The Only Chemoreceptor Encoded by che Operon Affects the Chemotactic Response of Agrobacterium to Various Chemoeffectors

Chemoreceptor (also called methyl-accepting chemotaxis protein, MCP) is the leading signal protein in the chemotaxis signaling pathway. MCP senses and binds chemoeffectors, specifically, and transmits the sensed signal to downstream proteins of the chemotaxis signaling system. The genome of Agrobacterium fabrum (previously, tumefaciens) C58 predicts that a total of 20 genes can encode MCP, but only the MCP-encoding gene atu0514 is located inside the che operon. Hence, the identification of the exact function of atu0514-encoding chemoreceptor (here, named as MCP514) will be very important for us to understand more deeply the chemotaxis signal transduction mechanism of A. fabrum. The deletion of atu0514 significantly decreased the chemotactic migration of A. fabrum in a swim plate. The test of atu0514-deletion mutant (Δ514) chemotaxis toward single chemicals showed that the deficiency of MCP514 significantly weakened the chemotactic response of A. fabrum to four various chemicals, sucrose, valine, citric acid and acetosyringone (AS), but did not completely abolish the chemotactic response. MCP514 was localized at cell poles although it lacks a transmembrane (TM) region and is predicted to be a cytoplasmic chemoreceptor. The replacement of residue Phe328 showed that the helical structure in the hairpin subdomain of MCP514 is a direct determinant for the cellular localization of MCP514. Single respective replacements of key residues indicated that residues Asn336 and Val353 play a key role in maintaining the chemotactic function of MCP514.


Introduction
Chemotaxis is an adaptive behavior of motile bacteria moving along the concentration gradient of chemoeffectors towards an optimal environment [1,2]. This adaptive behavior is regulated by a two-component system composed of histidine kinase CheA and chemotaxis response regulator protein CheY [3,4]. When a chemical attractant exists in the environment, methyl-accepting chemotaxis protein (MCP) can bind the attractant, and the ligand binding will change the conformation of MCP; then, through a coupling protein CheW, the kinase activity of CheA will be suppressed. The suppression of CheA activity will delay the transfer of the phosphoryl group to the response regulator protein CheY, resulting in the decrease in phosphorylated CheY [5,6]. The absence of phosphorylated CheY makes less change of the rotational direction of flagella and thus keeps bacteria swimming to the attractant. On the contrary, when the concentration of attractant goes down (or repellent goes up), MCP activates the kinase activity of CheA. Self-phosphorylated CheA passes the phosphoryl groups to CheY. Phosphorylated CheY will bind to the flagellar motor and change the rotational direction of flagella frequently, which causes cell tumbling and changes direction away from the adverse environment [7,8].
Many bacteria possess chemotactic behavior for a large number of chemicals, such as different nitrogen or carbon nutrient substances, different environmental pollutants, or different signal chemicals released by their ecological partners [9][10][11][12][13]. To sense the vast ganisms 2021, 9, x FOR PEER REVIEW 3 of 14 chemotaxis signal transduction mechanism of A. fabrum. In this study, we firstly constructed the atu0514-deletion mutant Δ514 and the complemented strain Δ514-C and tested the effect of atu0514-deficiency on the chemotactic response of A. fabrum C58. We also identified the key residues that affect the cellular localization and function of MCP514.

Primers, Plasmids, Bacterial Strains and Growth Conditions
The primers, plasmids and bacterial strains used in this study are listed in Supplementary Tables S1 and S2. Lysogeny broth (LB) liquid or agar medium was used to grow E. coli at 37 °C [35]. A. fabrum was grown in MG/L or AB-sucrose liquid or agar medium at 28 °C [36,37]. Concentrations of ampicillin and kanamycin used for E. coli were 100 and 50 µg/mL, respectively. Concentrations of kanamycin and carbenicillin for A. fabrum were 100 µg/mL.

Primers, Plasmids, Bacterial Strains and Growth Conditions
The primers, plasmids and bacterial strains used in this study are listed in Supplementary Tables S1 and S2. Lysogeny broth (LB) liquid or agar medium was used to grow E. coli at 37 • C [35]. A. fabrum was grown in MG/L or AB-sucrose liquid or agar medium at 28 • C [36,37]. Concentrations of ampicillin and kanamycin used for E. coli were 100 and 50 µg/mL, respectively. Concentrations of kanamycin and carbenicillin for A. fabrum were 100 µg/mL.

DNA Manipulations
DNA manipulations followed the standard molecular protocols [35]. Plasmid isolation was performed with the TIANprep Mini Plasmids Kit (TIANGEN BIOTECH Corporation, Beijing, China). PCR products obtained by Veriti 96-well cycler (Thermo Fisher Scientific Inc., Waltham, MA USA) and DNA fragments were purified from agarose gels by using the TaKaRa MiniBEST Agarose Gel DNA Extraction Kit (TaKaRa Corporation, Dalian, China). Plasmids were transferred into E. coli competent cells by heat-shock [35] and into A. fabrum by the Eppendorf electroporation instrument Eporator ® (Eppendorf AG, Hamburg, Germany) [36].

