Duration of symptoms prior to partial meniscectomy is not associated with the expression of osteoarthritis genes in the injured meniscus

Abstract

While there is emerging data on how duration of symptoms prior to surgery relates to outcomes of patients undergoing arthroscopic partial meniscectomy, little is known about how duration of symptoms relates to the biology of the knee in these patients. The purpose of this study was to test the hypothesis that duration of symptoms prior to arthroscopic partial meniscectomy is associated with expression of osteoarthritis (OA)-related genes in the meniscus. We collected resected meniscus from patients (N = 76) undergoing clinically indicated arthroscopic partial meniscectomy from knees without advanced degenerative changes. RNA from 64 patients was analyzed for 28 candidate OA transcripts by real-time polymerase chain reaction (PCR). RNA was also probed for identification of novel genes by RNA microarrays in 12 patients followed by validation of selected candidates by real-time PCR. The association of gene expression with duration of symptoms prior to surgery was tested. Additional screening was performed with known OA genetic risk alleles assembled from published literature and with gene transcripts differentially expressed between non-OA and OA menisci. Duration of symptoms did not predict expression of OA genes in the meniscus, other than limited association with CXCL3, BMP2 and HLA-DQA1. Microarray identified new genes/pathways with unknown role(s) in meniscus injury and OA and validation of a subset of genes by real-time PCR showed expression pattern highly concordant with the microarray data. While duration of symptoms prior to arthroscopic partial meniscectomy does not significantly alter expression of OA related genes, the association with novel genes/pathways deserves further investigation.

INTRODUCTION

Meniscus tears are a common injury, often treated with arthroscopic partial meniscectomy^1–3. They are likely to be an important early event in the initiation and propagation of osteoarthritis (OA) in the knee^4–7 as they are known to predispose individuals to develop knee OA^8–13. At least 50% of individuals with meniscal tears have radiographic evidence for OA within 10 to 20 years after injury¹⁴. The molecular basis for the initiation and progression of OA following meniscal injury is still largely unknown. A number of factors such as age, body mass index (BMI), tear pattern, concomitant chondral and ligamentous lesions and joint degeneration have been attributed to changes in the molecular biology (i.e. gene expression) of a torn meniscus^15–19.

There are limited data on if and how duration of symptoms influences the clinical outcomes of patients with meniscus tear. While earlier surgical treatment of meniscus tears may increase the likelihood of repair versus resection^20,21, some studies have shown that duration of symptoms is not relevant to clinical outcomes^{22; 23}. Contrasting evidence from other studies suggests that a shorter duration of symptoms prior to partial meniscectomy is associated with a greater improvement in clinical symptoms after surgery^{24; 25}. One study has shown that gene expression in the synovial fluid from patients with meniscus tears is associated with duration of symptoms²⁶ and gene expression has been shown to be associated with time from injury in ACL tears²⁷. The purpose of the current study is to determine if and how duration of symptoms influence the biology of the torn meniscus. We hypothesize that duration of symptoms prior to arthroscopic partial meniscectomy is associated with expression of OA-related genes in the meniscus which could at least partially explain or contribute to worse outcomes in patients with longer duration of symptoms prior to arthroscopic partial meniscectomy^{24; 25}. Understanding if and how gene expression in the meniscus varies with duration of symptoms is important to better understand how the knee responds to meniscus injury and to appropriately interpret other studies of meniscus gene expression.

MATERIALS AND METHODS

Patients

Tissue samples were collected in accordance with the procedures approved by the Institutional Review Board of the study institution. Patients (N = 76) diagnosed with symptomatic tear in the posterior horn of the meniscus not amenable to repair and scheduled for arthroscopic partial meniscectomy were consented prior to surgery (Table 1). None of the patients had any other injury or advanced degenerative changes in the knee based on X-ray or MRI. Patients were excluded from the study if they had more than grade 2 degenerative changes on X-ray or grade 3 or 4 chondrosis on MRI. At the time of surgery, the knee was assessed for any evidence of articular cartilage injury or degeneration. Patients were excluded if the knee had any evidence of cartilage damage greater than focal grade 2 lesions. The surgeon filled out a form with the date, age, sex, and intra-meniscal location of origin for each sample.

