Journal of Biomolecular Structure & Dynamics, ISSN 0739-1102 Volume 9, bsue Number 5 (1992), "Adenine Press (1992).

Model for the Porphyrin-DNA Binding Site: ENDOR Investigations ofCu-Porphyrins Binding to DNA Steven P. Greiner 1 R.W. Kreilick 1 and Luigi G. Marzilli2 7Department of Chemistry University of Rochester New York 14627

Ro~hester,

2

Department of Chemistry Emory University Atlanta, Georgia 30322

Abstract 0

Proton ENDOR has been observed from frozen solutions (ca. 38K of copper meso-(4-Ntetra-methylpyridyl)porphyrin (CuTMpyP(4))complexed with Salmon sperm DNA in water and D 20. Lines from exchangeable protons of the DNA bases have been observed in these ENDOR spectra. Analyses of these ENDOR data show that the separations of these DNA protons from the copper atom are between 3.76 and 3.84A with angles of 19.5 to 22.5 degrees between the Cu-H vectors and the gz axis. A distant ENDOR response has also been observed from phosphorous nuclei in the DNA backbone. We estimate that the phosphorous atoms producing this ENDOR signal are 7.5-10 A from the copper center of the porphyrin. These END OR data combined with results from an earlier NMR investigation (1) have been used to construct a computer simulated model ofthe binding site in which the porphyrin is partially intercalated and extends into the major groove of DNA. The two GC base pairs at this site are slightly inequivalent. For each, the G imino proton and one of the C amino protons are at appropriate positions to account for the ENDOR signals arising from exchangeable protons. It is unlikely that this inequivalence would persist at room temperature where dynamic processes would give an apparently symmetric interaction. Although the model accounts for all reported experimental data involving tetracationic porphyrin species which have been suggested to be intercalators, it is not a unique solution. )

Introduction Recent investigations of porphyrin-DNA interactions have shown that the cationic porphyrin, meso-(4-N-tetra-methylpyridyl)porphyrin (TMPyP(4)) and, in some cases, itsCu(II)complex(CuTMpyP(4))mayintercalateinDNA(l-17).AnNMRstudyofthe interaction ofTMpyP(4) with poly(dG-dC) and oligodeoxyribonucleotides showed that this porphyrin preferentially binds at a 5'CG3' site (1). Such porphyrins also bind to poly(dA-dT) but optical and circular dichroism spectra as well as other physical studies indicate that the porphyrin is bound on the outside of this polymer (4) or is intercalated to a very limited extent (2,5,6,8,9,10,13,14). A variety of molecules are known to intercalate in DNA but none have the steric bulk of these porphyrins.

837

838

Greiner et a/.

A local denaturing of the DNA helix is necessary for these porphyrins to enter the DNA helix (2,19). Mter the helix opens and the porphyrin moves into position, the basepairs would close upon the molecule and the helix would reform (2,10). This type of DNA breathing (18) is necessary for full intercalation of the porphyrin but may not be necessary to describe other modes of binding. We have conducted an Electron Nuclear Double Resonance (ENDOR) investigation of the interaction ofCuTMpyP(4) with Salmon sperm DNA. Earlier reports from one of our laboratories (19,20) have shown that ENDOR spectra from polycrystalline materials can be analyzed to yield information about the relative geometry of nuclei near the paramagnetic copper atom. We have taken and analyzed ENDOR spectra from CuTMpyP(4) and from copper meso-(2-N-tetra-methylpyridyl) potphyrin (CuTMpyP(2)) in aqueous glasses in the presence and absence of DNA. We have also analyzed data from copper tetraphenylpotphyrin (CuTPP) as a reference for analysis of spectra from the pyridyl substituted porphyrins. Our initial analysis of these copper complexes in the absence of DNA allowed us to determine changes in the spectra induced upon formation of the DNA-porphyrin complex. The spectrum of CuTMpyP(4) showed a signal from an additional proton when DNA was added while the spectrum ofCuTMpyP(2) was unchanged. Earlier investigations have indicated thatCuTMpyP(4)bindstoDNAbyintercalationwhileCuTMpyP(2)maybebound on the surface (5). The steric bulk of the ortho methyl groups ofCuTMpyP(2) keeps the plane of the pyridyl rings twisted at large angles with respect to the porphyrin plane and it is energetically unfavorable for this molecule to fit into the DNA structure. Our results are in accord with these observations. The spectra from the CuTMpyP(4)DNA complex also showed a signal from phosphorous atoms in the DNA backbone. We have analyzed these ENDOR data to yield a geometric model for the manner in which CuTMpyP(4) binds to DNA. Theory

