A Dual-Function Micropatterned Gold Platform for Cell Patterning and Optical Imaging Applications

Abstract

How can we reliably study cellular interactions and molecular responses in well-defined microenvironments while maintaining compatibility with high-sensitivity optical imaging techniques? Addressing this question is critical for understanding tissue development, disease progression, and cellular heterogeneity. Despite the promise of Raman spectroscopy for label-free molecular analysis, a major bottleneck lies in the lack of platforms that combine precise spatial control of cell placement with seamless integration into optical workflows. Here, we present CELLPAC a high-throughput, biocompatible cell-patterning platform that integrates micropatterned gold substrates, self-assembled PEG monolayers, and c-RGD peptides to create microscale adhesive domains for controlled cell confinement. CELLPAC enables the localization and co-localization of single or multiple human keratinocytes into tailored geometric patterns, supporting detailed studies of intercellular communication and migration. Crucially, the platform’s gold component supports surface-enhanced Raman spectroscopy (SERS), and our measurements reveal ~3-fold signal enhancement for intrinsic cellular biomolecules. This dual functionality allows CELLPAC to not only direct cell behavior with high spatial fidelity but also to probe associated molecular processes. With its modular design and optical compatibility, CELLPAC provides a versatile foundation for studying curvature-guided migration, biochemical signaling, and other microenvironment-dependent cell responses advancing biomedical research in areas such as tissue engineering, drug screening, and biosensing.

Graphical Abstract

Introduction

The complex interactions between cells and surrounding microenvironment regulates the development and homeostasis, and abnormal interactions can contribute to a wide range of diseases, including hepatic fibrosis, congenital muscular dystrophy, and osteoarthritis.(1, 2) During these processes, cells navigate an extracellular matrix (ECM) that is not only compositionally diverse and heterogeneous but also hierarchically organized, presenting them with various biochemical surfaces and topologies.(3, 4) Beyond genetic and biochemical signaling pathways, the physical attributes of tissue and its surrounding milieu play a significant role in shaping cellular responses.(5) Importantly, pathological interactions between cells and the microenvironment are associated with disease progression such as cancer metastasis.(6) For instance, early-stage breast cancer cells mechanically align collagen fibers in the tumor microenvironment (7) to facilitate efficient trafficking of exosomes, which subsequently activate fibroblasts to become cancer-associated fibroblasts. It has also been observed that metastatic cancer cells are softer, enabling them to navigate confined spaces in tissues during invasion.(8) Yet, given the complexity of the tissue landscapes in the human body, many of these interactions still remain unexplored. Systematically studying how cells respond to distinct physical cues in their microenvironment is required to gain critical insights into disease progression and inspire targeted strategies for therapeutic intervention.(9) At the molecular level, these cell–cell and cell–matrix interactions frequently involve exchanges of proteins, metabolites, and other signaling molecules. Raman spectroscopy is particularly well-suited for probing such interactions due to its capacity for molecular fingerprinting and its inherent resistance to photobleaching, offering a powerful window into the dynamic processes underlying tissue development and disease.

Cellular micropatterning technology has emerged as a functional tool (10, 11) to enable the systematic studies of cell-microenvironment interactions. It enables manipulation of cell culture substrate to create designated areas, or micropatterns, on which cells can attach, thereby restricting their location and shape. (12–14) Thus far, several cell micropatterning methods have been developed, with most efforts being focused on mechanical microcontact printing method.(15, 16) In this method, a polydimethylsiloxane (PDMS) stamp exhibiting desired microstructures is used to transfer ECM proteins onto the culture substrate.(17, 18) Nevertheless, limitations including initial stamp fabrication requirements, inconsistent protein transfer, and restricted versatility for complex geometries or dynamic microenvironments hinder its broad adoption. Direct photo-patterning circumvents stamp fabrication but often involves specialized chemistry and potential biocompatibility concerns due to photosensitive materials or fluorophores.(19, 20) Yang et al. proposed the use of a digital micromirror device-based modulating projection printing (DMPP) system for the fabrication of poly(ethylene) glycol diacrylate (PEGDA) hydrogel microstructures. (21, 22) While this method offers rapid patterning with high precision, it inherently depends on complex instrumentation which may limit its accessibility to standard cell biology laboratories. Furthermore, the reliance on free-radical polymerization and photoinitiators introduces challenges in ensuring uniformity and reproducibility of the resulting patterns.

