Two young genes reshape a novel interaction network in Brassica napus

Summary

• New genes often drive the evolution of gene interaction networks. In Brassica napus, the widely used genic male sterile breeding system 7365ABC is controlled by two young genes, Bnams4^b and BnaMs3. However, the interaction mechanism of these two young genes remains unclear.
• Here, we confirmed that Bnams4^b interacts with the nuclear localised E3 ligase BRUTUS (BTS). Ectopic expression of AtBRUTUS (AtBTS) and comparison between Bnams4^b-transgenic Arabidopsis and bts mutants suggested that Bnams4^b may drive translocation of BTS to cause various toxic defects.
• BnaMs3 gained an exclusive interaction with the plastid outer-membrane translocon Toc33 compared with Bnams3 and AtTic40, and specifically compensated for the toxic effects of Bnams4^b. Heat shock treatment also rescued the sterile phenotype, and high temperature suppressed the interaction between Bnams4^b and BTS in yeast. Furthermore, the ubiquitin system and TOC (translocon at the outer envelope membrane of chloroplasts) component accumulation were affected in Bnams4^b-transgenic Arabidopsis plants.
• Taken together, these results indicate that new chimeric Bnams4^b carries BTS from nucleus to chloroplast, which may disrupt the normal ubiquitin–proteasome system to cause toxic effects, and these defects can be compensated by BnaMs3−Toc33 interaction or environmental heat shock. It reveals a scenario in which two population-specific coevolved young genes reshape a novel interaction network in plants.

Introduction

Genes that have originated recently in the relevant evolutionary timescale are called new genes. They usually originate from exon shuffling, retroposition, gene duplication, mobile element, lateral gene transfer, gene fusion/fission and de novo origination (Long et al., 2003; Kaessmann, 2010; Cui et al., 2015). New genes create novel genetic and phenotypic diversity in organisms (Chen et al., 2013; Long et al., 2013; Zhang & Long, 2014). To acquire new molecular and cellular functions, new genes must integrate into ancestral gene interaction networks or reshape new gene interaction networks (Chen et al., 2010, 2012; W. Zhang et al., 2015). Several case studies have shown that new genes can integrate into local ancestral gene interaction networks and acquire male reproductive and chromosome segregation functions in fruit fly (Chen et al., 2012; Ross et al., 2013), mating function in budding yeast (Li et al., 2010), and pollen development and biochemical pathway formation in plants (Matsuno et al., 2009; Weng et al., 2012). However, it remains unknown if two or more young genes can coevolve into a novel interaction network in plants.

Chloroplasts need 2000–3000 different proteins, of which c. 95% are nucleus-encoded and must be imported into the chloroplast, which is surrounded by a double membrane envelope (Sugiura, 1989; Martin et al., 1998, 2002; Timmis et al., 2004; Sommer & Schleiff, 2014). Two translocon complexes, the TOC and TIC (translocons at the outer- and inner-envelope membranes of chloroplasts, respectively) complexes, mediate the translocation of preproteins into the chloroplast stroma with some molecular chaperones (Lee et al., 2014; Paila et al., 2015). The core TOC complex contains Toc33/Toc34, Toc75, and Toc159 (Kikuchi et al., 2006; Chen & Li, 2007). Toc33/Toc34 and Toc159 are both GTP-binding proteins as receptors for preproteins (Hirsch et al., 1994; Kessler et al., 1994; Reumann et al., 2005; Kessler & Schnell, 2009). Toc75 is a β-barrel protein deeply embedded in the lipid bilayer, forming the protein-conducting channel (Hinnah et al., 1997; Inoue & Potter, 2004). The ubiquitin–proteasome system (UPS) is a specific protein degradation pathway mediating 80–85% protein degradation in eukaryotes, with target proteins in the nucleus, cytoplasm, and on membrane surfaces (Bachmair et al., 2001; Vierstra, 2003; Schwechheimer & Schwager, 2004; Dreher & Callis, 2007). Ubiquitin ligase E3 recruits specific substrates and transfers ubiquitin to target proteins. Then the ubiquitinated target protein is recognised by the 26S proteasome for degradation (Smalle & Vierstra, 2004; Finley, 2009). Recent studies have revealed that the chloroplast protein import machinery is regulated by the UPS. Unimported chloroplast preproteins and chloroplast outer-membrane proteins can be ubiquitinated for degradation by various E3 ubiquitin ligases (Lee et al., 2009; Ling et al., 2012; Ling & Jarvis, 2015a; Woodson et al., 2015; H. Zhang et al., 2015).

Currently, the Brassica napus (canola or rapeseed) genic male sterile three-line breeding system 7365ABC, which consists of male sterile line 7365A (Bnams3ms3ms4^bms4^b), maintainer line 7365B (BnaMs3ms3ms4^bms4^b), temporary maintainer line 7365C (Bnams3ms3ms4^cms4^c), and restorer lines (BnaMs3Ms3ms4^bms4^b, Bnams3ms3ms4^ams4^a, and BnaMs3Ms3ms4ams4a), has been widely used for commercial hybrid production, due to its remarkable breeding advantages (Chen et al., 1998; Huang et al., 2007; Xiao et al., 2008; Stiewe et al., 2010; Xia et al., 2012). The fertility of 7365ABC is controlled by Bnams4^a/Bnams4^b/Bnams4^c on chromosome A07 and BnaMs3/Bnams3 on chromosome C09. The explicit−implicit relationships between these genes are Bnams4^a > Bnams4^b > Bnams4^c and BnaMs3 > Bnams3 (Dun et al., 2011; Xia et al., 2016). Bnams4^b is a new chimeric gene (fusion of four segments: AA1, AA2, NO, and AA3) originated via exon shuffling c. 4.6 Ma (Xia et al., 2016). It encodes a deleterious chimeric protein that localises to the chloroplast, affects normal function of the chloroplast to cause aborted anthers and albinotic seedlings, and high temperature treatment can partially restore male fertility caused by Bnams4^b (Zhu et al., 2010; Xia et al., 2016). Bnams4^a is a suppressor gene and may reduce the transcription level of Bnams4^b through DNA methylation (Xia et al., 2016; Wang et al., 2018). Bnams4^c is a null allele (Xia et al., 2016). BnaMs3 (homologous gene of AtTic40) is a neofunctionalised new gene relative to Bnams3, it also encodes a chloroplast-localised protein and can rescue the sterile phenotype caused by Bnams4^b with coevolution (Dun et al., 2011, 2014). We introduced the young genes Bnams4^b and BnaMs3 into Arabidopsis, and successfully reproduced the same genetic model as in B. napus (Xia et al., 2016). However, the mechanisms of Bnams4^b sterility and BnaMs3 fertility restoration are still unclear.

