Transcriptional regulation of ZIP genes is independent of local zinc status in Brachypodium shoots upon zinc deficiency and resupply

The biological processes underlying zinc homeostasis are targets for genetic improvement of crops to counter human malnutrition. Detailed phenotyping, ionomic, RNA-Seq analyses and flux measurements with 67Zn isotope revealed whole plant molecular events underlying zinc homeostasis upon varying zinc supply and during zinc resupply to starved Brachypodium distachyon (Brachypodium) plants. Although both zinc deficiency and excess hindered Brachypodium growth, accumulation of biomass and micronutrients into roots and shoots differed depending on zinc supply. The zinc resupply dynamics involved 1893 zinc-responsive genes. Multiple ZIP transporter genes and dozens of other genes were rapidly and transiently down-regulated in early stages of zinc resupply, suggesting a transient zinc shock, sensed locally in roots. Notably genes with identical regulation were observed in shoots without zinc accumulation, pointing to root-to-shoot signals mediating whole plant responses to zinc resupply. Molecular events uncovered in the grass model Brachypodium are useful for the improvement of staple monocots.

The zinc effect on root growth of Brachypodium plants (Fig. 1B) differed to shoot responses. Under 209 deficiency, root fresh and dry weight, as well as total root length were reduced by 15.2 to 16.5% (Fig.  210 3A-C). Lateral root length was driving the difference in total root length (Fig. 3D), as it was reduced by 211 17% while primary root lengths were similar under deficiency and control. Zinc-deficient plants had 212 developed one less nodal root (Fig. 3E) and lower total nodal root length (Fig. S1B) than control plants. 213 Zinc excess plants had 22.3 to 38% lower fresh and dry root weight than control and zinc-deficient 214 plants ( Fig. 3A-B), exhibiting stronger responses than the deficiency treatment for these traits. Plants 215 had reduced total root length (Fig. 3C), and slightly (but non-significantly) longer primary roots (Fig.  216   3D). Total length of lateral roots of zinc excess plants was 5.2% lower than control, but 14.2% higher 217 than in zinc-deficient plants (Fig. 3D). Finally, zinc excess fully inhibited nodal root growth ( Fig. 3E and 218 Zinc deficiency and/or excess also affected iron, manganese, copper, calcium and magnesium 227 concentrations in Brachypodium tissues (Fig. S2). Manganese and copper were slightly but significantly 228 less abundant in roots upon deficiency (Fig. S2C,E). Root iron and shoot copper concentrations of zinc-229 deficient plants were 40.7% increased or 25.3% decreased, respectively (Fig. S2A,F). Notably, upon zinc 230 excess, manganese and copper root concentrations were 3.4 and 2.1-fold reduced compared to 231 control, respectively (Fig. S2C,E). Finally, calcium and magnesium were 33% and 35.1% higher in shoots 232 of zinc excess plants, respectively (Fig. S2H,J). 233

Rapid ionome dynamics was observed in Brachypodium roots upon zinc deficiency and resupply 234
During zinc resupply of zinc deficient roots, we observed gradual accumulation of zinc through time. 235 Zinc increase was modest and non-significant after 10 and 30 min, but reached 3.7-fold after 8 h (Fig.  236 5A). In contrast to the roots, shoot zinc had no consistent increase within the 8 h of re-supply (Fig. 5A). 237 Zinc resupply affected the whole ionome ( Fig. 5 and Fig. S3). The root iron concentration rose gradually 238 until 2 h parallel to increased zinc level (Fig. 5B). Copper and manganese root concentrations displayed 239 different dynamics to zinc and iron with higher levels after 10 min, but a severe and transient drop at 240 the 30 min time-point ( Fig. 5C-D). Changes in the shoot ionome were minor (Fig. 5B-D). while resupply samples are progressing with time towards the right along the x-axis (Fig. 6B). 254 Differentially expressed genes (DEG) were identified [adjusted p < 0.05 and log2(fold change) > 1] in a 255 selection of 9 out of 21 possible contrasts between the seven treatments for roots and shoots. The 9 256 contrasts included comparisons of zinc deficiency and excess to the control (2 comparisons), of zinc 257 resupply time-points to deficiency (4 comparisons), and between consecutive resupply time-points (3 12 comparisons) (Fig. 6C). 