Identification and functional characterization of two heme uptake systems required for virulence of the fish pathogen Flavobacterium psychrophilum

Flavobacterium psychrophilum, a member of the phylum Bacteroidetes, causes bacterial cold-water disease in salmonids worldwide. Infected fish exhibit tissue damages and develop hemorrhagic septicemia. The molecular factors involved in host colonization are mostly unknown. In this study, we identified two TonB-dependent receptors that are required for iron homeostasis and virulence in rainbow trout (Oncorhynchus mykiss). Transcriptional analyses revealed that their expression is tightly controlled and upregulated under specific iron sources and concentrations. Characterization of single and double deletion mutants showed that the two receptors act without redundancy: BfpR is required for optimal growth in the presence of high hemoglobin level, while HfpR confers the capacity to acquire nutrient iron from heme or hemoglobin under iron scarcity. The gene hfpY, co-transcribed with hfpR, encodes a protein belonging to the HmuY family but with low sequence conservation compared to other members of this family. We demonstrated that HfpY is able to bind heme in vitro and contributes significantly to host colonization and disease severity in experimentally infected rainbow trout. Overall, these results are consistent with a model in which both BfpR and Hfp systems promote heme uptake and respond to distinct signals to adapt iron acquisition to the different stages of pathogenesis. Our findings give insight into the molecular basis of F. psychrophilum pathogenicity and point to the newly identified heme outer membrane receptors as promising targets for antibacterial development. Author summary Infectious diseases are a major threat for human and animal health. Bacterial pathogens impact aquaculture, a sector that accounts for half of the human fish consumption, and mechanistic insights into pathogenesis allow for improved control strategies. Pathogens have to retrieve essential nutrients to multiply. Host-derived heme constitutes the largest iron source during infection and pathogens have evolved fine-tuned iron acquisition systems to overcome host scavenging mechanisms while preventing iron toxicity. These mechanisms are poorly understood in pathogens of the family Flavobacteriaceae that have a critical impact on the fish farming, animal welfare and environment. Here, we report the identification and characterization of two heme uptake systems important for adaptation under distinct iron sources conditions encountered during infection and for virulence and host colonization in Flavobacterium psychrophilum, a serious pathogen of salmonids.

174 fragments (306, 609 and 950 bp) of the upstream region of bfpR were cloned into a transcriptional 175 reporter plasmid and a significant promoter activity was detected only for the one carrying the 176 longest fragment when colonies were grown in TYES agar supplemented with 1% hemoglobin (Fig   177 2B). These results indicate that the majority of bfpR mRNAs are monocistronic and may originate 178 from processing of a longer transcript starting 673 bp upstream of bfpR. 179 We therefore compared the effect of hemoglobin supply on bfpR mRNA level by RT-qPCR 180 assays (Fig 2C). Addition of large amounts of hemoglobin in TYES broth led to 10-fold increase 181 of bfpR mRNA level. This increase was not observed for the upstream gene THC0290_1678.
182 Altogether, we concluded that a primary transcript encompassing THC0290_1678 and bfpR is 183 synthetized then probably subject to maturation, resulting in the bfpR mRNA that accumulates in 184 presence of large amounts of hemoglobin.

185
186 BfpR contributes to growth in the presence of high amounts of 187 hemoglobin 188 On the basis of its expression profile, we analyzed the role of BfpR on the growth of F.
189 psychrophilum under high-hemoglobin condition. The wild-type strain exhibited a slight delay in 190 the presence of 1% hemoglobin (equivalent of 620 µM heme), but final viable cell counts were 191 higher than in TYES broth (F5 in Supporting Information), indicating that hemoglobin availability 192 can enhance bacterial growth. In these conditions, the growth rate of bfpR was reduced compared 193 to that of the wild-type and partially restored in the complemented bfpR strain (Fig 3).  showed that ectopic expression of bfpR in this complemented bfpR strain resulted in reduced 195 mRNA level compared to the wild-type (F6 in Supporting Information), indicating that adequate 197 hfpRbfpR and bfpR were similar. These results showed that among the two TBDRs, only BfpR 198 is required for optimal growth in the presence of high amounts of hemoglobin. 219 Expression of the hfp locus was then monitored under iron-depleted and -replete conditions 220 using a transcriptional reporter (P hfpR -mCh) and RT-qPCR assays. Addition of the iron-chelator 221 ethylenediamine-N,N′-bis (2-hydroxyphenylacetic acid) (EDDHA) in TYES broth resulted in a 222 strong increase of promoter activity, which was countered by FeCl 3 supply in a dose-dependent 223 manner (Fig 4B). Promoter activity was also limited by heme or hemoglobin supply, although with 224 a lower efficiency than when FeCl 3 was used ( Fig 4B). RT-qPCR confirmed up-regulation of the 225 3 genes composing the hfp operon under iron limitation and the effect of hemoglobin supply (Fig   226 4C). Moreover, a 5-fold down-regulation was measured in the presence of high level of hemoglobin 227 in iron-replete TYES broth (Fig 4C), which is consistent with the down-regulation previously 228 observed by microarrays in the presence of blood [31].

