Introduction

Nitric oxide (NO) is a small, highly diffusible molecule with a myriad of biological functions across phylogenies. NO possesses a number of key features that collectively make this molecule ideally suited to its cellular signalling functions. NO is a lipophilic diatomic gas under atmospheric conditions. It has a relatively small Stoke’s radius and this in combination with its neutral charge that facilitates rapid membrane diffusion (Goretski and Hollocher 1988). Furthermore, the presence of an unpaired electron in NO supports its high reactivity with oxygen (O2), superoxide (O2 −), transition metals and thiols, which largely shape its cellular functions within the cell. The generation of NO by plants was first reported by Klepper (1979), significantly earlier than the discovery of NO biosynthesis in animals during the mid 1980s. However, the route for NO synthesis in animals is now well established: NO is generated during the conversion of l-arginine to citrulline by a family of nitric oxide synthase (NOS) enzymes (Palmer et al. 1993; Furchgott 1995).

In total, seven possible mechanisms for RNI synthesis in plants have been proposed, including enzymatic and non-enzymatic sources (Gupta et al. 2011). Interestingly, a NOS-related enzyme has been identified in the single-celled green alga, Ostreococcus tauri, which exhibited NOS activity in vitro and possessed similar properties to animal NOS enzymes in terms of the K m for arginine and the rate of NADPH oxidation (Foresi et al. 2010). However, genes encoding these enzymes have not been identified in higher plants despite the completion of numerous genome projects. Nevertheless, the production of NO and citrulline from Arg by higher plant extracts has been described and established animal NOS inhibitors strikingly diminished this activity (Delledonne et al. 2001; Durner et al. 1998; Corpas et al. 2006). These data suggest the existence of a plant NOS, which must be structurally distinct from the mammalian and O. tauri enzymes.

NO derived from nitrite through the enzymatic activity of nitrate reductase (NR) is also thought to constitute a key potential source of this molecule (Yamasaki 2000; Wilson et al. 2008; Kolbert et al. 2010). However, the efficiency of this reaction is low and it requires small oxygen tensions and high nitrite concentrations (Rockel et al. 2002). Thus, while NO clearly accumulates in response to developmental and environmental cues, the source(s) of NO biosynthesis remain controversial.

A role for NO in plant biology was first uncovered when this gaseous signalling molecule was found to be required for the establishment of disease resistance (Delledonne et al. 1998; Durner et al. 1998). Subsequently, the emerging evidence has suggested that NO also underpins a plethora of activities integral to plant growth and development (Lamattina et al. 2003). In this context, NO is thought to promote germination in the dark (Beligni and Lamattina 2000), diminish hypocotyl elongation (Beligni and Lamattina 2000), promote the growth of lateral roots and inhibit primary root elongation in the differentiation zone (Correa-Aragunde et al. 2004; Lombardo et al. 2006; Fernàndez-Marcos et al. 2012). Furthermore, NO has been proposed to control the onset of senescence (Guo and Crawford 2005), cell death (Zago et al. 2006) and to be a key regulator of flowering time (He et al. 2004). Studies utilising alfalfa suspension cultures have also implied a function for NO in the activation but not the progression of the plant cell division cycle (Ötvös et al. 2005). Taken together, these data imply that NO may have a highly complex role in the regulation of plant growth and development. However, the mechanism(s) that underpin these numerous and diverse functions remain to be established.

In animals, S-nitrosylation is emerging as a prototypic redox-based post-translational modification, which underpins NO signal function during many cellular responses (Wang et al. 2006; Yun et al. 2011). This process occurs following the addition of an NO moiety to the thiol (SH) group of a specific regulatory cysteine, within the target protein, forming an S-nitrosothiol (SNO). The site of SNO formation is routinely but not exclusively embedded within an S-nitrosylation motif (Stamler et al. 1997). This post-translational modification can subsequently regulate protein function. For example, the cysteine protease, caspase-3, required for programmed cell death in animals, is inactivated by S-nitrosylation (Melino et al. 1997). In contrast, de-nitrosylation via an unknown mechanism, leads to caspase-3 activation (Mannick et al. 1999).

Recently, a number of plant proteins have been shown to be specifically S-nitrosylated in vitro following exogenous NO exposure (Lindermayr et al. 2005). One of these proteins was methionine adenosyltransferase 1 (MAT1), which catalyses the synthesis of the ethylene precursor S-adenosylmethionine. S-nitrosylation of MAT1 at Cys114 resulted in inhibition of this enzyme in vitro (Lindermayr et al. 2006). Furthermore, Arabidopsis thaliana metacaspase 9 (AtMC9) zymogen is S-nitrosylated at its active site cysteine in vivo, which suppressed both AtMC9 autoprocessing and proteolytic activity (Belenghi et al. 2007). More recently, it has been documented that S-nitrosylation of TIR1 might be associated with auxin signalling (Terrile et al. 2012). Furthermore, in vivo S-nitrosylation of AtRBOHD at Cys890 functions to regulate NADPH activity and thus the production of reactive oxygen intermediates (ROIs) during the pathogen triggered hypersensitive cell death response (Yun et al. 2011).

