Magnetic Resonance Imaging Visualization of Hyaluronidase in Ovarian Carcinoma

Hyaluronan, also known as hyaluronate or hyaluronic acid, is a high molecular weight linear glycosaminoglycan composed of repeating disaccharides of glucuronic acid and N-acetylglucosamine. Hyaluronan is an important structural element in cartilage, synovial fluid, skin of vertebrates, and vitreous humor of the eye (1). Hence, hyaluronan plays an important role in maintenance of intact architecture in normal tissues; it absorbs a large volume of water and creates a gel-like environment. Hyaluronan affects physiologic and pathologic processes that require cell movement, such as tissue organization and morphogenesis, wound healing, inflammation, angiogenesis, and tumor metastasis.

High molecular weight hyaluronan is degraded by hyaluronidase. Low molecular weight degradation products are known to stimulate endothelial cell proliferation and promote neovascularization, whereas the intact high molecular weight hyaluronan is antiangiogenic (2). Hyaluronidase was first identified in 1928 as a “spreading factor” of viral agents and was later characterized as hyaluronan-degrading enzyme and named hyaluronidase (3–5). Hyaluronidase activity is high in necrotic tumors and elevated in the sera of cancer patients (6).

Six hyaluronidase-like genes, coding for different enzymes with varying substrate specificities, based on their enzymatic activity and optimal pH for activity are identified in mammals (7). The main hyaluronidases that were studied are hyaluronidase-1, hyaluronidase-2, and PH-20. Hyaluronidase-1, the major hyaluronidase found in the plasma and urine is a 57-kDa protein composed of a single peptide chain of 49 kDa, with ∼8 kDa of post-translational glycosylation. Hyaluronidase-1 is also a lysosomal enzyme that can cleave hyaluronan to small tetrasaccharides and disaccharides (8) and has ∼40% identity to the enzyme PH-20 found mainly in sperm (7). Hyaluronidase-1 is also known as LUCA-1 defined by functional tumor suppressor activity (9). The activity of hyaluronidase-1 in the serum is suppressed by IαI, a potent inhibitor of hyaluronidase (10).

Hyaluronidase-2 is generally anchored to the plasma membrane by a glycosylphosphatidylinositol link. It cleaves high molecular weight hyaluronan to ∼20-kDa fragments (50 disaccharides), which are proangiogenic. Hyaluronidase-2 overexpression accelerates tumor formation of murine astrocytoma cells (11) but can also accelerate apoptosis (12). It seems that both hyaluronidase-1 and hyaluronidase-2 can function either as oncogene or as tumor suppressor gene product. Secretion of hyaluronidase by cancer cells can contribute to their aggressiveness and invasiveness. Thus, hyaluronidase provides the intermediate hyaluronan fragments that induce angiogenesis (13). Moreover, secretion of hyaluronidase by the cancer cell enables digestion of the hyaluronan barrier and thus facilitates invasion of tumor cells to neighboring organs and tissues. It is reported that hyaluronidase activity in ovarian cancer tissue is significantly higher than in endometrial cancer tissue. A significant correlation is found between hyaluronidase activity and metastasis of ovarian cancer (14). Moreover, treatment of conditioned medium with hyaluronidase increases the adhesion of ovarian cancer tumor cells to mesothelial monolayer that serves as a model for metastatic dissemination in the peritoneal cavity (15). Because hyaluronan, hyaluronidase, and CD44 (hyaluronan receptor) are involved in the progression of ovarian carcinoma (16–19), noninvasive detection of the presence and activity of hyaluronidase may give indication for the process and presence of metastasis.

Several methods are used for the detection of hyaluronidase, including quantitative spectroscopic ELISA-like assay (20), a microtiter-based assay for hyaluronidase activity (21), the Morgan-Elson reaction (22) and its fluorimetric version (23), chromatography (24), and particle exclusion assay (25). However, none of these methods can be applied in vivo for noninvasive imaging.

The goal of this work was to develop a novel method for the detection of hyaluronidase by magnetic resonance imaging (MRI). A new contrast material was constructed by linking of gadolinium-diethylenetriaminepentaacetic acid (GdDTPA)–tagged hyaluronan to nontoxic agarose beads. We report here that hyaluronidase secreted by ovarian carcinoma cells caused significant activation of the contrast material, manifested by alternations in R₁ and R₂ relaxation rates, in vitro as well as in the surrounding of s.c. ovarian tumor xenografts. Because hyaluronidase activity is correlated with the progression and metastatic spread of ovarian carcinomas to the peritoneum, this method could potentially be further developed for clinical molecular imaging of hyaluronidase.

