Role of IAA and Primary Metabolites in Two Rounds of Adventitious Root Formation in Softwood Cuttings of Camellia sinensis (L.)

Abstract

The type of adventitious root formation of tea softwood cuttings is different from that of single node cuttings. In addition to the callus at the base of cuttings, the adventitious roots are also formed in the upper cortex of softwood cuttings. In order to find out the similarities and differences between the above two types of adventitious roots of softwood cuttings and the influencing factors for the differences, an integrated analysis of plant hormones, untargeted metabolomics, and transcriptomics of the softwood cutting stems at different positions is performed. The phytohormone results show that IAA plays a leading role, and a high ratio of in vivo auxin to GA3 or ABA facilitated root formation. The ratios of IAA/GA3 and IAA/ABA in the upper and base rooting parts of cuttings are both higher than the middle non-rooting transition zone. Differences in metabolites indicate that 73 metabolites are involved in the formation of adventitious roots in cuttings. Compared with the middle non-rooting transition zone, most saccharides are downregulated in the upper and base rooting parts of softwood cuttings, and the saccharides in the base rooting parts of cuttings are more consumed than that in the upper rooting parts. Most organic acids in the callus at the base of cuttings show an upward trend, while those in the upper rooting parts show a downward trend. Furthermore, coniferyl alcohol is the key metabolite for adventitious root formation in the upper and base rooting parts of cuttings. Transcriptome results show 1099 differentially expressed genes (DEGs) are obtained, and KEGG enrichment analysis show that these DEGs are significantly enriched in phenylpropanoid biosynthesis, plant hormone signal transduction, and flavonoid biosynthesis pathways. Based on weighted gene co-expression network analysis (WGCNA), two key modules which have a highly positive correlation with IAA are identified. In summary, maintaining the balance of endogenous hormones and sufficient nutritional elements is very important for adventitious root formation in tea plants. Clarifying the material basis of softwood cutting rooting of tea plant is of great significance to improve the cutting survival rate of tea plant (especially difficult rooting varieties) and shorten the breeding cycle.

1. Introduction

Tea cultivars (Camellia sinensis (L.) O. Kuntze.) are popularly propagated by cuttings in China. According to the length of cuttings, tea cutting can be divided into softwood cuttings and single node cuttings. Among them, the single node cuttings are about 3 cm long, with a strong and full axillary bud and a mature leaf, and have higher reproduction coefficient. It is a reproduction method commonly used in major tea producing countries in the world. Shandong tea area is the tea area with the highest latitude in China (35–37° N), with cold and long winters. It takes more than one year for single node cuttings to be transplanted from cutting to out of the nursery. The single node cuttings hardly survive the winter in the Shandong tea area, the survival rate is low, and the application is greatly limited. The cutting branches selected for tea softwood cutting are semi lignified, and the length is greater than or equal to 15cm; which can form a complete root system in about 9 weeks. Due to the short seedling raising cycle, softwood cutting has been widely used and popularized in Shandong tea area.

The growth and development of adventitious root (AR) is the key for successful cutting propagation. It is generally believed that there are three types of plant cutting rooting: one is the callus rooting type, in which callus is formed at the cutting notch first, and then ARs are differentiated from callus, such as Drimys brasiliensis [1], Buxus sempervirens L. [2], etc. The second is the cortex rooting type, which does not form callus, and the ARs are directly formed by cells in the cambium, such as mulberry [3,4], etc. The third is the comprehensive rooting type, which merges the first two rooting types, such as poplar. The ARs of conventional mulberry hardwood cuttings are mainly the callus rooting type.

The rooting of tea plant single node cuttings is a typical callus rooting type. After the single node cuttings are inserted into the soil, the callus cork plasma membrane is first formed on the cut surface of the single node cuttings, and then the callus grows. The root primordia are extended from the lenticels or between the cuttings bark and the callus of the cuttings, then become young roots. Tea plant softwood cuttings include two types of rooting. The rooting type at the base incision of the softwood cuttings is the same as that of the single node cuttings, which is a callus rooting type and belongs to the indirect organogenesis pathway. The AR primordia of the green branches on the upper part of the cuttings are formed in the phloem parenchyma cells that on the inner side of the central cylindrical sheath and do not need to go through the stage of callus, which belongs to the direct organogenesis pathway. Softwood cuttings can shorten the seedling raising cycle because the cuttings have new shoots, and the root system is divided into upper and lower rounds. After ARs are formed, they can be out of the nursery, which shortens the growth time of new shoots. The characteristics of two-wheeled rooting make the roots of softwood cuttings have more layers and more developed than single node cuttings, and the contact area of ARs with the soil is enlarged, and it is easier to absorb water and mineral elements in the soil for the growth of cuttings.

