Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Point mutation wikipedia , lookup
Gel electrophoresis of nucleic acids wikipedia , lookup
Expanded genetic code wikipedia , lookup
Therapeutic gene modulation wikipedia , lookup
Deoxyribozyme wikipedia , lookup
Artificial gene synthesis wikipedia , lookup
DNA nanotechnology wikipedia , lookup
Nucleic acid double helix wikipedia , lookup
Volume 13 Number 7 1985 Nucleic Acids Research A comparison of the structure of echinomycin and triostin A complexed to a DNA fragment Giovanni Ughetto*, Andrew H.-J.Wang*§, Gary J.Quigley*, Gijs A.van der Marel"1", Jacques H.van Boom4" and Alexander Rich*§ *Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139, USA, and + Gorlaeus Laboratories, Leiden State University, Leiden, The Netherlands Received 2 January 1985; Accepted 19 February 1985 ABSTRACT Two members of the quinoxaline antibiotic family, echinomycin and triostin A, form crystals complexed to a DNA fragment with the sequence a(CpGpTpApCpG). The crystal structure of both complexes was solved by X-ray diffraction to near-atomic resolution. The two structures are similar to each other with differences in some details due to the shorter cross bridge of echinomycin. Both molecules act as bis intercalators surrounding the d(CpG) sequence at either end of the double helix. Alanine forms sequence—specific hydrogen bonds to guanines in the minor groove. The two central AT base pairs are held together by Uoogsteen base pairing with adenine in the svn conformation in both complexes. An octahedrally hydrated magnesium ion is found in the crystal lattice that plays an important role in organizing the lattice as well as stabilizing the complex by hydrogen bonding both to base pairs of UNA and to the quinoxaline ring nitrogen atoms in the major groove side of the UNA double helix. A functional description of the various amino acids in quinoxaline antibiotics is given, together with possible modifications that might affect biological activity. INTKODUCTION An important class of antibiotics are those that bind to UNA and thereby modify its biological activities. Streptomvces contain eight The quinoxaline antibiotics amino acids in a cyclic depsipeptide quinoxaline rings attached to them [1, 2 ] , These antibiotics derived from and have two are very active against gram positive bacteria and also have cytotoxic effects on tumor cells. Two members of this each by other molecules class are echinomycin and triostin A, which differ the nature of the sulfur-containing as shown in Figure 1. They act by binding replication as well as transcription [3, 4 ] . action. to with DNA is fundamental the deoxyhexanucleoside in from these to DNA and inhibit DNA [5]. to an understanding Previously, we have described bridge Currently, echinomycin used in clinical trials in human cancer chemotherapy interaction cross is being The nature of their of their mode of the structure of triostin A complexed pentaphosphate d(CpGpTpApCpG) [6], We have now solved the structure of echinomycin bound to the same oligonucleotide and here we compare the two molecular structures. © IRL Press Limited, Oxford, England. Both molecules appear to act as bis 2305 Nucleic Acids Research intercalators surrounding intercalated inside the sequence the double helical over the terminal CG base pair. d(CpG) tragment helix. In addition, central AX base pairs in echinomycin. interactions the unusual together Although results in as Hoogsteen because ot ot the similar shorter cross the many two base pairs [7] the two structures are to the ring stacks ot the DNA rearrangement modifying a major component van bridge der Waals of the stabilization between the The UNA fragment is considerably unwound and there are several conformational modifications to quinoxaline other ring and the minor groove in several details due This that provide drugs and UNA. is an base pairs. to each other, they differ found there that are now held instead of Watson-Crick one the This binding is sequence-specific hydrogen bonding between the peptide backbone double with while interaction with associated with alterations of base the antibiotic. These are also stacking due found to differ somewhat in the two structures due to the altered geometry ot the depsipeptide ring. Although differences these both activities [8] . in two quinoxaline their binding antibiotics constant to UNA are similar, and in there their are biological Thus, a detailed comparison ot the two types of complexes may be useful as it may be related to these differences. MATERIALS ANU METHODS The triostin A was a gift from Ur. T. Yoshida The echinomycin Cambridge. The earlier 19]. technique hexanucleoside described pentaphosphate before [6]. A 2 mM UNA hexamer, 20 mM sodium triostin A or echinomycin, and was as described unit cell lattices with dimensions triostin A asymmetric were & the synthesized using the vapor phase diffusion typical crystallization cacodylate space = 30.71, F222. 62.36, c For unit in both The crystals contained one echinomycin, = 61.70 the dimensions were a = 31.35, b = 62.38, and antibiotic molecule. (MPU). The crystals of both complexes group b - mixture (pH 7 ) , 10 mM MgCl 2 . 