Virginia Cornish
Helena Rubinstein Professor
Department of Chemistry

work : +1 212-854-5209

Cornish Group

Research: Chemical Complementation


Directed evolution offers a new approach for manipulating and understanding protein function.  Directed evolution involves generating large pools of protein variants and then assaying these variants en masse for the desired function.  While a powerful tool, directed evolution is restricted to enzymes that are inherently screenable or selectable--for example, enzymes where the product is fluorescent or an essential metabolite.  Thus, at the start of our research program, my laboratory has focused on developing a general, high-throughput assay for enzyme catalysis based on the yeast three-hybrid assay ("Chemical Complementation") that should allow directed evolution to be applied to a broad range of chemical reactions.  This assay detects enzyme catalysis of bond formation or bond cleavage reactions based on covalent coupling of two small molecule ligands in vivo.  The heterodimeric ligand reconstitutes a transcriptional activator, turning on transcription of a reporter gene.  Bond formation is detected as activation of an essential reporter gene; bond cleavage, repression of a toxic reporter gene.  The assay is high-throughput because it can be run as a growth selection where only the cells containing functional enzyme survive.  The assay can be readily extended to new chemistry simply by synthesizing dimeric ligands with different substrates as chemical linkers.  Such a general assay should find broad use not only in directed evolution, but also proteomics, drug discovery, and enzymology.


First, we reported the small molecule transcription assay that is the basis of chemical complementation.  This work was reported in J. Am. Chem. Soc. and featured in Chem. & Eng. News.  Then, we demonstrated chemical complementation using a well-studied enzyme-catalyzed reaction, b-lactam hydrolysis by a b-lactamase.  This work was published in Proc. Natl. Acad. Sci. USA and featured in both Proc. Natl. Acad. Sci. USA and Chem. & Eng. News.  The long-term goal is to apply this assay to directed evolution, enzymology, drug discovery, and proteomics.  Currently we are focused on two applications.  The first, a basic science application, to understand the molecular basis for the difference in chemical reactivity between two proteins believed to be evolutionarily related, a penicillin-binding protein and a b-lactamase.  The second, to use directed evolution to engineer glycosynthase enzymes that can be used for the synthesis of carbohydrates.  Recently, we have completed our first efforts in both areas. 

Back to Top  

Dexamethasone-Methotrexate Yeast Three-Hybrid Transcription Assay

[Published in H. Lin, W. Abida, R.T. Sauer, V.W. Cornish.  “Dexamethasone-Methotrexate:  An Efficient Chemical Inducer of Protein Dimerization In Vivo.”  J. Am. Chem. Soc., 122, 4247-4248 (2000).  Featured in Chem. & Eng. News, 78, 52 (2000).].


The first step to building the chemical complementation assay was to design a heterodimeric ligand that could efficiently reconstitute a transcriptional activator in vivo.  Essentially what this required was that we develop ligand-receptor pairs other than FK506 or FK506 analogs that could be used as “Chemical Inducers of Dimerization” (CIDs).  Methotrexate (Mtx) and dihydrofolate reductase (DHFR) were chosen because of the ease with which Mtx analogs can be synthesized and the high affinity of the Mtx-DHFR interaction (KD = 10 pM).  Dexamethasone (Dex) and the hormone-binding domain of the glucocorticoid receptor (GR) had been reported previously.  We demonstrated the efficacy of the Dex-Mtx CID based on activation of a lacZ reporter gene in a yeast three-hybrid assay.  The ease of synthesis of Dex-Mtx allows the linker between Dex and Mtx to be changed readily, and, thus, chemical complementation to be applied to a wide range of chemical reactions.  Proof of the advantages of this CID, GPC Biotech just reported in Chem. Biol. the use of our Mtx yeast three-hybrid system to discover both known and novel protein targets of cyclin-dependent kinase inhibitors. 

Back to Top  

Chemical Complementation: Proof-of-Principle with a beta-Lactamase Enzyme.

[K. Baker, C. Bleczinski, H. Lin, G. Salazar-Jimenez, D. Sengupta, S. Krane, V.W. Cornish.  "Chemical Complementation: A Reaction-Independent, High-Throughput Genetic Assay for Enzyme Catalysis."  Proc. Natl. Acad. Sci. USA, 99, 16537-16542 (2002).  Featured in a commentary in Proc. Natl. Acad. Sci. USA, 99, 16513-16515 (2002) and in Chem. & Eng. News, 81, 24, (2003).].

The well-studied enzyme-catalyzed reaction of cephalosporin hydrolysis by the P99 cephalosporinase was used to develop and demonstrate the chemical complementation assay.  A Dex-Cephem-Mtx CID with the cephem substrate as the linkage between Dex and Mtx was synthesized.  Using standard lacZ transcription assays, transcription activation was shown to be turned off by expression of the cephalosporinase enzyme, presumably because the enzyme catalyzes hydrolysis of the cephem bond.  To confirm that the change in transcription of the reporter gene was in fact caused by enzymatic turnover of the Dex-Cephem-Mtx substrate, lacZ transcription was shown to be unaffected by expression of an inactive P99:S64A mutant.  Finally, a lacZ screen was used to isolate the wild-type cephalosporinase from a pool of inactive variants.

Further, we have shown that the levels of lacZ transcription correlate with the catalytic efficiency of a series of cephalosporinase mutants [D. Sengupta et al. (V.W. Cornish), Biochemistry, 43, 3570-3581 (2004)].  The lacZ screen has been used to investigate the mechanism of cephalosporinase resistance to third-generation cephalosporin antibiotics [B. Carter and V.W. Cornish), unpublished results]. 

