Background

Rice seeds accumulate large amount of storage proteins, including glutelin, prolamin, and α-globulin, which supply nutrients for seed germination and seedling growth. Up to 80% of the total seed storage proteins are made up of glutelins which are important protein sources for human consumption because of their easy digestibility. Glutelins are synthesized as 57 kD precursors on the rough endoplasmic reticulum (RER) and transported to PBIIs by DV-mediated post-Golgi transport pathway or ER-derived precursor-accumulating compartments [1,2,3,4,5,6].

Rice 57H mutants are characterized by over-accumulation of 57 kD glutelin precursors in seeds, which are excellent genetic resources to dissect the glutelin vacuolar transport pathway. Up till now, nine 57H mutants have been reported in rice, including gpa1/glup4, gpa2/glup6, gpa3, gpa4/glup2, W379/glup3, esp2, glup1/esp5, Glup5, and glup7 [5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20]. Among which four cloned genes have been shown to regulate the proglutelin trafficking events. GPA4 encodes GOLGI TRANSPORT 1B which regulates protein export from the ER [17, 20]. GPA3, a plant-specific Kelch-repeat domain containing protein, acts as a scaffold to recruit guanine-nucleotide exchange factor (GEF) GPA2/OsVPS9A, which in turn activates a small GTPase GPA1/OsRab5a [5, 6, 16, 18, 19]. GPA3, GPA2, and GPA1 proteins form a functional complex on the DVs to regulate the vacuolar trafficking of proglutelins [6]. In addition, ESP2 encodes protein disulfide isomerase-like1–1 (PDIL1–1) which regulates the disulfide bond formation in ER [3]. W379 encodes a vacuolar processing enzyme which processes proglutelins into acidic and basic subunits in the PBII [8, 15]. Despite these advances, molecular mechanisms underlying glutelin trafficking are still elusive.

Na+/H+ antiporters (NHX antiporters) are H+-coupled cotransporters that transfer Na+ or K+ across membrane in exchange for H+ [21]. In plants, NHX antiporters are essential for cellular pH and ion homeostasis. They play important roles in various cellular processes, such as Na+, K+ movement, pH homeostasis, vesicular trafficking and protein targeting, stress response, plant growth and development [22,23,24,25,26,27]. Based on the subcellular localization, Arabidopsis NHX antiporters are classified into three subgroups which localized to the vacuoles (AtNHX1–4) [28,29,30], plasma membrane (AtNHX7/8) [24, 30, 31], and endosomal compartments (AtNHX5/6) [32], respectively. Particularly, AtNHX5 and AtNHX6 are localized at the Golgi, TGN and required for cell expansion, response of salt stress as well as vesicular trafficking [32]. These two proteins are also localized at the PVC and important for maintaining endomembrane luminal pH and receptor-mediated protein trafficking to the vacuole [33]. In addition, AtNHX5 and AtNHX6 are require for seed storage protein processing [34]. In rice, overexpression of OsNHX1 enhanced tolerance to salt stress [35]. However, the function of endosomal NHX proteins in rice remain largely unknown.

In this study, we report the functional characterization of a rice gpa6 mutant that accumulated a large amount of proglutelins in the mutant endosperm cells, and demonstrate that GPA6 encodes a Golgi-, TGN- and PVC-localized Na+/H+ antiporter OsNHX5 which is essential for endomembrane luminal pH homeostasis and proglutelin vacuolar trafficking.

