# Source paper archive

- Source title: Minimal requirements and evolution of polar membrane targeting in the land plant SOSEKI protein family
- Author(s): Andriy Volkov, Evgeniya Pukhovaya, Catherine Albrecht, Dolf Weijers
- Source landing URL: https://www.biorxiv.org/content/10.64898/2026.10.30.762996v1
- PDF URL: https://www.biorxiv.org/content/10.64898/2026.10.30.762996v1.full.pdf
- PDF SHA-256: c0b9acdf1f284d251427488782829b37a209da27450c86c6b364865f0602ce59
- Converter: pymupdf4llm 0.0.27
- Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version.

---

&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


1 **Minimal requirements and evolution of polar membrane targeting in the**


2 **land plant SOSEKI protein family**


3


4 Andriy Volkov &#91;1&#93;, Evgeniya Pukhovaya &#91;1,2&#93;, Catherine Albrecht &#91;1&#93;, and Dolf Weijers &#91;1,&#93; 

5


6 1, Laboratory of Biochemistry, Wageningen University, Stippeneng 4, 6708WE, Wageningen,

7 the Netherlands


8 2, Present address: Translational Plant Biology, Utrecht University, Padualaan 8, 3584CH

9 Utrecht, the Netherlands


10 - Corresponding author: &#91;dolf.weijers@wur.nl&#93;(mailto:dolf.weijers@wur.nl)


11


12 **Summary**


13 Accurate spatial distribution of polar proteins is essential for cell function, but mechanisms

14 underlying polar protein targeting in plants are poorly understood. Here, we dissect the

15 molecular mechanisms underlying the polar targeting of the SOSEKI1 (SOK1) protein in

16 _Arabidopsis thaliana_ . Using extensive mutagenesis and truncation analyses of the polarity

17 domain, we mapped the minimal polarity fragment necessary and sufficient for stable polar

18 targeting to a 25-amino-acid region. This region likely forms an amphipathic helix, and

19 mutations reveal that both its polar and apolar faces are essential for membrane association and

20 polarity. Using phylogenetic analysis, we reconstruct the evolutionary history of the minimal

21 polar targeting region across land plants. Using cross-species expression and protein

22 localization, we infer the ancestral polar SOK localization pattern and show how it remained

23 remarkably conserved, only to diversify upon gene duplication. This work identifies a polar

24 targeting peptide in plants, reveals a mechanism for polar membrane targeting of a plant protein

25 family, and demonstrates both conservation and diversification within this family.


26 **Running title**


27 SOSEKI polar targeting


28 **Keywords**


29 Plant cell polarity; SOSEKI; polar protein targeting; evolution


30 **Summary statement**


31 Mapping a minimal polar targeting region in SOSEKI reveals a key amphipathic helix.

32 Phylogenetic analysis illuminates an evolutionary path of conservation and diversification of

33 polar targeting.


34


1


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


35 **Introduction**


36 Cell polarity, the asymmetric distribution of cellular components, is indispensable for
37 organisms. It is crucial for oriented cell division, differentiation, and the development of
38 complex body plans (St. Johnston, 2018). While mechanisms underlying cell polarization and
39 polar protein localization in animals and fungi have been studied in detail (Chiou et al., 2017)
40 (Butler and Wallingford, 2017), much less is known about plant cell polarity (Muroyama and
41 Bergmann, 2019). Plants evolved their multicellularity independently of animals and fungi
42 (Coelho and Weijers, 2026), and animal and fungal cell polarity proteins and regulators are not
43 conserved (Muroyama and Bergmann, 2019)(Ramalho et al., 2021). However, while
44 polarization mechanisms are largely unknown in plants, there is a range of proteins that have
45 been shown to localize polarly within cells. These include integral membrane proteins such as
46 PIN-FORMED auxin transporters(Vanneste and Friml, 2009), and Boron transporters NIP5
47 and BOR1 (Takano et al., 2010), as well as peripheral membrane proteins such as BASL (Dong
48 et al., 2009), BRX (Breda et al., 2017) and SOSEKI (SOK)(Yoshida et al., 2019)(van Dop et
49 al., 2020). Most polar proteins in plants have close paralogs, and in several cases, different
50 members in each family display different polar localization patterns. This has been best
51 documented for PIN proteins, where the polar localization appears to be encoded by an
52 intracellular domain (Wiśniewska et al., 2006), but the underlying mechanism and the
53 evolutionary path to diversification is unknown.


54 SOK proteins have properties that make them especially suitable for studying polar
55 targeting mechanisms. SOK proteins are found across land plants, and in all studied species,
56 are polarly localized (van Dop et al., 2020; Yoshida et al., 2019). Importantly, the 5 Arabidopsis
57 SOK proteins each show different polar patterns ( **Fig. 1A-E** )(Yoshida et al., 2019), suggesting
58 both conservation of polar targeting in land plants, and specialization of the precise polar
59 domains. SOK proteins tend to localize to the edge between two specific faces rather than to a
60 defined face of the cell (Yoshida et al., 2019). For example, in _Arabidopsis thaliana_ root cortex
61 cells, SOK1 localizes only to the apical inner (center-facing) edge, while SOK2 localizes to the
62 basal inner edge. These localization patterns suggest that SOKs integrate cues from multiple
63 body axes (apical-basal and center-periphery axes) for their localization.


64 SOK proteins are peripheral membrane proteins with a conserved domain topology (van
65 Dop et al., 2020). Their defining characteristic is an N-terminal DIX-like domain. This domain
66 is exclusively found in SOK proteins and is highly conserved. The plant DIX domain is
67 structurally and functionally similar to animal DIX domains. Like its animal homologs, plant
68 DIX domains form head-to-tail polymers, binding SOK proteins into higher-order complexes.
69 Following the DIX domain is a long region with no known domains, although there are some
70 conserved sequence motifs. In the middle of the protein is a highly conserved cysteine. Finally,
71 there is a conserved DAxTQT motif just following the DIX domain, as well as an N-terminal
72 Zn finger motif (in some other SOKs, there is a KEY motif instead of the Zn finger).


73 Previous experiments revealed regions within the SOK1 protein that are critical for
74 polar targeting of SOK1. Firstly, the DIX domain is critical for focusing the protein on an edge.
75 Removing the domain (or interfering with its polymerization) causes the protein to spread out


2


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


76 over a larger area of the membrane, albeit retaining membrane localization and polarity (van
77 Dop et al., 2020). Secondly, the conserved cysteine is essential for membrane binding (van Dop
78 et al., 2020; Yoshida et al., 2019). We recently found that this cysteine serves as an acceptor for
79 post-translational S-acylation, through the activity of polar PAT enzymes (Pukhovaya et al.,
80 2026). As a result of this modification, an acyl chain is attached to the cysteine, which anchors
81 the SOK1 protein to the plasma membrane. In experiments, mutating this residue abolished the
82 polarity and membrane localization of SOK1, rendering the protein cytosolic. Lastly, protein
83 truncations and domain swaps identified a fragment of about 80 amino acids surrounding the
84 conserved cysteine to be critical for polarization (Yoshida et al., 2019). Loss of this ‘polarity
85 domain’ (PD) from SOK1 displaces the protein to the cytosol. Moreover, swapping the polarity
86 domain of SOK2 with the equivalent region from SOK1 results in a protein mimicking SOK1
87 localization, affirming the pivotal role of the polarity domain in selecting the appropriate edge.
88 In summary, previous data suggests that at a minimum, three components are needed to polarize
89 SOKs: DIX-mediated polymerization, the activity of the polarity domain, and S-acylation of a
90 cysteine within the polarity domain.


91 In this study, we explore the minimal mechanisms guiding polar SOK localization to the
92 membrane. We identify a minimal region within the Arabidopsis SOK1 PD that encodes an
93 amphipathic helix, whose amphipathicity is critical for membrane targeting. We next explore
94 the evolutionary trajectory that this PD followed in the SOK family, and show that most SOK
95 clades retain an ancestral polar localization pattern. This ancestral pattern diversified only after
96 gene duplications occurred in flowering plants. Our work therefore identifies a minimal
97 principle for polar membrane association of SOK proteins and reveals that an ancient, universal
98 polarization system controls the localization of SOK proteins in all plants.


99


100


3


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


101 **Results**


102 _**Oligomerization of the polarity domain is required for focused edge localization**_


103 Previous results demonstrated that Arabidopsis SOK1 and SOK5 proteins rely on their DIX
104 domain for focused localization to the edge, and that the native SOK1 DIX domain could be

105 substituted by a human Dvl2 DIX domain (van Dop et al., 2020) . Still, it remains unclear
106 whether only the oligomerization conferred by the DIX domain causes this effect, or whether

107 the DIX domain additionally encodes information pertinent to polar targeting. We first
108 confirmed loss of edge-polarized SOK1 localization in full-length SOK1 ( **Fig. 1F,G** ) lacking

109 its DIX domain ( **Fig. 1H** ) or carrying oligomerization mutations ( **Fig. 1I** ). While these mutants
110 still localize polarly to the membrane, they were more diffuse and less concentrated on the edge

111 compared to the wild-type control. We next replaced the DIX domain of SOK1 with unrelated
112 oligomerization domains ( **Fig. 1J** ): SAM-SOK1 (carrying a SAM domain from the

113 _Arabidopsis thaliana_ LFY protein)(Sayou et al., 2016), and PB1-SOK1 (carrying a PB1
114 domain from the _Marchantia_ _polymorpha_ ARF1 protein)(Kato et al., 2020), each fused to a

115 fluorescent protein. Both oligomerization chimeras localized polarly ( **Fig. 1L,M** ) and exactly
116 as the hDvl2-DIX swap ( **Fig. 1K** ). This observation indicates that oligomerization of the

117 polarity domain is sufficient for edge localization. The DIX domain does not encode any
118 positional information and is only responsible for focusing the protein on the edge, but not for

119 polarity itself. Instead, this polarity must be contained in the polarity domain of SOK1.


