Effects of PatU3 Peptides on Cell Size and Heterocyst Frequency of Anabaena sp. Strain PCC 7120

ABSTRACT

patU, one of the genes specifically found in filamentous cyanobacteria, is required for the pattern formation in heterocyst-forming species. In Anabaena sp. strain PCC 7120, patU is split into patU5 and patU3, and only patU3 is involved in heterocyst patterning. Here, we report that PatU3 is also involved in control of cell size. A patU3 deletion mutant showed remarkably smaller cell size and much higher heterocyst frequency than the wild type. Yeast two-hybrid and pulldown assays demonstrated a direct interaction between PatU3 and the cell division protein Ftn6. Without the N-terminal 16-amino-acid (aa) portion (MQERFQAVIKRRLQIH [the identified octapeptide is underlined]), PatU3 was no longer able to interact with Ftn6. This portion of PatU3 is also required for the interaction with PatN, a protein related to heterocyst differentiation/patterning. Addition of the 16-aa peptide or AVIKRRLQ-containing peptides restored the cell size and heterocyst frequency of a patU3 deletion mutant to normal or nearly wild-type levels. PatU3(1-16aa)-GFP, the N-terminal 16-aa sequence fused with green fluorescent protein (GFP), formed polar aggregates and peripheral patches in heterocysts of Anabaena sp. strain PCC 7120, whereas PatU3(1-198aa)-GFP showed a homogeneous distribution in the cytoplasm of all cells. The N-terminal AVIKRRLQ-containing sequence may function in intact PatU3, as a separate peptide, or both.

IMPORTANCE PatU (or split into PatU5 and PatU3) is distributed in almost all filamentous cyanobacteria, including those that do not form heterocysts (except Pseudanabaena); however, its functions other than heterocyst differentiation/patterning have not been reported before. In this study, we found that PatU3 in Anabaena sp. strain PCC 7120 is involved in cell size determination. The N-terminal 16-aa sequence of PatU3 is required for the control of cell size and interaction with the cell division protein Ftn6, and an octapeptide (aa 7 to aa 14) within the 16-aa sequence can restore the cell size (and heterocyst frequency) of a patU3 deletion mutant to normal. Such a peptide, if generated from PatU or PatU3 in vivo, may promote intercellular coordination in filamentous cyanobacteria.

INTRODUCTION

Cyanobacteria, the oxygen-evolving photosynthetic prokaryotes, are widely distributed in different types of ecological systems, such as the ocean, inland water bodies, and soil/rock surfaces, in all climate zones on earth. Morphologically, cyanobacteria can be unicellular, colonial, or (branched/unbranched) filamentous (1). Among filamentous cyanobacteria, a large group of phylogenetically related species can form heterocysts, specialized cells that fix N₂ under aerobic conditions. These N₂-fixing cells provide spatial separation of nitrogen fixation from photosynthesis, keeping nitrogenase from damage by oxygen (2). Anabaena sp. strain PCC 7120 (also called Nostoc sp. strain PCC 7120; here called Anabaena 7120) is a model strain for studies on heterocyst formation. It produces semiregularly spaced heterocysts along unbranched filaments under nitrogen-depleted conditions.

Some genes, such as hetR (3–6), hetZ-patU (4, 7), and patS or patX (here called patS/patX) (5, 8), are found specifically in filamentous cyanobacteria. In heterocyst-forming species, these genes play core regulatory roles in cell differentiation/patterning (9). In Anabaena 7120, patU is split into patU5 (alr0100) and patU3 (alr0101), and only patU3 is required for heterocyst pattern formation (7). Inactivation of patU3 in Anabaena 7120 leads to the phenotype of multiple contiguous heterocysts (Mch) under nitrogen-depleted conditions. Furthermore, PatU3 interacts with HetZ and modulates the expression of hetR and patS (9). Unlike patU3, patU5 is not required for heterocyst differentiation/patterning (7). Because the homologs of hetR, hetZ, and patU and counterparts of patS/patX are essentially universal in filamentous species (except Pseudanabaena) (4, 5), heterocyst-forming or not, it is postulated that the presence of these genes predated the origin of heterocysts in evolution. Therefore, they must be involved in functions other than control of heterocyst differentiation/patterning.

Cell size is one of the basic characteristics of bacterial species. Bacterial cell size is controlled by the rates of cell expansion and cell cycle progression, depending on nutrient availability, biosynthetic capacity, initiation of chromosomal replication, and the status of cell division (10, 11). Lowered nutrient availability and biosynthetic capacity reduce the cell size, while depletion of protein factors involved in DNA replication and septum formation, such as DnaA and FtsZ, leads to enlargement of cells. In some cases, cell size is also determined by asymmetric cell division prior to cell differentiation (12, 13).

