The genetic and molecular basis of flagellar motility has been investigated for several decades, with innovative research strategies propelling advances at a steady pace. Furthermore, as the phenomenon is examined in diverse bacteria, new taxon-specific regulatory and structural features are being elucidated. Motility is also a straightforward bacterial phenotype that can allow undergraduate researchers to explore the palette of molecular genetic tools available to microbiologists. This study, driven primarily by undergraduate researchers, evaluated hundreds of flagellar motility mutants in the Gram-negative plant-associated bacterium Agrobacterium fabrum. The nearly saturating screen implicates a total of 37 genes in flagellar biosynthesis, including genes of previously unknown function.
Flagellar motility is widespread in Gram-positive and Gram-negative bacteria, with motility systems in the Gram-negative enteric species Escherichia coli and Salmonella enterica being particularly well characterized. The gram-negative flagellum consists of an envelope-integrated basal body that drives rotary motion of an external hook-filament complex. The basal body includes a hollow rod connected to a series of rings embedded in each of the three envelope layers (inner membrane, peptidoglycan, and outer membrane). The rotary motion of the rod is driven by the flagellar motor, which consists of the rotor component along with stator modules that are mechanically stabilized by their association with the peptidoglycan wall. Flagellar motion is powered by the movement of protons through the stator modules and this movement is translated to locomotion by connection of the flagellar rod to a flexible extracellular hook that in turn connects to a propulsive filament. The spinning filament generates thrust by either clockwise or counterclockwise rotation (depending on the organism), and directionality of cell movement is ultimately controlled by a discontinuous pattern of "runs" (propulsive activity) and "tumbles" (reorientation by pausing or reversing flagellar rotation). The switching between runs and tumbles is directed by a chemosensory network consisting of chemoreceptors and associated "Che" proteins that transduce signals from attractants and repellents to the flagellar apparatus [[
Flagellar assembly in all bacteria is thought to be controlled by a cascade of regulatory events consisting of master transcriptional regulators and sensors of flagellar assembly status. For example, in the enteric bacteria, the FlhD4C2 heterohexamer [[
Flagellar Motility in the Rhizobiaceae features several characteristics not encountered in the enteric model organisms. First, Rhizobiaceae tend to encode multiple flagellin proteins. In A. fabrum, four flagellin-encoding genes (flaA, flaB, flaC, and flaD) are found. Of these, only flaA is required for motility, while at least one of the remaining flagellin genes must be co-expressed with flaA for normal motility [[
Here we report a large-scale transposon mutagenesis of A. fabrum and subsequent screen for motility-defective mutants. Nearly all of the responsible insertions fall within the motility cluster and highlight genes responsible for flagellar regulation and assembly, including several genes less well characterized or not previously assigned such functions. This work was carried out largely as a class project for a Brigham Young University microbial genetics course (MMBIO 360) involving 27 undergraduate students and two graduate teaching assistants.
