Heterocyst Septa Contain Large Nanopores That Are Influenced by the Fra Proteins in the Filamentous Cyanobacterium Anabaena sp. Strain PCC 7120

ABSTRACT

Multicellular heterocyst-forming cyanobacteria, such as Anabaena, grow as chains of cells forming filaments that, under diazotrophic conditions, contain two cell types: vegetative cells that perform oxygenic photosynthesis and N₂-fixing heterocysts. Along the filament, the intercellular septa contain a thick peptidoglycan layer that forms septal disks. Proteinaceous septal junctions connect the cells in the filament traversing the septal disks through nanopores. The fraCDE operon encodes proteins needed to make long filaments in Anabaena. FraC and FraD, located at the intercellular septa, are involved in the formation of septal junctions. Using a superfolder-green fluorescent protein (GFP) fusion, we found in this study that FraE is mainly localized to the poles of the heterocysts, consistent with the requirement of FraE for constriction of the heterocyst poles to form the “heterocyst neck.” A fraE insertional mutant was impaired by 22% to 38% in transfer of fluorescent calcein from vegetative cells to heterocysts. Septal disks were inspected in murein sacculi from heterocyst-enriched preparations. Unexpectedly, the diameter of the nanopores in heterocyst septa was about 1.5- to 2-fold larger than in vegetative cell septa. The number of these nanopores was 76% and 6% of the wild-type number in fraE and fraC fraD mutants, respectively. Our results show that FraE is mainly involved in heterocyst maturation, whereas FraC and FraD are needed for the formation of the large nanopores of heterocyst septa, as they are for vegetative cell nanopores. Additionally, arrays of small pores conceivably involved in polysaccharide export were observed close to the septal disks in the heterocyst murein sacculus preparations.

IMPORTANCE Intercellular communication, an essential attribute of multicellularity, is required for diazotrophic growth in heterocyst-forming cyanobacteria such as Anabaena, in which the cells are connected by proteinaceous septal junctions that are structural analogs of metazoan connexons. The septal junctions allow molecular intercellular diffusion traversing the septal peptidoglycan through nanopores. In Anabaena the fraCDE operon encodes septal proteins involved in intercellular communication. FraC and FraD are components of the septal junctions along the filament, whereas here we show that FraE is mainly present at the heterocyst poles. We found that the intercellular septa in murein sacculi from heterocysts contain nanopores that are larger than those in vegetative cells, establishing a previously unknown difference between heterocyst and vegetative cell septa in Anabaena.

INTRODUCTION

Multicellular, filamentous, heterocyst-forming cyanobacteria grow forming chains of cells that can be hundreds of cells long (1). The heterocysts are formed from vegetative cells in a developmental process that responds to nitrogen deprivation and follows a specific program of gene expression (2). This developmental process results in filaments that have CO₂-fixing vegetative cells and N₂-fixing heterocysts. In cyanobacteria such as the model organism Anabaena sp. strain PCC 7120 (here Anabaena), heterocysts constitute about 7% to 8% of all the cells and are spaced with a semiregular pattern along the filament (1). The cells in the filament share a continuous periplasm and are joined by proteinaceous septal junctions, formerly known as “microplasmodesmata” or “septosomes” (3, 4). The septal junctions have a role in the intercellular transfer of compounds that takes place in the filament to regulate heterocyst differentiation and to allow the exchange of nutrients between heterocysts and vegetative cells (5, 6).

Intercellular molecular exchange in filamentous cyanobacteria can be visualized by fluorescence recovery after photobleaching (FRAP) analysis using fluorescent markers such as calcein (7). Molecular transfer through the septal junctions takes place by diffusion (8), and the septal junctions appear to traverse the structures known as septal peptidoglycan disks through holes that have been termed nanopores (9, 10). The formation of mature intercellular septa, including septal junctions, requires numerous proteins that have been characterized mainly for Anabaena. SepJ, SepI, and HglK are important for the formation of fully developed septa (11–13), FraC and FraD are directly involved in septal junctions that have been visualized by electron cryotomography (14), AmiC-type amidases (AmiC1 and AmiC2) and a LytM-type regulator of AmiC1 appear to be involved in drilling the nanopores (15, 16), and further proteins, including the AmiC3 amidase (17), the peptidoglycan-binding protein SjcF1 (18), and the ABC transporter component GlsC (19), influence nanopore formation. Inactivation of the genes encoding these proteins generally affects intercellular molecular exchange and the formation of nanopores, but no mutant has been isolated that completely lacks nanopores or is completely blocked in intercellular molecular exchange.

