Monitoring extracellular matrix (ECM) components is one of the key methods used to determine tissue quality in three-dimensional scaffolds for regenerative medicine and clinical purposes. Raman spectroscopy can be used for non-invasive sensing of cellular and ECM biochemistry. We have investigated the use of conventional (confocal and semiconfocal) Raman microspectroscopy and fibre-optic Raman spectroscopy for in vitro monitoring of ECM formation in three-dimensional poly(ethylene oxide terephthalate)–poly(butylene terephthalate) (PEOT/PBT) scaffolds. Chondrocyte-seeded PEOT/PBT scaffolds were analysed for ECM formation by Raman microspectroscopy, biochemical analysis, histology and scanning electron microscopy. ECM deposition in these scaffolds was successfully detected by biochemical and histological analysis and by label-free non-destructive Raman microspectroscopy. In the spectra collected by the conventional Raman set-ups, the Raman bands at 937 and at 1062 cm−1 which, respectively, correspond to collagen and sulfated glycosaminoglycans could be used as Raman markers for ECM formation in scaffolds. Collagen synthesis was found to be different in single chondrocyte-seeded scaffolds when compared with microaggregate-seeded samples. Normalized band-area ratios for collagen content of single cell-seeded samples gradually decreased during a 21-day culture period, whereas collagen content of the microaggregate-seeded samples significantly increased during this period. Moreover, a fibre-optic Raman set-up allowed for the collection of Raman spectra from multiple pores inside scaffolds in parallel. These fibre-optic measurements could give a representative average of the ECM Raman signal present in tissue-engineered constructs. Results in this study provide proof-of-principle that Raman microspectroscopy is a promising non-invasive tool to monitor ECM production and remodelling in three-dimensional porous cartilage tissue-engineered constructs.
Articular cartilage is a highly specialized connective tissue with the function to provide a smooth and low-friction buffer between the bones of a joint and to distribute the load over the surface of joints during movement. In several pathological conditions, such as osteoarthritis and rheumatoid arthritis or trauma, cartilage shows limited capacity for regeneration due to poor cellularity and its avascular character [1,2]. The main features of cartilage are closely related to its three-dimensional matrix, which mainly consists of collagen type II, proteoglycans and water [3,4]. Therefore, evaluating the production levels of these essential extracellular matrix (ECM) components is a key to determine tissue quality in tissue-engineered constructs. Conventional quality testing methods such as immunohistochemistry, histology and microscopy techniques are all destructive and require tissue fixation, labelling, biochemical staining or cell lysis. By contrast, Raman microspectroscopy is a label-free technique, which does not require special sample preparation and can be used for non-invasive characterization of cell and tissue biochemistry . It has been demonstrated that a careful selection of suitable laser wavelengths and intensity eliminates cell damage allowing for the study of cells without inadvertently changing their phenotype or behaviour caused by photo damage [6,7]. This vibrational spectroscopic technique employs an inelastic scattering effect (the Raman effect) to generate a molecular fingerprint of the investigated samples based on detection of specific wavelength shifts caused by chemical bond vibrations.
Raman spectral studies have already been performed on structural analysis of collagen [8,9], sulfated glycosaminoglycans (sGAGs) and proteoglycans [10,11]. Furthermore, this technique has been successfully applied to study collagen-containing ECM in a medium-throughput culture system  and to monitor chondrocyte behaviour on bioactive scaffolds . In a fibre-optic configuration, Raman spectroscopy has already been used for monitoring tissues in joints in vivo  and to investigate the process of bone graft incorporation in bone reconstruction and repair in a transcutaneous manner . Other researchers have reported on in vivo measurements from the bladder and prostate , oesophagus , skin , cervix [19,20] and arteries using Raman spectroscopy .
In this study, we have investigated the use of label-free Raman microspectroscopy to detect ECM formation and to monitor the production levels of essential ECM components over time in three-dimensional porous poly(ethylene oxide terephthalate)/poly(butylene terephthalate) (PEOT/PBT) scaffolds using conventional and fibre-optic Raman set-ups. Previously, PEOT/PBT copolymers have been extensively studied and proved to be suitable for use as scaffold material in tissue engineering, both in vitro and in vivo [22–25], and reached clinical applications (PolyActive, IsoTis Orthopaedics SA) as dermal substitutes  and bone fillers [27,28].
