We first constructed the non-polar mutants ΔeryA and ΔeryH of B. abortus 2308. Using a chemically defined medium, we confirmed for ΔeryH (Barbier et al., 2014) and demonstrated for ΔeryA their inability to grow with erythritol as the sole carbon source (Supplementary Figure S1). We found no growth defect in a rich medium that lacked erythritol (2YT [10% yeast extract, 1% tryptone, 5% NaCl]) (Dozot et al., 2010), a result that makes broader metabolic defects in these mutants unlikely. We made the same observation using a ΔeryI mutant (not shown), which was as expected.

A critical requirement for using these mutants as reporters of erythritol availability was that ΔeryA and ΔeryH should be erythritol-tolerant and erythritol-sensitive, respectively. To prove these correlation, we grew ΔeryA and ΔeryH in 2YT medium that was supplemented with increasing concentrations of erythritol (Figure 1B). Whereas erythritol did not affect the growth of the mutant ΔeryA at the highest concentration tested, it markedly inhibited the growth of the mutant ΔeryH at a concentration as low as 0.05 g/L, which is in the inhibitory range for vaccine S19 (Keppie et al., 1967) and for an eryC mutant of B. suis 1330 (Burkhardt et al., 2005). Since we could restore the WT phenotype of the ΔeryH mutant by complementation (Figure 1B), we concluded that deletion of eryH caused both the inability to grow on erythritol and its toxicity. As expected from the activity of EryI (downstream of EryA, Figure 1A), the ΔeryI mutant was also inhibited by erythritol (not shown) and, as it phenocopied ΔeryH, below we present only the data obtained with the latter mutant.

Another critical requirement of any reporter system is specificity. To test the specificity, we studied the growth of WT and ΔeryH in 2YT medium supplemented with polyols (0.1 g/L) with structures close to erythritol (Supplementary Figure S2) that have been reported in male or pregnant female genital organs (Clark et al., 1967; Brusati et al., 2005; Jauniaux, 2005; Regnault et al., 2010; Larose et al., 2012). Since we did not detect any inhibitory effects for glycerol, ribitol, arabitol, xylitol, sorbitol, mannitol or dulcitol, the toxicity was specific for erythritol.

Since brucellae multiply intensively in bovine trophoblasts (Anderson and Cheville, 1986; Anderson et al., 1986a), we tested our reporting system in a suitable bovine trophoblastic cell line (Samartino et al., 1994). As shown in Figure 2A, no differences could be evidenced between the WT and the mutants at 2 h post infection (p.i.), a time when bacteria have not yet reached their replicative niche (Samartino et al., 1994). At later times, ΔeryH (erythritol-sensitive) but not ΔeryA (erythritol-tolerant) failed to multiply at the level of the WT strain. These results are consistent with the availability of erythritol in the replicative niche of B. abortus in bovine trophoblasts.

These observations and the recent description that Brucella can colonize human trophoblastic cell lines (Salcedo et al., 2013; Fernandez et al., 2016) prompted us to test our reporting system in trophoblastic cells other than those of bovine origin. When we infected human BeWo trophoblastic cells with B. abortus 2308 WT, ΔeryA or ΔeryH, we observed that the mutants were indistinguishable from the WT 2 h after infection and that only the multiplication of the erythritol-sensitive mutant ΔeryH was significantly reduced at later times (Figure 2B). These observations are in apparent conflict with previous studies that reported only very low erythritol concentrations in human fetal tissues (Keppie et al., 1965; Clark et al., 1967; Amin and Wilsmore, 1997). However, results obtained with cell lines do not necessarily reflect the in vivo situation and the parallelism with the results in bovine trophoblastic cells indicated that, while not being required for optimal bacterial replication, erythritol should be available in human BeWo cells at a concentration above that which is toxic for the sensitive mutant.

Macrophages are another cell type in which Brucella multiplies extensively. For this reason, RAW 264.7 murine macrophages and THP-1 human macrophage-like cells were infected with the erythritol catabolic mutants (Figure 3). Although both were able to invade these cells to the same extent as the WT (i.e., resulted in the same CFU/ml at 2 h p.i.), the erythritol-sensitive mutant was found in lower numbers 24 and 48 h after infection. These results are in agreement with those of Burkhardt et al. (2005) who showed that a B. suis 1330 ΔeryC mutant was erythritol-sensitive and attenuated in macrophages. Indeed, these authors also reported ΔeryC attenuation in mice and suggested an availability of erythritol in vivo.

