Ultrasound in Med. & Biol., Vol. 40, No. 2, pp. 389–399, 2014 Crown Copyright Ó 2014 Published by Elsevier Inc. on behalf of World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/$ - see front matter

http://dx.doi.org/10.1016/j.ultrasmedbio.2013.09.022

d

Original Contribution VEGFR2-TARGETED MOLECULAR IMAGING IN THE MOUSE EMBRYO: AN ALTERNATIVE TO THE TUMOR MODEL JANET M. DENBEIGH,* BRIAN A. NIXON,* JOHN M. HUDSON,* MIRA C. PURI,*y and F. STUART FOSTER* * Sunnybrook Research Institute, Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada; and y Lunenfeld-Tanenbaum Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada (Received 3 July 2013; revised 16 September 2013; in final form 18 September 2013)

Abstract—As a tumor surrogate, the mouse embryo presents as an excellent alternative for examining the binding of angiogenesis-targeting microbubbles and assessing the quantitative nature of molecular ultrasound. We establish the validity of this model by developing a robust method to study microbubble kinetic behavior and investigate the reproducibility of targeted binding in the murine embryo. Vascular endothelial growth factor receptor 2 (VEGFR2)targeted (MBV), rat immunoglobulin G2 (IgG2) control antibody-targeted (MBC) and untargeted (MBU) microbubbles were introduced into vasculature of living mouse embryos. Non-linear contrast-specific and B-mode ultrasound imaging, performed at 21 MHz with a Vevo-2100 scanner, was used to collect basic perfusion parameters and contrast mean power ratios for all bubble types. We observed a twofold increase (p , 0.001) in contrast mean power ratios for MBV (4.14 ± 1.78) compared with those for MBC (1.95 ± 0.78) and MBU (1.79 ± 0.45). Targeted imaging of endogenous endothelial cell surface markers in mouse embryos is possible with labeled microbubbles. The mouse embryo thus presents as a versatile model for testing the performance of ultrasound molecular targeting, where further development of quantitative imaging techniques may enable rapid evaluations of biomarker expression in studies of vascular development, disease and angiogenesis. (E-mail: [email protected]) Crown Copyright Ó 2014 Published by Elsevier Inc. on behalf of World Federation for Ultrasound in Medicine & Biology. Key Words: Micro-ultrasound, Molecular imaging, Mouse embryo, Microbubble contrast agent, Vascular endothelial growth factor receptor 2.

logic and pathologic processes (Cosgrove and Lassau 2010). What is more, advances in molecular biology have made it possible to conjugate these microbubbles to ligands (peptides, antibodies, glycoproteins) (Inaba and Lindner 2012; Voigt 2009) that enable specific binding to receptors. These may actively and selectively accumulate within the vasculature at targets associated with angiogenesis, inflammation, ischemia-reperfusion injury, thrombi and plaque (Klibanov 2009; Voigt 2009), and it has been suggested that targeted contrastenhanced ultrasound imaging may be used to detect and quantify expression levels of molecular markers on the vascular endothelium (Deshpande et al. 2011; Gessner and Dayton 2010). Over the last decade, contrastenhanced targeted ultrasound imaging has been found to have potential in molecular imaging with longitudinal assessments of inflammation and tumor angiogenic marker expression levels (Deshpande et al. 2011, 2012; Warram et al. 2011) and has been used to investigate mechanisms responsible for angiogenesis in disease models (Anderson et al. 2011; Fischer et al. 2010) and

INTRODUCTION Ultrasound is well established as an imaging tool that provides both anatomic and functional assessment of tissues. In cardiovascular imaging, ultrasound is highly effective as a means of imaging the heart in real time and in the measurement of the hemodynamics of blood flow via the Doppler effect (Cosgrove and Lassau 2010; Garcia et al. 2011). Ultrasound’s diagnostic capabilities have been further augmented by the introduction of microbubble contrast agents (Jansson et al. 1999). These small (Klibanov 2009), reflective (Liang and Blomley 2003), oscillating bubbles (see Lindner 2004 for a comprehensive review) have been established as excellent vascular tracers for the assessment of important physio-

Address correspondence to: Janet M. Denbeigh, Sunnybrook Health Sciences Centre, 2075 Bayview Avenue, S640, Toronto, ON Canada M4 N 3 M5. E-mail: [email protected] Conflicts of Interest: F.S. Foster acknowledges his role as consultant to VisualSonics Inc. 389

390

Ultrasound in Medicine and Biology

to evaluate pro-/anti-angiogenic therapy efficiency (Inaba and Lindner 2012). These approaches are still, however, largely qualitative. It remains to be seen whether molecular ultrasound will be able to provide quantitative information in a reliable and translatable manner. Many attempts have been made to quantify molecular ultrasound imaging of endothelial markers in tumor microvasculature, but the inherent heterogeneity of these systems has made it difficult to evaluate whether the observed changes in marker expression are truly being measured (Deshpande et al. 2011; Moestue et al. 2012). The highly regulated and controlled processes of normal angiogenesis and vasculogenesis in the mouse embryo, however, make it an excellent surrogate for these processes. As an additional benefit, advances in mouse genomics have allowed the identification and manipulation of hundreds of genes that influence mouse development and disease (Garcia et al. 2011), resulting in a number of useful transgenic strains. The versatility of the murine embryo thus makes it an ideal medium for studying the behavior of angiogenesis-targeting microbubbles, for optimizing current detection methods and for assessing the quantitative nature of molecular ultrasound. A number of groups have previously explored the potential of introducing microbubbles into the mouse embryo (Endoh et al. 2002; Aristiz
abal et al. 2007). Most recently, Bartelle et al. (2012) used in utero microultrasound-guided injections of avidin microbubbles into cardiac ventricles to assess binding within BiotagBirA transgenic embryos, focusing primarily on analyzing vascular anatomy and imaging elements not displayed on the luminal surface of vascular endothelial cells. Building on these studies, we aimed to establish the suitability of the mouse embryo as a vehicle for studying the performance of ultrasound molecular targeting by testing the kinetic behavior and reproducibility of targeted microbubble imaging in our model. Here, we present a robust method for introducing microbubbles into isolated living embryos. This offers freedom in terms of injection control and positioning, improved reproducibility of the imaging plane without obstruction and simplified image analysis and quantification. We measure a number of basic perfusion parameters for three types of microbubbles, examine possible influencing factors related to microbubble binding in living lategestational-stage embryos and test the reproducibility of our technique. We also present data indicating that molecular ultrasound imaging of endogenously expressed endothelial cell surface markers in the developing mouse embryo is possible, confirmed through the binding of targeted microbubbles to vascular endothelial growth factor receptor 2 (VEGFR2), a key molecule involved in the process of angiogenesis (Ferrara 2001) expressed directly

