Evaluation of enzymatic proteoglycan and collagen degradation in human articular cartilage using ultrashort echo time (UTE)-based biomarkers: a feasibility study

Abstract

Objective:

To investigate the feasibility of quantitative 3D ultrashort echo time (UTE)-based biomarkers in detecting proteoglycan (PG) loss and collagen degradation in human cartilage.

Methods:

A total of 104 cartilage samples were harvested for a trypsin digestion study (n=44), and a sequential trypsin and collagenase digestion study (n=60), respectively. 44 cartilage samples were randomly divided into a trypsin digestion group (tryp group) and a control group (PBS group) (n=22 for each group) for the trypsin digestion experiment. The remaining 60 cartilage samples were equally divided into four groups (n=15 for each group) for sequential trypsin and collagenase digestion, including PBS+Tris (incubated in PBS, then Tris buffer solution), PBS+30U col (incubated in PBS, then 30U/ml collagenase (30U col) with Tris buffer solution), tryp+30U col (incubated in trypsin solution, then 30U/ml collagenase with Tris buffer solution), and tryp+Tris (incubated in trypsin solution, then Tris buffer solution). The 3D UTE-based MRI biomarkers included T₁, multi-echo T₂*, adiabatic T_1ρ (AdiabT_1ρ), magnetization transfer ratio (MTR), and modeling of macromolecular proton fraction (MMF). For each cartilage sample, UTE-based biomarkers (T₁, T₂*, AdiabT_1ρ, MTR, and MMF) and sample weight were evaluated before and after treatment. PG and hydroxyproline assays were performed. Differences between groups and correlations were assessed.

Results:

All evaluated biomarkers were able to differentiate between healthy and degenerated cartilage in the trypsin digestion experiment, but only T₁ and AdiabT_1ρ were significantly correlated with the PG concentration in the digestion solution (p=0.004 and 0.0001, respectively). In the sequential digestion experiment, no significant differences were found for T₁ and AdiabT_1ρ values between PBS+Tris and PBS+30U col groups (p=0.627 and 0.877, respectively), but T₁ and AdiabT_1ρ values increased significantly in the tryp+Tris (p=0.031 and 0.024, respectively) and tryp+30U col groups (both p<0.0001). Significant decreases in MMF and MTR were found in the tryp+30U col group compared with the PBS+Tris group (p=0.002 and 0.001, respectively).

Conclusion:

AdiabT_1ρ and T₁ have the potential for detecting PG loss, while MMF and MTR are promising for the detection of collagen degradation in articular cartilage, which could all facilitate earlier, non-invasive diagnosis of osteoarthritis.

Introduction

Osteoarthritis (OA), one of the most common joint diseases of the musculoskeletal system, is a multifactorial disease characterized primarily by degeneration and loss of hyaline articular cartilage [1]. Hyaline articular cartilage is a dense connective tissue composed of chondrocytes and an extracellular matrix (ECM) that includes water, a collagen fibril network, and proteoglycan (PG) aggregates [2]. The onset of OA is associated mainly with biochemical alterations in the cartilage, with the PG loss in articular cartilage hypothesized to be an initiating factor [3, 4]. During these early stages of OA, collagen quantity has not been shown to be severely affected [3, 5]. While PG loss can be reversed [6, 7], it has been suggested that breakdown of the collagen framework during later stages of the disease marks the point of no return in OA progression [8]. Noninvasive detection of PG loss and collagen degradation by imaging would help early-stage diagnosis [9], making timely treatment of the disease while it is still reversible a clinical possibility.

