Dual Peptide-Functionalized Hydrogels Differentially Control Periodontal Cell Function and Promote Tissue Regeneration

Abstract

Restoring the tooth-supporting tissues lost during periodontitis is a significant clinical challenge, despite advances in both biomaterial and cell-based approaches. This study investigated poly(ethylene glycol) (PEG) hydrogels functionalized with integrin-binding peptides RGD and GFOGER for controlling periodontal ligament cell (PDLC) activity and promoting periodontal tissue regeneration. Dual presentation of RGD and GFOGER within PEG hydrogels potentiated two key PDLC functions, alkaline phosphatase (ALP) activity and matrix mineralization, over either peptide alone and could be tuned to differentially promote each function. Hydrogel matrix mineralization, fostered by high concentrations of GFOGER together with RGD, identified a PDLC phenotype with accelerated matrix adhesion formation and expression of cementoblast and osteoblast genes. In contrast, maximizing ALP activity through high RGD and low GFOGER levels resulted in minimal hydrogel mineralization, in part, through altered PDLC pyrophosphate regulation. Transplantation of PDLCs in hydrogels optimized for either outcome promoted cementum formation in rat periodontal defects; however, only hydrogels optimized for in vitro mineralization improved new bone formation. Overall, these results highlight the utility of engineered hydrogel systems for controlling PDLC functions and their promise for promoting periodontal tissue regeneration.

1. Introduction

The periodontal tissues, including the fibrous periodontal ligament (PDL) and mineralized cementum, mediate the attachment of teeth to surrounding alveolar bone[1]. These tissues are destroyed by periodontitis, a widespread inflammatory disease occurring in response to bacterial biofilms. While some reconstructive therapies show promise for regaining optimal tooth support and function[2], current clinical treatments have significant limitations[3], thereby motivating the design of tissue engineering approaches.

The PDL and its resident cells, here termed PDL cells (PDLCs), are critical for maintaining the fibrous-mineralized periodontal interfaces and repairing the periodontal tissues following injury. PDLCs are a heterogeneous population that include multipotent cells uniquely capable of differentiating and forming new cementum, bone, and PDL[4, 5]. PDLCs are also characterized by expression of tissue non-specific alkaline phosphatase (TNAP or ALP), an enzyme typically associated with mineralizing cells and tissues[6, 7]. Isolated and cultured PDLCs have been shown to possess stem cell properties and PDLC transplantation is a promising approach for periodontal tissue regeneration[8, 9]. Despite this promise, the scaffold properties required to maximize the regenerative potential of transplanted PDLCs are not fully understood. Furthermore, the factors that regulate in vivo PDLC activity, including maintaining the fibrous but ALP-rich PDL while supplying a pool of differentiated osteoblasts and cementoblasts, remain unclear.

Guiding cell-material interactions is a foundational principle underlying tissue engineered scaffold development. Cells bind to their extracellular matrix (ECM) via integrins, paired transmembrane proteins that translate physical and biochemical signals to regulate cell activity[10, 11]. Multiple integrins exist and bind various peptide motifs within ECM proteins, lending both specificity and diversity to cellular interactions. Specific ECM protein compositions define the periodontal tissues and are spatially and temporally regulated during periodontal wound healing[12], suggesting that identification and replication of key ECM binding cues can be leveraged to guide periodontal cell fate and tissue repair.

Collagen type I is the primary organic component of cementum and alveolar bone and comprises the majority of fibrillar collagen in the PDL[13]. The PDL is also rich in fibronectin and vitronectin[14] which bind collagen and play important roles in collagen fibril formation and cell migration[15]. The RGD peptide motif is a ubiquitous integrin-binding peptide present within numerous ECM proteins, including fibronectin and vitronectin[16]. Multiple integrins bind RGD, with integrins $\text{αvβ3}$ and $\text{α5β1}$ best characterized for stromal cells such as PDLCs[17]. Integrin interactions with collagen type I are mediated by integrin $\text{α2β1}$ binding the GFOGER peptide sequence[18]. Peptide ligands for integrin $\text{α2β1}$-specific binding contain GFOGER flanked by (Gly-X-Y)_n repeats, assemble into a triple helix in solution, and can be incorporated on surfaces or within hydrogels by inclusion of reactive groups or peptides near the N- or C-terminus[19]. GFOGER peptide presentation on or within diverse biomaterials has shown specific promise for promoting osteogenic cell differentiation and bone formation[20, 21].

Collagen is closely paired with fibronectin, vitronectin, and other non-collagenous RGD-containing proteins in the ECM, suggesting that cells can simultaneously bind both ligands on different proteins. Studies examining the effects of dual collagen and fibronectin/vitronectin protein-based RGD ligand presentation have shown enhanced cell binding and adhesion formation when both ligands were present[22, 23]. Biomaterial functionalization with multiple cell-binding peptides is an emerging approach[24], but few studies have investigated combined presentation of collagen-derived peptides and RGD to promote cell differentiation and mineralized tissue formation.

Poly(ethylene glycol) (PEG) hydrogels are widely used as ECM mimetics and cell transplantation scaffolds, harnessing well-defined properties that allow for precise control over hydrogel mechanical properties, degradation, and cell-hydrogel binding[25, 26]. We have recently investigated the ability of PEG hydrogels to modulate PDLC behavior through controlled introduction of three-dimensional ECM cues, showing a significant upregulation in ALP activity when PDLCs were cultured within matrix metalloproteinase (MMP)-degradable hydrogels[27]. Surprisingly, ALP activity was decoupled from mineralization of the hydrogel matrix, mimicking a key feature of PDLCs within the PDL, and suggesting that these hydrogels could be used to further define the ECM cues that regulate PDLC activity and promote their regenerative potential. Thus, the goal of the current study is to utilize controlled presentation of RGD and GFOGER peptide ligands within PEG hydrogels to interrogate ALP activity and matrix mineralization and identify promising hydrogel scaffold designs for periodontal tissue regeneration.

2. Materials and Methods

2.1. Synthesis of PEG macromers and Peptides

PEG-norbornene was synthesized from 8-arm PEG hydroxyl (JenKem Technology, TX, USA) and 5-norbornene-2-carboxylic acid (Alfa Aesar, MO, USA) using N,N′-dicyclohexylcarbodiimide (Alfa Aesar) as the coupling agent[26]. Norbornene functionalization of PEG arms was determined with ¹H-NMR [CDCl3]: $\text{δ}$ = ~3.6 for PEG ether protons, $\text{δ}$ = 5.9–6.3 for norbornene vinyl protons, with PEG macromers having ≥90% functionality used for all experiments. The MMP-degradable peptide crosslinker GKKCGPQGIWGQCKKG, integrin-binding peptide peptides RGD (GCGRGDSG) and GFOGER (GCG(GPP)₅GFOGER(GPP)₅GCG), and scrambled control peptide (Ctrl, CGRDGSG) were synthesized using a Liberty 1 Microwave Assisted Peptide Synthesizer (CEM, NC, USA) or purchased from Genscript (NJ, USA) and characterized prior to use as previously described[28]. GFOGER triple helix formation in solution and stability at 37 °C was confirmed using a circular dichroism spectrophotometer (J-1000, Jasco, MD, USA) (Supplemental Figure 1).

