3‐dimensional visualization of implant‐tissue interface with the polyethylene glycol associated solvent system tissue clearing method

Abstract Objectives Dental implants are major treatment options for restoring teeth loss. Biological processes at the implant‐tissue interface are critical for implant osseointegration. Superior mechanical properties of the implant constitute a major challenge for traditional histological techniques. It is imperative to develop new technique to investigate the implant‐tissue interface. Materials and methods Our laboratory developed the polyethylene glycol (PEG)‐associated solvent system (PEGASOS) tissue clearing method. By immersing samples into various chemical substances, bones and teeth could be turned to transparent with intact internal structures and endogenous fluorescence being preserved. We combined the PEGASOS tissue clearing method with transgenic mouse line and other labelling technique to investigate the angiogenesis and osteogenesis processes occurring at the implant‐bone interface. Results Clearing treatment turned tissue highly transparent and implant could be directly visualized without sectioning. Implant, soft/hard tissues and fluorescent labels were simultaneously imaged in decalcified or non‐decalcified mouse mandible samples without disturbing their interfaces. Multi‐channel 3‐dimensional image stacks at high resolution were acquired and quantified. The processes of angiogenesis and osteogenesis surrounding titanium or stainless steel implants were investigated. Conclusions Both titanium and stainless steel implants support angiogenesis at comparable levels. Successful osseointegration and calcium precipitation occurred only surrounding titanium, but not stainless steel implants. PEGASOS tissue clearing method provides a novel approach for investigating the interface between implants and hard tissue.


| INTRODUC TI ON
Dental and orthopaedic implants have revolutionized the treatment of patients with missing teeth or damaged bones and joints. 1 Success of implant placement is determined by their interactions with the host tissue occurred mostly at the implant-tissue interface. 2,3 Among the many biological processes of implant-tissue interface being evaluated, angiogenesis and osteogenesis are the two most important ones. [4][5][6] Implants interact with local vasculature and regulate the production of endothelial progenitor cells to form new blood vessels, which bring in oxygen, nutrition and stem cells to the interface. 7 Although a large number of studies have confirmed the close association between angiogenesis and osteogenesis during osseointegration process, 8,9 direct visualization of the two processes simultaneously remained extremely challenging due to limited research approaches.
Growing evidence has suggested that interplays between blood vessels with bone tissue occur in 3-dimensions and are essential for regulating local stem cell populations and various signal pathways. 10,11 Multiple transgenic mouse models have been developed to label different types of tissues with endogenous fluorescence, including GFP and tdTomato. 12 These mouse models provide possibility to investigate multiple events occurred at the implant-bone interface concomitantly.
Histological sectioning methods remain as the golden standards for visualization and quantitative measurement of peri-implant angiogenesis and new bone formation. 9,13 During the sample preparation, implants were removed to enable tissue embedding and sectioning, 14 which inevitably destroyed the integrity of the interface. 15 Alternatively, ground sections were employed to study implant-bone interface without decalcification treatment or removing implants. 15,16 Although calcein green dynamic labelling and several other histology staining could be performed, the hard tissue embedding process quenched endogenous fluorescence and compromised most immunofluorescence staining signals. 17,18 In addition, only very few ground sections could be achieved from a sample, which provided very limited information for the whole tissue. 19,20  Tissue clearing technique enables deep 3-D imaging of tissues with a confocal, two-photon or light-sheet microscope by turning them transparent. [21][22][23][24][25][26] Hard tissues are opaque because of mismatched refractive index (RI) among various components, including minerals, lipids, pigments and water. All tissue clearing methods followed similar principle which is to remove components blocking or diffracting the light. Transparency can finally be achieved after the tissue interstitial fluid is replaced with clearing medium with consistent RI. 27 Once the transparency being reached, images can be acquired even at several millimetres depth with high resolution.
Current tissue clearing methods can be classified into three major categories: (a) Organic solvent-based clearing methods, including DISCO series, 23,28,29 Fluoclear 22 and polyethylene glycol (PEG)-associated solvent system (PEGASOS). 30 (b) Aqueous reagent-based clearing methods, including Scale, 31 ClearT, 32 SeeDB, 33 CUBIC series. 25,34,35 (c) Hydrogel-based clearing methods, including CLARITY, 24 PACT. 21 Three major criteria to evaluate a clearing method include transparency outcome, fluorescent preservation and applicability of tissues. Overall, solvent-based methods achieve better transparency than other types. Aqueous methods achieve better fluorescence preservation than other methods. Several methods have been developed for clearing hard tissues, including PACT-deCAL, 21 mPACT, 36 CUBIC, 25 Bone CLARITY, 37 PEGASOS 30 and vDISCO. 28 Among them, PEGASOS has its unique advantages as it achieved favourable transparency of both soft and hard tissues with fluorescence preservation, relatively short time and low cost.
Intact mouse head, mandible bone with teeth, knee joint and long bone could be imaged with a two-photon microscope after tissue clearing process without sectioning. 30 In the current study, we introduced the application of the PEGASOS method on studying the implant-bone interface. A transgenic mouse model, Cdh5-Cre ERT2 ; Ai14 mouse line, was used to label blood vessels specifically. Angiogenesis and osteogenesis at the interface of stainless steel (SS) implants or titanium (Ti) implants were investigated. We demonstrated that 3-D imaging based on PEGASOS tissue clearing method is a useful new tool for investigating the implant-tissue interface.
Following that, titanium implant (0.6-mm-diameter titanium dentine pins, STABILOK) or stainless steel dentin pin (0.6-mm-diameter stainless steel dentin pins, STABILOK) was screwed into the extraction socket with ~1.5 mm depth and was cut at the level of gingiva level.

