Overexpression of Nitric Oxide Synthase Restores Circulating Angiogenic Cell Function in Patients With Coronary Artery Disease: Implications for Autologous Cell Therapy for Myocardial Infarction

CACs were isolated from 10 participants in each of the following 4 categories: healthy women aged <45 years, healthy men aged <45 years, healthy men and women aged ≥45 years, and men and women aged ≥45 years with documented CAD. Inclusion criteria for healthy participants included being 18 years or older and the absence of hypertension (blood pressure >140/90 mm Hg), dyslipidemia (low‐density lipoprotein >160 mg/dL), diabetes mellitus (fasting plasma glucose >126 mg/dL), cigarette smoking, evidence of coronary or peripheral artery disease, malignancies, terminal renal failure, acute inflammation, pregnancy, medication with statins, estrogen replacement, hormonal birth control, or erectile dysfunction medication. The young, healthy female participants underwent flow‐mediated dilation (FMD) measurement and blood draws within 3 days of the start of the menstrual cycle to minimize hormonal variability. The CAD group consisted of 10 patients recruited through the University of California, San Francisco (UCSF) Cardiology Clinic with angiographically documented CAD as defined by >70% stenosis of ≥1 coronary artery, receiving optimal medical therapy according to current secondary prevention guidelines.⁴¹ Informed consent was obtained in accordance with the UCSF Institutional Review Board and the Declaration of Helsinki. Subject characteristics of all participants with complete data are described in Table 1.

Endothelial function determined by FMD of the brachial artery was measured by using ultrasonography (Sonosite Micromax). Subjects rested in the supine posture in a room at 20° to 22°C for at least 10 minutes, fasted for ≥12 hours before the measurement, and refrained from caffeine and alcohol intake and exercise for ≥8 hours before the measurement. A cuff was placed distal to the probe and was inflated 250 mm Hg for 5 minutes. Diameter was measured before cuff inflation and immediately after cuff deflation, at 20, 40, 60, and 80 seconds. FMD was calculated as Percent change=(Peak diameter_postischemia−Diameter_baseline)/Diameter_baseline×100. All measurements were performed by the same examiner. Videos were analyzed offline by an automated analysis system (Brachial Analyzer, Medical Imaging Applications) as described earlier.¹⁷ Measurement of endothelium‐independent vasodilation, sometimes assessed in FMD studies by the administration of nitroglycerin, was not approved by the institutional review board and therefore was not attempted.

Blood samples (13 mL) were collected from the cubital vein on the morning of the ultrasound examination after a 12‐hours overnight fast to measure the serum lipid profile and other biochemical parameters. In a second visit that did not involve fasting, 250 mL of blood was collected for CAC isolation into Vacutainer CPT tubes (Becton Dickinson) with heparin. CACs were differentiated ex vivo from peripheral blood mononuclear cells by 7 days of culture as previously described.^{2, 7} Peripheral blood mononuclear cells were isolated through density centrifugation in Accuspin^™ System‐Histopaque‐1077 (Sigma‐Aldrich) and preplated on fibronectin‐coated dishes in endothelial cell basal medium (EBM‐2; Cambrex) supplemented with EBM‐2MV SingleQuot and 20% fetal bovine serum for 3 hours to remove platelets and shed endothelial cells, which were discarded. The initially nonadherent cells were harvested and frozen at a concentration of 1×10⁷ cells/mL in EBM‐2 with EBM‐2MV SingleQuot, 20% fetal bovine serum, and 10% DMSO for later use. Before subsequent experiments, the peripheral blood mononuclear cells were thawed and cultured on fibronectin‐coated dishes for 7 days as before.

