Cyclooxygenase-2 or tumor necrosis factor-α inhibitors attenuate the mechanotransductive effects of pulsed focused ultrasound to suppress mesenchymal stromal cell homing to healthy and dystrophic muscle

Abstract

Maximal homing of infused stem cells to diseased tissue is critical for regenerative medicine. Pulsed focused ultrasound (pFUS) is a clinically relevant platform to direct stem cell migration. Through mechanotransduction, pFUS establishes local gradients of cytokines, chemokines, trophic factors (CCTF) and cell adhesion molecules (CAM) in treated skeletal muscle that subsequently infused mesenchymal stromal cells (MSC) can capitalize to migrate into the parenchyma. Characterizing molecular responses to mechanical pFUS effects revealed tumor necrosis factor-alpha (TNFα) drives cyclooxygenase-2 (COX2) signaling to locally increase CCTF/CAM that are necessary for MSC homing. pFUS failed to increase chemoattractants and induce MSC homing to treated muscle in mice pretreated with ibuprofen (non-specific COX inhibitor) or etanercept (TNFα inhibitor). pFUS-induced MSC homing was also suppressed in COX2-knockout mice, demonstrating ibuprofen blocked the mechanically-induced CCTF/CAM by acting on COX2. Anti-inflammatory drugs, including ibuprofen, are administered to muscular dystrophy (MD) patients and ibuprofen also suppressed pFUS-induced homing to muscle in a mouse model of MD. Drug interactions with cell therapies remain unexplored and are not controlled for during clinical cell therapy trials. This study highlights potentially negative drug-host interactions that suppress stem cell homing and could undermine cell-based approaches for regenerative medicine.

INTRODUCTION

Noninvasively directing stem cell homing towards pathological sites to down-regulate inflammation, differentiate and/or stimulate repair is an obstacle to the future use of cell-based therapy [1]. The mechanisms of mesenchymal stromal cell (MSC) homing after intravascular infusion are similar to leukocyte homing to sites of inflammation, typically a passive or active process. Active homing of MSC involves tethering of its surface integrins to cell adhesion molecules (CAM) on vascular endothelial cells [2], followed by firm adhesion and chemoattractant gradients initiating transmigration into the parenchyma [3, 4]. Homing of the infused cells is inefficient [5, 6] and may require active molecular gradients that are present during acute but not chronic phases of pathology. Techniques need to be developed to improve cell homing to areas of pathology [7] without directly modifying cells or increasing their surface integrin expression [7, 8]. Diagnostic and therapeutic ultrasound are common techniques used in rehabilitative medicine and ablative therapy for various malignancies [9]. Pulsed focused ultrasound (pFUS) noninvasively localizes pressure waves to focally deposit energy in targeted regions and induce biological changes [10–13]. The tissue effects of FUS can be thermal or mechanical, which include acoustic cavitation and radiation forces (pressure) [14]. Here, pFUS was used to generate chemoattractant gradients composed of cytokines, chemokines, trophic factors (CCTF), and cell adhesion molecules (CAM) in skeletal muscle that were characterized by transient physiological changes without thermal or irreversible mechanical damage. pFUS transiently alters the tissue microenvironment to change levels of local chemoattractants that define a “molecular zip code” or a spatially-defined volume of tissue that where enhanced homing permeability and retention (EHPR) of infused MSC occurs [11, 13]. Molecular alterations to the microenvironment allowed significantly greater numbers of MSC to home to pFUS-treated muscle. Pretreating mice with ibuprofen (COX inhibitor) [15] or etanercept (TNFα inhibitor) [16] before pFUS significantly suppressed chemoattractant expression and MSC homing to treated muscle that was confirmed in COX2-knockout mice. Ibuprofen also had significant effects on MSC homing to dystrophic pFUS-treated muscle in a mouse model of muscular dystropy (MD). These results call attention to potential adverse effects of clinically relevant drugs on stem cell homing that could confound cell therapy approaches in regenerative medicine are currently are not controlled for in clinical trials.

MATERIALS AND METHODS

Animals

All animal studies were performed in accordance with the guidelines and regulations set forth our institutions Animal Care and Use Committee. The following mouse strains were used: C3H/HeNCrl (Charles River Laboratories, Wilmington, MA), B6;129P2-Ptgs2^tm1Unc (COX2^−/− knockout) (Jackson Laboratory, Bar Harbor, ME), or C57BL/10ScSn-Dmd^mdx/J (MDX). All mice were female and 20–25 g before experimentation. MDX mice were 9 weeks old at the time of experimentation.

Pulsed Focused Ultrasound

A modified Sonoblate 500 system (Focus Surgery, Indianapolis, IN) was used for all exposures. The device consisted of a concave therapeutic transducer (5 cm diameter, 4 cm focal length, 1 MHz operating frequency) and a collinear imaging transducer (8 mm aperture, 10 MHz). The shape of the focal zone was an elongated ellipsoid with an axial diameter (−3 dB) of 7.20 mm and radial diameter (−3 dB) of 1.38 mm.

