Stem Cell Microarrays for Assessing Growth Factor Signaling in Engineered Glycan Microenvironments

Abstract

Extracellular glycans, such as glycosaminoglycans (GAGs), provide an essential regulatory component during the development and maintenance of tissues. GAGs, which harbor binding sites for a range of growth factors and other morphogens, help establish gradients of these molecules in the extracellular matrix (ECM) and promote the formation of active signaling complexes when presented at the cell surface. As such, GAGs have been pursued as biologically active components for the development of biomaterials for cell-based regenerative therapies. However, their structural complexity and compositional heterogeneity make establishing structure-function relationships for this class of glycans difficult. Here, we describe a stem cell array platform, in which chemically modified heparinoid GAG polysaccharides were conjugated to a gelatin matrix and introduced into a polyacrylamide hydrogel network. This array allowed for direct analysis of HS contributions to the signaling via the FGF2-dependent mitogen activated protein kinase (MAPK) pathway in mouse embryonic stem cells. With the recent emergence of powerful synthetic and recombinant technologies to produce well-defined GAG structures, a platform for analyzing both growth factor binding and signaling in response to the presence of these biomolecules will provide a powerful tool for integrating glycans into biomaterials to advance their biological properties and applications.

Graphical Abstract

The present study describes the integration of glycosaminoglycan-protein conjugates into a hydrogel-supported stem cell microarray platform to analyze the activity of extracellular glycans in growth factor signaling. Such platforms can enable rapid development and optimization of functional glycomaterials for stem cell-based regenerative therapies.

Introduction

The development of biologically active materials that support cell adhesion and proliferation, while also providing signaling cues to guide cellular differentiation, has enabled the translation of the regenerative capacity of stem cells into clinical applications.^[1,2] The integration of various components of the native extracellular matrix into hydrogels has emerged as a major strategy for generating responsive materials for organoid and tissue engineering.^[3,4] Comprised of hydrated synthetic or biological polymer networks, hydrogels are commonly decorated with peptides or proteins for cell adhesion and supplemented with signaling molecules, such as growth factors (GFs), to promote signaling and differentiation toward desirable cell types.^[5,6,7]

Stem cell arrays, which allow for high-throughput analysis of cellular responses to their environment and culture conditions, have enabled the discovery and optimization of new biomaterials for cell-based applications.^[8] Such platforms have been particularly useful for examining the ability of various protein components of the ECM to enhance cell interactions and functions when introduced into hydrogels.^[9,10] Extracellular glycans, which also provide important biological functions in the ECM but are difficult to access in pure form synthetically or through isolation, have been comparatively less explored as components for biomaterials.^[11,12] For example, extracellular heparan sulfate (HS) polysaccharides, which belong to the family of glycosaminoglycans (Figure 1), are essential regulators of GF signaling and are being pursued as biologically active components of hydrogels for stem cell culture and tissue engineering.^[13] HS polysaccharides comprise chains of alternating N-acetylglucosamine and glucuronic acid residues, which undergo sequential enzymatic modifications to introduce N-sulfation and to partially epimerize GlcA into iduronic acid (IdoA).^[14] Additional O-sulfation is then introduced to produce sulfated domains harboring protein binding motifs. The compositional complexity of HS has made systematic structure-function analysis needed for their integration into biomaterials challenging. Recent advances in chemical^[15,16] and chemoenzymatic^[17,18] HS oligosaccharide synthesis as well as genetic engineering^[19] of HS biosynthetic pathways have produced increasingly large numbers of chemically well-defined HS structures available for examination in the context of biomaterial design.

Figure 1. Stem cell array for rapid analysis of growth factor signaling in engineered glycosaminoglycan (GAG) microenvironments.

A) GAGs covalently crosslinked to gelatin are arrayed and immobilized within a polyacrylamide (PAAm) hydrogel substrate. Signaling responses of embryonic stem cells grown on the GAG array after GF stimulation are assayed directly by immunofluorescence. B) Structures representing the five main families of GAG polysaccharides depicted using the symbol nomenclature for glycans (SNFG) notation.

