|Home | About | Journals | Submit | Contact Us | Français|
Cell death by apoptosis is critical in myocardial diseases, and non-invasive detection of early, reversible apoptosis might be useful clinically. Exogenous Annexin-V (ANX) protein binds membrane phosphatidylserine, which is externalized in early apoptosis. A molecular MRI probe was constructed with superparamagnetic iron oxide (SPIO) conjugated to recombinant human ANX (ANX-SPIO). Apoptosis was induced with doxorubicin, a cardiotoxic cancer drug, in culture in neonatal rat ventricular myocytes, cardiac fibroblasts, and mesenchymal stem cells, and in vivo in the mouse heart. ANX-SPIO was validated using T2*-weighted 3T MRI. ANX-SPIO produced T2* signal loss, reflecting iron content, that correlated highly with independent apoptosis markers; bound with high affinity to apoptotic myocytes by competition assay (Ki 69nM); detected apoptosis in culture much earlier than did TUNEL stain; and revealed fibroblast resistance to apoptosis. With apoptosis in vivo, ANX-SPIO produced diffuse myocardial T2* signal loss that correlated with increased iron stain and caspase activity. Treatment with an alpha-1-adrenergic agonist in vivo reversed apoptosis and eliminated the ANX-SPIO MRI signal. It is concluded that cardiac MRI of ANX-SPIO detects early, non-ischemic cardiac apoptosis in culture and in vivo, and can identify reversibly injured cardiac cells in diseased hearts, when treatment is still possible.
Cardiac cell death by apoptosis is a key pathogenesis in ischemic and non-ischemic cardiomyopathy (1), and cardiomyopathies are the main cause of heart failure, a major source of morbidity and mortality. Apoptosis was thought previously to be irreversible, but recent evidence indicates that early apoptosis is reversible, and that vulnerable cells could be rescued with timely intervention (2). The ability to monitor cardiac cell apoptosis, non-invasively and in real-time, longitudinally following coronary bypass surgery or angioplasty, cancer chemotherapy, or cardiac allograft rejection could reveal fundamental disease mechanisms, improve therapeutic monitoring, and guide future treatments.
Cardiac magnetic resonance imaging (MRI) has high spatial and temporal resolution, does not expose patients to damaging radiation, which is ideal for an approach involving serial, longitudinal studies, and can simultaneously define cardiac anatomy and function. Thus, we aimed to develop a molecular MRI probe strategy to detect early cardiac cell apoptosis.
An early event in apoptosis is translocation of phosphatidylserine (PS) from the cytoplasmic face of the membrane to the extracellular domain, where Annexin V (ANX), a soluble 36kDa protein, can bind PS with high affinity (3) (Figure 1). This principal underlies a well-validated, commercial assay for apoptosis using ANX conjugated to fluorescein isothiocyanate (ANX-FITC), and was first employed in MRI to detect tumor cell apoptosis, by conjugating a similar PS-binding molecule, synaptotagmin, to a negative MRI contrast agent, superparamagnetic iron oxide (SPIO) (4). However, application of this approach to the heart has been limited. Recently, other groups have conjugated ANX to cross-linked iron oxide, gadolinium chelates, and/or fluorochromes to image cardiac apoptosis in mouse models of ischemia-reperfusion, genetic injury, and atherosclerosis (5-8). Here we developed a different molecular MRI probe using two compounds approved individually by the FDA, human ANX and SPIO, and a clinical 3 Tesla MRI scanner. We introduce a robust method to conjugate ANX-SPIO, and validate the sensitivity and specificity of the probe for apoptotic cardiomyocytes in vitro and in vivo. To do this, we utilize a clinically important, apoptosis-specific model of heart disease (doxorubicin cardiomyopathy) (9) to determine whether ANX-SPIO could detect apoptosis in vivo, without the overwhelming degree of necrosis found in ischemia-reperfusion models. We show that ANX-SPIO is a reliable probe for myocyte apoptosis, and can detect reversal of apoptosis with appropriate drug treatment.
