2.1.

Synthesis of Cypate

The NIR fluorescent dye Cypate [Fig. 1(a) ] was prepared by using an improved method described in the literature.¹⁷ The compound was purified by flash chromatography and about $6\mathrm{g}$ (85%) of Cypate was obtained, which was further purified by recrystallization from 30% acetonitrile in water to give the desired compound in $>95\%$ HPLC purity.

2.2.

Linear Peptide Synthesis and Conjugation with Cypate

ACT APEX 396 peptide synthesizer was used to prepare the peptide by a standard fluorenylmethyl (Fmoc) protocol,¹⁸ as described previously.^{19, 20} Conjugation of the dye with the peptide at the $N$ -terminus was performed on solid support. After peptide assembly $\left(25\mu \mathrm{mol}\right)$ , the $N$ -terminal Fmoc group was removed with 20% piperidine in dimethylformamide (DMF). Cypate $\left(75\mu \mathrm{mol}\right)$ was preactivated with diisopropyl carbodiimide (DIC) $\left(125\mu \mathrm{mol}\right)$ in DMF $\left(2\mathrm{mL}\right)$ for $20\min$ and added to the resin-bound peptide. Cleavage of the conjugate from the resin and concomitant removal of the amino acid side-chain protecting groups were accomplished by adding a cleavage mixture $\left(1\mathrm{mL}\right)$ of TFA (85%), distilled water (5%), phenol (5%), and thioanisole (5%) to the resin and gently mixing the content for $3\mathrm{h}$ . The resulting green powder was purified by HPLC to give the desired compound Cypate-Gly-Arg-Asp-Ser-Pro-Lys-OH (Fig. 1(b), abbreviated Cyp-RGD; $6.4\mathrm{mg}$ , $4.84\mu \mathrm{mol}$ ) in 19% isolated yield. The molecular probe was characterized by ESI-MS, HPLC, and spectral analysis (Fig. 2 ).

Fig. 2

Absorption and fluorescence emission spectra of Cyp-GRD in 20% aqueous DMSO solution.

2.3.

Cell Culture

The human nonsmall cell carcinoma cells A549, which are positive for integrin expression,²¹ were purchased from ATCC and grown in $75\text{-}\mathrm{cm}$ tissue culture flasks or on Lab-Tek chambered slides in Ham’s F12K medium with $2\mathrm{mM}$ L-glutamine supplemented with $1.5\mathrm{g}∕\mathrm{L}$ sodium bicarbonate, 10% fetal calf serum, $100\text{units}∕\mathrm{ml}$ Penicillin, and $100\text{units}∕\mathrm{mL}$ Streptomycin.

2.4.

Cytotoxicity MTT Assay

Cytotoxicity assays were performed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) in vitro toxicology assay kit (Sigma, St. Louis, MO). Briefly, A549 cells were grown on 96 well plates to 75% confluence in phenol red free Dulbecco’s Modified Eagle’s Medium supplemented with 10% fetal calf serum, $100\text{units}∕\mathrm{mL}$ Penicillin, and $100\text{units}∕\mathrm{mL}$ Streptomycin. A concentration range of $0\text{to}125\mu \mathrm{M}$ Cyp-GRD was added to each well for a final volume of $100\mu \mathrm{L}$ /well and the cells were incubated for $24\mathrm{h}$ at $37°\mathrm{C}$ in 5% $\mathrm{C}{\mathrm{O}}_2$ . After incubation, the media was removed and replaced with $100\mu \mathrm{L}$ /well phenol red free medium. Absorbances at $570\mathrm{nm}$ and $690\mathrm{nm}$ were measured. MTT $\left(10\mu \mathrm{L}\right)$ was added to each well and the cells were incubated for $2\mathrm{h}$ . MTT solubilization solution ( $100\mu \mathrm{L}$ , formazan crystals dissolved in $0.04\mathrm{M}$ HCl in 2-propanol) was added to each well and formazan crystals were dissolved by vigorous mixing. Absorbances at $570\mathrm{nm}$ and $690\mathrm{nm}$ were measured. The $\Delta 570$ and $\Delta 690$ were determined by taking the difference of the absorbance before and after the addition of MTT. The percent of viable cells $B$ [Fig. 3(a) ] was determined according to:

Fig. 3

(a) Measurement of cytotoxicity of Cyp-GRD by the MTT assay at $24\mathrm{h}$ postincubation. (b) CyQUANT cell proliferation assay.

