Light Absorptive Properties of Articular Cartilage, ECM Molecules, Synovial Fluid, and Photoinitiators as Potential Barriers to Light-Initiated Polymer Scaffolding Procedures

Abstract

Objective

Many in vivo procedures to repair chondral defects use ultraviolet (UV)-photoinitiated in situ polymerization within the cartilage matrix. Chemical species that absorb UV light might reduce the effectiveness of these procedures by acting as light absorption barriers. This study evaluated whether any of the individual native biochemical components in cartilage and synovial fluid interfered with the absorption of light by common scaffolding photosensitizers.

Materials

UV-visible spectroscopy was performed on each major component of cartilage in solution, on bovine synovial fluid, and on four photosensitizers, riboflavin, Irgacure 2959, quinine, and riboflavin-5′-phosphate. Molar extinction and absorption coefficients were calculated at wavelengths of maximum absorbance and 365 nm. Intact articular cartilage was also examined.

Results

The individual major biochemical components of cartilage, Irgacure 2959, and quinine did not exhibit a significant absorption at 365 nm. Riboflavin and riboflavin-5′-phosphate were more effectual light absorbers at 365 nm, compared with the individual native species. Intact cartilage absorbed a significantly greater amount of UV light in comparison with the native species.

Conclusion

Our results indicate that none of the individual native species in cartilage will interfere with the absorption of UV light at 365 nm by these commonly used photoinitiators. Intact cartilage slices exhibited significant light absorption at 365 nm, while also having distinct absorbance peaks at wavelengths less than 300 nm. Determining the UV absorptive properties of the biomolecules native to articular cartilage and synovial fluid will aid in optimizing scaffolding procedures to ensure sufficient scaffold polymerization at a minimum UV intensity.

Introduction

Articular cartilage is an inhomogeneous biomolecular tissue characterized by 4 primary layers or subdivisions: (1) the superficial or tangential zone, (2) the middle or transitional zone, (3) the deep or radial zone, and (4) the underlying calcified cartilage zone.^1,2 The microstructural configuration of each of these layers varies little among different species and different joints within a single species.³ The major biochemical components of human and animal articular cartilage are collagen, chondroitin sulfate, aggrecan and other proteoglycans, fibronectin, and hyaluronate.^4,5 In vivo, synovial fluid, mainly composed of hyaluronic acid and lubricin, covers and adheres to the surface of the articular cartilage.

Since cartilage is an avascular tissue, the synovial fluid is the primary source of nutrients from which chondrocytes sustain themselves by molecular solute diffusion and fluid convection through the articular surface into the matrix.^6,7 When articular cartilage is damaged due to progressive arthritis or sudden trauma, the matrix is degraded and the biochemical linkages between matrix components are disrupted. Since the number of cells (chondrocytes) within the extracellular matrix (ECM) is relatively low compared with other tissues, chondrocytes lack the capacity to effectively repair the damaged matrix by synthesizing new ECM components. For this reason, artificial repair techniques are currently under investigation which could quickly and effectively treat damaged cartilage. One such tissue engineering technique, “scaffolding,” involves the development of a polymer lattice or scaffold—either porous (e.g., collagen sponge) or nonporous (e.g., hydrogel)—to replace or reinforce the damaged cartilage ECM.^8-11 This lattice is often assembled in situ by diffusing or injecting solubilized monomers or uncrosslinked polymers containing an ultraviolet (UV) photosensitizer into the damaged matrix or open defect, and then exposing the solution and cartilage to UV light to initiate the photosensitizer and crosslinking process until the polymer is crosslinked into a scaffold. The goal of this scaffolding process is to restore the structural and mechanical characteristics of the cartilage’s ECM and help facilitate the healing process.^7,10,12-15

As light passes through organic matter a percentage of that light is absorbed and the intensity of the transient radiation decreases. This is particularly true when UV and visible light traverse matter containing conjugated electron systems.^16,17 It follows that articular cartilage, a biphasic tissue composed of various organic compounds and water, will absorb UV radiation to some degree, as previously found.^18,19 Thus, light absorption is an important consideration for in situ formation and integration of cartilage scaffolding. UV-initiated scaffold assembly depends on the effective absorption of photons by photosensitizers that have been incorporated within the scaffold matrix. Other absorbing biomolecules, such as those within cartilage and synovial fluid, can interfere with the UV transmission and absorption of UV photons by these photosensitizers. Significant interference will hinder the effectiveness of the procedure and may require higher doses (intensity and time) of UV radiation to obtain the desired scaffolding results. This is problematic because minimum UV intensity should be used in these procedures to prevent excessive damage to the surrounding tissue and cells. Determining the UV absorptive properties of the biomolecules native to articular cartilage and synovial fluid will aid in optimizing scaffolding procedures to ensure sufficient scaffold polymerization at a minimum UV intensity.

