Regulation of fixLJ by Hfq Controls Symbiotically Important Genes in Sinorhizobium meliloti

The soil bacterium Sinorhizobium meliloti engages in a symbiosis with plants from genera Medicago, Melilotus, and Trigonella, which leads to the development of new plant organs (nodules). Inside nodules, rhizobia differentiate into bacteroids and carry out nitrogen fixation by converting atmospheric dinitrogen into ammonium. The establishment and productivity of S. meliloti–legume symbiosis depend on a signal exchange leading to the precisely timed regulation of gene expression in both partners (Gibson et al. 2008; Long 2001). The symbiosis is initiated when rhizobia chemotax along the gradient of flavonoids toward plant roots and, then, express early nodulation genes, resulting in the production of Nod factors (Peters et al. 1986). Nod factors, along with other cues, induce a number of physiological and anatomical accommodations in plants leading to intracellular invasion by rhizobia, their subsequent differentiation into bacteroids, and eventually, nitrogen fixation. Nitrogen fixation is catalyzed by nitrogenase, which is encoded by the bacterial nif genes (Zimmerman et al. 1983). While plants obtain fixed nitrogen from their bacterial symbionts, bacteria gain energy and C skeletons in the form of dicarboxilic acids (Yurgel and Kahn 2004). The plant host and bacteroids probably also exchange amino acids. For example, peas were shown to supply leucine, isoleucine, and valine to Rhizobium leguminosarum in exchange for aspartate (Lodwig et al. 2003).

Microarray studies comparing S. meliloti gene-expression patterns in free-living cells with those in bacteroids have shown that a number of genes are induced inside nodules (Ampe et al. 2003; Becker et al. 2004). The majority of these genes are regulated by the oxygen-sensing two-component regulatory system FixLJ, which plays a major role in controlling the expression of the late nodulation nif and fix genes within bacteroids (David et al. 1988). Concurrently on the plant side, changes in gene expression ensure production of plant leghemoglobins in the cytoplasm of the nodule cells containing bacteroids (Cutting and Schulman 1971; Zimmerman et al. 1983). Leghemoglobins have high affinity for oxygen and create a microaerobic environment, which is perceived by FixLJ and is suitable for the activity of the oxygen-labile nitrogenase (Fischer 1994).

FixLJ-mediated signal transduction has been well-studied in S. meliloti. The fix and nif genes mapped to the symbiotic megaplasmid (pSymA) (Barnett et al. 2001). S. meliloti fixL or fixJ mutants are able to form nodules, but no nitrogen fixation takes place within them (Fix⁻ phenotype) (Zimmerman et al. 1983). This regulatory cascade consists of the response regulator FixJ and its cognate sensor kinase FixL, transcriptional activators encoded by nifA and fixK, as well as nitrogenase genes (nifHDK) (Fischer 1994). The microaerobic environment induces the fixLJ operon within bacteroids, FixLJ then activates nifA and fixK genes, and NifA activates nitrogenase genes (David et al. 1988). In Azorhizobium caulinodans, Rhizobium leguminosarum, and S. meliloti, nifA and fixK are also subject to regulation by the RNA-binding protein Hfq (Kaminski et al. 1994; Torres-Quesada et al. 2010; Zhang and Hong 2009). In fact, recent studies identified Hfq as an important regulator of gene expression in the legume-rhizobium symbiosis (Barra-Bily et al. 2010; Gao et al. 2010; Torres-Quesada et al. 2010, 2014).

Hfq is a posttranscriptional regulator found in gram-negative and gram-positive bacteria. Hfq is a chaperone that is capable of actively unfolding small (s)RNA and messenger (m)RNA to either facilitate their base-pairing or block access to the ribosome-binding site (Brennan and Link 2007; De Lay et al. 2013; Soper et al. 2010). Binding of sRNA and mRNA to Hfq may also increase local concentrations of RNA and this alone may facilitate base pairing (Brennan and Link 2007). The proximal and distal faces of Hfq bind specific sequences in sRNA and mRNA. At least initially, uridine-rich sequences within sRNA bind to the proximal face and are thought to then move to the lateral site (rim), while adenine-rich ARN repeats in mRNA interact with the distal face of the protein (De Lay et al. 2013; Link et al. 2009; Schumacher et al. 2002). Functional importance of (ARN)n motifs in mRNAs was experimentally validated in the 5′ untranslated region (UTR) of Escherichia coli rpoS and glmS mRNA (Robinson et al. 2014) and within S. meliloti expR (Gao et al. 2015). Even though nearly all functionally important (ARN)n sites are within the 5′ UTR, internal Hfq-binding sites have been also identified (De Lay et al. 2013).

S. meliloti hfq mutants form nodules but are unable to regulate symbiotically relevant genes and exhibit a nitrogen-starvation phenotype when inoculated on alfalfa and Medicago truncatula roots (Barra-Bily et al. 2010; Gao et al. 2010; Torres-Quesada et al. 2010). Nodules formed by the hfq mutants were small and less pink due to the reduced levels of leghemoglobin. Bacteroids formed by the hfq mutants in the indeterminate M. truncatula nodules were similar to the type III and type IV bacteroids described by Vasse and associates (1990), similar to those formed by fixL and fixJ mutants (Gao et al. 2010; Vasse et al. 1990), therefore suggesting a link between the phenotypes controlled by Hfq and FixLJ.

In this study, we combined in silico predictions, genetic, and biochemical analyses to further characterize regulatory functions of the S. meliloti RNA-binding protein Hfq in M. truncatula nodules. We tested a hypothesis that the effects of Hfq on the expression of symbiotic genes are mediated by the direct interactions of Hfq with fixLJ.

RESULTS

Bioinformatic predictions and validation of the symbiotic Hfq regulon.

