An amidase is required for proper intercellular communication in the filamentous cyanobacterium Anabaena sp. PCC 7120

Significance

The filamentous cyanobacterium Anabaena has become a widely studied model to determine the molecular mechanisms involved in establishing and maintaining the pattern of heterocyst differentiation in response to the removal of fixed nitrogen from the environment. Heterocysts develop from vegetative cells, usually spaced about 10 cells apart, converting an oxic cell capable of division into an anoxic factory for nitrogen fixation that does not divide. Genetic analysis to elucidate the mechanisms of intercellular material exchange between heterocysts and vegetative cells is in an early phase. Here we show that an amidase is involved in the function of channels that penetrate the rigid peptidoglycan walls that separate cells in the filaments.

Abstract

Channels that cross cell walls and connect the cytoplasm of neighboring cells in multicellular cyanobacteria are pivotal for intercellular communication. We find that the product of the gene all1140 of the filamentous cyanobacterium Anabaena sp. PCC 7120 is required for proper channel formation. All1140 encodes an amidase that hydrolyses purified peptidoglycans. An All1140-GFP fusion protein is located at the Z-ring in the periplasmic space during most of the cell cycle. An all1140-null mutant (M40) was unable to grow diazotrophically, and no mature heterocysts were observed in the absence of combined nitrogen. Expression of two key genes, hetR and patS, was studied in M40 using GFP as a reporter. Upon nitrogen step-down, the patterned distribution of green fluorescent cells in filaments seen in the wild type were not observed in mutant M40. Intercellular communication in M40 was studied by measuring fluorescence recovery after photobleaching (FRAP). Movement of calcein (622 Da) was aborted in M40, suggesting that the channels connecting the cytoplasm of neighboring cells are impaired in the mutant. The channels were examined with electron tomography; their diameters were nearly identical, 12.7 nm for the wild type and 12.4 nm for M40, suggesting that AmiC3 is not required for channel formation. However, when the cell wall sacculi isolated by boiling were examined by EM, the average sizes of the channels of the wild type and M40 were 20 nm and 12 nm, respectively, suggesting that the channel walls of the wild type are expandable and that this expandability requires AmiC3.

The occurrence of multicellular organisms is one of the most significant steps in evolution (1, 2). One of the advantages of multicellularity is that an organism can differentiate specialized cells for different functions (3, 4). In prokaryotes, there are several independently evolved groups of multicellular organisms, including Actinobacteria, Myxobacteria, and the cyanobacteria (5). The cyanobacteria are a group of eubacteria that carry out oxygenic photosynthesis. They have the most diversified morphology among prokaryotes, ranging from unicellular cells to multicellular filaments with true branches (6). Anabaena sp. PCC 7120 (Anabaena 7120) is a filamentous cyanobacterium that can form heterocysts, providing an excellent model for studying cell differentiation and pattern formation (7–9). The importance of intercellular material exchange is evident in that heterocysts provide a micro-oxic environment for nitrogen fixation and supply nitrogenous compounds to the vegetative cells, whereas the vegetative cells perform oxygenic photosynthesis and supply sugars as energy and carbon skeleton to the heterocysts (7).

The heterocyst pattern is dependent upon intercellular communication according to Turing’s activator–inhibitor model, which requires the inhibitor to be diffusible (10, 11). Although many genes are involved in the regulation of heterocyst formation (7–9, 12, 13), hetR and patS are most important in heterocyst pattern formation. HetR is a transcription factor that controls the expression of other genes involved in heterocyst differentiation (14–18). The patS gene encodes a 17-aa peptide whose C-terminal pentapeptide is the inhibitor (19). The C-terminal peptide (RGSGR) prevents HetR from binding to DNA targets (16), leading to the suppression of heterocyst differentiation. In the case of heterocyst pattern formation, current evidence supports the view that the short peptide (E)RGSGR moves from heterocysts and proheterocysts to neighboring cells (19–23).

The detailed route of the PatS peptide movement between the cells has not been determined. Although it could move through the periplasmic space that is continuous and shared by all cells along the filaments of Anabaena 7120 (21, 24–27), we think that the PatS peptide and other metabolites move along the filaments through intercellular channels (26, 28–30). A recent electron tomography (ET) study has clearly established that channels penetrate the rigid peptidoglycan (PG) layer that separates cells in the filaments (30). The presence of nanopore pits on the PG septa between two cells (29) also strongly implies that there are cytoplasmic connections between two neighboring cells. The nanopores are located in the central areas of the septa. Formation of the nanopores on the septa between the cells requires amidases in both Anabaena 7120 and Nostoc punctiforme (29, 31, 32).

N-Acetylmuramyl-l-alanine amidases (Ami) cleave an amide bond between the N-acetylmuramic acid backbone (MurNAc) and l-alanine of the peptidoglycan (33). There are five subgroups of these enzymes in Escherichia coli: periplasmic AmiA, AmiB, AmiC, AmiD, and cytoplasmic AmpD (34–37). In Anabaena 7120, a mutant of amiC1 (alr0092) is unable to form heterocysts and loses intercellular communication. Two studies of the adjacent homologous gene amiC2 (alr0093) reported different results. Zhu et al. (38) showed that a mutant lacking alr0093 could not form mature heterocysts, whereas Berendt et al. (32) reported that a mutant lacking alr0093 showed no observable phenotype. An amiC2 mutant of N. punctiforme ATCC 29133 showed irregular cell-division planes and lacked both cell differentiation and intercellular communication through the cytoplasm (31). Recently, the 3D structure of AmiC2 from N. punctiforme was determined, and some structural features of the enzyme suggest that it has unique roles in cell-wall remodeling (39). Here, we show that all1140, encoding a different amidase-C, is required for intercellular material exchange in Anabaena 7120 and the differentiation of heterocysts.

