Disparate Response of Articular- and Auricular-derived Chondrocytes to Oxygen Tension

Abstract

Purpose/Aim

To determine the effect of reduced (5%) oxygen tension on chondrogenesis of auricular-derived chondrocytes. Currently, many cell and tissue culture experiments are performed at 20% oxygen with 5% carbon dioxide. Few cells in the body are subjected to this supra-physiological oxygen tension. Chondrocytes and their mesenchymal progenitors are widely reported to have greater chondrogenic expression when cultured at low, more physiological, oxygen tension (1–7%). Although generally accepted, there is still some controversy, and different culture methods, species, and outcome metrics cloud the field. These results are, however, articular chondrocyte biased and have not been reported for auricular-derived chondrocytes.

Materials and Methods

Auricular and articular chondrocytes were isolated from skeletally mature New Zealand White rabbits, expanded in culture and differentiated in high density cultures with serum free chondrogenic media. Cartilage tissue derived from aggregate cultures or from the tissue engineered sheets were assessed for biomechanical, glycosaminoglycan, collagen, collagen cross-links, and lysyl oxidase activity and expression.

Results

Our studies show increased proliferation rates for both auricular and articular chondrocytes at low (5%) O₂ versus standard (20%) O₂. In our scaffold free chondrogenic cultures, low O₂ was found to increase articular chondrocyte accumulation of glycosaminoglycan, but not cross-linked type II collagen, or total collagen. Conversely, auricular chondrocytes accumulated less glycosaminoglycan, cross-linked type II collagen and total collagen under low oxygen tension.

Conclusions

This study highlights the dramatic difference in response to low O₂ of chondrocytes isolated from different anatomical sites. Low O₂ is beneficial for articular-derived chondrogenesis but detrimental for auricular-derived chondrogenesis.

Introduction

The field of tissue engineering is starting to realize some of its promise with tissue engineered bladders surviving in patients for more than 11 years (1). Articular cartilage tissue engineering has, however, remained a challenge. Significant clinical success has been seen with autologous chondrocyte implantation (ACI) and matrix-induced autologous chondrocyte implantation (MACI) (2,3). However, histological analyses showed that more than 50% of patients had fibrous cartilage repair (3,4).

Current clinical approaches involve the expansion and implantation of articular chondrocytes harvested from a low-load bearing area of the joint (5). At least two issues may disadvantage the use of this donor site, 1) joint damage may have compromised the cells harvested from this site, 2) the cartilage is hypocellular, so cells isolated from small amounts of tissue must undergo significant expansion to give adequate numbers for ACI, MACI or tissue engineering. Chondrocytes undergo significant de-differentiation with increased population doublings, resulting in a cell that poorly re-differentiates into cartilage like tissue (6). Cartilage tissue has significant similarity across sources, i.e. articular, epiglottal, nasal, meniscal and auricular (7). This similarity raises the possibility of using non-articular sources for joint repair (8). Cartilage produced by chondrocytes of all five sources is generally avascular, hypocellular, and matrix-rich. The extracellular matrix has high glycosaminoglycan (9) and collagen content. However, while ostensibly similar, there are biochemical and mechanical differences between the sources and cartilage characteristic of hyaline, elastic or fibro-cartilage (7). An advantage to auricular cartilage is that it is easily harvested with minimal morbidity (10). Our group has had significant success in producing scaffold-free tissue engineered auricular cartilage sheets with favorable mechanical characteristics that could enable total joint resurfacing, tracheal and auricular reconstruction (11–15). Xu et al. (2004) observed that non-expanded porcine auricular chondrocytes embedded in fibrin gel and implanted subcutaneously in nude mice gave greater biomechanical and GAG values than articular or intercostal cartilage constructs (16). Clinical auricular reconstruction with subcutaneously implanted culture expanded auricular chondrocytes has also shown impressive results (17). In vitro tissue engineering of scaffold-free cartilage sheets for tracheal replacement found that the auricular source gave the greatest accumulation of GAG and collagen and had superior mechanical properties (13,14).

Oxygen tension has long been known to impact the development of tissues, with chondrogenesis being promoted by reduced oxygen tension (18). The in vivo partial pressure of oxygen in articular cartilage has been estimated to be ~1% at the growth plate and up to ~10% at the articulating surface (19). In nasal cartilage, it is reported to be 1–3% in vivo (20). We are not aware of any reported measurement or estimate for auricular cartilage. More physiological oxygen tensions have been broadly investigated in articular chondrocytes and mesenchymal stem cells for chondrogenesis (Table 1). In general, reduced oxygen tension enhances articular chondrogenesis, primarily in terms of gene expression and GAG accumulation (Table 1). The effect of oxygen tension is less clear with respect to collagen content, and studies investigating mechanical characteristics are rare. Current dogma is that low oxygen tension stimulates chondrogenesis (21–23). However, we are not aware of any investigations into the effects of low oxygen tension on auricular chondrocytes. Also, given the somewhat variable effects of oxygen tension reported (Table 1), the effect of low oxygen tension on articular chondrogenesis, especially in terms of mechanical strength and collagen cross-linking, deserved further analysis.

Table 1.

