2.1

Specimen preparation, bonding procedures and mechanical loading

Thirty non-carious human third molars were obtained with informed consent from donors (20–40 year of age), under a protocol approved by the Institution Review Board (405/CEIH/2017). Molars were stored at 4 °C in 0.5% chloramine T for up to 1 month before use. A flat mid-coronal dentin surface was exposed using a hard tissue microtome (Accutom-50; Struers, Copenhagem, Denmark) equipped with a slow-speed, water-cooled diamond wafering saw (330-CA RS-70300, Struers, Copenhagen, Denmark). A 180-grit silicon carbide (SiC) abrasive paper mounted on a water-cooled polishing machine (LaboPol-4, Struers, Copenhagem, Denmark) was used to produce a clinically relevant smear layer [ ] [step 1, graphical abstract (GA)]. The specimens were divided into two main groups (n = 15) based on the tested glass ionomer cement. A conventional glass ionomer cement, Ketac Bond (3 M Deustchland GmbH, Neuss, Germany) and a resin-modified glass ionomer cement, Vitrebond Plus (3 M Deustchland GmbH, Neuss, Germany) were employed (sep 2, GA). Glass ionomer cements were applied on dentin following the manufactureŕs instructions, and a flowable resin composite (X-Flow™, Dentsply, Caulk, UK) was placed incrementally in five 1 mm layers and light-cured with a Translux EC halogen unit (Kulzer GmbH, Bereich Dental, Wehrheim, Germany) for 40 s (step 3, GA). The detailed composition and application mode are shown in Table 1 .

The specimens were divided into three sub-groups, based on the type of mechanical loading that were applied: 1) Restored teeth stored in simulated body fluid solution (SBFS), for 24 h (no cycling). 2) Load cycling with sine waveform (259,200 cycles, 3 Hz,) (S-MMT-250NB; Shimadzu, Tokyo, Japan), and 3) Load with hold or sustained waveform, for 24 h (Instron 3345, Instron Corporation, Canton, MA, USA) (step 4, GA) (Fig. S1). After performing the load cycling test, all specimens were kept in SBFS at 37 °C, until completing a total time of 24 h of immersion in SBFS.

To proceed with the mechanical loading of samples, specimens were mounted in plastic rings using dental stone. A compressive load of 225 N was applied to the flat resin composite build-ups using a 5-mm diameter spherical stainless steel plunger, while immersed in SBFS and proceeded as in Toledano et al., 2014 [ ]. 225 N was selected as load cell (equivalent to “bite force”, in vivo ) for all mechanical loading stimuli, as the mean peak bite force of all bruxism events has been reported to be 22.5 Kgf, with a duration of 7.1 s [ ]. All specimens were longitudinally sectioned in 1.5-mm slabs from the central part of the specimen and polished through SiC abrasive paper from 800 up to 4.000-grit with a final polishing procedure performed with diamond pastes (Buehler-MetaDi, Buehler Ltd Illinois, USA), through 1 μm down to 0.25 μm (step 5, GA). The specimens were treated in ultrasonic bath (Model QS3, Ultrawave Ltd, Cardiff, UK) containing deionized water [pH 7.4] for 5 min at each polishing step.

2.2

Nano-DMA analysis and atomic force microscopy analysis (AFM) imaging

Five slabs of restored teeth were submitted to nano-DMA and AFM analysis. Property mappings were conducted using a HysitronTi 950 nanoindenter (Hysitron, Inc., Minneapolis, MN) equipped with nano-DMA III, a commercial nano-DMA package. The nanoindenter tip was calibrated against a fused quartz sample using a quasistatic force setpoint of 5 μN to maintain contact between the tip and the sample surface. A dynamic (oscillatory) force of 5 μN was superimposed on the quasistatic signal at a frequency of 200 Hz. Based on a calibration-reduced modulus value of 69.6 GPa for the fused quartz, the best-fit spherical radius approximation for tip was found to be 85 nm, for the selected nano-DMA scanning parameters. Modulus mapping of the samples was conducted by imposing a quasistatic force setpoint, F _q = 5 μN, to which we superimposed a sinusoidal force of amplitude F _A = 1.8 μN and frequency f = 200 Hz. The resulting displacement (deformation) at the site of indentation was monitored as a function of time. Data from regions approximately 50 × 50 μm in size were collected using a scanning frequency of 0.2 Hz. Viscoelastic data were acquired on the three different specimens and obtained from selected surface areas of the substrate using a rastering scan pattern. For each property map, 10 sets of 225 datapoints were used to obtain the median property value of a particular region of interest. That is, the 225 datapoints represent 1.47 × 1.47 = 2.15 μm ² of each 30 × 30 = 900 μm ² of the scan. The datapoints from 10 such non-overlapping squares were obtained for each zone at the bonded interface; thus, for each nano-DMA parameter, 30 values (3 specimens x 10 squares) were generated for each zone.

