ABSTRACT
The hydration in the minor groove of double stranded DNA fragments containing
the sequences 5
'
-dTTAAT, 5
'
-dTTAAC, 5
'
-dTTAAA and 5
'
-dTTAAG was investigated by studying the decanucleotide duplex
d(GCATTAATGC)
2
and the singly cross-linked decameric duplexes 5
'
-d(GCATTAACGC)-3
'
-linker-5
'
-d(GCGTTAATGC)-3
'
and 5
'
-d(GCCTTAAAGC)-3
'
-linker-5
'
- d(GCTTTAAGGC)-3
'
by NMR spectroscopy. The linker employed consisted of six ethyleneglycol units.
The hydration water was detected by NOEs between water and DNA protons in NOESY
and ROESY spectra. NOE-NOESY and ROE-NOESY experiments were used to filter out intense exchange cross-peaks and to observe water-DNA NOEs with sugar 1
'
protons. Positive NOESY cross-peaks corresponding to residence times longer than
~
0.5 ns were observed for 2H resonances of the central adenine residues in the duplex containing the sequences 5
'
-dTTAAT and 5
'
-dTTAAC, but not in the duplex containing the sequences 5
'
-dTTAAA and 5
'
-dTTAAG. In all nucleotide sequences studied here, the hydration water in
the minor groove is significantly more mobile at both ends of the AT-rich inner segments, as indicated by very weak or negative water-A 2H NOESY cross-peaks. No positive NOESY cross-peaks were detected with the G 1
'
H and C 1
'
H resonances, indicating that the minor groove hydration water near GC base
pairs is kinetically less restrained than for AT-rich DNA segments. Kinetically stabilized minor groove hydration water was
manifested by positive NOESY cross-peaks with both A 2H and 1
'
H signals of the 5
'
-dTTAA segment in d(GCATTAATGC)
2
. More rigid hydration water was detected near T4 in d(GCATTAATGC)
2
as compared with 5
'
-d(GCATTAACGC)-3
'
- linker-5
'
-d(GCGTTAATGC)-3'
, although the sequences differ only in a single base pair. This illustrates the
high sensitivity of water-DNA NOEs towards small conformational differences.
NMR spectrocopy offers unique possibilities to study the hydration of DNA in
aqueous solution at high resolution. By measuring the NOE between water protons
and DNA protons, individual hydration water molecules of the innermost layer of
hydration can be detected. The measurements are sensitive to the residence
times of the water molecules at their hydration sites in the 0.1-1 ns time regime (
1
,
2
). Based on a simple model of cross-relaxation, the sign of the water-DNA cross-peaks observed in NOESY spectra inverts for residence times of ~0.5 ns, while shorter and longer residence times result
in negative and positive NOESY cross-peaks respectively (
1
). Data are now available for several DNA fragments with B-type conformation (
3
-
7
) and for non-canonical DNA structures (
8
). NOE measurements performed with B-DNA indicated that the spine of hydration in the minor groove of AT-rich DNA segments, which had been observed by X-ray crystallography in single crystals (
9
-
11
), is characterized by water residence times >1 ns, while at all other hydration
sites the water molecules seem to exchange much faster (
4
,
7
). A comparison of the dodecamers d(GTGGAATTCCAC)
2
and d(GTGGTTAACCAC)
2
showed that an ordered spine of hydration with water residence times >0.5 ns is associated with the sequence 5'-dAATT, whereas the sequence 5'-dTTAA kinetically destabilizes hydration in the minor groove.
The study of DNA hydration is limited by the number of non-exchanging DNA protons with which NOEs with the hydration water can be
observed. The conclusions on the water residence times in DNA minor grooves
have mainly been based on NOEs observed with adenine 2 protons. The water-DNA NOEs with the 1' desoxyribose protons, which also point into the minor groove of B-type DNA structures, are usually obscured by overlap with
very big exchange cross-peaks from the hydroxyl protons from the 3'- and 5'-ends of the DNA fragments. The exchange cross-peaks can be removed by increased
temperature or the addition of exchange catalysts, like phosphate or ammonia,
which cause coalescence of the signals from the exchanging protons with the water resonance (
3
). However, under these conditions the exchange rates of the imino protons are
also increased (
12
), which makes it difficult to distinguish direct water-A 2H NOEs from imino proton-A 2H NOEs appearing at the water chemical shift because of rapid
exchange between imino and water protons. Here we applied recently developed
NMR experiments (
13
) which enable the observation of water-DNA NOEs with the 1' protons by separating the hydroxyl proton exchange peaks from the NOEs in the absence of exchange catalysts.