Unfortunately, the most widely used imaging modality in clinical practice for the evaluation of OA is plain radiography which can only depict the gross osseous changes and narrowing of joint space that occur in the late stages of OA [10]. While conventional magnetic resonance imaging (MRI) allows for better evaluation of morphological changes in cartilage with high tissue contrast, it still has some limitations. Articular cartilage, especially the deep radial and calcified layers, contains a significant amount of components with short T₂ relaxation times (<10 milliseconds), where the signal decays so quickly that little or no signal can be detected with conventional MRI sequences [11]. Previous studies have shown that quantitative MRI techniques based on Cartesian k-space trajectories with echo times (TEs) of several to tens of milliseconds, including T₁-, T₂-, and magnetization transfer (MT)-weighted imaging, are inconclusive in detecting early-stage changes in OA [12, 13]. Some ex vivo studies have indicated that T_1ρ may increase with PG loss and that T₂* may increase with collagen degradation [5, 14-17], but these studies were performed on high-field MR scanners, namely 9.4T and 4.0T, which are not typically used for human studies.

Three-dimensional ultrashort echo time (3D UTE) MRI sequences with TEs on the order of 10 microseconds allow for direct volumetric imaging and quantitative assessment of cartilage, including the deep layers, on clinical MR scanners [18-20]. In addition, a series of quantitative UTE MRI techniques has been developed, including actual flip angle imaging for T₁ mapping, fat-suppressed multi-echo T₂* mapping, adiabatic T_1ρ (AdiabT_1ρ), magnetization transfer (MT) imaging for magnetization transfer ratio (MTR), and modeling of macromolecular proton fraction (MMF) [21-25]. However, the capability of these UTE-based quantitative biomarkers in evaluating PG loss and collagen degradation in cartilage has not yet been systematically assessed.

The purpose of our study was to investigate whether quantitative 3D UTE-based biomarkers are sensitive to PG loss and collagen degradation induced enzymatically in human cartilage and to quantitatively evaluate the correlation between UTE-based biomarkers and biochemical components including PG and collagen to determine the specificity of these biomarkers in cartilage imaging.

Materials and Methods

A total of 104 healthy cartilage samples were harvested for a trypsin digestion study (n=44), and a sequential trypsin and collagenase digestion study (n=60), respectively. Trypsin is an enzyme that partly degrades PG while collagenase is an enzyme that degrades the collagen network. Collagenase may have deeper activity after PG removal as it’s too large to diffuse through the PG network. Several parallel experiments were performed and 12 or 16 cartilage samples were stacked in a 10 ml syringe in each experiment (Figure S1).

Sample Preparation and Trypsin Digestion

A total of 44 cartilage samples was harvested from three cadaveric specimens (two males, one female; 32.0±5.6 years old) using an osteochondral autograft transplant system (OATS) (catalog no. 10mm-AR-1981-10S; Arthrex, Naples, FL). The osseous components were removed using a scalpel. Before the baseline MRI scan, the cartilage samples were soaked in phosphate-buffered saline (PBS) for one hour to 1) reverse sample dehydration that had occurred during storage and preparation procedures, and 2) prevent water absorption when samples were later immersed in PBS or enzyme solutions. Samples were randomly divided into the enzyme group (tryp group) or the control group (PBS group) (n=22 for each group). Before the digestion step, all cartilage samples were weighed using a high-precision digital laboratory scale (AG245; Mettler Toledo, Columbus, OH) and underwent baseline scanning. Samples in the enzyme group were then incubated in a 3-ml 1mg/ml trypsin (catalog no. T1426; Sigma-Aldrich, St. Louis, MO)-in-PBS solution at 37°C for three hours to induce PG loss, while the control samples were immersed in 3-ml PBS solution at 37°C for three hours. At the end of the three-hour digestion period, samples were removed from the incubator and allowed to cool at room temperature for 30 minutes because T₁ relaxation measurements are dependent on temperature [26]. Finally, samples were scanned in the same order as at baseline. Surface water on the specimens was gently wiped away and the weight of each cartilage sample was measured and recorded once again. A diagram of the procedure is shown in Figure 1a.

Figure 1.

Experimental procedures of trypsin and sequential trypsin-collagenase digestion.

