Multi‐omics assisted identification of the key and species‐specific regulatory components of drought‐tolerant mechanisms in Gossypium stocksii

Cotton (Gossypium spp.) is one of the most important fibre crops globally, with the main cultivated species Gossypium hirsutum L. and G. barbadence L. contributing more than 90% of the total fibre production (Wang et al., 2019). However, low genetic diversity in germplasm has largely impeded the continuous improvement of cultivated cotton. Wild cotton species have desirable characters such as resistance to various biotic and abiotic stresses and high‐quality fibre products, which are therefore natural reservoirs for cotton genetic improvement (Wendel and Cronn, 2003).

Xinjiang is the earliest and, nowadays, the largest cotton‐planting region in China. However, as a typical arid region in Central Asia, water scarcity has severely hindered its cotton production. Thus, the breeding of drought‐tolerant cultivated species is currently one of the top priorities for China’s cotton production. Gossypium stocksii (2n = 2x = 26, E₁) (Figure 1a), native wild cotton to Eastern Africa, is tolerant to drought and cotton leaf curl disease (CLCuD). Till recently, some hybrids between G. stocksii and cultivated cotton species have been developed to improve the cotton resistance to drought stress and virus (Gill and Bajaj, 1996; Nazeer et al., 2014). Nevertheless, the molecular mechanism of G. stocksii stress resistance is still mostly unknown. To facilitate the genetics study of G. stocksii and for a further possibility to utilize its drought‐tolerant traits in breeding, herein we generated a high‐quality genome sequence of G. stocksii and surveyed its drought‐tolerant mechanisms based on integrated analyses of multi‐omics data.

Figure 1.

(a) Gossypium stocksii and its tissues of flower, boll and fibre. (b) Statistics of G. stocksii genome assembly and annotation. (c) Genomic synteny between G. stocksii and G. arboreum. (d) PEG‐simulated drought stress experiments. (e) Drought‐induced differentially expressed transcription factors. (f) Top enriched GO terms of drought‐induced DEGs. ‘*’ indicates the processes are enriched in both G. stocksii and G. arboreum. (g) Flowchart diagram of the establishment of TF–target interactions. (h) Identification of TFBS in conserved promoter regions. (i) Regulations between TFs with their targets enriched in drought‐induced DEGs in G. stocksii. (j) Function analyses of the top enriched drought‐responsive TFs. (k) Differentially expressed miRNAs in drought‐stressed G. stocksii. (l) qPCR analyses of miR164 and NAC1 expression after PEG treatment. (m) T‐plot of miR164:NAC1 regulation.

We used a hybrid strategy of genome sequencing and assembly in this work. Briefly, Pacbio long reads (102.8×) and Illumina short reads (123.5×) were used for contig construction, which gave rise to a total of 173 contigs (Figure 1b). These contigs were anchored and ordered using Hi‐C data, resulting in a chromosomal‐level genome of 127 scaffolds. The assembled genome is about 1.45 Gb in length, with about 99.2% accounting for 13 pseudochromosomes that are ranged from 83.8 to 129.5 Mb. BUSCO and CEGMA evaluation indicated the assembly’s completeness was about 95.4–98.4%, while 98.7% of the generated short reads could be mapped to the genome (coverage 99.9%). During the preparation of this manuscript, another genome assembly of G. stocksii was built towards a comprehensive understanding of G. stocksii resistance to CLCuD (Grover et al., 2021). Grover’s assembly shows comparable quality to our assembly. One of the main differences between the two projects is the strategy of gene annotation, i.e. Grover et al. used a series of RNA‐seq data from closely related species to assist the gene modelling and annotation. In contrast, we used the transcripts generated from four G. stocksii tissues (leaf, stem, root and flower) by RNA‐seq (one library for each tissue, 6.8–9.5 Gb) and Iso‐seq (one library for mixed tissues, 28.9 Gb). Eventually, this work identified a total of 46 224 protein‐coding genes, one‐third more than that of Grover’s assembly. The function of about 94.5% of genes identified in this work was assigned by homology search against the public databases, suggesting the high credibility of our annotation.

