A team of scientists combine forces to derive the first accurate 3D solution structure of a fascinating double-helical molecule that holds promise for applications in biomedicine and nanotechnology

Nature is sophisticated, says Catalina Achim. Over the course of several billion years, it has evolved remarkably efficient ways to transfer electrons from atom to atom within living organisms to produce energy from food. Our energy for getting up in the morning, for work and for pleasure all comes from these processes of “controlled burning” that depend on electron transfer.

PHOTO: Catalina Achim
group, Carnegie Mellon University

Catalina Achim (third from left), Carnegie Mellon University, with her laboratory group. “Our collaboration with PSC was very beneficial,” says Achim, “not only for the research itself but also for educational purposes. Working with Marcela Madrid, my graduate students learned how to do molecular dynamics simulations.”

“Just like we transfer electricity through power lines to heat and light our homes, we transfer electrons in our bodies to metabolize food,” says Achim, associate professor of chemistry at Carnegie Mellon University. “But nature does it very efficiently while we don’t yet know how to take oil and make our cars go without wasting a lot of energy. We know some of the basics of how electron transfer works, and many scientists study these processes so we can learn to apply them.”

PHOTO: Marcela Madrid, Pittsburgh Supercomputing Center, XSEDE Advanced User Support Services

Marcela Madrid, Pittsburgh Supercomputing Center, XSEDE Extended Collaborative Support Services

To that end, Achim has worked with a team and turned to XSEDE resources, including PSC scientist Marcela Madrid, and Pople, PSC’s SGI Altix system, to solve the structure of a fascinating “bio-mimetic” molecule called PNA, peptide nucleic acid. Their aim is to use PNA as a molecular “scaffold” — to organize metal ions so that they can transfer electrons as efficiently as biological electron-transfer molecules. Although PNA doesn’t exist in nature, it’s a close cousin to DNA and for Achim’s purposes has a special advantage — it doesn’t have charge. DNA’s helical strands have negative charges, due to repeating phosphate groups within the backbone, and when the four A-G-C-T bases pair up to form the DNA double-stranded helix, there’s electrostatic repulsion between the strands. “The DNA structure is overall always negatively charged,” says Achim.

DNA interested Achim as a scaffold for metal ions, but the negative charge and its sensitivity to chemical degradation in living organisms made it unsuitable. To circumvent these limitations, Achim and her colleagues turned to PNA, which they synthesize in her Carnegie Mellon laboratory — by replacing the DNA backbone with peptide-like groups. The resulting neutral double helix is more stable and better suited for research in electron transfer.

The necessary first step for electron-transfer research, though, was an accurate 3D structure of PNA in solution with water. For this task, Achim and her collaborators teamed with Madrid, an expert in molecular dynamics (MD), an approach that can find a molecule’s structure by computing the forces between its atoms. The results of their work, reported in Molecular Biosystems (2010), a journal of the Royal Society of Chemistry, provided for the first time the PNA structure in solution.

From this starting place, Achim and her colleagues can investigate potential applications of PNA that can open new understanding of electron transfer. Achim foresees, for instance, that PNA could be used to create “nanowires” — 100,000 times finer than a human hair — for quantum circuitry, in which the quantum characteristics of electrons hopping and tunneling can lead to electronics much faster than today’s integrated circuitry.

A Stiffened Backbone

PNA was first synthesized in 1991 by chemists in Denmark. Its similarity to DNA has made it a subject of study in various contexts. Being an inorganic chemist, Achim’s interest was to use it as a platform for transition metal ions, creating molecular complexes with various applications. “They may behave as molecular magnets,” she says, “or transport charge in a particular way, or act as catalysts.”

Superposition of 10 time-averaged methyl-substituted PNA structures from restrained MD.

The transition metals, which include copper, iron or ruthenium, easily transfer electrons and hence are ideal for the study of electron transfer via a molecular scaffold. “Our question,” says Achim, “was ‘What’s the best way in space to organize transition metal ions?’ Our idea was ‘Let’s use a scaffold that resists degradation and is easily modified in the lab.’” DNA’s drawbacks brought PNA to the fore.