Synthesis of hyaluronan-gadolinium-diethylenetriaminepentaacetic acid-beads. Hyaluronan-GdDTPA was synthesized as described (26). Briefly, hyaluronan (50 mg, extracted from human umbilical cord; Sigma Chemical Co., St. Louis, MO) was dissolved in MES (0.1 mol/L, 50 mL, pH 4.75, Sigma). The carboxyl groups of hyaluronan were activated by addition of N-(3-dimethylaminopropyl)-N-ethylcarbodiimide hydrochloride (EDC; 2.4 mg) followed by addition of ethylenediamine (2 mg) and stirring overnight at room temperature. The product was purified by dialysis against double-distilled water, reacted with DTPA dianhydride (18.7 mg, Sigma), and dissolved in filtered DMSO (55 mL) overnight at room temperature. After second dialysis, gadolinium (III) chloride (23.5 mg, Sigma) in double-distilled water was added, the mixture was stirred for 24 hours, and a third dialysis was done. Gadolinium content in the end product was measured by inductively coupled plasma-mass spectrometry (ICP-MS) showing ∼0.018 mg gadolinium bound to 1 mg hyaluronan (i.e., a ratio of 1:17 between gadolinium and hyaluronan disaccharides, respectively). The product was lyophilized and stored at room temperature in a desiccator and was reconstituted to a working concentration in double-distilled water.

Hyaluronan-GdDTPA was further covalently bound to avidin-agarose beads (bead size, 0.040-0.165 mm; Sigma). Avidin-linked agarose beads (50 μL bead suspension containing ∼6 × 10⁻⁹ mol avidin; Sigma) were mixed with 5-(biotinamido)pentylamine (2 μg; Pierce Biotechnology, Rockford, IL) in MES (pH 5.5, 0.1 mol/L) at room temperature to let the biotin in the 5-(biotinamido)pentylamine link to the avidin on the agarose beads. The carboxyl groups of hyaluronan (25 mg hyaluronan-GdDTPA) were activated in MES by EDC (2 mg) and added to the 5-(biotinamido)pentylamine–conjugated beads, stirred at room temperature overnight, and purified by dialysis against double-distilled water. In contrast with the source agarose beads, hyaluronan-GdDTPA-beads are highly hydrated and generate a clear and stable suspension, whereas the native agarose beads are nontransparent and tend to settle rapidly.

Synthesis of hyaluronan-Bodipy. Hyaluronan-Bodipy was synthesized as described previously (27) and purified by dialysis.

Cell culture. Human epithelial ovarian carcinoma cell lines ES-2 and OVCAR-3 (kindly provided by Prof. Steffen Hauptmann, Institute of Pathology, Rudolf-Virchow-Haus, Berlin, Germany) were cultured in DMEM supplemented with 10% FCS. Chondrocytes (RCJ-P, rat chondrocytes from fetal calvaria, batch 15.01.98; Prochon Biotec Ltd., Rehovot, Israel) were cultured in α-MEM supplemented with 15% FCS. All media were supplemented with 100 units/mL penicillin, 0.1 mg/mL streptomycin, 0.06 mg/mL amphotericin, and 0.292 mg/mL l-glutamine.

Particle exclusion assay. The very thick hyaluronan coat of the chondrocytes (typically ∼4.4 ± 0.7 μm; ref. 28) was used here as the substrate for analysis of its degradation by hyaluronidase-containing samples. Chondrocytes were grown to confluence in 10-cm culture dish, harvested, diluted 1:100, and seeded on a particle exclusion assay plate (35-mm culture dish with 14-mm microwell; MatTek, Ashland, MA). After overnight incubation, the medium was replaced by fresh medium, hyaluronidase in PBS, or by conditioned medium removed from different ovarian carcinoma cell lines for incubation of 1 hour in 37°C. This incubation allowed hyaluronidase in the different treatments to degrade the hyaluronan layer around the chondrocytes. After incubation, the medium was replaced by human RBC type O (Sigma) diluted in DMEM. The RBC are excluded from the hyaluronan around the chondrocytes and are used here to delineate the border of hyaluronan, which is a transparent gel, allowing it to become visualized by light microscopy. Samples were viewed in visible light using an Axioskop microscope (Zeiss, Oberkochen, Germany). Images were acquired using a scientific-grade cooled charged-coupled device camera (Model CC350, Photometrics Co., Tucson, AZ). The experiment was repeated five times, and in each, at least five cells from a particular treatment were measured.

