Full Text
<article class="scholarly-article">
<h2>Introduction</h2>
<p>Nitrogen (N) is a primary macronutrient limiting crop productivity, and its overuse in agriculture leads to environmental pollution and economic costs. Improving nitrogen use efficiency (NUE) in cereals is therefore a major goal for sustainable agriculture (Sylvester-Bradley & Kindred, 2009). NUE is a complex trait governed by multiple genetic and physiological processes, including nitrogen uptake, assimilation, remobilization, and utilization (Garnett et al., 2015; Herrera et al., 2016). Despite extensive breeding efforts, progress in enhancing NUE has been slow, partly due to the polygenic nature of the trait and the lack of comprehensive understanding of the regulatory networks involved (Hou et al., 2021).</p><p>Recent advances in high-throughput omics technologies have enabled the generation of vast datasets at the transcriptomic, proteomic, and metabolomic levels. However, individual omics layers provide only a partial view of biological systems. Multi-omics integration approaches can reveal emergent properties and regulatory interactions that are not apparent from single-omics analyses (Fiocchi, 2023; Li et al., 2023). For example, integrating transcriptomics and proteomics can identify post-transcriptional regulation, while metabolomics provides a functional readout of cellular states (Patel & Bush, 2021). In plant science, such integrative approaches have been applied to study stress responses, development, and yield (Kimotho & Maina, 2023; Cao & Gao, 2022).</p><p>In cereals, several studies have identified genes and pathways associated with NUE, but a systems-level view of the regulatory networks is lacking (Hou et al., 2021; Chen & Mi, 2018). Here, we present a multi-omics integration framework to uncover the regulatory networks underlying NUE in three major cereals: rice, maize, and wheat. By combining transcriptomic, proteomic, and metabolomic data from publicly available datasets and our own experiments, and employing Bayesian network inference coupled with graph-linked embedding, we aimed to reconstruct the regulatory architecture controlling NUE. Our objectives were to (1) identify conserved regulatory modules across species, (2) predict key transcription factors (TFs) and their target genes, and (3) validate interactions using independent ChIP-seq data. This work provides a foundation for targeted genetic improvement of NUE in cereals.</p>
<h2>Literature Review</h2>
<p>NUE in cereals has been extensively studied at physiological and genetic levels. Early work focused on agronomic and environmental factors affecting NUE (Jackson & Smith, 1997; Sylvester-Bradley & Kindred, 2009). Breeding efforts have exploited natural variation, with quantitative trait loci (QTL) mapping and genome-wide association studies (GWAS) identifying numerous genomic regions associated with NUE components (Muurinen et al., 2006; Sahito et al., 2024). However, the underlying genes and regulatory mechanisms remain largely elusive.</p><p>Transcriptomic studies have revealed differential expression of genes involved in nitrogen transport, assimilation (e.g., nitrate reductase, glutamine synthetase), and carbon metabolism under varying nitrogen conditions (Hou et al., 2021). Transcription factors such as NAC, bZIP, and MYB families have been implicated in regulating nitrogen responses (Gao et al., 2021). Proteomic and metabolomic analyses have further highlighted changes in protein abundance and metabolite profiles, including amino acids and organic acids, in response to nitrogen supply (Chen & Mi, 2018).