HetL provides immunity to HetR against PatS inhibition, and promotes pattern formation in the cyanobacterium Nostoc PCC 7120

Local activation and long-range inhibition are mechanisms conserved in self-organizing systems leading to biological patterns. A number of them involve the production by the developing cell of an inhibitory morphogen, but how this cell gets immune to self-inhibition is rather unknown. Under combined nitrogen starvation, the multicellular cyanobacterium Nostoc PCC 7120 develops nitrogen-fixing heterocysts with a pattern of a heterocyst every 10-12 vegetative cells. Cell differentiation is regulated by HetR which activates the synthesis of its own inhibitory morphogen (PatS), which diffusion establishes the differentiation pattern. Here we show that HetR interacts with HetL at the same interface as PatS, and that this interaction is required to suppress inhibition and to differentiate heterocysts. hetL expression is induced under nitrogen-starvation and is activated by HetR, suggesting that HetL provides immunity to the heterocyst. This protective mechanism might be conserved in other differentiating cyanobacteria as HetL homologues are spread across the phylum.


100
The HetR/PatS regulatory loop fits the local activation (HetR)/long range inhibition (PatS) module that 101 characterizes the Turing-adapted model of patterning (Brown & Rutenberg, 2014; A. Turing, 1952;A. 102 M. Turing, 1990) (Gierer & Meinhardt, 1972) (Figure 1A). In addition, the action of HetN and PatX as 103 However when produced in cells that have already initiated development (proheterocysts), PatS-5 is not 116 able to inhibit differentiation (Yoon & Golden, 2001) (Wu, Liu, Lee, & Golden, 2004). These 117 observations suggest that proheterocysts must acquire additional protection post-PatS export/processing. 118 The hetL gene (all3740) was unearth in a genetic screen aiming to identify factors involved in PatS 119 signaling (D. Liu & Golden, 2002). HetL is a single domain protein composed of 40 pentapeptides 120 (A(D/N)LXX), adopting a right-handed quadrilateral beta helix typical of an Rfr-fold common to all 121 pentapeptide repeats containing proteins (PRPs) (Ni, Sheldrick, Benning, & Kennedy, 2009). The 122 ectopic expression of hetL in a background of patS overexpression restores the ability of the strain to differentiate heterocysts (D. Liu & Golden, 2002). hetL overexpression stimulates differentiation also 124 when PatS-5 is added to the culture medium (D. Liu & Golden, 2002). Henceforward, HetL interferes 125 with PatS inhibition but the molecular mechanism involved is unknown. 126 This study aims to further explore the function of HetL in PatS signaling. We show that HetL interacts 127 with HetR at the same interface than PatS. This interaction is needed for HetR to escape PatS inhibition, 128 and does not inhibit its DNA-binding activity. Analyzing hetL transcription, we found that this gene is 129 induced shortly after initiation of cell differentiation and that HetR is required for its expression. Finally, 130 we show that the expression of hetL in heterocysts, but not in vegetative cells, is necessary to counteract 131 PatS inhibitory effect. We conclude that, by interacting with HetR, HetL interferes with PatS fixation 132 and therefore provides immunity to the developing cell.

