588–593
 1998 Oxford University Press
Nucleic Acids Research, 1998, Vol. 26, No. 2
Surface-directed DNA condensation in the absence of
soluble multivalent cations
Ye Fang and Jan H. Hoh*
Department of Physiology, Johns Hopkins University School of Medicine, 725 North Wolfe Street, Baltimore,
MD 21205, USA
Received September 2, 1997; Revised and Accepted November 18, 1997
ABSTRACT
Multivalent cations are known to condense DNA into
higher ordered structures, including toroids and rods.
Here we report that solid supports treated with
monovalent or multivalent cationic silanes, followed by
removal of soluble molecules, can condense DNA. The
mechanism of this surface-directed condensation
depends on surface-mobile silanes, which are
apparently recruited to the condensation site. The yield
and species of DNA aggregates can be controlled by
selecting the type of functional groups on surfaces,
DNA and salt concentrations. For plasmid DNA, the
toroidal form can represent >70% of adsorbed
structures.
INTRODUCTION
Within living cells and viruses, DNA molecules up to several
centimeters long can adopt highly concentrated, compacted states
with different levels of ordered structures (1). It is known that DNA
condensation can accelerate a number of biological processes, such
as DNA strand-exchange (2,3), ligation (4,5) and renaturation (6)
and catenation of closed circular DNA (7). Recently, cationic
liposomes (8,9) and polypeptides (10) have been applied to
condense DNA into various fine particles, and thus provide
efficient non-viral vectors for gene transfer to eukaryotic cells in
vitro and in vivo (11). The transfection efficiency was generally
parallel to the ability of these cations to condense DNA (9,10).
Thus, in vitro methods for producing and studying condensed
DNA have potential technological applications, and may provide
insight into the molecular mechanisms that control DNA
condensation in biological systems.
Extensive investigations have been made to understand the
physical properties and structure of condensed DNA phases in
vitro during the past several decades (see reviews 12–14). At
highly concentrated aqueous solutions, DNA molecules pack into
various ordered phases depending on the counterion strength and
species (15,16). For diluted DNA solutions, DNA molecules can
condense into toroids and rods, and the highly ordered DNA
condensation requires the presence of soluble multivalent cations
such as spermidine and Co(NH3)63+ (12–14,17–22). The DNA
condensation generally involves multi-molecular assembly for
relatively short DNA molecules, such as plasmid DNA of several
kilobase pairs (12). The condensation pathway of DNA solution
is 3-D, and not yet fully understood (23). On a reduced
dimensional system, DNA condensation might have different
behavior from those in bulk solutions. Recently Balhorn and
colleagues (24) used atomic force microscopy (AFM) (25) to
show that treatment of DNA adsorbed to a surface with a solution
of multivalent cations produced toroids with lateral dimensions
(i.e. diameters) similar to those in solution but with smaller
heights (the axial direction). AFM has also been used to show
another form of 2-D condensed DNA, which forms on supported
cationic membranes, in which molecules are extended but tightly
packed (26–28). Here we describe the formation of condensed
forms of DNA (toroids and rods), in the absence of soluble
multivalent cations, on a surface treated with cationic silanes and
propose that the condensation mechanism involves mobile
surface adsorbed silanes.
MATERIALS AND METHODS
Chemicals
3-Aminopropyl-triethoxysilanes (APTES) with the purities of 98, 96
and 95% were purchased from Sigma Chemical Co. (St. Louis,
MO), United Chemical Technologies (Bristol, PA) and Fluka
Chemical Co. (Milwaukee, WI), respectively. N-(2-aminoethyl)-3-aminopropyl-trimethoxysilane (AEA) of 94% purity and
3-[2-(2-aminoethyl)-ethylamino]-propyl-trimethoxysilane (AEEA)
of 90% purity were purchased from Fluka, and AEEA of 96% purity
was from United Chemical Technologies. All silylating agents were
used as received.
Lambda DNA was obtained from Sigma Chemical Co. (St. Louis,
Mo). A 3850 bp plasmid was purified from a bacterial lysate using
a Qiagen column (Qiagen Inc., Santa Clarita, CA) as described by
the manufacturer. DNA samples were diluted to the appropriate
concentration with deionized water (>18 MΩ, MilliQ-UV, Millipore
Co., Bedford, MA) before use.
