In Vitro MPIO Phantoms and MPIO Relaxivity

To estimate MPIO-induced susceptibility properties for the establishment of the in vivo experimental protocols, we prepared in vitro phantoms using different MPIO concentrations and quantified the MPIO-induced signal extinction in MRI via relaxometry according to the literature.¹⁸ A detailed description of the experimental setup and the MRI data acquisition and quantification can be found in Figure 1 and in the Methods section of the online Data Supplement.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. A, MPIO relaxivity demonstrates T2 rapid acquisition with relaxation enhancement images of the MPIO dilution series in tubes acquired at a TE of 15 ms. B, A gradient-echo image series used for r₂* determination with a TE of 3.5 ms; MPIOs typically cause a negative contrast, which can be recognized in a dose-dependent manner. The number of MPIOs and the concentration diluted are given, as well as below the images of the relative signal intensities compared with the tube containing no MPIOs. The relaxation rates R₂ (C) and R₂* (D), dependent on the contrast agent concentration, were fitted linearly (solid line) to gain the independent relaxivities r₂ and r₂*. The dashed lines depict the upper and lower 95% confidence limits.

Construction of the MRI Contrast Agent Specific for Activated Platelets

For construction of the contrast agent, autofluorescent cobalt-functionalized MPIOs (size, 1 μm) were conjugated to the histidine tag of the LIBS/control single-chain antibody following the protocol of the manufacturer (Dynal Biotech, Oslo, Norway). The detailed protocol is described in the supplementary Methods section.

Cloning of Truncated Mouse scuPA^LMW and Expression/Purification of Mouse scuPA^LMW in Drosophila S2 Cells

A detailed description of cloning, expression, and purification of low-molecular-weight single-chain urokinase-type plasminogen activator (scuPA^LMW) is given in the supplementary Methods section.

Adoption of the Carotid Artery Model for Reproducible Wall-Adherent Thrombosis

To obtain a semiocclusive platelet-derived thrombosis of the carotid artery, we first had to adopt the well-established model of ferric chloride-induced carotid artery thrombosis to obtain a reproducible semiocclusive thrombosis.^19,20 This was important to guarantee blood circulation and therefore contrast agent flow over the surface of the wall-adherent thrombus. Sample size was chosen according to our previous experience with the mouse carotid thrombosis model.^21,22 Ten- to 11-week-old male C57BL/6 mice weighing 24±0.2 g (Charles River, Sulzfeld, Germany) were used. Care and use of laboratory animals followed the national guidelines and were approved by the institutional animal care and ethics committees of the University of Freiburg (Freiburg, Germany). Mice were anesthetized by intraperitoneal injection of ketamine (100 mg/kg body weight) and xylazine (5 mg/kg body weight) and were then placed under a dissecting microscope. An incision of the skin was made superficially to the right common carotid artery. The fascia was bluntly dissected, and a segment of the right common carotid artery was exposed. Thrombosis was induced by applying a piece of filter paper (1×2 mm, GB003, Schleicher & Schuell, Dassel, Germany) saturated with ferric chloride (3% to 10% solution, Sigma Chemical Co, Munich, Germany) under the right carotid artery, which was removed after 3 minutes. After 1 hour, animals were terminally anesthetized with ketamine and xylazine and flushed with saline via the left ventricle; then, the injured carotid artery was removed.

In Vivo Experimental and MRI Protocol of Contrast Agent Injection and Thrombolysis

Mice were consecutively and randomly assigned to either the LIBS MPIO or the control MPIO group before the MRI was performed. After the wall-adherent thrombosis was induced with ferric chloride as described above, animals were transferred to the MRI system (94/20 Bruker BioSpec, Bruker, Germany). A venous catheter, connected with a 1-mL syringe containing 0.9% saline via a 120-cm-long tube (Portex, Smiths Medical International, Walford, UK), was placed in the tail vein. The animals were connected to an ECG and breathing-rate monitor and placed in the animal bed. Vital signs were monitored throughout the entire experiment, and body temperature was supported by a warm-water tube integrated in the animal bed. Anesthesia was continuously switched from ketamine to 1% to 1.6% isoflurane in O₂ starting with a heart rate rising above 190 bpm and maintaining a breathing rate of 60±20 breaths per minute. During the complete measurement, mice were placed in the center of a quadrature whole-body birdcage resonator (35-mm inner diameter). Imaging consisted of a pilot scan with 3 orthogonal slices followed by a respiration-gated coronal 2-dimensional gradient-echo sequence oriented parallel to the esophagus (Figure 2A) with an echo time (TE) of 4.4 ms, a repetition time (TR) of 250 ms, a flip angle of 40°, and a field of view of 25×25 mm. Via the coronal images, a volume (27×27×7.5 mm), including the bifurcation and the upstream carotid arteries beyond the lesion, was planned for the 3-dimensional (3D) gradient-echo sequence, monitoring the wall-adherent thrombosis. This 3D fast low-angle shot (FLASH) sequence had a TE of 2.8 ms, a TR of 20 ms, a bandwidth of 55 kHz, and an asymmetric echo with an echo position of 25%; a flip angle of 15° was chosen to gain good contrast between blood and surrounding tissue. With a matrix of 256×256×96, a resolution of 105×105×78 μm per pixel was achieved at a total acquisition time of 12 minutes 17 seconds.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. A and B, Transversal histology sections of injured carotid arteries stained for platelets with anti-CD41. Histology after in vivo MRI demonstrates wall-adherent nonocclusive thrombosis after ferric chloride treatment (A) and only a small remaining thrombus layer after thrombolysis (B). When the thrombosis is quantified in relation to the total vessel lumen, comparable thrombosis is visible for LIBS MPIO/control MPIO, which is significantly higher than the remaining thrombus after human/mouse urokinase treatment (C).

