The seeds of G. hirsutum cv. Zhongzhimian No. 2 (resistant cultivar) were sown in sterile mixed soil (vermiculite:humus = 1:1) in a greenhouse at 28°C under 16-h light/8-h dark conditions.

The seeds of A. thaliana ecotype Col-0 were sown on Murashige and Skoog (MS) medium, and then, the seedlings were planted in pots containing mixed soil (vermiculite:humus = 1:1) in a culturing room at 23°C under 16-h light/8-h dark conditions.

The Nicotiana benthamiana seedlings were grown in pots containing mixed soil (vermiculite:humus = 1:1) at 25°C for approximately 6 weeks, under 16-h light/8-h dark conditions in a greenhouse.

The defoliating isolate V991 of V. dahliae was grown on potato dextrose agar for 4 days at 25 °C and then further cultured in liquid Czapek medium for another 5 days. The spores were collected and resuspended in deionized water. For V. dahliae infections, the roots of cotton seedlings grown under hydroponic conditions for 2 weeks were inoculated with a spore suspension (10⁶ spores ml^–1), and harvested after 24 h for RNA extraction. For the inoculation of SINA-silenced cotton plants, the spore concentration was adjusted to the 10⁶ spores ml^–1 and injected into the hypocotyl at 1 cm below the cotyledons using a springe needle, approximately 3 μl per plant (Bolek et al., 2005). For Arabidopsis infections, 18-old seedlings were gently uprooted from soil, rinsed in sterile water and the roots were dip-infected with a V. dahliae spore suspension (4 × 10⁵ conidia ml^–1) for 2 min. Then, the plants were transferred into fresh steam-sterilized soil for the detection of disease symptoms.

The disease index (DI) was calculated using the following formula: DI = [(Σ disease grades × number of infected plants)/(total number of plants × 4)] × 100. After V. dahliae infection, seedlings were classified into five grades, 0, 1, 2, 3, and 4, on the basis of the disease severity as reported previously (Xu et al., 2014). Fungal recovery assays were performed as previously described (Fradin et al., 2009). Briefly, after surface sterilization, the first internode sections were cut into 3–5 mm slices from control and SINA-silenced cotton plants, and then, they were separately cultured on PDA medium. To determine the level of V. dahliae colonization, the longitudinal cross-sections of cotyledonary nodes were dissected and observed under a stereoscopic microscope (Leica, Wetzlar, Germany) (Fradin et al., 2009).

For fungal biomass quantification, stems of inoculated cotton plants and roots of inoculated Arabidopsis plants were collected for DNA extraction. The internal transcribed spacer (ITS) region of ribosomal DNA was targeted using the fungus-specific ITS1-F primer in combination with the V. dahliae-specific reverse primer STVE1-R (Hu et al., 2018). Primers for cotton Histone3 and Arabidopsis Actin2 genes were used as endogenous plant controls. The quantitative real-time PCR (qRT-PCR) analysis was conducted on genomic DNA (Santhanam et al., 2013). The primer sequences are listed in Supplementary Table 1.

pTRV1 and pTRV2 (Liu et al., 2002) vectors were used for Virus-Induced Gene Silencing (VIGS) experiments. The specific fragments of GhSINAs were amplified and inserted independently into pTRV2. The primer sequences are listed in Supplementary Table 1. The recombinant plasmids were transformed into Agrobacterium tumefaciens strain GV3101. Agrobacterium cultures (OD₆₀₀ = 1) harboring the pTRV1 and pTRV2-GhSINA plasmids were mixed at 1:1 ratios and infiltrated into two full cotyledons of 7-day-old cotton seedlings using a needleless syringe as described previously (Gao et al., 2011). Alternatively, cotyledons of seedlings were infected with the mixture by vacuum infiltration (Qu et al., 2012). The G. hirsutum CLA1 gene was used as the positive control for the silencing system. Approximately 7 days after the Agrobacterium-mediated transformation of cotton plants, the GhCLA1 gene-silenced plants displayed the photobleaching phenotype, suggesting that the VIGS experiment was performed well.

