Selective autophagy regulates chloroplast protein import and promotes plant stress tolerance

Abstract

Chloroplasts are plant organelles responsible for photosynthesis and environmental sensing. Most chloroplast proteins are imported from the cytosol through the translocon at the outer envelope membrane of chloroplasts (TOC). Previous work has shown that TOC components are regulated by the ubiquitin‐proteasome system (UPS) to control the chloroplast proteome, which is crucial for the organelle's function and plant development. Here, we demonstrate that the TOC apparatus is also subject to K63‐linked polyubiquitination and regulation by selective autophagy, potentially promoting plant stress tolerance. We identify NBR1 as a selective autophagy adaptor targeting TOC components, and mediating their relocation into vacuoles for autophagic degradation. Such selective autophagy is shown to control TOC protein levels and chloroplast protein import and to influence photosynthetic activity as well as tolerance to UV‐B irradiation and heat stress in Arabidopsis plants. These findings uncover the vital role of selective autophagy in the proteolytic regulation of specific chloroplast proteins, and how dynamic control of chloroplast protein import is critically important for plants to cope with challenging environments.

Heat and UV light stress promotes the NBR1‐mediated autophagy of specific chloroplast import apparatus components in Arabidopsis.

Introduction

Chloroplasts (and other plastids) are distinctive organelles in plants and algae which evolved through endosymbiosis from a cyanobacterial ancestor over 1 billion years ago (Jarvis & López‐Juez, 2013). Chloroplasts are responsible for photosynthesis, arguably the most important biological process, which converts CO₂ into organic matter. By doing so, they not only provide a food source for life on Earth but also release O₂ for respiration and control atmospheric CO₂ levels. In addition, chloroplasts have versatile functions, such as fatty acid and amino acid production and delivering responses to environmental cues (Rolland et al, 2012; Schwenkert et al, 2022). The establishment and maintenance of chloroplast functions are linked to the control of the organellar proteome.

The chloroplast proteome consists of about 3,000 proteins. Although the chloroplast genome does encode some of these proteins, over 90% of chloroplast proteins are encoded in the nucleus. They are synthesized in the cytosol in precursor form, and thus must be imported into chloroplasts posttranslationally and processed, in order to take on their final conformation and functions (Li & Teng, 2013). Protein import is mediated by two multiprotein complexes in the outer and inner membranes, called TOC and TIC (Translocon at the Outer/Inner envelope membrane of Chloroplasts), respectively (Shi & Theg, 2013; Richardson et al, 2017; Rochaix, 2022). Protein import initiates at the TOC complex, which comprises three key components: two GTPase‐domain receptors, Toc34 and Toc159, and one Omp85 (Outer membrane protein, 85 kD)‐related membrane channel, Toc75 (Demarsy et al, 2014).

Accumulating evidence suggests that chloroplast protein import is not a constitutive process, but, instead, can be rapidly regulated by the ubiquitin‐proteasome system (UPS). Such regulation is crucial to chloroplast biogenesis and operation in response to developmental and environmental cues (Ling & Jarvis, 2015a). The UPS is a major eukaryotic protein degradation system that implements the ubiquitin signal to turnover specific target proteins (Vierstra, 2009). Genetic analysis revealed that TOC components can be directly targeted for degradation by the proteasome, mediated by a chloroplast‐localized ubiquitin E3 ligase, SP1 (Ling et al, 2012). Recently, SP1 was shown to act in a novel pathway controlling chloroplast protein stability, named chloroplast‐associated protein degradation (CHLORAD) (Ling et al, 2019). CHLORAD regulates protein import to optimize the organellar proteome, and is thereby important for the interconversions that chloroplasts and other plastid types undergo (for example, during de‐etiolation and fruit ripening; Ling et al, 2012, 2021) and for abiotic stress tolerance (Ling & Jarvis, 2015b). Apart from CHLORAD, cytosolic ubiquitin‐dependent systems can also govern protein import, by regulating precursor stability (Lee et al, 2009; Grimmer et al, 2020), or by controlling the abundance of the Toc159 receptor before its entrance into the outer envelope membrane (Shanmugabalaji et al, 2018).

Autophagy is another conserved eukaryotic proteolytic pathway, and it is critical for responses to various forms of stress, such as nutrient deprivation or abiotic stress (Avila‐Ospina et al, 2014; Marshall & Vierstra, 2018; Chen et al, 2021). Cargos to be degraded by autophagy are sequestered into double‐membrane vesicles called autophagosomes, and then delivered to lysosomes or vacuoles as autophagic bodies for recycling. Autophagy‐related (ATG) proteins are essential for the formation of autophagosomes (Nakatogawa, 2020). Although autophagy was long thought to be responsible for non‐selective bulk degradation, it is clear now that autophagy can selectively eliminate unwanted or damaged proteins or organelles (Dikic & Elazar, 2018; Palikaras et al, 2018; Stephani & Dagdas, 2020; Nakamura et al, 2021). The most prevalent selective autophagy pathway requires the labelling of the substrate with specific signals (such as ubiquitin) that are recognized by autophagy adaptors, which guide the cargo to the autophagic membrane via their ATG8‐interaction motif (AIM) or ubiquitin‐interacting motif (UIM) (Kirkin & Rogov, 2019).

Compared with organelles in yeast and mammals, chloroplasts have been studied to a much lesser extent regarding autophagy. Given that they are unique to plants and algae, distinctive autophagy pathway components are expected to operate in chloroplasts. Two main types of chloroplast autophagy (or called chlorophagy) have been revealed: bulk degradation of Rubisco‐containing bodies during dark‐induced senescence (Ishida et al, 2008; Wada et al, 2009), and degradation of whole photodamaged chloroplasts (Izumi et al, 2017; Nakamura et al, 2018). However, little is known about the roles of selective autophagy in regulating chloroplast proteins (Otegui, 2018), and selective chlorophagy receptors or adaptors have not been identified (Kirkin & Rogov, 2019; Wan & Ling, 2022), with the exception of ATI1, a potential receptor for chloroplast degradation (Avin‐Wittenberg et al, 2012; Michaeli et al, 2014). Moreover, the role of chloroplast autophagy in various abiotic stress, such as heat and UVB, is poorly understood.

To improve understanding of the ubiquitin‐dependent proteolytic pathways acting on chloroplasts, here, we analyzed the effect of autophagy impairment on the stability of specific chloroplast proteins. We showed that protein import is regulated through selective degradation of TOC components by autophagy, in a process that is distinct from CHLORAD. We identified a selective autophagy adaptor, NBR1, and demonstrated that NBR1 mediates the degradation of the TOC complex in response to heat and UVB stresses. Collectively, our results reveal a novel mechanism of chloroplast protein quality control involving selective autophagy.

Results

TOC components are degraded by the autophagy pathway

Although chloroplasts were discovered as the target of autophagy more than a decade ago, the molecular mechanisms that govern this process have largely remained elusive. It is known that protein ubiquitination serves not only as a targeting signal for the UPS but also as a targeting signal for selective autophagy pathways (Kwon & Ciechanover, 2017). Our previous finding that TOC components can be ubiquitinated suggests a possible signal for the selective autophagy of chloroplast proteins. To shed light in this area, we performed a series of cellular, molecular and biochemical analyses.

First, to determine whether TOC components are autophagy cargo proteins, we transiently expressed yellow fluorescent protein (YFP)‐tagged Toc33, Toc159 and Toc75 in Arabidopsis protoplasts. The YFP tag was fused to Toc33 and Toc159 at their N termini, and inserted in‐between the transit peptide and the N terminus of the Toc75 mature sequence (Chen et al, 2016; Paila et al, 2016), so as not to interfere with proper membrane integration. In parallel, SP1–YFP and FAX1 (fatty acid export 1)‐YFP were expressed in protoplasts as control proteins; FAX1 is a chloroplast envelope protein involved in fatty acid metabolism. Both SP1 and FAX1 are CHLORAD substrates like TOC proteins (Sun et al, 2022), and previously we demonstrated that SP1 is not subjected to autophagic degradation (Ling et al, 2019). Protoplasts expressing all constructs were treated with concanamycin A (ConA), an inhibitor of tonoplast H⁺‐ATPase that blocks vacuolar proteolysis by increasing luminal pH (Nolan et al, 2017). As protoplastation and transfection cause abiotic stress (Ling & Jarvis, 2015b), autophagy is expected to be induced under such conditions. Indeed, treatment of cells expressing YFP‐tagged TOC proteins with ConA resulted in the accumulation of randomly moving YFP‐labeled puncta, which we interpret to be vacuolar autophagic bodies, in addition to the signals at chloroplasts; whereas treatment of cells expressing SP1‐YFP and FAX1‐YFP exhibited only signals around the chloroplasts (Fig 1A and B; Movie EV1). In the absence of ConA treatment, vacuolar YFP puncta were largely absent for all constructs (Fig EV1), supporting our interpretation of autophagic degradation of TOC proteins in the vacuole. As the YFP‐labeled puncta are similar or even smaller in size than typical autophagosomes, we infer that TOC proteins are most likely degraded by a macroautophagy pathway (Nakamura et al, 2018). To further confirm that TOC proteins are degraded by autophagy, we coexpressed YFP–Toc33, YFP–Toc159 or YFP–Toc75 together with the autophagosome marker CFP–ATG8 (ATG8a) in protoplasts, with ConA treatment. Confocal microscopy analyses demonstrated that ATG8 (blue puncta) and TOC proteins (green puncta) co‐localized in the vacuole (Fig 1C; Movie EV2), suggesting that turnover of TOC proteins is indeed associated with ATG8‐decorated autophagic particles.

Figure 1. Components of the TOC complex are selectively processed by autophagy.

• A, B
  Presence of TOC proteins in the central vacuole of protoplasts. Arabidopsis protoplasts expressing the YFP‐tagged chloroplast proteins were analyzed by confocal microscopy following ConA treatment for 4 hours. Representative images are shown (A). Chlorophyll autofluorescence was used to determine the localization of the YFP fluorescence signals relative to the chloroplasts. Signals for SP1–YFP and FAX1–YFP, which here are control proteins, were only present at the chloroplast envelope. Accumulation of YFP puncta in the central vacuole is indicated by the arrowheads. Scale bar, 10 μm. In addition, the numbers of YFP puncta in the central vacuole per protoplast were quantified (B). All values are means ± SEM from three experiments.
• C
  ATG8 and TOC proteins colocalized in the central vacuole of protoplasts. Arabidopsis protoplasts coexpressing the CFP–ATG8 and YFP‐tagged proteins were analyzed by confocal microscopy following ConA treatment for 4 hours. Representative images are shown. Accumulation of punctate bodies of overlapping fluorescence is indicated by the arrowheads. Scale bars: 10 μm.
• D
  Presence of TOC proteins in the central vacuole of Arabidopsis leaf and hypocotyl cells. Eight‐day‐old transgenic Arabidopsis seedlings stably expressing pToc33:YFP‐Toc33 or pToc75:YFP‐Toc75 were treated with ConA for 2 days, and then analyzed by confocal microscopy. Accumulation of YFP puncta in the central vacuole is indicated by the arrowheads. Scale bar, 10 μm.

