|Home | About | Journals | Submit | Contact Us | Français|
Reporter genes and associated enzyme activity are becoming increasingly significant for research in vivo. The lacZ gene and β-galactosidase (β-gal) expression have long been exploited as reporters of biological manipulation at the molecular level and a non-invasive detection strategy based on proton MRI is particularly attractive. 3,4-Cyclohexenoesculetin β-D-galactopyranoside (S-Gal®) is a commercial histological stain, which forms a black precipitate in the presence of β-gal and ferric ions suggesting potential detectability by MRI. Generation of the precipitate is now shown to cause strong T2* relaxation revealing β-gal activity. A series of tests with the enzyme in vitro and with tumor cells show that this approach can be used as an assay for β-gal activity. Proof of principle is shown in human breast tumor xenografts in mice. Upon direct injection of a mixture of S-Gal® and ferric ammonium citrate intense contrast was observed immediately in MCF7-lacZ tumors, but not in wild type tumors. S-Gal® activation in combination with ferric ions introduces a novel approach for assaying enzyme activity by MRI in vivo.
Reporter genes are routinely used to reveal genetic manipulations and have become a mainstay of molecular biology. More recently, they have been applied to in vivo investigations and fluorescent proteins and bioluminescence based on luciferase are effective for small animal investigations. However, the bacterial lacZ gene has been the most popular reporter with applications ranging from immunosorbent assays to in situ hybridizations, and evaluation of gene distribution. Indeed, lacZ has been used in clinical trials revealing regions of tissue transfection in biopsy specimens based on histological staining. As such many colorimetric stains and assays have been developed and are in routine use including reagents such as ONPG (nitrophenyl-β-D-galactopyranoside) which generates a yellow coloration, X-gal (4-chloro-3-bromoindole-galactose) which generates a blue stain, and S-Gal® (3,4-cyclohexenoesculetin β-D-galactopyranoside) which generates a black stain.
In vivo detection could be very valuable and several recent studies have reported novel substrates or novel applications of substrates allowing detection of β-galactosidase. β-gal shows broad substrate specificity and recent in vivo applications have exploited fluorescence of DDAOG (7-hydroxy-9H-(1,3-dichloro-9,9-dimethylacridin-2-one)) (1), bioluminescence of Lugal (6-o-β-galactopyranosyl-luciferin) (2), chemiluminescence of Galacton-Star Plus (3), photoacoustic tomography of X-gal (4), SPECT of 5-[I-125]iodoindol-3-yl-β-D-galactopyranoside ([I-125]IBDG) (5), and PET of 2-(4-[125I/123I]iodophenyl)ethyl-1-thio-β-D-galactopyranoside, 3-(2'-[F-18]fluoroethoxy)-2-nitrophenyl-β-D-galactopyranoside or 3-[C-11]methoxy-2-nitrophenyl β-D-galactopyranoside (6,7).
In terms of NMR Meade et al. (8) proposed a galactose capped gadolinium ligand (EgadMe) in 2000, which showed an increased relaxivity upon exposure to β-gal. Following direct intracellular injection into oocytes it could elegantly reveal cell lineage in developing tadpoles. Unfortunately, the requirement for direct intracellular injection precluded use in tumors. These studies prompted us to seek alternate NMR reporters.
We have presented a variety of substrates based on isomers and analogs of 4-fluoro-2-nitrophenyl-β-D-galactopyranoside (9–11), which exhibit 19F NMR chemical shift change due to β-gal activity. We have shown the ability to differentiate wild type (WT) and stably transfected lacZ expressing breast and prostate tumor xenografts implanted in mice using 19F NMR (12,13). Detection would ideally use systemic delivery of the reporter molecules, but to date we have achieved measurements based on direct intra tumoral injection. Furthermore, the signal to noise has generally limited investigations to spectroscopy, although we were able to achieve images of β-gal activity in phantoms comprising enzyme in solution or transfected cells (14,15). A proton NMR technique would be expected to provide greater signal to noise ratio.