Modification of surfaces
Rectangular (3 × 4 mm2) silicon substrates were cut from 4 inch
silicon wafers (Grade EPP, Wafernet Inc., San Jose, CA). The
silicon pieces were rinsed with ethanol, chloroform and water,
and then exposed to 254 nm UV light (UVO-Cleaner, Jelight
*To whom correspondence should be addressed. Tel: +1 410 614 3795; Fax: +1 410 614 3797; Email: [email protected]
589
Nucleic Acids
Acids Research,
Research,1994,
1998,Vol.
Vol.22,
26,No.
No.12
Nucleic
589
Figure 1. Chemical structures of organosilanes used in this work.
Company Inc., Laguna Hills, CA) for 10–20 min. The silicon
substrate is significant variability in the roughness, depending on
the source and batch. Therefore all batches of silicon substrate
were screened by AFM prior to use and only batches with low
surface roughness [<0.1 nm root mean squared (RMS)] were
used. Mica disks (12 mm diameter) were punched from sheets of
muscovite mica (grade V1-V2; Asheville-Schoomaker Mica Co.,
Newport News, VA). The mica was cleaved in air with a piece of
tape. Substrates obtained were used immediately. For silanization,
1% (v/v) of the appropriate silane was added to an aqueous solution
of 1 mM acetic acid (pH 3.5) and allowed to partially hydrolyze
for 5 min. Substrates were dipped vertically into the silane
solution and kept there for 10 min. The substrates were rinsed
under a flow (∼10 ml/s) of deionized water for 10–20 s to remove
all soluble silanes, and dried briefly with compressed air. Unless
specifically mentioned, these silane-coupled substrates were
taped onto metal disks and used the same day. Prior to use, they
were maintained under ambient conditions. To test for tightly
bound silanes, two simple approaches were used. The freshly
silanized substrates were either directly placed into a dissector
and stored for 2 days under vacuum (<1 torr), or placed into an
incubator at 100C for 2 h under ambient pressure. After these
treatments, the substrates were used directly for DNA adsorption
experiments.
DNA adsorption
Because of the poor spreading of DNA solution on the silanized
surfaces, a clean glass coverslip was placed on top of a 20 µl drop
of DNA aqueous solution to make it cover the entire substrate.
After 3 min the coverslip was rinsed off and the sample was
continuously rinsed with deionized water for 10 s and dried with
compressed air. To assess the effect of salt on the structure of
DNA on the surfaces, DNA samples were diluted into NaCl
solutions of appropriate concentration.
AFM imaging
A Nanoscope IIIa controller with multimode atomic force
microscope (Digital Instruments Inc, Santa Barbara, CA), with a
J type scanner (maximum xy range ∼150 × 150 µm) and an
ambient tapping mode cantilever holder, was used. Silicon
cantilevers with resonant frequencies ∼300 kHz were from
Digital Instruments (model TESP). All imaging was in ambient
tapping mode (29), and all images shown are height images. Scan
parameters were optimized for each experiments, but frequencies
were typically 3–5 Hz.
Figure 2. Typical ambient tapping mode AFM images of the 3.85 kb plasmid
DNA condensed on the APTES-Si (A), AEA-Si (B) and AEEA-Si (C), in the
absence of salt. All DNA concentrations were 0.5 µg/ml.
RESULTS
Short chain organosilanes can form covalent bonds with silanols
on oxidized silicon surfaces in the presence of water, and are
known to form polymerized networks (30). The organosilanes
used in this work, APTES, AEA and AEEA, have headgroups
with increasing number of amines (Fig. 1), and thus increasing
positive charge in an aqueous solution at neutral pH. Silicon
surfaces treated with these silanes show a slight increase in
surface roughness, from <0.1 nm RMS roughness for a clean
oxidized silicon surface to ∼0.15 nm RMS for a silanized surface.
However, the purity of the silanes is important. For APTES the
98% pure solution results in a smooth surface, while the 95% pure
solution produces a rougher surface often with small particles,
consistent with previous observations (31).
Silicon surfaces treated with APTES, AEA and AEEA all
condense plasmid DNA into higher ordered structures,
predominantly toroids and rods, during adsorption from an aqueous
590
Nucleic Acids Research, 1998, Vol. 26, No. 2
Figure 3. High magnification gallery of DNA toroids on the silanized silicon
surfaces. Toroids of plasmid DNA formed on AEEA-Si (A–D) and AEA-Si
(E–H) have similar morphologies. However, toroids of lambda DNA (48.5 kb)
formed on DETA-Si have a larger diameter and apparent height (I–K). Scale
bar is the same for all images.
