Suppressors of the Chloroplast Protein Import Mutant tic40 Reveal a Genetic Link between Protein Import and Thylakoid Biogenesis[CC-BY]

Genetic analyses identify the chloroplast proteins ALB4 and STIC2 as coactors in a specialized thylakoid protein targeting pathway, and a link between envelope translocation and thylakoid targeting. To extend our understanding of chloroplast protein import and the role played by the import machinery component Tic40, we performed a genetic screen for suppressors of chlorotic tic40 knockout mutant Arabidopsis thaliana plants. As a result, two suppressor of tic40 loci, stic1 and stic2, were identified and characterized. The stic1 locus corresponds to the gene ALBINO4 (ALB4), which encodes a paralog of the well-known thylakoid protein targeting factor ALB3. The stic2 locus identified a previously unknown stromal protein that interacts physically with both ALB4 and ALB3. Genetic studies showed that ALB4 and STIC2 act together in a common pathway that also involves cpSRP54 and cpFtsY. Thus, we conclude that ALB4 and STIC2 both participate in thylakoid protein targeting, potentially for a specific subset of thylakoidal proteins, and that this targeting pathway becomes disadvantageous to the plant in the absence of Tic40. As the stic1 and stic2 mutants both suppressed tic40 specifically (other TIC-related mutants were not suppressed), we hypothesize that Tic40 is a multifunctional protein that, in addition to its originally described role in protein import, is able to influence downstream processes leading to thylakoid biogenesis.


INTRODUCTION
Plastids are plant cell organelles that each form an enclosed compartment containing ;3000 different protein species, depending on the developmental stage and tissue-specific differentiation Leister, 2016). While a small fraction of plastid proteins are encoded on the plastid genome, most are encoded on the nuclear genome as preproteins with an N-terminal transit peptide and imported posttranslationally through the translocons at the outer (TOC) and inner (TIC) chloroplast envelope membranes (Jarvis, 2008;Li and Chiu, 2010;Shi and Theg, 2013;Demarsy et al., 2014;Paila et al., 2015). Plastid targeting specificity is conferred by the Toc34 and Toc159 families of TOC receptors in the outer plastid membrane, which can specifically interact with the transit peptides of preproteins. Each preprotein is then threaded through adjacent TOC and TIC channels in the outer and inner envelope membranes. The transit peptide is cleaved off by the stromal processing peptidase when the preprotein emerges on the stromal side of the channel, and the mature protein then either is folded to its native structure or engages downstream targeting pathways (Jarvis, 2008;Li and Chiu, 2010;Shi and Theg, 2013;Demarsy et al., 2014;Paila et al., 2015).
The identity and nature of the TIC channel is disputed and subject to current research. Suggested TIC channel subunits are Tic110 (Heins et al., 2002), Tic20 (Chen et al., 2002), and Tic21 (Teng et al., 2006). The C terminus of Tic110 has been reported to form either four transmembrane domains (Heins et al., 2002;Balsera et al., 2009) or a stromal scaffold for chaperone binding (Jackson et al., 1998;Inaba et al., 2003Inaba et al., , 2005. A recent detailed structural analysis of Tic110 supports the latter model (Tsai et al., 2013), which makes a channel function for Tic110 structurally unlikely. However, the severe embryo-lethal phenotype of Arabidopsis thaliana tic110 mutants and the proposed association of Tic110 with TOC proteins (Akita et al., 1997;Kovacheva et al., 2005) suggest an essential role of Tic110 in the late steps of protein import, e.g., in import propulsion, precursor processing or folding, or postimport inner membrane insertion. Recently, a 1 MD complex that associates with translocation intermediate precursor proteins was isolated from inner envelope membranes and found to comprise three novel components, Tic214, Tic100, and Tic56, as well as Tic20 and Tic21 (Kikuchi et al., 2013). Knockout mutants of all these components have severe phenotypes (Chen et al., 2002;Teng et al., 2006;Kikuchi et al., 2013); thus, the 1 MD complex seems to play an essential role in protein import. Importantly, Tic110 is not part of the 1 MD complex (Kikuchi et al., 2013). Instead, it interacts with Tic40 and the stromal chaperone Hsp93, which have been suggested to form a motor complex (Chou et al., 2003(Chou et al., , 2006. Tic40 is anchored in the inner membrane by a single N-terminal transmembrane domain, while its stromal C terminus comprises a putative tetratricopeptide repeat (TPR) domain and a domain also found in the Hsp70interacting protein (Hip) and Hsp70/Hsp90-organizing protein (Hop) cochaperones (Chou et al., 2003(Chou et al., , 2006Bédard et al., 2007;Kao et al., 2012). Interaction of Tic40 with Tic110 via its TPR domain may release preproteins from Tic110 to enable precursor processing, whereas the Tic40 Hip/Hop domain may stimulate the ATPase activity of Hsp93, possibly to provide a driving force for the import process (Chou et al., 2006). However, it was recently suggested that this energy is not used for protein import per se, but rather for quality control (Sjögren et al., 2014;Flores-Pérez et al., 2016) and/or precursor processing (Huang et al., 2016). Instead, plastid-localized cpHsp70 may be the dominant player in protein import propulsion (Su and Li, 2010;Liu et al., 2014;Huang et al., 2016).
Once a preprotein reaches the stromal side of the chloroplast envelope and its transit peptide has been cleaved off by the stromal processing peptidase, in many cases, the protein is further targeted to its correct subcompartment within the chloroplast. For example, some proteins are reinserted into the inner envelope membrane after first being released into the soluble stromal phase (Tripp et al., 2007;Chiu and Li, 2008;Viana et al., 2010). Other imported proteins, mainly those involved in the photosynthetic electron transfer reactions, are targeted via different pathways to the thylakoids (Celedon and Cline, 2013). Thylakoid lumen proteins usually carry further targeting information at their N-terminal ends, which directs their targeting via the chloroplast secretory (cpSec) pathway or the chloroplast twin arginine translocation (cpTat) pathway. Thylakoid membrane proteins may insert spontaneously, use one of the aforementioned pathways, or engage the chloroplast signal recognition particle (cpSRP) pathway (the latter is employed predominantly by lightharvesting chlorophyll binding proteins [LHCPs]) (Celedon and Cline, 2013).
In the cpSRP transport pathway, LHCP proteins are recognized by a bipartite cpSRP composed of cpSRP54, an ortholog of the bacterial SRP protein, and cpSRP43, a protein that specifically recognizes a targeting signal within LHCP termed L18 (Schuenemann et al., 1998;Tu et al., 2000;Stengel et al., 2008). The cpSRP cooperates in a GTP-dependent manner with cpFtsY, an ortholog of the SRP receptor, to deliver LHCP to ALBINO3 (ALB3), an ortholog of the bacterial YidC protein and a member of the Alb3/Oxa1/YidC insertase family. The ALB3 protein then mediates the insertion (and assembly) of the LHCP protein into the thylakoid membrane (Moore et al., 2000;Falk et al., 2010;Dünschede et al., 2011). Accordingly, Arabidopsis alb3 mutants display a strongly chlorotic/albino phenotype linked to defective LHCP biogenesis (Sundberg et al., 1997;Asakura et al., 2008); by contrast, those with defects affecting a second, paralogous protein called ALB4 appear relatively normal, suggesting that this factor is functionally different (Gerdes et al., 2006;Benz et al., 2009;Trösch et al., 2015). Recently, a novel component of the cpSRP pathway was identified: The LHCP targeting deficient (LTD) protein acts upstream of the cpSRP by mediating the transfer of the newly imported LHCP proteins from the TIC translocon to the cpSRP complex (Ouyang et al., 2011). Interestingly, Tic40 and Tic110 were both implicated in LTD association with the TIC, and in early recognition of LHCPs (Ouyang et al., 2011). Thus, Tic40 may have an additional role in the onward targeting of some thylakoid proteins, which is possibly analogous to its proposed function in the postimport reinsertion of other proteins into the inner envelope membrane (Chiu and Li, 2008).
More work is needed to clarify the role of Tic40 in chloroplast protein transport. Because Tic40 is encoded in Arabidopsis by a single gene that upon mutation causes a distinct yet viable chlorotic phenotype (Chou et al., 2003;Kovacheva et al., 2005), it is well suited for analysis using a forward-genetic suppressor screen. Here, we describe such a screen, along with the detailed characterization of the resulting stic1 and stic2 suppressor mutants.

