Arabidopsis WT (Col-0), OE-FLA12, and OE-FLA12 mutant/domain swap plants were grown under long day (16 h light/8 h dark) conditions at 22/18 °C in controlled environment rooms. N. benthamiana plants were grown in standard glasshouse conditions under long day (16 h light/8 h dark) conditions at 25/20 °C. The light intensity for Arabidopsis and tobacco growth was 120 µM/cm²/s.

A pGreenII0179 vector backbone (Hellens et al., 2000) was used for constructing proFLA12::His-YFP-FLA12 and proFLA12::His-YFP-FLA12 mutant variant vectors as used in previous study (Ma et al., 2022). The yellow fluorescence protein (YFP) used in this study is VENUS (Rekas et al., 2002). Supplementary Tables S1 and S2 provide the list of vectors and primers used for FLA12 domain mutation/deletion. Supplementary Tables S3 , S4 provide the list of vectors and primers used for FLA12 domain swaps. NEBuilder HiFi DNA Assembly kit (NEW ENGLAND Biolabs, NOTTING HILL Victoria, Australia) was used to construct vectors according to the manufacturer’s instructions. All vectors were confirmed by sequencing and then transformed into Agrobacterium strain AGL1. Arabidopsis plants were transformed using the flower dip method (Clough and Bent, 1998). Plants were then screened with hygromycin (35 mg/L).

The number of insertions and transgene copies (TC) were predicted based on segregation ratios in the T2 and T3 generation as outlined in Ma et al., 2023. Lines with 70%-85% survival ratio (3:1) in T2 were used for checking YFP signals in primary root vascular tissues. Plants from at least three independent T2 lines were transferred into soil and taken to the T3 generation for further phenotypic analysis. Morphological analysis of plants, including stem, silique and leaf length and width were performed on three biological replicates from three independent transformed lines. Statistically significant differences were determined using Student’s t-test, p < 0.05.

N. benthamiana leaf was infiltrated with Agrobacterium (OD_600 = 1.0 of each transformed vectors and p19) for imaging (Zeng et al., 2016). YFP (excitation 514 nm, emission 535 nm, detection 517-553 nm), α-ManI-ECFP (excitation 405 nm, emission 480 nm, detection 463-497 nm), and PIP2A-mCherry (excitation 594 nm, emission 632 nm, detection 604-660 nm) signals were detected in infiltrated N. benthamiana leaf of 2-3 days post infiltration using Zeiss LSM 780 laser scanning confocal microscope (Oberkochen, Germany).

Fresh stems were hand-sectioned and stained with Toluidine blue, phloroglucinol-HCl or Mäule reagent to visualize cell walls with an Olympus BX53 microscope (Tokyo, Japan) under bright field (Mitra and Loqué, 2014). At least three plant stems from three independent transformed lines were sectioned and measured for tissue organization analyses at the same developmental stages. Data are shown as mean ± SD. Student’s t-test was used for significance analysis with p < 0.05.

A 2 mm region at the base of stems at growth 6.9 (Boyes et al., 2001) when flowering complete, were chemically fixed, dehydrated, and embedded in LR white (Wilson and Bacic, 2012). Thin sections (~80 nm) were acquired for antibody labelling and post-stained (Wilson and Bacic, 2012). For antibody labelling, samples were incubated with anti-6x-His tag monoclonal antibodies (Invitrogen, # MA1-21315, Carlsbad CA, USA) with 1:100 dilutions for 2 h at room temperature. Samples were then washed and incubated with goat anti-mouse 18 nm gold conjugated secondary antibody (Jackson Immuno Research #115-215-166, West Grove, PA, USA) with 1:20 dilutions for 1 h at room temperature. Detection of ultrastructure, HIS-YFP-FLA12 and HIS-YFP-FLA12mutGPI subcellular location was performed for three biological replicates. Grids were imaged using a Jeol 2100 EM microscope (Tokyo, Japan) equipped with a Gatan Orius SC 200 CCD camera (Pleasanton, CA, USA). Gold signals were quantified manually from TEM images and data shown as mean ± SD.

Alcohol insoluble residue (AIR) was prepared from Arabidopsis stems (Pettolino et al., 2012). The Updegraff method was used for crystalline cellulose content measurement (Updegraff, 1969). Acetyl bromide method was used to detect lignin content according to (Chang et al., 2008). Three biological replicates from three independent transformed lines were measured. Data represent mean ± SD. Student’s t-test was used for significance analysis with p < 0.05.

Total proteins were extracted from 10-day-old Arabidopsis seedlings using extraction buffer (100 mM Tris-HCl, pH 8.8, 150 mM NaCl, 1mM EDTA, 10% (v/v) glycerol and 1X cOmplete protease inhibitor cocktail (Sigma, #11697498001, St. Louis, MO, USA). GFP-trap (ChromoTek, #gtma20, Bayern, Germany) was used to enrich proteins according to the manufacturer’s instructions. Denatured proteins were used for SDS-PAGE and protein blotting analysis with anti-GFP (Invitrogen, #MA5-15256, Carlsbad CA, USA). Two biological replicates from two independent transformed lines were performed.