In this study, we screened the interaction proteins of Bnams4^b and BnaMs3, discovered and confirmed that Bnams4^b and BnaMs3 interact with the E3 ligase BRUTUS (BTS) and plastid outer-membrane translocon Toc33, respectively. Further genetic experiments, gene and protein expression analysis suggested that Bnams4^b may drive translocation of BTS to cause vegetative and reproductive defects, and these defects can be compensated by neofunctionalised young gene BnaMs3 but not Bnams3, depending on BnaMs3−Toc33 interaction. This study provides further insight into how Bnams4^b and BnaMs3 work, and also provides evidence that two coevolved young genes can reshape a novel interaction network in plants.

Materials and Methods

Plant materials and growth conditions

Brassica napus NILs 7365AB (Bnams3ms3ms4^bms4^b/BnaMs3ms3ms4^bms4^b), temporary maintainer line 7365C (Bnams3ms3ms4^cms4^c), and the Arabidopsis thaliana ecotype Columbia-0 were used in this study. Transgenic Bnams4^b, Bnams4^bMs3 and Bnams4^bms3 plants were obtained via Agrobacterium-mediated transformation. The B. napus materials were planted in the experimental field under normal environmental conditions (Wuhan; 114.35°E, 30.48°N). The Arabidopsis plants were grown in controlled glasshouse conditions (16 h : 8 h, light : dark) at 21–23°C with 30–60% humidity. For the heat shock experiment, Arabidopsis plants containing Bnams4^b were treated under different conditions in an artificial climate chamber (MLR-351H; Sanyo, Osaka, Japan). High pH soil was generated by adding calcium oxide (7.8 g CaO kg⁻¹ dry soil) according to previous studies (Kim et al., 2006; Long et al., 2010).

Plasmid construction and plant transformation

All target fragments were amplified from cDNA or gDNA of 7365ABC and Arabidopsis plants with Phusion DNA Polymerase (F-530S; Thermo Scientific, Waltham, MA, USA), and then cloned into the binary vector pMDC83 using the same restriction endonucleases and T4 ligase. Subsequently, we confirmed that the constructed vectors were correct by restriction digestion analysis and sequencing. Finally, the constructed plasmids were introduced into Arabidopsis by Agrobacterium-mediated transformation using the floral dip method (Clough & Bent, 1998), or into B. napus by Agrobacterium-mediated transformation using the hypocotyls infection method (Yi et al., 2010).

Yeast-two-hybrid assays

Gal4-based yeast-two-hybrid analysis was performed using the Matchmaker Gold Yeast-Two-Hybrid System (PT4084-1; Clontech, San Francisco, CA, USA) and a B. napus bud-specific cDNA library according to the manufacturer's instructions. Full-length BnaMs3, Bnams3, AtTic40 and Toc33, amino acids 1–538 of Bnams4^b, amino acids 539–1386 of Bnams4^b, and amino acids 796–1255 of BTS were cloned into the pGBKT7 vector (GAL4-BD domain) as bait plasmids, and the pGADT7 vector (GAL4-AD domain) as prey plasmids, respectively. The bait plasmids and cDNA library or the pairs of constructs were co-transformed into AH109 yeast cells and then plated on synthetic dropout nutrient medium (SD/−Trp−Leu) or (SD/−Trp/−Leu/−His/−Ade+X-α-Gal) agar plates.

Pull-down assays

BnaMs3, Bnams3, AtTic40 and Bnams4^b without signal peptides were cloned into pGEX-4T-3 vectors as baits. Toc33 without signal peptide and amino acids 796–1255 of BTS were cloned into pET32a vectors as preys. All the constructs were transformed into the Escherichia coli BL21 (DE3) strain to produce recombinant proteins. The pGEX-4T-3 vector was used to express glutathione-S-transferase (GST) as a negative control. Bait and prey proteins were mixed at 4°C for 6 h and then purified using glutathione conjugated agarose beads (17-5132-01; GE Healthcare, San Ramon, CA, USA). In vitro binding was detected with western blot (Chou et al., 2006; Einarson et al., 2007; Zhao et al., 2017).

Bimolecular fluorescence complementation (BiFC) assays

BiFC assays were performed according to published methods (Waadt et al., 2008). The AA1 + AA2 sequence of Bnams4^b as well as full-length coding sequences of BnaMs3, Bnams3, AtTic40, Toc33 and BTS were PCR-amplified and then inserted into the expression vectors pVYNE(R) and pVYCE(R), which were fused with the N-terminal or C-terminal of yellow fluorescent protein (YFP), respectively. Arabidopsis leaf mesophyll protoplasts were transformed with the fusion constructs and control vector by polyethylene glycol/calcium-mediated transformation (Yoo et al., 2007). After 8–24 h of incubation for gene expression, YFP fluorescence signals and chlorophyll spontaneous fluorescence were detected and photographed using a confocal laser scanning microscope (Olympus, Tokyo, Japan).

Co-immunoprecipitation (Co-IP)

To obtain Pro35S::Bnams4^b-FLAG, Pro35S::BnaMs3-FLAG, Pro35S::Bnams3-FLAG, and Pro35S::AtTic40-FLAG plants, we cloned the coding sequences of Bnams4^b, BnaMs3, Bnams3 and AtTic40 with a C-terminal FLAG fusion sequence into the binary vector pCAMBIA2306, and then transformed into B. napus. Total plant protein was extracted with 15 ml IP buffer (50 mM Tris-HCl, pH 7.5; 150 mM NaCl; 1 mM DTT; 1 mM PMSF; 2 mM EDTA; 0.1% Triton X-100; 1× protease inhibitor cocktail) from 5 g of transgenic material. The extracts were incubated with 50 μl Anti-FLAG M2 Affinity Gel (Merck, Darmstadt, Germany) for 2–3 h at 4°C. After washing five times with ice-cold tris-buffered saline (TBS) buffer (50 mM Tris-HCl, pH 7.5; 150 mM NaCl), the protein complexes associated with FLAG-tagged target proteins were analysed by western blot using specific antibodies.

Real-time quantitative reverse transcription PCR

Total RNA was extracted from 0.1 g FW of various organs using the RNAprep plant kit containing the DNase I treatment reagent (DP441; TIANGEN, Beijing, China), and 2 μg of total RNA was used for cDNA synthesis using the ReverTra Ace RT-PCR system (FSK-100; Toyobo, Osaka, Japan) according to the manufacturer's instructions. qRT-PCR was performed using SYBR Green Real-Time PCR Master Mix (QPK-201; Toyobo) and the CFX96™ Real-Time system (Bio-Rad, USA). The data were analysed with CFX Manager Software according to the method (Livak & Schmittgen, 2001). ACTIN2 was used as an internal control. At least four biological replicates were performed. The primers used for qRT-PCR analyses are shown in Supporting Information Table S1.

