Identification of polyphosphate-binding proteins in Escherichia coli uncovers targets involved in translation control and ribosome biogenesis

Abstract

In many bacteria, polyphosphate kinase (PPK) enzymes use ATP to synthesize polyphosphate (polyP) in response to cellular stress. These chains of inorganic phosphates are joined by high-energy bonds and can reach hundreds of residues in length. PolyP plays diverse functions in helping bacteria adjust to changing environmental conditions. However, the molecular mechanisms underlying these functions are poorly understood. In eukaryotic cells, polyacidic serine- and lysine-rich (PASK) motifs of proteins can mediate binding to polyP chains. Whereas PASK motifs are relatively common in yeast and human cells, we report that these sequences are rare in bacteria commonly used for polyP research. Thus, to identify novel polyP-binding proteins in Escherichia coli, we carried out a screen and identified seven novel targets with links to translation control and ribosome biogenesis. For two targets, the GTPase activating protein YihI and the ribonuclease Rnr, we mapped the regions of polyP interaction to non-PASK sequences and identified lysine residues critical for binding. We found that deletion of rnr suppressed the slow-growth phenotype of Δppk mutants grown on minimal media. Conversely, ppk deletion resulted in decreased Rnr protein expression. These phenotypes were dependent on the polyP-binding region of Rnr but independent of polyP binding itself, suggesting a complex interplay between PPK and Rnr function in E. coli. Overall, our work provides new insights into the scope of polyP-binding proteins and extends the connections between polyP and the regulation of protein translation in E. coli.

Importance: In bacteria, polyphosphate (polyP) molecules are important regulators of cellular stress responses. Accordingly, cells that cannot make polyP display defects in processes that are important for bacterial survival, infection, and antibiotic resistance. The molecular mechanisms by which polyP exerts its functions are poorly understood. In eukaryotic cells, there has been much interest in the identification and characterization of polyP-binding proteins that act as effectors of polyP in vivo. By comparison, much less is known about polyP-binding proteins in bacteria. In this study, we take advantage of large-scale collections of Escherichia coli strains expressing epitope-tagged proteins to carry out the first systematic search for bacterial polyP-binding proteins. We describe seven novel polyP-binding proteins with links to ribosome biogenesis or translation. We further identify a complex genetic and molecular interplay between polyphosphate kinase, the enzyme that makes polyP, and the polyP-binding protein RNase R. Given the importance of translational control for bacteria survival, investigation of these pathways is expected to reveal new targets that can be leveraged for therapeutic exploration.

Keywords: E. coli; InfB; PPK; Rne; Rnr (VacB); SmpB; SrmB; YihI; YjeQ (RsgA); polyP-binding proteins; polyphosphate; ribosome; stress response.

Fig 1

Characterization of PASK sequences in E. coli. (A) Frequency of PASK motifs in bacteria. The number of proteins containing one or more PASK motifs (75% D/E/S/K content with at least one lysine within a 20 amino acid window) from reviewed proteomes of the indicated species was normalized by the total number of reviewed UniProt entries of each species. Underlying data for panel A can be found in Source Data 1. (B) Schematic of the in vitro polyP-binding assay. Whole-cell extracts incubated in the absence or presence of synthetic polyP (p700) were resolved using a Bis-Tris gel (sold under the name NuPAGE) electrophoresis. Target proteins were visualized by western blotting using an antibody toward an epitope tag or the endogenous protein. Proteins that have slower migration in the presence of polyP compared to in its absence are thought to bind polyP. (C and D) In vitro polyP binding to (C) YihI-SPA and (D) ZipA. Assays were conducted as described in panel B. In both cases, samples were resolved using NuPAGE and transferred to a PVDF membrane. YihI-SPA and ZipA were detected using anti-Flag or anti-ZipA antibodies, respectively. Ponceau S was used to show that samples migrated equally. Images are representative of results from ≥3 experiments.

Fig 2

PolyP binds a disordered lysine-rich region of YihI. (A) Schematic and amino acid distribution of full-length YihI (169 residues total). YihI has a C-terminal PASK motif and an N-terminal PASK-like region. The indicated amino acids distributed across the PASK or PASK-like sequences were targeted for mutagenesis experiments. An asterisk (*) is used to display the distribution of PASK (orange), PASK-like (blue), and other (black) lysine residues within YihI. (B) PolyP binds primarily via the N-terminus of YihI. In vitro polyP-binding assay was conducted (as described in Fig. 1B) using whole-cell extract expressing wild-type or K-R mutated GST-YihI. (C) The disorder propensity of YihI shows that the N- and C-termini are unstructured (>0.5). The graph shows the average (±SE) of computational prediction scores, represented as arbitrary units (A.U.), that were obtained using NetSurfP-3.0 (54), Metapredict (55), and IUPred3 (56). Underlying data for panel C can be found in Source Data 3. (D) The N-terminal PASK amino acids play a structural role in promoting polyP binding. Various GST-YihI mutants were grown and analyzed as described in panel B. D-N/E-Q = aspartic acid to asparagine/glutamic acid to glutamine; S-A = serine to alanine; D-A/E-L = aspartic acid to alanine/glutamic acid to leucine. For both panels B and D, samples were resolved using NuPAGE, transferred to a PVDF membrane, and probed using an anti-GST antibody. Ponceau S was used to show that samples migrated equally. Images are representative of results from ≥3 experiments.

