The ER luminal C-terminus of AtSec62 is critical for male fertility and plant growth in Arabidopsis thaliana

SUMMARY Protein translocation into the endoplasmic reticulum (ER) occurs either co- or post-translationally through the Sec translocation system. The Arabidopsis Sec post-translocon is composed of the protein-conducting Sec61 complex, the chaperone-docking protein AtTPR7, the J-domain-containing proteins AtERdj2A/B and the yet uncharacterized AtSec62. Yeast Sec62p is suggested to mainly function in post-translational translocation, whereas mammalian Sec62 also interacts with ribosomes. In Arabidopsis , loss of AtSec62 leads to impaired growth and drastically reduced male fertility indicating the importance of AtSec62 in protein translocation and subsequent secretion in male gametophyte development. Moreover, AtSec62 seems to be divergent in function as compared with yeast Sec62p, since we were not able to complement the thermosensitive yeast mutant sec62-ts . Interestingly, AtSec62 has an additional third transmembrane domain in contrast to its yeast and mammalian counterparts resulting in an altered topology with the C-terminus facing the ER lumen instead of the cytosol. In addition, the AtSec62 C-terminus has proven to be indispensable for AtSec62 function, since a construct lacking the C-terminal region was not able to rescue the mutant phenotype in Arabidopsis . We thus propose that Sec62 acquired a unique topology and function in protein translocation into the ER