2.5.

Cell Proliferation Assay

Cells were seeded at a concentration of $2.5\times {10}^4$ cells/well and grown in 96 well plates in the presence of varying concentrations of Cyp-GRD $(0\text{to}5\times {10}^{-4}\mathrm{M})$ and incubated for 24 or $48\mathrm{h}$ . Cell proliferation was measured by using the CyQUANT proliferation assay kit (Molecular Probes, Eugene, OR), as described elsewhere.²² Briefly, after incubation, the microplate was inverted to remove the media and frozen at $-70°\mathrm{C}$ overnight. The microplate was thawed at room temperature and $200\mu \mathrm{L}$ of the CyQUANT GR dye in cell-lysis buffer was added to each well. After incubating the mixture for $5\min$ at room temperature in the dark, the fluorescence was measured using a Synergy HT plate reader (BioTek, Winooski, VT) at $485\text{-}\mathrm{nm}$ and $520\text{-}\mathrm{nm}$ excitation and emission wavelengths, respectively. The results were expressed as percentages of untreated control [Fig. 3(b)].

2.6.

Animal Preparation

All in vivo studies were performed in compliance with the Washington University Animal Study Committee’s requirements for the care and use of laboratory animals in research. Nude mice $\left(18\text{to}22\mathrm{g}\right)$ were anesthetized with a xylazine/ketamine cocktail via intraperitoneal injection and placed in a supine position. The mice were injected subcutaneously in the lower back of the mice with $1\times {10}^6$ A549 cells. Tumor masses were palpable at $5\text{to}7\text{days}$ postimplant, reaching $5\mathrm{mm}$ in $10\text{to}15\text{days}$ . Prior to injection of the molecular probe, the nude mice were anesthetized as described above. The mice were injected retro-orbitally with the Cyp-GRD in 20% aqueous DMSO solution $\left(100\mu \mathrm{L}\right)$ at a dosage of $0.3\mu \mathrm{mol}∕\mathrm{kg}$ body weight of the rodent.

2.7.

In vivo Optical Imaging Studies

A schematic diagram of the eXplore Optix imaging system used for this study is shown in Fig. 4 . The distribution of Cyp-GRD in the A549 tumor-bearing mice were imaged by positioning the mice on the animal plate, which was heated to $36°\mathrm{C}$ in the eXplore Optix system (ART Advanced Research Technologies Inc., Montreal, Canada). Two-dimensional regions of interest were selected with the aid of a top real-time digital camera. The height of the animal was verified by a side digital camera to ensure the animal was in the correct plane of the imaging optics. The animal was automatically moved into the imaging chamber for scanning. Laser power and count time settings were optimized at $6.71\mathrm{e}\text{-}005\mu \mathrm{W}$ and $0.3\mathrm{s}$ per point. These parameters were kept constant throughout the imaging sessions.

Fig. 4

The animal is positioned on the animal table and a region of interest is selected via a top and side live camera (vision camera). The animal is then translated into the imaging section of the system where the optics are housed. Configured in reflection geometry, the illumination source and detector point(s) are scanned over the region of interest via galvanometer mirrors. Illumination is achieved via a pulsed laser diode. Photons are detected by filtering the appropriate wavelength onto a photomultiplier tube (PMT) detector and through a time-correlated single photon counting board (TCSPC).

Excitation and emission spots were raster scanned in $1.5\text{-}\mathrm{mm}$ steps over the selected region of interest to generate emission wavelength scans. In the topographic mode, each measurement consisted of a single source-detector pair distanced by $3\mathrm{mm}$ . Thus in this mode, the number of sources and detectors (source-detector pairs configuration) are equal. There were 216 such points taken for the scans performed in this experiment. This pattern was scanned over the mouse by moving the excitation beam controlled by galvomirrors. A $780\text{-}\mathrm{nm}$ pulsed laser diode with a repetition frequency of $80\mathrm{MHz}$ and a time resolution of $250\mathrm{ps}$ (PicoQuant GmbH, Berlin, Germany) was used to excite the molecular probe. The fluorescence emission at $830\mathrm{nm}$ was collected and detected through a fast photomultiplier tube (Hammamatsu, Japan) and a time-correlated single photon counting system (Becker and Hickl GmbH, Berlin, Germany). The data were recorded as temporal point-spread functions (TPSF) and the images were reconstructed as fluorescence intensity and lifetime maps.