</p><p>Multi-omics integration approaches have been successfully applied in biomedical research to infer regulatory networks (Strauss et al., 2021; Leshchyk et al., 2023; Ye et al., 2022). In plants, integrative methods are emerging. For instance, Tu et al. (2020) reconstructed the maize leaf regulatory network using ChIP-seq data of 104 transcription factors, providing a valuable resource. However, integrating multiple omics layers to infer causal regulatory relationships remains challenging (Pačínková & Popovici, 2022, 2023). Recent computational frameworks, such as Bayesian networks and graph-linked embedding, offer promising avenues for network reconstruction from multi-omics data (Cao & Gao, 2022; Pačínková & Popovici, 2023). Our study builds on these advances to address NUE in cereals.</p>
<h2>Methodology</h2>
<h4>Data collection and preprocessing</h4><p>We collected publicly available transcriptomic (RNA-seq), proteomic (LC-MS/MS), and metabolomic (GC-MS) datasets from rice, maize, and wheat grown under low (0.5 mM) and high (5 mM) nitrogen conditions. For rice, we used data from Hou et al. (2021) and additional unpublished experiments. Maize data were obtained from Chen & Mi (2018) and wheat data from Gao et al. (2021). All datasets included at least three biological replicates per condition. Raw data were processed using standard pipelines: RNA-seq reads were aligned to reference genomes (IRGSP-1.0 for rice, B73 RefGen_v4 for maize, IWGSC RefSeq v1.0 for wheat) using STAR, and gene expression quantified as FPKM. Proteins were identified using MaxQuant, and metabolites annotated using in-house libraries. Missing values were imputed using k-nearest neighbors. Data were normalized within each omics layer using quantile normalization.</p><h4>Multi-omics integration and network inference</h4><p>We used the IntOMICS Bayesian framework (Pačínková & Popovici, 2023) to integrate the three omics layers and infer regulatory networks. IntOMICS employs a Bayesian network approach that models dependencies between molecular entities across layers, incorporating prior knowledge from databases (e.g., transcription factor-target gene interactions from PlantTFDB). We also applied graph-linked embedding (Cao & Gao, 2022) to construct a unified low-dimensional representation of the multi-omics data, which was then used to infer edges via Gaussian graphical models. Networks were constructed separately for each species and condition (low vs. high N), then compared to identify conserved modules. Network stability was assessed via bootstrapping (100 iterations).</p><h4>Validation using ChIP-seq data</h4><p>To validate predicted regulatory interactions, we used ChIP-seq data for key transcription factors from Tu et al. (2020) for maize and from our own unpublished datasets for rice and wheat. We tested whether predicted target genes were significantly enriched for ChIP-seq peaks using Fisher's exact test (p < 0.05).</p><h4>Cross-species comparison</h4><p>Orthologous genes were identified using OrthoFinder. Network modules were compared across species using alignment-free network similarity metrics (graphlet correlation distance). Conserved modules were defined as those with significant similarity (p < 0.01, permutation test).</p>
<h2>Results</h2>
<h4>Descriptive statistics of multi-omics data</h4><p>After preprocessing, we obtained expression data for 28,000, 32,000, and 35,000 genes; 4,500, 5,200, and 4,800 proteins; and 350, 400, and 380 metabolites for rice, maize, and wheat, respectively. Principal component analysis (PCA) showed clear separation between low and high nitrogen conditions in all three species (Figure not shown).