What is the basis of protein stability and binding specificity? We are interested in using combinatorial and statistical methods to understand and re-engineer protein stability and function. Our lab uses a wide variety of chemical and biophysical techniques, including library construction and screening, protein purification, spectroscopic, crystallographic and calorimetric analysis of proteins, bioinformatics and synthetic organic chemistry.

Combinatorial biophysics: Library approaches to the 'inverse' protein folding problem

The basis of protein stability is a fundamental problem in biophysics, but we are still far from a precise, quantitative model. Mutations that lead to protein instability or misfolding are involved in many diseases (cancer, cystic fibrosis, amyloidosis, etc.), and we would like to improve the stability of proteins for therapeutic or industrial uses.

The advent of genomics and proteomics has significantly improved the throughput of DNA sequencing and protein analysis. We can now make large libraries of protein variants, sort them using screens and selections, and rapidly determine the sequences of the selected variants. Specific variants can be purified and analyzed for stability and structure using traditional biophysical tools (CD, NMR, and X-ray crystallography). By assuming that a protein function requires a particular structure, we now have an experimental approach to the 'inverse' folding problem: what are the sequences that adopt a particular fold?

Rop, A Model Four-Helix Bundle Protein

Our initial focus is on the four-helix bundle protein Rop, because it is small, simple, easy to work with--and because we have developed a cell-based screen for its function. We have made libraries of Rop variants in the hydrophobic core, in the loop and on the surface to understand the roles of amino acids in these positions in achieving a stable, folded protein.

(Left) Rop is a simple four-helix bundle protein that is well-characterized. Rop controls the copy number of ColE1 plasmids by binding to primer and inhibitor RNAs associated with the ColE1 origin. (Right) By expressing green fluorescent protein (GFP) from a ColE1 plasmid, we can quickly know if a Rop variant is active based on cellular fluorescence.

High-Throughput Thermal Scanning (HTTS)

The folded variants of Rop that emerge from these functional screens vary considerably in their biophysical properties. We invented a high-throughput method of measuring the melting temperatures of related proteins variants. The method is based on the binding of hydrophobic dyes (like ANS) to the protein as the sample is heated (sometime called differential scanning fluorimetry, or DSF), and it is interfaced with high-throughput protein production and purification. The melting temperatures determined from this method are closely related to those from CD thermal denaturation.

(a) Schematic of HTTS. Hydrophobic dyes are quenched when excluded from well-folded protein, but they become fluorescent when embedded in molten globules and thermal unfolding intermediates. The apparent melting temperatures from HTTS (b) are closely related to those measured by CD thermal denaturation (c). (d) It is straightforward to measure the Tm's of 96 proteins at a time with this method.

Rop Engineering and Topology

To facilitate studies of Rop variant, we engineered a cysteine-free Rop that is as stable as wild-type Rop and virtually identical in structure, but unfolds much more quickly. (Wild-type Rop is an anomalously slow unfolder.)

The X-ray crystal structure of C38A C52V Rop shows that there is minimal perturbation of the overall structure, except for a small displacement of Phe56 from the larger valine at position 52.

We have collaborated with Ashok Deniz and Scripps and Jose Onuchic at UCSD to study the topological preferences of Rop core mutants using single-molecule spectroscopy. Surprisingly, some active variants of Rop exist as two different topological variants (parallel and antiparallel) in solution, which appear to exchange without unfolding.

In buffer, wild-type Rop shows low FRET and inactive AI6 Rop shows high FRET, indicating that AI6 is in the non-native parallel topology. Surprisingly, the active AL6 variant is also parallel. However, in 0.6 M guanidine, AL6 is found as a mix of parallel and antiparallel topologies.

Tumor Suppressor Proteins and Split-GFP Reassembly

Besides Rop, we are using high-throughput methods to study the stability determinants of tumor suppressor proteins, whose cancer-related mutations are often destabilizing. For example, we have modified the split-GFP reassembly system to examine the association of BRCA1 cancer-related variants with its partner BARD1. We found that some, but not all, interface mutants of BRCA1 interfere with its BARD1 binding.

