I (0-0) No phosphate (M)-NDPA-AMP (M)-NDPA-ADP (P)-NDPA-ATP I (0-1)

Transcription

1 a) No phosphate b) CD (mdeg) (M)-NDPA-AMP (M)-NDPA-ADP (P)-NDPA-ATP Absorbance (norm.) I (0-1) I (0-0) No phosphate (M)-NDPA-AMP (M)-NDPA-ADP (P)-NDPA-ATP Supplementary Figure 1 Adenosine phosphates induced helical assembly of NDPA. a, CD and b, normalized absorption spectra of NDPA assemblies with various bound adenosine phosphates (5 x 10-5 M in aq. HEPES buffer). Normalized absorption spectra shows that binding of ATP or ADP leads to strong interchromophoric interactions among NDPA, AMP binding leads to weaker aggregates as evident from the I (0-0) /I (0-1) value AMP = 1.07, ADP = 0.95, ATP = 0.91.

2 C2 I Supplementary Figure 2 Conformations of NDPA. Two conformations of NDPA (C2 and I) shown with the NDPA plane perpendicular to the view (note that two chloride ions, linked to zinc, were added to have neutral molecules). Right: schematic representation of a stack of such molecules as viewed from the top, with a rotation angle of 60 : long axes of NDPA in black and red arrows show with which zinc atoms, zinc atom 1 has shortest distances: successively 2, 3 and 4. Two conformations of the NDPA molecule have been studied: the so-called C2 conformation, which possesses an axis of order 2, and the so-called I conformation, which possesses an inversion center (Supplementary Figure 2). The potential energy difference between both structures is negligible, around 0.1 kcal mol -1 in favor of I.

3 Supplementary Figure 3 Potential energy components. Evolution of the potential energy components per molecule in the NDPA-AMP systems during the MD run (block average over ten structures); for the four energy components, the energy scale spans 10 kcal, to easily compare the energy fluctuations. See Supplementary Note 1 and 2 for detailed explanation.

4 M P Supplementary Figure 4 MD snapshots of NDPA-AMP assemblies. Snapshots at the end of the MD run for M- helix (left) and P-helix (right) of NDPA-AMP assemblies (from C2 conformation). The Zn atoms are shown in purple balls. The difference in potential energy between these structures is around 2 kcal per mol per molecule in favor of the left-handed (M) assembly. The detailed atomistic structure analyses provided further insights into the differences in stability for the left-handed and right-handed helices. Two assemblies with opposite helicity are shown in this figure (last snapshot of MD run), both issued from C2 conformation: M corresponds to C2-L-1, and P to C2- R-2. From the potential energy estimates, these two structures possess a total energy difference of more than 2.1 kcal per mol per molecule, mostly originating from van der Waals interactions (see dark green line and light blue line in Supplementary Figure 3). As supported by the H-bond analysis, the left-handed assembly shows a higher global order than the respective right-handed assembly. On the basis of the modelled structures, several structural parameters have been computed: the average distances and the angles between neighbouring NDPA fragments and the so-called rotation strength, which is an estimate of the local chirality, see Supplementary Table 3. While the average angle between successive NDPA long axes are identical for the two structures (with opposite signs), i.e. around 56, the distances between successive NDPA differ: around 4.0 Å for M-helix while it is around 4.4 Å for the P-helix. Moreover, the deviations (deviation along the NDPA stack) are larger in the case of the P-helix, which is in line with the apparently more disordered global structure for P than for M, as depicted in this figure. The rotation strength provides direct information of the local chirality. The smaller deviation in the M-helix (Supplementary Table 3) further illustrates the differences in global order.

5 M P Supplementary Figure 5 MD snapshots of NDPA-AMP assemblies. Details of the inner structure of a snapshot at the end of MD for left-handed (left) and right-handed (right) NDPA-AMP assemblies, as extracted from a short stack in the middle of a snapshot at the end of MD. The Zn atoms are depicted in balls, with different color code depending on which side of the NDPA they are located, to show the chirality of the assembly. NDPA are depicted in thick sticks and H-bonds are shown in dashed light blue lines. The arrows guide the eye to show the chirality, the arrow pointing towards the back of the view. In the P-helix, a NDPA stacking defect appears in the middle of the stack. However, for the M-helix, the stacking of the NDPA is regular, the NDPA planes are all parallel. This difference in local order for the two types of helices can a priori be attributed to the intrinsic chirality of the ribose (D), which stabilizes more strongly the M-helix in the case of a single point (one phosphate) attachment. In the P- helix, the ribose moieties are oriented at the periphery of the stack, outside the tube formed by the Zn ions. However, in the M-helix they are within the tube formed by the Zn atoms (see Supplementary Figure 4, M versus P). They are closer to the NDPA cores, with which they can form a relatively high number of H-bonds. As the orientation of the ribose moieties in the M-helix make narrower tubes and improves the order of the molecules, partly thanks to hydrogen bonds with NDPA, the average number of - contacts per molecule is higher in M than in P (17.6 for M versus 15.3 for P). These results are in line with the more stabilizing van der Waals interactions for the M-helix compared to the P-helix.

