Biological membranes compartmentalize cells and, in concert with embedded and associated protein complexes, regulate transport and signal transduction between cells and cellular compartments. Membranes are composed of a phospholipid bilayer consisting of a hydrophobic core sandwiched between hydrophilic head groups. Phosphatidylethanolamine (PtdEtn) and phosphatidylcholine (PtdCho) are the two most abundant phospholipids found in eukaryotic membranes. Their polar head groups, phosphoethanolamine and phosphocholine, are synthesized in the earliest steps of the Kennedy pathway when ethanolamine and choline, respectively, are phosphorylated by choline/ethanolamine kinases (CEKs). Later in the pathway these phosphorylated head groups are enzymatically coupled to the diacylglycerol to create PtdEtn and PtdCho which are subsequently incorporated into membranes (Gibellini and Smith, 2010). Mutations in CEKs and other Kennedy pathway enzymes result in a wide range of phenotypes in plants as well as heritable genetic diseases in humans, highlighting the importance of this core biosynthetic pathway in all eukaryotes (Kwon et al., 2012; Lin et al., 2015, 2020; Yunus et al., 2016; Tavasoli et al., 2020; Tannert et al., 2021). Extensive efforts to characterize the Kennedy pathway have revealed that most organisms have multiple CEK isoforms encoded by distinct CEK genes.

The Arabidopsis genome encodes four CEKs numbered CEK1-4. These genes are expressed at different levels compared to each other and exhibit tissue-specific and environmentally responsive expression patterns, suggesting that each enzyme plays a distinct role in choline and ethanolamine metabolism (Tasseva et al., 2004; Lin et al., 2020). Examples of sub-specialization in this gene family include diverse substrate specificities among the CEKs as well as their differential effects on plant growth and development. In vitro, CEK1 and CEK2 have both choline and ethanolamine kinase activity while CEK3 and CEK4 have substrate specificity for choline and ethanolamine alone, respectively. Neither CEK1 or CEK2 exhibited detectable ethanolamine kinase activity in vivo, suggesting that CEK4 is the principal ethanolamine kinase in Arabidopsis (Lin et al., 2019, 2020). Similar enzymatic specialization has also been described in Glycine max, suggesting that it is widespread in plants (Wharfe and Harwood, 1979). Complementing this biochemical approach, reverse genetics has also been used to reveal the developmental and biochemical roles of the Arabidopsis CEKs. cek2 and cek3 single mutants have short roots while cek1 mutant roots are normal. This shows that at least two of the Arabidopsis CEKs are required for normal growth and development. Additional insights into CEK function have come from a chemical rescue experiment. In this experiment cek3 seedlings were grown on media containing phosphocholine, which rescued the cek3 short-root phenotype. This experiment’s results are consistent with the biochemical characterization of CEK3 as a choline kinase. cek mutants have also been used to provide in vivo support for the in vitro biochemical characterization. cek1 single mutant and cek1 cek3 double mutant plants were shown to have decreased choline kinase activity in vivo (Lin et al., 2020).

Compared to CEK1-3, CEK4 remains relatively uncharacterized. This is because cek4 loss of-function mutants are embryo lethal (Meinke et al., 2008; Lin et al., 2015; Meinke, 2020). The viability of cek1, cek2, and cek3 T-DNA loss-of-function mutants coupled with the lack of detectable changes in lipid content in their single mutants suggests that cek1-3 have partially overlapping functions in lipid metabolism and plant development (Lin et al., 2015, 2020). CEK4 stands out from the other Arabidopsis CEKs as the functional alleles of CEK1, CEK2, and CEK3 in cek4 mutants cannot compensate for the loss of CEK4 activity. Fundamental differences between CEK4 and the other CEKs are also suggested by the observation that CEK4 is phylogenetically distant from the other Arabidopsis CEKs (Lin et al., 2015). Combined with CEK4’s ethanolamine substrate specificity in vitro, it appears that CEK4 plays unique and important roles in plant biochemistry and development. These roles are only partially understood. CEK4 is expressed throughout the plant, including throughout heart stage and later embryos. High expression levels are found in the reproductive organs and in pollen, roots, and trichomes (Honys and Twell, 2004; Marks et al., 2009; Gilding and Marks, 2010; Lin et al., 2015). A 35S:CEK4 line that complemented cek4 embryonic arrest was shown to result in increased levels of both PtdEtn and PtdCho in seedlings and increased levels of PtdCho in mature siliques, consistent with a predicted function as a CEK (Lin et al., 2015). No further phenotypes were reported for this overexpression line and, to the best of our knowledge, no roles in post-embryonic plant development or responses to the environment have been described for CEK4.