Arrays comprising isolated or synthetic HS structures immobilized on glass surfaces are routinely used to profile the specificity of HS-binding proteins.²⁰ Developing of platforms to enable multiplexed, on-array analysis of HS-dependent cellular signaling could significantly streamline the discovery of biomaterials that capitalize on the regulatory functions of ECM glycans. An early example of arrays being used to evaluate the effects of HS structures on cellular responses came from Linhardt and co-workers, who studied proliferation of hydrogel encapsulated non-adherent Ba/F3 cells in the presence of chemically defined HS polysaccharides and fibroblast growth factors (FGFs).^[21] The cell-laden hydrogel droplets were printed on glass and exposed to combinations of HS and FGFs as soluble media supplements. Turnbull and his co-workers were able to directly observe activation of the mitogen activated protein kinase (MAPK) signaling pathway after FGF2 stimulation in Swiss 3T3 cells grown on arrays of oligosaccharides derived by partial heparin digestion. The cells were grown as a monolayer on HS oligosaccharides of increasing length (degree of polymerization, DP = 2–18) spotted and covalently immobilized on amine-functionalized glass via reductive amination. ^[22] MAPK activation was quantified by immunostaining for phosphorylation of Erk1/2 kinases and the magnitude of the observed signal scaled with oligosaccharide length.

To fully harness the multiplexing potential of these the array platform, strategies are needed to present HS structures to progenitor cells in a spatially isolated, yet addressable, format. Here, we present a method for the generation of hydrogel-based GAG microarrays for analysis of growth factor-mediated signaling in murine embryonic stem cells (ESCs). By arraying HS-protein conjugates on polyacrylamide hydrogels, we were able to generate stable ECM-mimetic microenvironments with the capacity to bind FGF2 and influence ESC signaling. The binding and activity of FGF2 in these cellular microenvironments was defined by the chemical composition of the HS polysaccharides.

Results

To develop an array for assessing stem cell signaling responses to engineered glycan ECM environments, we sought to present the glycans together with cell adhesion factors in microscopic islands separated by a non-adhesive surface. This would enable multiplexed analysis of a range of glycan structures while minimizing cellular crosstalk. After screening several common surface passivation strategies used for array construction (Figure S1), we found that a thin poly(acrylamide) hydrogel deposited on glass according to a method by Brafman et al.^[23] and spotted with a solution of gelatin (500 μg/mL) in PBS best supported ESC growth in well-separated colonies over 6 days in culture. The gelatin, a commonly used substrate for murine ESC culture, was loaded into the hydrogel in its dehydrated form, which allowed the protein to enter and become entrapped within the crosslinked polymer network.

To test whether GAG polysaccharides may similarly be arrayed and retained within the hydrogel, the dry acrylamide substrates were spotted with gelatin solutions (500 μg/mL) in PBS buffer (10 % glycerol, 0.003 % triton X-100) supplemented with increasing concentrations (50–750 μg/mL) of heparin (12 kDa) as a model HS glycan (Figure 2). Anticipating that the polysaccharide may diffuse out of the hydrogel network under cell culture conditions, we also included conditions where the heparin was crosslinked via its carboxylic acid groups activated in the form of N-hydroxysuccinimide (NHS) esters to the solvent exposed lysine residues in gelatin (Figure 1A). The heparin was activated by treatment with NHS in HEPES buffer (100 mM, pH = 7.4) in the presence of the coupling reagent, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), at 4°C for 18 hours and purified by size exclusion on a PD10 column to remove small molecule reagents and byproducts. We targeted low levels of crosslinking (~ 15 % carboxylic acid crosslinks per chain) to promote the retention of the GAG in the hydrogel network without compromising its GF-binding ability; however, we anticipate the actual frequency of crosslinks to be lower than our target due to competing hydrolysis of the activated NHS ester groups in the heparin chains under the reaction conditions.

Figure 2. Generation and characterization of polyacrylamide-gelatin GAG array substrates for stem cell culture.

A) Fluorescence micrographs and graph representations of FGF2 binding to heparin (Hep) on polyacrylamide-gelatin arrays printed at increasing Hep concentrations (c_Hep = 50 – 750 mg/mL) with or without N-hydroxy succinimide (NHS) crosslinking to gelatin. Assessment of FGF2 binding after 3-hour wash in PBS buffer indicates ~ 5 to 10-fold increase in Hep retention after crosslinking. B) FGF2-binding to crosslinked Hep arrays was assessed over 48 hrs under cell culture conditions. After initial decrease over the first 18 hours, the arrays retain ~ 40 % of FGF2 binding activity for up to 48 hours. C) FGF2, HA-binding protein (HABP), and the anti-CS antibody, CS56, bind selectively to CS, HA, and chemically desulfated heparinoids on the array. FGF2 bound to Hep and chemically desulfated heparinoids in the following order: Hep > 6OD-Hep > 2OD-Hep >> ND-Hep ~ NAc-Hep. Data represent the mean and standard deviation, n = 3 (and C) or 10 (B). Differences between + NHS and - NHS conditions (A) were assessed using a two-sided t-test, df = 8. Statistical analysis in B was performed using one-way ANOVA, p-values were determined by Tukey’s multiple comparisons test. (*p < 0.05, **p < 0.01, ***p < 0.001 ****p < 0.0001, ns = not significant).

