DSH5 , a dihydrosphingosine C4 hydroxylase gene family member, shows spatially restricted expression in rice and is lethal when expressed ectopically

Dihydrosphingosine C4 hydroxylase (DSH), a diiron-binding membrane enzyme, catalyzes the hydration of dihydrosphingosine and acyl-sphinganine to produce phytosphingosine and phytoceramide, respectively. Rice has two types of DSH homologs: general DSHs, namely DSH1, DSH2 and DSH4, and others that show spatial expression proﬁles, namely DSH3 and DSH5. The general DSHs exist in many plant species. These DSHs showed similarity in their functions and complemented the yeast sur2D mutation. In contrast, homologs of DSH3 and DSH5 were found only in monocot plants. Phylogenetic analysis placed these DSHs in different clades that are evolutionarily divergent from those of the general DSHs. DSH3 and DSH5 showed low-level expression. DSH5 expression was speciﬁcally in vascular bundle tissues. Ectopic expression of DSH5 induced a dwarf phenotype characterized by severe growth inhibition and an increase in the thickness of the leaf body caused by enlargement of bulliform cells in the leaves. However, no signiﬁcant difference was observed in the amount of sphingolipid species. DSH5 did not complement the yeast sur2D mutation, implying that DSH5 has little effect on sphingolipid metabolism. These ﬁndings suggested that DSH3 and DSH5 originated and diverged in monocot plants.


INTRODUCTION
Sphingolipids, a class of lipids derived from the aliphatic amino alcohol sphingosine, are ubiquitous constituents of eukaryotic cells that also occur in some bacteria (Lynch and Dunn, 2004). The biosynthesis of sphingolipids starts with the condensation of L-serine and palmitoyl-CoA to form 3-ketosphinganine (Tamura et al., 2001), which is then converted to dihydrosphingosine (DHS) by NADPH-dependent reduction at the C3 position (Beeler et al., 1998).
DHS and its derivatives are composed of sphingoid long-chain bases (LCBs) (Lynch and Dunn, 2004). The composition of LCBs is divided into three main types: mammalian, yeast and plant (Dickson, 1998;Lynch and Dunn, 2004). In mammals, sphingolipids mainly occur as ceramide derived from the N-acylation of DHS; these sphingolipids have both a long-chain fatty acid, ranging from 14 to 26 carbons in length, and a double bond at the C4-C5 position (Michel et al., 1997). In yeast, most DHS is hydroxylated at the C4 position and converted to 4-hydroxysphinganine, which is also known as phytosphingosine (PHS) (Dickson, 1998;Wright et al., 2003). In plants, a proportion of PHS is further modified by desatu-ration at the C8-C9 position following PHS synthesis, as occurs in yeast (Sperling et al., 1998). Plant ceramide (phytoceramide) is composed of these compounds, as is the mammalian-type ceramide (Lynch and Dunn, 2004).
The activity of DSH is believed to define the DHS/ PHS balance of sphingoid bases in yeast and of plant sphingolipids (Haak et al., 1997;Sperling and Heinz, 2003). DSH is considered a key enzyme in PHS synthesis; however, DSH (SUR2) activity, which is required for the hydroxylation of C-4 of the sphingoid moiety of ceramide, is not essential for the growth of Saccharomyces cerevisiae (Haak et al., 1997;Grilley et al., 1998). Two Arabidopsis thaliana DSH genes, SBH1 and SBH2, have been reported, and they complement yeast SUR2 gene function (Sperling et al., 2001). In Arabidopsis, the synthesis of trihydroxy LCBs plays a central role in maintaining growth and mediating the total content and fatty acid composition of sphingolipids. Double mutants, which completely lacked trihydroxy LCBs, were severely dwarfed, did not progress from vegetative to reproductive growth, and had enhanced expression of programmed cell death-associated genes. In these mutants, increases in sphingolipid content resulted from the accumulation of molecular species that have C16 fatty acids rather than very-long-chain fatty acids (VLCFAs) (Chen et al., 2008).