The fraCDE operon encodes proteins required to make long filaments, mainly under deprivation of combined nitrogen, in Anabaena (the term fra originates from the filament fragmentation phenotype of the mutants [20]). However, whereas FraC and FraD appear to be functionally related to each other, FraE is distinct (21). Thus, for example, fraC and fraD mutants form heterocysts that have a heterocyst neck (albeit somewhat altered) and show some basal nitrogenase activity, whereas a fraE mutant forms heterocysts that lack an evident heterocyst neck and shows no nitrogenase activity (21, 22). Additionally, the fraC and fraD mutants are more strongly affected in the exchange of calcein between vegetative cells than the fraE mutant (21). Immediately downstream of the fraCDE operon, in the opposite DNA strand, lies the fraF gene, which is required to restrict filament length (23). The operon has been reported to be expressed constitutively (21) or induced at low levels (approximately 2-fold) under nitrogen deprivation (24, 25), whereas fraF is induced 3- to 6-fold under N deprivation (24–26), producing a transcript that overlaps fraE (23). Induction of fraF is mediated by the NtcA-dependent transcriptional regulator NrrA (26) and takes place mainly in the heterocysts (23).

In Anabaena, proteins such as SepJ, SepI, and HglK have been localized to the intercellular septa (12, 13, 27). Among the proteins encoded in the fraCDE operon, FraC and FraD were also localized to the intercellular septa (21, 22), but the subcellular localization of FraE could not be determined. Because of a predicted cytoplasmic location of the C terminus of FraE, a FraE-GFP_mut2 construct was tried in Anabaena, but it did not show green fluorescent protein (GFP) fluorescence (V. Merino-Puerto and E. Flores, unpublished data). We have now rechecked the possible topology of FraE, and updated programs suggest a topology with seven transmembrane α-helices positioning the C terminus in the periplasm (see Fig. S1, upper part, in the supplemental material). Hence, in a new attempt to determine the subcellular localization of FraE, we have now used the superfolder-GFP (sfGFP) that folds efficiently in the periplasm (28). We found that FraE-sfGFP localizes predominantly at the heterocyst poles and, based on this observation, conducted further analysis of the phenotype of a fraE mutant, including the study of calcein transfer from vegetative cells to heterocysts and of the nanopores of heterocyst septal disks. This study unraveled that nanopores are larger in heterocysts than in vegetative cells, which led us to characterize further the nanopores in a fraC fraD deletion mutant.

RESULTS

Visualization of FraE-sfGFP.

An Anabaena strain that bears the fraE gene with the sf-gfp gene fused to its 3′ end through a DNA sequence encoding a 4-glycine linker was constructed and named CSSA22 (see Materials and Methods; also see Fig. S2 in the supplemental material). This strain could grow diazotrophically, indicating that the fusion protein was at least partially functional (Fig. S2C). To study the localization of FraE-sfGFP, strain CSSA22 and the wild type, used as a negative control, were incubated for 24 and 48 h in BG11₀ medium (lacking combined nitrogen) to induce heterocyst formation, and their filaments were analyzed by confocal microscopy. After 24 h of incubation, GFP fluorescence was observed in filaments of CSSA22 mainly at the heterocyst poles, although weaker signals were also observed at the septa between vegetative cells, whereas no GFP signal was observed in the wild type as expected (Fig. 1A). GFP signals in strain CSSA22 after 48 of incubation were weaker than after 24 h (Fig. 1B).

FIG 1.

Localization of FraE-sfGFP. (A and B) Visualization of Anabaena wild type (WT) and several samples of strain CSSA22 (producing FraE-sfGFP) after 24 h (A) or 48 h (B) of incubation in BG11₀ medium. Overlays of cyanobacterial autofluorescence (red) and sfGFP fluorescence (green), visualized by confocal microscopy, are shown. The arrows point to the areas of green fluorescence in heterocysts (some indicated as het). (C) Boxplot representation of the GFP fluorescence intensity at the subcellular locations indicated in the scheme. The values shown correspond to the fluorescence detected at each location, from which the average fluorescence of the wild type at the same location was subtracted. Student’s t test P values are indicated for the comparison of fluorescence at 24 and 48 h of incubation for each position. Fluorescence was significantly larger at position 1 than at positions 2, 3, and 4 and at position 3 than at position 4 for both 24-h and 48-h samples (Student’s t test, P < 0.01 in every case).

GFP fluorescence was quantified in selected areas of the filaments, i.e., at the heterocyst poles (area 1) and lateral walls (area 2) and at the vegetative cells’ intercellular septa (area 3) and lateral walls (area 4). This analysis showed the highest GFP fluorescence at the heterocyst poles, but it also showed that GFP fluorescence was detectable, at lower levels, from the heterocyst lateral walls and vegetative cells’ septa and lateral walls (Fig. 1C). At the heterocyst poles, GFP fluorescence was about 3.5-fold (at 24 h) or 2.4-fold (at 48 h) higher than at heterocyst lateral walls. In vegetative cells, FraE-sfGFP was also present at significantly higher levels at the intercellular septa than in the lateral location. However, in the vegetative cells, the levels in the septa were only about 1.5-fold (24 h) and 1.7-fold (48 h) higher than the levels in the lateral walls, which do not exceed the 2-fold-higher levels expected for the juxtaposition of the cytoplasmic membranes of the adjacent cells. Therefore, we conclude that FraE-sfGFP is distributed rather homogeneously in the periphery of vegetative cells, whereas it is significantly increased at heterocyst poles. Additionally, the intensity of GFP fluorescence at the heterocyst poles decreased significantly after 48 h of induction. Such a decrease at 48 h was not observed at significant levels at any of the other locations (Fig. 1C).

