2.1.

Materials

Core-shell QDs ( $\mathrm{ZnS}$ -capped $\mathrm{CdSe}$ , Ref. 15) were synthesized by College of Chemistry & Molecular Sciences, Wuhan University. Human hepatocellular carcinoma cell lines¹⁸ (HCCLM6) were provided by Liver Cancer Institute of Fudan University. EDC (1-(3-Dimethyl-aminopropyl)-3-ethylcarbodiimide hydrochloride, $>98\%$ ) and NHS ( $N$ -hydroxysuccinimide, $>98\%$ ) were purchased from Acros-Organics. Mouse anti-human monoclonal AFP antibody and RPMI-1640 culture medium were purchased from Sigma.

2.2.

QDs Bioconjugation and Animal Preparations

We used a $1.5\text{-}\mathrm{ml}$ solution of high-quality oil-soluble core-shell QD590 to synthesize water-soluble QDs according to a developed procedure.¹⁵ These activated dots modified with thioglycolic acid were dissolved in phosphate-buffered saline (PBS) ( $0.08\mathrm{mol}∕\mathrm{l}$ , pH 7.4) containing $50\mathrm{mmol}$ EDC and $5\mathrm{mmol}$ NHS. Then, the QDs were reacted with $20\mathrm{\mu }\mathrm{l}$ mouse anti-human monoclonal AFP antibodies at room temperature in a shaking incubator for 2 to $4\mathrm{h}$ . The final QD bioconjugates were purified by centrifugation at $6000\mathrm{g}$ for $10\min$ , and the suspension was dialyzed for 8 to $12\mathrm{h}$ . The QDs-Anti-AFP thus afforded were stored in refrigerator at $4°\mathrm{C}$ .

Using protocols approved by the Institutional Animal Care and Use Committee of Wuhan University, $\sim 10$ ⁶ prepared HCCLM6 cells were injected into a 6- to 8-week-old nude mouse subcutaneously. Tumor growth was monitored daily until it reached the acceptable size (0.5 to $1\mathrm{cm}$ in diameter). Then, the synthesized QDs-Anti-AFP probes were injected into the tail vein at $0.4\mathrm{nmol}$ for active targeting (four times). The time interval between each injection was about $24\mathrm{h}$ . The mice were placed under anesthesia by injection of 3% Nembutal at a dosage of $45\mathrm{mg}∕\mathrm{kg}$ for in vivo studies.

Some other mice bearing tumors of similar sizes (0.5 to $1\mathrm{cm}$ in diameter) were prepared under the same protocols for histological examination. The same dose $\left(0.4\mathrm{nmol}\right)$ of QDs and QDs-Anti-AFP were injected into different mice for passive and active targeting. After a 24-h probe circulation, the mice were killed and tumors and major organs were removed and frozen. Tissue collections were cryosectioned into sections ( $10\mathrm{\mu }\mathrm{m}$ thick), fixed with acetone at $0°\mathrm{C}$ , and examined with a confocal fluorescence microscope (Leica TCS-SP2-AOBS-MP, D-68165, Mannheim, Germany).

2.3.

Fluorescence Detection and Imaging

The fluorescence imaging system is shown in Fig. 1A . The pulsed laser was provided by a Ti:sapphire laser (Mira 900, Coherent). Samples were excited at the wavelength of $400\mathrm{nm}$ using an optical frequency-doubling system. The beam was focused on samples with a lens. The fluorescence over the entire emission range was collected by a liquid-nitrogen-cooled CCD-array spectrometer (ACTON, Spectrapro 2500I) along with imaging acquisition and analysis software. The following samples were prepared for fluorescence detection: (1) QDs and QDs-Anti-AFP and (2) QDs-tagged tumor tissue and normal tissue in a nude mouse [Fig. 1A].

Fig. 1

(A) Simple optical layout of fluorescence imaging system: A1, optical frequency double system; A2, lens; A3, liquid-nitrogen cooled CCD-array spectrophotometer. (B) and (C) Spectroscopic imaging of (B) QDs-tagged tumor tissue and (C) normal tissue. (D) and (E) Emission spectrum of (D) QDs-tagged tumor tissue and (E) normal tissue. All the images and the spectrums were captured from a nude mouse with the same exposure time.

2.4.

Site-by-Site Measurement for Distribution Analysis

We selected 2500 sampling sites covering a square area from the periphery to the central area in the tumor of a nude mouse, and each side of the square consists of 50 equidistant sites. An $X\text{-}Y$ axis translation stage was designed to shift the QD-labeled mouse automatically [Fig. 1A]. When the fluorescence emission of one site was collected, the stage was shifted $0.1\mathrm{mm}$ along one axis to focus the laser on the next site. After 2500 continuous collections, all the recorded emission peak intensities were processed by software to form a contour map (Fig. 2 ) to indicate the fluorescence intensity distribution of the detected area.

Fig. 2

Fluorescence intensity contour map of a detected cancerous area in the QDs-tagged tumor. Although all the sampling sites are in the tumor, their fluorescence intensities are different: cold colors represent lower fluorescence intensity in the area, while warm colors represent higher intensity. The detected cancerous area is from the periphery (corresponding to the lower left corner in the map) to the central area (corresponding to the upper right corner in the map) of the tumor. Thus, the map exhibits the inhomogeneous distribution of the fluorescence intensities in the detected area of the tumor.

3.1.