Volume 40, Number 2, 2014

on the embryonic vasculature, with concomitant study of the expression patterns of this clinically relevant target. METHODS Experimental preparation Wild-type CD-1 male and female Mus musculus (Charles River Laboratories, St. Constant, QC, Canada) were mated to produce staged embryos, with embryonic day (E) 0.5 defined as noon of the day a vaginal plug was observed. Before experimentation, dissection plates were created using a 1:8 volume ratio of curing agent to base (Sylgard 184 Silicone Elastomer Kit, Dow Corning, Midland, MI, USA). Glass needles (1 3 90-mm glass capillaries with filament) were pulled (PN-30, Narishige, East Meadow, NY, USA), and clear ultrasound gel (Aquasonic, Fairfield, New Jersey, USA) was centrifuged at 140g for 20 min and drawn into 30-mL syringes for easy application. Female luers (Cole Parmer, Montreal, QC, Canada) were attached to 400-mm pieces of polyvinyl chloride tubing (inner diameter 5 0.79 mm, VWR, Mississauga, ON, Canada). Finally, embryo medium was prepared and kept at 4 C: 89% Dulbecco’s modified Eagle medium with high glucose (Sigma, Oakville, ON, Canada), 9% fetal bovine serum, 1% 1 M Hepes, 1% penicillin-streptomycin (10,000 units penicillin, 10,000 mg streptomycin) (Gibco, Burlington, ON, Canada). An imaging platform (Integrated Rail System, VisualSonics, Toronto, ON, Canada) was arranged beneath the surgical microscope (Stereomaster, Fisher Scientific, Ottawa, ON, Canada). Syringes of ultrasound gel and phosphatebuffered saline (PBS) were pre-heated in a water-bath maintained at 45 C (Whiteley et al. 2006). Surgical procedure The surgical procedures performed in this study, adapted from Whiteley et al. (2006), were approved by the Animal Care Committee at Sunnybrook Research Institute (Toronto, ON, Canada). Embryos were removed at E15.5–E18.5. Pregnant mice were sacrificed by cervical dislocation, and uteri were removed and placed immediately in chilled embryo medium. Embryos were dissected out and kept on ice with fresh medium (changed every 1–1.5 h). Microbubble reconstruction and preparation Three types of microbubbles (MicroMarker, VisualSonics) were prepared: VEGFR2-targeted microbubbles (MBV); rat isotype immunoglobulin G2 (IgG2) control antibody-targeted microbubbles (MBC); and untargeted microbubbles (MBU). The agent (a dry-freeze powder) was reconstituted with 1 mL saline by slowly injecting it into the microbubble vial using a 21-G needle; the plunger was withdrawn, removing 1 mL of air, and the

VEGFR2-targeted molecular imaging in mouse embryos d J. M. DENBEIGH et al.

needle was removed; the vial was gently agitated and left to stand 8 min. Antibodies—VEGFR2-targeting anti-mouse CD309 [FLK1] biotin and rat isotype IgG2 control-targeting anti-mouse IgG2a biotin (eBioscience, San Diego, CA, USA)—were coupled to the lipid-shelled, perfluorocarbon-containing microbubbles through streptavidin (bubble)-biotin (antibody) interactions via the addition of 20 mg (recommended by MicroMarker) of antibody in 1 mL saline (final volume). A new syringe was filled with the antibody dilution and added to the appropriate vial in the manner outlined above; the final mixture (2 mL) was left to stand for 10 min and kept on ice. MBC served as a control for confirming the specificity of our MBV and ruling out non-specific Fc receptor binding or other cellular protein interactions. More detailed specificity tests have been conducted previously by Deshpande et al. (2011). On the basis of the assumption of complete surface conjugation, the average number of binding ligands (Deshpande et al. 2011) for the microbubbles is expected to be approximately 7600/mm2. Untargeted microbubbles did not contain streptavidin. The concentration and size distributions of each microbubble population were quantified using a Beckman Coulter counter (Multisizer 3, Beckman Coulter Canada, Mississauga, ON, Canada). Immediately before injection, a dilution of 5 3 106 MB/mL was prepared in saline and drawn into a syringe using a 21-G needle. The needle was removed, the luer and tubing were attached and the solution was pushed to the end, ensuring no air bubbles were generated. The syringe was then inserted into the syringe infusion pump (NE-1000, Biolynx New Era Pump Systems, Brockville, ON, Canada). A pulled glass needle was attached to the end of the tubing. Injection of microbubbles into embryos Each embryo, injected once with a single bubble type, was randomly selected and placed in a dissection dish. The yolk sac and amniotic membranes were cut from the anti-mesometrial side, and the embryo was gently removed. The embryo was positioned on its side with the placenta and umbilical vessels in front; the yolk sac and placenta were pinned in place and the embryo was washed with pre-warmed PBS, as illustrated in Figure 1. After revival (100% success rate within 40 s), the umbilical vein and associated vessels were identified. Pre-warmed ultrasound gel was used to cover and surround the embryo and topped with warm PBS. The glass needle was mounted on a plasticine base and inserted into the PBS. A placental vein was selected for injection, and the tip of the needle was trimmed to size (diameter: 50–100 mm) with Vannas-Tubingen scissors (Fine Science Tools, North Vancouver, BC, Canada). Microbubble solution was injected with the pump at 20 mL/min through the needle until all of the air was

391

Fig. 1. Experimental setup for the injection of microbubbles into isolated embryos. A 21-MHz linear array transducer (Vevo-2100) is positioned above an exteriorized living embryonic day 15.5 embryo as a 20-mL microbubble solution is injected into a placental vein using a glass cannula. Bar 5 10 mm.