In recent years, integrating cell patterning techniques with microelectronics has gained significant traction, particularly in developing multifunctional biochips that enable biosensing and advanced imaging-based assays.(23, 24) However, existing micropatterning platforms seldom incorporate surface-enhanced Raman spectroscopy (SERS) capabilities that could revolutionize the real-time decoding of cell–cell, and cell–ECM interactions due to its high sensitivity and molecular fingerprinting capabilities. To fully leverage SERS and other sophisticated optical analyses, robust patterning methods are required that selectively position cells on plasmonically active substrates without compromising cellular viability or experimental throughput. Yet many conventional micropatterning techniques rely on single-material interfaces, limiting their ability to support multimaterial constructs or conductive domains essential for advanced biosensing applications. Consequently, there is a growing need for a more versatile cell-patterning strategy one that maintains biocompatibility, accommodates multimaterial substrates, and supports a broad spectrum of cellular assays. Despite significant advances, there remains a critical gap in platforms that can simultaneously achieve high-fidelity cell patterning and enable high-sensitivity, label-free molecular readouts using SERS. Our work uniquely addresses this gap by introducing a multifunctional, biocompatible platform that couples spatial control with optical enhancement, advancing both fundamental cell studies and translational biosensing technologies.

Herein, we report the development of a highly robust and reproducible cell patterning platform, “CELLPAC: cell engineering with micropatterned gold films and PEG-assisted cues”, capable of precisely delineating cell populations from single-cell precision to complex multicellular arrangements. Leveraging a combination of micropatterned gold films, self-assembled monolayers (SAMs) of PEG, and cyclic Arg-Gly-Asp (c-RGD) peptide, we engineered micrometer-scale cell adhesive islands. These islands interface with non-adhesive regions by sharp boundaries, facilitating precise guidance for cell attachment and spatial organization. We demonstrate its versatility across various applications, including the patterning of single to multiple cells, conducting cell migration assays, and facilitating the co-localization of distinct cell types. While cellular micropatterning platforms utilizing PEG and RGD peptides have been previously reported, the integration of micropatterned gold films specifically for their intrinsic plasmonic properties and applicability for surface-enhanced Raman spectroscopy (SERS) imaging has remained largely unexplored. The novelty of CELLPAC lies in this unique integration, as gold films offer superior biocompatibility, facile bioconjugation, and critically, the capacity to enhance Raman signals due to surface-supported plasmons. Label-free SERS measurements on the CELLPAC platform revealed an approximately three-fold enhancement in intrinsic biomolecular signals, such as proteins and lipids, exhibiting uniformity with remarkably low variation. By seamlessly supporting both high-fidelity cellular patterning and advanced Raman-based optical imaging within a single, versatile system, CELLPAC enables multifaceted investigations into complex cellular processes. The simplicity, reproducibility, and high sensitivity of CELLPAC underscore its potential to become an indispensable tool for Raman-guided bioengineering applications, including bioprinting, cell-cell interaction studies, drug development, and single-cell analyses.

Results and discussion

CELLPAC design

The CELLPAC design involves two steps (Figure 1a). First, the fabrication of micropatterned gold films, which serves as the base of the platform. Second, the engineering of adhesive islands to facilitate cell adhesion.

Figure 1.

(a) Schematic depiction of the design of CELLPAC. (b) Bright-field images of a series of gold micropatterns exhibiting various geometrical topographies designed for cell patterning applications. Top row from left to right: a pattern of alternating triangular projections and recesses, a linear array of alternating diamond and circular microtopography. Bottom row from left to right: keyhole shape and sinusoidal wave pattern with uniform amplitude.

Fabrication of micropatterned gold substrates.

