Early Genetic Restoration of Lubricin Expression in Trangenic Mice Mitigates Chondrocyte Peroxynitrite Release and Caspase-3 Activation

Abstract

Objective

This study investigated the ability of endogenous lubricin secretion to restore joint health following a brief < 21 day, postnatal lubricin-null state, in a C57BL/6J Prg4 gene trap mouse under the control of cre-recombinase. Previously we showed that re-expression of lubricin at 21 days was partly restorative of joint lubrication.

Design

The tibio-femoral joints of adult C57BL/6J mice containing lubricin, lacking lubricin, and postnatally lacking lubricin until restoration of lubricin expression at 7 days or 14 days of age were evaluated ex vivo. At 8-weeks of age, whole joint coefficient of friction (COF), and caspase-3 activation were measured and the tibial-femoral joints histologically analyzed for degenerative changes, following progressive cyclic loading. The peroxynitrite content of femoral head cartilage from these mice prior to cyclic loading was measured.

Results

Mice that underwent gene recombination at 7 and 14 days of age did not reestablish low COF as joint cycling time increased and were histopathologically indistinguishable from the joints of lubricin-null littermates. However, cartilage from tibio-femoral joints that underwent recombination at 7 and 14 days of age had significantly fewer caspase-3 positive cells and significantly reduced peroxynitrite content compared to lubricin-null littermates.

Conclusions

The biological effects of lubricin, which include limiting inflammation via peroxynitrite production and caspase-3 activation, may be achieved without completely restituting low COF. However, fully recapitulating low COF may require undamaged cartilage surfaces or absence of biofouling, which may interfere with the activity of lubricin.

Introduction

Arthrosis in lubricin deficient mouse models can be characterized as having early cellular loss[1] that is directly attributable to the disturbance of the tribological systems that protect articular cartilage from locomotion induced friction.[2–4] Moreover, superficial zone chondrocytes located just beneath the cartilage surface possess limited replicative ability[5, 6] and likely sense enhanced mechanical strain from elevated friction,[7] in the absence of lubricin or its insufficiency. Friction incites the activation of caspase-3 and thus initiates arthrosis. By contrast, mouse models that spontaneously kill resident chondrocytes via endogenous diphtheria secretion have no demonstrable arthropathy at 9 months, unless a traumatic injury such as destabilization of the medial meniscus occurs.[8] Together these observations indicate that lubricin plays a fundamental role in protecting the integrity of the cartilage surface and the chondrocytes that are embedded in its matrix.

Lubricin-null mice clinically recapitulate camptodactyly-arthopathy-coxa varapericarditis (CACP) syndrome. CACP syndrome occurs in patients who lack functional expression of lubricin, a mucinous glycoprotein encoded by the gene PRG4. Patients with CACP syndrome develop premature joint failure[9] and synovial fluids from patients with CACP fail to reduce friction in synthetic[10] and cartilage bearings.[11] CACP patients have used non-steroidal anti-inflammatory medications for symptom relief, suggesting inflammation may play a role in the lubricin-null joint pathology, which hitherto was thought to be a non-inflammatory arthropathy.

Traumatized joints in humans and rodent models, though not lubricin-null, show reduced lubricin in synovial fluid and the superficial zone of cartilage.[12–15] Epidemiologically, traumatized joints are well known to enhance the risk for post-traumatic osteoarthritis.[16] Previously, transient deficiency of lubricin, when corrected by intra-articular delivery of exogenous lubricin, has been shown to be chondroprotective and to limit the extent of disease progression.[17–20] Recently, it was shown in 8-week-old, lubricin-null gene trap mice that recombination at 21 days produced less activation of caspase-3 in chondrocytes and displayed improved, but not normalized, histological features and whole joint coefficient of friction (COF) compared to their lubricin-null, non-recombined counterparts.[21] The aforementioned study also showed that restoring lubricin gene function prior to conception prevented arthrosis and joint disease.[21]

Presently, we hypothesized that the restoration of lubricin production in the same Prg4 gene trap mice (Prg4^GT/GT) at earlier time points, at 7 and 14 days, would prevent subsequent mechanical and biological joint degeneration, resulting in lower whole joint COF, decreased joint damage and a reduced percentage of chondrocytes displaying caspase-3 activation. This study also posited that earlier expression of lubricin would reverse the cartilage biofouling that is evident in lubricin-null mice and Prg4^GT/GT mice recombined at 21 days. Additionally, re-establishing low friction was theorized to reduce mechanical strain causing chondrocyte mitochondrial dysfunction as indicated by peroxynitrite production.

