The maize cultivar (Zea mays L. subsp. indentata (Sturtev.) Zhuk., accession number ZEA 3639, spring type, was provided by the Leibniz Institute of Plant Genetics and Crop Plant Research (IPK, Gatersleben, Germany). To investigate the effects of SAPs on seed germination and gene expression, an experiment was conducted with five treatment groups, as outlined in Table 1.

Three types of SAPs were evaluated in this study. Merck and SWT, which are synthetic fossil-based polymers composed of cross-linked sodium polyacrylate, and ABG, a bio-based superabsorbent derived from wood (lignin-based). The sodium polyacrylate structure of the synthetic SAPs contains sodium (Na⁺) as the neutralizing counter-ion, which can be exchanged or released upon hydration. Although the manufacturer did not disclose the specific ionic composition of the bio-based ABG, we evaluated its effects on seedling evolution and ion uptake, alongside those of the synthetic variants. For each SAP treatment, 30 coated seeds and 30 non-coated seeds (controls) were placed on individual germination paper sheets (Keimtestpapier gelb; 120 mm × 300 mm, 160 g/m²; Sartorius, Göttingen, Germany). SAP application rates followed manufacturer recommendations. Their amounts were normalized based on the distinct water absorption capacity of each polymer (See Supplementary Table S1) and the thousand-seed weight of the maize variety (TSW = 363 g) to ensure a comparable amount of available water per seed across the SAP treatments.

Each germination sheet for the SAP and CS treatments was moistened with 5 mL of distilled water to induce drought stress, while the CN sheets received 12 mL to represent well-watered conditions. The applied water volumes were determined based on the water-holding capacity of the germination paper, following standard water-soaking procedures (Rajendra Prasad, 2023). For the stress treatments (CS and SAPs), the appropriate moisture level was established through preliminary water-retention tests. All samples were incubated for 7 days in a climate-controlled chamber (BINDER KBW 720; Tuttlingen, Germany) at 25 ± 1°C under a 16/8 h light/dark photoperiod, following standard protocols for maize germination tests (Rajendra Prasad, 2023). At the conclusion of the incubation period, germination percentage and seedling biochemical parameters (total phenol content and sodium content) were recorded and prepared for subsequent gene expression analysis.

To evaluate the effects of different SAP treatments on seed germination behavior, we quantified and compared the proportions of normal, abnormal, and non-germinated seedlings. Seedlings were classified as normal if the radicle (embryonic root) and coleoptile (protective sheath) exhibited a straight primary root and showed no signs of necrosis in the cotyledons or root tissues. Seeds that displayed no visible germination, with no radicle and coleoptile, were classified as non-germinated. Abnormal seedlings were identified based on morphological anomalies, including looping or twisting of the primary root or cotyledons, absence of shoot or root structures, or the presence of necrotic tissue (Rajendra Prasad, 2023) (See Supplementary Figure S2).

A contingency table was constructed for the germination categories (abnormal, non-germinated, and normal) across treatments, with each treatment comprising controls and SAPs applied to 30 seeds. Pairwise comparisons of germination categories (normal, abnormal, and non-germinated) across treatments were performed using the G-test (log-likelihood ratio test), implemented in the pairwiseNominalIndependence function from the rcompanion package (Mangiafico, 2016) in R (R Core Team, 2021). P-values for the pairwise comparisons were adjusted for multiple comparisons using the false discovery rate (FDR) method (Benjamini and Hochberg, 1995). The significance of the pairwise comparisons was assessed using an adjusted p-value threshold of α = 0.05. Compact letter display codes were generated for each comparison using the cldList function from the rcompanion package to visualize significant differences between treatments within each germination category (Mangiafico, 2016).

Error bars represent the standard error for each proportion, computed using the binconf function in the Hmisc package in R.

Total phenolic content (TPC) was extracted from 200 mg of fresh, normal seedlings, with six replications. Samples were flash-frozen in liquid nitrogen and homogenized into a fine powder using a mixer mill (Retsch MM200, Haan, Germany) at 25,000 rpm for 10 min. The powder was subjected to a sequential extraction with three 1 mL aliquots of methanol (CH₃OH 95% (w/v)). After each methanol addition, the mixture was vortexed and centrifuged at 5,000 rpm for 5 min at 4°C (Rotina 380R, Hettich, Germany). The resulting supernatants were pooled for analysis. TPC was determined using the Folin-Ciocalteu colorimetric method (Anuar, 2014). A reaction mixture containing 0.5 mL of the methanolic extract, 0.5 mL of Folin-Ciocalteu reagent, and 0.5 mL of deionized water was briefly vortexed. After 30 seconds, 5 mL of sodium hydroxide (NaOH 1% (w/v)) was added. The solution was vortexed again and incubated for 30 min at room temperature (21 ± 1°C). Absorbance was measured at 760 nm using a UV-Visible spectrophotometer (UV-1800, Shimadzu, Germany). The absorbance values were used to quantify the total phenolic content by referencing a standard calibration curve prepared with gallic acid (R² > 0.99) (Supplementary Figure S1).

