Abstract
One of the hallmarks of tumor neovasculature is the prevalence of immature vessels manifested by the low degree of recruitment of vascular mural cells such as pericytes and smooth muscle cells. This difference in the architecture of the vascular bed provides an important therapeutic window for inflicting tumor-selective vascular damage. Here we demonstrate the application of gradient echo magnetic resonance imaging (MRI) for noninvasive in vivo mapping of vascular maturation, manifested by the ability of mature vessels to dilate in response to elevated levels of CO2. Histological α-actin staining showed a match between dilating vessels detected by MRI and vessels coated with smooth muscle cells. Switchable, vascular endothelial growth factor (VEGF)-overexpressing tumors (C6-pTET-VEGF rat glioma s.c. tumors in nude mice) displayed high vascular function and significant vascular damage upon VEGF withdrawal. However, damage was restricted to nondilating vessels, whereas mature dilating tumor vessels were resistant to VEGF withdrawal. Thus, MRI provides in vivo visualization of vascular maturity and prognosis of vascular obliteration induced by VEGF withdrawal.
INTRODUCTION
Remodeling of the vascular bed during angiogenesis initiates with the formation of a fragile endothelial capillary bed. Vascular maturation and stabilization is a secondary process, which involves the recruitment of vascular smooth muscle cells and pericytes. The existence of a window of plasticity in the immature vasculature allows proper adjustment of vessel density to tissue needs during angiogenesis. This plasticity window provides the possibility for selective obliteration of immature tumor neovasculature.
The therapeutic applicability of neovascular obliteration by antiangiogenic therapy depends on the fact that mature blood vessels are not susceptible to such damage (1, 2, 3) . One of the hallmarks of vessel maturation is the recruitment of pericytes and smooth muscle cells (4 , 5) . In the retina as well as in tumors, histological analysis has established that pericyte coating is associated with the resistance of vessels to VEGF 3 withdrawal (2 , 3) . Interference with the association between endothelial cells and pericytes results in severe vascular malformations and increased vascular fragility, as detected recently in mice deficient in tissue factor, LKLF, and TIE2 and its ligand, ANG-1 (6, 7, 8, 9) . It was therefore postulated that vascular maturation could provide a predictive marker for the sensitivity of neovasculature to VEGF withdrawal in antiangiogenic therapy and for vascular stabilization in proangiogenic therapy.
Damage of immature neovasculature was induced here by specific withdrawal of VEGF using switchable C6-pTET-VEGF glioma cells (3 , 10) . VEGF, a central growth factor participating in the induction of normal and pathological angiogenesis (11, 12, 13, 14, 15, 16) , is also essential for the maintenance of immature blood vessels (3 , 10 , 17) . In the C6-pTET-VEGF cells, VEGF165 is constitutively overexpressed in the absence of tetracycline, and overexpression can be switched off by the addition of antibiotics (10) . Previous histological analysis of s.c. C6-pTET-VEGF tumors in nude mice revealed endothelial detachment and TUNEL-positive apoptotic endothelial cells 24 h after VEGF withdrawal (10) . Thus, VEGF is an essential survival factor for newly formed tumor vasculature, as demonstrated previously for retinal angiogenesis (17) . After VEGF withdrawal, histologically intact vessels were frequently associated with smooth muscle cells, as detected by staining with a smooth muscle actin (3) .
In the study reported here, we used hypercapnia (elevated CO2) and hyperoxia (elevated O2) for in vivo MRI analysis of vascular maturation and functionality. The rationale for this experimental approach is that pericyte and smooth muscle cell-coated vessels, in contrast with immature endothelial capillaries, will dilate in response to hypercapnia (VD). However, all functional blood vessels will show a change in hemoglobin saturation in response to hyperoxia, reflecting the capacity of erythrocytes to deliver inhaled oxygen to the tumor vasculature (VF). Both VD and VF can be detected by gradient echo MRI using the intrinsic contrast generated by changes in deoxyhemoglobin, blood volume, and blood flow (18, 19, 20, 21, 22, 23) .