Materials and Methods

Lubricin gene trap and lubricin mutant mice

The gene trap (GT) allele produced β-galactosidase (lacZ) in place of Prg4, in cells that would normally express Prg4. Details describing this mouse been reported previously.[21] To excise the gene trap, the cre allele, ROSA26^CreERT2 (Jax strain 008423), was used which allows widespread excision of the gene trap following intraperitoneal injections of tamoxifen. Colonies of Prg4^GT/GT;ROSA26^CreERt2/+ (Prg4^GT/GT) mouse genotype (Jax strain 025740), lubricin knockout mice Prg4^−/− (Jax strain 025737), and lubricin wild type mice Prg4^+/+ (Jax strain 000664) were used for this study. Prg4^−/− and Prg4^+/+ mice were bred to produce Prg4^+/− mice. Mice of both sexes were used equally for experimentation. A single animal was considered an experimental unit, N=1. Three separate groups of mice from various genotypes were assessed for different experiments. Power analysis for the first group of mice was supported by the effect size in Hill et al.[21] The first group of mice (N=39 total mice) was used for biomechanical analysis and assessment of joint damage, caspase-3, and hyaluronic acid content. Scientists were blinded for all studies completed with the first group of mice. A second group of mice (N=25 total mice) was used to measure peroxynitrite. All mice from the first and second groups were euthanized at 8-weeks of age (weight ~20g). A third group of mice (N=4 total mice) was used for electron microscopy (EM). Mice used for EM were euthanized at 14 days of age (weight ~10g). All mice were housed 5 per Thoren cage with wood shavings for bedding. All mice had access to sterile water and pellet food at all times. Animals were randomly assigned to experimental groups. Mice from the second and third experimental groups were euthanized and processed for experimentation on the same day. Mice from the first experimental group were assessed in 9 batches of 4 mice and one batch of 3 mice. Three genotype control groups and two experimental groups were assessed for the first and second mouse groups. Two genotype control groups and two experimental groups were assessed for the third mouse group. All animal experiments were approved by the Institutional Animal Care and Use Committees at Rhode Island Hospital and were housed in an AAALAC accredited facility.

Induction of cre-mediated recombination using tamoxifen

Mice with the genotype Prg4^GT/GT were given daily, morning intraperitoneal (IP) injections of tamoxifen (0.1 mg/g) in corn oil (Sigma-Aldrich) for 10 consecutive days beginning at 7 or 14 days of age, to produce genotypes Prg4^GTR7D/GTR7D and Prg4^{GTR14D/GTR14D}, respectively, then allowed to age normally until euthanasia at 8-weeks of age. Injections of tamoxifen were given without anesthesia and within a biosafety cabinet. A separate group of control Prg4^GT/GT mice received injections of corn oil alone, under the same conditions as the tamoxifen injections, and were allowed to age normally for 8 weeks. In addition, another group of Prg4^GT/GT mice received tamoxifen earlier at post-natal day 2 for histological analysis via transmission electron microscopy (EM). These mice were then euthanized at day 14 and compared to Prg4^+/+, Prg4^GT/GT, and Prg4^GTR7D/GTR7D mice also euthanized at day 14. Mice receiving tamoxifen injections were monitored for any changes in behavior following IP injections.

Biomechanical analysis

Following euthanasia at 8-weeks of age for Prg4^GTR7D/GTR7D (N=11 mice and 22 legs), Prg4^{GTR14D/GTR14D} (N=11 mice and 22 legs), Prg4^GT/GT (N= 6 mice and 12 legs), Prg4^+/− (N=5 mice and 10 legs) and Prg4^+/+(N=6 mice and 12 legs), both tibio-femoral joints of each mouse were harvested and soft tissue was removed leaving the joint capsule intact. Joints were submerged in PBS, to prevent desiccation, and statically loaded with the weight of the pendulum (50g), which approximates 2 times body weight, for 8 minutes prior to each set of measurements. Whole joint coefficient (COF) was measured using a modified Stanton pendulum system, as previously described.[11, 22, 23]

Cyclic loading

Biomechanically tested tibio-femoral joints were subjected to 60-minute long ex vivo cyclic loading regimens using an active pendulum, which was continuously oscillated ±15° at a rate of 1.5 Hz[22] in 15-minute increments. During cyclic loading, the knee was entirely submerged in Dulbecco's Modified Eagle Medium (DMEM, Life Technologies) to prevent desiccation. COF was measured prior to cyclic loading and following each increment. All experiments were carried out at room temperature.

