(Received for publication, April 15, 1997, and in revised form, June 13, 1997)
From the Biophysical Laboratory, Russian Institute of
Agricultural Biotechnology, Timiryazevskaya Street 42, 127550, Moscow,
Russia, the § John Curtin School of Medical Research,
Division for Biochemistry and Molecular Biology, Nuclear Signaling
Laboratory, Canberra City, A.C.T. 2601, Australia, the
¶ Department of Biophysics, Biological Faculty, Moscow State
University, 119899 Moscow, Russia, and the
Institute for Medical
Chemistry, Szeged Medical University, Szeged, H-6720 Hungary
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 -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 (1-3). 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 e6 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 receptor-expressing cells (7-9). Photosensitization could be competed by incubating cells in the presence of an excess of unconjugated ligand (7-9), indicating that cellular uptake was receptor-dependent. Because only cells expressing specific receptors are targeted in this approach (see Refs. 7-9), 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 µm 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 e6 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 (7-9) 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 e6,
maximally reducing the EC50 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 basic design of the
conjugates used is shown schematically in Fig. 1. BSA was used as the
carrier in the case of the conjugates shown in A and
B, whereas -galactosidase or the
-galactosidase fusion
protein P10 were the carriers in C (see below). In the case of the
BSA-P11Lys-(chlorin e6)-insulin and BSA-P11Thr-(chlorin e6)-insulin conjugates (Fig. 1A), BSA was
cross-linked to chlorin e6 using
cyclo-hexyl-3(2-morpholinoethyl)carbodiimide metho-4-toluene sulfonate
(CMCS, Serva) in 10 mM sodium phosphate buffer, pH 7.5 (buffer A), at a ratio of BSA to chlorin e6 to CMCS of
1:30:300. The peptides P11Lys
(NH2-Pro-Lys-Lys-Lys-Arg-Lys-Val-Glu-Asp-Pro-Tyr-Cys-COOH) and P11Thr
(NH2-Pro-Lys-Thr-Lys-Arg-Lys-Val-Glu-Asp-Pro-Tyr-Cys-COOH) were linked individually to the amino groups of BSA-(chlorin
e6) using
3-maleimidobenzoyl-N-hydroxysuccinimide ester (MBS; Sigma) in buffer A containing 0.5 mM EDTA, whereas insulin (Sigma)
was separately linked to BSA using glutaraldehyde (Merck) (15). BSA-(chlorin e6)-peptide and BSA-(chlorin
e6)-insulin were then reacted individually with
N-succinimidyl 3-[2-pyridyldithio]propionate (SPDP; Sigma)
in 10 mM Hepes, 150 mM NaCl, pH 7.5, to yield
PDP derivatives. PDP-BSA-(chlorin e6)-peptide was reduced
with 50 mM dithiothreitol (Sigma) to yield HS-BSA-(chlorin
e6)-peptide, which was then reacted with PDP-BSA-insulin to
form the conjugate BSA-(chlorin e6)-peptide-BSA-insulin
(Fig. 1A). The final conjugates contained 3-4 peptide, 4-5
chlorin e6, and 4-5 insulin molecules per BSA molecule,
respectively.
In the case of the BSA-P101Lys-(chlorin e6)-insulin and BSA-P101Thr-(chlorin e6)-insulin conjugates (Fig. 1B), BSA was conjugated with chlorin e6 as above, and the peptides P101Lys (NH2-Cys-Gly-Pro-Gly-Ser-Asp-Asp-Glu-Ala-Ala-Ala-Asp-Ala-Gln-His-Ala-Ala-Pro-Pro-Lys-Lys-Lys-Arg-Lys-Val-Gly-Tyr-COOH) and P101Thr (NH2-Cys-Gly-Pro-Gly-Ser-Asp-Asp-Glu-Ala-Ala-Ala-Asp-Ala-Gln-His-Ala-Ala-Pro-Pro-Lys-Thr-Lys-Arg-Lys-Val-Gly-Tyr-COOH) were then cross-linked to BSA-(chlorin e6) through the MBS-activated amino-groups of BSA-(chlorin e6) in buffer A containing 0.5 mM EDTA. Insulin was reacted with sulfo-succinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate (Sigma; at a ratio of 1:1.5) in buffer A and then conjugated with dithiothreitol-pretreated BSA-(chlorin e6)-peptide (ratio of 15:1) under anaerobic conditions in buffer A containing 0.5 mM EDTA. Two series of constructs were made: BSA-P101Lys-(chlorin e6)-insulin (1:4:5:3), BSA-P101Thr-(chlorin e6)-insulin (1:4:7:3), and BSA-P101Lys-(chlorin e6)-insulin (1:1:11:2) and BSA-P101Thr-(chlorin e6)-insulin (1:1:8:2), where the numbers in parentheses indicate the ratio of the components in the conjugates.
In the case of the P10-(chlorin e6)-insulin and
-galactosidase-(chlorin e6)-insulin conjugates (Fig.