Identification of suppressor of tic40 Mutants
To identify mutant plants in which the chloroplast protein import defect displayed by tic40 is suppressed, ;40,000 tic40-4 seeds were mutagenized using EMS. The M1 plants were grown to maturity to produce the M2 generation of seeds, and the progeny of ;10,000 M1 plants was screened for putative suppressor of tic40 (stic) mutants. Because young tic40 seedlings display a reduced size and clear pale-green mutant phenotype (Chou et al., 2003;Kovacheva et al., 2005), it was assumed that suppression of tic40 would result in improved growth and a darker green phenotype. Thus, M2 seeds were germinated and grown in vitro for 10 d and then screened visually for seedlings that appeared significantly larger and greener than the tic40-like seedlings. This led to the identification of several putative stic mutants, which were thoroughly characterized genetically to exclude false positives, and further backcrossed to tic40-4 to remove unwanted background mutations.
The stic Mutations Suppress the Cellular and Organellar Structural Defects and Chloroplast Protein Import Defects of tic40 The tic40 mutant phenotype is characterized by a reduced number of chloroplasts per cell and by larger chloroplasts with a swollen, more spherical morphology as well as an elevated content of plastoglobules relative to wild-type organelles (Chou et al., 2003;Kovacheva et al., 2005). Furthermore, thylakoid ultrastructure in tic40 appears disorganized, with fewer and less stacked thylakoids (Chou et al., 2003;Kovacheva et al., 2005). To assess whether these defects were visibly reduced in the stic mutants, we employed light and transmission electron microscopy. Light micrographs of whole cotyledon cross sections revealed a disorganized architecture with reduced cell density in tic40 (Supplemental Figure 2). Remarkably, as exemplified by the data shown for stic1-1 tic40 and stic2-1 tic40, all of the stic mutants displayed a noticeable improvement in cellular architecture, with increased cell density and a more uniformly extended abaxial epidermal cell layer (Supplemental Figure 2). Transmission electron micrographs showed that the thylakoids are more developed in the stic mutants than in tic40, appearing more densely packed ( Figure 1C) and having a significantly greater number of interconnections ( Figure 1D). The increased cell density and improved thylakoid organization of the stic mutants, compared with tic40, most likely contribute to the larger size and greener appearance of the corresponding plants. However, the size and shape of the chloroplasts, the increased number of plastoglobules, and the number of chloroplasts per cell was not significantly different in the stic mutants relative to tic40 ( Figure 1C; Supplemental Figure 2).
To assess if the improvement in chloroplast development seen in the double mutants was linked to recovery of the tic40 protein import defect described previously (Chou et al., 2003;Kovacheva et al., 2005), we performed in vitro protein import assays with chloroplasts isolated from wild-type, tic40, stic1-1 tic40, and stic2-1 tic40 seedlings and radiolabeled Rubisco small subunit (SSU) precursor protein (Figures 1E and 1F). For both stic1-1 tic40 and stic2-1 tic40, the SSU precursor was imported significantly more efficiently than for the tic40 single mutant. The amount of mature protein detected at the final time point in each case was ;20 to 25% greater in the double mutants than in tic40, confirming that the suppression phenotypes are indeed linked to partial recovery of the tic40 protein import defect.

Molecular Identification of the stic1 and stic2 Loci
To identify the mutations responsible for the suppression effects, the stic mutants were crossed to Arabidopsis plants of the Landsberg erecta ecotype into which the tic40-4 mutation (originally of the Columbia-0 ecotype) had been introgressed. The F2 generations from these crosses were used initially to roughly map the mutations by assessing their linkage to polymorphic markers located on each arm of the five Arabidopsis chromosomes (Konieczny and Ausubel, 1993;Berendzen et al., 2005). This revealed that all five stic1 mutants carry a mutation on chromosome 1, while the two stic2 mutants mapped to chromosome 2 (Supplemental Figure 3). Further fine-mapping and sequencing, using the stic1-4 and stic2-1 alleles only, revealed potential splice-defect mutations in ALB4 (G to A at nucleotide 920 relative to the ATG) for stic1-4 ( Figure 2A) and in a gene with unknown function (At2g24020; G to A at nucleotide 670 relative to the ATG) for stic2-1 ( Figure 2B). ALB4 is a paralog of the ALB3 thylakoid membrane insertase and has been proposed to cooperate with ALB3 in the insertion, assembly, or stabilization of thylakoid membrane proteins (Benz et al., 2009;Trösch et al., 2015). Both ALB4 and ALB3 are members of the conserved Alb3/Oxa1/YidC family of protein insertases (Luirink et al., 2001;Kuhn et al., 2003;Yi and Dalbey, 2005). STIC2 is a homolog of bacterial ybaB, which encodes a protein implicated in membrane protein biogenesis (Skretas and Georgiou, 2010), as well as in DNA binding (Cooley et al., 2009;Jutras et al., 2012).
(C) Micrographs of typical chloroplasts are shown for each genotype, which were selected on the basis of a thorough analysis of three different plants per genotype. Bars = 2 mm. (D) To provide a quantitative measure of the extent of thylakoid development in the different genotypes, numbers of interconnecting stromal lamellae associated with each granum were counted. The presented data are from the analysis of a total of 60 grana proportionately representative of three typical chloroplasts from three different plants per genotype. Values are means and error bars correspond to the SE of the mean (n = 60).
(E) and (F) Analyses of chloroplast protein import efficiency in the stic mutants. In vitro import of radiolabeled Rubisco small subunit (SSU) precursor protein into chloroplasts isolated from 12-to 14-d-old plants of the following genotypes: wild type, tic40, and representative stic1 tic40 (stic1-1 tic40-4) (E) and stic2 tic40 (stic2-1 tic40-4) (F) double mutants. Top: Phosphor screen images of representative chloroplast protein import time-course experiments. Times shown indicate minutes (min) after the start of each import reaction. Positions of the precursor (p) and imported mature (m) forms of SSU are shown. Bottom: The amount of accumulated mature SSU protein at each time point was determined by quantifying the relevant band intensities in the top image and in two further similar experiments. Values were each expressed as a percentage of that obtained for wild-type chloroplasts at the final time point in each case. Values shown are mean percentages derived from three independent experiments; error bars show SE of the mean (n = 3). 2B). Using RT-PCR, the predicted splicing defects of the stic1-4, stic2-1, and stic2-2 alleles were confirmed, while the nonsense alleles stic1-1, stic1-3, and stic1-5 were shown to have less ALB4 transcript and the missense allele stic1-2 to have more ALB4 transcript relative to the wild type (Supplemental Figure 4).
In addition to the obviously aberrant transcripts, all splice-defect mutants produced transcripts of similar size to the relevant wildtype transcript, albeit at reduced levels (Supplemental Figure 4). However, in the stic1-4 mutant, these wild-type-like transcripts correspond to the loss of the first 1, 3, or 12 nucleotides of the (A) and (B) Schematic diagrams of STIC1 (A) and STIC2 (B) indicating the positions of the point mutations identified in the five stic1 mutants (stic1-1 to stic1-5) and the two stic2 mutants (stic2-1 and stic2-2). The T-DNA insertion sites for each stic T-DNA mutant (alb4-1, stic2-3, and stic2-4) as determined in this study are also indicated. Start and stop codons are shown. The gray boxes represent the 59 and 39 untranslated regions, whereas the black boxes and interconnecting black lines represent exons and introns, respectively. (C) and (D) Analyses of STIC protein expression in the stic mutants. Immunoblot analyses of total seedling protein extracts (40 mg per sample) from the different stic1 (C) and stic2 (D) mutants. Extracts from stic1 single mutants and wild-type control seedlings (C), or from stic2 tic40 double mutants and tic40 control seedlings (D), were analyzed. Antisera raised against Tic110, stromal Hsp70 (cpHsp70), ALB4, and STIC2 were used to detect the corresponding proteins, as indicated to the left of each blot. fourth exon. Similarly, in the case of stic2-1 and stic2-2, wildtype-like transcripts carried small deletions that led to frame shifts or deletions causing the removal of a few codons and/or missense mutations. For all mutants, although the RT-PCR results suggested that they may still produce some form of the relevant STIC protein, further analysis indicated that, with the exception of stic1-2, this is not the case (see "Analysis of STIC Protein Expression in the stic Mutants" section below).

Confirmation of Suppression by Independent T-DNA Mutants and Complementation
To confirm that STIC1/ALB4 and STIC2 loss of function is responsible for the suppression of tic40 in the stic mutants, we identified T-DNA mutants for each gene. For STIC1/ALB4, we obtained the alb4-1 (SALK_136199) mutant, which was characterized previously by Gerdes et al. (2006) and shown to have a significant knockdown effect on the expression of the ALB4 gene. We confirmed the presence and position of the T-DNA insertion ( Figure 2A) as well as the knockdown effect (30.3% ALB4 transcript relative to the wild type in 2-week-old seedlings, SE = 7.2%, n = 6). For STIC2, we identified two T-DNA mutants carrying insertions within the open reading frame of the gene: stic2-3 (SALK_001500) and stic2-4 (WiscDsLox 445D01) ( Figure 2B). All of these stic1 and stic2 T-DNA mutants were crossed to tic40, and corresponding double homozygous mutants (alb4-1 tic40-4, stic2-3 tic40-4, and stic2-4 tic40-3) were identified. The phenotypes of these T-DNA double mutants were then compared with those of the original stic1 tic40 and stic2 tic40 double mutants and found to be visibly indistinguishable (Supplemental Figures  5A and 5C). This phenotypic similarity was also demonstrated quantitatively by measuring chlorophyll concentrations in the rosette leaves of all genotypes (Supplemental Figures 5B and 5D).
Finally, we confirmed the identity of the STIC1 and STIC2 genes by complementation. The stic1-1 tic40-4 and stic2-2 tic40-4 double mutants were transformed with constructs that drive overexpression of STIC1/ALB4 and STIC2, respectively, each with a C-terminal FLAG tag (Supplemental Figure 6). The resulting transformants were phenotypically similar to tic40 plants (Supplemental Figures 6A and 6C) and were shown to express the respective FLAG-tagged protein (Supplemental Figures 6B and  6D). Thus, we concluded that the phenotypic reversion was caused by the expression of ALB4-FLAG and STIC2-FLAG, respectively.

