MicroRNA-223 demonstrated experimentally in exosome-like vesicles is associated with decreased risk of persistent pain after lumbar disc herniation

Inbred female Lewis rats were used in the experiments. The animals were sedated with isoflurane gas and anesthetized with urethane (~2 g/kg bodyweight, i.p.). Absence of paw withdrawal and ear reflex to pinch indicated adequate surgical anaesthesia. The rat core temperature was maintained at 36–37 °C with a feedback heating pad. A laminectomy, with a 1-mm lateral expansion on the left side for applying NP, was performed at vertebrae Th13-L1, corresponding to spinal cord segments L3–S1. The vertebral column was rigidly fixed by clamps rostral and caudal to the exposed spinal cord segments. The meninges, i.e., dura mater and arachnoidea, were punctured by a cannula and carefully removed by two tweezers. A section of 8–10 mm of the sciatic nerve was freed at the mid-thigh level and isolated from the surrounding tissue by a plastic film. A caudectomy was performed on genetically identically donor rats, and NP was harvested from 3 to 8 caudal intervertebral discs.

All animal experiments were approved by the Norwegian Animal Research Authority (NARA) and were performed in conformity with the laws and regulations controlling experiments and procedures on live animals in Norway. The rats were euthanized immediately after the end of the experiments.

A bipolar silver hook electrode (1.5 mm between the hooks) was placed proximal to the main branches of the sciatic nerve for electrical stimulation. A parylene-coated tungsten microelectrode with impedance of 2–4 MΩ (Frederick Haer & Co, Bowdoinham, USA) was lowered into the left dorsal horn by an electrically controlled micromanipulator (Märzhäuser Wetzlar GmbH & Co. KG, Wetzlar, Germany), and a reference electrode was placed subcutaneously. Extracellular single cell recordings were performed at depths of 100–600 µm from the surface of the spinal cord [29]. Only one cell was studied in each animal. The recorded signals were amplified, bandpass filtered with a half amplitude cut of 500–1250 Hz (Digitimer Ltd, Hertfordshire, UK), digitalized using a CED 1401 μ interface and displayed on a computer screen by means of the CED Spike 2.2 software (Cambridge Electronic design, Cambridge, UK). The sampling frequency was 20 kHz.

The A- and C-fiber responses were separated according to latencies, where spikes 50–300 ms after stimulus were defined as C-fiber responses. Single cell recordings were ensured by the amplitude and shape of the action potentials. Every 4th min a single test stimulus (2 ms rectangular pulse, 1.5 × C-fiber threshold) was applied to the sciatic nerve through the hook electrode. A pulse buffer connected to a stimulus isolator unit (NeuroLog System, Digitimer Ltd, Hertfordshire, UK) controlled the stimulus intensity. The C-fiber threshold was defined at the beginning of each experiment as the lowest stimulus intensity necessary to evoke a C-fiber response. Six stable C-fiber responses, varying less than 20%, served as a baseline for the subsequent experiments. Cells were rejected if the C-fiber response consisted of fewer than five or more than 20 spikes at the start of the recording.

Nucleus pulposus was harvested from 3 to 8 caudal vertebrae of genetically identical donor rats and carefully applied caudally to the recording electrode, covering the incoming spinal dorsal nerve roots [30]. The C-fiber responses were followed for 180 min after application of NP transplant covering 1–2 mm of the dorsal nerve roots, or 40 µL 0.6 mg/mL miR-223-3p in Invivofectamine (Invitrogen, Carlfbad, USA). In accordance with our earlier studies [30, 31], NP was applied in absence of nerve root compression. Application of 40 µL 0,9% NaCl served as controls.

As previously described [30], total RNA (RIN values were >7) was isolated from frozen NP tissue and converted into cDNA. The qPCR was performed in two parallels on a StepOnePlus qPCR machine (Applied Biosciences, USA). IL-6 primers (FW; TGCCCTTCAGGAACA; RV; AAGGCAGTGGCTGTC) were designed using Primer Express 2.0 (Applied Biosystems, California, USA) and checked for specificity by performing a BLAST search. Effort was made to design primers without non-specific binding (the melting curves indicated no bi-products). Target genes were normalized to β-actin (FW; CTAAG GCCAA CCGTG AAAAG A, RV; ACAAC ACAGC CTGGA TGGCT A) (internal reference).