Calcein transfer.

The intercellular exchange of calcein between vegetative cells of a fraE mutant, strain CSVT3 (fraE::pCSV3), has been previously determined as the exchange coefficient (E) and found to be affected in filaments grown in BG11 medium or incubated for 16 h in BG11₀ medium (21). Because of the preferential location of FraE at the heterocyst poles, we tested calcein transfer from vegetative cells to heterocysts in the wild type and the fraE mutant, repeating transfer between vegetative cells for comparison. The data were collected as the recovery rate constant R (19, 22) and are represented as boxplots (Fig. 2). Because the data showed a nonparametric distribution, statistical significance of the differences between mutant and wild type were assessed by the Mann-Whitney U test (29). The mutant generally showed intercellular calcein transfer at lower levels than the wild type, but this was especially significant (P = 0.048) in the transfer from vegetative cells to heterocysts at 24 h of incubation without combined N (Fig. 2).

FIG 2.

Intercellular transfer of calcein in the Anabaena wild type and the fraE mutant CSVT3. Shown are a boxplot representation and statistical analysis of calcein transfer between vegetative cells (Veg-Veg) or from vegetative cells to heterocysts (Veg-Het) in filaments grown in BG11 medium and incubated for 24 or 48 h in BG11₀ medium. The difference between the mutant and the wild type was assessed by the Mann-Whitney U test (P values shown at the top). Blue, quartile group 2 (from lower quartile [Q1] to median); yellow, quartile group 3 (from median to upper quartile [Q3]). Red diamonds, mean values.

Nanopores.

Because there is a relation between calcein transfer and the number of nanopores in septal peptidoglycan disks, we addressed the study of the nanopores in the fraE mutant strain CSVT3, and because of the specific localization of the FraE protein to the heterocyst poles, we first focused on the heterocyst-vegetative cell septa. For simplicity, we refer the septal disks of heterocyst-vegetative cell septa as “heterocyst septal disks” and the septal disks of vegetative cell-vegetative cell septa as “vegetative cell septal disks.” To isolate murein (peptidoglycan) sacculi from cell preparations enriched in heterocysts, filaments grown in bubbled BG11C medium and incubated for 24 h in bubbled BG11₀C medium (lacking combined N) were subjected to several rounds of mild passage through a French press to disrupt the vegetative cells and low-speed centrifugation to sediment the heterocysts, and the resulting heterocyst-enriched preparations were used to isolate peptidoglycan by protease treatment and hot SDS extraction (see Materials and Methods). A sacculus of a heterocyst (identified by its large size) and an adjacent vegetative cell including a heterocyst septal disk is shown in Fig. 3. As is illustrated below, it became immediately evident that the diameter of nanopores in the heterocyst septal disks was larger than that reported previously for vegetative cell septal disks (in which nanopores are about 15 to 20 nm in diameter [10, 12, 30]). We therefore investigated the septal disks from the heterocyst-enriched preparations and compared them to those found in murein sacculi from whole filaments grown and incubated under the same conditions (growth in bubbled BG11C medium and incubation for 24 h in bubbled BG11₀C medium). Because the larger size of heterocyst septal nanopores was previously unknown, we carried out the study not only in the wild type and the fraE mutant (strain CSVT3) but also in a ΔfraC ΔfraD double mutant (strain CSVT22).

FIG 3.

Murein sacculus of one heterocyst and an adjacent vegetative cell showing a septal peptidoglycan disk (SPD) with nanopores. Wild-type filaments were grown in bubbled BG11C medium and incubated for 24 h in bubbled BG11₀C medium, and murein sacculi were isolated from cell preparations enriched in heterocysts and visualized by transmission electron microscopy as described in Materials and Methods.

Some examples of vegetative cell and heterocyst septal disks from the three strains are shown in Fig. 4. We determined the diameter of the nanopores and noticed that for each strain, two populations of disks could be defined according to the mean diameter of their nanopores in the heterocyst-enriched preparations: a small fraction in which the mean nanopore diameter was between 12 and 20 nm and a large fraction in which the mean nanopore diameter was between 20 and 30 nm (Fig. 5A). In contrast, the whole-filament preparations (consisting mostly of vegetative cells) had almost exclusively disks with a mean nanopore diameter between 10 and 20 nm (Fig. 5B). This observation suggests that in the heterocyst-enriched preparations, the disks containing the small nanopores correspond to septal disks from contaminating vegetative cells, whereas the disks containing the larger nanopores correspond to heterocyst septal disks. We therefore characterized septal disks in the whole-filament and heterocyst-enriched preparations excluding from the latter the presumptive contaminating vegetative cell disks.