Figure 5. High magnification gallery of DNA rods on the silanized silicon
surfaces. Rods of plasmid DNA formed on APTES-Si (A–D), AEA-Si (E–H)
and AEEA-Si (I–L) have a wide range of dimensions. Rods of lambda DNA
(48.5 kb) formed on AEEA-Si are larger in all dimensions (M–N). Scale bar is
the same for all images.
Figure 4. Dimensions of plasmid DNA toroids on the silanized surfaces. (a) A
representative AFM image of DNA toroid shown in a reverse contrast. (b) A
section plot shows the dimensions measured. D represents to the center-to-center
distance taken at the highest point on the toroid. H is the apparent height of toroid
from the highest point to the substrate. (c) The height distribution shows a
maximum near 2 nm. (d) The distribution of diameters shows a peak near 40 nm.
For each toroid at least six measurements were made to give the average apparent
height of the toroid. The average diameter of each toroid was calculated dividing
the counter length along toroid at the highest point with 3.14.
solution with negligible amounts of salt (<0.1 mM) (Fig. 2). A
3.85 kb plasmid on APTES-Si, produced predominantly long bent
rods and condensed but amorphous structures. The AEA-Si
produced toroids and rods in roughly equal proportions. The rods
in this case were shorter and apparently more condensed than those
for APTES-Si. Plasmid DNA on AEEA-Si resulted in more toroids
(>70%) than rods. Lambda-phage DNA (48.5 kb) also condenses
onto all these surfaces.
The shape and geometries of the toroids formed on different
surfaces are similar for a given DNA (Fig. 3). For the plasmid DNA
the toroids have a diameter (center-to-center) of 45 ± 6 nm (24
measurements) and a height of 2.6 ± 0.6 nm (32 measurements)
(Fig. 4), and have a slightly irregular shape. The height contour
along the toroid, at the highest point, is also irregular. The measured
width of the toroids is tip shape dominated. For lambda DNA the
toroids are larger, a diameter of 90–130 nm and a height of ∼6 nm.
Figure 6. AFM images of DNA molecules on the APTES-Si surface after vacuum
treatment to remove non-covalently bound molecules. The molecules are largely
relaxed, although many are tightly coiled. (A) Lambda DNA (1 µg/ml),
(B) plasmid DNA (1 µg/ml).
591
Nucleic Acids
Acids Research,
Research,1994,
1998,Vol.
Vol.22,
26,No.
No.12
Nucleic
Figure 7. The condensation of lambda DNA on the AP-Si surfaces shown at
low and high DNA concentrations. (A) At 0.5 µg/ml the molecules appear
highly condensed. (B) However, at 5 µg/ml the condensed molecules are
surrounded by disordered and incompletely condensed DNA. This suggests
that the process is saturable at a low surface coverage.
In some cases, the lambda-DNA toroids seemed to spread
(Fig. 3I–K).
The rods are more variable than the toroids, with lengths of
140–180 nm and heights of 2–3 nm for the plasmid used
(Fig. 5A–L). The morphology of the rods is slightly different on
the different silanes, and appear most extended on APTES-Si, and
least extended on AEA-Si and AEEA-Si. Rods formed from
lambda DNA are much larger than the ones formed from
plasmids, with lengths up to ∼400 nm and heights of 3–8 nm.
(Fig. 5M–N).
One possible mechanism for surface-directed DNA condensation
is that some fraction of the silanes do not form covalent bonds with
the substrate, and thus are weakly bound and able to diffuse along
the surface and facilitate condensation. To test this possibility, we
applied two simple approaches to reduce or remove these noncovalent silanes; heat treatment (100C for over 2 h) or vacuum
storage (2 days) of the substrates. Both of these treatments
completely eliminated the condensation of the DNA on ATPES-Si
and AEEA-Si (AEA-Si was not tested). However, following
removal of non-covalent silanes the substrates bound DNA well,
suggesting the continued presence of bound silanes (Fig. 6).
