RNase III-dependent down-regulation of ftsH by an artificial internal sense RNA in Anabaena sp. PCC 7120


  • Yuan Gao,

    1. The State Key Laboratory of Freshwater Ecology and Biotechnology, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, Hubei, China
    Search for more papers by this author
  • Yangmin Gong,

    1. The State Key Laboratory of Freshwater Ecology and Biotechnology, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, Hubei, China
    Search for more papers by this author
  • Xudong Xu

    Corresponding author
    • The State Key Laboratory of Freshwater Ecology and Biotechnology, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, Hubei, China
    Search for more papers by this author

Correspondence: Xudong Xu, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, Hubei 430072, China. Tel.: +86-27-68780659; fax: +86-27-68780123; e-mail: xux@ihb.ac.cn


RNase III is a group of dsRNA-specific ribonucleases that play important roles in RNA processing and metabolism. Alr0280 and All4107 in Anabaena sp. PCC7120 are highly similar to RNase III enzymes. In vitro, recombinant Alr0280 showed RNase III activity. In the same cyanobacterium, the expression of ftsH (FtsH protease) could be suppressed by overexpression of an artificial sense RNA (aaftsH) that was complementary to aftsH, an internal antisense RNA. The aaftsH interference was abolished by inactivation of alr0280, the RNase III-encoding gene, and restored by complementation of the mutant. A cyanobacterial homolog to hen1, an RNA methyltransferase gene, may also be required for the aaftsH interference. This is the first report of RNase III-dependent sense RNA interference in cyanobacteria, and the underlying mechanism remains to be elucidated.


RNase III is a group of double-stranded (ds) RNA-specific endonucleases found in both prokaryotes and eukaryotes (Conrad & Rauhut, 2002). They are phosphodiesterases that produce 5′-monophosphate and 3′-hydroxyl termini with a 2-nt overhang at the 3′ end. A divalent metal cation is required for the catalytic activity (Dunn, 1982; Meng & Nicholson, 2008). RNase III can be divided into four classes based on the molecular weight and the polypeptide structure (Arraiano et al., 2010). Class 1 RNase III enzymes are composed of an endonuclease domain (endoND) and a dsRNA binding domain (dsRBD). Class 2 enzymes, called the Drosha family, contain two endoNDs and one dsRBD. Class 3 enzymes, called the Dicer family, include the largest RNase III that bear two endoNDs, one dsRBD, a DEAD-box helicase domain, and a PAZ domain. Class 4 only includes the mini-RNase III of Bacillus subtilis, which has an endoND but no dsRBD.

RNase III enzymes function as homodimers, in which the two ribonuclease domains form a single processing center, each cleaving one strand of dsRNA (Zhang et al., 2004). The endoND contains a stretch of highly conserved amino acid residues, ERLEFLGD, known as the RNase III signature motif that contributes to an important portion of the catalytic center (Blaszczyk et al., 2004). For example, in Aquifex aeolicus RNase III, this motif is located at amino acid (aa) residues 37–44. Of six aa residues that constitute the catalytic center, E37 is required for protein dimerization, E64 for substrate recognition and scissile-bond selection, E40, D44, D107, and E110 for hydrolysis of the scissile bond (Gan et al., 2006). In Escherichia coli (E. coli) RNase III, the 8 aa signature motif is located at residues 38–45, while the catalytic center is composed of E38, E65, E41, D45, D114, and E117 (Sun et al., 2004). Unlike endoND, dsRBD may not be essential for the activities of RNase III enzymes. A truncated form of E. coli RNase III lacking dsRBD accurately cleaved small processing substrates (Sun et al., 2001). Giardia intestinalis Dicer without the C-terminal dsRBD showed the catalytic activity in vivo and in vitro (Macrae et al., 2006).

RNase III enzymes play important roles in RNA processing, post-transcriptional gene regulation, and decay of RNAs (Nicholson, 1999). Especially, RNase III enzymes are involved in small interfering RNA (siRNA)-, microRNA (miRNA)-, or antisense RNA-mediated gene regulation (Ji, 2008). siRNAs- and miRNAs-mediated gene regulation, also called RNAi (RNA interference), is eukaryote specific, while antisense RNA-mediated gene regulation is found in both eukaryotes and prokaryotes (Brantl, 2002).

