GBPL3 is a candidate component of the nuclear basket/nucleoskeleton protein network

Previously, we used subtractive proteomics and proximity labeling proteomics to profile the plant nuclear envelope (NE)-associated proteome in Arabidopsis [44,45]. Among the identified candidates, the GUANYLATE-BINDING PROTEIN-LIKE 3 (GBPL3) protein attracted our attention. Although GBPL3 contains no predictable transmembrane (TM) domains, it was one of the top-ranked NE-associated proteins identified by the subtractive proteomics analysis (fold change = 580 and p-value = 1.2 × 10⁻⁹) with a high peptide-spectrum match (PSM) score (S1A Fig), suggesting that GBPL3 is an NE protein candidate of high confidence and likely an abundant component of the plant NE. On the other hand, proximity labeling proteomics revealed that GBPL3 was probed by multiple inner nuclear membrane baits, including SUN1, MAN1, PNET2_A, and PNET2_B, by main nucleoskeleton proteins, including CRWN1 and KAKU4, and by the NPC basket nucleoporin Nup82 (Fig 1A). In contrast, outer nuclear membrane protein baits, including SINE1 and WIP1, did not probe GBPL3. This evidence suggests that GBPL3 is likely a protein associated with the inner nuclear membrane.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. GBPL3 is a nuclear envelope protein located at the NPC basket.

(A) Bubble plot showing the reanalysis of the mass spectrometry data retrieved from previously published proximity labeling proteomics using different baits (PXD015919 and PXD026924). LFQ intensity values of GBPL3 protein are plotted. A heatmap of normalized PSM values of GBPL3 are shown on the right. For bait protein SUN1, SINE1, Nup82, and WIP1, biotin-treated NEAP1-BioID2 (Ctrl 1) and mock-treated YFP-BioID2 (Ctrl 2) were used as controls. For bait protein MAN1, PNET2_A, PNET2_B, CRWN1, and KAKU4, non-transformant plants (NT) (Ctrl 1) and mock-treated YFP-BioID2 (Ctrl 2) were used as controls. (B) Weighted gene coexpression heatmap of GBPL3 with all known genes of nucleoporin and nucleoskeleton in Arabidopsis. Mutual rank-based gene expression data using microarray and RNA-seq were retrieved from ATTED-II (version 11.0) and converted into network edge weights before visualizing into a clustered heatmap. (C) Prediction of CC domains and IDRs by Smart (http://smart.embl-heidelberg.de/) and D²P² (https://d2p2.pro/), respectively. (D) Fluorescence imaging showing subcellular localization of GBPL3 protein in root cells. One-week-old pGBPL3-mEGFP-GBPL3 transgenic plants with higher expression (left panel) and lower expression (right panel) levels are used for imaging. Bars = 10 μm. (E) Z-stack imaging followed by projection of root cells expressing mEGFP-GBPL3 in a line with lower expression. Bar = 10 μm. (F) Anti-GFP immunogold labeling and transmission electron microscopy analysis of pGBPL3-mEGFP-GBPL3 transgenic seedlings with low expression. Root tip cells were imaged. Red arrows indicate gold particles, and blue arrowhead pairs indicate nuclear pores. Bar = 1 μm. All underlying data in Fig 1 can be found in S5 Data. CC, coiled-coil; IDR, intrinsically disordered region; INM, inner nuclear membrane; LFQ, label-free quantitation; NLS, nuclear localization signal; NPC, nuclear pore complex; ONM, outer nuclear membrane; PSM, peptide-spectrum match.

https://doi.org/10.1371/journal.pbio.3001831.g001

To investigate the potential connection between GBPL3 and nuclear envelope components, we analyzed the transcriptional correlation of GBPL3 with genes that encode the nucleoskeleton and nucleoporins in Arabidopsis using ATTED-II [46]. The clustered heatmap showed that GBPL3 is strongly coexpressed and clustered into a small group with 4 nucleoskeleton genes CRWN1/2/3/4 and the NB gene NUA (Fig 1B). This result suggests a potential connection of GBPL3 with the nucleoskeleton and/or the NPC basket.

GBPL3 is conserved among eukaryotes and is an ortholog of human GBPs with 2 paralogs (GBPL1 and GBPL2) in Arabidopsis (S1B Fig and S1 Data). Besides the canonical N-terminal GTPase domain and the C-terminal helical domain (GBP_C) preserved in all GBPs, many plant GBPL proteins, including GBPL3, evolved a unique C-terminal extension that is annotated as a CC region but is also predicted to be intrinsically disordered (Figs 1C and S1B). Compared to ordered CC domains, disordered CC domains were found overrepresented in human proteins that function in actin filaments, cell junctions, macromolecular complexes, and nucleolus [47]. The disordered CC region of GBPL3 was recently demonstrated to mediate liquid–liquid phase separation (LLPS) of the protein [48]. Interestingly, part of the GBPL3 CC domain (562–784 aa) was predicted to have a structural similarity with the human Lamin_A protein (42–215 aa) by SWISS-MODEL (https://swissmodel.expasy.org) (S1C Fig), and in both Arabidopsis and humans, disordered CC domains are prevalent in the scaffolding NB protein (AtNUA/HsTPR) and nucleoskeleton protein (AtCRWN1/HsLamin_A) (Fig 1C), suggesting that this feature may be important for the NB and nucleoskeleton function. Together, the strong coexpression and domain similarity with NB and nucleoskeleton components led us to hypothesize that GBPL3 is functionally related to them.