Source data are available online for this figure.

Figure EV1. Subcellular localization analysis under control conditions related to Fig 1 .

Arabidopsis protoplasts expressing the YFP‐tagged chloroplast proteins were analyzed by confocal microscopy, performed as described for Fig 1A except that the protoplasts were treated by dimethylsulphoxide (DMSO) for 4 h, as a control for the ConA treatment. Representative images are shown. Chlorophyll autofluorescence was used to determine the localization of the YFP fluorescence signals relative to the chloroplasts. Overlay of the YFP images on matching chlorophyll images (Merge) clearly indicated that the YFP signals of Toc33, Toc75 and Toc159 were localized around the periphery of the chloroplasts, a pattern which is indicative of envelope localization (same as the SP1 and FAX1 controls). In contrast with the controls, YFP fused TOC proteins also presented in the central vacuole of protoplasts treated with tonoplast ATPase inhibitor, ConA (Fig 1A), suggesting that TOC complex is degraded by autophagy. Scale bar, 10 μm.Source data are available online for this figure.

To verify the above results, we also generated stable transgenic lines expressing YFP–Toc33 (pToc33:YFP–Toc33) and YFP–Toc75 (pToc75:YFP–Toc75) fusions under the control of the respective native promoters. By quantifying the accumulation of YFP–Toc33 and YFP–Toc75 in the transgenic lines, we showed that the levels of these YFP‐tagged proteins are similar to those of the endogenous proteins (Appendix Fig S1A–D). Both lines were grown in the dark with ConA treatment for 2 days to induce autophagy. In both cases, the accumulation of YFP‐labeled puncta in the vacuole was evident, confirming the turnover of TOC proteins in the vacuole (Fig 1D; Appendix Fig S2A and B, and Movie EV3). To directly determine whether the detection of YFP–TOC puncta was linked to autophagy, we generated Arabidopsis atg7‐2 autophagy mutant plants that stably express the pToc33:YFP‐Toc33 or pToc75:YFP‐Toc75 transgenes. ATG7 is the E1 enzyme for ATG8 activation, and the atg7 mutant was reported to compromise the process of autophagosome formation (Shin et al, 2014; Spitzer et al, 2015). In the atg7‐2 mutant, treated as described for Fig 1D, no obvious YFP‐labeled puncta could be observed in the vacuole (Figs 2A and EV2), consistent with the notion that TOC proteins were autophagic cargos.

Figure 2. Degradation of the TOC complex is affected upon autophagy inhibition.

• A
  Lack of TOC proteins in the central vacuole of atg7‐2 mutant hypocotyl cells. Eight‐day‐old transgenic Arabidopsis seedlings stably expressing pToc33:YFP–Toc33 and pToc75:YFP–Toc75 in the atg7‐2 mutant background were treated with ConA for 2 days, and then analyzed by confocal microscopy. Note that YFP–Toc33 and YFP–Toc75 bodies (arrowheads) were retained at plastid bridges and extensions, and not released into the vacuole, as occurs in the WT background (see Fig 1D, which acts as a positive control for this experiment). Scale bar, 10 μm.
• B, C
  Autophagy inhibition affected accumulation of TOC proteins. Eight‐day‐old wild type (WT) and atg7‐2 plants were treated respectively with proteasome inhibitor, bortezomib (Btz), cysteine protease inhibitor, E64d, or DMSO (Mock) for 1 day. Total protein extracts from these plants were analyzed by immunoblotting using antibodies as indicated. Actin acted as a loading control. Typical immunoblotting results are shown (B). Band intensities were quantified and normalized to equivalent data for Actin (C). Data are means ± SEM from three biological replicates. In (B), the arrowhead indicates full‐length Toc159, and the hollow arrow indicates a truncated form of Toc159. Molecular weight markers (sizes in kD) are indicated to the right of the images. Asterisks in (C) indicate significance according to an unpaired two‐tailed Student's t‐test (*significant at P < 0.05; **significant at P < 0.01).

Source data are available online for this figure.

Figure EV2. Quantification of YFP puncta in the vacuole related to Fig 2 .

Hypocotyl cells of transgenic Arabidopsis seedlings stably expressing pToc33:YFP–Toc33 or pToc75:YFP–Toc75 in the wild type and atg7‐2 mutant backgrounds were analyzed by confocal microscopy, performed as described for Figs 1D and 2A. The numbers of YFP puncta in the central vacuole per frame (4,000 μm² of hypocotyl section) were quantified. All the values are means ± SEM from three experiments.Source data are available online for this figure.

We next analyzed native TOC protein levels in wild‐type and atg7‐2 plants, grown in the dark and treated with or without bortezomib (a proteasome inhibitor) or E64d (a cysteine protease inhibitor, which allows us to check accumulation of autophagic cargo in the vacuole, like ConA) (Nolan et al, 2017; Ling et al, 2019). As expected, accumulation of all main TOC proteins was significantly elevated after bortezomib treatment (Student's t‐test, P < 0.05), since they are known UPS substrates (Ling et al, 2012; Fig 2B and C). Notably, TOC protein levels were similarly increased after E64d treatment, or in the atg7 mutant compared with the wild‐type plants (Fig 2B and C). In contrast, a component of the TIC complex, Tic110, was stable in both genotypes and under all treatments, suggesting that TOC proteins are selective autophagy cargos. The influence of autophagy on TOC proteins was further confirmed by ConA treatment, which had a similar effect to E64d (Appendix Fig S3A and B). In the atg7‐2 background or after E64d treatment, even higher TOC protein levels could be detected after bortezomib treatment, whereas the additivity of increased TOC protein levels was absent in atg7‐2 mutant treated with E64d (Fig 2B and C; Appendix Fig S3A and B), supporting the notion that the UPS and autophagy are complementary pathways controlling TOC components. These changes were not attributable to regulation at the mRNA level, because TOC transcript levels were comparable in the different genotypes and treatments (Appendix Fig S4A and B).

Identification of NBR1 as an autophagic adaptor for chloroplasts

To shed light on the molecular mechanism for the recognition of chloroplast cargos in selective autophagy, we sought receptors or adaptors of chloroplast proteins by screening for ATG8 interactors in chloroplasts. We first checked the proportion of cellular ATG8 protein localized in chloroplasts. Constructs encoding YFP–ATG8 or free YFP were transiently expressed in Arabidopsis protoplasts for confocal microscopy analysis. In contrast to free YFP, which only showed diffuse cytosolic localization, YFP‐ATG8 also presented puncta around the periphery of chloroplasts in a pattern indicative of envelope localization (Fig 3A, upper panels). Chloroplast localization of YFP‐ATG8 was clearer when the organelles were isolated from the protoplasts, whereas free YFP did not show such localization (Fig 3A, lower panels). To confirm the association of ATG8 with chloroplasts, the purified organelles, as well as their corresponding protoplasts were analyzed by immunoblotting (Fig 3B). By comparing with the Tic110 control protein that is exclusively localized in chloroplasts (Bédard et al, 2007; Ling et al, 2012), we estimated a large proportion of ATG8 interacted with chloroplasts, whereas the control free YFP protein was only marginally associated with chloroplasts (Fig 3C). Similar results were obtained when chloroplast‐associated ATG8 was detected using an ATG8 antibody (Appendix Fig S5A and B). These results implied that autophagy affects the chloroplast's proteins and functions extensively.

Figure 3. NBR1 is associated with autophagy at the envelope of chloroplasts.

• A–C
  Presence of autophagosome in the chloroplast envelope. Arabidopsis protoplasts expressing free YFP and YFP‐ATG8 (autophagosome marker) were analyzed by confocal microscopy. Corresponding chlorophyll autofluorescence images were employed to orientate the YFP signal in relation to the chloroplasts. For eliminating interference from cytosolic signals, intact protoplasts (upper panels) were ruptured using 10‐μm nylon mesh to isolate the chloroplasts. Retention of YFP signals in association with chloroplasts upon rupturing the cell (lower panels) clearly indicated that the YFP‐ATG8 was associated with chloroplasts. Representative images are shown (A). In addition, whole protein samples of protoplasts and purified chloroplasts were analyzed by SDS‐PAGE and immunoblotting, using antibodies against the YFP tag and Tic110 (a chloroplast marker). Typical immunoblotting results are shown (B). Band intensities were quantified and normalized to equivalent data for Tic110, to estimate protein abundance associated with chloroplasts (C).
• D, E
  Identification of ATG8's interactors associated with chloroplasts. Chloroplasts were isolated as in (B), and subjected to immunoprecipitation (IP) using anti‐YFP magnetic beads. Total lysate and IP eluate samples were analyzed by immunoblotting using anti‐YFP and anti‐Tic40 (a TIC component used as a negative control) antibodies (D). The IP elution samples shown in (D) were analyzed by LC–MS/MS. This identified NBR1 and ~ 150 chloroplast proteins, which were selectively listed in (E); a complete interactor dataset is provided in Dataset EV1. IEM, inner envelope membrane; OEM, outer envelope membrane.
• F
  Localization of NBR1–YFP in the chloroplast envelope. Arabidopsis protoplasts (upper panels) expressing NBR1–YFP, and chloroplasts (lower panels) isolated from the protoplasts were analyzed by confocal microscopy, in procedures similar to (A).
• G, H
  Presence of NBR1–HA in chloroplast fraction. The protoplasts were isolated from four‐week‐old transgenic Arabidopsis plants stably expressing p35S:NBR1–HA (NBR1–HA) and p35S:YFP–HA (YFP–HA), followed by chloroplast purification, respectively. Protein samples of whole protoplasts and isolated chloroplasts of NBR1–HA and YFP–HA were then analyzed by SDS–PAGE and immunoblotting, using antibodies against the HA tag and Tic40. Typical immunoblotting results are shown (G). Band intensities were quantified and normalized to equivalent data for Tic40, to estimate protein abundance associated with chloroplasts (H). Data are means ± SEM from three biological replicates. Scale bars, 10 μm.

