Engineering the mode of morphogenetic signal presentation to promote branching from salivary gland spheroids in 3D hydrogels

Abstract

Previously we developed a fibrin hydrogel (FH) decorated with laminin-111 peptides (L_1p-FH) and supports three-dimensional (3D) gland microstructures containing polarized acinar cells.

Here we expand on these results and show that co-culture of rat parotid Par-C10 cells with mesenchymal stem cells produces migrating branches of gland cells into the L1_p-FH and we identify FGF-7 as the principal morphogenetic signal responsible for branching. On the other hand, another FGF family member and gland morphogen, FGF-10 increased proliferation but did not promote migration and therefore, limited the number and length of branched structures grown into the gel. By controlling the mode of growth factor presentation and delivery, we can control the length and cellularity of branches as well as formation of new nodes/clusters within the hydrogel. Such spatial delivery of two or more morphogens may facilitate engineering of anatomically complex tissues/mini organs such as salivary glands that can be used to address developmental questions or as platforms for drug discovery.

Statement of Significance

Hyposalivation leads to the development of a host of oral diseases. Current treatments only provide temporary relief. Tissue engineering may provide promising permanent solutions. Yet current models are limited to salivary spheroids with no branching networks. Branching structures are vital to an effective functioning gland as they increase the surface area/glandular volume ratio of the tissue, allowing a higher output from the small-sized gland. We describe a strategy that controls branch network formation in salivary glands that is a key in advancing the field of salivary gland tissue engineering.

1. Introduction

Saliva is a viscous, clear, watery fluid secreted by the parotid, submandibular, sublingual and other minor mucous glands in the mouth. Saliva plays an essential role in maintaining the oral environment as it contains many antimicrobial molecules such as lysozyme, immunoglobulins, histatins, cystatins and lactoferrin that help to ward off infections while also facilitating palatability during the chewing of food [1]. Moreover, it aids mechanical cleansing and remineralization within the oral cavity [1]. Salivary gland function is mediated by the muscarinic M3 receptor, as stimulation of this receptor along with other efferent nerve signals mediated by acetylcholine increase salivary secretion and volume. Secretions can also be mediated through the beta-adrenergic receptors that induce a more proteinatious secretion [2].

Xerostomia (also known as dry mouth) is a condition associated with a change in saliva quality and quantity mostly affecting patients that have undergone head and neck proton radiation therapy. Acute xerostomia is attributed to off-target radiation-induced inflammation [3], and causes significant damage to the salivary glands, including loss of acinar cells and parasympathetic nerves [4] and damage to the ductal system [5]. On the other hand, late xerostomia is caused by gland fibrosis, is permanent and can occur up to one year after the radiation treatment [6,7]. Another cause of xerostomia is Sjögren Syndrome (SS), an autoimmune disease where the host immune cells target exocrine glands leading to hyposalivation [6,8,9]. Together, radiation therapy and SS cause loss in saliva quality and quantity that significantly decreases the quality of life of millions of patients. Treatments for hyposalivation are limited to medications (e.g., muscarinic receptor agonists that induce saliva secretion from residual acinar cells) and the introduction of artificial saliva, thus targeting relatively surface-level symptoms and providing only temporary relief. On the other hand, gene therapy has successfully increased salivary flow by enhancing expression of aquaporin-1 (AQP1, a water channel) in irradiated SG [10]; however, vital constituents of native saliva (e.g., mucins and amylase) are not replaced by this approach.

Salivary gland tissue engineering may provide a promising approach for regeneration and restoration of function for treatment of xerostomia. Previously, we have shown that both freshly isolated clusters of submandibular glands from mice [11] and rat parotid Par-C10 cells [12,13] were able to form spheroids characterized by the presence of tight junctions and open lumens, when grown on Growth Factor Reduced Matrigel® (GFR-MG). In this regard, Par-C10 cells were shown to be a good salivary gland model based on morphological (tight junction organization) and functional criteria (formation of transepithelial gradients and ion transport features that are required for saliva secretion) [13-15]. To avoid the tumorigenic properties of matrigel, we developed a matrigel-free, fibrin hydrogel (FH) that supports 3D gland microstructures containing polarized acinar cells [15]. This FH was prepared with fibrinogen that was conjugated with two peptides, RGD and YIGSR [16-18], derived from the laminin-111 chains, the major component of matrigel. When grown on laminin peptide conjugated fibrin hydroXgels (L1_p-FH), salivary gland cells formed spheroid clusters with open lumens and tight junctions [15,19]. In addition, conditioned medium from hair follicle derived mesenchymal stem cells (hHF-MSC) enhanced multi-lumen formation in mouse submandibular clusters grown on L1_p-FH [11]. These clusters demonstrated polarization with apical localization of ZO-1 and basolateral expression of E-cadherin [11], suggesting that diffusible factors secreted by the mesenchyme may promote salivary gland morphogenesis. Here we expand these studies to develop a system that promotes migration and branching of gland cells in 3D hydrogels by controlling the mode of presentation of two morphogenetic signals induced by Fibroblast Growth Factor-7 (FGF-7) and Fibroblast Growth Factor-10 (FGF-10) [18,20-28]. Particularly, these growth factors have shown to induce branching morphology in the distal bud tips of the lung and submandibular tissue explants [29]. Moreover, by controlling the mode of growth factor presentation and delivery, we can control the length and cellularity of branches as well as formation of new nodes/clusters within the hydrogel. This branching morphology is essential for salivary gland function as it allows the increase of saliva output without a major increase in glandular volume.