Immunohistochemistry

Mouse knees from each biomechanically tested specimen were processed for histology and active caspase-3 staining immediately following friction testing and cyclic loading, as previously described.[11] Samples were fixed in 10% formalin and decalcified using a solution of 0.48M EDTA adjusted pH to 7.1 with ammonium hydroxide at 4°C for 48 hours. Thin paraffin-embedded coronal sections (5 µm) were taken for histological analysis of caspase-3 activation and hyaluronic acid (HA) accumulation in the tibial-femoral joints of the harvested mouse limbs. Sections were heated to 60°C for 30 minutes, deparaffinized and hydrated in 3 changes of xylene and serial alcohol. Antigen retrieval was performed using a pepsin solution (ThermoScientific, Waltham, MA). Either a rabbit polyclonal antibody against active caspase-3 (ab13847, Abcam, Cambridge, MA) or a biotinylated recombinant antibody against HA binding protein (cat#AMS.HKD-BC41. AMSBIO, Cambridge, MA) at 1:100 dilution was added to the sections and incubated at 4°C overnight. After 3 washes with PBS, the sections were incubated with either a Cy3 goat anti-rabbit IgG (Life Technologies, Molecular Probes, Waltham, MA) at 1:50 dilution or a streptavidin-fluorescein conjugate antibody at a 1:100 dilution (Life Technologies, Molecular Probes, Waltham, MA), for 1 hr at room temperature, protected from light. The sections were washed 5 times using PBS, counterstained using Vectashield mounting media with DAPI (1.5µg/ml, Vector Laboratories Inc, Burlingame, CA). Images were captured with a 10× and 20× objective, using an Olympus BX51 microscope (Olympus America Inc., Center Valley, PA) using Image Pro Plus 7.0 software (Media Cyberkinetics, Bethesda, MD). Sections were also stained with hematoxylin and eosin, and were imaged at 10× and 20× using the same Olympus BX51 described above. Caspase-3 quantification was performed as previously described.[21]

Quantitative histologic assessment of joint pathology

Photomicrographs of hematoxylin and eosin stained coronal knee sections were scored using a previously described method used in the Prg4^(GT/GT) mouse.[21] Briefly, scoring assessed surface morphology, synovial hyperplasia, chondrocyte proliferation, and meniscal architecture. Two scientists (KL and GJ), who had been trained on the scoring system, were sent 10 photomicrographs per tibio-femoral specimen. The scientists were blinded with respect to the animals’ genotypes and treatment groups.

Peroxynitrite assay

Cartilage from femoral heads were obtained from 8-week old Prg4^+/+, Prg4^+/−, Prg4^GT/GT, Prg4^GTR7D/GTR7D, and Prg4^{GTR14D/GTR14D} mice, prior to friction testing and cyclic loading, and weighed (N=5 mice and 5 right legs for each genotype). Peroxynitrite was quantified using 2-[6-(4’-hydroxy) phenoxy-3H-xanthen-3-on-9-yl]-benzoic acid (HPF, Daiichi Pure Chemicals, Tokyo, Japan). Tissues were cultured in 96-well cell-culture plates with Dulbecco’s Modified Eagles Media (DMEM) as a control. The tissues were incubated at 37°C for 6 hours with HPF (10 µmol/L), protected from light. After incubation, the fluorescent HPF, activated by reactive oxygen species (ROS), was quantified in a fluorometer Spectra Max M2 fluorescence reader (Molecular Devices, Sunnyvale, CA) with an excitation wavelength of 490 nm and an emission wavelength of 515 nm. Peroxynitrite data were normalized to baseline DMEM fluorescence levels. The final results were normalized to femoral head weight.

