Molecular Imaging of Changes in the Prevalence of Vascular Endothelial Growth Factor Receptor in Sunitinib-Treated Murine Mammary Tumors

Cell Lines and Tumor Models

MDA-231/luc tumor cells (SibTech), a luciferase-expressing derivative of MDA-MB-231 human breast carcinoma; PAE/KDR cells expressing approximately 2 × 10⁵ VEGFR-2/cells (SibTech); and parental PAE cells (a generous gift of late Dr. Bruce Terman) were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (HyClone) and 2 mM l-glutamine (Invitrogen). MDA-231/luc and PAE/KDR were authenticated at the American Type Culture Collection Molecular Authentication Resource Center on July 20, 2009. The protocols for all animal studies were approved by the Stanford University Institutional Animal Care and Use Committee. The orthotopic MDA-231/luc model of xenogeneic breast cancer was established as described (9,10,12). Briefly, we injected 1.5–2 × 10⁶ MDA-231/luc tumor cells into the left axillary fat pad of 5- to 8-wk-old adult male severe combined immunodeficiency disease/Ncr (BALB/c background) mice (Taconic). Animals with tumors ranging between 200 and 500 mm³ (17–24 d after cell inoculation) were randomized into study groups. Treated animals received sunitinib (LC Laboratories) prepared as previously described (11), via gavage feeding, 80 mg/kg per day.

Tracers for Fluorescent Microscopy

scVEGF/Cy is produced in SibTech. The plasmid DNA–encoding scVEGF(2GA) mutant, scVEGF with the G107A and G218A amino acid substitutions, was generated by GeneWiz. The resulting mutant scVEGF(2GA) was produced at SibTech. For site-specific modification with the fluorescent dye AlexaFluor594-C₅-maleimide (Invitrogen), scVEGF or scVEGF(2GA) was deprotected as described elsewhere (9,10); incubated with 2 molar equivalents of AlexaFluor 594-C₅-maleimide for 1 h at room temperature in 0.1 M Tris-HCl and 1 mM ethylenediaminetetraacetic acid, pH 8.0; and purified by reversed-phase high-performance liquid chromatography on a Vydac Protein C4 300 Å 5-μm column (Grace Davison Discovery Sciences) using a gradient of acetonitrile in aqueous 0.1% trifluoroacetic acid. The purified conjugates were lyophilized and reconstituted in 50 mM Tris-HCl, pH 8. Functional activity of the resulting fluorescent tracers was tested as previously described (9,10).

In Vitro Cellular Uptake of scVEGF-Based Tracers

For quantitative uptake experiments, PAE/KDR and PAE cells were plated in black clear-bottom 96-well plates (BD Falcon), 20,000 cells per well. Twenty hours later, cells were shifted to fresh culture medium, with or without 1 μM sunitinib, and incubated for 1 h in a CO₂ incubator. For dose-dependence, varying scVEGF/Cy was added to cells in triplicate wells and incubated for 20 min under normal tissue culture conditions. For kinetics, 5 nM scVEGF/Cy was added to cells for varying times. Then the medium was removed, and the cells were washed extensively with phosphate-buffered saline and then with phosphate-buffered saline containing 0.5 M NaCl, fixed in fresh 4% ultrapure methanol-free formaldehyde (Polysciences), air-dried, and imaged in a Starion FLA-9000 device (Fujifilm Medical Systems USA). For fluorescent microscopy, cells were plated in 8-well chamber slides, preincubated for 1 h 20 h later with or without 1 μM sunitinib, and exposed to 5 nM scVEGF/Al594. Cells were washed and fixed in the same manner as for the kinetics, mounted in mounting medium for fluorescence with 4′,6-diamidino-2-phenylindole for nuclear counterstaining (Vector Labs), and observed in an AxioObserver microscope (Zeiss) using ×40 oil objective. For neutralization assays, a 2- to 5-fold molar excess of bevacizumab (Avastin; Genentech) or control human IgG (Sigma) was added to the fluorescent tracers scVEGF/Al594 or scVEGF(2GA)/Al594 in complete culture medium, incubated for 1 h at room temperature, and then added to PAE/KDR cells to a final scVEGF concentration of 5 nM. After a 20-min incubation, cells were washed, fixed, and mounted as described for kinetics.

