Minor pilins are involved in motility and natural competence in the cyanobacterium Synechocystis sp. PCC 6803

Cyanobacteria synthesize type IV pili, which are known to be essential for motility, adhesion and natural competence. They consist of long flexible fibers that are primarily composed of the major pilin PilA1 in Synechocystis sp. PCC 6803. In addition, Synechocystis encodes less abundant pilin‐like proteins, which are known as minor pilins. In this study, we show that the minor pilin PilA5 is essential for natural transformation but is dispensable for motility and flocculation. In contrast, a set of minor pilins encoded by the pilA9‐slr2019 transcriptional unit are necessary for motility but are dispensable for natural transformation. Neither pilA5‐pilA6 nor pilA9‐slr2019 are essential for pilus assembly as mutant strains showed type IV pili on the cell surface. Three further gene products with similarity to PilX‐like minor pilins have a function in flocculation of Synechocystis. The results of our study indicate that different minor pilins facilitate distinct pilus functions. Further, our microarray analysis demonstrated that the transcription levels of the minor pilin genes change in response to surface contact. A total of 122 genes were determined to have altered transcription between planktonic and surface growth, including several plasmid genes which are involved exopolysaccharide synthesis and the formation of bloom‐like aggregates.

The core components of T4P are widely conserved across different prokaryotic phyla (Denise et al., 2019). T4aP subunits of Synechocystis have been identified by homology to Pseudomonas aeruginosa (hereinafter designated Pseudomonas) and Myxococcus xanthus (hereinafter designated Myxococcus) pilus proteins (Bhaya et al., 1999(Bhaya et al., , 2000Yoshihara et al., 2001). PilQ is an integral outer membrane secretin and enables the emergence of the extending pilus fiber. The pilus platform is formed by the integral inner membrane protein PilC and traffic ATPases that facilitate the extension (PilB) or retraction (PilT) of the fiber. The PilMNO proteins align the pilus components throughout the periplasm. The most abundant structural protein of the pilus fiber is the major pilin PilA. PilA prepilins are inserted into the inner membrane, where they are processed by the bifunctional leader peptidase/methylase PilD and are subsequently inserted into the growing fiber (reviewed in Pelicic, 2008). In addition to the highly abundant major pilin, T4P-forming bacteria usually encode several, less abundant minor pilins with different functions.
Minor pilin genes are often clustered and specific subsets of genes are differentially expressed and likely coregulated. The Myxococcus genome contains three gene clusters each encoding a variant of the adhesin PilY1 and four minor pilins, the so-called core minor pilins.
These form a priming complex which is essential for T4aP pilus assembly. The complex of the minor pilins and PilY1 locates at the tip of the extended pilus and promotes adhesion to a surface (Treuner-Lange et al., 2020). In Pseudomonas, a similar minor pilin operon consisting of seven genes is under control of the transcription factor AlgR (Belete et al., 2008). The encoded set of minor pilins together with PilY1 was also suggested to function in pilus assembly (Nguyen et al., 2015) and to be incorporated into the pilus filament (Giltner et al., 2010). Further specific minor pilins (also called non-core minor pilins, Jacobsen et al., 2020) are involved in diverse pilus functions, such as aggregation, adhesion and natural transformation, in various bacteria (Ng et al., 2016;Winther-Larsen et al., 2001;Wolfgang et al., 1999), including the cyanobacterium Synechococcus elongatus PCC 7942 (Taton et al., 2020). The involvement of specialized minor pilins in DNA binding at the T4P tip was demonstrated for Vibrio cholerae (Ellison et al., 2018). In addition, the pilin-like proteins ComP of Neisseria meningitidis (Cehovin et al., 2013) and ComZ of Thermus thermophilus (Salleh et al., 2019) were suggested to interact with DNA at the pilus tip. Notably, these pilins are important for natural transformation but not for piliation.
PilA3 was misidentified and is now considered a Tat protein that has characteristics of both TatA and TatB of the Tat protein secretion system (Aldridge et al., 2008). All major and minor pilins contain a conserved PilD cleavage site (G|XXXXE) followed by a hydrophobic stretch (Linhartová et al., 2014). Most minor pilin genes of Synechocystis are organized in operons (Figure 1a). Based on the transcriptomic data by Kopf et al. (2014), the transcriptional unit (TU) TU763 comprises a polycistronic mRNA encoding the minor pilins PilA9, PilA10 and PilA11 together with the open reading frames slr2018 and slr2019. The minor pilin genes pilA5 and pilA6 constitute TU2300. The downstream pilA7 and pilA8 genes are most likely transcribed from a different promoter. To date, knowledge regarding minor pilin function in Synechocystis is scarce. Deletion mutants of pilA1, pilA10 and pilA11 are non-motile on agar plates (Bhaya et al., 2001;Yoshihara et al., 2001). Furthermore, it was demonstrated that deletion of the whole pilA9-slr2019 genomic region not only leads to loss of motility (Wallner et al., 2020) but also impairs flocculation (Conradi et al., 2019), which describes the aggregation of cells into floating assemblages in liquid culture. Hu et al. (2018) showed, that the antisense RNA PilR negatively regulates the amount of pilA11 mRNA and the corresponding protein. Thus, overexpression of the antisense RNA led to inhibition of motility in this study. In another study it was shown that PilA4 co-localizes with the pilus fiber if the major pilin PilA1 is present (Cengic et al., 2018).
As in many other bacteria, a functional T4P system is crucial for natural competence of Synechocystis (Yoshihara et al., 2001). The uptake mechanism has not been fully elucidated, but it is known that DNA is processed to a single-stranded form during DNA uptake (Barten & Lill, 1995) and that homologs of the proteins ComA (Slr0197) (Yoshihara et al., 2001;Yura et al., 1999) and ComF (Slr0388) (Nakasugi et al., 2006) are important for competence of Synechocystis cells. A protein with homology to ComEC was identified by Yoshihara et al. (2001), but they were not able to inactivate the corresponding gene sll1929.
T4P are also involved in surface attachment and biofilm formation, which are important in the life of many prokaryotes. Biofilms confer a fitness advantage over planktonic solitary cells by protection from diverse environmental stresses, better nutrient availability and drug and predator resistance (Laventie & Jenal, 2020). Surface recognition by different appendages, including T4P, triggers signal transduction cascades involving second messengers, quorumsensing systems, two-component systems and small regulatory RNAs (Laventie & Jenal, 2020). For Pseudomonas, which possesses T4aP and a single polar flagellum, it was suggested that pilus tension leads to conformational change of the T4P, which subsequently leads to 3ʹ,5ʹ-cyclic adenosine monophosphate (cAMP) production and activation of virulence programmes (Persat et al., 2015). Additionally, surface contact of the flagellum (Laventie et al., 2019), as well as surface adhesion of PilY1 together with a functional T4P (Rodesney et al., 2017), trigger the production of 3ʹ,5ʹ-cyclic dimeric guanosine monophosphate (c-di-GMP). Increased c-di-GMP levels promote surface acclimation by inhibiting flagella motility, thereby promoting faster T4P-dependent surface attachment and production of extracellular polymeric substances (EPS) (Rodesney et al., 2017).
In this study, we investigate the role of minor pilins in Synechocystis and show that the minor pilin PilA5 is involved in natural competence. Furthermore, our data imply that among others, genes encoding minor pilins, cell envelope structures and genes located on the plasmid pSYSM are major targets of a putative surface sensing system. We also investigated changes in second messenger production upon surface contact.