For sodium (Na⁺) analysis, the procedure was performed according to the manufacturer’s instructions for the Horiba LAQUAtwin Na⁺ ion-selective meter (Hildesheim, Germany). Specifically, 200 mg of fresh tissue from normal seedlings (six replications) was flash-frozen in liquid nitrogen and homogenized (Rotina 380R, Hettich, Kirchlengern, Germany) at 25,000 rpm for 10 min to obtain a fine powder. The homogenate was resuspended in 0.5 mL of distilled water and incubated at room temperature for 30 min with continuous agitation to ensure complete ion extraction. Insoluble material was sedimented by centrifugation at 10,000 rpm for 15 min (Rotina 380R, Hettich, Germany), and the Na⁺ concentration in the supernatant was quantified using the ion-selective meter. Results were expressed as parts per million per gram fresh weight (ppm).

TPC data were analyzed using one-way ANOVA in R software, followed by Tukey’s post hoc test to identify pairwise differences. Residual normality was evaluated using the Shapiro–Wilk test (Shapiro and Wilk, 1965), while homogeneity of variance was assessed with a Bartlett’s test (Bartlett, 1937). For Na⁺ analysis, due to variance heterogeneity, Welch’s ANOVA (Welch, 1951) was applied, followed by Games-Howell post hoc tests (Lee and Lee, 2018) to determine significant differences between treatments. A significance level of α = 0.05 was used for all comparisons.

Total RNA was extracted from plant tissues using the TIANGEN RNAprep Pure Plant Kit (Cat. no. 4992237, Beijing, China). Briefly, plant samples were ground to a fine powder in liquid nitrogen. For each 100 mg of tissue, 450 µL of buffer RL, containing 1% β-mercaptoethanol, was added. The mixture was then vortexed vigorously, and after an optional incubation at 56°C for 1–3 minutes, the lysate was transferred to an RNase-free filter column and centrifuged at 12,000 rpm for 2–5 minutes. The supernatant was collected for subsequent steps. To ensure high-quality RNA, a DNase I treatment was performed by adding 80 µL of the DNase I working solution (10 µL of DNase I stock and 70 µL of nuffer RDD) to the RNA sample for 15 minutes at room temperature. The RNA was then purified by washing with buffer RW1 and buffer RW, followed by elution with 30-100 µL of RNase-Free water.

Total RNA was extracted from plant tissues, and RNA integrity was assessed using the Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, USA). Messenger RNA (mRNA) was purified from total RNA using poly-T oligo-attached magnetic beads. After RNA purification and fragmentation, first-strand cDNA was synthesized using random hexamer primers. The second strand cDNA was synthesized using dUTP instead of dTTP to maintain strand specificity. The directional library was then constructed by performing end repair, A-tailing, adapter ligation, size selection, amplification, and purification. The final library was quantified using Qubit and real-time PCR, and its size distribution was analyzed using the Bioanalyzer. Following library quality control, the libraries were pooled based on effective concentration and targeted data amount and subjected to Illumina sequencing. Sequencing was performed using the Sequencing by Synthesis method, where fluorescently labeled dNTPs, DNA polymerase, and adapter primers were added to the flow cell for amplification. As each sequencing cluster extended its complementary strand, the addition of each fluorescently labeled dNTP released a corresponding fluorescence signal. The sequencer captured these signals and, using computer software, converted them into sequencing peaks to obtain sequence information for the target fragment. The mRNA extraction and library preparation were performed by Novogene (Munich, Germany).

Initially, the raw sequence reads underwent quality control using fastp version 0.24.1 (Chen et al., 2018). Adapter sequences were automatically identified and removed via the –detect_adapter_for_pe parameter, and poly-G/poly-A tails were trimmed using –trim_poly_g and –trim_poly_x, respectively. Furthermore, reads were trimmed from the 3’-tails to the front using the options –cut_tail and –cut_mean_quality 20 to achieve window mean qualities above 20. Afterwards, reads shorter than 50 bases were discarded (–length_required 50). Finally, bases of reads were classified as unqualified if their phred-score was below 20 (–qualified_quality_phred 20), and reads were discarded if the number of unqualified bases exceeded 30% (–unqualified_percent_limit 30).

