1.

Introduction

Articular cartilage provides a near-frictionless load-bearing surface for joints.^{1 2} It resists compressive forces and distributes loads via interactions between its primary macromolecular components, collagen and proteoglycan (PG).^{2 3} Loss of this function occurs as a result of pathological changes in collagen and PG, and can lead to joint degeneration, or osteoarthritis (OA),^{4 5 6} a progressive, potentially immobilizing, disease affecting millions of people.⁷ Accordingly, monitoring the molecular composition of cartilage in vivo is of great importance.

A significant impediment to the treatment of cartilage degeneration is the lack of methods to diagnose the disease process at an early stage and to assess therapeutic efficacy. Characterizing the tissue matrix and changes in collagen and PG in the greatest possible detail would be a key element of any such evaluation. Magnetic resonance (MR) imaging has emerged as a primary modality for such assessment, with anatomic definition of cartilage on a spatial scale of roughly 100 μm now being supplemented by use of MR techniques more or less specific for hydration and for collagen and PG content.^{8 9} However, to truly characterize the microstructure and biochemistry of cartilage, extraction of tissue for in vitro studies is still required. Histological evaluation provides details of microstructural changes, as well as a qualitative indicator of the distribution of macromolecules, by using appropriate stains.¹⁰ Biochemical assays can be used to obtain more quantitative information on collagen and PG content, but with loss of detailed spatial information. Clearly, an in vivo technique to obtain quantitative, high resolution compositional and structural information from native and repair cartilage would be invaluable.

In the current study, Fourier transform infrared imaging spectroscopy (FT–IRIS) was used to evaluate the composition of tissue engineered cartilage grown in a hollow fiber bioreactor (HFBR).¹¹ FT–IRIS is an extension of conventional infrared spectroscopy, whereby an FTIR spectrometer is coupled to an optical microscope equipped with an array detector. The methodology enables the evaluation of the relative amount, molecular nature, distribution, and orientation of the components of tissues.¹² Although the FT–IRIS technology is currently applicable only to in vitro analysis, the parallel development of an infrared fiber optic probe to evaluate cartilage in vivo ¹³ supports the eventual application of the FTIR imaging method to the clinical setting. A detailed understanding of the spatially localized spectra obtained from FT–IRIS is required in order to pursue this comprehensively. Here, we demonstrate the correspondence of FT–IRIS-determined PG content in cartilage to biochemically determined sulfated glycosaminoglycan (S-GAG) content and to histological PG evaluation.

2.

Materials and Methods

2.1.

HFBR Construction and Seeding

HFBR cartilage was grown from chondrocytes isolated from chick distal sterna as previously described.^{14 15 16 17} Briefly, the bioreactors were constructed from high purity glass tubing (inner diameter 4 mm, total length 60 mm) which had six porous polypropylene hollow fibers inserted (inner diameter 330 μm, 0.2 μm pores, wall thickness 150 μm). Each bioreactor was inoculated with 3×10⁷ cells through the rubber septum on the side port. At the end of the 4 week culture period, the HFBR cartilage was removed and sectioned into segments vertical to the fiber axis (Fig. 1). The composition of HFBR cartilage varies according to the proximity to the point of media influx or efflux,¹⁸ and accordingly, analyses were conducted on HFBR cartilage designated as either near the inflow or outflow region. Three to four tissues per group were utilized for FT–IRIS, histology, and biochemical analyses.

Figure 1

Schematic diagram of HFBR (adapted from Chen etal., see Ref. 18). Regions analyzed by FT–IRIS and biochemistry are highlighted.

2.2.

FT–IRIS Analysis

Tissues were fixed in EtOH and cetylpyridinium chloride (to preserve proteoglycan),¹⁹ embedded in paraffin, sectioned onto barium fluoride windows, and deparaffinized prior to FT–IRIS data acquisition. A BioRad (BioRad, Cambridge, MA) FTS-60A step-scanning Stingray 6000 FTIR spectrometer with a UMA 300A FTIR micrometer and a 64×64 MCT FPA detector (Stingray imaging spectrometer) was used to obtain 4096 FTIR spectra at 8 cm⁻¹ spectral resolution under N ₂ purge. Data were gathered from a 400×400 μm² region at 7 μm spatial resolution. FT–IRIS data were also collected at 8 cm⁻¹ spectral resolution using a Spectrum SpotLight FTIR imaging system (Perkin-Elmer, UK) to obtain IR images of the entire HFBR cartilage cross section. This system consists of an FTIR spectrometer coupled to an optical microscope with a 16×1 linear array of detector elements which has the capability to image a rectangular sample size up to several millimeters in length at a spatial resolution of 6.25 μm. Typically, images of the whole tissue section were acquired within ∼20 min. Data were analyzed with ISys 2.3 software (Spectral Dimensions Inc., Olney, MD). The primary macromolecular components of cartilage, collagen and PG, give rise to specific absorbances in the infrared, as previously described.²⁰ Integrated absorbance band areas were calculated and utilized as indicators of the quantity of component present. Collagen was monitored in the 1590–1720 cm⁻¹ spectral region (amide I absorbance), and proteoglycan [macromolecules consisting of S-GAGs linked to a core protein] in the 985–1140 cm⁻¹ spectral region. Absorbances in this region arise from vibrations of the C–O–C and C–OH bonds.²¹ Spectra were baselined and images based on the integrated area of the amide I region were created to determine type II collagen quantity and distribution. The ratio of the integrated area of the PG absorbance to the amide I collagen absorbance was evaluated in order to obtain relative quantity and distribution of the PG component (PG/Am I). The PG/Am I data were converted to % PG content based on standard mixtures of pure collagen and the proteoglycan aggrecan.²⁰ For comparison to histological data, PG content was quantitated in two regions for each section analyzed, near the cartilage outer surface, and in the center of the tissue [Figs. 2(a) and 2(e)]. For comparison to biochemical data, PG content was quantitated over the whole tissue section.

