Development and in Vitro Studies of Epidermal Growth Factor

Aug 15, 1996 - P.O. Box 535, S-751 21 Uppsala, Sweden, and Department of Pathology, Akademiska Hospital,. Uppsala University, S-751 85 Uppsala, Sweden...
0 downloads 0 Views 196KB Size
Download PDF
Bioconjugate Chem. 1996, 7, 584−591

584

Development and in Vitro Studies of Epidermal Growth Factor-Dextran Conjugates for Boron Neutron Capture Therapy Lars Gedda,*,† Pa¨r Olsson,† Jan Ponte´n,‡ and Jo¨rgen Carlsson† Division of Biomedical Radiation Sciences, Department of Diagnostic Radiology, Uppsala University, P.O. Box 535, S-751 21 Uppsala, Sweden, and Department of Pathology, Akademiska Hospital, Uppsala University, S-751 85 Uppsala, Sweden. Received February 21, 1996X

A delivery molecule for directed boron neutron capture therapy against epidermal growth factor (EGF) receptor-rich tumors, such as gliomas, squamous carcinomas, and breast cancers, is presented. EGF and sulfhydryl boron hydride (BSH) were covalently coupled to an allylated 70 kDa dextran chain to form a conjugate. Conjugates with low and high substitution rates of BSH, as well as without BSH, were investigated. The conjugate with a low amount of boron had approximately 6 BSH (72 boron atoms) per dextran, while the conjugates with higher amounts had an average substitution of 55 BSH (660 boron atoms) per dextran. The maximum substitution of boron to dextran in a single experiment was over 800 boron atoms. Binding, retention, and internalization of 125I-labeled conjugates were investigated on cultured human glioma cells. Binding of the conjugates was EGF receptor specific, but the amount of BSH coupled to dextran affected specificity, more than the presence of dextran. The nonspecific binding of the conjugates increased with the amount of attached boron. This was partly due to nonspecific adhesion to the plastic in the culture dishes. [125I]EGF-allyldextran with 6 BSH had a binding maximum after 4 h of continuous incubation and thereafter decreased in binding, while [125I]EGF-allyldextran with the higher substitution rate had a slow increase of binding during 24 h. Over 93% of the radioactivity bound to the cells was internalized, but the retention was quite poor. Only one-third of the cell-bound activity was still associated to the cells 4 h after incubation had ended. In conclusion, it is posible to load the conjugates produced with high amounts of boron, and they retained specificity for the EGF receptor and internalized into cultured cells. Theoretical calculations show that about 103 boron atoms per EGF-based conjugate are needed to give a satisfactory therapeutic response. These conjugates are within reach of that level.

INTRODUCTION

A major unresolved problem in cancer treatment is how to kill those tumor cells that remain after local therapy. The present work is an attempt to develop a therapeutically interesting substance planned to bind dispersed tumor cells that overexpress the epidermal growth factor receptor (EGFR). The basic principle was to synthesize a conjugate of EGF, dextran, and boron. The EGF ligand would deliver sufficient amounts of boron to EGFR-rich malignant cells. After boron capture of externally applied neutrons, locally emitted high energy particles would then specifically damage the target cells. The overexpression of EGFR in malignancies, such as gliomas, squamous carcinomas, and breast cancers, is due to gene amplification and/or increased transcription rates (Bigner et al., 1988; Collins and James, 1993). Such tumors are candidates for directed tumor cell specific therapy in which the receptors are the targets. This type of targeting therapy could hopefully become a supplement to conventional cancer therapy. Targeting agents suggested are antibodies (Mendelsohn, 1988; Ozawa et al., 1989; Vollmar et al., 1987; Wikstrand et al., 1995), fragments of antibodies (Reiter et al., 1994), or receptor-ligand conjugates (Andersson et al., 1992; Olsson et al., 1994). Growth factors carrying toxic agents could be superior to antibodies, due to their * Author to whom correspondence should be addressed (email [email protected]; telephone +46 18 183431; fax +46 18 183432). † Division of Biomedical Radiation Sciences. ‡ Department of Pathology. X Abstract published in Advance ACS Abstracts, August 15, 1996.

