- Email: [email protected]
Linear dichroism of the complex between the gene 32 protein of bacteriophage T4 and poly(1,N6-ethenoadenylic acid)
J. Mol. Biol. ( 1988) 204. 397-405
Linear Dichroism of the Complex between the Gene 32 Protein of Bacteriophage T4 and Poly( 1,N6-ethenoadenylic acid) H. van Amerongent,
M. E. Kuil, F. van Mourik and R. van Grondelle
Department of Biophysics Physics Laboratory of the Free University de Boelelaan 1081, 1081HV Amsterdam, The Netherlands (Received 22 March
1988, and in revised form
8 July
1988)
We performed linear dichroism measurements in compressed polyacrylamide gels on the complex between the helix-destabilizing protein of bacteriophage T4, GP32 and poly( 1,N6ethenoadenylic acid), which is used as a model system for single-stranded DNA. A strong hyperchromism for poly( 1,N6-ethenoadenylic acid) in the complex indicates a strongly altered conformation. The positive linear dichroism in the wavelength region where the bases absorb must be explained by a strong tilting of the bases in the complex. This finding is in accordance with results from earlier studies, using electric birefringence and circular dichroism measurements. Our measurements show that the angle between the bases and the local helix axis is 42( k6)‘. In addition, a pronounced contribution from the tryptophan residues of GP32 can be recognized, indicating that several of these residues have a specific orientation in the complex. The sign of the dichroism due to the tryptophan residues is the same as that due to the DNA bases. However, it is not sufficient to assume that all the observed dichroism is due to one or more intercalated tryptophan residues and there must be one or more additional tryptophan residues that make an angle of less than 40” with the local helix axis. Some possible structures of the DNA-protein complex are discussed.
1. Introduction
birefringence measurements suggest that the bases in single-stranded DNA with GP32 bound to it are strongly tilted with respect to the long axis of the complex, i.e. they make an angle of much less than 90” with this axis. A variety of experimental results such as the increase of fluorescence emission of poly( 1,N6-ethenoadenylic acid) (poly(~A): Toulme & Helene, 1980), the changes in circular dichroism spectra (Jensen et al., 1976; Anderson & Coleman, 1975) and absorption spectra (Jensen et al., 1976) of single-stranded DNA and synthetic polynucleotides also show that the structure of the polynucleotide is radically changed upon binding the protein, and these changes have often been ascribed to unstacking of the bases. An elaborate study of the DNA structure in complex has been done using circular dichroism and absorption spectroscopy (Scheerhagen et al., 1986a). From the measurements and corresponding model calculations it was concluded that the polynucleotide structure becomes rigid and regular; the bases are clearly t)ilted with respect to the long axis of the complex. If one assumes a helical arrangement of the bases in the complex, then the bases must be shifted towards the helix axis and/or neighbouring bases are less rotated with respect to each other as compared to
The protein GP32$ encoded by gene 32 of bacteriophage T4 is one example of single-stranded DNA binding proteins that are present in a variety of organisms (for a review, see Chase & Williams, 1986). It binds strongly and co-operatively to single-stranded nucleic acids (Alberts & Frey, 1970; Jensen et al., 1976) and protects single-stranded J)NA against nucleases (Curtis & Alberts 1976). It is essential for replication (Riva et al., 1970; Breschkin & Mosig, 1977), recombination (Tomizawa et al., 1966; Berger et al., 1969) and repair processes (Wu & Yeh. 1973). After binding GP32, polynucleotides have a strongly altered conformation, which is probably essential for most of the processes mentioned above. There is an increase in the length of the polynucleotides (single-stranded DNA) of at least 50%, as ascertained by electron microscopic (Jjelius et al., 1972) and hydrodynamic measurements (Scheerhagen et al., 1985a,b). Electric t Author to whom all correspondence should be sent. $ Abbreviations used: GP32, protein encoded by gene C?:!of’ bacteriophage T4; poly(&A), poly(l,N6ethenoadenylic acid); LD, linear dichroism. (K)22~2836/88/2203R7~09
$03.00/O
397
0
1988 Academic
Press Limited
398
H. van Amerongen
bases in the uncomplexed polynucleotide. These structural properties are consistent with the observed increase in length of the polynucleotide in the complex with GP32. In conclusion, substantial evidence exists that the position and orientation of the bases are altered when GP32 binds to the polynucleotide. Tn the complex there are close contacts between aromatic residues of the protein and the bases of the polynucleotide. It has been proposed that amino acid residues in the region of residue 72 to I16 (Williams et al., 1981) play an important role in the binding of the protein to single-stranded DNA. This region contains six tyrosine and two tryptophan residues. In agreement with results obtained by Anderson & Coleman (1975). nuclear magnetic resonance measurements show (Prigodich et al., 1984, 1986) that five tyrosine residues are in close contact with the bases; this is also the case for two phenylalanine residues (Prigodich et al., 1986). A role for at least one tryptophan in the binding process has been suggested (Toulme et al., 1984; CasasFinet et al., 1984), which probably corresponds to the finding that at least one tryptophan residue is close to the bases (Toulme & Helene, 1980; Prigodich et al., 1984; Le Doan et al., 1984; Khamis & Maki, 1986). In a recent study (Casas-Finet et al., 1988), it was concluded that at most two tryptophan residues per protein molecule are in close contact with the bases. Studies on model peptides have shown that intercalation does occur with the bases in single-stranded DNA but not in double-stranded DNA (Brun et al., 1975; Toulme & Helene, 1977; Mayer et al., 1979). Stacking interactions between the aromatic residues of the protein and the bases could possibly explain the fact that GP32 binds much more strongly to single-stranded DNA than to double-stranded DNA. From hydrodynamic measurements (Scheerhagen et al., 1985a,b), it appears that the hydrodynamic volume is larger than the volume that can be calculated using the mass of the complex and the partial specific volume. The difference can be explained by assuming an open solvated structure, possibly a superhelix (Scheerhagen et aZ., 1986a, 1985b). To obtain more direct information about the orientation of the bases and aromatic residues, especially tryptophan, in the complex, we performed linear dichroism (LD) measurements on the complex of GP32 and poly(sA) oriented in compressed polyacrylamide gels. This orientation method was first described by Abdourakhmanov et al. (1979). We used this method in a slightly modified way (van Amerongen et al., 1988), compressing a gel containing the complex in one direction (x) and allowing it to expand in one perpendicular direction (y). Rod-like molecules orient preferentially with their long axis along the y-axis. The measuring light is now incident along the z-axis and the difference in absorption, AA, for light polarized along the y-axis and the x-axis is measured. The measured amount of LD gives
et al.
information about the orientation of the bases and aromatic amino acid residues (see below). For the polynucleotide, we have chosen poly(sA) because it has a significant absorption band above 300 nm, well separated from the absorption bands due to the protein, which facilitates the interpretation of the LD spectra. Poll is thought to bind to GP32 in a manner similar to that of the polynucleotides, according to parameters that govern the binding process (Kowalczykowski et al., 1981; Toulme & Helene, 1980).
(a) Interpretation
qf LD
results
For rigid rod-like particles oriented in a polyacrylamide gel, which is compressed with a degree of compressiondiv;eiwidth of the gel before compression width after bY the compression) the following relation holds: cos2cr-1)
0(n),
(1)
where A is the absorption at a particular wavelength 1 when the gel is uncompressed; AA is the amount of dichroism, which is defined as the absorption along the y-axis of the gel (direction of expansion) minus the absorption along the x-axis (direction of compression); a is the angle between the transition dipole moment responsible for the absorption at the wavelength L and the orientation axis of the particle. The angle brackets ( .) denote an average for all transition moments. In the following we will omit these brackets for simplicity. When different transition moments absorb at 2, a weighted average has to be taken. Q(n) is the orientation function that describes how well the particles are oriented. It is 1 for perfect orientation and zero when there is no compression. Tn van Amerongen et al. (1988) Q(n) is given. Here we only give a useful approximation for low values of 72:
D(n)= AW ; c, L.l
(2)
which is correct within a small percentage below n = 1.7. For DNA and RNA and their complexes with the protein it is interesting to relate the amount of LD to the orientation of the bases or aromatic amino acid residues with respect to the orientation axis of the particle. In this case, cos2 c1 has to be replaced by cos2 fl x sin2 6 for transition moments lying in the plane of the bases or aromatic residues (Norden, 1978). The angle fl is defined as the angle between the orientation axis and the plane through the bases or the aromatic residues. The line about which the planes are tilted with respect to the orientation axis is called the inclination axis, and 6 is the angle between the transition moment and the inclination axis. When such a structure is organized as a (local) helix that is further organized into a superhelix where the local axis (e.g. helix axis) makes an angle y with the
Linear Dichroism of a DNA-Protein superhelix axis, the right-hand side of equation (1) has to be multiplied by $(3 cos2 y- 1) (NordBn, 1978). For elongated molecules with some flexibility the amount of LD will decrease when compared to a rigid molecule. This will probably depend on the length of the molecule (L) and its flexibility, which is related to its persistence length (P). This effect is accounted for in a formal way by multiplying the orientation function for a rigid rod by another factor Q(P, L)(O I @(P, L) I 1). This leads to the following relation :
3A(4
3A =+(3cos2pxsin26-1)
x @(n)O(P.L)+(3 In the following term @(P,L)+(3
notation, cos2 y-l).
(30s~y- 1).
(3)
Y will denote the product
Complex
399
remains constant. Light passes perpendicular to the z and y-direction in the z-direction through the press. The pathlength is only 5 mm to minimize the absorption of the gel. Absorption spectra were measured on a Gary 219 spectrophotometer interfaced to a HP85 computer. Spectra were baseline corrected. Accurate absorption spectra in gel were difficult to obtain because of a large contribution of the gel. Within experimental error, the spectrum of t’he DNA-protein complex in the gel phase was identical with that in solution after correct.ion for the gel contribution. However, for reasons of accuracy, spectra measured in solution and corrected for dilution upon going to the gel phase were used for the calculation of the reduced dichroism. LD spectra were measured on a modified (Bokma et al., 1987) Cary 61 spectropolarimeter, also interfaced to a HP85 computer. The crystal was modulated such as to give maximum LD signals. Calibration was done as described (van Amerongen et al., 1988). Absorption and LD spectra were further analysed on a VME 131 computer.
