Surface Immobilization Chemistry of a Laminin-Derived Peptide Affects Keratinocyte Activity

Many chemical routes have been proposed to immobilize peptides on biomedical device surfaces, and in particular, on dental implants to prevent peri-implantitis. While a number of factors affect peptide immobilization quality, an easily controllable factor is the chemistry used to immobilize peptides. These factors affect peptide chemoselectivity, orientation, etc., and ultimately control biological activity. Using many different physical and chemical routes for peptide coatings, previous research has intensely focused on immobilizing antimicrobial elements on dental implants to reduce infection rates. Alternatively, our strategy here is different and focused on promoting formation of a long-lasting biological seal between the soft tissue and the implant surface through transmembrane, cell adhesion structures called hemidesmosomes. For that purpose, we used a laminin-derived call adhesion peptide. However, the effect of different immobilization chemistries on cell adhesion peptide activity is vastly unexplored but likely critical. Here, we compared the physiochemical properties and biological responses of a hemidesmosome promoting peptide immobilized using silanization and copper-free click chemistry as a model system for cell adhesion peptides. Successful immobilization was confirmed with water contact angle and X-ray photoelectron spectroscopy. Peptide coatings were retained through 73 days of incubation in artificial saliva. Interestingly, the non-chemoselective immobilization route, silanization, resulted in significantly higher proliferation and hemidesmosome formation in oral keratinocytes compared to chemoselective click chemistry. Our results highlight that the most effective immobilization chemistry for optimal peptide activity is dependent on the specific system (substrate/peptide/cell/biological activity) under study. Overall, a better understanding of the effects immobilization chemistries have on cell adhesion peptide activity may lead to more efficacious coatings for biomedical devices.

A vast number of factors affect peptide immobilization quality such as environmental pH, surface topography, surface charge, surface polarity, peptide-peptide interactions, and immobilization chemistry [21,22]. These factors affect surface density, orientation/domain exposure, peptide intra-and intermolecular structure, and overall biological activity. One easily controllable factor in these systems is the immobilization chemistry [23]. In general, immobilization chemistries are divided into two classes based on whether they are chemoselective or not. Chemoselectivity is potentially important for peptide biological activity as a means to precisely control conformation vis-à-vis activity [24]. Silanization, a process which immobilizes an organofunctional alkoxysilane molecule via hydroxyl groups at the surface of the material which then reacts with peptides, is typically not chemoselective but is attractive because of its long history in biomaterials and relative ease of use [25,26]. On the other hand, strain-promoted, copper-free azide-alkyne click chemistry immobilizes azide or dibenzocyclooctyne (DIBO)-bearing peptides with chemoselectivity but requires additional expensive peptide synthesis steps [27,28]. Little attention has been paid to the comparison of biological outcomes from cell adhesion peptides immobilized with different chemistries despite decades of work immobilizing peptides on surfaces: this is in stark contrast to molecules other than peptides, like poly(ethylene) glycol (PEG) immobilization [29]. However, previous work has shown significant differences in antimicrobial activity between antimicrobial peptides immobilized with surface-initiated atom transfer radical polymerization (SI-ATRP) versus silanization [30].
Here, we aim to compare the physical and chemical properties and biological responses of a cell adhesion peptide immobilized using both silanization and click chemistry. We selected peri-implantitis, or dental implant infection, as a model disease in need of biomedical device innovations. Peri-implantitis rates vary based on many factors but a recent meta-analysis reports an implant-based prevalence of 9.25% and a patient-based prevalence of 19.83% [31]. Other biomedical devices suffer a similar fate [32]; implantable auditory assist devices demonstrate a 26% failure rate [33] and infections occur in up to 77% of individuals with bone-anchored orthopaedic devices [34]. We have selected the peptide LamLG3 identified [35] from a globular module (LG3) of laminin332 (LM332) capable of inducing hemidesmosome (HD) formation from oral keratinocytes in the soft tissue surrounding dental implants [5,36,37]. HDs are a transmembrane "link" or "rivet" between teeth and gingiva ( Figure 1), as the gingiva forms a protective, physical barrier for the tooth, or dental implant, against biofilm invasion [38,39]. Enhanced HD formation on dental implants may contribute to enhanced soft tissue healing around implants and improve their longevity.
Expected effects of click chemistry's chemoselectivity immobilization of the LamLG3 peptides are homogeneous structural display of the tethered peptides on the surface and less restricted peptide interactions with cells than non-selective silanization. Thus, we hypothesized that coatings of LamLG3 immobilized via click chemistry (D-LamLG3) upregulate hemidesmosome formation more in comparison to coatings of LamLG3 immobilized via silanization (S-LamLG3). Overall, a better understanding of immobilization chemistry effects on cell adhesion peptide activity may lead to more efficacious coatings for biomedical devices.