Strains used in this study are listed in S1 Table. The plasmid-cured A. fabrum strain UBAPF2 [[
Bacterial conjugation by triparental mating was carried out by first growing the six initial Sm-resistant A. fabrum strains (the recipients), and donor and helper E. coli strains (for plasmids and strains details see S1 and S2 Tables) separately as patches on LB-agar with appropriate antibiotics. Cells were collected with toothpicks and suspended in liquid LB to equivalent levels of turbidity. Six matings (one for each recipient) were set up by combining 70 μl of each suspension, plating cell mixtures on LB-agar, and allowing overnight incubation. Resulting lawns were collected by suspending in LB containing 15% glycerol. Aliquots were stored at -80°C. Transposants were selected by plating mating suspensions on LB-agar containing Sm and Nm. From each of the six matings, approximately 2 x 10
For enrichment of non-motile A. fabrum mutants from each transposon library, 1 μl of suspension (containing approximately 1 x 10
In the first round of Arbitrary (Arb) PCR, 1.5 μl of lysed, boiled cells from each strain was added to 15.4 μl of water, 2 μl of Taq buffer, 0.5 μl 10 mM dNTP, 0.15 μl Taq, 0.15 μl of 100 μM of forward primer (2100), and 0.3 μl of 100 μM reverse primer (2102 or 2103). PCR was carried out under the following conditions: after initial denaturation (94°C for 1 min), cycling 6 times (94°C for 15 sec, 33°C for 45 sec, 70°C for 45 sec), and cycling 30 times (94°C for 15 sec, 43°C for 30 sec, 70°C for 45 sec). In the second round of Arb-PCR, 0.9 μl of the amplified DNA from the first round of PCR was added to 16.1 μl of water, 2 μl of Taq buffer, 0.5 μl of 10 mM dNTP, 0.15 μl Taq, 0.15 μl of 100 μM forward primer (2101) and 0.15 μl of 100 μM reverse primer 2104. The second round of PCR was carried out under the following conditions: after initial denaturation (94°C for 1 min), cycling 30 times (94°C for 15 sec, 55°C for 15 sec, 70°C for 45 sec). Sanger Sequencing was carried out on the amplified PCR products to identify transposon insertion sites. The first-round arbitrary primer 2102 was normally used, but in cases of unacceptable product or low-quality sequence for a given mutant, the alternative primer 2103 was used (see S3 Table for primer sequences).
From the data obtained from Sanger Sequencing, the ...GAGACAG sequence at the end of the mini-transposon was located and the 30 nucleotides following this were used to find the position in the A. fabrum genome, using BLASTN (in GenBank accession AE007869.2) [[
Plasmids that contained homology regions for each target gene were created. Homology regions were designed to contain 300bp of DNA on each side of the gene of interest, maintaining the first and last several codons for that gene. Sequences for each target gene were retrieved from GenBank accession number AE007869.2. Parent plasmid pJG1108 containing XbaI-SalI-gus-sacB-kanR was digested with XbaI and SalI. Inserts were amplified from BB01 DNA using primers listed in S3 Table and prepared for 3-way ligations in which the joint between right and left homology regions is the 6-base sequence CCCGGG (XmaI, encoding Pro-Gly). Ligation products were transformed into E. coli DH5α and sequence verified. Deletion plasmids were transferred into A. fabrum strain BB01 by triparental mating, as described above for transposon mutagenesis and transconjugant clones were selected on LB-SmNm. Four individual colonies from each mating were carried over to selection on LB containing sucrose and X-Gluc (5-Bromo-4-chloro-3-indoxyl-beta-D-glucuronide cyclohexylammonium salt) (100 μg/mL). After 72 hours, cells from white colonies were suspended in PCR lysis buffer (5mM Tris pH 8.0, 2mM EDTA, 0.5% Triton X-100) and heated to 95°C for 5 minutes. Confirmatory PCR was carried out with primers listed in S3 Table. Successful deletion was verified by band down-shift on an agarose gel. Motility tests for deletion strains were carried out as described above.