| Minor pilins play roles in motility and surface attachment
In addition to the major pilin PilA1, Synechocystis encodes at least nine different minor pilins (Linhartová et al., 2014). Considering that the TU2300, encoding the minor pilins PilA5, PilA6 and the TU763, encoding PilA9, PilA10, PilA11 together with the hypothetical proteins Slr2018 and Slr2019, are inversely regulated in response to blue light (Wallner et al., 2020), we questioned if these sets of minor pilins are involved in different pilus functions. In contrast to the ΔpilA9-slr2019 mutant, which is non-motile, a ΔpilA5-pilA6 mutant strain showed a clear phototaxis response (Wallner et al., 2020). To identify other phenotypes, we first tested whether the ΔpilA5-pilA6 strain is impaired in flocculation. Figure 2 shows that the ΔpilA5-pilA6 strain exhibits a WT flocculation response and clearly differs from the non-flocculating and non-motile ΔpilA9-slr2019 and Δhfq mutant strains (Conradi et al., 2019;Wallner et al., 2020). The Δhfq strain was employed as a control. The function of the cyanobacterial homologue of the RNA chaperone Hfq is not well understood yet. However, it is known that deletion or mutation of the hfq gene in Synechocystis led to transcriptional changes of many genes, including minor pilin genes, several ncRNAs and the CRISPR systems and therefore this mutant seems to be dysregulated in many ways (Dienst et al., 2008). Hfq is an interaction partner of the secretion ATPase PilB1 and seems to co-localize with the pilus base F I G U R E 1 Major and minor pilins of Synechocystis. (a) Schematic representation of Synechocystis gene clusters encoding PilA1 and minor pilins. PilX homologues slr0226, slr0442 and sll1268 are not encoded in an operon or in proximity to other pilins. The proteins encoded by sll1696 of TU1330 and slr2019 of TU763 are not minor pilins. (b) Multiple alignment of the major pilin PilA1 and the proposed minor pilins. The potential cleavage site by PilD is marked with the dotted line and an arrow. The conserved glycine and glutamate at positions −1 and +5 in relation to the cleavage site are highlighted in turquois and blue, respectively. The newly proposed minor pilins PilA12 (Slr2018), Slr0226, Slr0442, and Sll1268 contain a hydrophobic amino acid substitution at position +5. The length of the full-length protein/mature protein is given next to the protein name in brackets. Hydrophobic amino acid residues are highlighted in grey (Schuergers et al., 2014). For our study, we used the Δhfq strain because inactivation of the hfq gene leads to non-motile and T4P lacking cells (Dienst et al., 2008).
To determine whether the attachment of T4P to surfaces and their dynamics is impaired in the ΔpilA5-pilA6 and ΔpilA9-slr2019 strains, we used fluorescent beads and monitored the retraction of the beads to the cell surface, as described by Nakane and Nishizaka (2017). To that end, a glass surface was coated with 4% collodion; thus, cells were not able to move, though they were still able to assemble and retract T4P. The bead assays suggest that the WT pili attach to nearby beads and transport them towards the cell (Figure 3a and Video S1). Interestingly, ΔpilA9-slr2019 mutant cells were not able to adhere to the collodion surface. However, in some cases, single cells were trapped on the glass slide, but they were not able to attach to the beads (Figure 3b and Video S2). In contrast, ΔpilA5-pilA6 mutant cells were able to retract beads towards the cell (Figure 3c and Video S3).
These results indicate that the lack of the pilA9-slr2019 gene cluster causes a defect in T4P function related to motility and attachment to biotic and abiotic substances, whereas the minor pilins PilA5 and PilA6 are dispensable for these processes.