Mutations found in breast cancer patients were made in the N-terminal RING domain of BRCA1, and it and the RING domain of BARD1 were fused to fragments of folding reporter GFP. Two variants, V11A and M18K, prevented binding and reassembly.

Leading References

• Magliery, T.J.; Lavinder, J.J. & Sullivan, B.J. (2011) "Protein stability by number: high-throughput and statistical approaches to one of protein science's most difficult problems," Curr. Opin. Chem. Biol. 15: 443-451, doi:10.1016/j.cbpa.2011.03.015. PDF
• Hari, S.B.; Byeon, C.; Lavinder, J.J. & Magliery, T.J. (2010) "Cysteine-free rop: A four-helix bundle core mutant has wild-type stability and structure but dramatically different unfolding kinetics," Protein Sci. 19: 670-679, doi:10.1002/pro.342. PDF
• Gambin, Y.; Schug, A.; Lemke, E.A.; Lavinder, J.J.; Ferreon, A.C.; Magliery, T.J.; Onuchic, J.N.; Deniz, A.A. (2009) "Direct single-molecule observation of a protein living in two opposed native structures," Proc. Natl. Acad. Sci. U.S.A.106: 10153-10158, doi:10.1073/pnas.0904461106. PDF
• Lavinder, J.J.; Hari, S.B.; Sullivan, B.J. & Magliery, T.J. (2009) "High-throughput thermal scanning: a general, rapid dye-binding thermal shift screen for protein engineering," J. Am. Chem. Soc., 131: 3794-3795, doi:10.1021/ja8049063. PDF
• Sarkar, M & Magliery, T.J. (2008) "Re-engineering a split-GFP reassembly screen to examine RING-domain interactions between BARD1 and BRCA1 mutants observed in cancer patients," Mol. BioSyst. 4: 599-605, doi:10.1039/b802481b. PDF
• Magliery, T.J. & Regan, L. (2004) “A cell-based screen for function of the four-helix bundle protein Rop: A new tool for combinatorial experiments in biophysics,” Protein Eng. Des. Select. 17: 77-83. PDF

Genomic design: Engineering enzymes from sequence statistics

Genomics efforts have yielded large numbers of putative proteins, many of which can be structurally or functionally categorized from sequence alone. In some cases, we can think of such families of sequences as a guide for engineering proteins and enzymes, since they represent the many ways nature solved a particular problem. Even very simple approaches, such as predicting stabilizing mutations from "consensus" sequences, have proven useful for improving enzyme stability or designing small protein motifs de novo.

Rather than merely asking what amino acid is the most common in a particular position in a protein family, we can ask more sophisticated questions like what is the distribution of amino acids, or how is the distribution of amino acids affected by the identity of other positions.

One question we can address this way is how correlated occurrences of amino acids lead to stability or function. We previously analyzed sequence co-variation in a small protein repeat motif, but we are now interested in studying sequence correlations in enzymes.

Consensus and Correlation in Triosephosphate Isomerase

We synthesized two related "consensus" variants of triosephosphate isomerase (TIM), using the most common amino acid at each positions in a "raw" and a "curated" database. We found that the curated consensus TIM had wild-type like properties--it was dimeric, stable and exhibited nearly diffusion-controlled kinetics. However, the raw consensus protein was monomeric, molten and weakly active. These proteins differed in only a small number of unconserved sites, but sites which are highly enriched in statistical correlations. We believe that "breaking" these coupled networks in the raw consensus resulted in non-native properties.

Two closely related consensus variants of TIM have drastically different properties, including oligomeric state and activity. The enzymes differ in a small number of unconserved positions, but those positions are enriched in residues that are strongly correlated to other positions.

We are examining the roles of correlated networks of residues in TIM, as well as other enzymes.