The detailed steps for fabricating micropatterned gold substrates are illustrated in Figure S1. The process began with the design of the desired geometric shapes using computer-aided design (CAD) software. This initial step enabled precise pattern customization specific to the intended application. After the patterns were designed, they were transferred onto UV photomasks, which are instrumental in shaping the exposure pattern of UV light in the subsequent photolithography steps. To create the adhesive islands, we used a light transmission glass coverslip with 50 nm thick gold film, selected for its common use in biomedical sensing and imaging, as well as gold’s well-documented biocompatibility and functionalization capabilities. A uniform coating of S1813 photoresist was then applied onto the gold-coated glass slide using spin coating. The photomasks were then placed on the photoresist-coated gold slides followed by controlled UV light exposure to make the exposed areas of the photoresist developable. Further, the slides were dipped in a CD26 developer solution, which removed the photoresist from the soluble regions, exposing the gold regions out of the desired pattern region. Gold etching was subsequently performed to remove this excess gold, preserving it only in the specified pattern. Finally, the remaining photoresist was removed using acetone, revealing the micropatterned gold substrate. The rightmost panel of Figure S1 and Figure 1b shows a diverse array of potential cell patterning substrate designs. The patterns showcase the ability of CELLPAC to engineer simple planar to increasingly complex two-dimensional structures with variable curvatures and angles. Each topography is engineered to investigate the influence of surface geometry on cell adhesion, orientation, and proliferation. Additionally, gold exhibits exceptional stability at room temperature, rendering micropatterned substrates highly durable and sustainable. This characteristic further emphasizes the versatility and reliability of the CELLPAC platform.

Fabrication of cell adhesive islands.

Creation of cell-adhesive surfaces is crucial for effectively capturing cells on a substrate and controlling their spatial arrangement. To achieve this, we engineered both cell adhesive and non-adhesive regions. PEG is a well-known biocompatible polymer widely used for creating biologically inert interfaces.(25, 26) In CELLPAC we initiated the process by fabricating non-adhesive regions on the gold micropatterns through a SAM of PEG, employing thiol-gold chemistry (Figure 2). Our coating solution comprised a mixture of PEG-thiol and biotin-PEG-thiol in a ratio of 9:1. Leveraging the inherently low cell adhesion properties of PEG, we hypothesized that the PEG coating would effectively impede cell attachment to the gold region of the micropattern, thereby directing adhesion exclusively to the glass region. The biotin-PEG-thiol fraction was incorporated to enable further surface functionalization. Upon immobilization on gold via thiol-gold chemistry, the biotin groups remained available for binding to Neutravidin, which in turn served as a molecular bridge for anchoring biotin-labeled ligands such as c-RGD. The biotin-PEG-thiol was included in the coating mixture to enable subsequent modification of the gold region with cell-adhesive ligands, using biotin-Neutravidin chemistry, which was performed after initial cell confinement on the glass region. Figures 2b and S2 illustrate bright field (BF) images of HeLa cells seeded onto a micropattern with the gold region coated with thiol-PEG, where cells adhered solely to the glass region of the pattern. To promote cell binding or initiate cell migration from the glass to the gold region of the micropattern, we employed biotin-Neutravidin conjugation chemistry to modify biotin-PEG-thiol on the gold region with biotin-conjugated c-RGD peptide. The c-RGD peptide, derived from the cell-binding domain of fibronectin protein, exhibits a high binding affinity towards αvβ3 and α5β1 integrins on the cell surface, rendering it an effective cell adhesive ligand.(27) Compared to the linear form of RGD peptide, the cyclic form has superior binding affinity and enhanced cell adhesion efficiency.(28) Therefore, c-RGD was used for creating cell adhesive regions. Following the coating with c-RGD, cells were able to adhere to the gold region (Figures 2c). It illustrates the broad adaptability of the developed technology to create customized cell patterning substrates engineered to meet specialized application requirements. To ascertain that cell adhesion on the gold region is indeed mediated by c-RGD, rather than nonspecific fouling, time-lapse movies were acquired (Video S1 and Figure S3) to monitor cells seeded onto a PEG-coated micropattern. The absence of cell migration or adhesion to the uncoated region, even after 360 minutes post-seeding, confirms that cell adhesion was indeed specifically mediated by c-RGD.