Western blotting

Total proteins were extracted from various plant organs or BL21 recombinant strains. Western blotting assays were performed as previously (Geoffroy et al., 1990). First, the protein samples in Laemmli buffer were denatured at 100°C for 10 min and then separated by SDS-PAGE. Second, the separated proteins were transferred to a polyvinylidene difluoride membrane, blocked with phosphate-buffered saline Tween (PBST) buffer (136.89 mM NaCl; 2.67 mM KCl; 8.1 mM Na₂HPO₄; 1.76 mM KH₂PO₄; 0.2% Tween 20) containing 3–5% nonfat powdered milk (A600669; Sangon Biotech, Shanghai, China), incubated with the specific primary antibody and HRP-labelled secondary antibody. Finally, the signals from secondary conjugated antibodies were detected using chemiluminescent Clarity™ Western ECL Substrate (1705060; Bio-Rad).

RNA sequencing, iTRAQ protein identification and data analysis

For RNA sequencing and iTRAQ protein identification, total RNAs and proteins were extracted from the same samples: bud and 4-wk old seedlings of Bnams4^b-transgenic, Bnams4^bMs3-transgenic and wild-type Arabidopsis plants. The samples were sequenced via Illumina HiSeq™ 2500 (San Diego, CA, USA) at the National Key Laboratory of Crop Genetic Improvement. iTRAQ protein identification was performed by Bo-Yuan Biotechnology company in Shanghai. Three biological replicates were carried out for each sample. Gene expression was estimated and normalised to normalised read counts (NRC) using DESeq2 (Love et al., 2014). Differentially expressed genes (DEGs) and differentially accumulated proteins (DAPs) were identified according to a fold change > 1.2 and FDR < 0.05 or P-value < 0.05. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the DEGs and DAPs was performed by TBtools (http://cj-chen.github.io/TBtools/) with a restrictive condition of FDR < 0.05.

Results

BnaMs3 rescues the embryo-lethal and chlorotic/albinotic phenotypes caused by Bnams4^b

Bnams4^b expresses in buds, rosette leaves, and cauline leaves, and causes aborted anthers and albinotic leaves (Xia et al., 2016). Here, we further found that the genotype Bnams3ms3ms4^bms4^b from 7365A × 7365B but no other genotypes (BnaMs3ms3ms4^bms4^b, BnaMs3Ms3ms4^bms4^b, and Bnams3ms3ms4^bms4^c) has embryo-lethal phenotype, and that the embryo abortion mainly occurs between 10 and 20 d after pollination (Fig. 1a,b,d,e). Expression level of Bnams4^b was highest in 10-d seed, and moderate in anther, 15-d seed, 20-d seed, and 25-d seed (Fig. 1c). In addition, the 10-d seed from 7365A × 7365B had higher expression level of Bnams4^b compared with seed from 7365A × 7365C (Fig. S1). Therefore, we speculated that the embryo-lethal trait may be tightly linked to Bnams4^b with a dosage effect.

Figure 1

Abnormal embryo maturation observed in Brassica napus and Arabidopsis thaliana. (a, b) Statistical analysis of seed setting percentage in different genotypes. Five individuals of each B. napus line were used for silique analysis. Five siliques for each individual were analysed. X-axis represents the days after pollination; Y-axis represents the seed setting percentage in different genotypes. n = 25, the data represent the means ± standard deviation (t-test: **P < 0.01). (c) Expression levels of Bnams4^b were detected in anther (from 1 to 2 mm bud) and silique or seed at different timepoints by real-time qRT-PCR analysis. Anther is from the 7365A line, silique and seed are from the 7365A × 7365B plants. The results are presented as relative transcript abundances. The data represent the means ± standard deviation of five biological replicates. (d, e) Opened siliques (d, c. 15 d after pollination; e, c. 20 d after pollination) from different genotypes are shown in B. napus. Red arrows indicate white or already dead seeds. AB, 7365A × 7365B; AB+, 7365A × 7365B+; AC, 7365A × 7365C; B+B+, 7365B+ × 7365B+. (f, g) Opened siliques (f, c. 9 d after pollination; g, c. 12 d after pollination) from different genotypes are shown in A. thaliana. Left, Bnams4^b-transgenic plants pollinated by BnaMs3-transgenic plants; right, Bnams4^b-transgenic plants pollinated by wild-type plants. Red arrows indicate white or already dead seeds. 7365A, Bnams3ms3ms4^bms4^b; 7365B, BnaMs3ms3ms4^bms4^b; 7365C, Bnams3ms3ms4^cms4^c; 7365B+, BnaMs3Ms3ms4^bms4^b.

To further explore whether Bnams4^b can cause embryo lethality, we performed a genetic experiment to investigate the relationship between Bnams4^b, BnaMs3 and the embryo-lethal phenotype in Arabidopsis. First, Bnams4^b and BnaMs3 were separately transformed into wild-type Arabidopsis, subsequently, male sterile plants expressing Bnams4^b were used as the female parent for hybridisation with the wild-type and BnaMs3-transgenic plants. Some embryos became white c. 9 d after pollination and aborted c. 12 d after pollination when the Bnams4^b-transgenic plants were crossed with wild-type plants (Fig. 1f,g). By contrast, the siliques of Bnams4^b-transgenic plants crossed with BnaMs3-transgenic plants showed no embryo lethality (Fig. 1f,g). In addition, we genotyped the F₁ population, because the Bnams4^b-transgenic plant contains at least one copy of Bnams4^b, normally the F₁ population should have at least half individuals containing Bnams4^b. Most individuals of the F₁ population from crosses between Bnams4^b-transgenic and BnaMs3-transgenic plants contained Bnams4^b, however, the F₁ population of Bnams4^b-transgenic crossed with wild-type plants only had c. 1/10 to 1/5 individuals containing Bnams4^b (Table 1). These data further suggest that the embryo-lethal trait was associated with Bnams4^b and can be rescued by BnaMs3.

Table 1. Number of individualsa with and without Bnams4^b in F₁ transgenic Arabidopsis.

Generations	Lines	Total	Bnams4^b contained	Bnams4^b not contained
Bnams4^b-T0 × WT	1	196	20	176
↓	2	251	39	212
F1	3	183	36	147
Bnams4^b-T0 × BnaMs3-T1	1	160	101	59
↓	2	220	154	66
F1	3	227	160	67

• The F₁ population contains three lines. Each line had several individuals.
• ^a Number of individuals for each line was determined by Bnams4^b-specific primers (ZA1-F/ZA1-R and RV-M/AQ1-R).

