Plasmodesmal connectivity in C4 Gynandropsis gynandra is induced by light and dependent on photosynthesis

In leaves of C4 plants the reactions of photosynthesis become restricted between two compartments. Typically, this allows accumulation of C4 acids in mesophyll cells and subsequent decarboxylation in the bundle sheath. In C4 grasses proliferation of plasmodesmata between these cell types is thought to increase cell-to-cell connectivity to allow efficient metabolite movement. However, it is not known if C4 dicotyledons also show this enhanced plasmodesmal connectivity and so whether this is a general requirement for C4 photosynthesis is not clear. How mesophyll and bundle sheath cells in C4 leaves become highly connected is also not known. We investigated these questions using 3D- and 2D- electron microscopy on the C4 dicotyledon Gynandropsis gynandra as well as phylogenetically close C3 relatives. The mesophyll-bundle sheath interface of C4 G. gynandra showed higher plasmodesmal frequency compared with closely related C3 species. Formation of these plasmodesmata was induced by light. Pharmacological agents that perturbed chloroplast development or photosynthesis reduced the number of plasmodesmata, but this inhibitory effect could be reversed by the provision of exogenous sucrose. We conclude that enhanced formation of plasmodesmata between mesophyll and bundle sheath cells is wired to the induction of photosynthesis in C4 G. gynandra.


DISCUSSION (1343 words) 33 48
• The mesophyll-bundle sheath interface of C4 G. gynandra showed higher 49 plasmodesmal frequency compared with closely related C3 species. Formation of 50 these plasmodesmata was induced by light. Pharmacological agents that perturbed 51 chloroplast development or photosynthesis reduced the number of plasmodesmata, 52 but this inhibitory effect could be reversed by the provision of exogenous sucrose. 53 54 • We conclude that enhanced formation of plasmodesmata between mesophyll and 55 bundle sheath cells is wired to the induction of photosynthesis in C4 G. gynandra. 56 57 INTRODUCTION 4 viridis compared with the C3 species rice and wheat. This increase in the C4 grasses was 93 due to a 2-fold increase in plasmodesmata numbers per pitfield, and a 5-fold increase in 94 pitfield area. In other C4 grasses substantial variation in absolute plasmodesmata frequency 95 was evident but they all possessed greater plasmodesmata frequency than C3 species 96 (Danila et al., 2018). 97 To our knowledge, the distribution of plasmodesmata at the mesophyll-bundle sheath cell 98 interface of C3 and C4 species has not been studied outside the grasses. Further, the cues 99 that underpin increased plasmodesmata formation are not known. Given the known variation 100 in how increased cell-to-cell connectivity is achieved in C4 grasses and the fact that they 101 carbon tabs (TAAB Laboratories Equipment Ltd), sputter-coated with a thin layer of carbon 188 ( 30 nm) to avoid charging and imaged in a Verios 460 scanning electron microscope at 4 189 keV accelerating voltage and 0.2 nA probe current using the concentric backscatter detector 190 in field-free (low magnification) or immersion (high magnification) mode (working distance Plasmodesmal frequency from 2D and 3D EM images was determined using published 205 methods (Koteyeva et al., 2014;Botha, 1992). Briefly, plasmodesmal frequency was 206 determined as the number of plasmodesmata observed per m of length of shared cell 207 interface between two cell types (mesophyllbundle sheath, mesophyllmesophyll, bundle 208 sheathbundle sheath). Plasmodesmata numbers and cell lengths were determined using 209 ImageJ software. Plasmodesmata were defined as dark channels in the EM images. 210 Depending on plasmodesmata orientation, the entire channel was sometimes not visible on 211 2D EM images, and so only channels that spanned more than half of the cell wall width were 212 counted. 213 214

Chlorophyll fluorescence measurement 215
Chlorophyll fluorescence measurements were carried out using a CF imager 216 (Technologica Ltd, UK) and image processing software provided by the manufacturer. 217 Seedlings were placed in the dark for 20 min evaluate dark-adapted minimum fluorescence 218 (Fo), dark-adapted maximum fluorescence (Fm) and then variable fluorescence Fv 219 (Fv = Fm-Fo). All chlorophyll fluorescence images of inhibitor-treated seedlings within each 220 experiment were acquired at the same time in a single image, measuring a total of 8 221 seedlings per treatment.