S1043-1802(96)00047-X CCC: $12.00

smaller size giving better penetration in tumor tissues and the fact that conjugated antibodies often have poor tumor specificity (Vaughan et al., 1987; Wheldon et al., 1988). The noxious agents in growth factor conjugates could be toxins (Cawley et al., 1980; Siegall et al., 1990), radioactive nuclides (Andersson et al., 1992; Capala et al., 1990), or stable nuclides for neutron activation (Barth et al., 1990; Carlsson et al., 1994). Neutron capture therapy was proposed 60 years ago (Locher, 1936) as a form of treatment for brain tumors and other malignancies. The principle is to deliver stable nuclides into the tumor that will react with a beam of low-energy neutrons to produce two high-energy fission particles. The nuclide most often used for boron neutron capture therapy (BNCT) is 10B, which captures a neutron to produce 7Li and 4He. The alpha particle has a range of 9 µm and the lithium particle 5 µm in tissue (Fairchild et al., 1990; Gabel et al., 1987). Since a mammalian cell diameter is about 10 µm, the particle may deposit virtually all of its lethal energy in a single cell that has bound or internalized 10B (Charlton, 1991). The probability for 10B to capture a neutron vastly exceeds that of other atoms in biological tissues, thus minimizing risk of unwanted side effects. Boron can rather easily be incorporated into a multitude of different chemical structures (Barth et al., 1990; Hawthorne, 1993), and a number of compounds have already been tested, such as sulfhydryl boron hydride (BSH), which consists of 12 boron atoms per molecule (Soloway et al., 1967). Gliomas have been a favorite target for experimental BNCT mainly because these malignancies are unusually difficult to cure in spite of the fact that they do not metastasize outside the skull. It is generally hypothesized that their intractability depends on early wide© 1996 American Chemical Society

Bioconjugate Chem., Vol. 7, No. 5, 1996 585

EGF−Dextran Conjugates for BNCT

spread dissemination of cancer cells in the brain, rendering local therapy inefficient. Only if specific targeting by therapeutic molecules is accomplished is there a chance of substantial improvement of cure rate. We have, in the present work, used the strategy of conjugating BSH to EGF via dextran. The use of EGFbased conjugates for BNCT has been described by Carlsson et al. (1994). EGF, consisting only of 53 amino acids (6 kDa), has a compact stable structure. Conjugation via its amino terminus to dextran and toxins without loss of biological activity of the ligand has been described by Andersson et al. (1991), Cawley et al. (1980), and Olsson et al. (1994). Dextran can easily be biochemically modified with a variety of methods to allow coupling to proteins and other substances (Holmberg and Meurling, 1993; Kohn and Wilchek, 1984). Furthermore, dextran stabilizes proteins against degradation (Andersson et al., 1991; Foster, 1975; Melton et al., 1987). In this study, the amino terminus of EGF was modified, using N-succinimidyl-3-(2-pyridyldithio)propionate (SPDP) to introduce a thiol group. The thiol group was then directly coupled to allyl groups present on 70 kDa dextran. Allyl groups were also used to attach various amounts of BSH to the dextran. Radical additions to carbon-carbon unsaturates were used to form sulfurcarbon bonds (Stacey and Harris, 1963). These EGFbased conjugates were tested for their receptor binding specificity and kinetics of internalization and retention in a cultured human malignant glioma cell line, which overexpresses the EGF receptor. EXPERIMENTAL PROCEDURES