The reference genome Zm-B73-REFERENCE-NAM-5.0 and corresponding gene and protein annotations were downloaded from Ensembl plant (Yates et al., 2021; Jiao et al., 2017). Filtered reads were aligned to the genome using STAR version 2.7.11b (Dobin et al., 2012) with default parameters. Base and read quality statistics were generated using fastp and FastQC version 0.12.1 (Andrews et al., 2010). Alignment statistics were generated using the bamqc and rnaseq commands of Qualimap version 2.3 (Okonechnikov et al., 2015). Consequently, all statistics were gathered using MultiQC version 1.28 (Ewels et al., 2016) for evaluation. Following the quality assessment using MultiQC and Qualimap, all samples met the established quality thresholds, and consequently, no samples were excluded or flagged for further analysis.

Sorting of alignment files was done using Sambamba version 1.0.1 (Tarasov et al., 2015) and gene-level quantification of alignment files was determined using featureCounts version 2.0.8 (Liao et al., 2013). Gene counts were extracted, and genes were filtered out if they were not abundant, i.e., if they did not have a raw count of at least 10 in at least 70% of the samples. This filtering was performed to increase statistical power by reducing the number of tests, thereby enabling a less stringent multiple-testing adjustment. Differential expression analysis using DESeq2 version 1.46.0 (Huber and Michael, 2017) in R version 4.4.3 (R Core Team, 2021) was conducted with CS as the reference group, applying FDR adjustment for p-values (padj) and log₂ fold change (LFC) shrinkage with apeglm (Zhu et al., 2018). Finally, for each comparison, we considered genes as differentially expressed if padj < 0.05 and if they were up- or downregulated by a factor of 2 or higher (|LFC| > 1). For each comparison, we performed functional enrichment for GO terms as well as KEGG Pathways on the DEG sets by using the gost function of the gprofiler2 R package (version 0.2.3) (Kolberg et al., 2020). Terms were considered significantly enriched if padj < 0.05. Arabidopsis thaliana orthologs and gene names were retrieved from MaizeGDB (MaizeGDB, 2025).

A heatmap visualizing gene expression, generated from z-scores calculated for each gene from the DESeq2 analysis (to account for differences in library sizes), was produced using the ComplexHeatmap package in R (Mangiola and Papenfuss, 2020). The z-score was computed for each sample and gene using the formula 1:

where X represents the raw count value, µ is the mean expression value of the gene across all samples, and σ is the standard deviation of the gene’s expression across all samples.

Germination success following SAP treatments was assessed by classifying seedlings into three categories: normal, abnormal, and non-germinated. The distribution of these outcomes is shown in Figure 1. An omnibus G-test was conducted to evaluate overall treatment effects on germination at a significance level α = 0.05. The results showed a reduction in the proportion of normal seedlings under all stress conditions, including SAP treatments and CS. Although there was no statistically significant difference among the SAP treatments, Merck showed a trend toward a higher germination rate (80%) than ABG (76%) and SWT (60%). In contrast, the SWT treatment negatively affected seedling development, resulting in a decrease in the proportion of normal seedlings and a corresponding increase in abnormal seedlings compared with the CN treatment leading to a 30 percentage points rise in non-germination. This elevated rate of morphological anomalies suggests a less favorable interaction under the tested conditions. Further details are provided in Supplementary Table S2. The ABG treatment performed at an intermediate level, with its germination profile not significantly different from that of CS. Regarding the non-germinated category, there were no statistically significant differences among any of the stress treatment groups. However, all stress treatments showed higher proportions of non-germinated seedlings than CN, with SWT showing the highest non-germination rate (10%) and significantly higher abnormal seedling rates. This suggests that while certain SAPs can support germination development, their ability to overcome initial germination inhibition under the applied stress conditions may vary fully.

To evaluate the biochemical impact of the SAP treatments, TPC, a common indicator of oxidative stress, was assessed in the seedlings. As illustrated in Figure 2A, statistical analysis revealed no significant differences in TPC levels among any of the treatment groups, including the controls. Despite this lack of statistical significance, some patterns were observed. Notably, the ABG treatment exhibited the lowest TPC values among all groups by 0.08 mg/mL, including the controls, potentially suggesting a stress-mitigating effect. The SWT treatment also resulted in TPC levels that were, on average, lower than the CS control. The MERCK treatment displayed an intermediate profile, with TPC levels generally positioned between those of the controls and the other SAPs. High variability in TPC levels was observed, particularly for ABG and MERCK. While SAPs appeared to modulate seedlings’ phenolic response, no statistically significant reduction in oxidative stress indicators was observed under the conditions of this study.