Results

gpa6 seeds accumulate proglutelins and develop abnormal endosperm

A 57H mutant named gpa6 was isolated during our continuous effort to dissect the glutelin trafficking pathway in rice. Unlike the transparent endosperm of wild type, gpa6 mutant endosperm appeared floury (Fig. 1a). Scanning electron microscopy (SEM) analysis revealed that gpa6 endosperm comprised round and loosely packaged compound starch granules instead of the tightly packaged, crystal-like structures observed in the wild type (Fig. 1b and c). Meanwhile, the 1000-grain weight was significantly decreased, and the amylose content was reduced approximately 20% and the lipid content was increased 81% in the gpa6 mutant. However, the total protein content in the endosperm was not changed (Additional file 8: Table S1). Compared with the wild type, gpa6 accumulated much higher level of unprocessed 57 kD proglutelins, accompanied by concomitant reduction of both 40 kD acidic and 20 kD basic subunits of the mature glutelins (Fig. 1d). The appearance of higher amount of proglutelins was further confirmed by immunoblotting using antibodies against 40-kD glutelin acidic subunits and 20-kD glutelin basic subunits (Fig. 1e). Time course studies revealed that the abnormal accumulation of proglutelins in gpa6 seeds occurred from ~ 12-DAF onwards (Additional file 1: Figure S1). In addition, the expression of representative genes coding for storage proteins, and the protein level of ER lumen BINDING PROTEIN1 (BiP1) and PDI1–1 showed almost no differences between gpa6 and wild-type seeds (Fig. 1e, Additional file 2: Figure S2), indicating that the accumulation of 57 KD proglutelins was not due to an increasement of storage protein coding gene expression and probably protein synthesis was normal as no ER stress was detected. Collectively these data suggested that in the gpa6 mutant might be defective in the trafficking of proglutelins.

Fig. 1
figure 1

Characterization of the gpa6 Mutant. a Transverse sections of representative wild-type (Indica cultivar N22) and gpa6 mutant dry seeds. Bars = 1 mm. b and c Scanning electron microscopy images of transverse sections of wild-type (b) and gpa6 mutant (c) seeds. Bars = 10 μm. d SDS-PAGE profiles of total seed storage proteins of the wild type and gpa6 mutant. pGlu, 57-kD proglutelins; αGlu, 40-kD glutelin acidic subunits; αGlb, 26-kD α-globulin; βGlu, 20-kD glutelin basic subunits; Pro, prolamins. e Immunoblot analysis of glutelins and the molecular chaperones BiP1 and PDI1–1. Arrowheads represent glutelin basic subunits. Arrows indicate the 57-kD proglutelins (red) and the glutelin acidic subunits (black). EF-1α was used as a loading control

Full size image

The gpa6 mutant is defective in post-Golgi trafficking of storage proteins in developing endosperm

To gain an overview of glutelin deposition, semi-thin sections (1 μm) of 12-DAF wild-type and gpa6 mutant endosperm were prepared and subjected to the immunofluoresence staining with specific antibodies against prolamins and glutelin acidic subunits (Fig. 2). In gpa6 mutant, the PBIIs (containing glutelins) were reduced to 53% of the wild type (Fig. 2a-f, g), while the PBI sizes (containing prolamins) were comparable (Fig. 2a-f, h). Large glutelin-containing paramural bodies (PMBs) were readily observable (Fig. 2f). In addition, α-globulins were also transported incorrectly to the PMBs rather than PBIIs in gpa6 (Additional file 3: Figure S3). Consistent with the proglutelin trafficking defects, Pectins labeled with JIM7 were seen to accumulate inside the PMBs in gpa6 rather than display an even distribution along the wild type cell wall (Additional file 4: Figure S4) [36].

Fig. 2
figure 2

Immunofluorescence microscopy of protein bodies in the subaleurone cells of the wild type and gpa6 mutant. a to f Immunofluorescence microscopy images of storage proteins in wild-type (a-c) and gpa6 (d-f) 12 DAF seeds. a, d Secondary antibodies conjugated with Alexa fluor 488 (green) were used to trace the antigens recognized by the anti-glutelin antibodies. (b, e) Secondary antibodies conjugated with Alexa fluor 555 (red) were used to trace the antigens recognized by the anti-prolamin antibodies. (c, f) Merged images. White arrows in (f) indicate the PMB structures. Bars = 10 μm (a-f). (g) and (h) Statistical analysis of the diameters of PBIIs (g) and PBIs (h). Values are means ± SD. **P < 0.01 (n > 350, Student’s t test)

Full size image

Next, subcellular observation by transmission electron microscopy (TEM) was performed using ultra-thin sections of 12-DAF developing endosperm. In wild-type endosperm cells, there are irregularly shaped, fully filled PBIIs and round spherical PBIs (Fig. 3a), however, in the gpa6 mutant, PBIIs were only partially filled with the storage proteins (Fig. 3b), which was accompanied by the presence of PMBs (formed by the clustered DVs) and secreted oval-shaped structures along the cell wall (Fig. 3g and i). To determine structural alterations in the endomembrane system that could account for the missorting of storage protein precursors to the apoplast, we analyzed 12-DAF developing endosperm by immunogold labeling. Indeed, in the gpa6 mutant, only part of glutelins was correctly transported to the PBIIs (Fig. 4a and b), meanwhile, mis-sorted DVs and PMBs were found to contain glutelins (Fig. 4c and d). These results demonstrated that proglutelin were mis-sorted, and delivered to the apoplast.