The full-length open reading frames (ORFs) of GhSINA7, GhSINA8, and GhSINA9 genes were cloned independently into the pMAL-C2x vector to generate maltose binding protein (MBP)-fusion proteins. The primers used in the assay are listed in Supplementary Table 1. Recombinant proteins were expressed in the Escherichia coli BL21 strain, in the presence of 0.5 μM isopropyl β-D-1-thiogalactopyranoside, purified by affinity chromatography using amylose resin (NEB, Ipswich, MA, United States) and used for in vitro ubiquitination analyses as described previously (Xie et al., 2002; Cho et al., 2008; Zhao et al., 2013). Purified fusion MBP-SINAs (3 μg) were incubated independently in 30 μl ubiquitination reaction buffer (50 mM Tris-HCl, pH 7.5, 5 mM MgCl₂, 2 mM DTT, and 2 mM ATP), 5 μg biotinylated-ubiquitin (Enzo, #BML-UW9920-0001), 100 ng E1 (Enzo, #BML-UW9920-0001) and 40 ng Human E2 (UbcH5b) at 30°C for 3 h (Enzo, #BML-UW9920-0001). The reactions were terminated by adding the 5 × sample buffer, and half of the mixtures were separated using 7.5% SDS-PAGE gel. The proteins were identified by western blotting using an anti-biotin antibody (Cell Signaling, 1:3,000 dilution). Images were visualized on Tanon-5200 Chemiluminescent Imaging System (AI600 UV, United States) using chemiluminescence following the manufacturer’s instructions (ECL; Amersham, Thermo).

The Y2H screening was constructed in accordance with the instructions of the Matchmaker Gold Yeast Two-Hybrid (Y2H) System (Clontech, Palo Alto, CA, United States) (Bai and Elledge, 1996). The full-length cDNAs of the GhSINAs were cloned independently into both the bait vector pGBDK7 and the prey vector pGADT7. The constructs were co-transformed into yeast strain AH109, and the co-transformed yeast colonies were streaked onto SD/-Leu/-Trp DO (DDO) medium. After growth at 30°C for 72 h, independent colonies of the same size were transferred to SD/-Leu/-Trp/-Ade/-His DO (QDO) medium supplemented with X-α-Gal (Clontech) to assess the pair-wise interactions among the GhSINA proteins. The primers for the Y2H constructs are listed in Supplementary Table 1.

Total RNAs were extracted from different tissues of cotton or Arabidopsis plants using TRIzol reagent (TIANGEN, Beijing, China) and treated with DNase I in accordance with the manufacturer’s protocol. In total, 1 μg of total RNA was reverse transcribed using a cDNA synthesis kit, version R323 (Vazyme, Nanjing, China). The RT-PCR analyses were performed as described previously (Hu et al., 2018). The qRT-PCR assays were performed using SYBR Green Real-time PCR Master Mix (Vazyme) on a LightCycler480 system (Roche, Germany). PCR amplification parameters were as follows: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. The cotton Histone3 or Arabidopsis Actin2 gene was used as an internal control. The primers used in the RT-PCR and qRT-PCR are listed in Supplementary Table 1.

The ORFs of the GhSINAs were cloned and inserted independently into the vector pCambia2300 (CAMBIA) containing the CAULIFLOWER MOSAIC VIRUS (CaMV) 35S constitutive promoter. The primer sequences are listed in Supplementary Table 1. All the constructs were confirmed by sequencing, and then introduced independently into A. tumefaciens GV3101. The Arabidopsis plants (ecotype Col-0) were transformed using the A. tumefaciens-mediated floral dip method (Zhang et al., 2006). Transgenic Arabidopsis were selected on MS medium containing kanamycin, and the selected transgenic seedlings were further screened using genomic PCR. Homozygous T₃ transgenic lines were generated for the functional analysis.

The ORFs of GhSINA7, GhSINA8, and GhSINA9 were fused independently to eGFP in the modified pCambia2300-eGFP expression vector. The primer sequences are listed in Supplementary Table 1. Agrobacterium cells (strain GV3101) containing recombinant plasmids were infiltrated into N. benthamiana leaves. The infiltrated plants were incubated for 48 h at 25°C under dark conditions. Fluorescence signals were then visualized with a confocal laser scanning microscope after three infiltrations (Olympus, Germany).

