In Vivo Luminescence Imaging of NF-κB Activity and Serum Cytokine Levels Predict Pain Sensitivities in a Rodent Model of Osteoarthritis

Authors


Abstract

Objective

To investigate the relationship between NF-κB activity, cytokine levels, and pain sensitivities in a rodent model of osteoarthritis (OA).

Methods

OA was induced in transgenic NF-κB–luciferase reporter mice via intraarticular injection of monosodium iodoacetate (MIA). Using luminescence imaging we evaluated the temporal kinetics of NF-κB activity and its relationship to the development of pain sensitivities and serum cytokine levels in this model.

Results

MIA induced a transient increase in joint-related NF-κB activity at early time points (day 3 after injection) and an associated biphasic pain response (mechanical allodynia). NF-κB activity, serum interleukin-6 (IL-6), IL-1β, and IL-10 levels accounted for ∼75% of the variability in pain-related mechanical sensitivities in this model. Specifically, NF-κB activity was strongly correlated with mechanical allodynia and serum IL-6 levels in the inflammatory pain phase of this model (day 3), while serum IL-1β was strongly correlated with pain sensitivities in the chronic pain phase of the model (day 28).

Conclusion

Our findings suggest that NF-κB activity, IL-6, and IL-1β may play distinct roles in pain sensitivity development in this model of arthritis and may distinguish the acute pain phase from the chronic pain phase. This study establishes luminescence imaging of NF-κB activity as a novel imaging biomarker of pain sensitivities in this model of OA.

Knee osteoarthritis (OA) is a leading cause of disability, with large associated economic costs. It is anticipated that 20% of adults will be affected by pain or disability associated with OA by the year 2030 ([1]). At the tissue level, the disease is associated with cartilage degradation, synovial hypertrophy, and subchondral bone remodeling. Clinically, the disease is diagnosed by radiographic changes that include joint space narrowing, presence of osteophytes, and synovial hypertrophy. While patients present with joint pain and impaired joint mobility ([2]), the mechanistic link between these features and tissue level changes is poorly understood ([3-5]).

Direct or indirect interaction of inflammatory cytokines (e.g., tumor necrosis factor α [TNFα], interleukin-1β [IL-1β], and IL-6) with knee joint afferents has been suggested as a possible contributor to pain in arthritis ([6]). The inflammatory cytokines TNFα and IL-6 have been demonstrated to play a direct role in the sensitization of nociceptors and the development of mechanical allodynia in preclinical models of inflammatory arthritis ([7, 8]). Elevated levels of these proinflammatory cytokines have been observed in OA patients ([6, 9, 10]), while TNFα and IL-1β antagonism may lead to substantial reductions in pain scores in patients presenting with an inflammatory component of OA ([11-13]). While the presentation of inflammation may vary widely across patients with OA and even within the time course of the disease, it is now thought that inflammatory cytokines play a key role in pain development in OA.

A key regulator of TNFα, IL-1β, and IL-6 is the transcription factor NF-κB. The NF-κB transcription factor family induces the expression of more than 150 genes that play a role in immunity, inflammation, antiapoptosis, cell proliferation, and the negative feedback of NF-κB ([14-17]). NF-κB activity can be induced by multiple cellular stimuli, including inflammatory cytokines (e.g., TNFα and IL-1β), extracellular matrix degradation products, and mechanical overload. Once activated, NF-κB induces target gene expression in multiple cell types, including synoviocytes, chondrocytes, and fibroblasts ([18-20]). In arthritis, NF-κB activity is associated with increased expression of proinflammatory cytokines (IL-1β, TNFα, and IL-6), metalloproteinases, chemokines, and inducible enzymes (cyclooxygenase 2) ([21-24]). Although the role of NF-κB in arthritis pain is not completely understood, the inhibition of key mediators of the NF-κB activation pathway leads to a reduction in pain in rodent models of arthritis ([25, 26]), suggesting that regulation of these key mediators is the means by which NF-κB facilitates the development and propagation of pain in arthritis.

