Perturbation of protein homeostasis brings plastids at the crossroad between repair and dismantling

The chloroplast proteome is a dynamic mosaic of plastid- and nuclear-encoded proteins. Plastid protein homeostasis is maintained through the balance between de novo synthesis and proteolysis. Intracellular communication pathways, including the plastid-to-nucleus signalling and the protein homeostasis machinery, made of stromal chaperones and proteases, shape chloroplast proteome based on developmental and physiological needs. However, the maintenance of fully functional chloroplasts is costly and under specific stress conditions the degradation of damaged chloroplasts is essential to the maintenance of a healthy population of photosynthesising organelles while promoting nutrient redistribution to sink tissues. In this work, we have addressed this complex regulatory chloroplast- quality-control pathway by modulating the expression of two nuclear genes encoding plastid ribosomal proteins PRPS1 and PRPL4. By transcriptomics, proteomics and transmission electron microscopy analyses, we show that the increased expression of PRPS1 gene leads to chloroplast degradation and early flowering, as an escape strategy from stress. On the contrary, the overaccumulation of PRPL4 protein is kept under control by increasing the amount of plastid chaperones and components of the unfolded protein response (cpUPR) regulatory mechanism. This study advances our understanding of molecular mechanisms underlying chloroplast retrograde communication and provides new insight into cellular responses to impaired plastid protein homeostasis.

. Wild-type and mutant seeds were grown on soil in climate Whole-mount preparation and optical microscopy 151 To analyse defects in embryo development, siliques of Col-0 and heterozygous PRPS1/prps1-2 plants 152 were manually dissected and cleared as reported in Tadini et al., 2018. Developing seeds were 153 observed using a Zeiss Axiophot D1 microscope equipped with differential interface contrast optics. 154 Images were documented with an Axiocam MRc5 camera (Zeiss).

178
In vivo Translation Assay 179 The in vivo translation assay was performed essentially as previously described .  Isolation of PRPS1-containing protein complexes 190 The isolation of PRPS1-containing complexes was performed according to previous works (Barkan,191 1998; Merendino et al., 2003). 100 mg of leaf fresh weight were ground in liquid nitrogen and codon (+ 4 bp from ATG, Fig.1 A, B). This event disrupts the PRPS1 reading frame and introduces   (Fig. 2 A). Interestingly, oePRPS1 seedlings were incapable of over-accumulating the PRPS1 313 protein ( Fig. 2 B), showing an accumulation level lower than the one observed in prps1-1 leaves, 314 despite the PRPS1 transcript level having been about two-fold the Col-0 control leaves (Fig. 2  and activity, as shown by oePRPL4 lines indistinguishable from Col-0 ( Fig. 2 A, B, C).

319
In order to investigate this aspect further, PRPS1 and PRPL4 coding sequences were cloned 320 into the pOp/LhG4 vector, which allows the inducible over-expression of the two genes once the 16 days (Fig 2 A). Conversely, when the growth medium was supplemented with 2µM DEX, 326 indPRPS1 seedlings showed a leaf virescent phenotype and a drop in photosynthetic performance, 327 together with a reduced accumulation of PRPS1 protein resembling the phenotype of prps1-1 and 328 oePRPS1 lines (Fig 2 A, B). This was despite PRPS1 transcripts accumulation to levels higher than 329 Col-0 control leaves (Fig. 2 C). On the other hand, the inducible overexpression of PRPL4 resulted 330 in an increased accumulation of PRPL4 transcripts and protein, similar to oePRPL4 seedlings ( Fig. 2 331 B, C), without any impact on chloroplast activity and leaf greening ( Fig. 2 A).

332
To understand how chloroplasts ultrastructure organization is affected by the increased  The over-accumulation of Arabidopsis PRPS1 protein inhibits Escherichia coli cell growth 392 Similarly to Arabidopsis, the depletion of the ribosomal protein S1 (Rps1) in E. coli cells leads to 393 lethality (Kitakawa & Isono, 1982). Moreover, the down-regulation as well as the over-accumulation acid sequence identified three S1 domains ( Fig. 5 A), in agreement with early analyses of spinach S1 401 protein (Franzetti et al., 1992). In addition, PRPS1 protein shows a high degree of identity (i.e. about 402 50%) with the S1 ribosomal proteins from cyanobacteria, which possess three S1 domains as well 403 (Sugita et al., 1995;Salah et al., 2009). In order to investigate the possible relationships between the 404 three S1 domains identified in PRPS1 and the six S1 domains in E. coli S1, the amino acidic sequences 405 of each S1 domain were aligned and clustered in a phylogenetic tree (Fig. S3). The resulting tree 406 showed that domains 1 and 2 of PRPS1 are more similar to the corresponding domains of S1, while 407 the PRPS1 domain 3 clusters together with the domains 3, 4 and 5 of S1.