Table 1:

Characteristics of study participants

Characteristic	Real-time PCR	Microarray	Combined
N	64	12	76
DOS (days, mean ± S.E.M.)	240.80 ± 39.52	324.06 ± 75.95	253.94 ± 35.36
Age (years mean ± S.E.M.)	45.28 ± 1.63	49.17 ± 2.96	45.89 ± 1.45
BMI (kg/m2, mean ± S.E.M.)	27.74 ± 0.63	26.89 ± 1.12	27.61 ± 0.56
Male/Female	45/19	7/5	52/24

DOS = duration of symptoms; S.E.M. = standard error of the mean; BMI = body mass index

Collection of duration of symptom data

Unlike ACL tears, which rarely occur without a distinct injury, meniscus tears may or may not have a distinct injury or event associated with the onset of symptoms. Therefore, studies refer to duration of symptoms rather than time from injury to determine the chronicity of meniscal injuries^{20; 22; 23; 25; 26; 21}. We prospectively recorded duration of symptoms data based on the clinical information provided at their initial presentation in the office as part of their routine clinical history.

Sample collection

A fragment of injured meniscus clinically indicated for resection was collected from the knee at arthroscopy. All tissue specimens were labeled with the date and a blinded sample number. The samples were placed into tubes of RNAlater solution (Thermo Fisher Scientific) in the operating room immediately after removal from the knee (within 1–2 minutes). The tubes were then stored in the refrigerator until a courier picked them up for refrigerated transport to the laboratory on the day of surgery. Samples were kept for first 24 h in the refrigerator to allow preservation of RNA and then stored at −80°C until total RNA preparation.

RNA extraction

The meniscus tissues were processed for RNA isolation using the recommended protocol¹⁵. Each tissue specimen was homogenized with use of a Polytron homogenizer (Brinkmann Instruments, Westbury, NY). Homogenized sample was transferred to microfuge tubes and incubated at room temperature for 5 min to allow the complete dissociation of nucleoprotein complexes. We added (200 μL) chloroform to each sample, which was then centrifuged at 12,000 g for 15 minutes at 4°C. The upper aqueous phase containing RNA was transferred to a new microfuge tube, and 500 μL isopropanol was added to precipitate RNA. The sample was then incubated at room temperature for 10 minutes followed by centrifugation at 12,000 g at 4°C for 10 minutes. After washing with 75% ethanol, RNA pellet was resuspended in RNase-free water, and finally incubated at 60°C for ten minutes. RNA was finally purified with the use RNeasy Mini Spin Columns (Qiagen) according to instruction manual. RNA amount was determined with the use of a Nanodrop spectrophotometer (Thermo Fisher Scientific, Wilmington, DE). On an average, we obtained per sample RNA concentration of 86.13 ± 10.01 ng/μl as measured by Nanodrop and RNA integrity number of 6.63 ± 0.12 as measured by Bioanalyzer (Agilent Technologies, Santa Clara, CA).

cDNA synthesis

Prior to reverse transcriptase-polymerase chain reaction (RT-PCR), we treated 200 ng of RNA with DNase I (Invitrogen) to remove any genomic DNA contamination. Briefly, amplification grade DNase I (1 U/μL), and 2 μL of DNase-I reaction buffer were added to RNA preparation in a total volume of 20 μL. The reaction was kept at room temperature for 15 min prior to inactivation of DNase-I by 2 μL of 25 mM EDTA at 65°C for 10 min. To synthesize first strand cDNA, the DNase-I-treated RNA was then reverse-transcribed using random hexamers and the SuperScript II First-Strand Synthesis System (Invitrogen) as per the manufacturer’s instructions. Briefly, a mixture of the random primer (50 ng/μL) dNTP mix (10 mM) and DNase-I-treated RNA was incubated at 65°C for 5 min. The reaction mixture was quenched on ice, and the constituents were collected by quick centrifugation. Reverse transcription was continued following addition of a master mix of 0.1 M DTT [dithiothreitol], 200 units/μL SuperScript II, and 5× first-strand buffer. The reaction mixture was incubated at 42°C for 50 min and terminated at 70°C for 15 min.

Real-time PCR

We examined the expression of 28 OA-related transcripts including proinflammatory cytokines (IL1A, IL1B, IL6, TNFA), aggrecanases (ADAMTS4, ADAMTS5), chemokines (IL8, CCL3, CCL3L1, CXCL1, CXCL3, CXCL6, CCL20), matrix components (BMP2, COL1A1, COL2A1, ACAN), metalloproteinases (MMP1, MMP3, MMP9, MMP13), and transcription factors (NFKB2, NFKBIA, IKBA). Custom-designed primers¹⁵ were purchased from Invitrogen. Gene expression analysis was performed on a 7500 Fast Real-Time PCR System (Applied Biosystems). The housekeeping gene used was glyceraldehyde 3-phosphate dehydrogenase (GAPDH) for normalization of fluorescence threshold (C_t) values of target genes. We used 10 μL of SYBR Green PCR Master Mix (Applied Biosystems), 1.5 μL of cDNA, and 200 nM of primers for each reaction. Amplification was performed with an initial activation step at 95°C for 10 min, followed by 40 cycles of denaturation at 95°C for 15 sec and annealing at 60°C for 60 sec. The C_t(fluorescence threshold) values for GAPDH and the genes of interest were normalized to GAPDH.