Reverse transcription-PCR. Total RNA was extracted using TRI-Reagent (Molecular Research Center, Inc., Cincinnati, OH) according to the manufacturer's instruction and reverse transcribed in 20 μL volume using RNase reverse transcriptase (SuperScript II, Invitrogen/Life Technologies, Inc., Carlsbad, CA) with 180 pmol hexamer random primer. Aliquots (2 μL) of the reverse transcription products were used for PCR. The following sense and antisense primers were used: hyaluronidase-2 sense 5′-GCTCAAGCCCACAGCACCAC-3′ and antisense 5′-CAGTGTCTCCAGCATGAACTGC-3′ and hyaluronidase-1 sense 5′-CACCGTCTGGAATGCAAACA-3′ and antisense 5′-GTTGGCTACCACATCGAAGACA-3′ (29). PCR variables were as follows: hyaluronidase-1, 94°C for 3 minutes, 31 cycles of 94°C for 20 seconds, 58°C for 30 seconds, and 72°C for 45 seconds followed by 72°C for 3 minutes; hyaluronidase-2, 94°C for 3 minutes, 24 cycles of 94°C for 20 seconds, 60°C for 30 seconds, and 72°C for 45 seconds followed by 72°C for 3 minutes. S16 was used for internal control: 5′-CGTTCACCTTGATGAGCCCAT-3′ and 5′-TCCAAGGGTCCGCTGCAGTC-3′.

Western blotting analysis. Confluent ovarian carcinoma cells OVACR-3 and ES-2 were washed with serum-free DMEM and incubated in serum-free medium for 24 or 48 hours. The medium was collected, centrifuged to remove cells, and concentrated using Amicon Ultra 10,000 MWCO (Millipore, Bedford, MA). Concentrated media were electrophoresed (12% SDS-polyacrylamide gel). Blocked membranes [5% milk in TBS-Tween 20 (TBST), 1 hour, room temperature] were incubated overnight at 4°C with either anti-hyaluronidase-1 or anti-hyaluronidase-2 polyclonal antibody (diluted 1:1,000, produced by Prof. R. Stern). Membranes were washed thrice with TBST and incubated with horseradish peroxidase–labeled antibodies (1:10,000; Zymed, Inc., San Francisco, CA). The immunoreactive bands were detected by commercial enhanced chemiluminescence (Santa Cruz Biotechnology, Santa Cruz, CA). The intensity of the signal was quantified by computerized densitometry (Quantity One, Bio-Rad, Hercules, CA).

Immunofluorescent cell staining. Polyclonal antibodies against hyaluronidase-1 and hyaluronidase-2 were kindly provided by R. Stern. Antibodies were prepared in rabbits using synthetic peptides from hyaluronidase-1 and hyaluronidase-2, corresponding to amino acids 104 to 120 and 100 to 116, respectively, as described previously (30). Cells seeded on 18-mm coverslips were cultured to confluence, washed twice in PBS, and fixed in 2.5% paraformaldehyde for 20 minutes in room temperature. Following three washes in PBS, the cells were permeabilized in 0.2% Triton (Sigma) in PBS and washed again. The coverslips were incubated with anti-hyaluronidase-1 or anti-hyaluronidase-2 antibody (diluted 1:200) for 1.5 hours at room temperature. Following three washes with PBS, the coverslips were incubated with secondary antibody (goat anti-rabbit FITC, Jackson ImmunoResearch, West Grove, PA) followed by 4′,6-diamidino-2-phenylindole (DAPI) nuclear staining (Sigma, 1:100). In control sections, only the secondary antibody was used.

Ovarian carcinoma tumor xenografts. All animal experiments were approved by the Institutional Animal Care and Use Committee.

