Quantitative Ultrasound Molecular Imaging

Ultrasound in Med. & Biol., Vol. -, No. -, pp. 1–19, 2015 Copyright Ó 2015 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.2015.04.011

d

Original Contribution QUANTITATIVE ULTRASOUND MOLECULAR IMAGING JAMES SHUE-MIN YEH,*yz CHARLES A. SENNOGA,zx ELLEN MCCONNELL,* ROBERT ECKERSLEY,z MENG-XING TANG,{ SUSSAN NOURSHARGH,*k JOHN M. SEDDON,x DORIAN O. HASKARD,* and PETROS NIHOYANNOPOULOS*y * National Heart and Lung Institute, Imperial College London, London, UK; y Department of Cardiology, Hammersmith Hospital, London, UK; z Imaging Sciences Department, Medical Research Council, Imperial College London, London, UK; x Department of Chemistry, Imperial College London, London, UK; { Department of Bioengineering, Imperial College London, London, UK; and k William Harvey Research Institute, Queen Mary, University of London, London, UK (Received 15 August 2014; revised 10 March 2015; in final form 21 April 2015)

Abstract—Ultrasound molecular imaging using targeting microbubbles is predominantly a semi-quantitative tool, thus limiting its potential diagnostic power and clinical applications. In the work described here, we developed a novel method for acoustic quantification of molecular expression. E-Selectin expression in the mouse heart was induced by lipopolysaccharide. Real-time ultrasound imaging of E-selectin expression in the heart was performed using E-selectin-targeting microbubbles and a clinical ultrasound scanner in contrast pulse sequencing mode at 14 MHz, with a mechanical index of 0.22–0.26. The level of E-selectin expression was quantified using a novel time– signal intensity curve analytical method based on bubble elimination, which consisted of curve-fitting the biexponential equation Itissue ðtÞ 5 Af e2lf t 1Ar e2lr t to the elimination phase of the myocardial time–signal intensity curve. Ar and Af represent the maximum signal intensities of the retained and freely circulating bubbles in the myocardium, respectively; lr and lf represent the elimination rate constants of the retained and freely circulating bubbles in the myocardium, respectively. Ar correlated strongly with the level of E-selectin expression (jrj.0:8), determined using reverse transcriptase real-time quantitative polymerase chain reaction, and the duration of post-lipopolysaccharide treatment—both linearly related to cell surface E-selectin protein (actual bubble target) concentration in the expression range imaged. Compared with a conventional acoustic quantification method (which used retained bubble signal intensity at 20 min post-bubble injection), this new approach exhibited greater dynamic range and sensitivity and was able to simultaneously quantify other useful characteristics (e.g., the microbubble half-life). In conclusion, quantitative determination of the level of molecular expression is feasible acoustically using a time–signal intensity curve analytical method based on bubble elimination. (E-mail: p. [email protected]) Ó 2015 World Federation for Ultrasound in Medicine & Biology. Key Words: Molecular imaging, Targeted microbubbles, Contrast agent, Contrast echocardiography, Echocardiography, Ultrasound imaging, Quantification, Time–signal intensity curve, Microbubble elimination, E-Selectin.

inflammation, angiogenesis and thrombosis (Yeh 2010), indicating its potential for clinical applications. The imaging technique, however, remains predominantly a semi-quantitative tool. Acoustic quantification has been assessed against independent (non-acoustic) methods, such as semi-quantitative immunohistochemistry and fluorescence immunohistochemistry (Behm et al. 2008; Kaufmann et al. 2007b; Korpanty et al. 2007; Lee et al. 2008; Leong-Poi et al. 2005; Liu et al. 2011; Mancini et al. 2013; Palmowski et al. 2008, 2009; Weller et al. 2003; Xie et al. 2011), illustrating at best the semiquantitative capability of the imaging technique. Only a handful of studies have used highly quantitative independent assays to assess acoustic quantification; these included quantitative radioactive assay (Bin et al. 2008), quantitative

INTRODUCTION Ultrasound molecular imaging has been achieved using echogenic microbubbles targeting molecules of interest (Lindner et al. 2001). After intravenous administration, the targeting bubbles circulate and accumulate in regions expressing the molecules of interest, depicted on ultrasound images as areas of bright signals localizing the molecules. This technique has allowed ultrasound molecular imaging of pathophysiological processes such as

Address correspondence to: Petros Nihoyannopoulos, National Heart and Lung Institute, Imperial College London, Du Cane Road, London W12 0NN, UK. E-mail: [email protected] 1

2

Ultrasound in Medicine and Biology

fluorescence imaging (Saini et al. 2013), quantitative chemiluminescence immunoblotting (Deshpande et al. 2011; Lyshchik et al. 2007) and quantitative fluorescence immunohistochemistry (Bachawal et al. 2013). However, the degree of correlation between the acoustic and nonacoustic methods varied from moderate to strong (Bachawal et al. 2013; Bin et al. 2008; Deshpande et al. 2011) or was not calculated (Korpanty et al. 2007; Lyshchik et al. 2007; Saini et al. 2013). Surprisingly, none provided scatterplots that could be used to examine the correlations claimed (Bachawal et al. 2013; Bin et al. 2008; Deshpande et al. 2011; Korpanty et al. 2007; Lyshchik et al. 2007; Saini et al. 2013). The conventional method of acoustic quantification of molecular expression in a tissue is based on the signal intensity of target-bound (retained) bubbles in the tissue at one time point after bubble administration, when the bubbles have attached to the molecular targets and unbound (freely circulating) bubbles have sufficiently (Lindner et al. 2000; Rychak et al. 2007; Stieger et al. 2008) or almost completely (Hernot et al. 2012; Tlaxca et al. 2013) cleared from the blood pool. Where the freely circulating bubble signal in the tissue remains significant at the time point chosen (more likely at earlier time points or with larger bubble dosages), it can be removed by subtraction to obtain the retained bubble-only signal (Lindner et al. 2000; Rychak et al. 2007; Stieger et al. 2008). This requires the use of high-power ultrasound that destroys all (retained plus freely circulating) bubbles in the acoustic field, followed by further imaging to obtain the freely circulating bubble-only signal as circulating bubbles re-fill the acoustic field. The subtraction step is not mandatory if the residual circulating bubble signal is minimal. Bubble signals can be obtained using bubble-specific or non-bubble specific imaging modes at high or low acoustic powers (Lindner et al. 2000; Rychak et al. 2007; Stieger et al. 2008). Although the conventional quantification method is widely used, it has not been formally validated. Limitations of the method include the arbitrary nature of the time point chosen for signal analysis, which varies widely in practice (ranging from 2 to 15 min post-bubble administration), yielding different values in the same subject. Furthermore, based on one single time point analysis, it is prone to error. Other quantification methods have been described; most are based on analysis of the time–signal intensity curve (TIC) (Behm et al. 2008; Carr et al. 2011; Chen et al. 2012; Fisher et al. 2002; Lindner et al. 1998; Sirsi et al. 2012). However, they remain untested for molecular quantification. Furthermore, they often require curve-fitting the entire TIC, which makes them prone to errors caused by signal saturation/bubble

Volume -, Number -, 2015

cloud attenuation in the early time points of the TIC, where bubble concentrations are high. As the level of molecular expression may reflect the state, type, prognosis or response to therapy of a disease, the ability of the imaging technique to quantitatively measure the level of molecular expression would increase its potential diagnostic power and breadth of clinical applications. In this study, we developed and tested a novel TIC-based method for acoustic quantification of molecular expression that does not require curve fitting the entire TIC. METHODS Antibodies MES-1 monoclonal antibody (mAb), a rat IgG2a,k against mouse Esel (Reynolds et al. 2006) and its F(ab0 )2 fragments were provided by D. Brown (UCB Celltech, UK). Reduced MES-1 F(ab0 )2 (containing two thiol groups per F(ab0 )2 from tris(2-carboxyethyl)phosphine hydrochloride reduction) was prepared as described in the Appendix. MEC13.3 mAb, a rat IgG2a,k against mouse platelet endothelial cell adhesion molecule-1 (PECAM-1) (BD Biosciences, UK), rat IgG2a,k isotype-negative control mAb (BD Biosciences) and biotinylated rabbit mAb against rat IgG2a (Vector Laboratories, UK) were purchased. Animals Wild-type (WT) mice were adult male C57BL/6J (Charles River, UK). Esel knockout (KO) mice were adult male Esel homozygote KO on C57BL/6J background (Labow et al. 1994), bred locally from mice donated by K. Norman and P. Hellewell (University of Sheffield, UK). All animal work was carried out under licences granted by the Home Office under the Animals (Scientific Procedures) Act 1986, with ethical approval obtained from Imperial College London’s Ethical Review Panel. Mouse model of lipopolysaccharide-induced inflammation (experimental endotoxemia) Wild-type and Esel KO mice were treated with 50 mg lipopolysaccharide (LPS) from Escherichia coli 0111:B4 (Sigma-Aldrich, UK), made up to a 200-mL volume in normal saline, by intraperitoneal injection to induce systemic inflammation (Eppihimer et al. 1996). Immunohistochemistry Immunohistochemistry was performed on acetonefixed cryosections of freshly harvested hearts of WT (with or without LPS pre-treatment) and Esel KO (LPS pre-treated) mice, using a standard protocol detailed in the Appendix. The primary antibodies used were MES-1 (for Esel), MEC13.3 (for PECAM-1, an

Quantitative US imaging d J. SHUE-MIN YEH et al.