Sodium accumulation in seedlings was analyzed to evaluate the impact of SAP treatments onion uptake (Figure 2B). While all SAP treatments, on average, increased Na⁺ content in seedlings compared to controls, statistical analysis revealed the following specific patterns. The ABG treatment resulted in significantly higher Na⁺ levels compared to the CN control. In contrast, no significant difference was observed between the fossil-based SAPs (SWT and MERCK) and the controls, a finding that may be influenced by high variability within these treatment groups. However, the consistent numerical elevation of Na⁺ in SWT and MERCK, relative to the controls, suggests a tendency toward increased sodium uptake. This overall pattern of increased sodium accumulation, particularly the significant increase with ABG, could indicate an additional osmotic or salt stress induced by the SAPs, which may have contributed to the observed impacts on seedling development.

A total of 129 million to 177 million raw reads per sample were obtained across five treatments, with two biological replicates per SAP treatment and per control group. After adapter removal and filtering of low-quality sequences, over 97% of the resulting reads were classified as high-quality. When mapped to the Zea mays reference genome, between 92.16% and 95.04% of the high-quality sequences were successfully aligned, indicating efficient mapping to the reference genome (See Supplementary Table S3).

Transcriptome analysis revealed that the most profound transcriptional shift occurred in response to drought stress itself, with the CN vs. CS comparison yielding the highest number of DEGs with 156 upregulated and 152 downregulated (Figure 3A). The application of SAPs distinctly modulated this response. Specifically, while MERCK and ABG induced a similar total number of DEGs against the stress control (32 and 33, respectively), their expression patterns differed significantly; for instance, MERCK showed a higher proportion of upregulated DEGs (20 up, 12 down). In comparison, the ABG response was more balanced (20 up, 13 down). In stark contrast, the SWT vs. CS comparison yielded the lowest transcriptional response, with only 21 total DEGs, suggesting a suppressed or failed molecular adaptation (See Supplementary Table S4).

The distribution of common and unique DEGs, visualized in the UpSet plot (Figure 3B), further clarifies these distinct strategies. The CN vs. CS comparison identified 290 unique DEGs, underscoring the extensive and specific transcriptional reprogramming induced by drought. Among the SAPs, the most significant overlap involved six DEGs shared between the MERCK vs. CS and ABG vs. CS comparisons, suggesting some convergence in their molecular mechanisms. These two treatments also shared several DEGs with the primary drought response (CN vs. CS), indicating they modify existing stress pathways. In contrast, the minimal number of DEGs and limited overlap seen in the SWT treatment reinforce the conclusion of a muted and distinct response. The fact that only one DEG was shared across all three SAP treatments indicates that each polymer elicits a distinct molecular adaptation in the seedling.

Functional enrichment analysis provided further insight into these responses. GO enrichment analysis identified a limited number of significantly enriched terms exclusive to the MERCK vs. CS comparison, suggesting specific pathway modulation by this SAP (see Supplementary Table S5). Notably, no significant GO terms were identified for the broad-spectrum CN versus CS drought response, nor for the DEGs of the ABG and SWT treatments. Similarly, KEGG pathway analysis identified only four significantly enriched pathways, all exclusive to the SWT vs. CS comparison (Supplementary Table S6), with no significant pathways observed in any other contrast. This general lack of widespread functional enrichment, particularly for the overall drought response, suggests that the seedling’s molecular adaptations may involve subtle, diverse, or novel pathways not fully captured by current annotations, rather than broad, well-defined functional categories.

Our results indicate that drought stress negatively affected maize germination quality, reflected in a higher proportion of abnormal and non-germinated seeds in CS compared to CN (Figure 1). This finding is consistent with the established mechanism in which drought reduces osmotic potential, preventing sufficient water uptake by the seed. This impairment slows the essential metabolic processes required for germination, thereby reducing germination success rates (Jaleel et al., 2009). Indeed, our observation of reduced germination success under stress aligns directly with previous studies on maize (Khodarahmpour, 2011).

