Identification of NUPs at PD

(A) Localization of NUPs from Physcomitrium patens at PD.

(B) Hierarchical clustering of PD scores of Arabidopsis NPC components found in PD proteome preparations from 5-week-old Arabidopsis shoot tissues. Clustering based on Euclidean distance of PD scores (v1.5), which were averages of at least three replicate scores from separate tissue preparations. Color scale represents PD score values. PD scores above the high-confidence threshold of 0.5 are indicated in shades of red. (C) Scheme of an NPC embedded in the nuclear envelope with all NUPs studied in this work. NUPs listed in B are green boxed. Green letters: NUPs significantly enriched at PD in transient localization assays (Figure 2B, 3B). Green boxed black letters: NUP was identified in the proteome, but was not enriched at PD in transient localization assays. Corresponding AGI locus identifiers in table S2. Cytoplasmic NUP is abbreviated as Cyt. NUP, and membrane anchor as Membr. anchor.

Dual localization of Arabidopsis FG NUPs

(A) Localization of selected Arabidopsis FG-NUPs after transient expression under the control of the β-estradiol-inducible XVE promotor in N. benthamiana epidermal cells using confocal microscopy. mVenus was fused to the C-terminus of different NUPs, except for NUP98a for which GFP was fused both N- and C-terminally (GFP-NUP98a-GFP). Aniline blue was infiltrated to detect callose in pit fields as a PD marker. Localization was repeated in three independent experiments with similar results. (B) Quantification of the PD localization by PD index. Red dashed line indicates the mean PD index of mScarlet3 (cytoplasmic control). PI, propidium iodide (‘membrane impermeant dye’). Significantly (p < 0.05) increased PD indices compared to the cell wall or cytoplasmic controls are indicated by * and #, respectively. Kruskal-Wallis Test p < 0.001, p-values of Bonferroni corrected pairwise test are summarized in table S2. Median is represented by vertical line inside the box, mean is represented by the bold +. Values between quartiles 1 and 3 are represented by box ranges, and 5th and 95th percentile are represented by error bars. At least 15 images from 3 biological replicates were analyzed for each NUP. Independent localization data are shown in Figure 1-figure supplement 3. (C-E) Arabidopsis NUP62-GFP plants expressing NUP62 fused to GFP under its own promoter. (C) The localization of NUP62-GFP in 11-day-old cotyledon epidermal cells was visualized by confocal microscopy (CSU10; 100x oil-immersion objective). Arrowhead: blue nucleus, white peripheral NUP62-GFP. (D) Localization of NUP62-GFP in cotyledonar epidermal cells of stably transformed 11-day-old (top) and 4-day-old (bottom) Arabidopsis lines. Cotyledons were immersed in aniline blue to detect callose in pit fields as PD markers. White arrowheads in D indicate overlay of NUPs and aniline blue. Scale bar: 10 µm. (E) Maximum intensity projection of 20 optical cross-sections taken at 1µm intervals of epidermal cotyledon cells expressing NUP62 fused to GFP under its own promoter (11-day-old seedling). Punctate localization of NUP62-GFP can also be found at the epidermis cell surface on the adaxial side of the cell.

Dual localization of Arabidopsis scaffold and transmembrane NUPs

(A) Localization of Arabidopsis transmembrane (CPR5) and scaffold (NUP43) NUPs after transient expression under the control of the β-estradiol-inducible XVE promotor in N. benthamiana epidermal cells using confocal microscopy. mVenus was fused to the C-terminus of different NUPs. Aniline blue was infiltrated to detect callose in pit fields as PD marker. White arrowheads indicate overlay with aniline blue. Localization was repeated in three independent experiments with similar results. (B) Quantification of the PD localization by PD index. Red dashed line indicates the mean PD index of mScarlet3 (cytoplasmic control). PI, propidium iodide (‘membrane impermeant dye’). Significantly (p < 0.05) increased PD indices compared to cell wall or cytoplasmic controls indicated by * and #, respectively. Kruskal-Wallis Test p < 0.001, p-values of Bonferroni corrected pairwise test are summarized in table S2. Median: vertical line inside box, mean: +. Values between quartiles 1 and 3: box ranges, and 5th and 95th percentile are represented by error bars. At least 15 images from 3 biological replicates were analyzed for each NUP. Independent data in Figure 1-figure supplement 3. Control identical to Figure 2B. (C) NUP43-mCitrine localization at PD at 1/10 β-estradiol concentration and shortened induction time (not exceeding 10h). White arrowheads: overlay with aniline blue. (D and E) Quantification of the fluorescence intensity of NUP43-mCitrine (D) and CPR5-FP fusions (mVenus and mCitrine) (E) in the periphery and the nucleus after induction by 20µM or 2µM β-estradiol. (F) Periphery/nucleus fluorescence ratios in plants expressing NUP43-mCitrine or CPR5-FP fusions after expression is induced by 2 or 20µM β-estradiol. Box plots in (D-F) are based on a smaller subset of confocal images that show nucleus and peripheral fluorescence in one image; for individual conditions: n ≥ 3 images from at least 2 independent repeats. Student’s t-test in (F) 2µM vs. 20µM estradiol: for NUP43, p = 0.26 and for CPR5, p = 0.45. (G) Localization of HOS1/ELYS-mCitrine and PDLP5-mScarlet3 co-expressed transiently under the control of the inducible XVE promoter in N. benthamiana epidermal cells. Aniline blue was infiltrated to stain callose that accumulates in pit fields. White arrowheads: overlay of HOS1/ELYS-mCitrine and PDLP5-mScarlet3 channels. Localization repeated in three independent experiments with comparable results.