2.2. Hydrogel Formation and Characterization

Hydrogels were formed by dissolving PEG macromers and peptides in PBS with 0.05% photoinitiator lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) then transferring 30 μL of the hydrogel solution to cylindrical molds (diameter 4.5 mm) and placing under UV light (365 nm, ~5 mW/cm²) for 3 minutes. Sub-stoichiometric ratios of crosslinker arms to PEG arms were employed to allow tethering of peptide ligands (0.6:1 thiol:ene ratio). Hydrogels were removed from molds and placed in Dulbecco’s phosphate buffered saline (PBS, Gibco, MA, USA) to swell overnight before characterization. Hydrogel stiffness was determined via unconfined compression with a MTS test frame (MN, USA) equipped with a 5 N load cell, and mesh size was estimated from the mass of swollen and lyophilized hydrogels as described previously[29].

Hydrogels were formulated to maintain a similar stiffness (~9 kPa) and mesh size (~19.5 nm) across all peptide conditions (Supplemental Figure 1). For hydrogels with 1.05 or 2 mM GFOGER, this necessitated a reduction in PEG content from 10 wt% to 9.5 and 9 wt%, respectively (Supplemental Figure 1).

2.3. PDLC Isolation and Characterization

Human PDLCs were obtained from the 3^rd molars of 6 donors (Supplemental Table 1) following informed consent (University of Rochester approved protocol RSRB00072932). After extraction, molars were immediately placed in Hank’s balanced salt solution (HBSS, Gibco) supplemented with 5% fetal bovine serum (FBS, R&D Systems, MN, USA) and 100 U/mL penicillin, 100 mg/mL streptomycin, and 0.25 μg/mL amphotericin B (1x Antibiotic-Antimycotic, Gibco). PDL tissues were removed from the middle third of roots with a scalpel, minced into 0.5 mm diameter pieces, rinsed twice in PBS, then digested in a solution of collagenase type 1 (900 U/mL, Gibco) and dispase type II (2.3 U/mL, Sigma) in PBS at 37 °C for 1 hour and passed through a 70 μm cell strainer to obtain single PDLC suspensions. PDLCs were plated in growth medium ($\text{αMEM}$ with nucleosides and Glutamax (Gibco), 20% FBS, and 1x Antibiotic-Antimycotic (Gibco)) on 6 well plates at CFU-F density (1000 nucleated cells/cm²). After 7–10 days, single cell-derived colonies were detached with 0.05% trypsin/EDTA (Gibco) and plated in 75 cm² flasks for expansion. PDLCs at passage 3 were characterized for cell surface marker expression and differentiation capacity, and PDLCs at passage 3–6 were used for all experiments. PDLCs from donors 1–3 were isolated and expanded separately for screening experiments while PDLCs from donors 4–6 were isolated together on the same day and pooled to provide a sufficient number of cells for downstream analyses.

2.4. PDLC Encapsulation and Culture

Hydrogel solutions were prepared in PBS with PDLCs suspended at a final concentration of 2×10⁶ cells/mL for screening experiments (donors 1–3) or 10×10⁶ cells mL for remaining experiments (donors 4–6) and exposed to UV light for 3 minutes. Hydrogels were transferred to basal medium ($\text{αMEM}$, 10% FBS) and allowed to equilibrate for 24 hours, after which medium was changed to osteogenic/cementogenic (inductive) medium (basal medium supplemented with 100 μM L-ascorbate 2-phosphate (Sigma), 10 mM $\text{β}$-glycerophosphate (Sigma), and 100 nM dexamethasone (Sigma)) which was then changed 2–3 times per week.

2.5. Hydrogel Peptide Screening and Analysis

PDLCs were cultured within hydrogels containing either RGD or GFOGER (single peptide) or RGD and GFOGER peptides together (dual peptide hydrogels) at a range of concentrations. The scrambled peptide was used alone as a control and for maintaining 4 mM total peptide concentration for all hydrogels. Alkaline phosphatase (ALP) activity was measured at 1 week and matrix mineralization was measured at 3 weeks using established methods[27]. Briefly, for ALP activity, hydrogels were removed from media, rinsed twice in PBS, homogenized in RIPA buffer with polypropylene pellet pestles (Sigma), and subjected to centrifugation at 14,000 g for 5 minutes. A portion of the supernatant was removed and incubated with p-nitrophenyl phosphate (PNPP substrate tablets, Fisher Scientific, MA, USA) dissolved in a diethanolamine buffer (1.0 M diethanolamine in ddH₂0 with 0.5 mM MgCl₂, pH 9.8) at a concentration of 1 mg/mL. Absorbance (405 nm) was read every minute for 10 minutes and the resulting slopes of absorbance versus time for samples or standards (calf intestinal alkaline phosphatase (Promega, WI, USA)) were used to determine ALP concentration. The remaining lysate was used to determine hydrogel DNA content with the PicoGreen^™ assay (Invitrogen, MA, USA) for normalization of ALP concentration to PDLC number. Matrix mineralization was determined by incubating hydrogels with 2 mM xylenol orange (Sigma) overnight at 37 °C followed by staining of PDLC nuclei with Hoechest 33342 (Invitrogen). Hydrogels were rinsed twice in PBS then imaged using a confocal microscope (Olympus FV1000) at 10x magnification. Single z-stack images were acquired at 100 μm intervals spanning the whole hydrogel height and analyzed using ImageJ and pooled to give a mineralized area (positive xylenol orange stain) normalized to PDLC number (nuclei). Hydrogels without PDLCs were also incubated in inductive medium for 3 weeks and stained with xylenol orange to confirm the absence of non-cell mediated hydrogel mineralization.

