2.1.

Materials

UV-visible and fluorescence spectra were recorded on a Perkin-Elmer Lambda 2 spectrophotometer and LS50B spectrofluorometer, respectively. ${}^1\mathrm{H}$ NMR spectra were recorded on a Bruker $500\text{-}\mathrm{MHz}$ instrument. Mass spectrometry analyses were performed at the Mass Spectrometry Facility of the Department of Chemistry, University of Pennsylvania. All chemicals and reagents were purchased from Aldrich (Milwaukee, Wisconsin). When necessary, solvents were dried before use. For TLC, EM Science TLC plates (silica gel 60 ${\mathrm{F}}_{254}$ ) were used.

2.2.

Synthesis of Bisoleate Conjugate of Silicon Tetra-tert-butyl-phthalocyanine, SiPcBOA

A suspension of silicon tetra-butyl-phthalocyanine dihydroxide $(200\mathrm{mg},0.25\mathrm{mmol})$ in $20\mathrm{mL}$ of 2-picoline was mixed with the oleoyl chloride $(300\mathrm{mg},1.00\mathrm{mmol})$ and stirred under argon for $2\mathrm{h}$ . The 4-dimethylaminopyridine ( $376\mathrm{mg}$ , $3.08\mathrm{mmol}$ ) was added to the mixture portion wise, which was kept well stirred under argon for an additional $36\mathrm{h}$ at $60°\mathrm{C}$ . On completion, the solvent was evaporated under reduced pressure, and the product was purified by column chromatography (silica gel-hexane: $\mathrm{C}{\mathrm{H}}_2{\mathrm{Cl}}_2=1:1$ ) to yield the desired conjugate ( $130\mathrm{mg}$ , $0.098\mathrm{mmol}$ , 39.1%). UV-vis [nm $\left(\epsilon \right)$ in $\mathrm{C}{\mathrm{H}}_2{\mathrm{Cl}}_2$ ]: 362 $(1.39\times {10}^5)$ , 620 $(5.32\times {10}^4)$ , 658 $(4.58\times {10}^4)$ , 691 $(2.54\times {10}^5)$ ; emission ${\lambda }_{\max }$ in $\mathrm{C}{\mathrm{H}}_2{\mathrm{Cl}}_2$ (excitation wavelength $680\mathrm{nm}$ ): $697\mathrm{nm}$ . ESI-MS calculated for ${\mathrm{C}}_{84}{\mathrm{H}}_{114}{\mathrm{N}}_8{\mathrm{O}}_4\mathrm{Si}$ : 1327.94, found: 1327.96 $(\mathrm{M}+)$ ; HRMS calculated for ${\mathrm{C}}_{84}{\mathrm{H}}_{114}{\mathrm{N}}_8{\mathrm{O}}_4\mathrm{Si}+\mathrm{Na}$ : 1349.8630, found: 1349.8643. ${}^1\mathrm{H}$ NMR ( $\mathrm{C}\mathrm{D}{\mathrm{Cl}}_3$ , $\delta$ ppm): 9.54–9.70 (m, 8H, Aromatic H), 8.42 (m, 4H, Aromatic H), 5.25 ( $2\mathrm{m}$ , each 2H, from oleoyl vinyl H), 1.67–2.01 (m, 52 H, from oleoyl chain H), 1.21–1.28 (m, 36H, Boc H), 0.90 (m, 4H, from oleoyl chain), 0.85 (t, 6H, from oleoyl chain terminal $\mathrm{C}{\mathrm{H}}_3$ ).

2.3.