2. Materials and Methods Unless stated otherwise, all measurements were done in the following buffer: 50 mM-NaCl, 1 mM-Na,HPO,, 0.1 mM-EDTA (pH 7.7) (buffer I). The protein GP32 was isolated as described (Scheerhagen et al., 1986a). The used: extinction coefficient was following &,,,=37x 103~-' cm-’ (J ensen et al., 1976). Poly(&A) was from PL Biochemicals (Woerden, The Netherlands) and was used without further purification. The extinction coefficient (per base). Ebb, = 3420 M-’ cm-‘, was obtained by comparing the absorption in buffer I with the absorption in the buffer used by Toulme & HBl&ne (1980), where an extinction coefficient of 3550 Mm 1 cm- ’ has been given. This latter value was taken from (Ledneva et al., 1978). where a value of 0.56 is found for the absorption ratio A(300 nm)/A(257 nm) (A is absorption), which is close to the value of 0.57 that we cm-’ has been find. Although a value of 5110 Me1 reported for this polynucleotide (Janik et al., 1973) and was used by Kowalczykowski et al. (1981) in GP32 binding studies. we have chosen to use the former value in view of the fact that the corresponding A(300 nm)/ ,4(260 nm) ratio of 0.374 (Janik et al., 1973) is much lower than in our case. I,-Trp was from Fluka AC Buchs SG and was dissolved in distilled water. Polyacrylamide gels were made using buffer I. and the final gels contained 145(/, (w/v) acrylamide from Serva (Heidelberg, West Germany). 0.550/, (w/v) N,N-methylenebisacrylamide (BTS), 0.075 “& N,N,N’,N-tetramethylparaphenylenediamine (TEMED) both from Merck-Schuchardt (Hohenbrunn, West Germany) and 0.1% (w/v) ammonium persulphate (APS) from *J. T. Baker Chemicals 1~.V. (Deventer, The Netherlands). All reagents To 750 ~1 of a were used without further purification. solution containing the complex of poly(~A) and GP32 (see also Results). an equal volume of buffer I containing proper amounts of acrylamide, BIS and TEMED was added. After gentle mixing of the solution, 15 ~1 of a 10% (w/v) APS solution was added and this solution was poured into a cast with the dimensions of the gel press. After about 1 h the polymerization was complete and the gel was placed into the press. This gel could be compressed continuously in one direction (2) from an initial width of 9 mm to a final width dfabout 5.5 mm. In this press the gel is allowed to expand in one perpendicular direction (y). where the volume of the gel
3. Results Complex formation between poly(&A) and GP32 could be followed by measuring the absorption spectra from 250 to 400 nm during titration, starting with poly(eA) adding GP32. In contrast to what was found in other measurements (Toulmt? & HBlBne, 1980), we observed a hyperchromism above 250 nm for the polynucleotide upon complex formation. The shape of the absorption difference spectrum (spectrum of the complex minus the spectrum of poly(&A) and that of GP32 measured in the unbound form) is almost identical with the induced difference in absorption, obtained when a poly(&A) solution is heated without the protein (not shown). No significant change in the protein absorption can be observed. A similar conclusion using other polynucleotides and the same or other single-stranded DNA-binding proteins was drawn elsewhere (Scheerhagen et al., 1986a). Therefore, we conclude that the absorption spectrum of GP32 remains unchanged in the complex and the absorption spectrum of poly(&A) in complex can be obtained by subtracting the absorption spectrum of free GP32. The resulting spectrum of bound poly(~A) is given in Figure 1, together with the spectrum of free poly(&A). In a saturated complex the amount of hyperchromism is about 36% in the peak at 267 nm at 12*6”C, which is even larger than the 24% obtained from heating to about 80°C. During tifrafion, the absorption in all the peaks increased almost linearly with the amount of GP32 added (data not shown) until saturation was reached. The titration was stopped when there was an excess of at least 40% of GP32 beyond the point of saturation. If the amount of hyperchromism is indicative of the amount of binding, we can determine (data not shown) the size of the binding site (number of nucleotides covered by one protein) by plotting the hyperchromism versu~~ the protein to nucleotide ratio in the same way as was done by van et al. (1987). Amerongen The value
H. van Amerongen
400
et al.
polynucleotides and single-stranded DNA (Bobst et al., 1982; Scheerhagen et al., 19866). Therefore, we shall use s = 10 for further calculations. However, we shall show that taking s = 8 does not significantly change the interpretations results. The saturated complexes described
of the I,D above
were
used to prepare gels as described in Materials and Methods. After compression of a gel, a clear LD spect’rum could be measured, as shown in Figure 2(a). Below 250 nm no measurements are 250
275
300
325
Wavelength
possible because of a large amount of scattering and strong absorption in this region. In addition t’o the spectral features due to the protein and the
350
(nm)
polynucleotide,
Figure 1. Absorption spectrum of poly(&A) free ( - ) and in complex with GP32 (---) at 12.6”C in buffer I. Corrections have been made for the protein contribution and dilution.
s = lO.Of
1.6 is an average
of four
titrations
with
the standard error reflecting the scatter in different experiments. Our result is clearly different from s = 5fl (Toulm6 & H618ne, 1980) and s = S&l reported by Kowalczykowski et al. (1981). But when the extinction coefficient used in our work is applied to the latter results, s = 9 + 1.5 is obtained, which is in reasonable agreement (see also Materials and Methods). The lower values obtained in fluorescence
titrations (Toulmk & H&l&e, 1980; Kowalczykowski et al., 1981) are an underestimate
due to difficulties associated with the specific conditions of these experiments (for example, see Bobst et aZ., 1982). On the other hand, s = 10 seems in good agreement with values obtained in recent studies for complexes of GP32 with different
contribution
the measured LD spect’ra show a from
the gel. An example
of an I,I)
spectrum for a blank gel is also shown in Figure 2(a) and is characterized by a monotonously decreasing dichroism, going where a minimum
from
400 nm
to about
252 nm,
is found. The LD of all gels seems to rise below this wavelength, but this is probably because very little light passes through the gel below this wavelength. The intensity of the whole spectrum increases when the gel is compressed further. The shape of the spectrum of a blank gel was found to be close to constant in different measurements, but the intensity changed somewhat from one measurement to another. Also the exact place of the slightly. When
minimum around 252 nm varied measurement,s were performed, the
spectra were corrected for the contribution of the gel in the following way: between 350 and 400 nm the spectrum was fitted with a combination of a straight horizontal line (accounting for a small drift of the apparatus) and a spectrum of a blank gel using a linear least-squares method. The contributions determined
in this manner
in this wavelength
region were used to correct the whole spectrum. The
1 3.5
x lo-’
-3.5
x 1o-3
-7.0
x lo-’
1 300
250
1 350
Wavelength
400
250
(nm)
300 Wavelength
350 (nm)
400
(b)
(a)
Figure 2. (a) (-) LD spectrum of poly(&A) +GP32 complex in a compressed polyacrylamide gel (n = 1.53). (---) LD spectrum of a blank gel compressed with a factor n = 1.53. (. .) LD spectrum of the same complex corrected for the baseline as described in the text (see Results). [poly(&A)] = 32.6 PM, [GP32] = 4.8 ,UM, t = 8°C. (b) (-) LD spectrum of poly(&A) +GP32 complex in a compressed polyacrylamide gel (n = 1.53) [poly(&A)] = 32.6 PM, [GP32] = 4.8 PM, t = 8°C. (---) LD spectrum
of poly(eA)
(n = 1.53) corrected
to 32.6 PM and multiplied
by a factor
of 3.75 (t = 11 T).
Linear
Dichroism
of a DNA-Protein
shape of the corrected LD spectra of the complex was almost the same at all degrees of compression (spectra not shown). For different gels the spectra varied somewhat in height, especially around 250 nm, where uncertainties due to the baseline corrections are largest. An LD spectrum of the complex corrected as described above is also given in Figure 2(a) with n = 1.53 at t = 8°C. The spectrum did not vary when raising the temperature to 25”C, reminiscent of results obtained with absorption and circular dichroism spectra of other polynucleotides in complex with GP32 (Scheerhagen et al., 1986a). The spectrum is positive and is markedly different from the spectra of GP32 and poly(&A) alone. The spectrum of free GP32 (not shown) is very small in comparison with the complex spectrum, and its contribution to the LD spectra of the gels containing the complexes can safely be neglected. For poly(.zA) (see Fig. 2(b)) the spectrum is negative everywhere, although there may be a contribution from a positive band around 290 nm. In the latter spectrum, the negative dichroism is in agreement with the expectation that the bases are oriented more or less perpendicular to the orientation axis of the polynucleotide with the most important transition moments, dipole corresponding to n--71* transitions, lying in the plane of the bases. The small amplitude of the LD is probably due to mobility of the bases and high flexibility of the sugar phosphate backbone. Above 300 nm the LD of the complex is positive, indicating that the bases are significantly tilted with respect t,o the orientation axis. The fine structure around 280 nm in the LD spectrum is completely different from the fine structure in the absorption spectrum of poly(~A), where the peaks are at different wavelengths. Therefore, the LD around 280 nm is probably due to the protein. In this wavelength region the protein absorption stems predominantly from the tryptophan residues. The peaks in the fine structure coincide remarkably well with those in the absorption spectrum of L-Trp measured in water (see Fig. 3). However, the clear fine structure that can be observed in the free tryptophan absorption spectrum and in the LD spectrum of the complex is present neither in the absorption spectrum of the protein (Fig. 3) nor in that of the complex. Therefore, we propose that it must be attributed to spectrally distinct and specifically oriented tryptophan residues in the GPSB-poly(~A) complex (see Discussion). A similar fine structure is also observed in the LD spectrum reported by Takahashi et al. (1987) for RecA protein from E. coli in complex with single-stranded DNA. The fine structure present in the absorption spectrum of poly(~A) in the region 255 to 280 nm cannot be discerned in the LD spectrum of the complex, indicating that the absorption in this region does not contribute significantly to the LD. The positive shoulder around 270 nm in the complex is also present in complexes with other polynucleotides (van Amerongen & van Grondelle,
250
Complex
275
401
300 Wavelength
325
350
(nm)
Figure 3. (-) LD spectrum of poly(&A)+GP32 complex in a compressed polyacrylamide gel (n = 1.53) [poly(cA)] = 32.6 PM, [GP32] = 4.8 p~.(. .) Absorption spectrum of L-Trp measured in water (20°C). (---) Absorption spectrum of GP32 in buffer I (20°C).
