Retina- and eye-derived endothelial cell growth factors: partial molecular characterization and identity with acidic and basic fibroblast growth factors

Two retina-derived growth factors have been isolated on the basis of their ability to stimulate the proliferation of capillary endothelial cells in vitro. Gas-phase sequence analysis identified the amino-terminal sequence of the major form of the mitogen as being identical with residues 1-35 of bovine basic fibroblast growth factor (FGF). Amino-terminal sequence analysis of the second form identified 28 residues that are indistinguishable from those of brain acidic FGF (residues 1-28). The possibility that these retina-derived endothelial cell growth factors are related to, if not identical with, basic and acidic FGF is supported by observations that they have similar molecular weights (15000-16000), similar retention behavior on all steps of chromatography (ion-exchange, heparin-Sepharose), and similar amino acid compositions and that they cross-react with antibodies to basic and acidic FGF. The eye-derived growth factors, like FGF, are potent stimulators of capillary endothelial cell growth in vitro. The results identify the major retina-derived endothelial cell growth factor as indistinguishable from basic FGF and demonstrate the presence of an acidic FGF in the eye. They suggest that at least some of the mitogenic, angiogenic, and neovascularizing activities described as being present in the retina are due to the existence of FGF in this tissue. The implications of this finding on the etiology and pathophysiology of vasoproliferative diseases of the eye are discussed.     

Pituitary fibroblast growth factors: immunocharacterization of an acid component and N-terminal sequence analysis of a truncated basic component

Extraction of bovine pituitaries at pH 7.0, in the presence or absence of protease inhibitors (PMSF, leupeptin, pepstatin A and EDTA) yielded both basic and acidic FGF components that were characterized by Western blotting and sequence analysis. Basic FGF comprised several components: an 18 KDa form that is similar, if not identical, to the basic FGF (1-146) already described; a 17 KDa form that is likely to be a new truncated molecular species (11-146) and a group of immunoreactive components of about 29 KDa. Acidic FGF showed several active components of pI 4.5-6.5. The most active component has a pI of approximately 5.0; molecular weight of 17 KDa and is shown, by Western blotting, to be similar to a truncated form of bovine brain acidic FGF. The biological activity of the latter component is shown to be neutralized by anti-brain acidic FGF antiserum.     

Localization of basic fibroblast growth factor to the developing capillaries of the bovine retina

The basic fibroblast growth factor (FGF) is a potent mitogen that has vascular endothelium as one of its principle target cells. Recent work has provided both the complete amino acid sequence of basic FGF and the nucleotide sequence of the genes for both human and bovine basic FGF. Although capillary endothelial cells have been shown to produce basic FGF in vitro and to deposit basic FGF in their extracellular matrix in vitro as well, no direct evidence yet exists for the distribution of basic FGF in vivo. Antipeptide antibodies were prepared against a 15-amino-acid sequence from the amino terminus of basic FGF in order to avoid cross-reactivity with acidic FGF, a protein with 55% overall homology to basic FGF. After affinity purification, these antisera were used to localize the basic fibroblast growth factor in the fetal and adult bovine retina. Immunoreactive material was found in capillaries of the inner nuclear layer, a capillary network undergoing development during the third trimester in the fetal bovine eye. Although the resolution of the technique does not permit a unique assignment of cellular localization, the presence of stain immediately adjacent to the lumen of capillaries suggests that capillary endothelial cells may produce the basic fibroblast growth factor in vivo during vascular development.     

Structural studies on rat prostatic binding protein. The primary structure of component C2 from subunit S

The amino acid sequence of component C2, the polypeptide specific for subunit S of prostatic binding protein, the major secretory glycoprotein of the rat ventral prostate, has been determined. Its structure was established using the manual Edman degradation on the most relevant fragments obtained by enzymatic digestion of the S-carboxamidomethylated component C2 and the native subunit S and by chemical cleavage of the remaining undigestible 'cores' with cyanogen bromide. Component C2 contains 92 amino acids corresponding to a molecular weight of 10619. It is a slightly acidic polypeptide in which the acidic and basic residues are unevenly distributed. The N terminus is blocked and three cysteine residues are almost evenly distributed over the peptide chain. A highly polar region is found in position 23-34 and two hydrophobic segments are located in the C-terminal part of the molecule. Component C2 is compared with component C1 of subunit F and their high sequence homology reveals an evolutionary relationship.     

