|Home | About | Journals | Submit | Contact Us | Français|
BM, NPW and JM performed experiments with Shigella. DFB and JAC provided technical support and advice for EMSAs and analysis of the fusions. JAS analysed the LPS, M-CP carried out EM, and FP performed the oxygen measurements. CMT and PS provided advice and overall direction, and wrote the paper.
Bacteria co-ordinate expression of virulence determinants in response to localised microenvironments in their hosts. Here we show that Shigella flexneri, which causes dysentery, encounters varying oxygen concentrations in the gastrointestinal (GI) tract, which govern activity of its type three secretion system (T3SS); the T3SS is essential for cell invasion and virulence1. In anaerobic environments (e.g. the GI tract lumen), Shigella expresses extended T3SS needles while reducing Ipa (Invasion plasmid antigen) effector secretion. This is mediated by FNR, a regulator of anaerobic metabolism that represses transcription of spa32 and spa33, virulence genes that the switch in secretion through the T3SS. We demonstrate there is a zone of relative oxygenation adjacent to the GI tract mucosa, caused by diffusing from the capillary network at the tips of villi. This would reverse the anaerobic block of Ipa secretion, allowing T3SS activation at its precise site of action, enhancing invasion and virulence.
Shigella virulence depends on its ability to enter epithelial cells by delivering Ipa effectors via its T3SS into the host cell cytoplasm1. Secretion through T3SSs is highly regulated. Initially, T3SS needle components are secreted until it reaches a pre-defined length 2,3. In inducing conditions, a switch then occurs allowing Ipa secretion through needles, mediating bacterial entry 4.
Using the rabbit ligated gastrointestinal (GI) loop model 5,6, we identified a colonisation-defective S. flexneri M90T mutant with a transposon in fnr, which encodes a regulator of anaerobic metabolism (supplementary Fig. 1). FNR is only active as a dimer containing an [4Fe-4S] cluster. The cluster dissociates in the presence of oxygen (O2), destabilising the dimer, with loss of FNR activity 7. M90TΔfnr was substantially attenuated for colonisation compared with the wild-type strain, M90T (competitive index [C.I.], 0.05), and this was restored by complementation (C.I. of M90TΔfnr pBM2, 0.6). Furthermore, we challenged intestinal loops with strains individually. Infection with M90T led to abscesses, and shortening and destruction of villi (Fig. 1a). These alterations were significantly reduced following infection with M90TΔfnr (p < 0.01, Fig. 1b and supplementary Table 1) and absent after challenge with a T3SS− mutant, M90TΔmxiD (Fig. 1c). Restoration of the invasive phenotype following complementation (M90TΔfnr pBM2) confirmed the requirement of FNR in vivo (Fig. 1d).
To define the contribution of O2 to Shigella virulence, epithelial cells were propagated in an aerobic environment, and either left there or transferred to an anaerobic cabinet for 30 min. and then challenged with bacteria grown in aerobic or anaerobic conditions, respectively. Bacterial entry into cells in an anaerobic cabinet was significantly higher than in an aerobic cabinet (p < 0.05, Fig. 1e). This was not caused by cell death or mediated by the host transcription factor HIF-1α8,9 (supplementary Fig. 2). Instead, the enhanced cell entry of Shigella in the absence of O2 was dependent on FNR (Fig. 1e).
To understand the basis of the increased cell entry, we investigated T3SS structure and function of Shigella grown with or without O2. With O2, M90T and M90TΔfnr secreted the effectors IpaB, IpaC and IpaD following exposure to the inducer of secretion Congo red 4 (CR). CR-induced secretion of Ipas by M90T was reduced in anaerobic conditions and was accompanied by increased intra-bacterial Ipa levels (p < 0.01, Fig. 2a and supplementary Fig. 3). The reduced effector secretion in the absence of O2 was not detected with M90TΔfnr (Fig. 2a), while complementation of the fnr mutant restored low level Ipa secretion in anaerobiosis as observed with M90T. Additionally, SEM demonstrated that the number of visible needle tips was significantly higher on bacteria after growth in anaerobic compared with aerobic conditions (supplementary Fig. 4). This was not associated with any detectable change in the lipopolysaccharide profile, which affects the presentation of T3SS needles 5 (supplementary Fig. 5). Instead, TEM revealed a 25% increase in the average needle length during growth in anaerobic compared with aerobic conditions (62 vs. 48 nm respectively, p <0.001, Fig. 2b), with less rigorous control of needle length in the absence of O2 (S.D. of needle length 28 and 11 nm in anaerobic and aerobic environments, respectively). In contrast, needles produced by M90TΔfnr were similar regardless of the presence of O2 (average length 53 and 58 nm with and without O2, respectively, Fig. 2c). Therefore in the absence of O2, FNR mediates reduced Ipa secretion and elongation of T3SS needles.