The resulting arrays were washed for 3 hours and probed with the heparin-binding FGF2 protein labeled with AlexaFluor 647 (FGF2-AF647) to detect the immobilized heparin and to assess the effect of crosslinking on its retention in the hydrogel (Figure 2A). Both immobilization strategies resulted in a dose-dependent FGF2 binding, with the NHS-crosslinked heparin providing ~ 5- to 10-fold higher signal. To further test the stability of the arrays, acrylamide substrates spotted with the NHS-activated heparin (500 μg/mL) in gelatin (500 μg/mL) were subjected to ESC culture conditions for 48 hours. Over this time, the arrays were probed with FGF2-AF647 to assess heparin retention (Figure 2B) After an initial decrease in binding activity during the first 18 hours, the heparin arrays remained stable with ~ 40 % of FGF2 binding activity being retained after 48 hours.

With a suitable method for heparin immobilization on the hydrogel substrates in hand, we aimed to test that the protein binding specificities of the crosslinked GAG structures within the hydrogel matrix are preserved (Figure 2C). Using the NHS-crosslinking strategy, we arrayed a panel of CS, DS, and HA GAGs as well as heparin polysaccharides chemically treated to selectively remove their 6-O-, 2-O-, and N-sulfates (6OD-Hep, 2OD-Hep, and ND-Hep, respectively). N-desulfated heparin, in which the exposed amino groups were capped as acetamides (NAc-Hep) to better represent native HS structures, was also included. The array was then probed with FGF2, a CS-specific antibody (CS56), and the hyaluronic acid binding protein (HABP). As shown in Figure 2C, FGF2 bound most strongly to the fully sulfated heparin. Removal of 6-O-, 2-O- and N-sulfates resulted in progressive loss of activity, which is in agreement with the known requirements of 2-O- and N-sulfation for FGF2 binding to HS.^[24,25,26] Likewise, CS56 and HABP proteins exhibited high specificity for CS and HA, respectively.

Having confirmed that NHS-crosslinking to the gelatin matrix enhances to stability of the GAG displays without altering the protein binding specificity of the polysaccharides, we set to evaluate the ability of these arrays to support ESC culture. For our cell model, we chose murine ESCs lacking the expression of Exostosin 1 (Ext1), which is a glycosyl transferase responsible for the assembly of HS chains.^[27] In the absence of this enzyme, the Ext1^−/− ESCs lack cell surface HS structures and are unable to engage a range of HS-dependent GFs, ^[28] including FGF2. As such, these mutant ECS are ideally suited to isolate the effects of the arrayed GAGs on FGF2 signaling from those of endogenous HS structures.

The envisioned on-array GF signaling assay would require that the cells formed near-confluent monolayer colonies on the printed heparin-gelatin spots after at least 2 days in culture. In order to suppress endogenous GF production and establish signaling activity baseline, the last 24 hours should be carried out under serum-free conditions. To optimize cell density and colony growth on the array, we seeded increasing number of Ext1^−/− ESCs on the substrates and grew them for 24 hours in embryonic culture media supplemented with leukemia inhibitor factor (LIF) and fetal bovine serum (FBS). The arrays were then washed, and the remaining bound cells were cultured for additional 24 hours in the absence of serum (Figure 3A). While seeding the stem cells too sparsely (20,000 cells/cm²) resulted in slow growth and irregular colony formation, too high seeding density (100,000 cells/cm²) led to rapid proliferation resulting in spot overgrowth and cell detachment. The intermediary seeding density (40,000 cells/cm²) produced consistent monolayers of Ext1^−/− ESCs (Figure 3A), which retained high levels of expression of the embryonic marker, Oct4, and showed no obvious signs of differentiation (via the neural marker, Nestin, Figure 3B). The optimized seeding conditions were further tested in the presence of immobilized heparin printed at 500 μg/mL concentration and under serum-free starvation conditions to ensure no negative effects of these conditions on cell adhesion and growth (Figure 3C).