229
We concluded from these results that 2 mRNAs are produced at the hfp locus: a primary 230 transcript encompassing hfpR, THC0290_1813 and hfpY, and a matured mRNA carrying hfpY only, 231 resulting in more abundant mRNA for hfpY than for hfpR. The expression level of these 2 RNAs is 232 tightly controlled in response to iron availability through the regulation of the hfpR promoter which 233 is activated by iron scarcity and downregulated in presence of iron sources.

234
235 HfpR and HfpY are required for heme utilization under iron scarcity 236 The effect of iron limitation on growth was first analyzed on the wild-type strain. Addition 237 of EDDHA, a chelator of ferric iron Fe(III), to TYES broth resulted in a strong growth inhibition, 238 confirming that iron is an essential element for F. psychrophilum. Growth was restored by the 239 supply of hemoglobin, hemin or FeCl 3 , indicating that the bacterium is able to scavenge these 240 different iron sources from the environment (Fig 5). We then analyzed the role of HfpR and HfpY by comparing the growth of deletion mutants 242 and respective complemented strains under iron scarcity (Fig 5). Supplementation by FeCl 3 fully 243 restored the growth of all strains (Fig 5). In contrast, growth inhibition was only partially restored 244 by hemoglobin and hemin supplementation in mutant hfpR which still exhibited a severe growth 245 delay compared to complemented hfpR and wild-type strains. hfpY was by far less affected but 246 its growth rate still significantly reduced compared to the wild-type in hemoglobin supplemented 247 medium containing EDDHA, and this growth defect was partially restored in the complemented 248 hfpY strain. In contrast, no growth defect was observed in hfpY when hemin was supplied instead 249 of hemoglobin ( Fig 5). Partial growth restoration in the ectopic complemented hfpY strain is likely 250 due to its reduced hfpY mRNA level compared to the wild-type as measured by RT

285
Survival curves were then compared in fry (mean weight, 5 g) after intramuscular injection 286 to quantity the virulence. At an infection dose corresponding to 10 LD50 of the wild-type strain, 287 100% of fish died rapidly in groups infected with the wild-type or complemented strains whereas 288 mortality was delayed and survival was significantly increased for fish infected with either hfpR 289 or hfpY (Fig 7B). A similar difference between strains was also observed using a lower infectious 290 dose (F10 in Supporting Information). Using this infection model, the survival curves of fish 291 inoculated with bfpR or the wild-type were similar. In addition, the levels of mortality observed 292 with hfpRbfpR and hfpR were similar (F10 in Supporting Information).

293
Since massive hemolysis may not be encountered in the fry infection model, we also 294 investigated the role of BfpR in virulence on larger fish. Two groups of juvenile rainbow trout 295 (mean weight, 65 g) were infected by intramuscular injection and the survival curves or bacterial 296 loads in blood, spleen and kidney 4-days post-infection were determined. In this model, bacteremia 297 and bacterial loads in spleen and kidney were consistently significantly lower in the group infected 298 with bfpR relative to the wild-type ( Fig 7C). The fish groups infected with bfpR displayed a 299 higher survival rate compared to those infected with the wild-type strain ( Fig 7D).

300
Altogether, the results showed that both Hfp and BfpR systems are necessary for full 301 virulence of F. psychrophilum in rainbow trout with no apparent functional redundancy.

304
Here, we report the identification of two heme/hemoglobin outer membrane receptors 305 conserved in the species and demonstrate that they act differently to support bacterial growth, under 306 high hemoglobin condition for BfpR and under iron scarcity for HfpR. We also show by 307 experimental infection in a natural host, the rainbow trout, that both receptors are required for full 308 virulence (Fig 8). 340 It is noteworthy that several hmuY transcripts are also produced in P. gingivalis, the monocistronic 341 form being more abundant than the bicistronic hmuY-hmuR one, suggesting a similar hemophore 342 upregulation in the two species despite distinct operonic structure [36]. In bacteria, the stability of 343 RNAs released by processing of a polycistronic RNA may differ, which affects the relative 344 expression levels of genes within a same operon. Our hypothesis is that a stable hfpY transcript is 345 released after maturation of hfpR-1813-hfpY mRNA, resulting in higher quantity of hfpY compared 346 to hfpR, independently of iron level. A similar regulation was recently described in Pseudomonas 347 aeruginosa for the hasR-hasA operon encoding the hemophore-dependent Has system [42].
348 Production of large amounts of hemophore likely provides a significant advantage to maintain 349 effective concentration relative to its cognate receptor while diffusing into the surrounding 350 environment.

373
In vitro, when ferric iron was deprived and heme or hemoglobin was the sole iron source, 374 the growth of F. psychrophilum was strongly impaired in the absence of HfpR, but only slowed in 375 the absence of HfpY. The importance of the hemophore was nevertheless obvious in vivo, the lack 376 of HfpY resulting in reduced mortality and bacterial loads in spleen. These results indicate that the 377 outer membrane receptor is able to independently capture and transport heme, although cooperation 378 with its cognate hemophore is needed in vivo, likely to outcompete the host sequestration systems 379 (Fig 8).   576 The THC0290_1813 transposition mutant was produced as part of a Tn4351 mutant library as 577 previously described [23,71]. Briefly, the pEP4351 suicide plasmid carrying