2. Materials and methods

2.1. Cell culture

The hHF-MSC were obtained and cultured under conditions previously described [30,31], passages 6–12 wild type and 9–11 FGF-7 knockout cells were used in the experiments. The Par-C10 cells [12] were cultured in DMEM/Ham’s F12 (1:1) (12,500,096, ThermoFisher Scientific, Grand Island, NY) containing 2.5% (v/v) fetal bovine serum (Atlanta Biologicals, Flowery Branch, GA) and the following supplements: 0.1 μM retinoic acid (R2625, Sigma-Aldrich, St. Louis, MO), 80 ng/mL epidermal growth factor (E4127, Sigma-Aldrich, St. Louis, MO), 2 nM triiodothyronine (T6397, Sigma-Aldrich, St. Louis, MO), 5 mM glutamine (included in the DMEM/F12 powder), 0.4 μg/mL hydrocortisone (H0888, Sigma-Aldrich, St. Louis, MO), 5 μg/mL insulin, 5 μg/mL transferrin, 5 ng/mL sodium selenite (ITS, I1884, Sigma-Aldrich, St. Louis, MO), and freshly added 100 μg/mL Gentamicin™ (15,750,060, ThermoFisher Scientific, Grand Island, NY), passages 40–60 were used in the experiments [13]. Cells were then filtered through a 0.4-μm nylon mesh, and approximately 2000 cells/well were plated on top of the indicated extracellular matrix.

2.2. Substrate preparation

Lab-Tek #1 Borosilicate Coverglass System Glass Chambers (ThermoFisher Scientific, Grand Island, NY) were pre-coated with 150 μL Growth Factor Reduced-Matrigel® (GFR-MG; Corning Life-sciences, Oneonta, NY): DMEM/F12 (2:1). L1_p-FH was prepared by mixing the L1_p-Fibrinogen [15] (120 μl, 2.5 mg/mL), Thrombin (30 μl, 2.5 mg/mL)) and εACA (1.5 μL, 2 mg/mL) in Tris buffered saline (TBS) with CaCl₂ (2.5 mM final concentration). The mixture was allowed to gel in a 37 °C incubator in the glass chambers for 1 h. Each layer was made out of 150 mL of gel. The chamber size is 1 cm² , yielding the initial gel layer a height of 1.5 mm.

2.3. Co-culture of the Par-C10 cells with mesenchymal stem cells

The experimental setup had hHF-MSC being seeded in GFR-MG at densities of 100,000, 50,000, 25,000, 10,000 and 0 cell/well (negative control). The cell laden mixture was allowed to gel in a 37 °C incubator for 1 h. Then Par-C10 cells were plated on top of the gel and cultured in Par-C10 medium [12,14] under 10% CO₂ . Sample morphology was evaluated by confocal microscopy after three days.

2.4. The effect of FGF-7 and FGF-10 in fibrin hydrogels on the Par-C10 cells

FGF-7 (K1757, Sigma-Aldrich, St. Louis, MO), FGF-10 (345-FG, R&D Systems, Minneapolis, MN) or pFGF-10 (fusion protein developed for this study) were added in the L1_p-conjugated FH (made similarly as above) at 50 ng/mL. FH were allowed to polymerize at 37 °C in an incubator for 1 h. After which the Par-C10 cells were plated on top of the FH and cultured in Par-C10 medium under 10% CO₂. Sample morphology was evaluated by confocal microscopy after three and six days.

2.5. Genome editing of mesenchymal stem cells using CRISPR sgRNA

Modified single guide RNA was purchased from Synthego (Redwood City, CA) to target the gene that encodes for FGF-7. The sgRNA was designed using the Synthego software to choose the guide with the lowest chance of off-target cuts. The sgRNA sequence that targets FGF-7 was 5′-UGCACAAAUGGAUACUGACA-3′. Cas9 enzyme (TrueCut™ Cas9 Protein v2, ThermoFisher Scientific, Grand Island, NY) and the sgRNA were allowed to complex at room temperature for 15–20 min in a 1:1 molar ratio to form the Ri-bonuleoprotein (RNP) complex. While the RNP complex was being formed, hHF-MSC (80,000-200,000 cells per reaction) were trypsinized and washed 4 times in PBS before the cells were re-suspended to a final concentration of 10,000-25,000 per μL in 8 μL buffer T provided in the Neon Transfection Kit from ThermoFisher Scientific, Grand Island, NY.