The objective of this study was to determine the UV light absorption properties of the major biochemical species present in cartilage and synovial fluid at 365 nm, the UV wavelength most commonly used in physiological photocrosslinking procedures, and compare them to the absorption properties of some commonly used photosensitizers in these procedures. We analyzed the ultraviolet-visible (UV-Vis) absorbance spectra for collagen types 1 and 2, chondroitin sulfate, sodium hyaluronate, fibronectin, proteoglycan, and synovial fluid in neutral, buffered solution. For each of these species the molar extinction coefficient and the absorption coefficient were calculated at 365 nm as well as at the wavelengths of peak absorption for a range of physiological concentrations. This process was then repeated for 4 photosensitizers: riboflavin, Irgacure 2529 (2-hydroxy-1-[4-(hydroxyethoxy)phenyl]-2-methyl-1-propanone), quinine, and a more water-soluble form of riboflavin, riboflavin-5′-phosphate. The first 2 of these are among the most cytocompatible photosensitizers commonly used in scaffolding procedures.^15,20 The absorptive properties of the native components were then compared with those of the photosensitizers to evaluate whether any components would be likely to interfere with the polymerization process. Finally, we spectroscopically evaluated the absorption properties of intact bovine cartilage (200-600 nm) to determine the potential for cartilage as a whole to pose a barrier to 365 nm light transmission as compared to its individual components in isolation. Similar studies have been reported on equine, porcine, and rabbit cartilage but these did not examine cartilage below 300 nm.^19,21-25

Materials and Methods

General

Chondroitin sulfate A sodium salt from bovine trachea, fibronectin from bovine plasma, proteoglycan from bovine nasal septum, collagen type 2 from bovine nasal septum, riboflavin, Irgacure 2959, riboflavin-5′-phosphate, quinine (Sigma-Aldrich, St. Louis, MO); rat-tail collagen type 1 in 0.02 M acetic acid (Advanced BioMatrix, San Diego, CA); sodium hyaluronate (Lifecore Biomedical, Chaska, MN); 1X phosphate buffered saline (PBS, Mediatech, Inc., Manassas, VA); all chemicals were used as received without further purification unless otherwise noted. Synovial fluid was extracted from 2 mature bovine knee joints, once from a normal knee and once from a knee with osteoarthritis (OA). The following solution concentrations were evaluated: 0.1%, 0.2%, 0.3% and 0.4% collagen type 1 in 0.02 M acetic acid; 0.03%, 0.04%, 0.05%, 0.1%, 0.2%, and 0.3% collagen type 2 in 0.04 M acetic acid; 0.01%, 0.02%, 0.03%, 0.04%, 0.05%, and 0.1% chondroitin sulfate in 1X PBS; 0.015%, 0.02%, 0.03%, 0.04% and 0.05% sodium hyaluronate in 1X PBS; 0.005%, 0.01%, 0.02%, 0.03%, 0.04%, 0.05%, and 0.1% fibronectin in 1X PBS; 1%, 2%, 3%, 4%, and 5% synovial fluid in 1X PBS; 0.025%, 0.05%, 0.1% 0.15% 0.2%, and 0.3% proteoglycan in 1X PBS; 2.66 µM, 5.31 µM, 7.97 µM, 10.63 µM, and 13.29 µM riboflavin in 1X PBS; 4.46 µM, 8.92 µM, 13.4 µM, 17.8 µM, 22.3 µM, and 44.6 µM Irgacure 2959 in 1X PBS; 2.01 µM, 4.03 µM, 6.04 µM, 8.06 µM, and 10.07 µM riboflavin-5′-phosphate in 1X PBS; 30.8 µM, 61.6 µM, 123.3 µM, 154.1 µM, and 308.2 µM quinine in 1X PBS.

Concentrations were chosen based on the order of magnitude for concentrations found in articular cartilage, used for biopolymerization, or in normal joints for all species except collagen.^20,26-29 More dilute collagen concentrations had to be used to minimize the degree of gelation (typical concentration in vivo 10%-20%).³⁰ PBS was utilized as a solvent for each assay in order to mimic in vivo conditions, except for collagen type 1 and type 2.

