Novel S-Gal^® analogs as ¹H MRI reporters for in vivo detection of β-galactosidase

Abstract

The quantitative assessment of gene expression and related enzyme activity in vivo could be important for the characterization of gene altering diseases and therapy. The development of imaging techniques, based on specific reporter molecules may enable routine non-invasive assessment of enzyme activity and gene expression in vivo. We recently reported the use of commercially available S-Gal^® as a β-galactosidase reporter for ¹HMRI, and the synthesis of several S-Gal^® analogs with enhanced response to β-galactosidase activity. We have now compared these analogs in vitro and have identified the optimal analog, C3-GD, based on strong T₁ and T₂ response to enzyme presence (ΔR₁ and ΔR₂ ∼ 1.8 times S-Gal^®). Moreover, application is demonstrated in vivo in human breast tumor xenografts. MRI studies in MCF7-lacZ tumors implanted subcutaneously in athymic nude mice (n = 6), showed significant reduction in T₁ and T₂ values (each ∼ 13%) 2 h after intratumoral injection of C3-GD, whereas the MCF7 (wild type) tumors showed slight increase. Thus, C3-GD successfully detects β-galactosidase activity in vivo and shows promise as a lacZ gene ¹HMR reporter molecule.

1. Introduction

One of the major challenges in the field of gene therapy has been the lack of methods to quantitatively assess the success of gene transfection and the longevity of gene expression [1]. Reporter genes are typically used to assess regulation and transfection of genes in biological pathways. Popular reporter genes are those that generate β-galactosidase (β-gal), β-glucuronidase, firefly luciferase, fluorescent proteins (e.g., green fluorescent protein), transferrin and ferritin [1–3]. Historically, the lacZ gene encoding β-gal has been the most extensively used reporter with applications ranging from basic science to translational and clinical trials for assaying protein expression and interactions [4,5].

There are various commercially available reagents (substrates) in regular use to detect β-gal expression in vitro by colorimetric stains and assays, such as nitrophenyl-β-d-galactopyranoside (generates yellow color) [6], 4-chloro-3-bromoindole-galactose (X-Gal^®, generates a blue stain) [7] and 3,4-cyclohexenoesculetin β-d-galactopyranoside (S-Gal^® - generates a black stain) [8]. The β-gal expression is also associated with cellular senescence [9] and X-Gal^® has been used to assay senescence-associated β-gal in vitro and ex vivo [10].

In recent years, development of novel reporter molecules and techniques for non-invasive in vivo detection of β-gal activity, as a marker of lacZ transgene expression, has been reported including optical [11–13], photoacoustic [14], radionuclide [15,16] and MR imaging (¹H [17–19] and/or ¹⁹ F [19–24]) methods. In a recent study, we demonstrated the ability to detect β-gal activity in lacZ-transfected breast cancer cells (MCF7) in vitro and in vivo using the commercially available S-Gal^® by ¹H MRI [25]. We also demonstrated the synthesis of several S-Gal^® analogs for improved response to β-gal activity [26]. In both cases, β-gal cleaves the substrate molecule at the C-O bond between β-d-galactopyranoside and aglycone, to release the aglycone, which forms a paramagnetic chelate in the presence of Fe³⁺ generating T1/T2/T₂^*-weighted ¹H MRI contrast. Fig. 1 illustrates the mechanism for ¹H-MRI detection of β-gal activity, using C3-GD as a model reporter. In the present study, we have compared the S-Gal^® analogs in vitro and shown application of the lead agent for detection of β-galactosidase activity in human breast tumor xenografts.

Fig. 1.

The illustration of mechanism for ¹H-MRI detection of β-gal activity, using C3-GD as a model reporter.

