Bioconjugate Chem. 1003, 4, 483-489
483
Basic Fibroblast Growth Factor-Pseudomonas Exotoxin Chimeric Proteins; Comparison with Acidic Fibroblast Growth Factor-Pseudomonas Exotoxin Susan L. Gawlak, Ira Pastan,' and Clay B. Siegall' Bristol-Myers Squibb, Pharmaceutical Research Institute, Molecular Immunology Department, 3005 First Avenue, Seattle, Washington 98121. Received May 19, 1993"
We have constructed growth factor-toxin chimeric molecules composed of basic fibroblast growth factor (bFGF) and two different binding mutant forms of Pseudomonas exotoxin termed bFGF-PE40 and bFGF-PE4E KDEL. The chimeric molecules were expressed in Escherichia coli and localized to both inclusion bodies and the spheroplast cytoplasm. The bFGF-toxin fusion protein that was isolated and purified from inclusion bodies was 3-fold more active in inhibiting protein synthesis than that purified from spheroplast cytoplasm. Immunoreactivity of purified bFGF-toxin fusion protein to anti-bFGF antibodies was similar to that of native bFGF, as determined by ELISA analysis. A variety of carcinoma cell lines were sensitive to bFGF-PE40 and bFGF-PE4E KDEL, including H3396 (breast), Hep G2 (hepatocellular), and A431 (epidermoid). The concentration of chimeric toxin that inhibited protein synthesis by 50% (ECa) was 110, 70, and 18 ng/mL for bFGF-PE40 and 15, 1, and 18 ng/mL for bFGF-PE4E KDEL. In comparison with fusion-toxins composed of acidic fibroblast growth factor (aFGF) and either PE40 or PE4E KDEL, bFGF-PE40 and bFGF-PE4E KDEL were similarly cytotoxic on most cell lines tested. Human aortic smooth muscle cells were sensitive to both bFGF and aFGF toxin fusion proteins. However, human aortic endothelial cells were sensitive to the bFGF-toxins but were resistant to both aFGF-toxin forms. Time course studies showed that bFGF-PE40 needed a 4-6-h exposure to target cells for peak inhibition of protein synthesis on both MCF-7 and A431 cells, while aFGF-PE40 was almost fully active within a 2-h incubation. The addition of heparin competed for the cytotoxic activity of bFGF-PE40 but not for aFGF-PE40 on MCF-7 and A431 cells. Cytotoxic forms of bFGF and aFGF may be useful in eliminating populations of cells that express both or one of these FGF receptors.
INTRODUCTION Basic fibroblast growth factor (bFGF, FGF-2) belongs to the family of FGF proteins which includes acidic fibroblast growth factor (aFGF, FGF-11, the oncogene products int-2 (FGF-3),and hstlK-fgf (FGF-41,and FGF5 , FGF-6, and keratinocyte growth factor (KGF, FGF-7) ( 1 , 2). Basic FGF has multiple activities, including stimulating some cell types to proliferate and others to differentiate (3,4). Basic FGF has also been shown to be a potent angiogenic factor (5). There are a t least five different genes that encode high affinity bFGF receptors (bFGFR). The receptors are transmembrane and display two or three immunoglobulinlike domains extracellularly and a tyrosine kinase domain intracellularly (6-11). There is evidence for the involvement of bFGF and its receptor in certain human tumors (12-1 7). Actively growingvasculature that suppo* tumor growth is dependent on growth factors including bFGF (18, 19). Basic FGF binds to both high and low affinity receptors (heparan sulfate molecules found in the extracellular matrix of cells and on cell surfaces) (20, 21). Several laboratories have used cell-specific reagents, including antibodies and growth factors, to deliver protein toxins to target cells (22, 23). These hybrid molecules bind to cells that display the receptor or antigen of choice
* Author to whom correspondence should be addressed.
+ Laboratory of Molecular Biology, National Cancer Institute, National Institutes of Health, 9OOORockville Pike, Bldg 37, Room 4E16,Bethesda, MD 20892. 0 Abstract published in Advance ACS Abstracts, October 1,
1993. 1043-1a
and are cytotoxic through the action of the protein toxin. We have previously reported on aFGF-Pseudomonas exotoxin molecules that are specifically toxic to tumor cells (24),smooth muscle cells (251,and endothelial cells (26)expressing aFGF receptors. Conjugates composed of bFGF and the plant protein toxin saporin (SAP) have been shown to kill many cell types including tumor cells (27-29), pancreatic islet cells (30),and vascular cells (31, 32). Pseudomonas exotoxin (PE), an ADP-ribosyl transferase, has three distinct domains (33, 34). Domain I is responsible for the binding of P E to its cell surface receptor. Domain I1 is essential for the translocation of P E to the cytosol, and domain I11 contains the ADP-ribosylation activity of PE, which is responsible for its cytotoxic effect. The enzymatic activity of domain I11 inhibits protein synthesis by catalyzing the transfer of the ADP-ribosyl moiety of oxidized NAD onto elongation factor 2. Elongation factor 2 is required for the addition of amino acids onto developing protein chains and its inactivation results in cell death. Many growth factors and cytokines, including interleukin-2, insulin-like growth factor I, transforming growth factor a,and interleukin-6, have been fused to Pseudomonas exotoxin and shown to be cytotoxic to the specific target chosen (35-39). In this report, chimeric proteins composed of bFGF and P E were made by fusing the cDNA encoding bFGF to both a gene fragment encoding a truncated form of PE, PE40 (devoid of domain I), and the gene encoding a binding mutant form of P E (containing mutations in domain I that destroy native receptor binding) and by expressing the gene fusions in Escherichia coli. The cytotoxic portion
~ ~ ~ ~ ~ ~ ~ ~ o ~ -0 o1993 ~ aAmerican ~ ~ o Chemical ~ . o o Society / o
484
Gawlak et ai.