Ultraviolet-Visible Spectrum Analysis

Quartz cuvettes were filled with the different concentrations of each sample and placed into a Perkin Elmer UV-Vis Lambda 12 spectrophotometer (Waltham, MA) for spectral analysis. The cuvettes had a 10-mm path length and 1-mm thick walls. UV-Vis absorption analysis was performed at wavelengths ranging from 200 to 750 nm. The absorbance spectra were then analyzed using Sigmaplot and OriginLab 8.5 softwares to calculate the absorbance, molar extinction coefficient, and the absorption coefficient. All spectra were collected and analyzed in triplicate.

The absorption of light by a species can be described by the absorbance (A) defined as

$$A=-\ln \left(I/I_0\right)$$

where $I_0$is the intensity of the incident light, $I$ is the intensity of the light transmitted through the species, which in our case are in solution, and I/I₀ is the transmittance (T).^31,32 The molar extinction coefficient (ε) is generally used to evaluate the absorptive behavior of a species in liquid solution and is defined by Beer’s law

$$A=\epsilon lc$$

where l is the path length or the distance light travels through the sample and c is the concentration of the solution containing the species of interest (expressed in molarity).

The transmission of light through a material is often described by the absorption coefficient (α) through a derivation of Equation (2)³³

$$\alpha =\frac{A}{d}$$

where d is the distance light penetrates into, or through, a material. The absorption coefficient can be expressed as a function of the extinction coefficient by combining Equations (2) and (3), such that

$$\alpha =\epsilon c$$

Unlike ε, α can be used to describe the light intensity attenuated by liquid and solid materials, and is a function of wavelength and concentration.³¹ In this study, $l=d$ since both represent the path length through the sample (i.e., 10 mm internal width of the cuvette).

Since cartilage is a biphasic tissue composed of solid and fluid phases (neither a pure liquid nor solid), the samples were evaluated by calculating both ε and α from Equations (2) and (4), respectively. For each species, the molar extinction coefficient, ε, as a function of concentration (M) and length (cm), was determined from the slope of a best-fit linear line to a plot of absorbance (A) versus concentration (c) at all absorption maxima and 365 nm. For comparison purposes, the absorption coefficient was calculated at reported concentrations for species in articular cartilage (3% chondroitin sulfate, 2% proteoglycan, 18% collagen, and 0.0002% fibronectin) and for each photosensitizer at its aqueous solubility saturation limit as specified by the specification documents or cited reference (2.25 × 10⁻⁴ M riboflavin, 3.39 × 10⁻² M Irgacure 2959, 0.201 M riboflavin-5′-phosphate, 1.54 × 10⁻³ M quinine) from Equation (4).^29,34-36

Amino acid sequences for collagen types 1 and 2 from bovine and rat were obtained using the ProtParam tool from the SIB Swiss Institute of Bioinformatics ExPASy Bioinformatics Resource Portal (http://wed.expasy.org/protparam/). Access numbers: bovine α1(II) collagen type 2, P02459; bovine α1(I) collagen type 1, P02453; bovine α2(I) collagen type 1, P02465; rat α2(I) type 1 collagen, P02466; rat α1(I) type 1 collagen, P02454.

Cartilage Slices

Three mature bovine knees were obtained from a local abattoir. Full-thickness 5-mm diameter plugs of articular cartilage were removed from the femoral trochlear notch, frozen, and later thawed to room temperature (n = 3; 1 plug harvested from each knee). The plugs were then sectioned using a freezing microtome into 100 µm thicknesses, starting from the articular surface and proceeding downward (the topmost section numbered as 1). Note that due to variability between knees the full thickness of the plugs was different (i.e., the thickness of articular cartilage from knee 1 was greater than knees 2 and 3). Seven slices were taken from each plug (cartilage thickness of plugs 2 and 3 was 700 µm), and the average absorbance calculated includes all 3 cartilage plugs.

Each slice was placed onto the side of a cuvette, inserted into the UV-Vis spectrophotometer, and spectral analysis performed from 200 to 600 nm. The data were smoothed in OriginLab using the fast Fourier transform (FFT) filter. To calculate the total absorbance of the full-thickness cartilage disk (A_T) at 365 nm, we assumed an absorbance model given by

$$A_{\mathrm{T}}=A_1+A_2+\cdots +A_n$$

where A_i is the absorbance of each slice (i) for n = 7 slices. This model allowed us to calculate the absorbance of the entire disk (n slices) as well as at any depth within the disk from the articular surface. The slope of absorbance versus depth (path length) through the disk yields an apparent total attenuation coefficient at 365 nm for the intact cartilage disk. Note that this value does not separate attenuation due to scatter and absorption.