The electroporation unit (Neon Electroporation Unit, ThermoFisher Scientific, Grand Island, NY) was set up at 990 V voltage, 40 ms pulse duration and 1 pulse. The cells in buffer T (8 μl were then added to the RNP complex (2 μL) to bring the total reaction volume of 10 μL. After electroporation, cells were immediately transferred to a pre-warmed 6-well plate for further expansion. The knockout efficiency was determined by comparing the secreted protein levels of the knockout cells to that of control hHF-MSC by a sandwich ELISA (Human FGF-7 Quantikine ELISA, R&D Systems, Minneapolis, MN).

2.6. Cloning for Human FGF-10

The cDNA sequence containing vector was purchased from DNASU (pDONR201-FGF-10, Tempe, AZ) and the primers that included the thrombin cutting site and fibrin binding domain were ordered from ThermoFisher Scientific, Grand Island, NY. The nucleotide sequence for the primers are listed, forward: 5′-ATATGGATCCCTGGTGCCGCGCGGCAGCGGCGGCGGCAGCAACCAGGAACAGGTGAGCCCGGGCGGCGGCAGCTGGAAATGGATACTGACACATTGTGCC-3′, reverse: 5′-ATATCTCGAGGCCGCCCTATGAGTGTACCACCATTGGAAGAAA-3′.

The sequence was ligated into the pET28a expression vector (Addgene, Watertown, MA) downstream of the histidine tag, which is downstream of the T7 promoter and is regulatable through an Isopropyl β- D-1-thiogalactopyranoside (IPTG) inducible system. The vector also contains a kanamycin resistance gene for selection of clones.

The FGF-10 gene was PCR amplified in the presence of 1% DMSO (due to high GC content in gene sequence) using the touchdown PCR protocol with annealing temperature at 72 °C to 52 °C (each cycle differing by 2 °C). The amplified FGF-10 gene was isolated from a 1% agarose gel that was run for 25 min at 100 V using standard extraction protocol and ligated to the pET28a vector between the BamHI and XhoI restriction sites downstream of the LacO operon with T4 ligase (New England BioLabs, Ipswich, MA). The ligation products were then heat treated at 65 °C for 5 min to inactivate the DNA ligase and used to transform the BL21 (DE3) strain of E. coli by heat shock at 42 °C for 40–45 s, followed by immediate incubation in ice for 2 min. The bacteria were then allowed to recover in S.O.C medium for an hour in a 37 °C incubator shaker (250 rpm) before plating in kanamycin-containing agar plates for 16 h at 37 °C. The colonies formed on the plate were expanded in 3 mL cultures of LB Broth with 0.05 mg/mL kanamycin for 16 h, and the plasmid was isolated (NucleoSpin^® Plasmid, Machery-Nagel, Bethlehem, PA). Successful cloning was verified by digestion of the vector with BamH1 and Xho1. A band at 750 base pairs in the agarose gel indicated a positive identity of the clone. The resultant colonies were screened for the level of protein production. Fourteen colonies were picked and grown in LB with kanamycin and then induced with IPTG overnight at 22 °C. Then bacteria were collected by centrifugation, lysed and protein production level was assessed using SDS PAGE gel. The highest producing clone was then stored at −80 °C in 10% glycerol.

2.7. Production of the recombinant pFGF-10 fusion protein

The bacterial stock was expanded overnight in two 3 mL start cultures at 37 °C; transferred to a beveled flask with 1 L of 25 g/L LB, 10 mL kanamycin (5 mg/mL) and 20 mL of 50 × Component A (0.05% Glucose and 0.5% Glycerol) and allowed to grow at 37 °C. When O.D. reached 0.6 - 0.8, the culture was induced with 100 μL of 1 M IPTG and switched to a 22 °C shaker incubator overnight. The cells were centrifuged and lysed with a lysis buffer containing 50 mM Tris, 500 mM NaCl (pH = 7.4) and lysozyme from chicken egg (Sigma-Aldrich, St Louis, MO) in 1% Triton X-100 The samples were then sonicated to break up the DNA, hence completing the cell lysis. Sample was centrifuged at 50,000 g to remove cell debris. Supernatant was collected and filtered using a 0.45 μm filter before being passed through a Histrap column (GE Healthcare Life Sciences, Pittsburgh, PA). The purified pFGF-10 was eluted using 500 mM imidazol and further dialyzed to physiolog-ical pH and osmolarity. This also served to remove excess imidazole that was used for elution of pFGF-10 from the Histrap column. Purified pFGF-10 was sterilized by passing through a 0.22 μM filter and protein concentration was measured using a sandwich ELISA kit (Human FGF-10 ELISA Kit, LifeSpan BioSciences, Seattle, WA). Protein activity was tested via a proliferation assay of Par-C10 cells and the protein was stored with 0.1% BSA (carrier protein) at −80 °C until use.