1.5 51o 2-methyl-2,4-pentanediol mixture was then equilibrated against 35% MPU. had orthorhombic (Shionogi, Osaka, Japan). from Ur. M. J. Waring, University of The complexes were crystallized as contained mM (Ciba-Geigy) was a gift 8, while the tor c = 61.26 ft. The strand of UNA as well The diffraction patterns produced by the two as one complexes were very similar to each other. Three-dimensional diffraction data diffractometer in the omega scan mode. 10°C and ad). 2,015 collected on a Nicolet P3 reflections were measured with an intensity greater than 1.0 The triostin A data collected in the same manner but at -16°C had 2,995 observed reflections. 2306 were The echinomycin data were collected at The structure of the triostin A complex was solved as Nucleic Acids Research NM,Vol iN-MeCy I-Ala Ou. Triostin A Future 1^: The structure of triostin A and echinomycin axe indicated. Triostin A has a OisulfiQe linkage between the cystine residues, whereas a thioacetal linkage is found in echinomycin. described earlier 5-bromocytosine by the method ot derivatives [6] . patterns, trial coordinates multiple Due to isomorphous the replacement similarity tor the echinomycin of the using diffraction structure were obtained from the triostin A structure and then they were refined at 1.8 8 resolution using the constrained were Konnert-Henarickson revealed in the asymmetric refinement unit of the the 137 found in the triostin A crystal. residual factor was 19.9% compared to 18.6% for tor the triostin A [10]. solvent molecules echinomycin crystal compared to After many cycles of refinement, the echinomycin at 88 1.67 % complex at 1.8 resolution. In 8 resolution the refinment procedure, a number of different constraints were imposed on the molecule and the results were studied. In all of placed on base pairing geometries. distance constraints structure revealed and placed echinomycin structure. the peptide Accordingly, and the similar A similar hydrogen weak constraints were the drug and the nucleic acid. bonding interactions between the That antibiotic type of refinement was carried out for the This yielded two hydrogen bonding interactions between nucleic constraints hydrogen bonds were peptides. between three hydrogen the nucleic acid. the refinements, In the triostin A structure there were no imposed bonding acid involving but the to be close interactions one of Ml of them was alamne implausibly and N3 of to 2.8 8 and this resulted between Both crystals were nearly 50% solvent the nucleic acid and there were large short. guanine in and three the solvent channels found weaving along the £ axis. The coordinates of both complexes are being published elsewhere. 2307 Nucleic Acids Research Figure 2: Van Her Waals diagrams showing the interaction of triostin A and echinomycin with the DNA duplex. The antibiotics are shaded with dark stippling. At the top of the diagram the quinoxaline rings can be seen protruding into the major groove of the helix. In the lower part of the diagram the cyclic peptide occupies virtually all of the minor groove. Carbon atoms are indicated by concentric circles, phosphorous atoms by concentric circles with cross bars, nitrogen atoms by stippled circles. Hydrogen atoms are not shown. RESULTS ANLI DISCUSSION The structure of the echinomycin complex to the DNA oligomer is broadly similar to that already reported for triostin A (Figure 2 ) . like triostin A, binds to the minor groove quinoxaline rings stacking on either side of the groove. helix. The although the The intercalated overall different similarity of the of the rings of structures contigurations two of the the two two CG base pairs at the end quinoxaline the Here echinomycin, DNA duplex with protrude into the is readily sultur-containing cross major apparent bridges can also oe seen in Figure 2. A more Both 2308 direct antibiotics comparison have a of the Oshaped two antibiotics form in which the is shown cyclic in Figure 3. depsipeptide is Nucleic Acids Research Ala Ser Figure 3: A superposition ot both the echinomycin (E, solid bars) and the triostin A molecules (X, open bars) is shown. In this view the quinoxal m e rings at the upper ana lower part of the diagram are pointing toward the reader ana we are looking toward the side of the cyclic peptide that is facing the nucleic acid. The two pairs of quinoxaline rings are virtually in the same position. There are only small deviations in the position of the various amino acids around the cyclic depsipeptide. The greatest deviation is found near the side chains of the cystine residues in which the disulfiae linkage of triostin A pushes the two vertical peptide units further apart than does the thioacetal cross bridge of echinomycin. organized each in other toward a rectangular and protruding the face ot form on with the the same the depsipeptide two quinoxaline side. In Figure that is exposed quinoxaline rings coming toward the reader. area of difference is near the center rings parallel 3, we right in side this of superposition the ring. The is to the DNA with of the molecule where founa distance at the Ca square (rms) deviation from the the shorter between the atom two cross The greatest of cystine two C a across the Dridge of echinomycin is 0.22 8 less than that root mean to looking It can be seen that the principal bridge in echinomycin brings the peptide