Nup98 mutation results in early senescence in Arabidopsis thaliana
Nup98 is a highly conserved nuclear pore protein in eukaryotes. In Arabidopsis thaliana, there are two homologs of the mammalian Nucleoporin98, Nup98a (At1g10390) and Nup98b (At1g59660), and they share high conserved amino acid sequence in the Phe-Gly (FG)-repeats and autoproteolytic domain (APD, Supplementary Fig. 1) [3, 6]. Nup98a was previously reported as DRACULA2 (DRA2), a regulator of the shade avoidance syndrome (SAS) in Arabidopsis  and immune responses to a rice fungal pathogen . To investigate Nup98 functions in plant development, we screened mutants of nup98a (SALK_080083, SALK_090744, SALK_023493, SALK_103803, SALK_015016) and nup98b (CS803848 and GABI_288A08) ordered from ABRC and GABI T-DNA mutant center, respectively. Homozygous lines were isolated for the following insertional mutants of SALK_103803, SALK_015016, GABI_288A08, and among them SALK_103803 and GABI_288A08 were the mutants reported by Parry . The T-DNAs in these homozygous mutants are inserted in the coding regions (Fig. 1A) and RT-PCR results demonstrated that these mutants were null alleles (Fig. 1B) consistent with the Parry’s results . We did not observe any obvious phenotypes on flowering and senescence in either the nup98a or nup98b single mutants when compared to wild type plants under long-day photoperiod conditions (Fig. 1C), as previous studies showed . As Nup98a and Nup98b share high conserved amino acid sequences (Supplementary Fig. 1), we tested the hypothesis that Nup98a and Nup98b acted redundantly with the nup98a nup98b double mutants we made by crossing nup98a to nup98b. Strikingly, both double mutants, nup98a-1 nup98b-1 and nup98a-2 nup98b-1, displayed similar early senescence phenotypes when compared to wild type plants (Fig. 1C and 1D). Also, the double mutant plants had additional phenotypes, such as smaller inflorescences, flowers, siliques and short stature and severe sterility when compared to wild type plants (Supplementary Fig. 2 and 3). We recently reported that the nup98a-1 nup98b-1 double mutants had early flowering phenotype . As expected, the mutated phenotypes observed in the double mutant plants were rescued by expressing only Nup98b (Fig. 1E). Our results demonstrate that Nup98a and Nup98b act redundantly.
To investigate if senescence phenotype was specific to the nup98a-1 nup98b-1 mutant, we selected another three nucleoporin mutants, nup96-1, nup160-1, and nup107-1, which showed flowering phenotypes in our previous report , to analyze the effect of other nuclear pore components on senescence. To our surprise, there is no early senescence phenotypes observed in these mutants (Supplementary Fig. 4), suggesting that some of nucleoporins may not be involved in the regulation of senescence and Nup98 likely had specific functions on this developmental event.
Nup98 gene may be involved at multiple pathways of senescence initiation
Because the early senescence observed in the nup98a-1 nup98b-1 double mutant could be a secondary effect of altered development, we further explored the role of Nup98a and Nup98b in senescence control. To date, at least 200 genes have been identified as regulating or participating in senescence in plants . We investigated and summarized literatures in Supplementary Fig. 5 focusing on the main genes, which showed that various endogenous and environmental cues, such as different hormones, sugar signalling, light and photoperiod conditions, and multiple stresses, trigger plant senescence and that many genes regulate senescence in multiple cross-talking pathways. To identified