The time-resolved fluorescence light intensity detected at the tissue surface is not only dependent on the concentration of fluorophores but is a function of the scattering and absorption coefficients at the excitation and emission wavelengths, as well as the quantum efficiency, fluorescence absorption, and the 3D location of the probes. This is because biological tissues scatter light in the visible and NIR wavelengths. Rather complicated theoretical models have been proposed to describe such processes.^{4, 23, 24, 25} However, for a small source-detector separation, for which the average penetration depth of photons was on the order of a few millimeters and for fluorescence lifetime (on the order of ns), the tail of the TPSF is dominated by the temporal delay caused by the lifetime (Fig. 5 ). Intuitively, it can be seen that the full-width half maximum (FWHM) of the underlying TPSF, without convolution with the lifetime decay, is much smaller than the lifetime decay when photons travel distances of the order of millimeters and the lifetime is larger than $0.35\mathrm{ns}$ . This can also be quantified by using a Born approximation to the fluorescence lifetime and simulating TPSF curves.²⁶ Thus, the resulting measured curves are asymptotically log-linear dominated by the lifetime decay. An estimate of the lifetime of the fluorophore (for each measurement point) is then obtained by measuring the decay of the TPSF and fitting the tail to the form:

Fig. 5

(a) A typical TPSF curve of Cyp-GRD from a tumor region in a nude mouse at $24\mathrm{h}$ postinjection. (b) Fluorescence lifetime distribution of Cyp-GRD in a nude mouse at $24\mathrm{h}$ postinjection. The large signal at $0\mathrm{ns}$ was obtained from tissues devoid of the probe.

The effect of the system impulse function has been reported elsewhere,²⁶ and was shown to have a negligible effect in that range of fluorophore lifetimes. Thus, the map of lifetime images was obtained by simply fitting the log-linear tail of the TPSF and obtaining the characteristic time at each point.

The in vivo fluorescence intensity and lifetime of the molecular probe were extracted from the TPSF using different properties of the accumulated curve [Fig. 5(a)]. Since the distances traveled by light were small in this case, the lifetime of Cyp-GRD was fully expressed by the decay of the curve at longer times. A histogram representing the distribution of the lifetimes in the image is shown in Fig. 5(b). To obtain this histogram, all of the zero lifetime regions, representing points with no signal from the molecular probe, were discarded. The fluorescence lifetime map was constructed based on the mono-exponential decay [Fig. 5(b)].

TPSFs were also used to reconstruct the 3D concentration of the probe.²⁷ Reconstruction was performed by discretizing the convolution expression of photon trajectories and lifetime delay and using the temporal moments to create a linear relation between the concentration of the fluorophore and the measurements. The photon trajectories here are modeled by taking into account the first order interaction between the light and the fluorophore (Born approximation). An expression for the Laplace transform of the detected light at position ${\vec{r}}_d$ given a delta excitation at position ${\vec{r}}_s$ is

Inversion was performed by minimizing the least square distance between the measurements and the model. The first two moments were used for inversion. To obtain unit consistency, the first moment was renormalized by the intensity in the inversion process. Details of the mathematics and procedures are described by Lam ²⁷ This approach avoided the necessity to compute an instrument calibration factor, with the drawback that a relative concentration was computed.

3.1.