Studies in animals have suggested that S-nitrosylated proteins are in dynamic equilibrium with de-nitrosylated proteins largely due to the action of glutathione (Liu et al. 2001). This antioxidant tripeptide reacts with SNOs to form S-nitrosoglutathione (GSNO) reconstituting the protein thiol as a consequence. Recently, an enzyme activity was purified, initially from Escherichia coli, which effectively turned over GSNO (Liu et al. 2001). An orthologue of the gene encoding this GSNO reductase (GSNOR) activity has now been identified in plants. Loss-of-function mutations in an Arabidopsis thaliana GSNOR (AtGSNOR1) resulted in elevated cellular levels of SNOs, while a mutation that conveyed over-expression of this gene decreased basal SNOs (Feechan et al. 2005); thus, implicating AtGSNOR1 in the control of total cellular SNO levels in plants. Importantly, the absence of AtGSNOR1 function compromised multiple modes of plant disease resistance (Feechan et al. 2005). Subsequently, this enzyme has also been shown to regulate thermotolerance and cell death control (Lee et al. 2008; Liu et al. 2001; Chen et al. 2009; Yun et al. 2011).

Previously, we identified a series of mutations in AtGSNOR1 (Feechan et al. 2005). While atgsnor1-1 and atgsnor1-2 increased the expression of AtGSNOR1, atgsnor1-3 was a null mutation. Furthermore, basal cellular SNO levels were decreased in atgsnor1-1 and atgsnor1-2 plants but strikingly increased in the atgsnor1-3 line. Here, we utilised these mutant lines to investigate the potential role(s) of AtGSNOR1 in plant growth and development. Our findings implicated AtGSNOR1 function in the control of shoot branching, hypocotyl growth, seed yield, flowering time and root development. Therefore, S-nitrosylation might regulate multiple facets of plant growth and development.

Materials and methods

Plant materials and growth conditions

The atgsnor1-1, atgsnor1-3 and atgsnor1-3R lines were described previously (Feechan et al. 2005). Seeds were sown on soil and vernalised at 4 °C for 2 days. Plants were grown in a controlled room at 22 °C under LD conditions (16-h light/8-h dark) or SD conditions (8-h light/16-h dark). The rosette leaves from 4-week-old plants of each genotype were stained with trypan blue to investigate mesophyll cell size under normal light microscopy.

For analysis of root and hypocotyl development, seeds were sown onto half-strength Murashige and Skoog media (Duchefa Biochemie, The Netherlands) supplemented with 1 %(w/v) sucrose or various hormones and with 0.8 %(w/v) agar. All MS plates were cold-treated in the dark at 4 °C for 2 days. For the hypocotyl length measurement, plates were incubated in either light or dark conditions for 7 days. 10-day-old light-grown seedlings were used for root length measurement and stained with toluidine blue for the root hair development assay.

Measurement of flowering time

Flowering time was measured by scoring the total number of rosette leaves, at bolting, excluding cotyledons. At least 17 plants grown under long- and short-day conditions were sampled for this purpose.

Gene expression analysis

Total RNA was extracted from rosette leaves of 3-week-old long-day grown plants using a Plant RNease kit (Qiagen, Crawley, UK) according to the manufacturer’s instruction. A 2 μg of total RNA was used for cDNA synthesis using the Omniscript RT kit (Qiagen, Crawley, UK) with oligo dT primers, according to the manufacturer’s instructions. Quantitative RT-PCR experiments were performed using gene-specific primers in a total volume of 20 μl containing 10 μl of 2X SYBR Green 1 Master (Roche Applied Science, UK), 0.4 μl of 10 μM gene-specific primers, and 5 μl of cDNA (20 ng/μl) on a Roche LightCycler480 (Roche Applied Science, UK). FLC and LFY gene expression were normalised relative to UBQ10, using a cDNA dilution series for each primer set in each experiment. The following primers were used: FLC forward primer GGAAGAAAAAAACTAGAAATCAAGCGAATTG, FLC reverse primer CGAGCTTTC TCGATGAGACCGTTG; LFY forward primer AAAGAACGGCTTAGATTATCTGTTCCACTTG, LFY reverse primer CATTTTTCGCCACGGTCTTTAGCA; UBQ10 forward primer TAAAAACTTTCTCTC AATTCTCTCTACCGTGA, UBQ10 reverse primer TTGTCGATGGTGTCGGAGCTTTC. Each RNA sample was assayed in triplicate. Data shown are representative trace from two independent biological replicates that gave very similar results.

