Abstract
purpose. Bioactive proteins such as interferon (IFN) have been reported to be
combined with water-soluble polymers, such as dextran, through metal
coordination, without need for complicated procedures. In the current
study, the targeting and inhibitory effects of IFN combined with
dextran on experimental choroidal neovascularization (CNV) were studied
in vivo.
methods. Interferon (IFN)β was conjugated to dextran, which has
metal-chelating, diethylenetriaminepentaacetic acid (DTPA) residues.
Based on metal coordination, conjugation of IFNβ with DTPA-dextran
resulted from simply mixing both substances in an aqueous solution
containing Zn2+. The effects of IFNβ on the proliferation
of human umbilical vein endothelial cells (HUVECs) and bovine retinal
pigment epithelial cells (BRPECs) were evaluated. To evaluate the
activity loss of IFNβ by conjugation, the effect of the conjugate on
HUVECs was compared with that of free IFNβ. Experimental CNV was
induced by subretinal injection of gelatin microspheres containing
basic fibroblast growth factor in rabbits. The rabbits with CNV were
intravenously treated twice weekly with 7.5 million
international units (MIU)/kg per day free IFNβ (for 4 weeks), with
IFNβ-DTPA-dextran conjugate containing 7.5 (for 2 weeks) or 0.75 (for
4 weeks) MIU/kg per day IFNβ, or with saline. The effects of these
substances were evaluated by fluorescein angiography and histology. To
observe the accumulation of conjugate, the doses of IFNβ in CNV
tissues were measured by enzyme-linked immunosorbent assay.
results. IFNβ inhibited the growth of HUVECs and enhanced the proliferation of
BRPECs. The conjugate seemed to preserve approximately 44% of IFNβ
activity. Although both doses of IFNβ-DTPA-dextran inhibited
progression of CNV in rabbits, longer term administration of a lower
dose of IFNβ-DTPA-dextran had a sustained inhibitory effect on
progression of CNV (P < 0.05). Histologic studies
revealed the inhibitory effect of IFNβ-DTPA-dextran on progression of
CNV. This conjugate prolonged the plasma half-life of IFNβ and
enabled IFNβ to accumulate in the CNV in rabbits.
conclusions. In this study, human IFNβ was successfully used to target CNV, an
enhanced antiangiogenic effect was achieved by combining it with
dextran, based on metal coordination. This targeted delivery of IFNβ
may have potential as a treatment modality for
CNV.
Age-related macular degeneration (AMD) is a major cause of
blindness in patients more than 50 years of age. Severe visual loss is
often caused by angiogenesis originating from the
choriocapillaris.
1 Laser photocoagulation, vitreous
surgery, radiation, and photodynamic therapy have been used to treat
AMD.
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 However, none of the current forms of therapy is
effective against blindness from choroidal neovascularization
(CNV).
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 These treatments may accelerate angiogenesis
or injure normal retinal tissue
17 and therefore cannot be
used to treat early-stage AMD in patients with good visual acuity. It
is important to develop potent antiangiogenic drugs to arrest AMD in
the incipient stage.
Several antiangiogenic drugs, such as interferon (IFN)-α,
thalidomide, and TNP-470, have been investigated.
18 19 20 21 22 Although in vivo studies using animal models of angiogenesis have
demonstrated the therapeutic feasibility of these drugs, recently
reported studies have shown that IFNα and thalidomide do not always
sufficiently inhibit the development of CNV
membranes,
23 24 25 primarily because the drugs generally do
not have organ-specific affinity and their in vivo half-lives are too
short. Moreover, systemic administration of these substances may result
in serious side effects when sufficient drug concentrations reach the
tissues with neovascularization, because angiogenesis occurs not only
at the pathologic site but also during wound healing and tissue
development.
20 24 This history of attempted development of
pharmaceutical therapy indicates that future novel drugs also may not
be efficacious. The development of drug delivery systems to facilitate
therapeutic efficacy and to minimize side effects is desired as eagerly
as the identification of a new potent antiangiogenic drug.
The newly formed vasculature in tumor tissue has high substance
permeability compared with that of normal tissues and incomplete
lymphatic systems. These features enable macromolecules to accumulate
and remain in the perivascular regions of solid tumors longer than in
normal tissue, which is referred to as the enhanced permeability and
retention (EPR) effect.