Here we describe the isolation of a viable EMS induced allele of CEK4 that appears to result in a partial loss of function. This novel allele reveals important post-embryonic developmental functions as well as a role in chilling responses for CEK4.

CEK4 is the least well characterized choline/ethanolamine kinase in Arabidopsis because it is required for embryonic viability and, up to now, only T-DNA loss of function alleles have been available. This has made it hard to establish what roles it plays in plant growth and development. In this study we isolated bst, a partial loss of function allele of CEK4, and used it to demonstrate that CEK4 plays roles in post-embryonic developmental patterning and cold responses.

CEK4 had previously been shown to be required for embryogenesis as cek4 T-DNA mutant embryos arrest early during development (Lin et al., 2015). While it is not surprising that a key enzyme in a core biosynthetic pathway would be an essential gene, there are three other CEK genes in the Arabidopsis genome which could compensate for a loss of CEK4 activity. The observation that they don’t suggests that CEK4 has specialized functions that overlap minimally, if at all, with those of the other CEKs. The bst allele of CEK4 behaves as a partial loss of function, demonstrated by the increased phenotypic severity when it is combined with a loss of function allele. bst is a point mutant, altering a key residue in the ATP binding pocket of the enzyme. The biochemical consequences of this change are uncharacterized but presumably the mutation results in decreased kinase activity due to altered ATP binding. This hypothesis could be tested in the future by directly assaying the effects of the bst T141I substitution on ATP binding and ethanolamine kinase activity, either in vitro or in vivo. We have not determined if bst mutant plants have alterations in their membranes. An analysis of any alterations may provide insights into the role that CEK4 plays in the biological pathways that are affected in bst mutants.

bumpy stem plants have pleiotropic phenotypes, consistent with CEK4 expression throughout plant development. These phenotypes include short roots and stems, swollen trichome bases, partial sterility, stem bumps at stem/pedicel junctions, and decreased chilling tolerance. These results suggest that the correct flux of metabolites through the Kennedy pathway is important not only for membrane biosynthesis but also for a variety of higher order processes in plants. Although we did not generate mechanistic insights into how bst results in decreased chilling tolerance, there are examples that illustrate the role of temperature dependent changes in phospholipid composition on temperature dependent responses. One example is temperature dependent flowering. The florigen FT has been shown to bind to membrane phospholipids at low temperatures, preventing it from moving and functioning as a signal. This serves as a mechanism that allows plants to delay flowering in response to chilling (Nakamura et al., 2014; Susila et al., 2021). Changes in the phospholipid composition of membranes could have direct functional effects on the protein machinery embedded in or associated with them which, in turn, may alter intra- and inter-cellular signaling (Champeyroux et al., 2020). In addition, there are many well established examples of lipids functioning directly as signaling molecules in plants (Wang and Chapman, 2013; Ruelland and Valentova, 2016). These examples include roles for lipids in responding to abiotic stresses such as chilling and freezing responses (Welti et al., 2002; Hou et al., 2016).