| μCT analysis
Mandible samples were placed in a 12.3 mm tube, and μCT scanning was performed using a SCANCO μCT35 device at Texas A&M University, College of Dentistry. The μCT images were acquired with the X-ray source at 70 kV voltage. The data were collected at a voxel size of 7 μm resolution. The reconstruction of 3-D images was performed with ImarIs 9.0 (Bitplane, AG, Zurich, Switzerland).

| PEGASOS tissue clearing process
Polyethylene glycol-associated solvent system tissue clearing was performed as previously described. 30

| 3-D reconstruction of images
Individual channels were merged with Imagej (NIH, Bethesda, MD, USA). Three-dimensional reconstruction and quantitative analysis were performed with ImarIs 9.0 (Bitplane). When performing reflection imaging, non-specific reflecting signal was also detected in bone tissue.
To remove the background noise, reflection signal from the implant on each optical slice was manually outlined to create a Surface. The reflection light channel was masked under the created "Surface," and a new individual channel was created to depict only the implant without background noise. Image stacks were reconstructed using the "volume rendering" function. Snapshot was generated using the "Snapshot" function. Movies were generated using the "Animation" function.

| Quantitative analysis of blood vessels volumes and bone volume
Quantitative analysis was performed with ImarIs 9.0 (Bitplane). A 150 × 150 × 150 μm stack in the thread grooves of each implant was defined as region of interest (ROI). Only the channel representing blood vessels was involved in analysis. Volume of selected blood vessels in the ROI was quantified using "Statistics" function. For each sample, at least four randomly selected ROIs were selected for quantification.
Bone volume was quantified based on the Second Harmonic Generation signal (SHG). ROI was generated in the SHG signal channel. Volume of selected structure in ROI was quantified using "Statistics" function.

| Analysis of vasculature-implant direct contact points
Stacks of 150 μm thickness containing both tdTomato and reflection signal channels are selected near the implant groove surface. Direct contact between blood vessels and implant can be visualized on individual optical slice. The number of direct contact points can then be measured for the entire image stack.

| BV/TV quantification
Bone volume/total volume (BV/TV) quantification in μCT image data was performed with ImarIs 9.0. BV/TV was defined as: the volume of selected high-density region/total volume of ROI.

| Statistic analysis
N numbers are displayed in the figures. Data are presented as mean ± standard deviation using Student's t tests or one-way

ANOVA. Statistical analysis was performed with Microsoft Excel and
GraphPad Prism.

| Data availability
The data that support the findings of this study are available from the corresponding author, HZ, upon reasonable request. F I G U R E 4 Three-dimensional analysis indicates the progressive angiogenesis and osteogenesis on the surface of titanium implants during the osseointegration process. Adult Cdh5-Cre ERT2 ; Ai4 mice (2 mo of age) were used for titanium implant placement. Samples cleared with decalcified polyethylene glycol (PEG)-associated solvent system method were imaged with a two-photon microscope. Optical stacks of 150 μm thickness were acquired to demonstrate blood vessels (tdTomato, red) surrounding the interface between implant (reflection image, blue) and mandible bone (second harmonic generation, green). A1-A3, Images were acquired with 10× objective 1 d after implant placement. A4-A5, Boxed areas in A1 and A2 were re-imaged with a 25× objective. A1-A3, Images were acquired with 10× objective 1 wk after implant placement. B4-B5, Boxed areas in B1 and B2 were re-imaged with a 25× objective. C1-C3, Images were acquired with 10× objective 2 wk after implant placement. C4-C5, Boxed areas in C1 and C2 were re-imaged with a 25× objective. D1-D3, Images were acquired with 10× objective 3 wk after implant placement. D4-D5, Boxed areas in D1 and D2 were re-imaged with a 25× objective. The diameter of the implant is 600 μm, and reflected light could not pass through the radius position. Therefore, no signal was detected