The EPR spectra of copper potphyrins exhibit nearly axial symmetry with a relatively large g anisotropy and smaller anisotropy from the copper hyperfine interaction. The spectrum of a non-crystalline sample shows contributions from all molecular orientations but at any given field value one often observes transitions from unique molecular orientations. The orientation ofthe molecular g axis system for this set of molecules can be determined if the principal components of the g and copper hyperfine tensors are known (19,20). The ENDOR spectrum taken at a given EPR field position therefore yields NMR transitions for a group of molecules at a fixed orientation of the g axis system with respect to the field. IfENDOR spectra are taken at a series ofEPR fields one obtains a set ofNMR spectra of molecules with different molecular orientations which can be analyzed to yield information about nuclear coordinates in the g axis system. The NMR spectra obtained with this technique show lines from all nuclei which interact with the unpaired electron spin of the copper atom. The electron-nuclear interaction consists of two terms: the Fermi contact interaction and the electronnuclear dipolar interaction. The Fermi contact interaction requires the electron

Porphyrin-DNA Binding Site

839

spin to reside in an s orbital ofthe nuclei in question and this term is generally small for nuclei which are not directly coordinated to the metal atom or in a conjugated system through which the spin can be de localized. The dipolar interaction varies as the inverse cube ofthe distance of the nuclear spin from the metal center and shifted NMR lines are resolved only for nuclei which lie within about a 7.5 Aradius of the metal center. One can generally use a point dipole approximation to analyze data from nuclei which are not coordinah:!d to the metal center and which have relatively small isotropic coupling constants (20). This approximation has been used for determination of the nuclear coordinates reported in this paper. The equation which relates the observed shifts ofENDORlines to the Fermi contact term Ap the electron-nuclear separation (r), and the direction cosines of the vector connecting the electron and nuclear spin (u) with respect to the gz axis is given by (19,21):

vd

en or

=[L[{Ms)Ln.g.-n.v.A . ]] i

\ geff

.i

.1 .1

1

1

.JI

[1]

with

[2] [3] For molecules with axial or near axial symmetry, the EPR signal is independent of the angle phi () in the gx- ~plane and the ENDOR spectra exhibit 2 dimensional powder patterns in which all values of phi () are allowed. One observes ENDOR peaks (turning points) when the derivative of the ENDOR absorption frequency with respect to theta and phi (9, ) is large. The EPR powder pattern allows selections of unique values of the angle (9) between the gz axis and the external field vector and ENDOR spectra taken at different EPR fields arise from selected sets of molecular orientations. If the principal components of the g tensor and copper hyperfine tensor are known one is able to determine the angle 9 for any field setting. If ENDOR spectra are taken at a series of different orientations one is able to use equation [1] with a computer iterative search (22) to determine Ap r, and the angle which the vector connecting the nuclear and electron spins makes with the g2 axis (9). This type of analysis has been carried out for CuTMpyP(4), CuTMpyP(2), and CuTPP in frozen glasses. The effect of the porphyrin ring's nitrogen hyperfine interaction was included in the calculations of the EPR angular selection. When the CuTMpyP(4)DNA complex was formed we observed a new signal from protons on the nucleic acid bases. An identical procedure was used to determine AF, r, and 9 for these protons. The complex also exhibited a line at the phosphorous Larmor frequency (23).

Greiner et a/.

840

This type of distant ENDOR signal is observed when the electron-nuclear interaction is strong enough to perturb the relative population of the electron spin energy states but too weak to shift the line outside of the line width of the signal at the Larmor frequency (23). We are able to observe shifted signals from nuclei greater then 7 A from the metal and estimate that the phosphorous nuclei must be between 7.5 to 10 A from the copper atom.

