Polar targeting of proteins – a green perspective Free

Most eukaryotic cells are organized along axes in space. Such coordinates may reflect the front and back or the top and bottom of a single cell. In multicellular organisms, cellular axes are often aligned with supracellular (tissue, organ or organismal) axes, allowing individual cells to coordinate their growth, division and differentiation with each other (reviewed in Ramalho et al., 2022). For this coordination to lead to organized multicellular development, each cell needs landmarks that reflect supracellular axes. Within cells, these landmarks are part of cell polarity systems that encompass the anisotropic distribution of components. Whereas any structure or molecule could serve as a polarity landmark, here we focus on proteins that are either embedded in the plasma membrane (PM) or located close to it in the cellular cortex.

In recent decades, much has been learned about the identity of such polarity proteins, the mechanisms that guide their localization and the pathways through which they influence cell function. However, most of what has been learned is derived from studies on either (model) animals (Kemphues et al., 1988; Wright et al., 2022) or fungi (Adams et al., 1990; Chiou et al., 2017), whereas much less is known about plants. Plants gained their multicellularity independently from animals and fungi, and this could have had a major impact on the evolution of cell polarity in the context of plant development, where the polarity pathways, proteins and mechanisms involved are known to be very different (Kania et al., 2014). For example, in contrast to animals, plants undergo continuous growth and formation of new structures post-embryonically throughout their life. This ongoing development might require more flexible polarity mechanisms in the meristems, where stem cells differentiate into new tissues and organs. Moreover, plant cells have specific components that are absent in the cells of other multicellular eukaryotes. These include cortical microtubules, cell walls and their modifications such as Casparian strips, which are asymmetrical band-like cell wall thickenings in the root endodermis that regulate water transport. These components require additional polarity mechanisms and partially explain the lack of conservation across kingdoms (if any factors were present in the unicellular last common ancestor of plants and animals). Plant cells and tissues also have intrinsic polarity, which enables plant polar proteins to polarize even outside their native expression domains (Chan et al., 2020; Wang et al., 2022; Yoshida et al., 2019). Understanding how these proteins interpret polarity signals in various tissues could provide insights into the evolution of multicellular tissues.

In this Review, we discuss insights into the mechanisms underlying polar protein targeting in plants. Unless otherwise specified, most of our discussion is based on findings from the best-studied model species Arabidopsis thaliana. In our discussion, we divide polar proteins into two subclasses: transmembrane proteins (those having at least one transmembrane helix) and cytosolic proteins (those that can still be associated with the PM at the cortical region of the cell and are thus also referred to as cortical proteins). Given the distinct routes for their translation and targeting, we discuss the principles guiding these two classes of proteins separately. We then reflect on the mechanisms that allow cells to establish and maintain polar membrane domains. Throughout, we discuss homologies and analogies between polar targeting mechanisms among eukaryotic clades.

Studies on the mechanisms of polar targeting of transmembrane proteins in plants have focused on a small set of proteins that are either transporters, receptor-like kinases, scaffolds or proteins that lack well-defined biochemical functions. The transporters include the ion transporters BOR1 (BORON TRANSPORTER 1) (Takano et al., 2010; Yoshinari et al., 2012), NIP5;1 (NOD26-LIKE INTRINSIC PROTEIN 5;1, also known as NIP5-1) (Wang et al., 2017), IRT1 (IRON-REGULATED TRANSPORTER 1) (Barberon et al., 2014), Oryza sativa (Os)Lsi1 (Low silicon protein 1, also known as NIP2-1) and OsLsi2 (Low silicon protein 2, also known as LSI2) (Konishi et al., 2023), as well as PIN (PIN-FORMED) auxin transporters (Marhava, 2022). The receptor-like kinases include IRK (INFLORESCENCE AND ROOT APICES RECEPTOR KINASE) (Campos et al., 2020; Goff and Van Norman, 2021 preprint; Rodriguez-Furlan et al., 2022), PXC2 (PXY/TDR-CORRELATED 2, also known as CANAR) (Goff and Van Norman, 2021 preprint; Hajný et al., 2020), MUS (MUSTACHES) and MUL (MUSTACHES-LIKE) (Xun et al., 2020), KOIN (KINASE ON THE INSIDE) (Rodriguez-Furlan et al., 2022), and SGN3 (SCHENGEN 3, also known as GSO1) (Alassimone et al., 2016; Fujita et al., 2020; Pfister et al., 2014). Additionally, CASP (CASPARIAN STRIP MEMBRANE DOMAIN PROTEIN) scaffolds (Roppolo et al., 2011) and TET1 (TETRASPANIN 1, also known as TORNADO 2), a protein with an unclear function (Konstantinova et al., 2024), have been studied in the context of polar targeting. Interestingly, several of these proteins, notably PIN proteins, have also been widely used as models for general membrane trafficking. It is therefore possible that the inferred mechanisms for non-polar membrane targeting are biased for components involved in polar targeting, and components in non-polar targeting might have been missed.

Polar targeting almost certainly uses many generic components of secretory systems, yet also has modifications that allow for spatial specificity. Some components rely on cytoskeleton orientation, such as vesicle transport during tip growth (Zhang et al., 2023) or exocyst localization (Vukašinović et al., 2017). However, in these known cases, the involvement of the cytoskeleton still requires the PM-localized protein regulators discussed in later sections, and we therefore do not discuss the role of these cytosolic structures here. We first briefly review the plant secretory system and then discuss the modifications and factors that allow polar targeting. For a detailed overview of the plant secretory system, we refer readers to other reviews (Aniento et al., 2022; Singh and Jürgens, 2018).