6 Supplementary Figure 6 MD snapshots of NDPA-ADP assembly: (M)-NDPA-ADP helix displayed with a depth cue intensity (right) and without a depth cue intensity (left). The arrow indicates the location of a ribose moiety, whose hydroxyl groups point to the oxygen atoms of naphthalenediimide cores. With ADP, there are two grafting points. To evaluate which pairs of complexes graft to ADP, we started from C2 AMP-substituted helices and noticed that the closest grafting points belong to adjacent molecules (1-2 motif). An ADP substituted M-helix and P-helix were then built and submitted to MM and MD simulations. Similar observations to AMP were also made with NDPA-ADP, both in terms of energy and morphology between the left-handed and right-handed helices, although the amplitude of the differences is smaller. The M- helix is slightly more stable than the P-helix, by 0.5 kcal per mol per molecule on average in the last 10 ns of MD, and the average percentage of oxygen atoms of NDPA involved in hydrogen bonding with hydroxyl groups of adenosine is 27% in the M-helix and 21% in the P-helix. The rotation angle between successive NDPA-ADP is around 58 ± 10, similar to what is found for NDPA-AMP (57 ± 11 ), i.e. around.3 NDPA per turn for both systems. The stacking distance of NDPA is slightly reduced from 4.0 Å to 3.8 Å, from AMP to ADP, likely due to the extra constraint imposed by ADP on adjacent molecules, which form dimers. The simulated CD spectrum for M-NDPA-ADP (Figure. 2, main article) shows a negative bisignated signal with zero crossing around 380 nm, similar to that with AMP.

7 . Supplementary Figure 7 Binding modes of ATP to NDPA. Top view and side view of representative and trimers of NDPA-ATP. The arrows indicate the angle (smaller than 90 ) between neighboring molecules. See Supplementary Note 3 for detailed explanation.

8 Supplementary Figure 8 Potential energy changes. Evolution of the potentiel energy (kcal per mol per molecule) during 20 ns dynamics for representative right-handed assemblies built from and motifs based on NDPA- ATP. Energetically, the trimers seem much more stable than the trimers; the energy difference between the most stable trimers of both motifs that we built are more than 4 kcal per mol per molecule. These structures are only a starting point, as they do not imply that their stacking is favorable. The assemblies are still clearly more stable than the assembly.

9 a) b) Supplementary Figure 9 MD snapshots of NDPA-ATP assembly: a, Details of the inner structure of snapshots at the end of the MD for (P) NDPA-ATP assemblies with the (left) and (right) binding motifs. b, Schematic representing top view of two stacked trimers, showing ATP forming motif in red, and the angles between successive NDIs along the stack (the NDIs are numbered according to their location in the trimer, from 1 for the molecule at the bottom to 3 for the molecule at the top. In the second trimer, the NDIs are numbered similarly from 1 to 3 ). See Supplementary Note 4 for detailed explanation.

10 O H O N H N H O N Supplementary Figure 10 Role of Hydrogen Bonding. Hydrogen bonding between adenosine and NDPA in NDPA-AMP, NDPA-ADP, and NDPA-ATP assemblies: radial distribution function (RDF) between the hydrogen atoms of the hydroxyl groups of adenosine and the oxygen atoms of NDPA (left); RDF between the hydrogen atoms of the amino groups of adenosine and the oxygen atoms of NDPA (right). Experimentally, for NDPA-ATP, P-helices are more stable than the M-helices, which is opposite to the situation with AMP and ADP. As the helical pitches of the NDPA stacks are similar in the three systems, the change of helicity has to be found in the adenosine phosphate complexes. Their organization is indeed very different in AMP or ADP systems on one side, and ATP systems on the other side. With AMP and ADP, the adenosine moieties are able to be close enough to the naphthalenediimide cores to form H-bonds with them in the case of left-handed structures, principally via the hydroxyl groups (the amino groups of adenosine marginally establish hydrogen bonds with NDPA). This figure shows a plot of radial distribution function (RDF) between the hydrogen atoms of the hydroxyl groups of adenosine and the oxygen atoms of NDPA. There is a sharp peak at around 2 Å, a distance characteristic of hydrogen bonding, both for AMP and ADP systems. These interactions, however, no longer exist in the ATP systems, as the adenosine moieties are pointing away. The trace only appears after about 8 Å in the RDF, indicating that the ribose moieties are far from the cores. As both right-handed and left-handed helices share that characteristic, the main interactions biasing the energy balance towards right-handed helices in ATP systems are to be found in the periphery of the stacks, at the level of the chiral adenosine moieties. However, due to large constraints where ATP binds to the zinc coordinated dipicolylethylenediamine receptors, slight changes in the conformation of these rigid parts of the assemblies and associated changes in long-range interactions involving the adenosine moieties can produce many local minima with energies within a few kcal per mol per molecule span. Because this conformational space is very difficult to explore exhaustively, we prevent ourselves here in comparing the P-helix versus M-helix from an energetic point of view.