The formation of AR is a complex developmental process and in response to stress conditions and regulated by many factors [5,6,7]. The change of endogenous phytohormone content plays an important role in the control of AR formation [8]. Auxin interacts with many plant hormones to mediate developmental processes, such as cell division, elongation, and differentiation [9,10]. In addition to auxin, ABA, zeatin, and SA may play a complementary role in the induction, initiation, and emergence of the developing AR [11]. Phenolic acids are the most common plant secondary metabolites and are involved in controlling adventitious root formation [12]. A close association has been reported between phenolic acids and rooting of Malus ‘Jork 9’ and Eryngium maritimum L. [13,14]. Flavonoids may affect AR development as auxin transport inhibitors or as antioxidants that may protect auxin pools from degradation, whereas carbohydrates are essential for energy metabolism and as biosynthetic units, thereby playing a central role in root development [15]. The number of DEGs involved in glycolysis decreased while the number of DEGs involved in phenylpropanoid biosynthesis increased following the AR formative process [16]. At the same time, the DEGs in the rooting process are also enriched in the plant hormone signal transduction pathway, and hormone signals regulate the formation of ARs [17]. The more obvious genes are related to early auxin response proteins, such as AUX/IAA, SAUR, and GH3. The AUX’s physiological role is to support plant growth and development, for which AUX transportation and signal transduction are essential [18].

At present, there are many studies on tea single node cuttings, but there are few studies on softwood cuttings, especially on two rounds of rooting. The similarities and differences between the two types of Ars of softwood cuttings are not clear. This study aims to clarify the similarities and differences between the two types of Ars in softwood cutting through phytohormone, transcriptome, and metabolome analyses, and analyze the metabolite change regularity and rooting mechanism of the two types of ARs in softwood cutting, so as to provide a theoretical basis for shortening the seedling raising cycle of tea softwood cutting and improving the survival rate of tea plant. At the same time, it has important practical significance for clonal tea breeding technology in North China.

2. Materials and Methods

2.1. Plant Material and Samples

The experiments were conducted in the greenhouse of Rizhao Wulianshan Tea Co., LTD. The materials for softwood cuttings were 15-year-old semi-lignified branches of Camellia sinensis cultivar ‘Shuchazao’ with a length of 20–30 cm. The extra leaves 3–5 cm above the base were removed and cut in a nursery pond with perlite as the growth substrate. The management of the cutting was according to DB 3711/T 96–2018. Before adventitious root formation (3 weeks after cuttings), cuttings were sampled for hormone and transcriptome analysis; after cuttings began to form adventitious roots (6 weeks after cuttings), cuttings were sampled for metabolome analysis. The sampling positions are shown in Figure 1A. The base parts of cuttings are differentiated into Ars (abbreviated as NL), the upper parts of cuttings (underside bark of the leaf axillary buds) were partially formed Ars (abbreviated as NU), and the middle non-rooting transition zone between NL and NU (abbreviated as NM). The obtained samples were immediately frozen in liquid nitrogen and stored at −80 °C for hormone content determination and metabolome and transcriptome analysis. Meanwhile, the samples at 3 weeks after cuttings were made into frozen sections and stained with toluidine blue for histological observation. All the tests were conducted in biological triplicates and 20 plants were taken from each replicate.

2.2. Metabolite Analysis Based on GC-TOF-MS

Metabolites extraction: 50 ± 1 mg sample was transferred into a 2 mL tube, and (methanol/distilled water (v:v) = 3:1) with 10 μL internal standard (adonitol, 0.5 mg/mL stock) were added. Samples were vortexed for 30 s and homogenized with ball mill for 4 min at 35 Hz, followed by ultrasonication for 5 min in ice water. After centrifugation at 4 °C for 15 min at 12,000 rpm, 50 μL supernatant was transferred to a fresh tube. After evaporation in a vacuum concentrator, 40 μL of methoxyamination hydrochloride (20 mg/mL in pyridine) was added and then incubated at 80 °C for 30 min, then derivatized by 60 μL of BSTFA regent (1% TMCS, v/v) at 70 °C for 1.5 h. Samples were then gradually cooled to room temperature.