| Role of minor pilins in natural competence
T4P are also known to be important for natural competence (Piepenbrink, 2019;Schirmacher et al., 2020;Yoshihara et al., 2001). Therefore, we investigated the ability of the minor pilin mutants described above to be transformed by exogenous DNA. We used a plasmid that enables integration via homologous recombination of a streptomycin resistance cassette into the chromosomal region of the gene encoding the small RNA PsrR1 (Georg et al., 2014). The ΔpilA9-slr2019 mutant strain was transformable with a transformation efficiency similar to that of the WT (Figure 4). The Δhfq strain was employed as a negative control, as this strain is not transformable (Dienst et al., 2008). Notably, the ΔpilA5-pilA6 mutant strain was not transformable (Figure 4). To discriminate the function of PilA5 and PilA6, we complemented the ΔpilA5-pilA6 strain with plasmids for the expression of pilA5, pilA6 or the whole pilA5-pilA6 operon ( Figure S1). To each coding sequence, the putative native promoter region (extending 450 bp upstream of the pilA5 start codon) was fused. Though the standard deviations of transformation efficiency were very high in these experiments, it is clear that only in the strains that contained the pilA5 gene, either the pilA5 gene alone or F I G U R E 2 Involvement of minor pilins in flocculation. (a) Representative flocculation assays for WT Synechocystis cells and deletion mutants Δhfq, ΔpilA9-slr2019, and ΔpilA5-pilA6. Chlorophyll fluorescence is shown in inverted greyscale; thus, areas with more chlorophyll appear darker. (b) Mean aggregation values of WT (n = 19), ΔpilA5-pilA6 (n = 19), Δhfq (n = 12), and ΔpilA9-slr2019 (n = 7). Three asterisks indicate statistical significance with p values ≤ .001, calculated via Student's t test in combination with pilA6, streptomycin-resistant clones were detectable ( Figure 4). Exconjugants containing the plasmid with only the pilA6 gene were not transformable (Figure 4), and Northern Blot analysis verified that pilA6 was transcribed in this strain ( Figure S2). Therefore, we conclude that the minor pilin PilA5 is essential for DNA uptake.

| Minor pilin mutants assemble thick pili on the surface of Synechocystis cells
To determine whether the lack of specific pilus functions is related to a defect in pilus assembly, we examined negative stained cells by transmission electron microscopy (TEM). Notably, we were able to detect thick pili in the WT and in the two minor pilin operon deletion mutants ( Figure 5). In all strains, we measured a diameter of the thick pili between 5.5 and 9 nm ( Figure S3), which is comparable to the 6 to 8 nm described by Bhaya et al. (2000). This result suggests that all minor pilin deletion mutants can assemble T4P. Moreover, immunodetection and quantification of the major pilin PilA1 in sheared pili fractions indicates, that piliation of ΔpilA9-slr2019, ΔpilA5-pilA6 and ΔpilA5-pilA6 complementation mutants is comparable to the WT ( Figure 6).
In addition to T4P, Synechocystis also possesses thin pili with unknown molecular composition and function. All mutant strains analyzed in our study were shown to produce these thin filaments ( Figure 5). Therefore, the minor pilins PilA5, PilA6, PilA9, PilA10 and PilA11 and proteins Slr2018 and Slr2019 are not involved in general formation of thin or thick pili. Notably, with our improved imaging procedure, we could also visualize thin pili on the Δhfq strain ( Figure 5f). This is in contrast to previous observations by Dienst et al. (2008), which described the mutant nonpiliated.
In summary, minor pilin deletion mutants assemble thick pili, and the presence of thin pili was not impaired in any of the constructed mutants. Therefore, loss of motility and flocculation in ΔpilA9-slr2019 and impaired natural competence in ΔpilA5-pilA6 did not correlate with loss of T4P or thin pili.

| Transcriptional changes in response to surface acclimation
Minor pilins are involved in attachment and surface motility. Therefore, we attempted to determine if their transcription is regulated upon surface contact and whether Synechocystis is able to respond to surface contact in general. In Pseudomonas, mechanosensing of a surface with T4P leads to upregulation of a cAMPdependent operon with significant transcriptional changes within 3 hr of surface contact (Persat et al., 2015). To evaluate the acclimation capacity of the slower growing Synechocystis, researchers usually analyze gene expression changes within 1 to 24 hr after transfer to new conditions (Hernández-Prieto et al., 2016). Therefore, we analyzed genome-wide transcriptional changes between Synechocystis planktonic cell cultures and cells on a surface (sessile culture) after 4 and 8 hr of surface incubation. We employed a microarray design that enables identification of all coding and non-coding RNA transcripts identified by previous transcriptomic studies (Klotz et al., 2016;Kopf et al., 2014;Mitschke et al., 2011). We defined a log 2 fold change (FC) ≥ |−0.8| and an adjusted p value < .05 as the threshold for a transcriptional change (comparable to Wallner et al., 2020). Based on F I G U R E 3 Pili can bind and retract fluorescent beads. Picture series of WT (a), ΔpilA9-slr2019 (b), and ΔpilA5-pilA6 (c) cells binding fluorescent beads in the medium. Cells are immobilized on a collodion surface and indicated by dotted circles. Representative pilus-bound beads are indicated by solid line circles. Scale bars: 3 µm. Original videos can be found in the supplementary data (Videos S1-S3) these criteria, the transcript accumulation of 122 genes changed after surface acclimation ( Figure S4, Tables 1 and 2). As depicted on the heatmap in green ( Figure 7) and in the volcano plot in blue ( Figure S4), genes on the 120 kbp pSYSM plasmid showed the highest differential transcription upon surface contact. The transcript levels of 10 genes on the pSYSM plasmid were downregulated, and three were upregulated, together representing approximately 10% of the 132 annotated genes on this plasmid (Kaneko et al., 2003).
One of the most upregulated TU at both time points after transfer of the cells to a surface is located on the pSYSM plasmid and encodes the two hypothetical proteins Slr5087 and Slr5088.
Notably, the mRNA of the minor pilin gene operon pilA5-pilA6 accumulated under sessile conditions, whereas the genes pilA9, pilA10, pilA11 and slr2018 showed higher transcript accumulation during the planktonic lifestyle. The downstream slr2019 open reading frame does not appear to be differentially transcribed based on our significance criteria, implying that this gene may not be part of TU763 or that there is posttranscriptional gene regulation.
Interestingly, the most strongly upregulated mRNA after 4 hr (but not after 8 hr) of incubation on an agar plate encodes the high affinity bicarbonate uptake system SbtA (Shibata et al., 2002). Further genes known to accumulate under inorganic carbon (C i ) limitation (e.g., sll1594, encoding the transcriptional regulator NdhR, and the hypothetical genes slr1634, slr1770, slr0226, slr0006, and slr2006 [Orf et al., 2016]) showed the same response in our microarray study, implying that cells experience a short term CO 2 limitation during surface acclimation. Other genes of components of the inducible bicarbonate/CO 2 uptake systems (e.g., ndhF3/D3, cmp, and cupA) show a similar altered transcription, however they did not meet our significance criteria.
Furthermore, most upregulated genes 4 hr after transfer to agar plates are sll0783, which is involved in polyhydroxybutyrate synthesis (Schlebusch & Forchhammer, 2010), and slr1667, a component of a putative chaperone usher pathway (Schuergers & Wilde, 2015) ( Table 1). Loci that show increasing transcript accumulation in sessile cultures over the 8-hr acclimation period are either located on pSYSM or are involved in nitrate or cyanate metabolic processes (e.g., nrtAB and cynS) or pilus function (such as the minor pilin genes pilA5 and pilA6) ( Table 1). A further operon with a similar transcription profile is slr1593/slr1594, which encodes an EAL domain protein and a CheY-like response regulator with unknown functions (Table 1).
The most downregulated genes in response to surface exposure are again transcribed from the pSYSM plasmid (Table 2). Other genes that show higher transcription in planktonic cultures than in sessile cultures encode sulfate and phosphate transport systems, proteins related to transcription/translation and hypothetical proteins ( Figure 7; Table 2). The transcription levels of some genes remain low or decrease further in surface-grown cells after 8 hr. Among them are the plasmid gene slr5055, encoding a glycosyltransferase of a newly identified EPS synthesis cluster (Maeda et al., 2021) and the polycistronic mRNA for minor pilins (pilA9-pilA11). These data were verified by Northern blot hybridizations ( Figure S5) which also show that at least the mRNA which hybridized to the pilA9 probe exhibited lower signal intensity under sessile conditions already at 10 min and 1 hr after surface contact ( Figure S5b).

