Arabidopsis FIP and FtsH sequences were obtained from TAIR¹. Transmembrane domains were predicted by TMPred².

Data from the AtGenExpress Project³ (Kilian et al., 2007) is part of the TAIR database (The Arabidopsis Information Resource, see foot note text 1) was analyzed with the package R/Bioconductor (Gentleman et al., 2004) and normalized by robust multi-array average [RMA; (Irizarry et al., 2003)].

FIP homologous and FtsH proteases type A sequences were identified through blastn algorithm, by means of reciprocal blast. The sequences were compared to the non-redundant (nr/nt) GenBank database⁴ and the Phytozome database⁵ and selected based on an adequate e-value threshold.

The AH109 strain of Saccharomyces cerevisiae (MATa trp1-901 leu2-3, 112 ura3-52 his3-200 gal4 gal80 LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7-lacZ) was used in all experiments. For the screening using an Arabidopsis library, the yeast strain containing the full-length FtsH5 as bait was employed. The LiAc/SS carrier DNA/PEG transformation method was used (Gietz and Schiestl, 2007). Plates were maintained for 14 days at 30°C. Positive colonies were rescreened on SC-leu-trp-his and SC-leu-trp-ade medium plates and incubated at 30°C for 48 h. Clone growth in both plates was considered positive and subjected to DNA sequencing.

An isolated colony of transformed E. coli strain BL21 was inoculated into LB medium containing the appropriate antibiotic and grown at 37°C and 200 rpm for 18 h. The preculture was inoculated into LB medium containing the appropriate antibiotics at a ratio of 1:100 of the total culture volume and incubated with agitation at 37°C until the optical density reached 0.6. Protein expression was induced with IPTG at a final concentration of 1 mM. The culture was incubated at 30°C for 4 h. The cells were centrifuged, and the pellet was frozen at -70°C.

FtsH5 was purified using a nickel resin Ni-NTA Spin Columns kit from QIAGEN according to the manufacturer’s instructions. For purification of the glutathione S-transferase (GST) GST-FIP fusion, Glutathione Superflow resin from QIAGEN was used according to the manufacturer’s instructions.

To each tube was added 50 ml of agarose beads and 1 μg purified FtsH5. To the first tube only was added incubation buffer (50 mM Tris–HCl, 200 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 1 mM DTT, 10 mM MgCl 2, pH 8.0). To the second tube was added 25 mg of the purified GST protein and incubation buffer. To the third tube was added 25 mg of the GST-FIP fusion and incubation buffer. All volumes were filled to 500 ml, followed by incubation for 4 h on ice with gentle agitation. The beads were then washed 5 times with 1 ml of incubation buffer and resuspended in 50 μl of SDS–PAGE sample buffer.

After SDS–PAGE, the samples were transferred to nitrocellulose membrane (Bio-Rad 0.45 μm). The membrane was then incubated in TBS-T buffer with 0.2% BSA and an adequate amount of primary antibody for 2 h. The membrane was then washed three times for 15 min with TBS-T buffer and used for incubation with an adequate amount of secondary antibody in TBS-T buffer with 0.2% BSA for 1 h. The membrane was then washed three times for 15 min with TBS-T buffer and used for film exposure after addition of 1 ml of substrate (Bio-Rad) for alkaline phosphatase.

Leaves of N. tabacum cv. SR1 were used for Agrobacterium-mediated transient transformation. A single Agrobacterium colony was incubated in five ml of LB medium containing appropriated antibiotics, 50 μM acetosyringone, and 10 mM MES (pH 5.6). The culture was incubated at 28°C for 16 h and 1.5 ml was centrifuged at 13,000 rpm for 1 min. One ml of 10 mM MgCl₂ was used to resuspend the pellet and the OD600 of the culture was adjusted to 0.2 and 100 μM of acetosyringone. N. tabacum leaves were infiltrated using a syringe without a needle. Mesophyll protoplasts were prepared as described by Carneiro et al. (1993). After 4 and 5 days of infiltration, the microscope analysis was conducted using an Olympus FV1000 confocal microscope. Excitation filters were, respectively, for GFP and chlorophyll autofluorescence: 488 and 635 nm; Emission filters were, respectively, for GFP and chlorophyll autofluorescence: 510–550 nm and 670–700 nm. Images were obtained with the following software: Olympus FluoView FV10-ASW.

