Evaluation of stemness properties of cells derived from granulation tissue of peri‐implantitis lesions

Abstract Objectives Peri‐implantitis (PI) is an inflammatory disease associated with peri‐implant bone loss and impaired healing potential. There is limited evidence about the presence of mesenchymal stromal cells (MSCs) and their regenerative properties within the granulation tissue (GT) of infrabony peri‐implantitis defects. The aim of the present study was to characterize the cells derived from the GT of infrabony PI lesions (peri‐implantitis derived mesenchymal stromal cells—PIMSCs). Material and Methods PIMSC cultures were established from GT harvested from PI lesions with a pocket probing depth ≥6 mm, bleeding on probing/suppuration, and radiographic evidence of an infrabony component from four systemically healthy individuals. Cultures were analyzed for embryonic (SSEA4, NANOG, SOX2, OCT4A), mesenchymal (CD90, CD73, CD105, CD146, STRO1) and hematopoietic (CD34, CD45) stem cell markers using flow cytometry. PIMSC cultures were induced for neurogenic, angiogenic and osteogenic differentiation by respective media. Cultures were analyzed for morphological changes and mineralization potential (Alizarin Red S method). Gene expression of neurogenic (NEFL, NCAM1, TUBB3, ENO2), angiogenic (VEGFR1, VEGFR2, PECAM1) and osteogenic (ALPL, BGLAP, BMP2, RUNX2) markers was determined by quantitative RT‐PCR. Results PIMSC cultures demonstrated high expression of embryonic and mesenchymal stem cell markers with inter‐individual variability. After exposure to neurogenic, angiogenic and osteogenic conditions, PIMSCs showed pronounced tri‐lineage differentiation potential, as evidenced by their morphology and expression of respective markers. High mineralization potential was observed. Conclusions This study provides evidence that MSC‐like populations reside within the GT of PI lesions and exhibit a multilineage differentiation potential. Further studies are needed to specify the biological role of these cells in the healing processes of inflamed PI tissues and to provide indications for their potential use in regenerative therapies.


| INTRODUCTION
Peri-implantitis (PI) is a biofilm-associated inflammatory disease resulting in progressive loss of supporting bone (Berglundh et al., 2018). Up to date there is no predictable therapy for PI. Due to implant surface characteristics and limited access to the microbial habitat, non-surgical therapy is frequently inefficient in ceasing the progression of the disease. Therefore, surgical intervention is required more often and at an earlier stage in PI lesions, when compared to periodontitis lesions (Heitz-Mayfield & Lang, 2010).
Similar to periodontitis, PI lesions are dominated by plasma cells and lymphocytes, but with larger shares of polymorphonuclear leukocytes and macrophages (Berglundh et al., 2004(Berglundh et al., , 2011Bullon et al., 2004;Cornelini et al., 2001;Sanz et al., 1991). Besides, it was documented that the size of PI lesions is more than twice as large as that noted at periodontitis sites (Carcuac & Berglundh, 2014). Moreover, PI lesions are characterized by a higher density of vascular structures lateral to the infiltrated connective tissue (Carcuac & Berglundh, 2014). Recently, we observed that granulation tissue (GT) harvested from periodontal pockets exhibit a cellular infiltrate reflecting a chronic inflammatory state (Apatzidou et al., 2018). Due to the presence of mesenchymal stromal cells (MSCs) and immunophenotypic characteristics similar to those found in clinically healthy periodontal tissues, it can be concluded that periodontal GT, albeit inflamed, retain regenerative healing potential (Apatzidou et al., 2018). For both, infrabony periodontitis and PI lesions, we could show that the preservation of GT during regenerative surgery may improve the treatment outcome from a clinical and radiographic perspective. Thus, application of the granulation tissue preservation technique (GTPT) resulted in significant clinical attachment gain and radiographic bone fill (Figure 1; Günay et al., 2013Günay et al., , 2019. The presence of mesenchymal stromal cells in the GT derived from infrabony PI defects (peri-implantitis derived mesenchymal stromal cells = PIMSC) has not been investigated so far. Therefore, the aim of the present study was to investigate whether cell cultures established from the GT of PI lesions show stemness properties, including expression of mesenchymal and embryonic stem cell markers and multipotent differentiation potential. Donos et al. investigated the gene expression profile following placement of dental implants in humans Donos, Retzepi, et al., 2011). They demonstrated that the biological processes of osteogenesis, angiogenesis and neurogenesis play a fundamental role in osseointegration. Although scientific evidence is lacking so far, it seems likely that the same biological processes are required for the regeneration of infrabony PI defects. Therefore, the current study selected the neurogenic, angiogenic and osteogenic pathways to show multipotency.