As FNR is a transcription regulator, we next examined mRNA levels of mxi/spa pathogenicity island genes and their regulators by qrt RT-PCR 10 (Fig. 3a). There was no significant alteration in mRNA levels of genes encoding effectors, T3SS components, or regulators in the presence or absence of O2. However there was a reduction in spa32 and spa33 mRNA levels (4.7- and 5.5-fold reduction respectively, p < 0.01) during anaerobic growth that was FNR dependent (Fig. 3a). Spa32 is required for control of needle length and the selection of substrates for secretion 2,3, while dysregulation of Spa33 expression blocks Ipa secretion 11. The activity of lacZ reporter fusions confirmed that the reduction of spa32 and spa33 mRNA levels is associated with decreased transcription in an FNR-dependent fashion (supplementary Fig. 6). The O2 regulation of spa32 and spa33 fusions was abolished by modifying the two predicted FNR binding sites 12 in the promoters of spa32 and spa33 (−156 and −67, and −205 and −116 upstream of the initiation codons, respectively, supplementary Fig. 6), while FNR binds sequences upstream of spa32 and spa33 (Fig. 3b). Binding upstream of spa33 was abolished by modification of the FNR binding sites (supplementary Fig. 7). These data demonstrate that FNR represses the Shigella virulence genes spa32 and spa33 in anaerobic conditions.
To establish whether FNR-mediated repression of spa32 and spa33 contributes to increased Shigella entry into cells in an anaerobic cabinet, we constructed M90T containing spa32 and spa33 under the control of either their native promoters (M90Tp32/33) or promoters with disrupted FNR boxes (M90Tp#32/#33). The presence in M90T of spa32 and spa33 alleles that are not repressed by FNR led to IpaB secretion during anaerobic growth (Fig. 3c), and loss of the increased entry of Shigella into cells in an anaerobic cabinet (invasion of strain with p#32/#33 vs. p32/33, p < 0.01, Fig. 3d); similar findings were obtained by inappropriate expression of spa32 and spa33 in anaerobic conditions with IPTG (supplementary Fig. 7).
To explain the enhanced entry of Shigella into epithelial cells following transfer into anaerobic cabinets, we hypothesized that bacteria encounter O2 in the vicinity of host cells that promotes T3SS activation. We used microelectrodes to directly measure concentrations of O2 above cells after they had been moved into an anaerobic cabinet 13. There was sufficient O2 (approximately 0.4%) 10 μm above cells to inactivate FNR 14 and allow IpaB secretion by S. flexneri (Fig. 4a and b). While no O2 was detected above cells following addition of the O2 scavenger deoxy-haemoglobin, O2 levels were unaffected by the addition of cytochrome c, which contains haem but does not bind O2 15. These data provide an explanation for the hyper-invasion of Shigella following transfer of cells to an anaerobic chamber, and indicate that remote from the cells, bacteria would assume a ‘primed state’ (with extended T3SS needles and impaired Ipa secretion), while residual O2 in proximity to cells would allow T3SS activation and enhanced entry.
To establish the relevance of these findings to Shigella invasion in vivo, we constructed two bacterial reporters for the presence of oxygen during host:pathogen interactions as introduction of a microelectrode into the GI lumen might disrupt O2 homeostasis. Formation of the GFP fluorophore requires covalent bonding with O2 16,17, and luciferase activity is dependent on O2 and ATP 18. M90TΔmxiD pFPV25.1 expresses GFP with O2-dependent fluorescence 16,17 while Lux-based luminescence of E. coli pXen5 provides a distinct reporter for the presence of O2 (supplementary Fig. 8 and 9); E. coli was used as the host because this construct is not functional in S. flexneri. The luminescence of anaerobically grown E. coli pXen5 was measured in intestinal loops with an intact vascular supply or in loops in which the vessels had been occluded by clamping prior to challenge. Bacteria emitted significant luminescence in intestinal loops with an intact vascular supply within 15 min, while barely any signal was detected from loops that had had their afferent vessels occluded (p < 0.001, Fig. 4c and supplementary Fig. 10). To determine the location of intra-luminal O2 within the GI tract, intestinal loops were infected with M90TΔmxiD pFVP25.1, and tissue recovered 18 h later. Even though bacteria were present throughout the gut lumen (detected with an anti-LPS pAb), only Shigella located within approximately 70 μm of the epithelial surface (particularly at the villous tips) emitted significant fluorescence (Fig. 4d). This zone of relative oxygenation is provided by the vascular supply to the GI tract as the GFP signal is abolished on clamping the afferent vasculature (supplementary Fig. 11).
