Using the nine Arabidopsis CAD protein sequences as queries, we searched the TAIR database¹ and identified another nine CAD-like proteins in Arabidopsis based on domain analysis (Figure 1A). The nine CAD-like proteins were originally annotated as aldehyde dehydrogenases. We then used these Arabidopsis CAD and CAD-like protein sequences to search against the proteins annotated in the Upland cotton genome [Gossypium hirsutum (AD1) “TM-1” genome NAU-NBI_v1.1_a1.1] (Zhang et al., 2015). All hits were further validated based on the presence of the two key domains (PF08240, alcohol dehydrogenase GroES-like domain; PF00107, zinc-binding dehydrogenase) responsible for the CAD catalytic activity. Finally, we identified 46 cotton CAD and CAD-like proteins which were classified into five groups (group I to V) according to the phylogenetic tree and the conserved motifs analysis of the amino acids of this gene family (Figure 1A and Supplementary Figure 1). Groups I, II, and III contain the nine Arabidopsis CADs previously reported and 19 GhCADs. Groups IV and V contain the nine Arabidopsis CAD-like proteins identified in this study and 27 GhCAD-like proteins (Figure 1A).

We investigated the tempo-spatial expression pattern of the 46 candidate GhCADs in root, stem, and leaf of 30-, 60-, and 90-day old cotton plants (cv. TM-1) by qRT-PCR. Of the 46 genes, 28 were found to be expressed in at least one of the samples tested (Figure 1B). Generally, the 28 GhCADs could be classified as low- (4), mediate- (17), and high-expression (7) three groups based on their overall expression levels in the tissues analyzed (Figure 1B). A specific expression pattern was observed for some GhCADs, for example, GhCAD27 and GhCAD43 seemed to be leaf specific, and GhCAD6 seemed to be mainly expressed in stem and root, while GhCAD35 showed an obvious developmental stage specific expression pattern (Figure 1B).

Lignin biosynthesis and deposition in plants induced by V. dahliae infection contribute to resistance to the pathogen (Zhang et al., 2019). To identify GhCADs responsible for V. dahliae induced lignin biosynthesis, we firstly identified four resistant cultivars, including Zhongzhimian 2 (ZZM 2), Han 333, GK 44, and GK 164, and four susceptible cultivars, including Jimian 11, 8891, Foster 6, and DH 966 from a collection of G. hirsutum cultivars (Supplementary Figure 2). ZZM 2 and Jiman11 have been previously reported to be resistant and susceptible to V. dahliae, respectively (Zhang et al., 2014; Zhang X.Y. et al., 2020). To exaggerate the expression of GhCADs after the induction of V. dahliae infection, we next compared the expression of the 28 GhCADs upon V. dahliae infection in the four resistant and the four susceptible cultivars, as well as TM-1, using plants at 18 days-post-infection (dpi) by qRT-PCR. PR10-5 and PR10-11 (Dowd et al., 2004), two pathogen defense-related genes, were used as positive controls. As expected, the expression level of PR10-5 and PR10-11 was highly up-regulated in all tested cultivars infected by V. dahliae (Figures 2A,B). Of the 28 GhCADs, three, including GhCAD35, GhCAD45, and GhCAD43, were consistently highly induced in almost all tested cultivars (Figures 2C–E), while the other GhCADs were either repressed or no consistent responsive trend in different cultivars after V. dahliae inoculation (Supplementary Figures 3, 4).

We further investigated the expression profiles of GhCAD35, GhCAD45, and GhCAD43, as well as PR10-5, 24 h after V. dahliae infection in the four resistant and the four susceptible cultivars. Interestingly, PR10-5 was significantly induced in all the four resistant but not in the four susceptible cultivars (Figure 3A). GhCAD35 and GhCAD45 were significantly induced in three (ZZM 2, Han 33, and GK 44) of the four resistant cultivars and one (8891) of the susceptible cultivars although they were repressed or not affected in different susceptible cultivars (Figures 3B,C); while the expression of GhCAD43 was only significantly induced in one (Foster6 24) of the cultivars (Figure 3D). Together, these results suggest a potential role of these three genes, particularly GhCAD35 and GhCAD45, in response to V. dahliae infection at the early infection stage.