Effect of reduced oxygen tension on in vitro chondrogenesis

Species	Cells	Differentiation Format	GAG	Collagen	Mechanical testing	Gene	Harvest time	Expansion Passages and O2	Differentiation O2	Reference
Bo	ACs	Alginate beads	+	+	N/A	+	d1, d3, d10, d17	None	2, 21%	(54)
Bo	ACs	Alginate beads	+	+	N/A	N/A	2 wk, 3 wk	P1 5, 21%	5, 21%	(55)
Bo	ACs	Pellet	−	=	N/A	+/−	d14	P1 1.5, 21%	1.5, 5, 21%	(56)
Bo	ACs	Alginate beads	N/A	N/A	N/A	=	d2, d7	None	0.1, 5, 10, 20%	(19)
Bo	ACs	Monolayer	N/A	+	N/A	N/A	d4	None	5, 21%	(57)
Bo	ACs	5mm diameter disc in agarose	=	−	=		d28	P1 21%	4, 21%	(58)
Bo	ACs, MSCs	Electrospun PCL	+	−	N/A	+	d0, d14, d28, d42	P1-2 21%	5, 20%	(59)
Bo	ACs	Agarose	−	−	−	−	d7, d14, d21, d28	None	5, 21%	(60)
Bo	ACs	Explant	−	N/A	N/A	N/A	d0, d1, d5, d7	None	6, 10, 24, 66, 91%	(61)
Bo	ACs	Self assembling	−	−	N/A	−	2wk, 4wk, 6wk	None	5, 20%	(62)
Ga	SCs	Monolayer	+	N/A	N/A	N/A	2wk, 4wk	None	5, 21%	(63)
Hu	ACs	Pellet	+	=	N/A	N/A	d21	P2-4, 5&20%	5, 20%	(6)
Hu	ACs	Micropellets	+	+	N/A	+	d4, d7, d11, d14	P3 2%	2, 20%	(64)
Hu	ACs	Alginate beads	+	N/A	N/A	N/A	d1, d2, d4	P3 5%	<1, 2, 5, 21%	(65)
Hu	ACs	fibrin glue, genipincross-linked	+	−	=	+	d1, 2.5wk, 7wk	P0 21%	5, 21%	(66)
Hu	ACs	Alginate beads	N/A	N/A	N/A	+	d5	P3 21%	5, 20%	(67)
Hu	ACs	Alginate beads	=	=	N/A	+	d21	P2-3 21%	5, 21%	(68)
Hu	MSCs	Alginate beads	+	=	N/A	+/−	d21	P5 21%	5, 21%	(69)
Hu	ACs	Alginate beads	+	=	N/A	+/−	d21	P2 21%	5, 21%	(69)
Hu	ESCs	Self assembling	−	+	+	N/A	4wk	1P* 2, 20%	2, 20%	(70)
Hu	NCs	Pellet	+	=	N/A	N/A	d21	P2 21%	1, 5.25, 21%	(45)
Hu	ACs	MPEG-PLGA	N/A	N/A	N/A	+	d1, d2, d6	P2 21%	1, 5, 21%	(71)
Hu	BM MSCs	Pellet	+/−	+/−	N/A	+/−	4wk	P2-4 21%	5, 20%	(72)
Hu	ACs	Alginate beads	+	N/A	N/A	+	d7,d14,d28	P3 20%	5, 20%	(73)
Hu	ACs	Pellet	+	+	N/A	+	d7, d14, d28	P3 5, 20%	5, 20%	(74)
Hu	ACs	Pellet	+	+	N/A	+	2wk, 4wk	P2 5, 19%	5, 19%	(75)
Hu	ACs	Monolayer	N/A	+	N/A	+	d3	P0-3 20%	1, 20%	(76)
Hu, Sus, Mu	AC explants	Explants	N/A	N/A	N/A	+	d3	P0-3 20%	1, 20%	(77)
Hu, Rb	ACs	Type I and III collagen sponge	+	+	N/A	+	d0, d3, d7, d14	P1 21%	3, 21%	(78)
Rb	Articular slices	Explant	−	N/A	N/A	N/A	d7	None	1, 5, 21, 35, 60, 90%	(79)
Rb	ACs	Pellet	+	+	N/A	+	d3, d7, d21	P1 5, 21%	5, 21%	(12)
Sus	ACs	Agarose	+	=	−	N/A	d21, d42	P1 20%	5, 20%	(80)
Sus	Fat pad MSCs	Agarose	+	=	+	N/A	d21, d42	P3 20%	5, 20%	(80)
Sus	AC explants	Explant	N/A	N/A	+	N/A	d6	None	5, 21%	(81)
Sus	BM MSCs	Agarose	+	=	+	N/A	d21, d42	P3 21%	5, 21%	(82)

Green indicates positive results, red indicates negative results and orange indicates no effect on chondrogenesis due to oxygen tension.

Hu = Human, Bo = Bovine, Ga = Chicken, Sus = Porcine, Mu = Murine; ACs = Articular Chondrocytes, MSCs = Mesenchymal Stem Cells, ESCs = Embryonic Stem Cells, NCs = Nasal Chondrocytes, SCs = Sternal Chondrocytes;* p38 human ESCs used, 1 passage (P) thereafter.

Based on the preponderance of evidence from previous studies on articular chondrocytes and their progenitor, mesenchymal stem cells, it was hypothesized that low (5%) oxygen tension would enhance chondrogenesis in terms of the accumulation of GAG and collagen, collagen cross-linking and mechanical strength for both articular and auricular chondrocytes.

Materials and Methods

Chondrocyte harvest, isolation, and expansion

Native ear (auricular) and humeral head (articular) cartilage tissues were harvested from adult, 12–16 month-old, male New Zealand White rabbits in accordance with the guidelines of the ACUC of Case Western Reserve University and the University of Washington. Chondrocytes were isolated from these tissues as previously described (12,13), plated at 5.7 × 10³ cells/cm² in T-175 cell culture flasks (431080; Corning, Lowell, MA) and grown in expansion medium (Dulbecco’s modified Eagle’s medium [DMEM] with 1g/L glucose (Invitrogen, Grand Island, NY) supplemented with 10% fetal bovine serum [FBS, Invitrogen; lot# 281497]) under standard conditions (humidified atmosphere, 37°C, 20% O₂, 5% CO₂). Medium was changed every 3 to 4 days. Chondrocytes were split from each prep at the end of primary (P0) and first passage (P1) into two equal groups and further expanded at 20% or 5% O₂ using a CO₂/N₂ incubator (MCO-18AIC; Sanyo, Tokyo, Japan). Medium changes, passaging, and the preparation for aggregate and engineered cartilage, were performed under standard O₂ conditions as previously described (12).