Recombinant human ANX protein was expressed using a commercial ANX clone (Genecopeia, Germantown, MD). Competent BL21 E. Coli were transformed with a DNA plasmid containing a glutathione-S-transferase (GST)-ANX (human) fusion cDNA and a 5′-T7 bacterial promoter; colonies were cultured overnight in 100μM Ampicillin/10μM IPTG medium; and plasmids were purified (Qiaprep Miniprep, Alameda, CA), confirmed by PCR (forward-5′-GCTGGCAAGCCACGTTTGG-3′, reverse-5′-TTCACTTCTGAGTTCGGCATG–3′), grown overnight (250ml), and isolated (Maxiprep, Qiagen). The ANX-GST fusion protein was purified on glutathione-agarose affinity columns (Thermo Scientific, Waltham, MA), confirmed by SDS-PAGE (Bio-Rad, Hercules, CA), and cleaved with enterokinase to remove the GST tag. Free ANX was purified by passing through a glutathione-agarose affinity column a second time to remove free GST, and quantified by Bradford protein assay.
We adapted a previous method to covalently link SPIO hydroxyl groups to Annexin V amine groups (10). SPIO particles (Feridex, Sigma, 11.2mg iron/ml, or Ocean Nanotech Inc., 5mg/ml) were oxidized 1h at 20°C in the dark in solution with sodium periodate (NaIO4), in a 4:1 weight:weight ratio (working concentration ~0.15M NaIO4), incubated 12h with purified ANX protein (1:1 ratio, weight:weight) in 0.15M sodium borate (NaBH4) at 20°C, further reduced 1h with sodium borohydride, and quenched with 0.15M Tris-HCl. Free ANX was separated from ANX-SPIO by centrifugation at 14,000xg (75μm pore Microcon filter, Millipore, Billerica, MA). The filter retentate (ANX-SPIO) was separated from unbound SPIO by size exclusion bead chromatography (Agarose Bead Technologies, Tampa, FL), and quantified by Bradford protein assay. ANX-SPIO was stored at −80°C at 2μg/μl in 0.1M Tris-HCl with 1% serum albumin and protease inhibitors (Halt Protease Inhibitor, Thermo Scientific, 1-50 μg/mg ANX).
For TEM, 40nm iron oxide nanoparticles (5mg Fe/ml) were diluted eight-fold in phosphate buffered saline (PBS), and 5μl of the nanoparticle suspension was drop cast onto copper TEM grids coated with carbon support film, incubated on the grid for 5min, rinsed with deionized water, and blotted dry with filter paper. TEM was done in a Tecnai G2 F20 FEG-TEM operated at 200kV.
To measure hydrodynamic size, SPIO and ANX-SPIO were separately diluted in PBS to an optical density of 0.1, and analyzed by DLS in a Zetasizer Nano DLS machine (Malvern Inc., Worcestershire, UK) as described (11). Monomodal peaks were obtained for each particle, and Z-average particle size and Polydispersity Index (PDI) were measured using 17 repetitions for each compound.
Neonatal rat ventricular myocytes (NRVMs) were isolated from day-old hearts, plated in 35mm dishes overnight in MEM with 5% calf serum and bromodeoxyuridine (BrdU) 0.1mM, and switched to serum-free medium with transferrin, insulin, and bovine serum albumin (12). Neonatal rat cardiac fibroblasts (FBs) were scraped from the preplates and grown in MEM with 5% calf serum, with an additional passage to remove NRVMs. Adult mouse bone marrow mesenchymal stem cells (mMSCs) were cultured in DMEM supplemented with 10% FBS and 1% penicillin–streptomycin (Invitrogen), and used at passage 20-30 (13).
In culture, apoptosis was induced by treatment for varying times at 37°C with DOX (Sigma, 1μM in 0.9% saline) (12). Apoptosis in culture was quantified using either MRI with ANX-SPIO (below), or using microscopy after TUNEL stain (APO-BrdU™ TUNEL Assay Kit, Molecular Probes/Invitrogen, Eugene, OR) or after stain with ANX-FITC (Roche, Switzerland). In vivo, male FVB/NJ mice age 12-16 weeks, were given 25mg/kg DOX intraperitoneal (ip) (200μl), a dose shown previously to cause apoptosis over days (14). The Stanford University Administrative Panel on Laboratory Animal Care approved animal protocols.