After transcardial perfusion with HBSS, NP tissue was harvested from 3 to 8 caudal vertebrae from the same donor rat and then bisected. One piece was immediately frozen after caudectomy (NP^nativ), and the other piece was applied onto the spinal dorsal nerve roots for 180 min before it was frozen (NP^exposed). For the NP graft in media, the tail of a donor rat was cut off at the base, skinned and placed in a sterile petri dish. The NP tissue was harvested from the caudal vertebrae in a LAF bench. The tissue was incubated in 0.5 mL of medium in a humidified 5% CO₂ incubator at 37 °C. Two groups of samples were provided, differing in the time spent in the incubator—5 min and 3 h. After incubation, the samples were spun first at 300g and then at 1000g in order to retain medium without any cells. miR extraction was performed on 200 µL of medium. In accordance with the manufacturer’s protocol, total RNA was isolated from frozen (−80 °C) tissue or 200 µL of medium using the miRNeasy micro kit (Qiagen) and converted into cDNA using the miScript HiSpec reverse transcription buffer (Qiagen).

The screening of miRs was performed using the Rat Pain: Neuropathic and Inflammatory MiScript miR PCR Array containing 84 miRs (cat.no. MIRN-120Z, Qiagen). cDNA from NP^nativ and NP^exposed were separately pooled before the analyses. Normalization was performed to the mean of all six snoRNA/snRNAs controls included in each array.

Based on the array analysis, miRs with a fold change >5 were selected for follow-up analysis with qPCR. In this case, the mean of the SNORD61 and SNORD68 reference genes was used for normalization of gene expression. All primers were predesigned and delivered by Qiagen. Then, 1 ng of cDNA was used in the qPCR reaction. qPCR was performed on an Applied Biosystems 7900 Real Time PCR System with the following conditions: 95 °C for 15 min, followed by 40 cycles at 95 °C for 5 s, 55 °C for 30 s, and 70 °C for 30 s.

Nucleus pulposus grafts were incubated in the serum-free culture media (Ham’s F-12 nutrient mixture) at 37 °C in a humidified 5% CO₂ incubator for 3 h. After incubation, media were centrifuged at 300g for 10 min and then at 1000g for 10 min to remove cells and other larger contaminants. Media were then centrifuged for 45-min at 10,000g. Pellets were discarded and supernatants were passed through a 0.45-µm filter. Supernatants were ultracentrifuged (Optima MAX-XP Benchtop Ultracentrifuge, Beckman Coulter, Brea, CA, USA) at 100,000g for 90 min to pellet ELVs. The pellet was washed once with PBS and centrifuged again at 100,000g for another 90 min. The ELV pellets were resuspended in PBS and stored at −21 °C for further analyses. As previously described [32], all centrifugation steps were performed at 4 °C.

The ELVs pellets resuspended in PBS were fixed in 4% formaldehyde/0.2% glutaraldehyde and placed on formvar-carbon-coated electron microscopy grids. The grids were then washed, and stained with a mixture of methylcellulose and uranylacetate. Samples were observed in a JEOL–JEM 1230 (JEOL Ltd., Tokyo, Japan) at 80 kV and pictures were acquired using a Morada camera and iTEM software (Olympus, Münster, Germany).

Nanoparticle tracking analysis was used to determine the size distribution and concentration of the ELV population. Exosome pellets were resuspended in PBS, vortexed thoroughly for 1 min and loaded into the NS300 instrument (Malvern Instruments, Worcestershire, UK) using the syringe pump. For all measurements, five videos, 60 s each, were acquired (infusion rate: 20 with camera settings: shutter, 600; gain, 350–450). Recorded data were processed with the NTA 3.0 software to obtain the size distribution and approximate concentration by tracking the centre of each particle under Brownian motion on a frame-by-frame basis.

