Cotton leaf curl Multan virus βC1 Protein Induces Autophagy by Disrupting the Interaction of Autophagy-Related Protein 3 with Glyceraldehyde-3-Phosphate Dehydrogenases

Autophagy plays an important role in plant–pathogen interactions. Several pathogens including viruses induce autophagy in plants, but the underpinning mechanism remains largely unclear. Furthermore, in virus– plant interactions, viral factor(s) that induce autophagy have yet to be identified. Here, we report that the βC1 protein of Cotton leaf curl Multan betasatellite (CLCuMuB) interacts with cytosolic glyceraldehyde-3-phosphate dehydrogenase (GAPC), a negative autophagic regulator, to induce autophagy in Nicotiana benthamiana . CLCuMuB βC1 bound to GAPCs and disrupted the interaction between GAPCs and autophagy-related protein 3 (ATG3). A mutant βC1 protein (βC1 3A ) in which I45, Y48, and I53 were all substituted with alanine (A), had a dramatically reduced binding capacity with GAPCs, failed to disrupt the GAPCs-ATG3 interactions and failed to induce autophagy. Furthermore, mutant virus carrying βC1 3A showed increased symptoms and viral DNA accumulation associated with decreased autophagy in plants. These results suggest that CLCuMuB βC1 activates autophagy by disrupting GAPCs–ATG3 interactions.


Introduction
Autophagy is a highly conserved mechanism that leads to the degradation and recycling of damaged or unwanted intracellular materials under stress conditions or during specific developmental processes (Klionsky and Codogno, 2013;Yin et al., 2016). Three major types of autophagy, i.e., macroautophagy, microautopahgy and chaperone-mediated autophagy, occur in eukaryotic cells (Klionsky, 2004;Massey et al., 2004).
Macroautophagy (hereafter referred to as autophagy) is mediated by autophagosome, a de novo-formed double-membrane vesicle. The outer membrane of a mature autophagosome fuses with the lysosome (in mammals) or vacuole (in yeast and plants) to release the sequestered cargo for breakdown by acid hydrolases (Mizushima et al., 2008;Ohsumi, 2001).
Increasing evidence shows that viruses manipulate or hijack the autophagy pathway to promote pathogenesis, demonstrating the pivotal role of autophagy in the arms race between hosts and viruses (Shoji-Kawata and Levine, 2009;Dong and Levine, 2013). In plants, autophagy plays an important role in disease resistance or susceptibility to various pathogens (Han et al., 2011;Ismayil et al., 2019). It contributes to hypersensitive response cell death, but restricts the spread of programmed cell death beyond the initial infection site (Liu et al., 2005;Hofius et al., 2009;Yoshimoto et al., 2009;Patel and Dinesh-Kumar, 2008). Cytoplasmic glyceraldehyde-3-phosphat dehydrogenases (GAPCs) interacts with autophagy-related protein 3 (ATG3) directly and negatively regulates autophagy and immunity. Silencing of GAPCs reduced TMV accumulation in N gene-containing plants and enhanced N gene-mediated cell death (Han et al., 2015). Autophagy has also recently been shown to represent an antiviral mechanism in plants (Nakahara et al., 2012;Haxim et al., 2017;Hafrén et al., 2017;Li et al., 2018;Hafrén et al., 2018). In addition, autophagy may directly or indirectly promote virus infection (Hafrén et al., 2017;Fu et al., 2018).
Given that autophagy has an important role in host-virus interactions, it is not surprising that some viruses have evolved strategies that modulate host autophagy for their own benefit. For instance, Turnip mosaic virus (TuMV) antagonizes NBR1-dependent autophagy during infection, thereby limiting its antiviral capacity. NBR1-independent bulk autophagy prevents premature plant death, extending the lifespan of virus reservoirs and particle production . Cauliflower mosaic virus (CaMV) P6 protein disrupts the interaction between viral P4 and host NBR1 to protect viral replication factory inclusions from autophagic degradation (Hafrén et al., 2017). Barley stripe mosaic virus (BSMV) γb interferes with the interaction of ATG7 with ATG8 in a competitive manner to suppress autophagy, thereby promoting viral infection (Yang et 3 al., 2018). Viral proteins can also promote autophagic degradation of components in RNA silencing pathway (Derrien et al., 2012;Cheng and Wang, 2017).
Betasatellites, such as Cotton leaf curl Multan betasatellite (CLCuMuB), are approximately half the size of the DNA genomes of their helper begomoviruses. They require the helper begomoviruses for replication and movement in plants and only encode a single multifunctional pathogenicity βC1 protein (Sattar et al., 2013). CLCuMuB βC1 functions as a viral suppressor of RNA silencing (VSR) (Amin et al., 2011). CLCuMuB βC1 enhances the accumulation of its helper virus CLCuMuV and is required for the induction of viral symptoms in plants (Jia et al., 2016). Geminivirus βC1 can also subvert ubiquitination to assist their helper viruses to infect plants (Jia et al., 2016;Shen et al., 2016). We reported that autophagic machinery targets and degrades viral βC1 through its interaction with autophagy-related protein 8 (ATG8) (Haxim et al., 2017).
Although we know that CLCuMuV infection activates autophagy and enhances the autophagic flux (Haxim et al., 2017), the viral factor responsible for autophagy activation and the underlying mechanism are unknown.
On the other hand, it is well-established that various plant pathogens can trigger autophagy (Liu et al., 2005;Hofius et al., 2009;Haxim et al., 2017;Hafrén et al., 2017;Li et al., 2018). However, in plants only one bacterial and one fungal protein have been shown to be responsible for autophagic activation, but the underlying mechanism(s) remain(s) to be characterized (Dagdas et al., 2016;Üstün et al., 2018). In addition, no viral protein has been reported to activate autophagy in plants. In this study, we demonstrate that CLCuMuB βC1 activates autophagy by interfering with the interaction between NbGAPCs and NbATG3.