Hyaluronic acid quantification

After detection with the HA binding protein antibody detailed above, 10× images for each biomechanically tested specimen were acquired using a fluorescence microscope Olympus BX51 microscope (Olympus America Inc., Center Valley, PA). These images were imported to ImageJ[24] and the mean intensity and % area occupied were calculated. Mean intensity values were corrected for background and normalized[25] using a 10% solution of standard fluorescein dye (Catalog#F1300, Thermo Fisher Scientific, Waltham, MA). Percent integrated density (optical density), defined as the corrected mean intensity times the % area occupied by HA in the superficial zone (0–10 µm), was reported.

Transmission electron microscopy

Prg4^+/+, Prg4^GT/GT, Prg4^GTR7D/GTR7D, and Prg4^GTR2D/GTR2D mice (N=1 mouse and N=2 legs for each genotype) were euthanized at day 14. These mice were processed for electron microscopy using methods previously published.[10]

Statistical analysis

Effects of genotype and cyclic loading duration on COF values were evaluated using a two-way mixed model repeated measures analysis of variance. Linear contrasts based on orthogonal polynomials were used to determine the significance of the linear effect of cycling duration on mean COF within each genotype with corresponding slope estimates derived based on mixed model regression. Simple effects were examined based on partial F-tests and pairwise comparisons among genotypes within cycling time were based on Fisher’s protected LSD. Comparisons for OA damage scores, caspase-3 positivity, HA binding protein quantity, and peroxynitrite content between genotypes were analyzed using a one-way ANOVA with pairwise comparison among genotypes based on Fisher’s protected LSD. All analyses were performed using SAS statistical software Version 9.4 (SAS Institute, Cary, NC) with statistical significance based on α=.05.

Results

All mice showed no adverse effects or changes in weight or microbiological status prior to sacrifice for use in experimentation.

Whole Joint Pendulum Effects of Early Gene Trap Recombination

There was evidence that changes in mean COF over cycling duration were genotype dependent (genotype × cycling duration interaction, p <.001). Significant increases in COF over cycling duration were observed for the Prg4^GTR7D/GTR7D (N=11) and Prg4^{GTR14D/GTR14D} (N=11) groups, (p <.001, Figure 1). In contrast, there were no significant changes in COF over cycle duration for the Prg4^+/+ (N=6), Prg4^+/−(N=5), and Prg4^GT/GT (N=6) mice (p=.12, p=.15 and p=.19, respectively). Thus, as cycling duration increased, the Prg4^{GTR7D/ GTR7D} and Prg4^{GTR14D/ GTR14D} mice shifted from displaying frictional behavior similar to lubricin wild type mice to displaying frictional behavior similar to lubricin-null mice.

Figure 1.

Coefficient of friction (COF) values at 0, 15, 30, 45, and 60 minutes cycling time for mouse knees evaluated with a whole joint pendulum system. Regression lines and Mean ±SE are plotted for each experimental group: GT/GT recombined at 7 days (white triangle), GT/GT recombined at 14 days of age (green diamond), GT/GT not recombined (red triangle), lubricin wild type (purple circle), and lubricin heterozygous mice (blue square). Significant increases in COF were observed for both GT/GT recombined at day 7 (p <.001) and day 14 (p <.001).

Prg4^GT/GT mice, while not displaying significantly higher friction values at time 0, did display significantly higher COF values for the remainder of the cyclic loading time periods (15, 30, 45, and 60 minutes) compared to the Prg4^+/+ and Prg4^+/− mice (Table 1). Recombined Prg4^GTR7D/GTR7D and Prg4^{GTR14D/GTR14D} mice exhibited different frictional behaviors over cycling duration. After the first 15 minutes of cyclic loading, Prg4^GTR7D/GTR7D mice exhibited significantly higher COF compared to Prg4^+/+ and Prg4^+/− mice (Table 1). In contrast, Prg4^{GTR14D/GTR14D} mice had significantly different friction compared to both Prg4^GT/GT and Prg4^GTR7D/GTR7D mice at 30 minutes of cyclic loading and displayed no significant difference from Prg4^+/+ mice until 45 minutes of cyclic loading, the point at which maximal differences between Prg4^+/+ and Prg4^+/− mice and the Prg4^GT/GT, Prg4^GTR7D/GTR7D, and Prg4^{GTR14D/GTR14D} mice were reached (Table 1). Thus, earlier recombination at day 7 after birth may have less effect at reducing friction than recombination at day 14 after birth.

Table 1.

