Aptamer-Crosslinked Microbubbles: Smart Contrast Agents for Thrombin-Activated Ultrasound Imaging

Abstract

Thrombosis, or malignant blood clotting, is associated with numerous cardiovascular diseases and cancers. This report describes a microbubble contrast agent that produces ultrasound harmonic signal only when exposed to elevated thrombin levels. Silenced initially, microbubbles activated in the presence of both thrombin-spiked and freshly clotting blood in three minutes with detection limits of 20 nM thrombin and 2 aM microbubbles.

Thrombosis, or malignant blood clot formation, has been associated with numerous cardiovascular diseases, including myocardial infarction, coronary artery disease, and deep venous thrombosis, the latter of which can develop into pulmonary embolism. The most common method of imaging thrombosis is to fill blood vessels with x-ray or ultrasound contrast media and locate clots as filling defects.[1] However, these methods do not provide biochemical information necessary to determine clot activity, which is vital for establishing a method of treatment, particularly to prevent acute active thrombosis from leading to pulmonary embolism. To analyze clot biochemistry, Weissleder, et al.[2] and Tsien, et al.[3] developed optical probes that sense thrombin through increased fluorescence emission; however, optical imaging is limited in its ability to image deep-lying blood vessels.

Ultrasound is an attractive imaging tool for detecting and monitoring cardiovascular health due to its widespread availability, safety, penetration depth, real-time image acquisition, millimeter resolution, and portability. Most importantly, ultrasound is currently the primary imaging tool for diagnosis of deep venous thrombosis. However, ultrasound is quite limited in its ability to discern abnormalities from surrounding tissue owing to the similar physical properties of soft tissues within the body.[4] A popular strategy to address this limitation is to administer an ultrasound contrast agent with a large acoustic impedance mismatch with the surrounding tissue. Although these agents allow the detection of thrombi,[1b] they cannot distinguish acute from chronic disease. To bias contrast enhancement towards sensing disease, researchers have attached ligands to ultrasound contrast to target fibrin or platelets within thrombi.[5] Such methods rely on the accumulation of microbubbles at the site of disease, tethering microbubbles by direct interaction of targeting ligands such as P-Selectin or GPIIb/GPIIIa, and they have proven to be effective at increasing the signal generated in specific areas. However, they are still limited in their ability to sense changes in biomarker concentration, and the total enhancement is theoretically limited by the surface area of the targeted tissue. The alternative method presented here seeks to improve the noninvasive detection of thrombosis by using a stimulus responsive or “smart” contrast agent that becomes active only in the presence of a biochemical environment specifically indicative of thrombus formation.[6] This strategy also reduces background signal and should even be able to image inflammation near the resolution limit of the instrument. We recently showed that DNA hybridization and dehybridization can be employed to reversibly change the way in which a microbubble responds to ultrasound.[7] Since the ultrasound pressure wave causes a lipid-shelled microbubble to undergo nonlinear size oscillations that result in the generation of nonlinear sound waves,[8] crosslinking and decrosslinking of the microbubble shell modulates the generation of these selectively-detectable harmonics.[9]

Here, we report the design, fabrication, and validation of microbubbles that only show ultrasound contrast activation at levels of thrombin associated with clot formation. First, the components of the microbubble shell were synthesized and validated for their ability to respond to thrombin. To form the shell that stabilizes the microbubbles, poly(acrylic acid) (PAA) was coupled to the amine of 1,2-distearoyl-sn-glycero-3-phosphoethanolamine (DSPE) via carbodiimide-mediated amidation, followed by a similar reaction to attach two amine terminated DNA strands (H₂N-CCAACCACAAAA and AAAACACCAACC-NH₂). Next, a DNA crosslinking strand was added with the sequence GGTTGGTGTGGTTGGTGTTTTTTTTTGTGGTTGGTGTGGTTGG. This thrombin aptamer crosslinking strand (TACS) contains a portion complementary to the polymer-DNA to induce crosslinking, as well as two flanking 15-base DNA aptamer sequences (GGTTGGTGTGGTTGG), each of which possesses a half maximal inhibitory concentration (IC₅₀) on the order of 25 nM for thrombin.[10] Nine of the fifteen aptamer bases are left unbound at the ends of the crosslinking strand so that the binding of thrombin results in the complete displacement of the polymer-DNA strands.