Mutagenesis and Complementation of Atu0514
Based on the principle of homologous recombination, we used the pEX18Km-derived gene replacement plasmids to construct the corresponding gene deletion mutants [38,39]. Plasmid pEX18Km carries both a positive selection marker (kanamycin resistance, K m R ) and counterselectable marker (suicide gene sacB) and cannot be replicated in A. fabrum. The positive selection marker allowed to select the transformants, in which the whole plasmid was integrated into the genome by the first homologous recombination. The counterselectable marker allowed to counterselect the transformants, in which both the target DNA fragment and the undesirable plasmid backbone were deleted from the genome by the second homologous recombination. The combined utilization of selectable and counterselectable markers can precisely generate an unmarked mutant without any undesirable DNA fragment ( Figure S1A). A. fabrum wild type C58 was used to construct the atu0514deletion mutant ∆514. A. fabrum cheW 1 -cheW 2 double-deletion mutant ∆w [40] was used to construct atu0514-cheW 1 -cheW 2 triple-deletion mutant ∆514∆w. The desirable mutant was screened using PCR ( Figure S1B) and verified by sequencing. The DNA fragment encoding amino acids 14-453 of MCP 514 was precisely deleted in both ∆514 and ∆514∆w mutants ( Figure S1C). The complementation of MCP 514 in the atu0514-deletion mutant was fulfilled by the introduction of plasmid expressing MCP 514 (or its variants). Gene fusions with the egfp as well as deletion constructs were created by overlap extension PCR, as described by Higuchi [41]. These egfp-fused genes were cloned into the modified vector pUCA19 to generate the plasmids expressing eGFP-fused proteins. When these eGFP-fused proteins were expressed in A. fabrum cells, the cellular localizations of these eGFP-fused proteins could be observed by using Zeiss confocal microscope LSM 880 NLO (Carl Zeiss AG, Oberkochen, Germany).

Chemotaxis Assays
The procedure of capillary assay was followed as described by Adler in 1973 [42], with minor modifications. A. fabrum cells were harvested from mid-log-phase culture by centrifugation at 4000 rpm for 3 min at room temperature (25 • C) and suspended in chemotaxis buffer (0.1 mmol/L EDTA, 10 mmol/L KH 2 PO 4 , pH 7.0) to an OD 600nm of 0.1. A. fabrum cell suspension (300 µL) was used to make a bacterial pond. The capillary tube was sealed at one end and filled with attractant at peak concentration dissolved in the chemotaxis buffer. The open end of the capillary tube was inserted into the bacterial pond and incubated for 1 h at room temperature (25 • C), then, the solution in the capillary tube was expelled and completely transferred into 1 mL of AB-sucrose medium. Dilutions were plated in duplicate on MG/L plates and incubated for 2 days at 28 • C. The colonies in the plates were counted and represented the number of cells attracted to the capillary tube.
The procedure of swim agar plate assay was followed that described by Merritt [43], with minor modifications. The tested strains were inoculated in AB-sucrose liquid medium and grown to the middle log phase, and then, the OD 600nm was adjusted to 0.6 by using AB-sucrose liquid medium. A total of 3 µL of bacterial suspension was dropped onto an AB-sucrose swim plate containing 0.2% agar, and 5 replicates were set. After incubation at 28 • C for 36-48 h, the diameter of the bacterial colony circle was measured, and the data were statistically analyzed.

Bacterial Two-Hybrid Analyses
The bacterial two-hybrid system from the Stratagene ® (Agilent Technologies Inc., Santa Clara, CA, USA) was used for testing protein-protein interactions. All the operations were conducted according to the manual. The open read frame (ORF) of atu0514 was inserted into the bait plasmid pBT to express λcI-MCP 514 (bait) fusion protein. ORFs of cheW 1 and cheW 2 were inserted to the target plasmid pTRG to express CheW 1 (target)-RNAP and CheW 2 (target)-RNAP fusion proteins, respectively. The interaction between bait (MCP 514 ) and target (CheW 1 or CheW 2 ) will take λcI and RNAP together to induce the expression of β-galactosidase, and thus, the bacterial colonies growing on plates containing 80 µg/mL X-gal will be blue. Otherwise, the bacterial colonies will be of normal color. Galactosidase activity was determined by the method of Miller [44].

