2.1.

Construction of Plasmids

DsRed2 was amplified with the forward and reverse primers $5^{\prime }$ -GGGAAGATCTATGGCCTCCTCCGAGAACGTCATCACCG- $3^{\prime }$ and $5^{\prime }$ -CGGGATCCGCCTCCCTCCAGGAACAGGTGGTGGCGGC- $3^{\prime }$ . The amplified fragment was digested with BglII and BamHI and cloned into pGEM-T. The forward and reverse primers of enhanced cyan fluorescent protein (ECFP) were $5^{\prime }$ -CGGGATCCGTGCCCCTTAGCCTGTACAGCGGATCTAGAATGGTGAGCAAGGGCGAGGAGC- $3^{\prime }$ and $5^{\prime }$ -CAAGTCGACCATGCGGGCGGCGGTCACGAACTCCAGCAGG- $3^{\prime }$ , respectively. The polymerase chain reaction (PCR) products of ECFP were cloned into pGEM-T (Promega). The cloned cDNA of DsRed2 was selected with a right-inserted direction, and then the cloned cDNA of ECFP was conformed by sequencing. Next, the ECFP PCR product was digested with BamHI and SalI and subcloned at the BamHI/SalI site of pGEM-T-DsRed2. This subclone was digested with BglII and SalI to produce the cDNA fragment of DsRed2-MSS-ECFP. This was subcloned into the pDisplay vector at the BglII and SalI sites, and also into pET-28a at the BamHI and XhoI sites.

2.2.

Expression and Purification of Recombinant Proteins

The cDNA of DsRed2-MSS-ECFP was subcloned into vector pET-28a and then was transformed into BL21 (DE3) E. coli (Novagen Corp.) to produce the recombinant 6His-tagged DsRed2-MSS-CFP. Bacteria was cultured to $A_{600}=1$ and induced with $1\mathrm{mM}$ IPTG (isopropyl-beta-D-thiogalactopyranoside) for $16\mathrm{h}$ at $22°\mathrm{C}$ . Then cells were disrupted by sonication in $30\mathrm{ml}$ lysis buffer ( $20\mathrm{mM}$ Tris. Cl, pH 7.9, $0.5\mathrm{M}$ $\mathrm{NaCl}$ , $1\mathrm{mM}$ PMSF) followed by centrifugation ( $\mathrm{12,000}\times \mathrm{g}$ , $40\min$ ). The recombinant protein was further purified by Ni-NTA resin (Qiagen). For further analysis, the recombinant protein was dialyzed in deionized water.

2.3.

Cell Culture and Transfection

Samples of human breast tumor MDA-MB 435s cells were purchased from the Culture Center of Wuhan University (Wuhan, China), and human hepatic cancer HepG2 cells were kindly provided by the Immunology Laboratory of Tongji Medical College of Huazhong University of Science and Technology. The cells were routinely cultured in Dulbecco’s modified eagle medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS). MDA-MB 435s and HepG2 cells were transfected with plasmid DMC by So-fast transfection reagent (Sunma Biotechnology Corp., Xiamen City, China), according to the instructions of the manufacturer. After transfection, DMC-expressing cells were screened by exposure to G418 (Sigma) $800\mu \mathrm{g}∕\mathrm{ml}$ for MDA-MB 435s cells and $400\mu \mathrm{g}∕\mathrm{ml}$ for HepG2 cells, and then stable DMC-expressing cells were cloned.

2.4.

Cellular Fluorescent Imaging

The fluorescent images of DMC-expressing cells were created using an FV1000 Laser Confocal Scanning Microscope (Olympus, Japan) with a Plan Apo $60\times$ oil immersion objective with a 1.42 numerical aperture. Each fluorescence channel was detected with a PMT (photo-multiplier tube), and Fluoview software was applied to this system. The CFP signals of the DMC-expressing cells were observed by using excitation light at $458\mathrm{nm}$ and emission at 465 to $485\mathrm{nm}$ . The DsRed2 signals of DMC-expressing cells were observed by using excitation light at $514\mathrm{nm}$ and emission at 580 to $650\mathrm{nm}$ .

2.5.