expelled from the tip and microbubbles were observed to flow freely, and then was stopped. The glass needle was gently inserted into the desired vessel and fixed in place for the entire imaging experiment. The transducer was immediately positioned above the embryo, and 20 mL of microbubble solution was injected at 20 mL/min using the syringe pump. We estimate the bolus injection accounts for 8%–15% of embryonic blood volume, with in vivo microbubble concentrations of approximately 4.0–7.6 3 105 MB/mL. Ultrasound imaging (perfusion or molecular) was then initiated. Post-imaging, the embryo was euthanized via decapitation with surgical scissors, or for cases of whole-body staining, anesthesia was induced by submerging the embryo in chilled PBS, where it was kept for 2 h to ensure death before transfer to fixative. For each subsequent embryo, a fresh dissection dish, needle, syringe and tubing segment were used, with preparation of the next batch of microbubble injection solution taking place during dissection and revival. Ultrasound imaging Ultrasound imaging was performed on a Vevo-2100 scanner (VisualSonics) using a 21-MHz linear array transducer (MS250, VisualSonics). All time-gaincompensation sliders were shifted to the exact middle position. Split screen B-mode and non-linear contrastspecific mode (standard settings) were employed, using the following parameters: frequency 5 18 MHz; lateral resolution 5 165 mm; axial resolution 5 75 mm; transmit power 5 4%; contrast gain 5 30 dB; 6- and 10-mm foci; and wide beam width, with destruction pulses of 3.7 MPa at 100% transmit power for 1 s (Deshpande et al. 2012). During perfusion studies (E17.5 embryos, n 5 11), 20-min cine loops captured the entire wash-in and dissipation (tail) of the microbubble bolus at a frame rate of

392

Ultrasound in Medicine and Biology

1 Hz. To assess binding in E15.5–E18.5 embryonic molecular imaging studies (separate cohort of embryos, n 5 114), microbubbles were allowed to circulate and adhere for 3 min 40 s post-injection, after which ultrasound imaging was initiated, and a ‘‘pre-destruction’’ acoustic response sequence was recorded at 29 Hz. At 4 min post-injection, a short burst of high acoustic pressure was employed, destroying all of the microbubbles in the imaging plane. This time point was selected to minimize the duration of each embryo imaging session as limited by the temporal window of embryo viability, while keeping in line with existing molecular ultrasound imaging protocols (Deshpande et al. 2011; Lyshchik et al. 2007). The subsequent ‘‘post-destruction’’ imaging sequence was assumed to contain only circulating bubble signals that had replenished the beam. Immunofluorescence staining The E15.5 embryos collected post-injection were immediately fixed in 4% chilled paraformaldehyde (Sigma) solution for 24 h and subsequently paraffinized, and 5-mm transverse sections were mounted on glass slides. Antigens were retrieved after rehydration using a proteinase K (Roche Diagnostics, Indianapolis, IN, USA) dilution (20 mg/mL). Endogenous fluorescence was quenched for 1 h with 3% H2O2 (Sigma) and serum blocked for 2 h (20% heat-inactivated sheep [HIS], 2% donkey and 1% heatinactivated fetal calf sera [Sigma]) before streptavidin/ biotin blocking (Vector Laboratories, Burlington, ON, Canada). Primary biotinylated rat anti-mouse CD309 (FLK1) (clone Avas12 a1, eBioscience) or rat IgG2a K-isotype control (eBioscience) antibody was incubated at a 1:400 dilution overnight at 4 C. Samples were rinsed with a 2% HIS solution in 0.1% Triton X-100 (Sigma) in PBS for 5 h and incubated overnight at 4 C with a 1:100 dilution of streptavidin-horseradish peroxidase (PerkinElmer, Woodbridge, ON, Canada). Additional washes were performed in 2% HIS and TNT buffer (0.1 M Tris, 0.15 M NaCl, 0.2% deionized-distilled H2O, pH 7.5 [Sigma]). Antibody signal was amplified during a 1-h tyramide-cy3 (PerkinElmer) incubation. Slides were added to 30% glycerol (Sigma) for clearing and dehydrated in increasing concentrations of ethanol (Sigma). Dako fluorescence mounting medium (Dako Canada, Burlington, ON, Canada) was used for mounting coverslips. Embryo samples were fluorescently imaged using a Mirax scanner (403, Carl Zeiss, Toronto, ON, Canada). DATA ANALYSIS AND MATHEMATICAL METHODS Analysis for perfusion imaging A number of perfusion parameters were derived from the time-intensity curves, including peak enhance-

Volume 40, Number 2, 2014

ment (PE), wash-in rate, time to peak (TTP) and binding coefficient (t). First, the contrast-specific and B-mode ultrasound image loops (Fig. 2a) were imported from the Vevo-2100 as raw data (beam formed, enveloped detected) into MATLAB (R2011a, The Mathworks, Natick, MA, USA) for quantification. It is assumed that the intensity of the contrast-mode signal is linearly related to the instantaneous microbubble concentration (Lampaskis and Averkiou 2010). Peak microbubble enhancement was assessed within the entire embryo with a 5 3 5pixel (0.06-mm2) grid to create parametric maps of the peak intensities (Fig. 2b). These were qualitatively assessed for homogeneity within the brain of the embryo. Regions of interest (ROIs) were drawn around the left and right hemispheres of the brain (see Fig. 2a), staying within the periphery of the skull, avoiding image artifacts and excluding large, visible vessels. Time-intensity curves were extracted for the ROIs of each embryo and plotted (Fig. 2c). A seven-point median filter was used to smooth each curve, from which the brain peak enhancement was determined. The wash-in rate, defined as the slope between 10% and 50% of the maximum PE, was also calculated. TTP was computed from the onset of the signal enhancement to the peak time. Lastly, the microbubble binding coefficient (t) was determined by fitting a single exponential decay function (E 5 exp(–t/t)) to the tail beyond the peak of the bolus curve using a non-linear least-squares algorithm in MATLAB. Results are presented as means 6 standard deviations, with average coefficients of variation (COVs) presented for each parameter. Differences in mean perfusion parameters across microbubble type were tested in PASW Statistics 18 (IBM, Armonk, NY, USA) using a univariate general linear model (perfusion parameter and microbubble type as fixed factors) and separate t-tests. A statistical power of 0.8 was achieved, and a two-sided p-value , 0.05 was considered to indicate statistical significance. Analysis for molecular imaging Molecular ultrasound imaging within the embryos was assessed by investigating the degree to which various experimental parameters might influence microbubble binding, including differences across embryonic stages and between litters, viability, measured as a function of heart rate and the order in which embryos were injected as well as microbubble type. Time-intensity plots were generated for 1.5-mm2 ROIs in the embryonic left and right brain hemispheres, and the ratio of the average signal intensities of ‘‘pre-destruction’’ to ‘‘post-destruction’’ sequences was used to produce a measure of the molecular signal called the contrast mean power ratio (CMPR). In a single imaging plane, five separate ROIs were placed within the left brain of each embryo to compute the COV within each embryo. These were

VEGFR2-targeted molecular imaging in mouse embryos d J. M. DENBEIGH et al.