Carlsen and coworkers previously described the use of a transgenic NF-κB–luciferase reporter mouse to study the transient kinetics of NF-κB activation after ultraviolet B radiation exposure and collagen antibody–induced arthritis ([17, 27]). This transgenic mouse contains NF-κB response elements upstream of the firefly luciferase gene, which allows NF-κB activity to be noninvasively and longitudinally quantified via luminescence imaging; this has been used to demonstrate NF-κB inhibition using a small molecule inhibitor of IKK-2 in a model of collagen antibody–induced arthritis ([28]).

For its repeatability and rapid onset of pathology, intraarticular injection of monosodium iodoacetate (MIA) has been widely used to induce arthritis ([26, 29-37]). Injection of MIA into the joint inhibits glycolysis in chondrocytes, which results in chondrocyte cell death. As a result, changes resembling human OA, including proteoglycan loss, collagen fibrillation, and subchondral bone remodeling, progress rapidly over a 4–6-week period ([29-34, 37, 38]). A biphasic pain response, which includes an early inflammation-associated pain phase followed by a late noninflammatory pain phase, has been documented in the MIA model ([36, 37]). The MIA model is of great interest for studying the transient role of NF-κB activity and cytokine levels in contributing to both inflammatory and noninflammatory pain in a model of OA.

In the present study, in vivo luminescence imaging of NF-κB activity was used to investigate the relationship between NF-κB activity, serum cytokines, and pain sensitivities in a rodent model of MIA-induced OA. This model was evaluated to test whether in vivo luminescence imaging could provide a spatially accurate measure of NF-κB activity in the knee joint and to test its relationship to the development of disease.

MATERIALS AND METHODS

Intraarticular injections and experimental time line

Transgenic BALB/c-Tg(NFκB-RE-luc) mice (ages 7–8 weeks; n = 24) (Taconic Farms), engineered to carry complementary DNA for firefly luciferase downstream of NF-κB response elements, were used. All procedures were performed with the approval of the Duke University Institutional Animal Care and Use Committee. Following a 1-week acclimation period, mice were anesthetized with 2% isoflurane inhalation. Mice (n = 12 per group) were injected intraarticularly with either MIA (5 μl, 10 mg/ml) or saline (5 μl) into the knee via a 30-gauge needle. One set of animals (n = 6 mice injected with saline and n = 6 mice injected with MIA) was killed on day 3 after injection, and a second set of animals (n = 6 mice injected with saline and n = 6 mice injected with MIA) was killed on day 28 after injection. These time points were selected to coincide with the early pain phase (day 3) and late pain phase (day 28) observed in the MIA model. Animals underwent in vivo luminescence imaging to quantify NF-κB activity before injection of MIA or saline and on days 1, 3, 7, 14, 21, and 28 after injection of MIA or saline. Mice underwent testing to measure mechanical allodynia and relative weight-bearing at these same time points. In a subset of animals, organs and relevant tissues were harvested for ex vivo imaging immediately postmortem. Serum was obtained from all animals (on day 3 or day 28) via cardiac puncture to assay for a subset of systemic proinflammatory and antiinflammatory cytokines.

In vivo luminescence imaging of NF-κB activity

At each imaging time point, mice were anesthetized and injected intraperitoneally with 150 mg/kg of D-luciferin (Caliper Life Sciences) prior to imaging. After 10 minutes to allow D- luciferin distribution, whole-body imaging of luminescence arising from NF-κB activity was performed on an in vivo imaging system (IVIS 100; PerkinElmer) with a 15-second exposure, determined from preliminary studies to avoid saturation of pixels in all images.

Six mouse cadavers were used to define an appropriate region of interest (ROI) for luminescence image analysis. Radiographs of mouse cadavers were obtained on an In Vivo FX Pro imaging system (Carestream Health) using the same positioning as for luminescence imaging. An ROI was selected (4 × 8 mm ellipse) that was anchored to the intersection of the ipsilateral hind limb and mouse body (Figure 1A) and encompassed the total knee joint in all animals.

Figure 1.