Figure 2

Cross sections of inflow and outflow tissue from HFBR cartilage. Unstained sections on barium fluoride windows and histologic Alcian blue-stained sections from inflow (a,b) and outflow (e,f) tissue. FT–IRIS images of inflow (c,d) and outflow (g,h) tissues created based on amide I absorbance and PG/Amide I absorbance, respectively. Regions labeled as “1” and “2” show where data were acquired from the “surface” and “center” of the tissues, respectively.

3.

Results

Distribution of collagen and proteoglycan components in FT–IRIS images of HFBR cartilage inflow and outflow sections are shown in Figs. 2(c), 2(d) and 2(g), 2(h), respectively. Although the collagen (amide I) images are qualitatively similar between the inflow and outflow sections, the PG/Am I images are quite different, with the inflow image reflecting a greater PG content. The Alcian blue-stained sections also reflected this pattern [Figs. 2(b) and 2(f)]. FT–IRIS spectra from regions towards the center of the cartilage and near the surface of the cartilage (adjacent to the HFBR inner surface) illustrate the higher absorbances in the PG sugar region, 985–1140 cm⁻¹, for the tissue near the tissue culture medium inflow as compared to tissue near the outflow, and for the tissue near the center of the HFBR as compared to the surface regions, within each of these tissue segments (Fig. 3). Quantitative analysis of the FT–IRIS data revealed a significantly higher PG content in the inflow versus outflow tissues at both positions evaluated, although this difference was more pronounced for the inflow tissues (Fig. 4). Comparison of PG/Am I values to histological determination of PG by integrated optical density calculation showed a significant correlation between these two methods (R=0.73, p=0.007) (Fig. 5). Biochemical determination of S-GAG content normalized by dry weight showed the highest content in the inflow tissue (42.6%) and the lowest in the outflow tissue (27.8%) [Fig. 6(a)]. Upon conversion of the FT–IRIS-determined PG values from the whole tissue to PG dry weight, the FT–IRIS data showed a similar difference of 45.0% PG in the inflow tissues versus 24.7% in the outflow tissues. However, the correlation between the biochemical and FT–IRIS data did not reach statistical significance due to sample number limitations (R=0.72, p=0.07) [Fig. 6(b)].

Figure 3

FT–IRIS spectra obtained from positions in the center and at the outer surface of inflow and outflow tissue engineered cartilage. There is a greater absorbance in the proteoglycan sugar region (985–1140 cm⁻¹) in the center of the tissue versus the surface, although this difference is more pronounced in the inflow tissues as compared to the outflow tissues.

Figure 4

Average FT–IRIS-determined % PG content from inflow and outflow tissues at positions in the center and near the surface of the cartilage. Inflow sections showed significantly higher PG content than outflow in both the center and outer surface. (dark shading, inflow; light shading, outflow).

Figure 5

Correlation between FT–IRIS-determined % PG content and histologic determination of integrated optical density on Alcian blue-stained sections (R=0.73, p=0.007) . Regions from both the center and the surface of tissue sections were included in the analysis.

Figure 6

(a) FT–IRIS-determined PG content converted to % PG/dry weight compared to biochemical determination of S-GAG content. Mean inflow and outflow values were significantly different by both biochemical and FT–IRIS measurements. In addition, there was no significant difference between biochemical and FT–IRIS-determined values for inflow or outflow tissues. (b) Correlation between FT–IRIS-determined PG content and biochemical S-GAG determination did not reach statistical significance due to sample number limitations (R=0.72, p=0.07) .

4.