After a baseline 3D volume was acquired, mice were injected via the tail vein catheter with either the LIBS MPIO or control MPIO contrast agent (4×10⁸ MPIOs) in a total volume of 200 μL saline; we previously used this dose and volume of MPIOs successfully in a study of in vivo vascular cell adhesion molecule-1 imaging.¹⁵ The 3D FLASH sequence was then repeated with equal parameters and geometry. For the LIBS MPIO and control MPIO groups, MRI scans were performed continuously for up to 108±5 minutes.

To observe MPIO-induced signal void and to track platelet dynamics in vivo, we performed thrombolysis in a second cohort of LIBS MPIO-injected animals by injecting either human urokinase (Medac, Wedel, Germany) or recombinant mouse urokinase. Therefore, after the strongest LIBS MPIO-induced signal extinction was observed by MRI, either 50 000 IU human urokinase or 1200 IU mouse urokinase was injected in a total volume of 250 μL via the tail vein catheter, and repetitive MRI sequences were performed.

Quantification of MPIO-Induced Signal Void

Image evaluation was Matlab based. For each time point, we calculated the quotient of mean signal intensity of the ROI in the thrombotic vessel divided by the mean of the region of interest in the contralateral, nonthrombotic vessel. These quotients were then normalized on the baseline 3D FLASH starting value, gaining a time series of the fractional amount of remaining signal after contrast agent injection. Further descriptions are given in the supplementary Methods section.

LIBS MPIO Binding to Old Thrombi

To evaluate whether LIBS MPIO can be used to detect old platelet-containing thrombi, we induced wall-adherent thrombi in the carotid artery of mice using the above protocol and injected LIBS MPIO either 24 hours, 48 hours, or 7 days after the injury (n=3 per group). Two hours after injection, animals were killed, carotid arteries were removed, and tissue was stained for platelets as described later. Thrombus size as a percentage of total vessel lumen was quantified for each group, as well as the number of MPIOs averaged per section.

Tissue Harvesting and Histology

After in vivo MRI was performed, animals were deeply anesthetized with ketamine and xylazine. Transcardiac perfusions via the left ventricle were carried out with saline. The injured carotid was removed, embedded in optical coherence tomography TissueTec (Sakura Finetec, Loeterwoude, the Netherlands), and frozen for histology.

Although thrombus weight would be a direct measure of thrombus size, the small size and the difficulty of reliably removing the thrombus in toto did not allow accurate quantification. For the detection and quantification of wall-adherent thrombosis by histology, mouse platelets were detected with rat anti-mouse GPIIb (CD41) polyclonal antibody (Clone MWReg30, GeneTex, Inc, San Antonio, Tex); primary antibody was detected with a rabbit anti-rat biotinylated secondary antibody (Vectastain ABC-AP Kit, Vector, Loerrach, Germany) and VectorRed (Vector, Germany). For each animal, 12 representative sections were chosen, and either the degree of thrombosis was quantified in percent of the total vessel lumen or the thrombus area was measured (in micrometers squared) with Axiovision Software (Carl Zeiss, Jena, Germany). For quantification of MPIOs within the wall-adherent thrombosis, the number of MPIOs was counted in the same representative sections used for the thrombosis quantification and averaged per section.

