<<

1 Catabolic pathway of in Anabaena involves a novel bifunctional

2 that produces from arginine

3

4 Mireia Burnat, Silvia Picossi, Ana Valladares, Antonia Herrero, and Enrique

5 Flores*

6 Instituto de Bioquímica Vegetal y Fotosíntesis, CSIC and Universidad de Sevilla, Américo

7 Vespucio 49, E-41092 Sevilla, Spain.

8

9 *Correspondence to: Enrique Flores ([email protected])

10

11

12

13 Key words: Archaea; Arginine catabolism; Arginine dihydrolase; Cyanobacteria; Ornithine

14 cyclodeaminase

1 15 Summary

16 Arginine participates widely in metabolic processes. The heterocyst-forming cyanobacterium

17 Anabaena catabolizes arginine to produce proline and glutamate, with concomitant release of

18 ammonium, as major products. Analysis of mutant Anabaena strains showed that this catabolic

19 pathway is the of two , agrE (alr4995) and putA (alr0540). The predicted PutA

20 protein is a conventional, bifunctional proline oxidase that produces glutamate from proline. In

21 contrast, AgrE is a hitherto unrecognized enzyme that contains both an N-terminal  propeller

22 domain and a unique C-terminal domain of previously unidentified function. In vitro analysis of

23 the proteins expressed in Escherichia coli or Anabaena showed arginine dihydrolase activity of

24 the N-terminal domain and ornithine cyclodeaminase activity of the C-terminal domain, overall

25 producing proline from arginine. In the diazotrophic filaments of Anabaena, -aspartyl-arginine

26 dipeptide is transferred from the heterocysts to the vegetative cells, where it is cleaved producing

27 aspartate and arginine. Both agrE and putA were found to be expressed at higher levels in

28 vegetative cells than in heterocysts, implying that arginine is catabolized by the AgrE-PutA

29 pathway mainly in the vegetative cells. Expression in Anabaena of a homologue of the C-

30 terminal domain of AgrE obtained from Methanococcus maripaludis enabled us to identify an

31 archaeal ornithine cyclodeaminase.

32

33

2 34 Introduction

35 Arginine (2-amino-5-guanidinopentanoic acid) is the with highest N:C ratio. This is

36 due to the presence of a guanidine group capping its 3-carbon aliphatic side chain. In addition to

37 being a proteinaceous amino acid, arginine has multiple metabolic uses in the living cell (Morris,

38 2016). For example, it is the for biosynthesis of nitric oxide, and its gives

39 rise to polyamines and metabolites of the urea cycle. In various bacteria, arginine can serve as a

40 source of carbon, nitrogen and energy. In cyanobacteria, oxygenic phototrophs, arginine

41 accumulates in a cell inclusion called cyanophycin (multi-L-arginyl-poly[L-]) that

42 can eventually be mobilized. Consistent with such diverse metabolic uses, arginine can be

43 metabolized by numerous , including those of catabolic pathways that make its nitrogen

44 atoms available for metabolism. Some of these pathways produce glutamate, which is a major

45 distributor of nitrogen in the cell (Lu, 2006). Commonly, first steps in catabolism of arginine

46 involve its conversion to ornithine, either directly by arginase (with concomitant release of urea)

47 or arginine amidinotransferase (with transfer of urea to an acceptor metabolite) or indirectly via

48 citrulline produced by arginine deiminase and further metabolized by ornithine

49 carbamoyltransferase (Lu, 2006). Ornithine is then converted to glutamate either via glutamate-

50 semialdehyde or via proline. In the latter case, ornithine cyclodeaminase, an enzyme discovered

51 in Clostridia (Costilow and Laycock, 1971; Goodman et al., 2004), produces the proline that is

52 further catabolized to glutamate by PutA, which is a bifunctional enzyme with proline

53 and 1-pyrroline-5-carboxylate dehydrogenase activities (Liu et al., 2017).

54 Some cyanobacteria grow as chains of cells (filaments) that, under nitrogen deprivation,

55 contain two cell types: vegetative cells that fix CO2 through oxygenic photosynthesis and

56 heterocysts that fix N2 (Flores and Herrero, 2010). In rapidly growing filaments, heterocysts and

3 57 vegetative cells are found in a ratio of about 1 to 10. The heterocysts provide the vegetative cells

58 with fixed nitrogen, and the vegetative cells provide the heterocysts with reduced carbon, mainly

59 in the form of sucrose, through intercellular molecular transfer (Nürnberg et al., 2015).

60 Cyanophycin accumulates conspicuously in the polar regions of the heterocysts, and degradation

61 of cyanophycin takes place in two steps: cyanophycinase produces a dipeptide, -aspartyl-

62 arginine, that is transferred to the adjacent vegetative cells where it is hydrolyzed by isoaspartyl

63 dipeptidase producing aspartate and arginine (Burnat et al., 2014). Those amino acids, together

64 with (Thomas et al., 1977), serve as nitrogen sources for the vegetative cells. Whereas

65 aspartate and glutamine can make their nitrogen atoms available for metabolism by glutamine-

66 oxoglutarate transamidase and , respectively (Thomas et al., 1977; Xu et

67 al., 2015), the pathway for arginine utilization was elusive.

68

69 Results

70 Physiological-genetic identification of the agrE-putA pathway

71 When filaments of the model heterocyst-forming cyanobacterium, Anabaena sp. strain PCC

72 7120 (hereafter Anabaena), were incubated in the presence of [14C]arginine, labelled proline and

73 glutamate were produced at substantial amounts, and ornithine was occasionally detected at low

74 levels (Fig. 1A; see also Burnat and Flores, 2014). We previously showed that a putative

75 ureohydrolase encoded in the Anabaena genome is an rather than an arginase and is

76 not required for the production of proline and glutamate from arginine (Burnat and Flores, 2014).

77 Open reading frame (ORF) alr4995 of the Anabaena genome has been discussed as a possible

78 arginine deiminase or arginine amidinotransferase (Schriek et al., 2007). A more detailed

79 analysis of Alr4995 showed however that it has two distinct domains (Fig. 2). The N-terminal

4 80 part of the protein is predicted to have the structure of an / propeller domain characteristic of

81 arginine deiminases and arginine amidinotransferases (Shirai et al., 2006), and the C-terminal

82 part is homologous to a Methanococcus protein annotated as -oxoglutarate

83 reductase/saccharopine dehydrogenase (LOR/SDH) protein 2 whose function was unknown (we

84 shall refer this Methanococcus protein as Mls2). We denoted ORF alr4995 agrE (from

85 arginine-guanidine removing enzyme) and constructed an Anabaena strain (agrE) with deletion

86 of an internal fragment of this gene (see Experimental procedures and Fig. S1). Neither proline

87 nor glutamate was observed in filaments of strain agrE when incubated in the presence of

88 [14C]arginine (Fig. 1A), indicating that protein AgrE is needed to catabolize arginine to proline

89 and glutamate. Anabaena filaments incubated with [14C]ornithine also produced labelled proline

90 and glutamate, and this production was also arrested in strain agrE (Fig. 1B), indicating that

91 AgrE also mediates catabolism of ornithine (Fig. 1C), which could be an intermediate in the

92 reaction that produces proline from arginine.

93 Because arginine deiminases produce citrulline from arginine, it was also possible that

94 citrulline rather than ornithine is an intermediate in the reaction catalyzed by AgrE, and that in

95 our in vivo assays with [14C]ornithine, [14C]citrulline was produced from [14C]ornithine by

96 ornithine carbamoyltransferase. We thereupon constructed an Anabaena strain (∆argF::C.K3)

97 with an insertion of gene-cassette C.K3 in the argF gene that encodes ornithine

98 carbamoyltransferase (Fig. S2). In this mutant, labelled proline and glutamate were produced

99 from [14C]ornithine as in the wild type (Fig. 3), indicating that conversion to citrulline was

100 unnecessary for ornithine catabolism. These observations suggested that AgrE can indeed use

101 ornithine as a substrate as well as produce it as an intermediate in the conversion of arginine into

102 proline. We hypothesize that (i) the N-terminal domain of AgrE catalyzes the formation of

5 103 ornithine from arginine, and (ii) the C-terminal domain of AgrE catalyzes the production of

104 proline from ornithine.

105 We then constructed derivatives of the Anabaena agrE mutant expressing Strep-tagged

106 Anabaena AgrE or Methanococcus Mls2 (Fig. S3 and S4, respectively). The Anabaena agrE

107 mutant complemented with the Strep-tagged AgrE produced [14C]proline and [14C]glutamate

108 from both [14C]arginine and [14C]ornithine, as the wild type (Fig. 4). The Anabaena agrE

109 mutant complemented with the Strep-tagged Mls2 produced [14C]proline and [14C]glutamate

110 from [14C]ornithine but not from [14C]arginine (Fig. 4). These results corroborate that AgrE is

111 involved in the production of proline from arginine in Anabaena and show that the

112 Methanococcus Mls2 protein provides Anabaena with the capability to synthesize proline from

113 ornithine. Therefore, Mls2 is an ornithine cyclodeaminase.