Results

UV-Vis absorption spectroscopy from 200 to 750 nm was performed for different concentrations of collagen types 1 and 2, chondroitin sulfate, fibronectin, synovial fluid, proteoglycan, riboflavin, riboflavin-5′-phosphate, Irgacure 2959, and quinine. The absorbance spectra for each sample are shown in Figures 1 - 10 , while the insets show the absorbance versus concentration plots at local absorbance maxima. Normal and arthritic synovial fluid were analyzed in this study. The spectra of normal synovial fluid are shown in Figure 5 while the spectra for arthritic synovial fluid are shown in the Supplementary Figure S1. Note that ε was calculated from the slope of the best-fit linear regression applied to the absorbance versus concentration plot, in accordance with Beer’s law (Equation 2).

Figure 1.

Absorbance spectra of rat tail collagen type 1 in 0.02 M acetic acid solution. The inset shows absorbance versus concentration measured at 273 nm with the linear best-fit line, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

Figure 2.

Absorbance spectra of bovine collagen type 2 in 0.04 M acetic acid solution. The inset shows absorbance versus concentration measured at 263 nm with the linear best-fit line, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

Figure 3.

Absorption spectra of chondroitin sulfate in phosphate buffered saline (PBS). The inset shows absorbance versus concentration measured at 263 nm with the linear best-fit line, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

Figure 4.

Absorbance spectra of fibronectin in phosphate buffered saline (PBS). The inset shows absorbance versus concentration measured at 282 nm with the linear best-fit line, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

Figure 5.

Absorbance spectra of bovine synovial fluid in phosphate buffered saline (PBS) from a normal joint. Insets show absorbance versus concentration measured at 283 nm with the linear best-fit line, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation. See Supplementary Figure S1 for absorbance spectra of bovine synovial fluid from an arthritic joint.

Figure 6.

Absorbance spectra of proteoglycan in phosphate buffered saline (PBS). The inset shows absorbance versus concentration measured at 281 nm with the linear best-fit line, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

Figure 7.

Absorbance spectra of riboflavin in phosphate buffered saline (PBS). The inset shows absorbance versus concentration measured at local absorbance maxima with the linear best-fit lines, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

Figure 8.

Absorbance spectra of riboflavin-5′-phosphate in phosphate buffered saline (PBS). The inset shows absorbance versus concentration measured at local absorbance maxima with the linear best-fit lines, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

Figure 9.

Absorption spectra of Irgacure 2959 in phosphate buffered saline (PBS). The inset shows absorbance versus concentration measured at local absorbance maxima with the linear best-fit lines, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

Figure 10.

Absorbance spectra of quinine in phosphate buffered saline (PBS). The inset shows absorbance versus concentration measured at local absorbance maxima with the linear best-fit lines, each point represents the average of 3 measurements with the error bars as ± 1 standard deviation.

In spectrophotometry, absorbance measurements that are greater than 1 AU and less than 0.1 AU are generally deemed as inaccurate due to potential interference of small amounts of ambient or scattered light.³² However to obtain more data points we increased the accepted absorbance measurement range to include values between 0.05 and 1.0 AU. With these measurement exclusion criteria, extinction and absorption coefficients were not calculated at 365 nm for any of the native species in articular cartilage at their physiological concentrations since all absorbance values were less than 0.05 AU. Photosensitizers Irgacure 2959 and quinine also had absorbance values outside of the acceptable range at 365 nm, and therefore extinction and absorption coefficients were not calculated. The spectra of sodium hyaluronate did not reveal any clear absorbance maxima and extinction and absorption coefficients were not calculated; however, light absorbance was observed near 200 nm (see Suppl. Fig. S2). Finally, through extrapolation the absorption coefficients were calculated for each of the native cartilage species at their physiological concentration, as well as for each photosensitizer at their aqueous saturation limits ( Tables 1 and 2 ). The extinction coefficients for articular cartilage components and synovial fluid were calculated using percent concentration rather than molarity due to variability in protein molecular weights.

Table 1.

Summary of Absorptive Characteristics for Native Biochemical Components of Articular Cartilage and Synovial Fluid.

