Lack of FIBRILLIN6 in Arabidopsis thaliana affects light acclimation and sulfate metabolism

Arabidopsis thaliana contains 13 fibrillins (FBNs), which are all localized to chloroplasts. FBN1 and FBN2 are involved in photoprotection of photosystem II, and FBN4 and FBN5 are thought to be involved in plastoquinone transport and biosynthesis, respectively. The functions of the other FBNs remain largely unknown. To gain insight into the function of FBN6, we performed coexpression and Western analyses, conducted fluorescence and transmission electron microscopy, stained reactive oxygen species (ROS), measured photosynthetic parameters and glutathione levels, and applied transcriptomics and metabolomics. Using coexpression analyses, FBN6 was identified as a photosynthesis-associated gene. FBN6 is localized to thylakoid and envelope membranes, and its knockout results in stunted plants. The delayed-growth phenotype cannot be attributed to altered basic photosynthesis parameters or a reduced CO2 assimilation rate. Under moderate light stress, primary leaves of fbn6 plants begin to bleach and contain enlarged plastoglobules. RNA sequencing and metabolomics analyses point to an alteration in sulfate reduction in fbn6. Indeed, glutathione content is higher in fbn6, which in turn confers cadmium tolerance of fbn6 seedlings. We conclude that loss of FBN6 leads to perturbation of ROS homeostasis. FBN6 enables plants to cope with moderate light stress and affects cadmium tolerance.

Plastid types vary in their FBN composition, suggesting that the various FBNs might have specialized functions in different classes of plastids (Singh & McNellis, 2011). Moreover, the FBNs have diverse molecular properties. Their molecular masses range from 21 to 42 kDa, their pI values vary between 4 and 9, and they possess different hydrophobicity profiles. This diversity suggests that each FBN family member has specific biological function(s) (Singh & McNellis, 2011;Lundquist et al., 2012). The first role assigned to members of the FBN family was an involvement in fibril structure formation. Thus, its founder member FBN1, which was purified from bell pepper chromoplasts together with carotenoids and polar lipids, can reconstitute the complete fibril structure in vitro (Deruere et al., 1994). Another proposed FBN function is based on the observation that some of the FBNs contain lipocalin domains, suggesting that FBNs may also play a role in metabolite transport (Singh & McNellis, 2011). Indeed, there is strong evidence that FBN4 (At3g23400) is involved in the partitioning of plastoquinone-9 (2,3-dimethyl-6-solanesyl-1,4-benzoquinone; PQ-9) between the plastoglobules and the rest of the chloroplast (Singh et al., 2012). Reduced PQ-9 levels are also found in Arabidopsis fbn5 knockout mutants, and interaction experiments, together with the seedling-lethal phenotype of fbn5 mutants, suggest that Arabidopsis FBN5 plays a critical role in PQ-9 biosynthesis, is essential for plant development (Kim et al., 2015), and plays a role as a transmitter of singlet oxygen (O) in the chloroplast stroma (Otsubo et al., 2018).
In addition to their proposed roles in structure formation and transport, it has been suggested that FBNs are involved in resistance to biotic stress (Singh et al., 2012) and in protecting PSII against photooxidative stress (Yang et al., 2006;Youssef et al., 2010;Otsubo et al., 2018). Thus, in Arabidopsis, ABA triggers accumulation of FBN1a (At4g04020), and ABA treatment and FBN1a overexpression both protect photosystem II (PSII) against photoinhibition triggered by light stress (Yang et al., 2006). Because no visually discernible growth phenotype was observed in Arabidopsis plants with either down or upregulated FBN1a levels, it was postulated that closely related members of the FBN family might functionally substitute for each other (Youssef et al., 2010). Consequently, RNA interference (RNAi) was employed to simultaneously downregulate FBN1a, FBN1b (At4g22240), and FBN2 (At2g35490) (Youssef et al., 2010). Under a combined high light and cold treatment, these plants display higher PSII photoinhibition, retarded shoot growth, lower anthocyanin accumulation, and an abnormal expression pattern of jasmonate-inducible genes. All these deficiencies can be neutralized by jasmonate treatment of RNAi plants, suggesting that FBN1 and FBN2 proteins modulate jasmonate biosynthesis during exposure to certain abiotic stresses (Youssef et al., 2010).
In this study, FBN6 was identified as a photosynthesis-associated gene by mining coexpression databases. To gain further insight into FBN6 function, an fbn6-1 transfer DNA (T-DNA) insertion mutant and artificial microRNA lines were characterized. Under normal growth conditions, fbn6 plants display delayed growth, have slightly reduced Chl contents, shorter primary roots, and are late flowering under long-day conditions. An FBN6-enhanced green fluorescent protein (eGFP) protein localizes to thylakoid and envelope membrane fractions, and FBN6 is needed to acclimate plants to moderate light stress. RNA sequencing (RNA-Seq) and metabolomics analyses point to accumulation of glutathione (GSH) in fbn6, which in turn confers cadmium (Cd) tolerance of fbn6 seedlings.
Arabadopsis thaliana plants were grown on potting soil (Stender, Schermbeck, Germany) under controlled glasshouse conditions (daylight supplemented with illumination from HQI Powerstar 400W/D, providing a total photon fluence of c. 120 µmol m À2 s À1 on leaf surfaces; 16 h : 8 h, light : dark cycle). Where indicated, seedlings were grown on agar (Sigma-Aldrich) plates containing half-strength Murashige and Skoog (MS) medium, 1.5% (w/v) sucrose and 0.3% (w/v) gelrite (Roth, Karlsruhe, Germany) at 22°C under 100 µmol m À2 s À1 provided by white fluorescent lamps. To investigate the effect of Cd on seedling growth, seeds were sown on MS medium supplemented with 50 or 250 µM cadmium chloride (CdCl 2 ), based on Cd concentrations applied in .

