In Vivo Imaging of Thrombin Activity in Experimental Thrombi With Thrombin-Sensitive Near-Infrared Molecular Probe
Objective— Thrombin, a serine protease, plays an important role in thrombosis as well as other cellular and developmental processes. In this study, we investigated the ability of a novel thrombin-activatable molecular probe to provide in vivo images of thrombin activity in experimental thrombi.
Methods and Results— The thrombin probe consists of a near-infrared (NIR) fluorochrome attached to a delivery vehicle via a thrombin-specific oligopeptide substrate. In human blood, endogenous thrombin activated the thrombin probe and increased the fluorescence signal by 18-fold (P=0.008). Hirudin, a specific thrombin inhibitor, suppressed probe activation by 82% (P=0.007). Imaging of in vivo thrombin activity was then investigated in acute experimental murine thrombosis models up to 12 hours. After systemic thrombin probe injection, focal NIR fluorescence signal enhancement was rapidly detected within acute and subacute thrombi. In contrast, no thrombosis signal enhancement was seen in similar experiments with a control NIR fluorochrome.
Conclusions— Thrombin activity can be imaged in vivo by using a novel thrombin-activatable and thrombin-specific NIR molecular probe. The thrombin probe could enhance the understanding of the role of thrombin in thrombogenesis and other homeostatic and pathological conditions.
Thrombin, a serine protease, is an important enzyme in a wide array of normal and pathological biological processes, including thrombogenesis,1 tumor invasion,2 embryogenesis,2,3⇓ angiogenesis,4 and tissue injury.5 Typically, detection of in vivo thrombin activity is made by assaying thrombin-associated activation molecules from circulating blood.6 Although such measurements provide evidence of systemic thrombin activity, less information is available regarding local thrombin activity within normal and diseased tissues. The ability to locally image thrombin activity in vivo could provide new insight into the effects of thrombin in a range of homeostatic and pathological conditions and allow more precise assessment of pharmacological therapies directed against thrombin.
See page 1753
With recent advances in near-infrared (NIR)-activatable fluorescent probe technology, in vivo imaging of protease activity is now possible.7–9⇓⇓ The activatable probes consist of NIR fluorochromes linked to a delivery vehicle via specific peptide sequences that serve as a substrate for the protease of interest. The proximity of individual fluorochromes results in quenching and suppression of the baseline NIR fluorescence (NIRF) signal. In the presence of the targeted proteolytic enzyme, the peptide substrate is cleaved, resulting in separation of the fluorochromes from the delivery vehicle. The fluorochromes consequently cease their quenching activity, increasing the NIRF signal by up to several 100-fold; thus, they serve as “molecular beacons.” Varieties of sensitive optical imaging systems are available for the detection of these molecular beacons in vivo, including surface-weighted fluorescence reflectance imaging,10 intravital fluorescence microscopy, epifluorescence (confocal and multiphoton), and fluorescence-mediated tomography (FMT). The latter in particular is of interest because the method is quantitative and capable of sensing fluorescence at depths of 7 to 14 cm11 and has recently been used to resolve protease activity in deep tissues in vivo.12
To determine whether this approach could be used to image thrombin activity in vivo, we recently synthesized a thrombin-activatable NIR molecular beacon and characterized its chemical behavior in vitro by using exogenous thrombin.13 In this present study, we first examine the specificity of the thrombin probe for endogenous thrombin within blood. We then investigate the ability of the thrombin probe to provide in vivo images of thrombin activity within experimental thrombi.
Synthesis of the Thrombin-Sensitive and Control Probes
The chemical synthesis of the fully assembled thrombin-sensitive NIRF probe and of a corresponding control probe has been presented elsewhere.13 Briefly, the probe consisted of a human thrombin-cleavable peptide ligand (Gly-d-Phe-Pip-Arg-Ser-Gly-Gly-Gly-Gly-Lys-Cys-NH2, where Pip indicates pipecolic acid)14 that contained an N-terminal NIR fluorochrome, Cy 5.5 (absorption maximum 675 nm, emission maximum 694 nm, Amersham-Pharmacia). The C-terminal of the peptide ligand was attached to a biocompatible long circulating delivery vehicle (molecular mass 500 kDa) consisting of a partially (≈25%) PEGylated poly-l-lysine.13 On average, each delivery molecule contained 23 reporter probes, resulting in efficient quenching of fluorescence. A control probe was synthesized identically to the thrombin probe except for a single amino acid substitution at the P1′ position, Gly-d-Phe-Pip-Arg-Pro-Gly-Gly-Gly-Gly-Lys-Cys-NH2, rendering it far less cleavable by thrombin.15 A second negative control probe consisted of the free Cy 5.5 fluorochrome alone.
