A new spray‐based method for the in‐vitro development of dry‐surface biofilms

Abstract The inanimate environment immediately surrounding the patient in healthcare facilities is a reservoir of microorganisms embedded in dry‐surface biofilms (DSB). These biofilms, first highlighted in 2012, are increasingly studied, but currently available in‐vitro models only allow for the growth of semi‐hydrated biofilms. We developed a new in‐vitro method under actual dehydration conditions based on the hypothesis that surface contamination is mainly due to splashes of respiratory secretions. The main objective of this study was to show that the operating conditions we have defined allowed the growth of DSB with a methicillin resistant Staphylococcus aureus strain. The second objective was to show that extended‐spectrum beta‐lactamase‐producing Enterobacteriaceae, that is, Klebsiella pneumoniae and Enterobacter cloacae were also able to grow such biofilms under these conditions. Monobacterial suspensions in sterile artificial saliva (SAS) were sprayed onto polyethylene surfaces. Nutrients and hydration were provided daily by spraying SAS enriched with 20% of Brain Heart Infusion broth. The primary outcome was mean surface coverage measured by image analysis after crystal violet staining. The method applied to S. aureus for 12 days resulted in reproducible and repeatable DSB consisting of isolated and confluent microcolonies embedded in extracellular polymeric substances as shown in scanning electron microscopy images. Similar DSB were obtained with both Enterobacteriaceae applying the same method. No interspecies variation was shown between the three strains in terms of surface coverage. These first trials are the starting point for a 3‐year study currently in process.


| INTRODUCTION
The inanimate environment immediately surrounding a patient is a well-recognized source of cross-transmission of microorganisms, including multiresistant bacteria. Indeed, microorganisms disseminated by the patient can remain in his environment for a long time despite regular cleaning and be transmitted to another patient, either by direct contact or by the hands of healthcare workers Otter et al., 2013;Weber et al., 2013). In 2012, Vickery et al. highlighted the persistence of methicillin resistant Staphylococcus aureus (MRSA) inside biofilms on dry clinical surfaces in intensive care units (Vickery et al., 2012). Three years later, such biofilms have been referred to as "dry-surface biofilm" (DSB) .
Unlike fully hydrated biofilms which have been studied for about 40 years (Costerton et al., 1987), the level of hydration in DSB is reported to be 57%-72% but extracellular polymeric substances are thicker (Almatroudi et al., 2018). It has been hypothesized that DSB develops through regular contact with surfaces by the hands of healthcare workers or the patient himself. The hydration phases would result from the diffusion of biological fluids (sweat, blood, urine, sputum, and so forth) or the regular use of chemicals when cleaning surfaces .
To study such biofilms, five laboratory models have been presented in the literature to date. The first, based on the dynamic CDC Biofilm Reactor (Goeres et al., 2005), was described by Almatroudi . It alternates daily hydration phases of several hours on polycarbonate coupons with drying phases of 42-66 h. It was then modified by two authors. Nkemngong first grew standard hydrated biofilms using the same reactor for 48 h on borosilicate glass coupons which were then dried for 24-120 h (Nkemngong et al., 2020). Amaeze et al. (2020) Table A3).

CHRISTINE ET AL.
| 3 of 11 created both a dynamic and a static model. Ledwoch et al. (2019) also alternated hydration and desiccation phases, each for 48 h, but used a sedimentation protocol on stainless-steel coupons inside cell culture plates. Watson et al.(2018), on the other hand, chose the drip flow reactor (Goeres et al., 2009) to first grow hydrated biofilms on stainless-steel coupons in 54 h and then let them dry for 48 h using an aquatic air pump. All of these models include a significant growth phase under fully hydrated conditions so that they lead to the development of semidehydrated biofilms. In addition, they all use rich culture media (Tryptic Soy Broth) to provide nutrients and hydrate the biofilms, which distances them from real-life conditions.
We report here the design and feasibility assessment of a spray-based method for the growth of dehydrated biofilms under real-life conditions, as close as possible to those encountered in a general intensive care unit. The aim of the study was twofold: (

| Bacterial strains
The method was first developed with a clinical strain of MRSA isolated from a blood sample of a patient with catheter-related bacteremia. It was then applied to two ESBL-PE, Klebsiella pneumoniae and Enterobacter cloacae, isolated from inanimate surfaces, respectively, a sink and a patient call button in the old disused intensive care unit of our hospital. These strains were part of our institutional collection and were stored at −70°C on glass beads. Each strain was tested separately.

| Sterile artificial saliva
The solution chosen to dilute the inoculum, provide nutrients, and hydrate the biofilms was sterile artificial saliva (SAS) prepared according to the Artisial ® drug formulation (Biocodex) ( Table A1). Each strain was first tested for survival and growth in this medium (Table A2).

| Inoculum
Each strain was first cultivated in Brain Heart Infusion (BHI) broth for 24 h at 37°C. The resulting suspension (containing approximately 10 9 CFU/ mL) was then diluted 1:1000 in SAS to obtain a final suspension of approximately 10 6 CFU/mL, verified by serial dilutions and spreading on Plate Count Agar. On Day 0, the inoculum was sprayed once on the inner surface of sterile 55 mm PE-LD Petri dishes.

| Nutrition and hydration
Nutrition and hydration of attached cells were carried out once a day, from Day 1 to endpoint, with two successive sprays of SAS enriched with 20% of BHI broth.

