3.1. Mutants Defective in Symbiotic Nitrogen Fixation and Showing Induced Defense Responses Are Deficient in the NAD1 Gene
A collection of
M. truncatula mutants showing defects in symbiotic nodule development and function was identified and described previously [
35]. One of the ineffective mutants in this mutant collection, termed 7Y, showed the symptoms of nitrogen starvation and developed slightly cylindrical yellowish or brownish nodules following inoculation with compatible rhizobia ([
35] and
Supplemenatary Figure S1). The brown pigmentation associated with strong autofluorescence that appeared after inoculation suggested the induction of defense responses in 7Y mutant nodules ([
35] and
Figure 1B,G). Based on genetic and sequence analyses described below, another allele of 7Y, the N5896A mutant showing similar nodulation phenotype (
Figure 1C,H), was identified in a separate symbiotic screen of a
M. truncatula Tnt1/MERE1 insertion mutant collection established in the Jemalong 2HA background [
43,
44].
Our previous study [
35] revealed that 7Y is not an allele of the previously described
dnf symbiotic mutants [
45], and identified the map position of the 7Y mutant locus on the upper arm of chromosome 7 of
M. truncatula between the genetic markers MtB243 and MtB183 ([
33] and
Figure S2A). A map-based cloning approach was applied to identify the gene affected in the
nad1-3 mutant. An extended segregating population of 727 F2 individuals was used to map the mutant locus between genetic markers EF4142291 and h2_96b16t19 (
Figure S2A). The analysis of the sequence of this genomic region of 153 kb revealed 24 predicted gene models (Mt4.0 JBrowse) [
46] including the nodule specific
NAD1 gene (
Medtr7g022640) [
21], providing a good candidate for the mutated gene in 7Y.
Oligonucleotide primers were synthesized for the
NAD1 gene and PCR reactions were carried out to amplify genomic fragments of
NAD1 from the 7Y mutant. The sequence analysis revealed a 50-bp deletion starting at 26 bp downstream of the predicted AUG start codon in 7Y; therefore, 7Y is hereafter termed
nad1-3 (
Figure 2A). The gene structure of
NAD1 available at the Medicago genome database (Mt4.0 JBrowse) is based on an expressed sequence tag (EST) sequence (EST483823; GenBank BG582085.1), predicting that the
NAD1 gene consists of four exons encoding a 70 amino acid long putative protein (Mt4.0 JBrowse and
Figure 2A). Because the 50-bp deletion is located in the first predicted intron of this
NAD1 gene model, cDNA of
NAD1 was generated and sequenced to search for other transcript versions. The sequence analysis revealed that the gene is actually composed of two exons and one intron (
Supplemenatary Figure S2A). A protein of 96 amino acid residues with two transmembrane domains is encoded in the first exon, as described by [
21]. Based on this gene structure, the 50-bp deletion in
nad1-3 mutant is located in the first exon of
NAD1 (
Figure 2A). This deletion generates a frameshift and a premature translation termination in the position 202 bp of the coding sequence, and, thus, a truncated and inaccurate protein with non-NAD1-specific amino acids from the position of the ninth residues.
To identify which
Tnt1/MERE1 insertion caused the deficiency in the ineffective symbiotic mutant N5896A, the co-segregation of flanking sequence tags (FSTs) and the mutant phenotype—nitrogen deficiency and accumulation of brownish pigmentation in the nodules—was analyzed in an F2 population generated by self-pollinating an F1 plant originating from a back-cross. The genetic analysis of 40 F2 plants, including seven homozygote mutants, revealed co-segregation of the mutant phenotype and the presence of a
Tnt1 retroelement inserted into the 5′-untranslated region (UTR) of the
NAD1 gene (
Figure 2A), indicating that this ineffective mutant carries an additional allele of
nad1; therefore, N5896A ineffective mutant hereafter termed
nad1-4. To analyze the effect of the
Tnt1 insertion in the 5′-UTR on the expression of
NAD1, the cDNA samples that were prepared from
nad1-4 nodules were used in RT-PCR which showed the absence of NAD1 transcript indicating that the
Tnt1 insertion in
nad1-4 abolished the expression and activity of
NAD1 (
Supplementary Figure S2B).