Results of multiple comparisons of mouse knee COF data by cycle time. Within each cycle time, group means not sharing a common letter are significantly different (Fisher’s LSD, p<0.05). Note, there were no significant group differences in mean COF at cycle time 0. However, as cycling time progresses to 15 minutes, the testing groups, GT recombined at 7 days (GT^R7D/ GT^R7D) and GT recombined at 14 days (GT^R14D/ GT^R14D) of age begin to have mean COF values that significantly deviate from the normal lubricin wild type (+/+) levels and mimic the lubricin-null GT not recombined (GT/GT) COF values. At time points 45 and 60 minutes, the GT^R7D/ GT^R7D and GT^R14D/ GT^R14D testing groups have synched with the COF values of the GT mice such that they are significantly different from both the lubricin wild type (+/+) and heterozygous (+/−) groups (p<.001).

Comparisons of Prg4 and Prg4 Gene Trap Mouse Knee COF by Cycle Time
	+/+ (n= 6)		+/− (n= 5)		GT/GT (n= 6)		GTR7D/ GTR7D (n= 13)		GTR14D/ GTR14D (n= 12)
Cycle time (min)	Mean (SD)		Mean (SD)		Mean (SD)		Mean (SD)		Mean (SD)		p-value
0	0.038 (0.007)	a	0.037 (0.014)	a	0.051 (0.010)	a	0.039 (0.006)	a	0.034 (0.007)	a	0.11
15	0.033 (0.006)	a	0.028 (0.007)	a	0.046 (0.006)	b	0.048 (0.010)	b	0.039 (0.007)	a,b	0.008
30	0.031 (0.005)	a,b	0.027 (0.005)	a	0.060 (0.035)	c	0.052 (0.011)	c	0.042 (0.010)	b	<.001
45	0.030 (0.007)	a	0.025 (0.003)	a	0.052 (0.005)	b	0.055 (0.019)	b	0.050 (0.013)	b	<.001
60	0.029 (0.005)	a	0.028 (0.010)	a	0.056 (0.007)	b	0.064 (0.016)	b	0.056 (0.017)	b	<.001

Histological Effects of Early Gene Trap Recombination

At 8-weeks of age and after 60 minutes of ex vivo loaded joint cycling, OA damage scores for mouse tibio-femoral joints were significantly higher for the Prg4^GT/GT (N=6), Prg4^GTR7D/GTR7D (N=11), and Prg4^{GTR14D/GTR14D} (N=11) groups compared to the Prg4^+/+ (N=6) and Prg4^+/− (N=5) mice (p<.001 for all comparisons, Figure 2). The Prg4^GTR7D/GTR7D and Prg4^{GTR14D/GTR14D} experimental groups displayed joint histopathology similar to the non-recombined Prg4^GT/GT littermates.

Figure 2.

Joint histopathology scoring was used to evaluate OA damage for mouse knees, following whole joint pendulum friction testing. GT/GT mice recombined at 7 days of age (GT^R7D/GT^R7D), GT/GT mice recombined at 14 days (GT^R14D/GT^R14D) of age, and GT/GT mice not recombined (GT/GT) display significantly higher OA damage scores compared to lubricin wild type (+/+) and lubricin heterozygous (+/−) mice (***p<.001 for all comparisons). Thus, GT^R7D/GT^R7D and GT^R14D/GT^R14D test groups display joint histopathology similar to the lubricin-null phenotype, GT/GT mice.

Transmission electron micrographs of the medial femoral condyles from 14-day old Prg4^+/+, Prg4^GT/GT, Prg4^GTR2D/GTR2D, and Prg4^GTR7D/GTR7D mice (N=1 for all groups) showed dark superficial zone, parallel collagen fiber alignment in the cartilage of Prg4^+/+ mice and highly disrupted, lighter collagen fiber alignment in Prg4^GT/GT mice (Figure 3). The Prg4^GTR2D/GTR2D and Prg4^GTR7D/GTR7D groups displayed much lighter collagen distribution compared to the Prg4^+/+ and Prg4^GT/GT mice. However, their collagen alignment was qualitatively similar to the Prg4^+/+ and less disrupted than the Prg4^GT/GT mice (Figure 3). Additionally, the Prg4^GTR2D/GTR2D and Prg4^GTR7D/GTR7D mice displayed less cartilage superficial zone thickness compared to the Prg4^GT/GT mice. Approximate values for the thickness of the parallel, superficial collagen layer in the articular cartilage, were 0.3 microns for Prg4^+/+, 2 microns for Prg4^GTR7D/GTR7D, 2 microns for Prg4^{GTR14D/GTR14D}, and 4 microns for Prg4^GT/GT mice. Overall, these data suggest that friction-induced structural changes in the cartilage superstructure were prevented by early re-expression of lubricin.