CPR5 topology and localization of CPR5 near PD orifices

Split-GFP membrane topology to test for the orientation of CPR5 in the nuclear pore membrane and the ER at PD. (A) (left) Fluorescence reconstitution at PD between a cytosolic non-fluorescent GFP domain lacking β-sheet 11 (GFP1-10; the circle missing a wedge) and a fusion of the GFP11 to the N-terminus of CPR5 (the red wedge). Since reconstitution occurs at PD, each cartoon indicates reconstitution in the intermembrane space between ER/desmotubule and plasma membrane. (right) Cytosolic GFP1-10 and GFP11fusion to C-terminus of CPR5. (B) (left) ER-retained GFP1-10-HDEL and GFP11 fused to N-terminus of CPR5. (right) ER-retained GFP1-10-HDEL and fusion of GFP11 to C-terminus of CPR5. Repeated independently 3 times with comparable results. (C) CPR5-mCitrine close to the orifices of PD by Structured Illumination Microscopy in N. benthamiana leaves co-infiltrated with XVE:PDLP5-mScarlet3 and XVE:CPR5-mCitrine. Prior to imaging, leaves were infiltrated with aniline blue. PDLP5-mScarlet3: turquoise, CPR5-mCitrine: yellow, aniline blue: magenta. Repeated independently four times with comparable results. (D) Cartoon approximately to scale, illustrating PD in pit field between adjacent cells with callose marked magenta, PDLP5 turquoise, CPR5 yellow.

Defects in cell-to-cell transport of kDa-sized cargo in cpr5 mutants

(A) Example images of intercellular 2xmEGFP movement in wild type (WT, left) and cpr5-1 (right) Arabidopsis leaves. Leaves were bombarded with gold particles coated with p35S:mEGFP-mEGFP DNA plasmids 24 h prior to the analysis by confocal microscopy (fusion protein abbreviated as 2xmEGFP). (B) Quantification of the intercellular spread of mEGFP by counting fluorescent cells in the GFP channel. n(WT) = 594 bombarded cells, 17 independent biological replicates: n(cpr5-1) = 278 bombarded cells, 6 independent biological replicates. n(cpr5-T3) = 99 bombarded cells, four independent biological replicates. a indicates significant difference to WT with p(cpr5-T3) = 0.0004; b indicates significant difference to WT with p(cpr5-1) < 10-15; p(cpr5-1 vs. cpr5-T3) = 0.0002 (Kruskal-Wallis test with Bonferroni correction for pairwise comparison). (C) Callose quantification. n(WT) = 4 independent experiments, n(cpr5-1) = 5 independent experiments, graph summarizes all experiments. Student’s t-test p = 0.48. (D) Sum projections showing SHR–GFP distribution in WT (left) and cpr5-1 mutant (right) root. White circles and boxes indicate positions of endodermis nuclei (E, arrows) and corresponding regions in the stele (S, white boxes). (E) Quantification of SHR–GFP endodermis:stele ratio (E:S) for WT and cpr5-1 mutant. n(WT) = 62 nuclei, n(cpr5-1) = 62 nuclei, with 3-4 nuclei analyzed per root. a indicates significantly decreased E:S in cpr5-1 mutant, based on Mann-Whitney U test with p = 0.0006. (F), Sum projections showing SHR–GFP distribution in WT (top) and cpr5-1 (bottom) background. (G), Quantification of relative SHR–GFP fluorescence recovery for WT and cpr5-1 mutant after 1 h. Fluorescence recovery was normalized to a control nucleus in the endodermis (see Methods and Figure 5-figure supplement 1F). n(WT) = 59 bleached nuclei, n(cpr5-1) = 59, per root 3-4 nuclei were bleached. Student’s t-test p = 0.037 indicating significantly decreased fluorescence recovery after 60 min. (H and I) PD permeabilities examined by CFDA/CFSE-based DANS dye loading assays. (H) Epifluorescence reporting on CF diffusion in the epidermis of the adaxial (left) and abaxial (right) side of the leaf. WT (top), 35S:PDLP5 (middle), and cpr5-1 (bottom). Bars, 1 mm. (I) Quantification of CF diffusion in WT and mutant lines. n(WT) = 49 plants, 7 independent biological replicates; n(35S:PDLP5) = 38 plants, 5 independent biological replicates; n(cpr5-1) = 29 plants, 5 independent biological replicates; n(cpr5-T3) = 15 plants, 5 independent biological replicates. Fluorescent areas on the abaxial side were identified using auto threshold and Fiji YEN-algorithm with user modifications. The same threshold setting was used for the adaxial side. The extent of dye diffusion was quantified by the ratio between the areal spread of fluorescence on the abaxial side and the areal spread of fluorescence on the adaxial side. a indicates significantly different at the α = 0.032 level according to Bonferroni-corrected pair wise comparison following Kruskal-Wallis analysis.

