Ziyang Xu

Rational Design of Irreversible Covalent Inhibitors for E3 Ubiquitin Ligase NEDD4-1

  • Faculty Advisor

    Alexander Statsyuk

Published On

May 2016

Originally Published

NURJ 2015-16
Honors Thesis


In this research project, we have used the Fragment Based Drug Discovery approach to design irreversible covalent inhibitors for NEDD4-1, an E3 ubiquitin ligase found to be upregulated in several forms of cancer, including Ewing Sarcoma. 100 small molecules that contain an electrophilic acrylate warhead have been synthesized to tether the catalytic Cys867 of NEDD4-1 HECT covalently. Each molecule was characterized with NMR kinetic studies to evaluate their reactivity with N-acetylcysteine methyl ester, and 10 compounds with similar cysteine reactivity were screened against NEDD4-1 HECT simultaneously. Two compounds have been identified (with a mass spectrometry based screening assay) to label NEDD4-1 HECT irreversibly. Mutagenesis studies performed showed that both molecules react with Cys627 of NEDD4-1 HECT rather than Cys867. Biochemical assays performed show that the initial hit inhibited NEDD4-1 meditated poly-ubiquitination in vitro at 1mM concentration and subsequent X-ray crystallographic studies of NEDD4-1 bound to one of the inhibitors (ZX1) shows that ZX1 binds the N-lobe of NEDD4-1 HECT at the ubiquitin-binding site. Structure Activity Relationship studies were conducted to enhance the potency of the initial hit, and a cyclopentylindole analogue (ZX2) was found to be much more potent than the initial hit. Fluorescence cross-polarization assays were finally used to determine the binding affinity KI and the inhibition rate constant kinact of ZX2. Additional SAR studies would be performed in the future to further enhance the potency and selectivity of the compound.


1.1 The Ubiquitin-Proteasome System (UPS)

The Ubiquitin-Proteasome System is a key strategy employed by eukaryotic cells to regulate proteins post-translationally (Ciechanover 2013). The UPS pathway is compromised in several diseases, including cancer(Wang 2007), neurodegenerative diseases (Tofaris 2011) and viral infection (Yasuda 2003). This thesis explores the design, synthesis, and characterization of irreversible covalent inhibitors for NEDD4-1 (neural precursor cell expressed developmentally down-regulated protein 4), an E3 ligase in the UPS pathway that regulates tumor suppressor PTEN (phosphatase and tensin homolog), α-synuclein, and ebola viral particles, among some of its other substrates.

The UPS pathway consists a cascade of E1-E2-E3 enzymes that relay ubiquitin molecules, consisted of 76 amino acid residues, to specific protein substrates (Figure 1.1) (Roos-Mattjus 2004). The E1 enzyme, or ubiquitin-activating enzyme, catalyzes a two-step reaction and consumes 1 ATP molecule. E1 enzyme would first bind to ubiquitin and ATP and catalyze adenylation of the C-terminus of ubiquitin, with the release of a pyrophosphate molecule. The catalytic cysteine residue on the E1 enzyme would subsequently react with the activated ubiquitin molecule, resulting in the release of 1 AMP molecule and the formation of thio-ester bond between the C-terminus of ubiquitin and the sulfhydryl group of the E1 catalytic cysteine (Lee and Schindelin 2008). In the next step, E2 enzyme, or ubiquitin-conjugating enzyme, catalyzes trans-thioesterification reaction by binding to both the E1 enzyme and the activated ubiquitin molecule, and allowing the E2 catalytic cysteine to react with the C-terminus of the ubiquitin. Free E1 enzyme is released in this process (Lee and Schindelin 2008). In the subsequent step, two classes of E3 enzymes, also known as ubiquitin ligases, catalyze the conjugation of the ubiquitin molecules to the selected substrates. The ring E3 enzymes catalyze the direct transfer of the ubiquitin molecule from E2 to the substrates by bring them in close proximity and allowing the lysine residue on the substrate to form an isopeptide bond with the C-terminus of ubiquitin (Berndsen and Wolberger 2014). The HECT E3, on the other hand, would form an E3-ubiquitin intermediate first when its catalytic cysteine residue reacts with the ubiquitin carboxy-terminus, releasing E2 in the process. The lysine residue on the substrate would subsequently react with this activated ubiquitin to form an isopeptide bond (Kamadurai, Qiu et al. 2013). Particularly, NEDD4-1 studied in this thesis is an HECT E3 ligase.