In plants, CADs catalyze the conversion of p-coumaraldehyde, coniferaldehyde, and sinapaldehyde into p- coumaryl-, coniferyl-, and sinapyl-alcohol, respectively, in the lignin biosynthesis pathway (Sibout et al., 2005; Li et al., 2012; Jun et al., 2017). To test whether GhCAD35, GhCAD45, and GhCAD43 have CAD catalytic activities, we generated recombinant GhCAD35, GhCAD45, and GhCAD43 fusion proteins with a GST tag (Figure 4D), incubated individual recombinant GhCAD35, GhCAD45, or GhCAD43 protein with each of the three substrates, p-coumaraldehyde, coniferaldehyde, and sinapaldehyde in vitro, and analyzed the final reaction products using HPLC. As shown in Figures 4A–C, GhCAD35, GhCAD45, and GhCAD43 all clearly showed CAD catalytic activities to efficiently catalyze p-coumaraldehyde, coniferaldehyde, and sinapaldehyde into their corresponding alcohol products, p- coumaryl-, coniferyl-, and sinapyl-alcohol, respectively.

To further address the question of the substrate preference and catalytic efficiency, we determined the K_m and V_max values of GhCAD35, GhCAD45, and GhCAD43 for the three substrates p-coumaraldehyde, coniferaldehyde, and sinapaldehyde, respectively. The three enzymes displayed slight differences in their kinetic parameters (Figure 4E). GhCAD45 generally showed a less affinity toward all the three substrates compared with GhCAD35 and GhCAD43. GhCAD35 showed a little higher affinity toward p-coumaraldehyde and sinapaldehyde than coniferaldehyde. GhCAD43 showed a little higher affinity toward coniferaldehyde and sinapaldehyde than p-coumaraldehyde; while GhCAD45 showed no clear affinity preference toward the three substrates (Figure 4E). The three enzymes showed no significant differences in their V_max values, although GhCAD45 generally showed higher catalytic rates toward all the three substrates compared with GhCAD35 and GhCAD43 (Figure 4E).

Together, these results demonstrate that GhCAD35, GhCAD45, and GhCAD43 have significant CAD catalytic activities to generate monolignol products.

To investigate whether GhCAD35, GhCAD45, and GhCAD43 are involved in Verticillium wilt resistance, we used the tobacco rattle virus (TRV)-based VIGS to knock down the expression of individual gene in the Verticillium wilt resistant G. hirsutum cultivar ZZM 2. The TRV:CLA construct was used as a positive control to validate the gene silencing system. As expected, TRV:CLA plants displayed a clear bleaching phenotype from the first pair of true leaves (Supplementary Figure 5A), indicating the VIGS system used here was effective and reliable. Next, we compared the expression level of GhCAD35, GhCAD45, and GhCAD43 in their corresponding VIGS plants with those in TRV:00 plants 2 weeks after Agrobacterium infiltration by qRT-PCR. The average expression level of GhCAD35, GhCAD45, and GhCAD43 in their corresponding VIGS plants was 21.3, 17.4, and 13.0% relative to that in TRV:00, respectively, indicating a successful silencing of the target genes (Supplementary Figures 5B,C). In addition, to check the specificity of the gene silencing in the VIGS plants, we also compared the expression of another three most homologous GhCADs of GhCAD35, GhCAD45, and GhCAD43, respectively, in the corresponding TRV:CAD35, TRV:CAD45 and TRV:CAD43 VIGS plants relative to those of TRV:00. As shown in Supplementary Figure 6, the expression of detected GhCADs were not altered in all the VIGS plants, suggesting TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants specifically silenced the expression of the target gene, respectively.