INTRODUCTION
Membrane as well as secretory proteins are synthesized on cytosolic ribosomes and need to be transported across or into the endoplasmic reticulum (ER) membrane for processing and for further transport through the Golgi apparatus. Translocation of pre-proteins can occur either co-translationally or in a post-translational manner (reviewed by Rapoport, 2007;Cross et al., 2009;Zimmermann et al., 2011;Linxweiler et al., 2017).
Co-translational protein transport into the ER involves the signal recognition particle (SRP) which is binding to the N-terminal signal sequence of the emerging polypeptide chain at the ribosomes thereby arresting elongation (Saraogi and Shan, 2011). Subsequently, the SRP binds the SRP receptor which is localized to the ER membrane and is composed of the soluble subunit SRa and the membrane anchoring SRb (Tajima et al., 1986;Schwartz and Blobel, 2003). Binding of the SRP to its receptor enables an interaction of the ribosome, still harbouring the nascent polypeptide chain, and the Sec translocation complex (Saraogi and Shan, 2011). Translation continues after dissociation of the SRP, while the emerging polypeptide is simultaneously translocated through the Sec61 channel (Saraogi and Shan, 2011).
However, most small pre-secretory proteins in yeast and mammals are imported post-translationally due to their signal sequence being transcribed only shortly before protein translation is entirely completed (Lakkaraju et al., 2012;Ast and Schuldiner, 2013). Fully translated pre-proteins are bound by cytosolic chaperones including heat shock proteins (HSPs) like HSP70, which prevent molecular crowding as well as premature folding of the polypeptide and facilitate guidance of the pre-protein to the Sec translocation complex (Zimmermann et al., 1988;Ng et al., 1996;Ngosuwan et al., 2003). In yeast, the Sec translocon is composed of the heterotrimeric complex Sec61 (composed of Sec61p, Sbh1p and Sss1p) forming the translocation channel and a tetrameric complex of Sec62p, Sec63p, Sec71p and Sec72p (Deshaies et al., 1991;Panzner et al., 1995;Harada et al., 2011). The Sec61 complex and Sec63p are involved in co-and post-translational protein translocation, whereas Sec62p, Sec71p and Sec72p are associated with post-translational translocation only (Deshaies et al., 1991;Brodsky and Schekman, 1993;Panzner et al., 1995). Together, Sec62p, Sec71p and Sec72p, probably act as a signal peptide receptor with Sec72p enabling interaction through its tetratricopeptide repeat (TPR) domain (Fang and Green, 1994;Feldheim and Schekman, 1994;Lyman and Schekman, 1997;Plath et al., 1998;Schlegel et al., 2007). Moreover, Yim et al. (2018) proposed that Sec62p, Sec63p, Sec71p and Sec72p associate in different combinations depending on the signal sequence characteristics of the translocation substrate. The positively charged Sec62p N-terminus is important for its interaction with Sec63p (Wittke et al., 2000;Willer et al., 2003;M€ uller et al., 2010;Jung et al., 2014) and recent studies even suggested a role for the Sec62p-Sec63p complex in membrane protein insertion and topogenesis of signal anchor proteins (Reithinger et al., 2013;Jung et al., 2014Jung et al., , 2019. Moreover, the Sec62p-Sec63p complex has been shown to interact with Sec61 through Sec63p thereby causing lateral channel opening and activating Sec61 for post-translational protein translocation (Harada et al., 2011;Itskanov and Park, 2019;Wu et al., 2019). For successful import into the ER the luminal HSP70-like chaperone Kar2p/BiP binds to the polypeptide chain, while it is still translocated by Sec61 to prevent it from sliding back into the cytosol (Brodsky and Schekman, 1993;Brodsky et al., 1995;Panzner et al., 1995;Osborne et al., 2005;Rapoport, 2007). For this purpose, Kar2p/BiP interacts with the J-domain of Sec63p activating its ATPase domain and thereby closing its peptide binding pocket (Brodsky and Schekman, 1993;Brodsky et al., 1995;Osborne et al., 2005;Rapoport, 2007). After complete translocation, peptide-bound Kar2p/BiP is released by ADP-ATP exchange (Osborne et al., 2005;Rapoport, 2007).
For mammalian cells, favouring co-translational protein import, the homologous Sec62 has been shown to be interacting with the ribosomal exit tunnel, suggesting a role of Sec62 in co-translational protein transport as well and therefore a gain of function in comparison with its yeast homologue (M€ uller et al., 2010). This interaction is enabled by two oligopeptide motifs that are exclusively found in mammalian Sec62 but not in its yeast or plant homologues (M€ uller et al., 2010). Nevertheless, Lang et al. (2012) demonstrated that silencing of SEC62 impairs the post-translational import of a small, pre-secretory protein into the human ER, while co-translational translocation as well as post-translational insertion of tail-anchored proteins remained unaffected indicating that ribosome binding of Sec62 might only be relevant in coordinating protein translocation across the ER membrane. Furthermore, Lakkaraju et al. (2012) found that mammalian Sec62 ensures efficient post-translational translocation of proteins shorter than around 160 amino acids and is crucial for proteins up to a length of 100 amino acids. Mammalian Sec62 and Sec63 are associated with the Sec61 channel (Meyer et al., 2000;Tyedmers et al., 2000) (composed of Sec61a, Sec61b and Sec61c being the mammalian homologues of Sec61p, Sbh1p and Sss1p, compare with Mart ınez-Gil et al., 2011) and might also be involved in recruiting BiP as well as translocation substrates to the translocon (M€ uller et al., 2010;Lakkaraju et al., 2012). The interaction between Sec62 and Sec61 is Ca 2+ -sensitive and mammalian Sec62, possessing a potential C-terminal EF hand motif, is important for Ca 2+ homeostasis (Linxweiler et al., 2013). An additional role of Sec62 in stress recovery has recently been shown by demonstrating that a C-terminal region of human Sec62 is required for delivering ER proteins to the autolysosomal system (Fumagalli et al., 2016).
However, there is only scarce knowledge concerning the ER protein translocation in plants. The Arabidopsis thaliana genome encodes three isoforms of the channel-forming Sec61a homologue, two isoforms of the Sec63 homologue, named AtERdj2A and AtERdj2B (Yamamoto et al., 2008), but only one Sec62 homologue . Yamamoto et al. (2008) demonstrated the importance of AtERdj2A in protein translocation reflected by drastic pollen germination defects in the respective T-DNA insertion line probably due to defective protein secretion, whereas AtERdj2B was proposed to have an auxiliary role only. Interestingly, neither AtERdj2A nor AtERdj2B was able to rescue the thermosensitive growth phenotype of the respective yeast mutant (Yamamoto et al., 2008). The ER localized TPR protein AtTPR7 has recently been described and its interaction with the cytosolic chaperones HSP70 and HSP90 was demonstrated, implying a role in guiding pre-proteins to the Sec posttranslocon (Schweiger et al., 2012. Strikingly, AtTPR7 was able to restore post-translational protein import in a Dsec71 yeast knockout mutant and was proposed to functionally replace Sec71p and the TPR protein Sec72p (Schweiger et al., 2012). An interaction between AtTPR7 and AtERdj2A/B as well as AtSec62 has additionally been shown providing further information regarding the composition of the Arabidopsis Sec post-translocon (Schweiger et al., 2012;. In this study, we characterize the ER membrane localized protein AtSec62 (At3g20920), which shares only 12% sequence identity with yeast Sec62p and 15% with its mammalian counterpart . We demonstrate that AtSec62 interestingly has a third transmembrane domain in contrast with its yeast and mammalian homologues Schekman, 1989, 1990;M€ uller et al., 2010) resulting in its C-terminus being exposed to the ER lumen. The atsec62 T-DNA insertion line displays defects in pollen development and pollen tube germination implying an important role of AtSec62 for male fertility probably due to influencing protein secretion similar to AtERdj2A (Yamamoto et al., 2008). Furthermore, we show that AtSec62 is not able to rescue the thermosensitive growth phenotype of the respective sec62-ts yeast mutant and that its C-terminus is essential for AtSec62 function. We therefore propose that the AtSec62 C-terminus acquired a unique function in plants and that AtSec62, as part of the Sec post-translocon in Arabidopsis, is critical for plant growth and male fertility.