In previous studies, the expression patterns of FLA11 and FLA12 were shown to differ in young stems (Ma et al., 2022). FLA11 was shown to have higher expression levels in xylem vessels (XVs) than FLA12, whereas FLA12 expression levels were higher in interfascicular fibres (IFs) (Ma et al., 2022). To investigate how the different FLA expression patterns relate to SCW development and plant growth, we conducted promoter swap experiments between FLA11 and FLA12. Expression of proFLA11::YFP-FLA12 and proFLA12::YFP-FLA11 proteins was confirmed by YFP signals present in XVs of roots ( Supplementary Figure S1 ). Phenotyping of proFLA11::YFP-FLA12 WT plants showed no obvious plant growth phenotypes compared to wild-type (WT) plants ( Figure 1 ) or secondary wall development phenotypes related to OE-FLA11 as shown in our previous study (Ma et al., 2023). Phenotyping analysis of proFLA12::YFP-FLA11 WT plants with one or two transgene copies (TC) showed that proFLA12::YFP-FLA11 had moderate effects on plant stem and silique length compared with OE-FLA11 (proFLA11::YFP-FLA11) that has significantly reduced stem length ( Figure 1 ; Table 1 ) (Ma et al., 2022).

Consistent with previous studies (Ma et al., 2022), histology analysis of young stems showed that OE-FLA11 can trigger early SCW development in both XV and IF cells. Phenotypic analysis of young stems (stage 6.0; Boyes et al., 2001) showed that proFLA12::YFP-FLA11 plants exhibited earlier development of SCW in IFs compared to WT and later development than OE-FLA11 ( Figure 2 ). Histological analysis of stage 6.9 mature stems was performed to investigate the development of xylem and morphology of mature XVs. proFLA12::YFP-FLA11 2TC plant stems had a similar number of XVs in each xylem bundle compared to WT ( Supplementary Figure S2 ). In comparison, the diameters of XVs in proFLA12::YFP-FLA11 2TC plant stems were significantly smaller compared to OE-FLA11 and WT plants ( Supplementary Figure S2 ).

In addition to stem phenotypes, proFLA12::YFP-FLA11 WT 2TC plants showed reduced rosette leaf length, a smaller length/width ratio, and ruffled leaf phenotype that were not observed in OE-FLA11 or WT plants ( Figure 3 ; Table 1 ). Analysis of leaf vein patterns showed that in rosette leaves of proFLA12::YFP-FLA11 2TC the density of tertiary and higher order veins was increased compared to OE-FLA11 and WT ( Figure 3 ). These results suggest that coordination of differential expression patterns of FLA11 and FLA12 is important for the timing and development of SCWs in stems and in rosette leaves.

To identify if domain features influence FLA12 function, OE-FLA12 protein domains were swapped with corresponding regions from FLA11 or FLA3 ( Supplementary Figure S3 ). The fasciclin (FAS) domain of FLA3 (type F) is distinct to that of FLA11 and FLA12 (type O) and constructs swapping either FAS11 or FAS3, into FLA12 were analysed. Plants with two transgene copies of pFLA12::FLA12-FAS11 and pFLA12::FLA12-FAS3 showed similar stem length as WT plants ( Figure 4A ; Table 2 ). Previous studies of FLA11 and FLA12 show location predominantly in the secondary wall and plasma membrane of fibre cells, respectively, as well as different degree of glycosylation in the AG2 region (Ma et al., 2022). The GPI-anchor and AG2 regions were swapped to generate plants containing pFLA12::FLA12-FLA11GPI, pFLA12::FLA12-FLA11linker+GPI, pFLA12::FLA12-FLA11AG2+linker+GPI and pFLA12::FLA12-FLA3AG+GPI constructs. Phenotyping of plants showed that pFLA12::FLA12-FLA11AG2+linker+GPI had shorter stem length compared to WT whereas all other constructs led to similar stem length as WT plants ( Figure 4A ; Table 2 ).

Mäule staining was further used to investigate cellular morphology in stem transverse sections and showed no obvious stem morphology or lignin composition changes of all transgene lines compared to WT ( Figures 4B–D ). Measurement of lignin content in mature stems of stage 6.9 (Boyes et al., 2001) showed that pFLA12::FLA12-FAS11, pFLA12::FLA12-FAS3, pFLA12::FLA12-FLA11GPI, pFLA12::FLA12-FLA11linker+GPI, and pFLA12::FLA12-FLA3AG+GPI plants similar to WT ( Table 2 ). In contrast, pFLA12::FLA12-FLA11AG2+linker+GPI plants showed higher lignin content compared to WT ( Table 2 ). Measurement of crystalline cellulose content in mature stems of stage 6.9 (Boyes et al., 2001) showed that only pFLA12::FLA12-FLA11GPI plants with a slight decrease compared to WT ( Table 2 ). These results suggest that the AG2 domain of FLA12 is the major region that differentiates FLA12 functions from FLA11 as only OE-FLA12-FLA11AG2+linker+GPI plants showed reduced stem length and increased stem lignin content phenotypes compared to WT.