FIG 4.

Nanopores in heterocyst and vegetative cell septal peptidoglycan disks of the Anabaena wild type and the fraE (CSVT3) and fraC fraD (CSVT22) mutants. Filaments were grown in bubbled BG11C medium and incubated for 24 h in bubbled BG11₀C medium, and murein sacculi were isolated from whole filaments (Veg [upper micrographs]) or heterocyst-enriched cell preparations (Het [lower micrographs]) and visualized by transmission electron microscopy as described in Materials and Methods. Arcs of pores outside the septal disk are indicated by white arc lines in the lower left part of the WT Het micrograph (more examples are shown in Fig. 6).

FIG 5.

Distribution of the septal disks from the Anabaena wild type and the fraE (CSVT3) and fraC fraD (CSVT22) mutants according to the mean size of their nanopores. Filaments were grown in bubbled BG11C medium and incubated for 24 h in bubbled BG11₀C medium, and murein sacculi were isolated from heterocyst-enriched cell preparations (A) or from whole filaments (B). Note the presence of two distinct populations of septal disks in the heterocyst-enriched preparation, the small population tentatively identified as disks from vegetative cells (Veg; nanopore mean diameter, about 17 nm) and the larger population as heterocyst-vegetative cell disks (Het; nanopore mean diameter, about 26 nm). The whole-filament preparations are mainly composed of vegetative cells, since heterocysts are typically about 7 to 8% of the cells.

The septal disks are generally round in appearance, and therefore, we characterized them by determining their diameter. For vegetative cell septal disks, the mean diameter was in the range of 0.85 to 0.97 μm, being 14% and 12% larger in the fraE and fraC fraD mutants, respectively, than in the wild type (Table 1). Nanopore diameter in the vegetative cell septal disks of the wild type (mean, 12.6 nm [Table 1]) was smaller than previously described for this strain of Anabaena in BG11₀ medium (about 20 nm [12]). The reason for the small diameter of these nanopores is unknown but may be related to rapid growth in the bubbled BG11₀C medium used in this study. (A parallel analysis of the septal disks from whole filaments of Anabaena grown in shaken cultures with BG11₀ medium [air levels of CO₂] corroborated previously published data; see footnote b to Table 1.) The diameter of the nanopores in vegetative cell septal disks was about 30% and 20% larger in the fraE and fraC fraD mutants, respectively, than in the wild type (Table 1). Finally, nanopore number in the wild type (mean, 48.5 nanopores/disk) was similar to that previously described for this strain of Anabaena in BG11₀ medium (40 to 49 nanopores/disk [12, 31]). The fraE mutant showed a number of nanopores per vegetative cell septal disk similar to that of the wild type (Table 1), whereas the fraC fraD mutant showed a significantly lower number (mean, 7.3 nanopores/disk), as described previously for cells grown in BG11 medium (7.27 nanopores/disk [10]). In summary, the vegetative cell septal disks of filaments grown in bubbled BG11₀C medium have similar characteristics in the wild type and the fraE and fraC fraD mutants except that the nanopores are somewhat wider in the mutants and the number of nanopores is much smaller in the fraC fraD mutant.

TABLE 1.

Septal disk and nanopore parameters in samples from vegetative cells and heterocyst preparations of the Anabaena wild type and mutant strains CSVT3 and CSVT22

Sample type and parameter analyzeda	Value for Anabaena strain (genotype)
	PCC 7120 (WTd)	CSVT3 (fraE::pCSV3)	CSVT22 (ΔfraC ΔfraD)
Vegetative cell septal disksb
Septal disk diam (mean ± SD; μm) (n)	0.850 ± 0.053 (7)	0.969 ± 0.112 (15)	0.952 ± 0.261 (11)
Student’s t test P, mutant vs WT		0.0158	0.3263
Nanopore diam (mean ± SD; nm) (n)	12.6 ± 2.2 (137)	16.2 ± 2.2 (319)	15.2 ± 5.0 (62)
Student’s t test P, mutant vs WT		1.43 × 10−46	4.99 × 10−7
Nanopores/disk (mean ± SD) (n)	48.5 ± 27.3 (8)	47.9 ± 19.9 (15)	7.3 ± 6.5 (11)
Student’s t test P, mutant vs WT		0.9550	0.0001
Heterocyst septal disksc
Septal disk diam (mean ± SD; μm) (n)	1.049 ± 0.124 (20)	1.109 ± 0.211 (12)	0.630 ± 0.119 (17)
Student’s t test P, mutant vs WT		0.3150	3.13 × 10−12
Student’s t test P, heterocyst vs vegetative cell	0.0004	0.0363	0.0001
Nanopore diam (mean ± SD; nm) (n)	26.0 ± 3.3 (270)	25.2 ± 4.5 (190)	28.1 ± 7.8 (42)
Student’s t test P, mutant vs WT)		0.0306	0.0019
Student’s t test P, heterocyst vs vegetative cell	5.41 × 10−152	9.6 × 10−115	1.29 × 10−17
Nanopores/disk (mean ± SD) (n)	64.0 ± 25.8 (20)	48.9 ± 11.2 (12)	3.60 ± 5.3 (17)
Student’s t test P, mutant vs WT		0.0660	3.69 × 10−11
Student’s t test P, heterocyst vs vegetative cell	0.1699	0.8802	0.1174

