Publications

The normal expression, cell surface localization, and function of the murine high density lipoprotein receptor scavenger receptor class B type I (SR-BI) in hepatocytes in vivo, and thus normal lipoprotein metabolism, depend on its four PDZ domain (PDZ1-PDZ4) containing cytoplasmic adaptor protein PDZK1. Previous studies showed that the C terminus of SR-BI ("target peptide") binds directly to PDZ1 and influences hepatic SR-BI protein expression. Unexpectedly an inactivating mutation in PDZ1 (Tyr(20) → Ala) only partially, rather than completely, suppresses the ability of PDZK1 to control hepatic SR-BI. We used isothermal titration calorimetry to show that PDZ3, but not PDZ2 or PDZ4, can also bind the target peptide (K(d) = 37.0 μm), albeit with ∼10-fold lower affinity than PDZ1. This binding is abrogated by a Tyr(253) → Ala substitution. Comparison of the 1.5-Å resolution crystal structure of PDZ3 with its bound target peptide ((505)QEAKL(509)) to that of peptide-bound PDZ1 indicated fewer target peptide stabilizing atomic interactions (hydrogen bonds and hydrophobic interactions) in PDZ3. A double (Tyr(20) → Ala (PDZ1) + Tyr(253) → Ala (PDZ3)) substitution abrogated all target peptide binding to PDZK1. In vivo hepatic expression of a singly substituted (Tyr(253) → Ala (PDZ3)) PDZK1 transgene (Tg) was able to correct all of the SR-BI-related defects in PDZK1 knock-out mice, whereas the doubly substituted [Tyr(20) → Ala (PDZ1) + Tyr(253) → Ala (PDZ3)]Tg was unable to correct these defects. Thus, we conclude that PDZK1-mediated control of hepatic SR-BI requires direct binding of the SR-BI C terminus to either the PDZ1 or PDZ3 domains, and that binding to both domains simultaneously is not required for PDZK1 control of hepatic SR-BI.

Reactive molecules have diverse effects on cells and contribute to several pathological conditions. Cells have evolved complex protective systems to neutralize these molecules and restore redox homeostasis. Previously, we showed that association of nuclear factor (NF)-erythroid-derived 2 (E2)-related factor 2 (NRF2) with the nuclear matrix protein NRP/B was essential for the transcriptional activity of NRF2 target genes in tumor cells. The present study demonstrates the molecular mechanism by which NRP/B, via NRF2, modulates the transcriptional activity of antioxidant response element (ARE)-driven genes. NRP/B is localized in the nucleus of primary brain tissue and human neuroblastoma (SH-SY5Y) cells. Treatment with hydrogen peroxide (H(2)O(2)) enhances the nuclear colocalization of NRF2 and NRP/B and induces heme oxygenase 1 (HO1). Treatment of NRP/B or NRF2 knockdowns with H(2)O(2) induced apoptosis. Co-expression of NRF2 with members of the Kelch protein family, NRP/B, MAYVEN, or MAYVEN-related protein 2 (MRP2), revealed that the NRF2-NRP/B complex is important for the transcriptional activity of ARE-driven genes HO1 and NAD(P)H:quinine oxidoreductase 1 (NQO1). NRP/B interaction with Nrf2 was mapped to NRF2 ECH homology 4 (Neh4)/Neh5 regions of NRF2. NRP/B mutations that resulted in low binding affinity to NRF2 were unable to activate NRF2-modulated transcriptional activity of the ARE-driven genes, HO1 and NQO1. Thus, the interaction of NRP/B with the Neh4/Neh5 domains of NRF2 is indispensable for activation of NRF2-mediated ARE-driven antioxidant and detoxifying genes that confer cellular defense against oxidative stress-induced damage.

The network organization of type IV collagen consisting of α3, α4, and α5 chains in the glomerular basement membrane (GBM) is speculated to involve interactions of the triple helical and NC1 domain of individual α-chains, but in vivo evidence is lacking. To specifically address the contribution of the NC1 domain in the GBM collagen network organization, we generated a mouse with specific loss of α3NC1 domain while keeping the triple helical α3 chain intact by connecting it to the human α5NC1 domain. The absence of α3NC1 domain leads to the complete loss of the α4 chain. The α3 collagenous domain is incapable of incorporating the α5 chain, resulting in the impaired organization of the α3α4α5 chain-containing network. Although the α5 chain can assemble with the α1, α2, and α6 chains, such assembly is incapable of functionally replacing the α3α4α5 protomer. This novel approach to explore the assembly type IV collagen in vivo offers novel insights in the specific role of the NC1 domain in the assembly and function of GBM during health and disease.

