Recordings of cellular autofluorescence were performed upon small clusters of type-1 cells (approx. 2–10 cells) or on single superior cervical ganglion (SCG) neurons. Signal levels were typically low when working with small clusters (2–3) of type-1 cells but could be easily resolved and recordings lasting ≥ 20 min with less than 25% signal loss were readily achievable. A large part of the autofluorescence at these wavelengths is due to mitochondrial NADH; NAD is non-fluorescent and cytosolic NADH levels are normally low. Mitochondrial NADH can be oxidised to NAD by the uncoupling of mitochondrial respiration. Application of 1 μm FCCP caused a rapid reduction in autofluorescence in all type-1 cells/clusters tested to 74% of control (±2.5%, n= 5, P < 0.005) indicating that approximately 25% of autofluorescence under control conditions was due to mitochondrial NADH (Fig. 1). Just as uncoupling promotes the conversion of mitochondrial NADH to NAD, inhibition of electron transport is expected to promote conversion of mitochondrial NAD into NADH. As expected, application of electron transport inhibitors increased autofluorescence in all type-1 cells tested to 144% (±7.9%, n= 6, P < 0.05) of control with cyanide (1–2 mm) and 136% (± 1.8%, n= 8, P < 0.005) of control with rotenone (1 μm). Similar results have also been obtained with other inhibitors of electron transport, including azide (data not shown) and hydrogen sulphide (Buckler, 2012). Uncouplers and electron transport inhibitors can therefore be used to define a range for the mitochondrial NADH signal between conditions of maximal reduction and maximal oxidation (see Fig. 1A).
In preliminary experiments we found that superfusing type-1 cells with a moderately hypoxic Tyrode (equilibrated with 2.5% oxygen) also caused a rapid and reversible increase in autofluorescence in all cells/clusters tested (to 123 ± 1.6% of control, n= 7, P < 0.001). This effect of hypoxia was abolished in the presence of cyanide (Fig. 1A, n= 4), rotenone (Fig. 1B, n= 4) and FCCP (Fig. 1, n= 4). These data indicate that the increase in autofluorescence during hypoxia is due to an increase in mitochondrial NADH.
We next sought to define the oxygen sensitivity of changes in mitochondrial NADH in type-1 cells. Experiments were conducted by equilibrating normal Tyrode with a range of different gas mixtures containing 0.5, 1, 2.5 and 5% oxygen (all with 5% CO2). Measurements of in the recording chamber, made using a 50 μm fibre optic oxygen probe (PreSens), returned values of 3.8, 7.6, 19 and 38 mmHg with each of these gas mixtures, respectively. Anoxic solutions were prepared by equilibrating Tyrode with 5% CO2/96% N2 for > 10 min followed by addition of 100–200 μm Na2S2O4 before use. Exposure of type-1 cells to hypoxic/anoxic Tyrode resulted in an increase in cellular autofluorescence (P < 0.001, n= 9, RM-ANOVA) and this was significant at every level of hypoxia tested (P < 0.001, n= 9, post hoc testing by Holm–Sidak method; see Fig. 2A). The increase in fluorescence at each level of hypoxia was expressed as a percentage of the increase seen in anoxic solution (relative to control) and is plotted in Fig. 2C (data points in anoxia are defined as 100% and those in normal Tyrode (= 150 mm Hg) as 0%). Fitting a rectangular hyperbola (see Methods) to these data returned an R2 value of 0.84 and a P50 (the at which a half maximal effect would be observed) of 15 mmHg.
Hypoxia is known to promote Ca2+ influx in type-1 cells (Buckler & Vaughan Jones, 1994), elevation of [Ca2+]i promotes mitochondrial Ca2+ uptake and elevation of mitochondrial [Ca2+]i activates NADH producing dehydrogenases (see Discussion). The effects of hypoxia upon NADH could therefore be secondary to these events rather that an inherent property of mitochondrial function. We therefore repeated the above experiments under Ca2+-free conditions wherein hypoxia has minimal effect upon cytosolic Ca2+ in rat type-1 cells (Buckler & Vaughan Jones, 1994). We performed 17 recordings of NADH autofluorescence in type-1 cells using the above gas mixtures. In 11 recordings we also included a more severe hypoxia (0 oxygen, = 2 mm Hg) and in six recordings we included a solution equilibrated with 10% oxygen. Again there was a significant increase in autofluorescence at all levels of hypoxia tested including 10% oxygen (P < 0.05, RM-ANOVA with Holm–Sidak post hoc testing against control; see Fig. 2B). The increase in fluorescence at each level of hypoxia was again expressed as a percentage of the increase seen in anoxic solution (relative to control) and plotted as a function of (Fig. 2D). Fitting a rectangular hyperbola to these data returned an R2 value of 0.87 and a P50 of 46 mm Hg.