114 ORF alr0540 of the Anabaena genome encodes a putative PutA protein (bifunctional

115 and 1-pyrroline-5-carboxylate dehydrogenase). To investigate whether

116 this enzyme is responsible for the second part of the arginine catabolism pathway, production of

117 glutamate from proline, we created an alr0540::C.S3 insertional mutant of Anabaena (Fig. S5).

118 This mutant did not produce [14C]glutamate from [14C]arginine or [14C]ornithine, but instead

119 accumulated [14C]proline (Fig. 1A, B). To test directly the role of alr0540 in proline catabolism,

120 we additionally tested the fate of [14C]proline in the wild-type and mutant cells. Whereas the

121 wild type produced [14C]glutamate and, further, [14C]glutamine –by amination of [14C]glutamate

122 by , the alr0540::C.S3 mutant did not produce [14C]glutamate or

123 [14C]glutamine when supplied extracellularly with [14C]proline (Fig. 1D). Hence, alr0540

124 encodes a PutA protein of Anabaena.

125

6 126 In vitro activity of AgrE

127 The complete AgrE protein and its N-terminal AT/ADI domain were produced, separately, in

128 Escherichia coli as His-tagged proteins, and the products of in vitro assays were analyzed by

129 HPLC after derivatization with phenylisothiocyanate (Heinrikson and Meredith, 1984). When

130 incubated with arginine in a buffer, each of these proteins produced ornithine (Fig. 5A, B;

131 enzymatic reactions tested are summarized in Table 1.1 to 1.3), but no simultaneous production

132 of urea (determined colorimetrically) was observed as it would have been expected from an

133 arginase-like reaction. Instead, in addition to ornithine, production of ammonium was found. The

134 whole protein produced 1.71 ± 0.16 mol of NH4+ per mol of ornithine (mean and SD, n = 4

135 independent enzyme preparations), and the N-terminal domain produced 1.84 mol of NH4+ per

136 mol of ornithine. These results indicate that the N-terminal domain of AgrE is an arginine

137 dihydrolase, reminiscent of the succinylarginine dihydrolase activity of some / propeller

138 proteins (Shirau et al., 2006). The same conclusion has been recently reported for a homologous

139 protein from the unicellular cyanobacterium Synechocystis sp. PCC 6803 (Zhang et al., 2018).

140 No production of proline from ornithine was, however, observed with the whole His-tagged

141 AgrE protein assayed under different incubation conditions.

142 We next produced and isolated a Strep-tagged AgrE protein from an Anabaena strain

143 carrying the corresponding gene construct on a plasmid (Fig. S3). Using arginine as substrate,

144 this protein was observed to produce ammonium and ornithine with a stoichiometry of 1.91 mol

145 NH4+ per mol of ornithine (Fig. 5C and Table 1.4). Using ornithine as substrate, this protein,

146 incubated in the presence of a reducing agent and 1 mM NAD+, was observed to produce

147 ammonium and proline with a stoichiometry of 1.16 ± 0.12 mol NH4+ per mol of proline (n = 4)

148 (Fig. 5D and Table 1.4). The reaction was shown to require NAD+ that could be partially

7 149 substituted by NADP+ (Table 1.4), but did not require ATP which instead was inhibitory (Table

150 1.5). Importantly, ornithine cyclodeaminases are known to use NAD+ as a catalytic

151 (Goodman et al., 2004). These results indicate that AgrE can catalyze the cyclodeamination of

152 ornithine. Because the N-terminal, / propeller domain is an arginine dihydrolase, and because

153 the C-terminal, LOR/SDH-like domain is homologous to the Methanococcus Mls2 protein that

154 provides Anabaena with the capability to produce proline from ornithine (Fig. 4D), we ascribe

155 the ornithine cyclodeaminase activity of AgrE to the C-terminal domain of the protein (Fig. 2).

156 Using arginine as substrate some preparations of Strep-tagged AgrE were observed to

157 produce more than 2 mol NH4+ per mol of ornithine (see Table 1.6). This result suggested that

158 the Strep-tagged protein isolated from Anabaena is able to carry out the two consecutive

159 reactions of AgrE, also producing ammonium from ornithine, although the concomitant

160 production of proline could not be determined because of coincidence of proline and arginine in

161 the chromatography. We then carried out assays analyzing the products of the reaction with a

162 different amino acid derivatization and HPLC method (Fabiani et al., 2002). For these assays, we

163 used Strep-tagged AgrE isolated from Anabaena in the presence of a reducing agent and NAD+,

164 which were the conditions that permitted to observe consistently the ornithine cyclodeaminase

165 activity. Using arginine as substrate, the production of ammonium, ornithine and proline could

166 be detected (Fig. 6 and Table 1.7). The amount of ornithine produced was 1.65-fold higher than

167 the amount of proline, and the amount of ammonium produced (29.2 nmol min-1) corresponded

168 to that expected from the production of 2 mol of NH4+ per mol of ornithine (2 x 7.62 nmol min-1)

169 plus 3 mol of NH4+ per mol of proline (3 x 4.62 nmol min-1), according to the reaction depicted

170 in Fig. 2. Hence, the studies carried out in vitro with different versions of the AgrE protein

8 171 corroborate that this enzyme can catalyze the conversion of arginine into proline with

172 concomitant release of ammonium.

173

174 Physiology and cell specificity of the AgrE-PutA pathway

175 We have recently described a pathway producing sym-homospermidine from arginine in

176 Anabaena (Burnat et al., 2018). The first enzyme in this pathway is arginine decarboxylase

177 encoded by speA (all3401), and an speA mutant is drastically hampered in diazotrophic growth

178 apparently because of a requirement for polyamines during heterocyst differentiation (Burnat et

179 al., 2018). Here we studied the growth rate constants of the agrE and putA mutants in liquid

180 medium. As shown in Fig. 7A, both mutants showed substantial diazotrophic growth.

181 Nonetheless, strain agrE was frequently observed to have increased granulation in the

182 cytoplasm that can correspond to accumulation of cyanophycin (Fig. 7B), which may reflect a

183 restricted catabolism of arginine in this mutant. To investigate whether further metabolic

184 pathways can use arginine in Anabaena, we created an agrE speA double mutant by transferring

185 an speA inactivating construct (speA::C.S3) to the agrE strain (Fig. S6). Filaments of strain

186 agrE speA::C.S3 did not generate any significant 14C-labeled product when incubated in the

187 presence of [14C]arginine (Fig. 8A), indicating that no other major pathway degrading arginine is

188 expressed in Anabaena under the growth conditions tested. Like the speA mutant, the double

189 agrE speA::C.S3 mutant was unable to grow diazotrophically (Fig. 8B).

190 In diazotrophically-grown filaments of Anabaena, hydrolysis of -aspartyl-arginine takes

191 place mainly in vegetative cells rather than in heterocysts (Burnat et al., 2014). To investigate the

192 cell type in which the AgrE-PutA pathway is active, we prepared an Anabaena strain producing

9 193 AgrE fused to superfolder green fluorescent protein (sf-GFP) (Fig. S7). In filaments incubated

194 under diazotrophic conditions, the AgrE-sf-GFP fusion protein accumulated in vegetative cells at

195 substantially higher levels than in heterocysts (Fig. 9A,B). Expression of the putA gene was

196 studied by northern blot analysis with RNA isolated from whole filaments incubated with

197 different nitrogen sources and from purified heterocysts (Fig. 9C). The heterocyst RNA was

198 validated by hybridization with probes of the nifH gene (encoding a nitrogenase component,

199 expressed only in heterocysts) and the rbcL gene (encoding a subunit of ribulose-1,5-

200 bisphosphate carboxylase/oxygenase, expressed only in vegetative cells) as described previously

201 (Picossi et al., 2005). The putA gene is 2,991-bp long, and a transcript of about 3.6 kb was

202 observed with RNA from whole filaments but not from heterocysts (Fig. 9C), showing that that

203 putA is expressed specifically in vegetative cells. These results together suggest that arginine is

204 degraded by the AgrE-PutA pathway mainly in vegetative cells.