Under steady conditions (application of a quasistatic force) the indentation modulus of the tested sample, E , can be obtained by application of different models that relate the indentation force, F , and depth, D [ ]. Complex modulus ( E ^* ) and tan delta ( δ ) values were calculated as in Toledano et al., 2017 [ ]. Measurements were recorded at the glass-ionomer cement layer (GI), on either the hybrid layer (HL) or bottom of hybrid layer (BHL), and on intertubular dentin (ID), as represented in Fig. S1. For intertubular dentin, discrete indentations were made in locations that were at least 3 μm from a peritubular cuff or previous indents. Statistical analysis was performed with ANOVA and Student Newman Keuls multiple comparisons tests. P < 0.05 was set for significance.

An atomic force microscope (AFM Nanoscope V, Digital Instruments, Veeco Metrology group, Santa Barbara, CA, USA) equipped with a Triboscope indentor system (Hysitron Inc., Minneapolis, MN) was employed in this study for topography analysis. The imaging process was undertaken inside a wet cell in a fully hydrated state (deionized water), using the tapping mode, with a calibrated vertical-engaged piezo-scanner (Digital Instrument, Santa Barbara, CA, USA). A 10-nm-radius silicon nitride tip (Veeco) was attached to the end of an oscillating cantilever that came into intermittent contact with the surface at the lowest point of the oscillation. Changes in vertical position of the AFM tip at resonance frequencies near 330 kHz provided the height of the images registered as bright and dark regions. 50 × 50 μm digital images were recorded from each resin-dentin interface, with a slow scan rate (0.1 Hz). To facilitate dentin surfaces observation, AFM images were tilted using a specific software (NanoScope Analysis v. 1.40, Bruker Corporation, Billerica, MA, USA) (step 6, GA).

2.3

Raman spectroscopy and hierarchical clusters analysis

The same dentin slabs were submitted to Raman analysis using a dispersive Raman spectrometer/microscope (Horiba Scientific Xplora, Villeneuve d́Ascq, France). A 785-nm diode laser through a ×100/0.90 NA air objective was employed. Raman signal was acquired using a 600-lines/mm grafting centered between 400 and 1700 cm ⁻¹ . Chemical mapping of the surfaces was performed. For each specimen, two areas 12 × 12 μm of the surfaces at different sites were mapped using 0.5 μm spacing at X and Y axis (625 points per map). The output from a clustering algorithm was basically a statistical description of the cluster centroids with the number of components in each cluster. The biochemical content of each cluster was analyzed using the average cluster spectra. The natural groups of components (or data) based on some similarity and the centroids of a group of data sets were found by the clustering algorithm once calculated by the software and the Hierarchical Cluster Analysis (HCA). The observed spectra were described at 400–1700 cm ⁻¹ with 10 complete overlapping Gaussian lines, suggesting homogeneous data for further calculations [ ]. At this point, the mineral (relative presence of minerals and crystallinity) and the organic components (normalization, crosslinking, nature and secondary structure of collagen) of dentin was assessed as in Timlin et al. [ ], Kunstar et al. [ ] and Toledano et al. [ ] (Step 6, GA).

