Molecular Imaging of Infarcted Heart by Biofunctionalized Gold Nanoshells

The unique combination of physical and optical properties of silica (core)/gold (shell) nanoparticles (gold nanoshells) makes them especially suitable for biomedicine. Gold nanoshells are used from high‐resolution in vivo imaging to in vivo photothermal tumor treatment. Furthermore, their large scattering cross‐section in the second biological window (1000–1700 nm) makes them also especially adequate for molecular optical coherence tomography (OCT). In this work, it is demonstrated that, after suitable functionalization, gold nanoshells in combination with clinical OCT systems are capable of imaging damage in the myocardium following an infarct. Since both inflammation and apoptosis are two of the main mechanisms underlying myocardial damage after ischemia, such damage imaging is achieved by endowing gold nanoshells with selective affinity for the inflammatory marker intercellular adhesion molecule 1 (ICAM‐1), and the apoptotic marker phosphatidylserine. The results here presented constitute a first step toward a fast, safe, and accurate diagnosis of damaged tissue within infarcted hearts at the molecular level by means of the highly sensitive OCT interferometric technique.


Introduction
Medical imaging of damaged myocardial tissues constitutes a valuable prognostic tool for patients with ischemic heart disease. Localization of myocardial damage is typically achieved by using either nuclear or magnetic resonance imaging (MRI) contrast agents. Some of them, such as those containing gadolinium as MRI contrast agent, are contraindicated for patients with renal impairment, whereas others, such as gamma ray-emitting tracers, imply the use of ionizing radiation that could lead to the appearance of undesirable collateral effects. [1] In order to avoid those limitations, new biocompatible contrast agents [2] are being developed for myocardium imaging with alternative technologies. [3] Among them, optical nanoparticles capable of absorbing, scattering, or emitting IR light have emerged as a reliable alternative. [4] The good penetration of IR light into tissues allows for high-resolution depth imaging. [5] In addition, the nonionizing character of IR light minimizes the collateral effects caused during imaging experiments. Of special relevance are gold nanoparticles (AuNPs) due to their unique combination of properties. [6] These include excellent long-term stability, fast and inexpensive synthesis, and facile surface biofunctionalization. [7] Reports of in vivo toxicity and biodistribution of large gold nanostructures (as the ones presented herein) in mice propose that nanoparticles between 60 and 200 nm are mainly accumulated in the liver and the spleen. [8] Studies carried out for long times after the injections of AuNPs solutions conclude that gold is cleared from the liver within the first month, but further research into clearance by the spleen is necessary since the literature does not provide a unique conclusion. [8d,8e] Even if there is a clear accumulation in both organs, no histological or biochemical changes have been reported up to date. A simple in vitro toxicity test of the commercial AuNPs used along this work is provided elsewhere, [9] showing that no in vitro toxicity of the AuNPs is noticed.
The morphology-dependent plasmon resonance of AuNPs leads to an enhanced optical response at selected wavelengths that has been extensively employed in preclinical studies. [10] AuNPs have been used for cardiovascular studies enabling, for instance, imaging of brain vasculature, thrombus detection by photoacoustic imaging, angiogenesis monitoring, drug delivery to myocardium tissues, enhancement of myocardial regeneration, and retinal vessels visualization. [11] Recently, it was demonstrated how an appropriate biofunctionalization of AuNPs allows imaging of myocardial scarring by using them as contrast agents in computed tomography, i.e., not by the noninvasive optical methods. [12] Optical coherence tomography (OCT) is a pure optical technique capable of in-depth tissue visualization that is based on the analysis of backscattered IR radiation by tissues. Originally used for ophthalmology, [13] OCT is now being used in hemodynamic and interventional cardiology units. [14] Indeed, clinical equipment already exists combining OCT optical systems with catheters for intracoronary imaging. Due to their large scattering cross-sections, AuNPs of different morphologies were previously proposed as efficient intravascular OCT contrast agents. For instance, gold nanoroses have been used as contrast agents for OCT imaging of atheroma plaques in rabbit aorta. [15] Among the great variety of AuNPs, gold nanoshells (GNSs) demonstrated to be the best contrast agents among those commercially available. [16] Nevertheless, a systematic comparison of the scattering properties of differently sized and shaped AuNPs (commercially and not commercially available) can be seen in the Supporting Information ( Figure S1, Supporting Information). Besides, GNSs' optical properties can be easily tuned in the synthesis process. [7e] Furthermore, it has also been postulated that spherical nanoparticles, as GNSs, could be especially indicated for in vivo molecular imaging. Indeed, when intravenously administered, spherical NPs interact more efficiently with cells than other geometries [17] such as elongated NPs, because the latter are extended by the flow. [18] In addition, GNSs combine their large scattering crosssection at near-infrared wavelengths of interest with high photothermal conversion efficiency, making possible their use as contrast agents not only in conventional OCT [19] but also, in pho-tothermal OCT. [19a] Finally, a very important aspect; the possible clinical application of GNSs as therapeutic agents has recently been demonstrated, supporting the feasibility of the near future application of GNSs in the clinics. [20] Despite all these advantages and promising results, the use of GNSs for imaging of the infarcted myocardium has not been explored yet.