Fluorescence Spectroscopy

The purification recombinant protein was incubated with $1\mu \mathrm{M}$ purified mature MMP2 (with a buffer of pH 7.5, $0.1\mu \mathrm{M}$ ${\mathrm{ZnSO}}_4$ , and $10\mathrm{mM}$ ${\mathrm{CaCl}}_2$ ). The reaction mixtures were transferred to a 1-cm cuvette of a spectrofluorometer (LS-50B, Perkin-Elmer, Norwalk, Connecticut) at $37°\mathrm{C}$ . Fluorescence emission spectra from 460 to $650\mathrm{nm}$ were recorded upon excitation at $433\mathrm{nm}$ (5-nm bandwidth).

2.6.

Western Blotting

The purification DsRed2-MSS-CFP was incubated with or without $20\mathrm{nM}$ of the mature MMP2 for $30\min$ and then immediately incubated for $10\min$ at $95°\mathrm{C}$ . The sample solutions were loaded for sodium dodecylsulfate-polyacrylamide gel electrophoresis. (SDS-PAGE) (10%). Proteins were separated in the SDS-PAGE at $120\mathrm{v}$ for 2 to $3\mathrm{h}$ and subsequently transferred onto a polyvinylidene difluoride membrane (PVDF; Amersham) with a semidry blotting system. Following blotting, the membrane was probed with an anti-GFP (green fluorescent protien) antibody (1:2000; Clontech) in Tris-buffered saline with 0.5% Tween-20 (TBST). The immunoblot was then probed with horseradish peroxidase-conjugated secondary goat anti-rabbit IgG antibody (1:5000; Bio-Rad), and the bands were detected using the ECL Western blotting analysis system (Amersham).

2.7.

Chick Embryo Chorioallantoic Membrane (CAM) Model and Tumor Fluorescent Imaging

Fertile chick eggs were incubated at saturation humidity for $72\mathrm{h}$ at $37.8°\mathrm{C}$ . The embryos were then explanted into a shell-less culture using a method adapted from our previous report.¹¹ Briefly, a 10-cm-deep concave well was made using plastic food wrap fixed with elastic bands in the mouth of a polystyrene water cup ballasted with $30\mathrm{ml}$ of water. After $72\mathrm{h}$ incubation, the shells were then cracked against the edge of a glass vessel and the contents poured into the wells. Nearly always, the contents settled with the blastodisc uppermost and central. Then the cup was sealed with food wrap and placed in an incubator ( $37°\mathrm{C}$ , 60% humidity). At day 10, $1\times {10}^6$ stable DMC-expressing MDA-MB 435s cells were explanted into CAM, and then GM6001 (Calbiochem Corp.), a kind of potent wide MMP inhibitor,¹² was add into $100\mu \mathrm{l}$ PBS solution. The solution was then added on the surface of the CAM near the tumor at a concentration of $0.4\mu \mathrm{mol}$ GM6001 per egg. As a negative control, the same number of cells was explanted into CAMs without GM6001. After three days of incubation, the fluorescence signals of the tumors were detected using a Leica MZ FL III fluorescence stereomicroscope with a mercury lamp (Leica, Germany). The DsRed2 (excitation filter $546∕12\mathrm{nm}$ ; barrier filter $560\mathrm{LP}\mathrm{nm}$ ) and CFP (excitation filter $436∕20\mathrm{nm}$ ; barrier filter $480∕40\mathrm{nm}$ ) fluorescence-imaging pictures of the tumors were recorded with a color CCD (Evolution VF, Olympus Corp.)

3.1.

In Vitro Monitoring of MMP2 Cleavage of DsRed2-MSS-CFP Fusion Protein with Spectroscopic Measurements and Western Blot Analysis