Microarrays

RNA from 12 samples was prepared essentially as reported previously²⁸. Prior to microarray hybridization, the RNA integrity numbering system was used to assess the quality of the total RNA samples by Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA). RNA was processed per manufacturer’s protocol using Affymetrix Gene 2.0 ST platform (Affymetrix Inc., Santa Clara, CA) to undertake transcriptome expression analysis as previously described²⁸. Total RNA was amplified by Ovation Pico WTA Amplification System (NuGen, Inc., San Carlos, CA) and biotinylated cDNA synthesis was undertaken according to the standard NuGen Ovation Pico WTA System Amplification system using 50 ng total RNA. NuGen Encore Biotin Module kits were then used to fragment and label 5 μg of cDNA hybridized into the GeneChip Hybridization Oven 640 for 18 hours on Gene 2.0 ST array. Then Affymetrix Gene 2.0 microarrays were washed and processed in the Affymetrix Fluidics Station 450. Finally, GeneChips were scanned using the Affymetrix GeneChip 3000 7G Scanner.

The expression data were analyzed and processed with the GeneChip Command Console (Affymetrix, Inc., Santa Clara, CA) and the raw (.CEL) files generated were analyzed using Expression Console software with Affymetrix default RMA (Robust Multichip Analysis) gene analysis settings. Probe summarization RMA, quality control analysis, and probe annotation were performed according to recommended guidelines (Expression Console Software, Affymetrix, Inc., Santa Clara, CA). Data were normalized and log₂ transformed before downstream analysis, which were done using Partek Genomic Suite software version 6.4 (Partek Inc. St. Louis, MO). Data were adjusted for age, BMI, and sex because initial analysis showed that these are confounding factors with different confounding effects for meniscus^{16; 17} We set an FDR (false discovery rate) at 0.1 (10% level) and P value set at 0.05. All microarrays were performed at Genome Technology Access Center of the study institution. To facilitate the interpretation of differentially expressed transcripts, we used gene ontology analysis using GeneGo MetaCore tools to classify according to their biological functions (https://portal.genego.com/). The raw array data were deposited into the National Center of Biotechnology Information Gene Expression Omnibus (GEO) repository and are accessible under the accession number (GSE66635) (http://www.ncbi.nlm.nih.gov/projects/geo).

Comparison with OA risk-alleles and OA transcripts

As the risk-alleles associated with OA may affect the expression level of their tagged genes, we assembled a list of susceptibility alleles (54 genes) from the available literature²⁹ and expression levels were probed in our dataset. We collected the genes around OA-associated genetic risk-alleles from genome wide association studies (GWAS) that were reported in published GWAS studies²⁹ and screened in our microarray dataset to investigate if the OA-associated genes were changes with duration of symptoms (Supplementary Table 1). We also used a previously published list of OA-related transcripts (63 transcripts) generated by microarray analysis of healthy and OA cartilage³⁰ to evaluate our microarray dataset (Supplementary Table 2).

Comparison with OA transcripts from meniscus tear and OA patients

We also screened our own previously published list of 176 gene transcripts generated by microarray analysis of non-OA meniscus (resected during arthroscopic partial meniscectomy) and OA meniscus (resected at time of total knee replacement)³¹, to evaluate if any of these transcripts found in our previous analysis were associated with duration of symptoms (Supplementary Table 3).

Validation of microarray expression data by real-time PCR

We validated the expression of 8 selected genes by real-time PCR (using methods described above) that were significantly differentially correlated with duration of symptoms (4 genes were positively correlated; 4 genes were negatively correlated) from microarray analysis. We performed real-time PCR on all 12 samples on which microarray was conducted using custom-designed primers (Table 2). PPIA (peptidyl prolyl isomerase A) was used as housekeeping gene.