The present study started with the observation of minor groove hydration water
with residence times >0.5 ns in the 5'-dTTAA segment of d(GCATTAATGC)
2
, which shows that a 5'-dTTAA sequence does not automatically infer a kinetic
destabilization of the spine of hydration in the minor groove. To investigate
the effect of the nucleotides surrounding the 5'-dTTAA segment, the two DNA fragments 5'-d(GCATTAACGC)-3'-linker-5'-d(GC- GTTAATGC)-3'
and 5'-d(GCCTTAAAGC)-3'-linker-5'-d(GC- TTTAAGGC)-3' were
synthesized, which contain all four possible sequences 5'-dTTAAT, 5'-dTTAAC, 5'-dTTAAG and 5'-dTTAAA. The linker between the two
DNA strands consists of six ethyleneglycol units connected to the two DNA
strands via a phosphate group: -PO
2
-O-(CH
2
CH
2
O)
6
-PO
2
-. The use of such a synthetic linker enables one to synthesize the DNA
duplex in one single strand, resulting in an accurate 1:1 ratio of the
complementary strands and in thermally stabilized duplexes (
14
,
15
). It has been demonstrated that the linker employed does not significantly
alter the structure of the surrounding stretch of DNA duplex. The three
oligonucleotides were used to investigate the sequence dependence of the minor
groove hydration. The results indicate that the mobility of the minor groove
hydration water molecules, as measured by the sign and size of water-DNA NOEs, reports on small conformational differences with a sensitivity comparable with that of chemical shift measurements.
NOEs between the
1
H NMR signal of the water and the DNA are used to probe the presence of
hydration water molecules near the DNA protons. The NOEs give rise to an exchange of magnetization between water and DNA protons which is manifested as cross-peaks in two-dimensional NOESY (
16
) and ROESY (
17
) spectra. Positive NOESY cross-peaks are observed if the dipole-dipole interaction lasts for longer than ~500 ps (
2
; see Discussion). Negative water-DNA NOESY cross-peaks indicate exchange of the hydration water within ~100 ps. The NOE cross-peaks are always negative in ROESY. Cross-peaks due to proton exchange between the water and
the DNA are positive both in NOESY and ROESY. Note that a negative NOE leads to
a positive NOESY cross-peak and vice versa. In the following we always refer to the sign of the
cross-peaks rather than the sign of the NOEs.
The self-complementary DNA sequence 5'-d(G
1
C
2
A
3
T
4
T
5
A
6
-
A
7
T
8
G
9
C
10
)-3' and the singly cross-linked DNA sequences 5'-d(G
1
C
2
A
3
T
4
T
5
A
6
A
7
C
8
G
9
C
10
)-3'-linker-5'-d(G
11
C
12
G
15
- T
14
T
15
A
16
A
17
T
18
G
19
C
20
)-3' (decamer 1) and 5'-d(G
1
C
2
C
3
T
4
T
5
-
A
6
A
7
A
8
G
9
C
10
)-3'-linker-5'-d(G
11
C
12
T
13
T
14
T
15
A
16
A
17
G
18
G
19
- C
20
)-3' (decamer 2) were synthesized on a Pharmacia DNA synthesizer. The
linker group -PO
2
-O-(CH
2
CH
2
O)
6
-PO
2
- was introduced as a phosphoramidite building block. It is commercially available from Glen Research (Serling, VA) and was used as supplied.
After cleavage of the protecting groups by heating the protected oligonucleotides in concentrated ammonium hydroxide, the samples were purified by anion exchange chromatography on Q-Sepharose columns with a 0.5-2.0 M NaCl gradient. To remove any salts which could catalyze proton exchange, the samples were
desalted by repeated dialysis against water and subsequently ultrafiltrated
with a 200 mM NaCl solution and finally pure water to remove excess salt. The
lyophilized samples were dissolved in a mixture of 90% H
2
O/10% D
2
O and the pH adjusted to 7.0. The final concentration of the samples was 1.2 mM DNA duplex for d(GCATTAATGC)
2
, 2 mM for decamer 1 and 1.8 mM for decamer 2.