For RT-PCR experiments, total RNA was extracted from 3-week-old long-day grown plants according to the time course (0, 4, 8, 12, 16, 20, 24 h). Primary cDNA was prepared from 2 μg of total RNA using the Omniscript RT kit (Qiagen, Crawley, UK) in a 10-μl reaction volume, and 2 μl of the reaction mixture was used for subsequent RT-PCR in a 50-μl reaction volume. Primers used for CO were described previously (Putterill et al. 1995; Wang and Tobin 1998; Chou et al. 2001). The following primers were used for ACTIN: ACTIN forward primer 5′-CATCAGGAAGGACTT GTACGG, ACTIN reverse primer 5′-GATGGACCTGACTCGTCATAC-3′.

Rosette area measurement and microscopy

Leaves from 4-week-old plants of each genotype were stained with trypan blue. The rosette leaves mounted on slides were measured by aerial photography. The photos were analysed using Photoshop and Image Tool: photos were transformed into black and white images by selecting the blue rosette areas using the magic wand tool in Photoshop 3.0 (Adobe Systems Incorporated, 2004). Rosette area was then measured on the transformed image using the ‘find object’ command in Image Tool (Wilcox et al. 1995), thresholding the image and selecting the white rosette area. A control rosette leaf marker of known area was included for each photo and used for correlation of the pixel number to the actual area.

Statistical analysis

All data were analysed by the student t test and the statistical significance was set at P < 0.05 and P < 0.01.

Results

atgsnor1-3 plants exhibit increased shoot branching

Enhanced development of shoot branching was apparent early after the transition to flowering in the atgsnor1-3 line (Fig. 1a–d). In order to determine the effects of the atgsnor1 mutations on shoot architecture more precisely, wild-type and atgsnor1 mutant plants were grown to maturity and their shoots closely examined. The atgsnor1-1 line produced slightly more vegetative, leaf bearing nodes, before floral transition compared to wild-type plants (Fig. 1e). However, the number of vegetative nodes that formed on atgsnor1-3 plants was 22 % less than those that were produced on wild-type plants. All these nodes have the potential to form a first order lateral inflorescence (Stirnberg et al. 2002). Quantification of this trait revealed that there was small but significant difference between atgsnor1-1, atgsnor1-3R and wild-type plants, whereas the atgsnor1-3 line produced a 194 % increase in first-order lateral inflorescences (Fig. 1f). We also determined the number of cauline and inflorescence nodes in these plant lines. With respect to cauline nodes, the number in atgsnor1-1 and atgsnor1-3 plants were increased 250 and 510 %, respectively, relative to wild type (Fig. 1g). In a similar fashion, the number of floral nodes in atgsnor1-1 and atgsnor1-3 plants was increased by 215 and 428 %, respectively, relative to wild type (Fig. 1h).

Fig. 1
figure 1

Impact of atgsnor1-3 on shoot branching. a–d Images of wild-type Col-0 (a), atgsnor1-1 (b), atgsnor1-3 (c) and atgsnor1-3R (d) plants at 6 weeks of age grown in long days. e Number of vegetative nodes in the given plant genotypes. f Number of first order lateral branches, as shown in the listed plant lines. g, h Number of cauline (g) and floral (h) nodes in the given plant genotypes. i The ratio of the total number of branches (first and higher order) to the number of first order branches as shown in the given lines. j, k Total shoot FW and lateral shoot FW per total shoot FW in the indicated plants. l Mean lateral inflorescence lengths at consecutive node positions of plants grown in long photoperiods, 9 days after the primary inflorescence started elongating. For nodes carrying a vegetative axillary shoot, lateral inflorescence length was scored as 0. Error bars represent 95 % confidence limits. These experiments were repeated at least twice with similar results. *P < 0.05, **P < 0.01 (t test)

Full size image

To quantify higher order branching, the ratio of the total number of branches (first and higher order) divided by the number of first-order branches was calculated for each shoot. These data suggested that atgsnor1-1 exhibited a small increase in this ratio compared to wild type, while atgsnor1-3 showed a striking enhancement (Fig. 1i). We also compared shoot growth in wild-type and mutant plants in terms of total fresh weight (FW) and FW distribution between the branches relative to the primary shoot (i.e. primary inflorescence and primary leaves). The total FW of shoot growth was similar between atgsnor1-1, atgsnor1-3R and wild-type plants; however, it was reduced by 47 % in the atgsnor1-3 line (Fig. 1j). Furthermore, the FW distribution between the primary shoot and cognate branches was not significantly different between atgsnor1-1, atgsnor1-3R and wild-type plants (Fig. 1k). In contrast, there was striking variation in this trait between atgsnor1-3 plants compared to these other plant lines. Therefore, the atgsnor1-3 plants had a significantly greater lateral shoot FW per total shoot FW compared to wild type. Thus, resource allocation between the primary shoot axis and lateral shoots was dramatically altered in atgsnor1-3 plants.