chains closer together. deviation are atoms of on the cystine in triostin A. The the position of all of the atoms from each other, excluding the side chains of the two cystines in both antibiotics, is 0.39 8. The side approximately those chain into projecting directly toward orientation two groups, away. the The nucleic of the amino those projecting methyl acid, side while chains the acids toward ot cystine the can be nucleic alanine are side chains divided acid and projecting are both 2309 Nucleic Acids Research Figure 4.: A schematic diagram illustrating the manner in which the quinoxaline antibiotics are inserted into the DNA duplex. The numbering system of the nucleotides in the duplex is shown together with the designation given to the different crosshatched quinoxaline rings. There is a two-fold axis in the middle of the molecule, so that half of the complex is the asymmetric unit. projecting while the away. The serine side valine residues chains are have a somewhat used directly tangential position, the amide bond to form that links the quinoxaline ring to the lactone ester linkage. ORGANIZATION OF THE COMPLEX The which general also form shows of the the complex numbering cross-hatched quinoxaline is shown diagramatically system rings surround used in the identical the center on the outer edge of the CG base pairs. molecule lies on a crystallographic is polynucleotide. to the lower halt. and 4, The the two GC base pairs at either with intercalation between AT and CG base pairs near of quinoxaline in Figure end stacking In the crystal, the two-fold rotation axis, so the upper half A detailed comparison of the interactions between the antibiotic and the nucleic acid fragment is shown in Figure 5. There complex: are three major interactions hydrogen bonding, that intercalation are and involved van der in stabilizing Waals the interactions. There is hydrogen bonding between both alanine residues and the guanine of the two CG base pairs in stippling in Figure 5. the minor have a two-fold symmetry tetrapeptide. This deviation the axis due sulfur to in the hydrogen side the backbone a two-told of the bonding of the diagram Nil of are alanine axis In echinomycin, the asymmetry on the G12, from atoms. also evident right These shown enhanced with from Figure 2, the antibiotic does not axis even though it is built out of two repeats of a positioning of two—fold groove. As is evident is most cross bridge. interactions. forms apparent there could not two hydrogen is hydrogen bonded This asymmetry In Figure 5, the bonds with to the in the be a true the is alanine guanine guanine N3 a ton, while the carbonyl oxygen of alanine receives a hydrogen bond from the guanine N2 atom. 2310 The NH-N hydrogen bond is 2.68 8 for echinomycin compared to 2.91 8 Nucleic Acids Research A) van der Waals contacts Echinomycin Figure 5.: Skeletal diagrams showing the interaction of (A) triostin A or (B) echinomycin with one half of the DNA duplex. The quinoxaline antibiotic is drawn with solid black lines while the DNA is drawn with open lines. Oxygen atoms are shaded black, nitrogen atoms are stippled and sulfur atoms are cross hatched. The numbers represent the distance between atoms in Angstroms illustrating van der Waals contacts of 3.5 8 or less between the antibiotic and the nucleic acid. Nineteen such contacts are found in triostin A compared to 21 in the echinomycin complex. Hydrogen bonds between the alanine residues and the guanine bases are emphasized by stippled shading surrounding the hydrogen bonds. 2311 Nucleic Acids Research tor the triostin A complex. The guanine N2 to the alanine carbonyl oxygen is a long hydrogen bo no in both cases with 3.38 A tor echinomycin and 3.07 8 in triostin A. The other alanine in Figure 5 has only one hydrogen bond trom NH of alanine to the N3 of guanine G2 which compared to 3.01 8 in the triostin A complex. is 2.99 8 in the the echinomycin In both structures, there is no hydrogen bond between the N2 amino group of guanine G2 and the alanine oxygen. In the triostin A complex, the distance between the N2 of guanine G2 to the alanine carbonyl oxygen is 4.1 8 while it is 3.6 8 in the echinomycin complex. Inspection of the two complexes in Figure S reveals that the nucleic acid has undergone a considerable Alanine plays a central chain points toward and G2 on results in one of its role in this reorganization, internal side of the ring tilt and Cll and G12 between the structure. since the C the polynucleotide wedging between a considerable cytosine Cll. reorganization on plane methyl side the sugar residues Cl the of other guanine side. base This G12 and In the echinomycin complex, these two bases have 11.3° dihedral angle between them in contrast to the 17° found in the triostin A complex. the other side of the double helix, there G2 and cytosine Cl is a tilt of 16.5° between in the echinomycin complex, compared the triostin A complex. The differences between On guanine to the 26° found in these tilts are due to their high sensitivity to the exact position of the alanine methyl side chains. Because the same of this tilt, Cl is hydrogen bonded plane. These are tilted 14° from to G12 but each other is not in the lying on echinomycin complex compared to 20° in the triostin A. Likewise, G2 is hydrogen bonded to Cll but is tilted 14° in the echinomycin complex compared triostin A complex. found for Despite the hydrogen bonding interactions between two quinoxaline to 23° seen in the this tilting of the bases, normal distances rings, 013 and 014, are the bases. virtually parallel However, with each are the other, with only a 7° deviation in echinomycin and a 6° deviation in the triostin A. Stacking of bases is generally considered a stabilizing interaction with interplanar spacings between unsaturated ring systems near 3.4 8. However, in these antibiotic—DNA complexes several interplanar distances have increased so that there is a decrease in stabilization. The partly over cytosine Cll but the distance between echinomycin under planes the is complex. 