Source data are available online for this figure.

We then performed co‐immunoprecipitation (co‐IP) using YFP‐Trap magnetic beads followed by mass spectrometry, using protein extracts prepared from the chloroplast fraction of cells expressing YFP‐ATG8 or YFP‐HA (Fig 3D). Using purified chloroplasts allowed us to focus on chloroplast ATG8 interactors, and to avoid detecting autophagy cargos in other organelles, because ATG8 is an abundant cellular constituent distributed throughout the nucleocytosolic compartment. After normalization and excluding proteins bound non‐specifically with YFP–HA, 150 proteins were identified as putative ATG8‐interacting proteins associated with chloroplasts (Fig 3E; Dataset EV1). The candidate list included chloroplast proteins belonged to various functional categories, and they are localized in different suborganellar compartments, which could potentially be cargo proteins. As expected, TOC components were detected as interactors of ATG8, further confirming that they are autophagic cargos. In addition, among the list was NBR1 (Next to BRCA1 gene 1), a plant homolog of the mammalian autophagic cargo adaptor SQSTM1/p62 (Svenning et al, 2011). Considering the crucial role of p62 in mammals, it is surprising that Arabidopsis nbr1 mutants do not show obvious defects under normal growth conditions (Zhou et al, 2013). Recently, NBR1 was reported to be responsible for selective autophagy in plants under stress conditions (Ji et al, 2020; Thirumalaikumar et al, 2021). However, the role of NBR1 in the selective autophagy of chloroplasts, if any, is unknown.

To assess whether NBR1 acts as an autophagy adaptor for chloroplast proteins, we first examined the association of NBR1 with chloroplasts in cells expressing NBR1–YFP. Confocal microscopy analysis demonstrated strong accumulation of YFP‐labeled puncta in the cytosol, adjacent to chloroplast envelope where the TOC apparatus resides (Fig 3F, upper panel). The association of NBR1–YFP with chloroplasts became clear when the organelles were isolated from the cells (Fig 3F, lower panel). Such NBR1‐chloroplast association was further verified by immunoblotting analysis of isolated chloroplast fractionations, which indicated that a considerable proportion of NBR1 is localized in the chloroplast (Fig 3G and H; Appendix Fig S5A and B). Collectively, these data implied that chloroplast proteins, particularly those localized at the envelope membrane, are likely selectively targeted by NBR1 for autophagic degradation.

NBR1 associates with TOC components

To test whether NBR1 is an adaptor for the selective autophagic degradation of TOC proteins, we first assessed co‐localization using protoplasts coexpressing YFP–TOC protein fusions and NBR1–CFP. While TOC fusion proteins still produced a typical distributed signal at the periphery of the chloroplasts, similar to when they are expressed alone (Fig EV1; Appendix Fig S6), all three TOC proteins (green puncta) also partially co‐localized with NBR1 (blue puncta) at chloroplasts, in a similar punctate pattern to NBR1 alone (Figs 4A and 3F; Appendix Fig S6). In contrast, the control fusion proteins SP1–YFP and FAX1–YFP displayed only distributed (non‐punctate) chloroplast envelope fluorescence, even when coexpressed with NBR1–CFP (Fig 4A). Chloroplast colocalization of NBR1–CFP with TOC–YFP fusion proteins was clearer when the organelles were isolated from the protoplasts, whereas SP1–YFP and FAX1–YFP did not show such colocalization with NBR1–CFP (Appendix Fig S7). These results suggest that NBR1 is able to recruit TOC proteins specifically. To further monitor co‐localization in the vacuole, we also generated transgenic lines coexpressing mScarlet‐tagged NBR1 and the YFP‐tagged TOC proteins (i.e., p35S:NBR1‐mScarlet/pToc33:YFP–Toc33 and p35S:NBR1–mScarlet/pToc75:YFP–Toc75), by crossing. Plants were grown in the dark for 2 days with ConA treatment in order to induce autophagy and enhance the vacuolar fluorescence signals. As shown in Fig 4B and Movie EV4, the TOC (green) and NBR1 (red) signals co‐localized in randomly moving vacuolar puncta that were most likely autophagic bodies. Such vacuolar puncta are similar in size to the ATG8‐TOC puncta in Fig 1D, but appear smaller than the Rubisco‐containing bodies (RCBs) described previously (Ishida et al, 2008; Wada et al, 2009), implying that they may contain distinct chloroplast cargos. While RCBs are known to contain various stromal proteins, we propose that the TOC‐containing puncta contain mainly TOC complexes, although we cannot rule out the possibility that there might be other chloroplast proteins in the TOC‐containing puncta.

Figure 4. NBR1 targets TOC components for autophagy.

1. NBR1 and TOC proteins colocalized at the chloroplast envelope of protoplasts. Arabidopsis protoplasts coexpressing the YFP‐tagged chloroplast proteins and NBR1‐CFP were analyzed by confocal microscopy following ConA treatment for 4 h. Representative images are shown. Chlorophyll autofluorescence was used to determine the localization of the YFP fluorescence signals relative to the chloroplasts.
2. Colocalization of NBR1 and TOC proteins in the central vacuole of Arabidopsis leaf cells. Eight‐day‐old transgenic Arabidopsis seedlings stably coexpressing p35S:NBR1‐mScarlet and either pToc33:YFP–Toc33 or pToc75:YFP–Toc75 were treated with ConA for 2 days, and then analyzed by confocal microscopy. Accumulation of punctate bodies of overlapping fluorescence in the central vacuole are indicated by arrowheads.
3. BiFC results showed interaction between NBR1 and TOC proteins. cYFP and nYFP refer to C‐terminal YFP fragment and N‐terminal YFP fragment, respectively. Reconstitution of YFP fluorescence was assessed after transient coexpression of the indicated pairs of fusion proteins. CDKA1 and OEP7 were used as negative controls for OEM and cytosolic proteins, respectively. Representative images are shown. Scale bars, 10 μm.
4. The frequency of protoplasts showing a BiFC signal in (C) was quantified. All data are means ± SEM from three experiments.
5. NBR1 coimmunoprecipitated with TOC proteins. Proteins were extracted from 8‐day‐old transgenic Arabidopsis seedlings stably expressing p35S:NBR1–HA or p35S: NBR1‐ΔUBA‐HA, or control seedlings only expressing p35S:YFP–HA, following UVB treatment for 4 h. Obtained protein extract was analyzed by IP using anti‐HA resin, and analyzed by immunoblotting of four endogenous proteins or HA‐tagged proteins. Tic40 acted as negative controls. The asterisks indicate nonspecific cross‐reacting bands. Arrow indicates the NBR1‐HA band. The arrowhead indicates full‐length Toc159 protein. Molecular weight markers (sizes in kD) are indicated to the right of the images.

Source data are available online for this figure.

To determine whether NBR1 interacts with the TOC complex, we carried out bimolecular fluorescence complementation (BiFC) experiments in Arabidopsis protoplasts. The NBR1 coding sequence was cloned into a vector that C‐terminally appends the C‐terminal half of YFP (cYFP); and a negative control construct encoding CDKA1, a cytosolic protein, was similarly generated. These cYFP constructs were coexpressed with various other constructs encoding the core TOC components (Toc33, Toc159 and Toc75) bearing the complementary, N‐terminal moiety of the YFP protein (nYFP); or with a negative control construct encoding OEP7, a non‐TOC outer envelope membrane protein, bearing the nYFP moiety appended C‐terminally. This method can report on protein–protein interactions via the detection of YFP fluorescence, caused by reconstitution of functional YFP from the nYFP and cYFP fragments. Remarkably, NBR1 was found to physically associate with all tested TOC proteins, whereas CDKA1 barely interacted with the TOC proteins (Fig 4C and D). Moreover, these interactions displayed a punctate pattern adjacent to the chloroplast, placing them in an ideal subcellular context for the in situ regulation of the chloroplast protein import machinery by the cytosolic autophagy components. Conversely, OEP7–nYFP could not restore a fluorescence signal with NBR1–cYFP, suggesting selective targeting of the TOC apparatus by NBR1.

To verify the interactions, we employed co‐IP assays, which were conducted using transgenic Arabidopsis plants stably expressing HA‐tagged NBR1. In these experiments, NBR1–HA was enriched using anti‐HA magnetic beads, and all of the core TOC components were detected in the fractions that co‐eluted with NBR1–HA, using antibodies against the native proteins, but not in control elutions (Fig 4E). The specificity of these TOC component interactions was demonstrated by the absence of a control protein, Tic40, from the eluted fractions. Thus, we concluded that NBR1 may regulate TOC proteins directly.

NBR1 mediates autophagic degradation of the TOC complex during UVB stress

Previously, we showed that the TOC apparatus is controlled by SP1‐dependent UPS action, in response to abiotic stresses such as salt and osmotic stresses (Ling & Jarvis, 2015b). Intriguingly, although chloroplast functions are apparently crucial under light and temperature stresses (Schwenkert et al, 2022), no obvious effect of SP1 was observed in these conditions, implying the involvement of alternative or complementary pathways (Ling & Jarvis, 2015b). The finding that NBR1 controls the TOC complex motivated us to investigate whether NBR1‐mediated selective autophagy is responsible for the chloroplast responses to light and temperature stresses. Thus, we grew atg7‐2 and nbr1‐2 mutant and NBR1 overexpressor (NBR1–OX) Arabidopsis plants under abiotic stress conditions, starting with ultraviolet‐B (UVB) treatment (5 μE/m²/s) to mimic natural high‐light stress (Shi & Liu, 2021). Both atg7 and nbr1 mutant plants failed to cope with these conditions, whereas NBR1–OX plants were more stress tolerant than wild type, as judged by plant size and chlorophyll content (Fig 5A and B).

Figure 5. NBR1 is involved in the autophagic degradation of TOC components during UVB stress.