205

206 Discussion

207 Our work has identified in the heterocyst-forming cyanobacterium Anabaena an arginine

208 catabolism pathway that produces glutamate mediated by the products of only two genes, agrE

209 (alr4995 in the Anabaena genome) and putA (alr0540). Figure 10 depicts this pathway along

210 with the homospermidine biosynthesis pathway that we have recently described in Anabaena

211 (Burnat et al., 2018). Among the pathways that catabolize arginine producing glutamate and

212 ammonium, the use of ornithine as an intermediate is common. However, ornithine is normally

213 produced by an arginase that also releases urea or by a dihydrolase if arginine is activated with

214 succinyl-CoA (Lu, 2006; Vander Wauven and Stalon, 1985). AgrE is unique in having arginine

215 dihydrolase activity, which can be ascribed to the N-terminal domain of the protein (Fig. 2; see

10 216 also Zhang et al., 2018). In many bacteria, ornithine (or succinyl ornithine) is further catabolized

217 through glutamate semialdehyde (or succinyl glutamate semialdehyde) to glutamate (Lu, 2006),

218 but in a few bacteria including some clostridia and Agrobacterium a specific ornithine

219 cyclodeaminase produces proline (Costilow and Laycock, 1971; Goodman et al., 2004), which is

220 then oxidized to glutamate by the bifunctional proline oxidase PutA (Lu, 2006). As we have

221 shown in this work, AgrE from Anabaena bears a C-terminal domain that carries out an ornithine

222 cyclodeaminase-like reaction producing proline, although AgrE is not homologous to

223 conventional ornithine cyclodeaminases and therefore represents a new type of protein with this

224 activity. Proline produced by AgrE in Anabaena is further oxidized to glutamate by PutA.

225 Overall, AgrE and PutA represent a straightforward pathway to make available to metabolism

226 the four nitrogen atoms of arginine, three as ammonium ions and one as glutamate (Fig. 10).

227 As observed repeatedly in this work, when using arginine as substrate, AgrE can release

228 ornithine in vitro. It is likely that ornithine release from AgrE can also take place in vivo, since

229 [14C]ornithine has been occasionally observed in our experiments of [14C]arginine catabolism by

230 Anabaena (see, e.g., Fig. 4A). Nonetheless, in vitro, AgrE can produce proline from arginine as

231 shown in Fig. 6, implying that the C-terminal domain of AgrE can directly use ornithine

232 produced by the N-terminal domain. This observation raises the possibility that AgrE carries out

233 substrate channeling, a process by which an intermediate is transferred from one enzyme active

234 site to another without diffusing out of the enzyme (Liu et al., 2017). If this were the case, the Km

235 of the enzyme for ornithine would be expected to be higher than for arginine. Consistently,

236 available information permits to estimate a Km of AgrE for arginine of about 6 mM and for

237 ornithine of about 50 mM (Table 1). Although further work will be necessary to increase our

238 knowledge of the biochemistry of AgrE, we speculate that channeling in AgrE may prevent the

11 239 use of ornithine in arginine biosynthesis under physiological conditions that favor arginine

240 catabolism, thus avoiding the operation of a futile cycle. A similar situation, channeling of the

241 intermediate P5C/GSA, may occur in the second enzyme of this pathway, PutA (see Fig. 10),

242 thus avoiding the use of the intermediate in proline biosynthesis. Substrate channeling has been

243 investigated comprehensively in other bacterial PutA proteins (Liu et al., 2017).

244 In Anabaena, arginine can be taken up from the outer medium by a high-affinity ABC

245 transporter (Pernil et al., 2008) or can be produced from cyanophycin (Burnat et al., 2014).

246 Under diazotrophic conditions, -aspartyl-arginine released from cyanophycin in the heterocysts

247 is transferred to the vegetative cells, in which it is split producing aspartate and arginine (Burnat

248 et al., 2014). The agrE and putA genes are expressed mainly in the vegetative cells of

249 diazotrophic filaments, which is consistent with the idea that the AgrE-PutA pathway will

250 contribute to the utilization by the vegetative cells of the nitrogen fixed in the heterocysts. This

251 pathway is however not essential for diazotrophic growth. The use of arginine as substrate for the

252 biosynthesis of homospermidine may also contribute to support growth of the vegetative cells

253 under diazotrophic conditions, since, in addition to producing polyamines, the homospermidine

254 biosynthesis pathway releases urea and ammonia that can be used in metabolism (Fig. 10). For

255 this, urea is further degraded to two molecules of ammonia and CO2 by urease, which is present

256 in Anabaena (Valladares et al., 2002).

257 The agrE and putA genes are present in a large number of sequenced cyanobacterial

258 genomes (tested at https://img.jgi.doe.gov/cgi-bin/m/main.cgi) indicating that the agrE-putA

259 pathway is widely distributed in these organisms. However, these genes are not present in

260 cyanobacteria such as Synechococcus elongatus or the marine Prochlorococcus spp. and

261 Synechococcus spp., which do not produce cyanophycin. These observations suggest a role of the

12 262 agrE-putA pathway in catabolism of arginine made available by mobilization of cyanophycin.

263 Previous work done with Synechocystis (which also accumulates cyanophycin) suggested a role

264 of the proline biosynthesis protein ProC in production of glutamate during arginine and ornithine

265 catabolism, but the proC mutant still showed substantial production of proline implying the

266 existence of another enzyme synthesizing proline during arginine catabolism (Quintero et al.,

267 2000). The Synechocystis ORF Sll1336 encodes a 705-amino acid residue protein that is 74 %

268 identical to Anabaena AgrE, and could therefore be responsible for the observed production of

269 proline in the Synechocystis proC mutant.

270 Whereas the putA gene is of wide distribution in the living world, agrE is restricted to

271 cyanobacteria and a few other bacteria. Notably, a protein homologous to the C-terminal domain

272 of AgrE is found in Methanococcus maripaludis and related archaea. These organisms have been

273 known to express ornithine cyclodeaminase activity (Graupner and White, 2001), but no gene

274 encoding a conventional ornithine cyclodeaminase is found in their sequenced genomes leading

275 to the idea that they express a different type of ornithine cyclodeaminase. Hence, our work

276 showing that the M. maripaludis protein homologous to the C-terminal part of AgrE provides

277 Anabaena with the capability of producing proline from ornithine has also identified the gene

278 encoding archaeal ornithine cyclodeaminase. The M. maripaludis gene encoding this protein,

279 MMP1218, appears to be not essential in either minimal medium or a complex rich medium

280 (Sarmiento et al., 2013). Our results identifying this gene’s product as an ornithine

281 cyclodeaminase could open the possibility of performing a specific test of its role in the

282 physiology of this type of archaea.

283

284

13 285 Experimental procedures

286 Strains and growth conditions

287 Anabaena sp. strain PCC 7120 was grown axenically in BG11 medium (containing NaNO3),

288 BG110 medium (free of combined nitrogen) or BG110NH4+ medium (BG110 containing 4 mM

289 NH4Cl and 8 mM TES-NaOH buffer, pH 7.5). In every case, ferric citrate replaced the ferric

290 ammonium citrate used in the original recipe (Rippka et al., 1979). For plates, medium was

291 solidified with 1% separately autoclaved Difco agar. Cultures were grown at 30 ºC in the light

292 (25 μmol photons m-2 s-1), with shaking (80-90 rpm) for liquid cultures. Alternatively, cultures

293 (referred to as bubbled cultures) were supplemented with 10 mM of NaHCO3 and bubbled with a

294 mixture of CO2 and air (1 % v/v) in the light (50 to 75 mol photons m-2 s-1). For mutants

295 described below, antibiotics were used at the following concentrations: erythromycin (Em), 2 μg

296 mL-1 for liquid cultures and 5 μg mL-1 for solid media; neomycin (Nm) at 20 μg mL-1 for both

297 liquid and solid media; streptomycin sulfate (Sm) and spectinomycin dihydrochloride

298 pentahydrate (Sp), 5 μg mL-1 each for both liquid and solid media. DNA was isolated from

299 Anabaena sp. by the method of Cai and Wolk (1990). The ΔargF::C.K3 mutant is an arginine

300 auxotroph and was supplemented with 0.5 mM L-arginine.

301 Escherichia coli strain DH5α was used for plasmid constructions, strains HB101 and

302 ED8654 for conjugations with Anabaena sp., and strain BL21(DE3) for overexpression of

303 proteins. They were grown in Luria-Bertani medium supplemented when appropriate with

304 antibiotics at standard concentrations.

305

306 Plasmid constructions and genetic procedures

14 307 Open reading frame (ORF) alr4995 of the Anabaena (Kaneko et al., 2001) was

308 inactivated by removing an internal fragment of 1,755 bp. DNA fragments upstream (490 bp)

309 and downstream (500 bp) from the central region of the gene were amplified by PCR using DNA

310 from Anabaena as template and primers alr4995-3/alr4995-4 and alr4995-5/alr4995-6 (all

311 oligodeoxynucleotide primers are listed in Table S1). The external primers, alr4995-3 and

312 alr4995-6, included a PstI-site in their 5’ ends, and primers alr4995-4 and alr4995-5 were

313 complementary in their 5’ ends and joined together by the megaprimer PCR protocol. The DNA

314 fragment from this gene was cloned in pSPARK (Canvax Biotech S.L., Spain) and its sequence

315 was corroborated by sequencing. The cloned fragment was then transferred as a PstI-ended

316 fragment to PstI-digested pCSRO (Merino-Puerto et al., 2013), producing pCSDH8.