CLCuMuB βC1 activates plant autophagy.
We previously showed that CLCuMuV infection induces autophagy but did not identify any viral protein responsible for this induction (Haxim et al., 2017). Here, we investigated whether βC1 affected autophagy in Nicotiana benthamiana. To achieve this, we used Cyan Fluorescent Protein (CFP)-tagged NbATG8f (CFP-NbATG8f) as a marker to visualize autophagosomes in the cytoplasm as well as autophagic bodies in the vacuole (Wang et al., 2013). In plants expressing HA-tagged βC1 (HA-βC1), numerous CFP-labeled autophagic bodies were observed compared to plants expressing the HA-tagged C-terminal luciferase (HA-cLUC) control ( Figures 1A). The relative autophagy level in HA-βC1 expressing plants was 2-to 3-fold greater than the control ( Figure 1B). Transmission electron microscopy confirmed βC1-mediated autophagic induction, and showed that overexpression of HA-βC1 increased the number of autophagic structures compared to the control ( Figure 1C and 1D). In addition, HA-βC1 expression reduced the accumulation of the 4 autophagy substrate, Joka 2/NBR1 (see below). These results indicated that CLCuMuB βC1 induces autophagy in N. benthamiana.

CLCuMuB βC1 Interacts with NbGAPCs.
Since CLCuMuB βC1 interacts with the autophagy-related protein ATG8 (Haxim et al., 2017), we investigated whether βC1-mediated autophagic induction was dependent on the βC1-ATG8 interaction. For this, we tested whether βC1 V32A , a βC1 point mutant that cannot interact with ATG8, can induce autophagy in plants. Expression of HA-βC1 or HA-βC1 V32A resulted in a similar level of CFP-labeled autophagic bodies (Supplemental Figure 1), revealing that βC1-mediated autophagic induction does not depend on its interaction with ATG8.
To identify the mechanism underlying activation of autophagy by βC1, we employed affinity purification using GFP-Trap followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis to identify plant proteins that interact with βC1 (see Methods for details). When LC-MS/MS polypeptide profiles were searched against N. benthamiana protein databases, we identified many potential βC1-interacting proteins including autophagy-related protein 8 (ATG8). The ATG8-βC1 interaction has been reported (Haxim et al., 2017), suggesting that our screen is successful. Particularly, the partial sequence of cytosolic glyceraldehyde-3-phosphate dehydrogenase (GAPC) was identified with a high score (Supplemental Figure   2). Interestingly, this peptide sequence exists in all three N. benthamiana GAPC homologues (NbGAPC1-3) (Han et al., 2015).