Surface Characterization
2.2.1. X-ray Photoelectron Spectroscopy (XPS)-An X-ray photoelectron spectroscopy (XPS) spectrometer (PHI 5000 VersaProbe III, ULVAC Inc., Chigasaki, Japan) was used to assess the surface elemental composition with a monochromatic Al Kα X-ray source (45°, 50 W, 1486.6 eV, sampling area; 200 μm diameter spot) of the peptideimmobilized surfaces and controls. Survey spectra were collected with a step size of 1.0 eV and a pass energy of 280.0 eV with charge compensation. Spectra were calibrated to the C 1s signal at 284.8 eV using the associated software (MultiPak, 9.6.0).

Water Contact Angle (WCA)-Water contact angles (WCAs)
were obtained using the sessile-drop method (>18 MΩ deionized water; 2 μL droplet) to assess surface wettability using a contact angle goniometer (DM-CE1, Kyowa, Niiza, Japan). Dynamic contact angles were also determined for 60 s every 1 s. Three disks per group were analyzed.

Coating Durability-D-LamLG3
and S-LamLG3 coatings were incubated (37 °C) in artificial saliva (1700-0305, Pickering Laboratories, Mountain View, CA, USA; pH = 6.8) for up to 73 days to assess their simulated intra-oral durability. Disks were periodically removed, thoroughly rinsed in de-ionized (DI) water, and desiccated for subsequent analysis by XPS (as described before). Two disks with three XPS analysis spots were used per timepoint.

Relative Amount of Surface
Peptide-In order to compare the amount of KN3-LamLG3 and 2K-LamLG3 peptides on D-LamLG3 and S-LamLG3, respectively, disks were rinsed in phosphate-buffered saline (PBS) five times and 1% Triton X-100 after coating synthesis. Relative peptide concentration was then determined using a commercially available micro bicinchoninic acid kit (BCA; 23235, Thermo-Fisher, Waltham, MA, USA) by reacting the washed, desorbed disk in BCA solution A standard curve of both D-LamLG3 and S-LamLG3 was prepared and measured on a plate reader (Synergy HT, Biotek, Winooski, VT, USA) similar to past work [42,43]. Three disks per group were analyzed.

Oral Keratinocyte Response
Immortalized TERT-2/OKF-6 (BWH Cell Culture and Microscopy Core, Boston, MA, USA) oral keratinocytes from non-cancerous tissue from the floor of a human mouth were cultured in defined keratinocyte serum-free media (Gibco, Waltham, MA, USA) with 1% penicillin/ streptomycin (Gibco) under standard cell culture conditions [44].

Lactase Dehydrogenase (LDH)
Release-Initial cell viability was determined through lactate dehydrogenase (LDH) release. Cells were seeded as previously described. Disks were then transferred to a new wellplate. A CyQUANT colorimetric assay (C20300, Thermo-Fisher) was used to quantify the amount of LDH in solution, per the manufacturer's instructions (n = 8).

Immunofluorescence and Oral Keratinocyte Hemidesmosome
Formation-Oral keratinocytes were seeded and cultured for one day as described and then fixed for ten minutes in 4% paraformaldehyde or ice-cold methanol (both Fisher Scientific, Waltham, MA, USA). Disks laden with cells were immersed in 5% bovine serum albumin (BSA) in PBS and then probed with a primary mouse monoclonal antibody (paraformaldehyde fixed) for integrin β4 [critical early marker for HD assembly [45]) NB10065599; Novus Biologicals, Littleton, CO, USA; 1:500] or primary rabbit polyclonal antibody (methanol fixed) for collagen XVII [(important late marker for HD assembly [46]) ab28440; Abcam, Cambridge, UK; 1:500] for 1 h at room temperature. Samples were immersed in an antimouse (A-11005; Invitrogen; 1:500) or antirabbit secondary (ab97037; abcam; 1:500) after extensive washing for 3 h. Samples were counterstained with DAPI. Total fluorescent intensity in each field was quantified after accounting for secondary-only controls at constant microscope settings across all samples (n = 5, 3 FOVs each). Micrographs (×10) Micrographs were obtained with an upright fluorescent microscope (DM 6B, Leica, Wetzlar, Germany) and analyzed in ImageJ (NIH, 2.0.0).