Design-Expert Software (Stat-Ease, MN, USA) was used to analyze dual peptide hydrogel responses ALP activity and matrix mineralization. Peptide concentrations (0.1, 1.05 and 2 mM) were coded (–1, 0, +1) and outcomes analyzed as a face-centered composite using a response surface model (Equation 1).

$$y={\beta }_0+{\beta }_1{x}_1+{\beta }_2{x}_2+{\beta }_{12}{x}_1{x}_2+{\beta }_{11}{x}_1^2+{\beta }_{22}{x}_2^2+\in$$ (1)

For the response, $y,{\beta }_1$ and ${\beta }_2$ were the coefficients for peptide main effects, RGD and GFOGER, respectively. ${\beta }_{12}$ was the coefficient for the interaction effect RGD*GFOGER and ${\beta }_{11}$ and ${\beta }_{22}$ were coefficients for quadratic effects RGD² and GFOGER². ANOVA was used to identify statistically significant coefficient estimates. Outcomes for donors 1–3 were also pooled and the resulting response surfaces used to identify dual peptide compositions estimated to optimize ALP activity (ALP optimized – AO) or matrix mineralization (mineralization optimized – MO).

2.6. Protein Isolation and Western Blotting

Pooled PDLCs (donors 4–6) were encapsulated in AO or MO hydrogels and cultured in inductive medium. At the specified timepoint, hydrogels were rinsed twice in PBS and then homogenized in RIPA buffer supplemented with a protease/phosphatase inhibitor cocktail (Cell Signaling Technology, MA, USA). Lysates from three individual hydrogels were pooled and transferred to an Amicon ultracentrifugation tube (3 kDa MWCO) and centrifuged at 14,000 g for 20 minutes. Total protein in each sample was determined using the Pierce^™ BCA protein assay (Thermo Scientific, MA, USA). Protein was suspended in Laemmli buffer (Bio-Rad, CA), and loaded into NuPage^™ 4 to 12% Bis-Tris gels (Invitrogen) at 15 μg protein per lane. Protein was transferred to PVDF membranes (iBlot^™ 2, Invitrogen) using an iBlot^™ 2 gel transfer device (Invitrogen). Membranes were blocked in 5% nonfat milk (Bio-Rad) for 1 hour, then incubated with rabbit primary antibodies (Cell Signaling Technology, Supplemental Table 2) overnight at 4 °C, followed by application of goat anti-rabbit secondary IgG-HRP conjugate antibody (Bio-Rad) for 2 hours at RT. Blots were reacted with SuperSignal^™ West Femto substrate (Thermo Scientific) and imaged on a ChemiDoc MP (Bio-Rad). ImageJ was used to quantify protein levels.

2.7. PDLC Immunofluorescence Staining and Analysis

Hydrogels were rinsed twice in PBS and fixed in 4% paraformaldehyde for 15 minutes, then rinsed three times. Blocking was performed using 5% BSA with 0.1% Triton X-100 (Sigma) for 1 hour. Hydrogels were stained with an antibody against active $\text{β1}$ integrin (mouse anti-human 12G10, 10 μg/mL, Invitrogen) for 24 hours at 4 °C, rinsed three times with 3% BSA, 0.05% TritonX-100, followed by staining with Alexa Fluor^™ 488 goat anti-mouse IgG (2 μg/mL), Alexa Fluor^™ 568 phalloidin, and DAPI (Invitrogen). Hydrogels were imaged using a spinning disc confocal microscope (Andor Dragonfly, Oxford Instruments, UK) equipped with a Nikon Plan Apo 60X/1.20 water immersion objective. Z-stack images of individual PDLCs of the hydrogel were taken at 0.29 μm intervals within 50–150 μm from the hydrogel surface. Maximum intensity z-projections of the $\text{β1}$ integrin channels was produced, background subtraction was performed with a rolling ball radius of 30 pixels, and the integrated density was acquired for each PDLC using ImageJ. A set threshold for the $\text{β1}$ integrin channel was then applied to all images and the particle analysis tool in ImageJ was used to count the discrete stained regions per PDLC to give an estimate of the number of active $\text{β1}$ integrin clusters.

2.8. PDLC mRNA Isolation and qPCR

At the specified timepoint, hydrogels were rinsed in PBS and transferred into hydrogel digestion buffer (1000 U/μL collagenase II, Invitrogen) and incubated at 37 °C for 30–45 minutes until the hydrogel was no longer visible. PDLCs from two hydrogels were then pooled and rinsed twice in PBS before suspension in TRIzol (Invitrogen). Following phase separation with chloroform, mRNA was transferred to spin columns (E.Z.N.A.^®, Omega Bio-Tek, GA, USA) for rinsing and DNAse treatment. 1000 ng/μL mRNA was then reverse transcribed to cDNA with iScript^™ cDNA synthesis kit (Bio-Rad). qPCR was performed with custom primers (Supplemental Table 2) and PowerUp^™ SYBR^™ green mastermix, and relative gene expression was calculated using RPL32 as the housekeeping gene[30].

2.9. Determination of Hydrogel Calcium and Collagen Content

After 3 weeks of culture, hydrogels were rinsed in PBS, homogenized in RIPA buffer and centrifuged at 14,000 g. A portion of the supernatant was removed to determine DNA concentration. The remainder of the supernatant and hydrogel was subjected to assays to determine calcium or collagen content. For calcium content, 1 mL 0.6 M HCl was added to each homogenized hydrogel and incubated at 4 °C for 24 hours then centrifuged at 14,000 g. 4 μL supernatant or standard (CaCl₂, Sigma) was reacted with 50 μL each of calcium reagent (55 μM o-cresolphthalein complexone (Sigma) and 8 mM 8-hydroxquinoline (Sigma) in ddH₂0, pH 0.7) and buffer (488 mM 2-amino-2-methyl-1-propanol (Sigma) in ddH₂0) for 10 minutes, and absorbance read at 570 nm. For collagen content, 1 mL of 1 mg/mL pepsin (Sigma) in 0.5 M acetic acid was added to homogenized hydrogels which were then incubated at 4 °C for 24 hours. After centrifugation at 14,000 g, the supernatant was filtered with Amicon ultracentrifugation tubes (100 kDa MWCO, MilliporeSigma, USA) to remove non-collagenous serum proteins[31] and the retentate rinsed twice in 0.5 M acetic acid. 30 μL sample or standard (rat tail collagen, Corning, NY, USA) was diluted in 70 μL 0.5 M acetic acid, incubated with 1 mL Picosirius red dye (Rowley Biochemical, MA, USA) for 30 minutes at room temperature, then centrifuged at 14,000 g for 10 minutes. The resulting collagen-dye pellet was rinsed with ice cold acid-salt solution (0.5 M acetic acid, 0.9 M NaCl) and then dissolved in 0.5 M NaOH to read absorbance at 555 nm.