The available genome data for three cotton species, G. hirsutum (AD1; ZJU assembly), G. arboreum (A2; CRI assembly) and G. raimondii (D5; JGI assembly), were collected from the CottonFGD database¹. The amino acid sequences of five A. thaliana SINAs (AtSINAT1, AtSINAT2, AtSINAT3, AtSINAT4, and AtSINAT5) were used as queries to identify candidates in three cotton protein databases using the BlastP program. The e-value was set at 1e-10. Redundant sequences were removed. The candidates were then filtered to confirm the presence of the conserved RING finger (IPR013083) and SINA domains (PF03145.16; IPR004162) using the Pfam database² and InterPro database³. Subsequently, the coding sequences of all putative SINA genes in G. hirsutum were further verified by cloning and sequencing. For the phylogenetic analysis, all the SINA protein sequences of cotton and Arabidopsis (Supplementary Table 2) were employed to construct an unrooted phylogenetic tree by the neighbor-joining method with 1,000 bootstrap replicates using the MEGA 7.0 software (Hall, 2013; Kumar et al., 2016). For the sequence logo analysis, a multiple sequence alignment was performed using the ClustalX 2.0 software (Larkin et al., 2007). The conserved RING finger and SINA domain sequences of upland cotton SINAs were aligned, and the results were used as the input file in the online tool WEBLOGO⁴ (Crooks et al., 2004).

To investigate the roles of SINA ubiquitin ligases in cotton, 12 G. arboreum, 12 G. raimondii and 24 G. hirsutum SINA genes were identified in the cotton genome database. Then, all the G. hirsutum SINA genes were cloned and sequenced. As a result, the coding sequence of GhSINA8 acquired from sequencing ten clones was shorter compared with its initial annotation in the genome database, while was similar with GaSINA8 and GrSINA8. Therefore, the coding sequence of GhSINA8 acquired from cloning was used in the study. These SINAs encoded two typically conserved domains: a RING finger domain (located toward the N-terminus) and a SINA domain (located toward the C-terminus) (Figure 1A). To examine the features of the homologous domain sequences and the conservation frequency of each residue within the RING finger and large SINA domains, multiple sequence alignments were conducted to generate sequence logos in cotton (Figure 1B). In general, the RING finger domain contained 39 conserved basic residues, whereas the basic region of the SINA domain had 200 conserved residues, which were responsible for interactions with specific target substrates. To further study the evolutionary relationships among SINA proteins in cotton (G. hirsutum, G. arboreum, and G. raimondii) and Arabidopsis, a phylogenetic tree was constructed. All the identified SINA proteins were divided into two groups, with 32 cotton SINA genes (8, 8, and 16 from G. raimondii, G. arboreum, and G. hirsutum, respectively) and 3 Arabidopsis SINA genes in Group I, and 16 cotton SINA genes (4, 4, and 8 from G. raimondii, G. arboreum, and G. hirsutum, respectively) and 2 Arabidopsis SINA genes in Group II (Figure 1C). Twice the number of SINA genes was found in upland cotton than in the diploid cottons G. arboreum (AA) and G. raimondii (DD), which is consistent with the allotetraploid cotton G. hirsutum (AADD)-producing polyploidization event being derived from the natural hybridization of diploid progenitors resembling G. arboreum and G. raimondii. The number of SINA genes is significantly higher in cotton than in Arabidopsis, indicating that SINA gene family members expanded distinctly in the ancestors of cotton and during cotton genome evolution.

The germ tube germination and hyphal growth could be detected on the roots of cotton seedlings at 24 h after inoculation with GFP-labeled V. dahliae (Zhao et al., 2014; Li et al., 2016). To determine whether the expression levels of GhSINA genes changed after 24 h in response to V. dahliae infection, their transcript levels in the roots of the resistant upland cotton cultivar Zhongzhimian No. 2 were investigated. Because of the highly similar sequences of homoeologous gene pairs (similar values > 97%) (Supplementary Figure 1), it was difficult to differentiate between the homoeologs using quantitative real-time PCR (qRT-PCR). Consequently, they were amplified together. Overall, the accumulation level of the combined GhSINA gene transcripts was greater than the control (Figure 2A). Compared with the mock-inoculated controls, the expression levels of GhSINA7, GhSINA8, and GhSINA9 were induced by approximately 3. 5-, 3. 2-, and 2.0-fold, respectively, at 24 h after pathogen inoculation. We then focused on of the roles of GhSINA7, GhSINA8, and GhSINA9 in cotton defense against V. dahliae.

