Landoltia punctata 0202 line and Lemna minor ZH0403 callus were stored in the duckweed resource bank at the Chengdu Institute of Biology, Chinese Academy of Science (Li et al., 2021; Tian et al., 2021). Genomic DNA was extracted from the Landoltia punctata 0202 line cultured in Hoagland medium (15 g.L^-1 sucrose, pH 5.00 ± 0.05). The callus was cultured in Murashige and Skoog (MS) medium with 0.22 mg.L^-1 2,4-Dichlorophenoxyacetic acid, 0.45 mg.L^-1 6-Benzylaminopurine, 30 g.L^-1 sucrose, and 3.5 g.L^-1 gellan gum and then used for genetic transformation. These mediums were placed in a system of photon flux density cycle of 100-120/0 µmol^-2s^-1 (16 h light/8 h dark) and a temperature cycle of 25°C/15°C (16 h day/8 h night).

The 2000 bp promoter DNA sequences upstream from specific genes, sucrose transporter 2 gene promoter named LpSUT2 (Landoltia_punctata_GLEAN_10005267), AGPS1 gene promoter named LpA6804 (Landoltia_punctata_GLEAN_10016804), Rubisco1 gene promoter named LpR1090 (Landoltia_punctata_GLEAN_10021090), Rubisco2 gene promoter named LpR1091 (Landoltia_punctata_GLEAN_10021091), ubiquitin-like protein gene promoter named LpU4817 (Landoltia_punctata_GLEAN_10004817), and ubiquitin-1 promoter named LpU9400 (Landoltia_punctata_GLEAN_10019400) were extracted from the genome of L. punctata by TBtools software (Chen et al., 2020). The cetyltrimethylammonium bromide (CTAB) method was used to extract DNA from L. punctata (Allen et al., 2006). Next, DNA sequences of these promoters were cloned using high PCR fidelity KOD Plus enzymes (TOYOBO, Japan) and recombinant primers ( Supplementary Table 1 ). These amplified products and the linearized vector (the binary vector pCambia2301:35S:eGFP were digested by the restriction enzymes PvuII and PstI) were connected by ClonExpress^® MultiS One Step Cloning Kit (Vazyme, Nanjing, China) and then sequenced.

We referenced previous studies for the protein extraction methods of transgenic duckweed and non-transgenic duckweed (Khvatkov et al., 2018). According to the previously established method (Mahmood and Yang, 2012; Khvatkov et al., 2018), we detected the 35S: ChIL-2:His protein by the Western blot at 100 mg.L^-1 and 500 mg.L^-1 geneticin (G418) from the 35S: ChIL-2:His transgenic duckweed (data were not published). Anti-HIS mouse monoclonal antibody (diluted 1:2,000; Sangon Biotech, China) and Anti-HRP rabbit polyclonal antibody (diluted 1:10,000; Sangon Biotech, China) were used in this work. The blots were visualized using ECL Western blotting detection reagents (Cytiva, America).

Promoter activity was often studied directly using a reporter gene (eGFP or GUS) in stable or transient transformation assays (Nuccio, 2018). As a reporter gene, β-glucuronidase (GUS) is suitable for macroscopic observation of tissue structure in promoter-initiated gene expression. However, it is not suitable for real-time observation of gene expression due to its requirement for steps such as hydrolysis reaction and destaining. On the other hand, enhanced green fluorescent protein (eGFP), as a reporter gene, enables precise mapping at the living cell level, making it suitable for observing the more subtle real-time positional structure of promoter-initiated gene expression. Therefore, we chose eGFP as the reporter gene to provide a more detailed insight into the gene expression initiated by the candidate promoters. To investigate the candidate promoters activity in dicot, we used 35S, LpSUT2, LpR1091, LpA6804, LpU9400, LpR1090, and LpU4817 promoters to respectively drive the reporter gene eGFP, and transiently expressed each in Nicotiana tabacum leaves and left for 2 days. This method was carried out as described in previous studies (Yao et al., 2015). Then we used a laser scanning confocal microscope to observe the fluorescence signal of eGFP.

The vectors of the 35S:eGFP and the LpSUT2:eGFP were introduced into duckweed’s callus following Agrobacterium tumefaciens strain GV3101-mediated transformation. Then, follow the previously established method to screen and regenerate these calluses (Tan et al., 2022). Briefly, put these callus in MS medium (screening 6 weeks) and 1/2 Schenk and Hildebrandt (SH) medium (regeneration 4-5 weeks). Specially, the two-stage culture medium (MS for screening and 1/2 SH for regeneration) contained G418 concentrations of 50, 100, 200, 300, 400, and 500 mg.L^-1, respectively.

