Cell Therapy with Human Dermal Fibroblasts Enhances Intervertebral Disk Repair and Decreases Inflammation in the Rabbit Model

Cell Culture

The nHDFs isolated from human foreskin were purchased from Invitrogen Life Technologies (Carlsbad, California, United States). The cells were cultured in Dulbecco's modified Eagle medium (high glucose) supplemented with 1% penicillin–streptomycin, 25 mM HEPES (all from Invitrogen Life Technologies) and 10% fetal bovine serum (Omega Scientific, Tarzana, California, United States). On the day of injection, nHDFs were labeled with infrared dye for cell tracking purposes. Monolayer cells were detached with trypsin, labeled with CellVue NIR815 Fluorescent dye (LI-COR Biosciences, Lincoln, Nebraska, United States) as recommended by the manufacturer, and suspended in serum free media at a concentration of 1.0 × 10⁷ cells/mL.

Rabbit Disk Degeneration and Cell Treatment

New Zealand white rabbits (Myrtle's Rabbitry, Thompson Station, Tennessee, United States) weighing ∼2.5 to 3 kg were used in this study with the approval of the Institutional Animal Care and Use Committee (Fig. 1). Under general anesthesia and using aseptic conditions, a left abdominal incision was made and the ventral surfaces of four consecutive lumbar IVDs (L2–L3, L3–L4, L4–L5, and L5–L6) were exposed. Using an 18-gauge needle, the annulus fibrosus (AF) was punctured to a depth of 5 mm in the ventral aspect into the nucleus pulposus (NP). The needle was rotated 360 degrees. Suction was applied by withdrawing through a 5-mL syringe for 10 seconds to denucleate the IVD. A vascular staple was placed on the psoas muscle at the L4–L5 level as a marker. The surgical wound was repaired in layers. Meloxicam (1.5 mg) was also given orally (1 day before surgery and 2 to 3 days after the operation). An analgesic (buprenorphine HCl 0.01 to 0.03 mg/kg) was given up to twice daily for 2 to 3 days, when needed, in consultation with the veterinary staff. After recovery from anesthesia, the rabbits were returned to their cages and mobilized ad lib.

OPEN IN VIEWER

Fig. 1

Study design. New Zealand white rabbits (n = 16) underwent surgery to induce disk degeneration by annular puncture on four disks (L2–L3, L3–L4, L4–L5, and L5–L6). Four weeks later, rabbits underwent a second surgery and were randomized to receive either neonatal human dermal fibroblasts (nHDFs) or saline treatment in the injured disks (L2–L3, L3–L4, L4–L5, and L5–L6). L1–L2 and L6–L7 intervertebral disks (IVDs) from each rabbit were left uninjured and untreated. Rabbits were euthanized at 2 weeks (n = 4, two per group) or 8 weeks (n = 12, six per group) post-treatment. X-ray images were obtained before the first surgery, 4 weeks after disk injury, and 8 weeks post-treatment. After euthanasia, rabbit spines underwent magnetic resonance imaging (MRI) and infrared imaging. Individual IVDs were further isolated to study changes in gene expression and collagen staining.

Four weeks postinjury, an X-ray was taken to confirm degeneration of the disks. Then, a right abdominal incision was made. The ventral surfaces of the L2–L3, L3–L4, L4–L5, and L5–L6 IVDs were exposed. Treatments with nHDFs or saline were randomized between rabbits. A volume of 8 µL (the maximum volume that we could easily inject into the rabbit lumbar disks ex vivo) of either nHDF suspension or saline was injected into each degenerated rabbit IVD with an Exmire syringe (Ito Corporation, Fuji, Japan) and 27-gauge needle. The postoperative procedures are described above.

Radiographic Analysis

X-ray images of sedated rabbits were obtained before the surgery, 4 weeks after disk injury, and 8 weeks post-treatment. Disk height indexes (DHIs; n = 12) were calculated by three orthopedic researchers who were blinded to the treatment groups using the method of Lü et al with a slight modification (DHI = IVD height/adjacent IVD body height). ¹⁶ The average IVD height was calculated by averaging the measurements obtained from the anterior, middle, and posterior portions of the IVD and dividing that by the average of adjacent vertebral body heights. Changes in the DHI of injected disks were expressed as %DHI and normalized to the measured preoperative DHI (%DHI = postoperative DHI/preoperative DHI × 100) as previously described. ¹⁷