MRI and Histology of Human Eversion Endarterectomy Specimens

As proof of concept for the transfer of this technique into a human setting, we incubated human eversion endarterectomy specimens with either LIBS MPIOs or control MPIOs and performed ex vivo MRI and histology. Fresh human endarterectomy specimens of 2 patients with symptomatic carotid disease (both with sudden onset of hemiparesthesia) were transferred into PBS buffer and immediately imaged on the 9.4-T MRI system using a quadrature coil (inner diameter, 35 mm) and a high-resolution T2-weighted rapid acquisition with relaxation enhancement protocol. Imaging parameters included a 45-ms TE_eff, 6000-ms TR, 45° flip angle, TurboFactor of 8, 2 averages with an image resolution of 117 μm×117 μm×1 mm, and ≈7 minutes of acquisition time. Thereafter, specimens were incubated with either LIBS MPIOs or control MPIOs for 15 minutes under continuous rotation and extensively washed with PBS afterward. Imaging was then repeated using the above-mentioned MRI protocol. Afterward, samples were embedded in paraffin and sliced into 10-μm-thick sections. Immunohistochemistry for platelets in areas of atherosclerotic plaques and LIBS MPIO binding (as detected by MRI) was performed with a monoclonal mouse anti-human CD62 antibody (1:100, R&D Systems, Abingdon, UK) and a biotinylated goat anti-mouse IgG (Vector, Burlingame, Calif) serving as secondary antibodies. Peroxidase reaction was performed with Vectastain RTU Elite ABC reagent and Vector NovaRed (both Vector).

Statistical Methods

Relaxivities and their errors are derived via linear regression on the exponentially fitted relaxation rates. The impact of LIBS MPIOs on MRI signal void and histological MPIO counts and the influence of thrombolysis on histologically measured vessel occlusion were tested with an exact, unpaired, 2-sided permutation test. Associations between variables were investigated by correlation analyses. Test results were considered significant at values of P<0.05.

The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.

Relaxivity Measurements With MPIO Dilution Series

Relaxivity (r) measurements yield information on the contrast agent characteristics of the MPIOs. In this context, r₂* is especially of interest because it is directly responsible for the observed signal attenuation. The longitudinal relaxivity measurements of the MPIOs revealed a negligible r₁ of 0.05±0.01l mmol⁻¹ · s⁻¹. The T2-weighted dilution series images acquired at a TE of 15 ms with a spin-echo sequence (Figure 1A) and the T2* gradient-echo images acquired at a TE of 3.5 ms (Figure 1B) depict examples of the signal void induced through evenly distributed MPIOs.

The spin-echo sequence yields an r₂ of 121.2±6.8l mmol⁻¹ · s⁻¹ at an R of 0.992 (Figure 1C). The relaxivity r₂* (494.4±25.0l mmol⁻¹ · s⁻¹; R=0.994) is, as expected for the susceptibility effect induced by MPIOs, higher than r₂ (Figure 1D). Calculated with the equation above, we would expect a 30% signal void in a FLASH sequence with a TE of 3.5 ms at a concentration of 0.21 mmol/L Fe, which corresponds well with the signal values displayed in Figure 1.

Results of Establishment of Wall-Adherent Thrombosis

After performing serial histology of carotid arteries exposed to ferric chloride concentrations of 3% to 10%, we observed reproducible platelet-containing wall-adherent thrombosis partially occluding (15% to 45%) the total vessel lumen in animals exposed with 6% ferric chloride (Figure 2A). Importantly, this semiocclusive thrombosis remained unchanged 1 hour after surgery as detected by a time course in histology (data not shown), which is an important prerequisite because in vivo MRI and contrast agent injection were started about this time point. Thus, we assured blood and contrast agent flow over the surface of the thrombosis.

In Vivo MR Platelet Imaging of Thrombosis and Thrombolysis

Carotid artery injury using 6% ferric chloride exposure for 3 minutes was performed without any side effects in mice. After the mouse was placed in the MRI scanner, a first MRI sequence was performed to obtain images before contrast agent injection. Thereafter, either LIBS MPIOs or control MPIOs were injected via the tail vein catheter 77±3 minutes after the carotid artery injury. No animal showed signs of MPIO-associated toxicity. Repeated MRI sequences were obtained every 12 minutes. After harvesting the carotid arteries and performing immunohistochemistry, we selected animals with a nonocclusive wall-adherent thrombus of 15% to 40% (mean, 21.3±3.9%) for further MRI data quantification (LIBS MPIOs, n=8; control MPIOs, n=6). The degree of thrombosis did not differ significantly between LIBS MPIOs and control MPIOs (Figure 2C).