Infarct imaging at molecular level with GNSs requires their functionalization with suitable ligands for specific targets overexpressed in damaged myocardial tissues. Ischemia induces complex changes in cardiomyocytes that result in the upregulation and overexpression of different targets. Thus, different approaches do exist, among all of them inflammation and damage targeting result of special relevance: i) Ischemia-induced inflammation can be imaged with nanoparticles functionalized with peptides or antibodies dedicated to targeting the intercellular adhesion molecule (ICAM-1). ICAM-1 plays an important role in a number of cellular events, including adhesion of leukocytes to the endothelium followed by their emigration into sites of inflammation. [21] ICAM-1 is expressed on the surface of a wide variety of cell types. [22] In cardiomyocytes, mRNA and protein levels of this adhesion molecule are up-regulated after acute ischemia, [22] both in animals [23] and in humans. [24] This up-regulation is mediated by an increase in pro-inflammatory cytokines such as IL-6, [25] TNF-, [26] or IL-1 . [23c] It was reported that ICAM-1 overexpression mediates, at least in part, the ischemia-induced damage of myocardial tissue, and that inhibition of its expression in the heart dramatically reduces both the infiltration of immune cells and the infarct size. [22] ii) Targeting the ischemia-induced apoptosis of cardiomyocytes by anti-phosphatidylserine (anti-PS) functionalized probes constitutes an alternative route for imaging the infarcted heart. Apoptosis takes place during reperfusion and is associated with loss of membrane phospholipid asymmetry, which is one of the main mechanisms that allow macrophages to recognize cells that are undergoing apoptosis. [27] The remodeling of the cell membrane in apoptotic cells is associated with a reduction of amino phospholipid translocase, resulting in translocation of phospholipids such as PS and phosphatidylethanolamine into the outer leaflet of the cell membrane. [28] This translocation makes the recognition of PS by macrophages possible in order to rapidly eliminate the apoptotic cells. [29] The aim of this work is to take advantage of the large IR scattering cross-section of GNSs combined with state-of-the art intravascular OCT systems in order to visualize the damaged myocardium after an ischemic event. Selective accumulation of GNSs at damaged myocardial sites was achieved by proper surface decoration with both anti-PS and anti-ICAM-1 ligands, providing them with high affinity to cardiomyocytes subjected to both inflammation and apoptosis, respectively. Ex vivo images of infarcted hearts perfused with these functionalized GNSs were obtained. The differences between the OCT images of the healthy and infarcted myocardium were critically analyzed and the potential of GNSs as molecular contrast agents for heart studies have been evaluated.