All experiments were recorded at 4oC on a Bruker DMX-600 NMR spectrometer equipped with a Q-switch probe head (
18
). Two-dimensional NOESY, ROESY, NOE-NOESY and ROE-NOESY experiments were recorded for each of the three DNA
fragments. NOESY and ROESY spectra were recorded using previously published
pulse sequences, where the water signal is suppressed before acquisition by a
spin-lock pulse (
19
,
20
). In addition, the quality factor of the probe head was switched low during the
evolution time and during the NOESY mixing time, to optimize the sensitivity
and resolution of the water-DNA cross-peaks (
18
). The NOE-NOESY and ROE-NOESY pulse schemes were modified from previously published sequences (
13
).
Figure
1
shows the pulse sequences of the NOE-NOESY and ROE-NOESY experiments. The water is selectively excited by a 90o Q-switched selective pulse (
13
). The following 90o pulse in the NOE-NOESY sequence (Fig.
1
A) converts the water magnetization into longitudinal magnetization. A homospoil pulse is used to suppress
radiation damping (
21
) during the mixing time [tau]
m1
. During this mixing time, magnetization is transferred from the water protons
to the DNA protons by NOE or chemical exchange. The following pulses represent a conventional NOESY pulse sequence, where the water signal is left unexcited by the jump-return sequence (
22
) after the mixing time [tau]
m2
. This scheme assumes that the water magnetization has largely returned to
equilibrium by the end of [tau]
m2
. To support the return of the water magnetization by radiation damping, a
relatively long mixing time [tau]
m2
is used (200 ms in the present experiments), no homospoil pulse is applied
during [tau]
m2
and the phase of the 90o pulse after t
1
is shifted by 45o with respect to the phase of the 90o pulse before t
1
(
23
). In the ROE-NOESY experiment, the water magnetization is spin-locked by the spin-lock pulse together with the flanking 90o pulses (
24
) for the duration of the mixing time [tau]
m1
during which magnetization is transferred to the DNA protons by the water-DNA NOEs (Fig.
1
B). In both experiments, the Q-factor of the probe head is switched low during t
1
to avoid broadening of the water signal in the F
1
dimension by radiation damping. The results of the NOE-NOESY and ROE-NOESY experiments are two-dimensional spectra in which the water-DNA NOE and exchange cross-peaks are on the diagonal and off-diagonal peaks arise from a second NOE
transfer step to further DNA protons. The off-diagonal peaks aid in the assignment of overlapping direct water-DNA NOE cross-peaks on the diagonal.
Nearly all
1
H resonances were assigned for the three DNA sequences using conventional assignment strategies. The assignment of 3'H signals, which are degenerate with the water resonance at 4oC, was confirmed by spectra recorded at 15oC, at which temperature the water resonance was sufficiently
shifted to lift the degeneracy. Based on these assignments, the water-DNA cross-peaks were assigned in the NOESY, ROESY, NOE-NOESY and ROE-NOESY cross-sections.The assignments are available as supplementary
material in the on-line version of this article.
Figure
2
compares the NOESY cross-section through the water line along the F
2
frequency axis with the cross-section taken along the diagonal of the NOE-NOESY experiment for the DNA fragment d(GCATTAATGC)
2
. Both cross-sections display the water-DNA NOEs. The most intense cross-peaks in the NOESY cross-section (Fig.
2
A) are from chemical exchange. They are from the imino proton of the terminal
base pair at 13.13 p.p.m., the 5' and 3' hydroxyl protons at 6.03 and 6.52 p.p.m. and the labile amino protons of the terminal base pairs at 7.3 and 8.24
p.p.m. These exchange peaks are absent from the diagonal through the NOE-NOESY spectrum (Fig.
2
B), because their magnetization exchanges back to the water signal during the
second mixing time [tau]
m2
(see Materials and Methods). All intense NOE cross-peaks from the NOESY cross-section of Figure
2
A are reproduced in the diagonal cross-section through the NOE-NOESY spectrum (Fig.
2
B). Different peak intensities in NOESY and NOE-NOESY are expected, because different protons relax to different extents
during the evolution time t
1
and the mixing time [tau]
m2
of the NOE-NOESY experiment. Yet, the intensities of the NOEs with the base protons
vary by <3-fold and the signs of the cross-peaks are conserved. Most importantly, removal of the biggest
exchange cross-peaks enables observation of water-DNA NOEs with the 2H of adenine 6 and the 1' desoxyribose protons between 5 and 6.5 p.p.m. (Fig.
