Introduction

Trichomes are small specialized organs composed of either single or multiple cells, and are classified into non-glandular and glandular trichomes (Hülskamp 2004; Olsson et al. 2009). They protect plants from insects, radiation damage and viruses (Traw and Bergelson 2003). Glandular trichomes are generally known as “cell factories”, because they have the abilities to synthesize and store great amount of secondary metabolites, such as artemisinin (Wang 2014). These secondary metabolites, as medicines and foods, have great commercial value (Johnson 1975; Peiffer et al. 2009). Previous researches indicate that one of the most effective methods to increase natural metabolites production is to increase the density of the glandular trichomes (Graham et al. 2010; Yan et al. 2017). Nevertheless, the molecular mechanism on glandular trichome development remains largely unclear.

Artemisia annua L. is a traditional Chinese herbal plant, a natural source plant of artemisinin, a sesquiterpene lactone endoperoxide. The World Health Organization (WHO) has proposed Artemisinin-based combination therapies (ACTs) as the treatment for Plasmodium falciparum malaria and Plasmodium vivax malaria resistant to chloroquine (White 2008; WHO 2017). In addition, artemisinin has been reported to have anti-tumor, hypolipidemic, anti-schistosomiasis effects (Efferth 2006; Singh and Lai 2004; Tang et al. 2010; Xiao 2005). Artemisinin, acting as a promising natural broad-spectrum drug, has great application prospects and commercial value. Although, semi-synthetic artemisinin in yeast has already been established (Paddon et al. 2013; Ro et al. 2006), this technology still cannot replace A. annua as the main and commercial source of artemisinin due to the high production cost (Ro et al. 2006). However, the low content of artemisinin (0.1–1.0% DW) in the natural source plant is a limiting factor for the low cost artemisinin production (Kumar et al. 2004). Therefore, it is urgent and necessary to attain the increment content of artemisinin in A. annua.

After past efforts, the artemisinin biosynthesis pathway has been revealed (He et al. 2017). Previous studies have indicated that the artemisinin synthase genes ADS, CYP71AV1, DBR2 and ALDH1 are specifically expressed in the glandular trichomes (Jiang et al. 2014; Liu et al. 2016; Wang et al. 2013). Trichomes on the surfaces of A. annua leaves are divided into two types: the glandular trichomes and the T-shaped trichomes, which can be easily distinguished due to their phenotypic specificity (Duke and Paul 1993; Schilmiller et al. 2008). Earlier experiments have proved that to attain the increment density of the glandular trichomes in A. annua is an effective approach for increasing the production of artemisinin (Shi et al. 2018; Yan et al. 2017, 2018).

Furthermore, it has been shown that a number of transcription factors (TFs) could regulate the development of the glandular trichomes in A. annua. AaHD1 was identified as a positive regulator of the JA-mediated initiations of glandular trichomes, and overexpression of AaHD1 increased the artemisinin contents in A. annua (Yan et al. 2017). AaMIXTA1 has been found to play an important role in cuticle development and trichome initiation, and AaMIXTA1-overexpression transgenic lines exhibited more cumulative amount of artemisinin than that of the control (Shi et al. 2018). Additionally, AaHD8 has been demonstrated to actively regulate leaf cuticle development and forms a regulatory complex with AaMIXTA1 to promote glandular trichome initiation and cuticle development (Yan et al. 2018).