The magnitude and timing of axillary shoot growth is dependent upon the position of nodes along the axis of the shoot, this ultimately gives rise to a characteristic apical–basal pattern (Hempel and Feldman 1994; Stirnberg et al. 1999). During the reproductive phase of growth, there is a basipetal progression of lateral inflorescence outgrowth from axillary shoot meristems that arise even in the axis of the youngest rosette leaf primordia. To assess the potential impact of atgsnor1 mutations on lateral shoot development, the phyllotactic sequence of wild-type and atgsnor1 mutant plants was determined. We also dissected the leaves with their associated axillary shoots from the shoot axis and recorded the growth of axillary shoots at consecutive node positions 9 days after the primary inflorescence started elongating. In Arabidopsis, axillary shoots are connected with their subtending leaves as they arise from cells at the base of the leaf. In all the genotypes examined the pattern of lateral shoot growth was similar, with four apical leaves subtending lateral inflorescences of similar length and inflorescence length declining progressively at more basal positions (Fig. 1l). However, in atgsnor1-3 and atgsnor1-1 plants lateral shoot length at the more apical node positions was significantly shorter than that found in wild-type plants.

atgsnor1-3 plants exhibit reduced hypocotyl elongation

The leaves of atgsnor1 mutant plants appeared significantly different to wild-type plants with respect to their surface areas. For the quantification of leaf area exhibited by these Arabidopsis lines we utilised Image Tool (Wilcox et al. 1995), to accurately measure the leaf surface area. Our analysis suggested there was no significant difference in rosette leaf area between atgsnor1-1 and wild-type plants (Fig. 2a). However, the rosette leaf area in atgsnor1-3 plants is 141 % greater than that of wild-type plants. Conversely, atgsnor1-1 plants exhibit a leaf surface area that is only 90 % that of wild type.

Fig. 2
figure 2

AtGSNOR1 controls plant growth. a Rosette leaf area of 4-week-old plants in the given genotypes. b, c Hypocotyl length in the plant genotypes grown in dark (b) and light (c) conditions for 7 days. Error bars represent 95 % confidence limits. These experiments were repeated at least twice with similar results. **P < 0.01 (t test)

Full size image

Exogenous NO application has been reported to reduce hypocotyl growth in both Arabidopsis and lettuce seedlings (Beligni and Lamattina 2000). We, therefore, determined the length of hypocotyl growth both in the dark and in the light of these Arabidopsis lines. There was no significant difference between the hypocotyl length of either atgsnor1-1 or atgsnor1-3R plants when grown in the dark for 7 days compared to wild-type plants (Fig. 2b). Conversely, hypocotyl elongation in the atgsnor1-3 line under these conditions was reduced by 74 % relative to wild type. Furthermore, atgsnor1-3 plants also exhibited a significant decrease in hypocotyl length when grown in the light for 7 days, showing a reduction of 41 % compared to the wild-type Col-0 line (Fig. 2c).

Silique size and seed production are altered in atgsnor1 mutants

The length of fully developed siliques on atgsnor1 mutant plants was significantly altered compared to wild type. Quantification of this trait showed that silique length in atgsnor1-3 and atgsnor1-1 plants was only 52 and 82 % of wild type, respectively (Fig. 3c). Both atgsnor1 mutants could function efficiently as either pollen donors or receivers. However, atgsnor1-3 plants have very short stamens which do not function as effective self-pollinators (Fig. 3a, b).

Fig. 3
figure 3

Impact of mutations in atgsnor1 on flower morphology and seed yield. a, b Image of Arabidopsis flowers to show stamen size in atgsnor1-1 mutants relative to wild-type plants. Scale bars are 1 mm. c, d mutations in atgsnor1 influence silique size and the number of seed developing in a green silique relative to wild-type plants. Scale bars are 5 mm (c) and 2 mm (d). e The number of seeds per silique produced by the listed plant genotypes. f mg fresh weight per ten siliques generated from atgsnor1 mutant and wild-type plants. Error bars represent 95 % confidence limits. These experiments were repeated twice with similar results. **P < 0.01 (t test)

Full size image

In the light of these results we also determined the seed yield of the Arabidopsis lines. The number of seeds per silique produced in atgsnor1 mutant was significantly altered relative to wild type. In this context, atgsnor1-3 and atgsnor1-1 plants produced only 30 and 87 %, respectively, of the seeds produced by wild-type plants (Fig. 3e). When pre-matured seeds within the green silique were quantified by weight, atgsnor1 plants showed a significant difference compared to wild type (Fig. 3f). atgsnor1-3 and atgsnor1-1 plants produced only 33 and 84 %, respectively, of the mg fresh weight of seed generated in wild-type plants. There was no significant difference between the weight of individual seeds in these genotypes (data not shown). To investigate whether atgsnor1 mutant yields developmentally normal seeds in siliques, we sliced the middle of the silique. The atgsnor1-1 mutant produced similar seed in the pre-matured green silique compared to wild type. However, atgsnor1-3 showed seed abortion in the small silique as compared to wild type (Fig. 3d). Taken together these results demonstrate that atgsnor1 mutations affect seed production, stamen length and silique size compared to wild type.