| Identification of putative new minor pilins of Synechocystis
Our microarray analysis identified several chromosomal genes that showed a comparably stable transcriptional change during the 8 hr of surface acclimation. Among these genes, slr0226, as well as pilA5 and pilA6, exhibited increased transcription under sessile conditions, whereas slr0442 and the pilA9-pilA11 transcripts accumulate in planktonic cultures over the whole sampling time (Tables 1 and 2).
Synteny analysis using the tool FlaGs (Saha et al., 2020, Table S6) demonstrated that in many cyanobacteria, a hypothetical protein ( Figure S6; Table S1) is encoded directly upstream or downstream of F I G U R E 5 Electron micrographs of different negatively stained Synechocystis strains. WT (a, e), T4P lacking deletion mutant Δhfq (b, f) and deletion mutants ΔpilA9-slr2019 (c, g) and ΔpilA5-pilA6 (d, h) were negatively stained with 1% uranyl acetate. Whole cells are depicted in (a-d) (scale bar 2 µm) and ultrastructural details of pili are shown in (e-h) (scale bar 200 nm) with distinct types of thick pili (black arrow) and thin pili (arrowhead) pilA5 homologues from other cyanobacteria ( Figure S6a, Table S1).
Such a gene was not found in proximity to pilA5 in Synechocystis.
However, using a protein sequence of this hypothetical gene (WP_008273982.1) as query, we identified in PHI-BLAST searches three Synechocystis genes with e-values less than 6 x 10 -11 , which are slr0226, slr0442 and sll1268. Further analysis showed that pilins are often encoded in genetic proximity to the homologues of these three genes ( Figure S6b-d, Table S1). Slr0226, Slr0442, and Sll1268 are annotated as hypothetical proteins, but all contain a conserved PilX-N-terminal domain (HHsearch probability > 95%; Figure S7). PilX is a minor pilin from Pseudomonas that is essential for motility and is implicated as a key promoter of pilus assembly (Giltner et al., 2010).
In order to reveal a possible function of these putative minor pilin genes, we created knock-out mutants of the pilX-like genes slr0226 (pilX1), slr0442 (pilX2) and sll1268 (pilX3) ( Figure S8). First, we checked the piliation of the Synechocystis pilX mutant strains with TEM ( Figure 8). Clearly, all three pilX deletion mutants present thick and thin pili on their surface ( Figure 8). Phototaxis experiments revealed that the mutant strains ΔpilX1 and ΔpilX3 are motile, whereas the ΔpilX2 deletion mutant showed highly reduced motility ( Figure 9a). Notably, pilX2 shows the same transcription pattern as pilA9-slr2019 in our transcriptomic analysis. Further, we tested the newly created ΔpilX deletion mutants for flocculation and discovered that none of the pilX deletion mutants was able to flocculate ( Figure 9b). A transformation test, however, revealed that all pilX mutants are transformable ( Figure 9c). Thus, the pilX2 and pilA9-slr2019 mutants have a similar phenotype, whereas ΔpilX1 and ΔpilX3 show a new combination of phenotypic characteristics. They are not able to form flocs, though they are still motile and transformable. A summary of the phenotypes of minor pilin mutants analyzed in our study in comparison to the WT and a ΔpilA1 strain is given in Table 3.

| Possible functions of pSYSM-encoded genes
Some of the most differentially transcribed genes between planktonic and surface-grown cells were encoded on the pSYSM plasmid (Tables 1 and 2). Ten genes with significantly lower transcript accumulation upon surface contact belong to a gene cluster consisting of 19 genes involved in biosynthesis of the recently identified sulphated extracellular polysaccharide synechan. This polysaccharide was shown to be essential for aggregate formation in planktonic cultures (Maeda et al., 2021).
The three genes with upregulated transcription in sessile lifestyle could not be classified into a specific biological process (slr5087, slr5088, slr5037; Table 1). As the genes slr5087 and slr5088 showed the highest upregulation upon surface contact, we sought to test the phenotypes of their respective deletion mutant, Δslr5087-5088 ( Figure S9). We performed phototaxis experiments and flocculation assays. However, we could not detect any difference between the mutant and the WT ( Figure S9). Notably, phototactic movement of this mutants and the pilA5-pilA6 minor pilin mutants was reduced compared to the WT. We often observe this phenomenon for different mutants which might be due to secondary effects of the antibiotic resistance cassette insertion. We could show that the complementation of the pilA5-pilA6 deletion restored natural competence but also resulted in retarded movement ( Figure S10). These differences are difficult to quantify and may vary between different experiments. Therefore, based on macroscopic motility on plates, we do not expect a possible minor involvement of these genes in motility, though we are not able to completely exclude such effects at the moment.