Anesthesia in mice used 2.5–3% isoflurane in O₂ for induction and 1.5–2.5% isoflurane in O 2 for maintenance. Exposures were carried out as previously described [10, 11, 17]. Mice were placed in a custom-built restrainer attached to a 3-D positioning unit with their hamstrings submerged in degassed water maintained at 37 °C. The right hamstrings of mice were positioned in the center of the focal zone using the imaging transducer. For pFUS exposures, 6 raster points in a 2×3 matrix (2 mm elemental spacing) were treated. Each raster point received 100 FUS sonication pulses with the following parameters: acoustic power, 40W; pulse repetition frequency, 5 Hz; and duty cycle, 5% (10 ms ON and 190 ms OFF). The total treatment time was ~2 minutes. Mice were dried and placed in a warm recovery cage following sonications.

Mice were euthanized at various time points post-pFUS for tissue collection. For molecular analyses, muscle from the pFUS-treated and untreated contralateral hamstrings were collected. Sham control mice (transducer power = 0 W) were also euthanized for tissue collection (n=6). Harvested muscles were frozen in liquid N₂ and stored at −80 °C. For histological analyses, mice were perfused with ice-cold phosphate buffered saline (PBS) containing 4% paraformaldehyde (PFA). Hamstrings were dissected and fixed in an excess volume of PBS containing 4% PFA for 24 hr. Samples were then transferred to a solution of 30% sucrose and allowed to equilibrate for 24–48 hr before embedding in OCT and cryosectioning.

MSC culture and injections

Human MSCs from 23 year old female [18] (provided by NIH Center for Bone Marrow Stromal Cell Transplantation under an approved Intramural Research Branch protocol http://sigs.nih.gov/bmsctc) were culture-expanded in α-minimum essential medium (Life Technologies, Carlsbad, CA) with 20% lot-selected fetal bovine serum (Gemeni Bio-products, Sacramento, CA) at 37 °C under 5% CO₂ and 95% air. Early passage cells (3–5) were used for studies. For i.v. infusions, MSCs were detached with TrypLE (Life Technologies, Carlsbad, CA) and resuspended at 10⁷ cells/mL in Hank's Balanced Salt Solution (Life Technologies) containing 10 U/mL Heparin (sodium salt) (Hospira, Lake Forest, IL). MSC (1×10⁶ in 100 μL) were injected into the lateral tail vein either 45 min before pFUS, or 3, 8, or 16 hr post-pFUS. For experiments that assessed the impact of drugs on MSC homing, 1×10⁶ cells were injected approximately 4 hr post-pFUS. For all MSC homing experiments, sodium nitroprusside (Hospira, Lake Forest, IL) was intravenously injected in the lateral tail vein at a dose of 1 mg/kg in PBS 5 minutes before MSC injection. pFUS-treated and untreated contralateral hamstrings for all experiments were harvested 24 hr post-MSC injection and processed for histological analyses.

Administration of ibuprofen and etanercept

Ibuprofen (Perrigo, Allegan, MI) was administered orally at 30 mg/kg 15 min prior to pFUS. Etanercept was injected intraperitoneally on days −3 and −1 prior to pFUS at 4 mg/kg.

Proteomic analyses

Frozen hamstring tissue was mechanically homogenized with Cell Lysis Buffer (Cell Signaling Technology) containing a protease inhibitor cocktail (Santa Cruz Biotechnoloy, Santa Cruz, CA). Insoluble components in samples were separated by centrifugation at 14000 rpm for 20 minutes at 4 °C. Total protein content in samples was determined using a bicinchononic acid (BCA) assay (Thermo Scientific, Waltham, MA). Samples were analyzed by multiplex ELISA kits (Bio-Plex Pro Mouse Standard 23-Plex Group I and 9-Plex Group II) (Bio-Rad, Hercules, CA) at total protein concentrations of 2 mg/mL. Bio-Plex kits were read on a Bio-Plex 200 instrument (Bio-Rad). Stromal cell-derived Factor 1 (SDF-1α), stem cell factor (SCF), insulin-like growth factor 1 (IGF-1) and cell adhesion molecule (ICAM-1 and VCAM-1) analyses were performed using ELISA (RayBiotech, Inc., Norcross, GA) with a protein concentration of 1.25 mg/mL and read on a spectrophotometric plate reader (Spectra Max M5, Molecular Devices, Sunnyvale, CA). All assays were performed according to manufacturers' instructions.

Western Blotting

Protein samples (20 μg; n = 3 per time point) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions on Novex Bis-Tris gels (4–12% acrylamide, Invitrogen, Carlsbad, CA) and then transferred to polyvinyldiene fluoride (PVDF) or nitrocellulose membranes. Membranes were blocked using 5% bovine serum albumin (BSA) in Tris buffered saline (TBS) containing 0.05% Tween-20 (TTBS) at room temperature for 1 hr. Membranes were hybridized with a rabbit anti-mouse IgG primary antibody against HSP-70 (Abcam, Cambridge, MA) at a dilution of 1:1000 overnight at 4 °C in TTBS containing 1% BSA. Hybridization with a secondary antibody was done for 1 hr at room temperature using a horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Amersham Biosciences, Piscataway, NJ) at a 1:5000 dilution. Blots were developed with enhanced chemiluminescence reagents (Life Technologies, Carlsbad, CA) for 2 min at room temperature and exposed on autoradiograph films (Carestream Health, Rochester, NY). Loading control, β-actin, was detected using a mouse anti-mouse β-actin IgG (1:5000 dilution) primary antibody and an HRP-conjugated goat anti-mouse IgG secondary (Amersham, 1:10000 dilution).