The life of a transmembrane protein in plants (Fig. 1) begins from its translation and transport to the endoplasmic reticulum (ER), which can occur either in a signal recognition particle (SRP)-dependent co-translational manner or through post-translational control (for example, the guided entry of tail-anchored proteins pathway; Rao et al., 2016). From the ER, transmembrane proteins are trafficked through the Golgi and trans-Golgi network (TGN) and sorted to the PM. Following their targeting to the PM, proteins can either be removed and sent back to the endomembrane system or directed to the vacuole for degradation (Kasai et al., 2011; Muntz, 2007). Endocytic removal from the PM occurs through clathrin-dependent or -independent pathways (reviewed in Rodriguez-Furlan et al., 2019). Clathrin-coated vesicles require adaptor complexes that bind PM phospholipids, clathrin and the cargo. The adaptor complexes in plants include both widely conserved complexes (such as AP-2; Yamaoka et al., 2013) and plant-specific ones (such as TPLATE; Gadeyne et al., 2014). During membrane targeting or recycling, the initiation of vesicles at the Golgi is controlled by ADP-ribosylation factor (ARF) GTPases that recruit coat proteins and are activated by ARF guanine-nucleotide-exchange factors (GEFs). The ARF-GEFs in plants – in contrast to those in other eukaryotes – consist of only two groups: in Arabidopsis, one of these groups is represented by GNOM (GN), GNOM-LIKE1 (GNL1) and GLN2, and the other by BIG proteins (BIG1–BIG5; reviewed in Pipaliya et al., 2019; Singh and Jürgens, 2018).

One of the most studied trafficking regulators in the context of polar targeting is the ARF-GEF GN. This protein was first discovered based on its embryonic mutant phenotype (Mayer et al., 1991). In gnom embryos, the polarity and, consequentially, the function of the auxin transporter PIN1 are disrupted, suggesting a role for GN in PIN protein membrane targeting or recycling (Steinmann et al., 1999). GN is thought to regulate only basal (root apex-facing) membrane targeting of PIN1 and PIN2 in root cells (Kleine-Vehn et al., 2008). Indeed, treatment of plants with the ARF-GEF inhibitor brefeldin A (BFA) causes PIN proteins to accumulate in large endomembrane compartments dubbed BFA bodies (Geldner et al., 2001; Morris and Robinson, 1998; Paponov et al., 2020). GN is responsible for the sensitivity of PIN proteins to BFA, as indicated by the finding that an engineered BFA-insensitive GN mutant version rescues the BFA sensitivity of basal-localized PIN proteins (Geldner et al., 2003; Kleine-Vehn et al., 2008). The strongest effect observed in the gnom mutant is on PIN1 localization, whereas the impact on PIN2, PIN3 and PIN7 localization is less pronounced (Guo et al., 2014), suggesting that GN is not the only trafficking factor for PIN proteins.

In Arabidopsis, GNL1 and GNL2 appear to have different functions. GNL2 is expressed only in pollen and is essential for its germination (Jia et al., 2009; Richter et al., 2007). Like GN, GNL1 is ubiquitously expressed (Richter et al., 2007) but only localizes to endomembranes, whereas GN is also observed in the cytosol and at the PM (Adamowski et al., 2024). Even though GNL1 has been shown to be involved in PIN polarity together with GN (Doyle et al., 2015), they are not completely redundant and are likely involved in different trafficking pathways (Richter et al., 2007). Indeed, analysis of IRK and KOIN receptor-like kinases suggests that GN and GNL1 might regulate opposing polar PM domains in endodermal cells (discussed in detail below) (Rodriguez-Furlan et al., 2023).

In addition to regulating the localization of PIN proteins, GN and GNL1 are involved in the polar trafficking and localization of other transmembrane proteins as well as soluble membrane-associated proteins in different tissues, such as BOR1, IRK and KOIN in roots (Yoshinari et al., 2021; Rodriguez-Furlan et al., 2023), and BASL (BREAKING OF ASYMMETRY IN THE STOMATAL LINEAGE) in stomata (Wang et al., 2022). For some of these examples, GN likely acts together with members of the PRAF (prenylated RAB acceptor-1 domain family)/RLD (RCC1-like) family that localize to the PM and endosomes in a manner similar to GN (Furutani et al., 2020; Wang et al., 2022). PRAF/RLD proteins are essential for the polarity of BASL and PIN3 in stomata (Wang et al., 2022). In lateral roots, they link the polarity of the cytosolic protein LAZY1 to the polarity of PIN3 at the PM (Furutani et al., 2020). Furthermore, GN is also involved in the trafficking of non-polar targets such as BRI1 (BRASSINOSTEROID INSENSITIVE 1) (Geldner et al., 2007; Irani et al., 2012; Xue et al., 2019). Finally, several polar proteins like BRX (BREVIS RADIX), PAX (PROTEIN KINASE ASSOCIATED WITH BRX) (Marhava et al., 2018) and NIP5;1 (Takano et al., 2010; Wang et al., 2017) have been shown to be BFA-sensitive in roots, but the direct connection to GN has not yet been investigated. Whether these BFA and GN effects are direct or mediated by changes in auxin distribution through the change of PIN localization is still an open question.

In bryophytes (liverworts, mosses and hornworts), PIN proteins seem to have different polarization mechanisms (Tang et al., 2023), even though overexpression of Marchantia (liverwort) PIN can partially rescue the Arabidopsis pin1 mutant (Fisher et al., 2023). This might mean that restriction to the polar PM domain is not always required for PIN function. In Physcomitrella patens (moss), trafficking of PINA involves BFA-insensitive components (Yáñez-Domínguez et al., 2023). This could also reflect a difference in GN function: the GN family might have diverged only in angiosperms, whereas GN in other plant species might have a different or more generic function, perhaps reflecting the ancestral role. It appears that the relationship between PIN polarity and GN trafficking has evolved recently, potentially indicating a more recent co-option of GN in polar trafficking pathways in addition to its general and ancestral functions in cellular trafficking.

