully replicate produce leaves, besides mimicking the leave topological properties and regulating the PDMS surface hydrophobicity to match that of natural leaves, the arti cial phylloplanes should also represent the leave surface chemical composition, mainly the wax composition. For that purpose, a chemical solution of different long-chain hydrocarbons mixed with chloroform (Table 1) was used to coat the arti cial phylloplanes using a spin coating process. The compounds in the chemical solution were chosen to represent the key compounds of the epicuticular wax according to the work of Lu et al. [33] . After coating, to exam if the PMDS replica surface has been coated with the compounds, a FTIR analysis was performed. Figure 5 shows the infrared spectra for PDMS without wax coating, and for the arti cial phylloplanes of the 'Romaine' lettuce and 'Carmel' spinach coated with wax. Each of the infra-red (IR) active functional groups is highlighted with a band depending on the functional group region. As seen on Figure 5, the alkene bands are at 3090 cm -1 , ketone bands are at 1750 cm -1 , and PDMS silicone groups at 1020-1074 cm -1 . The appearance of the new alkene and ketone bands on the IR spectra of PDMS replica con rms that the wax compounds are deposited on PDMS surfaces. The epicuticular wax in the form of a mixture of small crystals and amorphous layer can also observed on the natural and arti cial spinach leaves in the SEM images in Figure 6 E-F.
2.4 Attachment of E. coli O157:H7 and L. innocua to natural and arti cial spinach leaves.
A comparison of the attachment of E. coli O157:H7 and L. innocua to natural surface and arti cial phylloplanes of the 2 produce types is shown in Table 2. No signi cant difference between the attachment of E. coli O157:H7 and L. innocua to surfaces of the natural and the hydrophobic arti cial phylloplanes, respectively, was found. This nding suggests that bacterial cells may have a similar interaction with the PDMS leaf replica developed in this study and that of a natural biological leaf surface, at least regarding attachment of the bacterial cells to two kinds of surfaces.
The surface hydrophobicity is shown to affect bacterial attachment. Between the 2 arti cial produce surfaces, signi cantly more (P < 0.05) cells were attached to hydrophilic (WCA = 70) surfaces than on the hydrophobic surfaces (WCA = 110). Similarly, and between the natural leaf surface and the hydrophobic arti cial surfaces, signi cantly more (P < 0.05) cells were found on the fresh produce surfaces (WCA = 74 for spinach and WCA =71 for lettuce) than on the hydrophobic PDMS replica (WCA = 110) of them. These ndings are in agreement with Crick et al. [42] who evaluated the effect of hydrophobicity of various surfaces on the attachment of E. coli and S. aureus and found that the hydrophobicity of PDMS reduced the attachment of both types of bacteria compared to hydrophilic surfaces such as glass. The attachment of E. coli O157:H7 onto fresh and arti cial spinach leaves is shown in Figures 6A and 6D. Some attached E. coli cells can be identi ed in Figures 6A and 6D. No signi cant differences in the attachment patterns between fresh and arti cial phylloplanes can be found.

Reusability of arti cial phylloplanes
In order to prove that the arti cial phylloplanes of lettuce and spinach will work as an effective low-cost platform to study factors that promote bacterial attachment, we evaluated the reusability of arti cial phylloplanes after exposing them to two rounds of disinfection with ethanol and two rounds of heat sterilization using an autoclave.
As seen in Figure 7, no signi cant changes in surface hydrophobicity (water contact angle) were observed when the hydrophilic and hydrophobic spinach arti cial phylloplanes were disinfected with 70% (v/v) ethanol or disinfected by two rounds of sterilization at 121°C for 30 min (P > 0.05). Although changes in surface hydrophobicity of up to 7° ± 2° were observed for lettuce arti cial phylloplane disinfected with 70% ethanol and sterilization at 121°C for 30 min (P < 0.05), these changes did not cause the sample to become hydrophobic. Lastly, no signi cant differences in surface hydrophobicity were observed when lettuce hydrophobic arti cial phylloplanes were disinfected with ethanol or sterilization (P > 0.05) Furthermore, as seen in Figure 8, the FTIR spectra shows that after exposing the arti cial phylloplanes of lettuce and spinach to two rounds of sterilization at 121°C for 30 min, no changes in epicuticular wax composition were observed. The signal of IR-active functional groups from the surfaces compared was still identi able. Thus, the arti cial phylloplanes can be used for at least three times for experiments of bacterial attachment.