1C), the carrier molecules (either the NLS-containing
-galactosidase fusion protein P10 (10), containing a variant of T-ag
amino acids 111-135:
Ser-Ser-Asp-Asp-Glu-Ala-Thr-Ala-Asp-Ala-Gln-His-Ala-Ala-Pro-Pro-Lys-Lys-Lys-Arg-Lys-Val-Glu-Asp-Pro, or
-galactosidase itself) were expressed in bacteria and
prepared by affinity chromatography as previously (10, 16, 17). Chlorin e6 was initially reacted with 1,6-diaminohexane (Sigma)
using CMCS (ratio of chlorin e6 to 1,6-diaminohexane to
CMCS of 1:100:100) in 10 mM sodium phosphate buffer, pH
7.0, and CMCS was subsequently used to link aminochlorin e6
with the P10/
-galactosidase carrier in the same buffer (ratio of
carrier to aminochlorin e6 to CMCS of 1:30:300). After
successive reaction of insulin with citraconic anhydride (18) (Serva)
and SPDP (ratio of 1:6) in 25 mM Hepes, 150 mM
NaCl, pH 7.5 (buffer B), citraconic-insulin-PDP was linked to
P10-chlorin e6 or
-galactosidase-chlorin e6
(ratio of 15:1) in buffer B, after which the citraconic groups were
removed (18). Two series of constructs were made: P10-(chlorin
e6)-insulin (1:5:7),
-galactosidase-(chlorin
e6)-insulin (1:5:7), and P10-(chlorin e6)-insulin (1:3:8) and
-galactosidase-(chlorin
e6)-insulin (1:2:8).
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 e6). The conjugates were purified by dialysis or column chromatography.
Visualization and Quantitation of Photoactivation Activity in SituCells were incubated with conjugates or chlorin
e6 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 2,7
-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 2
,7
-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 e6 or
chlorin e6-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).
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 e6, then washed, and irradiated using a slide projector (96 kJ/m2). 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% N-lauroyl-sarcosine (Sigma) in phosphate-buffered saline and absorbance at 630 nm measured (22) in the case of C6 cells.
The goal of this study was to enhance the photodynamic
activity of chlorin e6 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 (Ser111/112) (14) but inhibited by
phosphorylation by the cyclin-dependent kinase cdc2 at
Thr124 adjacent to the NLS (16, 17). The NLS peptide
P101Lys (see "Materials and Methods") includes the CKII site but
contains nonphosphorylatable alanine residues in place of
Ser120, Ser123, and Thr124, whereas
the control peptide P101Thr is identical to P101Lys except that it
contains threonine in place of the critical Lys128 residue,
which abolishes nuclear targeting activity (14, 23). Shorter peptides
containing the wild type (P11Lys) or Thr128 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 Ser120,
Ser123, and Thr124 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
(7-9), such as those of the PLC/PRF/5 human hepatoma (8) or C6 rat
glioma cell lines (24).
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 e6)-insulin construct showed nuclear
levels of specific fluorescence (Fn of 3.47 arbitrary units) 15 times higher than those treated with either chlorin e6 alone or
the BSA-P101Thr-(chlorin e6)-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 with
P101Lys (Fn/c of 0.88 compared with 0.48 for the P101Thr construct).
Similar analysis for the NLS-containing P10-(chlorin
e6)-insulin construct indicated nuclear levels
significantly higher than those for the NLS-deficient
-galactosidase-(chlorin e6)-insulin construct (not shown).
Enhanced Photosensitization Conferred by Nuclear Targeting Internalizable Conjugates Containing Chlorin e6
Photodynamic activity was measured in colony
formation tests using PLC/PRF/5 and rat glioma C6 cells,
photoactivation (at a dose of 96 kJ/m2) being carried out
18 h after the addition of conjugates or free chlorin
e6. In all experiments, the NLS-containing constructs were
more effective than those lacking NLSs or containing nonfunctional (Thr128-substituted) NLSs, exhibiting much lower
EC50 values (see Table I).
The photodynamic activity of the P11Lys-containing conjugate (EC50 of 29 nM) was more than 10 times higher
than that of chlorin e6 alone (EC50 of 350 nM), whereas the NLS-deficient P11Thr peptide-containing conjugate exhibited reduced activity (EC50 of 80 nM) (Table I). The photodynamic activity of P101-containing
conjugates was higher (EC50 of 23 nM for
PLC/PRF/5 cells) than that of those containing P11Lys, whereas the
substitution of Lys128 by Thr greatly increased the
EC50 (>150 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
e6)-insulin, exhibiting an EC50 value of 0.13 nM, compared that of 2 nM for the NLS-deficient
-galactosidase-(chlorin e6)-insulin construct, in colony
formation tests with PLC/PRF/5 cells, which is over 2400 times lower
than the value for free chlorin e6 (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 e6
within the conjugate does not appear to be critical for its
photodynamic activity, e.g. the EC50s for
-galactosidase-(chlorin e6)-insulin (2 chlorin
e6 residues) and for BSA-(chlorin e6)-insulin (16 chlorin e6 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 e6 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 NLS-function 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.
We thank Tatyana Kiseleva for excellent technical assistance.