Results

Inactivation of all1140 (AmiC3) of Anabaena sp. PCC7120.

In a screen of a mutant library of Anabaena 7120 for genes that are involved in heterocyst formation, we found that an insertion mutant of all1140 was incapable of diazotrophic growth. The gene all1140 and five other genes in Anabaena 7120 (alr0092, alr0093, all4294, all4998, and all4999) encode proteins that belong to the N-acetylmuramoyl- l-alanine amidases based on a cyanobacterial genome database search (cyanobase; genome.microbedb.jp/cyanobase). All the amidases except All4294 have an AmiC domain. The genes alr0092 and alr0093 encode AmiC1 and AmiC2, respectively; their catalytic AmiC domain is located at the C terminus (32). The protein encoded by all1140 has an AmiC catalytic domain located at the N terminus along with two adjacent PG-binding domains at the C terminus (Fig. 1A), and the domain arrangement in All1140 is conserved in heterocystous cyanobacteria (Fig. S1). The AmiC domain of All1140 is 36.1% and 33.7% identical to that of AmiC1 and AmiC2, respectively, and is 32.3% identical to AmiC in E. coli.

Fig. 1.

Measurement of recombinant AmiC3 activity with isolated peptidoglycans. (A) Schematic of the Prosite domains of AmiC3 of Anabaena 7120. The AmiC domain is in red, and the two PG domains are in green. (B–D) Analyses of the enzymatically digested products of PGs from S. aureus (B), E. coli (C), and Anabaena 7120 (D) with liquid chromatography (black traces). The red and green traces represent control experimental analyses.

Fig. S1.

BLAST search analysis of cyanobacterial genomes using the sequence of AmiC3 from Anabaena 7120 as probe. The search criteria required that the proteins have a catalytic domain near the N terminus and two PG-binding domains in tandem near the C terminus.

All1140 protein hydrolyzes PG. Recombinant All1140 was produced in E. coli and purified (Fig. S2A). The recombinant protein was incubated with PG isolated from Staphylococcus, E. coli, and Anabaena 7120, and the hydrolyzed products were analyzed by liquid chromatography. The results (Fig. 1 B–D) show that the recombinant All1140 has PG hydrolysis activity; we named this protein “AmiC3.” A mutant of Anabaena 7120 was constructed in which the all1140 gene was replaced by the streptomycin-resistance (Sm^R) cartridge (Fig. S2C), confirmed by Southern hybridization (Fig. S2B). The mutant strain is named “M40.”

Fig. S2.

Production of AmiC3 and construction of an all1140 mutant. (A) SDS/PAGE analysis of the AmiC3 produced in E. coli. Lanes 1 and 2, total cellular proteins of E. coli strain BL21(DE3) expressing the all1140 gene from Anabaena 7120 before (lane 1) and after (lane 2) the addition of isopropyl β-d-1-thiogalactopyranoside (IPTG). Lane 3 shows purified recombinant AmiC3; lane M shows molecular mass standards with their molecular masses on the right. (B) Analysis of all1140 inactivation by Southern hybridization. Total DNA isolated from the wild-type and the M40 mutant strains were digested with CalI and EcoRI. The fragments were separated by agarose electrophoresis and transferred to nitrocellulose paper. The fragments containing a portion of all1140 downstream were detected by random primer synthesized probes. The sizes of hybridized fragments are shown on the left. (C) Schematic drawing of the replacement of a part of the all1140 region. The expected sizes of ClaI/EcoRI-digested fragments are shown below. (D) Schematic drawing of the all1140 genes for complementing M40 to obtain C40 (Upper) and C40G (Lower).

Growth of the wild-type and M40 strains in BG11 (with nitrate) was measured (Fig. S3A). M40 had a longer lag time and a somewhat slower growth rate than the wild-type strain. Under these experimental conditions, the doubling time of M40 was 31.9 ± 1.3 h, whereas the doubling time of the wild-type strain was 27.1 ± 0.7 h. M40 was not able to grow when N₂ was the sole nitrogen source, and no heterocysts could be detected after a nitrogen step-down (Fig. 2A). The growth phenotype of the M40 mutant was largely restored to that of the wild-type strain when M40 was complemented with a wild-type all1140 gene (C40). C40 had a doubling time of 28.8 ± 1.7 h, and it restored the ability to develop heterocysts in a semiregular distribution along the filaments (Fig. 2A). The frequency of heterocysts in the filaments of different strains 36 h after nitrogen starvation is shown in Fig. S3B. The heterocyst frequency of C40 was 7.1 ± 0.9%, nearly the same as that of the wild-type strain (8.7 ± 1.2%).

Fig. S3.

Growth and heterocyst development of Anabaena strains. (A) Growth curves of the Anabaena 7120 wild-type (circles), M40 (squares), and C40 (triangles) strains. (Inset) Semilog plot of exponential growth phases. (B) Heterocyst frequencies of the wild-type, M40, and C40 strains 36 h after the removal of combined nitrogen. More than 800 cells were counted before and after stepdown. (C) Microscopic images of the strains expressing hetR under control of the petE promoter in wild-type and M40 backgrounds in the presence (0 h) or absence (24 h) of nitrate. (Upper) Bright-field microscopic images. (Lower) Images of photosynthetic pigments. Arrows indicate positions of heterocysts. (Scale bars: 5 μm.) (D) Heterocyst frequencies of the strains expressing hetR under control of the petE promoter in wild-type and M40 backgrounds in the presence (0 h) or absence (24 h) of nitrate.

Fig. 2.