Tumor-Specific Molecular Probe (Cyp-GRD)

Indocyanine green (ICG) is the only NIR fluorescent dye currently approved for use in humans by the US Food and Drug Administration. For this reason, it is widely used in contrast agent-mediated NIR optical imaging of human patients.^{28, 29} Unfortunately, ICG is a nonspecific contrast agent and does not possess reactive functional groups needed to conjugate it with biological carriers to enhance target specificity in vivo. These limitations have led to the development of reactive NIR fluorescent dyes, such as Cypate and other carbocyanines for labeling biomolecules. Cypate has similar spectral and biodistribution properties similar to ICG.^{20, 30} Cypate is primarily excreted by the hepatobiliary system and requires conjugation to a disease-specific carrier to increase the specificity of the molecular probe. Recently, we found that a Cypate-labeled hexapeptide, abbreviated Cyp-RGD (Fig. 1), was preferentially retained in ${\alpha }_{\mathrm{v}}{\beta }_3$ integrin receptor-positive tumor (A549).³¹ Its retention in this tumor can be inhibited by RGD peptide ligands known to bind with high affinity to this receptor. In contrast to the widely used cyclic RGD peptides, which require an additional cyclization step to obtain the peptide carrier, the hexapeptide used to prepare Cyp-GRD is linear, which facilitates the complete preparation of the tumor-specific molecular probe on solid support. Figure 2 shows that the spectral properties of Cyp-GRD $({\lambda }_{\max }^{ab}785\mathrm{nm}$ , ${\lambda }_{\max }^{em}817\mathrm{nm}$ , ${\epsilon }_{\max }^{ab}\mathrm{237,680}{\mathrm{cm}}^{-1}{\mathrm{Mol}}^{-1})$ in 20% aqueous dimethyl sulfoxide (DMSO) are similar to those of the nonconjugated dye Cypate $({\lambda }_{\max }^{ab}786\mathrm{nm}$ , ${\lambda }_{\max }^{em}811\mathrm{nm}$ , ${\epsilon }_{\max }^{ab}\mathrm{244,400}{\mathrm{cm}}^{-1}{\mathrm{Mol}}^{-1})$ , which has been used to prepare several biocompatible optical molecular probes.^{19, 20} Previous studies have shown that aqueous DMSO solution boosted fluorescence intensity and was well tolerated by rodents.^{20, 30} The spectral widow of Cyp-GRD allows its use for optical imaging with the fixed wavelength time-domain DOT system. Therefore, Cyp-GRD is attractive for NIR fluorescence lifetime imaging study because of the ease of synthesis, high uptake by the human lung epithelial carcinoma in mice, and suitable spectral properties.

Fig. 1

Structures of (a) Cypate and (b) Cyp-GRD.

3.2.

Cytotoxicity and Cell Proliferation Effects of Cyp-GRD

Molecular contrast agents for in vivo imaging should enhance the localization and assessment of the functional status of diseases without interfering with the inherent cell growth and functions. Cytotoxicity and proliferation tests provide key information on the suitability of molecular probes for in vivo applications. It was necessary to assess these parameters because Cyp-GRD is a new molecular probe. The assays were analyzed at $24\mathrm{h}$ and $48\mathrm{h}$ because we typically complete our in vivo optical imaging studies within $36\mathrm{h}$ postinjection of the molecular probe. The MTT test, which determines cell viability, measures the metabolic activity of mitochondrial dehydrogenases.³² Incubation of varying concentrations of Cyp-GRD with A549 cells for $24\mathrm{h}$ and subsequent MTT assay showed that the molecular probe was not cytotoxic to these cells up to $1\times {10}^{-4}\mathrm{M}$ [Fig. 3(a)]. Similarly, the CyQUANT cell proliferation assay, which provides cell density information by quantifying the nucleic acids present in a given cell population, showed that Cyp-GRD did not induce cell proliferation at $24\mathrm{h}$ post-incubation [Fig. 3(b)]. A transient increase in cell density was observed between $1\times {10}^{-11}$ and $1\times {10}^{-10}\mathrm{M}$ after incubating A549 cells with Cyp-GRD for $48\mathrm{h}$ . This increase returned to normalcy at $1\times {10}^{-8}\mathrm{M}$ of the probe. Because most imaging studies are expected to be completed within $24\mathrm{h}$ after administering the probe to living organisms, the MTT and CyQUANT assays demonstrate the biocompatibility of Cyp-GRD for in vivo imaging. The results also suggest that a wide range of Cyp-GRD’s concentrations are safe for use in animals.

3.3.