Immunohistochemistry

Fluorescence immunohistochemistry (fIHC) was performed (n=3–6 mice per time point) to detect MSC, NFκB, and COX-2 in tissue. Tissue sections equilibrated in 30% sucrose (see above) were embedded in optimal cutting temperature (OCT) compound and sectioned to a thickness of 10 μm. Tissue sections processed through standard protocol and non-specific binding was blocked using Super Block (Thermo Scientific, Waltham, MA) for 10 min. Sections were washed in TBST and incubated with a mouse anti-human mitochondrial IgG (HuMito) (Abcam) at a dilution of 1:800, rabbit anti-NFκB IgG (Abcam) at a dilution of 1:1000, or rabbit anti-COX2 IgG at a dilution of 1:600 overnight at 4 °C. Isotype staining used mouse or rabbit IgG at identical concentrations to primary antibodies. Secondary antibody hybridization used an AlexaFluor 647-conjugated mouse anti-rabbit IgG (Life Technologies, Carlsbad, CA) at dilutions ranging from 1:1000 to 1:200 for 2 hr at room temperature. Sections were washed thoroughly with TTBS and glass coverslips were affixed with mounting medium containing ProLong Gold antifade reagent and 4',6-diamidino-2-phenylindole (DAPI) (Life Technology, Carlsbad, CA). Relative IHC intensity was measured as pixel intensity and normalized to untreated (contralateral hamstring) muscle signal using identical image acquisition settings. Muscle from MDX mice was examined using conventional IHC due to increased levels of autofluorescence. Endogenous peroxidase signal was inactivated with 1% H₂O₂. After primary incubation with the HuMito primary, a biotinylated secondary antibody (1:2000 dilution) was applied. Slides were then incubated with horseradish peroxidase (HRP)-conjugated streptavidin and signal was developed with 3,3'-diaminobenzadine. MSC were detected by IHC and counted by surveying 10 high powered fields-of-view (HPF) in both pFUS-treated and contralateral hamstrings.

Microscopy

Bright-field microscopy was performed with an Aperio ScanScope CS equipped with a 20× air objective (NA=0.75, Leica Microsystems, Buffalo Grove, IL). Fluorescence imaging was performed on an upright laser scanning confocal microscope (series 710, Carl Zeiss, Oberkochen, Germany) using Plan-Apochromat objectives (20× air, N.A. = 0.8). Illumination was provided by an argon-ion (Lasos, Jena, Germany), diode, and diode-pumped solid-state lasers (Roithner Lasertechnik, Vienna, Austria). Excitation for DAPI and AlexaFluor 647 was performed using laser lines at 405 and 633, respectively.

Statistical Analyses

All values are presented as mean±SD. Statistical analyses and data presentation were performed with Excel (version 14.2.3, Microsoft, Redmond, WA) or Prism (version 5, GraphPad Software, Inc., La Jolla, CA). Pairwise comparisons were made using Student's t-tests and multiple comparisons were made using one-way analysis of variance (ANOVA) with Bonferroni post-hoc tests to compare means.

RESULTS

Proteomic analyses of pFUS-induced molecular changes

Proteomic analyses of muscle homogenates after pFUS revealed significant changes (increases and decreases) over time in CCTF and CAM (p<0.05) compared to sham controls (Time=0) (Figure 1b,c). CCTF and CAM levels (listed in Supplementary Table 1) returned to control values by 60 hours. Significant increases in TNFα and LIF were observed 10 minutes post-pFUS and returned to control levels by 30 minutes. Following the increase in TNFα that occurred at 10 minutes, the global changes in CCTF and CAM expression between 0.5–60 hours followed a biphasic pattern: 1) initial increases of several chemoattractants from 0.5–4 hours followed by 2) delayed increases in many other chemoattractants between 8–48 hours (Figure 1c). The initial increase in TNFα post-pFUS (10 minutes) is the most likely candidate to initiate the CCTF cascade and CAM. LIF expression at 10 minutes post-pFUS may represent an anti-inflammatory response to mechanical forces in muscle [19, 20].

Figure 1. pFUS elicits a transient microenvironmental response that can be utilized on to enhance MSC homing to treated tissue.

(a) One hamstring in each mouse (n=5–6/time point) was treated with pFUS. Hamstrings from treated and contralateral control limbs were harvested at various time points for molecular analyses. (b) Time courses of CCTF quantities in muscle homogenates after pFUS (Y axes are in pg/mL; X axis are hours post-pFUS; Time 0 represents sham control). Asterisks indicate significantly elevated levels (p<0.05) identified by ANOVA and Bonferroni post-hoc tests. (c) Heat map depicting fold changes in CCTF and CAMs over time after pFUS to muscle. Protein levels measured by ELISA (shown in [b]) were normalized to sham control values (Time=0). Asterisks indicate significantly elevated levels (also shown in [b]) (d) Stack plot displaying which CCTF and CAM are significantly elevated at each time point. Note the initial increase in TNFα and LIF at 0.17 hrs (i.e.,10 minutes) after pFUS and the biphasic response of CCTF and CAM over time (see Supplementary Table 1 for abbreviations).