26 27 28 29 It has been demonstrated
that conjugation of drugs with water-soluble polymers such as
poly(vinyl alcohol), poly(ethylene glycol), and dextran prolongs the
circulating life of these drugs and increases the accumulation of drugs
in tumor mass, according to the pharmacokinetics of
macromolecules.
30 31 32 This passive drug targeting is
supposed not only to facilitate the treatment effect, but also to
attenuate adverse effects, because substances combined with large
molecules are nearly unable to access nontargeted tissues across
capillaries with normal permeability, and the dose necessary for
systemic administration decreases in association with the enhanced
therapeutic effect. In ocular angiogenesis, choroidal neovascular
membranes (CNVMs) have anatomic characteristics similar to tumor
tissue, because the retinal tissue surrounding the CNVMs has
prelymphatic systems,
33 and CNVMs are stained by
fluorescein in late-phase fluorescein angiography. Based on these
concepts, we demonstrated that passive targeting of the antiangiogenic
agent TNP-470 to experimental CNV through chemical conjugation with a
water-soluble polymer significantly enhances its therapeutic potential
for treating AMD.
34
However, in bioactive proteins, chemical conjugation causes inevitable
loss of protein activity, because it involves complicated multistep
procedures.
35 36 It is therefore of utmost importance to
develop a new and simple method for conjugating proteins to carriers
without using chemical coupling. We developed a new method of
preparation of the drug–polymer conjugate based on metal coordination,
which has been conventionally used in metal-chelating affinity
chromatography to separate proteins and peptides.
37 38 We
have already applied this conjugation method to tumor necrosis factor
(TNF)-α and IFNα to demonstrate enhanced antitumor and antiviral
effects.
39 40
In the present study, we attempted the targeted delivery of IFNβ,
which is expected to exhibit stronger antiangiogenic effects than
IFNα, to treat AMD. We investigated the in vitro effect of IFNβ on
the proliferation of vascular endothelial cells and retinal pigment
epithelial (RPE) cells and the inhibitory effect of IFNβ on the
progression of experimental CNV in rabbits through simple mixing with
dextran, involving diethylenetriaminepentaacetic acid (DTPA) residues.
We also evaluated the retention effect of conjugate with IFNβ in CNV
lesions.
IFNβ and basic fibroblast growth factor (bFGF) were kindly
supplied by Toray (Kanagawa, Japan) and Kaken Pharmaceutical (Osaka,
Japan), respectively. Dextran with an average molecular weight of
200,000 was purchased from Nacalai Tesque (Kyoto, Japan);
dimethylaminopyridine and zinc chloride (ZnCl2)
from Wako Pure Chemicals Industries (Osaka, Japan); and DTPA anhydride
from Dojindo Laboratories (Kumamoto, Japan). All other chemicals were
reagent-grade products obtained commercially.
For metal chelation, DTPA residues were introduced to a hydroxyl
group of dextran, as described previously.
39 Briefly, 662
mg DTPA and 22.6 mg dimethylaminopyridine were added to 500 mL
dehydrated dimethyl sulfoxide containing 500 mg dextran. The solution
mixture was agitated at 40°C for 16 hours, followed by dialysis
against distilled water for 2 days and freeze drying to obtain dextran
with DTPA residues introduced (DTPA-dextran). DTPA introduction was
quantitated using a conductivity meter (model DS-12; Horiba, Kyoto,
Japan). DTPA residues were estimated to be introduced into 15% of the
repeated glucose residues of dextran.
IFNβ was conjugated to DTPA-dextran (IFNβ-DTPA-dextran) in
ZnCl
2 aqueous solution. Briefly, the dose of
IFNβ required for one intravenous injection in a rabbit was diluted
with distilled water containing DTPA-dextran to 1.35 mL final volume.
To this resultant solution, 0.15 mL ZnCl
2 aqueous
solution was added (ratio of IFNβ, DTPA-dextran, and
ZnCl
2 = 1 million international units [MIU]:1.5
mg:0.126 mg). The mixture was gently agitated at 25°C for 1 hour to
allow IFNβ to conjugate to the DTPA-dextran under
Zn
2+ coordination
(Fig. 1) . Before intravenous administration, 1.5 mL of 1.8 N normal saline was
added to this solution. For in vitro use, this solution was diluted
with culture medium by 0.1% or less.