Spectral Characteristic

The QDs and the QDs-Anti-AFP probes have the same narrow and symmetric emission spectrums (Fig. 3 ). The corresponding emission peaks are at $590\mathrm{nm}$ and half-maximum values are only $20\mathrm{nm}$ . The QDs-Anti-AFP probes used in this study exhibit the same excellent spectral property as the original QDs. Note that the exposure times of QDs $\left(8\mathrm{ms}\right)$ and QDs-Anti-AFP $\left(10\mathrm{ms}\right)$ were different for spectral matching.

Fig. 3

Fluorescence spectral comparison between QDs and QDs-Anti-AFP.

3.2.

In vivo Cancer Detection

To investigate the specific binding of QDs-Anti-AFP to hepatoma in vivo, we used QDs-Anti-AFP to target AFP for active targeting.^{2, 19} AFP is the main component of mammalian fetal serum, which is synthesized by visceral endoderm of yolk sac and by fetal liver cells. The alteration of AFP blood level is an important marker for hepatocellular carcinoma, so that the accumulation and retention of AFP at the site of tumor is the basis of immunofluorescence detection and targeted therapy for hepatic cancer.²⁰

The fluorescence spectral feature of QDs-tagged tumor tissue almost fits that of the original QDs, and the quantitative spectral comparison [Figs. 1D, 1E] demonstrate that QD emission in tumor can be clearly distinguished from tissue autofluorescence. The spectroscopic imaging of the tumor tissue [Fig. 1B] exhibits strong QD fluorescence at $590\mathrm{nm}$ and relatively weak tissue autofluorescence. In contrast, the spectroscopic imaging of the normal tissue [Fig. 1C] shows only autofluorescence background. These results indicate that the QDs-Anti-AFP probes have specific and attractive binding to hepatoma in vivo. Note that the QD probes in nude mice were excited with femtosecond laser pulses in our study, so that the sampling sites for spectroscopic imaging are only a part of the tumor, not the whole mouse. The use of a macroillumination system^{1, 2, 21} should be an adapted alternative for whole animal studies.

In addition, the control study results with the same dose of nontargeted QDs and QDs-Anti-AFP are shown in Fig. 4 ( $0.4\mathrm{nmol}$ injected plus a 24-h latent period of probe circulation). Little QD accumulation was found in tumors due to passive tumor targeting, while QDs-Anti-AFP were detected by their clear characteristic fluorescence. Note that nonspecific liver and spleen uptake was apparent in organ uptake studies (image not shown), which is similar to that of QD-PSMA biconjucgates.²

Fig. 4

Histological examination of QD passive and active targeting ability. The same dose of QDs and QDs-Anti-AFP ( $0.4\mathrm{nmol}$ and 24-h circulation) were injected in different mice bearing hepatoma of similar sizes (0.5 to $1\mathrm{cm}$ in diameter). The images were obtained from $10\text{-}\mathrm{\mu }\mathrm{m}$ thin tissue sections on a fluorescence microscope. Little QD accumulation was found in tumors due to passive tumor targeting, while QDs-Anti-AFP were detected by their clear characteristic fluorescence.

3.3.

Distribution Analysis

It has been shown that the QD probes can be targeted to tumor cells through both passive and active mechanisms (effected by cell membrane alignment along adhesive surfaces¹⁹), but passive targeting (due to permeation and retention of QDs at tumor sites ^{1, 2, 22, 23}) is much less efficient than active targeting (Fig. 4) (due to QD-antibody binding to tumor-specific antigens²). Thus, the necrotic cancer cells lacking specific antigen are almost deprived of the ability to bind to the QD probes, and the distribution of QD probes in tumor is mainly attributed to the density of cancer cells with specific antigens for active targeting. Consequently, a fluorescence intensity contour map (Fig. 2) was plotted by using the site-by-site measurement method already discussed for distribution analysis. Different colors in the map show the inhomogeneous distribution of cancer cells (with probes) in the detected area of the tumor. Since cold colors represent a lower fluorescence intensity in the area, and lower intensity corresponds to fewer QD probes among the sampling sites, it is clear that the cancer cells per field in the central area were less than the periphery area of the tumor. Although some singular points (such as the orange and red areas in the lower right-hand corner and the upper middle region of the map) do not match the whole tendency of color changes because of the complexity of tumor’s structure, the fluorescence intensity contour map exhibit that the number of cancer cells (with probes) per field is lower inside than those outside.

Note that the result does not provide an absolute intensity comparison among these sites because several factors (such as complexity of tumor’s structure and optical density of sampling sites) are difficult to normalize or calibrate. Instead, it is just a qualitative fluorescence intensity comparison, demonstrating the distribution of the cancer cells with QD probes in the tumor. Similarly, histological examinations have shown that the density of the vascular distribution in tumor is also inhomogeneous. It accumulates in the periphery and is scarce or absent in the central area, so that the process of nutrient competition causes more necrotic cells inside than outside in the tumor.²⁴ Further, the distribution result is consistent with the universal dynamics theory of tumor growth.²⁵ Unfortunately, the maximum shifting distance of the automatic $X\text{-}Y$ axis translation stage in our study is only $5\mathrm{mm}$ , not enough for whole tumor studies. However, this problem could be solved with improved experimental conditions. Notwithstanding this limitation, these results demonstrate the QD probes are specific and sensitive enough for distribution analysis of a tumor in vivo.