A condensation mechanism that depends on surface mobile
silanes should be saturable prior to complete coverage of the
surface by DNA. Adsorption of lambda DNA onto ATPES-Si at
low (0.5 µg/ml) or high (5 µg/ml) concentrations shows that the
higher DNA concentration results in incomplete condensation,
591
Figure 8. High concentrations of NaCl inhibit condensation and also affects the
dimensions of toroids formed. (A) Lambda DNA condensed on a AEEA-Si
surface in the presence of 250 mM shows incomplete toroid formation. (B–D)
show the toroids of plasmid DNA on AEA-Si in the presence of 100 mM NaCl,
while (E–G) present the toroids of the circular DNA on AEEA-Si in the
presence of 200 mM NaCl. All DNA concentrations were 1 µg/ml. Scale bar
is the same for (B–G).
indicating saturation while the surface coverage is still relatively
low (Fig. 7).
The concentration of NaCl in DNA solution can also affect the
DNA condensation on these silane modified surfaces. For lambda
DNA on the AEEA-Si, the presence of 250 mM NaCl results in
incomplete condensation (Fig. 8A). The effect of NaCl on
surface-directed condensation also depends on the silane types.
For a plasmid DNA there is detectable inhibition of condensation
at 100–200 mM NaCl for AEEA-Si, and 50–100 mM NaCl for
AEA-Si. In the case of toroids, the partial condensation results in
a larger diameter (Fig. 8B–G). However, completely relaxed
DNA is not seen on any of these substrates at NaCl concentrations
up to 500 mM. It is interesting to note that some of the partially
condensed toroids, particularly Figure 8C, have a nodular
appearance. This may indicate the presence of a late intermediate
during the condensation, or structural subunit of the condensed
DNA.
DISCUSSION
DNA molecules are known to adsorb well onto mica surfaces in the
presence of divalent cations (32), but adsorb poorly or not at all onto
UV-cleaned silicon surfaces even in the presence of divalent cations
such as Mg2+ and Ni2+. Mica surfaces modified with cationic silanes
are also known to bind DNA, and have been used to immobilize
DNA for AFM imaging (31). However, condensation of DNA in
these cases was rare (31; our data not shown). Balhorn and
co-workers (24) have demonstrated condensation of DNA adsorbed
to a surface in the presence of soluble multivalent cations
(protamine). However, pretreatment of the substrate with the
592
Nucleic Acids Research, 1998, Vol. 26, No. 2
protamine prior to DNA adsorption only resulted in partial
condensation of some molecules. In solution, condensed forms of
DNA such as toroids and rods have been shown to form under a
wide range of conditions, but to our knowledge always require the
presence of a soluble multivalent cation (11–13). The surfacedirected DNA condensation described here is unique in that it does
not require such soluble cations. The silanes used are cationic, but
following adsorption to the surface, soluble molecules are removed
by vigorous rising of the surface and air drying. Further, any
soluble residues that might remain on the surface after rinsing
would be diluted into a large volume upon addition of the DNA
solution, and thus have a very low concentration. The silanes used
have cationic valences, assuming complete ionization, from one
for APTES to three for AEEA. However, because the silanes can
oligomerize, the exact nature of the molecules on the surface is not
known.
Surface-directed condensation of the DNA using cationic
silanes is highly effective, and the fraction of condensed
molecules on a surface is often >95%. There is a range of
morphologies represented, including rods of varied length and
toroids. The diameters of the toroids formed from plasmid DNA
(∼45 nm) are very close to those formed in solution (11–12).
However, on a silanized surface lambda DNA forms toroids that
are 90–130 nm in diameter, much larger than any that have
previously been described. Further, the heights (the axial
direction) of the toroids on a surface is much smaller, 2–6 nm for
the DNA’s used here, than in solution, where they are ∼25 nm
(11–12). Because the cationic head groups on the silanes are
bound to the surface there are no such groups in solution to
support growth in the axial direction (away from the surface).
Hence the increased radii of the toroids formed from lambda
DNA suggests that the interaction between the DNA and the
cationic groups is either required for toroid growth, i.e. it is not
enough to just nucleate the growth, or more generally that the
energetics of the DNA-cation interaction dominate the energetic
factors that favor axial growth (such as minimizing solvent
contact area).
The mechanism of surface-directed DNA condensation by
cationic silanes appears to involve mobile surface adsorbed
molecules (Fig. 9). In such a mechanism, DNA molecules adsorb
to the surface by interacting with the head groups of the cationic
silanes. This DNA molecule then recruits surface mobile silanes,
by trapping silanes or binding silanes as the DNA molecule
moves, and a 2-D condensation follows. As noted above, DNA
does not condense onto untreated silicon surface, and the
concentration of any cationic species in solution is negligible after
rinsing, drying and rehydration. Hence the presence of cationic
silanes at the surface is required. The mobility of the silanes is
suggested by two lines of evidence. First, curing the substrates at
high temperature is a common step in silanizing surfaces, which
drives the formation of covalent bonds with the silicon oxide.