Discussion

Characterization of molecular changes in native, repair, and tissue engineered cartilage is a critical element in the development of therapeutic approaches to OA. Previous studies from our laboratories have monitored type II collagen and PG components in cartilage in a semi-quantitative fashion,^{14 20} and the current study now extends that work by evaluation of differences in PG content between biochemically different samples at 6.25 μm spatial resolution. The current data demonstrate that quantitative FT–IRIS-determined PG content in tissue sections is consistent with bulk biochemical measurements of S-GAG content in tissue engineered cartilage, and correlates with histological determination of PG content in tissue sections. These results support the use of FT–IRIS as an important tool in the development and monitoring of therapeutics for cartilage disorders. Moreover, since FT–IRIS analysis is based on molecular alterations in tissue, it may be possible to discern pathologic or therapeutic effects before macroscopic changes are observed.

Cartilage is spatially heterogeneous, and superimposed degradation processes greatly increase heterogeneity. Therefore, it is important to quantitate the molecular composition at high spatial resolution. Historically, evaluation of the spatial distribution of PG content in cartilage has been done by analysis of appropriately stained histologic sections,²³ the most commonly used stains being Safranin O and Alcian blue. Safranin O is a cationic dye that binds stoichiometrically to polyanions, e.g., to the keratin and chondroitin sulfate moieties of proteoglycans.²⁴ Based on this assumption, this method has been used for semiquantitative analysis of PG content in cartilage.^{25 26 27 28} However, actual quantification of PG in tissue sections using this method and most other histological methods is challenging due to lack of reproducibility of staining intensity from sample to sample, and to variations of section thickness. Alcian blue staining has also been used frequently to identify PG or hyaline cartilaginous tissue.^{29 30 31} The primary difficulty with this staining method is that the staining intensity is very pH dependent, and also varies with the concentration of ions used in the staining solution.³² Thus, PG quantification from experiment to experiment can be even more problematic than with Safranin O. To minimize those effects in the current study, the tissue engineered cartilage sections were stained using the same preparation of Alcian blue dye on the same day. In spite of careful preparation, such histological studies are still limited by potential variability in section thickness. Even though all sections were cut at a set thickness of 10 μm, the actual thickness of the sections can vary by 20% (data not shown) due to inherent limitations in the precision of the microtome. Ultimately, many of the histological studies of PG content, including the one in the current study, evaluate PG content in a relative, and not absolute, manner.

Some of the limitations associated with histological analysis of PG content are avoided by using FT–IRIS to determine PG content. In particular, variations associated with section thickness can be minimized by the inclusion of an appropriate standard,³³ or by normalization of absorbance bands of interest to a standard absorbance band from the tissue embedding medium that is related to section thickness. Alternatively, in the current study, we calculated the ratio of the PG sugar absorbance to the amide I absorbance, thereby factoring out section thickness. Furthermore, these ratios have previously been correlated to specific quantities of aggrecan and collagen mixtures in a series of model compounds.²⁰

Another potential strength of the FT–IRIS method is that it monitors PGs based on molecular vibrations, and therefore can reflect changes in molecular structure. Although we currently have only investigated total PG content based on absorbance of the sugar components of the molecule, earlier studies have identified proteoglycans and their components, such as hyaluronic acid, chondroitin-4-sulfate and chondroitin-6-sulfate, based on differences in spectral features.^{34 35} It is very likely that these spectral features will change as the molecular structure of PGs is altered through degradative disease processes,³⁶ and we envision ultimately being able to utilize the spectral changes to monitor PG degradation.

The morphology and composition of repair and engineered cartilage is likely to be predictive of its ultimate in vivo functionality,³⁷ so that developing the ability to assess these parameters in the clinical setting is of great importance. To date, magnetic resonance imaging (MRI) has been the technique of choice for defining cartilage anatomic features such as volume, thickness and surface characteristics.^{38 39} In addition, emerging MRI techniques⁹ permit the delineation of more specific biochemical aspects of cartilage matrix, including hydration, PG content,⁴⁰ and collagen content. Recently, MRI has also been applied to the evaluation of a model of repair cartilage.^{17 18} While it is clear that MRI has the potential for early detection of OA and evaluation of compositional properties of repair tissue, the optimal MRI methods towards these goals have not yet been established. In addition, there are certain limitations inherent to MRI techniques as currently implemented. Chief among these is the relative lack of specificity of several of the MRI outcome measures, in spite of their high sensitivity. In addition, alterations in net matrix composition are likely to occur later than molecular changes in component macromolecules. Finally, the spatial resolution of MRI studies is limited; currently, 100 μm isotropic resolution would be considered to be beyond realistic clinical capabilities.

In summary, evaluation of native, repair and engineered cartilage at the molecular level is of critical importance, as is development of novel modalities of investigation. The current study demonstrates for the first time that FT–IRIS determination of PG correlates with the gold standard technique of histological determination of PG, and yields quantitatively similar results to biochemical determination. The FT–IRIS technique, and the related fiber optic technology appropriate for clinical use,¹³ with its potential for quantitative tissue analysis, sensitivity to early molecular changes, and high degree of spatial resolution, may represent an important approach to clinical cartilage assessment.