LDL Reconstitution and Characterization

LDL, purchased from Lund-Katz’ laboratory at the Children’s Hospital of Philadelphia (Philadelphia, Pennsylvania) was isolated from fresh plasma of healthy donors by sequential ultracentrifugation as described previously.¹⁶ LDL reconstitution with SiPcBOA was performed following a minor modification of the method of Krieger.¹⁵ Briefly, LDL $\left(1.9\mathrm{mg}\right)$ was lyophilized with $25\text{-}\mathrm{mg}$ starch, and then extracted three times with $5\mathrm{mL}$ of heptane at $-5°\mathrm{C}$ . Following aspiration of the last heptane extract, $6\mathrm{mg}$ of SiPcBOA was added in $200\mu \mathrm{L}$ of benzene. After $90\min$ at $4°\mathrm{C}$ , benzene and any residual heptane were removed under a stream of ${\mathrm{N}}_2$ in an ice salt bath for about $45\min$ . The r-SiPcBOA-LDL was solubilized in $10\text{-}\mathrm{mM}$ Tricine, pH 8.2, at $4°\mathrm{C}$ for $24\mathrm{h}$ . Starch was removed from the solution by a low-speed centrifugation $(500\times \mathrm{g})$ followed by a $20\text{-}\min$ centrifugation $(6000\times \mathrm{g})$ . The reconstituted LDL was stored under an inert gas at $4°\mathrm{C}$ . Similarly, r-SiPcBOA-AcLDL was also prepared from SiPc-BOA and acetylated LDL (AcLDL, Biomedical Technologies, Incorporated, Stoughton, Massachusetts). The protein content of the specimen was determined by the Lowry method.¹⁷ The absorption spectrum of SiPc-BOA was measured after extraction with a chloroform and methanol mixture (2:1), and probe concentration was calculated based on the following formula: $C=(A∕\epsilon )\times D$ , where $C$ is the concentration of the probe, $A$ is the OD value, $\epsilon$ is the extinction coefficient, and $D$ is the dilution fold. Probe/protein molar ratio was calculated using the molecular mass of the ApoB-100 protein $\left(514\mathrm{kDa}\right)$ , knowing that one LDL particle contains only one ApoB-100.

2.4.

Cell Preparations

$\mathrm{Hep}{\mathrm{G}}_2$ tumor cells, which were obtained from van Berkel’s laboratory from the University of Leiden in the Netherlands, were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), $2\mathrm{mM}$ L-glutamine, $10\text{-}\mathrm{mM}$ HEPES, with $100\text{-}\mathrm{U}∕\mathrm{mL}$ penicillin G sodium and $100\text{-}\mu \mathrm{g}∕\mathrm{mL}$ streptomycin sulfate. Cells were grown at $37°\mathrm{C}$ in an atmosphere of 5% $\mathrm{C}{\mathrm{O}}_2$ in a humidified incubator.

2.5.

Confocal Microscopy Studies

For confocal microscopy studies, $\mathrm{Hep}{\mathrm{G}}_2$ cells were grown in 4-well Lab-Tek chamber slides (Naperville, Illinois) at a density of 40,000 cells/well. Experiments were started, after two quick washes with preincubation medium [medium with 0.8% (w/v) BSA instead of FBS], by the addition of preincubation medium containing the indicated amounts of r-SiPcBOA-LDL/AcLDL and/or unlabeled LDL. After a $4\text{-}\mathrm{h}$ incubation at $37°\mathrm{C}$ , the cells were washed three times with ice-cold PBS and fixed for $15\min$ with 3% formaldehyde in PBS at room temperature. Then the chamber slides were mounted and sealed for confocal microscopy analysis. Confocal microscopy was performed with a Leica TCS SPII laser scanning confocal microscope (Heidelberg, Germany). Filter settings were $633\mathrm{nm}$ for excitation and $638\text{to}800\mathrm{nm}$ for emission.

2.6.

Electron Microscopy Studies

Five microliters of the reconstituted LDL suspension were placed on carbon-coated 200 mesh copper grids and allowed to stand for $5\min$ . Excess sample was wicked off with lens paper and 2% saturated aqueous uranyl acetate was applied to the grid in 5 consecutive drops within $20\mathrm{s}$ . The stain was then drained off with filter paper and the grid was air dried. Digital images were taken using JEOL JEM 1010 electron microscope (JEOL-USA, Incorporated, Peabody, Massachusetts) at $80\mathrm{kV}$ using AMT 12-HR software aided by a Hamamatsu CCD Camera. All related supplies were purchased at Electron Microscopy Sciences (Fort Washington, Pennsylvania).

2.7.

