Advanced glycation end products cause RAGE‐dependent annulus fibrosus collagen disruption and loss identified using in situ second harmonic generation imaging in mice intervertebral disk in vivo and in organ culture models

Abstract Aging and diabetes are associated with increased low‐back pain and intervertebral disk (IVD) degeneration yet causal mechanisms remain uncertain. Advanced glycation end products (AGEs), which accumulate in IVDs from aging and are implicated in diabetes‐related disorders, alter collagen and induce proinflammatory conditions. A need exists for methods that assess IVD collagen quality and degradation in order to better characterize specific structural changes in IVDs due to AGE accumulation and to identify roles for the receptor for AGEs (RAGE). We used multiphoton microscopy with second harmonic generation (SHG), collagen‐hybridizing peptide (CHP), and image analysis methods to characterize effects of AGEs and RAGE on collagen quality and quantity in IVD annulus fibrosus (AF). First, we used SHG imaging on thin sections with an in vivo dietary mouse model and determined that high‐AGE (H‐AGE) diets increased AF fibril disruption and collagen degradation resulting in decreased total collagen content, suggesting an early degenerative cascade. Next, we used in situ SHG imaging with an ex vivo IVD organ culture model of AGE challenge on wild type and RAGE‐knockout (RAGE‐KO) mice and determined that early degenerative changes to collagen quality and degradation were RAGE dependent. We conclude that AGE accumulation leads to RAGE‐dependent collagen disruption in the AF and can initiate molecular and tissue level collagen disruption. Furthermore, SHG and CHP analyzes were sensitive to collagenous alterations at multiple hierarchical levels due to AGE and may be useful in identifying additional contributors to collagen damage in IVD degeneration processes.


| INTRODUCTION
Back pain is a leading cause of global disability commonly associated with intervertebral disk (IVD) degeneration. 1 IVD degeneration is an aberrant, cell-mediated response to progressive structural failure. 2 The relationship between back pain and IVD degeneration is complex, resulting in an etiology that is multifactorial and heterogeneous, highlighting a need to understand structural changes and biological responses. Aging and diabetes are identified as risk factors associated with increased IVD degeneration and back pain. [3][4][5][6] These associations may be attributed to chronic proinflammatory conditions, yet these associations are confounded by environmental and genetic factors, making causal relationships difficult to identify. [7][8][9] A leading hypothesis for a relationship between diabetes and IVD degeneration is the formation and accumulation of advanced glycation end products (AGEs) in diabetic IVD tissue. 10,11 AGEs are highly oxidant compounds that accumulate in aging and are implicated in diabetic complications that are known to cause structural and biological alterations to collagen and the extracellular matrix (ECM). 12 There is mounting evidence for a causal relationship between IVD degeneration and AGEs. AGEs can accumulate in spinal tissues from aging, high-AGE (H-AGE) diets (eg, highly processed western diets) and diabetes, and are associated with structural changes in the IVD including decreased glycosaminoglycan content, [13][14][15] increased vertebral bone changes, 16 and increased collagen degradation. 15 In addition, the receptor for AGEs (RAGE) has been observed to initiate an NF-kB mediated inflammatory response in both human and mice IVD tissue exposed to AGEs. 17,18 The specific structural changes to the IVD ECM due to AGE exposure in the presence of RAGE are not well-understood and we believe this is partly due to limitations in the methods used to identify early degenerative changes to the ECM that mark the initiation of a degenerative cascade. Recently, we demonstrated that dietary accumulation of AGEs in the IVD increased levels of molecular level collagen degradation, 15 highlighting the direct contributions that AGEs can make to IVD degeneration. We believe that an improved understanding of the factors that cause early degenerative changes and the events that are subsequent to initiation of these changes may inform methods to detect and prevent such contributors to IVD degeneration. This motivates the development of tools to characterize changes in collagen quality and quantity in response to AGEs, as well as an investigation into the role of RAGE in initiating these structural changes.