SP6 polymerase transcription kit (Promega) was used to produce radiolabeled precursors. The RNA was translated with a wheat germ kit (homemade) with ³[H]leucine (Cline, 1986). 60 mM leucine in 2× IB (IB, import buffer; 1× = 50 mM HEPES/KOH, pH 8.0, 0.33 M sorbitol) was used to dilute the products of translation, prior to use.

For chloroplasts isolation, nine to ten-day-old Pisum sativum cv. Laxton’s Progress 9 seedlings were used according to Cline (1986) and chlorophyll was determined as described (Arnon, 1949). Intact chloroplasts were used to produce lysates, which was the base to produce thylakoids and stroma (Cline et al., 1993). Radiolabeled precursors importation into intact chloroplasts, chloroplast lysates, or thylakoids (0.33 mg chlorophyll/ml or equivalent) was conducted at 25°C, 5 mM MgATP, and 70–100 μE m^-2sec^-1 white light (Cline et al., 1993). Lysis of recovered chloroplasts occurred by adding 20 mM HEPES/KOH, pH 8.0 on ice for 5 min. Centrifugation for 8 min at 3200 g was employed to separate thylakoids from stroma. IB was used to wash and 100 mM NaOH to extract the thylakoid membrane. Thermolysin treatment occurred according to Summer et al. (2000), using 1 μg thermolysin per μg chlorophyll for 40 min at 4°C and the thermolysin was inactivated with IB with 14 mM EDTA. SDS–PAGE and fluorography was used to analyze the samples.

Multiple sequence alignments (MSAs) were generated with the TranslatorX server⁶ (Abascal et al., 2010) using MAFFT v.7 software (Katoh and Standley, 2013) and curated in Jalview v.2.8.0 or Gblocks [standard parameters; (Castresana, 2000; Talavera and Castresana, 2007)] for large amount of data. MSAs were visualized in Jalview v.2.8.0 (Waterhouse et al., 2009) and trimmed for short, redundant, largely incomplete, and poorly aligned sequences.

Bayesian phylogenetic inference was conducted using MrBayes v.3.2.1 (Ronquist and Huelsenbeck, 2003) for the tree topology and branch length based on MSAs and substitution models. For this procedure, 2 × 10⁶ MCMC (Markov chain Monte Carlo) generations were produced and sampled at each 1000, yielding 2000 estimates with 25% discarded as burn-in. The trees were visualized and manipulated with FigTree⁷.

Arabidopsis thaliana ecotype Col-0 is the wild-type used in this study. The mutant T-DNA lines used herein were obtained from ABRC (Arabidopsis Biological Resource Center) for the FIP gene (Salk_080769C and Salk_069143C). To identify homozygous in the F3 generation, PCR was performed with three primers (left border 5′-AACTGCATTCCCGATCCTCT-3′; right border 5′-AAATCCTGCTCCGTCACATT-3′and LBb1.3 5′-ATTTTGCCGATTTCGGAAC-3′) using DNA samples from the leaves of three-week-old plants as template. The plants were grown in soil under control conditions (22°C, 16 h/8 h light/dark, 120 μmol m^-2 s^-1).