317 Conjugation of Anabaena with E. coli HB101 carrying the cargo plasmid (pCSDH8) with

318 helper and methylation plasmid pRL623 was effected by the conjugative plasmid pRL443,

319 carried in E. coli ED8654, and performed as described (Elhai et al., 1997) with selection for

320 resistance to Sm and Sp. Exconjugants were spread on BG110NH4+ medium supplemented with

321 5% sucrose (Cai and Wolk, 1990), and individual SucR colonies were checked by PCR looking

322 for clones that had replaced the wild-type locus by a locus bearing the deletion. The genetic

323 structure of selected clones was studied by PCR with DNA from those clones and primers

324 alr4995-6/alr4995-7 and alr4995-7/alr4995-8. A clone homozygous for the mutant

325 was named strain CSMI43 (Δalr4995).

326 The plasmid carrying fusion gene alr4995-sf-gfp was prepared as follows. A 603-bp

327 fragment from the 3’-terminal part of alr4995 was amplified by PCR using Anabaena DNA as

328 template and primers alr4995-9 (which contains a HindIII site) and alr4995-10 (which lacks the

329 stop codon of the gene and contains a BsaI site in its 5’ end), and the resulting fragment was

15 330 cloned as a HindIII/BsaI-ended fragment in HindIII/BsaI-digested pCSAL39 producing plasmid

331 pCSMI72 that carries the fusion of the sf-gfp gene to the 3’ end of alr4995 (pCSAL39 is a

332 pMBL-T- derived vector that contains the sf-gfp gene, a sequence encoding a 4-Gly linker and a

333 BsaI site in its 5’ end). The insert of pCSMI72 was corroborated by sequencing and the resulting

334 fusion was transferred as a KpnI-ended fragment to KpnI-digested pCSV3, which provides

335 resistance to Sm and Sp (Valladares et al., 2011), producing pCSMI73. This plasmid, which

336 bears the alr4995-sf-gfp fusion gene, was transferred to Anabaena by triparental mating as

337 described above, with selection for SmR SpR. Insertion into alr4995 and segregation of

338 chromosomes carrying the fusion was confirmed by PCR using template DNA from exconjugant

339 clones and primers alr4995-11 and gfp-6 for testing insertion of sf-gfp, and alr4995-11 and

340 alr4995-6 for testing segregation of the mutated chromosomes. A homozygous clone bearing the

341 alr4995-sf-gfp construct was named strain CSMI32.

342 ORF alr4907 (argF) encodes ornithine carbamoyltransferase (Kaneko et al., 2001). To

343 inactivate alr4907, two DNA fragments, one encompassing 522 bp from sequences upstream

344 from the central region of the gene and the other including 646 bp from sequences downstream

345 from the central region of the gene, were amplified by PCR using DNA from Anabaena as

346 template and primer pairs alr4907-1/alr4907-2 and alr4907-3/alr4907-4, respectively (alr4907-2

347 and alr4907-3 bear SmaI sites at their 5’ ends). The two DNA fragments, joined together by the

348 megaprimer PCR protocol, were cloned into pSPARK producing plasmid pCSMI67 and, after

349 corroboration by sequencing and digestion with SmaI, were ligated to SmaI-ended gene cassette

350 C.K3 encoding Kmr/Nmr, producing plasmid pCSMI68. The insert of the resulting plasmid,

351 excised with SacI, was transferred to SacI-digested pCSRO (Merino-Puerto et al., 2013),

352 producing pCSMI69. Conjugation of Anabaena with E. coli HB101 carrying pCSMI69 and

16 353 helper and methylation plasmid pRL623 was effected by the conjugative plasmid pRL443,

354 carried in E. coli ED8654, and performed as described (Elhai et al., 1997) with selection for

355 resistance to Nm. Exconjugants were isolated, and double recombinants were identified as clones

356 resistant to Nm, resistant to sucrose (Cai and Wolk, 1990), and sensitive to Sm and Sp, for which

357 the resistance determinant was present in the vector portion of the transferred plasmid. The

358 genetic structure of selected clones was studied by PCR with DNA from those clones and

359 primers alr4907-5/alr4907-6 and alr5045-5/alr4907-4. Clones homozygous for the mutant

360 chromosomes were only obtained for the indirect orientation of the C.K3 cassette, and were

361 named strain CSMI30b (Δalr4907::C.K3).

362 ORF alr0540 (putA) encodes proline oxidase. To inactivate alr0540, a 2.3-kb fragment

363 from the central part of alr0540 was amplified by PCR using Anabaena DNA as template and

364 primers PA-1 and PA-2, and the resulting fragment was cloned into pGEM-T producing plasmid

365 pCSS6. Gene cassette C.S3 encoding Smr Spr was introduced in the HindIII site of pCSS6,

366 producing plasmid pCSS9. The insert of the resulting plasmid, excised with PvuII, was

367 transferred to NruI-digested pRL278, resulting in plasmid pCSS11. Conjugation of Anabaena

368 and identification of exconjugants was performed as for the CSMI30b mutant (but using Smr Spr

369 for positive selection). A clone homozygous for the mutant chromosomes was named CSS5.

370 ORF all3401 (speA) encodes arginine decarboxylase, and the arginine decarboxylase

371 deletion mutant (∆speA::C.S3) has been described previously (Burnat et al., 2018). To create a

372 double ∆agrE ∆speA::C.S3 mutant, plasmid constructs used to create the speA mutant were

373 transferred to Anabaena ∆agrE by triparental mating, as described above, with selection for SmR

374 SpR. Exconjugants were isolated, and double recombinants were identified as clones resistant to

375 Sm and Sp, resistant to sucrose, and sensitive to Nm, for which the resistance determinant was

17 376 present in the vector portion of the transferred plasmid. The genetic structure of selected clones

377 was studied by PCR with DNA from those clones and appropriate primer pairs (Fig. S6). A clone

378 homozygous for both mutant chromosomes was obtained and named strain CSMI43-MI35

379 (∆agrE ΔspeA::C.S3).

380

381 Construction of strains producing Alr4995 (AgrE)

382 For overexpression of protein Alr4995 (AgrE), two strategies were used. For the first strategy

383 (expression in E. coli), the 6xHis tag was added in frame to alr4995 and the N-terminal AT/ADI

384 domain of alr4995. Using DNA from Anabaena as template and primers alr4995-11 and

385 alr4995-12 (which includes a PstI site close to its 5’ end), and alr4995-11 and alr4995-13 (which

386 also includes a PstI site close to its 5’ end), the alr4995 gene and the AT/ADI domain of alr4995

387 were amplified by PCR. The PCR products were PstI-digested and cloned into SfoI/PstI-digested

388 pPROEX-HTb expression vector, producing plasmids pCSMI81 and pCSMI82, respectively,

389 whose inserts were corroborated by sequencing. These plasmids were introduced into E. coli

390 BL21(DE3) by electroporation.

391 The second strategy, used for overexpression of Alr4995 in Anabaena and for

392 complementation of the Anabaena alr4995 deletion mutant, was performed using Strep tag II

393 and a plasmid replicating in Anabaena. Using DNA from Anabaena as template and primers

394 alr4995-19 (which includes an NdeI site close to its 5’end) and alr4995-17 (which includes a

395 BamHI site close to its 5’ end), ORF alr4995 was amplified by PCR and ligated into pSPARK

396 (Canvax, Biotech SL) producing plasmid pCSMI93, whose insert was corroborated by

397 sequencing. This insert was excised from pCSMI93 by digestion with NdeI/SalI, and transferred

398 to NdeI/XhoI-digested pCMN28b (which is a vector derived from pET28b in which the His tag

18 399 has been replaced by a Strep-tag linker, as described previously [Napolitano et al., 2013]),

400 producing plasmid pCSMI94. Plasmid pCSMI94 was BglII-digested, receded 3’ends were filled

401 with Klenow fragment, and BamHI-digested. The resulting excised fragment was transferred to

402 SmaI/BamHI-digested pRL3845 replacing all1711 by alr4995, producing pCSMI95. (pRL3845

403 is a CmR EmR-plasmid that contains a PglnA-all1711 construct and can replicate in Anabaena

404 [López-Igual et al., 2012].) As is the case for all1711 in pRL3845, alr4995 in pCSMI95 is

405 expressed from the Anabaena glnA promoter. This plasmid was conjugated into Anabaena wild

406 type and strain CSMI43 (Δalr4995) as described above, with selection for EmR, producing

407 strains CSMI40 and CSMI43-C, respectively.

408

409 Construction of strains producing LOR/SDH protein 2 of Methanococcus

410 Cloning of the LOR/SDH protein 2 of Methanococcus (hereafter protein Mls2) was performed as

411 above for alr4995, with some modifications. Using genomic DNA from Methanococcus

412 maripaludis strain S0001 (which is derived from the wild-type M. maripaludis S2; William B.

