|Home | About | Journals | Submit | Contact Us | Français|
Proteases play important roles in a variety of pathologies from heart disease to cancer. Quantitative measurement of protease activity is possible using a novel spectrally matched dual fluorophore probe and a small animal lifetime imager. The recorded fluorescence from an activatable fluorophore, one that changes its fluorescent amplitude after biological target interaction, is also influenced by other factors including imaging probe delivery and optical tissue absorption of excitation and emission light.
Fluorescence from a second spectrally matched constant (non-activatable) fluorophore on each nanoparticle platform can be used to correct for both probe delivery and tissue absorption. The fluorescence from each fluorophore is separated using fluorescence lifetime methods.
Proteases, enzymes that break down proteins, play an important role in a variety of pathologies including cancer and cardiovascular disease , the two leading causes of death the United States . In cancer, high levels of protease activity are associated with invasive and metastatic cancers . In cardiovascular disease, proteases are active in apoptosis, tissue remodeling, and modification of cardiac proteins .
Protease activity has been used to visualize these diseases with protease activatable probes. Because these probes are injected in a quenched state and activated in vivo, the background signal from nonspecific probe uptake is minimized for accurate measurements of low levels of activity . This makes it possible to accurately delineate tumor margins in vivo and quantify that the protease activity in tumors is more than ten times the level in healthy tissue using ex vivo samples . Imaging of protease activity in cancer can be used as a screening tool or for measuring treatment efficacy . Atherosclerosis treatment decisions may be made based on in vivo evaluation of plaques by fluorescence angioscopy .
While these probes are useful for locating lesions, the measured fluorescence cannot be used to quantify protease activity because the measured fluorescence is a combination of probe delivery, probe activation, and optical tissue absorption. Probe activation can be measured independently by using a dual fluorophore probe. In these probes, a constitutively active fluorophore is used to measure probe distribution and optical absorption. An activatable fluorophore measures the combined effect of distribution, optical absorption, and activation. The percentage of activation is the ratio of the two signals , since the other terms cancel out due to a common carrier platform.
The accuracy of the dual fluorophore technique is limited by the matching of optical absorption of tissue at the excitation and emission wavelengths of the two fluorophores. The quantitative protease activity measurement system presented in this paper guarantees that optical absorption cancels out exactly in ratiometric imaging. The system, shown in Fig. 1, consists of the two parts: a nanoparticle imaging probe and an imager. The probe is small enough to travel throughout the animal and interact with biomolecules. Spectrally matched fluorophores are used to report distribution and activation. The imager measures fluorescence from the two fluorophores using the property of fluorescence lifetime by taking nanosecond images of a large imaging area.
Fluorescence lifetime imaging is an instrumentation challenge because the fluorescence lifetimes of fluorophores suitable to in vivo imaging varying from hundreds of picoseconds to a few nanoseconds. The Small Animal Lifetime Imager (SALI) captures images for measuring these lifetimes and the fluorescence from multiple fluorophores by combining nanosecond excitation pulses with nanosecond exposures. Two of these nanosecond images are sufficient to calculate the lifetime of a single exponential decay or the fluorescence from two fluorophores with known lifetimes .
The small animal lifetime imager (SALI) captures fluorescence lifetime images of in vitro and in vivo samples. This system comprises three parts: a nanosecond illuminator, imaging optics, and a commercial nanosecond imager as shown in Fig. 2.
The nanosecond illuminator is based on an electronically pulsed laser diode . A custom designed printed circuit board houses the timing circuitry. Inverters A, B, and C form an oscillator with the 47 kΩ resistor and 10 pF capacitor setting the frequency to 2 MHz. The 5.6 kΩ resistor and inverters D and E introduce the delay required between triggering the nanosecond imager and exciting the fluorophores. The 10pF capacitor and the 560 Ω resistor transform the square wave into a pulse train. Inverter F and the power mosfet switch the 658nm laser diode. The laser beam is diffused to illuminate the entire sample by a square-pattern engineered light diffuser (Thorlabs ED1-S20.) The size of the illumination area is then adjusted by the placement of the sample with respect to the illuminator.