Conclusions
In this study, a double-casting method to fabricate arti cial phylloplanes that mimic with high delity the physical, chemical, and biological characteristics of fresh leaves of lettuce and spinach was developed with a soft polymer (PDMS). The surface hydrophobicity of the PDMS fresh produce leaf replica was manipulated with addition of non-ionic surfactant with different HLB values to match the hydrophobicity of produce leaves. A method was developed to coat the PDMS leaf replica with the main epicuticular wax compounds extracted and identi ed from the natural spinach and lettuce leaves to replicate the chemical composition of the natural leaf surfaces. Similarities in bacterial attachment patterns between the fresh produce leaves and arti cial phylloplanes were observed. The PDMS leaf replicas are reusable, economical, and recyclable. The arti cial produce leaf phylloplanes can be used as platform to investigate the interactions between bacteria and produce phylloplanes, and to develop new or enhanced fresh produce decontamination strategies.

Greenhouse production of leafy vegetables
"Romaine" lettuce (Lactuca sativa L.) and "Carmel" spinach (Spinacia oleracea L.) were used in this study. They were grown in a greenhouse as previously described [37] . Brie y, lettuce and spinach cultivar seeds purchased from Johnny's Selected Seeds (Winslow, ME) were germinated in 32-cell plant plug trays lled with Sunshine LC1 (Sun Gro Horticulture, Vancouver, British Columbia, Canada) professional soil mix. Seedlings were grown in a greenhouse at University of Illinois under a 25°C/17°C and 14 h/10 h day/night temperature regimen with supplemental lighting. Twenty days post-germination, the seedlings were transferred to 4-liter pots. Leaf tissues from the "Carmel" spinach plants were harvested 40-45 days after sowing seeds and that from the "Romaine" lettuce plants were harvested 50-65 days after sowing seeds. For this study, leaves were harvested at market maturity. Since commercial crop seeds were purchased from a seed company, it is not applicable to the IUCN policy state as it does not involve any risk of extinction.

Preparation of epicuticular chemical solution.
Based on the information presented by Lu et al. (2015), a wax solution containing the key epicuticular wax compounds of "Carmel" spinach leaves were prepared by mixing chloroform with 22% (w/v) of the alkane octadecenol, 54% (w/v) of the fatty alcohol 1-hexacosanol, and 24% (w/v) of the fatty acid myristic acid. The mixture was placed in airtight containers and stirred for 1 hour in a water bath (70 °C ± 2°C). Similarly, a wax solution containing the key epicuticular wax compounds of "Romaine" lettuce leaves was prepared by mixing chloroform with 41% (w/v) of the alkane heneicosane, 20% (w/v) of the fatty alcohol 1-hexacosanol, and 39% (w/v) of the fatty acid myristic acid. The mixture was placed in airtight containers and stirred for 1 hour in a water bath (70 °C ± 2°C). To prevent evaporation and precipitation of epicuticular wax, the wax solutions were kept in airtight containers and placed in a darkroom until further use.

Con rmation of deposition of epicuticular chemical solution on surfaces
Using a Pasteur pipette, approximately 200 mL of epicuticular chemical solution was placed in direct contact with attenuated total re ectance (ATR) crystal on a multibounce plate at controlled ambient temperature (25 °C). An FTIR spectrometer (Thermo Nicolet Nexus 670) connected to the software SPECTRUM ® was used during FTIR data collection. FTIR spectra were recorded from 8 scans at a resolution of 4 cm −1 at 4000-400 cm −1 . These spectra were subtracted against background air spectrum.
After every scan, a new reference air background spectrum was taken. The ATR plate was carefully cleaned in situ by cleaning the sample holder with ethanol twice and dried with soft tissue paper before placing the next sample. Cleanliness was veri ed by collecting a background spectrum and compare to the previous one. These spectra were recorded as absorbance values at each data point in triplicate. from each arti cial leaves with and without epicuticular chemical solution were excised and taped (3M, Minnesota, USA) to a microscope glass slide exposing the adaxial surface of leaves and covered with aluminum foil to prevent contamination with debris and dust particles. Water contact angle of all surfaces was obtained using a goniometer (KSV Instruments, Stockholm, Sweden) model CAM 200. Using a calibrated pipette, 5 mL of deionized water was placed at the center of each disk and within 20 seconds ve contact angle readings were measured.