• A, B
  Phenotypes of seedlings under UVB stress. Briefly, 8‐day‐old WT, atg7‐2, nbr1‐2 and NBR1‐OX (overexpressor) seedlings grown under normal condition were subjected to UVB stress, with UVB treatment for 3 h and recovery for 7 days. Representative plant images (A) and total chlorophyll content values (B) are shown; the latter provide a convenient metric for biomass. The chlorophyll values for the mutants and overexpressor plants were statistically significantly different from that for the WT (Student's t‐test; *significant at P < 0.05; ***significant at P < 0.001).
• C, D
  TOC protein levels were regulated by autophagy in response to short‐term UVB stress. Eight‐day‐old plants of indicated genotypes were subjected to UVB stress for the indicated times, and then analyzed by immunoblotting using 10 μg total protein extract loaded per lane (C, upper). Coomassie staining indicated that the treatments did not change the overall protein profile (C, lower). Band intensities were quantified and normalized to equivalent Actin data (D) Time 0 was taken as 100%. Unpaired two‐tailed Student's t‐tests revealed that the Toc159 protein level in nbr1‐2 (P < 0.05) and NBR1–OX (P < 0.05), the Toc75 protein level in atg7‐2 (P < 0.05), nbr1‐2 (P < 0.01), and NBR1–OX (P < 0.05), and the Toc33 protein level in atg7‐2 (P < 0.05), nbr1‐2 (P < 0.01) and NBR1–OX (P < 0.05), were significantly different from levels in the WT at the time of 1 day recovery from UVB stress.
• E, F
  Toc159 was modified by both K48 and K63 ubiquitin linkage types. Eight‐day‐old WT plants were grown under normal growth condition (−), or subjected to UVB stress (+), with UVB treatment for 4 h. The protein extract from the plants was analyzed by IP using native Toc159 antibody, or its preimmune serum as a negative control, and analyzed by immunoblotting using antibodies as indicated to the left of the images (E). Tic40 acted as a negative control. The asterisk indicates a truncated Toc159 form without A domain (Agne et al, 2010). Molecular weight markers (sizes in kD) are indicated to the right of the images. Please note that ubiquitin antibodies tend to produce stronger signals for hyper‐ubiquitinated species than for hypo‐ubiquitinated or mono‐ubiquitinated species (based on the number of epitopes), which contrasts with the behavior of the Toc159 antibody. Thus, the signals of the two antibodies are not expected to fully overlap. Poly‐ubiquitinated Toc159 may not be detected by the Toc159 antibody due to the much lower abundance of these species compared with the mono‐ubiquitinated or unmodified forms. This is why the poly‐Ub K48 and K63 signals appear at higher molecular weight positions than the Toc159 signals. Relative Toc159 amounts modified by K48 or K63 ubiquitin linkage in plants with or without UVB stress were assessed by quantifying the smeary band intensity of the IP samples in α‐K48 or α‐K63 blots (F). All data are means ± SEM from three experiments.

Source data are available online for this figure.

To understand the molecular basis for stress tolerance mediated by NBR1, we began by examining the levels of various chloroplast proteins in atg7‐2, nbr1‐2 and NBR1–OX plants by immunoblotting, under UVB stress conditions, as the primary function of selective autophagy is to control protein abundance. For this work, we focused on short‐term UVB stress (eight‐day‐old seedlings were exposed to UVB for 3 h, and then recovered under normal condition for 1 day), to avoid strong morphological changes that might lead to pleiotropic consequences, and to aid identification of the primary effects of the stress. The results revealed that protein abundance of TOC components declines significantly (P < 0.05) 1 d after the UVB treatment in the wild type, although no obvious effect was detected immediately after the UVB exposure (Fig 5C and D). This response was NBR1 and autophagy dependent, as it did not occur in the atg7 and nbr1 mutants; whereas the TOC proteins reached even lower levels in NBR1–OX plants. In contrast with the TOC proteins, a component of TIC apparatus did not change in abundance in response to the stress (Fig 5C and D). In addition, inspection of the protein banding pattern upon Coomassie‐staining revealed that the effect on TOC proteins was unlikely to be a general, damage‐related phenomenon affecting many proteins (Fig 5C). Semi‐quantitative RT–PCR analyses did not reveal significant changes in TOC transcript levels under the conditions employed, ruling out the possibility that the NBR1/autophagy‐dependent changes of TOC protein levels were an indirect consequence of transcriptional effects (Appendix Fig S8).

NBR1 typically recognizes ubiquitinated cargo proteins (Svenning et al, 2011; Sun et al, 2021b). To investigate whether ubiquitination of TOC proteins occurs in response to UVB stress, to stimulate selective autophagy, we assessed the ubiquitination status of the TOC components in vivo, with or without UVB treatment. Although polyubiquitination has the potential to trigger protein degradation through either the UPS or autophagy, distinct degradation signals involving specific linkage types within the ubiquitin chains are typically involved. For example, K48‐linked polyubiquitin typically induces proteasomal proteolysis, whereas K63‐linked chains can facilitate the autophagic degradation of protein substrates (Kwon & Ciechanover, 2017). Although TOC proteins are known targets of ubiquitination, the type of ubiquitin chain is unknown. To elucidate this, we purified Toc159 by immunoprecipitation using anti‐Toc159 antibody (and the preimmune serum as a negative control) from wild‐type Arabidopsis plants subjected to UVB exposure (alongside control plants), and the samples were subsequently analyzed by immunoblotting. Remarkably, high‐molecular‐weight smears were clearly visible in the precipitates of Toc159 protein, in blots using two linkage‐specific ubiquitin antibodies, indicating that Toc159 is polyubiquitinated via both K48 and K63 linkages (Fig 5E). Moreover, not only the amount of detectable K48 ubiquitination but also that of K63 ubiquitination, was dramatically enhanced when the plants were exposed to UVB stress (Fig 5E and F). In contrast, Tic40, serving as a negative control, was not detected in the Toc159 precipitates. These results provided further strong support for the conclusion that chloroplast proteins are selectively regulated by ubiquitination and autophagy. Elevated K63 polyubiquitination of Toc159 upon UVB stress suggested that ubiquitination is involved in activating the autophagic degradation of TOC components.

In mammal, evidences indicate that NBR1 binds ubiquitinated proteins through its C‐terminal UBA domains (Kirkin & Rogov, 2019). To assess whether the ubiquitination of TOC components is a prerequisite for their recruitment by NBR1, we employed co‐IP assay using an NBR1–ΔUBA (deletion of both UBA domains) variant. As shown in Fig 4E, NBR1–ΔUBA resulted in strongly reduced binding with all three TOC protein, compared with full‐length NBR1, indicating that the interaction between NBR1 and TOC substrates depends on ubiquitination.

As NBR1 was shown to bind both ATG8 and TOC proteins, we hypothesized that NBR1 serves as an intermediary between ATG8 and TOC proteins in selective autophagic degradation. To test this hypothesis, we employed co‐IP assays, which were conducted using wild type, nbr1‐2 mutant and NBR1‐OX Arabidopsis protoplasts expressing YFP‐tagged ATG8. In these experiments, the interactions between YFP–ATG8 and TOC proteins were strongly increased by NBR1 overexpression, but largely diminished in nbr1‐2 mutant (Appendix Fig S9). These results suggest that NBR1 is responsible for the recruitment of ATG8 to TOC proteins for degradation.

NBR1 mediates autophagic degradation of the TOC complex during heat stress

Temperature stress was reported to impair chloroplast protein import, through an unknown mechanism (Dutta et al, 2009). Thus, we next turned our attention to heat stress. In this case, we incubated soil‐grown wild‐type, atg7‐2, nbr1‐2 and NBR1–OX mature Arabidopsis plants under 42°C for 17 h and then returned them to the normal growth conditions (22°C) for 2 days before their visible phenotypes were observed. Paralleling the UVB stress results, atg7 and nbr1 mutants failed to develop under these conditions, and exhibited severe leaf death, whereas NBR1 overexpressors were more stress tolerant than wild type, showing much bigger shoot size (Fig 6A).

Figure 6. NBR1 is involved in the autophagic degradation of TOC components during heat stress.

• A
  Phenotypes of plants under heat stress (HS). Briefly, 25‐day‐old WT, atg7‐2, nbr1‐2 and NBR1‐OX (overexpressor) plants grown in soil under normal condition were subjected to heat stress, with 42°C treatment for 17 h and recovery for 2 days. Representative plant images are shown.
• B, C
  TOC protein levels were regulated by autophagy in response to short‐term heat stress. Eight‐day‐old plants of indicated genotypes grown in vitro were subjected to short term heat stress, recovered for the indicated times, and then analyzed by immunoblotting using 10 μg total protein extract loaded per lane (B, upper). Coomassie staining indicated that the treatments did not change the overall protein profile (B, lower). Band intensities were quantified and normalized to equivalent Actin data (C). Time 0 was taken as 100%. Data are means ± SEM from three biological replicates. Unpaired two‐tailed Student's t‐tests revealed that the Toc159 protein level in nbr1‐2 (P < 0.05) and NBR1–OX (P < 0.01), and the Toc75 protein level in NBR1–OX (P < 0.05), were significantly different from levels in the WT at Time 12 h; and that the Toc159 protein level in atg7‐2 (P < 0.001), nbr1‐2 (P < 0.01), and NBR1‐OX (P < 0.001), and the Toc75 protein level in atg7‐2 (P < 0.01), nbr1‐2 (P < 0.01), and NBR1–OX (P < 0.01), were significantly different from levels in the WT at Time 24 h.

Source data are available online for this figure.

As in the previous experiment, immunoblotting was performed to investigate the effect of short‐term heat stress on TOC proteins of in vitro grown seedlings, and the results were essentially identical to those obtained for UVB stress: the TOC apparatus was specifically and quickly depleted under stress conditions, through the action of NBR1‐mediated autophagy; whereas a TIC component and a chloroplast interior protein, AtpA, remained stable in response to stress (Fig 6B and C). Notably, such short‐term heat stress did not cause obvious growth defect in the seedlings of all tested genotypes, ruling out the possibility that the changes of TOC protein levels were merely consequences of developmental abnormality. Differences in the abundance of the TOC proteins between genotypes and conditions were not linked to transcriptional changes between the genotypes, as the mRNA levels of the corresponding proteins were similar, as revealed by semi‐quantitative RT–PCR (Appendix Fig S10). Collectively, our results indicate an important role for selective autophagy of chloroplast proteins in plant responses to UVB and heat stresses.

Degradation of TOC components mediated by NBR1 impacts protein import

We previously reported that SP1 acts to degrade the TOC apparatus under certain abiotic stresses to limit protein import (Ling & Jarvis, 2015b). To test whether altered NBR1 expression, and corresponding TOC protein level changes, can also influence chloroplast protein import efficiency, we first compared the import capabilities in vitro of chloroplasts purified from unstressed wild‐type, nbr1‐2 mutant, and NBR1 overexpressor plants. Isolated chloroplasts were incubated with the precursor of the Rubisco small subunit (SSU), and import efficiency was assessed by quantifying the amount of mature, processed proteins in the organelles (Fig 7A). In this analysis, nbr1 mutant chloroplasts displayed slightly elevated import efficiency of SSU, relative to wild type, whereas NBR1 overexpressor chloroplasts tended to show a decline in import capacity (Fig 7A).