Once the initial substrate-ubiquitin conjugation has taken place, this process could be repeated to build a poly-ubiquitin chain on the substrate. Specifically, the C-terminus of the newly added ubiquitin molecule would be conjugated to one of the lysine residues on the preceding ubiquitin. Since the ubiquitin molecule contains several lysine residues, poly-ubiquitin chains that involve different lysine residues could be targeted to different locales. For example, poly-ubiquitin chain that involves K48 linkage would target the substrates to the 26S proteasome for degradation, while K63 chain targets the substrates to cellular lysosome. Number of the ubiquitin molecules in the chain could also determine destinations of the substrates. For example, when PTEN is poly-ubiquitinated (K48), it would be degraded by the proteasome; alternatively when PTEN is mono-ubiquitinated, it would be transported into the nucleus to regulate gene transcription.

1.2 Structure of NEDD4-1 and its Implications in human diseases

1.2.1 Domain Structure of NEDD4-1

The domain structure of NEDD4-1 is well characterized—it contains a C2 domain, 4 WW domains and a catalytic HECT domain (Figure 1.2a and b)(Ingham 2004). The C2 domain regulates NEDD4-1 functions by constitutively binding to the HECT domain and auto-inhibiting the ligase activity of NEDD4-1. When intracellular Ca2+ level is elevated, Ca2+ binds to the C2 domain and causes the release of the HECT domain and activation of the ligase. The four WW domains of NEDD4-1 interact with proline-rich motifs, such as PPxY, and render substrate specificity. Finally, the catalytic HECT domains consists of an N-lobe at the N-terminus and a C-lobe at the C-terminus linked through a flexible hinge. Published Crystal structure of NEDD4-1 bound to UbcH7 (an E2 enzyme) suggests that an incoming E2 enzyme would dock at the N-lobe of the HECT domain (Figure 1.3). C-lobe of the HECT, on the other hand, contains the active site. Upon docking of the E2 enzyme, the flexible hinge would rotate about so as to bring the catalytic Cys867 on the C-lobe into close proximity with the activated ubiquitin molecule bound to the E2 enzyme, and a trans-thioesterification reaction would subsequently charge the catalytic Cys867 with ubiquitin. Once the substrate has been mono-ubiquitinated, the newly added ubiquitin molecule would dock at the ubiquitin-binding domain on the N-lobe, and the catalytic Cys867 would be charged again with the second ubiquitin for chain extension (Figure 1.4). This ubiquitin-binding domain on the N-lobe is therefore crucial to the processivity of NEDD4 ligase. It is reported that another cysteine residue Cys627 is also present at this site (Ingham 2004).

1.2.2 Implications of NEDD4-1 in Parkinson’s Disease

α-synuclein, a hallmark and contributing factor of Parkinson’s disease, is found to be regulated by NEDD4-1 mediated poly-ubiquitination (Stefanis 2012). NEDD4-1 recognizes the carboxy-terminus of α-synuclein and attaches a K63-linked poly-ubiquitin chain to its lysine residue, thereby targeting α-synuclein to cellular lysosome for degradation(Sugeno 2014). NEDD4-1 is down-regulated in neurons that contain α-synuclein aggregates. It is hoped that a pharmacological probe synthesized in this project could be used to study if selective inhibition of NEDD4-1 mediated poly-ubiquitination could impede degradation of α-synuclein in neurons.

1.2.3 Role of NEDD4-1 in Ebola Viral Budding

A study by Yasuda et. al. reports that NEDD4-1 facilitates budding of Ebola Viral-Like Particles (VLP) possibly through the interaction of its WW domains with the PPxY motifs on the viral particles (Yasuda 2003). It is not known whether NEDD4-1 mediated ubiquitination is crucial to this pathway, although Rsp5p, an ortholog of NEDD4-1 in yeast, could potentially ubiquitinate the viral particle VP40 in vitro (Harty, Brown et al. 2000). The pharmacological probe synthesized in this study selectively inhibits the NEDD4-1 poly-ubiquitination pathway but leaves the WW domains in tact. This compound can be used to investigate if poly-ubiquitination of VP40 is indeed necessary for viral-budding.