413 Whitman laboratory, University of Georgia) as template and primers LOR/SDH-4 (which

414 includes a BamHI site close to its 5’end) and LOR/SDH-5 (which includes an NdeI site close to

415 its 5’ end), the Mls2-encoding gene was amplified by PCR and ligated into pSPARK producing

416 plasmid pCSMI99, whose insert was corroborated by sequencing. This insert was excised from

417 pCSMI99 by digestion with NdeI/SalI and transferred to NdeI/XhoI-digested pCMN28b

418 producing plasmid pCSMI100. This plasmid was XbaI-digested, filled in with Klenow fragment

419 and BamHI-digested. The resulting excised fragment was transferred to SmaI/BamHI-digested

420 pRL3845 replacing all1711 by Mls2, producing pCSMI101. This plasmid was conjugated into

19 421 Anabaena strain CSMI43 (Δalr4995) as described above, with selection for EmR, producing

422 strain CSMI43-CMls2.

423

424 Growth tests, microscopy and northern blot analysis

425 Protein concentration and chlorophyll a (Chl) content of the cultures were determined by a

426 modified Lowry procedure (Markwell et al., 1978) and by the method of Mackinney (1941),

427 respectively. The growth rate constant ( = ln2/td, where td is the doubling time) was calculated

428 from the increase of protein content, determined in 0.2-mL samples, of shaken liquid cultures.

429 Cultures were inoculated with cells containing about 5 µg of protein mL-1 and grew

430 logarithmically until reaching about 40 µg of protein mL-1.

431 For growth tests on solid media, cultures grown in BG11 medium were harvested and

432 washed three times with BG110 medium, and dilutions were prepared in BG110 medium.

433 Samples of 10-μl of the resulting suspensions were spotted on agar plates with different nitrogen

434 sources and incubated at 30 °C in the light (25 μmol photons m-2 s-1).

435 GFP fluorescence was analyzed by confocal microscopy. Samples from cultures of strain

436 CSMI32 or Anabaena wild type (used as a control) grown in bubbled cultures with BG11 or

437 BG110 medium were visualized using a Leica HCX PLAN-APO 63X 1.4 NA oil immersion

438 objective attached to a Leica TCS SP2 confocal laser-scanning microscope. GFP was excited

439 using 488-nm irradiation from an argon ion laser. Fluorescence emission was monitored by

440 collection across windows of 500-520 nm (GFP imaging) and 630-700 nm (cyanobacterial

441 autofluorescence). Under the conditions used, optical section thickness was about 0.4 µm. GFP

442 fluorescence intensity was analyzed using ImageJ 1.43m software.

20 443 Northern analysis was performed using a DNA probe of putA obtained with primers PA-1

444 and PA-2 (Table S1) and control probes of the nifH, rbcL and rnpB genes as described

445 previously (Picossi et al., 2005).

446

447 Amino acid catabolism

448 Cells grown in BG11 medium were harvested by centrifugation at 4,000 rpm at room

449 temperature, washed twice with 25 mM N-tris(hydroxymethyl)-methylglycine (Tricine)-NaOH

450 buffer (pH 8.1), and resuspended in the same buffer. The uptake assays were carried out at 30 °C

451 in the light (white light from fluorescent lamps, about 175 μmol photons m-2 s-1) and were started

452 by mixing a suspension of cells (2.1 mL) containing 5 to 10 μg of Chl· mL-1 with a solution (0.1

453 mL) of L-[U-14C]arginine (274 mCi·mmol-1), L-[1-14C]ornithine (57.1 mCi·mmol-1), both

454 purchased from Perkin Elmer (Massachusetts, USA), or L-[U-14C]proline (257 mCi·mmol-1)

455 from Amersham Life Science (UK). The final concentration of both arginine and ornithine in the

456 experiments was 1 μM, and the final concentration of proline was 1.9 µM. The amount of

457 metabolite taken up in the assays was estimated in 0.5-mL samples of the cell suspensions. The

458 samples were filtered (0.45 μm-pore-size Millipore HA filters were used) and the cells on the

459 filters were washed with 5 to 10 mL of 5 mM Tricine–NaOH buffer (pH 8.1). The filters

460 carrying the cells were then immersed in 5 mL of scintillation cocktail, and their radioactivity

461 was measured. Retention of radioactivity by boiled cells was used as a blank.

462 To determine metabolites produced from the provided labeled amino acid, samples of 0.5

463 mL of the cell suspension were immediately (<15 s) mixed, without filtering the cells, with 1.5

464 mL of water at 100 °C and further incubated for 5 min in a bath of boiling water. The resulting

465 suspensions were centrifuged, and samples from the supernatants were analyzed by thin-layer

21 466 chromatography (TLC) and electronic autoradiography as described previously (Burnat and

467 Flores, 2014).

468

469 Expression and purification of Anabaena Alr4995

470 Plasmids pCSMI81 and pCSMI82 contain the alr4995 gene and the 5’ terminal part of this gene

471 encoding the AT/ADI domain, respectively, fused in frame to a sequence encoding a 6xHis tag

472 under an IPTG inducible promoter. These plasmids were transferred to E. coli BL21-lacIq. A

473 pre-inoculum of this strain grown overnight in LB medium supplemented with 50 μg of

474 ampicillin (Ap) mL−1 (50 μg of Km mL−1 for pCSMI94) and 1 % glucose was washed with LB

475 medium and used to inoculate 0.5 L of LB medium + Ap or Km. The culture was incubated at 37

476 °C up to an OD600 of 0.6, and protein expression was induced by addition of 1 mM isopropyl-β-

477 D-1-thiogalactopyranoside (IPTG). After 2-3 h at 37 °C, cells were collected by centrifugation at

478 4,000 x g (10 min, 4º C), washed with a buffer containing 100 mM Tris-HCl (pH 8.0) and 150

479 mM NaCl and centrifuged at 7,000 g (10 min, 4 ºC). Cells were resuspended with lysis buffer

480 (200 mM Tris-HCl [pH 8.0], 500 mM NaCl, 5% glycerol, and 5 mM imidazole) at 5 mL g−1 of

481 cells. DNase I and protease inhibitor cocktail complete Mini EDTA-free (Roche) were added just

482 before breakage of the cells by passage twice through a French pressure cell at 20,000 psi. After

483 centrifugation at 27,216 g (30 min, 4 °C), the 6xHis-Alr4995 or 6xHis-AT/ADI Alr4995 protein

484 was purified from the supernatant by chromatography through a 5-mL His-Select column from

485 Sigma, using imidazole in the same buffer described above to elute the retained proteins.

486 Samples obtained after purification were subjected to SDS-PAGE to confirm the presence of the

487 proteins (see Fig. S8A, B).

488 Anabaena strain CSMI40 carries pCSMI95 in which alr4995 is cloned fused to a

22 489 sequence encoding a Strep-Tag II as described above. Cyanobacterial cells were incubated in 50

490 mL BG11 medium supplemented with 5 µg Em mL-1 for 5-7 days, and then, transferred to 800

491 mL of bubbled BG11 medium supplemented with 2 µg Em mL-1, and incubated for 5-7 days.

492 After harvesting by filtration, cells were resuspended in 50 mM Tris-HCl (pH 8), 200 mM NaCl

493 and 20 % glycerol. Protease inhibitor cocktail complete Mini EDTA-free (Roche) was added just

494 before breakage of the cells by passage twice through a French pressure cell at 20,000 psi. After

495 centrifugation at 27,216 g (30 min, 4 °C), soluble fractions were subjected to purification using

496 StrepTrap HP columns and 0.53 mg mL-1 desthiobiotin in the same buffer as above as eluent (GE

497 Healthcare), and the protein was desalted using PD-10 columns (see Fig. S8C). After initial

498 checking of different conditions, 1 mM NAD+ and 10 mM dithiothreitol (DTT) were added to

499 lysis and purification buffers.

500

501 In vitro enzymatic assays of Alr4995

502 To study the enzymatic activity in vitro, L-arginine and L-ornithine supplied at different

503 concentrations were used as substrates. Different buffers and co-factors were tried: 100 mM Tris-

504 HCl (pH 9) buffer, 100 mM phosphate buffer (pH 7.5), 10 mM dithiothreitol (DTT), 10 mM β-

505 mercaptoethanol, and NAD+ at different concentrations. Anoxic conditions obtained by flushing

506 argon through solutions were also tried. Substrates and products were analyzed by HPLC, except

507 urea that was determined colorimetrically (Boyde and Rahmatullah, 1980). The HPLC method

508 used by default involved a derivatization of amino acids with phenylisothiocyanate (PITC)

509 (Heinrikson and Meredith, 1984) and was performed as described (Burnat et al., 2014). An

510 alternative HPLC method involved derivatization with 9-fluorenylmethyl-chloroformate and was

511 performed as described by Fabiani et al. (2002).

23 512

513 Acknowledgements

514 We thank Dorothee Heinemann for her participation in the construction of the agrE mutant,

515 Ignacio Luque (CSIC, Sevilla, Spain) for useful suggestions, William B. Whitman (University of

516 Georgia, Athens, GA) for a sample of DNA from Methanococcus maripaludis, and Anthony J.

517 Michael (University of Texas Southwestern Medical Center, Dallas, TX), Vicente Rubio (CSIC,

518 Valencia, Spain) and C. Peter Wolk (Michigan State University, East Lansing, MI) for critical

519 reading of the manuscript. We also thank the editor, James A. Imlay, for excellent suggestions.