1215 and 976 unique DEG were identified in roots and shoots, respectively. 298 259 genes were common among roots and shoots, meaning that 1893 unique DEG appeared in 9 contrasts 260 (Data S2). The steady-state responses to zinc deficiency and excess mobilized less DEG than the 261 dynamic response to zinc resupply. There was little overlap in the zinc resupply response between 262 roots and shoots (Fig. 6D). The transcriptional response to zinc resupply was rapid and massive in roots, 263 with the up-or down-regulation of > 450 genes within 10 min (Fig. 6C), with only a small overlap (2.4%) 264 with static deficiency response (Fig. 6D). This latter figure was higher for shoots (9.8%) but differential 265 expression between deficiency and 10 minutes of zinc resupply concerned many genes as well. In roots 266 and shoots, the response to zinc resupply continued to mobilize new genes with time, but slowed 267 down, with a remarkable low number of DEG between the 2-and 8 h time-points ( Over-represented biological processes (BPs, p-value < 0.05) were identified in most contrasts (Fig. 7,  272 Data S3). 273 The "zinc ion transport" BP was strongly overrepresented among DEGs in roots and shoots of deficient 274 plants (0 vs 1.5 µM zinc), with only up-regulated genes, whereas overrepresentation of catabolism, 275 oxidation-reduction and response to chemical processes was observed in response to excess (20 vs 1.5 276 µM zinc) in roots, driven by down-regulated genes only (Fig. 7A). 277 The dynamic response to zinc resupply mobilized many more BPs. In roots, multiple enriched BPs 278 corresponding to up-regulated genes were noticeable (high density red color, Fig. 7A) at the 10 min 279 and/or 30 min time-points compared to deficiency. These BPs were mainly related to signaling, 280 different metabolisms, and stress and hormone responses. "Transcription" as well as other signaling-281 related BPs were enriched only after 10 min resupply. Noticeably, a single enriched BP, "divalent metal 282 transport", corresponded to down-regulated genes at 10 min (blue cell at "10 min vs 0 µM zinc"). This 283 item, as well as "zinc ion transport", was also enriched with down-regulated genes (blue cells) after 2 284 h of zinc resupply compared to deficiency. As expected, the genes corresponding to the "zinc ion 285 transport" BP were strongly up-regulated at deficiency, but were down-regulated within 2 h upon 286 resupply. Finally, a single or no BP were enriched in "30 vs 10 min" and "8 vs 2 h" consecutive time-287 point comparisons, respectively, whereas a shift in the zinc resupply response was observed between 288 30 min and 2 h, with many of the early up-regulated genes being down-regulated in that interval (see 289 the blue cells in the "2 h vs 30 min" comparison, Fig. 7A). 290 In shoots (Fig. 7B, Data S3), the most striking observation was an enrichment of "zinc ion transport", 291 "divalent metal ion transport" and "cation transmembrane transport" BPs, corresponding to down-292 regulated genes within 10 min of resupply, suggesting that a quick transcriptomic regulation of zinc 293 transporter genes preceded zinc re-entry in shoots ( zinc/iron/copper/manganese/cadmium homeostasis/resistance, based Phytozome BLAST annotation 303 (Table 1). As Brachypodium is a relatively new model and metal homeostasis studies on this species 304 are scarce, the majority of these annotations were based on sequence or domain similarities with 305 genes/proteins of other species, especially Arabidopsis and rice. Among these 27 genes, 19 genes were 306 differentially expressed in roots only, 2 in shoots only and 6 in both tissues (Table 1) To confirm the V-shape expression pattern of ZIP genes (Fig. 8), a fully independent experiment with 336 the same design was conducted, except with the exclusion of zinc excess (Experiment 2, Methods). 