The Small Regulatory Antisense RNA PilR Affects Pilus Formation and Cell Motility by Negatively Regulating pilA11 in Synechocystis sp. PCC 6803

Pili are found on the surface of many bacteria and play important roles in cell motility, pathogenesis, biofilm formation, and sensing and reacting to environmental changes. Cell motility in the model cyanobacterium Synechocystis sp. PCC 6803 relies on expression of the putative pilA9-pilA10-pilA11-slr2018 operon. In this study, we identified the antisense RNA PilR encoded in the noncoding strand of the prepilin-encoding gene pilA11. Analysis of overexpressor [PilR(+)] and suppressor [PilR(−)] mutant strains revealed that PilR is a direct negative regulator of PilA11 protein. Although overexpression of PilR did not affect cell growth, it greatly reduced levels of pilA11 mRNA and protein and decreased both the thickness and number of pili, resulting in limited cell motility and small, distinct colonies. Suppression of PilR had the opposite effect. A hypothetical model on the regulation of pilA9-pilA10-pilA11-slr2018 operon expression by PilR was proposed. These results add a layer of complexity to the mechanisms controlling pilA11 gene expression and cell motility, and provide novel insights into how sRNA and the intergenic region secondary structures can work together to discoordinatly regulate target gene in an operon in cyanobacterium.


INTRODUCTION
Cyanobacteria are ancient organisms that perform oxygenic photosynthesis (Waterbury et al., 1985). According to endosymbiotic theory, plant chloroplasts originated from cyanobacteria (or a cyanobacteria-like organism) through primary endosymbiosis. Many cyanobacteria move by gliding, swimming, or twitching (Waterbury et al., 1985;Häder, 1987). Gliding motility is a slow, uniform, forward motion, which parallel to the cell's longitudinal axis on a solid surface (Häder, 1987). This type of motion is occasionally interrupted by reversals in filamentous cyanobacteria such as Phormidium uncinatum and Anabaena variabilis (Häder, 1987). Several marine species of unicellular Synechococcus show swimming motility through liquids at a rate of 25 µm s −1 (Waterbury et al., 1985). Twitching motility is small and intermittent translocation on a solid surface with frequent changes in direction (Henrichsen, 1972). Synechocystis sp. PCC 6803, a model unicellular cyanobacterium, exhibits twitching motility on an agar plate or glass slide (Stanier et al., 1971;Ng et al., 2003).
The transcriptome analysis using differential RNA sequencing in Synechocystis sp. PCC 6803 (Xu et al., 2014;Hu et al., 2017) revealed a low-abundance asRNA encoded in the noncoding strand of pilA11, which was named PilR. In this study, we determined the molecular functions of PilR by identifying its target gene, pilA11. An analysis of mutants with either elevated or reduced levels of PilR expression showed that PilR plays a key role in pilus formation and cell motility by negatively regulating pilA11 expression.

Characterization of PilR
Differential RNA sequencing of the Synechocystis sp. PCC 6803 (hereafter Synechocystis) transcriptome detected an asRNA, designated PilR, with a transcription start site at the 3 ′ end of the pilA11 gene, but on its complementary strand. PilR was determined to be 210 long using Northern blot ( Figure 1A) and RACE analyses, and its transcription start site was mapped to nucleotide nt758305 in the sequenced genome. PilR extends from positions 677-886 in the coding sequence of pilA11 (Figure 1B), and can be folded into two extended stem regions, with a terminal loop at each ending as predicted by the mfold software (http://www.bioinfo.rpi.edu/applications/mfold/; Figure 1C), such loops structures are believed to be involved in RNA-RNA interactions and therefore functionally related to the hypothetical trans-acting function (Dühring et al., 2006).
We further constructed background control, PilR overexpression [PilR(+)], and PilR suppression [PilR(−)] strains (Supplementary Figure S1) for investigating the relationship between PilR and its target gene, pilA11. PilR appears to be a negative regulator of pilA11 expression during the exponential growth phase, as revealed through qRT-PCR and immunoblot analyses of these strains (Figures 2A,E). As shown in Figure 2A, the levels of asRNA PilR were 4.28-fold higher in the PilR(+) strain than in the control, and 0.25-fold lower in the PilR(−) strain than in the control (Figure 2A, white speckled bars; * * P < 0.01). By contrast, the levels of pilA11 mRNA were 0.45-fold lower in the PilR(+) strain than in the control, and 2.06-fold higher in the PilR(−) strain than in the control (Figure 2A, gray bars, * * P < 0.01). Similarly, the levels of PilA11 protein were 0.55-fold lower in the PilR(+) strain than in the control, and 1.25-fold higher in the PilR(−) strains than in the control ( Figure 2E).
Frontiers in Microbiology | www.frontiersin.org low steady-state expression level, negatively regulates the amount of pilA11 mRNA and PilA11 protein in Synechocystis.