Analysis of STIC Protein Expression in the stic Mutants
Immunoblot analyses of whole-seedling protein samples from all stic1 mutants showed that none of the stic1 EMS alleles, with the exception of stic1-2, accumulates the full-length STIC1/ALB4 protein ( Figure 2C). Because the ALB4 protein expressed in the stic1-2 mutant carries a G397S amino acid substitution due to the stic1-2 missense mutation (Figure 2A), it is evident that this mutation impedes the function of the ALB4 C terminus and that an intact C terminus is important for ALB4 function. As reported previously, the alb4-1 T-DNA mutant expresses the ALB4 protein at a reduced level (Gerdes et al., 2006). It is noteworthy that the antibody used to detect the ALB4 protein was raised against only the 154 most C-terminal amino acids of the protein, so the accumulation of truncated forms of the STIC1/ALB4 protein cannot be ruled out, especially as most of the mutants carry nonsense mutations ( Figure 2A).
We also tested the expression of the STIC2 protein by immunoblotting ( Figure 2D). Surprisingly, despite the complete lack of intact STIC2 transcript in the stic2-3 and stic2-4 T-DNA mutants (Supplemental Figure 4D), we found that all stic2 alleles accumulate small amounts of a protein that is comparable in size with STIC2 and cross-reacts with the anti-STIC2 antibody raised against the full predicted mature STIC2 protein (amino acids 49 to 182 relative to the start codon) ( Figure 2D).
The Arabidopsis genome encodes a close homolog of STIC2, named STIC2-Like (STCL; At4g30620), with which STIC2 shares both structural (similar exon structure; 78% amino acid sequence identity) and functional similarity (Supplemental Figure 7). A phylogenetic analysis of STIC2/STCL-related sequences from different species confirmed that this gene family is ubiquitously distributed in plants and that the two Arabidopsis genes arose from a duplication that occurred in an early ancestor of the Brassicaceae (Supplemental Figure 8 and Supplemental Data Set 1). Given the high level of similarity between the two proteins, the cross-reacting band observed in the stic2 mutants is most likely STCL, not STIC2. Because the intensity of this band in the stic2-1 and stic2-2 mutants is comparable to that observed in the stic2 T-DNA mutants ( Figure 2D), we conclude that these point mutants do not accumulate significant amounts of STIC2 protein. This point was further supported by the similar analysis of plants deficient in both STIC2 and STCL (Supplemental Figure 9).

Phenotypic Characterization of the stic1 and stic2 Mutants
Given that the stic1 tic40 and stic2 tic40 mutants are visibly identical and contain similar concentrations of chlorophyll (Figures 2E and 2F), we hypothesized that the STIC proteins act in a common process (e.g., protein targeting to the thylakoids) such that the removal of one or the other protein disrupts this process. To begin to address this possibility, we generated a stic1 stic2 tic40 triple mutant, reasoning that in an alternative scenario where stic1 and stic2 suppress tic40 by disrupting different processes, their suppression effects would likely be additive. The fact that the phenotype and chlorophyll content of the triple mutant were identical to that of each double mutant, as shown in Figures 2E and 2F, supported our hypothesis that STIC1/ALB4 and STIC2 cooperate in a common process.
To determine if the stic1 and stic2 mutations affect chloroplast and plant development in isolation, we outcrossed the mutants to wild-type plants to obtain the single stic1 and stic2 mutants. As shown in Figures 2E and 2F, both single mutants look identical to, and have essentially the same chlorophyll content as, wild-type plants. This suggests that the loss of STIC proteins may be compensated for by redundancy or the operation of parallel pathways or that STIC functions become critical only under certain circumstances. The fact that the stic1-1 mutation was not found to have a visible effect on plant development is in agreement with the findings of Gerdes et al. (2006) and Trösch et al. (2015), who characterized the alb4-1 mutant, but in contrast to the stunted growth phenotype observed by Benz et al. (2009).
We also generated stic1 stic2 double mutant plants by outcrossing the stic1 stic2 tic40 triple mutant to the wild type. Like the stic1 and stic2 single mutants, the double mutant was phenotypically indistinguishable from wild-type plants under normal growth conditions ( Figures 2E and 2F). This observation is consistent with the notion that the STIC1/ALB4 and STIC2 proteins act in a common pathway, the disruption of which is potentially compensated for by parallel processes or can be tolerated under normal conditions.

Localization of the STIC1 and STIC2 Proteins
Since the STIC2 protein is predicted to carry a transit peptide, we wished to determine if the protein is indeed targeted to chloroplasts. Thus, we generated a construct for the transient expression in protoplasts of STIC2 fused to a C-terminal YFP tag. A similar construct was made for STIC1/ALB4, although this protein was previously established as a chloroplast protein that associates with the thylakoid membranes (Gerdes et al., 2006). In addition, a free-YFP construct and a previously generated Tic110-YFP fusion construct (Bédard et al., 2007) were used as cytosol and chloroplast envelope markers, respectively. Using confocal microscopy to image transfected protoplasts, these control proteins displayed fluorescence patterns typical of cytosol and envelope localization, respectively ( Figure 3A). By contrast, the ALB4-YFP signal closely overlapped with the chlorophyll autofluorescence signal, indicating that the protein likely accumulates in the thylakoids as previously reported (Gerdes et al., 2006). The STIC2-YFP protein also accumulated in the chloroplasts, confirming the TargetP (Emanuelsson et al., 2000) prediction, but the fluorescence signal in this case displayed a different profile, being concentrated in areas of the chloroplasts where the chlorophyll fluorescence signal was least intense, which suggests that the protein accumulates in the stroma ( Figure 3A). A similar STCL-YFP fusion displayed the same localization profile (Supplemental Figure 7D).
To confirm the localizations of the STIC proteins using an alternative approach, chloroplasts were isolated from seedlings expressing the ALB4-FLAG and STIC2-FLAG fusion proteins described earlier and then subfractionated by sucrose densitygradient centrifugation to obtain stromal, envelope, and thylakoidal fractions. Immunoblotting analysis of the different fractions using antibodies against the FLAG tag and the following control proteins enabled us to assess the localization of the STIC proteins biochemically; the controls were Tic110 (envelope), LHCP (thylakoid), ALB3 (thylakoid), and the small subunit of Rubisco (SSU; stroma). In accordance with the microscopy data, the ALB4-FLAG protein was found exclusively in the thylakoid fraction ( Figure 3B), while the STIC2-FLAG protein was detected exclusively in the stromal fraction ( Figure 3C). The control proteins largely behaved as expected, although we did detect some Tic110 in the thylakoid fractions, indicating that these were somewhat contaminated with envelopes ( Figures 3B and 3C), as is often the case in such subfractionation experiments.
To assess the possibility that the localization of STIC1/ALB4 is dynamic, enabling colocalization with Tic40 in the envelope at specific developmental stages, we similarly analyzed chloroplasts isolated from plants of different ages, including some that were of approximately the same age as those used for the microscopy (5 weeks old; Supplemental Figure 10). In each case, we failed to detect any ALB4-FLAG protein in the envelope fraction, indicating that the protein is stably confined to the thylakoid membrane.
The accumulation of STIC2 in the stroma is not surprising because the protein does not possess any hydrophobic regions that might form a transmembrane helix to anchor it in a membrane (Krogh et al., 2001). However, a previously reported chloroplast proteome study suggested that some STIC2 protein is associated with the thylakoids (Peltier et al., 2004), presumably as a peripheral protein via protein-protein interactions. The fact that no STIC2-FLAG protein was detected in the thylakoids in our study suggests that either the chloroplast thylakoid fraction analyzed by Peltier et al. (2004) was slightly contaminated with stromal proteins, or the amount of STIC2-FLAG that associates with the thylakoids is so low as to be beyond the limits of detection in our assay.

Chloroplast Morphology and Ultrastructure of the stic Mutants
Because STIC1/ALB4 and STIC2 are both chloroplast proteins, we sought to gain further insight into their functions by characterizing chloroplast ultrastructure in corresponding single mutant plants, which at a macroscopic level are indistinguishable from the wild type. As shown in Figure 4A (upper micrographs), we observed that the cotyledon chloroplasts of stic1 and stic2 seedlings have an unusual morphology: In both cases, the chloroplasts appeared swollen and less appressed to the cell periphery than wild-type chloroplasts. The shape difference was quantified by measuring the length and width of chloroplast cross sections and comparing the length/width ratio with that of the wild type ( Figure  4B). In both mutants, this ratio is roughly three-quarters that of wild-type chloroplasts, confirming that they are more spherical.
Upon observation at higher magnification, the thylakoid membrane networks in the stic mutants appeared to be as developed and complex as those of wild-type chloroplasts ( Figure  4A, lower micrographs): Comparable grana of variable thickness were seen in all genotypes, interconnected by networks of stromal lamella membranes, dispersed throughout each chloroplast cross section. Nonetheless, the lamellae were found to be less parallel and tightly packed (i.e., more loosely distributed) in the mutants, possibly due to the rounder shape of the chloroplasts. Interestingly, a significantly higher number of plastoglobules (lipid bodies) were visible in the mutant chloroplasts, as illustrated in Figure 4C, pointing toward a thylakoid biogenesis defect (Rudella et al., 2006).
In relation to stic1, these observations are consistent with those made by Gerdes et al. (2006) and Trösch et al. (2015) using the alb4-1 T-DNA insertion mutant. The appearance of the stic mutant chloroplasts is reminiscent of the previously described swollen chloroplast phenotype of tic40 plants (Kovacheva et al., 2005;Bédard et al., 2007), although in that case the thylakoid membrane system was much less developed. The chloroplast morphology defect is possibly more similar to that observed in the vipp1 knockdown mutant (Zhang et al., 2012), which is believed to have an inner envelope membrane integrity defect.