CD-1 female nude mice 12 weeks old were inoculated with 3 × 10⁶ ES-2 cells. MRI studies were done 3 weeks after inoculation. The mice were anesthetized with i.p. ketamine (75 mg/kg; Ketaset, Fort Dodge Laboratories, Fort Dodge, IA) and xylazine (3 mg/kg; XYL-M, VMD, Arendonk, Belgium). A vein flow (24GA, BD Neoflon, Becton Dickinson, Helsingborg, Sweden) was inserted s.c. for local administration of hyaluronan-GdDTPA-beads (5 mg/mL).

In vitro magnetic resonance imaging studies. Hyaluronan-GdDTPA-beads (1 mg/mL) were incubated with hyaluronidase for 15 minutes at 37°C. The enzyme used was either bovine testes hyaluronidase (Sigma, type IV-S from bovine testes, 0.3 mg/mL) or hyaluronidase secreted to the conditioned medium of ovarian carcinoma cell lines.

MRI measurements were done on a 400-MHz (9.4 T) vertical wide-bore DMX spectrometer equipped with a microimaging attachment with a 5-mm Helmholtz radiofrequency coil or on a horizontal 4.7 T Biospec (Bruker, Karlsruhe, Germany). For R₂ measurements, multiecho spin echo sequence was used to acquire eight consecutive echoes with interecho spacing of 10 ms [echo time (TE) 10, 20, 30, 40, 50, 60, 70, and 80 ms, repetition time (TR) 2,000 ms, two averages, field of view (FOV) 1 × 1 cm, slice thickness 1 mm, matrix 128 × 128, spectral width 50,000 Hz]. For R₁ measurements, spin echo images were acquired at eight different TRs ranging between 2,000 and 200 ms (two averages, FOV 1 × 1 cm, slice thickness 1 mm, matrix 128 × 128, spectral width 50,000 Hz).

For the kinetic analysis, the sample was equilibrated to 37°C within the spectrometer before the beginning of the measurements. R₂pre was measured and the enzyme was injected to the test tube through a catheter and R₂post was measured.

In vivo magnetic resonance imaging measurements. MRI experiments were done on a horizontal 4.7 T Biospec spectrometer using an actively radiofrequency decoupled 1.5 cm surface coil embedded in a Perspex board and a birdcage transmission coil. Anesthetized mice were placed supine with the graft located above the center of the surface coil. The mice were immobilized using adhesive tape and covered with a paper blanket to reduce temperature drop during the measurement. For R₂ measurements, multiecho spin echo images were acquired with TR 500 ms, TE 10 to 80 ms, slice thickness 1 mm, and FOV 3.5 × 3.5 cm. For R₁ measurements, a series of T₁-weighted three-dimensional gradient-echo images, with pulse flip angles of 5°, 15°, 30°, 50°, and 70°, were acquired to determine the R₁ relaxation rate before administration of the contrast material. Variables were as follows: TR 10 ms, TE 3.561 ms, two averages, spectral width 50,000 Hz, FOV 3.5 × 3.5 × 1.75 cm (obtained by acquisition using an actively decoupled surface coil), matrix 128 × 128 × 64, zero filled to 128 × 128 × 128. Consecutive postcontrast T₁-weighted three-dimensional gradient-echo images were acquired 2 to 30 minutes after the contrast agent was administered (pulse flip angle 15°).

Analysis of the magnetic resonance imaging data. MRI data were analyzed on a personal computer using Matlab (The Math Works, Inc., Natick, MA). Images acquired with eight different TE values were used for generation of R₂ maps by nonlinear least-square pixel-by-pixel fitting to a single exponent (Eq. A):

where I is the measured signal intensity for each TE and A is the fitted steady-state signal intensity in fully relaxed images. R₂ values within a region of interest were averaged and SD was calculated.

R₁ relaxation rates for the in vitro experiments were derived by nonlinear single exponential fitting of images acquired at different TRs (Eq. B):

where I is the measured signal intensity for each TR. A, B, and R₁ were derived from optimization of the curve fitting; A is the steady-state signal intensity in fully relaxed images and B is ∼1 for steady-state saturation.