A, Selection of the region of interest (ROI) in the mouse knee. A radiograph (i), reflected light image of the ROI (ii), and an overlay of the ROI upon a luminescence image of the mouse limb (iii) are shown. B, In vivo imaging of NF-κB activity over 28 days in the knees of mice injected with monosodium iodoacetate (MIA) and mice injected with saline. C, Quantitative fold change in knee ROI luminescence (NLUM) compared to preinjection (preinj) levels over 28 days. Bars show the mean ± SD (n = 6 mice per group). # = P < 0.05 versus preinjection levels; ∗ = P < 0.05 versus levels in saline-injected mice.

At each time point, the ROI for luminescence (LUMROI) was measured as photons/second/cm2 for all animals. Luminescence data for each animal at each time point was normalized to preinjection values from the same animal (NLUM) according to the following equation:

display math

Ex vivo luminescence imaging of NF-κB activity

Immediately postmortem, the ipsilateral and contralateral tibiofemoral knee joints, patella, and adjacent muscle tissue were harvested for immediate ex vivo luminescence imaging in a subset of animals (n = 18). Tissue specimens were placed in separate wells of a 12-well tissue culture plate, and each well was supplemented with 300 μg/ml of D-luciferin in Dulbecco's phosphate buffered saline (Gibco). After 5 minutes, the tissue culture plate was imaged following the same procedure as for the whole body on an IVIS 100 (PerkinElmer; 15-second exposure). Selected ROIs around each tissue specimen were measured as photons/second/cm2 (LUMKNEE, LUMPATELLA, LUMMUSCLE).

Pain-related sensitivity (mechanical allodynia).

Mice were acclimated to a wire-bottomed cage and exposed to the von Frey testing procedure over a period of 3 days prior to the start of the experiment. Mechanical allodynia was evaluated in the ipsilateral hind paw by touching the plantar surface with von Frey filaments (0.07–4.0 gm; Stoelting) using the “up-down” method ([39]). Briefly, filaments ranging from 0.07 gm to 4.0 gm were applied to the ipsilateral hind paw for 3–4 seconds, starting with the 0.6-gm filament. A positive response (paw flick, lick, or vocalization) resulted in moving to the next weaker filament, while a lack of response resulted in moving to the next stronger filament. Data for paw withdrawals following filament application for a minimum of 4 filaments were used to calculate the 50% withdrawal threshold (50%WT) in units of gram-force. Data were collected preinjection and on days 3, 7, 14, 21, and 28 after injection and normalized to the preinjection levels for the same animal and limb (N50%WT); data are presented as a percentage of the preinjection 50%WT according to the following equation:

display math

Weight bearing

Mouse hind limb weight bearing was determined using an incapacitance meter (IITC Life Science). Mice were placed in an angled plexiglass chamber with the 2 hind paws resting on 2 independent force plates, the front paws resting on the plexiglass ramp, and the mouse facing forward. Force (gram-force) readings were obtained over a 3-second interval for both the ipsilateral (WBipsi) and contralateral (WBcontr) limb; 3 independent measurements were obtained for each mouse at each time point and averaged. Percentage weight-bearing was normalized (NWB) by dividing the weight-bearing on the ipsilateral limb by the total weight-bearing for both hind limbs according to the following equation:

display math

Measurement of serum cytokine concentrations

The following serum cytokine concentrations were quantified using multiplex and single immunoassays (Mouse Proinflammatory 7 plex and Mouse MCP-1 single plex; Meso Scale Discovery): IL-1β, IL-6, IL-10, IL-12p70, CXCL1, TNFα, interferon-γ (IFNγ), and monocyte chemotactic protein 1 (MCP-1). Measurements were performed on a Sector Imager 2400 (Meso Scale Discovery). The mean intraassay coefficients of variation for the 7-plex proinflammatory and MCP-1 immunoassays were 3.3% and 2.8%, respectively. The lower limit of detection for each cytokine was as follows: for IL-1β, 0.21 pg/ml; for IL-6, 3.07 pg/ml; for IL-10, 0.37 pg/ml; for IL-12p70, 9.73 pg/ml; for CXCL1, 0.27 pg/ml; for TNFα, 0.63 pg/ml; for IFNγ, 0.06 pg/ml; and for MCP-1, 1.46 pg/ml. One-half of the lower limit of detection was used for any value below the level of detection for the purpose of performing statistical analyses. Concentrations of serum TNFα and IL-12p70 were below the lower limit of detection in a majority of samples and were not included in the analysis.