The status of the Shigella T3SS is precisely modulated by ambient O2 tensions which vary at specific sites in the GI tract. In the anaerobic lumen of the intestine, T3SS needles will be extended but not competent for secretion. In contrast, the relatively aerobic zone adjacent to the mucosal surface would allow activation of the T3SS on bacteria at this site (Fig. 4e). This zone of oxygenation depends on the vascular supply probably by diffusion of O2 from the capillary network at villous tips 19. Expression of elongated needles would optimise the presentation of the T3SS translocon for engaging host cells 20. Furthermore precise regulation of secretion of effectors would allow bacteria to deliver maximal amounts of Ipas at the exact time and place in vivo needed to trigger efficient manipulation of the cytoskeleton and cell entry 21.
The presence of available O2 in localised regions in the intestine may be under-appreciated and could have important biological consequences. This ‘aerobic zone’ might dictate the outcome of interactions with the extensive intestinal microbiota, acting as an innate immune barrier to protect the mucosal surface from anaerobic bacteria, while being recognised as a signal to promote invasion by pathogens such as Shigella, Salmonella and Listeria monocytogenes 22. FNR and possibly other regulators (such as the ArcBA system) may mediate other effects aside from bacterial invasion, especially in the context of the inflammatory response to infection. Similar mechanisms of modulating virulence in response to O2 may also have evolved in other intestinal pathogens, as FNR boxes are present upstream of genes required for T3SS function in Yersinia spp., and serovars of Salmonella enteritidis (supplementary Table 2).
All strains and plasmids in this study are described in Supplementary Table 2. E. coli was grown in Luria Bertani broth while Shigella was propagated in Trypticase soy media (TCS, Oxoid) or on TCS plates with Congo Red (0.01%, Sigma). Experiments under different oxygen (O2) concentrations were performed in an anaerobic cabinet (Whitley DG250), or in a cabinet in which O2 tensions can be controlled (Plaslabs). Antibiotics were added at the following concentrations: chloramphenicol, 50 μg/ml; ampicillin, 100 μg/ml; kanamycin 50 μg/ml. For LPS analysis, whole cell lysates of bacteria grown in liquid culture to an O.D. A600 of <0.2 were prepared in LPS loading buffer, treated with proteinase K, then subjected to Tricine SDS-PAGE analysis 23, and visualised by silver staining (Pierce).
The regulatory regions of spa32 and spa33 (from 300 bp upstream of the start codon to 15 bp downstream) were amplified by PCR with NG900/901 and NG902/903, respectively, and ligated into the SmaI and BamHI sites of pRS415 24 (supplementary Fig. 6). All oligonucleotides are shown in supplementary Table 4). For transcriptional fusions, the regulatory regions of spa32, spa33 and P6 (located upstream spa47 25 and supplementary Fig. 6) were amplified with primers SG87/88, SG81/89 and SG94/26, and ligated in pQF50.1 which contains a promoterless lacZ as a reporter (kind gift from C. Parsot 26). Putative FNR binding sites upstream of spa32 and spa33 were modified by a two step-PCR procedure 27. spa32 FNRbox1 (TTGAAGAAATTCAA, inverted repeats underlined) and spa32 FNRbox2 (TTGATTAAAAGAAA) were altered to TCGAGGAAATACAG and CTCATTAAACGGAA using primers SG25/26 and SG27/28, respectively (for details see supplementary Fig. 6). spa33 FNRbox1 (TTGATAGCAGTCAA) and spa33 FNRbox2 (TTGACGCTAACGAA) were altered to TCGATAGCACAGCA and TCCACGCTAAGGAA with primers SG29/30 and SG31/32, respectively. To generate p32/33, the open reading frames and promoters of spa32 and spa33 were amplified with SG13 and SG14, and ligated into pBBR1MCS-4. The FNR boxes in this plasmid were modified as above to generate p#32/#33. For IPTG inducible over-expression, spa32 and spa33 were amplified with NG1390/1391 and NG1369/1370, and ligated into the NdeI/BamHI sites of pET21b and the NdeI/XhoI sites of pET28b (Invitrogen), respectively. Constructs were verified by DNA sequencing.