Experimental The CuTMpyP(4) and CuTMpyP(2) chloride salts were obtained from Mid-Century. The Salmon sperm DNA samples were prepared as described previously (3) and were dissolved in a PIPES 10 buffer. The EPR/ENDOR samples were prepared by adding 1 ml ofdeionized water or D20 to the appropriate amount ofporphyrin and stirring slowly for one hour. The DNA solution was added to these samples. The TPP sample was dissolved in toluene d-6 with about 5 percent CDC13 which made a good glass upon freezing. All samples were made to be paramagnetically dilute, typically about 5 millimolar of copper complex. The solutions were estimated to be 44 Mm in DNA so that the "r" value (24), the number of moles of bound porphyrin per mole of total basepairs was ca. 0.05. At these low values of"r" all the porphyrin is expected to be bound and remain paramagnetically dilute. Thus small differences in porphyrin concentration only effect the overall signal to noise and not the position of peaks in the ENDOR spectra. The EPR and ENDOR spectra were acquired on a modified BRUKER ER200-D spectrometer. Experimental conditions which stayed consistent for all ENDOR spectra consisted of 19Khz FM frequency modulation (30Khz frequency excursion) and RF power over 100 watts while saturating the EPR transition. Microwave power typically was between 0.2 and 4 milliwatts. Other parameters such as temperature (near 38 K were allowed to vary slightly for different samples. Further information about the experimental setup and apparatus is given by Hurst (25). 0

)

An important requirement for angle selected ENDOR experiments involves selection of field locations where the ENDOR spectra allow observation of changes in line positions as a function offield or selector angle. One must select field values that minimize overlap of the ENDOR lines in order to monitor shifts as a function of selector angle. Any location of the EPR spectrum can be used as long as the angle selection (governed by molecular orientation effects within the glassy sample) are accounted for (26) and the above criteria is fulfllled. In the experiments reported here, ENDOR field positions near the low field turning point, and also at intermediate values within the EPR powder pattern were used. These field values were chosen because they gave well resolved ENDOR spectra which allowed assignment of ENDOR lines to specific protons. The EPR and ENDOR experimental parameters reported in Tables I and II were obtained by using computer simulation of experimental spectra. Programs for simulation ofENDOR and EPR spectra have been written in both Fortran and the ASYST programming languages. Most simulations were conducted with 80286 or 80386 IBM type PCs. These programs included terms to account for the effect of both the copper and nitrogen hyperfine interactions on angular selection.

841

Porphyrin-DNA Binding Site Table I EPR Parameters for the Cu Porphyrins Compound

gz

gx,y

A. (Cu)a

N·Y(Cu)a

CuTPP CuTMpyP(4) CuTMpyP(4) /DNA CuTMpyP(2) CuTMpyP(2) /DNA

2.185 2.209 2.195 2.212 2.203

2.050 2.055 2.055 2.055 2.055

631 598 615 593 605

42.6 40.8,40.5 42.8 40.1,40.7

a The coupling constants are in Mhz

Table II ENDOR Parameters for the Cu Porphyrins Compound CuTPP: This Work CuTPP: Single Crystal"

Nuclei

~(MHz)

en (deg)

Pyrrole H o-phenyl H

1.17 -0.09

91.7 62.5

Pyrro1e H

X-ray

Pyrrole H o-phenyl H

CuTMpyP(4)

Pyrro1e H o-pyridy1 H

4.97 5.59 5.16

1.32

5.18 5.54 1.17 -0.09

90.0 64.0

4.96 5.74

CuTMpyP(4) /DNA

Pyrrole H DNAH

1.25 0.2

92.0 18-38

4.74 3.7-4.1

CuTMpyP(2)

Pyrrole H

1.17

94.9

4.95

a Reference 27 h Reference 28

Computer modeling was carried out with a molecular graphics program written using the ASYST programming language using an IBM type PC. This program utilizes atomic coordinates from x-ray crystallographic data to ftx the relative position of atoms included in the model. A perspective transform is used to give a 3-D view on a 2-D surface. The program allows rotation and translation of a given molecule so the porphyrins can be easily positioned with respect to the DNA One is also able to rotate the complete porphyrinDNA complex to view the molecules from different orientations.

Results and Discussion The ENDOR spectra ofCuTMpyP(4) and CuTMpyP(2) in frozen glasses were acquired at about 38K. The EPR spectrum ofCuTMpyP(4) in the presence of Salmon sperm DNA is shown in Figure l. This figure also shows the field positions at which ENDOR data were acquired. The EPR spectrum of CuTMpyP(4) in the absence of DNA was almost identical to this spectrum and ENDOR data were taken at nearly the same field positions in order to have data sets which reflected the same angular selection. The EPR parameters for the various complexes are listed in Table I.

Greiner eta/.

842 c

I d

I

I

I

I

I

I

I

2600.

I

I

I

I

I

I

I

I

I

I

2800.

I

I

I

I

I

I

I

I

I

I

3000.

I

I

I

I

I

I

I

I

I

I

3200.

I

I

I

I

I

I

I

I

I

3400.

