Disruption of nucleoid expanded conformation by toxic aberrant proteins synthesized in Escherichia coli

Aminoglycoside antibiotics interfere with selection of cognate tRNAs during translation, resulting in the production of aberrant proteins that are the ultimate cause of the antibiotic bactericidal effect. To determine if these aberrant proteins are recognized as substrates by the cell’s protein quality control machinery, we studied whether the heat shock (HS) response was activated following exposure of Escherichia coli to the aminoglycoside kanamycin (Kan). Levels of the HS transcription factor σ32 increased about 10-fold after exposure to Kan, indicating that at least some aberrant proteins were recognized as substrates by the molecular chaperone DnaK. To investigate whether toxic aberrant proteins therefore might escape detection by the QC machinery, we studied model aberrant proteins that had a bactericidal effect when expressed in E. coli from cloned genes. As occurred following exposure to Kan, levels of σ32 were permanently elevated following expression of an acutely toxic 48-residue protein (ARF48), indicating that toxic activity and recognition by the QC machinery are not mutually exclusive properties of aberrant proteins, and that the HS response was blocked in these cells at some step downstream of σ32 stabilization. This block could result from halting of protein synthesis or from radial condensation of nucleoids, both of which occurred rapidly following ARF48 induction. Nucleoids were similarly condensed following expression of toxic aberrant secretory proteins, suggesting that transertion of inner membrane proteins, a process that expands nucleoids into an open conformation that promotes growth and gene expression, was disrupted in these cells. The 48-residue ARF48 protein would be well-suited for structural studies to further investigate the toxic mechanism of aberrant proteins.

. Since each affected ribosome likely synthesizes a unique 58 set of aberrant proteins, it has not been possible to directly study the sequence and structural 59 determinants of aberrant protein toxic potential. Earlier work suggested that the diverse population of 60 aberrant proteins may act in concert to disrupt cell membrane integrity, based on the finding that 61 membrane permeability to ions increased following exposure to aminoglycoside antibiotics [8]. This led 62 to the suggestion that aberrant proteins might be incorporated into the membrane, forming ion-63 conducting channels [7]. However, these putative channels were not isolated or directly characterized.

64
For aberrant proteins to have toxic effects they must escape detection and destruction by the 8 159 in 100 µl buffer 1 containing 40 mM glycine. Bound proteins were then eluted in 50 µl of buffer 1 160 containing 150 mM imidazole.

191
Early log phase E. coli cells were exposed to the aminoglycoside kanamycin (Kan), and to the 192 non-aminoglycoside antibiotics tetracycline (Tc) and chloramphenicol (Cm). Cell division was halted soon 193 after exposure to all three antibiotics over a range of concentrations, as shown by leveling of the growth 194 curves (Fig 1, panels A-C). σ32 levels in cells after 20, 60 or 120 min of antibiotic exposure were then 195 determined by Western blotting. σ32 accumulated to high relative concentration in cells exposed to 196 Kan, even at a low dose (3 µg/ml) that slowed but did not completely halt cell division, whereas σ32 was 197 not detected in cells exposed to Cm or to any but the highest concentration (143 µg/ml) of Tc (Fig 1D;198 only the 60 min time point is shown). Quantification of the blotting data (Fig S1) showed that σ32 levels 199 in Kan-treated cells increased by up to 12-fold in comparison to untreated control cells. Importantly, in 200 contrast to the transient increase in σ32 concentration that occurs during an effective HS response, σ32 201 concentration remained permanently elevated in cells exposed to Kan (Fig 1E). These results indicated 202 that aberrant proteins synthesized in Kan-treated cells can be recognized as substrates by DnaK, raising 10 203 the question whether toxic species therefore might escape recognition or degradation by QC factors.

204
The permanent increase in σ32 concentration also suggested that some step downstream of σ32 205 stabilization required for increased production of pro-survival HS proteins was blocked in these cells. following ARF48 expression from a multi-copy plasmid (Fig 2D) and cell viability decreased about 10,000-226 fold by 2 h after induction (Fig 2E). ARF48 expression therefore mimics the bactericidal effect of 227 aminoglycoside antibiotics [6]. By contrast, cells remained viable for several hours following growth 228 arrest by 30 µg/ml of the bacteriostatic antibiotic Cm (Fig 2E). At higher concentration, Cm was 229 increasingly toxic (Fig 2E   To determine whether ARF48 was recognized as a substrate by DnaK in vivo, we monitored σ32 316 concentration as described above. As in Kan-treated cells, the σ32 concentration increased following 317 ARF48 induction and then remained at elevated concentration indefinitely, presumably until cell death 318 (Fig 4B). σ32 concentration increased to a lesser degree following ARF-NR induction but did not increase 319 following induction of ARF-DA (Fig 4B). ARF48 and ARF-NR therefore were recognized as substrates by 320 DnaK in vivo. These proteins also were toxic when overexpressed in an E. coli mutant that lacks 321 functional DnaK [34] however (Fig S3B), indicating that DnaK itself was not required for the toxic effect.