Figure 3.

Electron micrographs were used to qualitatively assess the superficial zone of the medial femoral condyle of Prg4 expressing and Prg4 gene trap mouse knees at 14 days after birth. Collagen fiber alignment was stained with tannic acid in black. The lubricin-null GT not recombined (GT/GT) has disrupted collagen fiber alignment compared to the normal lubricin wild type (+/+) control. GT/GT mice recombined at 2 days (GT^R2D/GT^R2D) and 7 days after birth (GT^R7D/GT^R7D) display more parallel collagen fiber alignment similar to the +/+ mice but less collagen density (as evident through the gray vs. black coloring of the collagen at the articular surface). GT/GT cartilage displayed a thicker collagen structure along with collagen disruption compared to the +/+ cartilage. This thickening is seen to some extent in GT^R2D/GT^R2Dand GT^R7D/GT^R7D but not to the same extent as the GT/GT cartilage that had not been recombined. An amorphous lamina splendin surface layer, demarcated by black triangles, sits atop the collagen type II fibers. Images were taken at X13500 and scale bar is 0.2 microns.

Biological Effects of Early Gene Trap Recombination

A significant reduction in caspase-3 activation in recombined mice, Prg4^GTR7D/GTR7D (N=11) and Prg4^{GTR14D/GTR14D} (N=11) compared to the non-recombined, Prg4^GT/GT(N=6), littermates was observed following 60 minutes of cycling (p=.003, p<.001). The Prg4^GT/GT mice also showed significantly more caspase-3 activation compared to Prg4^+/− (N=5) and Prg4^+/+ (N=6) mice (p=.025, p<.001, Figure 4). Thus, Prg4^GTR7D/GTR7D and Prg4^{GTR14D/GTR14D} experimental groups, despite being born lubricin-null, display similar levels of caspase-3 activation found in wild type mice.

Figure 4.

Prg4 gene trap (GT) mouse knees, following whole joint pendulum friction testing, display reduced caspase-3 activation upon recombination. GT/GT mice recombined at 7 days of age (GT^R7D/GT^R7D), GT/GT mice recombined at 14 days (GT^R14D/GT^R14D) of age, lubricin heterozygous mice (+/−), and lubricin wild type mice (+/+) have significantly lower caspase-3 positive cells compared to the lubricin-null (GT/GT) mice (**p=.003, ***p <.001,*p=.025, ***p<.001). The GT^R7D/GT^R7D and GT^R14D/GT^R14D test groups, despite being born lubricin-null, display levels of caspase-3 positivity similar to levels of lubricin producing mice.

Peroxynitrite content in femoral head cartilage (N=5 for all groups) was significantly lower in Prg4^GTR7D/GTR7D and Prg4^{GTR14D/GTR14D}, as well as, the Prg4^+/+ and Prg4^+/− controls compared to the non-recombined Prg4^GT/GT mice ((p <.001, p<.001) and (p<.001, p<.001, Figure 5)) respectively. There were also significant differences between Prg4^GTR7D/GTR7D and Prg4^{GTR14D/GTR14D} mice compared to the Prg4^+/+ mice (p=.035, p=.002, Figure 5). This indicates that the peroxynitrite content in the mouse joints is dependent upon gene trap recombination, with both early recombination groups (days 7 and 14) exhibiting peroxynitrite values that were not significantly different from Prg4^+/− mice (p>0.05, p>0.05, Figure 5).

Figure 5.

Peroxynitrite concentrations were assayed for the femoral head cartilage of GT mice recombined at day 7 (GT^R7D/GT^R7D) and 14 (GT^R14D/GT^R14D) after birth. All testing groups, GT^R7D/GT^R7Dand GT^R14D/GT^R14D, as well as the lubricin wild type (+/+) and heterozygous (+/−) controls had significantly lower peroxynitrite content compared to the lubricin-null (GT/GT) mice (***p<.001 for all) respectively. There were also significant differences between GT^R7D/GT^R7D and GT^R14D/GT^R14D mice compared to the +/+ mice (*p=.035, **p=.002,).

