A Fluorescent Analog of Batimastat Enables Imaging of α-Secretase in Living Cells

Abstract

The ADAM family of metalloproteases cleave a diverse range of transmembrane substrates, resulting in the release of their soluble ectodomains. This process of protein shedding, termed α-secretase processing, is involved in many facets of both normal and disease related cellular function. While the processing of substrates has been well documented, the regulation and trafficking of the ADAMs are less well understood. Tools that allow for the study of ADAMs under their native environment will allow for a better understanding of their regulation and activity. Here we describe the design and evaluation of a novel fluorescent analog of a well-characterized ADAM inhibitor, Batimastat. This probe exhibited similar activity for inhibiting α-secretase processing in cells as did Batimastat. Importantly, this probe specifically labeled ADAMs fluorescently in both fixed and living cells, enabling the possibility to study the trafficking of α-secretase proteins in a dynamic environment.

The ADAMs (A Disintegrin And Metalloproteinase) are a family of cell surface transmembrane proteases that function in the shedding of diverse transmembrane proteins including growth factors, cytokines, receptors and adhesion molecules.¹ These cleavage events, termed α-secretase cleavage, play an important role in multiple aspects of basic cell biology.² For instance, cleavage of one protein, Notch, is critical for cell differentiation and development, while the processing of another protein, Amyloid Precursor Protein (APP), has implications for Alzheimer’s disease (AD) pathology.^3,4 Furthermore, certain mutations found in ADAM10 have been linked to late onset AD.^5,6 There is also emerging evidence for ADAMs as a potential cancer biomarker in cancer diagnosis and predicting patient outcome.^7–9

The ADAMs are expressed as an immature pro-enzyme that undergoes a Furin-like cleavage event^10,11 which liberates an inhibitory pro-domain, leading to an active protease.^12–14 There are both constitutive and regulated forms of α-secretase activity, which are putatively catalyzed by ADAM10¹⁵ and ADAM17,¹⁶ respectively. The regulation of these proteases is complex and still poorly understood. Chemical probes that target these enzymes could serve as valuable tools to help understand how they are regulated in a dynamic cellular environment.

Small fluorescent probes have been used to reveal the location of proteins within their native environment by fluorescence microscopy. Such fluorescent small-molecule probes that bind specifically to a protein of interest (e.g., kinases,¹⁷ monoamine oxidase A,¹⁸ legumain,¹⁹ cycloxygenase-2,²⁰ etc.) may offer specific advantages over other methods to track proteins (e.g., fluorescent proteins tags and immunofluorescence). Unlike antibodies, small molecule probes can be readily cell-permeable, which may be useful for dynamically interrogating the intracellular location and movement of proteins in living cells and in vivo. In contrast to tools that exploit the expression of proteins fused to a fluorescent tag (e.g., EGFP), small-molecule probes do not require genetic manipulation of cells that can introduce broad artifacts in protein expression, stability, and localization,²¹ and they generally do not alter the structure of the protein, as tags may do.

Here we describe the synthesis and evaluation of a fluorescent analog of an ADAM inhibitor in order to develop an ADAM-specific, small fluorescent probe for live-cell imaging.

Results and Discussion

In order to develop a fluorescent probe that could target ADAMs, we based our probe design on the structure of Batimastat (BB94, Fig. 1), a potent synthetic inhibitor of ADAMs.²² BB94 has a low molecular weight and exhibits a peptide-like collagen motif. It comprises a peptide backbone and a hydroxamic acid group, which bind to ADAMs through coordination with the catalytically active zinc atom.²³ Modified BB94 has been reported previously for the development of an inhibitor-tethered resin for isolation of BB94-targeted proteins from cell lysates.²⁴ Previous work from structural studies of BB94 bound to a metalloproteinase suggest that the thienyl moiety of BB94 could be replaced with other functionalities without significantly affecting binding to the ADAMs.²⁵ We, therefore, envisioned that we could replace the thienyl moiety of BB94 with a 2-aminoethanethiol linker, which could later be used for attachment of a fluorophore through reaction of the amine in 1 with an activated carboxylic acid (Fig. 1).

