Multicompartmental Scaffolds for Coordinated Periodontal Tissue Engineering

Abstract

Successful periodontal repair and regeneration requires the coordinated responses from soft and hard tissues as well as the soft tissue–to–bone interfaces. Inspired by the hierarchical structure of native periodontal tissues, tissue engineering technology provides unique opportunities to coordinate multiple cell types into scaffolds that mimic the natural periodontal structure in vitro. In this study, we designed and fabricated highly ordered multicompartmental scaffolds by melt electrowriting, an advanced 3-dimensional (3D) printing technique. This strategy attempted to mimic the characteristic periodontal microenvironment through multicompartmental constructs comprising 3 tissue-specific regions: 1) a bone compartment with dense mesh structure, 2) a ligament compartment mimicking the highly aligned periodontal ligaments (PDLs), and 3) a transition region that bridges the bone and ligament, a critical feature that differentiates this system from mono- or bicompartmental alternatives. The multicompartmental constructs successfully achieved coordinated proliferation and differentiation of multiple cell types in vitro within short time, including both ligamentous- and bone-derived cells. Long-term 3D coculture of primary human osteoblasts and PDL fibroblasts led to a mineral gradient from calcified to uncalcified regions with PDL-like insertions within the transition region, an effect that is challenging to achieve with mono- or bicompartmental platforms. This process effectively recapitulates the key feature of interfacial tissues in periodontium. Collectively, this tissue-engineered approach offers a fundament for engineering periodontal tissue constructs with characteristic 3D microenvironments similar to native tissues. This multicompartmental 3D printing approach is also highly compatible with the design of next-generation scaffolds, with both highly adjustable compartmentalization properties and patient-specific shapes, for multitissue engineering in complex periodontal defects.

Introduction

Nearly half of individuals in the United States aged 30 y and older have periodontitis, resulting in the destruction of supporting periodontal tissues around the teeth and often leading to tooth loss (Kinane et al. 2017). Successful regeneration of periodontal tissues requires a coordinated response among soft tissues (periodontal ligament [PDL]), hard tissues (alveolar bone, cementum), and the recapitulation of the graded mineral distribution at the interface (Hurng et al. 2011; Ivanovski et al. 2014). Common tissue engineering strategies with singular material properties have enabled promotion of selective cellular repopulation (Gonzalez-Fernandez et al. 2019). However, the regeneration of periodontium often leads to unpredictable clinical outcomes and can address only a small proportion of presenting defects (Larsson et al. 2016). Recent advances in the field have seen the emergence of tissue engineering strategies capable of recapitulating the complex spatial and temporal periodontal defect healing events, which will potentially lead to predictable periodontal treatments.

A multicompartmental scaffold can be defined by the variation within the physical architecture and/or the biochemical composition of the fabricated construct (Ivanovski et al. 2014), which will resemble to some extent the structural organization and/or biochemical composition of the native tissue (Jeon et al. 2014). Adapted from orthopedics, multicompartmental scaffolds have been recognized as having significant potential to enable periodontal tissue engineering (Jeon et al. 2014; Vaquette et al. 2018). Compartmentalization is particularly desirable for mimicking hierarchical structures of periodontium, which has the potential to provide precise presentation of regulatory cues within each engineered compartment, allowing the scaffold to direct cells to form appropriate tissue types in the correct anatomical locations (Gonzalez-Fernandez et al. 2019). A wide range of manufacturing technologies have been used to fabricate multicompartmental scaffolds to better mimic the physically relevant microenvironments and hierarchical structures of periodontium (Latimer et al. 2021). However, previous multicompartmental scaffold strategies have several limitations, including low resolution of features, porosities not well matched to tissue, inflexible shapes, and difficulties eliciting cohesion between different compartments (Hao et al. 2016).

Melt electrowriting (MEW) is a recently developed 3-dimensional (3D) printing technology (Kade and Dalton 2021), which overcomes several of the main limitations associated with solution electrospinning and fused deposition modeling (FDM). Compared to solution electrospinning, MEW can facilitate the fabrication of highly ordered and larger pore-sized architectures in a solvent-free process, avoiding the potential damage of solvents negatively affecting cells (Eichholz and Hoey 2018). When compared to FDM, MEW allows the fabrication of micron- to nanodiameter filaments (Robinson et al. 2019). These ultrathin filaments not only lead to higher resolution but also result in increased surface-to-volume ratios of the filaments, which promote cell/tissue ingrowth (Abbasi et al. 2019). In addition to these advantages, MEW can also achieve the cohesion between adjacent compartments via the continuous melting, cooling, and solidification of polymers during the electrospinning.