Fluorescence Microscopy
For microscopy observation, agrobacterial cells from the mid-log-phase cultures were added to the center of the slides. A coverslip was placed on top of the culture droplet. The edges of the coverslip were sealed with acrylic polymer to prevent drying. A. fabrum cells were visualized by a Zeiss LSM 880 NLO system (Carl Zeiss AG, Oberkochen, Germany) using an Ar laser (excitation wavelength of 488 nm and emission wavelength of 500 to 550 nm) and a ×100 oil immersion objective. The images were analyzed and edited using ZEN lite (Blue edition), version 3.2 (Carl Zeiss AG, Oberkochen, Germany).

Statistical Analysis
The quantitative data shown in this study were the means with the standard deviations (SDs), which were derived from at least three independent experiments conducted in triplicate. Differences among bacterial strains were compared using one-way analysis of variance (ANOVA), followed by the Tukey test for multiple comparisons. The statistical analysis was conducted using Microsoft Office Excel's data analysis tool (2019 version) (Microsoft Corporation, Redmond, WA, USA).

MCP 514 Is a Cytoplasmic Chemoreceptor, but Localized at Cell Poles
By SMART analysis, MCP 514 has a total of 514 amino acids and carries three conserved domains, protoglobin domain, HAMP domain and cytoplasmic signal domain (SD), in the order from N-terminal to C-terminal ( Figure 1C). The protoglobin domain of MCP 514 shares 13.51% sequence identity with the LBD of HEMAT, a cytoplasmic chemoreceptor from B. subtilis [21], through the SWISS-Model homology search. According to Alexandre's heptapeptide classification, cytoplasmic signal domain (SD) belongs to the 36H family [45]. The two best-studied MCPs of the 36H family are Tsr and Tar proteins of E. coli. Further analysis on the secondary structure of MCP 514 by SOSUI [46] and SPLIT [47] shows that MCP 514 does not contain the hydrophobic domain, indicating that MCP 514 does not possess a transmembrane (TM) region and belongs to an IVa cytoplasmic chemoreceptor.
Transmembrane chemoreceptors are mainly localized at cell poles, but the cytoplasmic chemoreceptors have a wider cellular localization mode, ranging from co-localization with transmembrane chemoreceptor arrays to a diffuse cytoplasmic distribution [48]. The localization of some cytoplasmic chemoreceptors is associated with the physiology and life cycle of bacteria [22]. According to previous classification [9], MCP 514 should be classified into the category of the cytoplasmic chemoreceptor due to the lack of a transmembrane region, but we do not know the cellular localization of MCP 514 . To observe the cellular localization of MCP 514 , the enhanced green fluorescent protein (eGFP) was fused to the Nterminus of MCP 514 because only the N-terminally GFP-tagged MCP 514 was functional [49]. This eGFP-MCP 514 fusion protein was expressed in the MCP 514 -deficient strain ∆514. Figure 2A showed that the eGFP-MCP 514 fusion protein is localized at the poles of the A. fabrum cell. However, the eGFP-MCP 514 fusion protein is distributed in the whole cell of E. coli ( Figure 2B), verifying that MCP 514 lacks a transmembrane domain. These data also imply that the polar localization of MCP 514 in A. fabrum cell requires the assistance of other A. fabrum proteins.
the N-terminus of MCP514 because only the N-terminally GFP-tagged MCP514 was functional [49]. This eGFP-MCP514 fusion protein was expressed in the MCP514-deficient strain ∆514. Figure 2A showed that the eGFP-MCP514 fusion protein is localized at the poles of the A. fabrum cell. However, the eGFP-MCP514 fusion protein is distributed in the whole cell of E. coli ( Figure 2B), verifying that MCP514 lacks a transmembrane domain. These data also imply that the polar localization of MCP514 in A. fabrum cell requires the assistance of other A. fabrum proteins. Bacterial cells were grown to middle-log phase and observed by using confocal laser-scanning microscopy.

MCP514 Significantly Affects the Chemotactic Response of Agrobacterium Fabrum
Since MCP514 is localized at cell poles, we next try to test the effect of MCP514 on the chemotactic response of A. fabrum. When bacterium grows on the swim agar plate, the utilization of nutrient substances by bacterium will result in a nutrient concentration gradient. Bacterium with chemotaxis will move outward along the nutrient concentration gradient and grow a big colony. Consequently, the overall chemotactic response to nutrient substances can be characterized by measuring the colony size in the swim agar plate [10]. A. fabrum has only one CheA, and CheA is the key component of a chemotaxis system. The deletion of the cheA gene will completely abolish the chemotactic response of A. fabrum, and thus, the CheA-deficient mutant Δa was used as a control of chemotaxis deficiency [40]. As shown in Figure 3, the deficiency of MCP514 significantly attenuates the overall chemotactic response of A. fabrum to nutrient substances. The complementation of MCP514 by the introduction of MCP514-expressing plasmid can fully restore the chemotactic response of the MCP514-deficient mutant to the level of the wildtype, confirming the role of MCP514 in the chemotactic response. These results also provided evidence that the deletion of atu0514 did not have a polar effect on the rest of the che operon. Previous research showed that A. fabrum has chemotaxis toward sugars, amino acids, organic acids and phenols [29,30]. The traditional capillary assay is an effective method for quantifying the chemotaxis ability of bacteria [42]. The chemotactic responses of wildtype C58, mutant Δ514 and complemented strain Δ514-C to four various chemicals (1 µmol/L sucrose, 1 mmol/L valine, 1 mmol/L citric acid and 0.1 µmol/L acetosyringone (AS)) were measured by using the traditional capillary method. As shown in Figure 4, the deletion of atu0514 significantly weakens the chemotaxis of A. fabrum toward these four different types of chemicals. However, the chemotaxis of strain Δ514 toward these chemicals does not completely disappear, indicating that MCP514 is not the receptor directly recognizing these substances but affects the chemotaxis efficiency in other ways.