Because of the large number of CAC donors and assays, it was not feasible to count postculture CACs that were positive for standard markers. However, in 3 of our previous studies,^{16, 17, 18} we isolated human CACs from small groups of donors and reported that after 7 days of culture, >95% of them were double‐positive for Ulex europeus agglutinin lectin binding and acetylated low‐density lipoprotein uptake. In 2 of the studies,^{16, 18} we reported that the cell populations expressed CD45 (85–93%), KDR (24–49%), CD31 (49–63%), CXCR4 (57–59%), CD14 (68–70%), and CD11b (54–55%); with extremely low incidence (≈1%) of postculture cells expressing CD34 or CD133. CACs isolated from healthy young volunteers and older CAD patients in our earlier pilot study¹⁶ did not differ significantly in expression of CD45 or CD31 postculture.

Migration of CACs was quantified with a transwell chemotaxis assay by using a modified Boyden chamber. Per our previously published protocol,¹⁸ 600 μL of EBM‐2 media with or without 50 ng/mL VEGF (Sigma) was added to the bottom of a 24‐well transwell chamber plate (Corning). Then, 2×10⁴ CACs (post 7‐day culture) were resuspended in 100 μL EBM‐2 supplemented with 0.5% bovine serum albumin, added to each migration insert (8‐μm pores; Corning), and placed in the companion 24‐well tissue culture plate. Each sample was loaded in triplicate inserts. Cell migration occurred during a 6‐hours incubation at 37°C. Cells attached to the underside of the insert membrane were fixed in 4% formaldehyde, and cells attached to the topside of the membrane were removed with a cotton swab. The membrane was removed, mounted on a glass slide, and stained by using Hoechst 33342. Fluorescence microscopy was used to capture 5 random fields (×10 objective) per membrane, and results were expressed as the average of the number of cells visualized per field.

CAC adhesion to endothelial cells and association with endothelial tubules were performed as previously described.⁹ For adhesion, 2×10⁵ human umbilical vein endothelial cells (HUVECs) at passages 3 to 5 were seeded per well in 4‐well glass slides. At 36 hours later, some wells were pretreated with 1 ng/mL tumor necrosis factor‐α (BD Biosciences) for 12 hours; other wells were not pretreated. Post–7‐day CACs in suspension were incubated with Vybrant^® DiI Cell‐Labeling Solution (2.5 μg/mL in PBS; Life Technologies) for 5 minutes at 37°C followed by 15 minutes at 4°C and then were added to the wells containing HUVEC monolayers and incubated for 3 hours. After the nonattached cells were washed out with PBS, the HUVECs and adhered CACs were fixed with 4% paraformaldehyde and counted in 10 random fields.

For the HUVEC tube‐forming assay, 4‐well glass slides coated with Matrigel (BD Bioscience) were put in the incubator at 37°C for 30 minutes to allow solidification. DiI‐labeled CACs (2×10⁴) were coplated with 4×10⁴ HUVECs and incubated at 37°C for 6 hours to allow the HUVECs to form tubes. The percentage of CACs associated with tubes⁹ was determined in 10 random fields by counting the number of stained cells colocalizing with tubes and dividing by the total number of stained cells, and then multiplying by 100.

Total RNA from cultured CACs was isolated by using an RNeasy Mini Kit (Qiagen) or RNAqueous‐Micro Kit (Ambion). cDNA was transcribed from RNA with iScript cDNA Synthesis Kit (Bio‐Rad). Real‐time polymerase chain reaction (PCR) was then performed on a sequence‐detection system (Prism 7900; Applied Biosystems) and the TaqMan PCR Core Reagent kit (Applied Biosystems). β‐Actin was used as the endogenous control to normalize the amount of cDNA added to each reaction for analysis of eNOS expression; primers were purchased from Applied Biosystems. Human β‐glucuronidase was the endogenous control for CD14, KDR, CD31, and CD45 expression; primers were purchased from Integrated DNA Technologies. PCR was performed at the Genome Analysis Core Facility, Helen Diller Family Comprehensive Cancer Center, UCSF.

eNOS level in undiluted cell lysate was quantified by the use of ELISA (R&D Systems) in samples prepared from 1×10⁶ cultured CACs in 1 mL of lysis buffer from the manufacturer. Each sample was assayed in duplicate. To normalize the eNOS protein levels, total protein concentrations were measured by using Coomassie Plus–The Better Bradford^™ Assay Kit (Pierce). eNOS in Western blots was detected by using eNOS/NOS Type III antibody (1:1000, BD Transduction Laboratories) followed by corresponding IRDye 800CW goat anti‐mouse second antibody (1:5000; LI‐COR Biosciences) and visualized and quantified by using an Odyssey® infrared imaging system (LI‐COR).