Given the complex set of cellular types found across interfacial tissues (e.g., periodontal ligament fibroblasts in PDL, osteoblasts in alveolar bone), establishing similar cellular populations is essential to engineering interfacial tissues (Zhou et al. 2020). Since the 3D coculture of periodontal ligament cells and osteoblasts in a FDM scaffold has been demonstrated as practical in achieving specific coordination (Vaquette et al. 2012), 2 cell types, human periodontal ligament fibroblasts (hPDLFs) and osteoblasts, were selected for this study. Specifically, primary human calvarial osteoblasts (HCOs) were chosen for the 3D coculture due to a similar osteoblast maturation process to osteoblasts derived from alveolar bones, which is known as intramembranous ossification (Matalová et al. 2015).

This study included the design and fabrication of multicompartmental scaffolds by MEW. The biological rationale for this approach arises from the variations in tissue types present in the bone–PDL complex. Based on previously published work from our group (Park et al. 2010; Park et al. 2012; Pilipchuk et al. 2016) and the rationale, we have further refined the design of multicompartmental scaffolds. This generation of multicompartmental scaffolds comprises 1) a dense mesh bone compartment, 2) a highly aligned PDL compartment, and 3) a highly interconnected transition region bridging the bone and PDL compartments. Refined design of geometry within tricompartmental scaffolds could then achieve coordinated proliferation and differentiation of primary human osteoblasts and hPDLFs for establishing 3D tissue-engineered periodontal tissue–like constructs. It was also demonstrated in a 3D coculture model that the initial cell–scaffold interaction is crucial for obtaining the mineral gradient transition with long-term incubation.

Materials and Methods

Three different constructs (Fig. 1, Appendix Table 1, Appendix Materials and Methods 1), containing 1 or multiple compartments, were designed as mesh structures (height ca. 1 mm). Tricompartmental constructs were designed with 3 different tissue-specific compartments: 1) a dense mesh bone compartment (250-µm filament spacing, 90° layer-to-layer rotation), 2) a PDL compartment (500-µm filament spacing, 0° layer-to-layer rotation) mimicking aligned ligaments, and 3) a transition region (500-µm filament spacing, 90° layer-to-layer rotation) that bridges the bone and PDL compartments. The transition region aims at initially promoting distinct and separate regions of growth in the other 2 compartments and then subsequently allows for cellular cross-communication between compartments in order to promote integration between the bone and PDL after cell maturation (Vaquette et al. 2019). The design of bicompartmental scaffolds was adapted from previous studies (Vaquette et al. 2012; Vaquette et al. 2019) possessing a bone compartment and a highly connected square-mesh region, without a highly aligned compartment. The monolithic scaffolds possessed homogeneous structures (identical to the transition regions in the other scaffolds) as a control group. The monolithic scaffolds also serve as an analog to the homogeneous materials currently used in clinical settings. Use of the bioprinting platform (3D Discovery; RegenHu) allowed all constructs to be 3D printed in a single step without manual adjustment or calibration, with a typical printing time within 10 min for each. Other details of materials and methods are listed in the Appendix.

Figure 1.

Scaffold design. (A) A schematic illustration of the biological basis to use multicompartmental scaffolds to elicit coordinated responses from both ligamentous- and bone-derived cells for periodontal tissue engineering and the melt electrowriting (MEW) fabrication of a multicompartmental scaffold. (B) A diagram detailing the monolithic, bicompartmental, and tricompartmental scaffold layering and structure. (C) A 3-dimensional rendering of the individual compartments for each scaffold type with simplified layering.

Results

MEW Scaffolds with Highly Ordered and Porous Structures

Micro–computed tomography (µCT) analysis was used to observe the 3D pore morphology and measure scaffold porosity. Shown in Appendix Figure 1A, the whole scaffold constructs exhibited highly ordered structures in all compartments. The individual unit cells of the scaffold (Appendix Fig. 1B) were easily isolated from the µCT scanning file, demonstrating the tessellated and cuboid pore geometries. The results of µCT analysis indicate that all groups achieved similarly high porosities varying between 89% and 90% (Appendix Table 1).