To characterize the symbiotic Hfq regulon, bioinformatic predictions of its potential members were carried out (the PERL script can be found in the Supplementary Materials file). Of the transcripts identified in this bioinformatics analysis, Sma0854 (nodG), Smc00241 (transcriptional regulator), Sma1227 (fixJ), Sma2379 (catalase), Smc01033 (arylesterase), and small regulatory RNA SmrC15 were previously reported as differentially regulated by Hfq in shake cultures and SmrC15 was also shown to be under the Hfq control in nodules (del Val et al. 2007; Gao et al. 2010; Torres-Quesada et al. 2014). There is only partial overlap with the previously reported Hfq regulon; however, it is important to point out that previous high-throughput studies were carried out in shake cultures. However, only transcripts reported to be differentially regulated during the symbiosis were included in this in silico screen. The list of genes resulting from the bioinformatic analysis was further narrowed to 11 genes with known or hypothesized functions in the specificity of symbiotic interactions and efficiency of nitrogen fixation (Table 1) (Ampe et al. 2003; Becker et al. 2004; Bobik et al. 2006). It includes nodulation genes nodI, nodJ, nodG, symbiotic and nitrogen fixation genes fixJ, nifK, and noeB, EPSII regulatory gene mucR, putative regulatory genes Sma0281, Smc00241, and Smc00370, and putative pilus gene Sma1564. To test the hypothesis that Hfq affects accumulation of the corresponding transcripts, reverse transcription-quantitative polymerase chain reaction (RT-qPCR) was carried out using RNA extracted from nodules formed by the wild-type S. meliloti and its isogenic hfq mutant. Two S. meliloti sRNAs, SMrC15 and SMrC45, known to interact with Hfq (Gao et al. 2010; Torres-Quesada et al. 2010; Voss et al. 2009) served as controls. SMrc15 is a homolog of AbcR1 in Agrobacterium tumefaciens and AbcR1, which was shown to control the uptake of the nonproteinogenic γ-aminobutyric acid (Wilms et al. 2011), a plant-generated signaling molecule recognized by Agrobacterium tumefaciens. Even though Hfq binds to both of these sRNAs, SMrC15 is expressed in nodules, while SMrC45 is known not to be expressed strongly under the same conditions (del Val et al. 2007).

Table 1. Hfq-responsive expression of Sinorhizobium meliloti genes in young nodules on main roots of Medicago truncatula seedlings

Gene ID	Gene name	Fold change in gene expressiona
SMa1227	fixJ	13.3 ± 0.5
SMa0829	nifK	34.35 ± 0.4
SMa0774	noeB	6.9 ± 0.9
SMa0863	nodJ	2.0 ± 0.3
SMa0864	nodI	1.8 ± 0.2
SMa0854	nodG	3.2 ± 0.4
SMc00058	mucR	1.5 ± 0.1
SMa0281	Sma0281	1.5 ± 0.4
SMc00241	Smc00241	1.9 ± 0.4
SMc00370	Smc00370	1.1 ± 0.2
SMa1564	Sma1564	1.3 ± 0.2
SMrc15	Smrc15	19.9 ± 3
SMrc45	Smrc45	2.1 ± 0.8

^aWild-type versus Δhfq mutant. The fold change in the expression of target genes relative to the expression of the internal control gene (SMc00128) was documented. The mean fold change in the expression of each target genes was calculated using cycle threshold (Ct) values, using the 2^−ΔΔCt method (Livak and Schmittgen 2001), and was normalized to the expression of SMc00128 as previously described (Gao et al. 2010). Errors represent standard error of the mean. Each mean value was from three different cDNA samples and quantitative polymerase chain reactions (n = 3), as described before (Gao et al. 2010). Each cDNA sample was made from pooled RNA of three different nodule RNA preparations for each tested strain, as in (Ampe et al. 2003). The experiments were repeated twice.

View as image HTML

Accumulation of mRNA corresponding to the late nodulation genes (fixJ, nifK, and noeB) was clearly increased in the wild-type background (Table 1). Accumulation of mRNA of early nodulation genes (nodI, nodJ, nodG) in the nodules was only modestly affected by the deletion of hfq; however, these early nodulation genes are not strongly expressed inside the nodules in the first place (Ampe et al. 2003). No strong effects of hfq on the accumulation of mRNA of mucR, Sma1564, Smc00370, Smc00241, and Sma0281 were observed (Table 1). Consistent with previous reports that hfq affected accumulation of the small RNA SMrC15, corresponding changes in accumulation of this sRNA were observed in the nodules formed by the wild-type S. meliloti but not by the hfq mutant (Table 1).

While positive effects of Hfq on the expression of fixK were previously reported (Torres-Quesada et al. 2010), Hfq control of either fixJ or nifK has not been previously demonstrated. With subsequent experiments, we decided to determine whether Hfq directly interacts with its targets (fixJ, nifK, noeB) or whether the regulatory effects are mediated via the two-component system FixLJ. To test this hypothesis, we first determined whether Hfq can directly bind to fixL, fixJ, nifH, nifK, and noeB transcripts in vitro. As shown in Figure 1, Hfq was capable of binding to these five transcripts (and SmrC45 RNA control) in vitro. In control experiments, Hfq weakly interacted with poly-C₃₀ RNA (Fig. 1; Supplementary Fig. S1). However, binding of poly-C₃₀ to Hfq appears nonspecific, based on the fact that the appearance of the shift was not dependent on the concentration of poly-C₃₀ (Fig. 1G). The addition of unlabeled fixL (a presumed direct target of Hfq) did not disrupt the weak shift in the poly-C₃₀+Hfq reactions (Fig. 1G), which is consistent with the nonspecific binding of poly-C₃₀ to Hfq. However, when unlabeled poly-C₃₀ was incubated with biotinylated fixL, it partially disrupted binding of fixL to Hfq only when poly-C₃₀ was added in at least 10-fold excess. However, unlabeled fixL was a much stronger competitor than poly-C₃₀. These observations are consistent with previously reported weak interactions of Hfq and poly-C RNA fragments (Zhang et al. 2002). Therefore, with follow-up experiments, we attempted to determine whether interactions between Hfq and fixL, fixJ, nifH, nifK, and noeB occur during normal symbiotic interactions.

Download as PowerPoint

fixLJ mRNA coimmunoprecipitates with Hfq.

To follow up on the results of the RT-qPCR and in vitro RNA-binding experiments, coimmunopreciptation (co-IP) experiments were carried out. A S. meliloti strain carrying N-terminally hexahistidine-tagged hfq on a medium–copy number plasmid was used. Functionality of this construct has been previously confirmed, as this plasmid-borne tagged hfq construct was shown to complement wild-type levels of nodulation and growth under a variety of conditions (Gao et al. 2010).