Magnetic Resonance Imaging Analysis

Magnetic resonance imaging (MRI) was performed on the rabbit spines removed en bloc using a 1.5-T MRI machine (Siemens, Malvern, Pennsylvania, United States) with a hand coil. Transverse relaxation time (T2)-weighted sections in the sagittal plane were obtained in the following settings: fast spin echo sequence with time to repetition, 4,000 milliseconds; time to echo, 97 milliseconds; slice thickness, 3.5 mm; field of view, 250, and matrix, 256 × 320. MRIs (n = 12) were graded by five blinded independent readers according to the modified Thompson scale as described previously. ¹⁸

MRI indexes (n = 12) were calculated as described by Sobajima et al. ¹⁹ , ²⁰ MRIs were transformed into picture files using Image J software (National Institutes of Health, Bethesda, Maryland, United States). NPs were outlined to define the regions of interest. High signal intensity area, highest signal intensity, and average signal intensity of this region of interest were calculated. The MRI index was defined as the product of the high signal intensity area and average signal intensity. MRI indexes were normalized to the uninjured intact disks in the same animal.

Cell Tracking

The spine segments and individual disks were scanned with the Odyssey infrared imaging system (LI-COR Biosciences, Lincoln, Nebraska, United States) to detect signals at the 700- and 800-nm wavelengths. Infrared fluorescence intensity counts per cubic millimeter of the individual disks were determined using the Odyssey imaging software (LI-COR Biosciences) and exported to Microsoft Excel (Redmond, Washington, United States). The infrared fluorescence intensities between the 2- and 8-week points were compared using a nonparametric t test.

Gene Expression Assays

Total RNA was processed for each IVD. Real-time PCR analysis was performed as described in previous studies. ⁸ Taqman gene expression assays were used for analysis or normalization for the following genes: rabbit COL1A2, COL2A1, IL-8, HPRT1 and 18 S rRNA (Oc03396113_m1, Oc03396134_m1, Oc03397860_m1, Oc03399461_m1, Hs99999901_s1; Applied Biosystems, Foster City, California, United States). A PrimeTime Mini qPCR Assays were designed for rabbit chemokine (C-C motif) ligand 2 (CCL2) and rabbit COL3A1 (CCL2 primers: 5′-TCT TGT CCA GGTTGG CAATG-3′ and 5′-CCC AAA GAA GCT GTG ATC TTC A-3′; CCL2 probe: 5′-/56-FAM/CCA AGC AGA /ZEN/AGT GGGTCC AGG ATG /3IABkFQ/-3′; COL3A1 primers: 5′-ACA GCCTTG CGT GTT CTATAT T-3′ and 5′-GGC AGA AGG AAA CAG CAA ATT C-3′; COL3A1 probe: 5′-/56-FAM/TAC ACA GTT /ZEN/CTG GAG GAT GGCTGC /3IABkFQ/-3′ [Integrated DNA Technologies, Coralville, Iowa, United States]). DataAssist Software (Applied Biosystems) was used to calculate the relative gene expression using the comparative CT (ΔΔCT) method. Intact controls were used as reference samples.

Immunostaining for Collagen Types I and II

Intervertebral disk segments (n = 12) were fixed with 10% formaldehyde, embedded in paraffin, and sectioned (5 μm). Sections were deparaffinized, treated with Proteinase K, incubated with 0.5% hydrogen peroxide, and blocked with 5% horse serum before incubating overnight with mouse anti-collagen type I antibody (1:1,000; C2456, Sigma, St. Louis, Missouri, United States) or mouse anti-collagen type II antibody biotinylated (1:2,000; no. 7006, Chondrox, Redmond, Washington, United States). Because the anti-collagen type II antibody was already biotinylated, a secondary antibody incubation was not necessary for these sections. For collagen type I staining, the sections underwent secondary antibody incubation with biotinylated horse anti-mouse immunoglobulin G (1:400). All the antibodies were diluted in PBS with 3% horse serum. Vectastain ABC kit (Vector Labs, Burlingame, California, United States) and 3,3′-diaminobenzidine (DAB) were used to develop the immunostaining. The sections were counterstained with hematoxylin solution (VWR, Radnor, Pennsylvania, United States). Negative controls were performed by omitting the primary antibodies.

The degree of collagen types I and II immunopositive staining was assessed by scoring the staining intensity and the number of stained areas. Representative images were taken at 200× magnification on a Nikon Eclipse 50i microscope using NIS-Elements Imaging Software (Nikon, Tokyo, Japan).

Statistical Methods

The Kruskal-Wallis test and Mann-Whitney test were used to analyze nonparametric data (MRI grading) for the effect of treatment. The changes in disk height indexes were assessed by the Spearman rank correlation test. The MRI indexes were compared using the Student t test. The infrared fluorescence intensities were compared using a nonparametric t test. The differences were considered significant when the p value was equal to or below 0.05.