Nucleic acid extraction
For DNA isolation, leaf tissue was homogenized in extraction buffer containing 200 mM Tris hydrochloride (Tris-HCl), pH 7.5, 25 mM NaCl, 25 mM EDTA, and 0.5% (w/v) sodium dodecyl sulfate (SDS). After centrifugation, DNA was precipitated from the supernatant by adding isopropanol. After washing with 70% (v/v) ethanol, the DNA was dissolved in distilled water.
For RNA isolation, frozen tissue was ground in liquid nitrogen (N). Total RNA was extracted with the RNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. RNA quality and concentration and the A 260 /A 280 ratio were assessed by agarose gel electrophoresis and spectrophotometry. Isolated RNA was stored at À80°C prior to use.

Generation of artificial microRNA-mediated knockdown lines
Knockdown fbn6-amiR mutants were generated using an artificial microRNA (amiRNA)-mediated knockdown technique (Schwab et al., 2006). The Web MicroRNA Designer program (http:// wmd3.weigelworld.org/) was applied to generate two amiRNA constructs targeting different regions of FBN6. The amiR constructs were generated by PCR from pRS300 template vector with each primer combination of A + IV, III + II, and I + B mentioned in Table S1. The three products were amplified via PCR with A + B primers. The final products were cloned into pDONR207 via BP reactions (Thermo Fisher Scientific, Waltham, MA, USA) and were subsequently cloned via LR reactions (Thermo Fisher Scientific) into pALLIGATOR3 (Bernaudat et al., 2011). Arabidopsis transformation was conducted by the floral dip method using Agrobacteruim tumefaciens GV3101 strain (Clough & Bent, 1998). The individual T 1 seeds were selected by GFP fluorescence, and independent T 2 or T 3 transgenic lines were used for phenotypic analysis. Knockdown levels of T 2 fbn6-amiR lines were confirmed by reverse transcription (RT)-PCR and real-time PCR with the gene-specific primers listed in Table S1.

Expression and intracellular localization of fluorescence fusions
For overexpression of FBN6 in Col-0, the AT5G19940.1 coding region was amplified from cDNA by PCR (see Table S1 for primer information). The PCR product was cloned with GATEWAY technology into pB7FWG2 to generate a fusion with eGFP under the control of the Cauliflower mosaic virus 35S promoter. The construct was introduced into Col-0 plants by floral dip (Clough & Bent, 1998).
For co-transformation experiments, the VTE1 coding region was amplified from cDNA with primers listed in Table S1. The NcoI-digested product was cloned in frame 5 to the dsRED gene in the vector pGJ1425 (Jach et al., 2001). Intact A. thaliana protoplasts were prepared as described (Dovzhenko et al., 2003). For transient gene expression assays, 5 9 10 5 protoplasts were transfected with 20 lg of plasmid DNA by polyethylene glycol-mediated DNA uptake (Koop et al., 1996) and cultured for 16 h at 22°C in the dark.

Leaf pigment analyses
For Chl extraction, approximately 10 mg of leaf tissue from 4wk-old plants was ground in liquid N in the presence of 80% (v/ v) acetone. After removal of cell debris by centrifugation, absorption was measured with an Ultrospec 3100 pro spectrophotometer (Amersham Biosciences, Freiburg, Germany). Pigment concentrations were calculated following Lichtenthaler (1987).

Chl fluorescence analysis
In vivo Chla fluorescence of whole plants was recorded using an ImagingPAM Chl fluorometer (Heinz Walz GmbH, Effeltrich, Germany). Plants adapted to dark (for 20-30 min) were exposed to a pulsed, blue measuring beam (1 Hz, intensity 4; F 0 ) and a saturating light flash (intensity 4) to obtain F v /F m = (F m À F 0 )/ F m (maximum quantum yield of PSII). Plants were exposed to actinic light (111 lmol m À2 s À1 ) to record the induction curve of the effective quantum yield of PSII (ɸ II ). To quantify nonphotochemical quenching (NPQ) of Chl fluorescence (NPQ; ðF m À F 0 m Þ=F 0 m ), in vivo Chla fluorescence was measured using the ImagingPAM as well as the Dual-PAM 100 (Heinz Walz GmbH).

RNA sequencing and data analysis
Total RNA from plants was isolated using Trizol (Invitrogen) and purified using Direct-zol TM RNA MiniPrep Plus columns (Zymo Research, Irvine, CA, USA) according to the manufacturer's instructions. RNA integrity and quality were assessed by an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA). Ribosomal RNA depletion, generation of RNA-Seq libraries, and 150 bp paired-end sequencing on an Illumina HiSeq 2500 system (Illumina, San Diego, CA, USA) were conducted at Novogene Biotech (Beijing, China) with standard Illumina protocols. Three independent biological replicates were used per genotype.
RNA-Seq reads were analyzed on the Galaxy platform (Afgan et al., 2016) as described in Xu et al. (2019). Sequencing data have been deposited in National Center for Biotechnology Information's Gene Expression Omnibus (Edgar et al., 2002) and are accessible through GEO Series accession no. GSE125515.

Gas exchange measurements
CO 2 exchange measurements were conducted on whole rosettes from 4-wk-old plants using the portable GFS-3000 system (Heinz Walz GmbH). Conditions within the cuvette were set to 22°C, 60% relative humidity, and ambient CO 2 concentrations. The impeller speed was set to 7 and the flow rate to 750 µmol s -1 . Monitoring of the light curve was started with darkened rosettes. When the CO 2 and water system parameters had stabilized, the light level was progressively increased. The CO 2 assimilation rate per unit weight was calculated with the software GFS-WIN v.3.50b (Heinz Walz GmbH).