Evidence for a polypeptide segment at the carboxyl terminus of recombinant human gamma interferon involved in expression of biological activity

A panel of 18 murine monoclonal antibodies was raised in BALB/c mice to the full-length, 146 amino acid residue recombinant human gamma interferon (rHuIFN gamma-A). Two monoclonal antibodies, designated 47N3-6 and 30N47-1, were purified from ascites tumors and further characterized. Antibody 47N3-6 neutralized both the antiviral and antiproliferative activities of rHuIFN gamma-A. Both Western blotting and enzyme-linked immunosorbent assays indicated that antibody 47N3-6 could bind to rHuIFN gamma-A as well as to a genetically engineered truncated form lacking the first three amino-terminal residues (rHuIFN gamma-D) but did not recognize a genetically engineered variant terminating at residue 131 (rHuIFN gamma-B). This antibody also demonstrated binding to a 15 amino acid residue oligopeptide, designated F-1, corresponding to residues 132-146 at the carboxyl terminus of rHuIFN gamma-A. Chemical cleavage of peptide F-1 with cyanogen bromide produced two fragments that were separated by reversed-phase high-pressure liquid chromatography. Dot-blot analysis indicated that antibody 47N3-6 could bind to a fragment, KRKRSQHse, derived from residues 132-137 of rHuIFN gamma-A, but could bind only weakly to the cyanogen bromide fragment corresponding to residues 138-146. It was consistent with these results that antibody 47N3-6 demonstrated binding to a form lacking the five carboxyl-terminal amino acids (rHuIFN gamma-D') but did not bind to a synthetic polypeptide corresponding to residues 138-146. Peptide F-1 exhibited neither antiviral nor antiproliferative activity, and it did not antagonize the antiviral activity of rHuIFN gamma-A.(ABSTRACT TRUNCATED AT 250 WORDS)     

Amino-terminal sequence of a large form of basic fibroblast growth factor isolated from human benign prostatic hyperplastic tissue

Homogenization of human benign prostatic hyperplastic tissue in high ionic strength alkaline buffer containing protease inhibitors resulted in the isolation of a 17,400 molecular weight growth factor. When tissue was homogenized in ammonium sulfate at pH 4.5 without protease inhibitors a smaller, 16,600 dalton, growth factor was isolated. Both growth factors reacted with antisera against synthetic peptides whose sequences corresponded to the amino-terminal (1-12), Internal (33-43) and carboxyl-terminal (135-145) portions of basic fibroblast growth factor (bFGF). This suggested that the smaller growth factor was not a truncated form of (1-146) bFGF and that the larger growth factor may contain additional sequences. Amino-terminal sequencing showed the larger growth factor to have the sequence: Ala-Ala-Gly-Ser-Ile-Thr-Thr-Leu-Pro-Ala-Leu-Pro-Glu-Asp-Gly-Gly-Ser-Gly- Ala-Phe-Pro-. These results show that the larger growth factor is an 8 amino acid extended from of (1-146) bFGF and it is likely that the smaller growth factor is a proteolytic cleavage product of the larger growth factor produced during the extraction procedure.     

Purification and partial characterization of a mitogenic factor from bovine liver: structural homology with basic fibroblast growth factor

Two mitogenic peptides in bovine liver extract were purified to apparent homogeneity by monitoring the purification steps with two in vitro bioassays; one based on stimulation of adult bovine aortic arch endothelial cell proliferation and the other incorporation of [3H]thymidine to mouse fibroblast 3T3 cells. The purification procedure involved cation-exchange chromatography followed by affinity chromatography on heparin-Sepharose and two steps of reversed-phase HPLC. The purified material showed the same biological activity as pituitary basic fibroblast growth factor (FGF). Amino acid analyses of the purified mitogen yielded a similar, but not identical composition to that of bovine pituitary basic FGF(1-146) reported previously. Gas-phase microsequencing identified two sequences in equal amounts in the purified preparation. Furthermore, the sequencing results are in accord with the theoretical data obtained when two truncated forms of basic FGF, corresponding to FGF(12-146) and (16-146), are being sequenced simultaneously. Basic FGF(12-146) is a novel truncated form of basic FGF which has not been isolated before although the (16-146) fragment has been found previously in kidney, corpus luteum, and adrenal. SDS-PAGE analysis could not separate the two forms and showed that both migrated as a protein of about 15,100 daltons, which is slightly smaller than intact basic FGF(1-146) (16,200 daltons). These results, taken together, indicate that at least some of the mitogenic activity in liver may be derived from basic FGF-related polypeptides.     