GBPL3 localizes to the NPC

To validate this hypothesis, we first verified the subcellular localization of GBPL3. We generated Arabidopsis stable transgenic lines expressing mEGFP-GBPL3 fusion protein. Although the construct was driven by the GBPL3 native promoter, we were able to detect a difference in the mEGFP-GBPL3 expression level in independent transgenic lines based on immunoblots (S1D Fig), potentially due to different copy number or insertion loci of the transgene. Using confocal fluorescence microscopy, we found that in most transgenic lines with relatively low expression, mEGFP-GBPL3 is enriched at the nuclear membrane with some signal in the nucleoplasm (Fig 1D). However, in transgenic lines with high expression levels, mEGFP-GBPL3 stained the nuclear rim in a relatively weak manner but predominantly labeled bright and large droplet-like condensates at the nuclear periphery (Fig 1D), in agreement with the previous report that GBPL3 is capable of undergoing LLPS and forming biomolecular condensates upon overexpression [48]. These GBPL3 localization patterns were also observed in mCherry-GBPL3 lines, confirming that the observation is not tag specific (S1E Fig). Because transgenic lines displaying both localization patterns can fully complement gbpl3 mutant phenotypes (see below), we conclude that GBPL3 is functionally localized to the nuclear periphery but forming large condensates is not essential for GBPL3 function, at least under normal conditions. However, we cannot exclude that LLPS of GBPL3 may still occur at a level beyond detection by fluorescence microscope in low expression lines.

When we performed z-stack imaging of root cells using low expression lines, small punctate structures were captured at the nuclear surface (Fig 1E), similar to those formed by some plant nucleoporins (e.g., CPR5 and Nup155) [49], suggesting that GBPL3 is enriched in the NPC. To confirm this, we performed immunogold labeling and transmission electron microscopy using mEGFP-GBPL3 transgenic lines with low expression. In nearly all samples examined, we found that gold particles are predominantly associated with the inner nuclear membrane, close to nuclear envelope openings indicative of nuclear pores (Figs 1F and S1F). Together, these results support that GBPL3 localizes at the inner nuclear membrane and is highly enriched at the NPC.

GBPL3 physically interacts with both the nuclear basket and the nucleoskeleton

To understand the function of GBPL3 at the nuclear periphery, we performed proximity labeling proteomics using GBPL3 as bait to profile the in vivo GBPL3 proxitome. GBPL3 was fused with TurboID, an engineered promiscuous biotin ligase that mediates efficient biotinylation of proximal proteins. The fusion construct was driven by the GBPL3 native promoter, and we showed separately that this promoter can drive ubiquitous expression of GUS in Arabidopsis seedlings (except in the hypocotyl) (S2A Fig). For proximity labeling, we chose 1-week-old pGBPL3-GBPL3-TurboID-3HA transgenic seedlings from a line with relatively low GBPL3 expression and treated them with free biotin for 4 h followed by biotin depletion. The resulting biotinylated protein collection was affinity purified using streptavidin-coated beads. Two biological replicates were used, and wild-type non-transgenic plants were sampled in parallel as control. Immunoblot using 2% affinity-purified samples confirmed inducible and efficient protein biotinylation by GBPL3-TurboID, including GBPL3 itself (S2B Fig). Samples were then digested with trypsin on beads and purified before being subjected to label-free quantitative mass spectrometry (LFQMS), from which we collected both label-free quantitation (LFQ) intensity and PSM values. By applying fold-change > 4 and p-value < 0.01 as cutoffs compared to control samples, we identified 243 and 216 significantly enriched proteins using LFQ and PSM data, respectively (Fig 2A and S2 Data). We merged the 2 datasets and obtained 270 unique protein candidates, of which more than 70% are present in both datasets.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. GBPL3 physically interacts with NPC basket and nucleoskeleton components.

(A) Volcano plots of candidates identified by proximity labeling proteomics using GBPL3 as bait. Analyses based on the LFQ intensity and PSM data are shown on the left and right sides, respectively. The overlap of candidates identified by both analyses is shown in the middle. Non-transformants (WT) treated with biotin were used as control, and 2 biological replicates are included in the analysis. Significantly enriched proteins were selected using PSM > 4, fold-change > 4, and p-value < 0.01 as cutoffs and represented by red dots. The nuclear basket and nucleoskeleton components were text labeled. Underlying data can be found in S5 Data. (B) Heatmap showing normalized and averaged PSM values of all nucleoskeleton and nucleoporin proteins identified by GBPL3, Nup82, and KAKU4 using proximity labeling proteomics. Blue and green dots indicate p-value < 0.01 and 0.05, respectively. Nup82 and KAKU4 data were retrieved from PXD015919 and PXD026924, respectively. (C) A model diagram showing that the nuclear basket and the nucleoskeleton comprise a protein meshwork underneath the NPC and the position of GBPL3 at a more distal region of the NB. (D) Transient protein expression of GBPL3 and NPC basket components Nup136/Nup82 and nucleoskeleton components CRWN1/CRWN3 in N. benthamiana. The upper 2 panels show individual protein expression, and the lower 4 panels show mCherry-GBPL3 coexpression with CRWN1-YFP, CRWN3-GFP, Nup136-YFP, or Nup82-GFP. Bars = 10 μm. (E) Yeast-two-hybrid analysis using GBPL3-C (GBP_C domain and IDRs) and Nup82 as the bait and Nup136, Nup82, CRWN1, CRWN2, CRWN3 as the prey. Zygote yeasts were grown on DDO (SD/–Leu/–Trp) and QDO (SD/–Leu/–Trp/–His/–Ade) media. Empty vectors were used as negative controls. (F) BiFC assay between GBPL3 and CRWN1, KAKU4, Nup136, or Nup82. Indicated BiFC constructs were transiently expressed in N. benthamiana leaf epidermal cells. Free-mCherry was coexpressed to label the nuclei. Bars = 10 μm. BiFC, bimolecular fluorescence complementation; LFQ, label-free quantitation; NB, nuclear basket; NPC, nuclear pore complex; PSM, peptide-spectrum match.