MCP 514 Significantly Affects the Chemotactic Response of Agrobacterium Fabrum
Since MCP 514 is localized at cell poles, we next try to test the effect of MCP 514 on the chemotactic response of A. fabrum. When bacterium grows on the swim agar plate, the utilization of nutrient substances by bacterium will result in a nutrient concentration gradient. Bacterium with chemotaxis will move outward along the nutrient concentration gradient and grow a big colony. Consequently, the overall chemotactic response to nutrient substances can be characterized by measuring the colony size in the swim agar plate [10].
A. fabrum has only one CheA, and CheA is the key component of a chemotaxis system. The deletion of the cheA gene will completely abolish the chemotactic response of A. fabrum, and thus, the CheA-deficient mutant ∆a was used as a control of chemotaxis deficiency [40]. As shown in Figure 3, the deficiency of MCP 514 significantly attenuates the overall chemotactic response of A. fabrum to nutrient substances. The complementation of MCP 514 by the introduction of MCP 514 -expressing plasmid can fully restore the chemotactic response of the MCP 514 -deficient mutant to the level of the wildtype, confirming the role of MCP 514 in the chemotactic response. These results also provided evidence that the deletion of atu0514 did not have a polar effect on the rest of the che operon. Previous research showed that A. fabrum has chemotaxis toward sugars, amino acids, organic acids and phenols [29,30]. The traditional capillary assay is an effective method for quantifying the chemotaxis ability of bacteria [42]. The chemotactic responses of wildtype C58, mutant ∆514 and complemented strain ∆514-C to four various chemicals (1 µmol/L sucrose, 1 mmol/L valine, 1 mmol/L citric acid and 0.1 µmol/L acetosyringone (AS)) were measured by using the traditional capillary method. As shown in Figure 4, the deletion of atu0514 significantly weakens the chemotaxis of A. fabrum toward these four different types of chemicals. However, the chemotaxis of strain ∆514 toward these chemicals does not completely disappear, indicating that MCP 514 is not the receptor directly recognizing these substances but affects the chemotaxis efficiency in other ways.

Both CheW 1 and CheW 2 Interact with MCP 514 but Do Not Affect the Cellular Localization of MCP 514
It is known that CheW couples CheA to chemoreceptors and forms stable ternary signaling complexes with chemoreceptors and CheA [3]. To confirm the role of MCP 514 in the chemotaxis signal transduction pathway of A. fabrum, the bacterial two-hybrid system was used to test the interaction between MCP 514 and two CheWs. As shown in Figure 5A, the colors of the colonies representing MCP 514 /CheW 1 and MCP 514 /CheW 2 interaction are bluer than that of the negative control. The quantification of β-galactosidase activity also showed that the β-galactosidase activities of the colonies expressing these two tested protein pairs (MCP 514 /CheW 1 and MCP 514 /CheW 2 ) are significantly higher than that of the negative control ( Figure 5B). This indicates that MCP 514 protein interacted with both CheW 1 and CheW 2 proteins.   The bars paired with "*", "**" and "***" marks represent that they are different in a statistical manner at p < 0.05, 0.01 and 0.001, respectively, via the one-way ANOVA, followed by Tukey test. The strain names in the horizontal axis are the same as in Figure 3.
the cells in the capillary tube with attractant minus the cells in the capillary tube without attractant. The data represent the means ± SDs from three independent experiments with triplicate. The bars paired with "*", "**" and "***" marks represent that they are different in a statistical manner at p < 0.05, 0.01 and 0.001, respectively, via the one-way ANOVA, followed by Tukey test. The strain names in the horizontal axis are the same as in Figure 3.