Figure 2.

(a) Schematic illustration of the fabrication of cell-adhesive and non-adhesive regions. The gold region of the micropattern was coated with a mixture of PEG and biotinylated-PEG (9:1 ratio) to prevent cell adhesion on gold. Cells were then seeded onto the pattern, forming a confluent layer on the glass region. To facilitate cell adhesion on the micropatterned area, c-RGD peptide was conjugated using biotin-Neutravidin bonding. BF images of HeLa cells fabricated on gold micropatterns with gold region coated with (b) PEG and (c) after c-RGD coating.

Modulating cell confinement from single to clustered arrangements using CELLPAC

Next, we examined CELLPAC’s ability to precisely organize cells into prescribed shapes and patterns. This involved organizing cells both individually and/or collectively, as shown in Figures 3 and S4. Utilizing CELLPAC, we accomplished high-efficiency cell patterning in uniform array-like structures, spanning an area of approximately ~ 4 × 10⁵ μm², and encompassing various geometric shapes such as triangles, circles, and squares. Gold micropatterns were designed using the photolithography method as described in the previous section, then appropriate numbers of cells were introduced to completely cover the adhesive regions, with some excessive cells not adhered to the surface. The non-adherent cells were subsequently removed by washing using the culture media. We observed exclusive adhesion of the cells to the intended region coated with c-RGD, as confirmed by imaging the focal-adhesion organization by GFP-tagged paxillin. Fluorescence imaging of GFP–paxillin (Figure 3a) reveals focal adhesions concentrated along the island periphery, with pronounced enrichment at geometric vertices. Notably, with the triangle and circular designs, we observed that 100% of the coated areas were populated by cells (Figure S4(a-b and d). For smaller patterns intended for singular cells (shown in Figure S4c), 86% occupancy by cells was achieved, indicating the platform’s high efficiency. The results also showcase the uniqueness of the CELLPAC platform which is its high degrees of freedom in generating arbitrary geometries and sizes. This can be easily accomplished by modifying the design using a CAD software. By changing the area and the shape, the number of cells included in an island can range from single to multiple (Figures 3b, S4d, and S4e).

Figure 3.

(a) Geometry-dependent confinement of single U2OS cells on micropatterned adhesive islands and corresponding focal-adhesion organization. Top row: GFP–paxillin revealing focal adhesions (green) accumulating along island boundaries and intensifying at the vertices of polygonal shapes. Bottom row: GFP–paxillin fluorescence signal overlaid on transmitted-light images highlights cell confinement within each motif. (b-c) The distribution of actin in keratinocytes seeded into the triangular confinement. From left to right: BF image of the confined cells, F-actin stained using phalloidin Alexa 647, nucleus stained using DAPI, and overlay of the fluorescence images.

Confinement geometry and available adhesive area markedly shape cytoskeletal architecture. We therefore analyzed F-actin organization across triangular islands of increasing size (Figures 3b-c, S5). Regardless of island size, phalloidin-Alexa 647 staining revealed a pronounced cortical actin rim outlining the adhesive perimeter. In single-cell islands (Figure 3b), this rim formed an unbroken ring, whereas in multi-cell islands (Figures 3c and S5), thick actin bundles accumulated along cell–cell junctions and at the triangle’s corners, generating tension-bearing stress fibers. Quantification of fluorescence intensity of F-actin from the edge and inside region of cells showed higher intensity at the edges (Figure S5c). Such spatial redistribution enables epithelial cells to accommodate external constraints, preserve shape, and sense substrate geometry processes essential for wound healing and stem-cell differentiation. The presence and patterned localization of stress fibers validate that our confinement platform elicits the expected mechanobiological response and offers a quantifiable link between geometry and cytoskeletal tension. While understanding the underlying molecular pathways is beyond this methods-focused study, CELLPAC’s high-throughput, RGD-selective patterning capable of producing single-to multi-cell motifs of diverse shapes in a single run provides an ideal foundation for future mechanistic investigations and for high-content cytoskeletal screens, tissue-engineering constructs, and in-vitro cell-function assays.