Light microscopic images of Anabaena 7120. (A) Bright-field (Upper) and pigment fluorescence (PF) (Lower) microscopic images of the wild-type, M40 mutant, and the C40 strains. All images were taken from cultures that had been deprived of combined nitrogen for 24 h. (B and C) GFP fluorescence (Left) or pigment fluorescence (PF) (Right) images of Anabaena 7120 strains that had a gfp gene under control of the patS promoter (P_patS-GFP) (B) or the hetR promoter (P_hetR-GFP) (C). The strains were either in a wild-type background (Upper) or in a M40 background (Lower). Images were taken from cultures that had been deprived of combined nitrogen for 12 h (B) or 24 h (C). Arrows indicate the positions of heterocysts (A and C) or proheterocysts (B). (Scale bars: 5 μm.)

Fusions of the gfp gene to the promoters of hetR and patS (PhetR-gfp and PpatS-gfp) were constructed to examine the expression of hetR and patS. In the wild-type background, little GFP fluorescence is observed in strains with these gfp fusions under nitrogen-replete conditions. A typical pattern of GFP-fluorescent cells can be observed in the two strains with gfp fusions in the wild-type background after nitrogen step-down (Fig. 2 B and C). On the other hand, in strain M40 the expression of hetR increased even in the presence of combined nitrogen. GFP fluorescence was observed in nearly all the cells in the M40 strain and became brighter in all cells after nitrogen step-down (Fig. 2C). However, no pattern of differentiating cells with high GFP fluorescence was observed in the M40 strain. The expression pattern of patS is shown in Fig. 2B. Little GFP fluorescence could be observed in the strain with PpatS-gfp in the presence of combined nitrogen. The cells remained dim after the removal of combined nitrogen, and no typical pattern of GFP fluorescent cells in the filaments was observed (Fig. 2B).

Subcellular Localization of AmiC3-GFP.

Although there is no signal peptide or other domain that might target AmiC3 to periplasmic space, proteomic analysis showed that AmiC3 (All1140) existed in the outer membrane fraction of Anabaena 7120 (40). In E. coli, AmiC is located in the periplasm, and it is recruited to the Z-ring during cell division (41). To study the subcellular location of AmiC3, we constructed an all1140-gfp (AmiC3-GFP) fusion and transformed it into the wild-type and M40 strains. We first isolated different cell fractions and used immunoblotting to determine which fractions contained the fusion protein. In strain C40G, obtained by complementing M40 with the all1140-gfp fusion, the majority of AmiC3-GFP was found in outer membrane/PG fractions; a minor portion of AmiC3-GFP could be detected in the cytoplasmic fraction (Fig. 3A and Fig. S4). In one control, a strain that expressed the gfp gene from the rbcL (ribulose-1,5-bisphosphate carboxylase/oxygenase; Rubisco) promoter, GFP was detected only in the cytoplasmic fraction, demonstrating that various membranes were not contaminated with GFP. Fig. 3B shows the subcellular location of AmiC3-GFP in the filaments of these strains grown in BG11 medium (with nitrate). The GFP fluorescence images in strain C40G (Fig. 3B) show that AmiC3-GFP forms a ring structure at midcell positions in growing cells, possibly a result of association with the Z-rings in cell division (42–45). The AmiC3-GFP rings at the septa between two separated or nearly separated daughter cells were double-layered, suggesting that the AmiC3-GFP rings were split by the formation of septa. The diameter of the AmiC3-GFP rings became smaller as two daughter cells separated, and the ring eventually disappeared from the septal regions. In the absence of combined nitrogen, C40G formed heterocysts (Fig. 3C) and grew diazotrophically. In nitrogen-deprived cultures of C40G, the AmiC3-GFP fluorescence rings were observed on both sides of a proheterocyst, whereas no AmiC3-GFP ring was seen on either side of mature heterocysts (Fig. 3C). Because the septal ring is dependent upon FtsZ, we compared the AmiC3-GFP ring with the FtsZ-GFP ring. Although both rings are located in the septal regions (Fig. 3D), the AmiC3 ring persisted longer in cell cycles. Another difference was observed in the stationary phase: FtsZ-GFP was located mostly in the cytoplasm, whereas AmiC3-GFP was located mostly in the periplasmic space (Fig. 3E).

Fig. 3.

Subcellular localization of AmiC3 of Anabaena 7120. (A) Localization of AmiC3 by cell fractionation and immunoblotting. The cytoplasmic (cyto) and outer membrane and PG (OM) fractions of the wild-type and C40G strains and of a wild-type strain containing a gfp gene under control of the rbcL promoter (P_rbcL-gfp) were obtained. The proteins in the fractions were separated by SDS/PAGE followed by transfer to a nitrocellulose membrane. Antibodies against GFP were used for detecting GFP and AmiC3-GFP. The molecular mass standards are shown on the left. (B) Subcellular localization of AmiC3-GFP. Confocal image of Anabaena 7120 C40G containing an all1140-gfp fusion gene (B, 1) and its tilted (15°) image (B, 2). B, 3 and B, 4 are the enlarged images of the AmiC3-GFP ring in the red squares of B1 and B2, respectively. (C) Light microscopic images of C40G filaments containing proheterocyst (C, 1) or heterocyst (C, 2). Panels show bright-field, GFP, pigment fluorescence (PF), and a merged GFP and pigment fluorescence (PF+GFP) images. Arrowheads indicate the positions of a proheterocyst or a heterocyst. (D and E) Comparison of FtsZ-GFP rings (D, 1 and E, 1) and AmiC3-GFP rings (D, 2 and E, 2) in exponential growth phase (D) and in stationary phase (E). E, 3 and E, 4 are enlarged images of the images in red squares of panels E, 1 and E, 2, respectively. Arrows in D indicate the positions of newly formed septa. (Scale bars: 5 μm.)