PilA11 Localization
To visualize the influence of PilR on PilA11 protein, we observed the in vivo localization of PilA11 protein in control, PilR(+), and PiIR(−) strains using immunofluorescence and confocal microscopy. PilR(+) cells exhibited weaker fluorescence compared to the control strain, but PilR(−) cells had significantly enhanced fluorescence compared to the control (Figures 3, 4; Table 1, * * P < 0.01), in agreement with their protein levels ( Figure 2E). When enlarged and visualized at the single cell level, the fluorescence signal could be observed clearly at the cell surface of all strains (Figure 3), as expected given its role in cell motility. We also observed high levels of fluorescence in all three strains in dividing cells (Figure 4, Supplementary Figure S2), suggesting that PilA11 protein is enriched at the cell surface during division regardless of PilR levels.

Pilus Formation and Cell Motility of the PilR Mutant Strains
Previous studies showed that PilA11 was an essential protein for cell motility and thick pili formation (Bhaya et al., 2001;   Panichkin et al., 2006). To test the effect of PilR on PilA11 protein function, we examined cell motility and thick pili formation in the PilR strains. Cyanobacterial cells were grown under normal conditions, and cells were maintained in the presence of 20 µg·mL −1 kanamycin. However, for eliminating possible phenotypic alterations due to the antibiotic, the final cultures without kanamycin were used in the experiments. We examined control, PilR(+), and PiIR(−) cells with an electron microscope to investigate whether changes in PilR expression affected the formation of pili. As shown in Figure 5, the pili of PilR(+) cells were significantly thinner and fewer than the well-developed, normal pili of the control cells. Pili of PilR(−) cells were thicker and denser than those of the control. The number and diameter of pili of fifty individual cells each of the control, PilR(+), and PiIR(−) strains were examined and the results confirmed that PilR overexpression or suppression significantly affected pilus formation ( Table 2).
To investigate the effects of PilR on cell growth and motility in Synechocystis, we analyzed the three strains under normal conditions in liquid culture or on agar plates. As shown in Figure 6A and Table 3, when cultured in liquid BG11, the growth rate and pigmentation were not affected in either the PilR(+) or PilR(−) strains. We examined the effects of PilR overexpression or suppression on motility by monitoring the colonies shape formed on agar plates. The colonies of the PilR(−) strain were larger and more diffuse than the control colonies, whereas those of the PilR(+) strain were smaller and more centralized ( Figure 6B).
We then performed agar surface-based phototaxis assays to clarify the differences in motility between the control, PilR(+), and PiIR(−) strains ( Figure 6C). When the cells were dot plated and exposed to a unidirectional light source for 7 days, the control and PilR(−) mutant strains showed positive phototactic movement, with the latter having a stronger phenotype. The PilR(+) mutant strain showed almost no sign of phototactic movement (Figure 6C), suggesting that PilR expression influences Synechocystis cell motility in a concentration-dependent manner.