MATERIALS AND METHODS
Animal Protocols.
C6-pTET-VEGF cells were derived as reported previously (10) . Cells were cultured in DMEM supplemented with 5% FCS (Biological Industries Israel), 50 units/ml penicillin, 50 μg/ml streptomycin, and 125 μg/ml fungizone (Biolab Ltd.) with the addition of 1 μg/ml tetracycline (Sigma). The s.c. tumors were generated by inoculation of cells in the lower back of male CD1-nude mice (2-month-old mice; body weight, 30 g; 106 cells/mouse). MRI experiments were initiated 16 days after inoculation. VEGF overexpression was switched off by adding tetracycline (100 μg/ml + 5% sucrose) to the drinking water of the mice, as reported previously (10) . Control mice were given water with 5% sucrose for 48 h. Animal experiments were approved by the Animal Committee of The Weizmann Institute of Science.
MRI Measurements.
MRI experiments were performed on a horizontal 4.7 T Bruker Biospec spectrometer using an actively RF decoupled surface coil (2 cm in diameter) imbedded in a Perspex board and a bird cage transmission coil. Mice were anesthetized (75 mg/kg ketamine + 3 mg/kg xylazine, i.p.) and placed supine with the tumor located at the center of the surface coil. Functionality and maturation of the neovasculature were determined from gradient echo images acquired during the inhalation of air, air-CO2 (95% air and 5% CO2), and oxygen-CO2 (95% oxygen and 5% CO2; carbogen). The different gas mixtures were administered to the mice via a home-built mask. Four images were acquired at each gas mixture (117 s/image; slice thickness, 0.5 mm; TR = 230 ms; spectral width, 25,000 Hz; field of view, 3 cm; 256 × 256 pixels; in plane resolution, 110 μm; TE = 10 ms; two averages). Other experimental details were as reported previously (18) .
Data Analysis.
MRI data were analyzed on an Indigo-2 work station (Silicon Graphics) using Paravision software (Bruker) and Matlab (The Math Works Inc.).
VF was derived as reported previously
(24)
from images acquired during the inhalation of oxygen-CO2 (95% oxygen and 5% CO2) and air-CO2 (95% air and 5% CO2) using the following equation:
Where Ioxygen-CO2 and Iair-CO2 are the mean signal intensity during the inhalation of oxygen-CO2 and air-CO2, respectively; Y is the fraction of oxyhemoglobin; b is the volume fraction of blood; and CMRI = 599 s−1 at 4.7 T
(18)
. This parameter measures the capacity of erythrocyte-mediated oxygen delivery from the lungs to each pixel in the image
(18)
. The sensitivity for detection of functional vessels will be reduced for highly oxygenated blood and might therefore differ between arterial and venous vessels. It is important to note, however, that oxygenation of s.c. blood in anesthetized mice is relatively low (17.7 and 15.7 mmHg in s.c. arterioles and venules, respectively; Ref.
25
).
VD was derived from air and air-CO2 images using the following equation:
in which Iair is the mean signal intensity during inhalation of air. Positive VD corresponds to increased signal intensity by hypercapnia due to increased blood flow.
In addition to this analysis, a pixel-by-pixel t test analysis of VD and VF was performed showing a map of significance of change in signal intensity between images acquired during inhalation of air and air-CO2 and between images acquired during inhalation of air-CO2 and oxygen-CO2, respectively. Results analyzed using both approaches yielded a similar distribution for mature and functional vessels (Fig. 1) ⇓ . Data are presented in color overlayed on the gray scale baseline image for values of VD > 0.01, VF > 0.006, and P < 0.05.