Histologic analysis

Both ipsilateral and contralateral limbs were dissected to remove excess skin and musculature while preserving the knee joint capsule. The intact joints were fixed in 10% formalin for 3 days, decalcified for 5 days in Cal-EX decalcifying solution (Fisher Scientific), and embedded in paraffin. Sections (8 μm thick) were obtained from the coronal plane within the load-bearing region of the joint and mounted at 80-μm intervals. Three slides covering the central sections of each joint were stained with Safranin O–fast green. A single section representing the most severe evidence of arthritic changes was selected for each joint and was scored according to a modified Mankin scale ([40]). Images were randomized, and 2 scorers (BAM and RDB) who were blinded with regard to treatment group assigned grades, by consensus, to the tibial and femoral cartilage in the medial and lateral compartments. Ordinal scores were assigned for 7 parameters related to degeneration, including cartilage structure (score range 0–11), loss of Safranin O staining (range 0–8), tidemark duplication (range 0–3), fibrocartilage formation (range 0–2), chondrocyte clones in uncalcified cartilage (range 0–2), hypertrophic chondrocytes in calcified cartilage (range 0–2), and subchondral bone thickness (range 0–2). Scores were summed for the lateral femur, medial femur, lateral tibia, and medial tibia for a maximum possible score of 120 for each knee.

Statistical analysis

All statistical analyses were performed using JMP Pro software. Continuous, normally distributed data (N50%WT, NWB, and NLUM) were analyzed by two- way analysis of variance (ANOVA) with Tukey's post hoc test, treating day and group (MIA or saline) as factors. Significance was tested at α = 0.05. Ordinal (Mankin score) and non–normally distributed data (serum cytokine levels) were analyzed by Friedman's test. Post-hoc analysis with Wilcoxon's signed rank tests was conducted with Bonferroni correction for 6 comparisons (α = 0.008).

To test for the hypothesized relationships between NF-κB activity (NLUM), serum levels of cytokines (IL-6, CXCL1, IL-1β, MCP-1, IFNγ, and IL-10), and pain-related measures (N50%WT and NWB), multivariate linear regression was performed using both day 3 and day 28 data (all measures collected on these days) for both mice treated with MIA and mice treated with saline. The significance of the regressions was reported with Bonferroni correction for 45 comparisons (α = 0.001). All variables except N50%WT and NWB were log- transformed (base 10) prior to statistical analysis.

To test for the ability of luminescence imaging of NF-κB activity and serum cytokine measures to account for pain-related sensitivity in the MIA model, N50%WT and NWB were modeled using partial least squares regression to produce a linear model from the log-transformed variables NLUM, IL-6, CXCL1, IL-1β, MCP-1, IFNγ, and IL-10. A threshold for variable influence on projection (VIP) was set at 0.8 for inclusion in the final linear model ([41]). The VIP parameter is a summarizing tool that describes the relative importance of the predictors in order to rank the importance of each predictor in relation to a chosen response variable. Both VIP and scaled and centered regression coefficients are reported for independent variables included in the final model. K-fold cross-validation with 6 subsamples was used to calculate Q2, an estimate of the predictive power of the models.

To test for relationships that were specific to pain phase (early [day 3] or late [day 28]), additional multivariate linear regressions were performed to test for an association of 50% withdrawal threshold (N50%WT) with independent variables (NLUM, IL-6, IL-1β, and IL-10) on day 3 or day 28. The significance of the regression was reported with Bonferroni correction for 4 comparisons (α = 0.0125) for both day 3 and day 28. Linear regression was performed to test for an association of weight bearing (NWB) with independent variables (NLUM, IL-6, CXCL1, and MCP-1) on day 3 or day 28. The significance of the regression was reported with Bonferroni correction for 4 comparisons (α = 0.0125) for both day 3 and day 28. Independent variables were log-transformed (base 10) prior to statistical analysis.