2312 ana 3.97 pyrimidine 3.46 8 in 8 for ring the the of triostin adenine echinomycin A quinoxaline complex. A10, complex but and ring 014 lies the rings is 3.68 8 for the the Similarly, distance 3.53 8 in 014 between the stacks their triostin A Nucleic Acids Research VAN HER WAALS INTERACTIONS A major stabilizing interaction between the antibiotic ana DNA is due to the large number of van der Waals interactions which are shown in Figure S for both complexes. distances These der Waals interactions. echinomycin acid the echinomycin that are 3.S 8 or less. van subtle In and triostin in the to the A changes in the alanine side chains complex chain has moved over Cll. serine so that changes echinomycin relative antibiotic ana triostin group of the van antibiotics oxygen 01' of sugar G2. binding. The total represent the nucleic with the clearly in association In in contact with the furanose altered ot seen as the alanine the interactions an oxygen contact atom is of in side ring of the sugar residue G12 have positioning of the with found the valine adenine interactions between between nucleic and All) to those seen the in the for methyl acid. In the is 2.86 8 between an alanine methyl group and In echinomycin, the closest contact and oxygen 02 of cytosine shorter than normal van der Waals distances and they reflect the are 19. shown for which to Waals stacking associated in echinomycin compared between the N-methyl group of valine addition of different there side of Figure S. contacts are der Waals and der to ring relative to triostin A. There are fewer closest van three van der Waals contacts; however, number and T3 ring backbone the right it is now the 21 of differences This can be seen most triostin A the closest distance an the antibiotic antibiotic Likewise, residues G2 A. The of between triostin A with in echinomycin. residues the two additional Some of the contacts lengthened a number shown on triostin A complex, there are only the echinomycin are of the van der Waals contacts of the different sulfur cross bridges. with exclusive reveals positioning there In the triostin A complex, are A detailed comparison shifts due distances complex, energy of the binding is thus is 2.90 Cl. 8 Both are the tightness of due to a complex stabilizing and destabilizing interactions. STACKING OF THE BASES The stacking of the bases ana the quinoxaline the approximate helix axis. A complex; rings is best viewed This is illustrated in Figure 6 for the the results for echinomycin are similar. in which the base pairs G2-C11 down triostin Figure 6A shows the way stack upon C1-G12 with considerable unwinding as the twist angles between the base pairs is only 10° compared to the normal 36° that is found in B-DNA. In Figure 6B, the quinoxaline rings Q13 and 014 (see Figure 4) have been added to Figure 6A so that their position relative to the base pairs can be seen. The quinoxaline rings stack upon both bases 2313 Nucleic Acids Research Future 6_: Stacking interactions in the triostin A-UNA complex. All views project along the helix axis of the complex. (A) Base pair G2-C11 over C1-G12. There is an unwinding angle of 26° between base pairs. (B) The quinoxaline rings 013 and 014 are drawn on either side of the base pairs shown in ( A ) . The solid quinoxaline ring is closer to the reader while the dashed one is further away. It can be seen that the quinoxaline rings span both bases in the base pair with some overlap. (C) The stacking interactions on either side of quinoxaline ring 014 are shown. The base pair T3-A10 is closer to the reader while the base pair C11-G2 is further away, behind the quinoxaline ring. It can be seen that the quinoxaline ring stacks very well with adenine A10 in the svn conformation and its attached carbonyl group sits over T3. If adenine A10 were in the anti conformation, there would be virtually no stacking. (D) The intermolecular stacking of quinoxaline rings between the different complexes. This stacking stabilizes the organization of these complexes into long rods that pass through the lattice. 