For complementing mutants ΔATU0568, ΔATU0583, ΔflgN, and ΔmotF in A. fabrum, plasmid pKJ056 was used (S1A Fig). This plasmid allows expression of downstream genes by read-through transcription of the kanamycin-resistance (kanR) gene. All of the genes to be complemented were amplified from the A. fabrum C58 genome with their respective primers (S3 Table). pKJ056 was digested with EcoRI and BamHI as were all the amplified products and then ligated together. Ligated plasmids were transformed into E. coli DH5α and sequence verified. For visNR complementation experiments, three replicative plasmids were created that contain visN, visR, or visNR under the control of the P
A screen for motility mutants in A. fabrum was carried out by first mutagenizing Sm-resistant derivatives of the plasmid-cured strain UBAPF2 [[
Libraries enriched for non-motile mutants were plated to single colonies, and these were screened colony-by-colony for motility defects. Only strongly non-motile mutants were carried forward in our analysis. With the enrichment strategy described above, over 25% of colonies assayed were strongly defective in motility (S2 Fig). Around 500 such mutants were streaked to isolation and retested for motility. Of these, 360 confirmed mutants were used to map transposon insertion sites by arbitrarily primed PCR. This analysis revealed 314 unique insertions possibly associated with the motility defect. Most of these (301 insertions) were confined to a known cluster of flagellar motility genes, comprising nucleotides (
Graph: Fig 1 Annotated motility cluster in A. fabrum.Shown in gray are genes that were not disrupted in the screen. Shown in white are genes that were disrupted in the screen and are of strongly suspected function. Shown in black are genes that were disrupted in the screen and were subjected to further analysis. White dots on top of the genes show that a transposon landed in rightward orientation (referring to kanR transcription), while black dots indicate that a transposon landed in leftward orientation. A 16-kb region not shown in the figure includes genes that are not predicted to be involved in flagellar assembly, and no genes in this region were disrupted in the screen.
Genes hit in our screen represent 37 proteins: 33 with known or suspected functions. At the time this screen was carried out, four of these proteins were of unknown function. Most of these proteins map to specific components of a model Gram-negative bacterial flagellum, as depicted in Fig 2. These components include the flagellar secretion apparatus (FliP, FliI, FlhB, FliQ, FlhA, FliR), the cytoplasmically localized C ring (FliM, FliN, FliG), the inner membrane-localized MS ring (FliF), the proximal rod junction (FliE), the proximal rod (FlgB, FlgC, FlgF), the P ring (FlgI), the L ring (FlgH), the distal rod (FlgG), core stator proteins (MotA, MotB), stator-associated proteins (MotC, MotE, FliL), hook (FlgE), hook-filament junction (FlgK, FlgL), and filament (FlaA). Other genes disrupted in the screen encode proteins that may play transient roles in directing flagellar assembly, including the rod capping protein (FlgJ), the hook capping protein (FlgD), the hook length regulator (FliK), and the P ring assembly chaperone (FlgA). Disrupted genes with transcriptional or translational regulatory functions include the Class IB transcriptional regulator Rem, and the functionally coupled translational regulators FlbT and FlaF.
DIAGRAM: Fig 2 The putative structure of the A. fabrum flagellum and the proteins from which it is constituted.The diagram includes the orientation of the flagellum with respect to major cell envelope components: Outer membrane (OM), peptidoglycan (PG), and the inner membrane (IM). Also labeled are the main functional units of the flagellum and the specific proteins from which they are made, all of which were implicated in this genetic screen. Proteins with regulatory or unknown functions are listed in the box.
At the time this screen and analysis were carried out, four genes were of unknown function, and they were subjected to further analysis. These genes were designated ATU0568, ATU0583, ATU0585, and ATU8132 (according to the naming system in GenBank accession AE007869.2, and see Fig 1). A BLASTP search revealed that ATU0568 contains a DUF4231 domain from the SLATT superfamily, which contains a pair of N-terminal transmembrane helices and a helical C-terminal cytoplasmic region [[
A recent study by Sobe et al. [[
Graph: Fig 3 Synteny of four gene cluster required for motility in Sinorhizobium meliloti and Agrobacterium fabrum.ATU0585 was renamed flgN and ATU8132 was renamed motF.
The four focal genes discussed above (ATU0568, ATU0583, flgN, motF) were deleted in a manner intended to eliminate possible polar effects. This was done by removing most of the gene but retaining the first and last several codons of each coding sequence. We refer to these as "in-frame deletion" strains. All four of these strains exhibited motility defects similar to those of the rest of the motility mutants identified in the transposon screen. These deletion strains could be restored to normal motility by complementation plasmids constitutively expressing the corresponding genes (Fig 4).