Specificity of the Suppression Effects of stic1 and stic2
To determine if the suppression effects of the stic mutations are specific to tic40, or more general, we crossed the single stic1-1 or stic2-1 mutants to several other protein import mutants: ppi1, toc75-III-3, hsp93-V, and tic110. All of these mutants display a degree of chlorosis linked to defective import, with the lesions occurring at either the outer membrane (ppi1 is a null mutant for the Toc33 receptor; toc75-III-3 is a point mutant of the Toc75 channel) or the inner membrane (hsp93-V is a null mutant for the Hsp93/ ClpC chaperone; tic110 is a heterozygous mutant for a chaperone recruitment scaffold protein) (Jarvis et al., 1998;Kubis et al., 2003;Kovacheva et al., 2005;Stanga et al., 2009;Huang et al., 2011). Double mutants were generated and their phenotypes were (A) Confocal microscopy analysis. Arabidopsis protoplasts transiently expressing the following proteins were analyzed by confocal microscopy: YFP (Free YFP), Tic110-YFP, STIC1/ALB4-YFP, and STIC2-YFP; all of the YFP fusion proteins carried a C-terminal YFP tag. Representative protoplasts are shown. In each case, YFP fluorescence, chlorophyll autofluorescence, merged YFP and chlorophyll fluorescence, and bright-field images are shown. White arrowheads in the STIC2-YFP micrographs indicate areas of the chloroplasts where the YFP signal is clearly accumulated but a strong chlorophyll signal is lacking. Bars = 12.68 mm. (B) and (C) Chloroplast subfractionation analysis. Chloroplasts isolated from transgenic seedlings overexpressing C-terminally FLAG-tagged STIC1/ALB4 (B) or STIC2 (C) were separated into stroma (S), envelope (E), and thylakoid (T) enriched fractions prior to immunoblotting (;15 mg of each sample was analyzed). An anti-FLAG antibody was used to detect the STIC1/ALB4-FLAG and STIC2-FLAG proteins, while antisera raised against Tic110, SSU, LHCP, and ALB3 were used to detect the corresponding reference proteins. compared with those of the corresponding single import mutants (Supplemental Figure 11). Most of the double mutants analyzed were double homozygotes, but because the tic110 mutation is embryo lethal in the homozygous state and causes chlorosis in the heterozygous state (Inaba et al., 2005;Kovacheva et al., 2005), the corresponding double mutants were heterozygous for tic110.
As exemplified for the import mutants in combination with stic1-1 (Supplemental Figure 11A), the double mutants appeared in all cases identical to the corresponding single mutants, regardless of the severity of the chlorotic phenotype displayed. To quantitatively assess these observations, chlorophyll concentrations in each pair of single and double mutants were measured and used to calculate average ratios (Supplemental Figure 11B). In all cases, the ratios obtained were approximately equal to 1, indicating no difference between the double and single mutants. A similar result was obtained for the import mutants in combination with stic2-1 (Supplemental Figure 11C). These results contrast markedly with those obtained for the stic tic40 double mutants, where the equivalent chlorophyll content ratio values obtained were ;1.8 to 1.9. Thus, the results clearly indicate that the stic mutations are not general suppressors of import deficiency. We conclude that the suppression effects mediated by stic1 and stic2 are quite specific for tic40, implying a close functional connection between the STIC proteins and Tic40.

Genetic Interactions between the stic Mutations and Thylakoid Protein Targeting Mutations
Given the similarity between STIC1/ALB4 and ALB3 and the wellestablished role of the latter in thylakoid protein biogenesis, and the genetic evidence pointing toward roles for the STIC proteins in a common process ( Figures 2E and 2F), we wished to investigate if the STIC proteins act in thylakoid protein transport. To begin to address this possibility, further genetic interaction tests were performed, this time with mutants for components of the different thylakoid targeting pathways; i.e., the cpSec, cpTat, and cpSRP . Roughly equal numbers of chloroplasts from two different seedlings per genotype were analyzed. The length and width values were used to derive ratios, as an indicator of chloroplast shape, and the average ratio (6SE) for each genotype is shown. (C) Analysis of plastoglobule numbers. Using the electron micrographs described in (A), the number of plastoglobules per chloroplast cross section in a total of n different organelles was counted (values of n are shown in the figure). Roughly equal numbers of chloroplasts from three different seedlings per genotype were analyzed. The average number (6SE) for each genotype is shown.
The cpsecA1 and hcf106 mutants have seedling-lethal, albino phenotypes similar to that of the alb3 mutant of the cpSRP pathway, and so cannot survive when grown on soil but can be maintained in vitro on medium containing sucrose (Liu et al., 2010). It was recently shown that the alb4-1 mutation interacts synergistically with the alb3-1 null mutation, producing seedlings that are even more chlorotic and weak than the single alb3 mutant (Trösch et al., 2015). This was interpreted to indicate that there is a degree of functional redundancy between the two affected proteins and that ALB4 participates in the cpSRP pathway or in a parallel pathway that performs a similar role. In marked contrast, no synergistic effects were observed when stic1 and stic2 were combined with the cpsecA-1 and hcf106 mutations, as the double mutants were phenotypically indistinguishable from the corresponding albino single mutants ( Figures 5A and 5B). If we assume that the cpSec and cpTat pathways are not completely blocked in these mutants, respectively, then the failure of the stic mutations to accentuate the secA1 and hcf106 phenotypes suggests that the STIC proteins do not participate in these pathways.
The cpsrp54, cpsrp43, and cpftsY mutants are all chlorotic, but they are viable and phenotypically less severe than the albino mutant alb3: The cpsrp54 and cpsrp43 mutants produce rosette leaves that are initially yellow but then become progressively greener ( Figures 5C and 5D; Supplemental Figures 12B to 12D) (Amin et al., 1999;Klimyuk et al., 1999;Yu et al., 2012), whereas cpftsY is more severely chlorotic, remains yellow throughout development, and displays stunted growth ( Figures 5E and 5F) (Asakura et al., 2008). As shown in Figures 5C to 5F, the stic cpsrp54 and stic cpftsY double mutants were all significantly smaller and more chlorotic than the corresponding single mutants, indicating that both stic mutations interact synergistically with cpsrp54 and cpftsY. By contrast, similar synergistic interactions were not detected in the stic1 cpsrp43 and stic2 cpsrp43 double mutants (Supplemental Figures 12B to 12D), implying that the STIC proteins do not act in the canonical cpSRP43 pathway for LHCP targeting.
Overall, these genetic data strongly support the notion that the STIC proteins act together in a specialized cpSRP pathway that is distinct from the cpSec and cpTat pathways. This conclusion is consistent with the thylakoid morphological defects ( Figure 4A) and the accumulation of plastoglobules ( Figure 4C) seen in the stic mutants, and with the genetic evidence suggesting that the two proteins act in a common process (Figures 2E and 2F).

STIC2 Interacts with the C Termini of ALB4 and ALB3 in Vitro
We wished to investigate whether the STIC proteins can interact physically with each other. To address this possibility, we employed in vitro pull-down experiments using recombinant proteins expressed in bacteria ( Figures 6A to 6C). For this work, we used an N-terminally tagged form of the STIC1/ALB4 C terminus (His-ALB4C; residues 345 to 498 of the preprotein), representing most of the stromal C-terminal domain (Falk et al., 2010), and an N-terminally tagged form of the mature STIC2 protein (GST-STIC2; residues 49 to 182 of the preprotein) ( Figure 6A, left side). Equal amounts of soluble lysate from bacteria expressing each protein were mixed together, and then the His-ALB4C protein was affinity purified. The GST-STIC2 protein copurified efficiently with His-ALB4C in the absence of significant amounts of any other protein, providing strong evidence for a direct interaction between STIC2 and the C terminus of STIC1/ALB4 ( Figure 6A, left side). By contrast, when His-ALB4C was replaced in the assay by the His-tagged LTD stromal targeting factor (Ouyang et al., 2011), significant GST-STIC2 copurification was not detected ( Figure  6B). This demonstrated that GST-STIC2 does not associate nonspecifically with the affinity resin used to purify the His-tagged proteins, confirming that the detected copurification ( Figure 6A) reflects a genuine protein-protein interaction.
The stic1-2 mutant carries a glycine-to-serine missense mutation (G397S) that affects the stromal domain of STIC1/ALB4 employed in the in vitro pull-down experiments. In view of the evidence indicating that STIC1/ALB4 and STIC2 act in a common pathway (Figure 2), we reasoned that a consequence of this mutation might be to perturb the interaction between the STIC proteins. To address this possibility, we introduced the G397S mutation into the pull-down construct (His-ALB4C-G397S) and used it in further pull-down studies with the GST-STIC2 protein ( Figure 6A, right side). Although identical amounts of His-ALB4C-G397S and His-ALB4C were employed, in simultaneous experiments conducted in parallel, substantially reduced binding of GST-STIC2 was observed with His-ALB4C-G397S relative to wild-type His-ALB4C (on average, 17% over the four elutions, SE = 4%). These data support the notion that STIC2 interacts specifically with the C terminus of STIC1/ALB4 and indicate that the Gly-397 residue is important for this interaction.
The functional differences between the ALB3 and ALB4 proteins are believed to be mainly attributable to their different stromal C termini. While ALB3 contains two conserved motifs (Motifs II and IV) important for association with the LHCP targeting factor cpSRP43, ALB4 lacks these domains (Falk et al., 2010). Gly-397 of ALB4 has been proposed to be part of Motif III that is present in both ALB4 (residues 397 to 414) and ALB3 (residues 386 to 403) (Falk et al., 2010). Thus, it is possible that this motif is also important for STIC2 interaction with ALB3. Therefore, to determine if the C terminus of ALB3 can also directly interact with GST-STIC2, we conducted further pull-down studies using an N-terminally tagged form of the ALB3 C terminus (His-ALB3C; residues 361 to 462 of the preprotein). As shown in Figure 6C, His-ALB3C also efficiently pulled down the GST-STIC2 protein, indicating that STIC2 interacts not only with STIC1/ALB4 but also with ALB3, further supporting the notion that STIC2 plays a role in protein targeting to the thylakoids.