408
The possible functional homology between the two proteins was then investigated by 409 introducing the PRPS1 gene in E. coli cells, for over-expression and complementation assays. To 410 experimentally test the ability of PRPS1 to complement S1 functions and to repress cell growth when 411 over-accumulated in bacterial cells, both rpsA (encoding the S1 protein) and PRPS1 coding sequences 412 were cloned into pQE31-pREP4 plasmid system under the control of pT5-lacO promoter in pQE31. inhibited due to the excessive amount of S1 protein (Fig 5 B, C). Strikingly, the over-accumulation 434 of PRPS1 protein was effectively repressing the bacterial growth and led to S1 protein depletion, too 435 ( Fig. 5 B, C). It is worth noting, that the overaccumulation of PRPS1 Arabidopsis protein was able to 436 repress the growth of E. coli cells, comparably to S1 overaccumulation, even when the overexpression 437 of either rpsA or PRPS1 genes was achieved in the C-1a E. coli strain, devoid of the conditional 438 depletion system of the endogenous S1 protein (Fig. S4 A). To better understand whether PRPS1 immunoblotting to detect either S1 or PRPS1 protein localisation. Interestingly, PRPS1 was retrieved 443 in both ribosome-bound and -unbound fractions, similarly to S1 from E. coli, suggesting that the 444 Arabidopsis PRPS1 can compete for the ribosome core with E. coli S1 protein (Fig. S4 B). Taken 445 together, these data indicate that PRPS1 protein is able to inhibit E. coli growth when over-expressed 446 in addition to the endogenous S1, whereas it is unable to functionally replace the E. coli endogenous 447 S1 protein. by the inhibitory role of PRPS1 protein over-accumulation on E. coli cell growth ( Fig. 5 and Fig. S4).

453
In order to investigate the molecular mechanism responsible for controlling PRPS1 protein infiltration, unlike the control sample (Fig. 6 A). Intriguingly, the most relevant chloroplast stromal  Table S2), obtained by comparing data from 486 indPRPS1-DEX with indPRPS1+DEX, and indPRPL4-DEX with indPRPL4+DEX (Fig. 7 A). unique DEGs in each group (Fig. 7 B, see also indPRPS1, 3 were upregulated in indPRPL4 (Table S2 G).  Table   506 S3). On the other hand, the repressed biological functions found in indPRPS1 were related to the 507 production of secondary metabolites, defence against herbivores and oxidative stress such as 508 "glucosinolate biosynthetic process" (GO:0019761), "sulfur compound biosynthetic process" 509 (GO:0044272), "cell redox homeostasis" (GO:0045454) and "response to wounding" (GO:0009611), 510 to cite a few of them (Fig. 8 B; see also Table S4). Interestingly, the high enrichment in GO terms comparable to prps1-1 behaviour (Fig 9 A, B).  dexamethasone-mediated induction led to significant changes also at proteomic level (Fig. S8).