DISCUSSION
Many species of cyanobacteria move by gliding, twitching, or swimming. Unlike Escherichia coli and Chlamydomonas reinhardtii, which use flagella, cyanobacteria use a pilus apparatus for motility (Waterbury et al., 1985;Häder, 1987). The  oxygenic phototrophic cyanobacterium Synechocystis exhibits twitching motility (Stanier et al., 1971;Ng et al., 2003). Two morphologically distinct pilus types, thick and thin, exist in wildtype Synechocystis cells. Thick pili, possibly encoded by pilA1, are similar to type IV pili in many functional and morphological characteristics. Thin pili with smaller diameter are shorter than typical type IV pili (Bhaya et al., 2000). Type IV pilus biogenesis requires a complex polypeptides assemblage, located in the cytoplasmic membrane, the periplasm, or the outer membrane, for post-translational modification (e.g., PilD), assembly and export (e.g., PilC, PilQ) (Bhaya et al., 2000).. Type IV pili subunits (i.e., PilA) have a conserved, hydrophobic α-helix domain at the N-terminus, which consists of 20-25 amino acids that forms the hydrophobic pilus core (Proft and Baker, 2009). pilA1 is responsible for the structure, motility, and transformation efficiency of thick pilus (Bhaya et al., 1999;Yoshihara et al., 2001). Mutants with disrupted pilA2, which encodes a second pilin-like protein, are still motile with normal cell-surface pili morphology and density. By contrast, inactivation of pilD, which encodes the leader peptidase, or pilC, which encodes a protein required for pilus assembly, abolishes cell motility and causes the absence of both pilus morphotypes (Bhaya et al., 2000). In our study, suppression of pilA11 by overexpression of its antisense RNA PilR [PilR(+)] leads to thin, sparse pili, whereas PilR suppression [PilR(−)] has the opposite effect ( Figure 5). This indicates that pilA11 may also contribute to thick pilus biogenesis and, indeed, the ratio of thick pili to all pili was altered in the PilR mutant strains.
A locus containing five genes (pilA9-pilA10-pilA11-slr2018-slr2019) was discovered in Synechocystis in an analysis of transposon-generated mutants (Bhaya et al., 2001). The protein encoded by pilA10 shows weak similarity to members of the PilA-like protein family (Bhaya et al., 1999), but the proteins encoded by pilA11 and slr2018 lack obvious functional motifs and have no clear homologs in protein databases. We observed that expression of pilA11, but not pilA9, pilA10, or slr2018, is negatively affected by its antisense sRNA PilR (Figure 2), and that overexpression of PilR disturbs the biogenesis of the thick pilus morphotype (Figure 5), which affects cell motility (Figure 6). These results suggest that PilR negatively regulates pilA11 and has an important function in cell motility.
SpkA and the ATPase PilT are required for motility in Synechocystis (Kamei et al., 2001b;Okamoto and Ohmori, 2002). Bhaya et al. showed that the pilA10, pilA11, and slr2018 genes were essential for cell motility (Bhaya et al., 2001). This study found that on 1% agar-solidified plates the pilA11 asRNA overexpressor PilR(+) colonies are small and distinct, whereas the suppressor PilR(−) colonies are large and diffuse ( Figure 6B). spkA::Cm r mutant cells, in which the pilA9-pilA10-pilA11-slr2018 operon is down-regulated, also form distinct colonies similar to PilR(+) (Panichkin et al., 2006). The observed phenotype of PilR(+) mutant strains is consistent with a previous analysis showing that the putative pilA9-pilA10-pilA11-slr2018 operon in Synechocystis might be involved in the formation of thick pili (Panichkin et al., 2006). However, the effect that expression of the operon has on Synechocystis cells has not been deciphered until now. sRNAs regulate gene expression from polycistronic messages through a variety of mechanisms (Balasubramanian and Vanderpool, 2013), which can alter expression of select genes in an operon by inhibiting the translation of genes or by altering the stability of mRNA, resulting in discoordinate regulation of the target mRNA (Møller et al., 2002;Kalamorz et al., 2007;Desnoyers et al., 2009). Alternatively, they can affect expression of all genes in an operon by sRNA-mRNA interactions, causing coordinate regulation (Rice and Vanderpool, 2011;Lu et al., 2012). The 210 nt sRNA PilR was identified as an asRNA by repressing expression of pilA11 gene, but not pilA9, pilA10, or slr2018, in Synechocystis (Figure 2). It is speculated that PilR prevents pilA11 gene expression by selective mRNA degradation. This type of regulation represses pilA11 gene expression, but allows continued synthesis of other Type IV pilin-like proteins.
FIGURE 7 | Hypothetical model on the regulation of pilA9-pilA10-pilA11-slr2018 operon expression by PilR. (A) A secondary structure of the intergenic region between pilA11 and slr2018 cistrons (represented by hairpin structure in panel B) as determined by the mfold software (http://www.bioinfo.rpi.edu/applications/mfold/). (B) Normally, single-stranded pilA9-pilA10-pilA11-slr2018 mRNA is translated into three type 4 pilin-like protein, PilA9, PilA10, PilA11 and the unknown protein Slr2018 (upper panel). While the degradation of pilA11 mRNA is typically triggered by sRNA PilR base pairing to the target mRNA, and results in translational silencing and promoted degradosome-dependent endonucleolytic cleavage of the pilA11 portion of the mRNA (middle panel). The degradosome is stopped and discharged by the hairpin structure between the pilA11 and slr2018 cistrons and thus selectively prevents the slr2018 mRNA from degradation (lower panel).
To further elucidate the discoordinate regulation mechanism of the cis-type PilR transcript to pilA11 gene, a closer sequence inspection is needed in both target gene pilA11 and asRNA PilR, their cleavage by RNases. By sequence analysis, a strong secondary structure (Figure 7A) in the intergenic region between the pilA11 and slr2018 cistrons was predicted. Add up to the present understanding of the function of PilR in the regulation of pilA11 gene expression, a hypothetical model on the regulation of pilA9-pilA10-pilA11-slr2018 operon expression by PilR was proposed (Figure 7). Normally, single-stranded pilA9-pilA10-pilA11-slr2018 mRNA is translated into three type IV pilin-like protein, PilA9, PilA10, PilA11 and the unknown protein Slr2018 (Figure 7B, upper panel). While the degradation of pilA11 mRNA is typically triggered by sRNA PilR base pairing to the target mRNA, and results in translational silencing and promoted degradosome-dependent endonucleolytic cleavage (Kaberdin et al., 2011;Balasubramanian and Vanderpool, 2013) of the pilA11 portion of the mRNA, presumably by RNase E (Figure 7B, middle panel). The degradosome is stopped and discharged by the hairpin structure ( Figure 7A) between the pilA11 and slr2018 cistrons and thus selectively prevents the slr2018 mRNA from degradation ( Figure 7B, lower panel). The suggested model provides novel insights into how sRNA and the intergenic region secondary structures can work together to discoordinatly regulate target gene in an operon in Synechocystis (Figures 2A,E).
Cis-encoded asRNA transcripts typically regulate target gene expression either negatively, as is the case for IsrR (Dühring et al., 2006) and As1_flv4 , or positively, as is the case for PsbA2R, PsbA3R , and RblR (Hu et al., 2017). These asRNAs are involved in various processes (Kopf and Hess, 2015), but prior to this study, no asRNAs related to cell motility were known. Our results show that the asRNA PilR is encoded by the antisense strand of the pilA11 gene and negatively regulates its expression through complementary base pairing. Downregulation of this member of the putative pilA9-pilA10-pilA11-slr2018 operon reduces the number and thickness of the pili, affecting cell motility. These results add a new layer of complexity to the regulatory mechanisms controlling pilA11 gene expression and function and hence cell motility.