Multiphoton microscopy is a powerful tool to analyze the ECM of collagenous tissues due to the emission of the second harmonic generation (SHG) signal from fibrillar collagen. The intensity of the SHG signal is related to several biological characteristics of collagen including collagen orientation, 19,20 collagen fibril diameter, 21 total collagen content, 22,23 collagen degradation, 24,25 and collagen crosslinking. 26,27 In addition to the quantification of mean SHG intensity, second order statistics (eg, entropy, energy, and inertia) of SHG images proved useful in characterizing collagen structure in tendon, cardiac, and tumor tissues. [28][29][30][31][32] Percent area calculations of positive SHG pixels above a threshold value has been used extensively to assess the total amount of collagen. [33][34][35][36] SHG is particularly amenable to investigations of the annulus fibrosus (AF) in the IVD since the AF is abundant in collagen type I. We believe that the development of these image analysis tools for application assessing collagen quality and quantity in the AF will provide methods that improve understanding of collagen changes in early IVD degeneration and also increase understanding of ECM changes in response to AGE exposure and accumulation.
This two-part study used in vivo and ex vivo IVD model systems with wild type and RAGE-knockout (RAGE-KO) mice in order to investigate changes in AF collagen quality and degradation in response to AGE challenges using SHG methods ( Figure 1). We first hypothesized that loss of SHG intensity and collagen degradation due to H-AGE diet will result in AF collagen fibril disruption and a loss of overall collagen.
To test this hypothesis we utilized a dietary mouse model, validated previous SHG Intensity and collagen-hybridizing peptide (CHP) results, 15 and performed new SHG image analysis to measure changes to fibril disruption and collagen quantity. The second hypothesis was that this early degenerative cascade is a result of resident IVD cell activity via the recognition of AGE ligands through RAGE, leading to degradation of the surrounding collagen. We isolated cell-mediated effects from biomechanical or systemic level effects by using an ex vivo organ culture model to evaluate the effects of AGEs and RAGE in WT and RAGE-KO mice. Using the organ culture model system, we could assess the role of IVD cells in early degenerative collagenous changes in the absence of systemic alterations (eg, hormonal, immune and/or biomechanical) to the animal from AGEs. 37 Collagen degradation, quality, and quantity were observed using CHP and SHG imaging methods and quantitative image analysis using intensity and morphometric approaches to measure differences between groups. Furthermore, SHG analyzes were performed with thin sections for the in vivo model and with in situ imaging in the organ culture to determine if both methods can characterize ECM alterations, and to provide a proof-of-concept for the potential of nondestructive situ imaging with SHG.

| Study design
Aim 1 characterized collagen quality and quantity in the AF of IVD tissues from mice fed either a high or low-AGE diet for 6 months. Aim 2 characterized collagen quality and quantity in the AF of IVD tissues using an ex vivo organ culture model of both WT and RAGE-KO IVDs.
