A hybrid cultivar of Broussonetia kazinoki × B. papyifera [1,11] was obtained from the Institute of Botany, Chinese Academy of Sciences (116° 10′-116° 20′ E, 39° 55′-49° 5′ N) and used in the study. The seedlings with uniform growth were cultivated using the cuttings propagation technique which resulted in the plantlets with identical tissue-culture batches. The plants were cultivated for 1 month in a greenhouse (day natural light: about 5500 lx; day/night temperature: 25°C/20°C; relative humidity: 75%). Thereafter, stems with original length and diameter of 25–35 cm and 2–4 mm, respectively, were trimmed separately into 10 cm segments to simulate the field harvesting and trigger re-sprouting.

After growing for 2-week, healthy plants were divided into six groups and exposed to different salt concentration gradients. A half-strength modified Hoagland’s solution was used as a normal nutrient solution, NaCl was added to prepare 0, 50, 100, 200, 300, and 400 mmol/L solution that was coded as CK, Y50, Y100, Y200, Y300, and Y400, respectively. Based on the growth performance and previous experience, the CK, Y50, and Y300 were selected for further studies. The control, and moderate and high salt treatment groups were designated as samples CK, B1 and B2, respectively. For every group, there were three replicates (i.e., 1, 2 and 3) and 20 plants were used per replicate. The nutrient solution with the corresponding NaCl concentration was irrigated for 12 days to maintain the water content of the matrix at approximately 60% to 90%. Significant phenotypic differences were observed after 12 days of salt stress, which was followed by transcriptome and physiology analysis. In this study, perlite was used as pot media to grow the paper mulberry plants. All the other management practices were kept similar except NaCl application.

The leaf samples were collected from the salt-treated and control groups, and the plant RNA Kit (Aidlab China) was employed to extract the total RNA following the specific protocols. The quality of RNA samples was determined using Agilent 2100 Bioanalyzer. For quantification of extracted RNA, NanoDrop Technologies ND-2000 was used followed by the elimination of genomic DNA (gDNA) with Takara DNase I. Each experiment consisted of three biological replicates.

The RNA-Seq library was constructed using an Illumina TruSeq™ sample preparation kit (Illumina, San Diego, CA, USA), following the manufacturer’s instructions. To isolate polyA mRNA from total RNA, oligo (dT) was used, followed by fragmentation. Furthermore, a synthesis kit (Invitrogen, CA, USA) was utilized to synthesize the double-stranded cDNA, and the above-described fragments were utilized as templates. Then, agarose gel electrophoresis (AGE) was performed to select the target fragments, followed by amplification with NEB Phusion DNA polymerase. Finally, to process the eventual library, the Illumina HiSeq platform was employed. Each experiment consisted of three biological replicates.

The low-quality and adaptor reads were removed from the raw transcriptome data to obtain clean reads. Following preprocessing, the obtained clean reads based on SS-treatments (BP1, BP2) and control (CK) samples were subject to de novo assembly to acquire the unigenes with Trinity [27,28]. The BLASTX software was used to search all the unigenes for NCBI NR, SwissProt, COG, and KEGG public databases with an E-value of less than 1.0 × 10⁻⁵. The BLAST2GO (https://www.blast2go.com/) program [29], and KEGG database (http://www.genome.jp/kegg/) [5] provided the Gene Ontology (GO) terms and KEGG pathways enriched by those assembled unigenes.

To measure the transcript expression levels, fragments per kilobase of exon per million mapped fragments (FPKM) values were calculated. Additionally, RSEM (http://deweylab.biostat.wisc.edu/rsem/) was applied in calculating gene expression [30]. Further, the R package EdgeR (Empirical Analysis of Digital Gene Expression) software was used for differential analysis of transcript expression between SS-treatments and CK samples [31]. A significant difference was screened upon the thresholds of |log2FC (CT/CK)| ≥ 1 and false discovery rate (FDR) < 0.05. Then GO analysis on differentially expressed genes (DEGs) was conducted using KOBAS (http://kobas.cbi.pku.edu.cn/expression.php) and Goatools (https://github.com/tanghaibao/Goatools) [32]. Furthermore, the same method was used to analyze KEGG pathways enriched by those identified DEGs. The DEGs were considered to be significantly enriched when a Bonferroni-corrected p-value ≤ 0.05 was obtained.