We believe one of the most fruitful resources for developing novel in vivo imaging agents is the body of knowledge associated with traditional histology and pathology. Review of the histology literature suggested S-Gal® as a possible substrate for 1H MRI (16). Upon cleavage by β-galactosidase the aglycone of S-Gal chelates ferric ions (Fe3+) to produce an intense black stain, which is not only visible, but also paramagnetic (Figure 1). We now report investigations to characterize the development of 1H MRI contrast based on S-Gal as a substrate for β-gal. We also demonstrate the ability to identify WT versus stably expressing lacZ tumors based on direct intra tumoral injection.
S-Gal® (3,4-cyclohexenoesculetin-β-D-galactopyranoside, Figure 1), S-Gal sodium salt and ferric ammonium citrate (FAC) were purchased from Sigma-Aldrich (St. Louis, MO). We used the sodium salt throughout the studies, since it is far more water soluble and designate it as S-Gal. MR images were obtained using a Varian unity INOVA 4.7 T horizontal bore MR system equipped with actively shielded gradients (22 G/cm 12 cm bore) or a 9.4 T vertical bore spectrometer equipped with a micro imaging system. All animal experiments were approved by the UT Southwestern Institutional Animal Use and Care Committee.
S-Gal (300 µg/ml) and FAC (500 µg/ml) were blended with LB (luria broth) agar medium. E. coli and E. coli induced by IPTG (isopropyl β-D-1-thiogalactopyranoside; 30 µg/ml) to express β-galactosidase were cultured in 20 mm NMR tubes overnight and distilled water was added on top of the E. coli before imaging at 9.4 T (TR=1 s, TE=20 ms, FOV=20 mm × 20 mm, with matrix=512×512 and slice thickness 150 µm).
various concentrations of FAC (0–15 mM), sodium S-Gal (0–30 mM) and 0.4–1.25 units β-gal (Sigma G6008 from E. coli, 504 units/mg protein) were mixed in buffer at pH 7.2. After 2 minutes, 50 µl 1% agarose was added and cooled to gel at room temperature. T2-weighted images were acquired with TR=2 s and various TE (12 –100 ms) and T1-weighted images with TE=15 ms and various TR (300 – 5,000 ms), all with a 30×30 mm field of view (128×128 in plane resolution) and 2 mm slice thickness, using a spin echo sequence.
human breast cancer MCF7 cells (ATCC, Manassas, VA) were transfected to generate a stable β-galactosidase expressing MCF7-lacZ subline by introducing the E. coli lacZ gene, as described previously (13). The cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum with 100 units/ml of penicillin, 100 units/ml streptomycin, and cultured in a humidified 5% CO2 incubator at 37 °C. To ensure continued selectivity 200 µg/ml of G418 antibiotic was added to the DMEM medium for MCF7-lacZ cell growth.
For the MRI study, the cells were grown to about 80% confluence, trypsinized, and washed with PBS. A mixture of 1×106 MCF7 or MCF7-lacZ cells in PBS with S-Gal (150 µg yielding 9 mM) and/or FAC (250 µg yielding 2.4 mM) was placed in a 10 mm NMR tube as an interlayer between 1% low gelling temperature agarose. T2* values were measured after 30 hr at 37 °C with various TE values from 3 ms to 40 ms using a gradient echo sequence (TR=1 s, flip angle=20°). In subsequent tests varying numbers of MCF7-lacZ cells (0–1.5×106) were used and mixed with agarose to evaluate the kinetics of contrast development. To evaluate the rate of contrast development 5×106 MCF7-lacZ cells were placed on agarose and incubated with FAC (8.5 mM) and S-Gal (32 mM) at 37 °C. MRI was performed at 4.7 T with a 2.5 cm quadrature birdcage coil. A multi-slice gradient echo sequence was used for measurements (TR/TE=1000/30 ms, in plane resolution 80×60 µm, slice thickness = 2 mm) and T1, T2 and T2* values were measured after various incubation times ranging from 0 to 24 hrs at 37 °C.