https://doi.org/10.1371/journal.pbio.3001831.g002

Among the significantly enriched candidates, we found all 5 known nucleoskeleton proteins in Arabidopsis (CRWN1, CRWN2, CRNW3, CRWN4, and KAKU4). Remarkably, GBPL3 also specifically probed all reported NB components (NUA, Nup136, Nup82, Nup50a, Nup50b, and Nup50c) but barely labeled other NPC constituents (Fig 2B). This result provides strong support that GBPL3 is part of the intimately associated NB and nucleoskeleton protein meshwork underneath the NPC. Consistently, the FG basket nucleoporin Nup82 probed GBPL3, nucleoskeleton proteins, and other basket nucleoporins (Fig 2B). However, Nup82 also probed a significant number of core scaffold nucleoporins, suggesting that compared to Nup82, GBPL3 may polarly localize to a more distal region of the NB away from the NPC core (Fig 2C).

To confirm the association between GBPL3 with the NPC basket and the nucleoskeleton, we transiently coexpressed GBPL3 with the NB nucleoporin Nup136 and Nup82 or nucleoskeleton protein CRWN1 and CRWN3 in N. benthamiana. Overexpression of Nup136 or Nup82 alone led to the formation of condensates at the nuclear periphery (Fig 2D), consistent with a recent report [43]. Those condensates overlapped with GBPL3 condensates (Fig 2D), suggesting that GBPL3 can form condensates at the NPC basket. On the other hand, CRWN1 and CRWN3 did not form condensates when expressed alone; however, they were recruited to GBPL3 condensates upon coexpression (Fig 2D). To determine whether GBPL3 directly interacts with Nup82/Nup136 or CRWNs, we performed yeast two-hybrid (Y2H) assays. While Nup136 displayed auto-activation as bait, we found that GBPL3 C-terminus, including the GBP_C and the disordered CC domain, is sufficient to interact with CRWN2 and weakly interact with Nup82 in yeasts (Fig 2E). The NB scaffolding nucleoporin NUA was not included in the Y2H assay because we were not able to obtain the full-length cDNA of the gene due to its large size. Lastly, we performed bimolecular fluorescence complementation (BiFC) assay and confirmed the in vivo association between the full-length GBPL3 and Nup82/Nup136/CRWNs, which occurs in condensate structures (Fig 2F). Together, these results led us to conclude that GBPL3 physically associates with both the NB and the nucleoskeleton in plants.

GBPL3 can recruit the transcription apparatus and RNA processing machinery

Besides NB and nucleoskeleton components, gene ontology (GO) analysis revealed that GBPL3 proxitome is highly enriched in proteins that function in chromatin modification and organization, transcription initiation and regulation, and RNA splicing and processing (Fig 3A), which account for over 65% of the identified GBPL3 proxitome (177/270). To better categorize the function of these proteins, we built a protein–protein interaction network of these 177 GBPL3 interactors using the STRING database, which further classified them into 14 protein complexes or functional groups with 2 main molecular functions: transcription (83 proteins) and RNA processing (94 proteins) (Fig 3B). The transcription-related category includes chromatin remodelers, histone modifiers, transcription initiator factor II (TFII) proteins, key transcription coactivators—the Mediator complex and the SAGA complex, as well as TFs and co-regulators that regulate major phytohormone responses, including auxin, gibberellin, cytokinin, salicylic acid, and jasmonic acid. The RNA processing category contains RNA helicases, mRNA splicing machinery, and other RNA processing proteins. Notably, these proteins are not enriched in the Nup82 proxitome (S3A Fig), suggesting that the GBPL3-defined NB region may be specifically enriched with molecular apparatus that is involved in active chromatin remodeling, transcription activation/repression, and RNA processing.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. GBPL3 is capable of recruiting transcription regulators and RNA processing machinery.

(A) GO enrichment analysis of GBPL3 proxitome. GO terms under the 2 major molecular functions identified (transcription and RNA processing) are presented in a bubble plot. Underlying data can be found in S5 Data. (B) GBPL3 PPI network that contains 177 GBPL3 proximal proteins involved in transcription and RNA processing identified in (A). Cytoscape was used to construct and visualize the GBP3 PPI network. GBPL3 interactors were clustered and color-coded manually based on published protein functions. Node sizes represent scores of PPIs retrieved from the STRING database (https://string-db.org/). (C) Transient expression of selected GBPL3 proximal proteins involved in transcription regulation and RNA processing in N. benthamiana. The left column shows the individual expression of GFP-tagged GBPL3 proximal proteins, and the right 3 columns show coexpression with mCherry-GBPL3. CELL DIVISION CYCLE (CDC5), a putative TF, was included as a negative control. Bars = 10 μm. GO, gene ontology; PPI, protein–protein interaction; TF, transcription factor.