the potential link to early senescence, we measured mRNA abundance of important senescence-associated genes in different pathways the nup98a-1 nup98b-1 double mutant plants (Supplementary Fig. 5) at ZT0 (Zeitgeber 0, the time point of light on) and ZT16 (the time point of light off) in plants grown under long day conditions by real time quantitative PCR (RT-qPCR). As shown in Fig. 2, many genes tested had different transcript abundances in the nup98a-1 nup98b-1 mutant when compared to wild type plants, although different patterns were observed. In the first category, WRKY53 (WRKY DNA-BINDING PROTEIN53), SAG13 (SENESCENCE-ASSOCIATED GENE13), WRKY6, NAC1 (NAM, ATAF, and CUC) and NAC2 displayed higher transcript abundances at both ZT0 and ZT16 (Fig. 2A), suggesting that genes of stress (WRKY53, WRKY6, and NAC1) and SA (WRKY53, WRKY6, and NAC2) pathways were related to the nup98a-1 nup98b-1 double mutant senescence phenotypes. In the second category, SAG12, NAP (NAC DOMAIN CONTAINING PROTEIN), SAG2 and CAT1 (CATALASE1), also genes in stress and SA pathways, were only increased at either ZT0 or ZT16 (Fig. 2B). In contrast, in the third category, SAUR36 (SMALL AUXIN UPREGULATED36), WRKY70, ARP4 (ACTIN-RELATED PROTEIN4), SEN1 (SPLICING ENDONUCLEASE1), and COI1 (CORONATINE INSENSITIVE1) were decreased in abundance at either ZT0 and/or ZT16 (Fig. 2C), indicating that auxin (SAUR36) and jasmonate (COI1) may be negatively related to the nup98 senescence phenotypes. The abundance of AGL15 (AGAMOUS-LIKE15), EBP1 (ERBB-3 BINDING PROTEIN1), RPS6a (RIBOSOMAL PROTEIN6a), and NPR1 (NONEXPRESSOR OF PR-GENES1) had opposite changes at ZT0 and ZT16 (Fig. 2D), suggesting the function of these genes on Nup98 was dependent on circadian rhythm.
This expression profile suggested that there were at least three characteristics of senescence in the nup98 mutant. Firstly, several pathways, mainly stress and SA pathways, were involved in regulation of senescence processes in the nup98a-1 nup98b-1 double mutant. Secondly, different genes functioned in their own special modes, positively or negatively at different phases (morning or afternoon phases). Thirdly, some of genes, such as WRKY53, WRKY6, NAP1, and SAG2, may play roles in multiple pathways. The results indicated that the senescence phenotypes of the nup98 double mutant was resulted from multiple pathways. Many genes showed circadian expression patterns, consistent with our previous report that Nup98 genes participated in regulation circadian clock .
Starch metabolism is impaired in the nup98a-1 nup98b-1 double mutant
During photosynthesis, glucose is synthesized and stored as starch, which is degraded at night. Starch synthesis, degradation and metabolism pathways involve a number of genes in plants (Supplementary Fig. 6) [18, 19]. Firstly, we checked starch homeostasis in the nup98a-1 nup98b-1 double mutant plants that were grown under 12 h light/12 h dark conditions (at ZT0, dawn, and ZT12, dusk) (Fig. 3). All plants accumulated starch at dusk, however, double mutant plants accumulated much more starch when compared to wild type plants as determined by iodine staining (Fig. 3A) and starch quantitative assay (Fig. 3B). The more intense signals in older, 21 and 28-day, double mutant leaves may be a consequence of starch accumulating over time. Therefore, the double mutant displayed much more starch not only at dawn but also at dusk when compared to wild type plants and this was much clearer in 14 and 21-day old seedlings (Fig. 3A and 3B). Together, these results suggested that starch metabolism was impaired in the nup98a-1 nup98b-1 double mutant.
