Both plant and bacterial nitrate reductases contribute to nitric oxide 1 production in Medicago truncatula nitrogen-fixing nodules 2

Nitric oxide (NO) is a signalling and defence molecule of major importance in living 3 organisms. In the model legume Medicago truncatula , NO production has been detected in 4 the nitrogen fixation zone of the nodule, but the systems responsible for its synthesis are yet 5 unknown, and its role in symbiosis is far from being elucidated. In the present work, using 6 pharmacological and genetic approaches, we explored the enzymatic source of NO production 7 in M. truncatula – Sinorhizobium meliloti nodules, under normoxic and hypoxic conditions. 8 When transferred from normoxia to hypoxia, nodule NO production was rapidly increased, 9 indicating that NO production capacity is present in functioning nodules and may be promptly 10 up-regulated in response to decreased oxygen availability. Contrary to roots and leaves, 11 nodule NO production was stimulated by nitrate and nitrite, and inhibited by tungstate, a 12 nitrate reductase inhibitor. Nodules obtained with either plant nitrate reductase RNA 13 interference double knockdown ( MtNR1/2 ), or bacterial napA - and nirK -deficient mutants, or 14 both, exhibited reduced nitrate or nitrite reductase activities and NO production levels. 15 Moreover, NO production in nodules was found to be inhibited by electron transfer chain 16 inhibitors, and nodule energy state (ATP/ADP ratio) was significantly reduced when nodules 17 were incubated in the presence of tungstate. Our data indicate that both plant and bacterial 18 nitrate reductase and electron transfer chains are involved in NO synthesis. We propose the existence of a nitrate-NO respiration process in nodules which could play a role in the 20 maintenance of the energy status required for nitrogen fixation under oxygen-limiting 21 conditions. measured using either DAF-2, or CuFL probes, respectively; D, nodule NO content. NO production and content are expressed as relative fluorescence units. Nodules were incubated under 21% oxygen (control, Ctrl), in the presence of either 100 µM cPTIO, 50 mM sucrose (Suc), 300 µM KCN, or under 1% O2. Data are the means ± SD of 5 (A, B), 4 (C), and 3 (D) independent experiments assayed in duplicates. Significant difference * (P=0.05), or ** (P=0.01), when compared with the control (Ctrl) according to Student's t-test.

Nitric oxide (NO) is a gaseous intracellular and intercellular signalling molecule in 3 mammals with a broad spectrum of regulatory functions in various physiological processes 4 (Ignarro, 1999). There is increasing evidence of a role for this molecule in plant growth and 5 development (del Rio et al., 2004;Delledonne, 2005). NO appears to have signalling 6 functions in the induction of cell death, defence genes and interaction with reactive oxygen 7 species during plant defence against pathogen attack (Wendehenne et al., 2001;Delledonne, 8 2005). Similarly, there is evidence of a role for NO in plants exposed to abiotic stress such as 9 osmotic stress, salinity or high temperature (Gould et al., 2003). However, NO synthesis in 10 plants is still a matter of debate (Besson-Bard et al., 2008;Corpas et al., 2009;Moreau et al., 11 2010). Possible generating systems have been proposed: NO synthase-like proteins (Corpas et 12 al., 2009), nitrate reductase (NR) (Dean and Harper, 1988;Rockel et al., 2002), polyamine 13 oxidase (Yamasaki and Cohen, 2006), nitrite-NO reductase (NI-NOR) (Stohr and Stremlau, 14 2006), but convincing evidence of their involvement in the purposeful generation of NO in 15 vivo is still lacking (Zemojtel et al., 2006;Moreau et al., 2010). 16 Contrary to that in pathogenic situations, the interaction between legumes and soil 17 bacteria of the Rhizobiaceae family leads, after extensive recognition of both partners, to the 18 establishment of a symbiotic relationship characterized by the formation of new differentiated 19 organs called nodules, which provide a niche for bacterial nitrogen fixation. Functional 20 nodules result from the combination of developmental and infectious processes: bacteria 21 released in plant cells differentiate into bacteroids with the unique ability to fix atmospheric 22 nitrogen via nitrogenase activity (Oldroyd and Downie, 2008). As nitrogenase is strongly 23 inhibited by oxygen, nitrogen fixation is made possible by the microaerophilic conditions 24 prevailing in the nodule (Millar et al., 1995). 25 Several lines of evidence have demonstrated the occurrence of NO production during 26 the legume -rhizobium symbiosis. NO was transiently observed in Lotus japonicus and 27

Medicago sativa roots within the few hours after infection with Mesorhizobium loti and 28
Sinorhizobium meliloti strains respectively (Shimoda et al., 2005;Nagata et al., 2008). NO 29 has also been involved in the auxin-controlled formation of M. sativa and M. truncatula 30 nodules (Pii et al., 2007). NO formation has been also detected in functional M. truncatula -31 S. meliloti nodules: NO detection was restricted to the fixing zone of the nodule (Baudouin et 32 al., 2006). Finally, NO has been shown to be produced by mature Glycine max -33 Bradyrhizobium japonicum nodules in response to flooding conditions (Meakin et al., 2007;1 Sanchez et al., 2010). A wide modulation of NO-responsive genes has also been detected 2 during the establishment of a functioning nodule, pointing to a possible contribution of NO in 3 nodule metabolism (Ferrarini et al., 2008). Moreover, it has been shown that leghemoglobin 4 (Lb), the hemoprotein