atgsnor1 mutations impact flowering time

NO has been reported to play an important role in flowering time (He et al. 2004). Therefore, we investigated the potential function of AtGSNOR1 in this important aspect of plant development. A well-established indicator for flowering time is the number of leaves produced on the primary shoot before the first flower is initiated. Plants that flower early have fewer leaves at this point in development. Flowering time under long-day conditions in atgsnor1-1 plants, which have reduced SNO levels, was slightly delayed compared to wild type (Fig. 4a). In contrast, flowering in atgsnor1-3 plants was accelerated with respect to wild type. Similar results were obtained when flowering time was determined under short-day conditions (Fig. 4b).

Fig. 4
figure 4

AtGSNOR1 regulates flowering time through the autonomous and photoperiod pathways. a, b Rosette leaf number in the given plant genotypes at the transition to flowering following growth in long (a) and short (b) days. c FLC expression of 3-week-old plant in the indicated plant genotypes. d Expression of LFY in 3-week-old plants of the given genotype. e LFY gene expression of 3-week-old plants treated with 100 μM of GA3 for 4 days. f Expression profile of CO transcript accumulation during a 24 h photoperiod in the shown plant lines. Error bars represent 95 % confidence limits. These experiments were repeated at least twice with similar results. *P < 0.05, **P < 0.01 (t test). LD long day, SD short day

Full size image

We next investigated which components of the flowering time regulatory system are affected by AtGSNOR1. The vernalisation and autonomous pathways converge on a floral repressor, FLOWERING LOCUS C (FLC) (Michaels and Amasino 1999) and reduced FLC expression promotes flowering (Michaels and Amasino 1999; Sheldon et al. 1999). We, therefore, analysed the expression of this marker gene in atgsnor1 mutants. FLC expression was significantly reduced in atgsnor1-3 plants compared to wild type, while in the atgsnor1-1 line FLC transcript accumulation was similar to that of wild type (Fig. 4c).

We also investigated the expression of the floral meristem identity gene, LEAFY (LFY). This gene has a key function in flower initiation and consequently LFY expression increases gradually before flowering (Blazquez et al. 1997). However, the amount of LFY transcripts was similar in all atgsnor1 mutants as compared to wild type (Fig. 4d). To identify whether atgsnor1 mutants flower through the GA-mediated pathway, we applied 100 μM of GA3 on the leaf surface of 3-week-old plants of the given phenotypes and investigated LFY gene expression through quantitative RT-PCR using untreated and treated samples with GA3. Analysis of quantitative RT-PCR indicated that there was no significant difference in these genotypes (Fig. 4e).

With respect to the GA pathway, exogenous GA application promoted flowering in a similar fashion in atgsnor1-3, atgsnor1-1 and wild-type plants (data not shown). Furthermore, the basal expression of the GA marker gene, AtGA20ox1, which encodes a gibberellin 20-oxidase (Hay et al. 2002), was similar in all these lines and neither of the atgsnor1 mutants was more sensitive to GA (data not shown). Collectively, these data suggest that AtGSNOR1 operates in parallel to this flowering time pathway.

We also assessed if AtGSNOR1 controlled the photoperiod pathway. CONSTANS (CO) is an important marker for this pathway, which functions as a link between the circadian clock and the control of flowering time (Suarez-Lopez et al. 2001; Yanovsky and Kay 2002). Increased CO expression leads to early flowering while decreased CO expression results in late flowering. As CO expression is controlled by a diurnal rhythm (Yanovsky and Kay 2002) a decrease in CO expression could reflect either a phase shift or a reduction in amplitude or an impact on both of these parameters. We, therefore, determined CO transcript accumulation over a 16-h light/8-h dark cycle in the atgsnor1 lines. The overall level of CO expression was similar in atgsnor1-1 and wild-type plants (Fig. 4f). Conversely, in the atgsnor1-3 line CO transcript accumulation was increased relative to wild type between the 12- and 24-h time points. The phase of CO expression, however, was not altered in atgsnor1-3 plants compared to wild type.

AtGSNOR1 regulates root development and cell size

The phenolic molecule salicylic acid (SA) is known to have functions in the control of cell growth (Vanacker et al. 2001). For example, SA is thought to regulate leaf cell enlargement, endoreduplication and/or cell division (Vanacker et al. 2001). As AtGSNOR1 regulates both SA biosynthesis and SA signalling (Feechan et al. 2005; Malik et al. 2011), we investigated if cell growth was altered in atgsnor1 mutant lines. Analysis of trypan blue stained leaf mesophyll cells revealed that this cell type was significantly larger in atgsnor1-3 plants compared to wild type (Fig. 5a). However, in atgsnor1-1 mutant plants mesophyll cells were very similar to that of wild type.

