Fig 1.
SFRP2-targeted and control ultrasound contrast agents were designed with NeutrAvidin™.
(A) Sulfhydryl-activated lipid solution was formulated by incorporating a thiol functionalized PEG-lipid (DSPE-PEG2000-thiol) at 0.2 mole percent in our lipid solution. Thiol-activated contrast agent was prepared by shaking the lipid solution in 3 ml glass vials containing perfluorobutane gas (see Methods for details). (B) NeutrAvidin™ functionalized contrast agent was prepared by forming a covalent crosslink between maleimide-activated NeutrAvidin™ and thiol-activated contrast agent. We estimated the thiol content of our sulfhydryl-activated contrast agent and tested 3 and 10-fold molar excess of maleimide-activated NeutrAvidin™. (C) Anti- SFRP2 targeted or control anti-chicken IgY targeted microbubbles were prepared by adding a 3-fold molar excess of either biotinylated SFRP2 antibodies or biotinylated antibody to chicken IgY to NeutrAvidin™ functionalized, ultrasound contrast agents.
Fig 2.
3-dimensional molecular imaging protocols estimated video intensity of free flowing contrast, with and without destruction of microbubbles.
Prior to injection of targeted contrast agent, anatomical features were recorded with a B-mode scan (black arrow). The first contrast-specific, Cadence™ contrast pulse sequencing scan was performed in the absence of contrast (blue line capped with open circle) to establish the average video pixel intensity (VI) of ROIs at baseline. (A) Time-intensity curves (TICs). A ‘non-destructive imaging scheme’ captured multiple Cadence™ mode scans during the first 2 minutes post-injection to capture wash-in of the contrast bolus, and then every two minutes between 2 to 20 minutes post-injection. Average video pixel intensities from a representative study were plotted, showing total signal from tumor ROIs (black line) and from non-tumor ROIs (dotted blue line). The TIC from non-tumor ROIs was used later to approximate the signal contributed by freely flowing contrast. (B) An image—destroy—image’ scheme was used to differentiate targeted contrast bound within tumor endothelium from targeted contrast flowing freely in the vasculature. A Cadence™ mode scan was captured at a pre-determined time (10 minutes, pre-destruction scan, blue arrow) to determine the average video pixel intensity in the tumor ROI. This was followed immediately by a high mechanical index (D-Color) scan to destroy all bound and freely flowing contrast within the tumor ROI (red arrow, destructive scan). A pause of 60 seconds allowed freely flowing contrast to re-enter the tumor vasculature, and a final Cadence™ mode scan was captured (12 minutes, post-destruction scan, blue arrow) to estimate the average VI contributed by freely flowing contrast. The average video pixel intensity contributed by contrast bound to endothelium was calculated by subtracting the average VI of the post-destruction scan from the average VI of the pre-destruction scan. Average video pixel intensity for all scans was reported after subtracting the average video pixel intensity of the baseline scans.
Fig 3.
Time-Intensity Curves (TICs) generated from a representative animal that received both SFRP2-targeted, and IgY-targeted control contrast.
(A) SFRP2-targeted contrast: solid blue line represents intensity within the tumor ROI, dotted black line represents intensity within a non-tumor ROI, red filled circles represent the difference between the tumor and non-tumor ROI intensities, which we term ‘free-flowing-corrected’ TIC (ffc-TIC). (B) IgY-targeted contrast: solid green line represents intensity within the tumor ROI, dotted black line represents intensity within a non-tumor ROI, purple open circles represent the difference between the tumor and non- tumor ROI intensities, which we term ‘free-flowing-corrected’ TIC (ffc-TIC). (C) The wash out of contrast from tumor ROI was modeled by one-phase exponential decay for IgY-targeted contrast (raw data, open green circles; green line, best-fit model), and by a plateau followed by one-phase exponential decay for SFRP2-targeted contrast (raw data, solid blue circles; blue line, best-fit model). (D) The difference between the models depicted in panel (C) was plotted with open black triangles. Note the maxima between 6–10 minutes. (E) The ffc-TICs for SFRP2, and IgY-targeted contrast were fitted to curves. A one-phase association model (grey line) fit the wash in portion of the ffc-TIC for SFRP2-targeted contrast (red filled circles), and for IgY-targeted contrast (open purple circles). The wash out of SFRP2-targeted ffc-TIC was best fit with a linear regression model (blue line), while the wash out of IgY-targeted ffc-TIC was best fit with a one-phase exponential decay model (green line). (F) The best-fit model for the IgY-targeted ffc-TIC was subtracted from the best-fit model for SFRP2-targeted ffc-TIC, and was plotted (red filled circles with red line). This produced a TIC representing the signal intensity within the tumor ROI that could be attributed to binding of contrast specifically mediated by the SFRP2 antibodies used to formulate the SFRP2-targeted contrast. N = 5 animals.