The importance of SBH genes in Arabidopsis growth can be explained by the plant's requirement for the ceramide species and by the substrate specificity of ceramide synthases. The Arabidopsis ceramide synthases LOH1 and LOH3 but not LOH2 are essential for plant growth (Markham et al., 2011;Ternes et al., 2011). The knockdown loh1-1 loh3-1 double mutant shows a less drastic reduction in VLCFA-containing sphingolipids and reduced growth. loh2 mutant seedlings show both a strong reduction of 16:0 fatty acids in ceramide and other sphingolipids and a significant increase in VLCFA-containing sphingolipids. These results suggest the existence of two classes of ceramide synthases that have different substrate specificities, one preferring both C16 fatty acids and dihydroxy-LCBs and the other preferring both VLCFAs and trihydroxy LCBs (Markham et al., 2013). In contrast, no phenotype is detected in mutant yeast cells with a deletion of either LAG1 or LAC1/DGT1, which are homologs of the LOH genes, whereas lag1Δ dgt1Δ cells display growth defects (Barz and Walter, 1999), suggesting functional redundancy. The unique evolutionary differentiation of plant LOH genes may provide roles for DSH genes in the control of the ceramide biosynthesis pathway in plants.
The rice genome contains five homologous DSH genes, DSH1-DSH5, whose gene products show sequence similarity to yeast SUR2. We have found that DSH1 and DSH4 complement the yeast sur2D mutation. We have also reported that DSH1 is essential for fertility even though rice has five similar DSH genes in its genome. RNAi-mediated knockdown of DSH1 stimulated the up-regulation of DSH2 in vegetative organs but not in reproductive organs (Imamura et al., 2007). DSH3 and DSH5 are diverged from other DSH family members, having relatively low sequence conservation with the others. These observations suggest that there is functional conservation but physiological differentiation among the members of the rice DSH gene family. Here, we report functional differences among rice DSHs and discuss the evolutionary divergence within the DSH gene family in plants.

MATERIALS AND METHODS
Plant materials and growth conditions Oryza sativa L. cv. Nipponbare was used as the wild-type plant. Seeds were germinated at 30 °C in a dark chamber. The seedlings were grown under continuous light (13 μmol m − 2 s − 1 ) for three days. The rice plants were cultivated in a growth chamber with 80% humidity under long-day conditions that consisted of a cycle of 14 h of light (30 °C, 270 μmol m − 2 s − 1 ) and 10 h of dark at 25 °C.
RNA quantification using real-time RT-PCR Realtime RT-PCR (reverse transcription-PCR) was carried out according to previously described methods (Imamura et al., 2007). The amounts of DSH1 (Os06g0227200), DSH2 (Os02g0745700), DSH4 (Os07g0627700) and DSH5 (Os01g0658600) transcripts were determined as values relative to those of the Actin1 transcript (acc. no. X16280). Detailed procedures are given in the Supplemental Methods.
Creation of the gene for DSH5 transformants For the forced expression of DSH5, the coding region was inserted into the region following the CaMV 35S promoter in pGWB2 (Nakagawa et al., 2007). For the ectopic expression of DSH5, the region for the DSH5 ORF was prepared by PCR from the corresponding cDNA using the primers 5'-CACCATGGGGTTCATCAGCGGCGAG-GAGC-3' and 5'-CAACGATCGCCACGTTCTCCTG-3', and then inserted into pENTR-D-TOPO. The resultant plasmid was used for construction of the appropriate plasmids by LR clonase using pGWB2 (Nakagawa et al., 2007).

Analysis of LCBs and sphingolipid composition
Yeast sur2D mutants containing the expression plasmid for each of DSH1, DSH2, DSH4 or DSH5 were prepared using a pYES2 vector. The LCBs in the transformants were analyzed using HPLC as described previously (Imamura et al., 2007). The sphingolipid composition in plant leaves was analyzed according to the methods of Ishikawa et al. (2016). Leaves were freeze-dried prior to the analysis.