Exosome-like vesicles were lysed in lysis buffer containing 50 mM Tris–HCl, 300 mM NaCl, 1 mM EDTA, 0.5% Triton X-100 (pH 7.4) in the presence of Halt™ protease inhibitor cocktail (ThermoFisher Scientific, Waltham, MA, USA). ELV proteins were separated by SDS-PAGE in 4–20% TGX™ precast gels (BioRad, Hercules, CA, USA) after solubilization in laemmli sample buffer. Proteins were transferred to PVDF membranes using a Tranfer-Blot Turbo Transfer Pack (BioRad). Membranes were blocked in 5% (w/v) non-fat dry milk dissolved in PBS containing 0.1% (v/v) Tween-20. Blots were incubated with the specified primary and secondary antibodies. Blots were visualized with Amersham™ ECL™ Prime Western blot detection (GE Healthcare, Little Chalfont, UK) on an Amersham AI600RGB imaging instrument (GE Healthcare).

The antibodies used for western blot analysis were as follows: mouse anti-tsg101 (Santa Cruz Biotechnology Inc., Dallas, TX, USA); rabbit anti-CD9 (Abcam, Cambridge, UK); rabbit anti-alix (Merck Millipore, Darmstadt, Germany); and HRP-conjugated secondary antibodies (Cell Signalling Technology, Danvers, MA, USA). All antibody dilutions (1:500 for the primary and 1:5000 for the secondary antibodies, respectively) were prepared in blocking solution.

Participants with lumbar radicular pain were recruited from Oslo University Hospital (Ullevål) during 2007–2009. A total of 122 patients were included in the intended follow-up assessment, 8% were lost during the follow-up, and 112 patients were ultimately assessed at the 12-month follow-up. All participants received written information and signed an informed consent form. The study was approved by the Norwegian Social Science Data Services and Norwegian Regional Committee for Medical Research Ethics (REK 214/1725).

The inclusion criteria were patients aged 18–60 years, LDH on magnetic resonance imaging with corresponding radicular pain, and positive straight leg raise test results. A positive straight leg raise test was defined by pain radiating into one or both legs when the examiner slowly raised the straightened limb until 60°. The test was performed in the supine position and supplemented with slight dorsiflexion of the foot. The exclusion criteria were lumbar spinal stenosis, previous spinal surgery for a herniated disc at the same level (to rule out radiating pain from epidural fibrosis), fusion surgery at any level in the lumbar spine, generalized musculoskeletal pain (widespread musculoskeletal pain in three out of four body quadrants), inflammatory rheumatic disease, diabetic polyneuropathy, cardiovascular disease (New York Heart Association III and IV), cancer, psychiatric disease, drug abuse and alcoholism, recent surgery (within 1 month), pregnancy, poor proficiency in the Norwegian language, and non-European-Caucasian ethnicity. Patients with cauda equina syndrome were excluded.

At the initial screening visit and 1-year follow-up, the patients underwent a standardized clinical examination, which included an assessment of sensory and motor function, and standardized registrations of pain and function. Pain intensity was recorded using the visual analogue scale (VAS) with anchor values from 0 (no pain) to 10 (worst imaginable pain) over the past week during activity.

Demographic data and the patient’s smoking status were obtained upon inclusion in the study. Treatment (surgery or conservative) was determined during the initial screening visit, where surgical treatment was given to patients with persistent radicular pain lasting for more than 8 weeks, neurological deficits (sensory changes, muscle weakness or depressed or absent deep tendon reflexes) and corresponding magnetic resonance imaging findings in the anticipated location. Non-surgical treatment was given to patients who did not clearly meet these criteria. Non-surgical treatment included a brief cognitive intervention, activity guidance during the acute phase of lumbar radicular pain, and physiotherapy for most patients.