unpublished results) and we do not assign it to poly(&A). The observed LD spectrum of the complex can reasonably be assigned to a superposition of only the long wavelength band around 307 nm due to poly(&A) and contributions from the tryptophan residues. It would not be very realistic to try a detailed fit for the whole spectrum, in view of the large number of different aromatic groups in the complex (bases, tyrosine and tryptophan residues). Moreover, even in a single residue more transitions can contribute to the absorption in this wavelength region, e.g. in tryptophan, in which the absorption around 280 nm is due to two almost degenerate transition moments with different orientations. To interpret the spectra in a more quantitative way we use the fact that the LD at 315 nm is exclusively due to poly(.zA) and that the positive LD at 280 nm can be ascribed predominantly to tryptophan residues. The contribution from the long wavelength band of poly(~A) to the LD at 280 nm is estimated to be about 5%, assuming a Gaussian shaped band, centred at 307 nm. In Figure 4 the amount of dichroism at 280 and 315 nm is plotted versus n to show the orientational behaviour of the complexes in gel. The drawn lines represent the orientation function @(n) for a rigid rod multiplied by a constant to give a reasonable fit with the experimental data. In all measurements the amount of dichroism increases somewhat more slowly than would be expected in the case of a rigid rod. This may be due to some uncertainties in the baseline corrections, but it may also indicate that @(L/P) is dependent on n. However, as a first approximation we assume that @(L/P) is constant at all values of n. This will not have much impact on our final conclusions. Averaging A.A/(A@(n)) for different values of n, we have at least a lower limit for AA/(A@(n)O(L/P)) (i.e. taking @(L/P) = 1).
402
H. van Amerongen
O-060
7 G Q
0.045
0.030
0.015
C I.0
I.2
I.4
I.6
”
Figure 4. (A) AA(280 nm) of poly(~A)+GP32 complex at different values of n divided by A of the protein in the complex (see the text). The fit represents the curve 0.183 m(n). (0) AA(315 nm) of the same complex divided by A(315 nm) of the complex. The fit represents the curve 0.143 Q(n).
The absorption at 315 nm is exclusively due to poly(aA) and at this wavelength we find that AA/(A@(n)) = 0*15+0*02. The error is the standard error in the results of four different series of measurements. Variations in one series of measurements due to deviation from the rigid rod curve have not been taken into account. The scatter in the results will be due partly to some uncertainty in the baseline corrections. The uncertainty in AA due to some uncertainty in the calibration of the LD apparatus is about 3%, whereas the values of n are accurate within a small percentage. The error in the value of A is small. At 280 nm almost all of the LD can be ascribed to the protein. Subtracting 5% of the signal (estimated contribution from the long wavelength band of poly(aA)), we obtain AA/(A@(n)) = 0*18&0*02. In this case, A is the absorption of the protein in the complex, which has been calculated using a value of 10 for the size of the binding site, and using the fact that the protein absorption at this wavelength does not change upon binding. Taking s = 8 results in AA/A@(n) = 0.14f0.02. Note that the choice of the value of s does not influence the value AA/A for the poly(aA) bases.
4. Discussion Complex formation between poly(&A) and GP32 leads to an induced hyperchromism for poly(eA), indicating that the structure of poly(aA) is altered in the complex. Hyperchromism is also induced for other polynucleotides (poly(dA) and poly(rA)) and single-stranded DNA (Newport et aZ., 1981; Scheerhagen et al., 1986a) upon binding of GP32. From the increased poly(&A) fluorescence upon binding of GP32 Toulme & Helene (1980) arrived at a similar conclusion, but they did not observe the hyperchromism. The authors used a different buffer and probably performed their measurements at a
et al.
higher temperature, where the free poly(aA) is less hypochromic. But this can only explain less than half of the difference and we do not have an explanation for the remaining discrepancy. The changes in the absorption spectrum after complex formation are very similar to the changes obtained by heating solutions of poly(aA). However, t’he increase in absorption induced by binding is even higher than that upon heating to about 8O”C, an arbitrary high temperature. The size of the binding site we obtained is lO*O( + 1.6) nucleotides. It is not the intention of this work to determine the value of 8, but we performed the titrations to be sure that LD measurements were performed with a saturated complex. In fact, the value of s does not influence the value of AA/A for the bases. However, it is important for the determination of the absorption of the protein in the complex. For our calculations we have chosen s = 10, in agreement, with our findings. In view of the contradictory values reported for s (see Results) we show that with s = 8 the same essential conclusions can be drawn concerning the orientation of the tryptophan residues in the complex. (a) Qualitative
interpretation of the LD spectrum of the complex
In gels, the complexes do not dissociate, as can be concluded from the fact that the LD spectrum of the complex is totally different from the spectra of GP32 and poly(&A) alone. The LD spectra have to be corrected for the contribution of the gel. This can lead to some uncertainty in the spectra, which is probably reflected in the scatter of the results obtained in different experiments. However, the main characteristics of the spectra are very reproducible and the accuracy in the amount of LD is reasonable. A striking result is the fact that the LD signal above 300 nm is positive in the complex, where it is negative for poly(&A) alone. This shows t,hat the orientation of the bases must change significantly when the complex is formed. The predominant transition moments in these planar aromatic molecules are due to X-Z* transitions and are polarized in the plane of the bases. This is in agreement with the negative LD for free poly(eA), where the bases are more or less perpendicular to the long axis of the polynucleotide (with the possible exception of an out-of-plane transition moment around 290 nm). This means that the positive LD above 300 nm must be explained by a strong tilting of the bases (b < 54.7”). This strong tilting can explain why the negative birefringence measured for single-stranded DNA (Scheerhagen et al., 1985a) reverses sign in the complex with GP32 and, in addition, is in agreement with the CD calculations done for poly(rA) and for poly(dA) in complex with GP32 (Scheerhagen et al., 1986a). A second remarkable characteristic of the LD spectrum is the fine structure around 280 nm and, as indicated in Results, this feature must be ascribed to the protein. The major contribution to
Linear Dichroism of a DNA-Protein the absorption around this wavelength stems from the indole groups of the tryptophan residues. In an indole group two transition moments, L, and L,, contribute to the absorption near 280 nm. These moments are nearly perpendicular to each other (Yamamoto & Tanaka, 1972). Different shapes and relative intensities have been proposed for both bands (Yamamoto & Tanaka, 1972; Valeur & Weber, 1977), and the contribution of the L, band to the intensity at 280 nm varies in both studies from about 20 to 35%. The L, band is held to be responsible for most of the fine structure. As reported by Sun & Song (1977), the absorption characteristics of an indole group can change markedly when different solvents are used. In the protein the different indole groups are probably not all in the same environment, resulting in a smooth absorption spectrum GP32, both in the free and in the complexed form, which does not have such definitive fine structure as Trp. Also wavelengthselected optically detected magnetic resonance measurements reveal the existence of environmentally distinct and spectrally different types of Trp in GP32. One of these types represents tryptophan residues in close contact with the bases (Khamis & Maki, 1986). As was described in Results, the absorption difference spectrum due to complex formation shows no changes attributable to the protein; therefore, one would expect also a relatively smooth spectral contribution of the trypophan residues to the LD, if all tryptophans were oriented in the same way with respect to the long complex axis. However, in the LD spectrum there is a clear fine structure, indicating that not all indole groups have the same orientation with respect to the long axis of the complex. At this moment it is difficult to give a more detailed interpretation of the tine structure in the LD spectra around 280 nm, because of the many different transition moments that contribute, but, nevertheless, from the positive sign of the LD we can conclude that the average angle between the indole rings and the long axis of the complex is certainly smaller than 54.7”. (b) Estimation of the value of the parameter Y In the following, $&
we will use the values: (315 nm) = 0.15
Complex
403
superhelix, the term Y must be 1, which is an absolute upper limit for Y. The complex is possibly folded into a superhelix (Scheerhagen et al., 1986a, 1985b) and a value of 40” has been proposed for y (Scheerhagen, 1986). This would lead to Y < 0.4. In addition, flexibility leads to a further decrease of the value for Y. A rough estimate of the contour length of the complex can be made, using the fact that the polynucleotide contains on the average about 500 bases. From a rise per base in the complex of 0.44 nm (Scheerhagen et al., 1986a) a lower limit for the contour length of about 220 nm is found. This is much larger than the persistence length P, which is certainly lower than 50 nm and probably is about 30 nm, as can be concluded from light-scattering experiments (M. E. Kuil et al., unpublished results). Therefore, the rigid rod approximation is not valid, which will further lower the term Y. On the other hand, we can estimate a lower bound for Y, assuming that all indole rings are oriented with respect to the local helix axis in such a way as to give a maximum amount of LD at 280 nm. This is the case when b = 0 for all indole groups (P(indole)), 6 = 90” for the L, transition moment (&LO)) and 6 = 0” for the L, transition moment (6(L,)). We also assume that 20% of the absorption at 280 nm is due to the L, band. A higher percentage leads to a larger value of the lower bound of Y. This chosen conformation leads to AA/A@(n) = 2.1 at 280 nm when Y = 1. To explain the measured value of 0.18, Y must therefore be at least 0.09. This is an absolute lower bound for Y. However, in this extreme case one would not observe such distinct fine structure in the LD spectrum. To obtain a significant fine structure, at least one but probably more Trp residues cannot contribute significantly to the positive LD in the way described above. When only four Trp residues contribute to the positive LD, then AA/A@(n) would maximally be about 1.7 at, 280 nm, corresponding to a minimum value for Y of 0.11, which is still a rather extreme value in view of the extreme orientations for the other four Trp residues. So the real value of Y will be somewhere between the extreme values of 0.11 and 1.OObut in view of the foregoing discussion it seems reasonable to assume that its real value is somewhere near 0.25, corresponding to a superhelix with y = 35” and a value for @(L, P) of 0.5, for example.