This reduces the number of ‘free’ silanes. Vacuum treatment
would be expected to have the same effect by causing the small
molecular weight organic-silanes to evaporate from the surface.
That these two treatments eliminate the condensation of DNA
onto a surface, but bind relaxed DNA well, is consistent with the
presence of non-covalent silanes at the surface. Second, the
condensation is saturable at high DNA concentrations and low
surface coverage. Therefore there is a limiting reagent that does
not remain uniformly distributed over the surface. In the context
of the mechanism proposed here, it is interesting to note that
Figure 9. The proposed mechanism is based on the surface diffusion of cationic
silanes. A population of the silanes covalently attach to the silicon oxide surface
through conventional silane chemistry, while another population loses its
reactivity by the formation of tri-silanol groups. The silanes are shown
schematically as a circle with plus signs to represent the cationic head group,
a zigzag line to represent the alkane chain and a terminal silicon. The tri-silanol
terminated molecules adsorb to the hydrophobic surface, through the alkane
chains on the silane, and can readily diffuse along the surface. However, the
hydrophobic interaction with the surface prevents desorption into the bulk
solution at any appreciable rate. When molecules do desorb, their concentration
in the large volume of solution is negligible. DNA initially becomes attached
to the surface at a fixed or mobile silane. Diffusing silanes then are captured by
the DNA by a diffusion limited process, and toroids and other higher order
structures form by mechanisms similar to those that occur in solution.
diffusion of cations adsorbed to the DNA surface has also been
proposed as a critical step in the formation of electrostatic
correlation forces, which result in an attractive force between two
DNA helices that is critical to condensation (33).
A question with regard to the surface-mobile-silane mechanism
is how the silanes remain adsorbed to the surface during an
aqueous rinse, but not so tightly bound that they do not move
laterally on the surface. One possibility is that the hydrophobic
portions of the silanes interact with hydrophobic surface while the
headgroup is solvated. Despite the positive charge on these
silanes, they all produce surfaces that are highly hydrophobic
based on contact angle and liquid spreading measurements (not
shown). Hence, a hydrophobic interaction between the substrate
and the silane would allow lateral mobility along hydrophobic
regions while preventing desorption into an aqueous solution
(Fig. 9). This would suggest an approach to produce other
reagents that are capable of facilitating surface-directed DNA
condensation. Further, such a mechanism would explain why
pretreatment of a surface with protamine results in poor
condensation (24); there would not be a strong interaction holding
the protamine close to the surface that allowed lateral movement.
Therefore protamine molecules would either desorb when the
protamine containing solution was removed , or if it bound, would
bind to a point, preventing lateral movement.
The condensation of DNA on cationic silane surfaces is
controlled by two of the parameters studied here, the valency of
the silane and the NaCl concentration in the aqueous phase. The
valency of the silane affects the morphology and relative yields
of rods versus toroids. However, the valency is linked to
increasing hydrocarbon content in the molecules and thus the
control of morphology may be more complicated than just
valency. High concentrations of NaCl appear to inhibit the
593
Nucleic Acids
Acids Research,
Research,1994,
1998,Vol.
Vol.22,
26,No.
No.12
Nucleic
condensation process, resulting in partially condensed structures.
The effect of NaCl was also dependent on the type of silane used.
CONCLUSION
We have shown that silicon oxide surfaces treated with cationic
organic-silanes can efficiently condense DNA into higher ordered
structures, such as toroids and rods, in the absence of soluble
cationic silanes. These structures appear to form by a mechanism
that involves the adsorption of a DNA molecule to the surface,
and recruitment of surface mobile silanes, to form a condensed
structure. The morphology and assembly of the condensed
structures can be controlled by varying the silanes used and the
ionic composition of the solutions. These results offer insight into
how DNA condenses into higher ordered structure, and suggest
ways to form novel DNA structures.
ACKNOWLEDGMENTS
We thank Dr Roger Reeves for providing the plasmid used here.
This work was supported by NIH grant HGO1518.
REFERENCES
1
2
3
4
van Holde,K.E. (1988) Chromatin, Springer-Verlag, New York.
