**4. Materials and Methods**

#### *4.1. Plant Materials and Treatments*

Three cassava genotypes, W14, SC124, and Arg7, were planted in the greenhouse of the Chinese Academy of Tropical Agricultural Science (Haikou, China). Their characteristics were described in our previous studies [38,55]. Stem segments containing three nodes were cut from eight-month-old cassava plants and planted in pots, as described in Hu's study [51]. The transcripts of W14 and Arg7 *MePOD* genes in stems and leaves after being planted for 90 days and middle storage roots after being planted for 150 days were examined by RNA-Seq. After W14, SC124, and Arg7 were cultured for 90 days, they were subjected to drought stress by withholding water for 12 days, after which their leaves and roots were sampled to study the transcriptional responses by RNA-Seq. To examine the expression profiles of *MePOD* genes after the plants were exposed to stress and related signaling treatments, the 60-day-old Arg7 variety was treated with 100 µM MeJA for 0, 2, 6, 10, and 24 h; 300 mM NaCl for 0 h, 2 h, 6 h, 3 days, and 14 days; a low temperature (4 ◦C) for 0, 2, 5, 15, and 48 h; 100 µM SA for 0, 2, 6, 10, and 24 h; 200 mM mannitol (to induce osmotic stress) for 0 h, 2 h, 6 h, 3 days, and 14 days; 100 µM ABA for 0, 2, 6, 10, and 24 h; 10% H2O<sup>2</sup> for 0, 2, 6, 10, and 24 h; or *Xam* for 0, 2, 6, 12, and 24 h. Ten-month-old cassava storage roots (CSR) were cut into 5-mm-thick slices and placed into Petri dishes containing wet filter paper for 0, 6, 12, and 48 h to study the expression changes in *MePOD* genes via RNA-Seq during CSR deterioration. All samples were frozen immediately in liquid nitrogen and stored at −80 ◦C for RNA-Seq and qRT-PCR.

#### *4.2. Identification and Phylogenetic Analysis of PODs in Cassava*

*MePOD* genes were identified in cassava on the basis of homology with 73 POD protein sequences from the *Arabidopsis* genome database (available online: http://www.arabidopsis.org/index.jsp) and 138 POD protein sequences from the rice genome database (available online: http://rice.plantbiology. msu.edu/index.shtml) [56,57]. The Hidden Markov Model-based search (HMMER: http://hmmer.wustl. edu/) profile of these confirmed POD proteins was constructed to search the cassava genome hub (available online: http://www.phytozome.net/cassava.php) [58]. Finally, all predicted POD protein sequences were further examined by CDD (available online: http://www.ncbi.nlm.nih.gov/cdd/) and PFAM (available online: http://pfam.sanger.ac.uk/) after being checked by BLAST analyses [59,60]. All the predicted cassava POD genes identified from HMMER and BLAST were confirmed only if they included the POD special domains examined by SMART (available online: http://smart.emblheidelberg.de/) [61]. Multiple sequence alignment of all predicted MePOD protein sequences was performed with Clustal W in BioEdit software [62]. The phylogenetic tree of the full-length MePOD protein sequences was created using MEGA 5.0 (University College Dublin, Dublin, Ireland) with the neighbor-joining (NJ) method, and bootstrap analysis was conducted with 1000 replicates [63].

## *4.3. Protein Properties and Structure Analyses of PODs in Cassava*

The ProtParam database (available online: http://web.expasy.org/protparam/) was used to predict the properties, including amino acid numbers, molecular weights (MW), and isoelectric points (pI), of all presumed POD proteins [64]. The motifs were analyzed using the MEME program (available online: http://meme-suite.org/tools/meme), in which the maximum number of motifs was set to 10, the optimum width of motifs was set to 15–50 amino acid residues, and the other settings were kept at default values [65]. Subsequently, these 10 motifs were annotated in InterProScan (available online: http://www.ebi.ac.uk/Tools/pfa/iprscan/) [66]. The gene structures of each MePOD were investigated using GSDS software (available online: http://gsds.cbi.pku.edu.cn/) using each MePOD's genomic DNA sequence and its corresponding CDS sequence, which were retrieved from the cassava genome database [67].

#### *4.4. Chromosomal Location and Duplication Pattern Analyses*

According to the results of BLASTN in the Phytozome 12.0 cassava database, *MePOD* genes were mapped to different chromosomes. On the basis of the calculated value of nucleotide sequence similarity and the phylogenetic relationship of cassava *POD* genes, paralogous genes were identified. The gene duplication pattern of paralogous *MePOD* genes was determined by the following two criteria: (1) the identity of the aligned region was >90% and (2) the alignment covered >90% of the longer gene. Circos software (Canada's Michael Smith Genome Sciences Center, Vancouver, Canada) was used to draw the duplication events of *MePOD* genes [68,69]. The values of nonsynonymous substitution (Ka) and synonymous substitution (Ks) were calculated suing DnaSP 5.0 software [70]. Ka/Ks rate > 1 indicates positive evolution, Ka/Ks rate = 1 indicates neutral evolution, and Ka/Ks rate < 1 indicates negative evolution [71].

## *4.5. Transcriptome Analyses of PODs in Cassava*

RNA-Seq was used to determine the expression of cassava *MePOD* genes. Total RNA was isolated from frozen stems, leaves, roots, and storage roots using the plant RNeasy extraction kit (TIANGEN, Beijing, China) and quantified with a NanoDrop 2000c (Thermo Scientific Inc., Waltham, MA, USA). Total RNA (3 µg) of each sample was used to construct the cDNA library according to the Illumina instructions and then sequenced using an Illumina GA II (Illumina Inc., San Diego, USA). The original data processing and analysis methods were described in our previous study [72].

## *4.6. Quantitative Real-Time PCR Analyses*

Leaf samples of the Arg7 variety subjected to MeJA, SA, ABA, NaCl, low temperature, mannitol, H2O2, or *Xam* treatments were collected to perform qRT-PCR analysis. Total RNA (1 µg) from each sample was used to synthesize the first-strand cDNA using oligo-dT primer by SuperScript reverse transcriptase (Takara, Dalian, China). The cDNA product was diluted to 50 ng·µL −1 , and 1 µL was used for qRT-PCR. The qRT-PCR reaction mixtures (20 µL) contained 0.6 µL of each gene-specific primer (300 nmol·µL −1 ), 10 µL of 2× FastFire qPCR PreMix (Tiangen, Beijing, China), and 7.8 µL of RNase-free water. The qRT-qPCR thermal cycling included cDNA denaturation at 95 ◦C for 1 min, with 40 cycles of 95 ◦C for 5 s and 60 ◦C for 15 s in the Mx3005P Real-Time PCR System (Agilent Inc., Palo Alto, CA, USA) with the SYBR green method. The β-tubulin gene of cassava was chosen as an internal control. All qRT-qPCR experiments were performed in triplicate, and the gene-specific primers

used in expression analysis are listed in Table S7. The data obtained from the qRT-qPCR were analyzed with Tukey's post-hoc ANOVA in SPSS 22.0 (SPSS Inc., Chicago, USA) (*P* < 0.05) after fold treatment with the 2−44*C*<sup>t</sup> method.