R₁pre maps for the in vivo experiments were derived by a nonlinear best fit to Eq. C:

where I is the signal intensity as a function of the pulse flip angle α; TR = 10 ms; the pre-exponent term, M₀, includes the spin density and the T₂ relaxation, which are assumed to be constant. A single R₁ value was derived for each voxel (31). Postcontrast R₁ values (R₁post) were calculated from precontrast and postcontrast three-dimensional gradient-echo signal intensities (Eq. D):

Hyaluronan-gadolinium-diethylenetriaminepentaacetic acid-beads as novel contrast material for detection of hyaluronidase by magnetic resonance imaging. Particles presenting local high concentration of hyaluronan, serving as MRI detectable substrate for hyaluronidase, were generated by cross-linking hyaluronan-GdDTPA to agarose beads (Fig. 1A). The product, hyaluronan-GdDTPA-beads, contained ∼1.8% (w/w) gadolinium as determined by ICP-MS. The specific R₁ relaxivity of the hyaluronan-GdDTPA-beads was measured to be 11 mmol/L⁻¹ s⁻¹ per gadolinium, and the R₂ relaxivity of the hyaluronan-GdDTPA-beads was measured to be 25 mmol/L⁻¹ s⁻¹ per gadolinium (Fig. 1B). To check whether chemical modification of hyaluronan can still be recognized and cleaved by hyaluronidase, we used a fluorescently tagged hyaluronan. Hyaluronan-Bodipy incubated with bovine testes hyaluronidase or double-distilled water was loaded on Sephadex G-25 columns and eluted with double-distilled water. Aliquots were collected from the two columns and fluorescence was measured at 538 nm (Fig. 2A). The undigested hyaluronan-Bodipy was eluted with the void volume, whereas the hyaluronan-Bodipy that was incubated with hyaluronidase showed delayed elution of the lower molecular weight degradation products. Hence, the modification of hyaluronan did not prevent the ability of hyaluronidase to recognize and cleave it.

Hyaluronan-gadolinium-diethylenetriaminepentaacetic acid-beads allow detection of hyaluronidase activity by magnetic resonance imaging. Suspension of hyaluronan-GdDTPA-beads in culture medium (plus 15% FCS) was used to test the ability of MRI to detect hyaluronidase-dependent changes in R₁ and R₂ relaxation rates of water (Fig. 2C and D). The R₁ measured for hyaluronan-GdDTPA-beads incubated with bovine testes hyaluronidase in serum-containing medium was 0.84 s⁻¹ (15 minutes at 37°C). This was significantly higher relative to control hyaluronan-GdDTPA-beads suspended in medium without hyaluronidase that had R₁ of 0.75 s⁻¹ (n = 6; two-tailed unpaired t test; P < 0.009; Fig. 2C). The R₂ measured for hyaluronan-GdDTPA-beads incubated with bovine testes hyaluronidase in serum-containing medium was 3.4 s⁻¹, this was significantly higher relative to control hyaluronan-GdDTPA-beads suspended in medium without hyaluronidase that had R₂ of 3.2 s⁻¹ (n = 6; two-tailed unpaired t test; P < 0.009; Fig. 2D). The rate of the enzymatic reaction was measured by injecting hyaluronidase into preheated tube (37°C) that contained hyaluronan-GdDTPA-beads and measuring the changes in R₂. The enzymatic reaction was quite rapid and was over within few minutes (Fig. 2B).

Activity of hyaluronidase secreted by human epithelial ovarian carcinoma cells. Particle exclusion assay was used to evaluate the biological activity of hyaluronidase secreted by two human epithelial ovarian carcinoma cell lines, OVCAR-3 and ES-2 (Fig. 3A-E). The chondrocytes used in this assay produce a thick pericellular coat composed of high molecular weight hyaluronan. RBC, which are added to the culture plate, are excluded by the hyaluronan layer, thereby delineating it and allowing its visualization (Fig. 3A). After addition of commercial hyaluronidase, the high molecular weight hyaluronan is degraded, allowing the RBC to approach the chondrocyte membrane (Fig. 3B). The activity of hyaluronidase in medium conditioned by OVCAR-3 and ES-2 cells was examined by addition of this medium to the chondrocytes (Fig. 3C and D, respectively). Conditioned medium from ES-2 cells but not from OVCAR-3 cells degraded the hyaluronan layer, showing a significantly higher activity of hyaluronidase that is secreted to the culture medium by ES-2 cells (n = 5; two-tailed unpaired t test; P = 0.02; Fig. 3E).