SINA proteins form homodimers as well as heterodimers to accomplish their biological functions (Den Herder et al., 2008; Yang et al., 2015). To determine whether the three candidate GhSINA proteins had the ability to undergo homo- and/or heterodimerization, the pair-wise interactions of GhSINA7, GhSINA8, and GhSINA9 were examined using an Y2H assay. The full-length cDNAs of GhSINA7, GhSINA8, and GhSINA9 genes were fused independently to the DNA-binding domain (BD) bait vector pGBKT7 or the GAL4 activation domain (AD)-containing prey vector pGADT7, and their associations were determined. As shown in Figure 2B, each GhSINA interacted with itself and with the other two GhSINAs to form homo- and heterocomplexes, respectively.

E3 ligases bind to E2 ubiquitin-conjugating enzymes and have the functional enzyme activity for self-ubiquitination. To determine whether the GhSINA7, GhSINA8, and GhSINA9 proteins, which are up-regulated in the presence of V. dahliae, have E3 ligase activities, we expressed SINAs fused to MBPs in E. coli and affinity-purified the MBP-SINAs from the soluble fractions. In the presence of human E1 (UBA1), E2 (UbcH5b), biotinylated-tagged ubiquitin (Bt-Ub) and MBP-SINAs, high molecular mass self-ubiquitinated smear ladders were detected using an anti-biotin antibody (Figures 3A–C, Lane 1), indicating that MBP-SINAs were ubiquitinated. There was no polyubiquitination signal when MBP-SINAs were replaced by MBP or when the E1, E2 or biotinylated-tagged ubiquitin was absent from the reaction (Figures 3A–C, Lanes 2–6). The results implied that the three SINA proteins possess E3 ubiquitin ligase activities.

Ubiquitination usually occurs in the nucleus and cytoplasm to control nuclear and cytoplasmic proteins, respectively (Tanaka et al., 2005; Heck et al., 2010; Yoo et al., 2013). To determine the subcellular localizations of GhSINA7, GhSINA8, and GhSINA9 proteins, the GhSINA7/8/9-eGFP fusion constructs under control of the CaMV 35S constitutive promoter were transiently expressed separately in N. benthamiana leaves. As shown in Figure 4, the green fluorescence of free eGFP was observed in entire cells. Significantly, the three SINA proteins were present in nucleus, which is consistent with their functions in the ubiquitination pathway.

An overexpression strategy was used to assess the functions of the GhSINA genes in defense responses. Owing to the long cotton transformation process, the model plant Arabidopsis was used in this experiment. More than 18 transgenic Arabidopsis lines heterologously overexpressing GhSINA7, GhSINA8, and GhSINA9, independently, were obtained. The two independent homozygous T₃ lines with the highest SINA expression levels were selected for the phenotypic analysis (Figure 5A). The disease resistance levels of SINA-overexpression transgenic plants against V. dahliae were assessed at 18-day after planting. Disease symptoms were observed at 10 days post-inoculation, and overall, the leaves of transgenic plants showed more resistance to V. dahliae, with less wilting, chlorosis, early senescence and necrosis, than those of the wild-type (WT) Col-0 (Figure 5B). The disease tolerance of the GhSINA7 transgenic line increased compared with WT, but it was much lower than in the GhSINA8 and GhSINA9 transgenic lines. The necrosis rates of diseased GhSINA transgenic lines were significantly lower than that of WT (Figure 5C). Moreover, the fungal biomass analysis confirmed that less fungal DNA accumulated in the roots of the transgenic plants, especially in the GhSINA8 and GhSINA9 transgenic lines (Figure 5D). Thus, the ectopic overexpression of GhSINAs conferred greatly enhanced Verticillium wilt resistance in Arabidopsis compared with WT.