ensuring an oxygen flux to the bacteroids in the microaerophilic 5 conditions of the nodule, has the capacity to scavenge NO (Herold and Puppo, 2005), which 6 suggests that it participates in the protection of bacteroids against the inhibition of nitrogenase 7 by NO (Trinchant and Rigaud, 1982;Shimoda et al., 2008;Kato et al., 2009). In the same 8 way, NO has been shown to induce expression of non-symbiotic haemoglobin genes in L. 9 japonicus (Shimoda et al., 2005), and overexpression of class 1 plant haemoglobin gene 10 appeared to enhance symbiotic nitrogen fixation activity between L. japonicus and M. loti 11 (Shimoda et al., 2008). Similarly, M. truncatula or M. sativa nodules induced by a S. meliloti 12 mutant strain deficient in the flavohemoglobin Hmp, a well known NO-degrading enzyme, 13 showed a decreased nitrogen fixation efficiency (Meilhoc et al., 2010). The sources of NO in 14 symbiotic nodules are still unclear. In bacteria, the denitrification pathway is the main route 15 for NO production identified to date (Zumft, 1997), and it was assumed that it could be a 16 source of NO in symbiotic nodules. This was recently demonstrated by Delgado's team which 17 found that B. japonicum periplasmic nitrate and nitrite reductase are the main source of NO 18 production in soybean nodules in response to flooding (Sanchez et al., 2010). On the plant 19 partner side, a NO synthase-like activity has been measured in lupine nodules (Cueto et al., 20 1996), and it was suggested that such an enzyme could be responsible for NO production in 21 nodule infected cells (Baudouin et al., 2006). On the other hand, it has been known for a long 22 time that nodules, grown aseptically in the absence of a source of combined nitrogen, exhibit 23 high NR activity (Cheniae and Evans, 1960), and it was asked whether nodule NR activity 24 could be involved in functioning nodules (Kato et al., 2009). However, no evidence for the 25 contribution of either a NOS-like enzyme or the NR to NO production in the plant partner was 26 brought to date. 27 As underground organs, nodules may be submitted to flooding or drought episodes. 28 Hypoxia is a major determinant in the adverse effects of waterlogging on plants (Mommer et 29 al., 2004). Hypoxic stress has pronounced effects on mitochondrial function, both from the 30 perspective of oxygen limitation and from increased production of compounds that compete at 31 the oxygen binding site. Among these compounds, NO has been demonstrated to be produced 32 in hypoxic roots through a mechanism called "nitrate-NO respiration", which involves the 33 non-symbiotic haemoglobin, the nitrate reductase (NR) and electron transfer chain (ETC) 34 the NO produced and released from entire nodules with a less invasive method to keep the 23 nodule structures intact and maintain the micro-aerophilic conditions inside the nodule. Entire 24 nodules were taken from the roots, and incubated in a detection medium containing the DAF 25 probe. In these conditions, after a 45 to 60 min transient equilibration period, the production 26 of NO, when measured under either 21% or 1% O 2 , was found to be linear for at least 4 h 27 ( Fig. 1-A). Consequently, NO production was routinely measured between 2 and 4 h of 28

incubation. 29
To test the experimental protocol of NO measurement and the specificity of DAF for 30 NO, nodule NO production was analyzed in various conditions ( Fig.1-B). When measured in 31 the presence of 2-[4-carboxyphenyl]-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (cPTIO, a 32 NO scavenger), DAF fluorescence was 73 % inhibited compared to the control showing that 33 the major part of the DAF fluorescence was due to the production of NO itself, and not to 1 products interacting with DAF-2 (Planchet and Kaiser, 2006). The presence of 50 mM 2 sucrose (Suc) in the incubation medium did not modify NO production, indicating that 3 nodules were not sugar starved during the experiments. In the presence of KCN, a lethal and 4 potent inhibitor of many haem-containing enzymes, NO production was 83% inhibited (Fig.  5 1-B). The remaining NO production measured with KCN treatment (about 1 fluorescence unit 6 h -1 mg FW -1 ) was considered as the background in our experimental system, and was 7 subtracted in the following experiments. Under 1% oxygen (12 µM O 2 ), NO production was 8 1.7 fold increased compared to the control ( Fig. 1 A and B), indicating that DAF fluorescence 9 may be used for NO measurement in hypoxic conditions. Moreover, measurement of NO 10 production by samples containing 5 to 120 mg of fresh nodules showed that neither O 2 , nor 11 DAF-2, were limiting in our experimental conditions (data not shown). In addition, to test 12 further our experimental system, NO production was also analyzed with the CuFL fluorescent 13 probe, which is known to react rapidly and specifically with NO itself (Lim et al., 2006). The 14 absence of CuFL toxicity for nodules was first verified (see Material and methods), and NO 15 production was then measured in the conditions tested above. The results were similar to that 16 obtained with the DAF probe ( Fig. 1-C). Finally, NO production was compared to the amount 17 of NO measured in the extracts of nodules crushed immediately after a 4-h incubation period 18 with the above tested effectors. NO measured in tissue extracts ( Fig. 1-D) were found to 19 correlate NO production ( Fig. 1-B and C), which confirmed whole organ assays. Taken 20 together, these data show that DAF fluorescence may be considered as a good indicator of NO 21 produced and released from nodules, and that the protocol used in this study is efficient to 22 assess NO production by symbiotic nodules. A similar protocol was recently used to estimate 23 nodular NO accumulation in soybean nodules (Sanchez et al., 2010). All the experiments of 24 NO production reported in this work were first carried out with DAF-2, and then repeated 25 with CuFL with similar results. For simplicity, only the data obtained with DAF-2 are 26 presented below. 