For the co-IP experiments, frozen nodules were ground, cells were lysed, and the lysates were incubated with the anti-His antibody. RNA was heat-released from the Hfq-antibody complexes following extensive washes. sRNA SmrC15 and SmrC16 were readily detectable in the fraction that was heat-released from the anti-His/Hfq^His-RNA complexes (Fig. 2), consistent with previous reports that these two sRNA interacted with Hfq and were expressed in nodules (del Val et al. 2007; Gao et al. 2015; Torres-Quesada et al. 2010). Furthermore, Smrc45 sRNA was not significantly enriched in the same fraction, consistent with the report that, even though this sRNA interacts with Hfq under laboratory conditions, it is not expressed in nodules (del Val et al. 2007). Finally, a nonspecific control Smrc7 sRNA in the Hfq co-IP experiment was enriched by 3.1 ± 0.7 fold (Fig. 2B). Smrc7 was selected as control because it is known to be expressed in both free-living bacteria and in Sinorhizobia nodules, with strong increase in expression within mature nodules, as demonstrated by Northern hybridization (del Val et al. 2007). Neither the accumulation of the Smrc7 transcript nor its half life in S. meliloti was affected by an hfq mutation (Gao et al. 2010; Torres-Quesada et al. 2014) (Fig. 2C). Therefore, co-IP experiments were technically sound.

Download as PowerPoint

The fixLJ mRNA was found in the fraction that was heat-released from the anti-His/Hfq^His complex from nodules (10 and 18 days postinoculation [dpi]) (Fig. 2) but not in the similarly prepared control samples containing untagged Hfq protein (Fig. 2). Transcripts corresponding to nifHDK mRNA and noeAB were enriched only approximately 1.5 to fourfold by co-IP, similar to that of nonspecific control Smrc7 RNA (3.1 ± 0.7). Expression levels of nifK and noeB can play an important role in determining their binding to Hfq in planta; more abundant transcripts may be overrepresented in the pool of mRNA bound and subsequently released from Hfq. For example, expression levels of nifK in mature nodules could be quite high. The accumulation of nifK was strongly induced in M. truncatula nodules (6.1-fold in 18-dpi nodules and 9.6-fold in 8-dpi nodules in microarray experiments [Ampe et al. 2003]). NifK protein is also abundant in the nodules (Djordjevic 2004). Recovery of the nifK transcripts from the pool of transcripts bound to Hfq was higher (3.6 ± 0.9) than the background. In contrast, expression levels of noeB in mature nodules are probably low, even though they were detectable by microarray (Bobik et al. 2006). NoeA and NoeB proteins are not as abundant as NifK (Djordjevic 2004).

Collectively, results of the RT-qPCR experiments and in vivo co-IP suggest that, under symbiotic conditions, Hfq strongly associates with fixLJ transcripts and the Hfq does not associate with nifHDK or noeAB transcripts under the same conditions. It is possible that Hfq exerts its control over nifHDK and noeAB via FixLJ (Bobik et al. 2006; Gong et al. 2006). Electrophoretic mobility shift assay (EMSA) experiments (Fig. 1) confirmed the results of the bioinformatic analysis, showing fixL, fixJ, nifH, nifK, and noeB contain sites that can be bound by Hfq under some conditions (i.e., in vitro). However, given that only fixJ bound to Hfq in planta and was subsequently released from this chaperone in vitro, it is also possible that binding of these transcripts to Hfq under symbiotic conditions may not be sufficiently strong to be detected under these experimental conditions.

Partial restoration of the Fix⁻ phenotype of hfq with fixLJ.

To test the hypothesis that the symbiotic defects of hfq are mediated by FixLJ, we carried out epistasis experiments. A 2.5-kb fragment of S. meliloti DNA containing a wild-type copy of the fixLJ operon and a presumed fixL promoter was cloned into a broad host range–medium copy number plasmid (Supplementary Fig. S2). The accumulation of fixLJ transcripts in strains 8530Δhfq pFixLJ (hfq, fixLJ+) and 8530Δhfq pHfq (hfq, hfq+) was compared with that in the hfq mutant carrying only the vector pBBR1MCS-5 (Table 2). Levels of fixLJ transcripts in 8530Δhfq pFixLJ in nodules (10 and 18 dpi) were 4.3- to 16-fold higher than in the control strain. However, this represented only partial restoration compared with the strain in which the hfq mutation was complemented with a plasmid-borne copy of hfq (Table 2). These results further support the hypothesis that a functional hfq is required for the posttranscriptional stability of fixLJ.

Table 2. Expression of fixL and fixJ in shake cultures and in planta

Sample	Relative expressiona
	8530 Δhfq pHfq		8530 Δhfq pFixLJ
	fixL	fixJ	fixL	fixJ
Shake cultureb	1.5 ± 0.7	1.3 ± 0.2	1.3 ± 0.3	1.7 ± 0.3
10 dpi	27.0 ± 2	11.3 ± 0.6	16 ± 2	11.3 ± 0.6
18 dpi	24 ± 1.0	27 ± 0.8	10.5 ± 0.5	4.3 ± 1

^aMean fold change relative to 8530Δhfq pBBR1MCS-5. The fold change in the expression of target genes relative to the expression of the internal control gene (SMc00128) was calculated. The mean fold change in the expression of each target gene was calculated using cycle threshold (Ct) values, using the 2^−ΔΔCt method (Livak and Schmittgen 2001), and was normalized to the expression of SMc00128 as previously described (Gao et al. 2010).

^bIn the shake culture experiment, cells were harvested at OD₆₀₀ = 0.5 to 0.6 each cDNA sample was made from a pooled RNA of three different RNA preparations for each tested strain as described in (Gao et al. 2010). Errors represent standard error of the mean. Experiments were repeated twice. dpi = days postinoculation.