The PDZ1 domain of the four PDZ domain-containing protein PDZK1 has been reported to bind the C terminus of the HDL receptor scavenger receptor class B, type I (SR-BI), and to control hepatic SR-BI expression and function. We generated wild-type (WT) and mutant murine PDZ1 domains, the mutants bearing single amino acid substitutions in their carboxylate binding loop (Lys(14)-Xaa(4)-Asn(19)-Tyr-Gly-Phe-Phe-Leu(24)), and measured their binding affinity for a 7-residue peptide corresponding to the C terminus of SR-BI ((503)VLQEAKL(509)). The Y20A and G21Y substitutions abrogated all binding activity. Surprisingly, binding affinities (K(d)) of the K14A and F22A mutants were 3.2 and 4.0 μM, respectively, similar to 2.6 μM measured for the WT PDZ1. To understand these findings, we determined the high resolution structure of WT PDZ1 bound to a 5-residue sequence from the C-terminal SR-BI ((505)QEAKL(509)) using x-ray crystallography. In addition, we incorporated the K14A and Y20A substitutions into full-length PDZK1 liver-specific transgenes and expressed them in WT and PDZK1 knock-out mice. In WT mice, the transgenes did not alter endogenous hepatic SR-BI protein expression (intracellular distribution or amount) or lipoprotein metabolism (total plasma cholesterol, lipoprotein size distribution). In PDZK1 knock-out mice, as expected, the K14A mutant behaved like wild-type PDZK1 and completely corrected their hepatic SR-BI and plasma lipoprotein abnormalities. Unexpectedly, the 10-20-fold overexpressed Y20A mutant also substantially, but not completely, corrected these abnormalities. The results suggest that there may be an additional site(s) within PDZK1 that bind(s) SR-BI and mediate(s) productive SR-BI-PDZK1 interaction previously attributed exclusively to the canonical binding of the C-terminal SR-BI to PDZ1.

The transcription regulatory protein PAX3 binds to cognate DNA sequences through two DNA-binding domains, a paired domain and a homeodomain, and has important functions during neurogenesis and myogenesis. In humans, mutations in the PAX3 gene cause Waardenburg syndrome, whereas a chromosomal translocation that generates a PAX3-FOXO1 fusion gene is associated with the development of alveolar rhabdomyosarcoma. We have determined the crystal structure of the human PAX3 homeodomain in complex with a palindromic DNA containing two inverted TAATC sequences at 1.95 A resolution. Two homeodomains bind to DNA as a symmetric dimer, inducing a 3 degrees bend in the DNA helix. The N-terminal arm of the homeodomain inserts into the minor groove and makes direct and water-mediated interactions with bases and the sugar-phosphate backbone. The recognition helix fits directly into the major groove, and an elaborate network of structurally conserved water molecules mediates the majority of protein-DNA interactions. The structure elucidates the role of serine 50 in selection of the CG sequence immediately 3' of the TAAT motif by PAX class homeodomains and provides insights into the molecular mechanisms by which certain Waardenburg syndrome-associated missense mutations could destabilize the fold of the PAX3 homeodomain whereas others could affect its interaction with DNA.

A number of germ-line mutations in the BRCA1 gene confer susceptibility to breast and ovarian cancer. However, it remains difficult to determine whether many single amino-acid (missense) changes in the BRCA1 protein that are frequently detected in the clinical setting are pathologic or not. Here, we used a combination of functional, crystallographic, biophysical, molecular and evolutionary techniques, and classical genetic segregation analysis to demonstrate that the BRCA1 missense variant M1775K is pathogenic. Functional assays in yeast and mammalian cells showed that the BRCA1 BRCT domains carrying the amino-acid change M1775K displayed markedly reduced transcriptional activity, indicating that this variant represents a deleterious mutation. Importantly, the M1775K mutation disrupted the phosphopeptide-binding pocket of the BRCA1 BRCT domains, thereby inhibiting the BRCA1 interaction with the proteins BRIP1 and CtIP, which are involved in DNA damage-induced checkpoint control. These results indicate that the integrity of the BRCT phosphopeptide-binding pocket is critical for the tumor suppression function of BRCA1. Moreover, this study demonstrates that multiple lines of evidence obtained from a combination of functional, structural, molecular and evolutionary techniques, and classical genetic segregation analysis are required to confirm the pathogenicity of rare variants of disease-susceptibility genes and obtain important insights into the underlying pathogenetic mechanisms.