| PEGASOS tissue clearing method efficiently renders mandible bones and teeth transparent
beyond 500 μm depth ( Figure 1K). 3-D reconstruction of the reflection signal depicted half side of the implant surface ( Figure 1L, Movie S1). Working distance of ZEISS 25×/0.8NA objective is no more than 500 μm. When using a 10×/0.3NA objective with 5.2 mm working distance, tdTomato and SHG signal could be detected as deep as 800 μm ( Figure 1O-Q).
We imaged a 3413 × 2317 × 400 μm volume with a 10×/0.3NA objective using a two-photon microscope. Alveolar bone and dental root, which were enriched with Collagen I, could be detected with SHG. Enriched blood vessels were detected surrounding the implant and within the bone marrow space ( Figure 1M). Selected area near the implant surface was re-imaged with a 25×/0.8NA objective.
Enriched blood vessel and bone were clearly detected on the implant surface with direct contacts with the implant surface (arrows in Figure 1N, Movie S2).

| PEGASOS clearing process preserves intact bone-implant interface
It remains controversial whether decalcification compromises the bone-implant interface. In addition, it remains unknown whether SHG signal can truly display the bone structure with sufficient details. To test this, we harvested a mandible sample 1 month after titanium implant placement. μCT images were acquired prior to clearing process. Next, the sample was cleared following the PEGASOS decalcification method and SHG signal was imaged.
We were able to locate an identical anatomical region in both μCT and SHG image dataset for comparison. Both μCT and SHG signal revealed trabecular bone organization near the implant surface ( Figure 2A,B). Overlaying the two images showed complete overlapped details from the two datasets ( Figure 2C). Boxed area on the implant surface was zoomed in. In μCT image, bone on the implant surface could not be distinguished from the implant due to metal halation artefact ( Figure 2D). 39,40 In contrast, SHG signal image showed no such artefact and the boundary between bone and implant could be clearly identified ( Figure 2E). Overlaying of the two images indicates nearly all structures revealed with μCT analysis were also displayed in SHG image ( Figure 2F). In summary, PEGASOS decalcification method preserves intact bone-implant interface and SHG signal reveals better bone tissue details than μCT analysis.

| PEGASOS renders non-decalcified mandibles partially transparent and enables visualization of calcein green signal
Calcein green labelling is a routine technique for investigating dynamic osteogenic activity of the bone tissue by detecting their calcium precipitation and is not compatible with decalcification treatment. 41 Mandible bones cleared with PEGASOS method without decalcification achieved only partial transparency. We tested whether calcein green labelling signal can be visualized in 3-D. Adult Cdh5-Cre ERT2 ; Ai14 mice of 6-8 weeks of age were induced with tamoxifen.
Seven days later, calcein green was injected and mice were sacri-