Table 2:

List of gene-specific primers for real-time PCR to validate microarray experiment

GeneBank ID	Gene symbol	Gene name	Primer	Sequence	Location (from)	Location (to)	Product length
NM_000053.4	ATP7B	ATPase copper transporting beta	Forward	5’-CCTGGTCCTGGCACTGATTT-3’	4085	4104	127 bp
			Reverse	5’-CACAGACACAGAGGAGGCTG-3’	4211	4192
NM_053039.2	UGT2B28	UDP glucuronosyltransferase family 2 member B28	Forward	5’-AGGACTCCACTGCAAACCTG-3’	860	879	136 bp
			Reverse	5’-TACGTTGGCCCTTTCTGCTG-3’	995	976
NM_001257268.2	RD3L	Retinal degeneration 3 like	Forward	5’-CTGACAGAGGGAGTGAGCAA-3’	834	853	90 bp
			Reverse	5’-TCTTGGAGCTGTCACCTTGG-3’	923	904
NM_020209.4	SHD	Src homology 2 domain containing transforming protein D	Forward	5’-CTGTCCCTCTGCAAGGAAGG-3’	1558	1577	106 bp
			Reverse	5’-ACTTCAGATGCAGGAAGCCC-3’	1663	1644
NM_001002917.1	OR8D1	Olfactory receptor family 8 subfamily D member 1	Forward	5’-CCCTAGCTGTTGCTGTCTCC-3’	632	651	110 bp
			Reverse	5’-AGCCATGAGATGAGAGCTGC-3’	741	722
NM_000905.4	NPY	Neuropeptide Y	Forward	5’-ACCTCATCACCAGGCAGAGA-3’	256	275	105 bp
			Reverse	5’-CAAGCCGAGTTCTGGGAACA-3’	360	341
NM_015916.5	CALHM2	Calcium homeostasis modulator family member 2	Forward	5’-GCTCAGGTATGAGTCCCAGC-3’	1113	1132	107 bp
			Reverse	5’-GGTAGCTGAGTGGTGAGCAG-3’	1219	1200
NM_001006933.1	TCEAL3	Transcription elongation factor A like 3	Forward	5’-CCCAAGGACTCTCAGGAGGA-3’	557	576	91 bp
			Reverse	5’-CCTCTTGAGCCCTTGACACA-3’	647	628
NM_021130.4	PPIA	Peptidylprolyl isomerase A	Forward	5’-TCTGCACTGCCAAGACTGAG-3’	430	449	117 bp
			Reverse	5’-TGGTCTTGCCATTCCTGGAC-3’	546	527

bp = base pair

Statistical analysis

We tested the correlation between the expression level of OA-related genes and duration of symptoms. Patient age and BMI have been shown to affect gene expression levels in the meniscus and modify the correlation^15–18. Characteristics such as age, BMI and duration of symptoms of the study participants are shown as mean ± standard error of the mean. To control for these effects, we computed Spearman rank partial correlations between gene expression level and symptom duration that controlled for age-corrected BMI values. To correct BMI values, we first regressed BMI on age, and then for each patient we added the residual to the average BMI. Calculations were performed in SPSS version 12.0 (IBM, North Castle, NY), using P values of 5% as evidence of statistical significance. Pearson’s correlation coefficients were computed for 8 genes, which were validated by real-time PCR using GraphPad Prism v7.03 (San Diego, CA).

RESULTS

Patient characteristics

The overall cohort consisted of 76 patients (24 females and 52 males) with symptomatic medial meniscus tears indicated for arthroscopic partial meniscectomy (Table 1). The mean age of the patients was 45.89 ± 1.45 years and mean BMI was 27.61 ± 0.56 kg/m². Mean duration of symptoms was 253.94 ± 35.36 days. The real-time PCR cohort included 64 patients (19 females and 45 males) with a mean age of 45.28 ± 1.63 years and mean BMI was 27.74 ± 0.63 kg/m². Mean duration of symptoms was 240.80 ± 39.52 days. The microarray array cohort had 12 patients (5 females, 7 males) with a mean age of 49.17 ± 2.96 years, mean BMI of 26.89 ± 1.12 kg/m² and mean duration of symptoms of 324.06 ± 75.95 days. Some of these data have been used as the basis for previous publications^{15; 31}, none of which included any information or analysis related to duration of symptoms.

Association of OA-related transcripts with duration of symptoms

We observed that none of the genes tested by real-time PCR were significantly associated with duration of symptoms (Table 3). While none of the genes tested reached a formal level of significance, only MMP13 was significant at a 10% level of significance, however, the correlation coefficient was very small (r = 0.22). Thus, these data showed no correlation of OA genes with duration of symptoms.

