Detection of proteolytic activity by covalent tethering of fluorogenic substrates in zymogram gels

Introduction

Zymography is a modified SDS-PAGE technique that allows for the separation of proteases by molecular weight and the detection of proteolytic activity through degradation of large, native proteins incorporated into the polyacrylamide network [1–3]. Gelatin is the most commonly used protein in this technique, enabling the measurement of the active and pro-form of the gelatinases, MMP-2 and -9 [4]. Zymography has also been adapted to measure other proteases, including MMP-1, -7, -8, -12 and -13 [3], as well as cathepsins K, L, S and V [5], by incorporating proteins such as casein [6] and collagen [7,8]. In addition to protein substrates, previous studies have also incorporated carbohydrates and lipids into polyacrylamide gels in order to measure the activity of amylases and lipases [9]. Nonendopeptidases have also been measured through the inclusion of degradable substrates into the developing buffer solution, which can be detected following their reaction with a protein dye [10]. However, this technique is still not applicable to a vast majority of the currently identified 500+ proteases, due in large part to the limited number of substrates that have been successfully incorporated into acrylamide gels. The utility of zymography would be greatly expanded if it was amenable to a wider range of degradation substrates.

Incorporation of fluorescent peptide substrates into polyacrylamide gels is a promising alternative to native, full-length proteins, and would greatly expand the number of proteases detectable by zymography. Peptide substrates afford researchers the benefit of being highly tunable – a myriad of peptide substrates have already been developed for cleavage by a wide range of proteases. Combining fluorescent, protease-sensitive substrates and zymography was first demonstrated by Yasothornsrikul and Hook in 2000, in which they included a fluorogenic peptide labeled with a MCA (4-methyl-coumaryl-7-amide) group into the acrylamide solution prior to polymerization [11]. This method of incorporating synthetic substrates into electrophoresis gels, however, has not been widely implemented. We speculate that this may be due in part to the transient nature of physical entrapment of a small molecule (i.e., the fluorescent peptide) within the hydrogel structure. This limitation was alluded to by the authors themselves, in which they state that the gels should be prepared the same day as electrophoresis, and recommend the use of short incubation times and immediate detection of fluorescence [11]. To overcome this limitation, we have developed a method for covalently coupling fluorescent peptides within the acrylamide structure, enabling the stable incorporation of these degradable moieties and increasing the utility of zymographic techniques for the study of a wider range of proteases.

Materials & methods

Fluorescent peptide synthesis

Peptides (GGPQG↓IWGQK^Dde(PEG)₂C [abbreviated as QGIW throughout the text and figures to denote the P2-P2’ residues]; GPLA↓C^pMeOBzlWARK^Dde(PEG)₂C [abbreviated as LACW throughout the text and figures to denote the P2-P2’ residues];↓ indicates cleavage site) were synthesized using Fmoc solid-phase peptide synthesis (CEM Liberty Blue Peptide Synthesizer, NC, USA) using a Rink Amide MBHA resin (EMD-Millipore, MA, USA). The QGIW peptide is a commonly used, collagen I-derived sequence [12]. LACW is an artificial, commercially available peptide that has been optimized for MMP-14 detection [13]. All Fmoc-protected amino acids and (PEG)₂ spacers (Fmoc-8-amino-3,6-dioxaoctanoic acid) were purchased from Chem-Impex (Chem-Impex, IL, USA). Peptides were then functionalized with a quencher (dabcyl) and fluorophore (fluorescein) as previously described [14]. Briefly, a dabcyl succinimidyl ester (Anaspec, CA, USA) was coupled to the amino terminus of the peptide in dimethyl formimide (DMF) with six equivalents N,N’-diisopropylethylamine and reacted overnight. An orthogonally protected lysine (K^Dde) was then deprotected twice in 2% hydrazine monohydrate in DMF for 10 min and a fluorescein NHS ester (Anaspec) was coupled to the resultant free amine in the same manner as dabcyl. The peptides were cleaved from the resin beads by incubating in a cleavage cocktail of trifluoroacetic acid, phenol, triisopropyl silane and water (95/2.5/1.25/1.25 v/v) for 3 h at room temperature, and precipitated in chilled diethyl ether three times. Peptides were purified by reverse-phase high-performance liquid chromatography (Hitachi, IL, USA) (52–70% gradient of acetonitrile) and molecular weight (QGIW: 1884 Da; LACW: 2336 Da) verified by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (Bruker, MA, USA).