Analyte	Absorbance Maximum (nm)	Extinction Coefficient (%−1 cm−1)	Physiological Concentration in Articular Cartilage (%)a	Coefficient of Absorption at Physiological Concentration (cm−1)
Collagen Type I	273 ± 1	0.82 ± 0.04	1829	14.8 ± 0.7
Collagen Type II	263 43	1.88 ± 0.07	1829	34.0 ± 1.0
Fibronectin	282 ± 2	14.0 ± 1.0	0.234	2.8 ± 0.2
Chondroitin Sulfate	263 ± 1	12.8 ± 0.1	329	384 ± 0.3
Proteoglycan	281 ± 1	1.57 ± 0.05	229	3.1 ± 0.1
Bovine Synovial Fluid	283 ± 2	0.14 ± 0.02	100	14 ± 2

^aProtein compostion in articular cartilage is based on estimates from cited sources.

Table 2.

Summary of Absorptive Characteristics for Common Photosensitizers.

Analyte	Absorbance Maxima (nm)	Extinction Coefficient (×103 M−1 cm−1)	Maximum Water Solubilty (M)	Coefficient of Absorption at Maximum Solubility (cm−1)
Riboflavin	223 ± 1	39.6 ± 0.2	2.25 × 10−4	8.91 ± 0.05
	267 ± 1	42.0 ± 0.3		9.45 ± 0.07
	374 ± 1	14.0 ± 0.2		3.15 ± 0.05
	444 ± 1	15.7 ± 0.2		3.53 ± 0.05
Riboflavin-5′-phosphate	224 ± 1	38.1 ± 0.4	2.01 × 10−1	7660 ± 80
	269 ± 1	42.0 ± 0.2		8440 ± 40
	378 ± 1	13.2 ± 0.5		2700 ± 100
	452 ± 1	15.8 ± 0.3		3180 ± 60
Quinine	285 ± 1	3.4 ± 0.1	1.54 × 10−3	5.2 ± 0.2
	334 ± 2	4.6 ± 0.1		7.1 ± 0.2
Irgacure 2959	224 ± 2	12.6 ± 0.2	3.39 × 10−2	427 ± 7
	285 ± 1	21.1 ± 0.2		715 ± 7

Finally, the absorbance spectrum of each cartilage slice was averaged across the 3 knees analyzed. Figure 11 shows the average absorbance spectrum of each slice (slice 1 represents the first slice, which is the topmost slice). Figure 12 shows the cumulative light absorbance as a function of depth into the cartilage (A) and the percentage of light absorbed as a function of depth (B) starting from the top and moving deeper into the cartilage; 100 µm represents the first (top-most) slice.

Figure 11.

Average absorption spectrum of each 100 µm cartilage slice (n = 3; 1 from each of 3 knees). The inset shows a more detailed view of the 200- to 350-nm region, where absorption peaks are more clearly observed. Note that slice 1 represents the first 100 µm slice and each successive slice is moving deeper into the cartilage toward the bone (i.e., slice 7 is the furthest from the articular surface).

Figure 12.

(A) Cumulative absorbance of cartilage with depth determined using Equation (5). The slope of the best-fit linear regression was used to calculate the apparent attenuation coefficient. (B) Percentage of light absorbed by articular cartilage as a function of depth; note that more than 95% of incident light is absorbed within the first 300 µm. Each point represents the average of 3 measurements (n = 3; 1 from each of 3 knees); error bars represent ± 1 standard deviation.

Discussion

In our study, we measured the absorption spectra for the native components of articular cartilage, synovial fluid, and four commonly used photoinitiators. We also analyzed the light absorption through slices of native bovine articular cartilage. Our data are in agreement with previously measured absorption spectra for some of these species; however, they were measured separately without comparisons.^20,37-44 Here we analyzed all of the major native species within the cartilage ECM, together with synovial fluid and several photoinitiators, and compared their absorptive properties. Our results indicate that none of the individual native species in cartilage will interfere with the absorption of UV light at 365 nm by these commonly used photoinitiators.