Bloconjmte Chem., Vol. 4, No. 6, 1993
A.
Y
7
YY
I
pbFQF
bFQF
u
pCSF40
I
E =EcORI
I
bFQF
pBWlO
PE4E KDEL
pBW11
B. bFGF LINKER PE40 1 2 34 153154155 1 2 3 253 254 608609610611 612613 M A A G...........A K S -----K L M-----A E E G G............ P R E D L K
bFGF LINKER PE4E KDEL 1 2 34 153154155 1 2 3 4 5 608 M A A G...........A K S ----K L M-----A E E A F............P K
D
E
L
Figure 1. (A) Schematic diagram for the construction of plasmid pBW10, encoding the basic FGF-PE40 protein and plasmid pBWll encoding the basic FGF-PE4E KDEL protein and (B) linear structure of the bFGF-PE40 and bFGF-PE4E KDEL fusion proteins. Details are shown at the junction between bFGF and PE. Mutated areas are also shown.
of the P E molecule was able to enter the cells through bFGF receptors and was cytotoxic to various cell types expressing the bFGF receptor. We compared the cytotoxic activities of the bFGF-toxin molecules to aFGF-toxin molecules and determined their differential cytotoxic activities toward a variety of human cancer and vascular cells. EXPERIMENTAL PROCEDURES Reagents. The human basic fibroblast growth factor gene encoded in plasmid pbFGF, bFGF protein, and rabbit anti-basic FGF neutralizing antibody were purchased from (R & D Systems, Minneapolis, MN). Q-Sepharose and heparin-Sepharose were purchased from Pharmacia (Uppsala, Sweden). Immunoblots were performed using rabbit anti-bovine basic FGF antibodies (Upstate Biotechnology, Inc., Lake Placid, NY). Rabbit polyclonal anti-PE antibody was kindly supplied by Dr. David FitzGerald, National Institutes of Health (Bethesda, MD). ABC immunoblot kits were purchased from Vector Laboratories (Burlingame, CA). IPTG (isopropyl l-thio-P-D-galactopyranoside) was purchased from Strategene (La Jolla, CAI. PHIleucine was purchased from New England Nuclear (Boston, MA). Cell Culture. All tumor cells were from human carcinomas. MCF-7 (breast), A431 and KB (epidermoid), and Hep G2 (hepatocellular) cells were obtained from ATCC (Rockville, MD). RCA (colon) cells were obtained from M. Brattain, Baylor University, (Houston, TX), and L2987 (lung), H3606 (melanoma), H3396 (breast), and H3719 (colon) were obtained from I. Hellstrom, BristolMyers Squibb (Seattle, WA). The tumor cell lines were cultured in RPMI 1640 supplemented with 10% fetal
bovine serum (FBS), 2 mM L-glutamine, and 50 units/mL penicillin/streptomycin a t 37 OC with the exception of Hep G2 cells, which were cultured in DMEM. Human aortic smooth muscle cells and endothelial cells were purchased as primary cultures from Clonetics, Inc. (San Diego, CA) and cultured in smooth muscle growth medium (SMGM) or endothelial growth medium (EGM). SMGM is composed of modified MCDB 131 media, 10 ng/mL EGF, 2 ng/mL bFGF, 5% FBS, 1 pM dexamethasone, 0.05 mg/ mL gentamycin, 0.05 pg/mL amphotericin-B. EGM is composed of modified MCDB 131media, 10 ng/mL EGF, 1.0 pg/mL hydrocortisone, 2 % FBS, 0.4% bovine brain extract, 0.05 mg/mL gentamycin, and 0.05 pg/mL amphotericin-B. Construction of bFGF-PE40 and bFGF-PE4E KDEL. The expression plasmids encoding bFGF-PE40 and bFGF-PE4E KDEL were constructed as shown in Figure 1. The bFGF gene was amplified using PCR, resulting in a 500 bp DNA fragment. The 5’PCR primer 5’-AGC-TTA-CCT-CAT-ATG-GCA-GCC-GGG-3’ incorporated an NdeI restriction site and the 3’ PCR primer
5’-GAA-TTC-GGA-TCC-AAG-CTT-GCT-CTTAGC-AGA-3‘ a HindIII site. The PCR-amplified DNA fragment was digested with NdeI and HindIII, followed by ligation into either a similarly digested plasmid encoding PE40 or PE4E KDEL. In this case, the PE40 and PE4E KDEL plasmids contained the aFGF gene (pCS F40 and pCS F4E KDEL, respectively), which was excised by the NdeIIHindIII digestion and replaced with the bFGF gene by ligation. The expression plasmids encoding bFGFPE40 and bFGF-PE4E KDEL, were termed pBW 10 and pBW 11, respectively. PE4E KDEL encodes a P E gene in which the binding domain is mutated a t four residues
Bioconjugate Chem., Vol. 4, No. 8, 1993 485
bFGF-Toxin vs aFGF-Toxin
