H. caspica seeds were harvested from extremely saline-alkali and arid areas in the Gurbantunggut Desert in Xinjiang, China. Healthy seeds were sown in pots with the substrate (perlite: vermiculite: flower soil = 1:1:3) and cultivated under natural light and suitable temperature (25°C - 28°C). Eight-week-old H. caspica plants were exposed to 600 mM NaCl, 1000 mM mannitol, 0 °C freezing stress, 100 μM methyl viologen (MV) and 300 μM abscisic acid (ABA) for 0, 3 and 24 h, and assimilating branches were taken and put in liquid nitrogen for subsequent qRT-PCR assays.

The HcmiR169b mature sequence was obtained from a small RNA library derived from H. caspica roots (Yang et al., 2015). The HcmiR169b precursor sequence was obtained through homologous cloning and 5’-RACE nested amplification using the H. caspica cDNA as template. Based on the EST sequence of HcNFYA1 in the H. caspica transcriptome data, the full length of HcNFYA1 gene was cloned from the H.caspica cDNA using the SMARTer RACE 5’/3’ Kit (Takara, Japan).

Amino acid sequences of NFYA family members were downloaded from the PlantTFDB plant transcription factor database (http://planttfdb.gao-lab.org/index.php) (Jin et al., 2017). Multiple comparisons of amino acid sequences were performed using DNAMAN software (LynnonBiosoft, USA). Phylogenetic analysis was applied using the proximity method (1000 replicates) with MEGA 11 software (Mega Limited, New Zealand).

As shown in the Supplementary Figure S1 , the method in this study was modified according to the RNA ligase mediated (RLM)-cDNA end rapid amplification (RACE) technique developed by (Wang and Fang, 2015). T4 RNA ligase (Ambion, Canada) was used to add an adapter to the 3’ hydroxyl end of H. caspica RNA. A specific primer (GSP) complementary to the adapter was used for reverse transcription to obtain cDNA with the adapter sequence. Two rounds of nested PCR amplification was carried based on the cDNA and the second round of amplification products were cloned into the pMD-19T vector for sequencing, and the sequencing results were analyzed to determine the cleavage site of HcNFYA1 by HcmiR169b. The primers used in this study were listed in Supplementary Table S1 .

Total RNAs were extracted from plant samples by RNA prep pure Plant Kit (Tiangen, China). The stem-loop method was used for microRNA reverse transcription, with HcU6 and AtU6 serving as internal controls (Chen et al., 2005), and protein-coding genes were reverse-transcribed using One-Step gDNA Removal and cDNA Synthesis SuperMix (Transgen, China), with HcUBQ10 and Atactin as internal controls (Zhang et al., 2015). qRT-PCR was performed in a CFX96 Touch Real-Time PCR System (Bio-Rad, USA) using a PerfectStart Green qPCR SuperMix (Transgen, China) with three biological replicates per sample. The relative expression of genes was calculated by 2^−ΔΔCT comparison method (Livak and Schmittgen, 2001). The primers were shown in Supplementary Table S1 .

The HcNFYA1 open reading frame (ORF) region was inserted into the plant expression vector pCAMBIA1301-1-GFP (Liu et al., 2009). HcNFYA1 was fused to GFP for expression and transformed into onion epidermal cells using the Agrobacterium GV3101-mediated transient transformation system. Onion epidermal cells were stained with DAPI and observed under a LSM800 Confocal Microscope (Zeiss, Germany).

The ORF region of HcNFYA1 was constructed into the pGBKT7 vector containing the GAL4 DNA binding domain. The recombinant plasmid was transformed into the Y2H Gold yeast strain. Transformants were screened in SD/-Trp-His medium containing X-α-Gal and HcTOE3 was used as a positive control (Yin et al., 2021). The primers were listed in Supplementary Table S1 .

The precursor sequence of HcmiR169b and the ORF region of HcNFYA1 were constructed into plant expression vector pCAMBIA2300, and transformed into Arabidopsis wild type (Columbia) by inflorescence infection (Zhang et al., 2006). Positive lines were screened by kanamycin primarily. The selected T₃ generation transgene Arabidopsis lines with single copy were verified by genomic PCR and qRT-PCR.

Arabidopsis seeds were sterilized with sodium hypochlorite (10%) and 75% ethanol (90%) for 5 min and sown in 1/2 MS medium. After 3 days of vernalization at 4°C with a 16 h/8 h light/dark photoperiod, petri dishes with plants were put in the growing chamber at 22°C. After two weeks of growth, the plants were transferred to pots filled with the substrate (perlite: vermiculite: flower soil = 1:1:3) for cultivation.

Arabidopsis germination experiments were conducted in 1/2 MS medium containing NaCl (125 mM), mannitol (250 mM), and ABA (0.5 μM, 0.75 μM). The seed germination rates were recorded daily, and the cotyledon greening rates were measured after 7 days.

