Evolutionary Analysis of Six Gene Families Part of the Reactive Oxygen Species (ROS) Gene Network in Three Brassicaceae Species

Abstract

Climate change is expected to intensify the occurrence of abiotic stress in plants, such as hypoxia and salt stresses, leading to the production of reactive oxygen species (ROS), which need to be effectively managed by various oxido-reductases encoded by the so-called ROS gene network. Here, we studied six oxido-reductases families in three Brassicaceae species, Arabidopsis thaliana as well as Nasturtium officinale and Eutrema salsugineum, which are adapted to hypoxia and salt stress, respectively. Using available and new genomic data, we performed a phylogenomic analysis and compared RNA-seq data to study genomic and transcriptomic adaptations. This comprehensive approach allowed for the gaining of insights into the impact of the adaptation to saline or hypoxia conditions on genome organization (gene gains and losses) and transcriptional regulation. Notably, the comparison of the N. officinale and E. salsugineum genomes to that of A. thaliana highlighted changes in the distribution of ohnologs and homologs, particularly affecting class III peroxidase genes (CIII Prxs). These changes were specific to each gene, to gene families subjected to duplication events and to each species, suggesting distinct evolutionary responses. The analysis of transcriptomic data has allowed for the identification of genes related to stress responses in A. thaliana, and, conversely, to adaptation in N. officinale and E. salsugineum.

1. Introduction

The current climate change increases intense episodes of drought and floods also leading to higher salinity rates at the soil surface. In the case of flooding or high salt concentration, plants must withstand partial or total immersion, as well as osmotic stress leading to Reactive Oxygen Species (ROS) homeostatic variations involving oxidoreductases (ORs) [1,2].

To adapt to various biotic and abiotic stresses, plants have evolved very complex regulatory mechanisms that can modulate the cellular concentration of ROS, including superoxide anion (O₂^λ−), hydrogen peroxide (H₂O₂), and hydroxyl radical (HO^λ). High levels of ROS are toxic and lead to oxidative damage. When their concentrations are controlled, ROS participate in some biological processes such as cell growth, programmed cell death, and signaling [3]. ROS homeostasis is determined by the interplay between ROS-producing and scavenging mechanisms controlled by haem and non-haem peroxidases, as well as other ORs that are part of the so-called ROS gene network [4]. These proteins belong to more than 100 classes and are encoded by multigenic families comprising two to more than one hundred members (https://peroxibase.toulouse.inrae.fr/, accessed on 30 November 2023). Members of this ROS gene network have been detected in many species [5], and some families such as superoxide dismutases (SODs) and peroxiredoxins have been found in all kingdoms. Regulation of ROS homeostasis by the ROS gene network is a major player in salinity and hypoxia acclimation and adaptation [6,7,8].

Six gene families belonging to this ROS gene network have been selected for this study. (i) The ascorbate peroxidases (APxs), which are encoded by small multigenic families (1–10 members), are only present in chloroplastic organisms and are highly conserved between species. They play a key role in H₂O₂ homeostasis [9]. (ii) The monofunctional (typical) catalases (Kats) are present in all aerobic organisms in which they transform H₂O₂ into H₂O and oxygen [10]. The number of Kat genes can vary from one to three copies. (iii) The class III peroxidases (CIII Prxs), which are encoded by large multigenic families (2–150 members), are only found in Viridiplantae and their genes have been subjected to numerous duplication events, which can be species specific [11]. CIII Prxs are mainly predicted as cell wall proteins and participate in many different biological processes such as cell wall elongation and stiffening, or protection against pathogens [12,13]. (iv) The α-dioxygenases (DiOxs) belong to small families (1–2 members). Their genes are only found in plants and are rarely subjected to duplication [14]. They catalyze the initial step of the α-oxidation of polyunsaturated fatty acids. Their activation is part of the defense mechanisms induced to protect cells from oxidative stress [15]. Moreover, their enzymatic activity is part of the plant responses to saline stresses [16]. (v) The NADPH oxidases (RBOHs) are encoded by small multigenic families (1–10 members) highly conserved between species. They are transmembrane proteins located at the plasma membrane producing O₂^λ− extracellularly [17]. (vi) SODs are encoded by small multigene families (1–10 members). They catalyze the dismutation of O₂^λ− into H₂O₂. In plants, there are three SOD classes (FeSOD, MnSOD and CuZnSOD), and two SOD-related classes: SOD-like (SDL) and copper chaperone for SOD (CCS) [18].

Brassicaceae is a large plant family composed of 57 tribes, 349 genera and 4140 species [19]. It is highly studied for its fundamental, evolutionary and agronomical aspects. In the NCBI database (https://www.ncbi.nlm.nih.gov/genome/?term=Brassicaceae, accessed on 30 November 2023) [20], 96 genomes are available, as well as 237,803 Sequence Read Archives (SRAs), mainly from Arabidopsis thaliana (166112) and Brassica napus (21486). For this study, three Brassicaceae species have been chosen based on the following criteria: genomic sequences and transcriptomic data availability, taxonomic proximity, and capacity to adapt to hypoxia or salt stress. A. thaliana has been chosen as a “reference” species (not halophytic, not resistant to hypoxia), Eutrema salsugineum (formerly Thellungiella halophila) as a halophytic species, and Nasturtium officinale (watercress) as a semi-aquatic species adapted to hypoxia.