Co-localization of diverse cell types for generating patterned co-cultures

Next, we sought to demonstrate that CELLPAC platform can spatially organize different types cells in a co-culture setting with precision. Common practices of co-cultures rely on the self-organization of different cell types to form physiologically relevant patterns.(29) But not many immortalized cell lines are capable of self-organization when subjected to cocultures, limiting the utility of these cells in applications such as drug screening in tissue-mimetic in vitro systems (30, 31). To illustrate a foundational step toward overcoming this challenge, we designed gold micropatterns with different topographical features (Figure 4), and modified surface affinity through selective coating techniques. The design of the patterns was chosen as a simple demonstration of the potential to deposit different cell types in a format appropriate for biosensor or drug screening applications.

Figure 4.

BF and fluorescence images of two cell types deposited in different spatial arrangements using the CELLPAC platform. The cells that appear green in color are GFP⁺ keratinocytes; the non-fluorescent cells are GFP^- keratinocytes. The first column’s BF images were captured at low exposure to highlight cells within glass regions, whereas the second column’s BF images, taken at high exposure, reveal cells positioned on the gold surface. The third column integrates these images, presenting a comprehensive view of cell distribution across the entire micropattern.

Following the micropattern design, we coated the gold regions with a PEG-thiol and biotin-PEG-thiol mix, promoting selective adherence of GFP-expressing (GFP⁺) keratinocytes to form a dense monolayer on the glass areas, while the PEG’s non-adhesive properties left the gold regions unoccupied. Further, the introduction of c-RGD converted these areas into cell-adhesive zones, allowing us to seed GFP^- keratinocytes subsequently, resulting in GFP⁺ and GFP^- populations occupying separate regions (Figure 4). To rule out the potential quenching effect of fluorescence by the gold surface, we conducted a control experiment. In this experiment, GFP⁺ cells were seeded on an uncoated micropattern, where cells adhered to both the glass and gold regions. The results showed minimal variation in fluorescence intensity (Figure S6), indicating that the presence of gold did not interfere with or quench the fluorescence emitted by the cells. This confirmed that the gold surface has no significant impact on the fluorescence signal. While not demonstrating functional crosstalk between cell types, this proof-of-concept co-patterning establishes a technical foundation for future co-culture assays. The CELLPAC platform’s ability to spatially organize distinct cell populations offers a promising basis in the cocultures is an attractive feature for studies of interactions between different cell types.

Cell migration assay using the CELLPAC platform

Curvatures presented in the tissue environment to cells regulate mechanosignaling linked to important biological processes. For instance, the ECM mimicking nanostructures like grooves and fibers have been known to direct stem cell adhesion and maturation (32, 33). It was also observed that cell-cell adhesion and myosin activity regulate cortical actin assembly in mammary gland epithelium on curved surface.(34) Beyond the simple curvatures reported in the past studies, complex topographical patterns can be found in the body too. One well-known example is dermal papilla.(35) Our CELLPAC platform can readily mimic complex topographical patterns.

Emulating the geometry of dermal papilla, we developed a sinusoidal-shaped micropattern. To examine how cells navigate across contoured surfaces, keratinocyte cells were seeded on the glass region of the micropattern such that a confluent layer is formed (Figure 5). Following coating by c-RGD cells were imaged at different time points to monitor their migration from glass to the gold region (Figure 5a). We observed that keratinocytes migrated more rapidly onto negative curvature regions compared to positive curvature regions, as estimated from the area covered by cells at different time-points (Figure S7). Interestingly, cell advancement on positive curvatures stalled temporarily, resuming only after adjacent negative curvature regions were fully occupied by cells, underscoring a curvature-dependent migration hierarchy. The CELLPAC platform thus enables controlled initiation and progression of cell migration by spatially confining initial seeding, temporally modulating surface adhesiveness via c-RGD coating, and incorporating topographical cues through curvature, allowing systematic investigation of migration dynamics. The preliminary results serve as a precursor to in-depth studies into the mechanisms dictating curvature-influenced cellular dynamics. While the underlying causes of these behaviors remain to be fully elucidated, CELLPAC’s versatile design capabilities offer a promising avenue for future research.