566
The significantly enriched GO terms retrieved by analysing the up-regulated proteins detected in 567 indPRPL4 samples were mainly associated with mitochondrial electron transport chain and 568 mitochondrial ribosomes (Table S9). Moreover, GO annotation revealed that several DAPs in 569 indPRPL4 seedlings belong to "response to stress" (GO:0006950), as also detected by the 570 transcriptome analysis (Fig. 11 C). The GO annotation tool allowed to locate DAPs in indPRPL4 571 plants especially in the nucleus, chloroplast and cytoplasm (Fig. 11 D). Overall, our results, in Intriguingly, the reduced accumulation of S1 protein can also be obtained when PRPS1 gene 600 expression, and the consequent transcript accumulation, is both constitutively increased in oePRPS1 601 and indPRPS1 + DEX seedlings (Fig. 2) and induced in indPRPS1 leaf discs for 24 hours (Fig. 4). -DEX controls (Fig. 2, Fig. 4).  and PRPS5, was reduced, as observed in indPRPS1 + DEX leaf discs (Fig. 4), indicating that the 638 decreased abundance of PRPS1 protein has a deleterious effect on plastid ribosome stability. In 639 agreement with that, the pulse-labelling experiment conducted on indPRPS1 leaf material infiltrated 640 with DEX for 6 hours showed a severe inhibition of plastid protein translation and a larger fraction 641 of PRPS1 protein freely associated to mRNA and not bound to actively translating ribosomes (Fig.   642 4). These findings explain the defects in leaf greening observed in the different Arabidopsis lines 643 ( Fig. 1 and Fig. 2) and support the role of PRPS1 as a stringently regulated translation factor rather 644 23 than a "real" ribosomal protein, given its weakly and reversible association with the 30S subunit, 645 similarly to previous observations in E. coli (Delvillani et al., 2011). 646 S1 protein is the closest PRPS1 homologue in E. coli, and it is encoded by the essential gene 647 rpsA (Kitakawa & Isono, 1982). To gain possible insights on PRPS1 function, we attempted to rescue 648 E. coli cell lethality due to S1 depletion and to phenocopy the bacteriostatic effects of S1 649 overaccumulation, by modulating the level of PRPS1 protein in E. coli cells (Fig. 5). Our data clearly 650 showed that PRPS1 is not able to functionally replace the endogenous S1, as cells depleted of rpsA 651 but with moderate amount of PRPS1, failed to grow (Fig. 5). According to previous studies, S1 652 protein exerts its functions in relationship with the specializations of its six S1 domains (Salah et al.,653 2009). The interactions with the ribosome relies on S1 domains 1 and 2 (Giorginis & Subramanian, 654 1980), whereas the ability to bind mRNAs has been associated with the S1 domains 3, 4 and 5 655 (Subramanian, 1983). As for domain 6, if removed together with S1 domain 5, the initiation of

658
According to our in silico analysis, S1 domains 1 and 2 of PRPS1 appeared to be more similar to the 659 respective domains 1 and 2 of S1, while domain 3 clustered together with domains 3, 4 and 5 of S1 660 (Fig. S3). The fact that PRPS1 is a smaller protein and possesses three out of the six S1 domains 661 identified in S1, of which none could be associated with those required for translation initiation in E. 662 coli, could explain the inability of PRPS1 to functionally replace S1 protein. Nonetheless, PRPS1 663 overaccumulation blocked cells growth as much as S1 (Fig. 5, Fig. S4). It has been shown that S1 664 over accumulation inhibits translation since the excess of "free" S1 interacts with mRNAs, unfruitful. In this scenario PRPS1 would be able to inhibit E. coli growth by competing with the 669 endogenous S1 protein and inhibiting ribosome activity (Fig. S4). Nevertheless, PRPS1 is likely 670 capable of binding E. coli mRNAs with the S1 domains 2 and 3, since E. coli and plastid mRNAs concomitantly leads to the decreased accumulation of S1 protein over time (Fig. 5 C), mimicking the 676 role of S1 as feedback effector of its own regulation at the translational level (Skouv et al., 1990).

677
Such regulatory mechanism has a different spatial constraint in photosynthetic eukaryotes translational regulatory mechanism, in response to PRPS1 overexpression, mediated by the plastid 684 proteostasis machinery (Fig. 6). In chloroplasts, the major soluble stromal protease is the CLP

709
On the other hand, the constitutive over-accumulation of PRPL4 protein upon induction didn't 710 affect neither the plastid translation nor the chloroplast ultrastructure (Fig. 2, Fig. 3, Fig. 4).

711
Interestingly, chaperones both resident in the cytoplasm and in plastids were progressively up-712 regulated as PRPL4 levels increased, indicating the activation of protein homeostasis mechanisms to 713 cope with the increased protein amount (Fig. S7). These different responses to the overexpression of 714 two plastid ribosomal proteins could be due to their intrinsic features and activities, having PRPS1 hormones, such as "photoperiodism, flowering", "vegetative to reproductive phase transition of 739 meristem" and "cellular response to hormone stimulus" (Fig. 8 and Table S3), and by the concomitant 740 repression, amongst others, of genes involved in "cell redox homeostasis", "response to wounding", 741 "response to oxidative stress", "response to water deprivation". Accordingly, the circadian clock In contrast, the results obtained from the transcriptomic and proteomic analyses performed on 755 indPRPL4 + DEX leaf discs showed markedly different results ( Fig. 7-8, Fig. 10-11). The 756 overaccumulation of PRPL4 promoted abiotic stress responses and protein homeostasis such as 757 "response to hydrogen peroxide", "response to high light intensity", "response to heat", "response to 758 oxidative stress" and "protein folding" (Fig. 8 C).  The data that support the findings of this study are openly available in Gene Expression Omnibus at

802
The authors declare no conflicts of interest.