Figure 1: Experimental EPRspectrum ofCuTMpyP(4)-DNAcomplex. The arrows show the EPR fields where ENDOR spectra were acquired. Microwave frequency: 9.00 GHz, temperature ca. 38K, field positions are a: 2630 gauss, b: 2762 gauss, c: 3126 gauss, d: 3155 gauss and e: 3271 gauss.

The ENDORspectra ofCuTMpyP(4) in the presence and absence ofDNAin an H 20 glass are shown in Figure 2. Spectra in a D 20 glass are shown in Figure 3. The x axis in these Figures show the deviation from the proton Zeeman frequency (Mhz ). The ENDOR spectra ofCuTPP, CuTMpyP(4), and CuTMpyP(2) are readily analyzed to yield the isotropic and dipolar couplings of the pyrrole and ortho-aromatic protons via spectral simulation techniques (26). The aromatic rings of these compounds are twisted with respect to the porphyrin plane so there is little spin delocalization into these rings. The isotropic coupling of the aromatic protons is small because of this large twist angle. The meta and para protons of the aromatic rings are too distant (greater than 7 Afrom the copper center to yield a measurable shift of the ENDOR lines and accurate data could only be obtained for the ortho protons. The dipolar couplings were analyzed to give the Cu-H separations (r) and the angle the Cu-H vector makes with the gz axis en. Values for AF, r, and en are listed in Table II. Data from X-ray data (27) and from a single crystal ENDOR study ofCuTPP (28) are given in this table for comparison. The values from our analysis of disordered samples are in good agreement with the single crystal values, even though there need not be any strong correspondence between crystal and amorphous sample magnetic parameters due to site dissimilarity in amorphous samples. The Cu-proton separations, en and isotropic coupling constants are similar for all of these molecules. When DNA is added to an Hp solution ofCuTMpyP(4) one observes a new ENDOR

843

Porphyrin-DNA Binding Site 0

0

8

_,,, b

I

I! -1.7

c

1.7

I

-

1.7

-1.7

-

I 1.7

-1.7

I

e

-1.7

ll.

I

I

D

D.M

O,P O,P

-1.7

I

I

-t.7

-1.7

I 1.7

~-

1,7

-1.7

I

1.7

Figure 2: Left; Experimental ENDORspectra of the CuTMpyP(4)-DNAcomplex in Hp. Right; Experimental ENDOR spectra ofCuTMpyP(4). The abscissa is the distance from the proton Larmor frequency in MHz. The solid and dashed lines in the spectra on the left indicate theoretical peak positions from guanine imino protons located at 3.76 A and 3.82 A with 9" values of 19.2 and 21.7 degrees respectively. The arrows in the spectra on the right indicate the ortho pyridyl and pyrrole protons (0 and P respectively). The letters indicate the EPR field position where the ENDOR data was acquired (Figure 1).

signal from protons on the DNA base pairs. This signal disappears in D 20 solution indicating that it arises from exchangeable amino or imino protons. The signals from DNA protons are relatively broad. These lines are marked with the theoretical line positions calculated from our binding model, with lines on the left side of

Greiner et at.