Aggregate culture and harvest

Aggregates were made from TC-expanded cells as described previously (6). Briefly, 250,000 cells were dispensed into several wells of a sterilized (autoclaved) polypropylene v-shaped bottom 96-well plate (Phenix, Hayward, CA), centrifuged at 590 RCF for 10 min, and placed in chondrogenic medium (serum-free DMEM with 4.5g/L glucose, containing 1% sodium pyruvate, and 1% penicillin/streptomycin [all from Invitrogen], 1% ITS-premix [BD Bioscience], 100 nM dexamethasone [Sigma-Aldrich], and 37.5 µg/mL L-ascorbate-2-phosphate [Wako chemicals]. The aggregates were cultured under the same oxygen condition as that used in expansion, and medium was changed every other day. At 21 days, the aggregates were collected and processed for combined GAG and DNA analysis (3 aggregates from each group), for combined collagen and collagen cross-links analysis (3 aggregates from each group), for lysyl oxidase (LOX) activity assay (3 aggregates from each group), for LOX gene expression analysis (3 aggregates from each group, stored in RNA Later® [Qiagen] at −80°C), and fixed in neutral buffered formalin for histology (1 aggregate from each group). Wet weights of all aggregates were taken after blotting excess media using filter paper.

Scaffold-free tissue-engineered cartilage sheet culture

Chondrocytes were applied to a custom double-diffusion biochamber with porous (10 µm pore diameter) polyester membrane (PET 1009030; Sterlitech, Kent, WA), which had been precoated with 5 µg/ml fibronectin (BD Bioscience, San Jose, CA), as previously described (11), at a density of 3.13 × 10⁶ cells/cm². The cells were cultured under the same O₂ condition as in expansion with bioreactor medium (serum-free DMEM with 4.5 g/L glucose (Invitrogen, Grand Island, NY), containing 1% ITS-premix (BD Bioscience, San Jose, CA), 100 nM dexamethasone (Sigma-Aldrich), 37.5 µg/mL L-ascorbate-2-phosphate (Wako chemicals, Osaka, Japan), and 1% sodium pyruvate, 1% non-essential amino acid, 1% glutamax, and 1% Penicillin/Streptomycin (all from Invitrogen, Grand Island, NY)). The medium was changed every other day. After 3 weeks, cartilage sheets were removed from the biochamber and allowed to free float until 6.5 weeks. Punches (5 mm in diameter) were taken for the mechanical testing from each sheet. Combined GAG/DNA analysis was made in triplicate (3 mm diameter punches), as was collagen/collagen cross-link analysis. Histology was performed on a single neutral buffered formalin fixed 3 mm punch from each sheet. The sheets were made in duplicate from 4 donors.

Assessment of population doublings

After tissue culture expansion under low (5%) or standard (20%) oxygen tension, at the end of P1 and P2 stages, the number of viable cells under both conditions were counted by trypan blue exclusion with a hemocytometer to determine cell doublings per day during expansion, as previously described (24), i.e., Population doubling rate (PD/day) = (log₂ (#cells at the end of passage/#cell seeded))/days in culture.

Biomechanical analysis for scaffold-free engineered cartilage

Mechanical properties of the engineered cartilage sheets, Poisson’s ratio and aggregate modulus, were determined by indentation testing, as previously described (15). Samples were adhered to a stainless steel platen using cyanoacrylate. This maintains the no-slip and no-fluid-flow boundary conditions at the cartilage-substrate interface, which are used as constants in data processing to obtain mechanical properties. During testing, samples were bathed in phosphate-buffered saline. A cylindrical porous stainless steel tip with a diameter of 1.07 mm was used to apply tare and test loads to samples, and displacement was measured using a linear variable transducer. After displacement reached equilibrium under the tare load, the test load was applied and displacement over time was recorded. A 1g test load was chosen to limit the engineered cartilage compressive strain to ~20%. Aggregate modulus and shear modulus were determined as parameters of biphasic stress-strain equations fit to test load-displacement data as described (25,26).

Biochemical analyses

GAG, hydroxyproline (HDP) and DNA analyses

Samples of engineered cartilage were analyzed for their biochemical content. Initially GAG/DNA were assessed as previously described (27), and, later, a combined GAG/HDP/DNA assay was developed. In both cases, samples were digested in papain solution (200 µl of 25 µg/mL papain, 2 mM cysteine, 50 mM sodium phosphate, and 2 mM EDTA adjusted to pH 6.5 [all from Sigma-Aldrich]) at 65° for 3 h. At this point, for the combined GAG/HDP/DNA assay, half of the sample was taken into a new tube and processed as described below. For the GAG/DNA assay portion, papain digested solutions were incubated with 2 volumes of 0.1 M sodium hydroxide (NaOH) for 30 min at room temperature. Two volumes of neutralizing solution was added (5 M sodium chloride and 100 mM disodium phosphate dibasic, pH 7.2; brought to a final concentration of 0.1 M hydrochloric acid). Neutralized samples and a chondroitin sulfate (Seikagaku Chemicals) standard curve were mixed with Safranin-O solution (0.05 % in 50 mM sodium acetate) in duplicate on a dot blot apparatus with a 0.45 µm nitrocellulose membrane (162-015; Bio-Rad, Hercules, CA). Safranin-O was extracted from the nitrocellulose membrane by incubation of the punched dots with 1 ml of 10 % cetylpyridinium chloride (Alpha Aesar), 37 °C for 20 min. This was pipetted into a clear 96-well polystyrene microplate in triplicate, measured by absorbance (536 nm; OPTImax tunable microplate reader with Softmax pro control software ver. 4.3; Molecular Devices LLC., Sunnyvale, CA), and the GAG concentration calculated from the standard curve. For DNA content, aliquots of the neutralized papain digested samples and a calf thymus DNA standard curve were transferred to a black 96-well plate in triplicate, mixed with Hoechst dye (33258; Sigma-Aldrich, 1 mg/ml stock diluted 1/1500 in 0.2 M pH 8 phosphate buffer), and read using a fluorescent plate reader (Ex 365 nm, Em 460 nm; Fusion with Instrument control application ver. 4.02; PerkinElmer Inc., Boston, MA). HDP was assessed on the portion aliquoted prior to alkalization and neutralization using the method previously described (15). Briefly, the samples were hydrolyzed overnight (1 ml 6 M HCl; 110 °C in boil-proof polypropylene tubes), cooled and tubes uncapped then dried at 65–70 °C. Samples and hydroxyproline standards were resuspended in ddH₂O and mixed with 1 volume of 0.15 M copper sulfate, 1 volume of NaOH and incubated at 40 °C for 5 min. To this, 1 volume of hydrogen peroxide (6%) was added and incubated for 10 min at 40 °C. After cooling, 4 volumes of sulfuric acid (6 M) and 2 volumes of Ehrlich’s reagent (5% p-dimethyl-amino-benzaldehyde [Sigma] in 60 % n-propanol, 26 % perchloric acid, 14 % water) were added and incubated at 70 °C for 16 min before cooling and transferred, in triplicate, to a 96-well plate. Absorbance was read at 492 nm and hydroxyproline content calculated from the standard curve. Collagen content was estimated from the hydroxyproline concentration with the conversion factor of 7.6 (28).