https://doi.org/10.1371/journal.pbio.3001831.g003

To confirm the association of GBPL3 with these transcription and RNA processing factors, we selected 11 well-characterized candidates, which have been reported to participate in the regulation of plant development and responses to stress-related phytohormones and various environmental stimuli [50–60]. They include chromatin remodelers ACTIN-RELATED PROTEIN 4 (ARP4), ANTHESIS PROMOTING FACTOR 1 (APRF1), and PWWP DOMAIN INTERACTOR OF POLYCOMBS1 (PWO1), transcription regulators TBP-ASSOCIATED FACTOR 5 (TAF5), ARABIDOPSIS RESPONSE REGULATOR 1 (ARR1), ABA-RESPONSIVE KINASE SUBSTRATE 2 (AKS2), C-TERMINAL DOMAIN PHOSPHATASE-LIKE 1 (CPL1), and TOPLESS-RELATED 1 (TPR1), and mRNA binding or processing proteins F OLIGOURIDYLATE BINDING PROTEIN 1C (UBP1C), SUPPRESSORS OF MEC-8 AND UNC-52 1 (SMU1), and SNW/SKI-INTERACTING PROTEIN (SKIP). We fused these proteins with mEGFP and transiently expressed them in N. benthamiana. When expressed alone, 6 of the fusion proteins formed spontaneous nuclear condensates while the other 5 diffused in the nucleus (Fig 3C). However, all of them colocalized with GBPL3-containing nuclear condensates upon coexpression with mCherry-GBPL3 (Fig 3C), suggesting that GBPL3 is capable of recruiting those chromatin remodelers, transcription regulators, and mRNA processing machinery to the nuclear periphery. These recruitment events were observed in the form of condensates with a wide range of sizes during transient overexpression (Fig 3C). Therefore, it is reasonable to speculate that GBPL3 may recruit those factors to the NB through LLPS. In line with this idea, most of these proteins contain intrinsically disordered domains predicted by multiple algorithms (S3B Fig).

Together, these results support a hypothesis that the NPC basket region in plant cells is associated with active transcription reprogramming coupled with mRNA processing, and NB-associated GBPL3 appears to be sufficient for recruiting relevant molecular apparatus for this process.

GBPL3 is required for plant development and reproduction

To investigate the functional importance of GBPL3, we obtained 3 available T-DNA insertional mutant lines from ABRC, including SAIL_635_G05, SALK_016366 (gbpl3-3), and SALK_139144 (gbpl3-4). However, none of the 3 alleles exhibited observable developmental defects, nor did they show decreased GPBL3 expression measured by RT-qPCR using the whole seedling or young leaf tissue (S4A–S4C Fig). We then generated new gbpl3 mutant alleles using CRISPR/Cas9. We designed 3 single-guide RNAs (sgRNAs) that target the coding region of the very N-terminus of GBPL3, its GBP domain, and its CC domain, respectively (Fig 4A). A total of 4 different CRISPR mutant alleles were obtained, and we named them gbpl3-6 through gbpl3-9 (Fig 4A). In the gbpl3-6 and gbpl3-8 mutant, Cas9-mediated genome editing resulted in a single nucleotide insertion that led to premature termination of the GBPL3 protein (1,082 aa) at 14 aa and 696 aa, respectively. The gbpl3-7 allele contains a 19-bp deletion that results in a premature stop codon at 116 aa. Lastly, Cas9 caused a 12-nucleotide in-frame deletion at 600 bp downstream from the start codon in the gbpl3-9 mutant, which led to a 4 amino acid deletion in the GBP domain (S4D Fig). We found that all these 4 alleles showed developmental defects. Particularly, gbpl3-6, gbpl3-7, and gbpl3-8 showed short roots and stunted growth (Fig 4B–4D). They are also completely sterile and are defective in pollen grain formation (Fig 4E and 4F). These phenotypes are consistent with the GBPL3 expression pattern observed in the pGBPL3-GUS lines (S2A Fig) and indicate that GBPL3 is required for both normal vegetative growth and reproduction. The gbpl3-9 mutant shows relatively weak growth defects and is semi-sterile, likely a partial loss-of-function allele (Fig 4B–4E). It has been reported that Thr75 in human GBP1 (the equivalent to Thr107 in GBPL3) is highly conserved in GTP-binding proteins and its mutation affects GTP hydrolysis but does not change GTP-binding affinity and dissociation rate [61]. These mutant alleles support that both the GTPase activity and the disordered CC region are essential for GBPL3 function.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. GBPL3 is required for plant development.

(A) Four gbpl3 mutant alleles generated by CRISPR/Cas9. The upper panel shows the GBPL3 protein domain structure with 3 sgRNAs that target different regions. The lower panel describes gene editing details in different gbpl3 alleles. (B) Five-day-old WT and gbpl3 mutant seedlings. All 4 CRISPR gbpl3 mutants were confirmed by sequencing. Bar = 1 cm. (C) Four-week-old soil-grown plants. Bar = 1 cm. (D) Measurement of primary root length of 5-day-old seedlings and rosette diameter of 4-week-old plants (n = 15 biological replicates). Student t tests were performed. ****p-value < 0.0001; ns indicates not significant. (E) Representative siliques from 6-week-old plants. Bars = 1 cm. (F) Alexander’s staining of anthers and pollen grains in WT and gbpl3 mutants. Bars = 100 μm. (G) Complementation of the gbpl3-7 mutant by expressing native promoter-driven mEGFP-GBPL3 or mCherry-GBPL3 (T3 generation). Four-week-old soil-grown plants are shown in the upper panel. Bar = 1 cm. The lower panel shows fluorescence imaging of GBPL3 localization in root cells of 5-day-old seedlings in the complemented lines. Bars = 10 μm. (H) Volcano plot showing DEGs (fold-change > 2 and p-value < 0.05) in the gbpl3-7 mutant determined by RNA-seq analysis. Up- and down-regulated DEGs are displayed as pink and cyan dots, respectively. One-week-old seedlings were used (n = 3 biological replicates). (I) GO term enrichment analysis of gbpl3-dependent DEGs. All underlying data in Fig 4 can be found in S5 Data. DEG, differentially expressed gene; GO, gene ontology; sgRNA, single guide RNA.