Sample Preparation and Sequential Trypsin-Collagenase Digestion

A total of 60 cartilage samples were harvested from five donors (four males, one female; 46.2±11.2 years old). The cartilage samples were soaked in PBS and inhibitor solutions, including E-64 (catalog no. S7379; SelleckChem, Houston, TX), 4-(2-aminoethyl)-benzenesulfonyl fluoride (AEBSF) (catalog no. S7378; SelleckChem), and Ilomastat/GM6001 (catalog no. S7157; SelleckChem) for two hours to rehydrate the samples and prevent autolysis. Samples were then rinsed in PBS solution for one hour to stop the inhibitors, weighed, and underwent baseline MR scanning. Next, 30 samples were incubated in a 3-ml 1mg/ml of trypsin-in-PBS solution (tryp group) and the remaining 30 samples were incubated in PBS at 37°C for three hours (PBS group), similar to the trypsin digestion procedure described above. After samples had cooled at room temperature for 30 minutes, samples were weighed again, and a post-digestion MR scan was performed. After scanning, samples were rinsed in PBS for one hour.

Next, the tryp group and the PBS group were divided to form four subgroups (n=15 samples each) for subsequent Tris and collagenase incubation. One tryp subgroup and one PBS subgroup were incubated in 30U/ml collagenase (30U col) (catalog no. C0773; Sigma-Aldrich) with Tris buffer solution made from 50 mM Tris buffer and 2 mM calcium acetate buffer (termed tryp+30U col and PBS+30U col groups, respectively). The remaining tryp and PBS subgroups were incubated in only Tris buffer solution at 37°C for 24 hours (tryp+Tris and PBS+Tris, respectively). Tris buffer solution will not have profound effect on PG or collagen content of cartilage. Following this incubation period, samples in all four groups were weighed for the third time and underwent MRI scanning.

A diagram of the procedure is shown in Figure 1b.

MRI Scanning

Cartilage samples were stacked in a 10-ml syringe filled with perfluoropolyether (Fomblin; Ausimont, Thorofare, NJ) and this particular order was maintained across scans (Figure S1). In addition, a small notch was carved into the samples to maintain orientation across scans. All imaging was performed on a 3T clinical MRI scanner (MR750; GE Healthcare Technologies, Milwaukee, WI) using a homemade birdcage coil (25.4 mm inner diameter) for both RF transmission and signal reception. The basic 3D UTE sequence employs a unique data sampling trajectory scheme that samples MRI data along twisting paths on evenly spaced cone surfaces [27]. The following four imaging protocols were performed: A) 3D UTE with actual flip angle imaging and variable flip angles (3D UTE AFI-VFA) with four flip angles (FAs) of 5°, 10°, 20°, and 30°) and a TR of 20 ms [24] for accurate T₁ measurement; B) fat-suppressed 3D multi-echo UTE imaging with six TEs of 0.032, 4.1, 8.1, 12.1, 16.1, and 32 ms for single-component T₂* measurement; C) 3D UTE with adiabatic T_1ρ preparation for AdiabT_1ρ measurement with spin-locking times (TSLs) of 0, 12, 24, 36, 48, 72, and 96 ms [22]; and D) 3D UTE-MT imaging with three saturation pulse powers (θ = 400°, 600°, and 800°) and five frequency offsets (Δf = 2, 5, 10, 20, and 50 kHz) to measure the MTR and MMF [28]. Other imaging parameters included: field of view (FOV) = 5 cm, matrix = 160×160, and slice thickness = 0.5 mm, 60 slices. The total scan time was 78 minutes.

Biochemical Analysis

After MR imaging, the glycosaminoglycan (GAG) concentration in the digestion solutions was determined by the dimethyl-methylene blue (DMMB) method [29] using a standard curve of chondroitin 4-sulfate from bovine cartilage (catalog no. C9819; Sigma-Aldrich). Absorbance was measured at 525 nm using a plate reader (SpectraMax 340PC; Molecular Devices, San Jose, CA). To measure the collagen content of all the samples, a standard hydroxyproline assay (catalog no. K555-100; BioVision, Milpitas, CA) was used to measure hydroxyproline, an indicator of collagen quantity [30]. All samples were spectrophotometrically quantified at 560 nm absorbance.