BiFC Analysis of the Interaction between ALB3, ALB4, and STIC2
To provide in vivo corroboration of the interactions observed in vitro ( Figures 6A to 6C), we employed bimolecular fluorescence complementation (BiFC) in transfected Arabidopsis protoplasts (Lee et al., 2008). Thus, the STIC coding sequences were inserted into vectors that generate fusions to the N-terminal or C-terminal parts of yellow fluorescent protein (nYFP and cYFP, respectively). Various nYFP and cYFP fusion vector pairs were then cotransfected and imaged by confocal microscopy. In this system, YFP fluorescence is only observed when the nYFP and cYFP parts are brought into close proximity by the interaction of the fused proteins (in this case, the STIC1/ALB4, STIC2, and ALB3 proteins), enabling reconstitution of a complete, functional YFP protein  (Lee et al., 2008). As a positive control for transfection, the ALB4-YFP plasmid was used on its own ( Figure 7); this control produces a stronger signal than BiFC because it does not depend on the interaction of two proteins.
Protoplasts that were transfected with both STIC2-nYFP and ALB4-cYFP constructs produced a clear fluorescence signal that overlapped with the chlorophyll signal, as observed in the positive control (Figure 7). Neighboring untransfected protoplasts or protoplasts transfected with only one of the plasmids did not produce a fluorescence signal, although chlorophyll fluorescence was observed (Supplemental Figure 13). In the reciprocal experiment, protoplasts doubly transfected with STIC2-cYFP and ALB4-nYFP produced an identical signal. When ALB4 was replaced with the mutated form (G397S), fluorescence signals were  , and/or His-ALB3C (C). In each experiment, equal amounts of lysate containing GST-STIC2 and one of the Histagged proteins were incubated together, and then the His-tagged protein was purified using Ni-NTA resin prior to analysis by SDS-PAGE. In each gel, individually purified protein samples from the relevant lysates were included as reference controls (left side, with angled labels); the asterisks indicate a truncated form of the GST-STIC2 protein that appeared in the lysates to a variable degree. The flow-through (FT) samples contained proteins not bound by the Ni-NTA resin, whereas the imidazole elution fractions (E1-E4) contained proteins that were bound by the resin. All of the gels were stained with Coomassie blue. For all samples, 10 mL was loaded per lane. (D) Co-IP analysis. Experiments were performed with chloroplasts isolated from 12-to 14-d-old, in vitro-grown wild-type plants overexpressing either STIC1/ALB4-FLAG (left side) or STIC2-FLAG (right side). Chloroplasts were solubilized with DDM directly after isolation (left side in each case) or after treatment with 0.25 to 0.50 mM of the membrane-permeable cross-linker DSP (right side in each case) and subjected to anti-FLAG immunoprecipitation. All samples were analyzed by immunoblotting. Amounts equivalent to ;25 mg of protein for solubilized chloroplast lysate (SL) and for the co-IP flow-through (FT) were loaded, whereas one-third or one-sixth of elution (E) samples was loaded for the STIC1/ALB4-FLAG and STIC2-FLAG samples, respectively. Antisera raised against STIC1/ALB4, STIC2, ALB3, Tic40, and LHCP or FNR were used to detect the corresponding proteins. BiFC analysis of protein-protein interactions was performed by studying Arabidopsis protoplasts coexpressing proteins fused to complementary N-terminal (nYFP) and C-terminal (cYFP) fragments of the YFP protein. The images shown are representative confocal micrographs of protoplasts expressing STIC1/ ALB4 with a complete YFP tag (ALB4-YFP; this served as a positive control) or the following fusion protein pairs: STIC2-nYFP with ALB4-cYFP, also observed ( Figure 7); detection of this interaction may reflect the ability of mutant ALB4 to residually bind STIC2 ( Figure 6B) and be aided by the overexpression of both interaction partners. Alternatively, if ALB4-G397S-cYFP homooligomerizes with endogenous ALB4 or heterooligomerizes with ALB3, interaction of STIC2-nYFP with the endogenous proteins may be sufficient to result in fluorescence due to close proximity of the YFP moieties. Clear fluorescence signals were also observed for both STIC2-ALB3 combinations (STIC2-nYFP with ALB3-cYFP; STIC2-cYFP with ALB3-nYFP; Figure 7), while no signals were observed in the negative control, STIC2-nYFP with PSI-D-cYFP (Figure 7). The PSI-D-cYFP fusion is a valid negative control here, as it was shown to be expressed and able to interact with PSI-D-nYFP in chloroplasts (Figure 7; Supplemental Figure 13) (Xia et al., 1998;Kudla and Bock, 2016).
Overall, our BiFC data confirm the data obtained from the in vitro pull-down assays and provide clear in vivo support for the interaction of STIC2 with both ALB4 and ALB3.

STIC2 Can Be Cross-Linked to Both ALB4 and ALB3 in Coimmunoprecipitation Experiments
To further corroborate the positive interaction results from the in vitro pull-down and in vivo BiFC studies, a complementary approach was used: Anti-FLAG coimmunoprecipitation (co-IP) from chloroplasts isolated from wild-type plants that overexpress ALB4-FLAG or STIC2-FLAG proteins ( Figure 6D). Using freshly isolated chloroplasts solubilized with the nonionic detergent dodecyl maltoside ( Figure 6D, left side of each panel), the ALB4-FLAG and STIC2-FLAG proteins were efficiently immunoprecipitated, eluted from the anti-FLAG resin with SDS buffer, and detected with ALB4 and STIC2 antisera, respectively. However, no significant interaction between the STIC proteins was detectable, nor was there any significant interaction between the STIC proteins and ALB3 or Tic40. The photosynthetic proteins LHCP and ferredoxin-NADP reductase (FNR) used as controls in these assays were also absent from the eluates.
Our failure to detect interactions in our initial studies prompted us to repeat the experiments following treatment with the membrane-permeable chemical cross-linker dithiobis succinimidyl propionate (DSP), which can facilitate the detection of transient or weak interactions that might otherwise be lost during the purification procedures ( Figure 6D, right side of each panel). In these experiments, the control proteins LHCP and FNR were again absent from the eluted fractions, providing a clear indication of specificity. No interactions between the STIC proteins and Tic40 were detected, but on this occasion, the ALB3 protein was coimmunoprecipitated with STIC2-FLAG and with ALB4-FLAG in accordance with the results of Trösch et al. (2015) (Figure 6D). Moreover, the STIC2 protein copurified with STIC1-FLAG, and the STIC1/ALB4 protein copurified with STIC2-FLAG, in the reciprocal experiments, confirming that the two STIC proteins can indeed interact physically with each other.

DISCUSSION
Our screen for suppressors of the tic40 mutant identified two loci, corresponding to the STIC1/ALB4 gene and the previously uncharacterized STIC2 gene. Genetic analyses implied that STIC1/ALB4 and STIC2 cooperate in a common pathway ( Figure  2), a notion that was supported by the fact that the proteins interact physically with each other (Figures 6 and 7). Thus, we concluded that the two proteins act together in thylakoid membrane biogenesis, as elaborated below.

STIC1/ALB4 Is a Member of a Conserved Family of Membrane Protein Biogenesis Factors
ALB4 was discovered as a paralog of ALB3 in Arabidopsis, with both proteins belonging to the Alb3/Oxa1/YidC family of membrane insertases (Gerdes et al., 2006). Mutants of ALB3 are particularly affected in the accumulation of the light-harvesting system and display a strongly chlorotic/albino phenotype (Sundberg et al., 1997;Bellafiore et al., 2002;Asakura et al., 2008). By contrast, alb4 mutant plants show no macroscopic defects, although the thylakoids in such mutants are structurally aberrant (Figures 2 and 4) (Gerdes et al., 2006). In addition to its role in the insertion of imported LHCPs, ALB3 is also thought to participate in the targeting of cotranslationally inserted proteins (e.g., D1), interacting with ribosome nascent chains together with VIPP1 and cpSecY (Pasch et al., 2005;Walter et al., 2015). Thus, it is likely that ALB4 plays a specialized role in the insertion and/or assembly of a subset of proteins (Trösch et al., 2015).
The phenotypic differences between alb3 and alb4 mutants may reflect expression level differences between ALB3 and ALB4, functional differences between ALB3 and ALB4, and/or the operation of parallel pathways (Trösch et al., 2015). It has been suggested that ALB4 stabilizes assembly intermediates of the ATP synthase complex (Benz et al., 2009) and that it participates in the insertion or assembly of cytochrome b 6 f complex components (Trösch et al., 2015). Thus, ALB4 may have a chaperone-like function, assisting the folding and assembly of transmembrane proteins into complexes, which is another role generally attributed to the Alb3/Oxa1/YidC protein family (Wang and Dalbey, 2011).