There are other trafficking and recycling factors involved in polar protein localization. The ARF-GEFs BIG1–BIG5 take part in the trafficking of PIN proteins (Singh and Jürgens, 2018; Zhang et al., 2020), either together with GN or independently, depending on the cellular context. The adaptor complex AP-2 is involved in the trafficking of BOR1 (Yoshinari et al., 2019), NIP5;1 (Wang et al., 2017) and PIN proteins (Fan et al., 2013), whereas the AP-1 and AP-3 complexes recognize PIN1 (Sancho-Andrés et al., 2016). Dynamin-related protein DRP1A, which helps vesicle budding together with its partner ARF GTPase-activating protein (GAP) VAN3 (also known as AGD3) (Naramoto and Kyozuka, 2018; Sawa et al., 2005), is essential for the localization of BOR1 (Yoshinari et al., 2016) and for proper positioning of PIN1 and PIN2 after cell division (Mravec et al., 2011). DRP1A acts to restrict BOR1 polar localization in the membrane (Yoshinari et al., 2016), which is an important distinction from general polar trafficking. In this example, recycling probably creates the bias of BOR1 localization towards certain membrane regions. PIN protein polarity and trafficking are also affected by flippases, which can change the phospholipid content of bilayers and thus influence recycling (Zhang et al., 2020). It has been shown that PIN proteins form clusters on the PM in a phosphoinositide-dependent manner (Li et al., 2021), suggesting their preference for a certain phospholipid content, which might guide vesicle targeting. The small RAB GTPase RABA2a contributes to the recycling and secretion of non-basal (both non-polar and polar) PM proteins via its interaction with two soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) proteins, VAMP721 and SYP121, that induce membrane tethering (Li et al., 2017; Pang et al., 2022). However, most of these components are likely not specific to polarity because there is clear evidence for non-polar cargos of these pathways (Di Rubbo et al., 2013; Ekanayake et al., 2021; Kitakura et al., 2017; Mishev et al., 2018; Wang et al., 2023a; Xue et al., 2019).

In addition to recycling, direct exocytosis can create an asymmetry in membrane protein composition. The plant small GTPase RABA5c regulates trafficking directed to the cell edges in lateral root cells (Kirchhelle et al., 2016). Among the cargos of RABA5c vesicles are receptor-like proteins RLP4 and RLP4-L1, which play essential roles in cell wall sensing and integrity at the cell edges. RABA5c localizes at the cell edges, forming a negative feedback loop for the trafficking machinery (Elliott et al., 2024). The role of exocytosis has also been shown in the case of CASP proteins (Barbosa et al., 2023; Kalmbach et al., 2017). The EXO70A1 exocyst complex subunit marks the Casparian strip (the band-shaped cell wall thickening consisting of lignin and suberin) region on the membrane of endodermal cells before the appearance of the CASP domain and guides CASP protein exocytosis. Once at the PM, CASP proteins repel EXO70A1 and thus allow the dynamic formation of the Casparian strip (Barbosa et al., 2023) (Fig. 1). This mechanism is specific to only one out of 23 EXO70 subunits in Arabidopsis, and probably requires specific RABA GTPases on the CASP protein-carrying vesicles that act as activators of exocytosis. This suggests the existence of different exocytosis routes for different cargos using different exocyst complex components in plant cells. Moreover, the RABA2a–SNARE complex can act independently of the exocyst complex in the secretion of non-basal proteins, revealing additional independent secretion pathways (Pang et al., 2022).

Evidence for the importance of seemingly general factors, including the abovementioned adaptor complexes, ARF GTPases and dynamin, in the trafficking of polar proteins is also abundant in animal models (Gravotta et al., 2019; Lin et al., 2015; Nakayama et al., 2009; Osmani et al., 2010; Shafaq-Zadah et al., 2012; Tower-Gilchrist et al., 2019; Wang et al., 2009). However, like in plants, it is often unclear how direct and at which trafficking step these factors control polar targeting. Thus, the question of how transmembrane polar proteins reach their polar PM domain is to a large degree unanswered. One notable exception is the exocyst complex, which is often tethered to specific PM domains to guide targeted exocytosis in both plant and animal models (reviewed in Polgar and Fogelgren, 2018). How exocyst components are themselves polarized is less clear, although direct interactions between exocyst components and specific lipids or polarity regulators such as Par3 (also known as PARD3) have been reported (Ahmed and MacAra, 2017; Synek et al., 2021).

In summary, transmembrane proteins have complex targeting mechanisms that can be controlled both at the stage of exocytosis and endocytic recycling to confine the polar domain. Some of these trafficking routes or their combinations might be specific for polar PM domains, but more knowledge of vesicle cargos and membrane protein content is needed to draw a full picture of the polar targeting of transmembrane proteins.

There are various examples of soluble polar proteins without transmembrane domains that peripherally associate with the PM. These proteins, referred to as cortical proteins, include BRX, BRX-LIKE (BRXL) proteins, BASL, OCTOPUS (OPS) and OCTOPUS-LIKE (OPL) proteins in stomatal and vascular development (Breda et al., 2017; Marhava et al., 2018; Muroyama et al., 2023; Wallner et al., 2023); SOSEKI (SOK) proteins in various tissues (van Dop et al., 2020; Yoshida et al., 2019); and SGN1 (also known as PBL15) in the root endodermis (Alassimone et al., 2016). Such cortical proteins use diverse targeting strategies that can often be dynamically regulated by cellular factors in response to developmental or environmental cues. These include direct interactions with phospholipids via specific lipid-binding domains or less specific basic hydrophobic protein motifs, which can be attenuated by phosphorylation due to steric hindrance or introduction of a negative charge; hydrophobic post-translational modifications (PTMs) such as acylation, which allows proteins to be anchored in the membrane; or direct interactions with other polar proteins (Fig. 2).

The localization of soluble proteins to the cortex, regardless of targeting strategy, likely changes their function by altering their conformation, diffusion kinetics or molecular neighborhood. Whereas none of these mechanisms are exclusive to polar proteins, reversible anchoring to the PM is essential in all known polarized systems, where both association with and dissociation from the membrane can create asymmetries. In well-characterized yeast and animal polarity models, it is known that polarization dynamics and outcomes heavily depend on the ability of multiple interacting factors to reversibly associate with the membrane (reviewed in St Johnston, 2018). In contrast, membrane targeting mechanisms used by plant polar proteins remain mostly elusive. Below, we review the known targeting mechanisms exploited by plant cortical polar proteins.

Cortical proteins can interact directly with negatively charged phospholipids through hydrophobic and electrostatic interactions using specialized domains or charged basic hydrophobic (polybasic) motifs (reviewed in Nakamura, 2017) (Fig. 2A). The phospholipid content of the membrane is not equally distributed (Lebecq et al., 2023; Marhava et al., 2020; Mei et al., 2012; Zhou et al., 2020) and shows some asymmetry, which could support asymmetric protein distribution in the cell, although direct evidence for this in plants has not been reported. Phospholipids can also regulate membrane trafficking (Liu et al., 2023; Xing et al., 2021); biased lipid composition can create hotspots for endocytosis or exocytosis (Synek et al., 2021), and could thus affect polar proteins indirectly.

