Fibronectin Splice Variation in Human Knee Cartilage, Meniscus and Synovial Membrane: Observations in Osteoarthritic Knee

Abstract

Objective

Fibronectin (FN) is a widely expressed molecule that can participate in development of osteoarthritis (OA) affecting cartilage, meniscus, and synovial membrane (SM). The alternatively spliced isoforms of FN in joint tissues other than cartilage have not been extensively studied previously. The present study compares FN splice variation in patients with varying degrees of osteoarthritic change.

Design

Joint tissues were collected from asymptomatic donors and patients undergoing arthroscopic procedures. Total RNA was amplified by PCR using primers flanking alternatively spliced Extra Domain A (EDA), Extra Domain B (EDB) and Variable (V) regions.

Results

EDB⁺, EDB⁻ and EDA⁻ and V⁺ variants were present in all joint tissues, while the EDA⁺ variant was rarely detected. Expression levels of EDB⁺ and EDV⁺ variants were similar in cartilage, synovium and meniscal tissues. Synovial expression of V⁺ FN in arthroscopy patients varied with degree of cartilage degeneration. Two V⁻ isoforms, previously identified in cartilage, were also present in SM and meniscus.

Conclusions

Fibronectin splicing in meniscus and SM bears striking resemblance to that of cartilage. Expression levels of synovial V⁺ FN varied with degree of cartilage degeneration. V⁺ FN should be investigated as a potential biomarker of disease stage or progression in larger populations.

Introduction

Osteoarthritis (OA), historically considered a disease of cartilage extracellular matrix (ECM) degeneration, is now appreciated to involve pathologic changes in all connective tissues of the joint, including the synovial membrane (SM), bone, and meniscus¹. Changes in these tissues reflect ECM molecular reorganization, turnover and repair responses. One widely expressed molecule that may play a pivotal role in all of these processes is fibronectin (FN).

FN is a ubiquitously expressed glycoprotein found in ECM that plays important roles in cell adhesion, migration, and differentiation ². Although encoded by a single gene, multiple alternative splicing products exist. Alternative splicing occurs at three major sites: the Extra Domain A (EDA), Extra Domain B (EDB) and Variable (V) regions². Alternative splicing at the EDA and EDB regions results in inclusion or exclusion of one single protein domain in its entirety. In contrast, multiple splice sites within the V region exist and thus multiple V-region variants are observed. Splice variation is associated with different tissue distributions, developmental phases, and functional consequences. FN containing the EDA, EDB and portions of the V region are expressed during embryonic development, but are not highly expressed in adult tissues except during aging ³, wound-healing ⁴, malignancy^{5, 6} and inflammation ⁷. Splice variants containing the EDB region have been described as a marker of angiogenesis ⁸. The EDA region plays a role in inflammatory responses ⁹, and contributes to FN-mediated cellular adhesion and myofibroblast differentiation ^10,11. The V region contains integrin and proteoglycan binding sites and thus also contributes to cell adhesion ¹². This region, also called the type III connecting segment (IIICS), promotes secretion of FN dimers contributing to FN forms found in serum ¹³. In adult human cartilage, a unique splice variant is expressed lacking the entire V-region as well as flanking domains ([V +III15 +I10]⁻ FN) ¹⁴. This isoform may in part prevent inflammatory cell adhesion and invasion into mature articular cartilage matrix ¹⁵.

In the setting of arthritis, variation due to alternative splicing has been observed. FN including the EDA and EDB domains are increased in cartilage from patients with both OA and rheumatoid arthritis (RA) ^{16, 17}. EDA⁺ isoforms have been detected in synovial fluid (SF) from OA and RA patients,¹⁸ and alterations in V region splicing have been demonstrated in cartilage from OA patients²⁶. However, splicing in joint tissues other than cartilage has not been well described previously, despite the importance of tissues such as synovium and meniscus in disease initiation and progression. The ability of FN fragments to potentiate cartilage matrix damage is well established ^{19, 20}, and recombinant FN fragments can induce proteolytic enzyme production by cartilage explants in vitro ²¹. But the contribution of isoforms to these FN-fragment driven activities is not understood.

A better understanding of the tissue sources of FN splice variants within the joint and variation with disease-stage in OA is necessary to determine their contributions to pathogenesis in OA. As OA involves all joint tissues ¹, the present studies were carried out to define FN splice variation in cartilage, SM and meniscal tissues, and determine whether splicing varies according to stage of disease.