The activity of thrombin towards these DSPE-PAA-DNA conjugates was studied by attaching a fluorescent dye, 6-carboxyfluorescein (FAM) to the end of the phospholipid-polymer-DNA, and its corresponding quencher, 4-(4-dimethylaminophenyl) azobenzoic acid (Dabcyl) was appended to the other.[11] When separate, the FAM fluorescence emission is bright and unaffected by the presence of the Dabcyl in solution (Figure 1b). After addition of TACS, the two strands are pulled together such that the dye and quencher come into close proximity, and the fluorescence emission decreases. Addition of thrombin in a 1:1:1:1 ratio of thrombin:DSPE-PAA-DNA-FAM:DSPE-PAA-DNA-Dab:TACS results in the partial recovery of fluorescence emission, indicating that the thrombin is preferentially removing the crosslinking strands from the DNA conjugates. The incomplete recovery of fluorescence is most likely due to the equilibrium established by the PAA-DNA:TACS:thrombin system.

Figure 1.

Design of aptamer-activated microbubbles. (a) Schematic representation of fluorescence activation by thrombin binding. (b) Thrombin aptamer crosslinking strand (TACS). (c) Fluorescence emission spectra of DSPE-PAA-FAM and DSPE-PAA-Quencher (blue), after addition of TACS (red), and addition of thrombin (green) (λ_ex = 465nm). (d) Microbubble suspension after sonication. (e) Bright field and (f) fluorescence images of crosslinked bubbles. (g) Bright field and (h) fluorescence images of crosslinked bubbles after addition of thrombin. Fluorescence increases after TACS interacts with thrombin. Scale bar = 5 µm.

To form microbubbles, the DSPE-PAA-DNA conjugates were then melted with 1,2-palmitoyl-sn-glycero-3-phosphocholine (DPPC) and 1,2-palmitoyl-sn-glycero-3-phosphatic acid (DPPA) in 20 mM Tris Acetate buffered saline. This mixture was probe sonicated under a headspace of perfluorobutane gas to form microbubbles mostly in the range of 1–5 µm at a concentration of approximately 10⁸ bubbles per milliliter (Figure S1). Addition of the TACS did not affect the bubble concentration significantly. Fluorescence microscopy of the microbubble suspension revealed highly-dampened green fluorescent rings that correlated with the bright field images of the microbubble suspension. After introduction of thrombin at 100 nM, a significant increase in green fluorescence was observed on the bubble surface, indicating that the removal of the crosslinker occurred even when the DNA was incorporated into a self-assembled encapsulating shell. Thrombin-induced removal of the TACS did not significantly affect the microbubble concentration.

The effect of thrombin activation on the acoustic contrast properties of the microbubbles was studied through the use of homebuilt instrumentation designed to measure the generation of harmonic signals (Figure 2). A dilute sample of microbubbles in phosphate buffered saline contained within a submerged transfer pipette bulb in a large water tank was insonated with a 2.25 MHz single sine pulse sent through a focused single element transducer at 290 kPa. Signals generated by the oscillation of the insonated microbubbles were then collected by a 5 MHz unfocused transducer angled perpendicular to the ultrasound wave, followed by subtraction of a blank sample containing only PBS. The average nonlinear power output was determined by integrating the signal from 4.0 to 6.0 MHz, which was outside the pulse width of the original sine pulse. By comparing the power output generated around the 5 MHz peak listening frequency before and after thrombin addition, we could determine that the interaction between the TACS-microbubbles and thrombin had a significant effect on the harmonic signals generated by the microbubble, increasing the side scatter signal 10 dB over non-activated bubbles. To further confirm that this effect was specific to the thrombin-TACS interaction, we also measured the signal generated by microbubbles crosslinked with a DNA strand in which the thrombin binding aptamer sequence was replaced with a mixed sequence (GGTTGGGTGTGTGTG) with no affinity for thrombin (hereby SACS). The resulting signal from the SACS microbubbles showed little change in side scatter signal power before and after thrombin addition, demonstrating that the signal enhancement observed shows excellent specificity for thrombin. Interestingly, the power generated by the bubbles crosslinked with the SACS DNA strand prior to thrombin addition was slightly higher than the corresponding signal generated by the TACS bubbles. This can be attributed to the greater repetition of sequence motifs within the TACS strand, enabling crosslinks to be more easily formed within the encapsulating shell. In the case of the SACS strand, hybridization between the FAM/Dab strands and the crosslinker must be well-aligned, while for the TACS strands, partial hybridization may occur at the hanging aptamer tails.