In Vitro PDT Studies Using r-SiPcBOA-LDL as a Photosensitizer

Flasks containing approximately $2\times {10}^6$ $\mathrm{Hep}{\mathrm{G}}_2$ cells were incubated for $5\mathrm{h}$ at $37°\mathrm{C}$ in preincubation medium with no drug, $8\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein (equivalent to $50\text{-}\mu \mathrm{g}∕\mathrm{mL}$ SiPcBOA), or $50\text{-}\mu \mathrm{g}∕\mathrm{mL}$ SiPcBOA. [The effective concentration of the SiPcBOA is calculated as follows: $(50\mu \mathrm{g}∕\mathrm{mL})∕(1328\mathrm{g}∕\mathrm{mol})=37\mu \mathrm{M}$ , where 1328 is the molecular weight of SiPcBOA.] Cells were washed with $10\text{-}\mathrm{mL}$ HBSS and subsequently incubated for $30\min$ with fresh preincubation medium. After this incubation, cells were again washed with HBSS, collected, and resuspended at a concentration of $1\times {10}^6\text{cells}∕\mathrm{mL}$ . Aliquots of the drug-exposed cells were transferred to individual $60\times 15\text{-}\mathrm{mm}$ dishes and treated with PDT at $5\mathrm{mW}∕{\mathrm{cm}}^2$ to a total fluence of $1\mathrm{J}∕{\mathrm{cm}}^2$ $\left(200\mathrm{s}\right)$ , $2.5\mathrm{J}∕{\mathrm{cm}}^2$ $\left(500\mathrm{s}\right)$ , $4\mathrm{J}∕{\mathrm{cm}}^2$ $\left(800\mathrm{s}\right)$ , or $5\mathrm{J}∕{\mathrm{cm}}^2$ $\left(1000\mathrm{s}\right)$ . Light at $680\mathrm{nm}$ was delivered using a KTP Yag-pumped dye module (Laser Scope, San Jose, California). Additionally, controls were prepared containing either $8\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein ( $50\text{-}\mu \mathrm{g}∕\mathrm{mL}$ SiPcBOA) or $50\text{-}\mu \mathrm{g}∕\mathrm{mL}$ SiPcBOA with no light dose exposure, no drug with a $5\text{-}\mathrm{J}∕{\mathrm{cm}}^2$ light dose exposure, or neither drug nor light exposure. Dishes $(100\times 20\mathrm{mm})$ were plated in triplicate and placed in a $37°\mathrm{C}$ incubator (5% $\mathrm{C}{\mathrm{O}}_2$ ) for $11\text{days}$ . Following incubation, dishes were rinsed with PBS and allowed to air dry. Subsequently, cells were fixed and stained with methylene blue, then colonies were counted. This experiment was replicated three times.

2.8.

Statistics

Analyses of the clonogenic assays were performed using STATA software (STATA Corporation, College Station, (Texas) and data were plotted in SigmaPlot (SPSS, Incorporated, Chicago, Illinois). The outcome variable, surviving fraction, was log-transformed and data were fit by linear regression. The addition of a quadratic term to the model was tested, but a likelihood ratio test indicated no significant support for the square term. The outcome was analyzed using the multiple regression procedure, treating drug as a class variable and light as continuous. Significance of effects for light and drug combinations was determined by t-test. Controls for light-alone (at the highest dose tested), drug-alone (free and conjugated), and untreated controls showed no difference and were averaged to generate a control plating efficacy to which experimental data were compared.

3.1.

Design and Synthesis of SiPcBOA

The aim of this work is to improve the labeling (reconstitution) efficiency of LDL-based PS for achieving high probe to protein payload and to provide targeting of PS to LDLR. To achieve this aim, PS should be neutral, highly soluble in nonpolar solvent, have minimal aggregation, and contain a suitable linker for conjugation to a lipid anchor to prevent the dye from dissociating from LDL and nonspecifically binding to phospholipid bilayers on cellular membranes. For these purposes, we designed a new PS for LDL reconstitution based on SiPc. Because Si coordination allows the binding of two axial oleate ligands, these ligands will then create steric hindrance on each side of the Pc ring, thereby limiting stack aggregation. Therefore, we anticipate a large increase in the PS payload of LDL.

