Arabidopsis thaliana Columbia-0 (Col-0) accession was used for all experiments. Plants were cultured in a growth chamber controlled at 22°C with 60% relative humidity under 12 hours light and 8 hours dark cycle as described previously (Ishiyama et al., 2004a). The light intensity was 40 µmol m^-2 s^-1. Plants grown for two weeks under sterile conditions on MGRL (Molecular Genetics Research Laboratory) medium containing 7 mM nitrate as the nitrogen source (Fujiwara et al., 1992) were subjected to nitrogen starvation for three days prior to the treatment and then transferred to the medium without nitrogen or with 10 mM ammonium chloride (Ishiyama et al., 2004a). Roots of wild-type or transgenic Arabidopsis plants treated with ammonium for 6 hours were used for quantitative real time PCR analysis and visualization of GFP reporter activity. Each medium without nitrogen or with 10 mM ammonium chloride was prepared by adding 10 mM potassium chloride or 10 mM ammonium chloride, respectively, to replace 7 mM potassium nitrate in the MRGL medium.

The fusion gene constructs with various lengths of GLN1;2 (At1g66200), GLT1 (At5g53460) and GLU2 (At2g41220) promoters were generated as follows. The 5’-intergenic regions upstream of coding sequences of GLN1;2, GLT1 and GLU2 were amplified by polymerase chain reaction (PCR) from genomic DNA of Arabidopsis Col-0 accession and cloned as promoter fragments to generate the GFP reporter fusion constructs. PCR was carried out using KOD plus DNA polymerase (Toyobo, Osaka, Japan) and pairs of forward and reverse oligonucleotide ( Table 1 ). The forward primers were designed for amplification of promoter fragments starting 3,583-bp upstream of the translation initiation site of GLN1;2, 2,814-bp, 2,200-bp, 2,100-bp, 2,000-bp, 1,930-bp, 1,730-bp, and 1,050-bp upstream of GLT1, and 1,358-bp upstream of GLU2. Among these forward primers, GLN1;2P3583L_F designed for the amplification of GLN1;2 promoter region has an overhang of a HindIII site (AAGCTT) at the 5’-end. The rest of the forward primers designed for the amplification of GLT1 and GLU2 promoter regions have an overhang of a BamHI site (GGATCC) at the 5’-end. The reverse primers were designed to have the complementary sequences with the 5’-untranslated regions immediately upstream of the translation initiation sites of GLN1;2, GLT1 and GLU2. These reverse primers have an overhang of an NcoI site (CCATGG) at the 5’-end. The ATG in the NcoI site is the translation initiation site for GFP. The amplified PCR products were subcloned into pCR-Blunt II-TOPO (Thermo Fisher Scientific K.K., Tokyo, Japan), and fully sequenced to confirm the identity. These promoter fragments were then cut out as a HindIII-NcoI fragment (for GLN1;2) or BamHI-NcoI fragments (for GLT1 and GLU2) and cloned into respective restriction sites of pTH-10KI, replacing the cauliflower mosaic virus (CaMV) 35S promoter, to obtain the promoter:GFP:teminator cassettes. pTH-10KI is the modified version of CaMV35S-synthetic GFP (sGFP, S65T) vector(Chiu et al., 1996; Niwa et al., 1999) and has a full EGFP coding sequence (Takara Bio Inc. Shiga, Tokyo) between the 35S promoter and the nopaline synthase terminator (NosT). Finally, the promoter:GFP : NosT cassettes created in pTH-10KI were cut out as a HindIII-EcoRI fragment (for GLN1;2) or BamHI-EcoRI fragments (for GLT1 and GLU2) and cloned into pBI101 (Takara Bio Inc.). These binary vector plasmids were introduced into Agrobacterium tumefaciens GV3101 (pMP90) by freeze-thaw method as previously described (Ishiyama et al., 2004a). Arabidopsis plants were transformed according to the floral dip method (Clough and Bent, 1998). Transgenic plants were selected on GM medium (Valvekens et al., 1988) containing 50 mg/L kanamycin sulfate. Kanamycin-resistant T2 progenies were used for analyses.

GFP visualizations in whole plants transformed with the promoter:GFP gene constructs were performed as described previously (Maruyama-Nakashita et al., 2004). Expression of GFP in transgenic plants was visualized using an image analyzer FluorImager 595 under 488 nm excitation (Molecular Dynamics, Sunnyvale, CA, USA). GFP and autofluorescence of the plant were detected by 530DF30 and 610RG filters, respectively (Maruyama-Nakashita et al., 2005). Relative intensity of GFP signals was quantified with IMAGEQUANT software (Molecular Dynamics). Fluorescence of GFP in transgenic plants were observed under a BX61 microscope equipped with a FV500 confocal laser scanning system and 505-525-nm band pass filter (Olympus, Tokyo, Japan), as described previously (Ishiyama et al., 2004a).