Bacterial cell division is a multistep process, including at least two major steps: (i) FtsZ ring formation and determination of cell division site and (ii) synthesis of septal peptidoglycan (10, 14, 15). Among the cell division proteins so far identified, Ftn6 is one of those unique to cyanobacteria. Inactivation of ftn6 led to the elongation (in one dimension) of cells in Synechococcus elongatus strain PCC 7942 (Synechococcus 7942) (16, 17) and Anabaena 7120 (16) and to the enlargement (in three dimensions) of cells in Synechocystis sp. strain PCC 6803 (Synechocystis 6803) (18). Ftn6 localizes to the septum at midcell and interacts physically with FtsZ (18). In Synechococcus 7942, Ftn6 is required for FtsZ ring formation (17); however, in Synechocystis 6803, Ftn6 is only required for proper localization of the FtsZ ring (18). Because its N-terminal portion is predicted to be a DnaD-like domain, it may also play a role in coordination of replication initiation and cell division (19).

In our studies of patU3, we found that patU3 deletion mutants showed smaller cell size than the wild type. Further investigations demonstrated that PatU3 interacts with Ftn6 and that the interaction depends on the N-terminal 16-amino-acid (aa) portion of PatU3. Most interestingly, an 8-aa sequence within this portion was sufficient to restore the cell size of a patU3 deletion mutant to normal and greatly reduced the heterocyst frequency of the mutant.

RESULTS

Interaction of PatU3 with the cell division protein Ftn6.

Compared to the wild type (WT), the patU3::C.K4 mutant (called MTpatU3-Km, with base pairs [bp] 37 to 641 of the 777-bp-long patU3 coding region replaced by C.K4) shows much higher heterocyst frequency after nitrogen step-down, producing multiple contiguous heterocysts (Mch). In addition, most cells of the patU3 mutant are smaller than wild-type cells under nitrogen-replete and -depleted conditions (Fig. 1). We measured x axis (parallel to filament) and y axis (perpendicular to filament) sizes of the cells (both vegetative cells and heterocysts). In each case, the peak size distribution of the mutant shifted to the smaller value side relative to that of the wild type (Fig. 2), and the difference between the two strains was statistically significant (see Fig. S1 in the supplemental material for all cells and Fig. S2 for heterocysts) (two-tailed t test, P < 0.01).

FIG 1.

Photomicrographs of the wild-type (WT) and patU3 mutant strains of Anabaena 7120. Heterocyst frequencies and percentages of multiple contiguous heterocysts (Mch) are indicated. +N, cultured in BG11; −N 24h, 24 h after nitrogen step-down. Arrowheads point to heterocysts. The scale bar applies to all panels. Het. freq., heterocyst frequency.

FIG 2.

Cell size distributions of Anabaena 7120 and patU3 mutants under nitrogen-replete and -depleted conditions. For each sample, 1,000 cells were measured. (A) x axis (parallel to filament) size; (B) y axis (perpendicular to filament) size.

Cell division is one of the important aspects for determination of cell size. We wondered whether the effect of PatU3 on cell size is associated with a protein involved in cell division. Using a yeast two-hybrid system, we tested the interactions between PatU3, predicted cell division proteins in Anabaena 7120, and identified Ftn6 as the candidate (Fig. 3A; Fig. S3). Using pulldown assays, we confirmed the PatU3-Ftn6 interaction (Fig. 3B). In addition, we tested the interactions between PatU3 and some proteins involved in heterocyst differentiation/pattern formation and identified HetZ (Fig. S3) (9) and PatN (Fig. S4). PatN is a protein that may be involved in heterocyst pattern formation (20). PatU3 also showed interaction with itself (Fig. S3 and S5A and C); therefore, it may form dimers or multimers in Anabaena cells.

FIG 3.

Yeast two-hybrid and pulldown assays showing the PatU3-Ftn6 interaction. (A) Yeast two-hybrid assays. pGBK-Lam (lamin C) + pGADT7-T (T-antigen), the negative control; pGBKT7-53 (p53) + pGADT7-T, the positive control; mut1 to mut16, mutated versions of patU3 with consecutive and nonoverlapping 15-codon (mut1, with the start codon untouched) or 16-codon (mut2-mut16) deletions. The 16 deletions within PatU3, a 258-aa protein, are Δaa_2-16, Δaa_17-32, Δaa_33-48, etc. (B) SDS-PAGE (left) and Western blot (right) analyses of the pulldown assay. Anti-HA monoclonal antibody was used for Western blot detection. 1, EF-Ts(HA)-Ftn6; 2, MBP·Bind resin + MBP-PatU3 + EF-Ts(HA)-Ftn6; 3, MBP·Bind resin + MBP-PatU3 + EF-Ts(HA); 4, MBP·Bind resin + MBP + EF-Ts(HA)-Ftn6; 5, These should not be in bold face. Please correct to light face.