Significant increases in levels of many factors were observed from 0.5–4 hours post-pFUS. These included interleukins (IL) 12p40, 13, and 17, monocyte chemoattractant protein (MCP-1), monokine induced by gamma-interferon (MIG), regulated on activation, normal T-cell expressed and secreted (RANTES), monocyte colony stimulating factor (M-CSF), vascular endothelial growth factor (VEGF), stromal cell-derived factor 1 alpha (SDF-1α), insulin-like growth factor (IGF-1), and vascular cell adhesion molecule (VCAM) (Figure 1b,c). TNFα signaling likely enhanced MSC homing by increasing MCP1, RANTES, and CAM. Moreover, the known chemoattractant SDF-1α [21, 22] was elevated, a possible indirect consequence of TNFα signaling. Decreases in the pro-inflammatory factors IL1β, IL2, IL6, Interferon (IFN)-γ, and macrophage inflammatory protein (MIP)-1α were observed between 0.5 and 8 hours post-pFUS. However, IL1β, IL2, IL6, and MIP1α levels significantly increased from 16–48 hours in concert with other CCTF and CAM in the treated muscle.

The relative nadir in CCTF and CAM expression from 2–8 hours post-pFUS (Figure 1c) demonstrates a rapid decrease in primarily pro-inflammatory factors and subsequent shift towards expression of anti-inflammatory factors. Eight to 36 hours post-pFUS was characterized by significant increases (p<0.05) in IL1β, IL2, IL4, IL5, IL6, IL9, IL10, IL 13, IL 15, IL17, RANTES, VEGF, M-CSF, IFNγ, keratinocyte-derived chemokine (KC), MIP-2, platelet-derived growth factor (PDGF) and basic fibroblast growth factor (bFGF), along with intercellular adhesion molecule (ICAM) and VCAM levels. Few CCTF (MIP1α, MIP-2, M-CSF, IL 3, and IL 17) remained elevated >36 hours post-pFUS and the return of the molecular profile to baseline levels by 60 hours.

Heat shock protein-70 (HSP-70) expression did not increase through 48 hours post-pFUS (Supplementary Figure 1a), indicating minimal temperature elevations (<2–3°C) occurred in pFUS-treated muscle [23]. This suggests pFUS effects in muscle were nonthermal and therefore, primarily mechanical. pFUS did cause transient interstitial edema without micro-hemorrhage (Supplementary Figure 1b) that persisted up to 16 hours post-pFUS and was consistent with previous observations [10, 17].

MSC homing to pFUS-treated muscle

Based on the biphasic response of chemoattractants following pFUS, we investigated the optimal time of MSC infusions to maximize homing (Figure 2). Following iv injection at various time points before and after pFUS, significantly more (p<0.01) MSC were found in pFUS-treated muscle compared to untreated contralateral controls (Figure 2c). The greatest numbers of MSC per 10 HPF were observed in pFUS-treated muscle when cells were infused either prior to pFUS (72.7±24.0) or 3 hours post-pFUS (75.3±38.0). MSC injected at 8 (45.3±25.3) or 16 hours (36.7±23.6) post-pFUS and, respectively) homed to treated muscle in significantly greater numbers (p<0.05) compared to untreated contralateral control (14.0±13.52), but the magnitude of homing was also significantly less than when MSC were injected at 3 hours post-pFUS or before pFUS. While homing is diminished at later time points, EHPR of MSC still occurs when with injections performed 16 hours post-pFUS.

Figure 2. MSC home to pFUS-treated muscle when injected prior to pFUS and up to 16 hours post-pFUS.

(a) Mice (n=6 per group) were given 10⁶ human MSC intravenously either 45 min before, or 3, 8, or 16 hours post-pFUS to muscle. MSC were quantified with fluorescence immunohistochemistry (fIHC) using an anti-human-mitochondria antibody (HuMito). MSC were counted in 10 high powered fields-of-view from 3 sections per animal. (b) High-magnification images of HuMito staining in muscle from mice that did not receive human MSC (left panel), and HuMito and Isotype staining in mice that received MSC. Scale bars = 20 μm (c) fIHC and (d) quantification of MSC homing after infusing cells at various times before or after pFUS. MSC stained with an anti-human mitochondria antibody are shown in red. Scale bars = 50 μm. The greatest amount of MSC homing was observed when cells were injected either 45 minutes before or 3 hr after pFUS. However, statistically significant increases in MSC homing were also observed when MSC were injected at 8 or 16 hr after pFUS (statistical significance following ANOVA with Bonferroni post-tests is indicated by asterisks and crosses).

Drug interactions with the biological effects of pFUS

Due to similar CCTF profiles being reported after exercise exertion [24, 25], we further investigated the biological responses to pFUS that would be associated with the observed changes in CCTF post-pFUS. Ingenuity Pathway Analysis of upregulated CCTF revealed that canonical NFκB and COX2 signaling pathways were likely important for changes in muscle following pFUS (Supplementary Figure 2). Moreover, mechanical muscle stretching during strenuous exercise increases COX2 expression that can involve NFκB [26, 27]. Therefore, C3H hamstring muscles were harvested at 0.5, 4, and 8 hours post-pFUS for expression of COX2 and NFκB. fIHC demonstrated significantly increased expression of cytosolic COX2 (Figure 3a,d) [28] from 0.5–4 hours post-pFUS and significant elevations in nuclear and peri-nuclear NFκB at 4 hours post-pFUS (Figure 4a,d). These findings were further confirmed by western blotting (Figures 3f and 4f). Expression of both declined to baseline by 8 hours (p<0.05).