To better understand the mechanisms of plant adaptation to salt stress or hypoxia, we have performed two complementary approaches. On the one hand, we have performed a complete and exhaustive phylogenomic analysis to see whether the number of genes encoding each of the above-described OR families was correlated to the capacity of the plants to cope with salt stress or hypoxia. In this respect, to complement the available genomic sequence of N. officinale, we provide new genomic data resulting from the assembly of larger DNA fragments compared to a previous study [21]. As a tetraploid species, N. officinale was also used to study the pseudogenization process/rate of two gene copies (ohnologs) in the six OR gene families. On the other hand, taking advantage of the availability of transcriptomics data, we have looked at the changes in the level of expression of the selected ORs to see if the regulation of some of them is modified depending on the capacity of the plants to cope with salt stress or hypoxia.

2. Results and Discussion

2.1. A New Genome of N. officinale

When this work started, no genomic data were available for N. officinale. Commercially available seeds were sown, and the genomic DNA was extracted from young leaves of 17 d old plantlets. The quality of the DNA has been checked and most of the DNA fragments were larger than 90 kb (Figure S1). The HiFiasm de novo assembler, initially designed for PacBio HiFi reads, produced a complete primary sequence comprising 2667 scaffolds, with the largest one at almost 12 Mb and more than 50% larger than 1 Mb (N50) (Figure S2). The overall coverage of the genome was 57x The calculated assembly size of N. officinale (primary total length of 647 Mb) was twice as big as the heterozygous genome size of 395 Mb estimated with the jellyfish tool [22], consistent with the value of 377 Mb previously determined by flow cytometry [23]. The observed difference suggested that the sequenced DNA could correspond to an admixture of genotypes. In addition, a BUSCO of 98.8% was calculated for the primary assembly with the brassicales_odb10.2020-08-05 lineage as a reference. This result indicated a high completeness of the genome sequence. A duplicated BUSCO of 74.1% could also be calculated, which was consistent with the tetraploidy of N. officinale [24].

At the time that we had finalized the genomic sequence of N. officinale, another sequencing project was released [21]. This former sequencing was performed with Illumina short reads, resulting in a number of scaffolds four times higher than in our work (10793). The value of N50 was 94.2 kb and the number of contigs was 14,564 (https://www.ncbi.nlm.nih.gov/datasets/genome/GCA_900406445.1/, accessed on 6 November 2023). The overall sequence was 216.1 Mb in length. This size was smaller than that obtained with our sequencing data, consistent with the fact that our data corresponded to several populations, most probably three, with a major one named Pop 1 (Table S1). This was mostly due to the use of commercial seeds containing mixed populations. In addition, pairs of genes with a lower percentage of identity between them than with genes from different populations (90% versus 98%, respectively) have been considered as ohnologs and numbered A and B. Ohnologs are actually paralogs originating from a whole-genome duplication event [25]. The BUSCO and the coverage of the former genome were comparable with those of the genome we sequenced, with values of 98.3% and 59.0 x, respectively. As an example, only three sequences (NoffAPx02-1A, NoffPrx02-3B and NoffRboh[P]07-1B) were not found in this previous genome (Table S1).

As mentioned above, we could conclude that the new genomic data provided by this study correspond to three populations of plants with a major one (Pop 1 in Tables S1 and S2). The availability of the former genomic sequence of N. officinale allowed us to clarify the distribution of the ohnolog and homolog genes between these populations (Tables S1 and S2). The complementary analyses of the two genomes show the presence of pairs of ohnologs and pseudogenes, as well as the gain and loss of orthologs in the gene families of interest (see Section 2.2). On the other hand, our genome with much longer scaffolds allowed for the determination of more precise gene distribution in the genome in the case of genes located in close proximity (Table S1). Indeed, the 189 genes identified in N. officinale in this study have been detected in 81 scaffolds in the case of our assembly vs. 159 in the case of the former genome. For example, the A copies of NoffPrx69, 70, 71, 72, 73 (our work) are located on the same scaffold (ptg000119), whereas the corresponding genes of the previous genome are located on three different scaffolds (186 g, 3 g and 655 g) (Table S1). A similar situation was found for the B copies of the same genes. The localization of these five Prx genes on a single scaffold reflects the distribution of the ortholog genes in the A. thaliana genome.

The next step of this study was to perform a comparative phylogenetic analysis of six families of genes encoding proteins of the ROS network in three Brassicaceae adapted to different environmental conditions, N. officinale, E. salsugineum and A. thaliana.

2.2. Gene Duplications, Gains and Losses

A. thaliana and N. officinale belong to the Camelineae and Cardamineae tribes, respectively, which have diverged 20–30 MYA [26,27]. E. salsugineum, which belongs to the Eutremeae tribe, has diverged earlier, i.e., 24–35 MYA. A. thaliana and E. salsugineum are diploid species. It was recently shown that E. salsugineum did not experience an independent whole genome duplication event [28]. Conversely, N. officinale is a tetraploid. However, since this genome duplication event, some ohnologs could have been lost or pseudogenized.

In E. salsugineum, 10 APxs, 3 Kats, 74 CIII Prxs, 2 DiOxs, 11 RBOHs and 9 SODs have been found, including only eight pseudogenes for CIII Prxs (

Table 1. APxs, Kats, CIII Prxs, DiOxs, RBOHs and SOD genes have been annotated from the three Brassicaceae species (Table S2). Open reading frames with relevant predicted functional domains and pseudogenes (Ψ, corresponding to truncated amino acid sequence, nucleotidic sequences with in-frame stop codons, frameshifts or gaps) are indicated. Gene gains and losses have been determined using the A. thaliana orthologs as references. Only A and B genes of the major population of N. officinale have been considered for the calculations (Pop 1, Table S2).

