Structural basis of disease-causing mutations in hepatocyte nuclear factor 1beta

Abstract

HNF1beta is an atypical POU transcription factor that participates in a hierarchical network of transcription factors controlling the development and proper function of vital organs such as liver, pancreas, and kidney. Many inheritable mutations on HNF1beta are the monogenic causes of diabetes and several kidney diseases. To elucidate the molecular mechanism of its function and the structural basis of mutations, we have determined the crystal structure of human HNF1beta DNA binding domain in complex with a high-affinity promoter. Disease-causing mutations have been mapped to our structure, and their predicted effects have been tested by a set of biochemical/ functional studies. These findings together with earlier findings with a homologous protein HNF1alpha, help us to understand the structural basis of promoter recognition by these atypical POU transcription factors and the site-specific functional disruption by disease-causing mutations.

Figure 1

(A) Disease-causing mutations found in HNF1β. Unlike nonsense and frameshift mutations, point mutations are clustered within the DNA binding domain. Point mutations found in the DNA binding domain are indicated in black, and the remaining regions are shown in gray. (B) Sequence alignment of human HNF1β and HNF1α DNA binding domains. Pα1 through Pα5 make up POU_S, while Hα1 through Hα3 make up POU_H. Invariant and variant residues are colored blue and brown, respectively. Point mutations are shown in red above (HNF1β) and below (HNF1α) the sequences. The numbers in front of each sequence indicate the beginning number of the amino acids, and the core domains and the nuclear localization signal (NLS) are highlighted by green and pink shadow boxes, respectively.

Figure 2

(A) Comparison of HNF1β-DNA (purple) and HNF1αDNA (gold) (60) complexes. The two structures are superimposed through their POU_H domains of the first molecule (left). After the operation, the POU_S domains of the first molecule also superimpose very well, while the second molecules are 9.8° apart from each other. (B) Superposition of HNF1β homeodomain/DNA complex crystal structure (blue) with the NMR ensemble structures of HNF1β homeodomain without DNA (pink, PDB code 2DA6). The N-terminal arm containing two key arginine residues (R235 and R232) that make contacts with DNA through the minor groove can be seen below the helices on the left.

Figure 3

Schematic summary of HNF1β-DNA interactions and disease-causing mutations. Residues from molecules 1 and 2 are shown in pink and green, respectively. POU_H and POU_S domain residues are shown as rectangles and ovals, respectively; filled shapes indicate disease-causing mutations. DNA bases within the consensus recognition motif are colored blue.

Figure 4

Mapping of disease-causing mutations on the HNF1β structure (upside-down view from Figure 2). Ribbon representation of HNF1β is shown in pink (POU_S) or red (POU_H), and DNA is in orange. Side chains of residues affected by diabetes-associated missense mutations are displayed in green (predominantly affecting DNA binding), purple (interdomain interactions), or light blue (protein stability).

Figure 5

Close-up views of each mutation and surrounding residues. The mutated residues are labeled in light blue, while the surrounding residues are labeled in white. DNA bases that make sequence-specific interactions are also labeled in yellow. Ionic interactions and hydrogen bonds are indicated by dotted lines. POU_S is shown in pink, while POU_H is shown in red.

Figure 6

DNA binding assay. The amount of loaded protein sample is shown in the top box (Coomassie blue staining). In the main gel, the lower bands correspond to free DNA while the upper bands represent the shifted HNF1β/DNA complex. Wild type (WT) is shown in the second lane as a control, and R165H is missing from the experiment due to an extremely low yield during purification.

Figure 7

Lifetime of the HNF1β mutants compared with wild type. HeLa cells were labeled with [³⁵S]methonine/cysteine for 30 min (0 h) and pulse–chased for various lengths of time (3−18 h) in the presence of excess nonradioactive methionine/cysteine. Samples were immunoprecipitated under normal stringency conditions with polyclonal anti-HNF1β antibody and subjected to SDS–PAGE followed by autoradiography.

Figure 8

Overall transcriptional activity by the mutants compared to wild type. CTL in the first lane refers to an empty vector, and all data have been normalized against firefly/Renilla luciferase activity. Protein expression levels were assessed by Western blot and there is a high degree of correlation between these in vivo expression levels and the findings from the protein lifetime assays (Figure 7). Actin, a housekeeping gene product, was used as a loading control.