In C57BL/6 mice that were infected intraperitoneally with either B. abortus 2308 WT or the erythritol catabolic mutants, the CFU numbers/spleen of both ΔeryA and ΔeryH were significantly lower than those of the WT 3 days after infection (Figure 4). After 9 days, only the erythritol-sensitive mutant showed significantly reduced CFU numbers, and consistent with this, splenomegaly was lower in the corresponding group of mice. Although splenomegaly was reduced in the ΔeryH-infected group, no differences in the CFU numbers of the mutants were observed 30 days after infection. These results are also in line with those of Burkhardt et al. (2005) who also found reduced CFU/spleen for the B. suis ΔeryC at early times (7 days or less) but not 28 days after infection. On the other hand, they are only in partial agreement with those of Sangari et al. (1998). These authors found that the CFU/spleen of a B. abortus 2308 erythritol sensitive Tn5 mutant and the parental strain were similar 7, 14, and 28 days after infection. The discrepancy is thus limited to the results obtained 7 days after infection but as earlier times were not studied and the CFU/spleen at longer times coincide with the our observations, it seems possible that the differences could relate to the protocols (breed of mice [BalB/c versus C57BL/6], Tn5 polarity effects on regulators downstream and possibly others).

Bearing in mind the properties of the mutants in our reporting system, these results suggest that the availability and importance of erythritol as a carbon source in the spleen of mice changes during the course of infection. During a splenic infection everything including the bacterial load, the type of infected cells, the type of recruited cells, the splenic microarchitecture and the immune environment is evolving in a dynamic way. Therefore, it may be not so surprising that 3 days after infection (that is to say before the peak of splenic infection and before any detectable splenomegaly) Brucella is in a compartment where erythritol is available (attenuation of ΔeryH) but also required (attenuation of ΔeryA); later on, changes in one or several parameters (i.e., infected cells or immune environment) might lead to a change in the nutrients that are available which makes erythritol, while available (attenuation of ΔeryH) less needed (no attenuation of ΔeryA). Afterward, erythritol availability would progressively dwindle because both mutants reached a WT-like bacterial load at 30 days p.i.

The well-known genital tropism of Brucella prompted us to investigate erythritol availability in the mouse conceptus. We infected C57BL/6 mice in either early (6 days) or late (14 days) pregnancy with B. abortus WT, ΔeryA and ΔeryH (Bosseray, 1980). Then, we determined the CFU in fetuses, placenta and fetal envelopes 1 and 9 days p.i., which in both groups corresponds to post-conception day 15. The results (Figure 5) showed a systemic distribution of the three strains in the conceptus with the placenta being be the most heavily infected organ, as expected from its barrier function (Bosseray, 1980). For the mice that were infected at day 6 post-conception, only the erythritol-sensitive strain ΔeryH was present in significantly lower numbers in the fetus and placenta. When mice were infected late in pregnancy, only ΔeryH was attenuated, and attenuation was observed in all tissues. These results strongly suggest that erythritol is available but not crucially required for B. abortus to multiply in the murine conceptus.

The polyol pathway has been suggested to be involved in the synthesis of erythritol in mammalian tissues (Pearce et al., 1962) and the key enzyme, AR, can catalyze the conversion of erythrose into erythritol (Hayman and Kinoshita, 1965; Kardon et al., 2008). Thus, we examined whether there was a connection between AR and the ability of Brucella to multiply in cells and tissues.

First, we studied whether AR is expressed in some of the cells in which our reporting system indicated the presence of erythritol. We found that the enzyme was detectable by immunofluorescence and that AR gene expression was observed in uninfected RAW 264.7 macrophages that were cultured in standard media; no noticeable changes were noted when these cells were infected by Brucella (data not shown). Since AR is induced by hyperglycemia (≥20 mM glucose; 5 mM being normoglycemic) (González et al., 1984; Tawata et al., 1992) and the DMEM-HG culture medium contains 25 mM glucose, we also measured the dependence of the expression of gene Akr1b3 (which codes for mouse AR) on glucose as an indirect and complementary test for AR activity. In RAW 264.7 macrophages that were grown in a range (2.8 [0.5 g/L], 5.6 [1.0 g/L] and 25 mM [4.5 g/L]) of glucose concentrations, we found that expression of Akr1b3 was modulated by glucose (Figure 6A), a result that strongly suggests that AR is active in these cells in culture. Although the glucose concentration did not affect the multiplication of the ΔeryH mutant during the first 2 h after infection, its inhibition was significant at 24 and 48 h (Figure 6B) suggesting that erythritol is available in cells that are grown in the range of physiological glucose concentrations that induce AR gene expression.