https://doi.org/10.1371/journal.pbio.3001831.g004

The above phenotypes can be fully complemented by pGBPL3-mEGFP-GBPL3 or pGBPL3-mCherry-GBPL3 in the gbpl3-7 mutant background (Figs 4G and S4E). Notably, this complementation is independent of the formation of GBPL3 condensates, suggesting that forming observable GBPL3 condensates is not required for plant development under normal conditions and consistent with the previous report that GBPL3-dependent biomolecular condensates are largely induced by certain biotic stresses [48].

In addition to phenotypic characterization, we also performed whole-genome RNA-seq analysis to understand the transcriptome changes in gbpl3-7 compared to WT plants. We identified 778 up-regulated genes and 284 down-regulated genes (fold-change > 2 and p-value < 0.05) in 1-week-old gbpl3-7 seedlings compared to WT (Fig 4H and S3 Data). GO term enrichment analysis suggests that both abiotic and biotic stress-related responses are spontaneously activated to certain degrees in the gbpl3-7 mutant (Fig 4I). The mutant may also suffer from dysregulation of iron homeostasis and nutrient deficiency and other developmental defects based on the functional classification of down-regulated genes. Using RT-qPCR in the gbpl3-6 background, we confirmed that some immune-related genes are indeed significantly up-regulated, and genes that are involved in regulating iron homeostasis and transport of nitrate are down-regulated compared to WT (S4F Fig). Together, the genetic and transcriptome analyses indicate that GBPL3 is required for plant development and plays a role in maintaining nutrient homeostasis and suppression of stress responses under normal conditions. Notably, transcriptome responses (mainly up-regulated genes) in the gbpl3 mutant overlapped considerably with GO terms enriched in the nucleoskeleton mutant crwn1 crwn2 (S4G Fig and S3 Data) [53,62], suggesting a functional connection between GBPL3 and the nucleoskeleton.

GBPL3 functions with NB and nucleoskeleton components to differentially regulate plant development and stress responses

To investigate the functional connection between GBPL3 and the NPC basket and the nucleoskeleton, we crossed gbpl3-7 with NB nucleoporin and nucleoskeleton mutants. We found that GBPL3 genetically interacts with the NPC basket components Nup82 and Nup136 in a different manner. Even though the single mutant of nup82-2 and nup136-3 (a weak allele) did not show observable growth phenotype, we found that nup82-2 dramatically enhances the stunted growth of gbpl3-7, whereas nup136-3 has a very limited effect on the phenotype (Fig 5A and 5B). The gbpl3-7 nup82-2 double mutant phenotype is further enhanced in the gbpl3-7 nup82-2 nup136-3 triple mutant, which displays highly restricted growth and early senescence and cannot survive until reproduction. These results support a functional connection of GBPL3 with the NB and also suggest that Nup82 and Nup136 are only partially redundant, but together with GBPL3, are essential for plant survival.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. GBPL3 functions with nuclear basket and nucleoskeleton components to differentially regulate plant development and stress responses.

(A) Four-week-old plants of WT, nup136-3, nup82-2, nup136-3 nup82-2 (upper panel) and gbpl3-7, gbpl3-7 nup136-3, gbpl3-7 nup82-2, and gbpl3-7 nup136-3 nup82-2 (lower panel) are shown. (B) Statistics of rosette diameter from 4-week-old plants. Bars represent means ± SD (n = 8 biological replicates). Similar results have been obtained 3 times. (C) Genetic interaction between gbpl3-7 and crwn1 mutant. Five-day-old seedlings are shown. (D) Genetic interaction between gbpl3-7 and crwn2/crwn3. Four-week-old plants are shown. (E) Statistics of rosette diameter from 4-week-old plants. Bars represent means ± SD (n = 8 biological replicates). (F) Images and statistics of trichomes on the third rosette leaves of 4-week-old plants. (G) Four-week-old WT, crwn4, gbpl3-7, and gbpl3-7 crwn4 plants. Enlarged images show visible small chronic lesions on gbpl3-7 crwn4 but not gbpl3-7 plants. Bars = 10 μm. (H) Relative expression levels of SA-responsive defense maker gene (PR2, ICS1, EDS5) in WT, gbpl3-7, crwn2, gbpl3-7 crwn2, crwn3, gbpl3-7 crwn3 and crwn4, gbpl3-7 crwn4 measured by quantitative RT-PCR (n = 2 biological replicates). The expression level was normalized to that in WT. Two-way ANOVA was performed (*p < 0.05, ***p < 0.001, ****p < 0.0001). ns indicates not significant. All underlying data in Fig 5 can be found in S5 Data.