Data analysis

The analysis algorithm was written in MATLAB (The MathWorks Inc., Natick, MA) using the Levenberg-Marquardt method for non-linear least-squares curve fitting and was executed offline on DICOM images obtained by the protocols described above. Regions of interest (ROIs) were manually drawn on the first UTE image of each series, then copied to each of the subsequent images. The mean intensity within each ROI was used for curve fitting. T₁ was analyzed from the 3D UTE-AFI and UTE-VTR sequences using single-component fitting [29]. T_1ρ values were obtained using a fitting algorithm that was developed in-house and based on the equation, $\mathrm{S}(\text{TSL})\propto \frac{e^{-TSL∕T_{1\rho }}\times (1-e^{-(TR-TSL)∕T_1})}{1-e^{-TSL∕T_{1\rho }}\times e^{-(TR-TSL)∕T_1}\cos (\alpha )}\times \sin (\alpha )+C$ [22]. The UTE-MT data set was analyzed for MTR [28] and MMF based on a two-pool model [31]. A single-component fitting model, (S(TE) ∝ exp (−TE/T2*) + C), was utilized for T₂* decay analyses [32].

Three consecutive slices at the center of each sample were used for global ROI analysis (Figure S1). T₁, T₂*, AdiabT_1ρ, MTR, and MMF values were calculated for all cartilage samples before and after enzymatic digestion. The figures of non-digested and sequentially digested cartilage samples for T₁, T₂*, AdiabT_1ρ, and MT were shown in Figure S2.

Statistical analysis

SPSS software (v. 24; IBM, Armonk, NY) was used for statistical analysis, with different tests used to analyze the initial trypsin digestion experiment and the sequential trypsin-collagenase experiment.

Kolmogorov-Smirnov test was performed to determine the normality and the data are all normally distributed. Two-sided paired t-tests were used to compare variations between the quantitative UTE biomarkers and sample weights before and after digestion in the initial trypsin experiment. Independent t-tests were used to compare the differences in digestion ratio between these the tryp and PBS groups, while Pearson correlation coefficients were used to evaluate the relationship between UTE biomarkers and biochemical quantification of PG concentration.

For the sequential trypsin-collagenase digestion experiment, ANOVA and LSD (least significant difference) tests were used to compare differences between the UTE biomarkers, digestion ratios, and sample weights among the four subgroups that included tryp+30U col, PBS+30U col, tryp+Tris, and PBS+Tris. A p-value less than 0.05 was considered a significant difference.

Results

Figure 2 shows the differences in weight and quantitative UTE-based biomarkers, including T₁, T₂*, AdiabT_1ρ, MTR, and MMF, for trypsin -treated and control groups before and after digestion. For the control group, there were no significant differences in the weights of cartilage samples (p=0.78) nor in the T₁ (p=0.106), AdiabT_1ρ (p = 0.892), MTR (p = 0.316), or MMF (p = 0.406) values after soaking in PBS for three hours (all p-values>0.05). For the cartilage samples in the enzyme digestion group, sample weights decreased significantly (p<0.0001), T₁ and AdiabT_1ρ increased significantly (both p<0.0001), and MTR (p<0.0001) and MMF (p=0.005) decreased significantly after trypsin incubation. T₂* values decreased significantly in the control group and increased significantly in the enzyme-treated group after digestion (p=0.014 and 0.005, respectively).

Figure 2.