Collagen type 1 had a peak of maximum absorbance at 273 ± 1 nm; collagen type 2 did not have a clear absorbance maximum and 263 nm was chosen based on previous work.⁴³ Collagen type 2 was found to have a higher extinction coefficient compared with collagen type 1 ( Table 1 ). The light absorption behavior of proteins can be predicted by the number of tyrosine, tryptophan, and cysteine residues present in the polypeptide chain.⁴⁵ Collagen type 1 is composed of 2 α1(I) chains and 1 α2(I) chains, or 3 α1(I) chains.⁴⁶ Bovine α1(I) collagen 1 chain contains 0.5% tyrosine, 0% tryptophan, and 0% cysteine, while the α2(I) chain contains 0.1% tyrosine, 0% tryptophan, and 0% cysteine. Rat-tail α chains for collagen 1 are similar to bovine, 0% tryptophan, 0% cysteine, and 0.5% tyrosine in the α1(I) but with 0.2% tyrosine in the α2(I) chains. Therefore, bovine and rat-tail collagen type 1 are relatively similar in tyrosine, tryptophan, cysteine composition. Collagen type 2 is composed of three α1(II) chains⁴⁶; in bovine collagen type 2 these contain 0.7% tyrosine, 0.5% tryptophan, and 0.6% cysteine. In considering the differences in composition of collagen types 1 and type 2, it would be predicted that collagen type 2 would be a better light absorber than collagen type 1. Our data supports this result. This is interesting as native articular cartilage is composed mainly of collagen type 2 while cartilage scar tissue is mainly collagen type 1.⁴⁷

The absorption coefficient of riboflavin at 365 nm was reported at 0.02% concentration as 11.2 cm⁻¹.^42,48 Based on our absorption coefficient per concentration value in Table 2 , the absorption coefficient for riboflavin at 0.02% was calculated as 6.91 ± 0.05 cm⁻¹. This difference could be attributed to the previous study’s use of solvent comprised of 20% DextranT500 dissolved in 0.9% sodium chloride solution, as opposed to the PBS used in this study.⁴⁸ Extinction coefficients for riboflavin in water were calculated by Koziol³⁸ to be 28,500 M⁻¹ cm⁻¹, 31,600 M⁻¹ cm^-1, 10,300 M⁻¹ cm⁻¹, and 12,100 M⁻¹ cm⁻¹ at 223 nm, 268 nm, 374 nm, and 449 nm, respectively, in reasonable agreement with data obtained here in PBS ( Table 2 ). The extinction coefficient for quinine in 0.1 N sulfuric acid was calculated by Chen³⁷ to be 5020 M⁻¹ cm⁻¹ at 345 nm; in good agreement with our value of 4600 M⁻¹ cm⁻¹ at 334 nm. Bradshaw et al.⁴⁹ found the extinction coefficient of fibronectin at 280 nm to be 704,975 M⁻¹ cm⁻¹, assuming a molecular weight of 550 kDa, this is about 13%⁻¹ cm⁻¹, in agreement with our measured value of 14%⁻¹ cm⁻¹.

Collagen types 1 and 2, and fibronectin had absorption peaks around 280 nm, which is a characteristic of aromatic amino acids in proteins.^45,50-52 Normal and arthritic synovial fluid also had a peak at 283 nm ( Fig. 5 and Suppl. Fig. S1), likely due to dissolved proteins such as lubricin, collagen, and those associated with various proteoglycans.^26,29 The small absorption peak found at 410 nm in the arthritic sample is most likely attributed to hemoglobin, which has been found to be a strong absorber in small concentrations.^53,54 The normal synovial fluid did not show this peak at 410 nm, which was clear and colorless while the arthritic fluid was clear and slightly pink. The extinction coefficients for both synovial fluids were also close to each other at 283 nm, 0.14 ± 0.02%⁻¹ cm⁻¹ for normal synovial fluid versus 0.1575 ± 0.0007%⁻¹ cm⁻¹ for the arthritic sample. Proteoglycan had a peak at 281 nm ( Fig. 6 ), suggesting the presence of aromatic amino acids. Sodium hyaluronate does not absorb visible light but absorbs light at the end of the UV range (Suppl. Fig. S2), and was excluded from the analysis since no clear absorbance maximum was observed. This is not surprising since this particular species does not demonstrate any conjugated systems. Even though chondroitin sulfate is not a protein and has no conjugated systems, it had an absorption peak at 263 nm ( Fig. 2 ). This may be due, in part, to the presence of undissociated biomolecules.^55,56 Previous work has also shown the absorbance spectrum of chondroitin sulfate to have this peak around 260 nm.⁴¹

Of all the species studied, only riboflavin ( Fig. 7 ) and riboflavin-5′-phosphate ( Fig. 8 ) have defined absorption peaks near 365 nm. Quinine has a significant absorption peak at 334 nm, and although it does not have an absorption maximum at 365 nm ( Fig. 10 ) it has been successfully used as a photosensitizer in some applications of UV photopolymerization.⁵⁷ Even though Irgacure 2959 did not have an absorption peak at 365 nm, it did have one at 285 nm ( Fig. 9 ). None of the native species in articular cartilage had an absorption peak at 365 nm. However, they all exhibit absorption peaks at other wavelengths within the UV-Vis spectrum, except sodium hyaluronate ( Table 1 ), but are not relevant with regard to photopolymerization in biomedical applications. While photopolymerization generally proves effective in the UV-Vis range, it is primarily used only with 350 nm or higher wavelengths because of the cytotoxicity induced from shorter wavelength light.^58,59 With regard to these applications, based on our analysis none of the individual native species are expected to significantly interfere with absorption of UV light in solution alone at 365 nm.