11 (P)-NDPA-ATP (P)-NDPA-ATP + 1 eq. AMP a) 8 b) 0.8 CD (mdeg) 0 CD (mdeg) 0.4 (P)-NDPA-ATP (P)-NDPA-ATP + 1 eq. AMP AMP X (M)-NDPA-AMP (P)-NDPA-ATP Supplementary Figure 11 Competitive binding between AMP and ATP. a, CD and b, absorption spectra of (P)- NDPA-ATP (5 x 10-5 M) with different equivalents of AMP. Lack of any significant change in the signal confirms inability of AMP to replace ATP competitively. Schematic on the right side depicts the same process.

12 Absorbance a) b) I (0-1) I(0-0) (M)-NDPA-ADP ATP (P)-NDPA-ATP Absorbance I (0-1) I (0-0) (M)-NDPA-AMP ATP (P)-NDPA-ATP Supplementary Figure 12 ATP based competitive replacement. Changes in the absorption spectra of a, (M)- NDPA-ADP and b, (M)-NDPA-AMP on competitive replacement with multivalent ATP (5 x 10-5 M in aq. HEPES buffer). Replacement of AMP and ADP by ATP show similar CD signal changes and thus cannot differentiate the two processes. But a closer look at the absorption spectra can distinguish the two processes. Whereas, removal of ADP by ATP in a) do not show any significant change in absorption spectra, a significant decrease in the intensity along with 2 nm red shift of absorption peak at 380 nm was observed when AMP was substituted by ATP as seen in b). This high band intensity at 380 nm, characteristic of weak NDI aggregates, are expected from AMP with one coordinating site, which lack the ability to clip chromophores for better stabilization. Whereas ADP and ATP with two and three point of attachments can clip to stabilize the chromophore aggregates, thus justifying the observed variation in absorption spectra (Supplementary Figure1b).

13 8 (P)-NDPA-ATP 4 CD (mdeg) 0-4 (M)-NDPA-AMP mole fraction of ATP Supplementary Figure 13 Chiral amplification experiment. Majority rules experiment with a mixture of AMP and ATP upon binding to NDPA (5 x 10-5 M in aq. HEPES buffer). The total concentration of phosphates is chosen such that all the binding sites are occupied and CD intensity was monitored at the CD maxima of 390 nm (positive) and 394 nm (negative). Zero CD intensity do not appear at 0.5 mole fraction of ATP due to non-racemic nature of the (M)-NDPA-AMP and (P)-NDPA-ATP helices which is reflected in their unequal maximum CD intensity (Supplementary Figure 1a).

14 a) b) o C 20 o -6 C 30 o C 40 o C CD (normalized) ln k Time (min) 40 C 30 C 20 C 10 C /T (K -1 ) c) 1.0 CD 390 nm) 0.5 Trial 1 Trial Time (min) Supplementary Figure 14 Kinetic analysis of the state-1 state-2 stereomutation process of (P)-NDPA-ATP assembly: a, Time dependent CD intensity at 390 nm monitored at different temperature (c = 5 x 10-5 M, 1 eq. ATP, with 0.28 U ml -1 CIAP) and corresponding fits (solid lines) using 1 st order kinetic model and b, the resultant Arrhenius plot. Table in c shows the increase in rate constant with temperature, which almost doubles with every 10 K rise. These data corresponds to the Figure 4d in main text. To show the reproducibility of the kinetic analysis, ATP hydrolysis was repeated multiple times and a sample data is presented in c). CD signal variation of two identical samples of (P)-NDPA-ATP with CIAP = 0.42 U ml -1 at 35 C was repeated and their respective 1 st order kinetic fit (solid line) is presented in c). We notice that the error margins in these rate constant values are within 5 % as calculated over repeated experiments (K trial1 =0.331 min -1 and K trial2 =0.333 min -1 ). All CD signals were normalized between zero and one for ease of fitting into kinetic model.

15 D ATP eq., Temp a) 15 o C b) 0, 15 o c) C o C 1, 15 o C 1, 95 o C CD (mdeg) d) 1.0 ATP eq., Temp e) 382 1, 15 o C Absorbance (norm) evident that as temperature 340 increases, the CD 380 signal vanishes Cooling along with blue shift in absorption spectra (382 nm to Absorbance Wavelenght (max, nm) , 15 o C A Temperature ( o C) CD (mdeg) f) 10 CD (mdeg) 0 Heating Cooling Temperature ( o C) ATP : CIAP 1, 95 o 1 : 0 U/ml C 1 : 0.28 U/ml 0.8 1,15 o C Supplementary Figure 15 Temperature dependent helical assembly. Temperature 5 dependent changes in a, CD Heating 381 and b, 0.5absorption spectra upon heating and cooling (P)-NDPA-ATP assembly (5 x 10-5 M in aq. HEPES buffer). c, Temperature dependent variation in CD intensity at 390 nm upon one complete cycle of heating and cooling. It is 380 nm). Upon cooling, the signal red shifts again to 382 nm along with recovery of CD signal. The vibronic features Time (sec) of the absorption spectra at high temperature do not resemble that of a monomeric NDPA. These observations indicate that the decrease in CD signal with increasing temperature at both (1) and (2) state of (P)-NDPA-ATP (Figure 4d, main text) is due to weaker ATP binding at elevated temperatures (and not their detachment as monomeric absorption features were not observed).