Positive signals were detected when βC1 was combined with all three NbGAPCs (Figure 2A), due to the reconstitution of the luciferase activity by βC1-NbGAPCs interactions. We also detected the signals in the positive control (NbATG3-nLUC/NbGAPC1-cLUC). However, no signal was detected in any of the negative controls (HA-cLUC + nLUC-βC1) ( Figure 2A). We further confirmed the interaction of βC1 with NbGAPCs by co-immunoprecipitation (Co-IP) assays. HA-βC1 co-immunoprecipitated with both NbGAPC1-GFP and NbGAPC2-GFP fusion proteins. However, HA-cLUC did not co-immunoprecipitate with NbGAPC1-GFP or NbGAPC2-GFP ( Figure 2B).
To identify the subcellular localization of the βC1-NbGAPCs interactions, citrine yellow fluorescent protein (YFP)-based Bimolecular Fluorescence Complementation (BiFC) assays (Han et al., 2015) were 5 performed. βC1 was fused with the N-terminal domain of YFP (nYFP) to generate nYFP-βC1 and GAPC1 and GAPC2 were tagged with the C-terminal domain of YFP (cYFP) to generate NbGAPC1-cYFP and NbGAPC2-cYFP, respectively. nYFP-βC1 and 35S-nYFP were transiently co-expressed with NbGAPC1-cYFP or NbGAPC2-cYFP in N. benthamiana. Positive interactions (yellow fluorescence) were observed for nYFP-βC1 with both NbGAPC1-cYFP and NbGAPC2-cYFP, and such interactions occurred in the cytoplasm of leaf cells ( Figure 2C). To confirm the specific localization of this interaction, we performed cell fractionation and immunoblotting assays. Total protein extracts from leaves co-expressing HA-βC1 and NbGAPC1-GFP were separated into soluble and microsomal membrane fractions by ultracentrifugation.
Both βC1 and NbGAPC1 were mostly detected in the soluble fraction ( Figure 2D).
Together, these results suggest that CLCuMuB βC1 specifically interacts with NbGAPCs.

CLCuMuB βC1 interferes with the interaction between NbGAPCs and NbATG3.
GAPCs interact with autophagy-related protein 3 (ATG3) to negatively regulate autophagy in plants (Han et al., 2015), whilst CLCuMuB βC1 induces autophagy and interacts with NbGAPCs as shown above. Based on these observations, we hypothesized that βC1 could interfere with the interaction of ATG3 with GAPCs to relieve the negative regulatory role of GAPCs on autophagy. To test this hypothesis, we examined the effect of βC1 on ATG3-GAPC interactions using BiFC assays. For this, we co-expressed NbATG3-nYFP and NbGAPC1-cYFP, or NbGAPC2-cYFP with HA-βC1 or HA-cLUC in N. benthamiana leaves. Stronger signals were detected in tissue expressing NbATG3YFP and NbGAPC1-cYFP or NbGAPC2-cYFP in the presence of HA-cLUC than HA-βC1 ( Figure 3A and 3B). The protein levels of NbGAPC1-cYFP, NbGAPC2-cYFP and NbATG3-nYFP were similar between the two groups ( Figure 3C).
We further confirmed these results using a co-IP assay. For this, we co-expressed NbATG3-Myc with NbGAPC1-GFP or NbGAPC2-GFP in the presence of HA-βC1 or HA-cLUC in N. benthamiana.