Human umbilical vein endothelial cells (HUVECs) were purchased
from Kurabo (Okayama, Japan). HUVECs were grown in monolayer cultures
in medium (HuMedia EG-2; Kurabo) containing 1% fetal bovine serum.
Bovine RPE cells (BRPECs) were obtained from bovine eyes, as described
previously.
34 The cells of passages 3 to 5 were used for
the experiments. HUVECs and BRPECs were maintained in 10-cm cell
culture dishes.
To assess the original effect of IFNβ on cell viability, a
tetrazolium-based colorimetric assay (XTT assay; Boehringer
Mannheim, Tokyo, Japan) method was used according to the
instructions of the manufacturer.
41 HUVECs and BRPECs were
plated into 96-well cell culture plates (Iwaki Glass, Tokyo, Japan),
and IFNβ (0.1 IU/mL to 1.0 MIU/mL) was added to the cultures the next
day. After 5 days, 50 μL of XTT solution was added to the culture.
After an additional 4-hour incubation, the absorbance at 450 nm was
determined by spectrophotometry (model DU-64; Beckman Instruments,
Tokyo, Japan). Each experiment was performed in quadruplicate and
repeated three times.
To evaluate the loss of activity of IFNβ by conjugation with dextran,
HUVECs were incubated for 2 days with free IFNβ (0.1 MIU/mL),
IFNβ-DTPA-dextran with the same concentration of IFNβ, the mixture
of IFNβ and DTPA-dextran without ZnCl2, or
DTPA-dextran and ZnCl2 without IFNβ. As
described previously, cell viability was measured using the XTT assay.
Each experiment was performed in quadruplicate and repeated four times.
Eighty-eight pigmented rabbits, weighing 1.9 to 2.8 kg, were
used in this study. The animals were treated according to the ARVO
Statement for the Use of Animals in Ophthalmic and Vision Research.
Only the right eye of each rabbit was used. The rabbits were
anesthetized with intramuscular ketamine (5 mg/kg) and xylazine (2
mg/kg). Topical 1% tropicamide and 2.5% phenylephrine hydrochloride
were instilled for mydriasis during surgery and fluorescein angiography
and to observe the fundus. The rabbits were killed with an overdose of
intravenous pentobarbital sodium.
Experimental CNV in rabbits was induced as previously
reported.
42 In this model, CNV is induced by bFGF, a
direct angiogenesis factor, to exclude inflammation as the predominant
angiogenic stimulus.
In 45 (70.3%) of the 64 eyes that received the bFGF-impregnated
microspheres, CNV lesions developed with mild or moderate fluorescein
leakage 3 weeks after induction. Because the fluorescein leakage in
eyes persists for 4 to 6 weeks,
42 we used these eyes in
the following in vivo studies.
To determine the antiangiogenic efficacy of IFNβ-DTPA-dextran,
rabbits were treated with saline (n = 9, 4 weeks), free
IFNβ (single dose, 7.5 MIU/kg; n = 5, 4 weeks), or
IFNβ-DTPA-dextran with 0.75 (n = 8, 4 weeks) and 7.5
(n = 10, 2 weeks) MIU/kg of IFNβ. Each substance,
beginning 3 weeks after implantation of the gelatin microspheres, was
administered intravenously twice weekly. Fluorescein angiography was
repeated weekly. To evaluate the treatments’ effects, all angiographic
images taken 7 to 9 minutes after injection to document the degree of
late fluorescein leakage were converted to digital images, and the area
of fluorescein leakage was quantified in a masked fashion by computer
(Image 1.52; NIH Image is provided in the public domain by the National
Institutes of Health, Bethesda, MD and is available at
http://rsb.info.nih.gov/nih-image/). To prevent variability during
image processing, the density of the choroidal background on the images
from the same eye was harmonized without changing the contrast. Each
measurement was scored as follows: 0, no or faint fluorescein leakage
(less than choroidal background in density; 1, mild leakage (as low as choroidal background in density); 2, focal intense leakage (20%
or less than the area with microspheres); 3, intense leakage of
moderate size (20%–50%); and 4, intense leakage of large size (50%
or more).