Specimens were analyzed with SHG and CHP and image processing algorithms to assess mean SHG intensity, SHG entropy as a measure of collagen fiber disruption, percent SHG area fraction as a measure of collagen quantity, and CHP intensity as a measure of collagen damage. Aim 1 used thin sections for SHG and CHP analyzes. Aim 2 used in situ SHG analyzes and used thin sections for CHP analyzes. Further, Aim 1 used animals of both sexes, and since sex had no significant effect for AF parameters we pooled for sex. Due to the lack of a significant sex effect in the AF parameters a single sex (males) was selected in Aim 2 studies to minimize variance and size differences.  15 Media was exchanged daily and the culture plate was kept on a rocker in an incubator at 37 C, 5% CO 2 , 5% O 2 , and > 90% humidity. All IVDs underwent a 24 hour preconditioning period in control media, after which the experimental group received AGE-BSA supplemented media. Following the culture period, a subset of IVDs each were immediately fixed in 4% formaldehyde, resin embedded, and sectioned for CHP assessment. The remaining IVDs were flash frozen in liquid N 2 and stored in −80 C until in situ multiphoton imaging. Similar organ culture procedures were previously applied to mice IVDs in our lab demonstrating metabolically active cells and no evidence for apoptosis. 39 Cell viability was validated in this study F I G U R E 1 Study design. A, Dietary mouse model and B, ex vivo organ culture was used to assess the role of AGEs and RAGE in the initiation of early degenerative changes in the IVD. In situ SHG imaging was performed to assess collagen structure and amount. CHP was performed to assess molecular collagen damage. AGEs, Advanced glycation end products; CHP, collagen-hybridizing peptide; IVD, intervertebral disk; RAGE, receptor for AGEs; RAGE-KO, RAGE-knockout; SHG, second harmonic generation; WT, wildtype using double staining of thiazolyl blue tetrazolium bromide (MTT, Sigma-Aldrich, St. Louis, Missouri) to stain metabolically active cells and 4 0 ,6-diamidino-2-phenylindole (DAPI, Roche Diagnostics, Germany) to stain cell nuclei, as described. 40 Live cells were confirmed on a single motion segment by the presence of dual staining robust MTT staining that aligned closely with DAPI staining (Supplementary Figure 1).

| In situ multiphoton microscopy
Prior to imaging, whole IVDs from organ culture were washed in dd-H2O and halved using a size 10 surgical scalpel along the midsagittal plane. In situ imaging was performed on an Olympus FV1000

| Histological multiphoton microscopy
PMMA embedded IVDs from the in vivo dietary mouse model were sectioned mid-sagittally at 5 μm and used for multiphoton microscopy.
The multiphoton setup was used as described above and optical sections with a step size of 1.5 μm were taken through a depth of~10 μm.

| Image analysis and quantification
Intensity and morphometric evaluation was performed using a custom Matlab (The Mathworks Inc., Natick, Massachusetts) code to quantify the SHG intensity, percent area, and entropy of collagen in the IVD.
Background subtraction was first performed on SHG images and a 500 x 500 pixel 2 region of interest in the central anterior AF was selected for evaluation. The mean intensity of pixels within this region of interest was measured for comparison among the groups. To assess changes in the morphology of collagen structure independent of changes in intensity, a histogram equalization was performed and the entropy of the images was measured to determine the complexity of the collagen structures. Lower entropy values indicate a loss in the complexity of the AF structure, indicating a disruption to normal fibril and fiber architecture. These images were then binarized with the built in Otsu segmentation function to determine SHG signal above an automated threshold. The percent area of positive pixels was calculated as a measure of percent area of collagen and compared among groups.

| Collagen hybridizing peptide
To assess IVDs for molecular damage to collagen, PMMA-embedded mid-sagittal lumbar sections were stained using collagen hybridizing peptide-biotin conjugate (B-CHP; BIO300, 3Helix Inc; Salt Lake City, Utah) or fluorescein conjugate (F-CHP; FLU600, 3Helix Inc; Salt Lake City, Utah) to detect the presence of collagen damage as described previously. 24,25,41 This procedure includes the creation of a 2 μM   In response to a H-AGE diet, IVDs exhibited significantly decreased mean SHG intensity (percent difference − 25.8%; P < .033; Figure 2A). CHP analyzes identified significantly increased signal intensity demonstrating increased collagen damage (percent difference 63.5%; P < .006; Figure 2B). 15 Texture analysis of the SHG images demonstrated significantly decreased SHG entropy in the H-AGE group, indicating fibril level disruption of the collagen ECM in the anterior AF (percent difference − 6.76%; P < .009; Figure 2C).

| Statistics
Finally, the percent area of SHG signal was significantly but slightly decreased in the H-AGE group (percent difference − 1.19%, P < .043), suggesting a subtle decrease in the total amount of collagen in the anterior AF ( Figure 2D).