Statistical Analysis
Arithmetic mean values with one standard deviation on the mean are reported. Differences in mean between groups were assessed with a one-way analysis of variance (ANOVA)

Physical and Chemical Characterization of Peptide Coatings
Water contact angle measurements were performed (Figure 2A,C) after each iterative synthesis step for synthesizing surfaces immobilized with 2K-LamLG3 and KN3-LamLG3 peptides using silanization (S-LamLG3) and click chemistry (D-LamLG3). Marked differences were noted in the WCAs at equilibrium between S-LamLG3 (ca. 40°) and D-LamLG3 (ca. 60°) likely due to the differences in underlying chemistry (CPTES vs. CPTES and DIBO, respectively; shown in Figure 2B) or peptide conformation differences. The differences in peptide ability to restructure on the surface when in contact with water are also highlighted by the higher dynamic response (larger change in contact angle over time) for the D-LamLG3 surfaces compared to S-LamLG3 surfaces. In addition, a longer period to reach equilibrium in D-LamLG3 compared to S-LamLG3 was noted. The approximately 22 nmol per disk detected in both groups is similar to previous reports by other groups [47,48]. Despite these difference in WCAs, no differences were seen in the amount of peptide on the surfaces via a BCA assay ( Figure 2D). XPS results ( Figure 3A) further confirmed the formation of a LamLG3 coating through detection of Nitrogen (N 1s peak) from the amino acid backbone of the peptide on S-LamLG3 and D-LamLG3 surfaces. Silicon (Si 2p) was also present, likely from the underlying glass substrate [49].

Durability of Peptide Coatings
In order to determine the simulated oral cavity durability of our immobilized peptides, we incubated our coatings in simulated saliva for up to 73 days and then periodically performed XPS [ Figure 3B (D-LamLG3) and Figure 3C (S-LamLG3)]. In both cases, the N 1s/C 1s signal significantly decreased following 28 days of incubation compared to day 0. However, this decrease in N 1s/C 1s signal was less pronounced in D-LamLG3 compared to S-LamLG3. No statistically significant differences were seen when comparing N 1s/C 1s between S-LamLG3 and D-LamLG3 at each timepoint ( Figure S3). The underlying, more hydrophobic DIBO molecules than the CPTES molecules, and its associated hindrance of interactions with water molecules, are a plausible cause for the higher resistance to coating degradation of the D-LamLG3 surfaces. The CPTES molecules in S-LamLG3 surfaces are susceptible for hydrolysis and, thus, detachment from the surface [50]. Despite this, a notable peptide-associated signal remained after 73 days in both peptide-coated surfaces.

Keratinocytes Reponse on Peptide Coatings with Different Peptide Immobilization Chemistry
Oral keratinocytes were cultured on our coatings to determine the differences between S-LamLG3 and D-LamLG3 in promoting keratinocyte proliferation ( Figure 4) and hemidesmosome formation ( Figure 5). After one day of culture, proliferation was increased in terms of both increased numbers of cells ( Figure 4A) and metabolic activity ( Figure 4B) for both S-LamLG3 and D-LamLG3 compared to controls (pGlass, pGlass-sill, and pGlass-DIBO). No differences were seen in cytotoxicity (LDH release; Figure 4C). However, proliferation was nearly significant (p = 0.06; number of cells) or significantly (metabolic activity) increased on S-LamLG3 compared to D-LamLG3, suggesting that differences in immobilization methods affect resulting cellular behavior. An interesting observation was the marked reduction in keratinocyte proliferation on DIBO control surfaces compared to all other groups, which has not been previously reported.
Our targeted biological activity, hemidesmosome formation on dental implant surfaces (summarized in Figure 1), was finally evaluated. Semi-quantitative immunofluorescence was performed for both collagen XVII ( Figure 5A,B; representative micrographs in Figure S4) and integrin β4 ( Figure 5C,D; representative micrographs in Figure S5). Similar to proliferation, HD formation was significantly increased on both S-LamLG3 and D-LamLG3 compared to all controls. However, HD formation was significantly higher on S-LamLG3 compared to D-LamLG3, further emphasizing differences in immobilization methods affect resulting cellular behavior.