In both groups, preinjection MRI allowed clear identification of the carotid artery by the bright-blood signal inside the vessel (Figure 3B and 3C). After contrast agent injection, an initial signal drop occurred in both groups after 12 minutes that possibly was attributable to MPIOs temporally trapped in vortexes at the thrombus. However, the visible signal decrease was stronger in LIBS MPIOs compared with control MPIOs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. In vivo MRI after carotid artery injury. A, Localizer sequence with labeled anatomic landmarks: the noninjured left carotid artery (a), trachea (b), and injured right carotid artery (c). B and C, Transverse sections demonstrate the injured right carotid artery (red circle), which is well perfused despite wall-adherent thrombosis (negative control: noninjured left carotid artery [green circle]). After injection of control MPIOs, signal intensity in both vessels is similar at 12, 48, and 72 minutes. After LIBS MPIO injection, there is increasing signal drop at 12, 24, and 72 minutes compared with preinjection images and the noninjured left carotid artery. This represents the typical susceptibility artifact caused by MPIOs in T2*-weighted MRI and therefore indicates LIBS MPIO binding.

Pharmacological thrombolysis was used to further evaluate whether LIBS MPIO-induced signal void intensity reports reliably on thrombus size, particularly size reduction. Furthermore, pharmacological thrombolysis is a good experimental model for one of the potential applications of the presented platelet-targeted imaging, which is the monitoring of the therapeutic success or failure of therapeutic approaches in thrombotic disease. Because the decrease in signal intensity in this first cohort of LIBS MPIO animals was clearly visible after 48 minutes (Figure 3), we decided to use this time point for thrombolytic therapy in the second cohort. Therefore, 48 minutes after the injection of LIBS MPIOs, urokinase injection was performed. For the MRI data evaluation, only animals with an initial LIBS MPIO-induced signal drop comparable to the first cohort of LIBS MPIO animals were included in the thrombolysis group (Figures 4 and 5⇓). Thus, we guaranteed that these animals had a degree of wall-adherent thrombosis comparable to the first cohort. After MRI data quantification in those animals, we observed a significant signal increase in the animals injected with human urokinase after 24 minutes (n=5) and in animals with mouse urokinase 36 minutes after injection of urokinase (n=5; Figures 4C and 5⇓) compared with the LIBS MPIO signal curve of the first cohort. Immunohistochemistry of animals confirmed an expected and significantly smaller thrombus after thrombolysis in such animals (Figure 2B and 2C; P<0.05). However, there was minor residual wall-adherent platelet aggregation in this group (Figure 2B), possibly explaining why the contrast effects did not dissipate completely and return to the level of the control MPIO group immediately after thrombolysis. Because signal intensity almost reached baseline values at the time point of thrombolysis in the control MPIO group and therefore the effect of thrombolysis would not have been detectable, we did not perform thrombolysis in control MPIO animals.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Treatment with intravenous mouse urokinase. Enlarged transverse sections of the injured right carotid artery. A, The vessel before injection, with a marked and progressive diminution in signal after LIBS MPIO injection between 12 and 48 minutes (B). Thrombolysis was performed after the 48-minute scan. Images obtained after intravenous application of urokinase show the reappearance of the vessel lumen signal over time indicated by (C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5. Quantification of signal voids. The quotient of signal intensity in the injured and noninjured vessel of mice has been normalized, providing a comparable start level of signal void at the time point immediately before injection (t−12 indicates 12 minutes before contrast agent injection). After contrast agent injection (t+12 to t+108 indicate 12 to 108 minutes after contrast agent injection), an initial signal drop occurred in both groups after 12 minutes, possibly attributable to MPIOs temporally trapped in vortexes at the thrombus, which returns to baseline over time (blue line). After LIBS MPIO injection, a signal decrease can be observed, which is significant compared with control MPIO, between 36 and 96 minutes after contrast agent injection (red line), and the signal returns toward control MPIO baseline values at 108 minutes. Animals with LIBS MPIO injection before thrombolysis (green and yellow lines) demonstrate a similar signal decrease after LIBS MPIO injection, which suggests a similar degree of thrombosis in those animals. After intravenous thrombolysis (black arrow at 48 minutes), human and mouse urokinase-treated animals show a signal increase toward control MPIO baseline, which is significantly different from LIBS MPIOs at 36 minutes after injection of mouse urokinase (yellow line) and at 24 minutes for human urokinase (green line).

When animals first showed signs of hemodynamic instability, the running scan was interrupted and not used for data evaluation. For later time points (t+96 minutes and t+108 minutes), we therefore have smaller numbers of animals (n=3 for control MPIO, LIBS MPIO, and human urokinase; n=4 for mouse urokinase). Although limited by the maximal time of animals in the magnetic resonance scanner, the latest time point (t+108 minutes; Figure 5) demonstrated adaption of the signal effect of animals with LIBS MPIO injection toward control MPIOs and toward a steady state.