Figure 3. pFUS significantly increased expression of COX2 in targeted muscle and pFUS-induced expression is blocked by pretreatment with ibuprofen or etanercept.

(a,d) COX2 was significantly elevated as early as 0.5 hours post-pFUS with significant expression declined by 8 hours after pFUS. (b, e) Pretreatment with ibuprofen (30 mg/kg) or etanercept (100 μg) prior to pFUS blocked pFUS-induced COX2 expression. Scale bars = 100 μm. (c) Magnified views of COX2 fIHC showing staining with isotype (left panel) and primary antibody 4 hr post-pFUS (right panel). Scale bars = 10 μm. (f) Western blotting for COX2 4hr post-pFUS compared to untreated contralateral muscle. Statistical significance (p<0.05) is indicated by asterisks after ANOVA or t-test for pairwise comparisons; n=5–6 per time point.

Figure 4. pFUS significantly increased expression of NFκB in targeted muscle and pFUS-induced expression is blocked by pretreatment with ibuprofen or etanercept.

(a,d) NFkB was significantly elevated as early as 0.5 hours post-pFUS with significant expression declined by 8 hours after pFUS. (b, e) Pretreatment with ibuprofen (30 mg/kg) or etanercept (100 μg) prior to pFUS blocked pFUS-induced NFkB expression. Scale bars = 100 μm. (c) Magnified views of NFkB fIHC showing staining with isotype (left panel) and primary antibody 4 hr post-pFUS (right panel). Scale bars = 10 μm. (f) Western blotting for NFkB 4hr post-pFUS compared to untreated contralateral muscle. Statistical significance (p<0.05) is indicated by asterisks after ANOVA or t-test for pairwise comparisons; n=5–6 per time point.

To further explore the effects of pFUS on COX pathways, mice (n=6) were pretreated with ibuprofen, which is often used to treat symptoms following exercise [15]. Ibuprofen decreased COX2 (Figure 3b,e) and NFκB (Figure 4b,e) expression in the treated muscle to similar levels as the contralateral hamstring. Pretreatment with etanercept, which binds thereby suppressing the effects of TNFα, also resulted in decreased expression of COX2 and NFκB to baseline levels (Figures 3b,e and 4b,e). Based on these results, mice (n=6 per time point) were treated with ibuprofen or etanercept and muscles were then harvested at several time points after pFUS to characterize CCTF/CAM expression. Comparing the CCTF expression profiles following pFUS with either ibuprofen (Figure 5b and Supplementary Figure 3) or etanercept (Figure 5b and Supplementary Figure 4) to pFUS-treated mice without drugs demonstrated clear differences in response to pFUS without any compromise to the tissue (Supplementary Figure 1b). Ibuprofen or etanercept blocked the increases in TNFα and LIF, resulting in no subsequent significant increases in CCTF or CAM at any of the time point post-pFUS (p<0.05). Ibuprofen or etanercept had no effect on CCTF expression alone (Supplementary Figure 5).

Figure 5. Mechanotransductive effects of pFUS are inhibited by ibuprofen or etanercept.

(a) Mice (n=6 per time point) were pretreated with either ibuprofen (30 mg/kg orally) 30 minutes prior to pFUS or etanercept (100 mg/kg ip) at 3 days and 1 day prior to pFUS. Muscles were harvested at various time points for molecular analyses. (b) Heat maps depicting the temporal fold changes of CCTF and CAMs following pFUS to muscle when mice were pretreated with either ibuprofen or etanercept. Protein levels were measured by ELISA and normalized to sham control values (See Supplementary Figures 3 and 4 for raw data). No significant increases in expression of any CCTF or CAM were observed at any time point by ANOVAs with Bonferroni post-hoc tests.

Drugs inhibit MSC homing to healthy and dystrophic skeletal muscle

Ibuprofen or etanercept treatment prior to pFUS significantly inhibited MSC homing to healthy pFUS-treated muscle compared to pFUS-treated muscle that was not exposed to drugs (Figure 6b,c). pFUS without pretreatment with drugs did not enhance MSC homing to skeletal muscle in COX2^−/− mice compared to untreated contralateral controls (Figure 6b,c). To investigate whether ibuprofen and etanercept affected MSC directly, a trans-well migration assay was performed with SDF-1α in the bottom well with either ibuprofen or etanercept and MSC in top well. There were no significant effects on cell migration (Supplementary Figure 6) suggesting that ibuprofen or etanercept act on the host tissue rather than directly influencing MSC. Lastly, we investigated potential inhibition of MSC homing by drugs in a mouse model of MD. Dystrophic MDX mice treated with or without ibuprofen were administered pFUS followed by MSC. Significantly greater MSC homing to dystrophic pFUS-treated muscle was observed in mdx mice that did not receive ibuprofen. However, ibuprofen pretreatment entirely inhibited MSC homing to pFUS-treated dystrophic muscle (Figure 7).