| RAGE dependent increase in collagen damage at multiple hierarchical levels in an IVD ex vivo organ culture model
In situ SHG 3D reconstruction and maximum intensity z-projections were used to evaluate the overall tissue morphology of the outer AF ( Figure 3A,B). Culture with AGE-BSA significantly decreased SHG intensity of the anterior AF in the WT IVDs (percent difference − 62.8%; P < .001; Figure 3C). The key result is that AGE-BSA caused a significant reduction in SHG intensity in WT while AGE-BSA did not cause a significant reduction in SHG intensity for RAGE-KO.
It is also notable that SHG intensity values for WT and RAGE-KO had different baseline levels, which were significantly different when comparing between Control WT and AGE-BSA RAGE-KO (P < .017).
F I G U R E 2 H-AGE diet caused increased collagen damage, increased fibril disruption, and decreased total collagen amount. A, SHG images of the central anterior AF region were quantified for SHG intensity. (n = 10, n = 11). B, CHP images of the anterior AF indicated collagen damage in the central anterior AF (n = 12, n = 11). SHG images were C, equalized and then D, segmented to perform morphometric analysis and entropy and percent area was quantified as a measure of collagen fibril disruption and total amount of collagen respectively (n = 10, n = 11). P values represent two-tailed t test comparing the L-AGE and H-AGE groups. AF, annulus fibrosus; AGEs, Advanced glycation end products; CHP, collagen-hybridizing peptide; SHG, second harmonic generation To assess the role of RAGE in changes to collagen we evaluated RAGE-KO and determined that RAGE-KO was protective against this loss of SHG intensity (P > .999). We therefore directly measured changes in collagen degradation using CHP analysis, and determined that AGE-BSA caused significantly increased the binding of CHP to collagen indicating greater collagen damage (percent difference 29.8%, P < .030; Figure 4A). RAGE-KO IVDs were not affected by AGE-BSA media (P > .999; Figure 4B), supporting the concept that AGEs caused RAGE-dependent molecular degradation of collagen ( Figure 4C).
To determine if the damage to the molecular structure of collagen was linked to the higher order levels of collagen fibril structure, we evaluated SHG entropy. Texture analysis showed a significantly decreased SHG Entropy of the WT IVDs cultured in AGE-BSA (percent difference − 7.86%; P < .012; Figure 5A). SHG Entropy was not affected by AGE-BSA media in the RAGE-KO groups (P > .999; F I G U R E 3 AGE-BSA caused reduced SHG intensity in WT but not RAGE-KO mice. A, SHG 3D reconstruction of ex vivo organ culture of WT and RAGE-KO IVDs to observe collagen microstructure with minimal processing of tissue. B, A maximum intensity z-projection of the central anterior AF was created for image analysis. C, SHG intensity was quantified and two-way ANOVA with Bonferroni post hoc (n = 5). P < .017 represents contribution of RAGE-KO genotype to decreased SHG Intensity, while P < .001 compares SHG Intensity of WT IVDs cultured in AGE-BSA supplemented media. AF, annulus fibrosus; AGEs, Advanced glycation end products; IVD, intervertebral disk; RAGE, receptor for AGEs; RAGE-KO, RAGE-knockout; SHG, second harmonic generation; WT, wildtype F I G U R E 4 AGE-BSA caused collagen damage in WT but not RAGE-KO mice. CHP analysis was performed on A, WT and B, RAGE-KO IVDs cultured in AGE-BSA or BSA control media. Blue indicates DAPI labeled cell nuclei, green represents fluorescein labeled CHP molecules incorporated into damaged collagen. Two-way ANOVA with Bonferroni multiple comparison P value is presented (n = 5). AGEs, Advanced glycation end products; CHP, collagen-hybridizing peptide; IVD, intervertebral disk; RAGE, receptor for AGEs; RAGE-KO, RAGE-knockout; WT, wildtype Figure 5B). Finally, AGE-BSA media caused no changes were detected in the amount of collagen (P > .576; Figure 5C,D).