Cyanobacteria are a group of oxygenic photosynthetic prokaryotes. Certain N2-fixing species of Anabaena/Nostoc has been proposed to be the ancestor of chloroplasts (Deusch et al., 2008). Antisense RNAs have been found in different groups of cyanobacteria (Csiszar et al., 1987; Dühring et al., 2006; Hernández et al., 2006, 2010; Steglich et al., 2008; Georg et al., 2009), and their regulatory effects have been shown in at least two species (Dühring et al., 2006; Hernández et al., 2006, 2010). In a previous report, we described a small internal antisense RNA (aftsH) of all3642 (ftsH, encoding a transmembrane protein with cytosolic AAA-ATPase and Zn2+-metalloprotease domains) in Anabaena sp. PCC 7120 (Anabaena 7120) (Gong & Xu, 2012). When overexpressed, aftsH inhibited the expression of ftsH gene. Here, we report that overexpression of the sense RNA, which could form duplex RNA with the endogenous aftsH, also suppressed the expression of ftsH. Of two predicted RNase III proteins, the one showing activity in vitro was required for the down-regulation of ftsH by the artificial sense RNA. Consistently, a Hen1 homolog appeared to promote the RNase III- and sense RNA-mediated gene regulation.

Materials and methods

Strains, culture conditions and transformation

Anabaena 7120 was obtained from the Freshwater Algae Culture Collection of Institute of Hydrobiology, Chinese Academy of Sciences. The wild-type and its derivatives were cultured in BG11, as previously described (Ning & Xu, 2004). Conjugative transfer of plasmids into Anabaena 7120 was performed, as described by Elhai and Wolk (Elhai & Wolk, 1988). Anabaena mutant was generated by sacB-based positive selection of double-crossover homologous recombination (Cai & Wolk, 1990). Complete segregation of the mutant was confirmed by PCR.

Plasmid construction

Molecular manipulations were performed using standard methods or per manufacturers′ instructions. PCR fragments cloned into pMD18-T (Takara) were confirmed by sequencing. Plasmid construction processes are detailed in Supporting Information, Table S1 in supporting information but briefly described below.

  1. Plasmids used to inactivate alr0280 or alr3730 in Anabaena 7120. DNA fragments carrying alr0280::Emr was generated using two T-vectors, pHB518 and pHB576, as described by Zhang et al. (2007). alr3730::Emr was generated by positioning a CmrEmr cassette between the 5′-end and 3′-end fragments of alr3730 in the T-vector pMD18-T (Takara, Japan) and deleting the Cmr marker. alr0280::Emr and alr3730::Emr were cloned into a sacB-based positive-selection vector (Cai & Wolk, 1990), respectively, producing pHB3315 and pHB4057.
  2. Plasmids used to produce recombinant Alr0280, Alr1158, or All4107 in Ecoli. DNA fragments containing alr0280, alr1158, or all4107 were cloned into pMD18-T, then excised and cloned into pET21b, producing pHB4046, pHB4556, or pHB4047.
  3. Plasmids used to overexpress aaftsH in Anabaena. PrbcL-aaftsH generated by overlap PCR (Horton et al., 1989) was cloned into pMD18-T, then the omega cassette (Prentki & Krisch, 1984) was inserted downstream of PrbcL-aaftsH. PrbcL and aaftsH-oop generated by PCR were cloned into pMD18-T, then PrbcL was excised and cloned upstream of aaftsH-oop. PrbcL-aaftsH-omega and PrbcL-aaftsH-oop were cloned into pDC8 (Zhang et al., 2007), respectively, producing pHB3144 and pHB3920.
  4. Plasmids used to complement the mutants and overexpress aaftsH. Addition of alr0280 with upstream sequence to pHB3920 produced pHB4315 for complementing the alr0280 mutation and overexpressing aaftsH. Addition of alr3730 with upstream sequence to pHB3920 produced pHB4327 for complementing the alr3730 mutation and overexpressing aaftsH.
  5. Plasmids used to detect antisense transcription within all3642. DNA fragments generated by PCR were first cloned in pMD18-T, then excised and cloned upstream of gfp in pHB912 (SmaI- and Sse8387I-cut pHB1071) (Wang & Xu, 2005), generating plasmids pHB2919 to pHB2923.