To investigate the responses of gene silencing to V. dahliae infection, V. dahliae was inoculated in the TRV:00, TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants 2 weeks after Agrobacterium infiltration. The disease symptoms were examined at 18 dpi. As shown in Figure 5A, very weak disease symptoms were observed in TRV:00 plants; while obvious and typical disease symptoms of Verticillium wilt, including leaf chlorosis, wilting, and defoliation were observed in TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants. We further quantified the ratio of different disease-grades and disease index (DI) to evaluate the degree of Verticillium wilt. Half or more of the TRV:GhCAD35 (60%), TRV:GhCAD45 (55%), and TRV:GhCAD43 (50%) plants displayed the severe grades 3 and 4, while no TRV:00 plant of these grades were observed and only 15% of TRV:00 plants displayed the weak grades 1 and 2 (Figure 5B). In addition, TRV:GhCAD35 plants (DI: 60.0) displayed slightly more severe disease symptoms compared with TRV:GhCAD45 (DI: 56.3) and TRV:GhCAD43 (DI: 50.0) plants (Figure 5B). Furthermore, we compared vascular browning of TRV:00, TRV:CAD35, TRV:CAD45, and TRV:CAD43 plants using longitudinal sections of stems. The vascular tissues of TRV:GhCAD45 and TRV:GhCAD43 plants, especially TRV:GhCAD35, displayed obviously dark-brown, typical symptom of V. dahliae infection, while no such symptom was observed in TRV:00 plants (Figure 5B). These results indicate that silencing GhCAD35, GhCAD45, or GhCAD43 compromised the resistance to V. dahliae infection in ZZM 2.

In addition, we chose GhCAD8 and GhCAD1, which were repressed by V. dahliae infection in most cultivars 18 dpi (Supplementary Figure 4), and analyzed their roles in resistance to V. dahliae infection by examining their VIGS plants. Compared to TRV:00 plants, TRV:GhCAD8 plants displayed slight susceptibility to V. dahliae infection, and TRV:GhCAD1 plants showed no Verticillium wilt disease symptoms as indicated by plant growth, quantification of disease grades and DI, and observation of vascular tissues 18 dpi (Supplementary Figure 7). Together, these results suggest that GhCAD35, GhCAD45, and GhCAD43 are positively involved in Verticillium wilt resistance in G. hirsutum.

Verticillium dahliae induced lignification has been reported to confer resistance to Verticillium wilt in plants (Smit and Dubery, 1997; Gayoso et al., 2010). To evaluate the involvement of GhCAD35, GhCAD45, and GhCAD43 in lignin biosynthesis, we quantitatively measured the composition of the three lignin monomers, i.e., H-, G-, and S-unit, assayed by thioacidolysis and analyzed by gas chromatography mass spectrometry (GC-MS) using stems of TRV:00, TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants with or without infection by V. dahliae (48 h after inoculation). In TRV:00 plants, V. dahliae infection significantly induced lignin biosynthesis and deposition (+60.0%), mainly the G- and S-unit lignin (+68.4% and +57.9%, respectively) (Figures 6B,C). Compared to TRV:00 plants, TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants showed a decrease of the total lignin content, mainly due to decreased G- and S-unit, especially under V. dahliae infected condition (Supplementary Table 1). In addition, the levels of the G-unit and S-unit were increased in V. dahliae infected TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants compared to their corresponding control plants, despite not as significant as that in TRV:00 plants (Figures 6B,C). No significant difference was observed for the effect of V. dahliae infection on the H-unit lignin between VIGS plants and controls (Figure 6A). Cell wall residue (CWR) mainly indicates the ratio of lignin in the whole tissue. Notably, CWR showed a slight but significant increase in TRV:00 plants after V. dahlia inoculation; in contrast, a decrease trend of CWR was observed in the V. dahliae infected TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants compared to their corresponding controls (Figure 6D).