Results and Discussion
In this work, we used commercial GNSs provided by Nanocomposix. Figure 1a shows a typical transmission electron microscopy (TEM) picture of the GNSs, revealing their homogenous size distribution as well as their spherical shape. They consist of a 200 nm diameter silica core surrounded by a 20 nm thick gold shell (see schematic drawing in the inset of Figure 1b). These GNSs show a broad plasmonic extinction band spanning from the visible up to the IR that leads to a high extinction coefficient at the operating wavelength of cardiovascular OCT systems (1320 nm as indicated in Figure 1b). Figure 1c schematizes our approach for the detection of damaged cardiomyocytes. It implies, as stated previously, the use of anti-PS or anti-ICAM-1 functionalized GNSs to target apoptosis and inflammation, respectively. A detailed description of the functionalization procedure is included in the Experimental Section. Figure 2a shows comparative Fourier-transform Infrared (FTIR) spectra corresponding to GNSs, before functionalization (i.e., with only lipoic acid on their surface), and after anti-ICAM-1 and anti-PS functionalization. Lipoic acid, as a ligand, displays a characteristic broad O-H stretch vibration peaking at around 3425 cm −1 , a C=O stretch at 1676 cm −1 , and a C-O vibration at 1085 cm −1 . i) After functionalization with the peptide (anti-ICAM-1), the broad O-H band has increased relative to the other bands, due to the presence of more hydrogen bonding in the peptide. Additionally, the C=O double bond is slightly shifted to shorter wavenumber (1646 cm −1 ), which is typically observed when moving from the acid to the amide. [30] The C-O vibration is drastically reduced in intensity, revealing a relative decrease of C-O bonds, which also goes along with a shift from acid to amide. ii) The functionalization of the GNS with polyethylene glycol (PEG; used to link the nanoparticle to anti-PS) shows a broad band between 497 and 717 cm −1 which is characteristic of the Au-S bond between the thiol group on the PEG and the GNS. It also shows an O-H stretch at 3407 cm −1 . The addition of the anti-PS can be characterized by a broad band at 620-880 cm −1 and a narrow peak at 3371 cm −1 which is characteristic of the NH bond. The C=O stretch is common between the PEGfunctionalized GNS and the GNS functionalized with PEG and anti-PS.
Both the anti-ICAM-1 or anti-PS surface decorations should make the specific targeting of damaged cardiomyocytes possible, as illustrated in Figure 1c.    Figure 1a). Anti-ICAM-1 and anti-PS functionalization increases this hydrodynamic radius up to about 300 and about 600 nm, respectively. Despite this substantial increment in the hydrodynamic size, the biofunctionalized GNSs kept their stable colloidal character, without evidence for agglomeration over weeks. On the other hand, the uptake of these rather large GNSs by the reticuloendothelial system is a possible concern since it might reduce the circulating time of the GNSs, compromising the specific labeling. This aspect should be tested in future experiments.
To verify the molecular specificity of the anti-PS GNSs, in vitro experiments regarding endothelial cells (human umbilical vein endothelial cells) cultures were carried out. We compared the dark field microscopy images of healthy and apoptotic (dimethyl sulfoxide (DMSO) treated) cell cultures after being incubated with functionalized (anti-PS) and nonfunctionalized GNSs. The highest amount of GNSs (red spots) attached to endothelial cells is found in the apoptosis-induced culture incubated with anti-PS functionalized GNSs (see Figure S2 in the Supporting Information). Figure 3a shows a schematic representation of the Langendorff model used in this work. Experimental details can be found in the Experimental Section. Briefly, the rat heart is mounted in the perfusion system and local ischemia is induced by a ligature of the left anterior descending coronary artery. The occlusion time was varied between 0 and 60 min. The ischemic area was detected by staining heart slices with tryphenyl tetrazolium chloride ( Figure 3b). After coronary occlusion, hearts were perfused either with a solution without GNSs (vehicle) or with a solution containing biofunctionalized GNSs.
As mentioned in the introduction section, PS translocates to the outer leaflet of the cell membrane of apoptotic cells. In order to test this phenomenon, we stained PS by perfusing ischemic hearts with fluorescent Annexin V (see a detailed description in the Experimental Section). As expected, our results show an increased Annexin V staining in the infarcted tissue compared to the healthy one ( Figure 3c). Green areas in Figure 3c are due to the fluorescent Annexin V staining, while blue dots represent the cells nuclei due to 4′,6-diamidino-2-phenylindole (DAPI) staining.