At inclusion and 12-month follow-up, venous blood was collected and kept on ice for 45 min. After centrifugation at 2000g and 4 °C and for 10 min, the supernatant serum was collected and stored in aliquots at −80 °C until further analysis. Serum was used to analyse the miRs. Samples with visible haemolysis were excluded. Total small RNAs were extracted from 200 µL serum using the miRNeasy serum plasma isolation kit (Qiagen) according to the manufacturer’s protocol. Synthetic C-elegans (C) miR-39-3p (Qiagen) was spiked-in at a final concentration of 1.6 × 10⁸ copies/μL after the initial denaturation, prior to extraction, to correct the extraction efficiency. Total RNA was eluted in 14 µL of RNase-free water. A fixed volume of 7 µL of eluate was used as input for the cDNA synthesis. RNA was converted to cDNA using the qScript microRNA cDNA synthesis kit (Quanta). qPCR was performed with PerfeCta SYBR green supermix on an Applied Biosystems 7900 Real Time PCR System with the following conditions: 95 °C for 2 min, followed by 40 cycles at 95 °C for 5 s, 60 °C for 15 s, and 70 °C for 15 s. Data were normalized to the spiked-in C miR-39-3p.

Of the 112 patients with both baseline and 12-months data, 7 and 3 samples were excluded due to visual haemolysis at inclusion and 12 months, respectively. In addition, 1 sample was missing at inclusion, 2 samples were missing at 12 months and 2 samples could not be analysed due to undetectable miR levels at 12 months. In total, this resulted in miR data for both inclusion and 12 months from 97 patients.

As previously described [9], IL-6 concentrations in the serum of the same 97 patients were determined by using commercial ELISA kits (Human ultra-sensitive kits for IL-6, Invitrogen Corporation, CA, USA). Following the manufacturer’s instructions, 100 µL of the serum samples or standards were added to each well in microtiter plates that were pre-coated with an antibody specific to IL-6 and incubated at room temperature. Serum collected at the two time points for each patient was analysed on the same microtiter plate.

The spinal nociceptive activity, i.e., the C-fiber response, was presented as percent of baseline. Baseline recordings were converted to an average of three consecutive responses, whereas the recordings after NP application were converted to an average of nine consecutive responses, producing two pre-NP/miR-233-3p and five post-NP/miR-233-3p values. The effect of NP or miR-233-3p on spinal nociceptive response over time was compared to the control by repeated measures analysis of variance (rmANOVA). When the sphericity assumption was not met, a Greenhouse–Geisser correction was applied.

For miR arrays, data analysis was performed using the miScript miR PCR Array Web-based software delivered by Qiagen (http://​pcrdataanalysis.​sabiosciences.​com/​mirna). In brief, Δ C _tvalue for each miR profiled in the plate is calculated using the formula ΔC _t = C _t ^miRNA  – AVG C _t ^{SN1/2/3/4/5/6} , where AVG C _t ^{SN1/2/3/4/5/6} is the mean of all six snoRNA/snRNAs controls included on each array. ΔΔC _t for each miR across the two groups of samples is calculated using the formula: ΔΔC _t = ΔC _t (sample) −ΔC _t (control). The fold-change was calculated as 2^−ΔΔCt.

In the qPCR analysis, fold change values for each sample were defined by the expression of the target gene normalized to the expression of the mean of the reference genes SNORD61 and SNORD68 in the animal studies and C miR-39-3p in the patients. The abundance of each miR was calculated by the comparative Ct method and 2^−ΔCt values in animals were analysed by the two-tailed unpaired Student’s t test and linear mixed model.

Regarding the clinical arm of the study, the change in IL-6 and miRs from inclusion to 12 months was analysed by the paired Student’s t test. A two-sided Pearson correlation test was performed to examine the relationship between miR-223-3p at inclusion and IL-6 at 12 months and the delta miR values versus delta VAS values. In addition, the miR levels were analyzed by linear regression adjusted for sex, age and smoking. Baseline differences were tested by Pearson Chi square, unpaired Student’s t test and Mann–Whitney U test.