322
These results support the model that the QC machinery capacity to process aberrant proteins is 323 exceeded when ARF48 is overexpressed, and further suggest that the KL decapeptide mediates ARF48 324 interaction with some other factor(s) in addition to DnaK, which leads to cell death.

326
Stabilization of σ32 typically indicates that HS gene expression has been activated [10], which 327 normally would result in increased production of HS proteins that promote cell survival during periods of 328 proteotoxic stress. However, cell viability decreased following ARF48 induction (Fig 2E) despite the 329 increased concentration of σ32 in these cells, therefore indicating that the HS response was not 330 effective. To further investigate this, we monitored the concentration of several HS proteins, including 331 DnaK, DnaJ and GroEL, following ARF48 induction. Levels of these HS proteins essentially were 332 unchanged following ARF48 induction but increased markedly when non-induced cultures were grown 333 at 45°C for 10 min (Fig 4C). The HS response therefore functioned normally in these cells prior to ARF48 induction, suggesting that some step downstream of σ32 stabilization must be blocked following ARF48 335 induction.

336
To investigate whether the stabilized σ32 that accumulated following ARF48 expression could 337 associate with RNA polymerase, we expressed ARF48 in an E. coli strain that produces RNA polymerase 338 containing a 6-histidine-tagged βʹ subunit [20]. Following ARF48 induction, RNA polymerase was purified 339 from cell lysates by metal affinity chromatography, and the sigma factors associated with the enzyme 340 were then analyzed by Western blotting. RNA polymerase isolated from cells following ARF48 induction 341 indeed was associated with an increased amount of σ32 protein (E-σ32) in comparison to cells 342 expressing ARF-NR or ARF-DA (Fig 5), indicating that the stabilized σ32 that accumulates following 343 ARF48 expression is functionally intact. The failure of cells to increase production of HS proteins 344 following ARF48 expression therefore must result from a block in either transcription of HS genes by E-345 σ32 or translation of HS protein mRNAs by cell ribosomes. The initial decrease in culture OD600 following ARF48 induction (Fig 2D) suggested that some 357 cell lysis might have occurred. However, the OD600 eventually stabilized (Fig 2D), and cell proteins were 358 not detectably released into the culture medium (not shown). To further investigate the effect of ARF48 359 expression on cell membrane integrity, cells were incubated with the DNA-binding fluorescent dye 360 Hoechst 33342 (H33342). Following ARF48 induction, cells accumulated approximately 8-times more 361 H33342 than non-induced (NI) cells, based on the exposure times required to produce images of equal 362 fluorescence intensity (Fig 6). Enhanced uptake of H33342 does not necessarily indicate the membrane 363 is damaged however, because some bacteria can actively exclude H33342 by an efflux pumping 364 mechanism [36]. Whether this efflux pumping mechanism also is present in E. coli has not yet been 365 determined. Importantly, the microscopy also showed that nucleoids were radially condensed by 5-10 366 min after ARF48 induction, whereas the nucleoids in non-induced cells had a radially expanded 367 conformation (Fig 6 (Fig 7A). Cell uptake of H33342 increased markedly after 1-2 h of induction, and by 4 h many 398 nucleoids had a radially condensed conformation (Fig 7B). variants, but the smallest proteins, which were truncated after AP residues 149 or 241, had the most 414 acutely toxic effect (Fig 8A). Enhanced accumulation of the larger variants ( Fig 8C) suggests they may 415 have self-aggregated before entering the Sec pathway. Deletion of 5 residues (LLPLL) from the signal 416 peptide hydrophobic core (Δhc) attenuated the toxic effect of all truncated variants (Fig 8B), therefore 417 indicating that Sec pathway localization is required for the toxic effect of these proteins. Nucleoids were 418 rapidly condensed following expression of the acutely toxic variants (Fig 8E) and the effect was 419 suppressed by the Δhc mutation (Fig 8D). These results support the conclusion that the toxic effect of 420 aberrant proteins can result from disruption of transertion-mediated expansion of nucleoids into an