Tracking of nHDFs Injected into Rabbit Degenerating IVDs In Vivo

To determine if the cells transplanted into the degenerating IVD remained in the IVD, the nHDFs were labeled with infrared dye and tracked at the end of treatment. The infrared dye-labeled nHDFs were transplanted into the degenerating rabbit IVDs at a concentration of 10⁷ cells/mL. After euthanasia, isolated spines and individual IVDs were scanned with an infrared scanner 2 and 8 weeks after transplantation. As seen in Fig. 2, the rabbit spine and disk contours were detected in the 700-nm wavelength channel (represented in red). Injected cells were detected in the 800-nm wavelength channel and represented in green or yellow (yellow represents overlapping signals of the 800- and 700-nm wavelengths). At both 2 weeks (Fig. 2, A’ and A”) and 8 weeks (Fig. 2, B’ and B”) after transplantation, infrared dye-labeled cells were detected in the spines and individual IVDs. The average intensity of the IVDs was 226,555 counts at 2 weeks post-treatment (n = 2 rabbits) and 75,239 counts at 8 weeks post-treatment (n = 3 rabbits). Although there was a threefold decrease in signal intensity from 2 to 8 weeks, this data suggests that some of the cells injected into the IVD remained in the IVD for up to 8 weeks.

OPEN IN VIEWER

Fig. 2

Neonatal human dermal fibroblasts were tracked after injection into the degenerating rabbit intervertebral disk in vivo. Infrared dye-labeled neonatal human dermal fibroblasts were injected into the degenerating rabbit disks and imaged at 2 and 8 weeks post-treatment. The rabbit spine and disk contours were detected in the 700-nm wavelength channel represented in red. The injected cells were detected in the 800-nm wavelength channel and represented in green or yellow (yellow represents overlapping signals of the 800- and 700-nm wavelengths). Coronal images of the spine and transverse images of the individual disks were taken at (A’ and A”) 2 weeks post-treatment and at (B’ and B”) 8 weeks post-treatment. (C) Average integrated signal intensity of the infrared fluorescence in the 800-nm wavelength of the intervertebral disks. Error bars represent the standard error of the mean.

Radiographic Assessment 8 Weeks Post-Treatment

DHIs were calculated by three orthopedic researchers who were blinded to the treatment groups. Four weeks after disk injury, a reduction of disk height occurred due to injury-induced degeneration. Representative lateral radiographs are shown in Fig. 3 (left panel). DHI at 4 weeks after injury had decreased by ∼30% compared with the initial DHI before surgery. Eight weeks after saline treatment, there was a small increase in the DHI but this increase was not significant. In the disks that were treated with nHDFs, there was a 10% restoration in the DHI. This difference was significant in comparison with the DHI before treatment (p ≤ .05; Fig. 3, right panel).

OPEN IN VIEWER

Fig. 3

Lateral radiographs and changes in disk height index. (Left) Representative serial lateral radiographs of a lumbar spine of a rabbit obtained initially presurgery, 4 weeks after injury, and 8 weeks after treatment. (Right) Average disk height indexes were calculated presurgery, 4 weeks after injury, and 8 weeks after treatment in disks that were intact, injured, and treated with neonatal human dermal fibroblasts (nHDFs) and injured and treated with saline. The average disk height index was normalized to the presurgical disk height index. *Significant difference between the two time points: 4 weeks after injury and 8 weeks after nHDF treatment (p < 0.05). Error bars represent the standard error of the mean.

Magnetic Resonance Imaging Assessment 8 Weeks Post-Treatment

At 8 weeks post-treatment, rabbit spines underwent MRI. Representative MRIs are shown in Fig. 4A. Four levels (L2–L3, L3–L4, L4–L5, and L5–L6) were injured and treated and the other levels (L1–L2 and L6–L7) were left uninjured and intact. MRI grading was calculated using the Modified Thompson Grading Scale by five independent orthopedic researchers who were blinded to the treatment groups. The NPs in the intact uninjured disks showed lower average MRI grades and higher MRI signal intensities than those in the injured nHDF-treated or saline-treated disks (Fig. 4B). The average MRI grade for the nHDF-treated disks was 3.5 and the average for the saline-treated disks was 2.8. The nHDF-treated disks had higher MRI grades and lower signal intensities than the saline-treated IVDs.

OPEN IN VIEWER

Fig. 4

Magnetic resonance imaging (MRI) analysis 8 weeks after neonatal human dermal fibroblast (nHDF) or saline treatment. (A) Representative MRI scans of the lumbar spine of rabbits at 8 weeks post-treatment. (B) Average MRI grading using the Modified Thompson Grading Scale. (C to F) Highest signal intensity, average signal intensity, and high signal intensity areas were calculated to determine the MRI index (nucleus pulposus [NP] average signal intensity × NP area). Error bars represent the standard error of the mean.