16 a) b) Absorbance (norm.) CIAP 0 U/ml 0.28 U/ml 0.56 U/ml 0.70 U/ml CD (mdeg) NDPA + CIAP NDPA Monomer CIAP NDPA Monomer Supplementary Figure 16 Enzyme interaction with NDPA. a, Absorption and b, CD spectral variations on addition of CIAP to NDPA. All measurements were done with 5 x 10-5 M solution in aq. HEPES buffer. Schematic on right side depicts the influence of CIAP on state of NDPA assembly. Absence of any significant change in absorption spectra and the lack of CD signal confirm that the CIAP do not have any specific interactions with NDPA.

17 a) Absorbance (norm.) 1.0 I (0-0) I (0-1) Pi 0 eq. 1 eq. 2 eq. 3 eq. NDPA Monomer Pi (PO 4 ) 3- (rac)-ndpa-pi b) I (0-0) Absorbance NDPA 1.0 I (0-1) (P)-NDPA-ATP (rac)-ndpa-pi after stereomutation 3 eq. Pi (PO 4 ) (P)-NDPA-ATP CIAP (rac)-ndpa-pi Supplementary Figure 17 Pi binding of NDPA. a, Absorption spectra of NDPA upon binding with Pi [(PO 4 ) 3- ]. Changes in the ratio of band intensity suggest significant interchromophoric interactions in NDPA on binding to Pi. b, Changes in the absorption spectra upon enzymatic hydrolysis of ATP with CIAP and its comparison with Pi bound NDPA. Schematic on right side depicts the respective processes. All experiments were at 5 x 10-5 M NDPA in aq. HEPES. Absorption spectra obtained after complete hydrolysis is significantly different from that of NDPA alone without phosphates, indicating that it is not monomeric. We see that after stereomutation, the absorption spectra show blue shift of band along with change in the ratio of band intensity indicating weakening of aggregates. As Pi is shown to strongly interact with NDPA, 3 eq. of Pi released upon complete hydrolysis of ATP can bind to the NDPA resulting in racemic assembly due to achiral nature of Pi.

18 (M)-NDPA-AMP (rac)-ndpa-pi a) 2 0 t = 20 min -2 t = 0 min CD (mdeg) (M)-NDPA-ADP (rac)-ndpa-pi b) c) CD 394 nm U/ml 0.42 U/ml 0.56 U/ml Time (min) Absorbance (norm.) 0.5 I (0-0) I (0-1) (M)-NDPA-AMP CIAP (rac)-ndpa-pi d) 4 CD (mdeg) min 0 min CD 394 nm e) f) U / ml 0.84 U / ml Time (min) Absorbance (norm.) I (0-1) I (0-0) (M)-NDPA-ADP CIAP (rac)-ndpa-pi Supplementary Figure 18 Kinetic analysis of NDPA bound AMP and ADP hydrolysis. Time dependent CD spectra showing the racemisation process of a, (M)-NDPA-AMP (c = 5 x 10-5 M) with CIAP (0.42 U ml -1 ), d, (M)- NDPA-ADP (c = 5 x 10-5 M) with CIAP (0.84 U ml -1 ) at 35 C. Schematic shows the pictorial representation of the enzymatic AMP and ADP hydrolysis. b, and e, show the corresponding CD intensity monitored at 394 nm probing the racemisation process of (M)-NDPA-AMP and (M)-NDPA-ADP respectively, with varying concentrations of CIAP at 35 C. c, and f, show changes in absorption spectra corresponding to hydrolysis of AMP/ADP prebound to NDPA with CIAP (0.70 U ml -1, 35 o C). We notice that hydrolysis leads to weakening of aggregates as seen from the change in ratio of the I (0-0) /I (0-1) but doesn t show monomeric features, suggesting Pi binding to the stacks.

19 a) b) I 1.0 (0-1) Absorbance (norm.) c) 0.5 I (0-0) CIAP NDPA (M)-NDPA-AMP (rac)-ndpa-pi after stereomutation 1 eq. Pi (PO 4 ) 3- Absorbance (norm.) 1.0 I (0-1) NDPA CIAP I (0-0) (M)-NDPA-ADP (rac)-ndpa-pi after stereomutation 2 eq. Pi (PO 4 ) 3- (M)-NDPA-AMP (rac)-ndpa-pi (M)-NDPA-ADP Supplementary Figure 19 Absorption changes during stereomutation: Changes in the absorption spectra upon enzymatic hydrolysis of a, (M)-NDPA-AMP and b, (M)-NDPA-ADP with CIAP (0.7 U ml -1, 35 C) and its comparison with completely racemized state (rac)-ndpa-pi obtained by action of CIAP as well as the Pi bound NDPA. Absorption spectra obtained after complete hydrolysis is very different from that of NDPA alone without phosphates, as seen from their I (0-0) /I (0-1) value indicating that it is not monomeric. Also the spectra after stereomutation matches well with the Pi bound NDPA stacks confirming that NDPA is indeed aggregated due to Pi binding.