For GC-TOF-MS analysis, a 7890 system (Agilent Technologies, Santa Clara, CA, USA) equipped with a DB-5 ms capillary column (30 m× 250 μm× 0.25 μm) was used in the splitless injection mode with a helium flow at 1.0 mL/min. The temperature of the GC injector was maintained at 280 °C, and the following column temperature program was employed: initial temperature, 50 °C (hold for 1 min); increased to 310 °C at a rate of 10 °C/min (held for 8 min). The injection, transfer line, and ion source temperatures were 280 °C, 280 °C, and 250 °C, respectively. The energy was −70 eV in electron impact mode. The mass spectrometry data were acquired in full-scan mode with the m/z range of 50–500 at a rate of 12.5 spectra per second after a solvent delay of 6.35 min. The data were normalized before analysis, and the method adopted was internal standard normalization, that is, each metabolite was divided by the peak area of the internal standard metabolite.

2.3. Phytohormone Determination

The concentrations of IAA, ABA, GA, and ZR were determined by enzyme linked immunosorbent assay (ELISA) [19,20,21]. A sample of 0.2–1.0 g was weighed, to which was added 2 mL extraction solution (80% methanol, containing 1 mmol/L BHT), ground into a homogenate in an ice bath, transferred to a 10 mL test tube, shaken, and extracted at 4 °C for 4 h. After centrifugation for 8 min at 3500 rpm, the supernatant was taken and the volume recorded. The supernatant was passed through a C-18 solid phase extraction column and then concentrated and dried in vacuo, and the volume was made constant with the sample diluent. The maximum concentration of the standard curve of IAA and ABA was 50 ng/mL, ZR was 10 ng/mL, and the maximum concentration of GA was 10 ng/mL. The 8 concentrations (including 0 ng/mL) were sequentially diluted twice. The competition conditions were 37 °C and 0.5 h (used 96 Orifice plate). The secondary antibody was added at 37 °C, and incubated for 0.5 h. Color reagent (10–20 mg o-phenylenediamine (OPD) dissolved in 10 mL substrate buffer (5.10 g C₆H₈O₇·H₂O + 18.43 g Na₂HPO₄·12H₂O + 1000 mL distilled water + 1 mL Tween-20, pH 5.0) and 4 μL 30% H₂O₂) was added after washing the plate for color development. The concentration of the standard substance and the OD value of each sample was determined at 490 nm on an enzyme-linked immunoassay spectrophotometer, and the hormone content was calculated according to the logit curve.

2.4. RNA Extraction, Sequencing, and Transcriptome Assembly and Annotation

Total RNA was isolated from the three samples using the RNAprep Pure Plant Kit (Tiangen, Beijing, China). Nine cDNA libraries (each sample with concentration > 20 ng/μL) were built using the PCR-cDNA Sequencing Kit (SQK-PCS109) and PCR Barcoding Kit (SQK-PBK004) and sequenced on the Nanopore PromethION plantform. ONT RNA-Seq raw reads were first filtered and high quality reads obtained, and rRNA were removed after mapping to rRNA database. Then, low-quality reads, short reads, and reads with adaptors were discarded to obtain clean data. Clusters of transcripts were obtained after mapping to reference genome of R. Camellia sinensis with Mimimap2 [22], and consensus isoforms were obtained after polishing within each cluster using the Pinfish package. All unigenes were annotated by BLASTx search against the NCBI, Swiss-Port, KEGG, and KOG/COG databases [23,24,25]. Gene expression levels were calculated as CPM values, and the expression of differentially expressed genes (DEGs) with fold change (FC ≥ 1.5) and p value (p value < 0.01) based on DESeq were screened as differentially expressed. A total of 12 DEGs were selected and examined using RT-qPCR to validate the reliability of the RNA-Seq results.

2.5. Co-Expression Network Construction

Construction of the network was performed using the R package WGCNA to identify modules of highly correlated genes based on the gene data. Firstly, it was assumed that the gene network follows a scale-free distribution, and then the adjacency matrix was further transformed into a topological overlap matrix (TOM), so as to establish a hierarchical clustering tree [26]. The min module size was 50, and the merge cut height was 0.5 to produce different modules. After the preliminary module division, the modules with similar expression patterns were combined according to the module similarity to form a dynamic tree. The subsequent analysis was carried out according to the combined module [27].

2.6. Statistical Analysis

Metabolite data analysis, including peak extraction, baseline adjustment, deconvolution, alignment, and integration, was finished with Chroma TOF (V 4.3x, LECO) software [28] and LECO-Fiehn Rtx5 database was used for metabolite identification by matching the mass spectrum and retention index.