Fig 4.
The relationship between microbubble acoustic signal, and log compressed video pixel intensity.
(A) The acoustic signal from microbubbles in arbitrary acoustic units (AU) was log compressed from a dynamic range of 80 dB to an 8-bit range (0–255) of video pixel intensities (VI). Given video pixel intensities (VI), log decompression with Eq 1 was used to estimate the original acoustic signal in AU, and was plotted to show their non-linear relationship over the entire 8-bit range. (B) A linear relationship existed between video pixel intensities (VI), and their corresponding acoustic signals (AU), when VI was < 20. This linear relationship represented a range of acoustic signals limited to ~ 0.025–0.055 AU. (C) The ratio of any two video pixel intensities (VI) was not related linearly to the ratio of their estimated acoustic signals. (D) However, the ratio between log-decompressed acoustic signals (AU2 ÷ AU1) was related linearly to Δ VI = (VI2 –VI1). Computing ΔVI resulted in a metric that was proportional to the ratio of the original acoustic signals: a natural consequence of the Law of Exponents. Linear regression provided the best-fit equation: (AU2 ÷ AU1) = 0.0529 × ΔVI + 0.9455, with R2 = 0.992 describing the relationship between ΔVI, and (AU2 ÷ AU1) when ΔVI < 20.
Table 1.
Best-fit values described nonlinear curves for wash out of contrast.
Table 2.
Best-fit values described models for wash out of contrast from non-tumor ROI.
Table 3.
Best-fit values described models for ffc-TICs.
Fig 5.
Average video pixel intensity increased with higher levels of NeutrAvidin™ labeling.
Significantly higher tumor video pixel intensity was observed when SFRP2-targeted contrast was created using 10-fold molar excess of maleimide-activated NeutrAvidin™ compared to 3-fold molar excess. N = 8 animals.
Fig 6.
Average video pixel intensity increased with increased microbubble dose.
Significantly higher video pixel intensity was observed in tumors when animals received 5 × 107 SFRP2-targeted microbubbles compared to 5 × 106 SFRFP2-targeted microbubbles. N = 10 animals.
Fig 7.
Ultrasound molecular imaging of animal receiving SFRP2-targeted and control IgY-targeted contrast.
A white dashed line outlines tumors. The contrast-specific signal (green) was superimposed over the b-mode image (grey). At 30 seconds, average video pixel intensity was similar between control and SFRP2-targeted contrast. The contrast-specific video intensity was retained in tumors at much higher levels when using the SFRP2-targeted contrast compared to the IgY-targeted contrast.
Fig 8.
Verifying performance of optimized molecular imaging reagents, and protocol in vivo.
Non-identical superscripts indicate a statistically significant difference in means. Both IgY-targeted (a, p = 0.03), and SFRP2-targeted (c, p < 0.001) contrast produced significantly higher video pixel intensities in tumor ROI compared to peri-tumoral regions (b). SFRP-2 targeted contrast (c, p < 0.001) produced significantly higher signal intensity in tumor ROI than the control IgY-targeted contrast (a). Average video pixel intensities were corrected for free flowing contrast by subtracting the post-destruction VI from the pre-destruction VI as shown in Fig 2B for the ‘image—destroy—image’ method. N = 10 animals.