In vivo analysis of VF and VD. a, gradient echo image of a C6-pTET-VEGF tumor acquired 48 h after VEGF withdrawal; b, photograph of the same tumor showing the large vessels clearly visible by MRI. Functionality and maturation of the neovasculature were derived from such gradient echo images acquired during inhalation of air, air-CO2, and oxygen-CO2. Representative data of a single slice from two different mice are presented (first mouse, a, b, and e–l; second mouse, m–p). Data were analyzed by two approaches. Color-coded VF maps (e, i, m, and o) and VD maps (f, j, n, and p) were derived, and values above a threshold (VF > 0.006 and VD > 0.01) were overlaid on a gray scale image. A pixel-by-pixel t test analysis was used to determine functionality (g and k) and maturation (h and l), and color-coded Ps for P < 0.05 were overlaid on the gray scale image. c, values for VD and VF are defined in the coded color scale. The black bar scale of 5 mm applies for e–l. d, color scale Ps for the t test analysis of vascular maturation and functionality. e–h, maps derived for VEGF overexpression, i.e., before the administration of tetracycline. i–l, maps derived 48 h after switching off VEGF overexpression by the addition of tetracycline. Arrows point to mature vessels, which are resistant to VEGF withdrawal; arrowheads point to immature vessels showing loss of function upon VEGF withdrawal. m–p, VF (m and o) and VD maps (n and p) acquired before (m and n) and 10 h after (o and p) VEGF withdrawal. The black bar in m is a 5-mm scale for m–p.
The change in signal intensity due to hypercapnia (VD) was found to be predominantly a result of a change in the apparent T1 relaxation due to a change in blood flow. On the other hand, the signal change due to hyperoxia (VF) was due to a change in T2* due to a change in blood oxygenation. 4
RESULTS
Vascular functionality and maturation in C6-pTET-VEGF tumors were determined by MRI at three time points, immediately before and 10 and 48 h after the administration of tetracycline (Figs. 1 ⇓ 2 ⇓ 3 ⇓ ; Tables 1 ⇓ and 2 ⇓ ). Control C6-pTET-VEGF tumors were imaged at two time points (48 h apart) without the addition of tetracycline (Table 2) ⇓ . The tumors were then fixed for histological analysis. Mice were imaged during inhalation of air, air-CO2, or oxygen-CO2, and the images were used for derivation of the VF and VD maps (see “Materials and Methods”).
MRI-detected vascular maturation coincides with resistance to VEGF withdrawal. VF and VD maps were acquired before and after VEGF withdrawal (± tetracycline). The mean ± SE (n = 6) values of VD and VF of four mice (one to two slices/mouse) are plotted for regions of interest selected at three locations: a, normal tissue, aproximately 7 mm from the border of the tumor; b, the tumor periphery; and c, the tumor center. Paired t test analysis was applied to evaluate the effects of VEGF withdrawal. A highly significant loss of VF after the administration of tetracycline was measured at the tumor center (c; ***, P = 0.005, paired Student t test; n = 6). This loss of function correlated with vascular immaturity manifested by the significantly reduced mean VD in the tumor center (c) relative to the tumor periphery (b; **, P = 0.01). Surprisingly, VEGF withdrawal resulted in elevated VD in the tumor center as well as in normal tissue (a and c; *, P = 0.05).
Vascular dilation detected by MRI corresponds to histologically mature vessels. a, MRI analysis of VF 48 h after tetracycline administration. b, VD of the same tumor. c, H&E staining of a section in the same plane as the MRIs shown in a and b. d, magnification of the boxed region in c shown with the arrowheads in a and b. This region maintains VD and VF and contains large peripheral blood vessels leading into the tumor. e, anti-smooth muscle actin staining confirms that these blood vessels (arrows in d and e) that retain VD are covered by smooth muscle. f, lectin staining of endothelium shows that the peripheral and functional blood vessels are intact (black arrows), whereas those immediately adjacent but inside the tumor are not intact (white arrow). g, a higher magnification of lectin staining inside the tumor demonstrates that staining is primarily of vessel remnants damaged by VEGF withdrawal, and not intact blood vessels. *, necrotic region of the tumor aids in orienting the histology with the MRI (a, b, c). h, TUNEL analysis of programmed endothelial cell death (arrows) in the same region as shown in g. i, magnification of one capillary from h showing TUNEL-positive endothelial cells surrounding a lumen with a RBC (arrowhead). Lectin, anti-smooth muscle actin, and TUNEL staining were done as reported previously (3) .