Murine α-Shigella LPS pAbs 28 were diluted 1:500 for immunofluorescence microscopy. FITC-conjugated goat anti-rabbit and Phalloidin-Rhodamine red X (RRX)-conjugated donkey anti-mouse antibodies (Jackson Immunoresearch Antibodies) were used at a final dilution of 1:500. Dapi (Invitrogen) was used at 1μg/mL. Rabbit α-FNR pAbs (kind gift from Professor Jeff Green, University of Sheffield, UK), anti-RecA (MBL), α-IpaB, α-IpaC and α-IpaD 29 were used at a 1:10,000 final dilution. HRP-conjugated α-His mAb (Qiagen) was used at a final dilution of 1:2,000.
HeLa epithelial cells were grown to semi-confluency in aerobic conditions in 24-well tissue culture plates. Before challenge, cells were either maintained under aerobic conditions or transferred to an anaerobic cabinet (Whitley DG250) for 30 min. Cells were challenged at an MOI of 100 with bacteria grown in liquid culture under aerobic or anaerobic conditions. For aerobic growth, bacteria were grown in air overnight in 5 ml LB with shaking in a 37 °C incubator. Cultures were then diluted 1 in 100 into 5 ml fresh media in a 20 ml universal and then grown with vigorous shaking until the O.D. A600 reached 0.2. For anaerobic growth, bacteria were propagated in exactly the same way except cultures were performed in an anaerobic cabinet in media which had been pre-incubated in the cabinet for at least 16 hrs. Bacteria were spun onto cells by centrifugation at 2,000 x g for 10 min. For adhesion assays, cells were immediately washed three times in PBS, then fixed in 3% paraformaldehyde (PFA) for 15 min.
To measure bacterial entry, cells were challenged as above, incubated for 30 min at 37°C, after which gentamicin (50 μg/ml) was added for 1 hr to kill extracellular bacteria. Cells were washed with PBS, lysed with 500 μL 1% saponin in PBS, and bacteria recovered by plating. Results are expressed as the number of CFU/ml of cell lysate, and are the average of at least three independent experiments performed in triplicate. Statistical significance was calculated using the Student’s T-test. When required, IPTG (1 mM final concentration) was added to the cells in tissue culture wells at the time of bacterial challenge. Cells were exposed to mimosine (Sigma, final concentration 800 μM), which inhibits degradation of HIFα for 1 h before challenge with bacteria 30. The constitutively HIFα –expressing epithelial cell lines Rcc10 VHL−/− and corresponding control VHL+/+ (kindly provided by P. Maxwell 8,31) were cultured and infected as above.
For immunofluorescence microscopy, cells were grown on 12 mm glass coverslips, challenged as above, washed, then fixed in 3% paraformaldehyde (PFA) for 15 min, and washed again. Coverslips were incubated for 30 min with primary antibodies diluted in 10% horse serum/0.1% saponin in PBS, washed with PBS containing 0.1% saponin, then incubated with the secondary conjugated antibodies for 30 min. Samples were washed three times in 0.1% saponin in PBS, then in PBS, and finally in water, mounted in Prolong (Molecular Probes), and examined by laser-scanning confocal microscopy (Zeiss Axiovert 100M or LSM510).
Cell viability was determined by staining with 0.01% propidium iodide (PI, Sigma) in PBS for 10 min 32. PI only penetrates into and stains the DNA of non-viable cells. As a positive control, cells were killed by incubation in 3% PFA for 15 min. Fluorescence was measured using a FACS flow cytometer (BD systems) recording at least 10,000 events. Data were analysed with CellQuest Pro software (BD Biosciences), and expressed as percentage survival.
To evaluate the response of bacterial O2 reporters, M90TΔmxiD pFPV25.1 or E. coli pXen were grown anaerobically in a controlled atmosphere chamber (Plaslabs) in TCS media at 37 °C until the O.D. A600 reached 0.2. The cabinet was then adjusted to 5%, 10%, 15% or 20% O2, with the ambient oxygen tension monitored with an oxygen meter. At each step, bacteria and PFA were incubated for 15 min. To measure fluorescence an aliquot of the culture was mixed with an equal volume of equilibrated PFA. For luminescence, bacteria were transferred into an air tight tube within the chamber. Fluorescence of bacteria was assessed by confocal microscopy, and the results are the average of at least 150 bacteria from three different fields. Luminescence was measured with an IVIS camera (Xenogen). Experiments were performed on three different occasions.