ftn6 was first identified in Synechococcus 7942, and inactivation of ftn6 led to elongation of all cells in this strain (16, 17). In an ftn6 (all1616) mutant of Anabaena 7120, some cells were elongated or enlarged, while most cells remained unchanged (16). To confirm the effect of Ftn6 on cell size in Anabaena 7120, we generated the Ω-P_petE-ftn6 strain (called the P_petE-ftn6 strain), in which Ω-P_petE was inserted immediately upstream of the coding region of ftn6, so that the only copy of ftn6 was expressed from P_petE, a cupric ion-responsive promoter. Stem-loop structures at both ends of the Ω cassette are designed to terminate background transcription (21). Downstream of ftn6 is a gene oppositely oriented (and, therefore, exhibits no polar effect). The mutant cells were grown in BG11 with Cu²⁺ and then transferred to Cu²⁺-free BG11 as previously described (22) (see Materials and Methods). In BG11 with Cu²⁺, the mutant showed no difference in cell size from the wild type; in Cu²⁺-free BG11, most cells were significantly elongated (Fig. 4 and 5; Fig. S6). A reverse transcriptase quantitative PCR (RT-qPCR) analysis showed that the relative mRNA level of ftn6 was reduced by 96% after removal of Cu²⁺ (from 1.00 ± 0.10 to 0.04 ± 0.01). The removal of Cu²⁺ also decreased heterocyst frequency at 24 h after nitrogen step-down, but the heterocyst frequency gradually increased to the wild-type level in the next 24 h (Fig. S7).

FIG 4.

Photomicrographs of Anabaena 7120 and the P_petE-ftn6 strain in BG11 with (thereby inducing P_petE) or without (switching off P_petE) cupric ion.

FIG 5.

Cell size distributions of Anabaena 7120 and the P_petE-ftn6 strain in Cu²⁺-free BG11 medium (switching off P_petE). For each sample, 2,000 cells were measured. (A) x axis size; (B) y axis size.

Role of the N-terminal portion of PatU3 in interaction with Ftn6 and control of cell size.

Since PatU3 evidently interacts with Ftn6, we wondered which portion of PatU3 is required for the interaction. Using a yeast two-hybrid system, we explored the interaction using mutated versions of PatU3 with consecutive and nonoverlapping 15-codon or 16-codon deletions (Fig. 3). Without the N-terminal 15-aa sequence (next to the methionine residue encoded by the start codon), PatU3 no longer interacted with Ftn6. The same sequence was required for the interaction with PatN (Fig. S4A) and possibly was partially required for the interaction with HetZ (Fig. S5B and data not shown; hetZ+mut1 reproducibly showed ca. 24-h delay in growth compared to other two-hybrid combinations) but not for the self-interaction of PatU3 (Fig. S5A).

To examine the role of the N-terminal portion of PatU3 in regulation of cell size and heterocyst frequency, we generated an in-frame mutant of Anabaena 7120 in which the N-terminal 15 codons immediately following the start codon of patU3 were deleted, called the MTpatU3del4-48 mutant (nucleotides 4 through 48 were deleted). Under both nitrogen-replete and -depleted conditions, the MTpatU3del4-48 mutant was similar in cell size to the MTpatU3-Km mutant (Fig. 1 and 2). The heterocyst frequency of the MTpatU3del4-48 mutant was much lower than that of the MTpatU3-Km mutant but higher than that of the wild type (t test; P < 0.01); the MTpatU3del4-48 mutant also showed an Mch phenotype, although quantitatively at a much lower level than the MTpatU3-Km mutant (t test; P < 0.01) (Fig. 1).

Effects of N-terminal peptides of PatU3 on the cell size and heterocyst frequency of a patU3 deletion mutant.

Because the MTpatU3del4-48 strain showed the same cell size as the MTpatU3-Km strain, the role of PatU3 in regulation of cell size may be completely dependent on the N-terminal 16-aa sequence (including the methionine residue encoded by the start codon). To test this hypothesis, we constructed pHB6143, a pDU1-based plasmid (capable of replication in both Anabaena and Escherichia coli) that expresses the N-terminal 16-aa portion of PatU3 from P_patU5 (retaining patU5) and introduced the plasmid into MTpatU3-Em, which is identical to the MTpatU3-Km strain except for the replacement of C.K4 with C.CE2. The presence of pHB6143 lowered the heterocyst frequency in the patU3 mutant from more than 2 times higher than that of the wild type to significantly lower (Fig. 6). The plasmid also brought the cell size of the mutant up from significantly below that of the wild type to nearly wild-type levels (Fig. 7; Fig. S8). We also complemented the MTpatU3-Em mutant with pHB6166 and pHB6371, two pDU1-based plasmids that carry P_ntcA-patU3 (1-16 aa) and P_ntcA-patU3 (full length), respectively. Compared to pHB6143, pHB6166 showed similar effects on the heterocyst frequency (Fig. 6) and cell size (Fig. 7 and 8; Fig. S8) of the mutant, while pHB6371 brought the cell size up to a similar level (Fig. 7 and 8; Fig. S8) but decreased the heterocyst frequency to a much lower level (Fig. 6).

