Preventing Accidental Heterocyst Development in Cyanobacteria

The filamentous cyanobacterium Anabaena can form heterocysts specialized in N₂ fixation, mostly through a cascade of transcriptional activation in response to the nitrogen starvation signal 2-oxoglutarate. It is reported now that a transcription repressor, CalA, acts as a safety device to prevent heterocyst development under certain conditions where the 2-oxoglutarate level may touch the threshold to trigger unnecessary initiation of heterocyst development.

ABSTRACT

The filamentous cyanobacterium Anabaena can form heterocysts specialized in N₂ fixation, mostly through a cascade of transcriptional activation in response to the nitrogen starvation signal 2-oxoglutarate. It is reported now that a transcription repressor, CalA, acts as a safety device to prevent heterocyst development under certain conditions where the 2-oxoglutarate level may touch the threshold to trigger unnecessary initiation of heterocyst development. Such a control may increase the fitness of Anabaena under a constantly changing environment.

TEXT

Labor division and cell-cell interaction are characteristic of certain cyanobacteria, making these organisms a wonderful prokaryotic model for the understanding of multicellular behaviors, from basic mechanisms to evolutionary biology. The genetic model most frequently used for such studies is Anabaena (Nostoc) sp. strain PCC 7120 (here Anabaena), a filamentous and facultative diazotrophic cyanobacterium (1). Anabaena can use either combined nitrogen, such as ammonium or nitrate, or atmospheric N₂ for its growth, with a hierarchy in the preference of the nitrogen sources: ammonium is the most economic nitrogen source, followed by nitrate, which needs to be reduced to ammonium by nitrate reductase and nitrite reductase (Fig. 1). N₂ is the most costly nitrogen source since its reduction into ammonium catalyzed by nitrogenase consumes ATP and requires a microoxic environment for the protection of nitrogenase against oxygen inactivation. In the case of Anabaena, the microoxic environment is provided by the heterocyst, a cell type specialized in N₂ fixation. Heterocyst differentiation is triggered by the deprivation of combined nitrogen in the growth medium, and the process takes 20 to 24 h. Thus far, the model on the initiation of heterocyst differentiation relies mostly on transcriptional activation, in particular a regulatory circuit composed of 2-oxoglutarate (2-OG) acting as a nitrogen starvation signal and two transcription factors, NtcA (a receptor of 2-OG) and HetR (master regulator of heterocyst differentiation), together activating heterocyst differentiation (2). Now, in a study reported in this issue, Higo et al. propose that beyond the activating mechanism summarized above, a repression mechanism may exist to prevent accidental development of heterocysts under certain conditions (3). The authors found that calA, encoding a transcriptional repressor essential in cyanobacteria, exercises a negative control over the initiation of heterocyst development in the presence of nitrate that normally inhibits heterocyst differentiation. When calA is knocked down, the inhibitory effect of nitrate on heterocyst differentiation is partially relieved, and this effect seems to be amplified with a carbon oversupply. Although the data reported are short of providing all the necessary evidence for such a negative-control mechanism involved in the initiation of heterocyst differentiation, the concept itself and its significance should prompt the scientific community to investigate into the matter. calA is also known in other cyanobacteria as cyabrB, and in the paper in question, both names are used. For simplicity and with respect to those authors who reported previous works in Anabaena under the name calA, we suggest keeping the original name.

FIG 1.

Mechanism of heterocyst development in Anabaena. (A) Fluorescent image of Anabaena filaments grown under diazotrophic conditions, with the presence of heterocysts (indicated by arrowheads) (for details of fluorescence labeling, see reference 31). (B) Steps of heterocyst development. Only major steps most closely related to the topic are depicted. Anabaena can use combined nitrogen sources, such as nitrate or ammonium, or N₂ under combined nitrogen deprivation. In the latter case, 2-oxoglutarate (2-OG), as a carbon skeleton for ammonium assimilation, accumulates and activates the transcription factor NtcA. HetR is also a transcriptional activator and a master regulator specific for heterocyst development. Inhibitory signals derived from PatS and PatX act on HetR to determine the heterocyst pattern along the filaments. hetP and hetZ are two direct targets of HetR, involved in the commitment step, making heterocyst development irreversible. At the later stages, morphogenesis takes place with the formation of the polysaccharide layer (HEP layer) and the glycolipid layer (HGL layer) at the heterocyst cell wall. HepA regulates the formation of the HEP layer. CalA negatively controls hetP and hepA. trpE is required for tryptophan synthesis from glutamine, and it affects the 2-OG pool and thus also heterocyst development. Dotted lines represent steps that need further experimental confirmation. The binding of HetR on the hepA promoter is weak in vitro and thus may not be significant in vivo (indicated by a question mark).