120


4


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


121 **Figure 1: Oligomerization is a requirement for SOSEKI polar membrane localization**


122 ( **A** - **E** ) Localization of Arabidopsis SOK1 ( **A** ), SOK2 ( **B** ), SOK3 ( **C** ), SOK4 ( **D** ) and SOK5 ( **E** ), fused to
123 YFP, and expressed from the _RPS5A_ promoter, in Arabidopsis root tips. Coordinates are indicated in ( **A** ) and
124 interpretations of polar localization patter indicated beneath each root cross-section. ( **F** ) Cartoons of SOK1
125 protein variants expressed as RPS5A-driven YFP fusions in ( **G** - **I** ). ( **J** ) Cartoons of the fusions between
126 oligomerization domains and the SOK1 polarity domain (PD), expressed as RPS5A-driven mNeonGreen
127 fusions, shown in ( **K** - **M** ). ( **N** ) Cartoons of the tandem multimeric PD versions, expressed as RPS5A-driven
128 mNeonGreen fusions ( **O** - **Q** ). The number of transgenic lines with the pattern shown (n) is shown as a fraction
129 of the total (t) number of lines analyzed (n/t) in each lower right corner. Scale bar for panels ( **A** - **E** ) is 20 µm
130 and for ( **G** - **I**, **K** - **M** and **O** - **Q** ) 10 µm.


131 Oligomerization allows clustering of polarity domains, thus increasing the avidity of the SOK
132 complex for the membrane. To test if increasing avidity up to 6 times that of the monomeric
133 wild-type protein would allow for focused polar edge localization, we expressed (non134 oligomerizing) tandem fusions of the SOK1 polarity domain. Specifically, we expressed three
135 constructs, solely consisting of two (dimer), four (tetramer), or six (hexamer) tandem fused
136 copies of the SOK1 polarity domain (2xS1-PD, 4xS1-PD, 6xS1-PD; **Fig. 1N** ). All three
137 constructs localized polarly to the membrane but diffusely ( **Fig. 1O-Q** ), similar to the non138 oligomerizing SOK1-H29D/D78A mutant ( **Fig. 1I** ). This demonstrates that significantly more
139 than six units are needed for edge localization, and/or oligomerization serves to not only
140 increase the effective number of polarity domains but (also) contributes to the localization
141 pattern in another way.


142 _**Mapping of a minimal polarity domain**_


143 Having established that polar targeting of SOK1 is encoded in its polarity domain, we further

144 delineated the minimal requirements for polar targeting within this domain. Previously, the

145 boundaries of the polarity domain of SOK1 were based on a series of deletion mutants and a

146 domain swap between SOK1 and SOK2 (Yoshida et al., 2019). However, it remains unclear

147 whether the entire 80-amino-acid fragment is required for polar localization. We substituted

148 parts of the SOK1 polarity domain with flexible glycine-serine linkers in blocks of 10 amino

149 acids, yielding eight mutants covering the entire polarity domain except the essential CG motif

150 ( **Fig. 2A** ), and expressed each as a fusion to the hDvl2-DIX domain and a fluorescent protein

151 ( **Fig. 2B-J** ). We scored the localization as membrane-associated if at least some signal was

152 concentrated at the polar membrane domain, even if the cytosolic background signal was higher

153 than in the controls. Note that the use of the FastGreen marker for transformation, which drives

154 strong green fluorescence in seeds (Shimada et al., 2010), causes persistent accumulation of

155 large fluorescent particles in some roots that can not be fully spectrally separated from the

156 mNeonGreen used for protein imaging ( **Fig. 2,3,5** ). Hence, we disregard such large cytosolic

157 structures and focus only on membrane-associated and homogeneous cytosolic fluorescence.

158 Even if there was variation in these patterns among individual transgenics, the interpretation of

159 the localization pattern was supported by all lines tested. Mutants replacing the first 40 amino

160 acids hardly differed from the non-mutated control (scans 1-4; **Fig. 2C-F** ). However, scans 5

161 and 6 dramatically decreased membrane association and localized to the cytosol instead ( **Fig.**

162 **2G,H** ). Of the remaining mutants, scan 7 showed a mixed cytosolic and polar membrane

163 localization, with a weak membrane localization and a significant cytosolic fraction ( **Fig. 2I** ),


5


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


164


165 **Figure 2: Identification of a minimal region for SOK1 polar membrane targeting**


166 ( **A**, **K** ) Cartoon showing the substitution scans ( **A** ) and truncations ( **K** ) generated within the SOK1 PD,
167 expressed as _RPS5A_ -driven fusions to hDvl2-DIX and mNeonGreen and localization of the substitution
168 scans ( **B** - **J** ) localization and truncations ( **L** - **U** ) in Arabidopsis root tips. The number of transgenic lines with
169 the pattern shown (n) is shown as a fraction of the total (t) number of lines analyzed (n/t) in each lower right
170 corner. Scale bar for panels ( **B** - **J** and **L-U** ) is 10 µm.


171 while scan 8 localized polarly ( **Fig. 2J** ). Interestingly, all the proteins investigated in this

172 experiment were either membrane-bound and polar, or cytosolic, and we never found

173 membrane-localized but apolar mutants, or mutants whose polar domain was modified. In
174 summary, this experiment demonstrates that about 30 amino acid residues contained in scans

175 5-7 are necessary for polar edge localization, while the others are dispensable.


176 To investigate which region within the polarity domain is sufficient for polar targeting, we

177 generated N- and C-terminally truncated versions of the SOK1 polarity domain ( **Fig. 2K** ), and

178 likewise expressed each as a fusion to hDvl2-DIX and a fluorescent protein. Each fragment


6


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


179 again conferred either polar membrane localization, cytosolic localization, or a mix of both

180 ( **Fig. 2L-U** ). The minimal fragment required for consistent polarization is encoded by 25 amino

181 acids, namely residues 213-237 of the SOK1 protein (Fragment 8; **Fig. 2S** ). A C-terminally

182 extended Fragment 3 did however, appear to subtly enhance membrane localization relative to

183 Fragment 8 ( **Fig. 2N** ), suggesting that there is a quantitative contribution of the residues C
184 terminal to Fragment 8. Nonetheless, given that the 25 amino acids are sufficient for polar

185 targeting, we consider it the minimal polarity fragment (MPF).


186 _**A predicted amphipathic helix in the SOK1-MPF mediates polar targeting**_


187 Having identified the minimal SOK1 fragment required for stable polar targeting, we wondered

188 how such a short fragment confers membrane localization and polarity. We analyzed sequence

189 features and used prediction tools to explore the nature of this region, and used the extended

190 motif (Fragment 3; **Fig. 2K,N** ). Both JPred and AlphaFold predict an α-helix within this

191 fragment ( **Fig. 3A,B** ). We also noticed the abundance of charged amino acids ( **Fig. 3C** ): five

192 out of 30 residues are positively charged (R and K) and only two are negatively charged (D).

193 Closely examining the predicted AlphaFold structure of this polarity fragment showed that

194 most of the charged amino acids are on one face of the α-helix, while the other face is abundant

195 in hydrophobic amino acids, including two highly hydrophobic leucines ( **Fig. 3C,D** ). A helical

196 wheel projection showed clear amphipathicity of the helix through grouping of positive

197 residues on one face of the helix and hydrophobic residues on the other ( **Fig. 3E** ). Such

198 amphipathic α-helices (AHs) are well-known to be involved in the membrane binding of

199 various proteins (Lee et al., 2026) . These proteins typically rely on positive charges (conferred

200 by R and K residues) on one face of an AH for recruitment to the negatively charged

201 phospholipid head groups (Giménez-Andrés et al., 2018). The hydrophobic face of the helix

202 ensures association within the lipid bilayer through long-chain aliphatic residues (frequently

203 L). This configuration of amino acids confers high hydrophobic moments to the helix. In our

204 case, the hydrophobic moment is as high as 0.378 µH, which is typical for membrane
205 associated helices (Eisenberg et al., 1982). The molecular dynamics tool PMIpred predicts

206 strong binding of the AH to negatively charged membranes with a high free energy of

207 membrane binding ( **Fig. S1** ). These simulations of binding kinetics suggest that the SOK1 MPF

208 acts to bind membranes.


209 We next directly tested the role of amphipathicity within this predicted AH by targeted

210 mutations in the context of Fragment 1 ( **Fig. 2K,L** ; **Fig. 3F** ). Mutating the two leucines on the

211 hydrophobic face of the helix (L227A/L230A) completely abolished membrane association

212 ( **Fig. 3G** ). Similarly, to manipulate the hydrophilic face of the AH, we mutated various charged

213 and hydrophilic residues. Mutants with altered positive charges (K216A; R228A/N229A;

214 K232A; K216A/R228A/K232A) were similarly cytosolic and failed to localize to the

215 membrane ( **Fig. 3H-K** ). The K232A mutant showed an intermediate effect by localizing both

216 polarly to the membrane and to the cytosol ( **Fig. 3J** ), suggesting it is less critical than K216.

217 Other residues within the helix are also essential, as demonstrated by the cytosolic mutant

218 A223N/S224A/H225A ( **Fig. 3L** ), while mutations C-terminal of the cysteine (and therefore

219 outside of the amphipathic helix) did not affect localization (L236N/D237A; T238G;


7


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


220 N239A/D240A; **Fig. 3M-O** ). Together, these findings show that hydrophobicity and positive

221 charge on either face of the AH, respectively, are essential for SOK1 polar membrane

222 association.


223


224 **Figure 3: An amphipathic helix drives SOK1 polar membrane targeting**


225 ( **A** ) Secondary structure prediction (JPRED) for the SOK1 PD region, showing an alpha helix (orange)
226 within the CPF region (boxed). ( **B** ) AlphaFold2 prediction of the SOK1 structure, showing high-confidence
227 DIX domain structure and a low-confidence helix within the MPF region (both highlighted). ( **C** ) Detail of
228 the MPF region, showing the predicted helix with side chains, colored by physicochemical properties. ( **D** )
229 Alpha helix viewed from different angle, with its sequence underneath, and residues color-coded by
230 properties. ( **E** ) Helical wheel projection of the helix showing polar/basic and hydrophobic faces. Calculated
231 amphipathic helix properties are indicated. ( **F** ) Cartoon showing the mutations generated within the SOK1
232 PD, expressed as _RPS5A_ -driven fusions to hDvl2-DIX and mNeonGreen, and localization of mutated
233 fragments ( **G** - **O** ) in Arabidopsis root tips. The number of transgenic lines with the pattern shown (n) is shown
234 as a fraction of the total (t) number of lines analyzed (n/t) in each lower right corner. Scale bar for panels
235 ( **G** - **O** ) is 10 µm.