and L+!(280 nm) = 0.18 A@(n) to estimate average /? values of the bases and indole groups with respect to the assumed local helix axis, using the relation: AA ~=3(3cos2psinZ6-l)Y. A@(n) An initial estimate of Y is needed to determine the product term cos’ j? sin’ 6. When the complex behaves as a rigid rod, which is not folded into a
(c) Orientation of the bases and tryptophan with respect to the local helix axis
residues
The bases of the polynucleotide are probably organized in a regular structure (Scheerhagen et al., 1986a) and all contribute approximately in the same way to the total amount of LD. In Table 1 we give the average angles between the bases and the local helix-axis when different values for 6,,, are assumed with Y = O-25. It will be shown below that, within certain limits, other values of Y lead to similar values of 8. The same is done for the indole
H. van Amerongen
404
Table 1 Angles of bases and indole groups with the local helix axis B(indole) 6 3,5 9o” 70‘ 50”
/+A)
6(L,)
m= 5
m= 4
m = 3
m= 2
m= 1
47” 43” 27”
90” 70” 50”
38” 35” 19”
35” 31” 11”
30” 25” NP
17” NP NP
NP NP NP
Average values for the angle /I between the bases/indole groups and the assumed local helix axis, using Y = 0.25. Average values of /?(indole) have been calculated, assuming that m indole groups contribute to the positive LD. It is assumed that the L, band leads to 80% of the Trp absorption at 280 nm. A lower percentage will lead to lower average angles &indole). NP indicates that the corresponding conformation is not possible.
groups with different values for 6(&J, again assuming that L, is responsible for 80% of the absorption at 280 nm. A lower percentage leads to values of fi closer to 0”. We consider different cases where a number of indole groups, m, contribute significantly to the positive LD and then calculate the average orientation of these indole rings. In some cases the observed amount of LD cannot be explained with the chosen values for m, 6 and Y: because the term cos’ B would be larger than 1 (eqn (3)). It is directly clear from Table 1, that more than one Trp contributes to the positive LD and even two residues seems highly unlikely. Also the angle between the bases and the local helix axis must be lower than 48”. To be more specific we will use the fact that no fine structure below 300 nm is observed that can be ascribed to poly(~A). Therefore, the corresponding transition moments around 270 nm do not contribute significantly to the LD. If we call the angle of the corresponding transition moment with the inclination axis 6,,, and assume that this transition moment is also oriented in the plane of the base, then 6 2,0 = &,, rf: x and x denotes the angle between both transition moments of poly(~A). The value of x is about 55”, which is deduced from fluorescence depolarization measurements of the free nucleotides in high concentrations of glycerol (data not shown). Because there seems to be no dichroism due to poly(&A) around 270 nm, we have the following relation for /?(&A) and ha15: 3 cos’ /I sin’
(6,,,
+x)-
1 = 0.
Together wit*h the known amount of dichroism at 315 nm we can calculate I, 8315 and 6,,, for a chosen value of Y. For Y = 0.25, these values are, for respectively, 44”, 72“ and 127”. Assuming instance that the complex is completely rigid and that there is no superhelix ( Y = l), we calculate the values 48”, 75” and 130”. However, if we take the minimum value of 0.11 for Y. we obtain the values 36”, 80” and 135”. So, even assuming rather extreme values for Y does not lead to large variations in the calculated
et al.
angle between the bases and the local helix axis. The safest estimation for this angle is 42( +6)‘, but most probably it is somewhere between 40” and 45”. Because we do not know the orientation of the transition moments in the molecular plane, we are not able to determine the approximate orientation of the inclination axis and, therefore, we cannot determine the values of twist and tilt of the bases. Tt is argued by Casas-Finet et al. (1988) that at least one and at most two tryptophan residues are in close contact with the bases. If one Trp molecule would intercalate, as is mentioned as a possibility in Introduction, it would make about the same angle with the local axis as the bases. If Y = 0.25 and p = 40”, such a single intercalating Trp molecule can (if 6( L,) = 90”) explain at most 17 o/o of the observed dichroism at 280 nm and two tryptophan residues at most 34%. So, there must be other nonintercalating tryptophan residues that contribute to the positive dichroism, and the indole groups of these Trp residues make, on average, a smaller angle with the local helix axis as compared to the assumed intercalated tryptophan(s). Note that any other choice of 6(L,), either for the possibly intercalated or for the remaining residues, tryptophan residues will lead to only more extreme values of p (i.e. closer to 0”) for the non-intercalated groups. A choice of s = 8 for the binding site size will reduce AA/A in the protein band, but an intercalated tryptophan residue explains maximally 21 y. of the observed dichroism and therefore the conclusions are not significantly altered. The main conclusions of this work are as follows: (I ) in the GP32-single-stranded complex the polynucleotide assumes a non-random specific, conformation characterized by a strong base tilt, with B = 42( +6)“, but most probably between 40” and 45”. (2) Probably, at least more than one tryptophan residue contributes in a specific way to the LD. Part of the protein LD may be due to one or more intercalated Trp residues, but this cannot explain all the observed dichroism and the indole of the non-intercalating tryptophan groups residues make an angle with the complex axis that is almost certainly smaller than t’hat of possibly intercalated residues. We are performing similar measurements with GP32 complexed to polynucleotides containing other bases, in both their riho and their deoxyribo forms (van Amerongen & Van Grondelle unpublished results). Although the spectra show more overlap with that of the protein, an advantage of these polynucleotides is that the directions of the various transition moments are available. Using poly(~A) as a starting point, we can be more specific about the DNA conformation in terms of base twist and tilt and hopefully find further details concerning the orientation of the tryptophan residues. GP32 was skilfully isolated by J. den Blaauwen. This work was supported by the Netherlands Foundation for Pure Research via the Dutch Foundation of Biophysics.
Linear
Dichroism
of a DNA-Protein
References Abdourakhmanov, I. A., Ganago, A. O., Erokhin, Y. E., Solov’ev, A. A. & Chugunov, V. A. (1979). Biochim. Biophys. Acta, 546, 183-186. Alberts, R. M. & Frey. L. (1970).Nature (London), 227, 1313-1328. Anderson, R. A. & Coleman, ,I. E. (1975). Biochemistry, 14, 5485-5491. Berger, H., Warren. A. ,J., Fry, K. E. (1969). J. Virol. 3, 171-175. Robst, A. M.. Langemeier, P. W., Warwick-Koochaki, P. E.. Hobst, E. V. & Ireland, J. C. (1982). J. Biol. Chem. 257, 6148-6193. Rokma. tJ. T.; ,Johnson, W. C. Jr & Blok, J. (1987). Biopolymers,
26, 893-909.
Breschkin, A. M. & Mosig, G. (1977). J. Mol. Biol. 112, 2955308. Krun. F., Toulm&. J-,I. & HPlBne, C. (1975). Biochemistry, 14, 558-563. (‘asas-Finet, J. R.. ToulmC, ,J.-J., Cazenave, C. & Santus. R. (1984). Biochemistry, 23, 1208-1213. Casas-Finet. J. R., ToulmB, J.-J., Santus, R. & Maki. A. H. (1988). Eur. J. Biochem. 172, 64-646. Chase, ,J. W. & Williams, K. R. (1986). Annu. Rev. Biochrm. 55, 103-136. (‘urtis. M. *I. & Albrrts, B. (1976). J. Mol. Biol. 102, 793% 816. Delius. H.. Mantell. X’. ,J. & Alberts, B. (1972). J. Mol. Biol. 67. 341-350.
Janik, B.. Sommer. R. G., Kotick, M. P., Wilson, D. P. & Chem. Phys. 5, Erickson. R.
Complex
405
Newport, J. W., Lonberg, N., Kowalczykowski, S. C. & von Hippel, P. H. (1981). J. Mol. Biol. 145, 105-121. Eordbn, B. (1978). Appl. Spectrosc. Rev. 14(2), 157-248. Prigodich, R. V., Casas-Finet, J., Williams. K. R., Konigsberg, W. & Coleman, CT.E. (1984). Biochemistry,
23, 522-529.
Prigodich, R’. V., Shamoo, Y., Williams. K. R.. Chase, J. W., Konigsberg, W. H. & Coleman, J. E. (1986). Biochemistry,
25, 3666-3672.
Riva, S., Cascino, A. & Geiduschek, E. P. (1970). J. Mol. Biol. 54. 85-102. Scheerhagen, M. A. (1986). Doctoral Thesis, Vrije Universiteit. Scheerhagen. M. A., van Amerongen, H., van Grondelle, R. & Blok, tJ. (1985a). FEBS Letters, 179, 221-224. Scheerhagen. M. A., Kuil, M. E., van Grondelle, R. & Blok, ,J. (1985b). FEBS Letters, 184, 221-225. Scheerhagen. M. A., Bokma, J. T., Vlaanderen. C. A.. Blok, J. & van Grondelle, R. (1986a) Hiopolymers, 25, 1419-1448. Scheerhagen. M. A., Vlaanderen, C. A., Blok, ,J. & van Grondelle, R. (19866) J. Hiomol. Struct. Dyn. 3, 887898.
Sun, M. & Song, P.-S. (1977). PhotochewL. Photobiol.
25.
3-9.
Takahashi,
M., Kubista,
M. C NordBn. B. (1987). .I. Biol.
Chem. 292, 8109-8111.
Tomizawa, J. I.. Anraku, N. & Twama. Y. (1966). J. Mol. Biol. 21. 247-253. ToulmP, ?J.-J. & H&ne, C. (1977). J. Biol. (‘hem. 252, 2446249.
Toulmk.
J.-J.
& HBlBne, CZ. (1980). Biochim.