The VIGS strategy, which is an efficient method for transient silencing of genes and widely used in cotton research (Li et al., 2019; Long et al., 2019), was employed to investigate the roles of GhSINA genes. Thus, we used VIGS to specifically silence each of the three GhSINA genes (homoeologous gene pairs silenced simultaneously) to study their functions during cotton responses to V. dahliae. The construct TRV:GhCLA1, which produces an obvious photobleaching phenotype when silenced, and empty TRV:00 were used as the positive and mock controls, respectively. At 7 days after agroinfiltration, the leaves of cotton plants injected with TRV:GhCLA1 displayed the expected photobleaching phenotype (Supplementary Figure 2). Additionally, the tender leaves of cotton seedlings infiltrated with different constructs were sampled for RNA isolation and qRT-PCR analyses. The expression levels of GhSINA7, GhSINA8, and GhSINA9 dramatically decreased in their respective VIGS-treated plants compared with the TRV:00 plants (Figure 6A). To investigate the specificity of the VIGS-mediated suppression of the three SINAs, the transcription levels of the non-targeted remaining two SINA genes, which shared high similarity levels (Supplementary Figure 1) with the silenced GhSINA-coding sequence, were detected in VIGS plants. The expression levels of two non-targeted SINA genes were not affected in each of the specifically silenced plants (Supplementary Figure 3).

Subsequently, the control and silenced plants were challenged by V. dahliae. Approximately 2 weeks later, the gene-silenced cotton seedlings, especially those containing TRV:GhSINA8 and TRV:GhSINA9, displayed more severe leaf withering, yellowing and defoliation symptoms, and even death, than control plants (TRV:00) (Figure 6B). The DIs were calculated, and the results indicated that most of the TRV:GhSINA8 and TRV:GhSINA9 plants developed severe disease lesions (Figure 6C). Furthermore, more necrotic vascular tissues were found in GhSINA8 and GhSINA9-silenced cotton plants than in TRV:00 plants (Figure 6D). The fungal recovery assays confirmed that GhSINA8- and GhSINA9-silenced plants were subjected to more fungal colonization than control plants (Figure 6E). Correspondingly, the fungal biomasses were dramatically greater in GhSINA8- and GhSINA9-silenced plants than in control plants (Figure 6F). Collectively, these observations demonstrated that the silencing of GhSINA7, GhSINA8, or GhSINA9 inhibited the plant immune system and enhanced the susceptibility to V. dahliae.

The SINA E3 ubiquitin ligases are ubiquitous moderators that regulate plant growth and stress responses at the post-translational level (Shu and Yang, 2017; Zhang et al., 2019). Here, we report, for the first time, 24 SINA members containing highly conserved RING finger and SINA domains in upland cotton. The qRT-PCR analysis of cotton samples taken at 24 h after V. dahliae inoculation indicated that the GhSINA7, GhSINA8, and GhSINA9 transcript levels were dramatically upregulated compared with uninfected controls (Figure 2A). A phylogram separated the G. hirsutum SINA7, GhSINA8, and GhSINA9 proteins into Group 1, which contains the ortholog of the SINAT5 gene in Arabidopsis (Figure 1B). SINAT5 is a versatile regulator of plant developmental processes, including lateral root production, flowering time control and nodulation formation (Xie et al., 2002; Park et al., 2007; Den Herder et al., 2008). Thus, the SINA E3 ligases of Group 1 may, by modifying distinct substrates, have diverse functions in Arabidopsis and cotton.