View as image HTML

The expression of fixLJ from pFixLJ in 8530Δhfq partially rescued nitrogen starvation of M. truncatula, as evidenced by better growth and green color of the plants. However, restoration of the phenotype was incomplete (Fig. 3A). At 18 dpi, average shoot height of the plants inoculated with 8530Δhfq pFixLJ was 3.3 ± 0.3 cm (n = 13), which was significantly higher (P < 0.05) than average shoots of plants infected with 8530Δhfq (2.2 ± 0.3 cm, n = 15) but less than that of plants infected with 8530Δhfq pHfq (4.0 ± 0.3 cm, n = 15). Nodules formed by 8530Δhfq pFixLJ (10 dpi) were ovoid and clearly more advanced than the spherical nodules formed by 8530Δhfq (Fig. 3B), however, they were not as developmentally advanced as the nodules formed by 8530Δhfq pHfq (Fig. 3B). Furthermore, the intensity of the pink color (indicative of the abundance of leghemoglobin) was higher in the nodules formed by 8530Δhfq pFixLJ and the sizes of the nodules were significantly bigger than the nodules formed by the mutant (Fig. 3C; Supplementary Table S1).

Download as PowerPoint

Hfq controls nifHDK via FixLJ in planta.

Because fixLJ supplied in trans partially complemented the symbiotic defect of the hfq mutant, we tested whether fixLJ can control the nifHDK operon involved in the synthesis of nitrogenase in the hfq mutant background. These experiments tested whether the regulation of nifHDK genes is entirely through FixLJ or whether Hfq also contributes to the posttranscription regulation of nifHDK in nodules. To this end, levels of nifHDK transcripts in shake cultures and in nodules formed by S. meliloti 8530Δhfq pFixLJ and by S. meliloti 8530Δhfq pBBR1MCS-5 (vector control) were measured. Only a modest (1.7 to 2.6-fold) effect of pFixLJ on the accumulation of these transcripts was observed in midlog shake cultures (Table 3). However, nif genes are known not to be expressed in well-oxygenated laboratory cultures (David et al. 1988; Fischer 1994). Substantial increases in the accumulation of these transcripts were observed in nodules formed by the hfq mutant carrying pFixLJ. Levels of nifHDK transcripts in nodules formed by 8530Δhfq pHfq were higher than those in the nodules formed by 8530Δhfq pFixLJ. In 18-dpi nodules, levels of nifH transcripts in 8530Δhfq pHfq were 2.1-fold higher than in 8530Δhfq pFixLJ and the level of nifDK was 3.5-fold higher than in the control, while the difference in the accumulation of the nifK transcript was approximately 2.2. Similar trends were observed in 8- and 18-dpi nodules. However, pFixLJ in the hfq background did not fully complement the mutation. This is either due to the effects of Hfq on nifH and nifDK that are independent of FixLJ or, alternatively, the fixLJ transcript may not be fully stabilized in the hfq background, even when supplied in trans. Nevertheless, it is clear that, at least in part, the regulatory effects of Hfq on the nitrogenase genes are through the control of fixLJ.

Table 3. Expression of nifH, nifD and nifK in Sinorhizobium meliloti 8530Δhfq in shake culture and in nodules (epistasis experiments)

Sample	Relative expression (fold change)a
	8530Δhfq pHfq			8530Δhfq pFixLJ
	nifH	nifDK	nifK	nifH	nifDK	nifK
Shake cultureb	1.7 ± 0.4*	2.6 ± 0.7	2.0 ± 0.3	1.2 ± 0.3	2.0 ± 0.2	1.6 ± 0.9
8 dpi	45.3 ± 1.9	36.5 ± 2.0	21.6 ± 1.4	26 ± 2.8	10.6 ± 0.3	7.7 ± 1.8
10 dpi	45.3 ± 4	25.6 ± 1.3	34.3 ± 1.5	28.6 ± 1.3	6.6 ± 1.2	16.0 ± 1.2
18 dpi	270.6 ± 1.2	68.6 ± 2.2	69.1 ± 1.4	127.1 ± 1.87	19.7 ± 2	30.5 ± 2.7

^aThe fold change in the expression of target genes relative to the expression of the internal control gene (SMc00128) was calculated. The mean fold change in the expression of each target genes was calculated, using cycle threshold (Ct) values using the 2^−ΔΔCt method (Livak and Schmittgen 2001), and was normalized to the expression of SMc00128, as previously described (Gao et al. 2010).

^bIn the shake culture experiment, cells were harvested at an optical density at 600 nm = 0.5 to 0.6. Each cDNA sample was made from a pooled RNA of three different RNA preparations for each tested strain, as described by Gao et al. (2010). Errors represent standard error of the mean. Experiments were repeated twice.

View as image HTML

To further support the results of the RT-qPCR experiments, a merodiploid nifH::gus reporter fusion was constructed in S. meliloti 8530 Δhfq. Expression of the reporter was tested in Δhfq and in the Δhfq strain carrying pFixLJ. As shown in Table 4 and consistent with the results of the RT-qPCR analyses, pFixLJ increased expression of the nifH::gus reporter.

Table 4. Endosymbiotic β-glucuronidase (GUS) activity of nifH-gus promoter fusion

Strain	GUS activitya
	10 dpi	18 dpi
8530Δhfq-0102	76 ± 33	113 ± 17
8530Δhfq-0102 pFixLJ	1,192 ± 233	1,104 ± 218

^aMeasured as nanomoles per minute per optical density at 595 nm × 1,000. dpi = days postinoculation. Activities are expressed as previously described (Bittner and Oke 2006). Averages of three independent assays for three different nodule samples are shown. Nodules were harvested from primary infection zones that are susceptible to infection (Caetano-Anollés and Bauer 1988; Gao et al. 2010). Errors represent standard error of the mean.

View as image HTML

Characterization of Hfq binding motifs upstream and within the fixLJ open reading frame (ORF).

Collectively, co-IP, gel shifts, gene expression, and epistasis experiments indicated that Hfq likely interacts with fixLJ. However, in the co-IP experiments, fixL mRNA was enriched 44-fold within the Hfq^his-bound fraction, while fixJ mRNA was enriched only about 15-fold. Therefore, to better understand how Hfq impacts fixLJ, we focused on the characterization of the effects of this RNA-binding chaperon on fixL. Two putative Hfq-binding (ARN)₂ motifs (motif 1, AAGACG and motif 2, AAAAAG) were detected upstream of and within fixL (Fig. 4A). Motif 1 is identical to one of the two functional Hfq-binding sites in the S. meliloti expR (Gao et al. 2015). There were at least three other (ARN)n regions within fixLJ transcript (data not shown); however, their effects on interactions with Hfq were not tested in this study.