520 Work was supported by grant nos. BFU2014-56757-P and BFU2017-88202-P (EF) and

521 BFU2016-77097 (AH) from the Spanish Government, co-financed by the European Regional

522 Development Fund. The authors have no conflict of interest to declare.

523

524 Author contributions

525 MB designed and performed work related to AgrE and drafted parts of the manuscript; SP

526 designed and performed work related to PutA; AV performed biochemical work related to AgrE;

527 AH designed and supervised research; EF conceived the study, designed and supervised

528 research, and wrote the manuscript; all authors analyzed results and made manuscript revisions.

529

24 530 References

531 Boyde TRC, Rahmatullah M (1980) Optimization of conditions for the colorimetric determination of citrulline, 532 using diacetyl monoxime. Anal Biochem 107(2):424-431. 533 Burnat M, Flores E (2014) Inactivation of agmatinase expressed in vegetative cells alters arginine catabolism and 534 prevents diazotrophic growth in the heterocyst-forming cyanobacterium Anabaena. MicrobiologyOpen 3(5):777- 535 92. 536 Burnat M, Herrero A, Flores E (2014) Compartmentalized cyanophycin metabolism in the diazotrophic filaments of 537 a heterocyst-forming cyanobacterium. Proc Natl Acad Sci USA 111(10):3823-3828.

538 Burnat M, Li B, Kim SH, Michael AJ, Flores E (2018) Homospermidine biosynthesis in the 539 cyanobacterium Anabaena requires a deoxyhypusine synthase homologue and is essential for normal 540 diazotrophic growth. Mol Microbiol 109(6):763-780. 541 Cai YP, Wolk CP (1990) Use of a conditionally lethal gene in Anabaena sp. strain PCC 7120 to select for double 542 recombinants and to entrap insertion sequences. J Bacteriol 172(6):3138-3145. 543 Costilow RN, Laycock L (1971) Ornithine cyclase (deaminating). Purification of a protein that converts ornithine to 544 proline and definition of the optimal assay conditions. J Biol Chem 246:6655-6660. 545 Elhai J, Vepritskiy A, Muro-Pastor AM, Flores E, Wolk CP (1997) Reduction of conjugal transfer efficiency by 546 three restriction activities of Anabaena sp. strain PCC 7120. J Bacteriol 179(6):1998-2005. 547 Fabiani A, Versari A, Parpinello GP, Castellari M, Galassi S (2002) High-performance liquid chromatography 548 analysis of free amino acids in fruit juices using derivatization with 9-fluorenylmethyl-chloroformate. J 549 Chromatogr Sci 40(1):14-18. 550 Flores E, Herrero A (2010) Compartmentalized function through cell differentiation in filamentous cyanobacteria. 551 Nat Rev Microbiol 8(1):39-50. 552 Goodman JL, Wang S, Alam S, Ruzicka FJ, Frey PA, Wedekind JE (2004) Ornithine cyclodeaminase: structure, 553 mechanism of action, and implications for the mu-crystallin family. Biochemistry 43:13883-13891. 554 Graupner M, White RH (2001) Methanococcus jannaschii generates L-proline by cyclization of L-ornithine. J 555 Bacteriol 183(17):5203-5. 556 Heinrikson RL, Meredith SC (1984) Amino acid analysis by reverse-phase high-performance liquid 557 chromatography: precolumn derivatization with phenylisothiocyanate. Anal Biochem 136(1):65-74. 558 Kaneko T, Nakamura Y, Wolk CP, Kuritz T, Sasamoto S, Watanabe A, Iriguchi M, Ishikawa A, Kawashima 559 K, Kimura T, Kishida Y, Kohara M, Matsumoto M, Matsuno A, Muraki A, Nakazaki N, Shimpo S, Sugimoto 560 M, Takazawa M, Yamada M, Yasuda M, Tabata S (2001) Complete genomic sequence of the filamentous 561 nitrogen-fixing cyanobacterium Anabaena sp. strain PCC 7120. DNA Res 8(5): 227-253. 562 Liu LK, Becker DF, Tanner JJ (2017) Structure, function, and mechanism of proline utilization A (PutA). Arch 563 Biochem Biophys 632:142-157.

25 564 López-Igual R, Lechno-Yossef S, Fan Q, Herrero A, Flores E, Wolk CP (2012) A major facilitator superfamily 565 protein, HepP, is involved in formation of the heterocyst envelope polysaccharide in the cyanobacterium 566 Anabaena sp. strain PCC 7120. J Bacteriol 194 (17): 4677-4687. 567 Lu CD (2006) Pathways and regulation of bacterial arginine metabolism and perspectives for obtaining arginine 568 overproducing strains. Appl Microbiol Biotechnol 70:261-272. 569 Mackinney G (1941) Absorption of light by chlorophyll solutions. J Biol Chem 140:315-322. 570 Markwell MAK, Hass SM, Bieber LL, Tolbert NE (1978) A modification of the Lowry procedure to simplify 571 protein determination in membrane and lipoprotein samples. Anal Biochem 87: 206-210. 572 Merino-Puerto V, Herrero A, Flores E (2013) Cluster of genes that encode positive and negative elements 573 influencing filament length in a heterocyst-forming cyanobacterium. J Bacteriol 195(17): 3957-3966. 574 Morris SM Jr (2016) Arginine Metabolism Revisited. J Nutr 146(12):2579S-2586S. 575 Napolitano M, Rubio MA, Camargo S, Luque I (2013) Regulation of internal promoters in a zinc-responsive operon 576 is influenced by transcription from upstream promoters. J Bacteriol 195 (6):1285-1293. 577 Nürnberg DJ, Mariscal V, Bornikoel J, Nieves-Morión M, Krauß N, Herrero A, Maldener I, Flores E, Mullineaux 578 CW (2015) Intercellular diffusion of a fluorescent sucrose analog via the septal junctions in a filamentous 579 cyanobacterium. mBio 6(2):e02109. 580 Pernil R, Picossi S, Mariscal V, Herrero A, Flores E (2008) ABC-type amino acid uptake transporters Bgt and N-II 581 of Anabaena sp. strain PCC 7120 share an ATPase subunit and are expressed in vegetative cells and heterocysts. 582 Mol Microbiol 67(5):1067-80. 583 Picossi S, Montesinos ML, Pernil R, Lichtlé C, Herrero A, Flores E (2005) ABC-type neutral amino acid permease 584 N-I is required for optimal diazotrophic growth and is repressed in the heterocysts of Anabaena sp. strain PCC 585 7120. Mol Microbiol 57(6):1582-92. 586 Quintero MJ, Muro-Pastor AM, Herrero A, Flores E (2000) Arginine catabolism in the cyanobacterium 587 Synechocystis sp. strain PCC 6803 involves the urea cycle and arginase pathway. J Bacteriol 182(4):1008-15. 588 Rippka R, Deruelles J, Waterbury JB, Herdman M, Stanier R (1979) Generic assignments, strain histories and 589 properties of pue cultures of cyanobacteria. J Gen Microbiol 111(1): 1-61. 590 Sarmiento F, Mrázek J, Whitman WB (2013) Genome-scale analysis of gene function in the hydrogenotrophic 591 methanogenic archaeon Methanococcus maripaludis. Proc Natl Acad Sci USA 110(12):4726-31. 592 Schriek S, Rückert C, Staiger D, Pistorius EK, Michel KP (2007) Bioinformatic evaluation of L-arginine catabolic 593 pathways in 24 cyanobacteria and transcriptional analysis of genes encoding enzymes of L-arginine catabolism 594 in the cyanobacterium Synechocystis sp. PCC 6803. BMC Genomics 8:437. 595 Shirai H, Mokrab Y, Mizuguchi K (2006) The guanidino-group modifying enzymes: structural basis for their 596 diversity and commonality. Proteins 64(4):1010-23.

597 Thomas J, Meeks JC, Wolk CP, Shaffer PW, Austin SM (1977) Formation of glutamine from [13N]ammonia,

598 [13N]dinitrogen, and [14C]glutamate by heterocysts isolated from Anabaena cylindrica. J Bacteriol 129(3):1545- 599 55.

26 600 Valladares A, Montesinos ML, Herrero A, Flores E (2002) An ABC-type, high-affinity urea permease identified in 601 cyanobacteria. Mol Microbiol 43(3):703-15. 602 Valladares A, Rodríguez V, Camargo S, Martínez-Noël GMA, Herrero A, Luque I (2011) Specific role of the 603 cyanobacterial PipX factor in the heterocysts of Anabaena sp. strain PCC 7120. J Bacteriol 193(5):1172-1182. 604 Vander Wauven C, Stalon V (1985) Occurrence of succinyl derivatives in the catabolism of arginine in 605 Pseudomonas cepacia. J Bacteriol 164:882–886.