Complementation test in yeast Complementation by DSH1, DSH2, DSH4 and DSH5 of the yeast sur2D mutation was determined as described previously (Imamura et al., 2007). The cultured cell suspension was diluted with YNB medium (Becton Dickinson Japan, Tokyo) to adjust its concentration to OD 600 = 0.3, and then portioned to a series of 1:2 dilutions ranging from 1:2 to 1:32. Each aliquot was inoculated onto YNB plates, in which 4 μg/ml syringomycin E was supplemented as required. Cells were grown at 30 °C.
Phylogenetic analysis Homologs of DSH1, DSH2, DSH3, DSH4 and DSH5 in various plant species were queried using the BLAST program and the DDBJ (http:// www.ddbj.nig.ac.jp/). The amino acid sequences were aligned using MUSCLE (Edgar, 2004), and the maximum likelihood tree was constructed using MEGA7 (Kumar et al., 2016) based on the JTT matrix-based model (Jones et al., 1992) with 1,000 bootstrap replicates.

RESULTS
Phylogenetic analysis of DSHs There are five DSH genes in the rice genome: DSH1 to DSH5. Homologs of DSHs are widely distributed in the plant kingdom. We constructed a phylogenetic tree of the DSHs of various plant species including a moss (Physcomitrella patens) and a fern (Selaginella moellendorffii). The tree showed that the homologs could be divided into two groups based on their evolutionary relationship; one group had subfamilies containing homologs of DSH1, DSH2 and DSH4,     and the other contained those of DSH3 and DSH5 (Fig.  1A). Homologs of DSH1, DSH2 and DSH4 exist in many plant species. However, homologs of DSH3 and DSH5 were found only in monocot plants (Fig. 1B). This suggested that they originated after the divergence of monocot plants in angiosperms.
Expression patterns of rice DSH genes Among the DSH genes, we have determined that DSH1 (Os06g0227200), DSH2 (Os02g0745700) and DSH4 (Os07g0627700) are abundantly expressed, but we have not detected any DSH3 (Os06g0330400) transcripts (Imamura et al., 2007). To date, the rice expression database RiceXPro (http://ricexpro.dna.affrc.go.jp/) has shown that DSH3 is specifically but weakly expressed both in the roots during vegetative stages and in the pistil, whereas DSH1, DSH2 and DSH4 are highly expressed in many organs ( Supplementary Fig. S1). However, there are no registered cDNA data and no reports on the expression profile for DSH5 (Os01g0658600) in the database.
In this work, we detected DSH5 transcription. Realtime RT-PCR detected transcripts of the DSH5 gene specifically in the leaves (Fig. 2). The spatial pattern of DSH5 gene expression was analyzed using a GUS reporter gene. GUS activity was detected in the cells of vascular bundle tissues in the leaves, roots, glumes and filaments of the stamen (Fig. 3).
To determine functional similarity between rice DSH5 and other DSHs, we transformed the yeast sur2D mutant with rice DSH1, DSH2, DSH4 and DSH5 cDNAs. We detected phytosphingosine (PHS) and dihydrosphingosine (DHS) in the transformants of the sur2D mutant in which either DSH2 or DSH4 was expressed. However, No PHS was produced in the yeast transformant containing DSH5 (Fig. 4A). The PHS/DHS ratio was higher in yeast cells containing DSH4 than those containing DSH1 and DSH2 (Fig. 4B). The phenotype of the yeast sur2D mutation can be determined by the acquisition of tolerance to syringomycin E (Grilly et al., 1998). As shown in Fig.  4C, the sur2D mutant containing DSH4 lacked tolerance to syringomycin E, and that containing DSH1 was less tolerant than the sur2D mutant. However, the mutant containing DSH5 maintained obvious tolerance to syringomycin E, indicating that DHS5 did not complement the sur2D mutation. These results suggest that DSH5 has a different function from DSH1, DSH2 and DSH4.