endothelial marker) and rat IgG2a,k isotype-negative control mAb. Color was developed using 3,30 -diaminobenzidine with hematoxylin counterstaining. Reverse transcriptase real-time quantitative polymerase chain reaction Wild-type mice were pre-treated with LPS as described above. The duration between LPS treatment and animal sacrifice for immediate tissue harvesting was denoted as LPSTime. RNA extraction and reverse transcriptase real-time quantitative polymerase chain reaction (RT-qPCR) for Esel and hypoxanthine phosphoribosyltransferase I (HPRT-I, a housekeeping gene) in the hearts were performed; only the apical halves of the harvested hearts were used to avoid measuring Esel from the great vessels. Note that Esel is expressed on endothelial cells and not on other cell types. RT-qPCR was performed using standard protocols and kits according to the manufacturers’ instructions. All PCRs were carried out in triplicate on a 96-well plate. Means of the replicates were used. Esel mRNA concentration was expressed as percent HPRT-I. Further details are provided in the Appendix. Microbubble preparation Esel-targeting microbubbles were prepared by grafting MES-1 F(ab0 )2 to phospholipid bubbles, as detailed in the Appendix. Briefly, native bubbles were first produced by sonicating C3F8-sparged aqueous suspension containing 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC; Avanti Polar Lipids, Alabaster, AL, USA), 1,2distearoyl-sn-glycero-3-phosphoethanolamine-N-(maleimide[polyethylene glycol]-2000) (DSPE-PEG2000-Mal; Avanti Polar Lipids), monostearate poly(ethylene)glycol (PEG40-stearate; Sigma-Aldrich) and the fluorescent dye 1,10 -dioctadecyl-3,3,30 ,30 -tetramethylindocarbocyanine perchlorate (DiI; Molecular Probes, UK) at a 75:9:14:2 molar ratio. Reduced MES-1 F(ab0 )2 containing two thiol groups per F(ab0 )2 was then linked to the maleimide on the bubble shell outer surface by maleimide–thiol conjugation to produce the Esel-targeting bubbles. The conjugation reaction ratio was 4.338 3 106 F(ab0 )2 molecules per bubble. The Esel-targeting bubbles had a mean (standard error of the mean [SEM]) diameter of 2.2 (0.2) mm; 98.6% or 100% of the bubbles were less than 6 or 10 mm in diameter, respectively. These Esel-targeting bubbles have previously been validated in vitro and in vivo and found to be specific and effective for real-time ultrasound molecular imaging of Esel in the heart and other tissues in mice (Yeh 2010). Ultrasound molecular imaging Fifteen WT and eight Esel KO mice were imaged. All were pre-treated with LPS. Ultrasound imaging was performed under intraperitoneal xylazine/ketamine gen-

3

eral anesthesia. The Acuson Sequoia 512 clinical ultrasound system (Siemens, Mountain View, CA, USA) equipped with a 15L8-s linear array transducer was used. Gel (Gel for Ultrasonic & Electrical Transmission, Henleys Medical, UK) was coupled between the shaven skin and the transducer. Electrocardiogram (ECG) electrode pads (Ambu Blue Sensor P, Ambu, UK) were applied to the paws and connected to the ultrasound machine. Imaging was maintained in the parasternal short axis (PSA) view of the heart by fixing the transducer in position with a free-standing clamp. Contrast pulse sequencing (CPS) mode imaging at 14-MHz and low power (mechanical index [MI] 5 0.22–0.26) was used. The focus point was placed at the level of the inferior wall, which was typically at a depth of 10 mm (mean 5 9.5 mm, median/mode 5 10 mm, range 5 7.5–12.5 mm). Magnified images of the heart with enhanced resolution were obtained by activating the ‘‘regional expansion selection’’ (RES) function. The RES box (z7 3 7 mm) spanned a depth of 4–16 mm amongst the animals used in this study. Gain and other settings were fixed: time gain 5 0% (depth gain control sliders at extreme left position), CPS gain 5 8, fundamental 2D gain 5 15 dB and dynamic range 5 55 dB. TEQ tissue equalization technology was not used. Bubble signals were presented in heated object scale. Baseline images (before bubble injection) were first acquired, then a stopwatch was started. At 10 s on the stopwatch, 108 Eseltargeting bubbles (in 100-mL volume made up with normal saline) were injected as an intravenous bolus over 1–2 s through a cannula in the tail vein. This was followed by a 100-mL normal saline flush at 20 s on the stopwatch. Continuous ultrasound imaging was performed and recorded as 3-s digital clips, starting at time 0 on the stopwatch, and repeated at pre-defined time intervals to follow the whole lifetime of the bubble bolus. Animals received only one dose of bubbles to eliminate any carryover effects from previous bubble dosing (e.g., blocking of Esel binding sites). The duration between LPS treatment and the administration of bubbles was denoted as LPSTime. The heart rate (HR) for each animal was averaged from all HRs recorded during cardiac imaging. All animals were sacrificed at the end. Further details are provided in the Appendix. For data plotting and analysis, the time of bubble administration (10 s on the stopwatch) was taken as time 0. Acoustic quantification End-diastolic image frames of the heart in the PSA view were selected and aligned; those that could not be aligned (e.g., because of large movement artifact) were excluded. The video densitometric method was used to quantify CPS bubble signal intensities in regions of interest (ROIs) off-line, using YABKO software

4

Ultrasound in Medicine and Biology

(Charlottesville, VA, USA). One ROI was placed in the anterior wall of the myocardium (M) and another adjacent to it in the left ventricular (LV) cavity (C). M and C spanned mean depths of 4.5–6.5 mm (range: 3–9 mm) and 5.5–6.5 mm (range: 3.5–9.5 mm), respectively, from the probe surface. These regions were chosen because they were consistently least/minimally affected by ultrasound attenuation in all animals. The CPS video signal intensities (VIs) were ‘‘linearised’’ by logdecompression. The formula used for this by the YABKO software was 
Linearised VI 5 255310

VI22553dynamic range 20 255

according to the manufacturer. Linearised VI(I) was expressed in arbitrary acoustic units (AU). I values of several end-diastolic image frames within the 3-s recording period of each time point were averaged (except those in the first 6 s post-bubble administration because of the rapid changing signals), then subtracted by average I of the baseline (before bubble administration) images. Novel method. TIC analysis based on bubble elimination. TICs of the myocardium and LV cavity were constructed by plotting the baseline-subtracted I values against time post-bubble administration. The exponential functions of targeting bubble elimination, Itissue ðtÞ 5 Af e2lf t 1Ar e2lr t and Ic (t) 5Ace2lct (see Theory of Mathematical Model for derivation of these eqns [13] and [7], respectively), were then curve-fitted to the elimination phase of the TICs of the myocardium and LV cavity, respectively, using Prism 5.01 software (GraphPad, La Jolla, CA, USA). To satisfy the condition of linear acoustics (assumption 2 of the mathematical model), only time points with signal intensities in the linear acoustic range (where bubble concentrations had much decreased such that signal saturations/bubble cloud attenuations were minimal) were curve-fitted. In this study, the linear acoustic range was taken as I 5 0–5 AU, approximated from an in vitro study of bubble concentration vs. signal intensity in a beaker, using the same ultrasound settings and similar imaging depths (7–13 mm in vitro vs. 4–16 mm in vivo) as used for in vivo imaging (Supplemental Fig. A1, see online version at http://dx.doi.org/10.1016/j.ultrasmedbio. 2015.04.011). Time points with signal intensities #5 AU in the ascending and signal-saturated/attenuated (because of the initial high bubble concentrations) parts of the TIC were excluded (see Results for a description of the various parts of a TIC). Effectively, the time points curve-fitted were those with signal intensities #5 AU in the descending part of the TIC. This was done by limiting the earliest time point for curve fitting, which varied

Volume -, Number -, 2015

amongst animals and was frequently z2–6 min postbubble administration. In Figure 1 are examples from a WT animal and KO animal. Further details are provided in the Appendix. Af, Ar, Ac, lf, lr and lc were parameters of the curve fit. The true zero time for the start of the elimination phase, which determines Ar, Af and Ac, is unknown. Thus, for consistency, this zero time was arbitrarily taken as time 0 when the bubbles were injected. Ar represented the maximum signal intensity (proportional to maximum concentration) of the retained bubbles in the myocardium, from elimination phase analysis. It was assessed for quantification of the targeted molecule (Esel) expression in the myocardium. Af and Ac were the maximum signal intensities (proportional to maximum concentrations) of the freely circulating bubbles in the myocardium and LV cavity, respectively, from elimination phase analysis. The relative myocardial blood volume (rMBV) was estimated as zAf/Ac. lr was the elimination rate constant of the retained bubbles in the myocardium, while lf and lc were the elimination rate constants of the freely circulating bubbles in the myocardium and LV cavity, respectively. The acoustic half-life of the retained or circulating bubbles in the myocardium, or circulating bubbles in the LV cavity (central blood pool), was calculated as ln 2/l, where l 5 lr, lf or lc, respectively. Conventional method. Retained bubble signal intensity at a late time point. Baseline-subtracted I of the myocardium at 20 min 10 s post-bubble administration (R20) was assessed for quantification of Esel expression in the myocardium. This represented the retained bubble signal intensity in the myocardium at 20 min 10 s postbubble administration. As the residual circulating bubble signal intensity in the LV cavity (blood pool) was absent/ minimal by z20 min post-bubble administration in all animals, the circulating bubble signal intensity in the myocardium by this time was considered negligible; therefore, its subtraction from R20 was not required (e.g., a circulating bubble signal intensity ,0.1 AU in the LV cavity would contribute an undetectable circulating bubble signal intensity of ,0.005–0.024 AU in the myocardium, assuming a rMBV of 5%–24%) (Coulden 1997; Streif et al. 2005; Waller et al. 2000). Data analysis in acoustic quantification of Esel expression in the heart. Ar and R20 were correlated against the levels of Esel expression in the heart, in terms of LPSTime and Esel mRNA levels determined by qRTPCR. Both were shown to be suitable surrogate quantifiers for the cell surface Esel protein (actual bubble target) concentration (see Reverse Transcriptase Real-Time Quantitative Polymerase Chain Reaction in the Results). Because of the relatively long duration of each imaging study (mean z 45–60 min from the time of bubble

Quantitative US imaging d J. SHUE-MIN YEH et al.

