Introduction

Fumarate hydratase (FH) and succinate dehydrogenase are two integral enzyme components of the Krebs cycle, and besides their essential role in the TCA cycle, can act as tumour suppressors (King et al. 2006). The FH gene codes for fumarate hydratase (or fumarase; EC 4.2.1.2), which catalyzes the stereospecific, reversible hydratation of fumarate to L-malate. The FH gene localized at 1q42.1 codes for differentially processed, but sequence-wise identical cytosolic and mitochondrial forms. Whereas the mitochondrial enzyme is part of the TCA cycle, the cytosolic form is thought to utilize fumarate derived from different sources. Deficiency in FH activity causes an impaired energy production by interrupting the flow of metabolites through the Krebs cycle. Accumulation of fumarate is thought to competitively inhibit 2-oxo-glutarate dependent dioxygenases that regulate hypoxia inducible factor (HIF), thus activating oncogenic hypoxia pathways (Ratcliffe, 2007).

Due to their essential role in energy production, enzyme deficiencies result in early onset of severe encephalopathy (Kerrigan et al. 2000). Accordingly, autosomal recessive fumarate hydratase deficiency (FHD) caused by mutations in the FH gene results in fumaric aciduria, and common clinical features observed are hypotonia, failure to thrive, severe psychomotor retardation, seizures, facial dysmorphism and brain malformations. Interestingly, whereas homozygous FH mutations predispose to fumaric aciduria, several heterozygous FH mutations are known to be involved in the autosomal dominant syndrome of multiple cutaneous and uterine leiomyomata (MCUL1) (Tomlinson et al. 2002). Affected individuals develop benign smooth muscle tumours of the skin, and females develop fibroids of the uterus. When co-existing with an aggressive form of renal cell carcinoma (papillary renal type II cancer or renal collecting duct cancer) it is also known as hereditary leiomyomatosis and renal cancer (HLRCC) syndrome. In MCUL1/HLRCC germline mutations in FH are detected in the majority of the screened cases. To date, 107 variants have been described, of which 93 are thought to be pathogenic (Bayley et al. 2008). The most common types are missense mutations (57%), followed by frameshift and nonsense mutations (27%), as well as diverse deletions, insertions and duplications.

Here we present the crystal structure of human fumarase at 1.95 Å resolution and summarize structure-activity correlation between observed mutations and clinical phenotypes.

Materials and methods

Expression, purification & crystallization

DNA fragment encoding the fumerase domain of human FH (aa 44-510; GenBank entry 19743875) was subcloned into pNIC28-Bsa4 vector incorporating an N-terminal His6-tag. The plasmid was transformed into BL21(DE3)-pRARE, cultured in Terrific Broth at 37°C, and induced with 0.1 mM IPTG overnight at 18°C. Cells were homogenized in lysis buffer (50 mM K-phosphate pH 7.5, 500 mM NaCl, 1 mM TCEP), centrifuged to remove cell debris, and the supernatant was purified by Nickel affinity (HisTrap Crude FF) and size exclusion (HiLoad 16/60 Superdex S200) chromatography. Purified protein was concentrated to 12.6 mg/ml and stored in 10 mM HEPES pH 7.5, 150 mM NaCl, 5% (w/v) glycerol and 0.5 mM TCEP at -80°C. Crystals were grown by vapour diffusion at 20°C in sitting drops mixing 150 nl protein and 150 nl reservoir solution containing 20% (w/v) PEG 3350, 0.2 M sodium acetate, 10% (w/v) ethylene glycol and 100 mM Bis-Tris propane pH 7.5. Crystals were cryo-protected in mother liquor containing 25% (w/v) glycerol and flash-frozen in liquid nitrogen.

