supplementary results: surface plasmon resonance (spr)...supplementary results: surface plasmon...
TRANSCRIPT
Supplementary results: Surface Plasmon Resonance (SPR)
Experimental details:
For SPR experiments, the K1 sequence was ordered biotinylated on the 3’ end from Integrated DNA
Technology (Leuven, Belgium) with standard desalting. Prior to SPR experiments, ESI‐MS analysis was
performed to check the purity of the oligonucleotides provided.
The experiments were done on a BIAcore T200 (GE Healthcare, Uppsala, Sweden). The ESI‐MS buffer
(150 mM NH4OAc) was used for all experiments containing 0, 1 or 3 mM Mg(OAc)2.
Hairpin K1 was diluted at 1 µM and then folded (65°C for 5 min, 4°C for 1 min, the rest of the procedure
at room temperature). K1 was then diluted at 50 nM and immobilized at a flow rate of 5 µL/min on a
streptavidin sensorchip (GE Healthcare, Uppsala, Sweden) to reach around 250 RU. All channels were
grafted in the same way. All complementary hairpins i.e. UC, CC, CU and UU, were diluted from 4 µM
to 31.25 nM. Multi‐cycle kinetics of UC, CC, CU and UU were carried out in the appropriate buffer at a
flow rate of 20 µL/min. Regeneration was achieved with a 20 µL pulse of 3 mM EDTA, followed by one
20‐µL pulse of distilled water and finally several pulses of the running buffer.
The global fitting (Figure S1) was done using the BIAeval software 3.1. The model used was 1:1, plus
the bulk signal (RI) which can happen when the refractive index of the analyte solution and the flow
buffer are different.
To obtain the Req, we had to extrapolate the association curves to an infinite time. The association
parts of the sensorgrams were analyzed with Table curve 2D v5.01 (Systat Software Inc.), and fitted
with two first‐order formation kinetics (equation: 1 1 . The total
extrapolated Req was obtained by summing the fitting parameters A+B+D.
The graphs of Req as a function of [ligand] were fitting was done using a 1:1 Langmuir binding model
(Equation: ), using Sigmaplot v12.5 (Systat Software Inc.).
Results:
Supplementary Figure S1. SPR results. Sensorgrams obtained for the titration of hairpin K1 attached
to the sensorship, and each of the four variants of the hairpin K1’ injected in the microfluidics
system. The fits in black were obtained with global fitting with a 1:1 binding model, plus a variable
parameter for a jump of refractive index. The 1:1 binding model could not fit the data appropriately.
The association is characterized by at least two rate constants.
Supplementary Figure S2. KD determination by SPR. Equilibrium dissociation constants obtained by
analyzing the steady‐state of surface plasmon resonance experiments. The steady‐state RU of each
sensorgram was obtained by fitting the association part with two first‐order association rates, and
Req is calculated as the sum of the RUs extrapolated at infinite time. Req is plotted as a function of the
ligand concentration, and fitted by a 1:1 binding model to obtain the KD values indicated on the
figure.
Supplementary results: UV‐melting
Experimental details:
Melting experiments were performed on a UV mc² spectrophotometer (SAFAS, Monte Carlo, Monaco)
equipped with a Peltier temperature controller, in 580‐µL quartz cuvette (Hellma, Müllheim,
Germany). UV melting experiments were performed with K1short and all four complementary hairpins
i.e UC, CC, CU and UU. The solutions were prepared at 1 µM of each complementary hairpin in 150
mM NH4OAc. Three different concentrations of magnesium acetate were used: 0, 800 µM and 3 mM.
The temperature ramp was set at 0.2°C/min from 4°C to 90°C. The absorbance was followed at 260
nm. The results are presented as the first derivative curves of the absorbance by the temperature. Tm
are determined using the maximum of these curves.
Results:
Supplementary Figure S3. UV‐melting. First derivative of the absorbance measured at 260 nm, as a
function of the temperature. The first local maximum corresponds to the disruption of the kissing
complex base pairs. The second local maximum obtained at >50 °C corresponds to the melting of the
individual hairpins. The solutions without magnesium and with 800 µM magnesium were prepared
in the same conditions as for mass spectrometry. The stability ranking is CC > UC > CU > UU (melting
not observed, likely completed below 4°C) in both cases, in line with the mass spectrometry results.
