(A) Genomic structure of SRO5 (At5g62520) and P5CDH (At5g62530) genes. Arrows indicate the direction of transcription. Thick and thin solid lines represent ORF and UTR regions, respectively. Sequence of the 24-nt SRO5-P5CDH nat-siRNA is aligned with P5CDH mRNA. The three nucleotides marked by a line on top were deduced from RNA blot analysis and confirmed by primer extension assays (Table S1).
(B) Accumulation of the SRO5-P5CDH nat-siRNA is specifically induced by salt stress. (Left panel) Twenty-day-old soil grown plants were exposed to the following stress or hormone treatments: dehydration (withholding watering for 10 days); ABA (spraying with 100 μM ABA with samples collected 24 hr later); and NaCl (watering with 300 mM NaCl for 2 days). (Right panel) Ten-day-old seedlings exposed to the following treatments in liquid media for 6 hr: Control (nutrient medium), 150 mM NaCl, 15 mM LiCl, 30% (w/v) PEG-8000, 250 mM Mannitol, 100 μM ABA, or 10 μM ACC.
(C) Northern blot of small RNA extracted from control or salt-treated (150 mM, 6 hr) seedlings. The sense and antisense probes are described in the text.

(A and B) Northern analysis of small RNAs from various mutants and wild-type controls. One hundred micrograms of low-molecular-weight RNA was loaded per lane. The blots were hybridized with an oligonucleotide probe corresponding to the sequence of the 24-nt nat-siRNA. The number below the blot indicates relative abundance (RA) of the nat-siRNA with respect to the loading control miR172 or siRNA 1003 when miR172 is not present.
(C and D) Northern blot analysis of P5CDH and SRO5 mRNAs. Twenty-five micrograms of total RNA was probed with P5CDH 5′- or 3′-specific probes or SRO5 probe. The P5CDH 5′ probe includes the 50 bp 5′UTR and the first 400 bp of ORF. The P5CDH 3′ probe corresponds to the last 200 bp of the 3′UTR and detects a cleavage fragment of approximately 700 bp. The P5CDH 3′ probe also hybridized with the ~1.2 kb SRO5 mRNA (not shown). The SRO5 probe detects a transcript of approximately 1.2 kb.

(A) P5CDH mRNA cleavage sites detected by RLM-5′ RACE. Numbers indicate the frequency of cleavage at each site. The positions of predicted 21-nt nat-siRNAs are indicated.
(B) Diagram showing the sequences of 24-nt and 21-nt nat-siRNAs aligned with the P5CDH mRNA (upper panel) and detection of the 21-nt nat-siRNAs (lower panel). Sequences in italics are 21-nt nat-siRNA predicted but not confirmed by Northern blots. One hundred micrograms of low-molecular-weight RNA per lane was probed with oligo probes, 1, 2, and 3, complementary to 21-nt nat-siRNAs numbered. The number below the blot indicates relative abundance (RA) of the nat-siRNA with respect to the loading control miR172.
(C and D) Accumulation of 21-nt nat-siRNA number 2 in different mutants. The number below the blot indicates relative abundance (RA) of the nat-siRNA with respect to the loading control miR172 or siRNA 1003.
(E) Alignment of the 24-nt nat-siRNA region in wild-type and mutated (mSRO5 andmP5CDH) versions of SRO5 and P5CDH used for transient expression in N. benthamiana. Two mutations were introduced on either side of the site of 24-nt nat-siRNA-directed P5CDH cleavage.
(F) Coexpression of various combinations of wild-type and mutant SRO5 and P5CDH constructs in N. benthamiana. As a control, SRO5 and P5CDH were also coexpressed with an unrelated GFP transcript. Small RNA blots were prepared from leaves collected 3 days after infiltration and probed with oligoes complementary to wild-type 24-nt SRO5-P5CDH nat-siRNA, the mutated 24-nt SRO5-P5CDH nat-siRNA sequence (m24-nt nat-siRNA), and a mixture of three 21-nt P5CDH nat-siRNAs (probes 1, 2, and 3; Figure 3B).