 Back to Top  

Molecular Basis for the Difference in Reactivity Between Penicillin-Binding Proteins and beta-Lactamases.

[S. Goldberg, W. Iannuccilli, T. Nguyen, J. Ju, V.W. Cornish.  "Identification of Residues Critical for Catalysis in a Class C b-Lactamase by Combinatorial Scanning Mutagenesis."  Protein Science, 12, 1633-1645 (2003)B.F. Gherman, S.D. Goldberg, V.W. Cornish, R.A. Friesner.  "Mixed Quantum Mechanical/Molecular Mechanical (QM/MM) Study of the Deacylation Reaction in a Penicillin Binding Protein (PBP) Versus in a Class C b-Lactamase."  J. Am. Chem. Soc., accepted.  S. Goldberg, V.W. Cornish.  "Reenacting the Evolution of a b-Lactamase from a Penicillin-Binding Protein."  in preparation. Featured in Chem. & Eng. News, 81, 35-36, 38-40 (2004).].

Directed evolution offers a new tool for understanding protein structure and function by allowing us to watch proteins acquire new function through mutation of the amino acid sequence.  Here, in collaboration with the Friesner group at Columbia, we are combining directed evolution with computational studies to understand the difference in reactivity between two proteins believed to be evolutionarily related.  Penicillin-binding proteins (PBPs) are the targets of b-lactam antibiotics, and b-lactamases are the bacterial resistance enzymes that hydrolyze and inactivate these antibiotics.  b-Lactamases are believed to have evolved from an ancestral PBP.  Interestingly, the two proteins have conserved three-dimensional structures and active-site residues.  Both are serine-protease type enzymes.  PBPs are inactivated by b-lactams, while b-lactamases turn them over, because PBPs are inefficient catalysts of acyl-enzyme hydrolysis.  The rate constant for acyl-enzyme hydrolysis differs by six-orders of magnitude for these two proteins with typical substrates.  QM/MM calculations of the ground states and transition states for catalysis of this step by the R61 transpeptidase PBP and the P99 b-lactamase suggest that the difference in reactivity stems from the ability of the active-site Tyr general base to interact with a stabilizing hydrogen-bonding network in the acyl-enzyme complex.  “Isostere Scanning” of the P99 b-lactamase and directed evolution of the PBP support this hypothesis.

Back to Top  

Engineering Glycosynthases for Use in Carbohydrate Synthesis

[H. Lin, H. Tao, V.W. Cornish.  "Directed Evolution of a Glycosynthase Via Chemical Complementation." J. Am. Chem. Soc., 126, 15051-15059 (2004).]

Conversion of a  glycosidase to a glycosynthase.  A glycosidase has two acidic residues in the active site, one acting as the general acid/base, and the other acting as the nucleophile to attack the anomeric position where cleavage of the glycosidic bond is going to occur.  If the nucleophilic acidic residue is changed to a small hydrophobic residue, such as a Ser or Ala, the protein will not be able to hydrolyze glycosidic bonds since it lacks the nucleophile.  But it can act as a glycosynthase, accepting a glycosyl fluoride as the glycosyl donor and forming glycosidic bonds with suitable glycosyl acceptors.

Despite their fundamental role in biological processes and potential use as therapeutics, it still remains difficult to synthesize carbohydrates.  Enzymes, with their control of both regio- and stereo-chemistry, provide an obvious alternative to traditional small molecule chemistry for carbohydrate synthesis.  Recently, Withers and co-workers demonstrated that retaining glycosidases can be re-engineered to glycosynthases simply by mutating the nucleophilic Glu residue at the base of the active site to a small hydrophobic residue and using an a-fluoro donor.  While several retaining glycosidases have now been reported, these glycosynthase variants are not sufficiently active for use on a preparative scale.  Directed evolution should offer a powerful approach for improving the activity of these enzymes and even modifying their substrate specificity; however, there are no natural selections for glycosynthase activity.  Chemical complementation should open up directed evolution to carbohydrate chemistry.  In this first paper, we showed that chemical complementation can link Cel7B:E197A glycosynthase activity to LEU2 transcription activation in vivo using Dex disaccharide acceptor and Mtx disaccharide a-fluoro donor substrates.  The LEU2 selection was then used for an E197 saturation library, yielding mutations that increased the activity of the glycosynthase and that increased the expression levels of the protein. Back to Top  

Ligand-Receptor Binding

[K. deFelipe, E. Althoff, B. Carter, V.W. Cornish.  "Correlation between ligand-receptor affinity and the transcription readout in a yeast three-hybrid system," Biochemistry, 43, 10353-10363.]

In addition to catalysis, we are exploiting the yeast three-hybrid assay to study ligand-receptor binding.  It was proposed that the yeast three-hybrid assay could be used to identify the protein targets of small molecule drugs.  The high-affinity of the Mtx/DHFR CID has improved the sensitivity of this assay and allowed us to go beyond initial proof-of-principle experiments.  We have shown that the Mtx yeast three-hybrid system can detect small molecule-protein interactions with dissociation constants up to ca. 100 nM.  In collaboration with Prof. Young-Tae Chang at New York University, we have applied the Mtx yeast three-hybrid system to an important pharmaceutical target, cyclin-dependent kinases (unpublished data).

Back to Top  

Home > Research > Chemical Complementation