Download as PowerPoint

To test whether motifs 1 and 2 have an Hfq-dependent role in the expression of fixL, we constructed a set of S. meliloti translational fixL::gus reporters that lack either one or both of the motifs. These fusions were borne on integrative plasmids pFixL24, pFixL30, pFixL2a, and pFixL2 (Fig. 4B). Each of the fixL-gus fusions was mobilized into either the wild type or an hfq mutant of S. meliloti, generating single-copy merodiploid fixL⁺ fixL::gus reporters integrated into pSymA (Fig. 4C). The expression of fixL gene was analyzed using RT-qPCR and with β-glucuronidase (GUS) assays (Figs. 4D and E). In the absence of hfq, expression of fixL was reduced by 60- to 100-fold for all constructs. In the wild-type background, expression of fixL was the highest in the construct carrying an intact upstream region of fixL (including both motifs). Mutations of either of the motifs reduced fixL expression. Deletion of both motifs had the strongest effect on the expression of fixL (Fig. 4D).

Subsequent GUS assays of the fixL-gus fusion activities confirmed RT-qPCR results (Fig. 4E). Motifs 1 and 2 were critical to the expression of fixL in the wild type but not in hfq backgrounds. Deletion of motif 1 has a stronger (approximately 20%) effect on the expression of fixL than the deletion of motif 2 (Fig. 4E). It is possible, however, that the reduction in the expression of fixL following mutagenesis of these motifs could also be due to the mutations in the promoter region of fixL. To test this hypothesis and to determine whether Hfq induces expression from the fixL promoter, a pSymA-integrated transcriptional fixL-gus fusion, fixL23, was tested for Hfq-induced GUS activity in planta.

A deletion of hfq resulted in an approximately 1.4 reduction in the activity of the fixL transcriptional fusion in nodules 10 dpi and approximately 2.3 reduction in the activity 18 dpi (Table 5). This ability of Hfq to exert transcriptional control over fixL expression explains activation of the fixL translational reporters in the hfq-dependent but motif-independent manner (Figs. 4D and E). However, because Hfq is not a DNA-binding protein, this activation of the fixL transcriptional fusion suggests that there may be another layer of fixL regulation, with Hfq controlling another upstream regulator, in addition to controlling posttranscriptional regulation of fixL.

Table 5. Endosymbiotic β-glucuronidase (GUS) activity of fixL-gus promoter fusion

Strain	GUS activitya
	10 dpi	18 dpi
8530 pFixL23	398 ± 83	375 ± 96
8530Δhfq pFixL23	280 ± 94	164 ± 51

^aMeasured in nanomoles per minute per optical density at 595 nm × 1,000. dpi = days postinoculation. Activities are expressed as previously described (Bittner and Oke 2006). Averages of three independent assays for three different nodule samples are shown. Nodules were harvested from primary infection zones that are susceptible to infection (Caetano-Anollés and Bauer 1988; Gao et al. 2010). Errors represent standard error of the mean.

View as image HTML

DISCUSSION

RNA-binding chaperon Hfq has already been shown to effect global changes in S. meliloti gene expression in shake cultures (Gao et al. 2010). Even though symbiotic defects of the hfq mutant are well-documented and include defects in nitrogen fixation and assimilation of fixed nitrogen (Barra-Bily et al. 2010; Gao et al. 2010; Torres-Quesada et al. 2010), mechanisms by which Hfq controls these functions are not clear. High throughput studies in nodules are technically complicated. Therefore, we carried out bioinformatics predictions in an attempt to delineate the symbiotic Hfq regulon. Our in silico predictions that fix and nif genes are among at least 11 plausible Hfq target genes in M. truncatula nodules were supported by previous observations that the phenotypes of plants infected with the hfq mutants and ultrastructure of bacteroids are similar to those infected with S. meliloti fixLJ and some nif mutants (Gao et al. 2010; Hirsch and Smith 1987; Vasse et al. 1990). Therefore, we focused on testing the hypothesis that Hfq controls fix and nif genes during the symbiosis.

RT-qPCR analysis of the expression of these genes in the nodules formed by S. meliloti revealed that the deletion of hfq resulted in a reduced expression of fixJ, nifK, and noeB compared with the wild type. In the nodules formed by S. meliloti carrying hfq^His (10 and 18 dpi), only fixLJ transcripts coimmunoprecipitated with Hfq, while the same experiments did not identify nifK or noeB transcripts as strongly bound to Hfq. Because the two-component system FixLJ is known to control nifK and noeB, it is reasonable to hypothesize that Hfq likely controls these via FixLJ. This hypothesis was then tested in epistasis experiments with fixLJ borne on pBBR1MCS-5. This vector appears to be stable and replicated in rhizobia for at least 4 weeks after inoculation (Gao et al. 2010) and in shake cultures (Gao et al. 2015). Consistent with our hypothesis, plasmid-borne fixLJ was able to partially restore hfq mutation both in alleviating the nitrogen-deficiency symptoms and in the expression of the downstream nifHDK operon.

Because earlier work demonstrated that S. meliloti Hfq exerts posttranscriptional control over regulatory gene expR mRNA (Gao et al. 2015) via two Hfq-binding sites with sequences similar to the (ARN)n motif of Escherichia coli (Link et al. 2009), we searched for (ARN)n motifs within the fixLJ transcript. Two (ARN)₂ motifs were found in positions that are likely consequential for posttranscriptional regulation of the fixLJ operon. Motif 1 begins 42 nt upstream of the fixL ORF and motif 2 begins at the fourth nucleotide in the ORF. Integrated translational fusions confirmed Hfq-dependent functionality of both motifs. While mutations of either one of the motifs resulted in about a twofold reduction of the activity of the translation fusion (compared with the wild type), mutagenesis of both motifs resulted in an approximately sixfold reduction of the activity in M. truncatula nodules. The mechanism of this regulation is not yet known; however, Hfq is known to be able to relax folding of mRNA around translation initiation site (Soper et al. 2010) and this may allow presentation of the fixL start codon AUG (Fig. 5). Alternatively, Hfq is also known to prevent degradation of the bound transcripts by RNase by masking multiple plausible recognition sites of this enzyme (Baumgardt et al. 2014). Homologs of RNase recognition sites are found in fixL mRNA, including one in motif 1 and another in motif 2 (Fig. 5), although we do not yet know whether these recognition sites are truly functional RNase recognition sites.