Direct evidence for methylation of arginine residues in high molecular weight forms of basic fibroblast growth factor.

Basic fibroblast growth factor (bFGF) is a heparin-binding angiogenic polypeptide mitogen. Protein sequence analysis of bFGF isolated from tissue sources initially established that it is composed of 146 amino acids (apparent Mr 18,000). More recently larger apparent molecular weight forms have been identified and partially characterized. In addition, these high molecular weight forms (apparent Mr 22,000 and 25,000) have been shown to localize preferentially to nuclear fractions of transfected cells. In this report we demonstrate that the higher molecular weight, amino terminally extended forms of bFGF contain methylated arginine residues. The demonstration is based on 1) amino acid sequence analysis of a protein known to contain methylated arginine (myelin basic protein) and a comparison with amino acid sequence analysis of trypsin-derived fragments of the high molecular weight bFGF purified from guinea pig brain and 2) the ability to label in vivo the high molecular weight forms of bFGF with S-adenosyl-L-(methyl-3H)-methionine, the substrate of arginine-protein methylase I. These results are suggestive of a role of arginine methylation in directing nuclear localization of certain forms of bFGF.     

Both the sequence and length of the C terminus of PEN-2 are critical for intermolecular interactions and function of presenilin complexes

Presenilin 1 or presenilin 2, nicastrin, APH-1, and PEN-2 form high molecular weight complexes that play a pivotal role in the cleavage of various Type I transmembrane proteins, including the beta-amyloid precursor protein. The specific function of PEN-2 is unclear. To explore its function and intermolecular interactions, we conducted deletion and mutagenesis studies on a series of conserved residues at the C terminus of PEN-2. These studies suggest that: 1) both the presence and amino acid sequence of the conserved DYLSF domain at the C terminus of PEN-2 (residues 90-94) is critical for binding PEN-2 to other components in the presenilin complex and 2) the overall length of the exposed C terminus is critical for functional gamma-secretase activity.     

Structure-function analysis of human interleukin-2. Identification of amino acid residues required for biological activity

To locate functional domains of the interleukin-2 (IL-2) protein, a cDNA clone encoding biologically active human IL-2 was mutagenized using synthetic oligonucleotides to incorporate defined amino acid substitutions and deletions in the mature protein. The IL-2 analogs were then produced in Escherichia coli and assayed for the ability to induce proliferation of IL-2-dependent cells and the ability to compete for binding to the IL-2 receptor. Our analysis of over 50 different mutations demonstrated that the integrity of at least three regions of the IL-2 molecule is required for full biological activity: the NH2 terminus (residues 1-20), the COOH terminus (residues 121-133), and 2 of the 3 cysteine residues (58 and 105). Deletion of the NH2-terminal 20 amino acids or the COOH-terminal 10 amino acids resulted in the loss of greater than 99% of bioactivity and binding. Amino acid substitutions at specific positions in these regions also resulted in proteins which retained less than 1% activity. The NH2 terminus and an adjacent internal region were recognized by neutralizing anti-IL-2 antibodies. In combination with the results from epitope competition analysis with neutralizing antibodies, these data are consistent with the IL-2 protein being folded such that the NH2 terminus, the COOH terminus, and the internal 30- to 60-region are juxtaposed to form the binding site recognized by the IL-2 receptor.     

Phosphorylation-dependent conformational transition of the cardiac specific N-extension of troponin I in cardiac troponin