(57, 246, 247, and 249) to glutamate, thereby abrogating normal P E binding. The KDEL portion of the mutant PE molecule encodes the endoplasmic reticulum retention sequence and has been shown to increase specific cytotoxicity toward certain target cells (40). Expression and Isolation of RecombinantProteins. Basic FGF-PE40 and bFGF-PE4E KDEL were expressed in E. coli BL21 (XDE3)by culturing in Superbroth (Digene, Inc., Silver Spring, MD) containing 75 pg/mL of ampicillin a t 37 "C until Azw is approximately 1.3, followed by induction with isopropyl 1-thio-8-D-galactopyranoside (IPTG) to a final concentration of 1mM. The cells were harvested after 90 min, washed in sucrose buffer (20% sucrose, 30 mM Tris-HC1 (pH 7.4), 1 mM EDTA), and osmotically shocked in ice-cold HzO. The cell pellet was resuspended in TE buffer (50 mM Tris, pH 8.0/1 mM EDTA), sonicated three times (30 s each), and centrifuged a t 30 000 rpm (Beckman Ti45 rotor) for 45 min. The pellet (inclusion bodies) was denatured in 7 M guanidine hydrochloride, 20 mM Tris-HC1 (pH 7.4), and 1 mM EDTA, sonicated as before, and refolded by rapid dilution in phosphate-buffered saline (PBS) supplemented with 0.4 M L-arginine. Alternatively, the supernatant (spheroplast cytoplasm) was used as the source of the protein. The bFGF-toxin proteins were extensively dialyzed against 0.02 M Tris-HC1 (pH 7.4) and were purified by anion-exchange (Q-Sepharose) with a Pharmacia fast protein liquid chromatography (FPLC) system, as described previously (24). Heparin-AffinityChromatography. Fractions pooled from Q-Sepharose columns were further purified by heparin affinity chromatography. Heparin-Sepharose beads (Pharmacia) were washed extensively with 0.4 M NaC1/0.01 M Tris HC1, pH 7.4. The sample was adjusted to 0.4 M NaCl and batch adsorbed to the heparinSepharose (1.5 g dry weight/100 mg bFGF-toxin) for 18 h a t 4 "C (with gentle rocking). The bFGF-toxin bound beads were washed stepwise with 0.4 M NaCV0.01 M TrisHC1 (pH 7.4) and 0.6 M NaCV0.01 M Tris HC1 (pH 7.4). The bound bFGF-PE forms were eluted with 1.2-2.5 M NaCU0.01 M Tris HC1 (pH 7.4). Recovery of purified bFGF-PE40 after heparin affinity was 2.5% of the crude protein applied to the column. In comparison, the recovery of aFGF-PE40, was approximately 10% of the crude protein. Inhibition of Protein Synthesis Assay. Tumor cells (106 cells/mL) in growth media were added to 96-well flat bottom tissue culture plates (0.1 mL/well) and incubated at 37 "C for 16 h. Dilutionsofgrowthfactor-toxins (bFGFPE40, bFGF-PE4E KDEL, aFGF-PE40, and aFGFPE4E KDEL) were made in leucine-free RPMI (assay media) and 0.1 mL was added to each well for 20 h a t 37 "C. Each dilution was done in triplicate. The cells were pulsed with i3H1leucine (1pCi/well) for an additional 4 h a t 37 "C. The cells were lysed by freezing-thawing and harvested using a Tomtec cell harvester (Orange, CT). Incorporation of I3H1leucine into cellular protein was determined using an LKB Beta-Plate liquid scintillation counter. A time course study was also done in which the FGF-toxins were incubated with the cells for 1,2,4, and 6 h before removal and replacement with fresh growth media and further incubated for a total of 20 h prior to the measurement of cellular protein synthesis as described above. The effects of the FGF-toxin proteins on protein synthesis of aortic endothelial cells and aortic smooth muscle cells were also studied. Primary human smooth muscle cells (SMC) and primary human endothelial cells
(EC) were cultured in SMGM or EGM, respectively. The FGF-toxins dilutions were prepared in EGM or SMGM. The incorporation of I3H1leucineinto cellular protein was done exactly as described for the tumor cells. bFGF ELISA. Two-fold dilutions of bFGF, bFGFPE40, and bFGF-PE4E KDEL in PBS (50 pLlwel1) were incubated in Dynatech Immulon I1 96 well plates for 2 h a t 37 "C. Acidic FGF-PE40 and aFGF-PE4E KDEL were similarly prepared and used for comparison. The plates were washed five times in PBS, blocked with PBS, 3% gelatin for 2 h a t 37 "C, and washed five times in PBS. Polyclonal rabbit anti-bFGF (human) antibody (Collaborative Research Inc., Bedford, MA), 1 pg/mL in PBS, 1%BSA (50 pL/well) was incubated for 1 h a t 37 "C and washed in PBS, 0.2% Tween 20, 1%BSA (PTB buffer). Anti-rabbit IgG alkaline phosphatase conjugate (Pierce, Rockford, IL) was added to each well (50 pL of a 1:lOOO dilution in PBS, 1%BSA) and incubated a t 37 "C for 1 h. Plates were washed five times with PTB, three times with phosphatase buffer (PBS, 75 mM Tris, 0.1 M NaC1, 5 mM MgC12, pH 9.41, and reacted with p-nitrophenyl phosphate (Sigma, St. Louis, MO), freshly made in phosphatase buffer (100 pL), for 30 min a t 37 "C. The reaction was stopped by the addition of 2 N NaOH (100 pL/well). The plates were analyzed a t 405 nm using a microplate reader (Molecular Devices, Menlo Park, CAI.