Several recent studies have shown the importance of regulating the direct interactions between polar proteins and lipids. For example, the AGC family of kinases, which includes the PIN regulators PID (PINOID), D6PK (D6 PROTEIN KINASE) and PAX, has been shown to interact with different phospholipids, including phosphatidic acid and phosphatidylinositol (4,5)-bisphosphate [PI(4,5)P₂] (Barbosa et al., 2016; Marhava et al., 2020; Simon et al., 2016; Wang et al., 2019). These interactions occur via lysine-rich polybasic motifs and influence PIN protein auxin transport activity and polarity. During root hair and trichome initiation, ARO (ARMADILLO REPEAT ONLY) proteins, which are usually cytosolic, are recruited to phosphatidic acid in an asymmetric manner, where they recruit GAPs for ROP (RHO OF PLANTS) small GTPases to the complex with ROP1, spatially restricting ROP1 signaling at the membrane (Kulich et al., 2020). Direct interactions of ROP proteins with anionic phospholipids allow the formation of membrane nanoclusters where the ROP proteins are enriched and their activity is regulated in a spatial manner (Aphaia et al., 2021 preprint; Platre et al., 2019; Smokvarska et al., 2021). Finally, a basic hydrophobic patch is also essential for the membrane association of LAZY3 and LAZY4 (collectively LAZY3/4), which regulate root bending in response to gravity (gravitropic response). However, other factors, such as protein–protein interactions, also contribute to the movement of LAZY3/4 proteins to the PM after a change in root growth direction (Nishimura et al., 2023).

Even though non-polar cortical proteins also use direct phospholipid interactions as a recruitment mechanism, the transient association between polar proteins, negatively charged phospholipids and phospholipid-modifying enzymes could create a local self-reinforcing asymmetry. For example, the direct interaction of the PIN regulator PAX with phosphatidylinositol 4-phosphate 5-kinase (PIP5K, herein referring to PIP5K1 and PIP5K2) creates a polar enrichment of PI(4,5)P₂, thus recruiting more BRX–PAX complex and maintaining low PIN1 levels (Marhava et al., 2020). Phospholipid-modifying enzymes, such as phospholipase D, are components of signaling pathways generating phosphatidic acid as a secondary messenger. Direct interaction between phosphatidic acid and PID activates PIN2 phosphorylation, thus modifying its activity and redistribution (Wang et al., 2019). Direct lipid binding and recruitment of lipid-modifying enzymes have been reported for several animal polarity proteins (Claret et al., 2014; Dong et al., 2020; Gervais et al., 2008; Krahn et al., 2010; Scholze et al., 2018), underscoring the importance of the interplay between lipids and proteins to generate positive feedback during polarization.

Understanding the contribution of lipid asymmetries to cell polarity has long been hindered by a lack of reliable methods to map the spatial distribution of membrane lipids. Some membrane lipid sensors are now available, such as sensors for negatively charged phospholipids that are based on the protein domains that bind them (Platre et al., 2018), and can be used in super-resolution imaging. However, sensor binding can also interfere with the phospholipid dynamics and cause side effects (Bayle et al., 2021). Characterizing the localization and dynamic behavior of lipid-modifying enzymes and lipid-binding domains will be important steps to tackle this issue and will help us understand whether lipid asymmetries generated during vesicular sorting are present and important.

In addition to interacting with lipids, cortical proteins can be recruited to the membrane by direct binding to other proteins (either polar or non-polar) (Fig. 2B). This is exemplified by numerous interactions in the well-described polarization of stomatal precursor cells. In this context, BRXL proteins polarize several hours before the asymmetric cell division of stomatal precursor cells and regulate the polarity of the daughter cells (Muroyama et al., 2023). BRXL proteins are essential for promoting the cortical polarization of BASL, which otherwise localizes to the nucleus and cytosol (Rowe et al., 2019 preprint). Notably, BASL also contributes to focusing BRXL protein localization, indicating a mutual dependency (Rowe et al., 2019 preprint), and together they are essential for the polar cortical recruitment of POLAR (Houbaert et al., 2018). BASL also recruits the mitogen-activated protein kinase kinase kinase (MAPKKK) YODA, and POLAR recruits the GSK family kinases BIN2 (also known as ASK7) and SK12 (also known as ASK3) (Houbaert et al., 2018; Zhang et al., 2015), which together ensure asymmetries in cell division orientation and fate specification. Novel methods that couple proximity labeling with mass spectrometry show promise in identifying novel interactions that help shape the composition of polar domains. Using this technique, new components of polar domains in stomata have been identified, including SOK3 (SOSEKI 3), AN (ANGUSTIFOLIA) and DLC1 (DYNEIN LIGHT CHAIN 1), which were previously known for playing a role in polarity in other tissues (Wallner et al., 2024). These studies bring us closer to describing general tissue context-independent regulators of polarity.

Examples outside stomata include the recruitment of AN by SOK1 proteins in roots, which likely depends on the local concentration of SOK1 (van Dop et al., 2020). A DIX domain in SOK1 proteins allows their polymerization and is essential for polarizing AN, in a similar way to how polymerization of the animal DIX-domain protein Dishevelled2 (Dvl2) has been shown to increase its avidity for the interactor Axin1 (Kan et al., 2020). In shoots, the BRX domain-containing BRXL4 and RLD family proteins are polarly recruited by LAZY1, whose localization is modulated by gravity to control shoot and root branching angle (Che et al., 2023; Furutani et al., 2020). In lateral roots, the regulation of branching angle is achieved by RLD proteins through the control of PIN1 and PIN3 localization to the direction of gravity (Furutani et al., 2020). In stem branches, BRXL4 causes delocalization of LAZY1 to the nucleus, where it suppresses its own transcription and thus inhibits anti-gravitropic growth (Che et al., 2023). These examples demonstrate the complexity and tissue specificity of relationships between polar proteins, making it difficult to define hierarchies and upstream regulators of polarity.