Figure 3. Optimization of conditions for embryonic stem cell (ESC) culture on GAG microarrays.

A) Ext1^−/− ESCs were seeded at densities of 2 × 10⁴, 4 × 10⁴, 10 × 10⁴ cells/cm² on poly(acrylamide) substrates printed with gelatin (0.5 mg/mL). The cells were assessed for growth and colony morphology over 48 hrs by optical microscopy (scale bar = 500 μm). B) Ext1^−/− ESCs were cultured on gelatin arrays in embryonic media containing LIF for 48 hrs. The cells retained high levels of pluripotency (Oct4) with no significant spontaneous neural differentiation (Nestin) (scale bar = 200 μm). C) Immobilized heparin (500 μg/mL) does not significantly alter Ext1^−/− ESCs adhesion, growth, or colony formation on the array. Cells were seeded at 4 × 10⁴ cells/cm² and cultured for 48 hrs under embryonic conditions (+ LIF), with the last 24 hrs under serum-free conditions to minimize autocrine GF signaling activity. (scale bar = 500 μm)

To establish whether the array format is suitable for directly assessing changes in stem cell signaling in the engineered glycan microenvironments, we chose to examine the activation of the MAPK pathway in response to stimulation with exogenous FGF2. The requirement for HS in the formation of a signaling complex between FGF2 and its receptor, FGFR, has been well established and the signaling response is accompanied by well-characterized changes in the phosphorylation status of downstream kinases (i.e., Extracellular regulated kinase 1 and 2, Erk1/2).^{[ 29,30,31]}

For the on-array FGF2 signaling assay, Ext1^−/− ESCs (40,000 cells/cm²) were seeded on spots printed with gelatin (500 μg/mL) with or without NHS-crosslinked heparin (500 μg/mL) and grown for 48 hours under the optimized embryonic culture and starvation conditions (Figure 4A). The cells were then placed in a fresh serum-free media containing FGF2 (0.5 ng/mL) and stimulated for 15 min at 37 °C. The cells were fixed, permeabilized, and immunoassayed for MAPK activity using antibodies against Erk1/2 proteins and their phosphorylated forms (pErk1/2). Fluorescence from the arrayed cells was detected using a microarrays scanner (Figure 4B) and validated via fluorescence microscopy (Figure 4C). The ratios of fluorescent signals corresponding to the phosphorylated-ERK1/2 (pERK, green) and total ERK1/2 (ERK, red) proteins was used to quantify the signaling response (Figure 4B). We used soluble heparin (s-Hep, 5μμg/mL), which is known to restore MAPK activity in Ext1^−/− ESCs,^[19] as a positive control and a benchmark in our assay. While ERK1/2 protein levels were similar across all conditions, only Ext1^−/− ESCs stimulated with FGF2 in the presence of immobilized or soluble heparin showed significant increase in Erk1/2 phosphorylation (Figure 4B). We performed fluorescent microscopy imaging (Figure 4C) and image J analysis (Figure S2) to confirm the co-localization of the pERK and ERK signals and to validate our quantification scheme based on signal detection via microarray scanner. We observed somewhat lower levels of MAPK activity on the arrayed heparin compared to its soluble form (Figure 4B). This may be due to a more limited accessibility of the immobilized heparin to only a subset of FGFRs localized to the point of cell contact with the array. FGF2 stimulation of Ext1^−/− ESCs grown on gelatin spots containing increasing amounts of immobilized heparin (i-Hep, 0–500 μg/mL) showed a heparin dose-responsive ERK1/2 phosphorylation (Figure 4D).

Figure 4. Analysis of MAPK signaling in Ext1^−/− ESCs on heparin array.