Total RNAs were extracted using the Eastep Super Total RNA Extraction Kit (Promega, Shanghai, China) according to the manufacturer’s instructions. Next, we examined their concentration and quality by NanoDrop 2000 (Thermo, USA). All RNA samples were treated with RNase-free DNase to eliminate DNA contamination. The cDNA was synthesized from 1 μg of total RNAs following the manufacturer’s protocol of the GoScript™ Reverse Transcription System (Promega, Shanghai, China). These samples were harvested and immediately frozen in liquid nitrogen, subsequently stored in a -80°C freezer until being used. The L. minor 18S gene was used as an internal control. Primer sequences were listed in Supplementary Table 1 . The qRT-PCR was operated by CFX Maestro Real-Time PCR (Bio-Rad, USA), using the SsoAdvanced™ Universal SYBR Green Supermix (Bio-Rad, USA). The PCR reaction system, procedures, and data analysis followed the SsoAdvanced™ Universal SYBR Green Supermix protocol.

We treated the 35S:eGFP transgenic duckweed at 500 mg.L^-1 G418 with 45.64 mg.L^-1 zebularine (a methylation inhibitor) for 8 days. Then, the transgenic duckweed from the other treatment groups was collected together. Transgenic duckweed genomic DNA was extracted using the CTAB method. The methylation status of promoters was determined by methylation-specific PCR. Briefly, 2 μg of genomic DNA was bisulfite-treated with the Zymo DNA Modification Kit (Zymo Research, Orange County, CA, USA). Bisulfite-treated DNA was amplified using primers specific for either methylated or unmethylated DNA (Baik et al., 2021). The sequences of the first set of methylated-specific (M) primers and unmethylated-specific (U) primers for the promoters are listed in Supplementary Table 1 .

In this work, the fluorescence signal of eGFP protein in all transgenic plants was observed using a laser scanning confocal microscope. To maintain the consistency of the parameters in the obtained fluorescence images, all parameters were set as follows: equipment, Leica TCS SP8; excitation and emission wavelengths, Ex488 nm/Em505-530 nm; software, LAS AF; values of intensity, 30%; the master gain, 800 V; digital gain, 1; pinhole, 1 AU.

We used GraphPad Prism (Version 8.0) and SPSS 26 software for data analysis.

To find a promoter with wide application potential in plants, we selected six duckweed endogenous promoters in contrast to the widely utilized 35S promoter from cauliflower mosaic virus. In parallel, we used the 35S promoter and replaced the 35S promoter with the candidate promoters to drive the eGFP gene into the binary vector pCambia2301 ( Figure 1A ). To test whether the candidate promoters from monocots were suitable in dicots, we transiently expressed these plasmids were transiently expressed in dicots Nicotiana tabacum leaves for 2 days with Agrobacterium tumefaciens-mediated infiltration (Batistič et al., 2012). The images were acquired using consistent acquisition parameters. Under the same validation conditions, we observed that only the eGFP fluorescence signal initiated by the LpSUT2 promoter exhibited comparable intensity to that of the 35S promoter, indicating similar strong initiation abilities. The weaker eGFP fluorescence signals were initiated by the other endogenous promoters (LpR1091, LpA6804, LpU9400, LpR1090, and LpU4817 promoter) ( Figure 1B ). It is possible that they have a limited initiation ability or tissue-specific expression; further studies are necessary to distinguish these possibilities. Based on this result, only the LpSUT2 promoter fulfilled our objective of screening endogenous promoters with strong initiation ability and practicality. Subsequently, we concentrated on comparing the initiation activity and stability of the LpSUT2 promoter and the 35S promoter in duckweed.

Recent studies showed that duckweed has been a widely used bioreactor to express the alpha-interferon, M2e peptide of avian influenza virus H5N1 (Cox et al., 2006; Firsov et al., 2015), etc. Meanwhile, the 35S promoter has been proven to drive the expression of the target proteins in duckweed. However, the 35S promoter also raises concerns about its biosafety since it comes from a virus (Ho et al., 1999), and the low activity in the cell to recognize the sequence as foreign inactivates it (Elmayan and Vaucheret, 1996). Therefore, we expect that finding a plant endogenous promoter will become a more effective tool to produce recombinant proteins. To evaluate the activity of the LpSUT2 promoter in monocot duckweed, we stably transformed the 35S:eGFP (positive control) and the LpSUT2:eGFP vectors into duckweed by Agrobacterium tumefaciens mediated infiltration. Geneticin G418 is an aminoglycoside antibiotic that inhibits protein synthesis by binding to the 30S subunit of the ribosome of a cell. G418 is commonly employed in transgenic plant screening. In our stable transformations in duckweed, G418 was also employed as an antibiotic selection marker. We followed the G418 resistance marker carried by the pCambia2301. Consequently, plants were selected by the addition of G418 to the medium. We collected the images under the same acquisition parameters. Notably, the LpSUT2 promoter was highly active in transgenic duckweed fronds and stronger than the 35S promoter by 100 mg.L^-1 G418 screening ( Figure 2 ).