Figure 5.

(a) BF images of cell migration on sinusoidal-shaped gold micropattern captured at different time points (Scale bar −150 μm). (b) Sequential high-resolution fluorescence images capturing the organizational dynamics of F-actin and nuclei in keratinocytes post-migration initiation.

Label-free SERS imaging of human keratinocytes using the CELLPAC platform

To demonstrate the utility of the CELLPAC platform for SERS imaging, we performed label-free Raman mapping of human keratinocytes cultured on micropatterned gold substrates (Figures 6 and S8). Raman spectra were collected in both the fingerprint (500–1800 cm⁻¹) and high wavenumber (2500–3000 cm⁻¹) regions. A significant enhancement of Raman intensity was observed, with signals from the gold-patterned regions exhibiting approximately a three-fold higher intensity compared to those from glass-patterned regions (Figures 6d–f). Notably, prominent enhancements were detected at peaks corresponding to proteins (amide I band at 1665 cm⁻¹, Figure 6b) and lipids (C–H stretching at 2855 cm⁻¹, Figure S8). Importantly, the Raman intensities demonstrated remarkable spatial uniformity across gold-patterned regions, reflected by a low standard deviation (±0.0138) and coefficient of variation (1.39%) for the amide I band (1665 cm⁻¹), confirming consistent SERS enhancement (Figure 6f). The enhanced SERS signal on the CELLPAC platform arises from the planar gold microfeatures with well-defined boundaries. These regions promote electromagnetic field localization at the edges, which act as ‘hot zones’ for Raman enhancement. Furthermore, the smooth and continuous gold layer supports uniform enhancement and consistent signal collection from cell–substrate interfaces. This uniformity underscores the reproducibility and reliability of the CELLPAC platform, effectively addressing one of the critical limitations inherent to conventional SERS substrates. The label-free SERS measurements suggest that the CELLPAC approach enables robust, sensitive, and reproducible characterization of multiple biomolecular signatures, thus offering a powerful tool for investigating cell confinement effects, migration patterns, and biomolecular distributions without reliance on external labeling.

Figure 6.

(a) BF image illustrating keratinocyte coverage across both gold and glass regions of the micropatterned substrate. (b–c) SERS intensity maps showing spatial distribution of characteristic Raman signals for proteins (amide I peak at 1665 cm⁻¹) and lipids (C–H stretching at 2855 cm⁻¹), indicating substantial signal enhancement on the gold region. (d–e) SERS spectra of cells on gold and glass regions, highlighting significantly higher Raman intensities from cells positioned on gold substrates. The pink and purple shaded regions mark the analyzed protein peak (1665 cm⁻¹). (f) Quantitative comparison of the amide I peak intensities between gold and glass regions, demonstrating significant enhancement on gold (****p < 0.0001). Scale bar: 10 μm. Error bars represent mean ± s.d.

Conclusion

Trapping and patterning cells is essential for probing biophysical cues, studying intercellular interactions, and analyzing single-cell behaviors. In this work, we developed CELLPAC, a cell patterning platform, through the integration of micropatterned gold films, SAMs of PEG, and c-RGD peptides. This combination enabled the creation of microscale extracellular matrix mimicking analogs with well-defined adhesive and non-adhesive boundaries, allowing exceptional spatial control over cell adhesion and organization. Our results demonstrate that CELLPAC supports both single- and multi-cellular patterning, facilitates advanced migration assays, and enables precise co-localization studies across diverse cell types. Importantly, the biocompatible gold substrate allows seamless integration with optical imaging modalities. As a key finding, label-free SERS imaging on CELLPAC yielded approximately three-fold signal enhancement, validating the platform’s suitability for high-sensitivity, spatially resolved spectroscopic analysis of live cells. These findings establish CELLPAC as a dual-function platform that enables both physical manipulation of cell behavior and concurrent molecular characterization. Table S1 provides a comparative analysis of CELLPAC against existing micropatterning and SERS platforms. The unique capability of CELLPAC positions it as a valuable tool for investigating disease-relevant phenomena, such as migration-induced metabolic shifts or curvature-driven changes in protein expression, in well-defined microenvironments. Its simplicity, adaptability, and analytical power underscore CELLPAC’s potential for broad impact in tissue engineering, drug screening, and cell-based diagnostics.