^aThe strains were grown in bubbled BG11C medium and incubated for 24 h in bubbled BG11₀C medium for isolation of murein sacculi from whole filaments (vegetative cell septal disks) or heterocyst-enriched preparations (heterocyst septal disks).

^bA parallel analysis of the septal disks from whole filaments of wild-type Anabaena grown in shaken cultures with BG11₀ medium (air levels of CO₂) showed a septal disk diameter of 1.431 ± 0.413 μm (n = 22), 48.7 ± 12.3 nanopores per septal disk (n = 21), and a nanopore diameter of 20.3 ± 3.2 nm (n = 383).

^cThe heterocyst-enriched preparations contained some septal disks that, as defined for Fig. 5 (disks with mean nanopore diameter ≤ 20 nm), should be ascribed to vegetative cells and therefore were not included in this analysis. These “contaminant” vegetative septal disks were 5 out of 25 disks in the wild type, 4 out of 16 disks in CSVT3, and 3 out of 20 disks in CSVT22.

^dWT, wild type.

The diameters of heterocyst septal disks were similar in the wild type and the fraE mutant, 1 to 1.1 μm, but clearly smaller in the fraC fraD mutant, about 0.63 μm (Table 1). The mean diameter of nanopores of the heterocyst septal disks was only slightly smaller in the fraE mutant, and somewhat larger in the fraC fraD mutant, than in the wild type (Table 1). However, the number of nanopores per septal disk was, compared to the wild type, about 76% for the fraE mutant and about 6% for the fraC fraD mutant, showing a reduction in the number of nanopores in the former mutant and confirming the strong reduction of nanopore numbers in the latter mutant (10, 30).

The diameter of the septal disks in the wild type and the fraE mutant was about 15 to 20% larger in the heterocysts than in the vegetative cells, but in the fraC fraD mutant it was about 34% smaller in the heterocysts than in the vegetative cells (Table 1). As mentioned above, the nanopore diameter was clearly larger (about 1.5- to 2-fold larger) in the heterocyst than in the vegetative cell disks (Table 1), which of course represents an even larger nanopore area. Finally, the number of nanopores was somewhat larger (1.32-fold larger) in heterocyst than in vegetative cell septal disks of the wild type but not of the mutants (Table 1). Thus, the most significant difference between heterocyst and vegetative cell septal disks is the larger size of nanopores in the former, whereas the small size of the heterocyst septal disks in the fraC fraD mutant is also to be noted.

Arcs of pores.

Outside of the septal peptidoglycan disks, some pores organized in arc lines were observed in the heterocyst-enriched preparations but not in the murein sacculi from whole filaments (for an example, see Fig. 4, WT Het). Such arrays of pores were observed outside the heterocyst disks in 5 out of 20 micrographs for the wild type, 1 out of 12 micrographs for the fraE mutant, and 5 out of 17 micrographs for the fraC fraD mutant (some examples are shown in Fig. 6). It should be noted, however, that we did not look specifically for such structures, and therefore their distribution between strains and in different preparations should be considered preliminary. The approximate diameter of these pores in the wild type was 13.8 ± 2.5 nm (n = 22), and it was not significantly different in the mutants. Thus, these pores are smaller than those in the nearby heterocyst disks. Interestingly, in the fraC fraD mutant, in which the heterocyst septal disks were smaller than in the wild type, the arcs of pores were more separated from the disk than in the wild type (Fig. 6).

FIG 6.

Arcs of pores outside the heterocyst septal peptidoglycan disks of the Anabaena wild type (WT) and the fraE (CSVT3) and fraC fraD (CSVT22) mutants. Filaments were grown in bubbled BG11C medium and incubated for 24 h in bubbled BG11₀C medium, and murein sacculi were isolated from heterocyst-enriched cell preparations and visualized by transmission electron microscopy as described in Materials and Methods. Arcs of pores outside the septal disks are indicated by white or black arc lines.