The luxABCDE operon present on pXen5 (Xenogen) was introduced into E. coli by transformation. Luminescence was measured using an IVIS camera (Xenogen), and images quantified using the Living Image 3.1 software (Xenogen) and expressed as a Total flux (p/s), as recommended by the manufacturer.
For TEM, bacteria were grown on solid medium overnight and a single colony was re-suspended in distilled water. Bacteria were negatively stained with 2% uranyl acetate on glow discharged copper grids. The samples were observed in a Jeol 1200EXII or a JEM 1010 microscope (Jeol Company, Tokyo Japan) at 80-kV with an Eloise Keenview camera. Images were recorded with Analysis Pro Software version 3.1 (Eloise SARL, Roissy, France). The length of at least 200 needles was measured on three separate occasions, and the results presented using a non-linear regression model (Loess, locally weighted scatterplot smoothing, GraphPad, Prism software).
For SEM, bacteria were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.2) for 1 h on coverslips coated with poly L-lysine, washed three times in 0.2 M cacodylate buffer (pH 7.2), then post-fixed for 1 h in 1% (wt/vol) osmium tetroxide in 0.2 M cacodylate buffer (pH 7.2). Samples were rinsed with distilled water, dehydrated through a graded series of ethanol then critical point dried with CO2. Dried specimens were sputtered with 10 nm gold palladium with a GATAN Ion Beam Coater and examined with a JEOL JSM 6700F field emission scanning electron microscope operating at 5 kV. Images were acquired with the upper SE detector and the lower secondary detector.
Bacteria were grown in 500 ml TCS liquid media in the presence or absence of O2 until the O.D. A600 reached 0.2; IPTG was added to a final concentration of 1 mM as required. Bacteria were harvested by centrifugation for 15 min at 3,000 x g, washed, then re-suspended in 10 ml PBS. Cells were spun again for 5 min at 3000 x g, and re-suspended in 200 μl of PBS. Secretion was induced by adding Congo Red (CR, final concentration, 0.05%) under aerobic or anaerobic conditions, and incubated for 20 min at 37 °C. Bacteria (representing the cellular fraction) were collected by centrifugation at 3000 x g, and the supernatants (representing the secreted fraction) retained. Total protein concentrations were measured by the method of Bradford (Biorad). Proteins were separated by 10% SDS-PAGE and either visualised by Coomassie blue staining or transferred to nitrocellulose membranes using semi-dry transfer, and incubated with the primary antibodies diluted in PBS/5% milk/0.01% Tween20 (Sigma) for 1 h. Membranes were washed in PBS three times, then incubated with secondary antibodies for 1 hr before washing. Antibody binding was detected with chemiluminescence (ECL kit, GE Healthcare). Quantification of expression levels was performed by densitometry of scanned blots using the ImageJ software and performed at least on two independent occasions. Statistical significance was calculated by the Student’s T-test.
Gel retardation assays were carried out as described previously using purified FNRD154A, an allele that forms dimers under aerobic conditions 33. In brief, purified 300 bp spa32 and spa33 promoter fragments (supplementary Fig. 6), and the modified alleles were end-labelled with [γ-32P]-ATP with 10 U/μl T4 polynucleotide kinase (New England BioLabs). Approximately 0.5 ng of each labelled fragment was incubated with varying amounts of FNRD154A in 10 mM potassium phosphate (pH 7.5), 100 mM potassium glutamate, 1 mM EDTA, 50 μM DTT, 5% glycerol and 25 μg ml−1 herring sperm DNA. After incubation at 37°C for 20 min, samples were separated on 6% polyacrylamide gels containing 2% glycerol. Gels were analysed using a Bio-Rad Molecular Imager FX and Quantity One software (Bio-Rad).
Bacteria were grown in liquid LB media until an OD600 0.2. The cultures were immediately transferred to a new tube containing a 1 in 5 volume of 5% phenol pH 4.3 (Sigma) in ethanol. Samples were kept on ice for 30 min, then bacteria harvested by centrifugation at 10,000 x g for 10 min. The pellets were re-suspended in TE buffer containing 50 mg/ml lysozyme. RNA was isolated with the RNeasy method (Qiagen), and quantified by measuring ratio of the O.D. A260/A280 of samples. DNA was digested by treatment with DNase I (Sigma), then cDNA was synthesized with reverse transcriptase (Quantitect, Qiagen). polA, ipaB, ipaC, ipaD, mxiH, spa47, spa13, spa32, spa33, spa24, virF, and virB were amplified using the SensiMix DNA kit (Quantance) with primers described in supplementary Table 4 with a Rotor-gene RT-PCR machine (Corbett Research). Results are the average of duplicate experiments each performed on three independent occasions. Data were analysed with Rotor Gene 3000 (Corbett Research) software.
