Molecular Imaging of the Paracrine Proangiogenic Effects of Progenitor Cell Therapy in Limb Ischemia

Abstract

Background

Stem cells are thought to enhance vascular remodeling in ischemic tissue in part through paracrine effects. Using molecular imaging, we tested the hypothesis that treatment of limb ischemia with multipotential adult progenitor cells (MAPC) promotes recovery of blood flow through the recruitment of pro-angiogenic monocytes.

Methods and Results

Hindlimb ischemia was produced in mice by iliac artery ligation and MAPC were administered intramuscularly on day 1. Optical imaging of luciferase-transfected MAPC indicated that cells survived for 1 week. Contrast-enhanced ultrasound on day 3, 7 and 21 showed a more complete recovery of blood flow and greater expansion of microvascular blood volume in MAPC-treated mice than in controls. Fluorescent microangiography demonstrated more complete distribution of flow to microvascular units in MAPC-treated mice. On ultrasound molecular imaging, expression of endothelial P-selectin and intravascular recruitment of CX₃CR-1-positive monocytes was significantly higher in MAPC-treated than control groups at day 3 and 7 after arterial ligation. Muscle immunohistology showed a >10-fold greater infiltration of monocytes in MAPC-treated than control-treated ischemic limbs at all time points. Intravital microscopy of ischemic or TNF-α-treated cremaster muscle demonstrated that MAPC migrate to peri-microvascular locations and potentiate selectin-dependent leukocyte rolling. In vitro migration of human CD14+ monocytes was 10-fold greater in response to MAPC-conditioned than basal media.

Conclusions

In limb ischemia, MAPC stimulate the recruitment of pro-angiogenic monocytes through endothelial activation and enhanced chemotaxis. These responses are sustained beyond MAPC lifespan suggesting that paracrine effects promote flow recovery by rebalancing the immune response toward a more regenerative phenotype.

The immune response plays an important regulatory role in postnatal modification of the vascular system that occurs in response to ischemia. Inflammation and vascular remodeling not only share common signaling pathways, but there is also evidence that certain inflammatory cells such as monocytes can promote arteriogenesis in ischemic tissues.^1–3 In particular, a subset of “pro-angiogenic” monocytes may serve as a source for both pro-angiogenic cytokines and growth factors, and may participate in requisite remodeling of the extracellular matrix.^1,4,5

In recent years, there has been growing interest in using stem cells and differentiated progenitor cells as a treatment option for the growing number of patients with severe ischemic coronary artery disease (CAD) and peripheral arterial disease (PAD) who are not candidates for revascularization procedures. Understanding the mechanism by which these cells promote vascular remodeling is particularly important since clinical trials have varied considerably in their results.⁶ Irrespective of whether or not they are incorporated into the native tissues, stem cells are able to act through local paracrine effects by secreting pro-inflammatory growth factors, cytokines, and chemokines.^7,8 In animal models of myocardial infarction, improvement in ventricular function with allogeneic mesenchymal stem cell therapy has been linked with increased monocyte but not T-cell infiltration.⁹ Yet little is known about the mechanisms of monocyte recruitment with stem cell therapy or its effect on vascular remodeling. In this study, we hypothesized that progenitor cell therapy enhances vascular remodeling, in part, by upregulation of monocyte recruitment in ischemic tissue and rebalancing toward a more reparative immune response. To test this hypothesis, cell therapy with xenogeneic circulating multipotential adult progenitor cells (MAPC) was performed in a murine model of limb ischemia.¹⁰ Non-invasive contrast-enhanced ultrasound perfusion and molecular imaging were used to evaluate in vivo temporal changes in blood flow, endothelial activation (P-selectin expression), and intravascular recruitment of a pro-angiogenic subset of monocytes. For the latter, targeted imaging of the fractalkine receptor (CX₃CR-1) was performed in order to specifically detect a population monocytes (Ly-6C^lo, CX₃CR-1^hi, Arginase-1-positive) that have been implicated in some studies as being immunomodulatory and pro-angiogenic in mice.^1,4 Intravital microscopy and cell migration assays were used as complementary techniques to evaluate the effect of MAPC on endothelial-leukocyte interaction and monocyte chemotaxis.

METHODS

Hindlimb Ischemia

The study was approved by the Animal Care and Use Committee at Oregon Health & Science University. All studies with the exception of optimal imaging were performed in C57Bl/6 mice (Jackson Laboratories) age 8–10 weeks (n=77). Mice were anesthetized with inhaled isoflurane (1.0–1.5%). Unilateral hindlimb ischemia was produced by ligation of the distal common iliac artery and the origin of the epigastric artery through a midline abdominal incision using sterile technique. Mice were then recovered and buprinorphine HCL (0.2 mg/kg IM) was administered for analgesia.

