Peptide nucleic acids: Mechanism of tight binding and application in artificial nuclease

Peptide nucleic acids (PNA) are mimics of nucleic acids with a peptidic backbone. Duplexes and triplexes formed between PNA and DNA or RNA possess remarkable thermal stability, they are resistant to nuclease cleavage and can better discriminate mismatches. Understanding the mechanism for the tight binding between PNA and oligonucleotides is important for the design and development of better PNA-based drugs.^ We have performed molecular dynamics (MD) simulations of 8-mer PNA/DNA duplex and two analogous duplexes with chiral modification of PNA strand (D- or L-Alanine modification). MD simulations were performed with explicit water and Na$\sp{+}$ counter ions. The 1.5-ns simulations were carried out with AMBER using periodic boundary and particle mesh Ewald summation. The point charges for PNA monomers were derived from fitting electrostatic potentials, obtained from ab initio calculation, to atomic centers using RESP. Derived charges reveal significantly altered charge distribution on the PNA bases and predict the Watson-Crick H-bonds involving PNA to be stronger. Results from NMR studies investigating H-bond interactions between DNA-DNA and DNA-PNA base pairs in non-polar environment are consistent with this prediction. MD simulations demonstrated that the PNA strand is more flexible than the DNA strand in the same duplex. That this flexibility might be important for the duplex stability is tested by introducing modification into the PNA backbones. Results from MD simulation revealed dramatically altered structures for the modified PNA-DNA duplexes. Consistent with previous NMR results, we also found no intrachain hydrogen bonds between O7$\sp\prime$ and N1$\sp\prime$ of the neighboring residues in our MD study. Our study reveals that in addition to the lack of charge repulsion, stronger Watson-Crick hydrogen bonds together with flexible backbone are important factors for the enhanced stability of the PNA-DNA duplex.^ In a related study, we have developed an application of Gly-Gly-His-(Gly)$\sb3$-PNA conjugate as an artificial nuclease. We were able to demonstrate cleavage of single stranded DNA at a single site upon Ni(II) binding to Gly-Gly-His tripeptide and activation of nuclease with monoperoxyphthalic acid. ^

DNA-DAMAGING REACTIONS OF NEOCARZINOSTATIN (ANTIBIOTICS, DNA REPAIR, CYTOTOXICITY)

Sensitive assays utilizing a cell-free and an intracellular system were employed to study the molecular bases of the DNA-damaging reactions of neocarzinostatin (NCS). In the cell-free DNA system, super-helical form I DNA from the bacteriophage PM2 was used as the substrate. The three forms of DNA present after treatment with NCS were separated by agarose gel electrophoresis. When NCS-damaged DNA was assayed under neutral conditions, there was a progressive decrease in the amount of surviving form I DNA and a corresponding increase in form II (nicked, relaxed circular) DNA, but very little increase in form III (linear duplex) DNA. This indicates that NCS introduces primarily single-strand breaks. However later studies showed that there were some site-specific double-strand breaks mediated by NCS on PM2 DNA. Seven such specific sites were mapped on the PM2 genome. When the damage was assayed under nondenaturing alkaline conditions or with the apurinic/apyrimidinic endonuclease IV, there was a slightly greater decrease in the amount of surviving form I DNA compared with neutral conditions indicating the presence of some alkali-labile sites.^ NCS-mediated DNA damage and repair were examined with cultured Chinese hamster ovary (CHO) cells using either alkaline elution for analysis of single-strand breaks or neutral elution for analysis of double-strand breaks. Most of the strand breaks introduced by NCS were capable of being rejoined. However, there was a small amount of residual DNA damage remaining unrejoined at 24-hr after removal of the drug. The amount of residual DNA damage was higher in a CHO mutant cell line (EM9) having a higher sensitivity to killing by NCS than its parental strain (AA8). Other lesions, DNA-protein complexes and alkali-labile sites, were detected after NCS treatment but they constituted only a small fraction of the DNA damage.^ Based on the above information, it can be postulated that NCS introduces some very lethal DNA damage. It is likely that the lethal lesions are a subset of the total DNA lesions representing the residual DNA damage. This DNA damage may be composed of site-specific, unrejoinable double-strand breaks and are thus the primary lesion leading to NCS-mediated lethality.^

Selection bias and the cross-validation of regression models for prediction

Strategies are compared for the development of a linear regression model with stochastic (multivariate normal) regressor variables and the subsequent assessment of its predictive ability. Bias and mean squared error of four estimators of predictive performance are evaluated in simulated samples of 32 population correlation matrices. Models including all of the available predictors are compared with those obtained using selected subsets. The subset selection procedures investigated include two stopping rules, C$\sb{\rm p}$ and S$\sb{\rm p}$, each combined with an 'all possible subsets' or 'forward selection' of variables. The estimators of performance utilized include parametric (MSEP$\sb{\rm m}$) and non-parametric (PRESS) assessments in the entire sample, and two data splitting estimates restricted to a random or balanced (Snee's DUPLEX) 'validation' half sample. The simulations were performed as a designed experiment, with population correlation matrices representing a broad range of data structures.^ The techniques examined for subset selection do not generally result in improved predictions relative to the full model. Approaches using 'forward selection' result in slightly smaller prediction errors and less biased estimators of predictive accuracy than 'all possible subsets' approaches but no differences are detected between the performances of C$\sb{\rm p}$ and S$\sb{\rm p}$. In every case, prediction errors of models obtained by subset selection in either of the half splits exceed those obtained using all predictors and the entire sample.^ Only the random split estimator is conditionally (on $\\beta$) unbiased, however MSEP$\sb{\rm m}$ is unbiased on average and PRESS is nearly so in unselected (fixed form) models. When subset selection techniques are used, MSEP$\sb{\rm m}$ and PRESS always underestimate prediction errors, by as much as 27 percent (on average) in small samples. Despite their bias, the mean squared errors (MSE) of these estimators are at least 30 percent less than that of the unbiased random split estimator. The DUPLEX split estimator suffers from large MSE as well as bias, and seems of little value within the context of stochastic regressor variables.^ To maximize predictive accuracy while retaining a reliable estimate of that accuracy, it is recommended that the entire sample be used for model development, and a leave-one-out statistic (e.g. PRESS) be used for assessment. ^