Effects of hyperoxia and hypercapnia on signal intensity in C6-pTET-VEGF tumors
Reproducibility of the assessment of VD and VF in C6-pTET VEGF tumors
Overexpression of VEGF (i.e., before the addition of tetracycline) led to the generation of fully functional blood vessels, as evidenced by the significant signal enhancement (P = 0.0014; n = 4) detected upon inhalation of oxygen-CO2 relative to inhalation of air-CO2 (Fig. 1, e and g ⇓ ; Fig. 2, b and c ⇓ ; Table 1 ⇓ ). The extent of VF, which was highest at the tumor periphery and lowest in normal tissue, was in agreement with the spatial distribution of blood vessels (Fig. 1b) ⇓ .
Images acquired during inhalation of air and air-CO2 (VD) showed a significant change in signal intensity for normal vessels and vessels in the tumor periphery (P = 0.04; n = 4; Table 1 ⇓ ). VD in response to hypercapnia was significantly lower in the tumor center relative to the tumor periphery (Fig. 1, f and h ⇓ ; Fig. 2, b and c) ⇓ . Maturation of the tumor vasculature appeared to proceed from the margins of the tumor inward and could represent secondary migration of supporting mural cells such as pericytes and vascular smooth muscle cells. The lack of response to CO2 (low VD) within the tumor was not due to hypovascularity, as evidenced by the high density of functional vasculature (high VF; Fig. 1, e and g ⇓ ). The bimodal response in the VF and VD maps enabled us to define two distinct populations of functional tumor blood vessels that differ in their degree of maturation.
Forty-eight h after switching off VEGF overexpression by adding tetracycline to the drinking water, VF within the treated tumors was significantly reduced (P = 0.005; Fig. 1, i and k ⇓ , arrowheads; Fig. 2c ⇓ ). Damage to the vascular bed detected by MRI was not immediate, and, accordingly, no reduction in VF was detected 10 h after the administration of tetracycline (Fig. 1, m and o ⇓ ; Table 2 ⇓ ), a time point at which no damage to endothelial cells could be detected in histological sections. The results at 48 h are consistent with impaired oxygen delivery and massive vascular damage associated with VEGF withdrawal for tumor blood vessels. TUNEL-positive staining of endothelial cells was observed in tumor regions in which MRI measurements showed a loss of VF (Fig. 3) ⇓ . However, 48 h after VEGF withdrawal, no TUNEL staining was detected in the tumor cells, thus showing the direct and selective effect of VEGF withdrawal on endothelial cell survival (Fig. 3) ⇓ .
VEGF withdrawal did not lead to complete collapse of the entire tumor-induced vasculature. Vessels at the immediate periphery of tumors and a few vessels within the tumors effectively managed to escape destruction by VEGF withdrawal and continued to show similar oxygen-dependent signal modulation and significant VF (Fig. 1, i and k) ⇓ . We therefore examined the VD maps to test the hypothesis that vascular maturation is correlated with vessel resistance. Indeed, in all cases, tumor and normal vessels showing CO2-induced dilation (high VD) were refractory to vascular occlusion induced by VEGF withdrawal (Fig. 1 ⇓ , arrows). Furthermore, vascular maturation (VD) detected by MRI matched the maturation defined by the smooth muscle cell coating of vessels, as revealed in corresponding histological sections of the same tumors (Fig. 3) ⇓ .
DISCUSSION
The induction of angiogenesis in solid tumors initiates with the stimulation of proliferation and migration of endothelial cells. A pivotal role has been implicated for VEGF and its receptors in these early stages of vascular sprouting (26) . However, recent studies suggested that the role of VEGF is not completed with the establishment of a primary endothelial plexus and the initiation of perfusion. In fact, VEGF is essential for maintaining the viability and functionality of endothelial cells in these immature vessels, and acute withdrawal of VEGF results in endothelial programmed cell death and vascular collapse, as reported previously (2 , 3 , 10 , 17) and shown here.