Plants overexpressing FIP were generated by cloning the full-length sequence of AtFIP (At5g02160) next to the CaMV35S promoter in the pK7WG2 Gateway vector. The sequence of AtFIP containing 390 bp was amplified using primers 5′-CACCATGACGATCGCACCGGCATTG-3′ and 5′-TGATTTATCAATCTGGTTAAGC-3′ and Platinum DNA Polymerase Taq High Fidelity (Thermo). The amplicon was cloned into the pENTR/D-TOPO vector (Gateway) according to the manufacturer’s instructions and subjected to DNA sequencing. The amplicon was transferred to the overexpression vector pK7WG2 using the recombination enzyme LR Clonase II (Thermo). Agrobacterium tumefaciens strain GV3101 was used to insert the plasmid construction into the Col-0 plants by the floral dip method (Clough and Bent, 1998). Plants overexpressing FIP were selected from the F3 generation in agar plates containing Murashige and Skoog (MS) medium half strength (PhytoTechnology Lab.) and 50 mg/L of kanamycin. After 2 weeks of growth on selective plates under control conditions (22°C, 16 h/8 h light/dark, 120 μmol m^-2 s^-1), the green seedlings were transferred to soil.

For light stress treatment, three-week-old plants of FIP overexpressing (OE) lines, fip knockdown mutant, and wild-type (WT) growing in soil under control conditions (22°C, 16 h/8 h light/dark, 120 μmol m^-2 s^-1) were transferred to high light conditions (22°C, 16 h/8 h light/dark, 400 μmol m^-2 s^-1) for 11 days. The plants were watered every 2 days throughout the period.

For stress assays with seedlings, seeds of OE lines, fip knockdown mutant, and WT were surface-sterilized, placed on 0.5X (half strength) MS agar plates (PhytoTechnology Lab.) and kept at 4°C for 2 days. The plates were then transferred to control conditions (22°C, 16 h/8 h light/dark, 120 μmol m^-2 s^-1) for 7 days, and then the seedlings were subjected to different stress treatments for more 10 days. The concentrations of paraquat (Methyl Viologen) for oxidative stress were 0.01, 0.05, 0.1, and 0.2 μM. The concentrations of mannitol for osmotic stress were 25, 50, 100, and 150 mM. The concentrations of NaCl for salt stress were 25, 50, 100, and 150 mM. The average root length was calculated considering the difference between the initial and final root length of all repetitions of one independent treatment. One-way ANOVA followed by Tukey’s pairwise (P < 0.05) was applied to show significant differences amongst all plants in different concentrations of the same treatment. Mannitol treatment with three-week-old plants was conducted using 300 mM for 24 h, and the leaves were then collected for isolation of the RNA and real-time RT-PCR analysis.

Total leaf RNA extraction used TRIzol reagent (Invitrogen) followed by a TURBO DNase (Thermo) treatment. The first-strand cDNA was obtained using 1 μg of total RNA, ImProm-II Reverse Transcriptase (Promega), and primer oligo(dT). The DNA isolation was performed with Phenol reagent according (Edwards et al., 1991) method.

Real-time PCR analysis was conducted using Maxima SYBR Green/ ROX qPCR Master Mix 2X (Thermo) with the Applied Biosystems StepOne real time PCR system equipment. The relative expression was calculated with the ΔCP method using ACTIN expression as the reference gene (Pfaffl, 2001). Significant differences between control condition and treatment were indicated by asterisks using Student’s t-test (P < 0.05). One-way ANOVA followed by Tukey’s pairwise (P < 0.05) was applied to show significant differences amongst all plants in different treatments. The measurements were performed using six biological replicates. The genes analyzed by PCR were as follows: ACTIN7 (At5g09810) 5′-CTAGAGACAGCCAAGAGCAGTTC-3′ and 5′-GTTTCATGGATTCCAGGAGCTTC-3′; FIP (At5g02160) 5′-CATTCCCGATCCTCTTCAAA-3′ and 5′-CGAGTCCATTGCAGTTAGCA-3′; FtsH5 (At5g42270) 5′-TTGCTGCTAGACGTGAGCTT-3′ and 5′-TGGATCATACTCCGGCATAA-3′; D1 (AtCg00020) 5′-TCCGGAACTGCCATTCTAAC-3′ and 5′-TCCGATTCCAAAGTTCGTTC-3′; HSP60-2 (At2g33210) 5′-CCGCATTAGTTGATGCTGCAAGTG-3′ and 5′-CGTTGGAATCTCAGTCACAACTGC-3′; AOX1a (At3g22370) 5′-AGCATCATGTTCCAACGACGTTTC-3′ and 5′-GCTCGACATCCATATCTCCTCTGG-3′ and Cu-Zn-SOD (At1g08830) 5′-GGAACTGCCACCTTCACAAT-3′ and 5′-TCCAGTAGCCAGGCTGAGTT-3′.