atsec62 displays vegetative and generative growth defects
We identified an atsec62 T-DNA insertion line with an insertion in exon 1 of AtSEC62 and homozygous mutant plants were selected by PCR using the oligonucleotides AtSEC62-5 0 UTR-f, AtSEC62-Exon2-r and Gabi-LB8409 (Figure 1a,b). Even though no obvious aerial phenotype was observable during initial growth, atsec62 plants clearly displayed an impaired growth in comparison with wild-type plants after 4 weeks ( Figure 1c). However, 2-week-old at-sec62 seedlings already showed an altered root morphology in contrast with Col-0 with their primary root length being reduced by 31% but a nearly 50% increase regarding the number of lateral roots ( Figure S1). RNA was extracted from leaves of 4-week-old plants and the absence of AtSEC62 transcript in atsec62 was confirmed by RT-PCR ( Figure 1d). Immunodetection using an AtSec62 specific antibody also revealed the absence of AtSec62 protein in microsomal membranes isolated from atsec62 leaves (Figure 1e). For complementation analysis, we generated a construct coding for the full-length AtSEC62 under the control of the endogenous promoter. Following stable transformation and selection of plants, we isolated two lines of homozygous atsec62 plants carrying the construct coding for endogenous AtSEC62, pAtSEC62::AtSEC62 #14-6 and pAtSEC62::AtSEC62 #27-2 (Figure 1b,d). In both lines, the atsec62 growth phenotype was completely rescued by expression of endogenous AtSec62 (Figure 1c).
During initial screening for atsec62 plants, we observed a very low number of homozygous mutant plants and consequently examined the segregation pattern of progeny plants derived from heterozygous atsec62 plants (+/À). Normal Mendelian segregation would imply 50% heterozygous progeny, 25% wild-type and 25% homozygous progeny. Interestingly, plants displayed only 8.83% homozygous progeny, but 40.38% wild-type and 50.79% heterozygous progeny (n = 317), significantly differing from the expected ratio (Table 1a). When grown on half-strength Murashige and Skoog (½MS) medium containing 10 lg ml À1 sulfadiazine, only plants carrying the T-DNA insertion should be resistant, resulting in 75% of resistant progeny plants derived from atsec62 (+/À). However, we observed only 58.63% and therefore assumed that the transmission of the AtSEC62 mutation was affected (Table 1b). Male or female gametophyte mutants typically display a segregation ratio of 1:1 (Liu and Qu, 2008;Drews and Koltunow, 2011), but progeny plants derived from atsec62 (+/À) rather segregated in a 3:2 ratio (Table 1b).

Defects in pollen development and tube germination lead to decreased male fertility in atsec62
Besides the distorted segregation ratio, we observed aborted and mostly empty siliques for atsec62 providing further evidence for a gametophytic defect in the mutant ( Figure 2a). For further analysis of potential gametophytic defects, we examined the siliques of atsec62 (+/À), which should display a reduced seed set or defective seeds if the female gametophyte was non-functional (Liu and Qu, 2008;Drews and Koltunow, 2011). As we observed no difference between wild-type siliques and those of heterozygous mutant plants, we conclude that female gametophytes in atsec62 are still functional.
We then tested pollen viability by Alexander staining which allows discriminating between viable (violet staining) and non-functional, greenish pollen grains (Alexander, 1969). When staining anthers of opening buds, the majority of atsec62 anthers appeared rather greenish with only occasionally faint violet staining and no pollen-like structures visible, whereas viable pollen grains were present in Col-0 and the complemented lines ( Figure 2b). Strikingly, when staining anthers of already open flowers, corresponding to dehiscent wild-type anthers, we observed some pollen-like structures clustered together in atsec62 anthers ( Figure 2b). Besides general defects in pollen development probably resulting in coherent pollen-like structures, we therefore propose that in atsec62 anther and pollen development are severely delayed in comparison with the female gametophyte and therefore proper pollination does not occur. In vitro pollen germination was subsequently examined by incubating pollen grains on solid medium for 22 h at either 25°C or 37°C. Interestingly, less pollen were released from atsec62 than from Col-0 anthers and the germination rate of atsec62 pollen at 25°C was reduced by 57% in comparison to wild-type pollen, while mutant pollen were hardly germinating at 37°C (Table 2). We therefore assume that the absence of AtSec62 also has a great impact on pollen tube germination, which is likewise reflected in the expression pattern of AtSEC62 with the highest expression being observed in the mature pollen (Arabidopsis eFP browser, http://bar.utor onto.ca/, Winter et al., 2007). Due to defective pollen development and pollen tube germination, atsec62 displays a drastically reduced male fertility resulting in only 3.79 AE 2.42 mg seeds/plant (n = 28) in comparison with Col-0 plants yielding 440.50 AE 26.33 mg seeds/plant (n = 7) (shown as mean AE SD, P = 4.86 9 10 À40 , compare with Figure 2c).
atsec62 is sensitive towards high-temperature and ER stress For further investigation of heat sensitivity of atsec62, we conducted a high-temperature stress treatment according to Yang et al. (2009). Seeds on ½MS medium were stratified for 3 days at 4°C, germinated at 37°C for 48 h and then transferred to 22°C. Germination was monitored 5 days after stress treatment, while survival rate was determined after 12 days growth at 22°C. Without heat treatment, the germination rate of atsec62 seeds is already reduced by 15% in comparison with Col-0 seeds, while there is no further effect on the germination rate upon high-temperature stress (Table 2). Whereas 25.96% of germinated atsec62 plants suffered and died from the previous treatment, wildtype plants survived (Table 2). In addition to the reduced tolerance of atsec62 against heat, we observed an increased susceptibility towards ER stress, when growing plants on ½MS medium containing dithiothreitol (DTT) ( Figure S2).