Gene Family		E. salsugineum	A. thaliana	N. officinale
APx	Total gene number	10	8 + 1 Ψ	12 + 2 Ψ
	% Ψ	0	11.1	14.3
	Gene gain	2	0	0
	Gene loss	0	0	2
	% gene lost + Ψ	0	nd	25.0
Kat	Total gene number	3	3	5
	% Ψ	0	0	0
	Gene gain	0	0	0
	Gene loss	0	0	1
	% gene lost + Ψ	0	nd	16.6
CIII Prx	Total gene number	66 + 8 Ψ	73 + 2 Ψ	111 + 17 Ψ
	% Ψ	10.8	2.7	14.1
	Gene gain	8	5	5
	Gene loss	3	0	16
	% gene lost + Ψ	15.9	nd	24.5
DiOx	Total gene number	2	2	3
	% Ψ	0	0	0
	Gene gain	0	0	0
	Gene loss	0	0	1
	% gene lost + Ψ	0	nd	25.0
RBOH	Total gene number	11	10	15 + 4 Ψ
	% Ψ	0	0	21.0
	Gene gain	1	0	0
	Gene loss	0	0	1
	% gene lost + Ψ	0	nd	25.0
SOD	Total gene number	9	9	17 + 1 Ψ
	% Ψ	0	0	5.5
	Gene gain	0	0	0
	Gene loss	0	0	0
	% gene lost + Ψ	0	nd	5.5

and Table S2). The orthogroup analysis showed that only three CIII Prxs orthologs have been lost compared to A. thaliana, meaning that 15.9% of the orthologs of the CIII Prxs have been lost or are in a process of pseudogenization in E. salsugineum. No loss was observed for the five other gene families studied.

In N. officinale, 14 APxs, 5 Kats, 128 CIII Prxs, 3 DiOxs, 19 RBOHs and 18 SODs have been found, including 2, no, 18, no, 4 and 1 pseudogenes, respectively (Table 1 and Table S2). The orthogroup analysis showed that 2, 1, 16, 1, 1 and no orthologs, respectively, have been lost compared to A. thaliana, meaning that 23 to 25% of the ohnologs of five out of the six OR families of interest have been lost or are on the way to be lost by pseudogenization in N. officinale. Most of the gain or loss events are not common to N. officinale and E. salsugineum consistent with their distinct tribal classification: only 24% of the events observed in N. officinale also occurred in E. salsugineum (Table S2). However, some events are common (Prx04, 16, 17, 38, 44 and 58) and these events could be associated with a gene-specific selection pressure.

Two CIII Prx genes (Prx16 and 22) have been detected in Camelineae and Cardamineae while they are missing in Eutremeae (not found in E. salsugineum and Eutrema halophilum). The four CIII Prx genes (Prx08, 14, 63 and 68), which are missing in N. officinale and E. salsugineum, have been gained in A. thaliana through recent tandem, segmental or chromosomal duplications [29]. The chronology of these events is confirmed by the presence of Prx63 and Prx68 only in A. thaliana and in Arabidopsis species (e.g., Arabidopsis lyrata) and their lack in the other Brassicaceae tribes (e.g., Prx63 and 68 are not detected in Capsella rubella and Brassica napus). Since the five other gene families (APxs, Kats, DiOxs, RBOHs and SODs) are much smaller than the CIII Prx families, they could be less subjected to the gain or loss of genes or pseudogenization events. Alternatively, the higher evolution rate of CIII Prxs could also be related to the diversity of their functions vs. the more specific functions of the five other gene families.

Independently of the sequence, the percentages of identity between the protein sequences of a given ortholog between the three species are very similar, with an average value of 85.7 ± 7.4% (Table S2). In addition, the percentages of identity are in the same range between (1) the orthologs, on the one hand, and (2) the ohnologs on the other hand, with an average value of 89.0% ± 10.3% for the latter, meaning that the selective pressure is sequence dependent, but similar at the intra- or inter-species levels. The Prx35 sequences are among the less conserved between species exhibiting 88.0% identity between ohnologs and 87.0% identity between orthologs (N. officinale vs. A. thaliana). However, they show 93.0% identity intra-tribe (i.e., between Camelineae), as shown by a BLAST query using the Redoxibase (https://peroxibase.toulouse.inrae.fr/; accessed on 30 November 2023). Conversely, the Prx42 protein sequences are highly conserved between the three species exhibiting 98.0% identity between the N. officinale ohnologs, 97.0% identity between A. thaliana and N. officinale orthologs, and 97.0–98.0% with the protein sequences of their orthologs from other Brassicaceae tribes (Arabideae, Brassiceae, Thlaspideae). This could be related to the fact that Prx42 have crucial functions. Indeed, we have not yet been able to isolate prx42 mutants (F. Passardi and C. Dunand, unpublished work).

The Ka/Ks ratio reflects the molecular evolution rate of a gene family. The ratio between the non-synonymous (Ka) and synonymous (Ks) substitution rates can be superior, inferior or equal to 1, indicating positive, purifying, or neutral selection, respectively [30]. The Ka/Ks ratios were calculated independently for each OR gene family (Figure S3). All of them were smaller than 1 (mean values between 0.03 and 0.19). It means that the mutations detected between orthologs and ohnologs are mostly synonymous, indicating that the evolution of the different OR families was driven by purifying selection pressure. This is probably related to the fact that the regulation of ROS homeostasis is highly critical, thus leading to low rates of evolution in the OR family genes. In addition, the percentage of identity is lower when one ortholog or one ohnolog is a pseudogene. These two results confirmed the hypothesis that the divergence between sequences (orthologs and ohnologs) is sequence dependent, thus possibly related to selective pressure. A similar evolutive scenario has already been described for the non-specific lipid transfer protein gene family in Nicotiana tabacum with a high percentage of identity between homologous sequences of three Nicotiana sp. and very low Ka/Ks ratios [31].