Supplementary results: mass spectrometry
Supplementary Figure S4. ESI‐MS spectra of the full length target K1 (a‐c) and the shorter target
K1short (e‐f), illustrated for the complex UC. Spectra a) and d) are without magnesium (150 mM
NH4OAc only). Spectra b) and e) are with 200 µM Mg(OAc)2. Spectra c) and f) are with 800 µM
Mg(OAc)2. On the full length sequences, the magnesium adducts make it difficult to quantify the free
target K1. With the shorter target K1’short, there is no such interference.
Supplementary results: molecular modeling
Supplementary Figure S5. RMSD over time for the 4 kissing complexes: evolution of the RMSD over
time for each complex in sodium (34 Na+ ions and ~8900 water molecules) for 200 ns (black curves)
and in magnesium (30 Na+ ions, 2 Mg2+ ions and ~8900 water molecules) for 300 ns (red curves). The
mean and the standard deviation are shown on the top of each graph for the respective times in ns.
Supplementary Figure S6. Density of Na+ and Mg2+ ions around the kissing complexes. The receptor
K1 is in grey and ligands K1’ are in cyan. The surface shape of the kissing complex is shown as
transparent. In yellow mesh is the density of the Na+ ions calculated by grid from AmberTools with an
isovalue of 400, and in solid yellow is shown the isovalue of 100. In red mesh is the density of the Mg2+
ions calculated by grid from AmberTools with an isovalue of 50, and in solid red is shown the isovalue
of 150. We observe that the same pockets are sampled by the two types of ions: a main pocket in the
loop‐loop region and secondary pockets next to the major groove of the hairpins and on the exit of
the K1 loop across all four complexes.
Supplementary Figure S7. Opening of the U7•G30 base pair. The system UC is shown with the K1
receptor in gray and the K1’ ligand in cyan. The residues U7 and G30 are labeled. The open conformation
in a) is shown with the U7 positioned outside the loop‐loop interaction and in b) the close conformation
with U7 base paired with G30. For checking convergence the MDs were extended to 400 ns each. We
followed the average distance of the two hydrogen bonds between U7 and G30 in sodium (black) and
magnesium (red) environment and plotted it as a time series in c).
Supplementary Figure S8. Hydrogen bond network in sodium 50 to 200 ns. The target K1 on the
right hand side of each complex contains residues 19 to 36. The ligands K1’ contain residues 1 to 18.
The kissing complexes are named according to the identity of their bases in positions 7 and 12,
respectively. For example, in panel (a), the kissing complex UC contains a uracil (U) in position 7 and
a cytosine (C) in position 12. The four panels show the hydrogen bond network obtained by MD for
each complex in sodium ions (from 50 to 200 ns). The colors are proportional to the presence or
absence of the bond, from blue 0% (absence) to red 100% (present). The residues differing in the
four kissing complexes in position 7 and 12 are in bold and underlined. The hydrogen bonds between
two bases are represented by solid lines. The dashed lines represent the hydrogen bond between
the ribose and the phosphate groups.
Supplementary Figure S9. Hydrogen bond network between the two hairpins. The hydrogen bond
network between the two hairpins (K1 in gray and K1’ in cyan) is shown. In licorice is represented
the residues involved in the hydrogen bond: U7, C6 from the ligands K1’ and G24, G25 from the
receptor K1. The network is shown as dotted black line between the ribose of the residues G24 and
C6 and the phosphate group of the residues G25 and U7.
Supplementary Figure S10. Base slide in the MD in magnesium. The helicoidal parameters were
calculated on the linearized sequence shown in a). The steps which yield to a change along the 4
systems are shown in 3D structure in b) the steps G30pC31.G24pN7 (N for Cytosine or Uracil) and
the step C31pG32.C23pG24 and in c) the steps G25pC26.G11pN12 (N for Cytosine or Uracil) and the
step C27pG26.G10pG11 and their corresponding twist value (in degrees) as a boxplot in d) and e)
respectively. The gray line in the boxplots indicates the canonical value of the slide parameter (‐1.5°)
in an A‐helix. In f) is shown a schematic representation of two base pairs representing a slide along
the y‐axis.