2. Material and methods
2.1. Casting of agarose microwell arrays
Agarose microwell arrays were prepared with a soft-lithography technique. Polydimethylsiloxane negative moulds were used to routinely cast the arrays, with microwells with a diameter of 200 µm, in 3% agarose gel as described earlier . Ultrapure agarose (Invitrogen, Carlsbad, CA, USA) was dissolved with a concentration of 3% w/v in sterile phosphate-buffered saline (PBS) solution (Gibco, Carlsbad, CA) by heating. Seven millilitres of dissolved agarose were pipetted into each well of a six-well tissue culture plate in which the moulds were previously placed (one mould per well) and centrifuged briefly to remove air bubbles. After setting in 4°C for 30 min, the solidified gels with the microwell arrays were removed from the wells and separated from the mould using a spatula. A cylindrical puncher was used to cut out the microwell arrays to fit a standard 12-well plate.
2.2. Fabrication of poly(ethylene oxide terephthalate)/poly(butylene terephthalate) scaffolds
300PEOT55PBT45 (PolyActive300/55/45, PolyVation, The Netherlands) is a block copolymer with a weight ratio of 55 to 45 for the two components (PEOT and PBT), respectively, and a molecular weight of the starting poly(ethylene glycol) segments of 300 Da used in the copolymerization process. Three-dimensional regular grids were fabricated by three-dimensional fibre deposition with a bioscaffolder (SysENG, Germany) with a fibre to fibre distance of 800 µm, a fibre diameter of approximately 200 µm and a layer thickness of 150 µm. Cylindrical porous scaffolds (6 mm in diameter by 4 mm in height for the semiconfocal Raman study and 4 mm in diameter by 3 mm in height for the confocal and the fibre-optic Raman study) were punched out of the three-dimensional regular grids. These scaffolds were sterilized in 70% ethanol two times for 30 min each, washed in PBS first for 5 min and additionally for other 30 min two times and finally incubated in culture medium overnight prior to cell culture.
2.3. Isolation of bovine chondrocytes and cell culture
Primary bovine chondrocytes were isolated from articular cartilage derived from the femoral-patellar groove of a 10-month-old calf by digestion with 420 Units ml−1 collagenase type II (Worthington Biochemical, Lakewood, NJ, USA). The freshly isolated passage 0 (P0) chondrocytes were cultured in chondrocyte medium (CM) consisting of Dulbecco's modified Eagle medium (DMEM; Gibco, Carlsbad, CA, USA) containing 10% v/v fetal bovine serum (South American origin; Biowhittaker, Lonza, Verviers, Belgium), 100 Units ml−1 penicillin G (Invitrogen, Carlsbad, CA, USA), 100 µg ml−1 streptomycin (Invitrogen), 0.1 mM non-essential amino acids (Sigma, St Louis, MO, USA), 0.4 mM proline (Sigma) and 0.2 mM l-ascorbic-acid-2-phosphate (Sigma) at 37°C under a humidified atmosphere of 5% CO2.
After being cultured on tissue culture plastic (T-flask; Nunc; Thermo Fischer Scientific, Roskilde, Denmark), in control samples chondrocytes (P1) were seeded on scaffolds in CM. To compare samples seeded with single chondrocytes and with chondrocyte aggregates (confocal Raman study), P1 chondrocytes were used to seed directly onto scaffolds, while in parallel cell aggregates were made using agarose microwell arrays in CM (2865 microwells and 1 × 106 cells per agarose microwell array) for subsequent scaffold seeding. Chondrocytes seeded in these microwells spontaneously formed microaggregates (approx. 350 cells per microaggregate) at the bottom of the microwells after 12 h. Subsequently, at 1 day of culture, the chondrocyte microaggregates were flushed out with CM from the agarose microwell arrays, collected and seeded on scaffolds.