Histology

The tissues for histology were fixed with 10% formalin, dehydrated through a graded series of alcohol washes, and embedded in paraffin. Paraffin sections were stained with safranin-O for GAG detection, and Fast Green was used as a counterstain. For immunohistochemistry, slides were deparaffinized, hydrated, and antigen retrieval performed by incubation with pronase (1 mg/ml in PBS containing 5 mM calcium chloride; Sigma-Aldrich) for 10 min at room temperature. Sections were then incubated with PBS/BSA for 10 min at room temperature, then overnight at 4 °C with primary antibody (mouse anti-type II collagen – DSHB II-II6B3, 1/500 in PBS/BSA, mouse anti-typeX collagen – kind gift of Gary Gibson (Henry Ford Hospital, Detroit, MI) - 1/500 in PBS/BSA, mouse anti-elastin – millipore MAB2503, 1/200 in PBS/BSA). After 3 washes, secondary antibody (biotin conjugated goat anti-mouse; Cappel 1/200 in PBS/BSA) was added and sections incubated for 1 h at room temperature. After 3 washes, sections were incubated with streptavidin-HRP (Gibco, 1/200 in PBS/BSA) at room temperature for 30 min. Slides were washed 3 times and detection performed using Vector VIP peroxidase (SK-4600; Vector Labs). Slides were then counterstained with Fast Green (0.0125% in 25 % ethanol, 0.875 % glacial acetic acid) 10 s, then dehydrated and mounted with Permount (Fisher Scientific).

Collagen cross-link analysis

Samples were dried, weighed and acid hydrolyzed in 6M HCl, 110 °C for 24h. Hydroxylysyl pyridinoline (HP) and Lysyl pyridinoline (LP) cross-linking residues were resolved and quantified by C-18 reverse phase HPLC with fluorescence detection (excitation 297 nm, emission 396 nm) and total collagen content was determined as described (29).

Collagen heteropolymer analysis

Samples were dried using a kimwipe and weighed. The heteropolymeric collagen network formed in the samples was depolymerized in equal volumes of 0.5 M acetic acid containing 100 µg/ml pepsin for 18 h at 4 °C. Extracted collagen was analyzed by SDS-PAGE and the separated collagen chains visualized by Coomassie blue staining. Pepsin-extracted type II collagen from rabbit articular cartilage was use as a control. The separated collagen chains were also blotted onto PVDF membrane and probed with mAb 1C10 to identify α1(II) chains. After stripping, the western blot was sequentially probed with mAb 10F2 and pAb 5890 to identify collagen chains cross-linked to the C-telopeptide of α1(II) and to the N-telopeptide of α1(XI) chains. As we have described before, this validates if a heteropolymer of type II and type XI collagen had formed (30).

LOX gene expression analysis

Total RNA was extracted from aggregate samples. Aggregates were thawed in RNA later solution, tissue was transferred to an RNase free custom made stainless steel 24-well pulverizer on dry ice, and pulverized. 300 µL of solution D (4 M guanidium thiocyanate, 25 mM sodium citrate, pH 7.0, 0.5% sarcosyl with DEPC-H₂O, 0.2M sodium acetate (pH 2.0), 7.2% 2-mercaptoethanol [Fisher Biotech, Fair Lawn, NJ]) was added to the crushed sample. This solution would freeze in the well; then, the pulverizer was transferred to a water bath (65 °C, 5–7 min). After all the solutions were thawed, the lysate/homogenates were transferred and further homogenized with QIAshredder (Qiagen), then mixed with phenol/ chloroform/ isoamyl alcohol solution, and total RNA purified following conventional phenol/chloroform extraction methods. RNA purity and concentration was confirmed by NanoDrop analysis (260 nm/280 nm ratio). Reverse transcription reaction was performed with an equal amount of RNA from both hypoxic and standard samples according to the manufacturer’s protocol (Random hexamers; Applied Biosystems [ABI], Foster city, CA). The real-time PCR to detect LOX and 18S gene expression was performed with SYBR green (Maxima SYBR Green qPCR Master Mix; Fermentas/ Thermo Scientific, Eugene, OR) using a 7500 HT real-time PCR system (ABI). Samples were initially denatured at 95 °C for 2 min and then subjected to 40 repeated cycles of: denature at 95 °C for 15 s, annealing at 55 °C for 45 s, elongation at 72 °C for 30 s. Expression levels of LOX were normalized to 18S rRNA levels for each sample. The primer sets were designed as follows: LOX-forward; GACCCGTACAACCCCTACAAG; LOX-reverse; AGCACCCTGTGGTCATAGTCTC; 18S forward; CTCAACACGGGAAACCTCAC; 18S reverse; TTATCGGAATTAACCAGACAAATCG. The human 18S primers (31) were tested in silico against the rabbit RNA sequence (NR_033238.1 using Primer-BLAST) and for primer dimers using Eurofins Oligo Analysis Tool; 100% sequence coverage and acceptable self-complementarity were found. Experimentally, 18S primers were 99.8% efficient for a 5-log dilution of rabbit cDNA with a single melting point of 81.89 °C. LOX primers were designed and tested using Primer-Blast against the predicted rabbit sequence (XM_002710146.1) and acceptable self-complementarity was also found. Experimentally, LOX primers were 98.7% efficient for a 5-log dilution of rabbit cDNA with a single melting point of 86.57 °C.