The SINA genes encode C3HC4-type RING E3 ligases that are often active as dimers to sustain their own stability in vivo and perform different biological functions (Ning et al., 2011; Yang et al., 2015). The human homolog of SEVEN IN ABSENTIA (hSiah1) was found to oligomerize with itself and heterodimerize with other Sina and Siah proteins in upper eukaryotic cells (Depaux et al., 2006). Oligomerization of a protein does not seem to compete for binding of the substrates, but rather be involved in the formation of a higher ubiquitylation complex, assembling E2 ligases and proteins to be degraded (Depaux et al., 2006; Liew et al., 2010; Yang et al., 2015). Therefore, dimerization of SINA protein may allow simultaneous interaction with multiple proteins. In fact, the RING finger domain is required for the dimer formation of C3HC4 RING finger E3 ubiquitin ligases (Lee et al., 2009; Zhang et al., 2019). C3HC4 RING E3 ligases, such as Siah (Polekhina et al., 2002), RNF4 (Plechanovová et al., 2011), HAF1 (Yang et al., 2015), cIAP (Mace et al., 2008), SlSINA (Wang et al., 2018), MtSINA (Den Herder et al., 2008), and MdSINA (Li et al., 2020), can self-interact to form homodimers or heterodimers with other RING E3 ligases. Because two RING finger domains may be required to spatially accommodate E2 (Duncan et al., 2010; Pao et al., 2018), dimer formation probably depends on the ubiquitination activities of RING E3 ligases in vivo. Using the Y2H assay, we found that GhSINA7, GhSINA8, and GhSINA9 physically interact with themselves or with the remaining two GhSINAs to form homo- or heterodimers (Figure 2B). Thus, GhSINA E3 ubiquitin ligases may form homo- or heterodimeric complexes to perform the ubiquitination function.

Ubiquitin-mediated proteolysis plays a pivotal role in plant adaptability to changing environmental conditions, including exposure to a wide range of pathogens, such as bacteria, viruses and fungi, and insects. The RING-type E3 ubiquitin ligases control nuclear proteins targeted by ubiquitin-proteasome system activities during plant-pathogen interactions. In Arabidopsis, the E3 ligase MIEL1 negatively regulates hypersensitive cell death by ubiquitinating MYB30 (Marino et al., 2013). The transcription factor MYC2, as a master regulator targeted by E3 ligase PUB10, coordinates plant defense and development by repressing defense-related jasmonic acid- and ethylene-responsive genes, respectively (Boter et al., 2004; Lorenzo et al., 2004; Jung et al., 2015). BOI1 (MYB transcription factor BOS1-interacting E3 ligase 1) is involved in the regulation of signaling responses downstream of pathogen perception or cell death (Luo et al., 2010). The Magnaporthe oryzae effector AvrPiz-t represses the detection of pathogen-associated molecular pattern-triggered immunity by suppressing E3 ubiquitin ligase APIP6 in rice (Park et al., 2012). The receptor kinase XA21, regulated by E3 ligase XB3, provides resistance to rice bacterial blight disease caused by Xanthomonas oryzae pv. oryzae (Huang et al., 2013). In wild grapevine (Vitis pseudoreticulata), the RING-type E3 ubiquitin ligase EIRP1 mediates the degradation of transcription factor VpWRKY11 and attenuates the expression of jasmonic acid-responsive genes, which enhances resistance to fungal infections (Yu et al., 2013).

Verticillium wilt is caused by a highly aggressive fungal pathogen, resulting in severely reduced cotton fiber quality and yield worldwide (Fradin and Thomma, 2006; Cai et al., 2009). At almost 2 weeks after inoculation, susceptible cotton plants show visual disease symptoms of cotyledon wilting, leaf chlorosis and seriously stunted growth (Cox et al., 2019). Transgenes, VIGS and infection phenotypes are available approaches used to illuminate the molecular bases of cotton defense against V. dahliae. As shown in Figures 5, 6, the ectopic overexpression of GhSINA7, GhSINA8, and GhSINA9 genes enhanced the tolerance of the transgenic Arabidopsis to V. dahliae. Knockdowns of GhSINA7, GhSINA8, and GhSINA9 genes had compromised resistance to V. dahliae. These findings on the loss- and gain-of-functions of expressed SINAs in response to V. dahliae infection were consistent with GhSINAs being associated with plant defense against V. dahliae. Hereafter, stable GhSINA-overexpression and genome-edited transgenic upland cotton lines will be produced, which will help to ascertain the functions of GhSINAs in response to V. dahliae. Additionally, the specific substrates promoted by GhSINAs for degradation by mono- or polyubiquitination to mediate cotton responses to V. dahliae at the posttranscriptional level is still an urgent topic to investigate.