2.4. Cell seeding of poly(ethylene oxide terephthalate)/poly(butylene terephthalate) scaffolds
Each scaffold was placed aseptically into a well of a non-tissue culture treated 24-well (in the semiconfocal Raman study) or 48-well plate (in the confocal and fibre-optic Raman study; both Nunc, Thermo Fisher Scientific, Roskilde, Denmark).
In the semiconfocal Raman study, human fibronectin (300 µg ml−1, BD Biosciences), which is known to support cell adhesion , was applied to the chondrocyte suspension before seeding onto the scaffolds. The cell–fibronectin suspension, 3 × 106 cells in a final volume of 50 µl for each scaffold, was pipetted onto scaffolds and left to completely adsorb into the porous structure. Samples were left in an incubator at 37°C under a humidified atmosphere of 5% CO2 for 45 min. Subsequently, 2 ml of culture medium was carefully added drop wise. After 7 and 21 days of culture, the samples were studied using Raman microspectroscopy. Chondrocytes of C-28/I2 human immortalized chondrocyte cell line were also seeded on PEOT/PBT scaffolds. C-28/I2 cells were used as a negative control for matrix formation, since they were previously described showing mainly proliferation and low gene expression involved in ECM synthesis and turnover . C-28/I2 cell-seeded samples were cultured in CM under a humidified atmosphere of 5% CO2 at 37°C and analysed by Raman microspectroscopy after 21 days of culture period.
In the confocal Raman study (single cells versus microaggregates study) and fibre-optic Raman study, before transferring the PEOT/PBT scaffolds into 48-well plates, a layer of 3% w/v agarose gel was deposited onto the bottom of each individual well. A circular hole with a diameter slightly bigger than the scaffold was punched out from the middle of the gel in each well and the scaffolds were placed into these holes (one scaffold per well/hole). These holes serve as barriers to prevent cell suspension from flowing out of the scaffold, providing efficient cell seeding. For fibre-optic Raman measurements, only single cell-seeded samples were used. The single chondrocyte or microaggregate suspension (both without fibronectin) consisting of 500 000 cells, or 500 000 cells in microaggregates in a final volume of 30 µl per scaffold, were seeded onto scaffolds drop wise. Subsequently, samples were left undisturbed for 45 min and then 1 ml of culture medium was carefully added drop wise. Cell-seeded samples were cultured in chondrogenic differentiation medium consisting of DMEM supplemented with 100 Units ml−1 penicillin G, 100 µg ml−1 streptomycin, 40 µg ml−1 proline, 0.2 mM l-ascorbic-acid-2-phosphate, 50 mg ml−1 insulin-transferrin-selenite Premix (Bexton Dickinson), 100 µg ml−1 sodium pyruvate (Sigma), 10 ng ml−1 transforming growth factor-β3 (Invitrogen) and 0.1 µM dexamethasone (Sigma), at 37°C under a humidified atmosphere of 5% CO2. After 7, 14 and 21 days of culture, analysis of ECM formation was performed by Raman microspectroscopy, total DNA quantification, glycosaminoglycan (GAG) and hydroxyproline detection assay, scanning electron microscopy (SEM) and histology.
2.5. Semiconfocal Raman microspectroscopy
Samples for the semiconfocal Raman study were monitored by using a Nikon E600 epi-fluorescence microscope (Nikon, Inc., Melville, NY, USA) modified for near-infrared (NIR) Raman spectroscopy in house which was previously described . Briefly, a 785 nm Kaiser Invictus laser was line-focused (Kaiser Optical Systems, Inc., Ann Arbor, MI, USA) onto the samples using a 20×/0.75 NA S Fluor objective. Raman-scattered light was collected through the same 20×/0.75 NA S Fluor objective and dispersed through a spectrograph (HoloSpec f/1.8, Kaiser Optical Systems, Inc.). Prior to Raman measurements, samples were transferred onto fused silica slides. Raman signal was collected for 180 s on a charge-coupled device (CCD) detector optimized for NIR wavelengths (DU401-BR-DD, Andor Technology, Belfast, UK). Raman transects consisted of 126 Raman spectra arranged at equidistant points along a line through the specimen. Compared with confocal Raman systems (point-mapping mode), the confocality is reduced along the slit. Three transects were collected at three various locations (in three pores) across the surface of each scaffold sample and time point (one sample per time point). A total of 378 (3 × 126) Raman spectra were collected from each scaffold and time point in this study. Bright field micrographs were collected with a stereomicroscope (Nikon, SMZ800) and with a Plan Apo 1×/0.1 NA objective from each scaffold at all time points.