The eyes treated with saline (n = 2) and
IFNβ-DTPA-dextran (n = 2) were studied histologically.
These rabbits were killed 7 weeks after implantation of the
microspheres. The eyes were enucleated and immediately placed in 2.5%
glutaraldehyde and 2% formaldehyde in 0.1 M phosphate-buffered saline
(PBS; pH 7.4) for 15 minutes. The cornea, lens, and vitreous were
carefully removed and placed in the fixative for an additional 24 hours
at 4°C. The area in which the hydrogel had been implanted was
dissected from each eye under the dissecting microscope. Each specimen
was embedded in paraffin. Sections 2- to 3-μm thick were prepared and
stained with hematoxylin-eosin for light microscopy.
To investigate the plasma half-life, blood samples were obtained
and centrifuged 1, 2, and 3 days after intravenous administration of
7.5 MIU/kg IFNβ (n = 12) or IFNβ-DTPA-dextran, with the
same amount of IFNβ (n = 12). The plasma samples were
diluted with PBS containing 1 M NH4Cl and
analyzed by enzyme-linked immunosorbent assay (ELISA). The dose of
IFNβ-DTPA-dextran in the CNV lesions was also measured by ELISA.
Other rabbits with CNV were treated with 7.5 MIU/kg IFNβ (n= 4) or IFNβ-DTPA-dextran with the same amount of IFNβ
(n = 5). Twenty-four hours later, these rabbits were killed
with an overdose of intravenous pentobarbital sodium, and the eyes were
immediately enucleated. The chorioretinal tissue (5 mm in diameter
including CNV lesions) was punched out after freezing on acetone ice
and homogenized with PBS containing 1 M NH4Cl and
0.1% 3-[(3-cholamidopropyl) dimethylammonio]-1-propane sulfonate.
The tissue homogenates were centrifuged at 3000 rpm, and the
supernatant was collected. The remaining tissue was freeze dried and
weighed. The concentration of IFNβ in this supernatant was then
measured by ELISA.
All data are mean ± SEM. Student’s t-test was
used to analyze the in vitro data and the recovery of IFNβ from CNV
lesions. The Wilcoxon signed rank test was used to compare the results
of fluorescein angiography at different time points in each group and
the Mann-Whitney test to compare the results at each time point with
the control. Differences were considered statistically significant at P < 0.05.
RPE cells may play a partial role in vascular nonperfusion and the
reconstruction of the outer BRB, as well as in the progression of
CNVMs.
43 44 Proliferation of RPE cells may promote the
regression of CNVMs by covering the neovascular vessels. We
investigated the effect of IFNβ on the proliferation of HUVECs and
BRPECs
(Fig. 2) . At concentrations of more than 10 IU/mL, IFNβ inhibited the
proliferation of HUVECs. Cytostatic inhibition was observed, with
complete inhibition at concentrations higher than 0.1 MIU/mL. However,
IFNβ enhanced the proliferation of BRPECs in a dose-dependent manner.
If these properties of IFNβ were reflected in the in vivo relation
between choroidal endothelial cells and RPE cells, they would be
advantageous in the treatment of AMD.
At the concentration of 0.1 MIU/mL, whereas IFNβ reduced the cell
viability of HUVECs to 68% of the control with cytostatic inhibition,
the conjugate retained approximately 44% of this activity of IFNβ.
The solution of DTPA-dextran and ZnCl
2 without
IFNβ exhibited no inhibitory effect on HUVECs
(Fig. 3) .
A CNV rabbit model was prepared by local injection of gelatin
microspheres containing bFGF into the subretinal space. We demonstrated
that controlled-release gelatin microspheres enable bFGF to induce
neovascularization from the choriocapillaris and that the pathologic
course is similar to that of human AMD.
42 In this model,
CNV was observed 3 weeks after the injection and developed for
approximately 4 weeks. We designed the administration regimen to treat
CNV in the incipient stage. The rabbits with CNV received intravenous
injections of IFNβ-DTPA-dextran containing 0.75 or 7.5 MIU/kg IFNβ
for 4 or 2 weeks, respectively, or free IFNβ (7.5 MIU/kg) for 4
weeks. All substances were administered intravenously twice weekly.