3.2. Both Exons of NAD1 Are Required for Complete Rescue of the Symbiotic Phenotype of nad1-3 and nad1-4 Mutants
In order to test whether the entire transcript of
NAD1 harboring two exons or other alternative transcripts are able to restore the symbiotic phenotype of
nad1-3 and
nad1-4, we carried out genetic complementation experiments using the following constructs: the genomic copy of
NAD1, the first exon (containing the entire coding sequence for the protein), the first exon with a 78-bp deletion mimicking the first intron predicted at the Medicago genome database based on the BG582085 EST and the whole gene with the same 78-bp deletion fused to the native
NAD1 promoter (
Figure S3). These constructs were introduced into
nad1-3 and
nad1-4 roots using
Agrobacterium rhizogenes mediated hairy root transformation. The roots were inoculated with
S. medicae strain WSM419 (pXLGD4), and the symbiotic phenotype was assessed for the presence of infected cells in the nodules with the help of X-gal staining, as well as for the vanishing of the brown pigmentation, the hallmark of the
nad1 mutation. Brownish empty nodules were only formed on the
nad1-3 roots transformed either with the empty vector (
Supplementary Figure S3D–F) or the constructs harboring the 78-bp deletion that mimics the first intron prediction at the
Medicago genome database, based on the BG582085 EST sequence (
Supplementary Figure S3L,N). This result proved that the lack of this sequence presumed to be the “first intron” abolishes the function of NAD1, so it is an integral part of the first exon. The mixture of wild-type nodules with proper zonation (
Supplementary Figure S3J) and nodules showing brown pigmentation (
Supplementary Figure S3K) developed on
nad1-3 roots transformed with the first exon harboring the coding sequence for the NAD1 protein, suggesting the loss of proper regulation of
NAD1 in the absence of the rest of the transcript and/or the intron and/or the 3′-UTR to restore the symbiotic phenotype. The nodules on
nad1 mutants transformed with the complete
NAD1 gene were pink, indicating that they were functional nodules (
Supplementary Figure S3G). The X-gal stained nodules showed the typical zonation of the indeterminate nodules with invaded cells by rhizobia (
Supplementary Figure S3I,M), demonstrating that the complete gene is essential for the full rescue of the symbiotic phenotype of
nad1-3.
The results of the complementation experiments confirmed the proper structure of the NAD1 gene, and, in addition, the requirement of the non-coding second exon for the complete capacity of NAD1 to restore the symbiotic phenotype was demonstrated. These results point out the potential regulatory function of exon2 and/or 3′UTR of NAD1, which, in turn, suggests that the BG582085 sequence correspond to either an aberrant transcript or an alternative spliced product of the NAD1 gene.
3.3. NAD1 Is Expressed in Infected Cells and the Gene Product Localizes to the Endoplasmic Reticulum
The expression of
NAD1 is nodule specific and its activation requires the formation of symbiotic nodules [
21]. The expression of
NAD1 was monitored during the nodule development using quantitative RT-PCR. The
NAD1 is expressed at a low level at 4 dpi with rhizobia and a strong increase in expression was detected at 6 dpi, and subsequent time points in wild type nodules (
Figure 3A). In
nad1-3 nodules, the
NAD1 was activated similarly, but it was expressed at a lower level when compared to wild type samples between 8 and 21 dpi (
Figure 3A). We detected a great decline in the
NAD1 expression 21 dpi, which is probably associated with the advanced stage of the necrotic phenotype of
nad1-3 nodules.
To further analyze the expression pattern of
NAD1, its promoter fused to the β-glucuronidase reporter gene was introduced into wild-type
M. truncatula roots using
A. rhizogenes-mediated hairy root transformation. The roots were inoculated with
S. medicae strain WSM419 (pXLGD4) and nodules on transformed roots were monitored for GUS activity using 5-bromo-6-chloro-3-indolyl β-D-glucopyranosiduronic acid cyclohexylammonium salt (Magenta-Gluc) substrate and then stained for β-galactosidase activity to visualize the presence of rhizobia at 14 dpi. GUS activity was found in the cells of the invasion zone, the intermediate zone, and the nitrogen fixation zone, as well (
Figure 3B). All of these cells were occupied by rhizobia (
Figure 3C), indicating that
NAD1 is expressed in the infected cells of the symbiotic nodule.