Materials and Methods

Cartilage, meniscus and SM specimens

Human articular cartilage, meniscus and SM tissues were collected from four asymptomatic organ donors via the Gift of Hope Organ & Tissue Donor Network (Elmhurst, IL, USA). Specimens were obtained within 24 hours of death, and donated according to Gift of Hope policies. Despite no history of clinical arthritis, cartilage degeneration and bone changes were observed within the joints: all four donors exhibited grade 2 osteoarthritic changes (cartilage fibrillation or fissuring with or without small osteophytes) as evaluated by the modified Collins score ²², although some areas of grade 0 (normal appearing, “non-lesional”) cartilage could be identified.

SM biopsies were obtained during arthroscopic knee procedures, from patients enrolled in the IRB approved Rush Arthritis and Knee Injury Repository; all patients gave informed consent. Patients with ligament tears are excluded from this repository, so the majority of specimens are derived from subjects undergoing partial meniscectomies for degenerative menisci, or loose body removals. Cartilage integrity in these patients was graded intra-operatively using the Outerbridge score, ²³ which ranges from 0 (normal) to 4 (full-thickness cartilage erosion). SM specimens from 15 patients were chosen, representing the range of Outerbridge grades from 0 to 4 (three patients in each grade) (Figure 1A–E). Patient characteristics are presented in Table 1.

Figure 1. Spectrum of cartilage pathology in arthroscopy patients from whom synovial membrane specimens were obtained.

For arthroscopy patients donating synovial membrane, integrity of each articular surface in the knee was graded using the Outerbridge system³² by the surgeon intra-operatively. Pictures representative of each grade are shown (panels A–E). (A) Grade 0: normal cartilage. (B) Grade 1: softening, swelling, minor surface roughening. (C) Grade 2: fibrillation and/or fissuring in an area half an inch or less in diameter. (D) Grade 3: fibrillation and/or fissuring in an area greater than half an inch in diameter. (E) Grade 4: fissuring and erosion of cartilage down to subchondral bone. m: meniscus, f: femoral condyle, t: tibial plateau.

Table 1.

Characteristics of patients undergoing arthroscopic procedures who contributed synovial membrane for analysis.

Outerbridge grade	Gender (M/F)	Ages
0	2M/1F	30,39,46
1	3M	48,49,54
2	3M	41,49,56
3	3M	42,48,52
4	1M/2F	46,60,65

^*Outerbridge grade refers to worst score of all cartilage surfaces in the knee (femoral, patellar, and tibial).

RNA extraction

SM from arthroscopy patients were collected from the operating room, rinsed in sterile PBS to remove contaminating synovial fluid, stabilized in RNAlater solution (Ambion, Inc), and then stored at −80°C until processing. Tissue specimens from the organ donors were treated similarly.

Total cellular RNA was isolated using Trizol reagent (Life Technologies, Rockville, MD), after homogenization. Total RNA was further purified using the RNA Easy kit (Qiagen, Valencia, CA). All RNA was treated with RNase-free DNase I (Qiagen).

Reverse Transcriptase Polymerase Chain Reaction (RT-PCR)

1μg of total RNA from each sample was converted to complementary DNA (cDNA) using EcoDry Premix (Oligo dT, Clontech Laboratories, Inc.) Polymerase chain reaction (PCR) was performed with MyFi DNA Polymerase Mix (Bioline USA Inc.). Primers were designed flanking human FN alternatively spliced regions (EDA, EDB and V), and sequences reported previously ²⁴. Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was amplified as a control. The PCR cycle consisted of an initial denaturation of 95°C for 1 min followed by 94°C for 15 sec, annealing at 55°C–61°C for 15 sec, and extension at 72°C for 30 sec. This cycle was repeated 25–35 times. PCR products were mixed with orange loading dye (LI-COR; Biosciences, Lincoln, NE), separated in 1.2–2% agarose gels, and analyzed using an Odyssey Infrared Imaging System. Where relevant, signal intensities were determined using LI-COR imaging software, and density of specific transcript variants normalized to GAPDH for quantification. Comparison of signal intensity of amplification products from patients with different degrees of cartilage degeneration was performed with the Kruskal-Wallis test followed by Dunn multiple comparison test. Finally, PCR products amplified with the EDV primer pair generated a number of bands. Two unexpected bands were gel purified using QIAquick Gel Extraction Kit (Qiagen), and sequencing of these and other PCR products was performed by Big Dye Terminator v.3.1 Cycle Sequencing Kit (Applied Biosystems, ABI, Foster City, CA). The sequences were analyzed in an ABI-3730 DNA Analyzer.