8


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


236 _**The minimal polar targeting region shares conserved features in SOSEKI proteins**_


237 The five Arabidopsis SOSEKI proteins differ in subcellular targeting (Yoshida et al., 2019)

238 ( **Fig. 1A-E** ), which indicates that the targeting mechanisms and sequences must have diverged.

239 Previously, we reconstructed the evolutionary history of the SOSEKI protein family (van Dop

240 et al., 2020) (Ramalho et al., 2021). However, the density of sampling in some clades was too

241 limited to infer patterns of evolution at the scale needed to interpret evolution of the polarity

242 domain. We therefore developed a more extensive phylogenetic tree ( **Fig. 4A,B** ) using the most

243 recent available genomic data. This phylogeny includes 1086 SOK sequences from 199 plant

244 species. Relative to previous studies (Ramalho et al., 2021)(van Dop et al., 2020), this

245 phylogeny includes a large number of angiosperm sequences, as well as additional species of

246 lycophytes and bryophytes. Additionally, by reconciling the SOK gene tree with a known

247 species tree, we resolved recent and ancient evolutionary events. Our data are consistent with

248 previous analyses (van Dop et al., 2020), but add detail about the origin of SOK orthologs. We

249 infer that a single ancestral SOK (AncSOK) in early embryophytes (land plants) founded the

250 single bryophyte SOK ortholog group, BryoSOK. These BryoSOKs form a sister clade to all

251 tracheophyte (vascular plant) SOKs, which independently evolved from AncSOK. The single

252 AncSOK underwent a series of duplications and speciations in the tracheophyte lineage. From

253 these events, several ortholog groups arose, of which the lycophyte-specific LycoSOK is the

254 earliest-diverging group. All euphyllophyte SOKs form a single EuSOK ortholog group. This

255 group diversified into several subgroups. At the base of euphyllophytes, the ancestral EuSOK

256 duplicated, forming the two major ortholog groups EuSOK-A (containing AtSOK1) and

257 EuSOK-B (containing AtSOK2-5). Both groups are broadly found across all euphyllophyte

258 lineages. We further detected a recent diversification of EuSOK-B genes into three subgroups.

259 A series of duplications at the base of spermatophytes (seed plants) and angiosperms (flowering

260 plants) formed three angiosperm-specific ortholog groups, namely EuSOK-B2 (including

261 AtSOK2), EuSOK-B3 (including AtSOK3), and EuSOK-B4 (including AtSOK4/5). All the

262 major angiosperm groups (including basal angiosperms, monocots, and dicots) retained SOKs

263 from each of these three groups, suggesting that each group is relevant for flowering plants.


264 Since different SOK groups, even the more recently diverged ones, have proteins with
265 different localization patterns, we aimed to identify group-specific differences in the polarity

266 domains of SOKs from the groups. We analyzed the regions in BryoSOKs, LycoSOKs,

267 EuSOK-As, and EuSOK-Bs that were in equivalent positions to the 80-amino-acid broader

268 polarity domain of SOK1. Subsequently, we estimated the evolutionary conservation of

269 residues within these domains using ConSurf. Within the polarity domain, we found a

270 conserved region common to all SOK proteins, including the acylated cysteine ( **Fig. 4C** ). We

271 refer to this region as the Conserved Polarity Fragment (CPF). Except for BryoSOKs, the

272 consensus sequence of this region is always VLxxLxxCG ( **Fig. 4D** ). ConSurf also predicted

273 functional importance for this CPF in all SOK groups, supporting its critical role in SOK

274 function. Detailed motif analysis revealed differences in the CPF sequence between the SOK

275 groups **(Fig. 4D** ). In LycoSOKs and EuSOK-Bs, the CPF is abundant with hydrophobic

276 residues (such as leucine, phenylalanine and methionine). In contrast, the CPF of EuSOK-As


277


9


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


278


279 **Figure 4: SOK phylogeny identifies conserved motifs within the polarity domain**


280 ( **A** ) Simplified phylogenetic tree for the SOSEKI protein family, identifying deeply conserved clades in
281 Bryophytes (BryoSOK), Lycophytes (LycoSOK), Euphyllophytes (EuSOK), subclasses within
282 Euphyllophytes (EuSOK-A, EuSOK-B) and within Angiosperms (EuSOK-B2, -B3 and -B4). Bootstrap
283 values are indicated at each node. ( **B** ) Abbreviated phylogenetic tree highlighting all relevant clades, and
284 species are color-coded by phylogenetic distribution (see legend at bottom). Arabidopsis (AtSOK) sequences
285 are marked, and sequences selected for functional analysis are marked with triangles. The scale bar shows
286 branch length relative to the substitution rate. ( **C** ) Alignment of the Polarity Domain in all 5 Arabidopsis
287 SOK proteins, as well as individual Selaginella and Physcomitrium shown (unaligned) SOK proteins (thus
288 reflecting all clades), highlighting conserved motifs and residues through ConSurf color coding. ( **D** )
289 Sequence logos for the identified conserved motifs in each SOK subclasss.


290


291 invariably contains a positively charged residue, such as arginine, which is likely a derived

292 form. Interestingly, this arginine is indispensable for polarization of AtSOK1 ( **Fig. 3I,K** ). The

293 CPF of BryoSOKs appears less conserved, but nonetheless often consists of hydrophobic and

294 positive residues. Unlike all other SOKs, the cysteine in BryoSOKs is not usually followed by

295 a glycine. In summary, detailed phylogenetic and motif analysis identified both conserved and

296 divergent features among SOSEKI protein groups.


10


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


297 _**Broad conservation of ancestral SOK polarity**_


298 Having identified a minimal polar targeting fragment in Arabidopsis SOK1, and having

299 mapped the evolution of this region, we next asked how its function evolved. We selected 13

300 SOK polarity domains from 12 species covering phylogenetic clades from BryoSOKs to

301 EuSOKs ( **Fig. 4B** ). Each was expressed as a fusion to hDvl2 DIX and mNeonGreen, and we

302 analyzed localization in Arabidopsis roots ( **Fig. 5** ). To test if the polar patterns of SOK proteins

303 are preserved within their PD region beyond AtSOK1, we also included the AtSOK3 PD in this

304 analysis. We found this PD to recapitulate the all-edges polar pattern of full-length AtSOK3

305 ( **Fig. 5** ). In the set of orthologous PD fragments, we found that, except for one of the fern

306 EuSOK-B sequences (CtSOKB1) and an angiosperm euSOK-B4-type sequence (LaSOKB4
307 1), which were both (mostly) cytosolic, all proteins polarly localized to the membrane of

308 epidermis and cortex cells. All but two localized to the inner membrane, similar to AtSOK1

309 ( **Fig. 1A** ). The Marchantia SOK domain (MpSOK), localized to the outer basal membrane

310 domain. The _Zea mays_ EuSOK-B3-type protein localized throughout the membrane and was

311 more concentrated at cell edges (ZmSOKB3-1), exactly like AtSOK3 ( **Fig. 1C** ). The

312 EuSOKB4-type sequence LaSOKB4-1showed some weak membrane association and behaved

313 similarly to AtSOK4, which is distinctly less polar than other SOK proteins in Arabidopsis

314 ( **Fig. 1A-E** )


315 In summary, this data shows that most tested SOK polarity domains confer the same

316 polar inner membrane localization in Arabidopsis roots, likely reflecting an ancestral character

317 state. While there may be variation in all clades beyond the sequences we experimentally tested,

318 we find that functional diversification within the polarity domain occurred within the euSOK
319 B clade, likely giving rise to the diverse SOK patterns observed in Arabidopsis.


320


321 _**Diversification of SOK polarity through the polarity domain**_


322 Our comparative analysis demonstrated that the ancestral SOK localization corresponds to the

323 inner lateral domain in Arabidopsis roots, and that later diversification in the EuSOK-B clade

324 generated novel polar localizations. However, MpSOK is a remarkable exception: It localizes

325 to the exact opposite domain when expressed in Arabidopsis roots ( **Fig. 5** ), compared to all
326 other non-EuSOK-B domains. This suggests that the MpSOK protein has undergone

327 modifications that changed its polar membrane targeting. We have recently shown that MpSOK

328 polar targeting depends on the conserved Cysteine in the polarity domain (Pukhovaya et al.,

329 2026), similar to AtSOK1 and AtSOK5 (van Dop et al., 2020). It is therefore an open question

330 whether the opposite localization of MpSOK in Arabidopsis roots reflects a different wiring of

331 the cellular polarity system, or just altered behavior when expressed in Arabidopsis. To address

332 this question, we performed the reciprocal experiment and expressed AtSOK1 in Marchantia

333 thallus, driven by the EF promoter. In this tissue, native MpSOK polarizes towards the

334 meristematic notch in a supra-cellular, tissue-wide pattern (van Dop et al., 2020) ( **Fig. 6A** ).

335 Strikingly, within the same cells, AtSOK1 polarized in the opposite pattern, away from the


11


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


336 meristematic notch ( **Fig. 6B** ). Thus, the opposite polarity between AtSOK1 and MpSOK is

337 conserved in the two species, suggesting that the entire cell polarity system that guides SOK

338 polarization is rewired in Marchantia.


339


340 **Figure 5: Evolution of polar targeting through the SOK polarity domain**


341 Schematic phylogeny of the SOK family in land plants, with the ancestral proteins (AncSOK, EuSOK)
342 indicated at the relevant nodes and extant SOK classes marked on branches. Microscopy images show the
343 fluorescence of chimaeric proteins carrying hDvl2-DIX fused to the PD domain of 12 land plant SOK
344 proteins (see legend Figure 4 for identity and abbreviations), fused to mNeonGreen and expressed from the
345 RPS5A promoter in Arabidopsis root tips. The number of transgenic lines with the pattern shown (n) is
346 shown as a fraction of the total (t) number of lines analyzed (n/t) in each lower right corner. Scale bar for all
347 microscopy panels is 10 µm.