Likewise, previous works postulated the overexpression of ICAM-1 in myocardial tissue after an ischemic event. In order to confirm such overexpression, we measured by Western Blot analysis the ICAM-1 protein content in control hearts and in hearts subjected to local ischemia-reperfusion (IR) for 15, 30, 45, or 60 min. A detailed description of the experimental procedure can be found in the Experimental Section. Data included in Figure 3d reveal a slight, but significant, overexpression of ICAM-1 in ischemic hearts compared to the controls. The Western blot images included in Figure 3d correspond to representative examples extracted from the complete assay that is provided in the Supporting Information ( Figure S3, Supporting Information). Furthermore, ICAM-1 overexpression was also assessed by immunofluorescence in slices of hearts subjected to 60 min of focal ischemia, showing increased ICAM-1 staining in ischemic tissue compared to nonischemic tissue ( Figure 3e). The slight overexpression of ICAM-1 detected by Western blot may be explained by the short duration of the ischemic event, since it is reported that ICAM-1 levels increase slowly in the damaged tissue and remained elevated for several days. [22] This is at variance to the case of the AT1R protein, an excellent label of damaged cardiovascular tissues due to its high overexpression after acute myocardial infarction. Previous works have reported an almost linear increment of AT1R in infarcted hearts with the ischemia duration. [31] Although we did not find a clear evidence that longer ischemia events lead to larger overexpression of PS and ICAM-1 proteins, imaging experiments were all carried out by using hearts subjected to 60 min ischemia events.
For visualization of the damaged myocardium sites, the catheter of the intravascular OCT system was introduced into the left ventricle of the rats' heart after being subjected to the infarct, as is schematically represented in Figure 4a. Technical specifications of the OCT system used in this work are given in the Experimental Section. Under the used experimental arrangement, the cross-sectional images account for the OCT signal provided by both healthy and infarcted tissues. In presence of functionalized GNSs at the damaged myocardium tissues, the OCT intensity is expected to increase due to their enhanced backscattering crosssection ( Figure 4b).

Figure 5a
includes the cross-sectional OCT images of a control heart subjected to a 60 min ischemia. The control case corresponds to the infarcted heart but just perfused with vehicle (i.e., with a solution without GNSs). The area affected by the ligation of the anterior descending coronary artery is indicated as "infarcted," whereas the myocardium tissue not irrigated by the occluded artery is indicated as "healthy." In the control case, the OCT brightness obtained in the infarcted and healthy areas seems to be very similar. The comparable OCT signal obtained from healthy and damaged myocardium in absence of GNSs is further evidenced in Figure 5b, that shows in-depth OCT intensity averaged profiles obtained in both the healthy and infarcted tissues. We should mention here that these profiles correspond to an average over ten individual radial intensity depth profiles. In order to quantify the total OCT intensity in both healthy and infarcted myocardium, we define the integrated OCT signal as where I OCT (z) is the OCT intensity at depth z (z ranging from 0 up to 2 mm). In the control heart (i.e., without post-infarct GNSs perfusion), we have estimated that the integrated OCT signals at the healthy and infarcted tissues are I c OCT (health) = 0.80 ± 0.07 and I c OCT (Inf ) = 0.76 ± 0.04, respectively (Figure 6a). In I c OCT (inf ) For the control heart (i.e., without any contrast agent), we have obtained ΔI norm,c OCT ≈ −4.9% (Figure 6b). Thus, the careful analysis of depth profiles reveals that the OCT intensity is slightly larger in the healthy tissues. In absence of any exogenous contrast agent, the OCT signal provided by a given tissue depends on its scattering and extinction properties. Therefore, experimental data suggest that the backscattering coefficient of the myocardium decreases due to the ischemic event. This is, indeed, in agreement with previous results published by Abookasis et al. who found that the wavelength-dependence of the reduced scattering (i.e., scatter power) of brain tissues decreases significantly after an ischemic event. [32] Not only that, our findings are also in good agreement with the results published by Akter and co-workers who, using a single-reflectance fiber probe, found that the scattering coefficient of the liver decreased during an ischemic event. [33] The small difference between the OCT integrated signal from healthy and infarcted tissues clearly indicates that detection of infarcted sites by OCT requires the use of contrast agents that would enhance the OCT signal difference between infarcted and healthy tissues.