The βC1-GAPC interaction is important for βC1-induced autophagy
To explore the potential role of the βC1-NbGAPCs interaction in autophagy activation, we used βC1 deletion mutants to characterize the region required for binding to NbGAPCs in a yeast two-hybrid assay.
Given that βC1 3A interacts weakly with NbGAPC1, we hypothesized that the βC1 3A mutant may not interfere with the interaction between NbGAPC1 and NbATG3. To test this, we performed a co-IP assay in which NbGAPC1-GFP was co-expressed with NbATG3-Myc in the presence of HA-βC1 or HA-βC1 3A in N.
benthamiana. Less NbATG3-Myc was co-immunoprecipitated by NbGAPC1-GFP in the presence of HA-βC1 but not with HA-βC1 3A compared to that in the presence of the HA-nLUC control ( Figure 4A). We also tested the effect of several other mutations on the ATG3-GAPC interaction. Mutations (I45A, Y48A, I53A) in βC1 have little effect on the ATG3-GAPC interaction, while a strong effect was observed with the Y42A mutation (Supplemental Figure 8; Figure 4A). BiFC assays confirmed that, in contrast to wild-type βC1, βC1 3A failed to disrupt the ATG3-GAPC interactions ( Figure 4B and 4C). The protein levels of NbGAPC1-cYFP, NbGAPC2-cYFP, and NbATG3-nYFP were similar in the BiFC assays ( Figure 4D).
Because the natural host of CLCuMuV is cotton, we tested the interaction between βC1 and cotton GAPC (GhGAPC). Indeed, βC1 also bound to GhGAPC and disrupted the interaction between GhGAPC and cotton ATG3 (GhATG3). Co-IP assays showed that HA-βC1, but not cLUC, was co-immunoprecipitated with GhGAPC-GFP fusion protein. Further, HA-βC1 3A dramatically decreased the βC1-GhGAPC interaction (Supplemental Figure 9A). Moreover, less GhATG3-Myc was co-immunoprecipitated by GhGAPC-GFP in the presence of HA-βC1 but not with HA-βC1 3A (Supplemental Figure 9B).
These results indicated that the interaction of βC1 with GAPC is critical for disruption of the interaction of ATG3 with GAPC.
In addition, we examined the role of the βC1-NbGAPCs interaction in βC1-mediated autophagy induction using the βC1 3A mutant. Few autophagic bodies were observed in leaf tissue agroinfiltrated with HA-βC1 3A or the HA-cLUC control. By contrast, a high level of autophagic bodies accumulated in leaf tissue expressing HA-βC1 ( Figure 5A and 5B). We also measured the autophagy flux by quantifying the autophagy substrate Joka2/NBR1, which is degraded in the vacuole in an autophagy-dependent manner (Xu et al., 2017).
These data suggest that the CLCuMuB βC1-GAPCs interaction is important for βC1-mediated autophagy induction.

CLCuMuB βC1 induces autophagy independent of its VSR activity
Since CLCuMuB βC1 also suppresses gene silencing, we investigated whether βC1-mediated autophagy induction is dependent on its activity as viral suppressor of RNA silencing (VSR). We compared the VSR activity of wild-type βC1 and the βC1 3A mutant by co-expressing them with GFP. Analysis of the agroinfiltrated plants under UV illumination at 4 dpi showed that leaf tissues expressing both HA-βC1 and HA-βC1 3A proteins showed similar GFP fluorescence. The immunoblot results indicate that VSR activity was not affected in mutant βC1 3A protein (Supplemental Figure 10). These results indicate that βC1-activated autophagy is independent of its VSR activity.