As these data suggest an extraordinary degree of oxygen sensitivity in type-1 cells we performed control experiments using another cell type. For this we chose sympathetic neurons from the SCG. These cells were chosen for a number of reasons: (i) the carotid bodies are themselves regarded as part of the sympathetic nervous system (sympathetic paraganglia), (ii) sympathetic neurons are not known to have any oxygen sensing role and (iii) their location in close proximity to the carotid body plus the ability to dissociate neurons from these ganglia using the same enzymatic solutions as those used for dissociating carotid body cells enabled us to ensure that the procedure for preparing isolated SCG neurons was almost identical to that used for type-1 cells (see Methods).
The effects of hypoxia upon autofluorescence in SCG neurons was studied in Ca2+-free media as above. Unlike type-1 cells, hypoxia had very little effect upon autofluorescence in SCG neurons whereas anoxia caused a robust increase in autofluorescence, and FCCP caused a robust decrease in fluorescence (Fig. 3A). Data obtained in the presence of hypoxia was again converted into a percenatge increase relative to that caused by anoxia. Analysis of these data by RM-ANOVA revealed a significant effect of oxygen overall (P < 0.001, n= 10), although post hoc testing revealed that 2.5% oxygen had no significant effect on fluorescence and that whist there was a statistically significant effect of both 0.5 and 1% oxygen (P < 0.05) these effects were very small (7 and 3% of the response to anoxia, respectively). These data are plotted as a function of in Fig. 3B. A rectangular hyperbola fitted to these data returned an R2 value of 0.99 and a P50 of 0.3 mmHg, a value 100-fold less than that obtained in type-1 cells under otherwise identical recording conditions. Data from type-1 cells (from Fig. 2D) are reproduced in Fig. 3B (in grey) to facilitate comparison with that obtained from SCG neurons.
The significance of these observations is expounded upon further in the Discussion. Suffice to say that our data on type-1 cells are dramatically at variance with conventional wisdom regarding the normal oxygen requirements of mitochondrial respiration but confirm the earlier work of Duchen & Biscoe (1992a) on rabbit type-1 cells. We therefore sought to investigate this phenomenon further.
Mitochondrial membrane potential
Mitochondrial membrane potential was measured using the dye Rh123 in de-quench mode (see Methods). Recordings of ψm were performed under control conditions and in Ca2+-free Tyrode to prevent voltage-gated Ca2+ influx during exposure to either hypoxia or FCCP (Buckler & Vaughan Jones, 1998) as Ca2+ influx has been shown to depolarise type-1 cell mitochondria (Duchen & Biscoe, 1992b). FCCP at 1 μm was applied briefly (for approx. 20 s) both before and after testing the effects of hypoxia upon ψm. Application of FCCP caused a rapid increase in Rh123 fluorescence due to mitochondrial depolarisation and redistribution of Rh123 from mitochondria to cytosol. This response was assumed to represent the maximum (100%) Rh123 fluorescence attainable with complete mitochondrial depolarisation. Measurement of Rh123 fluorescence in the presence of 150 mmHg O2 was taken as the minimum level of Rh123 fluorescence (0%). Responses to hypoxia were then quantified on this relative 0–100% scale (see Fig. 4B). Hypoxia and anoxia caused a significant increase in Rh123 fluorescence both in the presence (P < 0.001, n= 11 RM-ANOVA) and in the absence of external calcium (P < 0.001, n= 11 RM-ANOVA on ranks). This effect was significant (P < 0.05 Holm–Sidak post hoc test) for all levels of hypoxia in the presence of extracellular calcium and 2.5% oxygen and below in the absence of extracellular calcium (Dunnett's post hoc test; note that although data obtained at 5% oxygen failed to reach statistical significance in this test, an increase in Rh123 fluorescence was seen in 10 of the 11 recordings). The increase in fluorescence at each level of hypoxia is plotted as a function of in Fig. 4C and D. Fitting a rectangular hyperbola to these data returned an R2 value of 0.96 and a P50 of 3.1 mmHg in the presence of external calcium and an R2 value of 0.85 and a P50 of 3.3 mmHg in the absence of external calcium. Similar results were also obtained in the presence of extracellular calcium and the calcium channel antagonist Ni2+, and in a subset of data in normal Ca Tyrode representing recordings from single cells only (see supplemental material).
