Identification of three regions essential for interaction between a sigma-like factor and core RNA polymerase

Abstract

The cyclic interactions that occur between the subunits of the yeast mitochondrial RNA polymerase can serve as a simple model for the more complex enzymes in prokaryotes and the eukaryotic nucleus. We have used two-hybrid and fusion protein constructs to analyze the requirements for interaction between the single subunit core polymerase (Rpo41p), and the sigma-like promoter specificity factor (Mtf1p). We were unable to define any protein truncations that retained the ability to interact, indicating that multiple regions encompassing the entire length of the proteins are involved in interactions. We found that 9 of 15 nonfunctional (petite) point mutations in Mtf1p isolated in a plasmid shuffle strategy had lost the ability to interact. Some of the noninteracting mutations are temperature-sensitive petite (ts petite); this phenotype correlates with a precipitous drop in mitochondrial transcript abundance when cells are shifted to the nonpermissive temperature. One temperature-sensitive mutant demonstrated a striking pH dependence for core binding in vitro, consistent with the physical properties of the amino acid substitution. The noninteracting mutations fall into three widely spaced clusters of amino acids. Two of the clusters are in regions with amino acid sequence similarity to conserved regions 2 and 3 of sigma factors and related proteins; these regions have been implicated in core binding by both prokaryotic and eukaryotic sigma-like factors. By modeling the location of the mutations using the partial structure of Escherichia coli sigma70, we find that two of the clusters are potentially juxtaposed in the three-dimensional structure. Our results demonstrate that interactions between sigma-like specificity factors and core RNA polymerases require multiple regions from both components of the holoenzymes.

Figure 1

Two-hybrid deletion constructs. (A) Mtf1p deletions were fused to the LexA DNA-binding region and tested for interaction with a full-length Rpo41p two-hybrid construct . (B) Rpo41p deletions were fused to the Vp16 activator region and tested for interaction with a full-length LexA:Mtf1p construct. Strains, vectors, and protocols are described in Materials and Methods. Numbers refer to amino acid positions retained in the constructs. Shaded boxes in the full-length maps of MTF1 and RPO41 refer to regions conserved between σ factors and Mtf1p (Jang and Jaehning 1991) or between Rpo41p and T7 RNA polymerase (Masters et al. 1987; Jaehning 1993).

Figure 2

Confirmation of the expression of the two-hybrid constructs. Two-hybrid fusion proteins were detected in extracts of whole-cell yeast grown at 30°C. Rpo41p fusion proteins were visualized using an anti-Rpo41p antibody, whereas Mtf1p fusion proteins were detected with an anti-LexA antibody (Materials and Methods). Equal amounts of yeast cell extracts were analyzed for each sample. (A) Mtf1p carboxy-terminal deletion constructs. (B) Rpo41p deletion constructs. (C) Mtf1p point mutations that are defective for interaction with Rpo41p in the two-hybrid system.

Figure 3

Plasmid shuffle isolation of PCR-generated mtf1 mutants. Plasmids, strains, and protocols are described in Materials and Methods. A haploid strain (yJH71) containing a single functional copy of MTF1 on a plasmid was transformed with plasmids bearing mutant mtf1 genes on a LEU2 selectable plasmid. The transformants were plated onto glucose medium containing 5-FOA to select for the loss of the URA3 plasmid bearing the wild-type copy of MTF1. The mutant alleles were then tested for Mtf1p function by plating the cells on a nonfermentable carbon source (YPG) . Petite and ts petite mutants were identified.

Figure 4

Ts mtf1 mutants are defective in all classes of mitochondrial gene transcription. Wild-type (wt) and ts petite strains bearing mtf1 mutations L53H and I154T were grown in minimal glucose medium at 30°C and shifted to 37°C. Total RNA was harvested from cells immediately before and at the indicated time intervals after the shift to the nonpermissive temperature. (A) Blots of the isolated RNAs were hybridized with labeled 14S rRNA, COB, or tRNA^THR(ACN) oligonucleotide probes to analyze the abundance of mitochondrial transcripts. (B) Hybridization to the blots was quantitated by PhosphorImager analysis to directly compare levels of transcripts. The blots were hybridized to an 18S rRNA oligonucleotide probe to normalize for loading and transfer. Strains, plasmids, and protocols are described in Materials and Methods.