FIG 6.

Effects of PatU3 peptides on the heterocyst frequency (percent) of the MTpatU3-Em strain. Sequences of PatU3 peptides are listed to the left of the chart. −N 24 h, 24 h after nitrogen step-down; −N 48 h, 48 h after nitrogen step-down. Asterisks indicate significant differences from the value of MTpatU3-Em at 24 h or 48 h (t test; P < 0.01).

FIG 7.

Photomicrographs of Anabaena 7120 and MTpatU3-Em with or without the octapeptide PatU3N8-VI and the complemented strains MTpatU3-Em(pHB6143), MTpatU3-Em(pHB6166), and MTpatU3-Em(pHB6371). Cells were cultured in BG11. pHB6143, pHB6166, and pHB6371, plasmids that carry P_patU5-patU5-patU3 (1-16 aa), P_ntcA-patU3 (1-16 aa), and P_ntcA-patU3, respectively.

FIG 8.

Cell size distributions of Anabaena 7120 and MTpatU3-Em with or without the octapeptide PatU3N8-VI and MTpatU3-Em complemented with plasmids, as shown in Fig. 7. For each sample, 2,000 cells were measured. (A) x axis size; (B) y axis size.

We have previously shown that the addition of a 13-aa peptide, PatS-13, to the medium can inhibit target gene expression in Anabaena 7120 (23). The addition of N-terminal peptides of PatU3 to the medium may restore the cell size of the MTpatU3-Em mutant to normal as well. Because heterocyst frequency was more easily analyzed than cell size, we first tested the effects of PatU3 peptides on heterocyst frequency. PatU3 peptides were added to cultures of the patU3 mutant before and after nitrogen step-down at a final concentration of 10 μM (see Materials and Methods). As shown in Fig. 6, PatU3N16 (the N-terminal 16-aa peptide) and PatU3N15 (the 15-aa peptide without the methionine residue) both reduced the heterocyst frequency to nearly wild-type levels; PatU3N13, without the two amino acid residues at the C-terminal end of PatU3N15, showed a similar result. We then tested all 6 octapeptides (PatU3N8-I to PatU3N8-VI) within the 13-aa sequence. Only PatU3N8-VI (AVIKRRLQ) reduced the heterocyst frequency to nearly the wild-type level. No septapeptide or hexapeptide derived from AVIKRRLQ showed any such effect on the heterocyst frequency. Based on this, we further analyzed the effect of PatU3N8-VI on cell size (Fig. 7; Fig. S8). Addition of PatU3N8-VI to the medium restored the cell size of the MTpatU3-Em mutant to nearly normal. The effects of PatU3N8-VI on the MTpatU3-Em mutant were very similar to those of plasmid pHB6143 on the same strain. However, PatU3N8-VI showed no effects on the wild type with respect to cell size (Fig. 7; Fig. S9) and heterocyst frequency (not shown). It should be noted that the MTpatU3-Em and MTpatU3-Km deletion mutants lack the 13- to 214-aa region of PatU3, in particular the LQ of the octapeptide (AVIKRRLQ).

Localization of PatU3-GFP fusions.

In light of the dependence of PatU3-Ftn6 and PatU3-PatN interactions on the N-terminal 16-aa sequence of PatU3 and the cellular effects of N-terminal peptides, we investigated how this portion would be localized in Anabaena cells if it is fused to green fluorescent protein (GFP) with or without the rest of PatU3.