348


12


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


349 **Discussion**


350 In this study, we investigated the mechanisms underlying the polar localization of cortical SOK
351 (SOK) proteins using the SOK1 protein from _Arabidopsis thaliana_ as a model. Through
352 extensive mutagenesis, we probed the contributions of electrostatic interactions, specific amino
353 acid sequences, and polymerization to the polar targeting of SOK1. We determined that polar
354 targeting of SOK1 requires two components: the minimal polarity fragment and DIX-mediated
355 polymerization. Our analysis of various mutants of the SOK1 polarity domain provided
356 significant mechanistic insights. Most strikingly, we identified the precise minimal fragment
357 required for polar localization of SOK1. This minimal SOK1 polarity motif, encoded by amino
358 acids 213-237, is required and sufficient to target the chimeric protein to the inner apical edge
359 of cortex cells when polymerized (e.g. via a DIX domain). To our knowledge, this is the first
360 polar targeting sequence described in plant cells. Within this minimal polarity fragment, a
361 combination of hydrophilic, positively charged, and hydrophobic residues polarly target the
362 protein to the membrane.


363 _In silico_ analysis of the minimal polarity fragment revealed that it encodes an
364 amphipathic helix (AH), one face of which is hydrophobic, while the other face is positively
365 charged. Such AHs are commonly found in membrane-binding proteins throughout the
366 domains of life, from viruses to eukaryotes (Giménez-Andrés et al., 2018). The hydrophobic
367 side of the helix is buried within membrane lipids, while the positively charged face interacts
368 with the negative charges of the head groups. These amphipathic properties are essential for
369 membrane association: removing positive residues in the polar face or removing hydrophobic
370 residues in the apolar face of the AH of the polarity motif disrupts membrane targeting. An
371 intriguing aspect of our findings is the potential interplay between the amphipathic helix and
372 the post-translational modification of the Cysteine at its terminus. Previous studies of SOK1
373 showed that this residue is essential for membrane association (van Dop et al., 2020). More
374 recently, we identified polarity-localized Protein Acyltransferase (PAT) enzymes that can target
375 SOK1 to the PM in heterologous systems (Pukhovaya et al., 2026). Our data support a model
376 where the AH is required for initial association of SOK1 to the PM, followed by PAT-dependent
377 acylation. An attractive hypothesis is that the AH directly interacts with the membrane, yet with
378 low affinity. This would bring SOK1 in close proximity to the PAT enzymes that subsequently
379 stably anchor SOK1 to the PM. The effective reduction in diffusion from 3D to 2D may
380 increase the probability of oligomerization, which lastly concentrates SOK1 protein in the cell
381 edge.


382 There is a precedent for this interplay between an AH and PAT activity: a similar
383 topology of an AH adjacent to a palmitoylated cysteine has been studied in several proteins.
384 For example, the intracellular loop of the human sodium-calcium exchanger protein NCX1
385 contains an AH that regulates palmitoylation of adjacent cysteines. Mutations disrupting the
386 AH also abolished palmitoylation. Moreover, fusion of this AH and the adjacent cysteine with
387 YFP resulted in palmitoylation and membrane targeting of the fusion protein, demonstrating
388 that this structure is a signal for S-acylation by DHHC-PAT enzymes (Plain et al., 2017). In a
389 similar study, Lin et al. (Lin et al., 2015) identified an AH within the _Legionella_ effector protein


13


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


390


391 **Figure 6: Divergence and evolution of SOK polar targeting**


392 ( **A**, **B** ) Localization of MpSOK ( **A** ) and AtSOK1 ( **B** ), each expressed as a fusion to mCitrine from the _EF_
393 promoter in Marchantia thallus, shows coordinated polarization towards (MpSOK) and away from
394 (AtSOK1) the meristematic notch (green dot). Insets show cells in equivalent position with the direction of
395 the notch marked by an arrow. ( **C** ) Reconstruction of the evolution of SOK polarity in land plants. The tree
396 is a combined protein sequence tree and species tree, and shows the trajectory of speciations (regular nodes)
397 and duplications (squares at nodes). Inferred ancestral SOK polarity (AncSOK), as well as extant polarity
398 patterns (right) are indicated. Scale bar in ( **A**, **B** ) is 10 µm.


399


400 GobX. The association of this AH with the plasma membrane causes S-acylation of a cysteine
401 within the AH and membrane association of the protein.


402 Whether this three-step model (1: AH-mediated PM association; 2: PAT-mediated
403 acylation; 3 – DIX-mediated oligomerization) is sufficient to explain the native SOK patterns
404 is an open question. It is possible that the AH in fact offers the interaction interface with the
405 PAT enzyme, not the PM. Likewise, it is possible that either global or localized de-S-acylation
406 may be required to maintain polarized SOK localization, analogous to the localized association
407 and global dissociation of the yeast polar protein Cdc42 (Woods et al., 2016).


408 The identification of the minimal targeting region in SOK1 also allowed exploring the
409 evolution of the polar targeting signals in the family. Through phylogenetic analysis and motif
410 and sequence analysis, we found that SOK proteins fall into several ortholog groups, and that
411 within angiosperms, the EuSOK-B group diversified into three distinct subgroups. A key
412 feature shared by all SOKs is the presence of a Conserved Polarity Fragment (CPF) within their
413 polarity domains (PDs), which drive their polarization. Synthesizing our findings, we propose
414 a scenario for the evolutionary history of polar SOK proteins ( **Fig. 6B** ). Consistent with prior
415 studies (Ramalho et al., 2021)(van Dop et al., 2020), our phylogenetic analysis shows that all
416 extant SOKs originate from a single ancestral protein, AncSOK, at the base of land plants. This
417 AncSOK gene evolved into BryoSOKs in bryophytes and into tracheophyte (vascular plant)
418 SOKs. In vascular plants, a gene duplication event in early euphyllophytes gave rise to two


14


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


419 major ortholog groups: EuSOK-A and EuSOK-B. These two groups have been broadly retained
420 across different euphyllophyte lineages. A subsequent series of duplications in the EuSOK-B
421 lineage resulted in three angiosperm-specific subgroups: EuSOK-B2, -B3, and -B4, which are
422 retained in most angiosperms. Thus, while bryophytes possess at least one BryoSOK gene,
423 ferns have at least two (EuSOK-A and EuSOK-B), and angiosperms typically have at least four
424 SOK genes (one EuSOK-A and three EuSOK-Bs).


425 The evolutionary diversification of SOK proteins was accompanied by changes in their
426 polar localization. Ancestral SOK proteins likely exhibited “inner” membrane localization, as
427 inferred from the polarization patterns of modern BryoSOK, LycoSOK, EuSOK-A and fern
428 EuSOK-B proteins. All tested polarity domains from these groups polarized to the inner
429 membrane when expressed in _A. thaliana_ root cells. This suggests that inner localization, or its
430 equivalent in other species, was the ancestral state in AncSOK and was largely retained
431 throughout SOK evolution. What exactly this “inner” domain corresponded to in the earliest
432 land plants with vastly different morphology, remains an open question.


433 As EuSOK-Bs diversified into the angiosperm-specific subgroups their polar
434 localization patterns shifted. For example, the inner basal localization of AtSOK2 (EuSOK435 B2) and the uniform localization of AtSOK3 (EuSOK-B3) are newly evolved patterns. These
436 shifts in localization are likely caused by the novel sequence motifs that evolved around the
437 ancient CPF, possibly modifying interactions with additional cellular components. We note that
438 this diversification appears restricted to the group that underwent gene duplication events. This
439 suggests that duplications liberated SOK copies to diversify polarity patterns, whilst retaining
440 at least one copy that kept the ancestral pattern. An open question is whether the _de novo_
441 polarity patterns serve specialized functions, and if so, which function. Loss-of-function
442 analysis will be required to answer this question.


443 Different SOK ortholog groups have evolved distinct patterns of charge and
444 hydrophobicity within their polarity domains, reflecting underlying evolutionary trajectories.
445 All EuSOK-A proteins encode amphipathic helices at the CPF. In AtSOK1, this amphipathic
446 helix presumably enables the S-acylation of the conserved cysteine, which anchor the protein
447 to the membrane. We argue that this mechanism likely extends to other EuSOK-A proteins as
448 well. However, our phylogeny indicates that the amphipathic helix-dependent mechanism is an
449 innovation specific to EuSOK-As, which is remarkable, since EuSOK-As seem to have
450 retained the ancestral targeting localization properties of AncSOK. In contrast, EuSOK-B
451 proteins and the earlier-diverging LycoSOK, retain an ancestral configuration of a hydrophobic
452 CPF surrounded by positive charges. BryoSOKs differ even further, lacking a clear predicted
453 helix and instead forming extended, unstructured loops with high positive charge around their
454 CPF. Despite these differences, all studied MBMs drive polar localization, indicating that the
455 conserved positive charge around the MBM is the primary driver of polarization across all SOK
456 proteins. Interestingly, comparative proteomic analysis revealed that both Arabidopsis and
457 Marchantia SOK proteins associate with PAT enzymes, and require the conserved Cysteine in
458 the PD for polar membrane localization (Pukhovaya et al., 2026). Thus, while the initial
459 targeting mechanisms to the PM (or PAT enzyme) may differ, there is deep conservation of the
460 second and third steps in polar targeting.


15


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


461 In summary, our work identifies a minimal targeting region within SOK1, suggests a
462 three-step polar targeting mechanism, and reveals the ancestral properties of SOK proteins and
463 their evolutionary patterns. We anticipate that further analysis of the biochemical and
464 biophysical mechanisms of SOK-PM interaction and loss-of-function analysis of the broader
465 SOK family will be excellent starting points to understand general mechanisms underlying
466 plant cell polarity.


467


16


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


468 **Materials and Methods**


469 **Plants**


470 _Arabidopsis thaliana_ plants of the Columbia-0 ecotype were used for all experiments. For
471 plate-based experiments, sterile seeds were grown on ½ MS + 1% sucrose medium and 1 %
472 Agar (Daichin Agar, Duchefa). Seeds were stratified for 1-3 days at 4 °C, then placed vertically
473 in a growth chamber in a 16 h / 8 h light / dark cycle at a temperature of 22°C. For propagation,
474 5-10 day old seedlings were transferred from plates to soil.