AtSec62 localizes to the ER membrane and has three transmembrane domains
Besides its role during plant growth and development, we were interested in the orientation of AtSec62 within the ER membrane. It has recently been suggested that AtSec62 has a third predicted transmembrane domain , which would alter its topology in comparison with its yeast and mammalian counterparts Schekman, 1989, 1990;M€ uller et al., 2010). For further investigation, we compared the amino acid sequences of different Sec62 homologues including higher plants (A. thaliana, Oryza sativa and Zea mays), the moss Table 1 Segregation analysis for heterozygous atsec62 (+/À) plants. (a) Segregation of atsec62 (+/À) in comparison with Col-0 and pAtSEC62::AtSEC62 #14-6 (#14-6) as well as pAtSEC62:: AtSEC62 #27-2 (#27-2). Progeny genotype was determined by PCR using AtSEC62 and T-DNA specific oligonucleotides. (b) Segregation on sulfadiazine (10 lg ml À1 ) was examined by investigating the number of resistant (Sulfa R ) and sensitive (Sulfa S ) plants. P-value for expected segregation, validation by chi-squared analysis ( Physcomitrella patens, the unicellular green algae Chlamydomonas reinhardtii as well as mammals (Homo sapiens and Mus musculus), Drosophila eugracilis and yeast (Saccharomyces cerevisiae). Interestingly, AtSec62 shares only 12% and 15% sequence identity with its yeast and mammalian counterparts . Two transmembrane domains were predicted for all analyzed Sec62 homologues, whereas a third transmembrane domain was only predicted for plant Sec62 containing many hydrophobic amino acids (Figure 3a, for complete sequence alignment see Figure S3). Furthermore, the C-terminal region of AtSec62 harbours predicted N-glycosylation sites with the consensus motif N-X-S/T (with X being any amino acid except proline) (compare with Marshall, 1972), which are absent in yeast and mammals (compare with Figure S3). We investigated AtSec62 topology by using the full-length AtSec62 as well as a truncated AtSec62 lacking the C-terminal region and the third transmembrane domain (D248À365), named AtSec62ÀDTMD3/C ( Figure 3b). We initially expressed N-and C-terminal GFP-fusion constructs in tobacco, named GFPÀAtSEC62 and AtSEC62ÀGFP, to test whether GFP-fusion would affect AtSec62 targeting and to confirm its ER localization. Both constructs and a construct coding for an ER marker protein fused to mCherry (compare with Nelson et al., 2007) were transformed into Agrobacterium and bacterial cells carrying the respective constructs were co-infiltrated into tobacco leaves. Fluorescence was monitored in intact leaves or isolated protoplasts. Both GFP-fusion proteins were localized to the ER, indicating that fusion of a GFP-tag would not affect AtSec62 targeting (Figures 4a-c and S4).
To confirm the altered topology of AtSec62, we made use of a Split-GFP Gateway system including plasmids coding for cytosolic or ER luminal GFP1-10 (Xie et al., 2017). In addition, we generated constructs with an N-or C-terminal fusion of the eleventh GFP b-sheet to AtSec62, named GFP11-AtSec62 and AtSec62ÀGFP11, or to AtSec62ÀDTMD3/C, named GFP11-AtSec62-DTMD3/C and AtSec62ÀDTMD3/CÀGFP11. After co-infiltration into tobacco leaves, we detected fluorescent signals in intact leaves as well as in isolated protoplasts (Figures 4d and  S5). Only upon co-expression with the cytosolic GFP1-10, there was an ER localized fluorescent signal detectable for GFP11ÀAtSec62, whereas fluorescence for AtSec62ÀGFP11 was only observable when co-expressed with the ER luminal GFP1-10ÀHDEL. In contrast, GFP11ÀAtSec62ÀDTMD3/ C and AtSec62ÀDTMD3/CÀGFP11 produced a detectable signal only with cytosolic GFP1-10. We therefore demonstrated that AtSec62 has three transmembrane domains   resulting in its N-terminus facing the cytosol, while its Cterminus is located in the ER lumen ( Figure 4e).
AtSec62 cannot rescue the growth phenotype of the yeast mutant sec62-ts As we have shown that AtSec62 has an altered topology compared to its yeast homologue, we further tested whether AtSec62 and Sec62p are conserved on a functional level. For this purpose, we conducted complementation analysis using the thermosensitive yeast mutant sec62-ts and the wild-type strain W303. Due to a thermosensitive Sec62p version, normal growth of sec62-ts is only enabled at permissive temperature (28°C) but not at restrictive temperature (37°C). We tried to complement this mutant with full-length AtSEC62, AtSEC62-DTMD3/C as well as with endogenous Sec62p as positive control. Strikingly, growth of sec62-ts at 37°C was still inhibited for mutants carrying the construct coding for AtSec62 and AtSec62ÀDTMD3/C or the empty vector control, whereas growth was fully restored upon expression of the endogenous Sec62p (Figure 5a). The presence of AtSEC62 and AtSEC62-DTMD3/C transcript was confirmed by RT-PCR (Figure 5b).