393

Fig. 2. Perfusion parameters for embryonic day (E) 17.5 embryos derived from time-intensity curves. (a) B-Mode image of E17.5 embryo with a region of interest (ROI) within the left brain. (b) Perfusion map of peak enhancement within the same embryo, with significant intensity (in arbitrary units, in red) within the heart and carotid arteries (arrowheads). (c) Intensity plot of microbubbles as a function of time within a single ROI in the embryonic brain. Perfusion parameters are identified on the graph, including peak enhancement (PE), wash-in rate (slope), time to peak (TTP) and binding coefficient (t). Bar 5 5 mm.

averaged to compute an overall ‘‘within embryo’’ variability (COVw). From the distribution of CMPRs for each group of microbubble type and embryo stage, categorical COVs were averaged to calculate the mean variability across embryos as well (COVa). Significance of microbubble type and embryonic stage to CMPR was evaluated using a general linear model for repeatedmeasures analysis, with a Student-Newman-Keuls test for multiple comparisons (Kusuoka and Hoffman 2002). Microbubble type and stage were defined as betweensubject factors, with brain hemisphere (right, left) as a within-subject factor. Multiple regression analysis was used to mathematically describe the relationships of heart rate, order of injection, litter, bubble type and embryonic stage to CMPR. A two-sided p-value , 0.05 was considered to indicate statistical significance. All data are reported as means 6 standard deviations, where n is the number of embryos studied, unless otherwise noted. Analyses were conducted in PASW Statistics 18. Embryos were excluded from analysis if there was profuse bleeding during injection, if the animal moved significantly during imaging (e.g., twitching) or if the injection needle was blocked. In some cases (,5%), repositioning of the needle was required before initiating the bolus injection. As this minor complication was easily resolved, these embryos were included in the final results. Error bars plotted in figures represent one standard deviation of the mean. RESULTS We established a new method to image perfusion of microbubbles within living embryos, facilitating investi-

gation of the behavior and circulation times of various microbubble types. Embryos were consistently revived up to at least 3 h after initial isolation with chilling, and we observed that after dissection of the yolk sac, the embryonic heart would beat for at least 20 min. This heart rate, 44 6 17 beats/min on average during molecular imaging studies, was qualitatively observed to decrease over time. The median diameter of microbubbles was 0.78 6 0.12 mm, and the average diameter was 1.77 6 0.12 mm. The average microbubble vial concentration was 2.46 6 1.05 3 107 MB/mL. Non-linear contrast ultrasound imaging confirmed the presence of microbubbles in these animals throughout each experiment, as illustrated in Figure 3, where a number of time points are sampled before, during and after MBU injection in an E17.5 embryo. Rapid wash-in of the microbubbles can be observed within the hearts and livers of all animals with prolonged wash-out of the bubbles, particularly evident within the liver. We identified the brain, where there was minimal tissue motion, as the ideal target for our microbubble studies. Moreover, the brain exhibits regular vascularity compared with more complicated arrangements found elsewhere (e.g., liver), making it possible to avoid large vessels that would require filtering for perfusion analysis. Reproducibility in ultrasound probe placement and plane selection is also possible because of the well-defined structure of the skull. Finally, brain histogenesis begins after mid-gestation (Yamada et al. 2008), making the expression of angiogenic markers in this tissue likely. The presence of VEGFR2 was verified via immunofluorescence staining of brain tissue. We chose to target VEGFR2 in our initial experiments

394

Ultrasound in Medicine and Biology

Volume 40, Number 2, 2014

Fig. 3. Injection and perfusion of microbubbles in an embryonic day (E) 17.5 embryo across multiple time points. Ultrasound images of untargeted microbubbles (MBU) within an E17.5 embryo at various time points. (a) B-Mode ultrasound image at t 5 0 s, before injection. (b–h) Subsequent non-linear contrast mode ultrasound images. (b) t 5 0 s, with depth scale in millimeters along the side. (c) t 5 20 s, at commencement of injection. (d) t 5 40 s, during injection. (e) t 5 50 s. (f) t 5 80 s, post-injection. (g) t 5 120 s, entire embryo is perfused. (h) t 5 840 s, during wash-out of microbubbles. Arrowheads: R 5 ribs, Li 5 liver, Ht 5 heart, Br 5 brain. Bar 5 5 mm.

because of the clinical importance of this marker as an anti-angiogenic target (Shojaei 2012), highlighted by the advent of a VEGFR2-targeting microbubble (BR55) entering clinical trials (Wijkstra et al. 2012). Perfusion imaging In a litter of E17.5 embryos, microbubble kinetics was assessed for MBU (nU 5 4), MBC (nC 5 4) and MBV (nV 5 3) bubbles. Parametric maps (Fig. 2b) of peak intensities achieved within the embryos exhibit maximal enhancement within the heart and the carotid arteries. Within the brain itself, enhancement was symmetric across hemispheres and homogeneous within the outer cerebral cortex, diencephalon and midbrain. A number of perfusion parameters were extracted from the timeintensity curves plotted for each ROI. The results for PE, wash-in rate, t and TTP are summarized in Table 1. Corresponding within-litter COVs are also reported.