As a second control, the accumulation of nonfunctionalized GNSs has been tested. For this purpose, a heart subjected to a lipoic acid-coated GNSs dispersion perfusion has been imaged. Lipoic acid does not provide a specific functionalization besides enhancing the aqueous colloidal stability of GNSs. Figure 5c,d shows the cross-sectional image and in-depth averaged intensity profile of a heart after perfusion with lipoic acid-coated GNSs, respectively. Further analysis of these profiles verifies the slightly larger scattering of healthy tissue seen in the control heart (vehicle solution) as also manifested in Figure 6a (please remember that this larger scattering is due to the tissue scattering but not to the contrast). In particular, I LA OCT (inf ) = 0.73 ± 0.02 and I LA OCT (health) = 0.80 ± 0.01 (LA stands for lipoic acid), which leads to a normalized OCT contrast of ΔI norm, LA OCT ≈ −9% (Figure 6b), as defined in expression (2). Figure 5e shows the cross-sectional image of an infarcted heart subjected to a perfusion with anti-PS functionalized GNSs. In this case, the infarcted tissue shows a larger OCT signal than the healthy one, indicating an enhanced backscattering signal of IR light. This is further evidenced in Figure 5f that includes the OCT intensity profiles as obtained from both healthy and infarcted tissues (profiles have been obtained by averaging the OCT signal along ten radial scans). We attribute the larger OCT signal to the presence of anti-PS functionalized GNSs and to their high scattering cross-section at the OCT wavelength. In this case, the OCT signal accounts for the backscattering signal generated by tissues plus that generated by GNSs adhered to the damaged cardiomyocytes. Such adhesion reveals the cell damage (apoptosis) caused by the 60 min ischemic event. Analysis of the profiles given in Figure 5d leads to I PS OCT (inf ) = 1 ± 0.03 and I PS OCT (health) = 0.89 ± 0.04, where the superscript PS stands for the anti-PS functionalization (see Figure 6a). The normalized OCT contrast is ΔI norm, PS OCT ≈ +13%. Note that the presence of the GNSs at the damaged tissues has turned the normalized OCT contrast from negative (for the vehicle and lipoic acid-coated GNSs solutions) to positive (in the heart perfused with anti-PS GNSs).
Finally, Figure 5g corresponds to the OCT cross-sectional images of an infarcted heart after being subjected to a reperfusion with anti-ICAM-1 functionalized GNSs. Figure 5h shows the averaged in-depth profiles of the OCT intensity as obtained from both healthy and infarcted tissues. Again, the perfusion with functionalized GNSs provides a clear enhancement of the OCT signal generated at the infarcted myocardial tissues sites in respect to the healthy sites. As a matter of fact, the analysis of the experimental data included in Figure 5g leads to I IC OCT (inf ) = 1.00 ± 0.02 and I IC OCT (health) = 0.82 ± 0.01, where the superscript IC stands for the anti-ICAM-1 functionalization (see Figure 6a). The normalized OCT contrast, as defined in expression (2), is ΔI norm,IC OCT ≈ +22% (Figure 6b). Consequently, it seems that, in the infarcted myocardium, both the anti-ICAM-1 and anti-PS functionalization produce an enhancement in the OCT signal that is clearly detectable by OCT.
Although further experiments, including in vivo, would be necessary to verify this conclusion, it seems that targeting inflammation could be as good as targeting apoptosis for infarcted myocardium imaging at the molecular level. Figure 6a,b demonstrates how intensity-based analysis of OCT images can be used for detection of ischemic tissues by using GNSs as contrast agents. However, the presence of GNSs in a given necrotic area should not only produce an intensity enhancement but also a reduction in the light penetration depth. In fact, GNSs do not only scatter but also absorb the OCT IR laser radiation. Thus, the presence of GNSs in a tissue does not only increase its effective backscattering cross-section but also its effective extinction coefficient ( eff ext ) and so, reducing the light penetration depth. In a first-order approximation, the depth dependence of the OCT signal is given by where d eff = 1∕ eff ext is the effective penetration length. Thus, the presence of GNSs should cause a reduction in d eff . The penetration depths of OCT radiation in healthy and infarcted tissues in the four cases under study were calculated by fitting the I OCT (z) experimental profile data to expression (3). In all the cases, the ischemic process leads to a decrease in d eff when compared to the values of the healthy tissue, as expected. Therefore, the change in d eff can be used as a complementary indicator of the infarcted myocardium. As for the normalized OCT contrast (ΔI norm OCT ), we can now define the normalized change in the OCT penetration length From data given in Figure 6c (Figure 6d). This enhancement also indicates the presence of GNSs in the infarcted tissues, this fact being in good agreement with the results obtained by intensity-analysis of OCT images.
Therefore, data included in Figure 6 indicate that detection of infarcted myocardium using functionalized GNSs as contrast agents can be achieved either by intensity analysis or by the determination of the effective penetration length of OCT radiation.