MAPC Administration

Human MAPC (Athersys Inc., Cleveland, OH) were isolated and expanded as described previously were stored in liquid nitrogen until the day of use.¹¹ These cells are positive for CD29, CD49c, CD90 and MHC class I; negative for HLA class II CD34, CD45, CD80 and CD106; and have been shown to produce endothelial tube formation in vitro and in vivo.^11,12 After thawing and washing, cell viability was assessed by trypan blue exclusion. One day after arterial ligation, the proximal ischemic hindlimb was injected with either 1×10⁶ MAPC suspended in 25 μL of saline (n=28) whereas control animals received either sham saline injection (n=18) or no injection (n=22). Injections were made into the deep portion of the proximal hindlimb adductor muscles using high frequency (40 MHz) ultrasound guidance (Vevo 770, VisualSonics Inc.). In order to image the spatial distribution of the cell injectate in the first 5 treated animals, 1×10⁴ non-targeted microbubbles were added to the cell suspension. Immediately after injection, ultrasound imaging was performed with both high-frequency (40 MHz) 2-D imaging and contrast-specific low-power imaging (see below). Sequential short-axis planes were acquired from the inguinal fold to the knee using 0.25 mm adjustments in the elevational plane direction. High-frequency anatomic and contrast ultrasound image sets were digitally co-registered and 3-D rendered (OsiriX v3.5) to evaluate the spatial distribution of the injection.

Microbubble Preparation

Non-targeted lipid-shelled decaflourobutane microbubbles were prepared by sonication of a gas-saturated aqueous suspension of 2 mg/mL distearoylphosphatidylcholine and 1 mg/mL of distearoylphosphatidylethanolamine-PEG(2000). For molecular imaging, biotinylated microbubbles were prepared by adding 0.4 mg/mL distearoylphosphatidylethanolamine-PEG(2000)-biotin. Biotinylated rat anti-mouse antibodies against P-Selectin (RB40.34) or CX₃CR-1 (sc30030, Santa Cruz) was conjugated to the microbubble surface via a streptavidin linkage as previously described.¹³ Microbubble concentration was measured by electrozone sensing (Multisizer III, Beckman Coulter).

Flow Cytometry

Microbubble attachment to monocytes was assessed by flow cytometry using bone marrow cells collected from the long bones of either wild-type mice or NR4A1 nuclear receptor-GFP reporter mice (NR4A1-GFP mice; kindly provided by Dr. Kristin A. Hogquist, University of Minnesota). This nuclear receptor has been shown to regulate monocyte differentiation to the CX₃CR-1^high Ly6C^low or patrolling monocyte phenotype.^14,15 Fcγ receptors were blocked (2.4G2, BD Biosciences) and cells were stained with a phycoerythrin-labeled anti-mouse CD115 mAb (AFS98, Biolegend) to identify monocytes. For wild-type animals, FITC-labeled anti-mouse Ly-6C (AL-21, BD Biosciences) was used to discriminate monocyte subsets. Cells were analyzed by flow cytometry (FACSCalibur BD Biosciences) either alone or after exposure to pressure-deflated CX₃CR-1-targeted or control microbubbles labeled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine 4-chlorobenzenesulfonate (DiD). Positive interaction was determined by fluorescent intensity beyond the 95% exclusion gate for control experiments with non-targeted microbubbles.

Perfusion Imaging

CEU perfusion imaging of the ischemic limb was performed at days 3, 7 and 21 after ligation with a linear-array transducer at 7 MHz (Sequoia 512, Siemens Medical Systems). The non-linear fundamental signal component for microbubbles was detected using multipulse phase-inversion and amplitude-modulation imaging at a mechanical index [MI] of 0.18 and a dynamic range of 55 db. Blood pool signal (I_B) was measured from the left ventricular cavity at end-diastole during an intravenous microbubble infusion rate of 1×10⁶ min⁻¹. The infusion rate was then increased to 1×10⁷ min⁻¹ and the proximal hind-limb adductor muscles were imaged in three transverse planes between the inguinal fold and the knee. Images were acquired at a frame rate of 2 Hz immediately after a brief high-power (MI 1.9) destructive pulse sequence and time-intensity data were fit to the function:

$$\text{y}=\text{A}(1-{\text{e}}^{-\beta \text{t}})$$

where y is intensity at time t, A is the plateau intensity, and the rate constant β represents the microvascular flux rate.¹⁶ Skeletal muscle microvascular blood volume (MBV) was quantified by:

$$\text{A}/(1.06\times {\text{I}}_{\text{B}}\times \text{F}\times 1.1)$$

where 1.06 is tissue density (g/cm³), F is the scaling factor (10) that corrected for the different infusion rate for measuring I_B in order to avoid dynamic range saturation, and 1.1 is a coefficient to correct for murine sternal attenuation measured a priori.¹⁷ MBF was quantified by the product of MBV and β.¹⁶ On day 21 only, CEU was performed both at rest and during electrostimulated (5 mA) contraction of the adductor muscle group at 2 Hz. Because of the limited number of times that jugular cannulation for CEU could be performed for each animal, a separate group of 6 C57Bl/6 mice underwent perfusion imaging only at baseline and day 1 after ligation to determine the extent of initial ischemia.