RESULTS Construction of bFGF-Pseudomonas Exotoxin Gene Expression Plasmids. Plasmids containing synthetic DNA encoding the human bFGF gene and two mutated forms of the gene coding for Pseudomonas exotoxin were constructed as described in Experimental Procedures (Figure 1A). The plasmids pBWlO (bFGF-PE40) and p B W l l (bFGF-PE4E KDEL) were expressed under control of the bacteriophage T7 promoter (41). Both expression plasmids encode the mature bFGF gene (amino acids 1-155), followed by a three amino acid linker that added the residues lysine, leucine, and methionine (KLM), and one of two mutant P E forms. PE40 encodes amino acids 1,2,3,and 253-613 of the native PE protein sequence (42). PE4E KDEL encodes amino acids 1-608 with residues KDEL on the carboxyl terminus of the protein sequence and glutamate substitutions a t codons encoding residues 57,246,247, and 249 (43). The carboxyl terminal KDEL sequence represents the endoplasmic reticulum retention sequence and is used in place of the native PE terminus, REDLK. The addition of the KDEL sequence has previously been shown to increase the cytotoxic activity of P E against target cell lines (40). The four-glutamate substitution (all in PE domain I) inactivates normal P E binding activity and makes this molecule analogous to the PE40 molecule in which domain I is deleted (44). A depiction of the linear protein structure of bFGF-PE40 and bFGF-PE4EKDEL, detailing the fusion protein junction and mutated PE regions, is shown in Figure 1B. Expression and Purification of bFGF-Toxin Proteins. bFGF-PE40 and bFGF-PE4E KDEL fusion proteins were expressed in E. coli upon IPTG induction. The fusion proteins localized to the soluble (spheroplast cytoplasm) and insoluble fraction (inclusion bodies) of the spheroplast. The insoluble fraction was denatured in guanidine hydrochloride and renatured by rapid dilution in PBS containing L-arginine as described in Experimental Procedures. Both the soluble and renatured protein from the inclusion bodies were dialyzed against 20 mM TrisHC1, pH 7.4, and initially purified using Q-Sepharose (anion-exchange) chromatography. The peak fractions
486
B.
A. 1
Gawlak et al.
Bioconlugate Chem., Vol. 4, No. 6, 1993
2 3 4 5 6 M k D a
1
2 3 4 5 6
kDa
Table I. Cytotoxic Activities of bFGF-PE40 and aFGF-PE40 Chimeric Proteins on Various Carcinoma Cell Lines
“I
i
i1
2009760-
-97 -68
43-
-29
-43
29 18-
Figure 2. SDS-PAGE analysis of bFGF-PE forms following
heparin-Sepharose purification, stained with Coomassie Blue: lane 1, bFGF-PE40 (insoluble-refoldedfusion protein); lane 2, bFGF-PE40 (soluble); lane 3, bFGF-PE4E KDEL (insolublerefolded fusion protein); lane 4, bFGF (native); lane 5, PE40; lane 6, aFGF-PE40 (insoluble-refolded fusion protein). 0.4
In
,
A. I
68
33
16.5
8.25
4.13
2.06
1.03
0
2 0
B. 0.3
02
0.1
0.0 68
33
16.5
8.25
4.13
2.06
1.03
bFGF equivalents, nM
Figure 3. Immunoreactivityanalysisof (A)bFGF and (B)soluble
and insoluble-refolded bFGF-PE40 and aF’GF-PE40, on ELISA plates coated with polyclonal rabbit anti-bFGF antibody and probed with anti-rabbit IgG antibody. Soluble bFGF-PE40 represents fusion protein that was isolated and purified from the soluble E. coli spheroplast cytosol. Insoluble-refolded bFGFPE40 represents fusion protein that was isolated from the E. coli inclusion bodies, denatured, renatured, and purified. were pooled and the proteins were further purified using heparin affinity chromatography. Purified bFGF-PE40 and bFGF-PE4E KDEL migrated a t 55 and 81 kDa, respectively, equivalent to their predicted molecular sizes as visualized by Coomassie-stained SDS-PAGE (Figure 2). The majority of the bFGF-PE40 fusion protein eluted between 1.2 and 1.5 M NaC1. The fractions eluting a t 1.2 and 1.5 M NaCl were pooled, concentrated, and quantitated by Bradford analysis (45). bFGF ELISA. To determine that the bFGF fusion proteins were folded in the correct fashion, anti-bFGF antibody immunoreactivity assays were performed. Comparison of the same molar amounts of bFGF-PE40, bFGFPE4E KDEL, and bFGF (native) were used in the assay. The bFGF domains of native bFGF (Figure 3A) and bFGFPE40 from both soluble and insoluble preparations (Figure 3B) were functional, as determined by their immunoreactivity to anti-bFGF antibody in an ELISA experiment. The immunoreactivityof bFGF-PE40 that was renatured followingguanidine hydrochloride treatment was increased relative to the fusion protein that was purified from the soluble fraction of the spheroplasts. The bFGF-toxin