A commercial gated-optical-intensifier camera, the Picostar HR (LaVision), captures the nanosecond exposures when triggered by an electrical signal from the illuminator. A programmable delay in the Picostar HR varies the delay between excitation and exposure across a 20 ns range. The fluorescence light first passes through a chromatically corrected lens assembly for focusing and aperture control (Schneider 17mm/f1.4 Xenoplan.) The light is then infinity focused by a 40 mm achromatic lens (Edmund Optics.) A six-position motorized filter wheel (Thorlabs FW102B) selects the wavelengths with a 680 nm-710 nm bandpass filter (Omega Filters 3RD680-710) used for far-red imaging. The image is then focused using another 40 mm achromatic lens (Edmund Optics) on the Picostar HR.
Operation of the SALI was first verified by imaging a reflective white surface with the filter wheel turned to an empty slot. In this configuration, the excitation beam is imaged. Images were taken at delays at one-nanosecond intervals with respect to the excitation pulse. The mean signal intensity was then measured within a region of interest in the illumination area using Matlab (Mathworks.) This measured pulse waveform is plotted in a sold line in Fig. 3. The pulse is 10ns wide with both on and off transients faster than one order of magnitude per nanosecond, sufficient for measuring single nanosecond fluorescent lifetimes.
Two fluorophores with published lifetimes were then imaged with the bandpass filter replaced. ATTO 647 and ATTO 647N are far-red fluorophores from ATTO Tec. The fluorophores were placed in black 384-well plates. As with the excitation pulse, images were taken at one-nanosecond delay intervals and the mean intensity calculated. Plotting the intensity on a log scale versus time on a linear scale as in Fig. 3 creates a straight line from the single exponential decay as seen in both fluorophores. The slope of the line is the fluorescent lifetime. The lifetimes of 2.5ns and 3.8ns agree well with the manufacturer’s reported values.
The SALI was then used to measure unknown fluorescent lifetimes, the fluorescent lifetimes of an activatable probe before and after activation. This activatable probe is a member of a class of enzyme-activatable fluorescence probes using an iron-oxide nanoparticle, cleavable peptide linker, and fluorophore . The iron-oxide nanoparticle is used as a quencher and nanoparticle platform. The peptide linker is chosen for enzyme specificity. The fluorophore is chosen for fluorescence properties, generally excitation and emission wavelengths. In this work, a long fluorescence lifetime fluorophore was used to investigate changes in fluorescence lifetime during activation. A poly-arginine linker was used because of its susceptibility to trypsin cleavage .
This single fluorophore activatable probe in Fig. 4(A) was synthesized by first making CLIO nanoparticles . A mixture of ferrous and ferric iron chlorides was dissolved with T-10 dextran. The pH was then modified with ammonia to form the particles. The resulting material was further reacted with epichlorohydrin to crosslink the protective dextran coating, followed with ammonia to provide amine attachment points.
The cleavage peptide (GRRRRGC) (Tufts University Peptide Core Facility) was labeled with ATTO 647N by reacting 1 mg of the lyophilized peptide with 100 μL of ATTO-647N-NHS ester (1 mg/200 μL DMSO) overnight, using 2 μL of diisopropylethyl amine as a catalyst. After the reaction, 1 mL of PBS and 10 μL of 0.5M TCEP (tris(t-carboxyethyl) phosphine) solution were added to form the labeled peptide stock solution.
The labeled cleavage peptide was attached to CLIO by first activating 10 mg of CLIO with 9 mg of SIA (succinimidyl iodoacetate) in 100 μL of DMSO. After 15 minutes at room temperature, the CLIO was purified by size exclusion chromatography (sephadex G-25 resin) in PBS. Activated CLIO was then mixed with 450 μL of the labeled cleavage peptide. This was allowed to react overnight at room temperature, and again purified by size exclusion chromatography (sephadex G-50 resin) in PBS.