In Vitro Thrombin Probe Experiments Using Human Blood
To demonstrate thrombin probe activation by endogenous human thrombin, blood was obtained from a healthy male donor (aged 33 years) after informed consent and in accordance with a human studies protocol approved by the Institutional Review Board at Massachusetts General Hospital. In a clear-bottomed 96-well plate (Corning Costar), 0.2 nmol thrombin probe or control probe was pipetted into separate wells, with or without 25 μg hirudin, a direct thrombin inhibitor (Aventis). Blood was then added to each well in 100-μL aliquots. Experiments (n=4) were performed at room temperature with the use of a reflectance NIRF imaging system.10 Each plate was placed in the center of the imaging field, and serial light images (exposure time 75 ms) and NIRF images (exposure time 30 seconds) were obtained over a 24-hour period.
In Vivo Thrombin Probe Experiments in Murine Thrombosis Models
To determine the ability of the thrombin probe to image in vivo thrombin activity within experimental thrombi, we obtained C57BL/6 and BALB/c mice (National Cancer Institute, Bethesda, Md) to create experimental models of hematoma and thrombosis. Mice were cared for according to our institution’s animal facility guidelines. Mice were anesthetized with an intraperitoneal mixture of 100 mg/kg ketamine (Bedford Labs) and 10 mg/kg xylazine (Ampro Pharmaceutical). At the end of the experiment, mice were euthanized by cervical dislocation. All animal studies were approved by the Subcommittee on Research Animal Care at Massachusetts General Hospital.
Mouse Hematoma Model
In this model, a hematoma was created by distal tail vein amputation (n=5). After anesthesia, each mouse was positioned on its back, and the subxiphoid area was shaved to facilitate left ventricular injection. The thrombin probe (2 nmol in 200-μL volume) was then injected into the left ventricle to avoid potential probe contamination by tail vein injection. After 5 minutes of probe circulation, the distal tail vein was cut with a surgical blade. After hematoma formation, light images (exposure time 75 ms) and NIRF images (exposure time 2 minutes) were acquired at 5- to 10-minute intervals for up to 1 hour by using the reflectance imaging system described below.
Mouse Intravascular Thrombosis Model
In this model, an intravascular thrombosis was created by topical application of 10% ferric chloride (FeCl3) solution.16 After anesthesia, mice (n=3) were positioned supine, and their groins were shaved with the use of hair clippers. After a midline skin incision, the superficial femoral artery and vein were exposed by blunt dissection. Care was taken to dissect the adventitial layer from the femoral vessels, and a 2-mm×5-mm strip of filter paper (Whatman #1, Whatman Inc) soaked in 10% FeCl3 (Fisher Scientific) was then applied to the superficial femoral vessels for 3 minutes. After the experiments, the mice were euthanized and perfused with 4% formalin. After pathology confirmed consistent femoral vein thrombosis, 10 additional mice underwent the same protocol, followed by systemic injection of 2 nmol of the thrombin probe or control free fluorochrome. The protocol was first tested by preinjection of either the thrombin probe or control probe in 2 separate mice, followed by thrombus formation. To then simulate a clinically relevant scenario, probes were injected, after thrombus formation, at either the 1-hour time point (thrombin probe, n=3; control fluorochrome, n=3; acute thrombus group) or the 12-hour time point (thrombin probe, n=2; subacute thrombus group). After probe injection, the mouse was placed onto the inverted fluorescence microscope, and intravital fluorescence microscopy was performed. Serial light images (exposure time 75 ms) and NIRF images (exposure time 10 to 60 seconds) were obtained every 15 minutes for a period of 2 hours. After the imaging was performed, a segment of the femoral vessels with thrombosis was resected and snap-frozen in liquid nitrogen. Adjacent frozen sections (9-μm thickness) were obtained for correlative fluorescence microscopy and histopathologic analysis. To verify that FeCl3 did not itself activate the thrombin probe, in vitro experiments (n=4) were performed with 0.1 nmol of the probe added to wells containing either 50 μL of 10% FeCl3 solution or distilled water. Serial NIR reflectance images (exposure time 30 seconds) were obtained over 2 hours.
