Elongation factor P modulates Acinetobacter baumannii physiology and virulence as a cyclic dimeric guanosine monophosphate effector

Significance

Many bacteria use the second messenger cyclic diguanosine monophosphate (c-di-GMP) to control diverse biological functions. Here, we show that A1S_2419, an elongation factor P (EF-P) from the human pathogenic bacterium Acinetobacter baumannii, is a c-di-GMP effector protein. It controls the translation efficiency of the target proteins with consecutive prolines. Intriguingly, A1S_2419 modulates biofilm formation, motility, and virulence in A. baumannii in a c-di-GMP–dependent manner. This article describes a novel c-di-GMP signaling mechanism with which a pathogen uses EF-P to integrate the second messenger signal turnover and translational regulation of target genes.

Abstract

Cyclic diguanosine monophosphate (c-di-GMP) is widely used by bacteria to control biological functions in response to diverse signals or cues. A previous study showed that potential c-di-GMP metabolic enzymes play a role in the regulation of biofilm formation and motility in Acinetobacter baumannii. However, it was unclear whether and how A. baumannii cells use c-di-GMP signaling to modulate biological functions. Here, we report that c-di-GMP is an important intracellular signal in the modulation of biofilm formation, motility, and virulence in A. baumannii. The intracellular level of c-di-GMP is principally controlled by the diguanylate cyclases (DGCs) A1S_1695, A1S_2506, and A1S_3296 and the phosphodiesterase (PDE) A1S_1254. Intriguingly, we revealed that A1S_2419 (an elongation factor P [EF-P]), is a novel c-di-GMP effector in A. baumannii. Response to a c-di-GMP signal boosted A1S_2419 activity to rescue ribosomes from stalling during synthesis of proteins containing consecutive prolines and thus regulate A. baumannii physiology and pathogenesis. Our study presents a unique and widely conserved effector that controls bacterial physiology and virulence by sensing the second messenger c-di-GMP.

The second messenger bis-(3′-5′)-cyclic guanosine monophosphate (c-di-GMP) controls the bacterial biofilm and motility switch and many other biological functions, including quorum sensing signal production and virulence (1–4). It is enzymatically synthesized from two molecules of guanosine-5'-triphosphate (GTP) by diguanylate cyclases (DGCs) harboring GGDEF domains and degraded to the linear dinucleotide pGpG or to guanosine monophosphate (GMP) by c-di-GMP–specific phosphodiesterases (PDEs) containing EAL or HD-GYP domains (5, 6). Many bacterial species contain multiple proteins involved in c-di-GMP synthesis and degradation, indicating the importance and diverse roles of the c-di-GMP signaling system in bacterial physiology (7, 8).

The c-di-GMP metabolic enzymes are readily predictable because of the characteristic GGDEF, EAL, and HD-GYP domains. However, it is challenging to decipher how c-di-GMP could act on target genes, because this signal has diverse receptors or effector proteins with unconserved binding sites and motifs (5). A few c-di-GMP receptor families, including PilZ domain receptors, inhibitory site receptors, and EAL domain receptors, can be predicted on the basis of primary sequences (5). There are also some unpredictable c-di-GMP receptors that have binding sites without defined consensus motifs that include various kinds of transcriptional regulators and proteins with diverse and unrelated functions (3–5, 9–14). Recently, the cis-2-dodecenoic acid receptor RpfR was found to be a c-di-GMP sensor, and binding of c-di-GMP by RpfR decreased the affinity of the RpfR-GtrR complex to target promoter DNA (15, 16). In addition, members of a particular class of riboswitches were also found to bind to c-di-GMP (17, 18).

Proline contains a unique constrained ring structure that makes it a poor peptide bond donor and acceptor (19). Polyproline stretches cause ribosome pausing and influence the translation elongation rate (20). In addition to polyproline-containing proteins, specific subsets of proteins containing diprolyl motifs (XPP/PPX) also cause ribosome pausing (21). Elongation factor P (EF-P) binds the stalled ribosomes and facilitates peptide bond formation between prolines by forcing the prolines to adopt an alternative conformation (22, 23).

Here, we found that A1S_2419, an EF-P, controls biological functions, including biofilm formation, motility, and virulence, in the human pathogen Acinetobacter baumannii. A1S_2419 controls the translation efficiency of target proteins with consecutive prolines, and the binding of c-di-GMP enhances the function of A1S_2419 to promote translation efficiency. Thus, A1S_2419 serves as a unique sensor and effector of c-di-GMP. Taken together, our findings have unveiled a novel effector of c-di-GMP, which enables integration of the intracellular c-di-GMP signal into translational machinery to modulate bacterial physiology and virulence.

Results

c-di-GMP Is Involved in the Regulation of Biological Functions and Virulence in A. baumannii.