We constructed pDU1-based plasmids, pHB6141 and pHB6142, to express the N-terminal 16-aa sequence and 192-aa sequence, respectively, of PatU3 (full length, 258 aa) as translational fusions with GFP in Anabaena 7120. The fused genes were expressed from P_patU5 (retaining patU5), and the copy numbers of the two plasmids (4.75 ± 0.01 and 3.30 ± 0.19 relative to rnpB in chromosomal DNA) were comparable to each other. Before nitrogen step-down (0 h), PatU3(1-16aa)-GFP encoded by pHB6141 was occasionally expressed in some vegetative cells; after 12 h of nitrogen deprivation, it was expressed in differentiating cells and heterocysts (Fig. 9; Fig. S10). In contrast, PatU3(1-192aa)-GFP, encoded by pHB6142, was expressed in all cells weakly at 0 h, 12 h, and 48 h but moderately at 24 h (Fig. 9; Fig. S10). We should note that when expressed from the high-copy-number plasmids, PatU3(1-16aa)-GFP delayed heterocyst differentiation in Anabaena 7120 (significantly higher percentages of proheterocysts in the strain with pHB6141 than that in the wild type) (Fig. 9A and C), while PatU3(1-192aa)-GFP inhibited differentiation (Fig. 9A).

FIG 9.

Expression of PatU3(1-16aa)-GFP and PatU3(1-192aa)-GFP in Anabaena 7120 at 24 h after nitrogen step-down. pHB6141, the plasmid for expression of PatU3(1-16aa)-GFP; pHB6142, the plasmid for expression of PatU3(1-192aa)-GFP. Means ± SD are relative copy numbers of the plasmids. Arrowheads point to heterocysts and proheterocysts. (A) Light (I) and GFP fluorescence (II) photomicrographs of wild-type (WT) Anabaena 7120 and derivative strains that express the fusion proteins. (B) The structure of PatU3(1-16aa)-GFP and PatU3(1-192aa)-GFP. (C) The effect of PatU3(1-16aa)-GFP on heterocyst differentiation. Heterocyst frequency is the percentage of (heterocysts + proheterocysts) among all cells; the percentage of proheterocysts is [proheterocysts/(heterocysts + proheterocysts)] × 100. Asterisks indicate significant differences from the wild type (t test; P < 0.01).

In addition to the expression pattern along filaments, subcellular localizations of PatU3(1-16aa)-GFP and PatU3(1-192aa)-GFP were dramatically different from each other. Under a fluorescence microscope, PatU3(1-16aa)-GFP was found most often at the polar and subpolar regions of cells, while PatU3(1-192aa)-GFP was homogenously distributed in the cytoplasm (Fig. 9; Fig. S10); in a small number of cells, PatU3(1-16aa)-GFP was distributed in the whole cytoplasm (Fig. S10). Using confocal laser scanning microscopy, we further observed that PatU3(1-16aa)-GFP formed polar aggregates and peripheral dots in heterocysts (Fig. 10A). Under a superresolution confocal microscope, the peripheral dots of GFP fluorescence were found to be irregularly shaped flat patches (Fig. 10B).

FIG 10.

Localization of PatU3(1-16aa)-GFP in heterocysts of Anabaena 7120(pHB6141) at 24 h after nitrogen step-down by confocal laser scanning microscopy. Arrows point to heterocysts. (A) A filament imaged on a Leica TCS SP8 microscope. I, bright-field image; II, autofluorescence image; III to X, successive GFP fluorescence images at different focal depths. (B) Two heterocysts (I and II) imaged on a Zeiss LSM 880 microscope.

We also constructed pHB5563, a plasmid similar to pHB6141 and pHB6142 but with GFP translationally fused to the full-length PatU3. A plasmid called pHB1466 (7), with gfp transcriptionally fused to P_patU5-patU5-patU3, was used as a control. Anabaena 7120(pHB1466) and Anabaena 7120(pHB5563) showed heterocyst differentiation at 24 h after nitrogen step-down (Fig. S11), although heterocyst differentiation was slightly delayed in the strain carrying pHB5563. The full-length PatU3-GFP translational fusion encoded by pHB5563 was weakly expressed in all cells, while gfp cotranscribed with patU5-patU3 in pHB1466 was strongly expressed in heterocysts and proheterocysts (Fig. S11). Relative to the great difference between the two strains, the copy numbers of pHB1466 (2.76 ± 0.05) and pHB5563 (1.88 ± 0.07) are comparable to each other. A combination of the results shown with Anabaena strains harboring pHB6142 or pHB5563 indicated that the full-length PatU3 (or the main portion, if processed) is nonspecifically distributed in all cells along the filaments.

DISCUSSION

As one of the important players in heterocyst patterning (7), PatU3 interacts with HetZ and modulates the expression of hetR and patS (9). In this report, our new evidence indicates that PatU3 also interacts with Ftn6, a cell division protein, and that an 8-aa sequence in its N-terminal portion is involved in the control of cell size.