Fluorescent substrate gel preparation

The azido-PEG3-maleimide heterobifunctional crosslinker was prepared as detailed by the manufacturer (Click Chemistry Tools, AZ, USA). A two-layer 10% polyacrylamide resolving gel was cast at room temperature in a 1.5 mm mini gel cassette (Thermo Fisher Scientific, PA, USA) (Figure 1). The first layer (5 ml) was prepared using a 40% acrylamide/bis-acrylamide solution (19:1; Thermo Fisher Scientific) as per the method described by Laemmli [15] and allowed to polymerize for 1 h. The second layer (3 ml) was identical to the first, but contained 2 mM azido-PEG3-maleimide crosslinker and 100 μM fluorescent peptide, and was allowed to polymerize for 1 h at room temperature. A 4% stacking gel was cast in the remaining portion of the gel cassette. All gels were stored at 4°C in PBS for no more than 3 weeks.

Enzyme analysis

Serial dilutions of type I bacterial collagenase enzyme (20 mg/ml) (Life Technologies, PA, USA) were dissolved in 3X zymography sample buffer (62.5 mM Tris-HCl, 25% glycerol, 4% SDS, 0.01% Bromophenol Blue; Bio-Rad, PA, USA) under nonreducing conditions and subjected to electrophoresis in peptide gels at 120 V for 1.5 h at 4°C. Following electrophoresis, gels were washed three times for 15 min each at room temperature under gentle agitation in wash buffer containing 2.5% Triton X-100 in 50 mM Tris-HCl, pH 7.5. Gels were then transferred to a developing buffer solution, 1 μM ZnCl₂ and 5 mM CaCl₂ in 50 mM Tris-HCl, pH 7.5, overnight at 37ºC under gentle agitation [1]. Fluorescent images of the gels were captured at 24 h using a GE Amersham Typhoon 9410 Gel Imager (Excitation 488/Emission 521). This procedure was repeated using purified enzymes, MMP-2 (EMD-Millipore) and MMP-9 (EMD-Millipore) and recombinant human MMP-14 catalytic domain (31 kDa; R&D Systems, Minneapolis, MN, USA). MMP-2 and -9 were activated by 10 mM 4-aminophenylmercuric acetate (Sigma-Aldrich) for 1 and 2 h at 37°C, respectively. rhMMP-14 was activated by furin convertase (R&D Systems) at 37°C for 2 h.

Cell culture

HT1080 fibrosarcoma (American Type Culture Collection, VA, USA; RRID: CVCL_0317) and MDA-MB-231 breast adenocarcinoma (American Type Culture Collection; RRID: CVCL_0062) cell lines were purchased from the American Type Culture Collection and cultured in RPMI-1640 media (Life Technologies) supplemented with 10% FBS, 1% L-glutamine and 1% penicillin/streptomycin. All cells were grown in humidified chambers (5% CO₂ at 37°C). For zymographic analyses, cells were seeded at a density of 40,000 cells/cm² in six-well plates for 24 h in 10% FBS media, washed with PBS twice and then cultured in serum-free media for an additional 24 h. Media samples were collected and concentrated using 10-kDa Amicon Ultra Centrifugal Filter Units (EMD-Millipore). Protein content was quantified by μBCA according to manufacturer protocols (Thermo Fisher Scientific).

Gelatin zymography

30 μg of protein isolated from conditioned cell media were loaded under nonreducing conditions onto precast gelatin zymography gels (10% polyacrylamide, 0.1% gelatin; Thermo Fisher Scientific) and electrophoresed at 120 V for 2 h at 4°C. Gels were washed and incubated in developing buffer as described for peptide gels. After 24 h, gels were stained with 0.5% (w/v) Coomassie Brilliant Blue (Thermo Fisher Scientific) in an aqueous solution of 40% methanol and 10% acetic acid for 2 h. Gels were then destained in the same solution, but without Coomassie Blue, for 10 min [1]. Gels were imaged with a FluorChem E gel imager (Protein Simple, CA, USA) using the UV transilluminator (365 nm).