Transcription analyses

qRT-PCR and Northern blot hybridization were performed, as previously described (Zhang et al., 2007; Gao & Xu, 2012) with modifications. Total RNA extracted from Anabaena 7120 with TRIzol reagent was treated with RNase-free DNase I to eliminate chromosomal DNA. For qRT-PCR, total RNA was reverse-transcribed, and samples were tested in triplicates. Two independent experiments showed consistent results. rnpB (RNase P subunit B) (Vioque, 1992) was used as the internal control. Primers for rnpB and all3642 were rnpB-RT-1/rnpB-RT-2 and 3642-RT-1/3642-RT-2, respectively. For Northern blot hybridization, DNA probes were prepared by PCR using primers rnpB-N1/rnpB-N2 or 3642-N1/3642-N2 and labeled by incorporation of digoxigenin-dUTP.

Production, purification, and activity assays of recombinant proteins

His6-tagged RNase III was overexpressed in Ecoli BL21 (DE3) containing the plasmids pHB4046, pHB4556, or pHB4047, purified from total soluble proteins using His·Bind Purification Kit (Novagen) and desalted using microcon YM-3 or YM-10 (MILLIPORE) per manufacturers′ instructions. Proteins were stored in 20 mM Tris-HCl (pH 8.0), 50 mM NaCl, 1 mM EDTA, 0.5 mM DTT, and 50% glycerol at −20 °C. Protein concentrations were determined by the Bradford method (Kruger, 2002).

Four micrograms of recombinant proteins were incubated with 0.5 μg of 30-bp biotin-labeled dsRNA or 30-nt ssRNA (Fig. S1) in reaction buffer containing 30 mM Tris-HCl (pH 8.0), 100 mM KCl, 0.1 mM EDTA, 0.1 mM dithiothreitol (DTT), 10 mM MgCl2 at 25 °C for different periods of time. The reaction mixture was then transferred to an ice bath and mixed with an equal volume of loading buffer (95% formamide, 20 mM EDTA, 0.025% SDS, 0.025% xylene cyanol, 0.025% bromophenol blue). After heating at 95 °C for 5 min, samples were separated by electrophoresis on a 15% polyacrylamide-urea gel and electro-blotted onto Immobilon-Ny+ membranes (Millipore). Detection of the biotin-labeled RNA was performed using North2South Chemiluminescent Detection Kits (Thermo).


alr0280 encodes RNase III in Anabaena 7120

In the CyanoBase (http://genome.kazusa.or.jp/cyanobase), alr0280 and all4107 in Anabaena 7120 were annotated as RNase III-encoding genes. A homolog search in NCBI GenBank (http://blast.ncbi.nlm.nih.gov/Blast.cgi) also identified alr1158 as a candidate RNase III gene. Alr0280 and All4107 both contain an N-terminal endoND and a C-terminal dsRBD, while Alr1158 only possesses an endoND (Fig. S2A). Bacterial endoNDs possess a stretch of conserved residues, which is known as the RNase III signature motif (Blaszczyk et al., 2004). An alignment of partial sequences in predicted RNase III enzymes from Anabaena and other bacteria is shown in Fig. S2B. Alr0280 possesses the 8-aa motif, ERLEFLGD, All4107 has two substitutions, while Alr1158 does not have such a motif. In E. coli RNase III, E38, E41, D45, E65, D114, and E117 are required for the catalytic activity. Corresponding to D114, Alr0280 has a serine residue as the substitute; to E65, All4107 has a valine residue as the substitute.

These three genes were cloned and expressed in E. coli. The recombinant Alr0280 was purified by Ni2+-affinity chromatography to homogeneity, while the recombinant All4107 and Alr1158 were only partially purified (Fig. S3). These proteins were assayed in vitro for RNase III activity. A 30-bp biotin-labeled dsRNA and a 30-nt biotin-labeled ssRNA were used as the substrates. As shown in Fig. 1, the recombinant Alr0280 cleaved dsRNA. After denaturation of the dsRNA products, a biotin-labeled RNA strand slightly larger than 20 nt was detected. Neither Alr1158 nor All4107 showed RNase III activity under the same conditions. Also, none of the three proteins showed activity to cleave the 30-nt ssRNA (Fig. S4). Of the three genes, at least alr0280 encodes RNase III.