27 The production of NO in normoxic and hypoxic conditions was tested also in root 28 segments (without nodule) and leaf disks. In normoxia, NO production of roots and leaves 29 was respectively 35% and 70% lower than that found with nodules, and this production was 30 not increased in hypoxic conditions ( Fig. 2A). It may be noted that, whatever the organ, 31 fluorescence was 90-95% reduced when measured at 1% O 2 in the presence of 100 µM 32 cPTIO, which indicates that most of the fluorescence was related to NO production. These 33 data show that, contrary to roots and leaves, nodules are able to overproduce NO within hours 34 following transition from normoxia to hypoxia. To assess the sensitivity of the nodules to 1 changing pO 2 conditions, nodule NO production was measured during rapid normoxic (21% 2 O 2 ) to hypoxic (1% O 2 ) transition, and vice versa. As reported in figure 2B, NO production 3 rate exhibited a two-fold increase within 2 to 4 min after 21% to 1% O 2 transition, and 4 decreased within 5 min upon return to 21% O 2 conditions. As previously observed (Fig. 1), 5 the addition of 300 µM KCN to the incubation medium almost totally abolished the 6 production of NO (Fig. 2B). These results underline the flexibility and the reversibility of 7 nodule NO production regarding the oxygen environment, and indicate that nodules are able 8 to quickly respond to changes in partial oxygen pressure (pO 2 ). 9

II
Nitrate reductase activity is involved in nodule NO production 11 DAF fluorescence was analyzed with nodules incubated in the presence of NR 12 effectors, under either 21% or 1% O 2 . As shown in Fig. 3, in the presence of 10 mM nitrate 13 (NO 3 -), the substrate of NR, NO production was 2.2-fold increased, both in normoxia and 14 hypoxia, suggesting that NR is possibly involved in NO production. To further test this 15 hypothesis, NR activity was inhibited with the use of tungstate (Tg), an inhibitor of NR 16 (Harper and Nicholas, 1978). In these conditions, Tg significantly reduced NO production, 17 both in the control and in the presence of nitrate (Fig. 3). This means that NR is involved in 18 the production of NO either directly, or indirectly via the production of nitrite, the product of 19 NR. Moreover, on the basis of Tg-inhibition results, it may be concluded that the increase in 20 NO production observed under hypoxia was due to NR activity, since NO production was 21 inhibited to similar values (1.5 Fluo. unit h -1 mgFW -1 ) both in normoxia and hypoxia (Fig. 3). 22 This indicates that NR activity contributes more importantly to NO production under hypoxia 23 than under normoxia. In the presence of 1 mM nitrite (NO 2 -), nodule NO production increased 24 3.6-and 4.0-fold under normoxic and hypoxic conditions, respectively (Fig 3). However, it 25 was not inhibited by Tg in the presence of NO 2 -, which indicates that NR is involved in NO 26 production through the reduction of nitrate in nitrite, but does not produce NO directly. In 27 addition to NR, xanthine oxidase (a MoCo-enzyme like NR) has been reported to reduce NO 2 -28 into NO (Millar et al., 1998;Li et al., 2001). As xanthine oxidase is also inhibited by Tg, the 29 question arose as to whether the production of NO, and its inhibition in the presence of Tg, 30 could be due to xanthine oxidase activity rather than that of NR. To answer this question, we 31 analyzed the effect of allopurinol, an inhibitor of xanthine oxidase (Atkins et al., 1988), on 32 NO production. As reported in Fig. 3, allopurinol did not modify NO production under either 33 21% or 1% O 2 , which excludes the contribution of xanthine oxidase in the synthesis of NO. 34 The involvement of NR activity in the generation of NO has been already investigated 1 in various plant organs and tissues, and it was concluded that it contributes -directly or 2 indirectly-to NO production in roots and leaves (Dean and Harper, 1988;Rockel et al., 2002;3 Gupta et al., 2005;Planchet et al., 2005). Thus, to assess the possible contribution of NR in 4 NO production in other M. truncatula organs than nodules, root segments and leaf disks were 5 also analyzed for NO production in the presence or absence of NR effectors. It resulted that, 6 under either 21% or 1% O 2 , the production of NO was not affected by the addition of NO 3 -, 7 NO 2 or Tg in the incubation medium (Fig. S1). This means that, contrary to what happens in 8 nodules, NR is not involved in the production of NO in the roots and leaves of M. truncatula 9

plants. 10