2.6. Confocal Raman microspectroscopy
In the confocal Raman study (single cells versus microaggregates study), measurements were performed using a home-built confocal Raman spectrometer as described earlier [12,33,34] on the tissue-engineered constructs. Briefly, a krypton ion laser (Coherent, Innova 90K, Santa Clara, CA, USA) with an emission wavelength of 647.1 nm was used as the excitation source. A 40×/0.75 NA Olympus UIS2 objective (UPlanFLN, Olympus, Hamburg, Germany) was used to illuminate the sample as well as to collect the Raman-scattered photons in epi-detection mode. The scattered light was filtered by a RazorEdge filter (Semrock, Rochester, NY, USA) to suppress reflected laser light and Rayleigh-scattered light, and focused onto a pinhole of 15 μm diameter at the entrance of an imaging spectrograph (HR460; Jobin-Yvon, Paris, France), which contained a blazed holographic grating with 600 grooves mm−1. The spectrograph dispersed the Raman-scattered photons on an air-cooled electron-multiplying CCD detector (Newton DU-970N, Andor Technology). The system provided a spectral resolution of 1.85–2.85 cm−1 per pixel over the wavenumber range from −20 to 3670 cm−1. Raman images were acquired from the samples by recording a full spectrum from each position of the laser beam guided by the displacement of the scanning mirror in the area of interest on the samples. Each Raman scan resulted in 1024 spectra. These measurements were performed in an area of 30 × 30 µm with a spectral resolution of 310 nm per scan, an accumulation time of 1 s per step and an excitation power of 35 mW. Three Raman images were collected from three pores of each sample, and triplicates of samples were measured each time point. The samples were scanned by using a scanning mirror system (SM, MG325D and G120D, General Scanning, Bedford, MA, USA). Toluene, a Raman calibration standard, which has accurately known peak frequencies (521, 785, 1004, 1624, 2921 and 3054 cm−1), was used for wavenumber calibration of the spectra. Prior to Raman measurements, samples were washed with PBS, fixed in 10% formalin for 10 min, washed extensively with PBS afterwards and placed onto UV grade calcium fluoride slides (Crystran Ltd, UK).
2.7. Raman data analysis
Spectra obtained from the semiconfocal Raman study were corrected for curvature, dark current and variations in the CCD quantum efficiency. A mean spectrum was calculated and baseline-corrected from each transect after correction for the fused silica background.
Preprocessing of spectra obtained from the confocal Raman study (single cells versus microaggregates study) was performed as described previously [34–36]. The spectra were preprocessed by: (i) removing cosmic ray events, (ii) subtracting the camera offset noise (dark current), and (iii) calibrating the wavenumber axis. The well-known band positions were used to relate wavenumbers to pixels. The frequency-dependent optical detection efficiency of the set-up was corrected using a tungsten halogen light source (Avalight-HAL; Avantes BV, Eerbeek, The Netherlands) with a known emission spectrum. The detector-induced etaloning effect was compensated by this procedure. After data correction, the spectra of the Raman scans on respective measurement days were averaged to generate mean spectra for each sample.
In both the semiconfocal and confocal Raman study, semiquantitative univariate data analysis was performed by selecting specific vibrational bands of collagen or/and sGAGs in the averaged spectra from each time point and integrating each band after local baseline subtraction. Subsequently, normalized band-area ratios were obtained from the collagen and proteoglycan bands separately over the integrated band of the phenylalanine ring breathing mode at 1001 cm−1. The calculations were made using normalized band-area ratios, not absolute values, so the system is a semiquantitative rather than a quantitative monitoring of the collagen and sGAG formation . All data manipulations were performed using routines written in Matlab 7.4 (The Math Works Inc., Natick, MA, USA).