Principal component analysis (PCA) and orthogonal partial least squares analysis (OPLS-DA) were performed using R (3.3.2) package PCA analysis and ropls in R software platform. Then, the different metabolites were screened according to the size of variable contribution value (VIP value) and the significance of component change (p < 0.05), and the R (3.3.2) package cluster profiler was used to enrich the annotation results of KEGG. The data analysis was performed via one-way analysis of variance (ANOVA) with SPSS 18.0 software (SPSS Inc., Chicago, IL, USA). The tables were drawn by Microsoft Excel, and Adobe Photoshop CS6 was used to perform necessary processing and integration of pictures and tables.

3. Results

3.1. Phenotypes and Phytohormones Changes of Different Cutting Nodes

There are obvious differences in anatomical structure among the three parts of the softwood cuttings after 3 weeks (Figure S1). Callus was formed at the base of cuttings, and root primordia both appeared at the upper and base parts (NU, NL) of cutting, but there were almost no differentiated root primordia detected in the middle transition zone (NM) of cuttings. After 6 weeks, two types of rooting appeared on the softwood cuttings of tea plant. The middle transition zone of cuttings was the ordinary stem and no roots. The upper and base parts of cutting were located above and below the middle transition zone, respectively, and both had ARs. There were fewer ARs at NL, but they were thick and long. Compared with NL, the ARs at NU were more vigorous, but shorter, and were mostly distributed on both sides of the stem segment (Figure 1A).

To explore the diverse functions of phytohormones in adventitious roots (AR) formation, the contents of 4 phytohormones in the 3 samples were determined. The results revealed that most phytohormones were remarkably different among the 3 samples. Specifically, 3-indoleacetic acid (IAA), gibberellin A3 (GA3), abscisic acid (ABA), and zeatin riboside (ZR) in NU, NM, and NL had similar change patterns, showing ABA > IAA > ZR ≈ GA3. The contents of IAA and ABA in NU were the highest among the 3 samples. Meanwhile, the content of ABA in NL was significantly lower than the other two. However, the content of GA3 in NM was higher than that in NU and NL (Figure 1B). In addition, the ratio of IAA/GA3, IAA/ABA, IAA/ZR was analyzed, and the ratio of this 4 phytohormones also had obvious differences. Among the 3 samples, the difference in the ratio of IAA/GA3 was the largest. NU and NL were the rooting parts. The ratio of IAA/GA3 in NU and NL was significantly higher compared with NM, while the ratio of IAA/GA3 was significantly higher in NU than in NL, indicating that the ratio of IAA/GA3 was more essential for rooting in NU. Moreover, the ratio of IAA/ZR in NU was significantly higher, which was lower in NL compared with NM. Meanwhile, the ratio of IAA/ABA in NL and NU was significantly higher than in NM (Figure 1C).

3.2. Global Profiling of Metabolites and Expression of Genes Based on PCA

To reveal the material basis and gene regulation level of rooting in different parts of tea plant softwood cuttings, the transcriptomic sequencing and metabolomic analysis of the three samples were performed. At the metabolic level, a total of 454 metabolites were detected from the three samples. The peak areas of those metabolites were used for PCA. The first two principal components explained 88.6% of the total variance (79.0% by PC1 and 9.6% by PC2). The results showed that NL and NU were located in the negative half axis and NM was located in the positive half axis in the direction of PC1, indicating that in the comparison of the three, the metabolites of NL and NU were more similar in species and contents, but the two had a big difference with NM (Figure 2A). At the transcriptomic level, there were nine cDNA libraries from three samples (in three biological replicates per sample), including NL, NU, and NM (Table S1). A total of 53,471 expressed genes was detected in all the samples. The CPM values of these expressed genes were used for PCA. The first two principal components explained 74% of the total variance (47.4% by PC1 and 26.6% by PC2). The results showed that NL formed an independent cluster in the direction of PC2, but the three samples were not distinguished in the direction of PC1, and NU and NM partially overlapped (Figure 2B). The PCA results of metabolome data and transcriptome data showed that NU was similar with NL at metabolic level, but NU was similar with NM at the transcriptomic level.

3.3. Differential Key Metabolites and Expression Profiling of Genes in Different Cutting Nodes

In order to figure out the differences at the metabolic and gene levels of NL, NU, and NM, 73 known differentially accumulated metabolites (DAMs) were identified in NM vs. NL, NM vs. NU, and NU vs. NL, divided into 10 categories. Among them, five categories account for more than 5% of the DAMs, including saccharides (27%), organic acids (19%), alcohols (11%), amino acids and derivatives (10%), and phenolic acids (7%) (Figure 3A). A total of 1099 DEGs were obtained. Between the non-rooting and basal rooting, there were 594 DEGs, of which 386 DEGs were upregulated and 208 DEGs were downregulated. Between the non-rooting and upper rooting, there were 205 DEGs, of which 94 were upregulated and 111 were downregulated. Between the upper rooting and basal rooting, there were 681 DEGs, of which 450 were upregulated and 231 were downregulated (Figure 3B).