AtSec62 C-terminus is required for its function in plants
As the AtSec62 C-terminus is facing the ER lumen in contrast to yeast and mammalian Sec62 homologues, we were interested in whether the C-terminus has an essential function in protein translocation across the ER membrane in Arabidopsis. To investigate the role of the AtSec62 C-terminus, we performed complementation studies of atsec62 with a construct coding for AtSEC62ÀDTMD3/C (compare with Figures 3 and 6a) driven by the endogenous promoter. We identified two lines of homozygous mutant plants carrying the respective construct, pAtSEC62:: AtSEC62-DTMD3/C #30-23 and pAtSEC62::AtSEC62-DTMD3/C #12-22 (Figure 6b). Strikingly, the atsec62 phenotype was not rescued in these lines (Figure 6c,d), demonstrating that AtSec62ÀDTMD3/C cannot complement atsec62 and thereby indicating an essential role of the AtSec62 C-terminus in Arabidopsis. The presence of respective AtSEC62-DTMD3/C transcripts was confirmed by RT-PCR (Figure 6e), proving that the observed phenotype was not caused by lack of gene expression.

DISCUSSION
In this study, we demonstrated that AtSec62 has three transmembrane domains and is critical for plant growth and male fertility in Arabidopsis. Moreover, the luminal exposed C-terminus is crucial for AtSec62 function and probably acquired a unique function in protein translocation into the ER, which was also indicated by unsuccessful complementation in yeast and sequence comparison with other Sec62 homologues.
The isolated atsec62 T-DNA insertion line displayed delayed pollen development and impaired pollen tube germination leading to defects in male transmission. This phenotype resembles other mutant lines possessing a defective secretory pathway (Jakobsen et al., 2005;Yamamoto et al., 2008;Conger et al., 2011;Maruyama et al., 2014;Vu et al., 2017). For example, aterdj2a and atsec24a-1 display drastic defects in pollen germination leading to male sterility (Yamamoto et al., 2008;Conger et al., 2011). AtSEC24A is probably part of the COPII coat thereby playing an important role in the secretory pathway in plants by being involved in selective cargo binding during ER export (Kuehn et al., 1998;Conger et al., 2011). The quadruple mutant cnx1 crt1 crt2 crt3, lacking ER resident calreticulin and calnexin, both involved in protein quality control, shows reduced fertility due to reduced pollen viability and tube growth (Vu et al., 2017). Moreover, fusion of polar nuclei in female gametophytes and pollen tube growth in bip1 bip2 are affected, whereas bip1 bip2 bip3 pollen are not even viable (Maruyama et al., 2010(Maruyama et al., , 2014. Jakobsen et al. (2005) showed that MIA, involved in vesicle trafficking from the ER to the plasma membrane, is required for proper protein secretion and mia-1 has smaller leaves in comparison with wild-type plants and defective pollen which are not released and not germinating in vitro leading to reduced fertility. mia-1 still harbours remnants of the pollen mother cell and the callosic wall attached to pollen grains preventing proper development and release of the pollen (Jakobsen et al., 2005). Degradation of callose and the pollen mother cell wall is indispensable for male gametophyte development and depends on proper secretion from tapetal cells (Stieglitz, 1977;Rhee et al., 2003;Lu et al., 2014). Huang et al. (2013) have previously shown that lipid transfer proteins, which are secreted from the tapetum to become part of the microspore surface, are required for pollen exine formation. We assume that the lack of AtSec62 not only disturbs ER protein translocation of various polypeptides but subsequently also leads to defects in secretion during vegetative growth and male gametophyte development as already proposed for AtERdj2A by Yamamoto et al. (2008). Defective pollen development and reduced seed germination match the AtSEC62 expression pattern with the highest expression level observed in mature pollen and higher expression in seeds and during anther development than in vegetative tissue (Arabidopsis eFP browser, http://bar.utoronto.ca/, Winter et al., 2007). The altered root morphology in atsec62 might likewise be a result of disturbed ER protein translocation also interfering with hormone signalling and root nutrient uptake, for example regarding phosphate (compare with P erez- Torres et al., 2008;Malhotra et al., 2018), which in return also affects growth of aerial plant tissues thereby contributing to the observed phenotype. However, in contrast with aterdj2a (Yamamoto et al., 2008), we still obtained homozygous atsec62 plants and observed atsec62 (+/À) progeny plants not segregating in the expected ratio of 1:1 for usual gametophytic mutants. We therefore presume that the defects caused by the absence of AtSec62 can partially be bypassed or compensated by an alternative pathway. Another Arabidopsis mutant line, Atget1-1, has a similar, though more severe vegetative phenotype in comparison with atsec62, when additionally overexpressing AtGET3a (Xing et al., 2017). Respective plants display a dwarf phenotype, shorter roots as well as a reduced number of siliques and seeds (Xing et al., 2017). AtGET1 and AtGET3a are both part of the cytosolic/ER Guided Entry of Tail-anchored proteins (GET) pathway in Arabidopsis, which facilitates the membrane insertion of tail-anchored proteins into the ER membrane (Srivastava et al., 2017;Xing et al., 2017). As no severe phenotype is observable for Atget1-1 alone (Srivastava et al., 2017;Xing et al., 2017), Xing et