Fig. 5
figure 5

atgsnor1 mutants are altered in mesophyll cell size and root architecture. a Size of mesophyll cells in 4-week-old wild-type and atgsnor1 mutant lines revealed by trypan blue staining. Scale bar is 100 μM. b, c Root architecture (b) and primary root length (c) of 10-day-old light-grown seedlings on 1/2 MS medium containing 1 % sucrose in the indicated plant lines. d Root hair development in wild-type and atgsnor1 mutant lines. 10-day-old light-grown seedlings were stained with toluidine blue for this experiment. Scale bar is 100 μM. Error bars represent 95 % confidence limits. These experiments were repeated three times with similar results. **P < 0.01 (t test)

Full size image

We also investigated the possible impact of mutations in AtGSNOR1 on root development. In this context, atgsnor1-3 plants exhibited a lack of visible lateral root development. Also, both the atgsnor1-1 and atgsnor1-3 lines exhibited reduced primary root length compared to wild type (Fig. 5b). Thus, primary root length was reduced in atgsnor1-1 and atgsnor1-3 lines by 27.3 and 72.7 %, respectively (Fig. 5c). Interestingly, mutations in AtGSNOR1 also influenced root hair development. The atgsnor1-1 line exhibited strikingly elongated root hairs relative to wild-type plants (Fig. 5d). Conversely, in atgsnor1-3 plants root hairs were reduced in stature.

atgsnor1-3 conveys auxin sensitivity and delayed germination

As loss and gain-of-function mutations in AtGSNOR1 had a pleiotropic effect on plant development we investigated the response of the atgsnor1 lines towards a variety of plant hormones. Neither atgsnor1-1, atgsnor1-3 nor atgsnor1-3R plants showed any significant differences to ethylene, methyl-jasmonate, gibberellin or cytokinin, when germinated in the presence of a series of concentrations of these hormones (data not shown). In contrast, in the presence of 1 μM of the synthetic auxin 2,4-D in only 7 % of atgsnor1-3 seedlings had cotyledons emerged compared to 93 % for wild-type plants (Fig. 6a). The extent of cotyledon emergence in atgsnor1-1 and atgsnor1-3R plants was not significantly different from the wild-type line.

Fig. 6
figure 6

atgsnor1-3 conveys sensitivity to 2,4-D and delays germination. a Sensitivity of the indicated Arabidopsis genotypes to 2,4-D. b Impact of mutations in AtGSNOR1 on seed germination. Error bars represent 95 % confidence limits. These experiments were twice with similar results. *P < 0.05 (t test)

Full size image

Interestingly, the atgsnor1-3 mutation also resulted in delayed germination. Thus, when grown on MS plates the germination frequency at 24 h post-sowing for atgsnor1-1, atgsnor1-3R and wild-type plants was 63, 52 and 51 %, respectively (Fig. 6b). However, the germination frequency for the atgsnor1-3 line under these conditions was only 18 %.

Discussion

Shoot branching has a key function in the elaboration of adult plant body plans. Variation in the pattern of axillary shoot meristem initiation and activity largely underpins the diversity in plant shoot architecture which permits plants to adapt to their local environments (Sussex and Kerk 2001). Loss of AtGSNOR1 function had a profound effect on shoot architecture. In atgsnor1-3 plants shoots were dramatically shorter than wild type, as exemplified by the length of the primary inflorescence. Furthermore, the atgsnor1-3 line showed a striking increase in cauline and floral nodes compared to wild type. Also, atgsnor1-3 plants also exhibited a significant increase in higher order branching, reflected in an elevation of the ratio of total branches to the number of first-order branches. These phenotypes of atgsnor1-3 plants contrasted with other well-characterised branching mutants (Reintanz et al. 2001; Stirnberg et al. 2002; Beveridge et al. 2003). For example, the more axillary growth 1 (max1) and max2 mutants, that exhibited an increased number of shorter shoot branches, produced a similar number of vegetative nodes to wild type (Stirnberg et al. 2002). Moreover, the max1 and max2 mutants also showed a similar level of higher order branching relative to wild-type plants. Thus, while the major impact of max1 and max2 is on first-order branching, atgsnor1-3 predominately affects higher order branching. Determination of the phyllotactic sequence of shoot emergence at consecutive node positions revealed, however, that the atgsnor1-3 mutation did not interfere with the growth phase characteristic patterns of lateral development. Rather, atgsnor1-3 affected the extent of axillary growth. This observation paralleled that found for other mutants altered in shoot branching (Stirnberg et al. 2002). Apically synthesised auxin has long been known to inhibit lateral shoot growth (Thimann and Skoog 1933). Auxin is not thought to enter the bud, rather its inhibitory effect is thought to be mediated by a secondary messenger (Morris 1977). As might be anticipated, strong alleles of auxin resistant (axr) Arabidopsis mutants exhibit increased shoot branching (Lincoln et al. 1990). However, counter intuitively atgsnor1-3 plants, which are auxin sensitive, also exhibit this morphological phenotype. Thus, AtGSNOR1 might control shoot branching by regulating a step downstream of auxin in this particular signalling pathway. However, auxin-dependent signalling is not the only pathway that represses axillary bud outgrowth. Cytokinin or carotenoids transported from the root are both thought to suppress shoot growth (Cline 1994; Sorefan et al. 2003). Hence, AtGSNOR1 could also function in one or more of these two signalling pathways.