136
HetL interacts with HetR without inhibiting its DNA binding activity 137 To get further insights into HetL function, we wondered whether its activity would be mediated by its 138 direct interaction with HetR. To test this, we used the bacterial two hybrid assay (BACTH), which is 139 based on the reconstitution of adenylate cyclase (CyA) activity by two interacting proteins that bring the 140 T18 and T25 domains of CyA into close proximity (Karimova, Pidoux, Ullmann, & Ladant, 1998). T18 141 and T25 were fused to N-terminus coding sequences of HetR and HetL, and the dimerization ability of 142 HetR was used as an internal control for this assay. The data of Figure 1B show that HetL displays a 143 strong interaction with HetR. Interestingly, it seems that HetR-hood domain is sufficient to mediate the 144 interaction of HetR to HetL. In addition, this experiment indicated that HetL is able to form dimers (or 145 oligomers) ( Figure 1B). 146 To validate this interaction, we developed a BioLayer interferometry (BLi) assay. For this purpose, HetR 147 and HetL proteins were produced and purified using affinity chromatography (Supplementary Figure  148   1A). HetR was biotinylated and immobilized on streptavidin biosensors as the ligand, while HetL was 149 used as the analyte. Upon addition of HetL, a concentration-dependent association was recorded and 150 decreased during the dissociation step corresponding to the washing of the sensor, testifying for a direct 151 interaction between HetL and HetR ( Figure 1C). The estimated dissociation constant (KD) of the HetR-152 HetL interaction was 6 µM. The interaction between HetR and HetL observed in the BACTH assay was 153 hence confirmed by BLi. As HetR acts by directly binding to promoters of a subset of its target genes, we tested if the interaction 155 with HetL would impact its DNA binding activity. To this end, we conducted electrophoretic mobility 156 shift assay (EMSA) using the hetP promoter as a target (Huang et al., 2004) (Hu et al., 2015). The 157 previously reported ability of PatS-5 to inhibit HetR DNA binding activity was used as a control (Huang 158 et al., 2004) (Hu et al., 2015). In the presence of HetL, HetR was still able to interact with hetP promoter 159 and the complex formed was higher than the one formed by HetR alone ( Figure 1D). This result 160 indicates that, contrary to PatS-5 binding, the interaction between the two proteins does not inhibit the 161 activity of HetR. Heterocysts are presented in green. Vegetative cells in brown. various concentrations of HetL at 2.5, 5, 10 and 20 µM followed by a 120 second dissociation step. Each 177 curve represents the average of two experiments minus the control experiment. As a negative binding 178 control, HetL 20µM was added to the empty biosensor devoid of HetR.

179
(D) EMSA assay of HetR (1 µM) with hetP promoter (50 nM) in the presence of or PatS-5 or PatS-6 (1 180 µM) and HetL (4 µM). The hetP promoter incubated alone served as negative control (free DNA) and 181 HetL plus DNA as a specific control for HetR. 182 183 184 HetL and PatS interact with HetR at the same interface 185 Since HetL has been identified on the basis of counteracting PatS inhibition and as both PatS and HetL 186 interact with HetR at its hood domain, we hypothesized that HetL could interact with HetR at the same 187 interface than PatS, which would therefore interfere with its inhibiting action. To test this hypothesis, 188 we took advantage of the fact that HetR-PatS interaction involves the Hood domain of HetR and that 189 the residue R223 of HetR is required for this interaction. Interestingly, a variant of HetR bearing a 190 R223W substitution lost the ability to interact with HetL (Figure 2A). HetR (R223W) was still able to 191 form dimers (Figure 2A), which indicates that this variant is correctly folded. We conclude that the 192 absence of interaction between HetR (R223W) and HetL indicates that this residue, in addition to be 193  To gain further information about HetR:HetL interaction, we performed a site directed mutagenesis to 207 substitute the residue D151 from HetL to Alanine. Interestingly, this substitution impaired the binding 208 of HetL to HetR as revealed by BACTH assay (Figure 2A). Furthermore, HetL D151A variant was still 209 able to form dimers which is an important indication to rule out the possibility that the mutation 210 destroyed the fold of this variant (Figure 2A). Taken together, these results indicate that HetL and PatS-211 5 interact with HetR at the same interface, which implies that HetL and PatS must compete for the 212 interaction with HetR. To check this assumption, the interaction between HetR and HetL was analyzed 213 by BLi in the presence of increasing amounts of PatS-5. The data obtained clearly indicated that the 214 association between HetR and HetL is impaired from the addition of PatS-5 in a dose-manner response, 215 with a total inhibition effect obtained at a concentration of 5 µM ( Figure 2C). Similar results were 216 obtained with PatS-6 (supplementary figure 2A). 217 The effect of PatS on HetR-HetL interaction was further analyzed by two hybrid assays. For this, a 218 synthetic operon constituted of T18-hetL and patS-6 was constructed and used to question the interaction 219 with HetR. While adenylate cyclase activity was restored in bacteria producing T25-HetR and T18-HetL 220 fusions, the interaction between HetR and HetL was abolished when PatS-6 was produced 221 Figure 1B), which is a further demonstration of the fact that PatS and HetL share the 222 same interaction within HetR. The interference of PatS with the HetR-HetL complex was also observed 223