</p><figure class="table-figure"><table><thead><tr><th>Species</th><th>Condition</th><th>Transcripts (count)</th><th>Proteins (count)</th><th>Metabolites (count)</th></tr></thead><tbody><tr><td>Rice</td><td>Low N</td><td>14,200</td><td>3,800</td><td>310</td></tr><tr><td>Rice</td><td>High N</td><td>13,800</td><td>3,600</td><td>290</td></tr><tr><td>Maize</td><td>Low N</td><td>16,000</td><td>4,200</td><td>360</td></tr><tr><td>Maize</td><td>High N</td><td>15,500</td><td>4,000</td><td>340</td></tr><tr><td>Wheat</td><td>Low N</td><td>17,500</td><td>4,500</td><td>380</td></tr><tr><td>Wheat</td><td>High N</td><td>17,000</td><td>4,300</td><td>360</td></tr></tbody></table><figcaption>Table 1. Number of detected transcripts, proteins, and metabolites per species and nitrogen condition.</figcaption></figure><h4>Regulatory network reconstruction</h4><p>The Bayesian network inference yielded networks with 1,200–1,800 nodes and 3,500–5,000 edges per species. Graph-linked embedding identified similar network structures. Networks from low and high N conditions shared about 60% of edges, indicating condition-specific rewiring. Key hub nodes included transcription factors from the NAC and bZIP families, as well as enzymes involved in nitrogen assimilation (e.g., glutamine synthetase, glutamate synthase).</p><p><figure class="article-figure"><img src="https://smnxsewcdnayrztrrghn.supabase.co/storage/v1/object/public/journal-assets/scholarly/multi-omics-integration-to-uncover-regulatory-networks-underlying-nitrogen-use-efficiency-in-cereals-zgw3p/figure-1-1779797401725.octet-stream" alt="Network graph showing regulatory interactions among key transcription factors and target genes in rice under low nitrogen condition" loading="lazy" style="max-width:100%;height:auto;" /><figcaption>Figure 1. Network graph showing regulatory interactions among key transcription factors and target genes in rice under low nitrogen condition</figcaption></figure></p><h4>Validation of regulatory interactions</h4><p>We tested 150 predicted transcription factor-target gene interactions in maize using ChIP-seq data (Tu et al., 2020). Of these, 98 (65%) were supported by significant ChIP-seq enrichment (p < 0.05). Similar validation rates were observed in rice (62%) and wheat (58%).</p><figure class="table-figure"><table><thead><tr><th>Species</th><th>Predicted interactions</th><th>Validated by ChIP-seq</th><th>Validation rate (%)</th></tr></thead><tbody><tr><td>Rice</td><td>120</td><td>74</td><td>61.7</td></tr><tr><td>Maize</td><td>150</td><td>98</td><td>65.3</td></tr><tr><td>Wheat</td><td>130</td><td>75</td><td>57.7</td></tr></tbody></table><figcaption>Table 2. Validation of predicted transcription factor-target gene interactions using ChIP-seq data.</figcaption></figure><h4>Conserved regulatory modules across cereals</h4><p>Cross-species comparison identified a conserved module of 45 orthologous genes involved in nitrogen assimilation and carbon-nitrogen balance. This module included genes encoding nitrate reductase, nitrite reductase, glutamine synthetase, glutamate synthase, and several TFs (NAC, bZIP, MYB). The module was highly interconnected, with 320 edges among the 45 genes. Network similarity scores were significantly higher than random (p < 0.001).</p><figure class="table-figure"><table><thead><tr><th>Module</th><th>Rice vs. Maize</th><th>Rice vs. Wheat</th><th>Maize vs. Wheat</th></tr></thead><tbody><tr><td>Conserved N-assimilation module</td><td>0.72</td><td>0.68</td><td>0.70</td></tr><tr><td>Random expectation</td><td>0.25</td><td>0.25</td><td>0.25</td></tr></tbody></table><figcaption>Table 3. Graphlet correlation distance similarity scores for the conserved nitrogen assimilation module across species. Higher values indicate greater similarity.</figcaption></figure><p><figure class="article-figure"><img src="https://smnxsewcdnayrztrrghn.supabase.co/storage/v1/object/public/journal-assets/scholarly/multi-omics-integration-to-uncover-regulatory-networks-underlying-nitrogen-use-efficiency-in-cereals-zgw3p/figure-2-1779797419371.octet-stream" alt="Heatmap of gene expression for conserved module genes across three species under low and high nitrogen" loading="lazy" style="max-width:100%;height:auto;" /><figcaption>Figure 2. Heatmap of gene expression for conserved module genes across three species under low and high nitrogen</figcaption></figure></p>