2.8. Western Blot

Wild type FGF-10 along with pFGF-10 were run in separate lanes in a 20% w/v Tris-Glycine gel for 90 min. The contents of the gel were transferred to a Trans-Blot® Turbo™ Mini-Size Nitrocellulose Transfer Stack (Bio-Rad Laboratories, Hercules, CA) via the use of the Trans-Blot® Turbo™ Transfer System (Bio-Rad Laboratories, Hercules, CA). The nitrocellulose membrane was blocked for 1 h with 5% milk powder in TBST. The membrane was then incubated overnight with 1:1000 rabbit anti-FGF-10 polyclonal antibody (ABN44, EMD Millipore, Burlington, MA) in 5% milk powder in TBST at 4 °C on a rocker. The membrane was washed 3 times with TBST and then incubated with 1:2000 anti-rabbit IgG, HRP-linked antibody (7074S, Cell Signaling Technology, Danvers, MA) for 2 h at room temperature on a rocker. The membrane was washed 3 times with TBST and then incubated with a 1:1 mixture of SuperSignal™ West Pico PLUS Luminol/Enhancer Solution (1,863,098, ThermoFisher Scientific, Grand Island, NY) and SuperSignal™ West Pico PLUS Stable Peroxide Solution (1,863,099, ThermoFisher Scientific, Grand Island, NY) for 5 min. After which the membrane was imaged on auto exposure time using the ChemiDoc™ Imaging System from Bio-Rad Laboratories, Hercules, CA.

2.9. Growth factor release from the fibrin hydrogel

The growth factors (FGF-7/FGF-10) were embedded in the L1_p-FH at 50 ng/mL. The pFGF-10 (NQEQVSP-FGF-10, 50 ng/mL) was conjugated into L1p-FH in the presence of Factor XIII (4PEU/mL). Each gel (100 μL per well) was allowed to polymerize at 37 °C for 1 h before 200μl of DMEM/F12 containing 1% BSA was added to the gel. At different times (30 min, 1, 2, 3, 4, 5, 6, 9, 12 and 24 h) all 200 μL of medium was collected and replenished in full. The concentration of each growth factor was measured by sandwich ELISAs using the same kits and the cumulative release was plotted over time.

2.10. Immunocytochemistry and confocal microscopy

The cells were fixed with 4% paraformaldehyde in Phosphate Buffered Saline (PBS) for 20 min at room temperature after which they were permeabilized using 0.1% Triton X-100 in PBS for 10 min and blocked with 5% goat serum in PBS for 2 h at room temperature and incubated at 4 °C on a rocker overnight with primary antibody diluted in blocking buffer (1:200 dilution, rabbit anti ZO-1 (61–7300, ThermoFisher Scientific, Grand Island, NY), 1:400 mouse anti TMEM16A (ab 190,721, Abcam Cambridge, MA), or 1:200 rabbit anti Ki-67 (MA5-14,520, Ther-moFisher Scientific, Grand Island, NY). Cells were then washed 5 times with PBS and secondary antibody diluted in blocking buffer (1:200 Alexa Fluor™ 488/568/647 goat anti rabbit/rabbit/mouse (A32731/A11011/A32728, ThermoFisher Scientific, Grand Island, NY respectively) was added and incubated for 2 h at room temperature on a rocker. Alexa Fluor™ Phalloidin 488/594 (A12379 /A12381 ThermoFisher Scientific, Grand Island, NY) was diluted 1:200 in PBS containing 1% BSA and incubated for 2 h at room temperature on a rocker. Cells were counterstained with Hoechst-33,342 dye (1:2000 in PBS, H1399, ThermoFisher Scientific, Grand Island, NY) or TO-PRO™-3 iodide (1:1000 in PBS, T3605, ThermoFisher Scientific, Grand Island, NY) for 5 or 30 min, respectively. Cells were visualized and images were acquired using the Zeiss LSM 510 point scanning confocal microscope (Zeiss, Thornwood, NY). Images in the z-stack were first inverted and then analyzed by using the LSM extension package of ImageJ (NIH, Bethesda, MA). The size of the nodes were quantified as the average diameter of five diameters taken at the focal plane of the largest diameter and 2 focal planes (1 μm/focal plane) above and below it, for the Day 3 aggregate and similarly for the Day 6 aggregate that had at least one branch connected to it. The length of the branching networks was quantified as the distance between its origin and end-point. The thickness of each branch was calculated at five different equidistant points along the branch from the top to bottom and their average was taken as the overall thickness of the branch. The depth of invagination was calculated as the maximum distance from the top of the aggregate to the lowest point of the aggregate on the Z-axis.

2.11. Statistical analysis

Data are means ± SEM of results from three or more experiments. Statistical significance (p<0.05) was calculated from one-way ANOVA comparing three or more groups, with a post-hoc t-Test along with the Bonferroni correction done to test for significance between the groups. To calculate significant differences between two groups only, Student’s t-Test was used.