2.3. Differentially Expressed OR Genes upon Salt Stress

The differential expression of the OR genes of interest has then been analyzed upon salt stress in A. thaliana and E. salsugineum. Two experiments have been selected: (i) A. thaliana seedlings have been submitted to salt stress (50 mM NaCl) for 10 d in in vitro cultures [32], and (ii) 6-week-old E. salsugineum seedlings have been irrigated over 24 h with a 300 mM NaCl solution [33]. Both RNA-seq analyses were performed on roots.

Under these conditions, 46 OR genes with a significant FC (fold change) value have been detected (Figure 1). Eight CIII Prxs (two are up-regulated and six are down-regulated) and one Kat show a change in their expression level in a similar way in both plants (Table S3). Conversely, one CIII Prx is up-regulated in A. thaliana (AtPrx10), whereas its ortholog is down-regulated in E. salsugineum (TsPrx10), and one APx is down-regulated in A. thaliana (AtAPx01), whereas its ortholog is up-regulated in E. salsugineum (TsAPx01) (

Table 2. OR genes regulated in contrasted manner between A. thaliana and E. salsugineum upon salt stress. The gene nomenclature is that of the online database Redoxibase (https://peroxibase.toulouse.inrae.fr/; accessed on 30 November 2023). Differential expression data are presented as log₂(FC).

E. salsugineum Redoxibase ID	E. salsugineum log2(FC)	A. thaliana log2(FC)	A. thaliana Redoxibase ID
TsPrx10/Thhalv10011619	−3.01	15.87	AtPrx10/At1g49570
TsAPx01/Thhalv10008402m	0.67	−1.41	AtAPx01/At1g07890

).

A total of 21 genes were specifically regulated in E. salsugineum, among which 6 were up-regulated and 15 down-regulated (

Table 3. APxs, Kats, CIII Prxs, DiOxs, RBOHs and SODs differentially expressed upon salt stress in E. salsugineum and A. thaliana. The gene nomenclature is that of the online database Redoxibase (https://peroxibase.toulouse.inrae.fr/; accessed on 30 November 2023). Differential expression data are presented as log₂(FC). ‘No transcript’ indicates that no transcripts were detected, and ‘FDR > 0.05’ that the calculated FDR is greater than 0.05 (not statistically significant). (+) and (−) in the second [34] and third [36] columns indicate that the proteins were found to be up- or down-accumulated upon salt stress, respectively.

E. salsugineum Redoxibase ID	E. salsugineum log2(FC)	A. thaliana log2(FC)	A. thaliana Redoxibase ID
TsPrx02/Thhalv10008157	−2.88	FDR > 0.05	AtPrx02/At1g05250
TsPrx03/Thhalv10008189	−0.69	FDR > 0.05	AtPrx03/At1g05260
TsPrx11/Thhalv10018830	0.97	FDR > 0.05	AtPrx11/At1g68850
TsPrx12/Thhalv10018753	−2.07	FDR > 0.05	AtPrx12/At1g71695
TsPrx25/Thhalv10016924	−0.89	FDR > 0.05	AtPrx25/At2g41480
TsPrx27/Thhalv10021152	−1.96	FDR > 0.05	AtPrx27/At3g01190
TsPrx28/Thhalv10021155	−2.02	no FDR	AtPrx28/At3g03670
TsPrx30/Thhalv10021123	−0.85 (+)	FDR > 0.05	AtPrx30/At3g21770
TsPrx39/Thhalv10028797	−2.80	FDR > 0.05	AtPrx39/At4g11290
TsPrx45/Thhalv10025699	−1.65	FDR > 0.05	AtPrx45/At4g30170
TsPrx56/Thhalv10014085	−1.44	FDR > 0.05	AtPrx56/At5g15180
TsPrx58-1/Thhalv10014083	−2.15 (−)	FDR > 0.05	AtPrx58/At5g19880
TsPrx59/Thhalv10014080	−2.84	FDR > 0.05	AtPrx59/At5g19890
TsPrx66/Thhalv10014117	2.43	FDR > 0.05	AtPrx66/At5g51890
TsPrx69/Thhalv10004557	−0.62	FDR > 0.05	AtPrx69/At5g64100
TsAPx-R/Thhalv10025686m	0.96	FDR > 0.05	AtAPx-R/At4g32320
TsAPx02/Thhalv10021381m	1.87 (+)	FDR > 0.05	AtAPx02/At3g09640
TsDiOx01/Thhalv10020279	−3.33	FDR > 0.05	AtDiOx01/At3g01420
TsDiOx02/Thhalv10018266	0.97	FDR > 0.05	AtDiOx02/At1g73680
TsFSD01-1A/Thhalv10026218	1.11	FDR > 0.05	AtFSD01/At4g25100
TsMSD01-1A/Thhalv10021449	0.61	FDR > 0.05 (+)	AtMSD01/At3g10920
no sequence		−4.04	AtPrx08/At1g34510
TsPrx21/Thhalv10016921	FDR > 0.05 (−)	−1.74	AtPrx21/At2g37130
TsPrx34/Thhalv10010494	FDR > 0.05	−1.69 (+)	AtPrx34/At3g49120
TsPrx35/Thhalv10010538	FDR > 0.05	−3.61	AtPrx35/At3g49960
TsPrx42/Thhalv10025680	FDR > 0.05	−1.29	AtPrx42/At4g21960
TsPrx50/Thhalv10025717	no transcript	1.61	AtPrx50/At4g37520
TsPrx54/Thhalv10013916	no transcript	−1.97	AtPrx54/At5g06730
TsPrx60/Thhalv10015253	no transcript	−5.01	AtPrx60/At5g22410
TsRboh03/Thhalv10012613	FDR > 0.05	−1.46	AtRboh03/At5g51060
TsRboh04/Thhalv10003619	FDR > 0.05	−1.40	AtRboh04/At5g47910
TsKat02/Thhalv10025015	no transcript (+)	1.32	AtKat02/At4g35090
TsCSD01/Thhalv10008927	FDR > 0.05 (+)	1.80	AtCSD01/At1g08830
TsCSD02/Thhalv10017094	FDR > 0.05	1.77	AtCSD02/At2g28190
TsCCS01/Thhalv10008232	FDR > 0.05	1.92	AtCCS01/At1g12520