Figure 6. Ibuprofen or Etanercept suppress MSC homing to normal pFUS-treated muscle.

(a) Mice (n=6 per group) were pretreated with either ibuprofen or etanercept prior to pFUS. Three hours after pFUS, mice received 10⁶ MSC intravenously. Treated and contralateral hamstrings were harvested 24 hr after cell infusions and quantified. (b) MSC homing to pFUS-treated muscle is disrupted when mice are pretreated with ibuprofen or etanercept and MSC do not home to pFUS-treated muscle in COX2-knockout mice. Since ibuprofen is a nonspecific COX inhibitor, MSC homing in COX2-knockout mice (no drug treatments) was also investigated and they were found not to home to pFUS-treated muscle. (c) Quantification of pFUS-induced MSC homing to skeletal muscle in normal mice (pFUS alone), normal mice treated with ibuprofen or etanercept, or COX2-knockout mice that received no drugs. Significant inhibition (p<0.05) of MSC homing to pFUS-treated muscle was observed compared to normal pFUS-treated mice (wild-type without drugs). There were no differences in the numbers of MSC observed in contralateral control muscle from any group. Statistical significance (p<0.05) is indicated by asterisks after ANOVA with Bonferroni post-hoc tests; scale bars = 50 μm.

Figure 7. Ibuprofen suppresses MSC homing to dystrophic pFUS-treated muscle.

(a) MDX mice (n=6 per group) were administered pFUS to the hamstring. One group was treated with Ibuprofen 15 min prior to pFUS and the other was not. Three hours after pFUS, MDX mice received 10⁶ MSC intravenously. Treated and contralateral hamstrings were harvested 24 hr after cell infusions and quantified. (b) MSC homing to dystrophic pFUS-treated muscle is disrupted when mice were pretreated with ibuprofen. (c) shows MDX muscle stained with istoype antibodies. (d) Quantification of pFUS-induced MSC homing to dystrophic muscle in MDX mice with and without ibuprofen pretreatment. Significant inhibition (p<0.05) of MSC homing to the pFUS-treated hamstring in ibuprofen-treated was observed compared to pFUS-treated mice without ibuprofen. There were no differences in the numbers of MSC observed in contralateral control muscle from both groups. Statistical significance (p<0.05) is indicated by asterisks after ANOVA with Bonferroni post-hoc tests; scale bars = 50 μm.

DISCUSSION

The major findings of this study are as follows: 1) The mechanical effects of pFUS induced acute changes (within 10 minutes) in the tissue microenvironment by increasing TNFα followed by biphasic expression of CCTF and CAM; 2) pFUS mechanical effects lead to increased COX2 expression in muscle; 3) infusion of MSC before or up to 16 hours post-pFUS significantly increased the homing of cells in muscle; and 4) Clinically relevant drugs blocked the molecular responses to pFUS and significantly decreased MSC homing to both healthy and dystrophic muscle.

The ultimate challenge in targeting stem cells is to noninvasively target them to deep structures, enabling direct interaction or paracrine effects for decreasing inflammation, stimulating repair, or differentiation of the cells into appropriate cell elements to facilitate cell-based therapy [1]. Enhancing stem cell homing to targeted organs primarily depends on molecular inflammatory changes in the host. Increased CAM on vascular surfaces interact with integrins on the MSC [29], leading to cellular transmigration into the parenchyma. Following muscle exercise in mdx/scid mice, increased expression of VCAM on endothelial surfaces enhanced homing of intra-arterially infused human CD133⁺ stem cells in the dystrophic muscle [30]. With skeletal muscle injury, the dynamic nature of expression of various cytokines and trophic factors [26, 27] (including TNFα) [22, 24], originating from various tissue components has not been well elucidated [31, 32]. The molecular and morphological changes observed in muscle with inflammation or strenuous exercise [33] can be used as a basis for extrapolating the transient biphasic (defined as an initial increase followed by a nadir and second increase) molecular effects of pFUS in muscle [34–36].

We previously proposed that pFUS-induced tissue alterations presumably occur through the process of mechanotransduction from non-thermal ultrasound effects (i.e., acoustic radiation or acoustic cavitation forces) that induce molecular changes in cells culminating in the increased expression of CCTF/CAM [10, 11, 13]. pFUS to muscle resulted in a complex interaction of multiple factors that has also been associated with skeletal muscle following strenuous exercise [24, 25, 34], injury [31], inflammation [32, 37], and degenerative diseases [32, 38, 39]. Increased interstitial edema observed with strenuous exercise [33] was also observed with pFUS [10, 17] where tissue alterations resolved over time [40]. The increase in interstitial spaces observed within the muscle following pFUS may reflect early increases in VEGF levels that generate fenestrations in endothelial barriers and cause leakage of fluids into the parenchyma [41]. pFUS effects in muscle maybe an end result of the mechanochemical changes transmitted through multiple cellular structures within the targeted areas [42, 43].