RESULTS

NF-κB activity in mice injected with MIA

A transient increase in NF-κB activity luminescence (NLUM) (∼4–5 fold) was observed following MIA injection into the mouse knee joint (Figures 1B and C). On day 1, luminescence was observed consistently and was broadly distributed in the limbs of mice injected with MIA. The intensity of luminescence peaked at the knee ROI on day 1 postinjection and slowly and steadily declined thereafter (Figure 1B). For mice injected with MIA, luminescence (NLUM) was significantly increased in the knee ROI on day 1 and day 3 compared to preinjection values (P < 0.0001 for both, by ANOVA) (Figure 1C), but returned to preinjection levels by day 28. Additionally, luminescence (NLUM) was significantly increased on day 3 in the knees of mice injected with MIA compared to those injected with saline (P = 0.043 by ANOVA).

Correlations between in vivo luminescence and ex vivo luminescence of NF-κB activity

In vivo luminescence (LUMROI) was significantly correlated with ex vivo tibiofemoral knee joint luminescence (LUMKNEE) (R2 = 0.71, P < 0.0001) and ex vivo patella luminescence (LUMPATELLA) (R2 = 0.52, P < 0.0001), with evidence of weaker correlations to ex vivo muscle luminescence (LUMMUSCLE) (R2 = 0.16, P < 0.016) (Figure 2). This observation confirms that the tibiofemoral joint was the primary source of NF-κB–generated luminescence in vivo.

Figure 2.

A, Ex vivo luminescence imaging of ipsilateral and contralateral knee joints, patella, and adjacent muscle tissue from a mouse injected with monosodium iodoacetate. Linear correlation coefficients for the correlation between ex vivo luminescence and in vivo knee region of interest (ROI) luminescence (LUMROI) are shown. The values in parentheses are the P values for correlation. B, Linear correlation between in vivo knee ROI luminescence and ex vivo knee luminescence (LUMKNEE).

Serum cytokine levels

Serum IL-6 levels were transiently increased following injection of MIA into the mouse knee joint (Figure 3). Serum IL-6 levels were significantly increased on day 3 in animals injected with MIA compared to those injected with saline (P = 0.004 by Friedman's test) and were significantly increased on day 3 compared to day 28 in animals injected with MIA (P < 0.0001 by Friedman's test). There was no evidence of between-group or time-related differences in serum levels of CXCL1, MCP-1, IFNγ, IL-1β, or IL-10 (P > 0.05 by Friedman's test).

Figure 3.

Serum levels of interleukin-6 (IL-6), CXCL1, monocyte chemotactic protein 1 (MCP-1), interferon-γ (IFNγ), IL-1β, and IL-10 on day 3 and day 28 in mice injected with monosodium iodoacetate (MIA) and mice injected with saline. Data are shown as box plots. Each box represents the 25th to 75th percentiles. Lines inside the boxes represent the median. Lines outside the boxes represent the maximum and minimum values, excluding outliers. The circle indicates an outlier. ∗ = P < 0.05 versus all other groups.

Histologic evidence of arthritis

Histologic scoring indicated that the knees of mice injected with MIA showed the greatest change in proteoglycan loss, fibrillation/clefts of the articular cartilage surface, and hypertrophic chondrocytes in the calcified cartilage (Figure 4A). On day 28, the total modified Mankin score for the limbs of mice injected with MIA was significantly higher than that for the limbs of mice injected with saline (P = 0.005 by Friedman's test) (Figure 4B).

Figure 4.