In plants, a large number of proteins containing zinc finger domain are called as zinc finger proteins. Zinc finger proteins are involved in lots of biological processes, such as tissue development, regulating gene transcription, and stress responsiveness (Berg and Shi 1996; Jenkins et al. 2005; Moore and Ullman 2003). Stress associated proteins (SAPs) containing A20/AN1 zinc finger motif are identified from both prokaryotes and eukaryotes, which are shown to primarily participate in the responses to plant abiotic stresses (Jin et al. 2007; Mukhopadhyay et al. 2004; Solanke et al. 2009; Vij and Tyagi 2006). OSiSAP1, the first A20/AN1 zinc finger protein in plants, is identified from rice and the OSiSAP1-overexpression tobacco showed an increase in the tolerance to salt, water deficit and cold stress (Mukhopadhyay et al. 2004). In Arabidopsis, AtSAP5 had been proved to have the activity of E3 ubiquitin ligase and its expression was induced by various kinds of stresses. The AtSAP5-overexpression transgenic plants showed a stronger tolerance to dehydration pressure than the control (Kang et al. 2011). In addition, SAP genes from alfalfa, tomato and banana were observably induced by environmental factors, such as drought, heat stress, salt stress, oxidative stress, osmotic pressure and mechanical damage (Charrier et al. 2012; Solanke et al. 2009; Sreedharan et al. 2012). Besides, SAPs were also reported to be functional in biological stress responses. Liu et al. identified OsDOG from rice, and the OsDOG expression was induced by the exogenous gibberellin (GA) treatment. The OsDOG-overexpression transgenic lines showed that OsDOG negatively regulated the height and internodes through inhibiting cell elongation (Liu et al. 2011). In Arabidopsis, overexpression of AtSAP9 enhanced the tolerance to dehydration and sensitivity to ABA (Kang et al. 2017; Tyagi et al. 2014). Furthermore, SAPs have been shown to participate in plant development through bio-stress responses (Tyagi et al. 2014). Nevertheless, the function of SAPs in trichome development has not been reported yet.

In this study, we have identified two SAP transcription factors, AaSAP1 and AaSAP2, from A. annua encoding proteins with A20 and AN1 zinc finger motifs. The qRT-PCR analysis showed that the expressions of both AaSAP1 and AaSAP2 were induced by the treatment of MeJA. In A. annua, up-regulation or down-regulation of the transcriptional level of AaSAP1 led to an increase or reduction in both the density of the glandular trichomes and the artemisinin contents, respectively. However, compared with wild-type plants, there was no significant difference on both glandular trichomes density and the artemisinin contents of the AaSAP2 transgenic lines. Taken together, these results indicate that AaSAP1 positively regulates the development of the glandular trichomes in A. annua.

Materials and methods

Plant materials and hormone treatments

The A. annua seeds used for the stable transformation in this study was “Huhao 1”, which has been screened and developed by our laboratory for many years, and its artemisinin content was 0.8–1.0% DW (Shen et al. 2016). Nicotiana benthamiana was used for the transient transformation experiments. Plants were grown in the greenhouse at 25 °C under a 16/8 h light/dark photoperiod at 120 mE m−2 s−2.

For hormone treatments, 1-month-old wild A. annua seedlings were uniformly sprayed with methyl jasmonate (MeJA, 100 μM), abscisic acid (ABA, 100 μM) and gibberellic acid (GA3, 100 μM) solution, and 1% DMSO solution was used as a mork. The young leaves were sampled at 0 h, 0.5 h, 1 h, 1.5 h, 3 h, 6 h, 9 h, 12 h and 24 h after the treatments.

Isolation and characterization of AaSAP1 and AaSAP2

HMM search was carried out using the A20 zinc finger (PF01754) and AN1 zinc finger (PF01428) HMM models downloaded from the Pfam database against our A. annua protein sequence database (Vij and Tyagi 2006). The open reading frames of the predicted SAPs were obtained from the A. annua genome (Shen et al. 2018). The RPKMs of SAPs in distinct tissues of A. annua were obtained from the different tissue transcriptome database. Hierarchical cluster analysis was generated as described before (Zhang et al. 2015). The selected SAPs were further analyzed by using the ClustalX multiple alignment program. The sequences of Arabidopsis SAP genes were obtained from the TAIR (the Arabidopsis Information Resource) database. The SAP protein sequences from both A. annua and Arabidopsis were aligned using ClustalX. The neighbor-joining (NJ) phylogenetic tree was constructed using MEGA7. Bootstrap analysis was set to use 1000 replicates. To obtain ORF of candidate genes, amplification was carried out using cDNA synthesized from 0.5 μg of total RNA extracted from the leaves of A. annua as a template. The primers used in this study are shown in Table 1.