STIC2 Is Homologous to the Bacterial Protein YbaB
The STIC2 protein and its homolog STCL were previously uncharacterized, but they share homology with the bacterial protein YbaB (Supplemental Figures 7 and 8). Although YbaB itself ALB4-G397S-cYFP, ALB3-cYFP, or PSI-D-cYFP (the latter served as a negative control); STIC2-cYFP with ALB3-nYFP; and PSI-D-nYFP with PSI-D-cYFP (which demonstrated that the negative control protein PSI-D-cYFP can be expressed and is competent to reconstitute YFP when associated with an appropriate interaction partner). For each protoplast, YFP fluorescence, chlorophyll autofluorescence, merged YFP and chlorophyll fluorescence, and bright-field images are shown. Bars representing 5 or 10 mm are shown for each set of images. is far from well characterized, it has been implicated in membrane biogenesis (Skretas and Georgiou, 2010): Coexpression of the native ybaB gene enhanced the accumulation of membraneintegrated heterologous proteins up to 10-fold in Escherichia coli. Although DNA binding activity has also been suggested for YbaB (Cooley et al., 2009;Jutras et al., 2012), the overall negative charge of STIC2 (charge at pH 7.0, -3.93; pI 4.95) and STCL (charge at pH 7.0, -1.94; pI 5.40) make this property seem unlikely. The crystal structure of YbaB shows a tweezer-like dimer formation (Lim et al., 2003), which based on homology modeling analysis is likely to be shared by STIC2 and STCL (Supplemental Figure 14) (Kelley et al., 2015).

STIC Proteins Cooperate in a Common, cpSRP-Related Thylakoid Protein Targeting Pathway
Because STIC1/ALB4 has already been assigned a function in thylakoid protein biogenesis (Trösch et al., 2015), it is likely that STIC2 participates in the same process. This notion was further supported by our phenotypic analysis of the single stic mutants. Although the stic mutants were visibly indistinguishable from the wild type, they had swollen chloroplasts with disordered (less parallel) thylakoids and an accumulation of plastoglobules (Figures 2 and 4). These phenotypes are consistent with those of the alb4-1 knockdown mutant described previously (Gerdes et al., 2006), and overall they support our hypothesis that STIC1/ALB4 and STIC2 cooperate in thylakoid membrane biogenesis.
Our data showed that STIC2 is a stromal protein lacking detectable association with the thylakoid membranes ( Figure 3). With STIC1/ALB4 being an integral thylakoid membrane protein, this makes a stable interaction between STIC1/ALB4 and STIC2 seem implausible. Indeed, the addition of a cross-linker was required to detect coimmunoprecipitation of the two proteins ( Figure 6), suggesting that the interaction is not structural but functional and of a transient nature. STIC2 potentially acts upstream of STIC1/ALB4 as a stromal sorting factor that delivers targets to the membrane. The STIC1/ALB4 C terminus can bind STIC2 in vitro, and Gly-397 of mature STIC1/ALB4 is important for this interaction in vitro ( Figure 6). In fact, Gly-397 may be essential for ALB4 function because stic1-2 (which carries the G397S substitution) has an identical tic40 suppression phenotype to nonsense mutants such as stic1-5 (Figures 1 and 2; Supplemental Figure 5). It is also pertinent that both ALB4 and STIC2 can interact with ALB3 in vitro and in vivo (Trösch et al., 2015) (Figures 6 and 7). Given that direct STIC2-ALB3 and STIC2-ALB4 interactions were both detected, one may speculate that STIC2 mediates the interaction between ALB3 and ALB4.
The interactions with ALB3 suggested a functional link between the STIC proteins and the cpSRP pathway. In support of this notion, the phenotypes of the cpsrp54 and cpftsy mutants became much more severe when combined with either stic1 or stic2 ( Figure  5). No such effects were observed when stic1 and stic2 were crossed to mutants affected in protein import (Supplemental Figure 11) or the cpSec and cpTat pathways (Figure 5), indicating that the observed synergistic effects are highly specific for the cpSRP pathway. However, the lack of physical interaction between STIC2 and LTD (Figure 6), and the absence of synergistic genetic interaction between the stic mutants and a null cpSRP43 mutant (Supplemental Figure 12), indicates that STIC2 and ALB4 do not participate in the main transport pathway for delivering LHCP proteins. Thus, STIC1/ALB4 and STIC2 may cooperate in thylakoid membrane biogenesis as part of a cpSRP43independent cpSRP pathway for a specific subset of clients or a putative alternative LHCP transport pathway (Tzvetkova-Chevolleau et al., 2007). Interestingly, STIC2 has a similar expression profile to some nucleus-encoded chloroplast ribosomal protein genes (e.g., PSRP4 and PRPL35) (Aoki et al., 2016), suggesting another possible role in the cotranslational targeting via ALB4 or ALB3. Accordingly, cpSRP54 and cpFtsY are both implicated in cotranslational insertion via ALB3, whereas cpSRP43 is not (Amin et al., 1999;Nilsson et al., 1999;Walter et al., 2015).
Regardless of their precise nature, the roles of the STIC proteins in thylakoid protein transport must be such that they can, if necessary, be compensated for by other factors, as the stic mutants do not display severe thylakoid defects. It will be necessary to identify the substrates of the STIC proteins before we can fully understand their roles and elucidate how their loss results in the suppression of tic40.