Scanning electron microscopy (SEM) and bright field microscopy were used to evaluate the ordering of structures and fineness of filaments. Top-down SEM micrographs (Fig. 2A1–A3) display consistent filament spacing within each. The gap between adjacent filaments varied at most by 4% for all compartments across all scaffolds (Appendix Fig. 3, Appendix Table 2). SEM analysis resulted in the determination of a mean diameter of 22.3 ± 3.7 µm, presenting an appropriately sized substrate for cell adhesion and growth (Appendix Discussion 1). All regions of the constructs could be observed in the cross-sectional SEM micrographs (Fig. 2C1–C3), clearly demarking the differences between the three types of scaffolds.

Figure 2.

Scanning electron microscopy (SEM) and brightfield microscopy–based image analyses of scaffold architecture features. (A) Top-down images of the lattice-like structure of the scaffolds. (B) Magnified images of individual scaffold compartments. (C) Cross-sectional views of the scaffolds detailing the layering and structures (blue: bone compartment; red: periodontal ligament compartment; gray: transition region). (D) Brightfield microscopy–based pore size analysis of monolithic, bicompartmental, and tricompartmental scaffolds.

Bone Compartments Promote Short-Term Preosteoblast Proliferation

Confocal laser scanning microscopy (CLSM) and WST-8 assay were used to assess the short-term attachment and proliferation of preosteoblasts through observation at days 3 and 7 after incubation (Fig. 3A, D). The scaffolds with bone compartments (bi- and tricompartmental) possessed cellular proliferation that spanned the entire perimeter of the bone compartment region of the scaffold at day 3, while the monolithic scaffold (no bone compartment) presented only minimal attachment and growth. The overall cell number (Fig. 3B) and Ki-67⁺ rate (Fig. 3C) were approximately 1.5-fold higher in the bi- and tricompartmental scaffolds when compared to the monolithic, indicating faster proliferation rate. At day 7, the scaffolds with bone compartments expressed extensive growth that spanned the entirety of the pores, while the monolithic system expressed much lower growth that resulted in incomplete coverage of the pore perimeter. The bone compartment containing scaffolds had a more than 70% increase in cell number by day 7, over the growth evaluated in the monolithic scaffold.

Figure 3.

Cell proliferation results. (A) Z-stack fluorescence images and 2-dimensional large images of preosteoblast cells labeled to present Ki-67⁺ (a proliferating indicator) cells (red), conjugated phalloidin for actin staining (green), and nuclear stain (blue) within the bone compartment of monolithic, bicompartmental, and tricompartmental scaffolds via confocal laser scanning microscopy. (B) Cell number via nuclei counting (n = 3 per group). (C) Proportion of Ki-67⁺ cells (n = 3 per group), achieved by calculating the ratio between the Ki-67–positive cells and the total cell count. (D) Cell proliferation evaluated by WST-8 assay (n = 3 per group). Two-way analysis of variance with Bonferroni’s post hoc test. *P < 0.05. **P < 0.01. ***P < 0.001. ****P < 0.0001. Of note, the dense mesh structure of the bone compartment accelerated preosteoblast proliferation at days 3 and 7.

PDL Compartments Facilitate Long-Term Formation and Orientation of Collagen

The influence of PDL compartment structure on hPDLF orientation was first investigated to assess early stage effects on the hPDLF culture. Within the tricompartmental scaffold (containing PDL compartment), highly aligned cellular structures were observed as early as 3 d (Appendix Fig. 2).

To further assess the effectiveness of PDL compartments in the facilitation of the formation of functional PDL-like tissues, 3D orientation of hPDLFs and type I collagen were measured at 14 and 21 d. Collagen was selected due to its unique status as the most abundant extracellular component in human PDL tissues (Ho et al. 2010). At day 14 (Fig. 4A), it was clear that the initial alignment of hPDLFs was effectively maintained on the tricompartmental scaffolds (P < 0.0001) and facilitated the orientation of collagen-enriched fibers (P < 0.05). The hPDLF alignment results were consistent with the impact of the PDL compartment on initial growth in the day 3 and 7 cell studies. At day 21, the alignment of type I collagen and the formation of collagen-rich fibers was quite evident for the tricompartmental scaffolds (Fig. 4B, P < 0.0001), with minimal random distribution of extracellular matrix. However, the scaffolds lacking PDL compartments (monolithic/bicompartmental) were observed to possess a distribution of type I collagen that was not well established into highly aligned fibers.