2314 Nucleic Acids Research although there is somewhat better stacking upon the pyrimidine rings than the purine ring of the base pair. Figure 6C shows the way in which the CG ana AT base pairs surround the quinoxaline CU.4. As was the case in the triostin A complex, the central two AT base pairs in the echinomycin complex are pairs. Instead, the thymine residues form described adenine two no longer residues hydrogen by Hoogsteen [7] . held together are in bonds as Watson-Crick the svn conformation, with the adenine in These Uoogsteen base pairs involve base and a the manner the N6 amino group of adenine hydrogen bonding to thymine 04, while the thymine N3 donates a in hydrogen in echinomycin distance bonding complex Cl' to are to Cl' adenine similar from the N7. to The those sugar hydrogen in atoms the bond triostin across the lengths A AT base pair shorter tor Hoogsteen base pairs than for Watson-Crick base pairs. the effect of bringing the polynucleotide The is 2 S This has closer to the quinoxaline makes it possible for there to be a large number the complex. rings and of stabilizing van der Waals contacts as shown in Figure S. As shown in Figure 6C, the quinoxaline ring stacks directly upon the pyrimidine ring of adenine Alt) but does not stack at all with pyrimidine T3. The base pairs G2'C11 ring ana A10'T3 have a twist angle of 26° and therefore an unwinding angle of 10°. Studies of DNA angle of 45 in solution with echinomycin have to 55° per echinomycin molecule in the echinomycin complex to make is an unwinding [8]. an estimate angle of 10° between the and tinally an unwinding be another comparable angle of 26° between of this in the complex. may assume that the the 1U° unwinding angle at unwinding angle pairs between the quinoxaline angle the quinoxaline is residue 014, there would also the other end of the molecule. of 26° associated with residues will be found that There the two CG base pairs bracketed In a continuous DNA molecule, unwinding unwinding information two AT base pairs, also 10° between the AT base pairs ana the CG base pairs around by the two quinoxaline rings. revealed an We can use the found around the two in solution. 014 is We CG base Likewise, likely to be representative, and for two rings we come to a total of 46° of unwinding. We have ignored the unwinding found between the two Hoogsteen base pairs found in the center of the complex. However, as we do not know how many Hoogsteen base pairs will fora in the vicinity of the intercalative estimate their contribution to unwinding. total unwinding would be close to 6 6 ° . the average only one extra site, it is difficult to If there were two on each side, the However, if we assume that there is on Hoogsteen base pair on either side of the 2315 Nucleic Acids Research 2316 Nucleic Acids Research quinoialine close to ring, then the observed the estimate value. of the unwinding The unwinding data of angle of 56° would the complex be is thus in general accord with the unwinding measurements observed in solution. FACTORS STABILIZING THE LAITICE There are two fundamental of both the stacking 6U). echinomycin interaction Secondly, forming successive between these sheets. ana the linear Successive layers. interactions used in building up the triostin A complexes. two-fold rod-like sheets related elements are There Q13 is an quinoxalines pack parallel tilted lattices end-to-end relative (Figure to each to each Figure 7A shows a view down the c axis with the A-DNA complexes organized along the diagonals of the unit cell. other other in triostin Each isolated figure represents the superposition of two complexes packed at an angle of 54° to each other ana stacked on the quinoxaline residues Q13. The interactions between which shows a view down the diagonal the a axis. elements can be seen in Figure 7B, The rods alternate in their with one rod tipping toward the reader at the top of the diagram rod tipping away from the reader at the top of the, diagram. In Figure 7B it should be noted that there are two kinds of interaction between rods. One set of interactions antibiotic-DNA residues has a hydrated magnesium (stippled regions) while close lntermolecular sulfur—to-sulfur contact A more magnesium detailed ions view contribute diagram of Figure 8. of to the manner the lattice orientation and the next the sheets of complex between the the other interaction has a (4.0 8 ) . in which the octahedrally interaction is seen in hydrated the stereo Two opposite faces of the octahedron form hydrogen bonds with the base pair G2-C11 and quinoxaline QJ.4 from complexes that are oriented at an angle to each other in the diagonal array. The six water molecules are bound with hydrogen bond lengths of 2.80, 2.83, 2.VI 8 from water molecules to quinoxaline N3, 06 of G2 and N7 of G2 respectively. while the magnesium complex plays a key role It should in organizing be noted the that lattice by Figure 2: The lattice packing of the triostin A-DNA complex. (A) The view down the .c axis is shown ana the lattice is outlined. Two different levels of complexes are shown, which are organized as rod-like arrays that run across the diagonals of the figure. The t> axis is vertical and ji axis is horizontal. The rods along the diagonals are stabilized by the stacking interactions between the quinoxaline rings at the outer edges of the complex. (B) The outlined lattice is shown viewed down the a axis, b is vertical, c is horizontal. The long rod-like arrays of complexes are tipped so that alternate ones are successively pointing toward and away from the reader. There are two different types of interactions between the rods, one of which is stabilized by the octahedrally coordinated magnesium ions (some