NIR Reflectance Imaging
Thrombi were imaged by using a custom-built optical imaging system.10 The system contained a 150-W halogen lamp with a Cy 5.5 excitation filter (610 to 650 nm, Omega Optical). The imaging area was homogeneous within ±10% over a 5×5-cm field of view. The NIRF signal was detected by a 12-bit CCD camera (Kodak) equipped with a special C-mount lens and a Cy 5.5 bandpass emission filter (680 to 720 nm, Omega Optical). Images were digitally acquired as 16-bit images. To account for light source variation across the field of view, a pure white light image (mask image) was obtained before each NIRF image. The acquired NIRF images were then divided by the mask image to correct for light variation. Although reflectance imaging is semiquantitative because of an inability to resolve light from different depths, this calibration allowed images acquired at different times to be assessed for relative changes in NIRF signal intensity.10
For visualization of intravascular thrombi, mouse femoral vessels were viewed in the phase-contrast or NIRF mode (680-nm long-pass emission filter, Omega Optical) by using an inverted epifluorescence microscope (Zeiss Axiovert) with a cooled CCD camera (Sensys, Photometrics) interfaced to a Power Macintosh G4 computer (Apple Computer). Control fluorescence images were obtained with the fluorescein channel (emission bandpass filter 515 to 565 nm, Omega Optical). Light and fluorescence images were digitally acquired as 16-bit images.
Statistical Analyses and Image Analyses
Data are presented as mean±SD. Student t tests were used for comparisons between data; all t tests were unpaired unless further specified. A value of P<0.05 was considered statistically significant.
Images acquired on the fluorescence reflectance system were mask-corrected as described above and then analyzed by using a commercially available software package (Kodak Digital 1D Software). For each image of blood clots in the well plates, a circular region of interest (ROI) defined by the well (Figure 1B) was manually traced, and this identical ROI was copied and used to measure the signal intensity from all wells within 1 image. The area of the ROI varied by <3% between images. For the in vivo hematoma model, a freehand ROI was drawn around the margins of the hematoma, defined by the strong NIRF signal (Figure 2, right). A second ROI of similar size was manually drawn on the adjacent tail to measure signal differences between the hematoma and background tissue.
Images acquired on the fluorescence microscopy system were analyzed with the use of IP Laboratory Spectrum software (version 3.5.2, Scanalytics) to compare areas of NIRF signal enhancement within venous segments. Images were then correlated by histology and fluorescence microscopy of frozen sections. The fusion images demonstrating thrombin activity were created by transferring the NIRF image to NIH Image (version 1.62, National Institutes of Health) to generate a color lookup table in a region of interest. This functional image was superimposed on the corresponding light image with the use of Photoshop software (version 5.5, Adobe Systems).
In Vitro Human Blood Thrombin Activation of Thrombin Probe
Activation of the thrombin-sensitive NIRF probe was first studied in vitro with the use of human blood (n=4 experiments). Endogenous thrombin from the blood activated the thrombin probe, and by 24 hours, the NIRF signal had increased by 18-fold (P=0.008 by paired t test, Figure 1). In contrast, the NIRF signal from the control probe (a single amino acid substitution in the thrombin probe peptide substrate) was 7-fold lower than the thrombin probe group (44.2±11.6 versus 292.7±81.2 arbitrary units, P=0.008), consistent with its known relative inability to serve as a thrombin substrate.15 Pretreatment with hirudin suppressed thrombin probe activation and the NIRF signal, further confirming the specificity of the probe for the thrombin enzyme (82.4% signal reduction at 24 hours, P=0.007). The initial NIRF signal was first seen at the margin of the semisolid blood clot, and then it progressively spread throughout the well over time. Thrombin activation of this probe in the blood clots was slower than in similar experiments performed with purified thrombin in buffer,13 possibly because of slower diffusion in the semisolid blood clot. To confirm that the detected NIRF signal was due to specific Cy 5.5 fluorescence, the emission filter was changed from a 700±20-nm bandpass filter to a 800±20-nm bandpass filter to eliminate Cy 5.5 light. This emission filter change decreased the NIRF signal by 97.2%.