In addition to having a basic hydrophobic domain or using another protein as a membrane anchor, proteins can be subject to hydrophobic PTMs that regulate their recruitment to membranes. There are several hydrophobic modifications (reviewed in Chamberlain and Shipston, 2015; Shang et al., 2022), including the well-studied S-acylation (or specifically S-palmitoylation), N-myristoylation and C-terminal prenylation. These three modifications involve the enzymatic addition of long-chain fatty acid moieties – namely palmitic acid, myristic acid, or a geranylgeranyl or farnesyl residue, respectively – to specific amino acid residues, such as cysteines in the case of S-acylation or prenylation, and N-terminal glycines in N-myristoylation. Whereas S-acylation is a reversible modification, with removal of the modification catalyzed by thioesterases (reviewed in Li et al., 2022) (Fig. 2C), N-myristoylation of glycines is considered irreversible (Fig. 2D) (Traverso et al., 2008).

Several plant polar proteins have been shown or are predicted to be S-acylated, suggesting a prominent role for this reversible lipidation in their localization, similar to what has been described for some protein polarity mechanisms in animals (Aramsangtienchai et al., 2017; Chen et al., 2016; He et al., 2014; Matakatsu and Blair, 2008).

ROP GTPases, as well as their homologs in animals (Berzat et al., 2005; Navarro-Lérida et al., 2012; Wirth et al., 2022), exhibit two types of acylation: either stable prenylation (type I ROP GTPases) or S-palmitoylation (type II ROP GTPases) in their C terminus, and reversible S-palmitoylation of the conserved G domains (Sorek et al., 2011a). In the regulation of leaf epidermal cell shape, ROP6 (also known as ARAC3) is palmitoylated in response to auxin, leading to the formation of active ROP6 nanoclusters on the PM (Pan et al., 2020). The stable modification of the C-terminal end (either prenylation or S-palmitoylation, depending on the family member) seems less important for ROP protein function (Sorek et al., 2011b). Thus, dynamic membrane association seems more relevant than stable association for ROP protein polarity.

A similar requirement of reversible S-palmitoylation has been suggested for the endodermal cytosolic polar kinase SGN1. Mutation of two potential S-palmitoylation sites in its N terminus (Alassimone et al., 2016) or treatment with the inhibitor of lipid metabolism 2-BP abolishes SGN1 membrane localization. These S-acylation sites are essential for membrane association but not sufficient, indicating the existence of additional mechanisms or enzyme recognition motifs.

Dynamic S-palmitoylation also appears to be relevant for the polar membrane association of SOK proteins (van Dop et al., 2020; Yoshida et al., 2019). If a conserved S-acylation candidate CG motif is mutated in SOK1 or SOK5, the proteins no longer associate with the PM. Notably, adding a myristoylation motif to the N terminus of SGN1 or SOK1 leads to apolar distribution at the PM, showing that the reversibility of acylation might be required for the asymmetry (Fujita et al., 2020; Yoshida et al., 2019). The importance of polar distribution for the activity of these proteins is evident from the phenotypic change in cells expressing N-myristoylated SGN1 or SOK1. Normally, SGN1 regulates the position and cell wall lignification of the Casparian strip in endodermal cells, whereas N-myristoylated SGN1 causes ectopic lignin and suberin overaccumulation (Alassimone et al., 2016; Fujita et al., 2020). In the case of SOK1, overexpression normally leads to ectopic cell divisions in the root epidermis and cortex, whereas N-myristoylated SOK1 is unable to induce such a defect (Yoshida et al., 2019).

An S-palmitoylation site is also important for the membrane targeting of LAZY2 (NGR1) (Kulich et al., 2024). A cysteine mutation in LAZY2 leads to a significant increase in its cytosolic fraction and a loss of its ability to rescue the disrupted gravitropic response in the triple mutant ngr1/2/3. This mutation effect is only observed in combination with mutations in two polybasic regions, suggesting that multiple localization mechanisms enhance each other. In other members of the LAZY clade, such as LAZY1, this S-palmitoylation site is not conserved (Kulich et al., 2024); instead, LAZY1 carries a potential N-myristoylation site (Yoshihara and Spalding, 2020). The reversibility of LAZY2 S-acylation is likely less important for its function in the gravitropic response than its PM localization, since mutant versions with irreversible N-myristoylation or a C-terminal prenylation site localize to the PM and restore the gravitropic response in the ngr1/2/3 mutant background (Kulich et al., 2024).

BRX family proteins also have two potentially acylated cysteines in their C terminus (Koh et al., 2021; Rowe et al., 2019 preprint). In root primary phloem (protophloem) cells, localization of BRX is unaffected when the protein is fused to an N-myristoylation signal; however, the role of S-palmitoylation has not been investigated directly (Koh et al., 2021). In the same tissue, N-myristoylation also does not affect the localization or function of OPS, a polar protein that localizes in the opposite manner to BRX (Breda et al., 2017). Notably, a double mutation of the putative S-acylated cysteine residues in BRX expressed in stomata, where it can substitute its homologs BRXL2, BRXL3 and BRXL4, significantly perturbs its membrane association and biological function, even though fusion with an N-myristoylation signal does not disturb function (Rowe et al., 2019 preprint). Whether the same or different mechanisms act on the homologs in a different tissue context is not yet completely clear.

Although the different sensitivity of polar protein localization to the addition of an N-myristoylation modification is not well understood, it might depend on the strength and rates of membrane association of each protein. In support of this, BRXL proteins insensitive to N-myristoylation-dependent loss of polarity remain polarized in one of the daughter cells after cell division in stomata (Muroyama et al., 2023), whereas SOK proteins that are mislocalized by N-myristoylation have to polarize de novo after division in roots (Yoshida et al., 2019). However, these examples come from different tissues (stomata and root meristems) and might also reflect the tissue specificity of membrane association mechanisms.

Collectively, these observations indicate that S-acylation is likely an important mechanism for localizing polar proteins in plant cells. An attractive hypothesis would be that S-acylation directly instructs polarity via asymmetric activity or localization. However, the enzymes that catalyze S-acylation (S-palmitoyl acyltransferases, or PATs) are not well-studied in plants: their substrates and cellular localization have not been comprehensively described yet in vivo, apart from individual examples (Batistič, 2012; Liu et al., 2024; Wan et al., 2017). Several PATs are known to have developmental roles in processes that require polarity, such as gametophyte development (Li et al., 2019), root hair growth (Hemsley et al., 2005) and shoot branching (Guan et al., 2022). In mammalian epithelial cells, the PAT DHHC5 (also known as ZDHHC5) localizes to lateral membranes, forming cell–cell contacts where it finds its substrate ankyrin-G (also known as ANK3), which assures the stability of the lateral membrane (He et al., 2014). The same PAT S-palmitoylates the polarity regulator Cdc42 in the brain (Wirth et al., 2022). Identifying the PATs responsible for S-acylation of specific polar proteins and characterizing their distribution and activity is likely to reveal mechanistic insights into polar targeting in plant cells.