Both CheW1 and CheW2 Interact with MCP514 but do Not Affect the Cellular Localization of MCP514
It is known that CheW couples CheA to chemoreceptors and forms stable ternary signaling complexes with chemoreceptors and CheA [3]. To confirm the role of MCP514 in the chemotaxis signal transduction pathway of A. fabrum, the bacterial two-hybrid system was used to test the interaction between MCP514 and two CheWs. As shown in Figure 5A, the colors of the colonies representing MCP514/CheW1 and MCP514/CheW2 interaction are bluer than that of the negative control. The quantification of β-galactosidase activity also showed that the β-galactosidase activities of the colonies expressing these two tested protein pairs (MCP514/CheW1 and MCP514/CheW2) are significantly higher than that of the negative control ( Figure 5B). This indicates that MCP514 protein interacted with both CheW1 and CheW2 proteins. and CheW was tested by using a bacterial two-hybrid assay. All the operations followed the manual. The interaction between two tested proteins will induce the expression of β-galactosidase, and bacterial colonies will show blue color on the X-gal indicator plate. (A) Colony color on the X-gal indicator plate. (B) The activity of β-galactosidase reporter induced by the interacting proteins. Data are the means ± SDs from three independent experiments with triplicate. The bars paired with "*" and "**" marks represent that they are different in a statistical manner at p < 0.05 and 0.01, respectively via the one-way ANOVA, followed by Tukey test. Positive control, interaction between two known proteins (LGF2 and Gal11 p ) provided by the manufacturer; MCP514/CheW1, interaction between MCP514 and CheW1; MCP514/CheW2, interaction between MCP514 and CheW2; negative control, without any interacting proteins.
Both CheW1 and CheW2 interact with MCP514 and most of the ternary MCP-CheW-CheA complexes are localized in the cell poles. However, MCP514 lacks a transmembrane domain. It is unknown whether the cellular localization of MCP514 is dependent on the ternary MCP-CheW-CheA complex. Therefore, we tested the effects of the CheW deficiency on the cellular localization of MCP514. To determine whether CheW affects the cellular localization of MCP514, a plasmid expressing eGFP-MCP514 fusion protein was transferred into the atu0514-cheW1-cheW2 triple deletion mutant Δ514Δw. Fluorescence observation showed that MCP514 in the CheW-deficient strain is still localized at cell poles, indicating that CheW deficiency does not affect the polar localization of MCP514 ( Figure 6). and CheW was tested by using a bacterial two-hybrid assay. All the operations followed the manual. The interaction between two tested proteins will induce the expression of β-galactosidase, and bacterial colonies will show blue color on the X-gal indicator plate. (A) Colony color on the X-gal indicator plate. (B) The activity of β-galactosidase reporter induced by the interacting proteins. Data are the means ± SDs from three independent experiments with triplicate. The bars paired with "*" and "**" marks represent that they are different in a statistical manner at p < 0.05 and 0.01, respectively via the one-way ANOVA, followed by Tukey test. Positive control, interaction between two known proteins (LGF2 and Gal11 p ) provided by the manufacturer; MCP 514 /CheW 1 , interaction between MCP 514 and CheW 1 ; MCP 514 /CheW 2 , interaction between MCP 514 and CheW 2 ; negative control, without any interacting proteins.
Both CheW 1 and CheW 2 interact with MCP 514 and most of the ternary MCP-CheW-CheA complexes are localized in the cell poles. However, MCP 514 lacks a transmembrane domain. It is unknown whether the cellular localization of MCP 514 is dependent on the ternary MCP-CheW-CheA complex. Therefore, we tested the effects of the CheW deficiency on the cellular localization of MCP 514 . To determine whether CheW affects the cellular localization of MCP 514 , a plasmid expressing eGFP-MCP 514 fusion protein was transferred into the atu0514-cheW 1 -cheW 2 triple deletion mutant ∆514∆w. Fluorescence observation showed that MCP 514 in the CheW-deficient strain is still localized at cell poles, indicating that CheW deficiency does not affect the polar localization of MCP 514 (Figure 6).

Helical Structure of Hairpin Subdomain Is Required for the Cellular Localization of MCP 514
Due to the lack of a transmembrane domain in MCP 514 and the evidence that CheW deficiency does not affect the polar localization of MCP 514 , it is most likely that MCP 514 is localized in the cell poles via interacting with other MCPs. Results from E. coli MCPs showed that the hairpin subdomain of MCP is a coiled-coil of two antiparallel helices with a 'U-turn' and two hairpin subdomains form a supercoiled four-helical bundle, which makes MCP form homodimeric molecules [50]. The dimers of different MCPs can form mixed trimers of dimers via the interactions between their highly conserved helical bundle, and several residues play key roles in the formation of trimer ( Figure S1) [51][52][53].