Riophys.
Acta, 606, 95-104.
Toulm6. J-
Edited by P. H. won Hippel
Linear Dichroism of the Complex between the Gene 32 Protein of Bacteriophage T4 and Poly( 1,N6-ethenoadenylic acid) H. van Amerongent,
M. E. Kuil, F. van Mourik and R. van Grondelle
Department of Biophysics Physics Laboratory of the Free University de Boelelaan 1081, 1081HV Amsterdam, The Netherlands (Received 22 March
1988, and in revised form
8 July
1988)
We performed linear dichroism measurements in compressed polyacrylamide gels on the complex between the helix-destabilizing protein of bacteriophage T4, GP32 and poly( 1,N6ethenoadenylic acid), which is used as a model system for single-stranded DNA. A strong hyperchromism for poly( 1,N6-ethenoadenylic acid) in the complex indicates a strongly altered conformation. The positive linear dichroism in the wavelength region where the bases absorb must be explained by a strong tilting of the bases in the complex. This finding is in accordance with results from earlier studies, using electric birefringence and circular dichroism measurements. Our measurements show that the angle between the bases and the local helix axis is 42( k6)‘. In addition, a pronounced contribution from the tryptophan residues of GP32 can be recognized, indicating that several of these residues have a specific orientation in the complex. The sign of the dichroism due to the tryptophan residues is the same as that due to the DNA bases. However, it is not sufficient to assume that all the observed dichroism is due to one or more intercalated tryptophan residues and there must be one or more additional tryptophan residues that make an angle of less than 40” with the local helix axis. Some possible structures of the DNA-protein complex are discussed.
1. Introduction
birefringence measurements suggest that the bases in single-stranded DNA with GP32 bound to it are strongly tilted with respect to the long axis of the complex, i.e. they make an angle of much less than 90” with this axis. A variety of experimental results such as the increase of fluorescence emission of poly( 1,N6-ethenoadenylic acid) (poly(~A): Toulme & Helene, 1980), the changes in circular dichroism spectra (Jensen et al., 1976; Anderson & Coleman, 1975) and absorption spectra (Jensen et al., 1976) of single-stranded DNA and synthetic polynucleotides also show that the structure of the polynucleotide is radically changed upon binding the protein, and these changes have often been ascribed to unstacking of the bases. An elaborate study of the DNA structure in complex has been done using circular dichroism and absorption spectroscopy (Scheerhagen et al., 1986a). From the measurements and corresponding model calculations it was concluded that the polynucleotide structure becomes rigid and regular; the bases are clearly t)ilted with respect to the long axis of the complex. If one assumes a helical arrangement of the bases in the complex, then the bases must be shifted towards the helix axis and/or neighbouring bases are less rotated with respect to each other as compared to
The protein GP32$ encoded by gene 32 of bacteriophage T4 is one example of single-stranded DNA binding proteins that are present in a variety of organisms (for a review, see Chase & Williams, 1986). It binds strongly and co-operatively to single-stranded nucleic acids (Alberts & Frey, 1970; Jensen et al., 1976) and protects single-stranded J)NA against nucleases (Curtis & Alberts 1976). It is essential for replication (Riva et al., 1970; Breschkin & Mosig, 1977), recombination (Tomizawa et al., 1966; Berger et al., 1969) and repair processes (Wu & Yeh. 1973). After binding GP32, polynucleotides have a strongly altered conformation, which is probably essential for most of the processes mentioned above. There is an increase in the length of the polynucleotides (single-stranded DNA) of at least 50%, as ascertained by electron microscopic (Jjelius et al., 1972) and hydrodynamic measurements (Scheerhagen et al., 1985a,b). Electric t Author to whom all correspondence should be sent. $ Abbreviations used: GP32, protein encoded by gene C?:!of’ bacteriophage T4; poly(&A), poly(l,N6ethenoadenylic acid); LD, linear dichroism. (K)22~2836/88/2203R7~09
$03.00/O
397
0
1988 Academic
Press Limited
398
H. van Amerongen
bases in the uncomplexed polynucleotide. These structural properties are consistent with the observed increase in length of the polynucleotide in the complex with GP32. In conclusion, substantial evidence exists that the position and orientation of the bases are altered when GP32 binds to the polynucleotide. Tn the complex there are close contacts between aromatic residues of the protein and the bases of the polynucleotide. It has been proposed that amino acid residues in the region of residue 72 to I16 (Williams et al., 1981) play an important role in the binding of the protein to single-stranded DNA. This region contains six tyrosine and two tryptophan residues. In agreement with results obtained by Anderson & Coleman (1975). nuclear magnetic resonance measurements show (Prigodich et al., 1984, 1986) that five tyrosine residues are in close contact with the bases; this is also the case for two phenylalanine residues (Prigodich et al., 1986). A role for at least one tryptophan in the binding process has been suggested (Toulme et al., 1984; CasasFinet et al., 1984), which probably corresponds to the finding that at least one tryptophan residue is close to the bases (Toulme & Helene, 1980; Prigodich et al., 1984; Le Doan et al., 1984; Khamis & Maki, 1986). In a recent study (Casas-Finet et al., 1988), it was concluded that at most two tryptophan residues per protein molecule are in close contact with the bases. Studies on model peptides have shown that intercalation does occur with the bases in single-stranded DNA but not in double-stranded DNA (Brun et al., 1975; Toulme & Helene, 1977; Mayer et al., 1979). Stacking interactions between the aromatic residues of the protein and the bases could possibly explain the fact that GP32 binds much more strongly to single-stranded DNA than to double-stranded DNA. From hydrodynamic measurements (Scheerhagen et al., 1985a,b), it appears that the hydrodynamic volume is larger than the volume that can be calculated using the mass of the complex and the partial specific volume. The difference can be explained by assuming an open solvated structure, possibly a superhelix (Scheerhagen et aZ., 1986a, 1985b). To obtain more direct information about the orientation of the bases and aromatic residues, especially tryptophan, in the complex, we performed linear dichroism (LD) measurements on the complex of GP32 and poly(sA) oriented in compressed polyacrylamide gels. This orientation method was first described by Abdourakhmanov et al. (1979). We used this method in a slightly modified way (van Amerongen et al., 1988), compressing a gel containing the complex in one direction (x) and allowing it to expand in one perpendicular direction (y). Rod-like molecules orient preferentially with their long axis along the y-axis. The measuring light is now incident along the z-axis and the difference in absorption, AA, for light polarized along the y-axis and the x-axis is measured. The measured amount of LD gives
et al.
information about the orientation of the bases and aromatic amino acid residues (see below). For the polynucleotide, we have chosen poly(sA) because it has a significant absorption band above 300 nm, well separated from the absorption bands due to the protein, which facilitates the interpretation of the LD spectra. Poll is thought to bind to GP32 in a manner similar to that of the polynucleotides, according to parameters that govern the binding process (Kowalczykowski et al., 1981; Toulme & Helene, 1980).
(a) Interpretation
qf LD
results
For rigid rod-like particles oriented in a polyacrylamide gel, which is compressed with a degree of compressiondiv;eiwidth of the gel before compression width after bY the compression) the following relation holds: cos2cr-1)
0(n),
(1)
where A is the absorption at a particular wavelength 1 when the gel is uncompressed; AA is the amount of dichroism, which is defined as the absorption along the y-axis of the gel (direction of expansion) minus the absorption along the x-axis (direction of compression); a is the angle between the transition dipole moment responsible for the absorption at the wavelength L and the orientation axis of the particle. The angle brackets ( .) denote an average for all transition moments. In the following we will omit these brackets for simplicity. When different transition moments absorb at 2, a weighted average has to be taken. Q(n) is the orientation function that describes how well the particles are oriented. It is 1 for perfect orientation and zero when there is no compression. Tn van Amerongen et al. (1988) Q(n) is given. Here we only give a useful approximation for low values of 72:
D(n)= AW ; c, L.l
(2)
which is correct within a small percentage below n = 1.7. For DNA and RNA and their complexes with the protein it is interesting to relate the amount of LD to the orientation of the bases or aromatic amino acid residues with respect to the orientation axis of the particle. In this case, cos2 c1 has to be replaced by cos2 fl x sin2 6 for transition moments lying in the plane of the bases or aromatic residues (Norden, 1978). The angle fl is defined as the angle between the orientation axis and the plane through the bases or the aromatic residues. The line about which the planes are tilted with respect to the orientation axis is called the inclination axis, and 6 is the angle between the transition moment and the inclination axis. When such a structure is organized as a (local) helix that is further organized into a superhelix where the local axis (e.g. helix axis) makes an angle y with the
Linear Dichroism of a DNA-Protein superhelix axis, the right-hand side of equation (1) has to be multiplied by $(3 cos2 y- 1) (NordBn, 1978). For elongated molecules with some flexibility the amount of LD will decrease when compared to a rigid molecule. This will probably depend on the length of the molecule (L) and its flexibility, which is related to its persistence length (P). This effect is accounted for in a formal way by multiplying the orientation function for a rigid rod by another factor Q(P, L)(O I @(P, L) I 1). This leads to the following relation :
3A(4
3A =+(3cos2pxsin26-1)
x @(n)O(P.L)+(3 In the following term @(P,L)+(3
notation, cos2 y-l).
(30s~y- 1).
(3)
Y will denote the product
Complex
399
remains constant. Light passes perpendicular to the z and y-direction in the z-direction through the press. The pathlength is only 5 mm to minimize the absorption of the gel. Absorption spectra were measured on a Gary 219 spectrophotometer interfaced to a HP85 computer. Spectra were baseline corrected. Accurate absorption spectra in gel were difficult to obtain because of a large contribution of the gel. Within experimental error, the spectrum of t’he DNA-protein complex in the gel phase was identical with that in solution after correct.ion for the gel contribution. However, for reasons of accuracy, spectra measured in solution and corrected for dilution upon going to the gel phase were used for the calculation of the reduced dichroism. LD spectra were measured on a modified (Bokma et al., 1987) Cary 61 spectropolarimeter, also interfaced to a HP85 computer. The crystal was modulated such as to give maximum LD signals. Calibration was done as described (van Amerongen et al., 1988). Absorption and LD spectra were further analysed on a VME 131 computer.