Extraction of total RNA, reverse transcription, and real-time PCR were performed as described previously (Maruyama-Nakashita et al., 2004). Total RNA was isolated using RNeasy Plant Mini Kit (Qiagen, Hilden, Germany), and treated with DNaseI (Thermo Fisher Scientific K.K.). Reverse transcription was carried out using Superscript II reverse transcriptase (Thermo Fisher Scientific K.K.) with oligo-d(T)12-18 priming. Real time PCR was carried out using SYBR green PCR master mix and GeneAmp 5700 Sequence Detection System (Applied Biosystems, Foster City, CA, USA). Gene specific primer pairs used are shown in Table 2 . Calculation of mRNA contents was carried out using UBQ2 (At2g36170) as a constitutive internal control.

Yeast one-hybrid (Y1H) screening was performed using the Matchmaker Gold Yeast One-Hybrid Library Screening System (Takara Bio Inc.). The 41 bp ammonium-responsive region (–3,604 to –3,564-bp) found in the GLN1;2 promoter was used as a bait fragment. The bait fragment was cloned into the pAbAi vector (Takara Bio Inc.). The vector was linearized with BbsI and subsequently transferred to Saccharomyces cerevisiae strain Y1H Gold with a LiAc method. Auto-activation was checked by the growth of the bait strain on the medium containing various concentration of Aureobasidin A (AbA). The prey library used in this study was the complete Arabidopsis transcription factor collection (Mitsuda et al., 2010). Approximately 3.7 × 10⁴ transformants were initially screened on SD/−Leu medium containing 500 ng mL⁻¹ AbA for 5 days at 30°C to test the possible interaction. Preys were identified from the positive colonies by DNA sequencing.

Arabidopsis has three GOGAT genes which are the NADH-dependent GOGAT coded by GLT1 and two ferredoxin (Fd)-dependent GOGAT coded by GLU1 and GLU2 in its genome. The qPCR analysis showed that GLT1 was the major GOGAT expressed in Arabidopsis roots in addition stems, flowers, and siliques when the plants were grown hydroponically ( Figure S1A ). In contrast, GLU1 was mainly expressed in young rosette and matured leaves ( Figure S1A ). GLU2 was expressed in both roots and aboveground tissues, while the amount of the transcripts was lower than the other two GOGAT ( Figure S1A ). Our previous study indicates that GLN1;2 is the only form of GS showing increased expression in roots in response to ammonium supply among the five cytosolic GS (GS1: GLN1;1 – GLN1;5) and one plastidial GS (GS2: GLN2) in Arabidopsis (Ishiyama et al., 2004a). To test whether expression of the three GOGAT genes is stimulated by exogenous supply of ammonium, nitrogen-starved Arabidopsis seedlings were exposed to 10 mM ammonium chloride for 6 hours as described previously (Ishiyama et al., 2004a) ( Figure S1B ). Relative to no nitrogen control, the transcript level of GLT1 encoding the NADH-GOGAT significantly increased in roots in response to ammonium ( Figure S1B ). Thus, the ammonium response of gene expression was similar between GLT1 ( Figure S1B ) and GLN1;2 (Ishiyama et al., 2004a). In contrast, the transcript levels of GLU1 and GLU2 encoding the Fd-GOGAT did not change when ammonium was supplied to the roots of nitrogen-starved seedlings ( Figure S1B ).

To investigate cell-type specific ammonium responses of GLN1;2 and GLT1 expression in Arabidopsis, transgenic lines carrying the GLN1;2 or GLT1 promoter:GFP fusion gene constructs were grown with or without ammonium supply ( Figure 1 ). Five independent transgenic lines expressing GFP under control of 5,697-bp GLN1;2 promoter (Konishi et al., 2017) and 2,814-bp GLT1 promoter, respectively, were analyzed to assess their ammonium response in roots. Fluorescence scanning of whole seedlings revealed marked increase in the intensity of GFP signals in roots of both transgenic lines with 6 hours of ammonium treatment following nitrogen starvation ( Figure 1 ). Based on these findings, we carried out confocal laser microscopy analyses of these transgenic lines to identify cell types of the roots where the promoter activity of GLN1;2 and GLT1 are present and can be seen as fluorescence of GFP. Under the control of the GLN1;2 promoter, only faint signals of GFP were detected in the epidermis of elongation zone as well as in the cortex of mature zone of roots when there was no nitrogen supply ( Figures 2A, E ). In contrast, strong GFP signals were detected upon ammonium supply, located specifically in the cortex of the mature zone ( Figure 2B ) and in the epidermis and cortex of the elongation zone of roots ( Figure 2F ). In the GLT1 promoter:GFP lines, fluorescent signals derived from GFP expression were present in the pericycle cells of both elongation and mature zones of roots with no nitrogen supply ( Figures 2C, G ), although even stronger signals were detected in the cortex of matured zone ( Figure 2D ) and in the epidermis and cortex of elongation zone of roots ( Figure 2H ) upon ammonium supply. In root tips, almost no or very faint GFP signals were detected in both promoter:GFP lines ( Figures 2I–L ). These results show that both GLN1;2 and GLT1 are expressed in the surface cell layers of roots, particularly in the epidermis and cortex of roots in response to ammonium supply.

