Defining the molecular basis of Arf and Hdm2 interactions.

Understanding the interaction of Arf and Hdm2 has recently become a central issue in cancer biology. In response to hyperproliferative signals, p14(Arf) stabilizes p53 by binding to Hdm2 and inhibits the ubiquitination and subsequent proteosome-dependent degradation of p53. The medical importance of the Arf-Hdm2-p53 regulatory system is highlighted by the finding that either p53 or p14(Arf) are lost or modified in virtually all human cancers. Isolated Arf and Hdm2 domains are dynamically disordered in solution, yet they retain the ability to interact in vitro and in cellular assays. Upon binding, domains of both Arf and Hdm2 undergo a dramatic transition from disordered conformations to extended structures comprised of beta-strands. The presence of domains from both proteins are necessary and sufficient for the formation of the highly stable extended beta structures. We have mapped sites within Arf and Hdm2 that interact at a resolution of five amino acid residues using surface plasmon resonance. Surface plasmon resonance and circular dichroism spectropolarimetry confirm the presence of multiple interaction domains within each protein. Both p14(Arf) (human) and p19(Arf) (mouse) interact with Hdm2 through two short motifs present in their N termini. The Arf interacting region of Hdm2 is also composed of two short sequences located in the central acidic domain, between residues 235-264 and 270-289. The binding-induced structural transition is also induced by short peptides, 15 amino acids in length, that contain the binding motifs. Micro-injection and live cell imaging of proteins tagged with fluorescent labels was used to confirm the in vivo function of the interaction domains. Arf and Hdm2 thus appear to interact through a novel mechanism that exerts control over the cell division cycle. The novel molecular mechanism of interaction and the limited size of the protein domains involved provide opportunities for the development of anticancer therapeutics.

Cloning, overexpression, and purification of aminoglycoside antibiotic nucleotidyltransferase (2'')-Ia: conformational studies with bound substrates.

Aminoglycoside nucleotidyltransferase (2'')-Ia [ANT (2'')-Ia] was cloned from Pseudomonas aeruginosa and purified from overexpressing Escherichia coli BL21(DE3) cells. The first enzyme-bound conformation of an aminoglycoside antibiotic in the active site of an aminoglycoside nucleotidyltransferase was determined using the purified aminoglycoside nucleotidyltransferase (2' ')-Ia. The conformation of the aminoglycoside antibiotic isepamicin, a psuedo-trisaccharide, bound to aminoglycoside nucleotidyltransferase (2' ')-Ia has been determined using NMR spectroscopy. Molecular modeling, employing experimentally determined interproton distances, resulted in two different enzyme-bound conformations (conformer 1 and conformer 2) of isepamicin. Conformer 1 was by far the major conformer defined by the following average glycosidic dihedral angles: PhiBC = -65.26 +/- 1.63 degrees and PsiBC = -54.76 +/- 4.64 degrees. Conformer 1 was further subdivided into one major (conformer 1a) and two minor components (conformers 1b and 1c) based on the comparison of glycosidic dihedral angles PhiAB and PsiAB. The arrangement of substrates in the enzyme.metal-ATP.isepamicin complex was determined on the basis of the measured effect of the paramagnetic substrate analogue Cr(H2O)4ATP on the relaxation rates of substrate protons which were used to determine relative distances of isepamicin protons to the Cr3+. Both conformers of isepamicin yielded arrangements that satisfied the NOE restraints and the observed paramagnetic effects of Cr(H2O)4ATP. It has been suggested that aminoglycosides use both electrostatic interactions and hydrogen bonds in binding to RNA and that the contacts made by the A and B rings to RNA are the most important for binding [Fourmy, D., Recht, M. I., Blanchard, S. C., and Puglisi, J. D. (1996) Science 274, 1367-1371]. Comparisons based on the determined conformations of enzyme-bound aminoglycoside antibiotics also suggested that interactions of rings A and B with enzymes may be the major determinant in aminoglycoside binding to enzymes [Serpersu, E. H., Cox, J. R., DiGiammarino, E. L., Mohler, M. L., Ekman, D. R., Akal-Strader, A., and Owston, M. (2000) Cell Biochem. Biophys. (in press)]. The conformation of isepamicin bound to the aminoglycoside nucleotidyltransferase (2' ')-Ia, determined in this work, lent further support to this theory. Furthermore, comparison of enzyme-bound conformations of isepamicin to the RNA-bound conformation of gentamycin C1a also showed remarkable similarities between the enzyme-bound and RNA-bound aminoglycoside antibiotic conformations. These studies should aid in the design of effective inhibitors possessing a broad range of aminoglycoside-modifying enzymes as targets.

Solution structure of the p53 regulatory domain of the p19Arf tumor suppressor protein.

Arf is a tumor suppressor that regulates p53 function and is a frequent target for loss in human cancers. Through two novel mechanisms, Arf inhibits the oncoprotein Hdm2, a negative regulator of p53. (1) Arf inhibits the E3 ubiquitin ligase activity of Hdm2 that leads to p53 degradation, and (2) Arf sequesters Hdm2 within nucleoli. These activities of Arf promote p53-mediated cell cycle arrest and apoptosis. Fundamental to these processes are interactions between Arf and Hdm2. Here we show that a peptide containing the 37 N-terminal amino acids of mouse Arf (mArfN37) localizes to nucleoli, sequesters Hdm2 within nucleoli, and causes cell cycle arrest. Circular dichroism and NMR spectroscopy show that mArfN37 is largely unstructured under aqueous conditions; however, the peptide adopts two alpha-helices (helix 1, residues 4-14; and helix 2, residues 20-29) in 2,2,2-trifluoroethanol (TFE). Each helix contains an amino acid motif that is repeated twice in mArfN37, once in each helix. The two helices, however, do not interact but are connected by an apparently flexible linker. The repeated motif contains Arg residues spaced by a hydrophobic segment that may be involved in Hdm2 recognition and binding. The RRPR nucleolar localization signal, contained within residues 31-34, appears to be disordered under all conditions. The identification of two Arf structural modules suggests that short peptides containing the repeated motif may function as Arf mimics and may allow the design of small molecule Arf mimics in the future.