A) Ext1^−/− ESCs were seeded (4 ×10⁴ cells/cm²) and grown under embryonic conditions (+ LIF) for 48 hrs, with the last 24 hrs under serum starvation. The cells were then stimulated with FGF2 (0.5 ng/mL) for 15 mins. Levels of pERK and total ERK were assessed via immunofluorescence using an array scanner or microscopy. B) Fluorescence images and bar graph representations of arrays stained with anti-pERK (green) and anti-ERK (red) antibodies. Enhanced ERK phosphorylation in response to FGF2 stimulation was observed only in the presence of immobilized or soluble heparin (i-Hep or s-Hep). C) Fluorescence micrographs of ESC colonies after FGF2 stimulation analyzed by microscopy. (Scale bar = 250μm) D) Dose response in FGF2 stimulated ESCs grown on arrays printed at increasing concentrations of heparin (c_Hep = 0 – 500 mg/mL). No significant increase of MAPK signaling was observed in colonies cultured on immobilized heparin in the absence of FGF2 (500 mg/mL, grey bar). Data and error bars represent mean values and standard deviations for at least 11 replicate colonies per condition. Statistical analysis was performed using one-way ANOVA, p-values were determined by Tukey’s multiple comparisons analysis. (*p < 0.05, **p < 0.01, ***p < 0.001 ****p < 0.0001, ns = not significant)

Conclusions

We have developed an array platform for direct, rapid and multiplexed profiling of extracellular glycosaminoglycan activity on growth factor signaling in live embryonic stem cells. The arrays were generated by printing and physisorption of GAG polysaccharides chemically crosslinked with extracellular matrix proteins onto polyacrylamide hydrogel substrates. The crosslinking facilitated the retention of the polysaccharides on the array during cell culture without altering the protein-binding specificity of the polysaccharides. The immobilized GAG structures were able to facilitate GF-mediated activation of signaling events in live embryonic stem cells which could be detected and quantified using immunofluorescence. This array offers a convenient platform to systematically analyze biological activities of extracellular GAGs in stem cell signaling and to accelerate the development of new bioactive glycomaterials for stem cell-based therapeutic applications by capitalizing on the rapidly expanding repertoire of available synthetic,^[16] chemoenzymatic,^[17,18] and recombinant glycosaminoglycan structures.^[19]

Experimental Section

General chemistry procedures

All chemicals, unless stated otherwise, were purchased from Sigma Aldrich. Purchased starting materials were used as received unless otherwise noted. Heparin and desulfated heparinoids were purchased from Iduron (Manchester, UK). The selectively desulfated heparinoids originated from the unmodified heparin used in this study. Iduron reported the average molecular weight of the parent heparin as 12,000 g/mol. Solvent compositions are reported on a volume/volume (v/v) basis unless otherwise noted.

Instrumentation

Nuclear magnetic resonance (NMR) spectra were collected on a Bruker 300 MHz NMR spectrometer. Spectra are reported in parts per million (ppm) on the δ scale relative to the residual solvent as an internal standard. Brightfield images of live cells were taken using ZEISS Axio Observer microscope. Fixed cells were fluorescently imaged using a Keyence BZX-700 fluorescent microscope. Microarray slides used for protein binding assays were assessed using an Axon GenePix 4000B microarray scanner (Molecular Devices). All microarray experiments were performed using steel spring ProPlate gaskets (Gracebiolabs), which were attached to the array slide.

Preparation of NHS-activated glycosaminoglycans

A HEPES buffer (100 × 10⁻³ _M, 15.0 mL, pH 7.4) was prepared. Then, GAG (0.17 mmol) was dissolved in HEPES buffer (200.0 μL), 4 equiv. per GAG chain (0.68 mmol) of NHS was added via an aliquot from a stock NHS solution in HEPES and stirred overnight at 4 °C to afford heparin NHS-ester. The activated NHS ester solution was diluted in MQ water, loaded onto a PD-10 column, and eluted with 2.5 mL Milli-Q water (MQ H₂O). The solution was lyophilized to afford intermediate heparin NHS-ester (2.0 mg). The resulting heparin-NHS product was dissolved in a PBS solution containing gelatin (10.0 mg/mL, 200 μL). The reaction was allowed to proceed overnight. The crosslinked glycosaminoglycan product was purified through a PD-10 column (2.5 mL loading volume, 2.5 mL elution volume). The resulting solution was lyophilized to afford purified gelatin containing crosslinked glycosaminoglycans as a white solid. The same stoichiometry was used for all glycosaminoglycan conjugates.

Glass slide cleaning

Untreated 25×75 mm glass microscope slides were loaded into a steel slide rack and submerged in a crystallization dish filled with MQ H₂O. The slide rack was washed five times with water, allowing the slides to remain in the last water wash for 30 minutes on a rocker. After 30 minutes, the water was removed and replaced with acetone. This solution rocked for 30 minutes, covered. The acetone was then removed and replaced with methanol, was once more rocked for 30 minutes, covered. The methanol was replaced with a solution of NaOH (0.05 _M), and rocked for 2 hours. Slides were then rinsed three times in MQ H₂O and subsequently spin-dried. The slides were then lightly blow-dried using 0.22 μm filtered air. Once dried, slides can then be placed into a vacuum oven to dry at 70 °C and safely stored for up to a month.