Stabilization of the vascular bed is achieved by recruitment of pericytes and smooth muscle cells. Vascular maturation has long been recognized as an integral component of angiogenesis (5 , 27) . Some of the molecular signals that can promote vascular maturation have been revealed recently. For example, platelet-derived growth factor has been implicated in the proliferation and migration of pericytes (28) , and HB-EGF has been demonstrated to induce VEGF secretion by vascular smooth muscle cells (29) . TIE2 and its ligands, ANG-1 and ANG-2, were recently shown to control vascular maturation (30 , 31) . ANG-1 promotes the association of vessels with pericytes and appears in the later stages of angiogenesis, whereas its inhibitor, ANG-2, induces dissociation of pericytes. ANG-2 is thus elevated in the early stages of angiogenesis and in immature neovasculature.
Angiogenesis in the normal ovarian cycle is characterized by tight regulation of pericyte migration, which correlates well with the patterns of expression of ANG-1 and ANG-2 (32) . In contrast, in tumors, the vascular bed is frequently immature, with many vessels devoid of pericytes or smooth muscle cells. Recent studies showed that hypervascularized tumors were characterized by elevated expression of ANG-2 (33 , 34) . Thus, the low degree of maturation of the tumor vasculature provides an attractive window for specific obliteration of the tumor neovasculature (3) . This finding calls for efficient noninvasive methods that would enable us to classify the degree of vascular maturation in tumors for patient selection and follow-up of treatment.
Vascular functionality and maturation were differentially mapped here by the change in MRI signal intensity in response to hyperoxia, which changed hemoglobin saturation (VF), and hypercapnia, which affected signal intensity due to the change in blood flow (VD). VF mapping by MRI using hyperoxia is a valid approach for s.c. tumors in anesthetized mice, in which hemoglobin in arterial blood is not fully oxygenated. It must be noted, however, that in well-oxygenated organs, VF can be assessed by alternative MRI methods, including signal changes in response to hypoxia, as demonstrated previously for the brain (35) , and arterial spin labeling (36) or by tracking exogenously administered contrast material. Mapping of vascular maturation by signal changes associated with increased flow in response to hypercapnia, as reported here, is not sensitive to blood oxygenation and should therefore be applicable in any organ.
We show here that tumor vascular maturation, which is manifested by smooth muscle cell recruitment, can be detected by MRI through the response to CO2-induced VD. Moreover, we show that the maturation detected by MRI marks the window of susceptibility of tumor vessels to VEGF withdrawal. Thus, noninvasive high-resolution MRI provides an in vivo approach for three-dimensional assessment of VF and maturation (VD). MRI could provide a tool for prediction of vascular response to anti-VEGF treatment of tumors as well as for noninvasive assessment of vascular maturation during proangiogenic therapy.
Acknowledgments
We thank Prof. Eli Keshet, Prof. Yoram Salomon, and Dr. Peter Bendel for helpful suggestions.
Footnotes
-
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-
↵1 Supported by research grants from the Israel Science Foundation and NIH Grant RO1 CA75334-01A1 (to M. N.). R. A. is a recipient of a fellowship from the Charles Clore foundation. M. N. is incumbent of a Research Career Development Award from the Israel Cancer Research Fund.
-
↵2 To whom requests for reprints should be addressed. Phone: 972-8-934-2487; Fax: 972-8-934-4116; E-mail: lhneeman{at}wiccmail.weizmann.ac.il
-
↵3 The abbreviations used are: VEGF, vascular endothelial growth factor; MRI, magnetic resonance imaging; ANG, angiopoetin; TUNEL, terminal deoxynucleotidyl transferase-mediated nick end labeling; VF, vascular function; VD, vasodilation; TE, echo time; LKLF, lung kruppel-like factor; TR, repetition time; RF, radio frequency; HB-EGF, heparin binding like growth factor.
-
↵4 H. Dafni and M. Neeman, unpublished observations.
- Received February 16, 1999.
- Accepted August 6, 1999.
- ©1999 American Association for Cancer Research.
References
Volume 59, Issue 19