2. Materials and Methods Unless stated otherwise, all measurements were done in the following buffer: 50 mM-NaCl, 1 mM-Na,HPO,, 0.1 mM-EDTA (pH 7.7) (buffer I). The protein GP32 was isolated as described (Scheerhagen et al., 1986a). The used: extinction coefficient was following &,,,=37x 103~-' cm-’ (J ensen et al., 1976). Poly(&A) was from PL Biochemicals (Woerden, The Netherlands) and was used without further purification. The extinction coefficient (per base). Ebb, = 3420 M-’ cm-‘, was obtained by comparing the absorption in buffer I with the absorption in the buffer used by Toulme & HBl&ne (1980), where an extinction coefficient of 3550 Mm 1 cm- ’ has been given. This latter value was taken from (Ledneva et al., 1978). where a value of 0.56 is found for the absorption ratio A(300 nm)/A(257 nm) (A is absorption), which is close to the value of 0.57 that we cm-’ has been find. Although a value of 5110 Me1 reported for this polynucleotide (Janik et al., 1973) and was used by Kowalczykowski et al. (1981) in GP32 binding studies. we have chosen to use the former value in view of the fact that the corresponding A(300 nm)/ ,4(260 nm) ratio of 0.374 (Janik et al., 1973) is much lower than in our case. I,-Trp was from Fluka AC Buchs SG and was dissolved in distilled water. Polyacrylamide gels were made using buffer I. and the final gels contained 145(/, (w/v) acrylamide from Serva (Heidelberg, West Germany). 0.550/, (w/v) N,N-methylenebisacrylamide (BTS), 0.075 “& N,N,N’,N-tetramethylparaphenylenediamine (TEMED) both from Merck-Schuchardt (Hohenbrunn, West Germany) and 0.1% (w/v) ammonium persulphate (APS) from *J. T. Baker Chemicals 1~.V. (Deventer, The Netherlands). All reagents To 750 ~1 of a were used without further purification. solution containing the complex of poly(~A) and GP32 (see also Results). an equal volume of buffer I containing proper amounts of acrylamide, BIS and TEMED was added. After gentle mixing of the solution, 15 ~1 of a 10% (w/v) APS solution was added and this solution was poured into a cast with the dimensions of the gel press. After about 1 h the polymerization was complete and the gel was placed into the press. This gel could be compressed continuously in one direction (2) from an initial width of 9 mm to a final width dfabout 5.5 mm. In this press the gel is allowed to expand in one perpendicular direction (y). where the volume of the gel
3. Results Complex formation between poly(&A) and GP32 could be followed by measuring the absorption spectra from 250 to 400 nm during titration, starting with poly(eA) adding GP32. In contrast to what was found in other measurements (Toulmt? & HBlBne, 1980), we observed a hyperchromism above 250 nm for the polynucleotide upon complex formation. The shape of the absorption difference spectrum (spectrum of the complex minus the spectrum of poly(&A) and that of GP32 measured in the unbound form) is almost identical with the induced difference in absorption, obtained when a poly(&A) solution is heated without the protein (not shown). No significant change in the protein absorption can be observed. A similar conclusion using other polynucleotides and the same or other single-stranded DNA-binding proteins was drawn elsewhere (Scheerhagen et al., 1986a). Therefore, we conclude that the absorption spectrum of GP32 remains unchanged in the complex and the absorption spectrum of poly(&A) in complex can be obtained by subtracting the absorption spectrum of free GP32. The resulting spectrum of bound poly(~A) is given in Figure 1, together with the spectrum of free poly(&A). In a saturated complex the amount of hyperchromism is about 36% in the peak at 267 nm at 12*6”C, which is even larger than the 24% obtained from heating to about 80°C. During tifrafion, the absorption in all the peaks increased almost linearly with the amount of GP32 added (data not shown) until saturation was reached. The titration was stopped when there was an excess of at least 40% of GP32 beyond the point of saturation. If the amount of hyperchromism is indicative of the amount of binding, we can determine (data not shown) the size of the binding site (number of nucleotides covered by one protein) by plotting the hyperchromism versu~~ the protein to nucleotide ratio in the same way as was done by van et al. (1987). Amerongen The value
H. van Amerongen
400
et al.
polynucleotides and single-stranded DNA (Bobst et al., 1982; Scheerhagen et al., 19866). Therefore, we shall use s = 10 for further calculations. However, we shall show that taking s = 8 does not significantly change the interpretations results. The saturated complexes described
of the I,D above
were
used to prepare gels as described in Materials and Methods. After compression of a gel, a clear LD spect’rum could be measured, as shown in Figure 2(a). Below 250 nm no measurements are 250
275
300
325
Wavelength
possible because of a large amount of scattering and strong absorption in this region. In addition t’o the spectral features due to the protein and the
350
(nm)
polynucleotide,
Figure 1. Absorption spectrum of poly(&A) free ( - ) and in complex with GP32 (---) at 12.6”C in buffer I. Corrections have been made for the protein contribution and dilution.
s = lO.Of
1.6 is an average
of four
titrations
with
the standard error reflecting the scatter in different experiments. Our result is clearly different from s = 5fl (Toulm6 & H618ne, 1980) and s = S&l reported by Kowalczykowski et al. (1981). But when the extinction coefficient used in our work is applied to the latter results, s = 9 + 1.5 is obtained, which is in reasonable agreement (see also Materials and Methods). The lower values obtained in fluorescence
titrations (Toulmk & H&l&e, 1980; Kowalczykowski et al., 1981) are an underestimate
due to difficulties associated with the specific conditions of these experiments (for example, see Bobst et aZ., 1982). On the other hand, s = 10 seems in good agreement with values obtained in recent studies for complexes of GP32 with different
contribution
the measured LD spect’ra show a from
the gel. An example
of an I,I)
spectrum for a blank gel is also shown in Figure 2(a) and is characterized by a monotonously decreasing dichroism, going where a minimum
from
400 nm
to about
252 nm,
is found. The LD of all gels seems to rise below this wavelength, but this is probably because very little light passes through the gel below this wavelength. The intensity of the whole spectrum increases when the gel is compressed further. The shape of the spectrum of a blank gel was found to be close to constant in different measurements, but the intensity changed somewhat from one measurement to another. Also the exact place of the slightly. When
minimum around 252 nm varied measurement,s were performed, the
spectra were corrected for the contribution of the gel in the following way: between 350 and 400 nm the spectrum was fitted with a combination of a straight horizontal line (accounting for a small drift of the apparatus) and a spectrum of a blank gel using a linear least-squares method. The contributions determined
in this manner
in this wavelength
region were used to correct the whole spectrum. The
1 3.5
x lo-’
-3.5
x 1o-3
-7.0
x lo-’
1 300
250
1 350
Wavelength
400
250
(nm)
300 Wavelength
350 (nm)
400
(b)
(a)
Figure 2. (a) (-) LD spectrum of poly(&A) +GP32 complex in a compressed polyacrylamide gel (n = 1.53). (---) LD spectrum of a blank gel compressed with a factor n = 1.53. (. .) LD spectrum of the same complex corrected for the baseline as described in the text (see Results). [poly(&A)] = 32.6 PM, [GP32] = 4.8 ,UM, t = 8°C. (b) (-) LD spectrum of poly(&A) +GP32 complex in a compressed polyacrylamide gel (n = 1.53) [poly(&A)] = 32.6 PM, [GP32] = 4.8 PM, t = 8°C. (---) LD spectrum
of poly(eA)
(n = 1.53) corrected
to 32.6 PM and multiplied
by a factor
of 3.75 (t = 11 T).
Linear
Dichroism
of a DNA-Protein
shape of the corrected LD spectra of the complex was almost the same at all degrees of compression (spectra not shown). For different gels the spectra varied somewhat in height, especially around 250 nm, where uncertainties due to the baseline corrections are largest. An LD spectrum of the complex corrected as described above is also given in Figure 2(a) with n = 1.53 at t = 8°C. The spectrum did not vary when raising the temperature to 25”C, reminiscent of results obtained with absorption and circular dichroism spectra of other polynucleotides in complex with GP32 (Scheerhagen et al., 1986a). The spectrum is positive and is markedly different from the spectra of GP32 and poly(&A) alone. The spectrum of free GP32 (not shown) is very small in comparison with the complex spectrum, and its contribution to the LD spectra of the gels containing the complexes can safely be neglected. For poly(.zA) (see Fig. 2(b)) the spectrum is negative everywhere, although there may be a contribution from a positive band around 290 nm. In the latter spectrum, the negative dichroism is in agreement with the expectation that the bases are oriented more or less perpendicular to the orientation axis of the polynucleotide with the most important transition moments, dipole corresponding to n--71* transitions, lying in the plane of the bases. The small amplitude of the LD is probably due to mobility of the bases and high flexibility of the sugar phosphate backbone. Above 300 nm the LD of the complex is positive, indicating that the bases are significantly tilted with respect t,o the orientation axis. The fine structure around 280 nm in the LD spectrum is completely different from the fine structure in the absorption spectrum of poly(~A), where the peaks are at different wavelengths. Therefore, the LD around 280 nm is probably due to the protein. In this wavelength region the protein absorption stems predominantly from the tryptophan residues. The peaks in the fine structure coincide remarkably well with those in the absorption spectrum of L-Trp measured in water (see Fig. 3). However, the clear fine structure that can be observed in the free tryptophan absorption spectrum and in the LD spectrum of the complex is present neither in the absorption spectrum of the protein (Fig. 3) nor in that of the complex. Therefore, we propose that it must be attributed to spectrally distinct and specifically oriented tryptophan residues in the GPSB-poly(~A) complex (see Discussion). A similar fine structure is also observed in the LD spectrum reported by Takahashi et al. (1987) for RecA protein from E. coli in complex with single-stranded DNA. The fine structure present in the absorption spectrum of poly(~A) in the region 255 to 280 nm cannot be discerned in the LD spectrum of the complex, indicating that the absorption in this region does not contribute significantly to the LD. The positive shoulder around 270 nm in the complex is also present in complexes with other polynucleotides (van Amerongen & van Grondelle,
250
Complex
275
401
300 Wavelength
325
350
(nm)
Figure 3. (-) LD spectrum of poly(&A)+GP32 complex in a compressed polyacrylamide gel (n = 1.53) [poly(cA)] = 32.6 PM, [GP32] = 4.8 p~.(. .) Absorption spectrum of L-Trp measured in water (20°C). (---) Absorption spectrum of GP32 in buffer I (20°C).