Fluorescence Lifetime and Intensity Imaging of Cyp-GRD Distribution in Mice

Accurate measurement of the distribution of Cyp-GRD in animals by optical methods can be used to localize integrin receptor-positive tumors and characterize tumor heterogeneity. We used a time-domain optical imaging system for this study (Fig. 4). A typical in vivo TPSF curve of Cyp-GRD shown in Fig. 5(a) was used to obtain the fluorescence intensity and lifetime maps. A histogram representing the distribution of the lifetimes in the image is shown in Fig. 5(b). Variations of this distribution were observed at different regions of the mice.

The fluorescence lifetime image (FLI) of the whole mouse at $24\mathrm{h}$ postinjection showed that the molecular probe selectively localized in the target tumor tissue and in the primary excretion organs, the liver and kidneys (Fig. 6 ). The FLI also delineated the liver from the left kidney but not the right kidney. This image was in agreement with mouse anatomy because the right kidney is closer to the liver than the left kidney. Note that the higher values for lifetime in the left part of the liver were due to noisy signals (lower fluorescence) in that region. Points with inadequate fluorescence to evaluate lifetime were not represented in Fig. 6.

Fig. 6

NIR fluorescence lifetime image of Cyp-GRD distribution in an A549-tumor-bearing mouse at $24\mathrm{h}$ postinjection.

The mean lifetime average of the tumor over eight localized scan points was $1.03\mathrm{ns}$ , with a minimum $\tau$ of $0.76\mathrm{ns}$ and a maximum $\tau$ of $1.5\mathrm{ns}$ . However, the mean lifetime average of the liver over eight scan points was $0.80\mathrm{ns}$ , with a minimum $\tau$ of $0.74\mathrm{ns}$ and a maximum $\tau$ of $0.85\mathrm{ns}$ . The tumor had a higher lifetime average, which can be related to higher pH or hypoxic status.³³ A previous study had shown that lifetime of protoporphyrin IX increased from $5\mathrm{ns}$ in normal tissue to $18\mathrm{ns}$ in neoplastic tissue.⁸ Therefore, lifetime maps can be used to delineate tumor tissues in vivo.

A similar FLI biodistribution pattern was obtained by fluorescence intensity imaging (Fig. 7 ). The molecular probe preferentially localized in the tumor tissue, the kidneys (the major excretion organ), and to some extent the liver. The ex vivo image of the organ parts showed that the amount of Cyp-GRD in the liver was low. Optimization of the molecular probe design can lower the amount of the fluorescent probe retained by the kidneys after $24\mathrm{h}$ of administering the contrast agent.

Fig. 7

Fluorescence intensity image of the distribution of Cyp-GRD in an A549 tumor-bearing mouse at $24\mathrm{h}$ postinjection. Note that the tumor has a significant fluorescence signal.

Figures 8 and 9 were constructed by showing the intensity either in slices cut along the axial axis or through a ray-casting method on the volume. The main advantage of using moments for the inversion was the possibility to deconvolve additively the effects of the impulse response function (IRF) as well as the lifetime decay. To compare tumor and normal tissues, the images were represented as 2D slices in the $Z$ axis, starting at $Z=0\mathrm{mm}$ (dorsal skin of the mouse) to about $7\mathrm{mm}$ deep (Fig. 8).^{27, 34} The images gave detailed information on the distribution of the molecular probe in the tumor relative to surrounding normal tissues. Particularly, the ratio of the tumor-to-kidney or liver increased in the $Z$ direction, with the largest contrast observed between 3 and $6\mathrm{mm}$ deep. Table 1 summarizes the depth recovery parametric information obtained by taking the maximal voxel of the volumetric reconstruction for each organ.

Fig. 9

Volumetric representation through ray-casting of the volume of the tumor and major excretion organs.

Fig. 8

2D slices of the image from Fig. 6 reconstruction in the $Z$ direction.

Table 1

Parametric information from the reconstruction, the maximal voxel in the region under consideration.

A 3D volumetric representation of the probe distribution in the tumor and major excretion organs is shown in Fig. 9. This volumetric rendition of the biodistribution was obtained through ray-casting, which is similar to a maximum intensity projection.