To understand what is happening, let us go briefly through the model of heterocyst development and the coupling mechanism of carbon and nitrogen metabolism (Fig. 1). Heterocysts constitute 5 to 10% of all cells of the filaments and are semiregularly intercalated among vegetative cells. This distribution of heterocysts along Anabaena filaments represents a simple, one-dimensional pattern in developmental biology (1, 4). After the perception of the nitrogen starvation signal 2-OG by NtcA (5, 6), heterocyst development is initiated through the autoregulation of ntcA and hetR as well as their mutual regulation (7–9). HetR then starts to accumulate in developing cells and activates, through a mechanism that still needs to be clarified, the expression of patS (10). HetR acts cell autonomously, while PatS- and PatX-derived signals are supposed to act in a non-cell-autonomous manner, diffusing from developing cells to neighboring vegetative cells (10, 11). HetR forms a homodimer, and each monomer is composed of an N-terminal DNA-binding domain (DBD), a central flap domain, and a C-terminal hood domain (12, 13). Peptide signals containing the RGSGR motif bind to each HetR monomer at the lateral cleft of the hood domain, triggering a conformational change of the flap domain and destabilizing the HetR-DNA complex (13). Thus, cell-cell interaction under the joint action of HetR and diffusible inhibitory signals leads to the formation of a heterocyst pattern and the determination of cell fate (Fig. 1). Two direct targets of the HetR transcriptional activator have been confirmed, hetZ and hetP (14–18). How HetP and HetZ regulate heterocyst development remains unknown, but they have redundant functions, since depending on the expression levels, they can bypass, alone or together, the requirement of HetR in heterocyst development. HetP and likely HetZ represent a transition step from initiation toward commitment during heterocyst development (15–18) (Fig. 1). The commitment point occurs around 8 h after nitrogen stepdown and is defined as the nonreturn point in heterocyst development where the addition of a combined nitrogen source can no longer push a developing cell back to its vegetative state (19). After the commitment, morphogenesis starts with the expression of genes involved in the synthesis of the polysaccharide layer and the glycolipid layer added in the heterocyst cell wall, contributing, together with other physiological changes, to the formation of a microoxic environment so that nitrogenase can fix N₂ when filaments are grown under aerobic conditions (1). HepA is required for the synthesis of the polysaccharide layer (20).

The nitrogen starvation signal 2-OG, a trigger of heterocyst differentiation derived from the incomplete Krebs cycle in cyanobacteria, is also a carbon skeleton for ammonium assimilation through the glutamine synthetase-glutamate synthase (GS-GOGAT) cycle. Thus, 2-OG stands at the crossroads of carbon and nitrogen metabolism (4, 21, 22). Another metabolite, 2-phosphoglycolate (2-PG), a product of the oxygenase activity of Rubisco under carbon limitation, is considered a carbon starvation signal in cyanobacteria (22). Because of the intimate coupling of carbon and nitrogen metabolism, both carbon and nitrogen input can cause changes in the ratios of 2-OG and 2-PG, which then act allosterically on transcriptional factors such as NtcA (a receptor of 2-OG) and NdhR (a receptor of both 2-OG and 2-PG) in order to keep carbon and nitrogen metabolic balance (6, 22). The carbon-nitrogen metabolic coupling also means that nitrogen starvation corresponds to carbon oversupply and vice versa. The nature of the nitrogen source also affects the nitrogen-carbon metabolic balance. Indeed, because of the complexity and the cost linked to the nitrogen assimilation processes in cyanobacteria, the 2-OG concentration of the cells is lowest when ammonium is directly used as a nitrogen source, highest under N₂ fixation, and at an intermediate level when nitrate is present in the growth medium (4, 23). Consequently, the repressive effect of nitrate on heterocyst differentiation is less strong than that of ammonium; similarly, a high CO₂ supply can also boost heterocyst development (24).

calA is an AbrB-type transcription factor gene, highly conserved and considered essential in cyanobacteria (3). The lack of efficient genetic tools and the polyploid nature of freshwater cyanobacteria hinder our ability to reveal the function of essential genes in these organisms. So far, one approach used was the copper-inducible promoter of petE, which in theory can turn on/off gene expression, thus allowing examination of gene function (8). However, copper, as a trace metal, is difficult to deplete from growth media and culture glassware, and therefore, such a system is not stringent enough and sufficiently reliable to allow the autopsy of the essential gene function. This may explain why no particulate phenotype was observed when the petE promoter replaced the native promoter of calA in Anabaena (25). Two genetic tools have recently been developed to strengthen our ability to manipulate essential genes in cyanobacteria. One used in the study by Higo et al. is the dCas9-based CRISPR interference (CRISPRi) system combined with a double control with the copper-inducible petE promoter and the anhydrotetracycline-inducible tetR promoter (26). The second tool relies on CRISPR/Cpf1-assisted gene replacement on the chromosome and the control of gene expression by a synthetic promoter under the control of theophylline alone or with copper (27). In both cases, the expression of target genes can be turned down to a sufficiently low level so that their function can be examined. The availability of these genetic tools should facilitate functional dissection of essential genes in cyanobacteria.