https://doi.org/10.1371/journal.pbio.3001831.g005

Genetic interactions between GBPL3 with nucleoskeleton genes are even more intriguing. We found that gbpl3-7 displayed distinct phenotypes when crossed with single mutants of 4 CRWN paralogs, with the gbpl3-7 crwn1 double mutant showing the most pronounced phenotype-seedling lethality (Fig 5C), suggesting a prominent/broader role of CRWN1 among the CRWN family [62,63]. Both the gbpl3-7 crwn2 and gbpl3-7 crwn3 double mutants showed stunted growth and early senescence, which is dramatically enhanced in the gbpl3-7 crwn2 crwn3 triple mutant, implicating a redundant role of CRWN2 and CRWN3 when acting together with GBPL3 (Fig 5D and 5E). The triple mutant plants also appear to suffer from abiotic stresses, indicated by anthocyanin accumulation. However, the gbpl3-7 crwn2 mutant is different from the gbpl3-7 crwn3 mutant in that the former is significantly smaller than the latter. More importantly, the gbpl3-7 crwn3 mutant exhibited a unique trichome-less phenotype (Fig 5F), suggesting that the cell division pathway may be specifically affected in this mutant. In contrast to the above gbpl3-7 crwn mutants, the gbpl3-7 crwn4 double mutant displayed no obvious phenotype compared to the gbpl3-7 single mutant during early developmental stages (Fig 5E and 5G); however, gbpl3-7 crwn4 plants showed chronic lesions on leaves after 5 weeks (Fig 5G), a typical phenotype caused by an elevated accumulation of salicylic acid. In line with this observation, we detected significant up-regulation of SA-responsive defense maker genes, including PR2, ICS1, and EDS5 in the gbpl3-7 crwn4 double mutant but not in the gbpl3-7 crwn2 and gbpl3-7 crwn3 double mutants (Fig 5H). Our observation of distinct gbpl3 crwn double mutant phenotypes raises an intriguing hypothesis that CRWNs are not redundant in their functional connection with GBPL3. We propose that GBPL3 may work with different CRWNs, which is required for distinct developmental processes and suppression of abiotic and biotic stress responses.

Taken together, these genetic results strongly support the functional connection of GBPL3 with the NPC basket and the nucleoskeleton. They also reinforce the previously established role of Nup82/136 and CRWNs in regulating plant biotic and abiotic stress responses [42,62,64] and place GBPL3 as an essential component and a critical regulator of the NB and nucleoskeleton-dependent transcription control of growth and stress in plants.

Plant material and growth conditions

All Arabidopsis plants used in this study are Col-0 background. T-DNA insertion lines, including SAIL_635_G05, gbpl3-4 (SALK_139144), gbpl3-3 (SALK_016366), nup136-3 (SALK_076476), nup82-2 (SALK_024526), were obtained from Arabidopsis Biological Resource Center (ABRC). The gbpl3-6, gbpl3-7, gbpl3-8, and gbpl3-9 mutants were generated by CRISPR/Cas9-mediated gene editing. The crwn1, crwn2, crwn3, crwn4, crwn1 crwn4, and crwn2 crwn3 mutant seeds were kindly provided by Dr. Eric Richard at Boyce Thompson Institute. All double or triple mutants of gbpl3-7 with nup82/136 or crwns were obtained by genetic crossing and confirmed by PCR-based genotyping. pGBPL3: GBPL3-TurboID-3HA transgenic lines are in WT background and pGBPL3: mEGFP-GBPL3 and pGBPL3: mCherrry-GBPL3 transgenic lines are in gbpl3-7 mutant background. Four pGBPL3: mEGFP-GBPL3 lines with lower expression were identified, and 3 lines with higher expression were identified. All transgenic plants were generated by floral dip transformation using Agrobacterium tumefaciens GV3101 and selected using Basta herbicide by spraying. All primers used for genotyping were listed in S4 Data. All seeds were surface sterilized and germinated on half-strength Murashige and Skoog (1/2 MS) agar plates after stratification at 4°C for 3 days. Seven-day-old seedlings grown on 1/2 MS medium were transferred to soil for further growth. Arabidopsis and Nicotiana Benthamiana plants were grown under a 16 h light/8 h dark-light cycle at 22°C.

Plasmid construction

Gateway (Thermo Fisher) and In-fusion (Vazyme) technology were used for cloning unless otherwise specified. To generate constructs for protein localization, coexpression, BiFC, complementation, proximity labeling assay, mCherry, mEGFP, n/c-YFP, and TurboID-3HA were cloned into entry vector pBSDONOR p4r-2 or pBSDONOR p1-4, and cDNA fragments of Nup136, Nup82 or genomic DNA fragments of GBPL3, CRWN1, CRWN3 were cloned into entry vector pBSDONOR p1-4 or pBSDONOR p4r-2 using BP reaction (BP clonase II). The native promoter of GBPL3 (1,314 bp upstream of the start codon) was amplified and cloned into empty vector pEG100 by In-fusion cloning (ClonExpress II One Step Cloning Kit, Vazyme). All fusion constructs were generated by combining 1 pBSDONOR p1-4 construct, 1 pBSDONOR p4r-2 construct, and an original or modified pEG100 destination vector by LR reaction (LR clonase II). To generate the construct for histochemical staining, the cDNA fragment of GUS was amplified and inserted into entry vector pBSDONOR p1-2 and further cloned into modified pEG100 containing the native promoter of GPBL3 using LR reaction. To generate constructs for components of the GBPL3 proxitome, cDNA fragments of ARP4, APRF1, PWO1, TAF5, ARR1, AKS2, CPL1, TPR1, UBP1C, and SMU1 were cloned into a modified pEG100 vector containing GFP by In-fusion cloning. For PWO1, genomic DNA fragment was cloned into pBSDONOR p1-4, which was subsequently combined with pBSDONOR p4r-2-mEGFP into pEG100 using LR reaction. All binary expression constructs are under 35S promoter except GBPL3, which is driven by its native promoter. To generate constructs for yeast-two hybrid assay, cDNA fragments of GBPL3-C (GBPL3 without the GTPase domain), Nup136, Nup82, CRWN1, CRWN2, CRWN3 were subcloned into pGBKT7 or pGADT7 by In-fusion cloning. CRISPR/Cas9 constructs for gene editing were described previously [44]. Briefly, single guide RNA (sgRNA) sequences that target GBPL3 were created using the online webserver CRISPOR (http://crispor.tefor.net/). sgRNA scaffold was then amplified using pCBC-DT1T2 vector as the template and inserted into an egg cell promoter-driven pHEE401 binary vector by golden gate assembly. All primers used for cloning were listed in S4 Data.