We present here the solution structure for the bisphosphorylated form of the cardiac N-extension of troponin I (cTnI(1-32)), a region for which there are no previous high-resolution data. Using this structure, the X-ray crystal structure of the cardiac troponin core, and uniform density models of the troponin components derived from neutron contrast variation data, we built atomic models for troponin that show the conformational transition in cardiac troponin induced by bisphosphorylation. In the absence of phosphorylation, our NMR data and sequence analyses indicate a less structured cardiac N-extension with a propensity for a helical region surrounding the phosphorylation motif, followed by a helical C-terminal region (residues 25-30). In this conformation, TnI(1-32) interacts with the N-lobe of cardiac troponin C (cTnC) and thus is positioned to modulate myofilament Ca2+-sensitivity. Bisphosphorylation at Ser23/24 extends the C-terminal helix (residues 21-30) which results in weakening interactions with the N-lobe of cTnC and a re-positioning of the acidic amino terminus of cTnI(1-32) for favorable interactions with basic regions, likely the inhibitory region of cTnI. An extended poly(L-proline)II helix between residues 11 and 19 serves as the rigid linker that aids in re-positioning the amino terminus of cTnI(1-32) upon bisphosphorylation at Ser23/24. We propose that it is these electrostatic interactions between the acidic amino terminus of cTnI(1-32) and the basic inhibitory region of troponin I that induces a bending of cTnI at the end that interacts with cTnC. This model provides a molecular mechanism for the observed changes in cross-bridge kinetics upon TnI phosphorylation.     

Dual activities of 22-24 kDA basic fibroblast growth factor: inhibition of migration and stimulation of proliferation

Basic fibroblast growth factor (FGF2) is synthesized as four isoforms with molecular weights of 24, 22.5, 22, and 18 kDa, with each of the three higher molecular weight forms (hmwFGF2) produced by the initiation of translation at one of three upstream CUG codons. We have shown that bovine arterial endothelial cells export the high molecular weight forms of FGF2 (hmwFGF2) in a 17beta-estradiol-dependent manner (Piotrowicz et al., 1997, J Biol Chem 272:7042-7047). To determine whether the hmwFGF2 forms affected cell behavior after release, we evaluated the effect of recombinant hmwFGF2 on the growth and migration of endothelial cells and mammary carcinoma cells (MCF-7). Treatment with the recombinant protein resulted in the inhibition of endothelial cell migration by 45% and MCF-7 cell migration by 70%. HmwFGF2-dependent inhibition was observed when endothelial cell migration was stimulated by 18 kDa FGF2 or vascular endothelial growth, and MCF cell migration was stimulated with insulin-like growth factor. In each case, inclusion of an antibody against the 55 amino acid amino terminal end of 24 kDa FGF2 abrogated the inhibition of migration, while antibodies to the 18 kDa FGF2 domain had no effect. When endothelial cells were cultured under conditions which promoted export of hmwFGF2, a 40% decrease in motility was observed which was reversed by the antibodies to the 24 kDa FGF2. Thus, both recombinant and endogenously produced hmw-FGF2 are capable of inhibiting migration. In contrast to the ubiquitous effect on migration, hmwFGF2 had no effect on endothelial cell growth but stimulated MCF-7 growth equally as well as the 18 kDa FGF2 (threefold). Antibodies to the 18 kDa domain of 24 kDa FGF2 blocked the growth-promoting activity of hmwFGF2, but those to the amino terminal end were ineffective. These data suggest that hmwFGF2 has dual activities, an inhibitory effect on cell migration and a growth-stimulating effect. The two activities can be localized to different parts of hmwFGF2: inhibitory activity to the amino terminal 55 amino acids (which are absent from the 18 kDa FGF2) and growth-promoting activity to the 18 kDa domain. Therefore, the ratio of hmwFGF2 and 18 kDa FGF2 in the extracellular space may provide a mechanism of control for angiogenesis and mammary tumor development.     

Identification of desmoglein, a constitutive desmosomal glycoprotein, as a member of the cadherin family of cell adhesion molecules.

Monoclonal antibodies to the constitutive desmosomal glycoprotein desmoglein were characterized whose epitopes are located intracellularly, i.e., in the cytoplasmic portion of this molecule, and contribute to the structure of the desmosomal plaque. Using one of these antibodies (DG3.10), a peptide was isolated from a proteolytic digest of desmoglein purified from isolated bovine muzzle demosomes, and its amino acid sequence was determined. In comparisons of this sequence with the amino acid sequence of desmoglein as deduced from the sequence of cDNA clones from the same tissue, encompassing most of approximately 7.6 kb mRNA and the complete coding region of 959 residues (calculated molecular weight approximately 102,400), the DG3.10 epitope was identified in a region starting 163 amino acids before the carboxy terminus in the first of four consecutive repeats of a homologous element of 29 +/- 1 amino acids. This topological information, together with the identification of a single hydrophobic region of sufficient length to provide a transmembrane segment and of several extended regions showing high sequence homology to various cadherins, has allowed the construction of a model of the molecular organization of desmoglein. We conclude that desmoglein is a member of the cadherin family of cell adhesion glycoproteins which is characterized by an unusually long cytoplasmic domain which exceeds those of the cadherins by more than 275 amino acids, contains special repetitive elements and spans the desmosomal plaque at least once.     