We have previously shown FLA12 plays a role regulating cellulose levels, likely in response to mechanical stress, whereas FLA11 regulates lignin levels and composition (Ma et al., 2022). The location of FLA12 is predominantly at the PM whereas FLA11 is largely released into the SCWs (Ma et al., 2022). The location of FLA12 at the cell surface and potential to act as a surface sensor led us to investigate FLA12 further. FLA12 mutant variants were generated to determine the influence on trafficking, location and biological functions. Mutant variants were generated in the AG glycomotifs that direct O-glycosylation at either the N-terminus (mutAG1), the C-terminal AG region (mutAG2) or both regions (mutAG1 + 2), variants with the GPI-anchor signal sequence removed (mutGPI), mutation of three predicted N-glycosylation motifs within the FAS1 domain by replacing NXS/T with AXA (mutNglyA, a non-conservative mutation which will prevent N-glycosylation or binding site function) and replacing NXS/T with QXS/T (mutNglyB, a conservative mutation which should retain function if a binding site) ( Figures 5A, B ).

In order to investigate the influence of these changes on post-translational modification, trafficking and location of FLA12, pro35S::YFP-FLA12 mutant variants were transiently expressed in N. benthamiana leaves. A previous study showed that tagRFP-FLA12 proteins located both at PM and cell wall of N. benthamiana leaf epidermal cells (Pinski et al., 2021). In this study, N. benthamiana leaf epidermal cells co-expressing each YFP-FLA12 mutant variant together with a Golgi marker (α-ManI-ECFP) and a PM marker (PIP2A-mCherry) were examined by fluorescence microscopy for the location of the fluorescence signals. YFP signals were observed at the cell wall, PM, and Golgi in plasmolyzed epidermal cells of leaves expressing YFP-FLA12 ( Supplementary Figure S4A ). YFP signal in cells expressing YFP-FLA12 were shown to largely overlay with the Golgi marker and a minor component with the PM marker. YFP signals in mutAG1, mutAG2 and mutAG1 + 2 fusion proteins showed similar location patterns to FLA12 ( Supplementary Figures S4B–D ). In contrast, YFP signals in mutGPI were shown to principally overlap with the PM marker ( Supplementary Figure S4E ). YFP signals in mutNglyA and mutNglyB were similar to FLA12 ( Supplementary Figures S4F–G ). The sub-cellular location of the FLA12 mutant variants supports the role of the GPI-anchor for correct trafficking and location of FLA12.

Transient expression of pro35S::YFP-FLA12 in N. benthamiana epidermal cells, which only contain primary walls, may not reflect the location of FLA12 in cells with secondary walls. Constructs containing the FLA12 mutated variants driven by the endogenous FLA12 promoter were therefore stably transformed into WT Arabidopsis plants to enable investigation of FLA12 glycosylation and sub-cellular locations in stems. All transgenic plants carrying FLA12 mutations were confirmed with DNA sequencing and showed expression of YFP signals in XVs of seedling primary roots as expected for the FLA12 promoter ( Supplementary Figure S5 ). No obvious plant growth or stem length phenotypes were found between the OE-FLA12 mutant variants and either WT or OE-FLA12 plants of stage 8.0 and later ( Supplementary Figure S6 ). Western blot analysis showed that all fusion proteins were expressed in comparable amounts ( Figure 5C ). OE-FLA12mutAG2, OE-FLA12mutAG1 + 2 and OE-FLA12mutN-glyA only showed sharp bands around 60 kDa, the expected size of HIS-YFP-FLA12 fusion proteins without glycosylation ( Figure 5C ). OE-FLA12, OE-FLA12mutAG1 and OE-FLA12mutN-glyB showed a smeared band around 80-110 kDa that likely corresponds to AG glycosylated proteins, suggests the main O-glycosylation occurs on AG2 ( Figure 5C ). OE-FLA12mutGPI showed two sharp bands around 50-60 kDa. The sharp bands in OE-FLA12mutN-glyA and OE-FLA12mutN-glyB had an obvious size shift (~5 kDa) to a smaller size, indicating that FLA12 is likely N-glycosylated ( Figure 5C ).

Transient expression of pro35S::YFP-FLA12mutGPI in N. benthamiana indicated the GPI-anchor influences FLA12 trafficking and location ( Supplementary Figure S4 ). The PM/wall was difficult to distinguish in the epidermal cells of N. benthamiana even with plasmolysis, therefore the effect of the GPI-anchor on FLA12 location in SCWs was investigated in stably transformed Arabidopsis plants in IF cells in stems using transmission-electron microscopy (TEM). Immuno-gold labelling was used to detect the HIS-tagged FLA12 in ultrathin transverse sections of stems in the base region at growth stage 6.9 (Boyes et al., 2001). In OE-FLA12 lines, immuno-gold labelling was detected mostly at the PM-wall (SCW) interface ( Figure 6 ). In contrast, the immuno-gold labelling in mutGPI plants was present mostly in cytoplasm ( Figure 6 ).