| Establishment of PIMSC cultures
The human PIMSC cultures used in the present study were established from PI lesions of four male donors aged 52-68 years. All donors were systemically healthy, not taking any medication and not consuming any alcohol. Two of the donors were smokers with a daily consumption of more than 15 cigarettes; the other two donors were non-smokers. All patients received non-surgical therapy to reduce local signs of inflammation and to facilitate the surgical intervention.
Before surgery, a residual PI lesion with a pocket probing depth ≥6 mm, bleeding on probing, and a radiographically evident infrabony defect had to be present . After mobilization of the muco-periosteal flap, the inflammatory GT was collected from the defect using curettes and scalers. In particular the GT from the bottom of the PI lesion was harvested. Cell cultures were established using the enzymatic dissociation method, as described previously (Bakopoulou et al., 2010). Briefly, the tissue samples were cut into small pieces and digested in alpha-minimal essential medium (alpha-MEM, Gibco, Grand Island, NY) supplemented with 3 mg/mL collagenase type I (Gibco/Life Technologies, Paisley, Scotland) and 4 mg/mL dispase II (Sigma-Aldrich, Steinheim, Germany) for 1 h at 37 C. Filtration through a strainer with a pore size of 70 μm (EASYstrainer, Greiner bio-one, Frickenhausen, Germany) eliminated tissue debris.
The resulting single-cell suspension was seeded into cell culture flasks containing complete culture medium (CCM). This consisted of alpha-MEM supplemented with 15% fetal bovine serum (FBS, Biochrom, Berlin, Germany), 100 U/mL penicillin (Biochrom), 100 μg/mL streptomycin (Biochrom), 2.5 μg/mL amphotericin B (Capricorn Scientific, Ebsdorfergrund, Germany), and 100 μM L-ascorbic acid phosphate (Sigma-Aldrich). The cells were incubated in humidified atmosphere at 37 C in 5% CO 2 and the first medium change was carried out after 24 h to remove the abundance of erythrocytes. After reaching 80%-90% confluency, cells were collected by treatment with a 0.25% tryp-

| Neurogenic differentiation
For the induction of neurogenic differentiation, PIMSCs were seeded into six-well plates coated with 0.1% gelatin (Sigma-Aldrich) at 1 × 10 5 cells / well and grown in neurogenic differentiation medium (NDM). This consisted of neurobasal A medium (Gibco) supplemented with B27 supplement (2% v/v, Gibco), 2 mM L-glutamine (Gibco), 20 ng/mL recombinant human epidermal growth factor (rh-EGF, Biochrom), 40 ng/mL recombinant human basic fibroblast growth factor (rh-bFGF, Biochrom), 100 U/mL penicillin, 100 μg/mL streptomycin, F I G U R E 1 Non-surgical and surgical treatment of peri-implantitis. (a) Purulent peri-implantitis lesion at implant regio 12. (b) Clinical view after removal of the cemented crown. Note the increased probing pocket depth at the buccal aspect of the implant. (c) Clinical situation 2 weeks after non-surgical treatment using an air-polishing device. Note the resolution of inflammation signs. (d) After mobilization of the mucoperiosteal flap, a circumferential infrabony defect was detected. The surgical treatment included decontamination of the implant surface and application of enamel matrix derivatives. (e and f) Apart from a minor soft tissue dehiscence at the mesial papilla, an uneventful wound healing was observed 1 week after surgery. (g and h) Bland mucosal conditions 1 and 2 years after surgery without formation of mucosal recession. (i and j) A pronounced infrabony component was present on the baseline X-ray. A significant bone fill was achieved 2 years after surgery and 2.5 μg/mL amphotericin B. Cells were cultured for 5 weeks and the NDM was changed every 2-3 days. Cultures exposed to CCM were used as negative control. Neurogenic differentiation was assessed by observation under an inverted microscope for the detection of morphological changes towards a neuron-like phenotype and by real-time reverse transcriptase polymerase chain reaction (qRT-PCR) for the expression of neural markers, including neurofilament light polypeptide (NEFL), neural cell adhesion molecule 1 (NCAM1), tubulin beta 3 class III (TUBB3) and enolase 2 (ENO2).