475 For generating transgenic plants, we used the floral dip method (Clough &amp; Bent, 1998). In
476 brief, _Agrobacterium tumefaciens_ (strain GV3101 pSoup) carrying the construct of interest
477 were grown to an OD of 0.5-0.8. Then, the cultures were diluted with an equal volume of 10%
478 sucrose 0.1% Silwet L-77 solution. Plants with young bolts (corresponding to about 7-week479 old plants) were then dipped in the resulting solution, allowed to recover for 1 day in a dark
480 humid chamber, then replaced into the growth rooms.


481 Successfully transformed T 1 plants were selected using an appropriate method depending on
482 the construct used. Most of the constructs in this study use the FastGreen selection
483 marker(Shimada et al., 2010) . In this case, we selected green-fluorescent seeds under a
484 stereomicroscope. For constructs with kanamycin resistance, T 1 seeds were plated on plates
485 supplemented with the 50 mg/L antibiotic. Surviving transformants were selected 10 days later.


486 _Marchantia polymorpha_ Takaragaike-1 (Tak-1) was used as the wild-type accession for all
487 Marchantia lines and experiments. Plants were cultured on ½ strength Gamborg B5 medium
488 (Duchefa G0209) at 22°C under constant white light (40 μmol photons m⁻² s⁻¹).


489 Marchantia Tak-1 spores were transformed with the above constructs using _Agrobacterium_
490 _tumefaciens_ strain GV3101:pMp90, following a published protocol (Ishizaki et al., 2008).
491 Transformants were selected on ½ strength B5 medium supplemented with 10 mg/L
492 hygromycin and 100 μg/mL cefotaxime.


493 **Design and generation of expression constructs**


494 Polarity domain constructs. All polarity domain constructs were expressed as mNeonGreen495 tagged fusions using the RPS5a promoter, which drives expression in the root tips. To this end,
496 we created a new expression vector. A C-terminal mNeonGreen tag was codon-optimized for
497 plants and introduced into the previously published pPLV28 vector (De Rybel et al., 2011),
498 which yielded the vector pNeonGreenKan. To avoid antibiotic-based selection and allow
499 imaging of primary transformants, we replaced the resistance marker in this vector with a
500 FastGreen selection marker. This new vector, pNeonGreenFG, was used as a backbone for
501 further cloning rounds.


502 Constructs were generated as hDVL2-DIX DIX and mNeonGreen fusions, unless stated
503 otherwise. For each construct, a synthetic DNA gBlock (Integrated DNA Technologies)
504 encoding a variant of the SOK1 polarity domain, along with the hDvl2-DIX domain fragment,
505 were used for cloning. These two regions were spaced with a five-amino-acid GGGSG linker.


17


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


506 We assembled the constructs into the PstI site of pNeonGreenFG using HiFi cloning (New
507 England Biolabs).


508 Polarity domain mutants. Targeted mutations were introduced into the polarity domain of
509 SOK1 (amino acids 173-252). We substituted isolated amino acids with alanines (unless
510 otherwise stated). To avoid disturbing the structure of the protein by introducing long stretches
511 of hydrophobic alanines, we substituted longer fragments with flexible glycine-serine linkers.
512 All mutant polarity domains were also expressed as hDvl2-DIX-polarity domain-mNeonGreen
513 fusions as described above. Similarly, to manipulate the amphipathic properties of the SOK1
514 polarity motif, we introduced mutations in the SOK1 203-242 fragment, which we expressed

515 as hDvl2-DIX-SOK1 203-242 -mNeonGreen fusions.


516 Polymerization domain constructs. The DIX domain in SOK1 was replaced with other
517 polymerization domains and expressed as mNeonGreen-tagged fusions. We made three such
518 constructs. The first replaces the SOK1 DIX domain with the DIX polymerization domain from
519 the human DVL2 protein. Note that this construct corresponds to a C-terminal truncation of
520 the previously published hDvl2-SOK1 construct (van Dop et al., 2020). The second construct
521 uses a SAM polymerization domain (Sayou et al., 2016) form the _Arabidopsis thaliana_ LFY
522 protein (amino acids 37-158), which we obtained directly from cDNA. The third construct uses
523 a PB1 polymerization domain from the _Marchantia polymorpha_ ARF2 protein (Kato et al.,
524 2020). Each construct was assembled in the HpaI restriction site of pNeonGreenK using HiFi
525 cloning (new England Biolabs) and two inserts: a polymerization domain fragment and a
526 fragment encoding the SOK1 polarity domain (amino acids 116-252).


527 Tandem oligomers. Two, four or six copies of the polarity domain were spaced with a flexible
528 linker. The resulting tandem dimer, tetramer and hexamer were expressed with the same
529 mNeonGreen tag but without any polymerization domain. To avoid issues resulting from
530 repetitive DNA sequences, the individual units of the oligomers were encoded using different

531 codons.


532 Cross-species polarity domain constructs. We expressed hDvl2-SOK polarity domain533 mNeonGreen (mNG) fusion proteins under the _RPS5a_ promoter as described above. The
534 species and sequences are listed in **Table S2** .


535 An overview of all constructs used in this study can be found in Error! Reference source not

536 found. **Table S1** .


537 **Microscopy and image analysis**


538 Unless stated otherwise, we localized the various SOK proteins in roots of 5-day-old T 1 or T 2
539 seedlings, which represent independent insertion events of the transgene. To visualize cell
540 outlines, cell walls were stained with 20 µg/ml Propidium iodide for 3 minutes. Then, the root
541 tips were imaged on a Leica SP8 Dive multiphoton microscope using a Leica HC PL IRAPO
542 40x/1.10 water immersion objective. The settings of the microscope were as follows: excitation
543 940 nm, emission detection range: 500-560 nm (mNeonGreen) and 600-700 nm (propidium
544 iodide) using HyD detectors with 10% Smart gain, scan format 1024 x 1024 pixels, 400 Hz.


18


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


545 Images were acquired using the LAS X software, and processed in Fiji (Schindelin et al., 2012).

546 .


547 All images were only adjusted for contrast, brightness, and orientation, and these adjustments
548 were applied uniformly to each image. Images with very low or no expression of the transgene
549 were discarded from analysis. For each construct, one representative image is shown. The total
550 number of independent lines imaged for each construct is indicated in each figure panel.


551 **Bioinformatic tools**


552 The 3D structure of SOK1 (UniProt ID A0A178VZT8) was obtained from the AlphaFold
553 Protein Structure Database (Stroe et al., 2021)(Varadi et al., 2022). Molecular graphics and
554 analyses of the polarity domain structure were performed in UCSF ChimeraX version
555 1.7.1(Meng et al., 2023; Pettersen et al., 2021). HELIQUEST (Gautier et al., 2008) was used
556 to perform amphipathicity analysis, hydrophobicity calculations and helical wheel projections.
557 This analysis was performed using a 14 amino acid window. Prediction of peptide-membrane
558 binding dynamics was performed using PMIpred (van Hilten et al., 2024).


559 **Analysis of SOK sequence data**


560 We compiled a set of SOK sequences from various public databases (see Supplementary
561 Methods for details on sequence sources). Most databases have Pfam/InterPro annotations,
562 which allowed us to obtain proteins containing annotated SOSEKI DIX domains (Pfam ID:
563 PF06136). Where no such annotations existed ( _Metasequoia glyptostroboides, Lupinus_
564 _angustifolius_ and _Lycopodium clavatum_ ) we queried the genomes using BLASTP to identify
565 sequences highly homologous to the _A. thaliana_ SOK1 protein. Since genome sequences are
566 available for few bryophyte species, we supplemented our dataset with additional bryophyte
567 SOK sequences from the OneKP dataset (van Dop et al., 2020). We filtered out sequences that
568 did not match a SOK DIX domain using PFAMScan (Madeira et al., 2024)(Mistry et al., 2020),
569 as well as sequences with less than 200 amino acids (median sequence length before filtering:
570 427 amino acids). The resulting dataset contained 1086 sequences from 199 species.


571 Sequences were aligned using MAFFT L-Ins-2 (Katoh et al., 2017). This multiple sequence
572 alignment (MSA) was used to build a phylogenetic tree using FastTree version 2.1.11 (Price et
573 al., 2009). No trimming of the input sequences was performed. We used a JTT + gamma
574 substitution model, which Smart Model Selection for PhyML (Lefort et al., 2017) showed as
575 optimal. The resulting tree was used as a starting tree for GeneRax to infer a maximum
576 likelihood species-tree-aware phylogeny. We ran GeneRax with the undated gene tree evolution
577 model (UndatedDL) and optimized the tree topology using SPR moves with a maximum radius

578 of 6.


579 We retrieved a species tree from OpenTree Taxonomy version 3.6
580 &#91;(https://zenodo.org/records/3937751). There were a few polytomies in the species tree. We&#93;(https://zenodo.org/records/3937751)
581 resolved these polytomies for downstream use by grouping pairs of branches into artificial
582 clades. Furthermore, we rerooted the species tree to reflect the monophyletic origin of
583 bryophytes, which is currently the consensus topology of land plant phylogeny (Harris et al.,
584 2020; Harris et al., 2022; Su et al., 2021). The species tree, corrected gene tree and the


19


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


585 reconciled species-gene-tree showing duplications and losses can be found in the
586 Supplementary materials. Gene trees are visualized and annotated using iTOL (Letunic and
587 Bork, 2024). The reconciled gene family tree was visualized with ThirdKind version 3.6.8
588 (Penel et al., 2022).


589 **Data processing and visualization**


590 Data was processed using R version 4.3.3 (Team, 2014) and RStudio version 2023.12.1+402
591 (Team, 2020). Graphics were generated using ggplot2 (Wickham, 2016). Images were
592 processed using Fiji/ImageJ (Schindelin et al., 2012).


593 **Evolutionary conservation analysis**


594 We used ConSurf (Ashkenazy et al., 2016; Berezin et al., 2004; Yariv et al., 2023) to calculate
595 residue-level conservation of each ortholog group. For this, we selected the respective ortholog
596 sequences of bryoSOKs, lycoSOKs, angiosperm sequences of euSOK-A and the three euSOK597 B groups. Each sequence group was aligned using MAFFT. This alignment was used to
598 calculate conservation scores using the ConSurf web service.