Transmission electron microscopy
Pieces of primary leaves were fixed with 2.5% (v/v) glutaraldehyde in fixative buffer (75 mM sodium cacodylate, 2 mM magnesium chloride, pH 7.0) and ultrathin sections were prepared as described (Romani et al., 2015).

Metabolite analysis
Fresh material (25 mg) was ground in 180 µl cold (À20°C) methanol containing 5 ll ribitol (0.2 mg ml À1 in water) and 5 ll 13 C-sorbitol (0.2 mg ml À1 in water) as internal standards for the relative quantification. Samples (at least five independent replicates) were processed as described (Roessner et al., 2001;Lisec et al., 2006;Erban et al., 2007). Finally, 1 ll of each sample was injected into a GC-time-of-flight MS system (Pegasus HT; Leco, St Joseph, MO, USA). Sample derivatization and injection were performed by an autosampler system (Combi PAL; CTC Analytics AG, Zwingen, Switzerland). Helium was used as carrier gas at a constant flow rate of 1 ml min À1 . Gas chromatography was conducted on an Agilent GC (7890A; Agilent) using a 30 m VF-5ms column with 10 m EZ-Guard column. The injection temperature of the split/splitless injector was set to 250°C, as well as the transfer line and the ion source. The initial oven temperature (70°C) was continuously increased to a final temperature of 350°C by a ramp of 9°C min À1 . Solvent delay was set to 340 s. Metabolites were ionized and fractionated by an ion pulse of 70 eV, and mass spectra were recorded at a rate of 20 s À1 within a scan range of 35-800 m/z. Chromatograms and mass spectra were evaluated using CHROMATOF 4.5 and TAGFINDER 4.1 software (Luedemann et al., 2008). Nonpolar compounds were extracted as described (Espinoza-Corral et al. 2019).

Reactive oxygen species staining of rosette leaves
Reactive O species (ROS) staining was done essentially as described before (Lundquist et al., 2013).

Quantification of glutathione
Extraction and determination of total and oxidized GSH content was essentially done as described (Queval & Noctor, 2007). Standard curves of GSH and oxidized GSH disulfide (GSSG) were prepared, and 25 mg of plant material was used.

Data analysis
The significance of differences in gene expression in real-time PCR, root length, flowering time, and GSH content was evaluated by Duncan's multiple range test, Holm-Sidak test, or by Student's t-test, as indicated in the figure legends.

Results
FBN6 is co-regulated with photosynthesis genes One means of identifying genes coding for proteins in particular pathways is the mining of coexpression data over a range of tissues and conditions (Usadel et al., 2009). Here, we used photosynthesis genes as query genes in the ATTED-II COEXSEARCH tool (Aoki et al., 2016) in an attempt to identify previously unrecognized photosynthesis-associated genes. We found that FBN6 is coexpressed preferentially with genes coding for components of the photosynthetic machinery, like PSI reaction-center proteins, and components of the O-evolving and light-harvesting complexes (e.g. PsbY, PsbO1 and O2, Lhcb2.4, and Lhcb6) ( Fig. 1a; Table S2). Furthermore, we used the CORrelation NETworks tool CORNET2.0 (De Bodt et al., 2012), with which we constructed a condition-independent coexpression network (i.e. one based on the screening of all deposited microarray data, with no bias towards any particular type of condition/treatment) for 10 of the 14 FBNs (FBN1a, FBN1b and FBN3a and FBN3a were disregarded by CORNET2.0 because of their sequence similarity) and the FBN-like gene At1G18060 (Lundquist et al., 2012) (Fig. 1b). The FBN11 gene is an outlier in this network and is only coexpressed with one gene of unknown function. The genes FBN5 and FBN8 are predominantly coexpressed with genes for chloroplast protein homeostasis and gene expression, respectively. Other FBN genes are coexpressed with genes for chloroplast proteins that are involved in several processes. The FBN6 gene is the only FBN that is coexpressed chiefly with genes for photosynthesis proteins (Fig. 1b).
To assess to what extent FBN messenger RNA (mRNA) levels are regulated in response to various experimental perturbations, the GENEVESTIGATOR Perturbations Tool was applied. FBN11 expression is most susceptible to perturbations: 285 conditions provoke at least a two-fold change in transcript level (Fig. 2).

