Nuclear targeting of chlorin e6 enhances its photosensitizing activity.

Although photosensitizers, molecules that produce active oxygen species upon activation by visible light, are being extensively used in photodynamic therapy to treat cancer and other clinical conditions, problems include normal cell and tissue damage and associated side effects, which are attributable in part to the fact that cytotoxic effects are largely restricted to the plasma membrane. We have previously shown that the photosensitizer chlorin e6 has significantly higher photosensitizing activity when present in conjugates containing specific ligands and thus able to be internalized by receptor-expressing cells. In this study we use insulin-containing conjugates to which variants of the simian virus SV40 large tumor antigen nuclear localization signal (NLS) were linked to target chlorin e6 to the nucleus, a hypersensitive site for active oxygen species-induced damage. NLSs were either included as peptides cross-linked to the carrier bovine serum albumin or encoded within the sequence of a beta-galactosidase fusion protein carrier. The results for photosensitization demonstrate clearly for the first time that NLSs increase the photosensitizing activity of chlorin e6, maximally reducing the EC50 by a factor of over 2000-fold. This has wide-reaching implications for achieving efficient cell type-specific photodynamic therapy.

Photosensitizers such as porphyrins are molecules that produce active oxygen species upon activation by visible light and are currently being extensively used in photodynamic therapy to treat cancer and other clinical conditions (1)(2)(3). Because normal cells are able to accumulate porphyrins, however, and porphyrins are only excreted slowly from the body, prolonged skin photosensitization as well as other effects can be a problem (4,5), leading to normal cell and tissue damage (6). A high priority with respect to photodynamic therapy is accordingly to increase the specificity of the uptake of photosensitizers in particular target cells, thereby enabling the active dose of porphyrins administered to patients to be reduced. We have previously shown that the photosensitizer chlorin e 6 has significantly higher photosensitizing activity when present in conjugates containing specific ligands such as insulin or concanavalin A and thus is able to be internalized by receptorexpressing cells (7)(8)(9). Photosensitization could be competed by incubating cells in the presence of an excess of unconjugated ligand (7)(8)(9), indicating that cellular uptake was receptor-dependent. Because only cells expressing specific receptors are targeted in this approach (see Refs. [7][8][9], it is clear that it enables selectivity in terms of the cell types targeted for photosensitization. Due to the fact that injury induced by singlet oxygen comprising 80% of all of the active oxygen species generated upon porphyrin activation is localized within less than 0.1 m of the site of its production, the effect of photosensitizers is integrally dependent on their site of cellular accumulation (1). Although most porphyrins such as chlorin e 6 and hematoporphyrin derivatives localize largely at the plasma membrane (10,11), it is known that intracellular sites, and particularly the nucleus, are much more sensitive sites for photodynamic damage (10,12,13). Consistent with this, our previous results (7)(8)(9) indicate that the enhancement of photosensitization effected by internalizable conjugates is directly attributable to their ability to be internalized (and thereby damage intracellular sites), because treatments preventing internalization severely reduce photosensitization. The directed delivery of photosensitizers to particularly sensitive subcellular organelles such as the nucleus using specific targeting signals would seem to be a key to performing efficient photodynamic therapy.
Here we use insulin-containing conjugates to which variants of the simian virus SV40 large tumor antigen (T-ag) 1 nuclear localization signal (NLS) (14) were linked to target the photosensitizer to the nucleus. NLSs were either included as peptides cross-linked to the carrier bovine serum albumin (BSA) or encoded within the sequence of a ␤-galactosidase fusion protein carrier. Results for photosensitization demonstrate clearly for the first time that NLSs increase the photosensitizing activity of chlorin e 6 , maximally reducing the EC 50 by a factor of over 2000-fold, and confirm that the nucleus is a hypersensitive site for photodynamic action. These observations have implications for achieving efficient cell type-specific photodynamic therapy.
The average ratio of components in conjugates was calculated using both the molecular weight estimation from polyacrylamide gel electrophoresis and the optical density measurement at 400 nm (for chlorin e 6 ). The conjugates were purified by dialysis or column chromatography.