2.8. Fibre-optic Raman spectroscopy and data analysis
Fiber-optic Raman measurement was carried out using a custom designed fibre-optic Raman probe. Details of the Raman probe design have been previously described [15,37]. Briefly, the fibre-optic Raman probe consisted of 50 collection fibres that were arranged into 10 branches. The fibres employed in the probe design were silica fibres with core diameter of 100 µm. Each branch housed five fibres and terminated in a stainless steel ferrule with an outer diameter of 1.25 mm. The Raman probe was used with a custom designed aluminium probe holder to facilitate contact of each fibre with the scaffold. The Raman collection probes were bundled at the distal end into a linear array for coupling to the Raman spectrograph (RamanRxn1, Kaiser Optical Systems Inc.). Laser illumination was achieved using a 300 µm core optical fibre that terminated in a stainless steel ferrule (outer diameter of 400 µm) and coupled to an 830 nm laser source. Raman data were acquired with the fibre-optic Raman probe and processed with Matlab—including silica subtraction—to generate 10 Raman spectra for all collection probes. Data processing included dark subtraction, removal of spikes due to cosmic rays and correction for grating induced curvature. Silica contribution from the optical fibres was subtracted by derivative subtraction of silica background spectra that were acquired by reflecting the excitation laser from a frosted aluminium surface into the optical fibres. Further processing included baseline subtraction using a fifth-order polynomial . The 10 Raman spectra were averaged to generate mean spectra for each sample (each single cell-seeded scaffold).
2.9. Glycosaminoglycan detection assay
Following 7 and 14 days of culture, samples from the confocal Raman study (single cells versus microaggregates study) were washed with PBS and frozen at 80°C. After thawing, the constructs were digested for 16 h at 56°C with 1 mg ml−1 proteinase K (Sigma-Aldrich) in Tris/EDTA buffer (pH 7.6). This solution contained 185 µg ml−1 iodoacetamine and 10 µg ml−1 Pepstatin A (both Sigma-Aldrich). Quantification of total DNA was done using the CyQuant DNA assay (MolecularProbes) and a spectrophotometer (excitation 480 nm, emission 520 nm, Victor3, Perkin Elmer). The amount of GAG was determined spectrophotometrically after reaction with dimethylmethylene blue dye (Sigma-Aldrich) in a 10 mM hydrochloric acid (HCl) solution containing 3.04 g l−1 of glycine and 2.37 g l−1 of NaCl (pH 3). A microplate reader (Multiskan GO, Thermo Fisher) was used to determine the absorbance at 520 nm. The amount of GAG was calculated using a standard of chondroitin sulfate (Sigma-Aldrich).
2.10. Hydroxyproline assay
Hydroxyproline assay was carried out on the same proteinase K-digested samples as for the GAG assay and DNA quantification. From the digested sample, 15 µl was transferred to a Teflon-capped glass bottle, 15 µl of concentrated HCl (12 M) was added and samples were hydrolysed at 120°C for 3 h. The complete supernatant was transferred to a 96-well plate and left to evaporate at 60°C. Subsequently, 100 µl of chloramine T/oxidation buffer mix was added to each well and incubated at room temperature for 5 min. Finally, 100 µl of 4-dimethylaminobenzaldehyde reagent was added to each including the hydroxyproline standard and incubated for 90 min at 60°C. A microplate reader (Multiskan GO, Thermo Fisher) was used to determine the absorbance at 560 nm.
In the confocal Raman study (single cells versus microaggregates study), after Raman measurements, samples were embedded in glycol methacrylate for histological analysis, sectioned at 10 µm intervals and stained with Safranin-O (Sigma) for visualization of sGAGs, and haematoxylin and eosin (H&E; Sigma) for visualization of the nuclei and cytoplasm.