LOX activity assay

The aggregates for LOX activity assay were homogenized in 4M urea in 0.02M borate buffer (pH 8.2) at 4°C and were centrifuged at 15,000 RCF for 30 min at 4°C, and the protein concentration of the supernatants measured using the bicinchoninic acid assay according to the manufacturer’s instructions (Promega). Aliquots of 10 µg protein extract were analyzed immediately for the assay with or without LOX-specific inhibitor β-propionitrile (BAPN; Sigma-Aldrich). The sample solution was added to the final reaction mixture (50 mM sodium borate [pH 8.2], 1.2 M urea, 50 µM N-acetyl -3,7-dihydroxyphenoxazine (Amplex Red; Molecular Probes-Invitrogen), 0.1 U/ml horseradish peroxidase (HRP; Sigma-Aldrich), and 10 mM 1,5-diaminopentane substrate (Cadaverine dihydrochloride; Sigma-Aldrich) in the presence or absence of 500 µM BAPN (Alfa-Aesar, Ward hill, MA), as previously described (32,33). LOX acts on Cadaverine (pseudosubstrate) and produces H₂O₂, and the HRP-catalyzed oxidation of N-acetyl-3,7-dihydroxyphenoxazine by H₂O₂ produces fluorescent resorufin. The fluorescence was read every 20 min for 2 hours at 37°C using a multimode plate reader (Fusion with Instrument control application ver. 4.02; PerkinElmer Inc., Boston, MA). LOX activity was calculated from the slope which was corrected for non-specific signal by subtracting BAPN-containing values from BAPN-free values, and normalized to the DNA content.

Data analysis and statistics

Each data point represents the mean for replicates within that experiment (replicates ≥ 2); data are then summarized as mean ± standard deviation for each group. An experiment (n) is defined by rabbit donor and passage; numbers of experiments are indicated below graphs and all n are ≥ 4. Experiments were performed at passage 1 and 2. Differences between groups and conditions were assessed by one-way ANOVA, with Sidak’s correction for multiple tests. For tests between the two oxygen tensions with the same passage/donor, a ratio paired t-test was used. Statistical analyses were performed using Graph Pad Prism 6 (Graph Pad Software, Inc.). Values of p < 0.05 were considered to indicate statistically significant differences; * p < 0.05, ** p < 0.01, ***p < 0.0001.

Results

Proliferation rate

Chondrocytes expanded at a faster rate when cultured at 5% O₂ than at 20% O₂ for both auricular and articular chondrocytes. Although the magnitude of this difference was small, its effect is exponential. The proliferation rate of auricular chondrocytes at 5% O₂ was 0.51 ± 0.13 PD/day, 3% higher than cells from the same rabbit at 20% O₂ (0.50 ± 0.13 PD/day; Fig. 1). The proliferation rate of articular chondrocytes at 5% O₂ was 0.28 ± 0.08 PD/day, 16% higher than cells from the same rabbit cultured at 20% O₂ (0.24 ± 0.07 PD/day; Fig. 1). Auricular chondrocytes grew significantly more quickly than articular chondrocytes at either oxygen tension (p < 0.0001).

Figure 1.

Comparison of proliferation rates under different oxygen tensions. Low oxygen tension increased the rate of proliferation for both auricular and articular chondrocytes. This increase was greater for articular chondrocytes (16% vs. 3%). Auricular chondrocytes proliferated 2-fold faster than articular chondrocytes at 20% O₂ and 1.8-fold faster at 5% O₂ (p < 0.0001). Symbols indicate the mean for an experiment, and lines connect experiments. Bars show mean ± S.D.

Engineered cartilage: Aggregate wet weight, GAG, and collagen content

Wet weights of auricular-derived aggregates showed no significant differences between the two oxygen tensions, while articular-derived aggregates at 5% O₂ were significantly heavier than those at 20% O₂ (Fig. 2). The GAG content of auricular aggregates cultured in 5% O₂ was significantly decreased compared to those at 20% O₂ (Fig. 3), while articular-derived aggregates cultured at 5% O₂ tended to have greater GAG content then those cultured at 20% O₂ (Fig. 3). GAG content for auricular chondrocytes cultured at 20% O₂ was significantly higher than all other conditions (p < 0.0001). Collagen content of auricular aggregates was also significantly decreased through culture at 5% O₂ vs 20% O₂, with no apparent effect on articular chondrocytes (Fig. 4).

Figure 2.

Aggregate wet weights from chondrocytes grown and differentiated at 5% and 20% O₂. Wet weight was not significantly different between aggregates cultured at either oxygen tension for auricular chondrocytes but was significantly enhanced by low oxygen tension for articular chondrocytes. Auricular chondrocytes produce aggregates that are significantly heavier regardless of oxygen tension (p < 0.0001). Symbols indicate the mean for an experiment, and lines connect experiments. Bars show mean ± S.D.

Figure 3.

GAG accumulation in aggregates. Auricular chondrocyte aggregates cultured under low oxygen tension accumulated significantly less GAG than those cultured at standard O₂. Articular chondrocytes displayed the opposite trend and accumulated significantly more GAG at low O₂. Auricular chondrocyte aggregates grown at standard O₂ had significantly more GAG than articular chondrocyte aggregates (p < 0.0001). Symbols indicate the mean of an experiment, and lines connect the experiment. Bars show mean ± S.D.

Figure 4.