| Osteogenic differentiation
For the induction of osteogenic differentiation, PIMSCs were seeded into six-well plates at 1 × 10 5 cells / well and expanded in CCM until they reached confluency. Subsequently, cells were exposed to osteogenic differentiation medium (ODM) consisting of CCM supplemented with 5 mM β-glycerol phosphate (Sigma-Aldrich), 1.8 mM monopotassium phosphate (KH 2 PO 4 , Sigma-Aldrich), and 10 nM dexamethasone (Sigma-Aldrich). Cells were cultured for 4 weeks and the ODM was changed every 2-3 days. Osteogenic differentiation was analyzed using the Alizarin Red S (AR-S) method to identify the mineralized matrix. Briefly, the culture dishes were washed two times with PBS without Ca 2+ and Mg 2+ (Biochrom), and cells were fixed with 10% (w/v) neutral buffered formalin solution (Sigma-Aldrich) for 1 h at room temperature (RT). Afterwards, the culture dishes were washed two times with distilled water and staining was performed using 1% AR-S (pH 4.0, Sigma-Aldrich) for 20 min at RT. The culture dishes were washed four times with 2 mL distilled water to eliminate unspecific staining and mineralized deposits were visualized and photographed under an inverted microscope equipped with a digital camera (Olympus Optical Co., Ltd., Japan). Quantification of mineralized matrix was performed by the AR-S extraction method.
After aspiration of the distilled water, 1.5 mL cetylpyridinium chloride buffer (CPC, 10%, w/v, dissolved in 10 mM disodium monohydrogen phosphate, pH 7) was added to the culture dishes for 2 h at 37 C. Aliquots of 200 μL were transferred to a 96-well plate. A microplate spectrophotometer was used to measure the optical absorption at 550 nm (Spectra Max 250, MWG Biotech, Sunnyvale, CA). In addition, the qRT-PCR was applied to evaluate the expression of osteogenic markers, including runt-related transcription factor 2 (RUNX2), bone gamma-carboxyglutamate protein or osteocalcin (BGLAP), bone morphogenic protein 2 (BMP2) and alkaline phosphatase (ALPL).

| Quantitative real-time reverse transcription polymerase chain reaction
A two-step quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) was applied to analyze changes in gene expression during neurogenic, angiogenic and osteogenic differentiation. The entire RNA was isolated from the cell cultures using the RNeasy Plant Mini Kit (Qiagen, Hilden, Germany). The genomic DNA was eliminated through on-column digestion (RNase-free DNase Set, Qiagen). A microplate reader (Synergy H1, BioTek, Bad Friedrichshall, Germany) was used for the measurement of the RNA concentration.
The cDNA was synthesized using 1 μg of isolated RNA and the QuantiTect Reverse Transcription Kit (Qiagen). The QuantiTect SYBR Green PCR Kit, the QuantiTect Primer Assays (Table 1) and the Rotor-Gene Q cycler (all from Qiagen) were used for the amplification and real-time quantification of cDNA targets. The PCR reactions involved an initial activation of the HotStarTaq DNA polymerase (at 95 C for 5 min) and 40 cycles of denaturation (at 95 C for 5 s), annealing and extension (at 60 C for 10 s). The specificity of the reaction products was confirmed by a standard melting curve. LinRegPCR was applied to perform baseline correction, to determine the window-of-linearity, and to analyze the PCR efficiency per sample and per group (Ruijter et al., 2009). Actin beta (ACTB), beta-2-microglobulin (B2M), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 18S ribosomal RNA (RRN18S), succinate dehydrogenase flavoprotein subunit (SDHA2), and tyrosine 3-monooxygenase/ tryptophan 5-monooxygenase activation protein zeta (YWHAZ) were used as housekeeping genes. The two most stable housekeeping genes were selected by geNorm and used to normalize the adjusted qPCR data (Vandesompele et al., 2002). The delta delta CT method was applied to calculate fold changes in gene expression (Pfaffl, 2001).