Expression and secretion of hyaluronidase-1 and hyaluronidase-2 by ovarian carcinoma cells. The expression of hyaluronidase-1 and hyaluronidase-2 was measured in the two human epithelial ovarian carcinoma cell lines, OVCAR-3 and ES-2, using reverse transcription-PCR (RT-PCR) analysis. No significant differences in the mRNA levels of hyaluronidase-1 and hyaluronidase-2 were detected between the two cell lines (Fig. 3F and G). The cells were also stained with anti-hyaluronidase-1 and anti-hyaluronidase-2 antibodies (Fig. 4). Those staining showed that the amount of intracellular hyaluronidase-1 and hyaluronidase-2 in OVCAR-3 and ES-2 cells was similar. In contrast to the similarity in hyaluronidase mRNA expression levels and the immunohistochemistry, a large difference was detected in the amount of hyaluronidase secreted by OVCAR-3 and ES-2 cells. The amount of hyaluronidase secreted to the conditioned medium of the cells was analyzed by Western blot using anti-hyaluronidase-1 and anti-hyaluronidase-2 antibodies. In accord with the higher activity of hyaluronidase (Fig. 3A), medium conditioned by ES-2 cells contained high levels of both hyaluronidase-l and hyaluronidase-2, whereas in OVCAR-3 conditioned medium both proteins were undetectable (Fig. 3H). Although ES-2 cells exhibited higher rate of proliferation in 10% FCS, they showed decreased cell survival in serum-free medium relative to OVCAR-3 cells. Thus, conditioned medium was collected in most experiments 24 hours after serum removal, when both cell lines were viable. Traces of hyaluronidase-1 and hyaluronidase-2 could be detected in medium from OVCAR-3 cells by using double the amount of protein (data not shown).

Magnetic resonance imaging detection of hyaluronidase secreted by ovarian carcinoma cells. ES-2 and OVCAR-3 cells were grown to confluence in a 96-well plate for 24 hours. Hyaluronan-GdDTPA-beads (1 mg/mL) were added to the cells without changing the culture medium (DMEM + 10% FCS) such that the hyaluronidase secreted to the medium was not washed out.

Hyaluronidase expressed by tumor cells induced an elevation in R₁ and R₂ (Fig. 5). R₁ of fresh medium was 0.45 ± 0.04 s⁻¹, R₁ of OVCAR-3 medium was 0.47 ± 0.03 s⁻¹, and R₁ of ES-2 medium was 0.51 ± 0.04 s⁻¹. R₂ of fresh medium was 2.42 ± 0.06 s⁻¹, R₂ of OVCAR-3 medium was 2.48 ± 0.03 s⁻¹, and R₂ of ES-2 medium was 2.77 ± 0.01 s⁻¹. Therefore, ES-2 cells that showed high secretion of biologically active hyaluronidase in the particle exclusion assay (Fig. 3) also showed the higher increase in R₁ and R₂ relaxation rates (Fig. 5). We found significant difference in R₁ and R₂ between ES-2 and OVCAR-3 (n = 3; two-tailed unpaired t test; P = 0.005) and between ES-2 and the negative control (n = 5; two-tailed unpaired t test; P = 0.0002).

The change in relaxation rate of hyaluronan-GdDTPA-beads could be induced by hyaluronidase that was secreted to the medium and reacted with the contrast material to degrade it. Alternatively, the change could be attributed to hyaluronidase that was anchored to the cells surfaces (32). To differentiate between these two options, we repeated the experiment with the medium alone. Conditioned medium from ES-2 or OVCAR-3 cells was depleted of residual cells by centrifugation and hyaluronan-GdDTPA-beads were added. This approach resulted in R₂ relaxation changes that were similar to those measured in the presence of cells. Hyaluronan-GdDTPA-beads suspended in conditioned medium of ES-2 cells showed larger increase in R₂ relative to that observed for OVCAR-3 cells (n = 6; two-tailed unpaired t test; P < 0.0001). Thus, the R₂ effect detected by MRI was consistent with the different levels of biologically active hyaluronidase secreted by these two human ovarian carcinoma cell lines.