Preparation of Radiolabeled ^99mTc-scVEGF

scVEGF (SibTech) was deprotected and radiolabeled, using ^99mTc-tricine as a precursor complex as described previously (10), yielding ^99mTc-scVEGF tracers that are stable in circulation. Briefly, lyophilized tin-tricine reagent was reconstituted with 1.0 mL of degassed saline to give a final SnCl₂·2H₂O concentration of 0.6 mg/mL and a final tricine concentration of 20 mg/mL (pH 7.1). Tin-tricine (10 μL) was added to 30–50 μg of premixed deprotected scVEGF and 370–555 MBq (10−15 mCi) of ^99mTc-pertechnetate in a final volume of 50–100 μL. After 30 min of incubation at 37°C, radiolabeled protein, named ^99mTc-scVEGF, was purified by gel filtration on PD-10 Sephadex-25 columns (GE Healthcare). Specific activities ranged from 4.8 to 7.4 MBq/μg (130–200 μCi/μg) of protein, with a radiopurity of approximately 95%, as determined by instant thin-layer chromatography using phosphate-buffered saline as a solvent.

SPECT and Autoradiography

For SPECT, 74–148 MBq (2–4 mCi) of ^99mTc-scVEGF per mouse was injected via the tail vein. SPECT images were obtained at 1 h after injection using a small-animal SPECT γ-camera (A-SPECT, LumaGEM; Gamma Medica) with the following parameters: 360° rotation, 64 steps, 30 s per step, 0.5-mm pinhole aperture, 64 × 64 image matrix, and a 2.7-cm field of view. For autoradiography, tumors and pectoral muscle (from the contralateral chest wall) were snap-frozen immediately after imaging, cryosectioned (60-μm thickness), and exposed to a phosphor storage screen for 16 h. The phosphor screen images were read out with a laser digitizer at a pixel dimension of 50 μm. Region-of-interest analysis of radiotracer activity was performed using ImageQuant TL software (GE Healthcare) with 4 rectangular areas per tumor rim, center, and contralateral pectoral muscle. The maximal counts per pixel of the rim and the average counts per pixel of the center of the tumor were normalized to corresponding contralateral pectoral muscle uptake (minimum background counts per pixel). The uptake of the rim (maximum) and center of each tumor were then divided by the contralateral pectoral muscle uptake and presented as rim-to-background and center-to-background ratios, respectively.

Immunohistochemical Staining

Tumor cryosections (thickness, 20 μm) were stained for VEGFR-2 (Flk-1, rat monoclonal antibody; BD Pharmingen) and CD31 (PECAM, rat monoclonal antibody; BD Pharmingen). Apoptosis was assessed by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) (Apo-BrdU-IHC In Situ DNA Fragmentation Assay Kit; BioVision) according to the manufacturer's instructions. Immune complexes were visualized by chromogenic assay using Vectastatin Elite ABC Kit (Vector Labs) and VIP Substrate Purple Kit (Vector Labs). For quantitative analysis, color images were acquired with ×4 objective, exported into Adobe Photoshop, and converted into gray-scale images, and a threshold to the same dynamic range was applied. The percentage of black pixels corresponding to immunostained area was determined using histogram analysis either for the whole images or for multiple small rectangular areas that were selected to cover the whole image.

Statistical Analysis

Data that were normally distributed were compared by standard t tests for differences among means. Immunohistochemical data that were not normally distributed were analyzed with the nonparametric Kruskal–Wallis rank test, and the significance levels of differences between group medians were determined by applying a Z test with Bonferroni adjustment for multiple comparisons.