Bacteria were grown to an OD600 0.2 in 10 ml of TCS media and harvested by centrifugation at 5,000 x g for 10 min. Pellets were re-suspended in 500 μl of PBS, sonicated and further centrifuged at 13,000 x g for 20 min and the supernatants collected. β-galactosidase (βgal) activity was measured at 37°C by assaying degradation of ortho-nitrophenyl-β-galactoside (ONPG) 34. The protein concentration of lysates was determined by the method of Bradford, and the results are the mean values of three measurements performed on two independent samples, and are expressed in Miller Units/mg of total cellular protein.
For evaluation of the virulence and measurement of competitive indices (C.I.), experiments were performed on four or more independent occasions in at least two loops in four animals. Bacteria were grown in liquid media for three hours at 37°C, and a total 107 CFU of bacteria consisting of equal numbers of two strains in 500 μl of physiological water were injected into loops. Bacteria were recovered from the loops by plating homogenates to solid media with (to recover mutant strains) or without antibiotics (to recover the mutant and wild-type strains). The competitive index was calculated as the ratio of mutant to wild-type bacteria recovered from animals divided by the ratio in the inoculum; the results are the average of at least four samples originating from four different rabbits.
For histopathological analysis, ileal loops were fixed in 4 % buffered formalin, embedded in paraffin, 5 μm sections obtained, and stained with haematoxylin-eosin-safranin or Giemsa. The extent of damage to villi was quantified by measuring their volume-to-length ratio and the presence of indentations, ruptures of the mucosal surface, and abscesses. Results are the average of measurements from at least 25 vili from two independent samples for each strain 35. For immunohistochemical staining, samples were washed in PBS, incubated at 4°C PBS containing 12% sucrose for 90 min, then in PBS with 18% sucrose overnight, and frozen in OCT (Sakura) on dry ice. 7 μm sections were obtained using a cryostat CM-3050 (Leica). The immunostaining was performed as for immunofluorescence of infected cells, and the antibodies were used at the same concentrations. Stained samples were examined by laser-scanning confocal microscopy (Zeiss Axiovert 100M or LSM510), as described above.
The vascular supply to intestinal loops was interrupted by exposing the afferent mesenteric vessels which were occluded with a clamp. To detect luciferase and GFP activity in the loops, bacteria were grown under anaerobic conditions for 4 hrs in LB media then 107 CFU injected into the lumen of individual ligated loops. The total luminescence of each injected loop was measured with an IVIS camera system (Xenogen) and expressed as a Total flux (p/s), as recommended by the manufacturer. To detect GFP fluorescence, the corresponding loops were sectioned and fixed in 3 % PFA and processed as above. Stained samples were observed by laser-scanning confocal microscopy (Zeiss Axiovert 100M). The results are the average of duplicate experiments each performed on three independent occasions.
HeLa cells were grown at 37 °C in room air supplemented with 5% CO2. Plates with the cells were transferred into an anaerobic cabinet, and media discarded and replaced with media that had been equilibriated in the cabinet for over 18 hrs. An oxygen microelectrode sensor (Oxy-4, Unisense, Aarhus, Denmark) connected to a PA-2000 (Unisense) was advanced into the media to 10 μm from the top of the cells. The oxygen sensor was calibrated at 37 °C as described previously 36. Measurements were performed in triplicate on three occasions at 30 and 60 mins after replacement of the media. For experiments in the presence of deoxy-haemoglobin (Hb, Sigma) and cytochrome c (cyt, Sigma), saturated solutions were prepared in deionised water, and added to cells at a final concentration of 6 μM to media kept under aerobic or anaerobic conditions.
This work was supported by the Fondation pour la Recherche Médicale, the Royal Society, an ERC Advanced Grant (HOMEOEPITH), and the European Union (QLRT-1999-00938). Work in CMT’s laboratory is supported by the Wellcome Trust and The Medical Research Council, and PJS is a Howard Hughes Medical Institute scholar. Imaging was performed at the PFID station (Institut Pasteur). We are grateful to Professor Jeff Green (University of Sheffield) for the α-FNR pAbs and advice. Rachel Exley, David Holden and Katrina Ray provided helpful suggestions and reviewed the manuscript; we thank Marie-Anne Nicola for technical help.