Overlay of Physcomitrium patens FG-NUPs with the PD marker aniline blue

Localization of FG-NUPs from P. patens in N. benthamiana epidermal cells visualized by confocal microscopy. FG-NUPs were fused at their C-terminus to mVenus (yellow) and expressed transiently under the control of the inducible XVE promotor. Aniline blue (purple) was infiltrated to stain callose, which accumulates in pit fields. Arrowheads indicate overlay between NUPs and aniline blue. For each NUP, a similar localization pattern appeared in at least three independent experiments. NC = nucleus. Scale bar: 10 µm.

Arabidopsis shoot proteome and the comparison to other PD proteomes

(A), Venn diagram defining high-confidence PD proteins from previous publications, e.g., confirmed Arabidopsis PD proteins, and high confidence PD proteins identified in Physcomitrium, and in both Physcomitrium and Arabidopsis leaves (Gombos et al., 2023; Johnston et al., 2023; Kirk et al., 2022). These proteins were then used as high-confidence data set to establish the PD probability threshold and were labelled as ‘PD proteins’ in file S3. 43% of these proteins were quantified in our Arabidopsis shoot PD proteome. (B), PD Scores of nuclear pore components. PD Scores (v 1.5) are shown as box plots for high-confidence PD proteins (PD proteins), typical co-purifying proteins from plastid and mitochondria identified in Arabidopsis PD preparations (likely contaminants) and proteins of the nuclear pore complex (NUP). Dotted lines indicate two thresholds based on the frequency distribution for contaminant proteins and PD proteins. The median is shown as a gray line, the mean value is denoted by a + symbol, and the whiskers cover the 95thpercentile. PD Scores above 0.83 (the upper dotted line) indicate high-confidence (HC) PD proteins; scores between 0.63 and 0.83 (between the two dotted lines) signify medium-confidence (MC) PD proteins; and scores below 0.63 are considered low confidence (LC) as their PD Scores overlapped with 75% of the contaminants. These confidence clusters were annotated in file S3. The final Arabidopsis shoot PD proteome consists of proteins with PD scores above 0.63, which are also quantifiable in the PD fraction. Capital letters represent group statistical comparisons determined by Oneway-ANOVA. The p-values of pairwise comparison using Student’s t-test, are indicated on the top of the boxes. (C), Overlap between the Arabidopsis shoot PD proteome and other PD proteomes. Proteins with a PD score >0.63 and quantifiable in PD fraction (Arabidopsis shoot PD proteome) were compared with nine representative PD proteome studies. In the current Arabidopsis shoot PD proteome, 80% of the proteins were previously identified in published PD proteomes. Combining PD enrichment and data analysis thresholding allowed us to determine the confidence levels of PD proteins and obtain a “cleaner” PD proteome. In total, we identified 5145 proteins from PD, CW, and TC fractions, with 3347 proteins having a PD score >0.63. Among them, 3062 proteins are quantifiable in the PD fraction, constituting our Arabidopsis shoot PD proteome. Among these, 2268 proteins are categorized as high-confidence PD proteins, while 792 proteins fall into the medium-confidence category (file S3). Additionally, we discovered 20 nuclear pore proteins (NUPs), with 15 of them surpassing the threshold and being included in the final PD proteome. Among these, 11 are classified as high-confidence PD proteins, while 4 are medium-confidence PD proteins (file S3, Figure 1-figure supplement 2B). (D) Comparison of Arabidopsis shoot PD proteomes with various published PD proteomes. 80% of the proteins had been identified in previous studies.