2.12. Scanning electron microscopy
Cell morphology and attachment were characterized by SEM analysis with a Philips XL 30 ESEM-FEG. Samples (from the confocal Raman study) were fixed for 30 min in 10% formalin. Subsequently, the samples were dehydrated in sequential ethanol series and critical point dried from liquid carbon dioxide using a Balzers CPD 030 critical point dryer. The constructs were gold sputter coated (Cressington) prior to SEM analysis.
2.13. Statistical analysis
The results are presented as mean ± standard deviation (s.d.). Experimental data were analysed for statistical significance using a Student t-test. Statistical significance was set to p-value < 0.05 (*).
3. Results and discussion
The aim of this study is to investigate the use of conventional (semiconfocal and confocal) and fibre-optic Raman microspectroscopy for monitoring ECM formation during cell culture in three-dimensional fibre-deposited porous PEOT/PBT scaffolds. Moreover, chondrocytes have been seeded as single cells and as microaggregates onto the PEOT/PBT scaffolds, to investigate whether there are differences in ECM formation between single cell- and cell aggregate-seeded scaffolds. The rationale of using microaggregates instead of single chondrocytes was to initiate increased ECM formation. It is known that three-dimensional aggregation is an essential step in chondrogenesis and sustains matrix production of chondrocytes [39–41].
In the semiconfocal Raman study, Raman transects from PEOT/PBT scaffolds seeded with bovine chondrocytes were acquired after 7 and 21 days of culture period using a conventional Raman set-up. Mean Raman transects of these samples showed increased intensities of the bands at 937 and 1062 cm−1 from day 7 to day 21 of the culture period (figure 1a,b, truncated). These bands correspond to collagen (proline, hydroxyproline and C–C vibrations of collagen backbone)  and chondroitin sulfate and proteoglycans (including aggrecan) [10,43,44], respectively, which are essential components of the ECM in cartilage. It was previously described that the band of collagen found at 937 cm−1 can be used as a Raman marker for collagen-containing ECM formation over time in chondrocyte pellet cultures . Furthermore, chondrocyte behaviour on bioactive scaffolds has already been monitored by using Raman spectra from the ECM deposited onto such scaffolds and by assessing Raman bands indicative for collagen and amide I in the spectral region between 1200 and 1800 cm−1 . In order to quantify changes in intensities of bands at 937 and 1062 cm−1, we used semiquantitative univariate data analysis. The results of the univariate data analysis are shown in figure 1c with the normalized band-area ratios for the band signifying collagen and proteoglycans. Subsequently, band-area ratios were obtained from collagen and proteoglycan bands separately over the integrated band of the phenylalanine ring breathing mode at 1001 cm−1 . Phenylalanine was found to be the best-resolved band and is not sensitive to local chemical environments . The normalized band-area ratio for both collagen and proteoglycan content increased from day 7 to day 21 of culture period. Mean Raman transects were also obtained from PEOT/PBT scaffolds without cells (figure 2a,b, truncated) as reference spectra. Comparing these spectra with the mean Raman spectra of cell-seeded scaffolds in figure 1a,b, Raman contributions of PEOT/PBT polymer can be distinguished at 630 and 854 cm−1 and between 1100 and 1800 cm−1 in Raman spectra from cell-seeded scaffolds. The area of measurement on ECM overlaps in some cases with part of the polymer scaffold during Raman imaging, leading to Raman contributions from the polymer in the obtained spectra. Yet, PEOT/PBT-specific bands can be easily recognized and do not interfere with ECM-specific bands at 937 and 1062 cm−1.
Raman transects from PEOT/PBT scaffolds seeded with the C-28/I2 human immortalized chondrocyte cell line, used as negative control for ECM formation, were collected after 21 days of culture period similar to bovine chondrocyte-seeded samples. The average of the Raman transects of these samples did not show bands at 937 cm−1 (collagen) and at 1062 cm−1 (proteoglycans), even after 21 days of culture (figure 3). Furthermore, bright field microscopy of these samples showed that no ECM was present (figure 4a) which confirmed the analysis done based on Raman spectra. By contrast, the bright field micrographs of bovine chondrocyte-seeded samples (figure 4b) showed collagen-like fibre formation at day 7 of the culture period indicative of ECM production. Moreover, the amount of ECM further increased in time completely covering the scaffolds at day 21 (figure 4b).