Materials and methods

Materials

The 22 mm square coverslips with 50 nm gold were purchased from Platypus Technologies, LLC. Isopropyl alcohol (IPA) was purchased from VWR International, LLC. CD26 developer was purchased from Rohm & Haas Electronic Materials LLC. Acetone and sulfuric acid (95.0–98.0%) were purchased from Sigma-Aldrich. The photomask was purchased from Front Range Photomask (Las Vegas, NV, United States). Hydrogen peroxide (30%) was purchased from Fisher Chemical. Biotin-PEG-SH (MW 2,000) and mPEG-thiol (MW 2,000) were purchased from Laysan Bio Inc. cyclo [Arg-Gly-Asp-D-Phe-Lys(Biotin-PEG-PEG)] (c-RGD) was purchased from Biosynth. Ethyl alcohol (200 proof, absolute, anhydrous) was purchased from Pharmco. 1× PBS (pH 7.4), Dulbecco’s Modified Eagle Medium (DMEM) (1×), penicillin-streptomycin (10000 U mL⁻¹), and NeutrAvidin protein were purchased from Thermo Fisher Scientific Inc. Fetal bovine serum (FBS) was purchased from Corning. 4 % paraformaldehyde (PFA) was purchased from Biotium.

Methods

Fabrication of gold pattern

The 22 mm square coverslip with 50 nm gold was rinsed by isopropyl alcohol (IPA) for 20 s and then dried by blowing compressed N₂ to the coverslip. Next, the gold surface of the coverslip was adjusted to face up and the coverslip was then plasma cleaned for 5 mins with the power of 150 watts. S1813 photoresist was subsequently coated on the gold surface of coverslip at 4000 rpm for 1 min. The coverslip was then baked at 115 ℃ for 1 min on the hotplate. After that, the coverslip was exposed to UV light with a photomask on top of the coverslip, so the patterns on the photomask were transferred to the photoresist on the coverslip. Photoresist developer CD26 was used for incubating with the exposed photoresist on top of the coverslip for 1 min and the excessive CD26 on the coverslip was then washed by deionized (DI) water. The gold surface was exposed to air in the region where the photoresist was developed by CD26. The gold region that was exposed to air was etched by incubating the coverslip with gold etchant for 20 s. The excessive gold etchant on the coverslip was then washed by DI water. Because a titanium adhesion layer was deposited in between the gold layer and coverslip glass when the gold coated coverslip was fabricated, the coverslip was subsequently incubated with hydrochloric acid (HCl) for 5 s to remove the titanium that was exposed to air. Then the coverslip was washed by DI water to remove the excessive HCl. After that titanium region was removed, the coverslip’s glass region previously below that titanium region was exposed to air.

For the photoresist that was not exposed to UV light, it was removed by incubating the coverslip with acetone for 1 min. After this step, the gold region previously below the unexposed photoresist was exposed to air. The coverslip with the desired gold pattern was finally rinsed by IPA for 20 s and then dried by blowing compressed N₂ to the coverslip.

PEG coating of gold micropatterns

To render the gold regions of the micropattern non-adhesive to cells, thiolated PEG was employed. Prior to PEG coating, the gold micropatterns were cleaned using piranha solution for 5 minutes. A mixture of biotin-PEG-SH (MW 2,000) and mPEG-thiol (MW 2,000) was prepared in anhydrous ethanol at a 9:1 ratio, achieving a final PEG concentration of 10 mg/mL. The cleaned micropatterns were then immersed in the prepared PEG solution under a nitrogen atmosphere and incubated for 48 hours. Following incubation, the micropatterns were rinsed twice with ethanol, followed by three rinses with deionized water. The patterns were subsequently air-dried and prepared for cell incubation.