Figure 7. Functional analysis of NBR1 reveals roles in chloroplast protein import and development.

• A
  Analysis of protein import efficiency in vitro. Chloroplasts were isolated from 12‐day‐old WT, nbr1‐2 and NBR1–OX seedlings grown in vitro under normal condition, and were subjected to in vitro protein import assay by using 35S‐labeled Rubisco small subunit (SSU) precursor protein as the import substrate, followed by corresponding quantification of the maturation of the radiolabeled precursor protein. A representative phosphor screen image is shown (upper left), and Coomassie staining of Rubisco large subunit (LSU) indicated equal loading (lower left); Times indicate minutes after the start of each import reaction. IVT, in vitro translation product used as a control. Together with similar images from two additional experiments, this was used to conduct the quantitative analysis shown (right); the amount of imported, mature protein in each sample was expressed as a percentage of that present at the final time point for the WT. Data are means ± SEM from three biological replicates.
• B, C
  Analysis of protein import efficiency in vivo. The 35S:cpGFP probe was analyzed in the 8‐day‐old WT, ppi1, nbr1‐2 and NBR1–OX backgrounds grown in vitro under normal condition (B) or UVB stress (C). UVB stress was conducted by UVB treatment for 4 h followed by recovery for 1 day. The ppi1 mutant lacks Toc33 and has constitutively reduced chloroplast protein import (Jarvis et al, 1998). In each case, GFP signal recovery was measured following photobleaching, in hypocotyls of seedlings. Recovery was normalized using data from unbleached cells, and is expressed relative to the average normalized GFP signal across 4–5 cell fluorescence measurements taken 3 s apart before bleaching began. Data are means ± SEM from three biological replicates.
• D–H
  Chloroplast development in WT, nbr1‐2 and NBR1–OX seedlings under UVB stress was studied. Cotyledons of 8‐day‐old plants grown in vitro under normal condition (D) or UVB stress (E) were analyzed by transmission electron microscopy, and representative images are shown. UVB stress was conducted in the same procedures as in (C). In (D) and (E), the upper images are at the same magnification (scale bar, 2 μm); and below these are high‐magnification (5×) images corresponding to the boxed regions above. These and similar micrographs of chloroplasts were used to determine chloroplast shape (F), the number of thylakoid lamellae per granum (G) and the number of starch granules per chloroplast (H) in each genotype, with or without UVB stress. Asterisks indicate significance according to an unpaired two‐tailed Student's t‐test (*significant at P < 0.05; ***significant at P < 0.001; ****significant at P < 0.0001). Values are means ± SEM from 30 to 50 chloroplasts.

Source data are available online for this figure.

These data suggested that NBR1 may play a nuanced role in optimizing chloroplast protein import when plants are grown under normal conditions. However, changes of NBR1 expression had strong effects on TOC levels in plants grown under stress conditions (Fig 5D), implying that NBR1‐mediated regulation of protein import may be especially important during stress. We attempted similar in vitro protein import assays using plants kept under stress conditions, but chloroplasts isolated from these plants proved to be incompetent for such assays, possibly due to fragility caused by stress damage. Thus, we conducted in vivo import experiments based on fluorescence recovery after photobleaching (FRAP) assay. For this purpose, we first introduced a chloroplast‐targeted marker protein, cpGFP (Ishida et al, 2008) into the wild type, nbr1‐2, plastid protein import 1 (ppi1) mutant and NBR1–OX plants by crossing. The ppi1 mutant is a knockout of Toc33 in which TOC activity is constitutively impaired, due to the permanent loss of Toc33 (Jarvis et al, 1998). The cpGFP marker is a GFP protein harboring a chloroplast transit peptide, which is stably expressed in plants driven by the constitutive 35S promoter. The chloroplast transit peptide enables cpGFP to be exclusively localized within the chloroplast stroma. In the FRAP assay, the chloroplast GFP signal in a cell was partially depleted by photobleaching using the confocal scanning laser microscope, enabling newly imported GFP molecules to be monitored as fluorescence signal recovery, due to new chromophore formation (Craggs, 2009). Thus, even when the plants were grown under stress conditions, it was possible to conduct non‐destructive in situ assessments of relative chloroplast protein import efficiency from different genotypes.

When plants were grown under normal conditions, no clear differences in chloroplast import efficiency could be observed in nbr1‐2 and NBR1–OX leaves, as estimated by signal recovery (Fig 7B). By contrast, following exposure to UVB treatment, protein import was greatly enhanced in the nbr1 mutant and dramatically reduced in NBR1–OX plants, compared with that in wild type (Fig 7C; Appendix Fig S11; Movies EV5 and EV6). In fact, stressed NBR1 overexpressors showed a defect in protein import comparable to that seen in ppi1. This implies that NBR1 is a strong regulator of protein import under stress conditions. Significantly, mutant and overexpressor plants displayed no obvious growth defects after the short‐term UVB stress employed here. Thus, protein import was indeed specifically regulated by selective autophagy in response to stress, and not indirectly influenced as a result of growth differences between the genotypes under such conditions.

Altered NBR1 expression affects chloroplast morphology and photosynthesis

Protein import is essential for chloroplast biogenesis and maintenance (Jarvis & López‐Juez, 2013). Plants with compromised protein import typically exhibit defective chloroplast morphology, with regard to chloroplast size and shape and thylakoid membrane structure (Jarvis et al, 1998; Bauer et al, 2000; Kovacheva et al, 2005). We hypothesized that NBR1's role under stress is to quench chloroplast function by suppressing protein import. To test this possibility, we first analyzed chloroplast ultrastructure by transmission electron microscopy in nbr1‐2 and NBR1–OX plants, either kept under normal conditions or treated with short‐term UVB stress. Although chloroplasts of nbr1 mutant and NBR1 overexpressor plants had chloroplast morphology similar to the wild type in the absence of stress (Fig 7D), consistent with our earlier finding that altered NBR1 levels have minor consequences under such conditions, the importance of NBR1 was clearly revealed following stress treatment (Fig 7E). Strikingly, NBR1–OX chloroplasts were frequently deformed in shape, as illustrated by the decreased length/width ratio of the organelles (Fig 7F). In addition, they had severely disrupted thylakoid networks, as indicated by the significantly reduction in thylakoid stacking compared with the wild type (Fig 7G). Moreover, after the UVB exposure, the number of starch granules in NBR1–OX chloroplasts was significantly decreased compared with the wild type (Fig 7H), suggesting lower carbon fixation efficiency. On the contrary, nbr1 mutant chloroplasts appeared to respond even less to the UVB stress than wild type, regarding thylakoid complexity and starch granule number of the organelles.

Abundances of TOC proteins, such as Toc159 are dramatically reduced in ppi1 mutant, and mutations of CHLORAD components, SP1 or SP2 can partially suppress the ppi1 phenotype, possibly due to the increase of other TOC protein levels (Ling et al, 2012, 2019). To confirm the role of NBR1 in protein import genetically, we crossed nbr1‐2 mutant with ppi1, and phenotypically characterized the resultant double mutant. Intriguingly, it appeared greener than the ppi1 single mutant, although nbr1 single mutant was indistinguishable from the wild type (Fig 8A). Correspondingly, the leaf chlorophyll concentration of ppi1 nbr1 was also significantly increased relative to ppi1 (Fig 8B). To assess whether the phenotypic suppression observed in ppi1 nbr1 was linked to changes in the abundance of TOC proteins, we analyzed the double mutant and control plants via immunoblotting. The double mutant displayed a clear increase in the abundance of Toc159 compared with ppi1 (Fig 8C and D); note that possible effects of nbr1 on Toc33 could not be assessed due to the complete inactivation of the TOC33 gene in ppi1. These results indicated a genetic link between NBR1 and the chloroplast protein import apparatus.

Figure 8. NBR1 mutation suppresses the phenotype of the plastid protein import mutant ppi1 .

• A, B
  Phenotypes of 2‐week‐old WT, nbr1‐2, ppi1 and ppi1 nbr1‐2 seedlings grown on soil. Representative plant images (A) and chlorophyll contents (B) are shown. The chlorophyll values for the ppi1 and ppi1 nbr1‐2 plants were statistically significantly different (Student's t‐test; ****significant at P < 0.0001). Values are means ± SEM from 20 plants.
• C, D
  TOC protein accumulation was improved in ppi1 nbr1‐2 relative to ppi1. Two‐week‐old plants of indicated genotypes were subjected to immunoblotting analysis (C). Tic40 acts as a negative control, which does not change in response to the ppi1 mutation. Numbers over the lanes indicate relative loading amounts of total protein extracts, and 100% = 10 μg. Band intensities were quantified and normalized to equivalent Tic110 data (D). The values of Toc159 protein abundances for the ppi1 and ppi1 nbr1‐2 plants were statistically significantly different (Student's t‐test; ****significant at P < 0.0001). Data are means ± SEM from three experiments.
• E–G
  Chloroplast development in ppi1 and ppi1 nbr1‐2 seedlings was studied. Cotyledons of 2‐week‐old plants grown under normal condition as in (A) were analyzed by transmission electron microscopy, and representative images are shown (E). In (E), the upper images are at the same magnification (scale bar, 1 μm); and below these are high magnification (3×) images corresponding to the boxed regions above. These and similar micrographs of chloroplasts were used to determine chloroplast size (F) and the number of thylakoid lamellae per granum (G) in each genotype. Asterisks indicate significance according to an unpaired two‐tailed Student's t‐test (****, significant at P < 0.0001). Values are means ± SEM from 30 to 40 chloroplasts.

Source data are available online for this figure.

To assess whether the aforementioned chloroplast morphology changes are linked to the role of NBR1 in TOC regulation, we also analyzed chloroplast ultrastructure in ppi1 and ppi1 nbr1‐2 mutants under normal growth conditions. Although chloroplast size in these two mutants was not significantly different, the addition of the nbr1 mutation partially recovered the defect of ppi1 in thylakoid stacking (Fig 8E–G). These results support the hypothesis that TOC regulation by NBR1 influences chloroplast morphology.