Leading References

• Sullivan, B.J.; Durani, V. & Magliery, T.J. (2011) "Triosephosphate Isomerase by Consensus Design: Dramatic Differences in Physical Properties and Activity of Related Variants," J. Mol. Biol., 413: 195-208, doi:10.1016/j.jmb.2011.08.001. PDF
• Magliery, T.J.; Lavinder, J.J. & Sullivan, B.J. (2011) "Protein stability by number: high-throughput and statistical approaches to one of protein science's most difficult problems," Curr. Opin. Chem. Biol. 15: 443-451, doi:10.1016/j.cbpa.2011.03.015. PDF
• Magliery, T.J. & Regan, L. (2004) “Beyond consensus: statistical free energies reveal hidden interactions in the design of a TPR motif,” J. Mol. Biol. 343: 731-745. PDF

Functional scanning mutagenesis

One of the most powerful methods for identifying residues important for structure, function or binding is alanine scanning mutagenesis, where positions in a protein are mutated to alanine and then tested for loss of activity or stability. Recently, robust methods have been introduced to expand the genetic code of bacteria and yeast, allowing some interesting "unnatural" amino acids to be site-specifically interested into proteins in living cells. Among those unnatural amino acids are ketones, which can be selectively modified, and benzophenones, which can be used as a photoaffinity label to trap interacting proteins in cells.

Synthetic Stop-Codon Scanning

One serious drawback to scanning mutagenesis is that it is labor-intensive: it takes a lot of work to create all the mutants. We have developed a very general DNA-synthetic approach to create versatile mutant-scanning libraries, with minimal reagent synthesis and full compatibility with commercially-available DNA synthesizers. These libraries can be used to scan natural or unnatural amino acids through proteins to aid in determining structures or binding partners, especially for proteins that are recalcitrant to more standard methods.

As a test, we scanned Ala residues through the Rop binding site, and found that in vivo screening did not perfectly match in vitro gel-shift results.

We also found that Bpa crosslinking of Rop trapped a tetramer, which is not detectable by equilibrium methods of detecting protein interactions.

Leading References

• Nie, L.; Lavinder, J.J.; Sarkar, M.; Stephany, K. & Magliery, T.J. (2011) "Synthetic approach to stop-codon scanning mutagenesis," J. Am. Chem. Soc. 133: 6177-6186, doi:10.1021/ja106894g. PDF
• Magliery, T.J. (2005) “Unnatural protein engineering: producing proteins with unnatural amino acids,” Med. Chem. Rev. Online 2: 303-323. PDF

Engineering paraoxonase I

One very practical use of our technology for improving protein stability and properties, and of our scanning mutagenesis technology, is to engineer proteins for better drug-like properties. As participants in the NIH Center of Excellence for Catalytic Bioscavenger Medical Defense Research, we are using protein engineering to understand better the mechanism of paraoxonase I (PON1), and to engineer it for improved properties as a therapeutic agent. PON1 shows therapeutic potential against organophosphorus nerve agents and pesticides, but it is not active enough against agents and not very drug-like. This work is a collaboration with Christopher Hadad at Ohio State, P. George Wang at Georgia State, Douglas Cerasoli at the U.S. Army Medical Research Institute of Chemical Defense, and Danny Tawfik and Joel Sussman at the Weizmann Institute in Israel.

Mutations at different positions in PON1 have dramatically different effects on the activities toward different substrates. In particular, mutations of His115 to Trp in both human and engineered chimeric PON1 ablate esterolytic activity but increase activity toward V agents.

The Center grant supporting this work was recently renewed, led by Cerasoli and Magliery. The first round of funding was led by David Lenz, who retired from USAMRICD in September, 2011, and Cerasoli. It involved partners at Ohio State, the Weizmann, Arizona State BioDesign Institute and HBRI in La Jolla, Ca. The grant is funded through the NINDS CounterACT Research Network.

Leading References

• Otto, T.C.; Harsh, C.K.; Yeung, D.T.; Magliery, T.J.; Cerasoli, D.M. & Lenz, D.E. (2009) "Dramatic difference in organophosphorpus hydrolase activity between huam and chimeric recombinant mammalian paraoxonase-1 enzymes," Biochemistry 48: 10416-10422, doi:10.1021/bi901161b. PDF
• Harel, M., Tawfik, D.S., et al. (2004) "Structure and evolution of the serum paraoxonase family of detoxifying and anti-atherosclerotic enzymes," Nat. Struct. Mol. Biol. 11: 412-419.