PGR5-LIKE A AT1G65230
Nitrogen compound metabolism (4/10)  Table S2. (b) Coregulation gene network derived from condition-independent coexpression analysis using the CORrelation NETworks tool (CORNET 2.0; https://bioinformatics.psb.ugent.be/cornet; De Bodt et al., 2012). Only strong coregulations of the top 10 genes with a Pearson correlation coefficient (corr. coeff.) ≥0.8 are shown. The relative size of a sector on the pie chart represents how much a protein is predicted to be localized to the compartment indicated. FBN1a, FBN1b, FBN3a, and FBN3b were ignored by CORNET 2.0, as no reliable probe sets were identified with this program. mRNA, messenger RNA; PS, photosystem; (x/y), x out of y genes are encoding proteins related to the process category indicated. conditions. This expression profile is shared with numerous photosynthesis-associated genes.
Identification and phenotypic analysis of the fbn6-1 mutant and fbn6 artificial microRNA lines To learn more about the physiological functions of FBN6, lossof-function fbn6-1, fbn6-2, and fbn6-3 mutants were identified. In all mutants, the T-DNA insertion (fbn6-1) and the Dissociation (Ds) transposon (fbn6-2, GT_5_46738, and fbn6-3, GT_5_46794) were supposed to be located in the first exon of FBN6. However, we could only isolate homozygous fbn6-1 mutants (Fig. 3a), and the Ds transposon insertions could not be confirmed. In fbn6-1, FBN6 transcripts comprising sequences 5 of the insertion point were reduced to 1% of Col-0 levels and none were detectable 3 of the insertion (Fig. S1a). Growth of fbn6-1 plants was retarded, as reflected in smaller rosettes and a reduction in rosette fresh weight to less than half that of Col-0 in 26-d-old plants (Fig. 3b). Moreover, fbn6-1 plants were late flowering under long-day photoperiods (Fig. 3b), though not under short-day conditions (8 h : 16 h, light : dark cycles; Fig. S1b). Furthermore, leaves of fbn6-1 appeared slightly paler than those of the wild-type (WT). Indeed, their Chl content was slightly but significantly reduced in fbn6-1 plants, whereas the Chla/Chlb ratio was increased (Fig. 3d). Segregation analysis of the F 2 offspring of fbn6-1 back-crossed to Col-0 indicated that only one T-DNA inserted in fbn6-1 was causing the fbn6-1-specific phenotype (Table S3). To ultimately confirm that the altered activity of AT5G19940 was responsible for the mutant phenotype of fbn6-1, independent knockdown mutants of FBN6 were generated by amiRNA (Schwab et al., 2006) targeting the second (amiR1) and third exon of AT5G19940 (amiR2) (Fig. S2a). Knockdown of FBN6 was confirmed by RT-PCR ( Fig. S2b) and real-time PCR analysis (Fig. S2c). Subsequent phenotypic analysis showed that amiR1 and amiR2 behaved like fbn6 plants (Figs 3b,c,S2d,e). This indicates that the fbn6 phenotype is indeed caused by knockout or knockdown of the AT5G19940 gene.

Photosynthetic properties of fbn6-1 mutant plants
To examine the effect of FBN6 disruption on photosynthesis, Chla fluorescence parameters were investigated. Under normal growth conditions, basic photosynthetic parameters, such as maximum (F v /F m ) and effective (Φ II ) quantum yields of PSII, were slightly elevated in all developmental stages relative to Col-0 investigated (Fig. 4a,b), indicating that PSII and the electron transport chain are fully functional in fbn6-1 and amiR1 and amiR2 plants. Also, a minor increase in NPQ was observed in fbn6-1 mutants, both at the level of induction kinetics (111 µmol m À2 s À1 ) and in response to increases in light intensity ( Fig. S3a,b). This indicates that a slightly larger fraction of the absorbed light energy is dissipated as heat, but it does not explain the growth phenotype of plants lacking FBN6.
To determine the abundances of photosynthetic proteins in fbn6-1, immunoblot analysis was performed on total protein extracts from Col-0 and fbn6-1 leaves. Representative subunits of PSI and PSII accumulated to similar levels in Col-0 and the fbn6-1 mutant (Fig. 4c)including PsaD and PsbO, whose genes are also coexpressed with FBN6 (see Fig. 1a). Only levels of RbcL, the large subunit of Rubisco, were reduced in fbn6-1 to approximately 75% of WT amounts (Fig. 4c). We therefore measured CO 2 assimilation rates in Col-0 and fbn6-1 plants exposed to different light intensities up to 1000 µmol m À2 s À1 . However, no differences in CO 2 assimilation rates could be detected between Col-0 and fbn6-1 rosettes (Fig. 4d).
Hence, loss of FBN6 has only a marginal effect on basic photosynthetic parameters, and CO 2 assimilation capacity is unaltered.
FBN6 is localized to thylakoid and envelope membranes FBN6, together with FBN3a, FBN3b, FBN-like, and most probably FBN9, was found to be thylakoid localized (Lundquist et al., 2012). Moreover, FBN6 is the only FBN isoform that was found to be enriched in purified envelope fractions (Ferro et al., 2010), a localization that was recently supported by Bouchnak et al. (2019). To experimentally test for the subcellular location of FBN6, the fluorescence distribution in Col-0 protoplasts overexpressing the FBN6-eGFP fusion was investigated. The eGFP fluorescence signals colocalized with the Chl autofluorescence ( Fig. 5a), confirming the localization of the FBN6 fusion protein to chloroplasts. However, the fluorescence signals were detected only occasionally as evenly distributed signals, and more often or exclusively as small dots suggestive of plastoglobules (Fig. 5a). To define the nature of the dots, and thereby confirm its localization to plastoglobules, FBN6-eGFP was cotransformed into A. thaliana protoplasts with a VTE1-dsRED construct. The localization of VTE1 (VITAMIN E DEFICIENT 1; a tocopherol cyclase) to plastoglobules was previously established in studies involving mass spectrometry and yellow fluorescent protein fusions (Vidi et al., 2006;Lundquist et al., 2012). Indeed, FBN6-eGFP signals were found in dots together with the dsRED signal of VTE1. Merging of both signals confirmed colocalization of FBN6 and VTE1, and thus suggested a localization of FBN6
To further explore the issue of FBN6's localization, chloroplasts from Col-0 plants overexpressing FBN6-eGFP were prepared and thylakoid, stroma, and envelope-enriched fractions were isolated (Fig. 5b). These fractions were subjected to immunoblot analysis, and the enrichment of chloroplast subcompartments was tested by monitoring thylakoid light-harvesting Chla/b-binding protein (LHCP), stromal RbcL, plastoglobular protein 18 (PG18; Espinoza-Corral et al., 2019), and the translocon at the inner and outer envelope membrane proteins Tic110 and Toc64, respectively. RbcL was detected mainly in the stromal fraction, Tic110 and Toc64 mainly in the envelope fraction, PG18 nearly exclusively in the thylakoid fraction, and LHCP was most prominent in the thylakoids, but was also detected to a weaker extent in the envelope fraction (Fig. 5b), showing a contamination of the envelope fraction with thylakoids, but overall corroborating the enrichment of the respective fractions. Using this approach, FBN6-eGFP was mainly detected in the thylakoid and, to a lesser extent, envelope membrane fractions. Comparing relative intensities of eGFP signals with those of LHCP, it was concluded that FBN6 was localized to thylakoids and the envelope, albeit to a weaker extent to the latter subcompartment. In a second fractionation approach, enrichment of selected fractions from a sucrose gradient used to fractionate into thylakoids and plastoglobules was tested by monitoring LHCP and PG18. LHCP was detected mainly in fractions 25-33 (Fig. 5c) and PG18 mainly in fraction 1. However, FBN6 was only detected in fractions co-migrating with LHCP, suggesting that FBN6 is not localized to plastoglobules, corroborating the previous finding of Lundquist et al. (2012).
Because FBN6-eGFP was localized to thylakoid and envelope membranes, we asked whether the growth phenotype of fbn6-1 plants might be related to alterations in chloroplast ultrastructure. However, transmission electron microscopy (TEM) of Pigments were acetone-extracted, measured spectrophotometrically, and concentrations were determined as described (Lichtenthaler, 1987). The data are shown as mean values AE SD from four different plant pools. Significant differences between the data pairs were identified by the Student's t-test, and significant differences (P < 0.05) with respect to Col-0 are denoted by the asterisks. ultrathin sections of primary leaves showed that the ultrastructures of chloroplasts and also that of plastoglobules were virtually indistinguishable between Col-0 and fbn6-1 (Fig. S4).
In summary, FBN6-eGFP expression leads to localization of the fusion protein to thylakoid and envelope membranes, but chloroplast ultrastructure in fbn6-1 mutants is indistinguishable from WT under normal growth conditions.
FBN6 is needed to acclimate plants to moderate light stress GENEVESTIGATOR analysis indicated that FBN6 mRNA levels are downregulated after 1200 µmol m À2 s À1 (HL) treatment. To confirm this, 29-d-old Col-0 plants grown under standard lighting conditions (120 µmol m À2 s À1 ; GL) were exposed for 2 h to the 10-fold higher light level (HL) and real-time PCR analysis showed that FBN6 transcripts were indeed downregulated after HL treatment (Fig. S5a). Apparently, FBN6 transcripts must be downregulated to accommodate the plant to the unfavorable HL condition, and we therefore asked whether a complete lack of FBN6 might help the plants to cope even better with HL stress. However, F v /F m values of fbn6-1 plants exposed to 2 h of HL stress were only slightly lower than in Col-0 plants (Fig. S5b).