al. (2017) proposed that overexpression of AtGET3a in the Atget1-1 background caused trapping of tail-anchored proteins in the cytosol and consequently also disturbed a potential alternative insertion pathway, which might be the post-translational Arabidopsis Sec translocon also involving AtSec62. In atsec62, the GET pathway should still be functional and we speculate that it might therefore partially compensate the lack of AtSec62 in post-translational translocation and insertion of membrane proteins explaining the less severe vegetative growth phenotype in comparison with plants overexpressing AtGET3a in the Atget1-1 background. We assume that the GET pathway might be more root specific, reflected by the root hair phenotype only observable for Atget1-1 but not for at-sec62 (Xing et al., 2017; Figure S1d). Moreover, the GET pathway can probably not compensate for lack of AtSec62 regarding the protein import of pre-secretory proteins resulting in impaired growth and drastic defects during male gametophyte development in atsec62. However, it still remains unclear whether there are indeed several distinct, maybe even overlapping ER translocation pathways in plants.
Besides impaired growth under normal conditions, we observed that atsec62 plants are more sensitive towards high-temperature stress than wild-type plants as shown by reduced pollen tube germination and survival after heat treatment. These observations might be due to a generally reduced fitness of atsec62 or due to an involvement of AtSec62 in plant thermotolerance. Yet, we observed no effect of high-temperature treatment on atsec62 seed germination rate probably due to storage compounds already being present in seeds and therefore low ER import rates, which increase during further vegetative growth. As we additionally observed a higher susceptibility towards ER stress, we assumed that AtSec62 might also be involved in the unfolded protein response in plants (reviewed by Strasser, 2018), probably by translocating respective components across or into the ER membrane. However, it might also play a more substantial role during ER stress (recovery) similar to its mammalian homologues (Linxweiler et al., 2013;Fumagalli et al., 2016).
Former studies have already pointed out the low homology between AtSec62 and Sec62p as well as mammalian Sec62 , especially regarding the diverse C-terminus which features potential N-glycosylation sites in plants, indicating exposition to the ER lumen rather than the cytosol. Indeed, we demonstrated that AtSec62 has a third transmembrane domain in contrast to its yeast and mammalian homologues Schekman, 1989, 1990;M€ uller et al., 2010) resulting in the AtSec62 C-terminus facing the ER lumen. As AtSec62ÀDTMD3/C is not sufficient for proper AtSec62 function as shown by unsuccessful complementation in at-sec62, we propose that the C-terminus plays an essential role in protein translocation into the ER, maybe by interacting with other Sec translocon components or even translocation substrates. Identifying additional AtSec62 interaction partners besides AtTPR7  within the Sec translocation complex or beyond will be a challenging task in the future providing further insights into ER protein translocation in Arabidopsis thereby also explaining the observed atsec62 phenotype. In addition, future studies might reveal whether AtSec62, similar to Sec62p (compare with Panzner et al., 1995), functions solely in the post-translational translocation pathway or also contributes to the co-translational pathway as proposed for mammalian Sec62 (M€ uller et al., 2010). We showed that AtSec62 as well as AtSec62ÀDTMD3/C cannot rescue the thermosensitive growth phenotype of the yeast mutant sec62-ts in contrast to the endogenous Sec62p, whereas recent studies by Zhou et al. (2016) have proven that Sec62 is conserved at least in other fungal species like the phytopathogen Magnaporthe oryzae with MoSec62 restoring growth of thermosensitive Dsec62 yeast. Human and Drosophila Sec62 homologues were also able to complement respective yeast mutants (Noёl and Cartwright, 1994;M€ uller et al., 2010). Due to the cytosol exposed C-terminus of Sec62p having an essential function in yeast, probably by contributing to signal peptide recognition (Deshaies and Schekman, 1990;Wittke et al., 2000), AtSec62 as well as AtSec62ÀDTMD3/C might be unable to restore growth of sec62-ts, even though AtSec62-DTMD3/C topology resembles the one of Sec62p. This might also be due to their inability to form proper complexes with other components of the yeast Sec translocon as already proposed by Yamamoto et al. (2008) after performing complementation analysis of sec63-1 yeast mutants using AtERdj2A. AtERdj2A and AtERdj2B display only around 20% sequence identity to their yeast and human counterparts, similar to AtSec62 sharing only 12% and 15% sequence identity with Sec62p and human Sec62 (Yamamoto et al., 2008;. Even regarding the predicted Sec62 domain (compare with , AtSec62 shows only 14% sequence identity with its yeast and human homologues. Notably, also yeast Sec62p and the well studied human Sec62 share only 19% sequence identity with respect to the conserved Sec62 domain.
Summarizing, presented findings support the idea of plant Sec62 having acquired a unique function during evolution. Although we prove the importance of the luminal exposed AtSec62 C-terminus, especially in plant male fertility, its exact function in ER protein translocation still remains unclear and will be an interesting issue to address in future studies.