Helical Structure of Hairpin Subdomain is Required for the Cellular Localization of MCP514
Due to the lack of a transmembrane domain in MCP514 and the evidence that CheW deficiency does not affect the polar localization of MCP514, it is most likely that MCP514 is localized in the cell poles via interacting with other MCPs. Results from E. coli MCPs showed that the hairpin subdomain of MCP is a coiled-coil of two antiparallel helices with a 'U-turn' and two hairpin subdomains form a supercoiled four-helical bundle, which makes MCP form homodimeric molecules [50]. The dimers of different MCPs can form mixed trimers of dimers via the interactions between their highly conserved helical bundle, and several residues play key roles in the formation of trimer ( Figure S1) [51][52][53].
To determine if the hairpin subdomain of MCP514 affects the cellular localization of MCP514, key residue Phe328 in the hairpin subdomain was changed to Ala, Pro or Trp. These three single-residue-substituted MCP514 variants were fused to eGFP, respectively. Fluorescence observation showed that the substitution of Phe328 for Pro causes the MCP514 diffusion in the cytoplasm ( Figure 7C), whereas the replacement of Phe328 by Ala or Trp does not affect the polar location of MCP514 ( Figure 7A,B). Proline is a constraint on the formation of helix. Replacement of Phe328 by Pro will destroy the helical structure of the hairpin subdomain. These results demonstrated that the helical structure of the hairpin subdomain is required for the cellular localization of MCP514.  To determine if the hairpin subdomain of MCP 514 affects the cellular localization of MCP 514 , key residue Phe328 in the hairpin subdomain was changed to Ala, Pro or Trp. These three single-residue-substituted MCP 514 variants were fused to eGFP, respectively. Fluorescence observation showed that the substitution of Phe328 for Pro causes the MCP 514 diffusion in the cytoplasm ( Figure 7C), whereas the replacement of Phe328 by Ala or Trp does not affect the polar location of MCP 514 ( Figure 7A,B). Proline is a constraint on the formation of helix. Replacement of Phe328 by Pro will destroy the helical structure of the hairpin subdomain. These results demonstrated that the helical structure of the hairpin subdomain is required for the cellular localization of MCP 514 . Figure 6. Cellular localization of MCP514 in CheW-deficient A. fabrum mutant. Plasmid expressing eGFP-MCP514 fusion protein was introduced into the atu051-cheW1-cheW2 triple-deletion mutant Δ514Δw. Cells were grown to the middle-log phase and observed by using confocal laser-scanning microscopy.

Helical Structure of Hairpin Subdomain is Required for the Cellular Localization of MCP514
Due to the lack of a transmembrane domain in MCP514 and the evidence that CheW deficiency does not affect the polar localization of MCP514, it is most likely that MCP514 is localized in the cell poles via interacting with other MCPs. Results from E. coli MCPs showed that the hairpin subdomain of MCP is a coiled-coil of two antiparallel helices with a 'U-turn' and two hairpin subdomains form a supercoiled four-helical bundle, which makes MCP form homodimeric molecules [50]. The dimers of different MCPs can form mixed trimers of dimers via the interactions between their highly conserved helical bundle, and several residues play key roles in the formation of trimer ( Figure S1) [51][52][53].
To determine if the hairpin subdomain of MCP514 affects the cellular localization of MCP514, key residue Phe328 in the hairpin subdomain was changed to Ala, Pro or Trp. These three single-residue-substituted MCP514 variants were fused to eGFP, respectively. Fluorescence observation showed that the substitution of Phe328 for Pro causes the MCP514 diffusion in the cytoplasm ( Figure 7C), whereas the replacement of Phe328 by Ala or Trp does not affect the polar location of MCP514 ( Figure 7A,B). Proline is a constraint on the formation of helix. Replacement of Phe328 by Pro will destroy the helical structure of the hairpin subdomain. These results demonstrated that the helical structure of the hairpin subdomain is required for the cellular localization of MCP514.