599


600 **Acknowledgements**


601 The authors are grateful to Robertas Ursache and Niko Geldner for sharing plasmids, and for

602 members of our team for helpful discussions.


603


604 **Competing interests**


605 The authors declare no competing interests.


606


607 **Funding**


608 This work was supported by the European Research Council (ERC; AdG “DIRNDL”; contract

609 number 833867) to D.W. and the Sectorplan Biology of the Dutch Ministry of Education,

610 Culture and Science to D.W..


611


612 **Data and resource availability**


613 All primary data are available upon request from the corresponding author. Multiple sequence

614 &#91;alignments are available at https://doi.org/10.6084/m9.figshare.28883615.v2. Sequences used&#93;(https://doi.org/10.6084/m9.figshare.28883615.v2)

615 are available at &#91;https://doi.org/10.6084/m9.figshare.28883615.v2. Phylogenetic trees are&#93;(https://doi.org/10.6084/m9.figshare.28883615.v2)

616 available at &#91;https://itol.embl.de/tree/193142201179291961725458366,&#93;(https://itol.embl.de/tree/193142201179291961725458366)

617 &#91;https://itol.embl.de/tree/8610422102213921726438937&#93;(https://itol.embl.de/tree/8610422102213921726438937) and

618 &#91;https://doi.org/10.6084/m9.figshare.28883615.v2. Plasmid sequences are available at&#93;(https://doi.org/10.6084/m9.figshare.28883615.v2)

619 &#91;https://doi.org/10.6084/m9.figshare.34862703&#93;(https://doi.org/10.6084/m9.figshare.34862703)


20


620


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


21


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


621 **References**


622 **Ashkenazy, H., Abadi, S., Martz, E., Chay, O., Mayrose, I., Pupko, T. and Ben-Tal, N.**

623 (2016). ConSurf 2016: an improved methodology to estimate and visualize evolutionary

624 conservation in macromolecules. _Nucleic Acids Res._ **44**, W344–W350.


625 **Berezin, C., Glaser, F., Rosenberg, J., Paz, I., Pupko, T., Fariselli, P., Casadio, R. and Ben-**

626 **Tal, N.** (2004). ConSeq: the identification of functionally and structurally important residues

627 in protein sequences. _Bioinformatics_ **20**, 1322–1324.


628 **Breda, A. S., Hazak, O. and Hardtke, C. S.** (2017). Phosphosite charge rather than shootward

629 localization determines OCTOPUS activity in root protophloem. _Proceedings of the_

630 _National Academy of Sciences_ **114**, E5721–E5730.


631 **Butler, M. T. and Wallingford, J. B.** (2017). Planar cell polarity in development and disease.

632 _Nat. Rev. Mol. Cell Biol._ **18**, 375–388.


633 **Chiou, J.-G., Balasubramanian, M. K. and Lew, D. J.** (2017). Cell Polarity in Yeast. _Annu._

634 _Rev. Cell Dev. Biol._ **33**, 77–101.


635 **Coelho, S. M. and Weijers, D.** (2026). Convergent paths to multicellular complexity in

636 photosynthetic eukaryotes. _Curr. Biol._ **36**, R668–R676.


637 **De Rybel, B., van den Berg, W., Lokerse, A., Liao, C. Y., van Mourik, H., Moller, B., Peris,**

638 **C. L. and Weijers, D.** (2011). A versatile set of ligation-independent cloning vectors for

639 functional studies in plants. _Plant Physiol_ **156**, 1292–1299.


640 **Dong, J., MacAlister, C. A. and Bergmann, D. C.** (2009). BASL Controls Asymmetric Cell

641 Division in Arabidopsis. _Cell_ **137**, 1320–1330.


642 **Eisenberg, D., Weiss, R. M. and Terwilliger, T. C.** (1982). The helical hydrophobic moment:

643 a measure of the amphiphilicity of a helix. _Nature_ **299**, 371–374.


644 **Evolution 1641-1650. https://doi.org/10.1093/molbev/msp077 Price M. N., 26(7), Dehal, P.**

645 **S. and Arkin, A. P.** (2010). FastTree 2 – Approximately Maximum- Likelihood Trees for

646 Large Alignments. _PLoS One_ **5**, e9490.


647 **Gautier, R., Douguet, D., Antonny, B. and Drin, G.** (2008). HELIQUEST: a web server to

648 screen sequences with specific alpha-helical properties. _Bioinformatics_ **24**, 2101–2102.


649 **Giménez-Andrés, M., Čopič, A. and Antonny, B.** (2018). The Many Faces of Amphipathic.


650 **Harris, B. J., Harrison, C. J., Hetherington, A. M. and Williams, T. A.** (2020).

651 Phylogenomic Evidence for the Monophyly of Bryophytes and the Reductive Evolution of

652 Stomata. _Current Biology_ **30**, 2001-2012.e2002.


653 **Harris, B. J., Clark, J. W., Schrempf, D., Szöllősi, G. J., Donoghue, P. C. J., Hetherington,**

654 **A. M. and Williams, T. A.** (2022). Divergent evolutionary trajectories of bryophytes and


22


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


655 tracheophytes from a complex common ancestor of land plants. _Nat. Ecol. Evol._ **6**, 1634–

656 1643.


657 **Ishizaki, K., Chiyoda, S., Yamato, K. T. and Kohchi, T.** (2008). Agrobacterium-mediated

658 transformation of the haploid liverwort Marchantia polymorpha L., an emerging model for

659 plant biology. _Plant Cell Physiol._ **49**, 1084–91.


660 **Kato, H., Mutte, S. K., Suzuki, H., Crespo, I., Das, S., Radoeva, T., Fontana, M.,**

661 **Yoshitake, Y., Hainiwa, E., van den Berg, W., et al.** (2020). Design principles of a minimal

662 auxin response system. _Nat. Plants_ **6**, 473–482.


663 **Katoh, K., Rozewicki, J. and Yamada, K. D.** (2017). MAFFT online service: multiple

664 sequence alignment, interactive sequence choice and visualization. _Briefings in_ .


665 **Lee, S., Le Roux, A.-L., Goudge, M., Mors, M., Vanni, S., Roca-Cusachs, P. and**

666 **Bahmanyar, S.** (2026). Screening of amphipathic helices identifies features linked to inner

667 nuclear membrane properties. _Nat. Commun._ **17**,.


668 **Lefort, V., Longueville, J.-E. and Gascuel, O.** (2017). SMS: Smart Model Selection in

669 PhyML. _Mol. Biol. Evol._ **34**, 2422–2424.


670 **Letunic, I. and Bork, P.** (2024). Interactive Tree of Life (iTOL) v6: recent updates to the

671 phylogenetic tree display and annotation tool. _Nucleic Acids Research, 52(W1),_ .


672 **Lin, Y. H., Doms, A. G., Cheng, E., Kim, B., Evans, T. R. and Machner, M. P.** (2015). Host

673 Cell-catalyzed S-Palmitoylation Mediates Golgi Targeting of the Legionella Ubiquitin

674 Ligase GobX. _J Biol Chem_ **290**, 25766–25781.


675 **Madeira, F., Madhusoodanan, N., Lee, J., Eusebi, A., Niewielska, A., Tivey, A. R. N.,**

676 **Lopez, R. and Butcher, S.** (2024). The EMBL-EBI Job Dispatcher sequence analysis tools

677 framework in 2024. _Nucleic Acids Res._ **52**, W521–W525.


678 **Meng, E. C., Goddard, T. D., Pettersen, E. F., Couch, G. S., Pearson, Z. J., Morris, J. H.**

679 **and Ferrin, T. E.** (2023). UCSF ChimeraX: Tools for structure building and analysis.

680 _Protein Science_ **32**, e4792.


681 **Mistry, J., Chuguransky, S., Williams, L., Qureshi, M., Salazar, G. A., Sonnhammer, E.**

682 **L. L., Tosatto, S. C. E., Paladin, L., Raj, S., Richardson, L. J., et al.** (2020). Pfam: The

683 protein families database in 2021. _Nucleic Acids Res._ **49**, D412–D419.


684 **Muroyama, A. and Bergmann, D.** (2019). Plant Cell Polarity: Creating Diversity from Inside

685 the Box. _Annu Rev Cell Dev Biol_ **35**, 309–336.


686 **Penel, S., Menet, H., Tricou, T., Daubin, V. and Tannier, E.** (2022). Thirdkind: displaying

687 phylogenetic encounters beyond 2-level reconciliation. _Bioinformatics_ **38**, 2350–2352.


688 **Pettersen, E. F., Goddard, T. D., Huang, C. C., Meng, E. C., Couch, G. S., Croll, T. I.,**

689 **Morris, J. H. and Ferrin, T. E.** (2021). UCSF ChimeraX: Structure visualization for

690 researchers, educators, and developers. _Protein Science_ **30**, 70–82.


23


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


691 **Plain, F., Congreve, S. D., Yee, R. S. Z., Kennedy, J., Howie, J., Kuo, C. W., Fraser, N. J.**

692 **and Fuller, W.** (2017). An amphipathic α-helix directs palmitoylation of the large

693 intracellular loop of the sodium/calcium exchanger. _J Biol Chem_ **292**, 10745–10752.


694 **Price, M. N., Dehal, P. S. and Arkin, A. P.** (2009). FastTree: Computing Large Minimum

695 Evolution Trees with Profiles instead of a Distance Matrix. _Molecular Biology and_ .


696 **Pukhovaya, E. M., Albrecht, C., van Dop, M., Jones, V., Roosjen, M., Volkov, A., Ramalho,**

697 **J. J., Mutte, S., Su, C., Strutt, H., et al.** (2026). Systematic proteomics identifies a

698 conserved mechanism for polar targeting of plant cortical proteins.


699 **Ramalho, J. J., Jones, V. A. S., Mutte, S. and Weijers, D.** (2021). Pole position: How plant

700 cells polarize along the axes. _Plant Cell_ **34**, 174–192.


701 **Sayou, C., Nanao, M. H., Jamin, M., Posé, D., Thévenon, E., Grégoire, L., Tichtinsky, G.,**

702 **Denay, G., Ott, F., Peirats Llobet, M., et al.** (2016). A SAM oligomerization domain

703 shapes the genomic binding landscape of the LEAFY transcription factor. _Nat. Commun._ **7**,

704 11222.