Plant material and growth conditions
Wild-type A. thaliana (ecotype Columbia, Col-0) and the atsec62 T-DNA insertion line were grown on soil either under greenhouse conditions or in the climate chamber under a 16 h light/8 h dark cycle at 22°C/18°C and 100 lmol photons m À2 sec À1 . Plants for segregation analysis and high-temperature treatment were grown on ½MS medium (pH 5.7) with or without 10 lg ml À1 sulfadiazine, additionally containing vitamins, 1% sucrose and 0.6% (w/v) Gelrite, whereas plants for root analysis were grown on ½MS medium (pH 5.7) and 1.0% (w/v) Phytoagar, additionally supplemented with DTT (1.5 mM, 2 mM) for ER stress treatment. Seeds were sterilized (70% (v/v) ethanol, 0.05% (v/v) Triton X-100), prior to three times washing in 100% (v/v) ethanol. Seeds were stratified at 4°C for 2-3 days in the dark. For complementation analysis, we generated constructs under the control of the endogenous promoter (base pairs 800-1899 of genomic AT3G20920), which was narrowed down by using PlantPAN2.0 (http://plantpan2.itps.ncku.edu. tw/), coding for the full-length genomic AtSEC62 or the truncated AtSEC62-DTMD3/C (amino acids 1-247). Stable transformation of Arabidopsis plants was conducted by floral dip according to Clough and Bent (1998) using Agrobacterium tumefaciens GV3101. Nicotiana benthamiana was grown in soil under greenhouse conditions.

Preparation of microsomal membranes
Microsomal membranes were prepared from 4-week-old Arabidopsis plants. All steps were carried out at 4°C. Plant material was frozen in liquid nitrogen, ground in 50 mM TrisÀHCl (pH 7.5), 2 mM MgCl 2 , 100 mM KCl and 0.5 M sucrose and filtered through gauze. To remove chloroplasts and mitochondria, the plant lysate was centrifuged once at 4 200 g for 10 min, at 10 000 g for 10 min and at 22 000 g for 30 min. To pellet microsomal membranes the remaining supernatant was centrifuged at 100 000 g for 1 h. For further purification of microsomal membranes, the resulting pellet was resuspended in 50 mM TrisÀHCl (pH 7.5), 2 mM MgCl 2 , 100 mM KCl and 12% (w/v) sucrose, loaded onto a step gradient (50%, 30% and 20% sucrose) and centrifuged at 100 000 g for 1 h. The fraction between 50% and 30% sucrose was transferred to a new tube and washed once at 100 000 g for 1 h. Microsomal membranes were resuspended in 25 mM TrisÀHCl (pH 7.5), 0.33 M sucrose and 0.5 mM DTT.

SDS-PAGE and immunoblotting
Proteins were separated on 12% polyacrylamide. Subsequently, proteins were transferred onto PVDF membranes by semi-dry blotting and immunodetection was visualized using enhanced chemiluminescence. AtSec62 antibody was raised against the recombinant AtSec62 N-terminus (amino acids 1À160) fused to a C-terminal His-tag and was generated in rabbits by BioGenes (https://www.biogenes.de/).

