A possible use of melatonin in the dental field : protein adsorption and in vitro cell response 1 on coated titanium 2

30 Melatonin (MLT) is widely known for regulating the circadian cycles and has been studied for its 31 role in bone regeneration and inflammation. Its application as a coating for dental implants can 32 condition the local microenvironment, affecting protein deposition on its surface and the cellular 33 and tissue response. Using sol-gel coatings as a release vehicle for MLT, the aim of this work was 34 to assess the potential of this molecule in improving the osseointegration and inflammatory 35 responses of a titanium substrate. The materials obtained were physicochemically characterized 36 (scanning electron microscopy, contact angle, roughness, Fourier-transform infrared 37 spectroscopy, nuclear magnetic resonance, Si release, MLT liberation, and degradation) and 38 studied in vitro with MC3T3-E1 osteoblastic cells and RAW264.7 macrophage cells. Although 39 MLT application led to an increased gene expression of RUNX2 and BMP2 in 10MTL, it did not 40 improve ALP activity. On the other hand, MLT-enriched sol-gel materials presented potential 41 effects in the adsorption of proteins related to inflammation, coagulation and angiogenesis 42 pathways depending on the dosage used. Using LC-MS/MS, protein adsorption patterns were 43 studied after incubation with human serum. Proteins related to the complement systems (CO7, 44 IC1, CO5, CO8A, and CO9) were less adsorbed in materials with MLT; on the other hand, 45 proteins with functions in the coagulation and angiogenesis pathways, such as A2GL and PLMN, 46 showed a significant adsorption pattern. 47


Introduction 61
Dental implantations have become a standard procedure in oral rehabilitation, representing a 62 reliable treatment with many advantages. However, implant failure still occurs, particularly in 63 patients with poor osseointegration capability (e.g. patients with osteoporosis), prompting the 64 need for bioactive surfaces that accelerate this process [1]. 65 Titanium (Ti) and its alloys are commonly used in dental implants due to their high degree of 66 biocompatibility. However, these materials have the limitation of being relatively bioinert and 67 various methodologies are being studied to confer them bioactive properties. increment of bone regeneration [9]. Additionally, MLT has been studied for its anti-inflammatory 82 potential leading to the downregulation of TNFα, IL-1β, IL-6 [10,11], and iNOS [11,12], can 83 either stimulate or inhibit angiogenesis [13,14], and it has an antioxidant potential [15]. 84 Considering the effects of this molecule in bone and inflammatory responses, MLT has become 85 a particularly attractive molecule to use in implants. 86 Upon implantation, blood/implant interactions lead to immediate protein adsorption onto the 87 implant surface and consequently developing a provisional matrix on and around the biomaterial. 88 The type, level, and surface conformation of the adsorbed proteins will determine the biological 89 response and the ultimate implant outcome [16]. This adsorption is dependent on the surface 90 properties of the material, such as wettability, roughness, and charge [17,18]. Thus, these 91 parameters can ultimately have a determining role not only in the initial immune responses but 92 also in other processes, such as coagulation, fibrinolysis, and the earlier stages of osteogenesis 93 [19]. 94 Biomaterial cytotoxicity was assessed following the ISO 10993-5:2009 (Annex C) norm. 164 MC3T3-E1 cells (1x10 5 cells cm -2 ) were seeded on 96-well NUNC plates (Thermo Fisher  165 Scientific, Waltham, MA, USA) for 24 h. The materials were also incubated for 24 h in 48-well 166 NUNC plates (Thermo Fisher Scientific) in DMEM with 1% of penicillin/streptomycin and 10% 167 FBS. Then, the cell culture medium was replaced with the medium exposed to the materials 168 followed by an incubation of 24 h. To measure cell viability, the CellTiter 96® Proliferation 169 Assay (MTS) (Promega, Madison, WI) was used according to manufacturer's guidelines. As a 170 negative control, wells with only cells were used. As a positive control, cells were incubated in 171 latex, a compound well known for being cytotoxic. The material was considered cytotoxic when 172 presented cell viability below 70%. 173

Cell proliferation 174
To measure the effects of the biomaterials in cell proliferation, the alamarBlue™ cell viability 175 reagent (Invitrogen, Thermo Fisher Scientific) was used. MC3T3-E1 cells were cultured in 24-176 well NUNC plates (Thermo Fisher Scientific) at a density of 3.5x10 4 cells cm -2 . After culturing 177 for 1, 3, and 7 days, cell proliferation was evaluated following the manufacturer's protocol. 178 Additionally, an essay without cells was carried out to verify that the tested materials did not 179 affect the alamarBlue™ cell viability reagent. 180