Strains and Growth Conditions
Wild-type Synechocystis was cultured as described previously (Hu et al., 2017). Solid medium supplemented with 1% (w/v) or 0.5% (w/v) agar is used to observe colony morphology for motility evaluation. After generating mutant strains (see below), kanamycin (20 µg/mL) was added to the growth medium to identify the transformed cells. Antibiotics were excluded during phenotyping to avoid interactions. The colonies were trained for 7 days under lateral illumination with a white fluorescent lamp (∼30 µmol photons·m −2 ·S −1 ).

RNA Extraction and Northern Blot Analysis
Total RNA extraction was performed as described previously (Hu et al., 2017). Northern blot analysis was performed as previously described (Hu et al., 2014). [γ-32 P] ATP (PerkinElmer, USA) was used for the labeling of probes. DNA oligonucleotides used for the Northern blot analysis are listed in Table 4.

′ -and 3 ′ -Rapid Amplification of cDNA Ends (RACE)
The 5 ′ end and 3 ′ end RACE was performed according to Hu et al. (2014). All oligonucleotides and primers used in the RACE analysis are listed in Table 4.

qRT-PCR Validation
The qRT-PCR analysis was done by standard procedure using cDNA as previously described (Hu et al., 2014(Hu et al., , 2017. All data are shown as the mean ± SD (n = 5). All primers used for the analysis are listed in Table 4.

Electron Microscopy
Specimens were prepared for electron microscopy using the conventional negative staining procedure (Bhaya et al., 1999) with the following modifications. Briefly, a 200 µL drop of sample solution was adsorbed onto a glowdischarged carbon-coated copper grid for 10 min, stained with two drops of freshly prepared 2% phosphotungstic acid (pH 7.0), and examined using a Hitachi HT7700 microscope.

Immunofluorescence Microscopy
Microscopy analysis of cells grown in liquid BG11 medium was carried out using a confocal scanning microscope (Leica TCS SP8, Wetzlar, Germany). PilA11 localization was performed as previously described (Miyagishima et al., 2005;Zhan et al., 2016). The fluorescence intensity was measured and quantified using the Quantity One software package, version 4.0.0 (Bio-Rad, http:// www.bio-rad.com/) in 50 individual cells for each of the three strains.

Measurements of Chlorophyll a and Carotenoid Contents
Chlorophyll a and carotenoid contents were measured as described (Zhang et al., 2013). Chlorophyll a (Ca) and total carotenoid (Cc) contents were calculated according to Equation (1) (Harmut and Lichtenthaler, 1987

Calculating the Number and Diameter of Pili
Number and diameter of pili were calculated for the crosssections of 50 cells of electron microscopy using ImageJ software. All values are means ± standard deviation. Significant differences between the control (CON) and test values were determined using a one-way ANOVA. * * P < 0.01 vs. CON.

AUTHOR CONTRIBUTIONS
JH and QW: Designed the study; JH, JZ, HuiC, CH, and QW: Collected, analyzed, and interpreted the data; JH, HuaC, and QW: Wrote the manuscript.