Osmotic pumps (Alzet #1002, Cupertino, CA) loaded to deliver 10ng/kg/d A61603 (Tocris #1052, Ellisville, MO) or vehicle (0.9% NaCl with 100μM ascorbic acid) were implanted subcutaneously under isoflurane anesthesia, on the same day as DOX injection.
We used a 3 Tesla GE Signa EXCITE scanner and a one-in-diameter custom loop surface coil. In culture, cells were labeled with ANX-SPIO (10μg protein/ml medium except as noted), or an equivalent weight of free SPIO (5μg), for 15min at 37°C, washed 4x with PBS, overlaid with 1ml of 1% agarose to prevent diffusion of the SPIO nanoparticles, and imaged in the dishes. Alternately, cells were washed and trypsinized after DOX and ANX-SPIO labeling, then pelleted in Eppendorf tubes. Culture dishes or tubes were placed into a copper sulfate agar phantom, and gradient-echo (GRE) spoiled gradient sequences were used with repetition time (TR) 550ms, echo time (TE) 10ms, flip angle 15°, matrix 256×256, field-of-view 3cm × 3cm, slice thickness 1.3mm, number of acquisitions (NEX) 2, spacing 0mm (15). Axial images through the cell layer on the plate were analyzed for T2* signal loss, using regions-of-interest (ROIs) of a fixed area (0.8mm2), and adjusting for background using control ROI signal from the agar phantom. Background noise was measured from the signal of the air surrounding the phantom. All in vitro MRI experiments were performed at a temperature of 21°C and 3T field strength to ensure consistent T1/T2 properties of the copper sulfate agarose gel phantom.
We calculated Contrast-to-Noise Ratio (CNR)=(SI sample ROI-SI control ROI)/(SD of SI of background noise), where SI is signal intensity, and SD is standard deviation.
We used 3T field strength to measure relaxivities r1 and r2 for both nanoparticles, functionalized ANX-SPIO and non-functionalized SPIO. Each was diluted to (16) 0-18mM in HEPES-buffered saline at room temperature in Eppendorf tubes in a copper sulfate phantom agar. T1 values were obtained using a spin echo (SE)-inversion recovery sequence: TE minimum, TR 2550ms, FOV 10cm2, matrix 512×512, thickness 5mm, and NEX 1, with TI sampling at 50, 400, 1100, and 2500ms. T2 values were obtained using an SE sequence: TR 4000ms, FOV 10cm2, matrix 256×192, thickness 5mm and NEX 1, with TE sampling at 15, 30, 45, and 60ms. T1 values were computed using a reduced-dimension nonlinear least-squares algorithm with phase restoration (17), and T2 values were computed using linear regression. Linear regression (GraphPad Prism) was done with R1 (1/T1) and R2 (1/T2) plotted versus 5 different Fe concentrations, ranging from 0 to 18mM, with the resulting slope values (mM−1s−1) estimating r1 and r2.
In vivo, ANX-SPIO, or free SPIO as a control, were injected by tail vein after DOX, as in Figure 7 legend, and MRI was done at intervals, using cardiac and respiratory gating, under anesthesia with isoflurane by nose cone (2% induction, 1% maintenance). Mouse temperature was kept physiological with a heating pad and rectal probe. Heart rates ranged between 300 and 400 bpm, with core temperature between 95 and 97°F, and total acquisition time was 3 to 3.5 min per slice, depending on cardiac gating efficiency. T2*-weighted GRE sequences were used in the short and long axes to detect T2* signal loss in multiple imaging planes, using TR 100ms, flip angle 60°, matrix 256×256, slice thickness 0.8mm, field-of-view 4-8cm2, NEX 6, TE variable 5-20ms.
Myocardial iron measurement was adapted from a clinical cardiac MRI protocol (18). Integrated signal values from the areas of signal loss were measured with a circular ROI of 9 pixels (0.8mm2) taken from the septum of each heart during diastole, with MRI acquisition TEs from 5 to 20ms (19). Background signal was subtracted using an ROI of matched area from non-cardiac muscle within the same image frame. As the TE was increased, T2* signal loss intensified in an exponential decay pattern, which was curve-fitted using GraphPad Prism (GraphPad Software, La Jolla, CA).