Supplementary Figure S11. Inter‐base pair twist in the MD in sodium. The helicoidal parameter is calculated on the linearized sequence shown in a). The steps which yield to a change across the four kissing complexes are shown in 3D structure in b) the steps G30pC31•G24pN7 (N for cytosine or uracil) and the step C31pG32•C23pG24 and in c) the steps G25pC26•G11pN12 (N for cytosine or uracil) and the step C26pC27•G10pG11 and their corresponding twist value (in degrees) as a boxplot in d) and e) respectively. The gray line in the boxplots indicates the canonical value of the twist parameter (32°) in an A‐helix. In f) is shown a schematic representation of two base pairs adopting a relative twist around the z‐axis.
Supplementary Figure S12. Base slide in the MD in sodium. The helicoidal parameters were
calculated on the linearized sequence shown in a). The steps which yield to a change along the 4
systems are shown in 3D structure in b) the steps G30pC31.G24pN7 (N for Cytosine or Uracil) and
the step C31pG32.C23pG24 and in c) the steps G25pC26.G11pN12 (N for Cytosine or Uracil) and the
step C27pG26.G10pG11 and their corresponding twist value (in degrees) as a boxplot in d) and e)
respectively. The gray line in the boxplots indicates the canonical value of the slide parameter (‐1.5°)
in an A‐helix. In f) is shown a schematic representation of two base pairs representing a slide along
the y‐axis.
Supplementary Figure S13. Energy decomposition per residue for MDs in Sodium. The
decomposition per‐residue in kcal/mol along the sequence of the kissing complex is shown, where
N7 and N12 can be either a uracil or a cytosine for a window of 50 to 200 ns. The symbols; triangle,
cross, dot and star, and the blue, orange, green, red lines represent the UC, CC, CU and UU systems,
respectively. The residues that show different energy are labeled on the graph.
Table S1: MMPBSA calculations (data of main text Figure 5a). Thermodynamics calculations on the
four kissing complexes in different environment (with and without magnesium ions). The ΔGassoc
obtained by MMPBSA (kcal/mol), is reported with its relative mean error on different window
frames (for sodium MDs, 50 to 100 ns, 100 to 150 ns , 150 to 200 ns ‐ for the magnesium MDs we
added two windows from 200 to 250 ns and 250 to 300 ns). The global MMPBSA is calculated from
50 to 200 ns and 50 to 300 ns for the MD in sodium and in magnesium respectively.
Name Ions MMPBSA ΔG (kcal/ mol)
50‐100 ns
MMPBSAΔG (kcal/ mol)
100‐150 ns
MMPBSAΔG (kcal/ mol)
150‐200 ns
MMPBSA ΔG (kcal/ mol)
200‐250 ns
MMPBSAΔG (kcal/ mol)
250‐300 ns
MMPBSAΔG (kcal/ mol) global
UC 34 Na+ ‐25.4 ± 0.2 ‐19.7 ± 0.3 ‐20.8 ± 0.3 NA NA ‐23.2 ± 0.2
30 Na+ , 2 Mg2+
‐22.0 ± 0.2 ‐21.3 ± 0.2 ‐21.8 ± 0.3 ‐18 ± 0.3 ‐17.9 ± 0.2 ‐20.2 ± 0.1
CC 34 Na+ ‐21.7 ±0.3 ‐23.5 ± 0.2 ‐23.6 ± 0.3 NA NA ‐22.8 ± 0.2
30 Na+ , 2 Mg2+
‐23 ± 0.3 ‐26.7 ± 0.3 ‐24.7 ± 0.3 ‐26.5 ± 0.2 ‐27.8 ± 0.2 ‐26.3 ± 0.1
CU 34 Na+ ‐18.3 ± 0.2 ‐19.1 ± 0.2 ‐19.6 ± 0.2 NA NA ‐18.3 ± 0.2
30 Na+ , 2 Mg2+
‐18.3 ± 0.2 ‐17.8 ± 0.2 ‐17.8 ± 0.2 ‐18.6 ± 0.2 ‐19.5 ± 0.2 ‐18.3 ± 0.1
UU 34 Na+ ‐13.7 ± 0.3 ‐9.3 ± 0.3 ‐9.8 ± 0.2 ‐17.1 ± 0.2 NA ‐9.3 ± 0.3
30 Na+ , 2 Mg2+
‐14.9 ± 0.3 ‐11 ± 0.3 ‐15.6 ± 0.2 ‐15.4 ± 0.2 ‐14.3 ± 0.2 ‐15.5 ± 0.1