Table 1 Nucleotide sequence of primers used in this study
Full size table

Relative expression analysis via qRT-PCR

Samples of different tissues (stem, root, young leaf, old leaf, bud, flower and trichome) were gathered from the 5-month-old A. annua as reported (Fu et al. 2017). The total RNA was extracted using plant RNAprep pure Plant kit (Tiangen, China). The DNase treatment has been finished before we got the total RNA. According to the manufacturer’s instructions, 80 μL DNase I containing 10 μL DNase and 70 μL RDD was added into the spin columns CR3. The spin columns were kept for 15 min at the room temperature. The quality and quantity of RNA samples were examined using agarose gel electrophoresis (AGE) and the Nanodrop microvolume spectrophotometer. Subsequently, the RNA was reversely transcribed in a 20 µL reaction mixture as previously described. The qRT-PCR analysis was performed as described before (He et al. 2017). The β-ACTIN was used as a standard control in each experiment. Three biological replicate experiments were performed on each sample. Relative expression level analysis was performed as mentioned previously (Livak and Schmittgen 2001; Shen et al. 2016). All primers for qRT-PCR are shown in Table 1.

Subcellular localization of AaSAP1 and AaSAP2

For transient expression, the full-length sequences of AaSAP1 and AaSAP2 without the stop codon were inserted into the pHB-YFP vector to generate pHB-AaSAP1-YFP and pHB-AaSAP2-YFP fusion constructs. Agrobacterium tumefaciens (strain GV3101) harboring these plasmids were separately cultured overnight, further diluted to OD600 = 0.6 using MS liquid medium, and then infiltrated into N. benthamiana leaves along with GV3101 (with p19) strain. The YFP signal was observed using a confocal laser microscope.

Plasmid construction and A. annua transformation

The ORF of AaSAP1 was ligated into SacI and BamHI restriction sites of pHB vector, and the pHB-AaSAP1-YFP construct was used for generating overexpression transgenic plants. The 500 bp fragment encoding the non-conserved domain of AaSAP1 was cloned into the pHB vector between SpeI and BamHI for down-regulating the AaSAP1 expression in A. annua. For the overexpression and antisense constructs of AaSAP2, the construction steps were similar to those for AaSAP1. The recombinant plasmids were introduced into A. tumefaciens (strain EHA105), and then transformed into A. annua as previously reported (Zhang et al. 2009).

For the transformation of A. annua, the seeds were soaked in 75% ethanol for 1 min, followed by rinsing once with the sterile water. Seeds were then treated with 20% sodium hypochlorite solution for 10 min, and washed under sterile water for 4–5 times. The sterilized seeds were plated on MS0 medium (about 55–70 seeds per culture dish) and cultured in an incubation room with a 16 h light/8 h dark cycles at 25 °C. When the seedlings reached a length of 4–5 cm, the leaves were cut out, soaked with EHA105 solution, and then placed on the co-cultivation medium (1/2 MS + 100 μmol/L Acetosyringone) in the dark for 3 days. The infected explants were transferred to the germination screening medium MS1 containing 0.05 mg/L naphthalene-1-acetic acid, 5 mg/L 6-benzoyladenine, 500 mg/L carbenicillin and 50 mg/L hygromycin and the root induction medium MS2 containing 250 mg/L carbenicillin. Subsequently, the generated seedlings were transplanted into the soil for domestication. The detailed MS0, MS1 and MS2 medium used in the study were configured as described previously (Liu et al. 2016; Ma et al. 2019).

The glandular trichome density counting

The fully-expanded leaves (leaf9) were chosen from both the transgenic and wild type A. annua to analyze the glandular trichome density. The glandular trichomes of the leaves were observed by Olympus fluorescence microscope (OLYMPUS BX51), and photographed at the top, middle and bottom of the paraxial surface of each leaf. The counting standard is to calculate the number of glandular trichomes in the range of 1 mm2 using ImageJ. Each sample was taken from the same site of three independent transgenic lines.