Research
New Phytologist display darker stains (Fig. S5c). This might suggest that plants lacking FBN6 cannot properly adjust their ROS levels to HL conditions. However, under prolonged light stress, both genotypes responded in a similar manner (Fig. S5d).
Plants at 26 d old were also exposed to more moderate light stress (400 µmol m À2 s À1 ; ML) combined with mild heat stress of 28°C for several days. After 6 d of ML, the leaf petioles of Col-0, fbn6-1, amiR1, and amiR2 plants were shortened, leaf area expanded, and the maximum quantum yield of PSII was mildly reduced, but otherwise Col-0 plants still looked healthy (Fig. 6a). By contrast, the primary leaves of plants lacking FBN6 displayed small necrotic lesions and the reduction in F v /F m was more pronounced (Fig. 6a). Furthermore, older leaves of fbn6 mutant lines almost completely failed to induce anthocyanin accumulation: anthocyanin content was only 10% of WT levels (Fig. 6a).
To determine the impact of FBN6 deficiency on the ultrastructure of chloroplasts after 6 d of ML, TEM of ultrathin sections of Col-0 and fbn6-1 primary leaves was conducted (Fig. 6b). Col-0 leaves showed well-developed lenticular chloroplasts with starch accumulating. Occasionally, grana thylakoids were somewhat distorted, indicating that Col-0 chloroplasts were under mild stress, but plastoglobules were present in normal numbers and size. The fbn6-1 chloroplasts tended to be more nearly spherical in shape and contained no starch, and the diameter of grana stacks was smaller than in the WT. Obviously, the number of plastoglobules in fbn6-1 chloroplasts increased, the   (Chl) were isolated from a 4-wk-old CaMV35S:: FBN6-eGFP (Col-0) plants and fractionated into thylakoids (Thy), stroma (Str), and envelope (Env). These fractions were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis, transferred to a polyvinylidene difluoride membrane, and exposed to antibodies raised against GFP (to detect the FBN6-eGFP fusion protein), Tic110 and Toc64 (as controls for the envelope fraction), PG18 (a plastoglobule-localized protein), RbcL (as a control for stromal proteins), or lightharvesting Chla/b-binding protein (LHCP; as a control for the thylakoid fraction). Quantification of signals relative to the whole chloroplast fraction (=1.00) is provided below each panel and serves as an enrichment factor. The Ponceau Red (Ponc.)stained blot served as the loading control. (c) Chloroplast membranes were isolated from 5-wk-old CaMV35S::FBN6-eGFP (Col-0) plants. The samples were fractionated on a discontinuous sucrose gradient, and fractions containing the same volume were taken from the top (number 1) to the bottom of the gradient. Selected fractions were treated as described in (b)  plastoglobules clustered, were larger, and variable in size. Along with the larger size, the core of the plastoglobules stained lighter, whereas the periphery was electron dense (Fig. 6b), indicating that fbn6-1 chloroplasts in primary leaves already experience stress under moderate higher light conditions.
Identification of other mutants with an fbn6-1-like phenotype To identify mutants with an fbn6-1-like phenotype, we filtered a comprehensive dataset of genes with a loss-of-function mutant phenotype in A. thaliana published in 2012 (Lloyd & Meinke, 2012) and identified 16 mutants (Table S4). The majority (12) of the mutants are impaired in genes for chloroplast-localized proteins. One of the mutants harbors a T-DNA insertion in LHCB6, which is coexpressed with FBN6 (see Fig. 1b). Nearly half of the tagged genes encode proteins involved in ROS production/homeostasis, like the chloroplast ferredoxin AtFD2 and the chloroplast NADPH-dependent thioredoxin reductase NTRC, and the cytosolic ascorbate peroxidase APX1 which scavenges H 2 O 2 (Dietz et al., 2016). Indeed, genes encoding the following ROSassociated proteins are among the top 30 genes coexpressed with FBN6: thioredoxin F2, the GSH peroxidase GPX1, and the (cation exchanger) CAT1-interacting protein CXIP1 (AtGRXcp), which is a chloroplast glutaredoxin and is critical for protection against protein oxidative damage (Cheng et al., 2006). Furthermore, mutants defective in tocopherol (vitamin E) biosynthesis, specifically vte2 and vte4 (Bergmuller et al., 2003;Sattler et al., 2004), have an fbn6-like phenotype. Here, it is interesting to note that VTE5 is also coexpressed with FBN6 (see Fig. 1). Taken together, these data suggest that changes in ROS homeostasis might be responsible for impaired light acclimation and the delayed growth phenotype of fbn6-1.