20 a) 0.42 U / ml b) 0.28 U / ml c) U / ml U / ml 0.56 U / ml CD (normalized) Time (min) CD (normalized) Time (min) CD 394 nm) Time (min) Trial 1 Trial 2 Supplementary Figure 20 Kinetic analysis of NDPA bound ADP and AMP hydrolysis. Time dependent CD signal change on action of varying CIAP concentration to a, (M)-NDPA-ADP, b, (M)-NDPA-AMP at 35 C and their respective 1 st order kinetic fit (solid line). a, Correspond to kinetic decay data of Supplementary Figures 18e and b, for Supplementary Figures 18b. CD signals were normalized between zero and one for ease of fitting into kinetic model. To show the reproducibility of the kinetic analysis, AMP hydrolysis was repeated multiple times and a sample data is presented in c). CD signal variation of two identical samples of (M)-NDPA-AMP with CIAP = 0.40 U ml -1 at 35 C was repeated and their respective 1 st order kinetic fit (solid line) is presented in c). We notice that the error margins in these rate constant values are within 5 % as calculated over repeated experiments (K trial1 =0.116 min -1 and K trial2 =0.118 min -1 ).

21 a) b) Rate Constant (k, min -1 ) AMP ADP CIAP (U / ml) CD (mdeg) (M)-NDPA-AMP (M)-NDPA-ADP Time (sec) Supplementary Figure 21 Comparative Enzymatic hydrolysis. a, Plot of hydrolysis rate constants of NDPA bound AMP and ADP (k amp and k adp ) with varying concentration of CIAP at 35 C as shown in Supplementary Table 5. We see that at a given CIAP concentration, k amp is higher than k adp confirming that ADP hydrolysis is much slower than AMP. Confirming the same fact b, shows decay kinetics by plotting time dependent variation in CD signal of (M)-NDPA-AMP and (M)-NDPA-ADP (c=5 x 10-5 M)with CIAP (0.84 U ml -1 ) at 35 o C monitored at 390 nm.

22 Absorbance I (0-1) (2) I (0-0) 0.6 (1) a) b) 0.3 (1) (2) Absorbance I (0-1) I (0-0) (3) (2) (3) (2) Supplementary Figure 22 Absorption changes in stepwise hydrolysis of ATP: Variation in absorption spectra upon stepwise hydrolysis of ATP during a, 1 2 and b, 2 3 (c = 5 x 10-5 M, 35 C with CIAP (0.7 U ml -1 )). We see that the transformation from 1 2 (when the CD signal reverses in Fig. 4c) do not show any change in absorption spectra as expected of the transition from ATP ADP (Supplementary Figure 12a). However, the later process from 2 3 (when the CD signal goes to zero in Fig. 4f), we notice changes in vibronic features of NDPA absorbance (I (0-0)/I (0-1) value) as expected of the transition from ADP AMP Pi (Supplementary Figure 12b).

23 Time ATP + CIAP(t 4 ) ATP + CIAP(t 3 ) ATP ATP + CIAP(t 2 ) ADP ATP + CIAP(t 1 ) AMP ATP ppm Supplementary Figure P NMR signature of enzymatic process. Time dependent changes in the 31 P NMR spectra of ATP in presence of CIAP and its comparison with ATP without CIAP. We notice that the peak at ppm (characteristic of ATP, marked with red star symbol)vanishes soon after addition of CIAP, whereas the other two (-11.2 ppm, ppm, marked with blue triangle symbol) remain which is characteristic of ADP, which further changes with time to give only AMP signal followed by conversion to Pi and adenosine. This enzymatic ATP hydrolysis without NDPA also follows the same pathway and confirms our finding by circular dichroism in presence of NDPA.

24 a) 0.24 b) c) 0.12 A 820 nm (M)-NDPA-AMP (M)-NDPA-ADP A 820 nm ATP without NDPA (P)-NDPA-ATP A 820 nm 8 4 ADP without NDPA (M)-NDPA-ADP Time (min) Time (min) Time (min) Supplementary Figure 24 Inorganic method for Pi estimation. Pi assay results using Chen s method to determine the amount of Pi released on enzymatic hydrolysis. a) shows comparative hydrolysis kinetics of NDPA bound AMP and ADP. We see that ADP hydrolysis is indeed very slow when compared with AMP as proven previously by monitoring CD signal (Supplementary Figure11, Supplementary Figure12). These measurements were done with CIAP (0.84 U ml -1 ) and NDPA (c = 5x10-5 M) at 35 C. Comparative hydrolysis rate of b) ATP and c) ADP with and without NDPA, as probed by Pi assay. It shows that phosphate hydrolysis is slightly faster in free state compared to NDPA bound state. All measurements were done with CIAP (0.84 U ml -1 ) and NDPA (c = 5x10-5 M) at 35 C. This difference in the rate of bound and unbound phosphates could be due to easy access of unbound phosphates to enzyme. However, unlike the helix reversal measurements for kinetic analysis, Chen s assay is not an in situ measurement due to highly acidic conditions required for metal complex formation, so aliquots were taken at regular intervals for measurements. In addition, this method can only probe the net Pi released and cannot distinguish the stepwise hydrolysis of ATP via ADP and AMP to Pi.