DISCUSSION

Using a FraE-sfGFP fusion, we have shown that the FraE protein is mainly located at the heterocyst poles in filaments of Anabaena that had been incubated for 24 h in the absence of combined N, whereas it is present at lower levels in other cellular locations (i.e., heterocyst lateral walls and vegetative cell septa and lateral walls). At 48 h of incubation without combined N, the GFP fluorescence decreased at the heterocyst poles, whereas remained without significant changes at the other positions. Although the construct introduced into the chromosome to generate the fraE-sf-gfp fusion may affect the fine regulation of fraE expression (23), increased transient localization at the heterocyst poles is consistent with an important role of FraE in the formation of the heterocyst neck. That is, when FraE is missing (as in the fraE mutant), the heterocyst neck is not formed (22). The impaired transfer of calcein from vegetative cells to heterocysts (about 22% and 38% lower in the fraE mutant than in the wild type at 24 h and 48 h of N deprivation, respectively) and the low number of nanopores found in the heterocyst septal disks (24% lower in the fraE mutant than in the wild type) are also consistent with the absence of a properly formed heterocyst neck in the fraE mutant. This specific function of FraE in heterocyst differentiation is different from the role in the filaments of FraC and FraD, which are components of the septal junctions that can be involved in the transfer of metabolites such as sucrose between cells, including transfer to heterocysts (14, 30). Hence, our observations explain the different phenotypes of the fraC fraD and fraE mutants that have been described earlier (21, 22). On the other hand, FraE is also observed in the heterocyst lateral membranes and in the vegetative cells’ lateral and septal membranes. A role for FraE along the filament is therefore also possible, but this should be a minor role since, for example, molecular exchange between vegetative cells is less affected in the fraE mutant than in the fraC or fraD mutant (21).

BLASTp analysis identified FraE as an integral membrane component (transmembrane domain [TMD]) of an ABC transporter. More precisely, FraE is most similar to the type IV pilus biogenesis protein PilI from Myxococcus xanthus (Transporter Classification Database [TCDB] number 3.A.1.132.12; http://tcdb.org/), which is a component of an ABC exporter necessary for pilus assembly and pilus subunit export (32). On the other hand, structures recently added to databases have permitted us to identify FraE as a protein with structural similarity to the integral membrane component (Wzm) of the ABC transporter involved in export of the O-antigen polysaccharide of Gram-negative bacteria (see Fig. S1, lower part, in the supplemental material). This exporter has a wide channel that permits the translocation of the linear polysaccharide across the cytoplasmic membrane (33). Taking into consideration both similarities (to PilI and to Wzm), we speculate that FraE might be involved in the translocation of a macromolecule outside the Anabaena cytoplasm.

ABC transporters generally comprise a membrane complex and two ATPase subunits (the nucleotide-binding domains [NBD], frequently a homodimer), in addition to a periplasmic substrate-binding domain in the case of bacterial uptake transporters (34). Whereas a full membrane complex might be formed by a FraE homodimer, the question arises which protein encoded in the Anabaena genome could be an NBD partner of FraE to conform a functional ABC exporter. A hypothetical possibility is GlsC, an ABC transporter NBD subunit in Anabaena that appears to be a multitask NBD subunit, necessary for glucoside uptake but also for maturation of the intercellular septa, as it is necessary to produce a normal number of nanopores (19). Further work will be necessary to establish an association between FraE and GlsC.

We have visualized septal peptidoglycan disks from whole filaments (composed mostly of vegetative cells) and from heterocyst-enriched preparations of Anabaena. Although the presence at the heterocyst poles of the constriction that represents the heterocyst neck makes the heterocyst septa appear smaller than the septa between vegetative cells in transmission electron microscopy (see, for examples, references 35 to 37), the disks are somewhat larger in the heterocyst-vegetative cell septa than between vegetative cells of wild-type Anabaena (Table 1). This suggests that the peptidoglycan disk is a distinct structure in the septa, which may not occupy the apparent whole septa between vegetative cells.

An unexpected finding of this research has been that the nanopores are larger in heterocyst septal disks than in vegetative cell septal disks. This observation indicates that the nanopores in the heterocyst septa are not identical to those present in vegetative cell septa. We do not know, however, whether the heterocyst septal disks contain a modified version of the vegetative cell nanopores or the nanopores are completely different in the two types of septal disks. Nonetheless, the strong effect of the deletion of fraC and fraD on the number of nanopores in both types of disks suggests that they are related. The AmiC1 and AmiC2 amidases, which are involved in drilling the nanopores, and the AmiC3 amidase, which influences nanopore size, have been shown to be present at the septal areas of developing heterocysts (17, 36), where they could be involved in drilling or modifying the nanopores of heterocyst septa. Differences between the junctions (described as “channels”) between vegetative cells and between vegetative cells and heterocysts have been previously noticed (38). However, in that case, the channels between heterocysts and vegetative cells were described as thinner than the channels between vegetative cells. The channels remaining in the fraC fraD mutant were, however, wider than the wild-type channels (38), as observed in this study for the nanopores (Table 1).