Myocardial caspase 3 and 7 activities were measured using the Caspase-Glo assay kit (Promega). Hearts were removed under deep anesthesia with isoflurane, rinsed in saline, frozen in liquid nitrogen, and stored at −80°C, then homogenized (25mM HEPES, pH 7.5, 5mM MgCl2, 1mM EGTA, 1mM Pefablock, and 1μg/ml each pepstatin, leupeptin, aprotinin), and centrifuged (15min, 13,000rpm, 4°C). The supernatant (cytosol) was adjusted to 1mg/ml protein and stored at −80°C. An equal volume of reagent and 100μg cytosolic protein were added to a white-walled 96-well plate and incubated 1h at room temperature, and luminescence was measured in triplicate, minus background.
Hearts were fixed in 1% paraformaldehyde, embedded in paraffin, sectioned at 100μm, and stained for iron using a diaminobenzidine-enhanced Prussian blue protocol, or sectioned at 8μm and stained for collagen with sirius red.
Results are mean ± SEM. Significant differences (p<0.05) were tested using analysis of variance (ANOVA) and Bonferoni post-test for more than two groups or time points, and Student's unpaired t-test for two groups. Linear regression tested for association between MRI CNR and ANX-FITC cell counts. Goodness-of-fit to one- or two-site models for ANX-SPIO competition binding analysis was compared by F test, and dissociation constant (Ki) was calculated (GraphPad Prism v4.0).
Human ANX conjugated to GST was generated using an E. Coli expression system (Figure 2A). PCR of positive colonies using primers for the GST-ANX fusion construct revealed an expected 964bp product (Figure 2B). GST-ANX fusion protein from a large-scale preparation was isolated on glutathione-agarose affinity columns, as verified by Coomassie blue stained gels for the 63kDa protein (Figure 2C). The GST tag was removed by enterokinase digestion and re-passage through a glutathione affinity column to remove free GST. Purified ANX (36kDa) was conjugated with SPIO, and the ANX-SPIO product was purified. Yield from a 250ml bacterial culture was ~10mg GST-ANX and ~6mg purified ANX.
The 40nm SPIO nanoparticles contain ~1.2 × 106 Fe atoms per nanoparticle (20) (www.oceannanotech.com). Conjugation produced ANX protein-SPIO iron with a 1:300 moles protein:moles iron ratio, and 2:1 mg:mg weight ratio. TEM confirmed homogeneous SPIO nanoparticles, with a 40nm diameter (Figure 1B). From measured protein and iron concentrations, the molar ANX:SPIO stoichiometry was ~14:1. Annexin V molecules have a 2-3nm hydrodynamic diameter (21), and therefore, a 14:1 molecular ratio would predict an increase of ~35nm in ANX:SPIO particle size compared to SPIO alone. DLS supported this size estimate, as SPIO particle hydrodynamic size was 46nm (PDI=0.07), whereas ANX-SPIO conjugate size was 78nm, with a single peak and a narrow PDI (0.13) (Figure 1C). This molecular ratio was proportionally higher than previously synthesized ANX conjugates with smaller nanoparticles (e.g., 4-5:1 ANX:iron oxide ratio) (22,23).
Molar SPIO relaxivities r1 and r2 were 2.28±0.14mM-1s-1 and 426±38mM-1s-1, whereas ANX-SPIO r1 and r2 were 8.6±0.61mM-1s-1 and 326±16mM-1s-1. At an iron concentration of 1.25μg/ml, ANX-SPIO R1 was slightly higher than SPIO (0.84±0.28s-1 vs. 0.56±0.03s-1, N=3, p<0.05); however, R2 was not significantly different between ANX-SPIO and SPIO (13.9±0.5s-1 vs. 12.5±0.3s-1, N=3, p>0.05). These data indicated that the T2-reducing properties of SPIO were not altered significantly by ANX conjugation.