For the salt treatment, four-week-old Arabidopsis plants were irrigated with 300 mM NaCl for 7 days, photographed, dried at 80°C for 24 hours, and weighed for the dry weight (10 biological replicates). 300 mM NaCl-treated Arabidopsis leaves were collected for qRT-PCR (24 h) and measuring some physiological and biochemical parameters (3 d).

For the drought treatment, Arabidopsis plants were planted in pots containing the same substrate weight and re-watered for 3 days after stopping irrigation for 7 or 9 days. The plants were photographed and counted for survival numbers (three independent experiments with 40 biological replicates each). Arabidopsis leaves were collected when irrigation was stopped for 5 days for physiological index testing. Leaves were taken for qRT-PCR analysis after 24 h of treatment when Arabidopsis was irrigated with 20% PEG6000.

Adult Arabidopsis leaves were submerged with 1 mg/mL of Evans blue, DAB, and NBT solution and stained for 2 h at 37°C in the dark. The leaves were placed in absolute alcohol at boiling temperature for 30 min to remove chlorophyll. 10 blades were run per treatment.

Whole Arabidopsis plants were soaked in 50 ml ultrapure water for 24 h. The solution conductivity (C1) was measured and the solution conductivity (C2) was measured again after boiling for 1 h using a Conductivity Meter (DSS-307, China). C1/C2×100% was calculated as the relative electrolyte leakage.

The measurements of chlorophyll, MDA, H₂O₂, O 2 − and proline contents as well as the detection of POD, APX and SOD enzyme activities were performed according to the manufacturer’s instructions (Solarbio, China). Soluble protein content was determined using the BCA Soluble Protein Content Kit (Addison, China). The content of ABA was determined by ABA ELISA assay kit (Saipei, China). Per sample had three biological replicates for each treatment, and 0.1 g of Arabidopsis leaves were collected for each biological replicate.

Arabidopsis leaves were taken and placed in an area with good airflow. The initial fresh weight (W₀) and the fresh weight of the leaves were measured every 5-30 minutes (W_t), and 100%-W_t/W₀ was called as the water loss rate.

Arabidopsis leaves were immersed in stomatal opening buffer (5 mM MES, 10 mM KCl, 50 mM CaCl₂, pH 5.6) for 2 h, transferred to a solution containing 0 mM/300 mM mannitol or 0 μM/30 μM ABA for 2 h and then observed under a microscope for photographs. The aspect ratio was analyzed using Image J software (National Institutes of Health, USA). Three Arabidopsis leaves were run for each treatment, and at least 50 clear stomata in the field of view were obtained for each leaf.

SPSS Statistics 20 (IBM, USA) was employed for data analysis. All experiments were performed at least 3 times. Experimental data of gene expression in H. caspica and aspect ratio with Arabidopsis leaves for stomatal aperture analysis were assessed by Student’s t-test (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001). All other data were assessed using Duncan’s test (p<0.05).

HcNFYA1 was predicted to be targeted by HcmiR169b in our previous work (Yang et al., 2015). Here, RLM-RACE experiment was conducted to verify their targeting between HcmiR169b and HcNFYA1 (Rhoades et al., 2002; Wu et al., 2012). The results of agarose gel electrophoresis were shown in the Supplementary Figure S2 . The final products by sequencing indicated that HcmiR169b cleaves the 3’UTR region of HcNFYA1 with 100% (6/6) efficiency ( Figure 1A ).

Eight-week-old H. caspica were exposed to salt ( Figure 1B ), drought ( Figure 1C ), cold ( Figure 1D ), and oxidative stress ( Figure 1E ) as well as exogenous hormone ABA ( Figure 1F ). After 3 and 24 hours of treatment, individual assimilating branches were extracted and used for qRT-PCR analysis. The results showed that the expressions of HcmiR169b and HcNFYA1 presented a significant negative correlation, HcmiR169b was significantly inhibited and HcNFYA1 was notably induced under different abiotic stress treatments. HcmiR169b and HcNFYA1 appeared to respond to various abiotic stresses, including salt and drought stress; HcNFYA1 might be controlled by HcmiR169b under these adverse conditions.

The coding sequence of HcNFYA1 is 903 bp in length, and its encoded protein contains 300 amino acids. HcNFYA1 protein contained the binding sites for NFYB/C and CCAAT ( Supplementary Figure S3A ). Based on a phylogenetic analysis of conserved amino acids found in HcNFYA1 and members of the NFYA family of other plant species, including Arabidopsis, soybean, maize, rice, and Beta vulgaris, HcNFYA1 was clustered with BvNFYA1 ( Supplementary Figure S3B ).