| DISCUSSION
Developing improved methods to assess early degenerative changes to IVD collagen can advance our understanding of IVD degeneration mechanisms and identify opportunities for intervention. This study evaluated the role of AGEs and RAGE in driving early IVD degeneration processes in mice. We characterized IVD collagen by applying multiphoton microscopy with SHG on IVD thin sections and in situ explants along with CHP and image processing analyzes. This in vivo dietary model was previously shown to increase IVD AGE accumulation without systemic obesity or diabetes. 15 The current study identified AF collagen damage at the molecular and fibrillar levels, and collagen loss. Organ culture results demonstrated that these collagen molecular and fibrillar changes were IVD specific, not requiring the presence of an immune system or other factors present in the in vivo model. Furthermore, since RAGE-KO was protective of most collagen changes, we conclude that the observed collagen structural damage was RAGE dependent and suggested that these effects are due to catabolic processes in addition to crosslinking.
Dietary AGEs as well as diabetes in mice can cause AGE accumulation in IVDs and endplate with a shift toward catabolism and early degenerative changes. 13,43 AGE accumulation in IVD in rodent models of diabetes and H-AGE diet was further demonstrated to result in IVD AGE accumulation as well as stiffening providing further evidence of crosslinking from glycation. 14,15 This study builds on that knowledge and supports the concept of glycoxidation by demonstrating H-AGE diet and AGE supplemented IVD organ culture causes increased catabolism with AF collagen degradation, fibril disruption and loss of total collagen that is RAGE dependent (Figure 6). This study, together with the literature therefore indicates that IVD AGE exposure, as might be expected from H-AGE diets or diabetes, can play a causal role in early IVD degeneration. It is interesting to consider that severe AF damage and degeneration, as can occur in mice from severe puncture injuries, can result in reduced IVD stiffness from severe AF F I G U R E 5 AGE-BSA caused disrupted collagen fibril structure in WT but not RAGE-KO mice but no changes in percent area collagen. Morphometric analysis of A, Intensity equalized SHG images and B, Binarized in situ SHG images of mouse central anterior AF were used to quantify, C, SHG entropy as a measure of fibril disruption, and D, Percent area of SHG signal as a measure of total amount of collagen. Two-way ANOVA with Bonferroni multiple comparison P value is presented (n = 5). AF, annulus fibrosus; AGEs, Advanced glycation end products; RAGE-KO, RAGE-knockout; SHG, second harmonic generation; WT, wildtype disruption in contrast to the literature showing IVD AGE accumulation can result in increased IVD stiffening. [44][45][46] We therefore believe the AF damage is relatively mild from this dietary intervention and organ culture model and did not progress to the point where collagen damage dominates crosslinking to reduce stiffness. Nevertheless, the AF damage observed was detectable and likely to accumulate and predispose to accelerated IVD degeneration.
AGEs have been previously shown to stimulate expression of inflammatory catabolic enzymes MMP1, MMP3, and MMP9 in IVD cells. [47][48][49] AGEs also cause oxidative stress and proinflammatory cytokine production, 12 which are known drivers of IVD degeneration. [50][51][52] Increased carbonylation (a form of glycation) can also directly cause structural disruption to molecular collagen, increasing susceptibility to degradation by MMPs. 53 However, in the absence of initial molecular damage to collagen, glycation formation has also been shown to protect collagen from degradation via MMPs. 54 Since IVD collagen degradation and fiber disruption occurred both in vivo as well as in an ex vivo model, we conclude that these crosslinking and catabolic processes occurred in an IVD specific way, due in part to IVD cells in these living models. Furthermore, RAGE-KO protected IVDs from collagen damage, which supports that this is a RAGE-dependent process.