Results

FN splice variants in different joint tissues

Figure 2 shows the fibronectin splicing profile of RNAs extracted from knee SM, meniscus, and articular cartilage from one of the four organ donors with Collins Grade 2 cartilage changes (fibrillation and fissuring). Cartilage was obtained from both lesional cartilage (LC) and normal appearing cartilage (NC) from the tibial surface, and SM was only obtained from three of these donors. mRNA splice variants containing the EDB region were detectable in cartilage, and were faintly detectable in meniscus (M) and synovial membrane (SM) as well. EDB⁺ FN was present in both normal and lesional cartilage. In contrast, the EDA⁺ transcript (410 bp band) was only faintly detectable in one SM specimen (Figure 2, lane 2) and was absent in meniscus and cartilage. Amplification with V-region primers demonstrated multiple variants. Bands corresponding to multiple V⁺ (1076 to 1243 bp, Figure 2) ²⁵ and V⁻ variants were detectable in all three tissues. A 473 bp band corresponding to the [V+III15+I10]⁻ splice variant was detectable in meniscus, lesional (LC) and normal-appearing cartilage (NC), and faintly in SM. We also observed two bands of apparent sizes 638 and 719 bp in meniscus and cartilage from the donors. After amplification of RNA from normal cartilage using our V-region flanking primers, we purified and sequenced these two bands. These bands were confirmed to be [V+III15]⁻ and [V+I10]⁻ forms as described by Parker et al.²⁶.

Figure 2. Fibronectin splice variants in meniscus (M), synovial membrane (SM), normal cartilage (NC) and lesional cartilage (LC).

Tissues were obtained from four asymptomatic organ donors with Collins grade 2 degenerative changes in the joints. PCR reactions were carried out as described to amplify transcripts containing the alternative spliced regions EDA, EDB, and the V-region. EDB and EDA RT-PCR amplification revealed two forms, representing transcripts with or without the EDA or EDB domains. V region RT-PCR amplification revealed multiple V⁺ and V⁻ transcripts, including the [V+III-15 +I10]⁻, [V+III15]⁻ and [V+I10]⁻ forms. A: Representative PCR products separated by agarose gel electrophoresis from a single donor are depicted. B: Relative expression of V⁺ variants normalized to GAPDH by densitometry. There was a tendency for SM expression to be higher than expression in NC (n=4, P=0.08). C: Relative expression of EDB⁺ variants normalized to GAPDH. Differences between tissues were not statistically significant (n=4, p=0.55).

Among the donor joint tissues, there was a trend for higher synovial expression of EDV+ transcripts (Figure 2b; mean EDV/GAPDH ratio±SD: 0.12±0.01) compared to normal cartilage (0.055±0.01, ANOVA P=0.04, Tukey’s multiple comparisons test P=0.08). Expression levels of EDB+ mRNA (Figure 2c) were similar in cartilage, meniscal and synovial tissues (Figure 2c). Levels of EDB⁺ and EDV⁺ transcripts did not differ between lesional and non-lesional areas of cartilage (P=0.97 for EDB⁺ and P=0.98 for EDV⁺).

FN splice variation in synovial membrane from patients undergoing arthroscopic meniscectomy, with various stages of cartilage degeneration

Given the emerging importance of synovial membrane in OA ²⁷, we next examined FN splice variation specifically in this tissue. Samples from 15 patients undergoing meniscal arthroscopy for degenerative meniscal tears (n=13) or loose body removals (n=2) were evaluated. Patients’ ages ranged from 30 to 65 years, 12 male and 3 female (Table 1). Cartilage integrity in these patients was graded by the surgeon ²⁸ as presented in Figure 1A–E. In each patient, all cartilage surfaces of the knee joint were evaluated, and patients were stratified according to the worst score. Synovial specimens were chosen from fifteen patients representing the spectrum of degeneration from normal to grade 4 (3 patients from each grade). All 15 expressed V⁻ variants including most prominently the 884 bp V⁻ form, as well as the “cartilage-specific” [V+III15+I10]⁻, [V+III15]⁻ and [V+I10]⁻ variants. Figure 3A shows EDV PCR products from a representative set of five patients. We further quantified the density of V⁺ isoforms (1076 to 1243 bp isoforms) normalized to GAPDH. There was a trend for synovial V⁺ variant expression to increase with increasing cartilage Outerbridge score (Figure 3C, n=3 in each group, P=0.09). Figure 3B depicts synovial EDB PCR products from a representative set of five patients with Outerbridge scores ranging from 0 to 4. EDB⁺ to GAPDH ratios were quantified in all 15 patients, but there was no significant association with stage of cartilage degeneration (Figure 3D, P=0.45). All tissues expressed EDA⁻ variants; only one synovial specimen expressed EDA+ variant (data not shown).