Determination of surface roughness
Produce leaf surface samples and arti cial leaves were prepared following the same procedure used for contact angle measurement and surface roughness was measured as previously described by Lu et al. [33] . A confocal microscope (NanoFocus, μSurf explorer) was used to determine 3-dimensional surface parameters. Area-average root mean square roughness (-S q bar) was obtained from the average of a number of linear root mean square roughness S q measured from the 3-D image reconstructed from 2-D laser confocal images over an area of . Image analysis was done using software Mountains (Digital Surf, France) 4.5.3 Attachment of Escherichia coli O157:H7 and Listeria innocua to arti cial phylloplanes vs leafy greens surfaces Prior to inoculation, romaine lettuce and spinach leaves were cleaned by rinse step with sterile Milli-Q water to remove debris and patted dried using Kim wipes® (Kimberly-Clark, TX). Each arti cial phylloplane was sterilized using 10 min of UV light. A diluted bacteria solution was prepared by diluting 1 mL of E. coli O157:H7 and L. innocua inoculum in 9 mL of 1X PBS buffer (Initial Inoculum E. coli = 7.0 Log 10 PFU/ml and L. innocua = 8.4 Log 10 PFU/mL). Using sterile tweezers each piece was transferred to an empty sterile petri dish and 100 μL of each bacteria solution in PBS buffer was spot inoculated at 10 different spots on adaxial surface. The petri dish was loosely capped and incubated for 2 hours at 25 °C ± 1 °C in a biological cabinet. After the incubation, the samples were transferred to a sterile container with 1X PBS buffer at ratios of 1:10 (surface: buffer solution) and agitated for 1 min to remove loosely attached bacteria. Afterwards, each sample was transferred to a sterile sampling bag containing 1X PBS buffer and pummeled for 1-minute to remove all bacteria attached to the surface. The remaining supernatant was collected, spread in selective media incubated for 24 hours at 37°C.

Reusability of the arti cial phylloplanes
To determine PDMS-based arti cial phylloplane surfaces endurance to commonly used disinfection practices, they were exposed to two different disinfection procedures. Changes in surface hydrophobicity and epicuticular composition were evaluated. The PDMS-based surface samples were immersed twice in 70% ethanol (v/v) for 36 hours, air dried inside a safety cabinet for 2 hours and stored at 25 °C for 24 hours prior analysis. In addition, PDMS-based surface samples were placed inside an autoclave at 121 • C for 30 min and were stored at 25 °C for 24 hours prior analysis. Surface hydrophobicity was determined following the previously described procedure, while epicuticular composition was determined following procedure in 2.4.1.

Scanning electron microscopy
Surface characterization was carried using a scanning electron microscope. Microimages of the epicuticular surfaces were taken using a FEI Quanta FEG 450 ESEM (Hillsboro, OR, USA). The images were captured under low vacuum at 20 kV and at 400´, 800´, and 1200´ resolution from at least three different samples.

Statistical analysis.
The experiments were performed with a complete randomized design (CRD) with each treatment conducted three times. Bacterial counts were subjected to log transformation before statistical analysis. Data were analyzed using a general linear model available in SAS version 9.1 (SAS Institute, Raleigh, NC, USA), and with Origin-Pro 2016 (OriginLab Corporation, MA, USA). Mean separation was determined using Tukey's test with α= 0.05. Relationships were considered signi cant when the P value was < 0.05.
Tel: 217-244-2571      Con rmation of deposition of epicuticular wax on arti cial phylloplane using FTIR *Highlighted zones indicate the presence of IR-active functional groups. The alkene band is at 3090 cm-1, the ketones band is at 1750 cm-1, PDMS silicone groups at 1020-1074 cm-1.