The authors have declared that no competing interests exist.
Conceived and designed the experiments: MKA MAL RPM. Performed the experiments: MKA LL JM. Analyzed the data: MKA MAL RPM. Contributed reagents/materials/analysis tools: MAL LL JM. Wrote the paper: MKA RPM.
Current address: Jordan University of Science and Technology, Irbid, Jordan
Small animal imaging provides diverse methods for evaluating tumor growth and acute response to therapy. This study compared the utility of non-invasive optical and ultrasound imaging to monitor growth of three diverse human tumor xenografts (brain U87-luc-mCherry, mammary MCF7-luc-mCherry, and prostate PC3-luc) growing in nude mice. Bioluminescence imaging (BLI), fluorescence imaging (FLI), and Power Doppler ultrasound (PD US) were then applied to examine acute vascular disruption following administration of arsenic trioxide (ATO).
During initial tumor growth, strong correlations were found between manual caliper measured tumor volume and FLI intensity, BLI intensity following luciferin injection, and traditional B-mode US. Administration of ATO to established U87 tumors caused significant vascular shutdown within 2 hrs at all doses in the range 5 to 10 mg/kg in a dose dependant manner, as revealed by depressed bioluminescent light emission. At lower doses substantial recovery was seen within 4 hrs. At 8 mg/kg there was >85% reduction in tumor vascular perfusion, which remained depressed after 6 hrs, but showed some recovery after 24 hrs. Similar response was observed in MCF7 and PC3 tumors. Dynamic BLI and PD US each showed similar duration and percent reductions in tumor blood flow, but FLI showed no significant changes during the first 24 hrs.
The results provide further evidence for comparable utility of optical and ultrasound imaging for monitoring tumor growth, More specifically, they confirm the utility of BLI and ultrasound imaging as facile assays of the vascular disruption in solid tumors based on ATO as a model agent.
Vascular disruption has been proposed as a therapy for solid tumors based on the principle of starving cancer of a supply of nutrients
Optical imaging is particularly easy to implement and both BLI and FLI have become routine in cancer research notably for examining tumor growth, spread and response to therapy
While optical imaging is valuable in the pre-clinical setting, application to patients is less feasible and in this case ultrasound imaging offers relatively cost-effective measurement of both anatomy and vascular perfusion. New pre-clinical instruments offer high frequency transducers and 40 MHz Power Doppler ultrasound allows non-invasive detection of blood flow changes following a treatment. The use of PD to detect such changes
In this study, arsenic trioxide (ATO) was used as a model vascular disrupting agent (VDA) on three different human tumor types including brain, breast, and prostate cell lines and assessed using BLI, FLI, and PD US. We investigated correlations between BLI and FLI signal intensities in dual reporter gene transfected cells and explored the correlation between BLI and Power Doppler ultrasound to reveal the acute effects of ATO on tumor vasculature.
U87 MG (human brain tumor, ATCC, Rockville, MD) and MCF7 (human breast tumor, ATCC) cell lines were transfected sequentially to stably express luciferase and mCherry genes and cultured using DMEM medium. PC3 (human prostate tumor, ATCC) cells were transfected with luciferase gene alone and cultured using HAM'S F12 medium according to ATCC cell culture protocol. Both media were supplemented with 10% FBS, 1%
A stock solution of 1% ATO was prepared by dissolving As2O3 (99.995% trace metals basis- 202673-5G; Sigma-Aldrich, St. Louis, MO) in distilled water with continuous stirring for five days and gentle heat for 2 hrs on the last day and then stored at 4°C. A second stock solution was prepared with 5% dextrose in 0.9% NaCl and used to dilute the ATO solution to the desired concentration. Dextrose was added to reduce the toxicity of ATO, as suggested
All procedures were approved by the UT Southwestern Institutional Animal Care and Use Committee. Twenty one athymic nude mice (12 male and 9 female, 5–6 weeks old) were obtained from National Cancer Institute (NCI, Frederick, MD) and housed in a specific pathogen free facility. Tumors were generated by implanting cells of the three tumor cell lines U87-luc-mCherry, MCF7-luc-mCherry, and PC3-luc subcutaneously. U87 cells (1×106 cells/mouse) were implanted on the back of the mice, while MCF7 and PC3 (2×106 cells/mouse for each line) were implanted in the right flank. Caliper measurements were applied using the ellipsoid volume equation (height×width×length×π/6) to estimate tumor volume. Six MCF7-Luc-mCherry mice were assigned for histology. Various doses of ATO in the range 5 to 10 mg/kg or saline were administrated IP.
BLI and FLI were performed using IVIS® Spectrum or Lumina II instruments (Caliper Life Sciences, Hopkinton, MA). On each occasion FLI was performed before BLI to avoid possible interference. Optimal detection of mCherry was achieved with λex = 570 nm and λem = 620 nm with f-stop 1, pixel binning 8 and 0.5 s exposure time.
For BLI, sodium
Ultrasound imaging was performed using a VisualSonics Vevo 770 High-Resolution Imaging System (Visual Sonics Inc, Toronto, Ontario, Canada) with a 40 MHz probe in B mode to identify the tumor region. Power Doppler mode was additionally performed at various time points (pre, 2 hrs, 4 hrs, 6 hrs, 24 hrs after drug or saline) to quantify the blood flow in each single slice of the tumor to provide a volumetric quantification.