The expression level of DEGs was measured by RT-qPCR to validate the transcriptomic data. The reaction was performed using specific primers designed by the Platinum SYBR Green qPCR Supermix-UDG (Invitrogen) and Light Cycler 480 (Roche Diagnostics, Germany) using the specific protocols. Three technical replicates were set for each sample, while the amplification specificity was verified based on the dissociation curves. The 2^−ΔΔCT method [33] was used for calculating the relative expression of the genes, and EF1 was used as the reference gene [25,19] (Supplementary Table S3).

Paper mulberry plants showed visible morphological changes in their growth and development. The paper mulberry plants where NaCl was applied had resulted in drying and drooping of leaves. Additionally, the entire plant showed apparent phenotypic damage as the duration of SS treatment was increased. In response to 400 mmol/L NaCl stress, plants exhibited severe leaf wilting, along with drying of entire leaves or leaf margins (Fig. 1F). Furthermore, complete leaf withering was observed on the 12-day of SS treatment at 400 mmol/L NaCl stress (Fig. 1F).

In this study, total leaf RNA was isolated from SS-treated (NaCl) and CK (control) paper mulberry plants. The Illumina platform was employed to perform the RNA-sequencing (RNA-seq) of three replicates in SS-treated and CK groups that were named B1–1, 2, 3, B2–1, 2, 3, and CK-1, 2, 3, respectively. For the control CK samples, 7.76, 8.64, and 7.93 Gb clean data was obtained (Table 1), whereas, from SS-treated samples, 8.31, 7.84, 8.20, and 7.10, 8.77, and 8.49 Gb clean data was obtained (Table 1). Among those nine samples, GC concentration was around 47.51%. Similar data were obtained from the biological replicates (Table 1). Further, the clean transcriptomic data were assembled with the Trinity program for these nine samples [28]. A total of 45,550 transcripts were obtained from the initial assembly, and the N50 was 2,243 bp. The mean length of these transcripts was 1,200 bp, along with the longest one being 17,035 bp (Table 2 and Supplementary Table S1). Meanwhile, a total of 35,068 unigenes (N50 = 2261) were obtained collectively, with the length ranging from 201 to 17,035 bp and the average size being 1,055 bp (Table 2).

As illustrated in Fig. 2a, many unigenes (35,068; 93.25%) had a length of <3000 bp. In general, the number of unigene reduced with an increase in gene length. Among the obtained unigenes, most have a length of 1,000–2,000 bp (17,035) which suggested a high-quality assembly. All the raw sequence reads were imported into the NCBI short read archive (SRA) (BioProject accession number, PRJNA516316).

To identify the functions of unigene in paper mulberry, the multiple complimentary annotation approaches were used, and the assembled unigenes were searched in SwissProt, NCBI non-redundant (Nr), GO, KEGG, KOG, COG, and Pfam databases using BLASTX algorithm, and the E-value was set at <1.0 × 10⁻⁵. Hence, the known proteins were identified in one or more databases corresponding to 23,421 unigenes (Fig. 2a, Table 3, Supplementary Table S5, and Supplementary Fig. S1). Among these unigenes, 15,047 showed apparent hits (E-value <1e⁻⁵) within the GO database, whereas 22,884 were within the Nr database. Thus, the known proteins were identified for 17,608, 7,199, and 15,658 unigenes in SwissProt, COG, and KEGG databases, respectively, whereas 11,647 unigenes remained unmatched with sequences from 7 databases.

A BLASTX similarity search analysis was performed based on the Nr database, and the E-value was set to <1e⁻⁵. The greatest annotation proportion (n = 22,884 unigenes) was observed in the Nr database compared with the remaining six databases. Altogether, these unigene sequences showed the closest BLASTx matching against gene sequences to Morus notabilis (57.39%), Trema orientale (5.71%), Arabidopsis thaliana (5.41%), and Parasponia andersonii (3.17%) (Fig. 2b). The paper’s mulberry unigenes were similar to Morus notabilis genes, which indicated that Morus notabilis genomes and transcriptomes may be used as a reference in further research.