We measured the transcript abundance of genes involved in starch metabolism  by quantitative RT-PCR (Fig. 4) and found that many genes had significantly lower abundant transcripts at least one time point in the double mutant when compared to wild type plants. These genes encoded enzymes for the degradation of starch not only in the early steps of starch degradation in chloroplasts , such as GWD1 (α-GLUCAN WATER, DIKINASE1; or STARCH EXCESS1, SEX1), β-BAM1 (β-AMYLASE1), BAM3, BAM5, BAM6, BAM7, BAM8, SEX4, and LSF1 (LIKE SEX FOUR1), but also in the later steps of starch degradation, such as LDA (LIMIT-DEXTRINASE), AMY1 (α-AMYLASE) and AMY2 (), DPE1 (DISPROPORTIONING ENZYME) and DPE2, PHS1 (α-GLUCAN PHOSPHORYLASE)and PHS2 . Time-dependent low-expression of these genes suggested they were in the control of circadian clock, since Nup98 is involved in circadian regulation in Arabidopsis . These genes function at different steps . GWD1 is α-glucan water, dikinase and phosphorylates glucosyl residues of amylopectin at the C-6 position, the initiation step for starch degradation. BAMs are a family of β-amylase breaking down the α-1,4-linked glucose chains. LSF1 and SEX4 releases phosphate bound at C-6 and C-3 of glucosyl residues. Both ISA (ISOAMYLASE 3) and LDA (LIMIT-DEXTRINASE) hydrolyze α-1,6 branch points but show different substrate specificities. AMYs act on α-1,4-linkages releasing linear α-1,4-linked oligosaccharides and branched α-1,4- and α-1,6-linked oligosaccharides. DPEs (DISPROPORTIONATING ENZYMEs) Transfer glucose/α-1,4-linked glucan moiety from a donor glucan to an acceptor, releasing the non-reducing end glucose/glucan moiety. PHSs (α--GLUCAN PHOSPHORYLASEs) acts on the non-reducing end of α-1,4-linked glucose. Lower abundance of these genes would be expected to results in a starch-excess phenotype in the nup98a-1 nup98b-1 double mutant (Fig. 3) as these enzyme mutants . We also found that there were some genes, which did not show a significant change in mRNA abundance. These genes included GWD2, GWD3, ISA3 (ISOAMYLASE3) and AMY3, suggesting the effect of Nup98a/b on starch metabolism was limited to some specific enzymes.
We also measured the abundance of genes related to photosynthesis and sugar metabolism by RT-qPCR in the nup98a-1 nup98b-1 double mutant and wild type plants (Supplementary Fig. 7). In terms of photosynthesis related genes, the decrease of mRNA abundance of LHCA (PHOTOSYSTEM I LIGHT HARVESTING COMPLEX GENE A) and LHCB (PHOTOSYSTEM I LIGHT HARVESTING COMPLEX GENE B) was observed in the nup98a-1 nup98b-1 double mutant at different time points, e.g., LHCA1/2 and LHCB1.1 at ZT0 and LHCA1 and LHCB1.4 at ZT16. We also observed the decrease of mRNA abundances of KIN10 and KIN11, two sugar signaling genes in the double mutant when compared to wild type plants. Both genes delay plant senescence [22, 27, 28] and therefore the reduced abundance may be associated with earlier senescence (Supplementary Fig. 5). We also observed slightly-increased mRNA abundance of HXK1 at dusk (ZT16) and this may contribute to earlier senescence via the cytokinin signaling pathway . Unexpectedly, mRNA abundance of TPS1, a senescence activator , was reduced in the nup98a-1 nup98b-1 double mutant compared to wild type plants, suggesting that T6P (trehalose-6-phosphate) was not related to senescence of the nup98a-1 nup98b-1 double mutant. The results indicated that starch synthesis and sugar signalling were impaired in the double mutant.
Exogenous sugar rescues the early senescence in the nup98a-1 nup98b-1 double mutant
Based on our results above, we interpreted that the carbon or energy supply was impaired in nup98a-1 nup98b-1 double mutant plants. We tested the idea by supplying exogenous carbon in the form of sucrose in growing medium to see if the early senescence phenotype in the nup98a-1 nup98b-1 double mutant plants could be rescued. Our results showed that sucrose and MS nutrients can support the double mutant plants growing well even though they were weak compared to wild type plants (Fig. 5). Both plants can complete their life cycles on medium containing agarose supplement with sucrose and nutrients. Then, we allowed double mutant and control plants grown in MS medium until inflorescence emergence and then transferred them to soil. As expected, the double mutant grew well as wild type plants did on MS medium before transplanting (Supplementary Fig. 8). However, after transferring to soil, senescence symptoms on mutant plants’ leaves quickly appear at day 6, and the mutant plants wilted at day 30 (Supplementary Fig. 8). If seeds were sown on medium containing only MS nutrients or sucrose, both wild type plants and the nup98 mutant cannot survive as plants grow agarose medium without any supplements (Supplementary Fig. 9).