606 Xu X, Gu L, He P, Zhou R (2015) Characterization of five putative aspartate aminotransferase genes in the N2- 607 fixing heterocystous cyanobacterium Anabaena sp. strain PCC 7120. Microbiology 161(6):1219-1230. 608 Zhang H, Liu Y, Nie X, Liu L, Hua Q, Zhao GP, Yang C (2018) The cyanobacterial ornithine-ammonia cycle 609 involves an arginine dihydrolase. Nat Chem Biol 14(6):575-581. 610

27 611 Figure legends

612

613 Fig. 1. Catabolism of [14C]arginine, [14C]ornithine and [14C]proline in Anabaena. (A, B)

614 Visualization of catabolic products. Filaments of Anabaena grown in BG11 medium were

615 incubated with 1 µM L-[U-14C]arginine (A) or L-[1-14C]ornithine (B) for 10 min and extracted.

616 Samples were subject to thin layer chromatography (TLC) and visualized by electronic

617 autoradiography as described in Experimental procedures.. Empty ovals indicate the approximate

618 locations of the spots of Pro and Glu (not produced in the corresponding mutant). Note that

619 citrulline and arginine are synthesized from ornithine in the agrE mutant. (C) Schematic

620 showing the putative metabolic path of the labeled amino acids. (D) Catabolism of 1.9 µM L-[U-

621 14C]proline in Anabaena wild type and the putA::C.S3 mutant (incubation time, 15 min).

622 Arrowheads point to the origin of the chromatography.

623

624 Fig. 2. AgrE of Anabaena (Alr4995) is a 703-amino acid residue protein containing two well

625 differentiated domains: an N-terminal / propeller domain (amino acid residue 10-269) and a

626 C-terminal domain homologous to Methanococcus LOR/SDH protein 2 (amino acid residue 285-

627 694), here identified as an ornithine cyclodeaminase and in which two further homology domains

628 are detected. The proposed reactions of the AgrE protein and their stoichiometry are shown.

629

630 Fig. 3. Production of 14C-labeled metabolites from L-[14C]arginine and L-[1-14C]ornithine in

631 filaments of strain CSMI30b (argF::C.K3). Suspensions of filaments grown in BG11 medium

632 and containing 5 to 10 μg of Chl·ml-1 were incubated for 10 min with 1 μM L-[14C-(U)]arginine

633 (A) or 1 μM L-[1-14C-(U)]ornithine (B). Metabolites in the cell suspensions were extracted and

28 634 analyzed by TLC and autoradiography as described in Experimental procedures. The amino

635 acids identified were: Arg, Cit, Pro, Glu, Gln and Orn. Note that the glutamine and citrulline

636 spots overlap. White triangles point to the origin of the chromatography. (C) Alternative pathway

637 via citrulline that is ruled out by analysis of the argF mutant (i.e., if this pathway were operative

638 in Anabaena, [14C]proline and [14C]glutamate would not be produced from [14C]ornithine in the

639 argF mutant).

640

641 Fig. 4. Production of 14C-labeled metabolites from [14C]arginine or [14C]ornithine in filaments of

642 the Anabaena agrE mutant complemented with the Anabaena ArgE protein (A, B) or with the

643 Methanococcus Mls2 protein (C, D). Filaments grown in BG11 medium supplemented with 2 µg

644 Em/mL were incubated for 10 min with 1 μM L-[U-14C]arginine (A, C) or L-[1-14C]ornithine (B,

645 D). Metabolites in the cell suspensions were extracted and analyzed by TLC and autoradiography

646 as described in Experimental procedures. The amino acids identified were: Arg, Cit, Pro, Glu,

647 Gln and Orn. Note that the glutamine and citrulline spots overlap. Arrowheads point to the origin

648 of the chromatography. Empty ovals indicate the approximate locations of the spots of Pro or

649 Glu (not produced in the indicated mutant).

650

651 Fig. 5. In vitro activity of the Anabaena AgrE protein and of its N-terminal domain. The

652 indicated proteins were produced and isolated from E. coli or Anabaena as described in

653 Experimental procedures and Fig. S8. The reactions were carried out in 100 mM phosphate

654 buffer (pH 7.5) supplemented with the indicated substrate and, when indicated, 10 mM -

655 mercaptoethanol and 1 mM NAD+, and the reaction products were analyzed by HPLC after

29 656 derivatization with phenylisothiocyanate (Heinrikson and Meredith, 1984). The molar ratios of

657 products detected are indicated.

658

659 Fig. 6. Whole activity of the Strep-tagged AgrE protein isolated from Anabaena. The protein

660 (10.75 µg mL-1) was assayed in 100 mM phosphate buffer (pH 7.5) supplemented with 1 mM

661 NAD+, 1 mM -mercaptoethanol and 40 mM L-arginine. (A) HPLC analysis of the 60-min

662 reaction products after derivatization with 9-fluorenylmethyl-chloroformate (Fabiani et al.,

663 2002). Arg, substrate; FMOC, 9-fluorenylmethyl-chloroformate; React., unknown reactant;

664 Proline (Pro*) and ornithine (Orn*) are noted to partially overlap reactants. (B) Quantification of

665 the reaction products after subtraction of time zero values. The reaction rates were calculated

666 from the slope of the lines: 29.20 nmol ammonium min-1, 42.51 nmol ornithine min-1 and 25.77

667 nmol proline min-1.

668

669 Fig. 7. Physiology of Anabaena (WT) and mutant strains CSMI43 (agrE) and CSS5

670 (putA::C.S3). (A) Growth rate constants determined in liquid media. Mean and SD (n = 3 to 6).

671 nd, not determined. (B) Filaments from cultures incubated for 6 days in BG11 medium and

672 visualized by light microscopy. Note the presence of increased granulation in strain CSMI43.

673 Magnification, 100x.

674

675 Fig. 8. Study of an Anabaena agrE ΔspeA::C.S3 double mutant. (A) Catabolism of 1 µM

676 [14C]arginine for 10 min in wild-type Anabaena and the agrE ΔspeA::C.S3 double mutant, and

677 analysis by TLC of labeled products. Incubation with wild-type filaments that had been subjected

678 to boiling is shown as a negative control. (B) Growth tests of the indicated strains on solid

30 679 medium with ammonium (BG110NH4+), nitrate (BG11) or N2 (BG110) as the nitrogen source.

680 Each spot was inoculated with an amount of cells containing the indicated amount of chlorophyll

681 a (Chl), and the plates were incubated under culture conditions for 10 days and photographed.

682

683 Fig. 9. Expression of agrE and putA. (A) Cellular localization of AgrE in the filaments of

684 Anabaena. Filaments of strain CSMI32 (alr4995-sf-gfp) grown in bubbled BG110 medium for 24

685 h or 48 h, were visualized by confocal microscopy and their GFP fluorescence was quantified.

686 Average background fluorescence from wild-type cells (lacking sf-GFP) was subtracted. VC,

687 vegetative cell; H, heterocyst. (B) Quantification of the sf-GFP fluorescence in cells of strain

688 CSMI32. Average background fluorescence from wild type cells was subtracted. Figures are the

689 mean and standard deviation of the mean of the fluorescence recorded in cells grown in bubbled

690 BG110 medium for 24 h (365 vegetative cells and 39 heterocysts counted; Student’s t test p = 5.1

691 x 10-37) or 48 h (367 vegetative cells and 37 heterocysts counted; Student’s t test p = 8.3 x 10-58).

692 (C) Northern analysis of the expression of the alr0540 gene in Anabaena cells grown with

693 ammonium (NH4+), nitrate (NO3-) or N2 as the nitrogen source, as well as in isolated heterocysts

694 (Het). The probe used is shown in the top scheme. The arrowhead points to a transcript of about

695 3.6 kb. As loading control, the membrane was hybridized with a rnpB probe. As controls for the

696 quality of the RNA isolated from heterocysts, the same membrane was hybridized with a nifH

697 and a rbcL probe. (Data of the heterocyst controls are from Picossi et al., 2005.)

698

699 Fig. 10. The arginine metabolism pathways of Anabaena. The homospermidine biosynthesis

700 pathway is represented to the left and the arginine-glutamate catabolic pathway to the right.

701 SpeA, arginine decarboxylase (all3401 gene product); SpeB, agmatinase (alr2310); SpeY,

31 702 deoxyhypusine synthase-like protein (alr3804); AgrE, arginine-guanidine removing enzyme

703 (alr4995); PutA, bifunctional proline oxidase (alr0540) that carries out two sequential reactions:

704 proline dehydrogenase and glutamate--semialdehyde dehydrogenase; [H], reducing power,

705 predictably in the form of FADH2 (product of the proline dehydrogenase reaction) and NADPH

706 (product of the glutamate--semialdehyde dehydrogenase reaction). The positions of ornithine

707 (Orn) as an intermediate in AgrE and of 1-pyrroline-5-carboxylate/glutamate--semialdehyde

708 (P5C/GSA) as intermediates in PutA are indicated.