Bnams4^b leads to male sterile and chlorotic/albinotic phenotypes, and BnaMs3 can restore male fertility (Dun et al., 2011, 2014; Xia et al., 2016). However, whether the chlorotic/albinotic phenotypes can be restored by BnaMs3 is unknown. First, we obtained BnaMs3-transgenic and Bnams3-transgenic plants with hygromycin resistance. Next, Bnams4^b with kanamycin resistance was introduced into wild-type, BnaMs3-transgenic, and Bnams3-transgenic plants through Agrobacterium-mediated transformation to generate transgenic Arabidopsis lines (Bnams4^b, Bnams4^bMs3 and Bnams4^bms3). Almost no chlorotic/albinotic phenotypes (only two slightly chlorotic plants) was found in the Bnams4^bMs3-transgenic plants, while the other two genetic backgrounds both had a high proportion of chlorotic/albinotic plants (Table 2). These results indicate that BnaMs3 can rescue the chlorotic/albinotic phenotypes caused by Bnams4^b in Arabidopsis.

Table 2. Phenotypic statistics of transgenic plants with various genotypes in Arabidopsis.

Genotypes	Total	Green	Chlorotic	Albino
Only Bnams4^b	69	42	18	9
Both Bnams4^b and BnaMs3	58	56	2	0
Both Bnams4^b and Bnams3	55	34	15	6

• Individuals for each genotype were determined by primers specific for Bnams4^b (ZA1-F/ZA1-R and RV-M/AQ1-R), BnaMs3 (BJ4-2/BJ20 and RV-M/G4JC-1F) and Bnams3 (APJC6/BJ20 and M13-47/AGJC-R).

Bnams4^b interacts with BTS and BnaMs3 interacts with Toc33

Bnams4^b has multiple deleterious effects on plant growth and development, BnaMs3 suppresses these effects. To explore whether Bnams4^b and BnaMs3 function depending on protein interactions, we performed yeast-two-hybrid screening with Bnams4^b-1 (amino acids: 1–538), Bnams4^b-2 (amino acids: 539–1386), BnaMs3 and Bnams3 as baits and a floral bud-specific cDNA library of B. napus as prey. Bnams4^b was not directly used as a bait owing to its toxicity. 203, 152, 84, and 41 positive clones were obtained for Bnams4^b-1, Bnams4^b-2, BnaMs3 and Bnams3, respectively. Plasmids from all the positive clones were extracted and transformed into E. coli for sequencing. Frameshift mutations, self-activating and duplicated clones were excluded. We found that Bnams4^b-1 interacts with BnaBTS, and BnaMs3 interacts with BnaToc33. In addition, Bnams4^b and BnaMs3 are both chloroplast-localised proteins (Dun et al., 2011; Xia et al., 2016). To avoid missing the chloroplast membrane proteins, we cloned the open reading frames of 17 known Arabidopsis chloroplast outer and inner-membrane proteins as preys, and performed yeast-two-hybrid analysis using Bnams4^b-1, Bnams4^b-2, BnaMs3 and Bnams3 as baits. The results showed that BnaMs3 interacts specifically with AtToc33 (Fig. 2a). Then we performed directed yeast-two-hybrid point-to-point assays. Consistent with the library screening results, Bnams4^b-1 interacted with BnaBTS and AtBRUTUS (AtBTS), while BnaToc33 and AtToc33 interacted with BnaMs3, but not Bnams3 and AtTic40 (Fig. 2b).

Figure 2

Yeast-two-hybrid analyses of Bnams4^b, BnaMs3, Bnams3 and AtTic40. (a) Directed yeast-two-hybrid analysis using 17 known Arabidopsis chloroplast membrane proteins with Bnams4^b-1, Bnams4^b-2, BnaMs3 and Bnams3. Description of interacting colony is indicated in bold black. (b) Various pairs co-transformed into AH109 yeast cells and evaluated on the selective medium SD/−Trp/−Leu (indicates whether transformations were successful) and SD/−Trp/−Leu/−His/−Ade+X-α-Gal (indicates interactions). BD-53/AD-RecT was the positive control, and BD-Lam/AD-RecT was the negative control. AD, pGADT7/activation domain; Ade, adenine; BD, pGBKT7/binding domain; His, histidine; Lam, human lamin C; Leu, leucine; RecT, recombination of the SV40 large T antigen; SD medium, synthetic defined medium; Trp, tryptophan.

To confirm the yeast-two-hybrid results, we performed a GST pull-down experiment in vitro. GST-tagged Bnams4^b, BnaMs3, Bnams3 and AtTic40 and His-tagged BnaBTS and BnaToc33 were produced and purified from E. coli using agarose resin. GST protein was purified as a negative control for this assay. The glutathione sepharose bead pull-down and immunoblotting assays showed that GST-Bnams4^b pulled down His-BnaBTS, and His-BnaToc33 was pulled down by GST-BnaMs3 but not GST-Bnams3 and GST-AtTic40 (Fig. 3a).

Figure 3

Yeast-two-hybrid results were confirmed in vitro and in vivo. (a) In vitro glutathione S-transferase (GST) pull-down assays confirm interactions of Bnams4^b with BnaBTS, and BnaMs3 with BnaToc33. The GST pull-down products were detected by western blots using anti-GST and anti-His antibodies. GST tag alone was a negative control. Input, samples processed without pull-down. (b) Co-immunoprecipitation analysis of the interactions between Bnams4^b and BnaBTS, BnaMs3 and BnaToc33. Total plant proteins were immunoprecipitated using Anti-FLAG M2 Affinity Gel and detected by anti-FLAG, anti-BTS and anti-Toc33 antibodies. Input, samples processed without precipitation. Lanes 1–4: proteins from Pro35S::Bnams4^b-FLAG, Pro35S::BnaMs3-FLAG, Pro35S::Bnams3-FLAG, and Pro35S::AtTic40-FLAG transgenic Brassica napus plants. (c, d) Bimolecular fluorescence complementation (BiFC) assays in Arabidopsis protoplasts transformed with different combinations of the constructs. Enhanced yellow fluorescent protein (YFP) indicates protein–protein interactions. Truncated Bnams4^b (AA1 + AA2) was used. CE, the C-terminal fragments of YFP; NE, the N-terminal fragments of YFP. Bars, 5 μm.

We also verified our results in vivo using BiFC assays by Arabidopsis protoplast transient transformation. Bnams4^b had toxic effects on protoplast cells, so truncated Bnams4^b (AA1 + AA2), BnaBTS, BnaMs3, Bnams3, AtTic40, and BnaToc33 were fused to the N-terminal and C-terminal fragments of enhanced YFP under the control of the Pro35S promoter. After they were co-transformed into Arabidopsis protoplast cells, reconstructed YFP signals were detected in the cells co-expressing with (N-terminal fragment of enhanced YFP) NE-Bnams4^b and (C-terminal fragment of enhanced YFP) CE-BnaBTS, CE-Bnams4^b and NE-BnaBTS, NE-BnaMs3 and CE-BnaToc33 as well as CE-BnaMs3 and NE-BnaToc33. However, no YFP signal was detected in protoplasts containing other co-transformed plasmids (Fig. 3c,d). These results indicate that Bnams4^b interacts with BnaBTS, and BnaMs3 interacts with BnaToc33.