Fig. S4.

(A) Isolation of the outer membrane and PG of Anabaena 7120. Band 1 is the thylakoid membrane layer. The pellet contains the outer membrane and PG. (B) SDS/PAGE analysis of the centrifugation fractions of Anabaena 7120: cytoplasm (lane 1), thylakoid membranes (lane 2), band 2 (lane 3), and pellets containing outer membranes and PG (lane 4). Molecular mass standards (lane M) with their molecular masses are shown on the left side of the gel. (C) Immunoblotting analysis of cytoplasmic (lanes 1, 3, and 5) and membrane/PG (lanes 2, 4, and 6) proteins. Lanes 1 and 2: fractions from C40G; lanes 3 and 4: fractions from the wild-type strain expressing the gfp gene under control of the promoter Prbcl (large subunit of RubisCO); lanes 5 and 6: the wild-type fractions. The molecular mass standards were separated in lane M, and their molecular masses are shown on the left of the gel. (Upper) SDS-Page analysis. (Lower) Immunoblotting with antibodies against GFP. The positions of GFP and AmiC3-GFP are shown on the right.

Intercellular Molecule Movement and Septal Nanopores (Channels).

To determine whether AmiC3 is required for intercellular communication, we investigated whether small molecules are able to move between the cells in Anabaena 7120. A nonfluorescent acetoxymethylester (AM) derivative of calcein was loaded into the cytoplasm of Anabaena 7120 (28) where it was converted into hydrophilic green-fluorescent calcein (622.5 Da) (46). The fluorescence recovery after photobleaching (FRAP) method was used to determine the capacity of the wild-type and M40 strains to support calcein movement (Fig. 4). In the calcein-loaded filaments of the wild-type strain, fluorescence of a photobleached cell could recover with a t_1/2 of 2 s (Fig. 4A and Table S1). In the calcein-loaded filaments of M40, the fluorescence of a bleached cell did not recover during the entire observation period of 80 s. We also investigated fluorescence recovery when the cells were loaded with fluorescent esculin, which has a molecular mass of 340.3 Da, smaller than calcein (Fig. 4B). Fast recovery of esculin fluorescence with a t_1/2 of 2 s was observed in the wild-type strain. The degree of the recovery was ∼50%. In M40, a small (18%) recovery of esculin fluorescence was observed with a recovery t_1/2 of 9 s (Table S2).

Fig. 4.

FRAP of Anabaena 7120 strains. The wild-type and mutant M40 strains were loaded with calcein (A, 1) or esculin (B, 1) before photobleaching. As indicated by arrowheads, a cell was photobleached, and its fluorescence was monitored continuously. Images before photobleaching (Pre) and at 0, 4, 20, and 80 s after photobleaching are shown. A, 2 and B, 2 show the kinetics of fluorescence intensities as a function of the recovery time of filaments loaded with calcein and esculin, respectively. (Scale bars: 5 μm.)

Table S1.

Kinetics of FRAP of the probe calcein, comparing wild-type Anabaena and the mutant M40 that lacks the amidase AmiC3

	Recovery time t1/2, s		Recovery degree, %
	WT7120	M40	WT7120	M40
1	2.36	>80	53.8	0
2	1.58	>80	46.8	0
3	1.92	>80	51.8	0
4	1.52	>80	69.4	0
5	4.36	>80	50.8	0
6	1.44	>80	59.4	0
7	0.89	>80	56.1	0
8	3.3	>80	52.3	0
	2.17 ± 1.14		55.1 ± 6.89	0

Table S2.

Kinetics of FRAP of the esculin probe, comparing wild-type Anabaena and the mutant M40 that lacks the amidase AmiC3

	Recovery time t1/2, s		Recovery degree, %
	WT7120	M40	WT7120	M40
1	1.33	5.88	55.6	19.5
2	1.44	19.5	32.7	31.2
3	1.25	11.2	59.0	16.8
4	3.52	4.63	34.4	12.4
5	4.19	5.18	45.1	20.3
6	1.42	9.67	36.2	18.6
7	0.94	15.91	66.5	23.1
8	2.3	4.26	50.8	8.36
	2.17 ± 1.14	9.53 ± 2.13	47.5 ± 4.42	18.8 ± 2.24

The lack of intercellular movement of calcein in M40 was investigated further using EM. In transmission EM (TEM), images of thin sections after chemical fixation show that the channels that physically connect two vegetative cells are present in both the wild-type and M40 strains (Fig. 5 A and B). Connections between a vegetative cell and heterocyst were observed also (Fig. S5). Next, high-pressure freezing combined with ET was used to investigate the channels of the wild-type and M40 strains; the results are shown in Fig. 5 C and D. The septal nanopores that are required for the formation of channels between cells are present as opaque areas of PG wall in perpendicular sections of the septa. Our measurements show an average diameter of 12.7 nm and 12.4 nm for the wild-type and M40 strains, respectively. The channels could be observed best when the images are rotated 90° (Fig. 5 C, 2 and D, 2). The septal nanopores were examined further with isolated cell-wall sacculi from Anabaena 7120 strains according to Lehner et al. (29); results are shown in Fig. 5 C and D. Although arrays of nanopores are observed in the septa of both the wild-type and M40 strains (Fig. 5 E, 2 and F, 2), the sizes of nanopores in the isolated sacculi of these two strains are quite different: The average diameter of the nanopores in the M40 strain is 11.7 nm, similar to that obtained by ET, whereas the average diameter of the nanopores of the wild-type strain is 20.1 nm, significantly larger than that obtained by ET.

Fig. 5.