5

Fig. 1. Novel time–signal intensity curve (TIC) analysis based on bubble elimination. TICs of the myocardium and left ventricular (LV) cavity from a wild-type (WT) (a, b) and Esel knockout (KO) (c, d) mouse, showing biphasic and monophasic signal decay in the myocardium (tissue) and LV cavity (central blood pool) in the elimination phase, respectively. Linear (top row) and semi-logarithmic (bottom row) plots are shown. The exponential functions of bubble elimination are curve-fitted (solid lines) only to signal intensities in the linear acoustic range in the elimination phase (solid points). Af, Ar, Ac and lf, lr, lc are parameters of the curve fit, where Af, Ar and Ac are intercepts at t 5 0 and lf, lr and lc are slopes (on semi-logarithmic plot) of the respective exponents. As shown graphically, the myocardial bubble signal can be decomposed into its separate components—circulating and retained bubble signals—allowing quantification of molecular expressions from the latter. Error bars represent standard deviations.

administration to tissue harvesting) with respect to the kinetics of Esel expression in the mouse model used, it was not possible to correlate Ar or R20 of individual animals against its own Esel expression measured ex vivo by RT-qPCR. This was because Esel expression decreased rapidly with time such that the retained bubble signal intensities reflected Esel expression around the time of bubble administration rather than tissue harvest. The solution was to use a standard curve of LPSTime vs. Esel mRNA level from the hearts of 42 mice to estimate the Esel mRNA level in the heart of each animal imaged at known LPSTime. Finally, as Esel expression in the myocardium was essentially global and uniform in the mouse model (see Immunohistochemistry in the Results), the quantification of Esel expression from the anterior myocardial wall by ultrasound (Ar and R20) and that from the apical half of the heart by qRT-PCR were regarded as equivalent. Statistics Pearson’s correlation and linear or non-linear regression analysis was performed, where appropriate.

Student’s t-test or analysis of variance with Tukey’s post hoc analysis was used for significance testing, where appropriate. p , 0.05 was taken to indicate statistical significance. THEORY: MATHEMATICAL MODEL OF TARGETING MICROBUBBLE ELIMINATION IN VIVO Elimination of circulating bubbles from the blood pool The general equation for a kinetic process, describing how fast the mass (X) of a compound decreases per unit time (t), is (Riviere 1999) dXðtÞ 5 2KX n dt

(1)

where K is a constant reflecting the rate of the kinetic process, and n is the order of the process. Pharmacokinetic studies suggest that the elimination of circulating bubbles from the blood pool in vivo is a first-order (linear) kinetic process (Yeh 2010). Thus, if we let n 5 1, eqn (1) becomes

6

Ultrasound in Medicine and Biology

dXc ðtÞ 5 2Kc Xc dt

(2)

where subscript c refers to the circulating bubbles in the blood pool. If we integrate eqn (2), we have  ðt ðt
dXc ðtÞ (3) dt 5 ð2Kc Xc Þdt dt 0 0 Solve eqn (3) by Laplace transformation to obtain Xc at time t, Xc ðtÞ 5 Xc0 e2Kc t

Xc ðtÞ Xc0 2Kc t 5 e V V Cc ðtÞ 5 Cc0 e

process, and the bubbles do not transfer from one population to another in the elimination phase, then eqn (1) can be extended as dXtissue ðtÞ 5 2Kf Xf 2Kr Xr dt

(8)

Solving by integration and Laplace transformation, followed by conversion of bubble mass to concentration and signal intensity as before, yields  ðt
ðt   dXtissue ðtÞ dt 5 2Kf Xf 2Kr Xr dt (9) dt 0 0

(4)

where Xc0 and Xc(t) are the mass of the circulating bubbles in the blood pool at times 0 and t, respectively. Convert mass X to concentration (C) by division with volume (V) in eqn (4),

2Kc t

Volume -, Number -, 2015

(5)

where Cc0 and Cc(t) are the concentration of the circulating bubbles in the blood pool at times 0 and t, respectively. Assuming linear acoustics (i.e., bubble concentration is linearly related to bubble ultrasound signal intensity), eqn (5) is rewritten as Ic ðtÞ 5 Ic0 e2lc t

(6)

Ic ðtÞ 5 Ac e2lc t

(7)

or

where Ic0 (hAc) and Ic(t) are the ultrasound signal intensities of the circulating bubbles in the blood pool at times 0 and t, respectively. lc is the rate constant for the signal intensity decay. The mono-exponential function (eqn [7]) models the elimination of circulating bubbles from the blood pool. It may be used to predict or curve-fit the elimination phase of the TIC of circulating bubbles in the blood pool, such as the LV cavity (Fig. 1).

Elimination of circulating and retained bubbles from a tissue Let us assume that in the elimination phase, the tissue contains two bubble populations: (i) bubbles that circulate freely (circulating bubbles); and (ii) bubbles that are retained in the tissue because of target/ non-specific binding (retained bubbles). If the elimination of each bubble population is a first-order kinetic

Xtissue ðtÞ 5 Xf0 e2Kf t 1Xr0 e2Kr t

(10)

Ctissue ðtÞ 5 Cf0 e2Kf t 1Cr0 e2Kr t

(11)

Itissue ðtÞ 5 If0 e2lf t 1Ir0 e2lr t

(12)

Itissue ðtÞ 5 Af e2lf t 1Ar e2lr t

(13)

where, as before, X, C, I and A are the bubble mass, concentration, signal intensity and simplified notation for I0, respectively; K and l are the rate constants; subscripts tissue, f, r and 0 refer to tissue, circulating bubbles, retained bubbles and time 0, respectively. The bi-exponential function (eqn [13]) models the elimination of circulating and retained bubbles from a tissue. It may be used to predict or curve-fit the eliminationphase TIC of targeting bubbles in a tissue of interest (Fig. 1). It separates tissue bubble signal intensity into those of the circulating and retained bubbles, each described by its respective exponential term. Note that under lowpower ultrasound imaging conditions, elimination of the retained bubbles is slower than that of the circulating ones, so the exponential term with the larger elimination rate constant is assigned to the circulating bubbles. Model assumptions 1. Both circulating and retained targeting bubbles are eliminated by first-order kinetic processes. This is supported by the exponential nature of their signal decay in the elimination-phase TIC (Riviere 1999). 2. Linear acoustics apply (i.e., bubble concentration is proportional to bubble ultrasound signal intensity). This has been reported in vitro or in vivo for free and attached bubbles at low to moderate concentrations under different ultrasound imaging modes (Lankford et al. 2006; Raher et al. 2007; Skyba et al. 1994; Yeh 2010). 3. Further accumulation of targeting bubbles in the tissue of interest (non-reticuloendothelial tissue) is

Quantitative US imaging d J. SHUE-MIN YEH et al.

7

negligible in the elimination phase. Previous intravital microscopy observations indicated that bubble retention ceased to increase significantly soon (e.g., z2 min) after bubble bolus administration (Yeh 2010). 4. Bubble internalization into cells or transmigration into tissue interstitium is infrequent, so there exists essentially only one type (kinetically homogenous) of retained bubble population in the tissue of interest. 5. Detachment of attached bubbles is infrequent, so the transfer of bubbles from the retained population to the circulating one is negligible.

Quantitative real-time ultrasound molecular imaging

Assumptions 4 and 5 are supported by our previous intravital microscopy observations of the Esel-targeting bubbles in inflamed mouse cremasters (Yeh 2010).

Imaging. Esel expression in the WT myocardium was visualised in real time, characterised by the persistence of bubble signals in the myocardium (resulting from the retention of attached bubbles) beyond the clearance of circulating bubbles from the LV cavity (blood pool). The persistence of bubble signals in the KO myocardium was low/minimal (Fig. 4), suggesting high specificity of the Esel-targeting bubbles.

RESULTS Ex vivo analysis of Esel expression Immunohistochemistry. Frozen-section immunohistochemistry revealed that Esel was expressed in the hearts of the WT animals pre-treated with LPS (n 5 35 mice, LPSTime 5 4–9 h). The spatial distribution was essentially uniform throughout the myocardium, but limited to the capillaries and post-capillary venules. Esel was undetectable in LPS pre-treated Esel KO mice pre-treated with LPS (n 5 10 mice, LPSTime 5 4–7 h) and untreated WT mice (n 5 5 mice) (Fig. 2). Reverse transcriptase real-time quantitative polymerase chain reaction. RT-qPCR revealed that the concentration of Esel mRNA in the heart decreased exponentially with time after z3 h post-LPS pre-treatment in the WT mice, reaching very low levels by z9 h (n 5 42 mice, LPSTime 5 3.2–15.7 h) (Fig. 3a). This trend was similar to that of the cell surface Esel protein concentration determined using intravenous radiolabelled mAb (Eppihimer et al. 1996) in the same strain, sex and age of mice, with the same dose and route of LPS treatment (Fig. 3b). The relationships between the concentrations of protein and mRNA (Fig. 3c, derived from the lines of best fit in Fig. 3a, b) and between the concentration of protein and LPSTime (Fig. 3b) were both curvilinear, with an approximately linear relationship in the range of LPSTime values or mRNA concentrations used in acoustic quantifications of this study (i.e., LPSTime 5 4–6 h, Esel mRNA concentration 5 50%–220% HPRT-I). Both LPSTime and Esel mRNA concentration were therefore used as surrogate quantifiers of the protein (actual bubble target) concentration in assessing acoustic quantifications of Esel.

Animals. Imaging was performed on 15 WT and 8 Esel KO animals. Three WT animals were excluded from quantitative analysis due to (i) bubble dosing error (one animal); (ii) uncertainties regarding the Esel expression level (two animals); and (iii) one of these three animals could not be adequately curve-fitted for TIC analysis, possibly due to severe attenuation artefact. Thus, quantitative analysis was performed on 12 WT and 8 KO mice. The LPSTime ranges were 3.9–6 and 4.5–5.7 h in the WT and KO groups, respectively.