2.4

Confocal microscopy evaluation

Additional teeth, 6 samples per group, were used in this part of the study. In half of the specimens, previous to the cement application, glass ionomers were doped with 0.05 wt% Rhodamine-B (RhB: Sigma-Aldrich Chemie Gmbh, Riedstr, Germany). The pulpal chamber was filled with 1 wt% aqueous/ethanol fluorescein (Sigma-Aldrich Chemie Gmbh, Riedstr, Germany) for 3 h [ ]. The rest of the molars were immersed in 0.5 wt% xylenol orange solution (XO: Sigma-Aldrich Chemie Gmbh, Riedstr, Germany), excited at 514-nm for 24 h at 37 °C (pH 7.2). Specimens were copiously rinsed with water and treated in an ultrasonic water bath for 2 min. The specimens were cut in resin-dentin slabs and polished using ascending grit SiC abrasive papers (#1200 to #4000) on a water-cooled polishing device (Buehler-MetaDi, Buehler Ltd. Lake Bluff, IL, USA). A final ultrasonic cleaning (5 min) concluded the specimen preparation. Analysis of bonded interfaces were performed by dye assisted confocal microscopy evaluation (CLSM), and attained by using a confocal laser scanning microscope (SP5 Leica, Heidelberg, Germany) equipped with a 20×, 40× and 60× oil immersion lenses. Fluorescein was activated by blue light (488–495 nm) and emited yellow/green (520 nm), while the ultramorphology evaluation (cement-diffusion) was executed using Rhodamine excitation laser. Rhodamine was excited using green light (540 nm) and emitted red in color (590 nm). CLSM images were obtained with a 1 μm z-step to optically section the specimens to a depth up to 12–10 μm below the surface. The z-axis scans of the interface surface was arbitrarily pseudo-colored by the same operator for better exposure and compiled into single projections using the Leica image-processing software SP2 (Leica, Heidelberg, Germany). The resolution of CLSM images was 1024 × 1024 pixels. Five optical images were randomly captured from each resin-dentin interface, and micrographs representing the most common features of nanoleakage observed along the bonded interfaces were selected [ ]. Fluorescences were or not separated into spectral regions, allowing that the operator has a full control of the region of the light spectrum directed to each channel (step 6, GA).

2.1

Specimen preparation, bonding procedures and mechanical loading

Thirty non-carious human third molars were obtained with informed consent from donors (20–40 year of age), under a protocol approved by the Institution Review Board (405/CEIH/2017). Molars were stored at 4 °C in 0.5% chloramine T for up to 1 month before use. A flat mid-coronal dentin surface was exposed using a hard tissue microtome (Accutom-50; Struers, Copenhagem, Denmark) equipped with a slow-speed, water-cooled diamond wafering saw (330-CA RS-70300, Struers, Copenhagen, Denmark). A 180-grit silicon carbide (SiC) abrasive paper mounted on a water-cooled polishing machine (LaboPol-4, Struers, Copenhagem, Denmark) was used to produce a clinically relevant smear layer [ ] [step 1, graphical abstract (GA)]. The specimens were divided into two main groups (n = 15) based on the tested glass ionomer cement. A conventional glass ionomer cement, Ketac Bond (3 M Deustchland GmbH, Neuss, Germany) and a resin-modified glass ionomer cement, Vitrebond Plus (3 M Deustchland GmbH, Neuss, Germany) were employed (sep 2, GA). Glass ionomer cements were applied on dentin following the manufactureŕs instructions, and a flowable resin composite (X-Flow™, Dentsply, Caulk, UK) was placed incrementally in five 1 mm layers and light-cured with a Translux EC halogen unit (Kulzer GmbH, Bereich Dental, Wehrheim, Germany) for 40 s (step 3, GA). The detailed composition and application mode are shown in Table 1 .

The specimens were divided into three sub-groups, based on the type of mechanical loading that were applied: 1) Restored teeth stored in simulated body fluid solution (SBFS), for 24 h (no cycling). 2) Load cycling with sine waveform (259,200 cycles, 3 Hz,) (S-MMT-250NB; Shimadzu, Tokyo, Japan), and 3) Load with hold or sustained waveform, for 24 h (Instron 3345, Instron Corporation, Canton, MA, USA) (step 4, GA) (Fig. S1). After performing the load cycling test, all specimens were kept in SBFS at 37 °C, until completing a total time of 24 h of immersion in SBFS.

