Heat shock transcription factor σ32 co-opts the signal recognition particle to regulate protein homeostasis in E. coli

Abstract

All cells must adapt to rapidly changing conditions. The heat shock response (HSR) is an intracellular signaling pathway that maintains proteostasis (protein folding homeostasis), a process critical for survival in all organisms exposed to heat stress or other conditions that alter the folding of the proteome. Yet despite decades of study, the circuitry described for responding to altered protein status in the best-studied bacterium, E. coli, does not faithfully recapitulate the range of cellular responses in response to this stress. Here, we report the discovery of the missing link. Surprisingly, we found that σ(32), the central transcription factor driving the HSR, must be localized to the membrane rather than dispersed in the cytoplasm as previously assumed. Genetic analyses indicate that σ(32) localization results from a protein targeting reaction facilitated by the signal recognition particle (SRP) and its receptor (SR), which together comprise a conserved protein targeting machine and mediate the cotranslational targeting of inner membrane proteins to the membrane. SRP interacts with σ(32) directly and transports it to the inner membrane. Our results show that σ(32) must be membrane-associated to be properly regulated in response to the protein folding status in the cell, explaining how the HSR integrates information from both the cytoplasm and bacterial cell membrane.

Figure 1. Homeostatic control circuits of σ³².

(A) Current and (B) revised model for activity and degradation control of σ³². The revised model incorporates SRP-mediated trafficking of σ³² to the membrane. Interactions validated in vitro are shown as solid lines; those inferred from in vivo data are shown as dashed lines. Newly identified interactions are shown in red.

Figure 2. σ³² binds to Ffh.

(A) Schematic representation of E. coli SRP (Ffh+4.5S RNA), indicating experimentally confirmed functions associated with each domain. (B) σ³² co-immunoprecipitates with Ffh and FtsY in vivo, but σ⁷⁰ does not. Immunoprecipitations of Ffh or FtsY were carried out on lysates of Δσ³² and ΔftsH cells grown to exponential phase. Immunocomplexes were isolated, analyzed by SDS-PAGE, and immunoblotted with anti-σ³² and anti-σ⁷⁰ antibodies. Proteins from approximately 15-fold more cells were loaded onto the gel for the immunoprecipitated samples against σ³² and σ⁷⁰ as compared with the lysate samples. (C) Protein–protein interaction analysis indicates that σ³² binds to Ffh, but not FtsY. Purified FtsY and Ffh were run on a 10% SDS-PAGE gel, transferred to nitrocellulose, re-natured, and incubated with purified WTσ³². The Coomassie-stained gel (left) and the nitrocellulose blot probed with polyclonal anti-σ³² antibodies (right) are shown. (D) σ³² binds to the M-domain of Ffh. Ffh, partially digested by endopeptidase V8, was resolved on a 10% SDS-PAGE gel, transferred to nitrocellulose, and incubated with σ³². The Coomassie-stained gel (left) and the nitrocellulose membrane, containing transferred Ffh fragments, probed against σ³² (right) are shown.

Figure 3. In vivo cross-linking between σ³² and Ffh.

(A) Detection of a cross-linked product following UV irradiation in whole cells. Cells of CAG48238/pEVOL-pBpF/p6XH-rpoHT52amber (lanes 1, 2, 5, and 6) and CAG48238/pEVOL-pBpF/p6XH-rpoH (lanes 3, 4, 7, and 8) were grown at 30°C in L-medium supplemented with 0.02% arabinose, induced with 1 mM IPTG for 1 h, and UV-irradiated for 0 or 10 min as indicated. Total cellular proteins were acid-precipitated and analyzed by SDS-PAGE and immunoblotting with anti-Ffh and anti-σ³² antibodies. (B) Immunoprecipitation with anti-Ffh reveals a unique cross-linked product that interacts with anti-σ³². Supernatants of sonically disrupted UV-irradiated cells were subjected to immunoprecipitation with anti-Ffh antibodies. Immunocomplexes were solubilized in SDS sample buffer, analyzed by SDS-PAGE, and immunoblotted with anti-Ffh and anti-σ³² antibodies. Proteins from approximately 4.4-fold more cells were loaded onto the gel for the immunoprecipitated samples as compared with the whole cell samples. (C) Purification of 6×H-σ³² from UV-irradiated cells reveals a band that interacts with anti-Ffh. Supernatants of sonically disrupted UV-irradiated cells were subjected to TALON affinity chromatography, and bound proteins were eluted with 300 mM imidazole. Proteins in the eluate were acid-precipitated and analyzed by SDS-PAGE and immunoblotting with anti-Ffh antibodies. Proteins form approximately 20-fold more cells were loaded onto the gel for the TALON-affinity isolated samples as compared with the whole cell samples.

Figure 4. SRP (Ffh+4.5S RNA) preferentially interacts with WTσ³².

(A) A₂₈₀ elution profiles of WTσ³², I54Nσ³², Ffh, and SRP alone or in complex. WTσ³² or I54Nσ³² was incubated with a 10-fold molar excess of purified SRP on ice for 10 min, and complexes were analyzed by gel filtration on a Superdex 200 PC3.2/30 column. Protein elution was monitored by A₂₈₀. Gel filtration of purified WTσ³², I54Nσ³², and SRP alone was carried out to determine the migration of each individual protein on the column. (B) Eluted fractions were separated on SDS-PAGE and probed with polyclonal antibodies against Ffh and σ³²; Western blots of σ³² are shown. Experiments were performed at least four times, and a representative experiment is shown.

Figure 5. σ³² is partially membrane associated.

The extent of association of σ³² and the β′ subunit of RNA polymerase with the membrane fraction was determined by quantitative immunoblotting of the soluble and nonsoluble fractions. Membrane association of σ³² and β′ was assessed in several relevant strain backgrounds. In addition to endogenous σ³², all strains contained a plasmid-encoded variant of σ³² lacking its 21 C-terminal amino acids (σ³²Δ21aa). Ectopically expressed σ³²Δ21aa or I54Nσ³²Δ21aa were present at levels comparable to native σ³² and were distinguished from endogenous σ³² on a 12% SDS-PAGE gel. All fractionation experiments were performed ≥8 times, and % fractionation was calculated from experiments where probed cytoplasmic (RuvB) and membrane (RseA) proteins separated properly.

Figure 6. The SecY translocon plays a role in chaperone-mediated activity control of σ³².

(A) Schematic of SecY topology in the IM by highlighting in yellow the locations/allele names of the mutated residues used in this study . The region that interacts with FtsY (Domain C5) is boxed in green. (B) Mutations in secY show higher σ³² activity and affect chaperone-mediated activity control of σ³². The activity of σ³² was measured in WT and secY mutant cells growing at 30°C in LB medium (column 1) or in the same cells following induction of DnaK/J (column 2) or GroEL/S (column 3). Activity is calculated as the differential rate of β-galactosidase synthesis from a chromosomal P_tpG-lacZ reporter in each cell type relative to that of WT cells.