| Statistical analysis
All assays were performed with cell cultures of all four donors. There was a significant inter-individual biological variability with regard to the qRT-PCR results. Therefore, the standardization method described by Willems et al. was applied (Willems et al., 2008). This method consisted of a logarithmic transformation, mean centering and T A B L E 1 QuantiTect Primer Assays (Qiagen) used for the qRT-PCR analyses

| Immunophenotypic characterization
The PIMSC cultures used in this study were highly positive (>98%

| Time-course gene expression of neurogenesis-related markers
The observed morphological changes of PIMSCs were further corroborated with gene expression of the neuron-specific markers NEFL, NCAM1, TUBB3 and ENO2, assessed by qRT-PCR ( Figure 5).

| Time-course gene expression of osteogenesis-related markers
The osteogenic differentiation was verified by the expression profiles of the osteogenesis-related markers ALPL, BGLAP, BMP2 and RUNX2 ( Figure 7). In cultures grown in ODM, the expression of ALPL was significantly increased at day 3, 7, 10, 14 (p < 0.01, respectively) and 21 (p < 0.05), the expression of BGLAP was significantly elevated at day 21 (p < 0.05), and the expression of BMP2 was significantly upregulated at day 7, 14 and 28 (p < 0.05, respectively). The expression of RUNX2 did not significantly change. Cultivation in CCM led to a significant upregulation of the ALPL (day 3, p < 0.05) and BMP2  Two of our four tissue donors were smokers. Studies assessing the influence of smoking and nicotine on the properties of MSCs suggest that smoking results in a reduced immunomodulatory capacity, an impaired osteogenic differentiation potential in vitro, and poorer bone regeneration capacity in vivo (Cruz et al., 2019;Sreekumar et al., 2018;Zhao et al., 2018). However, we could not find any difference in the experimental data that would indicate an influence of the smoking behavior of the donor.