Fluorescein angiography revealed that IFNβ-DTPA-dextran significantly
inhibited progression of CNV, whereas free IFNβ had no significant
effect
(Figs. 4 5) . When the conjugate containing a higher IFNβ dose (7.5 MIU/kg) was
intravenously injected twice weekly for 2 weeks, a significant
inhibitory effect on progression of CNV was observed 5 weeks after the
induction of CNV, but CNV recurred with the end of treatment
(Fig. 5C) .
In contrast, 4-week administration even with the lower IFNβ dose
(0.75 MIU/kg) exhibited a prolonged inhibitory effect
(
P < 0.01;
Fig. 5D ). These findings demonstrate that
the conjugate of IFNβ and DTPA-dextran based on metal coordination
successfully inhibited CNV in rabbits compared with free IFNβ, in a
manner dependent on the injection frequency and dose.
After injection of the gelatin microspheres, fibrovascular
membranes, which are mainly composed of RPE-like cells, an
extracellular matrix, and new vessels, were observed beneath the
retina. Numerous capillaries had newly formed from the
choriocapillaris. After treatment with the IFNβ-DTPA-dextran
conjugate, the subretinal membrane consisted mainly of a large number
of RPE-like cells with pseudoacinar structures, and the newly formed
capillaries were hardly visible in the membrane
(Fig. 6) .
To investigate the retention’s effect, an experimental CNV model
in rabbits was used. After the CNV lesions were examined by fluorescein
angiography 3 weeks after the subretinal injection, rabbits were killed
1 day after intravenous injection of free IFNβ or
IFNβ-DTPA-dextran. The eyes were enucleated and the chorioretinal
tissues were dissected and homogenized. The concentration of IFNβ in
the supernatant of this homogenate was measured by ELISA. More IFNβ
was detectable in rabbits injected with IFNβ-DTPA-dextran than in
those injected with free IFNβ
(Fig. 7) . The plasma half-life of IFNβ in the terminal-elimination phase
increased to 16.9 hours with the DTPA-dextran conjugation with metal
coordination, whereas that of free IFNβ was 4.5 hours. Therefore, the
presence of IFNβ-DTPA-dextran was prolonged in the CNV lesions
because of the prolonged plasma half-life and the EPR effect, compared
with free IFNβ.
CNV is a major cause of visual loss, and a new therapy for CNV in
AMD has been eagerly sought. Recently, antiangiogenic drugs, such as
IFNα and thalidomide, have been used to suppress the progression of
CNVMs. However, systemic IFNα is not effective in patients with
exudative AMD.
24 Thalidomide also did not prevent CNV’s
recurrence in patients with punctuate inner choroidopathy, and
difficulties in its continuous use have been reported, including such
side effects as drowsiness, constipation, and peripheral
neuropathy.
23 25
These treatment failures may occur partly because there is no
organ-specific affinity and because of the short plasma half-lives of
low-molecular-weight drugs. A new treatment for AMD may be developed
without overcoming these limitations. To the best of our knowledge, the
application of drug-targeting technology to treat ocular angiogenesis
has been reported only by our group.
34 Our previous
studies have demonstrated that various water-soluble polymers with a
molecular weight of approximately 200,000 accumulates in tumor tissue
in significantly higher amounts and for longer periods than larger or
smaller polymers.
28 29 Based on these results and the
hypothesis that CNV may have surroundings similar to that of
neovascular vessels in tumors, we have demonstrated that the chemical
conjugation of TNP-470 with poly(vinyl alcohol) with a molecular weight
of 220,000 accumulates in the CNV lesion and inhibits
progression.
34 Therefore, we used dextran, with an average
molecular weight of 200,000, in the present study. Dextran has been
used as a plasma expander to prevent blood platelet
aggregation.
45 Recently, another study showed that dextran
with molecular weights of 70,000 and higher tends to remain in the
perivascular area longer than several weeks,
46 which makes
it possible to enhance the targetability of antiangiogenic drugs to new
vessels.