Figure 4.

Cytoskeleton and collagen alignment. Day 14 (A) and day 21 (B) confocal laser scanning microscopy images of actin (red, used for cytoskeleton alignment calculations) and collagen (green, used for collagen orientation analysis). Results for the aspect ratio of the actin (cytoskeleton) and collagen (the most abundant matrix component in human periodontal ligament tissue) presented for each day (n ≥ 6 per group). One-way analysis of variance, with post hoc Tukey’s multiple comparisons. *P < 0.05. ****P < 0.0001. Of note, a significantly higher degree of aligned, high aspect ratio features was observed for the tricompartmental scaffold at both times.

Tricompartmental Scaffolds Elicit Calcium Gradients in Long-Term Coculture of HCOs and hPDLFs

A modified 3D coculture system (He et al. 2012; Cooper et al. 2014) was used to assess the cellular and mineral distribution of multiple cell types across each scaffold. This was done to test if the multicompartmental scaffold can elicit coordinated differentiation and organization of HCOs and hPDLFs. Cellular and calcium distributions from cocultured samples could be clearly observed in cross-sectional images (Fig. 5) taken 28 d after incubation.

Figure 5.

Coculture of primary human periodontal ligament fibroblasts (hPDLFs) and human calvarial osteoblasts (HCOs) on scaffolds. (A1–C1) The cross-sectional schematic of scaffold target structure. (A2–C2) The cross-sectional view of hematoxylin and eosin staining (thickness: 5 µm) showing the organization of hPDLFs and HCOs. (A3–C3) The cross-sectional view of scaffolds after Von Kossa staining (thickness: 5 µm) showing the differentiation and mineralization of hPDLFs and HCOs with projected histograms (A4–C4) illustrating cell distribution (pink curve) and calcium deposition (black curve) in each level. (D1) Total calcium deposition quantification (n = 3 per group). Two-tailed unpaired Student’s t test. *P < 0.05. Of note, the tricompartmental scaffold presented the most coordinated organization and differentiation after 28 d.

Von Kossa staining of tricompartmental scaffolds (Fig. 5C) revealed robust mineralization in the bone compartment and stretched fibrous cells in the PDL compartment, consistent with preosteoblast results obtained from early time points. It could be observed from both hematoxylin and eosin (H&E) and Von Kossa staining that cells penetrated the transition region of the scaffolds. Taken together, these findings demonstrate that the transition region was interconnected with other compartments and could allow for cellular cross-communication between compartments over time. Pointedly, the intermediate region of the tricompartmental scaffolds resulted in ligamentous insertions with a gradual transition of biomineralization, spanning from calcified HCOs to uncalcified hPDLFs. For monolithic scaffolds, a random distribution of cell growth was observed across the entire structure and minimal biomineralization occurred over 28 d (Fig. 5A). In bicompartmental scaffolds (Fig. 5B), bone-like growth was found in both bone and PDL compartments, leading to a bimodal distribution of calcium deposition, which was also demonstrated by the osteogenic protein staining (Appendix Fig. 6). The graded mineral distribution obtained from tricompartmental scaffolds successfully mimicked the key features of native interfacial tissues in periodontium, an effect that is difficult to achieve with mono- or bicompartmental platforms.

Discussion

Regenerating periodontal tissue requires new bone formation and reestablishment of functional PDL attachment between the bone and the tooth root surface. This transition from soft tissue to bone has proven to be challenging to reconstruct with homogeneous materials due to the hierarchical structure, gradations in mineral content, and multiple cellular populations, which require a highly coordinated spatiotemporal healing responses (Vaquette et al. 2019). In this study, a MEW multicompartmental structure was proven to coordinate the proliferation and differentiation of multiple cell types in vitro, including both soft tissue– and hard tissue–derived cells. This also promoted a gradated transition from calcified to uncalcified regions, effectively recapitulating the key features of a healthy periodontium. Our findings clearly demonstrate that initial cell–scaffold interactions are critical to drive desired tissue-like responses.