To further examine the effects of GhCAD35, GhCAD45, and GhCAD43 on lignin deposition upon V. dahliae infection, we sectioned the stems of the TRV:00, TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants at 18 dpi, and checked deposition of lignin in xylem based on the autofluoresence under UV light and by phloroglucinol-HCl staining. Under the control (ddH₂O) conditions, no difference in lignin deposition was observed in stems among TRV:00, TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants (Figure 7); however, under the V. dahliae infection conditions, the xylem width of the stems from the TRV:GhCAD35, TRV:GhCAD45 and TRV:GhCAD43 plants was much thinner compared to that of TRV:00 indicated by autofluorescence under UV illumination and the color stained by phloroglucinol-HCl (Figure 7). Moreover, the V. dahlia infection could be observed surrounding the vessels in the V. dahlia infected VIGS plants, indicating spreading of the V. dahlia infection from the root to the shoot through vessels (Figure 7). These results were further confirmed by the quantification of the lignin composition of H-, G-, and S-unit. As shown in Table 1, under control (ddH2O) conditions, the contents of the three units in TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants were slightly lower than that of TRV:00; however, upon V. dahliae infection, all the three units significantly reduced at 18 dpi in TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 compared with TRV:00 plants. Moreover, the reduction of the S unit was more severe than the H- and G-units, leading to a significantly reduced S/G ratio in TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants but not in TRV:00 plants (Table 1). In addition, to check whether the silencing of GhCAD genes disrupts the expression of phenylpropanoid metabolic and lignin biosynthesis genes, we checked the expression of several genes in phenylpropanoid metabolic pathway, such as GhPAL1, Gh4CL2, GhHCT, GhCOMT1, GhCCR1, etc. However, no clear induction or repression was observed for the investigated genes (Supplementary Figure 8).

Taken together, these results demonstrate GhCAD35, GhCAD45, and GhCAD43 positively regulate V. dahliae induced lignification in the stems of Upland cotton.

Salicylic acid and JA play key roles in plant defense against pathogen infection (Shaban et al., 2018; Song et al., 2020). Previous studies have shown that disturb of lignin metabolism in the cell wall of root and stem tissues in G. hirsutum results in the change of SA synthesis and signaling upon V. dahliae infection. Therefore, we investigated the effects of silencing GhCAD35, GhCAD45, or GhCAD43 on the SA biosynthesis and signaling. As shown in Figure 8A, SA content was significantly increased in TRV:00 plants 8-h- and 12-h-post inoculation of V. dahliae (Figure 8A); however, SA contents were lower or less induced in the TRV:GhCAD35, TRV:GhCAD45, and TRV:GhCAD43 plants compared to that of TRV:00 plants after V. dahliae infection. We further investigated the expression of SA biosynthesis gene ICS1, and SA signaling genes PR1, PR5, and NPR1 12-h-post V. dahliae infection by qRT-PCR. Consistently, the expression levels of ICS1, PR1, PR5, and NPR1 were all induced by V. dahliae infection in TRV:00 plants; while in TRV:GhCAD35 and TRV:GhCAD45, the expression levels of ICS1, PR1, and PR5 were inhibited or no significant change after V. dahliae infection. In TRV:GhCAD43, a less induction or no significant change were observed for the expression of ICS, PR1, and PR5 after V. dahliae infection compared to those of TRV:00 plants (Figures 8B–D). We also investigated the effects of silencing GhCAD35, GhCAD45, or GhCAD43 on the gene expression of JA signaling components, such as GhJAZ3, GhJAZ10, GhMYB2-2, and GhMYB2-3. As shown in Supplementary Figure 9, GhJAZ10 was slightly induced after V. dahliae infection in TRV:GhCAD35 and TRV:GhCAD45, but the other genes did not show consistent changes in the GhCAD silencing plants after V. dahliae infection compared to that of control. These results indicate that silencing GhCAD35, GhCAD45, or GhCAD43 might affect the defense induced SA responses at the early infection stage.