The observed expression pattern of
NAD1 was in agreement with RNA-sequencing (RNA-seq) data of different nodule zones obtained by laser-capture microdissection [
47]. The
NAD1 gene is induced in the infection zone and reached its maximum activity in the transition between the infection and nitrogen fixation zones, and maintained in the nitrogen fixation zone (
Figure 3D). Based on the RNA-seq data, the
DNF2 gene also required for the suppression of plant immunity during nitrogen fixing symbiotic interaction [
19], shows co-expression with
NAD1 in the zones of
M. truncatula nodules (
Figure 3D). The expression pattern of
NAD1 in the nodule zones indicates the continuous requirement of
NAD1 in infected nodule cells during the symbiotically active lifetime of the nodule.
The sub-cellular localization of the proteins might help to elucidate their functional properties. The NAD1 is predicted to have two transmembrane domains [
21], suggesting to be localized to the plasma membrane and/or subcellular membrane compartment. In order to study the localization of NAD1, constructs coding for
NAD1 proteins tagged either C- or N-terminally with GFP or c-myc epitopes were created and introduced into
nad1-3 roots using hairy root transformation. Unfortunately, none of the constructs restored the symbiotic phenotype, indicating that these tags were probably interfered with the function of NAD1 (data not shown). To further investigate the subcellular localization of NAD1, the constructs of N- and C-terminal GFP fusions of the full-length
NAD1 cDNA were expressed transiently under the control of the CaMV p35S promoter in
Nicotiana benthamiana leaves. According to the detected GFP fluorescence in the leaf epidermal cells, NAD1 was found to be associated exclusively with the endoplasmic reticulum (ER) membrane network (
Figure 4). In order to confirm the localization pattern of the NAD1 protein, the p35S::GFP-NAD1 fusion was co-transformed with organelle marker constructs [
48] harboring well-established targeting sequences fused to mCherry. We could not detect any significant overlap with the Golgi, tonoplast and peroxisome-specific markers (data not shown), however NAD1 co-localized with the ER-specific marker (
Figure 4C).
NAD1 codes for a small protein with a 10.7 kDa molecular weight. It has been generally admitted that the diffusion limit set by the nuclear pore for protein is 60 kDa. Moreover, the nuclear localization of a GFP3 oligomer protein, whose size is around 90 kDa, was observed recently [
49]. Thus, GFP fusions below this size without any localization signal, theoretically, can enter into the nucleus by passive diffusion. When considering the fact that GFP-NAD1 chimeric protein is still under this size limit (38 kDa), there was a theoretical chance that GFP carries the small NAD1 into the nucleus. To exclude this possibility, we co-expressed the p35S::GFP-NAD1 constructs with the pUbq10::RFP-NSP1 [
40], showing strong nuclear localization. According to this experiments, the GFP-NAD1 fusion protein was clearly excluded from the nucleus (
Figure 4F).
Our transient co-localization studies in
N. benthamiana leaves suggesting the association of the NAD1 signal to the endoplasmic reticulum (ER) is in agreement with the previous results that were obtained from localization studies in restored functional nodules and in
Arabidopsis protoplasts [
21]. In this previous study, the flag-tagged NAD1 restored the symbiotic phenotype of
nad1 nodules, indicating that the flag-tag did not interfere with the activity of NAD1 and the more sensitive immunofluorescence assay detected the signal of NAD1 on ER. The ER localization of NAD1 proved by three independent methods supports plausibility of the association of NAD1 to the ER. The ER serves many general functions in the cell, and have a role in protein and lipid biosynthesis and transport, protein folding, signaling (calcium storage) [
50], and even in immunity [
51], which can inspire several theories about the actual function of NAD1 in the root nodule symbiotic interaction.
