The single molecule force unzipping experiments allowed us to extract the DNA base pair free energies at various salt concentrations. The heterogeneous salt correction found in our measurements is a remarkable result and it could be due to electrostatic effects. The NNBP formed as a combination of purine-purine or pyrimidine-pyrimidine (-YY- or -RR-, i.e., AA/TT, AG/TC, CC/GG, GA/CT) differ most from the UO homogeneous salt correction than mixed purine-pyrimidine combinations (-RY- and -YR-). A difference between these combinations can be observed in how charges (e.g., hydrogen bond acceptor and donor groups) are distributed along the major groove of the double helix. The latter have charged groups that tend to be uniformly distributed between the two strands along the major groove, whereas the former have donor and acceptor groups unevenly distributed between the two strands. The specific salt correction found in our measurements could be consequence of how monovalent cations bind the two strands along the major groove. There are precedents to such results. Sugimoto and collaborators  have reported that cation binding is correlated to duplex stability. Computer simulations have identified acceptor groups in guanine (N,O) and adenine (N) as preferential cation binding sites . Our experimentally determined specific salt corrections might be interpreted as a corroboration of such hypothesis.
An alternative explanation could be that the heterogeneity of the salt correction is consequence of sequence specific elastic properties of the ssDNA. Previous studies of ssDNA elasticity  have shown how the contour length of the ssDNA gradually increases from low to high forces suggesting a conformational transition of the sugar pucker that goes from the A-form (C-endo) to the B-form (C-endo) at high forces (see Sec. 3.1 and Fig. 3.4c for a detailed description of the sugar puckering). A similar phenomenon has been reported in recent studies of homopolymeric RNA sequences  that reveal conformational transitions under tension (in the form of force plateaus) in poly-C and poly-A sequences but not for poly-U sequences (studies for poly-G sequences were not available). For random ssDNA sequences one might expect that such specific effects are averaged out and a gradual A
B transition is observed. Conformational transition effects are also observed in the elastic measurements of ssDNA when varying the salt. As shown in Table 5.1, the value of the interphosphate distance increases when decreasing the salt (it changes from 0.59 nm at high salts to 0.67 nm at low salts) suggesting a similar sugar pucker (B
A) conformational transition for the ssDNA. How this consideration could affect the interpretation of the heterogeneous salt correction? In other words, might it be possible that the contribution of each pair of stacked bases to the Kuhn length is different? This result has never been reported before in any experimental study of the ssDNA elasticity [48,139]. It represents a completely different interpretation of the salt dependence and there is no reason why such hypothesis should be rejected. In order to determine the dependence of the Kuhn length with the sequence, it would be necessary to perform a detailed study of the elastic properties of periodic ssDNA sequences along the lines of what has been recently reported for homopolymeric RNA sequences . A detailed investigation should be carried out for homopolymeric DNA sequences containing all possible combinations of stacking bases. Candidate sequences should be poly-dA, poly-dC, poly-dT and other periodic sequences such as poly-(dA-p-dT), poly-(dA-p-dT-p-dC). In the end, the Kuhn length associated to each combination of stacked bases (16 in total) could be written as a constant value plus a base (or sequence) dependent correction. Although an exhaustive research of the elastic response of different ssDNA sequences would shed light into this issue, this is beyond the scope of the present work.
JM Huguet 2014-02-12