2. Materials and methods

2.1. General

3,4-Cyclohexenoesculetin β-d-galactopyranoside (S-Gal^® sodium salt), and ferric ammonium citrate (FAC) were purchased from Sigma-Aldrich (St. Louis, MO). Monogalactopryanosides 7-O-(β-d-galactopyranosyl)-8-hydroxy-3, 4-cyclohexenocoumarin (C1-MGD), 7-O-(β-d-galactopyranosyl)-8-hydroxy-4-methylcoumarin (C2-MGD), 7-O-(β-d-galactopyranosyl)-6-hydroxy-4-methylcoumarin (C3-MGD), 7-O-(β-d-galactopyranosyl)-8-hydroxy-6-methoxycoumarin (C4-MGD) and digalactopryanosides 7,8-di-O-(β-d-galactopyranosyl)-3, 4-cyclohexenocoumarin (C1-GD), 7,8-di-O-(β-d-galactopyranosyl)-4-methylcoumarin (C2-GD), 6,7-di-O-(β-d-galactopyranosyl)-4-methylcoumarin (C3-GD), 7,8-di-O-(β-d-galactopyranosyl)-6-methoxycoumarin (C4-GD) (Fig. 2) were synthesized by us as described previously [26]. All MR studies were performed on a Varian INOVA 4.7 T horizontal-bore MR system (200 MHz for ¹H) equipped with actively shielded gradients. T₁ mapping was performed using a spin echo sequence with TE = 12 ms and varying TR (0.2–6 s), while T₂ mapping was performed using a spin echo sequence with varying TE (12–200 ms) and TR = 6 s.

Fig. 2.

The structures of mono- and di-galactoside analogs of S-gal^®.

2.2. Detection of β-gal activity in vitro

For in vitro measurements, each reporter molecule solution (15 mM in de-ionized water, 40 μL) was mixed with FAC (5 mM, 40 μL) in 2% agar solution with or without β-gal (5 units, E801A, Sigma-Aldrich, St. Louis, MO) and placed in a 4 × 4 cut section of a 96 well plate. The prepared phantom was placed on a heating block maintained at 37 °C for 30 mins for the enzymatic reaction to stabilize. T₁ and T₂ maps of the phantom were acquired using a 3 cm solenoid coil (home-built) at 37 °C. MRI parameters: FOV 40 mm × 40 mm, matrix size 128 × 128, slice thickness 1 mm.

2.3. In vivo detection of β-gal activity in MCF7 and MCF7-lacZ tumors

All in vivo studies were performed with approval from the UT Southwestern Medical Center Animal Care and Use Committee. MCF7 (wild-type 1.5 × 10⁶) and MCF7-lacZ (2 × 10⁶) cells were implanted subcutaneously in the thighs of athymic nude mice (N = 6) on contralateral sides. MRI studies were performed when the tumor size reached 0.75 cm³. Mice were anesthetized with 1.5% isoflurane in air and placed in a Litzcage volume coil (Doty Scientific Inc, Columbia, SC) for MRI. Animal body temperature was maintained at 37 °C using a circulating warm water pad. Baseline (pre-injection) T₁ and T₂ maps of tumor containing slices were acquired using MRI parameters: FOV 100 mm × 50 mm, matrix size 256 × 128, slice thickness 1 mm, 13 slices. Following baseline imaging, the mouse was removed from the scanner and 25 μL of a solution containing 15 mM C3-GD and 5 mM FAC in water was injected intra-tumorally in a fan pattern in a sagittal plane using a fine 32G Hamilton^® needle into each tumor. After repositioning the mouse carefully, T₁ and T₂ maps were acquired 1 and 2 hours post injection. Multiple slices were acquired and those with injected contrast agent were identified based on the T₂- and T₁-weighted images (to delineate tumor boundary and locate the injected Fe³⁺ ions, respectively) for analysis. Tumor voxels from the injected slice of all animals were pooled with respect to imaging time point (baseline, 1 h and 2 h post injection) and statistical analysis with one-way ANOVA followed by Bonferroni's multiple comparison test was performed using GraphPad Prism (GraphPad Software Inc., La Jolla, CA).

2.4. Histology

Following MRI, the tumors were excised and fixed in 4% formalin. These tumor sections were embedded in paraffin and then cut into 5 μm-thick sections. These sections were further stained with 4-chloro-3-bromoindole-galactoside (X-Gal^®) and counter stained by nuclear fast red to reveal β-gal activity and nuclear locations respectively.