In Vivo Activation of Thrombin Probe in Experimental Thrombosis
To determine whether the human-targeted thrombin probe could be activated by murine thrombin in experimental thrombosis models, we first verified probe activation by endogenous thrombin within mouse blood. In vitro experiments demonstrated substantial thrombin probe activation and suppression by hirudin (data not shown), in accord with the known high degree of conservation between human and mouse thrombin.17
No adverse effects to the animals were noted with the injection of the thrombin probe, consistent with previous experience.7 In the first in vivo thrombosis model, distal tail vein amputation resulted in the formation of a macroscopic hematoma (n=5). Activation of the thrombin probe occurred within minutes and was manifested as a bright fluorescent signal in the hematoma (Figure 2). Compared with the adjacent tissue background, the NIRF signal from the hematoma was nearly doubled (hematoma signal increase of 94±108% over the tissue signal, P=0.039 by paired t test) and remained elevated for at least 1 hour.
To further investigate thrombin activation of the probe in vivo, we next created an intravascular thrombosis in 3 mice by using a topical application of 10% FeCl3 to the superficial femoral vessels.16 Venous thrombosis was evident within minutes. The vein became larger and darker in FeCl3-soaked segments compared with control segments and was densely thrombosed with erythrocytes (Figure 3). The degree of thrombosis seen was variable in this FeCl3 model of thrombosis. In all cases, portions of the main femoral vein and smaller side branches all had occlusive thrombi. However, areas of nonocclusive thrombi were also noted within the main femoral vein (Figures 3B and 4⇓D), possibly because of larger vessel size, variable amounts of adventitia on the exposed femoral vein limiting the amount of exposure to FeCl3, or lower FeCl3 concentration at the margins of the filter paper.
We next investigated thrombin activity in the FeCl3 model by use of the thrombin probe. In all mice injected with the thrombin probe (n=6), focal NIRF was noted in multiple areas of venous microthrombi (Figures 4B through 4E). Importantly, we detected thrombin activation in thrombi at clinically relevant injection time points and did not require preinjection of the probe. The NIRF signal arose from the body and edges of occlusive thrombi, particularly within smaller venous segments, and was detectable for hours, presumably because of less washout of the cleaved fluorochrome from smaller vessels. In addition, NIR signal enhancement was also noted in areas of nonocclusive thrombi in the main femoral vein (Figure 4D). The signal was highest around the luminal and vascular margins of nonocclusive thrombi; this finding was likely due to greater local thrombin activity and/or low flow at the edge of the thrombus surface. Interestingly, we also detected thrombin activation and focal NIRF signal enhancement in subacute thrombi at the 12-hour time point (Figure 4E). The degree of NIRF signal enhancement was less than that in the acute thrombus group, presumably because of declining thrombin activity within the aging thrombus and/or decreased probe delivery due to vessel obstruction. In contrast to the thrombin probe, all mice injected with the control NIR fluorochrome (n=4) revealed minimal focal fluorescence within thrombi, consistent with its expected lack of specificity for thrombi (Figure 4F).
Fluorescence microscopy of a cross section of occlusive thrombi within venous side branches confirmed thrombin probe activation within microscopic areas of thrombosis (Figure 5). A strong fluorescence signal within the thrombus was seen in the NIRF channel, but not the FITC channel, confirming that the signal emanated from the NIR fluorochrome. The NIRF signal was fairly homogenous within occlusive branch thrombi, suggesting that the cleaved fluorochrome remained associated with the thrombus, presumably because of minimal residual flow in the small branch vessels. Finally, to confirm that thrombin probe activation was not due to FeCl3, we measured the NIRF signal in thrombin probe wells containing either FeCl3 or distilled water. Thrombin probe wells containing FeCl3 actually had a lower NIRF signal (P<0.01 versus distilled water), demonstrating that, if anything, FeCl3 inhibited the thrombin probe signal.