(Supplementary
in EMSA experiments where the addition of increasing amounts of PatS-5 abolished the binding of 224 HetR-HetL to the hetP promoter ( Figure 2D). Altogether, these data support the hypothesis of a 225 competition between PatS and HetL for the interaction with HetR. Note that, to distinguish the HetL/HetR-DNA and HetR-DNA complexes the concentration of HetL 242 used in this experiment was lower than that shown in Figure 1. The interaction between HetL and HetR is required to HetR function in vivo 268 As hetL was discovered on the basis of its capacity, when overexpressed, to suppress the inhibitory 269 effect of PatS, we used this approach to evaluate the impact of the interaction between HetR and HetL  Figure 4B). HetL overproduction also allowed the ability of the strain to form heterocysts when PatS-281 5 was added in the culture (Supplementary figure 2). The percentage of the heterocysts formed in the 282 hetL overexpressing strain was equal to that of the wild type strain (Table 1). Interestingly, the 283 recombinant strain overproducing HetL (D151A) was not able to form heterocysts when PatS was 284 overproduced or upon addition of PatS-5 in the growth medium ( Figure 4B, Supplementary figure 2). 285 The interaction of HetL with HetR is therefore needed to counteract PatS inhibition. 286 Golden, 2002). We therefore chose quantitative RT-PCR approach to analyze hetL transcription during 299 the differentiation process. The hetP gene whose transcription is activated by HetR 8 hours after nitrogen 300 step-down was used as an internal control for this experiment (Mitschke et al., 2011). RNAs were 301 collected from the wild type strain and the hetR mutant at different times after nitrogen starvation and 302 the transcript levels of the two genes were expressed relatively to their amount at time zero. Results 303 reveal that hetP expression was, as expected, strongly induced starting from 8 hours after nitrate 304 depletion. Contrary to the 10-fold induction in the wild type strain, the expression of hetP did not 305 significantly increase in the hetR strain which is consistent with the activation of this gene by HetR 306 (Figure 5A). The hetL gene showed a similar transcription profile to that of hetP, yet its expression 307 level was much lower. In the wild type strain, a 3.5-fold increase of hetL transcripts was observed 8 308 hours after nitrogen step-down and was maintained up to 24 hours. In the hetR mutant, no induction present in the hetL promoter. We concluded that, even if low, the transcription of hetL gene is induced 315 early during the differentiation program and that HetR is required for hetL activation but its action is 316 likely indirect and mediated by another factor. 317 From the results presented above, it can be deduced that HetL acts in the heterocyst. To further confirm 318 this assumption, we expressed hetL either from the rbcL promoter, which is specific to vegetative cells, 319 or from the patS promoter which is expressed in the heterocysts early after nitrogen stepdown. The 320 ability of HetL to suppress heterocyst inhibition triggered by the addition of PatS-5 was analyzed. 321 Results show that the strain expressing the PpatS-hetL gene was able to form heterocysts even in the 322 presence of PatS-5, but when expressed from the rbcL promoter, hetL was unable to prevent the 323 inhibitory effect of PatS-6 ( Figure 5B). This result is in favor of HetL acting specifically in the 324 heterocyst. HetL ( Figure 6A). In addition, the strain expressing the PpetE-hetL gene was able to form heterocysts 340 even in the presence of the HRGTGR peptide (Figure 6B), while the overproduction of HetL (D151A) 341 did not allow to bypass PatX-6 inhibition since heterocysts were not observed ( Figure 6B). From these 342 results, we conclude that HetL provides immunity to the developing cells against the two inhibitory 343 peptides involved in pattern formation. 344 the others, such HetL, are integrally formed by PR domains (Supplementary Table 1). Given the high 360 sequence identity shared by these PRPs (32% in average), predicting functional specificity based on 361 sequence similarity is not possible. Because hetL mutant does not show any specific differentiation 362 phenotype (D. Liu & Golden, 2002), we wondered if this could be due to a cross-complementation with 363 another PRP. In this regard, we analyzed the capacity of HetR to interact with some HetL homologs. 364 For this, we chose All3256 and All4303 because they share closest features with HetL. They have a 365 similar size (237, 268, 213 amino acids respectively), a similar organization of the PRs domains, and 366 the three are predicted to be cytosolic ( Figure 7A). Figure 7B shows the results of a BACTH assay 367 questioning the putative interaction of these proteins with HetR. Only All4303 displayed interaction 368 with HetR, and even if this interaction is two-fold weaker than that of HetR-HetL it is significant 369 compared to the negative control. This experiment indicates that at least one among the 31 PRPs is able 370 to interact with HetR, which makes a cross-complementation of the hetL mutation with all4303, or 371 another PRP coding gene a possible scenario. functions from previous valuable studies that are fundamental to our investigation. In particular, as the 397 unique structural fold of HetR with its two exposed domains (flap and hood) was proposed to favor 398 protein-protein interactions, we questioned the possible interaction of HetR with proteins involved in 399 patterning. In this context, HetL was found to interact strongly with HetR without abolishing its DNA-400 binding activity, which suggests that HetR-HetL complex may be active regarding gene regulation in 401 vivo (Figure 1). A possible mechanism to explain the role of HetL in PatS signaling is the titration by 402 HetL of the inhibiting peptide. This hypothesis was ruled out since ITC assay did not show any 403 interaction between HetL and PatS (Figure 3). Alternatively, a site-exclusion mechanism can be 404 proposed for HetL. The observation that HetL and PatS-5 (or PatS-6) interact with HetR at the same 405 interface is in agreement with this suggestion (Figure 2B). In Bli assays, increasing concentrations of 406