Disruption of βC1 binding to GAPCs reduces CLCuMuV-induced autophagy and enhances viral infection
To explore the role of the βC1-NbGAPCs interaction during CLCuMuV (CA) infection, we generated a CLCuMuB mutant (β 3A ) virus by replacing βC1 with the βC1 3A mutant and inoculated the mutant virus (CA+β 3A ) or wild-type virus (CA+β) onto N. benthamiana leaves. First, we tested the GAPC-ATG3 interaction during viral infection and found that less NbATG3-Myc was co-immunoprecipitated by NbGAPC1-GFP in plants infected with wild-type virus (CA+β) than mutant virus (CA+β 3A ) (Supplemental Figure 11). At 12 days post inoculation (dpi), the CA+β 3A mutant virus caused more severe leaf curling symptoms compared to wildtype CA+β virus ( Figure 6A), and viral DNA accumulated 4-fold higher in plants infected with CA+β 3A mutant 8 compared to CA+β wild-type ( Figure 6B). The leaf curling symptoms caused by CA+β 3A appeared 2 days earlier than those caused by CA+β ( Figure 6C). Since βC1 3A has a much lesser ability to bind to NbGAPC1 and induce autophagy than βC1, we hypothesize that the enhanced viral infection of this mutant virus is due to decreased autophagic activation. We tested the autophagy level of CA+β 3A -infected plants using the CFP-NbATG8f marker. Less CFP-labeled autophagic bodies were seen in CA+β 3A -infected plants than in CA+βinfected plants ( Figure 6D and 6E). We also measured the autophagy flux in CA+β 3A or CA+β infected plants using Joka2/NBR1 as an indictor. CA+β 3A -infected plants accumulated more Joka2/NBR1 than CA+βinfected plants ( Figure 6F). Interestingly, noninfected control plants accumulated more Joka2/NBR1 and less CFP-labeled autophagic bodies than both CA+β-and CA+β 3A -infected plants ( Figure 6D, 6E and 6F).
We also confirmed that CA+β 3A mutant virus is more infectious than wild-type CA+β virus in cotton plants. At 17 dpi, 60% of CA+β 3A , but only 20% of CA+β infiltrated plants showed disease symptoms. Further, CA+β 3A caused more severe leaf curling symptoms in cotton plants than wild-type CA+β (Supplemental Figure 12A), and viral DNA accumulated 5-fold higher in plants infected with CA+β 3A mutant than CA+β wild-type (Supplemental Figure 12B).
These results suggest that disruption of βC1 binding to GAPCs reduces virus-induced autophagy and enhances virus infection.