Figure 2.

Characterization of the effect of thrombin on microbubble acoustic properties. (a) Signal acquired from microbubbles crosslinked with TACS before and after addition of thrombin. Curves shown are the mean of 10 sequential trials. (b) Signal acquired from microbubbles crosslinked with DNA strand containing a scrambled aptamer sequence (SACS), before and after addition of thrombin. (c) Comparison of the change in the reflected power after thrombin addition between TACS bubbles and SACS bubbles. Error bars show 95% confidence interval. (d) Schematic of home-built side-scatter measurement system.

In order to measure the dynamic detection range of these microbubbles, different concentrations of thrombin were added to microbubble suspensions and the ultrasound signal was measured as before (Figure 3a). At 10 nM, enhancement was essentially zero and comparable to that of microbubbles unexposed to thrombin (Figure 3b), while a very sharp and dramatic increase in power output was observed as concentration was increased from 10 to 25 nM. Interestingly, increasing thrombin levels beyond this inflection point had little effect on the total received signal but resulted in noticeable sharpening of the 2^nd harmonic peak. This response suggests that at lower concentrations of thrombin, TACS strands are not fully removed from the encapsulating shell, leaving the shell with regions of increased and decreased stiffness. This shell heterogeniety results in more heterogeneous bubble dynamics and thus a broadband signal. As the thrombin concentration increases, TACS strands are completely removed from the bubble shell, and full, unhindered oscillation can occur, producing very sharp harmonic peaks. It should be emphasized that since in phase inversion mode all frequencies outside the sending pulse contribute to the detected signal, broadband signal generation can produce a very highly detectable signal. Secondly, because the microbubbles can generate nonlinear signal with only partial shell softening, the received signal exhibits a nonlinear dependence on thrombin concentration, so the microbubble “turns on” at concentrations greater than about 20 nM. Thus while it is difficult to differentiate between thrombin concentrations of 100 nM or 200 nM, it is relatively easy to distinguish thrombin concentrations of 10 and 25 nM. This correlates extremely well with expected levels of thrombin in blood clots, which begin clotting at a thrombin concentrations of about 25 nM and increase further as the clot grows.[12] Thus the microbubble remains dormant under normal conditions, but when thrombin levels exceed the threshold for blood clotting, the microbubble produces a strong “on” signal. We hypothesize that activation will begin as the bubbles enter the local environment of an acute clot, and aptamer-thrombin interaction will increase as the bubbles pass through the vasculature of the clot, where thrombin concentration is expected to be highest, followed by continued activation in eddy currents on the downstream side of the clot. Enhanced signal should then be seen either within a clot, or immediately after the microbubbles clear the clot.

Figure 3.

Analysis of thrombin-dependent ultrasound signal. (a) Average signal acquired from microbubbles after addition of thrombin in concentrations from 10 nM to 250 nM, with ten trials per concentration. (b) Change in side-scatter power from 4.0–6.0 MHz after addition of varying concentrations of thrombin. Error bars show 95% confidence interval.