Fig. 1.

Structures of BB94 and synthetic route for the formation of reactive BB94 analog 1 and probe 5. i) NH₂OH.HCl, HATU, DBU, TEA, DMF, rt, 16h, (47%); ii) 2-aminoethanethiol, K₂CO₃, THF, rt, 24h, (45%); iii) TFA, DCM, 0°C to rt, 4h, (Qt.); iv) 5-carboxyfluorescein, PyBOP, TEA, DMF, rt, 16h, (29%).

Intermediate 2 was initially synthesized from D-Leucine using the previously described route (7 steps).²⁴ Conversion of the carboxylic acid in 2 to the corresponding hydroxamic acid in 3 was accomplished using HATU-mediated coupling in presence of hydroxylamine (Fig. 1). The conversion of hydroxamic acid 3 to thioether 4 was carried out with 2-aminoethanethiol in a Michael-type reaction with high stereocontrol as previously reported.^24,26,27 After Boc-deprotection of 3 using trifluoroacetic acid, the resulting amine 1 was engaged in a PYBOP-mediated coupling with 5-carboxyfluorescein (5-FAM) to form fluorescent probe 5 (see supporting information, Fig. S1). Excitation and emission spectra of probe 5 were recorded and showed retention of the fluorescent properties of the parent fluorophore (see supporting information, Fig. S2).

In order to evaluate the activity of fluorescent BB94 analog 5, we initially examined whether this molecule retained the capability to inhibit α-secretase activity in living cells. CHO cells that stably express the canonical α-secretase substrate APP (7W cells)²⁸ were treated with either probe 5 or BB94 as a control. Conditioned media from treated cells were examined for the levels of APPsα and APPsβ (the products of the α-secretase and β-secretase cleavages of APP, respectively) by a multi-analyte ELISA (Fig. 2A–B); furthermore, APPsα levels were confirmed by Western blot (Fig. 2C, D). Lysates from these cells were also analyzed for the levels of APP, the γ-secretase component Nicastrin (NCT), and GAPDH as a loading control by Western (Fig 2D).

Fig. 2.

Probe 5 inhibition of α-secretase activity on 7W cells as compared to BB94. (A–B) 7W cells were treated with increasing amounts of either BB94 or probe 5 (0–20μM), conditioned media from treated cells were analyzed by an MSD ELISA for APPsα (A) and APPsβ (B). Experiments were performed once in duplicate. (C–D) Conditioned media along with cellular lysates were probed for the protein levels of APPsα, APP, GAPDH, and Nicastrin (NCT) by Western blot (D) and Western blot levels of APPsα were quantified in (C).

We found that probe 5 exhibited indistinguishable activity for inhibiting the production of APPsα as compared to the parental compound BB94 (Fig. 2A and 2C), with IC50 values of 2.6 and 2.3 μM, respectively. These data suggest that modification of BB94 with a fluorophore as in probe 5 does not appreciably affect the capability of this molecule to engage its target in living cells. As a control, we found no statistically significant effect (P = 0.2191) of probe 5 (0–20 μM) on production of APPsβ in these cellular activity studies (Fig. 2B), demonstrating that probe 5 does not have any off-target inhibition of β-secretase. Furthermore, probe 5 did not exhibit general toxicity to 7W cells as compared to BB94, which were both well tolerated by the cells (see supporting information, Fig. S3).

In order to evaluate the utility of probe 5 for imaging studies, we first stained methanol-fixed 7W cells with a high concentration (50 μM) of probe 5. Treatment with probe 5 displayed primarily a perinuclear staining pattern (likely endoplasmic reticulum) along with lesser staining throughout the cell body (Fig. 3A). This staining pattern was very similar to what was observed when cells underwent fluorescent immunostaining for ADAM10, one of the primary α-secretases¹⁵ (Fig. 3B).

Fig. 3.