As with the free dye, the lifetime of this probe was measured by imaging 50 μL of probe in a black 384-well plate. Trypsin was added to the wells to create concentrations ranging from 0 to 6 μM. The SALI was then used to measure fluorescence intensity in each 1 ns window from 0 to 20 ns with laser excitation between 2 ns and 9 ns. The linear plot in Fig. 4(B) demonstrates four-fold activation with increasing concentrations of trypsin. Plotting the same data with a log scale for fluorescence intensity demonstrates that the fluorescence lifetime, the slope of the straight-line decay in this plot, is unchanged during activation. A pure exponential decay with a 3.9 ns time constant is plotted in Fig. 4(C) for comparison to probes.
Because the lifetime is unchanged with activation, it is possible to apply the recently developed technique of fluorescent lifetime multiplexing to activatable probes. Fluorescence lifetime imaging was first applied to in vivo imaging to generate contrast between normal and diseased tissue [12–14]. It was then applied to exogenous fluorophores to increase the signal-to-background ratio . Recently, a new application has been developed, fluorescent lifetime multiplexing, the independent measurement of fluorescence from multiple fluorophores in a single spectral window [16–17].
A variety of methods have been used for fluorescent lifetime multiplexing. In this work the two-window method is used . The fluorescence is measured in two time windows of equal length, Fwin1 and Fwin2. The time between the start of the first window and the start of the second window is twin. Equations for the fluorescence in each window are written as the sum of fluorescence with fluorescence lifetimes of τ1 and τ2 and relative amplitudes A1 and A2 respectively.
The two equations are then solved for A1 and A2.
This multiplexing method was verified using free dyes in the SALI. The spectrally matched fluorophores measured independently for Fig. 3, ATTO 647 and ATTO 647N, were used for the multiplexing experiment. First, it was determined that fluorescence intensity varied linearly with fluorophore concentration in the range from 0.5 μM to 4.5 μM for both fluorophores when 50 μL of fluorophore solution is imaged in a standard a black 384-well plate with the SALI. Below this range, the fluorescence signal was indistinguishable from noise. Above this range, fluorescence increases more slowly, likely due to significant absorption of excitation and emission light.
Then, a phantom for testing multiplexing was developed using a four-by-four grid of wells in a 384-well plate. Each well contained a mixture of the two fluorophores, as shown in Fig. 5. Each row had a constant concentration of ATTO 647 increasing from no fluorophore in the bottom row to 3 micromolar in the top row. Each column had a constant concentration of ATTO 647N increasing from no fluorophore in the right column to 3 micromolar in the left column. Because these fluorophore intensities are within the linear range, the fluorescence from each fluorophore should vary with fluorophore concentration.
The well plate was imaged in the SALI at 2 ns and 11 ns following excitation. The images in Fig. 5 demonstrate that the 2 ns image shows signal from both fluorophores, while the contribution from the short lifetime fluorophore, ATTO 647, is decreased in the 11 ns image. Applying Eqns. (3) and (4) to each pixel independently in the measured images produced the calculated images of fluorescence from the two fluorophores. The agreement between the calculated fluorescence and fluorophore concentration demonstrates the success of this technique.
The two technologies of activation and fluorescent lifetime multiplexing offer even greater possibilities when used with a dual fluorophore probe. Dual fluorophore probes make it possible to separate the activation signal from probe delivery and optical tissue absorption. The operation of dual fluorophore probes can be seen by considering the variables effecting in vivo fluorescence intensity. The fluorescence that reaches an imager from an in vivo fluorophore is given by Eq. (5) with geometric terms independent of wavelength and quantum yield included in the constant k. The illumination photons, I(λex), pass through tissue where a fraction of the photons, ab(λex), are absorbed before they interact with the fluorophore. The light that reaches the fluorophore excites a number of fluorophores proportionate to the number of fluorophores, n, and the extinction coefficient, ε(λex). The quantum yield, (x), which may be a function of activation, determines the percentage of excited fluorophores that emit photons. A fraction, ab(λem), of these emitted photons are then absorbed by the tissue.
Activation of activatable probes changes the quantum yield, (x). Unfortunately, variations in the other variables also affect the signal. Uneven tissue delivery of the nanoparticles changes n. Different tissue absorption due to either depth of probe or tissue type can change ab(λ). Dual fluorophore probes attempt to solve this problem by taking the ratio of two fluorescence intensities, Fact and Fconst.