Graph: Fig 4 In-frame deletions and complementation experiments of the four genes subjected to this study.Motility assay for strains with each in-frame deletion and their corresponding complementation assays. Swim rings were imaged 48 h after inoculation. Shown are the averages of swim ring diameter for four replicates per strain in millimeters (mm) and standard deviation from the mean. Statistical analysis is shown in S3 and S4 Figs. Strategy used for in-frame deletions is depicted in S5A Fig.
Several insertions associated with motility defects occurred in intergenic regions within the motility cluster (see Fig 1). These are generally interpreted to result in mis-expression of a nearby gene (by disrupting a promoter, overexpression, or antisense RNA expression). It was less straightforward to explain the three intergenic insertions occurring between the genes flaD and ATU0568. As discussed below, flaD is not required for motility, though ATU0568 is. These "flaD-ATU0568" intergenic insertions are intriguing because they occur over a region of 161 bp, with the two insertions farthest from ATU0568 (called IG1 and IG2) transcribing toward flaD (see Fig 1). There is an alternative start codon located 69 bp upstream of the annotated start codon of ATU0568. However, this is not predicted to be disrupted by either insertion (IG1 and IG2) since the closest insertion (IG2) is located 105 bp upstream of this alternative start codon. Sequence specific details of this intergenic region are provided in S5B Fig. Within this region, there is also a predicted open reading frame in the reverse orientation that slightly overlaps with ATU0568 that could explain this motility phenotype. To investigate whether an unannotated gene exists in this region, the sequence disrupted in IG1 and IG2 was manipulated by deleting 10 bp and replacing it with a 6-bp XmaI sequence. This changes the frame and sequence in this region. These two deletion strains exhibited normal motility (S6 Fig), suggesting that something specific to the original transposon insertions, which was not recapitulated by the 10-bp deletions, contributed to the motility defect.
To begin exploring the functions of the four genes analyzed in this study (ATU0568, ATU0583, flgN, and motF), we used the in-frame deletion strains to test whether the mutant phenotype could be reversed by intergenic suppression. Genetic suppression was only observed for the ΔmotF strain. This suppression phenomenon is shown in Fig 5, where extended cultivation of ΔmotF in motility agar results in a blebbing pattern not seen in a non-suppressible mutant such as ΔflaF. Clones extracted from these zones of resumed motility exhibit near-wild-type motility upon retesting. Unlike wild-type cells, however, these suppressor strains seem to generate hypermotile derivatives, as evidenced by a blebbing pattern around the normal ring of motile cells. Future investigation to determine the molecular basis of ΔmotF suppressibility will allow us to make connections between this new gene and previously studied motility functions. For instance, the study by Sobe et al. [[
Graph: Fig 5 ΔmotF has a suppressible phenotype.Strains were inoculated and allowed to swim for 3 (top) or 6 (bottom) days. The ΔflaF strain serves as a non-suppressible control. Below each strain is the average of swim ring diameters in millimeters (mm) for four replicates per strain and the standard deviation from the mean. Statistical analysis is shown in S7 Fig.
A readily noticeable discrepancy between our screening data (Fig 1) and known flagellar motility pathways in Rhizobiaceae is the absence of any mutants in the master regulator genes visN and visR. These two genes are adjacent to one another and encode a likely dimeric LuxR family transcriptional regulator required for motility in several Rhizobiaceae species including S. meliloti and A. fabrum [[
Graph: Fig 6 Characterization of the ΔvisNR deletion strain.(A) Complementation test in which ΔvisNR strains harbor plasmids indicated in S1 Fig. Shown are the averages of swim ring diameters in millimeters (mm) in four replicates per strain and standard deviation from the mean. Statistical analysis is shown in S9 Fig. (B) The left image shows a plate with ΔvisNR colonies, and the right image shows a plate with parent strain BB01. Both strains were grown under the same conditions and images are shown at the same scale.