844

~'----------~'------~~~' ~'~----~~~~--------J

·-1.1

0

1.1 -1.1

0

1.1

c

~~rz----------~L~--------~1 ~'-...-------~----------~1 -1.7 0 1.7 -1.7 1.7

D

~'-1.1 r..------=-~' ----.....-1'.1.7 -1.7 ~=----"'="' ------J.J 0 0 1.7 _.:....!

~~~--------~~----------~~L~~------------~----------------~ -1.7 0 1.7 -1.7 0 1.7 Figure 3: Left; Experimental ENDOR spectra of the CuTMpyP(4)·DNA complex in D 20. Right; Experimental ENDOR spectra ofCuTMpyP(4). The abscissa is the distance from the proton Larmor frequency in Mhz. The letters indicate the EPR field position where the ENDOR data was acquired (Figure l).

Porphyrin-DNA Binding Site

845

Figure 2. These peaks are not modulation broadened. Their breadth is indicative of a superposition of lines from a series of protons with slightly different hyperfine interactions. There are many protons on the neighboring basepairs and sugars which are available to interact with the unpaired electron. However, these protons are somewhat removed from the copper atom and exhibit small hyperfine couplings near the proton Larmor frequency. Overlap of this distribution of signals broadens the lines near the Larmor frequency. The lines observed from the porphyrin protons in the D 20 solution are also broadened but the unique signals from exchangeable DNA protons are not observed. One also observes a distant ENDOR signal from the water molecules at the proton Zeeman frequency which overlaps with any lines near the center of the spectra. This signal overlaps with the lines from the ortho aromatic protons and makes analysis of data from these protons inaccurate. The same spectra acquired from frozen D 20 solution shows small changes in the measured hyperfine splittings of the orthopyridyl protons. We were unable to assign these small shifts to actual coordinate changes of the orthopyridyl protons because we were unable to accurately follow peak movement as a function of field. Nevertheless we attribute this as evidence that a slight flattening of the pyridyl groups occurs upon intercalating in DNA For small changes in twist angles, one expects that the change in the distance between the orthopyridyl proton and the copper atom will be too small to give an appreciable changes in the shift of the lines in the ENDOR spectra. The EPR spectral parameters show small increases in Az for copper and" for nitrogen, with a small decrease in gz. These changes indicate some redistribution of the spin density on formation of the complex. A signal at the phosphorous Larmor frequency is also observed. Figure 4 shows phosphorous ENDOR spectra taken at different EPR field positions. The signal to noise of these spectra is poor due to the small number of nuclei producing the signal and the weak polarization ofthe spins in a distant ENDOR response. The lines shift with the external field and remain at the phosphorous Larmor frequency proving that these signals come from phosphorous nuclei. The phosphorous is distant enough from the copper center that we are unable to measure a shift but close enough so that the electron-phosphorous interaction changes the population of the spin states. We estimate the copper-phosphorous separation should be between 7.5 and 10 A to yield this type of result. The ENDOR spectrum of CuTMpyP(2) in the presence of DNA did not show the signal from the exchangeable protons observed in the DNA-TMpyP(4) complex. The EPR spectral parameters ofDNA-CuTMpyP(2) showed a small decrease in gz and a small increase in~ for copper indicating that there is some interaction between DNA and the porphyrin complex. Previous experiments (5) indicated that this molecule does not intercalate in DNA and our results substantiate this finding. The small changes in & and Az can be accounted for by formation of a complex in which the copper porphyrin binds loosely on the exterior of the DNA helix. The results ofCarvlin (5) and Pasternack (2) indicate that TMpyP(4) and CuTMpyP(4)

846

Greiner eta/.

6.9

6.9

6.8

5.05

5.05

5.25

5. 25

5.45

5.65

5.85

5.45

5.65

5.85

:Fieqae&y, MHz Figure 4: Phosphorous ENDOR spectra from the CuTMpyP(4)-DNA complex taken at EPR field positions c and d as indicated in Figure I.

partially or fully intercalate in DNA with some specificity for regions containing GC base pairs. The NMR studies (1) of a group of oligodeoxyribonucleotides with metal free porphyrins indicated that the porphyrins bind preferentially to 5'CG3' sites. The chemical shifts of protons on bases in the binding site were large enough so that lines from these nuclei could not be observed or, alternatively, the porphyrin facilitated exchange with H 20. A shifted signal from imino protons on base pairs on either side of the binding site was observed. The observed shift can be correlated with porphyrin ring currents to show that these protons must be above or below the middle region of the aromatic porphyrin ring. A model for the porphyrin binding site must account for the ENDOR signal from the exchangeable amino or imino protons, the ENDOR signal from phosphorous and the NMR signal from the imino protons on base pairs adjacent to the binding