Two Key Residues of Hairpin Subdomain Play a Key Role in Maintaining the Chemotactic Function of MCP 514
The cellular localization of MCP 514 is dependent on the hairpin subdomain, which is involved in the interactions between different MCPs, as well as the interactions with the CheA and CheW [54]. The chemotactic signal is collaboratively transduced by the mixed MCP teams, and all signals from different MCPs will converge to CheA [51]. It is reasonable that residues involving in the interaction between MCPs may affect the chemotactic function of MCP 514 .
To further determine the key residues of MCP 514 that are involved in the trimer contact, we aligned the sequence of MCP 514 with the sequences of Tsr and Tar from E. coli [55] and chose Phe328, Asn336, Glu340, Arg343 and Val353 of MCP 514 as the target residues of the site-directed mutation ( Figure S2). Five residues were respectively replaced by alanine to generate five single residue-substituted MCP 514 variants. These MCP 514 variants were expressed in the MCP 514 -deficient mutant (∆514) by the introduction of the MCP 514 variantexpressing plasmid, respectively. Colonies of the MCP 514 -deficient mutant expressing various MCP 514 variants are shown in Figure 8A. The diameters of these tested strain colonies were used to quantify the effects of various MCP 514 variants on the chemotactic response of A. fabrum [43]. As shown in Figure 8, two single residue-substituted MCP 514 variants, MCP 514 N336A and MCP 514 V353A , are unable to restore the chemotactic response of the MCP 514 -deficient mutant to the level of the wildtype, demonstrating that residues Asn336 and Val353 play a key role in maintaining the chemotactic function of MCP 514 .
CheA and CheW [54]. The chemotactic signal is collaboratively transduced by the mixed MCP teams, and all signals from different MCPs will converge to CheA [51]. It is reasonable that residues involving in the interaction between MCPs may affect the chemotactic function of MCP514.
To further determine the key residues of MCP514 that are involved in the trimer contact, we aligned the sequence of MCP514 with the sequences of Tsr and Tar from E. coli [55] and chose Phe328, Asn336, Glu340, Arg343 and Val353 of MCP514 as the target residues of the site-directed mutation ( Figure S2). Five residues were respectively replaced by alanine to generate five single residue-substituted MCP514 variants. These MCP514 variants were expressed in the MCP514-deficient mutant (Δ514) by the introduction of the MCP514 variant-expressing plasmid, respectively. Colonies of the MCP514-deficient mutant expressing various MCP514 variants are shown in Figure 8A. The diameters of these tested strain colonies were used to quantify the effects of various MCP514 variants on the chemotactic response of A. fabrum [43]. As shown in Figure 8, two single residue-substituted MCP514 variants, MCP514 N336A and MCP514 V353A , are unable to restore the chemotactic response of the MCP514-deficient mutant to the level of the wildtype, demonstrating that residues Asn336 and Val353 play a key role in maintaining the chemotactic function of MCP514. The swim-ring diameters of these tested A. fabrum strains on the swim plate. The data represent the means ± SDs from five independent experiments in triplicate. The bars paired with "*", "**", "***" and "****" marks represent that they are different in a statistical manner at p < 0.05, 0.01, 0.001 and 0.0001, respectively, via the one-way ANOVA, followed by Tukey test. C58, A. fabrum wildtype C58 strain; Δ514, MCP514 deficient mutant; Δ514-C, Δ514 mutant complemented with native MCP514; MCP514 F328A , MCP514 N336A , MCP514 E340A , MCP514 R343A and MCP514 V353A represent Δ514 mutant complemented with the corresponding single residue-substituted MCP514 variants, respectively; Δa, CheA deficient mutant.

Discussion
In natural environments, A. fabrum is usually distributed around the rhizosphere of a plant [24]. Chemotaxis is the important initial step for A. fabrum to infect the plant host [56]. Only when A. fabrum correctly recognizes and responds to the chemical signals released by the plant host can it contact the host plant, infect the host and start the tumorigenic processes [27,[57][58][59]. MCPs are the first components of the chemotaxis system. The recognition of chemicals by these proteins is the initial stage of chemotaxis signaling transduction. Adaptive modification of the conserved glutamate domain of the MCP signal domain ensures that they are highly sensitive to different concentrations of chemoeffec- Figure 8. Effects of MCP 514 variants on the chemotactic response of A. fabrum. The test procedure was the same as described in Figure 3. (A) Typical colonies of these tested A. fabrum strains. (B) The swim-ring diameters of these tested A. fabrum strains on the swim plate. The data represent the means ± SDs from five independent experiments in triplicate. The bars paired with "*", "**", "***" and "****" marks represent that they are different in a statistical manner at p < 0.05, 0.01, 0.001 and 0.0001, respectively, via the one-way ANOVA, followed by Tukey test. C58