It was previously revealed that potential c-di-GMP metabolic enzymes are involved in the regulation of biofilm formation and motility in A. baumannii (24). To verify the existence of intracellular c-di-GMP, the collected extract of A. baumannii American Type Culture Collection (ATCC) 17978 cells was analyzed by using liquid chromatography coupled with tandem mass spectrometry. As shown by triple-quadrupole mass spectrometry (SI Appendix, Fig. S1A), the simultaneous appearance of three characteristic mass transitions (m/z 691/152, m/z 691/248, and m/z 691/540) verified the existence of c-di-GMP in A. baumannii cells.

To further confirm that changes in the intracellular level of c-di-GMP could affect the cellular functions of A. baumannii, we expressed in trans the rocR and wspR genes from Pseudomonas aeruginosa that encode a well-characterized c-di-GMP phosphodiesterase and synthase, respectively, in the wild-type strain of A. baumannii. As expected, the expression of the c-di-GMP degradation protein RocR caused a decrease in the intracellular c-di-GMP level and resulted in decreased biofilm formation capacity but enhanced the ability to move on a semisolid surface termed “surface motility” (25) (SI Appendix, Fig. S1 B–D). In contrast, overexpression of wspR increased the intracellular concentration of c-di-GMP in A. baumannii and promoted biofilm formation but reduced motility (SI Appendix, Fig. S1 B–D). A. baumannii usually causes lung infections (26). The human lung adenocarcinoma cell line A549 was used as an in vitro model to test the effect of intracellular c-di-GMP levels on the cytotoxicity of A. baumannii. It was revealed that in trans expression of rocR in the wild-type strain decreased cytotoxicity by 33%, whereas overexpression of wspR increased the cytotoxicity of A. baumannii by 121% (SI Appendix, Fig. S1E). These results suggest that c-di-GMP signaling regulates the important biological functions and pathogenicity of A. baumannii.

c-di-GMP Metabolic Enzymes Affect Biological Functions through c-di-GMP.

There are 11 potential c-di-GMP metabolic enzyme-encoding genes that have been identified in the genome of A. baumannii ATCC 17978 (24). Six genes encoding the GGDEF domain were located at the A1S_0751, A1S_1067, A1S_1695, A1S_2506, A1S_2986, and A1S_3296 loci. Two genes encoding the EAL domain were located at the A1S_1254 and A1S_2422 loci, and three genes encoding a GGDEF and an EAL domain were located at the A1S_0546, A1S_1949, and A1S_2337 loci. We found a new GGDEF domain–encoding gene at the A1S_3345 loci by using the Basic Local Alignment Search Tool (BLAST; https://blast.ncbi.nlm.nih.gov/Blast.cgi) algorithm (SI Appendix, Table S1), but we did not find any gene encoding HD-GYP domain in the genome of A. baumannii ATCC 17978.

We then investigated the possible relationship between the potential c-di-GMP metabolic enzymes and intracellular levels of c-di-GMP in A. baumannii. Considering the existence of compensatory mechanisms and functional redundancy in bacteria, we constructed a DGC-null mutant of A. baumannii 17978 by deleting all GGDEF domain–encoding genes. Starting with deletion of A1S_2506, we then sequentially deleted the other GGDEF domain–encoding genes, including A1S_0751, A1S_1067, A1S_1695, A1S_2986, A1S_3296, and A1S_3345, and the GGDEF domain–encoding regions of A1S_0546, A1S_1949, and A1S_2337 in A. baumannii ATCC 17978, generating the Δ10DGC deletion mutant strain. The Δ10DGC deletion mutant strain exhibited stronger motility and an obvious reduction in both biofilm formation and cytotoxicity compared with the wild-type strain (Fig. 1 A–C). Consistent with this, the intracellular c-di-GMP level in the Δ10DGC mutant was almost undetectable (Fig. 1D).

Fig. 1.

Phenotypes of multiple c-di-GMP metabolic enzyme-encoding gene deletion mutants. The Δ10DGC strain was constructed by deleting A1S_0751, A1S_1067, A1S_1695, A1S_2506, A1S_2986, A1S_3296, and A1S_3345 and the GGDEF domain-encoding regions of A1S_0546, A1S_1949, and A1S_2337. The Δ4PDE strain was generated by deleting A1S_1254, and the EAL domain-encoding regions of A1S_0546, A1S_1949, and A1S_2337. The biofilm formation (A), surface-associated motility (B), cytotoxicity (C), and intracellular c-di-GMP level (D) of DGC- and PDE-null mutants were measured. For ease of comparison, the c-di-GMP level of each mutant was normalized to that of the wild-type (WT) strain. The results were analyzed using one-way ANOVA for at least three independent experiments. ***P < 0.001. Error bars indicate the standard deviations (SD).