unpublished results) and we do not assign it to poly(&A). The observed LD spectrum of the complex can reasonably be assigned to a superposition of only the long wavelength band around 307 nm due to poly(&A) and contributions from the tryptophan residues. It would not be very realistic to try a detailed fit for the whole spectrum, in view of the large number of different aromatic groups in the complex (bases, tyrosine and tryptophan residues). Moreover, even in a single residue more transitions can contribute to the absorption in this wavelength region, e.g. in tryptophan, in which the absorption around 280 nm is due to two almost degenerate transition moments with different orientations. To interpret the spectra in a more quantitative way we use the fact that the LD at 315 nm is exclusively due to poly(.zA) and that the positive LD at 280 nm can be ascribed predominantly to tryptophan residues. The contribution from the long wavelength band of poly(~A) to the LD at 280 nm is estimated to be about 5%, assuming a Gaussian shaped band, centred at 307 nm. In Figure 4 the amount of dichroism at 280 and 315 nm is plotted versus n to show the orientational behaviour of the complexes in gel. The drawn lines represent the orientation function @(n) for a rigid rod multiplied by a constant to give a reasonable fit with the experimental data. In all measurements the amount of dichroism increases somewhat more slowly than would be expected in the case of a rigid rod. This may be due to some uncertainties in the baseline corrections, but it may also indicate that @(L/P) is dependent on n. However, as a first approximation we assume that @(L/P) is constant at all values of n. This will not have much impact on our final conclusions. Averaging A.A/(A@(n)) for different values of n, we have at least a lower limit for AA/(A@(n)O(L/P)) (i.e. taking @(L/P) = 1).
402
H. van Amerongen
O-060
7 G Q
0.045
0.030
0.015
C I.0
I.2
I.4
I.6
”
Figure 4. (A) AA(280 nm) of poly(~A)+GP32 complex at different values of n divided by A of the protein in the complex (see the text). The fit represents the curve 0.183 m(n). (0) AA(315 nm) of the same complex divided by A(315 nm) of the complex. The fit represents the curve 0.143 Q(n).
The absorption at 315 nm is exclusively due to poly(aA) and at this wavelength we find that AA/(A@(n)) = 0*15+0*02. The error is the standard error in the results of four different series of measurements. Variations in one series of measurements due to deviation from the rigid rod curve have not been taken into account. The scatter in the results will be due partly to some uncertainty in the baseline corrections. The uncertainty in AA due to some uncertainty in the calibration of the LD apparatus is about 3%, whereas the values of n are accurate within a small percentage. The error in the value of A is small. At 280 nm almost all of the LD can be ascribed to the protein. Subtracting 5% of the signal (estimated contribution from the long wavelength band of poly(aA)), we obtain AA/(A@(n)) = 0*18&0*02. In this case, A is the absorption of the protein in the complex, which has been calculated using a value of 10 for the size of the binding site, and using the fact that the protein absorption at this wavelength does not change upon binding. Taking s = 8 results in AA/A@(n) = 0.14f0.02. Note that the choice of the value of s does not influence the value AA/A for the poly(aA) bases.
4. Discussion Complex formation between poly(&A) and GP32 leads to an induced hyperchromism for poly(eA), indicating that the structure of poly(aA) is altered in the complex. Hyperchromism is also induced for other polynucleotides (poly(dA) and poly(rA)) and single-stranded DNA (Newport et aZ., 1981; Scheerhagen et al., 1986a) upon binding of GP32. From the increased poly(&A) fluorescence upon binding of GP32 Toulme & Helene (1980) arrived at a similar conclusion, but they did not observe the hyperchromism. The authors used a different buffer and probably performed their measurements at a
et al.
higher temperature, where the free poly(aA) is less hypochromic. But this can only explain less than half of the difference and we do not have an explanation for the remaining discrepancy. The changes in the absorption spectrum after complex formation are very similar to the changes obtained by heating solutions of poly(aA). However, t’he increase in absorption induced by binding is even higher than that upon heating to about 8O”C, an arbitrary high temperature. The size of the binding site we obtained is lO*O( + 1.6) nucleotides. It is not the intention of this work to determine the value of 8, but we performed the titrations to be sure that LD measurements were performed with a saturated complex. In fact, the value of s does not influence the value of AA/A for the bases. However, it is important for the determination of the absorption of the protein in the complex. For our calculations we have chosen s = 10, in agreement, with our findings. In view of the contradictory values reported for s (see Results) we show that with s = 8 the same essential conclusions can be drawn concerning the orientation of the tryptophan residues in the complex. (a) Qualitative
interpretation of the LD spectrum of the complex
In gels, the complexes do not dissociate, as can be concluded from the fact that the LD spectrum of the complex is totally different from the spectra of GP32 and poly(&A) alone. The LD spectra have to be corrected for the contribution of the gel. This can lead to some uncertainty in the spectra, which is probably reflected in the scatter of the results obtained in different experiments. However, the main characteristics of the spectra are very reproducible and the accuracy in the amount of LD is reasonable. A striking result is the fact that the LD signal above 300 nm is positive in the complex, where it is negative for poly(&A) alone. This shows t,hat the orientation of the bases must change significantly when the complex is formed. The predominant transition moments in these planar aromatic molecules are due to X-Z* transitions and are polarized in the plane of the bases. This is in agreement with the negative LD for free poly(eA), where the bases are more or less perpendicular to the long axis of the polynucleotide (with the possible exception of an out-of-plane transition moment around 290 nm). This means that the positive LD above 300 nm must be explained by a strong tilting of the bases (b < 54.7”). This strong tilting can explain why the negative birefringence measured for single-stranded DNA (Scheerhagen et al., 1985a) reverses sign in the complex with GP32 and, in addition, is in agreement with the CD calculations done for poly(rA) and for poly(dA) in complex with GP32 (Scheerhagen et al., 1986a). A second remarkable characteristic of the LD spectrum is the fine structure around 280 nm and, as indicated in Results, this feature must be ascribed to the protein. The major contribution to
Linear Dichroism of a DNA-Protein the absorption around this wavelength stems from the indole groups of the tryptophan residues. In an indole group two transition moments, L, and L,, contribute to the absorption near 280 nm. These moments are nearly perpendicular to each other (Yamamoto & Tanaka, 1972). Different shapes and relative intensities have been proposed for both bands (Yamamoto & Tanaka, 1972; Valeur & Weber, 1977), and the contribution of the L, band to the intensity at 280 nm varies in both studies from about 20 to 35%. The L, band is held to be responsible for most of the fine structure. As reported by Sun & Song (1977), the absorption characteristics of an indole group can change markedly when different solvents are used. In the protein the different indole groups are probably not all in the same environment, resulting in a smooth absorption spectrum GP32, both in the free and in the complexed form, which does not have such definitive fine structure as Trp. Also wavelengthselected optically detected magnetic resonance measurements reveal the existence of environmentally distinct and spectrally different types of Trp in GP32. One of these types represents tryptophan residues in close contact with the bases (Khamis & Maki, 1986). As was described in Results, the absorption difference spectrum due to complex formation shows no changes attributable to the protein; therefore, one would expect also a relatively smooth spectral contribution of the trypophan residues to the LD, if all tryptophans were oriented in the same way with respect to the long complex axis. However, in the LD spectrum there is a clear fine structure, indicating that not all indole groups have the same orientation with respect to the long axis of the complex. At this moment it is difficult to give a more detailed interpretation of the tine structure in the LD spectra around 280 nm, because of the many different transition moments that contribute, but, nevertheless, from the positive sign of the LD we can conclude that the average angle between the indole rings and the long axis of the complex is certainly smaller than 54.7”. (b) Estimation of the value of the parameter Y In the following, $&
we will use the values: (315 nm) = 0.15
Complex
403
superhelix, the term Y must be 1, which is an absolute upper limit for Y. The complex is possibly folded into a superhelix (Scheerhagen et al., 1986a, 1985b) and a value of 40” has been proposed for y (Scheerhagen, 1986). This would lead to Y < 0.4. In addition, flexibility leads to a further decrease of the value for Y. A rough estimate of the contour length of the complex can be made, using the fact that the polynucleotide contains on the average about 500 bases. From a rise per base in the complex of 0.44 nm (Scheerhagen et al., 1986a) a lower limit for the contour length of about 220 nm is found. This is much larger than the persistence length P, which is certainly lower than 50 nm and probably is about 30 nm, as can be concluded from light-scattering experiments (M. E. Kuil et al., unpublished results). Therefore, the rigid rod approximation is not valid, which will further lower the term Y. On the other hand, we can estimate a lower bound for Y, assuming that all indole rings are oriented with respect to the local helix axis in such a way as to give a maximum amount of LD at 280 nm. This is the case when b = 0 for all indole groups (P(indole)), 6 = 90” for the L, transition moment (&LO)) and 6 = 0” for the L, transition moment (6(L,)). We also assume that 20% of the absorption at 280 nm is due to the L, band. A higher percentage leads to a larger value of the lower bound of Y. This chosen conformation leads to AA/A@(n) = 2.1 at 280 nm when Y = 1. To explain the measured value of 0.18, Y must therefore be at least 0.09. This is an absolute lower bound for Y. However, in this extreme case one would not observe such distinct fine structure in the LD spectrum. To obtain a significant fine structure, at least one but probably more Trp residues cannot contribute significantly to the positive LD in the way described above. When only four Trp residues contribute to the positive LD, then AA/A@(n) would maximally be about 1.7 at, 280 nm, corresponding to a minimum value for Y of 0.11, which is still a rather extreme value in view of the extreme orientations for the other four Trp residues. So the real value of Y will be somewhere between the extreme values of 0.11 and 1.OObut in view of the foregoing discussion it seems reasonable to assume that its real value is somewhere near 0.25, corresponding to a superhelix with y = 35” and a value for @(L, P) of 0.5, for example.