The cDNA of DsRed2-MSS-ECFP were constructed into pET-28a, which is able to express the recombinant DsRed2-MSS-CFP in E. coli BL21 (DE3) when induced with isopropyl-β-D-thiogalacto-pyranoside (IPTG) [Fig. 1a ]. Then the recombinant protein was purified through an Ni-NTA column after being induced with $1\mathrm{mM}$ IPTG. The peptide of VPLSLYSG in MSS was chosen because of its high ${\mathrm{k}}_{\mathrm{cat}}∕{\mathrm{K}}_{\mathrm{m}}$ value for MMP2.¹³ To determine the cleavage sensitivity of the fusion protein DsRed2-MSS-CFP to MMP2, we first examined it in vitro. As show in [Fig. 1b], fluorescence resonance energy transfer (FRET) from CFP to DsRed2 took place¹⁴ when the DsRed2-MSS-CFP was intact. If MMP2 cleaved the linker peptide between CFP and DsRed2, FRET would be abolished. The changes of FRET in response to MMP2 cleavage were measured with a spectrofluorometer at $37°\mathrm{C}$ . Figure 1c shows the emission spectra of the purified DsRed2-MSS-CFP fusion protein with an excitation wavelength of $433\mathrm{nm}$ . The two emission peaks of DsRed2-MSS-CFP occurred at $476\mathrm{nm}$ and $580\mathrm{nm}$ , corresponding to peaks of CFP and DsRed2, respectively. This result indicated that FRET from CFP to DsRed2 occurred and that the DsRed2-MSS-CFP fusion protein was intact. After adding $20\mathrm{nM}$ MMP2 into the solution of DsRed2-MSS-CFP, the emission peak of DsRed2-MSS-CFP at the wavelength of $580\mathrm{nm}$ disappeared $15\min$ later. Furthermore, Western blotting analysis [Fig. 1d] demonstrated that DsRed2-MSS-CFP was cleaved by MMP2. In vitro analysis indicated that DsRed2-MSS-CFP could be used as a fluorescence sensor to detect MMP2 activity. For in vitro monitoring of MMP2 cleavage of the DsRed2-MSS-CFP fusion protein with spectroscopic measurements, we have to use the FRET property of DsRed2-MSS-CFP. Otherwise, the DsRed2/CFP ratio of the integrated or cleaved DsRed2-MSS-CFP fusion protein will be the same in the solution.

Fig. 1

Schematic drawing and analysis of fusion protein DsRed2-MSS-CFP in vitro. (a) Schematic representation of the 6His-tagged DsRed2-MSS-CFP. The cDNA of DsRed2-MSS-CFP were constructed into pET-28a, which is able to express the recombinant 6His-tagged DsRed2-MSS-CFP in E.coli BL21 (DE3) when induced with IPTG. (b) Conceptual drawing of a fluorescent biosensor DsRed2-MSS-CFP for detecting MMP activity. When the DsRed2-MSS-CFP is intact, the donor CFP absorbs energy from the excitation source $\left(433\mathrm{nm}\right)$ and then nonradiatively transfers the energy to the acceptor DsRed2, which in turn emits a fluorescent photon $\left(580\mathrm{nm}\right)$ . When DsRed2-MSS-CFP is cleaved by MMPs, the fluorescence intensity of the donor’s emission wavelength will increase at $475\mathrm{nm}$ , and little fluorescence of the acceptors’ emission wavelength can be deteced at $580\mathrm{nm}$ . (c) The purification DsRed2-MSS-CFP was incubated with $20\mathrm{nM}$ of the mature MMP2 in a buffer of pH 7.5, $0.1\mu \mathrm{M}$ ${\mathrm{ZnSO}}_4$ , and $10\mathrm{mM}$ ${\mathrm{CaCl}}_2$ . The reaction mixtures were transferred to a 1-cm cuvette of a spectrofluorometer (LS-50B, Perkin-Elmer, Norwalk, Connecticut) at $37°\mathrm{C}$ . Fluorescence emission spectra from 460 to $650\mathrm{nm}$ excited at $433\mathrm{nm}$ (5-nm bandwidth) were recorded before and after reacting 5 and 15 minutes later, respectively. The spectra were normalized at $475\mathrm{nm}$ . (d) A Western blot was used to confirm that DsRed2-MSS-CFP was cleaved by MMP2. Lane 1 is a Western blot of the DsRed2-MSS-CFP in the absence of MMP2. Lane 2 is the product of DsRed2-MSS-CFP in the presence of MMP2.

3.2.