Fig. 3
figure 3

Ultrastructure of subaleurone cells of developing endosperm of the wild type and gpa6 mutant. a and b Two types of protein bodies were observed in wild-type (a) and gpa6 mutant (b) endosperm. Bars = 2 μm. CW, cell wall. c Wild type. Bars = 400 nm. d DVs bud off from the Golgi in the gpa6 mutant. Red Arrow indicates enlarged DVs. Bars = 400 nm. e Large clusters of DVs (blue arrow) in the gpa6 mutant. Bars = 400 nm. f and g Electron micrographs showing that DVs can fuse with the PM (f) and expel their contents into the apoplast forming oval-shaped structures (Arrowheads) (g) in the gpa6 mutant. Bars = 1 μm. h Two DVs are fused with each other (rectangular box) in the gpa6 mutant. Bars = 1 μm. (i) The PMB structures in the gpa6 mutant. Bars = 1 μm

Full size image
Fig. 4
figure 4

Immunoelectron microscopy localization of glutelins in rice endosperm cells. a Glutelins were accumulated in PBIIs in the wild-type endosperm cells. CW, cell walls. Bars = 500 nm. b Size-reduced PBII containing glutelins. Bars = 500 nm. c Glutelins in DVs (red arrows) and oval-shaped structures (black arrows). Bars = 500 nm. d Glutelins in the PMBs. Bars = 500 nm. 10-nm gold particle conjugated secondary antibodies were used in (a–d)

Full size image

Proglutelins are delivered to the PBII via DV-mediated trafficking pathway [5, 6], therefore, DV morphology are carefully examined. Statistical analysis showed that the average diameter of DVs near the Golgi apparatus, and presumably just budding from the Golgi apparatus in gpa6 is 193 nm (n = 75, Fig. 3d), which is larger than that of wild type (average 154 nm, n = 56, Fig. 3c), and these DVs tends to cluster in the cytosol (Fig. 3e). In addition, numerous DVs at an average diameter of 198 nm were found just fused with the plasma membrane, and kept spherical shapes (Fig. 3f), gradually, the secreted storage proteins became oval-shaped (Fig. 3g). Occasionally, fusion between two DVs was detected (Fig. 3h), this might account for the slight size increasement of DVs along the plasma membrane (198 nm) versus newly budded ones (193 nm). In short, DVs in gpa6 mutant are enlarged.

Altogether, these results demonstrated that in gpa6 mutant enlarged glutelin-containing DVs were mis-sorted to the apoplast, thus led to the reduction of PBII size.

Map-based cloning of GPA6

The gpa6 mutant was isolated from a 60Co-irradiated population of indica variety N22. Genetic analysis revealed that the mutant phenotype was inherited as a recessive mutation (Additional file 9: Table S2). For map-based cloning, we crossed gpa6 with the japonica variety Nipponbare to generate 208 F2 recessive individuals. The GPA6 locus was mapped to chromosome 9 and further fine-mapped to a 98-kb region (Fig. 5a). DNA sequencing revealed a 7 bp deletion in the sixth exon of Os09g0286400, generating a premature stop codon that led to a truncated product with 199 amino acids (Fig. 5b). Three independent transgenic lines bearing Ubiquitin promoter driven Os09g0286400 open reading frame (ORF) rescued the gpa6 mutant phenotypes, including the floury appearance of endosperm (Fig. 5c), the accumulation of proglutelins (Fig. 5d) and the abnormal glutelin deposit pattern (Fig. 5e and f). Therefore, Os09g0286400 is the gene responsible for gpa6 mutant phenotypes.