The above results indicate that GhCAD35, GhCAD45, and GhCAD43 are positive regulators of resistance to V. dahliae by affecting lignin biosynthesis and deposition. To further confirm this conclusion, we heterogeneously overexpressed GhCAD35, GhCAD45, or GhCAD43 in Arabidopsis to examine their effects on V. dahliae infection. We analyzed two independent Arabidopsis transgenic lines overexpressing GhCAD35, GhCAD45, or GhCAD43, and challenged the transgenics with V. dahliae. As shown in Figures 9A–C, compared to the control 35S:GFP plants, all Arabidopsis transgenic lines displayed overexpressed expression of corresponding GhCADs and enhanced resistance to V. dahliae infection, as indicated by plant growth, disease infection grades, and disease index statistics, although the resistance of the transgenics overexpressing GhCAD43 was not as significant as those of the transgenes overexpressing GhCAD35 or GhCAD45. Furthermore, we chose one line of each GhCADs overexpressing transgenic Arabidopsis plants and measured the H-, G-, and S-units of lignin contents in their stems. As shown in Supplementary Figure 10, S-unit of lignin was significantly elevated in the GhCADs overexpressing transgenic Arabidopsis plants compared with that of control, suggesting overexpression of GhCAD35, GhCAD45 or GhCAD43 enhanced the lignin biosynthesis in Arabidopsis.

Nine Upland cotton (G. hirsutum L.) accessions, including four (Zhongzhimian 2, Han 333, GK 44, and GK 164) resistant to V. dahliae and five (Jimian 11, DH 966, Foster 6, 8891, and TM-1) susceptible to V. dahliae, were used in disease assay. Plants were grown at 25°C in a growth room with 80–100 μmol⋅m^–2⋅s^–1 light and a 16-hr light/8-hr dark photoperiod. Two-weeks-old cotton plants were used in V. dahliae infection with the strain of V991.

The V. dahliae resistant cultivar Zhongzhimian 2 (ZZM 2) was used in the VIGS experiment and disease assay. The experiment was performed in a growth room with a temperature of 23–24°C, light intensity of 100 μmol⋅m^–2⋅s^–1, and a photoperiod of 16-hr light/8-hr dark.

The Arabidopsis (Col-0 background) plants were grown at 22°C in a growth chamber under 100 μmol⋅m^–2⋅s^–1 light and a 16-hr light/8-hr dark photoperiod. Arabidopsis transformation was performed using the flower dipping method as described previously (Clough and Bent, 1998).

Two approaches were adopted to identify cotton CAD genes in G. hirsutum. First, nine bona fide Arabidopsis CAD proteins were used to search Arabidopsis TAIR database (see text footnote 1) and identified another nine CAD-like proteins in Arabidopsis based on domain analysis. Arabidopsis CAD and CAD-like protein sequences were used as queries to search against the annotated proteins of G. hirsutum (Gossypium hirsutum (AD1) ‘TM-1’ genome NAU-NBI_v1.1_a1.1) (Zhang et al., 2015) downloaded from the CottonGen database² by local blast tool with the default settings (E-value < 1^e–10). All hits were considered as potential candidate CAD genes. Second, the hidden Markov model (HMM) profile of the two conserved domains (PF08240, alcohol dehydrogenase GroES-like domain; PF00107, zinc-binding dehydrogenase)³⁴ (Potter et al., 2018) was used to search the G. hirsutum (Gossypium hirsutum (AD1) ‘TM-1’ genome NAU-NBI_v1.1_a1.1) (Zhang et al., 2015) using the HMMER 3.0 software. All candidate sequences derived from the two searches were subjected to domain analysis using the InterProScan⁵ and SMART⁶ tools with the default parameters. Only protein sequences with both domains were considered to be CADs. The gene ID and corresponding gene name for all GhCADs are listed in Supplementary Table 2.

The V. dahliae strain V991 was cultivated on potato dextrose agar (PDA) plate at 25°C for 4–5 days, and was then transferred to potato dextrose broth (PDB) liquid media and incubated on a shaker (120 rpm, 25°C) for 3 days. The conidia of V. dahliae were collected and adjusted to a concentration of ∼1 × 10⁷ spores/mL. Cotton seedlings were inoculated by root dipping with the inoculum. Three replicates were set up for each genotype with at least 20 plants in each replicate.

Disease index was calculated according to the method descripted previously (Miao et al., 2019). Disease symptom severity was scored as level 0 to level 4 at 18 dpi according to the percentage of wilted true leaves and leaf chlorosis or defoliation. A score of zero means no visible wilting or chlorosis symptoms; a score of 1–4 indicates 0–25% (inclusive), 25–50% (inclusive), 50–75% (inclusive), and 75–100% of wilted true leaves and leaf chlorosis or defoliation, respectively. DI was calculated according to the following formula:

n represents disease levels ranging from 0 to 4.