Statistical analysis 224
In violin plots, the middle line represents the median, the box and whiskers represent the 225 25 to 75 percentile and minimum-maximum distributions of the data. Letters show the 226 statistical ranking using a one-way ANOVA and post hoc Tukey test (different letters indicate 227 differences at P<0.05). Values indicated by the same letter are not statistically different. 228 Data was analyzed using RStudio 2022.07.2+576. 229

RESULTS 230
Plasmodesmata frequency is higher in C4 G. gynandra leaves compared with C3 A. 231

thaliana and T. hassleriana 232
We first explored whether the increased plasmodesmal connectivity between mesophyll 233 and bundle sheath cells found in C4 grasses was also present in the C4 dicotyledon 234 Gynandropsis gynandra. Transmission electron microscopy was used to examine the 235 mesophyll-bundle sheath cell interface in mature leaves of G. gynandra and the closely 236 related C3 species Tarenaya hassleriana (also a member of the Cleomaceae) as well as C3 237 Arabidopsis thaliana. Plasmodesmata were more abundant between mesophyll and bundle 238 cells in C4 G. gynandra compared with both C3 species (Fig. 1). Increased physical 239 connectivity was specific to this interface, and no obvious increases were detected at the 240 mesophyll-mesophyll or bundle sheath-bundle sheath cell interfaces in any species 241 (Supporting Information Fig. S1). 242 To quantify plasmodesmata numbers between mesophyll and bundle sheath cells, we 243 conducted serial block-face scanning electron microscopy (SBF-SEM). SBF-SEM offers 244 excellent resolution in 3D and has previously been used to quantify plasmodesmata in other 245 systems (Ross-Elliott et al., 2017; Paterlini and Belevich, 2022). Thin sections prepared from 246 fully expanded true leaves of G. gynandra, T. hassleriana and A. thaliana were imaged, and 247 an area of the mesophyll-bundle sheath cell interface identified for serial block face 248 sectioning ( Fig. 2a-c). From each species, between 281-438 serial transverse sections per 249 mesophyll-bundle sheath cell interface were collected and compiled into videos (Supporting 250 Information Videos S1-3). Using these SBF-SEM sections we quantified plasmodesmata 251 frequency by determining the number of plasmodesmata per length of mesophyll-bundle 252 sheath cell interface imaged in 3D (Fig. 2d). In C4 G. gynandra, plasmodesmata were visible 253 in almost every mesophyll-bundle sheath cell interface assessed such that only 20 out of 254 467 contained no plasmodesmata (Fig. 2d). In contrast, in the two C3 species 255 plasmodesmata were not detected in the majority of interfaces (263/367 for T. hassleriana, 256 628/886 for A. thaliana). Because plasmodesmata appear in clusters (pitfields) rather than 257 being equally distributed, a wide range of plasmodesmal frequencies per section were 258 observed between mesophyll and bundle sheath cells in all three species. However, there 259 were more sections with higher frequencies observed at the mesophyll-bundle sheath 260 interface of C4 G. gynandra, and this resulted in a 13-fold increase in the mean frequency 261 compared with C3 T. hassleriana and C3 A. thaliana (Fig. 2d). Plasmodesmal frequencies 262 between mesophyll and bundle sheath cells of the C3 species T. hassleriana and A. thaliana 263 were not significantly different to each other and were low compared with C4 G. gynandra.