). Among them, there are 15 CIII Prxs, 2 APxs, 2 DiOxs and 2 SODs. On the other hand, 14 genes were specifically regulated in A. thaliana, 5 were up-regulated and 9 were down-regulated (Table 3). Among them, there are nine CIII Prxs, two RBOHs, one Kat and three SODs. Some of the encoded proteins were found to be up- or down-accumulated by quantitative proteomics in E. salsugineum 4-week-old plants subjected to a 24 h watering with a 300 mM NaCl solution [34]. Some of these proteomics data show discrepancies with transcriptomics data, e.g., TsAPx02, TsPrx21, TsKat02 and TsCSD01 in Table 3 (see also [33,34]). This kind of difference has been described many times (for a review, see [35]).

AtPrx01, AtPrx44 and AtPrx73 are repressed by salt stress in the roots of both species (Table S3). These genes are mainly expressed in roots and control root hair length in A. thaliana [37]. AtPrx34, which is required for root elongation [38], is repressed in A. thaliana, but not differentially expressed in E. salsugineum. The observed modifications of CIII Prx expression are thus with a reduction of root growth in A. thaliana upon salt stress. Indeed, the salt stress response is a complex one in roots with a first phase of growth inhibition, followed by a second one of growth recovery [39]. In E. salsugineum, it was shown that salt stress results in an increased oxidation level of 159 Cys residues within 107 different proteins including proteins of the ROS network (TsPrx30, 49 and 71; TsKat02; TsCSD01). Interestingly, the CIII Prx homologous to TsPrx03, 07 and 56 are down-regulated in Mesembryanthemum crystallinum, a plant highly tolerant to salinity, after exposure to salt stress like in E. salsugineum, suggesting a role for these genes in adaptation to salinity [40].

In addition, none of the genes duplicated in E. salsugineum (TsPrx22, 33 and 38; APx03; Rboh07) or of the A. thaliana genes corresponding to genes lost in E. salsugineum (AtPrx22, 33 and 38) have a modified level of expression upon salt stress, suggesting that they play no critical role for the adaptation to salt stress (Table S2).

2.4. Differentially Expressed OR Genes upon Hypoxia

Another pair of plants has been selected to study the differential expression of the OR genes of interest upon hypoxia: A. thaliana, which is sensitive to hypoxia, and N. officinale, which is tolerant. The A. thaliana data originate from a comprehensive meta-study constituted by 29 RNA-seq datasets selected through a keyword search [41]. Instead of providing FC values, these datasets offer a HN-score, reflecting the trend of a gene being up- or down-regulated across all the experiments included in the study. These scores enabled us to compare the way in which a given gene is regulated, although they do not provide information on the extent of differential expression between control and treated plants. For N. officinale, the hypoxia stress was induced by submerging plants at the five or six-leaf stage in water for 24 h and the RNA-seq experiments were performed on the stems [42].

Upon hypoxia, 84 genes for which a significant FC value or a HN-score different from zero were found, as well as 28 genes of N. officinale for which no transcripts were detected (Table S3). Altogether, 36 genes were regulated in A. thaliana and N. officinale upon hypoxia, among which 10 and 12 were down- or up-regulated in both plants, respectively (Figure 2). Eleven genes were down-regulated in A. thaliana, but up-regulated in N. officinale. In contrast, three genes were up-regulated in A. thaliana but down-regulated in N. officinale (APx02, Rboh04 and FSD01) (

Table 4. OR genes regulated in contrasted manner between A. thaliana and N. officinale upon hypoxia. The gene nomenclature is that of the online database Redoxibase (https://peroxibase.toulouse.inrae.fr/; accessed on 30 November 2023). Differential expression data are presented as log₂(FC).

N. officinale Redoxibase ID	N. officinale log2(FC)	A. thaliana HN-score	A. thaliana Redoxibase ID
NoffPrx09-1B	5.56	−3	AtPrx09/At1g44970
NoffPrx16-1A	3.18	−5	AtPrx16/At2g18980
NoffPrx22-1A	6.16	−1	AtPrx22/At2g38380
NoffPrx23-1A	6.36	−2	AtPrx23/At2g38390
NoffPrx27-1A	3.83	−2	AtPrx27/At3g01190
NoffPrx45-1A	2.87	−4	AtPrx45/At4g30170
NoffPrx56-1A	4.27	−2	AtPrx56/At5g15180
NoffPrx64-1A	1.92	−2	AtPrx64/At5g42180
NoffPrx67-1A	4.59	−2	AtPrx67/At5g58390
NoffAPx02-1A	−3.56	1	AtAPx02/At3g09640
NoffRboh04-1A	−0.40	26	AtRboh04/At5g47910
NoffCCS01-1A	1.97	−3	AtCCS01/At1g12520
NoffCSD03-1A	0.74	−1	AtCSD03/At5g18100
NoffFSD01-1A	−5.38	3	AtFSD01/At4g25100

). In addition, 34 genes were down-regulated and 12 up-regulated in A. thaliana upon hypoxia, whereas only 2 genes were specifically up-regulated in N. officinale.