337 Quantitative RT-PCR was used to profile expression of selected genes: (i) BdZIP4, BdZIP7 and BdZIP13 338 present in RootZIP and ShootZIP clusters, (ii) BdIRT1 and BdHMA1 present in the RootZIP cluster only 339 and (iii) the NAS gene that was not present in either of these clusters. Complete consistency was 340 observed between RNA sequencing and qPCR data for all six genes in root and shoot tissues in 341 deficiency, resupply and control (Fig. 9). 342 The reproducible V-shape expression pattern of ZIP family and 15 other genes in the shoot gene cluster 343 #4 long before zinc influx could be detected in shoots (Fig. 5) was puzzling. A possibility was that a tiny 344 amount of zinc was reaching the shoot tissues rapidly, in an amount lower than the ICP-OES detection 345 limit, and was responsible for local transcriptional regulation for these genes. To enable distinction 346 between zinc still present in shoots after 3 weeks of deficiency (~75 ppm, Fig. 5A) and resupplied zinc, 347 67 Zn, a non-radioactive zinc isotope, was used for resupply (Experiment 3, Methods). Note that 1 and 348 5 h time-points were added to refine the dynamics information. To increase sensitivity and enable 349 detection of zinc isotopes, 67 Zn concentration measurements were obtained using ICP-MS. 350 To ensure that 67 Zn has the same physiological effect as naturally abundant zinc, we first analyzed zinc-351 responsive genes by qPCR (Fig. S8 compared to Fig. 9). The V-shape expression pattern of BdZIP4, 352 Next, 67 Zn accumulation in isotope-labelled samples was examined (Fig. 10, Data S6). In roots, a gradual 361 and significant increase of 67 Zn concentrations was observed with time upon resupply to deficient 362 plants (Fig. 10A, Data S6). The gain in sensitivity compared to Experiment 1 was evident: a significant 363 zinc concentration increase was measured from 10 min (Fig. 10A), when such a change was only 364 detected after 2 h in our initial kinetics (Fig. 5A). In contrast, 67 Zn accumulation in shoots was only 365 detected after 5 h (Fig. 10B). Examining shoot to root 67 Zn ratios confirmed that starting from a higher 366 67 Zn shoot accumulation in deficiency, 67 Zn resupply mostly triggered root accumulation up to 5 h 367 before the ratio stabilized (Fig. 10C).

Zinc deficiency and excess impact growth and development in Brachypodium 377
In shoots, increased leaf number was peculiarly associated with reduced total leaf area, total leaf 378 biomass and dry weight per leaf in zinc-deficient plants (Fig. 2B-D and Fig. S1A). Leaf number is known with unfavorable zinc conditions may prove to be disadvantageous to logging in various crops and thus 405 further increase of yield loss (in addition to the physiological zinc effects). It would therefpre be 406 interesting to look for variation in nodal root allocation in response to zinc among Brachypodium 407 accessions, as was found for water supply (Chochois et al., 2015). 408

Interaction of zinc and other metal homeostasis 409
Zinc excess had no impact on iron root and shoot levels in Brachypodium (Fig. S2A) and no enrichment 410 for iron homeostasis genes was observed in the transcriptomic response to zinc excess (Fig. 7 (Fig. S6). With other ZIPs sharing a similar 428 expression pattern, BdIRT1 may be involved in iron and zinc transport, and be responsible for higher 429 accumulation of iron upon zinc deficiency (Fig. S2A), as well as for the parallel increase of zinc and iron 430 uptake at early time-points upon zinc resupply (Fig. 5A-B). 431 Competition in root uptake between zinc and manganese/copper was possibly regulated by the same 432 (or another set of) ZIP transporters (Fig. S2C,E). In rice and wheat, similar competition was reported regulated by zinc supply: it was slightly more expressed in zinc-deficient shoots compared to control 455 plants and displayed a very flattened V-shape dynamics upon zinc resupply (Fig. S10A). AtbZIP19 and 456 AtbZIP23 are proposed to be specialized in either roots or shoots, respectively (Arsova et al., 2019; 457 Sinclair et al., 2018). BdbZIP9 was more expressed in shoots than roots (Fig. S10A). Interestingly, 458 20 another bZIP gene, Bradi1g29920 (BdbZIP8 in Phytozome v.12.1), was majorly expressed in roots (Fig.  459   S10B) and, although it was not