PatS-5 interfered with HetR-HetL interaction, and in EMSA experiments addition of PatS-5 abolished 407
the formation of HetR-HetL complex in a concentration-dependent manner (Figure 2C-D). In addition 408 to confirming the docking model predicting a same HetR-interaction interface for PatS and HetL, these 409 data imply that the concentration of HetL in the (pro)heterocyst must be higher than that of PatS peptide, 410 which is plausible because (i) PatS peptide is diffusible (Wu et al., 2004;Yoon & Golden, 2001), (ii) 411 HetL features do not include any motif or domain that predicts a putative translocation/secretion or 412 association with membranes that could decrease its amount in the producing cell. Moreover, as hetL 413 transcription is activated by HetR (Figure 4), HetL protein is likely to accumulate in the cytoplasm of 414 differentiating cells. 415

416
Alternatively, it can be assumed that in the (pro)heterocyst the affinity of HetR for HetL must be higher 417 than that for PatS. Bli and ITC experiments showed that in vitro, the affinity of HetR to PatS is 10-fold 418 higher than for HetL, (compare data of Figure 2C and Figure 3). One might speculate that in vivo, and 419 especially in the (pro)heterocyst, the affinity of HetR for HetL increases either due to a modification of 420 HetR or to its interaction with another factor. HetR has been shown to be regulated by phosphorylation 421 with HetR, are rather in favor of functional redundancy. A global transcription study using RNA 434 sequencing has shown that all4303 transcription is induced in response to nitrogen starvation (Flaherty,435 Van Nieuwerburgh, Head, & Golden, 2011). The hetL gene has also been reported in this study to be 436 induced after nitrogen starvation, which comforts our data. The other PRPs coding genes whose 437 transcription has also been reported to be regulated in this condition are reported in Supplementary 438 Table 1. At this stage of our investigation, we cannot rule out that homologs of HetL might also be part The presence of heterocysts was confirmed by microscopy. 500 Conjugation of Nostoc was performed as described in reference (Cai & Wolk, 1990). Briefly, E. coli 501 strains (bearing the replicative plasmid and the RP-4 conjugative plasmid) grown to exponential growth 502 phase, were mixed to an exponentially grown Nostoc culture. The mixture was plated on BG11 plates 503 and antibiotics were added 24 hours later for plasmid selection. 504