Pooled human umbilical vein endothelial cells (HUVECs) were purchased from Cambrex, cultured in EBM‐2 (supplemented with Singlequots 5% FBS) and used no later than passage 5 as positive controls for eNOS immunoblots.

Thawed peripheral blood mononuclear cells were seeded in a 6‐well plate and, at day 6, washed with phosphate‐buffered saline; the medium was replaced with 1 mL fresh culture medium and incubated overnight. At day 7, medium was collected and frozen at −80°C for measurement of liberated NO. NO levels were evaluated by measuring the intermediate and end products, nitrate and nitrite (NO_x). NO_x was reduced by using vanadium (III) and hydrochloric acid at 90°C. NO was purged from the solution, resulting in a peak of NO for subsequent detection by chemiluminescence (NOA 280; Sievers Instruments). The emitted light signal was recorded by dedicated software as the NO content (micromoles per liter).

Stimulation of eNOS was achieved by treatment of CACs with 10 μmol/L ionomycin for 2 hours (Sigma‐Aldrich).

Adenoviruses Ad.CMV‐eNOS and Ad.CMV‐GFP (green fluorescent protein) (S.M. Black, PhD, unpublished) and Ad‐eNOS‐siRNA (gift from David Fulton⁴²) were used to transduce CACs. CACs were infected at day 5 of culture with Ad.CMV‐GFP, Ad.CMV‐eNOS, or Ad‐eNOS‐siRNA at a multiplicity of infection of 100 or 500. After overnight incubation, the cells were washed twice with phosphate‐buffered saline, fresh culture medium was added, and cells were used for the relevant experiments at day 7 of culture.

All animal procedures were approved by the UCSF Institutional Animal Care and Use Committee and performed in accordance with the recommendations of the American Association Accreditation of Laboratory Animal Care. SCID mice (C.B‐17, male, 10 weeks) were obtained from Taconic. Group size ranged from 6 to 10 depending on the number of individual donor CACs available; vehicle control group was n=5.

MI was surgically induced through permanent coronary artery ligation as we have described previously.⁴³ Briefly, mice were anesthetized with 2% isoflurane during surgery and received subcutaneous buprenorphine (0.1 mg/kg) for analgesia at the time of surgery and at the end of the day. The heart was exposed via parasternotomy, and the left anterior descending coronary artery was permanently ligated ≈3 mm below the tip of the left atrium.

End‐diastolic and end‐systolic volumes and left ventricular ejection fraction (LVEF) were assessed by using echocardiography with 1.25% isoflurane (Vevo660 micro‐ultrasound system; VisualSonics Inc) as we have previously described.^{27, 44} For CAC injections into myocardium, 10⁵ viable CACs in Hanks balanced salt solution (HBSS) were split into 2 injections of 5 μL and injected into myocardium at the infarct border zone with use of a 30‐gauge needle under ultrasound visualization with 2% isoflurane at 3 days post‐MI, according to our previously published procedure.⁴³ Echocardiography was performed and interpreted by a blinded investigator. Infarct size was calculated through histological measurement of the fibrotic zone according to our published protocol.⁴⁵