| Spraying technique
The

| Biofilm analysis
At the endpoint, Petri dishes were rinsed with sterile distilled water and the attached biomass was stained with Crystal Violet (RAL Diagnostics). One-quarter of the surface was randomly selected online (https://www.random.org), observed in its entirety by light microscopy (×500, immersion objective), and 20 photographs were taken (ICC50E; Leica); 10 of these were randomly selected online for the calculation of the MSC by image analysis using Scion Images software (Scion Corporation) (Table A3).
In addition, one sample per strain was randomly chosen and observed by scanning electron microscopy (SEM) (FEI Quanta 250 FEG; ThermoFisher Scientific). The analysis method is illustrated in Figure 2.

| Data analysis
Quantitative data were expressed as mean ± standard deviation (SD). To examine for between-run, within-run, and interspecies variations, an analysis of variance (ANOVA) was performed (release 9.4; SAS Software).  (Table 1). Moreover, the evolution of the percentage of total surface coverage over time follows a linear trend line with a regression coefficient of 0.98 and a slope of 2.23% (Figure 3). This seems close to the linear part of growth kinetics of oropharyngeal biofilms expressed as percentages of coverage as a function of time (Leonhard et al., 2018) but this will need to be investigated in future experiments.
Due to the slow growth of the biofilm caused by our experimental conditions (nutrient-poor medium, hydration and nutrient spray, and long drying phases), the experiments were stopped before a mature biofilm was obtained. The MSC measured on Day 12 was about 500 times higher than the detection limit (defined in Table A3) of the method and the biomass (adherent cells + extracellular polymeric substances) covered more than 20% of the total surface. In addition, we observed that Day 12 corresponded to the time needed to double the MSC measured on Day 1.
For all these reasons, we considered the biomass obtained sufficient to perform further experiments.
Note, MRSA DSBs analyzed at the endpoint were made of isolated and confluent microcolonies embedded in an external matrix, which was confirmed by SEM observations (Figure 4). In this way, we nutrient medium which exist in other models Amaeze et al., 2020;Nkemngong et al., 2020) but not in reality.
For biofilm analysis, MSC was chosen as the primary outcome as it gives an objective representation of surface conditions while allowing for biomass quantification (Table A3). Other authors assessed biofilms using surface coverage measurements (Flockton et al., 2019;Marion-Ferey et al., 2003;Mountcastle et al., 2021) but the most widely studied criterion in the literature remains bacterial cultivability Amaeze et al., 2020;Ledwoch et al., 2019;Nkemngong et al., 2020;Watson et al., 2018). However, the absence of culturable bacteria in a biofilm does not mean that there is no biofilm because many phenotypic modifications occurring within the biofilm can transiently reduce cultivability (Fux et al., 2005).
This parameter will be thoroughly studied in future experiments.
Finally, we determined the between-run variation (reproducibility) and within-run variation (repeatability) with MRSA strain to validate the method. As the ANOVA tests did not reach significance, neither between-run nor within-run variation was statistically demonstrated (Table 2). However, we note a relatively high coefficient of variation in both between and within-run studies, F I G U R E 5 Dry-surface biofilms obtained with extended-spectrum beta-lactamase-producing Enterobacteriaceae (ESBL-PE) strains at the endpoint. (a) Enterobacter cloacae; (b) Klebsiella pneumonia. One representative picture of a sample observed by light microscopy (×500) and scanning electron microscopy (×2000, ×5000, and ×10000) is given. Focused is done on confluent microcolonies, and isolated microcolonies and bacteria.
which may be due to the manual spraying technique. Indeed, neither the spray itself nor the operator can deliver a perfectly repeatable volume. This represents the main limitation of this first version of our method. Our operating conditions allowed for MRSA strain to grow reproducible and repeatable DSBs that look like environmental DSBs.

| Secondary objective: To apply the method to ESBL-PE
Our method was applied to two ESBL-PE, Klebsiella pneumoniae, and Enterobacter cloacae. Both had already shown their ability to attach to inert surfaces under real-life conditions. The study of ESBL-PE DSBs was of major interest to us because they are widespread in the hospital ecosystem of Martinique. Moreover, their persistence in DSBs has not yet been studied.
Both BLSE-PE strains were able to grow DSBs using this spray method ( Figure 5). They both consisted of isolated and confluent microcolonies embedded in an external matrix.
To confirm these conclusions based on microscopic observations, we studied interspecies variation by performing three independent experiments, one per strain, with six samples each.
The MSCs (N = 6) obtained with each of the three strains were compared by ANOVA ( T A B L E A3 Coverage measurement using Scion Images softwareImage analysis using the "slice density" option and "analyze" tool Step Image Description Step 1: Original photograph (uncompressed TIFF format) after rinsing the surface and staining with crystal violet (light microscopy ×500) Isolated sessile bacteria, isolated or confluent microcolonies appear in violet.
Step 2: The photograph converted into a red and gray image The "density slice" option produces a black-andwhite copy and uses a red/grayscale to separate the colored biomass from the surface background.
Step 3: Coverage measurement 0.27 cm 2 The "analyze" tool measures exclusively the surface covered by the red-colored biomass, expressed in cm². The final result is calculated by taking into account the magnification factor.
Note: Image analysis using the "slice density" option and "analyze" tool. CHRISTINE ET AL.