Because of the frame rates employed for this portion of the study, we were unable to measure heart rate in the embryos. Multiple comparisons between subject effects (perfusion parameter, bubble type) indicated that there is a significant (p 5 0.048) effect for microbubble type. This was observed in t-tests performed on each pairing of microbubble type, where both t and TTP for MBV are significantly different from the values for MBC or MBU. The t-test also indicated that MBU have a significantly different wash-in rate compared with MBV. Molecular imaging We assessed the binding capabilities of microbubbles, tested influencing factors and determined whether molecular imaging of endogenously expressed angiogenic markers with microbubbles within embryos was feasible and reproducible. This work was performed in 15 litters (average size: 9 6 3 embryos) of mice across

VEGFR2-targeted molecular imaging in mouse embryos d J. M. DENBEIGH et al.

395

Table 1. Embryonic day 17.5 perfusion parameters Microbubble type

p value (t-test)

Perfusion parameter

MBU

MBC

MBV

U vs. C

U vs. V

C vs. V

Peak enhancement, PE (a.u.) Wash-in rate (a.u./s) Binding coefficient, t (s) Time to peak, TTP (s)

0.427 6 0.063 0.008 6 0.002 289 6 93 53.75 6 7.96

0.490 6 0.079 0.009 6 0.002 288 6 124 51.75 6 4.46

0.499 6 0.064 0.011 6 0.002 481 6 188 45.00 6 5.33

0.10 0.18 0.83 0.55

0.06 0.01 0.03 0.03

0.81 0.09 0.02 0.03

Univariate general linear model: Between-subject effects

Perfusion parameter Microbubble type Perfusion parameter Microbubble type

df

F

p-value

3 2 6

1257.26 3.19 3.25

,0.01 0.048 0.01

a.u. 5 arbitrary units, df 5 degrees of freedom. Means 6 standard deviations derived from time-intensity plots for all embryo regions of interest. Microbubble type is found to be a significant factor for perfusion parameters, with MBV having significantly different t and TTP as found from the t-tests. The average coefficients of variation are 15%, 19%, 12% and 38% for PE, wash-in rate, TTP and t, respectively.

four embryonic stages of development (E15.5, E16.5, E17.5 and E18.5), with three types of microbubbles (MBV, MBC and MBU). Non-linear contrast-enhanced ultrasound imaging signals were collected, and the contrast mean power ratios for both brain hemispheres were computed for all embryos. Variability within five brain ROIs of a single embryo (COVw) was on average 15%, whereas CMPR variability across sets of embryos (COVa) was found to be 33%. Average CMPRs for each category are plotted in Figure 4, with the youngest embryos exhibiting minimal differentiation between bubble groups, reflected in an insignificant p 5 0.13. All subsequent stages, however, exhibited a greater than twofold increase in the mean VEGFR2-targeted CMPR compared with control bubbles. The p values were ,0.001, 0.003 and ,0.001 for E16.5, E17.5 and E18.5 embryos, respectively. Statistical analysis of all embryo CMPRs revealed a significant effect for microbubble type, F(2,101) 5 54.48, p , 0.001. No effect was found for embryonic stage alone, F(3,101) 5 0.79, p 5 0.502, or between embryonic stage and microbubble type, F(6,101) 5 1.57, p 5 0.164. CMPRs compiled according to bubble type, regardless of stage, are summarized in Figure 5. These ratios indicate that targeted microbubble binding within the mouse embryo is detectable, and that MBV bind in greater numbers compared with MBU and MBC. The relationship between CMPRs and potential explanatory variables, including litter, embryonic stage, heart rate, order of injection and microbubble type, was also examined. A multiple regression model with all predictors yielded R2 5 0.41, F(5,107)5 14.65, p , 0.001. Insignificant contributions were observed from litter (p 5 0.877), embryonic stage (p 5 0.799), heart rate (p 5 0.664), and order of injection (p 5 0.423), whereas a significant coefficient for

microbubble type (p , 0.001) was obtained, indicating that of the variables examined, only bubble type is a predictor of CMPR. Immunofluorescence staining Immunofluorescence staining was performed on transverse sections of paraformaldehyde-fixed E15.5 embryos. The pattern of VEGFR2 staining closely matched published observations (Allen Institute for Brain Science 2012), with intense staining detected in vasculature-rich areas including the heart, lungs and liver, as illustrated in Figure 6 for an E15.5 embryo. Within the brain, VEGFR2 expression was identified within the medulla oblongata, with pronounced staining through the fourth ventricle and around the perimeter of the cerebellar primordium and midbrain. Signal was absent in the mesencephalic vesicle and cerebral aqueduct. Immunofluorescence staining was used for qualitative purposes only, because of the non-linear tyramide amplification techniques used to detect antibody signals, and clearly confirms the presence of the vascular target VEGFR2 within brain tissues of murine embryos. DISCUSSION Ultrasound imaging has found widespread use in morphologic, functional and hemodynamic phenotyping of embryonic mice, as described by Kulandavelu et al. (2006). Contrast-enhanced imaging within embryos, however, has the potential to affect our understanding of the developing cardiovascular system, basic vascular biology and genetic models of heart disease. Most importantly, it might also serve as an excellent model in which to study the performance of ultrasound molecular targeting. In this study, we developed a unique method for

396

Ultrasound in Medicine and Biology

Volume 40, Number 2, 2014

Fig. 4. Targeted microbubble binding ratios for embryonic day (E) 15.5–E18.5 embryos. Summary of the average contrast mean power ratios (CMPRs) computed for each microbubble type—vascular endothelial growth factor receptor 2 (VEGFR2)-targeted microbubbles (MBV); rat isotype immunoglobulin G2 (IgG2) control antibody-targeted microbubbles (MBC); and untargeted microbubbles (MBU)—and stage. Results are plotted in order of increasing gestational stage, starting with E15.5 embryos (MBV,E15.5 [n 5 9, 3.37 6 1.16], MBC,E15.5 [n 5 10, 2.30 6 1.13], MBU,E15.5 [n 5 10, 1.96 6 0.51]). All embryos beyond E16.5 (MBV,E16.5 [n 5 8, 4.69 6 1.64], MBC,E16.5 [n 5 9, 2.08 6 0.74], MBU,E16.5 [n 5 8, 1.77 6 0.40]) and older (MBV,E17.5 [n 5 9, 4.18 6 2.01], MBC,E17.5 [n 5 9, 1.86 6 0.49], MBU,E17.5 [n 5 9, 1.95 6 0.42] and MBV,E18.5 [n 5 10, 4.30 6 1.70], MBC,E18.5 [n 5 10, 1.55 6 0.33], MBU,E18.5 [n 5 13, 1.41 6 0.22 ]) exhibit a significant increase in the mean VEGFR2-targeted CMPR compared with the others. p , 0.001, p 5 0.003 and p , 0.001 for E16.5, E17.5 and E18.5 embryos, respectively.