Ultrasound TICs of the myocardium (tissue) and LV cavity (central blood pool). Three sequential phases were discernible from the TICs following intravenous bolus administration of the targeting bubbles (Fig. 4b, d): 1. Bolus phase: Lasting a few seconds, characterised by an initial rapid rise in bubble signal intensity, reaching a peak or saturation/attenuation level. 2. Distribution phase: Lasting up to 1–2 min (taking a few circulatory cycles to complete), characterised by a short, rapid decrease in bubble signal intensity as bubbles were diluted by mixing and distribution to the tissues. High bubble concentration resulted in signal saturation/bubble cloud attenuation, obscuring re-circulation peaks. As the bubble concentration decreased further over time, the bubble signal intensity was frequently observed to paradoxically increase as attenuation decreased, giving rise to a second lower peak within z0.5–1 min of bubble administration (arrow in inset to Fig. 4b, d). 3. Elimination phase: Lasting several minutes, characterised by a long, slow, exponential decrease in bubble signal intensities. In the LV cavity which contained only circulating bubbles, the bubble signals became undetectable sooner in the WT than KO animals because of Esel-mediated depletion of the circulating bubbles from the blood pool in the former. In all animals, the circulating bubbles were essentially undetectable by 20 min post-bubble administration. In the WT myocardium, the bubble signals persisted beyond the disappearance of the circulating bubble signals in the LV cavity, because of Esel-mediated bubble retention in the myocardium. In contrast, bubble signals in

8

Ultrasound in Medicine and Biology

Volume -, Number -, 2015

Fig. 2. Frozen-section immunohistochemistry of the heart. Representative examples from wild-type (WT) and Esel knockout (KO) mice with/without lipopolysaccharide (LPS) pre-treatment 6 h beforehand. Magnification: 2003. Low-power magnification (low mag): 403. Box 5 optical field for the 2003 magnification in the WT mouse pretreated with LPS.

the myocardium disappeared sooner than those in the LV cavity in the KO animals, as expected from an rMBV of z5%–24% (Coulden 1997; Streif et al. 2005; Waller et al. 2000) and the absence of significant non-specific bubble retention. Acoustic quantification of Esel expression. Novel method. The elimination-phase TIC of the myocardium fitted well to Itissue ðtÞ 5 Af e2lf t 1Ar e2lr t (eqn [13]) in both WT (R2 5 0.88) and KO (R2 5 0.87) animals (Table 1). Ar, the maximum retained bubble signal intensity (concentration) in the myocardium, was significantly higher in the WT than KO animals. Importantly, Ar correlated strongly with LPSTime in

the WT (r 5 20.87, p , 0.0005) but not KO (r 5 20.08, p 5 0.86) mice (Fig. 5a). It also correlated strongly with Esel mRNA concentrations (r 5 0.84, p , 0.001) (Fig. 5b). Both LPSTime values and Esel mRNA concentrations were in the range where their relationships with the cell surface Esel protein (actual bubble target) concentration were linear (see Reverse Transcriptase Real-Time Quantitative Polymerase Chain Reaction). Compared with the conventional quantification method, Ar had a greater dynamic range (Fig. 5a, b vs. Fig. 5c, d) and was able to detect and quantify lower levels of Esel expression (Supplemental Fig. A2) than R20. In essence, Ar was superior to R20.

Quantitative US imaging d J. SHUE-MIN YEH et al.

9

Fig. 3. Temporal expression of Esel in the heart. (a) Esel mRNA concentration versus duration of post-lipopolysaccharide treatment (LPSTime). Each point represents one animal. Exponential lines of best fit with 95% confidence interval (CI) are shown. (b) Cell surface Esel protein concentration versus LPSTime. Each point (error bars) represents the mean (95% CI) of 5, 5, 4 and 5 mice at LPSTime 5 3, 5, 8 and 24 h, respectively. Exponential lines of best fit and 95% CI are shown. (c) Esel mRNA concentration versus cell surface Esel protein concentration. Percentage injected dose of radioactivity per gram of tissue (%ID/g tissue).

Conventional method. R20, the retained bubble signal intensity at 20 min 10 s post-bubble administration, was significantly higher in the WT than KO animals (Table 1). It correlated strongly with LPSTime in the WT (r 5 20.78, p , 0.005) but not KO (r 5 20.05, p 5 0.9) mice (Fig. 5c). It also correlated strongly with the concentration of Esel mRNA (r 5 0.81, p , 0.005) (Fig. 5d). However, closer examination of the scatterplots revealed that R20 failed to detect or quantify low levels of Esel expression (in the range 60%–110% HPRT-I or LPSTime 5–6 h, where R20 # 0.24 AU), applicable to half of the WT animals (Supplemental Fig. A2). Acoustic quantification of other characteristics (novel method). Circulating bubbles. The eliminationphase TIC of the LV cavity (central blood pool) fitted well to Ic ðtÞ 5 Ac e2lc t (eqn [7]) in both WT (R2 5 0.78) and KO (R2 5 0.91) animals (Table 1). There was no significant difference amongst the WT lf and lc and KO lf and lc (p 5 0.93 [analysis of variance]). Changes in lf followed those in lc (Fig. 6a); both correlated with HR (Fig. 6b, c), consistent with increasing clearance of the circulating bubbles as a result of the HR-mediated increase in cardiac output (cardiac output 5 HR 3 stroke volume), delivering more bubbles per unit time to the acoustic field, reticuloendothelial tissues and lungs (the

major sites of bubble elimination). The half-life of the circulating bubbles ranged from 1 to 5 min (mean 5 2 min) (Table 1). Af and Ac were lower in the WT group than in the KO group, consistent with Esel-mediated depletion of the circulating bubbles from the blood pool. As expected, both Af and Ac decreased with decreasing LPSTime (increasing Esel expression) in the WT animals; this trend was absent in the KOs (Fig. 6d, e). In both groups of animals, changes in Af essentially followed those in Ac (Fig. 6f). This and the near equivalence of lf to lc are consistent with changing circulating-bubble concentrations in the tissue (myocardium) being linked to those in the central blood pool (LV cavity). However, the magnitude of Ac in the WT group was much lower than expected relative to Af. This was likely due to ultrasound attenuation in the LV cavity, caused by significant bubble retention in the myocardium overlying the LV cavity. This resulted in overestimations of the rMBVs using Af/Ac in WT mice, where the values (mean 5 50, range 5 3%–98%) frequently exceeded the physiological range of 5%– 24% (Coulden 1997; Streif et al. 2005; Waller et al. 2000). In contrast, estimation of rMBVs using Af/Ac in the KO mice yielded values (mean 5 17, range 5 7%– 28%) within the physiological range (Table 1).

10

Ultrasound in Medicine and Biology

Volume -, Number -, 2015

Fig. 4. Real-time ultrasound molecular imaging of Esel expression in the heart. (a, c) Sequential 14-MHz contrast pulse sequencing images of the heart in end-diastole in the parasternal short axis view, in a wild-type (WT) (a) and Esel knockout (KO) (c) mouse, both pre-treated with lipopolysaccharide. (b, d) Time–signal intensity curves of the left ventricular (LV) cavity (region of interest C) and myocardium (region of interest M) for the respective WT (b) and KO (d) animals. Error bars represent standard deviations.

Retained bubbles. The elimination rate constant of the retained bubbles in the myocardium (lr) decreased with increasing Ar (Fig. 7). The relationship was nonlinear and could be empirically fitted to an exponential function (lr 5 0:45e20:64Ar , R2 5 0.77) or sigmoidal function (lr 5 1/[1 1 100.26Ar 1 0.3], R2 5 0.87) (n 5 20 [12 WT and 8 KO mice]). This suggested that the lower the maximum retained bubble concentration (or lower targeted molecule concentration), the shorter the half-life of the retained bubbles. The half-life of the retained bubbles in the myocardium ranged from 3 to 18 min (mean 5 7 min) in the WT mice and from 1 to 14 min (mean 5 4) in the KO animals (Table 1).

DISCUSSION In this study, quantitative determination of Esel expression levels in the heart was achieved acoustically using a novel TIC analytical method based on bubble elimination. The maximum retained bubble signal intensity (Ar) correlated strongly with the level of Esel expression in terms of LPSTime or Esel mRNA concentration (jrj.0:8), both in the range linearly related to the cell surface Esel protein concentration. Although the conventional quantification method based on retained bubble signal intensity at a single late time point (R20) also allowed quantitative determination of Esel expression (jrj 5 0:8), the corresponding scatterplots revealed that

Quantitative US imaging d J. SHUE-MIN YEH et al.

11

Table 1. Acoustic quantifications WT (n 5 12) Region of interest LV cavity Myocardium

Parameter Goodness of fit (R2): Ic ðtÞ 5 Ac e2lc t Goodness of fit (R2): Itissue ðtÞ 5 Af e2lf t 1Ar e2lr t

Circulating bubbles LV cavity Ac (AU) lc (/min) Bubble half-life (min)* Myocardium Af (AU) lf (/min) Bubble half-life (min)* rMBV (zAf/Ac) Retained bubbles Myocardium Ar (AU) lr (/min) Bubble half-life (min)* Myocardium R20 (AU)

Esel KO (n 5 8)

Mean (SEM)

Range

Mean (SEM)

Range

p (WT vs. Esel KO)

0.78 (0.06) 0.88 (0.03)

0.23–0.98 0.7–0.96

0.91 (0.02) 0.87 (0.04)

0.78–0.97 0.62–0.97

0.13 0.92

9.7 (2.1) 0.5 (0.06) 1.6 (0.2) 4 (0.7) 0.45 (0.08) 2.1 (0.3) 0.5 (0.08)y

2–25.3 0.22–0.93 0.7–3.2 0.2–8.2 0.14–1.12 0.6–5.1 0.03–0.98y

36.8 (7.6) 0.5 (0.06) 1.5 (0.2) 5.4 (0.9) 0.5 (0.07) 1.6 (0.2) 0.17 (0.02)

12.6–69.5 0.29–0.76 0.9–2.4 2.8–9 0.27–0.79 0.9–2.5 0.07–0.28

,0.001 0.99 0.7 0.25 0.64 0.32 ,0.005y

2.3 (0.4) 0.13 (0.02) 6.9 (1.3) 0.46 (0.16)

0.8–4.2 0.04–0.24 2.9–17.6 0.01–1.61

0.4 (0.1) 0.28 (0.06) 4.4 (1.5) 0.06 (0.03)

0–1 0.05–0.63 1.1–14.3 20.01 to 0.24

,0.005 ,0.05 0.22 ,0.05

KO 5 knockout; LV 5 left ventricular; WT 5 wild type. * The group mean bubble half-life in vivo was calculated from ln 2/l value of individual animals and is not expected to be equal to ln 2/group mean l. y rMBV is overestimated in the WT group because of signal attenuation in the LV cavity caused by significant bubble retention in the myocardium overlying the LV cavity.