present among initially identified DEG (1.9-fold down-regulation 10 min 460 after resupply, Data S1), it displayed the same V-shape expression pattern as ZIP cluster genes upon 461 zinc resupply, suggesting that BdbZIP8 may be involved in zinc homeostasis in Brachypodium. deficiency (1.6 fold) in roots, then transiently down-regulated upon zinc resupply before peaking at 475 after 8 h (Fig. S10C). This up-regulation may be responsible for zinc re-entry observed in shoots after 5 476 h of resupply (Fig. 10) (Table 1) was up-regulated at the 10 min and 30 min resupply time-points and 481 gradually down-regulated thereafter (Fig. S6, Cluster #7). It may serve as a minimal shoot zinc supplier 482 when the HMA pump is down-regulated (Song et al., 2010). 483 21 In contrast to zinc, copper and manganese concentrations changed quickly upon zinc resupply. Both 484 metals experienced an increase at 10 min and then a decrease at 30 min, the inverse of the V-shape of 485 ZIP clusters in root and shoot, although it was only significant in root (Fig. 5C-D). OsNRAMP5 is 486 suggested to function in manganese distribution from root into shoot (Yang et al., 2014). The zinc-487 responsive NRAMP gene (homolog of OsNRAMP6, Bradi1g53150) may serve the same function in 488 Brachypodium. Its severe induction at 10 min time-point and with excess zinc, where manganese 489 concentration is lowered (Fig. S10D) can support its role in manganese root-to-shoot translocation. On 490 the other hand, OsATX1, homolog of ATOX1-related copper chaperone, was reported to have an 491 important role in root-to-shoot copper translocation (Zhang et al., 2018) and to interact with multiple 492 rice HMA pumps, probably to transfer copper to these pumps. There are seven ATOX1-related genes 493 in the metal list, some of which were immediately regulated by zinc resupply (Clusters #4 in Fig. S6, 494 cluster #8 in Fig. S7). Rapid induction of the NRAMP gene and several ATOX1-related genes (Fig. S5), in 495 contrast to the late induction of AtHMA4 homolog (Bradi1g34140), might explain the efficient 496 regulation of manganese and copper concentration in shoot, compared to zinc. 497

Zinc shock appears to be the first transcriptomics response upon Zn resupply to deficient roots 498
Expression patterns of the root ZIP cluster genes ( Fig. 8 and 9) were in stark contrast to observations 499 made in Arabidopsis. In Brachypodium, genes within this cluster were highly expressed at zinc 500 deficiency, rapidly down-regulated after 10 min resupply, then up again after 30 min, thus displaying 501 a V-shape expression pattern (Fig. 8 and 9). This response occurred in roots as zinc concentration was 502 steadily increasing upon resupply (Fig. 5A and 10A). In Arabidopsis was observed an initial up-503 regulation in roots of multiple metal homeostasis genes and proteins, including ZIPs, after 10 min of 504 resupply of zinc-starved plant before a down-regulation from 30 min (Arsova et al., 2019). 505 The V-shape expression pattern of the ZIP cluster genes in roots implies that zinc influx into roots of 506 zinc-starved plants is sensed as a zinc stress, similar to a zinc excess. This sensing then initiates within 507 10 minutes down-regulation of zinc uptake genes in roots. Such zinc shock response was described in 508 22 the yeast Saccharomyces cerevisiae (MacDiarmid et al., 2003;Simm et al., 2007). Thereafter, upon 509 sensing yet below-sufficient zinc levels in root tissues, ZIP genes are re-up-regulated at 30 minutes 510 followed by more classical down-regulation with increasing zinc concentrations in tissues at later time-511 points. The response to zinc resupply in roots therefore occurs in two phases (Fig. 6A,D), an initial and 512 rapid phase (10-30 minutes) combining zinc shock response as well as zinc reuptake supported by 513 intense signaling (Fig. 7A), and a later phase (2-8 hours) which corresponds to a slow return to a 514 sufficient state. Although they display very different dynamics, two phases are also observed in 515 response to zinc resupply in Arabidopsis (Arsova et al., 2019). 516 517

Early transcriptomic response of zinc transporter genes in shoots mirrors the root pattern and is 518