What Is the Functional Relationship between the STIC Proteins and Tic40?
It has long been assumed that Tic40 functions as part of a protein import motor, along with Tic110 and Hsp93 (Chou et al., 2003(Chou et al., , 2006Kovacheva et al., 2005). However, the absence of these components from the newly identified 1 MD TIC complex cast some doubt on this model (Kikuchi et al., 2013). The recent finding that Hsp93 may perform a protein quality control role during import, in partnership with the ClpP protease (Flores-Pérez et al., 2016), with Hsp70 performing the motor function instead (Shi and Theg, 2010;Su and Li, 2010), suggests that Tic40 may function immediately downstream of import rather than during the import process per se. This view fits well with the hypothesis that Tic40 participates in the postimport targeting of proteins to the inner envelope and thylakoids (Chiu and Li, 2008;Ouyang et al., 2011).
Assuming that Tic40 plays a key role in sorting as proteins exit the import process, its loss may be accompanied not only by the disruption of import (Chou et al., 2003;Kovacheva et al., 2005), but also by the mistargeting of certain proteins. For example, proteins normally targeted to the inner envelope or thylakoids may enter noncognate targeting pathways with nonproductive outcomes in the absence of Tic40. If such mistargeting leads to thylakoid membranes that are held in close physical proximity to the TIC channel, further blockage of import may result. It is conceivable that STIC1/ALB4 and STIC2 act in such a noncognate pathway and that their loss causes a reduction in mistargeting events leading to partial suppression of the deleterious effects of tic40 on protein transport and plant development.
A Link between Chloroplast Swelling, Plastoglobule Accumulation, and Loss of Membrane Integrity?
Intriguingly, the swollen appearance of the tic40 chloroplasts and the accumulation of plastoglobules are largely unaffected in the suppressor mutants (Figure 1), and these are particular characteristics that tic40 happens to share with both stic single mutants ( Figure 4) and which are not generally seen in other importdefective mutants (e.g., ppi1) (Jarvis et al., 1998). This phenotype is very similar to one observed upon moderate heat stress (40°C) treatment of Arabidopsis leaves, which is partially reversed after only 30 min of recovery, suggesting that the effect is primarily physicochemical in nature (Zhang et al., 2010). Moderate heat stress may cause increased membrane fluidity and, thus, greater proton leakage into the stroma, leading to acidification, osmotic stress, and organelle swelling.
Interestingly, a vipp1 mutant was also reported to display a swollen chloroplast phenotype with disorganized, "ruffled" thylakoids (the VIPP1 protein, like its bacterial ortholog PspA, plays an important role in membrane maintenance) Zhang et al., 2012). In this mutant, spherical chloroplast morphology was linked to increased osmotic pressure inside the organelles, as the chloroplasts could be returned to their normal elliptical shape by increasing the osmotic concentration of the cytosol (Zhang et al., 2012). Again, the hypertonic state of the chloroplasts was attributed to membrane leakiness. Thus, it is conceivable that the chloroplast swelling seen in tic40 occurs due to defects in inner envelope and/or thylakoid membrane integrity, while the less drastic swollen chloroplast phenotype seen in the stic single mutants may be associated with an effect on the thylakoid membranes only. The latter is supported by data showing reduced photosynthetic performance in alb4 mutants (Benz et al., 2009;Trösch et al., 2015).
Plastoglobules are structures formed on thylakoid membranes that contain lipids (carotenoids, plastoquinone, and tocopherols) as well as lipid biosynthesis enzymes such as tocopherol cyclase (Steinmüller and Tevini, 1985;Austin et al., 2006). They are commonly observed in mutants with defects in thylakoid membrane biogenesis (e.g., var2) and following various biotic and abiotic stresses (Takechi et al., 2000;Bréhélin et al., 2007;Bréhélin and Kessler, 2008), and may be a mechanism to temporarily store lipids (e.g., tocopherols) in order to protect the thylakoid membranes from oxidative damage. Thus, the plastoglobule accumulation seen in the tic40 and stic mutants is consistent with our hypothesis that these mutants experience membrane instability and leakage.
Tocopherol biosynthesis is initiated in the inner envelope membrane (Froehlich et al., 2003;Motohashi et al., 2003), so plastoglobules were suggested to form a metabolic link between the envelope and thylakoids (Ytterberg et al., 2006). Our observation that stic1 and stic2 (mutants that result in thylakoid biogenesis defects) can suppress tic40 reveals a further link between the two membrane systems and sheds new light on the interconnectedness of protein import and thylakoid protein biogenesis. We expect that the stic mutants will provide an invaluable resource for further investigations on how these crucial processes are coordinated in the future.
The only instance where another ecotype was used was in the generation of hybrid mapping populations, where the stic tic40-4 mutants were crossed with Landsberg erecta plants in which the tic40-4 mutation had been introgressed; this was done by sequential backcrossing (10 times) of the original tic40-4 Col-0 mutant with Landsberg erecta ecotype plants.
In vitro-grown plants were generally grown on Murashige and Skoog (MS) plant growth media (0.5-13 MS salts, 0.5% [w/v] sucrose, 0.05% [w/v] MES-KOH pH 5.7, and 0.6% [w/v] phytoagar). However, very chlorotic mutant seedlings (cpsecA-1, hcf106, and cpftsY-1/2) were usually transplanted to 3% (w/v) sucrose MS media, after ;10 d on normal media, for further growth. Plants on MS media were grown in Percival growth chambers with ;100 mmol m 22 s 21 fluorescent tube lighting (Philips) at 20°C under long-day (16 h light, 8 h dark) conditions. Seeds were sterilized by shaking for 5 min in 70% (v/v) ethanol containing 0.05% (v/v) Triton X-100, followed by shaking for 10 min in 100% ethanol, and then air-dried on sterile filter paper prior to sowing on MS media. Seeds on MS media were stratified for 2 d at 4°C before being transferred to the growth chamber. Normally, plants were transferred to soil after ;10 d of growth on MS media and kept either in growth cabinets, controlled environment rooms, or greenhouses under long-day (16 h light, 8 h dark) conditions. Chlorophyll measurements were performed using a SPAD-502 meter (Konica Minolta) using soil-grown plants as described previously .
When necessary, the following antibiotics were included in the medium at the indicated concentrations: 50 mg/mL of kanamycin monosulfate (Melford Laboratories) was used to select for the SALK T-DNA insertion lines; 10 mg/mL of DL-phosphinothricin (Duchefa) was used to select for SAIL and WiscDsLox lines; and 15 mg/mL of hygromycin B (Duchefa) was used for transgenics generated with the pH2GW7-FLAG plasmid.

Characterization of the stic T-DNA Mutants
To confirm the position of the T-DNA insertion in each stic2 T-DNA mutant (stic2-3 and stic2-4), the upstream flanking sequence of the insertion in each case was amplified using primers specific for the T-DNA left border (LB) and for the STIC2 gene. In the case of stic2-3, primer SALK LBb1 (59-GCGTGGACCGCTTGCTGCAACT-39) was used in combination with the forward primer Stic2 Pro F (59-CCTTTGTTAGGTCATGAC-39), which binds upstream of the STIC2 coding sequence (CDS; in the putative promoter region). In the case of stic2-4, the LB-specific primer WiscDsLox LB (p745) (59-AACGTCCGCAATGTGTTATTAAGTTGTC-39) was used in combination with the forward primer T29E15.22 F1 (59-GCTTCCGGTTTTA-TCTTCTCG-39). The PCR products obtained were cloned and sequenced to obtain precise positions of insertion ( Figure 2B). For the stic2-4 T-DNA mutant, the downstream T-DNA flanking sequence was also amplified with the LB-specific primer in combination with . Therefore, in this case, the LB is found on each side of the insertion site, indicating that the plant contains an inverted repeat T-DNA insertion. However, for the stic2-3 T-DNA mutant, we failed to amplify the downstream flanking T-DNA sequence with either RB or LB primers, indicating that in this case the inserted T-DNA is likely truncated. For routine genotyping, the initial primer combinations were used to detect the insertion, and in each case the forward primer was used with the reverse primer T29E15.22 R3 to detect the wild-type STIC2 gene. We proceeded to confirm that these T-DNA insertion lines segregated 3:1 for kanamycin (stic2-3) or phosphinothricin (stic2-4) resistance to ensure that they each contained just a single T-DNA insertion.

Constructs and Vectors Used for in Vivo Studies
The CDSs of ALB3, STIC1/ALB4, STIC2, and STCL were amplified from wild-type cDNA (ALB3 and STIC1/ALB4) or obtained from the ABRC (cDNA clones: NM_001202657, STIC2; NM_119208, STCL). They were amplified to engineer AttB1 and AttB2 Gateway recombination sites at their 59 and 39 ends, respectively, without a stop codon, and were then cloned into the pDONR201 Gateway entry vector using Gateway technology (Invitrogen, ThermoFisher).
The CDS PCR products were further amplified with the AttB1/B2 primer pair (AttB1, 59-GGGGACAAGTTTGTACAAAAAGCAGGCT-39; AttB2, 59-GGGGACCACTTTGTACAAGAAAGCTGGGT-39) to add the full AttB recombination sequences prior to cloning. The ALB3, ALB4, STIC2, and STCL CDS inserts were then subcloned into the C-terminal YFP tag vector p2GWY7 (Karimi et al., 2002) to produce the ALB3-YFP, ALB4-YFP, STIC2-YFP, and STCL-YFP constructs, respectively. Also, the ALB4-CDS, STIC2-CDS, and STCL-CDS-STOP inserts were cloned into the C-terminal FLAG tag vector pH2GW7-FLAG, a vector derived from the original pH2GW7 vector (Karimi et al., 2002(Karimi et al., , 2005; the pH2GW7-FLAG vector was built by replacing the original Gateway cassette and 35S tobacco mosaic virus terminator of pH2GW7 with the Gateway cassette followed by a C-terminal FLAG tag and the octopine synthase terminator from the pEarleyGate302 (Earley et al., 2006). The YFP constructs were used for transient expression in protoplasts, while the FLAG constructs were used for stable, Agrobacterium tumefaciens-mediated transformation of wild-type and stic1-1 tic40-4 (with ALB4-FLAG) or stic2-1 tic40-4 (with STIC2-FLAG or STCL-CDS-STOP) plants. Arabidopsis plants were transformed by Agrobacterium infiltration using the previously described floral dip method (Clough and Bent, 1998). The Tic110-YFP construct used as controls for protoplast transfection in the localization studies (Figure 3) was described previously (Bédard et al., 2007).

Recombinant Protein Expression and STIC2 Antibody Production
The vector used here to express the soluble, 154-residue ALB4 C-terminal domain with an N-terminal 6xHis tag (His-ALB4C) was previously described and used to produce the ALB4 antibody (Trösch et al., 2015). The His-ALB4C-G397C expression vector is essentially equivalent, except that the ALB4 coding sequence was amplified from stic1-2 mutant cDNA and carries the point mutation resulting in the G397C substitution. The His-ALB3C clone was produced by amplifying the coding sequence for the C-terminal 102 residues of ALB3 from wild-type cDNA using the following primers: ALB3-His-F (59-AAGGATCCAATATGGATGGATGAAAACG-CAAGC-39) + ALB3-His-R (59-TCGTCGACCTATACAGTGCGTTTCCGCT-39). The obtained PCR fragment was digested with BamHI and SalI and cloned into the pQE-30 expression vector (Qiagen). The His-ALB4C, His-ALB4C-G397S, and His-ALB3C fusions were all expressed in XL1-Blue Escherichia coli cells (Agilent Technologies) grown in medium containing 0.4% (w/v) glucose followed by induction with 1 mM isopropyl b-D-1-thiogalactopyranoside (IPTG). The construct used to produce His-LTD (in the pET28a vector) was previously described (Ouyang et al., 2011).
The CDS encoding the predicted mature form of STIC2 (i.e., amino acids 49 to 182) was amplified using the primers STIC2-GST-2-F (59-AAG-GATCCGTGAATGGATTATTTG-39) and STIC2-C-SalI-R (59-AAGTC-GACCCTTCATTCCTTCGCTG-39) and transferred to the pGEX-6P1 as a BamHI-SalI fragment to produce the GST-STIC2 fusion. The GST-STIC2 construct was expressed in BL21 E. coli cells and induced with 0.2 mM IPTG, and the protein was purified from bacterial lysate produced by sonication in lysis buffer (50 mM NaH 2 PO 4 and 300 mM NaCl, pH 8.0). Lysate was applied to polypropylene columns containing glutathione agarose resin (Sigma-Aldrich), before washing with wash buffer (50 mM NaH 2 PO 4 and 300 mM NaCl, pH 8.0) and protein elution by cleavage overnight at 4°C with PreScission protease (GE Healthcare) in cleavage buffer (50 mM Tris-HCl, pH 7.0, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, and 0.01% [v/v] Triton X-100). The purified protein was sent to Harlan Sera-lab for antibody production in rabbits, and the antiserum was affinity purified using the original antigen.