Second, we examined the AR and bacterial distributions in the conceptuses of mice that were infected intraperitoneally with B. abortus 2308-mCherry at 6 days post conception. Bacteria and AR-positive cells were located preferentially in the junctional zone of the placenta just underneath the decidua basalis and in a cellular sheet surrounding the fetus apposed on the internal face of the distended decidua parietalis (Figure 7A). Infected cells were positive for cytokeratin 7 (Figures 7B,C) and thus trophoblastic in nature (Croy et al., 2013), probably being trophoblastic giant cells (Kim et al., 2005). As expected, some infected trophoblasts were also scattered in the decidua after mid-gestation (Hu and Cross, 2010). Bacteria were found almost exclusively in AR-positive cells (Figure 8).

Finally, we investigated mouse spleens 9 days after infection, a time at which the erythritol-sensitive strain ΔeryH showed attenuation. We found intense AR-staining in the red pulp (Supplementary Figure S3), with small clusters of CD11b-positive cells [often iNOS positive and corresponding to granulomas (Copin et al., 2012)], which are frequently associated with AR. In contrast, we hardly detected AR in spleens of non-infected mice; when we did, AR was mostly restricted to scattered CD11b-negative cells in the red pulp.

Escherichia coli DH10B were grown in LB medium. B. abortus 2308 Nal^R and derived strains were grown at 37°C in rich medium 2YT (16 g/L bacto tryptone 10 g/L yeast extract and 5 g/L NaCl; BD Difco) or in a chemically defined medium (Barbier et al., 2014) composed of 2.3 g/L K₂HPO₄; 3 g/L KH₂PO₄; 0.1 g/L Na₂S₂O₃; 5 g/L NaCl; 0.2 g/L nicotinic acid; 0.2 g/L thiamine; 0.07 g/L pantothenic acid; 0.5 g/L (NH₄)₂SO₄; 0.01 g/L MgSO₄; 0.1 mg/L MnSO₄; 0.1 mg/L FeSO₄; 0.1 mg/L biotin and 2 g/L of erythritol. Growth was monitored using an automated plate reader (Bioscreen C, Lab Systems) following the OD (600 nm) with continuous shaking at 37°C. The growth rate was calculated as follows: (In (ODt2) − In (ODt1)) / (t2 − t1). The Δt was set for 7 h, i.e., approximately two division times, and incremented over the log phase every 0.5 h (0–7 h; 0.5–7.5 h,…) resulting in a set of values whose mean is the average growth rate, μ. When required, the medium was supplemented with chloramphenicol (20 mg/ml), nalidixic acid (25 mg/ml), sucrose (5%), agar (15 g/L, BD Difco) and polyols (concentrations annotated in the manuscript). Unless otherwise stated, reagents were purchased from Sigma–Aldrich.

Construction of an mCherry-producing B. abortus 2308 strain was performed following the same procedure that was validated for B. melitensis (Copin et al., 2012). Construction of the in-frame deletion in eryA was done following a previously described strategy (Barbier et al., 2014). Briefly, approximately 750 bp upstream and downstream of BAB2_0372 were amplified by PCR from genomic DNA of B. abortus 2308. The obtained PCR products were, respectively, flanked by SpeI/BamHI (SpeI_F: 5′-ACTAGTCTTGGCGGAAACTTGACTGG-3′; BamHI_R: 5′-ATACGCGGATCCGCGATAACGCATGGCTGACACAGG-3′) and BamHI/SphI restriction sites (BamHI_F: 5′-TATCGCGGATCCGCGTATGGCAAATAAGGAAACATTGAATG-3′; SphI_R: 5′-GCATGCGCGCTTGTCGTGGTTCTG-3′). A third PCR joined the two fragments together using primers SpeI_F and SphI_R, which was followed by ligating this product into an EcoRV-digested pGEM plasmid (Promega). After sequence verification (Beckman Coulter Genomics), the ±1500 bp insert was excised as a SpeI – SphI fragment and cloned into a pNPTS138 suicide vector (Kan^R, Suc^S). The acquisition of this vector by Brucella after mating with conjugative S17 E. coli was selected by kanamycin and nalidixic acid resistance. The loss of the plasmid concomitant with either a deletion or a return to WT phenotype was then selected on sucrose. Mutants were identified using PCR with primers that were located external to the deletion. The ΔeryH and ΔeryI strains were previously characterized (Barbier et al., 2014). For complementation, eryH was amplified by PCR from genomic DNA of B. abortus 2308 as a BamHI/XhoI fragment (BamHI_F: 5′-gcgggatccatgaccaaattctggattgg-3′; XhoI_R: 5′-ttaattcgcttgaaccttggctcgagccg-3′). Fragments were cloned into an EcoRV-digested pGEM, sequenced and then transferred into a pBBR1MCS1 (Cm^R). The ΔeryH strain with the construct was then transformed by conjugation with the construction and selected for with the newly acquired resistance to chloramphenicol.