709

710

32 711 Table 1. Activities of the AgrE proteina.

Assays Enzyme Substrate Additions/conditions Products Ratio set prep (nmol/µg protein)

NH4+ Orn Pro NH4+/ NH4+/ Orn Pro

1 AgrEb 40 mM Arg - 323.5 200.2 1.62 0.5 mM NAD+ 390.1 203.6 1.92

2 AgrEb 40 mM Arg - 304.3 173.4 1.75 4 mM Arg - 139.4 80.2 1.74

3 AT/ADI 40 mM Arg - 90.12 48.85 1.84 domainc

4 AgrEd 40 mM Arg - 214.8 112.6 1.91

40 mM Orn 10 mM β-m-etOH + 1 mM NAD+ 17.31 13.05 1.33 1 mM FAD+ 0 0.33 - 1 mM NADP+ 4.26 2.72 1.57 1 mM PLP 0 0 -

5 AgrEd 10 mM β-m-etOH + 0.1 mM DTT + 1 mM NAD+ + 1 mM Orn - 0 0.96 - 1 mM Orn 5 mM ATP 0 0 - 10 mM Orn - 27.87 26.56 1.05 10 mM Orn 5 mM ATP 0 4.14 - 50 mM Orn - 86.29 74.33 1.16 50 mM Orn 5 mM ATP 41.31 37.71 1.10

6 AgrEd 40 mM Arg 10 mM β-m-etOH + 188.0 77.15 2.44f 1 mM NAD+, 37 ºC

7e AgrEd 40 mM Arg 1 mM β-m-etOH + 162.98 42.51 25.77 2.0g 3.0g 1 mM NAD+

712 a Assays were performed with 100 mM phosphate buffer (pH 7.5) at 30 ºC. Exceptions to these standard conditions 713 are described in the table. No activity was found at pH 9. Figures for NH4+, ornithine (Orn) and proline (Pro) 714 correspond to products detected by HPLC 60 min after the addition of the substrate. DTT, dithiothreitol; β-m-etOH, 715 β-mercaptoethanol; PLP, pyridoxal 5-phosphate. 716 b His-tagged AgrE protein isolated from E. coli (see Fig. S8A). 717 c His-tagged AgrE N-terminal (AT/ADI) domain isolated from E. coli (see Fig. S8B). 718 d Strep-tagged AgrE protein isolated from Anabaena (see Fig. S8C). The protein used in Assays set 5, 6 and 7 was 719 isolated in the presence of 10 mM DTT and 1 mM NAD+. 720 e The products of this assay were analyzed by the derivatization and HPLC methods described by Fabiani et al. 721 (2002). 722 f This figure exceeds the stoichiometry of 2 mol NH4+/mol of ornithine expected for the arginine dihydrolase 723 reaction, which may imply that the reaction proceeded further to proline releasing extra ammonium. 724 g The amount of NH4+ observed in this assay (162.98 nmol/µg protein) matched that expected from the detected 725 amounts of ornithine (42.51) and proline (25.77), which together make 162.33 nmol/µg protein according to the 726 equation: nmol NH4+ = 2 [nmol ornithine] + 3 [nmol proline]. 727

33 A [14C]Arginine B [14C]Ornithine C WT WT Pro Pro [14C]Arginine Orn Arg

Glu Glu [14C]Ornithine

ΔagrE ΔagrE AgrE Orn

Arg Cit Arg

[14C]Proline putA::C.S3 putA::C.S3 Pro Pro Orn PutA Arg

[14C]Glutamate

D [14C]Proline

WT putA::C.S3

Pro Pro

Gln

Glu a/b propeller LOR/SDH-like protein domain N C LOR/SDH NAD-dependent n-domain DHS

Arg dihydrolase Ornithine cyclodeaminase Arg Orn Pro

+ + 2 NH4 + CO2 NH4

+ mol NH4 produced = 2 x (mol ornithine) + 3 x (mol proline) A [14C]Arginine C

[14C]Arginine [14C]Ornithine Pro Orn ArgF Gln/Cit Arg [14C]Citrulline Glu AgrE

B [14C]Ornithine

[14C]Proline Pro Orn PutA

Glu [14C]Glutamate 14 A [ C]Arginine B [14C]Ornithine

ΔagrE + AgrE ΔagrE + AgrE

Pro Pro

Orn Orn Gln/Cit Gln/Cit Arg Arg

Glu Glu

14 C [ C]Arginine D [14C]Ornithine

ΔagrE + Mls2 ΔagrE + Mls2

Pro

Arg Orn

Glu A His-tagged a/b propeller domain B His-tagged whole AgrE protein

6xHis a/b propeller 6xHis a/b propeller LOR/SDH-like protein tag domain tag domain N C N C

40 mM L-arginine substrate; 40 mM L-arginine substrate; protein isolated from E.coli. protein isolated from E.coli. 0 min 0 min Arg Arg ) ) units units arbitrary ( arbitrary ( 60 min 60 min Arg Arg Orn Orn

+ 1.84 NH /Orn Fluorescence + Fluorescence 4 1.75 NH4 /Orn NH + + 4 NH4

0 10 20 30 (min) 0 10 20 30 (min)

Strep-tagged whole AgrE protein Strep II a/b propeller LOR/SDH-like protein tag domain N C Muestra: 5 0 Vial: 6 Vol iny. 10 µL C Secuencia: C:\EZChrom Elite\Enterprise\Projects\Aminoacidos D PICT\Sequence\parada 1'0.seq Usuario: Carlos 40 mM L-arginine substrate; 40 mM L-ornithine substrate + 1 mM NAD+ + Fecha: 06/07/2017 9:55:11 protein isolated from Anabaena 10 mM b-m-etOH + protein from Anabaena

2000 2000 0 min 0 min Arg Orn 1500 1500 ) ) 1000 1000 mAU mAU units Muestra: 5 60 units Vial:500 16 500 Vol iny. 10 µL Secuencia: C:\EZChrom Elite\Enterprise\Projects\Aminoacidos PICT\Sequence\parada 1'0.seq Usuario:0 Carlos 0 Fecha: 06/07/2017 9:54:03 arbitrary arbitrary ( ( 0 5 10 15 20 25 30 Minutes 2000 60 min 2000 60 min Arg Orn Orn

1500 1500

+

Fluorescence 1000 1000 1.33 NH /Pro Fluorescence mAU mAU 4 1.91 NH +/Orn NH + 4 500 4 500 + Pro NH4

0 0

0 5 10 15 20 25 30 0 10 Minutes 20 30 (min) 0 10 20 30 (min) Muestra: arg s9 60 Vial: 3 Vol iny. 10 µL Secuencia: C:\EZChrom Elite\Enterprise\Projects\Aminoacidos PICT\Sequence\parada 1'0.seq Usuario: Carlos Fecha: 12/11/2018 12:49:01 A 500 500 Arg FMOC ) 400 Orn* 400

units 300 300 mAU mAU

200 200 React. Fluorescence arbitrary 100 100 ( + Pro* NH4

0 0 0 5 10 15 20 25 30 0 10 Minutes 20 30 (min)

B 2000 + NH4

1500 (µM)

1000 product

Reaction Orn 500

Pro

0 0 20 40 60 80 Reaction time (min)

-500 A Growth rate, µ (day-1)

+ BG110NH4 BG11 (Nitrate) BG110 (N2) Set 1 PCC 7120 (WT) 0.53 ± 0.16 0.86 ± 0.19 0.87 ± 0.56 CSMI43 (ΔagrE) 0.56 ± 0.08 0.91 ± 0.42 0.83 ± 0.51 Set 2 PCC 7120 (WT) nd 0.86 ± 0.10 0.50 ± 0.17 CSS5 (putA::C.S3) nd 0.82 ± 0.04 0.36 ± 0.05

B PCC 7120 CSMI43 (WT) (DagrE) A Anabaena WT ΔagrE ΔspeA::C.S3 Boiled filaments

Agm Pro

Orn Arg Gln/Cit Arg Arg Glu

B + BG110NH4 BG11 BG110 PCC 7120 (WT)

ΔagrE ΔspeA::C.S3

ΔagrE

ΔspeA::C.S3

(ng Chl) 10 5 2.5 1.25 0.6 10 5 2.5 1.25 0.6 10 5 2.5 1.25 0.6 A 70 C 60 50 alr0540 alr0541 40

fluorescence 30 PA-1 PA-2 20 2.3-kb probe 10 Relative Het N NO - NH + 0 2 3 4 putA VC VC VC VC H VC VC VC VC

B 120 vegetative cells heterocysts 100

80 rnpB

60 fluorescence nifH 40

Relative rbcL 20

0 24 h -N 48 h -N Arginine SpeA

+ 2 NH4 + CO2 Agmatine CO 2 AgrE Orn NH + SpeB 4 Urea Proline

Putrescine [H] Putrescine SpeY PutA P5C/GSA

+ [H] NH4 Sym-homospermidine Glutamate