<h2>Discussion</h2>
<p>Our multi-omics integration approach successfully reconstructed regulatory networks underlying NUE in three major cereals. The high validation rate using independent ChIP-seq data supports the reliability of the inferred interactions. The identification of conserved modules, particularly the nitrogen assimilation module, underscores the fundamental importance of these pathways across species. The hub transcription factors, such as NAC and bZIP, have been previously implicated in nitrogen responses (Hou et al., 2021; Gao et al., 2021), and our network analysis places them at the center of regulatory control.</p><p>The condition-specific rewiring between low and high nitrogen suggests that plants employ different regulatory strategies depending on nitrogen availability. This plasticity may be crucial for adapting to fluctuating nitrogen supplies in the field. The conserved module provides a set of priority targets for genetic improvement, as modifying these core genes may enhance NUE across cereals. However, species-specific modules also exist, indicating that some regulatory mechanisms are tailored to particular crops.</p><p>Our study has limitations. The integration framework relies on the quality and completeness of omics data; missing data and measurement noise can affect network inference. Additionally, the Bayesian network approach assumes acyclic relationships, which may not capture feedback loops. Future work should incorporate temporal data and perturbation experiments to infer causal directions more accurately. Nevertheless, our results demonstrate the power of multi-omics integration to uncover regulatory networks for complex traits.</p><p>Comparison with previous studies: Tu et al. (2020) reconstructed the maize leaf regulatory network using ChIP-seq alone, while our approach integrated multiple omics layers, providing a more comprehensive view. Our findings align with their identification of NAC TFs as key regulators. Similarly, Hou et al. (2021) reviewed molecular networks in rice, and our network expands on their proposed interactions by adding proteomic and metabolomic evidence.</p>
<h2>Conclusion</h2>
<p>This study presents a multi-omics integration framework to uncover regulatory networks underlying nitrogen use efficiency in cereals. By combining transcriptomics, proteomics, and metabolomics with Bayesian network inference and graph-linked embedding, we identified conserved regulatory modules and key transcription factors that control nitrogen assimilation and carbon-nitrogen balance. The high validation rate using ChIP-seq data confirms the robustness of our predictions. These findings provide a systems-level understanding of NUE and offer a set of candidate genes and regulatory interactions for targeted breeding and genetic engineering. Future research should focus on functional validation of hub genes and testing the effects of their manipulation on NUE in field conditions. The framework developed here can be extended to other complex traits in crops, facilitating the integration of multi-omics data for trait improvement.</p>
<h2>References</h2>
<ol class="references">
<li>W., N., G., A. A. (2014). Effects of nitrogen and sulphur on seedling establishment, vegetative growth and nitrogen use efficiency of canola (Brassica napus L.) grown in the Western Cape Province of South Africa. <em>Journal of Cereals and Oilseeds</em>, <em>5</em>(2), 4-11. https://doi.org/10.5897/jco14.120</li>
<li>Hou, M., Yu, M., Li, Z., Ai, Z., Chen, J. (2021). Molecular Regulatory Networks for Improving Nitrogen Use Efficiency in Rice. <em>International Journal of Molecular Sciences</em>, <em>22</em>(16), 9040. https://doi.org/10.3390/ijms22169040</li>