Porphyrin-DNA Binding Site

847

site. Coulombic attraction between the positively charged pyridyl nitrogens and the negatively charged phosphate groups should be optimized by the model. The ENDOR shifts observed at a given EPR field depend on both the Cu-H separation and the angle between the Cu-H vector and the gz axis (9n). If normal hydrogen bonding between base pairs in the binding site is assumed, then the relative position of the guanine amino and imino protons and the cytosine amino protons is fixed. One is therefore able to determine both the Cu-H separation and en for all amino and imino protons when the porphyrin is placed at a fixed position in the helix between GC base pairs. We used data from the crystal structure of ethidium cation intercalated in CpG to establish the relative positions of the base protons (29,30). We choose d(GCG)(S'-3') duplexed to its complement for our model. These porphyrins are large sterically bulky molecules and full intercalation into DNA can only be accomplished by breaking hydrogen bonds between base pairs with some distortion of the DNA helix. After the DNA had opened, the porphyrin could fit between the base pairs, and the DNA would close around it If the porphyrin is fully and symmetrically intercalated, the configuration of base-pairs is similar to that found when ethidium ion binds to DNA (18). The base-pairs are separated by 6.8 Aand the helix is unwound by 26 degrees at the binding site. An alternate mode of porphyrin-DNA binding is via partial intercalation. In this case, less distortion of the DNA helix is necessary, as the porphyrin fits in the major groove with two pyridyl group inserted between the base pairs with their positively charged nitrogen atoms near the phosphate atoms in the DNA backbone. The base pair separation and the degree to which the helix is unwound is the same as the fully intercalated case. Figure 5 shows a schematic representation of the dipolar interactions of the Cu atoms unpaired electron with the various guanine and cytosine amino and imino

Figure 5: Schematic representation of the dipolar interaction of the Cu's electron and various protons on the bases.

Greiner et al.

848 Table III Cu-Proton Separations and Proton Position

Guanine I: Amino Guanine 2: Amino Guanine 1: Imino Guanine 2: Imino Cytosine 1: Amino Cytosine 2: Amino

en Values for the Fully and Partially Intercalated Models Fully Intercalated eon r,A

Partailly Intercalated eon r,A

4.5 4.5 3.7 3.7 4.3 4.3

5.06 5.79 3.76 3.82 3.80 3.84

39.5 39.5 16.4 16.4 34.0 34.0

52 45 19.2 21.7 20.9 22.4

protons. Table III gives electron-proton separations and values for en for fully and partially intercalated models. The bases are symmetrically disposed about the Cu atom in the fully intercalated model and the guanine and cytosine amino and imino protons on the two sides ofthe Cu atom are at equal separations and angles. There is a slight asymmetry in the partially intercalated model which leads to non-equivalence of the proton positions. The fully intercalated model has two guanine imino protons at 3.7 A from the copper with a en of 16.4 de~rees. The two closest cytosine amino protons are separated from the copper by 4.3 A with a en of34 degrees while the two closest guanine amino protons are 4.5 A from the copper with a en of39.5 degrees. The ENDOR shifts computed from these distances and angles give a poor match to the experimental results for data taken at EPR positions a and b. One predicts that the signal for the guanine imino protons should be shifted from the signal from the cytosine amino protons by 910Khz and the guanine amino protons by 1 Mhz at EPR position a. The predicted shift differences are 280Khz and 460 Khz at position b and then fall within the line width (100Khz) for EPR positions c-e. The signal to noise is relatively poor for spectra taken at EPR positions a and b but we should have been able to detect individual signals with the predicted differences in shifts. Our data, therefore, do not support this mode of binding. The partially intercalated model allows the porphyrin to enter the DNA helix through the rnajor groove and places the copper atom about 1 A from the helix axis. The guanine imino protons on the two sides of the porphyrin have Cu-H separations of3.76 and 3.82A with en values of19.2 and 21.7 degrees. The guanine amino protons are separated from the copper atom by 5.79 and 5.06 A with en values of 45 and 52 degrees. The cytosine amino protons have Cu-H separations of3.80 and 3.84 A with en values of20.9 and 22.4 degrees. The ENDOR shift differences (average= ca 40 Khz) of the guanine imino and cytosine amino protons are within the experimental line width for each of the EPR positions where ENDOR data were taken and this model is consistent with the ENDORdata. The solid and dashed lines in the spectra on the left of Figure 2 indicate the theoretical peak positions of the guanine imino protons located on the DNA. The theoretical positions of the cytosine amino resonances are similar but are omitted for clarity. The resonances closest to the