are stippled) that are located between every other pair of rods. 2317 Nucleic Acids Research Figure 8: A stereo diagram illustrating the manner in which the hydrated magnesium ion with octahedral coordination torms hydrogen bonds to two different antibiotic-DNA complexes. In the unrefined structure, the Mg-ILO complex was very close to a regular octahedron. During the refinement, the Mg-0 distances in the complex were all constrained to remain at 2.0 A in a regular octahedral geometry. The magnesium ion lies on the b axis which is a two-fold rotation axis. Opposite octahedral faces which are related by the two—fold axis form three hydrogen bonds to each DNA quinoxaline complex. The water molecules are hydrogen bonded to both the intercalating quinoxaline rings as well as the CG base pairs in the center of the complex. bridging 7B, it the successive can also serve layers to of the DNA-drug stabilize the complexes complex in as shown in Figure solution using hydrogen bonding interactions from one face of the octahedron. the three It is possible that the role of the metal ion complex in stabilizing the complex by hydrogen bonding both to the base pairs of UNA and to the intercalated quinoxaline ring in the major have groove observed side ot the DNA double helix is a more a similar between daunomycin noticed that ana other interaction d(CUTACG) metal ions in the crystal (unpublished such as general one. structure data). calcium of the Furthermore, and barium ions We complex we have in the crystallization mixture produced different crystal lattices in accordance with these observations. SPECIFICITY OF THE INTERACTION OF QUINOXALINE ANTIBIOTICS WITH DNA Binding preference studies tor contents [8]. of both triostin A double-stranded DNA, and echinomycin with triostin A to DNA have preferring shown a higher GC Footprinting studies have shown that echinomycin and triostin A cover a binding site of tour to six Dase pairs [11, 12, 13] and the tightest binding was found when the sequence d(CpG) was located in the central two base pairs. The outer base pairs usually had AT or TA sequences. in agreement with the details of this structure. of 2318 the central two base pairs is determined This result is The CpG sequence by the specificity hydrogen bonding Nucleic Acids Research interaction different guanine of the sides of alanine the specificity the guanine adenine bond. G2 with helix. for the has not a It necessarily similar N3 the is alanine The other alanine does residue double is high bonds with guanine. residues guanine bases interesting residue to position it that note the two that the that forms two hydrogen that forms a single define on for hydrogen guanine. could In receive bond to tact, an the hydrogen However, the footprinting experiments [11, 12, 13] show that all of the strong binding sites have the sequence CG in the center of the four base pairs ana only a tew weak binding sites have a sequence TG. The asymmetry we observe in the hydrogen bonding to guanine residues may not necessarily be an inherent feature of these quinoxaline antibiotics that the UNA fragments we are using are but may reflect short and the quinoxaline the fact ring 013 is not intercalating the base pair that normally stacks on the other end. The strongest binding sites have a preference for AT base pairs flanking the central CG sequence. base pairs. However, Hoogsteen base pairs can be formed readily with AT in order to form them with CG base pairs must be protonated on N3 in order to hydrogen bond with the cytosine the guanine N 7 . This apparently occurs less readily because flanking CG base pairs occur largely in the weak binding sites rather this suggests that Hoogsteen base strands CG pair geometry approximately Watson-Crick than the Hoogsteen 2 hydrogen 8 strong binding base brings pairs the closer to bonding were two sugar each present can in residues other and sites [12] . form than this in However, solution. on the they turn The opposite would be facilitates if a large number of van der Waals interactions which has the effect of stabilizing the binding. The interaction of these quinoxaline antibiotics with the double DNA fragment make are used to make it possible a bis to to the triostin A. An exact because of echinomycin the these particular intercalating antibiotic. pseudo-two-fold rotation axis due This may be related ask why molecule, symmetry axis the actual of the covalent acids The DNA double helix has a strands. in the molecular formula is not found dispositions helical amino to the anti-parallel nature of the two-fold two-fold skewed eight in the triostin A cystine bonding of side the of the structure chains. In the thioacetal bridge on ribosomes but rather on special enzyme prevents a two-fold axis. These antibiotics are not made systems that can accept not only L-ami no various other amino acid derivatives. D—amino acid ana two N-methylated acids In the amino but present acids. also D— amino instance Only acids there alanine or is one has no 2319 Nucleic Acids Research Table I Functional Descriptions o£ Amino Acids in Qninoxaline Antibiotics Ami no Acid Role £ l Backbone Role of Side Chain Role Qi Modification L-alanine Cu methyl group separates adjacent sugars, wedges guanine base out of plane; van der