The differences of weights and quantitative UTE-based biomarkers, including T₁, T₂*, AdiabT_1ρ, MTR, and MMF for the control and enzyme-treated (trypsin) groups before and after digestion are shown in a through f, respectively. For the control group, the weights of cartilage samples and T₁, AdiabT_1ρ, MTR, and MMF values are not significantly different after soaking in PBS for three hours (all p-values>0.05). For the trypsin group, weights significantly decrease after enzymatic digestion. T₁ and AdiabT_1ρ increase significantly, while MTR and MMF decrease significantly after digestion. T₂* decreases significantly in the control group but increases significantly in the enzyme-treated group after digestion. (n.s., non-significant; *, 0.01<p<0.05; **, 0.001<p<0.01; ***, 0.0001<p<0.001; ****, p<0.0001)

The ratio differences of weights and quantitative UTE-based biomarkers are shown in Figure S3.

Figure 3 displays the correlations of the change ratios of quantitative UTE biomarkers after digestion with the PG concentrations found in the digestion solutions, which were standardized according to the weight of the samples. The change ratios of T₁ (p=0.004) and T_1ρ (p= 0.0001) were positively correlated with the PG concentration (r=0.74 and 0.86, respectively). No significant correlations were found between the PG concentrations and the change ratios of T₂* (p=0.058), MTR (p=0.061), or MMF (p=0.111).

Figure 3.

The correlations of the change ratio of quantitative UTE biomarkers including T₁, T₂*, AdiabT_1ρ, MTR, and MMF after digestion with the PG concentration in the digestion solutions, which was standardized with cartilage weights, are shown in a, b, c, d, and e, respectively. The change ratios of T₁ and AdiabT_1ρ are positively correlated with the PG concentration (p<0.0001) and the correlation coefficients are 0.74 and 0.86, respectively. No significant correlations are found between the change ratios of T2*, MTR, and MMF with the PG concentration (all p-values>0.05).

Figure S4 shows the weights of the four cartilage sample groups that were part of the sequential trypsin-collagenase digestion experiment, including PBS+Tris, PBS+30U col, tryp+30U col, and tryp+Tris. There was no significant effect on cartilage weights after soaking in the PBS+Tris or PBS+30U col groups after initial soaking in PBS for 3 hours (p=0.054 and 0.582, respectively), nor was there any significant effect on weights in the PBS+Tris or tryp+Tris groups after secondary soaking in Tris buffer solution for 24 hours (p=0.189 and p=0.098, respectively), Significant weight loss occurred after both the tryp+30U col and PBS+30U col groups’ digestions in 30U collagenase for 24 hours (both p<0.0001) and after the tryp+30U col and tryp+Tris groups’ initial digestion in trypsin for 3 hours (both p<0.0001).

Figure 4 shows the changes in quantitative UTE-based biomarkers throughout sequential processing steps for the four subgroups involved in the sequential trypsin-collagenase experiment. For T₁ and AdiabT_1ρ, there was no significant difference between the PBS+Tris and PBS+30U col groups (p=0.627 and 0.877, respectively), whereas T₁ and AdiabT_1ρ values increased significantly in the tryp+Tris (p=0.031 and 0.024, respectively) and tryp+30U col groups (both p<0.0001), with the latter group showing more obvious changes. The MTR values of the PBS+30U col and tryp+Tris groups were not significantly different from the values of the PBS+Tris group (p=0.335 and 0.086, respectively). Similarly, the MMF values of the PBS+30U col and tryp+Tris groups were not significantly different from the values of the PBS+Tris group (p=0.582 and 0.083, respectively), which was considered a control. However, MTR and MMF values decreased significantly in the tryp+30U col group (p=0.002 and 0.001, respectively). No significant differences in T₂* values were observed among groups.

Figure 4.