Discussion
In this study, we established that CLCuMuB βC1 induces autophagy by interacting with the negative autophagy regulator GAPCs to disrupt the interaction between NbGAPCs and NbATG3. We identify a plant viral protein responsible for the autophagy induction and uncover the underlying mechanistic basis of pathogen-mediated autophagy activation in plants.
Autophagy has been reported to participate in plant-pathogen interactions. Pathogens have evolved the ability to hijack autophagy and degrade host defense components for their own benefits. For instance, the VSR protein P0 of Polerovirus is thought to mediate autophagic degradation of ARGONAUTE 1, a component of the cellular RNAi-based antiviral defense machinery (Derrien et al., 2012). Similarly, the VSR protein VPg of Turnip mosaic virus is also reported to mediate the degradation of the cellular RNAi-based antiviral defense component suppressor of gene silencing 3 (SGS3), partially via autophagy (Cheng and Wang, 2017). The VSR protein βC1 of Tomato yellow leaf curl China virus can mediate autophagic degradation of SGS3 through the calmodulin-like protein rgsCAM (Li et al., 2017). The movement protein of Rice stripe virus (RSV) induces autophagic degradation of remorin and overcomes remorin-mediated inhibition of viral movement by binding 9 to and interfering with the S-acylation of remorin, thus facilitating virus infection (Fu et al., 2018). In these cases, viruses hijack autophagy to target host defense-related factors for degradation. However, host autophagy can also target viral virulence factors for degradation. The VSR protein 2b of Cucumber mosaic virus (CMV) is thought to be indirectly targeted by autophagy for its degradation through its binding to host ATG8-binding protein rgs-CaM (Nakahara et al., 2012). In this study, we show that CLCuMuB βC1 induces autophagy. Since CLCuMuB βC1 is directly targeted by autophagy for its degradation through its direct binding to ATG8 (Haxim et al., 2017), it seems that CLCuMuB βC1 activates autophagy and thereby degrades the viral effector itself.
TuMV VPg and 6K2 proteins inhibit NBR1-dependent autophagy antiviral defense by blocking the degradation of viral protein HCpro . BSMV γb inhibit the interaction between ATG7 and ATG8 to suppress autophagy and promote viral infection (Yang et al., 2018). In contrast to these pathogen effectors that antagonize or inhibit plant autophagy, both CLCuMuV infection and CLCuMuB βC1 activate autophagy (Haxim et al., 2017; this study). Considering TuMV not only hijacks autophagy to degrade plant antiviral protein SGS3 (Cheng and Wang, 2017) but also inhibits autophagy to block the degradation of viral effector (Li et al., 2017), it will be interesting to determine whether there are CLCuMuV protein(s) that suppress autophagy antiviral defense.
Host autophagy can also be induced during plant-pathogen interactions (Liu et al., 2005;Hofius et al., 2009;Haxim et al., 2017;Hafrén et al., 2017). To date, only two pathogen effectors, PexRD54 from Phytophthora infestans and HopM1 from Pseudomonas, have been shown to activate autophagy (Dagdas et al., 2016;Üstün et al., 2018). Interestingly, no viral effectors have been shown to activate autophagy, even though autophagy is activated by the infection of several viruses, including CaMV, TuMV, and geminiviruses (Hafrén et al., 2017;Li et al., 2018;Haxim et al., 2017). More importantly, the mechanism underlying activation of autophagy during plant-pathogen interactions is generally unknown in plants. In this study, we reveal that CLCuMuB βC1 activates autophagy by interfering with the interaction between NbGAPCs with NbATG3 through its binding to NbGAPCs. In this study, we also determined that the autophagy level in mutant CA+β 3A virus-infected plants is higher than that in noninfected control plants. It is possible that other viral 10 factors except βC1 could also induce autophagy. Geminiviral βC1 is a multifunctional pathogenicity protein. βC1 is crucial for producing viral disease symptoms and enhancing viral DNA accumulation (Briddon et al., 2001;Jia et al., 2016), and complements the movement function of some bipartite begomoviruses (Saeed et al., 2007). βC1 can also subvert ubiquitination to enhance virus infection (Jia et al., 2016). Furthermore, βC1 plays important roles in suppressing posttranscriptional gene silencing (PTGS) (Amin et al., 2011). In this study, we showed that the mutant protein βC1 3A , which cannot induce autophagy, has a similar PTGS suppression activity to wild-type βC1 (Supplemental Figure 8). This indicates that CLCuMuB βC1 protein induces autophagy independent of its VSR functionality.
In summary, we describe a first viral factor that is able to activate plant autophagy and the underlying activation mechanism. CLCuMuB βC1 functions as a viral autophagy activator by interfering with the interaction between GAPCs with ATG3 through its interaction with GAPCs. Interestingly, CLCuMuB βC1 is targeted and degraded by autophagy through its interaction with ATG8, and autophagy functions as an antiviral mechanism against CLCuMuV in plants ( Haxim et al., 2017). Strikingly, the mutant CA+ β 3A virus that is unable to induce autophagy causes more severe viral symptoms and increases viral DNA accumulation to a greater extent than wild-type virus ( Figure 6A), consistent with the antiviral role of autophagy (Haxim et al., 2017). Since CLCuMuB βC1 is a viral virulence factor and is targeted by autophagy for degradation, CLCuMuB βC1-mediated autophagy may reduce its viral virulence and enable successful infection during the plant-virus co-evolution. In addition, it is also possible that CLCuMuB βC1-mediated autophagy helps host cell survival for better viral propagation and transmission. Indeed, autophagy is reported to extend plant lifespan and to play a proviral role during some virus infections (Hafrén et al., 2017).