Molecular Imaging

Molecular imaging of P-selectin and CX₃CR-1 were performed on the same days as CEU perfusion imaging (3, 7, and 21 days after ligation). Intravenous injections of targeted microbubbles (1×10⁷) were performed in random order. Low mechanical index (MI 0.18) images were acquired 8 min after each injection and the signal from retained microbubbles alone was determined as previously described by subtracting signal from the few remaining circulating microbubbles.¹³ This was accomplished by digitally subtracting averaged frames obtained >10 seconds after destroying microbubbles within the beam volume with high mechanical index (1.2) imaging. Data were averaged for two adjacent but not-overlapping short-axis planes. All retained microbubbles within the limb were destroyed between injections by high power continuous imaging.

Fluorescent Microangiography

Functional microvascular angiography was assessed in 3 mice from each treatment group at day 7 and 21. For these studies, mice received heparin (1,000 u/Kg) followed by 200 μL of a 0.5 μM solution of FITC-labeled lycopersicon esculentum lectin (Sigma Aldrich) by intravenous route. Five minutes after injection, the descending aorta was cannulated through an abdominal approach, and perfusion fixation of the lower limbs was performed by infusion of 4% formalin at a perfusion pressure of 90–100 mm Hg. Tissues were embedded in egg albumin-gelatin matrix and 3-dimensional confocal fluorescent microscopy (FW1000, Olympus) was performed on 100 μm sections using a full z-plane depth.

Optical Imaging

Optical imaging was performed to assess MAPC survival and function post-injection in 6 immune-competent mice with spontaneous albinism (Crl:CD1) and 6 beige severe combined immunodeficiency (SCID) mice with defective NK cells (CB17.Cg-Prkdc^scidLyst^bg/Crl). For these experiments, 2×10⁶ rat MAPC stably transfected with firefly luciferase using a lentivirus vector were injected 1 day after arterial ligation. At day 1, 2, 3, and 7 after injection, animals were anesthetized with inhaled isofluorane and luciferin (150 μg/g) was injected by I.P route. Optical imaging (IVIS Spectrum, Caliper Life Sciences) was performed 15 min after luciferin injection using medium binning and data were expressed as photons/s/cm².

Immunohistochemistry

Histology was performed on perfusion-fixed paraffin-embedded sections. Sections were stained with Hematoxylin & Eosin for evaluation of inflammatory cell infiltration. Immunohistochemistry was performed for monocytes/macrophages with a rat anti-mouse Mac-2 monoclonal antibody (M3/38, eBioscience) with an ALEXA Fluor-488-labeled secondary antibody (Invitrogen). The spatial extent of positive staining was quantified by a pixel intensity threshold program (NIH Image-J) and expressed as a percent of the total muscle area per section. Histologic evaluation of the spatial relation between MAPC and monocytes was performed by Mac-2 staining in muscle injected with MAPC pre-labeled with dioctadecyl-tetramethylindocarbocyanine perchlorate (DiI). Rabbit primary antibodies were used to stain for CX₃CR-1 (ab8021, Abcam), arginase 1 (LS-B4789, LifeSpan Biosciences Inc.), macrophage F4/80 (ab74383, Abcam), CD335 (bs2417-r, Bioss), and CD31 (SP38, Novus Biologicals); with anti-rabbit secondary antibody detection with ALEXA Fluor-555-labeling (Invitrogen) or 3,3′-diaminobenzidine chromagen (Vector Labs).

Intravital Microscopy

Intravital microscopy was performed to further assess the effects of MAPC on vascular leukocyte recruitment. Mice were randomized to one of the following treatment regimens performed two days prior to intravital microscopy: (A) intrascrotal injection of 1×10⁶ MAPC alone (n=3); (B) intrascrotal injection of 0.5 mg TNF-α alone (n=6); (C) intrascrotal injection of 0.5 mg TNF-α (Sigma) followed 1 hr later by 1×10⁶ MAPC (n=7); (D) left cremasteric ischemia produced by the isolated interruption of flow to the left internal iliac artery and pudic-epigastric trunk and sham intrascrotal saline injection (n=4); or (E) cremasteric ischemia followed 1 hr later by 1×10⁶ MAPC (n=5). MAPC were fluorescently-labeled prior to injection with 5μM dioctadecyl-tetramethylindocarbocyanine perchlorate (DiI). Two days later, mice were anesthetized with an intraperitoneal injection of ketamine hydrochloride, xylazine and atropine. A cremaster muscle was exteriorized and mounted on a custom stage and intravital microscopy (Axioskop2-FS, Carl Zeiss, Inc., Thornwood, New York) was performed with a saline-immersion objective (×40/0.5 numerical aperture). Video recordings of ≥5 venules (diameter 20–40 μm) were made with a CCD camera (C2400, Hamamatsu Photonics) within 30 min of exteriorization. Centerline RBC velocity (V_b) was measured using a dual-slit photodiode (CircuSoft Instrumentation). Venular diameters (d) were measured off-line using video calipers. The distance traveled by individual rolling leukocytes was divided by the elapsed time to calculate rolling velocity. The number of rolling leukocytes (r_n) crossing a line perpendicular to the vessel over 1 min were counted and leukocyte rolling flux fraction (the percent of leukocytes passing through a venule that are rolling) was calculated by:

$$r_n/(0.25\pi d^2·V_b·60·C_L)$$

where C_L is the systemic blood leukocyte concentration. In 4 additional mice, the degree of cremasteric ischemia was determined by ligation described above, placement of a catheter in the carotid artery which was advanced into the aorta, and intra-aortic injection of 4×10⁵ fluorescently-labeled 15 μm microspheres (DYE-TRAK Persimmon, Triton Technology) over 1 min. After 10 min, both cremaster muscles were removed and weighed. The degree of ischemia was quantified by the ratio of left to right cremaster fluorescent activity normalized to weight.

Monocyte Migration

Conditioned media was collected from MAPC plated at 1 × 10⁶ cells/well grown under standard conditions for three days in the absence of serum. Human CD14+ monocytes were isolated from whole blood using a magnetic isolation kit (Dynabeads FlowComp, Invitrogen) and were confirmed to be CD14+ by flow cytometry. Either basal non-conditioned or MAPC-conditioned media (750 μl) was placed in the bottom of 24-well 3 μm pore Transwell plates (Costar) and 250 μl of non-conditioned serum-free media containing 1.25×10⁵ CD14⁺ monocytes were added to the inserts and monocyte migration to the wells was determined after 1 hr. Experiments were performed in triplicate with monocytes from different donors.

Cytokine Assay

A multi-cytokine Ab array (RayBio® Biotin Label-based Human Antibody Array 1, Ray Biotech) was used to measure human cytokines from serum-free 3-day MAPC-conditioned medium.

Statistics

Analysis was performed using SAS 9.1. Comparisons between the treatment cohorts with regards to perfusion, molecular imaging data, and histologic analysis were made with one-way ANOVA or repeated measures ANOVA for longitudinal differences and, when significant (p<0.05), post-hoc analysis with unpaired Students t-test and Bonferroni correction for multiple comparisons was performed. Intravital microscopy data were non-normally distributed and were compared by Mann-Whitney rank-sum test except for leukocyte rolling data which were compared using unpaired Student’s t-test or Kruskal-Wallis for evaluation of treatment of treatment effects. A Fisher’s exact test was used to analyze differences in proportions.

RESULTS

Microvascular Remodeling and Flow Recovery

Arterial ligation resulted in a dusky color in the ischemic limb but distal toe necrosis did not occur. Microvascular blood flow in the ischemic hindlimb measured by CEU was reduced by approximately 75% by arterial ligation (Figure 1A). Three-dimensional rendering of the spatial distribution of the MAPC cell suspension after ultrasound-guided intramuscular injection on day 1 indicated that the suspension had distributed longitudinally in the deep portions of the proximal hindlimb adductor muscle group, extending from the inguinal region to knee (Figure 1C). Injection of MAPC on day 1 resulted in a more rapid and complete recovery of microvascular blood flow in the ischemic leg compared to either sham saline injection or non-treated control mice (Figure 1A). Blood flow during contractile exercise on day 21 was also significantly greater in MAPC compared to sham-treated control animals (2.8±0.5 vs 1.6±0.4 mL/min/g, p<0.05) Parametric CEU analysis indicated that functional microvascular blood volume was greater in MAPC-treated animals than in controls at day 3, 7, and 21 after ligation; and only in the MAPC-treated group did microvascular blood volume exceed that at baseline (Figure 1B). Microvascular fluorescent angiography performed by in vivo lectin staining illustrated a greater number of capillary units functionally perfused and an increase in the number of small to medium transverse or bridging arterioles in MAPC-treated compared to control groups at day 7 and 21 (Figure 1D). At day 21, muscle capillary density and capillary-to-myocyte ratio were slightly increased in the ischemic tissue from all groups compared to non-ischemic limbs (Supplemental Figure 1); however, there were no significant differences between MAPC-treated and untreated ischemic muscle.

Figure 1.

Microvascular perfusion imaging and fluorescent microangiography. (A) Mean (±SEM) microvascular blood flow in the hindllimb adductor muscles at baseline (BL) and after arterial ligation. *p<0.05 vs both control groups (B) Mean (±SEM) microvascular blood volume in the hindlimb adductor muscles. *p<0.05 vs both control groups. (C) Three-dimensional ultrasound composite image illustrating the distribution of MAPC injectate from the contrast-enhanced color-coded regions in the individual 1 mm elevational planes denoted 1 through 4. (D) Examples of fluorescent microangiograms from a PBS-treated control and MAPC-treated limb 7 days after ligation using similar optical settings. The number of subjects for each data point is listed in the Supplement Table.

