Visualization of Cells at the Solid-microbe Interface Using Confocal Reection Microscopy With Index-matching Materials

Herein, we demonstrated that the use of index-matching materials (IMM) allows direct visualization of microbial cells maintained at a solid-liquid interface through confocal reection microscopy (CRM). The RI mismatch induces a background reection at the solid-liquid interface, which dwarfs the reection signals from the cells and results in low-contrast images. We found that the IMM suciently suppressed the background reection at the solid-liquid interface, facilitating the imaging of microbes at the solid surface using CRM. Further, we succeeded in temporal imaging of initial biolms directly colonizing the IMM with CRM in a tag free fashion, and thus, it is highly advantageous for probing the dynamics of biolm formation, along with visualization of environmental organisms and newly isolated bacteria, for which transformation methods are dicult to establish.


Introduction
Confocal re ection microscopy (CRM) has been applied to visualize the three-dimensional distribution of intact microbiological samples. Distinct from the better-known uorescence confocal laser scanning microscopy, CRM detects incident light scattered by opaque samples, thereby eliminating the need for epi uorescent tagging 1,2 . CRM, thus, offers a unique solution to analyze the three-dimensional structure of microbial communities such as microcolonies, bio lms, and biofouling-communities in a tag-free fashion 3,4 .
CRM imaging of cells directly adhered to a solid surface however, is often hindered by the refractive index (RI) mismatch between the solid and the liquid phases. Many bacteria exist in a surface-associated aggregate referred as bio lm rather than planktonic state, and bacterial adhesion to the surfaces is a key process for understanding the dynamics of bio lm formation 5,6 . Microscopic studies of bio lm formation typically involve maintaining bio lms on a glass surface for optical access 5,6,7 . RI mismatch at the glassmedium interface causes a strong background re ection of incident light, which dwarfs the re ection signals of cells. For example, the RI of the coverslip and culture mediums such as Luria-Bertani (LB) medium and M9 minimal medium are around 1.5 and 1.3, respectively 8,9 . This difference between the two RIs is large enough to cause background re ection at the interface, and the CRM image at the glass-liquid interface thus suffers from poor contrast 3 .
Here, we demonstrate the use of index-matching materials (IMM) to allow for the direct visualization of microbial cells maintained at a solid-liquid interface using CRM. We examined the effectiveness of two different materials, as IMMs, namely uorinated polymers and hydrogels. We overlaid the IMMs on the glass surface and showed that both materials su ciently canceled the background re ection at the solidliquid interface. We demonstrated that, with the aid of IMMs, CRM can be used to visualize the entire bio lm structure including the parts directly interacting with the substrate surface, without the use of any tags.

Results
IMMs cancel background re ection. The use of an IMM successfully canceled the background re ection at the interface of water and solid surfaces. Figure 1 shows the placement of the IMM and the experimental set-up, where the solid surface with a paint mark is illuminated using a laser and re ected light is detected as a signal in an inverted confocal system. We measured the level of background signals at the solid-liquid interface, with and without uorinated acrylic polymer (MY-133-EA; My Polymers Ltd., RI=1.333) overlaid on glass as an IMM. Figure 2a and 2b show the Z-stack images near the interface without and with the MY-133-EA, respectively. The use of the MY-133-EA makes the paint mark clearly visible, while it is hardly visible without the MY-133-EA owing to the strong background re ection along the Z-axis from the interface. Figure 2c shows the average signal intensities of the eld of view along the Z-axis with and without MY-133-EA. Without the MY-133-EA, a strong peak of background signal appears at the interface. This signal is almost eliminated with the use of MY-133-EA. These results con rm that the IMMs chosen for this experiment can counter the background re ection at the solid-liquid interface.
Cell imaging at the interface between liquid and IMM. IMMs allowed the clear visualization of Schizosaccharomyces pombe cells directly resting on glass using CRM (Fig. 3). We obtained Z-stack images of cells on the glass or on the IMM and represented them as 3D projections of the cells. The cells on the glass were hardly imaged in the range of -0.5-2.0 mm, along the Z-axis, from the interface (Fig.   3a), and the resulting 3D projections were far from the actual cell morphologies (Fig. 3b). In contrast, cells imaged using MY-133-EA were successfully imaged without background re ection (Fig. 3c), and thus, the resulting 3D projections clearly resembled cell morphologies (Fig. 3d). Similar 3D projections were also achieved using other IMMs, including an acrylamide-based hydrogel and an agar hydrogel (Fig. 4). These results indicate that a lower background re ection with the use of IMMs overlaid on the glass results in better visualization of the cells with CRM.
Temporal monitoring of initial bio lms. We temporally monitored the initial bio lm formation on the IMM using CRM (Fig. 5). We obtained Z-stack images of the initial bio lms on MY-133-EA at predetermined time points, up to 11 h, and reconstructed them into 3D projections of bio lm formation. The 3D projections successfully represented the initial bio lm formation process, where cells adhered to the solid phase (Fig. 5a), grew (Fig. 5b), and formed further microcolonies (Fig. 5c, d).