In this study, we carried out a comprehensive forward genetic screen for motility mutants in Agrobacterium fabrum in which 37 genes were identified as being required for motility based on strong loss-of-motility phenotypes. Based on the suspected functions of proteins encoded by these genes, nearly every molecular component generally required for flagellar assembly in Gram-negative bacteria was identified, in addition to several Alphaproteobacteria-specific functions and four proteins less well characterized. These four proteins (ATU0568, ATU0583, FlgN, and MotF) are all found within a 16-kb region on the right side of the motility gene cluster [[
Of the four flagellar filament genes in A. fabrum (flaA, flaB, flaC, and flaD), only flaA was identified in this screen, which was expected based on previous work showing that this is the only required flagellin, with the others serving subsidiary functions. Earlier studies have shown that ΔflaA mutants form straight flagellar filaments that result in very slow tumbling motion [[
Three of our transposon mutants had insertions in three unique locations between the genes flaD and ATU0568. Within this intergenic region of 279 bp, they were located at position 84, 105 and 245. The transcription from the transposon read to the left for insertions at 84 and 105 and right for insertion 245. With these insertions spread so broadly across this region and the upstream flanking gene (flaD) not required for motility, we can only speculate how these transposon insertions bring about loss of motility. For two of these (IG1 and IG2), disruption of a 10-bp segment corresponding to the wild-type sequence, did not noticeably affect motility. We suspect that these intergenic insertions may disrupt an unusually large regulatory region upstream of ATU0568 or may disrupt or mis-regulate some independent and unannotated feature required for motility.
Mutagenesis by transposon insertion can have polar effects on polycistronic operons. In this screen, with modest transcription emanating from the kanR gene of the transposon into the genome, we did not expect polar effects when the transposon was transcribed inserted in the same direction of the disrupted gene; but we expected possible polar effects when the transposon was transcribed in the opposite direction of the gene. We also observed that there was a strong directionality bias for transposon insertion in some genes such as flaA and ATU0568, but for most genes implicated in the study, insertions in both directions could be found.
The regulatory genes rem, flbT, and flaF were all hit in this screen. The Rem protein is a Class I transcriptional regulator that activates the expression of Class II structural and regulatory genes [[
The Agrobacterium fabrum C58 genome encodes roughly 20 methyl-accepting chemotaxis protein (MCP) homologs, and 9 che genes for chemotactic control of flagellar activity [[
S1 Fig. Parent plasmids used for complementation experiments.
(A) Diagram of parent vector used to make complementation plasmids for ΔATU0568, ΔATU0583, ΔflgN, and ΔmotF strains. Each of the four genes were constitutively expressed by read-through transcription of the kanR gene. (B) Plasmid used as parent vector for visNR complementation derivates. visN, visR, and visNR labels show where each gene was inserted; transcription of visN, visR, and visNR is driven by the native Pvis promoter.
(TIF)
S2 Fig. A transposon screen for motility mutants in A. fabrum aided by pre-enrichment.
(A) Transposon delivery plasmid used to carry out the mutagenesis. TE labels show the location of the repeated transposon elements. (B) Triparental mating scheme involving helper plasmid pRK600 and A. fabrum recipient strain UBAPF2 (SmR). (C) Enrichment strategy enabling high-efficiency screening for motility mutations.
(TIF)
S3 Fig. Quantification of swim ring diameter of knock-out strains shown in Fig 4.
BB01 is the wild-type control. Values are the averages of four swim rings per strain. Error bars show standard deviation from the mean. Different letters denote statistically significant differences (P < 0.05) according to Tukey multiple comparison test.
(TIF)
S4 Fig. Quantification of swim ring diameter of the motility complementation test shown in Fig 4.
Values are the averages of four replicates per strain. Error bars show standard deviation from the mean. Different letters denote statistically significant differences (P < 0.05) according to Tukey multiple comparison test.