We also compared shoot growth in the mutants and wild-type lines in terms of total FW and FW distribution between the primary shoot (i.e. primary inflorescence and primary leaves) and the branches. While the atgsnor1-1 line exhibited an increase in total shoot FW relative to wild type, atgsnor1-3 plants showed a significant reduction. The absence of AtGSNOR1, therefore, significantly reduces shoot production as determined by FW, while overexpression of AtGSNOR1 promotes an increase in shoot FW. Taken together these data suggest that AtGSNOR1 contributes to the control of total shoot FW production in Arabidopsis. Furthermore, the FW distribution between primary shoot and the branches is also impacted by mutation of AtGSNOR1. In the atgsnor1-3 line total FW of lateral branches is strikingly increased relative to the primary inflorescence. Thus, AtGSNOR1 also regulates resource allocation between primary and lateral branching. This result contrasts with those of other shoot branching mutants such as max1 and max2 (Stirnberg et al. 2002) which promote outgrowth of a greater number of first order lateral branches, but have little effect on the overall resource allocation between the primary shoot axis and lateral shoots.

The size of atgsnor1-3 leaves were significantly greater than wild type, suggesting a role for SNO signalling in the control of leaf organ size. Furthermore, hypocotyl elongation in either the light or the dark was reduced in atgsnor1-3 plants. It has been reported previously that exogenous NO application can reduce hypocotyl elongation in both Arabidopsis and lettuce seedlings grown in the dark (Beligni and Lamattina 2000).

Indole acetic acid (IAA) treatment of cucumber induced a transient increase of endogenous NO in the basal region of the hypocotyl (Pagnussat et al. 2002), suggesting this might be an important region for growth-related NO synthesis. Thus, NO generated in this area may control hypocotyl elongation either in the light or the dark by regulating protein function via the S-nitrosylation of critical regulatory thiols.

Somewhat surprisingly, seed yield was decreased in both atgsnor1-3 and atgsnor1-1 plants to 30 and 87 % of wild-type levels, respectively. In the context of atgsnor1-3 plants this reflects the presence of very short stamens which do not function as effective pollinators and the seed abortion in the half size of silique as compared to wild type. Gibberellin has been reported to play an important role in controlling seed development and silique size (Singh et al. 2002). Our data suggests that S-nitrosylation may, therefore, regulate GA levels or GA signalling during seed and silique development.

In contrast, atgsnor1-1 plants express constitutive broad spectrum pathogen resistance (Feechan et al. 2005). A common trait associated with this phenotype is reduced yield, thought to reflect the increased resources being channelled into defence relative to plant growth and reproduction (Grant et al. 2003). These data may explain in part why atgsnor1-1 plants produce only 87 % of the seed of the wild-type line.

The major developmental transition in flowering plants is the switch from vegetative to reproductive development. The timing of this transition is important to both coincide with favourable conditions for seed set and to orchestrate synchronous flowering in out-crossing species (Simpson and Dean 2002). This transition is thought to be under the control of four major signalling pathways (Simpson and Dean 2002). The photoperiod and vernalisation pathways integrate external signals into the decision to flower (Gendall et al. 2001; Onouchi et al. 2000). In contrast, the autonomous and GA pathways function independently of environmental cues (Michaels et al. 1999; Reeves and Coupland 2001). Recently, NO has also been reported to control flowering time (He et al. 2004). In atgsnor1-1 plants, flowering time in both long- and short-day conditions, determined by the number of leaves on the primary shoot before the first flower was initiated, was slightly delayed relative to wild type. Conversely, in the atgsnor1-3 line flowering time in both long and short days was accelerated. Collectively, these data suggest that AtGSNOR1 can modulate flowering time irrespective of day length.

The vernalisation and autonomous flowering pathways converge on the floral repressor, FLC (Michaels and Amasino 1999; Sheldon et al. 1999) and autonomous pathway signalling components normally function to limit the accumulation of FLC mRNA (Michaels and Amasino 1999; Sheldon et al. 1999, 2000). While FLC expression appeared similar to wild type in atgsnor1-1 plants, the expression of this key regulatory gene was strongly repressed in the atgsnor1-3 line. Because many of the effects of NO are light dependent, we also investigated if AtGSNOR1 is involved in the photoperiod pathway. CO is an important marker for this pathway: increased CO expression leads to early flowering while decreased CO expression results in late flowering. While the level of CO expression in atgsnor1-1 plants was similar to wild type, the expression of this gene in atgsnor1-3 plants was increased. However, the phase of 123CO expression, which is controlled by a diurnal rhythm, was not altered in atgsnor1-3 plants.