To proceed with the mechanical loading of samples, specimens were mounted in plastic rings using dental stone. A compressive load of 225 N was applied to the flat resin composite build-ups using a 5-mm diameter spherical stainless steel plunger, while immersed in SBFS and proceeded as in Toledano et al., 2014 [ ]. 225 N was selected as load cell (equivalent to “bite force”, in vivo ) for all mechanical loading stimuli, as the mean peak bite force of all bruxism events has been reported to be 22.5 Kgf, with a duration of 7.1 s [ ]. All specimens were longitudinally sectioned in 1.5-mm slabs from the central part of the specimen and polished through SiC abrasive paper from 800 up to 4.000-grit with a final polishing procedure performed with diamond pastes (Buehler-MetaDi, Buehler Ltd Illinois, USA), through 1 μm down to 0.25 μm (step 5, GA). The specimens were treated in ultrasonic bath (Model QS3, Ultrawave Ltd, Cardiff, UK) containing deionized water [pH 7.4] for 5 min at each polishing step.

2.2

Nano-DMA analysis and atomic force microscopy analysis (AFM) imaging

Five slabs of restored teeth were submitted to nano-DMA and AFM analysis. Property mappings were conducted using a HysitronTi 950 nanoindenter (Hysitron, Inc., Minneapolis, MN) equipped with nano-DMA III, a commercial nano-DMA package. The nanoindenter tip was calibrated against a fused quartz sample using a quasistatic force setpoint of 5 μN to maintain contact between the tip and the sample surface. A dynamic (oscillatory) force of 5 μN was superimposed on the quasistatic signal at a frequency of 200 Hz. Based on a calibration-reduced modulus value of 69.6 GPa for the fused quartz, the best-fit spherical radius approximation for tip was found to be 85 nm, for the selected nano-DMA scanning parameters. Modulus mapping of the samples was conducted by imposing a quasistatic force setpoint, F _q = 5 μN, to which we superimposed a sinusoidal force of amplitude F _A = 1.8 μN and frequency f = 200 Hz. The resulting displacement (deformation) at the site of indentation was monitored as a function of time. Data from regions approximately 50 × 50 μm in size were collected using a scanning frequency of 0.2 Hz. Viscoelastic data were acquired on the three different specimens and obtained from selected surface areas of the substrate using a rastering scan pattern. For each property map, 10 sets of 225 datapoints were used to obtain the median property value of a particular region of interest. That is, the 225 datapoints represent 1.47 × 1.47 = 2.15 μm ² of each 30 × 30 = 900 μm ² of the scan. The datapoints from 10 such non-overlapping squares were obtained for each zone at the bonded interface; thus, for each nano-DMA parameter, 30 values (3 specimens x 10 squares) were generated for each zone.

Under steady conditions (application of a quasistatic force) the indentation modulus of the tested sample, E , can be obtained by application of different models that relate the indentation force, F , and depth, D [ ]. Complex modulus ( E ^* ) and tan delta ( δ ) values were calculated as in Toledano et al., 2017 [ ]. Measurements were recorded at the glass-ionomer cement layer (GI), on either the hybrid layer (HL) or bottom of hybrid layer (BHL), and on intertubular dentin (ID), as represented in Fig. S1. For intertubular dentin, discrete indentations were made in locations that were at least 3 μm from a peritubular cuff or previous indents. Statistical analysis was performed with ANOVA and Student Newman Keuls multiple comparisons tests. P < 0.05 was set for significance.

An atomic force microscope (AFM Nanoscope V, Digital Instruments, Veeco Metrology group, Santa Barbara, CA, USA) equipped with a Triboscope indentor system (Hysitron Inc., Minneapolis, MN) was employed in this study for topography analysis. The imaging process was undertaken inside a wet cell in a fully hydrated state (deionized water), using the tapping mode, with a calibrated vertical-engaged piezo-scanner (Digital Instrument, Santa Barbara, CA, USA). A 10-nm-radius silicon nitride tip (Veeco) was attached to the end of an oscillating cantilever that came into intermittent contact with the surface at the lowest point of the oscillation. Changes in vertical position of the AFM tip at resonance frequencies near 330 kHz provided the height of the images registered as bright and dark regions. 50 × 50 μm digital images were recorded from each resin-dentin interface, with a slow scan rate (0.1 Hz). To facilitate dentin surfaces observation, AFM images were tilted using a specific software (NanoScope Analysis v. 1.40, Bruker Corporation, Billerica, MA, USA) (step 6, GA).