3.4. The Defense Responses Are Induced Simultaneously in nad1 and dnf2 Mutants
The accumulation of brown pigmentation that was reported in the nodules of the
M. truncatula dnf2 mutant [
19] was similar to those detected in
nad1-3 and
nad1-4 nodules (
Figure 1). To characterize in more details and discriminate them if possible, the progression of the rhizobial infection in the nodules was analyzed and defense responses were compared in these and in generated double mutants. For this, longitudinal sections of wild-type,
nad1-3, nad1-4,
dnf2-1, and
nad1-3/
dnf2-1 double mutant nodules at 14 dpi with
Sinorhizobium medicae WSM419 (pXLGD4) constitutively expressing the
lacZ gene were stained for β-galactosidase activity and were investigated by light and fluorescence microscopy. Wild-type nodules showed the typical zonation of indeterminate nodules with fully infected cells in the nitrogen fixation zone (
Figure 1A), but
nad1,
dnf2, and
nad1/dnf2 mutant nodules did not show strong β-galactosidase activity, indicating the low occupancy of rhizobia in nodule cells (
Figure 1B–E). The nodulation phenotype of the single
nad1,
dnf2, and the
nad1/dnf2 double mutants were indistinguishable, indicating that the two mutations affect the same or very similar pathways. Staining with X-gal did not clearly reveal the zonation of the mutant nodules, therefore, they were further analyzed by confocal laser scanning microscopy using the nucleic acid-binding dye SYTO13 [
52]. Wild-type and mutant nodules did not show differences in the infection zone and the transition between the infection and nitrogen fixation zones (
Figure 1F–J). In wild-type nodules, differentiated bacteria in the interzone were oriented toward the vacuoles (
Figure 1K). In contrast, in the last layers of infected cell in the interzone of mutant nodules, rhizobia were disordered and slightly elongated (
Figure 1L–O). The mutant nodules in the fixation zone were devoid of bacteria (
Figure 1G–J). Moreover this region displayed strong autofluorescence in all of the mutant nodules (
Figure 1G–J), as described previously for
nad1 nodules [
21,
35], suggesting the accumulation of phenolic compounds. The presence of phenylpropanoids was confirmed with the potassium permanganate/methylene blue staining procedure that reveals polyphenolics with blue coloration and with the toluidine blue dye which stained phenolic compounds dark greenish (
Figure 5J–L,P–R and
Figure 6A–E). The deposition of polymeric phenols is considered as a defense response [
53], and, therefore, the induction (or lack of suppression) of defense responses can be anticipated in
nad1,
dnf2, and
nad1/dnf2 mutant nodules. To assess the activation of plant defense responses in
nad1-3,
dnf2, and
nad1-3/dnf2 nodules, we monitored the transcriptional activation of defense-related marker genes at 14 dpi with
S. medicae strain WSM419 using quantitative RT-PCR. A
chitinase (
Medtr3g118390), the
NDR1 (a Non-race-specific Disease Resistance,
Medtr5g076170), a
flavonol synthase (
Medtr5g055680), a
PR10 (
Medtr2g035150), a
plant invertase (
Medtr4g101760), and a
Kunitz-type trypsin inhibitor (
Medtr6g078250) defense-related genes were up-regulated in the
nad1-3,
dnf2-1, and
nad1-3/dnf2-1 double mutant nodules (
Figure 1P,Q), confirming induced defense responses, similarly, as it was found previously in the
M. truncatula dnf2 [
19] and
symcrk [
20] mutants.