Materials. EGF, tissue culture grade, of murine origin (Janssen, Geel, Belgium) was labeled by 125I (Amersham Int plc, Little Chalfont, Buckinghamshire, U.K.). N-Succinimidyl-3-(2-pyridyldithio)propionate (SPDP; Pharmacia, Uppsala, Sweden) was dissolved in ethanol to a concentration of 4 mM and used for thiolation of the growth factor. Dextran (Sigma, St. Louis, MO; industrial grade) with an average molecular mass of 70 kDa was used. The compound disodium undecahydromercapto-closo-dodecarborate (BSH; Na2B12H11SH or sulfhydryl boron hydride) was obtained from Callery Chemical Co., Pittsburgh, PA. Inductively coupled plasma-atomic emission spectrometry (ICP-AES) was used for boron analysis (Lindstro¨m et al., 1994). Elemental analysis of dextrans was performed by Micro Kemi AB, Uppsala, Sweden. NAP-5 columns (Sephadex G-25 gel, Pharmacia, Sweden) were used for separation of conjugates from low molecular weight compounds and Sephadex G-50 gel (Pharmacia, Sweden) for separations of conjugates from unconjugated growth factor. Radioactivity measurements were performed on an automatic gamma counter and liquid scintillation counter (LKB Wallac Oy, Turku, Finland). The cell line U-343 MGaCl2:6 is a subclone of the U-343 MG family of human malignant glioma lines. It has approximately 2 × 105 EGFRs per cell (Westermark et al., 1982). Cells were grown in 35 mm Petri dishes with a cell density of 104-105 cells/cm2. Binding of EGF to U-343 MGaCl2:6 is not cell density dependent in this range (Capala et al., 1990). Ham’s F10 medium supplemented with 10% FCS, L-glutamine (2 mM), and PEST (penicillin 100 IU/mL and streptomycin 100 µg/mL) was used, all from Kebo, Stockholm, Sweden. The cells were incubated at 37 °C in humidified air containing 5% CO2 if not stated otherwise. Methods. Modification of Dextran. Allyl groups were introduced into dextran as described by Holmberg and Meurling (1993). Briefly, 2 g of 70 kDa dextran, 0.5 g of

Scheme 1

NaOH, and 20 mg of NaBH4 were mixed in 15 mL of distilled water at 40 °C. Allyl bromide (3.5 g; 2.5 mL) was added, and the reaction was allowed to proceed at 60 °C for about 3 h. After neutralization, dextran was purified by precipitation with ethanol three times and thereafter lyophilized and stored at -70 °C. The degree of allyl substitution was calculated by elemental analysis of carbon and hydrogen in the allylated and native dextran. Introduction of BSH to Allyldextran (Scheme 1). Two slightly different methods were used. Dextran-BSH(A): 20 mg of BSH and 20 mg of (NH4)2S2O8 were mixed with 20 mg of allyldextran dissolved in 200 µL of distilled H2O. Dextran-BSH(B): 20 mg of BSH and 1 mg of K2S2O8 were mixed with 20 mg of allyldextran dissolved in 200 µL of distilled H2O. Both reactions continued for 2 h at 50 °C with stirring. Thereafter, dextran-BSH was purified from uncoupled BSH and other low molecular weight compounds on a NAP-5 column (equilibrated with distilled H2O). Nine hundred microliters of the high molecular weight fraction was eluted instead of the recommended 1.0 mL to avoid any contaminants of low molecular weight compounds. The amount of boron added to the dextran was determined with ICP-AES, and substitution rates were calculated. Thiol- and 125I-Labeling of EGF. EGF (5 µg in 25 µL of 0.25 M phosphate buffer, pH 7.5) was mixed with a 10-fold molar excess of SPDP for 30 min. Thereafter, 125I (37 MBq) and chloramine T (20 µg in 10 µL of 0.5 M phosphate buffer, pH 7.5) were added and mixed for 1 min. The reaction was interrupted with sodium metabisulfite (50 µg in 25 µL of 0.5 M phosphate buffer, pH 7.5). Thiol groups on [125I]EGF were generated by reducing the introduced SPDP with 10 mM dithiothreitol at pH 3.5 (0.1 M NaAc buffer). The low pH was used to prevent reduction of native disulfide bonds present in the growth factor. Thiolated [125I]EGF is hereafter referred to as [125I]EGF-SH. Excess reagents and low molecular weight substances were separated from macromolecules on a NAP-5 column (equilibrated with distilled water). Nine hundred microliters of the high molecular weight fraction was eluted. Optimization of the [125I]EGF-SH Coupling to Allyldextran. Coupling of [125I]EGF to allyldextran or dextran-BSH (Scheme 2) was done according to the same principle as coupling of BSH to allyldextran. To obtain acceptable yield of conjugate and preserve good binding to cellular EGF receptors, different coupling conditions were applied. [125I]EGF-SH (150 µL; 0.8 µg) and 20 mg of allyldextran were mixed in 650 µL of distilled water with either (a) 1 mg of K2S2O8 at 30 °C, (b) 1 mg of K2S2O8 at 40 °C, (c) 1 mg of K2S2O8 at 50 °C, (d) 1 mg of (NH4)2S2O8 at 50 °C, or (e) 20 mg of K2S2O8 at 50 °C.