Table 3:

Partial correlation coefficients for 28 genes tested in relation to duration of symptoms (real-time PCR)

Gene symbol	Gene name	Correlation coefficient	P value
IL1A	Interleukin 1 alpha	0.07	0.569
IL1B	Interleukin 1 beta	0.10	0.439
IL6	Interleukin 6	0.07	0.593
TNFA	Tumor necrosis factor alpha	0.10	0.431
ADAMTS4	Disintegrin and metalloproteinase with thrombospondin motif 4	−0.06	0.644
ADAMTS5	Disintegrin and metalloproteinase with thrombospondin motif 5	0.11	0.401
BMP2	Bone morphogenetic protein 2	0.13	0.306
MMP1	Matrix metalloproteinase 1	0.10	0.439
MMP3	Matrix metalloproteinase 3	0.22	0.080
MMP9	Matrix metalloproteinase 9	0.02	0.903
MMP13	Matrix metalloproteinase 13	0.21	0.104
IL8	Interleukin 8	0.14	0.285
CCL3	C-C motif chemokine ligand 3	0.16	0.197
CCL3L1	C-C motif chemokine ligand 3 Like 1	0.02	0.878
CXCL1	C-X-C motif chemokine ligand 1	0.07	0.572
CXCL3	C-X-C motif chemokine ligand 3	0.07	0.570
CXCL6	C-X-C motif chemokine ligand 6	0.05	0.677
CCL20	C-C motif chemokine ligand 20	−0.05	0.709
COL1A1	Collagen, type I, alpha 1	0.17	0.178
COL2A1	Collagen, type II, alpha 1	0.10	0.416
ACAN	Aggrecan	0.09	0.465
NFKB1A	Nuclear factor kappa B subunit 1	0.10	0.447
NFKB2	Nuclear factor kappa B subunit 2	0.13	0.315
IKBA	nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha	0.15	0.248
ADIPOQ	Adiponectin	0.03	0.799
APLN	Apelin	−0.09	0.499
LEP	Leptin	0.14	0.277
RETN	Resistin	−0.13	0.311

Association of microarray data with duration of symptoms

We explored the microarray dataset in an attempt to see if any of the aforementioned genes were significantly associated with duration of symptoms. Although none of the real-time PCR showed significant association with duration of symptoms (Table 4), we found two of these 28 genes were present in the microarray analysis. While BMP2 showed negative correlation with duration of symptoms (r = −.63, P = 0.035), CXCL3 was positively correlated (r = 0.73, P = 0.004).

Table 4:

Number of genes tested for relationship with duration of symptoms

Screening cohort	Original		Validation
Analysis method	Real-time PCR	Microarray	Microarray	OA GWAS	OA transcripts	Meniscus genes (elevated in OA)	Meniscus genes (repressed in OA)
Genes tested	28	27,641	28	54	63	80	96
Genes passed	0	903	2	1	2	2	1
Percentage (%)	0.00	3.27	7.14	1.85	3.17	2.50	1.04

Microarray identified genes and pathways associated with duration of symptoms

The microarray analysis identified 903 gene transcripts associated with duration of symptoms (343 gene transcripts were positively associated while 560 gene transcripts were negatively associated) (Table 5; Supplementary Table 4). These gene transcripts represented several important biological processes, as 44 biological processes were positively associated with duration of symptoms while 38 were negatively associated with duration of symptoms (Table 6).

Table 5:

Selected gene transcripts associated with duration of symptoms (microarray)