A, Representative Safranin O–fast green–stained histologic sections from saline-injected mice and monosodium iodoacetate (MIA)–injected mice on day 3 and day 28 after injection. B, Total Mankin score for the limbs of saline-injected and MIA-injected mice on day 3 and day 28 after injection. ∗ = P < 0.05 versus saline-injected mice, by Friedman's test. C and D, Normalized measures of mechanical allodynia (50% withdrawal threshold [N50%WT]) (C) and weight bearing (NWB) (D) in the ipsilateral limbs of MIA-injected mice and saline-injected mice over 28 days. Values in B–D are the mean ± SD. In C and D, # = P < 0.05 versus preinjection (preinj) values; ∗ = P < 0.05 versus saline-injected mice, by two-way analysis of variance.

Behavioral assessments

Injection of MIA into the mouse knee resulted in the induction of significantly elevated sensitivity to mechanical stimuli (allodynia) in the ipsilateral limb (Figure 4C). Compared to saline-injected mice, MIA-injected mice had significantly decreased 50% withdrawal threshold normalized to preinjection values (N50%WT) for the hind limbs on day 3 (P < 0.0001 by ANOVA), day 7 (P = 0.0043 by ANOVA), day 21 (P = 0.0021 by ANOVA), and day 28 (P = 0.0033 by ANOVA). The N50%WT values for the limbs of MIA-injected mice were also significantly decreased on day 3 (P < 0.0001 by ANOVA) and day 7 (P < 0.0001 by ANOVA) compared to preinjection values (Figure 4C). On day 28, MIA-injected limbs showed a trend toward decreased N50%WT compared to preinjection levels, but the difference failed to reach the threshold for statistical significance (P = 0.082 by ANOVA). The percent of weight bearing on the ipsilateral limb significantly decreased in MIA-injected mice on day 3 (P = 0.0002 by ANOVA) and day 7 (P = 0.035 by ANOVA) compared to preinjection values and on day 3 (P = 0.023 by ANOVA) compared to saline-injected mice (Figure 4D).

Relationships between NF-κB activity, serum cytokine levels, and pain-related sensitivities

NF-κB activity, serum cytokine levels, and pain-related sensitivities (Figures 5 and 6) were all significantly associated in the MIA model. After adjusting for multiple comparisons by Bonferroni correction, multivariate linear regression identified a correlation for NF-κB activity (luminescence) and pain-related sensitivity data when analyzed independently of pain phase (i.e., early [day 3] or late [day 28]). NF-κB activity (NLUM) was strongly negatively correlated with the 50% withdrawal threshold (N50%WT) (R = −0.70, P < 0.0001) (Figure 5). No evidence of additional significant correlations between NF-κB activity (NLUM), mechanical allodynia (N50%WT), weight-bearing (NWB), and serum levels of IL-6, CXCL1, IL-1β, MCP-1, IFNγ, and IL-10 was detected when these variables were analyzed independently of pain phase.

Figure 5.

The 50% withdrawal threshold plotted against NF-κB activity (LUMROI) for the ipsilateral limbs of monosodium iodoacetate–injected mice and saline-injected mice at all time points (preinjection [preinj] and days 3, 7, 14, 21, and 28).

Figure 6.

A, Partial least squares linear regression models for mechanical allodynia (50% withdrawal threshold [N50%WT]) and weight bearing (NWB) in monosodium iodoacetate (MIA)–injected mice and saline-injected mice. B and C, Absolute values of multivariate linear correlations for variables in respective partial least squares linear models with mechanical allodynia (B) and weight bearing (C) in MIA-injected mice and saline-injected mice by day. ∗ = P < 0.05. NLUM = NF-κB activity luminescence; IL-6 = interleukin-6; MCP-1 = monocyte chemotactic protein 1; VIP = variable influence on projection. Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/doi/10.1002/art.38279/abstract.