A complementary anti-FLAG Co-IP approach was used to further confirm the interaction results. We transformed Pro35S::Bnams4^b-FLAG, Pro35S::BnaMs3-FLAG, Pro35S::Bnams3-FLAG, and Pro35S::AtTic40-FLAG constructs into B. napus and obtained stable transgenic lines. All the Pro35S::BnaMs3-FLAG, Pro35S::Bnams3-FLAG, and Pro35S::AtTic40-FLAG plants grew normally; however, most Pro35S::Bnams4^b-FLAG plants were chlorotic/albinotic lethal, only a small portion of the plants survived but were sterile. Subsequently, the protein complexes associated with FLAG-tagged target proteins were immunoprecipitated using Anti-FLAG M2 Affinity Gel (Merck) from total plant protein, and then analysed by western blot using specific antibodies. Immunoblot analysis showed that Bnams4^b was associated with BnaBTS, and BnaMs3 was associated with BnaToc33 (Fig. 3b). Taken together, our results provided further support for the interactions between Bnams4^b and BnaBTS, BnaMs3 and BnaToc33.

Determining the functions of the four fusion segments of Bnams4^b

Bnams4^b is fusion of four segments: AA1, AA2, NO, and AA3 (Xia et al., 2016). To investigate the functions of the four fusion segments, we produced various deletion and replacement constructs. ProBnams4^b::Bnams4^b, ProBnams4^b::AA1, ProBnams4^b::AA2, ProBnams4^b::NO, and ProBnams4^b::AA3 were made using the full-length or partial cDNAs of Bnams4^b, ProBnams4^b:: Bnams4^b⇆mtHsc70 was made using Bnams4^b with AA3 replaced by mtHsc70-1, ProBnams4^b::Bnams4^bΔAA1, ProBnams4^b::Bnams4^bΔAA2, ProBnams4^b::Bnams4^bΔNO, and ProBnams4^b::Bnams4^bΔAA3 were made using Bnams4^b lacking the AA1, AA2, NO, and AA3 segments respectively (Fig. 4a). The constructs were transformed into Arabidopsis. Only ProBnams4^b::Bnams4^bΔAA3, ProBnams4^b::Bnams4^b, and ProBnams4^b:: Bnams4^b⇆mtHsc70 plants exhibited sterile and chlorotic/albinotic phenotypes. Interestingly, all the ProBnams4^b::Bnams4^bΔAA3 plants showed chlorotic/albinotic phenotypes, but only part of ProBnams4^b::Bnams4^b and ProBnams4^b::Bnams4^b⇆mtHsc70 plants had chlorotic/albinotic phenotypes (Fig. 4a,b). Next, we determined which segment of Bnams4^b is required for interaction with BnaBTS by yeast-two-hybrid analysis. AA2 but not AA1, NO, or AA3 directly interacted with BnaBTS (Fig. 4c). The yeast-two-hybrid results were further confirmed using GST pull-down assay, immunoblot analysis showed that the truncated AA2 interacted with BnaBTS (Fig. 4d).

Figure 4

Functional regions of Bnams4^b. (a) Schematic diagram showing the various deletion and replacement constructs (left), and statistical analysis of the chlorotic/albinotic and sterile phenotypes (right). P, ProBnams4^b; mtHsc70, the homologous sequence of AA3 from AtmtHsc70-1. (b) Chlorotic/albinotic and sterile phenotypes of ProBnams4^b::Bnams4^b, ProBnams4^b::Bnams4^b⇆mtHsc70, and ProBnams4^b::Bnams4^bΔAA3 plants. Red arrows indicate the already dead albinotic seedlings. (c) BnaBTS interacts with AA2 but not AA1, NO or AA3 in yeast-two-hybrid assays. (d) Interaction between AA2 and BnaBTS was confirmed by GST pull-down in vitro. GST tag alone was the negative control. Input, samples processed without pull-down.

Determination of the core site of BnaMs3

Six amino acid substitutions present in the C-terminal of BnaMs3, comprising the TPR (Tetratricopeptide Repeat) and Hop domains, cause the neofunctionalisation of BnaMs3 (Dun et al., 2014). To further explore the BnaMs3 functional variant amino acid sites, six site-directed mutagenesis expression vectors for BnaMs3 were constructed, named P307, P321, P343, P378, P386 and P408, with each amino acid replaced by that of Bnams3, to transform wild-type Arabidopsis and then cross with Bnams4^b-transgenic plants. Phenotypic observation and acetic carmine staining showed that BnaMs3 lost its phenotypic rescue function after mutation of amino acid 321 (Fig. 5a,b; Table S2). We have known that BnaMs3 interacts with Toc33. In order to study whether the six mutation sites of BnaMs3 affect this interaction, we used six point mutations on BnaMs3 (P307, P321, P343, P378, P386 and P408 with each amino acid replaced by that of Bnams3) and BnaToc33 for yeast-two-hybrid assays, and found that only P321 could not interact with BnaToc33 (Fig. 5c). These results suggest that amino acid 321 is essential for the neofunction of BnaMs3.

Figure 5

Fertility of hybrids and yeast-two-hybrid analyses with site-directed mutagenesis of BnaMs3. (a) Fertility observations on representative site-directed mutagenesis hybrids in Arabidopsis thaliana. (b) Acetic carmine dyed anthers coming from (a). Bars, 50 μm. (c) Protein interactions were confirmed by yeast-two-hybrid analyses. P307, P321, P343, P378, P386 and P408, BnaMs3 with each amino acid replaced by that of Bnams3.

Mislocalization of AtBTS leads some transgenic Arabidopsis plants to exhibit sterile/semisterile phenotypes

If the effects of Bnams4^b depends on its interaction with BTS, introduction of Bnams4^b into bts mutants will not lead to sterility. Unfortunately, the BTS knockout mutant was embryo lethal, therefore, we could not perform this experiment. Bnams4^b localises to the chloroplast, while AtBTS mostly localises to the nucleus (Kobayashi et al., 2013; Xia et al., 2016). Ectopic expression of BTS may produce certain phenotypes. In order to verify this hypothesis, ProBnams4^b::AtBTS⇆SP was constructed, using the coding sequence of AtBTS, but with the N-terminal signal peptide substituted by the Bnams4^b N-terminal signal peptide. As a control, ProBnams4^b::AtBTS was constructed. These two binary constructs were transformed into Arabidopsis, 23 ProBnams4^b::AtBTS⇆SP and 27 ProBnams4^b::AtBTS transgenic plants were identified. Three ProBnams4^b::AtBTS⇆SP lines showed sterile phenotype, and four ProBnams4^b::AtBTS⇆SP lines were semisterile (Fig. 6a). However, no obvious phenotypes of ProBnams4^b::AtBTS plants were observed compared with wild-type plants. These results support our hypothesis that the altered localisation of AtBTS affected plant fertility as in Bnams4^b-transgenic Arabidopsis plants.

