**Background:** Prior drought stress can alter plant response patterns and increase tolerance to subsequent stress, a phenomenon known as 'drought memory.' While progress has been made in understanding drought memory in model plants, the mechanisms in psammophytes (sand-dune plants) remain largely unknown. Agriophyllum squarrosum is a pioneer annual plant widely distributed in Asian deserts with outstanding water use efficiency. This study aimed to dissect the drought memory mechanism of A. squarrosum and determine discrepancies between two ecotypes adapted to different water availability.
**Methods:** Two contrasting ecotypes were studied: AEX from a semi-arid site (Horqin Sand Land, 485 mm annual precipitation) and WW from an arid region (Tengger Desert, 166 mm annual precipitation). One-month-old seedlings were subjected to three rounds of dehydration-rehydration cycles (4 days water suspension followed by 5 days rehydration). Physiological measurements included soil moisture content (SMC), leaf relative water content (RWC), and water loss of isolated leaves. For transcriptomic analysis, mature leaves were collected from control (R0), first dehydration (S1), and third dehydration (S3) treatments. RNA sequencing generated 86.8 Gb clean reads (average 7.23 Gb per sample, Q30 average 94.2%), which were de novo assembled into 52,925 unigenes, with 32,488 (61.4%) annotated in at least one database. Drought memory genes (DMGs) were identified as drought-responsive genes with significantly different expression between S1 and S3, classified into four types: [+/+], [−/−], [+/−], and [−/+]. Ortholog groups were sought between A. squarrosum and three previously studied species (A. thaliana, Z. mays, P. virgatum). Co-expression analysis between drought memory transcription factors (TFs) and other DMGs was conducted using Pearson correlation coefficient. Protein-protein interaction (PPI) networks were constructed using STRING and visualized with Cytoscape.
**Key Results:** Physiological monitoring showed that WW had stronger and longer-lasting drought memory than AEX. In AEX, a consecutive decline of RWC in R2 and S3 was observed, indicating attenuation in drought memory function, while no such decline was found in WW. A total of 5,379 (12.5%) and 4,436 (10.7%) genes were drought-responsive in AEX and WW, respectively. DMGs comprised 1,776 (33.0%) of drought-responsive genes in AEX and 1,384 (31.2%) in WW. The [+/−] subgroup dominated in both ecotypes (48.4% in AEX, 39.9% in WW). Comparison across species identified 257 shared DMG ortholog groups (OGs) among all five tested species, 406 OGs unique to A. squarrosum, and 655 and 339 OGs unique to AEX and WW, respectively. Shared DMGs were enriched in phenylpropanoid biosynthesis, MAPK signaling, and saccharide metabolism. A. squarrosum-specific DMGs were enriched in responses to heat, high light intensity, hydrogen peroxide, and water deprivation. WW-unique DMGs were concentrated in photosynthesis-related functions. PPI network analysis revealed heat shock proteins (HSPs) as hub genes in both ecotypes. Co-expression analysis identified 17 TFs in AEX and 20 TFs in WW that showed 'on-off' controlling patterns, where pairs of TFs appeared to function as molecular switches regulating the expression of DMG sets. For example, in shared DMGs in AEX, a module of AsGBF3-AsODORANT1-AsWRKY53-AsARF19 controlled 30 DMGs; in WW, an AsNF-YA4-AsRAV1 module controlled 48 DMGs. DMGs subjected to TF switches included heat shock proteins, chaperonins, late embryogenesis abundant proteins, cytochrome P450, MAPKs, and glutathione S-transferases. Q-PCR validation of 10 TFs confirmed high correlation with RNA-seq data.
**Clinical Implications:** This is a plant biology study with no direct clinical implications. However, the findings provide valuable molecular resources for understanding stress resistance mechanisms in plants, which could inform breeding programs for drought-tolerant crops and pseudocereals, particularly for cultivation in sandy and arid lands. The proposed regulatory module—where TF pairs act as molecular switches controlling drought memory gene expression—offers a conceptual framework for understanding how plants balance growth and stress defense under recurring drought conditions.