Mice were killed at day 28 post‐MI/day 25 postimplantation, and hearts were frozen in freezing isopentane and cryosectioned at 10 μm. Sections were fixed in 1.5% formaldehyde for 15 minutes and then blocked with normal goat serum staining buffer for 30 minutes without detergent (2% normal goat serum and 0.02% sodium azide in phosphate‐buffered saline). For quantitation of blood vessels and CACs, sections were further blocked with use of the Avidin and Biotin Blocking System (DAKO) each for 10 minutes. Slides were then incubated in primary antibody for 1 hour, consisting of rat anti–human leukocyte antigen (HLA) monoclonal antibody (Bio‐Rad; 1:400) and mouse anti–smooth muscle actin monoclonal antibody (clone 1A4; ICN MP Biomedicals; 1:400), as well as biotin conjugated BS‐1 isolectin B4 (Sigma‐Aldrich, 1:100). Negative controls lacking primary antibody/lectin were always performed. Sections were rinsed in staining buffer and then incubated for 1 hour at room temperature with Alexa 488 goat anti‐rat antibody (1:250), Alexa 546 goat anti‐mouse antibody (1:200), or Alexa 660 streptavidin (1:100; secondary antibodies from BD Pharmingen). Alternatively, for initial detection of CACs 3 days after implantation, CACs had been stained with cytoplasmic Cell Tracker Green (Life Technologies) according to the manufacturer's instructions before implantation. These sections were stained as described here earlier only with HLA antibody. The slides in all cases were then rinsed, mounted, and viewed with a Nikon E800 fluorescence microscope by using Openlab software (Improvision).

For quantitation of capillary density represented by total endothelial stain, photomicrographs were taken with a ×20 objective, and endothelium‐stained regions were automatically selected by using ImagePro software (Media Cybernetics), normalized for regions of visual fields lacking tissue and expressed as percentage of the field that stained positive for endothelium. For quantitation of arterioles and venules, the number of structures in the same photomicrographs staining positive for smooth muscle actin were counted, normalized for regions of visual fields lacking tissue, and expressed as number of objects per field.

For in vitro/ex vivo experiments, data are presented as mean±SEM. Planned comparisons central to the research design were tested by using a family‐wise α of 0.05 as specific contrasts within an ANOVA test for each biological outcome, with group status as the independent variable (with the use of SPSS software). The a priori hypotheses were that FMD and CAC function would be greatest in young men and women, significantly decreased in healthy older individuals, and significantly further decreased in CAD patients. Because we did not make specific hypotheses about differences between younger men and women, we used a more stringent critical α of 0.025 when comparing each of the younger groups separately (men or women only) with the older groups. For all other pairwise differences reported, Tukey post hoc corrections were used. We used Pearson correlations to investigate associations between functional properties of the groups.

For mouse cardiac functional experiments, data are presented as mean±SD. To compare means between multiple (>2) groups on days 2 and 28 post MI, SPSS was used to apply 2‐way repeated‐measures ANOVA, and pairwise significance was determined with Tukey post hoc correction. An interaction between group and time was included in the model. To compare means between day 2 and day 28 post MI in each group, 2‐tailed paired t test was applied. P<0.05 regarded as significant.

Characteristics of the study subjects are shown in Table 1. Systolic blood pressure showed increases associated with male sex, age, and CAD. However, diastolic blood pressure in the CAD group was comparable to that of the healthy older group and was not significantly higher than that in the younger groups. The CAD group, in which all patients were being treated with statins, exhibited lower total cholesterol and low‐density lipoprotein than the older healthy group. Notably, these conditions did not prevent brachial artery FMD, a measure of endothelial function, from being significantly lower in the CAD group than in both young healthy groups and the aged healthy group (Figure 1).