Alkaline phosphatase activity assay 181
To evaluate the effects of the materials in the mineralization capability of osteoblastic cells, the 182 conversion of p-nitrophenylphosphate (p-NPP) to p-nitrophenol was used to assess the alkaline 183 phosphatase (ALP) activity. MC3T3 cells were seeded onto the distinct surfaces in 24-well 184 NUNC plates (Thermo Fisher Scientific) at a density of 3.5x10 4 cells cm -2 . After culturing for 14 185 and 21 days, cells were rinsed twice with Dulbecco's phosphate-buffered saline (DPBS; Thermo 186 Fisher Scientific), immersed in lysis buffer (0.2% Triton X-100, 10 mM Tris-HCl, pH 7.2) and 187 incubated at 4°C for 10 minutes. Following centrifugation (7 min, 14000 rpm, 4°C), 100 µL of p-188 NPP (1mg mL -1 ) in substrate buffer (50 mM glycine, 1 mM MgCl 2 , pH 10.5) was added to 100 189 µL of the supernatant. After 2 h of incubation in the dark (37°C, 5% CO 2 ), the absorbance at 405 190 nm was measured using a microplate reader. Alkaline phosphatase activity was calculated using 191 a p-nitrophenol in 0.02 mM sodium hydroxide standard curve. A Pierce BCA assay kit (Thermo 192 Fisher Scientific) was used to calculate total protein content in the sample and to normalize ALP 193 levels. The experiment was carried out in triplicate. 194

RNA extraction and cDNA synthesis 195
To evaluate the effects on the gene expression of osteogenic and inflammatory targets,  E1 cells were seeded on the discs in 48-well NUNC plates (Thermo Fisher Scientific) at a density 197 of 3.5x10 4 cells cm -2 for 7 and 14 days. RAW264.7 were seeded at a density of 30x10 4 cells cm -2 198 for 1 day and 1.5x10 4 cells cm -2 for 3 days. In each plate, wells without any material were used 199 as control of culture conditions. Total RNA was extracted using TRIzol (1M guanidine 200 thiocyanate, 1M ammonium thiocyanate, 3M sodium acetate, 5% glycerol, 38% aquaphenol). 201 Briefly, 300 µL of TRIzol were added to the samples, and then they were incubated at room 202 temperature for 5 min. The experiment was carried out in quadruplicate. Real-Time PCR System (Applied Biosystems®, Thermo Fisher Scientific) at 95°C for 30s, 228 followed by 40 cycles of 95°C for 5s, 60°C for 34s, 95°C for 15s and 60°C for 60s. The data were 229 obtained using the StepOne Plus™ Software 2.3 (Applied Biosystems®, Thermo Fisher 230 Scientific). Fold changes were calculated using the 2-ΔΔ Ct method and the data was normalized 231 in relation to the blank wells (without any material). Six technical replicates for each sample were 232 measured.

Cytokine quantification by ELISA 236
To evaluated the influence of the materials in tumor necrosis factor (TNF)-α and interleukin 4 237 (IL-4) production, RAW264.7 cells were seeded in 48-well NUNC plates (Thermo Fisher 238 Scientific) a density of 30x10 4 cells cm -2 for 1 day and 1.5x10 4 cells cm -2 for 3 days. Then, the cell 239 culture media was collected and frozen until further analysis. The concentration of these cytokines 240 was determined using an ELISA (Invitrogen, Thermo Fisher Scientific) kit and according to the 241 manufacturer's instructions. 242

Adsorbed protein layer 243
For obtaining the proteins adsorbed by the material surface, discs doped with MLT were incubated 244 for 3 h (37 °C, 5% CO 2 ) in 24-well NUNC plates (Thermo Fisher Scientific) with 1 mL of human 245 blood serum from male AB plasma (Sigma-Aldrich). After incubation, the serum was removed 246 and the discs were washed five times with ddH 2 O and once with 100 mM NaCl, 50 mM Tris-247 HCl, pH 7.0 to eliminate non-adsorbed proteins. The materials were washed once with an elution (0.5 M triethylammonium bicarbonate buffer (TEAB), 4% of sodium dodecyl sulfate (SDS), 100 249 mM of dithiothreitol (DTT)) to obtain the adsorbed protein layer. The analysis was made in four 250 independent replicates and each replicate was a pool of four discs. A Pierce BCA assay kit 251 (Thermo Fisher Scientific) was used to calculate total protein content in the serum. were considered statistically significant at p ≤ 0.05 (*), p ≤ 0.01 (**), and p ≤ 0.001 (***). 269