Sunitinib Does Not Inhibit Uptake of scVEGF-Based Tracer

Sunitinib inhibits VEGFR-2 tyrosine phosphorylation with a median effective concentration of 0.01 μM (11). Because conflicting reports on the role of tyrosine phosphorylation in VEGFR-2 internalization exist (13–15), we tested whether sunitinib could inhibit the VEGFR-2–mediated uptake of scVEGF-based fluorescent tracers by porcine aortic endothelial cells PAE/KDR expressing high levels of VEGFR-2 (9). Although in these cells complete inhibition of scVEGF-induced VEGFR-2 tyrosine phosphorylation was reached in the presence of 100 nM sunitinib (Supplemental Fig. 1; supplemental materials are available online only at http://jnm.snmjournals.org), we found that 1 μM sunitinib did not affect dose-dependent intracellular accumulation of scVEGF/Cy fluorescent tracer by PAE/KDR cells (Fig. 1). Although sunitinib induced a small but significant difference in kinetics of scVEGF/Cy accumulation in the first 5–10 min, this difference disappeared after 15–20 min of incubation (Supplemental Fig. 2).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

Receptor-mediated uptake of scVEGF/Cy in presence of sunitinib. After 1 h of preincubation with 1 μM sunitinib, PAE/KDR and PAE cells in triplicate wells were exposed to varying concentrations of scVEGF/Cy with 1 μM sunitinib (Su-PAE/KDR and Su-PAE). Control PAE/KDR and PAE cells preincubated without sunitinib were exposed to same concentrations of scVEGF/Cy without sunitinib (Cont-PAE/KDR and Cont-PAE). After 20 min of exposure, cells were washed, fixed, and imaged. Error bars are shown only for data in presence of sunitinib. fl. u. = fluorescent units.

To track internalized scVEGF by fluorescent microscopy, we used fluorescent scVEGF/Al594 tracer. scVEGF/Al594 was internalized by PAE/KDR cells both with and without sunitinib treatment and was readily detected in specks of increased fluorescent signal accumulated in the perinuclear area (Supplemental Fig. 3). As expected, control PAE cells did not accumulate any detectable amounts of the tracer (data not shown). In contrast to sunitinib, bevacizumab, a VEGF-specific antibody that prevents VEGF binding to VEGFR, completely blocked scVEGF/Al594 uptake at a 2:1 bevacizumab-to-tracer molar ratio (Supplemental Fig. 4). Notably, bevacizumab did not block the internalization of scVEGF(2GA)/Al594 (Supplemental Fig. 4). That tracer is based on a scVEGF mutant with the G107A and G218A amino acid substitutions that correspond to the G88A substitutions in each subunit of native VEGF, which completely abrogates bevacizumab binding without affecting VEGFR-2 binding (16).

Taken together, the results obtained with fluorescent scVEGF-based probes indicate that it is unlikely that sunitinib would directly inhibit VEGFR-mediated internalization of ^99mTc-scVEGF tracer in vivo. We recently found that another inhibitor of VEGFR tyrosine kinase activity, pazopanib (17), does not inhibit receptor-mediated internalization of scVEGF-based tracers (18), supporting the notion that the use of such tracers is compatible with VEGFR inhibitory drug therapy.

Sunitinib Causes Rapid Decline in ^99mTc-scVEGF Tumor Uptake

To assess early effects of sunitinib treatment, MDA-231/luc tumor–bearing mice were given sunitinib at 80 mg/kg daily for 4 d. As an inhibitory dose of sunitinib is maintained in circulation for at least 12 h after drug treatment (11), we allowed for a 1-d recovery after the last sunitinib dose before imaging. Treated and control mice were injected with ^99mTc-scVEGF, imaged, and sacrificed; after sacrifice, tumors were harvested and sectioned for autoradiographic analyses of ^99mTc-scVEGF uptake and immunohistochemical staining for VEGFR-2 and the pan-endothelial marker CD31.