As in the case of the salt stress, it was possible to identify genes specifically regulated in A. thaliana upon hypoxia (

Table 5. APxs, Kats, CIII Prxs, DiOxs, RBOHs and SODs differentially expressed upon hypoxia stress in N. officinale and A. thaliana. The gene nomenclature is that of the online database Redoxibase (https://peroxibase.toulouse.inrae.fr/; accessed on 30 November 2023). Differential expression data are presented as log₂(FC). Differential expression data are presented in log₂(FC) for N. officinale and HN-score for A. thaliana [42]. ‘No transcript’ indicates that no transcripts were detected, ‘FDR > 0.05’ that the calculated FDR was greater than 0.05 (not statistically significant), and [P] that no transcripts were detected because the gene was annotated as a pseudogene.

N. officinale Redoxibase ID	N. officinale log2(FC)	A. thaliana HN-score	A. thaliana Redoxibase ID
NoffPrx38-1A	2.42	0	AtPrx38/At4g08780
NoffRboh05	1.38	0	AtRboh05/At1g19230
NoffPrx02-1A	no transcript	−4	AtPrx02/At1g05250
NoffPrx04-1B	FDR > 0.05	15	AtPrx04/At1g14540
NoffPrx05-1A	no transcript	6	AtPrx05/At1g14550
NoffPrx07-1A	no transcript	−4	AtPrx07/At1g30870
no sequence		−13	AtPrx08/At1g34510
NoffPrx11-1A	no transcript	−2	AtPrx11/At1g68850
NoffPrx12-1A	FDR > 0.05	−9	AtPrx12/At1g71695
NoffPrx15-1A	no transcript	−3	AtPrx15/At2g18150
NoffPrx17-1A	FDR > 0.05	−1	AtPrx17/At2g22420
NoffPrx21-1A	FDR > 0.05	2	AtPrx21/At2g37130
no sequence		−5	AtPrx24/At2g39040
NoffPrx25-1A	no transcript	−11	AtPrx25/At2g41480
NoffPrx28-1A	no transcript	5	AtPrx28/At3g03670
NoffPrx30-1A	FDR > 0.05	−3	AtPrx30/At3g21770
NoffPrx32-1A	FDR > 0.05	−2	AtPrx32/At3g32980
NoffPrx33-1A	FDR > 0.05	−1	AtPrx33/At3g49110
NoffPrx[P]34-1A	[P]	5	AtPrx34/At3g49120
NoffPrx35-1B	no transcript	−6	AtPrx35/At3g49960
NoffPrx39-1A	FDR > 0.05	−2	AtPrx39/At4g11290
NoffPrx40-1A	FDR > 0.05	−1	AtPrx40/At4g16270
NoffPrx44-1A	no transcript	−3	AtPrx44/At4g26010
NoffPrx46-1B	FDR > 0.05	10	AtPrx46/At4g31760
NoffPrx53-1A	no transcript	−3	AtPrx53/At5g06720
NoffPrx57-1A	no transcript	−2	AtPrx57/At5g17820
NoffPrx58-1A	FDR > 0.05	2	AtPrx58/At5g19880
NoffPrx59-1A	no transcript	6	AtPrx59/At5g19890
NoffPrx60-1A	no transcript	−10	AtPrx60/At5g22410
NoffPrx61-1A	no transcript	6	AtPrx61/At5g24070
no sequence		−11	AtPrx63/At5g40150
NoffPrx65-1A	no transcript	−1	AtPrx65/At5g47000
NoffPrx72-1A	FDR > 0.05	−2	AtPrx72/At5g66390
NoffPrx73-1A	FDR > 0.05	−4	AtPrx73/At5g67400
NoffAPx-R01-1A	FDR > 0.05	−2	AtAPx-R/At4g32320
NoffAPx03-1A	FDR > 0.05	−1	AtAPx03/At4g35000
NoffAPx06-1A	FDR > 0.05	−5	AtAPx06/At1g77490
NoffDiOx01-1A	FDR > 0.05	−1	AtDiOx01/At3g01420
NoffRboh03-1A	no transcript	3	AtRboh03/At5g51060
NoffRboh06-1A	FDR > 0.05	1	AtRboh06/At1g64060
NoffRboh07-1B	FDR > 0.05	−1	AtRboh07/At4g25090
NoffRboh09-1B	no transcript	3	AtRboh09/At4g11230
NoffKat01-1B	FDR > 0.05	−2	AtKat01/At1g20630
NoffKat02-1A	FDR > 0.05	−4	AtKat02/At4g35090
NoffCSD02-1A	FDR > 0.05	−1	AtCSD02/At2g28190
NoffFSD02-1A	FDR > 0.05	−2	AtFSD02/At5g51100
NoffFSD03-1A	FDR > 0.05	−4	AtFSD03/At5g23310
NoffMSD01-1A	FDR > 0.05	−1	AtMSD01/At3g10920

). Thirty-five genes were down-regulated, among which included 23 CIII Prxs, 3 APxs, 1 DiOx, 1 Rboh, 2 Kats and 4 SODs. Fifteen genes were up-regulated among which nine were Prxs and three were Rbohs. Conversely, only two genes were specifically up-regulated in N. officinale: Prx38 and Rboh05.