MAPC Survival in the Ischemic Limb

Optical imaging of luciferase activity after injection of luciferase-transfected MAPC indicated that cells remained primarily within the injected limb (Figure 2). Activity declined progressively by two orders of magnitude between day 1 and day 7 suggesting either loss of cells or loss of protein production. The pattern of decay was not substantially different for immune-competent and NK-deficient SCID mice, suggesting clearance of MAPC occurs similarly in the presence or absence of immune rejection. At day 7, luciferase activity was not significantly different from background levels in the majority of mice in both groups.

Figure 2.

Mean (±SEM) photon flux from luciferase activity (log scale) after injection of MAPC transfected with firefly luciferase in (A) immune competent, and (B) NK-deficient SCID mice (n=6 for each condition). The images illustrate examples of optical imaging of MAPC luciferase activity localized at the site of injection in the proximal ischemic limb over time.

Molecular Imaging of Endothelial Activation and Monocyte CX₃CR-1

CEU molecular imaging of P-selectin was used to assess endothelial response (Figure 3A and 3B). P-selectin-targeted signal in the ischemic limb was significantly greater in MAPC-treated compared with both control groups at 3 and 7 days after arterial ligation. In MAPC-treated animals, signal was not confined to the initial distribution of cell injection and instead was uniformly distributed throughout the hindlimb adductor muscle group. P-selectin signal subsequently decreased by day 21 for all groups. There was a trend toward higher P-selectin signal in sham saline-injected versus non-treated control limbs at day 3 which did not reach statistical significance after correction for multiple comparisons. Monocyte CX₃CR-1 signal on CEU molecular imaging in the ischemic limb was significantly higher for MAPC-treated compared to control groups at all time intervals (Figure 3C) and was also distributed uniformly throughout the muscle. There was no significant decline in CX₃CR-1 signal between day 3 and 21. Preferential attachment of the CX₃CR-1-targeted microbubbles to the intended monocyte population was verified by flow cytometry (Supplemental Figures 2–4). For wild-type CD115-positive monocytes, there was greater attachment of DI-D-labeled CX₃CR-1-targeted microbubbles to LyC6^low than LyC6^high cells indicated by almost 2-orders of magnitude greater fluorescent intensity, and a higher percent of cells interacting with labeled microbubbles (70% vs 15%). Similarly, for CD115-positive monocytes from transgenic mice co-expressing GFP with the NR4A1 nuclear receptor that promotes the CX₃CR-1^high Ly6C^low phenotype,^14,15 fluorescent intensity from interaction with CX₃CR-1-targeted microbubbles was approximately 6-fold greater for NR4A1^high than NR4A1^low cells. There was little attachment to monocytes for control non-targeted microbubbles.

Figure 3.

Molecular imaging for P-selectin and CX₃CR-1. (A) Mean (±SEM) signal intensity from CEU molecular imaging with P-selectin-targeted microbubbles in control non-injected, sham PBS-injected, and MAPC-treated limbs. (B) Examples of P-selectin targeted imaging in the transvers-axis plane at day 7 illustrate the spatial distribution of signal in the proximal adductor muscle group. The B-mode 2-D image for the MAPC-treated mouse is provided at the top for anatomic reference (femoral acoustic shadowing at the right of the image). (C) Mean (±SEM) signal intensity from CEU molecular imaging with CX₃CR-1-targeted microbubbles. (D) Microscopy showing attachment of CX₃CR-1-targeted microbubbles (dark spheres) to murine homotypic monocyte aggregates (top); and lack of microbubble (fluorescently labeled green with DiO) attachment to a Ly-6c^hi (phycoerythrin-positive) monocyte. Scale bar = 10 μm. *p<0.05 vs. control and PBS; †p<0.05 vs. PBS. Number of animals for each group is provided in the Supplement Table.

Leukocyte Rolling on Intravital Microscopy

To corroborate the P-selectin expression patterns seen on molecular imaging, intravital microscopy was performed two days after intrascrotal injection of MAPC to assess selectin-dependent leukocyte rolling. When given in the absence of TNF-α pretreatment or ischemia, MAPC remained attached to the outer surface of the cremaster muscle and did not migrate intramuscularly. With either TNF-α pretreatment or cremasteric ischemia, MAPC migrated into the muscle and the majority assumed a perivascular arrangement around almost all venules and many arterioles (Figure 4A and Supplement Videos). Venular leukocyte rolling flux fraction was higher and leukocyte rolling velocity was slower in animals treated with TNF-α and MAPC two days prior compared to sham-treated control animals receiving TNF-α alone (Figure 4B and 4C). Extravasated leukocytes were frequently observed in the proximity of perivascular MAPC (Figure 4D). There was no major difference in leukocyte rolling in venules with high versus low density of MAPC. For ischemia studies, flow in the ischemic left cremaster muscle was 40±18% of that in the contraleratal control cremaster. In the ischemic cremaster muscles, venular leukocyte rolling velocity was slower in muscles treated with MAPC than for sham-injected controls (Table 1). Although leukocyte flux fraction was not significantly different between the groups, MAPC treatment resulted in greater flow velocity, greater flow rate, and hence a higher total number of rolling leukocytes. Together, intravital microscopy data suggest that in the presence of ischemia or a pro-inflammatory milieu, MAPC are able to migrate to a perivascular location where they promote leukocyte recruitment and probably migration.