Tumstatin is an angiogenesis inhibitor that binds to alphavbeta3 integrin and suppresses tumor growth. Previous deletion mutagenesis studies identified a 25-aa fragment of tumstatin (tumstatin peptide) with in vitro antiangiogenic activity. Here, we demonstrate that systemic administration of this tumstatin peptide inhibits tumor growth and angiogenesis. Site-directed mutagenesis identified amino acids L, V, and D as essential for the antiangiogenic activity of tumstatin. The tumstatin peptide binds to alphavbeta3 integrin on proliferating endothelial cells and localizes to select tumor endothelium in vivo. Using 3D molecular modeling, we identify a putative interaction interface for tumstatin peptide on alphavbeta3 integrin. The antitumor activity of the tumstatin peptide, in combination with bevacizumab (anti-VEGF antibody), displays significant improvement in efficacy against human renal cell carcinoma xenografts when compared with the single-agent use. Collectively, our results demonstrate that tumstatin peptide binds specifically to the tumor endothelium, and its antiangiogenic action is mediated by alphavbeta3 integrin, and, in combination with an anti-VEGF antibody it exhibits enhanced tumor suppression of renal cell carcinoma.

The interaction of the breast tumor suppressor BRCA1 with the protein BARD1 results in the formation of a heterodimeric complex that has ubiquitin ligase activity and plays central roles in cell cycle checkpoint control and DNA repair. Both BRCA1 and BARD1 possess a pair of tandem BRCT domains that interact in a phosphorylation-dependent manner with target proteins. We determined the crystal structure of the human BARD1 BRCT repeats (residues 568-777) at 1.9 A resolution. The composition and structure of the BARD1 phosphoserine-binding pocket P1 are strikingly similar to those of the BRCA1 and MDC1 BRCT domains, suggesting a similar mode of interaction with the phosphate group of the ligand. By contrast, the BARD1 BRCT selectivity pocket P2 exhibits distinct structural features, including two prominent histidine residues, His685 and His686, which may be important for ligand binding. The protonation state of these histidines has a marked effect on the calculated electrostatic potential in the vicinity of P2, raising the possibility that ligand recognition may be regulated by changes in pH. Importantly, the BARD1 BRCT structure provides insights into the mechanisms by which the cancer-associated missense mutations C645R, V695L, and S761N may adversely affect the structure and function of BARD1.

Cyclin K and the closely related cyclins T1, T2a, and T2b interact with cyclin-dependent kinase 9 (CDK9) forming multiple nuclear complexes, referred to collectively as positive transcription elongation factor b (P-TEFb). Through phosphorylation of the C-terminal domain of the RNA polymerase II largest subunit, distinct P-TEFb species regulate the transcriptional elongation of specific genes that play central roles in human physiology and disease development, including cardiac hypertrophy and human immunodeficiency virus-1 pathogenesis. We have determined the crystal structure of human cyclin K (residues 11-267) at 1.5 A resolution, which represents the first atomic structure of a P-TEFb subunit. The cyclin K fold comprises two typical cyclin boxes with two short helices preceding the N-terminal box. A prominent feature of cyclin K is an additional helix (H4a) in the first cyclin box that obstructs the binding pocket for the cell-cycle inhibitor p27(Kip1). Modeling of CDK9 bound to cyclin K provides insights into the structural determinants underlying the formation and regulation of this complex. A homology model of human cyclin T1 generated using the cyclin K structure as a template reveals that the two proteins have similar structures, as expected from their high level of sequence identity. Nevertheless, their CDK9-interacting surfaces display significant structural differences, which could potentially be exploited for the design of cyclin-targeted inhibitors of the CDK9-cyclin K and CDK9-cyclin T1 complexes.