Plasmid construction: 505
All the plasmids used in this study are listed in Supplementary file 1 506 The strategy used for plasmid construction is briefly described below. All the recombinant plasmids 507 were analyzed by sequencing. 508 The open reading frame of hetR gene (alr2339) was amplified using the hetR dh fw T25 and hetR dh rv 510 primers and cloned into the PstI and EcoRI restriction sites of the pKT25 expression plasmid. 511 The open reading frame of hetL gene (all3740) was amplified using the hetL dh fw T25 and hetL dh rv 513 primers and cloned into the PstI and EcoRI restriction sites of the pKT25 expression plasmid. 514 pXX3: pKT25-hetLD151A 515 The pXX2 plasmid was used as template to substitute the D151 residue of HetL to Alanine. For this the 516 primers Mut hetLD151A fw and Mut hetLD151A rv were used as primers in a megapriming PCR assay. 517 The open reading frame of hetR gene (alr2339) was amplified using the hetR dh fw T18 and hetR dh rv 519 primers and cloned into the PstI and EcoRI restriction sites of the pUT18C expression plasmid. 520 pXX5: pUT18C-hetRR223W 521 The pXX4 plasmid was used as template to substitute the R223 residue of HetR to Tryptophan. For 522 this the primers Mut hetRR223W fw and Mut hetRR223W rv were used as primers in a megapriming 523 PCR assay. The open reading frame of hetRhood (encoding for HetRhood from Y215 to R296) was amplified using the 538 hetRhood dh fw and hetRhood dh rv primers and cloned into the PstI and EcoRI restriction sites of the 539 pUT18 expression plasmid. 540 The open reading frame of all3256 gene was amplified using the all3256 dh fw T18 and all3256 dh rv 542 primers and cloned into the PstI and EcoRI restriction sites of the pUT18C expression plasmid. 543 The open reading frame of all4303 gene was amplified using the all4303 dh fw T18 and all4303 dh rv 545 primers and cloned into the PstI and EcoRI restriction sites of the pUT18C expression plasmid. 546

pXX12: pET28a-hetL-his 547
The open reading frame of hetL gene (all3740) was amplified using the hetL pET28 fw and hetL pET28 548 rv infusion primers and cloned into the XhoI and NcoI restriction sites of the pET28a expression plasmid 549 using the In-Fusion technology (Takara In-Fusion® HD Cloning kit). The open reading frame of hetL gene (all3740) was amplified using the hetL pRL fw and hetL pRL rv 556 infusion primers and cloned into the BamHI restriction site of the pRL25T-PpetE replicative plasmid in 557 Nostoc (Takara In-Fusion® HD Cloning kit). 558 The open reading frame of hetL gene mutated to encode for a D151A substitution was amplified using 560 the hetL pRL fw and hetL pRL rv infusion primers with pKT25-hetLD151A as template and cloned into 561 the BamHI restriction site of the pRL25T-PpetE replicative plasmid in Nostoc (Takara In-Fusion® HD 562 Cloning kit). 563 pSC1: pRL25T-PpatS-hetL 564 The promoter of patS gene was amplified using PpatS fw and PpatS rv, the hetL coding region was 565 amplified using hetL PpatS fw and hetL PpatS rv. Nostoc genomic DNA was used as template  The promoter of rbcL gene was amplified using PrbcL fw and PrbcL rv, the hetL coding region was 570 amplified using hetL PrbcL fw and hetL PrbcL rv. Nostoc genomic DNA was used as template 571 for both amplifications. The two amplicons were cloned into the BamHI restriction site of the 572 pRL25T plasmid using the In-Fusion technology.