Densitometry & statistical analysis

All protein and fluorescent bands were quantified using ImageJ 1.43 software (NIH). Graphs were created using GraphPad Prism 7.0 software (GraphPad Sofware, Inc., CA, USA). Plots are of band intensity normalized to the highest concentration. EC₅₀ values were calculated using Prism software after fitting values to a four parameter variable slope curve. At least three independent experiments were carried out for each experiment and data are shown as mean ± SD.

Results & discussion

Incorporation of fluorescent peptides using an azido-PEG3-maleimide crosslinker

Expanding the library of degradable substrates that can be incorporated into zymography gels would enable the detection of many more proteases than is currently attainable. To achieve this objective, we have developed a technique to covalently incorporate synthetic, fluorescent peptides with well-defined degradation characteristics using an azido-PEG3-maleimide chemical crosslinker. Maleimide groups react with thiol moieties in the C-terminal cysteine residue [16]. Azide groups covalently attach to the polyacrylamide side-chains, although this mechanism is not well understood [17]. This crosslinking reaction immobilizes the fluorescent substrate within the polyacrylamide matrix, allowing for the detection of proteolytic activity (Figure 1). To test the feasibility of this approach, we incorporated a widely used collagen I-derived peptide sequence, GGPQG↓IWGQK(PEG)₂C (abbreviated as QGIW throughout the text and figures to denote the P2-P2’ residues), functionalized with a fluorophore (fluorescein) and quencher (dabcyl) pair on either side of the cleavage site (indicated by ↓) [18]. This modified version of the fluorescent, MMP-sensitive peptide was previously developed for covalent incorporation in norbornene-functionalized poly(ethylene glycol) hydrogels, enabling facile detection of cellular collagenase activity with an approximate tenfold increase in fluorescence intensity upon cleavage [14]. While these peptides were synthesized and purified in-house using standard solid-phase peptide synthesis and N-hydroxysuccinimide (NHS)-ester-amine chemistry for fluorophore coupling, modified peptides can also be purchased through a number of commercial peptide synthesis companies. We have found that adapting peptide sequences from commercially available fluorescent protease substrates is often a successful strategy for developing usable peptides that are adequately quenched and soluble in standard buffers.

Polyacrylamide gels were modified with the QGIW peptide (100 μM) through covalent attachment with the azido-PEG3-maleimide (2 mM) crosslinker by addition of these molecules to the acrylamide monomer solution (Figure 1). The peptide-modified zymogram gel consisted of three layers – a standard stacking gel layer, and two resolving gel layers. The upper resolving gel contained the fluorescent degradable peptide substrate, while the lower resolving gel was a standard polyacrylamide gel. A multilayer gel approach was used to reduce the total amount of fluorescent peptide and crosslinker utilized in each gel. To determine the optimal time of electrophoresis, a molecular weight ladder was run concurrently with samples until the 28-kDa standard was at the bottom of the light orange peptide-modified resolving gel (as compared with the colorless nonmodified resolving gel). Most proteases fall between 30 and 100 kDa, thus this electrophoretic migration time maximizes the number of detectable proteases and the separation between bands. Following electrophoresis of serial dilutions of bacterial type I collagenase enzyme, the gels were washed three times in a Triton-X buffer to remove the SDS and incubated for 24 h at 37°C in a development buffer, enabling the proteases to renature and degrade the peptide substrate. Using a standard gel imager (excitation 488/emission 521), fluorescent bands were clearly visible (Figure 2B). To determine if the crosslinker was necessary for the observed protease bands, collagenase enzyme was electrophoresed in polyacrylamide gels made with the fluorescent peptide but without the crosslinker. In these gels, no bands were visible at any time point (24, 48, or 72 h) following development (Figure 2A). These results indicate that covalent attachment of thiol-functionalized fluorescent peptides using the azido-PEG3-maleimide crosslinker is a necessary and feasible strategy for retaining substrates within the gel matrix and for visualizing proteolytic activity.