independent of local zinc concentration 519 Strikingly, shoot ZIP cluster genes ( Fig. 8 and 9) displayed a V-shape expression pattern as in roots ( transporter genes in shoots appears to be independent of local zinc concentration and to be 524 coordinated with roots in Brachypodium and we propose that zinc re-entry in roots initiates a root-to-525 shoot signaling that instigates a distant transcriptomic response (Fig. 11). 526 Shoot transcriptome response, independent from shoot nutrient concentration, was reported upon 527 nitrogen resupply to nitrogen-starved maize plants (Takei et al., 2002). In roots, several signaling-528 related BPs were enriched at 10 min and 30 min time-points (Fig. 7A, dense red area, up-regulated 529 genes), while this response was delayed in shoots where metal transport (Fig. 7B, in blue, down-530 regulated genes) was among the few enriched BPs after 10 min (Fig. 7B). It is therefore tempting to 531 speculate that the root-to-shoot signaling directly represses expression of metal transporter genes in 532 shoot, rather than activating local signaling pathways in shoot. Supporting this idea is that the "RNA 533 23 metabolism" BP was also enriched (Fig. 7B, in red, up-regulated genes), 10 min after resupply in shoots. 534 Several transcription factors were found in this enriched BP, and belong to different superfamilies such 535 as B3, AP2, WRKY and bZIP (Bradi4g02570, Data S3). These TFs may potentially regulate ZIP genes in 536 Brachypodium shoots upon zinc resupply. 537 Long-distance signaling mechanisms known in plants include electric or hydraulic signaling, calcium 538 waves propagated by calcium-dependent protein kinases and calmoduline proteins, ROS waves, sugar 539 signaling, hormonal signaling and mobile mRNA (Shabala et al., 2016). Among the signaling-related 540 DEG and enriched BPs at 10 min in root, multiple genes connected to these processes are present and 541 constitute candidates for producing root-to-shoot signals (Data S5). 542 Long distance or systemic signaling is known to contribute to metal homeostasis regulation. It was 543 suggested that AtMTP2 and AtHMA2 transcript levels in roots are regulated by shoot zinc 544 concentration in Arabidopsis, in contrast to ZIP genes being controlled by the local zinc status (Sinclair 545 et al., 2018). Designing an experiment testing our model of root-to-shoot signaling upon nutrient 546 resupply is a challenge. Split-root experiments were successful to disentangle local versus systemic 547 signals regulating the response to iron deficiency (Schikora & Schmidt, 2001;Vert et al., 2003;Wang 548 et al., 2007). To study systemic shoot-to-root signaling, reciprocal grafting of mutant and wild-type 549 roots and shoots, and foliar nutrient supply are popular methods (Sinclair et al., 2018;Tsutsui et al., 550 2020). Conversely, treating half of the root system with deficient medium allows detecting a shoot 551 deficiency response while still sufficiently supplied by the other half of the root, and thus characterizing 552 a root-to-shoot deficiency signal. In the case of a long-distance signal triggered upon resupply of 553 deficient plants, it is delicate to distinguish signaling from delayed nutrient flux in such experimental 554 setups and alternative approaches will need to be designed to identify the putative signal. 555 In summary, our study revealed the complexity of the zinc homeostasis network in Brachypodium by 556 comparing static and dynamic responses to zinc supply. We identified a short-lived zinc shock response 557 to zinc resupply in roots and hypothetical long-distance zinc signaling that could be important in 558 24 realistic field resupply conditions. The study also showed that Brachypodium responds phenotypically 559 and genetically to changes in zinc supply, and represents a valuable model of staple grass crops to 560 examine zinc homeostasis that contrasts with the widely studied model Arabidopsis. Differences in 561 zinc/iron interactions and in dynamics of transcriptional changes upon zinc resupply reveal the 562 diversity of zinc homeostasis mechanisms among plant species.