For BLI and FLI regions of interest (ROI) were chosen for each tumor. Total photon flux (φ = photon count/(image acquisition time x area)) was calculated for each mouse using the optical instrumentation software (Living Image 4.2). Signal intensities obtained for control and treated mice with different doses were compared before and after treatment. For PD US an ROI was selected in each single slice using the Vevo 770 V3.0.0 software and 3D reconstruction was performed to measure the tumor size. The percent of vascularity (PV) for power Doppler mode was calculated for the entire tumor volume using the ultrasound manufacturer's software.
Student's t-tests were used for comparisons between signal intensities for BLI and FLI, and PV values for PD ultrasound before, and after treatment. P<0.05 was considered significant. The square of the correlation coefficient (
CD31 staining: after imaging at each time point, the blue fluorescent dye Hoechst 33342 (10 mg/kg, Molecular Probes, Eugene, OR, USA) was injected into the tail vein of selected anesthetized mice, and the tumors were excised 1 min later. Tumor tissue was frozen in OCT (Sakura Finetek, Torrance, CA, USA) and then stored at −80°C. Cryosections (8 µm) were cut and fixed with 4% paraformaldehyde for 15 mins at room temperature. The tissue was then washed three times in PBS for 5 minutes. After blocking with normal goat serum for 3 hours, the slides were incubated with primary rat anti-mouse CD31 antibody (1∶1000, BD Pharmingen, USA) overnight at 4°C. Slides were rinsed three times at 5-minute intervals with PBS and incubated with Alexa Fluor 488 goat anti-rat antibody (1∶1000, Molecular Probes, Eugene, OR) for two hours in the dark. The slides were washed 5 times using PBS for 5 minutes each cycle. The slides were prepared with fluorescent mounting medium (Dako North America, Carpentaria, CA, USA), and imaged using an LSM 510 Meta confocal microscope (Carl Zeiss Microscopy, Germany). Overlay analysis of the CD31 antibody with Hoechst 33342 was performed using ImageJ (NIH).
Caspase-3 activity: additional sections were blocked with normal donkey serum for 3 hours, and slides incubated with cleaved caspase-3 (Asp175) primary antibody (1∶4000, Cell Signaling Technology, Beverly, MA, USA) overnight at 4°C. Slides were rinsed three times at 5 minutes each with PBS and incubated with Alexa Fluor 488 donkey anti-rabbit antibody (1∶1000, Molecular Probes, Eugene, OR, USA) for two hours in the dark. Further preparation was performed as for CD31 above. Photomicrography was achieved using a Leica DM2000 photomicroscope equipped with brightfield epi-fluorescence, incident angle darkfield illumination and an Optronics Microfire digital CCD color camera interfaced with Macintosh G4 computer. Images were captured using PictureFrame 2.0 acquisition and software (Optronics, Inc. Goleta, CA, USA).
H&E: Whole mount serial sections were cut at 8 µm and microscopy performed with the Leica DM2000. Imaging used a Microtek ArtixScann 400tf film scanner with PathScan Enabler slide holder (Meyer Instruments, Houston, TX, USA).
Six mice implanted with U87-luc-mCherry cells were imaged using FLI and BLI over a period of 30 days (
A) FLI (λEx = 570 nm and λEm = 620 nm) showed tumor growth on the back of a nude mouse. B) Corresponding BLI acquired 10 mins after administration of sodium
Following the control study with saline the same mice were used to evaluate response to ATO. Mice bearing U87-luc-mCherry tumors were injected IP with doses of ATO ranging from 5 to 10 mg/kg and sequential FLI and BLI acquired up to 24 hrs (
Optical imaging was performed at various times before and after administration of ATO. On each occasion FLI was performed first and then
MCF7-luc-mCherry showed similar behavior to U87 in terms of FLI and BLI signal versus tumor growth (
With respect to saline injection highly reproducible BLI, FLI and PD US images were observed over a period of 24 hrs following injection of saline IP (
A group of nine MCF7-mCherry-luc tumors was repeatedly observed by FLI, BLI, and PD US following injection of 100 µl saline IP. A) Sequential FLI for a representative mouse, B) BLI from the same mouse showing images acquired 10 mins after administration of fresh luciferin on each occasion, c) PD US showing MIP (maximum intensity projection) observed at 40 MHz, D) Variation in dynamic bioluminescent signal intensity from the same tumor as in A,B and C, on five sequential occasions over 24 hrs, E) Comparison of signal stability based on FLI (red
Repeat imaging every 2 hrs with any of the modalities was highly successful in the control state. Past experience has shown that repeated long term anesthesia with respect to vascular disrupting agents could cause animal death and therefore separate mice were imaged 2, 4, and 6 hrs after administration of ATO with each tumor serving as its own control. At each time point following administration of ATO, light emission was significantly reduced, as shown for a representative mouse in
A) Variation in bioluminescent signal intensity from a representative tumor on sequential occasions over 24 hrs: blue ♦ baseline; red ▪ 4 hrs; green ▴ 24 hrs. B) BLI acquired 10 mins after administration of luciferin. C) PD US images are presented as single slice (left) and PD maximum intensity projection (right) before and 4 hrs after treatment with ATO. D) Comparison of PD US and BLI in MCF7-mCherry-luc tumors as fractional signal versus baseline.