In a different approach, in Anabaena cylindrica, the “microplasmodesmata” observed by freeze fracture electron microscopy in the exoplasmic fracture face were described as showing two possible sizes in vegetative cells but only one in heterocysts (3). Interestingly, the microplasmodesmata in heterocysts would correspond to the larger ones observed in vegetative cells. These observations made us to recheck all our data of vegetative cell septal disk nanopores from BG11-grown filaments (to ensure the absence of any significant number of heterocyst septal disks), and we found a possible polynomial (binomial or trinomial) distribution of nanopore sizes, with the smaller nanopores being more abundant than the larger ones (Fig. S3). Hence, our data support those of Giddings and Staehelin (3) that suggested the presence of microplasmodesmata of two sizes in vegetative cells. Whether the two types of microplasmodesmata and the two (or three) types of nanopores correspond to different types of septal junctions is unknown. It should be noted that microplasmodesmata as defined by Giddings and Staehelin (3) likely correspond to proteinaceous septal junctions (5), and therefore, our discussion implies that differently sized septal junctions are held by differently sized nanopores. The presence of more than one type of septal junction has been previously suggested based on the differences between sepJ and fraC fraD mutants in the transfer of fluorescent markers (22, 30). However, in an electron cryotomography study of Anabaena, only one type of septal junctions was resolved (14). Further work will be necessary to establish the nature of the septal junctions at the heterocyst-vegetative cell septa.

Another outcome of this research has been the visualization of arcs of small pores outside the septal peptidoglycan disks specifically in the heterocyst-enriched preparations. In contrast to the wild type, in which the arcs of pores are typically adjacent to the septal disks, in the fraC fraD mutant (which forms small disks), the arcs of pores are distant from the disks, which suggests a localization independent of the disk. Therefore, the formation of these arrays of pores may be unrelated to the formation of the septal disks. While their role in the heterocysts is currently unknown, the pores in these arc arrays are reminiscent of those in the “junctional pore complex” that has been proposed to mediate slime secretion for gliding motility in cyanobacteria (39). Anabaena is described as nonmotile (40), and accordingly, we have not observed such pores in vegetative cells. Instead, if the similarity to the pores in the junctional complexes were confirmed, these arcs of pores might have a role in the production of the thick polysaccharide envelope that is assembled around the heterocyst neck. Polysaccharide synthesis and export in a heterocyst-forming cyanobacterium have been recently suggested to involve a system of the Wzx/Wzy/Wzc/Wza type (41). The Wzc/Wza polysaccharide export complex that traverses the peptidoglycan in the periplasm of Gram-negative bacteria has a width of about 12 nm (42), indicating that a structure of this type could fit in the small pores (about 13.8 nm in diameter). Interestingly, Wzc (Alr2833)- and Wza (All4388)-like proteins are involved in the production of the heterocyst-specific polysaccharide in Anabaena (43–45), but further work will be necessary to establish a relation between these proteins and the small pores and, generally, to explore the role of the arcs of pores.

MATERIALS AND METHODS

Strains and growth conditions.

Anabaena (also known as Nostoc) sp. strain PCC 7120 was grown photoautotrophically in liquid BG11 medium (containing NaNO₃ as the nitrogen source and in which ferric ammonium citrate of the original recipe [40] was replaced with ferric citrate) or BG11₀ medium (lacking NaNO₃) under constant white light (15 to 25 μmol of photons m⁻² s⁻¹) with continuous shaking at 30°C. Alternatively, cultures (referred to as bubbled cultures) were supplemented with 10 mM NaHCO₃ (BG11C or BG11₀C medium) and bubbled with a mixture of CO₂ and air (1% [vol/vol]) in the light (50 to 75 μmol of photons m⁻² s⁻¹). To induce heterocyst formation, filaments of an exponentially growing culture were harvested by centrifugation, washed three times in BG11₀ or BG11₀C medium, and resuspended in the same medium. Plates of solid medium were prepared with 1% separately autoclaved agar and BG11 or BG11₀ medium. Anabaena strain CSVT3 (fraE::pCSV3 [21]) was routinely grown in BG11 medium supplemented with streptomycin sulfate (Sm) and spectinomycin dihydrochloride pentahydrate (Sp), each at 2.5 μg per ml (liquid medium) or 5 μg per ml (solid medium).

Escherichia coli DH5α was used for plasmid constructions, and strains HB101 and ED8654 were used for conjugation with Anabaena. They were grown in LB medium, supplemented when appropriate with antibiotics at standard concentrations (46).