| Concentrations of the second messenger nucleotides cAMP and c-di-GMP change upon acclimation to a surface
Previously, it was shown that the two sets of minor pilins analyzed in this study are differentially transcribed in a c-di-GMP-dependent manner (Wallner et al., 2020). Therefore, we analyzed the intracellular accumulation of c-di-GMP upon the transfer of cells from a planktonic lifestyle to a surface. Cells were treated as for RNA sampling. We were not able to detect significant changes (p values ≤ .05) in the cellular c-di-GMP level after planktonic cells were incubated for 4 and 8 hr F I G U R E 6 PilA1 quantification of different minor pilin mutants. Representative Western Blot analysis of sheared pili fractions of the WT, the Δhfq, the ΔpilA9-slr2019, the ΔpilA5-pilA6 deletion mutants, as well as the ΔpilA5-pilA6/+pilA5, ΔpilA5-pilA6/+pilA6 and ΔpilA5-pilA6/+pilA5-pilA6 reinsertion mutants with an antibody against the major pilin PilA1 and relative PilA1 levels to the mean WT PilA1 level (n = 3 on an agar plate ( Figure 10). Therefore, an altered c-di-GMP content appears not to be directly responsible for the transcriptional changes detected in the microarray analysis at these time points. However, el-  (Wallner et al., 2020) and altered gene transcript levels upon surface contact indicates only a small overlap or even opposite changes between these two analyses (see Section 3). Further, we revealed a rapid response of cAMP accumulation upon surface contact and an approximately 3.6-fold elevated cAMP level after 4 hr on agar compared to planktonically grown cells.
Taken together, we detected a very rapid increasing response of the second messengers cAMP and c-di-GMP within the first minutes after surface contact. However, the intracellular second messenger levels after 4 and 8 hr were almost indistinguishable between sessile and planktonic cells.

| Response of Synechocystis to surface contact
Our transcriptome analyses demonstrated that minor pilin genes are targeted by a putative surface response system in Synechocystis  Table 1). The accumulation of these two transcripts which encode five different minor pilins was also reduced in Synechocystis mutants lacking the homolog of the RNA chaperone Hfq (Dienst et al., 2008), the transcription factors LexA (Kizawa et al., 2016) and SyCRP1 (Yoshimura, Yoshihara, et al., 2002)  responding genes in our microarray study, which suggested a considerably larger response to that signal compared to c-di-GMP-based regulation. However, the transcriptional changes between surface versus planktonic growth in the current study and high c-di-GMP versus low c-di-GMP conditions (Wallner et al., 2020) do not support the idea that this second messenger is directly or solely involved in controlling gene transcription in response to a surface. For example, the pilA9-slr2019 operon was downregulated during surface incubation (Table 2), whereas a high c-di-GMP content has been observed to lead to upregulation of this mRNA (Wallner et al., 2020). Laventie et al. (2019) showed for Pseudomonas that the c-di-GMP concentration increased within seconds upon surface contact, thereby leading to pilus assembly and enhanced surface attachment. We were also able to measure an increase in c-di-GMP and cAMP levels at shorter time points, such as 10 min or 1 hr after transfer to sessile conditions, suggesting that there is a short-term surface response in Synechocystis involving second messenger molecules. However, whether and how such a signal triggers some of the specific gene expression changes we observed after 4 and 8 hr of surface acclimation needs to be studied in the future. At least for pilA9 such a short response on transcriptional level was detected ( Figure S5b), but a more comprehensive analysis on the timing of acclimation The two CRP-like transcription factors SyCRP1 and SyCRP2 have been determined to be involved in second messenger signalling and are known to control the pilX2 and the pilA9-slr2019 operon (Song et al., 2018;Wallner et al., 2020;Yoshimura, Yoshihara, et al., 2002). Other studies predicted the binding of SyCRP1 to the promotor regions of pilA9-2019, pilA5-pilA6, pilX2 and pilX3 operons or genes. This was confirmed by in vitro binding studies for the pilA5-pilA6, pilA9-2019, pilX2 and pilX3 promoters (Omagari et al., 2008;Song et al., 2018). Together, these data suggest that cAMP is an important signal for the expression of minor pilins. It remains to be elucidated whether Synechocystis can sense a surface contact mechanically via its T4P. In P. aeruginosa, surface contact and shear stress generate tension in the pili leading to elevated levels of the second messengers cAMP and c-di-GMP (Gordon & Wang, 2019).
Further, we are not able to exclude that the cells also respond to the high culture density which is achieved on the surface of an agar plate. On one hand, high cell density affects light, nutrient and CO 2 availability, on the other hand, a high concentration of cells on a surface makes not only surface contact, but also a contact between the cells (e.g., via pili) more likely. In general, transcript accumulation of the minor pilin genes changes in response to several stress factors, e.g., the amount of both pilA5-pilA6 and pilX1 mRNAs is also upregulated under C i conditions, stationary phase, heat stress and in the dark, whereas pilA9-2019 and pilX2 accumulation responds to low temperature and nitrogen limitation (Eisenhut et al., 2007;Kopf et al., 2014). This suggests a complex regulatory network on the level of transcript accumulation and/or stability for the minor pilin genes.