3. Results

3.1. Human mesenchymal stem cells induce branching networks Par-C10 cells on GFR-MG

In our previous work [11], we showed that soluble signals from the human hair follicle derived mesenchymal stem cells (hHF-MSC) enhanced multi-lumen formation in submandibular gland (SMG) cell clusters. These results prompted us to hypothesize that co-culture of salivary gland cells with hHF-MSC may promote gland morphogenesis in vitro. To generate a gradient of diffusible factors secreted by hHF-MSC, we embedded hHF-MSC in Growth Factor Reduced Matrigel® (GFR-MG) at the indicated cell densities and Par-C10 cells were plated on top of GFR-MG at 2000 cells per cm² [22] (Fig. 1). After three days, in the absence of hHF-MSC, the Par-C10 cells formed spheroid-like clusters (Fig. 1A), consistent with our previous work. Interestingly, the presence of hHF-MSC led to migration and proliferation of Par-C10 cells into the hydrogel forming a branching network, which showed increased branch network length with increasing hHF-MSC cell density, as seen in the phase (Fig. 1B-E and 1G) and confocal images (Fig. S1).

Fig. 1.

hHF-MSC increase branching morphogenesis of Par-C10 cells. (A-E) Co-culture of Par-C10 cells (2000 cell/cm²) seeded on GFR-MG with hHF-MSC at the indicated cell number per well. (F) Co-culture of Par-C10 cells (2000 cell/cm²) with FGF-7 KO hHF-MSC (10⁵ cell/well). Scale Bar = 1000 μm. (G) Quantification of the total length of the branching networks per well. Statistical analysis was performed by employing one-way Anova followed by a Tukey post-hoc test. Significance between groups were calculated using t-Test: Two-Sample Assuming Unequal Variances between the groups with n = 3.

3.2. FGF-7 secreted by hHF-MSC is necessary for formation of Par-C10 branching networks

Interestingly, the ELISA results showed that the Fibroblast Growth Factor 7 (FGF-7), a known gland morphogen [21,22,29], was secreted at high levels by hHF-MSC (Fig. S2C), suggesting that it may be at least one of the morphogens that lead to branching of gland cells on matrigel. To evaluate this hypothesis, we knocked-out (KO) the FGF-7 gene in hHF-MSC using the CRISPR-Cas9 sgRNA and verification of the successful knock-out was done via ELISA using cell supernatant (Fig. S2C). Indeed, FGF-7 KO hHF-MSC failed to support branching network formation of the Par-C10 cells on GFR-MG, even at the highest cell density (Figs. 1F and S2), indicating that FGF-7 secreted by hHF-MSC was necessary for Par-C10 migration and branching.

3.3. Differential effects of FGF-7 vs. FGF-10 on salivary gland morphogenesis

In our previous work, we showed that FH alone failed to induce acinar structures of Par-C10 cells, except when decorated with two laminin-111 peptides (L1_p), namely RGD and YIGSR (L1_p-FH). In the presence of these two peptides, the cells formed organized cell clusters with small lumens, similar to GFR-MG. Here, we employed this system to evaluate the effect of FGF-7 in gland morphogenesis. Indeed, after three days in culture, the Par-C10 cells displayed spheroid-like structures on L1_p-FH containing FGF-7 (Fig. 2, A-D). We also tested FGF-10 (KGF-2) (Fig. 2, E-H), a closely related family member of FGF-7, which binds to FGFR2-IIIb. We found that FGF-10 also increased spheroid formation and that the size of the spheroids was larger than those treated with FGF-7 (Fig. 2I). However, no branching was observed at this time, even in the presence of either of these growth factors.

Fig. 2.

FGF-10 promotes larger Par-C10 spheroids than FGF-7. Par-C10 cells were cultured for three days on L_1p-FH in the presence of (A-D) FGF-7 (50ng/ml) or (E-H) FGF-10 (50 ng/mL). Samples were stained for ZO-1 (green) and nuclei were counterstained with To-Pro-III iodide (blue). (A, E) The top view of the self-assembled Par-C10 spheroid. (B, F) The 3D rendering of the Par-C10 spheroid. (C, G) The middle section of the confocal image of the Par-C10 spheroid. Scale bar = 50 μm for FGF-7 and bar = 100 μm for FGF-10 treatment. (D, H) A 3D rendering showing the depth map of the Par-C10 spheroid in the L_1p-FH. Red indicates the top of the spheroid and blue indicates the bottom. Scale bar = 80 μm for FGF-7 and 70 μm for FGF-10 treatments. Panel images were all rendered from confocal images. (I) Quantification of the average diameter of the aggregates. * denotes p<0.01. Statistical analysis was done using t-Test: Two-Sample Assuming Unequal Variances (n = 3). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Interestingly, a longer exposure time (6 days) to FGF-7 or FGF-10 induced significant morphogenetic changes (Fig. 3A-L). Specifically, in the presence of FGF-7 (Fig. 3A-D), the cells formed a highly branched interconnected network that had multiple branches and nodal points that showed considerable penetration into the gel. In contrast, FGF-10 (Fig. 3E-H) induced a network with fewer but thicker branches, which contained more cells, indicating higher cell proliferation (Fig. 3M and 3N). However, the branches did not penetrate to the same extent as FGF-7-treated cells, indicating that FGF-10 did not support migration to the same extent as FGF-7 (Fig. 3O).