The present study demonstrates that thrombin, the central protease of the coagulation cascade, can be functionally imaged in vivo by using a novel NIR thrombin-activatable reporter. In vitro experiments first confirmed that the probe was specifically activated by endogenous thrombin within blood. In experimental murine thrombosis models, thrombin activation of the probe resulted in focal NIRF signal enhancement in occlusive and nonocclusive thrombi. Thrombin activity was detected within acute and subacute thrombi with the use of probe injections at clinically relevant time points, did not require preinjection of the probe, and could be rapidly detected in vivo by fluorescence reflectance imaging systems.
Although in vitro thrombin activity has been detected for many years with the use of chromogenic or fluorogenic substrates,18,19⇓ current experimental results now show that thrombin activity can be imaged in vivo by using the NIRF-activatable probe scheme previously developed in our laboratory.7–9⇓⇓ The ability to successfully image thrombin activity in vivo is likely due to several synergistic factors. First, thrombin has a high affinity for the peptide substrate on the thrombin probe, allowing rapid cleavage of the substrate and consequent cessation of quenching of the probe.13 Second, optical imaging in the NIR range is ideal for imaging through biological tissue, because water, oxyhemoglobin, and deoxyhemoglobin have local absorption minima in this range, permitting more efficient transmission of the NIRF signal.20 Furthermore, fluorescence detection systems have high intrinsic sensitivity and can detect fluorochromes at nanomolar concentrations.21 Finally, FeCl3 injury of the vascular wall is thought to acutely produce high levels of tissue factor,22 leading to thrombin generation and subsequent activation of the thrombin-sensitive probe and generation of the NIRF signal.
By demonstrating that thrombin activity can be imaged within thrombi, the thrombin probe may also serve as a targeted thrombosis-imaging agent. Thrombi can be imaged passively (eg, negative contrast enhancement on angiographic methods) or directly (eg, noncontrast MRI23–25⇓⇓ or contrast-targeted thrombosis-imaging methods).26–28⇓⇓ Other contrast-targeted methods typically rely on the binding of the contrast agent to a stable structure (such as platelets or fibrin) and, thus, may have a higher sensitivity for detecting all thrombi but, consequently, are less able to differentiate between acute versus chronic thrombi. Furthermore, because binding contrast agents detect the presence of a stable target, they are less suited to imaging enzyme activity or short-lived molecules like thrombin. In contrast, by using the thrombin-activatable probe, a single enzymatically active thrombin molecule can cleave multiple probe substrates, allowing for the measurement of thrombin activity and amplification of the NIRF signal in the relatively short life of thrombin. Furthermore, using thrombin as a molecular target for thrombosis detection allows the preferential detection of biologically active thrombi (eg, acute thrombi with high thrombin activity). The ability to distinguish biologically active thrombi is of substantial clinical importance, inasmuch as fibrinolytic resistance markedly increases with thrombus age.29
Although the in vitro blood studies demonstrated that the probe is specific for thrombin, background nonspecific protease cleavage will contribute to a small amount of the measured NIRF signal. From an imaging standpoint, a possible cause of diminished NIRF signal enhancement in the in vivo intravascular thrombosis model may have been the washout of the cleaved fluorochrome, particularly in nonocclusive thrombi, in which flow rates are likely to be higher. As opposed to thrombin, which acts in the local extracellular thrombotic milieu, other activatable probe proteases (eg, cathepsins7,8⇓ and matrix metalloproteinases9) cleave their respective probes within the internal cell environment, which may minimize NIR fluorochrome washout. Alternative models of thrombin expression may help to overcome this limitation. Finally, although the thrombin probe may be useful for imaging acute and subacute thrombi, it may be less useful for imaging older chronic thrombi that have less residual thrombin activity.