10
SBF-SEM provides an excellent 3D view of plasmodesmata frequency and distribution 265 but is relatively low throughput and so limited numbers of cell interfaces can be visualised 266 per unit time. We therefore used 2D electron microscopy to further explore the high 267 occurrence of plasmodesmata at the mesophyll-bundle sheath cell interface of C4 G. 268 gynandra. Large areas of leaf sections were imaged at high resolution using 2D Scanning 269 Electron Microscopy (SEM) mapping such that automated serial imaging at 10,000X 270 magnification and subsequent image stitching enabled visualization of plasmodesmata at 271 numerous interfaces of the same 2D section (Fig. 3a). Representative SEM maps in which 272 cell interfaces (mesophyll-bundle sheath, mesophyll-mesophyll and bundle sheath-bundle 273 sheath) were pseudocoloured according to plasmodesmal frequency, and consistent with 274 the 3D SBF-SEM analysis reported above, illustrated that plasmodesmata were specifically 275 enriched at the mesophyll-bundle sheath interface of C4 G. gynandra (indicated by the 276 numerous green-coloured interfaces). In contrast, frequency was lower and more uniform 277 between all cellular interfaces in the C3 species (indicated by the pink and orange 278 pseudocoloured cell interfaces) (Fig. 3a). Plasmodesmata frequencies were quantified from 279 at least three SEM maps originating from three independent plants (biological replicates; 10-280 40 individual mesophyllbundle sheath, mesophyll-mesophyll and bundle sheath-bundle 281 sheath cell interfaces per biological replicate). This showed that plasmodesmata numbers 282 between mesophyll and bundle sheath cells were more than 8-fold higher in C4 G. gynandra 283 compared with both C3 species. The three cellular interfaces (mesophyll-bundle sheath, 284 mesophyll-mesophyll and bundle sheath-bundle sheath) in both C3 species had similar 285 plasmodesmal frequencies. Interestingly, plasmodesmal frequency of all three types of cell 286 interface in G. gynandra was significantly higher than that of the corresponding interface in 287 each of the two C3 species. For example, the mesophyll-mesophyll and bundle sheath-288 bundle sheath interfaces were approximately 3 to 4-fold and 2-fold higher in C4 G. gynandra 289 compared with T. hassleriana and A. thaliana respectively indicating that cell-to-cell 290 connectivity is generally enhanced between photosynthetic cells of the C4 species (Fig. 2b). 291 Plasmodesmata frequencies estimated from analysis of numerous mesophyll-bundle sheath 292 cell interfaces using this 2D SEM mapping were not statistically different from the 293 frequencies obtained from multiple serial sections of the mesophyll-bundle sheath interface 294 using SBF-SEM (Supporting Information Fig. 2). To allow greater replication and sampling 295 subsequent analysis was therefore carried out with the 2D SEM mapping technique. 296 To investigate the relationship between increased frequency of plasmodesmata at the 297 mesophyll-bundle sheath interface and pit fields, we visualized pitfields using SEM by clearly visible at the mesophyll-bundle sheath interface in all species, but unlike the previous 300 work in grasses individual plasmodesmata within the pitfields could not be distinguished 301 (Supporting Information Fig. 3a). When we measured the mean area of pitfields in each 302 species there was no clear difference. This suggests that the increased plasmodesmata 303 frequency at mesophyll-bundle sheath in G. gynandra most likely results from increased pit 304 field numbers per cell interface rather than enlarged pit fields that contain more 305 plasmodesmata (Supporting Information Fig. 3b). cotyledons showed that Kranz anatomy was already partially developed in 3-day-old dark 315 grown seedlings (Fig. 4a). For example, veins were closely spaced, and bundle sheath cells 316 contained abundant organelles. However, after 24 h of light cotyledons had almost doubled 317 in size and substantial cell expansion and formation of air spaces was evident (Fig. 4a). 318 High-resolution 2D SEM maps from cross sections of at least three cotyledons (biological 319 replicates) of G. gynandra were obtained at 0 h, 24 h and 48 h after transfer to light. In dark-320 grown seedlings plasmodesmal frequency at mesophyll-bundle sheath, mesophyll-321 mesophyll, and bundle sheath-bundle sheath were similar (n = 204) (Fig. 4c,d). However, 322 after light induction plasmodesmal frequency increased 1.7-fold after 24 h and 2.5-fold after 323 48 h between mesophyll and bundle sheath cells of G. gynandra ( Fig. 4b-d). There was also 324 a small increase in plasmodesmata numbers between mesophyll cells after light exposure. 325 These responses were specific to de-etiolation because growth in the dark for 48 h did not 326 increase plasmodesmata numbers (Supporting Information Fig. 4a-d). These data indicate 327 that as with true leaves, cotyledons of G. gynandra develop high plasmodesmal connectivity 328 between mesophyll and bundle sheath cells, and that this takes place rapidly in response to 329 light. We conclude that light is a crucial developmental cue for the formation of secondary Norflurazon treatment generated seedlings with white cotyledons, consistent with 348 compromised carotenoid accumulation (Fig. 5a). Etioplast ultrastructure was largely 349 unaffected by the inhibitors (Fig. 5b). After 48 h of light cotyledons of controls and DCMU-350 treated seedlings were green and etioplasts had developed into chloroplasts (Fig. 5a,b). 351 Norflurazon and lincomycin-treated seedlings had pale cotyledons even after light induction 352 and the etioplast-to-chloroplast development was arrested (Fig. 5a,b). To confirm that each 353 inhibitor had the expected effect on chloroplast function we used chlorophyll fluorescence 354 imaging to quantify Fv/Fm which provides a read-out for the maximum quantum efficiency of 355 Photosystem II. Each inhibitor drastically reduced Fv/Fm compared with controls ( Fig. 5c,d).  None of the three inhibitors affected plasmodesmal frequency at any cell interface in dark-361 grown seedlings (Fig. 5e-g). However, despite cotyledon expansion being unaffected by the 362 inhibitors during de-etiolation (Supporting Information Fig. 5a) plasmodesmal frequencies 363 did not increase significantly in seedlings treated with norflurazon, lincomycin or DCMU (Fig.  364 5e-g, Supporting Information Fig. 5b). In summary, inhibitors that perturbed the etioplast-to-365 chloroplast transition or blocked photosynthetic electron transport, reduced light-induced 366 plasmodesmata formation between mesophyll and bundle sheath cells of C4 G. gynandra. 367 We conclude that chloroplast function, and in particular photosynthetic electron transport, 368 play important roles in controlling the formation of secondary plasmodesmata in the C4 leaf.
The inhibitory effect of DCMU on plasmodesmata formation could be associated with 370 signaling from a dysfunctional photosynthetic electron transport chain, or because less 371 photosynthate is produced. To test the latter hypothesis plants were grown on sucrose 372 during DCMU treatment. No distinguishable effects on phenotype of the seedlings or 373 etioplast-to-chloroplast development were detected (Fig. 6a,b) and provision of sucrose did 374 not rescue the reduction in Fv/Fm caused by DCMU (Fig. 6c,d). However, when we quantified 375 plasmodesmal frequencies in a total of 1655 cell interfaces DCMU-treated seedlings 376 supplemented with sucrose had plasmodesmal frequencies at the mesophyll-bundle sheath 377 interface comparable to those in untreated seedlings (Fig. 6e, p > 0.05) indicating full rescue 378 by sucrose of the DCMU-induced inhibition of plasmodesmata formation. Thus, when 379 photosynthetic electron transport is inhibited, sucrose is sufficient to restore plasmodesmata 380 formation at the mesophyll-bundle sheath cell interface of G. gynandra. 381