705 **Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T.,**

706 **Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al.** (2012). Fiji: an open-source

707 platform for biological-image analysis. _Nat. Methods_ **9**, 676–682.


708 **Shimada, T. L., Shimada, T. and Hara-Nishimura, I.** (2010). A rapid and non-destructive

709 screenable marker, FAST, for identifying transformed seeds of Arabidopsis thaliana. _Plant_

710 _J._ **61**, 519–28.


711 **St. Johnston, D.** (2018). Establishing and transducing cell polarity: common themes and

712 variations. _Curr. Opin. Cell Biol._ **51**, 33–41.


713 **Stroe, O., Wood, G., Laydon, A., Žídek, A., Green, T., Tunyasuvunakool, K., Petersen, S.,**

714 **Jumper, J., Clancy, E., Green, R., et al.** (2021). AlphaFold Protein Structure Database:

715 massively expanding the structural coverage of protein-sequence space with high-accuracy

716 models. _Nucleic Acids Res._ **50**, D439–D444.


717 **Su, D., Yang, L., Shi, X., Ma, X., Zhou, X., Hedges, S. B. and Zhong, B.** (2021). Large-Scale

718 Phylogenomic Analyses Reveal the Monophyly of Bryophytes and Neoproterozoic Origin

719 of Land Plants. _Mol. Biol. Evol._ **38**, 3332–3344.


720 **Takano, J., Tanaka, M., Toyoda, A., Miwa, K., Kasai, K., Fuji, K., Onouchi, H., Naito, S.**

721 **and Fujiwara, T.** (2010). Polar localization and degradation of Arabidopsis boron

722 transporters through distinct trafficking pathways. _Proceedings of the National Academy of_

723 _Sciences_ **107**, 5220–5225.


724 **Team, R. C.** (2014). R: A language and environment for statistical computing. _MSOR_

725 _connections_ **1**,.


726 **Team, R. S.** (2020). RStudio: Integrated Development Environment for R.


24


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


727 **van Dop, M., Fiedler, M., Mutte, S., de Keijzer, J., Olijslager, L., Albrecht, C., Liao, C.-**

728 **Y., Janson, M. E., Bienz, M. and Weijers, D.** (2020). DIX Domain Polymerization Drives

729 Assembly of Plant Cell Polarity Complexes. _Cell_ **180**, 427-439.e12.


730 **van Hilten, N., Verwei, N., Methorst, J., Nase, C., Bernatavicius, A. and Risselada, H. J.**

731 (2024). PMIpred: a physics-informed web server for quantitative protein–membrane

732 interaction prediction. _Bioinformatics_ **40**,.


733 **Vanneste, S. and Friml, J.** (2009). Auxin: a trigger for change in plant development. _Cell_ **136**,

734 1005–1016.


735 **Varadi, M., Anyango, S., Deshpande, M., Nair, S., Natassia, C., Yordanova, G., Yuan, D.,**

736 **Stroe, O., Wood, G., Laydon, A., et al.** (2022). AlphaFold Protein Structure Database:

737 massively expanding the structural coverage of protein-sequence space with high-accuracy

738 models. _Nucleic Acids Res._ **50**, D439–D444.


739 **Wickham, H.** (2016). _ggplot2: Elegant Graphics for Data Analysis_ .


740 **Wiśniewska, J., Xu, J., Seifertová, D., Brewer, P. B., Růžička, K., Blilou, I., Rouquié, D.,**

741 **Benková, E., Scheres, B. and Friml, J.** (2006). Polar PIN Localization Directs Auxin Flow

742 in Plants. _Science (1979)._ **312**, 883–883.


743 **Woods, B., Lai, H., Wu, C.-F., Zyla, T. R., Savage, N. S. and Lew, D. J.** (2016). Parallel

744 Actin-Independent Recycling Pathways Polarize Cdc42 in Budding Yeast. _Current Biology_

745 **26**, 2114–2126.


746 **Yariv, B., Yariv, E., Kessel, A., Masrati, G., Chorin, A. B., Martz, E., Mayrose, I., Pupko,**

747 **T. and Ben-Tal, N.** (2023). Using evolutionary data to make sense of macromolecules with

748 a “face-lifted” ConSurf. _Protein Science_ **32**, e4582.


749 **Yoshida, S., van der Schuren, A., van Dop, M., van Galen, L., Saiga, S., Adibi, M., Möller,**

750 **B., ten Hove, C. A., Marhavy, P., Smith, R., et al.** (2019). A SOSEKI-based coordinate

751 system interprets global polarity cues in Arabidopsis. _Nat. Plants_ **5**, 160–166.


752


753


754


755


756


757


25


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


758 **Supplementary Materials**


759


760


761 **Figure S1: Predicted membrane binding properties of the SOK1 MPF**


762 Plots depicting the membrane-binding probability of the SOK1-MPF as a function of membrane curvature
763 (left) or vesicle radius (right). SOK1-MPF is indicated with a red dot (left) or line (right), relative to reference
764 non-binder, sensor and binder peptides.


765


26


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


766 **Table S1:** List of all constructs used in this study, with protein sequences


|Protein|Plasmid|Fig.|Expressed protein (schematic)|
|---|---|---|---|
|hDvl2-DIX-S1-&lt;br&gt;PD|AV87|1K|`&#91;hDvl2-DIX&#93;-&#91;linker&#93;-&#91;SOK1-PD&#93; -&#91;mNeonGreen&#93;`|
|MpARF2-PB1-&lt;br&gt;S1-PD|AV89|1L|`SNKTSQNFQQGPVRSYTKIHKQGSFGRSIDVQSYDGYTDLLRKVENMFELNGELFDKKSGWQLVYTDHEDDVLLVGDDPWMEFVSCVRTLRLLSPGEASSS`&lt;br&gt;`GKSGQSHDEDAGAGKDGAKRCDSSSPSAGARGDDM-&#91;linker&#93;-&#91;SOK1-PD&#93; -&#91;mNeonGreen&#93;`|
|LFY-SAM-S1-PD|AV88|1M|`PQTAAFGMRLGGLEGLFGPYGIRFYTAAKIAELGFTASTLVGMKDEELEEMMNSLSHIFRWELLVGERYGIKAAVRAERRRLQEEEEEESSRRRHLLLSAA`&lt;br&gt;`GDSGTHHALDALSQEGLSEEP-&#91;linker&#93;-&#91;SOK1-PD&#93; -&#91;mNeonGreen&#93;`|
|2xS1-PD|AV107|1O|`&#91;SOK1-PD&#93;- &#91;SOK1-PD&#93; -&#91;mNeonGreen&#93;`|
|4xS1-PD|AV108|1P|`&#91;SOK1-PD&#93;- &#91;SOK1-PD&#93;- &#91;SOK1-PD&#93;- &#91;SOK1-PD&#93; -&#91;mNeonGreen&#93;`|
|6xS1-PD|AV109|1Q|`&#91;SOK1-PD&#93;- &#91;SOK1-PD&#93;- &#91;SOK1-PD&#93;- &#91;SOK1-PD&#93;- &#91;SOK1-PD&#93;- &#91;SOK1-PD&#93; -&#91;mNeonGreen&#93;`|
|original|AV92|2B|`&#91;hDvl2-DIX&#93;- &#91;SOK1-PD&#93; -&#91;mNeonGreen&#93;`|
|Scan1|AV93|`2C`|`&#91;hDvl2-DIX&#93;-`**`GGSGSSGGGS`**`QRDGSTENGSGNDVESGRPSVSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGGLDTNDAVLVPLNKSRSG-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|Scan2|AV94|2D|`&#91;hDvl2-DIX&#93;-KKQDPKTVSG`**`GGSGSSGGGA`**`GNDVESGRPSVSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGGLDTNDAVLVPLNKSRSG-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|Scan3|AV95|2E|`&#91;hDvl2-DIX&#93;-KKQDPKTVSGQRDGSTENGS`**`GGSGSSGGSG`**`VSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGGLDTNDAVLVPLNKSRSG-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|Scan4|AV96|2F|`&#91;hDvl2-DIX&#93;-KKQDPKTVSGQRDGSTENGSGNDVESGRPS`**`GGSGSSGGGS`**`IKNKSYSSVRASHVLRNLMKCGGLDTNDAVLVPLNKSRSG-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|Scan5|AV97|2G|`&#91;hDvl2-DIX&#93;-KKQDPKTVSGQRDGSTENGSGNDVESGRPSVSSTTSSSSY`**`GGSGSSGGGS`**`ASHVLRNLMKCGGLDTNDAVLVPLNKSRSG-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|Scan6|AV98|2H|`&#91;hDvl2-DIX&#93;-KKQDPKTVSGQRDGSTENGSGNDVESGRPSVSSTTSSSSYIKNKSYSSVR`**`GGSGSSGGGS`**`CGGLDTNDAVLVPLNKSRSG-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|Scan7|AV99|2I|`&#91;hDvl2-DIX&#93;-KKQDPKTVSGQRDGSTENGSGNDVESGRPSVSSTTSSSSYIKNKSYSSVRASHVLRNLMKCG`**`SGSSGGGS`**`LVPLNKSRSG-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|Scan8|AV100|2J|`&#91;hDvl2-DIX&#93;-KKQDPKTVSGQRDGSTENGSGNDVESGRPSVSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGGLDTNDAV`**`GGSGSSGGGS`**`-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|Fragment 1|AV137|2L|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|Fragment 2|AV138|2M|`&#91;hDvl2-DIX&#93;-SSSSYIKNKSYSSVRASHVLRNLMKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|Fragment 3|AV139|2N|`&#91;hDvl2-DIX&#93;-IKNKSYSSVRASHVLRNLMKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|Fragment 4|AV140|2O|`&#91;hDvl2-DIX&#93;-YSSVRASHVLRNLMKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|Fragment 5|AV141|2P|`&#91;hDvl2-DIX&#93;-ASHVLRNLMKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|Fragment 6|AV142|2Q|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGGLD-&#91;mNeonGreen&#93;`|
|Fragment 7|AV143|2R|`&#91;hDvl2-DIX&#93;-SSSSYIKNKSYSSVRASHVLRNLMKCGGLD-&#91;mNeonGreen&#93;`|
|Fragment 8|AV144|2S|`&#91;hDvl2-DIX&#93;-IKNKSYSSVRASHVLRNLMKCGGLD-&#91;mNeonGreen&#93;`|
|Fragment 9|AV145|2T|`&#91;hDvl2-DIX&#93;-YSSVRASHVLRNLMKCGGLD-&#91;mNeonGreen&#93;`|
|Fragment 10|AV146|2U|`&#91;hDvl2-DIX&#93;-ASHVLRNLMKCGGLD-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD K216A|AV147|`3H`|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKN`**`A`**`SYSSVRASHVLRNLMKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD A223N&lt;br&gt;S224A H225A|AV148|3L|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVR`**`NAA`**`VLRNLMKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD L227A&lt;br&gt;L230A|AV149|3G|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVRASHV`**`A`**`RN`**`A`**`MKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD R228A&lt;br&gt;N229A|AV150|3I|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVRASHVL`**`AA`**`LMKCGGLDTNDAV-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD K232A|AV151|3J|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVRASHVLRNLM`**`A`**`CGGLDTNDAV-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD L236N&lt;br&gt;D237A|AV152|3M|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGG`**`NA`**`TNDAV-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD T238G|AV153|3N|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGGLD`**`G`**`NDAV-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD N239A&lt;br&gt;D240A|AV154|3O|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKNKSYSSVRASHVLRNLMKCGGLDT`**`AA`**`AV-&#91;mNeonGreen&#93;`|
|hDvl2-DIX-S1-&lt;br&gt;PD K216A&lt;br&gt;R228A K232A|AV155|3K|`&#91;hDvl2-DIX&#93;-VSSTTSSSSYIKN`**`A`**`SYSSVRASHVL`**`A`**`NLM`**`A`**`CGGLDTNDAV-&#91;mNeonGreen&#93;`|
|AtrSOK1|AV174|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`IHENPSASIRQNLSLKIIQKEAELPENSNPPEFSSKPSSLKAEDQNGISNGPKKSGASSSSLHMFRNLLSCGGVETNESKIKMMSKNNPNR-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|AtSOK3|AV158|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`RSKTPIETFSRGVSTDEDVSSEPETSENNLVSEASCAGKERESAEISRNSVSPPFSNSASSLGGKTDTLESLIRADVSKMNSFRILEQEDVRMPAIPRLRA`&lt;br&gt;`SNMLMQLISCGSISVKDNNF-&#91;mNeonGreen&#93;`|
|CpSOK1|AV171|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`TSPTSPKIANQEKHRNAASQTAQREPASQLKDVTTDCNLEPATTMTSKKDEMTSVSPLPERTPRSEGCTRRRIWEKEIDKNAIYTQFVSQKNGSDVDERPG`&lt;br&gt;`SAASEECSGSNSCVSGPDDPFLYGVLRKATRLGSSFRPRLCRGVDVVDSTRTKTKLHYRAKSKGEGHVKQPSSFIDEPSSKNGTTSPAEVNSETKSY-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|CtSOKB1|AV164|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`QRVSSTSVREASSLDCEVSLHSAMLDLCREEILQPLYSTSSSTCTSTHSASSSSITSSSVDAHQNCITGKPCCSAEDQKLEKLLTSTPRVQLGKEALLNDE`&lt;br&gt;`DTSAAYFAANGNDQGHNTCGMSAGFKGTSMQTTEFAKGIKPSFSSSTVKSQVMGLSSNSSKFLHLISCGRGLEIINPSDSHIQRSSSMNSICHTDSAPGTQ`&lt;br&gt;`SASLLINKHNGF-&#91;mNeonGreen&#93;`|
|CtSOKB2|AV167|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`RSWMKTAIHKLEERNDKFYQHNSSRQGVDEEFYKDLSSVDRNADTDFDRRLTNNSTASMAVEDAHSSPEAVGKHGRGKHFSVVSFLQFMSCGGLDIKEQML`&lt;br&gt;`PMSLYRLRSRGTPAVANQLRMSNYVLSPEHPLPDSHEACGSSEVRNLWSRISKTKSL-&#91;mNeonGreen&#93;`|