(A) Proline content in 10-day-old seedlings with or without 6 hr of exposure to 150 mM NaCl. Values represent means ± SD from three independent experiments.
(B) Response of sgs3 and nrpd1a to NaCl and exogenous proline. Seedlings were germinated and grown for 7 days in 75 mM NaCl or 5 mM proline for Col-O and sgs3, and 75 mM NaCl or 2 mM proline for C24 and nrpd1a. Photos show representative seedlings.
(C) Tolerance to NaCl or exogenous proline quantified as the percentage of 7-day-old seedlings with green expanded cotyledons. Values represent means ± SD from three independent experiments.

(A) Diagram of T-DNA insertion mutants isolated.
(B) Northern blot of P5CDH and SRO5 mRNAs in wild-type, p5cdh, and sro5 T-DNA lines. RNA (20 μg per lane) from 10-day-old seedling with or without 6 hr exposure to 150 mM NaCl were probed with P5CDH 5′ probe or SRO5 probe.
(C) Northern blot of small RNAs. Eighty micrograms of low-molecular-weight RNA extracted from the same samples as in (B) was blotted and probed with an oligoprobe complementary to the 24-nt SRO5-P5CDH nat-siRNA. The number below the blot indicates relative abundance (RA) of the nat-siRNA with respect to the loading control siRNA 1003.
(D) Proline accumulation in wild-type, p5cdh, and sro5 T-DNA lines after 6 hr exposure to 150 mM NaCl. Data are means ± SD (n = 10).
(E) Phenotypes of seedlings grown on 75 mM NaCl or 5 mM proline for 7 days.
(F) Proline response quantified as the percentage of seedlings with green expanded cotyledons. Data are means ± SD (n = 10).
(G) NaCl response quantified as seedling fresh weight relative to control not treated with NaCl. Data are means ± SD (n = 10).

(A) Hypersensitivity of sro5 mutant to oxidative stress induced by 100 μM exogenous H₂O₂.
(B) ROS staining in seedling roots after exposure to 100 mM NaCl for 12 hr.
(C) H₂O₂ accumulation quantified in seedlings exposed to 100 mM NaCl for 12 hr. Data are means ± SD (n = 10).
(D) Regulation of P5CDH and SRO5 mRNAs by oxidative stress. Seedlings were treated or not treated (control) with 6 hr exposure to 10 mM H₂O₂.
(E) Accumulation of 24-nt SRO5-P5CDH nat-siRNA is induced by H₂O₂. Small RNA (100 μg per lane) from the same tissues used in (D) were probed with an oligonucleotide complementary to the 24-nt SRO5-P5CDH nat-siRNA. The number below the blot indicates relative abundance (RA) of the nat-siRNA with respect to the loading control miR172.

Salt stress induces ROS accumulation which in turn induces the expression of SRO5. This initiates a series of nat-siRNA processing steps (shown in the shaded area of the model) that result in the downregulation of P5CDH. When SRO5 is induced, dsRNA can form by annealing of the SRO5 and P5CDH transcripts (note that poly-A tails of the mRNAs are omitted for clarity), and this initiates DCL2-dependent formation of 24-nt nat-siRNA. This initial cleavage of the P5CDH mRNA causes phased generation of 21-nt nat-siRNAs by a DCL1-dependent mechanism and additional cleavage of the P5CDH transcript. RDR6, SGS3, and NRPD1A are suggested to be involved in the formation of both 24- and 21-nt SRO5-P5CDH nat-siRNAs. The importance of P5CDH downregulation by nat-siRNAs is further illustrated by the functional relationship between the SRO5 and P5CDH proteins (portion of model outside the shaded area). Degradation of P5CDH mRNA decreases P5CDH activity and proline degradation. Decreased proline degradation contributes to salt-induced proline accumulation and salt tolerance but, at the same time, leads to the ROS accumulation caused by buildup of the proline catabolic intermediate P5C. This ROS accumulation is potentially harmful but is also a signal to initiate further stress responses. The SRO5 protein counteracts P5C-induced ROS accumulation, thus completing a regulatory loop that allows fine tuning of ROS accumulation and stress response.