In this study, we demonstrated that a simple mixing procedure enabled
bioactive protein to conjugate to macromolecules, based on metal
coordination. In our previous studies, the coexistence of a
polysaccharide with DTPA residues and metal ions, such as
Zn
2+ and Cu
2+, shifted the
peak of IFN or TNF shown in gel filtration chromatography to a shorter
retention time, which suggested that these bioactive proteins could be
conjugated to DTPA-polysaccharide, based on metal
coordination.
39 40 This metal ion-coordinated
protein-DTPA-polysaccharide conjugate was structurally stable in PBS.
In metal-chelating affinity chromatography, IFNβ that was bound to
the gel through Zn
2+ coordination was released
slightly by application of serum (data not shown). Therefore, it is
possible that the conjugate may release the protein gradually by
displacing or chelating agents in the plasma and target tissue.
However, bodily distribution and in vivo studies have revealed a
prolonged plasma half-life and targeting of bioactive proteins and
their biological activities only by mixing proteins and
DTPA-polysaccharide in an aqueous solution containing metal ions.
Therefore, it is likely that the metal coordination bond enables
proteins to conjugate to a polysaccharide strongly enough to carry them
to the target tissue without dissociation in the body.
The present study demonstrated that the conjugation of IFNβ with
dextran could prolong the plasma half-life of IFNβ, enhance the
accumulation of IFNβ in CNV lesions, and inhibit progression of CNV
in rabbits. When IFN was chemically conjugated by the cyanuric chloride
method to another polysaccharide (pullulan) with a high inherent
affinity for the liver, IFN activity decreased by 9% or
less.
35 36 If hexamethylendiamine was introduced between
pullulan and IFN molecules as a spacer group to increase the distance
between these large molecules, IFN activity was recovered to
18.8%,
36 which suggests that this loss of IFN activity
results not only from complicated procedures but also from steric
hindrance by conjugated polymer. Thus, conjugation may reduce the
biological activity of IFNβ because of steric hindrance by dextran.
In vitro, IFNβ-DTPA-dextran had approximately 56% of the apparent
activity loss of IFNβ through conjugation. As described previously,
when the liberation of IFNβ from the conjugate in culture medium is
taken into consideration, the activity loss by conjugation may be
greater. However, even if this activity loss is considered, the
conjugate was still highly effective in suppressing the progression of
CNV. The in vivo efficacy of the conjugate in suppression of CNV is
ascribed to the prolonged plasma half-life and enhanced targetability
of IFNβ to CNV. IFNβ that is retained longer and possibly slowly
liberated in the CNV lesion may easily interact with its receptor on
the cell surface. Because the efflux of free IFNβ from the target
tissue is relatively faster than that of the conjugate with a higher
molecular weight, the conjugate may continue to release IFNβ slowly,
different from its activity in vitro. Another possibility is that the
internalization and metabolization of IFNβ after binding to the
receptor may be suppressed because of its binding to dextran. The
IFNβ molecule may be released from the receptor without
internalization and again bind to another receptor. Enhanced
concentrations and reuse effects in CNV lesions probably enabled IFNβ
to be remarkably effective in vivo.
IFNβ promoted the progression of RPE cells, whereas it inhibited the
proliferation of HUVECs. Although further in vitro studies should be
undertaken in human RPE cells and choroidal endothelial cells, these
effects of IFNβ may be more helpful in causing the regression of CNV,
because the reconstruction of the outer BRB must have the proliferation
of RPE cells and because RPE cells may have some effects on apoptosis
in developed CNV.
43 44 Oral administration of 200 mg zinc
sulfate daily for 24 months increased the mean serum zinc level by 37%
but did not produce a significant change in red blood cell counts and
hemoglobin levels or other side effects.
47 Rebizak et
al.
48 investigated the feasibility of DTPA-introduced
dextran to prolong the plasma half-life of DTPA and gadolinium, which
has been used for magnetic resonance imaging. Thus, the content of
Zn
2+ and dextran with DTPA residues is expected
to be low enough that no adverse effects occur. Although further animal
studies should be performed to investigate the toxicology, immunology,
dose, and frequency of use of conjugate, this novel passive targeting
system for bioactive proteins may eliminate the problems encountered in
treating patients with this serious ocular disease.