2.10. ENPP1 Activity and Pyrophosphate Concentration

Ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1) activity was determined following homogenization of hydrogels in RIPA. The ENPP1 substrate, thymidine 5’-monophosphate p-nitrophenyl ester sodium salt (Sigma), was dissolved at 1 mg/mL in diethanolamine buffer, and relative ENPP1 activity was measured via absorbance at 405 nm and normalized to hydrogel DNA content using PicoGreen (Invitrogen). For measurement of pyrophosphate (PP_i) concentration, hydrogels were transferred to new wells with 500 μL inductive medium and cultured for 24 hours. 40 μL of medium or standard (sodium pyrophosphate tetrabasic, Sigma) was then combined with 10 μL each of PPiLight^™ converting and detection reagent (Lonza, MD, USA) in 96 well plates and luminescence was read once every minute for 10 minutes. The slope of luminescence versus time was used to determine pyrophosphate concentrations.

2.11. Rat Periodontal Fenestration Defect Creation and Hydrogel-Mediated PDLC Transplantation

A rat periodontal fenestration model[32] was adapted for transplantation of hydrogels with encapsulated PDLCs using a University of Rochester Committee of Animal Research-approved protocol (UCAR-2020–012) and carried out in compliance with ARRIVE guidelines[33]. 18 male nude rats (Crl:NIH-Foxn1rnu, ages 6–7 weeks (180–200 grams body weight) were acquired from Charles River (MA, USA). Buprenorphine SR 1 mg/kg (ZooPharm^™, WY, USA) and carprofen 5 mg/kg (Rimadyl, Zoetis, NJ, USA) was administered subcutaneously for analgesia and isofluorane for anesthesia. Single defects were made in the right mandibles of each rat via extraoral incisions. A surgical drill (Elcomed, W&H, AT) and round burs with copious saline irrigation were used to create a defect in the mandibular bone, 3 mm in width, 2 mm in height and 1 mm depth, removing the buccal root and exposing the distal root of the 1^st molar. Hand instruments were used to remove PDL, and cementum, and superficial dentin on the distal root. Twelve rats randomly received either 5 μL of AO or MO hydrogel solution containing 1×10⁴ suspended PDLCs (donors 4–6), with hydrogels formed in situ using a spot UV illuminator (3 minutes, 365 nm, ~5 mW/cm²) (AmScope, CA, USA). Defects were created in six additional rats and left empty (untreated) with the same UV illumination applied. Muscle and skin were closed separately with Vicryl Rapide and Ethilon sutures (Ethicon, NJ, USA), respectively. Carprofen was administered daily for the first three days post-operatively and the rat diet was supplemented with soft food (Nutra-Gel, Bio-Serv, NJ, USA) for the first week.

2.12. Rat Sample Preparation and Analysis

After 3 weeks of healing, rats were euthanized with CO₂ inhalation. Right mandibles were removed together with soft tissues overlying the defect and fixed in 10% neutral buffered formalin for 3 days. Mandibles were scanned with a Scanco Medical VivaCT 40 cone-beam $\text{μCT}$ (55 kV, 145 $\text{μA}$, 300 ms integration time, resolution of 10.5 μm voxels, 2048 samples, 1000 projections over 180°) then decalcified in 14% EDTA with glacial acetic acid (pH 7.4 – 7.6) for 3 weeks. Mandibles were then embedded in paraffin and 5 μm sections were acquired in the coronal plane.

The $\text{μCT}$ system software (Scanco v6.50) was used to measure bone defect fill (bone volume/total volume) and bone mineral density within the fenestration defect by a blinded observer. The defect region of interest (ROI) was defined using the lateral, superior, and inferior borders of the bone defect, the surface of the distal root, and the external bone surface. In the case the bone defect borders were not identifiable, the ROI was defined using the instrumented surface of the distal root to set the superior and inferior border and setting the mesial defect border 2 mm from the root center and the distal border 1 mm from the root center. The threshold used to distinguish bone was a linear attenuation coefficient of 2.24^-cm. A standard Gaussian filter with a sigma of 0.8 and a support of 1 pixel was used for noise reduction. $\text{μCT}$ Images were also reconstructed using Amira software (Thermo Fisher) to visualize the defect area and newly formed bone.

Three paraffin sections per root, each separated by at least 50 μm in distance, were deparaffinized, rehydrated, and stained with hematoxylin and eosin then scanned with a Cytation5 imaging system (Agilent, CA USA). ImageJ was used to measure total root length and root length with new cementum. Immunohistochemical staining was performed on adjacent paraffin sections. After deparaffinization and rehydration, antigen retrieval was performed with a citrate buffer (10 mM sodium citrate, 0.05% Tween 20, pH). Sections were blocked with BLOXALL solution (Vector Labs, CA, USA) and CAS-Block^™ (Thermo Fisher) then incubated with primary antibodies against bone sialoprotein (BSP) (WVID1(9C5), 1:100, Developmental Studies Hybridoma Bank, IA, USA) or Ku80 (C48E7, 1:600, Cell Signaling Technology) for 2 hours at RT. After rinsing, ImmPRESS IgG polymer reagents (Vector Labs) were applied for 30 minutes followed by application of ImmPACT DAB EqV working solution (Vector Labs). Slides were rinsed, counterstained with hematoxylin, and scanned. Analysis of human Ku80+ nuclei in defects was performed on three separate sections per root using ImageJ with an adapted protocol[34] using set color threshold values for hematoxylin and DAB channels and the same nuclei size range applied for all sections.

2.13. Statistical Analysis

Data was analyzed using Prism software (GraphPad, CA, USA) and presented as mean ± standard deviation. For animal experiments, a power analysis was performed based on previous studies[35–37] to determine sample size. Data distributions were analyzed using the Shapiro-Wilk test. For single PDLC level immunofluorescence analyses, outliers were identified and removed using the robust regression and outlier tool (ROUT) in Prism. Unpaired t tests were used for comparisons of two independent groups. One- and two-way ANOVA was used for comparisons with three or more groups, using a post hoc Dunnett correction when comparing all peptide groups to control (0 mM, Ctrl) or a post hoc Tukey correction when making comparisons between all independent groups. P<0.05 was considered statistically significant.

Results and Discussion

3.1. Isolation of PDLCs and Encapsulation in Peptide-Functionalized Hydrogels

PDLCs were isolated from third molars of six human donors and cultured individually (donors 1–3) or pooled at the time of isolation then cultured (donors 4–6). PDLCs were uniformly positive for stromal cell markers CD90 and CD105 and negative for endothelial marker CD31 and hematopoietic marker CD45 while expression of multipotency markers CD146 and STRO-1 varied between donors (Supplemental Table 1). Greater than 95% of all donors’ PDLCs expressed integrins $\text{α2}$, $\text{α5}$, and $\text{β1}$, as assayed by flow cytometry while expression of integrin $\text{αvβ3}$ ranged from 51–95% (Supplemental Table 1). PDLCs from all donors showed multi-differentiation capacity and demonstrated changes in cementoblast/osteoblast gene expression when cultured in inductive medium (Supplemental Figure 1).