An in-frame deletion mutant strain was also constructed by deleting EAL domain–encoding gene A1S_1254 and the EAL domain–encoding regions of A1S_2337, A1S_1949, and A1S_0546, but not that of A1S_2422. A1S_2422 did not exhibit PDE activity and c-di-GMP binding ability in vitro, and in-frame deletion of it did not affect the intracellular level of c-di-GMP (SI Appendix, Fig. S2). In addition, the catalytic site and loop 6 motif of the A1S_2422 EAL domain are alanine residue (Ala⁶⁷⁰) and RDFASSMYS⁶²¹, respectively, which deviate from glutamine residue (Glu³⁵²) and DDFGAGYSS³⁰³ of a typical EAL domain–containing protein such as RocR (27, 28) (SI Appendix, Table S1). The generated Δ4PDE mutant strain showed impaired motility and clearly enhanced biofilm formation and cytotoxicity (Fig. 1 A–C). In contrast to the intracellular level of c-di-GMP in the Δ10DGC mutant strain, the level of c-di-GMP in the Δ4PDE mutant strain was increased to 201% of that in the wild-type strain (Fig. 1D).

Major DGCs and PDEs in A. baumannii.

To further study the enzymatic activity and roles of these c-di-GMP metabolic enzymes in A. baumannii, we complemented the Δ10DGC and Δ4PDE mutant strains with each single c-di-GMP metabolic enzyme. As shown in SI Appendix, Fig. S3A, in trans expression of A1S_1695, A1S_2506, and A1S_3296 increased the biofilm formation of the Δ10DGC mutant strains by 228%, 443%, and 352%, respectively. The motility of the Δ10DGC mutant strains was reduced by 35%, 44%, and 31% with overexpression of A1S_1695, A1S_2506, and A1S_3296, respectively (SI Appendix, Fig. S3B). The analysis results for the model human cell line showed that in trans expression of A1S_1695, A1S_2506, and A1S_3296 partially restored the virulence of the Δ10DGC mutant strains to the wild-type level (SI Appendix, Fig. S3C). We continued to test the potential role of the PDEs and found that in trans expression of A1S_1254 in the Δ4PDE mutant strain increased the motility of this mutant strain by 357% (SI Appendix, Fig. S3E) and restored biofilm formation and cytotoxicity to the wild-type level (SI Appendix, Fig. S3 D and F).

The in vitro DGC or PDE activity of A1S_1695, A1S_2506, A1S_3296, and A1S_1254 was also investigated. More than 86% of GTP was converted to c-di-GMP by the A1S_2506 protein within 5 min (SI Appendix, Fig. S4 A–C), while the amount of GTP converted by A1S_1695 and A1S_3296 was only ∼70% within 2 h (SI Appendix, Figs. S4 D–F and G–I). We observed that A1S_1254 degraded almost all of the c-di-GMP to GMP rather than pGpG within 5 min (SI Appendix, Fig. S4 J–L). Because A1S_0546, A1S_1949, and A1S_2337 all contain a GGDEF-EAL tandem, their DGC and PDE activities were also investigated. Both of the GGDEF domains of A1S_0546 and A1S_1949 showed a weak activity for converting GTP to c-di-GMP, while the in vitro DGC activity of A1S_2337 GGDEF domain was not observed (SI Appendix, Fig. S5 A–G). Similar to A1S_1254, the EAL domain of A1S_2337 degraded c-di-GMP to GMP, while the EAL domains of A1S_0546 and A1S_1949 hydrolyzed c-di-GMP to pGpG and further to GMP (SI Appendix, Fig. S5 H–N). It was also previously observed that some proteins containing the EAL domain hydrolyzed c-di-GMP to pGpG, and then pGpG was further degraded to GMP (28, 29). In this study, we observed that after incubation with the EAL domains of A1S_0546, A1S_1254, A1S_1949, and A1S_2337, c-di-GMP was finally hydrolyzed to GMP (SI Appendix, Fig. S5).

The effect on the c-di-GMP concentration of deleting each individual gene was also investigated. Interestingly, only the ΔA1S_1254 mutant showed a 40% increase, and the ΔA1S_1695, ΔA1S_2506, and ΔA1S_3296 mutants exhibited 16%, 20%, and 18% decreases, respectively, in the level of c-di-GMP compared with that of the wild-type strain (SI Appendix, Fig. S6A). These results agree well with the results of the phenotypic analysis. Consistent with the elevated intracellular c-di-GMP concentration, the ability of the ΔA1S_1254 mutant to form biofilms increased by 38%, but the motility was decreased by 25% (SI Appendix, Fig. S6 B and C). In contrast, ΔA1S_1695, ΔA1S_2506, and ΔA1S_3296 displayed 14%, 28%, and 15% decreases, respectively, in biofilm formation, but 16%, 43%, and 39% increases, respectively, in motility activity compared with the wild-type strain (SI Appendix, Fig. S6 B and C). Together, these results suggested that A1S_1695, A1S_2506, A1S_3296, and A1S_1254 are the main c-di-GMP turnover proteins in A. baumannii ATCC 17978 under the investigated experimental conditions.