The GO analysis was conducted to annotate the unigene functions related to paper mulberry. A total of 15,047 unigenes belonging to 55 functional subgroups were classified collectively (Fig. 2c) and were divided later into 3 GO categories using BLAST2GO [34]. These unigenes were related to biological processes (BPs), molecular functions (MFs), and cellular components (CCs), which were useful for gene annotation. For BP terms, these genes were mainly enriched into the metabolic process (7.490%), cellular process (6.686%) as well as multi-organism-process (2.340%). Furthermore, those genes were mostly enriched into catalytic activity (7.446%), binding (7.234%) together, and transporter activity (8.090%), related to MF terms. The genes associated with CC terms were mainly ‘cell’ (6.656%), ‘cell part’ (6.620%), and ‘organelle’ (2.192%).

In addition, the COG database was searched to predict and classify the unigene functions related to paper mulberry. Under the assumption that every protein from the COG database exhibited independent evolution from a certain ancestral protein, 7,199 unigenes were divided into 25 COG classification subgroups (Fig. 2d). Among these subgroups, post-translational modification, protein turnover, and chaperones (6.480%) occupied the greatest percentage, translation, ribosomal structure, and biogenesis were 5.990%, signal transduction mechanisms were 1.790%, and general function prediction was only 1.108%. The remaining unigenes could be classified as function unknown (2.430%) category. Nonetheless, there were few clusters in nuclear structure (n = 2 unigenes) and cell motility (7 unigenes) categories, separately. Meanwhile, there was no unigene classified under the extracellular structures category.

Further, the KEGG protein database was used to categorize paper mulberry unigene functions with an emphasis on biochemical pathways. For identifying the aforementioned pathways related to paper mulberry, KEGG analysis was conducted for the identified unigenes. As a result, the identified 15,658 unigenes were enriched into 143 KEGG pathways (Supplementary Fig. S1 and Tables 3 and 4 and Supplementary Table S6). Among them, carbon metabolism, plant-pathogen interaction, starch and sucrose metabolism, ribosome and amino acid biosynthesis pathways were most significantly enriched. Thus, these findings indicate that SS tolerance of paper mulberry mainly depends on the metabolism of large amounts of metabolites and energy-related processes.

The DEGs between SS-treated and CK samples were examined with de novo assembled transcriptome considered as reference. To comprehensively explore the levels of transcripts associated with SS tolerance of paper mulberry, DEGs were detected between SS-treated and CK paper mulberry samples. A total of 2,126 DEGs (selected upon |log2FC (CT/CK)| ≥ 1 and FDR < 0.05 thresholds) were detected between SS-treated and CK groups using edgeR software. Among these DEGs, 812 genes were up-regulated and 1,314 genes were down-regulated (Fig. 3) in the leaves of paper mulberry exposed to NaCl stress conditions (Supplementary Figs. S2 and S3).

The KEGG database was used to identify the significantly changed pathways (p ≤ 0.05) in the leaves of paper mulberry exposed to SS-treatment. As a result, 15 significant KEGG pathways were discovered (Table 4), including plant-pathogen interaction, plant hormone signal transduction, fatty acid metabolism, starch and sucrose metabolism, and MAPK pathway, which displayed the highest significance.

Most DEGs were enriched into the plant hormone signal transduction pathway (ko04075), which suggests that plant hormones have critical functions in resistance against SS in Paper mulberry. In addition, the plant-pathogen interaction pathway (ko04626) suggests the vulnerability of SS-challenged plants to pathogens. However, based on the MAPK pathway (ko04013), endogenous gene levels within SS-challenged plants are modulated via diverse signaling molecules to adapt to the SS environment. The fatty acid metabolism (ko01212), starch, and sucrose metabolism (ko00500) pathways were significantly enriched, since SS exposure results in altered plant metabolism for improving salt tolerance [35].