Ar exhibited a greater dynamic range, was more sensitive and was better able to quantify lower levels of Esel expression than R20. A further advantage of the novel method was that other useful characteristics, such as the retained or circulating bubble half-life, could be quantified simultaneously. The inferior performance of the conventional method in this study may, at least in part, be the result of an excessive decay of the retained bubbles from a relatively long wait time (20 min 10 s), by which time freely circulating bubbles cleared from the blood pool in all animals (i.e., when bubble signals in the LV cavity became undetectable). Such a wait time was longer than those employed by others (range 5 2–15 min, most frequently 4 or 10 min in mice [Bachawal et al. 2013; Kaufmann et al. 2007a; Leong-Poi et al. 2003; Sorace et al. 2012; Wei et al. 2014]) for the following reasons: (i) we used a higher bubble dose (1 3 108 bubbles) than did most investigators (range 5 8–1,000 3 105, most frequently 0.1–1 3 107 or 5 3 107 bubbles in mice [Andonian et al. 2009; Bachawal et al. 2013; Korpanty et al. 2007; Sorace et al. 2012; Wei et al. 2014]), to allow both methods to detect the wide range of Esel expression levels in the LPS mouse model; (ii) rapid bubble destruction followed by subtraction of the freely circulating bubble signal (allowing a shorter wait time, frequently employed by others using the conventional method) was not used because it would preclude direct comparison against the novel method in the same animal, and furthermore, the method has its own limitations (see Introduction); and (iii) differences in the experimental setup may have influenced the freely

circulating bubble half-life (e.g., imaging settings/protocol, host and bubble factors). This study indicated that for the retained bubbles, the higher its maximum concentration (Ar), the longer its half-life (ln 2/lr). The Ar vs. lr relationship was non-linear. As far as we know, this is the first in vivo demonstration of the ‘‘bubble–bubble protection’’ phenomenon previously described in vitro for bubbles attached to a surface (Klibanov et al. 1998). The shorter distance between neighbouring retained bubbles at higher concentrations may result in (i) increased acoustical interactions between adjacent bubbles causing mechanical responses such that the net diffusion of gas out of the bubble population is reduced; and/or (ii) reduced concentration gradients for gas diffusion out of the bubbles because of increased gas saturation in the micro-environment surrounding them. Our demonstration of such a phenomenon in vivo highlighted a potential weakness of the conventional acoustic quantification method; that is, by the time of signal sampling, the retained bubbles would have decreased by a different number of half-lives amongst subjects with different levels of target molecule expression. One consequence of this is the underestimation (or even missed detection) of targeted molecules in subjects with lower expression levels (lower maximum retained bubble concentrations), as shown in R20. In contrast, the novel method takes into account the retained bubble half-life, allowing a more sensitive and accurate quantification of the targeted molecule, as shown in Ar. This is important because the expression of targeted molecules may

12

Ultrasound in Medicine and Biology

Volume -, Number -, 2015

Fig. 5. Acoustic quantification of Esel expression. Wild-type (WT) mice (blue); Esel knockout (KO) mice (red); not applicable (NA). R20 is plotted as the mean with standard deviation (error bars). Pearson’s r is shown. N 5 12 WT and 8 KO mice.

vary and be relatively low in natural/human disease states. The concentration of retained microbubbles on a target tissue increases over the course of circulating bubble passage up to a point near the beginning of the elimination phase (e.g., z2 min post-bubble administration) (Yeh 2010). It is not clear how accurately one can determine the zero time for the retained bubble concentration decay. The same problem arises for the circulating microbubble concentration as well. For consistency, the zero time was arbitrarily taken as the time of bubble administration, giving the maximum limit for Ar and Af. The linear acoustic range in vivo was assumed to be 0–5 AU, approximated from a limited titration experiment in vitro applying the same ultrasound settings and similar imaging depths as for in vivo imaging (see Supplemental Fig. A1). It should be noted that this range can change for different ultrasound settings, imaging depths, tissues and bubbles. Although the mathematical model and linear acoustic range assumptions are not perfect, they appeared adequate in practice for acoustic quantification. The good curve fit, strong correlation of Ar with Esel expression and expected results in the analysis of circulating bubbles constitute

important preliminary evidence supporting the practical applicability of the mathematical model and linear range. Future studies would be desirable for further validation of the mathematical model and determination of the actual linear range in vivo in the ROI(s). Other TIC-based quantification methods, using different equations for curve fitting, have been described (Behm et al. 2008; Carr et al. 2011; Chen et al. 2012; Fisher et al. 2002; Lindner et al. 1998; Sirsi et al. 2012). However, unlike our method, their performance in molecular quantification has not been tested (Carr et al. 2011; Chen et al. 2012; Fisher et al. 2002; Lindner et al. 1998; Sirsi et al. 2012) or was only tested against independent (non-acoustic) assays semiquantitatively (Behm et al. 2008). Perhaps most importantly, all previous methods required curve-fitting the entire TIC, a process that renders them prone to errors caused by signal saturation/bubble cloud attenuation at early time points of the TIC, when bubble concentrations are high. Low bubble dosages would be required when using these methods, risking underestimation of the targeted molecules as a result of non-saturation of the targets by the bubbles. In contrast, our novel method required curve-fitting of only part of the elimination

Quantitative US imaging d J. SHUE-MIN YEH et al.

13

Fig. 6. Quantitative analysis of circulating bubbles. Wild type (WT) mice (blue); Esel knockout (KO) mice (red); excluded from correlation as outlier (cross). Pearson’s r is shown. N 5 12 WT and 8 KO mice.

phase. It is simpler and quicker and allows higher bubble dosages to be used. In due course, the method may be adapted for real-time volumetric 3-D ultrasound imaging. In this study, LPS-treated Esel KO mice with the same genetic background as the WT animals were used as controls, in preference to control bubbles bearing non-specific F(ab0 )2. This had the advantages of directly testing the targeting bubbles themselves for in vivo

specificity, and eliminated any differences in signal intensities caused by non-identical size distribution between the control and targeting bubbles. Furthermore, nonspecific F(ab0 )2 fragments may well not be inert and, hence, have their own contributions. Although testing control bubbles in the WTs may help eliminate the possibility that the increasing bubble retention in the WTs with decreasing LPSTime (increasing Esel) was related to

14

Ultrasound in Medicine and Biology

Fig. 7. Retained bubble Ar versus lr relationship. N 5 12 wildtype (WT) and 8 Esel knockout (KO) mice.

Volume -, Number -, 2015

will likely vary significantly in clinical practice. Here, calibration of bubble signal intensities (e.g., against bubble concentration) in the ROI(s) will be necessary for molecular quantification and interand intra-subject comparisons. 3. The applicability of the novel method remains to be tested in other tissues and diseases and for other molecular targets. In situations where bubble detachment, cellular internalization or transmigration into the tissue interstitium is substantial, an additional term(s) may be added in the mathematical model to account for them. CONCLUSIONS

increasing non-specific bubble interaction with adherent leukocytes, this was deemed unnecessary because such a phenomenon, where present, would have similarly been observed in the KOs (albeit to a lesser degree because of the less-adherent leukocytes), which was not the case (Fig. 5a, c). Limitations There are a number of limitations in this study: 1. A relatively large bubble dose was used, equivalent to 10-fold the maximum dose of Sonovue (a clinically approved non-targeting bubble) tested in humans without adverse effects. Like other investigators (Anderson et al. 2011; Bzyl et al. 2013), we did not observe significant adverse effects at such a dose, although its safety in humans is uncertain. It is worth mentioning that, because targeting microbubbles have not yet been approved for clinical work, no dose has been set by regulatory medicine agencies. The wide range of Esel expression required the use of a large bubble dose (for saturating Esel) for its quantification, by both methods. The effects of bubble sequestration by soluble Esel in the plasma, applicable if the latter retains the epitope recognised by the targeting ligand, might also be diminished by the use of excess bubbles. For clinical applications, a significantly lower dose is likely sufficient, because the levels of molecular expression are likely lower than the maximum examined in this study. Dose titration for optimal dosing would be desirable in future studies. 2. The retained bubble signal intensities (Ar) used to relate to the targeted molecule concentration will vary with the imaging depths, ultrasound settings, types of tissues and bubbles. We kept these variables as constant as possible in this study, but they

Quantitative determination of the level of molecular expression is feasible acoustically using a time–signal intensity curve analytical method based on microbubble elimination. Acknowledgments—We thank Joseph Boyle (Hammersmith Hospital, UK) for assistance in immunohistochemistry; David Cosgrove, the late Martin Blomley (Imperial College London) and David Dawson (Hammersmith Hospital) for scientific discussions; Sanjiv Kaul, Jonathan Lindner, Alexander Klibanov and Jiri Sklenar for J.Y.’s training in bubble targeting (University of Virginia, Charlottesville, VA, USA, 2003); Helene Houle and Pavlos Moustakidis (Siemens Medical Solutions, Riverside, CA, USA) for technical support with the Acuson Sequoia 512 (2003–2008); and Suzanne Schiller-Yeh (London, UK) for manuscript proofreading.—The work described here was supported by a Bristol–Myers Squibb Cardiovascular Prize Fellowship (J.Y.); British Heart Foundation Project Grant (P.N., M.B., D.O.H., S.N., J.S., C.A.S., E.M., J.Y.); Hammersmith Hospitals Trustees’ Research Committee Small Project Grant (P.N., M.B., D.O.H.); and professorial support from the British Heart Foundation (D.O.H.).