Collagen accumulation in aggregates. Auricular chondrocyte aggregates cultured under low oxygen tension had significantly less collagen content than those cultured at standard O₂. No consistent effect on collagen content was evident for articular chondrocytes. Symbols indicate the mean from an experiment, and lines connect the experiment. Bars show mean ± S.D.

Lysyl oxidase expression, activity and effect on collagen cross-links

Lysyl oxidase expression was unchanged in auricular chondrocyte aggregates grown at 5% O₂ vs. 20% O₂, while articular chondrocyte aggregates showed an increased lysyl oxidase expression at 5% O₂ vs. 20% O₂ (Fig. 5A). Although the gene expression of lysyl oxidase in auricular aggregates was unchanged by oxygen tension, its activity was significantly decreased in aggregates cultured at 5% O₂ vs. 20% O₂ (Fig. 5B). There was no apparent difference in the activity of lysyl oxidase in the articular chondrocytes at different oxygen tensions (Fig. 5B). This difference in lysyl oxidase activity is reflected in the density of the collagen cross-links found within the tissue, with auricular chondrocytes grown at 5% O₂ having significantly less collagen cross-links than those grown at 20% O₂ (Fig. 6). As expected from the LOX activity, there was no significant effect of oxygen tension on the collagen cross-links in articular chondrocyte matrix.

Figure 5.

Lysyl oxidase expression and activity in aggregates. (A) Oxygen tension had no consistent effect on lysyl oxidase expression in auricular chondrocyte aggregates. LOX trended toward higher expression at low oxygen tension with articular chondrocyte aggregates. Auricular chondrocytes had significantly higher gene expression than articular chondrocytes of lysyl oxidase at each oxygen tension (p < 0.05). (B) In terms of activity, there was a significant decrease in auricular chondrocytes grown under low oxygen tension but no significant effect on the articular chondrocytes. Lysyl oxidase activity was significantly higher per cell in auricular chondrocytes over articular chondrocytes only at 20% oxygen tension (p < 0.05). Symbols indicate the mean from an experiment, and lines connect the experiment. Bars show mean ± S.D.

Figure 6.

Collagen cross-link density in aggregates. Collagen cross-link density was significantly lower for auricular chondrocyte tissue grown at low oxygen tension, but no consistent change was seen for the articular chondrocytes. Auricular chondrocytes grown at standard O₂ had significantly more cross-linking (p < 0.01). Symbols indicate the mean from an experiment, and lines connect the experiment. Bars show mean ± S.D.

Histology

Safranin-O staining showed results compatible with total GAG/DNA analysis (Fig. 2), where 5% O₂ decreased the accumulation of GAG in the matrix of auricular aggregates but increased its accumulation in articular aggregates (Fig. 7A–D). Collagen type II is strongly expressed in auricular aggregates (Fig. 7E–F), but a reduced intensity at 5% O₂ is apparent at higher magnification (Fig. 8E–F). Collagen type II accumulation in the articular aggregates is greater in 5% O₂ conditions than 20% O₂, and appears more evenly distributed in the matrix (Fig. 7G–H). Type X collagen, commonly found in native auricular tissue, is downregulated in 5% O₂ vs. 20% O₂ and is more intracellular (Fig. 7I–J); higher magnification images are shown in Fig. 8I–J. Collagen type X staining is weak in articular chondrocytes and is predominantly cell associated. There is a semblance of more matrix accumulation at 5% O₂ vs. 20% O₂ (Fig. 7K–L). Higher magnification images confirmed this cell associated staining (Fig. 8K–L). Elastin, natively expressed by auricular chondrocytes, is weakly expressed by auricular chondrocytes across all conditions with only 20% O₂ micrographs showing any appreciable accumulation (Fig. 7M–P and Fig. 8M–P).

Figure 7.

Histology and immunohistochemistry. Safranin-O/Fast green staining shows greater accumulation of GAG in auricular chondrocyte aggregates differentiated at 20% O₂ vs. 5% O₂; the reverse is true for the articular cartilage aggregates. Collagen type II staining (E–H) shows no apparent difference in the auricular chondrocyte aggregates but is clearly stronger in articular chondrocyte aggregates at 5% O₂ vs. 20% O₂. Collagen type X staining (I–L) is stronger in auricular chondrocyte aggregates at 20% O₂ vs. 5% O₂ and is more matrix associated. In articular chondrocyte aggregates, there is a slightly stronger staining at 5% O₂ vs. 20% O₂. Elastin staining (M–P) was weak across all samples with a slightly greater accumulation at 20% O₂ for the auricular chondrocyte aggregates only.

Figure 8.

Higher magnification histology and immunohistochemistry. Safranin-O/Fast green staining shows greater accumulation of GAG in auricular chondrocyte aggregates differentiated at 20% O₂ vs. 5% O₂; the reverse is true for the articular cartilage aggregates (A–D). Collagen type II staining (E–H) shows no apparent difference in the auricular chondrocyte aggregates but is clearly stronger in articular chondrocyte aggregates at 5% O₂ vs. 20% O₂. Collagen type X staining (I–J) is stronger in auricular chondrocyte aggregates at 20% O₂ vs. 5% O₂ and is more matrix associated. In articular chondrocyte aggregates there is a slightly stronger staining at 5% O₂ vs. 20% O₂. Elastin staining (M–P) was weak across all samples with a greater accumulation at 20% O₂ for the auricular chondrocyte aggregates only.

Mechanical strength

In order to investigate the effect of low oxygen tension on mechanical strength, scaffold-free tissue engineered sheets were produced for mechanical testing. Auricular chondrocytes failed to produce testable sheets at low oxygen tension (3 of 3 donors), but articular sheets showed a significant increase in stiffness through culture at 5% oxygen tension (Fig. 9). Auricular chondrocyte sheets produced at 20% oxygen tension had aggregate moduli of 0.229 MPa ± 0.04 and native bovine articular cartilage had an aggregate modulus of 0.371 MPa ± 0.07.

Figure 9.