Venn diagrams were created to show the distribution of DAMs and DEGs among the three pairwise comparisons of NM vs. NL, NM vs. NU, and NU vs. NL. The 5, 5 and 16 DAMs overlapped between NM vs. NL and NM vs. NU, between NM vs. NU and NU vs. NL, and between NM vs. NL and NU vs. NL, respectively. Moreover, 16, 11, and 19 DAMs were exclusively expressed in the three pairwise comparisons, respectively (Figure 3C). There were 41, 65, and 243 DEGs overlapping between NM vs. NL and NM vs. NU, between NM vs. NU and NU vs. NL, and between NM vs. NL and NU vs. NL, respectively. In addition, 294, 83, and 357 DEGs were uniquely expressed in the three pairwise comparisons (Figure 3D). The Venn diagrams results of metabolome data and transcriptome data were consistent.

Furthermore, the enrichment analysis results of the KEGG pathway showed that the DAMs in the samples were mainly enriched in the pathway of “citrate cycle (TCA cycle)”, “biosynthesis of amino acids”, “pentose phosphate pathway”, and “carbon fixation in photosynthetic organisms” (Figure 3E) and the DEGs in the samples were mainly enriched in the pathway of “phenylpropanoid biosynthesis”, “plant hormone signal transduction”, and “flavonoid biosynthesis” (Figure 3F). It indicated that the expression changes of the genes associated with these classes might be very important for rooting since levels in sugars, phytohormones, organic acids, phenylpropanoids, and flavonoids were likely quite different among NL, NU, and NM.

3.4. Comparison of Key Metabolites and Related DEGs among Different Cutting Nodes

The types and concentrations of saccharides were significantly different in NU, NL, and NM. The 19 differential saccharides were obtained, and it was obvious from the heatmap that saccharide content in NL and NU was lower than that in NM. Compared with NU and NM, the contents of 3,6-anhydro-D-galactose, isomaltose, N-acetyl-beta-D-mannosamine, gentiobiose, allose, glucose, fructose, trehalose-6-phosphate, cellobiose, and raffinose were significantly lower in NL. Meanwhile, the content of ribose, ribonic acid, gamma-lactone, erythrose, and d-glucoheptose were lower in NU compared with NL and NM. Moreover, compared with NM, the contents of fructose were significantly lower in NL and NU than in NM (Figure 4A). The expression level of DEGs correlated with differential saccharides were significantly higher in NL than NM except 1,4-alpha-glucan-branching enzyme 1 (CSS0018649); among them, pectinesterase (CSS0012593) was upregulated in NU compared with NM (Figure 4B). Those DEGs were mainly enriched into the KEGG pathway of “starch and sucrose metabolism” (Figure 4C).

Organic acids were important metabolites. A total of 14 differential organic acids were identified. It could be seen from the heatmap that the contents of most differential organic acids in NL and NU showed the opposite trend. Overall, the contents of differential organic acids were higher in NL and lower in NU, while the contents of hydroxyrea, maleamate, and isocitric acid were significantly lower in NL (Figure 5A). There were 30 DEGs were correlated with differential organic acids in the heatmap (Figure 5B). The expression level of most DEGs were significantly high in NL, which were mainly low in NM and NU. In particular, the expression of homocysteine S-methyltransferase 2-like (CSS0050363), sorbitol dehydrogenase (CSS0020410), and inositol oxygenase 4-like (ONT.1996) was significantly upregulated in NL compared with NU (Figure 5B). The KEGG enrichment analysis results showed that the differential organic acids were mainly annotated in six pathways and the DEGs correlated with differential organic acids were enriched in the pathway of “cysteine and methionine metabolism” and “pentose and glucuronate interconversions” (Figure 5C).

A total of 5 differential phenolic acids and 4 differential flavonoids were detected. Compared with NM, the contents of coniferyl alcohol, (+/−)-taxifolin, and prunin were significantly lower in NL and NU. Moreover, the contents of arbutin, epicatechin, and (+)-catechin were significantly higher in NU and NM than in NL (Figure 6A). In addition, 35 DEGs correlated with phenolic acids and flavonoids were obtained, in which, most DEGs were significantly upregulated in NL than in NM and NU. Notably, dihydroflavonol-4-reductase (CSS0011557) was significantly downregulated in NL than in NM and NU. Compared with NM, the expression level of cinnamyl alcohol dehydrogenase 3 (CSS0018190), aldehyde dehydrogenase (CSS0028365), peroxidase (CSS0043574), and peroxidase P7-like (CSS0001194) were higher in NU (Figure 6B). The DEGs correlated with phenolic acids and flavonoids were mainly annotated in the KEGG pathway of “phenylpropanoid biosynthesis” and “flavonoid biosynthesis” (Figure 6C).