Agrobacterium-mediated transient expression of fluorescent proteins in tobacco
Infiltration of leaves from 4-to 6-week-old Nicotiana benthamiana plants, isolation of protoplasts and subsequent detection of fluorescence signals were performed as described previously (Koop et al., 1996;Schweiger and Schwenkert, 2014). However, cell cultures of Agrobacterium tumefaciens Agl1 carrying respective constructs were resuspended in infiltration medium containing 10 mM MgCl 2 , 10 mM MES (pH 6) but only 100 lM acetosyringone. For transient expression of respective fluorescent proteins, leaves of Nicotiana benthamiana were infiltrated with Agrobacterium. Fluorescence signals in leaves or isolated protoplasts were detected after 2-3 days of incubation by confocal laser scanning microscopy (Leica TCS SP5 CLSM, imaging medium: glycerol, Software: Leica Application Suite/Advanced Fluorescence) at RT (compare with Schweiger and Schwenkert, 2014). For localization and topology studies, the AtSEC62 or AtSEC62-DTMD3/C coding region was cloned into the binary Gateway vectors pB7FWG2 and pK7WGF2 (Karimi et al., 2002) or the appropriate Split-GFP Gateway vectors (Xie et al., 2017). The constructed Split-GFP plasmids encode the GFP b-sheets 1À10 (compare with Cabantous et al., 2005) being expressed either in the cytosol (GFP1À10) or in the ER lumen by an integrated ER retention signal (GFP1À10ÀHDEL) (Xie et al., 2017). The Gateway vectors also facilitate N-or C-terminal fusion of the eleventh b-sheet to AtSec62 (compare with Xie et al., 2017). The C-terminal AtSEC62-DTMD3/C fusion construct contained an additional C-terminal His-tag. The ER marker has been described recently (Nelson et al., 2007;Schweiger et al., 2012).

Yeast strains and complementation analysis
Complementation analysis in yeast was performed using the wildtype strain W303 and the thermosensitive mutant sec62-ts (BY4741; MATa; ura3D0; leu2D0; his3D1; met15D0; sec62-ts: kanMX; obtained from EUROSCARF, http://euroscarf.de/). Yeast cells were grown in YPD medium (1% (w/v) bacto yeast extract, 2% (w/v) bacto peptone, 2% (w/v) glucose) or in SCD medium (0.7% (w/v) yeast nitrogen base without amino acids, 2% (w/v) glucose, 0.2% (w/v) dropout mix without leucine). Solid media were supplemented with 2% (w/v) bacto agar. For complementation analysis, the yeast SEC62 and AtSEC62 coding region were cloned into the Gateway vector pAG425GPD-ccdB (Alberti et al., 2007). Yeast genomic DNA was isolated as previously described (Lõoke et al., 2011), however, 70% (v/v) ethanol was used for DNA precipitation. Competent yeast cell preparation and subsequent transformation were modified based on Staudinger et al. (1995). For generation of competent cells, yeast cells were grown in 50 ml YPD medium at 30°C and cells were harvested at OD 600 0.5-0.6 by centrifugation for 5 min at 700 g and 4°C. Cells were washed in 50 ml sterile water, prior to washing in 12.5 ml LiSorb (100 mM LiOAc, 10 mM TrisÀHCl pH 8, 1 mM ethylene diamine tetraacetic acid (EDTA) pH 8, 1 M sorbitol). Yeast cells were subsequently resuspended in 300 ll LiSorb and 42 ll carrier DNA (2 mg ml À1 denatured and sheared herring sperm DNA) were added. For yeast transformation, 5 ll plasmid DNA and 300 ll LiPEG (100 mM LiOAc, 10 mM TrisÀHCl pH 8, 1 mM EDTA pH 8, 40% (w/v) PEG 3350) were added to 50 ll competent cells and the mixture was incubated for 20 min at RT. 35 ll dimethyl sulphoxide (DMSO) were added prior to a heat shock at 42°C for 15 min. Yeast cells were pelleted at 700 g for 90 sec, resuspended in 0.9% (w/v) NaCl and spread on SCD medium plates. Yeast transformants were then grown in SCD medium at 30°C, OD 600 was adjusted to 1.0 and serial dilutions of each strain were spotted onto SCD medium plates. Yeast cells were then grown for 48 h at 28°C or 37°C.

Statistical analysis
Chi-squared test was used to assess deviations of the observed segregation pattern from the expected Mendelian ratios, whereas root and seed yield analysis were performed using Student's t-test (two-sided, equal variance). The significance threshold was set at P ≤ 0.05. Box plots were generated using BoxPlotR software (http://shiny.chemgrid.org/boxplotr/).

CONFLICT OF INTEREST
The authors declare no conflict of interest.

AUTHORS' CONTRIBUTIONS
MJM conducted and analyzed experiments, participated in designing experiments and wrote the manuscript. FAB and TB performed experiments; FAB analyzed experiments. SS supervised the study, conceived experiments and revised the manuscript. All authors read and approved the final version of the manuscript.

SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article. Figure S1. Root morphology of atsec62 in comparison to Col-0. Figure S2. atsec62 sensitivity towards ER stress. Figure S3. Amino acid sequence alignment of Sec62 homologues. Figure S4. Localization of AtSec62 GFP-fusion constructs in tobacco protoplasts. Figure S5. Expression of AtSec62 Split-GFP constructs in isolated tobacco protoplasts.