A range of stiffness-matched (~9 kPa) PEG hydrogels were formed with an MMP-degradable crosslinker and varying concentrations of RGD, GFOGER and/or scrambled (Ctrl) peptides (Supplemental Figure 2). PDLCs maintained >90% viability and spread within all hydrogels (Supplemental Figure 3). In single ligand hydrogels (RGD or GFOGER), the degree of PDLC spreading was dependent on both the peptide and peptide concentration, while spreading was similar for all dual peptide hydrogels at both early (1 week) and late (3 weeks) time points (Supplemental Figure 3).

3.2. Peptide Screening for Optimized Hydrogel Compositions

First, RGD or GFOGER alone were screened for their ability to alter ALP activity (at 1 week) and hydrogel matrix mineralization (measured at 3 weeks) at low, medium, and high concentrations (0.1, 1.05, 2 mM). For ALP activity, both the pattern and magnitude of the response were highly variable between donors, while mineralization tended to increase with higher concentrations of either RGD or GFOGER (Supplemental Figure 4).

RGD and GFOGER were next tested in combination to better mimic the periodontal ECM. Both ALP activity and mineralization were potentiated by dual peptide presentation with a distinct response pattern apparent for each outcome (Figure 1A-B). ALP activity increased from less than 2-fold for RGD or GFOGER alone versus controls to nearly 4-fold for high levels of RGD combined with low or moderate levels of GFOGER (Figure 1A), and mineralization increased up to 5-fold for moderate RGD and high GFOGER (Figure 1B). Combinatorial presentation outcomes further showed that low (0.1 mM) RGD or GFOGER concentrations had a minimal impact on ALP activity or matrix mineralization, respectively, and indicated a possible ligand saturation effect for high levels of RGD and GFOGER in combination (Figure 1A-B).

Figure 1.

PDLC activity in peptide-functionalized hydrogels. A) Fold-change ALP activity at 1 week and B) fold-change matrix mineralization at 3 weeks for high single peptide (2 mM) and dual peptide concentrations versus Ctrl. C-D) Regression coefficient estimates for fold-change ALP activity of individual donor PDLCs (C) and pooled values of donor 1–3 PDLCs (D). E-F) Regression coefficient estimates for fold-change mineralization of individual donor PDLCs (E) and pooled values of donor 1–3 PDLCs (F). G-H) Response surfaces for fold-change ALP activity (G) and fold-change matrix mineralization for pooled values of donor 1–3 PDLCs (H). N=3 hydrogels per donor. I) ALP activity and J) matrix mineralization in hydrogels optimized for ALP activity (AO – gray bars) or matrix mineralization (MO – white bars) using PDLCs from pooled donors 4–6. ALP activity was also measured at 1 week on tissue culture polystyrene using the same pooled PDLC population (black bar). K) Representative z-stack images of hydrogel matrix mineralization. Xylenol orange-stained calcium deposits are red, Hoechst 33342-stained PDLC nuclei are blue. Scale bars are 100 μm. L) Total calcium concentration in optimized hydrogels normalized to PDLC DNA content. N=6 hydrogels. A-B) One-way ANOVA with Dunnett post hoc. 1, 2, and 3 indicate p<0.05 for individual donor’s PDLCs versus individual Ctrl, a, b indicates p<0.05 for dual peptide pooled means versus 2 mM RGD (a) and 2 mM GFOGER (b) pooled means, **p<0.01, ***p<0.001 for pooled means versus Ctrl. C-F) ANOVA. I) One-way ANOVA with Tukey post hoc. J, L) unpaired t test. *p<0.05, ***p<0.001, ****p<0.0001 for between group comparisons.

Results for dual peptide presentation were fit to a response surface model, which confirmed that RGD and GFOGER together differentially affected ALP activity and matrix mineralization for individual donor (Figure 1C, E) and pooled PDLCs (Figure 1D-F). For ALP activity, RGD had an overall strong positive effect, while GFOGER had a small but significant inhibitory effect (Figure 1D). In contrast, GFOGER had a strong positive impact on matrix mineralization, together with a small positive effect for RGD (Figure 1F). Response surfaces were used to illustrate the divergent impact of dual peptide presentation and identify the dual peptide concentration estimated to give the maximal response for ALP activity (ALP optimized (AO): 2 mM RGD + 0.1 mM GFOGER) and mineralization (mineralization optimized (MO): 1.05 mM RGD + 2 mM GFOGER) (Figure 1G, H). To test the utility of this approach, pooled PDLCs from 3 additional donors (donors 4–6), serving as a representative heterogeneous PDLC population, were cultured in AO and MO hydrogels. ALP activity at 1 week was 3.5-fold greater in AO versus MO hydrogels and was 12- and 3.5-fold higher in AO and MO hydrogels compared to the same PDLCs grown on tissue culture polystyrene (Fig. 2I), similar to the previous study showing an overall elevation of PDLC ALP activity in MMP-degradable hydrogels vs. standard culture[27]. Matrix mineralization was 3-fold greater in MO hydrogels versus AO hydrogels (Fig. 2J, K) and was accompanied by significantly higher total hydrogel calcium content (Fig. 2L).

Figure 2.

PDLC integrin and focal adhesion expression in AO (gray bars) and MO hydrogels (white bars). A) Representative western blots of protein from AO and MO hydrogels at 3 and 7 days. B-E) Quantification of western blot signals at 3 (B, D) and 7 days (C, E) using $\text{β}$-Actin as loading control. N=6 replicates, n=18 hydrogels. F) Representative images of PDLCs in optimized hydrogels at 3 and 7 days with actin cytoskeleton stained red (Phalloidin), nuclei blue (DAPI), and active integrin $\text{β1}$ green. Scale bars are 20 μm. G-H) Total fluorescent integrin $\text{β1}$ signal given as integrated density for individual PDLCs at 3 (G) and 7 days (H). I-J) number of discrete integrin $\text{β1}$ stained clusters in individual PDLCs at 3 (I) and 7 days (J). N=4 hydrogels, n=32 PDLCs. Unpaired t test. *p<0.05, ***p<0.001, ****p<0.0001.