Activation of the thrombin-detecting microbubbles was imaged in blood using a Siemens Acuson Sequoia 512 ultrasound scanner in contrast-enhanced ultrasound mode. Microbubbles crosslinked with either TACS or SACS were diluted in EDTA-treated bovine blood to a concentration of 20 aM (10,000 bubbles per millliter) and imaged at 2.0 MHz and 140 kPa inside an acoustically transparent phantom before and after addition of 100 nM thrombin. Prior to thrombin addition neither crosslinked bubble sample showed significant signal inside the phantom walls (Figure 4). The few bright spots in these images are most likely the result of a few large outlier bubbles, as very large bubbles will scatter signal easily even with a stiffened encapsulating shell,[7] although in vivo large bubbles would be expected to be filtered by the lungs.[13] After the addition of thrombin, there was a substantial increase in harmonic signal for TACS microbubbles. There was no corresponding effect on microbubbles crosslinked by the scrambled SACS. This indicates that the activation is specific to the presence of thrombin. Brightness analysis using ImageJ showed a 10-fold increase in signal after thrombin addition (Figure 4e). This brightness value was similar to a sample containing the same concentration of PEG-DSPE stabilized bubbles, which is similar to an FDA-approved formulation and always retains shell flexibility (not shown). Unfortunately, earlier time points cannot be obtained using this method, as it is necessary to allow the microbubble suspension to settle and equilibrate in order to let the mixing currents subside. Our future investigations will work with a thrombosis flow model to further study the effects of microbubble activation immediately after exposure to thrombin.

Figure 4.

Contrast-enhanced ultrasound images of microbubble response to thrombin. Contrast-enhanced ultrasound images of TACS microbubbles EDTA-treated blood (a) prior to thrombin addition and (b) 3 min after thrombin addition. SACS microbubbles in EDTA-treated (c) prior to thrombin addition and (d) 3 min after thrombin addition. (e) Average of three brightness analyses inside phantom of microbubble images, with error bars representing 95% confidence interval.

To demonstrate activation of the stimulus-responsive bubbles with naturally-generated thrombin, TACS and SACS microbubbles were added to blood freshly drawn from a one year old female New Zealand white rabbit. As soon as blood was removed from the animal, visible clotting in the tube was observed by eye (Figure 5a). At identical imaging conditions (2.0 MHz, 140 kPa), TACS microbubbles showed a 4-fold increase in signal over SACS microbubbles (Figure 5). The reduced increase in signal can be attributed to the active clotting occurring in the signal; as clotting decreases the acoustic transparency of the medium and the resulting signal generated by the activated microbubbles is substantially muted. Because of this decreased acoustic transparency, a 100-fold increase in microbubble concentration for all samples was necessary to observe any ultrasound signal. Again, the signal generated is comparable to a PEG-coated, freely oscillating microbubble with a formulation similar to those currently being used clinically. This sharp difference in the smart and scrambled bubbles was visible in as few as three minutes; this lag time is required to allow background signal generated by sample mixing to subside. This result also confirms that microbubble activation will occur faster than potential degradation by nucleases. Studies are currently underway to validate bubbles in a rabbit model with induced thrombosis.

Figure 5.

(a) Contrast-enhanced imaging of microbubbles in freshly drawn rabbit blood. (b-d) Contrast enhanced ultrasound sonographs of (b) TACS crosslinked microbubbles, (c) SACS crosslinked microbubbles, and (d) PEG-coated, freely oscillating microbubbles in freshly-drawn rabbit blood, 5 min after mixing. (e) Average of three brightness analyses inside phantom of microbubble images, with error bars representing 95% confidence interval.

In summary, we have demonstrated the first example of a stimulus-responsive contrast agent that generates ultrasound signal only in response to levels of thrombin found in active blood clots while remaining dormant in normal blood conditions. Contrast-enhanced ultrasound is ideal for imaging thrombosis, since it is both the primary modality for diagnosis and is safe, inexpensive, portable, and noninvasive. The development of a thrombin-sensitive contrast agent will help clinicians determine the activity of a clot and initiate proper therapy accordingly. In this work, microbubbles were found to activate in a fast, efficient manner at clinically relevant levels of thrombin in both fresh and preserved blood. These stimulus-responsive contrast agents have excellent potential for improving cardiovascular imaging techniques and the diagnostic power of an important imaging modality.