Finally, it should be noted that efficient targeting of the infarcted tissues at the in vivo level would, in principle, benefit long circulation times that are not expected for our over 200 nm gold nanoparticles. In any case, we would recall that recent results have also demonstrated how efficient in vivo imaging of infarcted heart is possible by using nanoparticles with low circulation times. [31b]

Conclusion
In summary, the ability of biofunctionalized gold nanoshells for imaging of the infarcted myocardium has been explored. Two different strategies were followed: targeting inflammation and apoptosis, both taking place simultaneously in cardiomyocytes after an ischemic event. For that purpose, gold nanoshells have been properly functionalized to provide them with specific affinity to cardiomyocytes overexpressing either the ICAM-1 or the PS enzyme. Ex vivo results in rat hearts demonstrate the ability of both functionalizations to induce selective accumulation of gold nanoshells at damaged myocardial tissues. Such selective accumulation causes an enhancement in the backscattered light efficiency and, therefore, makes the infarcted tissue visible by IR intravascular OCT. Visualization of infarcted myocardial tissues is possible by two different approaches: either by the analysis of the total OCT signal or by the determination of the penetration length of OCT radiation into tissues.
These results pioneers the first demonstration of molecular imaging of infarcted hearts by combining an operational clinical technique (intravascular OCT) with optical nanoprobes currently undergoing clinical testing, such as gold nanoshells. This work constitutes the first step toward fast and accurate diagnosis of infarcted hearts at molecular level, by using nonionizing (safe) radiation and biocompatible gold nanoparticles. Once the potential of functionalized plasmonic nanoparticles for molecular imaging of infracted myocardium is demonstrated, their translation to the in vivo and to the clinics requires a comprehensive study on the toxicity and stability of these > 200 nm nanoparticles.

Experimental Section
Anti-PS Functionalization of Gold Nanoshells: For the anti-PS biofunctionalization, GNSs dispersions were employed in phosphate buffered saline (PBS) at an initial concentration of 0.05 mg mL −1 (8.5 × 108 particles mL −1 ), carboxyl thiolated PEG (CT-PEG 12) from Thermo Scientific with a MW 634.77 and stock concentration 0.105 m, and PEG was dissolved using 1.5 mL MilliQ-grade water (from now on just water). The 1-ethyl-3-(3ʹ-dimethylaminopropyl)carbodiimide (EDAC) buffer was prepared using 2% w/v 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) and 3% w/v N-hydroxysuccinimide (NHS) in PBS and anti-PS from Merck Millipore (1 mg mL −1 ) was used. First, the PEGylation of gold nanoshells was proceeded according to the following steps: i) 1.5 mL CT-PEG diluted in water (with a final concentration of 0.01 m) was added in a 1:1 ratio to a glass vial containing 5 mL of a GNS/PBS dispersion (0.05 mg mL −1 ) under sonication. This mixture was sonicated for 5 min, and then magnetically stirred for 1 h at room temperature with a speed of about 760 rpm. ii) The PEGylated GNSs were transferred to an Eppendorf tube and centrifuged at 5600 rpm for 10 min. Then the supernatant was discarded, the pellet was washed with water, and redispersed under sonication. The wash cycle was performed three times to remove the excess of PEG. iii) The final pellet was redispersed using 5 mL of Hank's balanced salt solution. Then 1 mL aliquot was removed for characterization by FTIR and DLS.
After PEGylation, the conjugation of anti-PS antibody (aPS) to the PEGgold nanoshells was proceeded using EDC according to the following steps: i) In a glass vial under a sterile fume hood, 4 mL of the GNS-PEG was mixed with 400 µL EDAC buffer. 40 µL of anti-PS solution was slowly added to this mix (1:0.1:0.01 ratio). ii) The dispersion was magnetically stirred at room temperature for 2 h at a speed close to 760 rpm. iii) After 2 h, the reaction was quenched with 80 µL 1 m hydroxylamine for 10 min. iv) The solution was transferred to 1 mL eppendorf tubes and centrifuged for 10 min at 5600 rpm. The supernatant was discarded, the pellet was washed with PBS, and redispersed under sonication. The washing cycle was repeated three times. The final pellet was redispersed in a total of 5 mL PBS and stored at 4°C. v) While scaling up the production of GNS-PEG-anti-PS, the same ratios of reagents as mentioned above were used.