Intriguingly, the contrasting levels of RGD and GFOGER peptides between optimized hydrogels parallel different ECMs present during periodontal wound healing. Initially, periodontal wounds fill with fibronectin- and vitronectin-rich granulation tissue, while collagen type I is restricted to bands along root surfaces and alveolar bone, sites of future cementum and bone formation[12, 38]. Cementum and alveolar bone surfaces also undergo transient resorption, exposing a collagen-rich surface thought to play a role in cementoblast and osteoblast attachment and differentiation[39]. Thus, replicating a collagen ligand-enriched matrix through high levels of GFOGER with RGD may provide a more specific signal for osteoblast or cementoblast PDLC differentiation, while high concentrations of RGD may signal to PDLCs to form non-mineralized PDL tissues or regulate mineralization by other cells. An additional consideration is that RGD binding sites can become exposed alongside GFOGER on denatured collagen[40], which is present in wounds and remodeling tissues[41]. Duplicating these collagen binding sites with partially denatured collagen can enhance cell adhesion and osteogenic differentiation in comparison to intact collagen[42], similar to the current results observed with dual RGD and GFOGER presentation within hydrogels.

3.3. Integrin Expression and PDLC Adhesions in Optimized Hydrogels

Integrin engagement and PDLC-matrix adhesions were investigated within optimized hydrogels to relate PDLC activity to physical interactions with the hydrogel matrix. The $\text{β}$ integrin subunit plays a distinct role in the transfer of physical signals from the ECM to cells[10]. Total $\text{β1}$ integrin protein levels were similar between AO and MO hydrogels at 3 days or 1 week of culture time points (Figure 2A, Supplemental Figure 5), while $\text{β3}$ integrin levels showed a slight but significant elevation in AO hydrogels at 3 days (Figure 2A-C). Focal adhesion kinase (FAK) is a key component of cell adhesions and has been specifically implicated in integrin-mediated cellular differentiation and osteogenesis[43, 44]. FAK protein levels were elevated 2.5-fold in MO over AO hydrogels at 3 days, while levels were similar at 1 week (Figure 2A, D). Activated FAK (pFAK (Tyr397)) was barely detectable at 3 days and was expressed at similar levels in both hydrogels at 7 days (Figure 2A, Supplemental Figure 5). 3D matrix adhesions, such as those formed by cells within collagen or fibronectin-based matrices, include $\text{β1}$ integrin and FAK[45, 46] but not necessarily pFAK or $\text{β3}$ integrins[47]. Delays in adhesion formation and FAK phosphorylation may also occur for cells within dense hydrogels which could necessitate remodeling and re-organization of surrounding integrin-binding ligands prior to development of sufficient cellular tension[48].

As $\text{β1}$ integrin can have both active and inactive conformations[49], immunofluorescence staining was performed with an active $\text{β1}$ integrin-specific antibody (12G10) (Figure 2F-J). At 3 days, PDLCs displayed an overall round shape with smaller projections into the hydrogel matrix and multiple distinct regions of integrin $\text{β1}$ staining on the cell surface (Figure 2F). By 7 days, PDLCs had elongated and continued to show multiple regions of active integrin $\text{β1}$ staining. Total active integrin $\text{β1}$ fluorescent signal was significantly higher in MO versus AO hydrogels at both 3 and 7 days (Figure 2G-H). The number of discrete integrin $\text{β1}$-stained regions, suggestive of integrin clusters, was also significantly greater in MO hydrogels at both time points (Figure 2I-J). Altogether, PDLCs in MO hydrogels with higher concentrations of GFOGER peptide exhibited enhanced and accelerated adhesion formation, characterized by elevated FAK at early timepoints and increased active $\text{β1}$ integrin.

3.4. PDLC Gene Expression Optimized Hydrogels

Gene expression was evaluated to determine if PDLC activity within either optimized hydrogel correlated with expression of key cementoblast/osteoblast markers. In AO hydrogels, expression of SP7 (Osterix) and IBSP (Bone sialoprotein 2, BSP) remained near baseline levels through 3 weeks (Figure 3A, B). In contrast, PDLCs in MO hydrogels showed dramatic increases in both SP7 and IBSP expression, with the former showing an 18-fold increase at 2 weeks and the latter increasing 17-fold at 3 weeks (Figure 3A, B). SPP1 (Osteopontin, OPN) showed opposing expression patterns between hydrogels, with expression initially suppressed below baseline in AO hydrogels before rising at 3 weeks, while remaining steady in MO hydrogels until 3 weeks when expression dropped 5-fold (Figure 3C). Critically, PDLC gene expression patterns in MO hydrogels parallel timelines of periodontal mineralized tissue formation. SP7 expression is transiently increased in PDLCs differentiating toward cementoblasts during root formation[50]. OPN is expressed early throughout periodontal wounds, while BSP expression peaks later and is restricted to areas of cementum and bone formation[51]. The opposing pattern of late SPP1 expression in AO and MO hydrogels also fits well with the mineralization-inhibiting role of OPN[52]. PDLCs in both hydrogels exhibited a similar pattern of RUNX2 expression, while MO hydrogels supported increased BGLAP, CEMP1, and CAP expression at 3 weeks (Supplemental Figure 6), further indicating that elevated hydrogel mineralization, but not necessarily ALP activity, identifies PDLCs with a cementoblast/osteoblast phenotype.

Figure 3.

PDLC gene expression and collagen content in AO (gray bars) and MO (white bars) hydrogels. A-E) PDLC gene expression relative to expression 1 day after encapsulation (baseline) using RPL32 as housekeeping gene. N=6 replicates, n=12 hydrogels. F) Total collagen content in hydrogels at 3 weeks of culture. N=6 hydrogels. A-E) Two-way ANOVA with Tukey post hoc, *p<0.05, ***p<0.001, ****p<0.0001 for between group comparisons, & p<0.01, % p<0.0001 for within group comparisons. F) Unpaired t test, ns: not significant

Periostin is a highly expressed PDL ECM protein, regulating collagen fibrillogenesis and maintaining the integrity of PDL attachment to cementum and bone[53]. In 2D culture, periostin can promote PDLC osteogenic differentiation through RGD-independent integrin $\text{αvβ3}$ binding[54]. Periostin (POSTN) expression was elevated 2–3 fold in MO hydrogels while remaining below baseline expression in AO hydrogels (Figure 3D), pointing to its potential as a PDLC differentiation marker in 3D matrices as well. COL1A1 expression increased over time in both hydrogels (Figure 3E), and total collagen content was similar between AO and MO hydrogels (Figure 3F). While the presence of collagen type I is required for ECM mineralization, non-collagenous proteins, including BSP and OPN, regulate mineralization of collagen fibrils[55]. Other genes previously associated with the regulation of periodontal mineralization, ASPN and S100A4[56, 57], showed limited differences in temporal expression patterns between optimized hydrogels (Supplemental Figure 6), suggesting that additional factors were regulating hydrogel mineralization.