25 5 (M)-NDPA-AMP + 1 eq. ATP (M)-NDPA-ADP + 1 eq. ATP CD (mdeg) Time (min) Supplementary Figure 25 Mechanistic insight on enzymatic action. Time dependent variation in CD signal of a) 5 x 10-5 M NDPA (aq. HEPES buffer) on competitive replacement of AMP and ADP by 1 eq. of ATP (CIAP=0.70 U ml -1 ) at 35 o C monitored at 390 nm. To understand if enzyme selectively hydrolyses the unbound or bound phosphates, CD kinetic measurements of (P)-NDPA-ATP in presence of other phosphates (one eq. of ATP followed by addition of 1 eq. of either AMP or ADP) were performed. As expected, due to competitive binding, ATP should be bound to NDPA to form (P)-NDPA-ATP stacks, whereas AMP or ADP must be free in solution. Gradual decrease in signal from time t=0 suggests that CIAP has no preferential action to unbound phosphates compared to bound ones. If enzymatic action were to be preferential to unbound phosphates, we should have obtained constant CD signal initially till all unbound phosphates are consumed and only then signal should have started to decrease.

26 Supplementary Table 1 Potential energy of NDPA-AMP: Average potential energy for the 8 types of AMP bound supramolecular assemblies during MD; the average potential energy of the most stable assembly has been set to zero, and that of other assemblies rescaled accordingly to allow direct energy comparison between the assemblies (R stands for right-handed helical assembly (P); L stands for left-handed helical assembly (M)). The standard deviation is around 1.8 kcal per mol for all types of assemblies. See Supplementary Note 1 for detailed explanation. Type of assembly Average potential energy (kcal per mol per molecule) C2-L C2-L-2 C2-R C2-R I-L I-L I-R I-R-2 2.6

27 Supplementary Table 2 Average percentage of oxygen atoms of NDPA involved in hydrogen bonding with hydroxyl groups of adenosine in AMP (C=O (NDPA)--- H-O (ribose)). Despite similar H-bonding energies between the assemblies, their hydrogen-bonding network is different, and thus the orientation of adenosine. Analysing hydrogen bonding between the hydroxyl groups of ribose and the oxygen atoms of NDPA during the last 10 ns of MD, it is found that left-handed assemblies develop up to two hydrogen bonds per molecule. These interactions can take place between molecules n and n+2 or n+3, thus contributing to long-range ordering of the assembly. Such hydrogen bonds are two to three times less numerous in the P-helix. Type of assembly Average proportion of oxygen atoms of NDPA involved in hydrogen bonding (%) C2-L-1 45 C2-L-2 39 C2-R-1 15 C2-R-2 15

28 Supplementary Table 3 Structural parameters of different types of helices in NDPA-AMP assemblies, with amplitude of deviations along the stack. See Supplementary Figure 5 for detailed explanation. The rotation strength between two neighbouring molecules is calculated as 1 x 2. r12/ 1 2, the mixed product between two successive dipoles along the stack. *The average number of - contacts is calculated from the total number of interactions at short distance (from 3.5 to 5.5 Å) between the four rings centers of adjacent NDPA molecules along the stack, divided by the total number of NDPA molecules. Structural parameter Left-handed helix (M) Right-handed helix (P) Average angle between NDPA ± ± 19.3 Average distance between NDPA 4.04 Å ± 0.42 Å 4.45 Å ± 0.92 Å Average «rotation strength» ± ± 0.21 Average number of H-bonds per NDPA-AMP in the stack Average number of - contacts per NDPA-AMP in the stack*

29 Supplementary Table 4 Kinetic analysis of the (1) (2) stereomutation process of (P)-NDPA-ATP assembly: Variation in rate constant of stereomutation in (P)-NDPA-ATP assembly upon increasing temperatures (c = 5 x 10-5 M, 1 eq. ATP, with 0.28 U ml -1 CIAP) as obtained by using 1 st order kinetic model. The rate constant increases with temperature, which almost doubles with every 10 K rise. These data corresponds to the Figure 4d in main text and Supplementary Figure 14. T (K) k ATP (min -1 )

30 Supplementary Table 5 Comparative rate constant of NDPA bound AMP and ADP hydrolysis at various CIAP concentrations and 35 C as obtained from the respective 1 st order kinetic fit shown in Supplementary Figure 20. CIAP (U ml -1 ) k AMP (min -1 ) k ADP (min -1 )