Biomechanical testing of articular chondrocyte sheets. Articular chondrocyte derived, scaffold-free, tissue engineered sheets have greater mechanical stiffness when cultured at low (5%) oxygen tension than at standard (20%) O₂. Auricular sheets did not form testable constructs at low oxygen tension. Symbols indicate the mean from an experiment, and lines connect the experiment. Bars show mean ± S.D.

Collagen heteropolymer analysis

Figure 10A shows that the major pepsin resistant Coomassie blue-stained band in both the auricular and articular chondrocyte cultures migrated identically to the α1(II) chain of type II collagen in the control. β1(II) chains (dimers of α1(II) chains) were also observed in all lanes. Two faintly stained bands migrating slightly slower than that of the α1(II) chains, best visualized in lane 6, were identified by mass spectrometry as the α1(XI) and α2(XI) chains of type XI collagen (data not shown) and also described in (34). No other major pepsin resistant bands were stained, indicating that type II collagen and type XI collagen were the major collagens synthesized by the cultured chondrocytes and the cartilage collagen phenotype was maintained even after 6.5 weeks in culture. The α1(I) and α2(I) chains of type I collagen were not detected in either tissue engineered cartilage. Although immunohistochemistry detected type X collagen in standard (20% O₂) auricular cartilage cultures (Figs. 7 and 8), no pepsin resistant Coomassie blue stained type X collagen chains were observed in the expected 55–60 kDa range (data not shown). The western blot shown in Fig. 10B confirms that the Coomassie blue stained bands were indeed α1(II) chains of type II collagen. Using mAb 1C10, which specifically recognizes type II collagen chains (35), intense staining of both the α1(II) and β1(II) chains were observed. Since equivalent loads were electrophoresed in all the lanes, densitometry of the Western blot revealed moderately higher levels of type II collagen (α1(II) + β1(II)) reactivity in the 20% O₂ auricular culture when compared to 5% O₂. However for the articular cartilage cultures, there were comparable levels at both oxygen tensions. This is consistent with the results in Fig. 4. In the articular cartilage lanes, mAb 1C10 also recognized low molecular weight fragments of α1(II) chains, and occasionally recognized higher molecular weight fragments of β1(II) chains, which have been identified as pepsin over-cleavage products of type II collagen (29).

Figure 10.

Type II and type XI collagen heterpolymer formation in aggregates. (A) Coomasie blue-stained SDS-PAGE gel of pepsin solubilized collagen showing α1(II), α1(XI), α2(XI) and α1(II) chains. (B) Western blot of samples equivalent to those in A and probed with anti-type II collagen antibody (1C10) confirmed type II collagen chains synthesized by auricular and articular chondrocytes under standard and low O₂ conditions. (C) Western blot of samples identical to those in B and probed with antibody 5890. This antibody specifically recognizes the N-telopeptide domain of α1(XI) collagen when cross- linked to chains of α1(II) and β1(II). (D) Western blot of samples identical to those electrophoresed in B (above) and probed with mAb 10F2. This antibody specifically recognizes the C- telopeptide domain of type II collagen when it is cross-linked to α1(II) collagen chains as we have shown before for murine cartilage (34). The antibody also detected the α1(XI) chain in articular chondrocyte matrix grown in 20% O₂ but not under 5% O₂.

Having quantitated that both hydroxylysyl pyridinoline and lysyl pyridinolines cross-link residues in these cultures (Fig. 6), we identified precise domains of collagen chains that are cross-linked using a recently refined western blot method (36). To show that type II and type XI collagen molecules in these cultures were stabilized by these cross-links, we used the pAb 5890 (34,36). As seen in Fig. 10C, this antibody recognized the α1(II) chains and the β1(II) chains of type II collagen. As we have shown before (34,36), this means that the N-telopeptide of the α1(XI) chain to which this antibody was raised was cross-linked to the helical lysine (L930) residue in a fraction of α1(II) chains of type II collagen molecules and thus a hetero-polymer of type II and type XI collagens had formed in the cultures. Under standard culture (20% O₂) conditions, auricular chondrocytes seem to have higher levels of this cross-link when compared to hypoxic conditions. This was not evident for articular chondrocyte cultures, where equivalent levels were observed. A faint reactivity of the α1(XI) chain was observed in standard (20% O₂) articular cartilage culture that probably indicated an N-telopeptide of one α1(XI) chain to α1(XI) helical lysine cross-link of another chain had formed and a homo-polymer of type XI collagen had also formed in these cultures.

Finally, as seen in Fig. 10D, Western blotting using mAb 10F2 (30,36) again recognized the α1(II) and β1(II) chains in all the cultures. This is evidence that the C-telopeptide of the α1(II) chain specifically recognized by this antibody was cross-linked to the helical lysine (L930) residues in the α1(II) collagen chains and, thus, type II to type II collagen cross-links had also formed in these cultures (30,36). Under standard (20% O₂) conditions both auricular and articular chondrocyte cultures seem to have a higher level of these cross-links when compared to hypoxic conditions (compare intensities in Figs. 10B, 10C and 10D). As explained in (36), it must be reiterated that pepsin-extracted α1(II) collagen is devoid of telopeptides unless they are cross-linked to the lysine residues in the helical regions of α1(II) chains. The data confirms that a heteropolymer of collagen type II and collagen type XI is formed in tissue engineered cartilage sheets from both auricular and articular chondrocyte sources, and a mature collagenous network of cross-linked fibrils had formed.

Discussion

The importance of the control of oxygen tension in cell and tissue culture is often overlooked. This report clearly shows a dramatic difference between the response of rabbit articular and auricular chondrocytes to low (5%) vs. standard (20%) oxygen tension during expansion and re-differentiation. While multiple studies have investigated the effect of oxygen tension on articular chondrocytes (Table 1), the effect of oxygen tension on in vitro auricular chondrogenesis is described here for the first time. We have also further clarified the data on articular chondrocyte biology at low (5%), more physiological, oxygen tension.