CEK4 has been shown to have ethanolamine kinase but no detectable choline kinase activity in vitro (Lin et al., 2020). This does not mean that mutations in CEK4 will only affect phosphoethanolamine production as phosphoethanolamine can be converted to phosphocholine by phosphoethanolamine N-methyltransferases (PMTs) (Figure 6). The observation that pmt1 mutants have short roots (Cruz-Ramírez et al., 2004) demonstrates that normal levels of phosphocholine, produced from phosphoethanolamine, are required for robust root growth. Arabidopsis pmt1 pmt3 double mutant plants have reduced phosphocholine levels and exhibit developmental phenotypes including small roots and shoots, but not bumpy stems, that can be rescued by choline or phosphocholine supplementation (Chen et al., 2018, 2019). pmt1 pmt2 pmt3 triple mutants are seedling lethal and exhibit up to 80% reductions in phosphocholine levels. This suggests that the bulk, if not all of, the phosphocholine found in Arabidopsis is produced from phosphoethanolamine and that decreased ethanolamine kinase activity will result in decreased phosphocholine levels (Liu et al., 2018, 2019; Chen et al., 2019). The conversion of phosphoethanolamine to phosphocholine by PMTs probably explains why CEK4 overexpression in Arabidopsis resulted in both increased PtdEth as well as increased PtdCho levels even though CEK4 only has ethanolamine kinase activity (Lin et al., 2015).

We tested the prediction that the bst short root phenotype would be rescued by supplementation with phosphoethanolamine, CEK4’s predicted product. Surprisingly, phosphoethanolamine only partially rescued bst root phenotypes and only did so at relatively high concentrations (1 mM). The same was true for ethanolamine, CEK4’s predicted substrate. We also supplemented bst seedlings with choline and phosphocholine. Rescue with both of these compounds was nearly complete at relatively low concentrations (100 μM). These unexpected results do not provide clear insights into bst’s root phenotypes; understanding them better will require a detailed metabolic analysis. One possible interpretation of these results is that, although CEK4 functions principally or solely as an ethanolamine kinase, much of the phosphoethanolamine it produces is subsequently used as a substrate for the choline branch of the Kennedy pathway (Figure 6). Since a reduction in PMT generated phosphocholine results in short roots (Cruz-Ramírez et al., 2004; Chen et al., 2019; Liu et al., 2019), the bst phenotype could be principally due to reduced levels of choline containing compounds as opposed to ethanolamine containing compounds. This could explain the rescue with lower concentrations of choline containing compounds than ethanolamine containing compounds. Presumably only some of the supplemental phosphoethanolamine would be converted to phosphocholine by PMTs, reducing the efficiency of rescue with ethanolamine containing compounds relative to choline containing compounds. This idea could be tested by supplementation with labeled compounds followed by a metabolic analysis. Combining such an analysis with chemical inhibition of PMTs (Liu et al., 2019) or by using bst pmt double mutants could further refine our understanding of the flux of metabolites in the Kennedy pathway on plant growth and development. When combined with mutations in other CEK genes, bst may also provide insights into the in planta functions of this important gene that are shared with the other CEKs.

The viability of the bst allele provides a useful tool for further investigation of CEK4 functions in development and responses to the environment. This study did not attempt to define the defects that underlie the developmental, reproductive, or chilling phenotypes we identified. Of special interest is the bumpy stem phenotype that this allele was named for. This phenotype is unusual and we are not aware of other Arabidopsis mutants that display this phenotype. Organ boundaries in plants are delineated by LOB gene expression (Shuai et al., 2002) and regulated by a complex genetic network that includes CUC and other genes (Hepworth and Pautot, 2015). Investigating the expression domains of LOB, CUC, and other boundary network genes in bst mutants could provide insights into the developmental signaling mechanisms that link phospholipid head group metabolism and the establishment of boundaries between organs.

Strains described in this manuscript have been deposited at the ABRC with accession numbers CS73151 (bst^Col-0) and CS73152 (bst^Ler). The sequence data generated in the course of mapping and cloning bst have been deposited in the NCBI SRA as BioProject PRJNA796015.

CR and MC conceived of and performed the experiments and edited the manuscript. NK conceived of and performed the experiments, supervised the work, wrote the original draft, and edited the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.