Download as PowerPoint

Hfq typically facilitates base-pairing between mRNA and sRNA bound to different sites (“faces”) of the protein. The sRNA that may be controlling expression of fixL in the Hfq-dependent manner is not yet known. Motif 2 (AAAAAG) appears to overlap a C/A-rich element (CAAAAAGC). Some C/A-rich elements were shown to function as sRNA-binding sites in target mRNA, as exemplified by the binding of sRNA GcvB (Sharma et al. 2011).

While this study provided a first glimpse into the symbiotic functions of Hfq by controlling fixLJ, much remains to be learned about the intricacies of Hfq regulation in planta. S. meliloti Hfq controls many genes, including small regulatory RNA expressed in nodules as well as sRNA that are induced in microaerobic environments (Torres-Quesada et al. 2014). While studies in microaerobic laboratory shake cultures are informative, they do not always reveal the complexity and consequences of the bidirectional exchange between the bacterium and its plant host. Further experiments are needed to continue defining the role of Hfq and the sRNA in controlling the symbiotic productivity.

MATERIALS AND METHODS

Bioinformatics analysis.

mRNA that were previously reported as expressed in nodules and in the bacteroids (Ampe et al. 2003; Becker et al. 2004; Bobik et al. 2006) were scanned with ad hoc PERL script to detect the presence of the previously reported mRNA-Hfq binding motif (ARN)n, where A is an adenine, R is any purine nucleotide (guanosine or adenosine), N is any nucleotide, and n is tract length from 2 to 8 (Link et al. 2009). It is of note that the biochemical evidence of S. meliloti Hfq binding to (ARN)n motifs is not available; however, site-directed mutagenesis suggests that S. meliloti Hfq interacts with (ARN)n-like motifs (Gao et al. 2015).

Bacterial culture conditions.

All bacteria were maintained as frozen glycerol stocks. Cultures of Escherichia coli were grown in Luria Bertani broth with appropriate antibiotics. S. meliloti cultures were started from individual colonies on fresh tryptone-yeast (TY) agar (Beringer 1974) plates, which were inoculated into 5 ml of TY broth with appropriate antibiotics. Cultures were grown at 200 rpm and 30°C until optical densities at 600 nm [OD₆₀₀] reached 0.5 to 0.6 (for experiments in midlog phase).

Plant material and inoculation protocol.

M. truncatula cv. A17 seedlings were aseptically germinated in the dark and were then transferred into growth pouches and were grown as previously described (Gao et al. 2005, 2010). Seedlings were watered with ¹/₄ strength of nitrogen-free Jensen’s medium. After 24 h of growth, plants were inoculated with the S. meliloti midlog cell suspensions (200 µl of water per seedling) near root tips as before (Gao et al. 2005, 2010). To ensure that nodules were developmentally synchronized (Caetano-Anollés and Bauer 1988), growth pouches were marked to indicate the position of root tips at the time of inoculation (RT1). After 36 h, positions of root tips were marked again (RT2). Eight, 10, and 18 dpi, nodules formed on or within the RT1-RT2 zone were separated from roots with a toothpick by a gentle push at the base of a nodule. Nodules were further processed or frozen in liquid nitrogen.

Strain and strain construction.

Strains used in this study are listed in Table 6. All S. meliloti mutants are isogenic derivatives of S. meliloti 8530 (Pellock et al. 2002). Sequences of primers used for the strain construction are in Supplementary Table S2.

Table 6. Bacterial strains and plasmids used

Strain or plasmid	Descriptiona	Reference
Sinorhizobium meliloti strains
Rm8530	Rm1021, expR+ Smr	Pellock et al. 2002
Rm8530 Δhfq	In-frame hfq deletion mutant, Smr	Gao et al. 2015
Rm8530xpFixL24	Translational fixL::gus reporter construct, pFixL24 integrated in pSymA in Rm8530, Nmr	This work
Rm8530xpfixL30	Translational fixL::gus reporter construct, pFixL30 integrated in pSymA in Rm8530, Nmr	This work
Rm8530xpfixL2a	Translational fixL::gus reporter construct, pFixL2a integrated in pSymA in Rm8530, Nmr	This work
Rm8530xpfixL2	Translational fixL::gus reporter construct, pFixL2 integrated in pSymA in Rm8530 Nmr	This work
Rm8530ΔhfqpfixL24	Translational fixL::gus reporter construct, pFixL24 integrated in pSymA, Hfq, Nmr	This work
Rm8530ΔhfxqfixL30	Translational fixL::gus reporter construct, pfixL30 integrated in pSymA, hfq, Nmr	This work
Rm8530ΔhfqfixL2a	Translational fixL::gus reporter construct, pFixL2a integrated in pSymA, hfq, Nmr	This work
Rm8530ΔhfqfixL2	Translational fixL::gus reporter construct, pFixL2 integrated in pSymA, hfq, Nmr	This work
Rm8530xpfixL23	Transcriptional fixL::gus reporter construct, pFixL23 integrated in pSymA in Rm8530, Nmr	This work
Rm8530ΔhfxqfixL23	Transcriptional fixL::gus reporter construct, pFixL23 integrated in pSymA, hfq, Nmr	This work
Rm8530-01/02	Transcriptional fixL::gus reporter construct, p01/02-nifH integrated in pSymA in Sm8530, Nmr	This work
Rm8530Δhfq-01/02	Transcriptional fixL::gus reporter construct, p01/02-nifH integrated in pSymA, Hfq, Nmr	This work
Escherichia coli strain
DH5α	Competent cells	Invitrogen
Plasmids
pBBR1MCS-3	Broad-host-range cloning vector, Tcr	Kovach et al. 1995
pBBR1MCS-5	Broad-host-range cloning vector, Gmr	Kovach et al. 1995
pRK600	Conjugal transfer helper plasmid, Gmr	Finan et al. 1986
pVMG	gus fusion vector, ColE1, oriT, Nmr	Gao et al. 2005
pHfq	1.6-kp SpeI-XbaI fragment of hfq from 1021, Tcr	Gao et al. 2010
p#5	HfqHis expression construct, Gmr	Gao et al. 2015
pFixLJ	2.5-kb SpeI-KpnI fragment of fixLJ gene, Gmr	This work
pMG78/80-17	340-bp A 370-bp S. meliloti fixL motif determinant in pVMG, intermediate plasmid	This work
pFixL23	A transcriptional fixL-gus fusion in pVMG, Nmr	This work
pFixL24	A translational fixL-gus fusion in pVMG, motif 1+, motif 2+, Nmr	This work
pFixL30	A translational fixL-gus fusion in pVMG, motif 1+, motif 2− Nmr	This work
pFixL2a	A translational fixL-gus fusion in pVMG, motif 1−, motif 2+ Nmr	This work
pFixL2	A translational fixL-gus fusion in pVMG, motif 1−, motif 2− Nmr	This work
pVMG01/02-nifH	A transcriptional nifH-gus fusion in pVMG, Nmr	This work