AtPrx07, AtPrx44 and AtPrx73 were found to be repressed in A. thaliana and not expressed in N. officinale (Table 5). AtPrx44 and AtPrx73 control root hair length in A. thaliana as shown by the phenotyping of mutants impaired in these genes and of over-expressors [37]. Yet, hypoxia inhibits root apical meristem (RAM) activity but increases root hair density and length [43].

AtRboh04/AtRBOHD was found to be up-regulated in many studies devoted to the analysis of the response of A. thaliana to hypoxia (HN-score of 26 out of the 29 analyzed sets of data), whereas it is weakly repressed in N. officinale (Table S3). This is consistent with the fact that the AtRboh04-mediated ROS burst induces genes required for hypoxia acclimation [44,45,46]. AtRboh04 expression is detected in all organs, and AtRBOH04 could be the source of H₂O₂ in both roots and shoots during hypoxia.

As previously shown [47], AtPrx04, AtPrx05 and AtPrx28 are induced in A. thaliana upon hypoxia, but they are not differentially expressed in N. officinale (Table 5). As for E. salsugineum upon salt stress, none of the three duplicated genes in N. officinale is induced upon hypoxia (NoffPrx17-2B, 44-2A and 44-2B), meaning that their role in the adaptation to hypoxia is not critical. In the same way, the loss of several gene copies does not prevent N. officinale from being adapted to hypoxia (NoffPrx04-1A, 05-1B, 30-1B, 32-1B, 46-1A, 58-1B, and 61-1B; NoffDiOx01-1B; NoffRboh07-1A). Their orthologs in A. thaliana undergo changes in their regulation being either up- or down-regulated, thus suggesting a role in the response to hypoxia.

The perception of both saline stress and hypoxia first occurs at the root level when plants are moderately overflooded, but developmental effects, such as growth inhibition, can be also observed in the aerial part of the plants. Changes in gene expression in response to these stresses can therefore be rapidly detected after perception in the root, but also in the shoot following signal transduction [36].

About one third of the A. thaliana OR genes differentially regulated upon hypoxia or salinity were actually responding to both stresses, among which were 2 APxs, 1 Kat, 17 Prxs, 4 SODs, and 2 Rbohs. AtPrx10 was identified as one of those genes and was previously shown to be involved in adventitious root formation, its mutant being impaired in root regeneration from leaf explants [48]. These genes could be part of a more general answer to abiotic stresses. None of them correspond to genes lost or gained in either genome. In addition, AtPrx62 and its N. officinale ortholog were both induced during hypoxia (Table S3). Now, AtPrx62 promotes root hair growth at low temperatures [49], which suggests that AtPrx62 could be associated with root growth regulation in response to different stresses and induced even in tolerant species.

The number of genes belonging to the six OR gene families expressed in the adapted species or in the sensitive one and differentially regulated upon salt stress is lower than upon hypoxia. This is probably because many more experiments have been included in the latter case. However, for each stress, it was possible to identify genes specifically regulated either in A. thaliana or in the tolerant species and genes regulated in both cases.

3. Materials and Methods

3.1. Growth of N. officinale

N. officinale seeds (NAST-OFFI, Semences du Puy, Le Puy-en-Velay, France) were sown directly on water-saturated soil and grown with 50% humidity, at 21 °C upon a 13 h light/11 h dark cycle. The leaves were collected after 17 d, frozen in liquid nitrogen and stored at −80 °C, prior to DNA extraction.

3.2. HMW DNA Extraction

DNA was isolated from young leaves using a Genomic-tips 500/G kit (Cat No./ID: 10262, QIAGEN, Courtabœuf, France) following the tissue protocol extraction. Briefly, 2 g of young leaf material was ground in liquid nitrogen with a mortar and pestle. After 3 h of lysis at 50 °C and one centrifugation step, DNA was immobilized on the provided column. After several washing steps, DNA was eluted from the column, then desalted and concentrated by alcohol precipitation. The DNA was resuspended in the EB buffer. The DNA amount was assessed using a NanoDrop One spectrophotometer (Thermo Fisher Scientific, Illkirch-Graffenstaden, France) and a Qbit 3 Fluorometer using the Qbit dsDNA BR assay (Invitrogen, Illkirch-Graffenstaden, France). The size of the DNA fragments was determined using the FemtoPulse system (Agilent, Santa Clara, CA, USA) (Figure S1).

3.3. N. officinale Genome Sequencing: HiFi PacBio Library Preparation

According to the manual ‘Procedure & Checklist—Preparing HiFi SMRTbell Libraries using the SMRTbell Express Template Prep Kit 2.0′ (PN 101-853-100, Pacific Biosciences of California, San Diego, CA, USA), the HiFi libraries were prepared with 10 µg DNA sheared by using the Megaruptor 1 system (Diagenode, San Diego, CA, USA) to obtain a 15–20 kb average size. Size-selected libraries were sequenced on the PacBio Sequel II system in CCS mode for 30 h (Pacific Biosciences of California). To assess the quality and the completeness of the genomes, we performed BUSCO (version 5.4.4) analysis with the brassicales_odb10.2020-08-05 lineage as a reference [50].

3.4. Genome Annotation

We have annotated the newly sequenced N. officinale genome and the one deposited at NCBI (http://www.ncbi.nlm.nih.gov/datasets/genome/GCA_900406445.1/, accessed on 7 February 2023, [21]) with Braker3 [51]. We have also performed a functional annotation with InterProScan [52]. The v1.0 version of the genomic sequence of E. salsugineum (Esalsugineum_173_v1.fa) and the corresponding annotation file (Thalophila_173_gene.gff) were downloaded from Phytozome (http://phytozome-next.jgi.doe.gov/info/Esalsugineum_v1_0, accessed on 11 April 2023) [53]. From the perspective of more exhaustive evolutionary analysis, the prediction of pseudogene sequences was performed with P-GRe [54] in order to detect potential gene losses (pseudogenization).

