Supplemental information

Yonggang Ke, Yan Liu, Junping Zhang and Hao Yan*

Experimental Methods
Synthesis and purification of DNA. Custom oligonucleotides were purchased from Integrated DNA
Technology (www.idtdna.com) and purified by 10% or 20% denaturing polyacrylamide gel
electrophoresis (PAGE). The concentration of each strand was estimated by measuring OD260.
Formation of the 4-, 8-, 12-helix bundle complexes. Each individual blunt-ended tile was assembled
by mixing a stoichiometric quantity of the strands involved in the tile in 1x TAE/Mg buffer (20 mM
Tris, pH 7.6, 2 mM EDTA, 12.5 mM MgCl2). The final concentration was 1.0 µM for each strand, and
the final volume was 60 µL. The oligo mixtures were cooled slowly from 90 °C to room temperature in
2 L water placed in a styrofoam box over 24 hours to facilitate strand hybridization. Non-denaturing
PAGE gels run in 1xTAE.Mg buffer were used to confirm the assembly of each blunt-ended tiles.
Non-denaturing PAGE and Ferguson analysis. 10 µL of each of the 4-HT, 8-HT and 12-HT DNA
complexes (1 µM) were mixed with 1 µl 10x native gel tracking dye (containing TAE.Mg, 50%
glycerol, and 0.2% each of Bromophenol blue and xylene cyanol FF) and then loaded into 4 Gels
containing 6%, 8%, 10% and 12% acrylamide (19:1 acrylamide:bisacrylamide). The gels were run in
1xTAE.Mg buffer on Amersham Ruby-600 electrophoresis unit at constant current of 30 mA per gel at
20 °C for 6 hours. The gels were stained with Stains-all (Sigma, 0.01% in 40:60 formamide:water),
destained by exposing to room light and scanned by a HP scanner. The migration distances were
measured from the digital pictures.
Melting temperature measurement. 1.4 mL of the 4-HT, 8-HT and 12-HT complexes (no sticky
ends, 0.1 µM of each strands in 1xTAE.Mg buffer) was annealed as described above and the sample
transferred to quartz cuvettes with the 1xTAE.Mg buffer used as a blank. Thermal melting of the
complexes were monitored at 260 nm on a Cary UV-vis spectrometer using built-in peltier temperature
control. The temperature was incremented at 0.1 °C/min.
Formation of the 2D array or DNA tubes. The tiles with the sticky ends are annealed similarly as
the blunt-ended complexes. The self-assembly of the tile complexes into 2D array or closed DNA tubes
occurs at the same annealing conditions.
AFM imaging. Imaging was performed under 1xTAE/Mg in a fluid cell on PicoPlus AFM
(Molecular Imaging) in AAC mode, using the tip on the thinner and shorter cantilever of the NP-S tips
(Veeco Inc.). A piece of freshly cleaved mica (Ted Pella, Inc.) was first assembled as the bottom of the
fluid cell on the sample plate. A 2 µL of 1 mM NiCl2 solution was spotted on mica, and left to adsorb on
the surface for 2 min. Then a 2 µL of the sample was added to the spot and left to adsorb on the surface
for another 2 min. Finally, 400 µL 1xTAE/Mg buffer was added onto the mica in the fluid cell. The
Ni2+ adsorbed on mica surface can help the DNA array stay on the surface in the scanning.
Results
Ferguson analysis. The migration speed of a charged molecule through the pores of a gel depends on
the voltage applied across the gel, the surface area of contact, and the pore size of the gel (or gel
concentration). The surface area of contact is proportional to the friction force that is related to the pore
size of the gel. Under the same gel running conditions, when the only variable is the pore size of the gel,
the slope of the Ferguson plot of log(mobility) as a function of gel concentration is proportional to its
friction constant28, thus yielding information about the surface area of a molecule. For the 4-HT, 8-HT
and 12-HT complexes, they are similar in length, but have a different number of helixes in the tile plane,
therefore their surface areas are nearly in the ratio of 1:2:3 in the planar arrangements. As observed, the
slope of the Ferguson plot (Figure S1A) goes more negative as the helix number increases. The increase

S1
of the friction constant is roughly linear to the number of helices in the complex (Figure S1B). This
suggests similar conformation of the three individual tiles.

Figure S1. Ferguson analysis. (A) Ferguson plots. Log(mobility) as a function of polyacrylamide
concentration is shown for 4-HT (black squares), for 8-HT (red circles) and 12-HT (green triangles).
(B) Analysis of the Ferguson slopes. The increase in relative friction constant with increasing numbers
of helical domains in the complex is roughly linear.

Figure S2. Thermal transition profiles. A, C and E show the relative change in optical density at 260
nm as a function of temperature for 4-HT, 8-HT and 12-HT, respectively. B, D and F show the
differential melting behavior of 4-HT, 8-HT and 12-HT, respectively.
Thermal transition profiles. Melting curves of DNA complexes provide a measure of stability and
cooperativity of the internal interactions indicated by the temperature at the transition midpoint and the
width of the transition. A higher transition temperature and a narrower width of the transition indicate
that the association enthalpy is higher and the complex has greater thermal stability. Figure S2 shows
the thermal transition profiles for the three complexes. It is apparent that the 4-HT has two transitions,
the most profound at ~58±7 °C, and the second at ~ 65 °C. This behavior is similar to the triple
crossover (TX) molecule studied before2, which is in a similar design with one helix narrower than the
4-HT. The melting transitions for the 8-HT and 12-HT complexes show single transitions that occur at
higher temperatures, ~65±5 °C and ~62±8 °C, respectively. The three complexes have similar GC
content (56%, 59% and 58%, respectively). Considering the tile structure design, the longest strand in
the 8-HT and 12-HT that weave though all the helixes, has to be broken up into two strands to be shorter
than 150 nt. The two nicks in the tile may make the tile plane flexible. In addition, the portion of shorter

S2
strands (<26 bases) increases from 33% to 42% from 4-HT to 12-HT. This may also affect the overall
thermal stability of the complex, as the shorter strands may become loose and dissociate from the
complex at lower temperatures.

DNA sequences and strand structures used:

S3
S4
S5
S6
S7
S8
S9
S10