Fig. 3.

Morphogen delivery through the hydrogel promotes Par-C10 cell migration and branching. Par-C10 cells were cultured on L_1p-FH containing (A-D) FGF-7; (E-H) FGF-10; or (I-L) both (50 ng/mL each) for six days. Samples were stained for ZO-1 (green) and F-actin (red); nuclei were counterstained with To-Pro-III iodide (blue). (A, E, I) The top view of the self-assembled Par-C10 spheroid. (B, F, J) The 3D rendering of the Par-C10 spheroid. (C, G, K) The middle section of the confocal image of the Par-C10 spheroid. Scale bar = 200 μm. (D, H, L) 3D rendering showing the depth map of the Par-C10 spheroid grown for six days with red indicating the top of the spheroid and blue indicating the bottom. Scale bar = 140 μm for FGF-7; 60 μm for FGF-10; and 70 μm for FGF-7 + FGF-10 treatments. Panel images were all rendered from confocal images. (M) Quantification of the percentage Ki-67 + nuclei per aggregate (* denotes p<4.6 × 10⁻⁴). (N) Quantification of the total number of Ki-67 + nuclei per aggregate (* denotes p<5 × 10⁻⁵). (M, N) Statistics done using t-Test: Two-Sample Assuming Unequal Variances (n = 3). (O) Quantification of the average penetration depth (* denotes p<2.9 × 10⁻³). Statistical analysis was performed by employing one-way Anova followed by a Bonferroni post-hoc test. Significance was calculated using t-Test: Two-Sample Assuming Unequal Variances between the groups with n = 3. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Measurements of node diameters showed that node size increased by approximately 2-fold in the presence of FGF-10 as compared to FGF-7, suggesting increased proliferation of cells in each node (Fig. 4A). Further addition of FGF-7 did not affect the node size, indicating no further effect on cell proliferation (Fig. 4A). The average length of branches was similar across all three conditions (Fig. 4B), but the average length of each branch per node was 2-fold higher in the presence of FGF-7, with no further change by the addition of FGF-10 (Fig. 4C), suggesting that FGF-7 might promote faster migration. Likewise, the total number of nodes per network was higher in the presence of FGF-7 as compared to FGF-10 (Fig. 4D). These results suggest that FGF-10 promotes proliferation, while FGF-7 supports migration and penetration of Par-C10 cells into the 3D hydrogels.

Fig. 4.

Quantification of branched networks. Par-C10 cells were cultured on L_1p-FH containing FGF-7, FGF-10 or both (50 ng/mL each) for six days. (A) The average diameter of the nodes at the indicated treatments (* denotes P<2 × 10⁻⁴). (B) Average length of branches (no statistical significance between groups). (C) Average length of the branches per node (* denotes p<3.9 × 10⁻⁴). (D) Average number of branches under each condition (* denotes p<4.4 × 10⁻³). (A-D) Statistic analysis was performed by employing one-way Anova followed by a Bonferroni post-hoc test. Significance was calculated using t-Test: Two-Sample Assuming Unequal Variances between the groups (n = 3).

3.4. Engineering FH for differential mode of FGF-7 and FGF-10 stimulation

Diffusion of FGF-7 from within the gel to the cells on the top surface generates a concentration gradient that promotes migration into the gel. However, sustained 3D cell migration and branching may require proliferation of the migrating cells, which could be achieved by a second morphogen such as FGF-10, when immobilized into the hydrogel. Therefore, we hypothesized that diffusible FGF-7 may promote directional migration, while FGF-10 conjugated into L1_p-FH may promote proliferation of the migrating cells, thereby sustaining migration and branching.

To this end, we engineered a fusion protein of FGF-10 with the Factor XIII recognition sequence, NQEQVSP (denoted as pFGF-10) which results in a higher molecular weight protein (wild type: 19.5 KDa v/s recombinant: 24.5 KDa) (Fig. 5A). In the presence of active FXIII, the fusion protein can be enzymatically conjugated into the FH during polymerization, as we demonstrated previously [32-34]. On the other hand, FGF-7 lacking the peptide sequence is free to diffuse in the hydrogel (Fig. 5B).

Fig. 5.