Figure 4.

Intravital microscopy data. (A) Illustration of a venule (delineated by dashed line) under transillumination (left) and fluorescent epi-illumination (right) illustrating perivascular localization of Di-I-labeled MAPC two days after intrascrotal injection of TNF-α and MAPC. (B) Mean (±SEM) leukocyte rolling flux fraction in cremasteric venules from animals treated with TNF-α alone (control, n=6) and TNF-α and MAPC (n=7) two days earlier. *p<0.05 vs control. (C) Histogram and median leukocyte rolling velocities in venules for control and MAPC-treated animals. Mann-Whitney p<0.05. (D) Image illustrating a perivenular Di-I-labeled MAPC (arrowhead) with extravasated or intravascular adhered leukocytes (arrows) in proximity. Scale bar = 20 μm.

TABLE 1.

Intravital Microscopy Data for Ischemic Cremaster Muscle Venules^*

	PBS Control (n=12)	MAPC (n=9)	P
Vessel diameter (μm)	30±9	34±10	0.40
Venular blood velocity (mm/s)	1.0±0.5	2.1±1.0	0.002
Venular shear rate (s−1)	585±437	1,196±797	0.09
Leukocyte rolling velocity (mm/s)†	38±4	43±6	0.04
Leukocyte rolling velocity/shear rate (mm)†	0.060±0.049	0.029±0.019	0.001
Leukocyte rolling flux fraction (×10−2)	7.2±6.6	5.0±5.3	0.45

PBS, phosphate buffered saline; LVEF, left ventricular ejection fraction;

^*Data are presented as Mean±SD.

^†Greater than 100 observations per group were made for rolling velocity data.

Monocyte Migration and Cytokine Production

Cell migration assays performed to further examine whether products from MAPC promote monocyte migration. Migration of human monocytes was much greater in response to MAPC-conditioned media compared to non-conditioned media (151±55 vs 6±3 cells per optical field, p<0.01). The cytokine array from conditioned media indicated the production of several key chemokines involved either directly or indirectly in monoycte chemotaxis and selectin expression including MCP-1, MIP-1α, IL-1α, VEGF-A, and M-CSF (Supplemental Figure 5), some of which were increased markedly by exposure of cells to cytokines known to be elevated in ischemic tissue.

Histology

On H&E staining of muscle from the ischemic hindlimb at day 3 and 7, there was a immune cellular infiltrate seen in all three treatment groups which was greater for the MAPC-treated group compared to control groups (Supplemental Figure 6). On quantitative analysis of monocyte Mac-2 staining, there were pronounced differences between groups. For all three post-ligation study points, the area staining positive for Mac-2-positive cells was greater in MAPC-treated compared with control groups (Figure 5A and 5B). For all groups, monocyte infiltration peaked at day 7. MAPC produced a more diffuse rather than focal monocyte infiltration indicated by the greater number of sections with Mac-2-positive cells in MAPC-treated compared to control muscle (Figure 5C). On immunohistochemistry at the early time intervals, there were more Arg1-positive monocytes in the MAPC-treated than in control ischemic muscle (Figure 5D and 5E). The majority of monocytes stained positive for CX₃CR-1 in all treatment groups. Although this did not allow us to confidently differentiate between CX₃CR-1^hi and CX₃CR-1^lo populations, MAPC-treated limbs were characterized by a greater number of cells in the muscle interstitium that stained strongly positive for CX₃CR-1 (Figure 5F). Dual fluorescent staining demonstrated that most MAPCs that could be identified at day 3 had CX₃CR-1-positive monocytes in proximity (Figure 5G).

Figure 5.

Monocyte immunohistochemistry. (A) Quantitative results of the area staining positive for monocyte Mac-2 from day 3, 7, and 21 (note different y-axis scales) (n=3 for each condition with >20 sections per subject analyzed). (B) Examples of Mac-2 staining at day 7 from MAPC-treated and sham PBS-treated ischemic muscle. (C) Percent of sections demonstrating any Mac-2-positive cells (≥80 fields for each condition). Histology from muscle at day 3 illustrating separate examples of arginase-positive (Arg1) cells (green) in MAPC-treated muscle (D) which were largely absent in untreated muscle (E). (F) Example of positive CX₃CR-1 staining of monocytes from a MAPC-treated limb on day 3. (G) Example illustrating colocalization of Mac-2-positive monocytes with a Di-labeled MAPC. *p<0.05 vs. both control groups.