To synthesize bisoleate-anchored SiPc, commercially available silicon tetra-tert-butylphthalocyanine dihydroxide was conjugated at the axial position with oleoyl chloride in the presence of 4-dimethylaminopyridine and 2-picoline. The desired conjugate, SiPcBOA, was obtained in 40% yield. This efficient synthetic pathway is depicted in Fig. 2 . The structure of this compound was confirmed by ${}^1\mathrm{H}$ NMR and high resolution mass spectroscopy analysis. Figure 3 shows the absorption, excitation, and fluorescence spectra of this new compound. It has a very intense absorption at $684\mathrm{nm}$ and emission at $692\mathrm{nm}$ , both within the NIR range.

Fig. 2

Synthesis of SiPc-BOA.

Fig. 3

Absorption (black), excitation (green), and fluorescence spectra (red) of SiPcBOA.

As shown in Fig. 2, this method has several distinct advantages. First, the reaction condition is very mild. It can be carried out in weak base (picoline, dimethylminopyridine) at warm temperatures $(<60°\mathrm{C})$ instead of at $150°\mathrm{C}$ in a much stronger base (sodium alcoholate), as is commonly used for the preparation of Pc derivatives. Second, the starting material, ${\left(\mathrm{t}\mathrm{Bu}\right)}_4\mathrm{Si}\mathrm{Pc}{\left(\mathrm{O}\mathrm{H}\right)}_2$ , is commercially available and consists of four lipophilic and bulky t-butyl groups at the peripheral position of the Pc macrocycle, further increasing its lipophilicity. Finally, the bisoleate anchor (BOA) is known to strongly associate with the lipid membrane, a characteristic similar to that of the cholesterol moiety. Therefore, for Pc LDL reconstitution, we expect that the bisoleate anchor will be an enhancement over the corresponding cholesterol oleate moiety.

3.2.

LDL Reconstitution and Characterization

Protein recovery determined by the Lowry method¹⁷ is an excellent assay for evaluating the success of the reconstitution. 55 to 70% protein recovery was observed for both r-SiPcBOA-LDL and r-SiPcBOA-AcLDL, which is better than that observed for r-Pyro-CE-LDL.¹⁰ The absorption spectrum for the recovered SiPcBOA after LDL reconstitution is the same as it was before reconstitution, indicating that absorbance measurement can serve as the basis for calculating the SiPcBOA concentration in the reconstituted LDL (payload). It was found that $\sim 3000$ to 3500 SiPc-BOA molecules were reconstituted into one LDL molecule core. Compared to the 50:1 probe:protein ratio for r-Pyro-CE-LDL we prepared previously, the new probe design reported here improved probe payload on each LDL nanoparticle by 60 fold.

3.3.

Confocal Microscopy Studies of the LDLR-Specific Uptake

To visualize LDLR-mediated internalization of r-SiPcBOA-LDL, we performed laser scanning confocal microscopy studies on $\mathrm{Hep}{\mathrm{G}}_2$ tumor cells. Figure 4 shows the confocal fluorescence images of $\mathrm{Hep}{\mathrm{G}}_2$ cells incubated with/without fluorescent probe (B, D, F, H, J) as well as corresponding bright field images (A, C, E, G, I). Figures 4(a) and 4(b) depict images of the cell alone, providing values for the fluorescence of the cells. When cells were incubated with $85\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein at $37°\mathrm{C}$ for $4\mathrm{h}$ , the fluorescence signal appears to be localized in the cytoplasm [Figs. 4(c) and 4(d)]. To determine the specificity of this r-SiPcBOA-LDL toward LDLR, three sets of control experiments were performed. When $\mathrm{Hep}{\mathrm{G}}_2$ cells were incubated with $85\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein plus 50-fold excess of unlabeled native LDL, complete fluorescence inhibition was observed [Figs. 4(e) and 4(f)]. When $170\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-AcLDL protein was incubated with $\mathrm{Hep}{\mathrm{G}}_2$ cells, despite the fact that the fluorophore concentration doubled, no fluorescence was observed. This is consistent with the inability of Ac-LDL to target LDLR [Figs. 4(g) and 4(h)]. Figures 4(i) and 4(j) show that incubation with $570\text{-}\mu \mathrm{g}∕\mathrm{mL}$ SiPcBOA alone (equivalent to $85\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein) did not lead to any observable fluorescence, indicating that no internalization occurred. Collectively, the previous experiments indicate that r-SiPcBOA-LDL was internalized into $\mathrm{Hep}{\mathrm{G}}_2$ tumor cells specifically via the LDL receptor pathway.