Physicochemical characterization 271
The sol-gel materials with MLT were successfully synthesized and well-adhering coatings were 272 obtained as it can be observed in SEM micrographs (Figure 1). In these images, it can be observed 273 that the sol-gel material has completely covered the Ti surface. Furthermore, the coatings seem 274 to have smoothed the initial morphology of the SAE treatment, accumulating more sol-gel in the 275 irregularities caused by the previous sandblasting. The band is composed of two peaks corresponding to vibrations of asymmetrical and symmetrical 298 tension of the bond C-H. The bond associated with the Si-CH 3 group appears around 1275 cm -1 299 [25]. These methyl-associated signals show that the integrity of organic species has been 300 maintained after processing. All identified signals are maintained and display similar intensity 301 when the MLT is incorporated into the sol-gel. However, the materials with MLT show bands 302 between 1500-1600 cm -1 , which corresponds to the CO group present in this molecule [26]. In 303 addition, the spectra of these materials show bands at 1610 cm -1 and 1555 cm -1 , which correspond 304 to N-H and C-N bounds present in MLT, correspondingly [23]. The intensity of these bands is 305 slightly more intense as the amount of melatonin increases. 306 Figure 2b represents 29 Si solid NMR spectra of 70M30T and 70M30T supplemented with MLT. 307 These spectra show T n signals from MTMOS and Q n signals from TEOS. The MTMOS spectra 308 show T 2 and T 3 signals with higher intensity of T 3 . Additionally, the spectra show Q 3 and Q 4 from 309 TEOS, with a signal more intense in Q 3 . It seems that the addition of MLT to the sol-gel network 310 did not affect the final crosslinking degree of structure.

Cytotoxicity, cell proliferation, and ALP activity 356
Neither of the materials in the study was cytotoxic (data not shown). Cell proliferation and ALP 357 activity assays did not show significant differences between the 70M30T with or without 358 melatonin (Figure 4)

Cytokine quantification by ELISA 384
To evaluate the effect of the materials with MLT on the inflammatory response, the 385 secretion of anti (IL-4) and pro-inflammatory (TNF-α) cytokines by RAW264.7 386 macrophage was quantified at 1 day and 3 days. The secretion of IL-4 did not show 387 differences at any of the times measured in any of the materials tested (Figure 6a). In 388 the case of TNF-α, the profile was similar at 1 day for all materials (Figure 6b). After 3 days of 389 culture, there is a general increase in the production of this cytokine; however, is significantly 390 lower in 1% MLT when compared to the 70M30T coating.