Waals bonding to sugars NH groups form hydrogen bond to both guanine N3; carbonyl group hydrogen bonds to one guanine N2 L-cystine Sulfur-containing cross linkage provides rigidity to cyclic peptide and is positioned on the side away from the base pairs NH groups are methylated; N-methyl gronp prevents NH carbonyl groups project hydrogen bonding and allows alanine carbonyl group in away from base pairs planar peptide linkage to be oriented for hydrogen bonding L-valino C methyl groups are in der Waals contact with sugar-phosphate linkage; other quinoxaline antibiotics exist with different side chains here, including more aliphatic groups NH groups are methylated; carboxyl groups form an ester linkage to serine side chain; one carbonyl group i s in contact with adenine base outside intercalative s i t e N-methyl groups are in close contact with cytosine rings, changing base planes; methyl groups prevent NH hydrogen bonding and distortion of the structure D-sorine Side chain provides ester linkage to form cyclic structure. NH gronp i s used to form peptide bond to quinoxaline, cannot be Nmethylated due to steric hindrance; carbonyl gronp projects away from base pairs. D-amino acid i s needed to position quinoxaline ring perpendicular to cyclic depsipeptide structure. modifications. in these individual central Table 1 contains a functional two quinoxaline antibiotics. side chains, the backbone role in both hydrogen description of the amino acids It summarizes or thioacetal the roles and the modification. bonding with the bases interaction of its methyl group with the sugars. forms a disulfide Cannot have modifications and maintain function linkage that of the Alanine plays a residues and in the In L-cystine, the side chain rigidifies the cyclic peptide. Its Nil group is methylated, which prevents the intramolecular hydrogen bonding that has been seen in the non-methylated analog of triostin A, TANDEM [14]. The NH group of cystine is linked to the carbonyl group of alanine planar peptide linkage. fixed by hydrogen carbonyl bonding to hydrogen through a Orientation of the alanine carbonyl group must not be of the cystine bond with guanine* NH group This in order to allow the is accomplished by blocking hydrogen bonding in the cystine Nil group through N-methylation. The N—methylation of the L-valine groups plays a different role, since it points directly contacts. stabilizing 2320 toward the cytosine In the absence of a methyl rings group and forms there tight one would van der Waals lose both the interaction due to the methyl group and create a hole that could Nucleic Acids Research not accomodate a destabilization. solvent molecule and thus would represent a focus of Host of the carbon atoms of the valine side chain are not in van der Waals contact with the sugar phosphate backbone, with the exception of the C Cl. atom of triostin A, which is in van der Waals contact with the sugar That contact echinomycin are is both lost first in the members echinomycin of complex. a family of Triostin quinoxaline A The other members of the family differ by having variations on the side of valine. This involves not and antibiotics. chain only the normal amino acid but also side chains in which there are additional carbon atoms that lengthen the side chains. or both valine side chains can be This is in agreement with the fact substituted that this for alloisoleucine is the only side One residues. chain of the peptide that can vary. The serine residue has a side chain that plays a crucial role because its hydroxyl group forms an ester linkage with form the cyclic structure. to position structure. to The alpha amino group of serine is used to form an amide bond to the quinoxaline ring. in order the carbonyl group of L valine the In this case, the D residue is essential quinoxaline rings at right angles to the cyclic Use of the L residue would have the rings in the wrong orientation for intercalation. The quinoxaline occurring Other drugs examples that antibiotics interact include the with represent DNA by another binding anthracyclines class onto of the daunomycin [IS] naturally minor and groove. adriamycin (manuscript in preparation), actinomycin D [16, 17] and nonintercalators as netropsin and distamycin [IB]. Binding to the minor groove has such fewer unique recognition features than does the major groove since, for instance, it is difficult However, to distinguish recent they bind analyses largely protein-UNA of AT from TA base bacterial to the major groove interactions, pairs by hydrogen bonding [19] . repressors [20]. it is possible that bound If that to DNA suggest that is a common feature small molecule binding may of be biased in favor of minor groove binding. The structure DNA have many first examples of both the echinomycin and the surprising features. First in which a peptide has mechanism for the sequence triostin A complexes with of all, it illustrates sequence-specific one of the binding to DNA. specificity involves hydrogen bonding The to bases and also the participation of a conformational change involving the Hoogsteen base pairs on the sequences first time structure on one in which Hoogsteen and it suggests that side of base this the intercalative pairs type have been of site. seen in conf ormational