27


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


767


768

|LTSOKB4-1|AV166|5|&#91;hDvl2-DIX&#93;-&lt;br&gt;RRRRAVAEEDEQKNGIEVSEIETEGERDPHVTCQNQSTELSIDEISPPPSSDSSPETLETMMKVNGRLGLRALFGTKENNLTTESDSSGKMRAPSILLQLL&lt;br&gt;SCGAVSFKESGVNKDQGF-&#91;mNeonGreen&#93;|
|---|---|---|---|
|MpSOK|AV170|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`STPFMLSPRRLMAALSSPSPEKKFGKLVHSSSTRSSTPSTSAASTQCEDSLPNINIRLAKQATCLSNFRLCGNANPHATDSRPDSPEHPLAA-`&lt;br&gt;`&#91;mNeonGreen&#93;`|
|MtSOK|AV172|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`TQPDDLPPLKSRTIPRIATSAQHAIDPAKGAPPTPEPSRAAVAPPTPAASSFAVAASDRKTLAKICLRRVGGKDDEPAVSSARLKSGHSFRQLLCGVVDVR`&lt;br&gt;`SPANGFSPEHQVSTPHIGSAGWLWRSKSKTSSERDKPERAGRDNIKSAHY-&#91;mNeonGreen&#93;`|
|MUSOK1|AV163|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`KEESPPLSITDNSFEKSASFETSEEFSRTSSYSKLSCKNNDVVKPENIKAPSKVPRHGKAKAGGASNVLRQLFTCGGLDTRDSRILAVKAITFNRLSNEDN`&lt;br&gt;`RSEGVVNNPPSSTAPTKSESR-&#91;mNeonGreen&#93;`|
|PpSOK2|AV175|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`TRPCFPGDPLIFMLKKATKFHRSQLCRKAEVEGSPCATVGKPSSSRHTSKIRHGSGPLKAYHGPHSSKSATQNSRPVTYETSAKESERDIRSLCQFLGASN`&lt;br&gt;`VGSVSQTINKEGTKKTGSARHIQAGSNDDFDFARHKEV-&#91;mNeonGreen&#93;`|
|SmSOK1|AV173|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`ERRHPEQLVPSVTPVGKGPSKHEGGKLKRRFGRQQAEGKGASPKVKMKHSAASCVFLHILSYCGGVDTKMQSRAPTEITSSEYSFDGGICRAENKMVMTSQ`&lt;br&gt;`NRSPQQY-&#91;mNeonGreen&#93;`|
|WsSOKA|AV168|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`RRVSSTSVNKNHHAGSSVIRGNLSKQLSKEFSNLPDVEKALIINQTSDLGEDSKAANVSSSNNTSLLNHMGSPVLSVCSAMSTRSTRFAREDNNRSSSAAV`&lt;br&gt;`KNSSKKNVSKPDGHHTHHKGSSHSSSTSTTSSAPTLLKQLLTCGNADVISESKPSDNVIMSSNALSAGVQTPSRLISTSGHQGGLTSSHDKKGSNSNNNNM`&lt;br&gt;`W-&#91;mNeonGreen&#93;`|
|ZmSOKB3-1|AV165|5|`&#91;hDvl2-DIX&#93;-`&lt;br&gt;`RDAAENHSTRVAGASTMDTTSDAEIQECHRRSSPLSPKVPETVRESPPAVCSSDAASPGGRVETLESLIRAEASRRSSFRTLEEGHMLGPVGVKLKPANLL`&lt;br&gt;`MQLITCGFVSVKEHRGF-&#91;mNeonGreen&#93;`|


28


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.10.30.762996; this version posted November 2, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.10.30.762996)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


769 **Table S2:** Polarity domains used in this study. Shown are abbreviated sample names used in
770 the main text (for bioinformatic analysis and microscopy of hDvl2-polarity domain-mNG
771 fusions), corresponding species and gene or sequence ID, as well as the amino acids
772 corresponding to the polarity domain. In some cases, we made more than one construct per
773 gene, and the respective polarity fragments are given in the table.


774


775


|Protein|Type|Species|GeneorSequenceID|Region|
|---|---|---|---|---|
|AtrSOK1|EuSOKA|_Amborella trichopoda_|AmTrH1.03G055500|192-282|
|AtSOK1|EuSOKA|_Arabidopsis thaliana_|AT1G05577|173-252|
|AtSOK3|EuSOKB3|_Arabidopsis thaliana_|AT2G28150|232-352|
|CpSOK1|BryoSOK|_Ceratodon purpureus_|CepurR40.11G122000|257-454|
|CtSOKB1|EuSOKB|_Ceratopteris_&lt;br&gt;_thalictroides_|LM_PIVW_2013750_&lt;br&gt;Ceratopteris_thalictroi&lt;br&gt;des_g.430|319-532|
|CtSOKB2|EuSOKB|_Ceratopteris_&lt;br&gt;_thalictroides_|LM_PIVW_2008561_&lt;br&gt;Ceratopteris_thalictroi&lt;br&gt;des_g.428|391-548|
|LaSOKB4-1|EuSOKB4|_Lupinus angustifolius_|Lupan_006625|193-311|
|MpSOK|BryoSOK|_Marchantia_&lt;br&gt;_polymorpha_|Mapoly0032s0007|423-515|
|MtSOK|BryoSOK|_Megaceros tosanus_|HW_UCRN_2009276_&lt;br&gt;Megaceros_tosanus_g.&lt;br&gt;336|241-391|
|MgSOK1|EuSOKA|_Metasequoia_&lt;br&gt;_glyptostroboides_|Mgl0103920|239-360|
|PpSOK2|BryoSOK|_Physcomitrium patens_|Pp3c2_20380|310-448|
|SmSOK1|LycoSOK|_Selaginella_&lt;br&gt;_moellendorffii_|412562|293-400|
|WsSOKA|EuSOKA|_Woodsia scopulina_|LM_YJJY_2005726_&lt;br&gt;Woodsia_scopulina_g.&lt;br&gt;459|391-593|
|ZmSOKB3-1|EuSOKB3|_Zea mays_|Zm00001eb039020|292-409|


29