1.2.4 Regulations of Tumor Suppressor Protein PTEN by NEDD4-1

Shi et. al. reports in a previous study that NEDD4-1 could antagonize PTEN and activate the IGF1-PIP3-AKT pathway and is potentially implicated in Ewing Sarcoma and breast cancer (Shi and Jiang 2014). Upon binding of IGF1 (Insulin Growth Factor 1), the tyrosine kinase receptor protein IGF1R would undergo auto-phosphorylation and become activated. The IGF1R-P would subsequently phosphorylate its downstream substrate IRS1 (Insulin Receptor Substrate 1) at its tyrosine residues. Phosphorylated IRS1 (or IRS1-P) contains a domain that specifically recruits kinase protein PI3K to the plasma membrane to convert the lipid phosphate PIP2 to PIP3 by hydrolyzing ATP. PIP3 acts as a messenger to activate the AKT kinase pathway, resulting in transcriptions of genes necessary for cell survival and proliferation. The membrane protein PTEN antagonizes this pathway as the following. First, as a protein tyrosine phosphatase, it expedites the dephosphorylation of IRS1-P to IRS1. Secondly, as a lipid phosphatase, it converts PIP3 to PIP2 and thereby inhibits AKT. Finally, the study shows that NEDD4-1 selectively inhibits the protein tyrosine phosphatase activity of PTEN and allows the accumulation of IRS1-P to amplify AKT signaling (Figure 1.5). It is known that some cancer cell uses this pathway to drive its proliferation through autocrine regulation. This type of cancer cell secretes insulin into the extracellular space. Binding of extracellular IGF1 to its IGF1R would in turn drive AKT signaling (Clemmons 2007). Details of how NEDD4-1 regulates PTEN in this specific pathway are not known. It has been hypothesized that NEDD4-1 could potentially antagonize the tyrosine phosphatase activity of PTEN by building K63 poly-ubiquitin chain. Our covalent processivity inhibitor of NEDD4-1 could be used to study the mechanism by which NEDD4-1 regulates PTEN in the IRS1-PI3K-AKT pathway and be fully optimized into a potential anti-cancer compound.

In another study reported by Trotman et. al., NEDD4-1 regulates PTEN in a different fashion (Trotman, Wang et al. 2007). Trotman argues that poly-ubiquitination of PTEN results in its proteasomal degradation, whereas mono-ubiquitination of PTEN results in its nuclear import and suppression of the AKT pathway (Figure 1.6). In either case, we predict that inhibiting poly-ubiquitination of PTEN would antagonize the AKT pathway and arrest cancer cell proliferation.

1.3 Fragment-Based Drug Design of Irreversible Covalent Inhibitors of NEDD4-1

1.3.1 Comparisons of High-Throughput Screening and Fragment-Based Drug Discovery towards the Design of Inhibitors

In High-Throughput Screening (HTS), tens of thousands or even millions of compounds are screened against the biological target. The average molecular weight of compounds used for HTS is around 500Da, and nano-molar binders of the biological target can be found (Ungermannova, Lee et al. 2013). In contrast, Fragment-Based Drug Discovery (FBDD) attempts to discover weak milli-molar binders with an average molecular weight of 200Da, and would subsequently make these “hits” more potent by growing or combining the fragments to make larger molecules (Murray and Rees 2009). The compound library used for FBDD could be much smaller—as few as 100 compounds could be screened against the biological target to find a potential inhibitor. By using smaller molecules, FBDD improves the “hit” rate by maximizing the ligand efficiency and minimizing unfavorable steric interactions. Additionally, since a smaller library could be used for initial screening, syntheses of the compounds and screening of the compounds against biological targets could be done manually.

1.3.2 Principles to the Design of NEDD4-1 Inhibitors

As an HECT E3 ligase, NEDD4-1 forms an ubiquitin-thioester intermediate before it relays the ubiquitin tag onto its substrates. As such, the catalytic cysteine residue (Cys867) is necessary for the ubiquitylation of NEDD4-1 substrates. A small molecule that covalently labels Cys867 will thus inhibit the functions of NEDD4-1. An inhibitor that reacts covalently and irreversibly with the proteins usually has an affinity much higher than those of reversible inhibitors. By designing irreversible inhibitors, a weak binder can be identified from the library in the initial screen and subsequently optimized to enhance its potency. Ideally, the small molecule should be a combination of an electrophilic warhead that reacts covalently with the cysteine residue through a Michael reaction, and a non-polar ring-based targeting element that docks the molecule selectively and potently at the catalytic site of NEDD4-1 prior to covalent inhibition.

Alternatively, we can also design inhibitors that target the cysteine residue 627, which is found at the processivity site of NEDD4-1 and is necessary for the poly-ubiquitylation of the NEDD4-1 substrates.