To rule out the potential effect of soil on senescence phenotypes observed above, we carried out another experiment to test if exogenous macro and micronutrients would complement the phenotypes observed in the mutant by continuously growing plants on medium at different strengths of sucrose and macro- and micro-nutrients (Supplementary Fig. 9). To our surprise, not only did sucrose suppress the early senescence phenotype in the double mutant but also macro- and micro-nutrients in the presence of sucrose. The lower strength nutrients (½ MS) enhanced the lower sucrose (1.5%) effect on suppressing senescence, suggesting that both energy supply and nutrients metabolism were impaired in the nup98a-1 nup98b-1 double mutant. We also tested if sucrose could recue the nup98a-1 nup98b-1 double mutant phenotype in soil. However, such an experiment failed and both the nup98a-1 nup98b-1 double mutant and wild type seedlings died, because sucrose enhanced pathogen growing (Supplementary Fig. 10).
Misexpression of starvation and senescence marker genes in the nup98a-1 nup98b-1 double mutant
Results above implied that the nup98a-1 nup98b-1 double mutant may suffer from sugar starvation or/and senescence. Next, we asked when starvation or senescence initiated in the nup98a-1 nup98b-1 double mutant. DORMANCY-ASSOCIATED PROTEIN-LIKE 1 (DRM1/DYL1, At1g28330) and DARK INDUCIBLE 6 (DIN6, At3g47340) [38, 39] are two well-studied sugar starvation gene markers, whereas SAG12 (At5g45890) and WRKY53 (At4g23810) are well-characterized senescence markers [40-42]. Autophagy is an important event occurring during sugar starvation and senescence [38, 43, 44], and AUTOPHAGY8a (ATG8a, At4g21980) and ATG8e (At2g45170) are two typical molecular indicators for autophagy in plants . Therefore, we investigated expression changes of these genes in the nup98a-1 nup98b-1 double mutant compared to that in wild type plants, and the results showed that they had different changes in a time- and developmental-dependent mode (Fig. 6). In the double mutant, DRM1 had significantly higher expression at ZT0, but lower at ZT12 from very early stage (day 5 after germination) (Fig. 6A). DIN is a light-repressed and dark-induced gene , and its high level of expression at ZT0 in the nup98a-1 nup98b-1 double mutant became obvious at day 15, but at ZT12 higher abundancy appeared earlier from day 10 (Fig. 6A). Compared to wild type plants, the senescence marker WARKY53 in the nup98a-1 nup98b-1 double mutant expressed higher at the early stage when DRM1 expression was in disorder (day 5) (Fig. 6B). SAG12 is a developmental controlled indicator for the later stage of senescence [47, 48]. We found that there was no much difference of SAG12 expression in the early stage (day 5) between the nup98a-1 nup98b-1 double mutant and wild type plants. However, SAG12 had a higher expression level in the nup98a-1 nup98b-1 double mutant at both ZT0 and ZT12 from day 10 (Fig. 6B). In the meanwhile, the two markers of autophagy, ATG8a and ATG8e, also had higher abundancy of mRNA in most of samples of the double mutant from day 10. Token together, our results showed that the nup98a-1 nup98b-1 double mutant appeared the sign of energy starvation, at least at molecular level, in early developmental stage when plants did not display visible senescence phenotypes. And these expression changes may have circadian and developmental characters. A previous report shows that different sugars (such as sucrose, glucose, and fructose) have different effects on the regulation of senescence . It may be a cue to study the function of Nup98 on senescence regulation.
Nup98 proteins mainly localize to the nuclear membrane and nucleoplasm
Nup98 is one of the mobile and peripheral FG (Phe-Gly domain) nucleoporins and is located at both the nuclear and cytoplasmic sides of the NPC central channel [4, 5]. Arabidopsis Nup98a (also known as DRA2) is also found distributing in different subcellular compartments . We constructed transgenic Arabidopsis plants expressing 35S::GFP:Nup98a and 35S::GFP:Nup98b and analyzed the subcellular localization of both translation fusion proteins. Not surprisingly, both proteins were distributed in the cytoplasm, the nucleoplasm and at the nuclear periphery (Fig. 7). We also observed no significant difference in the subcellular distribution of Nup98a and Nup98b and this is consistent with our observations of genetic redundancy. In conclusion, our combined results demonstrated that Nup98b proteins localized at both the nucleus and cytoplasm as Nup98a proteins  and their homologs in other organisms do [4, 5].