To comprehensively determine the regulatory roles of c-di-GMP in the transcriptome of A. baumannii ATCC 17978, we analyzed and compared the gene transcription profiles of the wild-type strain and the ΔA1S_1254 mutant strain by using RNA sequencing (RNA-seq). Differential gene expression analysis showed that 72 genes were upregulated and 66 genes were downregulated (log₂ fold-change ≥ 1.5) in the ΔA1S_1254 mutant strain compared with that in the wild-type strain (SI Appendix, Fig. S7 and Table S2). These genes were associated with a diverse range of biological functions, including cell motility, DNA replication, transport, transcription, membrane components, and signal transduction (SI Appendix, Fig. S7A and Table S2). qRT-PCR analysis of select genes confirmed the RNA-seq results (SI Appendix, Fig. S7B). These genes include the csu pilus (A1S_2213 to A1S_2218) and outer membrane protein A (A1S_2840) genes, which were shown to be involved in biofilm formation in A. baumannii (30, 31).

EF-P Rescues the Phenotypes of Δ10DGCs.

Because c-di-GMP plays a role in regulating the important biological functions and virulence of A. baumannii, we next searched for homologs of c-di-GMP effectors in the genome of A. baumannii ATCC 17978 to identify the potential downstream component of the c-di-GMP signaling system by using the BLAST program. We found 13 potential homologs of c-di-GMP effectors sharing 24.42–52.63% of their identity with the identified c-di-GMP effectors or receptors (SI Appendix, Table S3). We expressed these homologs in trans in the Δ10DGC mutant and measured the biofilm formation and motility of these overexpression strains. The results showed that in trans expression of A1S_2421, which contains a PilZ domain analyzed by the Pfam database (pfam.xfam.org), partially rescued the biofilm formation defect and reduced the motility of Δ10DGCs (SI Appendix, Fig. S8). Interestingly, in trans expression of the neighboring gene of A1S_2421 (A1S_2419) also showed a similar effect in restoring all the tested phenotypes of Δ10DGCs, including biofilm formation, motility, and cytotoxicity, which are all regulated by the c-di-GMP signaling system (Fig. 2). Because the PilZ domain is widely recognized as a c-di-GMP effector, we then focused our attention on A1S_2419.

Fig. 2.

EF-P complements the phenotypes of the Δ10DGC mutant strain. Genomic organization of the region encoding EF-P in A. baumannii ATCC 17978 (A) and domain structure analysis of EF-P (B). EFP, elongation factor P oligonucleotide binding domain; Elong-fact P_C, elongation factor P, C-terminal. The domains were analyzed by using SMART. Overexpression of efp promoted the biofilm formation (C), inhibited the surface-associated motility (D), and increased the cytotoxicity (E) of the Δ10DGC mutant. The results in C–E were analyzed using one-way ANOVA for at least three independent experiments. ***P < 0.001. WT, wild-type. Error bars indicate the standard deviations (SD).

A1S_2419 is an efp encoding EF-P (Fig. 2 A and B), which recognizes and binds to the E-site region of ribosomes when it translates proteins containing consecutive prolines (32). Previous studies have shown that EF-P plays an important role in bacterial motility, fitness, and pathogenicity (33–36). Proteome analysis revealed that A. baumannii ATCC 17978 contained a total of 1,367 XPPX motifs, and approximately 60 proteins contained 3 or more consecutive prolines (SI Appendix, Table S4), including BfmS and CsuD. BfmS is the sensor kinase of the two-component system BfmSR, which controls the production of outer membrane vesicles (37), and CsuD is an important constituent of the csu pilus (31). Both of those proteins have been identified as being involved in biofilm formation, motility, and virulence in A. baumannii (31, 37–39). Because there is no information on the relationship between EF-P and c-di-GMP, we continued to study the role of A1S_2419 in the c-di-GMP signaling pathways in A. baumannii.

EF-P Modulates c-di-GMP–Regulated Phenotypes.