All Brucella were handled under BSL-3 containment according to the Directive 98/81/CE du Conseil du 26 octobre 1998 and to a law of the Gouvernement wallon du 4 juillet 2002.

RAW 264.7 murine macrophages were routinely cultured in Dulbecco’s modified Eagle’s medium with high glucose (DMEM, Gibco) supplemented with 10% heat-inactivated fetal calf serum (FCS, Gibco). THP-1 human macrophage-like cells were cultured in RPMI 1640 medium (Gibco) that was supplemented with 10% FCS and 2 mM L-glutamine. Cells were differentiated into adherent monocytes by overnight treatment with 5 nM phorbol myristate (PMA). Bovine trophoblastic CL2 cells were kindly provided by Pr. Cynthia Baldwin (University of Massachusetts, Amherst, MA, United States) and cultured in RPMI 1640 supplemented with 10% FCS and 0.05 mM 2-mercaptoethanol (Gibco) as previously described (Parent et al., 2012). BeWo human trophoblastic cells (ATCC clone CCL-98) were cultured in DMEM-F12 Ham medium (Gibco) that was supplemented with 10% FCS and 2 mM L-glutamine as already described (Salcedo et al., 2013). All cells were maintained at 37°C with a 5% CO₂ atmosphere.

For experiments in which RAW 264.7 were cultured with various glucose concentrations, DMEM with no glucose (Gibco) was supplemented with the appropriate amount of filter-sterilized glucose. Cells were cultured under these conditions for at least 2 passages prior to infection. Infections were performed as described elsewhere (Salcedo et al., 2013; Zúñiga-Ripa et al., 2014). Briefly, cells were seeded in 24-well plates at an appropriate density (2.105 cells/ml THP-1 cells and 4.104 cells/ml for RAW 264.7 and trophoblastic cells) and infected 24 h later with a multiplicity of infection (MOI) of 100. The cells were centrifuged at 1000 RPM for 10 min at 4°C before being incubated for 1 h at 37°C with 5% CO₂; they were then washed with fresh medium and incubated for 1 h with medium containing 50 μg/ml of gentamicin. The medium was then replaced with a fresh medium that contains 10 μg/ml of this antibiotic. At 2, 24, and 48 h after infection, cells were washed with PBS and treated for 10 min at room temperature with PBS Triton X100 0.1%; after this treatment, the lysates were collected, diluted, plated on TSB and incubated at 37°C for approximately 3 days to enumerate the CFUs.

For immunofluorescence staining, cells were seeded on coverslip and treated as described previously (Francis et al., 2017). At the end of the process, the cells were fixed in 2% paraformaldehyde, pH 7.4, at 37°C for 15 min.

The procedures used and the handling of mice complied with current European legislation (directive 86/609/EEC) and the corresponding Belgian law “Arrêté royal relatif à la protection des animaux d’expérience du 6 avril 2010 publié le 14 mai 2010.” The Animal Welfare Committee of the Université de Namur (Belgium) reviewed and approved the complete protocols (Permit Number 16/277). All infections were performed at an Animal Biosafety Level 3 facility.