Immunohistological Visualization of Hyaluronic Acid

Hyaluronate was found to be present in all mouse tibio-femoral joints (Figure 6A). The recombined Prg4^GTR7D/GTR7D (N=11), Prg4^{GTR14D/GTR14D} (N=11), and non-recombined Prg4^GT/GT (N=6) mice displayed more hyaluronate accumulation at the articular surface and on the synovium (Figure 6A). When the hyaluronate was quantified, Prg4^GT/GT, Prg4^GTR7D/GTR7D, and Prg4^{GTR14D/GTR14D} displayed significantly higher optical density values for hyaluronate accumulation in the tibio-femoral joint compared to Prg4^+/+ (N=6) mice (p=.02, p=.01, p=.009, Figure 6B). This indicates that hyaluronic acid accumulates on articular surfaces that are continually and initially lubricin deficient.

Figure 6.

Immunohistochemistry of hyaluronic acid (HA) in mouse tibio-femoral joints, following whole joint pendulum friction testing. A) HA accumulated more on the surface of tissues that were lubricin deficient. Lubricin wild type mice (+/+) exhibited lower quantities of HA accumulation on the articular and synovial surfaces. Lubricin-null (GT/GT), GT mice recombined at day 7 (GT^R7D/GT^R7D), and GT mice recombined at day 14 (GT^R14D/GT^R14D) displayed much thicker and vibrant HA accumulation at the articular and synovial surfaces. B) Quantification of HA expressed as optical density. Lubricinnull (GT/GT), GT mice recombined at day 7 (GT^R7D/GT^R7D), and GT mice recombined at day 14 (GT^R14D/GT^R14D) displayed significantly higher optical density values for HA accumulation in the mouse tibio-femoral joint compared to lubricin wild type mice (+/+) (*p=.02, *p=.01, **p=.009).

Discussion

A significant increase in friction at longer cycling time points was commensurate with a significant increase in cartilage damage in mice with no restoration or early restoration of lubricin gene expression compared to lubricin wild type and heterozygous mice. It was also found that early restoration of lubricin expression in lubricin-null mice significantly decreased caspase-3 activity compared to lubricin-null mice that have not had their lubricin expression restored. This suggests that activating lubricin in lubricin-null mice at the earlier time point of 7 and 14 days does have an effect on reducing cellular stresses within the cartilage even though whole joint COF was higher. A reduction in caspase-3 activity upon activating lubricin in lubricin-null mice was also observed in 21 day-old mice.[21] However, the present data suggest that moving the time of recombination and lubricin re-expression did not have the combined effect of both preventing caspase-3 activation and lowering whole joint COF, as posited.

The number of caspase-3 positive cells was used as an indicator of apoptosis, as the disappearance of chondrocytes, counted three-dimensionally within cartilage, is associated with caspase-3 activation.[1] Excessive mechanical stimulation due to cartilage surface friction may alter integrin connections between chondrocytes and the extracellular matrix, which are important in preserving the chondrocyte phenotype[26, 27]. Alternatively, exaggerated mechanical strain transmitted into the cytoskeleton has been linked to activation of intrinsic apoptotic pathways [27–29] which in the present case is friction-induced. These pathways are initiated by mitochondrial dysregulation,[30, 31] which in our experiments appears associated with cartilage surface friction. Caspase-3 is activated by both extrinsic, via caspases-8 and 10, and intrinsic pathways, via caspases-9.[32]

Despite the lack of clinical inflammation in human cases of CACP and the paucity of inflammatory cells observed histologically in lubricin-null mice,[33] which recapitulate CACP, it appears that the peroxynitrite content in cartilage is associated with lubricin expression and likely indicative of mitochondrial stress in lubricin-null joints[34]. In this study, early restoration of lubricin gene expression in lubricin-null mice is correlated with decreased peroxynitrite levels, such that the two earliest recombination dates (7 and 14 days after birth) resulted in levels of peroxynitrite that were not significantly different than those detected in Prg4^+/− mice. Thus, the inflammatory component of the lubricin-null state is mitigated by early lubricin gene restoration.