Data collection & structure determination

Diffraction data to maximum resolution of 1.95 Å were collected on beamline X10A at the Swiss Light Source, and processed using the CCP4 Program suite (CCP4, 1994). FH crystallized in the trigonal space group P3221 (a = 180.5 Å, b = 180.5 Å, c = 114.6 Å, α = 90o, β = 90o, γ = 120o) with four molecules in the asymmetric unit. The structure of FH was solved by molecular replacement with PHASER (McCoy et al. 2005), using the yeast fumerase structure as search model (PDB code 1YFM). Initial automated model building was performed with ARP/wARP (Perrakis et al. 1999). This is followed by cycles of iterative manual model building using COOT (Emsley & Cowtan 2004) and restrained refinement using REFMAC5 with TLS parameters (Murshudov et al. 1997). The final structure was deposited in the Protein Data Bank (www.rcsb.org) under accession code 3E04 (Table 1).

Table 1 X-ray data collection and refinement statistics
Full size table

Results and discussion

Fumarases are divided into two distinct groups. Class I fumarases are iron-dependent iron-sulfur cluster containing, dimeric enzymes, whereas the class II enzymes, including human and other eukaryotic fumarases, are homotetrameric enzymes with a molecular mass of about 200 kDa. Class II fumarases are evolutionarily highly conserved enzymes, e.g. the pairwise identity between E. coli and human fumarase is about 60%. Every monomer exhibits a typical tridomain structure, with a central domain involved in subunit interaction, thus forming a typical bundle comprised of 20 α-helices (Fig. 1A). Previous crystallographic analyses have revealed two distinct sites (A and B) in E. coli fumarase that can bind carboxylic acids. Site A is formed from three different monomer chains and likely to be the catalytic site, whereas site B is thought to allosterically regulate activity (Rose and Weaver 2004).

Fig. 1
figure 1

Ribbon/surface diagram of human fumarate hydratase illustrating the tetrameric assembly of class II fumarases. Molecular surface representation is used to convey the overall shape of each monomer as well as the tetrameric assembly. Each monomer has been coloured distinctively, to facilitate visualization. Two monomers are represented using semi-transparent surfaces, to highlight the fold (represented as ribbons). One of the active sites is highlighted in red, showing contribution of three distinct subunits. The figures were generated using the program ICM (www.molsoft.com)

Full size image

A previous study correlated 27 distinct missense mutations to the E. coli fumarase structure (Alam 2005b), since then the list of mutations has doubled. To this end, 55 missense mutations in the human fumarase gene are now described. Here we correlate this updated list of mutations to fumarase deficiency, MCUL1 and HLRCC syndrome (Table 2) by using the human fumarase structure. Although not all of these novel mutations have been biochemically characterized, previous results suggest that FH activity is related to HLRCC (Alam 2005a), although other environmental or genetic factors likely play a role in the etiology of the disease. The clustering of mutational “hotspots” suggests enzyme activity relationships to phenotypic appearances. Figure 2 illustrates the clustering of FH mutations observed in FHD, MCUL1 and HLRCC. The large majority of mutations are located at evolutionarily highly conserved positions (Table 2) indicating that these mutations likely affect stability and/or activity of the enzyme. Two major clusters of mutations are observed; one is likely to affect structural integrity of the enzyme by interrupting inter or intrasubunit interactions (indicated in yellow in Fig. 2), whereas the other mutations are located around the active site and likely directly affect activity.

Table 2 Mutations observed in the human fumarase gene and association to disease. Abbreviations: CL: cutaneous leiomyoma; FHD: fumarate hydratase deficiency; HLRCC: hereditary leiomyomatosis and renal cell cancer; LCT: Leydig cell tumors; MCUL: multiple cutaneous and uterine leiomyomata; OMC: ovarian mucinous cystadenoma; RCC: renal cell carcinoma; STS: soft tissue sarcoma; UL: uterine leiomyomas; ULMS: uterine leiomyosarcoma
Full size table
Fig. 2
figure 2

Clustering of human fumarase missense mutations observed in FHD, MCUL1 and HLRCC. The active site is highlighted in cyan. Positions of amino acid mutations are indicated as small spheres and numbered according to Table 2. The positions around the active site are indicated in red, mutations affecting inter- or intrasubunit interactions are indicated in dark yellow. For clarity, one monomeric subunit is omitted

Full size image