Figure 6

Phenotypic characterisation of Bnams4^b ectopic expression plants, and iron deficiency tolerance of Bnams4^b-transgenic plants in Arabidopsis thaliana. (a) Some ProBnams4^b::AtBTS⇆SP plants exhibited sterile (middle) and semisterile (right) phenotypes. Red arrows indicate shortened siliques. (b) Two-wk-old bts, wild-type, and Bnams4^b-transgenic Arabidopsis plants grown on acidic (pH 5.6–5.7) or alkaline (pH 7.9–8.0) soil. (c) Eleven-d-old bts, wild-type and Bnams4^b-transgenic Arabidopsis plants grown on iron-sufficient or iron-deficient medium.

Bnams4^b-transgenic plants exhibit increased tolerance to iron deficiency in Arabidopsis like bts mutants

We found that Bnams4^b interacts with BTS, and previous studies have confirmed that partial loss-of-function bts mutant show increased tolerance to iron deficiency (Long et al., 2010; Selote et al., 2015). To test whether the growth of Bnams4^b-transgenic seedlings was affected by iron deficiency, we germinated the wild-type, bts mutant, and Bnams4^b-transgenic plants on alkaline soil (pH 7.9–8.0), in which iron is insoluble. The wild-type seedlings exhibited chlorotic phenotype on the high pH soil, consistent with limited iron availability, however, the bts mutant and Bnams4^b-transgenic seedlings grew greener and larger than wild-type. As a control, when grown under normal soil cultivation (pH 5.6–5.7), no obvious difference was observed among these plants (Fig. 6b). We also germinated the wild-type, bts mutant, and Bnams4^b-transgenic plants on iron deficiency and iron addition solid medium, and observed similar results (Fig. 6c). These findings suggest that the Bnams4^b-transgenic plants have iron deficiency tolerance like bts mutants.

High temperature affects the interaction between Bnams4^b and BnaBTS

Heat shock can partially restore male fertility caused by Bnams4^b in 7365A (Zhu et al., 2010). We further confirmed this result using Bnams4^b-transgenic Arabidopsis plants. The sterile plants were treated with different temperature (29–37°C) and humidity (25–85%) conditions over a range of time periods (2–72 h), and then transferred to normal temperature conditions for phenotypic observation. Heat shock treatment made some sterile flowers change into fertile ones (Fig. 7a). Comparing different treatment conditions, we found that the optimum conditions for fertility restoration were 34–35°C and 75–85% humidity for > 24 h (Table 3). In addition, fertility was identified each day after heat shock by acetic carmine staining. Interestingly, only the flowers that opened on the 7^th d could produce fertile pollen (Fig. 7b). These findings indicate that heat shock only affected buds at a specific developmental stage, which depended on strict temperature, humidity and treatment duration.

Figure 7

Fertility restoration of Bnams4^b-transgenic Arabidopsis plants by heat shock, real-time qRT-PCR and western blot analysis of gene and protein expression. (a) Heat shock treatment partially restored male fertility in Bnams4^b-transgenic Arabidopsis plants. Left, Bnams4^b-transgenic plant without heat shock treatment; middle, Bnams4^b-transgenic plant with heat shock treatment; right, wild-type plant. Red arrows indicate the normal siliques. (b) Acetic carmine dyed anthers coming from the flowers opened on each day after heat shock. Only 7^th d flowers could produce fertile pollen. Bars, 50 μm. (c) Yeast-two-hybrid analyses of Bnams4^b and BnaBTS under different incubation temperatures. (d, e) Real-time qRT-PCR analysis of Bnams4^b and BnaMs3. The data represent the means ± standard deviation of four biological replicates. Gene transcript levels from RNA-seq were normalised by NRC. 7365A-1/2/3, 1–2 mm, 2–3 mm, and 3–4 mm anthers from 7365A bud; 7365B-1/2/3, 1–2 mm, 2–3 mm, and 3–4 mm anthers from 7365B bud. (f) Western blot analysis of proteins extracted from seedling and bud. Actin used as control to normalise protein levels. Two lines were used for each genotype. The asterisk indicates nonspecific protein band.

Table 3. Statistical analysis of Bnams4^b-transgenic Arabidopsis plants with heat shock treatment.

		RH 75–85%	RH 50–60%	RH 25–35%
29°C	24H	0/6
31°C		0/5
32°C		0/10
33°C		1/12
34°C	2H	0/6
	6H	0/5
	10H	0/8
	14H	1/7
	18H	1/8
	22H	1/6
	24H	6/12	0/12	0/12
	48H	5/11	0/11	0/13
	72H	5/12	0/12	0/11
35°C	24H	6/12	0/13	0/12
	48H	6/13	0/12	0/12
	72H	5/12	0/12	0/12
36°C	24H	2/12
37°C		0/6

• A/B, fertile flowers/total flowers with heat shock treatment; H, treatment time (hour); RH, humidity.

Next, we investigated whether temperature could affect the interaction between Bnams4^b and BnaBTS by yeast-two-hybrid assays. The AH109 yeast cells co-transformed with BD-AA2 and AD-BnaBTS (BD-AA2/AD-BnaBTS) were incubated at 16°C, 23°C, 30°C, 35°C, 37°C, and 42°C. Yeast cells co-transformed with BD-53 and AD-RecT (BD-53/AD-RecT) were as a control. We found that BD-53/AD-RecT grew normally at 16–37°C, however, growth of BD-AA2/AD-BnaBTS was affected at 35°C and inhibited at 37°C (Fig. 7c). These results suggest that proper high temperatures disturb the interaction of Bnams4^b and BnaBTS.