Fluorescence imaging analysis

To image Arabidopsis plants, 5-day-old pGBPL3-mEGFP-GBPL3 transgenic seedlings with low or high expression were used. Coexpression and BiFC assay were performed using agrobacterium-mediated transient protein expression in N. benthamiana as previously described [45]. Agrobacterium carrying corresponding constructs were mixed and infiltrated using a needless syringe into leaves of 4-week-old N. benthamiana plants, and images were taken in leaf epidermal cells 2 days past infiltration. Fluorescence images were obtained using a Zeiss LSM880 inverted confocal microscope with the GaAsP detector.

Transmission electron microscopy and immunogold labeling

Root tips from Arabidopsis seedlings stably expressing mEGFP-GBPL3 were dissected and cryofixed by an HPM100 (Leica Microsystems). The frozen specimens were freeze-substituted at −80°C and embedded in HM20 resin (Electron Microscopy Sciences) at −45°C. After polymerization at −45°C, samples were washed with acetone and embedded in Lowicryl HM20 resin with an increasing gradient of 33%, 66%, and 100% HM20 concentration. After UV polymerization at −45°C, the root tips were sectioned and immunolabeled with a GFP antibody (Santa Cruz) as described previously [68]. Sample sections were observed under a transmission electron microscope after being stained with uranyl acetate and lead citrate. Gold particles in 3 individual roots of Arabidopsis seedlings expressing mGFP-GBPL3 and wild-type seedlings were counted. A total of 40 images were taken for each individual root, and gold particles located on the nuclear membrane (inner nuclear membrane, outer nuclear membrane, and nuclear pore complex), nucleoplasm, and the cytoplasm were counted.

Bioinformatic analysis

Proteomic data of GBPL3 were retrieved from our previous subtractive proteomics and proximity labeling proteomics using SUN1, MAN1, PNET2_A, PNET2_B, KAKU4, CRWN1, Nup82, SINE1, and WIP1 as baits [44,45] (PXD015919 and PXD026924). Samples were re-normalized before being visualized as box plots, bubble plots, and heatmaps using R packages ggplot2 and pheatmap. To generate the co-expression network of GBPL3 with known nucleoporin and nucleoskeleton genes, the averaged logic scores based on both microarray and RNA-seq data were retrieved from ATTED-II version 11.0 (https://atted.jp/) and visualized in a clustering heatmap using R. Protein domain prediction was performed in SMART (http://smart.embl-heidelberg.de/). Nuclear localization signals were predicted by cNLS Mapper (http://nls-mapper.iab.keio.ac.jp/cgi-bin/NLS_Mapper_form.cgi) with a score of more than 5.0. Intrinsically disordered regions (IDRs) were predicted by the Database of Disordered Protein Predictions (D²P²) (https://d2p2.pro/) and Predictor of Natural Disordered Regions (PONDR) (http://www.pondr.com/).

Phylogenetic analysis

Phylogenetic analysis of GBPL3 was performed as previously described [44]. Briefly, GBP protein sequences from various plant and animal species including Arabidopsis thaliana, Populus trichocarpa, Medicago_truncatula, Gossypium_raimondii, Brachypodium distachyon, Hordeum vulgare, Zea mays, Oryza sativa, Selaginella moellendorffii, Physcomitrella patens, Chlamydomonas reinhardtii, Caenorhabditis elegans, Homo_sapiens, Xenopus_tropicalis, Mus_musculus, Gallus gallus, Danio_rerio, and, Physcomitrium patens were downloaded from JGI database (https://genome.jgi.doe.gov/portal/). The sequence alignment was carried out using ClustalW2, and the phylogenetic tree was built by MEGA-X (https://www.megasoftware.net/) using the maximum likelihood method with 1,000 bootstrap replicates. All sequences of GBPL3 homologs are provided in S1 Data.

Histochemical staining with X-Gluc

To examine the expression pattern, GUS staining was performed as described previously [69]. Entire seedlings and tissues including leaf, stem, flower, anther, and silique from transgenic plants expression pGBPL3-GUS were used. All fresh tissues were incubated in the darkness overnight at 37°C following vacuum infiltrated in the staining buffer (100 mM sodium phosphate buffer (pH 7.0), 10 mM EDTA, 0.1% Triton X-100, 2 mM potassium ferrocyanide, 2 mM potassium ferricyanide, 1 mM X-gluc (5-bromo-4-chloro-3-indolyl glucuronide)) for 1 h. The stained seedlings or tissues were washed with 70% ethanol to remove chlorophyll before being imaged using Zeiss Lumar Stereoscope.