Figure 3. FN splice variants in synovial membrane (SM) from patients with various degrees of underlying cartilage degeneration who were undergoing arthroscopic procedures.

Grade refers to Outerbridge grade determined by the surgeon intra-operatively (0–4; 0 = normal, 4= full thickness cartilage erosion). A: All synovial specimens expressed various V⁺ and V⁻ forms, as well as the cartilage-specific [V+III15]⁻ and [V+III15+I10]⁻ variants. B: EDB⁺ and EDB⁻ were present in all SM. C: Relative expression of V⁺ variants normalized to GAPDH by densitometry. D: Relative expression of EDB⁺ variants normalized to GAPDH, analyzed by densitometry. P value calculated by ANOVA. Open triangles in panels C and D represent samples shown in images in panels A and B, respectively.

Comparison of synovial FN expression between patients and asymptomatic donors

Although different grading systems were applied to grade osteoarthritis changes in patients (Outerbridge²³) and donors (modified Collins²²), a limited comparison was done between synovial expression levels of EDB+ and EDV+ variants in patients with Outerbridge grade 2 changes (n=3, fibrillation and/or fissuring in an area half an inch or less in diameter), and donors, all of whom had modified Collins grade 2 changes (cartilage fibrillation or fissuring with or without small osteophytes). Synovial EDB+/GAPDH ratios were higher in grade 2 arthroscopy patients (0.10±0.02) compared with asymptomatic donors (0.04±0.01, P<0.01), while EDV⁺/GAPDH ratios were lower in patients (0.07±0.01) compared to the donors (0.11±0.01, P=0.03).

Discussion

Fibronectin is a versatile protein that is present in multiple forms due to alternative splicing of three regions: Extra Domain A (EDA), Extra Domain B (EDB) and Variable (V) regions. In humans, the V region can be spliced in at least five different ways leading to multiple products. To add to this diversity, in human cartilage the adjacent III-15 and I-10 domains can be spliced out, resulting in [V+III15+I10]⁻, [V+III15]⁻ and [V+I10]⁻ forms ^{14, 26}. In human intervertebral disc tissues ²⁴, we previously reported increased EDB⁺ and EDV⁺ variant species in the annulus fibrosus (AF) tissues that were associated with degree of disc degeneration ²⁴. Because there are similarities between the intervertebral disc and articular joints ²⁸, we conducted the current study to further investigate potential associations between FN splicing and knee joint OA.

FN splicing varies according to cell type, and can be altered in many disease states. However, with the exception of cartilage, patterns of splice variation in other human joint tissues have not been extensively characterized. We first investigated FN splice variants in SM, meniscus and cartilage obtained from four asymptomatic organ donors (Figure 2). Synovial splice patterns were further explored in biopsies from arthroscopy patients with various degrees of cartilage integrity loss (Figure 3). This analysis yielded a number of interesting findings.

EDB⁺ splice variants were detected in meniscus, SM and cartilage tissues from asymptomatic donors (Figure 2), and transcript levels of and EDB⁺ variants did not differ significantly among these tissues (n=4, p=0.55). In normal adult mice, expression of this isoform has been demonstrated in cartilaginous structures ²⁹, but meniscal expression has not been specifically described. To our knowledge, this is the first demonstration of the presence of EDB⁺ FN transcripts in human meniscal tissues, consistent with a previous report of this isoform in human articular cartilage ¹⁷. EDB⁺ FN was also found in synovium of patients undergoing arthroscopic procedures with and without evidence of early cartilage degeneration (Figure 3), and levels were higher than levels in synovium from the asymptomatic donors. A previous study demonstrated protein expression of EDB⁺ FN in SM from RA and OA patients, but did not compare levels with those in asymptomatic tissues ³⁰. The increased synovial levels in symptomatic patients compared with asymptomatic donors is intriguing, but needs to be confirmed by larger studies.