3. Results

3.1. Comparison of reporter molecules in vitro

T₁ and T₂ maps of the reporter molecules together with FAC in agar gel phantoms showed varying contrast in the presence of 5 units β-gal (Fig. 3). The largest responses in T₁ were observed for C3-GD, C2-MGD and C3-MGD with decreases of 1.12 s, 1.24 s, and 1.17 s, respectively, which is greater than 1.03 s observed using S-Gal^® (Table 1). Similarly, C3-GD and C3-MGD showed a decrease in T₂ of 45 ms and 95 ms respectively, which is significantly higher than 23 ms observed using S-Gal^® (Table 1). These values correspond to ΔR₁ and ΔR₂ about 1.8 times greater than S-Gal^® for C3-GD and ΔR₁ ∼ 2.8 and ΔR₂ ∼ 6.5 times S-Gal^® for C3-MGD. C3-GD was used for further in vivo studies due to better water solubility than C3-MGD.

Fig. 3.

In vitro lacZ gene reporter activity of various analogs of S-gal ^®. T₁ (top row) and T₂ (bottom row) maps of each sample containing 15 mM agent + 5 mM FAC in agarose without or with 5 units of β-gal enzyme (left and right sample in each box, respectively). Pronounced T₁ and T₂ effects are seen for the agents C3-GD and C3-MGD.

Table 1.

In vitro comparison of various analogs of S-gal^® as MR gene reporter molecules: T₁ and T₂ (mean ± standard deviation) of the synthesized gene reporter molecules (7.5 mM) in the absence or presence of β-galactosidase (5 units) in 1% agar containing ferric ammonium citrate (FAC, 2.5 mM).

Agent	T1 (s)		$\Delta {\mathrm{R}}_1^{\mathrm{\beta }-\text{gal}}$ (s-1)	T2 (ms)		$\Delta {\mathrm{R}}_2^{\mathrm{\beta }-\text{gal}}$ (s-1)
	w/o β-gal	w/β-gal		w/o β-gal	w/β-gal
C1-MGD	1.79 ± 0.28	1.46 ± 0.19	0.13	0.13 ± 0.04	0.09 ± 0.03	3.42
C2-MGD	1.83 ± 0.13	0.58 ± 0.05	1.18	0.12 ± 0.05	0.10 ± 0.03	1.72
C3-MGD	1.68 ± 0.12	0.51 ± 0.13	1.37	0.13 ± 0.05	0.03 ± 0.01	22.09
C4-MGD	1.78 ± 0.09	0.84 ± 0.05	0.63	0.13 ± 0.05	0.13 ± 0.03	0.09
S-Gal®	2.07 ± 0.12	1.04 ± 0.05	0.48	0.09 ± 0.03	0.07 ± 0.01	3.41
C1-GD	1.76 ± 0.13	1.11 ± 0.09	0.33	0.12 ± 0.03	0.12 ± 0.04	−0.84
C2-GD	1.81 ± 0.15	1.39 ± 0.09	0.17	0.12 ± 0.05	0.11 ± 0.03	0.76
C3-GD	1.83 ± 0.15	0.72 ± 0.12	0.84	0.11 ± 0.04	0.07 ± 0.01	5.97
C4-GD	1.85 ± 0.15	1.18 ± 0.06	0.31	0.13 ± 0.05	0.11 ± 0.04	1.67

Also listed are the corresponding differences in the relaxation rates due to β-galactosidase ( $\Delta {\mathrm{R}}_1^{\mathrm{\beta }-\text{gal}}$ and $\Delta {\mathrm{R}}_2^{\mathrm{\beta }-\text{gal}}$, respectively). T₁ and T₂ values of the agents C3-GD and C3-MGD decreased the most in the presence of the enzyme.

3.2. Evaluation of reporter molecule C3-GD in vivo

T₁ and T₂ maps of athymic nude mice bearing wild-type (WT) and lacZ-transfected MCF7 tumors before and after the injection of C3-GD (15 mM) + FAC (5 mM) solution (25 μL vol.) are shown in Fig. 4. T₁-weighted images showed small hyper-intensities, which were better perceived from the decrease in T₁ values, as seen in the T₁ maps of representative tumor slices (Fig. 3a). There was a small, but significant, difference between mean baseline T₁ (1.84 ± 0.28 s; WT (n = 6) vs. 1.88 ± 0.22 s; lacZ (n = 6); p < 0.0001; unpaired t-test) of the two tumor types (Table 2). Following injection of reporter molecule solution (C3-GD + FAC), T₁ decreased significantly in WT tumors (T₁ = 1.72 ± 0.29 s, p < 0.0001) after 1 h, but recovered back by the end of 2 h (T₁ = 1.97 ± 0.35 s, p < 0.0001) to exceed pre-injection values. In lacZ tumors, T₁ decreased significantly by the end of 1 h (T₁ = 1.69 ± 0.33 s, p < 0.0001) and showed continued decrease at 2 h (T₁ = 1.63 ± 0.25 s, p < 0.0001).

