A pot test was carried out in the glasshouse of Shanghai Jiao Tong University, China, by using 1-year-old S. superba saplings. For this purpose, 18 saplings, with same growth potential (about 35 cm high), were taken and randomly allocated into two groups: control (CK) and Mn concentration 100 mM (WT), in three replicates. Plant tissues were collected after 1, 5, and 10 days of transplantation in quarter-strength of Hoagland solution. The tissues were collected in liquid nitrogen and processed through the isolation of RNA at −80°C.

RNA extraction from the collected plant samples (0.2 g leaves) was performed with the RNeasy plus Mini Kit (Qiagen, Valencia, United States). Using the Nano drop ND-1000 (Nano drop Technologies, Rockland, DE, United States) spectrophotometer, its quality and quantity were calculated after observing RNA on agarose gel. By using Qubit^® RNA Assay Kit and RNA Nano 6000 Assay Kit, RNA quality has been identified clearly.

For sequencing, C2 sequencing reagents were used in Pacific Biosciences (PacBio) real-time sequencer. Purified RNA was utilized for synthesizing cDNA by SMRT PCR cDNA Synthesis Kit (Clontech, CA, United States). Full-length cDNA of different sizes was selected and cDNA libraries were constructed using the BluePippin^® (SageScience, Beverly, MA, United States). After BluePippin screening, the fragments were pacified to large-scale PCR to get sufficient total cDNA and evaluate using Qubit fluorometer (Life Technologies, Carlsbad, CA, United States). The libraries’ uniqueness was maintained by using Agilent Bio analyzer 2100 system and SMRT sequencing was achieved.

Sequence statistics were obtained by using the SMRT link 5.1 software. Relevant parameters such as no-polish TRUE, min-length 50, max-drop-fraction 0.8, min-predicted-accuracy 0.8, min-passes 2, min-z score -9999.0, and max-length 15000 were followed to produce CCS by subreading BAM files. The output was CCS.BAM files, which, using pbclassify.py, were categorized into full length and non-full length reads. Full and non-full length FASTA files were fed into the cluster stage, and additional nucleotide errors were corrected by LoRDEC software in the consensus readings. Unnecessary data in the corrected consensus reads were separated using CD-HIT (-c 0.95 -T 6 -G 0 -aL 0.00 -aS 0.99) to acquire final transcripts for the subsequent investigation.

We recognized functional annotations matching by searching to the non-redundant nucleotide database (Nr), Swiss-Prot, Pfam, non-supervised orthologous groups egg (NOG), clusters of orthologous groups (COG), eukaryotic orthologous groups (KOG), Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and BLAST alignment databases, and all transcript sequences were examined for homology (Bairoch and Apweiler, 2000; Li et al., 2002; Tatusov et al., 2003; Kanehisa et al., 2004).

Plant transcription factors were predicted using iTAK v1.7a (Zheng et al., 2016). Four tools, CNCIv2 (Sun et al., 2013), CPCvcpc-0.9-r2 with e-value “1e-10” (Kong et al., 2007), Pfam-scan (E 0.001 -domE 0.001) (Finn et al., 2016) and PLEKv1.2 with min length 200 (Li et al., 2002) were chosen to predict candidate lncRNAs. Transcripts predicted with coding capacity by all of the mentioned above tools were separated out, and those without coding prospective were considered as candidate set of lncRNAs.

Simple sequence repeats were identified by MISA v1.0¹ (Beier et al., 2017), with default parameters. MISA can recognize seven types of SSR, namely mono, di, tri, tetra, penta, and hexa nucleotide by analyzing transcript sequences. By using tool Batch primer 3, SSR primers were designed. DNA was extracted from S. superba for PCR amplification, and the product was extracted in 8% polyacrylamide gel.

All 36 saplings of control group (CK) and Mn-treated group (WT) were utilized by RNA extraction via RNeasy plus Mini Kit (Qiagen, Valencia, United States). The cDNA was synthesized using SMRT PCR cDNA synthesis kit (Clontech, United States). The qRT PCR was performed to see the expression of nine different transporter genes (Supplementary Table S5). The reaction mixture consisted of 7.8 μl of cDNA and 10 μl of 2× qPCR. The process involve initial denaturation at 95°C for 10 min, which is followed by 45 cycles of denaturation at 95°C for 15 s, while annealing and extension at 60°C for 60 1 min.

Plant tissues were used for RNA extraction and library construction of cDNA. After removing adaptor sequences, a total of 1,311,552,27 low-quality sequences were obtained. It comprised of 1–6 kb data. After self-correction of subread sequences (with minPasses = 39), overall 588,479 circular consensus sequence (CCS) were obtained. Moreover, 511,759 full-length non-chimeric (FLNC) picks were also identified. Iterative sequence clustering revealed 167,782 consensus isoforms, with a mean read length of 2187 bp. Together with non-FL sequence, the Quiver program corrected the consistent sequence in each cluster, yielding 163834 high-quality transcripts (high- quality isoforms) with more than 97.65% accuracy. In addition, the low-quality record was corrected with the corresponding sequencing data to enhance the accuracy of the isoforms. Finally, using CD HIT (3.3) tools, 97,520 bar non-redundant transcript sequences were obtained (Tables 1, 2 and Figure 1).