Experimental

Synthesis of DSPE-PAA

Synthesis of the DSPE-PAA conjugate was described previously.[7]

Synthesis of DSPE-PAA-DNA

DSPE-PAA (80.8 µg) was mixed with either 5’ Amine - CCA ACC CAC AAA - Dabcyl or 5’ Fluorescein - AAA CAA CCA ACC - Amine, (187.5 nmol, Integrated DNA Technologies) with N-hydroxysulfosuccinimide (8.70 mg, 40 µmol, Pierce) and 1-ethyl-3-(3-dimethylpropyl)carbodiimide hydrochloride (EDC, 3.83 mg, 20 µmol, Pierce) in 50 mM 2-(N-morpholino)ethanesulfonic acid (MES, Sigma-Aldrich) buffer (pH 4.95, 1 mL) under agitation at 60°C overnight. Product was subsequently purified by filtration through a regenerated cellulose filter (MWCO 10,000 kDa, Millipore) at 14,000g for 20 min three times. The retentate was subsequently lyophilized to produce a white powder. DNA attachment was quantified by comparing the UV-Vis absorbance at 260 nm (Beckman Coulter DU 730) with the mass of the collected powder. This procedure typically produced a product with an average of 2.7 DNA strands per DSPE-PAA.

Fluorescence Analysis

DSPE-PAA-DNA-FAM (20 pmol) and DSPE-PAA-DNA-Dabcyl (20 pmol) were mixed in 20mM Tris actetate buffered saline (200 µL) and measured by fluorometry (Perkin Elmer LS 55; λ_ex = 490nm). To achieve crosslinking and quenching, an equimolar amount of TACS was added, and the system was allowed to remain at RT for 30 min. To dissolve the tripartite DNA hybrid molecule, an equimolar amount of thrombin from human plasma (Sigma-Aldrich) was added.

Microbubble Formulation, Sizing and Crosslinking

Microbubbles were formulated and sized in a similar manner to that described previously.[14] Briefly, ten images per sample were taken at 20X in bright field at random locations in the sample. The microbubbles were counted and sized using ImageJ (NIH). Concentration was calculated by determining the area of spot size against known volumes of buffer. In order to add crosslinks to the encapsulating shell, a symmetric thrombin binding aptamer (IDT) was added to pre-formed microbubbles in a 3:1 molar ratio to each pair of fluorescence – quencher DNA strands. The bubbles were then allowed to sit at RT for 30 min to allow for the hybridization of the crosslinking strand to the complementary bound strands. Images were taken with a Nikon Eclipse TE200, for fluorescence images gain = 0.2, exposure = 5 set in Spot Basic.

Analysis of microbubble harmonic scattering

Samples were diluted to a concentration of 100,000 bubbles per mL in 2 mL of phosphate buffered saline within the bulb of a transfer pipette held upside down. The pipette bulb was fixed in a submerged position within a large fishtank centered to the focus of a 1” focused 2.25 MHz transducer (Panametrics V305), then insonated with a single pulse from a function generator (Tektronics TDS 2012C) at room temperature open to the atmosphere. Harmonic side scatter signal was collected by an unfocused 5 MHz transducer (Panametrics V308) perpendicular to the sending transducer, and amplified by 30 dB (Olympus 5072PR Pulser/Receiver). Focus was determined by maximizing the signal reflected off of a needle placed upright inside the pipet bulb prior to adding microbubbles.

Contrast-Enhanced Ultrasound Imaging in Whole Blood

Samples were imaged with a Siemens Acuson Sequoia 512 set in dual cadence mode (B-mode and contrast-enhanced ultrasound obtained simultaneously). The samples were diluted to a concentration of 10,000 bubbles per mL in 2 mL of EDTA-treated bovine whole blood (Sigma Aldrich), or a concentration of 1.0x10⁶ bubbles per mL in 2 mL of freshly drawn rabbit blood (1 year old, New Zealand White, 4.5 kg) within a stationary 15 mL centrifuge tube submerged within a large water tank at room temperature open to the atmosphere. Images were obtained at 2.0 MHz and MI = 0.10. Brightness analysis was performed with ImageJ (NIH) by integrating the brightness level inside the phantom minus an average of four background regions of the same size and shape at similar imaging depth, as done previously.[7]