The role of PatU3 in cell size determination was first noted due to the reduced cell size of a patU3 mutant of Anabaena 7120 compared to the wild type. Using a yeast two-hybrid system, we systematically tested cell division proteins and identified Ftn6 as one that interacts with PatU3. Deletion of patU3 in Anabaena 7120 reduced the cell size (in three dimensions), whereas switching off ftn6 in the P_petE-ftn6 mutant led to cell elongation (in one dimension). In certain cyanobacterial species, such as Synechococcus 7942, mutations in ftn6, ftn2, ftsZ, minE, and cdv1 all lead to cell elongation rather than enlargement (16, 17). Anabaena 7120 apparently belongs to this category, even though some enlarged cells were also found in ftn6 mutants (16) (see Fig. S12 in the supplemental material). One explanation for our results is that PatU3 regulates the cell size through an interaction with Ftn6. However, it is also possible that multiple pathways exist. Due to the complicated mechanisms for control of cell size, genetic manipulations of patU3, ftn6, or a combination of them are not sufficient to give a clear answer at this stage. Nonetheless, the effect of PatU3 on cell size in Anabaena 7120 is conclusive. Therefore, we further investigated which portion of PatU3 is involved in this effect.

Based on yeast two-hybrid assays, the N-terminal 16-aa sequence of PatU3 was shown to be required for the interaction with Ftn6. Deletion of the 15 codons following the start codon of patU3 in Anabaena 7120 (MTpatU3del4-48) showed the same effect on cell size as the deletion of most of the patU3 sequence (MTpatU3-Km and MTpatU3-Em mutants); addition of the N-terminal 16-aa peptide or plasmids expressing the peptide to the MTpatU3-Em mutant restored the cell size to normal. Compared to MTpatU3-Km (26.5% ± 0.9%), MTpatU3del4-48 mutant (13.3% ± 0.5%) was much closer to the wild type (10.1% ± 0.4%) in heterocyst frequency. Apparently, the N-terminal 16-aa region makes a low but evident contribution to the role of PatU3 in the control of heterocyst frequency. At the same time, it implies that PatU3 (without 2 to 16 aa) was produced in the MTpatU3del4-48 strain at a level comparable to that of PatU3 in the wild type (otherwise, the heterocyst frequency of the MTpatU3del4-48 mutant should be close to that of the MTpatU3-Km mutant). These results, taken together, indicate that the role of PatU3 in the control of cell size depends on this 16-aa region, while the role in the control of heterocyst frequency depends, to a much lower extent, on the same region. AVIKRRLQ, within this 16-aa region, was shown to be the minimum sequence for an exogenous peptide to restore the cell size to normal. Because PatU3 might have been expressed in the wild-type strain at a saturating level (alternatively, a further increase in cell size must be limited by some other factors), it was not surprising that the PatU3 octapeptide showed no effect on the cell size of the wild type.

The 15-codon deletion in patU3 slightly increased the heterocyst frequency in Anabaena 7120; however, the AVIKRRLQ-containing peptides restored the heterocyst frequency of the MTpatU3-Em mutant to nearly wild-type levels. We postulate that two or more regions of PatU3 are separately involved in heterocyst pattern formation, but the lack of the other portion(s) can be partially compensated for by excess N-terminal peptides. Therefore, addition of AVIKRRLQ-containing peptides, or expression of the N-terminal 16-aa peptide on a pDU1-based plasmid, largely complemented the patU3::C.CE2 mutation. Because the N-terminal sequence is also required for the interaction with PatN, and possibly HetZ, the effect of this portion on heterocyst frequency is not necessarily associated with Ftn6 and cell division.

In Anabaena 7120, patU3 is specifically transcribed in heterocysts and proheterocysts (7) (Fig. S11); however, the deletion of patU3 reduced the sizes of all cells (Fig. 1). Presumably, the background expression of this gene in vegetative cells could satiate the requirement for control of cell size. The PatU3(1-192aa)-GFP fusion protein (and, similarly, the full-length PatU3 fused to GFP), expressed under the control of the native promoter, was shown to be nonspecifically distributed in cells along filaments before and after nitrogen step-down. This is quite consistent with the effect of PatU3 on the sizes of all the cells, but the following question arises: why is the distribution of the fusion protein along filaments so different from what the transcription activity shows? Posttranslational regulation could be an answer. Unlike PatU3(1-192aa)-GFP, PatU3(1-16aa)-GFP expressed using the same promoter was localized in differentiating cells and heterocysts (occasionally found in vegetative cells before nitrogen step-down). Considering the difference between PatU3(1-16aa)-GFP and PatU3(1-192aa)-GFP, the posttranslational regulation, if any, may depend on the function of the 17- to 192-aa region of PatU3. We also noted that PatU3(1-192aa)-GFP reached the maximal level at 24 h after nitrogen step-down in the filaments and that this fusion protein showed a stronger inhibitory effect on heterocyst differentiation than the full-length PatU3-GFP fusion protein (consistent with their abundance in cells). These phenomena may reflect a more complex role of PatU3 in regulation of cellular activities in Anabaena.