In vivo detection of hyaluronidase by magnetic resonance imaging in ovarian carcinoma tumors. To examine the feasibility of the new contrast material in vivo, we administered it s.c. through a catheter to the vicinity of s.c. ES-2 tumors in immunodeficient nude mice and compared the activation of the contrast material with that observed in the absence of tumor (Fig. 6). Regions of interest were selected in the area of contrast material enhancement. The most significant changes in R₁ and R₂ were observed at early time points between the time of injection and the next scan (after 2.5 minutes). The rapid dynamics were consistent with rapid activation of the contrast material detected in vitro (Fig. 2B).

Longitudinal R₁ relaxation maps revealed significant changes associated with activation of hyaluronan-GdDTPA-beads in the vicinity of the ES-2 tumors [high ΔR₁; ΔR₁ = R₁(t = 2.5 minutes) − R₁(t = 0); four tumor-bearing mice and two control mice; P = 0.01 relative to control animal and 0.039 relative to area with contrast material near the tumor; two-tailed unpaired t test; Fig. 6E]. Activation of the contrast material was evident but smaller for R₂ relaxation [high ΔR₂; ΔR₂ = R₂(t = 2.5 minutes) − R₂(t = 0); P = 0.06 relative to control area in the tumor-bearing mice and 0.078 relative to control mice without tumor; two-tailed unpaired t test; n = 4 in each experimental group].

Molecular imaging aimed at detection of specific enzymatic reactions by MRI is a novel and exciting approach for noninvasive characterization of processes in living organisms (33). Previous agents developed for MRI analysis of enzymatic activity include substrates for β-galactosidase (34), proteases (33, 35), and iron-binding proteins, such as tyrosinase (36), the transferrin receptor (37, 38), and ferritin (39). Some of these agents exploit the change in relaxivity that occurs on enzymatic activity, affecting the interaction of water molecules with the contrast material. The large number of mechanisms influencing magnetic relaxivity offers many possibilities for targeting different enzymatic reactions. The aim of the present work was to develop contrast material for detection of hyaluronidase, a key enzyme that alters the angiogenic balance of the extracellular matrix by mediating breakdown of antiangiogenic high molecular weight hyaluronan into its proangiogenic low molecular weight products.

We report here a single step assay for detection of hyaluronidase, which allows noninvasive in vivo imaging by MRI. Using beads enveloped with hyaluronan-GdDTPA, we were able to detect changes in R₁ and R₂ relaxation rates after addition of either hyaluronidase or conditioned medium taken from human ovarian carcinoma cell lines that express and secrete hyaluronidase. In the presence of hyaluronan-GdDTPA-beads, the relaxation rates increased significantly after interaction with hyaluronidase. The observed changes in relaxation rate can be due to degradation of hyaluronan-GdDTPA but can also be due to binding of hyaluronidase to the contrast material. Hypothetically, degradation of hyaluronan can cause a release of GdDTPA or low molecular weight hyaluronan-GdDTPA, which might allow increased interaction with water and affect MRI contrast. Similarly, binding of hyaluronidase to hyaluronan could change the conformation of hyaluronan and expose GdDTPA to water possibly by interfering with a direct interaction of the negatively charged carboxyl groups of hyaluronan with the free coordination sites of gadolinium.

The ability to detect physiologic levels of hyaluronidase by MRI using hyaluronan-GdDTPA-beads was tested in vitro using human epithelial ovarian carcinoma cells. The expression of hyaluronidase-1 and hyaluronidase-2 was evaluated in OVCAR-3 and ES-2 human ovarian carcinoma cell lines. Although no significant differences were found in the level of mRNA in the two cell lines, major difference was detected in the protein level of hyaluronidase-1 and hyaluronidase-2 secreted to the culture medium. ES-2 cells secreted high level of both hyaluronidases to the medium; on the contrary, both hyaluronidase-1 and hyaluronidase-2 were undetectable in OVCAR-3 cells medium.

The biological activity of hyaluronidase, as manifested in particle exclusion assay by degradation of the hyaluronan coat of chondrocytes, revealed that ES-2 cells not only secrete more hyaluronidase than OVCAR-3 cells but that the secreted hyaluronidase is also active. Remarkably, the new MRI contrast material reported here, hyaluronan-GdDTPA-beads, allowed detection of hyaluronidase when added to solution and also when secreted by the human ovarian carcinoma cells. Moreover, the contrast material was sensitive to the different levels of hyaluronidase between the two ovarian cell lines. The measured changes in R₁ and R₂ were consistent with the Western blot and particle exclusion assay, all of which showed higher levels of hyaluronidase in the medium of ES-2 cells. The R₁ and R₂ changes induced by bovine testes hyaluronidase were larger than those induced by conditioned medium of the ovarian carcinoma cells, reflecting probably the concentration of active enzyme, the type of enzyme (hyaluronidase-1 and hyaluronidase-2 relative to testis hyaluronidase, respectively), as well as other modulators of hyaluronidase activity that are secreted by the cells.