MCF7 and MCF7-lacZ cells were respectively mixed with 10% Matrigel (BD Company, Franklin Lakes, NJ) and injected in the flanks of female SCID mice (average body weight 20 g). When the tumor size reached approximately 500 mm3 they were used for MRI. Mice were anesthetized with 1% isoflurane in air for MRI. Animal body temperature was maintained using a pad with circulating warm water. Anesthetized animals were positioned in a home-built 20-mm diameter RF volume coil and multi slice T2-weighted images (TR =1000 ms; TE =40 ms; field of view of 40 mm × 50 mm; slice thickness 2 mm) were acquired at 4.7 T. Five slices were acquired within an experimental time of 129 s. The mouse was removed from the magnet and FAC (0.5 mg) and S-Gal (1 mg) in 100 µl saline were administered separately by direct injection into each tumor. The mouse was repositioned carefully and MRI repeated over a period of 60 mins. Following MRI, the mouse organs and tumors were harvested and western blot and β-gal activity assay were performed on extracted protein.
The extracted protein of tumors was quantified by a protein assay (Bio-Rad, Hercules, CA) based on the Bradford method (17). Protein (30 µg) was added to each well, separated by 10% SDS-PAGE (Nu-PAGE) and transferred to membrane. Primary monoclonal anti-β-gal antibody (Promega, Madison, WI) and anti-actin antibody (Sigma, St. Louis, MO) were used at a dilution of 1:5000, followed by reactive protein detection using a horseradish peroxidase –conjugated secondary antibody and ECL detection (GE Healthcare, NJ).
The expression of β-gal was measured using the colorimetric β-gal assay kit (Promega) based on ONPG. The extracted protein was quantified by a protein assay (Bio-Rad) based on the Bradford method. β-Gal activity was expressed as units per mg protein.
The excised tumors were embedded in Tissue-Tek OCT (Miles Laboratory, Elkhart, IN) and frozen in liquid nitrogen. Cryostat sections were collected on gelatin-coated glass slides, and 10 µm sections stained with Nuclear fast red (Sigma) and X-gal for β–gal activity analyzed.
When β-gal expressing E. coli were grown on the surface of agar containing S-Gal and FAC a black precipitate was seen visually (Figure 2a) and signal loss was observed as contrast in T2*-weighted MRI (arrow, Figure 2b). No color or contrast was seen for control E. coli (Figure 2c&d). In gel phantoms R2-weighted contrast increased with FAC concentration (Figure 3a). R1 (=1/T1) and R2 (=1/T2) were essentially invariant with β-gal activity and S-Gal concentration in agar in the absence of FAC (data not shown). The presence of 1 mM FAC generated significant (p<0.01) increase in both R1 (Figure 3b) and R2 (Figure 3c), but the relaxation tended to plateau once the concentration of FAC exceeded 5 mM. At the highest concentrations of β-gal (1.25 units) R2 did increase to 15 mM (Figure 3c). Both R1 and R2 were found to be sensitive to the amount of β-gal activity. While ΔR2 was greater at higher β-gal activity reflecting additional precipitate, ΔR1 was found to be higher at lower β-gal activity. Changes in R1 with increasing S-Gal in the presence of β-gal were essentially independent of enzyme concentration (Figure 3d), whereas R2 showed a strong dependence (Figure 3e).
When S-Gal and FAC were added to the MCF7-lacZ cells, the signal loss (R2* contrast) was detected as a black interlayer in the agarose gel (Figure 4a). Contrast is caused by the formation of the black ferric complex and requires both S-Gal and FAC, since absence of either did not show R2* signal loss. The R2* signal loss increased with MCF7-lacZ cell number (Figure 4b) ranging from 7.2 s−1 in absence of cells to a plateau of about 30 s−1 for >300, 000 MCF7-lacZ cells after 30 hrs exposure to S-Gal and FAC. Maximum contrast was not seen immediately and it was found to develop over a period of hours for cells in culture. Both R2 and R2* contrast increased with the exposure time though neither showed appreciable change beyond 6 hrs.