Most interestingly, PatU3(1-16aa)-GFP formed polar aggregates and peripheral patches in differentiating cells and heterocysts. The subcellular localization of bacterial proteins is often mediated by the signals carried on the RNA sequences and/or the amino acid sequences (24). Hypothetically, the 1- to 16-aa region of PatU3 (and, possibly, the RNA sequence upstream of the GFP-encoding region) is responsible for the subcellular localization of PatU3(1-16aa)-GFP, but the 17- to 192-aa region of PatU3 masks the localization signal or contains a recognition site for enzymatic excision of the protein (separating GFP from the N terminus of PatU3); therefore, the GFP fluorescence generated by PatU3(1-192aa)-GFP is distributed homogeneously in the cytoplasm.

While nonspecific distribution along filaments is consistent with the role of PatU3 in the control of cell size, the upregulated transcription of the gene and the localization of the N-terminal 16-aa sequence in differentiating cells are consistent with the role of PatU3 in heterocyst patterning. The N-terminal amino acid sequence may function in intact PatU3 in vegetative cells and developing heterocysts, as AVIKRRLQ-containing peptides after processing of PatU3, or both. AVIKRRLQ in PatU/PatU3 is conserved (identical and positive) in most heterocyst-forming cyanobacteria but less conserved in filamentous species that do not form heterocysts (data not shown). This could be due to the coevolution between proteins. It remains possible that PatU or PatU3 contributes to the coordination of cellular activities both intracellularly and intercellularly in filamentous cyanobacteria.

MATERIALS AND METHODS

General.

Anabaena strains are listed in Table S1 in the supplemental material, along with antibiotic resistances. These strains were cultured in BG11 (25) in flasks at 30°C in the light of 30 μE m⁻² s⁻¹ on a rotary shaker. For derivative strains carrying antibiotic resistance markers, erythromycin (5 μg ml⁻¹), neomycin (20 μg ml⁻¹), or spectinomycin (10 μg ml⁻¹) was added to the medium as needed.

To induce heterocyst differentiation, Anabaena strains grown in BG11 (optical density at 730 nm [OD₇₃₀], 0.7 to 0.9) were collected by centrifugation, washed 3 times with BG11₀ (BG11 with omission of nitrate) (25), and resuspended in the same medium.

To switch off gene expression from the P_petE promoter, Anabaena cells grown in BG11 were harvested by centrifugation, washed 3 times with copper-free BG11, and subcultured in the same medium in transparent plastic bottles for two passages, as previously described (22).

To test the cellular effects of PatU3 peptides, Anabaena strains were grown in test tubes with BG11, starting at an OD₇₃₀ of 0.05, and grown in the light with occasional agitation. PatU3 peptides were added to the cultures in test tubes at a final concentration of 10 μM after 24 h of growth and added again at 72 h. At 120 h, cells were collected by centrifugation, washed 3 times with BG11₀, and resuspended in BG11₀ with the same peptides.

Analyses of cell sizes and heterocyst frequencies.

Cell size and heterocyst formation were observed under an Olympus BX41 microscope. Proheterocysts (pale green and enlarged compared to vegetative cells) and heterocysts (yellowish transparent, enlarged, with thickened wall and less autofluorescence than vegetative cells, forming polar nodules when fully mature) were recognized by morphology, assisted with autofluorescence. In strains with GFP expressed in developing heterocysts, the GFP fluorescence was also used to enhance the identification of proheterocysts. Heterocyst frequencies (including heterocysts and proheterocysts) were calculated from at least 300 cells for each sample, with 3 biological repeats (over 300 cells × 3), and the differences between strains were evaluated with t tests. On magnified photomicrographs,x axis/y axis cell sizes were measured from 1,000 or 2,000 cells (including vegetative cells and heterocysts) for each strain, or 300 heterocysts for each strain, and the differences between strains or conditions were evaluated with cell size distributions (Fig. 2, 5, and 8; Fig. S2) and two-tailed t tests (Fig. S1, S2, S6, and S8). The two-tailed t test analyses were performed using Prism (GraphPad software v.5) and SPSS v.18 software packages.

Fluorescence microscopy.

Fluorescence images were observed through an Olympus BX41 microscope, a Leica SP8 confocal microscope, and a Zeiss LSM880 (superresolution) confocal microscope.