To get insight into the drought‐resistant mechanisms of G. stocksii and its divergence during the evolutionary scenario, we comparatively analysed the transcriptomic response of G. stocksii and G. arboreum to PEG‐simulated drought stress. Unlike the extensive conservation in gene content and arrangement (Figure 1c), the transcriptomic response of G. stocksii and G. arboreum to drought stress was largely varied. After a 6‐h treatment of PEG‐6000 (20%) (Figure 1d), a total of 4484 and 2147 differentially expressed genes (DEGs) were induced in G. stocksii and G. arboreum seedlings (4‐week), respectively. More than 80% of these DEGs were conserved between G. stocksii and G. arboreum; however, only 1043 were regulated in the same manner. Exampled by the transcription factors (TFs), 414 out of the 449 differentially expressed TFs in G. stocksii were conserved in G. arboreum, while only 126 were differentially expressed in both species (Figure 1e). Function analyses revealed that several biological processes were altered in both G. stocksii and G. arboreum, such as protein dephosphorylation, plant‐type cell wall organization, microtubule‐based process and response to auxin (Figure 1f). However, about half of the top enriched biological processes were species specific. All these findings indicated the divergence of the drought‐tolerant mechanisms in G. stocksii and G. arboreum. Since most DEGs are phylogenetically conserved, this deviation might be largely ascribed to the variation in transcriptional regulation.

We then tried to identify the key regulators involved in G. stocksii drought tolerance through regulatory network analyses (Figure 1g). Based on the assessment of the deposition of H3K4me3, we found that most of the promoter regions (˜82%) in G. stocksii were located within 1.5 kb upstream of the transcription start sites (TSSs). The interactions between TFs and their targets were determined if the transcription factor binding sites (TFBSs) were identified within the conserved promoter regions of the target genes (Tian et al., 2020) (Figure 1h). To minimize the data redundancy, expression correlation between the TFs and target genes was analysed, and the interactions between co‐expressed pairs were retained. Chi‐square test analyses found that target genes of 69 TFs were enriched in the DEGs (Figure 1i). Biological roles for the top‐enriched TFs in inducing stress tolerance have been demonstrated in many plants, such as involvement of ABI5, ABF2 and GBF2 in ABA‐mediated regulation in response to drought or salinity stresses (Ashraf et al., 2018). Consistently, predicted targets of these TFs in G. stocksii were also enriched in the protein phosphorylation or dephosphorylation that are closely associated with the ABA signalling (Figure 1j).

We also assessed the expression of microRNAs (miRNAs) in the response of G. stocksii and G. arboreum to drought stress. After PEG‐treatment, 20 differentially expressed miRNAs (DE‐miRNAs) were found in G. stocksii, whereas none in G. arboreum (Figure 1k). A total of 44 target genes of the DE‐miRNAs in G. stocksii were determined based on the degradome signature. Interestingly, only miR164 and its target gene NAC1 were inversely correlated in expression in drought‐stressed G. stocksii (Figure 1l). The miR164:NAC module is conserved and involved in response to salt and drought stress in several plant species (Tang and Chu, 2017). In G. stocksii, interactions between miR164 and NAC1 and NAC100 were found (Figure 1m), while expression of NAC100 was not significantly altered in drought‐stressed G. stocksii plants. Moreover, comparative transcriptomics analyses also revealed that similar to G. arboreum, the expression of NAC1 in G. hirsutum was not induced by PEG‐stimulated drought stress (Zhang et al., 2015). Taken together, these results suggested that miR164:NAC1 module plays a species‐specific role in drought response in G. stocksii.

In conclusion, this work provided a high‐quality reference genome of G. stocksii and identified several vital regulatory elements involved in the response of G. stocksii to drought stress. These findings broaden the current knowledge of cotton biology and contribute to the future use of wild cotton to improve the agricultural traits of cultivated cotton species.

Conflict of interest

No conflict of interest declared.

Author contributions

Y. S. and J. S. conceived the study. D. Y., L. K., D. Z., Y. W., J. M. and Y. S. performed the experiments and data analyses. D.Y., Y. S. and L. K. wrote the manuscript.

Data availability statement

Genome sequence and annotation of Gossypium stocksii, and the raw RNA‐seq data have been deposited in NCBI under the projects PRJNA707270 and PRJNA712942.

Yu, D. , Ke, L. , Zhang, D. , Wu, Y. , Sun, Y. , Mei, J. , Sun, J. and Sun, Y. (2021) Multi‐omics assisted identification of the key and species‐specific regulatory components of drought‐tolerant mechanisms in Gossypium stocksii . Plant Biotechnol. J., 10.1111/pbi.13655