Analyses of septal nanopores and connections by EM. (A and B) TEM images of the wild-type strain (A, 1), its enlarged image (A, 2), the M40 mutant (B, 1), and its enlarged image (B, 2). (C–F) C, 1 and D,1 are ET images of samples of the wild-type and M40 strains, respectively, frozen at high pressure. C, 2 and D, 2 correspond to a 90° rotation around the y axis of the septa of C, 1 and D, 1, respectively. C, 3 and D, 3 are the enlarged images of middle parts of C, 1 and D, 1, respectively. E, 1 and F, 1 are the images of isolated sacculi from the wild-type and M40 strains, respectively. E, 2 and F, 2 are the images of septa with nanopores from the wild type and M40, respectively. All tomographic images are composed of 10 superimposed 2.2-nm tomographic slices. (Scale bars: 100 nm in C, 2; C, 3; D, 2; and D, 3; 500 nm in all other panels.)

Fig. S5.

(A) Observation of a connection between a heterocyst and a vegetative cell by TEM. (B) An enlarged image of the area within the square in A. The arrow indicates the connecting area. The lengths of the scale bars are shown in each panel.

FRAP in Nonheterocystous Strains of Cyanobacteria.

In a survey of the ability to recover fluorescence after photobleaching in filamentous cyanobacteria, we found two nonheterocystous strains that showed no calcein fluorescence recovery when a cell was bleached: Phormidium sp. and Oscillatoria sp. (Table 1 and Fig. S6). Septal nanopores were examined with the sacculi isolated from these two strains and a Microcoleus sp. strain, which showed fast fluorescence recovery (Fig. S6). The average diameters of the septal nanopores of Phormidium sp. and Microcoleus sp. were 12.5 nm (Fig. 6A) and 20.0 nm (Fig. 6B), respectively (Table 1). The nanopore in Oscillatoria sp. was unique in that it had only one central nanopore on a septum, and the diameter of the nanopore was 23.3 nm (Fig. 6C).

Table 1.

Measurement of diameters of septal nanopores and fluorescence recovery times of filamentous cyanobacteria

Species	Diameter of nanopores, nm	Recovery time t1/2, s
Anabaena sp.7120 WT	12.7 ± 1.3 (ET)*	-
Anabaena sp.7120 M40	12.4 ± 0.7 (ET)*	-
Anabaena sp.7120 WT	20.07 ± 0.42 (sacculi)†	2.17 ± 1.14
Anabaena sp.7120 M40	11.66 ± 0.53 (sacculi)†	>80
Anabaena cylindrica	25.55 ± 0.46†	2.06 ± 0.93
Microcoleus vaginatus	19.95 ± 0.81†	3.13 ± 2.67
Oscillatoria lutea	23.32 ± 0.52†	>80
Phormidium foveolarum	12.49 ± 0.45†	>80

Two methods were used in measurement of the nanopore diameters: observation of isolated sacculi with TEM and ET of cryopreserved filaments. For determination of fluorescence recovery time, filaments were loaded with calcein, and fluorescence recovery was measured after a cell was photobleached. The strains used were obtained from the Institute of Hydrobiology (FACHB), Chinese Academy of Sciences, Wuhan, China.

^*Data from ET images of samples frozen under high pressure.

^†Data from TEM images of purified PG sacculi.

Fig. S6.

Images of FRAP experiments in filamentous cyanobacteria. Shown from the top are Anabaena 7120 (A), Anabaena cylindrica (B), Phormidium foveolarum (C), Microcoleus vaginatus (D), and Oscillatoria lutea var. contorta (E). Arrowheads indicate the cells that were photobleached. (Scale bars: 5 μm.)

Fig. 6.

Analyses of septal nanopores from filamentous nonheterocystous cyanobacteria by EMy. (A–C) Images from Phormidium foveolarum (A), Microcoleus vaginatus (B), and Oscillatoria lutea var. contorta (C), respectively. (A, 1, B, 1, and C, 1) Isolated cell wall sacculi. (A, 2, B, 2, and C, 2) EM images of septa with nanopores. (A, 3, B, 3, and C, 3) Enlarged images of a nanopore in A, 2, B, 2, and C, 2, respectively. (Scale bars: 500 nm in A and B; 20 nm in C.)

Discussion

Like AmiC1 and AmiC2, AmiC3 of Anabaena 7120 had an AmiC domain, and its amidase activity was demonstrated by its ability to hydrolyze PG isolated from E. coli, Staphylococcus aureus, and Anabaena 7120 (Fig. 1). Computer modeling (Fig. S7) predicts that the catalytic domain of AmiC3 is very similar to that of the N. punctiforme AmiC2 (39). Fractionation of cellular proteins revealed that the AmiC-GFP fusion protein is located in the periplasmic space (Fig. 3A), even though there is no signal peptide in the primary sequence of AmiC3. Confocal microscopy demonstrated that AmiC3-GFP is associated with the septal rings (Fig. 3B). A similar situation was found for the AmiC protein of E. coli, which has no signal peptide but is located in the periplasmic space (41). In Anabaena 7120, AmiC3 is a persistent component of the septal ring: The association of AmiC3-GFP with the septal ring can be observed for nearly the entire period of septum formation. The behavior of the AmiC3-GFP ring during the cell cycle (Fig. 3B) indicates that AmiC3 is tightly associated with inward-growing cell septa even after the disassembly of the FtsZ-ring (Fig. 3D). It also shows that AmiC3 in Anabaena 7120 performs its function in a restricted area: the newly formed PG layer of the cell septa. The M40mutant lacking the all1140 gene encoding AmiC3 showed no fluorescence recovery after a cell loaded with calcein was photobleached (Fig. 4), indicating that cellular communication is impaired in the mutant.