2
B).
Figure
4
shows the diagonal cross-sections through the NOE-NOESY and ROE-NOESY spectra of d(GCATTAATGC)
2
. The positive peaks in the ROE-NOESY cross-section (Fig.
4
A) are either residual exchange peaks, like the signal of C10 NH
2
at 8.24 p.p.m., or from TOCSY relayed exchange peaks, like the signals of G1 5',5''H and C10 3'H, which arise during the ROE mixing time from a
TOCSY transfer with the rapidly exchanging 5' and 3' hydroxyl protons. Chemical exchange peaks tend to be more intense in the NOE-NOESY than in the ROE-NOESY cross-sections, partly because of the longer mixing time [tau]
m1
used in the NOE-NOESY experiments and partly because broad exchange peaks relax faster
during the ROESY spin-lock. Disregarding the NOEs with protons of the terminal base pairs, which
are likely to arise from interactions with rapidly exchanging DNA protons, all negative peaks in the ROE-NOESY spectrum represent direct NOEs with hydration water.
The protons A6 2H, A7 2H and A7 1'H are the only protons which yield positive peaks in the diagonal cross-section through the NOE-NOESY spectrum (Fig.
4
B) and which are neither from terminal base pairs nor from labile OH or NH
groups. In addition, there is a positive peak with A6 1'H which overlaps with a negative peak from G1 1'H (Fig.
3
B) and therefore does not appear in the diagonal cross-section of Figure
4
B. The intensity of the off-diagonal peak A7 1'H/2''H is too weak to be seen in Figure
3
B, but the A7 1'H resonance appears in the diagonal cross-section of Figure
2
B. Only the A6 NH
2
protons give a signal at a similar chemical shift, but the very broad line
shape of the amino protons excludes the assignment of the relatively narrow
peak at 6.13 p.p.m. to any other resonance than A7 1'H.
In a B-DNA type structure, the A 2H and the 1'H protons point into the minor groove of a B-DNA structure. The positive NOESY cross-peaks observed with the minor groove resonances of A6
and A7 indicate the presence of hydration water molecules with residence times
>0.5 ns. Therefore, a kinetically stabilized spine of hydration exists in the
minor groove of the 5'-dTTAA segment. The lower intensities of the water-1'H NOEs compared with the water-2H NOEs correlate with the longer
1
H-
1
H distances observed in single crystal structures between the water protons of
the spine of hydration and the 1' protons than between the water protons and the adenine 2 protons (
9
). Rapid hydration water exchange on a sub-nanosecond timescale is indicated near those protons for which the NOE-NOESY spectrum shows negative NOE peaks with the water signal. This includes all base and methyl protons pointing towards the major groove of the DNA, but
also some minor groove protons, like A3 2H, A3 1'H and T8 1'H (Figs
3
B and
4
B). The kinetically stabilized spine of hydration is thus strictly confined to the central 5'-dTTAA segment.
Like in d(GCATTAATGC)
2
, many of the water-DNA NOE cross-peaks in the NOESY and ROESY spectra of decamer 1 and decamer 2
were obscured by intense exchange cross-peaks (data not shown). Because of the structure stabilizing effect of the
hexaethyleneglycol linker in decamer 1 and decamer 2, the exchange peak of the
imino proton of G11 was much weaker than the exchange peak of the imino proton of G1. Only small exchange peaks were observed with the imino protons of all other nucleotides.
NOE-NOESY and ROE-NOESY spectra were recorded to unveil the NOEs with the adenine 2
protons in the minor groove. Figure
5
shows the low field regions of the diagonal cross-sections through the NOE-NOESY and ROE-NOESY spectra of decamers 1 (Fig.
5
A and B) and 2 (Fig.
5
C and D). All adenine 2 protons give intense negative cross-peaks in the ROE-NOESY experiments (Fig.
5
A and C), whereas positive, negative and vanishing NOE intensities are observed
for the same protons in the diagonal cross-sections through the corresponding NOE-NOESY spectra (Fig.
5
B and D). With few exceptions, all A 2H resonances are resolved. The exceptions
are the signals of A7 2H and A17 2H in decamer 1, which partially overlap with
the resonance of C10 5H, and the A17 2H resonance in decamer 2, which overlaps
with the signals of T5 6H, T15 6H and C20 6H.