Probe 5 fluorescent staining on 7W cells. (A) Fluorescence imaging on methanol-fixed 7W cells after incubation with 50 μM of probe 5; Scale bar = 20 μm (B) ADAM10 (PC528, Calbiochem) immunofluorescence on 7W cells; Scale bar = 20 μm (C) Fixed 7W cells were incubated with 25 μM BB94, 25 μM probe 5 or both simultaneously at a ratio of 1:1 or 1:5; probe 5:BB94; probe 5 concentration was kept constant at 25 μM; Scale bar = 20 μm (D) Quantitation of (C) using a one-way ANOVA with a Dunnett’s post-test comparing all conditions to Probe 5 control; **p<0.01, ***p<0.001 (E–F) 7W cells were transfected with siRNA targeting ADAM9, 10, 17, 9/10/17, BACE1 or no siRNA as a control, and stained with probe 5 and imaged on a Zeiss LSM-710 (E) and quantitated in (F). A one-way ANOVA using Dunnett’s post-test comparing all conditions to No siRNA control was performed; **p<0.01, ***p<0.001. (G) Cellular lysates of cells treated with siRNA were probed for the protein levels of ADAM9, ADAM10, and ADAM17 to confirm siRNA engagement of target.

In order to evaluate the specificity of probe 5 for BB94-specific targets, we repeated the labeling of cells at a lower concentration (25 μM) of probe 5 in the presence of the non-fluorogenic BB94 compound (Fig. 3C). Fluorescence of probe 5 in the cells was reduced to 44% and 23% of the non-competed probe 5 signal when BB94 was added in a 1:1 or 1:5 ratio of probe 5 to BB94 (Fig. 3D). These values are close to the expected theoretical values of 50% and 17% reduction in fluorescence, respectively, if we assume a similar binding affinity of probe 5 and BB94 to ADAM proteins. Therefore, these competition experiments confirm that probe 5 is highly specific for binding to BB94-cellular targets without significant off-target labeling of proteins.

To determine whether probe 5 staining was specific to the ADAMs, we examined the fluorescence staining of probe 5 on 7W cells that were treated with siRNAs targeting three different ADAMs (Fig. 3E). siRNA knockdown of ADAM9, 10, 17, or all three ADAMs was able to reduce the staining of probe 5 in cells to 72%, 62%, 76%, and 53% of no siRNA control respectively (Fig 3F). As a control, siRNA knockdown of BACE1, a β-secretase, did not reduce fluorescent labeling by probe 5, demonstrating that our compound has specificity for α-secretase. Western blot of cellular lysates generated from cells treated with the siRNAs confirmed knockdown of the respective ADAMs (Fig. 3G). In these experiments, we knocked down only three of many ADAM proteins that are known to be targeted by BB94.²⁹ We attribute the remaining fluorescence signal of cells stained with probe 5 after knockdown of ADAM9, 10, and 17 to the labeling of other metalloproteinases (i.e., ADAMs) that are typically targeted by BB94.

To determine whether probe 5 could be used as a tool to label ADAMs in living cells, live 7W cells were incubated with 100 μM of probe 5 in serum-free medium overnight. The following day, cells were imaged by confocal microscopy and the uptake of probe 5 was confirmed by the strong and uniform vesicular pattern of staining of the cells (Fig. 4A). We observed a perinuclear staining by probe 5 (Fig. 4B) that was similar to that observed in our previous staining of fixed cells (Fig. 3A), indicating that probe 5 can be used to stain α-secretase in living cells. As a control, we observed no fluorescence staining in live cells when we used the non-fluorescent parental compound, BB94 (data not shown).

Fig. 4.

Live cell imaging of probe 5 on 7W Cells. Live 7W cells were treated with 100 μM probe 5 (green channel) and Hoechst (blue channel) and imaged by confocal microscopy at low (A) and high (B) magnifications: Scale Bars = 200 and 50 μm for A and B, respectively.