About 93,362 ORFs were predicted using Trans Decoder software. Overall, 78,255 complete ORFs were recognized, and the complete ORFs were successfully analyzed for length distribution (Figure 2); 7,489 AS events were revealed out of all transcripts obtained by SMRT sequencing (Supplementary Table S1). Owing to the lack of an unavailable S. Superba reference genome, the classification of the forms of AS events in future research would be justified (Figure 2).

These are a group of non-coding poly-A RNAs that are involved in growth and stress response of the plant. In this work, we utilize four computational methods to recognize lncRNAs, associated with Pfam, CPC, CPAT, and CNCI databases. Overall, 8,551, 9,174, 20,720, and 18,669 lncRNAs were recognized in the Pfam, CPC, CPAT, and CNCI, respectively (Supplementary Table S2). By filtering the record of less than 300 bp, 2,011 transcripts were examined as lncRNAs by all four mechanisms (Figure 3).

These are the key regulators of gene expression and play a significant role in plant growth and development. In this study, TFs are divided into 64 families, and 9423 putative TFs were identified (Supplementary Table S3). TFs in S. superba transcriptome were mostly related to the RLK-Pelle-DLS (473,5.02%), bHLH (278,2.95%), MYB-related (230,2.44%), CAMK-CDPK (218,2.31%), C2H2 (218,2.31%), C3H (202,2.14%), RWP-RK (189,2.01%), bZIP (177,1.88%), TKL-PI-4 (159,1.69%), and NAC (152,1.61%) families (Figure 4).

All 16,3834 unique SMRT transcripts were practically annotated by seven data storage, such as GO, KOG, Pfam, Swiss-Prot, COG, Nr, and KEGG using BLAST (7) software [version 2.2.26] (Table 3). By comparing the transcript sequence to NR with homologous species, 68,630 genes were annotated (Figure 5).

Transcripts GO classification statistics demonstrated 71,839 unique genes, which were enriched in major categories of molecular function, cellular component, biological component, and catalytic activity (Figures 3, 5). This analysis also helped us to get transcripts COG classification statistics (Figure 6).

To examine the functional annotation and classification of S. Superba for further study, all the transcripts were checked against the COG database clusters². This investigation indicated that 38,914 transcripts were allocated into 24 categories (Figure 7). The highest group was general function prediction only (4,690, 11.1%), followed by signal transduction (4,291, 10.15%) whereas carbohydrate transport and metabolism was 3,981, 9.42%. Our results revealed six groups percentage were less than 1.00%, i.e., nuclear structure and modification, chromatin structure and dynamics including extracellular structure (Figure 7).

KEGG data storage interpreted a total of 67,426 sequences and plotted 367 operative categories in S. superba. Among them, metabolism was the largest category. The functional annotations of all 78,559 unique transcripts were detected in this work (Supplementary Table S4). A significant number of genes, especially interrelated in salt-tolerance and fatty acid component of S. superba were annotated, such as oxidative phosphorylation (1073), plant hormone signal transduction (506), fatty acid biosynthesis (246), biosynthesis of unsaturated fatty acids (94), and α-linolenic acid metabolism (199). We also recognized matches to our unique transcripts in clusters of orthologous class of proteins (COG) (44,376, 56.49%), Pfam database (41,535, 52.87%), and Swiss-Prot (58,535, 74.51%) (Table 4).

For the expression analysis of potential high expression genes, nine transporter genes were selected randomly for qRT PCR. Results of qRT PCR analysis showed the highest expression by Gene 6 (Na_Ca_ex), in Mn-treated group (WT) on day 1, 5, and 10, while the expression of CAX (CAX-interacting protein 4) was higher in Mn-treated group (WT) at day 5 (Figure 8).

After the screening of 95,258 obtained transcripts, about 58,396 possible SSRs were recognized from 29,075 transcripts. Among them 26,312 composed one SSR, and 19,007 contained two loci or more. In addition, 29,075 and 18,324 transcripts contained one SSR and at least two SSRs, respectively. Whereas observed compound formations were 18,324 SSRs. As shown in Table 5, the number of mono-, di-, tri-, tetra-, penta-, and hexa-nucleotide repeats were 44,045, 51,716, 14,521, 1,675, 753, and 988, accordingly. SSRs with 10 repeat units (16949, 14.9%) were the most abundant, followed by those with 6 (12563, 11%), 11 (10564, 9.2%), and 7 (7995, 87.03%) (Table 6).

By using Primer 3.0 software, 95,374 pair of primers were designed, and among them, 200 were randomly selected for PCR (Supplementary Table S6). The product of 180 PCR pairs was examined successfully, with an efficiency rate of 90%. However, the remaining 20 pairs of primer failed to amplify at various annealing temperatures.