Gene symbol	Gene name	P value	Correlation coefficient	Gene symbol	Gene name	P value	Correlation coefficient
Genes positively associated with duration of symptoms				Genes negatively associated with duration of symptoms
ATP7B	ATPase copper transporting beta 7	0.003	0.86	OR8D1	Olfactory receptor family 8 subfamily D member 1	0.001	−0.89
UGT2B28	UDP glucuronosyltransferase family 2 member B28	<0.0001	0.85	NPY	Neuropeptide Y	0.002	−0.88
RD3L	Retinal degeneration 3 like	0.001	0.85	FLJ42351	Uncharacterized LOC400999	0.001	−0.87
SHD	Src homology 2 domain containing transforming protein D	0.003	0.84	CALHM2	Calcium homeostasis modulator 2	0.001	−0.85
CSNK1E	Casein kinase 1 epsilon	0.002	0.83	TCEAL3	Transcription elongation factor A like 3	0.002	−0.84
PCF11	PCF11 cleavage and polyadenylation factor subunit	<0.0001	0.82	KIF1C	Kinesin family member 1C	0.002	−0.84
TMPRSS11BNL	TMPRSS11B N-terminal like	<0.0001	0.82	PLEKHN1	Pleckstrin homology domain containing N1	0.003	−0.83
CPLX4	Complexin 4	0.003	0.82	NCAPH2	Non-SMC condensin II complex subunit H2	0.004	−0.82
FAM71C	Family with sequence similarity 71 member C	0.004	0.80	CFHR3	Complement factor H related 3	0.004	−0.82
TRPC5OS	TRPC5 opposite strand	0.002	0.80	SART1	Squamous cell carcinoma antigen recognized by T-Cells 1	<0.0001	−0.82
ABCC13	ATP binding cassette subfamily C member 13	0.004	0.80	EIF1B-AS1	EIF1B antisense RNA 1	0.001	−0.82
OR5M8	Olfactory receptor family 5 subfamily M member 8	0.007	0.80	FAM196B	Family with sequence similarity 196 member B	0.004	−0.82
OR10A6	Olfactory receptor family 10 subfamily A member 6	0.009	0.79	RNA5SP354	RNA, 5S ribosomal pseudogene 354	0.002	−0.81
FAM41C	Family with sequence similarity 41 member C	0.006	0.79	CLEC4F	C-type lectin domain family 4 member F	0.007	−0.81
SLC10A2	Solute carrier family 10 member 2	0.008	0.78	NFATC2IP	Nuclear factor of activated T-cells 2 interacting protein	0.007	−0.81
SIGLECL1	SIGLEC family like 1	0.001	0.78	GFAP	Glial fibrillary acidic protein	0.005	−0.81
RNU1–23P	RNA, U1 small nuclear 23	0.013	0.77	ACTRT3	Actin related protein T3	0.004	−0.80
ETNK1	Ethanolamine kinase 1	0.001	0.77	NELFB	Negative elongation factor complex member B	0.005	−0.80
FBXO39	F-Box protein 39	0.006	0.77	GNAL	G protein subunit alpha L	0.004	−0.80
OR2J2	Olfactory receptor family 2 subfamily J member 2	0.012	0.77	MAD2L1	MAD2 mitotic arrest deficient-like 1	0.005	−0.80

Table 6:

Biological processes associated with duration of symptoms in meniscus

Processes	P value	FDR	Processes	P value	FDR
Processes positively associated with duration of symptoms			Processes negatively associated with duration of symptoms
muscle cell differentiation	9.138E-09	4.245E-05	regulation of synaptic transmission, glutamatergic	5.832E-10	3.423E-06
regulation of G-protein coupled receptor protein signaling pathway	2.133E-08	4.954E-05	activation of phospholipase C activity	3.176E-08	7.858E-05
muscle fiber development	3.760E-08	5.822E-05	activation of protein kinase activity	4.016E-08	7.858E-05
regulation of signaling	5.402E-08	6.273E-05	negative regulation of cAMP biosynthetic process	9.315E-08	1.132E-04
response to organic substance	8.437E-08	7.838E-05	negative regulation of synaptic transmission, glutamatergic	9.643E-08	1.132E-04
striated muscle cell differentiation	1.452E-07	9.832E-05	positive regulation of phospholipase C activity	1.648E-07	1.312E-04
regulation of cell communication	1.482E-07	9.832E-05	regulation of synaptic transmission	2.348E-07	1.312E-04
regulation of neurotransmitter secretion	2.077E-07	1.180E-04	regulation of phospholipase C activity	2.697E-07	1.320E-04
response to oxygen-containing compound	2.425E-07	1.180E-04	neurogenesis	5.700E-07	2.326E-04
regulation of cellular localization	2.540E-07	1.180E-04	positive regulation of arachidonic acid secretion	5.945E-07	2.326E-04
positive regulation of response to stimulus	3.982E-07	1.682E-04	positive regulation of phospholipase activity	8.374E-07	3.072E-04
negative regulation of cAMP metabolic process	5.217E-07	1.861E-04	negative regulation of adenylate cyclase activity	9.193E-07	3.174E-04
positive regulation of biological process	5.362E-07	1.861E-04	neural precursor cell proliferation	1.058E-06	3.449E-04
skeletal muscle fiber development	6.320E-07	1.861E-04	positive regulation of protein kinase activity	1.325E-06	3.950E-04
muscle cell development	6.874E-07	1.861E-04	regulation of blood circulation	1.410E-06	3.950E-04
smooth muscle contraction	7.363E-07	1.861E-04	regulation of cation channel activity	1.548E-06	3.950E-04
glutamate receptor signaling pathway	7.614E-07	1.861E-04	positive regulation of MAP kinase activity	1.772E-06	4.160E-04
myotube differentiation	8.870E-07	2.060E-04	regulation of protein kinase activity	2.287E-06	4.873E-04
response to lithium ion	9.677E-07	2.113E-04	positive regulation of MAPK cascade	2.302E-06	4.873E-04
G-protein coupled glutamate receptor signaling pathway	1.001E-06	2.113E-04	generation of neurons	2.325E-06	4.873E-04