Working range & time-dependence of QGIW gels

The working range of an assay is an indicator of the upper and lower bound of concentrations that can be reliably detected, typically defined as three standard deviations above the minimum and three standard deviations below the maximum detectable signals. To establish the working range of this assay, serial dilutions of type I collagenase enzyme (1.6–400 μg) were electrophoresed in QGIW gels. Fluorescent images were taken at 0, 2, 5 (data not shown), 24, 48 and 72 h after placement in developing solution. Bands were not detectable until the 24 h time point (Figure 2B). Densitometric evaluation of the ∼130-kDa band indicated that the band intensity increased logarithmically with increasing amounts of collagenase enzyme (Figure 2C). Two bands were detected from the collagenase samples at ∼130 kDa and ∼35 kDa. Type I bacterial collagenase enzyme isolated from Clostridium histolyticum can contain several collagenase isoforms of varying lengths and structures. Six isoforms were previously identified by Bond & Van Wart, of which collagenase ζ had a molecular weight of 125 kDa [19]. This is likely the isoform that is observed in our gels. In line with these observations, Bond & Van Wart also concluded that collagenase ζ had a preference for short, synthetic peptides over the native, triple-helical structure of collagen. The working range was determined to be 12–100 μg of collagenase enzyme using the QGIW gels, within which proportional changes in collagenase concentration could be reliably detected. Similar trends were observed with analysis of the smaller ∼35-kDa band (data not shown). It is also well known that crude collagenase solutions can contain several other proteases that might also be able to cleave the QGIW peptide. We speculate that this band likely corresponds to one of the other proteases present within the solution.

Zymography and protein quantification techniques often rely on the use of staining agents or antibodies. These signals can easily be altered based on staining, destaining and incubation times. To evaluate the stability of the fluorescent signal, band intensities were quantified at 48 and 72 h of incubation in developing buffer (Figure 2C). Analysis revealed that the band intensity remained concentration-dependent at extended time points, presenting a clear advantage over current techniques.

MMP detection in QGIW & LACW gels

Numerous fluorescent peptides have been developed to monitor the activity of a number of specific proteases. To test the adaptability of this technique to other peptide sequences, we synthesized a second peptide based on a substrate optimized for the detection of MMP-14 [13], a critical regulator of cellular migration and invasion [20–23]. GPLA↓C^pMeOBzlWARK(PEG)₂C (abbreviated as LACW throughout the text and figures to denote the P2-P2’ residues; ↓ indicates cleavage site) was functionalized with fluorescein and dabcyl as described for the QGIW peptide. Polyacrylamide gels were produced with the QGIW or LACW peptides, and purified MMP-2 (125 ng) and -9 (125 ng) and recombinant human MMP-14 catalytic domain (31 kDa; 125 ng) were electrophoresed in both gels (Figure 3A). QGIW peptide gels displayed high sensitivity towards MMP-9, resulting in the appearance of proMMP-9 (92 kDa) and MMP-9 (82 kDa) bands. The QGIW gel showed moderate sensitivity against MMP-2 (68 kDa), while recombinant MMP-14 was undetectable. MMP-9 was also a potent degrader of LACW peptide gels, resulting in the appearance of proMMP-9, MMP-9 and a third, lower molecular weight active product (∼35 kDa) that was previously described by Ries and colleagues [24]. MMP-2 (68 kDa) and rhMMP-14 (31 kDa) moderately degraded the LACW peptide. Although the LACW peptide was originally designed to be more sensitive towards MMP-14, it is possible that the recombinant, truncated version of the protein is less active against the peptide than its native, membrane-bound form. Alternatively, recombinant MMP-14 may less easily regain its structure and function after electrophoresis and renaturation than purified MMP-2 and -9. This is a common limitation of zymographic techniques as the renaturation step only partially restores catalytic activity of the proteases [25]. Given these results, we show the versatility of our system in expanding the library of detectable proteases through the incorporation of tunable peptide substrates into polyacrylamide gels.