The GA pathway is another genetically distinct mechanism which can control flowering time. Either GA or mutations which result in activation of GA signalling promote flowering (Jacobsen and Olszewski 1993). In contrast, mutations that block GA biosynthesis or signalling delay flowering (Wilson et al. 1992). GA promoted flowering in atgsnor1 mutants and wild-type plants in a similar fashion. Moreover, expression of the GA marker gene, AtGA20ox1, was similar in these Arabidopsis lines. Also, neither atgsnor1-1 nor atgsnor1-3 plants were more sensitive than wild-type plants to GA. LFY is a marker gene for flower initiation and integrates the long day (LD) and GA pathways (Piñeiro and Coupland 1998; Blazquez and Weigel 2000). GAs promote LFY expression to induce flowering in Arabidopsis (Blazquez et al. 1998; Eriksson et al. 2006; Mutasa-Göttgens and Hedden 2009). The expression of this floral integrator was similar in all atgsnor1 mutant lines relative to wild type. This data indicates that atgsnor1 mutants were not influenced by the GA pathway, therefore AtGSNOR1 appears to function in parallel to this pathway.

Collectively, these findings imply that increased SNO levels, resulting from the absence of AtGSNOR1 function, promote flowering in atgsnor1-3 plants through both the autonomous and photoperiod pathways. This data is unexpected because characterisation of the NO overproducer 1 (nox1) and Arabidopsis thaliana nitric oxide synthase 1 (atnos1) mutants, which have been reported to show increased and decreased NO accumulation, respectively (He et al. 2004; Guo et al. 2003), have suggested that NO negatively regulates flowering time (He et al. 2004). This discrepancy may reflect the differences in the concentration of NO bioactivity between these lines. For example, the level of FLC expression is dependent on the exogenous NO concentration: while 100 μM NO induced FLC expression, lower NO concentrations repressed the expression of this gene (He et al. 2004). Alternatively, manipulating total cellular NO or SNO levels may give different outcomes, in some cases. For example, PR1 gene expression is repressed in atgsnor1-3 plants which accumulate SNOs (Feechan et al. 2005), however, the application of exogenous NO donors has been reported to induce PR1 expression in tobacco (Durner et al. 1998).

Loss of AtGSNOR1 function has a striking impact on root architecture. In this context, absence of AtGSNOR1 activity strongly reduced primary root growth. The emerging evidence suggests NO is a key regulator of root architecture, where it probably acts as a component of the auxin signalling pathway (Pagnussat et al. 2002). Furthermore, in a similar fashion to increased endogenous SNO levels, exogenous NO application is known to negatively regulate primary root growth (Beligni and Lamattina 2001). However, exogenous NO application also promotes the development of lateral roots (Pagnussat et al. 2003), a phenotype not observed in atgsnor1-3 plants. The atgsnor1-1 line did, however, show increased root hair length, while the atgsnor1-3 line exhibited reduced root hair length compared to wild-type and atgsnor1-3R plants. The growth of root hairs in Arabidopsis is under redox control: reactive oxygen intermediates (ROIs) produced by an NADPH oxidase drive cell expansion through the activation of calcium channels (Foreman et al. 2003). Thus, it may not be surprising that changes in cellular SNO levels impact root hair development because cross-talk between ROI and NO signalling is well documented (Grant and Loake 2000; Delledonne et al. 2001; Yun et al. 2011). In this context, increasing levels of cellular SNOs have recently been shown to promote S-nitrosylation of the leaf NADPH oxidase, AtRBOHD, at Cys890, decreasing its activity (Yun et al. 2011). By extension enhanced SNO levels in the roots of atgsnor1-3 plants, which possess small root hairs, might drive S-nitrosylation of AtRBOHC, the NADPH oxidase implicated in root hair development (Foreman et al. 2003). Consequently, ROI synthesis might be blunted at key cellular locations in the roots of atgsnor1-3 plants, leading to inactivation of the calcium channels required for normal root hair development.

Collectively, our findings suggest that AtGSNOR1 function impinges upon a variety of programmes integral to plant growth and development. S-nitrosylation of cysteine thiols may therefore be a critical mechanism underpinning these processes. The pleiotropic effects resulting from the modification of cellular SNO levels, through AtGSNOR1 activity, may reflect a key role for SNOs in plant hormone signalling networks. In this context, atgsnor1-3 plants exhibit increased sensitivity to auxin. Furthermore, NO has been implicated in auxin signalling and the receptor for this phytohormone, TIR1, has recently been proposed to be S-nitrosylated (Terrile et al. 2012). Interestingly, deletion of GSNOR in mice impacted vascular function but did not result in any obvious defects in either development or reproduction (Liu et al. 2004). Hence, the regulation of protein function via S-nitrosylation, may have a significantly more expansive role within the biology of plants relative to animals.