Artemisinin content measurement

We collected the leaves of A. annua plants and then dried the leaves at 50–55 °C. The dried materials were ground to powder, and 0.1 g of powder (three biological replicates per sample) was extracted twice with 1 mL methanol (30 min each time) using an ultrasonic machine. Then we filtered the supernatant after centrifugation (12,000 rpm, 10 min). Samples were analyzed as previously described (Zhang et al. 2009).

Results

Isolation and characterization of AaSAP1 and AaSAP2

To identify SAPs in A. annua, HMM search was performed using HMM models (A20 zinc finger and AN1 zinc finger). According to the results, 16 unique genes encoding SAP proteins in A. annua were obtained. Because artemisinin is synthesized in the glandular trichomes (Jiang et al. 2014; Yan et al. 2017), we performed the hierarchical cluster analysis to study the expression patterns of SAP genes from A. annua. As shown in Fig. 1a, only two SAP genes (Aannua01707S242380 and Aannua02879S342930) exhibited high expression in the trichomes of A. annua. These two genes were cloned and assigned as AaSAP1 (Aannua01707S242380) and AaSAP2 (Aannua02879S342930) respectively. The ORF of AaSAP1 was 465 bp, encoding 154 amino acids, and the ORF of AaSAP2 was 510 bp, encoding 101 amino acids. Analysis of the sequences of AaSAP1 and AaSAP2 by ClustalW software revealed that both AaSAP1 and AaSAP2 contained A20 zinc finger and AN1 zinc finger domains (Fig. 1b). A phylogenetic tree was then carried out with 16 SAPs from A. annua and 18 SAPs from Arabidopsis. The result shows that AaSAP1 and AaSAP2 cluster with AtSAP5 which was involved in drought, cold, osmotic pressure and salt stress responses (Fig. 1c).

Fig. 1
figure 1

a Hierarchical cluster analysis of SAP genes in different tissues of A. annua. The top color scale represents the value of logarithmic conversion reads per kilobase per million mapped reads. b Alignment of AaSAP1 and AaSAP2 protein sequences with sequences from Arabidopsis (AT3G12630, O49663), Oryza sativa (XP_015651267, A2YEZ6), Zea mays (ABS83245). The two conserved zinc finger domains (A20 and AN1 zinc finger domains) are marked by red boxes. c Phylogenetic relationship between the SAP protein sequences of A. annua and Arabidopsis. This bootstrapped tree with 1000 replicates was made using Clustal W and MEGA 7 tools. (Color figure online)

Full size image

Localizations of AaSAP1 and AaSAP2

In order to investigate the localizations of AaSAP1 and AaSAP2, we fused these two genes with YFP respectively and carried out the transient expression analysis in tobacco leaves. The epidermal cells of tobacco leaves were detected by laser confocal microscopy. The results showed strong YFP fluorescence in both the nucleus and the cytoplasm of tobacco injected by AaSAP1-YFP plasmid, exhibiting that AaSAP1 was localized in both the nucleus and the cytoplasm, while AaSAP2 was localized in the nucleus (Fig. 2). The results indicate that these two SAPs are likely to have different functions.

Fig. 2
figure 2

Subcellular localization of 35S: AaSAP1-YFP and 35S: AaSAP2-YFP in tobacco leaf epidermal cells. 35S: YFP was used as a negative control. Scale bars: 50 µm

Full size image

Expression pattern of AaSAP1 and AaSAP2

In order to investigate the expression of AaSAP1 and AaSAP2, qRT-PCR was used to analyze the expressions of AaSAP1 and AaSAP2 in different tissues. The results showed that the highest expression levels of AaSAP1 were detected in the trichome, followed by flower, root and stem (Fig. 3a), and the AaSAP2 expression levels were also the highest in the trichome, followed by stem, young leaf and old leaf (Fig. 3b).