There are not many DNA protons in the minor groove which can be used as probes
for the presence and kinetic stability of a spine of hydration. Adenine 2H
protons are the only base protons in the minor groove for which water-DNA NOEs can readily be measured. Furthermore, the 1' protons of the deoxyribose moieties are in the minor groove at a similar distance from the water
molecules of the spine of hydration as the A 2H protons. The 1' protons can be used to report on the spine of hydration in GC-rich DNA segments. While many of the A 2H resonances may be
resolved in conventional NOESY and ROESY spectra, intense chemical exchange
peaks with hydroxyl and amino protons tend to obscure the NOEs with some of the
A 2H and almost all of the 1'H signals. In principle, the required separation of exchange cross-peaks and NOEs could be achieved by homonuclear three-dimensional NOESY-NOESY or ROESY-NOESY experiments. In the present work, more sensitive two-dimensional experiments, NOE-NOESY and ROE-NOESY, were used for this
purpose. The sensitivity of these experiments is sufficient to obtain diagonal
cross-sections with acceptable signal-to-noise ratios, where the water-DNA NOEs can be observed without overlap from intense
exchange peaks. Off-diagonal peaks, which help in the assignment of overlapping water-DNA cross-peaks, are observable at higher sample concentrations.
The sign of the water-DNA NOEs offers a straightforward criterium for fast or slow exchange of
the hydration water molecules near the observed DNA protons with respect to a
time scale of ~0.5 ns (
2
). Strictly speaking, the residence times deduced from the sign of the
intermolecular NOEs refer only to protons. Yet, the lifetime of water molecules
with respect to proton dissociation is by orders of magnitude longer than 1 ns
(
28
), so that the modulation of the intermolecular proton-proton vector must come about by the relative motions of the DNA and
entire water molecules (
1
).
The precise values of the water residence times are, however, difficult to
assess from the water-DNA NOEs alone. Only grossly oversimplifying relaxation models are
available (
2
). Based on a model of intermolecular diffusion (
29
), water-DNA cross-peaks with a positive sign in NOESY and of comparable intensity in
ROESY were usually interpreted as indicating hydration water with residence
times >1 ns (
1
,
3
-
8
). NOEs which were present in ROESY but absent in NOESY were interpreted by
hydration lifetimes of ~0.5 ns or by significantly increased mobility of the water molecules while
they are bound at their hydration sites (
4
,
7
). Recent
17
O and
2
H relaxation dispersion measurements indicate, however, that even the hydration
water molecules with the longest residence times exchange within 2 ns at 4oC (
30
). To account for these results, we now attribute residence times >0.5 ns to hydration water detected by sizeable positive water-DNA cross-peaks in NOESY. In the limit of very long residence times, rigidly
bound water would be expected to result in 2-fold faster cross-relaxation rates in ROESY than in NOESY. This prediction is
independent of the model used. Experimentally, the positive NOESY cross-peaks observed in the present study were consistently weaker than their ROESY counterparts, even though 2-fold longer mixing times had been used in NOESY than in ROESY (Figs
3
-
5
). This observation supports the finding that the water molecules of the spine
of hydration in the minor groove of B-DNA are characterized by much shorter residence times and/or smaller order
parameters than water molecules in the interior of proteins (
2
,
31
-
33
).
The data obtained in the present study show that the residence times of the
hydration water molecules in the minor groove depend on the nucleotide sequence
in a more complicated way than previously thought. A kinetically more
restrained spine of hydration water is indicated for the 5'-dTTAA segment in d(GCATTAATGC)
2
, whereas no positive water-A 2H NOEs had been observed for the corresponding segment in
d(GTGGTTAACCAC)
2
(
7
). Clearly, the sequence 5'-dTTAA alone does not determine the mobility of the hydration water
in the minor groove even in the center of the TTAA segment.
All DNA fragments studied here showed negative NOE-NOESY peaks for the A 2H resonances of the adenines next to GC base pairs, i.e.
any kinetic stabilization of the minor groove hydration seems to be confined to
the central part of the AT base pair segments. This situation differs from 5'-dAATT-containing DNA fragments, where stable spines of hydration
are also observed when the 5'-dAATT segment is immediately preceded and followed by GC base pairs
(
2
-
4
,
7
). Based on the observation of negative NOE-NOESY peaks with the G 1'H and C 1'H resonances, the minor groove of GC base pairs does not contain
kinetically stabilized hydration water.