Fig. S7.

Computer modeling of the 3D structure of the catalytic domain of AmiC3 of Anabaena 7120. (A) Alignment of amino acid sequences of the Anabaena 7120 AmiC3 catalytic domain and the N. punctiforme AmiC2 catalytic domain. Key amino acid residues are in red squares. (B) 3D structure of the Nostoc AmiC2 catalytic domain. (C) Computer modeling of the Anabaena 7120 AmiC3 catalytic domain based on the 3D structure of the Nostoc AmiC2 catalytic domain. (D) The structures shown in B and C are overlapped for comparison.

Because the cyanobacterial cells have a rigid PG layer in cell walls, the formation of the channels for material exchange between cells would require the formation of nanopores in septa, and therefore it is not surprising that amidases are involved in these processes. An early study with freeze-fracture EM showed that some filamentous cyanobacteria had channels between the cells (47). These channels were called “microplasmodesmata,” but their relationship to plant connections was not established in terms of detailed structure or function. Several recent studies demonstrated that nanopores or channels are indeed present in the septa of heterocystous cyanobacteria (30). These studies agreed that there is an array of nanopores in the septum, but the reported size of the nanopore differed in these studies. A diameter of 12 nm was found when ET was used to study the channels crossing the septum in Anabaena 7120 (30). When sacculi of N. punctiforme were isolated by boiling and were examined by EM, the observed diameter of septal nanopores was 20 nm (29). Because the size of nanopores is important for their function, we compared the two methods. We found that the different nanopore sizes reported by various laboratories do exist in the wild-type Anabaena 7120 when the two different methods mentioned above were used to observe the nanopores. However, when M40 lacking AmiC3 was studied, very similar sizes of nanopores were observed with the same two methods (Fig. 5). These results indicate that the function of AmiC3 is not directly involved in pore drilling as is AmiC2 of N. punctiforme. More importantly, we found that the septal walls of the wild-type and M40 strains respond differently to the treatment during sacculi isolation. The septal wall of the wild-type strain was expandable, and the expandability was dependent upon AmiC3. Apparently, the ability to expand is important for proper channel formation to allow molecules with certain mass to move between the cells. That some fluorescence recovery was observed when the cells of the M40 strain were loaded with esculin (Fig. 4) supports this suggestion. It is also interesting that Phormidium sp. (Fig. 6) and Microcoleus sp. have 12-nm and 20-nm nanopores on septa of isolated sacculi, respectively. The former displays no calcein fluorescence recovery, but the latter does. In addition to nanopore size, the number of nanopores on septa is also important. In Oscillatoria sp., which has just one nanopore at the center of the septum, no calcein fluorescence recovery was detected even though the nanopore has a diameter of 23 nm (Fig. 6).

The ability of a vegetative cell to differentiate and form heterocysts for nitrogen fixation represents an advanced feature of multicellularity in prokaryotes. Among many genes that are involved in heterocyst formation, hetR together with patS plays a central role in the process. The initiation of heterocyst differentiation requires up-regulation of hetR (48), which in turn regulates the expression of other downstream genes for heterocyst formation. However, even though hetR was up-regulated in the absence of combined nitrogen in the M40 strain (Fig. 2), the patS gene, which is a downstream gene regulated by hetR, was not up-regulated, and no morphological differentiation was observed (Fig. 2). Overexpression of hetR, which usually induces multiple contiguous heterocyst formation (15), did not induce the formation of heterocysts (Fig. S3) in the M40 strain. Because the molecular mass of the short C-terminal peptide of PatS [(E)RGSGR] is similar to that of calcein, its movement from cell to cell could be impeded in the M40 strain.

How do we explain the inability of the mutant strain M40, which fails to make the enzyme AmiC3, to differentiate heterocysts? Differentiation requires the activity of HetR (14). The hetR gene is transcribed in response to a regulatory cascade that begins with deprivation of fixed nitrogen, in particular a reduction in the ratio of glutamine to 2-oxoglutarate. The latter activates NtcA, which activates transcription of nrrA (49), which in turn activates transcription of hetR. Thus, HetR is produced in all the cells of the filament of Anabaena 7120. Therefore, one would expect that all cells should differentiate. However, one gene that is immediately activated by HetR is patS; its product, the protein PatS, whose C-terminal hexapeptide, ERGSGR, is released by proteolysis, binds to HetR, and causes it to dissociate from target DNA (8, 16, 19). HetR freed from DNA is destroyed rapidly by protease. Therefore, the critical gradient that determines the pattern of differentiated cells should be a gradient of PatS peptide whose concentration is highest next to a heterocyst and lowest halfway between two heterocysts (23, 28). The full PatS protein must be made in the heterocyst, where there is abundant active HetR. Logic suggests that the protease that generates the inhibitory peptide from PatS is located in or near the channels that connect heterocysts to vegetative cells. Our FRAP experiments indicate that the channels through the peptidoglycan layer are a correct size to transport the PatS peptide. To complete this model, we must conclude that the channels made in the mutant M40 strain are defective in PatS processing and transport, so inhibitory PatS levels remain in all cells. Without export of PatS there can be no functional HetR and no heterocysts.

Experimental Procedures

Gene Inactivation and Complementation.