Expression of the EDB domain in adult tissues has been described as a marker of angiogenesis ⁸, as EDB is upregulated around newly forming vascular structures in malignant and non-malignant tissues ^{31, 32}. In OA, the normally avascular articular cartilage becomes invaded by new vascular channels at the interface with the subchondral bone ³³ and angiogenesis is also observed in the meniscus ³⁴ and SM ³⁵. Use of a monoclonal antibody directed against the EDB domain of FN is being investigated as a therapeutic agent to target tumor vasculature ³⁶, and inhibition of angiogenesis has demonstrated utility in a rodent model of OA ³⁷. All tissues in our study expressed the EDB⁺ variant, including tissues from patients without visible cartilage damage, and non-lesional areas of cartilage from the donors. As expression of EDB⁺ variants was found in multiple joint tissues regardless of the presence or absence of cartilage damage, expression of this variant may not be related to structural disease in OA. The function of this splice variant in normal adult human cartilage is unknown.

Our analysis of V region splice variation yielded several interesting findings. First, several V⁻ splice variants were detectable in all joint tissues. Specifically, Figures 2 and 3 show significant SM and meniscus expression of three variants previously described as “cartilage-specific”: the [V120 +III15 +I10]⁻ [V+III15]⁻, and [V+I10]⁻ isoforms ^{15, 26}. Expression of these three variants was observed in SM from all fifteen arthroscopy patients. These biopsies were obtained from the suprapatellar pouch by direct visualization, and care was taken to ensure that only SM was included. Moreover, processing and analysis of synovial mRNA from patients was performed separately from the donor tissues, to prevent contamination from cartilage or meniscal mRNA specimens. Parker and colleagues described cartilage expression of the [V+III15]⁻ and [V+I10]⁻ isoforms in osteoporotic patients, with more limited expression in OA patients ²⁶. Non-cartilaginous joint tissues were not analyzed in their study. The potential functional role for these V⁻ variants in SM is unclear. Another novel finding was the variation of synovial expression levels of V⁺ transcripts according to tissue source and degree of underlying joint degeneration. V⁺ variants were identified in all three joint tissues (Figure 2), and were slightly higher in SM compared with cartilage. In SM from arthroscopy patients (Figure 3), expression levels of V⁺ variants tended to increase with increasing levels of cartilage damage. This finding is consistent with the hypothesis that the synovial reaction in OA is related to cartilage damage precipitated by aging or joint injury ²⁷. The V region of FN contains an α4β1 integrin binding site ¹², and is likely to be important in cell-cell interactions and cellular trafficking during wound healing. It is possible that the synovial expression of the V⁺ isoforms during early and intermediate stages of disease reflects a synovial response to cartilage damage.

We found no evidence of EDA⁺ variant expression at the mRNA level in meniscal or cartilage tissues, and little evidence in synovium, with the exception of low expression in one donor and one patient. Hines et al ³⁸ noted immunostaining for EDA+ FN protein in the synovial lining cells of patients with RA, but not in synovium from four OA patients. These findings are consistent with our current results, as it is possible that expression in the SM occurs in a small minority of OA patients or at very low levels and may be overlooked if analyzing very small numbers of specimens.

In this study, arthroscopy patients with normal cartilage, or with low grades of cartilage degeneration, were younger than those with higher grades of degeneration (Table 1). Because of the small sample number, we cannot rule out an effect of age on V⁺ splice variation in SM. Furthermore, the regulating factors for alternative splicing and the functional importance of these variants remain to be determined. In addition, although synovial expression of EDB⁺ and EDV⁺ variants differed in patients and asymptomatic donors, the value of this comparison is limited, owing to the small number of specimens, and the slight differences in the cartilage grading systems applied to patients and donors. The Collins grade applied to donor tissues accounts for both cartilage degeneration and bone changes, while the Outerbridge system applied to arthroscopy patients solely grades cartilage defects. Nevertheless, the current analysis extends our understanding of FN splice variation in two important ways. First, two specific V⁻ isoforms previously described only in cartilage, were in fact expressed in multiple joint tissues including meniscus and SM. These findings highlight the emerging importance of the meniscus and the SM as active tissues in OA ^{1, 27}. Second, synovial expression of V⁺ isoforms appeared to vary with degree of cartilage damage in patients undergoing arthroscopic knee procedures. These expression patterns suggest potential utility of FN isoforms as markers of disease stage in OA and in other forms of arthritis. Given the functional consequences of FN splicing on cellular responses and interactions with extracellular matrix, our observations may guide further investigation regarding the impact of FN splicing on reparative and inflammatory processes occurring in multiple joint tissues affected by arthritis and joint injury.