NRVMs were treated with either DOX (to induce apoptosis) or vehicle (VEH), labeled with ANX-SPIO, and transferred to tubes for coronal and axial MRI. MRI T2* signal loss, measured by CNR, was significantly (p<0.05) more intense in DOX-treated apoptotic myocytes labeled with ANX-SPIO than in non-apoptotic VEH-treated cells labeled with ANX-SPIO (Figure 3, lane 7 vs. lane 2). MRI T2* signal loss with ANX-SPIO in DOX-treated cells was also significantly greater (lower CNR value) than two other controls: DOX-treated apoptotic cells plus SPIO alone (lane 4), or VEH-treated cells without label (lanes 1 and 6). Signal loss with ANX-SPIO was dose-related, as both a 50% ANX-SPIO dose (lane 3) and a 50% DOX dose (0.5μM, lane 5) had intermediate CNRs (p<0.05 versus ANX-SPIO and versus VEH). These data suggested that ANX-SPIO labeled apoptotic cardiac myocytes specifically. A mild degree of signal loss was observed in control cells treated with ANX-SPIO (lane 2), which may reflect non-specific binding of ANX-SPIO versus a basal degree of apoptosis in cultured neonatal rat cardiomyocytes, which are known to have a limited lifespan in vitro.
To further test the specificity of ANX-SPIO for apoptotic cardiac myocytes, we compared labeling with ANX-FITC, an extensively validated, commercial marker for apoptosis. NRVMs were incubated with DOX for increasing times, to produce increasing numbers of apoptotic cells, then divided and stained with ANX-SPIO, or ANX-FITC. As shown in Figure 4A, the number of apoptotic cells detected by ANX-SPIO, quantified as CNR for T2* signal loss, correlated significantly with the number of apoptotic myocytes detected by ANX-FITC (R2=0.88, p<0.05).
To test ANX-SPIO affinity for apoptotic myocytes, cells were treated with DOX for 4h to cause maximum apoptosis, then stained with ANX-FITC in the presence of increasing concentrations of ANX-SPIO. As shown in Figure 4B, ANX-SPIO competed with high affinity to reduce the number of apoptotic cells labeled with ANX-FITC (Ki 69nM, R2=0.76). In a small fraction of cells (~20%), ANX-FITC was not displaced by ANX-SPIO, possibly due to the larger hydrodynamic diameter of ANX-SPIO (75nM vs. FITC 2nM) (24), indicating that ANX-SPIO might slightly underestimate the number of apoptotic cells.
Figure 4C illustrates ANX-FITC labeling of apoptotic myocytes (DOX), compared with control cells (VEH). Figure 4D illustrates the T2* signal loss with ANX-SPIO in DOX-treated cells, versus two independent controls, DOX-treated and labeled with SPIO alone, and VEH-treated labeled with ANX-SPIO.
In summary, these data further supported that ANX-SPIO stained apoptotic cardiac myocytes specifically and with high affinity.
A critical clinical need is to detect early apoptosis, before it is irreversible, and an advantage of ANX-labeling is affinity for externalization of PS, one of the earliest events after an apoptosis trigger (see Figure 1). To test if ANX-SPIO detected early apoptosis, we compared labeling with TUNEL, which detects DNA fragmentation, a late, terminal event in apoptosis. NRVMs were treated with DOX for increasing times, and then stained with either ANX-SPIO or TUNEL. Significant T2* signal loss from ANX-SPIO was observed as early as 30min after DOX, and peaked at 1h (CNR −80±6 at 1hr DOX vs. −16±5 baseline, N=3, p<0.05) (Figure 5). However, significant TUNEL staining was not observed until 2h DOX, and did not peak until 3h. Thus, ANX-SPIO detected apoptosis much earlier than did TUNEL.
To determine the lower detection limit using ANX-SPIO, serial dilutions of myocytes treated with DOX to induce apoptosis were labeled with ANX-SPIO, washed, and imaged on 35mm dishes. T2* signal loss was not detected when total cell counts were below 100 cells/dish, but discrete areas of T2* signal loss were observed at cell concentrations of 100-200/dish (data not shown). Light microscopy of the same fields revealed clumped cells, suggesting that discrete areas of T2* signal loss from ANX-SPIO binding arose from small populations of apoptotic cells (not shown).