Articular Chondrogenesis

In scaffold-free culture systems, using both the cell aggregate and sheet approaches to fabricate cartilage, articular chondrocyte constructs had increased GAG accumulation at 5% O₂. Oxygen tension of 5% did not have a positive effect on total collagen accumulation. This is in agreement with several studies (Table 1) including our own on human articular chondrocytes (6). Gel electrophoresis and western blotting also showed that 5% O₂ did not increase cross-linked type II collagen accumulation in the matrix (Fig. 10). In spite of this, mechanical stiffness of the engineered cartilage significantly increased at low O₂ (Fig. 9). This indicates that it may not be the absolute amount of collagen or proteoglycan in the matrix influencing these properties but rather the proportion of collagen to proteoglycan that accumulates in the matrix. Alternatively, it is due to the increase in GAG content while collagen content is maintained above a threshold level.

Lysyl oxidase is an important enzyme in the stabilization of the collagen network in cartilage, necessary in forming aldehyde-derived pyridinoline cross-links between collagen fibers (37). Low oxygen tension increased lysyl oxidase mRNA expression, but this did not translate into increased lysyl oxidase activity. The reason for this is not clear, but Makris et al.,(38)have proposed that lysyl oxidase activity could be reduced due to a lack of copper ions in defined culture medium. Nevertheless, articular cartilage-like pyridinoline cross-links were detected in the type II collagen under both oxygen conditions.

Auricular chondrogenesis

The dramatically different response of auricular chondrocytes to low oxygen tension was somewhat surprising. It is clear that auricular chondrocytes exhibit significantly lower GAG and collagen deposition when cultured at low oxygen tension in comparison to traditional cultures at standard (20%) oxygen tension (Figs. 3 and 4). Although gene expression of lysyl oxidase was comparatively high for auricular chondrocytes over articular chondrocytes in both conditions, lysyl oxidase activity was significantly decreased by low oxygen tension. The diminished lysyl oxidase activity correlated with a reduction in pyridinoline cross-links in collagen at low O₂ tension. Whether the reduction in collagen cross-linking was responsible for the decreased accumulation of GAG and collagen within the construct is not known, but it has been proposed by Asanbaeva et al. (2008) that these molecules are released into the culture media (39). Lysyl oxidase has also been shown to cross-link elastin (40), but as elastin was weakly expressed in chondrogenic culture, this is thought to have had minimal effect on the outcomes reported. Weak elastin staining was also reported by others in auricular chondrocyte constructs (41). The effect of oxygen tension on auricular chondrocytes may be species specific; as rabbits are known to thermo-regulate through their ears (42), the cartilage therein would be expected to experience greater oxygen tension at times. The difference in response could be due to the fact that auricular cartilage is of neural crest origin while articular cartilage is derived from the mesoderm (43,44). Interestingly, Malda et al., (2004) established that human nasal chondrocytes, which are also derived from the neural crest, are more chondrogenic when cultured under reduced oxygen conditions (45). What is evident from our study is that, under in vitro tissue culture conditions, oxygen tension has significant and differing effects on chondrogenesis from two anatomic regions in the same animal.

In a rabbit model (46) and minipig model (47) of articular repair, Lohan et al. found inferior healing in constructs seeded with auricular chondrocytes. Our study could explain their observations. We show that auricular chondrocyte constructs produced at low oxygen tension, as would be expected in the joint, were poorly chondrogenic. Auricular chondrocytes implanted subcutaneously both in humans (17) and in a xenogenic nude mouse model (16,48) seemed to perform well, although neither Xu et al. (16) nor Yokoyama et al. (48) tested the partial pressure of oxygen in their models. The partial pressure of subcutaneous oxygen is 40–50 mmHg which equates to 5.2–6.5% O₂ (49,50). Besides this, there are additional factors benefiting the delivery of oxygen to the in vivo construct, such as haemoglobin’s oxygen release profile (51) and fluid flow mass transfer effects (52), and these make comparisons of chondrogenesis with our static culture conditions difficult to interpret. Auricular chondrocytes also produce glycosaminoglycan and collagen rich constructs in vivo (16,48). Xu et al. presented data indicating that auricular-derived chondrocyte constructs outperformed costal and articular chondrocytes in terms of both GAG and mechanical strength but not in collagen content. This is corroborated by our data, with higher GAG, collagen, collagen cross-linking and collagen type II staining in auricular-derived constructs at standard O₂ when compared to articular chondrocytes. However, auricular chondrocyte sheets cultured in hypoxic conditions failed to produce mechanically testable sheets. It is likely that other differentiation culture systems will be similarly affected by oxygen tension and, like growth factor and media supplements, this is an essential component of in vitro developmental mimicry.

Whilst both articular and auricular cartilage are avascular themselves, the perichondium surrounding auricular cartilage is rich in blood vessels in both humans and rabbits (53). However, auricular cartilage does not benefit from mechanical perfusion mechanisms, creating fluid flow into and out of the tissue, thereby improving nutrient and waste exchange. Future studies investigating the in vivo partial pressure of oxygen in rabbit and human auricular cartilage would be of great use in defining an optimal in vitro culture system. In addition, studies using flow vs. compression to increase the mass transfer effect would investigate the potential differences in articular vs. auricular response to these different mechanisms of nutrition.

In conclusion, oxygen tension has significant effects on chondrocyte differentiation. Specifically, low oxygen tension is beneficial for articular chondrocyte accumulation of GAG, but its effects on collagen accumulation and cross-linking are negligible. Even so, mechanical properties were significantly improved. Scaffold free constructs formed from auricular chondrocytes at low oxygen tension, however, are deficient in GAG and collagen and are structurally insubstantial and mechanically unsound.

Abbreviations used

ACI

autologous chondrocyte implantation

BAPN

β-propionitrile

DEPC

diethylpyrocarbonate

GAG

glycosaminoglycan

HDP

hydroxyproline

ITS

insulin, transferrin, selenium

LOX

lysyl oxidase

MACI

matrix assisted autologous chondrocyte implantation