FDR = false discovery rate

Association of OA risk alleles and OA transcripts with duration of symptoms

Interestingly, we observed that only one (1.85%) out of 54 GWAS genes was significant with microarray data (ITIH4), and only two (3%) genes from 63 OA-transcripts were associated with duration of symptoms in our microarray analysis (Table 4), CXCL3 (r = 0.73, P = 0.004) and HLA-DQA1 (r = 0.74, P = 0.018).

Association of OA transcripts in meniscus with duration of symptoms

We have recently published a list of gene transcripts. Cross-check of the 176 previously published gene transcripts differentially expressed between non-OA and OA meniscus³¹, only 2 out of 80 transcripts that were more highly expressed in OA meniscus (2.5%) were correlated with duration of symptoms: LTF (r = 0.70, P = 0.021) and ID4 (r = 0.72, P = 0.011). From the 96 transcripts that were down-regulated in OA meniscus, only one gene (1%), namely D2HGDR (r = −0.72, P = 0.011), was associated with duration of symptoms (Table 4).

Validation of microarray data

Our data showed the expression patterns of the eight validated genes by real-time PCR were highly consistent with that in the microarray experiment as shown in Fig. 1. Four genes that showed significant and positive correlations with the duration of symptoms in microarray experiment also showed highly significant and positive correlations with the duration of symptoms in real-time PCR experiment: ATP7B (r = 0.87, P = 0.0002), UGT2B2B (r =0.92, P < 0.0001), RD3L (r = 0.85, P = 0.0005) and SHD (r = 0.83, P = 0.0008). Likewise, four genes that showed significant negative correlations with the duration of symptoms in microarray experiment also exhibited significant and strong negative correlations with the duration of symptoms in real-time PCR experiment: OR8D1 (r = - 0.91, P < 0.0001), NPY (r = - 0.74, P = 0.0061), CALHM2 (r = - 0.59, P = 0.0451) and TCEAL3 (r = - 0.85, P = 0.0005).

Figure 1: Validation of microarray expression data by real-time PCR.

We selected eight genes from microarray experiment to validate by real-time PCR. We observed that four genes namely ATP7B, UGT2B2B, RD3L and SHD, which were significantly positively correlated with duration of symptoms in microarray experiment were also significantly positively correlated with duration of symptoms in real-time PCR experiment. Similarly, four genes namely OR8D1, NPY, CALHM2 and TCEAL3, which were significantly negatively correlated with duration of symptoms in microarray experiment also exhibited significant and strong negative correlations with duration of symptoms in real-time PCR experiment. Pearson correlation for best-fitted curve (solid line) with 95% confidence interval (dotted lines) from the curve is shown for 12 patients.

DISCUSSION

There is little association between expression of OA-related genes in resected meniscus and duration of symptoms in patients undergoing arthroscopic partial meniscectomy. Although microarrays suggested a few OA genes such as BMP2 and CXCL3 may be associated with duration of symptoms, independent validation of these data by real-time PCR in a larger cohort did not hold true. Thus, duration of symptoms has little influence on the expression of classical OA genes in patients undergoing arthroscopic partial meniscectomy, even though other patient-related factors such as age, BMI, and sex, as well as tear pattern, are related to gene expression in the injured meniscus^15–19. While gene expression in anterior cruciate ligament tears is associated with time from injury²⁷ and gene expression of synovial fluid in patients with meniscus tears has been shown to have some association with duration of symptoms²⁶, this does not appear to be true for the meniscus itself.