586 Bioconjugate Chem., Vol. 7, No. 5, 1996 Scheme 2

Furthermore (f), 900 µL (5 µg) of [125I]EGF-SH and 20 mg of allyldextran were mixed with 20 mg of (NH4)2S2O8 at 50 °C. All reactions were continued for 5 h, and thereafter conjugates were separated from [125I]EGF-SH and other reaction compounds with Sephadex G-50. Fractions of 1 mL were collected and analyzed regarding radioactivity. The fractions containing conjugate were pooled and diluted to a radioactivity concentration of about 1 kBq/mL. These conjugates were tested in receptor binding assays. Preparation of Complete Conjugates. Conjugate I: 900 µL (5 µg) of [125I]EGF-SH was stirred with 900 µL (20 mg) of allyldextran and 1 mg of K2S2O8 at 50 °C for 5 h. Conjugate II: 900 µL (5 µg) of [125I]EGF-SH was stirred with 900 µL (20 mg) of dextran-BSH(B) and 1 mg of K2S2O8 at 50 °C for 5 h. Conjugate III: 900 µL (5 µg) of [125I]EGF-SH was stirred with 900 µL (20 mg) of dextran-BSH(A) and 1 mg of K2S2O8 at 50 °C for 5 h. All of these conjugates were separated from unconjugated growth factor and other low molecular weight reaction compounds with Sephadex G-50. Fractions of 1 mL were collected and analyzed regarding radioactivity. The fractions containing conjugate were pooled, diluted to a radioactivity concentration of 37 kBq/mL, and tested in receptor binding assays and retention studies. Receptor Binding Assays. Cells were presaturated with 0 or 1 µg/mL of nonradioactive EGF and thereafter incubated with culture medium containing either [125I]EGF-SH or any of the conjugates. The incubation time used for [125I]EGF-SH was 25 min, which is the binding maximum for native EGF. The conjugates where incubated for 3 h to enable them to reach near their binding maximum, since EGF-dextran conjugates have shown slower kinetics than EGF (Andersson et al., 1991). After incubation, the cells were washed six times with cold serum-free medium and detached with 0.5 mL of trypsinEDTA for 10 min at 37 °C. One milliliter of medium was added, and the cells were resuspended; 0.5 mL of the cell suspension was used for cell counting, and 1.0 mL was used for radioactivity measurement in a liquid scintillation counter. Displacement. Cells were washed with serum-free medium and incubated in triplicates with 0.5 mL of nonradioactive EGF (concentrations: 0, 10-3-103 ng/mL) and 0.5 mL of conjugate III or [125I]EGF. The cells were incubated during 25 min with [125I]EGF and for 3 h with conjugate III. The amounts added, with respect to EGF, were the same for both conjugate III and [125I]EGF. Culture Petri dishes without any cells were also incubated with conjugate III to examine the nonspecific binding to the dishes. After incubation, cells were thoroughly washed, trypsinized, and counted and the radioactivity was measured, as described above.

Gedda et al.