A linear model was constructed for mechanical allodynia (N50%WT) and weight bearing (NWB), using the noninvasive measure of NF-κB activity (NLUM) and serum cytokine level (Figure 6A). The model for mechanical allodynia described ∼75% (R2) of the variability in mechanical allodynia (N50%WT), using the variables NLUM (VIP = 1.7), IL-6 (VIP = 1.3), IL-1β (VIP = 0.8), and IL-10 (VIP = 0.9). Using K-fold cross-validation, the model predicted ∼67% (Q2) of the variability in the data. The model for weight bearing described ∼27% (R2) of the variability in weight bearing (NWB), using the variables NLUM (VIP = 1.6), IL-6 (VIP = 1.3), CXCL1 (VIP = 1.3), and MCP-1 (VIP = 1.1). However, using K-fold cross-validation, the model predicted only ∼7% (Q2) of the variability in the data.

Of the significant parameter estimates observed in the general linear model, only NF-κB–related luminescence (NLUM) was significantly correlated with mechanical allodynia (N50%WT) when analyzed independently of pain phase (Figure 5). Multivariate correlations were conducted to compare the significant variables in the linear models to mechanical allodynia and weight bearing by day. Adjusting for multiple comparisons, luminescence (NLUM) and the 50% withdrawal threshold (N50%WT) were strongly negatively correlated on day 3 (R = −0.78, P = 0.0029), serum IL-6 and N50%WT were strongly negatively correlated on day 3 (R = −0.67, P = 0.011), and serum IL-1β and N50%WT were strongly negatively correlated on day 28 (R = −0.72, P = 0.0088) (Figure 6B). NF-κB activity (NLUM) and ipsilateral weight bearing (NWB) were also strongly negatively correlated on day 3 (R = −0.71, P = 0.012) (Figure 6C).

DISCUSSION

We used luminescence imaging to investigate the relationship between NF-κB activity, cytokine levels, and pain sensitivities in a rodent model of OA. The tibiofemoral joint appeared to be the principal source of NF-κB–related luminescence following MIA injection in mice, and revealed strong relationships between local measures of NF-κB activity, pain sensitivity, and systemic measures of IL-6. These findings establish luminescence imaging of NF-κB activity as a novel imaging biomarker of pain-related sensitivities in a rodent model of OA.

Injection of MIA into NF-κB–luciferase reporter mice produced both histologic and behavioral changes associated with the development of arthritis ([26, 35]). Significant histologic changes were observed in the knees of MIA-injected mice by day 28 after injection. A biphasic pain response was observed, with an acute increase in mechanical sensitivity followed by a persistent increase in sensitivity in the limbs of MIA-injected mice compared to those of saline-injected mice on days 21 and 28. Similar biphasic pain responses that include an early inflammation-associated pain phase and a late noninflammation-associated pain phase have been reported with regard to weight bearing and spontaneous mobility in the MIA model ([36, 37]). Thus, the biphasic response observed in the mechanical sensitivity data is consistent with these previous findings for pain-related behavioral measures.

The biphasic response has been reported in the rat model at moderate concentrations of MIA (6–60 mg/ml) ([36, 37]), while high concentrations (50– 100 mg/ml) produce a pain response that is characterized by rapid onset and no recovery and is sustained chronically in both mice and rats ([33, 42, 43]). The concentration used in this study (10 mg/ml) was within the range expected to produce a biphasic pain response. It is important to note that, while present, the late pain phase observed in this study was moderate and was observed only in the mechanical allodynia measure. A more severe chronic pain response with regard to both mechanical allodynia and weight bearing would have likely been observed with injection of higher concentrations of MIA.

In vivo luminescence imaging provides a powerful tool to noninvasively quantify molecular events during musculoskeletal disease progression and therapeutic intervention. NF-κB activity was observed ex vivo in the tibiofemoral knee joint and patella with a strong correlation with the in vivo ROI luminescence measurements. The ex vivo imaging established the tibiofemoral joint, and to a lesser extent the patella, as the sources of the in vivo total knee ROI luminescence. In vivo luminescence imaging as used in the present study provides a spatially accurate measure of NF-κB activity in the knee that may be useful for the rapid screening of NF-κB–targeting therapeutics for the treatment of OA and other inflammatory diseases ([18, 44]).