NUPs are found in Arabidopsis shoot proteome and other PD proteomes. (A),

Comparison of NUPs in Arabidopsis shoot PD proteomes with published PD proteomes. Among the NUPs identified in the Arabidopsis shoot PD proteome, 73.3% (11 out of 15) were also detected in previously published PD proteomes. (B), Overview of identified proteins in the PD proteome. HC (PD score > 0.83), MC (0.83 > PD score > 0.63), and Unknown (PD score < 0.63). Purple-edged boxes in HC and MC mark proteins that meet Arabidopsis shoot PD proteome criteria (PD score > 0.63, quantifiable in PD fraction). Blue-edged boxes denote 16 NUPs within the Arabidopsis shoot PD proteome; Gray-edged boxes represent NUPs found also in other proteomes.

Normalized protein intensities of different protein groups in total extract (TC), cell wall fraction (CW) and PD enrichment (PD)

The group of likely contaminants contains plastidic and mitochondrial proteins. Bona fide PD proteins are PDLPs and MCTPs. Protein categorization was based on Mapman (Usadel et al., 2005). A Student’s t-test was used for statistical analysis.

Arabidopsis NUP-FP localization overlays with a PD marker

Localization of Arabidopsis NUPs in N. benthamiana epidermal cells visualized by confocal microscopy. NUPs were fused at their C-terminus to mVenus or mCitrine (yellow) and expressed transiently under the control of the inducible XVE promotor. Aniline blue (purple) was infiltrated to stain callose that accumulates in pit fields. Arrowheads indicate overlay between NUPs and aniline blue. NUPs shown here were found to have a significant increased PD index values compared to negative controls in Figure 2. For each NUP, the localization pattern was reproducible in at least in three independent experiments. NC, nuclei. Note that for comparison, we here show the same images for NUP98B and NUP62 from Figure 2A again. Scale bar: 10 µm.

Predicted structure of monomeric CPR5 and model for integration into ER-derived membranes

Topological studies using the split-GFP system (Figure 4A,B, Figure 4-figure supplement 3) are consistent with a localization of the CPR5 C-terminus in the ER lumen, while its N-terminus extends into the cytoplasm. (A), CPR5 is composed of a N-terminal, disordered region (residues 1-106), a soluble domain formed by 3 α-helices (107-192), linked by a second disordered region (193-265) to an approx. 115 Å long, central α-helix (266-342). Hydrophobic residues (red) of the central helix (TM1) and 4 shorter helices (TM2-TM5) form a transmembrane domain spanning approx. 35 Å, consistent with the expected thickness for ER-derived membranes. C-terminus and helical extensions of TM1 and TM2 reach into the ER lumen. (B), Structure colored by per-residue prediction confidence (pLDDT; dark blue: 90-100, light blue: 50-70, yellow: 50-70, red: < 50). Low confidence values (red) may indicate regions that are either intrinsically disordered or assume a structure when interacting with other proteins and therefore could partake in formation of complexes with NUPs. (C), View of transmembrane helical bundle from ER lumen (left) and cytoplasm (right). Amino acid side chains spanning between transmembrane helices are colored by hydrophobicity (blue: hydrophilic, red: hydrophobic). The CPR5 transmembrane domain is not predicted to form a solvent-accessible pore.

Phylogenetic analysis of CPR5 sequences

The accession numbers and the abbreviations corresponding to the mature protein sequences used for this analysis are given in table S7. The analysis was performed as described in the “Materials and Methods” section.

The C-terminus of CPR5 faces ER lumen

(A), Transient co-expression of GFP11-CPR5 and cytosolic GFP1-10 in N. benthamiana leaves, same cell as shown in Figure 4. Left, GFP channel. Reconstitution can be observed also at PD (arrowheads). Middle, aniline blue. Right, merge. (B), N. benthamiana epidermal cells transiently expressing CPR5-GFP11 and ER luminal GFP1-10-HDEL, same cell as shown in Figure 4. Left, GFP channel. Reconstitution can be observed also at PD (arrowheads); middle, aniline blue; right, merge. (C), Surface view of the same cell as in B, showing reconstitution in the ER.