The sequences of decamers 1 and 2 were chosen to find out how the water
residence times in the minor groove of a 5'-dTTAA segment are influenced by different nucleotides following the 5'-dTTAA segment. More rigid minor groove hydration is
indicated by positive NOESY peaks with all but one A 2H resonance for decamer 1, which
contains the sequences 5'-dTTAAC and 5'-dTTAAT (Fig.
5
B). For decamer 2, which contains the sequences 5'-dTTAAA and 5'-dTTAAG, the NOE-NOESY experiment yielded positive cross-peak intensity only for the central
adenine in the 5'-dTTAAA sequence (Fig.
5
D).
The picture is complicated by long range sequence effects, which are clearly
manifested in a comparison between the water-DNA NOEs of d(GCATTAATGC)
2
and decamer 1. Both nucleotide sequences are identical except for the base pair
at position 8. Yet, the NOESY cross-peak intensity with the A 2H at position 4 [A7 in d(GCATTAATGC)
2
and A17 in decamer 1 respectively] is quite different (Figs
4
and
5
B). Apparently, these water-DNA NOEs report on conformational differences caused by the base pair
substitution four positions further towards the 3'-end. The presence of conformational differences in the minor groove
is confirmed by the observation of, for example, a well-observable interstrand NOE between T5 1'H and A6 2H in d(GCATTAATGC)
2
at 15oC which was clearly absent in the NOESY spectrum of decamer 1 recorded
under the same conditions (data not shown). Furthermore, the
1
H chemical shifts are identical within experimental error for the first 4 bp,
but chemical shift differences of 0.08-0.16 p.p.m. were observed for the A 2H and 6NH
2
and the T 1'H and 2'H, 2''H resonances of the base pairs at position 5. This
indicates that the conformational differences start at base pair 5 and that the
water molecule detected by the NOE with the A 2H of base pair 4 is located
between base pairs 4 and 5, in agreement with single crystal X-ray data, where the innermost water molecules of the spine of hydration
tend to be between neighboring base pairs (
34
-
36
). Because the NOEs with the A 2H at position 4 are significantly weaker than
the corresponding ROEs in both DNA fragments, this hydration water must be
quite mobile on a sub-nanosecond timescale. In this time regime, a change in residence time by a
few hundred picoseconds is sufficient to explain the observed differences in
NOE intensities (
1
). Therefore, the water-DNA NOEs can report on small conformational differences in the minor groove with a sensitivity comparable with that of chemical shifts.
Although it is plausible that hydration water in the minor groove is immobilized
most when the minor groove is narrow, it is practically impossible to determine
the width of the minor groove with good accuracy by high resolution NMR
experiments (
37
). In addition, the number of single crystal structures solved by X-ray crystallography is too small to predict the width of the minor grooves
of the DNA fragments studied here. X-ray analyses showed a well-defined spine of hydration associated with a narrow minor groove in
d(CCATTAATGG)
2
, but not in d(CGATTAATCG)
2
(
10
,
11
). The differences in minor groove width for the central 5'-dTTAA segments were ascribed to crystal packing forces (
10
). Still, the spine of hydration in d(CCATTAATGG)
2
correlates closely with the present observation of relatively long hydration lifetimes in the minor groove of d(GCATTAATGC)
2
. Assuming a close correlation between minor groove width and residence times of the hydration water in the minor groove, our data would indicate wide minor grooves in GC-rich nucleotide sequences and narrow minor grooves in 5'-dAATT segments, while the width of the minor groove in 5'-dTTAA segments would depend on the flanking
nucleotide sequences. However, a much larger number of nucleotide sequences
needs to be studied to confirm this putative connection between hydration
lifetimes, minor groove width and nucleotide sequence, since the hydration lifetimes, like the minor groove widths, result from the cumulative effect of interactions between more than just four or five sequentially
neighboring base pairs.
The authors thank Bruker (Billerica, USA) for the generous loan of a probe head
with a Q-switch and Dr B.Halle for critical comments on the manuscript. Financial
support from the Swedish Natural Science Council (Project 10161) and from the Wallenberg Foundation is gratefully acknowledged.
+
Present address: Gymnasium Liestal, Department of Chemistry, CH-4410 Liestal, Switzerland
REFERENCES
Return