Discussion
In natural environments, A. fabrum is usually distributed around the rhizosphere of a plant [24]. Chemotaxis is the important initial step for A. fabrum to infect the plant host [56]. Only when A. fabrum correctly recognizes and responds to the chemical signals released by the plant host can it contact the host plant, infect the host and start the tumorigenic processes [27,[57][58][59]. MCPs are the first components of the chemotaxis system. The recognition of chemicals by these proteins is the initial stage of chemotaxis signaling transduction. Adaptive modification of the conserved glutamate domain of the MCP signal domain ensures that they are highly sensitive to different concentrations of chemoeffectors [60]. Although A. fabrum C58 contains 20 MCP-encoding genes, only one MCP-encoding gene (atu0514) is in the che operon [12]. It is rational that the MCP 514 encoded by the atu0514 gene may play a unique role in the signaling transduction of chemotaxis.
Our results show that MCP 514 deficiency significantly affects not only the overall chemotactic response of A. fabrum to nutrient substances ( Figure 3) but also the chemotaxis toward four various types of chemicals ( Figure 4). The ligand binding domain (LBD) of MCP 514 is predicted to be a protoglobin, which cannot bind these four tested types of chemicals ( Figure 1B). MCPs with different LBDs can form mixed trimers of dimers, and thousands of MCPs form a receptor cluster. The receptor cluster comprised of MCPs with different detection specificities collaboratively transduces the chemotactic signal in a team signaling model [51]. The effects of MCP 514 deficiency on the chemotaxis of A. fabrum toward four various types of chemicals demonstrate that MCP 514 is a very important member of the chemoreceptor signaling team, and the absence of MCP 514 will affect the signaling efficiency of the whole chemoreceptor signaling team, explicating the reason of the atu0514 gene locating the che operon.
Based on the sequenced genomes, 43% of archaeal and 14% of bacterial MCPs lack a transmembrane (TM) region. These TM-lacking MCPs are classified as cytoplasmic chemoreceptors. Unlike the transmembrane chemoreceptors, which are mainly located at the pole of the cell, cytoplasmic MCPs adopt more exotic locations [22]. Some cytoplasmic chemoreceptors are polarized at one end of the cell, such as, the HemAt of B. subtilis and the IcpA of S. meliloti [21,61]; other cytoplasmic chemoreceptors have both polar and diffuse states, for example, the AerC of A. brasilense was diffused in cytoplasm under an oxygen-rich environment but located in the cell polar under symbiotic nitrogen-fixing state, which helps cells adapt to hypoxic environments [19]. There are also some bacteria whose cytoplasmic chemoreceptors formed clusters in the cytoplasm, for example, the TlpC and TlpT of R. sphaeroides are located in the center of the cell by forming cytoplasmic clusters [62]. It is believed that the subcellular localization and distribution of the cytoplasmic chemoreceptor arrays are associated with the life cycle of bacteria [22]. Our results show that MCP 514 localizes cell poles. Although both CheW 1 and CheW 2 interact with MCP 514 , respectively, ( Figure 5), the cellular localization of MCP 514 is independent of CheW ( Figure 6), which is consistent with the previous results obtained from E. coli, in which trimers of homodimers of MCPs form clusters of MCPs and in turn recruit CheA/CheW to assemble MCP-CheW-CheA ternary complexes [52,54]. In combination with the previous studies on other cytoplasmic MCPs [22,62], our results could support that this TM-lacking chemoreceptor, MCP 514 , localizes the cell poles through interacting with other transmembrane chemoreceptors, although the experimental evidence may be required to demonstrate the interaction of MCP 514 with other MCPs.
According to the previous studies on Tsr and Tar of E. coli, the hairpin (HP) subdomain of MCP is not only the region forming homodimeric MCP but also the contacting sites for the formation of mixed trimers as well as the assembly of MCP-CheW-CheA ternary complexes [51,55]. Five conserved amino acid residues in the two antiparallel helices of E. coli Tsr (Phe373, Asn381, Glu385, Arg388 and Val353) are important for the cluster formation and function of MCP [51]. Hydrophobic interactions between the helices contribute to the main trimer packing forces [55]. Five corresponding residues of MCP 514 conducted the single residue of the respective substitution. Phe328 of MCP 514 is in the key site of the helical structure of the HP subdomain. The replacement of Phe328 by the helix destroyer, proline, results in MCP 514 diffusing in the cytoplasm. Replacement of Phe328 by alanine does not affect both the cellular location and chemotactic function of MCP 514 . These results demonstrate that Phe328 is not the direct determinant to the chemotactic function of MCP 514 , but indirectly affects MCP 514 function through stabilizing the helical structure. Amongst the five tested MCP 514 variants, three variants have the full function of the wildtype, which is slightly different from the results of E. coli Tsr [51].

Conclusions
MCP 514 is localized at cell poles, although it lacks a transmembrane region. The cellular localization of MCP 514 is independent of the assembly of MCP-CheW-CheA ternary complex but dependent on the interaction with other MCPs through the hairpin (HP) subdomain. Compared with other MCPs, MCP 514 plays a superior role in maintaining the signaling efficiency of the whole chemoreceptor signaling team. The helical structure of the hairpin subdomain is the prerequisite of MCP 514 cellular localization and functioning. The molecular mechanism of MCP 514 interacting with other MCPs and functioning in the chemotaxis signaling is similar to that of E. coli MCPs, although the roles of some key residues of MCP 514 in the transducing signal are slightly different from those of E. coli Tsr.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/microorganisms9091923/s1, Figure S1: Construction procedure, screening and verification of atu0514-deficient mutants. Figure S2: Sequence alignment of three A. fabrum MCPs and two E. coli MCPs. Table S1: Bacterial strains and plasmids used in this study. Table S2

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.