Yeast two-hybrid

Yeast two-hybrid analysis was performed using the Matchmaker GAL4-based Two-Hybrid System 2 (Clontech) according to the manufacturer’s instructions. The activation domain fusion (prey) constructs and the DNA-binding domain fusion (bait) constructs were transformed into yeast strains Y187 and AH109, respectively. The yeast cells of Y187 and AH109 expressing corresponding proteins were mated in 2 × YPDA medium at 30°C for 24 h before plated and selected on double (SD-Leu-Trp) and quadruple (SD-Leu-Trp-Ade-His) dropout media at 30°C for 3 to 5 days.

Immunoblot analysis

Total protein was extracted from transgenic seedlings using protein extraction buffer (50 mM Tris (pH 7.5), 150 mM NaCl, 0.5% Triton X-100, 0.5% Nonidet P-40, 0.5% Nadeoxycholate, plant protease inhibitor cocktail, 1 mM PMSF, and 40 μm MG132). Protein extracts were separated by SDS-PAGE, followed by immunoblot analysis using an anti-GFP antibody (Clontech, Cat #632381, dilution 1:5,000), anti-HA antibody (Roche, Cat #11867431001, dilution 1:5,000), or streptavidin-HRP (Abcam, Cat#7403, dilution 1:10,000).

Proximity labeling and affinity purification

The TurboID-based proximity labeling and affinity purification of biotinylated protein have been described previously with modifications [70]. Briefly, 1-week-old pGBPL3-GBPL3-TurboID-3HA transgenic seedlings and wild-type non-transformants were treated with 50 μm biotin for 4 h at room temperature. For each sample, 0.4 g of treated seedlings were harvested and frozen directly in liquid nitrogen. Two biological replicates were used for each sample. The material was grounded into fine powder, and total protein was extracted with protein extraction buffer (50 mM Tris (pH 7.5), 150 mM NaCl, 0.5% Triton X-100, 0.5% Nonidet P-40, 0.5% Nadeoxycholate, protease inhibitor cocktail, and 40 μm MG132). Protein extraction was subjected to 2 tandem PD-10 desalting columns (GE-Healthcare) to deplete free biotin. The following flow-through was collected for affinity purification. The eluted protein fraction was mixed with 50 μL pre-washed streptavidin-coated magnetic beads (Dynabeads MyOne Streptavidin C1, Invitrogen). The sample was incubated on a rotor wheel overnight at 4°C, and the beads were separated on a magnetic rack and washed 5 times with the protein extraction buffer.

On beads digestion, mass spectrometry, and proteomics analysis

For on-beads tryptic digestion, the streptavidin beads were washed with PBS buffer 3 times and incubated with 1 μg trypsin in 100 μL 50 mM triethylammonium bicarbonate buffer (TEAB) solution overnight at room temperature with gentle shaking. The resulting digests were separated from the beads on a magnetic rack and dried down by Speedvac. The peptides were dissolved with 10 μL 0.1% trifluoroacetic acid (TFA) and desalted using 10 μL C18 desalting ZipTips according to the manufacturer’s instructions. The purified peptides were then dried down and re-dissolved with 20 μL formic acid (FA) before being subjected to LC-MS/MS analysis. MS/MS spectra were searched against TAIR 11 database using Scaffold 5 and MSFragger 3.2 software with default criteria to harvest PSM and LFQ intensities, respectively. To analyze proximity labeling data, both PSM and LFQ data were used as input, and DESeq2 and DEP packages in R were applied to enrich GBPL3 proximal candidates with cutoffs: fold-change > 4, p < 0.01, and normalized PSM > 4. Volcano plot and heatmap were generated by R. To construct the protein–protein interaction networks of GBPL3, interactions were retrieved from the STRING database (Search Tool for the Retrieval of Interacting Genes/Proteins) (https://string-db.org/) and visualized by Cytoscape (version 3.9.0).

RNA extraction and real-time PCR

RNA was extracted using TRIzol reagent (Invitrogen, Cat #15596026) from about 0.1 g of 5-day-old seedlings. The first-strand cDNA was synthesized using the Maxima First Strand cDNA Synthesis with dsDNase Kit (Thermo Scientific, Cat #K1672). SYBR Green PCR Master Mix (Thermo Fisher, Cat # 4309155) was used for qPCR reaction, which was performed on a CFX96TM Real-Time PCR Detection System (Bio-Rad). ACTIN2 was used as a reference gene, and 2 to 3 biological replicates were used for each sample. All qPCR primers can be found in S4 Data.

RNA sequencing analysis

For RNA-seq, 1-week-old media-grown Col-0 and gbpl3-7 seedlings were collected. Total RNA was extracted using the Direct-zol RNA Miniprep kit (Genesee Scientific, Catalog #11–330) with a DNase treatment, following the manufacturer’s instructions. Library construction, quality control, and RNA sequencing were performed by Novogene (https://en.novogene.com/) on an Illumina HiSeq 2000. Each sequencing run results in 38 to 50 million raw reads per sample. Sequencing read counts were quality checked and trimmed to get rid of adaptor contaminations and low-quality reads. Cleaned reads were aligned to the reference genome of Arabidopsis (TAIR10) using Hisat2 (v2.1.0). SAM files were sorted and converted to BAM files using SAMtools (v1.14). Read counts of each gene were generated using HTseq-count with default parameters. DESeq2 package (v3.14) was utilized to analyze the difference in gene expression. Differentially expressed genes (DEGs) were determined with criteria of p_value < 0.05 and fold change > 2. Gene ontology annotation of DEGs was conducted using agriGO v2.0 (http://systemsbiology.cau.edu.cn/agriGOv2/). The volcano plot and bubble plot were generated using R package ggplot2 (v3.3.5).