In our study, the MERCK SAP showed a tendency to support normal seedling development, with a 20 percentage points higher germination rate than SWT and four percentage points higher than ABG, and fewer seedling abnormalities than other Stressed treatments (by 13%), suggesting a potentially beneficial role for MERCK under drought conditions. In contrast, the SWT treatment had a clear adverse impact, with the highest rate of abnormal seedlings (30%) and non-germination (10%), and the lowest proportion of normal seedlings (60%), suggesting that any potential benefits of water retention from SWT were negated by its inhibitory or incompatible properties, thereby exacerbating stress on germinating maize seeds, the ABG treatment was performed at an intermediate level of complexity. The general trend of slightly higher non-germinated seeds in SAP-treated groups compared to controls hints that initial germination inhibition under stress remains a challenge that these SAPs did not fully overcome.

Abiotic stresses, including drought, are known to induce oxidative stress in plants by overproducing ROS, which can damage vital cellular components. In response, plants typically enhance their antioxidant systems to tolerate this stress better, often leading to the accumulation of secondary metabolites, such as phenolic compounds (Rahimi et al., 2021). This defensive accumulation is a recognized marker of drought response, with numerous studies confirming that water deficit effectively increases TPC in plants (Mohammadi et al., 2020). Higher phenolic levels often correlate directly with enhanced antioxidant capacity (Mohammadi et al., 2020). However, contrary to the expected trend, our study did not find a significant difference in TPC levels across the treatments. This observation suggests that SAPs may have reduced the severity of drought stress, thereby limiting the triggers that typically stimulate phenolic production. However, these findings should be interpreted with caution. Since TPC was measured only at a single developmental stage (7-day seedlings), it is possible that the peak stress response occurred at a different time. Thus, while the stable TPC levels are consistent with a stress-mitigation hypothesis, they do not definitively prove that oxidative stress was entirely absent.

A critical finding of our study is that seedlings treated with SAPs exhibited significantly elevated internal sodium levels, with concentrations increasing by at least 77% compared to CN (18 ppm), reaching 32 ppm in ABG and SWT, and 35 ppm in Merck. This accumulation is consistent with the release of sodium ions from the polymer matrix during hydration, particularly for the sodium polyacrylate-based SAPs (Merck and SWT), which rely on Na⁺ dissociation for swelling. Interestingly, despite being a bio-based material, the ABG treatment also resulted in increased seedling Na⁺ levels. While this suggests that the bio-polymer formulation may contribute ions to the rhizosphere, we cannot rule out the possibility that SAPs altered the osmotic dynamics of the root zone, thereby influencing passive ion uptake mechanisms. Regardless of the specific mechanism, this inadvertent elevation of sodium imposed a secondary salinity stress on top of the primary drought condition. This dual-stress scenario is particularly detrimental, as high Na⁺ concentrations can disrupt osmotic and ionic homeostasis, inducing secondary oxidative stress (Bosnic et al., 2018). Maize is moderately sensitive to salinity, with germination and early seedling stages particularly vulnerable (Farooq et al., 2015; Khalid et al., 2023). This increased sensitivity can manifest not only as a failure to germinate but also as impaired post-germination development. For example, Hernandez et al. (2018) found that although germination percentage was unaffected by combined osmotic and metabolic stress, subsequent seedling growth and fresh weight were significantly reduced. These findings align with our observations of increased seedling abnormalities in SAP treatments. Thus, the elevated sodium load associated with SAP application, including the bio-based ABG, likely contributed significantly to these developmental setbacks, transforming what was intended as a drought-mitigation tool into an additional source of compound stress for maize seedlings.

Our transcriptome analysis provides a comprehensive understanding of how the distinct physiological outcomes observed in control (CN) and SAP-treated (MERCK, ABG, SWT) seedlings are driven by fundamentally different impacts on gene expression, affecting RNA metabolism, stress signaling, metabolic adjustments, and developmental regulation.

While this study offers valuable initial insights into maize transcriptomic responses to SAPs under drought, it has limitations. Our investigation used a single maize cultivar and two biological replicates per treatment. Furthermore, only three SAP formulations were evaluated, limiting insights into the full diversity of polymer chemistries. Future research should address these by incorporating diverse maize genotypes and increasing replication for enhanced statistical power. Broader screening of additional SAP chemistries is also crucial. Ultimately, future work should integrate multi-year, multi-site field trials with mechanistic studies to: (i) verify agronomic benefits of promising SAP formulations, (ii) identify the precise source of Na⁺ release (which appears linked to abnormal seedling development), and (iii) functionally validate candidate gene biomarkers via techniques like qPCR or CRISPR/Cas9. This validation is key for developing high-throughput screening platforms. It will help optimize SAP technologies for broader applicability across crops and combined-stress environments, complementing genetic improvements for durable drought resilience.