^aSm = streptomycin, Nm = neomycin, Gm = gentamicin, and Tc = tetracycline.

View as image HTML

Construction of pFixLJ.

A 2.5-kb DNA fragment containing fixLJ from wild-type S. meliloti 8530 was PCR-amplified with primers MG3078 and MG3131 and Taq DNA polymerase (New England Biolabs). The amplified fixLJ fragment was digested with SpeI and KpnI purified by Illustra GFX PCR DNA and the gel-band purification kit (GE Healthcare), and was then ligated into SpeI-KpnI–digested pBBR1MCS-5. The orientation of the fixLJ insert was determined by restriction digest and DNA sequencing.

Construction of gus fusions.

To construct fixL-gus fusions, a 370-bp region spanning the start codon of S. meliloti fixL (340-bp DNA upstream of ATG for fixL, which contains motif 1 plus 30 bp within the coding region (which contains motif 2), was amplified with primers MG3078 and MG3080, using Taq polymerase (New England BioLabs). The amplified fragment was purified by Illustra GFX PCR DNA and the gel purification kit (GE Healthcare) and was then cloned in front of the promoterless GUS gene (gus) in pVO155 between the SpeI and XbaI sites, using T4 DNA ligase to yield an intermediate plasmid, pMG78/80-17. pMG78/80-17 was digested with XbaI and was then treated by S1 nuclease (Promega), purified, and self-ligated, using T4 DNA ligase, resulting in derivatives, each with a unique deletion from the XbaI site of its parental plasmid. Sequencing of 11 of these derivatives confirmed transcriptional fixL-gus fusion plasmid pFixL23 with the junction sequence ATGCTTTCATAGGCGGATCTCGACCATGGTCCGT (start codons for fixL and gus are shown in bold italics and the stop codon TAG for fixL is shown in bold, the Shine-Dalgarno sequence for gus is underlined). Translational fixL-gus fusion plasmid pFixL24 with its in-frame junction sequence ATGCTTTCAAAAAGCATCTCGGTCCGTCCT (motif 2 is underlined) and ranslational fixL-gus fusion plasmid pFixL30 with its in-frame junction sequence ATGGTCCGTTTC were similarly constructed.

The motif 1 in pFixL24 was mutated (AAGACG to tAGtCG) by site-directed mutagenesis via PCR, using the following primer sets: MG3191and MG3192-tt, MG3193-tt and MG3196-BglII. PCR-generated motif 1 fragments were purified as above and were diluted and mixed, and the mixture was used as a template for a second round of PCR, using primer set MG3191-BamHI and MG3196-BglII. The PCR-generated motif 1 mutant fragment was digested with BamHI and BglII to replace the wild-type sequence in pFixL24 by cloning. This procedure yielded pFixL2a. DNA sequencing confirmed the presence of the mutations in motif 1. Motif 1 in pFixL30 was mutagenized as above, except pFixL30 was used in the process and that created pFixL2. PCR-generated mutations in motif 1 in pFixL2 were confirmed by DNA sequencing. fixL-gus fusions were introduced into S. meliloti by triparental matings with helper pRK600. Integration of each fixL-gus fusion was verified by PCR using fixL recombination-specific primer MG3140F and gus primer PVO155R2 (Gao and Teplitski 2008).

To construct nifH-gus fusion, a 230-bp S. meliloti nifH fragment (a 200-bp sequence upstream of nifH ATG plus 30 bp within the codon region) were PCR amplified with primers MG3101 and MG3102. The amplified fragment was purified and was cloned in front of the promoterless gus gene on pVMG (Gao et al. 2005), between SpeI and XbaI sites, using T4 DNA ligase. This yielded a transcriptional nifH-gus fusion pVMG01/02-nifH. The presence of the stop codon in frame was confirmed by DNA sequence analysis (ATGGCAGCTCTGCGTTTCTAGA, start codon of nifH is shown in bold and the stop codon in bold italics). pVMG01/02-nifH was introduced into S. meliloti genome by triparental mating as above. Correct integration of pVMG01/02-nifH was verified by PCR, using nifH-specific primer MG3141F and PVO155R2.

RNA isolation and RT-qPCR experiments.

The bacterial pellet from a 5-ml midlog culture or 200 mg of thin powder of nodules made by grinding them in liquid nitrogen was suspended in and incubated in 1 ml of TRIZOL (Invitrogen) for 5 min at room temperature. RNA was extracted according to Invitrogen’s instructions and the pellet was resuspended in nuclease-free water (Ambion). DNA was eliminated by treatment with 5 U of RNase-free DNase (Qiagen). RNA was further purified with RNeasy MinElut cleanup kit (Qiagen), according to the manufacture’s instructions, with one modification, i.e., the sample was brought up in 700 μl (instead of 250 μl) of 100% ethanol prior to binding to the column, to prevent exclusion of small RNAs. RNA was quantified with Bioanalyzer.