Since glutamine (or a post-glutamine metabolite) has been shown to induce NADH-GOGAT1 expression in rice (Hirose and Yamaya, 1999), we used methionine sulfoximine (MSX), an inhibitor of GS, to investigate whether Arabidopsis GS/GOGAT also fluctuates in expression depending on glutamine or a post-glutamine metabolite. The qPCR analysis was carried out in the presence or absence of 10 mM MSX. As shown previously (Ishiyama et al., 2004a) and in Figure S1 , a substantial increase in GLN1;2 and GLT1 expression was observed at 6-h after ammonium supply in the roots, but this increase was not observed when MSX was added prior to ammonium supply ( Figure S2 ). Over approximately 3-fold increase in GLN1;2 and GLT1 expression was observed after glutamine supply regardless of the MSX pre-treatment ( Figure S2 ). The expression of GLU2 and UBQ2 was stable after ammonium and glutamine supply with or without MSX ( Figure S2 ).

To determine the “ammonium-responsive region” (ARR) in the 5’-intergenic region upstream of the translation initiation site of GLT1, we created a 5’-deletion series of GLT1 promoter:GFP fusion constructs and introduced them into Arabidopsis by Agrobacterium infection. D1, D2, D3, D4, D5, D6 and D7 are the fusion constructs generated to have the GLT1 promoter regions from positions –2,814, –2,200, –2,100, –2,000, –1,930, –1,730 and –1,050 bp, respectively, cloned in front of GFP ( Figure 3A ). Transgenic plants generated with these constructs were grown with or without ammonium supply. Among them, D1, D2 and D3 showed 3.5-fold increase in GFP mRNA levels on ammonium compared to no nitrogen control ( Figure 3B ). In contrast, D4, D5, D6 and D7 showed no induction of GFP expression in response to ammonium ( Figure 3B ). The endogenous GLT1 mRNA levels increased in response to ammonium treatment by 3- to 4-fold relative to no nitrogen control in all GLT1 promoter:GFP lines ( Figure 3B ). Thus, the ammonium-responsive induction of endogenous GLT1 transcript expression was consistent with our previous observation (Konishi et al., 2014).

In our previous study, we found the ARR in the GLN1;2 promoter between the positions –3,604 and –3,564 bp upstream of the translation initiation site (Konishi et al., 2017). To explore potential transcription factor binding sites in the ARR, we performed yeast one-hybrid screening. A GAL4-AD library for yeast one-hybrid system consisting only of Arabidopsis transcription factors (Mitsuda et al., 2010) was screened using the 41-bp region of the GLN1;2 promoter required for the ammonium response (i.e., –3,604 to –3,564 bp) as a decoy sequence. Approximately 3.7 × 10⁴ yeast colonies were screened, and 24 positive clones were obtained, of which 22 gave the insert sequence information ( Table 3 ). Of these 22 Arabidopsis genes identified, 14 were of the AHL family and 3 were DF1. Other transcription factors such as TT16, TTG1, ZFP3, ZFHD2, and LBD22 were also identified as positive clones. Among these transcription factors, a DF1-binding sequences was found between the positions –3,604 and –3,564 bp of GLN1;2 promoter ( Figure 4A ). The DF1-binding sequence was also found within the ARR of GLT1 promoter ( Figure 4B ). AHL and ZFHD2 were predicted to recognize and bind TTTAATT in the decoy sequence ( Figure 4A ). In contrast, binding sites for TT16, TTG1, ZFP3, ZFHD2 and LBD22 were not found in the decoy sequence ( Figure 4A ).

To gain further insights into this ARR, we generated promoter:GFP lines designated T2 with a construct which has the 5’-promoter region of GLN1;2 starting from the position –3,583 bp and lacking the DF1-binding site ( Figures 4A , 5A ). T1 and T3 lines generated previously with the GLN1;2 promoter region with or without the ARR starting from –3,604 and –3,564 bp, respectively, were used in parallel to study the ammonium responsiveness of GFP reporter expression (Konishi et al., 2017). Among these GLN1;2 promoter:GFP lines, only the T1 lines showed a significant increase in GFP mRNA accumulation on ammonium relative to no nitrogen control ( Figure 5B ). In contrast, in the T2 and T3 lines, the ammonium treatment rather resulted in decreasing GFP mRNA accumulation levels down to 0.6- to 0.8-fold of no nitrogen control ( Figure 5B ). In all these GLN1;2 promoter:GFP lines, the endogenous GLN1;2 mRNA levels increased upon ammonium treatment by 3.0 to 4.4-fold relative to no nitrogen control ( Figure 5B ). At the cell-type levels, T1 responded to ammonium and expressed GFP in the epidermis of roots ( Figure 5D ), while T2 did not ( Figure 5H ). In both T1 and T2, the root tip did not respond strongly to ammonium ( Figures 5E, F, I, J ). The cell-type specific patterns of GFP expression identified in T1 lines resembled those with the –5,697 bp GLN1;2 promoter ( Figure 2 ), as both demonstrated strong ammonium response in the epidermis of elongation zone of roots, while they were not exclusively identical in all cell types. These results indicated that the ARR for ammonium-inducible expression is located between the positions –3,604 and –3,583 bp in the GLN1;2 promoter and between –2,100 and –2,001 bp in the GLT1 promoter, respectively.