To further investigate the differences between nodules, roots and leaves with regards 11 to NR-dependent NO production, we analyzed NR and nitrite reductase (NiR) activities in 12 these organs. When expressed as a function of fresh weight, NR activity was found to be 3 13 and 6-fold higher in nodules than in roots and leaves respectively (Table 1). These activities 14 are of the same order of magnitude than that measured in the nodules of other legumes such 15 as yellow lupine (Polcyn and Lucinski, 2001), faba bean and pea (Chalifour and Nelson, 16 1987), or soybean (Arrese-Igor et al., 1998). In the three organs, NiR activity was higher than 17 that of NR (Table 1). It has long been known that NO 2 is cytotoxic for plants, although the 18 molecular mechanism is still obscure, and a higher NiR versus NR activity, which avoids 19 NO 2 accumulation in the tissues, was classically observed in plants (Lucinski et al., 2002). 20 Interestingly, the NiR to NR activity ratio was found to be about 2 in nodules, and 8 in roots 21 and leaves (Table 1). This indicates that the nitrite-production versus nitrite-utilization 22 capacity is significantly higher in nodules than in roots and leaves, and underlines a possible 23 specific function of NR in the nodules. 24

III Both plant and bacteroid NRs contribute to NO production in nodules 26
In symbiotic nodules, NR activity has been generally found, with some exceptions, in 27 both the plant and the bacteroid partners (Lucinski et al., 2002). In the present work, to 28 determine if the NR-dependent production of NO observed in the nodules was due to either 29 one or both of the partners, we used a mutant approach. 30 Two NR genes have been identified in M. truncatula, NR1 (TC137636; 31 Mtr.10604.1.S1_at) and NR2 (TC130773; Mtr.42446.1.S1_at), which are both expressed at 32 detectable level in N 2 -fixing nodules (data not shown). To date, the main function of NR 33 identified in plants is its key role in the NO 3 to NH 4 + reduction pathway, which controls 34 nitrogen metabolism (Campbell, 1999). Thus, to assess the involvement of NR in the 1 production of NO by the nodules, without affecting the nitrogen metabolism in the whole 2 plant, we used a nodule-targeted RNA interference strategy. A RNAi M. truncatula MtNR1/2 3 double knockdown was constructed under the control of the zone III-specific promoter 4 MtNCR001 (Mergaert et al., 2003) (Fig. S2-A). In such a way, NR expression level was only 5 affected in the N 2 -fixing zone (zone III) of the nodule, avoiding any other effect that could 6 affect plant and nodule growth at early stages of development. Four weeks after inoculation, 7 MtNR1/2 RNAi transgenic roots did not show significantly modified phenotypes compared to 8 GUS RNAi control for plant growth and nodulation events (data not shown), but nodule size 9 was 30-40% reduced in the MtNR1/2 RNAi (Fig. S2-B and C). Measurement of the nitrate 10 reductase activity in this knockdown mutant line showed a 40 % decrease compared to the 11 GUS RNAi control nodules (Table 2). In the MtNR1/2 RNAi nodules, the production of NO 12 was found to be 46% decreased compared to that of control nodules, when measured under 13 either 1% O 2 (Fig. 4, Table 2), or 21% O 2 (Fig. S3). In addition, for both MtNR1/2 RNAi and 14 GUS RNAi control nodules, the production of NO was found to be increased by NO 2 and 15 inhibited by Tg, under 1% O 2 (Fig. 4) or 21% O 2 (Fig. S3). These results clearly indicate that 16 the decrease in NO production in knockdown nodules was related to the decrease in the plant 17 NR activity, and that the remaining NO production was dependent on bacteroid and/or 18 residual plant NR activities. 19 In bacteria such as S. meliloti, the denitrification process is known to generate NO as 20 an intermediate of NO 3 reduction to N 2 . NO 3 is firstly reduced to NO 2 by NR, and NO 2 is 21 then reduced to NO by NiR. To investigate the involvement of the bacteroid denitrification 22 pathway in the generation of NO, we analyzed NO production in nodules formed upon root 23 infection with S. meliloti napA and nirK mutant strains, impaired in NR and NiR activity 24 respectively. As reported in Table 3, NR and NiR activities were found to be respectively 25 37% and 38% reduced in napA and nirK nodules compared to wild type ones. As a control, no 26 NR or NiR activity was found in the bacteroid fractions extracted from napA and nirK mutant 27 nodules respectively (Table 3), which confirms the absence of NR or NiR activity in the 28 mutant strains. In both napA and nirK mutant nodules, the production of NO was decreased 29 by about 35% compared to that of wild type control, when measured under either 1% O 2 ( Fig.  30 5) or 21% O 2 (Fig. S3). Moreover, as observed in wild type nodules, NO production was 31 stimulated by NO 2 -, and inhibited by Tg, when measured under either 1% O 2 (Fig. 5), or 21% 32 O 2 (Fig. S3). These results indicate that the decrease in NO production in napA and nirK 33 mutant nodules was related to the absence of bacteroid NR and NiR activities respectively, 34 and that the remaining NO production was dependent on the plant partner NR and other 1 potential plant or bacteroid NO-producing activities. 2 MtNR1/2 and GUS RNAi transgenic roots were inoculated with S. meliloti wild type 3 and napA mutant strains to evidence a possible additive effect of plant and bacterial NR 4 mutations on NO production. In agreement with above presented data ( Fig. 4 and 5), NR 5 activity and NO production were decreased in both MtNR1/2 RNAi and napA mutant nodules 6 ( Table 2). The effects of the plant NR silencing and bacteroid NR mutations were found to be 7 partially additive in the MtNR1/2 RNAi/napA nodules, where NR activity and NO production 8 were decreased to 47% and 29% of their respective control (Table 2). Despite the absence of 9 fully additive effects at NR activity and NO production levels, which may probably be 10 explained by the up-regulation of complementary systems, these data confirm that NO 11 production in nodules is essentially related to the activity of NR. 12 Taken together, our data showed that, in M. truncatula -S. meliloti nodules, 1) both 13 the plant and the bacteroid partners produce