Migration of CACs toward VEGF has repeatedly been shown to correlate with cardiovascular health of the donor.^{6, 8, 16, 24} As shown in Figure 2A, the CACs from the young male group migrated toward VEGF more than did those from the CAD group (24.4±2.9 versus 10.5±1.3 per field, P=0.001), as did those from the young female group (19.3±2.9 versus 10.5±1.3, P=0.022; n=10/group for all migration experiments). Notably, as shown in Figure 2B, random migration in the absence of a VEGF gradient was also higher in young male CACs than in CAD CACs (12.3±1.6 versus 6.9±0.9, P=0.007) and higher in young female CACs than in CAD CACs (11.9±1.2 versus 6.9±0.9, P=0.012). In fact, when migration toward VEGF was normalized to random migration, no relationship achieved significance (not shown), demonstrating that intrinsic migration, rather than specific chemotactic response toward VEGF, is impaired in CACs from CAD patients.

Migration toward VEGF was significantly higher in the young male group compared with healthy older adults (24.4±2.9 versus 15.00±3.0 per field, P=0.015), but none of the other pairwise comparisons for VEGF or control migration approached significance (P>0.1 for all).

While late‐outgrowth EPCs have been reported to differentiate ex vivo into capillary‐like tubes, CACs by themselves do not form tubes.² However, CACs cocultured with endothelial cells will adhere to the endothelial tubes (Figure 3), a property that is decreased in CACs from diabetic subjects who exhibit other functional deficits.⁹ This property may reflect the homing of CACs to angiogenic endothelium and their association with the endothelial layer. When cocultured with HUVECs, association of CACs with tubes was slightly but significantly higher in the young healthy women (n=10) and men (n=9) relative to the CAD group (n=9) (89.6±2.3% and 88.8±2.4% versus 75.35±2.9%, P=0.001 and 0.003, respectively) but not relative to that of the older healthy group (83.14±3.7%; n=10, both Ps>0.1) (Figure 2C). Association of CACs was higher in the older healthy group than in the CAD group with a trend toward significance (P=.065; critical α was 0.05 for this relationship). We did not observe group‐specific pattern differences in the associated tubes (not shown).

We also examined whether age or CAD would reduce the number of CACs to adhere to quiescent or tumor necrosis factor‐α–activated endothelial monolayers, as described by Tepper et al⁹ for CACs from diabetic patients. We did not observe differences between groups in adhesion to monolayers (not shown), in agreement with Ward et al.²⁴

To determine how age and CAD influence production of NO from CACs, we measured both eNOS mRNA and eNOS protein in CAC lysates and measured nitrite in the culture medium as an indication of NO release from the cells. Figure 2D shows real‐time RT‐PCR results for eNOS in the CAC groups. The eNOS RNA level in the CAD group was significantly decreased relative to the young healthy female group (0.0010±0.0001 versus 0.0019±0.0003 relative unit, P=0.005; n=10 for all CAC groups). The differences between the CAD and young healthy male group (0.0017±0.0001) approached but did not reach significance (P=0.049; critical α was considered to be 0.025 for this relationship), whereas the similar difference between the older healthy (0.0017±0.0002) and CAD groups achieved significance with P=0.037 because the critical α was 0.05 (see Statistical Analyses). These results, along with the lack of a decrease in eNOS mRNA in the aged healthy group relative to either younger group (no difference approached significance), suggest that CAD was the driving effect. However, because the average age of the CAD group was greater than that of the aged healthy group, it is unclear whether this effect was driven by CAD per se or by even greater age (see Discussion).

In contrast, eNOS protein level measured with ELISA as picogram per microgram total cellular protein (Figure 2E) was higher in both healthy young women (2.2±0.3; n=9) and healthy young men (2.0±0.4; n=7) than in the CAD group (0.70±0.1; P=0.001 and 0.003, respectively) and the healthy older group (1.0±0.3; P=0.005 and 0.017) (each n=10). The interpretation of age as the driving factor rather than CAD is bolstered by lack of difference between the aged healthy and CAD groups (P=0.46). Further, NO_x (reflecting NO release) in conditioned culture media from the CACs (Figure 2F) showed an age‐related effect with almost identical values in the older healthy (0.20±0.03 nmol/10⁴ CACs) and CAD (0.18±0.04 nmol/10⁴ CACs) groups (both n=10) that were significantly lower than those in the healthy young female (0.40±0.06 nmol/10⁴ CACs; n=9) and healthy young male (0.43±0.07 nmol/10⁴ CACs; n=7, all P≤0.004 for all) groups.