In agreement with a previous report (10), ^99mTc-scVEGF tracer readily accumulated in tumors in control, but not in sunitinib-treated, mice (Fig. 2A). Autoradiography of serial tumor sections, compared with healthy contralateral tissue, confirmed a significantly higher accumulation of ^99mTc-scVEGF in the tumor area in control mice than in sunitinib-treated mice (Supplemental Fig. 5).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Decreased ^99mTc-scVEGF tracer uptake and immunostaining for VEGFR-2 and CD31 after 4 d of sunitinib treatment and 1 d of recovery (1 d after treatment). (A) Representative SPECT images obtained with ^99mTc-scVEGF tracer for treated and time-matched untreated control MDA-231/luc tumor–bearing mice. Tomographic images are from center of tumors located in left axilla and left anterior chest wall. Standard of activity (capillary tube, 1 percentage injected dose per milliliter) is seen adjacent to right axilla. Arrowheads mark location of left axillary tumor. (B) Quantitative analysis of relative decrease in VEGFR-2 and CD31 immunostaining at rim and center of tumors from treated vs. control mice. Rectangular fields (5–10) for rim and center areas on images shown in Supplemental Figure 6 were analyzed. Rx = treatment.

Sunitinib Causes Rapid Decline in VEGFR-2 and CD31-Positive Cells in Tumor Vasculature

As reported previously, of 2 main VEGFRs, only VEGFR-2 is expressed in MDA-231/luc tumors and localized to CD31-positive endothelial cells (9). Thus, we expected that a sunitinib-induced decrease in ^99mTc-scVEGF uptake might be due to the changes in VEGFR-2 prevalence, accessibility, or internalization or recycling in tumor endothelial cells. As sunitinib and other VEGF- and VEGFR-targeting drugs are known to induce a rather rapid regression of tumor vasculature (19–25), we assessed the effects of sunitinib on VEGFR-2 and CD31 prevalence in tumors harvested from control and treated mice. Immunostaining revealed extensive vascularization in control tumors, with somewhat higher VEGFR-2 and CD31 density at the tumor edge adjacent to the underlying muscle tissue (Supplemental Fig. 6). We observed occasional small TUNEL-positive areas; however, in general, the level of apoptosis was low (Supplemental Fig. 6).

Four-day sunitinib treatment led to a significant decrease of VEGFR-2 and CD31 immunostaining throughout the tumor section, including tumor edges (Supplemental Fig. 6). Interestingly, quantitative analysis of immunostaining indicated that the magnitude of the sunitinib-induced decrease in VEGFR-2 was lower than that for CD31, both for the rim and center areas (Fig. 2B). The latter finding might indicate that endothelial cells with the higher levels of VEGFR-2 expression were more resistant to sunitinib than the rest of endothelial cells. Sunitinib treatment did not lead to the appearance of TUNEL-positive apoptotic cells in amounts corresponding to the decrease in CD31- or VEGFR-2–positive cells (Fig. 2B), most probably because of rapid shedding and clearance of the early-stage apoptotic tumor endothelial cells (26).

Increase in ^99mTc-scVEGF Uptake After 3-Day Recovery from Sunitinib Treatment

Several groups reported revascularization of model and human tumors after the end of treatment with VEGF- and VEGFR-directed drugs (25,27–29). We tested if the earliest steps in this process were associated with enhanced VEGFR prevalence and therefore could be detected using ^99mTc-scVEGF. A group of mice was treated with sunitinib for 5 d and then allowed to recover for 3 d before imaging. Compared with after 1 d of recovery (Fig. 2A; Supplemental Fig. 5), after 3 d of recovery smaller differences existed in tumor size between treated and control mice as visualized by SPECT (Fig. 3A) or autoradiography (Fig. 3B). Despite the differences in sizes between tumors from treated and control mice, the enhanced tracer uptake at the edges of the tumors from recovered mice was clearly visible on all sections. As judged by the quantitative analysis of autoradiographs, a 3-d recovery period allowed for tracer accumulation in both rim and center to levels that were correspondingly 2.8-fold (P = 0.0036) and 1.5-fold (P = 0.016), respectively, higher than in a group with a 1-d recovery period (Fig. 3C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Enhanced ^99mTc-scVEGF uptake and immunostaining for VEGFR-2 and CD31 after 4 d of sunitinib treatment and 3 d of recovery (3 d after treatment) relative to 1 d of recovery (1 d after treatment). (A) Representative SPECT image of treated and control MDA-231/luc tumor–bearing mice with ^99mTc-scVEGF. Tomographic images are from center of tumors located in left axilla and left anterior chest wall. Standard of activity (capillary tube, 1 percentage injected dose per milliliter) is seen adjacent to right axilla. Arrowheads mark location of left axillary tumor. (B) Representative autoradiographs of serial 60-μm tumor sections of treated and control mice. (C) Quantitative analysis of autoradiographs. Increase in average radioactivity uptake per pixel relative to contralateral pectoral muscle was calculated using ImageQuant TL software. Differences between tracer uptake in control and treated animals after 1 d of recovery were statistically significant, with P = 0.0001 for both rim and center. Differences between tracer uptake in control and treated animals after 3 d of recovery were statistically significant, with P = 0.009 for rim and P = 0.004 for center. Differences between 1-d and 3-d recovery were statistically significant, with P = 0.0036 for rim and P = 0.016 for center. (D) VEGFR-2 and CD31 immunostaining, ×4 objective. m = mouse. Rx = treatment.