Discussion
Our specific aim was to compare the physiochemical properties and biological responses of LamLG3 peptide immobilized using silanization and click chemistry for improving dental implant outcomes through the upregulation of hemidesmosome formation in the soft tissue surrounding dental implants. The broader goal of this work was to use LamLG3 as a model peptide to study the effects of immobilization chemistry on cell adhesion peptide activity to engineer more efficacious coatings for biomedical devices. Our results show that the non-chemoselective route (silanization) resulted in increased keratinocyte proliferation and hemidesmosome formation compared to the chemoselective route (click chemistry).
Chemoselectivity of peptide immobilization for biological responses is an oft-cited design criteron to improve effectiveness for the desired biological activity with the selected peptides [51]. However, rationalization under the specific experimental conditions studied is rarely offered. Here, we show that this design criterion does not hold true under all conditions; indeed, here, the non-chemoselective route for immobilizing the LamLG3 peptides resulted in better biological outcomes (proliferation and HD formation). In some other systems, for example sensors [52] and antimicrobial surfaces [53], chemoselective immobilization results in better peptide activity. The ca. 20° difference in WCA between S-LamLG3 and D-LamLG3 and the longer time period for D-LamLG3 to reach equilibrium suggests that either underlying surfaces chemistry differences (presence of DIBO or not) and/or differences in number of residues where the peptide may be anchored (three for S-LamLG3 vs. one for D-LamLG3) may result in differences in peptide conformation, flexibility/mobility and/or orientation and subsequent cellular engagement. For example, recent work [54] has shown differences in osteogenic differentiation outcomes mediated by the identical osteogenic peptide immobilized through either amines or carboxyls on carbon nanotube surfaces. One potential cause for the differences we observed here is that KN3-LamLG3 peptides only provide one reactive site with the DIBO-modified surfaces, whereas 2K-LamLG3 peptides provide up to four (three free amines plus N-terminus) reactive sites with the CPTESmodified surfaces. As shown by dynamic WCAs, this can produce a more rigid molecular coating in the case of S-LamLG3 surfaces that might have facilitated appropriate cellular interactions. In spite of the speculative nature of this discussion, our results and results by others support that chemoselective peptide immobilization, in some but not all circumstances, can result in reduced peptide activity.
One design parameter for chemically immobilizing peptides is the spacer (or linker) design between the bioactive domain and the chemical domains necessary for anchorage. An ideal spacer maintains an independent structure and does not affect function of adjacent peptide domains; spacer type, length, and flexibility are generally tunable to achieve this goal [55]. While PEG spacers are very well studied for peptide immobilization, they require additional synthesis steps compared to amino acid-based spacers [10,56]. Previous research compared a classic amino acid spacer (GGG) and a novel spacer (GSGGG) with a "backbone bend" to separate an antimicrobial peptide domain from the surface. This novel GSGGG spacer showed enhanced antimicrobial activity compared to the classic spacer we used here [57,58]. Other work has compared a rigid spacer [(EAAAK) 4 ] against a flexible spacer [(GGGGS) 4 ]; the flexible spacer design showed more effective eukaryotic cell signaling but less effective antimicrobial activity [36,59]. Others have shown that flexibility per se is not the requisite design criteria but rather matching hydrophobicity/philicity of surrounding residues leads to maximal eukaryotic response [60]. While such general design principles are useful, fully optimized spacer design is likely system dependent.
Peptide orientation is another commonly cited design parameter when chemically immobilizing peptides. This is commonly modulated through changes in different chain positions, such as C-terminal, N-terminal and/or N-side-chain [10]. Our own work with the antimicrobial peptide GL13K has shown that orientation does not affect antimicrobial activity [61]. GL13K is antimicrobial both when immobilizedusingsilanizationandwhenrecombinantlyincorporatedintotheterminusofanelastin -like recombinamer [62]. On the other hand, LL37-another well-studied antimicrobial peptide for biomedical coatings-is sensitive to orientation [63]. Other work has shown that when an alpha-helical antimicrobial peptide is "standing" (helix z-axis perpendicular to surfaces), it interacts with bacterial cells faster than identical peptides "lying down" (helix zaxis parallel to surface) [64]. Similar orientation effects are well known for enzyme and the vastly explored RGD peptide immobilization [65,66]. Other factors such as the specific residue anchoring point (and number of them) affect activity as well, including peptide density [67,68]. Previous work has shown that peptide density, which we did not control here, can affect cell adhesion, differentiation, and focal adhesion formation [48,69,70]. Whether peptide density affects HD formation is unexplored to date. A number of strategies not discussed such as electric fields or chemical vapor deposition may also be viable for controlling peptide orientation [71][72][73]. On the whole, judicious selection of orientation is, again, likely system dependent. In particular for our D-LamLG3 system, future experiments are necessary to directly compare biological activity resulting from N -vs. C-terminus addition of the reactive azide.
An unexpected result of this work was the significantly reduced keratinocyte proliferation on pGlass-DIBO compared to all other groups; this result was reproducible under a number of different cell seeding schemes and we have seen this same trend with other cell types (unpublished observation). Others have reported the cytocompatibility of DIBO [74][75][76] under various conditions with various cells. Further work is necessary to discern this mechanism.
Overall, our results show that biological effects from cell adhesion peptides-LamLG3 in this case for dental implants-are dependent on the immobilization chemistry used. Additionally, the most effective immobilization chemistry for optimal peptide activity is heavily system dependent. Future work will include longer culture periods as HDs continue to form and mature and monitoring peptide degradation in human-isolated saliva with active enzymes. Despite the importance of early cell-peptide interactions dictating later cell fate, hemidesmosome formation does continue after one day of culture and later timepoints may provide additional data for better design of peptide coatings. Human saliva contains at least 1515 unique proteins [77], including enzymes, that made degrade immobilized peptides and render them less active or completely inactive. A more fundamental understanding of the many factors that go into cell adhesion peptide immobilization may result in faster translation of such proposed therapies through the reduction in variability and a better understanding of the entire system to enable industrial scale manufacturing.