Such NBR1‐dependent changes of chloroplast morphology suggested that autophagic regulation of protein import plays an important role in adapting chloroplast functions (e.g., photosynthesis) in response to stress. To address this hypothesis, we assessed photosynthetic performance in these plants by measuring the parameter F _v/F _m, which represents the maximal quantum yield of PSII (Ling et al, 2021). In line with the chloroplast morphology data, plants with altered NBR1 expression levels did not show obvious changes in the F _v/F _m value under the normal conditions used (Appendix Fig S12A). However, under UVB stress conditions, NBR1–OX plants showed clearly reduced F _v/F _m values (Appendix Fig S12B), suggesting that NBR1 acts to limit photosynthetic activity in response to stress. The abundancies of photosynthetic apparatus components PsaD and PsbP (both of which are imported into chloroplasts by the TOC apparatus) also declined markedly in the NBR1–OX plants under UVB stress conditions (Appendix Fig S12C–E). Such phenomena may be responsible for the reduced photosynthetic performance of those plants. Compared with NBR1–OX, ppi1 showed even stronger reductions in photosynthetic proteins, due to the complete loss of one key TOC receptor, Toc33 (as opposed to the more modest TOC reductions seen in stressed NBR1–OX plants). This explains the big phenotypical difference between ppi1 and NBR1–OX plants.

Discussion

Previous studies revealed that autophagy is involved in the regulation of chloroplast functions through the bulk degradation of chloroplast cargos or the turnover of whole photodamaged chloroplasts (Sun et al, 2021a). However, understanding of the mechanisms of chloroplast autophagy is still very preliminary, compared to autophagy in other organelles in yeast and mammals (Nakamura & Izumi, 2018; Zhuang & Jiang, 2019). In particular, little is known about the selective autophagy of chloroplast proteins due to a lack of identified receptors and adaptors. In this study, we present multiple lines of evidence showing that NBR1 acts as a selective autophagy adaptor for the TOC complex to regulate chloroplast protein import. This regulatory mechanism is physiologically important during stress responses.

Recent studies revealed the regulation of chloroplast TOC abundance by the UPS, which is critical for plastid biogenesis and plant stress tolerance (Ling et al, 2012, 2019). However, whether other ubiquitin‐dependent pathways, such as selective autophagy, are involved in TOC regulation has remained an open question. Data presented in this study reveal that TOC component abundancies are indeed controlled by selective autophagy. Evidence for this includes the confocal microscopy detection of vacuolar TOC protein‐containing puncta resembling autophagic bodies (Fig 1). Such puncta no longer existed when a key autophagy pathway component, ATG7, was missing (Fig 2A). Moreover, ATG8 was co‐localized with the vacuolar TOC‐containing bodies, demonstrating a functional association between TOC proteins and autophagy. Significantly, other envelope proteins such as SP1–YFP and FAX1–YFP were not localized in such TOC‐containing autophagic bodies in our experimental conditions, ruling out the possibility that they were simply cargos of the well‐known bulk chloroplast autophagic pathway induced upon starvation, described previously (Wada et al, 2009; Spitzer et al, 2015). Moreover, TOC proteins were specifically accumulated in wild‐type plants treated with autophagy inhibitor, or in autophagy‐deficient mutant (Fig 2B), confirming that TOC components are selective autophagy substrates.

In a screen for autophagic components in purified chloroplasts, we identified NBR1 as potential autophagy adaptor for the regulation of chloroplast cargos (Fig 3E). Although a role for plant NBR1 in selective autophagy was revealed some years ago, we have only begun to recognize its substrates in the past few years (Ji et al, 2020; Thirumalaikumar et al, 2021). Even less information exists concerning its substrates in chloroplasts. Subcellular fractionation analysis revealed that a large proportion of cellular ATG8 and NBR1 is associated with the chloroplast (Fig 3; Appendix Fig S5), suggesting that chloroplast cargos are predominant targets of autophagy under the tested conditions. In particular, we detected several chloroplast envelope proteins (e.g., Toc75 and Toc33) in the list of the candidate ATG8 interactors; and interaction of NBR1 with key TOC components, Toc33, Toc159 and Toc75, was validated through co‐localization, BiFC and co‐IP assays (Fig 4).

Based on previous mutant analyses, NBR1 was reported to be important under specific abiotic challenges, such as heat, oxidative, salt and drought stresses (Zhou et al, 2013), but its autophagic cargos under such conditions have remained largely unknown. Here, we demonstrated that nbr1 mutant plants are more sensitive to UVB and heat stresses, whereas NBR1 overexpressors are more tolerant of such stresses (Figs 5 and 6). Given that chloroplast functions are closely linked to both of these stresses (Dutta et al, 2009; Allorent et al, 2016), this implies that NBR1's role in chloroplasts is physiologically important. Indeed, TOC proteins are specifically regulated by NBR1 during short‐term UVB and heat stresses (Figs 5 and 6); and, consistently, nbr1 chloroplasts showed clearly enhanced import efficiency, whereas NBR1–OX chloroplasts showed compromised protein import, particularly under stress conditions, as indicated by in vitro and in vivo protein import assays (Fig 7). Furthermore, genetic experiment showed that nbr1 mutant significantly suppressed ppi1, a mutant with compromised TOC import machinery (Fig 8). We interpret that other TOC components such as Toc159 accumulated at higher levels in nbr1 ppi1 double mutant, and that this synthetically improves the double mutant phenotype relative to the ppi1 control.

Previous work revealed that the protein import TOC machinery is regulated by the UPS in response to developmental and environmental factors (Thomson et al, 2020), and that the control of TOC protein levels influences stress tolerance (Ling & Jarvis, 2015b). Our data here provide compelling evidence that selective autophagy also has a role in TOC regulation during abiotic stresses. Such regulation may be causally linked to the changes in photosynthetic protein abundance, photosynthetic performance, and chloroplast ultrastructure seen in nbr1 mutant and NBR1 overexpressor plants (Fig 7; Appendix Fig S12). Although plant growth may be compromised by NBR1‐mediated degradation of the TOC apparatus upon short‐term stress, due to reduced photosynthetic activity, such regulation may eventually ensure the survival of the plant under long‐term stress conditions, possibly by suppressing excessive reactive oxygen species (ROS) (Ling & Jarvis, 2015b; Figs 5A and 6A). Overall, our data further support the notion that the regulation of chloroplast protein import is crucial for plant stress responses and development, to meet the requirement for continuous remodeling of the organellar proteome (Demarsy et al, 2014).

However, it is clear that regulation of the TOC apparatus is only one of the roles of autophagy in stress tolerance, as autophagy has been shown to influence diverse cellular compartments under stress conditions (Marshall & Vierstra, 2018). Whether NBR1 interacts with other chloroplast proteins like ATG8 (Fig 3E) remains unknown, although NBR1 is very likely to have other chloroplast targets apart from TOC proteins. Nonetheless, we do expect that ATG8 has a broader range of interactors than NBR1, as ATG8 is very likely to work with other autophagy receptors and adaptors responsible for recruiting different chloroplast cargos.

One obvious question arising from these results concerns how NBR1 is induced to recruit TOC complexes under stress conditions, as persistent degradation of TOC proteins would clearly be harmful to plants (Jarvis et al, 1998; Bauer et al, 2000; Huang et al, 2011). Recently, systematic analyses of the chloroplast ubiquitinome demonstrated broad impacts of ubiquitination on chloroplast proteins (Li et al, 2022; Sun et al, 2022). Notably, all polyubiquitin linkage types were present in chloroplasts, and K63 was shown to be one of the most abundant types, second only to K48 (Sun et al, 2022), implying that ubiquitinated chloroplast proteins can have fates other than proteasomal degradation. Ubiquitination is a key mechanism governing cellular signaling, and K63‐linked polyubiquitination is particularly involved in the determination of autophagic substrate specificity in mammals and plants (Kwon & Ciechanover, 2017; Saeed et al, 2023). However, little is known about the role of K63 ubiquitination in plant autophagy, and it remained unclear how chloroplast protein cargos are regulated by this modification. The present study sheds light on these aspects, as it revealed that stresses like UVB exposure can cause intensive K63 ubiquitination of a key TOC component, Toc159. Interestingly, K48 ubiquitination of Toc159 was also largely enhanced by UVB, consistent with the previous finding that degradation of TOC proteins by the UPS is activated by abiotic stress (Ling & Jarvis, 2015b).

NBR1 is characterized by two domains: a ubiquitin‐binding UBA domain that binds ubiquitinated proteins, and an AIM domain that links cargos to autophagic membranes (Matsumoto et al, 2011; Svenning et al, 2011; Ustun et al, 2018). Intriguingly, plant NBR1 is frequently shown to target the substrates without its UBA sequences (Ji et al, 2020; Thirumalaikumar et al, 2021), so whether the interaction of plant NBR1 and its substrates depends on ubiquitination is still in doubt. However, we showed that deletion of the UBA domain in NBR1 indeed impaired the association between NBR1 and TOC proteins (Fig 4D). Furthermore, while expression of full length NBR1 was able to complement the heat tolerance defect in nbr1‐2, NBR1–ΔUBA failed to do so (Appendix Fig S13). These results suggest that ubiquitin binding is also crucial for substrate recognition and physiological function of plant NBR1, similar to its mammalian homologs p62 and NBR1. p62 preferentially recognizes K63 linkages through its UBA domain (Richter et al, 2016). Thus, K63 modification of Toc159 may lead to its exposure to the autophagy machinery, facilitating its recognition as a substrate. Ubiquitin linkage types are thought to be defined by specific E2/E3 combinations (Liwocha et al, 2021). Two E3 ligases has been uncovered to directly target chloroplast resident proteins, SP1 and PUB4 (Woodson et al, 2015). Although PUB4 was shown not to be responsible for autophagy (Woodson et al, 2015), SP1 mediates ubiquitination of TOC components (Ling et al, 2012), which may consequently promote their recognition by NBR1 and autophagic degradation. However, altering SP1 expression levels does not have obvious impacts during high light and heat stresses (Ling & Jarvis, 2015b). Therefore, it is very likely that other unknown E3 ligases may also exist to control chloroplast proteins.