Codon-based alignments of the coding regions were performed with MACSE v2 [55] for each orthogroup in each OR family, with default parameters, except the gap end penalty set to 8. Ka, Ks and subsequent Ka/Ks ratios were calculated from these alignments for each pair of sequences using the method of model averaging (MA) implemented in the KaKs_calculator 3.0 [56].

3.5. Phylogeny

Six gene families belonging to this ROS gene network have been extracted from the annotated genome using their PFAM ID. These families have been selected based on the two following criteria: the number of gene copies (possibly subjected to duplications), and the level of sequence conservation (possibly subjected to selective pressure). To precisely define the presence or absence of sequences in comparison with A. thaliana, the orthology and ohnology relationships were analyzed. All the protein sequences annotated and used in this study have been made available in the Redoxibase (https://peroxibase.toulouse.inrae.fr, accessed on 30 November 2023) [57]). Multiple alignments of the sequences were made with ClustalW [58]. Alignment curation and tree construction (PhyML) were made using NGPhylogeny.fr (https://ngphylogeny.fr/, accessed on 2 October 2023) [59]. In parallel, the orthogroups were inferred via the OrthoFinder software [60].

3.6. RNA-Seq Data

Three sets of RNA-seq data and one meta-analysis of RNA-seq have been used for this study. Plants were submitted to (i) salt stress for A. thaliana and E. salsugineum [32,33] or (ii) hypoxia stress for A. thaliana and N. officinale [41,42]. The details of the experimental setups are recalled as follows.

For salt stress, expression data for A. thaliana (Col-0 ecotype) [32] were obtained from roots grown in vitro for 10 d in half-strength Murashige and Skoog basal medium (1/2 MS) with and without 50 mM NaCl. For E. salsugineum (Shandong ecotype), expression data were collected from roots of plants germinated on 1/2 MS, transferred to soil after 7 d for a 5-week-growth period, and subjected to either 300 mM NaCl solution (salt treated) or water (control) irrigation for 24 h before sampling. These two NaCl concentrations have been described as stressful [61] but not toxic for the two species [32,34].

In the case of hypoxia, metadata for A. thaliana were sourced from 29 RNA-seq datasets identified through a keyword search in the Gene Expression Omnibus (GEO) [62]. Most samples consisted of whole seedlings or aerial organs. For N. officinale, expression data were obtained from plant stems that were initially germinated in sterile soil, grown until they reached the 5–6 leaf stage, and then sampled after 24 h of submersion to induce hypoxia.

3.7. Pipeline for the OR Families Expression Analysis

In hypoxia conditions for A. thaliana (meta-analysis), differential expressions have been calculated in the form of an HN-score (hypoxia–normoxia score), as computed by the authors [41]. So, no associated statistical values were provided. For the other differential analyses, FCs with a Benjamini–Hochberg adjusted p-value (FDR or false discovery rate) have been employed. Differential expressions utilized for N. officinale [42] and A. thaliana [32] under saline conditions were directly derived from the published data.

Regarding E. salsugineum, RNAseq libraries (SRR6837742, SRR6837743, SRR14804236, SRR14804235, available at NCBI) were used for conducting a differential analysis using the nf-core/rnaseq pipeline version 3.10.1 in Nextflow v22.12.0 [63] with default parameters [64,65]. Next, the counts were normalized using the Trimmed Mean of M-values (TMM) method from the EdgeR package [66,67]. Subsequently, a selection of expressed genes was performed, using a threshold of expression higher than 0.1 cpm (count per million) across all replicates. Following a second TMM normalization on the expressed genes, an analysis was conducted using the EdgeR package. The expression of the log₂FC threshold was set to 0 to distinguish the expressed genes (even if not differentially expressed) from genes with undetectable levels of expression.

To select the transcripts corresponding to the six investigated OR gene families, a BLAST analysis was carried out with the gene sequences annotated as described above.

4. Conclusions

This work has allowed for the expert annotation of 110 OR genes belonging to six different families in the diploid E. salsugineum and 188 in the tetraploid N. officinale. They were compared to the 108 already-annotated OR genes of A. thaliana. We could show that most of the orthologous sequences are conserved between the three species, which probably diverged 30 MYA, thus supporting the idea that OR families have critical functions for ROS homeostasis. As for other plants, polyploidy could have been an asset for the adaptation and evolution of N. officinale [68]. Also, we have observed that less than 30% of the ohnologs of N. officinale have been lost or pseudogenized, suggesting that the conservation of the two copies brought sufficient advantages (such as sub-functionalization or dose effect) to bypass the loss of duplicated genes [69].

Global gene expression analyses showed genes similarly regulated in the species subjected to comparable stresses. However, we also observed genes only regulated by stress in A. thaliana, which is sensitive to both stresses. Conversely, there were genes only regulated in the halophytic E. salsugineum upon salt stress or in the hypoxia-tolerant N. officinale upon hypoxia. In addition, among the six OR classes studied, CIII Prx genes represent two thirds of the genes analyzed and have been largely described as stress markers [12]. Their roles seem to be critical, especially upon hypoxia. It is also important to recall that the six OR classes analyzed in this work are members of the so-called ROS gene network, which contains dozens of other OR classes participating in ROS homeostasis during regular metabolism but also following stressing events. The OR genes highlighted by this work could be targets for functional analyses to better understand their roles either in the acclimation to hypoxia or salinity or in the stress responses.