Fig. 5
figure 5

Map-based cloning of GPA6. a Fine mapping of the GPA6 locus. The molecular markers and the number of recombinants are shown. b Gene structure and the mutation site in Os09g0286400. Os09g0286400 comprises 22 exons (closed boxes) and 21 introns (lines). ATG and TGA represent the start and stop codons, respectively. A 7 bp deletion in the sixth exon of Os09g0286400 in gpa6. c-f The Os09g0286400 ORF under the control of ubiquitin promoter rescues the grain appearance (c), the storage protein composition pattern (d), the ultrastructures of endosperm cells in the wild type (e) and L1 (f). L1 to L3 denote the grains from three independent T1 transgenic lines. Red arrows in (d) indicate the 57-kD proglutelins. Bars = 1 mm in (c). Bars = 2 μm in (e) and (f)

Full size image

GPA6 encodes OsNHX5 that localizes to the Golgi, TGN and PVC

GPA6 encodes a NHX antiporter homologous to the endosomal localized AtNHX5, and was named OsNHX5. OsNHX5 is predicted to have 9 putative transmembrane domains and the mutation in gpa6 leads to the deletion of the last four. Real-time PCR analysis revealed that OsNHX5 is expressed in all tissues examined (Fig. 6b). During endosperm development, the expression of OsNHX5 was low at the early stages, peaked at ~ 15-DAF, and decreased at ~ 18-DAF (Fig. 6b), which is correlated with the accumulation of glutelins. The rice genome has another putative endosomal antiporter OsNHX6 (Fig. 6a, Additional file 5: Figure S5). Although it is ubiquitiously expressed, its expression is much lower than OsNHX5 (ratio of expression level of OsNHX5/OsNHX6, root: 91; stem: 25; leaf: 232; leaf sheath: 15; panicle: 18; 12-DAF endosperm: 26) (Fig. 6b).

Fig. 6
figure 6

Phylogenetic analyses and spatial expression patterns of OsNHX5. a A neighbor-joining tree of OsNHX5 and its homologs. The tree was constructed using MEGA and bootstrapped with 1000 replicates. b Wild-type expression levels of OsNHX5 and OsNHX6 in various organs and different developmental stages of endosperm. R, root; S, stem; L, leaf; LS, leaf sheath; P, panicle; DAF, days after flowering. Actin1 was used as an internal control. For each RNA sample, three technical replicates were performed. Values are means ± SD

Full size image

To determine the subcellular localization of OsNHX5, OsNHX5 coding sequence was fused to the N-terminus of GFP to obtain p35S:OsNHX5-GFP construct. After transformed into gpa6, it completely rescued the mutant phenotype, indicating that OsNHX5-GFP is functional in vivo (Additional file 6: Figure S6). Unfortunately, GFP fluorescence in the transgenic lines was too weak to be detected. Transiently expressed OsNHX5-GFP in N. benthamiana leaf epidermal cells was examined instead, and OsNHX5-GFP was shown to partially colocalized with the Golgi, TGN and PVC markers, respectively (Pearson’s correlation coefficient [PSC], Golgi: PSC = 0.608 ± 0.08; TGN: PSC = 0.753 ± 0.14; PVC: PSC = 0.587 ± 0.11, Fig. 7). Similarly, OsNHX6 was localized to the Golgi, TGN, and PVC as well (Additional file 7: Figure S7).

Fig. 7
figure 7

Subcellular Localization of OsNHX5 in N. benthamiana protoplasts. a to c Confocal microscopy images showing that OsNHX5-GFP is localized as punctate signals in the cytosol and its distribution partially overlaps with the markers for Golgi (ST-mRFP [a]), TGN (mRFP-SYP61 [b]) and PVC (mRFP-VSR2 [c]). Bars = 10 μm (a-c)

Full size image

Given the fact that OsNHX5 mutation alone displayed phenotypes and OsNHX6 is much lower expressed, OsNHX5 might be a predominant endosomal localized NHX antiporter in rice.