Only one GWAS gene, namely ITIH4, was significant in our microarray dataset. ITIH4 is a member of an inter-alpha- trypsin inhibitor heavy chain H family. Expression of this gene has been shown to increase in OA cartilage and synovial fluid³² and in the serum of patients with rheumatoid arthritis, however, data regarding it is expression in a setting of a meniscus tear is lacking³³. A few genes with a known role in OA were found to be significantly associated with duration of symptoms as demonstrated by microarray analysis. While BMP2 showed negative correlation with duration of symptoms, CXCL3 and HLA-DQA1 were positively correlated. BMP2 is a member of bone morphogenetic proteins and is a potent inducer of bone and cartilage formation and a key player in skeletal homeostasis³⁴. Absence ofBMP2 has been reported to alter the extracellular matrix and shape of the meniscus, resulting in functional deficits in meniscus (and articular cartilage) that ultimately result in progressive OA³⁵. This suggests that decreasing levels of BMP2 with longer duration of symptoms could exacerbate joint pathology. CXCL3 is well known as powerful neutrophil chemoattractant and displays increased expression in OA³⁰. It has been shown that meniscus cells or explants express high levels of CXCL3 after treatment with pro-inflammatory stimuli³⁶. Increased levels of CXCL3 with longer duration of symptoms may reflect elevated inflammatory signals after meniscus injury and early evidence of OA. Finally, HLA-DQA1, a member of HLA class II alpha chain paralogues, has been implicated in susceptibility to knee OA³⁷. Increased expression of HLA-DQA1 with duration of symptoms suggests it could contribute to the development of OA following meniscal injury.

This investigation demonstrates that there is limited association of OA gene expression profiles of the injured meniscus with duration of symptoms at the time of surgery. While duration of symptoms has little association with expression of 28 transcripts commonly related to inflammation, OA and obesity, some non-OA transcripts associated with duration of symptoms identified by microarrays have been presented and deserve further detailed investigation.

The imprecision of the duration of symptoms with meniscus tears is an obvious limitation of the current study. Unlike other knee pathology such as ACL tears, patients with meniscus tears cannot always pinpoint the inciting injury. As such, it is possible the meniscus tear was present for some time prior to symptoms and patient recollection of the exact onset of symptoms is likely imprecise, particularly since studies have shown that a significant proportion of knees can have asymptomatic meniscus tears noted on MRI^{24; 25; 38}. While the duration of symptoms is admittedly not entirely accurate with regards to the presence of a meniscus tear, it does serve as a reasonable proxy for the length of time that the knee has been reactive to the meniscus tear. Previous studies have shown that duration of symptoms is a predictor of clinical outcomes following this surgery^{24; 25}. This makes duration of symptoms potentially relevant in terms of assessing the biological impact of the tear on the knee, despite its limitations. Moreover, while correlation analysis clearly demonstrates a linear relationship of gene expression with duration of symptoms, the proportion of gene changes over time may not have biological significance. It is important to note an inherent limitation of the current study design is that statistical significance does not always imply biological significance. Without analyzing the downstream effects of the significant genes, it is impossible to conclude whether the magnitude of the differences in gene expression are likely to have biological significance. Further studies are needed to answer that question for genes of interest. The small sample size may be not be adequately powered in terms of number of subjects or range of duration of symptoms to pick up all clinically relevant associations with duration of symptoms. The study may be especially underpowered at the extreme durations of symptoms and should be interpreted accordingly. This does not diminish the value of the significant relationships that were discovered, however. It is important to emphasize that these findings are in patients without OA or advanced degenerative changes in the articular cartilage of the knee, a different condition from the more common degenerative meniscus tears in knees with moderate, or advanced, joint degeneration. Although this limits the generalizability of these findings across all types of meniscus tears, it was by design in order to focus on the biology of knees early in the process of joint degeneration. While all samples came from the posterior horn and/or body of the medial meniscus, we did not adjust for the precise area of origin for each sample or the type of tear. We did not have any a priori rationale for why these variables would relate to the effect of duration of symptoms on gene expression. Moreover, given the large number of samples, we believe this heterogeneity is a strength and not a weakness as it makes our findings more generalizable. Future investigations should further refine our understanding of these relationships. Variation in the clinical history and/or presentation could be a significant confounder and deserves further consideration in future investigations. Finally, we do not have any clinical or radiographic follow-up to assess if and how differences in gene expression are associated with degenerative changes over time.

Despite these limitations, this is the first study to analyze the association of gene expression of the meniscus with duration of symptoms in patients undergoing arthroscopic partial meniscectomy. Commonly studied OA genes are not associated with duration of symptoms in these patients. This is important information, particularly for interpreting studies of meniscal gene expression, in order to better understand how meniscus tears impact the biology of the knee. A number of novel gene transcripts and biological processes associated with duration of symptoms based on microarray analysis warrant further study for their possible contribution to the development of OA following meniscal injury. Ultimately, this information has the potential to advance our understanding why and how OA develops in the knee after meniscal injury.