| Function of minor pilins in natural competence, motility, and flocculation
One of the mRNAs which accumulated in cells on a surface encodes the minor pilin PilA5 which is essential for natural competence. Within a biofilm it might be beneficial for the cell to take up DNA from adjacent cells, either as a nutrient or to gain new traits (Ibáñez de Aldecoa et al., 2017;Vorkapic et al., 2016). In freshwater environments, the concentration of extracellular DNA ranges from 1.75 to 7.7 μg/L (DeFlaun et al., 1986;Paul et al., 1989) which is an order of magnitude lower than in e. g in soil samples or in biofilms (Niemeyer & Gessler, 2002;Torsvik & Goksoyr, 1978). Thus, it seems more unlikely for planktonic, single cells to be in proximity of free-floating DNA in a water column; therefore, the demand for DNA uptake and PilA5 might be lower in a planktonic culture. In contrast to pilA5, the pilA9-slr2019 operon is dispensable for natural competence but is involved in motility, bead attachment and flocculation. During floc formation, cells presumably bind to other cells or cellular components excreted by adjacent cells, such as T4P or EPS (Trunk et al., 2018). We assume that the pilA9-slr2019 mutant can still retract its pili, as cells are still transformable. It seems that this mutant assembles pili that are not able to attach to a surface, other cells or beads. In Myxococcus, a complex of minor pilins (including a PilX homolog) is involved in priming pilus extension which is essential for assembling pili on the surface (Treuner-Lange et al., 2020). As we still detect thick pili on the surface of the analyzed mutants, we assume that the minor pilins and hypothetical proteins analyzed in our study are not essential for the assembly of a general priming complex. However, we hypothesize that the minor pilins can be part of a pilus structure. Indeed, the minor pilin PilA4 from Synechocystis was detected outside of the cell in a PilA1 dependent manner (Cengic et al., 2018) and PilA2 and PilA11 have been detected at the cell surface via affibodies against PilA2 and immunolabeling of PilA11 (Cengic et al., 2018;Hu et al., 2018). Notably, a mutant of the major pilin gene pilA1 lacks thick pili on the surface but is still able to flocculate (Conradi et al., 2019). Therefore, it remains unclear whether the minor pilins which are involved in aggregation are part of the major pilus structure or can form a distinct pilus. Hu et al. (2018) reported that specific repression of pilA11 mRNA via the naturally transcribed antisense RNA PilR leads to thinner and fewer pili, whereas depletion of PilR leads to overexpression of PilA11 and the opposite effects on pili.
However, we were not able to measure significantly smaller diameters of the T4P in our mutant which lacks all minor pilins encoded in the ΔpilA9-slr2019 operon ( Figure S3). Generally, measuring precise diameters from TEM images is rather inaccurate and was not the aim of this study.
Whether the minor pilins analyzed in our study can be inserted into a pilus fiber remains unclear. Therefore, we attempted to model their structure. Due to poor inter-and intraspecies conservation of the amino acid sequences, we were only able to model PilA1,  Figure S11). These proteins all form the hydrophobic N-terminal helix ( Figure S11), which is the most conserved trait within pilins (Giltner et al., 2012) and thus is essential for the insertion of pilins into the fiber. We surmise, that at least PilA5, PilA6 and PilA10 can be incorporated into the pilus fiber like PilA4 (Cengic et al., 2018) due to the conservation of the hydrophobic handle. Moreover, secondary structure prediction ( Figure S12) indicated hydrophobic handles for all known Synechocystis pilins and thus, all might be incorporated.
In addition to the assumption that PilA5 is incorporated into the pilus fiber, our results suggested that this protein facilitates DNA binding. However, we were not able to detect typical DNA binding motifs (Luscombe et al., 2000) or a region with enriched positively ΔpilX3 (n = 13) mutants was determined. The deletion mutant Δhfq lacking T4P was used as a negative control. Transformations were performed with different amounts of DNA ranging from 500 ng to 1 µg and different plasmids conferring different antibiotic resistances charged amino acids in the PilA5 sequence, as was shown for the minor pilin ComP of Neisseria meningitidis (Berry et al., 2016). In this study, the researchers identified a region with a highly positively charged surface, which is probably involved in the electrostatic attraction of negatively charged DNA. For Thermus thermophilus, recent studies demonstrated that this naturally competent, gramnegative bacterium assembles two distinct T4P filaments composed of two different pilins. The wider filament is required for natural competence, whereas the more narrow one is used for twitching motility (Neuhaus et al., 2020). For Synechocystis, we did not obtain evidence from electron microscopy indicating that the different minor pilins form distinct pili ( Figure S3). Additionally, our analysis of the general amount of the major pilin PilA1 in sheared pili fractions did not indicate a major difference in the level of piliation of the minor pilin mutants in comparison to the WT (Figure 6). These findings are supported by exoproteome analyses, which showed that the major pilin PilA1 is considerably more abundant in the exoproteome than minor pilins (Sergeyenko & Los, 2000). Therefore, we suggest that the minor pilins might be incorporated within a filament that is primarily composed of PilA1.

Taton et al. (2020) recently discovered indispensable proteins
(minor pilin PilA3 S.e ., RntA S.e . and RntB S.e ) for natural transformation of the cyanobacterium Synechococcus elongatus PCC 7942. RntB S.e . and PilA3 S.e . exhibit modest homology to the PilA5 and PilA6 proteins of Synechocystis (e-values between 1 x 10 -04 and 4 x 10 -06 , Table S2). RntA S.e . shows homology to the PilX domain containing minor pilin PilX1 (Table S2). Notably, we also found a PilX-N domain in Slr2018 which is encoded by the pilA9-slr2019 gene cluster Nevertheless, a multiple sequence alignment shows that the hydrophobic regions of PilX1, PilX2 and PilX3 are much more conserved to each other than to the hydrophobic region of Slr2018 ( Figure 1b).
Generally, PilX amino acid sequences are longer than those of the major pilin which is also true for the Synechocystis PilX-like proteins ( Figure 1b). Although PilX from Pseudomonas lacks glutamate at position +5 (E +5 ) of the PilD cleavage site, it was shown to be incorporated into the pilus fiber (Giltner et al., 2010). Pilins with substitution of E +5 to V +5 can still be cleaved and methylated (Strom & Lory, 1991). Moreover, it was hypothesized that pilins with a nonpolar substitution have an improved ability to leave the membrane during assembly (Giltner et al., 2012). Thus, these proteins were proposed to form the pilus tip complex. This property was shown for the minor pilin GspK of the type II secretion complex of E. coli,