First-strand cDNA was synthesized from 370 ng of total bacterial RNA using SuperScript VILO cDNA synthesis kit (Invitrogen). Gene-specific primers used for qPCR are listed in Supplementary Table S3. They were designed using Primer Expression 3.0 software. DNA sequences were downloaded from the LIPM Bioinformatics database. SMc00128 and the 16S rRNA genes were used as internal controls because cycle threshold (Ct) values of these genes are similar under a number of conditions (Becker et al. 2004) and did not differ in laboratory cultures of S. meliloti hfq mutant and its parental wild type (Gao et al. 2010). qPCR reactions were carried out with SYBR green PCR master mix (Applied Biosystems). Reactions were run on a StepOnePlus real-time PCR system (Applied Biosystems). Reaction conditions were: 40 cycles at 95°C for 15 s, 60°C for 60 s, and 72°C for 60 s, followed by a melt curve). RNA samples containing no RT were run as controls to ensure that samples were free from DNA contamination. Melt curve tests did not exhibit a second melting temperature for primer pairs used. qPCR data were processed using StepOne software version 2.2.2. Relative expression was calculated using the 2^−ΔΔCt method (Livak and Schmittgen 2001). Each sample contained pooled RNA preparations from three independent cultures, as before (Gao et al. 2010), or from nodules induced by three independent cultures of the same strain. Reactions were performed in triplicate and the values were averaged.

RNA co-IP.

Samples for the co-IP experiments from nodule lysates were prepared as previously described (Gao et al. 2015), with the following modifications. Briefly, 2 g of thin powder of nodules was suspended in 5 ml of cold cell extraction buffer (Invitrogen) and was lysed by sonication on ice. Debris were sedimented by centrifugation for 5 min at 800 × g. Five milliliters of the preparation, containing the Hfq^His-RNA complex, was incubated with 10 μg of mouse anti-His antibody (Invitrogen) for 3 h at 4°C with gentle rotation. RNA was coimmunoprecipitated onto Dynabeads with protein G (1.5 mg) (Life Technologies) by incubating consecutively with 1 ml of the RNA-Hfq–antibody complex with rotation for 10 min at room temperature. This was followed by five washes with phosphate buffered saline buffer. RNA was heat-released from Hfq^His by incubating the sample at 90°C for 2 min in water, as previously described (Gao et al. 2015; Schwab et al. 1997). Control experiments using extracts from the wild-type (i.e., hfq was not tagged with hexahistidine) nodule lysates were performed in parallel.

Purification of Hfq.

S. meliloti hfq was cloned into p15TV-L vector using the In-Fusion HD cloning kit (Clontech Laboratories) following the manufacturer’s instructions, which resulted in p15TV-L-hfq. BL21 (DE3) p15TV-L-hfq was grown to OD₆₀₀ = 0.6 at 37°C in 1 liter of terrific broth. Hfq expression was induced by the addition of isopropyl-β-d-thiogalactoside (1 mM final concentration) to the culture medium for 16 h at 16°C. Cells were harvested by centrifugation and were suspended in 30 ml of the lysis buffer (50 mM HEPES [pH 7], 300 mM NaCl, 10 mM imidazole, 2 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride). Cells were disrupted using French cell press and lysates were cleared by centrifugation. The supernatant was loaded onto Ni-NTA agarose column. The column was washed with 20 ml of the lysis buffer and, then, with 40 ml of wash buffer (50 mM imidazole). 6xHis-Hfq was eluted in 10 ml of 250 mM imidazole buffer. The eluent was then concentrated using Amicon Ultra-15 centrifugal filter units (3-kDa cutoff). 6xHis-Hfq was treated with TEV protease following the instructions from New England Biolabs to remove the hexahistidine tag. After the tag removal, the reaction was concentrated and imidazole was removed using Amicon Ultra-15 centrifugal filter units (3 kDa cut-off). It was then reloaded onto Ni-NTA agarose column to remove Tobacco etch virus (TEV) hexastidine tag and any undigested 6xHis-Hfq. After the purification, Hfq was stored in a storage buffer (25 mM HEPES, 150 mM NaCl, 0.5 mM EDTA, 15% glycerol, 50 mM arginine, 2 mM DTT) at −20°C.

RNA preparation for gel shift assays.

For mRNA synthesis, PCR products were prepared as follows: the forward primers (atgccgttgg)TAATACGACTCACTATAGGG(N)n containing 10 random nucleotides (to increase the yield or the transcription), followed by the T7 promoter and gene-specific sequences, and the reverse primer containing gene-specific sequences. All oligonucleotides are listed in Supplementary Table S4. mRNA was synthesized in vitro, using PCR templates and T7 RNA polymerase (New England Biolabs). DNA templates were then destroyed using the Turbo DNA-free kit (Ambion). mRNA was purified using a Megaclear kit (Ambion) and were labeled with biotin, using Biotin-16-dUTP (Roche Life Science).

Labeled mRNA was incubated with or without increasing amounts of purified Hfq in 20-μl reactions in the binding buffer (10 mM HEPES, 10 mM KCl, 10 mM NaCl, 1 mM MgCl₂, 0.5 mM EDTA, 2 mM DTT, 20% glycerol) for 20 min at 4°C. Where included, the “cold” competitor RNA was added to the reaction before the addition of Hfq. The samples were then mixed with glycerol to a final concentration of 10% and were loaded on native 6% polyacrylamide gel. Electrophoresis was performed in 1× tris-borate-EDTA buffer at 100 V for 70 min at 4°C and then transferred onto nylon membrane, using wet transfer (400 milliamps for 50 min). After the transfer, membranes were exposed to UV for 50 s. Biotin labeled bands were visualized using a Lightshift chemilunescent RNA EMSA kit (Thermo Scientific).

GUS assay.

Bacteria were isolated from nodules according to the method described by (Finan et al. 1991). GUS assays in S. meliloti strains carrying GUS fusions were performed with the substrate p-nitrophenyl-β-d-glucuronide, as previously described, and GUS activity calculations were carried out as according to Bittner and Oke (2006) and were expressed as nanomoles per minute per OD₅₉₅ × 1,000.

ACKNOWLEDGMENTS

This research was supported by a United States Department of Agriculture National Institute of Food and Agriculture grant 2015-67013-22837 to M. Teplitski and M. Gao. We gratefully acknowledge technical assistance by A. Benge and J. Mesa, who were both supported by University of Florida’s Institute of Food and Agricultural Sciences summer internships.