Transcriptome and proteome analyses of transgenic Arabidopsis plants

To get a better insight into the global effect of Bnams4^b on plant growth and development, we performed transcriptome and proteome analyses of Bnams4^b-transgenic, Bnams4^bMs3-transgenic, and wild-type plants by RNA-seq and the isobaric tags for relative and absolute quantitation (iTRAQ) technique, using seedling and bud tissue with three biological replicates for each experimental line. Comparing Bnams4^b-transgenic sterile plants with both Bnams4^bMs3-transgenic and wild-type fertile plants to get the intersections. There were different numbers of upregulated and downregulated DEGs and DAPs in the seedlings and buds (Tables S3, S4). We performed KEGG pathway enrichment analysis of the DEGs and DAPs. The DEGs were mainly involved in photosynthesis, energy and lipid metabolism, especially cutin, suberine and wax biosynthesis (Fig. S2a). This is consistent with the phenotypes of Bnams4^b-transgenic plants: chlorotic/albinotic seedlings, and sterile anthers with decreased wax and cutin contents (Xia et al., 2016). Interestingly, genes involved in the ‘ubiquitin system’ were enriched in M-up (Fig. S2a). We have confirmed Bnams4^b interacts with an E3 ligase BTS. These results suggest that Bnams4^b affect the normal ubiquitination process. The DAPs were also mainly involved in photosynthesis and energy metabolism (Fig. S2b).

Chloroplast protein translocation is a complex process that has been well described in Arabidopsis. Therefore, we analysed gene expression and protein accumulation of the known chloroplast outer and inner-membrane proteins. Changes in transcription levels were irregular; however, changes in protein accumulation were mainly related to the TOC complex, including Toc33, Toc34, Toc75-III and Toc159 (Table S5). These results suggest that chloroplast outer membrane translocation may be affected. In addition, we found that Bnams4^b expression is inhibited when BnaMs3 is present in our transcriptome data (Fig. 7d). We verified this result in transgenic Arabidopsis and B. napus near-isogenic lines (NILs) 7365AB by qRT-PCR analysis (Fig. 7d,e). We also detected protein expression abundance in Bnams4^b-transgenic, Bnams4^bMs3-transgenic and wild-type Arabidopsis plants by Western blot experiments. The results showed that Bnams4^b and BnaMs3 accumulated in the corresponding transgenic lines, furthermore, increased Toc33 accumulation was found in Bnams4^b-transgenic seedlings (Fig. 7f). Protein accumulation revealed by western blot was consistent with the material genotypes and proteome results.

Discussion

Segments of chimeric Bnams4^b cooperate to generate its toxicity and multiple factors contributed to its fixation

Bnams4^b is composed of four segments (AA1, AA2, NO, and AA3), which cause male sterility and chlorotic/albinotic seedlings (Xia et al., 2016). The segment AA1 exhibits high similarity with the N-terminal signal peptide of AT4G37510, a ribonuclease III family protein located in chloroplasts, and GFP fluorescence from AA1-GFP fusion protein was detected in chloroplasts (Xia et al., 2016). ProBnams4^b::Bnams4^bΔAA1 plants did not have sterile or chlorotic/albinotic phenotypes (Fig. 4a), indicating that AA1 is essential for Bnams4^b toxicity. The segment AA2 is homologous to the partial sequence of AT1G80070, a factor that influences pre-mRNA splicing and is required for embryonic development (Schwartz et al., 1994; Sasaki et al., 2015; Deng et al., 2016; Kanno et al., 2017). Yeast-two-hybrid and pull-down assays showed AA2, but not AA1, NO, or AA3, interacts with BTS (Fig. 4c,d), and there was no obvious phenotype in ProBnams4^b::Bnams4^bΔAA2 plants (Fig. 4a), indicating that AA2 is also necessary for Bnams4^b toxicity. Although the segment NO between AA2 and AA3 did not demonstrate homology with any reference gene sequence, however it is essential for Bnams4^b toxicity, like AA1 and AA2, because growth and development of ProBnams4^b::Bnams4^bΔNO plants was completely normal (Fig. 4a). The segment AA3 has high sequence similarity with AT4G37910, a mitochondrial heat shock protein 70-1 (mtHsc70-1) that could affect plant thermotolerance (Zhou et al., 2012). All the ProBnams4^b::Bnams4^bΔAA3 plants were chlorotic, but only a fraction of ProBnams4^b::Bnams4^b and ProBnams4^b::Bnams4^b⇆mtHsc70 plants had a chlorotic phenotype (Fig. 4a,b), these results indicate that AA3 can partially suppress the toxic effects of Bnams4^b. In short, AA1 is responsible for chloroplast localisation, AA2 interacts with BTS, NO performs an unknown function, and these three segments together lead to male sterility and chlorotic/albinotic seedlings, which was partially suppressed by AA3.

Most chimeric genes are rapidly eliminated from the genome, and only c. 1.4% of them are fixed by natural selection (Rogers et al., 2009; Ding et al., 2012; Long et al., 2013). Bnams4^b has detrimental effects on both plant vegetative growth and reproductive growth, therefore, its fixation should be a very complex process. Although Bnams4^b shows detrimental effects under normal conditions, it could contribute to stress resistance, such as tolerance to alkaline soil (Fig. 6b), this provides an original evolutionary force of natural selection for Bnams4^b. AA3 has a function in eliminating vegetative growth defects, to make plants grow successfully (Fig. 4a,b). High temperature restores plant fertility and makes Bnams4^b temporarily fixed (Fig. 7a,b). Finally, the stable fixation of Bnams4^b depends on the coevolved suppressor gene BnaMs3. This unveils a scenario for the natural selection and survival of a deleterious new gene.

Bnams4^b triggers translocation of BTS to reshape a novel interaction network

BTS is a nuclear localised E3 ligase. It binds to and degrades relative transcription factors in the iron-deficiency and drought responses (Long et al., 2010; Kobayashi et al., 2013; Selote et al., 2015, 2018; Hindt et al., 2017). However, Bnams4^b is a chloroplast-localised protein (Xia et al., 2016). Interestingly, our experimental results showed that Bnams4^b interacts with BTS (Figs 2b, 3a–c). So, we speculated that Bnams4^b might carry BTS to chloroplasts. There is some other evidence for this speculation. First, transcriptome KEGG analysis showed that genes involved in the ‘ubiquitin system’ were enriched in Bnams4^b-transgenic plants (Fig. S2a), indicating that the UPS was affected. Second, although total BTS protein accumulation did not change, the Bnams4^b-transgenic plants exhibited increased tolerance to iron deficiency, similar to bts mutants (Figs 6b,c, 7f), these results confirmed that not all BTS proteins are in the nucleus when Bnams4^b is present. Third, the ProBnams4^b::AtBTS⇆SP (signal peptide of AtBTS replaced by AA1) plants exhibited sterile/semisterile phenotypes (Fig. 6a), indicating that chloroplast localisation of AtBTS can affect plant fertility like in Bnams4^b-transgenic plants. Fourth, the ΔAA1 transformation experiment indicated that the chloroplast-localised signal peptide AA1 is necessary for the phenotypes from Bnams4^b (Fig. 4a). Last, the BiFC assays proved that Bnams4^b and BnaBTS interact in the chloroplast (Fig. 3c). When BTS is carried to chloroplast, it may affect chloroplast protein translocation by two pathways. One is that BTS directly recognises and degrades certain chloroplast preproteins by the UPS. The other is BTS affects the normal UPS associated with E3 ligase SP1 (suppressor of plastid protein import-1). SP1 can directly mediate ubiquitination of TOC components and promote their degradation at the chloroplast outer membrane (Ling et al., 2012; Ling & Jarvis, 2015a). We prefer the second hypothesis for the following reasons. In Bnams4^b-transgenic plants, protein levels of TOC complex members including Toc33, Toc34, Toc75-III and Toc159, were increased while transcription of these genes did not increase or even decreased (Fig. 7f; Table S5). This suggests that the turnover of TOC components was affected. Our previous studies have found that the sterile anther in 7365A displays delayed tapetal programmed cell death (PCD) (Dun et al., 2011; Zhou et al., 2011; Song et al., 2016), and another study reported that SP1 is involved in the regulation of PCD (Basnayake et al., 2011). So, SP1 mediating UPS may be affected in 7365A. In conclusion, Bnams4^b carries BTS to the chloroplast and reshapes a novel interaction network, in which BTS causes various toxic effects by disrupting normal chloroplast protein translocation at the chloroplast outer membrane.