| Deep imaging of cleared mandible bone samples with decalcification revealed angiogenesis and osteogenesis processes at the titanium implant-bone interface
Next, we investigated the osteogenesis and angiogenesis processes at the titanium implant-bone interface. On the next day of the titanium implant placement surgery (Day 1), few blood vessels were detected near the implant surface ( Figure 4A1,A3). Little SHG signal was detected near the implant surface ( Figure 4A2,A3). Enlarged images confirmed the lack of vasculature and bone near the implant surface ( Figure 4A4,A5). One week after procedure, significant F I G U R E 5 Three-dimensional quantitative analysis shows comparable angiogenesis but reduced osteogenesis on the surface of stainless steel implants than on titanium implants surface. Adult Cdh5-Cre ERT2 ; Ai4 mice (2 mo of age) were used for stainless steel implant placement. Samples with stainless steel implants were cleared with decalcified polyethylene glycol (PEG)-associated solvent system method and imaged in the same way as for the titanium implants. Optical stacks of 150 μm thickness were acquired to demonstrate blood vessels (tdTomato, red) surrounding the interface between implant (reflection image, blue) and mandible bone (second harmonic generation, green). (A1-A3) Images were acquired with 10× objective 1 d after implant placement. A4-A5, Boxed areas in A1 and A2 were re-imaged with a 25× objective. B1-B3, Images were acquired with 10× objective 1 wk after implant placement. B4-B5, Boxed areas in B1 and B2 were re-imaged with a 25× objective. C1-C3, Images were acquired with 10× objective 2 wk after implant placement. C4-C5, Boxed areas in C1 and C2 were re-imaged with a 25× objective. D1-D3, Images were acquired with 10× objective 3 wk after implant placement. D4-D5, Boxed areas in D1 and D2 were re-imaged with a 25× objective. E, Comparison of total blood vessel volumes around Ti and SS implants at different time points. The quantification results for Ti implants were from Two weeks after surgery, more blood vessels were detected within the thread grooves. Most of them were 2-5 μm in diameter and contacted directly with the implant surface ( Figure 4C1,C3,C4). Stronger SHG signal within the thread grooves was detected, suggesting more new bone formation within the thread grooves ( Figure 4C2,C5).
Three weeks after surgery, blood vessels were much enriched within the thread groove. Plenty of contacts were observed between blood vessels and the implant surface ( Figure 4D1,D3

| 3-D imaging of non-decalcified samples revealed distinct calcium precipitation activities surrounding titanium or stainless steel implants
To further test the mineral deposition activity near these two implant surfaces, we performed calcein green labelling to mark the calcium precipitation regions. Three weeks after titanium implant placement, abundant blood vessels were detected surrounding the titanium implant surface ( Figure 6A). Strong calcein green signal was detected suggesting highly active osteogenic process ( Figure 6A). Enlarged images indicated existence of calcein green signal immediately on the titanium implant surface in close association with blood vessels (Figure 6B-D). In contrast, little calcein green signal was detected surrounding the stainless steel implants 3 weeks after surgery ( Figure 6E). Although plenty of blood vessels were visualized on the stainless steel implant surface ( Figure 6F-H), little calcein green signal was detected surrounding blood vessels near the implant surface ( Figure 6G,H). Active calcein green signal was only detected at a distance from the stainless steel implant surface ( Figure 6E). resolution, which is far better than any current µCT equipment. In addition, SHG image has no halation surrounding metal implant, which is a common artefact for µCT analysis. 48,49 Although angiogenesis is known to be a prerequisite for osteo- Further investigation is needed to test this hypothesis.

| D ISCUSS I ON
Polyethylene glycol-associated solvent system method has its own limitations when imaging implant-tissue interface. First of all, PEGASOS treatment leads to differential shrinkage among soft and hard tissues, which may cause anisotropic distortion for organs composed of multiple tissue types. Second, although PEGASOS protects endogenous fluorescence better than other solvent-based clearing methods, it still compromises GFP or tdTomato fluorescence intensity significantly. Third, auto-fluorescence from bone marrow and muscle tissue may increase after clearing treatment, which deteriorates signal/noise ratio especially in deep region. 30 Our laboratory is working to improve PEGASOS method to overcome these limitations.
In the current study, we introduced the application of the PEGASOS tissue clearing method on studying bone-implant interface. By using Cdh5-Cre ERT2 ; Ai14 mouse model to label vascular endothelium, we demonstrated the angiogenesis and osteogenesis processes at the implant-bone interface. We showed that both titanium and stainless steel implants support angiogenesis, but only titanium implants support osteogenesis and osseointegration. 3-D multi-channel images of calcein green labelling with other signals further confirmed the distinct osteogenic activities on surfaces of two different types of implant. PEGASOS tissue clearing-based deep imaging provides a valuable new tool for studying tissue-material interactions and will help researchers to design better strategies for tissue engineering and regeneration.

ACK N OWLED G EM ENTS
We thank Kate Phelps and the UTSW Live Cell Imaging Facility for microscopy technical support. We thank Bridget Samuels for critical reading of the manuscript. We thank Ms. Meng Zhang and Ms. Evelyn Zhao for the support. This study was supported by the start-up funding from the Texas A&M University, the NIH/NIDCR K08 (K08DE025090), the NIH/NIDCR R21 (R21 DE027928), the NINDS R21 (5R21NS099950), American Heart Association Innovative Research Grant (17IRG33410377) and NINDS K99/R00 (R00NS073735) to Hu Zhao and Woo-Ping Ge.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.