CPR5-mCitrine localizes at the orifices of PD

(A), Cartoon showing analyzed PD: only PD most proximal to the cover glass and objective were analyzed to reduce effects caused by scattering of light due to cell wall. (B), SIM in N. benthamiana leaves co-infiltrated with XVE:PDLP5-mScarlet3 and XVE:CPR5-mCitrine. Prior to the imaging, leaves were infiltrated with aniline blue. PDLP5-mScarlet3 in turquoise, CPR5-mCitrine in yellow, and aniline blue in magenta. Comparable results were obtained in four independent experiments. Dashed lines indicate approximation of cell wall boundaries. (C), Cartoon of pit field and localization of PDLP5 and CPR5. Fluorescence reporting on CPR5 in (B) extends laterally 200-1000 nm in the epidermal cell wall. Previously, the lateral extension of individual PD pores in the cell wall was estimated to 100 nm for basal trichome cell walls and 187 nm for epidermal cell walls (Faulkner et al., 2008; Fitzgibbon et al., 2010). The resolution achieved with SIM was not high enough to resolve single PD or determine whether pit fields consist of several simple PD (as shown in C) or a few complex (H-shaped/twinned) PD. (D), SIM in N. benthamiana leaves co-infiltrated with XVE:PDLP5-mScarlet3 and XVE:CPR5-mCitrine. Green arrowheads indicate overlay between CPR5-mCitrine, PDLP5-mScarlet3 and aniline blue; white arrowheads indicate overlay between PDLP5-mScarlet3 and aniline blue, without CPR5-mCitrine. PDLP5-mScarlet3 in turquoise, CPR5-mCitrine in yellow, and aniline blue in magenta.

Intercellular transport assays in cpr5-1 mutants

(A), Example of three different clusters (1, 2, and 3) of fluorescent cells after bombardment in a WT Arabidopsis leaf. Each cluster reports a successful bombardment event. Left, GFP channel, right, mCherry channel. 24 h prior to the experiment, leaves were co-bombarded with gold particles coated with p35S:mEGFP-mEGFP (2xmEGFP). DNA plasmids and gold particles coated with p35S:mCherry-mCherry-mCherry (3xmCherry). (B), Profile plots across the different clusters, 1, 2, and 3 as shown in A. Yellow lines in A indicate the position for the profile. The bombarded cell marked with 3xmCherry overlays with the brightest cell in the GFP channel. (C), Six-day old seedlings with shr/pSHR:SHR-GFP in WT Col-0 background (left) and shr/pSHR:SHR-GFP in the cpr5-1 background (right) grown on ½ MS agar. (D), Quantification of the root length of 6-day old WT and cpr5-1 mutant background seedlings. n(WT) = 23 roots, four independent experiments, n(cpr5-1) = 16 roots, three independent experiments. a Significantly decreased root length in cpr5-1 mutant, based on Student’s t-test with p = 0.00002. (E), Quantification of SHR-GFP stele baseline fluorescence in cpr5-1 mutant and WT background. n(WT) = 62 stele ROIs, n(cpr5-1) = 62 stele ROIs, with 3-4 ROIs analyzed per root. n.s., no significant difference in fluorescence intensity based on Mann-Whitney U test with p = 0.074. (F), SHR–GFP fluorescence recovery after photobleaching (FRAP) in homozygous shr/pSHR:SHR-GFP root endodermal cells (arrow heads). Note that for comparison, we here show the same sum projections as in Figure 5F (top). Blue circles show the position of ROIs for FRAP. Green circle shows the position of ROI for control nucleus used for normalization (see Methods). Yellow square shows the position for mean background fluorescence used for correction (see Methods).

Genome-wide search for potential phase-separating proteins with FG repeats

The 15 genes with the highest number of FG motifs within a 200-amino acid window in the Physcomitrium genome. For the list summarizing the genome-wide search for FG containing genes, see file S1. For Arabidopsis NUPs see file S2.

List of known Arabidopsis NUPs and summary of localization results

p-values calculated from Bonferroni-corrected pairwise comparison between the tested NUP with propidium iodide marked by *; p-values calculated from Bonferroni-corrected pairwise comparison between the tested NUP and free mScarlet3 marked by #. Index is short for PD index (see also Figure 2 and 3). NA: not assessed. FG-NUPs in red.

NPC components identified in published cell wall/PD proteomes

Nuclear localization and observation of puncta in the cell periphery, consistent with PD localization by other laboratories

Membrane/microtubule anchor proteins in NPC and ciliary gates across kingdoms

Accession numbers of CPR5 homologs used for the phylogenetic analysis shown in figure S6

Primers used in this study