The luminal pH of Golgi, TGN and PVC is more acidic in the gpa6 protoplasts

In order to determine whether the cellular pH was affected in gpa6, we use noninvasive live-cell imaging to measure pH by the pHluorin-based pH sensor [37, 38]. pH sensors ManI-PRpHluorin, PRpHluorin-BP80 (Y612A) and PRpHluorin-AtVSR2 were used to measure pH of Golgi, TGN and PVC, respectively in the rice protoplasts [38]. The calibration curve was acquired by calculating pH-dependent fluorescence ratios (Fig. 8a). Our results showed that the pH of the Golgi, TGN, and PVC was more acidic in gpa6 (Golgi: 6.45 ± 0.17; TGN: 6.01 ± 0.16; PVC: 5.87 ± 0.19) than in the wild type (Golgi: 6.95 ± 0.21; TGN: 6.36 ± 0.18; PVC: 6.31 ± 0.14) (Fig. 8d) where the △pH = 0.35–0.50. Representative pseudocoloured images of PRpHluorin-BP80 (Y612A) were shown (Fig. 8b and c). These results clearly indicated that OsNHX5 regulates the pH homeostasis of the Golgi, TGN, and PVC in rice protoplasts.

Fig. 8
figure 8

OsNHX5 regulates the pH of Golgi, TGN and PVC. a In vivo calibration curve of pH. pH calibration was achieved by equilibrating intracellular pH with 10 μM nigericin, 60 mM KCl, and 10 mM MES/HEPES Bis-Tris-propane, pH 5.0 to 8.0. (mean ± SD; n ≥ 25 protoplasts). b and c Representative pseudocolored images of PRpHluorin-BP80 (Y612A) in wild-type (b) or gpa6 (c) protoplasts. Bars = 5 μm. d pH of the Golgi, TGN and PVC. (mean ± SD; n ≥ 25 protoplasts; **P < 0.01; Student’s t test)

Full size image

Discussion

gpa6 is defective in post-Golgi trafficking of storage proteins in rice endosperm cells

Characterization of 57H mutants facilitates us to understand the molecular mechanism of proglutelin trafficking and processing in rice endosperm cells. In this study, we isolated a 57H mutant gpa6 that accumulated large amount of proglutelins in rice endosperm. The comparable protein levels of BiP1 and PDI1–1 between wild type and the gpa6 mutant suggested a normal ER function in gpa6. Consistent with this notion, gpa6 developed normal ER-derived PBIs. In gpa6 mutant, large amount of DVs accumulated in the cytosol, suggesting that the trafficking rate is largely reduced. Furthermore, DVs were mis-sorted to the apoplast, leading to smaller PBIIs. These abnormalities are very similar to gpa1/osrab5a, gpa2/osvps9a, and gpa3/kelch mutants [5, 6, 16, 18, 19]. Taken together, gpa6 is defective in proglutelin post-Golgi vacuolar trafficking pathway in rice endosperm cells.

OsNHX5 is a predominant intracellular NHX antiporter that regulates luminal pH of several endomembrane compartments

Similar to animals, plant cells also have specific pH in different endomembrane compartments along the secretory pathway [37,38,39,40]. In this study, using noninvasive live-cell imaging and pH sensors, we found that the pH values of Golgi, TGN and PVC in wild-type rice protoplasts were not significantly different from their counterparts in Arabidopsis protoplasts (Golgi: 6.8; TGN: 6.3; PVC: 6.2) [38]. However, all three compartments had a more acidic pH (△pH = 0.35–0.50) in osnhx5, indicating that OsNHX5 plays an important role in the alkalization of these compartments. Given the fact that atnhx5 atnhx6 had a more acidic pH in those compartments as well (△pH = 0.25–0.40), endosomal antiporters seem to have conserved functions in rice and Arabidopsis cells.

In atnhx5 atnhx6 double mutant, lowered pH led to a compromised receptor–cargo association [33]. The phenotype of storage protein trafficking defects were similar between osnhx5 and atnhx5 atnhx6. Therefore, the acidification of Golgi, TGN, particularly PVC may result in reduced VSR (Vacuolar sorting receptor)-proglutelin association, although the receptors for proglutelins remains to be characterized. Previous studies showed that OsNHX5 has a K+ and Na+ transport activity [41], thus it is possible that ionic changes might also affect VSR-cargo interactions.