We then tried to prepare an efp-null mutant by inserting a gentamycin cassette into the gene or deleting the region encoding efp from the A. baumannii ATCC 17978 genome. However, we failed to obtain any mutant colonies. The results suggested that this gene may be essential for A. baumannii. To further confirm this possibility, A. baumannii cells were supplemented with a pME2 plasmid carrying the efp gene, which was expressed under the control of a vanillate-inducible promoter (3). In the presence of 1 mM vanillate, we successfully generated an in-frame efp deletion mutant (Δefp/Pvan-efp). The deletion mutant strain could not survive in the absence of vanillate (Fig. 3A), suggesting that efp is an essential gene for A. baumannii. The essentiality of EF-P is debated and may differ among organisms. For example, it is an essential gene for Neisseria meningitidis viability (40). However, in some other bacteria, including Escherichia coli, deletion of EF-P is not lethal, but it severely impairs bacterial growth (35, 36). As shown in Fig. 3B, the Δefp/Pvan-efp mutant exhibited an extended lag phase when cultured at 37 °C with shaking, even in the presence of 1 mM vanillate. In addition, the cell density of Δefp/Pvan-efp was significantly lower than that of the wild-type strain when the cells were cultured under stationary conditions (SI Appendix, Fig. S9). The results indicated that efp plays an important role in A. baumannii growth.

Fig. 3.

Effects of EF-P on A. baumannii physiology. (A) Growth of the wild-type (WT), Δefp/Pvan-efp, and Δefp/Pvan-efp complemented with efp under the control of its native promoter (Δefp/Pvan-efp(efp)) strains on the solid surface in the absence or presence of vanillate (1 mM). The plates were incubated for 12 h. The experiment was performed three times, and representative images from one experiment are shown. (B) The Δefp/Pvan-efp mutant had an extended lag phase compared with the wild-type strain. Cells were grown in LB broth supplemented with vanillate (1 mM) with shaking. (C) qRT-PCR analysis of efp expression in the Δ10DGC, Δ4PDE, and Δefp/Pvan-efp strains compared with that of the wild-type strain. (D) The total yield of EGFP with four proline residues at the N terminus (4P-GFP) was affected in the Δefp/Pvan-efp mutant. Data were analyzed using two-way ANOVA for three independent experiments. The fluorescence was measured when cells were grown to OD₆₀₀ = 1.0. a.u., arbitrary units. Biofilm formation (E), motility (F), and cytotoxicity (G) of the Δefp/Pvan-efp mutant were measured in the presence of 1 mM vanillate. For ease of comparison, the phenotypes of the Δefp/Pvan-efp mutant were normalized to those of the wild-type strain. The results in D–G are analyzed using one-way ANOVA for at least three independent experiments. ***P < 0.001. ns, not significant. Error bars indicate the standard deviations (SD).

In the presence of 1 mM vanillate, the transcription level of efp in the Δefp/Pvan-efp mutant was significantly lower than that in the wild-type strain (Fig. 3C). To test the impacts of efp on the yield of proteins containing consecutive prolines, we transformed a reporter plasmid containing enhanced green fluorescent protein (EGFP) or EGFP with four consecutive proline residues inserted at the N terminus (4P-EGFP) into the Δefp/Pvan-efp strain; we then measured the fluorescence intensity of EGFP or 4P-EGFP when cells were grown to an OD₆₀₀ (optical density at 600 nm) of 1.0. As expected, the EGFP fluorescence in the Δefp/Pvan-efp strain was similar to that in the wild-type strain, while the fluorescence of 4P-EGFP in the Δefp/Pvan-efp strain was reduced about 10-fold compared with that in the wild-type strain (Fig. 3D), confirming the EF-P role of A1S_2419. We also observed that the biofilm formation of Δefp/Pvan-efp was reduced by 77%, while its motility was increased by ∼24% (Fig. 3 E and F) compared with that of the wild-type strain. To test whether efp impacts the pathogenicity of A. baumannii, we investigated the wild-type strain and the Δefp/Pvan-efp mutant in a cell line infection assay. As shown in Fig. 3G, the cytotoxicity of the Δefp/Pvan-efp strain decreased to 45% of that of the wild-type A. baumannii strain. To verify that these phenotypic changes were due to deletion of the efp locus, we complemented Δefp/Pvan-efp strain with pAb-MCS plasmid containing efp and its native promoter to generate the complemented strain Δefp/Pvan-efp(efp). It was observed that complementation fully restored the phenotypes to levels comparable to those in the wild-type strain (Fig. 3 D and E). Together, these results showed that EF-P is an essential gene for A. baumannii and mediates c-di-GMP–regulated phenotypes.

EF-P Binds c-di-GMP.

Since EF-P modulates c-di-GMP–regulated functions and since it complemented the phenotypes of Δ10DGCs, we first speculated that c-di-GMP may regulate efp transcription. However, the transcriptional levels of efp were similar among the wild-type, Δ10DGC, and Δ4PDGE strains (Fig. 3C), and Western blotting assays showed that there is no difference among the EF-P levels in these strains (SI Appendix, Fig. S10). These results suggested that c-di-GMP may influence the activity of EF-P through ligand–protein interactions. Therefore, we then applied differential scanning fluorimetry (DSF) and microscale thermophoresis (MST) to test whether EF-P interacts with c-di-GMP. The DSF assay showed a marked shift in the melting curve of EF-P, suggesting that EF-P binds to c-di-GMP (Fig. 4 A and B). MST analysis revealed that this protein bound c-di-GMP with an estimated dissociation constant (KD) of 9.82 ± 0.54 μM (Fig. 4C).