2.3

Raman spectroscopy and hierarchical clusters analysis

The same dentin slabs were submitted to Raman analysis using a dispersive Raman spectrometer/microscope (Horiba Scientific Xplora, Villeneuve d́Ascq, France). A 785-nm diode laser through a ×100/0.90 NA air objective was employed. Raman signal was acquired using a 600-lines/mm grafting centered between 400 and 1700 cm ⁻¹ . Chemical mapping of the surfaces was performed. For each specimen, two areas 12 × 12 μm of the surfaces at different sites were mapped using 0.5 μm spacing at X and Y axis (625 points per map). The output from a clustering algorithm was basically a statistical description of the cluster centroids with the number of components in each cluster. The biochemical content of each cluster was analyzed using the average cluster spectra. The natural groups of components (or data) based on some similarity and the centroids of a group of data sets were found by the clustering algorithm once calculated by the software and the Hierarchical Cluster Analysis (HCA). The observed spectra were described at 400–1700 cm ⁻¹ with 10 complete overlapping Gaussian lines, suggesting homogeneous data for further calculations [ ]. At this point, the mineral (relative presence of minerals and crystallinity) and the organic components (normalization, crosslinking, nature and secondary structure of collagen) of dentin was assessed as in Timlin et al. [ ], Kunstar et al. [ ] and Toledano et al. [ ] (Step 6, GA).

2.4

Confocal microscopy evaluation

Additional teeth, 6 samples per group, were used in this part of the study. In half of the specimens, previous to the cement application, glass ionomers were doped with 0.05 wt% Rhodamine-B (RhB: Sigma-Aldrich Chemie Gmbh, Riedstr, Germany). The pulpal chamber was filled with 1 wt% aqueous/ethanol fluorescein (Sigma-Aldrich Chemie Gmbh, Riedstr, Germany) for 3 h [ ]. The rest of the molars were immersed in 0.5 wt% xylenol orange solution (XO: Sigma-Aldrich Chemie Gmbh, Riedstr, Germany), excited at 514-nm for 24 h at 37 °C (pH 7.2). Specimens were copiously rinsed with water and treated in an ultrasonic water bath for 2 min. The specimens were cut in resin-dentin slabs and polished using ascending grit SiC abrasive papers (#1200 to #4000) on a water-cooled polishing device (Buehler-MetaDi, Buehler Ltd. Lake Bluff, IL, USA). A final ultrasonic cleaning (5 min) concluded the specimen preparation. Analysis of bonded interfaces were performed by dye assisted confocal microscopy evaluation (CLSM), and attained by using a confocal laser scanning microscope (SP5 Leica, Heidelberg, Germany) equipped with a 20×, 40× and 60× oil immersion lenses. Fluorescein was activated by blue light (488–495 nm) and emited yellow/green (520 nm), while the ultramorphology evaluation (cement-diffusion) was executed using Rhodamine excitation laser. Rhodamine was excited using green light (540 nm) and emitted red in color (590 nm). CLSM images were obtained with a 1 μm z-step to optically section the specimens to a depth up to 12–10 μm below the surface. The z-axis scans of the interface surface was arbitrarily pseudo-colored by the same operator for better exposure and compiled into single projections using the Leica image-processing software SP2 (Leica, Heidelberg, Germany). The resolution of CLSM images was 1024 × 1024 pixels. Five optical images were randomly captured from each resin-dentin interface, and micrographs representing the most common features of nanoleakage observed along the bonded interfaces were selected [ ]. Fluorescences were or not separated into spectral regions, allowing that the operator has a full control of the region of the light spectrum directed to each channel (step 6, GA).