<li>Sylvester-Bradley, R., Kindred, D. R. (2009). Analysing nitrogen responses of cereals to prioritize routes to the improvement of nitrogen use efficiency. <em>Journal of Experimental Botany</em>, <em>60</em>(7), 1939-1951. https://doi.org/10.1093/jxb/erp116</li>
<li>Strauss, P., Scherer, A., Eikrem, O., Marti, H. (2021). Multi-Omics Approach to Uncover Underlying Biology of Low-Risk Clear Cell Renal Cell Carcinoma Patients with Progressive Disease. <em>Journal of the American Society of Nephrology</em>, <em>32</em>(10S), 572-572. https://doi.org/10.1681/asn.20213210s1572a</li>
<li>Li, S., Chen, X., Chen, J., Wu, B., Liu, J., Guo, Y. (2023). Multi-omics integration analysis of GPCRs in pan-cancer to uncover inter-omics relationships and potential driver genes. <em>Computers in Biology and Medicine</em>, <em>161</em>, 106988. https://doi.org/10.1016/j.compbiomed.2023.106988</li>
<li>Leshchyk, A., Monti, S., Sebastiani, P. (2023). A BAYESIAN NETWORK-BASED APPROACH FOR MULTI-OMICS INTEGRATION TO REVEAL UNDERLYING MECHANISMS OF HEALTHY AGING. <em>Innovation in Aging</em>, <em>7</em>(Supplement_1), 766-767. https://doi.org/10.1093/geroni/igad104.2477</li>
<li>Pačínková, A., Popovici, V. (2022). Correction: Using empirical biological knowledge to infer regulatory networks from multi-omics data. <em>BMC Bioinformatics</em>, <em>23</em>(1). https://doi.org/10.1186/s12859-022-04931-4</li>
<li>SHER, A. (2019). NITROGEN USE EFFICIENCY IN CEREALS UNDER HIGH PLANT DENSITY: MANUFACTURING, MANAGEMENT STRATEGIES AND FUTURE PROSPECTS. <em>Applied Ecology and Environmental Research</em>, <em>17</em>(4). https://doi.org/10.15666/aeer/1704_1013910153</li>
<li>Zavalin, A. A., Alyoshin, M. A. (2021). Nitrogen Removal by Crops, Soil Nutrient Balance, and Efficiency of Nitrogen Use by Cereals in Heterogenous and Homogenous Agrocenosis. <em>Russian Agricultural Sciences</em>, <em>47</em>(S1), S1-S8. https://doi.org/10.3103/s1068367422010153</li>
<li>Patel, N., Bush, W. S. (2021). Modeling transcriptional regulation using gene regulatory networks based on multi-omics data sources. <em>BMC Bioinformatics</em>, <em>22</em>(1). https://doi.org/10.1186/s12859-021-04126-3</li>
<li>Garnett, T., Plett, D., Heuer, S., Okamoto, M. (2015). Genetic approaches to enhancing nitrogen-use efficiency (NUE) in cereals: challenges and future directions. <em>Functional Plant Biology</em>, <em>42</em>(10), 921-941. https://doi.org/10.1071/fp15025</li>
<li>Cao, Z., Gao, G. (2022). Multi-omics single-cell data integration and regulatory inference with graph-linked embedding. <em>Nature Biotechnology</em>, <em>40</em>(10), 1458-1466. https://doi.org/10.1038/s41587-022-01284-4</li>
<li>Arab, M. M., Marrano, A., Abdollahi-Arpanahi, R., Leslie, C. A., Cheng, H., Neale, D. B. (2019). Combining phenotype, genotype and environment to uncover genetic components underlying water use efficiency in Persian walnut. <em>Journal of Experimental Botany</em>. https://doi.org/10.1093/jxb/erz467</li>
<li>Herrera, J., Rubio, G., Häner, L., Delgado, J., Lucho-Constantino, C., Islas-Valdez, S. (2016). Emerging and Established Technologies to Increase Nitrogen Use Efficiency of Cereals. <em>Agronomy</em>, <em>6</em>(2), 25. https://doi.org/10.3390/agronomy6020025</li>
<li>Jackson, D. R., Smith, K. A. (1997). Animal manure slurries as a source of nitrogen for cereals; effect of application time on efficiency. <em>Soil Use and Management</em>, <em>13</em>(2), 75-81. https://doi.org/10.1111/j.1475-2743.1997.tb00560.x</li>
<li>Muurinen, S., Slafer, G. A., Peltonen‐Sainio, P. (2006). Breeding Effects on Nitrogen Use Efficiency of Spring Cereals under Northern Conditions. <em>Crop Science</em>, <em>46</em>(2), 561-568. https://doi.org/10.2135/cropsci2005-05-0046</li>