Visualization and Quantitation of Photoactivation Activity in Situ-Cells were incubated with conjugates or chlorin e 6 for 18 h at 37°C, washed in RPMI 1640 medium (without phenol red, containing 2 mg/ml BSA, 25 mM Hepes, pH 7.5), incubated with 2Ј,7Ј-dichlorofluorescin diacetate (DCFD, Serva) for 5 min at 37°C, and then washed in the same medium and irradiated using a slide projector. Fluorescence due to the production of 2Ј,7Ј-dichlorofluorescein from DCFD through reaction with reactive oxygen species (19) was visualized either as described previously (9) or using CLSM (BIO-RAD MRC-600 CLSM) (16,17,20,21). Image analysis of CLSM files was performed using the NIH Image public domain software (16,17,20,21). Equivalent measurements were made of fluorescence in the nucleus (Fn), and cytoplasm (Fc) in cells treated without (background fluorescence) or with free chlorin e 6 or chlorin e 6 -containing conjugates. Background was subtracted from all values to yield specific fluorescence. The nuclear/cytoplasmic ratio (Fn/c) represents the fold accumulation in the nucleus (16,17,20,21).
Determination of Cell Viability-Two assays were used as previously (8,9). Cells were plated either in 12-well plates (80000 cells/well) or in 35-mm dishes (5000 cells/dish), 1 day after which they were incubated for 18 h at 37°C with various concentrations of conjugates or chlorin e 6 , then washed, and irradiated using a slide projector (96 kJ/m 2 ). In the case of the former, cells were trypsinized 1 h later and replated in 35-mm dishes, and the number of colonies were counted after 10 days, whereas in the latter, cells were not replated but simply fixed after 10 days and stained with methylene blue. Colonies were then either counted in the case of PLC/PRF/5 cells or dye extracted with 1% Nlauroyl-sarcosine (Sigma) in phosphate-buffered saline and absorbance at 630 nm measured (22) in the case of C6 cells.

RESULTS AND DISCUSSION
Construction of NLS-containing Internalizable Photosensitizing Conjugates-The goal of this study was to enhance the photodynamic activity of chlorin e 6 by directing its specific delivery to the nucleus through its incorporation into an internalizable construct containing an NLS. We used either BSA (Fig. 1, A and B) or ␤-galactosidase/␤-galactosidase fusion protein P10 (Fig. 1C) as carriers to which other components were covalently attached in three types of conjugates (see Fig. 1). NLSs were included either as peptides covalently coupled to BSA or within the coding sequence of P10. The NLSs were selected based on our previous work, which showed that nuclear targeting effected by the T-ag NLS (amino acids 126 -132) is enhanced markedly by the casein kinase II (CKII) phosphorylation site (Ser 111/112 ) (14) but inhibited by phosphorylation by the cyclin-dependent kinase cdc2 at Thr 124 adjacent to the Nuclear Targeting Enhances Chlorin e 6 Activity 20329 NLS (16,17). The NLS peptide P101Lys (see "Materials and Methods") includes the CKII site but contains nonphosphorylatable alanine residues in place of Ser 120 , Ser 123 , and Thr 124 , whereas the control peptide P101Thr is identical to P101Lys except that it contains threonine in place of the critical Lys 128 residue, which abolishes nuclear targeting activity (14,23). Shorter peptides containing the wild type (P11Lys) or Thr 128 derivative (P11Thr) of the T-ag NLS alone ("Materials and Methods") were also used. All peptides contained either N-or C-terminal Cys residues for cross-linking. The P10 ␤-galactosidase fusion protein contains T-ag amino acids 111-135 with Ala substitutions at Ser 120 , Ser 123 , and Thr 124 identical to those of P101Lys (14, 16, 17). Insulin was included in the constructs to confer binding to and internalization by receptor-containing target cells (7)(8)(9), such as those of the PLC/PRF/5 human hepatoma (8) or C6 rat glioma cell lines (24). NLSs Confer Nuclear Localization of Internalizable Conjugates Containing Chlorin e 6 -Subcellular localization of the BSA-peptide constructs was visualized by confocal laser scanning microscopy (CLSM) using DCFD (Fig. 2, B and D), which interacts with reactive oxygen species to yield a fluorescent product, thus enabling visualization of the precise subcellular sites of photoactivation/photo-oxidation in living cells (9). PLC/ PRF/5 cells treated with the NLS-containing BSA-P101Lys-(chlorin e 6 )-insulin construct showed nuclear levels of specific fluorescence (Fn of 3.47 arbitrary units) 15 times higher than those treated with either chlorin e 6 alone or the BSA-P101Thr-(chlorin e 6 )-insulin construct that contains a nonfunctional NLS (Fn of about 0.2). The extent of nuclear accumulation (Fn/c) relative to that in the cytoplasm was more than 85% higher than that for the functional NLS-containing construct   b Cells were incubated with varying concentrations of conjugates or chlorin e 6 , then washed, irradiated, and replated on dishes; after 10 days the number of colonies was counted.