SUPPLEMENTARY DATA Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.ultrasmedbio.2015.04.011.

REFERENCES 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. Andonian S, Coulthard T, Smith AD, Singhal PS, Lee BR. Real-time quantitation of renal ischemia using targeted microbubbles: invivo measurement of P-selectin expression. J Endourol 2009;23: 373–378. Bachawal SV, Jensen KC, Lutz AM, Gambhir SS, Tranquart F, Tian L, Willmann JK. Earlier detection of breast cancer with ultrasound molecular imaging in a transgenic mouse model. Cancer Res 2013;73: 1689–1698. Behm CZ, Kaufmann BA, Carr C, Lankford M, Sanders JM, Rose CE, Kaul S, Lindner JR. Molecular imaging of endothelial vascular cell adhesion molecule-1 expression and inflammatory cell recruitment during vasculogenesis and ischemia-mediated arteriogenesis. Circulation 2008;117:2902–2911. Bin JP, Yang L, Chen SM, Wu JF, Tang GQ, Wang YG, Liu YL. Quantification of the severity of microvascular inflammation with

Quantitative US imaging d J. SHUE-MIN YEH et al. microbubbles targeted to endothelial P-selectin [abstract]. Eur Heart J 2008;29:21. Bzyl J, Palmowski M, Rix A, Arns S, Hyvelin JM, Pochon S, Ehling J, Schrading S, Kiessling F, Lederle W. The high angiogenic activity in very early breast cancer enables reliable imaging with VEGFR2targeted microbubbles (BR55). Eur Radiol 2013;23:468–475. Carr CL, Qi Y, Davidson B, Chadderdon S, Jayaweera AR, Belcik JT, Benner C, Xie A, Lindner JR. Dysregulated selectin expression and monocyte recruitment during ischemia-related vascular remodeling in diabetes mellitus. Arterioscler Thromb Vasc Biol 2011;31: 2526–2533. Chen CC, Sirsi SR, Homma S, Borden MA. Effect of surface architecture on in vivo ultrasound contrast persistence of targeted sizeselected microbubbles. Ultrasound Med Biol 2012;38:492–503. Coulden R. Functional cardiac CT. In: Miles K, Dawson P, Blomely M, (eds). Functional computed tomography. Oxford: ISIS Medical Media; 1997. p. 133–156. 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. Eppihimer MJ, Wolitzky B, Anderson DC, Labow MA, Granger DN. Heterogeneity of expression of E- and P-selectins in vivo. Circ Res 1996;79:560–569. Feinstein SB, Voci P, Segil LJ, Harper PV. Chapter 27: Contrast echocardiography. In: Marcus ML, Skorton DJ, Schelbert HR, Wolf GL, (eds). Cardiac imaging: A companion to Braunwald’s heart disease. London: Saunders; 1991. p. 557–574. Fisher NG, Christiansen JP, Leong-Poi H, Jayaweera AR, Lindner JR, Kaul S. Myocardial and microcirculatory kinetics of BR14, a novel third-generation intravenous ultrasound contrast agent. J Am Coll Cardiol 2002;39:530–537. Hernot S, Unnikrishnan S, Du Z, Shevchenko T, Cosyns B, Broisat A, Toczek J, Caveliers V, Muyldermans S, Lahoutte T, Klibanov AL, Devoogdt N. Nanobody-coupled microbubbles as novel molecular tracer. J Control Release 2012;158:346–353. Kaufmann BA, Lewis C, Xie A, Mirza-Mohd A, Lindner JR. Detection of recent myocardial ischaemia by molecular imaging of P-selectin with targeted contrast echocardiography. Eur Heart J 2007a;28: 2011–2017. Kaufmann BA, Sanders JM, Davis C, Xie A, Aldred P, Sarembock IJ, Lindner JR. Molecular imaging of inflammation in atherosclerosis with targeted ultrasound detection of vascular cell adhesion molecule-1. Circulation 2007b;116:276–284. Klibanov AL, Ferrara KW, Hughes MS, Wible JH Jr, Wojdyla JK, Dayton PA, Morgan KE, Brandenburger GH. Direct videomicroscopic observation of the dynamic effects of medical ultrasound on ultrasound contrast microspheres. Invest Radiol 1998;33: 863–870. Klibanov AL, Gu H, Wojdyla JK, Wible JH Jr, Kim DH, Needham D, Villanueva FS, Brandenburger GH. Attachment of ligands to gasfilled microbubbles via PEG spacer and lipid residues anchored at the interface. Proc Int Symp Control Rel Bioact Mater 1999;26:230. Korpanty G, Carbon JG, Grayburn PA, Fleming JB, Brekken RA. Monitoring response to anticancer therapy by targeting microbubbles to tumor vasculature. Clin Cancer Res 2007;13:323–330. Labow MA, Norton CR, Rumberger JM, Lombard-Gillooly KM, Shuster DJ, Hubbard J, Bertko R, Knaack PA, Terry RW, Harbison ML. Characterization of E-selectin-deficient mice: Demonstration of overlapping function of the endothelial selectins. Immunity 1994;1:709–720. Lankford M, Behm CZ, Yeh J, Klibanov AL, Robinson P, Lindner JR. Effect of microbubble ligation to cells on ultrasound signal enhancement: Implications for targeted imaging. Invest Radiol 2006;41: 721–728. Lee DJ, Lyshchik A, Huamani J, Hallahan DE, Fleischer AC. Relationship between retention of a vascular endothelial growth factor receptor 2 (VEGFR2)-targeted ultrasonographic contrast agent and the level of VEGFR2 expression in an in vivo breast cancer model. J Ultrasound Med 2008;27:855–866. Leong-Poi H, Christiansen J, Heppner P, Lewis CW, Klibanov AL, Kaul S, Lindner JR. Assessment of endogenous and therapeutic ar-

15

teriogenesis by contrast ultrasound molecular imaging of integrin expression. Circulation 2005;111:3248–3254. Leong-Poi H, Christiansen J, Klibanov AL, Kaul S, Lindner JR. Noninvasive assessment of angiogenesis by ultrasound and microbubbles targeted to alpha(v)-integrins. Circulation 2003;107:455–460. Lindner JR, Ismail S, Spotnitz WD, Skyba DM, Jayaweera AR, Kaul S. Albumin microbubble persistence during myocardial contrast echocardiography is associated with microvascular endothelial glycocalyx damage. Circulation 1998;98:2187–2194. Lindner JR, Song J, Xu F, Klibanov AL, Singbartl K, Ley K, Kaul S. Noninvasive ultrasound imaging of inflammation using microbubbles targeted to activated leukocytes. Circulation 2000;102: 2745–2750. Lindner JR, Song J, Christiansen J, Klibanov AL, Xu F, Ley K. Ultrasound assessment of inflammation and renal tissue injury with microbubbles targeted to P-selectin. Circulation 2001;104:2107–2112. Liu H, Wang X, Tan KB, Liu P, Zhuo ZX, Liu Z, Hua X, Zhuo QQ, Xia HM, Gao YH. Molecular imaging of vulnerable plaques in rabbits using contrast-enhanced ultrasound targeting to vascular endothelial growth factor receptor-2. J Clin Ultrasound 2011;39:83–90. 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. Mancini M, Greco A, Salvatore G, Liuzzi R, Di Maro G, Vergara E, Chiappetta G, Pasquinelli R, Brunetti A, Salvatore M. Imaging of thyroid tumor angiogenesis with microbubbles targeted to vascular endothelial growth factor receptor type 2 in mice. BMC Med Imaging 2013;13:31. Palmowski M, Huppert J, Ladewig G, Hauff P, Reinhardt M, Mueller MM, Woenne EC, Jenne JW, Maurer M, Kauffmann GW, Semmler W, Kiessling F. Molecular profiling of angiogenesis with targeted ultrasound imaging: Early assessment of antiangiogenic therapy effects. Mol Cancer Ther 2008;7:101–109. Palmowski M, Peschke P, Huppert J, Hauff P, Reinhardt M, Maurer M, Karger CP, Scholz M, Semmler W, Huber PE, Kiessling FM. Molecular ultrasound imaging of early vascular response in prostate tumors irradiated with carbon ions. Neoplasia 2009;11:856–863. Raher MJ, Thibault H, Poh KK, Liu R, Halpern EF, Derumeaux G, Ichinose F, Zapol WM, Bloch KD, Picard MH, Scherrer-Crosbie M. In vivo characterization of murine myocardial perfusion with myocardial contrast echocardiography: Validation and application in nitric oxide synthase 3 deficient mice. Circulation 2007;116:1250–1257. Reynolds PR, Larkman DJ, Haskard DO, Hajnal JV, Kennea NL, George AJ, Edwards AD. Detection of vascular expression of E-selectin in vivo with MR imaging. Radiology 2006;241:469–476. Riviere JE. Compartmental models. In: Riviere JE, (ed). Comparative pharmacokinetics. Ames: Iowa State Press; 1999. p. 107–147. Rychak JJ, Graba J, Cheung AM, Mystry BS, Lindner JR, Kerbel RS, Foster FS. Microultrasound molecular imaging of vascular endothelial growth factor receptor 2 in a mouse model of tumor angiogenesis. Mol Imaging 2007;6:289–296. Saini R, Sorace AG, Warram JM, Mahoney MJ, Zinn KR, Hoyt K. An animal model allowing controlled receptor expression for molecular ultrasound imaging. Ultrasound Med Biol 2013;39:172–180. Sirsi SR, Hernandez SL, Zielinski L, Blomback H, Koubaa A, Synder M, Homma S, Kandel JJ, Yamashiro DJ, Borden MA. Polyplex-microbubble hybrids for ultrasound-guided plasmid DNA delivery to solid tumors. J Control Release 2012;157:224–234. Skyba DM, Jayaweera AR, Goodman NC, Ismail S, Camarano G, Kaul S. Quantification of myocardial perfusion with myocardial contrast echocardiography during left atrial injection of contrast: Implications for venous injection. Circulation 1994;90:1513–1521. Sorace AG, Saini R, Mahoney M, Hoyt K. Molecular ultrasound imaging using a targeted contrast agent for assessing early tumor response to antiangiogenic therapy. J Ultrasound Med 2012;31:1543–1550. Stieger SM, Dayton PA, Borden MA, Caskey CF, Griffey SM, Wisner ER, Ferrara KW. Imaging of angiogenesis using Cadence contrast pulse sequencing and targeted contrast agents. Contrast Media Mol imaging 2008;3:9–18.