Production of the fusion protein pFGF-10 (peptide-FGF-10) containing the fibrin binding domain. (A) Western blot showing the difference in molecular weight of the wild type FGF-10 versus pFGF-10 before the thrombin cleavage of the 6x His-tag. (B) Release of conjugated pFGF-10 and unconjugated FGF-7, FGF-10 over time from fibrin hydrogels. Statistical analysis was performed by employing one-way Anova followed by a Bonferroni post-hoc test. Significance was calculated using t-Test: Two-Sample Assuming Unequal Variances between the groups; * denotes p<1.92 × 10⁻¹⁰ , n = 3.

Indeed, pFGF-10 alone increased proliferation similar to FGF-10 (Fig. S3) but did not induce much branching and migration into the L1_p-FH (Fig. 6A-E). Then we set up a two-layer L1_p-FH gel, where pFGF-10 was conjugated in the top layer, while FGF-7 was embedded (no conjugation) in the lower layer (Fig. 6K-O). This was done to avoid the quick diffusion of the FGF-7 to the media and maintain the concentration gradient and therefore, migration into the gel. The addition of FGF-7 into the lower L1_p-FH layer induced migration and branching deep into the hydrogel, with migration paths as long as 140 μm from the surface. Interestingly, the branches of the network were thicker (Fig. 7D) as compared to FGF-7 alone, mostly due to higher cell proliferation induced by the conjugated pFGF-10. Interestingly, proliferation was not restricted to the migrating tip of the branch or the nodal head, but was observed throughout the branch, as evidenced by Ki67 positivity (Fig. 7A-C). In contrast, when FGF-7 was added in the media (no spatial gradient), branching was restricted to the surface of the L1_p-FH, with little penetration into the hydrogel (Fig. 6F-J). These results show that the direction and branching of gland cells into the hydrogel can be controlled by a combination of two morphogens; a diffusible one controlling migration and a spatially immobilized one controlling proliferation of the migrating cells.

Fig. 6.

Spatially controlled delivery of morphogens affects branching and node formation. (A-E) pFGF-10 at 50 ng/mL was conjugated to L1_p-FH. (F-J) FGF-7 was delivered at 50 ng/mL through the medium. (K-O) pFGF-10 was conjugated in the upper L1_p-FH layer and FGF-7 was mixed in the lower L1_p-FH layer. After six days, samples were stained for ZO-1 (green), F-actin (red), TMEM16A (pink) and nuclei were counterstained with Hoechst (blue). (B, G, L) The top view of the self-assembled Par-C10 spheroid. (C, H, M) The 3D rendering of the Par-C10 spheroid. (D, I, N) The middle section of the confocal image of the Par-C10 spheroid. Scale bar = 200 μm. (E, J, O) 3D rendering showing the depth map of the Par-C10 spheroid grown for 6 days with red indicating the top of the spheroid and blue indicating the bottom. Panel images were all rendered from confocal images. Scale bar = 60 μm for conjugated FGF-10; 80 μm for conjugated FGF-10 + FGF-7 in the medium above; and 140 μm for conjugated FGF-10 + diffusible FGF-7 from lower gel. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7.

Proliferation is sustained throughout the migrating branches. FGF-10 at 50 ng/mL was conjugated to the L1_p-FH and FGF-7 was delivered at 50 ng/mL through a lower L1_p-FH. After 6 days, samples were stained for F-actin (green), Ki67 (red), and nuclei were counterstained with To-Pro-III iodide (blue). (A) The top view of the branch of the self-assembled Par-C10 spheroid. (B) The 3D rendering of the Par-C10 spheroid. (C) A 3D rendering showing the depth map of the Par-C10 spheroid with red indicating the top of the spheroid and blue indicating the bottom. Scale bar = 200 μm. Panel images are all rendered from confocal images. (D) Average branch thickness. Statistical analysis was done using t-Test: Two-Sample Assuming Unequal Variances; * denotes p<3.9 × 10⁻¹³ , n = 3.

4. Discussion

In our previous work we showed that L_1p-FH is suitable for in vivo applications to promote salivary tissue regeneration as evidenced by formation of acinar gland organoids that were both structurally and functionally akin to native salivary gland [19]. These results were promising as the L1_p-FH scaffold promoted gland cell aggregation as well as lumen formation with expression of salivary gland differentiation markers, including amylase [14,15]. However, these organoids lacked the structural organization and branched morphology of salivary gland tissue. This phe-notype is seen in many organs of epithelial origin and plays a crucial role in enhancing tissue function by enhancing the tissue surface area [21,22,26,27]. The branching structure also leads to the development of specialized cellular compartments, each with its own unique function. The acinar cells that produce saliva, which is secreted with the aid of contractile myoepithelial cells; and the striated ducts that turn the saliva hypotonic while transporting it into the oral cavity [35].