carcinoma line, type MCF-7 (breast) L2987 (lung) KB (epidermoid) A431 (epidermoid) Hep G2 (hepatoma) RCA (colon) H3396 (breast) H3719 (colon)
bFGF-PEA0
dGF-PE40
30 250 350 18 70 200 110 500
30 150 75 10
20 900
200 400
Table 11. Cytotoxic Activities of bFGF-PE4E KDEL and aFGF-PE4E KDEL Chimeric Proteins on Various Cell Lines
carcinoma line, type bFGF-PE4E KDEL aFGF-PE4E KDEL 70 25 MCF-7 (breast) 50 KB (epidermoid) 150 18 15 A431 (epidermoid) 1 5 Hep G2 (hepatoma) 100 300 RCA (colon) 200 40 H3606 (melanoma) 15 30 H3396 (breast) 40 35 H3719 (colon) fusion proteins were immunoreactive to anti-bFGF antibody as well as bFGF (native). bFGF-PE4E KDEL was also immunoreactive with the anti-bFGF antibody equivalent to both native bFGF and bFGF-PE40 (data not shown). The assay was specific for bFGF, since aFGF (native) and aFGF-PE40 were not immunoreactive with the bFGF antibody in the ELISA assay. Cytotoxic Activity of bFGF-Toxin Fusion Proteins toward Cancer Cells. We compared the cytotoxic activity of bFGF-PE40 that was isolated from spheroplast cytoplasm and the inclusion bodies. Both proteins were purified with the same techniques (Q-Sepharose and heparin-Sepharose) and were equivalent in purity based on SDS-PAGE analysis (Figure 2). The fusion protein from inclusion bodies that was denatured with guanidine hydrochloride and refolded was 3-fold more potent in inhibiting protein synthesis in A431 cells than was the fusion toxin that was found in the soluble portion of the spheroplast compartment (10 ng/mL for refolded versus 30 ng/mL for soluble). Since renatured fusion protein was more active in both binding (Figure 3B) and cytotoxicity assays, all further experiments described in the paper use refolded material. We assayed a variety of cancer cell lines to determine protein synthesis inhibition activity of bFGF-PE40 (Table I). A431 cells were the most sensitive carcinoma line tested with an ECW value of 18ng/mL. MCF-7 breast carcinoma cells were also quite sensitive to bFGF-PE40, with an E C a value of 30 ng/mL. Hep G2 (hepatocellular), H3396 (breast), L2987 (lung), and KB (epidermoid) carcinoma cells were less sensitive, with ECMvalues of 70,110,250, and 350 ng/mL, respectively. The cytotoxic activity of bFGF-PE40 was compared to that of aFGF-PE40 (Table I). While the protein synthesis inhibition activity of both bFGF-PE40 and aFGF-PE40 forms were equivalent for many lines, all were slightly more sensitive to aFGF-PE40 except for H3396 and MCF-7, both breast carcinoma lines, which were more sensitive to bFGF-PE40. We also determined the cytotoxic activity of bFGFPE4E KDEL against the same cell lines with the addition of the melanoma line H3606. As shown in Table 11,bFGFPE4E KDEL was active in inhibiting protein synthesis in
Bioconlugate Chem., Vol. 4, No. 6, 1993 407
bFW-Toxln vs aFGF-Toxln c
a
7
I
z
lM) EO
a
z 0 t E .1
10
1
100
1000
nglml
Figure 4. Sensitivity of vascular cells to FGF-toxin fusion proteins: (A) primary human aortic smooth muscle cells and (B) primary human aortic endothelial cells. Both vascular cell lines were used at passage 5. Legend: bFGF-PE40 ( O ) ,bFGF-PE4E KDEL ( 0 ) ,aFGF-PE40 (0),aFGF-PE4E KDEL (B). 300
-E 5
.1
1
10
100
1000
.1
1
10
100
1000
I
bFOF-PE40, ng/ml
aFGF-PE40, ng/ml
Figure 6. Effect of heparin on the cytotoxic activity of bFGF-
PE40 versus aFGF-PE40 against MCF-7 cells. The cells were incubated with toxins in the presence of absence of heparin for 20 h followed by a 4-h incubation with [SHIleucine: (A) bFGFPE40 and (B) aFGF-PE40, with no heparin (B), 5 units/mL heparin (O),50 units/mL heparin (0).
300
A.
1
0.
2
4
6 2 0
1
2
4
6
2
0
Time (hours)
Figure 5. Cytotoxicity time course of bFGF-PE40 (0)and
aFGF-PE40 (B) on (A) MCF-7 and (B) A431 cells. FGF-toxins were incubated with tumor cells for 1,2,4, and 6, and 20 h. For the 1-6 h time points, supernatants were removed, fresh media was added, and cells further incubated for a total of 20 hours. Cells were then pulse labeledfor 4 h with [SHIleucineto determine cellular protein synthesis.