Figure 1.

RNase III activity assays. Alr0280 (a), Alr1158 (b), or All4107 (c) was incubated with 30-bp biotin-labeled dsRNA for different periods of time. RNA products were denatured, separated by urea-PAGE and electro-blotted onto Immobilon-Ny+ membranes. Lanes 1–5 in (a) to (c) incubated for 0, 20, 40, 60, and 90 min. (d) Sizing of the biotin-labeled RNA product generated by Alr0280. 1, 30-nt biotin-labeled ssRNA; 2, 30-nt biotin-labeled dsRNA excised with Alr0280; 3, 20-nt biotin-labeled ssRNA; 4, 10-nt biotin-labeled ssRNA. Sequences of the dsRNA or ssRNAs are shown in Fig. S1.

alr0280 is required for aaftsH-induced down-regulation of ftsH

RNase III enzymes play a role in antisense RNA-mediated gene regulation by cleaving the antisense RNA/mRNA duplex (Krinke & Wulff, 1987, 1990; Blomberg et al., 1990; Vogel et al., 2004; Huntzinger et al., 2005). Previously, we identified an antisense RNA called aftsH internal to all3642 (ftsH) in Anabaena 7120; Northern blot analysis showed that overexpression of aftsH from PrbcL (the promoter of rbcL that encodes the large subunit of ribulose-1, 5-bisphosphate carboxylase) suppressed the expression of ftsH (Gong & Xu, 2012). However, when overexpressed from PrbcL, the sense RNA that was complementary to aftsH, named aaftsH, also suppressed the expression of all3642 (Fig. 2).

Figure 2.

Effects of overexpression of aftsH or aaftsH on the expression of ftsH in Anabaena 7120. (a and b) Schematic diagrams showing aftsH within all3642 and clones for expressing aftsH or aaftsH from PrbcL in Anabaena 7120. Bent lines with arrowheads indicate start sites of transcription; stem-loop structures stand for transcriptional terminators of omega cassette or oop. (c) Northern blot analyses of ftsH transcripts and aftsH in Anabaena 7120 without (1) or with either PrbcL-aftsH-omega (2) or PrbcL-aaftsH-omega (3). Fusion of aftsH or aaftsH to a stretch of sequence downstream of transcriptional start site and a stretch of sequence before termination by the stem-loop structure of omega made the overexpressed RNA larger than the endogenous aftsH. The construct carrying PrbcL-omega showed no effect on ftsH mRNA (Gong & Xu, 2012). rnpB was used as the internal control.

To promote quantitative analysis of ftsH mRNA, we employed qRT-PCR to examine the regulation of mRNA level by aftsH or aaftsH using the primers indicated in Fig. 3. One primer was located within aftsH, while the other one was located in a region upstream of aftsH with very low background antisense transcription. Using gfp as the reporter gene, we detected the antisense transcription of a series of fragments within all3642. An additional antisense transcription was identified in a fragment located between chromosomal bp 4393694 and 4394264 (pHB2921), which was coincident with the antisense transcription from chromosomal bp 4394269 identified by Mitschke et al. (2012) (Fig. 3). However, this antisense transcription apparently terminated before crossing the region bracketed by the primers (Fig. 3). Consistent with this result, directional deep RNA sequencing had identified no additional antisense transcription in this region (Flaherty et al., 2011). Therefore, this pair of primers can be used to evaluate the mRNA level of all3642. qRT-PCR produced similar results with the Northern blot analysis, and the suppression of ftsH mRNA level by aaftsH was even stronger than that by aftsH (Fig. 4). Hereafter, this phenomenon is called aaftsH interference.

Figure 3.