16

Ultrasound in Medicine and Biology

Streif JU, Nahrendorf M, Hiller KH, Waller C, Wiesmann F, Rommel E, Haase A, Bauer WR. In vivo assessment of absolute perfusion and intracapillary blood volume in the murine myocardium by spin labeling magnetic resonance imaging. Magn Reson Med 2005;53: 584–592. Tlaxca JL, Rychak JJ, Ernst PB, Konkalmatt PR, Shevchenko TI, Pizarro TT, Rivera-Nieves J, Klibanov AL, Lawrence MB. Ultrasound-based molecular imaging and specific gene delivery to mesenteric vasculature by endothelial adhesion molecule targeted microbubbles in a mouse model of Crohn’s disease. J Control Release 2013;165:216–225. Waller C, Kahler E, Hiller KH, Hu K, Nahrendorf M, Voll S, Haase A, Ertl G, Bauer WR. Myocardial perfusion and intracapillary blood volume in rats at rest and with coronary dilatation: MR imaging in vivo with use of a spin-labeling technique. Radiology 2000;215:189–197.

Volume -, Number -, 2015 Wei S, Fu N, Sun Y, Yang Z, Lei L, Huang P, Yang B. Targeted contrastenhanced ultrasound imaging of angiogenesis in an orthotopic mouse tumor model of renal carcinoma. Ultrasound Med Biol 2014;40:1250–1259. Weller GE, Lu E, Csikari MM, Klibanov AL, Fischer D, Wagner WR, Villanueva FS. Ultrasound imaging of acute cardiac transplant rejection with microbubbles targeted to intercellular adhesion molecule1. Circulation 2003;108:218–224. Xie J, Liao Y, Yang L, Wu J, Liu C, Xuan W, Li M, Zhang L, Liu Y, Wu P, Bin J. Ultrasound molecular imaging of angiogenesis induced by mutant forms of hypoxia-inducible factor-1alpha. Cardiovasc Res 2011;92:256–266. Yeh JSM. Molecular imaging of inflammation using echocardiography. PhD thesis, Cardiovascular Medicine, National Heart and Lung Institute, Imperial College London. London: University of London; 2010.

Quantitative US imaging d J. SHUE-MIN YEH et al.

Appendix SUPPLEMENTAL METHODS Immunohistochemistry Immunohistochemistry was performed on acetonefixed cryosections of freshly harvested hearts of WT (with/without LPS pre-treatment) and Esel KO (pretreated with LPS) mice. After non-specific binding sites were blocked with 100 mL of 1:1,000 rabbit serum (Sigma-Aldrich) for 1 h at room temperature, sections were incubated for 1 h at room temperature with 100 mL of 0.067 mmol/L (0.01 mg/mL) primary antibody: MES-1 (for Esel), MEC13.3 (for PECAM-1, endothelial marker) or rat IgG2a,k isotype-negative control mAb. Each section was then incubated with 100 mL of 0.034 mmol/L (0.005 mg/mL) biotinylated secondary antibody (biotinylated rabbit mAb against rat IgG2a) for 60 min at room temperature. After blocking of endogenous peroxidase with 0.3% H2O2 methanol for 20–30 min at room temperature, a horseradish peroxidase-based detection system (Vectastain ABC kit, Vector Laboratories) was used with 3,30 -diaminobenzidine solution (SIGMAFAST DAB tablet, Sigma-Aldrich) as the chromogen substrate. Sections were counterstained using Harris’ modified hematoxylin solution (Sigma-Aldrich) and 1% NaHCO3, then dehydrated through 70%–100% ethanol, dried and mounted with Histomount (VWR, UK) and examined under light microscopy. Reverse transcriptase real-time quantitative polymerase chain reaction Freshly harvested tissues were kept in RNAlater solution (Ambion, UK) to preserve RNA in situ; total RNAwas subsequently extracted using TRIzol reagent (Invitrogen, UK) according to the manufacturer’s instructions. The yield of total RNA from the mouse heart was typically z1 mg pure RNA per 1 mg tissue, kept at concentrations over z1 mg/mL in molecular grade (RNase-free) H2O (Sigma-Aldrich). RT reaction for first-strand complementary DNA (cDNA) synthesis was performed using the Qiagen Omniscript Reverse Transcription kit (Qiagen, UK) according to the manufacturer’s instructions. The RT reaction mixture consisted of 1 mg total RNA, 2 mL 103 buffer RT, 2 mL dNTP mix (5 mmol/L each of dATP, dCTP, dGTP and dTTP), 1 mL (4 units) Omniscript reverse transcriptase, 2 mL (1 mg) oligo(dT)12–18 primer (Invitrogen) and molecular-grade H2O made up to a total reaction volume of 20 mL, incubated for 1 h at 37 C. This was followed by real-time qPCR (SYBR Green detection method) for Esel and HPRT-I, according to the manufacturers’ instructions. The following primer sequences were used. Esel forward primer: 50 -CTCATTGCTCTACTTGTTGATG-30 , Esel reverse primer: 50 -GCATTTGTGTTCCTGATTG-30 ,

17

HPRT-I forward primer: 50 -ATTAGCGATGATGAACCAG-30 , HPRT-I reverse primer: 50 -AGTCTTT(custom ordered from CAGTCCTGTCCAT-30 Invitrogen). The qPCR was carried on a 96-well 0.2-mL thin-wall PCR plate (Bio-Rad, UK) covered with an Optical Quality Sealing Tape (Bio-Rad), using the iCycler (iCycler iQ Real-Time PCR Detection System, Bio-Rad) according to the manufacturer’s instructions. The qPCR volume was 25 mL, consisting of 5 mL cDNA template (1:50 water dilution of the finished RT reaction), 0.5 mL (10 mmol/L) each of the forward and reverse primers for the respective gene, 6.5 mL molecular-grade H2O and 12.5 mL iQ SYBR Green Supermix (Bio-Rad). The qPCR cycling condition was as follows: initial 3-min denaturing step at 95 C (Well Factor analysis in the first 90 s); then 40 cycles of 15 s at 95 C, 1 min at 56 C; melt-curve analysis in 0.5 C steps (1 min denaturation at 95 C, 1 min reset at 56 C, then 80 cycles of 10 s at 60 C with 0.5 C increment for each cycle); final cooling step at 4 C. Triplicate Esel and HPRT-I were amplified on the same plate for each animal; no-template negative controls using molecular-grade H2O in place of cDNA template for both primer pairs were included in all plates. For data analysis, the threshold cycle (Ct ) was determined from the amplification plot using the iCycler iQ Optical System Software Version 3.0a (Bio-Rad). Wells with an abnormal amplification plot or melt curve were excluded. As PCR efficiency of the Esel and HPRT-I primer pairs differed by #5% (mean [standard deviation] 5 93 [4%] and 92 [3%], respectively; n 5 4 each), the comparative Ct method was used to estimate the amount of Esel mRNA relative to that of HPRT-I, using the formula: Esel mRNA (%HPRTI) 5 2DCt , where DCt 5 CtEsel -- CtHPRT-I ; subscripts refer to the genes of interest. The mean of replicates was used for each animal. Reduction of MES-1 F(ab0 )2 for microbubble conjugation MES-1 F(ab0 )2 was reduced using a 4 molar excess of tris(2-carboxyethyl)phosphine hydrochloride (SigmaAldrich): MES-1 F(ab0 )2 (83.3 mmol/L, 8.3 mg/mL) and tris(2-carboxyethyl)phosphine hydrochloride (333.3 mmol/L, 0.096 mg/mL) in Exchange Buffer (50 mmol/L 2-(N-morpholino)ethanesulfonic acid [Sigma-Aldrich], 2 mmol/L ethylenediaminetetraacetic acid [SigmaAldrich], pH 6) were incubated for 1 h at 37 C under constant agitation. The reaction volume ranged from 1 to 1.6 mL. The reaction was stopped by placing on ice and immediate purification of the reduced F(ab0 )2 using spin column gel filtration chromatography with a 5-mL Zeba Desalt Spin Column (size exclusion limit: 1000 Da) according to the manufacturer’s instructions (Perbio Science, UK), at 4 C. The spin column was previously equilibrated in cold Exchange Buffer. The degree of