Transcriptome changes point to an alteration of sulfate reduction in fbn6 plants
To obtain molecular insights into a potential function of FBN6 in ROS homeostasis and to follow the general RNA expression pattern of nucleus-and organelle-encoded genes, RNAs isolated from 3-wk-old WT and fbn6-1 plants were subjected to RNA-Seq. The mRNA levels of 194 and 118 genes were significantly (more than 1.5-fold) reduced or elevated, respectively (Table S5). Expression changes of several transcripts were confirmed by realtime PCR analysis of 3-wk-old Col-0, fbn6-1 and fbn6-amiR1 and amiR2 plants (Fig. 7a). Plastid and mitochondria-encoded transcripts were not changed. Gene Ontology (GO) analysis with DAVID (da Huang et al., 2009) identified among the nuclear-encoded upregulated genes in fbn6-1 no enriched GO terms in the 'cellular component' and 'molecular function' categories. But enrichment of genes encoding proteins for the 'biological process' GOs 'removal of superoxide radicals', 'chaperone-mediated protein folding', and those associated with response to stresses like high light, chitin, heat, cold, and salt stress were identified (Fig. S6a). This corroborates the aforementioned assumption that ROS homeostasis is perturbed in fbn6 plants. For the downregulated genes in fbn6-1, DAVID detected as highest fold enriched GOs in the three categories 'cellular component', 'biological process', and 'molecular function' as being 'chloroplast thylakoid', 'sulfate reduction', and 'adenylyl-sulfate reductase activity', respectively (Fig. S6b). In particular, the 80 and 120-fold enrichment of 'sulfate reduction' and 'adenylyl-sulfate reductase activity' was remarkable. Eighteen transcripts that were in several transcriptomics studies upregulated by sulfate deficiency were identified (Kopriva et al., 2015), and11 of these were not up but downregulated in the fbn6-1 mutant (Fig. 7b). This might indicate that the sulfate assimilation status in fbn6 is not deficiency but an excessive supply. Fig. 7(c) shows the transcriptome changes in the sulfate assimilation pathway in more detail. Sulfate assimilation is partitioned between the cytosol and chloroplast (Takahashi et al., 2011;Bohrer et al., 2014); and in fbn6-1, in particular, the transcripts for the chloroplast-localized ATP sulfurylase, adenosine 5 0 -phosphosulfate kinase, adenosine 5 0 -phosphosulfate reductase (APR), and sulfite reductase were downregulated. Those enzymes act in the sulfate assimilation pathway (Fig. 7c).
Taken together, these transcriptome changes point to an alteration of sulfate reduction when FBN6 is lacking.

Lack of FBN6 leads to higher glutathione accumulation and confers cadmium tolerance
To examine the metabolic consequences of a lack of fbn6, GC-MS analysis was conducted on 3-wk-old Col-0 and fbn6-1 mutant plants and 180 metabolites were identified (Table S6). Among the 36 metabolites that were significantly changed (> 1.5fold change; P < 0.05) in fbn6-1 compared with WT, glycine was 1.9-fold increased. Glycine is incorporated into c-glutamylcysteine to yield GSH (Fig. 7c). Because suppressed APR mRNA expression has been found previously to be caused by increased GSH levels (Koprivova & Kopriva, 2014;Fu et al., 2018), and indeed all three chloroplast-localized APRs were downregulated in fbn6-1 ( Fig. 7c; Table S5), we speculate that the increased glycine levels in fbn6 might result in increased GSH levels. To test this, GSH content of 3-wk-old Col-0, fbn6-1, and fbn6-amiR1 and amiR2 plants was measured with a plate reader assay (Queval & Noctor, 2007). Indeed, all fbn6 lines contained more total GSH (which is the sum of the reduced form of GSH and the oxidized GSSG) (Fig. 8a).
GSH is a key player in antioxidant mechanisms (Noctor et al., 2018) and plays, for example, a pivotal role in Cd detoxification and tolerance (Liu et al., 2016). We reasoned that the higher GSH content in fbn6 might lead to enhanced Cd tolerance. To test this, Col-0, fbn6-1, and fbn6-amiR1 and amiR2 mutant seeds were germinated on control MS medium and on MS medium containing 50 or 250 µM CdCl 2 , and phenotypes were scored after 14 d (Fig. 8b,c). When grown on MS plates, the primary root length of seedlings lacking FBN6 was reduced relative to Col-0 (Fig. 8c). By contrast, on MS supplemented with 50 µM CdCl 2 , root lengths of fbn6-1 and fbn6-amiR lines were approximately two-fold longer than for Col-0 (Fig. 8c), and growth of fbn6 seedlings was significantly better than that of WT seedlings on plates supplemented with 250 µM CdCl 2 . Thus, the mutants