To determine the subcellular localization of HcNFYA1, the HcNFYA1-GFP fusion protein was expressed in the onion epidermis using an Agrobacterium-mediated transient transformation system. Under laser confocal microscopy, the fluorescent signal of HcNFYA1-GFP overlapped with the 4-diamino-2-phenylindole (DAPI) signal were observed and appeared only in the nucleus ( Figure 2A ). This suggested that HcNFYA1 was translated and then translocated into the nucleus to function.

To identify whether HcNFYA1 functions as a transcription factor. The ORF region of HcNFYA1 was constructed into the yeast BD expression vector pGBKT7 and transformed into Y2H Gold cells. The HcNFYA1 protein, like the positive control (Yin et al., 2021), activated the expression of a downstream reporter gene, the protein encoded by this gene caused X-α-Gal to degrade and made transformed yeast show blue color ( Figure 2B ), indicating that HcNFYA1 had transcriptional activation activity.

To elucidate the function of the HcmiR169b/HcNFYA1 module in abiotic stress, we generated 35S:HcmiR169b and 35S:HcNFYA1 transgenic Arabidopsis homozygous lines with single copy by the methods of inflorescence infection and kanamycine screening. Based on the results of qRT-PCR assay for T₃ generation of transgenic plants, two individual transgenic lines (HcmiR169b OE5 and HcmiR169b OE11, HcNFYA1 OE2 and HcNFYA1 OE3) were selected for subsequent experiments; both of transgenic lines had high expressions for respectively transformed genes ( Figure 3A, D ). Because of the highly conserved mature sequences between HcmiR169b and AtmiR169b, a portion of AtmiR169b expression was also detected possibly ( Supplementary Figure S4 ). The expressions of its predicted target gene AtNFYA1/5 were reduced in 35S:HcmiR169b Arabidopsis by qRT-PCR assay ( Figure 3B, C ).

The seed germination was consistent between 35S:HcmiR169b Arabidopsis and WT on 1/2 MS medium ( Supplementary Figure S5A ). However, the germination rate of 35S:HcmiR169b seeds was lower when NaCl (125 mM) or mannitol (250 mM) was added to the medium in comparison with the WT ( Supplementary Figure S5B, C ), suggesting that HcmiR169b made Arabidopsis more sensitive to salt and drought stresses during the stage of seed germination. Compared with the WT, 35S:HcNFYA1 Arabidopsis seeds in 1/2 MS medium were exhibited slightly delayed germination ( Supplementary Figure S5D ). When NaCl or mannitol was added to the medium, the germination rate of 35S:HcNFYA1 Arabidopsis seeds was decreased compared to the WT ( Supplementary Figure S5E, F ). Overexpression of HcNFYA1 inhibited Arabidopsis seed germination under salt- and drought- stressed conditions.

Under salt stress of 300 mM NaCl, 35S:HcmiR169b Arabidopsis grew smaller and the leaves wilted more severely than WT, whereas 35S:HcNFYA1 Arabidopsis grew larger, leaves wilted less ( Figure 3E ). In addition, under salt stress, the dry weight of aboveground parts ( Figure 3F ), chlorophyll ( Figure 3G ) and soluble protein ( Figure 3H ) contents were lower in 35S:HcmiR169b Arabidopsis and higher in 35S:HcNFYA1 Arabidopsis, compared to those in the WT. In conclusion, HcmiR169b negatively regulated salt tolerance in Arabidopsis, whereas HcNFYA1 was positive in regulating salt tolerance in Arabidopsis.

To examine the role of Halostachys caspica miR169b/NFYA1 in response to drought stress, four-week-old Arabidopsis were stopped from irrigation untill different phenotypes emerged among them. Almost 7 days after ceasing to irrigate, 35S:HcmiR169b Arabidopsis showed leaf drying, and its survival rate after 3 days of rewatering was significantly lower than that of WT. Continuing with no irrigation for 9 days, 35S:HcmiR169b Arabidopsis leaves had completely dried out and wild-type Arabidopsis leaves were wilted, whereas 35S:HcNFYA1 Arabidopsis was still growing well. At this point, 3 days after rewatering, the survival rate of 35S:HcmiR169b Arabidopsis was already below 20%, 50% for wild-type Arabidopsis, and up to 80% for 35S:HcNFYA1 Arabidopsis ( Figure 4A, B ). Under drought stress, the drought-responding positive regulatory hormone ABA content was the highest in 35S:HcNFYA1 Arabidopsis and the lowest in 35S:HcmiR169b ( Figure 4C ). HcmiR169b appeared to be a negative regulator of drought resistance in Arabidopsis, whereas HcNFYA1 was a positive regulator.