investigating the behavior of microbubbles within living murine embryos in the context of microbubble kinetics and perfusion, validating targeted imaging of endogenous endothelial surface markers and examining potential confounding factors as a means of establishing the suitability of the mouse embryo as a model for studying the quantitative nature of microbubble binding. Our first objective in assessing the suitability of the murine model was to describe perfusion parameters, including PE, wash-in rate, TTP and t, using microbubbles within the embryo. The nature of the vascular arrangement in the embryo, with a major fraction of the cardiac output being diverted away from the lungs via the foramen ovale and the ductus arteriosus (Kaufman and Bard 1999), suggests that microbubble permeation and filtering may be unique in embryos. For this reason, standard curve fitting based on two-compartment models was not implemented, as the models do not account for these changes in vascular architecture and flow. Within the embryos, bubble type appears to be a factor in the measure t. We observed that MBV seem to persist longest in the embryos, as reflected by the largest t, suggesting that targeted bubbles are retained within the vasculature. In all embryos, we did observe some bubble retention

within the liver (not shown), a common observation sometimes attributed to trapping in the sinusoids or active retention by the reticuloendothelial system (Stewart and Sidhu 2006). Unexpected significant differences found in TTP and wash-in rates are most likely related to variations in embryonic heart rates, which unfortunately could not be measured accurately because of the slow frame rates employed, a consideration for future studies. This, and the technical challenge of maintaining ‘‘normal’’ embryonic physiology, makes a case for transitioning any work pursuing characterization of perfusion measurements into an in utero model. In utero injections may suffer from their own challenges, however, including guiding needles; maintaining hemodynamic viability in the mother and exteriorized embryos; motion artifacts; addressing long-term effects of anesthesia; and complications caused by bleeding (Yamada et al. 2008). Nevertheless, our results indicate reproducibility in our injection technique, and despite the somewhat altered physiology of our embryos, the time-intensity curves and perfusion parameters for our control and targeting microbubbles are comparable to those observed in normal and xenograft mouse tissues (Sugimoto et al. 2012; Schneider 2011). In terms of microbubble circulation then, the

VEGFR2-targeted molecular imaging in mouse embryos d J. M. DENBEIGH et al.

Fig. 5. Microbubble binding across all embryo stages. Contrast mean power ratios (CMPRs) across all stages are plotted for each bubble type, including vascular endothelial growth factor receptor 2 (VEGFR2)-targeted microbubbles (MBV, n 5 36, 4.14 6 1.78), rat isotype immunoglobulin G2 (IgG2) control antibody-targeted microbubbles (MBC, n 5 38, 1.95 6 0.78) and untargeted microbubbles (MBU, n 5 39, 1.79 6 0.45). The VEGFR2 CMPR is significantly different (p , 0.001) from the MBC and MBU CMPRs.

mouse embryo would appear to be a valid surrogate for tumor models. Second, we examined the feasibility of performing targeted contrast imaging of the mouse embryo vasculature using high-frequency ultrasound with the aim of investigating the significance of targeted microbubble binding, testing for factors that may influence binding and assessing the variability of our measures within and across embryos. In all cases, including control bubbles (both MBC and MBU), a CMPR . 1 was achieved. This is expected, as the microbubble concentration in the blood pool is somewhat reduced after a burst, and as evident from the time-intensity curves, microbubble intensity slowly diminishes over time. The control groups therefore provide a baseline against which the binding signal can be compared. CMPRs indicated that in all stages except E15.5, VEGFR2-targeted binding was significantly higher than that of MBC and MBU bubbles (p 5 0.003 or less). Our twofold increase in binding signal is a promising finding, being similar to that observed in various pre-clinical studies of angiogenesis (Moestue et al. 2012; Pysz et al. 2012; Willmann et al.

397

2008). The singularity of the E15.5 embryos, meanwhile, demands further attention. It is possible that VEGFR2 expression levels at E15.5 are altered or reduced compared with later gestational days and may not be sufficient to achieve the stoichiometry required for significant microbubble binding compared with the control bubbles (Greene et al. 2011). We also speculate that challenges with injecting the youngest embryos in a consistent manner may have contributed to this outcome. We therefore conclude that E16.5–E18.5 isolated embryos are optimal for future studies. Another major finding of our study is that litter, embryonic stage, heart rate and order of injection do not significantly predict CMPR. As a result, the only influencing factor is microbubble type (p , 0.001), with the compiled CMPR for MBV (4.14 6 1.78) being at least twofold greater than those for MBC (1.95 6 0.78) and MBU (1.79 6 0.45), suggesting that bubble binding is not influenced by outside biologic factors. We also assessed the variation in our ultrasound measurements. The average variations in PE, wash-in rate and TTP perfusion parameters between embryos were fairly small (,20%), with a considerably larger COV (38%) for t. Analysis of the CMPR COVs revealed, on average, 33% variation between embryos within a group (sorted according to bubble type). However, less variability was observed across regions of interest in a single embryo, on average 15%. This implies that although there may be variation between animals, exact placement of ROIs within the brain may be relaxed. This allows positioning and orientation of the embryo to be more flexible. In an effort to minimize experimental variability, we controlled for disparities in needle insertion and in microbubble populations by monitoring and adjusting initial vial concentrations while also focusing on the ratio ultrasound signals after an initial circulation time, when we assume sources of error related to concentration and injection technique are minimized. We cannot eliminate all contributors, however, as differences in position of mice with respect to the ultrasound beam and in embryo size and blood volume are likely, and irregularities in physiology and molecular expression patterns are expected as well. A number of limitations to our study can be identified. First, because the maturation of the placenta is not complete until E14.5 (Watson and Cross 2005), injection of contrast agents through a placental labyrinth is limited to embryos of mid- to late gestational age. Furthermore, the embryos are viable for a limited time, and experiments were confined to a 4-h window. Changes to the physiology of the animal could be expected during this time as it does not have a regular supply of oxygen and nutrients. Although we did not observe a significant impact of heart rate or injection order on bubble binding,