Conclusions
We immobilized a cell adhesion peptide, LamLG3, on model surfaces to upregulate hemidesmosome formation by oral keratinocytes with the ultimate goal of reducing dental implant peri-implantitis. The non-chemoselective immobilization route, silanization, produced LamLG3 coatings that induced significantly higher proliferation and hemidesmosome formation in oral keratinocytes compared to LamLG3 coatings obtained using chemoselective copper-free click chemistry. Our results emphasize that the chemical route to immobilize peptides on biomedical surfaces has significant effects on cell adhesion and that chemoselectivity of the immobilization route is not always beneficial for enhancing the biological effects of the immobilized peptides. A deeper understanding of the effects of the selected chemical route of immobilization on biological activity, which has to be tested for the specific combination of peptide and targeted cellular response, is needed to develop more effective biomedical device peptide coatings

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. Schematic of hemidesmosome formation by keratinocytes at the junctional epithelium of the oral mucosa (gum) to attach to the tooth.  Dynamic water contact angles for LamLG3-coated and non-coated, control surfaces (A); schematic illustrating immobilization site(s) and chemistries of D-LamLG3 (B, left) and S-LamLG3 (B, right); water contact angles at equilibrium (C); and amount of peptide on D-LamLG3 and S-LamLG3 surfaces using a BCA assay (D). Differences in mean between groups were assessed with a one-way analysis of variance (ANOVA) table followed by a Tukey's HSD (honest significant difference) post hoc test. A p value of <0.05 was considered statistically significant. Dissimilar letters denote statistically significant differences between groups. X-ray photoelectron spectroscopy spectra of peptide-coated (D-LamLG3, S-Lam-LG3) and non-coated, control surfaces. Full survey spectra of each coated and non-coated, control surface (A) and the ratio of N 1s X-ray photoelectron spectroscopy counts to C 1s counts (N 1s:C 1s) for D-LamLG3 (B) and S-LamLG3 (C) following up to 73 days in artificial saliva (37 °C). Differences in mean between groups were assessed with a one-way analysis of variance (ANOVA) table followed by a Tukey's HSD (honest significant difference) post hoc test. A p value < 0.05 was considered statistically significant. The asterisk (*) denotes when the N 1s/C 1s signal was statistically significantly reduced compared to day 0. Oral keratinocyte hemidesmosome formation [total collagen XVII intensity (A), collagen XVII intensity normalized to number of cells (B), total integrin β4 intensity (C), and integrin β4 intensity normalized to number of cells (D)] of coated and non-coated, control surfaces at 1 day. Differences in mean between groups were assessed with a one-way analysis of variance (ANOVA) table followed by a Tukey's HSD (honest significant difference) post hoc test. A p value of <0.05 was considered statistically significant. Dissimilar letters denote statistically significant differences between groups. Fischer et al. Page 18