Coating of cells on PEG-coated gold micropattern

HaCaT and HeLa cell lines were cultured in DMEM, supplemented with 10% FBS and 1% penicillin-streptomycin. Cultures were maintained at 37°C in a humidified atmosphere containing 5% CO₂. PEG-coated gold micropatterns were placed into 24-well plates and pre-incubated with DMEM containing 20% FBS for 20 minutes to render the exposed glass regions conducive to cell attachment. Following pre-incubation, HaCaT/HeLa cells, resuspended in serum-free DMEM, were seeded onto the micropatterns at a density sufficient to achieve greater than 90% confluence in the glass regions. The plates were then incubated for 3 hours to allow the cells to adhere firmly to the glass surface. After 3 hours, the culture medium was replaced with DMEM containing 10% FBS, and the cells were incubated for an additional 12 hours before further processing.

c-RGD coating of gold micropattern

To promote cell adhesion to the gold regions of the micropatterns or to induce cellular migration from the glass to the gold, biotin-labeled c-RGD was employed.(36) Initially, the gold micropatterns seeded with cells were coated with neutravidin. A 5 μL solution of 1 mg/mL neutravidin was added to the cells, which were incubated in 500 μL of DMEM supplemented with 2% FBS for 30 minutes. This step facilitated the binding of neutravidin to the biotin-PEG on the gold surface.

After the incubation period, the cells were gently washed three times with DMEM containing 2% FBS to remove any unbound neutravidin. Subsequently, a 5 μL solution of 1 mg/mL c-RGD was added, and the cells were again incubated in 500 μL of DMEM with 2% FBS for 30 minutes. Following this, the cells were washed three more times with DMEM containing 2% FBS to eliminate excess unbound c-RGD. Finally, the cells were resuspended in DMEM with 10% FBS and were used for further experiments.

Fluorescence and BF imaging

For nuclear staining, 4′,6-diamidino-2-phenylindole (DAPI) was added to the culture medium to a final concentration of 2 μM, and cells were incubated for 30 minutes prior to imaging. To visualize actin filaments, cells were stained with Alexa Fluor™ 647 Phalloidin (Thermo Fisher) according to the manufacturer’s protocol.

High-resolution fluorescence and brightfield images were acquired using a Leica SP8 confocal microscope, equipped with a 63× oil-immersion objective for fluorescence imaging and a 10× air objective for brightfield imaging. Live cell imaging was conducted with a stage-top incubator maintaining stable environmental conditions of 37°C, 5% CO2, and 35% humidity throughout the imaging session. For fixed cell imaging, cells were fixed with 4% PFA for 20 minutes, washed three times with PBS, and then imaged. BF images included in Figure 2 were acquired using a Nikon Eclipse Ni microscope with both transmission and reflection light sources.

Raman imaging

For Raman imaging, human keratinocyte cells were seeded onto micropatterned gold substrate with a seeding density such that a confluent layer of cells was formed. After 12 hours of cell seeding, cells were washed with PBS three times and then fixed with 4 % paraformaldehyde (PFA) for 20 min, resuspended in PBS, and then imaged using a Horiba confocal Raman microscope. Raman cellular imaging maps were acquired using 638 nm laser excitation and a 60× water immersion objective. For each acquisition point, the time and number of accumulations were 35s and 8, respectively. All spectra were processed using custom-written scripts in MATLAB for a tenth-order least-squares polynomial baseline correction with a moving window of 5.(37, 38)

Supplementary Material

ASSOCIATED CONTENT

Supporting Information. Schematic depiction of the fabrication steps of micropatterned gold substrates. Brightfield images of HeLa cells adhered to the micropattern. Control experiment showing no cell migration or cell adhesion on gold region of the micropattern in the absence of c-RGD coating. Fluorescence images showing the distribution of actin fibers in keratinocytes seeded into the triangular confinement. Control experiment showing keratinocytes on both gold and glass region of unmodified gold micropattern.