all the cell lines tested. Hep G2 cells were the most sensitive line tested with an ECw value of 1ng/mL. A431, H3396, and MCF-7 cells were all sensitive to bFGF-PE4E KDEL a t EC60 values less than 25 ng/mL. aFGF-PE4E KDEL was equally cytotoxic to Hep G2 cells and less toxic to H3396 breast carcinoma and H3606 melanoma cells than was bFGF-PE4E KDEL (Table 11). Sensitivity of Vascular Cells to FGF-Toxin Fusion Proteins. We also compared the sensitivity of primary human smooth muscle and endothelial cells to both bFGFand aFGF-toxin forms. Human aortic smooth muscle cells (passage 5) were more sensitive to bFGF-PE40 than to aFGF-PE40, and equally sensitive to bFGF-PE4E KDEL and aFGF-PE4E KDEL (Figure 4A). The human endothelial cells were very sensitive to both bFGF-PE40 and bFGF-PE4E KDEL (ECw values of 10 and 2 ng/mL, respectively). Alternatively, the human endothelial cells were quite resistant to both aFGF-PE40 and aFGF-PE4E KDEL (Figure 4B). Exposure Time for Cytotoxic Activity. To determine the time of exposure necessary for cytotoxic activity of bFGF-PE40 on target cells, we performed a time course protein synthesis inhibition assay. A431 and MCF-7 cells were incubated with bFGF-toxin for various time periods, washed free of the toxin, replenished with fresh growth media, and incubated for a sum total of 20 h. MCF-7 cells, treated with bFGF-PE40 were relatively insensitive to the toxin a t 1- or 2-h exposure times, and even a 6-h exposure time was still not as active as 20 h of exposure (Figure 5A). Alternatively, aFGF-PE40 was almost fully active in inhibiting protein synthesis in MCF-7 cells, with only a 1-h exposure time. With a 2-h exposure, protein synthesis was maximally inhibited. Using A431 cells, bFGF-PE40 was near full activity with a 4-h exposure (Figure 5B), as compared to greater than 6 h against MCF-7 cells. Acidic FGF-PE40 was almost
fully active with a 2 h exposure to the toxin, similar to the data obtained for activity in MCF-7 cells. Effect of Heparin on FGF-Toxins. Since heparin has a biological effect on the FGF family of proteins, we assayed whether heparin would potentiate, block, or have no effect on the cytotoxic activity of the FGF-toxin proteins. Coincubation with heparin (50 units/mL) blocked the ability of bFGF-PE40 to inhibit protein synthesis in MCF-7 cells (Figure 6A). Heparin a t 5 units/mL blocked the cytotoxic activity of bFGF-PE40 to a lesser degree than 50 units/mL. Alternatively, coincubation of heparin (at both 5 and 50 units/mL) and aFGF-PE40 produced a small increase in the inhibition of protein synthesis in the tumor cell line (Figure 6B). DISCUSSION We have constructed and characterized fusion proteins that are cytotoxic toward cells expressing bFGF receptors. These recombinant toxin molecules were produced by expressing gene fusions encoding bFGF and binding mutant forms of P E in E. coli. The hybrid bFGF-PE molecules accumulated intracellularly and were found in both soluble and insoluble forms. Both forms were purified by a two-step procedure using anion-exchange followed by heparin-affinity chromatography. The insoluble form was denatured and renatured prior to purification. Following purification, refolded bFGF-PE40 fusion protein was found to bind anti-bFGF antibody better than soluble bFGF-PE40 (Figure 3B), and to be more active in inhibiting protein synthesis than the intracellular, soluble form. The fusion protein isolated from the spheroplast cytoplasm may have contained mixed disulfides that reduced specific binding. With mixed disulfides in the fusion protein, the internalization ability of the growth factor may also have been affected. The heparin binding activity of the recombinant bFGF-toxin fusion proteins was retained, as evidenced by the heparin affinity of the material. However, retention of heparin affinity is no guarantee of bFGF receptor binding affinity. The bFGF-toxin fusion proteins were immunoreactive with anti-bFGF antibodies in an ELISA based assay (Figure 3). From this analysis, the bFGF fusion proteins display the bFGF epitope necessary for interaction with the anti-bFGF antibody. Thus, in this respect the refolded protein is similar in immunoreactivity to native bFGF. Specificity for bFGF was demonstrated since aFGF-toxin fusion proteins were not immunoreactive with the antibFGF antibody. Basic FGF-PE40 was effective a t inhibiting protein synthesis in a variety of tumor cell lines. In comparison with aFGF-PE40, bFGF-PE40 was generally found to be
400
Gawlak et al.
Bloconjugate Chem., Vol. 4, No. 6, 1993
a less potent cytotoxic reagent toward cancer cells with the exception of breast carcinoma and melanoma cell lines (Tables I and 11). This may be due to expression levels of the high-affinity receptors for the two FGF forms. The difference in activity may also be due to rates of internalization, as suggested by the time-course experiments (Figure 5A,B). The protein synthesis inhibition activity of aFGF-PE40 is near maximal within 2 h of exposure to target cells, as compared to 6 h for bFGF-PE40. Perhaps the molecular interaction of bFGF and the high-affinity bFGF receptor requires a longer time period to promote internalization than does the interaction of aFGF and its receptor. Since bFGF has been shown to be a potent angiogenic agent and its receptor is expressed on vascular cells (46, 47))we investigated its ability to inhibit protein synthesis of endothelial and smooth muscle cells. Both vascular cell types were sensitive to bFGF-PE40 and bFGF-PE4E KDEL (Figure 4). Similar cytotoxic activities have also been reported for bFGF-SAP (31,32). Alternatively, the two aFGF-toxin forms were active against smooth muscles cells but relatively inactive, as compared to the two bFGFP E molecules, against human endothelial cells (Figure 4B). Basic FGF can be internalized through both high and low affinity receptors (20,21). The low affinity receptors can be blocked by the addition of heparin (21). We found that heparin was able to compete with much of the ability of bFGF-PE40 to inhibit protein synthesis in MCF-7 breast carcinoma cells (Figure 6A). This suggests that binding to low-affinity receptors is important to facilitate optimized internalization. Against the same MCF-7 cells, aFGF-toxin forms were more potent in inhibiting protein synthesis in the presence of heparin (Figure 6B). It has been previously shown that heparin enhances the binding of aFGF to its high-affinity receptor (48). We conclude that bFGF linked to recombinant forms of PE are specifically cytotoxic to receptor-bearing target cells. We suggest that these chimeric toxins could be used to select cells deficient in FGF receptors to help identify the role of this receptor in controlling growth and differentiation. ACKNOWLEDGMENT We thank E. Kozlowski for oligonucleotide synthesis and Drs. P. Fell and K. E. Hellstrom for helpful suggestions and continued support. LITERATURE CITED (1) Burgess, W. H., and Maciag, T. (1989) The heparin-binding
(fibroblast) growth factor family of proteins. Annu. Rev.