18

Ultrasound in Medicine and Biology

reduction of the F(ab0 )2 (number of thiols per F(ab0 )2) was determined spectrophotometrically using Ellman’s test with Ellman’s reagent (Perbio Science) according to the manufacturer’s instructions with the following modifications: the Ellman reaction consisted of 2.5 mL Ellman’s reagent (10 mmol/L, 4 mg/mL), 7.5 mL Exchange Buffer and 90 mL of the purified reduced F(ab0 )2 in Exchange Buffer, incubated at room temperature covered with aluminium foil to minimise light exposure; absorbance at 412 nm (A412) was measured 24 min from the start of Ellman’s reaction, to determine the thiol concentration in the reduced F(ab0 )2 sample by reference to a standard curve of Ellman’s reaction with known concentrations of thiol-containing compound, L-cysteine hydrochloride (Perbio Science) in Exchange Buffer (pH 6); a duplicate Ellman ‘‘blank’’ reaction, where the Exchange Buffer was added in place of the reduced F(ab0 )2, was used for baseline subtraction of A412 from the test samples. The concentration of reduced F(ab0 )2 was determined from A280 in the absence of Ellman’s reagent (with the spectrophotometer zeroed using Exchange Buffer), as the Ellman’s reagent would interfere with A280. The degree of F(ab0 )2 reduction was calculated as thiol groups per F(ab0 )2 5 [thiol in mmol/L]/[F(ab0 )2 in mmol/L]. The reduced F(ab0 )2 contained two thiol groups per molecule of F(ab0 )2. The purified reduced F(ab0 )2 was kept at concentration of z80 mmol/L (8 mg/mL) in Exchange Buffer prior to conjugation with microbubbles. Microbubble preparation Native (non-conjugated) microbubbles were prepared by dispersing DSPC, DSPE-PEG2000-Mal, PEG40-stearate and DiI at a molar ratio of 75:9:14:2 in a small amount of cyclohexane:chloroform (1:2) solvent, in a 50-mL round-bottomed flask. Excess solvent was extracted using a stream of gaseous nitrogen. The lipid blend was then transferred to a freeze dryer and lyophilised to full dryness under a reduced atmosphere (1.3 3 104 Pa) at 278.5 C (using a jacket of dry ice). The dry powder (lyophilisate) was then dispersed in normal saline containing propylene glycol (PGNS: propylene glycol 1.37 mol/L (103.5 mg/mL), glycerol 1.37 mol/L (126.2 mg/mL), NaCl 0.116 mol/L (6.8 mg/mL), pH z 7.4) to a concentration of 4 mg/mL, homogenised by sonication in an ultrasonic bath at 60 C–65 C until transparent. Once fully dissolved, the solution was gently sparged with C3F8 gas (F2 Chemicals, UK). Microbubbles were then formed using a shear-mixing approach, by sonic dispersion of C3F8 using a Misonix 3000 sonicator (QSonica, Newtown CT, USA). The probe tip was positioned about 2 mm into the solution, and sonication was performed with the high-intensity ultrasound horn (20– 21 kHz) for 30–60 s at an acoustic power of approximately 120 W, with the initial temperature of the solution

Volume -, Number -, 2015

at z60 C. More C3F8 gas was sparged into the microbubble dispersion, and the vessel was capped and immediately plunged into ice-cold water (3 min) to dissipate the heat generated during the sonication process. Microbubbles produced were washed (purified) by centrifugation flotation at 1,000g and 4 C for 15–25 min, using a Beckman Coulter Allegra X-15R Centrifuge (Beckman Coulter, UK). Bubbles floated to the top of the sample vial after centrifugation, and the subnatant was removed and replaced with an equal volume of cold degassed normal saline (pH 7.4). The wash step was repeated seven times to remove unincorporated shell components and bubble fragments. To produce Esel-targeting microbubbles, these washed native microbubbles were added to reduced MES-1 F(ab0 )2 whilst mixing. Each reduced F(ab0 )2 molecule contained two thiol groups, prepared as described above. The conjugation reaction ratio was 4.338 3 106 F(ab0 )2 molecules per bubble (7.2 nmol F(ab0 )2 per 109 bubbles). The total number of DSPEPEG2000-Mal molecules in the aqueous lipid blend divided by the total number of bubbles (before wash) produced from the aqueous lipid blend 5 4.338 3 106. If #10% of the components in the aqueous lipid blend were incorporated into the bubble shell, and the molar ratio of the components on the shell remained close to that in the lipid blend (Klibanov et al. 1999), then a bubble of population mean size would contain #4.338 3 105 maleimide molecules, and the estimated F(ab0 )2:maleimide conjugation reaction molar ratio would be $10:1. The concentrations of bubbles and F(ab0 )2 in the conjugation reaction mixture ranged from 5 3 109 to 8 3 109/mL and from 35 to 60 mmol/L (3.5–6 mg/mL), respectively. The reaction mixture contained approximately twothirds volume of Exchange Buffer (pH 6) from the reduced F(ab0 )2 and one-third volume of normal saline (pH 7.4) from the washed bubbles. The conjugation reaction was incubated at 4 C for 30 min, continuously mixed gently on a vertically tilted rotating wheel. Bubble conjugation was terminated by adding 80 mmol/L N-ethylmaleimide (NEM, Sigma-Aldrich) dissolved in dry dimethyl sulfoxide (DMSO, Sigma-Aldrich) at 20 molar excess to F(ab0 )2; the reaction mixture was incubated at 4 C for 30– 60 min on the rotator. Typically, the concentrations of NEM and DMSO in the reaction mixture were z1 mmol/L and #1.7% (v/v), respectively. The bubbles were then washed four times with cold normal saline by centrifugation flotation as described above, at 160g and 4 C for 5 min. This removed unincorporated F(ab0 )2, unreacted NEM, DMSO and bubble fragments. To minimise bubble loss, all washes and incubations were performed with the bubble concentrations kept high ($1 3 109 bubbles/mL), under a C3F8 atmosphere (to reduce the concentration gradient for

Quantitative US imaging d J. SHUE-MIN YEH et al.

diffusion of C3F8 gas out of the bubbles) and at 4 C (to reduce the rate of gas diffusion). To preserve the bubbles, DiI fluorescent dye, the lyophilisate or bubbles were protected from light. Freshly prepared washed Esel-targeting bubbles were immediately divided into 20- to 50-mL aliquots, capped and sealed with Parafilm, then snap-frozen in liquid nitrogen and stored at 280 C until use. The integrity of the bubbles was checked under microscopy. Electrozone sensing using a Coulter Multisizer IIe equipped with a 30-mm-diameter orifice counting tube (Coulter Electronics, UK) was used for bubble counting and sizing, according to the manufacturer’s instructions. The concentration of subsequently thawed Eseltargeting bubbles ranged from 1 3 109 to 3 3 109 bubbles/mL amongst batches prepared at different times (the range was due mainly to batch variation of the bubble concentrations in the conjugation and wash steps, rather than the freeze–thaw process; the latter caused only 14% change). The targeting bubble size distribution was reproducible amongst the batches: the mean (SEM) bubble diameter was 2.2 (0.2) mm, and 98.6% or 100% of the bubbles were less than 6 or 10 mm in diameter, respectively (Yeh 2010). The freeze–thaw strategy allowed long-term storage of the bubbles and minimised variations in the bubble size distribution (affecting, e.g., ultrasound signal intensities) used in the experiments. Thawed left-over bubbles were not re-used. Ultrasound molecular imaging Fifteen WT and eight Esel KO mice all pre-treated with LPS were imaged. The tail vein was cannulated with a 24G 0.7 3 19-mm intravenous catheter (dead space z50 mL) (BD Medical, UK). General anaesthesia was achieved using intraperitoneal injection of 200–300 mL of a mixture containing 4.5 mmol/L (1 mg/mL) xylazine (Rompun, Bayer, UK) and 36.5 mmol/L (10 mg/mL) ketamine hydrochloride (Ketalar, Parke-Davis, UK) in normal saline. The chest was shaved. ECG electrode pads were applied to the paws and connected to the Acuson Sequoia 512 clinical ultrasound system, equipped with a 15L8-s linear array transducer (footprint 5 26 mm) and ‘‘Small Animal ECG Filter.’’ A layer of warm gel was coupled between the skin and ultrasound transducer. Imaging was maintained in the PSA view of the heart by fixing the transducer in position with a freestanding clamp. The following ultrasound settings were used: 14-MHz CPS mode, transmission power 5 9 dB yielding low MI of 0.22–0.26 (estimated by the scanner), dynamic range 5 55 dB, time gain 5 0% (depth gain

19

control sliders at extreme left position), CPS gain 5 8, fundamental 2-D gain 5 15 dB, colour map M:3 (bubble signal presented in heated object scale). TEQ tissue equalization technology was not used. Baseline images (before bubble injection) were first acquired. Then a stopwatch was started and 108 Esel-targeting bubbles (in 100-mL volume made up with normal saline) injected at 10 s via the tail vein catheter as an intravenous bolus over 1–2 s, followed by a 100-mL normal saline flush over 1–2 s at 20 s. Continuous ultrasound imaging was performed and recorded as 3-s digital clips to capture the whole lifetime of the bubble bolus, as follows. Continuous imaging was applied from time 0 to 1 min 23 s on the stopwatch, then for 3 s each time at (i) 1-min intervals from 2 min 20 s to 10 min 20 s; (ii) 2-min intervals from 12 min 20 s to 30 min 20 s; and (iii) 5-min intervals from 35 min 20 s to 60 min 20 s. Three-second digital clips for these were recorded at 10 s, 13 s and then at 10-s intervals from 20 s to 1 min 20 s; then at (i) 1-min intervals from 2 min 20 s to 10 min 20 s; (ii) 2-min intervals from 12 min 20 s to 30 min 20 s; and (iii) 5-min intervals from 35 min 20 s to 60 min 20 s. Imaging was terminated earlier if bubble contrast enhancement in the LV cavity (central blood pool) was no longer visible. All animals received only one dose of bubbles to eliminate any carryover effects from previous bubble dosing (e.g., blocking of Esel binding sites). The duration between LPS treatment and the administration of targeting bubbles was noted as the LPSTime. The HR for each animal was averaged from all HRs recorded at different time points during cardiac imaging. All animals were sacrificed at the end. Acoustic quantifications: Curve fitting the TICs with the exponential functions (mathematical model) of targeting bubble elimination A constant (S) may be introduced into the exponential functions to account for system noise not removed by baseline signal subtraction, that is, curve-fitting Itissue ðtÞ 5 A0f e2lf t 1A0r e2lr t 1Stissue (instead of Itissue ðtÞ 5 Af e2lf t 1Ar e2lr t ) and Ic ðtÞ 5 A0c e2lc t 1Sc (instead of Ic ðtÞ 5 Ac e2lc t ) to the eliminationphase TICs of the myocardium and LV cavity, then be respectively. Af, Ar and Ac  may 0 0  0 0 5 A 2 A = A 1A , obtained  using A S f tissue f f f r 0 0  0 0 Ar 5 Ar 2 Ar = Af 1Ar Stissue and Ac 5 A0c 2Sc ; respectively. In this study, both Stissue and Sc were negligible: mean (SD, range) 5 0.01 (0.01, 3 3 10218–0.02) AU and 0.01 (0.01, 2 3 10218–0.02) AU, respectively; n 5 20 each. In other words, Af zA0f ; Ar zA0r ; and Ac zA0c :