The ability to detect hyaluronidase activity in vivo was shown on s.c. ES-2 ovarian carcinoma tumors after s.c. administration of the contrast material in the tumor region. Due to rapid clearance of the contrast material after s.c. administration, activation of the contrast material was analyzed from the early increase in the relaxation rate measured from the first two time points after administration of the contrast material. A rapid increase in R₁ and R₂ relaxivities was detected in the periphery of the hyaluronidase-secreting tumor. The increased relaxivity induced by hyaluronidase could aid in detection of regions of enzyme activity. In the control regions and control mice, ΔR₁ and ΔR₂ were negative (i.e., there was a decrease in relaxation rate, consistent with the lack of activation and the rapid clearance of the s.c. administered contrast material, in the absence of a tumor).

This approach opens new possibilities for the study of ovarian carcinoma-induced angiogenesis. We have reported previously that human epithelial ovarian cancer cells show gonadotropin regulation of both vascular endothelial growth factor (VEGF; ref. 40) and CD44, the cell surface receptor for hyaluronan (41). Thus, elevated level of luteinizing hormone and follicle-stimulating hormone as occurs at menopause can lead to increased adhesion of ovarian cancer cells on hyaluronan-coated surfaces in the peritoneum as well as augmented expression of VEGF. However, elevated expression of VEGF is insufficient for inducing angiogenesis in the presence of high molecular weight hyaluronan. This was shown previously in the preovulatory ovarian follicle, where we showed that hyaluronan serves as an antiangiogenic shield, protecting the oocyte from premature vascularization in the presence of elevated levels of VEGF (42). Secretion of hyaluronidase by ovarian carcinoma cells can allow micrometastases to overcome this barrier and induce vascularization.

This study provides in vitro proof-of-principle and in vivo demonstration for the ability to detect hyaluronidase activity by MRI. For ovarian cancer, i.p. administration would allow the contrast material to adhere to the external surface of small tumor nodules and be activated by secreted hyaluronidase, enabling detection of hyaluronidase on the cell surface and the hyaluronidase secreted to the pleural effusion. Due to the particular dissemination pattern of ovarian cancer, i.p. administration of targeted chemotherapy is being evaluated as a route that would increase tumor targeting and reduce systemic toxicity (43). The contrast material reported here would highlight tumors that have the capacity to adhere to the peritoneal wall and activate angiogenesis by overcoming the antiangiogenic effect of high molecular weight hyaluronan. Because tumor-derived hyaluronidase has been implicated in tumor invasion across the mesothelial cell layer, a noninvasive imaging of this enzyme at sites of tumor implantation may aid in defining the risk for spread beyond the peritoneal cavity.

Clearly, development of this approach for other in vivo applications and particularly for i.v. administration would require significant modification of the carrier beads to make them more compatible for administration. The size of the hyaluronan-GdDTPA complex and the rapid degradation of hyaluronan in circulation by plasma hyaluronidase-1 (7) may limit i.v. administration.

In summary, we show here a novel contrast material for noninvasive MRI detection of hyaluronidase. Hyaluronan-GdDTPA-beads can help decipher the role of hyaluronan and its breakdown products in regulation of angiogenesis during normal development and cancer progression. Specifically, we anticipate that this approach will facilitate visualization and thereby the study of the role of hyaluronidase in peritoneal metastatic dissemination of ovarian carcinoma.

Grant support: National Cancer Institute grant RO1 CA75334 and Mark Family Foundation.

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.

We thank Profs. Lia Addadi and Benjamin Geiger (The Weizmann Institute of Science) for helpful discussions, Prof. Mala Mahendroo (Department of Obstetrics and Gynecology, University of Texas Southwestern Medical Center, Dallas, TX) for the primer sequences for human hyaluronidase-1, and Prof. Steffen Hauptmann for the human ovarian carcinoma cell lines ES-2 and OVCAR-3.