This study demonstrated a direct role of AGE formation and RAGE in causing IVD cells to adopt this degenerative phenotype from matrix catabolism (Figure 7). This study developed in situ SHG imaging methods as a step Texture analysis is a group of second order parameters that consider the intensity of individual pixels with respect to the pixels in its nearest neighborhood. Together this group of parameters is known as the gray level co-occurrence matrix (GLCM) which includes the entropy parameter included in this work. The other GLCM parameters are energy, inertia, inverse difference moment, and correlation. 58 While each parameter is known to give unique information about SHG texture, we used entropy in this study to assess collagen quality since it has been used to assess collagenous changes at the fibril level, it was sensitive during initial analysis of these parameters, and it is a readily available calculation in the Matlab image analysis toolbox. 28,29 Previous SHG image analysis techniques in the IVD have focused on the orientation of collagen fibrils throughout the AF under hyperphysiological loading. 19,20 This work represents the first time that texture analysis was utilized to characterize fibril disruption in the AF. A potential avenue of future work is the utilization of texture analysis on the spatial distribution of fibril orientations, which may aid in further characterizing early degenerative changes that are yet too subtle F I G U R E 6 Summary of in vivo dietary model and ex vivo organ culture results for WT and RAGE-KO mice. Results show a RAGE-dependent link between AGEs and collagen disruption at the molecular and fibril levels. AF, annulus fibrosus; AGEs, Advanced glycation end products; CHP, collagenhybridizing peptide; RAGE, receptor for AGEs; RAGE-KO, RAGE-knockout; SHG, second harmonic generation; WT, wildtype to be measured through means such as birefringence distributions or semiquantitative scoring.
In vivo and ex vivo organ culture results yielded similar findings with more severe alterations occurring in vivo. The degree of collagen loss observed in vivo was modest, and was not observed in the ex vivo culture. We attribute the greater severity of collagen damage in vivo to the substantially longer duration, which allowed damage to progress with subtle loss of collagen content. However, it is possible that these differences are also attributable to the additional systemic effects or biomechanical factors in vivo. Furthermore, potential differences in sensitivity are possible across thin section imaging vs in situ SHG imaging methods. For CHP measurements, the imaging parameters were different between the in vivo and ex vivo experiments in order to prevent oversaturation, and therefore it is best to compare CHP intensity values as relative within each experiment and not an absolute value. Nevertheless, even with some differences across model systems, many similarities were observed strengthening our confidence in our conclusions, while also showing that organ culture studies can be used to identify a mechanistic role for RAGE.
In addition, we observed a significant difference in SHG intensity in ex vivo control culture conditions when comparing WT and RAGE-KO IVDs ( Figure 3C). This baseline difference suggests a role for RAGE in the deposition and/or maturation of collagen in the AF, the implications of which remain unknown. Additional studies are required to investigate the role of RAGE in mediating the baseline structure of IVD collagen. Both sexes were assessed in Aim 1 and only males in Aim 2. Sex did not significantly contribute to the AF changes based on this statistical analysis. Since no sex differences were identified in our Aim 1 studies, we used a single sex for organ culture studies to reduce potential variability due to size and other factors. While this study is limited since we did not directly test for effects of AGEs on female IVDs in organ culture, we believe it is likely that females would also exhibit effects of AGES since female mice were reported to exhibit greater sensitivity than males in other spinal changes from dietary AGEs in our prior study. 15 Interestingly, this in vivo cohort showed sex-dependent effects on spine the biomechanical function. Therefore this study helps to clarify that the local effects of AGEs on AF collagen are likely not sex-dependent, while other spinal tissues (eg, vertebrae) and properties do have sexdependent effects of AGEs.
In conclusion, there is a causal relationship between AGEs, RAGE, and collagen degradation in the mice IVDs. The importance of the AGE-RAGE axis in IVD degeneration shown in this study may inform potential therapies for IVD degeneration in conditions where AGE accumulation is implicated, such as aging and diabetes. This study also demonstrated a method of in situ multiphoton microscopy in the mice IVD with quantifiable image analysis techniques for assessment of collagen quality. Collagen damage was also demonstrated with the novel CHP with direct imaging. These multiphoton microscopy and image analysis methods may be used in the assessment of collagen quality to further understand several degenerative cascades in the IVD.