MMP-9 sensitivity in peptide & gelatin zymography

Gelatin zymography is the current gold standard method by which MMP-9 activity is detected and serves as a highly sensitive standard for comparison. Serial dilutions of purified MMP-9 (1–250 ng) were electrophoresed in peptide and gelatin zymograms in order to compare their sensitivity in detecting the enzyme (Figure 3B). Plots of normalized band intensity were used to generate EC₅₀ values – the concentration that produces half the maximum signal (Figure 3C). LACW peptide gels had the lowest EC₅₀ value of 17.28 ng in comparison to gelatin zymograms with an EC₅₀ value of 31.85 ng. QGIW had the highest measured EC₅₀ value of 59.50 ng, indicating the lowest sensitivity. These results demonstrate that the use of artificial peptide sequences in zymography enables a measurement of enzyme activity that has sensitivity similar to the current gold standard, gelatin zymography, and the possibility to develop even more sensitive methods than is currently possible with native proteins.

Detection of cell-secreted protease activity

The ability to study specific proteases in the context of a complex mixture could provide tremendous insight into the secreted protease profile of various biological samples. Previously, zymography has been used to measure proteases in plasma [26], blood [27], tissue homogenates [28,29], conditioned cell media and lysates [29], and other bodily fluids to investigate the role of proteases in different physiological and pathophysiological processes. These efforts, however, have been hindered by the limited scope of current zymographic techniques. We sought to determine if our method was capable of separating a complex mixture of proteases in the conditioned media of HT1080 fibrosarcoma and MDA-MB-231 breast adenocarcinoma human cancer cell lines. 30 μg of protein from concentrated conditioned cell media were electrophoresed along with type I collagenase (100 μg) and MMP-9 (125 ng). Numerous bands were visible in fluorescent images of LACW gels taken after 24 h of incubation (Figure 4A), potentially corresponding to proMMP-9 (92 kDa), MMP-9 (82 kDa), MMP-9 (35 kDa) [24] and MMP-11 (47 kDa) [30]. QGIW gels displayed a single MMP-9 band in both cell types (Figure 4D). Gelatin zymography analysis (Figure 4G) showed increased proMMP-9, proMMP-2 (72 kDa) and MMP-2 (62 kDa) activity in HT1080 cell media as compared with MDA-MB-231 cell media. In comparison to LACW gels, which displayed numerous proteolytic bands, gelatin zymograms provided a narrower analysis of the cell-secreted protease profile, limited to the gelatinases. This highlights the ability of fluorescent substrate zymography to significantly expand the library of detectable proteases, facilitating more insightful and sensitive characterizations of the protease activity of biological samples.

To verify the identity of the fluorescent bands as belonging to the MMP family, experimental gels were incubated in buffer containing either 20 μM GM6001, a broad spectrum MMP inhibitor, or 10 μM E-64, a cathepsin inhibitor, as a negative control. The intensity of HT1080 and MDA-MB-231 cell media sample bands in LACW gels decreased under treatment by GM6001 (Figure 4B), but were unaffected by E-64 (Figure 4C), suggesting that the bands likely correspond to MMPs. Interestingly, GM6001 treatment of LACW gels did not affect the activity of 35-kDa MMP-9. In QGIW gels, MMP-9 activity bands were also ablated by GM6001 treatment (Figure 4E), but were unaffected by E-64 treatment (Figure 4F).

Here, we have developed a modified zymographic technique that utilizes a heterobifunctional crosslinker to covalently link fluorescent peptide substrates to a polyacrylamide polymer network. This coupling has enabled the separation of complex mixtures of cell-secreted enzymes and expanded the total number of detectable proteases using zymography. These results demonstrate that this process is modular in design, allowing for the incorporation of thiol-containing peptides and detection of various enzymes without the use of primers, antibodies or complex imaging systems. Finally, we demonstrate that this method is as sensitive as gelatin zymography in measuring MMP-9, and this sensitivity can be further enhanced by using alternative peptides.

Future perspective

This method has applications in studies of fundamental biology as well as in the development of biomaterials with well-defined degradation rates for tissue engineering. In biology, fluorescent peptide zymography can complement investigations into the role of proteases in disease, development and wound healing by capturing their compensation strategies, drug response and changes over the various stages of these processes. The use of fluorescent imaging to quantify protease activity, as opposed to the standard colorimetric readout of protein labeling in standard zymography, also opens up the possibility of using multiple substrates modified with different fluorescent dyes to create a multi-color system that would facilitate the simultaneous detection of numerous proteases.