Total RNA was extracted from cotton stems using a RNAprep Pure Plant Plus Kit (Polysaccharides and Polyphenolics-rich) (Tiangen, Beijing, China). Then 1 μg total RNA was taken for reverse transcription to synthesize cDNA using HiScript II Q Select RT SuperMix (Vazyme, Nanjing, China). The quantitative real-time PCR (qRT-PCR) experiment was conducted using ChamQ SYBR Color qPCR Master Mix (Vazyme, Nanjing, China) and analyzed using the Roche LightCycler 480 (Beijing, China). The thermal profile for qRT-PCR was 95°C for 2 min, followed by 40 cycles of 95°C for 15 s and 60°C for 30 s. Primers used for qRT-PCR are listed in Supplementary Table 3.

Full-length GhCADs were amplified with a KOD-Plus kit (Toyobo) using the cDNA generated from TM-1. The gene specific primers were designed according to the TM-1 genome sequence of NAU-NBI_v1.1. GhCADs were expressed in E. coli (BL21) by cloning individual GhCAD into a glutathione (GST)-tagged expression construct pGEX4T-1. The recombinant GST-GhCADs were purified using a column containing GST coupled with agarose medium. Western blot was used to detect the recombinant GST-GhCADs using anti-GST antibody on an SDS-PAGE Gel.

The reaction (200 μL) used in assay of GhCAD activity contains: 50 mM Tris HCl (pH = 7.5), 0.5 mM NADPH, 0.1 mM substrate (coniferyl aldehyde, p-coumaryl aldehyde, or sinapyl aldehyde), 2.0 mM dithiotreitol (DTT), and 1.3 μg recombinant GST-GhCAD protein. The reaction was incubated at 30°C for 30 min and terminated by adding 200 μL methanol. The reaction mixture was filtrated through a 0.22 μm organic diameter filter. A volume of 100 μL filtrated reaction mixture was then subjected to high-performance liquid chromatograph (HPLC) analysis with an XBridge BEH C18 column (2.5 μm, 3.0 × 150 mm XP). The experimental procedure established to separate the products (three alcohols) from the substrates (three aldehydes) was as follows: two chromatographic grade solvents, water (containing 1% formic acid) and acetonitrile, were used as the mobile phase A and B, respectively; the proportion of A:B was set to change from 95:5 to 5:95 from 0 to 30 min gradually; the flow rate was set to be 0.5 mL/min; the section of full wavelengths was set to be from 210 nm to 400 nm.

In order to analyze the GhCAD catalytic kinetics, nine concentrations (2, 6, 8, 17, 25, 33, 42, 50, and 100 μM) of each substrate were chosen to conduct the catalytic experiment as mentioned above, and the enzyme kinetic curves were fitted using the Origin 2018 software (Pan et al., 2014; Ma et al., 2018).

A gene-specific fragment encoding 100–200 amino acids was amplified from GhCAD35, GhCAD43, or GhCAD45, and cloned into the pTRV2 plasmid between the EcoRI and KpnI restriction sites. Primers used are listed in Supplementary Table 4. After confirmation by sequencing, the recombinant vector was introduced into Agrobacterium tumefaciens GV3101 by thermal activation. For VIGS, the Agrobacterium culture was resuspended in the VIGS buffer (10 mM MgCl₂, 10 mM MES and 200 μM acetosyringone) and mixed with Agrobacterium tumefaciens transformed with pTRV1 in equal proportion. After 3 h of incubation at room temperature, various Agrobacterium combinations were injected into cotyledons of 1 week old seedlings of ZZM 2. Two weeks later, the treated cotton seedlings were used in confirmation of silencing of the target gene (based on qRT-PCR) and then used in disease assay by inoculation of V. dahliae inoculum (as mentioned above).