Fig. 4.

c-di-GMP binds to EF-P (A1S_2419). (A) Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) of the purified EF-P protein. M, protein marker. (B) Protein melting curves of EF-P in the absence or presence of c-di-GMP (500 µM). The maximal fluorescence was normalized by 100%. PBS, phosphate-buffered saline; RFU, relative fluorescence units. Error bars indicate the standard deviations (SD). (C) MST analysis of the interaction of c-di-GMP and EF-P. Fnorm (‰) indicates the fluorescence time trace changes in the MST response. Error bars indicate the standard deviations (SD). (D) Isothermal titration calorimetry (ITC) analysis of the molecular interaction between c-di-GMP and EF-P of A. baumannii. The results in B and C are the mean ± SD of at least three independent experiments. N, binding stoichiometry; H, enthalpy.

To further confirm the binding between EF-P of A. baumannii and c-di-GMP, we used isothermal titration calorimetry analysis, and the result showed that EF-P tightly bound to c-di-GMP with an estimated KD of 5.9 μM (Fig. 4D). c-di-GMP is a ubiquitous second messenger and EF-P is widely conserved in diverse bacteria, so we investigated whether EF-P proteins of other bacterial species, including E. coli, P. aeruginosa, Burkholderia cenocepacia, and A. albensis, bind to c-di-GMP. Intriguingly, we found that the EF-P proteins from E. coli, P. aeruginosa, and B. cenocepacia did not bind to c-di-GMP, but the EF-P of A. albensis bound to c-di-GMP with an estimated KD of 10.7 μM (SI Appendix, Fig. S11). We also tested some other compounds such as GMP, adenosine monophosphate (AMP), and cyclic adenosine monophosphate (cAMP) and found that EF-P did not bind them (SI Appendix, Fig. S12), suggesting that the binding between EF-P and c-di-GMP is specific.

We then continued our attempt to identify the binding sites of c-di-GMP in the EF-P of A. baumannii. Autodocking analysis revealed five amino acid residues at positions Lys47(K47), Glu69(E69), Asp102(D102), Val127(V127), and Asn128(N128) that might be critical for the interaction between EF-P and c-di-GMP (SI Appendix, Fig. S13A). We then generated five single-point mutants (EF-P^K47A, EF-P^E69A, EF-P^D102A, EF-P^V127A, and EF-P^N128A) (SI Appendix, Fig. S13B). MST analysis showed that mutations at K47 and E69 remarkably weakened the binding between EF-P and c-di-GMP (SI Appendix, Fig. S13 F and G).

c-di-GMP Enhances the Modulatory Effect of EF-P on the Translation of Proteins Containing Consecutive Prolines.

We then investigated the influence of c-di-GMP on the function of EF-P to control the translation efficiency of the target proteins with consecutive prolines. As shown in Fig. 5A, there was no difference in the fluorescence of EGFP among the wild-type, Δ10DGC, and Δ4PDE mutant strains. The 4P-EGFP level in the Δ10DGC mutant and Δ4PDE mutant was 11% and 130% of that in the wild-type strain, respectively. Then, the Δ10DGC mutant was transformed with an EGFP reporter plasmid harboring WspR, the expression of which was controlled by the isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible tac promoter. As shown in Fig. 5B, the translation of 4P-EGFP was positively correlated with the concentration of IPTG, while the translation of EGFP was not affected by c-di-GMP. In parallel, we also analyzed whether 4P-EGFP translation in Δ10DGCs was impacted by efp under conditional expression. We observed that 4P-EGFP translation also occurred in an EF-P–dependent manner (Fig. 5C), suggesting that the translation of consecutive proline-containing proteins is modulated by EF-P, the activity of which was enhanced by c-di-GMP.

Fig. 5.

c-di-GMP influences the translation of polyproline motif-containing proteins through EF-P. (A) The fluorescence intensity of EGFP or 4P-EGFP in the wild-type (WT), Δ10DGC, and Δ4PDE strains. The 4P-EGFP production was recorded as a function of c-di-GMP (B) and EF-P (C) in the Δ10DGC mutant strain. WspR and EF-P are expressed from an IPTG-inducible tac promoter. The fluorescence was measured when cells were grown to OD₆₀₀ = 1.0. The results are analyzed using two-way ANOVA for at least three independent experiments. ***P < 0.001. a.u., arbitrary units. Error bars indicate the standard deviations (SD).