Prior studies indicate that cardiac fibroblasts (FBs) are resistant to apoptosis triggers as compared with myocytes (25). To test if ANX-SPIO could detect differences among cell types, we treated cultured NRVMs, rat neonatal cardiac FBs, and mouse mesenchymal stem cells (mMSCs) with DOX for increasing times, and labeled with ANX-SPIO for MRI. As shown in Figure 6, NRVMs were most sensitive to DOX-induced apoptosis, followed by mMSCs, whereas FBs were most resistant (time of maximum T2* signal loss NRVM, 0.6±0.2h; mMSC, 1.8±0.5h*; FB, 22±1.3h*; N=6 at each time, *p<0.01 vs. NRVM). Furthermore, ANX-FITC revealed almost identical time to maximum apoptosis (NRVM, 0.7±0.2h; mMSC, 1.4±0.3h*; FB, 20±1.2h*; N=3 at each time, *p<0.01 vs. NRVMs), verifying the close correlation of ANX-SPIO with this validated apoptosis marker.
We tested ANX-SPIO in vivo using a DOX-induced model of diffuse, non-ischemic, apoptotic cardiomyopathy. Cardiomyopathy is a limiting toxicity in cancer therapy with DOX, so that this model has clinical relevance (9). In the mouse, the DOX dose used (25mg/kg ip) causes within one week, extensive apoptosis, dilated cardiomyopathy, and mortality (14,16). ANX-SPIO was injected by tail vein 2d after DOX. Subsequent cardiac MRI revealed myocardial T2* signal loss as early as 30min after injection, which was stable between 6-8h and 7-10d, and dissipated between 10-14 days (data not shown). As shown in Figure 7, MRI done 2d after DOX and 6h after ANX-SPIO injection showed a diffuse, speckled pattern of myocardial T2* signal loss with ANX-SPIO (Figures 7A and 7C), that was much greater than seen with DOX-treated mice injected with SPIO alone as a control (Figures 7B and 7D) (CNR of septum −19±4 vs. −3±1, N=4, p<0.05).
Prussian blue histology documented increased iron stain in myocardium with T2* signal loss (Figure 7C vs. 7D), confirming higher uptake of ANX-SPIO in the injured myocardium. Blue iron stain had a linear appearance, suggesting myocyte surface binding of ANX-SPIO, as expected, and myocardial structure was normal, consistent with early apoptosis (Figure 7C). DOX-treated hearts also had increased activity of executioner caspases at 2d (see Figure 9A and below). Cardiac fibrosis was not seen until 7d after DOX (Figure 7E). These data indicated that the ANX-SPIO signal correlated with early, diffuse apoptosis.
We quantified iron content from ANX-SPIO using GRE sequences with increasing TE times (0-20ms), calculating decay of myocardial T2* signal intensity. As shown in Figure 8, the myocardial T2* decay constant with ANX-SPIO was significantly shorter in DOX-treated mice vs. VEH-treated, control mice, again indicating higher iron content and ANX-SPIO uptake in hearts treated with DOX.
In summary, these results indicated that MRI with ANX-SPIO could detect early cardiac apoptosis in vivo.
To further test ANX-SPIO specificity for apoptosis in vivo, and to assess potential clinical utility, we asked if treatment of apoptosis altered myocardial T2* signal with ANX-SPIO. To do this, we treated mice with the alpha-1-adrenergic receptor agonist A61603 (A6). We found recently that A6 at a low dose prevented DOX-induced cardiomyopathy (14), consistent with an earlier study with a different alpha-1-adrenergic agonist (16). As an independent marker for early apoptosis, we quantified caspase activation, which occurs in early apoptosis at the same time as PS externalization (26). Myocardial iron content increased significantly at 2d after DOX, as reflected in Figure 8 by the steep T2* decay and decreased T2* value, and this increase was eliminated by treatment with A6. Similarly, A6 eliminated apoptosis as quantified by increased myocardial caspase activity at 2d after DOX (Figure 9A). Simultaneously, the marked speckled ANX-SPIO MRI signal in the ventricular septum after DOX was also eliminated by A6 (Figure 9B). These data further supported the specificity of ANX-SPIO for early apoptosis in vivo, and indicated that ANX-SPIO might be useful to monitor therapy.