Despite their inherent differences in reaching the PM, some mechanisms that impact polar localization are shared between transmembrane and soluble proteins. PTMs such as phosphorylation or ubiquitylation can potentially influence the localization of proteins (Konstantinova et al., 2022; Offringa and Huang, 2013). Likewise, different types of polar proteins require maintenance of their polar membrane association and restriction of their lateral mobility and diffusion, which can be achieved by polymerization, interactions with other proteins, nanocluster formation, interactions with the cell wall or selective degradation. In some examples, such as the role of phospholipid-modifying enzymes in polarity (discussed below), cortical proteins play a role in maintaining the polar localization of transmembrane proteins, but their recruitment is influenced by the transmembrane polar proteins themselves (Fig. 3). This interplay between membrane and cortical polarity exemplifies the complex positive feedback mechanisms that amplify asymmetries and simultaneously hinder the clarification of cause–effect relationships.

Phosphorylation can induce profound changes in protein conformation, interactions, activity and localization (reviewed in Offringa and Huang, 2013). Kinases are central components in animal polarity pathways (Wu and Griffin, 2017), and growing evidence suggests that they also play an important role in the polarization of plant proteins. Phosphorylation regulates the recycling rate of the boron transporter NIP5;1, restricting its localization to the lateral outer domain of the PM (Wang et al., 2017). Although not formally demonstrated, a similar mechanism is likely to maintain the localization of BOR1 at the opposing inner domain (Takano et al., 2010). For the stomatal cortical polar protein BASL, phosphorylation reduces mobility and is essential for cortical membrane association (Zhang et al., 2015, 2016). The AGC kinase PAX is required for the PM association of BRX in root protophloem cells (Marhava et al., 2018) as well as its auxin-dependent membrane dissociation, which is driven by PAX-dependent phosphorylation (Koh et al., 2021). Amyloplast-to-PM movement of LAZY3/4 proteins in the gravitropic response of columella cells depends on phosphorylation by MKK5–MPK3, which increases LAZY3/4 affinity to the amyloplast coat proteins (Chen et al., 2023).

Eukaryotes have universal readers of phosphorylation – 14-3-3 proteins (Yaffe, 2002). In animals, different 14-3-3 proteins, including the PAR circuit component PAR-5, participate in the polarization of different cell types by altering the membrane association of polarity regulators in a phosphorylation-dependent manner (Benton and St Johnston, 2003; Hurd et al., 2003; Morton et al., 2002; Winter et al., 2012). Whereas no direct PAR-5 orthologs have been described in plants, there is evidence that 14-3-3 proteins control the localization of plant polar proteins. 14-3-3 proteins contribute to PIN protein polar trafficking (Keicher et al., 2017) and take part in phototropic responses that are dependent on NPH3 (NON-PHOTOTROPIC HYPOCOTYL 3, also known as RPT3) (Reuter et al., 2021; Sullivan et al., 2021). In this context, 14-3-3 proteins ‘read’ the phosphorylation status of NPH3, which is regulated by the kinase PHOT1 (PHOTOTROPIN 1) locally in a light-dependent manner, triggering its dissociation from the membrane and accumulation in cytoplasmic puncta (Sullivan et al., 2021). This 14-3-3-dependent relocalization is critical for NPH3 function, highlighting the importance of this mechanism in generating asymmetries that are shared between animals and plants. Even though not themselves polar, PHOT1 and NPH3 influence the polarity of PIN proteins. For other plant polar proteins, phosphorylation might be involved in controlling the interactions with other unknown regulators.

For PIN proteins, the role of phosphorylation has been extensively studied but remains incompletely understood. Like the phosphorylation of boron transporters, phosphorylation of PIN1 by CPK29 (CALCIUM-DEPENDENT PROTEIN KINASE 29) can be a mechanism to adjust recycling and restrict the polar domain, but in this case by regulating the PM-to-vesicle ratio (Lee et al., 2021). Phosphorylation by AGC kinases is proposed to define the direction of PIN protein polar targeting. For example, overexpression of the AGC kinase PID (Friml et al., 2004) or knockout of the corresponding PP2A phosphatase (Li et al., 2011) switches PIN1 polarity from basal to apical. PID and its homologs WAG1 (WAVY ROOT GROWTH 1) and WAG2 are only expressed in tissues where PIN proteins are apical, which supports their role in defining apical PIN protein polarity (Huang et al., 2010; Kleine-Vehn et al., 2009; Rahman et al., 2010). PIN3 phosphorylation by PID affects its dynamic polarization during the gravitropic response in roots (Grones et al., 2018). However, the role of phosphorylation in these processes has not been directly assessed. Intriguingly, antibody staining against phosphorylated PIN1 does not show polar distribution (Weller et al., 2017).

Of note, kinases can have phosphorylation-independent activities that impact the binding of polar proteins to other proteins or lipids, as is likely the case with interactions between PIN proteins and AGC kinases (Glanc et al., 2021) and between BRX and PAX (Marhava et al., 2018, 2020). To reconcile the phenotypes of AGC kinase mutants with the uniform distribution of phosphorylated PIN1, it has been suggested that instead of phosphorylation, the direct protein–protein interactions of PIN proteins with kinases are required for the restriction of their lateral diffusion in complex with members of the MAB4/MEL [MACCHI-BOU 4/MAB4(ENP1)-LIKE] protein family (also known as the NPY family) (Glanc et al., 2021). There is deep conservation of PIN phosphorylation sites across land plant species, despite differences in polarization mechanisms (Tang et al., 2023). This suggests that phosphorylation of PIN proteins serves functions beyond polar targeting and is supported by ample evidence for phosphorylation-dependent modulation of auxin transport by PIN proteins (Bassukas et al., 2022; Weller et al., 2017; Zourelidou et al., 2014).