The changes of quantitative UTE-based biomarkers including T₁, T₂*, AdiabT_1ρ, MTR, and MMF after the sequential digestions in four different groups are shown in a, b, c, d, and e, respectively. For T₁ and AdiabT_1ρ, there are no significant differences between the PBS+Tris and PBS+30U col groups, while T₁ and AdiabT_1ρ values significantly increase in the tryp+Tris and tryp+30U col groups, with the latter group being more obvious. The MTR and MMF values of the PBS+30 U col and tryp+Tris groups are not significantly different from the values of the PBS+Tris group, which is regarded as a control, but they decrease significantly in the tryp+30U col group. There are no significant differences in T₂* values observed among groups. (n.s., non-significant; *, 0.01<p<0.05; **, 0.001<p<0.01; ***, 0.0001<p<0.001; ****, p<0.0001)

The collagen digestion ratios of the four different groups, calculated as the hydroxyproline concentration in the digestion solution divided by the total concentration of hydroxyproline in both solution and samples are shown in Figure S5; p-values are listed in Table 1.

Table 1.

Collagen digestion ratios of four different groups.

Hydroxyproline Concentration(ug/ml)	Digestion Ratio, %	P-Values (LSD)
PBS+Tris	−0.02	-
PBS+30U col	18.94	<0.0001
tryp+Tris	1.50	0.643
tryp+30U col	40.30	<0.0001
P-value (ANOVA)	--	<0.0001

Collagen digestion ratios calculated by the hydroxyproline concentration in the digestion solution divided by the total concentration of hydroxyproline in both solution and samples. P-values were calculated by the ANOVA and LSD tests. Of note, the p-values of LSD tests listed here were the results compared with the PBS+Tris group.

Discussion

UTE MRI techniques have advantages in detecting signals from tissues with short T₂ relaxation times in the musculoskeletal system. UTE-based quantitative sequences allow for volumetric imaging of the whole cartilage, including the deep radial layer with short T₂ relaxation, potentially leading to a more precise and sensitive assessment of cartilage degeneration. This is the first study to systematically investigate the capability and potential sensitivity of quantitative UTE-based biomarkers to detect PG loss and collagen degradation. The results are likely to help in understanding the role of each UTE-based biomarker in the diagnosis of OA at its early stages.

The results of the trypsin digestion experiment demonstrate that PG loss was successfully induced (Figure S6). Sample weights and UTE-based biomarkers including T₁, AdiabT_1ρ, MTR, and MMF demonstrated no significant changes after soaking in PBS for three hours in the control group, indicating that PBS alone affected neither weights nor T₁, AdiabT_1ρ, MTR, and MMF values. On the other hand, significant increases in T₁ and AdiabT_1ρ, as well as a decrease in MTR and MMF, were observed after incubation in trypsin compared to baseline, which was consistent with previous studies that reported T₁ and AdiabT_1ρ increase with PG loss [5, 15, 33]. Lin et al. found that T₁ demonstrated the ability to discriminate between normal and trypsin-degraded cartilage with high sensitivity and specificity [33]. Although there is a decreasing trend in MTR and MMF with trypsin digestion, the correlation evaluation showed that only the change ratios of T₁ and AdiabT_1ρ were positively correlated with the PG concentration (p<0.0001), indicating that T₁ and AdiabT_1ρ may be more sensitive to PG loss in cartilage than other UTE biomarkers. For T₂*, both control and enzyme-treated samples showed significant changes after soaking in PBS or digestion with trypsin. The differences of change ratios (before vs. after) in weights and all imaging biomarkers between the control and enzyme-treated groups were statistically significant, demonstrating that all of the UTE biomarkers can differentiate healthy cartilage from tissue experiencing PG loss.

The weights of cartilage did not change significantly after soaking in PBS or Tris buffer solution during the sequential digestion experiment, but both trypsin and collagenase resulted in weight loss regardless of the point at which they were added, which was consistent with the trypsin digestion experiment. A previous study confirmed that the liquid mass of cartilage remained steady when rehydrated, then reached an equilibrium [34]. This is consistent with our observation that cartilage weights stayed the same after soaking in PBS or Tris buffer solution for several hours.