3.5. Co-Expression Network Analysis among Different Cutting Nodes

To explore the relationship between phytohormone content and gene expression, WGCNA was performed to construct a co-expression network (Figure 7). A total of 14 gene co-expression modules (labeled with different colors) were identified in the cluster dendrogram. The black module showed a remarkable correlation with IAA/ABA (p < 0.01, R > 0.8) and the dark gray module showed a correlation with IAA/GA3 and IAA/ZR (R > 0.6).

To further understand the function of genes in the black module and dark gray module, COG and KEGG enrichment analyses were performed. The results showed that these genes in the two modules were mainly involved in “carbohydrate transport and metabolism”, “general function prediction only”, “signal transduction mechanisms”, and “secondary metabolites biosynthesis, transport and catabolism” (Figure S2A,B). The KEGG enrichment analysis results showed that the pathways of gene enrichment were different in the two modules, but both were enriched in the pathway of “plant hormone signal transduction” (Figure S2C,D). These results suggest that these pathways were highly correlated with hormones and played critical roles in adventitious rooting. The genes involved in the plant hormones signal transduction pathway were analyzed and a total of 17 genes were identified. There were 9 auxin-related genes, including 7 auxin-responsive protein genes and 2 auxin transporter genes. Meanwhile, 1 cyclin D3-1, 2 two-component response regulator genes, and 2 ABA receptor genes were identified (

Table 1. The genes of the two modules involved in the plant hormones signal transduction pathway.

#ID	NM vs. NL		NM vs. NU		NU vs. NL		NR Annotation
	p Value	log2FC	p Value	log2FC	p Value	log2FC
CSS0005262	0.35	0.35	0.62	−0.13	0.14	0.60	DELLA protein 1
CSS0006056	8.55 × 10−4	1.10	0.53	0.18	4.15 × 10−3	0.92	pathogenesis related protein
CSS0008648	0.97	−0.01	0.32	−0.28	0.37	0.37	two-component response regulator ARR12
CSS0015005	1.19 × 10−5	1.24	0.89	0.04	6.77 × 10−5	1.20	auxin early response protein AUX/IAA7
CSS0017030	2.17 × 10−1	0.43	0.25	−0.33	0.01	0.88	auxin early response protein AUX/IAA4
CSS0017736	1.56 × 10−3	0.88	3.51 × 10−3	−0.74	1.77 × 10−9	1.75	PREDICTED: abscisic acid receptor PYL4-like
CSS0019988	0.59	0.20	0.17	−0.37	0.05	0.79	auxin early response protein AUX/IAA29
CSS0024392	0.12	0.59	0.47	−0.16	0.03	0.90	cyclin D3-1
CSS0033521	0.74	0.12	0.23	−0.23	0.10	0.60	auxin influx carrier component
CSS0047272	1.96 × 10−5	1.17	0.91	−0.03	1.96 × 10−5	1.23	abscisic acid receptor PYL4-like
CSS0050405	0.15	0.55	0.76	−0.08	0.09	0.69	regulatory protein NPR5-like isoform X1
ONT.5469	0.08	0.66	0.91	0.03	0.10	0.66	auxin-responsive protein IAA1-like
ONT.6169	4.15 × 10−3	1.03	0.58	0.16	0.02	0.85	PREDICTED: auxin-responsive protein SAUR40
CSS0030334	0.76	0.12	0.01	0.76	0.02	−0.87	auxin-responsive protein SAUR32-like
CSS0046696	0.92	−0.04	0.85	0.05	0.77	−0.12	auxin transport carrier
CSS0050314	0.25	−0.44	0.83	−0.06	0.34	−0.39	two-component response regulator ARR9-like
ONT.18362	0.21	0.47	0.04	−0.52	9.25 × 10−4	1.36	auxin-responsive protein IAA1-like

).