These findings can be used to optimize photoinitiated polymer/biopolymer scaffolding procedures in articular cartilage and other tissues to maximize functional performance and minimize the risk of damage to surrounding tissue by exposure to UV light. Our results indicate that at 365 nm UV wavelength riboflavin and riboflavin 5′-phosphate may be better photosensitizers compared with Irgacure 2959 and quinine, as the later do not absorb light well at this wavelength. Irgacure 2959 absorbs UV light at 285 nm, and others have cited an extinction coefficient for Irgacure 2959 at 365 nm although it is quite low, 4 M⁻¹ cm⁻¹.⁵⁸ Based on the design of our experiments and exclusion criteria an extinction coefficient was not calculated for Irgacure 2959. Although quinine has a strong absorbance in the UV range, it does not have a significant absorption at 365 nm. Thus, it would appear that photopolymerization procedures initiated by riboflavin will be more effective than those initiated by Irgacure 2959 and quinine if light absorption is the only consideration. However, energy level matching between photoinitiators and polymerizable groups would also need to be considered to determine appropriate photosensitizers. Unfortunately, this is beyond the scope of this study.

Water solubility is an important consideration for in vivo photopolymerization reactions. Riboflavin and riboflavin-5′-phospohate have similar molar extinction coefficients at 365 nm (13,000 ± 100 M⁻¹ cm⁻¹ and 12,900 ± 400 M⁻¹ cm⁻¹, respectively); however, there is a remarkable difference in water solubility. Since the water solubility is much greater for riboflavin-5′-phospohate, its absorption coefficient at maximum solubility for all absorption maxima was orders of magnitude greater than that for riboflavin ( Table 2 ). This was also true at 365 nm where the absorption coefficients were 2.93 ± 0.02 cm⁻¹ and 2590 ± 80 cm⁻¹ for riboflavin and riboflavin-5′-phosphate, respectively.

An inherent limitation to this study is that the spectroscopic data for each of the species was measured in an isolated environment (e.g., PBS or dilute acid). In vivo, each constituent is not dissolved in an isolated environment, but is interacting with other dissolved materials and these interactions can affect light absorption. Indeed, the effects of pH and solvent changes on the absorption properties of ionizable chemical species, such as polyaniline, 4-nitrophenol, and riboflavin, are documented.^38,60,61 Measuring light absorption under true physiological conditions would be optimal although difficult in practice given problems with solubility, as with collagen and other proteins. In addition, while being dissolved in solution there is no solid matrix as there would be in real tissue. In order to characterize the light absorbance of articular cartilage as whole, we collected the absorbance spectra of 100 µm thick articular cartilage slices to calculate the apparent attenuation coefficient and percentage of absorbed 365 nm UV light through cartilage. The average spectra of each 100 µm cartilage slice was nonzero at all wavelengths, likely due to a combination of scattering and absorption ( Fig. 11 ). It is interesting that the spectral baseline increases between slices 2 and 3 (200 µm and 300 µm into the cartilage from the top). This is most likely due to changes in cartilage composition (relative amounts of proteins and cells, and orientation of proteins). The superficial zone of articular cartilage, estimated as the top 10% to 20% of cartilage volume, displays tightly packed collagen fibers that run parallel to the articular surface. Deeper into the tissue, the middle zone, estimated as 40% to 60% cartilage volume, shows larger diameter collagen fibers that are randomly oriented, and deeper down toward the subchondral bone the collagen fibers continue to become thicker.^5,30 Furthermore, chondron size (cell and pericellular matrix) has also been noted to increase with depth into articular cartilage, 15 to 20 µm in the middle/transition zone to 30 to 40 µm in the deep zone.⁶² Our cartilage plugs were approximately 700 µm thick and based on estimates of the cartilage zones a transition from the superficial zone to the middle zone would occur around 150 µm. Therefore, it is possible that the increase in spectral baseline between slices 2 and 3 might be due to differences in cartilage structure and composition (increasing collagen diameter, changing collagen orientation, and increasing chondron size).