In this study we showed that a co-culture system of the Par-C10 cells with the hHF-MSC resulted in a highly branched 3D network of cells that increased in complexity with increased mesenchymal cell number, suggesting that soluble signals in the form of growth factors within the hHF-MSC might elicit the formation of a branched network. It is widely accepted that during embryogenesis there is dynamic signaling interplay between the mesenchyme and epithelium, which is mediated by soluble and ECM-immobilized factors. These morphogenetic factors including FGF-7 and FGF-10, are secreted by the mesenchyme and act through FGFR2-IIIb on the epithelium, resulting in the highly branched architecture of the gland [18,26,36]. Salivary gland branching morphogenesis in vivo is first seen in the developing mouse embryo at E13.5–14.5 at the pseudo-glandular stage [37] and is attributed to the Sox9+distal progenitors that proliferate and form branches under the influence of FGF-10 [24,27]. Interestingly, Sox9 expression in the submandibular gland epithelium appears to be regulated by FGF-7 [38], prompting us to examine the influence of both FGFR2-IIIb ligands in the formation of branched gland cell networks in vitro [26,36].

We showed that hHF-MSC express high levels of FGF-7 and FGF-7 gene knock out using sgRNA CRISPR/Cas9 [39,40] led to dramatic loss of hHF-MSC-mediated Par-C10 branching networks. This result clearly demonstrated that FGFR2-IIIb signaling is essential for gland cell migration and development of branching networks in a 3D environment. The small number of branches seen with FGF-7 KO hHF-MSC may be due to secretion of other soluble factors e.g. FGF-2 that can bind to FGFR2-IIIb and cause branching. However, FGF-7 and FGF-10 are clearly the two most potent activators of this signaling pathway, prompting us to test the effect of FGF-7 and FGF-10 on Par-C10 migration and formation of branching structures, when provided through the 3D hydrogel. Exposure to FGF-7 induced formation of branches that invaginated into the gel to a higher extent (i.e. longer branches, as compared to exposure to FGF-10). In contrast, incorporation of FGF-10 in L1_p-FH increased the node size and the number of branches initiating from them, albeit the length of these branches (or the penetration distance) was shorter. This is an intriguing result as both growth factors bind to the same receptor, yet they produce different cellular responses, both of which are essential during embryogenesis as they deter-mine the size and pattern of different organs [21,23,36]. The differential effects may be due to variations in the binding site or the binding kinetics of each growth factor to a cofactor (e.g. heparin) or receptor (FGFR2-IIIb) [22]. Indeed, mutations to the heparin binding domain of FGF-10 led to similar lacrimal gland branching as FGF-7 [22], suggesting that the strength of binding to heparin may affect receptor binding and subsequent pathway activation. Interestingly, when these growth factors were delivered together, both the length and size of branches increased, suggesting that downstream signaling pathways may be acting synergistically.

The differential affinity of FGF-7 vs. FGF-10 for heparin was shown to produce differential gradients that determine where branching or elongation occurs, thereby affecting gland branching morphogenesis [22]. To mimic such developmental gradients, we provided FGF-7 as a diffusible factor from a lower layer of FH and conjugated pFGF10 into the FH - via the factor XIII recognition domain that was fused to the FGF-10 sequence – to mimic its higher affinity for ECM. Diffusion of FGF-7 through the hydrogel to the surface where it bound to the cells generated a gradient that caused migration of multicellular branches. As cells moved into the gel, they came in contact with immobilized pFGF-10, which increased proliferation, generating new nodes reminiscent of acinar structures. As cells continued to migrate, new branches emerged from the nodes extending the network. Interestingly, Par-C10 cells did not move as individual cells but as a collective migrating group while maintaining their ZO-1 junctions, suggesting that the com-bination of FGF-7/pFGF-10 did not disrupt cell-cell interactions. Immobilizing FGF-10 generated thicker branches containing proliferating cells, possibly as a result of sustained signaling through FGFR2-IIIb. However, these structures still lacked lumen forming capability. Formation of such hollow structures may require conditions that activate programmed cell death possibly following the attenuation of the proliferation/branching signals [41,42]. Alternatively, formation of hollow structures may require the presence of vasoactive intestinal peptides such as those secreted by the innervating peripheral neurons and promote ductal growth while also leading to the formation of a contiguous lumen via a cyclic AMP/protein kinase A (cAMP/PKA)-dependent pathway [43].

5. Conclusion

In summary, we demonstrated that by controlling the gradients of two morphogens, FGF-7 and FGF-10, it is possible to influence gland cell migration and organization into branched networks of varying branch length and thickness. The diffusible FGF-7 may provide quick cell activation and initiation of migration into the hy-drogel; while the immobilized FGF-10 is accessed only when cells migrate into the gel and may help to sustain this migration by promoting proliferation. In places, the proliferation signals dominated shifting the balance toward formation of cell clusters or nodes, from which more cells branched out in the direction of increased morphogen concentration (i.e. FGF-7). The spatial delivery mecha-nism of two (or more) morphogens may provide insight into engineering geometrically complex mini glandular tissues to study development and function as well as in physiologically relevant platforms for drug testing and discovery.