Binding Patterns. Cells were continuously incubated for 1, 2, 4, 8, and 24 h with conjugates II and III. The radioactivity concentrations for both conjugates were 18.5 kBq/mL. After incubation, cells were thoroughly washed, trypsinized, and counted and the radioactivity was measured, as described above. Retention and Internalization. The method described by Haigler et al. (1980) was used. Cells were incubated for 4 h with conjugates II and III. The radioactivity concentrations for both conjugates were 37 kBq/mL. After incubation, cells were washed six times with cold serum-free medium and then incubated in ordinary culture medium for an additional 0, 4, and 8 h. The cells were washed, and retained cell-associated radioactivity was measured. Thereafter, cells were treated with 0.5 mL of 0.1 M glycine-HCl, pH 2.5, for 6 min at 4 °C, to peel membrane-bound activity off the cells. An additional 0.5 mL of glycine-HCl buffer was used to quickly wash the cells and to collect any remaining activity on the cell surface. The radioactivity, considered to be membrane bound, was measured in a well counter. The cells were then treated with 0.5 mL of 1 M NaOH for 60 min at 37 °C and an additional 0.5 mL of NaOH was used to wash remaining cells off the dishes. The cell-associated radioactivity, considered to be internalized, was measured in a well counter. RESULTS

Introduction of allyl groups into dextran increased its contents of carbon from 43 to 46%, corresponding to a substitution of 6.6 kDa and giving an average molecular mass of about 76.6 kDa. About 160 allyl groups were thus added to each dextran chain, which means that 37% of the glucose units became allylated. BSH coupling to dextran mediated by ammonium persulfate/dextran-BSH(A), resulted in a 10-fold higher boron content compared to mediation by potassium persulfate/dextran-BSH(B). On average, the substitutions corresponded to 660 and 72 boron atoms or 55 and 6 BSH, respectively. With dextran-BSH(A) a maximum of 827 boron atoms per dextran was recorded; that is, 69 BSH or 43% of the allyl groups were substituted with BSH. Variations in coupling conditions for EGF, when coupled to allyldextran, were performed to optimize the yield of conjugate as well as binding to the EGFR on cultured cells. Both of these parameters were taken into consideration when the coupling conditions for further generation of conjugates were chosen. Figure 1 shows how the yield of [125I]EGF-allyldextran increased with elevated temperature and increasing amount of radical initiators. However, as shown in Figure 2, the increase in yield accomplished by stronger or higher concentration of radical initiator was offset by decreased EGFR binding. Conjugation using 20 mg of ammonium persulfate is not shown since the resulting [125I]EGF-allyldextran did not bind at all. Temperature during coupling was only of minor importance for cellular EGFR binding. The best compromise was to use 1 mg of potassium persulfate at 50 °C for coupling [125I]EGF to allyldextran or boronated allyldextran, since this resulted in acceptable yield of conjugate and still good binding to the cells. Separations of complete conjugates on Sephadex G-50 are shown in Figure 3. The peaks from left to right correspond to conjugate, [125I]EGF, and low molecular weight substances. The gel-bed volume was not the same for the different chromatograms; hence, the conjugates were not eluted at the same fractions. The yield of conjugate seemed to be higher for production of conjugate

Figure 1. Chromatograms for separation of reaction mixtures with Sephadex G-50 after coupling of [125I]EGF to allyldextran under different reaction conditions. No boron was coupled in these reactions. (a) 1 mg of K2S2O8 at 30 °C; (b) 1 mg of K2S2O8 at 40 °C; (c) 1 mg of K2S2O8 at 50 °C; (d) 1 mg of (NH4)2S2O8 at 50 °C; (e) 20 mg of K2S2O8 at 50 °C; (f) 20 mg of (NH4)2S2O8 at 50 °C. One milliliter fractions were collected; fractions 22-28, 40-50, and 58-66 corresponded to [125I]EGF-allyldextran, [125I]EGF-SH, and low molecular weight compounds, respectively. Fractions 22-28 were pooled for further tests (Figure 2).

EGF−Dextran Conjugates for BNCT

Bioconjugate Chem., Vol. 7, No. 5, 1996 587

588 Bioconjugate Chem., Vol. 7, No. 5, 1996

Gedda et al.

Figure 2. Binding of [125I]EGF-allyldextran (pooled from fractions 22-28, in Figure 1), produced under different reaction conditions [(a) 1 mg of K2S2O8 at 30 °C; (b) 1 mg of K2S2O8 at 40 °C; (c) 1 mg of K2S2O8 at 50 °C; (d) 1 mg of (NH4)2S2O8 at 50 °C; (e) 20 mg of K2S2O8 at 50 °C] to cultured glioma cells with and without blocking of receptors, using 1 or 0 µg/mL EGF. Incubation time was 3 h, and added radioactivities for a, b, c, d, and e were 0.7, 1.0, 1.4, 1.7, and 1.6 kBq/mL, respectively. Each bar corresponds to the analysis of three dishes. Mean values and maximum variations are given. (f from Figure 1 is not shown since that conjugate had no specific binding at all to cells.)