Thus, our findings uncover a connection between the UPS and autophagy pathways regarding chloroplast import regulation for plant development and stress response (Appendix Fig S14). Why are TOC proteins regulated by two proteolytic pathways? One possibility is that the UPS and autophagy operate in a mutually compensatory manner to control the TOC complex. Such dual‐control scenarios provide regulatory resilience and often occur in important cellular processes in mammals, such as ER stress response (Kocaturk & Gozuacik, 2018). In agreement with this idea, treatment of autophagy mutants with proteasome inhibitor resulted in even higher levels of TOC proteins (Fig 2B), indicating that the UPS and autophagy may act collaboratively to control TOC proteins. Another possibility is that the UPS and autophagy have their own priorities, depending on the severity or type of stress, as was reported in mitochondria (Scarffe et al, 2014). It is noteworthy that UPS activity is mainly limited to degrading individual proteins, whereas autophagy can degrade protein aggregates, multiprotein complexes, or even whole organelles (Takahashi & Arimoto, 2021). As TOC components exist in a complex, autophagy might work more efficiently by delivering quick degradation of the whole TOC complex. This might happen under some stresses, for example, extreme light with high‐UVB damage, which potentially have consequences so severe or acute that they overwhelm the UPS‐dependent mechanism (Ling & Jarvis, 2015b). It is even possible that the turnover of TOC complexes is a prerequisite for the selective clearance of whole organelles by autophagy (Izumi et al, 2017), through a similar mechanism to that reported in mitochondria (Bertolin et al, 2013). However, in our stress conditions, no obvious whole‐chloroplast autophagy could be observed, for example, as indicated by the analysis of cpGFP transgenic lines (Appendix Fig S11). This is possibly because our UVB stress treatment was milder than conditions used previously (Izumi et al, 2017), and does not cause bursts of ROS strong enough to trigger degradation of the whole damaged organelle.

In conclusion, our work has uncovered how NBR1‐mediated selective autophagy plays a crucial role in the regulation of key TOC components and chloroplast protein import, which potentially contributes to plant stress tolerance. It also opens up the possibility that other chloroplast proteins may be cargos of NBR1 or other selective autophagy adaptors or receptors. This discovery reveals expanded complexity of chloroplast protein quality control systems, which are fundamental in plant physiology. These findings also suggest that the manipulation of selective chloroplast autophagy has potential applications in agriculture and in delivering carbon neutrality, through plant improvement (e.g., by enhancing environmental adaptation).

Materials and Methods

Plant materials and growth conditions

Arabidopsis thaliana Columbia‐0 (Col‐0) was used as the wild‐type line in this study. The autophagy‐deficient mutants atg7‐2 (GK655B06) and nbr1‐2 (GABI_246H08) were obtained from NASC (https://arabidopsis.info/). For the fluorescence microscopy analysis, pToc33:YFP‐Toc33 and pToc75:YFP‐Toc75 were crossed with atg7‐2 to generate the pToc33:YFP‐Toc33 atg7‐2 and pToc75:YFP‐Toc75 atg7‐2 lines. After stratification at 4°C for 3 days, plants were grown at 22°C under long day conditions (16 h light [100 μE/m²/s] and 8 h dark cycles) on Murashige & Skoog (MS) medium supplemented with 1% sucrose and 0.8% agar in petri plates. For the induction of autophagy, 8‐day‐old plants were transferred onto sucrose‐free MS liquid medium under dark conditions with bortezomib (5 μM), E64d (100 μM) or concanavalin A (ConA, 2 μM) and incubated for one to two additional days.

Gene identifiers

Gene sequences for the following proteins from A. thaliana were used experimentally in this study: NBR1 (At4g24690), Toc33 (At1g02280), Toc75 (At3g46740), Toc159 (At4g02510), OEP7 (At3g52420), ATG8 (ATG8a) (At4g21980), FAX1 (At3g57280) and Rubisco small subunit (SSU) (At1g67090).

Plasmid constructs

All primers used are listed in Table EV1. All Arabidopsis coding sequences (CDSs) were PCR‐amplified from Col‐0 cDNA, while CDSs of YFP and mScarlet were amplified from the p2YGW7 (Karimi et al, 2007) and NLS‐3mScarlet vectors (Andersen et al, 2018), respectively. The Gateway cloning system (Invitrogen) was used for most plasmid construction, and all entry clones were verified by DNA sequencing. For protoplast transient expression analyses, NBR1–YFP and FAX1–YFP were generated using the p2GWY7 vector; YFP–Toc33 and YFP–Toc159 was generated using the p2YGW7 vector (Karimi et al, 2007); NBR1–CFP was generated using the p2GWC7 vector (Karimi et al, 2007); YFP tag was inserted in between Toc75 transit peptide and the N terminus of its mature part as previously reported (Chen et al, 2016; Paila et al, 2016) using the InFusion cloning system (Clontech), and then cloned into the P2GW7 vector (Karimi et al, 2007) to generate the YFP‐Toc75 construct; haemagglutinin (HA)‐tagged NBR1 (NBR1‐HA) was generated using a modified p2GW7 vector (Karimi et al, 2007) providing a C‐terminal HA tag. Construct for the protoplast expression of free YFP was generated by cloning the YFP CDS into pBlueScript II SK− by using 5′ SalI and 3′ XbaI restriction sites.

To generate stable transgenic plants overexpressing NBR1 (NBR1–OX) or YFP, the corresponding CDSs without stop codon were amplified, and then cloned into the pH2GW7‐based 35S‐driven binary vector (Karimi et al, 2007) providing a C‐terminal HA tag (generating the p35S:NBR1–HA and p35S:YFP–HA constructs). Similar procedure was employed to generate the p35S:NBR1–ΔUBA–HA construct (residues 1–612) using the NBR1 CDS without the C terminal ubiquitin binding domains (Svenning et al, 2011). To generate C‐terminal mScarlet fused NBR1 (p35S:NBR1‐mScarlet) for plant stable transformation, coding sequence of NBR1 was amplified, and then cloned into a modified pH2GW7 vector providing a C‐terminal mScarlet tag. For transgenic plants stably expressing TOC33 or TOC75 genes driven by their native promotors, constructs were cloned into the binary vector pCambia1301 by using 5′ SalI and 3′ XbaI restriction sites through the InFusion cloning system. Promotor region of TOC33 (~ 2,000 bp upstream region) amplified from Col‐0 genomic DNA, together with the CDSs of YFP and TOC33, were used to generate pToc33:YFP–Toc33; and promotor region (~ 2,000 bp upstream region) plus transit peptide region of TOC75 amplified from Col‐0 genomic DNA, together with the CDSs of YFP and TOC75 mature part, were used to generate pToc75:YFP–Toc75.

The InFusion cloning system was used for most BiFC plasmid construction. CDS of NBR1 was cloned into KpnI site of the pSAT4A‐cEYFP‐N1 vector (Tzfira et al, 2005) for NBR1–cYFP; CDS of TOC33 or TOC159 was cloned into XhoI site of the pSAT4‐nEYFP‐C1 vector (Tzfira et al, 2005) for nYFP–Toc33 or nYFP–Toc159, respectively; and CDS of OEP7 was cloned into KpnI site of the pSAT4A‐nEYFP‐N1 vector (Tzfira et al, 2005) for OEP7–nYFP. The nYFP tag was inserted in between Toc75 transit peptide and the N terminus of its mature part using a similar procedure as described above for making YFP–Toc75, to generate the nYFP–Toc75 construct through the p2GW7 vector using the Gateway cloning system.

Transient assays and stable plant transformation

Protoplast isolation and transient assays were carried out as described previously (Wu et al, 2009; Ling et al, 2019). For YFP fluorescence and immunoprecipitation assays, 0.1 mL (10⁵) or 1 mL (10⁶) of protoplast were transfected with 5 μg or 100 μg of DNA, respectively; the fluorescence signals were analyzed after 15–18 h. When required, Bortezomib (Selleckchem) or ConA (APExBIO) was added to the protoplast culture medium to a final concentration of 5 or 1 μM, respectively.

The transgenic plants were generated by Agrobacterium‐mediated transformation (Huang et al, 2011). Transformants were selected using MS medium containing 50 μg/ml hygromycin B (for the pCambia1301‐ or pH2GW7‐based constructs). Approximately 10 independent T2 lines were analyzed per construct, and one representative line with a single T‐DNA insertion (which showed a 3:1 segregation on selective MS medium in the T2 generation) were chosen for further analysis.

UVB treatment

To examine UVB stress tolerance in Arabidopsis, plants were grown in soil under normal growth condition for 8 days, and then treated with white light (18 μE/m²/s) plus broadband UVB (5 μE/m²/s) as previous described (Liang et al, 2020) for 3 h. The treated plants were then allowed to recover for 7 days under normal growth condition prior to phenotypical analysis. Alternatively, treated plants were recovered for up to 1 day for immunoblotting analysis, to avoid strong growth defect which may lead to secondary effect. For the analyses of photosynthetic performance (F _v/F _m), chloroplast ultrastructure (TEM) and in vivo protein import (FRAP), 8‐day‐old plants grown in petri plates were treated with broadband UVB for 4 h, and recovered for 1 day before further analyses.

Heat treatment

To examine heat tolerance stress in Arabidopsis, plants were grown in soil under normal growth condition for 4 weeks, then incubated at 42°C for 17 h, and finally recovered under normal growth condition for 2 days before phenotypical analysis. For short‐term heat stress analysis, 8‐day‐old plants grown in petri plates were incubated at 42°C for 4 h, and then recovered for up to 1 day before immunoblotting analysis.

Measurement of photosynthetic performance

The UVB treated plants grown on agar plates were recovered for up to 1 day, and leaves were dark adapted for a minimum of 30 min before the photochemical activity was examined in individual leaves. Maximum photochemical efficiency of photosystem II (F _v/F _m) was determined by measuring chlorophyll fluorescence with a fluorescence imaging system (Closed FluorCam FC 800‐C/1010 GFP, Photon Systems Instruments, Czech Republic) as described previously (Liu et al, 2020). Three experiments were performed, and approximately 10 seedlings were analyzed per genotype in each experiment.

Chlorophyll measurements

To measure chlorophyII content, UVB‐treated plants grown in soil were recovered for up to 1 day, then harvested, weighted, frozen in liquid nitrogen and ground to powder using a motor with precooled pestle. Chlorophylls were extracted from ~ 10 mg frozen tissue powder in 1 ml 80% acetone through vigorous vortexing, followed by incubation at 4°C for 30 min with slow rotation. The extract was centrifuged at 12,000 g for 10 min, and the supernatant was used for chlorophyII determination. Chlorophyll concentrations were measured using a spectrophotometer as described previously (Lichtenthaler, 1987).

Microscopy

Fluorescence images were captured using TCS SP8 (Leica Ltd.) or SpinSR (Olympus Ltd.) laser‐scanning confocal microscope, as described previously (Hu et al, 2022; Liu et al, 2022). All experiments were conducted at least twice with the same results, and typical images are presented. All images were captured using the same settings to enable comparisons. For the observation of autophagosomes, Arabidopsis protoplasts expressing the YFP‐tagged TOC proteins were analyzed by confocal microscopy following ConA treatment for 4 h. Alternatively, 8‐day‐old transgenic Arabidopsis seedlings stably expressing YFP‐tagged TOC proteins were treated with ConA for 2 days, and then analyzed by confocal microscopy.