Fig. 4.

In vivo lacZ gene reporter activity of C3-GD. MRI of a representative nude mouse with wild type MCF7 tumor (left, dotted) and lacZ transfected MCF7 tumor (right, solid). A) Baseline T₁ weighted image and T₁ maps obtained before, 1 h after, and 2 h after direct intra tumoral injection of 15 mM C3-GD and 5 mM FAC (top row) and B) corresponding T₂ weighted images and T₂ maps (bottom row) showed decrease in the relaxation times in lacZ transfected tumors. C) X-gal and Nuclear fast staining of slices (whole mount) from the same wild type MCF7 (left) and MCF7-lacZ (right) tumors showed β-gal activity (intense blue stain from X-gal) for the MCF7-lacZ tumor section only.

Table 2.

In vivo lacZ gene reporter activity of C3-GD: T₁ and T₂ (mean ± standard deviation) obtained from ROI analysis of in vivo data (n = 6) from wild type and lacZ transfected MCF7 tumors.

	T1 (s)		$\Delta {\mathrm{R}}_1^{\mathrm{\beta }-\text{gal}}$ (s-1)	T2 (ms)		$\Delta {\mathrm{R}}_2^{\mathrm{\beta }-\text{gal}}$ (s-1)
	MCF7-WT	MCF7-lacZ		MCF7-WT	MCF7-lacZ
Baseline	1.84 ± 0.28	1.88 ± 0.22	−0.01	56 ± 16	51 ± 8	1.75
1 hr post injection	1.72 ± 0.29*	1.69 ± 0.33*	0.01	51 ± 16*	44 ± 10*	3.12
2 hr post injection	1.97 ± 0.35*	1.63 ± 0.25*	0.51	57 ± 18	44 ± 9*	5.18
Change over 1 hr	−6.5%	−10.1%	-	−8.9%	−13.7%	-
Change over 2 hr	7.1%	−13.3%	-	1.7%	−13.7%	-

Also listed are the corresponding differences in the relaxation rates (between wild type and lacZ tumor) due to β-galactosidase (ΔR₁^β-gal and ΔR₂^β-gal, respectively). MCF7-lacZ tumors showed a significant decrease in both T₁ and T₂ values at 2 h post injection compared to baseline.

Analysis of the same ROIs (as in T₁ analysis) on T₂ maps showed large decrease in T₂ for MCF7-lacZ tumors when compared to baseline (Fig. 4b). There was a significant difference in the baseline T₂ values of WT and lacZ transfected tumors (T₂ = 56 ± 16 ms (WT) vs. T₂ = 51 ± 8 ms (lacZ); p < 0.0001; unpaired t-test). Following administration of the solution of C3-GD + FAC, T₂ decreased significantly in WT tumors (T₂ = 51 ± 16 ms; p < 0.0001) at 1 h, but returned back to baseline value (T₂ = 57 ± 18 ms; p > 0.05) at 2 h (Table 2). In lacZ tumors, T₂ decreased significantly by the end of 1 h (T₂ = 44 ± 10, p < 0.0001) and remained unchanged at 2 h (T₂ = 44 ± 9, p < 0.0001). The X-gal and nuclear fast staining of tumor slices (Fig. 4) obtained following the MRI showed an intense blue stain representing β-gal activity only in the MCF7-lacZ tumor, and not in the MCF7 wild-type tumor (Fig. 4c).