In conclusion, we developed a cell-permeable, fluorophore-conjugated derivative of the α-secretase inhibitor Batimastat (BB94). Once conjugated to a fluorophore, the resulting probe showed no overt cytotoxicity and exhibited activity for inhibiting α-secretase activity that was indistinguishable from the parent compound. Probe 5 also demonstrated a strong selectivity for ADAM staining in both fixed and living cells. These results serve as the foundation for the development of chemical tools to help understand how ADAMs are regulated in a dynamic cellular environment using fluorescence imaging. The application of this probe to study the effects of various protein modulators and phorbol ester treatment on ADAM trafficking and regulation is currently under investigation.

Methods

Synthesis of probe 5

See supporting information for details.

Cell culture, Inhibitor treatments, and siRNA transfections

7W CHO cells were grown in standard DMEM plus 10% fetal bovine serum, 2 mM L-glutamine, 100 μg/mL streptomycin and 100 units/mL of penicillin at 37°C with 5% CO₂ atmosphere. For inhibitor treatments, BB94 or Probe 5 were diluted to the stated concentration in Opti-MEM (Life Technologies) and cells were treated overnight. The following day, conditioned media was collected and cells were harvested in 50 mM HEPES Buffer containing 150 mM NaCl and 1% CHAPSO detergent supplemented with protease inhibitor cocktail (Roche). Conditioned media (CM) and lysates were normalized to protein concentration as measured by a BCA assay (Thermo Fisher). For knockdown experiments, siRNA purchased from Dharmacon were transfected with the RNAiMAX reagent (Life Technologies) according to manufacturer’s specification. The next day, transfected cells were plated onto a 4-well chamber slide for immunofluorescence and a 6-well plate for biochemistry. 48 hours post transfection, cells in the 4-well chamber slide were fixed and stained and cells in the 6-well plate were lysed for Western blot.

Western Blot and ELISA

CM or cellular lysates were blotted as previously described.³⁰ Briefly, LDS Sample Buffer (Life Technologies) was added to samples to a final concentration of 1X and heated to 65°C for 5 minutes. Samples were then loaded onto a 4–12% Bis-Tris gel (MES buffer) and then transferred to Nitrocellulose Membrane. For detection of various proteins we used AB124695 for ADAM10; AB75609 for ADAM17; a polyclonal antibody from Cell signaling for ADAM9; a monoclonal antibody from BD Transduction Labs for Nicastrin; GAPDH antibody from Millipore; 6E10 or 1736 for APPsα; and an in-house generated antibody C7 for APP. Blots were developed by either ECL or the Odyssey infrared imaging system (LI-COR). For APPs ELISA, we used the multi-analyte ELISA kit from Meso Scale Discovery following manufacturer’s directions.

Immunofluorescence and Live Cell Imaging

For immunofluorescence of probe 5, 7W cells were fixed with ice cold methanol for 15 minutes, followed by blocking with 2% normal donkey serum with 0.1% Triton for 60 minutes. Probe 5 was diluted in either PBS or ethanol at the specified concentration (with or without competing BB94) and incubated overnight at 4°C. The next day, cells were washed with 80% ethanol followed by three washes in PBS. DAPI was added to the second PBS wash to stain for the nucleus. Fluorescence was detected by confocal microscopy using a 63X oil immersion objective on a Zeiss LSM-710. For live cell imaging, cells were incubated with 100 μM Probe 5 in Opti-MEM media overnight at 37°C with CO2. The following day cells were washed with fresh Opti-MEM media, treated with Hoechst for 10 minutes, and washed twice with Opti-MEM, and imaged with a 10X and 40X air objective.

Quantification and Statistical Analysis

All quantification of Western blots were performed using the Odyssey infrared imaging system (LI-COR), with data normalized to vehicle (DMSO) control. For immunofluorescence images, quantitation was done using ImageJ software, where individual cells were outlined and quantitated. All data were background subtracted and normalized to control. A one-way ANOVA with Dunnett’s post-test was performed with significance designated at p<0.05.