BnaMs3−Toc33 interaction integrates into the novel interaction network reshaped by Bnams4^b

Neofunctionalised suppressor gene BnaMs3 is a homologue of AtTic40 which has a controversial function in protein translocation across the chloroplast inner membrane (Chou et al., 2006; Dun et al., 2011; Kikuchi et al., 2013; Shi & Theg, 2013; Nakai, 2015, 2018; Richardson et al., 2017). We found that plastid outer-membrane translocon Toc33, responsible for the import of photosynthetic preproteins, interacts with BnaMs3 but not with Bnams3 and AtTic40 (Figs 2a,b, 3a,b,d). This finding suggests that BnaMs3 or at least part of BnaMs3 may be on the chloroplast outer membrane. We reviewed the discovery and identification process of Tic40 and noted extreme confusion when Tic40 was first isolated. Originally, Tic40 in B. napus was described as Com44/Cim44 both at the outer- and inner-envelope membranes of chloroplasts (Wu et al., 1994; Ko et al., 1995). Later, it was renamed Toc36, a 36 kDa TOC component at the outer-envelope membrane (Pang et al., 1997). Subsequently, a full-length pea clone encoding a 44 kDa protein was identified and termed Tic40, studies showed that Toc36 from B. napus was truncated at the N-terminal, and Tic40 was most likely located in the inner-envelope membrane of pea chloroplasts (Stahl et al., 1999). However, until now, the actual identities of the initially observed two 44 kDa proteins in the outer- and inner-envelope membranes have remained unclear (Nakai, 2015). In addition, there is no ubiquitin and proteasome inside the plastid, so the UPS only exists on the outer-membrane surface (Ling & Jarvis, 2015b). If BnaMs3 eliminates the toxic effects from the Bnams4^b and BTS mediated UPS, it should be on the outer membrane, which is consistent with the interaction between BnaMs3 and Toc33. The ages of BnaMs3 and Bnams4^b are almost the same (Xia et al., 2016). Therefore, this specific interaction between BnaMs3 and Toc33 probably coevolved with the Bnams4^b-BTS interaction. Interestingly, these two kinds of interaction can cause opposite phenotypes, which means that they are associated with each other directly or indirectly (possibly through UPS). Thereby, it may be concluded that BnaMs3 interacting with Toc33 integrates into the novel interaction network reshaped by Bnams4^b, which can eliminate all the toxic effects of Bnams4^b.

A putative model for the novel interaction network of young genes in 7365ABC

We propose a possible model for the Bnams4^b and BnaMs3 interaction network in the 7365ABC system. When Bnams4^b is absent, BTS localises in the nucleus and functions normally, and the translocation of preproteins into the chloroplast is normal (Fig. 8a). In plants containing Bnams4^b, Bnams4^b reshapes a novel interaction network by triggering translocation of BTS to the chloroplast. The novel interaction network could affect chloroplast protein translocation and ultimately lead to sterility, possibly by directly degrading certain preproteins or by disturbing the normal UPS (such as the SP1 mediated UPS) on the chloroplast outer membrane (Fig. 8b). Heat shock treatment prevents Bnams4^b−BTS interaction. Bnams4^b alone cannot affect chloroplast protein translocation, so the sterile phenotype was rescued (Fig. 8b). When BnaMs3 was present, both mRNA and protein of Bnams4^b were not detected (Fig. 7d–f). BnaMs3 interacts physically with Toc33 to integrate into the novel interaction network, which may directly send a signal to the nucleus to inhibit Bnams4^b expression, or remove Bnams4^b through some unknown mechanism, and the removal of Bnams4^b sends feedback signals to repress transcription of Bnams4^b in the nucleus. Therefore, the toxic effects of Bnams4^b were eliminated (Fig. 8b). However, the precise molecular mechanism of BnaMs3 rescuing the fertility of Bnams4^b remains unclear, and the strategy of BnaMs3 and Toc33 suppressing the mRNA and protein expression of Bnams4^b need to be further investigated. This model provides evidence that two coevolved young genes can indeed reshape a novel interaction network in plants.

Figure 8

Model showing the novel interaction network of young genes in 7365ABC. (a) No connection between BTS and chloroplast protein translocation under normal conditions. (b) Bnams4^b carries BTS to chloroplast to affect chloroplast protein translocation. Heat shock treatment suppresses interaction between Bnams4^b and BTS. BnaMs3 interacts physically with Toc33 to suppress Bnams4^b transcription and Bnams4^b accumulation. Putative or unverified pathways are indicated by broken lines. 26SP, 26S proteasome; HST, heat shock treatment; IMS, intermembrane space; OEM, outer-envelope membrane; Ub, ubiquitin.

Acknowledgements

We thank Hongbin Wang (Sun Yat-sen University) for providing bts mutant. We would like to thank Editage (www.editage.cn) for English language editing. This work was supported by the National Key Research and Development Program of China (2016YFD0100305), the National Key Research and Development Program of China (2016YFD0100804), the National Natural Science Foundation of China (31871657), and the National Natural Science Foundation of China (31501374).

Author contributions

ZZ performed most of the experiments and wrote the manuscript. YF participated in the heat shock experiment. JX assisted in doing part of the experiments. XG performed the site-directed mutagenesis of BnaMs3. KH offered help on bioinformatic analysis. ZW and JG gave useful advice and modified the manuscript. JW, BY, JS, CM and TF supervised this study. SX gave advice in the experiments and modified the manuscript. JT conceived and supervised the research and writing.