DISCUSSION

Stem cells and adult progenitor cells have been shown to promote beneficial vascular remodeling in ischemic tissues and in wound healing. One mechanism by which cell therapy potentiates angiogenesis is through their paracrine effects that involve the release of biomolecules that alter the local cellular and molecular environment.^7,8 In a model of limb ischemia that was intended to reproduce a clinically-relevant situation of severe reduction in resting flow, we used a panel of techniques including in vivo CEU perfusion and molecular imaging to temporally evaluate how MAPC therapy augments recovery of perfusion and influences the recruitment of monocytes thought to promote arteriogenesis in ischemic tissues. Our results suggest that this population of adult progenitor cells promote monocyte entry into the skeletal muscle in part via a two-step process that involves increased endothelial recruitment through P-selectin expression and enhanced monocyte chemotaxis.

Stem and progenitor cells have been shown to have somewhat varied immunomodulatory effects depending on the leukocyte cell type and local tissue environment.¹⁸ Because MAPC produce chemotactic and endothelial activating factors, it is not unexpected that under certain conditions MAPC have been shown to stimulate a monocyte/macrophage response.^10,19 Noteworthy findings of our study were the degree to which MAPC augmented monocyte infiltration, and how MAPC therapy dramatically altered the spatial characterization of the monocyte infiltrate, producing a much more uniform and diffuse distribution of monocytes than in untreated limbs. Monocytes that stained positively for Arg1 were preferentially recruited without an obvious increase in granulocyte infiltration on H&E staining, suggesting that cell therapy may rebalance the cellular immune response toward a more regenerative phenotype. A similar rebalancing of the cellular immune response has been described after injecting conditioned media from bone marrow-derived mesenchymal stem cells (BM-MSCs) in a model of wound healing.²⁰

One aim of this study was to define the mechanism of enhanced pro-angiogenic monocyte recruitment with MAPC therapy. For evaluating endothelial activation, we studied P-selectin due to its important role in initiating the immune response through leukocyte capture and rolling. We used a P-selectin-targeted contrast agent that has been validated in multiple different models of ischemic disease.^13,17,21 An advantage of using CEU in this particular study is that targeted microbubble signal reflects only luminal surface expression of P-selectin rather than that which is stored pre-formed within the Weibel-Palade bodies. We have previously shown that the P-selectin signal in this model of limb ischemia is primarily from endothelial activation and not from platelets.¹⁷ In MAPC-treated limbs, molecular imaging detected a significant increase in P-selectin expression at days 3 and 7. These findings were further substantiated by intravital microscopy where in venules of cremaster muscle that was either treated with TNF-α or was rendered ischemic, MAPC resulted in slower leukocyte rolling velocity and increased number of rolling leukocytes indicative of increased selectin expression.²² Intravital microscopy also provided convincing evidence that MAPCs promote the recruitment of certain immune cells by their ability to migrate to a perivascular position. Their migratory capacity may also explain why molecular imaging signal enhancement in MAPC-treated limbs was diffuse rather than localized to the region of cell distribution identified by imaging at the time of administration. An interesting and presently unexplained finding was the greater microvascular flow velocity in MAPC-treated ischemic muscles only several days after injection. This finding was corroborated by day 3 CEU perfusion data in ischemic limbs. These data imply a difference in the production of vasoactive compounds that may precede and promote structural vascular remodeling.

For temporal characterization of monocyte recruitment, we targeted microbubbles to CX₃CR-1 in an attempt to image a subpopulation of monocytes thought to be pro-angiogenic and reparative.^1,4,23 Flow cytometry demonstrated that these microbubbles preferentially attached to the intended population by binding with greater frequency to either Ly6C^lo monocytes or those with high NR4A1 expression. Molecular imaging signal for CX₃CR-1 was increased by MAPC therapy and was sustained for at least 21 days. From immunohistochemistry, CX₃CR-1 appeared to be almost exclusively from infiltrating mononuclear cells, the vast majority of which were monocytes. Expression of Arg1 by monocytes in the MAPC-treatment group further supports the idea that cell therapy enhanced the recruitment of Ly6C^lo CX₃CR-1^hi monocytes.

The use of targeted microbubbles for molecular imaging ensured that signal originated from cells that were being actively recruited in the vascular space rather than reflecting cumulative extravasation. In part this could explain why molecular imaging signal for monocytes underestimated the differences between treatment groups seen on histology. However, we believe that the greater treatment-related differences in monocyte recruitment on histology than molecular imaging could also be explained by the two step process of MAPC to increase endothelial cell adhesion molecule expression and to enhance monocyte chemotaxis. The notion that MAPC enhance monocyte chemotaxis was supported by data from migration assays, intravital microscopy, histology, and MAPC cytokine and chemokine array. Although CD4+ and CD8+ T-cells and natural killer cells could also have contributed to the CX₃CR-1 signal,²⁴ the degree of lymphocyte infiltration on histology was very small, suggesting that MAPC induce the specific recruitment of CX₃CR-1-positive monocytes.