Anti-ICAM-1 Functionalization of Gold Nanoshells: The commercial dispersion of GNSs coated with lipoic acid (10 mL, 0.5 mg) in PBS was centrifuged (4000 rcf, 30 min) and washed twice with DMSO (1.5 mL) and then carefully dispersed in DMSO (1.5 mL). NHS (5.66 mg, 49.18 µmol), followed by EDC (5.98 mg, 38.55 µmol) was added to the mixture and then careful sonication and stirring were alternated for 30 min. Afterward, the dispersion was centrifuged (4000, 20 min), washed twice with DMSO (1.5 mL), and then the supernatant solvent was removed and the activated GNSs were dispersed in borate buffer (1.5 mL, pH 8) and sonicated. The cLABL peptide (cyclo(1,12)PenITDGEATDSGC 0.33 mg, 0.28 µmol) was added and, after sonication for 30 min, the mixture was stirred overnight before being quenched with a few drops of 1 m glycine and stirred for another 30 min. Finally, the mixture was centrifuged (4000 rcf, 30 min), washed twice with PBS (2 and 1.5 mL), and then carefully redispersed in PBS (10 mL).
Langendorff Model of an Infarcted Heart: Male Sprague-Dawley rats (300 g body weight) were fed ad libitum with a standard chow and housed in quarters with a 12 h light cycle and under controlled conditions of humidity (50-60%) and temperature (22-24°C). All the experiments were conducted according to the European Union Legislation (Directive 2010/63/UE) and with the approval of the Animal Care and Use Committee of the Autonomous University of Madrid (approval number CEI-83-1537). The hearts were removed from rats anesthetized with intraperitoneal injection of sodium pentobarbital (100 mg kg −1 ) followed by an intravenous injection of heparin (1000 IU). Hearts were cannulated through the aorta and subjected to retrograde perfusion (11-15 mL min −1 , 70 mmHg) with Krebs-Henseleit buffer (95% oxygen and 5% carbon dioxide) (pH 7.3). Coronary perfusion and left ventricular pressures were measured through a transducer laterally connected to the perfusion cannula and through a latex balloon inserted in the left ventricle, respectively. After 30 min of equilibration, partial ischemia was induced by ligation of the left anterior descending coronary artery. The artery was kept occluded for 15, 30, 45, or 60 min. After ischemia, the ligation was released, and the heart was subjected to 30 min of reperfusion. Finally, the perfusion system was modified to a closed circuit, in which a dispersion containing 1 mL of a colloidal dispersion of GNSs in PBS (0.035 mg mL −1 ) was kept circulating with a residual volume of 9 mL for another 30 min. A control heart was also studied for the purpose of comparison. In this case, the perfusion was performed with a vehicle solution (i.e., without GNSs). Importantly, all animals used in this study were male as the estrous phase in female rats might affect the severity of ischemiareperfusion. Indeed, it is reported that female gender favorably influences the remodeling and the adaptive response of the heart to myocardial infarction.
Tetrazolium Staining: After 60 min of ischemia and 45 of reperfusion in the Langendorff system, hearts were frozen for 1 h at −20°C. Afterward, hearts were transversely cut into 1 mm slices and incubated with triphenyl tetrazolium chloride (T-8877, Sigma-Aldrich, St. Louis, MO, USA) 1% in PBS for 25 min at 37°C. The slices were then fixed in 4% paraformaldehyde (PFA) overnight at 4°C. Images were acquired using a Stereo Microscope (SMZ645, Nikon, Tokio, Japan).