Using the Olympus BX41 microscope, GFP fluorescence was visualized with a Sapphire GFP filter set (exciter D395/40, dichroic 425DCLP, and emitter D510/40) (Chroma Technology Corp., Brattleboro, VT) and autofluorescence visualized with a red long-pass WG fluorescence cube (bp 510 to 550; BA590) (Olympus Corp., Tokyo, Japan). On the Leica SP8 equipped with an HC PL APO 63×/1.4-numeric-aperture oil objective (Leica Microsystems, Wetzlar, Germany), fluorescence images were taken at 0.54-μm-depth intervals, using a 488 nm laser line for excitation and a photomultiplier tube detector for collection of emission (500 to 560 nm for GFP fluorescence, 600 to 700 nm for cyanobacterial autofluorescence). GFP fluorescence was also imaged using a Zeiss LSM 880 microscope equipped with an AiryScan detector, an argon laser (Melles-Griot) for 488 nm excitation, and a Zeiss Plan-Apochromat 63×/1.4-numeric-aperture differential interference contrast (DIC) M27 oil objective (Zeiss Microscopy, Jena, Germany), with acquisition modality set to spot acquisition with 1.56 μs per time point and the bit depth set at 8 bits.

Construction of plasmids and Anabaena strains.

Plasmid construction processes are described in Table S1. DNA fragments cloned by PCR were confirmed by sequencing and comparison to the known genomic sequence (https://www.ncbi.nlm.nih.gov/genome/13531?genome_assembly_id=300961).

Plasmids were introduced into Anabaena 7120 strains by conjugation (26). The MTpatU3del4-48 markerless in-frame deletion mutant was generated based on a two-step homologous recombination protocol involving the use of sacB (27), and the completely segregated mutant was confirmed by PCR. Plasmids and PCR primers are listed in Table S1.

Copy numbers of pDU1-based plasmids, relative to rnpB in chromosomal DNA, were evaluated by quantitative PCR (28) using primers pDU1-1/pDU1-2 and rnpB-1/rnpB-2 (Table S1).

RT-qPCR.

Total RNA was extracted from Anabaena cells grown in BG11 or copper-free BG11 using TRIzol reagent (Invitrogen, Carlsbad, CA). After removal of DNA with DNase RQ1 (Promega, Madison, WI), samples were extracted with TRIzol reagent again. RT-qPCR analyses were performed as described previously (29), using rnpB (subunit B of RNase P) as the internal control. PCR primers (indicated with “RT” in the name) are listed in Table S1. The relative mRNA levels of ftn6 are given as means ± standard deviations (SD) from 3 technical replicates.

Assays of protein interactions.

Yeast two-hybrid assays were performed using the Matchmaker two-hybrid kit (Clontech, TaKaRa Bio, Otsu, Japan) as described in a previous report (29), with patU3 or its mutated versions cloned on the prey vector pGADT7 and the other genes cloned on the bait vector pGBKT7. The mutated versions mut1 through mut16 are carried on plasmids pHB5131 through pHB5146, respectively (Table S1).

For pulldown assays, His₆-tagged MBP-PatU3 (pHB3883), EF-Ts(HA)-Ftn6 (pHB5085), EF-Ts(HA)-PatN (pHB5645), EF-Ts(HA)-PatU3 (pHB5084), MBP (pET28a-MBP), and EF-Ts(HA) (pHB5043) were expressed in E. coli BL21(DE3), purified using the His·Bind purification kit (Novagen, Merck KGaA, Darmstadt, Germany) and desalted according to the manufacturer’s protocols. MBP is derived from the maltose binding protein of Escherichia coli, whereas hemagglutinin (HA) is an epitope (YPYDVPDYA) derived from the influenza virus protein hemagglutinin. To assay the interaction between PatU3 and PatN, an MBP-PatU3 fusion protein or MBP (as the negative control) was bound to amylose resin (NEB, Ipswich, MA), and then the amylose resin beads were incubated with 100 μg of purified EF-Ts(HA)-PatN fusion protein or EF-Ts(HA) (as the negative control) in 0.5 ml of phosphate-buffered saline (PBS) (10 mM Na₂HPO₄, 20 mM NaCl, 68 mM KCl, 1.76 mM KH₂PO₄, pH 7.4) at 4°C for 1 h and washed with 1 ml of PBS buffer 10 times. The retained proteins were boiled off the beads in 100 μl of sample buffer for 10 min. Proteins were separated by SDS-PAGE (12%) and electroblotted onto an NC filter. EF-Ts(HA)-PatN and EF-Ts(HA) were detected with an anti-HA tag monoclonal antibody (Abcam, Cambridge, UK) and a horseradish peroxidase (HRP)-conjugated secondary antibody specific for mouse IgG (Thermo Fisher Scientific, Waltham, MA) and visualized by chemiluminescence generated by HRP in the presence of a luminol/enhancer solution and a stable peroxide solution (Thermo Fisher Scientific).