Porphyrin-DNA Binding Site

849

positions pointed to by these lines are assigned to the guanine imino and cytosine amino protons. The guanine amino protons are far enough from the copper atom that the predicted signals from these protons fall within the linewidth of the center lines in the spectra shown. If one could acquire ENDOR spectra at a field selector angle near 50 degrees, one would observe the maximum shift of the guanine amino protons well separated from the guanine imino and cytosine amino resonances. In this case, one might be able to resolve lines from the guanine amino protons, where there are no other overlapping resonances. Unfortunately, there is no clear location in the EPR powder pattern where selector angles are near 50 degrees and there is no interference from other (copper and nitrogen) spin manifolds. The EPR fields which were chosen for our study allow good angular selection but lead to overlap of proton ENDOR lines from the imino and amino protons. With the partially intercalated model, the guanine imino protons of the base pairs adjacent to the binding site lie 6.9 A above and below the plane of the porphyrin ring. A projection ofthe position of these protons on the plane of the porphyrin ring shows that they are displaced 1 Afrom the copper atom. Chemical shifts induced by porphyrin ring currents for protons at these positions are consistent with the experimentally observed chemical shifts of the imino protons (1). Dynamic processes at room temperature could yield a symmetric environment for these protons on the NMR time scale. The distance from the copper atom to the nonexchangeable protons on the backs of the bases is 6 angstroms or greater. Therefore these protons give small ENDOR shifts (under 400Khz) from the Larmor frequency. Moreover, the protons on the sugar moieties are even further from the copper atom (closer to the phosphorous atoms), and have negligible ENDOR shifts. Assignment of peaks near the center of the ENDOR spectra is difficult and inaccurate. Although these nuclei are still close enough to perturb the spin system of the unpaired electron, overlap of these signals with the broad central line makes exact assignment impossible. The maximum electron-nuclei separation which will produce a measurable shift is determined by the linewidth of the ENDOR signals and the degree of overlap with other signals. Unshifted signals may be observed if nuclei are close enough to the Cu to affect the spin polarization (distant ENDOR ). The distant ENDOR signal from the phosphorous atoms indicates a Cu-P separation of between 7.5-IOA. The partially intercalated model has phosphorous atoms at 9.06, 8.83, 10.2 and 10.7 A from the copper atom. Two positively charged pyridyl nitrogens are 3.8 Afrom the phosphorous atom while the other two pyridyl nitrogens are about 9 A from phosphorous. The fully intercalated model has averaged Cu-P distances of8.6 and 10.2 A The closest phosphorous to pyridyl nitrogen distance is 4.6 Ain this case, while the other two nitrogens are at about 8.1 Afrom the phosphorous atoms. The distant ENDOR signal observed for phosphorous could be accounted for by either of these models. One predicts greater Coulombic stabilization of the complex with partially intercalated TMpyP4 because of the inverse square law distance dependence of this interaction.

850

Greiner eta/.

Figure 6: Drawing of partially intercalated CuTMpyP(4) in DNA viewed from the side of the helix. The line gives the DNA helix axis. The protons and phosphorous nuclear spins which interact with the unpaired electron are shown.

The two models for site binding are nearly identical but subtle difference in proton positions can be determined from the ENDOR data taken at low temperatures. These experiments suggest that CuTMpyP(4) intercalates into DNA and is bound in an almost symmetric environment with the porphyrin ring centered about 1 Afrom the center of the helix. Figure 6 shows a molecular drawing of the side view of this partially intercalated TMpyP(4) between 5' CG3' base pairs. The proposed model is consistent with the ENDOR, EPR, and NMR data but does not exclude other types ofbinding. The ENDORdata would be identical for the case in which the porphyrin is partially intercalated between AT base pairs and GC base pairs were selected only because of previous experimental data which were consistent with intercalation of such highly cationic porphyrins in GC rich areas of DNA (1,2,5,6,8,14). It is also possible for a smaller percentage of the CuTMpyP(4) to be bound in different sites but not detected in the ENDOR spectra. Finally, it is possible that non-Watson Crick base pairing is induced in the base pairs at the binding site. In such a case it is conceivable that only one type of proton is in the correct position with respect to the copper atom to give the observed ENDOR spectrum.