Figure 5

Identification of point mutations in Mtf1p that fail to interact with Rpo41p. β-Galactosidase activity is shown for the LexA:Mtf1p construct alone and the LexA:Mtf1p construct with the VP16:Rpo41p construct. Interaction between the Mtf1p mutants and Rpo41p was determined for cells grown at 30°C (shaded bar). Activity for ts mutants Y42C and K157E is also shown for cells grown at 37°C (hatched bar) or 23°C (crosshatched bar), respectively, to demonstrate the temperature-sensitive nature of the interaction. β-Galactosidase activity is expressed in Miller units (Miller 1972).

Figure 6

Biochemical confirmation of the noninteracting Mtf1p mutations. GST fusion constructs of wild-type and the indicated mutant Mtf1ps were isolated by glutathione agarose chromatography. Whole-cell yeast extracts containing Rpo41p were loaded onto the GST–Mtf1p columns. The columns were washed to eliminate nonspecific binding, then step-eluted to release Rpo41p bound to the column. Rpo41p in the input, wash, and elution column fractions was detected by Western blot analysis using an anti-Rpo41p antibody. Protocols are described in Materials and Methods. The percentage of two-hybrid interaction relative to wild-type Mtf1p (from Fig. 5) is shown in parentheses for each mutant.

Figure 7

High pH inhibits the interaction of ts petite mutant Y42C with Rpo41p. The GST–Mtf1p fusion construct bearing mutation Y42C was isolated and tested for interaction with Rpo41p as outlined in the legend to Fig. 6. The columns and the whole-cell yeast extract containing Rpo41p were equilibrated with buffers of either pH 8.3 (as shown in Fig. 6) or pH 7.3. Protocols for the detection of Rpo41p in the eluted fractions are described in Materials and Methods.

Figure 8

Position of Mtf1p mutations that affect interactions with Rpo41p. Noninteracting Mtf1p mutations cluster in three regions designated A, B, and C. Regions with amino acid sequence similarity to σ factors are shaded (Jang and Jaehning 1991).

Figure 9

Alignment of the cluster A mutations with known core interaction regions of σ and σ-like factors, and location of this region on the structure of σ⁷⁰. (A) Core-binding regions identified in σ⁷⁰ [amino acids 361–417 (Lesley and Burgess 1989)], σ³² [amino acids 35–91 (Joo et al. 1997)], σ^E [amino acids 44–100 (Schuler et al. 1995)], Rap30 [amino acids 105–160 (McCracken and Greenblatt 1991)], Mtf1p [amino acids 9–66 (this work)] and σ⁵⁴ [amino acids 175–193 (Tintut and Gralla 1995)] are shown. A 30-amino-acid peptide from σ⁷⁰ with affinity for core polymerase (Lesley and Burgess 1989) is shown by a broken line above the σ⁷⁰ sequence. The position of mutations that reduce interaction with core polymerase are shown in green for σ^E, σ³², and σ⁵⁴, and in red for Mtf1p. Colored shading is used to indicate similarity between the amino acid sequences. (B) The locations of the noninteracting mutations shown in A are highlighted on the structure of region 2 of σ⁷⁰ as determined by Malhotra et al. (1996) using InsightII (Biosym, San Diego, CA). The Mtf1p noninteracting cluster A mutations are shown in red. Positions required for interaction in σ^E, σ³², and σ⁵⁴ are shown in green. Note that the carboxyl terminus of region 2.4 (marked with an arrow) is brought close to the position of the noninteracting mutations in regions 2.1 and 2.2 in the three-dimensional structure.