Discussion
In the present study, we imaged microbiological samples on solid surfaces using CRM, by eliminating the RI mismatch between the liquid and solid phases. This was achieved using IMMs as the solid phase. We found that the IMMs su ciently suppressed background re ection at the liquid-solid interface, resulting in high contrast images of cells directly on the solid phase.
RI-mismatch at the solid-liquid interface often induces undesired refraction or re ection in various optical devices such as particles used in a particle imaging velocimetry, optical lenses, and optical bers.
This issue can be eliminated by reducing the difference in RI between the liquid and solid phases using index-matched combinations, such as pairs of immersion oil and glass, mixture oil and epoxy, and p-cymene and polymethyl methacrylate [10][11][12] . Similar to this approach, the use of IMMs was effective in suppressing background re ection (Fig. 2). Thus, we successfully obtained high-contras images of cells, even at the solid surface, using CRM (Fig. 3-5). Quarter-wavelength anti-re ection (AR) coating is another general method used to avoid re ection at the solid-liquid interface. However, the AR coating is not suitable for CRM, which involves the usage of various excitation wavelengths depending on the sample because the AR effect directly depends on the combination of the coating thickness and wavelength of light, unless complex structures such as multilayered, graded, and moth-eye structures are involved 13,14 .
There are some limitations to this study. First, the solid phase is limited to IMMs. The main types of IMMs available so far are uorinated polymers and hydrogels 15,16 . Second, the thicknesses of the IMMs should be greater than the Z-resolution of the confocal systems. This prevents any leakage of re ection signals from the bottom to the top surfaces of the IMMs. Third, IMMs must be of high optical transparency to avoid any loss in the intensity of the incident laser and resultant signals.
CRM is used for temporal imaging of microbial communities in a tag-free fashion 3 . In this study, we successfully imaged microbial samples on a solid surface (Fig. 3,4) and temporally imaged the initial bio lm formation process (Fig. 5). This temporal imaging at the solid surface is highly advantageous for understanding the dynamics of bio lm formation by microbes for which transformation methods are di cult to establish. Thus, a potential application of our study is to probe how the dynamics of bio lm formation change depending on the chemical and physical properties of the solid surfaces.

Methods
Synthesis of IMM layer. We synthesized a layer of each IMM onto glass substrates, including a glass slip (Matsunami Glass Ind., Ltd., Japan, RI=1.5255) and a glass bottom dish (AGC Techno Glass CO., LTD., Japan). An MY-133-EA (MY Polymers Ltd., Israel) layer was prepared by photocuring, wherein we spincoated the MY-133-EA solution on the glass substrates at 4000 rpm for 30 s, followed by photo-initiated curing with a 365-nm lamp under an atmosphere of nitrogen to generate a layer of ~20-mm thick. Poly(Nhydroxymethyl acrylamide) hydrogel was prepared by chemical gelation 16 . A pre-gel aqueous solution consisting of 8 wt% N-hydroxymethyl acrylamide (Tokyo Chemical Industry Co., Ltd., Japan), N, N'-Methylenebisacrylamide (Tokyo Chemical Industry Co., Ltd., Japan), N, N, N', N'tetramethylethylenediamine (Tokyo Chemical Industry Co., Ltd., Japan), ammonium peroxodisulfate (Tokyo Chemical Industry Co., Ltd., Japan), and distilled water, was poured into a silicon frame (SYLGARD TM 184 silicone elastomer; Dow Corning Tray Co., Ltd., Japan) adhered to a glass bottom dish.
The frame was sealed with a plate and incubated at room temperature (25°C) for gelation. Agar gel was prepared by physical gelation 16 . The agar (0.8 wv%, Nacalai Tesque, Inc., Japan) was dissolved in PBS buffer (Fuji lm Wako Pure Chemical Co., Japan) while heating. A droplet of agar solution (150 mL) was placed on a coverslip and covered with another coverslip, followed by cooling at room temperature for gelation to make a layer of even ~0.5-mm thick. One side of coverslip was then slide off gently.
Strains and culture conditions. The strains used in this study were Schizosaccharomyces pombe JY1 and Pseudomonas aeruginosa PAO1 17,18 . S. pombe JY1 was cultured in yeast extract-peptone-dextrose (YPD) medium (BD Bioscience, USA) while shaking (190 rpm) overnight at 30°C. P. aeruginosa PAO1 was cultured in LB medium (Nacalai tesque, Inc., Japan) while shaking (190 rpm) overnight at 37°C. For bio lm formation, the culture of P. aeruginosa PAO1 was inoculated into fresh LB medium supplemented with 100 mM KNO 3 (Fuji lm Wako Pure Chemical Co., Japan) to adjust the optical density of the medium at 600 nm of 0.01 and was placed in a 25-μL frame-seal TM incubation chamber (Bio-Rad Laboratories, Inc., USA) to adhere to the MY-133-EA layer. The chamber was sealed with silicon resin and incubated at 37 °C under aerobic conditions. Experimental set up. Figure 1 illustrates the experimental setup. The sample was illuminated with a 561nm continuous wave laser. The re ected light passed through a half mirror and 1.2 Airy-unit (AU) pinhole and was detected with a photomultiplier tube in the inverted confocal system (NIKON A1; NIKON Solutions Co., Ltd., Japan). The average signal intensities in a eld of view, except for the region of paint mark, were calculated by processing the image using a custom MATLAB (MathWorks, Inc., USA) routine.
We reconstructed 3D projections from Z-stack images using NIS elements software (NIKON Solutions Co., Ltd., Japan).

Declarations Data availability
The data generated and analysed during the current study will be made available from the corresponding author on reasonable request.