Natural competence Flocculation
WT Note: Phenotypes of major and minor pilin mutants which have been analyzed in this and other studies. a Grigorieva and Shestakov (1982).
b Bhaya et al. (1999). F I G U R E 1 0 Intracellular cyclic nucleotide messenger concentrations in planktonic or sessile-grown Synechocystis WT cells. A planktonic preculture was cultivated either under planktonic or sessile conditions, and cAMP (a) and c-di-GMP (b) concentrations were determined after 10 min, 1, 4 and 8 hr. The bars display the mean concentration of cyclic nucleotide concentration normalized to whole cell protein amount. Error bars represent standard deviation. Experiments were performed as biological and extraction technical replicates. Significance was tested using a two-tailed Student's t test. * p ≤ .05, ** p ≤ .01, *** p ≤ .001, n.s., not significant which also lacks E +5 (Korotkov & Hol, 2008), and a similar PilX minor pilin found in Acinetobacter baumannii T4P (Piepenbrink, 2019). The three Synechocystis PilX homologues also have different hydrophobic amino acid substitutions at position E +5 (Figure 1b). Additionally, secondary structure predictions indicated the existence of hydrophobic N-terminal helices, which are characteristic of minor pilins and are vital for insertion into the pilus ( Figure S12). At the moment we are not able to predict where the Synechocystis minor pilins are positioned within the pilus fiber. new sulfated exopolysaccharide, called synechan has been described which is essential for the formation of bloom-like cell aggregates (Maeda et al., 2021). The genes encoding proteins involved in synechan synthesis and regulation, all but one are located on the pSYSM plasmid in a large cluster (sll5042-sll5060). Several of these mRNAs represent the most differentially regulated mRNAs in our microarray analysis (Table 2). Most interestingly, the respective mRNAs accumulate strongly in the planktonic culture when compared to the cells on a surface, similar to the pilA9-slr2019 and pilX2 mRNAs. Thus, the exopolysaccharide synechan as well as this specific set of minor pilins are expressed under the same conditions and are essential for cell-cell interaction. Therefore, it is tempting to speculate that these minor pilins can bind specifically the sulfated exopolysaccharide to form together bloom-like cellular assemblages or flocs. Alternatively, T4P pili could also self-interact as it was shown, e.g., for the DNA-

| Bacterial strains and culture conditions
The WT of Synechocystis sp. PCC 6803 employed in this study is motile and can grow photoautotrophically, mixotrophically and chemoheterotrophically on glucose. This strain was originally obtained from S. Shestakov (Moscow State University, Russia) in 1993 and was re-sequenced in 2012 (Trautmann et al., 2012).

| Construction of mutant strains
The deletion mutant ΔpilA5-pilA6 was created as described in Wallner et al. (2020). Three plasmids were constructed bearing either the single gene pilA5 or pilA6 alone or the operon pilA5-pilA6 under the control of the native promoter (P pilA5-pilA6 ). For a list of oligonucleotides used to generate all mutant constructs, see Table S3.
The gene fragments P pilA5-pilA6 -pilA5 and P pilA5-pilA6 -pilA5-pilA6 were amplified from WT chromosomal DNA with an SdaI recognition site at the 5ʹ-end (P1-P3) and inserted into the pJET 1.2 vector (CloneJET PCR Cloning Kit, Thermo ScientificTM, Germany). For the third construct, assembly cloning (AQUA cloning) (Beyer et al., 2015) was used to fuse the pJET 1.2 vector with the P pilA5-pilA6 fragment containing the SdaI recognition site at the 5ʹ-end and the pilA6 gene fragment (P4-P9). The subcloning vectors were cleaved with SdaI and HindIII restriction enzymes, and fragments were ligated into the pVZ322 vector (NCBI accession number AF100175), which was cleaved with the same restriction enzymes. The final plasmids were transferred into the Synechocystis deletion mutant ΔpilA5-pilA6 via triparental mating (Elhai & Wolk, 1988) with E. coli J53 (NCBI:txid1144303) harbouring the conjugative helper plasmid RP4 (NCBI:txid2503) and E. coli DH5α with the constructed plasmid. Verification of correct mutant strains was validated by PCR using primer pair P10-P11 ( Figure S1). A list of all plasmids and strains used in this study can be found in Tables S4 and S5. A Synechocystis knockout mutant of the slr5087-slr5088 operon located on the pSYSM plasmid was generated ( Figure S1d). First, we amplified chromosomal regions to be used as homologous recombination sites and added a NsiI restriction site to homologous region 1 (HR1) and overlapping regions using primer pairs P12-P13 and P14-P15. A combined homologous region sequence was obtained by overlap PCR using the two amplified DNA fragments and primer pair P12-P15. The construct was inserted into the pJET1.2 vector according to the manufacturer's instructions (CloneJET PCR Cloning Kit, Thermo ScientificTM, Germany). Furthermore, a chloramphenicol resistance cassette was amplified from the pACYC184 vector (GenBank: X06403.1), and NsiI restriction sites were added (P16-P17). This amplification product and the vector were cleaved by NsiI and ligated using T4 DNA ligase (New England BioLabs, Inc.).