In a separate follow‐up experiment, CAD CACs were still able to increase NO generation when stimulated with ionomycin (9.2±3.8 nmol/10⁴ unstimulated CACs versus. 13.5±5.3 nmol/10⁴ stimulated CACs; P=0.0003 by 1‐tailed paired t test).

A Pearson correlation matrix revealed similar correlations or lack thereof across the whole population (detailed results are shown in Table 2). CAC migration, dependent on intrinsic motility rather than chemotaxis, correlated significantly with eNOS protein levels (r=0.44, P≤0.01) with a borderline significant correlation with eNOS mRNA (r=0.27, P≤0.07). CAC migration also correlated with FMD (r=0.49, P≤0.01). However, the correlation between FMD and eNOS mRNA (r=0.20) or protein (r=0.15) did not reach significance, suggesting that of the outcomes analyzed, migration is the most predictive of overall function. This finding is in agreement with Britten et al,²¹ except that intrinsic motility appears to be more important than specific chemotaxis.

In a preliminary exploration of phenotypic changes in the CAD CACs that might underlie their poorer functional performance relative to CACs from the other groups, we assessed changes in expression of representative genes indicative of monocytic or endothelial characteristics, by using real‐time RT‐PCR analysis of pooled transcripts from each CAC group. Because we had observed only minor differences between young women and young men in our other functional characterizations of the individual CAC isolates, we combined them into a single young donor pool, yielding a total of 3 pools (aged <45 years, aged ≥45 years, and CAD patients). CAC isolates were grown individually in culture so that no single isolate would predominate the RNA complement as a result of faster growth, and the resulting unbiased RNA pools from the 3 groups were analyzed for relative expression of KDR, CD31, CD14, and CD45, normalized to the aged <45 years group (Figure 4). KDR was not detectable. However, while there was not a major shift in CD45 or CD31 expression among the groups, the CAD group had almost double the expression of CD14 of that of the other groups. CD31 went down slightly in that group. Because this experiment was hypothesis generating and involved pools of donor CACs rather than distinct values per donor, no statistical analysis was performed. The pooled approach makes it unclear whether this was because of an increase in CD14 expression by CAD CACs or the results of a larger percentage of CD14⁺ cells in this heterogeneous sample; this is a launching point for further study.

We ranked all 40 CAC isolates on the basis of migration and eNOS transcription, and selected 2 of the highest‐function isolates (both from young healthy females) and 2 of the lowest‐function isolates (both from CAD patients) for further in vivo experiments. Characteristics of the CACs and their donors are provided in Table 3.

To test our hypothesis that CAC ex vivo function and donor health status would positively correlate with therapeutic potential for MI, we implanted each of these 4 isolates into hearts of separate groups of immunodeficient SCID mice at 3 days after surgical induction of MI. Injection of vehicle (Hanks balanced salt solution [HBSS]) served as a negative control. Closed‐chest injection was targeted to the per‐infarct border zone of the left ventricular myocardial wall by microultrasound guidance by using a method that we have previously reported,⁴³ and that we have used successfully to demonstrate therapeutic effects of nonselected bone marrow cells to improve post‐MI left ventricular function.^{26, 27, 44, 46} As shown in Figure 5A with numerical details in Table 4, the LVEF in all groups significantly decreased on day 2 post MI (P<0.001), the day before therapy. From day 2 to day 28 post MI, LVEF in negative control hearts injected with HBSS continued to deteriorate (P<0.05). In contrast, LVEF improved (P<0.05) from day 2 to day 28 in mice injected with the 2 most functional donor‐specific CAC isolates. The 2 least‐functional donor‐specific isolates either only partially preserved LVEF (P=NS) or did not preserve LVEF (P<0.05). On day 28, overall comparison between the 5 recipient mice groups showed that LVEF in mice receiving either of the healthy CAC isolates was significantly higher than that in mice receiving 1 CAD CAC isolate or HBSS (P<0.05). The other CAD isolate did not achieve significant difference from 1 of the healthy CAC isolates but still differed from both of them by virtue of a lack of significant improvement in function over time. Similar results were obtained for end‐systolic volume and end‐diastolic volume, as shown in Figure 5B and 5C and Table 4.