Nonetheless, current experimental results suggest that the thrombin probe could have an important role in further elucidating the role of thrombin in thrombogenesis, particularly in conjunction with FMT, a new technology that allows quantitative measurement of fluorescence activity noninvasively.12 Because NIR light travels efficiently through tissue, FMT has the ability to noninvasively detect NIRF deep within tissue and has recently been used to resolve protease activity in vivo.12 Furthermore, recent simulations demonstrate that NIR light may be able to penetrate >10 centimeters,11 suggesting that FMT could ultimately be used in human subjects. By use of fluorescence reflectance imaging and FMT, imaging of thrombin probe activation may allow for serial in vivo study of thrombin activity within pulmonary, cerebral, and intracardiac thromboemboli and may be capable of assessing thrombin activity in response to antithrombotic drug therapies. Furthermore, the thrombin probe could potentially be used to study the role of thrombin in vivo in a wide range of biological processes, including tumor metastasis,2 embryogenesis,2,3⇓ angiogenesis,4 and cellular activation.5
In conclusion, we have imaged thrombin activity in vivo in experimental thrombi by using a novel thrombin-activatable NIR molecular probe. The thrombin probe could have widespread application for in vivo imaging of thrombin activity in various aspects of biology.
This study was supported by National Institutes of Health grants R33-CA-88365 (Dr Tung), RO1-HL-67768 (Dr Gerszten), P50-CA-86355 (Dr Weissleder), and R24-CA-92782 (Dr Weissleder). Dr Jaffer was supported in part by the William A. Schreyer Fellowship, Cardiology Division, Massachusetts General Hospital. The authors would like to thank Wendy Cohen, BS, for assistance with pathology.
↵*These authors contributed equally to the present study.
Received May 10, 2002; revision accepted July 29, 2002.
- ↵Griffin CT, Srinivasan Y, Zheng YW, Huang W, Coughlin SR. A role for thrombin receptor signaling in endothelial cells during embryonic development. Science. 2001; 293: 1666–1670.
- ↵O’Reilly MS, Pirie-Shepherd S, Lane WS, Folkman J. Antiangiogenic activity of the cleaved conformation of the serpin antithrombin. Science. 1999; 285: 1926–1928.
- ↵Patterson C, Stouffer GA, Madamanchi N, Runge MS. New tricks for old dogs: nonthrombotic effects of thrombin in vessel wall biology. Circ Res. 2001; 88: 987–997.
- ↵Tung CH, Mahmood U, Bredow S, Weissleder R. In vivo imaging of proteolytic enzyme activity using a novel molecular reporter. Cancer Res. 2000; 60: 4953–4958.
- ↵Farrehi PM, Ozaki CK, Carmeliet P, Fay WP. Regulation of arterial thrombolysis by plasminogen activator inhibitor-1 in mice. Circulation. 1998; 97: 1002–1008.
- ↵Banfield DK, MacGillivray RT. Partial characterization of vertebrate prothrombin cDNAs: amplification and sequence analysis of the B chain of thrombin from nine different species. Proc Natl Acad Sci U S A. 1992; 89: 2779–2783.
- ↵Morita T, Kato H, Iwanaga S, Takada K, Kimura T. New fluorogenic substrates for alpha-thrombin, factor Xa, kallikreins, and urokinase. J Biochem (Tokyo). 1977; 82: 1495–1498.
- ↵Toschi V, Gallo R, Lettino M, Fallon JT, Gertz SD, Fernandez-Ortiz A, Chesebro JH, Badimon L, Nemerson Y, Fuster V, Badimon JJ. Tissue factor modulates the thrombogenicity of human atherosclerotic plaques. Circulation. 1997; 95: 594–599.
- ↵Johnstone MT, Botnar RM, Perez AS, Stewart R, Quist WC, Hamilton JA, Manning WJ. In vivo magnetic resonance imaging of experimental thrombosis in a rabbit model. Arterioscler Thromb Vasc Biol. 2001; 21: 1556–1560.
- ↵Lanza GM, Wallace KD, Scott MJ, Cacheris WP, Abendschein DR, Christy DH, Sharkey AM, Miller JG, Gaffney PJ, Wickline SA. A novel site-targeted ultrasonic contrast agent with broad biomedical application. Circulation. 1996; 94: 3334–3340.
- ↵Flacke S, Fischer S, Scott MJ, Fuhrhop RJ, Allen JS, McLean M, Winter P, Sicard GA, Gaffney PJ, Wickline SA, Lanza GM. Novel MRI contrast agent for molecular imaging of fibrin: implications for detecting vulnerable plaques. Circulation. 2001; 104: 1280–1285.
- ↵Robinson BR, Houng AK, Reed GL. Catalytic life of activated factor XIII in thrombi: implications for fibrinolytic resistance and thrombus aging. Circulation. 2000; 102: 1151–1157.