A role for RNA metabolism in inducing the heat shock response

Abstract

Yeast HSF is constitutively trimeric and DNA bound. Heat shock is thought to activate HSF by inducing a conformational change. We have developed an assay in which we can follow a conformational change of HSF that correlates with activity and thus appears to be the active conformation. This conformational change requires two HSF trimers bound cooperatively to DNA. The conformational change can be induced in whole cell extracts, and is thus amenable to biochemical analysis. We have purified a factor that triggers the conformational change. The factor is sensitive to dialysis, insensitive to NEM, and is not extractable by phenol. It is small, and apparently not a peptide. Mass spectroscopy identifies a novel guanine nucleotide that tracks with activity on columns. This novel nucleotide, purchased from Sigma, induces the conformational change (although this does not prove the identity of the activating factor unambiguously, because Sigma's preparation is contaminated with other compounds). What is the source of this nucleotide in cells? Activity can be generated by treating extracts with ribonuclease; this implicates RNA degradation as a source of HSF-activating activity. The heat shock response is primarily responsible for monitoring the levels of protein chaperones; how can RNA degradation be involved? Synthetic lethal interactions link HSF activity to ribosome biogenesis, suggesting a possible model. Ribosomal proteins are produced in large quantities, and in excess of rRNA; unassembled r-proteins are rapidly degraded (t1/2 approximately 3 min). Unassembled r-proteins aggregate readily. It is likely that unassembled r-proteins represent a major target of chaperones in vivo, and for proteasome-dependent degradation. Interference with rRNA processing (e.g., by heat shock) requires hsp70s to handle the aggregation-prone r-proteins, and proteasome proteins to help degrade the unassembled r-proteins before they aggregate. A nucleotide signal could be generated from the degradation products of the rRNA itself.

FIG. 1

The effect of the M232V mutation. (A) Yeast carrying a HSE-lacZ reporter gene, the hsfJ-Δ disruption, and an episomal plasmid expressing either wild-type HSF or HSF^M232V were grown at 25°C, then assayed for β-galactosidase activity either directly or after 1 h at 37°C. (B) Protease-deficient yeast expressing either wild-type HSF or HSF^M232V were used to prepare whole cell extracts after growth at 25°C or at 25°C followed by a 30-min heat shock at 37°C. Extracts were then used for gel mobility shift DNA binding assay using a ³²P-labeled DNA probe (HSE4T) to which a single HSF trimer can bind. (C) The extracts from (B) were reanalyzed using a DNA probe (HSE6T) to which two trimers can bind simultaneously. The 25°C sample of wild-type was also assayed with DNA probe 4T, in order to identify the position of the DNA-bound trimer.

FIG. 2

Complex III correlates with HSF activity. (A) Cells carrying a HSE-lacZ reporter gene were grown at 25°C, then heat shocked at 37°C for 1 h, then returned to 25°C for recovery. Samples were taken at the indicated times for assay of (β-galactosidase activity. (B) Cells were grown as for (A), then used to prepare extracts for gel mobility shift DNA binding assay. The distribution of label in complexes I, II, and III was determined by phosphorimager, and the percentage of complex III calculated. (C) Cells were exposed to 6% ethanol or to 125 pM H202 for 30 min, then used to prepare extracts for gel mobility shift DNA binding assay. An extract from untreated cells (Ø) was run on the same gel for comparison.

FIG. 3

Heteromultiraer analysis of complex III. Extracts were prepared from heat-shocked cells carrying either truncated HSF (HSF^1–583) or full-length HSF (HSF^1–833), each with the M232V mutation. The extracts were used alone, or mixed, in a gel mobility shift DNA binding assay using Tris-glycine gel buffer. The positions of predicted mobilities for heteromultimers formed from six HSF monomers are indicated. The HSF^1–583 expression vector was prepared by inserting a 9-base oligonucleotide carrying an in-frame stop codon into a Stul restriction site. The stop codon was later discovered to be subject to translational readthrough [see (15)]; cells carrying this plasmid express a small amount of full-length readthrough product (detectable by Western blot, not shown), which accounts for the appearance of heteromultimers in the “1-583 only” lane.

FIG. 4

In vitro heat shock induces complex III. (A) Extracts from nonshocked cells were used in gel mobility shift DNA binding assays, in which the DNA binding reactions were performed at various temperatures. (B) A nonshocked extract was incubated in a 25°C DNA binding reaction for 30 min, then a 1000-fold excess of unlabeled HSE6T probe was added; incubation was continued at 25°C. At various times thereafter, samples were withdrawn and applied to a (running) gel for analysis. (C) Analogous to (B), except that the reaction was shifted to 37°C after addition of unlabeled competitor.

FIG. 5

HPLC fractionation of complex Ill-inducing activity. Partially purified complex Ill-inducing activity was loaded onto monoQ at 50 mM NaCl, washed with 100 mM NaCl, then eluted with a 100–500 mM NaCl gradient. Fractions were assayed for their ability to induce formation of complex III. Ø indicates the 25°C extract with no additions.

FIG. 6

Induction of complex ffl by ⁵G_2^3 and ribonuclease. (A) Commercially prepared ⁵G_2^3 was tested at 0.625, 1.45, and 2.5 mM. Untreated extract (Ø) and extract treated with the monoQ fraction from Fig. 5 were run in parallel. (B) ⁵G_2^3, was treated with RNAse T₁ then analyzed. (C) Extracts were treated with RNAse A or RNAse T₁ during the DNA binding reaction and assayed by gel mobility shift, with untreated extract (Ø) and extract treated with the monoQ fraction as controls