Osteoblasts and osteoprogenitor cells are key components of the alveolar bone regeneration and outperformed mesenchymal stem cells in a bony environment (Reichert et al. 2011). Previous studies have suggested that osteoblasts prefer scaffolds with a defined range of pore sizes to facilitate the formation of mineralized tissue, with evidence that pore sizes of 100 to 350 µm may be optimal for osteoblast proliferation and bone formation (Karageorgiou and Kaplan 2005; Kumar et al. 2016). The preliminary results (Appendix Fig. 4) indicated that the 250-µm bone compartments had osteogenic ability for preosteoblasts, outperforming larger pore sizes. In addition, orthogonal layer-to-layer filament orientation has been reported to result in significant upregulation of alkaline phosphatase (ALP) activity and mineralization when compared to random, 45°, and 10° orientation between layers (Eichholz and Hoey 2018). Thus, a square-mesh (x–y) pattern with a pore size of 250 µm was selected. The results of this work demonstrated that the mesh structure of the bone compartment could enhance the proliferation of preosteoblasts. The authors posit that this is due to 1) a denser mesh (higher ratio of vertices),2) smaller pore size (less turbulent), and 3) increased surface-to-volume ratio (nearly doubled for the bone compartment when compared to the transitional region) (Abbasi et al. 2019). Prior studies of multicompartmental scaffolds that possess bone compartment domains mainly employed FDM (Vaquette et al. 2012; Lee et al. 2014) and 3D wax printing (Park et al. 2010; Park et al. 2012). Both techniques tend to result in low porosity (<70%) and stiffnesses inappropriate for periodontal tissues. MEW scaffolds enable distinctly different MEW structures (fiber spacing from 0.2 to 1 mm) and possess similarly favorable porosities (93%–98%), a pliability appropriate for use in periodontal tissues, and mechanical properties that are sufficiently robust as to allow for handling in a medical setting (Visser et al. 2015).

For the design of the PDL compartment, it has been demonstrated that a large pore size enhanced the integration of the PDL with newly formed alveolar bone (Vaquette et al. 2012). Lee et al. (2014) further proved that a 600-µm microchannel for the periodontal compartment yielded aligned PDL-like collagen fibers that inserted into bone-like tissue. Combining the findings of previous works and the natural aligned architecture of human PDL, a parallel structure with 500-µm filament spacing was designed for the PDL compartment in this study, which can also provide perpendicular planes for collagen fibers inserting into hard tissues. There is a benefit to advancing periodontal tissue engineering strategies beyond a tissue formation focus into one with an emphasis on tissue functions (Steier et al. 2019). To that effect, this work not only demonstrated that the PDL compartment provided rapid biophysical cues for hPDLF alignment (Appendix Discussion 2) but also facilitated the orientation of collagen-enriched fibers (Fig. 4). Many current reports used 3D waxing printing or solution electrospinning to fabricate the PDL compartment within bi- and multicompartmental systems. These systems either provided topographical guidance for periodontal fiber orientation at the surface (Park et al. 2012; Pilipchuk et al. 2016) or facilitated the delivery of PDL cell sheets (Vaquette et al. 2012). These methods tended to form a 2-dimensional (2D) surface, whereas the design of the PDL compartments in this study is shown to elicit 3D formation of collagen-rich fibers, which were either parallel with each other or achieved perpendicular insertion into transitional region. Scaffolds without PDL compartments (monolithic/bicompartmental) displayed randomly distributed and nonaligned formation of collagen. These observations could be an indicator of poorly coordinated hPDLF growth. In the tricompartmental scaffolds, a series of transitions occur from small pore to large pore to large groove (Appendix Video). The large-pore transition region is specifically designed to mimic the woven fabric-like structures observed in the PDL inserts within alveolar bone (Hurng et al. 2011). By being interrupted by orthogonal layer-by-layer fibers when compared to the grooved PDL region, collagen fibers are inhibited from expressing a predominant alignment in the region that sits between PDL and bone compartment. This results in a transition space in which isotropic tissue growth may come to dominate, which is advantageous for bone formation and integration (Eichholz and Hoey 2018).