Glass slide silanization

Dried, etched slides in a steel rack were placed into a solution of 3-(trimethoxysilyl) propyl methacrylate in toluene (2 %), and rocked for 1 hour. The solution was then removed and the slides were washed three times in fresh toluene to remove residual 3-(trimethoxysilyl) propyl methacrylate. The slides were spin-dried, then blow-dried with 0.22 μm filtered air and placed in a desiccator overnight. For glutaraldehyde activation, slides were rocked for 2 hours in a solution of glutaraldehyde H₂O (0.05 %). The slides were spin-dried (500 rpm, 5 minutes), blow-dried with 0.22 μm filtered air, and placed in a desiccator overnight.

Deposition of acrylamide hydrogel on glass slides

An aqueous 30 % acrylamide solution was prepared by addition of acrylamide (2.85 g), to acrylamide/bisacrylamide (0.150 g, 19:1) to H₂O (10 mL). A separate solution of ammonium persulfate (APS) solution (10 % w/v) was prepared in H₂O. The polymerization solution was then prepared by combining MQ H₂O (985 μL), Acrylamide solution (500 μL), APS solution (15 μL), and tetramethylethylenediamine (0.6 μL). Immediately reagent addition, aliquots (110 μL) of the polymerization solution were placed in the center of each glutaraldehyde-activated methacrylate slide, and cover slip was placed on each slide. After 2 hours, and the slides were loaded into a steel slide rack with the coverslips still on, and allowed to sit in H₂O for 15 minutes, causing the coverslip to loosen on the slide as the hydrogel expands. Slides were removed from water, and using a razorblade, the coverslips were gently removed. The hydrogel exposed slides were then carefully reloaded into the steel slide racks and submerged in a crystallization dish filled with H₂O. The H₂O was replaced every 24 hours for a total of 48 hours of washing. After the 48-hour wash, the slides were spin-dried and placed hydrogel-side-up onto a slide warmer (50 °C, 10 minutes) to partially dehydrate slides for storage.

Printing of GAG arrays

Microarrays were printed using an SpotBot Extreme microarrayer (ArrayIt). Arrays were printed in 65 % humidity using 500 μm spot pins. While the number of spots varies from array to array, spacing between spots was consistently 1400 μm in arrays used for cellular culture. For protein binding assays, spots were spaced 750 μm apart. When designing the spot layout, the print parameter option MAUI4 was selected, and the lateral and vertical offset were 1 and 3 mm respectively. Arrays were printed with porcine gelatin (0.5 mg/mL, bloom 180) in PBS supplemented with glycerol (10 %) and triton X-100 (0.03 %). When concentration gradients we printed, the lowest concentration was always printed last. The gelatin was printed at 500 μg/mL.

After printing, slides are placed into a slide holder and allowed to dry overnight at 4 °C. Prior to use, slides were washed in MQ H₂O for 2 minutes by loading slides into a steel rack and rapidly and repeatedly dipping slides into a crystallization dish full of MQ H₂O. After this, slides were washed in PBS (3x, 15 mins) and then spin-dried. Following this, the slides are snap dried by placing the cells array-side-up onto a slide warmer (50 °C, 10 minutes).

Growth factor binding on arrays

Microarrays used for protein binding assays were equipped with a 4-well gasket chamber and then blocked for 45 minutes with a filtered PBS solution containing bovine serum albumin (BSA) (250 μL, 1 % BSA, 0.5 % tween-20). After blocking, protein binding incubations were performed at 4 °C for 90 minutes in blocking solution with AF647-FGF2 (10 × 10⁻⁹ _M). Between protein incubations, wells were washed four times with blocking solution. After all incubations, a final series of PBS washes (3x, 15 minutes) were performed, the slide was spin-dried, and scanned using a microarray scanner.

