1. The Bh3 Domain of Puma: Structure Determination and Small Molecular Inhibitor Design Nicki Zevola Bahar Lab Lab Meeting March 25, 2009

2. Intro. What are BCl-2 proteins? A class of pro-survival and pro-apoptotic proteins –Easily distinguishable –Pro-survival: Bcl XL, Bcl-w, MCl- 1, A1, Bax, Bak 2-4 BH domains –Pro-apoptotic: Bim, Bad, Bmf, Bid, Bik, Hrk, Noxa, Puma Only 1 BH domain, Bh3 Key interactions occur when pro-apoptotic proteins (e.g., Puma) inhibit pro-survival proteins (e.g., Bcl XL ) Crystal structure of BCl-xl (mouse) Crystal structure of Puma (mouse) 1 3 2 1

3. Project Overview Goal of Project: Develop small molecule mitochondrial-targeted drugs to inhibit the binding of the Puma Bh3 domain to Bcl-2 family proteins. Normally, Puma Bh3 domain:Bcl-2 protein binding antagonizes anti- apoptotic Bcl-2 proteins, inhibiting Apaf-1, and resulting in caspase release and apoptosis. –Studies with Puma-deficient mice have shown that Puma not only activates, but is required for, apoptosis induced by oncogenes and DNA-damaging agents. 1,2 1 Qui et. al. PUMA Regulates Intestinal Progenitor Cell Radiosensitivity and Gastrointestinal Syndrome. Cell Stem Cell. 2008. 2(6):576-583. 2 Yu J., Zhang L., Hwang PM, Kinzler K.W., Vogelstein B. Puma induces the rapid apoptosis of colorectal cancer cells. Mol. Cell. 2001. 7:3:673-82.

4. How We Will Find Small Molecular Inhibitors for Puma (Project Design) Analyze structure of Bh3 domain of Puma Determine key conserved interactions between Puma Bh3 domain and Bcl-2 proteins Develop pharmacophore models using conserved interactions Screen Zinc 8.0 database for lead- like compounds that match pharmacophore model Develop Puma models to prioritize hits In vitro testing

5. Step 1. Analyze structure of the Bh3 Domain of Puma

6. Day et. al.: How Puma Bh3 Domain Bh3-Mcl1 and Noxa Bh3-Mcl1 Structures were Obtained 1. Two samples made –1. 13 C, 15 N-labeled peptide + unlabeled Mcl-1 –2. Unlabeled peptide + labeled Mcl-1 2. NMR techniques used for resonance assignments 3. Structures evaluated for proper geometry –No distance violation > 0.2 Å –No angle violation > 5 o Why only the Bh3 Domain of Puma, Noxa with Mcl-1? Puma and Noxa have extremely unstable structures, and hence only the Bh3 binding domain structure has been accurately determined. In the Bahar lab, we have used the i-Tasser server (Zhang) to help ascertain the entire Puma structure (see next slide). (2rod) (2roc)

7. i-Tasser Sequence of hPUMA submitted to online server developed by Dr. Yang Zhang, Ph.D. (University of Kansas) –Finalist at CASP (Critical Assessment of Structural Predictions) Competition Server uses LOMETS (a form of threading) in structural prediction 5 models generated for hPuma –Problem: Best C-score: -3.46 Will be addressed later

8. Step 2. Determine key conserved interactions between Puma Bh3 domain and BCl-2 proteins

9. Day et. al.: Comparison of MCl-1 complexes with other BCl-2 proteins 1. Superimposition of Mcl-1:Puma with: BCl XL :Bim, A1:Puma, and Bcl XL –RMSD of backbone BCl XL :Bim and Mcl-1: Puma: 1.5 Å –Most notable structural difference: Bcl XL has a long loop between α1 and α2 not found in Mcl-1 or A1

10. Day et. al.: Comparison of Mcl-1 complexes with other Bcl-2 proteins (con’t) 2. Structural Alignment –Sequence comparison (excluding Bcl XL loop residues): Bcl XL,Mcl-1: 25.6% identity, 40.0% similarity A1, Mcl-1: 25.4% identity, 46.% similarity

11. Day et. al.: The Consensus Bh3 Domain Binding Motif The 13 residue consensus sequence that defines the Bh3 domain in PUMA is: φ1ΣXXφ2XXφ3ΣDZφ4L –where φ1-φ4 are hydrophobic residues –Σ are small residues –Z is usually an acidic residue –L is a hydrophilic residue capable of forming an intermolecular cap –D is the conserved aspartate Bh3 domains of Noxa (in yellow, left) and Puma (in pink, right).

12. in situ Mutagenesis Reveals Other Key Residues (Yu et. al., 2009) Already known from Day et. al.: conserved interactions with Bcl2 proteins Jian Yu et. al. (Univ. of Pittsburgh): 2 specific point mutations alter binding specificity upstream of the Bh3 binding domain! –These are therefore included in our pharmacophore model (next step) Images courtesy Gabriela Mustata, Ph.D.

13. Step 3. Develop pharmacophore models using conserved interactions

14. Pharmacophore Model Development Our superimposition of our hPuma-A1 (from i-Tasser) with hPUMA-Mcl1 (from 2roc) structures resulted in an RMSD of 0.55Å! Four key interactions of Bcl2 proteins and hPuma (see next slide): –Phe251.CB (Bcl2) – Leu141.CD1 (hPuma) –Ph395.CD1 (Bcl2) – Leu141.CD1 (hPuma) –Asp237.OD2 (Bcl2) – Asp142.NH1 (hPuma) –Arg244.NE (Bcl2) – Asp146.OD1 (hPuma)

15. Pharmacophore Model Development Slide courtesy Dr. Gabriela Mustata, Ph.D.

16. Step 4. Screen Zinc 8.0 database for compounds that match pharmacophore model

17. Zinc 8.0 Database Searching

18. Pharmacophore Search Results Thus Far Zinc 7.0 Lead-like Database Found 1 hit Zinc 8.0 Lead-like Database So far…Screened 24/37 of database (about 1.3 x 10 6 compounds) Found 20 hits thus far (some shown below)

19. Steps 5 and 6 (to come). Develop Puma models to prioritize hits; in vitro testing.

20. Future Steps

21. Difficulties Encountered Thus Far

22. Current Problem: i-Tasser model 5 models of hPuma generated by i-Tasser are not entirely satisfactory, as shown by Fast Contact server (Camacho et. al.) Current solution: Use MD Simulations for energy minimization of i-Tasser models, recalculate energies on Fast Contact server

23. Questions? Ideas? Comments? A special thank you: Dr. Ivet Bahar; Dr. Gabriela Mustata; Dr. Jian Yu; Dr. Joel Greenberger; Ahmet Bakan (for teaching MD Simulations); everyone else for listening,