Ectopic expression of DSH5 leads to a lethal phenotype DSH5 shows relatively low similarity to the other DSHs and has differences at many amino acid positions, although it possesses highly conserved histidine residues within the diiron center region (data not shown). To understand the functional difference between DSH5 and other DSHs, we created transformed rice plants (DSH5ox) that contained the DSH5 gene, whose expression was driven by the CaMV 35S promoter.
We obtained many DSH5-overexpressing transformants, but all of them showed a dwarf phenotype and their mature leaves were small. From the DSH5transformant callus, we regenerated numerous plants,  which were transplanted to pots for culture in a greenhouse when they reached a height of ~10 cm. However, the growth of these plants was severely reduced. Most of them stopped gaining height, and died during the vegetative stage. Plant height of the DSH5 transformant was severely reduced (Fig. 5A-5C). We observed that bulliform cells in the leaves were significantly enlarged, resulting in an increased thickness of the leaf body; however, no morphological change was observed in the vascular bundles in the leaves (Fig. 5D-5F).
We analyzed the amounts of representative species of sphingolipids in the transformants. However, no marked differences in the amounts of these species, or the ratio of the t18/d18 species, were detected between DSH5ox and WT plants (Fig. 5G). These results suggest that ectopic DSH5 expression had little effect on sphingolipid metabolism, even though significant morphological changes were observed.

DISCUSSION
DSH1, DSH2 and DSH4 showed functional similarity to each other. Yeast transformants containing these genes exhibited PHS synthesis activity. These DSHs complemented the yeast sur2D mutation. In contrast, DSH5 did not produce any PHS in the yeast sur2D mutant, and the yeast sur2D mutation was not complemented by the introduction of DSH5 (Fig. 4).
In our previous work, we did not detect any transcripts of DSH3 or DSH5 (Imamura et al., 2007). Currently, the established database RiceXPro indicates that DSH3 is specifically expressed in roots and pistils (Supplementary Fig. S1). However, there is no description of DSH5 gene expression in the database. In the present work, we detected transcripts of rice DSH5. Its expression pattern was highly specific to vascular bundle tissues (Fig. 3). Ectopic expression of DSH5 in transgenic rice plants resulted in severe growth inhibition in the vegetative stage, and these plants did not survive to the reproductive stage (Fig. 5); however, no evident alteration in the composition of sphingolipids occurred in the DSH5 transformants (Fig. 5). Therefore, we conclude that DSH5 has some unknown function that largely influences plant growth. These results imply that the DSH5 protein is active at the specific location where it is expressed, although its function is unclear.
DSH3 and DSH5 exhibit relatively low similarity with the general DSHs. Homologs of DSH3 and DSH5 are widespread in monocot plant species but were not found in any dicot species (Fig. 1). These homologs composed two different clades that had evolutionarily diverged. Brachypodium distachyon and Musa acuminata have multiple DSH3 homologs, but Zea mays and rice possess only a single counterpart; only one DSH5   homolog was found in the monocot plant species (Fig.  1). These homologs probably originated from the general DSHs and have become highly divergent in monocot plants. We propose the following history of DSH evolution in plants: an ancestral DSH enzyme originated in lower plants. DSH homologs were generated by duplication and dispersed in angiosperms. These homologs were established as general DSHs but then diverged into two major groups, one containing DSH1 and DSH2 and the other containing DSH4. Monocot plants acquired a second type of DSH homolog, comprising DSH3 and DSH5, which may have specialized roles, as they show distinctive expression patterns in specific regions of the plant body, namely the roots and bundle tissues in the leaves, respectively. Ectopic expression of DSH5 had lethal effects on plant growth (Fig. 5), which suggests that DSH5 homologs have unique characters that are different from those of the general DSHs.