There were no significant differences in T₁ or AdiabT_1ρ values between the PBS+Tris and PBS+30U col groups, but T₁ and AdiabT_1ρ values increased significantly in the tryp+Tris and tryp+30U col groups, indicating that trypsin alone, collagenase alone, and trypsin in combination with collagenase can cause changes in T₁ and AdiabT_1ρ. More obvious changes were observed after digestion in both trypsin and collagenase compared to trypsin alone, which can be explained by the fact that collagenase causes PG loss in addition to collagen degradation once it has diffused through the GAG network, consequently destroying the collagen framework [16, 35, 36]. There were MTR and MMF decreases in the PBS+30U col and tryp+Tris groups, but the differences did not reach statistical significance, whereas, in the tryp+30U col group, values were significantly lower compared to the PBS+Tris group. This non-significant decreasing trend was similarly observed in the trypsin digestion experiment and may be related to the addition of inhibitors in the sequential digestion experiment.

The biochemical results indicated that trypsin was able to digest PG without a profound influence on collagen, but that it could also make collagenase work more efficiently. The molecular weight of collagenase is around 63-130 kDa and it may be too large to diffuse through the GAG network of cartilage. However, following PG removal by trypsin, collagenase may be able to diffuse better [37-39], a notion that is supported by our results in the sequential digestion experiment where the collagen digestion ratio of the PBS+30U col group was 18.94% and tryp+30U col was 40.30%. Previously, Stenman et al. incubated bovine trypsin and human trypsin-2 directly with type II collagen substrate and found that bovine trypsin did not cleave type II collagen, whereas human type-2 trypsin could degrade the collagen network [40]. In our experiment, we digested the whole cartilage sample with trypsin, a methodology that may more closely resemble the natural degradation process of cartilage. The UTE-based biomarkers showed greater changes when the collagen digestion ratio was over 40% and exhibited minimal changes when the collagen digestion ratio was less than 18.94%. Furthermore, it seems that both collagen and PG contributed to MTR and MMF. Since there is much more collagen than PG in cartilage, collagen would be expected to make a greater contribution to MTR and MMF than PG.

There were some limitations in our study. First, this was an ex vivo study and the conclusions remain to be confirmed for in vivo studies. Second, the PG concentration in the digestion solution as opposed to the digestion ratio of PG was used to correlate with the changes in UTE-based biomarkers, which may have led to overestimation or underestimation of the correlation coefficients. However, we standardized the PG concentration with individual cartilage weights in order to minimize the effect of sample size. Third, the experimental conditions were slightly different between the two parts of the experiment (i.e., trypsin digestion and sequential trypsin-collagenase digestion). Inhibitors were only used in the sequential digestion experiment (as a means to prevent cartilage autolysis, as the incubation time for collagenase was much longer than trypsin), which may have affected our results. However, the overall changing trends of quantitative biomarkers were consistent. Finally, histological and immunohistochemical techniques were not employed in this study but might have been beneficial in demonstrating the spatial variation across cartilage samples. Especially for collagen digestion, not only the loss of collagen but also re-orientation of the collagen fibers may have contributed to the total quantitative parameter change.

In conclusion, UTE T₁ and AdiabT_1ρ relaxation times have the potential for detecting PG loss, while MTR and MMF are promising biomarkers for detecting collagen degradation in articular cartilage, which could all facilitate earlier, non-invasive diagnosis of osteoarthritis.

Abbreviations used:

UTE

ultrashort echo time

PG

proteoglycan

OA

osteoarthritis

ECM

extracellular matrix

TE

echo time

AdiabT_1ρ

adiabatic T_1ρ

MT

magnetization transfer

MTR

magnetization transfer ratio

MMF

macromolecular fraction

OATS

osteochondral autograft transplant system

PBS

phosphate-buffered saline

AFI

actual flip angle

VFA

variable flip angle

FA

flip angle

TSL

spin lock time

FOV

field of view

GAG

glycosaminoglycan

DMMB

dimethyl-methylene blue

ROI

regions of interest

LSD

least significant difference