Fig. 4

Confocal fluorescence images of $\mathrm{Hep}{\mathrm{G}}_2$ cells incubated w/wt fluorescent probes [(b), (d), (f), (h), (j)] as well as the corresponding bright field images [(a), (c), (e), (g), (i)]. (a) and (b) Cell alone control; (c) and (d) $\text{cell}+85\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein; (e) and (f) $\text{cell}+85\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein $+50$ -fold over excess native LDL; (g) and (h) $\text{cell}+170\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-AcLDL protein; (i) and (j) $\text{cell}+570\text{-}\mu \mathrm{g}∕\mathrm{mL}$ SiPcBOA (same amount of SiPcBOA as in $85\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein).

3.4.

Electron Microscopy Studies

A light scattering size scanner was originally used to measure the size of the SiPcBOA reconstituted LDL particle. However, we found that Pc absorption interferes with the laser wavelength used by the scanner; therefore, electron microscopy was used to directly visualize the LDL particles. As shown in Fig. 5 , the mean particle size of r-SiPcBOA-LDL was $23.2\pm 4.6\mathrm{nm}$ $(n=30)$ , which is about the same size as native LDL $(20\pm 2.7\mathrm{nm},n=37)$ .

Fig. 5

EM images of native LDL (left) and r-SiPc-BOA-LDL (right).

3.5.

In Vitro PDT Studies (Clonogenic Assay)

Figure 6 shows the in vitro PDT response of $\mathrm{Hep}{\mathrm{G}}_2$ cells to r-SiPcBOA-LDL and SiPcBOA using a clonogenic assay. At a dose of 4 and $5\mathrm{J}∕{\mathrm{cm}}^2$ , there was significantly more cell kill with $8\text{-}\mu \mathrm{g}∕\mathrm{mL}$ r-SiPcBOA-LDL protein (equivalent to $50\text{-}\mu \mathrm{g}∕\mathrm{mL}$ SiPcBOA) than with $50\text{-}\mu \mathrm{g}∕\mathrm{mL}$ SiPcBOA [ $\mathrm{t}\left(4\right)=3.85$ , $\mathrm{p}=0.02$ , and $\mathrm{t}\left(4\right)=2.79$ , $\mathrm{p}=0.05$ , respectively]. At the drug dose used, SiPcBOA induced limited cell kill, even at the highest light dose tested. Conversely, when r-SiPcBOA-LDL was used as a photosensitizer, increasing cell kill was detected with increasing light dose. The slopes of the linear regression fit to the logarithmic data for these two plots are significantly different from each other $[\mathrm{F}(1,28)=14.71,\mathrm{p}=0.0007]$ , indicating greatly enhanced efficacy of LDLR-targeted PDT in an LDLR-overexpressing cell line. Light and drug alone controls for both the free and conjugated photosensitizer show no difference compared to untreated controls. Currently, Pc4 is a leading phthalocyanine-based photosensitizer candidate for PDT.¹⁸ Compared to Pc4, our r-SiPcBOA-LDL nanoparticle requires a higher effective photosensitizer concentration for PDT-mediated cell kill. However, Pc4 is not a target-specific photosensitizer, whereas our agent is highly tumor-specific in $\mathrm{Hep}{\mathrm{G}}_2$ cells overexpressing LDLR. Since many cancer cells overexpress LDLR, we anticipate these novel nanoparticles can serve as a useful discriminatory vehicle for the delivery of PDT agents to tumor cells.

Fig. 6

In vitro PDT response of $\mathrm{Hep}{\mathrm{G}}_2$ cells to r-SiPcBOA-LDL (closed circles) and SiPcBOA (open circles). SEMs were generated for the arithmetic mean.