Unlike the design of the kinase inhibitors (many of which are substrate or transition state analogues of the enzyme and bear structural resemblances to ATP), our inhibitors ought to disrupt protein-protein interactions (Silverman 2014). To maximize our chance of finding a selective inhibitor from our library, we synthesized our compounds by coupling an electrophile that resembles the C-terminus of ubiquitin to different non-polar ring-based fragments from an unbiased commercially available library. In the process, a selective inhibitor that targets the cysteine residue 627 of NEDD4-1 has been found, characterized with X-ray Crystallography and optimized with a series of Structure-Activity Relationship (SAR) studies.

Download full thesis

Ziyang Xu


First and for most, I would like to acknowledge Dr. Alexander Statsyuk for his guidance and support throughout my undergraduate studies. Since I joined the Statsyuk laboratory in Spring 2012, Dr. Statsyuk had offered me extremely valuable advice on my research project and on my career development. This thesis work would not have been completed without his keen advice and insights. I would also like to thank Stefan Kathman, a graduate student with whom I have worked with on the NEDD4-1 project. I have learnt and refined my organic synthetic and analytical techniques working with him, and we have overcome many challenges together along the way. Specifically, while I accomplished the synthesis of the majority of the 100 compounds in our library, characterized some of them with NMR kinetic studies and conducted additional SAR studies to refine our “hit”, Stefan expressed NEDD4-1 HECT, screened our compounds against the protein, grew the crystals of NEDD4-1 inhibitor complex for X-ray crystallization, and designed various biochemical assays. Finally, I would like to acknowledge the funding and support of CLP Lambert Fellowship and thank Dr. Andrew Chan especially for providing me with this precious opportunity to work on a project that explores the intersection of science and medicine. I would also like to thank Northwestern University for the great scientific training I have received and for shaping my interests in academic medicine.


Anan, T. and M. Nakao (1998). Genes Cells 3(11): 751-763.

Berndsen, C. and C. Wolberger (2014). Nature Struc. & Mol. Biol. 21: 301-307.

Ciechanover, A. (2013). Bioorg. Med. Chem. 21: 3400.

Clemmons, D. (2007). Nature Reviews Drug Discovery 6: 821-833.

Congreve, M., et al. (2003). Drug Discov Today. 8(19): 876-877.

Fukami, T. and T. Yokoi (2012). Drug Metab Pharmacokinet. 27(5): 466-477.

Harty, R., et al. (2000). Proc Natl Acad Sci 97(25): 13871-13876.

Ingham, R. (2004). Oncogene 23: 1972-1984.

Kaiser, S., et al. (2011). Nature Methods. 8(8): 691-696.

Kamadurai, H., et al. (2013) Elife 2:e00828

Kathman, S., et al. (2014). J. Med. Chem. 57: 4969-4974.

Lee, I. and H. Schindelin (2008). Cell 134(2): 268-278.

Mari, S. (2014). Structure 22: 1639-1649.

Maspero, E. (2013). Nature Struc. & Mol. Biol. 20(6): 696-700.

Maspero, E., et al. (2011). Embo Rep 12: 342-349.

McRee, D. (1993). Practical Protein Crystallography. San Diego, Academic Press.

Murray, C. W. and D. C. Rees (2009). Nature Chem. 1: 187-192.

Roos-Mattjus, P. (2004). Ann Med. 36(4): 285-295.

Rossi, A. and C. Taylor (2011). Nature Protocol 6: 365-387.

Shi, Y. and X. Jiang (2014). Nature Struc. & Mol. Biol. 21(6): 522-527.

Silverman, R. (2014). San Diego, Academic Press.

Stefanis, L. (2012). Cold Spring Harb Perspect Med 2(2): a009399.

Sugeno, N. (2014). J Biol Chem. 289(26): 18137-18151.

Tofaris, G. (2011). Proc Natl Acad Sci 108(41): 17004-17000.

Trotman, L., et al. (2007). Cell 128(1): 141-156.

Ungermannova, D., et al. (2013). J Biomol Screen. 18(8): 910-920.

Wadsworth, W. S. and W. D. Emmons (1977). Org. React. 25: 73-253.

Wang, X. (2007). Cell 128(1): 129-139.

Yang, B. and S. Kumar (2010). Cell Death Differ. 17(1): 68.

Yasuda, J. (2003). J Virol 77(18): 9987-9992.

Zheng, T., et al. (2013). Bioconjugate Chem. 24(6): 859-864.