Camerini-Otero,R.D. and Hsieh,P. (1993) Cell, 73, 217–223.
Sikorav,J.-L. and Church,G.M. (1991) J. Mol. Biol., 222, 1085–1108.
Pheiffer,B.H. and Zimmerman,S.B. (1983) Nucleic Acids Res., 11,
7853–7871.
5 Zimmerman,S.B. and Harrison,B. (1987) Proc. Natl. Acad. Sci. USA, 84,
1871–1875.
6 Sikorav,J.L. and Church,G.M. (1991) J. Mol. Biol., 222, 1085–1108.
7 Krasnow,M.A. and Cozzarelli,N.R. (1982) J. Biol. Chem., 257, 2687–2693.
593
8 Felgner,P.L., Gadek,T.R., Holm,M., Roman,R., Chan,H.W., Wenz,M.,
Northrop,J.P., Ringold,G.M. and Danielsen,M. (1987) Proc. Natl. Acad.
Sci. USA, 84, 7413–7417.
9 Radler,J.O., Koltover,I., Salditt,T. and Safinya,C.R. (1997) Science, 275,
810–814.
10 Niidome,T., Ohmori,N., Ichinose,A., Wada,A., Mihara,H., Hirayama,T.
and Aoyagi H. (1997) J. Biol. Chem., 272, 15307–15312.
11 Lasic,D. (1997) Liposomes in Gene Delivery, CRC Press LLC., Boca
Raton, Florida.
12 Bloomfied,V.A. (1991) Biopolymers, 31, 1471–1481.
13 Bloomfied,V.A. (1996) Curr. Opin. Struct. Biol., 6, 331–341.
14 Marquet,R. and Houssier,C. (1991) J. Biomol. Struct. Dyn., 9, 159–166.
15 Livolant,F. (1991). Physica A , 176, 117–137.
16 Podgornik,R., Strey,H.H., Gawrisch,K., Rau,D.C., Rupprecht,A. and
Parsegian,V.A. (1996) Proc. Natl. Acad. Sci. USA, 93, 4261–4266.
17 Chattoraj,D.K., Gosule,L.C. and Schellman,J.A. (1978) J. Mol. Biol., 121,
327–337.
18 Gosule,L.C. and Schellman,J.A. (1976) Nature, 259, 333–336.
19 Manning,G.S. (1978) Q. Rev. Biophys., 2, 179–246.
20 Schellaman,J.A. and Parthasarathy,N. (1984) J. Mol. Biol., 175, 313–329.
21 Widom,J. and Baldwin,R.L. (1983) Biopolymers, 22, 1595–1620.
22 Wilson,R.W. and Bloomfiled,V.A. (1979) Biochemistry, 18, 2192–2196.
23 Yoshikawa,K., Yoshikawa,Y., Koyama,Y. and Kanbe,T. (1997) J. Am.
Chem. Soc., 119, 6437–6477.
24 Allen,M.J., Bradburyl,E.M. and Balhorn,R. (1997) Nucleic Acids Res., 25,
2221–2226.
25 Binnig, G., Quate, C.F. and Gerber, C.H. (1986) Phys. Rev. Lett., 56,
930–933.
26 Fang,Y. and Yang, J. (1997) J. Phys. Chem. B, 101, 441–449.
27 Fang,Y. and Yang, J. (1997) J. Phys. Chem. B, 101, 3453–3456.
28 MouJ., Czajkowsky,D.M., Zhang,Y. and Shao,Z. (1995) FEBS Lett., 371,
279–282.
29 Hansma,P.K., Cleveland,J.P., Radmacher,M., Walters,D.A., Hillner P.E.,
Bezanilla M., Fritz M., Vie D., Hansma H.G., Prater C.B., et al. (1994)
Appl. Phys. Lett., 64, 1738–1740.
30 Sagiv,J. (1980) J. Am. Chem. Soc., 102, 92–98.
31 Lyubchenko,Y.L., Gall,A.A., Shlyakhtenko,L., Harrington,R.E., Jacobs,B.L.,
Oden,P.I. and Lindsay,S. (1992) J. Biomol. Struct. Dyn., 10, 589–606.
32 Hansma,H.G. and Laney,D.E. (1996) Biophys. J., 70, 1933–1939.
33 Rouzina,I. and Bloomfield,V.A. (1996) J. Phys. Chem., 100, 9977–9989.