398

Ultrasound in Medicine and Biology

Volume 40, Number 2, 2014

Fig. 6. Vascular endothelial growth factor receptor 2 (VEGFR2) immunofluorescence staining of embryonic day (E) 15.5 embryos. Pronounced VEGFR2 expression is observed within vasculature-rich areas, including the heart, lung and liver. Magnification (2.53, dotted line) of the brain reveals staining of vascular structures within the entire brain (arrowheads). R 5 ribs, Li 5 liver, Ht 5 heart, CP 5 cerebellar primordium, CA 5 cerebral aqueduct, MO 5 medulla oblongata, MV 5 mesencephalic vesicle, M 5 midbrain. Scale bars indicated on figure.

it is worth investigating methods for maintaining embryo viability (e.g., humidified gas chamber [Garcia et al. 2011]), as the heart rates observed in our isolated animals were much lower than those previously observed in vivo (169–233 beats per minute for E10.5–E18.5) (Corrigan et al. 2010). In future studies, it would be valuable to track these embryonic heart rates throughout the imaging sessions as a measure of embryo viability, although hypoxic conditions (as sometimes found in tumors [H€ ockel and Vaupel 2001]) are not expected to drastically alter microbubble behavior (Mullin et al. 2011). We must also assume that shear stresses within these embryos are similar to those found in tumors, as there are few successful methods for measuring flow velocity profiles or shear stresses in vivo (Buchanan et al. 2013), whether for embryos (Jones et al. 2004) or for tumors (Monsky et al. 2002). Most importantly, we did not attempt to quantify the immunofluorescence staining for correlative purposes because of the non-linear amplification techniques required to detect our endothelial receptor. For full validation of this model, a comparison of molecular ultrasound measurements with established methods of biomarker quantification (e.g., half-maximal inhibitory concentrations) will be necessary and is the focus of future endeavors. CONCLUSIONS Our results indicate that the technical aspect of introducing microbubbles into the living embryo vasculature is achievable and reproducible. It is clear that the robust nature of this embryo model presents a unique opportunity for testing the performance of ultrasound molecular targeting. If we expand the scope of our work to include transgenic models of modified endogenous biomarker

expression, these methods may prove instrumental in assessing the quantitative measurement capabilities of molecular ultrasound and in optimizing current detection techniques. Application of these techniques might also be extended for the purposes of examining physiologic and molecular processes in embryonic models of vascular development and disease. This, in turn, might promote microbubble-enhanced ultrasound imaging as a clinical tool in the detection and localization of tumors and atherosclerosis, in the monitoring of molecular markers expressed on activated and inflamed endothelium (Kiessling et al. 2012) and in the assessment of patient suitability for various therapies and of progression and response to treatment. Acknowledgments—The authors thank Melissa Yin for assistance with microbubble handling and Ross Williams for contributions to the perfusion analysis and preparation of the article. This work was supported by the Terry Fox Programme of the National Cancer Institute of Canada.

REFERENCES Allen Institute for Brain Science. Allen developing mouse brain atlas. November 9, 2012. Available at: http://developingmouse. brain-map.org/data/Kdr.html?ispopup5true. Anderson CR, Hu X, Zhang H, Tlaxca J, Decleves AE, Houghtaling R, Sharma K, Lawrence M, Ferrara KW, Rychak JJ. Ultrasound molecular imaging of tumor angiogenesis with an integrin targeted microbubble contrast agent. Invest Radiol 2011;46:215–224. Aristiz
abal O., Williamson R., Turnbull D.H., 12 A-4 In vivo 3-D contrast-enhanced imaging of the embryonic mouse vasculature. In: Proceedings, 2007 Ultrasonics Symposium, New York City, October 28–31 2007. New York: IEEE, 2007:1073. Bartelle BB, Berr
ıos-Otero CA, Rodriguez JJ, Friendland AE, Aristiz
abal O, Turnbull DH. Novel genetic approach for in vivo vascular imaging in mice. Circ Res 2012;110:938–947. Buchanan CF, Voight EE, Szot CS, Freeman JW, Vlachos PP, Rylander MN. Three-dimensional microfluidic collagen hydrogels for investigating flow-mediated tumor-endothelial signaling and vascular organization. Tissue Engineering Part C Methods 2013 [EPub ahead of print].