3.5. PDLC Pyrophosphate (PP_i) Regulation in Optimized Hydrogels

Extracellular pyrophosphate (PP_i) is a negative regulator of mineralization, strongly inhibiting hydroxyapatite crystal formation and growth[58]. PP_i is produced from nucleotide triphosphates by the cell membrane enzyme ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1) and is transported to the extracellular matrix from intracellular PP_i pools by the transmembrane protein ANKH. ALP hydrolyzes PP_i to produce inorganic phosphate (P_i) [59] and enables the mineralization process of cementum and bone[7, 60]. ENPP1 and ANKH have also been implicated in regulating PP_i to P_i levels to maintain the ALP-rich but non-mineralized PDL[7].

When ALP activity was measured over the entire 3-week culture period, significant increases were noted at each time point in AO hydrogels, while only small, insignificant increases occurred in MO hydrogels (Figure 4A). In parallel, ENPP1 activity was consistently greater (Figure 4B) and extracellular PP_i levels were maintained at a relatively constant level through 3 weeks in AO hydrogels (Figure 4C). In contrast, PP_i levels were significantly higher in MO versus AO hydrogels until 3 weeks levels, when a significant decrease to 4.4 μM was observed, below the 5 μM level previously reported to be inhibitory for in vitro mineralization[52](Figure 4C). Distinct patterns were noted in the expression of PP_i-associated genes. At 1 week, ALPL expression was elevated nearly 40-fold in AO hydrogels alongside significantly greater ENPP1 and ANKH expression versus MO hydrogels (Figure 4D-F). As culture time progressed, ENPP1 and ANKH expression remained stable in AO hydrogels, while ALPL and ANKH expression increased in MO hydrogels (Figure 4D-F). Altogether, PDLCs in AO hydrogels maintained stable PP_i levels, despite elevated ALP activity, which contrasted with more dynamic PP_i levels in MO hydrogels where a reduction in PP_i corresponded with increased matrix mineralization.

Figure 4.

Pyrophosphate regulation in AO (gray bars) and MO (white bars) hydrogels. A) ALP activity, B) Relative ENPP1 activity and C) Extracellular pyrophosphate concentrations at 1, 2, and 3 weeks culture. N=6 hydrogels. D-F) PDLC gene expression relative to expression 1 day after encapsulation using RPL32 as housekeeping gene. N=6 replicates, n=12 hydrogels. Two-way ANOVA with Tukey post hoc. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 for between group comparisons, # p<0.05, & p<0.01, $ p<0.001, % p<0.0001 for within group comparisons.

To further elucidate the relationship between PDLC ALP activity, PP_i regulation, and mineralization, optimized hydrogels were cultured with different concentrations of $\text{β}$-glycerophosphate, which is rapidly converted to P_i by ALP[61]. At 1 week, both ALP and ENPP1 activity was uniformly elevated in AO hydrogels by all concentrations of $\text{β}$-glycerophosphate (Figure 5A, B). PP_i levels progressively increased in both hydrogels as the concentration of $\text{β}$-glycerophosphate increased, suggesting extracellular PP_i production may be a compensatory PDLC response to P_i during early culture (Figure 5C). Accordingly, introduction of the other components of inductive media, ascorbic acid and dexamethasone, alone led to significant increases in ALP and ENPP1 activity in AO hydrogels but only a small elevation in PP_i levels in both hydrogels (Supplemental Figure 7). By 3 weeks, ALP activity was elevated in AO hydrogels only by the highest concentration of $\text{β}$-glycerophosphate (Figure 5D) while ENPP1 activity showed dose-dependent increases in both hydrogels (Figure 5E). In parallel, PP_i concentrations progressively increased in AO hydrogels with increasing $\text{β}$-glycerophosphate levels, continuing the compensatory-like response seen at 1 week, but remained relatively stable and below 5 μM in MO hydrogels (Figure 5F). Finally, mineralization increased with increasing concentrations of $\text{β}$-glycerophosphate, reaching significance for MO hydrogels at 5 and 10 mM (Figure 5G, H). These findings, together, indicated that PDLC can respond to P_i to inhibit or enhance mineralization through PP_i regulation, and this response can be altered through presentation of integrin-binding peptides.

Figure 5.

PDLC response to $\text{β}$-glycerophosphate in AO (gray bars) and MO (white bars) hydrogels at 1 (A-C) and 3 weeks (D-F). A,D) ALP activity at 1 (A) and 3 weeks (D), B,E) Relative ENPP1 activity at 1 (B) and 3 weeks (E). C,F) Pyrophosphate concentrations at 1 (C) and 3 weeks (F). G) Mineralization at 3 weeks. H) Representative z-stack images of calcium deposits (Xylenol orange) in hydrogels at 3 weeks. PDLC nuclei are stained blue with Hoechst 33342. Scale bars are 100 μm. N=3–6 hydrogels. Two-way ANOVA with Tukey post hoc. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 for between group comparisons, & p<0.01, $ p<0.001, % p<0.0001 for within group comparisons.

3.6. Transplantation of PDLCs in Optimized Hydrogels

PDLCs with the distinct in vitro behaviors identified here, demonstrating either a propensity for elevated ALP activity or mineralization and expression of cementoblast/osteoblast genes, may also have potential for improving mineralized tissue formation in vivo. ALP activity and PP_i regulation are strongly implicated in the formation and maintenance of acellular cementum, the cementum subtype associated with inserting PDL fibers[62, 63]. Manipulation of extracellular PP_i levels is an emerging approach for periodontal tissue regeneration, with periodontal defects in Ank knockout and Ennp1 mutant (Enpp1^asj/asj) mice showing increased new cementum formation during wound healing[64], and delivery of recombinant human ALP to mouse defects promoting formation of both cementum and bone[65]. A direct role for transplanted cells in mineralized tissue formation has also been reported in various periodontal defect models[66, 67], pointing to the therapeutic potential of a hydrogel matrix that directs transplanted PDLCs toward a cementoblast/osteoblast phenotype. These findings, together with the results of the current study, motivated the testing of both optimized hydrogels for PDLC delivery to periodontal defects to determine the utility of this approach and the relevance of these in vitro PDLC phenotypes.

Mandibular fenestration defects were created in athymic rats and AO and MO hydrogels were used to transplant PDLCs (pooled donors 4–6). A single healing time point, three weeks post-surgery, was chosen for analysis to maximize the chances of identifying differential levels of bone and cementum formation[35–37] while minimizing the number of animals. All rats survived the surgical procedure with no adverse effects occurring during healing. Bone formation, as evaluated through $\text{μCT}$, was altered by hydrogel placement (Figure 6A-D). Bone regeneration in MO hydrogel-treated defects was significantly greater than all other groups, showing a 56% increase in defect fill over untreated controls and a 26% increase versus AO hydrogel-treated defects, while AO defects showed no difference compared to untreated defects (Figure 6D). The mineral density of newly formed bone was significantly greater in both hydrogel groups versus untreated defects (Figure 6E). Despite improved bone formation, total defect fill was 50% or less in all hydrogel-treated defects (Figure 6C). This may be attributed, in part, to the absence of any space-maintaining scaffold or barrier material overlaying the defect. Such materials play an important role in clinical periodontal defects[68], but were omitted in the present study to minimize the number of experimental variables.