This is the oligonucleotide lability may be 2321 Nucleic Acids Research seen elsewhere as well. On the structural level, the bis intercalative mode of action of the quinoxaline antibiotics that had been inferred from a variety of physical studies [1, 2] demonstrated here. accomodate itself The crystal It was well as model to the more rigid peptide structure found as building [21, 22] has been Finally, the ON A molecule is seen to be flexible enough to that of the the antibiotic in a variety of ways. triostin A alone has been solved recently [23]. dihedral angles of the peptide backbone in that structure are similar to those observed in the DNA-triostin A complex. The structure of a complex of this type provides us with information that can be used to direct the synthesis ot new antibiotics. At the present time, echinomycin in the human what is being cancers [5] . changes might we biological activity? introduced, one exists natural used in phase If we were introduce There II clinical going in order are of which is in variations in this a the its positions can be tew side chain could be decreased through should not produce a side of where chain changes where antibiotics. there These For example, change in the already variations the size of the the introduction of alanine significant of molecule, of valine family treatment the echinomycin to loox. for modifications could be further modified in a number of ways. This trials to redesign or glycine. interaction of the antibiotic with the UNA but might modify the pharmacodynamics of the molecule. Another example is to introduce positive charges by replacing a lysine residue at that position. This may increase DNA due to charge attraction. the binding of the antibiotic with the A negatively charged glutamic acid that position might also change the biological effectiveness Examination of intercalative larger the quinoxaline element overlaps intercalation. with with three the bases ring suggests fused rings would and lead biological a make the in somewhat for larger significantly stabilization due to Finally, if one or both of the alanine residue were changed to glycine, there will be no methyl group might modify that residue of the molecule. to a significantly activity of the to wedge between the base pairs, which modified antibiotic. interaction These that ideas would affect be explored can the by chemical synthetic methods to search for more effective antitumor agents. Acknowledgments: Institutes Office the of of Health, Naval Netherlands acknowledges This research was supported by grants from the National National Science Research, National Organization support from NATO for the and Advancement American Cancer We Society, Space Administration of Pure Research. ano Istituto Strutturistica Chimica, Nazionale Delle Richerche (Italy). 2322 Foundation, Aeronautics thank Dr. T. Yoshida and G.U. Consiglio of Shionogi Co., Nucleic Acids Research Osaka, Japan, tor the t r i o s t i n A ana Dr. M.J. Waring of the University of Cambridge, England, for the echinomycin. §To whom correspondence should be addressed 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. HCES Katagiri, K., Yoshida, T. and Sato, K. (1975)in Antibiotics III Mechanism of Action of Antimicrobial and Antitumor Agents, J.W. Corcoran ana F.E. Hahn, Eds., Ed., pp. 234-251, Springer-Verlag, Heidelberg, Waring, H.J. (197!') in Antiobiotics V Mechanism of Action of Antieukaryotic and Antiviral Compounds, F.E. Hahn, Ed., Ed., pp. 173-194, Springer-Verlag, Berlin, , part 2 Ward, D.C., Reich, E. and Goldberg, I.U. (1965) Science 149, 1259-1263. Sato, K., Shiratori, 0. and Katagiri, K.J. (1967) J. Antibiotics 20, 270-276. Foster, B.J., et al. (1985) , Submitted for publication Wang, A.H.-J., Ughetto, G., Uuigley, G.J., Hakoshima, T., van der Harel, G.A., van Boom, J.H. and Rich, A. (1984) Science 225, 1115-1121. Hoogsteen, K. (1959) Acta Crystallography 12, 822-823. Lee, J.S. and Waring, M.J. (1978) Biochem. J. 173, 115-128. van der Marel, G., van Boeckel, C.A.A., Wille, G. and van Boom, J.H. (1981) Tetrahedron Letters 22, 3887-3890. R. Srinivasan Ed., (1979)Biomolecular Structure: Conformation, Function and Evolution,, 1, Pergamon, Oxford, . Low, C.M.L., Drew, H.R. and Waring, M.J. (1984) Nucleic Acids Research 12, 4865-4879. van Dyke, M.M. and Dervan, P.B. (1984) Science 225, 1122-1127. Low, C.M.L., Olsen, K.K. and Waring, M.J. (1984) FEBS Letters 176, 414-420. Hossain, M.B. e£ al. (1982) J. Amer. Chem. Soc. 104, 3401-3408. Uuigley, G.J., Wang, A.H.-J., Ughetto, G., van der Marel, G., van Boom, J.H. ana Rich, A. (1980) Proc. Natl. Acad. USA 77, 7204-7208. Sobell, H.M., Jain, S.C., Sakore, T.D. and Nordman, C.E. (1971) Nature (London) New Biol. 231, 200-201. Xakusagawa, F.M., Dabrow, S., Neidle, S. and Berman, U.M. (1982) Nature (London) 290, 466-467. Zimmer, C. (1975) Progr. Nucl. Acid. Res. Mol. Biol. 15, 285-318. Seeman, N.C., Rosenberg, J.M. ana Rich, A. (1976) Proc. Natl. Acad. Sci. USA 73, 8U4-808. Pabo, C O . and Sauer, R.T. (1984) Ann. Rev. Biochem. 53, 293-321. Ughetto, G. and Waring, M.J. (1977) Mol. Pharmacol. 13, 579-584. Viswamitra, M.A. etal, (1981) Nature 289, 817-819. Sheldrick, G.M., Guy, J.J., Kennard, 0., Rivera, V., Waring, M.J. (1984) J. Chem. Soc. Perkin Trans. II 1601-1605. 2323