pFUS effects in muscle initiate a short-lived, primarily inflammatory, response profile followed by subsequent release of CCTF. Serial biopsies performed in subjects before and up to 24 hours after exercise indicated an early increase in TNFα followed by biphasic increase in mRNA expression of the cytokines IL6, IL8, IL10 [24]. Although increases in mRNA do not always correlate with protein expression, this observation provides insight into the CCTF changes following pFUS. The initial transient increase in TNFα (within 10 minutes) post-pFUS was likely responsible for initiating the cascade of changes observed in the muscle [26]. LIF was also elevated post-pFUS and most likely reflects a response in the muscle to an increased in TNFα [44]. Within 30 minutes post-pFUS, we observed an increased expression of factors (Figure 2a,b) that lasted for 2–4 hours followed by a nadir and a second increase in expression in CCTF and CAM. It is clear that pFUS initiates complex changes in the muscle microenvironment including the transient increased expression of SDF-1α, VEGF, and other factors thought to be important in MSC homing in tissues. Interleukin 17 (specifically IL17D) is preferentially expressed by skeletal muscle and the delayed (i.e., 4–36hrs) increased expression may indicate the direct interaction of the ultrasound energy with myocytes or other cells (i.e., endothelial cells or fibroblast) in muscle to induce observed changes in CCTF (e.g., IL6, IL1 β, G-CSF). Of note, the ELISA kit used in this study does not distinguish between subtypes of IL17. The second phase of CCTF and CAM expression occurring over 8–48 hrs represents a complex profile of pro-inflammatory and anti-inflammatory factors within the muscle.

The initial expression of predominantly pro-inflammatory factors was likely responsible for driving the second phase of CCTF originating from the muscle. However, the molecular response profile of CCTF and CAM described here may not encompass other factors that were not measured in the muscle following pFUS. The temporal changes following pFUS may depend on the viscoelastic properties of the treated tissue. These and other tissues and experimental conditions need to be further investigated in order to maximize homing of infused stem cells to targeted tissues.

Here, we show that MSC administered immediately prior to or 3 hours after pFUS resulted in maximum numbers of cells in muscle. Delaying infusion for up to 16 hours post-pFUS still resulted in significantly increased homing of MSC to muscle. These results demonstrate that pFUS primarily initiated active homing by the CCTF and CAM expression. The EHPR of pFUS-treated muscle does not require direct physical interaction between the US and infused MSC, nor do cells passively leak across severely compromised vascular barriers. Since pFUS treatment to targeted tissues does not alter the trapping of infused MSC in the vasculature of the lungs, liver, spleen, or marrow space [45–47], the transient sequestration in the capillary beds may delay the MSC movement into the circulation which coincides with the second peak increased expression of CCTF and CAM and results in EHPR of MSC in pFUS treated tissues. The local CCTF/CAM gradient within the tissue defines a transient biological address for cells to home and is termed a “molecular zip-code” for preferential MSC homing (vascular adhesion and transmigration) into the parenchyma [11]. The induced CCTF/CAM gradient following pFUS provides an added dimension to cellular therapy by potentially enabling infusions of complimentary or different cell products. It maybe possible to further enhance the delivery of multiple cell types in tissue by pFUS administering one cell type early (e.g., CD34+ cell) with MSC infused at later times. The combination of pFUS with different cell types may provide added cell delivery approached to cell therapy trials.

We have previously reported that pFUS exposures induce an EHPR effect to both normal muscles and kidneys resulting in similar findings with increased numbers of human (h) MSC in treated tissues [11, 13]. Moreover, we have demonstrated that multiple daily courses of pFUS coupled with MSC infusion results in significant amplification of the numbers of stem cells detected in targeted muscle compared to a single treatment. The expression profile of CCTF and CAM was dependent upon number of pFUS exposures performed daily in muscle indicating that the molecular changes can be modulated over time [11]. Combining pFUS with ultrasound micro-bubble (MB) contrast agents and cell infusions increased in selected trophic factors and greater numbers of stem cells compared to cells alone in models of acute myocardial infarction [48, 49] or acute kidney injury (AKI) [50] in the presence of active inflammation. We have recently reported that combining pFUS and hMSC infusion without the addition of MB in a nephrotoxin-induced AKI model resulted in improved renal function and survival compared to animals that received stem cells alone [51]. In our nephrotoxin-induced AKI model, pFUS alone did alter the expression of both pro-inflammatory and anti-inflammatory CCTF and CAM but did not alter renal function or tubule proliferation, rates of apoptosis or tubular necrosis. These results in our AKI model suggest that in the presence of active inflammation, pFUS was not detrimental to nor did it accelerate the clinical or pathological outcomes. These studies indicate that coupling pFUS with cell infusion will increase homing to targeted tissue in areas of active inflammation as well as improving functional outcomes.

It is also unknown if the EHPR effect in targeted tissues would vary between mouse (m) and hMSC due to the biological and functional differences in response to inflammation and across strains of animals [52]. We have observed no immunological complications infusing hMSC into mice and post-pFUS have reported similar numbers of hMSC homing to kidneys in immune-compromised (athymic nude mice) or immune-competent (C3H) mice [11, 13, 51]. Previous studies using pFUS with infusions of species-specific MSC or bone marrow mononuclear cells have also reported increased numbers of cells homing to targeted tissues [48, 49, 53, 54], consistent with findings in the current study. Further investigation will be needed to determine if there are significant differences in the amount of infused mMSC and hMSC homing to targeted tissue following pFUS and whether the CCTF/CAM changes in the host environment preferentially alters the immunological phenotype of the infused cells [52].