Fig 3

A screen for novel polyP-binding proteins in E. coli. (A) A total of 589 unique E. coli proteins were screened for polyP binding. Together, the SPA and TAP collection sets contain a total of 1,024 strains with C-terminal epitope tags encoded at the chromosomal loci of relevant open reading frames. Of these, 152 proteins are redundantly tagged between the SPA and TAP collection sets, and 291 (283 non-redundant) could not be screened for polyP binding because they could not be detected by western blotting, or the strain did not grow. (B) Seven novel polyP-binding proteins were identified by the screen. Proteins that shifted from the screen were reconfirmed using the in vitro polyP-binding assay. Samples were resolved using NuPAGE, transferred to a PVDF membrane, and probed using an anti-Flag antibody which detects the SPA-tag. Ponceau S was used to show that samples migrated equally. Images are representative of results from ≥3 experiments. (C) The seven polyP-binding proteins are involved in ribosome biogenesis or translation processes. General descriptions of each protein’s functions are provided.

Fig 4

Rnr is functionally regulated by polyP. (A) Loss of rnr partially rescues the slow-growth phenotype of ppk mutants. The indicated strains were serially diluted and spotted onto LB or MOPS plates and incubated at 37°C as indicated. Images are representative of results from ≥3 experiments. (B) Schematic of the functional domains of full-length Rnr (813 residues total). Rnr has two cold shock domains (residues 1–216), a nuclease domain (residues 217–643), an S1 domain (residues 644–730), and a basic domain (residues 731–813). (C) The basic domain of Rnr has a high disorder propensity (>0.5). The graph shows the average (±SE) of computational prediction scores, represented as arbitrary units (A.U.), that were obtained using NetSurfP-3.0 (54), Metapredict (55), and IUPred3 (56). Underlying data for panel C can be found in Source Data 3.

Fig 5

The Rnr S1 and basic domains are involved in polyP binding. (A and B) The characteristic NuPAGE shift is abrogated when the S1 and basic domains of Rnr are (A) truncated or (B) mutated. An in vitro polyP-binding assay was conducted using the indicated chromosomally truncated or K-R mutated Rnr strains. Full length (FL) represents the wild-type Rnr protein expressed in a background that is isogenic to the truncated and mutated strains (see “Bacterial strains” section of the “Materials and Methods” for details on how these strains were constructed). Samples were resolved using NuPAGE, transferred to a PVDF membrane, and probed using an anti-Rnr antibody. Ponceau S was used to show that samples migrated equally. Images are representative of results from ≥3 experiments. (C) Truncation but not K-R mutation of the S1 + basic polyP-binding domain partially rescues the slow-growth phenotype of ppk mutants. The indicated strains were serially diluted and spotted onto LB or MOPS plates and incubated at 37°C as indicated. Images are representative of results from ≥3 experiments.

Fig 6

PPK regulates Rnr expression at the translational level. (A and B) Western blot and quantification of WT and K-R mutated Rnr levels. Cells grown in LB were nutrient downshifted to MOPS media for 3 hours. Samples were resolved using a 4%–20% Criterion TGX Stain-Free gel, transferred to a PVDF membrane, and probed using an anti-Rnr antibody. Ponceau S was used to show equal loading. Three replicates of each condition were used for quantification as described in the “Materials and Methods” section. Mean values with SD are shown. Indicated are the P-values and nonsignificant differences (n.s.), calculated using two-way ANOVA with Tukey’s post hoc analysis. Underlying data for panel B can be found in Source Data 4. Western blot images are representative of results from ≥3 experiments. (C) Reverse transcriptase quantitative PCR measurements of Rnr transcript levels in WT and ∆ppk mutants. Cells were grown as described for panel A and B. Transcript levels were normalized to yqfB, and arnA served as a PPK-dependent positive control. Mean values with SD are shown. Indicated are the P-values and nonsignificant differences (n.s.), calculated using a two-tailed unpaired t-test with Welch’s correction. Underlying data for panel C can be found in Source Data 5. (D) Rnr turnover in WT and ∆ppk mutants during growth in MOPS media. Cells grown in LB and shifted to MOPS media for 30 minutes were treated with chloramphenicol (CM) to stop translation. Untreated controls were used to evaluate Rnr expression in the absence of translation shutoff. Cells were harvested at the indicated time points. Samples were resolved using 10% SDS-PAGE, transferred to a PVDF membrane, and probed using an anti-Rnr antibody. Images are representative of results from ≥3 experiments. (E) Model of how polyP regulates Rnr function. PolyP produced by PPK regulates translation but not transcription of Rnr through an unknown mechanism. Additionally, direct binding of polyP may regulate functions of Rnr that have yet to be investigated.