Confocal Microscopy and TEM
Agrobacterium tumefaciens GV3101 harboring the autophagy marker CFP-NbATG8f was infiltrated into leaves. About 48 h later, 20 mM protease inhibitor E-64d was infiltrated and 8 h later the samples were imaged using a Zeiss LSM 710 three-channel microscope with an excitation wavelength of 405 nm, and the emission was captured at 454 to 581 nm. TEM observation was performed as described previously (Wang et al., 2013).

Mass spectrometry analysis
Total protein was extracted as described previously (Han et al., 2015) from N. benthamiana leaves infected with GFP-βC1 or GFP constructs. Total protein extracts were incubated with GFP-Trap_A beads (ChromoTek). After washing, the purified GFP-tagged βC1 proteins were denatured at 98˚C, separated on a SDS-PAGE gel (12%), and visualized by silver staining. Protein bands were excised and in-gel-digested with trypsin (Promega) and the peptides were extracted twice with 1% (v/v) trifluoroacetic acid in 50% (v/v) acetonitrile aqueous solution for 30 min. All the peptides were subjected to LC-MS/MS analysis as described previously (Han et al., 2015).

LCI assays
LCI assays were performed as described (Ismayil et al., 2018). All combinations tested were agroinfiltrated into N. benthamiana leaves. The leaves were detached at 48 hpi, sprayed with 1 mM luciferin, and observed under a low-light cooled CCD imaging apparatus (iXon; Andor Technology). Photographs were 12 taken 5 min after exposure to luciferin.

Protein Analyses and Co-IP
For protein analysis, total protein extracts were prepared from N. benthamiana leaves and separated by SDS-PAGE for immunoblot analysis with the indicated antibodies as described (Zheng et al., 2019). For co-IP assays, total proteins from N. benthamiana leaves (~1 g tissue/sample) were extracted in ice-cold immunoprecipitation buffer (10% [v/v] glycerol,25 mM Tris,pH 7.5,150 mM NaCl,1×protease inhibitor cocktail [Roche], and 0.15% [v/v] Nonidet P-40). Protein extracts were incubated with GFP-Trap_A (ChromoTek) beads for 3 h at 4°C. The precipitates were washed four times with ice-cold immune precipitation buffer at 4°C and were analyzed on immunoblots using anti-Myc (Abmart), anti-HA (Cell Signaling Technology), or anti-GFP (ChromoTek) antibodies with dilutions of 1:3000.

Cell fractionation assay
Cell fractionation and immunoblotting assays were conducted as described previously (Chen et al., 2017).

BiFC Assays
Citrine YFP-based BiFC was performed as described . The experimental group and corresponding control group were inoculated on the same leaf to reduce differences of expression conditions. Live plant imaging was performed using a Zeiss LSM710 confocal microscope. Enhanced citrine YFP-derived fluorescence was acquired using a 514-nm laser and emission 519-to 587-nm filters. Images were analyzed with ZEN 2012 Light Edition.

Yeast two-hybrid screen and interaction assays
For the yeast two-hybrid interaction assay, CLCuMuB βC1 was amplified by PCR and cloned into the yeast vector pYL302 (Du et al., 2013) to generate the LexA DNA binding domain (BD) containing bait vectors.
NbGAPC1 was amplified and cloned into the B42 activation domain (AD)-containing vector pSAH20b. The interaction assay and yeast two-hybrid screen was performed as described (Wang et al., 2019).

Statistical analysis
The data were expressed as mean ± SEM from three independent experiments. Statistical analysis was performed using the Student t test with * P = <0.05.

50
(A) Co-IP assay shows that βC1, but not its mutants βC1 3A and βC1 Y48A , interferes with the interaction 51 between NbGAPCs and NbATG3. Total protein extracts were immunoprecipitated with anti-GFP beads 52 and monitored by immunoblotting (IB) using an anti-Myc antibody. Protein levels were assessed using