Proteomic analysis 396
The eluted proteins were analyzed by LC-MS/MS, followed by identification with Progenesis 397 QI software and DAVID system. Comparing MLT-enriched and the base sol-gel material, 26 398 proteins were differentially absorbed in the materials with MLT (Supplementary Table 1). The 399 formulation with 10MLT shows the higher amount of differently absorbed proteins, with 16 400 proteins being less adsorbed onto its surface and five showing more affinity. Among the proteins 401 with decreased adsorption, five are related to the complement system (CO7, IC1, CO5, CO8A, 402 and CO9). On the other hand, these surfaces lead to higher adsorption with CXCL7, which plays 403 a crucial role in neutrophil recruitment. Also related to immunological responses, the surface 404 1MLT and 5MLT showed a higher affinity with IGHA2, while 7.5MLT differentially absorbed 405 CO5, IC1, CO8A, and CXCL7. The glycoproteins VTCN and SEPP1 were significantly less 406 adsorbed in the material with 10MLT, while HEMO show higher affinity with the materials with 407 1MLT, 5MLT, and 10MLT. VTNC is known to inhibit/regulate the complement system activation. Depending on the concentration of MLT, the materials adsorbed fewer apolipoproteins 409 (APOA-I, APOF, APOL1, and APOC4) and PON1. These proteins are related to the metabolism 410 of high-density lipids. Regarding the coagulation process, HRG, HBB, PLMN, and KLKB1 were 411 differentially adsorbed: HRG was more adsorbed in 1MLT, while KLKB1 was more adsorbed in 412 5MLT. In 7.5MLT and 10MLT proteins related to this process presented less affinity with these 413 materials. Additionally, all materials except 7.5MLT showed a differential affinity with A2GL, a 414 protein-related with the angiogenesis processes. The materials 1MLT and 5MLT adsorbed more 415 of this protein, while 10MLT adsorbed less. ITIH2, ITIH4, and ITHI1, proteins from the inter-416 α-trypsin inhibitor family related to the hyaluronan metabolic process, were less adsorbed in the 417 materials with 10MLT. ATPA, a mitochondrial membrane complex that produces ATP from 418 ADP, was significantly more adsorbed in 1MLT. Table 1  In addition, it was found differential adsorption of complement system proteins. In the materials 493 with 7.5MLT and 10MLT, we could observe a decrease in the adsorption of complement C5 494 (CO5), complement component C8 alpha chain (CO8A), complement component (CO7) and 495 component complement (CO9). The activation of C5 initiates an assembly with late-phase 496 complement components, such as C6, C7, C8, and C9, leading to the formation of C5-C9 497 complex, a multimolecular structure that leads to the formation of the lytic complex that will be 498 responsible for the target cell lysis [34]. This is in agreement with the analysis PANTHER, which 499 shows that the proteins less adsorbed by these materials have functions associated with B cell 500 activation. The distinct complement pathways originate C3 and C4 fragments, which bind to 501 complement receptors CD21 and CD35, whose co-expression is limited to B cells and leads to 502 the enhancement of the activity of these cells [35,36]. On the other hand, vitronectin (VTNC) was 503 less adsorbed in the materials with 10MLT. This protein has been described as an inhibitor of 504 complement system action in bodily fluids [37]. Thus, the lower adsorption of complement 505 proteins associated with the lower adsorption of VTNC can explain how the release of TNF-α and 506 IL-4 cytokines by macrophage in contact with the 10MLT showed no statistical differences with 507 respect to the base coating. Although, the anti-inflammatory potential of MLT is well described 508 [11,38,39], its application in biomaterials can be dependent on the amount of hormone released 509 by the material over time, and further studies are needed. 510 Coagulation and angiogenesis are key processes in bone regeneration. Proteomic analysis showed 511 that MLT enriched materials differently adsorbed proteins related to both of these processes. In 512 this sense, A2GL, a protein implicated in angiogenesis [40], was found to be more adsorbed onto 513 the coatings 1MLT and 5MLT, but then, reduced its affinity with respect to the base material 514 when 10% of MLT was incorporated. In vitro, MLT was reported to inhibit angiogenesis in 515 cancer cells [41,42]. On the other hand, Ramírez-Fernandez et al. [14] reported that MLT 516 promoted this process in rabbit tibiae following implantation of melatonin implants. 517 Regarding the coagulation process, HRG, which modulates various components in the 518 coagulation cascade, such as heparin, increased its affinity for 1MLT. Similarly, KLKB1 was 519 significantly more adsorbed onto the material 5MLT. This protein activates the coagulation 520 cascade through the intrinsic pathway [43]. However, both KLKB1 and HRG reduced the affinity 521 by the material when 10% of MLT was added. 522 Fibrinolysis is a highly regulated enzymatic process of clot removal tightly related to blood 523 coagulation [44]. PLMN, a protein found less adsorbed onto 10MLT, has a role in tissue 524 regeneration by dissolving preformed fibrin clots and extracellular matrix components allowing 525 tissue remodeling [45]. These adsorption patterns are corroborated with PANTHER analysis, 526 which showed a general decrease in proteins with functions related to blood coagulation and 527 plasminogen activation. 528 MLT has a complex biological role and its potential effect on important pathways, such as 529 inflammation, coagulation, and angiogenesis, in the early stages of tissue regeneration, can 530 determine how these processes will be carried out around an implant. However, its specific 531 mechanism of action, timings, and doses needed to produce significant cellular effects still need 532 further studies. 533

Conclusions 535
In this article, we developed new coatings with MLT to be applied in titanium dental implants 536 using a hybrid sol-gel network as a release vehicle. The addition of MLT changed the superficial 537 parameters of the coatings, with the coatings supplemented with the hormone showing a lower 538 hydrophilia when compared to the base material. These materials revealed to be not cytotoxic and 539 showed an increase BMP2 and RUNX2 gene expression in 10MLT. However, osteoblastic cells 540 did not show an improvement in the capacity of proliferation and mineralization (ALP activity) 541 in vitro when exposed to the coatings. The proteomic analysis of protein adsorption onto the 542 materials showed differences in the adsorption patterns in proteins associated with the 543 complement pathway when MLT added and in a dose-response manner. This behavior can explain 544 the liberation of TNF-α, which was significantly lower in the 1MLT composition. In addition, it 545 was found differences in adsorption of proteins related to coagulation and angiogenesis, which 546 points out a possible effect of MLT in the activation and development of these pathways. 547 6. Acknowledgments 548