NO through NR-dependent processes, 2) NO is 14 mainly produced by the plant partner, and 3) around one third of the NO generated by the 15 nodule is produced by the bacteroid denitrification pathway. The maintenance of NO production, in the presence of both NO 2 and Tg (Fig. 3), 20 indicated that NR does not produce NO directly, but more probably produces NO 2 which in 21 turn is reduced to NO. Beside NR, root mitochondria have been reported to be able to reduce 22 NO 2 to NO at the expense of NADH under anoxic conditions, but not in air (Gupta et al., 23 2005;Planchet et al., 2005;Gupta and Kaiser, 2010). Here, we investigated the involvement 24 of the mitochondria in NO production through the use of various mitochondrial and 25 denitrification pathway inhibitors. As reported in Fig. 6-A, under either 21% or 1% O 2 , NO 26 production was 40% inhibited by rotenone, an inhibitor of the mitochondrial complex I and of 27 the bacteroid NADH-quinol oxidoreductase. In the presence of Antimycin A and 28 myxothiazol, two inhibitors of the complex III, NO production was 50-55% and 80% 29 inhibited in normoxic and hypoxic conditions, respectively ( Fig. 6-A). The NO production 30 insensitive to the inhibitors (approximately 2 Fluo. units h -1 mgFW -1 in both conditions) 31 accounted for the residual part of NO which production does not depend on electron transport 32 chain (ETC) functioning. This means that in normoxia, the production of NO largely depends 33 on mitochondrial and bacteroid ETC functioning, and that the increase in NO production 34 observed in hypoxic versus normoxic conditions was essentially contributed by mitochondrial 1 and bacteroid ETCs. Furthermore, NO production was found to be insensitive to the 2 uncoupler carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP) (Fig. 6-A), 3 indicating that it does not depend on the presence of the trans-membrane electrochemical 4 proton gradient. Similarly, NO production was found to be insensitive to propylgalate, an 5 inhibitor of the mitochondrial alternative oxidase (AOX), which indicates that AOX does 6 probably not contribute to NO production (data not shown). When nodules were incubated in 7 the presence of NO 2 in the incubation medium, NO production was increased in the control 8 condition, as already shown in Fig. 3, and was inhibited by the above tested inhibitors to the 9 same extent as in the absence of NO 2 - (Fig. S4). Moreover, as these inhibitors are not specific 10 for either mitochondria, or bacteroid ETC, the production of NO was also analyzed in nodules 11 issued from M. truncatula inoculated with S. meliloti nirK mutants, where bacteroid ETC 12 presumably does not produce NO. As shown in Fig. 6-B, the effects of all the inhibitors tested 13 on NO production were similar to that observed with WT nodules, under 21% as well as 1% 14 Taken together, these data show that nodule NO production is: 1) strongly (80%) or 16 partially (60%) inhibited by ETC inhibitors in hypoxia and normoxia, respectively, 2) 17 independent of the trans-membrane electrochemical proton gradient, 3) stimulated by NO 2 -18 supply, and 4) similarly inhibited by ETC inhibitors both in WT and nirK mutant nodules. 19 This means, first, that mitochondrial and bacteroid ETC are directly involved in the 20 production of NO in functioning nodules and, second, that NO is essentially produced through 21 both mitochondrial and bacteroid ETC in hypoxic conditions. 22 23 V Nitrate reductase activity is necessary to maintain nodule energy status 24 In the roots of plants submitted to hypoxia, a nitrate-NO respiration -involving the 25 sequential reduction of NO 3 into NO 2 and then NO, via NR and mitochondrial ETC-was 26 proposed to contribute to energy supply under microaerobic conditions (Igamberdiev and Hill, 27 2009). Above data demonstrating the involvement of NR in the production of NO raised the 28 hypothesis of a role of NR in energy functioning of symbiotic nodules. To test this 29 hypothesis, we analyzed the energy state (i.e. the ATP/ADP ratio) of nodules incubated in the 30 presence of NR effectors (Figure 7). Under 21% O 2 , ATP/ADP ratio was high (close to 6-7) 31 in the control nodules, indicating that ATP-regenerating processes were not limited. In the 32 presence of either NO 3 -, NO 2 -, or Tg plus NO 2 -, the ATP/ADP ratio was not significantly 33 modified, but it was 50% decreased when nodules were incubated with Tg only, or Tg plus 34 NO 3 - (Figure 7-A). This means that the inhibition of NR partially affects the energy state of 1 the nodule even in normoxic conditions. Under 1% O 2 , the ATP/ADP ratio of control nodules 2 was close to 4 (Figure 7-A), which indicates that the decrease in pO 2 from 21 to 1% 3 significantly affects the energy status of the nodules, but maintains it compatible with nodule 4 functioning. In these conditions, the presence of either NO 3 -, NO 2 -, or Tg plus NO 2 did not 5 modify significantly the energy status of the nodules, but Tg only, or Tg plus NO 3 triggered a 6 dramatic fall (95%) of the ATP/ADP ratio. These data clearly mean that, under 1% O 2 , ATP-7 regenerating processes almost entirely depend on the functioning of NR activity. into the same organ (Fig. 8), which complicates the analysis of the NO source. In the present 20 work, both pharmacological and genetic approaches were used to analyze the potential role of 21 NR in the production of NO. First, the increase in NO production upon nodule feeding with 22 either nitrate or nitrite, and its inhibition by Tg (Fig. 3), and second, the lower level of NO 23 production in MtNR1/2 RNAi (Fig. 4), and napA and nirK (Fig. 5) mutant nodules, provide 24 strong evidence for a NO 2 --dependent NO synthesis via the activity of NR. However, the 25 relief of the Tg-related inhibition of NO production by NO 2 - (Fig. 3) clearly indicates that NR 26 does not produce NO by itself, but NO 2 which is then reduced into NO. The inhibition of NO 27 production by ETC inhibitors such as rotenone, antimycin A, and myxothiazol (Fig. 6), but 28 neither by FCCP, nor by propylgalate, indicate that mitochondrial and bacteroid ETCs are 29 directly involved in the reduction of NO 2 into NO, probably at the cytochrome oxidase site. 30 Thus, in M. truncatula nodules, NO 3 may be reduced into NO in a two step mechanism 31 involving successively NR and ETC activities (Fig. 8). 