31 Supplementary Notes Supplementary Note 1 NDPA-AMP assembly: When grafting AMP, for the I conformer, the molecule loses its inversion center because of the chirality of AMP; however, the new structure obtained will still be referred to in the text as I, in reference to the symmetry of the NDPA core. For both C2 and I, two orientations of AMP are possible, and the molecules can assemble as left and right-handed helices. In total 8 systems were thus built: 2 types of NDPA conformations (I and C2) x 2 orientations of the AMP (1 and 2) with respect to the zinc atom x 2 left- or right-handed chirality of the NDPA stack (L and R). The conformers are thus named accordingly, such as C2-L-1 or I-R-2. In these assemblies, one of the two zinc atoms of a given molecule n is close to a zinc atom of molecules n+1 and n-1, but can also be close to a zinc atom of other molecules (for instance n-3 or n+3 if the rotation angle between successive molecules is 60 ; see Supplementary Figure 1). This information provides a clue about the possibilities of grafting that exist when replacing AMP with ADP or ATP, which have two or three grafting points, respectively. Molecular Dynamics is performed on these 8 types of supramolecular assemblies, each constituted of 32 NDPA molecules and 64 AMP molecules, so that each assembly is modelled in comparison to its mirror image structure, but with the ribose keeping a right-handed (D) chirality. To give insights into their relative stabilities, the average potential energy and standard deviation have been calculated after a sufficiently large relaxation time (the last 10 nanoseconds of MD for all structures, except O2-R-2 and I-L- 1, where the calculation was performed on the last 5 nanoseconds, as the assembly required more time to relax), and displayed in Supplementary Table 1. The potential energies show a clear energy segregation between the modelled stacks, on a scale of about 3.5 kcal per mol per molecule (i.e. nearly 112 kcal per mol for the 32 NDPA stack). The most striking segregation is between C2 and I families of assemblies, the least stable assemblies being those built from I conformations, where defects and stack bending appear during MD. This behavior reflects steric crowding in I systems, which explains this high level of energy segregation. Within C2 assemblies, the most stable structures are left-handed: from 1.5 up to 2.4 kcal per mol per molecule more stable than their right-handed counterparts (Supplementary Table 1). This is a large energy difference, since this potential energy is estimated per NDPA in the stack. Between the two C2 lefthanded assemblies, a small energy difference of 0.3 kcal per mol per molecule is estimated, suggesting that the orientation of AMP grafting on Zn atom does not appear very critical for left-handed (M) structure. Supplementary Note 2 (P)-NDPA-AMP vs (M)-NDPA-AMP: To evaluate the origin of the energy difference between right-handed and left-handed C2 assemblies of NDPA-AMP, we extracted the valence (bonding terms) and long-range (non-bonding) interactions from the energy profiles and displayed their evolution on a scale of around 10 kcal per mol per molecule, see Supplementary Figure 3. From the results

32 obtained, it is difficult to attribute the energy difference solely to a given type of interactions, as there are multiple cancellation effects. However, we can observe that the largest energy differences arise mainly from van der Waals interactions. Supplementary Note 3 NDPA-ATP assembly: With ATP, there are three grafting points, which give rise to much more different possibilities of grafting than with AMP. One possible type of assembly is formed of trimers interacting by long-range interactions, a trimer being composed of three stacked NDPA molecules bonded together by two ATP. Another possible type of assembly results from shifted ATP binding modes, i.e. where ATP molecules share three stacked NDPA molecules with two other ATP molecules, instead of one as in a trimer. As a result, all NDPA molecules are linked together to form a supermolecule. The assemblies with shifted binding modes that we built, however, showed a large level of disorder during MD, indicative that this type of organization is unfavorable. We thus focused on assemblies built from trimers. Despite this focus on a more limited set of conformers, there are still different possibilities to bind ATP to three NDPA molecules. The trimers are highly constrained, because the zinc complexes are rigid, and because these rigid entities are bonded together via ATP. Because of the different possibilities of binding, and the presence of strong geometric constraints, there are different local minima that are only accessible by building them systematically (this is opposite to flexible systems, where computational methods such as MD can be used to explore their potential energy surface). Due to this large number of possibilities, an exhaustive exploration was not possible. However, the simulation of about twenty trimers allowed us to distinguish two families of trimers, whose members have similar designs. The two families are characterized by the order of binding of the stacked molecules: or The x-y-z design means that molecule x is linked to molecule y, and molecule y is linked to molecule z, while x, y, and z refers to the order of the molecules along the stacking direction (Supplementary Figure 7). Supplementary Note 4 In Supplementary Figure 9, the right-handed and assemblies are displayed. The striking characteristic of assemblies is that the organization of the adenosine phosphate complexes is not helical. While the naphthalenediimide cores do form a right-handed helix, the adenosine phosphate complexes form two columns on both sides of the assembly (Supplementary Figure 9). As the pitch of the naphthalenediimide helix is similar, whatever the adenosine phosphate (6 molecules with ATP, versus 6.3 with AMP and ADP), as well as the average rotation angle between neighboring molecules (60 versus 57-58, respectively), the peculiar arrangement of the complexes is to be explained by symmetry. With AMP, the angle between the adenosine phosphate complexes is about 60, i.e. the same as the angle between the naphthalenediimide cores. With ADP, because each adenosine phosphate complex