The yeast two-hybrid system was employed to identify proteins involved in the stress response mechanism that potentially interacts with FtsH protein in chloroplasts, since FtsH is directly involved in stress response in plants. The complete Arabidopsis FtsH5 gene sequence was used as bait to transform Saccharomyces cerevisiae. Screening against an Arabidopsis library resulted in the identification of 48 positive candidates that activated the histidine and adenine reporter genes (Supplementary Table S1). Among them, a hypothetical plastidial protein (AT5G02160) named FIP (FtsH5 Interacting Protein) was chosen for further characterization due to its localization and potential regulatory role provided by the zinc-finger domain. FIP possesses an N-terminal transit peptide followed by a hydrophobic domain and a zinc-finger domain (Figures 1A,B). Despite the presence of a zinc-finger domain (C₄-type) with two CXXCXGXG conserved repeats, characteristic of DNAJ protein, the conserved J domain is absent in FIP. FtsH5-FIP interaction was confirmed using an in vitro GST pull-down assay (Figure 1C). The production of recombinant proteins used in this experiment is shown in Supplementary Figures S1A–D.

The subcellular localization of FIP in vivo was verified by fusing FIP with GFP. This construct was used to transiently transform Nicotiana tabacum leaves. FIP::GFP-transformed tissues displayed a GFP signal in chloroplasts (Figure 2A). To verify the intra-plastidial localization, the in vitro-translated precursor to FIP (pFIP; 13 kDa) was added to intact chloroplasts. FIP precursor was imported into the chloroplasts and processed to a 9 kDa mature form (Figure 2B, lane 2). The mature forms resided inside the organelle after treating recovered chloroplasts with thermolysin (Figure 2B, lane 3). FIP mature protein resided in the thylakoid fraction (Figure 2B, T, lane 5) rather than the stroma (Figure 2B, S, lane 4). Proper integration into the thylakoids was verified by protease treatment or extraction with 100 mM NaOH. Protease treatment produced a product of partial degradation of ∼5 kDa (Figure 2B, lane 6) and FIP was resistant to NaOH extraction (Figure 2B, TN, lane 7). These results indicated that FIP was properly integrated into thylakoid membrane.

As a first step in studying FIP evolution, we searched for FIP homologous in photosynthetic organisms and found that only mosses and higher plants carry homologous sequences. FIP amino acid alignment using higher plants sequences showed that the zinc-finger domain is conserved, in contrast to Physcomitrella used herein for comparisons (Figure 3A).

Considering FIP was found interacting with the FtsH5 protease, a phylogenetic tree for type A FtsHs was inferred and presented a monophyletic taxon composed of species possessing FIP protein, along with a few exceptions from species of the Chlorophyta division that lack FIP (Figure 3B). These findings indicated that type A FtsHs shared some features among these groups, which might have been present in FtsH proteases from the shared ancestors of green algae, mosses and higher plants before the advent of FIP.

To understand the role of FIP, plants with altered levels of FIP were analyzed. Two independent Arabidopsis knockdown mutants for fip were obtained from the ABRC (Salk_080769C and Salk_069143C). Both mutants have a T-DNA insertion located in the intron of the FIP gene, but different positions: +303 and +425 from the ATG (+1) and before the zinc-finger domain (Figure 4A). The T-DNA insertion was confirmed by RT-PCR using the primers displayed in Figures 4A,B and Supplementary Figure S2. We designated these mutants fip-1 and fip-2, respectively. Both mutants showed an extremely low level of FIP transcripts, as measured by RT-PCR (Figure 4C) and real time RT-PCR (Figure 5A). Two independent Arabidopsis transgenic OE lines FIP were obtained and referred to as OE-1 and OE-2. Both lines showed significantly higher transcript levels of FIP when compared with wild-type plants (Figures 4C, 5A).