c One day after plating into 35-mm dishes, cells were incubated with conjugates or chlorin e 6 , then washed, and irradiated. After 10 days, colonies were fixed and stained, and either colonies were counted in the case of PLC/PRF/5 cells, or dye was extracted (see "Materials and Methods") in the case of C6 cells and absorbance measured at 630 nm. d ND, not determined.

Enhanced Photosensitization Conferred by Nuclear Targeting Internalizable Conjugates
Containing Chlorin e 6 -Photodynamic activity was measured in colony formation tests using PLC/PRF/5 and rat glioma C6 cells, photoactivation (at a dose of 96 kJ/m 2 ) being carried out 18 h after the addition of conjugates or free chlorin e 6 . In all experiments, the NLS-containing constructs were more effective than those lacking NLSs or containing nonfunctional (Thr 128 -substituted) NLSs, exhibiting much lower EC 50 values (see Table I). The photodynamic activity of the P11Lys-containing conjugate (EC 50 of 29 nM) was more than 10 times higher than that of chlorin e 6 alone (EC 50 of 350 nM), whereas the NLS-deficient P11Thr peptide-containing conjugate exhibited reduced activity (EC 50 of 80 nM) ( Table  I). The photodynamic activity of P101-containing conjugates was higher (EC 50 of 23 nM for PLC/PRF/5 cells) than that of those containing P11Lys, whereas the substitution of Lys 128 by Thr greatly increased the EC 50 (Ͼ150 nM) (Table I). Similar results were obtained using C6 cells ( Fig. 3C and Table I). The fact that the P11Lys-containing BSA construct showed reduced efficiency compared with the P101Lys-containing construct implied that the additional T-ag sequences present in P101Lys enhanced nuclear import, presumably as a result of the presence of the CKII site. Consistent with this, the most potent photosensitizing conjugate was P10-(chlorin e 6 )-insulin, exhibiting an EC 50 value of 0.13 nM, compared that of 2 nM for the NLS-deficient ␤-galactosidase-(chlorin e 6 )-insulin construct, in colony formation tests with PLC/PRF/5 cells, which is over 2400 times lower than the value for free chlorin e 6 ( Fig. 3A and Table I). Like the P101Lys peptide, P10 contains the CKII site that enhances the rate of nuclear import by about 50-fold (14,16).
We measured the photodynamic activity of our conjugates using colony formation tests and applied the DCDF test permitting the determination of the subcellular localization of conjugates. Although the data are insufficient to enable a definitive conclusion to be drawn, it is interesting that the number of molecules of chlorin e 6 within the conjugate does not appear to be critical for its photodynamic activity, e.g. the EC 50 s for ␤-galactosidase-(chlorin e 6 )-insulin (2 chlorin e 6 residues) and for BSA-(chlorin e 6 )-insulin (16 chlorin e 6 residues) were approximately the same (Table I; see Ref. 9). In contrast, the number of insulin and NLS moieties does appear to be critical in determining photodynamic activity, where at least three moieties of each are preferable (see Table I).
In summary, targeting of chlorin e 6 to the nucleus through incorporation in NLS-containing conjugates increases its photodynamic activity by more than 2000-fold, confirming that the nucleus is a hypersensitive site for photodynamic action (10,12,13). NLS-containing photosensitizing constructs such as those described here should have important application in cell type-specific photodynamic therapy. Future work will include optimizing nuclear targeting through modification of NLSfunction modulating flanking sequences such as phosphorylation sites (20,25,26) and substituting insulin with alternative ligands to target photosensitizers specifically to particular tumor cell types.