To obtain the inoculum, bacteria from an overnight culture of Brucella in rich medium were pelleted, washed with RPMI 1640 and diluted in this medium. Intraperitoneal infection was carried out as previously described (Copin et al., 2012). Briefly, 500 μl of suspension (10⁵ CFU) was injected into groups of 8 to 12 C57BL/6 mice for each tested strain. Mice were euthanized 3, 9, and 30 days post-infection by cervical dislocation, the spleens were isolated, weighted and homogenized in 1 ml of PBS Triton X100 0.1%, and the CFU were counted on tryptic soy agar plates.

The procedure that was used to infect pregnant mice was adapted from previous reports (Bosseray, 1982; Kim et al., 2005; Pennington et al., 2012). Estruses of 6–14 weeks old C57BL/6 females were synchronized 3 days before mating pairs were set up with males that were 3–4 months old. Then, the presence of a vaginal plug was checked daily, and potentially fertilized females were isolated. That day corresponds to day 0 post-fecundation (PF). Four to five pregnant females were infected intraperitoneally with 500 μl of bacterial suspension that was prepared as previously described (10⁵ bacteria) at day 6 or 14 PF. At day 15 PF, mice were anesthetized with isoflurane (Zoetis) and euthanized by cervical dislocation. Conceptuses were removed from maternal uterine horns and transferred to sterile Petri dishes on ice, where they stayed for 15 min. Placenta, fetuses and surrounding fetal membranes were then further isolated and weighed. Tissues were homogenized in 1 ml PBS Triton X100 0.1% with an Ultra-Turrax homogenizer, the homogenates were serially diluted in PBS, and their CFU were counted.

Fetuses and spleens were fixed for 2 h at room temperature in 2% paraformaldehyde (pH 7.4), washed in PBS, and incubated overnight at 4°C in a 20% PBS-sucrose solution. The tissues were then embedded in Tissue-Tek OCT compound (Sakura) and frozen in liquid nitrogen, and cryostat sections (thickness, 5 μm for spleens and 10 μm for fetus) were prepared. For the staining, tissue sections were rehydrated in PBS and incubated in a PBS solution that contained 1% blocking reagent (PBS-BR 1%, Boehringer) for 20 min before they were incubated overnight in PBS-BR 1% containing mAbs or the following reagents: DAPI nucleic acid stain Alexa Fluor 350 or 488 phalloidin (Molecular Probes) to visualize the structure of the organ, and rat biotin-coupled anti-mouse CD11b (BD Pharmingen), rabbit anti-mouse iNOS (Calbiochem), rabbit anti-mouse Cytokeratin 7 (ab 181598, Abcam), and rabbit anti-mouse AR (CPA3124, Cohesion Biosciences) to stain the cells of interest. The samples were incubated with the appropriate secondary reagents [Alexa Fluor 568 streptavidin (Molecular Probes) or Alexa Fluor 647-coupled donkey anti-rabbit IgG (Molecular Probes)] for 2 h. Slides were mounted in Fluoro-Gel medium (Electron Microscopy Sciences, Hatfield, PA, United States). Labeled tissue sections were visualized with an Axiovert M200 inverted microscope (Zeiss, Iena, Germany) that was equipped with a high-resolution monochrome camera (AxioCam HR, Zeiss).

Images (1384 pixels × 1036 pixels, 0.16 μm/pixel) were acquired sequentially for each fluorochrome with A-Plan 10×/0.25 N.A. and LD-Plan-NeoFluar 63×/0.75 N.A. dry objectives and recorded as eight-bit gray-level ^∗.zvi files. At least three slides per organ were analyzed from three different animals, and the results are representative of two independent experiments.

For immunostaining of AR in RAW 264.7 murine macrophages, the primary antibody that was used was the same that was used for staining in mice with a goat anti-rabbit IgG Alexa 488 (Life Technologies) as the secondary antibody.

RNA from cells cultivated in a T75 flask was extracted with TriPure isolation reagent (Roche) according to the instructions of the manufacturer and DNA contamination was eliminated by incubation with DNase I (Fermentas). Then, RNA was first reverse transcribed (two steps) by SuperScript II (Invitrogen) into cDNA, which was then amplified in a LightCycler 96 Instrument (Roche) with FastStart Universal SYBR Green Master (Roche) as the fluorescent dye. The specificity of the SYBR Green assays was assessed by melting-point analysis and gel electrophoresis. The results were normalized using the housekeeping b-actin gene. Primer sequences are described in Supplementary Table S1.