Semi-quantitative PCR analysis visually confirmed the variation in FIP levels observed between knockdown mutants and OE lines. However, no differences were observed between the plants when comparing FtsH5, D1, or Actin transcripts under unstressed control conditions (Figure 4C). fip knockdown mutants and OE lines were phenotypically indistinguishable from wild-type when grown under control conditions (22°C, 16 h/8 h light/dark, 120 μmol m^-2 s^-1, Figure 4D).

Real-time RT-PCR was performed to evaluate the expression levels of FIP in OE lines, fip knockdown mutants, and WT plants exposed to high light and osmotic stresses. For the high light stress evaluation, three-week-old plants were transferred from a control condition (22°C, 16 h/8 h light/dark, 120 μmol m^-2 s^-1) to a high light condition (22°C, 16 h/8 h light/dark, 400 μmol m^-2 s^-1) for 11 days. In addition, three-week-old plants were submitted to a 24-h stress with 300 mM of mannitol.

Real-time RT-PCR results demonstrated that FIP was down-regulated in OE lines, fip knockdown mutants, and WT plants exposed to high light and osmotic stress (Figure 5B) when compared to control conditions. This expression pattern is consistent with the idea that FIP might have an inhibitory role in plants, since FIP levels are reduced during stress conditions. Considering that plants expressing different levels of FIP are not under the regulation of the same promoter (OE lines are controlled by the 35S promoter), it may indicate that FIP has some kind of post-translational regulation. The transcript levels of FtsH5 did not show significant difference between the plants expressing different levels of FIP under stress conditions (Figure 6A). The transcript levels of the stress-responsive genes Heat Shock Protein 60 (HSP60-2, Figure 6B), Alternative Oxidase 1a (AOX1a, Figure 6C) and Cu-Zn Superoxide Dismutase (Cu-Zn-SOD, Figure 6D) were up-regulated compared with the control condition in all plants, confirming the efficiency of the stress treatments. These results suggest that FIP is an abiotic stress-related gene.

Morphological and/or developmental alterations were evaluated in fip knockdown mutants and OE lines submitted to abiotic stress conditions. Differences in phenotype were observed in fip mutants submitted to a high light condition (22°C, 16 h/8 h light/dark, 400 μmol m^-2 s^-1 for 11 days) when compared with wild-type plants and OE lines growing under the same condition (Figure 7A). fip knockdown mutant plants became greener for a longer time, demonstrating an increased adaptation to high light conditions.

Seedlings of FIP OE lines, fip knockdown mutants, and wild-type (WT) plants were grown on agar plates containing 0.5X MS medium for 7 days and transferred to agar plates containing increasing paraquat, mannitol or NaCl concentrations for an additional 10 days. fip knockdown mutant seedlings growing in 0.1 μM of paraquat presented slightly improved leaf development compared with OE lines and WT plants (Figure 7B, left). No differences were visible between the leaves when the seedlings were grown in 0.2 μM of paraquat; however, the roots of fip mutants were longer and more branched than OE and WT seedlings (Figures 7B (right),C). No differences were visible between the leaves when the seedlings were grown in the presence of increasing amounts of mannitol (Figure 7D, left). A slightly improvement could be observed in roots length of fip mutants compared with OE and WT seedlings when plants were grown in 100 mM of Mannitol (Figures 7D (right),E). On the other hand, fip mutant seedlings leaves were clearly greener than the OE lines and WT plants in 100 mM of NaCl (Figure 7F, left) and the roots longer and more branched (Figures 7F (right),G). All together, these observations corroborate the low levels observed of stress-inducible genes in fip knockdown mutants when compared to OE lines and WT plants (Figures 6B–D), clearly demonstrating that FIP is involved in the abiotic stresses response.