33 involves two naphthalenediimide cores, the angle between the complexes is twice the angle between the naphthalenediimide cores, about 120 (or -60, considering that the molecules have a C 2 axis along the stacking direction). Thus, for left-handed assemblies, the adenosine phosphate complexes can be regarded as forming a left-handed helix with motifs separated by 120 (pitch of 3.1 adenosine phosphate complexes, i.e. 6.3 molecules), or a right-handed helix with motifs separated by 60 (pitch of 6.3 adenosine phosphate complexes, i.e molecules). With ATP, the angle between the adenosine phosphate complexes is three times the angle between the naphthalenediimide cores, about 180. As a result, the pitch of the assembly of adenosine phosphate complexes is of only two complexes, effectively corresponding to six NDPA molecules. But an assembly with a pitch of only two motifs can not be called an helix; the trimers are translational images from each other along the stacking direction. We then investigated in more details the R representative assembly. Ten structures extracted every nanosecond between 10 and 20 ns of MD were used to study the angle between neighboring molecules. Though the average angle is of 60, a more detailed analysis reveals that there are three population distributions: the angle between molecules 1 and 2, 1-2 is 81 ± 4. Similarly, 2-3 is 44 ± 4 (see Supplementary Figure 7). And 3-1 = 54 ± 4, 1 refering to a molecule belonging to a neigboring trimer (Figure 2, main article). The first two angles deviate from the angle found in the NDPA-AMP arrangement (~57 ), due to the constraints imposed by ATP. The third one, however, measured between molecules belonging to adjacent trimers, i.e. more free to rotate with respect to one another, is not much affected. This result indicates weak steric hindrance between trimers, which are thus able to stack while conserving the natural angle between single NDPA molecules. The sum of the three angles is 179, corresponding well to the symmetry condition necessary for having adenosine phosphate complexes forming columns instead of helices. Supplementary Methods Materials: CIAP (3 units mg -1 ) was purchased from Sisco Research Laboratory Pvt. Ltd. India. All other chemicals were purchased from the commercial sources and were used as such. Spectroscopic grade solvents were used for all optical measurements. Simulation of CD spectra: The calculation of the excitonic CD spectra involves two steps. First, the lowest 30 excited states of the 32 NDPA molecules extracted from the MD trajectories above are computed at the INDO/SCI level (using an active space of 30 occupied and 30 empty molecular orbitals). Then, an excitonic Hamiltonian encompassing a total of 32x30 basis functions (30 localized excitations per molecule) is built on the basis of INDO/SCI 1 excitation energies and exciton couplings. The latter are calculated as Coulomb interactions between transition densities, thus going beyond the usual point dipole

34 model. 2 Diagonalization of this Hamiltonian yields a set of 960 exciton states with energies ħ and wavefunctions >, for which the oscillator strength f and the rotational strength R are computed as: 3 f μ i, n 2 i, n µ i, n G 2,...Supplementary Equation 1 R h c μ i, n G G μ i, n j, n' µ i, n j, n' j, n' ( r n r n' ),.Supplementary Equation 2 where c is the speed of light, µ i,n the transition dipole moment from the ground state g> to the excited state i> of molecule n along the stack, μ ( i, n g h. c.) the corresponding dipole operator, G> the i, n µ i, n ground state of the helical stack (product state of all g>), and c i n the exciton state i, n i, n, wavefunctions expanded in terms of the c, eigenvectors. The absorption/cd response at input frequency i n is calculated on the basis of the oscillator/rotational strengths as: Abn( ) f G( ), Supplementary Equation 3 CD( ) R G( ),..Supplementary Equation 4 where G(- ) is a Gaussian function centered around with variance = 0.1 ev. The brackets denote a configurational average over the positional and energetic disorder as explored during the MD simulations. Here, a total of 8 supramolecular helical structures, each consisting of 32 (for NDPA-AMP and NDPA- ADP assemblies) or 33 (for NDPA-ATP assemblies) molecules, were used. This approach was found to yield CD spectra that are stable with respect to configurational averaging. Synthesis and Procedures: NDPA was synthesized based on the reported procedure and was accordingly characterized. 4, 5 Supplementary References: 1 Ridley, J. & Zerner, M. An intermediate neglect of differential overlap technique for spectroscopy: Pyrrole and the azines. Theor. Chim. Acta 32, (1973).

35 2 Beljonne, D., Cornil, J., Silbey, R., Millié, P., Brédas, J. L. Interchain interactions in conjugated materials: The exciton model versus the supermolecular approach, J. Chem. Phys. 112, (2000). 3 Spano, F. C., Meskers, S. C. J., Hennebicq, E. & Beljonne, D. Probing excitation delocalization in supramolecular chiral stacks by means of circularly polarized light: experiment and modeling,. J. Am. Chem. Soc. 129, (2007). 4 Kumar, M., Jonnalagadda, N. & George, S. J. Molecular recognition driven self-assembly and chiral induction in naphthalene diimide amphiphiles. Chem. Commun. 48, (2012). 5 Lee, H. N. et al. Pyrophosphate-selective fluorescent chemosensor at physiological ph: formation of a unique excimer upon addition of pyrophosphate. J. Am. Chem. Soc. 129, (2007).