As commented before, microscopy studies revealed that nodules of
nad1-3 and
nad1-4 mutants showed brown pigmentation that fluoresced, indicating the accumulation of phenolic compounds at 14 dpi (
Figure 1). To define at what stages of the symbiotic nodule development the defense responses are activated, the time course of phenolic compound accumulations in nodules was analyzed using potassium permanganate/methylene blue staining. In wild-type Jemalong nodules, cells were colonized by bacteria in the infection, inter and nitrogen fixation zones and no polyphenol accumulation could be detected at any time point of the analysis (
Figure 5A,D,G,J,M,P). In the mutant nodules, no polyphenolic staining was observed at 6 dpi (
Figure 5E,F). Blue precipitates, indicating that the presence of polyphenolics could be detected at 8 and 14 dpi that correspond to the sites of brown pigmentation (
Figure 5K,L,Q,R). Consistent with the accumulation of phenolic compounds, the expression of defense-related genes were strongly activated at 8 (the
chitinase and
PR10) and 10 dpi (
NDR1 and the
Kunitz-type trypsin inhibitor) in
nad1-3 mutant nodules, relative to wild-type (
Figure 5S,T).
To reveal any differences in the induction of plant defense responses, the kinetics of the appearance of brown pigmentation was analyzed in
nad1-3,
dnf2-1 and
nad1-3/dnf2-1 double mutants inoculated with
S. meliloti strain WSM419 (pXLGD4). Nodules were stained for β-galactosidase activity and the microscopy analysis revealed invaded nodule cells in the mutants at 6 dpi and the brown pigmentation appeared at 8dpi in
dnf2-1 and
nad1-3/dnf2-1 nodules, similar to
nad1-3 (
Supplementary Figure S4H,M,R). These results indicate that
nad1 and
dnf2 mutant nodules not only show comparable induction of defense-like responses but these reactions exhibit similar kinetics, as well (
Figure 5 and
Supplementary Figure S4).
3.5. Defense-Like Responses Are Induced upon the Internalization of the Symbiotic Nodule Cells by Rhizobia and Rapidly Induce the Degradation of Host and Bacterial Cells
Because the defense responses were induced between 6 and 8 dpi in
nad1 and
dnf2 nodules, we analyzed the viability of rhizobia using live and dead staining with the fluorescent nucleic acid-binding dye, SYTO9 and PI using confocal microscopy. SYTO9 detects living bacteria showing green fluorescence while plant nuclei and membrane compromised bacterial cells take up PI and show red fluorescence. At 6 dpi, viable green fluorescent bacteria were detected in wild-type and
nad1-3,
dnf2-1 and
nad1-3/dnf2-1 double mutant nodules and only the meristematic region composed of non-infected small cells rich in cytoplasm as well as plant nuclei displayed red fluorescence (
Supplementary Figure S5B,C,H,I,N,O,T,U). In the sections of 8 dpi wild-type nodules, bacteria fluoresced green (
Supplementary Figure S5E,F). In mutant nodules, a large area with autofluorescence appeared few layers below the invaded cells and only few cells containing dead red fluorescent bacteria could be observed in the transition zone between the invaded cells and the area accumulating phenolic compounds, suggesting the rapid disintegration of dead bacteria in nodule cells (
Supplementary Figure S5K,L,Q,R,W,X).
To further analyze the induced defense responses in
nad1-3 nodules at ultrastructural level, electron microscopy studies on 8 and 18 dpi nodules were carried out. The SEM analysis of developing 8 dpi wild-type nodules detected elongated bacteroids that were orientated towards the central vacuoles and were encompassed by cytoplasmic matrix in the cells of the first layers of the nitrogen fixation zone (
Supplementary Figure S6B,C). The nodule cells in the corresponding region in
nad1-3 nodules have thickened cell walls and they were either devoid of rhizobia or contained slightly elongated, disordered, and aggregated bacteria without surrounding cytoplasmic matrix, indicating the degradation of bacterial and plant cells (
Supplementary Figure S6E,F). Transmission electron microscopy images showed differentiated bacteroids, reaching 6–8 μm in length, in the nitrogen fixation zone of 18 dpi wild-type nodules (
Supplementary Figure S6G,J). In
nad1-3 nodules, cells in the last layers of infected cells had thickened cell walls and often cell wall-like deposits surrounding electron dense necrotic bacteria were detected (
Supplementary Figure S6H,I). In root distal part of the
nad1-3 nodules corresponding to zone III, disorganized cellular structure and hydrolyzed cell wall remnants were identified, indicating the necrosis of both the host cells and bacteroids (
Supplementary Figure S6K). Older cells in the proximal part of the nodules were almost empty, showing the advanced degradation of symbiotic cells in
nad1-3 nodules (
Supplementary Figure S6L).