III compared with production of conjugates I and II. Table 1 shows an overview of the conjugates. Binding of conjugates I, II, and III and [125I]EGF-SH to cultured cells is shown in Figure 4. Specific cell binding decreased when EGF was modified with allyldextran. When boron was bound to allyldextran, binding decreased even more, especially if large amounts of boron were added. Nonspecific binding seemed to increase when boron was bound to the dextran and was about 30% for conjugate II and 40% for conjugate III. These results showed that the presence of BSH had a negative effect on both receptor specific and nonspecific binding to cells by the conjugates. Figure 5 shows the displacement of conjugate III and [125I]EGF with increasing concentrations of native EGF. The radioactivity in culture dishes without cells corresponded to 40% of the radioactivity in culture dishes with cells incubated with both the conjugate and the highest concentration of native EGF. This background activity was subtracted from all dishes with conjugate III in the displacement test. As in Figure 4 the conjugate showed specific binding to the EGFR but it could not fully be displaced by EGF due to nonspecific binding to the cells of about 20%. The figure also shows that the amount of EGF needed to displace the conjugate is less than what was needed to displace [125I]EGF. The binding patterns for continuous incubation with conjugates II and III are shown in Figure 6. Conjugate II bound rapidly to the cells with a maximal binding at 4 h and thereafter slowly decreased. Conjugate III showed quite a different pattern with a peak already after 1 h and a slight decrease followed by a slow increase of binding from 4 h until, at least, 24 h. This indicated

Figure 3. Chromatograms for separation of boron-containing conjugates with Sephadex G-50 gel. One milliliter fractions were collected, and pools were used in further tests (Figures 4-6). Conjugate I (top), fractions 26-33 were pooled. This conjugate did not contain any boron. Conjugate II (middle), fractions 2633 were pooled. This conjugate contained 72 boron atoms on average. Conjugate III (bottom), fractions 24-28 were pooled. This conjugate contained 660 boron atoms on average.

that after 24 h there was more cell-associated activity with conjugate III than with conjugate II. The nonspecific binding for both conjugates was less or the same as shown in Figure 4. The retention curves of conjugates II and III in cultured cells are shown in Figure 7. Both conjugates showed similar patterns, though the binding was higher

Bioconjugate Chem., Vol. 7, No. 5, 1996 589

EGF−Dextran Conjugates for BNCT

Figure 4. Binding of conjugates I-III (pools from Figure 3) and [125I]EGF-SH to cultured glioma cells with and without blocking of receptors, using 1 or 0 µg/mL EGF. Incubation time was 3 h (25 min for [125I]EGF-SH), and added radioactivity was 37 kBq/mL in all cases. Each bar corresponds to the analysis of three dishes. Mean values and maximum variations are given.

Figure 5. Displacement of conjugate III and [125I]EGF with increasing concentrations of EGF. Added amounts of conjugate and [125I]EGF were the same, with respect to the growth factor. Incubation time was 25 min for [125I]EGF and 3 h for conjugate III. Each bar corresponds to the analysis of three dishes. Mean values and maximum variations are given.

Table 1. Schedule over Complete Conjugates Constructed and the Average Amount of Boron Bound to the Dextran conjugate

EGF

dextran

BSH (av)

boron atoms (av)

EGF-SH I II III

[125I]EGF [125I]EGF [125I]EGF [125I]EGF

allyldextran dextran-BSH(B) dextran-BSH(A)

6 55

72 660

for conjugate II. After 4 h, one-third of the original activity was still bound to the cells for both conjugates. After 8 h, only 15 and 25% of the radioactivity from conjugates II and III, respectively, was still associated to the cells. As shown in Table 2, the major part of this association was in the form of internalized radioactivity and