Structural and functional adaptation of Haloferax volcanii TFEα/β

Abstract

The basal transcription factor TFE enhances transcription initiation by catalysing DNA strand-separation, a process that varies with temperature and ionic strength. Canonical TFE forms a heterodimeric complex whose integrity and function critically relies on a cubane iron-sulphur cluster residing in the TFEβ subunit. Halophilic archaea such as Haloferax volcanii have highly divergent putative TFEβ homologues with unknown properties. Here, we demonstrate that Haloferax TFEβ lacks the prototypical iron-sulphur cluster yet still forms a stable complex with TFEα. A second metal cluster contained in the zinc ribbon domain in TFEα is highly degenerate but retains low binding affinity for zinc, which contributes to protein folding and stability. The deletion of the tfeB gene in H. volcanii results in the aberrant expression of approximately one third of all genes, consistent with its function as a basal transcription initiation factor. Interestingly, tfeB deletion particularly affects foreign genes including a prophage region. Our results reveal the loss of metal centres in Hvo transcription factors, and confirm the dual function of TFE as basal factor and regulator of transcription.

Figure 1.

Overview over the domain organization of archaeal TFE subunits and related eukaryotic proteins. TFEα-related proteins are composed of extended winged-helix domains (eWH), a zinc ribbon domain (ZR) and coiled coil helices (CC) (3–5,55). TFEβ-related proteins are composed of winged-helix (WH) and iron-sulphur cluster domains (FeS). Sequence alignments of the two metal binding domains (ZR and FeS) reveal the loss of cysteine residues (highlighted in yellow) required for metal coordination in H. volcanii TFEα and TFEβ. Key to sequences: H. sapiens TFIIEα (NCBI accession: P29083), S. cerevisiae Tfa1 (EDN59878), S. solfataricus TFEα (AAK40605), M. jannaschii TFE (Q58187), H. salinarum TFEα (AAG19231), H. walsbyi TFEα (ERG94676), H. volcanii TFEα (ELK55987), H. sapiens C39 (EAX10240), S. cerevisiae C34 (P32910), S. solfataricus TFEβ (AAK41220), H. salinarum TFEβ (WP_010902769), H. walsbyi TFEβ (ERG96516), H. volcanii TFEβ (ADE03923). The secondary structure of the H. sapiens TFIIEα ZR domain is based on Okuda et al. (43).

Figure 2.

Hvo TFEα and TFEβ form a dimeric complex in the absence of an FeS cluster. (A and B) Size exclusion chromatography of TFEα_His/TFEβ (A) and TFEα/TFEβ ΔWH_His (B) on a superose 12 h 10/30 column. Elution peaks of standard proteins are indicated. Peak fractions (highlighted in gray) were subjected to SDS-PAGE and Coomassie-staining (insert) to show symmetric elution of both subunits. (C) nESI mass spectrum of TFEα/TFEβ_His. The different charge state series observed are labelled with red, blue and green filled circles. Charge state series derived from additional cleavage of TFEα N-Met are labelled with open circles. Highlighted in gray is a zoom over the region that includes charge state (CS) +11. All detected masses indicate the absence of a cubane 4Fe–4S cluster (352 Da).

Figure 3.

The degenerate zinc-ribbon domain of TFEα is capable of zinc binding. (A and B) nESI native mass spectrum of monomeric wild type TFEα_TEV (A) and TFEα_TEV C138S (B) after incubation with zinc. Charge state series of TFEα_TEV with and without zinc are labelled with green and blue filled circles, respectively. Highlighted in gray is a zoom over the region that includes charge state (CS) +8. (C) Collision induced unfolding experiment of TFEα_TEV with and without zinc addition. CS+8 was isolated using the quadrupole and activated in the trap collision cell using a CE ramp (6–36 V) prior to ion mobility separation. The ion mobility cell separated the protein into two populations corresponding to folded (f) and unfolded (u) states.

Figure 4.

The C138S mutation does not cause larger structural perturbation of TFEα_TEV. Overlay 2D ¹⁵N–¹H HSQC spectra of TFEα_TEV (blue) and C138S variant (red) without Zn²⁺. The black arrow indicates the chemical shift perturbation caused by the C138S mutation. All data were recorded at 600 MHz, under the same conditions, number of scans and increments, at pH 7.5, 7% D₂O, 25°C.

Figure 5.

The tfeB gene encoding TFEβ is non-essential in H. volcanii. (A) Southern blot of EcoRV digest confirming deletion of tfeB in H2644 and H2645. (B) Deletion of tfeB confers a slow-growth phenotype in H2644 and H2645, versus the parent strain H53. Data from the mean of four repeats (and standard error) is plotted on a log₂ scale, the generation time in exponential phase is shown in bold (exponential growth phase is indicated by straight-line fit between perpendicular marks, R² = 0.99 for all strains). (C) The growth defect conferred by deletion of tfeB in H2644 is complemented by in trans expression of tfeB (eV, empty vector).

Figure 6.

Deletion of tfeB leads to widespread misregulation of transcription. (A) log₂-fold change in relative transcript abundance (P_adj < 0.01) of genes coding for the basal translation machinery, ribosomal proteins (black), translation initiation factors (green) and elongation factors (red). (B) Analysis arCOG category distribution in significantly misregulated genes (P_adj < 0.01). Significant enrichment or underrepresentation is indicated (two-sided Fisher's exact test with Bonferroni correction for multiple testing) with * and *** denoting P_adj < 0.05 and 0.001, respectively. (C and D) Heatscatter plot of rare codon frequency (defined as codons with less than 10% synonymous codon usage genome-wide) (C) and AT content of genes (D) against the log₂-fold change (P_adj < 0.01) in relative transcript abundance. (E) Plot showing the estimated transcript levels in parental strain H26 (transcripts per million, TPM, geometric mean for the two replicates) calculated for the transcription units of Hvo using RSEM (31) against the log₂-fold change in relative transcript abundance for the first cistron within the respective transcription unit (P_adj < 0.01). (F) BRE/TATA motif consensus in misregulated genes. In order to account for variable spacing between the TSS and BRE/TATA motifs we conducted a motif search using the MEME software version 4.11.4 (0 or 1 occurrence per sequence, 4 to 16 bp width, searching given strand only) (33). A background model based on the composition of the combined H. volcanii DS2 genome was employed. BRE/TATA motifs were discovered in all sequences.