This technique works well when R is relatively constant. The values of k, n, and ε(λex) are kept at a fairly constant ratio, but the absorbance terms, ab(λ), can lead to errors. The small amount of tissue between the illumination source and shallow fluorophores only creates a small variation between different wavelengths of light. The larger section of tissue between the illumination source and deeper fluorophores creates a larger wavelength dependent variation in excitation light. The cause of this variation can be seen by considering a homogenous sample. Here ab(λ) is given by Beer’s law in Eqn. (8), where α(λ) is the absorption coefficient. While the ratio of absorption coefficients is constant for the homogeneous sample, changes in position vary l changing the ratio of absorbance.
Fluorescent lifetime multiplexing solves this problem by forcing the excitation and emission wavelengths to be equal. In this case the absorbance terms cancel out exactly.
A comparison of spectrally separated ratiometric fluorescence imaging to lifetime separated ratiometric fluorescence imaging was performed using a tissue phantom and mixtures of free fluorophores. For spectral separation, Cy3.5 and ATTO 647 were used. For lifetime separation, ATTO 647 and ATTO 647N were used.
The tissue phantom was fabricated using murine blood mixed with agarose to mimic optical absorption of tissue. 600 μL of murine blood was combined with 16 mL of heated 2.5 % agarose (Sigma-Aldrich) in PBS along with an additional 3 mL of PBS, and molded in 100 mm Petri dishes (Fisher Scientific.)
The spectrally separated fluorophores were prepared by placing 25 uL of 2 uM Cy3.5 into each of 9 wells in a black 384 well plate. Then 25uL of 0, 2uM, and 4uM ATTO 647 were added to 3 wells each with the Cy3.5 forming fluorophore ratios of 0, 1, and 2. Two pieces of the tissue phantom were then placed atop the wells to create thicknesses of 2mm and 4mm as shown in Fig. 6(A). The fluorescence lifetime separated fluorophores were prepared in the same manner with each well containing 1uM ATTO 647N and varying concentrations of ATTO 647.
The fluorescence in each well was then measured in the SALI. The fluorescence-lifetime-separated fluorescence was measured using the fluorescence demultiplexing method in section 3. The spectrally separated fluorescence was measured using SALI as a steady-state fluorescence imager. A halogen illuminator (Edmund Optics) was used with a 550 nm to 580nm interference filter to excite the Cy 3.5. A 600nm to 650nm bandpass filter was used in the imaging path. The Picostar HR took 1ns exposure images at approximately 1 MHz rate independent of the illumination. The ATTO 647 fluorescence was measured with the same illumination and filtering as used in the fluorescence-lifetime-separated fluorescence measurement. The delay between excitation and imaging was set to maximize fluorescence intensity.
The fluorescence intensities for spectrally separated and fluorescence lifetime experiments are plotted in Fig. 6(B) and Fig. 6(C) respectively. In both cases, measured fluorescence from both fluorophores decreases with increased phantom thickness for the same fluorophore ratio, and the ATTO 647 fluorescence increases with increasing fluorophore ratio (i.e. increasing ATTO 647 concentration) for a given tissue phantom thickness. The difference between the spectrally separated and fluorescence lifetime experiments is in the ratios of the fluorescence. The measured fluorescence from Cy3.5 decreases more quickly with increasing phantom thickness than the ATTO 647 fluorescence. This difference in tissue absorbance at the different wavelengths causes the difference in measured fluorescence for the same fluorophore ratio as shown in Fig. 6(D). In contradistinction, the measured fluorescence ratios match the fluorophore ratios for the fluorescence lifetime separated fluorophores because the tissue absorption is matched as shown in Fig. 6(E).