Due to the subjectivity of MRI grading, MRI indexes were also included in our analysis. The highest signal intensity, average signal intensity, and high signal intensity area were calculated to determine the MRI index (Fig. 4C to 4F). The MRI index (defined as the NP area × NP average signal intensity) of each treated disk was then normalized to the intact disk in the same animal. On average, the MRI indexes showed that the saline-treated IVDs had higher signal intensities than the nHDF-treated IVDs, although this difference was not significant.

Collagen Gene Expression Analyses

Collagen types I and II are markers that are usually found in the NP and AF tissues, and collagen type III is a marker for fibrotic tissue. At 2 and 8 weeks after treatment, expression of COL1A2, COL2A1, and COL3A1 genes from rabbit IVDs treated with nHDFs and saline increased in comparison to uninjured controls (Fig. 5A to 5C). The expression of COL3A1 was highly upregulated in the nHDF-treated disks at 2 weeks, which later decreased to the same level as those treated with saline at 8 weeks. Cartilage and NP tissues usually contain a higher ratio of collagen type II over type I. The ratios of COL2A1 to COL1A2 gene expression were calculated and normalized to that of the saline-treated samples. At 2 weeks post-treatment, the ratio of COL2A1 to COL1A2 gene expression was similar between the nHDF (0.84) and the saline (1.00) treatment groups (Fig. 5C). At 8 weeks post-treatment, the ratio was higher in the IVDs treated with nHDFs (2.71) than in those treated with saline (1.00).

OPEN IN VIEWER

Fig. 5

Expression analysis of collagen genes using real-time polymerase chain reaction (PCR). RNA from uninjured untreated control disks and injured disks treated with neonatal human dermal fibroblasts (nHDFs) or saline was isolated for real-time PCR analysis. (A–C) At 2 and 8 weeks after treatment, expression levels of collagen types II, I, and III genes were analyzed. (D) The ratio of collagen II (Col II) to collagen I (Col I) gene expression was calculated and normalized to that of the saline treated disks. Error bars represent the standard error of the mean.

Collagen Types I and II Immunostaining in the IVD

Histologic analysis showed evidence of fibrocartilage formation in the injured and treated disks (data not shown). To determine if treatment led to a difference in fibrocartilage formation, sections were immunostained for both collagen types I and II and analyzed. Collagen type II was detected in the NP areas and collagen type I staining was found in the AF areas. Areas of new fibrocartilage formation had staining for both types of collagens. The disks treated with nHDF had a higher staining intensity and a larger number of areas positive for both collagen types I and II than those treated with saline. Positive immunostaining of the collagens was not distinct at low magnification (10×), therefore representative images in Fig. 6 are of high magnification (200×).

OPEN IN VIEWER

Fig. 6

Representative immunohistochemical staining of collagen types I and II. Collagen type I staining of the annulus fibrosus regions in disks treated with (B) neonatal human dermal fibroblasts (nHDFs) or (D) saline and (A, C) their respective negative controls. Collagen type II staining of nucleus pulposus regions of disks treated with (F) nHDFs or (H) saline and (E, G) their respective negative controls. Negative controls had primary antibodies omitted. Note: Staining of both collagen types I and II is more intense in nHDF-treated disks compared with those treated with saline. Scale bar = 50 μm.

Cytokine and Chemokine Gene Expression Analyses

For both nHDF- and saline-treated disks, the expression levels of inflammatory cytokine IL-8 and chemokine CCL2 genes were elevated compared with uninjured intact controls at 2 weeks (Fig. 7). Although the saline-treated disks showed a 4.5-fold increase in IL-8 and 15.4-fold increase in CCL2 gene expression, the nHDF-treated disks had only a 1.5-fold increase in IL-8 and 5-fold increase in CCL2 gene expression. Treatment with nHDFs reduced the expression of IL-8 and CCL2 genes by two-thirds.

OPEN IN VIEWER

Fig. 7

Expression analysis of inflammatory cytokine and chemokine genes using real-time polymerase chain reaction (PCR). RNA from uninjured untreated control disks and injured disks treated with neonatal human dermal fibroblasts (nHDFs) or saline was isolated for real-time PCR analysis. At 2 weeks after treatment, expression levels of interleukin-8 (IL8) and chemokine (C-C motif) ligand 2 (CCL2) genes were analyzed. Error bars represent the standard error of the mean.