Research
New Phytologist were more resistant to Cd stress. Of note is that the increased GSH content of fbn6 lines became even more prominent under 250 µM CdCl 2 stress than under control conditions (Fig. 8a,d).
In sum, these data show that a lack of FBN6 results in increased GSH levels and increased tolerance to Cd.

Using coexpression information to identify new pathway components
Approaches to the identification of novel associated genes with pathways of interest include guilt-by-association approaches (Higashi & Saito, 2013). This approach is based on the assumption that genes which are coexpressed (i.e. show similar expression patterns over a range of different tissues and conditions) are likely to encode proteins that act in the same or closely related biological pathway(s) (Higashi & Saito, 2013;Yonekura-Sakakibara & Saito, 2013). For example, transcriptome and coexpression network analyses were exploited to identify HEAT INDUCIBLE LIPASE1 (HIL1) as a remodeler of chloroplastic monogalactosyldiacylglycerol under heat stress, and to show that HIL1 homologue expression levels in various plants are tightly associated with chloroplastic heat stress responses (Higashi et al., 2018).
With the ongoing expansion of public gene expression repositories, the predictive power and usefulness of coexpression information increases (Rung & Brazma, 2013), and many more studies have since made use of the abundance of microarray data  Fig. 6 Behavior of Arabidopsis thaliana wildtype (Col-0) and fibrillin6-1 (fbn6-1) and two artificial microRNA (amiR1 and amiR2) mutant plants under moderate light stress.
(a) Plants were first grown for 26 d under normal growth light (120 µmol m À2 s À1 ) and then exposed to 400 µmol m À2 s À1 for the time periods indicated. Primary leaves of plants lacking FBN6 displayed small necrotic lesions (yellow arrow pointing to the enlargement of primary leaves displayed in square pictures) and maximum quantum yield of photosystem II (F v /F m ) was more pronounced. The data are shown as mean values AE SD of 8-10 different leaves. Significant differences between the data pairs were identified by the Student's t-test, and the asterisks denote significant differences (P < 0.05) with respect to Col-0.  . 1a). Thus, it is coexpressed with various genes coding for the core photosynthetic apparatus and regulators of photosynthesis, such as CURT1B (TMP14; see Table S2) or the O subunit of PSI that helps to balance excitation pressure between the two photosystems (Jensen et al., 2004). The case of psao is an example of a mutant that fails to display a dramatic phenotype under normal-growth light conditions, although it is impaired in a photosynthesis-associated protein.
Similarly, absence of Lhcb1 and Lhcb2 in antisense plants induces no significant growth phenotype, although Chl levels are Venn diagram depicting the degree of overlap between the sets of genes whose expression levels were altered by at least two-fold (up or down) in the fbn6-1 mutant compared with a robust set of sulfur-deficiency-induced genes extracted from Kopriva et al. (2015). (c) Sulfate assimilation leading to glutathione synthesis and sulfometabolites. Sulfate assimilation is partitioned between the cytosol and plastids (Takahashi et al., 2011;Bohrer et al., 2014). Sulfate is activated by reaction with ATP with the help of ATP sulfurylase (ATPS) to form adenosine 5 0 -phosphosulfate (APS), which is subsequently phosphorylated to 3 0 -phosphoadenosine 5 0 -phosphosulfate (PAPS) by APS kinase (APK) or reduced to sulfide by APS reductase (APR). The sulfide branch leads to glutathione synthesis in chloroplasts (Noctor et al., 2012). Red arrows, direction of transcript changes encoding the respective enzymes; red cross, inhibition; green triangle, activation.  (Andersson et al., 2003). The fbn6-1 mutant does show a phenotype under normal light conditions: it is stunted, contains slightly less Chl than Col-0, and is late flowering (see Figs 3, S1). But the fbn6-1 growth phenotype cannot be accounted for either by altered basic photosynthesis parameters or by a reduced CO 2 assimilation rate (see Fig. 4).