VEGFR2-targeted molecular imaging in mouse embryos d J. M. DENBEIGH et al. Corrigan N, Brazil D, McAuliffe F. High-frequency ultrasound assessment of the murine heart from embryo through to juvenile. Reprod Sci 2010;17:147–157. Cosgrove D, Lassau N. Imaging of perfusion using ultrasound. Eur J Nuclear Med Mol Imaging 2010;37:S65–S85. Deshpande N, Lutz AM, Ren Y, Foygel K, Tian L, Schneider M, Pai R, Pasricha PJ, Willmann JK. Quantification and monitoring of inflammation in murine inflammatory bowel disease with targeted contrast-enhanced US. Radiology 2012;262:172–180. Deshpande N, Ren Y, Foygel K, Rosenberg J, Willmann JK. Tumor angiogenic marker expression levels during tumor growth: longitudinal assessment with molecularly targeted microbubbles and US imaging. Radiology 2011;258:804–811. Endoh M, Koibuchi N, Sato M, Morishita R, Kanzaki T, Murata Y, Kaneda Y. Fetal gene transfer by intrauterine injection with microbubble-enhanced ultrasound. Mol Ther 2002;5:501–508. Ferrara N. Role of vascular endothelial growth factor in regulation of physiological angiogenesis. Am J Physiol Cell Physiol 2001;280: C1358–C1366. Fischer T, Thomas A, Tardy I, Schneider M, H€unigen H, Custodis P, Beyersdorff D, Plendl J, Schnorr J, Diekmann F, Gemeinhardt O. Vascular endothelial growth factor receptor 2-specific microbubbles for molecular ultrasound detection of prostate cancer in a rat model. Invest Radiol 2010;45:675–684. Garcia MD, Udan RS, Hadjantonakis AK, Dickinson ME. Live imaging of mouse embryos. Cold Spring Harbor Protocols 2011;2011. pdb. top104. Gessner R, Dayton PA. Advances in molecular imaging with ultrasound. Mol Imaging 2010;9:117–127. Greene JM, Dunaway CW, Bowers SD, Rude BJ, Feugang JM, Ryan PL. In vivo monitoring of fetoplacental Vegfr2 gene activity in a murine pregnancy model using a Vegfr2-luc reporter gene and bioluminescent imaging. Reprod Biol Endocrinol 2011;9:51. H€ ockel M, Vaupel P. Tumor hypoxia: definitions and current clinical, biologic, and molecular aspects. J Natl Cancer Inst 2001;93: 266–276. Inaba Y, Lindner JR. Molecular imaging of disease with targeted contrast ultrasound imaging. Transl Res 2012;159:140–148. Jansson T, Persson HW, Lindstr€om K. Estimation of blood perfusion using ultrasound. Proc Inst Mech Eng Part H 1999;213:91–106. Jones EAV, Baron MH, Fraser SE, Dickinson ME. Measuring hemodynamic changes during mammalian development. Am J Physiol 2004; 287:H1561–H1569. Kaufman MH, Bard JBL. The anatomical basis of mouse development. San Diego: Elsevier Science; 1999. Kiessling F, Fokong S, Koczera P, Lederle W, Lammers T. Ultrasound microbubbles for molecular diagnosis, therapy, and theranostics. J Nucl Med 2012;53:345–348. Klibanov A. Preparation of targeted microbubbles: Ultrasound contrast agents for molecular imaging. Med Biol Eng Comput 2009;47: 875–882. Kulandavelu S, Qu D, Sunn N, Mu J, Rennie MY, Whiteley KJ, Walls JR, Bock NA, Sun JCH, Covelli A, Sled JG, Adamson SL. Embryonic and neonatal phenotyping of genetically engineered mice. ILAR J 2006;47:103–117. Kusuoka H, Hoffman JIE. Advice on statistical analysis for circulation research. Circ Res 2002;91:662–671. Lampaskis M, Averkiou M. Investigation of the relationship of nonlinear backscattered ultrasound intensity with microbubble concentration at low MI. Ultrasound Med Biol 2010;36:306–312.

399

Liang HD, Blomley MJK. The role of ultrasound in molecular imaging. Br J Radiol 2003;76:S140–S150. Lindner JR. Microbubbles in medical imaging: current applications and future directions. Nat Rev Drug Discov 2004;3:527–532. Lyshchik A, Fleischer AC, Huamani J, Hallahan DE, Brissova M, Gore JC. Molecular imaging of vascular endothelial growth factor receptor 2 expression using targeted contrast-enhanced high-frequency ultrasonography. J Ultrasound Med 2007;26: 1575–1586. Moestue SA, Gribbestad IS, Hansen R. Intravascular targets for molecular contrast-enhanced ultrasound imaging. Int J Mol Sci 2012;13: 6679. Monsky WL, Carreira CM, Tsuzuki Y, Gohongi T, Fukumura D, Jain RK. Role of host microenvironment in angiogenesis and microvascular functions in human breast cancer xenografts: Mammary fat pad vs cranial tumors. Clin Cancer Res 2002;8:1008. Mullin L, Gessner R, Kwan J, Kaya M, Borden MA, Dayton PA. Effect of anesthesia carrier gas on in vivo circulation times of ultrasound microbubble contrast agents in rats. Contrast Media Mol Imaging 2011;6:126–131. Pysz MA, Guracar I, Tian L, Willmann JK. Fast microbubble dwell-time based ultrasonic molecular imaging approach for quantification and monitoring of angiogenesis in cancer. Quant Imaging Med Surg 2012;2:68–80. Schneider M. Ultrasound Contrast Agents. In: Kiessling F, Pichler BJ, (eds). Small animal imaging: Basics and practical guide. Berlin: Springer; 2011. p. 226–228. Shojaei F. Anti-angiogenesis therapy in cancer: Current challenges and future perspectives. Cancer Lett 2012;320:130–137. Stewart VR, Sidhu PS. New directions in ultrasound: Microbubble contrast. Br J Radiol 2006;79:188–194. Sugimoto K, Moriyasu F, Negishi Y, Hamano N, Oshiro H, Rognin NG, Yoshida T, Kamiyama N, Aramaki Y, Imai Y. Quantification in molecular ultrasound imaging: A comparative study in mice between healthy liver and a human hepatocellular carcinoma xenograft. J Ultrasound Med 2012;31:1909–1916. Voigt JU. Ultrasound molecular imaging. Methods 2009;48:92–97. Warram JM, Sorace AG, Saini R, Umphrey HR, Zinn KR, Hoyt K. A triple-targeted ultrasound contrast agent provides improved localization to tumor vasculature. J Ultrasound Med 2011;30: 921–931. Watson ED, Cross JC. Development of structures and transport functions in the mouse placenta. Physiology 2005;20:180–193. Whiteley K.J., Pfarrer C.D. and Adamson S.L., Vascular corrosion casting of the uteroplacental and fetoplacental vasculature in mice. In: Methods in molecular medicine: Vol. 121. Placenta and trophoblast: Methods and protocols. Totowa, NJ: Humana Press, 2006:371. Wijkstra H., Smeenge M., de la Rosette J., Pochon S., TardyCantalupi K. and Tranquart F., Targeted microbubble prostate cancer Imaging with BR55. In: Proceedings, 17th European Symposium on Ultrasound Contrast Imaging, Rotterdam, The Netherlands, 19–20 2012. Willmann JK, Paulmurugan R, Chan K, Gheysens O, Rodriguez-Porcel M, Lutz AM, Chen IY, Chen X, Gambhir SS. US imaging of tumor angiogenesis with microbubbles targeted to vascular endothelial growth factor receptor type 2 in mice. Radiology 2008;246:508–516. Yamada M, Hatta T, Otani H. Mouse exo utero development system: Protocol and troubleshooting. Congenit Anom (Kyoto) 2008;48: 183–187.