To further elucidate the mechanotransductive effects of pFUS results in CCTF and CAM changes, we examined a candidate common pathway based the current results. Mechanical stretching of muscles or myoblasts activates NFκB pathways [26, 55]. TNFα is a potent inducer of NFκB and its up-regulation can also increase COX2 expression alter prostaglandin synthesis, myogenesis, and fibrotic pathways in addition to driving expression of CCTF in muscle [56–59]. Based on the initial increase in TNFα following pFUS, it is possible that the expression of NFκB coincides with increased COX2 in muscle. To determine if the COX2 pathway was activated following pFUS, mice were treated with ibuprofen or etanercept. Ibuprofen significantly affects skeletal muscle adaptation to resistant exercise with associated decreases in TNFα, IL1b, and IL8 and prostaglandin synthase and inhibitor of nuclear factor kappa-B kinase subunit beta kinase essential for NFkB activation [27, 55, 60]. Ibuprofen also decreases inflammation and TNFα in mice with muscular dytrophy [61]. Etanercept inhibits both soluble and membrane-bound TNFα [62] to suppress signaling in muscle transplants [63]. Since TNFα induces the production of COX2, we examined the effect of drugs on the pFUS induced changes. The results clearly show that administering ibuprofen or etanercept before pFUS suppressed COX2, CCTF, and CAM expression as well as MSC homing to muscle. Furthermore, pFUS to muscle in COX2^−/− knockout mice did not result in an EHPR effect with only a few MSC were detected in treated and control muscle. These results indicate that the mechanical effects of pFUS in muscle works through pathways involving COX2 and NFκB activation leading to enhanced homing of MSC. While NFκB is one of many possible pathways activated by pFUS in muscle tissue, its effects can be centrally tied to the increased expression of many CCTF and CAM.

pFUS-induced MSC homing may also represent an in vivo platform to identify or screen drugs that would alter or interfere with active cell homing. Using pFUS in combination with MSC infusion and drugs can provide insight into possible micro-environmental changes that would augment stem cell homing as part of regenerative medicine strategies [11, 13]. Evaluating the effects of various agents on stem cell homing is usually performed in controlled model systems where drugs or factors that interfere with stem cell migration in vitro may have no effect in vivo [64]. For example, when MSC were exposed to either ibuprofen or etanercept were evaluated ex vivo (i.e., Boyden chamber), no differences in migration toward SDF-1α in vitro were detected. Further research will be needed to maximize cell homing following pFUS. A better understanding of the manner by which the ultrasound energy interacts with the tissues leading to the induced biological effects, will enable optimization of this approach for screening drug-host interactions as well as implementing cellular therapies.

Lastly, ibuprofen suppressed MSC homing to dystrophic muscle in which there was chronic ongoing inflammation. Anti-inflammatory agents (primarily corticosteroids) represent a standard-of-care for MD and are a critical constituent of MD therapy [65]. While it is unknown whether corticosteroids would also suppress MSC homing, other anti-inflammatory regimens are being widely investigated to due adverse side effects of long-term steroid use. The use of non-steroidal anti-inflammatory drugs such as ibuprofen, as well as TNFα inhibitors [66], has shown pre-clinical and clinical promise to slow MD disease progression. Moreover, pre-clinical and clinical studies have examined a variety of stem cell types for treating MD [56]. Relatively few studies have utilized MSC with most focusing on myoblasts and mesoangioblasts, that maybe susceptible to similar homing suppression in the presence of ibuprofen or other anti-inflammatory drugs. Adequate stem cell homing to dystrophic muscle remains problematic [67] and multiple direct injections to cover the affected muscle groups may be difficult to accomplish or impractical. We are presently investigating the CCTF and CAM response to pFUS in the MDX mouse, however our results indicate that pFUS significantly increased MSC homing to treated muscle compared to contralateral control. In addition, we observed a similar decrease in MSC homing to pFUS treated muscle in MDX and C3H mice that were administered ibuprofen indicating COX2 pathways and the associate molecular biological changes were stimulated in muscles with chronic inflammation. pFUS may represent a clinically relevant modality to target homing of cell therapies to dystrophic muscle, but this approach may be confounded by negative interactions with pharmacological agents. The effects from simultaneous use of pharmacological agents and stem cell therapies remain relatively unknown and currently not controlled for in clinical cell therapy trials.

Clinical translation of pFUS coupled with stem cell infusions will require consultation with regulatory authorities. Following pFUS, the targeted tissue is not damaged and is only transiently altered with changes in expression of CCTF and CAM. The stem cells used in a clinical trial would be unaltered since infusion would occur after pFUS. Therefore, only limited studies would probably be need to advance the coupling of pFUS with stem cell infusion into clinical trials. Since clinically relevant drugs completely disrupted the transient molecular zip code following pFUS, it is possible that stem cell homing maybe attenuated in clinical cell therapy trials. Therefore, drug-host effects on homing of therapeutic cells warrants substantial investigation and will likely need to be considered in the design of future clinical trials.