Detection of antisense transcription within all3642 using gfp reporter gene. The upper part shows DNA fragments within all3642 cloned upstream of gfp in Anabaena-Escherichia coli shuttle plasmids pHB2938 and pHB2919-pHB2923. These fragments were generated by PCR using the same primer at the 5′ end but different primers at the 3′ end. Bent lines with arrowheads indicate transcription at different start sites. The solid leftward line stands for transcription of all3642, and the solid rightward line stands for transcription of aftsH. Dashed lines with arrowheads stand for other transcriptional start sites within all3642 detected by RNA sequencing (Mitschke et al., 2012). Multiple transcriptional start sites of all3642 are not indicated here. The two arrowheads opposite to each other indicate primers (bracketing Anabaena chromosomal bp 4394859-4395029) for qRT-PCR (Table S1). all3642 is located at chromosomal bp 4395350-4393464, while aftsH is located at bp 4394967-4395199. The lower part shows bright-field (I) and GFP-fluorescence (II) microscopic images of Anabaena 7120 harboring these plasmids. Heterocysts were induced by nitrogen stepdown in BG11 minus nitrate.

Figure 4.

qRT-PCR analysis of ftsH mRNA showing the role of alr0280 and alr3730 in aftsH- or aaftsH-induced down-regulation of ftsH in Anabaena 7120. (a) Overexpression of aftsH in the wild type (wt) or alr0280 mutant. 1, wt; 2, wt with PrbcL-aftsH-omega; 3, alr0280::Emr; 4, alr0280::Emr with PrbcL-aftsH-omega. (b) Overexpression of aaftsH in the wild type or alr0280 mutant. 1, wt; 2, wt with PrbcL-aaftsH-oop; 3, alr0280::Emr; 4, alr0280::Emr with PrbcL-aaftsH-oop; 5, alr0280::Emr with alr0280-PrbcL-aaftsH-oop. (c) Overexpression of aaftsH in the wild type or alr3730 mutant. 1, wt; 2, wt with PrbcL-aaftsH-oop; 3, alr3730::Emr; 4, alr3730::Emr with PrbcL-aaftsH-oop; 5, alr3730::Emr with alr3730-PrbcL-aaftsH-oop.

We wondered whether Alr0280 was involved in the regulation by aftsH or aaftsH. An alr0280 mutant of Anabaena 7120 was generated by interrupting alr0280 with an Emr marker (DRHB3315, Table S1), and aftsH or aaftsH was overexpressed in the mutant as in the wild-type strain. Inactivation of alr0280 showed no influence on the regulation of ftsH mRNA level by aftsH (Fig. 4a) but almost abolished the regulation by aaftsH (Fig. 4b). In order to complement the alr0280 mutant, we cloned alr0280 with upstream sequence into the plasmid carrying PrbcL-aaftsH-oop and introduced the resulting plasmid into the alr0280 mutant. The addition of alr0280 restored the strong suppression of ftsH mRNA level by aaftsH in the alr0280 mutant (Fig. 4b). The aaftsH interference depended upon RNase III.

RNase III-dependent aaftsH interference may involve a Hen1 homolog

In Anabaena variabilis, there is a Hen1 homolog that is involved in RNA methylation and repair (Chan et al., 2009), and the Pnkp/Hen1 RNA repair system exhibits broad substrate specificity (Zhang et al., 2012). Because Hen1 in plants is required for accumulation of miRNA (Yu et al., 2005), we wondered if the Anabaena Hen1 is related to the aaftsH interference. In Anabaena 7120, alr3730 is the predicted Hen1-encoding gene. A hen1 mutant was generated by interruption of the gene with an Emr marker (DRHB4057, Table S1). For unknown reason, inactivation of alr3730 decreased the level of ftsH mRNA (Fig. 4c). In contrast to that in the wild type, introduction of PrbcL-aaftsH-oop on plasmid to the mutant up-regulated the expression of ftsH, and addition of alr3730 with upstream sequence to the plasmid partially restored the aaftsH interference (Fig. 4c). The Hen1 homolog appeared to play a role in the aaftsH interference.


RNase III enzymes are involved in generation of functional RNAs and gene regulation by noncoding RNAs therefore participate in many cellular activities (Nicholson, 1999; Blaszczyk et al., 2004). Even though many noncoding RNAs have been found in cyanobacteria, no study of cyanobacterial RNase III has been reported to date. In this study, we identified alr0280 as the RNase III-encoding gene in Anabaena 7120 and showed its role in down-regulation of ftsH by an artificial sense RNA.