In the confocal Raman study presented here, the seeding of single chondrocytes was compared with seeding of chondrocyte microaggregates on three-dimensional fibre-deposited PEOT/PBT scaffolds. In a previous study, similar three-dimensional polymer scaffolds were seeded with bovine articular chondrocytes in spinner flasks, so that stirred cell culture promoted the formation of cell aggregates as were used in this study . In that previous study , cell aggregation enhanced the kinetics of cell attachment without compromising the uniformity of cell distribution on the scaffolds. In our confocal Raman study, chondrogenic differentiation medium was used to stimulate proteoglycan synthesis from chondrocytes in three-dimensional cell culture in the same way as was done by Mauck et al. . Mean Raman spectra from single chondrocyte- and chondrocyte microaggregate-seeded scaffolds (figure 5) showed bands corresponding to phenylalanine (C–C aromatic ring) at 1001 cm−1 , lipids/proteins (CH2 bending mode) at 1448 cm−1  and amide I at 1657 cm−1 . Additionally, Raman bands at 937 cm−1 (collagen) and at 1062 cm−1 (proteoglycans) were also detected indicating ECM deposition. Although normalized band-area ratios for proteoglycan content did not change over time, collagen synthesis was found to be different in single chondrocyte-seeded scaffolds when compared with microaggregate-seeded samples (figure 5b,c). The normalized band-area ratios for total collagen content of single cell-seeded samples gradually decreased during a 21-day culture period (figure 5b), whereas the total collagen content of the microaggregate-seeded samples significantly increased (figure 5c) during this period. Several studies with chondrocytes have shown correlation between the aggregation of chondrocytes in multiple layers and enhanced cartilaginous tissue formation [41,50–52]. Raman data in this study also indicate that the use of chondrocyte microaggregates, instead of single chondrocytes, for cell seeding of scaffolds could enhance ECM formation and therefore could potentially lead to more functional cartilage tissue-engineered constructs.
ECM formation in the single chondrocyte-seeded scaffolds was also evaluated using fibre-optic Raman spectroscopy. The experimental fibre-optic Raman set-up for measurements can be seen in figure 6a. Average Raman spectra from single chondrocyte-seeded scaffolds cultured for 14 and 21 days mainly showed Raman contributions from the optical fibres (i.e. residual silica Raman spectra) even after silica background subtraction. However, Raman spectra of cell-seeded scaffolds showed Raman bands at 920 and 972 cm−1 corresponding to components of collagen, such as amino acid proline, which were not found in the spectra of the scaffold without cells. We show in figure 2 that PEOT/PBT spectra do not show Raman bands in the spectral region between 900 and 1100 cm−1. Therefore, bands arising in this region in spectra of bare scaffolds, as can be seen in figure 6c, originate from Raman scattering generated from the silica within the optical fibres, which were not completely removed by derivative silica subtraction algorithm performed with codes written in Matlab. A suitable algorithm in Matlab for more accurate silica background subtraction and/or further data processing using a multivariate statistical technique such as band target entropy minimization greatly reduces background bands and maximizes the Raman bands specific for ECM.
Additionally, we investigated ECM deposition in the single cell- and microaggregate-seeded samples by more conventional methods. GAG formation was successfully detected by biochemical analysis (GAG assay), both in the single cell- and microaggregate-seeded samples at day 7 and 14 of culture period; however, significant differences were not observed (figure 7a). The hydroxyproline analysis indicated that cells produce significantly more collagen-like matrix per cell when seeded on the scaffolds as single cells compared with microaggregates (figure 7b). Furthermore, higher collagen production per cell was found after 7 days compared with 14 days for single cells. By contrast, the absolute amount of total collagen was comparable for both time points in both conditions. The cell number, determined by DNA assay, showed an increase in single cell-seeded samples, whereas the DNA content for microaggregates remained stable during culture. This could explain the decrease in time for total collagen produced by single cell-seeded samples.