Figure 6.

Bone formation in periodontal fenestration defects at 3 weeks of healing. A-C) Representative $\text{μCT}$ renderings and coronal cross-sections of untreated defects (A) and defects receiving AO (B) and MO (C) hydrogels. Scale bars are 1 mm. D) Bone defect fill, given as the ratio of new bone volume to total defect volume in untreated defects (patterned bar) and defects treated with AO (gray bar) and MO hydrogels (white bar). E) Bone mineral density of new bone in defects. N=6 rats. One-way ANOVA with Tukey post hoc. *p<0.05, ***p<0.001.

Histologic sections showed no evidence of remaining hydrogels, suggesting that they fully degraded over the three week period. New cementum formation was limited on roots within untreated defects (Figure 7A.ia), while hydrogel-treated defects showed the formation of a thin layer of BSP-positive cementum on 45–75% of the root surfaces (Figure 7B.ia–C.ib). No areas of ankylosis between root and new bone were found in either control or treated defects. New PDL fibers were visible in portions of hydrogel-treated roots inserting into new cementum and/or bone (Figure 7B.ia–C.ia). In untreated defects, areas of disorganized cellular cementum and root resorption were noted adjacent to areas of new bone (Supplemental Figure 8), findings absent within hydrogel-treated defects.

Figure 7.

Histomorphometric analysis of cementum formation and human PDLC localization in fenestration defects at 3 weeks. A-C) Overview scans of coronal sections of the first molar distal root in untreated (A), and AO (B) and MO hydrogel-treated (C) groups stained with hematoxylin and eosin with the locations of insets ia-ic, ii, and iii indicated by black outlines. A.ia-C.ia) insets demonstrating new bone (*) and acellular cementum formation (arrows). A.ib-C.ib) Adjacent sections stained for bone sialoprotein (BSP) (brown). B.ic-C.ic, C.ii-iii) Adjacent sections stained for human nuclear protein Ku80 (brown). Scale bars in overview scans are 1 mm, scale bars in insets are 100 μm. D) Percentage of root surface with new cementum. E) Percentage of Ku80+ within defects. N=6 rats. D) One-way ANOVA with Tukey post hoc. ***p<0.001. E) Unpaired t test.

The persistence of transplanted PDLCs in hydrogel-treated defects was evaluated with a human-specific antibody against nuclear protein Ku80. Positively stained nuclei were present in all hydrogel-treated defects, ranging from 0.5 to 7.5% of all cells (Figure 7B.ic, C.ic, D). In some sections, Ku80+ nuclei were present within the PDL space and close to, but not contacting the root surface, and in connective tissues separated from new bone by a layer of host cells (Figure 7B.ic, C.ii-iii). Regions defined by more advanced formation of the bone-PDL-cementum complex often showed few or no Ku80+ nuclei (Figure 7C.ic).

The absence of Ku80+ osteocytes, cementocytes, or osteoblast- and cementoblast-like cells suggested that the transplanted PDLCs primarily played an indirect role in mineralized tissue formation in this model. Such a finding may account for the similar levels of cementum formation between hydrogel groups, despite elevated expression of cementoblast markers in PDLCs within MO hydrogels. Nevertheless, the presence of hydrogels and/or transplanted PDLCs with elevated ALP activity may have created a conducive environment for cementum formation without extensive root resorption or ankylosis, a putative function of in vivo PDLCs[7]. Host cell integrin binding to GFOGER peptides on MO hydrogels may also have played a role in the increased bone formation found in MO hydrogel-treated defects.

Similar cell localizations have been noted when human[69] or allogeneic rat[70, 71] PDLCs were transplanted into athymic rats. In contrast, one study identified transplanted allogeneic PDLCs as differentiated osteoblasts and cementocytes within immunocompetent rats[72], pointing to the impact of the host immune system on transplanted cell integration. Secreted factors by transplanted PDLCs have been shown to exert anti-inflammatory and immunomodulatory effects on host cells[73, 74], which likely influenced outcomes in the current study. In addition, recent studies have demonstrated the ability of scaffolds to alter the interplay between transplanted cells and/or host regenerative and immune cells in bone and periodontal defects to improve tissue formation[75, 76].

Further studies are required to fully investigate the predictive ability of in vitro hydrogel-based screening assays and the ability of hydrogel-mediated PDLC transplantation to alter periodontal tissue repair. The use of hydrogel-only controls and early and later time points may help elucidate how integrin-binding peptide presentation can alter host and/or transplanted periodontal cell crosstalk. The versatile nature of PEG-based hydrogels can allow further modification of the current hydrogel design to improve outcomes. Hydrogel degradation properties can be tuned to control encapsulated cell retention[77] and direct host tissue infiltration[26] to better coordinate the interplay between cells and scaffolds. Finally, layered hydrogel materials or the combination of hydrogels with stiff biomaterials have shown promise for multi-tissue repair in other models[78], pointing to the potential of engineered PEG hydrogels as a component of multiphasic periodontal scaffolds.

Conclusion

Overall, the presentation of RGD and GFOGER peptides in PEG hydrogels directed integrin interactions and enabled identification of two PDLC phenotypes with distinct patterns of gene expression and PP_i regulation. Functionalization of hydrogels with both RGD and GFOGER peptides elevated ALP activity and matrix mineralization over presentation of single peptides alone across heterogeneous PDLC populations. Increasing the concentration of integrin $\text{α2β1}$-binding GFOGER peptide alongside moderate RGD levels strengthened early PDLC adhesion formation, triggered elevated expression of cementoblast/osteoblast markers, and supported both in vitro matrix mineralization and in vivo bone formation. In contrast, high levels of RGD and low levels of GFOGER defined a PDLC phenotype with high ALP activity and a capability for maintaining stable hydrogel PP_i levels. Despite showing minimal in vitro hydrogel mineralization, ALP activity optimized hydrogels promoted formation of new cementum when used as a scaffold for PDLC transplantation. While in vitro assays used to identify the optimized peptide compositions did not fully replicate native PDLC activity, these studies were instrumental in identifying peptide compositions that supported distinct PDLC phenotypes and differentially affected periodontal wound healing. These results lend further support to this biomaterials approach for investigating the unique role of PDLCs and ALP in the PDL and engineering hydrogels to serve as scaffolds for PDLC transplantation.