and L+!(280 nm) = 0.18 A@(n) to estimate average /? values of the bases and indole groups with respect to the assumed local helix axis, using the relation: AA ~=3(3cos2psinZ6-l)Y. A@(n) An initial estimate of Y is needed to determine the product term cos’ j? sin’ 6. When the complex behaves as a rigid rod, which is not folded into a
(c) Orientation of the bases and tryptophan with respect to the local helix axis
residues
The bases of the polynucleotide are probably organized in a regular structure (Scheerhagen et al., 1986a) and all contribute approximately in the same way to the total amount of LD. In Table 1 we give the average angles between the bases and the local helix-axis when different values for 6,,, are assumed with Y = O-25. It will be shown below that, within certain limits, other values of Y lead to similar values of 8. The same is done for the indole
H. van Amerongen
404
Table 1 Angles of bases and indole groups with the local helix axis B(indole) 6 3,5 9o” 70‘ 50”
/+A)
6(L,)
m= 5
m= 4
m = 3
m= 2
m= 1
47” 43” 27”
90” 70” 50”
38” 35” 19”
35” 31” 11”
30” 25” NP
17” NP NP
NP NP NP
Average values for the angle /I between the bases/indole groups and the assumed local helix axis, using Y = 0.25. Average values of /?(indole) have been calculated, assuming that m indole groups contribute to the positive LD. It is assumed that the L, band leads to 80% of the Trp absorption at 280 nm. A lower percentage will lead to lower average angles &indole). NP indicates that the corresponding conformation is not possible.
groups with different values for 6(&J, again assuming that L, is responsible for 80% of the absorption at 280 nm. A lower percentage leads to values of fi closer to 0”. We consider different cases where a number of indole groups, m, contribute significantly to the positive LD and then calculate the average orientation of these indole rings. In some cases the observed amount of LD cannot be explained with the chosen values for m, 6 and Y: because the term cos’ B would be larger than 1 (eqn (3)). It is directly clear from Table 1, that more than one Trp contributes to the positive LD and even two residues seems highly unlikely. Also the angle between the bases and the local helix axis must be lower than 48”. To be more specific we will use the fact that no fine structure below 300 nm is observed that can be ascribed to poly(~A). Therefore, the corresponding transition moments around 270 nm do not contribute significantly to the LD. If we call the angle of the corresponding transition moment with the inclination axis 6,,, and assume that this transition moment is also oriented in the plane of the base, then 6 2,0 = &,, rf: x and x denotes the angle between both transition moments of poly(~A). The value of x is about 55”, which is deduced from fluorescence depolarization measurements of the free nucleotides in high concentrations of glycerol (data not shown). Because there seems to be no dichroism due to poly(&A) around 270 nm, we have the following relation for /?(&A) and ha15: 3 cos’ /I sin’
(6,,,
+x)-
1 = 0.
Together wit*h the known amount of dichroism at 315 nm we can calculate I, 8315 and 6,,, for a chosen value of Y. For Y = 0.25, these values are, for respectively, 44”, 72“ and 127”. Assuming instance that the complex is completely rigid and that there is no superhelix ( Y = l), we calculate the values 48”, 75” and 130”. However, if we take the minimum value of 0.11 for Y. we obtain the values 36”, 80” and 135”. So, even assuming rather extreme values for Y does not lead to large variations in the calculated
et al.
angle between the bases and the local helix axis. The safest estimation for this angle is 42( +6)‘, but most probably it is somewhere between 40” and 45”. Because we do not know the orientation of the transition moments in the molecular plane, we are not able to determine the approximate orientation of the inclination axis and, therefore, we cannot determine the values of twist and tilt of the bases. Tt is argued by Casas-Finet et al. (1988) that at least one and at most two tryptophan residues are in close contact with the bases. If one Trp molecule would intercalate, as is mentioned as a possibility in Introduction, it would make about the same angle with the local axis as the bases. If Y = 0.25 and p = 40”, such a single intercalating Trp molecule can (if 6( L,) = 90”) explain at most 17 o/o of the observed dichroism at 280 nm and two tryptophan residues at most 34%. So, there must be other nonintercalating tryptophan residues that contribute to the positive dichroism, and the indole groups of these Trp residues make, on average, a smaller angle with the local helix axis as compared to the assumed intercalated tryptophan(s). Note that any other choice of 6(L,), either for the possibly intercalated or for the remaining residues, tryptophan residues will lead to only more extreme values of p (i.e. closer to 0”) for the non-intercalated groups. A choice of s = 8 for the binding site size will reduce AA/A in the protein band, but an intercalated tryptophan residue explains maximally 21 y. of the observed dichroism and therefore the conclusions are not significantly altered. The main conclusions of this work are as follows: (I ) in the GP32-single-stranded complex the polynucleotide assumes a non-random specific, conformation characterized by a strong base tilt, with B = 42( +6)“, but most probably between 40” and 45”. (2) Probably, at least more than one tryptophan residue contributes in a specific way to the LD. Part of the protein LD may be due to one or more intercalated Trp residues, but this cannot explain all the observed dichroism and the indole of the non-intercalating tryptophan groups residues make an angle with the complex axis that is almost certainly smaller than t’hat of possibly intercalated residues. We are performing similar measurements with GP32 complexed to polynucleotides containing other bases, in both their riho and their deoxyribo forms (van Amerongen & Van Grondelle unpublished results). Although the spectra show more overlap with that of the protein, an advantage of these polynucleotides is that the directions of the various transition moments are available. Using poly(~A) as a starting point, we can be more specific about the DNA conformation in terms of base twist and tilt and hopefully find further details concerning the orientation of the tryptophan residues. GP32 was skilfully isolated by J. den Blaauwen. This work was supported by the Netherlands Foundation for Pure Research via the Dutch Foundation of Biophysics.
Linear
Dichroism
of a DNA-Protein
References Abdourakhmanov, I. A., Ganago, A. O., Erokhin, Y. E., Solov’ev, A. A. & Chugunov, V. A. (1979). Biochim. Biophys. Acta, 546, 183-186. Alberts, R. M. & Frey. L. (1970).Nature (London), 227, 1313-1328. Anderson, R. A. & Coleman, ,I. E. (1975). Biochemistry, 14, 5485-5491. Berger, H., Warren. A. ,J., Fry, K. E. (1969). J. Virol. 3, 171-175. Robst, A. M.. Langemeier, P. W., Warwick-Koochaki, P. E.. Hobst, E. V. & Ireland, J. C. (1982). J. Biol. Chem. 257, 6148-6193. Rokma. tJ. T.; ,Johnson, W. C. Jr & Blok, J. (1987). Biopolymers,
26, 893-909.
Breschkin, A. M. & Mosig, G. (1977). J. Mol. Biol. 112, 2955308. Krun. F., Toulm&. J-,I. & HPlBne, C. (1975). Biochemistry, 14, 558-563. (‘asas-Finet, J. R.. ToulmC, ,J.-J., Cazenave, C. & Santus. R. (1984). Biochemistry, 23, 1208-1213. Casas-Finet. J. R., ToulmB, J.-J., Santus, R. & Maki. A. H. (1988). Eur. J. Biochem. 172, 64-646. Chase, ,J. W. & Williams, K. R. (1986). Annu. Rev. Biochrm. 55, 103-136. (‘urtis. M. *I. & Albrrts, B. (1976). J. Mol. Biol. 102, 793% 816. Delius. H.. Mantell. X’. ,J. & Alberts, B. (1972). J. Mol. Biol. 67. 341-350.
Janik, B.. Sommer. R. G., Kotick, M. P., Wilson, D. P. & Chem. Phys. 5, Erickson. R.
Complex
405
Newport, J. W., Lonberg, N., Kowalczykowski, S. C. & von Hippel, P. H. (1981). J. Mol. Biol. 145, 105-121. Eordbn, B. (1978). Appl. Spectrosc. Rev. 14(2), 157-248. Prigodich, R. V., Casas-Finet, J., Williams. K. R., Konigsberg, W. & Coleman, CT.E. (1984). Biochemistry,
23, 522-529.
Prigodich, R’. V., Shamoo, Y., Williams. K. R.. Chase, J. W., Konigsberg, W. H. & Coleman, J. E. (1986). Biochemistry,
25, 3666-3672.
Riva, S., Cascino, A. & Geiduschek, E. P. (1970). J. Mol. Biol. 54. 85-102. Scheerhagen, M. A. (1986). Doctoral Thesis, Vrije Universiteit. Scheerhagen. M. A., van Amerongen, H., van Grondelle, R. & Blok, tJ. (1985a). FEBS Letters, 179, 221-224. Scheerhagen. M. A., Kuil, M. E., van Grondelle, R. & Blok, ,J. (1985b). FEBS Letters, 184, 221-225. Scheerhagen. M. A., Bokma, J. T., Vlaanderen. C. A.. Blok, J. & van Grondelle, R. (1986a) Hiopolymers, 25, 1419-1448. Scheerhagen. M. A., Vlaanderen, C. A., Blok, ,J. & van Grondelle, R. (19866) J. Hiomol. Struct. Dyn. 3, 887898.
Sun, M. & Song, P.-S. (1977). PhotochewL. Photobiol.
25.
3-9.
Takahashi,
M., Kubista,
M. C NordBn. B. (1987). .I. Biol.
Chem. 292, 8109-8111.
Tomizawa, J. I.. Anraku, N. & Twama. Y. (1966). J. Mol. Biol. 21. 247-253. ToulmP, ?J.-J. & H&ne, C. (1977). J. Biol. (‘hem. 252, 2446249.
Toulmk.
J.-J.
& HBlBne, CZ. (1980). Biochim.
Riophys.
Acta, 606, 95-104.
Toulm6. J-
Edited by P. H. won Hippel