Mice bearing MCF7 or MCF7-lacZ tumors were imaged at 4.7 T. Very strong R2 contrast was detected in the lacZ expressing tumors within 5 minutes of administering the contrast agent cocktail directly into the tumor (0.5 mg FAC and 1 mg S-gal in 100 µl solution, Figure 5). No R2 contrast was observed in the contra lateral wild type tumor, which was injected with the same contrast agents. Prior to administration of S-Gal the SNR for WT tumor was 31.8 and that for the lacZ tumor was 33.6. Following contrast agent administration the WT tumor remained essentially unchanged (SNR=28), while there was a signal loss in the lacZ tumor (SNR = 8.2) within 5 mins. Thus, contrast between WT and lacZ tumors changed from essentially unity to 3.5 accompanying the 75% loss of signal in the lacZ tumor. Similar results were seen in three pairs of tumors. In a second example presented in Figure 5 (c–e), the intense contrast observed 2 mins after injection declined over the next 60 mins. After MRI, the animals were sacrificed and tumors excised for histology, western blot and β-gal activity assay. Each technique confirmed that β-gal activity was much higher in MCF7-lacZ tumor than WT tumor (Figure 5f–j).
We have demonstrated the ability to detect β-gal activity in vivo based on contrast in 1H MRI R2-weighted images following administration of S-Gal and FAC. In vitro tests showed sensitivity of R1, R2 and R2 * to β-gal required the presence of S-Gal and ferric ions. Investigations with stably expressing MCF7-lacZ cells showed that the development of contrast is related to the number of cells and the exposure time. In vivo tissue expression of β-gal could be identified in MCF7-lacZ tumors compared with WT tumors based on direct intra tumoral injection of FAC plus S-Gal. Contrast developed immediately and β-gal activity was confirmed by ex vivo analysis including histology and western blot.
We first reported 1H MRI contrast based on the β-gal substrate S-Gal in 2004 (18). Initial experiments suggested that IP administration of S-Gal and ferric ammonium citrate could generate effective contrast in the tumor. However, multiple tests in mice with lacZ expressing breast, prostate and brain tumors failed to reproduce the contrast and we switched to direct tissue injection of S-Gal and ferric ammonium citrate, which provides robust and reproducible results.
Reporter assays for lacZ gene expression in vivo have become a popular area of research recently. Optical approaches have been demonstrated based on systemic administration of reporter molecules, e.g., Tung et al. (1) could identify β-gal expressing 9L-lacZ versus WT tumors growing as xenografts in the mammary fat pad of athymic nude mice based on a 50 nm red shift in fluorescent emission accompanying cleavage of DDAOG. We have observed light emission from MCF7-lacZ tumors growing in nude mice and throughout the body of black furry 129S-Gt(ROSA)26Sor/J mice following IV administration of the chemiluminescent substrate Galacton Star Plus (3).
For 1H MRI Chang et al. (19) reported differential β-gal induced signal change in CT26 wild type and lacZ expressing colon carcinoma xenografts in mice following IV administration of a galactose capped Gd-DOTA complex. However, a two fold difference in baseline signal intensity between the tumors made it difficult to rigorously evaluate the changes. In the studies reported here, baseline signal was quite similar in WT and lacZ tumors. Further developments were reported by Chauvin et al. (20), who presented a new β-gal activated PARACEST 1H MRI reporter molecule. Hanaoka et al. (21) have demonstrated a RIME (receptor-induced magnetization enhancement) approach whereby β-gal releases a Gd-containing moiety, which binds to human serum albumin generating enhanced relaxivity. A PARACEST approach to revealing β-gal activity was just reported based on [Yb(dota-αBz-βGal)]− in solution (20). To date each of these methods has been presented in vitro only.