Isolation and partial characterization of an endothelial cell growth factor from the bovine kidney: homology with basic fibroblast growth factor

Two kidney-derived mitogens have been isolated by ion exchange, heparin-Sepharose and reverse-phase high-performance liquid chromatography on the basis of their capacity to stimulate the proliferation of bovine vascular endothelial cells in vitro. Gas phase sequence analysis identified the amino terminal sequences His-Phe-Lys-Asp-Pro-Lys-Arg-Leu-Tyr-X-Lys-Asn-Gly-Gly-Phe-Phe-Leu and His-Phe-Lys-Asp-Pro-Lys-Arg-Leu, respectively. The sequences are identical to residues 16-32 and 16-23 of bovine basic pituitary Fibroblast Growth Factor (FGF). The possibility that these kidney-derived mitogens are related, if not identical, to pituitary basic FGF is supported by the observations that they have similar molecular weights (15-16 kDa), similar retention behavior on all steps of chromatography and similar amino acid compositions, and they share at least some structural homology. Moreover, the kidney-derived growth factors, like basic FGF, are potent stimulators of capillary endothelial cells, granulosa cells, adrenocortical cells and vascular smooth-muscle cells (ED50 = 50 pg). The results demonstrate the existence of a kidney-derived FGF and suggest that at least some of the mitogenic, angiogenic and neovascularising activities described to be present in the kidney are due to the presence of an FGF-like molecule in this tissue.     

Characterization of a high-molecular-weight form of epidermal growth factor in an extract of human urine

We purified from a side fraction of the commercial preparation of urokinase from large volumes of human urine a high-molecular-weight (HMW) form of human epidermal growth factor (hEGF). Sequence analysis of the amino terminus of the intact molecule and of two tryptic fragments and carboxypeptidase Y analysis revealed the molecule to correspond to residues 828-1023 of the hEGF precursor predicted by the nucleotide sequence of human renal hEGF mRNA, with hEGF forming its carboxyl terminus. HMW hEGF bound poorly to concanavalin A-agarose, quite avidly to wheat germ lectin-agarose, and completely to phenyl boronate-agarose, suggesting that it was O-glycosylated. Sephacryl S-200 chromatography of freshly-voided urine revealed mostly hEGF, with smaller amounts of a much higher molecular weight hEGF, but little material that was the size of the HMW hEGF we characterized. The large fragment we characterized presumably is cleaved from the larger form by enzyme(s) present in urine during the collection, storage, and processing of urine. We have confirmed that hEGF is synthesized as a large precursor molecule, as predicted by the nucleotide sequence of hEGF mRNA.     

Characterization of murine monoclonal antibodies directed against basic fibroblast growth factor

Monoclonal antibodies (McAbs) were developed that identify the complete (1-146 aa) and the NH2-terminal truncated (des 1-15) form of bovine basic fibroblast growth factor (bFGF). Four McAbs, designated McAbs 6, 8, 38, and 42, bind the complete form of bFGF found in bovine pituitary, brain, and adrenal gland. One of these McAbs, McAbs 42, also binds to the des 1-15 aa form of bFGF found in bovine adrenal gland, kidney, and corpus luteum. None of the McAbs binds bovine-brain-derived acidic FGF (aFGF). McAbs 6, 8, and 38 recognized the same epitope located within the first ten residues of the NH2-terminal of complete bFGF. McAb 42 recognizes a "core" epitope found on both the complete and des 1-15 aa bFGFs. The McAbs are murine IgGs with affinity constants of 10(7)-10(8) liter/M for bovine-pituitary-derived bFGF. McAbs 8 and 42 have been used in a two-site ELISA to detect the complete form of bFGF. The ELISA is sensitive to 38.5 fmole/well of bFGF and is not affected by the presence of calf serum or bovine-brain-derived aFGF. These McAbs should be useful in distinguishing the native and des 1-15 aa forms of bFGF from each other, and from aFGF and other growth factors.     