Localization of Fluorescence Sensor

MMPs are secreted as proenzymes and can degrade extracellular matrix (ECM) components once activated. To test the level of MMPs in the living cells, we chose a vector pDisplay (Invitrogen Corp.) that is a mammalian expression vector, allowing the protein under investigation to enter the secretory pathway and anchor on the cellular surface [Fig. 2a ]. Proteins expressed from pDisplay are fused at the N-terminus to the murine Ig $\kappa$ -chain signal sequence, which directs the protein to the secretory pathway, and at the C-terminus to the transmembrane domain of platelet derived growth factor receptor (PDGFR), which anchors the protein to the plasma membrane, displaying it on the extracellular side. The expressing probes anchor on the cellular surface, where fluorescence probes can be cleaved by secreted MMP [Fig. 2b]. After the DMC was cleavaged by MMPs, DsRed2 was cut off and free from the DMC-expressing cells, while CFP still remained in the cells. Thus, the MMP activity of tumor cells could be traced through monitoring the decrease of the fluorescent signal of DsRed2. To provide further support on the subcellular distribution of expressed fluorescence proteins, the vector of pDisplay-DMC was transfected into HepG2 cells with low expressing MMPs. As shown in Fig. 2c, fluorescent signals in the cellular secretory pathway and on the cellular surface were visualized in an optical section obtained by the confocal microscope.

Fig. 2

Schematic drawing and localization of fluorescence sensor DMC in living cells. (a) DMC consists of a secretion signal (SS) at the N-terminus and PDGF transmembrane (TM) domain at the C terminal. (b) Schematic representation of a part of DMC molecular localization on the cellular surface. Because the active MMP2 can cleave MSS in the DMC, the DsRed2 signal will disappear if the cell itself has highly expressed MMP2. (c) Confocal fluorescent images of DMC-expressing HepG2 cells. The CFP signal of DMC-expressing cells was observed by excitation light at $458\mathrm{nm}$ and emission at 465 to $485\mathrm{nm}$ . The DsRed2 signal of DMC-expressing cells was observed by excitation light at $514\mathrm{nm}$ and emission at 580 to $650\mathrm{nm}$ .

In Fig. 2c, the confocal fluorescent images of DMC-expressing HepG2 cells showed the most fluorescence signal of CFP and DsRed2 co-localized at the plasma membrane and the secretory pathway in the same single cells. Because HepG2 cells only low-express MMPs and were not a completely negative control, some regions in the plasma membrane at the tail end of the cell appear with only a CFP signal and no DsRed2 signal. Furthermore, DsRed2 required many hours to attain full fluorescence. CFP forms a chromophore rapidly after translation (half $\mathrm{time}=49\min$ ).¹⁵ DsRed2 takes longer to develop a coloration (half $\mathrm{time}=6.5\mathrm{h}$ ).¹⁶ If DsRed2 was used to tag the secretory protein, the fusion protein should be secreted prior to maturation for fluorescence.¹⁷ Thus, slow maturing of DsRed2 in living cells resulted in the different fluorescent intensity of CFP and DsRed2 inside the cell and at the cell surface in the same single cell, despite the fact that DMC-expressing HepG2 cells should contain equal parts of CFP and DsRed2.

3.3.

Using the DsRed2/CFP Ratio to Detect MMP Inhibitor Effects in Living Cells

The fluorescence sensor DMC was designed to trace MMP activity in living cells and small animals. When the cells excrete MMP2, the sensor DMC expressed on the surface of the cell plasma member or on the secretory pathway will be cleaved by the MMP2. Thus, DsRed2 is freed from the cells so that it is barely detected by fluorescent imaging. MDA-MB 435s cells are chosen because they highly express MMP2 and also produce MMP1, MMP3, MMP9, MMP14, MMP15, and MMP16.¹⁸ Figure 3 shows that the fluorescence signals of DMC-expressing MDA-MB 435s cells were simultaneously detected in the CFP and DsRed2 channels. MMP inhibitor GM6001 could dramatically increase the DsRed2 signal after $72\mathrm{h}$ of treatment. This indicated that GM6001 could efficiently inhibit the MMPs activity so that the new procreant DMC could be kept intact in living MDA-MB 435s cells.