4. Discussion

4.1. Relationship of Endogenous Phytohormone and AR Formation

The formation of adventitious roots (ARs) is essential for the successful propagation of plant materials, which is regulated by a number of internal and external factors, such as hormone level, nutritional status in cutting, and certain environmental stresses [29,30]. Auxins, mainly IAA, act with an array strongly non-linear effects of other phytohormones through complex crosstalk, modulating each other’s levels and actions at different levels [10,31,32]. In the study of Phyllostachys edulis ‘Pachyloen’, winter shoots had higher ZR/IAA and GA/IAA ratios than rhizomes and maternal bamboo organs [33]. The ratio of IAA/GA3 in the easy-to-root Camellia sinensis cultivar ‘Baihaozao’ was significantly higher than the cultivars ‘Zhongcha108’ and ‘Longjing43’ [34]. Moreover, the low ratio of auxin vs. cytokinin levels will facilitate root formation [35] and the ratio of IAA/ABA in cuttings was positively correlated with the rooting ability of cuttings [7,36]. In this study, the ratio of IAA/GA3 had significant difference in the different parts, in which the base and upper rooting parts were significantly higher compared with the middle non-rooting transition zone, and the ratio of IAA/GA3 was significantly higher in the upper rooting parts than that in the base rooting parts. Compared with the middle non-rooting transition zone, the ratio of IAA/ABA was higher in the base and upper rooting parts, and the ratio of IAA/ZR was significantly higher in the upper rooting parts but it was lower in the base rooting parts. These results suggest that higher ratio of IAA/GA3 and IAA/ABA promoted AR formation, which was consistent with previous studies. Meanwhile, the ratio of IAA/ZR is lower in the base rooting parts compared with that in the upper rooting and middle non-rooting transition zone, and this is due to the callus contained in the base of the cuttings, while the other two parts do not produce callus. In addition, the genes correlated with IAA/GA3 and IAA/ABA in this study, such as auxin/indole-3-acetic acid (AUX/IAA) genes [37], Gretchen Hagen 3 (GH3) genes [38,39], small auxin upregulated RNA (SAUR) genes [40], auxin carrier proteins [41], ARR and Cyclin D3-1 [42], have been proven to exert great influence on ARs formation.

4.2. Key Metabolites for AR Formation in Tea Softwood Cutting

Sufficient nutrients are one of the necessary conditions for cuttings to take roots. Soluble sugar is the main nutrient supply for plant growth and the direct energy source during the rooting of cuttings. Research has shown that the accumulation of carbohydrates was conducive to the formation of ARs of Pelargonium, Eucalyptus globulus, and Eucalyptus saligna [7,43,44,45]. In this study, saccharides were the most important DAMs, and there were significant differences in the comparison of rooting parts NL and NU with non-rooting part NM, and the downregulation was dominant. Meanwhile, the downregulation of saccharides in NL was more than that in NU. Both NL and NU need energy for rooting, but the AR formation in NL was based on callus. A large number of saccharides was consumed in this part to provide energy for the formation of callus and root development. The AR formation in NU was directly from the cortex, while rooting did not need to supply energy for the formation of callus, and the consumption of saccharides was relatively small.

Organic acids are involved in the metabolic process of many substances and are of great significance to the life activities of plants. Exogenous application of malic acid and citric acid could improve the rooting rate and root length of rose rootstock [46]. In this study, the content of organic acids in NU decreased and that in NL increased compared with NM. The types and contents of organic acids in NU were lower than those in NL, indicating that the higher content of organic acids promoted the rooting in NL. At the same time, the content of these organic acids in NU was low, which might be related to morphological upper respiration and metabolic cycle of other substances.

Phenolic acids are important secondary substances in plants, and their physiological effects are complex with hormones. The elongation of soybean roots could be significantly interrupted when the concentration of coumaric acid was greater than or equal to 0.25 mmol/L [47,48]. Phenolic acids, particularly monophenols, can inhibit rooting [49]. Concentrations of monophenolic sinapic acid and vanillic acid drop rapidly in terminal cuttings to reduce negative effects on rooting [50]. In previous studies on tea plants, it was found that phenols played a key role in AR induction, and phenols could be used as tea rooting promoters [51]. In our study, coniferyl alcohol, arbutin, and 4-hydroxy-3-methoxybenzoic acid were downregulated in NL and NU, of which coniferyl alcohol was the only common differential metabolites. Meanwhile, salicylic acid and arbutin were downregulated in NU vs. NL. This result is consistent with the conclusion that polyphenols inhibit rooting from other species. Only one phenolic acid substance was the same between NL vs. NM and NU vs. NM, so the decrease of coniferyl alcohol content should play a key role in promoting rooting. Flavonoids are secondary metabolites involved in plant development and defense and play an important role in organisms. Flavanol decreased from callus to rooting stage compared with all other compounds analyzed [52]. In the study of pear and papaya grafting, it showed that the accumulation of epicatechin above healing may be an indicator of transplantation failure [53]. In this study, there were significant differences in flavonoids between NL and NU. Compared with NM, epicatechin, (+)-catechin, and (+/−)-taxifolin were downregulated in NL but were unchanged in NU, and epicatechin and (+)-catechin were lower in NL than in NU. This suggests that the low flavonoids content might be crucial for rooting in NU and NL. Phenolic acids and flavonoids were downregulated in NU vs. NL, indicating that the two kinds of metabolites were more important in promoting rooting in NL.