The spectrally matched fluorophores were then used to synthesize a dual fluorophore fluorescent lifetime activatable probe (FLAP). The CLIO particles were synthesized as described in section 3. After creation of amine attachment points in that protocol, 10 mg of CLIO (Fe basis) dissolved in 2.1 mL PBS was reacted with 60 μL of ATTO 647-NHS ester (1 mg/300 μL in DMSO) and allowed to sit overnight at room temperature. An aliquot was purified by size exclusion chromatography (sephadex G-50 resin) to characterize the amount of dye directly bound to the nanoparticle. The cleavage peptide was then labeled with ATTO 647N and attached to the CLIO as before. The iron concentration and total dye concentration of both this conjugate and the aliquot of the ATTO-647-CLIO conjugate were measured by UV-Vis spectroscopy. The final product contains 1.97 mg Fe/mL, 17 μM ATTO 647, and 12 μM ATTO 647N.
Probe activation was measured using prepared concentrations of trypsin of 1, 2, 3, 4, and 5 μg/mL solutions. 10 μL of probe solution was placed in each of six wells in a black 384-well plate. 40 μL of PBS/trypsin solution was added to each well varying from pure PBS to 5 μg/mL trypsin. After allowing ten minutes for the trypsin to cleave the poly-arginine linker, the wells were imaged in the SALI. Fluorescent multiplexing analysis was performed to separate out the fluorescence signals from ATTO 647 and ATTO 647N. As shown in Fig. 7, the demultiplexing clearly separated a constant signal from a signal with only a 0.16 correlation to trypsin concentration from an activatable signal with a 0.92 correlation to trypsin concentration.
Protease activity can be quantified using a new spectrally matched dual fluorophore nanoparticle probe with a small-animal lifetime imager. Because the fluorescent lifetime is unchanged during activation of the CLIO based activatable probes used in this work, it is possible to separate fluorescence from spectrally matched fluorophores on a dual fluorophore probe. The use of spectrally matched fluorophores reduces the effect of tissue absorption on fluorescence ratios by 75 percent. The challenge of separating fluorescence from these fluorophores based on nanosecond time scale emission was solved by developing a small animal lifetime imager.
This system is now ready for in vivo imaging. Far-red fluorophores were chosen for optical transparency of tissue. Single fluorophore protease activatable probes based on similar technology have been previously used for in vivo imaging [7–8]. The SALI may now be used to improve the accuracy of these in vivo experiments characterizing tumors and atherosclerotic plaques.
This work was supported by NIH grants R01EB001872 “In Vivo Multichannel Fluorescence Imaging” and T32EB002102 “Postgraduate Program in Radiological Sciences”.
Christopher D. Salthouse received S.B., M.Eng., and Ph.D. degrees in electrical engineering from the Massachusetts Institute of Technology. During his doctoral research with Dr. Sarpeshkar at MIT, he developed micropower analog integrated circuits to improve cochlear implants. He is currently a research fellow at Massachusetts General Hospital developing new optical imaging technologies for biomedical applications.
Fred Reynolds received his B.S. degree in chemical engineering from The Cooper Union, and his PhD degree from Clarkson University. His thesis work under Dr. Yuzhou Li involved probing the polymerization mechanism of polystyrene nanoparticles. He is presently a staff scientist at Massachusetts General Hospital, providing chemical support and developing research tools for the imaging lab.
Lee Josephson received his BS in Chemistry from the University of Wisconsin, his PhD in Biochemistry from the State University of New York at Stony Brook and did a Postdoctoral fellowship at Harvard University in the laboratory of Dr. Guido Guidotti. His research interests include magnetic nanoparticles, fluorescent probe design, and the pharmacology of the viridin class of natural products of which wortmannin is a member. He is currently an Associate Professor at Harvard Medical School.
Umar Mahmood received his B.S. degree in chemistry from the California Institute of Technology and M.D. and Ph.D. degrees from Cornell University. His doctoral and postdoctoral work in biophysics were performed at Memorial Sloan Kettering Cancer Center and focused on tumor physiology studies using 31P NMR spectroscopy. His medical residency training was in Radiology at the Massachusetts General Hospital (MGH), Boston.
He is currently Associate Professor of Radiology, Harvard Medical School, and Associate Director of the Nuclear Medicine and Molecular Imaging Division at the MGH. His work focuses on the development of new technologies and applications across the electromagnetic spectrum for improved disease detection and disease characterization in people.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.