Computational models of auxin (Mironova et al., 2012; Wabnik et al., 2010) and boron (Sotta et al., 2017) transport have been generated and explored, and the findings of these studies support the hypothesis that transporter polarity is essential for creating intercellular molecular gradients. This does not necessarily mean that the transporter protein needs to have polar localization but does mean that, at least, the transporter activity must be polarized. In this context, the functions of PIN protein phosphorylation and interactions with kinases have been reconsidered, and the auxin transport activity of PIN proteins, rather than their polar targeting, has been proposed as the primary target of phosphoregulation (Weller et al., 2017). A broader function for phosphoregulation has also been proposed for the cortical polar protein OPS; disruption of OPS phosphorylation, but not its localization, has a severe phenotypic effect (Breda et al., 2017).

Overall, phosphorylation is an instrument that is used not only for the association of polar proteins with their targeting machineries, but also to direct their functions regardless of localization. Phosphorylation can add to polarity control by forming feedback mechanisms. The extent to which this happens in different contexts now needs to be further probed by employing phosphoproteomics and other proteomics techniques.

Subject to the laws of thermodynamics, proteins will diffuse away from a zone of high concentration. Membrane association eliminates one spatial dimension and thus dramatically limits the range of diffusion. Nonetheless, the membrane is a fluid, and membrane-associated proteins will diffuse laterally. Thus, the maintenance of polar distribution requires mechanisms that restrict diffusion of the protein within the membrane (Fig. 3). Accordingly, polar proteins in plants show slower membrane diffusion than non-polar proteins (McKenna et al., 2019; Takano et al., 2010). One interpretation of the reduced diffusion of polar proteins is that they might preferentially reside in membrane domains with a different composition – subdomains of the PM enriched with certain proteins and phospholipids, or links to the cell wall. Indeed, PIN2 forms PM clusters under certain environmental conditions (Ke et al., 2021), and the PIN1 and ABCB19 auxin transporters colocalize in the PM in detergent-resistant domains (Titapiwatanakun et al., 2009).

Conceptually, maintenance of polar domains can be achieved in several ways: there can be global targeting and local removal or only local targeting and maintenance of the distribution. This maintenance can be due to membrane organization, such as in microdomains, or by other mechanisms that promote the stability of protein–membrane interactions as described above (phosphorylation, acylation or interaction with other proteins). These different strategies have been described in specific contexts in plants. For example, the LAZY3 protein is selectively degraded outside of its polar domain to maintain an asymmetric distribution (Nishimura et al., 2023), whereas polymerization of SOK proteins is essential for their polar localization by promoting local self-reinforcement (van Dop et al., 2020). ROP GTPase regulation in various tissue contexts can also illustrate different modes of polar maintenance because both ROP activators (ROP-GEFs) and deactivators (ROP-GAPs) can have confined polar localization (Bouatta et al., 2024 preprint; Lauster et al., 2022), which also influences ROP GTPase membrane dynamics (Sternberg et al., 2021).

The cortical and transmembrane protein polarity mechanisms might converge on the regulation of polarity maintenance since cytosolic proteins have been shown to restrict transmembrane protein mobility, with feedback mechanisms regulating cortical protein polarity. As discussed above, polar proteins can have a preference for a specific phospholipid content in the PM and can also recruit soluble phospholipid-modifying enzymes that shape membrane composition. For example, PIN polarity is maintained by phospholipid-modifying enzymes in a positive feedback loop between recruitment and PM modification by the cytosolic enzymes PDK1 (3′-PHOSPHOINOSITIDE-DEPENDENT PROTEIN KINASE 1) and D6PK (Tan et al., 2020). PDK1 reads lipid content and phosphorylates D6PK, which regulates PIN protein polarity or activity. PIN1 localization in the vasculature is regulated by the cortical BRX–PAX–PIP5K module, which provides phospholipid modifications (Marhava et al., 2020). As another example of cortical and transmembrane polarity convergence, MAB4/MEL proteins reduce the mobility of phosphorylated PIN2 while also recruiting the AGC kinases that phosphorylate PIN2 (Glanc et al., 2021).

In plants, the cell wall can also act as a diffusion barrier. The cell wall can influence cell polarity, division orientation and cell fate specification, as has been reported in plants (He et al., 2007) and other kingdoms with different types of cell walls (Novotny and Forman, 1974; Roemer et al., 1996). Several studies demonstrate the influence of the plant cell wall on the PM localization of polar proteins. For instance, cell wall integrity is essential for the maintenance of PIN protein polarity; in the background of a cellulose synthase mutant repp3 and after chemical inhibition of cellulose synthesis, PIN1 basal polarity is affected (Feraru et al., 2011). The cell wall also restricts the size of membrane nanoclusters (McKenna et al., 2019), indicating that cell wall composition can affect the distribution of PM components. TMK1 also affects cell wall properties, suggesting a potential mechanism for auxin influence on polarity stability (Lin et al., 2021). At the cell edges, the extracellular domains of the receptor-like proteins RLP4 and RLP4-L1 interact with the cell wall. These proteins might participate in sensing cell wall mechanical properties and potentially contribute to the reorganization of polar membrane trafficking in response to changes in cell geometry and mechanics (Elliott et al., 2024). CASP proteins form membrane clusters that can be used to restrict the mobility and stimulate exocytosis of cell wall-modifying enzymes and thus mediate asymmetry in the cell wall content (Barbosa et al., 2023). For SOK proteins, the cell wall can also be a factor that restricts lateral diffusion or helps with membrane targeting (Yoshida et al., 2019). The association of SOK proteins with the PM is susceptible to osmotic shock and cell wall digestion, showing the importance of cell wall composition, cell geometry and mechanical forces for the maintenance of their polarity. The role of mechanical forces in polarity has been reviewed in greater depth previously (Gorelova et al., 2021).

Another mechanism for allowing the stabilization of polar PM domains can emerge from interactions between opposing membrane domains and the proteins that reside within them. This strategy is commonly employed in animal cells, where the segregation of proteins to opposing membrane domains and maintenance of this asymmetry depends on self-reinforcement interactions within the same domain and mutual exclusion between opposing domains (Lang and Munro, 2017). In the monocot model Brachypodium, the proteins PAN1 (PANGLOSS 1) and POLAR localize to complementary domains of the precusors of stomata helper cells (Zhang et al., 2022). In the pan1 mutant, POLAR loses its polarity whilst still being present on the PM. However, PAN1 polarity is independent of POLAR: it polarizes earlier and is expressed in all protodermal cells, whereas POLAR is only expressed in the stomatal lineage precursors.