To further characterize the relationship between EF-P function and c-di-GMP sensing, we chose the two variant proteins EF-P^K47A and EF-P^E69A that almost lost their activity to bind with c-di-GMP. It was observed that the function of both EF-P^K47A and EF-P^E69A was reduced for modulating the translation of 4P-EGFP, biofilm formation, and motility compared with the wild-type EF-P in the presence of c-di-GMP (SI Appendix, Fig. S14). We continued to construct a Δ10DGCΔefp/Pnat-efp^E69A mutant in which the EF-P^E69A was expressed from the native promoter in the pME2 plasmid. As expected, both EF-P and EF-P^E69A complemented the 4P-EGFP translation with similar activity in the absence of c-di-GMP (Fig. 6A). However, because EF-P^E69A was different from the wild-type EF-P, it could not sense c-di-GMP to modulate the translation of 4P-EGFP (Fig. 6B). Our results indicated that the function of EF-P is enhanced in the presence of c-di-GMP, while EF-P still has a basal level of activity in the absence of c-di-GMP.

Fig. 6.

Analysis of EF-P regulatory mechanism to modulate the translation of polyproline motif–containing proteins by sensing c-di-GMP. (A) The fluorescence intensity of EGFP and 4P-EGFP in the Δ10DGCsΔefp/Pnat-efp^E69A mutant was recorded after transforming with a plasmid containing EF-P or EF-P^E69A, which was expressed from an IPTG-inducible tac promoter. (B) The fluorescence intensity of EGFP and 4P-EGFP in the Δ10DGC mutant and Δ10DGCsΔefp/Pnat-efp^E69A mutant was recorded after transforming with a plasmid containing WspR, which was expressed from an IPTG-inducible tac promoter. The results are analyzed using two-way ANOVA for at least three independent experiments. ***P < 0.001. a.u., arbitrary units. Error bars indicate the standard deviations (SD).

To further confirm the influence of c-di-GMP on EF-P, we first measured the transcriptional levels of bfmS and csuD via both qRT-PCR and β-galactosidase activity assays because both BfmS and CsuD contain consecutive prolines (SI Appendix, Fig. S15 A and B and Table S4). Although the transcriptional level of csuD was dependent on c-di-GMP, we observed that bfmS transcription was similar in the wild-type, Δ10DGC, and Δ4PDE strains, suggesting that the transcription of bfmS was not dependent on c-di-GMP. Then, the BfmS protein levels in the wild-type, Δ10DGC, and Δ4PDE strains were determined by fusion of the genes and their native promoters with EGFP. As shown in SI Appendix, Fig. S15C, the fluorescence signals of BfmS fusion proteins were dependent on both c-di-GMP and EF-P. However, when the consecutive prolines of BfmS (PPP³⁹⁷) were mutated to AAA, there was no difference in the fluorescence of BfmS^AAA fusion protein among the wild-type, Δ10DGC, and Δ4PDE strains.

Discussion

As an important player in bacterial physiology, c-di-GMP not only determines the bacterial sessile-to-motile lifestyle transition but also binds to a range of effector components and controls diverse functions (41). A. baumannii ATCC 17978 harbors 12 potential c-di-GMP metabolic enzymes (SI Appendix, Table S1), and some of them were reported to be involved in the regulation of bacterial physiology, including biofilm formation and motility (24). However, the occurrence of GGDEF-EAL domain proteins does not imply the existence of c-di-GMP. For example, Staphylococcus has only remnants of c-di-GMP systems and cannot synthesize c-di-GMP, but the bacterium harbors a GGDEF domain protein to control biofilm formation through a c-di-GMP–independent mechanism (42). In this study, we isolated and characterized c-di-GMP in A. baumannii (SI Appendix, Fig. S1A) and found that this signal plays an important role in the regulation of biofilm formation, motility, and virulence (Fig. 1). Our findings provide additional evidence for the role of this kind of second messenger, which plays a vital role and is also widely present in bacteria.

Although there are 19 potential c-di-GMP metabolic enzymes in Dickeya dadantii strain EC1, it was found that only some of them (i.e., W909_14950, W909_10355, W909_02155, and W909_14945) were involved in c-di-GMP signaling (43), suggesting that not all GGDEF-EAL domain-containing proteins influence the intracellular c-di-GMP pool. In this study, we showed that only deletion of A1S_1695, A1S_2506, A1S_3692, and A1S_1254 caused apparent changes in the intracellular levels of c-di-GMP and phenotypes of A. baumannii under the investigated growth conditions (SI Appendix, Fig. S6). This might be a result of the compensatory mechanism and of functional redundancy, which were previously indicated to be responsible for the absence of phenotypic changes in some GGDEF-EAL protein deletion mutants of Sinorhizobium meliloti (1). Another explanation is that there could be a nonfunctional catalytic motif in these proteins that leads to loss of the c-di-GMP metabolic enzyme activity. For example, the GGDEF motif of A1S_2337 is replaced by NGDDF (SI Appendix, Table S1), resulting in the absence of DGC activity (24) (SI Appendix, Fig. S5 F and G). The structural model of RocR revealed that the helix loop 6 motif (DDFG(T/A)GYSS) of the EAL domain is essential for enzymatic activity (44), and the first glutamate residue in the EGVE motif is considered to be the catalytic base of the EAL domain (27). The helix loop 6 and the catalytic base of A1S_2422 were degenerated into RDFASSMYS and alanine, respectively (SI Appendix, Table S1), which might be the reason for the inactivation of the EAL domain of A1S_2422 (SI Appendix, Fig. S2). An additional possible explanation is that environmental factors also contribute to the deficiency in the activity of some c-di-GMP metabolic enzymes. For example, the PDE activity of PdeS is strongly inhibited by dephosphorylated EIIA^Glc in the presence of glucose (45). Together, these results suggest that the function of c-di-GMP metabolic enzymes is the outcome of evolution, metabolism, and environment combined.