There were “peaks and shoulders” within the spectra of the cartilage slices; notably around 340 nm, 280 nm, 260 nm, and 220 nm. The absorbance maximum near 280 nm is likely a result of aromatic amino acid absorbance; this is not surprising since the solid matrix of the tissue is mostly protein (60% of dry weight is collagen).^{5,45,50-52,63} Furthermore, the absorption peak near 220 nm is likely a result of absorption from the backbone of proteins in the tissue.^51,52 The peak near 260 nm could be attributed to nucleic acid absorbance and/or collagen absorbance since collagen types 1 and 2 absorb strongly around 260-270 nm.^43,63 The shoulder observed near 340 nm in Figure 11 was also observed by Ebert et al.¹⁹ in equine articular cartilage.

The apparent attenuation coefficient for cartilage was calculated to be 62.9 ± 0.9 cm⁻¹ at 365 nm ( Fig. 12A ). This value significantly deviates from the absorption coefficient value of zero for all of the individual components at 365 nm. Cartilage tissue has been described as “highly forward-scattering.”⁶⁴ The apparent attenuation coefficient reported here does not separate absorption from scatter. While this value does describe the attenuation of light as it travels through cartilage, this is an inherent limitation of this work and future studies should be conducted to separately measure absorbance and scatter to determine how each contributes to light attenuation in tissue.

In a study by Ebert et al.,¹⁹ equine articular cartilage slices (400 µm thick) had an absorption coefficient of about 6.5 cm⁻¹ and a scattering coefficient of about 42 cm⁻¹ at 365 nm. Our value of 62.9 cm⁻¹, while higher, is similar when both Ebert et al.’s scatter and absorbance are combined, 48.5 cm⁻¹.¹⁹ The additional differences may be due to differences in the age of the specimens, processing conditions, and method of evaluation (reflectance vs. absorbance measurements). In addition, Ebert et al.¹⁹ used a 400-µm thick specimen while 100-µm thick specimens were used in this work, potentially leading to differences in measurements. Finally, Madsen et al.²⁵ reported and cataloged the absorption and scattering coefficients from porcine nasal and shoulder cartilage, equine articular cartilage, and rabbit cartilage.^19,21-24 They found differences of 3 orders of magnitude between the absorption and scattering coefficients for cartilage depending on the tissue source and method of analysis.^21,25

The attenuation of 365 nm light with depth into cartilage from the surface is shown in Figure 12B . After 300 µm greater than 95% of light is absorbed. The attenuation of incident light is dependent on initial intensity and wavelength. Greater initial intensity will lead to greater penetration; however, this must be balanced with concomitant deleterious effects from greater radiation flux (i.e, tissue and cell damage). Ebert et al.¹⁹ found that light penetration depth was wavelength dependent; longer wavelengths had greater penetration, reflecting the greater absorbance and scattering that occur at shorter wavelengths, as observed in our study ( Fig. 11 ). Variations in penetration depth measurements may be due to differences in evaluation method, and cartilage age and species. It has been noted that cartilage can vary in optical properties between different parts within the same animal donor.²⁵

In comparing our measured absorption coefficients for each photosensitizer at 365 nm, the apparent attenuation coefficient of cartilage is approximately 95% higher than that of riboflavin. This suggests that cartilage will significantly interfere with the absorption of light by photosensitizers like riboflavin. However, due to the greater solubility of riboflavin-5′-phosphate it may be better able to compensate for the interference of cartilage. Although this assumes that light must travel through cartilage before reaching the photosensitizers, this may not always true.

Conclusion

This study compared the absorptive behavior of the major native biochemical species in articular cartilage and synovial fluid with that of the photosensitizers riboflavin, riboflavin-5′-phosphate, Irgacure 2959, and quinine. The absorptive properties of the individual cartilage components were then compared with those of intact cartilage samples. Within the context of our study, Irgacure 2959 and quinine were found to be poor absorbers of 365 nm light compared to riboflavin and its phosphate derivative. Our findings indicate that the native species of articular cartilage alone in solution are not significant light absorbers at 365 nm, and thus would seem not to interfere with absorption of light by photosensitizers. While the components did not significantly absorb light alone, the intact cartilage specimen showed distinguishable absorption of light at 365 nm that could interfere with the absorption of light by riboflavin, thereby hindering the effectiveness of cartilage repair procedures using UV-initiated photopolymerization. Furthermore, intact cartilage was observed to have defined absorption peaks below 300 nm, which were attributed to proteins and, potentially, nucleic acids. The results reported here may prove helpful in optimizing methods for in situ, in vivo chondral defect repair.