To improve the efficiency of screening transgenic positive plants and test the stability of the promoter after increasing antibiotic stress, we obtained multiple stable transgenic lines using six concentrations of G418 (50, 100, 200, 300, 400, and 500 mg.L^-1, respectively) during the screening and regeneration process ( Figure 3A ). Next, we used quantitative real-time PCR (qRT-PCR) on the stable transgenic lines from each of the six concentrations to detect the ability of promoters to initiate eGFP expression. The results showed that the expression level of eGFP driven by the 35S promoter decreased with increasing G418 concentration, while the expression level of eGFP driven by the LpSUT2 promoter did not significantly change and was all higher than that driven by the 35S promoter under each condition. Indeed, the expression level of eGFP in the LpSUT2:eGFP transgenic duckweed was 1.76 times that of the 35S:eGFP transgenic duckweed at 100 mg.L^-1 G418, and 6.18 times at 500 mg.L^-1 G418 ( Figure 3B ). Then, we observed the fluorescence signal of the eGFP protein using the laser scanning confocal microscope. The images were collected under the same parameters. Similar to the above qRT-PCR results, the fluorescence signal of the LpSUT2:eGFP transgenic lines had no significant change. However, the fluorescence intensity of the 35S:eGFP transgenic lines became weaker as the concentration of G418 increased. More importantly, the LpSUT2:eGFP transgenic lines fluorescence intensity was consistently stronger than that of the 35S:eGFP transgenic lines in each G418 concentration ( Figure 3C ). Therefore, these data demonstrate that the activity of the LpSUT2 promoter is higher than that of the 35S promoter in the monocotyledonous duckweed.

Based on the above interesting phenomenon, we hypothesized that the ability of the 35S promoter to initiate the expression of foreign proteins would decrease after increasing the antibiotic stress. To test the hypothesis, we used seven transgenic duckweeds with the 35S promoter driving chicken interleukin 2 (ChIL-2): His fusion protein in the pCambia2301 in the laboratory and cultured them at 100 or 500 mg.L^-1 concentrations of G418 (with none other treatments). Interestingly, similar to the 35S:eGFP results, we found that none of the ChIL-2:His fusion proteins in the transgenic duckweed were detectable when the G418 concentration changed from 100 mg.L^-1 to 500 mg.L^-1 ( Supplementary Figure 1 ). Hence, these results indicate that the 35S promoter initiation capacity is unstable under antibiotic stress, which is not conducive to increasing antibiotic stress to rapidly screen positive transgenic plants. In contrast to the 35S promoter, the LpSUT2 promoter exhibits a very stable initiation ability ( Figure 3C ).

Based on the previous experimental results, the expression of the 35S:eGFP decreased with increasing concentration of G418, while the LpSUT2:eGFP remained stable. However, all transgenic duckweed lines exhibit phenotypic similarity to the wild type. There was also not much difference between 100 and 500 mg.L^-1 G418 concentrations in the 35S:eGFP or LpSUT2:eGFP transgenic duckweed phenotypes. Duckweed did not exhibit abnormal growth phenotypes, such as regional albino or death ( Supplementary Figure 2 ).

In Arabidopsis, the 35S promoter was methylated, leading to the suppression of the overexpressed exogenous proteins (Lang et al., 2015). We predicted that the 35S promoter methylation would cause the down-regulation of eGFP expression. To test this prediction, we utilized methylation-specific PCR to detect methylation in the 35S promoter region in 35S:eGFP transgenic duckweed lines grown in 100 mg.L^-1 and 500 mg.L^-1 G418, as well as in the LpSUT2 promoter region in LpSUT2:eGFP transgenic duckweed lines grown in 100 mg.L^-1 and 500 mg.L^-1 G418. The results showed that methylation of the 35S promoter region in the 35S:eGFP transgenic duckweed line grown in 500 mg.L^-1 G418, and no methylation was detected in other transgenic duckweed lines ( Figure 4A ).

Zebularine is a potent DNA methyltransferase inhibitor known to reduce DNA methylation in plants (Baubec et al., 2009). In our study, we supplemented 45.64 mg.L^-1 Zebularine to inhibit methylation in the 35S:eGFP transgenic duckweed line continuously grown in 500 mg.L^-1 G418. After adding the methylation inhibitor, we found a reversion of the 35S promoter region to a non-methylated state ( Figure 4A ).

Additionally, to assess the impact of DNA methylation on the initiation ability of the 35S promoter, we quantified the expression levels of eGFP in the following groups: 35S-100, representing 35S:eGFP transgenic duckweed line grown in 100 mg.L^-1 G418; 35S-500, representing 35S:eGFP transgenic duckweed line grown in 500 mg.L^-1 G418; 35S-500-Z, representing 35S:eGFP transgenic duckweed line grown in 500 mg.L^-1 G418 with 45.64 mg.L^-1 zebularine. Compared with the 35S-100, the expression of the eGFP gene was down-regulated by 2.73 times in the 35S-500 and up-regulated by 8.01 times in the 35S-500-Z ( Figure 4B ). Then, we also observed the eGFP fluorescence signal in these groups. All images were acquired under the same parameters. Consistent with the above qRT-PCR results, 35S-500 shows a weaker fluorescence signal, and 35S-500-Z shows a stronger fluorescence signal compared to 35S-100 ( Figure 4C ). These data suggest that the methylation of the 35S promoter under high levels of antibiotic stress is a significant factor contributing to the reduction in the expression of foreign proteins. This implied that the 35S promoter was methylated and reduced activity, but the LpSUT2 promoter was not ( Figure 4A ). Hence, the LpSUT2 promoter is more advantageous for expressing foreign proteins in plants.