4. Discussion

We have demonstrated the relative relaxation response of various S-Gal^® analogs to β-galactosidase. ¹H MR showed both T₁ and T₂ response in vitro and the lead agent was examined in vivo. The mono- and di-galactopyranosides showed differential enhancements in response to β-gal in the presence of ferric ions (Fe³⁺ from FAC). The aglycone products readily form paramagnetic iron chelates causing pronounced T₁ and T₂ shortening in MR images. The mono-galactopyranosides always showed better T₁ and T₂ contrast when compared to their associated di-galactopyranosides: e.g., C2-MGD and C3-MGD showed greater T₁ & T₂ shortening than C2-GD and C3-GD, respectively (Fig. 2). This can be attributed to the need for two reactions on the di-galactopyranoside before liberating the aglycone to form the iron chelate. The mono and di-galactopyranosides display a variety of hydrolytic rate constants all of which lead to reaction completion times that are greater than the 30 min incubation time used here (even after accounting for the higher enzyme amount used in [26]). Thus, differences between various analogs may reflect differences in product generation at the 30 min time point apart from the relaxivities of the chelation complexes. C3-MGD and C3-GD both demonstrated pronounced T₁ and T₂ shortening in the presence of 5 units of β-gal, but given its better water solubility, C3-GD was used for in vivo studies. Our goal in this study was to compare the various synthesized gene reporter molecules and identify lead candidates in vitro followed by in vivo testing of detectability of reporter activity. With that in mind, the intra-tumoral injection route was preferred over i.v. delivery to ensure consistent deposition of equal amount of the agent in lacZ transfected and control tumors. In comparison, an i.v. delivery would have resulted in heterogeneous deposition of the agent due to variations in vasculature and confounded the testing by the inability to distinguish between poor delivery and poor in vivo activity. We do note that solubility was not an issue for intratumoral delivery, but it could be important for potential systemic delivery applications in the future. At the concentrations used for direct intra-tumoral injections in this study, either agent could have been used and given the higher contrast effect and lower solubility, we can expect better performance of C3-MGD for applications that permit intra-tissue injection of the reporter probes. Given that the chelation complex is identical for C3-MGD and C3-GD, other practical concerns such as hydrolytic rates, cellular uptake and retention may dominate the effective observed contrast in vivo.

Following intra-tumoral injection of the C3-GD/FAC combination in wild type tumors, we observed a decrease in T₁ and T₂ from baseline to 1 h followed by return to baseline values by 2 h. This initial decrease in T₁ and T₂ values can be attributed to the free ferric ions, which are eventually cleared. In the lacZ expressing tumors, paramagnetic complexes are formed due to the chelation of ferric ions by the β-gal-cleaved aglycone molecules. This causes a shortening of T₁ and T₂ when compared to baseline (pre-injection) measurements. Catechol moieties exhibit a high affinity for Fe³⁺ and the complex is expected to stabilize the ferric state preventing other changes under biological conditions [27,28]. The aglycone-Fe³⁺ complex is poorly soluble and is likely to precipitate in situ (source of the strong T₂ contrast), further aiding the stability of the complex in vivo. The clearance of the nano-scale precipitate is expected to be slow compared to small molecular chelates, as seen previously with S-Gal^® [25]. In similar experiments with S-Gal^® we previously noted intense contrast immediately after injection due to T₂ shortening, whereas after one hour, the contrast declined considerably, presumably due to clearance [25]. By comparison, the current experiments used only 8% the concentration of FAC and 20% as much substrate, yet achieved ∼13% reduction in baseline relaxation times over 2 h in the same tumor types. This indicates better retention and higher in vivo sensitivity to C3-GD in tumors compared to S-Gal^®. The lower dose of FAC needed to achieve detectable contrast suggests greater potential feasibility for systemic delivery of the substrate/FAC combination for assessment of lacZ gene expression in vivo, though this has not been tested.

By analogy, alternative sugar moieties could be used to develop substrates to report other enzymes such as glucoronidases and glucosidases. The ability to observe both T₁ and T₂ contrast adds considerably to the confidence of identifying β-gal activity. Otherwise tissue heterogeneity may mask contrast. We chose to deliver the reporter system intra-tumorally in this proof of principle study to separate the effect of delivery from the activity on the observed contrast. In its existing state, this system can be useful for pre-clinical investigations of development and improvement of novel gene-therapy approaches. For potential human use, the system will need to be refined to the point where even endogenous Fe (up-regulated in cancer and a target for chelation therapy) can provide sufficient contrast. While intra-tumoral injections are sub-optimal for human use, we note that all the recent studies on gene-therapy in the clinic have used intra-tumoral/intra-organ injection of the adenoviral vector [29]. Most β-gal reporters to date also have required direct intra-tumoral injection, but systemic administration would be advantageous and remains a goal.