Amino- and carboxy-terminal deletion mutants of Gs alpha are localized to the particulate fraction of transfected COS cells

To elucidate the structural basis for membrane attachment of the alpha subunit of the stimulatory G protein (Gs alpha), mutant Gs alpha cDNAs with deletions of amino acid residues in the amino and/or carboxy termini were transiently expressed in COS-7 cells. The particulate and soluble fractions prepared from these cells were analyzed by immunoblot using peptide specific antibodies to monitor distribution of the expressed proteins. Transfection of mutant forms of Gs alpha with either 26 amino terminal residues deleted (delta 3-28) or with 59 amino terminal residues deleted (delta 1-59) resulted in immunoreactive proteins which localized primarily to the particulate fraction. Similarly, mutants with 10 (delta 385-394), 32 (delta 353-384), or 42 (delta 353-394) amino acid residues deleted from the carboxy terminus also localized to the particulate fraction, as did a mutant form of Gs alpha lacking amino acid residues at both the amino and carboxy termini (delta 3-28)/(delta 353-384). Mutant and wild type forms of Gs alpha demonstrated a similar degree of tightness in their binding to membranes as demonstrated by treatment with 2.5 M NaCl or 6 M urea, but some mutant forms were relatively resistant compared with wild type Gs alpha to solubilization by 15 mM NaOH or 1% sodium cholate. We conclude that: (a) deletion of significant portions of the amino and/or carboxyl terminus of Gs alpha is still compatible with protein expression; (b) deletion of these regions is insufficient to cause cytosolic localization of the expressed protein. The basis of Gs alpha membrane targeting remains to be elucidated.     

Amino acid sequence of spinach ferredoxin:NADP+ oxidoreductase

The amino acid sequence of spinach ferredoxin: NADP+ oxidoreductase was determined by using overlapping sets of peptides derived by cleavage at arginyl or methionyl residues. The protein from different preparations varied in its length at the amino terminus. In the longest form the amino terminus is blocked with a pyroglutamyl residue, as determined by NMR. A single disulfide bond was placed between cysteine residues 132 and 137. The 314-residue sequence corresponds to a molecular weight of 35 317. The carboxyl-terminal half of the sequence has been fit to the electron density map of the NADP binding domain, revealing that this portion of the chain forms a typical nucleotide binding fold.     

A combined extended and helical backbone for Boc-(Ala-Leu-Ac7c-)2-OMe.

The structure of the peptide Boc-Ala-Leu-Ac7c-Ala-Leu-Ac7c-OMe (Ac7c,1-aminocycloheptane-1-carboxylic acid) is described in crystals. The presence of two Ac7c residues was expected to stabilize a 3(10)-helical fold. Contrary to expectation the structural analysis revealed an unfolded amino terminus, with Ala(1) adopting an extended beta-conformation (Phi=-93 degrees, psi=112 degrees). Residues 2-5 form a 3(10)-helix, stabilized by three successive intramolecular hydrogen bonds. Notably, two NH groups Ala(1) and Ac7c(3) do not form any hydrogen bonds in the crystal. Peptide assembly appears to be dominated by packing of the cycloheptane rings that stack against one another within the molecule and also throughout the crystal in columns.     

Identification of arfophilin, a target protein for GTP-bound class II ADP-ribosylation factors

Yeast two-hybrid screening of a human kidney cDNA library using the GTP-bound form of a class II ADP-ribosylation factor (ARF5) identified a novel ARF5-binding protein with a calculated molecular mass of 82.4 kDa, which was named arfophilin. Northern hybridization analysis showed high level arfophilin mRNA expression in human heart and skeletal muscle. Arfophilin bound only to the active, GTP-bound form of ARF5 and did not bind to GTP-ARF3, which is a class I ARF. The N terminus of ARF5 (1-17 amino acids) was essential for binding to arfophilin. The GTP-bound form of ARF5 with amino acid residues in the N terminus mutated to those in ARF4 (another class II ARF) also bound to arfophilin, suggesting it is a target protein for GTP-bound forms of class II ARFs. The binding site for ARF on arfophilin was localized to the C terminus (residues 612-756), which contains putative coiled-coil structures. Recombinant arfophilin overexpressed in CHO-K1 cells was localized in the cytosol and translocated to a membrane fraction in association with GTP-bound ARF5. ARF5 containing the N terminus of ARF3 did not promote translocation indicating that class II ARFs are specific carriers for arfophilin.