Protein purification: 574
For HetL purification, BL21DE3 strain containing the pXX12 plasmid was grown until an optical 575 density (OD 600nm) of 0.6. hetL induction was achieved by the addition of Isopropyl β-D-1-576 thiogalactopyranoside (IPTG, SIGMA) of 0.4 mM over night at 16 °C. Cells were harvested at 8000 577 rpm at 4˚C during 2 min. The pellet was re-suspended in 25 mL of lysis buffer (50 mM Tris HCl (pH 578 8), 0.3 M NaCl), and cells were disrupted using French press. After centrifugation at 8000 rpm for 30 579 min at 4˚C, the supernatant was loaded onto a column containing 1 mL of NiNTA agarose resin (Qiagen, 580 Hilden, Germany) pre-equilibrated with lysis buffer containing 10 mM Imidazole. The column was 581 rinsed with 10 mM and 35 mM Imidazole, both prepared in lysis buffer. Fractions were collected (in 582 200 mM Imidazole buffer, prepared in lysis buffer). The Imidazole was eliminated using the PD10 583 columns (GE Healthcare). The proteins were concentrated using Vivaspin columns (SIGMA) and 584 quantified using the Bradford assay (SIGMA). HetR purification was undergone as previously described 585 (Hu et al., 2015). 586

RNA preparation, reverse transcription, and quantitative Real-Time-PCR:
597 RNAs were prepared using the Qiagen RNA extraction kit (Qiagen) following the manufacturer's 598 instructions. An extra TURBO DNase (Invitrogen) digestion step was performed to eliminate the 599 contaminating DNA. The RNA quality was assessed by tape station system (Agilent). RNAs were 600 quantified spectrophotometrically at 260 nm (NanoDrop 1000; Thermo Fisher Scientific). For cDNA 601 synthesis, 1 µg total RNA and 0.5 μg random primers (Promega) were used with the GoScript™ Reverse 602 transcriptase (Promega) according to the manufacturer instructions. Quantitative real-time PCR (qPCR) 603 analyses were performed on a CFX96 Real-Time System (Bio-Rad). The reaction volume was 15 μL 604 and the final concentration of each primer was 0.5 μM. The qPCR cycling parameters were 95°C for 2 605 min, followed by 45 cycles of 95°C for 5 s, 55°C for 60 s. A final melting curve from 65°C to 95°C was 606 added to determine the specificity of the amplification. To determine the amplification kinetics of each 607 product, the fluorescence derived from the incorporation of BRYT Green ® Dye into the double-stranded 608 PCR products was measured at the end of each cycle using the GoTaq ® qPCR Master Mix 2X Kit 609 (Promega). The results were analysed using Bio-Rad CFX Maestro software, version 1.1 (Bio-Rad, 610 France). The RNA 16S gene was used as a reference for normalization. All measurements were carried 611 out in triplicate and a biological duplicate was performed for each point. The amplification efficiencies 612 of each primer pairs were 80 to 100%. All of the primer pairs used for qPCR are reported in 613 Supplementary file 1. 614

Bacterial two hybrid assays: 615
Bacterial two-hybrid assays were performed following the procedure described by Karimova et al 616 (1998) (Karimova et al., 1998). Briefly, after co-transforming the BTH101 strain with the two plasmids 617 expressing the T18-and T25-fusions, LB plates containing ampicillin and kanamycin were incubated 618 at 30° C for 2 days. For each assay, 10 independent colonies were inoculated in 3 ml of LB medium 619 supplemented with ampicillin, kanamycin and 0.5 mM IPTG, and incubated at 30°C overnight. ß-620 galactosidase activity was determined as previously described (Zubay, Morse, Schrenk, & Miller, 1972). 621 The values presented are means of 3 independent assays. 622