For construction of an Anabaena strain bearing a fraE gene with sf-gfp fused to it 3′ end, 591 bp of fraE was amplified by PCR using Anabaena DNA as the template and primers alr2694-5 and alr2394-6 (see Fig. S2 in the supplemental material), which lacked the stop codon and included the sequences needed for digestion with restriction enzymes to incorporate the fragment into plasmid pCSAL39, which contains the sf-gfp sequence and a sequence encoding a four-glycine linker upstream of sf-gfp. Cloning of the PCR product into pCSAL39 produced plasmid pCSSA37, which was corroborated by sequencing. Subsequently, the amplified fraE/4-Gly/sf-gfp fragment was transferred to plasmid pCSV3 (Sm^r Sp^r), resulting in plasmid pCSSA38, which was transferred to Anabaena by conjugation. Finally, some exconjugants were tested by PCR (see Fig. S2). The strain carrying the construct to produce the sfGFP attached to the carboxyl terminus of FraE was named CSSA22 and was routinely grown in BG11 medium supplemented with 2.5 μg of Sm and 2.5 μg of Sp per ml (liquid medium) or 5 μg of Sm and 5 μg of Sp per ml (solid medium).

Confocal microscopy.

Cultures were grown in liquid BG11 medium (supplemented with antibiotics for the CSSA22 mutant), harvested by centrifugation, washed with liquid BG11₀ medium, and reinoculated at 1 μg of chlorophyll a ml⁻¹ in liquid BG11₀ medium (without antibiotics) for 24 or 48 h. For detection of sfGFP fluorescence, a sample of cell suspension was spotted onto solid BG11₀ medium. Pictures were taken using a Leica HCX PL Apo 63×, 1.4-numerical-aperture (NA) oil immersion objective attached to a Leica TCS SP2 confocal laser scanning microscope. sfGFP was excited using 488-nm irradiation from an argon ion laser and visualized across a window of 500 to 540 nm. Cyanobacterial autofluorescence was collected using a window of 630 to 700 nm. Fluorescence intensities at the indicated locations were measured using ImageJ software. The numbers of measurements taken for each position, strain, and incubation time were 29 to 59 in heterocysts and 58 to 64 in vegetative cells.

Peptidoglycan sacculus isolation and visualization.

Preinocula of 25 ml were grown in liquid BG11 medium. Then, the filaments were grown in 800 ml of bubbled BG11C medium supplemented with antibiotics (as necessary) to about 3 to 4 μg of chlorophyll a ml⁻¹, harvested by centrifugation, washed with BG11₀C medium, and incubated in 800 ml of bubbled BG11₀C medium (without antibiotics) for 24 h. Finally, filaments were harvested by centrifugation and either used directly to isolate murein sacculi (mostly of vegetative cells) or subjected to heterocyst enrichment. For this, a filament pellet was resuspended in 10 ml of phosphate-buffered saline (PBS) and filaments were disrupted by passing through a French press at 3,000 lb/in² four times. Heterocysts were sedimented by centrifugation at 200 × g for 10 min at 4°C, with discarding of the supernatant after centrifugation and resuspension in PBS (4 to 6 times). The sacculi were isolated by protease treatment using 100 mg/ml of α-chymotrypsin (from bovine pancreas; Sigma) and hot detergent extraction with decreasing dilutions of SDS and strong stirring, as described previously (9, 31). The purified sacculi were deposited on Formvar/carbon film-coated copper grids and stained with 1% (wt/vol) uranyl acetate. All the samples were examined with a Zeiss Libra 120 Plus electron microscope at 120 kV.

FRAP analysis.

Aliquots (0.5 ml) taken from the cultures were incubated for 90 min with 10 μl of calcein AM (1 mg/ml in dimethyl sulfoxide; Molecular Probes, Invitrogen) under constant shaking in the dark at 30°C. Filaments were then harvested by gentle centrifugation to prevent fragmentation, washed with BG11₀ medium, and resuspended in approximately 0.1 ml of BG11₀ medium. Then, an aliquot was spotted onto a dry BG11₀ medium agar plate, and filaments were allowed to settle down by drying off excess liquid. Small agar blocks with labeled filaments were transferred to a custom-built and temperature-controlled sample holder. During the FRAP experiments the temperature was kept at 30°C. Cells were imaged with a Leica HCX PL Apo 63×, 1.4-NA oil immersion objective attached to a Leica TCS SP2 confocal laser scanning microscope with a 488-nm line argon laser as the excitation source; fluorescent emission was monitored by collection across a window of 500 to 520 nm and a 150-μm pinhole. After one or two prebleach images were acquired, the bleach was carried out by an automated FRAP routine as previously reported (7). Postbleach images were taken in XY-mode approximately every 1 to 2 s over a period of 20 to 30 s. For FRAP data analysis, kinetics of transfer of the fluorescent tracer was quantified and the recovery rate constant (R) was calculated as previously described (19).

Data availability.

Original confocal microscopy images, septal disk micrographs, and FRAP raw data will be made available upon request.