MPIO Quantification in Carotid Artery Thrombosis by Histology and Correlation to MRI Signal

Quantifying the number of MPIOs detected in the ferric chloride-induced thrombus (Figure 6A; arrows in the inlet depict the typical appearance of MPIOs in histology of frozen sections), we found a significantly higher number in the LIBS MPIO group compared with control MPIO and the thrombolysed groups of animals (P<0.01; Figure 6B). For the control MPIO group, we observed only random, probably nonspecifically attached MPIOs in the superficial, reticular thrombus network without penetration of MPIOs in deeper layers of the thrombus. Thrombus size in a representative selection of animals was measured in percent (Figure 6C) or micrometers squared (Figure 6D) of the total vessel lumen, and correlation with MRI signal extinction was highly significant. However, measurements of the percentage of thrombus occlusion yielded clearer differences compared with thrombus area because the first were not influenced as much by deformed vessel lumen or vessel shrinkage during processing of the histology specimens. When correlating these 2 measures of thrombosis in animals representing a wide range of thrombus size (2% to 40% of total vessel lumen) with the MRI signal void, we observed a strong correlation between those 2 parameters (Figure 6C and 6D). Although the luminal thrombus surface, which determines the number of adherent targeted LIBS MPIOs, may not perfectly reflect thrombus size, it seems to provide a sufficient correlation with thrombus size. In addition, volume average errors have to be taken into account in MRI studies validated by histology. However, the numbers of MPIOs detected in the thrombus correlated significantly with MRI-induced signal void (Figure 6E). These results indicate that the MRI signal void reflects the number of bound MPIOs and the extent of the wall-adherent platelet-derived thrombosis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6. A, Immunohistochemistry of a wall-adherent thrombus in an LIBS MPIO-injected animal. In the inlet, arrows depict the typical appearance of bound MPIOs on the thrombus surface in frozen sections; the thrombus area itself appears red in immunohistochemistry. B, Quantification of the MPIO number in the wall-adherent thrombosis shows significantly higher MPIO binding for LIBS MPIOs than for control MPIO/human urokinase/mouse urokinase. MRI signal void shows significant correlations to thrombus size in LIBS MPIO-injected animals when the thrombus area is quantified in percent of total vessel lumen (C) or in thrombus area by micrometers squared (D). E, The numbers of MPIOs in animals from B show a significant correlation to MPIO-induced signal void measured by MRI.

Detection of Activated Platelets in Old Thrombi

Thrombus size was small after 24 and 48 hours, and platelet-containing thrombus almost completely disappeared after 7 days (Figure I of the online Data Supplement; thrombus size after 24 hours, 3.5±0.6%; after 48 hours, 2.2±1.0%; and after 7 days, 0.2±0.1%; P<0.05 for 24 and 48 hours versus 7 days). Despite this limitation, however, comparison of thrombus size and numbers of MPIOs in histology again revealed a significant correlation between those 2 parameters (Data Supplement Figure, B). Thus, LIBS MPIOs seem to be able to attach to older thrombi. However, further systematic large-scale time studies need to determine a potential cutoff in thrombus age in regard to LIBS MPIO binding.

Confirmation of LIBS MPIO Binding to Symptomatic Carotid Atherosclerosis in Humans

As initial evidence of the potential for LIBS MPIO imaging in humans, we performed ex vivo MRI of human endarterectomy specimens from patients with symptomatic carotid atherosclerosis. Typical MPIO-induced signal void could be imaged on the surface of an atherosclerotic plaque after incubation with LIBS MPIOs (Figure 7A) but not after incubation with control MPIOs (Figure 7B). LIBS MPIO binding as detected by MRI can be further confirmed by immunohistochemistry for platelets, which shows dense binding of LIBS MPIOs to superficial platelet adhesion/aggregation on the surface of the atherosclerotic plaque (Figure 7C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7. MRI and histology of symptomatic human carotid plaques. A, Transversal MRI sections show the eversion specimen before and after incubation with LIBS MPIOs. Black arrows depict areas of contrast agent binding within the vessel lumen on the surface of the plaque. B, No luminal binding can be observed in a plaque incubated with control MPIOs before and after contrast agent incubation. C, Immunohistochemistry for platelets demonstrates the area depicted with black arrows in A. Platelets appear red; MPIOs appear as yellow round structures (typical appearance of MPIOs in paraffin-processed tissue in contrast to their appearance in frozen-section histology), confirming binding of LIBS MPIOs to areas of platelet adhesion/aggregation.