The effects of the lubricin-null phenotype, which include damage of the articular cartilage surfaces, are not derived from lack of hyaluronic acid (HA), the glycosaminoglycan responsible for the viscosity of synovial fluid.[35] The accumulation of HA in lubricin-null joints, and lubricin-null joints that have undergone early restoration of lubricin gene expression, appears to occur within the cleavable biofilm layer[21, 33] in the lubricin deficient mouse groups and is not as prominent in the Prg4^+/+ and Prg4^+/− mice. Thus, increased HA content in joints with compromised lubricin gene expression may actually increase whole joint friction and be detrimental to joint homeostasis. This observation confounds a recent theory that lubricin does not function as a lubricant but acts as an anchor for HA which is decorated with surface active phospholipids, in turn providing lubrication.[36] Aspirated synovial fluid from patients with CACP also does not demonstrate boundary lubricating ability in vitro in spite of having a supraphysiologic concentration of hyaluronate.[10, 37]

Mice with early restoration of lubricin gene expression have only partial preservation of the collagen ultrastructure[38] compared to lubricin wild type mice. This may be due to the direct protection of the superficial zone or early sparing of glycosaminoglycan loss, which occurs progressively in lubricin-null mice.[39] Additionally, mice with early restoration and mice without restoration of lubricin gene expression display increased thickness in the superficial collagen layer parallel to the articular surface. This increase in collagen layer thickness may be indicative of its disruption or of cartilage swelling, a phenomena detected in lubricin-null mice[1, 39] and a sign of arthrosis detected in human joints 2–3 years after anterior cruciate ligament reconstruction.[40] In a prior electron microscope study, the cartilage from newborn lubricin-null mice prior to weight bearing appear normal.[10] Newly expressed lubricin on a damaged surface may not be able to function as well as on an undamaged cartilage surface and thus lower COF to ~0.02 in a sustained manner. Overall, these findings indicate that lubricin is integral to preservation of the mechanical integrity of the mouse tibio-femoral joint.

Limitations of this study include recombination of the gene trap, which may not fully reconstitute lubricin expression until all 10 days of tamoxifen injections have been administered. Thus, damage from joints recombined at 7 or 14 days of age may be occurring over a period of 17 or 24 days respectively, as opposed to 7 or 14 days. Due to the 10 day lag time for full gene trap recombination in the transgenic mouse combined with the beginning of weight bearing at day 5 after birth, we may not be able to restore lubricin expression adequately before weight bearing begins. The electron micrographs for the Prg4^GTR2D/GTR2D mouse group illustrate that the lubricin-null condition has baseline cartilage ultra structural damage, within the context of early tamoxifen administration intended to restore lubricin expression. Thus, the use of a transgenic mouse that is lubricin-null at birth inherently limits this study.

Additionally, joint cycling duration for friction testing is completed ex vivo. In previous work, this particular gene trap mouse has been tribologically evaluated at day 21.[21] In this previous work, it was found that recombination at 21 days of age resulted in significantly lower friction compared to lubricin-null mice. However, that study grouped heterozygous and wild type mice together for lubricin producing controls and grouped Prg4^−/− and Prg4^GT/GT mouse controls together to represent the lubricin-null state. Thus, though COF and OA damage score values from this current work and the previous work by Hill et al[21] are similar, the differences in control groups may explain why we observed a smaller difference between recombined and non-recombined gene trap groups. Hill et al[21] also differs in that caspase-3 data were obtained from mice prior to cyclic loading while caspase-3 data were evaluated after 60 minutes of cyclic loading for the present study.

Conclusions

These studies indicate that the biological functionality of lubricin may be reconstituted in vivo through restoration of lubricin gene expression beginning at day 7 and 14 after birth. This early restoration of lubricin expression results in significantly decreased chondrocyte caspase-3 activity and significantly decreased peroxynitrite content compared to lubricin-null littermates that have not had lubricin expression restored. It appears that even partial restoration of a low COF can have an anabolic effect. However, the lubricin-null animals that undergo early restoration of lubricin gene expression appear to be unable to overcome the initial damage occurring within the first several days of ambulation coupled with the additional delay incurred as Cre-recombinase is activated to remove the gene trap. Thus, understanding the effect of a transient lubricin-null state in mice expressing lubricin normally from birth would be important in determining the effects of temporary lubricin reduction in larger mammals and humans. This scenario would recapitulate the effects of lubricin down regulation and later restoration of lubricin expression in joints that have undergone trauma or sports injury.[41] These models will provide a better understanding of post-traumatic osteoarthritis and how it might be mitigated via lubricin supplementation.[12]