All enzymes were purchased from Promega and used according to instructions. To construct amiC3 (all1140) mutants, a DNA fragment containing amiC3 was amplified by the PCR with primers P3 and P4, using total genomic DNA as template. All primers used in PCR are listed in Table S3; all PCR products were confirmed by sequencing. The PCR-generated fragment was cloned in pGEM-T vector (Promega) to generate plasmid pTv-amiC3. The plasmid was inversely amplified by PCR with primers P5 and P6. The generated fragment was ligated with a blunt-ended Sm^R cartridge encoding resistance to streptomycin. The resulting plasmid pAmiC3-Sm was digested with BglII and PstI, and the recovered fragment then was cloned into pRL277 (50) for transformation of Anabaena 7120 by conjugation. Segregation of the mutant amiC3 was confirmed by Southern hybridization as described by Huang et al. (16). Total genomic DNA of Anabaena 7120 was isolated with the E.Z.N.A. Plant DNA Miniprep Kit (Omega Biotek) and then was digested with ClaI and EcoRI. The digested fragments were separated on a 1.0% agarose gel before transfer onto nitrocellulose paper for hybridization. The DNA probe was synthesized by random primer extension using the template amplified with primers P7 and P8. The mutant with all1140 mutation was named M40. For complementation of M40, the amiC3 gene plus a 600-bp upstream section were amplified by PCR using primers P9 and P10 followed by insertion into a chromosome docking site according to Liu (51).

Table S3.

Sequences of the oligonucleotides used as primers or probes in this study

Oligonucleotide	Sequence
P1	gaagCATATG agatatggaattgatattgg
P2	gatcGGATCCctaaccgattagtctttgcc
P3	gaagAGATCTgctctcccctgctcccctgc
P4	gatcCTGCAGgttactttaattaaggaaag
P5	atgtacaccctcaaatgatttc
P6	tggtagtgatgagtgagagg
P7	ctcttcgcgatcgctaacttg
P8	gccaatagggtgggcaatgc
P9	gaagGGATCCgaacaatatcgccgtaatc
P10	gatcGTCGACctaaccgattagtctttgcc
P11	gaagGTCGACagtaaaggagaagaacttttc
P12	gatcGAATTCttatttgtatagttcatccatg
P13	gatcGTCGACaccgattagtctttgcc
P14	gaagGTCGACatgagtaaaggagaagaacttttc
P15	gaagGGATCCctgccaatgcagaaggttaaac
P16	gatcGTCGACacaaatagttgaatagcacg
P17	gaagGGATCCcataaagtttttgtctgatt
P18	gatcGTCGACaatcttaaaatcggtgaattac
P19	gaagGGATCCgatcttagccttaaacaag
P20	gatcGAGCTCctcctaacctgtagttttat
P21	gaagGAGCTCatgagtaacgacatcgatctg
P22	gatcGAATTCttaatcttcttttctaccaaac
P23	gaagGAGCTCatgacacttgataataaccaag
P24	gatcGTCGACatttttgggtggtcgccgtc

To construct P_hetR-gfp and P_patS-gfp as reporters of hetR and patS expression, the gfp gene was first amplified with primers P12 and P14, and the fragment was digested with SalI and EcoRI. It then was ligated to pAM505 to generate plasmid pAM505-gfp with the same enzymes. The hetR and the patS promoters (14, 19) were amplified by primers P15/P16 and P17/P18, respectively, and the generated fragments were cloned into the BamHI and SalI sites of pAM505-gfp, generating pAM505-P_hetR-gfp and pAM505-P_patS-gfp. They were transformed into the wild-type and the M40 strains.

To express the hetR gene from the petE promoter, the petE promoter (15) amplified with P19 and P20 was cloned into pAM505, generating plasmid pAM505-P_petE. The hetR gene amplified with P21 and P22 was cloned into the SalI and EcoRI sites of pAM505-P_petE, and the resultant plasmid pAM505-P_petE-hetR was transformed into the wild-type and M40 strains separately. Confocal images of Anabaena 7120 were obtained with a Zeiss LSM 710 NLO DuoScan System confocal microscope using a Plan-Apochromat 63×/1.40 iil differential interference contrast (DIC) M27 objective. Photosynthetic pigment fluorescence images (excitation: 561 nm; emission detection: 600–650 nm) and GFP fluorescence images (excitation: 488 nm; emission detection: 495–540 nm) were recorded.

Localization of AmiC3 and FtsZ.

To localize AmiC3, a gfp gene was fused to the C-terminal part of amiC3 as follows. A gfp gene was amplified with PCR by primers P11 and P12 using the plasmid pAM1951 (19) as template before it was ligated to pAM505 to generate plasmid pAM505-gfp-2. The P_amiC3-amiC3 was amplified using primers P9 and P13 and was cloned into the BamHI and SalI sites of pAM505-gfp-2. The resultant plasmid, pAM505-P_amiC3-amiC3-gfp, was transformed into the M40 strain by conjugation to generate the C40G strain. FtsZ was localized according to Sakr et al. (43). First, the ftsZ gene was amplified using primers P23 and P24 by PCR with Anabaena PCC 7120 genomic DNA as a template and was cloned into the SacI and SalI sites of pAM505-P_petE to generate the intermediate vector,pAM505-P_petE-ftsZ. Second, the gfp gene amplified with PCR by primers P11 and P12 was digested with SalI and EcoRI and then ligated to pAM505-P_petE-ftsZ; the resultant plasmid pAM505-P_petE-ftsZ-gfp was transformed into wild-type Anabaena PCC 7120.

The cell suspensions were spotted onto agar-coated glass slides [1.5% (wt/vol) Bacto-Agar in growth medium]. All measurements were carried out at room temperature (∼25 °C). 3D-SIM images were obtained on an N-SIM imaging system (Nikon) equipped with a 100×/1.49 NA oil-immersion objective (Nikon) and four laser beams (488 and 561 nm). Image stacks with an interval of 0.12/0.24/0.48 μm were acquired and computationally reconstructed to generate superresolution optical serial sections with twofold extended resolution in both x,y and z directions. The reconstructed images were processed further for maximum-intensity projections and 3D rendering with NIS-Elements AR 4.20.00 (Nikon).