Higo et al. used the CRISPRi approach to switch the expression of calA to a level low enough to obtain some phenotypes (3). Unfortunately, the authors could not confirm the essential function of calA, as one would expect, albeit the growth of the corresponding mutant under interference conditions was significantly inhibited. Thus, the control system still has room for further improvement, and this is a major shortcoming of the approach. When the phenotypes were examined, the results depended on the carbon and nitrogen regimes used in the culture. If nitrate was used as a nitrogen source, together with 1% CO₂ bubbling, the mutant could form a few heterocysts (0.6%, compared with less than 0.1% in the controls). When CO₂ was increased to 5%, about 2.9% heterocysts could be observed, relative to 0.4 to 0.6% in the controls. If ammonium replaces nitrate, heterocyst differentiation is still repressed. Thus, CalA suppresses heterocyst differentiation under conditions of carbon oversupply, but this effect is masked by ammonium but not by nitrate. It is already known that nitrate does not repress heterocyst differentiation completely, since the wild type can develop a few heterocysts under a nitrate regime, as also shown in the study by Higo et al. (3). It is also known that high CO₂ input leads to heterocyst differentiation as well (24). As summarized above, the concentration of the heterocyst differentiation signal 2-OG is lowest under an ammonium regime, highest under nitrogen starvation, and at an intermediate level in the presence of nitrate (4, 23). Thus, its level in cells of Anabaena filaments cultured with nitrate is just high enough to activate genes involved in nitrate uptake and assimilation through NtcA (2) yet just below the threshold concentration for triggering heterocyst differentiation. However, in such a situation, accidental differentiation may occur as a consequence of NtcA activation albeit at a lower level than that under combined nitrogen deprivation, accounting for the formation of a low percentage of heterocysts. Due to the coupling mechanism of carbon-nitrogen metabolism, carbon oversupply should exacerbate the nitrogen starvation status of the cells, causing the 2-OG level to increase further. Therefore, high CO₂ input weakens the repressive effect of nitrate on heterocyst development, which could explain why the phenotype of calA knockdown is more evident under such conditions. When ammonium is used as a nitrogen source, it drains the 2-OG pool faster, keeping its level below the threshold necessary for the initiation of heterocyst development even under carbon oversupply. Thus, the balance of carbon/nitrogen metabolism suffices to explain the difference in the phenotypes observed under different nitrogen regimes. Unless additional data are available, calling in an additional unknown mechanism as discussed by the authors may only add an unnecessary level of complexity.

At least two direct targets of CalA are identified, hetP and hepA, as CalA can bind to their promoter regions (3). The expression of these two genes is increased once calA expression is turned down, and this regulation is independent of HetR. The expression levels of hetR and hetZ are not affected. When calA is overexpressed, heterocyst development is inhibited, further confirming its role in the negative regulation of heterocyst formation. Strangely, while hepA expression is negatively affected by calA overexpression, that of hetP is little changed. This result is puzzling since both hetP and hepA are reported to be under the dual control of HetR and CalA (3, 28). The authors argue that HetR binds more strongly to the hetP promoter than that of hepA and thus outcompetes CalA for binding to the promoter region of hetP (3). Alternatively, direct control of hepA by HetR itself is questionable, since the binding affinity of HetR for the hepA promoter in vitro is too low to be significant in vivo, and the specificity of such binding in vitro still needs to be tested (28). Another puzzling question is how heterocyst differentiation is repressed by calA overexpression while both hetP and hetZ are still expressed. Indeed, these two genes, as direct targets of HetR, can bypass the requirement of the latter for heterocyst differentiation. When overexpressed alone or together, they can partly or completely rescue heterocyst differentiation in the absence of hetR (14–18). One hypothesis would be that heterocysts that developed under calA overexpression progressed beyond the steps controlled by hetZ and hetP but were blocked at the morphogenesis step through the action of CalA on hepA (Fig. 1). In summary, CalA negatively controls two steps in heterocyst development, the commitment step through hetP and the formation of the polysaccharide layer at the beginning of morphogenesis through hepA. The regulation of calA expression in different cell types during the course of heterocyst formation is not known. Higo et al. suggest that changes in the amounts of CalA examined by immunoblotting were not significant relative to the strong induction of HetR during heterocyst differentiation; they thus proposed a model according to which the accumulation of HetR after the induction of heterocyst differentiation outcompetes the repressive effect of CalA that remains little changed, thus allowing heterocyst differentiation to proceed. This may not be completely true. Since developing cells or mature heterocysts account for only 5 to 10% of the cells on the filaments, immunoblotting with total filaments could not tell the real story. Higo et al. observed a slightly low level of CalA in isolated mature heterocysts; in contrast, He and Xu reported a very dramatic decrease in the level of CalA in heterocysts compared to vegetative cells by using immunoblotting as well (25). Thus, despite certain discrepancies, both groups observed a downregulation of calA in mature heterocysts. Taken together, a more likely scenario as a working model is the following: once heterocyst development is initiated, the transcriptional activator HetR accumulates in developing cells, followed by or concomitant with a downregulation of the transcriptional repressor CalA. The combined effect of such regulations changes the relative ratio of HetR/CalA sufficiently to relieve the repressive effect of CalA on heterocyst development. In the presence of ammonium, the 2-OG level is sufficiently low to prevent the initiation of heterocyst development. When nitrate is used as a nitrogen source, or when carbon is oversupplied, a low percentage of cells may cause 2-OG to reach the threshold level for heterocyst differentiation to happen; the combination of both may even amplify the effect of relative nitrogen starvation, leading to more heterocyst differentiation, as observed by Higo et al. (3). The existence of CalA helps the filaments to limit the number of heterocysts formed when nitrate is still available, saving filament resources from the costly formation of heterocysts. The concept proposed by the authors, that CalA acts as a safety device preventing unnecessary heterocyst differentiation, is thus very attractive.