Immunohistochemical analysis of VEGFR-2 and CD31 in tumors harvested from imaged mice revealed higher levels of both markers after 3 d of recovery relative to 1 d of recovery (Fig. 3D). Notably, at the recovery stage, most VEGFR-2 and CD31 immunostaining was observed at the periphery of the tumor (Fig. 3D). In a separate group of mice both the decline and the recovery of VEGFR-2 and CD31 were characterized in more detail. As judged by quantitative analysis of immunostaining, both VEGFR-2 and CD31 declined rapidly from the beginning of treatment and then recovered as rapidly when the treatment was stopped (Figs. 4A and 4B; Supplemental Fig. 7).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4. 

Sunitinib-induced decline in VEGFR-2 and CD31 immunostaining and rapid posttreatment recovery of vascularization at tumor edge. (A and B) Quantitative analysis of immunostaining for VEGFR-2 and CD31 (representative fields are shown in Supplemental Fig. 7) and percentage of pixels positive for, respectively, VEGFR-2 and CD31 immunostaining in 3–10 rectangular fields per section, obtained from 2–3 mice. Boxes represent interquartile region where middle 50% of data points occur, error bars represent minimum and maximum values in each group, and lines show median of each group. Comparisons between groups are indicated by brackets, and significance was determined by applying Kruskal–Wallis Z test with Bonferroni adjustment for multiple comparisons. *P < 0.05. **P < 0.01. ***P < 0.001.

Imaging Effects of 2 Weekly Cycles of Sunitinib

In clinical practice, sunitinib and other VEGFR inhibitors are usually given in several cycles. In the initial assessment of a 2-cycle treatment, 5 mice were treated with sunitinib for 5 d, given 2 d of rest, and treated again for 5 d. Then 2 mice were imaged after 1 d and 3 mice after 3 d of recovery. Judging by SPECT with ^99mTc-scVEGF, after both 1 and 3 d of recovery there were mice with high and low levels of tracer uptake (Fig. 5A for 3-d recovery). The difference in tracer uptake was confirmed by autoradiography of serial sections of corresponding tumors (Fig. 5B). Tumors with high uptake (1 after 1 d of recovery and 2 after 3 d of recovery) had tracer uptakes similar to those of untreated controls (2 mice for each group). These preliminary data suggest that, at least in some mice, the sensitivity of tumor endothelium to sunitinib might be significantly decreased after 2 cycles of weekly treatment.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5. 

Two-cycle sunitinib treatment results in tumors with high and low levels of tracer uptake at 1 and 3 d after treatment. (A) SPECT of representative treated MDA-231/luc tumor–bearing mice with high and low levels of ^99mTc-scVEGF uptake after the following protocol: 5 d treatment plus 2 d recovery plus 5 d treatment plus imaging 3 d after treatment. Tomographic images are from center of tumors located in left axilla and left anterior chest wall. Standard of activity (capillary tube, 1 percentage injected dose per milliliter) is seen adjacent to right axilla. Arrowheads mark location of left axillary tumor. (B) Autoradiographs of serial 60-μm tumor sections from 2 cycles of treatment plus imaging 1 or 3 d after treatment. Rx = treatment; m = mouse.