Stems collected from the same position of different cotton plants were cross sectioned by hand with a section thickness of 100–120 μM. Sections were immersed in 1% phloroglucinol (0.1 g phloroglucinol in 10 mL 95%EtOH) for 3–5 min; afterward, sections were gently rinsed with water and acidified with 20% HCl solution. Finally, the sections were examined by inverted microscope.

Freeze-dried stem materials were fully ground to fine powder and extracted in 70% EtOH (5 × volume) at 65°C for 3 h. After repeated twice the extraction, the final extraction was carried out with chloroform/methanol (1:1 v/v) to get rid of water and pigments. The products were washed once by acetone (2–3 × volume) and dried for 3 days in an oven in a hood.

Lignin G-, S-, and H-units were depolymerized and separated with a streamlined thioacidolysis, and analyzed by gas chromatography–mass spectrometry (GC–MS) as previously described (Robinson and Mansfield, 2009). For each sample, 10 ± 1 mg of ground, oven-dried wood powder was filled into a 2-mL HPLC glass vial. The vials were blanked with nitrogen gas, added with 1 mL of freshly made reaction buffer [2.5% boron trifluoride etherate and 10% ethanethiol, in recently distilled dioxane (v/v)], and tightly capped. The vials were then placed in a dry heating block (98°C) for 4 h, with periodic manual agitation. The reaction was stopped by placing on ice for 15 min and adjusted to have a pH value between 3 and 4 by adding appropriate volume of 0.4 M sodium bicarbonate (∼0.3 mL, as determined by pH indicator paper). The solution was transferred to a new 10 mL vial. Internal standard (methyl heptadecanoate, 1 mg) dissolved in 1 mL methylene chloride was then added to each vial. To extract the reaction products from the aqueous mixture, the vials were recapped, vortexed and kept still for half an hour, phase-separating the upper (aqueous) and lower (organic, containing lignin breakdown products) phases. An aliquot (1.5 mL) of the organic phase was removed and simultaneously cleared of residual water and filtered by passing through a Pasteur pipette packed with a small tissue-paper plug and an inch (∼50 mg) of granular anhydrous sodium sulfate, and transferred directly into a 2-mL polypropylene microfuge tube. Samples were then collectively evaporated to dryness in a low pressure oven (50°C) overnight and resuspended in 0.5 mL of methylene chloride. A volume of 50 μL samples was transferred to new microcentrifuge tube and dried in the same condition as above; the dried samples were then combined with 50 μL of pyridine and 50 μL of N-methyl-N-(trimethylsilyl) trifluoroacetamide. After incubation for at least 5 h at room temperature, 1 μL of the reaction product was analyzed by GC-MS.

GC-MS was conducted on gas chromatography tandem mass spectrometry Xevo TQ-GC (Waters, United States). The analyses used were as previously described (Rolando et al., 1992). In brief, the GC-MS was equipped with a DB-5 column (30 m × 0.25 mm × 0.25 μm, Agilent Technologies). The oven initial temperature on the DB-5 column was maintained at 130°C for 3 min and then increased to 250°C at 3°C/min, and finally maintained 250°C for 5 min. The injector temperature was maintained at 250°C with a constant flow rate of 1.0 mL/min of helium. The inlet was operated in pulsed splitless injection mode. The GC–MS electron impact source was operated in scan mode with the MS source temperature at 280°C and MS Quad at 150°C.

Endogenous SA was extracted from the stems of cotton plants as previously described (Hu et al., 2018; Miao et al., 2019), and measured by LC-MS. In brief, three replicates, each with 100 mg (fresh weight) of stem samples, were harvested at 8 and 12 h after V. dahliae inoculation. The samples were ground to a fine powder in liquid nitrogen, mixed with 750 μL of ice-cold 80% methanol containing 1% acetic acid, and placed on a shaker for 16 h at 4°C in darkness. The samples were then centrifuged at 12,000 rpm (4°C) for 15 min, and the supernatant was collected and transferred into a new tube and was then evaporated to dryness, and the residue was reconstituted in 200 μL of 80% methanol. Finally, the extracts were analyzed using a Q Exactive Plus HPLC-MS system (Thermo, United States).