Relative BLI signal intensity observed for tumors 10 mins after administration of luciferin with respect to drug treatment in three different types: blue ♦ MCF7-mCherry-luc; green ▴U87-mCherry-luc; red ▪ PC3-luc; mean values ± SE.
Histology confirmed that perfusion was compromised 2 to 6 hrs post ATO. ATO is reported to cause both vascular effects and induce apoptosis, as confirmed by IHC in a group of five MCF7-luc-mCherry tumors, which were excised at various times after ATO (
Sections were obtained from a series of mice sacrificed at various times after ATO (8 mg/kg). Hoechst stain shows reduced perfusion 2–6 hrs following ATO and H&E stain shows increased necrosis after 24 hrs. Left column: vascular extent (CD31; green) and perfusion (Hoechst 33342; blue). Middle column: Caspase-3 activity indicating apoptosis. Right column: H&E.
Acute vascular response was observed in all three tumor types within two hours of administration of ATO IP. Non-invasive BLI and PD US each revealed vascular shutdown with considerable recovery 24 hr later. The effect was dose dependant in terms of fractional reduction in vascular perfusion and recovery.
All three imaging techniques (BLI, FLI, and US) accurately reported tumor growth by comparison with caliper measurements. The optical approaches have the advantage of revealing even sub-palpable volumes, though they do require that cells be transfected to express reporter genes. Tight correlation was observed between each of the techniques for measurements over a period of 30 days and volumes up to 200 mm3 (
In common with previous reports, ATO caused dose dependant vascular shutdown
Anti vascular effects of ATO were extensively reported based on uptake of intravenously injected 86RbCl or clearance of 99mTcO4− following direct intra tumor injection
Non-invasive assessment of vascular integrity is inherently more attractive and vascular disruption has been reported for several tumor types growing in diverse locations in various rodents. Using Laser Doppler flowmetry Hines-Peralta
Dynamic BLI is particularly facile to implement and allows high throughput analysis. Three to five mice may be observed simultaneously although each imaging session does require fresh administration of luciferin. Several reports indicate that IV infusion provides higher signal, though it is quite transient and requires the technical skill of IV tail vein injection on multiple, successive occasions
Dynamic BLI was previously used to characterize vascular disruption of MDA-MB-231-luc tumors following administration of CA4P and observations were validated by reference to dynamic contrast enhanced MRI and histology
Vascular shutdown was confirmed by histology based on distribution of the perfusion marker Hoechst 33342. Two hours after ATO administration (8 mg/kg) histology showed reduced perfusion (
Arsenic trioxide was used as a model agent here based on the reports that is causes acute vascular shutdown in solid tumors
In this study the effect of ATO was demonstrated non-invasively using optical and ultrasound imaging. The strong correlation between BLI as a pre-clinical tool and PD ultrasound as a potential clinical tool suggests the potential for both assessment of pre-clinical development of VDAs and a specific biomarker to demonstrate efficacy in patients.
Variation and reproducibility of optical signals with respect to growth in U87-mCherry-Luc tumors. Upper graphs: Variation of (A) FLI and (B) BLI photon signal intensities for a group of six tumors over a period of 30 days following implantation (mean ± SE). Lower graphs: Variation of (C) FLI and (D) BLI photon signal intensities for a group of six U87-mCherry-Luc tumors following administration of saline IP. Fresh luciferin was administered at each time point (mean ± SE).
(TIF)
FLI and BLI during growth of MCF7-Luc-mCherry and PC3-luc tumors. Dependence of signal intensity on tumor volume measured using calipers for individual MCF7-mCherry-luc tumors by (A) FLI (R2>0.9) and (B) BLI (R2>0.9); C) Correlation of FLI and BLI for MCF7-mCherry-luc tumors (R2>0.88). (D) Dependence of signal intensity on tumor volume measured using calipers for PC3-luc tumors (R2>0.9).
(TIF)
Correlation between B-mode US images and caliper-measured tumor volume: repeat images for a single MCF7-Luc-mCherry tumor and wire mesh analyses below. Graph shows strong correlation between US and caliper-measured tumor volumes (R2>0.8).
(TIF)
Vascular disruption assessed by BLI in PC3-luc tumor. Left) Variation in bioluminescent signal intensity from tumor on sequential occasions before and 2 hrs after administration of ATO (8 mg/kg IP). Right) Representative images acquired 10 mins after administration of fresh luciferin on each occasion.
(TIF)
Correlations between techniques.
(DOC)
We thank Drs. Dawen Zhao and Kate Luby-Phelps for valuable discussions and Jason Reneau and John Shelton for technical assistance. Routine histology/microcopy was kindly performed by the JAR Molecular Pathology Core at UT Southwestern Medical Center. The U87 MG cells were kindly provided to us by Dr. D. Saha.