| Mineralized tissue formation by PIMSC cultures
The definition of a cell population as MSCs involves the expression of specific epitopes. In the present study, we first performed an extensive immunophenotypic analysis with regard to several stem cell markers (Figures 3 and 4). These immunophenotypic profiles were overall similar in the analyzed PIMSC cultures. . Statistically significant differences to the reference value at baseline were identified by one-way ANOVA with Dunnett's multiple comparison test (*/ φ p < 0.05; **/ φφ p < 0.01; ***/ φφφ p < 0.001) comparative studies observed a higher expression of STRO1 in cells derived from inflamed periodontal sites (Li et al., 2014;Tomasello et al., 2017), others reported a higher expression in cells derived from healthy sites (Park et al., 2011;Tang et al., 2016). The percentage of STRO1 + cells ranged from 6.6% to 37.8% in these investigations. The latter is rather consistent with our results.
Overall, the data of the immunophenotypic analysis suggest that PIMSCs have properties of MSCs.
Our next aim was to investigate, if PIMSC cultures exhibit trilineage differentiation potential. Ivanowski et al. and Donos et al. investigated the biological processes associated with osseointegration of titanium dental implants in humans Donos, Retzepi, et al., 2011;Ivanovski et al., 2011). They reported that genes related to inflammation, angiogenesis, neurogenesis and skeletogenesis were temporally upregulated during the early stages of osseointegration. If neurogenesis, angiogenesis and osteogenesis are required for the osseointegration after implant placement, it appears conclusive that these processes are also required for the regeneration of PI defects. Therefore, our study addressed these three different pathways of differentiation: neurogenesis, angiogenesis and osteogenesis. From a morphological point of view, we observed that PIMSC cultures display heterogeneity. This is not surprising because we used the enzymatic dissociation method to establish PIMSC cultures. This resulted in the release of different cell types with varying size and morphology (Gronthos et al., 2000(Gronthos et al., , 2002. After induction of PIMSC cultures with respective differentiation media, the morphology of AR-S concentrations were significantly different at day 28 (one-way ANOVA with Bonferroni's multiple comparison test; **p < 0.01) deposits of calcium similar to bone nodules were observed in the cultures following staining with AR-S, but no sign of mineralization was detected in the control cultures grown in CCM (Figure 2d). In cultures exposed to ODM, the mineralized matrix started to form next to the cellular aggregates and gradually increased until it covered almost 100% of the monolayer at the end of the observation period.
Transcription profiling has broadened the knowledge about the biological processes and signaling pathways associated with the early events of wound healing Donos, Retzepi, et al., 2011). Cells exposed to NDM showed a continuously increasing expression of NEFL, NCAM1 and ENO2 throughout the entire differentiation experiments ( Figure 5). The NEFL gene encodes a type IV intermediate filament playing an important role in the intracellular transport of neurotransmitters to axons and dendrites (Leermakers & Zhulina, 2010). NCAM1 encodes a cell adhesion protein, which is involved in cell-to-cell and cell-to-matrix interactions during development and differentiation (NCAM1 Neural Cell Adhesion Molecule 1, 2020). ENO2 is known to be a useful index of neural maturation and a highly specific marker for neurons and peripheral neuroendocrine cells (Isgrò et al., 2015). Interestingly, the expression of TUBB3 did not significantly change in cells grown in NDM, but significantly decreased in cells grown in CCM. The same expression pattern was observed in ihPDLSC cultures during the neurogenic differentiation experiments Adam et al., 2020). As both alveolar bone and periodontal ligament originate from the neural crest, it is not surprising that cells isolated from these tissues exhibit a baseline expression of neural markers (Foudah et al., 2014;Heng et al., 2016).
Continuous exposure of PIMSC cultures to ADM led to a substantial time-dependent increase in the expression of angiogenesisrelated molecules, including PECAM1 and VEGFR2 ( Figure 6).
PECAM1 is a member of the immunoglobulin superfamily and likely involved in leukocyte migration, angiogenesis and integrin activation (National Center for Biotechnology Information, 2020). VEGF is a key regulator of angiogenesis (Neufeld et al., 1999). There are two tyrosine kinase receptors (VEGFR1 and VEGFR2) on endothelial cells that bind VEGF with high affinity. VEGF/VEGFR2-signaling is known to mediate a plethora of cellular functions involved in angiogenesis, like endothelial cell proliferation, migration, and survival (Abhinand et al., 2016;Chappell et al., 2009;Olsson et al., 2006). While the expression of VEGFR2 was gradually increased, the expression of VEGFR1 was markedly suppressed over time. VEGFR1 is assumed to affect vascular development by modulating VEGF-mediated VEGFR2 signaling (Chappell et al., 2009). The soluble form of VEGFR1 is known to act as a ligand sink that reduces the amount of VEGF available for VEGFR2 binding (Chappell et al., 2009). Accordingly, Shibuya described VEGFR1 as a negative regulator of angiogenesis (Shibuya, 2011).
The ability to regenerate bone is essential for the reosseointegration of an implant affected by PI. Therefore, we investigated the expression of several osteogenesis-related genes during osteoblast differentiation and bone matrix formation induced by dexamethasone, β-glycerol phosphate and inorganic phosphate. In vivo, the processes of osteoblast differentiation and bone formation are dynamically coordinated by transcription factors (like RUNX2), growth factors (like BMP2) and stage-specific signal transduction. RUNX2 is expressed by uncommitted mesenchymal cells, upregulated in preosteoblasts and immature osteoblasts, and finally downregulated in mature osteoblasts (Komori, 2010). In our study, the expression of RUNX2 did not significantly change, but exhibited a definite downregulation at day 21 and 28, when pronounced matrix mineralization occurred (Figure 7). At this stage, mature osteoblasts were likely to be present. ALPL is widely used as an early marker of osteoblast differentiation (Yang et al., 2009). In our study, the expression of ALPL was significantly increased within the first 21 days of induction and afterwards downregulated, when matrix mineralization largely covered the culture vessels. This was confirmed by Park et al., who reported a downregulation of ALPL expression during the mineralization process (Park et al., 2009). BGLAP is a non-collagenous protein associated with the late phase of osteoblast differentiation (Komori, 2010;Viereck et al., 2002). BGLAP is known to act as a regulator of bone mineralization (Neve et al., 2013). Accordingly, we observed a significant upregulation of BGLAP in the advanced stages of the experiments, when distinct matrix mineralization occurred. Bone morphogenic proteins (BMPs) belong to the transforming growth factor β superfamily. Of all BMPs, BMP2 appears to be the most potent inducer of bone formation (Ishikawa et al., 2007). In the present study, BMP2 showed a continuously increasing expression in cells grown in ODM. Interestingly, the expression of BMP2 was even stronger upregulated in cells grown in CCM. In this context, it is important to know that BMPs are not only involved in bone formation, but also in non-osteogenic developmental processes (Chen et al., 2004).
Taken together, the morphological changes and transcriptional data of our study indicate that PIMSCs have the potential to differentiate into neuron-, endothelial cell-, and osteoblast-like cells. Thus, preservation of GT during regenerative therapy of PI appears to be beneficial for the processes of re-innervation, re-vascularization and re-osseointegration. However, the mutual interaction between MSCs and the local inflammatory microenvironment must also be considered during the healing of PI lesions. There is evidence that an inflammatory microenvironment affects the proliferation, migration/homing, multilineage differentiation potential, and inflammatory cytokine pro-  (Lin et al., 2008). Molar teeth with class II furcation involvement were surgically treated using guided tissue regeneration and extracted after 6 weeks of healing.
The associated regenerating periodontal tissues were assessed using immunohistochemistry, flow cytometry and differentiation assays.
The authors reported that putative mesenchymal stromal cells were present in regenerating periodontal defects. The GT used for the present investigation was harvested from PI lesions, where a thorough non-surgical treatment was performed 3-6 weeks prior to the surgical intervention. The non-surgical treatment resulted in the resolution of inflammation signs and increased firmness of the peri-implant soft tissues. Therefore, the GT used in the present study may also be regarded as regenerating peri-implant tissues.