Biochem. 58, 575-606. (2) Basilico, C., and Moscatelli, D. (1992) The FGF family of growth factors and oncogenes. Adu. Cancer Res. 59,115-165. (3) Gospodarowicz, D. (1989) Fibroblast growthfactor. Crit. Reu. Ongogen. 1, 1-26. (4) Rifkin, D. B., and Moscatelli, D. (1989) Recent developments in the cell biology of basic fibroblast growth factor. J. Cell Biol. 109, 1-6. (5) Folkman, J., and Klagsbrun, M. (1987) Angiogenic Factors. Science 235,442-447. (6) Lee, P. D., Johnson, D. E., Cousens, L. S., Fried, V. A., and Williams, L. T. (1989) Purification and complementary DNA
cloning of a receptor for basic fibroblast growth factor. Science
245, 57-60. (7) Dionne, C. A., Crumley,G., Bello, F., Kaplow,J. M., Searfoss, G., Ruta, M., Burgess, W. H., Jaye, M., and Schlessinger, J. (1990) Cloning and expression of two distinct high-affinity
receptors cross-reactingwith acidic and basic fibroblast growth factors. EMBO J. 9, 2685-2692.
(8) Keegan, K., Johnson, D. E., Williams, L. T., and Haymas, M. J. (1991) Isolation of an additional member of the fibroblast
growth factor receptor family, FGFR3. Proc. Natl. Acad. Sci. U.S.A.88, 1095-1099. (9) Partanen, J., Makela, T. P., Eerola, E., Korhonen, J., Hirvonen,H., Claesson-Welsh,L.,and Alitalo, K. (1991)FGFR4, a novel acidicfibroblast growth fador receptor with a distinct expression pattern. EMBO J . 10, 1347-1354. (10) Pasquale, E. B. (1990). A distinctive family of embryonic protein-tyrosine kinase receptors.Proc. Natl. Acad. Sci. U.S.A. 87,5812-5816. (11) Johnson, D. E., and Williams, L. T. (1993). Structural and
functional diversity in the FGF receptor multigene family. Adu. Can. Res. 60, 1-41. (12) Halaban, R., Kwon, B. S., Ghosh, S., Delli Bovi, P., and Baird, A. bFGF as an autocrine growth factor for human melanomas. (1988). Oncogene Res. 3, 177-186. (13) Armstrong, E., Vainikka, S., Patanen, J., Korhonen, J., and Alitalo, R. (1992). Expression of fibroblast growth factor receptors in human leukemia cells. Cancer Res. 52,2004-2007. (14) Hattori, Y., Odagiri, H., Nakatani, H.,Miyagawa, K., Naito,
K., Sakamoto, H., Katoh, O., Yoshida, T., Sugimura, T., and Terada, M. (1990).K-sam,an amplified gene in stomach cancer, is a member of the heparin-binding growth factor receptor genes. Proc. Natl. Acad. Sci. U.S.A. 87, 5983-5987. (15) Luqmani, Y. A., Graham, M., and Coombes R. C. (1992) Expression of basic fibroblast growth factor, FGFRl and FGFR2 in normal and malignant human breast, and comparison with other normal tissues. Br. J. Cancer. 66,273-280. (16) Zagzag,D., Miller, D. C., Sato, Y .,Rifkin,D. B., and Burstein, I. (1990) Immunohistochemical localization of basic fibroblast growth factor in astrocytomas. Cancer Res. 50, 7393-7398. (17) New, B. A., and Yeoman, L. C. (1992) Identification of basic fibroblast growth factor sensitivity and receptor and ligand expression in human colon tumor cell lines. J. Cell. Phys. 150, 320-326. (18) Folkman, J., Watson, K., Ingber, D., Hanahan, D. (1989).
Induction of angiogenesis during the transition from hyperplasia to neoplasia. Nature 3339, 58-61. (19) Weidner, N., Semple, J. P., Welch, W. R., and Folkman, J. (1991). Tumor angiogenesis and metastasis: correlation in invasive breast carcinoma. N . Engl. J. Med. 324, 1-8. (20) Moscatelli, D. (1987) High and low affinity binding sites for basic fibroblast growth factor on cultured cells; absence of a role for low affinity binding in the stimulation of plasminogen activator production by bovine capillary endothelial cells. J. Cell. Physiol. 131, 123-130. (21) Roghani, M., and Moscatelli, D. (1992) Basic fibroblast growth factor is internalized through both receptor-mediated and heparan sulfate-mediated mechanisms.J . Biol. Chem.267, 22156-22162. (22) Pastan, I., Chaudhary, V., and FitzGerald, D. J. (1992) Recombinant toxins as novel therapeutic agents. Annu. Reu. Biochem. 61, 331-354. (23) Vitetta, E. S., and Thorpe, P. E. (1991) Immunotoxins containing ricin or ita A chain. Sem. Cell. Biol. 2, 47-58. (24) Siegall, C. B., Epstein, S., Speir, E., Hla, T., Forough, R., Maciag, T., FitzGerald, D. J., and Pastan, I. (1991) Cytotoxic
activity of chimeric proteins composed of acidic fibroblast growth factor and Pseudomonas exotoxin on a variety of cell types. FASEB J . 5, 2843-2849. (25) Biro, S., Siegall, C. B., Fu, Y-M., Speir, E., Pastan, I., and Epstein, S. E. (1992) In vitro effects of a recombinant toxin targeted to the fibroblast growthfactor receptor on rat vascular smooth muscle and endothelial cells. Circ. Res. 71, 640-645. (26) Merwin, J. R. , M. J. Lynch, J. A. Madri, I. Pastan, and C. B. Siegall. (1992) Acidic FGF-Pseudomonas exotoxin chimeric protein elicits anti-angiogenic effects on endothelial cells. Cancer Res. 52, 4995-5001. (27) Lappi, D. A. ,D. Martineau, and A. Baird. (1989) Biological and chemicalcharacterization of basic FGF-saporin mitotoxin. Biochem. Biophys. Res. Comm. 160,917-923. (28) Lappi, D., Maher, P. A., Martineau, D., andBaird,A. (1991) The basic fibroblast growth factor-saporin mitotoxin acta through the basic fibroblast growth factor receptor. J. Cell. Phys. 147, 17-26.