32 The use of either plant MtNR1/2 RNAi, or bacteria napA and nirK mutants, showed 1 that both the plant and the bacteroid partners are involved in the production of NO in the 2 nodule. Indeed, both plant and bacterial mutants exhibited decreased NO production (Figs. 4 3 and 5), and these effects were found to be additive in the MtNR1/2 RNAi/napA nodules 4 ( Table 2). The production of NO by the bacteroid partner was expected. Indeed, 5 denitrification activity has been shown to occur in S. meliloti bacteroids (O'Hara et al., 1983), 6 and NO is a well known intermediate product of the denitrification pathway (Zumft, 1997). 7 Moreover, it was recently described that bacteroid NR and NiR, products of the nap and nir 8 genes, contribute to the major part of the NO formed in soybean nodules, particularly under 9 hypoxic conditions (Meakin et al., 2007;Sanchez et al., 2010). However, evidence for the 10 involvement of the plant partner in NO production by nodules was still lacking. The 11 sensitivity of nirK mutant nodules to ETC inhibitors (Fig. 6) indicates that the mitochondrial 12 ETC is significantly involved in NO production. This observation is consistent with the fact 13 that root mitochondria of several species have been shown to be able to reduce NO 2 into NO 14 under anoxic or strongly hypoxic conditions (Gupta et al., 2005;Stoimenova et al., 2007;15 Gupta and Kaiser, 2010). 16 Taken together, our data show that in M. truncatula nodules, NO 3 reduction into NO 2 -

17
, and NO 2 reduction into NO, via the mitochondrial and bacteroid NR and ETCs pathway 18 (Fig. 8), constitute the main route for NO synthesis under hypoxic conditions, and contribute 19 to this synthesis in normoxic ones ( Fig. 3 and 6 shown to produce NO, and this production was increased in the G. max nodules when the 34 roots were submitted to a one-week hypoxia treatment in the presence of nitrate (Meakin et 1 al., 2007;Sanchez et al., 2010). Because of nitrogenase sensitivity to oxygen, legume nodules 2 are naturally hypoxic organs, with pO 2 in the range of nanomolar concentrations in the 3 infected region (Layzell and Hunt, 1990;Sung et al., 1991). Thus, the question was raised 4 whether nodule NO production is related to hypoxic conditions prevailing in nodules. In the 5 present work, using two different NO probes (DAF and CuFL), we showed that M. truncatula 6 nodules produced NO at a higher level than leaves or roots, and that this production may be 7 stimulated upon transition from normoxic to hypoxic conditions, contrary to what was 8 observed in leaves and roots ( Fig. 2A and S1). Considering the rapidity of the nodule 9 response to hypoxia (within few min, Fig. 2B), such an increase can hardly be explained by 10 an up-regulation of gene expression, but clearly indicates that NO production capacity is 11 already present in functioning nodules and may be promptly up-regulated to face a sudden 12 decrease in oxygen availability. 13 The data presented in this study raise the question of the role of such an NO 14 production process in microoxic symbiotic nodules. The presence of a gaseous diffusion 15 barrier in the inner cortex of the nodule and the respiration of bacteroids maintain naturally a 16 low oxygen pressure (5-60 nM O 2 ) within the infected cells of the nodules (Layzell and Hunt, 17 1990;Millar et al., 1995), and the pO 2 value can even fall to the nanomolar level in the 18 infected zone of nodules when the plants experience hypoxic environmental conditions. 19 Under conditions that limit oxygen availability, O 2 -dependent respiration of root 20 mitochondria declines below oxygen level required to saturate AOX and cytochrome c 21 oxidase (COX). The AOX Km value for oxygen is in the micromolar range (Millar et al., 22 1994;Affourtit et al., 2001), precluding AOX function under low oxygen pressure, whereas 23 that of COX is in the range of 80 to 160 nM (Hoshi et al., 1993;Millar et al., 1994), which 24 makes respiration possible under moderate hypoxic conditions. In symbiotic nodules, Lb 25 provides oxygen to bacteroids and host cell mitochondria which contain specific COX 26 pathway with high apparent affinity for oxygen (Km 50 nM, (Millar et al., 1995). However, 27 considering the oxygen dissociation constant value of Hb (2nM, (Duff et al., 1997), and the 28 very low oxygen concentration (nanomolar range) prevailing in infected cells, the question 29 arises whether oxygenic respiration can still fulfil ATP requirements for metabolic and 30 biosynthetic purposes in nodules submitted to hypoxia. It may be suggested that, under the 31 microaerobic conditions prevailing in nodules, the nitrate-NO respiratory pathway 32 (Igamberdiev and Hill, 2009;Igamberdiev et al., 2010), and references therein) may 33 contribute to energy supply in symbiotic N 2 -fixing nodules. Several