Determination of ICAM-1 Protein Content in Infarcted Hearts: 100 mg of cardiac tissue was homogenized using radio immunoprecipitation assay buffer. After centrifugation (12 000 rpm, 4°C, 20 min), the supernatant was collected and the total protein content was determined by the Bradford assay (Sigma-Aldrich, St. Louis, MO, USA). For each protein determination, 100 µg of total protein were loaded in each well and subjected to electrophoresis using resolving acrylamide (8-12%) sodium dodecyl sulfate gels (Bio-Rad, Hércules, CA, USA). Afterward, proteins were transferred to polyvinyl difluoride membranes (Bio-Rad, Hércules, CA, USA). Transfer efficiency was determined by Ponceau red dyeing (Sigma-Aldrich, St. Louis, MO, USA). Membranes were then blocked with tris-buffered saline containing 5% w/v nonfat dried milk and incubated with the primary antibody against ICAM-1 (Abcam #ab-33894; 1:1000). Membranes were subsequently washed and incubated with the secondary antibody conjugated with peroxidase (1:2000; Pierce, Rockford, IL, USA). Peroxidase activity was visualized by chemiluminescence and quantified by densitometry using BioRad Molecular Imager ChemiDoc XRS System (Hércules, CA, USA). Finally, in order to normalize each sample for gel-loading variability, membranes were incubated with a primary antibody against the constitutive protein GAPDH (1:1000; Ambion Life Technologies, Waltham, MA, USA). For each sample, the relative protein expression levels were calculated in relation to protein expression levels in nonischemic hearts. As shown in Figure S3 in the Supporting Information, all samples were run in the same assay.
Determination of PS by Staining with Fluorescent Annexin V: After 60 min of ischemia in the Langendorff system, hearts were perfused in a close circuit with Annexin V-iFluor 488 (ab219904, ABCAM, Cambridge, UK) diluted 1:2 in Annexin V buffer (10 × 10 −3 m HEPES, 140 × 10 −3 m NaCl, 2.5 × 10 −3 m CaCl 2 , pH 7,4) and 1:50 in 10 mL of Krebs buffer. After 45 min of perfusion with Annexin V, hearts were collected and fixed in 4% PFA overnight. Afterward the tissue was dehydrated, embedded in paraffin wax, and cut into 5 µm sections using a microtome. Sections were then displayed in superfrost slides, deparanized with xylol, rehydrated and washed with distilled water. Finally, sections were incubated with DAPI 1/1000 for 5 min and mounted using ProLong Gold (Thermo Fisher Scientific, Hampton, NA, USA).
Images were acquired using a Nikon Eclipse Ti2-U microscope with 40 × 0.65 NA objective (Nikon, Japan).
Determination of ICAM-1 by Immunofluorescence: After 60 min of ischemia and 45 of reperfusion in the Langendorff system, hearts were fixed overnight in 4% PFA. Afterward they were dehydrated, embedded in paraffin wax, and cut into 5 µm sections using a microtome. Sections were then displayed in superfrost slides, deparanized with xylol, rehydrated and washed in distilled water. Slices were then heated in citrate buffer pH 6 for antigen retrieval. Several washings were performed before blocking for 30 min with 5% bovine serum albumin in PBS and incubating with the primary antibody for ICAM-1 (ab206398, ABCAM, Cambridge, UK) 1:100 at 4°C overnight. Slides were then washed and incubated for 1 h with the fluorescent secondary antibody F488 (111-547-003, Jackson ImmunoResearch Europe Ltd., Cambridge, UK) diluted 1:300 in PBS. Nuclei were counterstained with DAPI 1/1000 for 5 min. Finally, slices were mounted whit Pro-Long Gold (Thermo Fisher Scientific, Hampton, NA, USA).
Images were acquired using a Leica SP5 confocal microscope fitted with a 40 × 1.25 NA objective (Wetzlar, Germany). Z-stacks were taken every 2 µm. Maximal projections of images were 3D reconstructed in wholemount views using FIJI for Windows 36bit (NIH, Bethesda, MA, USA).
OCT: The OCT system used all along this work was a commercially available cardiovascular OCT (CV-OCT) imaging system model Dragon-Fly (St. Jude Medical, St. Paul, MN, USA). Details about this system can be found elsewhere. [34] The CV-OCT system incorporated a compact superluminescent diode operating at a central wavelength of 1320 nm with a bandwidth of 200 nm. The IR radiation was coupled to a single mode fiber incorporated inside a 0.9 mm diameter catheter. The output end of the fiber was connected to a rotating reflector that deviated and scanned the 1320 nm radiation in an orthogonal plane in respect to the fiber. The reflector also collected the backscattered signal and coupled it into the single mode fiber. The single-mode fiber was optically connected to an interferometer working in the frequency domain, so that it could reconstruct cross-sectional images at the position of the reflector. The axial resolution of CV-OCT system was ≈15 µm, with a penetration length larger than 3 mm.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.