Furthermore, sGAG formation was also successfully shown by Safranin-O staining, and the nuclei and cytoplasm were visualized by H&E staining both in the single cell- and microaggregate-seeded samples at day 7 and 14 of culture period (figure 7c). The results of the Safranin-O indicated milder level of sGAG production in the microaggregate-seeded samples compared with the single cell-seeded ones, which can be explained by the fact that, especially after 14 days, the number of cells in the histological section is lower for microaggregate-seeded samples. This subtle difference in sGAG production was also observed by the GAG assay, although it was not statistically significant.
Furthermore, both the single cell- (figure 8a) and microaggregate-seeded sample (figure 8b) matrix-like structure formation could be observed by electron microscopy. Chondrocyte microaggregates adhering to the scaffold surface could be seen in the SEM micrographs of the microaggregate-seeded samples at day 7 of culture period. After 21 days of culture, the majority of microaggregates had a flattened appearance and collagen-like fibrous structures were observed in the vicinity of cells suggesting active ECM formation ultimately covering the surface of the scaffold used.
In this study, the use of conventional (semiconfocal and confocal) Raman microspectroscopy and fibre-optic Raman spectroscopy has been demonstrated for studying the formation of cartilage ECM in three-dimensional fibre-deposited PEOT/PBT scaffolds. The three different Raman microscope set-ups used in this study were all found to be feasible tools to monitor tissue formation inside tissue-engineered porous three-dimensional scaffolds. We studied ECM production of primary bovine chondrocytes seeded into three-dimensional fibre-deposited PEOT/PBT block copolymer constructs during three weeks of culture by Raman microspectroscopy and conventional bio- and histochemistry. The Raman bands at 937 and 1062 cm−1 corresponding to collagen and proteoglycans, respectively, were shown to be indicative for ECM production in the Raman spectra as analysed by fibre-optic Raman and confocal Raman measurements inside the pores of the three-dimensional tissue-engineered construct. Normalized band-area ratios for proteoglycan content did not show clear differences over time when comparing single chondrocytes with microaggregates seeded into samples. However, we observed that collagen synthesis was different in single chondrocyte-seeded scaffolds when compared with microaggregate-seeded samples. Normalized band-area ratios for collagen content of single cell-seeded samples gradually decreased during the 21-day culture period, whereas the collagen content of the microaggregate-seeded samples significantly increased during this period. Results obtained in this study suggest that seeding microaggregates, instead of single chondrocytes, seems to improve ECM formation and therefore potentially can be used to improve cartilage tissue-engineered constructs. Future studies will focus on studying matrix (and pericellular matrix) organization in scaffolds more in detail.
A limitation of the fibre-optic system was the inherent background signal generated by silica within the optical fibre. Careful data analysis using Matlab revealed that bands at 920 and 972 cm−1 attributed to ECM components were recoverable. The use of a new generation of fluorocarbon fibre-optic Raman probes, which generate a reference Raman band for quantitative Raman spectroscopy, will eliminate silica background signals and possibly lead to well-resolved Raman spectra of ECM [37,53]. If so and as opposed to the conventional Raman set-ups, which are only able to measure a relatively small volume of the scaffolds, this would allow collection of Raman spectra from multiple pores in tissue-engineered scaffolds with one measurement in parallel. This generates a representation of the overall Raman signal from ECM within these scaffolds.
In order to further improve such analysis, further investigation is needed to validate the potential of fibre-optic evaluation of cartilaginous matrix formation in tissue-engineered constructs. Additionally, more profound knowledge on the discrimination of specific collagen types is a key element to be considered in future studies regarding Raman measurements of cartilage ECM.
The authors gratefully acknowledge the funding from the Dutch Program for Tissue Engineering (DPTE) through grant no. TGT. 6737, from The Netherlands Institute for Regenerative Medicine (NIRM) through grant no. FES0908 and from the National Institutes of Health (NIH) through grant nos. R01AR055222 and R01AR056646. C-28/I2 human immortalized chondrocyte cell line was established using cells derived from tissue that was provided by the National Disease Research Interchange.
- Received May 22, 2013.
- Accepted June 10, 2013.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.