Fig. 3
figure 3

a, b The expression patterns of aAaSAP1 and bAaSAP2 in young bud (bud0), mature bud (bud1), old leaf, young leaf, flower, stem, root and trichome of A. annua measured by qRT-PCR. c, d The expressions of cAaSAP1 and dAaSAP2 in response to MeJA, ABA and GA3 treatments. The β-ACTIN was used as the internal reference. The mock used in this study was 1% DMSO solution. Values are mean ± SD (n = 3)

Full size image

MeJA, ABA and GA3 treatments were given to one-month-old seedlings, and the expressions of AaSAP1 and AaSAP2 were analyzed by qRT-PCR. The transcript level of AaSAP1 showed an increasing trend gradually, reached a maximum at 9 h, and then gradually decreased with the treatment of MeJA (Fig. 3c). ABA and GA3 also upregulated the AaSAP1 transcript level, with the highest point occurring 1.5 h after the treatment respectively (Fig. 3c). Similarly, the expression level of AaSAP2 also increased after the MeJA and GA3 treatments and reached a peak at 6 h and 1.5 h respectively (Fig. 3d). However, the ABA treatment had no significant effect on the expression level of AaSAP2 (Fig. 3d).

Transformation of A. annua

In order to further explore the functions of AaSAP1 and AaSAP2, the anti-sense (35S: anti-AaSAP1 and 35S: anti-AaSAP2) and overexpression (35S: AaSAP1 and 35S: AaSAP2) transgenic A. annua lines driven by the CaMV35S promoter were generated. Firstly, 55-70 sterilized A. annua seeds were sown on the medium (Fig. 4a), and then the generated seedlings were transformed with the constructed pHB-AaSAP1-YFP, pHB-anti-AaSAP1, pHB-AaSAP2-YFP and pHB-anti-AaSAP2 vectors in EHA105 strains using the leaf disc transformation (Fig. 4b–d). After generating roots, the transgenic plants were transferred into the soil. Totally, more than 30 independent A. annua transgenic plants for each vector were generated.

Fig. 4
figure 4

a The sterilized seeds were plated on MS0 medium. b The leaf discs from generated seedlings in co-cultivation medium. c The resistant shoots were generated on MS1 selection medium. d The small plantlets were transferred into MS2 rooting medium

Full size image

AaSAP1 promotes the development of glandular trichomes in A. annua

The qRT-PCR analysis showed that the AaSAP1 expression was dramatically decreased in AaSAP1-anti transgenic lines, and AaSAP1-OE transgenic lines exhibited a 69–73% increase of AaSAP1 transcript levels (Fig. 5a). The transcript levels of AaSAP2 was reduced to 40–56% of the control in AaSAP2-anti transgenic lines, and AaSAP2-OE transgenic plants also exhibited an increase in the AaSAP2 expression (Fig. 5b). The glandular trichomes were observed by Olympus fluorescence microscope, and it was found that, compared with the control, the density of the glandular trichomes in AaSAP1-OE transgenic lines was 1.3–1.5 times higher than those of the control, while the glandular trichomes density decreased by 22–54% in the AaSAP1-anti transgenic plants (Fig. 5c, d). However, the density of the glandular trichomes was no significantly different between AaSAP2 transgenic plants and the control (Fig. 5c, e). Taken together, these results indicate that AaSAP1 positively regulate glandular trichome development in A. annua.

Fig. 5
figure 5

a Expression levels of AaSAP1 in AaSAP1 anti-sense (AaSAP1-anti) and overexpressing (AaSAP1-OE) plants were analyzed using qRT-PCR. b Expression levels of AaSAP2 in AaSAP2 anti-sense (AaSAP2-anti) and overexpressing (AaSAP2-OE) plants were analyzed using qRT-PCR. EV was the empty vector as the control. The internal reference was β-ACTIN. c The glandular trichomes on the mature leaves from EV, AaSAP1-anti, AaSAP1-OE, AaSAP2-anti and AaSAP2-OE plants were captured by microscopy. Yellow autofluorescence represented glandular trichomes and red autofluorescence represented chlorophyll. Bars are given as 200 µm. d Density of glandular trichomes on mature leaves from EV, AaSAP1-anti and AaSAP1-OE plants were counted. e Density of the glandular trichomes on mature leaves derived from EV, AaSAP2-anti and AaSAP2-OE plants were counted. Values are mean ± SD (n = 3). Student’s t test compared with EV (*P < 0.05, **P < 0.01). (Color figure online)