Because migration positively correlated with eNOS mRNA level and therapeutic potential, we attempted to augment these qualities in the lowest‐functional CAC isolate (CAD) by adenoviral transduction to overexpress eNOS, compared with a GFP virus control. In the least‐functional CAC isolate (“CAD CACs #2” in Figure 5), eNOS was undetectable by immunoblotting, but after transduction with eNOS adenovirus, eNOS was clearly detectable at levels exceeding the natural level in HUVECs (Figure 6A), which we have reported to be higher than in CACs.¹⁶ NO release from eNOS‐transduced CAD CACs was elevated relative to that from untransduced and GFP‐transduced CACs of the same isolate (0.24±0.01 versus 0.15±0.01 and 0.14±0.03 μmol/L, both P<0.05; Figure 6B). Migration was substantially elevated in the eNOS‐transduced CAD CACs compared with nontransduced and GFP‐transduced CACs (7.1±0.8 versus 4.2±0.4 and 4.6±0.4 cells/high‐power field, both P<0.05; Figure 6C). When the CAD CACs were implanted intramyocardially into groups of SCID mice post MI, the therapeutic efficacy of the eNOS‐transduced CAD CACs was improved relative to the untransduced CAD CACs and HBSS (Figure 6D, values in Table 4).

To mechanistically confirm the central requirement for eNOS in these functional and therapeutic properties, we asked conversely whether the most‐functional CAC isolate (young healthy female) could be functionally and therapeutically impaired by eNOS knock‐down via small interfering RNA (siRNA) adenoviral transduction. Transduction of HUVECs with this virus at 500 multiplicity of infection reduced eNOS protein level to 22% of untransduced levels as determined by quantitative immunoblotting (Figure 7A). When healthy CACs (“Healthy CACs #2” in Figure 5) were transduced with the siRNA virus, NO release was markedly decreased compared with untransduced and GFP virus–transduced CACs of the same isolate (0.07±0.02 versus 0.39±0.04 and 0.36±0.06 μmol/L respectively, both P<0.05; Figure 7B). Migration was substantially reduced in the siRNA‐transduced CACs compared with the untransduced and GFP‐transduced CACs (6.3±1.3 versus 19.6±2.9 and 17.6±2.5 cells/high‐power field, respectively, both P<0.01; Figure 7C). When these healthy donor CACs were implanted into post‐MI mouse hearts, the therapeutic efficacy of the siRNA‐transduced CACs was impaired relative to the untransduced group but not to the level of the HBSS‐injected control group (Figure 7D, values in Table 4).

To confirm ability to detect human CACs in mouse heart tissue, a pilot experiment was performed in which CACs were preloaded with cytoplasmic Cell Tracker green dye before implantation into SCID mouse hearts. Immunofluorescence staining of tissue sections for HLA detected cells with cytoplasmic cell tracker and surface HLA staining (Figure 8). However, despite clear functional benefits of CAC implantation even 25 days postimplantation (28 days post MI), we did not detect CACs persisting in the tissue at that late time point.

Histological analysis of tissue sections at the infarct border zone of day 28 post‐MI hearts from different representative CAC implantation groups revealed no correlation with density of capillaries and arterioles (Figure 9). Expected relationships were observed for infarct scar size based on the cardiac functional data; however, the results did not approach significance because of the underpowered group size and the loss of several frozen negative control hearts (Figure 10; see Discussion).