Fig. 3

The fluorescence ratio images of DsRed2 to CFP reflected the effect of GM6001 inhibition on MMP activity in living cells. After $72\mathrm{h}$ treatment with $40\mu \mathrm{M}$ GM6001, cells were observed at the CFP channel and DsRed2 channel, and the images of control cells (without GM6001 treatment) also were recorded. The top row of images is the control group, and the bottom row of images is the GM6001-treated group. The fluorescence ratio images of DsRed2 to CFP were calculated with MATLAB 7.0, and pseudocolor represented the relative ratio value.

3.4.

Detecting MMP Inhibitor Effects in CAMs with DsRed2 and CFP Imaging

In vivo imaging of MMPs will be very useful for medical target assessment. In this experiment, we used the chick embryo CAM in a shell-less culture as our experimental model. This model has an advantage of allowing direct observation of fluorescence in the embryo surface without interference by the skin or other tissue. Monitoring the fluorescence signals of CFP and DsRed2 simultaneously reflected the MMP inhibitor effect during tumor growth on CAM. The fluorescent images of CAM were obtained after 3 days of treatment with or without GM6001 (Fig. 4 ). The tumor treated with GM6001 [the bottom row of Fig. 4a] yielded a brighter DsRed2 fluorescence signal than the control tumor [the top row of Fig. 4a]; however, the CFP signals of the two groups had no obvious difference. Thus, we still use the DsRed2/CFP ratio to more exactly reflect the effect of the MMP inhibitor [Fig. 4b], and the results were taken from experiments three times repeated. MMP inhibitor GM601 could obviously increase the DsRed2/CFP ratio because it inhibited the cleavage of the sensor by active MMPs.

Fig. 4

In vivo optical images for assessing the effect of MMP inhibitors. (a) MDA-MB 435s cells that expressed DMC were explanted on the shell-less chick embryo chorioallantoic membrane (CAM). The tumor was treated by MMP inhibitor GM6001 $0.4\mu \mathrm{mol}$ for 3 days, and GM6001 is absent in the control group. The MMP activity during tumor growth was reflected through monitoring the fluorescent signals of CFP and DsRed2 simultaneously. (b) The DsRed2/CFP ratio in CAMs were calculated after $0.4\mu \mathrm{mol}$ GM6001 treatment of the tumor for 3 days. Data represent three repetitions.

In this paper, we detected the MMP activity in vivo using a fluorescence stereomicroscope, a kind of optical imaging system with a larger imaging field and lower resolution than the confocal microscope. Although DMC is a FRET probe, it is difficult to detect a FRET signal in vivo by fluorescence stereomicroscope because of the low numerical aperture objective and because the light source is a mercury lamp. Furthermore, the FRET efficiency of DsRed2/CFP is much lower than YFP/CFP.¹⁰ Thus, we used the DsRed2/CFP ratio to reflect the MMP activity and did not detect the FRET efficiency of DMC.

Here, the CFP signal could be used as an internal reference to eliminate the error due to the difference of fluorescent sensor expression in various cells or the difference of imaging focus, and this could tell us where the tumor cells are if the cells highly secreted MMPs. This supposes that if we construct an expressing display vector containing only DsRed2-MSS, it also can be cleaved by MMPs. However, if DsRed2-MSS was transfected into the tumor cells with high secretory MMPs, we cannot detect an obviously fluorescent signal and we do not know whether this is due to the low transfection efficiency or high MMP activity. According to the mechanism of FRET, the fluorescence signal of CFP will increase if DMC was cleaved by MMPs. Because of this, the DsRed2/CFP ratio could generate a high contrast between its intact and cleaved components. If the cells low expressed MMPs, or if MMP activity was inhibited by an MMP inhibitor, DMC would stay intact and the cells could be detected with the fluorescence signals of both DsRed2 and CFP. The DsRed2/CFP ratio would be high. If the cells highly secrete MMPs, DMC would be cleaved, and the cells would appear with obviously decreased fluorescent signal of DsRed2 and lightly increased fluorescent signal of CFP, so the DsRed2/CFP ratio would decrease. We cannot directly compare the fluorescent signal of DsRed2 or CFP alone because the different expressing of the cells and the imaging focus will effect the fluorescent intensity of DsRed2 and CFP. Thus, the DsRed2/CFP ratio is a good approach to reflect the activity of MMPs in vivo.