FBN6 affects glutathione levels and cadmium tolerance
A large fraction of the mutants with an fbn6-1-like phenotype (see Table S3) are defective in ROS production/homeostasis. The chloroplast is considered to be the major source of ROS in plant cells. As a chloroplast-localized protein, FBN6 might be part of the ROS scavenging machinery. This would be compatible with our data, which point to an altered ROS homeostasis under HL conditions and altered transcript levels of genes for proteins involved in the removal of superoxide radicals in fbn6-1 plants even under standard growth conditions (see Figs S5, S6). This suggests that an ROS-dependent signaling pathway is activated in the fbn6-1 mutant under normally adequate growth conditions. Furthermore, anthocyanin levels in older fbn6-1 leaves are reduced to 10% of WT levels under moderate light stress, whereas younger leaf tissue of fbn6-1 mutants appears to accumulate normal levels of anthocyanins. This phenotype is reminiscent of that of the k1 k3 mutant, which is devoid of the plastoglobule kinases ABC1K1 and ABC1K3 (Lundquist et al., 2013), and of the behavior of the npq1 mutant (which lacks the xanthophyll cycle enzyme violaxanthin deepoxidase; (Havaux et al., 2000) and a mutant perturbed in two ascorbate peroxidases (Giacomelli et al., 2007).
The perturbation in ROS homeostasis and the higher GSH content might explain the delayed growth phenotype of fbn6 plants, as in the case of the apx1 mutant, in which GSH is also increased (Jiang et al., 2016). Indeed, the late-flowering (Pnueli et al., 2003) and short-root phenotypes of apx1 (Correa-Aragunde et al., 2013) are also recapitulated by fbn6-1 (see Figs 3,8). Notably, flowering time is delayed by exogenous and endogenous GSH supply (Cheng et al., 2015). GSH is used in several stress response pathways to detoxify ROS, xenobiotics, and certain heavy metals (Noctor & Foyer, 1998). Thus, it is most likely that the increased GSH content provokes Cd tolerance of fbn6 plants (see Fig. 8). Accordingly, plants lacking APX1 are more tolerant to selenium (Jiang et al., 2016), and the sultr1;1 sultr1;2 mutant with lower GSH content is more sensitive to Cd (Liu et al., 2016).  In general, oxidative stressinduced by treatment with paraquat, for instanceinhibits primary root elongation (Suzuki et al., 2013). Among the other fbn6-like mutants identified in the filtered dataset referred to earlier (Lloyd & Meinke, 2012) are the mutants vte2 (Sattler et al., 2004) andvte4 (Bergmuller et al., 2003). Notably, root elongation is severely reduced in vte2 plants and mildly compromised in the vte1 mutant (Sattler et al., 2004). Furthermore, VTE5 is coexpressed with FBN6 (see Fig. 1). Tocopherols are lipophilic antioxidants that are synthesized in all photosynthetic organisms; in plants, they are synthesized in plastids (Hussain et al., 2013). To mitigate the harmful effects of elevated ROS, plants possess a complex network of enzymatic and nonenzymatic antioxidant defense systems, of which the latter involves low-molecular-weight antioxidants such as tocopherol, ascorbate, and GSH (Hussain et al., 2013). When vte1 or vte2 leaf discs are simultaneously exposed to HL and low-temperature stress, they bleach and suffer from lipid photodestruction. Interestingly, this is not observed in whole plants exposed to long-term high light stress, unless the stress conditions are extreme (very low temperature and very HL), suggesting the availability of compensatory mechanisms for vitamin E deficiency under more physiological conditions (Havaux et al., 2005). The fbn6-1 mutant phenotype behaves very similarly to the WT under HL stress (see Fig. S5). It is conceivable that disruption of another part of the ROS network in combination with a lack of FBN6 might lead to a stronger phenotype, as has already been observed in other stress studies (Kanwischer et al., 2005;Giacomelli et al., 2007). For example, in vte1, ascorbate and GSH levels are increased. Whereas growth, Chl content, and photosynthetic quantum yield were very similar to WT in vte1, vtc1 (ascorbate deficient), cad2 (GSH deficient) and vte1vtc1 mutants, they were clearly reduced in vte1cad2 mutants, indicating that the simultaneous loss of tocopherol and GSH results in moderate oxidative stress, which in turn affects the stability and efficiency of the photosynthetic apparatus (Kanwischer et al., 2005).
Under a fluence of 400 µmol m À2 s À1 , although it has to be noted that FBN6 is not localized to plastoglobules (see Fig. 5), the plastoglobules in the fbn6-1 mutant become larger than in the WT, which is a further indication that FBN6 is needed to acclimate Arabidopsis to moderate light stress. Apple trees or Arabidopsis with reduced mounts of FBN4 are also susceptible to abiotic and biotic stresses (Singh et al., 2010). In leaves of apple fbn4 knockdown plants, the partitioning of PQ-9 between plastoglobules and the rest of the chloroplast seems to be disrupted; therefore, a failure to accumulate this antioxidant in plastoglobules might contribute to the increased stress sensitivity of fbn4 knockdown trees (Singh et al., 2012). Furthermore, plants in which mRNAs for FBN1a, FBN1b, and FBN2 were simultaneously downregulated by RNAi display higher PSII photoinhibition, retarded shoot growth, and lower anthocyanin accumulation under a combined HL and cold treatment (Youssef et al., 2010), and it was recently shown that FBN5 is involved in the acclimation to photooxidative stress (Otsubo et al., 2018).
Another example for the involvement of FBNs in plant growth regulation even under relatively nonstressful conditions is the FBN1 homologue C40.4 in potato. Under standard growth conditions, potato plants with reduced expression of C40.4 have reduced tuber size and yield, and their growth is stunted (Monte et al., 1999). On the other hand, overexpression of bell pepper FBN1 in tobacco results in a greater plant height and accelerated flowering under higher light intensities (300 µmol m -2 s -1 ), but not under lower light intensities (100 µmol m -2 s -1 ) conditions (Rey et al., 2000).
Thus, overall, a picture is emerging in which FBNs have important roles in plant responses to stresses.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.          (2012).

Table S5
Genes whose transcript levels were significantly regulated in 3-wk-old Arabidopsis thaliana fbn6-1 seedlings compared to Col-0. New Phytologist is an electronic (online-only) journal owned by the New Phytologist Trust, a not-for-profit organization dedicated to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews and Tansley insights.
Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged. We are committed to rapid processing, from online submission through to publication 'as ready' via Early View -our average time to decision is <26 days. There are no page or colour charges and a PDF version will be provided for each article.