lines of evidence argue in 34 favour of this hypothesis. In the present work we show that NRs and ETCs contribute to NO 1 production, via NO 3 and NO 2 reduction, particularly under hypoxic conditions. Similarly, in 2 soybean nodules, bacteroidal NR and NiR have been involved in NO production in response 3 to flooding conditions (Sanchez et al., 2010). Moreover, it has been shown that oxyLb, like 4 plant and animal class-1 haemoglobin, has the capacity to efficiently react with NO to 5 produce NO 3 with an elevated rate constant (Herold and Puppo, 2005). The NO generated at 6 either mitochondrial, or bacteroidal ETC level may therefore be oxidized by Lb into NO 3 -. It 7 should be mentioned that -considering the complexity of NO chemistry (Stamler et al., 1992)-8 the chemical forms and the mechanisms of NO diffusion or transport between the different 9 compartments (matrix, cytosol, periplasm, …) are still unknown. In the plant partner 10 particularly the nature and the importance of the NO flux between its production (COX in the 11 mitochondria) and oxidation (Lb in the cytosol) sites remained to be formally established and 12 estimated. However, different experiments carried out with either yeast, mammal, or plant 13 mitochondria (Castello et al., 2006;Stoimenova et al., 2007;Gupta and Kaiser, 2010) showed 14 that the NO produced by COX, in hypoxia or anoxia, may be detected by conventional 15 methods and partly quantified. The exchange of NO between mitochondrial matrix and 16 cytosol, or between the plant and bacteroid partners, may thus be reasonably hypothesized. 17 Thus, as summarized in Fig. 8, in parallel to the bacteroidal denitrification process, a plant 18 nitrate-NO respiration could be of importance in the micro-oxic nodules, particularly under 19 hypoxic conditions such as flooding, to maintain cell energy status and N 2 -fixing metabolism 20 when oxygen supply becomes limiting. The occurrence of such a mechanism is strongly 21 supported by the data on ATP and ADP measurements (Fig. 7), which show that the energy 22 status of the nodules depends either significantly, or almost entirely, on NR functioning under 23 normoxic, or hypoxic conditions, respectively. 24 The possible occurrence of the nitrate-NO respiration highlights potential new 25 functions for Lb and NR in N 2 -fixing nodules. Thus, in addition to its role in nitrogenase and denitrification pathway with nitrate. Similarly, it is well established that many symbiotic 29 associations between legumes and rhizobia are characterized by high NR activity (Cheniae 30 and Evans, 1960;Lucinski et al., 2002), and it was asked whether and how nodule NR activity 31 could be involved in functioning nodules (Lucinski et al., 2002;Kato et al., 2009). 32 Considering that the main route for nitrogen reduction in nodules is the bacteroid nitrogenase, 33 and not the NR-NiR pathway (Vance, 1990), an important function of the plant NR in the 34 nodule could be the reduction of NO 3 into NO 2 in the cytosol, to supply mitochondria and 1 COX with NO 2 -. The aim of the future prospects will be to demonstrate the functioning of 2 nitrate-NO respiration in N 2 -fixing nodules and the role of Lb and NR in this process, and to 3 consider the interplay between oxygen-dependent and nitrate-NO respirations for energy 4 regeneration processes in symbiotic nodules submitted to varying pO 2 conditions. Nodule, root and leaf samples were collected 4-5 weeks after inoculation, and either 21 immediately processed for NO quantification, or frozen into liquid nitrogen and stored at -22 80°C for further analysis. Bacteroids were prepared as previously described in (Trinchant et 23 al., 2004). 24 25

Construction of a binary vector for hairy roots transformation 26
For the RNAi construct, the CaMV 35S promoter (P35S) in pK7GWIWG2D(II),0 27 vector (VIB, Ghent, Belgium) was replaced by MtNCR001 promoter (Mergaert et al., 2003). 28 Following the nomenclature described for these binary vectors 29 (http://www.psb.ugent.be/gateway/index.php), we named our construction: 30 pK7GWIWG5D(II), where 5 is assigned for the promoter PMtNCR001. SacI and SpeI 31 restriction sites were added to PMtNCR001 by a PCR amplification with PMtNCR001SacI-F 32 and PMtNCR001SpeI-R primers, using as a template pENTL4L1-PMtNCR001. The resulting 33 1 promoter was done in three sequential subcloning steps, first a 2472 bp SacI-P35S:ccdB: 2 intron-MluI from pK7GWIWG2D(II),0 vector was subcloned in a modified Δ EagI pGEM-T ® 3 vector without SpeI site. Second, P35S was replaced by PMtNCR001 into SacI-SpeI sites. 4 Finally, pK7GWIWG5D(II) was obtained by insertion of the SacI-PMtNCR001:ccdB:intron-5 MluI cassette back into the original pK7GWIWG2D(II),0 vector. The primers used were: 6 PMtNCR001SacI-F 5'-GAGAGCTCGTTGTCCTTATTAGAGCGCCTA 7 PMtNCR001SpeI-R5'-GACTAGTTCTAGACCTTTGAACGTACTAAAGAGATT 8 Using M. truncatula cDNA as template, 432-bp and 441-bp fragments of MtNR1 9 (TC137636; Mtr.10604.1.S1_at) and MtNR2 (TC130773; Mtr.42446.1.S1_at) genes, 10 respectively were obtained via polymerase chain reaction (PCR) with specific primers: (1 kPa) for hypoxia treatment was chosen on the basis that pO 2 in most waterlogged soils 9 ranged from 5 kPa to zero (Gibbs and Greenway, 2003). The NO produced by the tissues and 10 released into the detection medium was measured using the fluorescence of the DAF probe. 