| CONCLUSION
Cells with MSC-like properties were isolated from the GT of PI lesions. These cells revealed the capacity to undergo neurogenic, angiogenic and osteogenic differentiation. These findings suggest that inflamed GT of PI lesions have a significant regenerative potential.
Further studies considering the immunomodulatory properties are necessary to specify the biological role of PIMSCs for the healing processes of inflamed peri-implant tissues and their potential application in regenerative treatment strategies.

ACKNOWLEDGMENTS
The authors would like to thank Dr. Matthias Ballmaier for his expertise during the flow cytometry experiments and analyses. The study was supported by the German Society of Dental, Oral and Craniomandibular Sciences (Deutsche Gesellschaft für Zahn-, Mund-und Kieferheilkunde).

CONFLICT OF INTERESTS
The authors declare that they have no competing interests.

AUTHOR CONTRIBUTIONS
Evangelia Gousopoulou conceived the study, performed the experiments, analyzed and interpreted the data and drafted the manuscript.

ETHICS STATEMENT
The present study has been approved by the Institutional Review Board (Ethics Committee of Hannover Medical School, reference number: 1096) and all donors signed an informed consent according to the Declaration of Helsinki.

DATA AVAILABILITY STATEMENT
The datasets generated and/or analysed during the current study are not publicly available but are available from the corresponding author on reasonable request.