The main new finding of this study is that ANX-SPIO, a molecular MRI probe, is sensitive and specific for early cardiac myocyte apoptosis in culture and in vivo, using T2*-weighted 3T MRI. We provide a simple, robust method to produce ANX-SPIO, and validate it extensively in culture and in vivo, using a model of DOX-induced cardiotoxicity. Our results suggest that ANX-SPIO might be useful clinically to detect reversibly injured, apoptotic myocytes when therapy is still possible, as shown here by treatment with an alpha-1-adrenergic receptor agonist.
In culture, ANX-SPIO bound specifically to apoptotic myocytes, and correlated extremely well with independent, established markers for apoptosis, the TUNEL stain and ANX-FITC. ANX-SPIO bound with high affinity to apoptotic myocytes by competition analysis (Ki 69mM), detected apoptosis in culture much earlier than did TUNEL stain, and revealed the expected differential apoptosis in myocytes and fibroblasts. With DOX-induced apoptosis in vivo, ANX-SPIO produced an early, diffuse, speckled myocardial T2* signal loss that correlated with increased iron stain and with an independent biochemical marker of early apoptosis, increased caspase activity.
Previously, a 9.4 Tesla small bore MRI scanner was used with an ANX- and Cy5.5-conjugated magneto-optical probe to detect T2* signal loss in mouse models of ischemia-reperfusion and genetic injury (6,7). Here, we used a clinical 3 Tesla MRI scanner with ANX-SPIO to detect diffuse apoptosis after a cardiotoxic cancer drug. A surprising finding was the robust signal with ANX-SPIO, despite that apoptosis is infrequent in the model, ~1% of cells, when quantified by a sensitive TUNEL assay (14). A potential explanation is found in the recent observation that TUNEL stain detects only 10% as many apoptotic cells as does staining for caspase activation, which is a more proximal event in the apoptosis cascade (27). Therefore, a robust signal with ANX-SPIO is anticipated even when apoptosis by TUNEL is negligible or low, because externalization of PS and caspase activation occur earlier in apoptosis (Figure 5 and (26)). Although PS can also be externalized in necrosis, the predominant mode of cell death in DOX cardiotoxicity is apoptosis (9), and ANX-SPIO uptake was associated with a significant DOX-induced increase in TUNEL staining (Figure 5) and caspase activity (Figure 9), specific markers for apoptosis.
Indeed, the concept that apoptosis is a sequential process, and that not all cells progress to its terminal stages as assayed by nuclear DNA fragmentation (TUNEL), underlies the crucial idea that apoptosis can be reversed by appropriate, well-timed therapy (2). This principal highlights the importance of an MRI molecular probe for early apoptosis, such as ANX-SPIO. Indeed, we show here that appropriate therapy with an alpha-1-adrenergic agonist can reverse both caspase activation and ANX-SPIO signal in DOX-induced apoptosis in vivo (Figure 9).
It remains to be determined whether ANX-SPIO might be appropriate for clinical use. Currently, we use ANX-SPIO at a dose that would translate in a 70kg man to about 100mg SPIO and 200mg ANX. By comparison, the dose of Feridex, an intravenously injected colloidal SPIO associated with dextran used for MRI of liver tumors, is ~1mg/kg, and human annexin V plasma levels are 2-15ng/ml (28), so current dosing does not seem unreasonable.
In summary, ANX-SPIO appears to be a useful molecular MRI probe to detect early apoptosis in vivo. MRI with ANX-SPIO might yield valuable insight into basic mechanisms of cardiovascular disease, and new diagnostic and therapeutic strategies. For example, ANX-SPIO might identify and monitor early cardiac injury in cancer chemotherapy. We demonstrated ANX-SPIO labeling in vivo with a clinically applicable 3 Tesla MRI sequence, highlighting the potential relevance. Future investigation is needed into the ability of ANX-SPIO to detect dynamic cell death and therapy in other injury models.
We thank Drs. Keh Kooi Kee, Ai Leen Koh, Ildiko Toma, and Jesse Jokerst for technical assistance.
Funding Sources: VA (PCS), NIH (JC, MY, JB, DN, PCY, PCS); American Heart Association, Western States Affiliate (RD); Sarnoff Cardiovascular Research Foundation (TC)
Conflict of Interest: none