Previous studies implicate that maintaining pH homeostasis of TGN is required for protein trafficking in Arabidopsis [42]. DVs are unique carriers for proglutelin transport in rice presumably budded from the TGN [5, 6]. It is worth noting that the average size of DVs newly budded from TGN is enlarged to about 193 nm in gpa6, which is much bigger than 154 nm in the wild type (Fig. 3), 160 nm in gpa2 [5], and 153 nm in gpa3 mutants [6]. Thus, pH of TGN seems to have an impact on DV size control, although the detail mechanism remains to be explored.

Conclusions

In summary, our studies demonstrated that OsNHX5 is localized to the Golgi, TGN and PVC to maintain pH homeostasis, which is important for DV-mediated glutelin trafficking in rice endosperm.

Methods

Plant materials and growth conditions

The gpa6 mutant was identified from a 60Co-irradiated mutant pool of the indica cultivar N22 (Nagina22, an Indian traditional variety). An F2 population was produced from gpa6 and a japonica variety Nipponbare for mapping. The seeds of all accessions were collected, stored and supplied by the State Key Laboratory of Crop Genetics and Germplasm Enhancement of Nanjing Agricultural University, Jiangsu, China. All plants were grown in the paddy field during the normal growing seasons or in a greenhouse at Nanjing, China. Developing seeds of wild type (N22) and gpa6 at 6–21 days after fertilization (DAF) were used in the experiments.

Protein Extraction from Rice seeds and immunoblot analysis

Total protein extraction and immunoblot assay were performed as described previously [16].

Microscopy

Scanning electron microscopy, transmission electron microscopy, light and immunofluorescence microscopy and immunogold labeling analysis were performed as described previously [5, 6, 16, 17].

Map-based cloning

To map the GPA6 locus, an F2 population was generated from a cross between the gpa6 mutant and a japonica variety Nipponbare. Total proteins were extracted from half of an individual rice seed and resolved by SDS-PAGE gel to monitor the accumulation of the proglutelins. Meanwhile, the other half of the identified mutant seeds with embryos was grown for DNA extraction. In total, 208 recessive individuals were used for fine mapping of GPA6. The primers used in fine mapping are listed in Additional file 10: Table S3.

Real-time RT-PCR analysis

Total RNA was extracted from different tissues using an RNA Prep Pure Plant Kit (TIANGEN). First strand cDNA was synthesized using oligo (dT)18 as the primer (TaKaRa). Three biological replicates of real-time RT-PCR were performed with SYBR Premix Ex Taq II (TaKaRa) on an Applied Biosystems 7500 Real-Time PCR System. The primer sequences used for PCR are listed in Additional file 12: Table S5.

Subcellular localization

For transient expression analysis in N. benthamiana leaf epidermal cells, the coding region of OsNHX5 or OsNHX6 was amplified and inserted into the binary vector pCAMBIA1305GFP to produce the OsNHX5-GFP or OsNHX6-GFP fusion construct (Additional file 11: Table S4). Construct were introduced into the Agrobacterium strain EHA105 and then used to infiltrate N. benthamiana leaves, as described previously [43]. N. benthamiana protoplasts were isolated using the same method used with Arabidopsis [44]. Fluorescence was observed using a confocal laser scanning microscope (Leica TCS-SP8).

pH measurements

The rice protoplasts were isolated from 10-days-old N22 and gpa6 seedlings. pH sensors ManI–PRpHluorin, PRpHluorin-BP80 (Y612A) and PRpHluorin–AtVSR2 were transformed into rice protoplasts as previously described [45, 46]. The PRpHluorin signals at emission wavelength of 500 to 550 nm were recorded with dual-excitation wavelength at 405 and 488 nm, respectively, and used to calculate the pH using the calibration curve. In vivo calibration was achieved from the same protoplasts expressing the PRpHluorin for pH measurement. Protoplasts were incubated in WI protoplast buffer (0.5 M mannitol and 20 mM KCl) with 25 μM nigericin, 60 mM KCl, and 10 mM MES/HEPES Bis-Tris-propane adjusted to different pH values ranging from 5.0 to 8.0 for each calibration point [33, 37, 38, 47,48,49]. Fluorescence was observed using a confocal laser scanning microscope (Leica TCS-SP8).