11 At various times, aliquots of the incubation medium were sampled and the fluorescence of 12 DAF-2T, the reaction product formed from DAF-2 and NO, was measured using a microplate 13 reader spectrofluorimeter (Cary Eclipse, Varian, Les Ulis, France), ex 495 nm/ em 515 nm. In 14 these conditions NO production and release was found to be linear between 1 and at least 4 h 15 of incubation. Assay blanks contained detection buffer and DAF, without nodules. 16 Alternately, NO production was measured in the same experimental system through the use of 17 CuFL fluorescent probe (Strem Chemicals, Bischheim, France) instead of DAF-2 in the 18 detection buffer. As CuFl is known to be a cell-permeant probe (Lim et al., 2006), its capacity 19 to penetrate into nodule cells, and its cytotoxicity were analyzed. After a 2h-incubation period 20 of entire nodules in the presence of 5 µM CuFl, nodules were excised into 100 µm thick slices 21 with a vibratom 1000 Plus (Labonord, Templemars, France), and analyzed with a Zeiss LSM 22 500 confocal laser microscope (Carl Zeiss SA, Le Pecq, France) as described in (Baudouin et 23 al., 2006). No fluorescence could be detected in nodule cells (Fig. S6-A), indicating that 24 CuFL probe, or its N-nitrosamine FL-NO derivative, did not penetrate into the nodule, and 25 could be used to measure the NO in the incubation medium. To test CuFL toxicity, the effects 26 of increasing concentrations (0, 2, 5, 10 and 20 µM) of CuFL were analyzed after 2h of 27 incubation on the nodule energy state (ATP/ADP ratio being used as a marker of cell 28 viability). Adenine nucleotides were extracted and analyzed as described below. No effect 29 was observed on ATP/ADP ratio (Fig. S6-B), which means that, in these conditions, CuFL 30 was not toxic for nodule cells. Thus, when assayed with CuFL, NO production was routinely 31 measured for 2 h with a probe concentration of 5 µM. 32 For rapid pO 2 transition (between 21% and 1% O 2 ) experiments, four to six nodules 33 were set in fluorescence cuvette containing 1 ml of detection medium, and NO production 34 was continuously measured on a kinetic mode using a Xenius spectrofluorimeter (SAFAS, 1 Monaco, Monte Carlo). pO 2 in the incubation medium was imposed by a permanent bubbling 2 of either ambient air or 1:99 % O 2 :N 2 (v/v) gaz stream. Incubation medium was continuously 3 homogenized using a non invasive stirring equipment during the assay. 4 5

Measurement of NO content 6
Ten to twenty mg of nodules, either freshly detached, or incubated for 4 h in the 7 presence or absence of effectors, were crushed with mortar and pestle in 200-300 µl of 8 detection medium in the presence of 10 µM DAF probe. The extract was centrifuged at 4°C 9 for 10 min, and the fluorescence of the supernatant was immediately measured as described 10 above. 11 12

Effects of effectors on NO production 13
The effectors tested on NO production were routinely used at the following 14 concentrations: 10 mM NaNO 3 , 1mM NaNO 2 , 1 mM NaTg, 1 mM allopurinol, 50 mM 15 sucrose, 300 µM KCN, 100 µM cPTIO, 10 µM rotenone, 25 µM antimycin A, 25 µM 16 myxothiazol, 1 mM propylgalate, and 10 µM FCCP. The effectors were added to the 17 detection buffer at the same time as nodules, and their effects on NO production were 18 measured after 2 to 4 h of incubation as described above. 19 20

Enzymatic activity measurements 21
Tissue samples were crushed at 4°C using an extraction buffer containing 25 mM Tris-22 HCl pH 8.5, 1 mM EDTA, 20 µM FAD, 1 mM DTT, 20 µM L-transepoxysuccinyl-23 leucylamido-[4-guanidino]butane (E64), and 2 mM phenylmethylsulfonyl fluoride (PMSF). 24 The extracts were centrifuged at 15000 g for 15 min, and used for nitrate and nitrite reductase 25 activities. 26 NR activity was assayed at 28°C by measuring NO 2 production. The reaction medium 27 (1 ml) contained enzymatic extract, 0.2 M Hepes pH 7.0, 15 mM KNO 3 , 250 µM NADH. dihydrochloride in water. After incubation for 30 min at ambient temperature, samples were 32 centrifuged for 10 min at 13000 g, and the absorbance of the supernatant was read at 540 nm. 33 Assay blanks contained enzymatic extracts boiled at 100°C for 3 min before the addition of 34 KNO 3 and NADH. To measure the inhibition of NR activity by tungstate, enzymatic extracts 1 were first preincubated with NaTg for 15 min at ambient temperature before activity 2 measurement. On the basis of inhibition experiment data (Fig. S5), a concentration of 1 mM 3 NaTg was routinely used for in vivo and in vitro experiments. To assess the effectiveness of 4 Tg in vivo, nodules or bacteroids were incubated for 4 h in the presence of 1 mM NaTg, 5 proteins were extracted and NR activity was measured as described above. 6 NiR activity was assayed at 28 °C by following nitrite consumption from the assay 7 mixture using the dithionite-methylviologen method. The reaction medium (1 ml mM Na 2 EDTA, in a mortar and pestle. After thawing, the extract was taken and the mortar 20 was rinsed with 200 µl of perchloric acid solution which was then pooled with the extract. 21 Sample was centrifuged, for 5 min at 13000 g. The supernatant was quickly and carefully 22 neutralized at pH 5.6-6.0 using a 2 M KOH -0.3 M MOPS solution. KClO 4 precipitate was 23 discarded by centrifugation (5 min, 13000 g). Adenine nucleotides of the supernatant were 24 measured in a luminometer (Bio-Orbit, Turku, Finland) using the ATPlite 1 step assay system 25 (ATPLT1STP-0509, Perkin Elmer, Inc., Waltham, MA, USA) according to manufacturer 26 instructions. 27 28

Supplemental data 33
The following materials are available in the online version of this article. Effector concentrations were 10 mM NaNO 3 (NO 3 -) 1 mM NaNO 2 (NO 2 -) and 1 mM sodium 24 tungstate (Tg). Data are the means ± SD of 5 independent experiments assayed in duplicates.