Water loss was faster in detached leaves of 35S:HcmiR169b Arabidopsis and slower in leaves of 35S:HcNFYA1 Arabidopsis compared with WT ( Figure 4D ). In detached leaves of Arabidopsis, the rate of water loss was primarily dependent on the degree of stomatal opening. After treatment with 300 mM mannitol, 35S:HcmiR169b Arabidopsis had significantly more open stomata than WT, and 35S:HcNFYA1 Arabidopsis had more completely closed stomata than WT ( Figure 4E-G ). This suggested that the HcmiR169b/HcNFYA1 module may regulate drought tolerance in Arabidopsis through stomatal activity.

Adversity stress leads to disruption of plant cell membrane and ROS homeostasis. We monitored the integrity of cell membranes in various types of Arabidopsis under salt and drought stress based on the Evans blue staining ( Figure 5A ), electrolyte leakage ( Figure 5D ) and malondialdehyde (MDA) content ( Figure 5E ) assays. The results showed that 35S:HcmiR169b Arabidopsis cell membrane was more severely damaged compared to the WT, while 35S:HcNFYA1 Arabidopsis cell membrane integrity was better. Diaminobenzidine (DAB) and nitro-blue tetrazolium (NBT) staining ( Figure 5B, C ), H₂O₂ and O 2 − content ( Figure 5F, G ) measurements with leaves in various Arabidopsis were performed. The results showed that 35S:HcmiR169b Arabidopsis accumulated more ROS and 35S:HcNFYA1 Arabidopsis increased less ROS compared to the WT under salt and drought stress.

Proline (Pro)is used as an osmoregulatory substance to keep the osmotic pressure stability in plants and maintain the integrity of the cell membrane (Székely et al., 2008). The Pro content ( Figure 5H ) and the expression of its key synthetic enzyme gene P5CS ( Figure 5L ) were lower in 35S:HcmiR169b Arabidopsis than those in the WT under salt and drought stress, and the opposite results were obtained from 35S:HcNFYA1 Arabidopsis. Antioxidant enzymes play an important role in scavenging plant ROS species, so the activities of antioxidant enzymes ( Figure 5I-K ) and their transcription levels of relative synthetic enzyme genes ( Figure 5M-O ) were examined in Arabidopsis. Under salt and drought stress, 35S:HcmiR169b Arabidopsis had the lowest antioxidant enzyme activities and gene expression at both protein and RNA levels, whereas 35S:HcNFYA1 Arabidopsis had the highest antioxidant enzyme activities, and WT Arabidopsis was in the middle. Under the control conditions, there was no significant difference in physiological indices and gene transcription levels between the transgenic (HcmiR169b, HcNFYA1 OE) Arabidopsis and the WT.

To investigate the regulatory mechanism in which the Halostachys caspica miR169b/NFYA1 module functions under salt and drought stress, we selected classical stress-responsive genes (RD29A, LEA3), salt stress-related genes (SOS3, NHX1), drought stress-related gene (DREB2A), ABA synthesis genes (NCED3, ABA1) and ABA signaling pathway positively regulatory genes (ABF1, RAB18, ABI5) for further analysis. Notably, in the controls, three genes, LEA3, SOS3, and ABF1, were significantly less expressed in 35S:HcmiR169b Arabidopsis and much more expressed in 35S:HcNFYA1 Arabidopsis than those in the WT, and all of them have CCAAT elements in their promoters ( Figure 6B, C, J ). Among them, ABF1 was directly regulated by NFYA family members in soybean (Yu et al., 2021).

Under salt and drought stress conditions, all genes were significantly more expressed in 35S:HcNFYA1 Arabidopsis than those in the WT, and the opposite was observed in the 35S:HcmiR169b Arabidopsis ( Figure 6 ). According to these results, HcNFYA1 can confer salt and drought tolerance to plants by activating downstream responsive genes; HcmiR169b reduced salt and drought resistance by silencing these pathways these genes involved in.

Exogenous hormone ABA reduces the seed germination rate of plants (Sah et al., 2016). To explore how the miR169b/NFYA1 module is involved in ABA signaling, the seeds of various types of Arabidopsis lines were sown on 1/2 MS medium containing 0, 0.5, and 0.75 μM ABA to observe their germination and cotyledon greening rates. Several types of Arabidopsis were capable of reaching 100% germination and cotyledon greening on ABA-free media; however, the germination rate and cotyledon greening rate of 35S:HcmiR169b Arabidopsis were significantly higher with exogenous hormone ABA treatments than without ABA treatment, while these growth parameters of 35S:HcNFYA1 Arabidopsis were significantly lower than WT ( Figure 7A-H ). ABA also induces the closure of plant stomata. Under 30 μM ABA treatment, stomatal closure in the leaves of 35S:HcmiR169b Arabidopsis was lower and stomatal closure of 35S:HcNFYA1 Arabidopsis was higher compared with WT ( Figure 7I-K ). Accordingly, HcmiR169b inhibited the ABA signaling pathway in Arabidopsis, while HcNFYA1 activated this pathway.