(TIF)
S5 Fig. Double-crossover strategy used for targeted in-frame deletions and sequence specific details of the flaD-ATU0568 intergenic region.
(A) The allelic replacement strategy used to construct deletion strains, using a plasmid that allows positive selection (kanR), negative selection (sacB), and color detection (gus). Left and right homology regions are indicated with striped blocks. "goi" refers to the gene of interest to be deleted. The arrow indicates the location of the XmaI sequence between the left and right homology regions. (B) DNA sequence of the intergenic region between flaD and ATU0568. Underlined are the stop and start codons respectively, as well as an alternative start codon for ATU0568. White dots show the insertion site of the transposon that landed in rightward orientation (referring to kanR transcription), while black dots indicate that the transposon landed in leftward orientation.
(TIF)
S6 Fig. Motility assay of intergenic regions (IG1 and IG2).
(A) Swim rings were imaged 48 h after inoculation. Shown are the averages of swim ring diameter for four replicates per strain in millimeters (mm) and standard deviation from the mean. Statistical analysis is shown in (B). (B) Different letters denote statistically significant differences (P < 0.05) according to Tukey multiple comparison test.
(TIF)
S7 Fig. Quantification of swim ring diameter of suppressor analysis 3 days (A) and 6 days (B) post-inoculation.
Swim rings from this suppressor analysis are shown in S7 Fig. Values are the averages of four replicates per strain. Error bars show standard deviation from the mean. Different letters denote statistically significant differences (P < 0.05) according to Tukey multiple comparison test.
(TIF)
S8 Fig. Cell-cell hyper-adherence test.
BB01 was modified to constitutively express lacZ from a synthetic transposon (lacZ+; blue colonies). This lacZ+ strain and ΔvisNR were grown as individual cultures in 5 ml of LB+Sm overnight at 30°C. Equal portions of these overnight cultures were mixed together, diluted, cultured for several hours, and plated on LB+Sm+X-Gal to test for cell hyper-adherence, which would have shown as sectored colonies.
(TIF)
S9 Fig. Quantification of swim ring diameter of complementation tests for the VisNR knock-out mutants in Fig 6A.
Shown are the averages of four swim rings per strain. Error bars show standard deviation from the mean. Different letters denote statistically significant differences (P < 0.05) according to Tukey multiple comparison test.
(TIF)
S1 Table. Bacterial strains used in this study.
(DOCX)
S2 Table. Plasmids used in this study.
(DOCX)
S3 Table. Primers used in this study.
(DOCX)
S1 File. Disrupted genes with two or more independent transposon insertions.
(XLSX)
S2 File. Disrupted genes with single transposon insertions.
(XLSX)
S3 File. Annotated motility cluster of Agrobacterium fabrum.
(XLSX)
S4 File. Plasmid sequences.
(DOCX)
We thank the Microbiology and Molecular Biology department at Brigham Young University for providing the facilities and instrumentation needed for this study. We also thank the editor and reviewers for their thoughtful and helpful comments.
By Diana G. Calvopina-Chavez; Robyn E. Howarth; Audrey K. Memmott; Oscar H. Pech Gonzalez; Caleb B. Hafen; Kyson T. Jensen; Alex B. Benedict; Jessica D. Altman; Brittany S. Burnside; Justin S. Childs; Samuel W. Dallon; Alexa C. DeMarco; Kirsten C. Flindt; Sarah A. Grover; Elizabeth Heninger; Christina S. Iverson; Abigail K. Johnson; Jack B. Lopez; McKay A. Meinzer; Brook A. Moulder; Rebecca I. Moulton; Hyrum S. Russell; Tiana M. Scott; Yuka Shiobara; Mason D. Taylor; Kathryn E. Tippets; Kayla M. Vainerere; Isabella C. Von Wallwitz; Madison Wagley; Megumi S. Wiley; Naomi J. Young and Joel S. Griffitts
Reported by Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author