Protein synthesis is dependent on ribosomal subunits, aminoacyl-transfer RNAs, messenger RNAs, and translation factors (46). EFs are a set of proteins that function at the ribosome to facilitate translational elongation from the formation of the first to the last peptide bond of a growing polypeptide during protein synthesis. There are 5 important EFs currently found in prokaryotes: EF-Tu, EF-Ts, EF-G, EF-P, and EF-4 (47–51). These factors play vital roles in the translation process and definitely modulate bacterial physiology. It was also revealed that some of these factors modulate the pathogenicity of pathogenic bacteria. Among them, EF-P is famous for its function in alleviating ribosome stalling during the translation of polyproline motifs (20, 50). In our study, we identified an EF-P (A1S_2419) that is essential for A. baumannii growth (Fig. 3A). Our research also showed that A1S_2419 not only facilitates the translational elongation of proteins with consecutive prolines but also binds to c-di-GMP and influences the bacterial physiology and virulence of A. baumannii (Figs. 2 and 3).

Studies over the last 30 years show that c-di-GMP commonly controls various biological functions by interacting with diverse receptor or effector proteins (3–5, 9–18). Interestingly, different from these paradigms, the findings from this study have unveiled a new regulatory mechanism with which c-di-GMP could modulate bacterial physiology and virulence through multiple signaling pathways. On one hand, c-di-GMP regulates the target gene at the transcriptional level through a potential PilZ domain–containing effector or other unknown effectors (SI Appendix, Figs. S8 and S15). This conclusion was supported by the fact that c-di-GMP controls the transcriptional levels of various genes, such as csuD and ompA (SI Appendix, Figs. S7 and S15 A and B and Table S2). On the other hand, c-di-GMP regulates the target gene at the translational level through an EF-P (Fig. 5). Furthermore, c-di-GMP could control target genes such as csuD at both the transcriptional and translational levels (SI Appendix, Fig. S15). Intriguingly, some transcriptional regulators might also be the targets controlled by c-di-GMP and EF-P at the translational level because they contain three consecutive prolines (SI Appendix, Table S4). Together, these results suggest that c-di-GMP uses a complex and unique regulatory mechanism in A. baumannii.

A BLAST search with A1S_2419 revealed that this kind of EF-P is highly conserved in diverse bacterial species including members of the genera Acinetobacter, Halomonas, Pseudoalteromonas, Vibrio, Xanthomonas, and Yersinia (SI Appendix, Table S5). In conclusion, our data suggest that EF-P is a unique effector of c-di-GMP that is present in various bacterial pathogens. Our findings may inspire further studies to investigate roles and mechanisms of EF-P in c-di-GMP signaling systems in diverse bacterial genera.

Materials and Methods

Bacterial Growth Conditions and Virulence Assays.

The bacterial strains used in this work are listed in SI Appendix, Table S6. A. baumannii ATCC 17978 strains were cultured at 37 °C in lysogeny broth (LB) Lennox. The following antibiotics were added when necessary: ampicillin, 100 μg/mL⁻¹; kanamycin, 50 μg/mL⁻¹; apramycin, 100 μg/mL⁻¹; and tetracycline, 12.5 μg/mL⁻¹. A cell line infection model was used for virulence assays following the methods indicated in SI Appendix.

Mutagenesis and Phenotype Analysis.

A. baumannii ATCC 17978 was used as the parental strain to generate an in-frame deletion mutant of efp, following the methods indicated in SI Appendix. Motility and biofilm formation assays were performed using previously described methods, which are provided in SI Appendix.

Protein Purification and Analysis.

Detailed descriptions are provided in SI Appendix. Briefly, open reading frames were amplified with the primers listed in SI Appendix, Table S7 and cloned into the expression vector pET-28a. The fusion gene constructs were transformed into E. coli strain BL21. Affinity purification of fusion proteins and binding assays were performed using previously described methods, which are provided in SI Appendix.

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.