Coculture experiments have been increasingly used to better re-create or reflect conditions in vivo. To determine whether the early architecture-induced cell behaviors corresponded to long-term commitment, we investigated the differentiation potential of multicompartmental constructs under a modified 3D coculture model (Appendix Fig. 5). This full-stack coculture study displayed responses that were consistent with the previous bone and PDL compartment studies. Insertion of ligamentous fibers from the PDL compartment toward the bone compartment was observed only within the tricompartmental scaffolds. Similarly, the calcium gradient was observed, with low calcium content in the PDL region, moderate in the transition region, and high content in the bone compartment. The presence of a transition region initially enabled good compartmentalization of the PDL and bone regions, which 1) prevented infiltration of osteoprogenitors into the PDL, thus minimizing the risk of developing ankyloses (Vaquette et al. 2019), and 2) resisted nonosteoblast interference with the bone compartment. Despite the ability to provide a “buffer” region immediately after seeding, the transition region still allowed for cellular cross-communication between bone and PDL compartments over time. This, in turn, led to the formation and insertion of PDL-like fibers into newly formed bone-like tissue. Historically, little emphasis has been put to the transitional structure for periodontal tissue engineering. This also means that the ability to increase the integration and crosstalk between soft tissues and bones has been somewhat limited to date. While some works included a fully porous (Park et al. 2012) PDL compartment, others used a compact PDL compartment with small pore size (10–20 µm) (Vaquette et al. 2019) to achieve this goal. The strategy selected for this work included the specific creation of a transition region, with evidence provided in this study that a large (but not fully open) porosity may be most appropriate for the intermediate area between bone and PDL compartments. Ultimately, it points to a “Goldilocks” zone for porosity that may be most desirable when attempting to achieve integration between soft and hard tissues within a single scaffold.

While cellular and calcium coculture results from tricompartmental scaffolds presented as a gradient, it is of note that the bicompartmental scaffolds led to bimodal distribution. This indicated the calcification of hPDLFs in the bicompartmental scaffold. This further reinforces the concept that 2 initial compartments and an appropriately designed interstitial region are needed to ensure protection of the hPDLFs in the PDL region. This result is consistent with a prior study, which found that the either direct or indirect coculture of bone marrow cells and PDLFs could lead to significantly upregulated osteogenic gene expression and increased mineralized matrix formation (Yu et al. 2015). This was not an issue of the tricompartmental scaffolds, where notable mineralization was not found in the PDL compartment. The authors propose that this is due to 1) transition region protection of the hPDLFs at early time points, 2) the highly aligned structures of the PDL compartment favoring maturation of hPDLFs, and 3) a sufficiently porous transition region to allow for PDL-like fiber insertion growth with maturation. The monolithic scaffold coculture results did not display any regionally confined calcification, with a random distribution observed. This finding is reasonable given the lack of appropriate compartments for the confinement and protection of either cell group.

This study mainly investigated the variation in the architecture of multicompartmental scaffolds. However, it may also be important to vary chemical composition within the construct, ideally to better mimic the physiological microenvironments of the periodontium. In the future, it would be promising to combine specifically tissue-engineered architectures with different biological microenvironments within a single construct.

Conclusion

In this study, 3D MEW multicompartmental structures were proven to coordinate the proliferation and differentiation of multiple cell types. It is the first reported attempt to employ MEW to fabricate multicompartmental scaffolds comprising bone compartment, PDL compartment, and transitional region in periodontal tissue engineering. The MEW technique is shown to overcome the main limitations of previous multicompartmental strategies, including low resolution, unmatched porosity, inflexible shape, and difficulty of cohesion between different compartments. Careful consideration and design of geometric factors promoted cell proliferation, cell expression, cell alignment, and extracellular biomolecule alignment in vitro, all in keeping with early stage periodontal regeneration in vivo. The 3D structure-induced cell behaviors promoted a gradient transition from calcified to uncalcified regions with longer-term growth, effectively recapitulating the key features of native interfacial tissues in periodontium. This multicompartmental approach reveals the importance of developing 3D tissue-engineered constructs that better mimic the physical structure of native tissues and advances the development of next-generation scaffolds for interfacial and multitissue engineering.

Author Contributions

Y. Yao, contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript; J.E. Raymond, contributed to analysis, interpretation, drafted and critically revised the manuscript; F. Kauffmann, S. Maekawa, J.V. Sugai, contributed to acquisition, interpretation, critically revised the manuscript; J. Lahann, contributed to interpretation, critically revised the manuscript; W.V. Giannobile, contributed to conception, design, data interpretation, critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.