To explore the functions of DEGs under SS conditions, GO analysis was performed using Goatools (p ≤ 0.05 by Fisher exact test). The results showed that DEGs were mostly associated with signal transduction, lipid metabolic process, hydrolase activity, carbohydrate metabolic process, oxidoreductase activity, response to stimulus, oxidation-reduction process, chloroplast, single-organism metabolic process, and membranes (Table 5). These findings suggest that DGEs associated with these pathways may also be important for the paper mulberry plants to adapt under NaCl stress conditions.

Transcription factors have a major role in the regulation stress-responsive genes. According to a report, TFs have a critical impact on salt stress signaling and trigger diverse downstream transcripts [12]. Based on the Arabidopsis database-predicted TFs belonging to 81 families, several genes encoding TFs belonging to 15 families were detected and were differentially expressed in response to NaCl stress in the leaves of paper mulberry (Supplementary Table S2–S4). The major transcription factor families expressed in this study include bHLH, MYB, NAC, bZIP, and WRKY.

In plants, auxin (indole-3-acetic acid) plays a critical role in growth and development processes such as cell division, leaf expansion, lateral root initiation, and response to stress [36]. According to the results of this study 16 genes related to the auxin pathway were detected and differentially expressed (Table 6); among these one gene encoded the auxin transporter-like protein (AUX), 2 were auxin-responsive GH3-like protein (GH3) genes, 4 were auxin-responsive factors (ARF), and 6 were auxin-induced protein IAA6 (AUX/IAA) genes. These genes were up-regulated under salt stress conditions in the leaves of paper mulberry plants. Moreover, 3 unigenes were also discovered that encoded SAUR; among them, 2 were up-regulated, whereas 1 was down-regulated. These observations suggest that the resistance of paper mulberry to salt stress may be associated with the auxin signaling pathways.

Kinases were previously reported to have critical functions in signaling that is achieved by protein phosphorylation to relay signals [37,1]. In SS-treated paper mulberry plants, 117 genes encode protein kinases that showed diverse expression patterns (Table 4). Additionally, based on the results of KEGG analysis it was observed that SS-treatment induced MAPK pathway (ko04013). Therefore, the abundant MAPK signaling-related genes that were identified from the paper mulberry transcriptome are potential candidate genes and seem to be involved in salt stress adaptation.

The metabolic processes in plants are changed under salt-stress conditions. According to the result of KEGG analysis, most of the DEGs were related to biosynthesis and metabolism. In this study, the changes in the secondary metabolites were observed when paper mulberry plants were exposed to salt stress; these changes include alteration of flavonoid biosynthesis (ko00941) and phenylpropanoid biosynthesis (ko00940) pathways. Overall, 2 pathways were mainly enriched by nine DEGs (Table 7) under salt stress conditions. Specifically, 2 DEGs that encoded flavonol synthase were enriched in the flavonoid biosynthesis pathway. The expression of one of these was up-regulated and one was down-regulated. Additionally, several genes enriched in the phenylpropanoid biosynthesis pathway showed down-regulation of expression. However, the mechanism of paper mulberry resistance to salt stress had a complicated process and involves several metabolic pathways. Therefore, it is difficult to reach conclusions only based on transcriptome data analysis because this clarifies part of the entire process. Thus, further precise molecular biology, proteomics, and genomic studies are required to clearly explain and verify the relevant results.

To validate the reproducibility and accuracy of RNA-Seq data, transcript levels were analyzed for 15 selected DEGs by qRT-PCR. According to the RNA-Seq data, out of these 15 unigenes, the expression of 14 unigenes was up-regulated and one unigenes was down-regulated. The results showed that salt stress-induced changes in the expression of tested genes in the leaf of paper mulberry were similar for RNA-Seq and qRT-PCR data. Thus it was concluded that the qRT-PCR analysis was consistent with RNA-Seq analysis and this confirmed the reliability of RNA-Seq data (Fig. 4, Supplementary Fig. S4).