To determine which cellular compartment contains AmiC3, cell fractionation was performed with the strain C40G and a strain expressing gfp according to Bölter et al. (52) and Moslavac et al. (40). Proteins in different fractions were analyzed by immunoblotting using antibodies against GFP.

Calcein Labeling and FRAP.

Calcein labeling was done according to Mullineaux et al. (28) and Lehner et al. (31). For calcein labeling, cultures were grown until the OD₇₃₀ was 0.6. Then 0.5 mL of culture was harvested by gentle centrifugation and was washed three times with fresh BG11 medium (6). The pellet was resuspended in 0.5 mL fresh BG11 medium and was mixed with 20 μL of calcein-Am (1 mg/mL in dimethylsulphoxide; Invitrogen Molecular Probes). The suspension was incubated at 30 °C in the dark for 90 min, and cells were washed three times with fresh BG11 medium. The suspension was incubated for another 90 min in the dark before imaging.

For FRAP, the suspensions were spotted onto agar-coated glass slides [1.5% (wt/vol) Bacto-Agar with growth medium]. All measurements were carried out at room temperature (∼25 °C). FRAP experiments were performed on an UltraVIEW VoX spinning disk confocal microscope (PerkinElmer) with a 100×/1.4 NA oil-immersion objective and a solid state 50-mW 488-nm laser line. Calcein fluorescence was imaged with 488-nm excitation and 525/50-nm emission filters. After three prebleach images were recorded using10% laser power, a region of interest was bleached used a single iteration with the 488-nm line operating at 100% laser power, which lasted 50–80 ms depending on the bleach region size. Fluorescence recovery was monitored at low laser intensity (10% of a 50-nW laser) over 80 s at 0.5-s intervals. FRAP data were analyzed using Velocity software.

Esculin Labeling and FRAP.

For esculin labeling (53), cell cultures were grown on BG11 medium until the OD₇₃₀ was 0.6. Then 0.5 mL of culture was harvested by gentle centrifugation and was washed three times with fresh BG11 medium. The cell pellet was resuspended in 0.5 mL fresh BG11 and mixed with 15 μL of saturated (∼5 mM) aqueous esculin hydrate solution (Sigma-Aldrich). The suspension was incubated at 30 °C in the dark for 30 min, and cells were washed three times with fresh BG11 medium. The suspension was incubated for another 15 min in the dark before imaging.

For FRAP, the suspensions were spotted onto agar-coated glass slides [1.5% (wt/vol) Bacto-Agar with growth medium]. FRAP experiments were performed as described above for calcein, except that the bleach used a 430-nm line and emission was recorded at 474/501 nm.

EM and Cryo-EM.

Peptidoglycan sacculi were prepared by the method of de Pedro et al. (54) and Lehner et al. (29). EM of purified PG sacculi was done according to Priyadarshini et al. (55) and Lehner et al. (31). The samples of ultrathin sections were obtained according to Fiedler et al. (56) and were observed with a JEM-1010 electron microscope (JEOL). Cryopreservation of Anabaena 7120 strains and EM tomography were performed as described by Omairi-Nasser et al. (30).

The experimental procedures of strains and culture conditions, protein purification, and enzyme assays are provided in SI Experimental Procedures.

SI Experimental Procedures

Strains and Culture Conditions.

The wild-type and mutant strains of Anabaena 7120 were grown in BG11 medium (6) with or without nitrate at 28 °C under cool white fluorescent light at the light intensities indicated in the text. Liquid cultures were bubbled gently with air plus 1% CO₂, and their growth was measured by monitoring OD at 730 nm. To induce expression of genes under control of the petE promoter of Anabaena 7120, copper-free media were prepared according to Buikema and Haselkorn (15), and genes were induced by the addition of appropriate concentration of CuSO₄. Heterocyst differentiation was induced by transferring the cells to a medium without combined nitrogen, resulting in a fixed nitrogen step-down. E. coli strains were grown in LB medium supplied with appropriate antibiotics. E. coli strain DH5α was used for all routine cloning, and strain BL21 (DE3) was used for the production of recombinant proteins.

Protein Overproduction and Enzyme Assays.

Overproduction of recombinant AmiC3 was performed as follows. A DNA fragment containing the amiC3 (all1140) gene of Anabaena 7120 was amplified by PCR with primers P1 and P2 and as cloned into pET-15b. The plasmid was transformed into BL21 (DE3) cells for overexpression. The recombinant AmiC3 with a His-tag was purified with a metal-chelating column. The activity of recombinant AmiC3 on PG was assayed by overnight incubation at 37 °C (final volume of 200 μL) of recombinant AmiC3 (100 μg of total protein) with insoluble purified peptidoglycan equivalent to 200 nmol of MurNAc in 50 mM sodium phosphate buffer (pH 7.4). The reaction mixture was centrifuged at 9,000 × g for 10 min to eliminate insoluble residual material, and the supernatant was reduced and lyophilized. This material was dissolved in 0.5 mL of 0.05% trifluoroacetic acid and then was applied to an Eclipse×DB-C18 5-μm column (4.6 by 150 mm) (Agilent Technology). Elution was performed with 0.05% trifluoroacetic acid for 10 min, followed by a linear gradient of acetonitrile from 0 to 10% during the next 50 min, at a flow rate of 0.6 mL/min, followed by a step of 10% (vol/vol) acetonitrile for 10 min. Elution was monitored at 215 nm. PG of S. aureus was purchased from Sigma. PG of E. coli and Anabaena 7120 were isolated and purified according to the methods described in earlier studies (57–59).