In Anabaena 7120, Alr0280, Alr1158, and All4107, all contain a region similar to endoND. However, Alr1158 lacks the 8-aa signature motif that is required for the formation of catalytic center (Blaszczyk et al., 2004), while All4107 lacks E65 (numbered as in E. coli RNase III) that is involved in substrate recognition and scissile-bond selection (Blaszczyk et al., 2001; Sun et al., 2004; Gan et al., 2006). Like All4107, Thermotoga maritime RNase III has a valine residue at the position corresponding to E65 but cleaves dsRNA in vitro: the role of E65 in RNase III activity could have been replaced by the adjacent glutamic acid residue (Sun et al., 2004; Nathania & Nicholson, 2010). In All4107, no such a glutamic acid residue can be found. In Alr0280, there is a substitution at the conserved site D114 (numbered as in E. coli RNase III), which is involved in hydrolysis of the scissile bond (Gan et al., 2006). As in Helicobacter pylori RNase III, it is substituted by serine. Due to a potential functional redundancy of E41 and D114 (Sun et al., 2004), the substitution at D114 may not necessarily eliminate the RNase III activity. Consistent with these analyses, Alr0280 showed RNase III activity in vitro, while the other two did not. Even so, we cannot exclude the possibility that All4107 has RNase III activity in vivo but with a different function.

Bacterial RNase III enzymes can play important roles in antisense RNA-mediated gene regulation (Krinke & Wulff, 1987, 1990; Blomberg et al., 1990; Vogel et al., 2004; Huntzinger et al., 2005). However, in Anabaena 7120, inactivation of alr0280 showed little effect on aftsH-induced down-regulation of ftsH. This could be due to the alternative mechanism that involves RNase E (Thomason & Storz, 2010). For example, in Salmonella enterica, AmgR-induced degradation of mgtC mRNA is RNase E dependent (Lee & Groisman, 2010). Usually, used as a control to show the effect of antisense RNA on target mRNA, overexpression of the complementary sense RNA should increase the mRNA level due to titration of the antisense RNA. In the unicellular cyanobacterium Synechocystis 6803, the sense RNA anti-isrR enhanced the expression of isiA by reducing isrR (Dühring et al., 2006). However, in Anabaena 7120, overexpression of aaftsH greatly reduced the level of ftsH mRNA. To our knowledge, this is the first report of down-regulation of gene expression by anti-antisense RNA in bacteria. Because inactivation of alr0280 eliminated the aaftsH interference and complementation of the mutant restored the interference, we conclude that the aaftsH interference is dependent on RNase III.

The aaftsH interference resembles RNA interference (RNAi) in some aspects: it is induced by a sense RNA that may form RNA duplex with an internal antisense RNA, dependent on RNase III activity and also dependent on a Hen1 homolog. However, no gene for argonaute protein (Hutvagner & Simard, 2008), the key player in RNAi, can be found in the genome of Anabaena 7120. Unlike the plant Hen1, the cyanobacterial homolog is in complex with Pnkp, and the Pnkp/Hen1 complex shows activity of ligase, in addition to methyltransferase, with colicin-cleaved tRNA (Chan et al., 2009) and probably many other small RNA as substrates (Zhang et al., 2012). Therefore, the aaftsH interference is probably based on a different mechanism.

Both aftsH and aaftsH are predicted to form extensive stem-loop structures (Fig. S5). Hypothetically, aaftsH and aftsH form dsRNAs that are processed by RNase III into small dsRNAs. If some of the small dsRNAs are converted into ssRNAs, those partially matching up with each other (for example, those from the stem-loop regions) could form stem-open loop structures. According to Zhang et al. (2012), the two ends of open loops can be efficiently ligated by the Pnkp/Hen1 complex. Due to methylation at the ligation site, such small RNAs may be accumulated and cause inhibition of translation on ftsH mRNA, leading to RNase E-dependent degradation of the mRNA (Thomason & Storz, 2010).

In Synechocystis 6803, anti-isrR does not induce the degradation of isiA mRNA (Dühring et al., 2006). Therefore, sense RNA interference may be limited to Anabaena sp. It remains to be tested whether artificial internal sense RNA or dsRNA can induce the degradation of other mRNAs in Anabaena 7120. At least, our results showed a key role of RNase III in the aaftsH interference. Also, for the first time, cyanobacterial Hen1 was related to the regulation of mRNA level.


This project was supported by the National Natural Science Foundation of China (Grant 31270132). The authors declare that there is no conflict of interest.