A static structure of PNA, from x-ray crystallographic study was available, but in biological applications in a water environment, PNA isn’t static; it’s flexible and mobile. To find its 3D structure in solution, Achim and her collaborators used a combination of NMR spectroscopy, computational quantum chemistry and a specialized form of MD.

A “Collaborative Research in Chemistry” grant from NSF supported this work, which depended on teamwork among Achim, Madrid and Achim’s partners at Carnegie Mellon and Duke University. First, Achim and her students synthesized several versions of PNA, each with a different chemical structure and flexibility in solution. A graduate student in Achim’s lab worked with Carnegie Mellon chemist Roberto Gil on the NMR spectroscopy of the synthesized PNAs, which provided a matrix of distances between protons in the molecules.

With software called AMBER, Madrid did “restrained” MD, a technique by which it is possible to find a family of 3D structures that fit the NMR data. “We did about ten of these simulations,” says Madrid, “and got a family of structures that are compatible with the experimental data.” This initial family of structures, however, had backbones that varied significantly from each other. These variations, the researchers realized, were attributable to intrinsic flexibility of the PNA backbone, which in turn limits the number of restraints that NMR can determine.

Fast turnaround of the many computations was essential to the success of the project.

So for the second part of the project, the researchers modified the PNA backbone — to make it more rigid, leading to more NMR restraints, which in turn would circumscribe the MD. The original PNA backbone contained several methylene groups — a carbon atom bonded with two hydrogens. Danith Ly, one of Achim’s collaborators, replaced one of the hydrogens of methylene groups with a methyl group — a carbon and three hydrogens. Because methyl groups occupy more space than a single hydrogen, they restrict flexibility, resulting in PNA with a stiffened backbone.

Side (top) and axial (bottom) view of PNA simulated by Achim, Madrid and colleagues. The substituted methyl groups are yellow. Other atoms are hydrogen (white), oxygen (red), nitrogen (blue), carbon (green).

Before doing MD with this revised PNA, Madrid calculated the changed charges between atoms due to the methyl group, which she needed as input to the MD “force fields” — mathematical expressions that facilitate MD by representing the quantum-mechanical energies between atoms. For this charge calculation, working with Achim’s grad students, Madrid turned to GAUSSIAN98 — a quantum chemistry program originally developed by Carnegie Mellon theoretical chemist John Pople, for which he received the 1998 Nobel Prize in Chemistry.

With revised charges and new NMR restraints, another round of MD computations produced a family of 3D structures that fit the NMR data and also fit well with the available crystallographic structure with much less variation in the backbone. “Handling the limitations imposed by the backbone,” says Achim, “was frustrating at the beginning, but it led us to the substitution with the methyl group and to understanding the PNA structure in very much detail. That’s the way research goes; you don’t know the answers when you start but everything falls in place like the letters in an interesting acrostic.”

The fast turnaround of the many computations, possible with the supercomputer named in honor of Pople and the help of Madrid, says Achim, was essential to the success of the project. “We are always anxious to get results and push the limits of our knowledge. Access to PSC computer resources and, even more, the help of Marcela made a huge difference in how fast we got the answers we were seeking.”

Looking Ahead

With the structure of their modified PNA in hand, Achim and her colleagues are proceeding with studies of metal-containing PNAs and electron transfer. “Now we can think about how the fundamental knowledge we have acquired can be translated into possible applications, both in nanotechnology as well in biology.”

In recent work, Achim and her students have created PNA molecules with copper, nickel and iron. David Waldeck, a University of Pittsburgh chemist, deposits those molecules on a gold electrode, creating self-assembled PNA monolayers. Waldeck then studies the electron flow through them. “He found that the mechanism of charge transfer in PNA,” says Achim, “is tunneling at short distances and hopping at long distances.”

Achim foresees that PSC’s new shared-memory system, Blacklight, will further advance her work with bio-mimetic molecules. “Based on the structures we have solved together, we have ideas about how to design additional nucleic-acid based structures. Those would be more complex, and require more computational effort and larger resources, so it is very exciting that these things come together. I look forward to continuing the collaboration, and to using this new resource.”