Increased plasmodesmata frequency is a conserved C4 trait 383
A critical feature of the C4 pathway is the spatial separation of biochemical processes 384 such that CO2 can be concentrated around RuBisCO. The consequence of this partitioning 385 of photosynthesis is an absolute requirement for the exchange of metabolites between cell 386 types. In C4 grasses this has long been associated with increased plasmodesmal frequency 387 between mesophyll and bundle sheath cells (Evert et al., 1977). Previous work quantified 388 plasmodesmata frequency at the mesophyll-bundle sheath cell interface of G. gynandra and 389 yielded comparable values for plasmodesmata frequencies as in our work (Koteyeva et al.,390 2014), but they did not quantify plasmodesmata in any other cell interface or compared 391 plasmodesmal frequency with related C3 species. Therefore, despite the very different leaf 392 morphology between monocotyledons and dicotyledons, our results reveal that increased 393 plasmodesmal connectivity between mesophyll and bundle sheath cells is likely a conserved 394 trait among C4 plants that separate photosynthesis between two cell types. In G. gynandra, 395 the mesophyll-bundle sheath interfaces had 8-13-fold higher plasmodesmata frequency 396 than those of the closely related C3 species T. hassleriana and A. thaliana (Fig. 1-3). This Compared with C4 grasses, a distinguishing feature of the increased plasmodesmal 414 frequency between mesophyll and bundle sheath cells of G. gynandra is that the increase 415 was not associated with any detectable increase in pitfield area compared with C3 T. hassleriana or C3 A. thaliana (Supporting Information Fig. 2). This suggests that the primary 417 mechanism for increased plasmodesmata numbers in G. gynandra is an increase in pitfields. 418 Since we were not able to visualise individual plasmodesmata within pitfields, we cannot 419 rule out that there is a higher frequency of individual plasmodesmata number within pitfields. 420 However, we consider this is unlikely since pitfield appearance was largely similar between 421 G. gynandra, T. hassleriana and A. thaliana (Supporting Information Fig. 2). In contrast, 422 increased plasmodesmal frequency in C4 grasses was accompanied by increases in pitfield 423 area such that were up to 5 times greater than those in C3 species ( Secondly, the basic structure of bundle sheath cells was already formed in dark grown 455 seedlings, and the formation of plasmodesmata was rapid. Our SEM mapping technique 456 provided sufficient resolution to observe branching in plasmodesmata (Fig. 2,4-6), but 457 interestingly we did not observe any structural differences between the plasmodesmata in 458 different cell interfaces. Although primary and secondary plasmodesmata can be sometimes 459 distinguished by structure, where secondary plasmodesmata are more branched, this is 460 highly dependent on other factors such as leaf age and sink-source transition (Roberts et 461 al., 2001). 462

A role for metabolism and organelles in formation of plasmodesmata 464
Our results suggest that chloroplasts, and more specifically photosynthesis, fuel the 465 formation of secondary plasmodesmata between mesophyll and bundle sheath cells in C4 466 G. gynandra. Inhibition of photosynthesis and chloroplast development through the 467 application of inhibitors greatly reduced plasmodesmata formation during deetiolation but 468 this effect could be rescued by the exogenous supply of sucrose (Fig. 5,6).