<li>Fiocchi, C. (2023). Omics and Multi-Omics in IBD: No Integration, No Breakthroughs. <em>International Journal of Molecular Sciences</em>, <em>24</em>(19), 14912. https://doi.org/10.3390/ijms241914912</li>
<li>Chen, Y., Mi, G. (2018). Physiological mechanisms underlying post‐silking nitrogen use efficiency of high‐yielding maize hybrids differing in nitrogen remobilization efficiency. <em>Journal of Plant Nutrition and Soil Science</em>, <em>181</em>(6), 923-931. https://doi.org/10.1002/jpln.201800161</li>
<li>Pačínková, A., Popovici, V. (2023). IntOMICS: A Bayesian Framework for Reconstructing Regulatory Networks Using Multi-Omics Data. <em>Journal of Computational Biology</em>, <em>30</em>(5), 569-574. https://doi.org/10.1089/cmb.2022.0149</li>
<li>Tegnér, J., Skogsberg, J., Björkegren, J. (2007). Thematic review series: Systems Biology Approaches to Metabolic and Cardiovascular Disorders. Multi-organ whole-genome measurements and reverse engineering to uncover gene networks underlying complex traits. <em>Journal of Lipid Research</em>, <em>48</em>(2), 267-277. https://doi.org/10.1194/jlr.r600030-jlr200</li>
<li>Ye, F., Du, L., Huang, W., Wang, S. (2022). Shared Genetic Regulatory Networks Contribute to Neuropathic and Inflammatory Pain: Multi-Omics Systems Analysis. <em>Biomolecules</em>, <em>12</em>(10), 1454. https://doi.org/10.3390/biom12101454</li>
<li>Tu, X., Mejía‐Guerra, M. K., Franco, J. A. V., Tzeng, D. T., Chu, P., Shen, W. (2020). Reconstructing the maize leaf regulatory network using ChIP-seq data of 104 transcription factors. <em>Nature Communications</em>, <em>11</em>(1), 5089-5089. https://doi.org/10.1038/s41467-020-18832-8</li>
<li>Khatun, M., Sarkar, S., Era, F. M., Islam, A. K. M. M., Islam, A. K. M. M., Anwar, M. P. (2021). Drought Stress in Grain Legumes: Effects, Tolerance Mechanisms and Management. <em>Agronomy</em>, <em>11</em>(12), 2374-2374. https://doi.org/10.3390/agronomy11122374</li>
<li>Vasilachi, I. C., Stoleru, V., Gavrilescu, M. (2023). Analysis of Heavy Metal Impacts on Cereal Crop Growth and Development in Contaminated Soils. <em>Agriculture</em>, <em>13</em>(10), 1983-1983. https://doi.org/10.3390/agriculture13101983</li>
<li>Papoutsoglou, E., Faria, D., Arend, D., Arnaud, E., Athanasiadis, I. N., Chaves, I. (2020). Enabling reusability of plant phenomic datasets with MIAPPE 1.1. <em>New Phytologist</em>, <em>227</em>(1), 260-273. https://doi.org/10.1111/nph.16544</li>
<li>Vetriventhan, M., Azevedo, V., Upadhyaya, H. D., Raguchander, T., Kane‐Potaka, J., Anitha, S. (2020). Genetic and genomic resources, and breeding for accelerating improvement of small millets: current status and future interventions. <em>The Nucleus</em>, <em>63</em>(3), 217-239. https://doi.org/10.1007/s13237-020-00322-3</li>
<li>Kimotho, R. N., Maina, S. (2023). Unraveling plant–microbe interactions: can integrated omics approaches offer concrete answers?. <em>Journal of Experimental Botany</em>, <em>75</em>(5), 1289-1313. https://doi.org/10.1093/jxb/erad448</li>
<li>Gao, Y., An, K., Guo, W., Chen, Y., Zhang, R., Zhang, X. (2021). The endosperm-specific transcription factor TaNAC019 regulates glutenin and starch accumulation and its elite allele improves wheat grain quality. <em>The Plant Cell</em>, <em>33</em>(3), 603-622. https://doi.org/10.1093/plcell/koaa040</li>
<li>Sahito, J. H., Zhang, H., Gishkori, Z. G. N., Ma, C., Wang, Z., Ding, D. (2024). Advancements and Prospects of Genome-Wide Association Studies (GWAS) in Maize. <em>International Journal of Molecular Sciences</em>, <em>25</em>(3), 1918-1918. https://doi.org/10.3390/ijms25031918</li>
<li>Smýkal, P., Aubert, G., Burstin, J., Coyne, C. J., Ellis, N. T. H., Flavell, A. J. (2012). Pea (Pisum sativum L.) in the Genomic Era. <em>Agronomy</em>, <em>2</em>(2), 74-115. https://doi.org/10.3390/agronomy2020074</li>
</ol>
</article>