Insert orientation was checked by sequencing. Finally, Synechocystis
WT was transformed with the plasmid. DNA of the mutants was extracted from cells by three repeated cycles of shock freezing (−80℃) and heating (60℃) for 10 min each. DNA was separated from cell debris by centrifugation at 3,000g for 3 min at 4℃. Correct construct insertion and full segregation were tested by colony PCR using the primer pair P18-P19 ( Figure S1e).

| Flocculation assays
Flocculation assays were mainly performed as described in Conradi et al. (2019). Briefly, cultures were diluted to an OD 750nm 0.25, and 6 ml were transferred to 6-well plates (Corning Costar®, non-

| Transformation assays
Natural transformation competence was tested with a suicide plasmid encoding streptomycin resistance. Cells were grown to log phase (OD 750nm 0.6 -0.8) in planktonic culture, and 15 ml were harvested by centrifugation (3,200g,RT). The cell pellet was resuspended in 300 µl BG11 medium. Then, 500 or 1,000 ng of plasmid DNA were added to the suspension, and cultures were incubated at 30℃ under white light (≈ 50 µmol photons m −2 s −1 ) without shaking. After three hours, cells were plated on sterile filters (0.2 µm pore size, MN615, Macherey-Nagel), placed on BG11 agar plates and incubated under standard conditions. After two days, the filters were transferred to new BG11 agar plates supplemented with 5 µg/ml streptomycin and incubated further until colonies appeared. Alternatively, instead of transferring the cells to the filter, they were streaked directly on an BG-11 agar plate. Then, a small filter paper (ø 2.9 cm) was placed in the middle of the plate and 25 µl of a streptomycin solution (25 µg/ ml) were added onto the filter to create a concentration gradient of the antibiotic. Plates were incubated as described above. Integration of the construct used led to deletion of the sRNA gene psrR1, which had no phenotypical effect under normal cultivation conditions (Georg et al., 2014).

| Pili quantification
Cells were plated on standard BG11 plates supplemented with 0.2% glucose and grown under standard growth conditions (see above).

| Fluorescent bead assay
Fluorescent bead assays were performed as described in Nakane and Nishizaka (2017) with slight changes. The coverslip was prepared as described by Nakane and Nishizaka (2017)

| Phototaxis experiments
Phototactic movement of cyanobacterial cells was analyzed on 0.5% agar plates containing BG11, 10 mM (w/v) TES buffer (pH 8.0), 0.3% (w/v) sodium thiosulphate and 11 mM glucose as previously described by Jakob et al. (2017). Cells were concentrated in a small volume of BG11, and 7 µl of cell suspension were spotted as droplets in triplicate on the phototaxis plate. Plates were then illuminated with unidirectional white light (≈5 µmol photons m −2 s −1 ) and incubated for 6-7 days at 30℃. Pictures were taken with a flatbed scanner.  Pinto et al. (2009) with modifications as described in Wallner et al. (2020).

| Microarray analysis
Removal of potential DNA from the RNA samples was performed using Ambion TurboDNase (Thermo Fisher Scientific, Germany) according to the manufacturer's instructions, but addition and incubation with Turbo DNase were performed twice. RNA was precipitated overnight with 3 M sodium acetate (pH 5.2) in 97% (v/v) ethanol.
Furthermore, RNA was collected by centrifugation at 20,800g for 30 min, washed with 70% ethanol and eluted in nuclease-free water. The concentration was checked on a NanoDrop ND2000 spectrophotometer (Thermo Fisher Scientific, Germany), and integrity was verified using a fragment analyser (Advanced Analytical in data sets S1-S5.

| Northern blot hybridization
Northern blot hybridizations were performed. RNA was extracted as described above. Seven or ten micrograms of RNA was separated on a denaturing 1.3% (w/v) agarose-formaldehyde electrophoresis gel and blotted onto a Roti-Nylon plus membrane (Carl Roth, Germany). The Northern blot was hybridized with probes generated by in vitro transcription of PCR fragments (P53-P60,

| Extraction of nucleotide second messengers
Cells for planktonic and sessile cultures were grown the same way as described for RNA extraction. To obtain measurable amounts of second messengers for sessile cultures, cells from two plates were scratched and suspended in 250 µl BG11 medium. Cells from 50 µl of this suspension were pelleted at 11,000g at 4℃ for 2 min for nucleotide extraction, and 50 µl were pelleted and used for determination of protein concentration. For planktonic samples, 10 ml of a liquid cell culture were pelleted for nucleotide and 1 ml for protein quantification. Cell pellets were resuspended in 300 µl extraction solution (acetonitrile/methanol/ water 2:2:1 [v/v/v]), incubated at 4℃ for 15 min and heated to 95℃ for 10 min. Samples were snap-cooled on ice and stored for further use at −20℃ or centrifuged at 21,000g at 4℃ for 10 min, and the supernatant was collected. The extraction was repeated twice from the pellets with 200 µl extraction solution each, without heat treatment. Supernatants were combined, and proteins precipitated by incubation at −20℃ overnight. The samples were centrifuged at 21,000g at 4℃ for 10 min, and the supernatant containing the extracted second messengers was air dried in a SpeedVac at 42℃. Quantification of c-di-GMP and cAMP was performed by HPLC/MS/MS analysis, as previously described (Burhenne & Kaever, 2013). Second messengers were normalized to the total protein amount. For protein quantification, cell pellets were resuspended in 50 µl PBS, supplemented with 0.7 volume of glass bead mix (0.1-0.11 and 0.25-0.5 mm) and vortexed for 60 s.
Then, the samples were snap-cooled at −80℃ twice and heated to 40℃ for 10 min each. After sedimentation of the beads, 2 µl of the supernatant were used for protein quantification using the Direct Detect system (Merck Millipore).

| Bioinformatic analysis
All bioinformatic tools used are listed in Table S6.

ACK N OWLED G EM ENTS
We thank D. Nakane (Gakushuin University), who established the bead assay in our lab, and A. Jakob (University of Freiburg) for performing the assays together with S. Oeser. We acknowledge V.

CO N FLI C T O F I NTE R E S T
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

AUTH O R CO NTR I B UTI O N S
SO performed experiments, analyzed and interpreted the data, wrote the manuscript and contributed to the design of the study.
TW performed experiments, interpreted the data and contributed to the design of the study. NS performed experiments, interpreted the data and contributed to the design of the study. LB performed experiments and analyzed data. SS performed experiments with SO and analyzed the data. SVA interpreted the data and contributed to the design of the study. AW contributed to the design of the study, interpreted data and wrote the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available