4.3. Expression of Genes Involved in Phenylpropanoid and Flavonoid Biosynthesis Associated with AR Formation

Based on the transcriptomic and metabolomic datasets, the KEGG network of phenylpropanoid biosynthesis and flavonoid biosynthesis with important DAMs and DEGs was delineated (Figure 8).

Phenylpropanoid biosynthesis and flavonoid biosynthesis are both important secondary metabolic pathways and many enzymes are involved in it to produce flavonoids, lignans, cinnamic acids, amides and other metabolites. 4-coumarate–CoA ligase (4CL) and caffeoyl-CoA O-methyltransferase (E2.1.1.104) were two important enzymes in the lignin formation [54]. Moreover, 4-coumarate–CoA ligase could also catalyze the synthesis of hydroxycinnamate-CoA thioesters, the precursors of other important phenylpropanoids, in two-step reactions involving the formation of hydroxycinnamate-AMP anhydride and then the nucleophilic substitution of AMP by CoA [55]. Peroxidases (E1.11.1.7) are heme-containing enzymes that catalyze oxidation of a diverse group of organic compounds and isoperoxidases involved in the polymerization of substituted cinnamyl alcohols to lignin [47,56]. Meanwhile, in the study of Betula pendula, it was proven that one isoform of peroxidase was a predictive marker of AR formation [57]. In this study, 4CL (CSS0011741 and CSS0016246), E2.1.1.104 (CSS0041921), and most E1.11.1.7 genes (CSS0001387, CSS0007582, CSS0021668, CSS0026112, CSS0017484, CSS0036797, CSS0040539, CSS0019982) were upregulated in NM vs. NL, while three E1.11.1.7 genes (CSS0001194, CSS0032480, and CSS0043574) were upregulated and two E1.11.1.7 genes (CSS0015732 and CSS0037341) were downregulated in NM vs. NU (Figure 8A). Furthermore, coniferyl alcohol were regulated by the DEGs, such as the genes encoding 4CL, peroxidase, and cinnamyl alcohol dehydrogenase (CAD) [58]. In the study of Z. bungeanum leaves, it was reported that annotated partial genes were related to flavonoid biosynthesis, such as 4CL, chalcone synthase (CHS), bifunctional dihydroflavonol 4-reductase genes (DFR), anthocyanidin synthase (ANS), and flavonol synthase (FLS) [59]. Those genes were upregulated in NU vs. NL in this study except for DFR (CSS0011557) (Figure 8B). The findings suggest that these changes of DEGs might regulate the decreased levels of coniferyl alcohol, (+)-catechin, and (-)-epicatechin and lignification of softwood cuttings to promote AR formation in NL and NU. In contrast, the upregulated genes in NL were more than those in NU, and the rooting in NL was significantly regulated by genes, which was consistent with the rooting situation of the two.

5. Conclusions

To sum up, in order to achieve the rooting state of tea softwood cutting, saccharide consumption is required to provide energy. Organic acids are formed and accumulated by saccharides transformation in callus, while they are consumed due to their participation in multiple ways in cortical rooting. Coniferyl alcohol and flavonoids all play a crucial role in AR formation. At the phytohormone level, the high ratios of IAA/GA3 and IAA/ABA both promote AR formation. Notably, compared with the upper rooting parts of cuttings, saccharides are lower, organic acids showed an opposite trend in the base rooting parts of cuttings (NL), and epicatechin, (+)-catechin, and (+/−)-taxifolin are only downregulated in NL. The expression level of genes is consistent with the above conclusion. This study can provide theoretical support for the subsequent research on the material basis of cutting rooting of difficult rooting tea varieties. In the study of cutting breeding, the cuttings can reach the rooting state by means of exogenous hormone regulation so as to improve the survival rate and shorten the breeding cycle, which is more beneficial to the softwood cutting of difficult rooting varieties. However, the molecular mechanism of two rooting types of softwood cutting needs to be further studied.