MIA injection in mice induced a transient increase in NF-κB activity that preceded histologic changes in the knee. Immediately after injection, elevated NF-κB activity was observed throughout the limb on day 1, was localized to the knee by day 3, and returned to preinjection levels by day 28. This pattern was consistent with the pattern previously reported for MIA injection, namely an inflammatory response in the first 1–3 days after injection, followed by a subsequent dissipation of inflammation ([29, 30]). Additionally, NF-κB activity has been demonstrated during inflammation in models of collagen-induced and adjuvant-induced arthritis ([28, 45]). A peak in joint swelling on day 1, a peak in synovial hyperplasia on day 3, and synovial inflammatory cell invasion characterize this inflammatory response through day 3 in the MIA model. It is possible that the day 3 time point chosen for this study may have missed the peak levels of inflammation and pain sensitivity, but the elevated NF-κB activity and IL-6 levels observed on day 3 indicate that this time point falls within the inflammatory period in this model. Our data suggest that the primary role of NF-κB in the MIA model is as a mediator of the early inflammatory response and may suggest that NF-κB–targeted therapeutics would be most effective during the early developmental phases of OA.

In this study, luminescence imaging of NF-κB activity and NF-κB–regulated serum cytokines were validated as biomarkers of pain-related sensitivities in the MIA model. NF-κB activity was very strongly correlated with mechanical allodynia and was the only measure with significant correlation with mechanical allodynia when analyzed independently of pain phase. A linear model was constructed that was capable of predicting ∼67% (Q2) of the variability in the mechanical allodynia data, which included luminescence imaging of NF-κB activity, serum IL-6, serum IL-1β, and serum IL-10 as predictive variables. NF-κB activity was identified as the most important variable in the model according to VIP. When these measures were correlated with mechanical allodynia by pain phase, NF-κB activity and serum IL-6 were strongly correlated with mechanical sensitivities in the early pain phase, and IL-1β was strongly correlated with mechanical sensitivities in the late pain phase. These data support the presence of an early inflammatory pain phase and a late noninflammatory pain phase in the MIA model of OA and suggest that distinct biomarkers can distinguish the 2 phases. These findings may indicate a distinct role for the NF-κB/IL-6 pathway in the acute pain sensitivities in arthritis and for IL-1β in the chronic pain sensitivities associated with joint and cartilage destruction. This is consistent with previous findings that both IL-6 and IL-1β are associated with increased pain sensitivities in the rodent knee joint ([7, 46]). Overall, our findings suggest that luminescence imaging of NF-κB activity and serum cytokine measures may be useful as a panel of noninvasive biomarkers of pain-related sensitivities in the MIA model, with biomarker subsets unique for distinguishing the distinct pain phases.

This study demonstrates the use of noninvasive in vivo luminescence imaging to measure NF-κB activity in a mouse model of OA and the comparison of key molecular events in arthritis to pain sensitivity development. MIA injection in mice induced a transient increase in NF-κB activity in the tibiofemoral joint at early time points, which was correlated with developing pain sensitivities in this model of OA. NF-κB activity and serum IL-6 were specifically related to the inflammation-associated acute pain phase, and IL-1β was related to the late pain phase following MIA injection. The demonstrated relationship between NF-κB activity, a subset of serum cytokines (i.e., IL-6, IL-1β, and IL-10), and mechanical allodynia suggests that NF-κB luminescence imaging and serum cytokine levels can be used as noninvasive biomarkers of pain sensitivities in this model. It will be important in future work to test how these results translate to human OA, and it is important to note that the transgenic nature of the luminescence imaging of NF-κB activity makes this imaging a challenge in the clinical setting. Our data suggest the feasibility and utility of luminescence imaging of NF-κB for rapid screening of targeted NF-κB antagonists for treating knee OA and other diseases driven by NF-κB.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Setton had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Bowles, Mata, Setton.

Acquisition of data. Bowles, Mata, Bell, Mwangi, Huebner, Kraus, Setton.

Analysis and interpretation of data. Bowles, Mata, Kraus, Setton.

Acknowledgments

We thank Stephen Johnson for assistance with the surgical procedures and Gregory Palmer for guidance on experiments with IVIS imaging.

Ancillary