Seedling Protein Extractions and Immunoblotting
Total protein extracts were prepared using a modified, previously described procedure (Kovacheva et al., 2005). Briefly, 100 mg of 10-to 14-dold seedlings was collected in a 2-mL Eppendorf tube, frozen in liquid nitrogen, and homogenized with a 5-mm stainless steel bead using a TissueLyser (Qiagen) by shaking at 12,000 rpm for 1 min. After pelleting the resulting tissue powder (pulse spin at 20,000g and 4°C), 0.2 mL protein extraction buffer (100 mM Tris-HCl, pH 6.8, 10% [v/v] glycerol, 2% [w/v] SDS, 1% [v/v] Triton X-100, 100 mM DTT, and 13 plant protease inhibitor cocktail [Sigma-Aldrich]) was added before mixing at 1400 rpm and 4°C for 20 to 30 min in a Thermomixer (Eppendorf). After spinning down the insoluble material (10 min at 20,000g), the supernatant was transferred to a new tube and a Bradford assay kit (Bio-Rad) was used to measure the protein concentration. Samples of equal protein concentration were prepared and used for analysis. SDS-PAGE and immunoblotting using PVDF (Millipore) membranes were performed employing standard procedures.

Transmission Electron Microscopy
For transmission electron microscopy, the cotyledons from 10-d-old plants were used. The tissue was prepared and imaged with the help of the University of Leicester Electron Microscope Laboratory (Faculty of Medicine and Biological Sciences). Cotyledons were excised from in vitrogrown seedlings and fixed overnight in 4% (v/v) glutaraldehyde/2% (v/v) formaldehyde in 0.1 M sodium cacodylate buffer with 2 mM calcium chloride (pH 7.2) and then washed in the same cacodylate/CaCl 2 buffer. The tissue was then fixed with 1% (w/v) OsO 4 /1.5% (w/v) potassium ferricyanide for 3 h, washed with distilled, deionized water, and finally tertiaryfixed with 2% (w/v) uranyl acetate for 1 h. The tissue was serially dehydrated through ethanol and propylene oxide and embedded in Spurr's modified low viscosity resin. Thin sections of ;80 nm thickness were cut using a Reichert Ultracut S ultramicrotome, collected onto copper mesh grids, and stained with Reynolds' lead citrate. The grids were viewed under a JEOL JEM-1400 electron microscope at 80 kV. Digital images were recorded using a SIS digital camera and iTEM software (Glauert and Lewis, 1998;Hyman and Jarvis, 2011).

Protoplast Transfection and Microscopy for Localization and BiFC Interaction Assays
Protoplasts from 4-to 5-week-old wild-type plants grown on soil were isolated using the tape Arabidopsis-sandwich method (Wu et al., 2009). Approximately10 5 protoplasts and 5 mg of plasmid DNA were used per (co)transfection in 40% (w/v) PEG-4000 solution (Wu et al., 2009). Samples were analyzed 16 to 18 h after (co)transfection using a Zeiss LSM 510 META confocal laser scanning microscope and a C-Apochromat 403/1.2-W Corr. objective. To detect YFP, 514-nm excitation from a 5-mW argon ion laser with an HFT 458/514 primary dichroic mirror and a 535-to 590-nm emission filter was used. To simultaneously detect chlorophyll autofluorescence, an NFT 635 visual long-pass filter was used. Images were processed with Zeiss LSM Image Browser software. Each (co)transfection was conducted three times, with the same result, and typical images are shown.

Chloroplast Isolation, Subfractionation, and Import Assays
Chloroplasts were isolated from 10-to 14-d-old in vitro-grown Arabidopsis seedlings as previously described (Aronsson and Jarvis, 2011). Chloroplasts were lysed hypotonically and subfractionated into soluble (stroma and intermembrane space), envelope, and thylakoid fractions by sucrose density gradient centrifugation as previously described (Flores-Pérez and Jarvis, 2017). In vitro chloroplast protein import assays were performed as described previously (Aronsson and Jarvis, 2011).

Co-IP Experiments
Chloroplasts were isolated from 14-d-old, in vitro-grown wild-type and STIC1-FLAG-or STIC2-FLAG-expressing seedlings. Amounts of chloroplasts equivalent to 0.4 mg of chlorophyll were cross-linked with 250 or 500 mM DSP (or mock-treated without DSP) on ice for 15 min in HMS buffer (50 mM HEPES-NaOH, pH 8.0, 3 mM MgSO 4 , and 0.33 M sorbitol). The DSP was then quenched by adding glycine (in HMS) to a final concentration of 50 mM, and the chloroplasts were further incubated on ice for 30 min after mixing. Chloroplast samples were split into two 0.2 mg chlorophyll aliquots, pelleted by spinning for 1 min at 10,000g, and frozen in liquid nitrogen. Chloroplast pellets of 0.2 mg chlorophyll were solubilized in 1.5 mL dodecyl maltoside (DDM) solubilization buffer (SB) (50 mM Tris-HCl, pH 7.4, 1% [w/v] DDM, 150 mM NaCl, 1 mM EDTA, 10% [v/v] glycerol, and 13 plant protease inhibitor cocktail [Sigma-Aldrich]) by rotating at 4°C for 1 h. The insoluble material was removed by centrifugation at 20,000g for 10 min at 4°C, and then the supernatants were mixed with 25 mL of preequilibrated anti-FLAG resin (Sigma-Aldrich) in a micro-spin column (Bio-Rad) by rotation for 2 h at 4°C. The resin was then pelleted by centrifugation at 500g for 1 min, and the flow-through was collected. The resin was washed twice with 0.5 mL of SB, and six times with 0.5 mL of wash buffer (SB with 0.1% [w/v] DDM instead of 1% [w/v] DDM). Bound proteins were eluted using 98 mL of SDS sample buffer without DTT (100 mM Tris-HCl, pH 6.8, 10% [v/v] glycerol, 2% [w/v] SDS, 1% [v/v] Triton X-100, 0.05% [w/v] bromophenol blue dye, and 13 plant protease inhibitor cocktail [Sigma-Aldrich]) by heating at 90°C and shaking at 600 rpm in a Thermomixer for 10 min. The eluate was then supplemented with 2 mL of 1 M DTT.

In Vitro Pull-Down Experiments
Recombinant His-tagged and GST-tagged proteins were expressed in 50 mL bacterial cultures induced with 0.2 to 1.0 mM IPTG. Bacterial lysates were prepared in 5 mL lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM imidazole, 0.01% [v/v] Triton X-100, 0.05 mg/mL lysozyme, and 13 Complete EDTA-free protease inhibitor cocktail [Roche Diagnostics]) from frozen bacterial pellets. Equal amounts (0.5 mL) of bacterial lysates from His-tag and GST-tag expressors were combined and mixed with 25 to 50 mL of Ni-NTA resin (Qiagen) in a micro-spin column (Bio-Rad) for 2 h by rotation at 4°C. The resin was pelleted by centrifugation at 500g for 1 min, and the flow-through was collected. The resin was then washed four to five times with 0.5 mL wash buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 20 mM imidazole). Bound proteins were eluted by washing four times with 50 mL of elution buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 250 mM imidazole, and 10% [v/v] glycerol), and each eluate was mixed with 15 mL 53 SDS sample buffer (500 mM Tris-HCl, pH 6.8, 50% [v/v] glycerol, 10% [w/v] SDS, 5% [v/v] Triton X-100, and 500 mM DTT). After electrophoresis, stained bands were quantified using ImageJ software .

Accession Numbers
The following gene codes can be used to access the full-length gene, cDNA, CDS, and protein sequences from TAIR: At1g24490 (STIC1/ALB4), At2g24020 (STIC2), and At4g30620 (STCL). For protein alignments, we used the STIC2 (NP_001189586) and STCL (NP_194791) sequences from the GenBank database. The STIC2 homologs used in the phylogenetic analysis are listed in Supplemental Data Set 2. Supplemental Figure 9. Analysis of stic2 stcl double mutants generated using a STCL amiRNA construct.

Supplemental
Supplemental Figure 10. Analysis of the localization of ALB4 at different stages of development.
Supplemental Figure 11. Specificity of the suppression effects of the stic1 and stic2 mutations.
Supplemental Figure 12. Immunoblot analysis of the cpSRP pathway mutants and phenotypes of the stic cpsrp43 double mutants.
Supplemental Figure 13. Control images for the in vivo analysis of interactions between STIC2 and the STIC1/ALB4 and ALB3 proteins.
Supplemental Figure 14. Structural models showing surface electrostatic potential of STIC2 and STCL relative to the E. coli homolog YbaB.

Supplemental Methods.
Supplemental Data Set 1. Alignment of the sequences of the STIC2/ STCL-type proteins used in the phylogenetic analysis.
Supplemental Data Set 2. Sequences of STIC2/STCL-type proteins used in the phylogenetic analysis.