In our experimental conditions, the cells in the invasion and the first layer of the intermediate zone of the indeterminate
M. truncatula nodules were invaded by rhizobia at 6 dpi, but no visible signs of induced defense response or induction of defense-related genes was observed in
nad1 nodules (
Figure 5,
Supplementary Figures S4 and S5). In contrast, the viability staining of bacteroids, the appearance of accumulated polyphenolics and the activation of defense-associated genes suggested the loss of the control over the plant immune responses in
nad1 nodules at 8 dpi. Moreover, electron micrographs of
nad1-3 showed the rapid death of rhizobia and the lysis of the host cell resulting in almost empty cells in the proximal part of the nodule, indicating the quick progression of necrosis in 8 dpi
nad1 nodules (
Supplementary Figure S6). Our microscopy analyses highlighted that bacteria not only filled the host cells in the transition zone between invasion and nitrogen fixation zone, but their differentiation was initiated in
nad1 nodules.
3.6. The Function of NAD1 Precedes DNF1 and NAD1 Acts Independently and Prior to Rhizobial BacA in the Symbiotic Process
To further explore at what stage the symbiotic interaction is blocked, the nodulation phenotypes of
nad1-3,
nad1-4,
nad1-3/dnf1-1,
nad1-3/dnf2-1 and
nad1-3/lin-2 mutant plants were analyzed after inoculation with rhizobial strains deficient in the production of the NF (
nodA) or the succinoglycan EPS I (
exoY) or defective in bacteroid differentiation (
bacA). To assess the induction of plant defense responses in wild-type and mutant nodules that were arrested at different stages of the symbiotic interaction, sections of the nodules were stained with toluidine blue and analyzed for the presence of phenolic compounds at 14 dpi. Inoculation of the symbiotically less effective rhizobial strain
S. meliloti 1021 [
54] caused a similar accumulation of polyphenolics in
nad1,
dnf2, and
nad1/dnf2 nodules (
Figure 6A–E) that was found with
S. medicae WSM 419, indicating that the
nad1 phenotype is not strain dependent. This allowed for us to analyze the symbiotic phenotype of wild-type and
nad1-3, nad1-4, dnf2-1, and
nad1-3/dnf2-1 mutants that were inoculated with symbiotic bacterial mutants generated in the
S. meliloti 1021 strain.
The
exoY mutant of
S. meliloti strain 1021 defective in the production of succinoglycan fails to initiate infection thread formation, hence ineffective nodules without rhizobial invasion are formed on
Medicago roots ([
55] and
Figure 6F). In mutant nodules, no accumulation of the phenolic compounds could be observed at 14 dpi, suggesting the requirement of bacterial colonization for the induction of plant defense in these symbiotic mutants (
Figure 6F–J).
The BacA protein protects
S. meliloti against the antimicrobial activity of NCR peptides [
8], and, therefore, BacA is essential for bacteroid development in galegoid legumes [
56].
bacA mutants induce indeterminate nodule formation on
Medicago roots but rhizobia are killed soon after release from infection threads prior to bacteroid differentiation [
57]. Accordingly,
bacA mutant rhizobia lysed and non-infected cells were observed in wild-type Jemalong nodules (
Figure 6K). This inefficient interaction does not trigger phenolic compounds production [
23]. In contrast, the nodules of
nad1, dnf2-1, and
nad1-3/dnf2-1 mutants that were elicited by
bacA show accumulation of phenolic compounds, indicating that the defense responses were rapidly activated following the release of mutant rhizobia into the host compartment (
Figure 6L–O).