Preparation of AF647-FGF2

Human FGF2 (100 μg) was dissolved in of 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (200 μL, 200 × 10⁻³ _M, pH 8.4). Then, a heparin solution (20 μL, 20 mg/mL) in MQ H₂O was added to the FGF2 solution and incubated for 10 minutes, after which a NHS-AF647 solution (2 μL, 10 mg/mL) in DMF was added to the solution. The reaction was gently rocked for 3 hours, and was quenched by the addition of glycine solution (80 μL, 20 mg/mL) in MQ H₂O. To purify the reaction, a heparin sepharose column (1 mL) was prepared and used using a wash buffer (0.5 _M NaCl, 0.2 % BSA, 20 × 10⁻³ _M HEPES, pH 7.4) followed by an elution buffer (3 _M NaCl, 2 % BSA, 20 × 10⁻³ _M HEPES, pH 7.4). The column was equilibrated with wash buffer, loaded and rinsed with 5 column volumes of wash buffer and eluted to yield purified FGF2-AF647.

ESC culture

Ext1^−/− mouse embryonic stem cells were a gift from Dr. Cathy Merry, University of Nottingham, UK. ESCs were cultured feeder free in treated plastic well plates (5 % CO_2, 37 °C). Cells were cultured in ESC maintenance media consisting of Knockout-Dulbecco’s modified eagle medium (KO-DMEM) supplemented with fetal bovine serum (10 %), non-essential amino acids, L-glutamine, 2-mercaptoethanol and LIF. Serum free media is of identical composition to ESC maintenance media except for the exclusion of FBS. Cells were passaged every other day and spilt at a ratio of 1:10 (10⁵ cells/well).

Sterilization of arrays

Arrays were sterilized for cell culture in a laminar flow tissue culture hood by placing arrays and autoclaved gaskets in ethanol for 5 minutes, followed by a sterile PBS wash. The gasket was then assembled onto the slide, and the slide is washed with PBS and left under UV light for at least 15 minutes this is repeated twice, each time with fresh PBS.

Seeding arrays

At least 20 minutes before seeding, LIF containing ESC maintenance media (500 μL) was added to each well of a 4 well gasket attached to an acrylamide slide with an array printed upon it. Cells were then seeded onto the array in a volume of 1 mL, bringing the final volume to 1.5 mL. At 24 hours after seeding, the outlines of gelatin spots became noticeable due to cells growing upon the spots, and the slide was washed once with KO-DMEM and the media appropriate for the desired experiment is added onto the plate in a 1 mL volume.

Growth factor stimulation

Cells were seeded onto microarray wells at a density of 4 × 10⁴ cells/cm² and allowed to adhere for 24 hours in ESC maintenance media. After this time, cells were washed with PBS once to removed unbound cells, and media was switched to serum free ESC media for the next day. Following serum starvation, cells were washed with DPBS and treated with serum free media containing various amounts of FGF2 with or without 5 μg/mL heparin. Immediately following starvation, cells were returned to the incubator for 15 minutes. After this incubation period, cells were placed directly onto ice for immunocytochemistry.

Immunocytochemistry

After stimulation, cells were immediately washed with cold DPBS and fixed for 10 minutes at room temperature in 4 % paraformaldehyde. Then cells were then washed 3x with cold PBS and cellular membranes were permeabilized using cold methanol for 20 minutes. Cells were then washed 3x with PBS and blocked for 1 hour at room temperature with immunocytochemistry (ICC) blocking buffer (3 % (w/v) BSA, 2 % goat serum). The appropriate primary antibody was applied overnight in ICC blocking buffer at 4 °C. Cells were washed 3x with PBS and corresponding secondary antibodies were applied for 1 hour at room temperature. Cells were washed 3x with PBS and nuclei were stained with hoescht for 15 minutes at room temperature. Cells were then washed 3x with PBS and mounted overnight at room temperature using ProLong Gold antifade (Cell Signaling Technology). The next day, cells were subjected to fluorescent microscopy imaging or scanner analysis using an Axon GenePix 4000B microarray scanner (molecular devices), equipped with a Cy3 and Cy5 filter.

Statistical Analysis

All mathematical analyses were performed using GraphPad Prism 9.0. The statistical significance between three or more groups was performed using the built‐in analysis (one-way ANOVA), and Tukey’s multiple comparison test was used to compare means of two groups. Stars and comparisons relate Tukey’s values from ANOVA results, α = 0.05. An unpaired two-tailed t-test was used to compare differences between two groups. Bar graph values represent mean ± SD. Thresholds for significance for all tests is set as *p < .05; **p < .01; ***p < .001; ****p < .0001. For detailed statistics for each experiment, with relevant p values, see Table S3 in the Supporting Information.