Complementary polar membrane domains are known in different Arabidopsis tissues, but it is still unclear whether their establishment is interdependent. In protophloem cells, the regulators OPS and BRX have opposing polarity but act genetically in parallel: OPS can rescue the brx mutant phenotype in roots, whereas BRX cannot rescue the ops mutant (Breda et al., 2017; Breda et al., 2019). A BRX–OPS fusion has a BRX-like rootward localization in both the brx background and the ops background, showing that the existence of the opposing OPS polar membrane domain is not necessary for the rootward BRX domain. This fusion can rescue the phenotype of both mutants, showing that OPS polarity is not functionally relevant (Breda et al., 2017). In stomata, OPL family proteins, which are homologs of OPS, form a domain opposite that formed by BRXL family proteins and BASL (Wallner et al., 2023). A BRXL2–OPL2 fusion protein shows mixed localization in stomata, which can be expected if the proteins mutually exclude each other. However, OPL polarity also does not depend on the presence of the opposing BRXL or BASL domain. Moreover, BRXL and OPL proteins are inherited by different daughter cells after asymmetric cell division and thus they do not rescue each other's mutants, showing the independent establishment and function of the opposing domains (Wallner et al., 2023).

In root cells, the inner–outer axis has opposing polar domains that are marked by several transmembrane proteins: BOR1 and NIP5;1 (Takano et al., 2010), and IRK and KOIN (Rodriguez-Furlan et al., 2022, 2023). BOR1 and NIP5;1 show opposing polarity even when expressed ectopically from early developmental stages (Liao and Weijers, 2018); however, it is unclear whether the establishment and maintenance of polarity of these two proteins is interdependent. IRK and KOIN show opposite domains only in the endodermis, whereas in other tissues they overlap. In newly divided cells, they both relocalize in a non-overlapping manner to the new membranes, which likely reflects different trafficking pathways contributing to membrane biogenesis. KOIN requires its kinase domain for its polarity, whereas IRK requires its extracellular domain, further supporting independent polarization mechanisms (Rodriguez-Furlan et al., 2022). Overall, we do not yet have general examples of opposing inner–outer domains in all cells.

To date, explicit examples of mutual exclusion have not yet been found in plants. However, we could still be missing the core components, and the proteins discussed here might be the dispensable ‘clients’ of polarity systems. Genetic redundancy and early mutant lethality may make it difficult to identify true regulators by genetic means. Nonetheless, the further study of currently known players will no doubt help uncover the principles of polarization.

Here, we have reviewed the modes and mechanisms underlying polar protein targeting in plants. Different mechanisms guide the PM localization of transmembrane and cytosolic polar proteins. Transmembrane proteins use different modifications of the general vesicle trafficking machinery, although there is not enough evidence yet to substantiate the existence of directed polar trafficking routes. Cytosolic proteins use various ways of direct or indirect phospholipid binding, achieving asymmetry through dynamic association and dissociation from the PM, interactions with other (polar) proteins or extracellular cues. Many open questions remain about how these general mechanisms align with cell asymmetry. Below, we highlight a few questions that we consider urgent priorities to be addressed (Fig. 4).

No specific ‘polar’ targeting mechanism has yet been described in plants (Fig. 4A,B). While we are still unable to identify the general set of factors defining polar coordinates, for every polar protein there is a specific set of polar targeting and localization mechanisms. Many polar proteins remain bound to the membrane and polar even when misexpressed in non-native tissues (Chan et al., 2020; Wang et al., 2022; Yoshida et al., 2019). This can be explained by either the ubiquitous expression of their specific regulators or the existence of a universal polarity system that all these proteins can read.

To date, there are only a few known mutations that result in the complete diffusion of polar proteins over the membrane in a non-polar way. Rather, known mutations or treatments disrupt membrane association, thus precluding analysis of roles in polarity. This underlines the importance of interactions with the membrane in the formation of cell asymmetries and the robustness of polarity regulation. In some cases, attempts to enforce non-polar membrane localization by adding an N-myristoylation signal do not affect polarity, showing either the robustness of polar membrane targeting to this modification or the existence of other protein concentration and diffusion-restriction mechanisms.

It is unclear how opposing polarity membrane domains are organized (Fig. 4C). There is evidence that successful formation of a polar domain can occur in the absence of a complementary domain. There is also evidence that polarity is not always essential for protein function, and proteins with opposing polarity can substitute for the functions of their opponents. This might mean that either we do not know the core regulators yet or, perhaps, no universal system exists.

Whether we are dealing with the establishment of polarity or its maintenance remains ambiguous in every differentiated tissue context. In many cases, we do not know the hierarchical order in which polar proteins are recruited to the PM and become polarized. This is partly due to genetic redundancy or mutant lethality, which prevent the assessment of downstream component localization in the absence of certain regulators. Comparing polar protein behavior during cell division or other processes with time resolution (Fig. 4D,E) can help unveil this order in cases where polarity is established de novo. For example, polarity can be directly visualized for the SOK proteins in dividing cells (Yoshida et al., 2019) and for the gravitropic response of LAZY1 (Nishimura et al., 2023). A colocalization study of PIP5K, PAX, BRX and PIN1 with a cell plate marker during the protophloem cell division has revealed that BRX arrives at the newly formed PM after PIP5K but before PIN1 (Wang et al., 2023b). Similarly, a quantitative assessment of the polarity levels of BRXL2 and BASL in the stomatal lineage has shown that BRXL2 polarizes at an earlier developmental stage than BASL (Gong et al., 2021). Understanding the influence of neighboring cells and external cues on the establishment of polarity in early embryogenesis or the de novo establishment of polarity after cell division can highlight the components that respond first and subsequently transfer signals to downstream polarity effectors. This approach has already been useful in the study of the response of neighboring cells to cell ablation, which has identified the role of RLP4 and RLP4-L1 in cell wall integrity sensing and cell edge maintenance (Elliot et al., 2024). Knowing the temporal dynamics of polarity events can assist in identifying regulators and clients of polarity and help to build a model of polarity throughout development.

We are grateful to Samuel W. H. Koh for critical reading of the manuscript.