To correlate the induction of
NAD1 and plant defense responses in nodules elicited by wild-type and symbiotically deficient rhizobial strains, the transcriptional activation of
NAD1 and the defense-related
PR10 was monitored in wild-type Jemalong roots or nodulated roots at 14 dpi. The basic expression level of the
PR10 gene detected in nodules elicited by mock or nodulation factor (NF) deficient
S. meliloti [
58] was suppressed following inoculation with wild-type rhizobia or bacteria defective in later stages of the symbiotic process, suggesting the suppression of
PR10 upon the initiation of infection thread development (
Figure 6P). The induction of the
NAD1 expression following the inoculation with
bacA mutant and wild-type
S. meliloti but not with
S. meliloti nodA and
exoY mutants indicated the need of bacterial release for the induction of
NAD1. These results showed that NAD1 starts to function after the uptake of rhizobia into the host cells but prior rhizobial differentiation in
M. truncatula nodules.
The characterization of the mutant phenotype of
nad1 and other ineffective symbiotic nodulation double mutants with
nad1 enables the determination of the functional hierarchy of the impaired genes. The symbiotic phenotype of single and double mutants of
nad1-3,
lin-2,
ipd3-1, and
dnf1-1 were analyzed at 14 dpi with
S. medicae (pXLGD4). The
lin-2 mutant is deficient in the early stage of the rhizobial symbiotic process and although nodule primordia are formed, the infection thread development is arrested in the root hairs ([
59] and
Figure 7B). The nodules on
ipd3 mutant roots are more developed when compared to
lin nodule primordia, but
ipd3 nodules are impaired in release of rhizobia from infection threads, and, thus, bacteria do not colonize nodule cells ([
60] and
Figure S7C). The
nad1-3/lin-2 and
nad1-3/ipd3-1 double mutants showed the same defects of infection and bacterial release as
lin-2 and
ipd3-1, respectively (
Figure 7F,G), without phenolics accumulation, indicating that bacterial release and invasion is required for the function of
NAD1.
The
DNF1 gene coding for a symbiosis-specific component of the secretory pathway is required for bacterial differentiation and symbiosome development and in
dnf1 mutant, released bacteria are blocked in the very early stage of bacteroid development [
6,
7]. In the undeveloped nodules of
dnf1, no distinct developmental zones could be observed, but nodule cells were colonized by rhizobia with no accumulation of phenolic compounds (
Figure 7D). The presence of phenolic compounds in the
nad1-3/dnf1-1 double mutants suggests the activation of plant defense responses (
Figure 7H). This was consistent with our previous findings that the proper acting of NAD1 is required before bacterial differentiation.
Our microscopy analysis of the nodulation phenotypes of single and double ineffective plant symbiotic mutants that were inoculated with wild-type and symbiotic mutant rhizobia allowed for us to position the
NAD1 gene in the symbiotic process. First of all, the nodulation phenotype of
nad1 and
nad1-3/ipd3-1 double mutants that were inoculated with
exoY mutant and wild-type rhizobia, respectively, indicated that bacterial release and colonization of nodule cells are required for the function of NAD1. The observation that
dnf2,
nad1, and
nad1/dnf2 nodules, as elicited by
bacA mutant rhizobia, did accumulate phenolic compounds suggested that the repression of plant defense responses is independent of the differentiation of rhizobia at this stage of the interaction. The symbiotic interaction is blocked at the same developmental stage and the expression of the defense-associated genes displayed the similar induction in
nad1/
dnf2 double mutant as in the single mutant nodules, suggesting that
NAD1 and
DNF2 operate close together in the same developmental pathway. The observation that the plant defense is suppressed irrespective of the differential status of bacteria was further evidenced by the microscopy analysis of the
nad1-3/ dnf1-1 double mutant that was elicited with wild-type bacteria. We detected brown pigmentation in
nad1-3/dnf1-1 double mutant nodules, although this defense reaction was not as robust as in
nad1-3 nodules, which probably correlates with the lower number of infected cells, and the reduced zonation of
dnf1 mutant nodules possessing mainly an enlarged invasion zone [
7]. This result is in agreement with the observation that
NAD1 expression was reduced in wild-type plants nodulated with
bacA mutant rhizobia. Taken together, these results indicate that the defense-like responses were induced irrespectively of the differential phase of rhizobia in
nad1-3 nodules, similarly to what was found in the
dnf2 mutant [
23].