Optical imaging of luciferase activity suggested that MAPCs did not persist beyond a week after injection. While previous studies suggest that NK cells could be responsible for early cell clearance,¹⁰ our optical imaging data and monocyte immunohistochemistry from SCID beige mice and immunohistologic staining for NK cells in wild-type mice (Supplemental Figure 7) indicate that most cell loss was not from immunologic “rejection” An intriguing observation was that certain indicators of immune activation, such as enhanced CX₃CR-1 signal on molecular imaging and monocyte recruitment on histology, persisted late (day 21) despite the evidence on optical imaging that few MAPC survived at one week. However, it is known that the monocyte inflammatory response in ischemic limb tissue is self-amplifying.²⁵ Many monocyte/macrophage cell products are capable of perpetuating endothelial activation and monocyte recruitment. Accordingly, we believe that a sustained monocyte response can be achieved with single administration of a chemokine stimulus such as intramuscular injection of MAPC. Although we did not specifically test which mediators were directly responsible for initiating these events, we did evaluate the MAPC secretome which indicated the presence of several key cytokines, CC-chemokines, and growth factors (MCP-1, MIP-1α, VEGF, M-CSF, and IL-1α) that have all been shown to directly or indirectly increase P-selectin expression and/or monocyte chemotaxis.^26–30

It should be noted that results from microvascular perfusion in this study alone are important. Muscle CEU perfusion imaging has been used to evaluate pro-angiogenic therapies in animal models of disease.³¹ As we have demonstrated, it provides information on perfusion at rest and during stress, and can measure the expansion of the microvascular blood volume which. CEU blood volume data in conjuction with fluorescent microangiography and capillary histology suggested that the MAPC treatment increased flow primarily through remodeling of the distal arteriolar network which enhanced distribution of flow to parallel microvascular units.

There are several limitations of the study that deserve mention. The lack of luciferase activity beyond day 7 on optical imaging may have resulted from altered protein production rather than cell loss, although histology supported the latter. Our data does not provide definitive information on how much of the vascular response to MAPC was attributable to monocyte recruitment. It should also be noted that there is some controversy regarding the identification of proangiogenic monocytes, with some reports suggesting that the “inflammatory subset” of monocytes that are CCR2^high may also contribute to angiogenesis in response to limb ischemia in mice.²⁵ The differences in flow and shear conditions could have influenced leukocyte rolling data on intravital microscopy experiments in ischemic muscle. However, data from TNF-α-treated muscle were generated under identical shear conditions and support the idea of enhanced rolling with MAPC therapy. Arguing against P-selectin as a mechanism for selective recruitment for “pro-angiogenic” monocytes, it has been shown that Ly-6C^low CX₃CR-1^high monocytes have slightly less cell surface expression of P-selectin glycoprotein ligand-1 when compared to their more inflammatory counterparts.³² However, we believe that P-selectin is important for the initial microvascular capture for all monocytes and selective recruitment of one subtype is likely to involve specific chemotactic signals.

In summary, we have demonstrated that adult progenitor cell therapy with MAPC potentiates the pro-angiogenic monoycte response in limb ischemia which may, in part, explain their ability to promote flow recovery. These data will likely provide justification for future studies to evaluate the molecular mediators that are responsible for monocyte recruitment and to develop models that can test how much of the vascular response is attributable to monocytes.

Clinical Perspective.

The number of patients with severe symptoms from coronary and peripheral artery disease who are not candidates for revascularization therapy due to age, comorbidities and/or diffuse distribution of disease is steadily growing. Pro-angiogenic cell therapy is a promising therapeutic alternative. However, the exact mechanisms by which stem cells promote vascular remodeling are unknown and are important for optimization of therapy in terms of cell type, dose and method of administration. In this study we used contrast ultrasound perfusion to demonstrate that multipotential adult progenitor cells (MAPC) given intramuscularly increased microvascular blood flow and volume in a murine ischemic limb model. On a structural level, the improvement in flow was due to arteriolar remodeling. Using both molecular imaging, histology, and intravital microscopy we showed that in the setting of ischemia MAPC migrate to a perivascular location where they recruit a specific population of monocytic cells that are recognized to be pro-angiogenic and important in wound healing. This process was accomplished by stimulation of endothelial cell adhesion molecules that participate in leukocyte recruitment and also by chemokine signaling of monocytes to migrate into tissue. These results add to the growing body of science that indicates that stem cells promote flow recovery more from their paracrine effects on host cells than from their engraftment into new blood vessels; and that a specific pro-angiogenic aspect of the inflammatory response is a mediator of this process.