Transmission electron microscopy was performed as described previously (Huang et al, 2011). Measurements were taken using at least 30 different chloroplasts per genotype and are representative of three individuals per genotype. Chloroplast shape was determined as described (Huang et al, 2011) based on the length/width ratio. Numbers of thylakoid lamellae per granal stack were counted as previously described (Huang et al, 2011; Ling et al, 2012). Numbers of starch granules were estimated as described (Crumpton‐Taylor et al, 2012).

Chloroplast isolation and in vitro protein import assay

Chloroplasts were isolated from 12‐day‐old in vitro grown plants (or, when stated, from protoplasts). Isolations and protein import were performed as described previously (Fitzpatrick & Keegstra, 2001; Aronsson & Jarvis, 2002; Huang et al, 2011). The presented import data are representative for four independent experiments.

SDS – PAGE, immunoblotting and immunoprecipitation

Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE) and immunoblotting were performed essentially as described before (Kovacheva et al, 2005). Primary antibodies were as follows. To detect TOC or TIC proteins, we employed: anti‐atToc75‐III POTRA‐domain antibody (Huang et al, 2011); anti‐atToc159 A‐domain antibody (Bauer et al, 2000); anti‐atToc33 pepitide antibody (Aronsson et al, 2003); anti‐atTic110 stromal antibody (Inaba et al, 2005); anti‐atTic40 stromal antibody (Kasmati et al, 2011). To detect proteins of other cellular compartments, we employed: ATG8 (cytosol; AS14 2769, Agrisera); NBR1 (cytosol; AS142805, Agrisera); PEPC (cytosol; AS09458, Agrisera) anti‐Actin (cytosol; LF208, Epizyme); anti‐PsaD (PSI subunit; PAB01004, Orizymes); anti‐PsbP (PSII subunit; PA02014, Orizymes) and anti‐AtpA (chloroplast thylakoid; PAB04001, Orizymes). Other primary antibodies employed were: K48‐linkage Specific Polyubiquitin antibody (8081S, CST); K63‐linkage Specific Polyubiquitin antibody (5621S, CST); anti‐HA tag antibody (sigma, H6908); anti‐GFP antibody (detects both GFP and YFP; Sigma, SAB4301138). We used Tic110 or Actin as loading controls.

Secondary antibodies were polyclonal goat anti‐rabbit IgG conjugated with horseradish peroxidase (Santa Cruz Biotechnology). Chemiluminescence was detected using ECL Plus Western Blotting Detection Reagents (GE Healthcare) and a Tanon 5200 imager (Tanon). Band intensities were quantified using ImageJ. Quantification data were from at least three experiments which showing a similar trend. Typical images are shown in all figures.

For the immunoprecipitation of NBR1–HA and NBR1‐ΔUBA‐HA, p35S:NBR1–HA and p35S:NBR1‐ΔUBA‐HA transgenic plants were used for subsequent analysis. To stimulate the recruitment of TOC components by NBR1, eight‐day‐old transgenic seedlings were exposed to UVB for 4 h before protein extraction. Total protein (~ 100 mg) was then extracted from these seedlings in IP buffer (25 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 10% glycerol, 1% Triton X‐100) containing 0.5% PIC (Sigma) and 5 μM bortezomib, and then centrifuged at 10,000 g for 10 min at 4°C. The clear lysate was then incubated with 50 μl anti‐HA beads (Thermofisher) at 4°C for 4–6 h with slow rotation. After three washes with 500 μl washing buffer (25 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% Triton X‐100), bound proteins were eluted by boiling in 2× SDS–PAGE loading buffer (50 mM Tris–HCl, pH 6.8, 20% glycerol, 1% SDS, 0.1 M DTT) for 5 min, and analyzed by SDS–PAGE and immunoblotting. Transgenic plants of p35S:YFP–HA were set as a negative control. A similar procedure was adopted for the immunoprecipitation of YFP‐tagged proteins, except that 50 μl anti‐GFP magnetic beads (ChromoTek) was used to precipitate YFP tagged proteins from isolated chloroplasts.

Ubiquitination of Toc159 in vivo was assessed as described previously (Ling et al, 2012), except that the protein was extracted from 8‐day‐old WT plants grown in petri plates treated with or without UVB for 4 h, and that anti‐atToc159 was used to precipitate endogenous Toc159 protein (or its preimmune serum as a negative control). Ubiquitinated Toc159 was detected by anti‐K48‐ubiquitin and anti‐K63‐ubiquitin antibodies.

Mass spectrometry analysis and function annotation

For the IP of YFP–ATG8, 1.5 ml protoplasts were used for the transient expression of YFP–ATG8 or free YFP as a negative control. Transfected protoplasts were let sit in the dark for 15 h followed by incubation with 5 μM bortezomib and 1 μM ConA for another 4 h. Protoplasts were equilibrated with HEPES‐sorbitol buffer (50 mM HEPES‐KOH pH 7.3, 330 mM sorbitol), and chloroplasts were isolated from the protoplasts as described previously (Fitzpatrick & Keegstra, 2001). IP was conducted as described above, and the eluted protein samples were analyzed by LC–MS/MS, which was performed by Shanghai Luming Biotechnology Co., Ltd.

Immunoprecipitated proteins were denatured and alkylated with 30 μl urea buffer (8 M urea, 50 mM ammonium bicarbonate [AB], 10 mM DTT, 0.2% protease inhibitor [Sigma, P8340]), and incubated at 55°C for 30 min, followed by adding 15 mM iodoacetamide aqueous solution for another 30 min in the dark at room temperature. The digestion solution was diluted to < 1 M urea upon addition of trypsin (Promega, V511A; 200 ng in 300 μl of 1 M urea, 25 mM AB, 1 mM CaCl₂), prior to incubation at 37°C overnight. After digestion, the eluate fractions were collected by centrifugation at 14,000 g for 5 min at 25°C.

Resulting tryptic peptides were analyzed on an EASY‐nano‐LC 1200 system (Thermo Fisher) connected to a Q Exactive mass spectrometer (Thermo Fisher) through an EASY‐Spray nano‐electrospray ion source (Thermo Fisher). The peptides were initially trapped on an RP‐C18 Trap column (100 μm i.d. × 20 mm, Thermo Fisher) using solvent A (0.1% formic acid in water). Trapped peptides were separated on an RP‐C18 Analysis column (75 μm i.d. × 150 mm, Thermo Fisher) using a linear gradient (length: 60 min, 18–30% solvent B [0.1% formic acid, 5% DMSO in acetonitrile], flow rate: 300 nl/min). The separated peptides were electrosprayed directly into the mass spectrometer operating in a data‐dependent mode. Full scan MS spectra were acquired in the Orbitrap (scan range 300–1,600 m/z, resolution 35,000, AGC target 3e6, maximum injection time 50 ms). After the MS scans, the 10 most intense peaks were selected for HCD fragmentation at 28% of normalized collision energy. HCD spectra were also acquired in the Orbitrap (resolution 17,500, AGC target 5e4, maximum injection time 50 ms).

The MS raw data for each sample were combined and searched using ProteomeDiscover (v2.4, Thermo Fisher) software against an Arabidopsis UniProt database (UniProt, 2021). Trypsin was chosen as the enzyme with a maximum of two missed cleavages allowed. Precursor and fragment mass error tolerances were set at 20 ppm and 0.1 Da, respectively. All matched MS/MS spectra were filtered by mass accuracy and matching scores to reduce protein FDR to ≤ 1%, based on the target‐decoy strategy using a reversed database. Relative protein quantification was performed through label‐free quantification analysis and the search engine was Andromeda. Quantitation was carried out only for proteins with two or more unique peptide matches.

Proteins identified by LC–MS/MS were manually annotated with respect to subcellular localization, and function. Information on protein subcellular localization was searched for sequentially and manually in several public databases (TAIR (Lamesch et al, 2012), UniProt (UniProt, 2021), NCBI (NCBI Resource Coordinators, 2018), PPDB (Sun et al, 2009)) and in the appropriate literature, in order to annotate well‐characterized chloroplast proteins. Alternatively, TargetP (Almagro Armenteros et al, 2019) prediction tools were used to test for the presence of a transit peptide (a typical feature of chloroplast proteins); or the SUBA4 database (Hooper et al, 2017) was used, which provides information on subcellular localization from published proteomic or fluorescent protein fusion datasets.

FRAP assay

FRAP of Arabidopsis hypocotyl chloroplasts was performed on a Leica TCS SP8 microscope using a 60×/NA1.2 water objective. Chloroplasts were bleached using a laser intensity of 20% at 488 nm with three repeats. Images were acquired using LAX software (Leica). Fluorescence recovery was recorded continuously for 300 s (3 s interval with 100 repeats) after bleaching. Analyses of the fluorescence intensity of bleached region, reference region and background region were carried out using cellSens imaging software (Olympus), and recovery curves were drawn using the cellSens FRAP plugin. The graph shown were finally drawn with GraphPad Prism software.

Statistical analysis

Statistical calculations (mean, standard error of the mean, t‐test) were performed using GraphPad Prism software. Statistical significance of differences between two experimental groups was assessed by using an unpaired two‐tailed Student's t‐test. Differences between two datasets were considered significant at P < 0.05.

Author contributions

Qihua Ling: Conceptualization; data curation; supervision; funding acquisition; validation; writing – original draft; project administration; writing – review and editing. Chen Wan: Conceptualization; data curation; formal analysis; investigation; visualization; methodology; project administration. Hui Zhang: Conceptualization; data curation; formal analysis; investigation; visualization; methodology; project administration. Hongying Cheng: Investigation; methodology. Robert G Sowden: Investigation; methodology. Wenjuan Cai: Investigation; visualization; methodology. R Paul Jarvis: Supervision; validation; writing – review and editing.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Table EV1

Movie EV1

Movie EV2

Movie EV3

Movie EV4

Movie EV5

Movie EV6

Dataset EV1

Source Data for Expanded View and Appendix

PDF+

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

Source Data for Figure 7

Source Data for Figure 8

The EMBO Journal (2023) 42: e112534

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Appendix. The mass spectrometry proteomics data have been deposited to the Integrated Proteome Resources (iProX, https://www.iprox.cn/) with the identifier PXD040844.