Cooperative signals governing ARF-mdm2 interaction and nucleolar localization of the complex.

The ARF tumor suppressor protein stabilizes p53 by antagonizing its negative regulator, Mdm2 (Hdm2 in humans). Both mouse p19(ARF) and human p14(ARF) bind to the central region of Mdm2 (residues 210 to 304), a segment that does not overlap with its N-terminal p53-binding domain, nuclear import or export signals, or C-terminal RING domain required for Mdm2 E3 ubiquitin ligase activity. The N-terminal 37 amino acids of mouse p19(ARF) are necessary and sufficient for binding to Mdm2, localization of Mdm2 to nucleoli, and p53-dependent cell cycle arrest. Although a nucleolar localization signal (NrLS) maps within a different segment (residues 82 to 101) of the human p14(ARF) protein, binding to Mdm2 and nucleolar import of ARF-Mdm2 complexes are both required for cell cycle arrest induced by either the mouse or human ARF proteins. Because many codons of mouse ARF mRNA are not recognized by the most abundant bacterial tRNAs, we synthesized ARF minigenes containing preferred bacterial codons. Using bacterially produced ARF polypeptides and chemically synthesized peptides conjugated to Sepharose, residues 1 to 14 and 26 to 37 of mouse p19(ARF) were found to interact independently and cooperatively with Mdm2, while residues 15 to 25 were dispensable for binding. Paradoxically, residues 26 to 37 of mouse p19(ARF) are also essential for ARF nucleolar localization in the absence of Mdm2. However, the mobilization of the p19(ARF)-Mdm2 complex into nucleoli also requires a cryptic NrLS within the Mdm2 C-terminal RING domain. The Mdm2 NrLS is unmasked upon ARF binding, and its deletion prevents import of the ARF-Mdm2 complex into nucleoli. Collectively, the results suggest that ARF binding to Mdm2 induces a conformational change that facilitates nucleolar import of the ARF-Mdm2 complex and p53-dependent cell cycle arrest. Hence, the ARF-Mdm2 interaction can be viewed as bidirectional, with each protein being capable of regulating the subnuclear localization of the other.

Aminoglycoside antibiotics bound to aminoglycoside-detoxifying enzymes and RNA adopt similar conformations.

Conformations of ribostamycin and isepamicin, aminoglycoside antibiotics, bound to an aminoglycoside antibiotic, 3'-phosphotransferase, were determined by transferred nuclear Overhauser effect spectroscopy and molecular modeling. Two major conformers of enzyme-bound ribostamycin, a neomycin-group aminoglyeoside were observed. The 3'- and 5"-OH groups (reactive hydroxyl groups) in the conformers are placed in approximate locations. One of the conformers is similar to the structure of paromomycin bound to a 27-nucleotide piece of ribosomal RNA that represents the A-site of the small ribosomal subunit, where rings A and C are in an orthogonal arrangement. Isepamicin, a kanamycin-group aminoglycoside antibiotic, also showed two major enzyme-bound conformations. Both conformations were similar to those observed for bound isepamicin in the active site of an aminoglycoside(6')-acetyl transferase-Ii. Conformations of other RNA-bound kanamycin-group aminoglycosides were also similar to the enzyme-bound conformations of isepamicin. These observations suggest that aminoglycosides may adopt similar conformations when bound to RNA and protein targets. This may have significant implications in the design of enzyme inhibitors and/or antibiotics.

Solution studies of isepamicin and conformational comparisons between isepamicin and butirosin A when bound to an aminoglycoside 6'-N-acetyltransferase determined by NMR spectroscopy.

NMR spectroscopy, combined with molecular modeling, was used to determine the conformations of isepamicin and butirosin A in the active site of aminoglycoside 6'-N-acetyltransferase-Ii [AAC-(6')-Ii]. The results suggest two enzyme-bound conformers for isepamicin and one for butirosin A. The dihedral angles that describe the glycosidic linkage between the A and B rings for the two conformers of AAC(6')-Ii-bound isepamicin were phi AB = -7.9 +/- 2.0 degrees and psi AB = -46.2 +/- 0.6 degrees for conformer 1 and phi AB = -69.4 +/- 2.0 degrees and psi AB = -57.7 +/- 0.5 degrees for conformer 2. Unrestrained molecular dynamics calculations showed that these distinct conformers are capable of interconversion at 300 K. When superimposed at the 2-deoxystreptamine ring, one enzyme-bound conformer of isepamicin (conformer 1) places the reactive 6' nitrogen in a similar position as that of butirosin A. Conformer 2 of AAC(6')-Ii-bound isepamicin may represent an unproductive binding mode. Unproductive binding modes (to aminoglycoside modifying enzymes) could provide one reason isepamicin remains one of the more effective aminoglycoside antibiotics. The enzyme-bound conformation of butirosin A yielded an orthogonal arrangement of the 2,6-diamino-2,6-dideoxy-D-glucose and D-xylose rings, as opposed to the parallel arrangement which was observed for this aminoglycoside in the active site of an aminoglycoside 3'-O-phosphotransferase [Cox, J. R., and Serpersu, E. H. (1997) Biochemistry 36, 2353-2359]. The complete proton and carbon NMR assignments of the aminoglycoside antibiotic isepamicin at pH 6.8 as well as the pKa values for it's amino groups are also reported.