Conclusions ENOORdata has been acquired and analyzed for copper complexes ofTPP, TMpyP(4), and TMpyP(2). These data allowed us to determine Ap. Cu-proton separations, and 90 for the pyrrole protons and the ortho aromatic protons. Complexes ofCuTMpyP(4)

Porphyrin-DNA Binding Site

851

and CuTMpyP(2) with Salmon sperm DNA were also investigated by ENDOR. The complex ofTMpyP(2) shows no additional ENDOR signals on addition of DNA, consistent with non-intercalative binding. In contrast, the complex of TMpyP(4) shows additional proton ENDOR signals which were tentatively assigned to guanine imino and cytosine amino protons. A distant ENDOR signal was also observed from phosphorous atoms in the DNA backbone. These ENDOR data were combined with earlier NMR data to construct a model for porphyrin-DNA binding in which the porphyrin is partially intercalated and extends into the major groove of DNA.

Acknowledgements This work was supported by National Institute of Health grants GM29222 (Emory University) and GM22793 (University ofRochester) We wish to thank D.L. Banville,JA Strickland, and W.D. Wilson for their help and stimulating discussions. Reference and Footnotes 1. Marzilli, L.G., Banville, D.L., Zon, G and Wilson, W.D.,J. Am. Chern. Soc. 108,4188 (1986). 2. Pasternack, RF., Gibbs, EJ. and Villafranca, JJ., Biochemistry 22,2406 (1983). 3. Banville, D.B., Marzilli, L.G. and Wilson, D.W., Biochem. and Biophy. Res. Comm. 113, 148 (1983). 4. Fiel, R.J., Howard, J.C., Mark, E.H. and Datta Gupta, N., Nucleic Acids Res. 6, 3093 (1979). 5. Marc Carvlin, Ph.D. Thesis, SUNY at Buffalo (1983). 6. Banville, D.L., Marzilli, L.G., Strickland, JA and Wilson, W.D., Biopolymers 25, 1837 (1986). 7. Pasternack, R.F., Gibbs, E.S., Gaudemer, A, Bassner, S., Depoy, L, Turner, D.H., Williams, A, Laplace, F., Lamsurd, M., Merienne, C. and Perree-Favet, M., J. Amer. Chern. Soc. 107, 8179 (1985). 8. Fiel, RJ., Carvlin, MJ. and Mark, E.H., Mol. Basis of Cancer. Part B., 215 ( 1985). 9. Carvlin, MJ., Mark, E.H., Fiel, R.J. and Howard, J.C., Nucleic Acid Res. JJ, 6141 (1983). 10. Carvlin, MJ. and Fiel, RJ., Nucleic Acid Res. JJ, 6121 (1983). 11. Carvlin, MJ., Datta-Gupta, Nand Fiel, R.J.,Biochem. Biophys. Res. Comm. 108,66 (1982). 12. Fiel, RJ. and Munson, B.R., Nucleic Acid Res. 8, 181 (1980). 13. Kelly, J.M., Murphy, MJ., McConnel, D.L. and OhUigin, C., Nucleic Acid Res. 13, 167 (1985). 14. Pasternack, R.F., Gibbs, E.S. and Villafranca, J.J., Biochemistry 22, 5409 (1983). 15. Dougherty, G., Pilbrow, J.R., Skorobogaty, A and Smith, T.D., J. Chern. Soc. Faraday Trans. II 81, 1739 (1985). 16. Pasternack, R.F., Sidney, D., Hunt, PA, Snowden, EA and Gibbs E.J., Nucleic Acid Res. 14, 3927 (1986). 17. Blom, N., Odo, J., Nakamoto, K and Strommen, D.P.,J. Phys. Chern. 90,2462 (1980). 18. Sobell, H.M., "Biological Macromolecules and Assemblies: Volume 2, Nucleic Acids and Interactive Proteins", edited by F. Jurnak and A McPherson; John Wiley and Sons, Inc., N.Y. (1984). 19. Hurst, G.C., Henderson, TA and Kreilick, R.W.,J. Am. Chern. Soc. 107,7294 (1985). 20. Henderson, TA, Hurst, G.C. and Kreilick, R.W.,J. Am. Chern. Soc. 107,7299 (1985). 21. Hutchison, C. and Mckay, D.B.,J. Chern. Phys. 66, 33ll (1977). 22. Henderson, TA, Ph.D. Thesis, University of Rochester ( 1985). 23. Schlick, S., Kevan, L., Toriyama, K and Iwasaki, M.,J. Chern. Phys. 74,282 (1981). 24. Pasternack, R.F., Gibbs, E.F. and Villafranca, JJ, Biochem. 22, 2406 (1983). 25. Hurst, G., Kraft, K., Schultz, R. and Kreilick, R.W.,J. Mag. Res. 49, 159 (1982). 26. Greiner, S.P. and Baumgarten, M., J. Mag. Res. 83, 630 ( 1989). 27. Silvers, SJ. and Tulinski, A, J. Am. Chern. Soc. 89, 3331 (1967). 28. Brown, T.G. and Hoffman, B.M., Mol. Phys 39, 1073 (1980). 29. Arnott, S, Dover, S.D. and Wonacott, AJ.,Acta. Cryst., 2192 (1969). 30. Jain T.D. and Sobel H.,J. Bio. Str. Dyn., ll79 (1984).

Date Received: January 6, 1992

Communicated by the Editor R.H. Sarma

Model for the porphyrin-DNA binding site: ENDOR investigations of Cu-porphyrins binding to DNA.

Proton ENDOR has been observed from frozen solutions (ca. 38K degrees) of copper meso-(4-N-tetra-methylpyridyl)porphyrin (CuTMpyP(4)) complexed with S...
3MB Sizes 0 Downloads 0 Views