Full size image

AaSAP1 increases artemisinin content in A. annua

It was reported that increasing the density of the glandular trichomes in A. annua was one of the effective approaches to increase the production of artemisinin (Graham et al. 2010; Shi et al. 2018). The artemisinin contents of A. annua transgenic plants were determined by HPLC. As shown in Fig. 6, the artemisinin contents in AaSAP1-OE transgenic lines were significantly increased, two fold of those in the control (7.88 mg/g DW), while the content of artemisinin was reduced by 5–22% of the control in the AaSAP1-anti transgenic lines. These results indicate that AaSAP1 is a positive regulator for the glandular trichome development, and is also valuable to be used in genetic engineering of A. annua for enhanced the production of artemisinin.

Fig. 6
figure 6

Artemisinin content of EV, AaSAP1-anti and AaSAP1-OE plants. Values are mean ± SD (n = 3). Student’s t test compared with EV (*P < 0.05, **P < 0.01)

Full size image

Discussion

Trichomes are plant-specific organs that protect plants from biotic and abiotic stresses, and are divided into the glandular trichomes and the T-shape trichomes in A. annua. Artemisinin is an important ingredient in the treatment of malaria, with huge annual demand. In order to enhance the content of artemisinin in A. annua, scientists all over the world have made a lot of efforts. Previous study has shown increasing the glandular trichome density can result in the increase of artemisinin content in the plant (Graham et al. 2010).

Plant growth and development frequently face challenges from variety of biotic and abiotic stresses. SAP is a category of zinc finger proteins that plays a crucial role in the abiotic stress response of various plants (Giri et al. 2011, 2013; Mukhopadhyay et al. 2004). In this study, we identified 16 SAPs from the A. annua protein database, two of which (AaSAP1 and AaSAP2) showed high expression in the glandular trichomes of A. annua by expression analysis. The sequences of both AaSAP1 and AaSAP2 had similarity to Arabidopsis AtSAP5 which involved in drought, cold, osmotic pressure and salt stress responses (Kang et al. 2011). Similarly, the SAP genes of alfalfa, rice, tomato and banana also respond to heat stress, salt stress, osmotic pressure, ABA and mechanical damage (Charrier et al. 2012; Solanke et al. 2009; Sreedharan et al. 2012; Tyagi et al. 2014).

Phytohormones such as MeJA, ABA and GA3 regulate a lot of fields of plant growth and development and response to biotic and abiotic stresses (Hong et al. 2012; Wasternack and Hause 2013; Yan et al. 2017; Zhu 2002). Our results indicate that, like the SAP gene described above, both AaSAP1 and AaSAP2 respond to the induction of MeJA and GA3. ABA also induced the expression of AaSAP1. However, the expression level of AaSAP2 did not change significantly after the treatment of ABA.

In order to explore the function of AaSAP1 and AaSAP2, AaSAP1 and AaSAP2 overexpressing and antisense transgenic A. annua plants were generated. We found that the glandular trichome density of AaSAP1 antisense plants significantly decreased compared to the empty control, but increased in the overexpressed plants. However, the density of the glandular trichomes did not change significantly in the AaSAP2 transgenic plants. In addition, HPLC analysis showed that the content of artemisinin increased in overexpressed plants and drastically decreased in antisense plants of AaSAP1. These results reveal that the overexpression and repression of AaSAP1 can affect the density of the glandular trichomes and artemisinin content in A. annua, and AaSAP1 is the first SAP gene found to affect the glandular trichome development. We also provide an effective strategy to improve the content of artemisinin by overexpressing AaSAP1 in A. annua by genetic engineering.