CalA is not the first player reported to suppress heterocyst differentiation in Anabaena. A trpE mutant displayed a similar, and even much stronger, phenotype compared to that resulting from a knockdown of calA (29) (Fig. 1). Anabaena has two genes encoding anthranilate synthase in the tryptophan biosynthesis pathway. One of the two, trpE, is a member of the hetR regulon (29). The trpE mutant forms about 8% heterocysts when cultured in the presence of nitrate, similar to the wild type under combined nitrogen deprivation; ammonium, on the other hand, can still repress heterocyst differentiation. Tryptophan synthesis requires glutamine as a precursor. According the proposed model (29), tryptophan can be transformed by tryptophan transaminase, in a 2-OG-dependent manner, to glutamate and 3-indole pyruvate. In the absence of TrpE, less 2-OG would be consumed, resulting in a transient accumulation of 2-OG. In this case, the filament will perceive the signal of nitrogen starvation, leading to heterocyst differentiation. Therefore, the mechanism of trpE and calA seems to be different. The hetY gene, when overexpressed, can also partly suppress heterocyst development, but the mechanism is unknown (30).

The study by Higo et al. raised some interesting questions for further investigation. For example, if calA expression could be turned down even lower, to a lethal level, could the phenotype related to heterocyst differentiation become stronger? If calA is downregulated during heterocyst development, what would the mechanism be? Could one test the hypothesis on the ratio of HetR/CalA in the regulation of heterocyst development? Can we rule out the possibility that CalA regulates nitrate uptake or assimilation, thus accounting for the difference in the phenotypes observed with ammonium and nitrate as nitrogen sources?

Despite all the questions raised, the shortcoming of CRISPRi, and several contradictions that need to be clarified in the future, the significance of the study by Higo et al. cannot be ignored. Except for several RGSGR motif-containing peptide signals involved in heterocyst patterning, heterocyst development is driven mostly by a cascade of transcriptional activation. A negative-control mechanism through CalA acting as a safety device to prevent unnecessary heterocyst formation constitutes another layer of regulation for the developmental process. The benefice of having such a control mechanism can be discussed in the ecological context. Freshwater cyanobacteria such as Anabaena in the natural environment should encounter constant changes in both the amount and the nature of nitrogen and carbon sources. A transition from one source to another, or variation in their concentrations, may cause the level of 2-OG to fluctuate constantly. Under certain conditions, the level of 2-OG may touch the threshold for the initiation of heterocyst differentiation, even though a combined nitrogen such as nitrate is still available. By controlling the expression of hetP at the commitment step, CalA can push the developmental process back, therefore saving resources from further development. Under combined-nitrogen deprivation, the 2-OG level may go beyond the threshold level, leading to an even higher level of HetR, which, together with the decrease of CalA through an unknown mechanism, will blow up the safety control of CalA, allowing heterocyst development to proceed. It is now becoming increasingly evident that carbon supply also affects heterocyst development. In view of our concept on the mechanism of carbon-nitrogen metabolic balance that is starting to emerge in the field of cyanobacterial studies, the initiation of heterocyst development should be conceived not just as a consequence of combined-nitrogen deprivation but rather as a result of a disbalance in carbon-nitrogen metabolism. Thus, more integrated approaches at the metabolic levels will be required for a better understanding of the mechanism of heterocyst development.

The views expressed in this article do not necessarily reflect the views of the journal or of ASM.