Bloconlugate Chem., Vol. 4, No. 0, 1993 489
bFGF-Toxln vs aFGF-Toxin
(29) Beitz, J. G., Davol, P., Clark, J. W., Kato, J., Medina, M.,
Frackelton, A. R., Jr., Lappi, D. A., Baird, A., and Calabresi, P. (1992) Antitumor activity of basic fibroblast growth factorsaporin mitotoxin in vitro and in vivo. Cancer Res. 52, 227230. (30) Beattie, G. M., Lappi, D. A., Baird, A., andHayek, A. (1990)
Selective elimination of fibroblasts from pancreatic islet monolayersby basic fibroblastgrowthfactor-saporinmitotoxin. Diabetes 39, 1002-1005. (31) Casscells, W., Lappi, D. A. Olwin, B. B., Wai, C., Siegman, M., Speir, E. H., Sasse, J., and Baird, A. (1992) Elimination of smooth muscle cells in experimental restenosis: Targeting of fibroblast growth factor receptors. Proc. Natl. Acad. Sci. U.S.A. 89, 7159-7163. (32) Lindner, V., Lappi, D. A., Baird, A., Majack, R. A., and Reidy, M. A. (1991) Role of basic fibroblast growth factor in vascular lesion formation. Circ. Res. 68, 106-113. (33) Pastan, I., and FitzGerald, D. (1991) Recombinant toxins for cancer treatment. Science 254, 1173-1177. (34) FitzGerald,D., and Pastan, I. (1991) Redirecting Pseudomonas exotoxin. Sem. Cell. Biol. 2, 31-37. (35) Chaudhary, V. K., FitzGerald, D. J., Adyha, S., and Pastan, I. (1987) Activity of a recombinant fusion protein between transforming growth factor alpha and Pseudomonas toxin. Proc. Natl. Acad. Sci. U.S.A. 84, 4538-4552. (36) Lorberboum-Galski, H., FitzGerald, D. J. P., Chaudhary, V., Adhya, S., and Pastan, I. (1988) Cytotoxic activity of an interleukin 2-Pseudomonas exotoxin chimeric protein produced in E. coli. Proc. Natl. Acad. Sci. U.S.A.85,1922-1926. (37) Siegall, C. B., Chaudhary, V. K., FitzGerald, D. J., and Pastan, I. (1988) Cytotoxic activity of an interleukin 6Pseudomonas exotoxin fusion protein on myeloma cells. Proc. Natl. Acad Sci. U.S.A. 85, 9738-9742. (38) Ogata, M., Chaudhary, V. K., FitzGerald, D. J. P., and Pastan, I. (1989) Cytotoxic activity of a recombinant fusion protein between interleukin 4 and Pseudomonas exotoxin. Proc. Natl. Acad. Sci. U.S.A. 86, 4215-4219.
(39) Prior, T. I., Hellman, L. J., FitzGerald, D. J., and Pastan, I. (1991) Cytotoxic activity of a recombinant fusion protein
between insulin-like growth factor I and Pseudomonas exotoxin. Cancer Res. 51, 4215-4219. (40) Seetharam, S. Chaudhary, V., FitzGerald, D., and Pastan, I. (1991) Increased cytotoxicactivity of Pseudomonas exotoxin and two chimeric toxins ending in KDEL. J.Biol. Chem. 266, 17376-17381. (41) Studier,F. W.,andMoffat,B.A. (1986) Useofbacteriophage T7 RNA polymerase to direct selective high-levelexpression of cloned genes. J.Mol. Biol. 189, 113-130. (42) Kondo, T., FitzGerald, D., Chaudhary, V. K., Adyha, S., and Pastan, I. (1988) Activity of immunotoxins constructed
with modified Pseudomonas exotoxin A lacking the cell recognition domain. J.Biol. Chem. 263, 9470-9475. (43) Chaudhary, V. K., Jinno, Y., Gallo, M. G., FitzGerald, D., and Pastan, I. (1990) Mutagenesis of Pseudomonas exotoxin in identification of sequences responsible for the animal toxicity. J. Biol. Chem. 265, 16306-16310. (44) Chaudhary, V. K., Jinno, Y., FitzGerald, D. J., and Pastan, I. (1990) Pseudomonas exotoxin contains a specific sequence at the carboxylterminus that is required for cytotoxicity.Proc. Natl. Acad Sci. U.S.A. 87, 308-312. (45) Bradford, M. M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of proteins utilizing the principle of protein-dye binding. Anal. Biochem. 72,248254. (46) Lindner, V., Majack, R. A., and Reidy, M. A. (1990) Basic
FGF stimulates endothelial regrowth and proliferation in denuded arteries. J. Clin. Invest. 85, 2004-2008. (47) Neufeld, G., and Gospodarowicz, D. (1988) Identification of the fibroblast growth factor receptor in human vascular endothelial cells. J. Cell. Phys. 136, 537-542. (48) Friesel, R., Burgess, W. H., Mehlman, T., and Maciag, T. (1986) The characterization of the receptor for endothelial cell growthfactor by covalentligand attachment. J.Biol. Chem. 261, 7581-7584.
