AccD1 And AccA1 from M. tuberculosis form A dodecameric MCC-type holo complex

Mycobacteria have an unusual redundancy of six putative carboxyltransferase genes that form high-molecular weight holo acyl coenzyme A carboxylase complexes with a complementary set of three biotin carboxylase genes. Most of these enzyme complexes use small fatty acid coenzyme A esters as substrate, to allow their extension by one methylene group via a carboxybiotin-mediated α-carboxylation reaction. Redundant occurrence of these complexes was assumed to be related to highly complex enzymatic requirements in lipid biosynthesis, as the mycobacterial thick cell wall comprises unusual very long chain fatty acids, including mycolic acid. We have solved two high-resolution crystal structures of the 350 kDa hexameric assemblies of two different acyl coenzyme A carboxylase hexameric assemblies, AccD5 and AccD6 [1; Anandhakrishnan et al., unpublished], and characterized these enzyme complexes functionally. In a second step we investigated the acyl coenzyme A carboxylase complex AccD1-AccA1 from Mycobacteria tuberculosis with hitherto unknown function. By using a metabolomics approach we found that AccD1-AccA1 is involved in branched amino acid catabolism, which was not investigated in mycobacteria before [Ehebauer et al, unpublished. Using an in vitro assay, we show that the enzyme complex uses methylcrotonyl coenzyme A as substrate]. We determined the overall architecture of the 700 kDa AccD1-AccA1 complex to be formed from three layers of a central AccD1 hexameric ring, flanked by two distal tiers composed of three AccA1 subunits each. Our electron microscopy data match the overall dimensions of a methylcrotonyl coenzyme A holo complex with known structure and thus support our functional findings. Our data suggest a unique functional role of the AccD1-AccA1 complex within the Mycobacterium tuberculosis acyl coenzyme A carboxylase interactome. Ultimately, it is our goal to solve this and related structures of ACCase holo complexes by high-resolution crystallography as well. The abstract is dedicated to Louis Delbaere with whom I shared time during my PhD at the University of Basel, Switzerland.

Serial femtosecond X-ray diffraction of in vivo crystals in intact yeast cells

Peroxisomes are membrane-enclosed organelles in eukaryotic cells with important roles in lipid metabolism and the scavenging of reactive oxygen species. Peroxisomes are capable of carrying an unusually high load of proteins, which under appropriate nutrient conditions results in the in situ crystallization of peroxisomal proteins in several yeast species and vertebrate hepatocytes [1,2]. In the methylotrophic yeast H.polymorpha, the predominant peroxisomal protein alcohol/methanol oxidase (AO) oligomerizes into octameric assemblies with a molecular mass of 600 kDa that spontaneously form 200-500 nm crystallites within peroxisomes [1]. We exposed H.polymorpha cell suspensions containing peroxisome-confined AO crystallites to femtosecond X-ray pulses at the Coherent X-ray Imaging (CXI) experimental endstation at the Linac Coherent Light Source. Peak detection routines mining the resulting scattering profiles identified >5000 Bragg-sampled diffraction patterns, providing the proof of concept that background scattering from the cells does not deteriorate the signal-to-noise ratio to an extent precluding observation of diffraction from individual AO crystallites. Summation patterns assembled from the individual frames match low-resolution powder diffraction patterns from concentrated suspensions of purified peroxisomes collected at the P14 beamline at the PETRAIII synchrotron, confirming that the observed diffraction mainly results from Bragg scattering of peroxisomal crystallites. To the best of our knowledge our data are the first to report room temperature X-ray diffraction from functional protein crystals in their native cellular environment. Currently the maximum resolution achieved in the diffraction patterns is limited to 20-15 Å. Future work will need to address improved sample preparation protocols in order to assess whether diffraction to a resolution sufficient to permit structure solution can be obtained. Protein crystal formation in vivo has been observed under physiological or pathological conditions in a number of other systems [3]. We hope that our results will help to establish serial femtosecond X-ray diffraction (SFX) as a method for structural characterization of cellular structures with crystalline content and provide a proof of concept for using in situ crystallization of proteins as a means to generate nanocrystalline samples for SFX.

DNA binding and dimerization specificities of Basic Zipper transcription factors

The ability of basic zipper transcription factors to form homo- or heterodimers provides a paradigm for combinatorial control of eukaryotic gene expression. In a first study, we clarified the specificity of the MIcrophthalmia-associated Transcription Factor [1]. To achieve this, we solved three crystal structures: two structures of MITF in complex with DNA duplexes encompassing two different target motifs (E-box and M-box) and one APO-structure. We then analyzed interactions between these DNA elements and several MITF mutants with documented mice phenotypes, using complementary techniques. The comparison of these experiments together with available biological data reveals the particular mechanism of DNA recognition by MITF. Moreover we demonstrated how a shift in the leucine zipper register limits the choice of the homotypic dimerization partner among the other b-HLH-Zip transcription factors. In a second study, we wondered how facultative dimerization results in alternative DNA-binding repertoires on distinct regulatory elements [2]. In this respect, the hematopoietic b-Zip transcription factor MafB, is a good model, since it has the ability to form homo- and heterodimers with a few other transcription factors. We first determined two high-resolution structures of MafB as a homodimer and as a heterodimer with c-Fos bound to variants of the Maf-recognition element (MARE). The two structures revealed several unexpected and specific coiled coil interactions. Based on these findings, we have engineered two MafB mutants with opposite dimerization preferences. One of them indeed showed a strong preference for MafB/c-Fos heterodimerization. In addition this variant enabled a selection of heterodimer- favoring over homodimer-specific MARE variants, demonstrating that protein/protein and protein/DNA interactions are interconnected. Our data provide a new concept for transcription factor design to selectively activate dimer-specific pathways and binding repertoires.