Systemic administration of reporters is our long-term goal, but most recent in vivo imaging of β-gal has required direct injection of reporter substrate into tissue of interest, e.g., tandem bioluminescence, photo acoustic spectroscopy, 19F NMR and nuclear medicine techniques (2,4–7). Direct injection avoids the complications of pharmacokinectics whereby the reporter molecule must reach the tissue of interest and then be present in sufficient concentration for evaluation. Systemic delivery is also subject to toxicity constraints of the substrate. We have been able to inject S-Gal IP or IV without obvious acute toxicity, but FAC appeared severely toxic at 50 mg/ml IV (unpublished data). We have now administered a mixture of FAC (20 mM, 0.5 mg) and sodium salt of S-Gal (20 mM, 1 mg) intra tumorally to about 10 tumor bearing mice with no obvious toxicity during the time prior to sacrifice (up to 4 days).
Interestingly, in presence of low enzyme concentrations, a larger ΔR1 and a lower ΔR2 was observed as the concentration of FAC was increased. We attribute this to formation of soluble paramagnetic macromolecular complexes of Fe3+ with the cleaved aglycone of S-gal that are more efficient in relaxing the surrounding water molecules than are Fe3+ ions. At higher enzyme concentrations, more complex is formed in the same incubation time, but it is more likely to precipitate into superparamagnetic nanoscale particles that have a stronger effect on R2. In vivo the contrast is seen immediately following direct intra tumoral injection, but it often cleared over the following hours (Figure 5 c–e). While the black precipitate is initially insoluble we note that it is soluble in ethanol.
Developing new reporter assays must compete with alternative technologies. Not only diverse substrates and modalities to detect lacZ, as outlined in the introduction, but also alternative reporter genes. Thus, both florescent proteins and bioluminescent imaging are particularly effective in small animals (22,23). Thymidine kinase and hNIS (human iodide symporter) have found success in larger animals and have been proposed for patients (24,25)). MRI contrast has been shown with respect to transgene expression of transferrin (26), ferritin (27), and MagA (28). Enhanced CEST accompanying expression of specific amino acids rich in exchangeable protons (e.g., polylysine) has been demonstrated in xenografts of LRP-expressing 9-L cells in the mouse brain revealing transgene activity (29). Meanwhile creatine kinase as a transgene was demonstrated using 31P NMR of liver 20 years ago (30) and 19F NMR effectively shows cytosine deaminase activity in the conversion of 5-fluorocytosine to 5-fluorouracil (31).
We believe that a repertoire of methods is particularly valuable and that 1H MRI contrast based on S-Gal represents a potentially viable approach. Beyond S-Gal as a reporter for lacZ and β-gal activity, we note the commercial availability of specific substrates for other enzymes such as glucose and glucuronidase, which may offer diverse scope for imaging. We have demonstrated the ability to identify lacZ tumors following intra tumoral injection. We note that Bengtsson et al. (32) recently used a rather similar approach based on S-Gal to pre-label bone marrow cells from ROSA mice and track them following implantation in mice. They reported superior contrast at higher field including 17.6 T, while we have undertaken studies at both 4.7 and 9.4 T
Our data show feasibility of S-Gal® as a proton MRI reporter for β-galactosidase at high field (9.4 T) and more importantly at 4.7 T in cultured cells and tumors growing in mice. S-Gal® is commercially available and readily enters cells. We believe this holds great promise as a novel MRI approach for imaging gene activity and detecting gene function. We have demonstrated contrast following direct injection into tissues. We are also investigating analogues of S-Gal with potential enhanced properties and the development of agents for systemic delivery will be crucial for widespread utility.
This work was supported in part by NCI R21 CA120774 in conjunction with the Small Animal Imaging Research Program (SAIRP NCI U24 CA126608). NMR experiments were conducted at the Advanced Imaging Research Center under NIH BTRP #P41-RR02584. We are grateful to Dr. Todd Soesbe for constructing the birdcage coil, Drs. Zhenyi Ma and Ammar Adam for technical assistance, and Dr. Jian-Xin Yu for stimulating discussions.