It is notable that the degree of depolarisation (approx. 22%) in complete anoxia was considerably less than that seen with FCCP despite the fact that anoxia should halt electron transport. In part this may simply be due to the fact that ψm did not appear to reach a steady state during the brief exposures to anoxia and so our measurement will underestimate the full effect of anoxia. The failure to rapidly and fully depolarise under anoxic conditions is probably due to reversal of the ATP synthase such that ATP is consumed in order to pump protons out of the inner matrix space and help maintain ψm.
Effects of hypoxia on electron transport
The mitochondrial depolarisation observed in the above experiments could either represent an unusually high degree of oxygen sensitivity in the electron transport chain, or it could be a secondary consequence of increased energy demand leading to increased ATP synthase activity and proton influx into the inner mitochondrial matrix. To evaluate the effects of hypoxia on mitochondrial electron transport alone we sought to isolate ψm and electron transport from changes in energy demand and the activity of ATP synthase by treating cells with the ATP synthase inhibitor oligomycin. We also increased passive proton leak by applying a low level of uncoupler. Under these conditions ψm reaches a steady state when the rate of proton translocation by complexes I, II and IV equals passive proton back flux via the uncoupler. Passive proton leak is dependent upon both ψm and the leak conductance to protons (gH+). If we assume that gH+ is relatively constant then steady-state ψm will be linearly related to proton extrusion. Thus, changes in steady-state ψm will reflect changes in electron transport.
Figure 5A shows the protocol employed in these experiments. Cells, bathed in Ca2+-free Tyrode throughout, were first exposed to 1 μm FCCP for 20 s to calibrate the Rh123. Cells were then exposed to 75 nm FCCP and 2.5 μg ml−1 oligomycin. Oxygen levels in this solution were then reduced, as indicated, in the range 5% to anoxia. Finally, the FCCP/oligomycin solution was removed and a second calibration in 1 μm FCCP performed. Both hypoxia and anoxia caused a brisk and reversible depolarisation in ψm that reached a steady state within 20 s (P < 0.001, n= 12, RM-ANOVA). This effect was significant (P < 0.001) at all levels of hypoxia tested. These data are plotted as a percentage of maximal depolarisation against in Fig. 5C (taking fluorescence recorded in the presence of 75 nm FCCP and oligomycin in air equilibrated solutions as the baseline, 0%). Fitting a rectangular hyperbola to these data returned an R2 value of 0.97 and a P50 of 5.4 mmHg.
The above data indicate that electron transport in type-1 cells would appear to have an unusually high degree of oxygen sensitivity. We therefore repeated the experiment with SCG neurons. The protocol used was identical to that above save for the use of fewer levels of hypoxia (1 and 0.5%). As can be seen from Fig. 5B anoxia caused a rapid depolarisation of ψm to a level comparable to that observed with 1 μm FCCP but little effect was discernible for either of the two hypoxic solutions. The extent of the increase in Rh123 fluorescence as a percentage of that caused by 1 μm FCCP was calculated (as above) and is plotted as a function of in Fig. 5D. RM-ANOVA showed an effect of oxygen on Rh123 fluorescence (P < 0.001, n= 7) but post hoc testing (Holm–Sidak) revealed that this was only significant for anoxia. In view of this we did not attempt to fit a rectangular hyperbola to the data (symbols in Fig. 5D are joined by straight lines).
Oxygen sensitivity of cytochrome oxidase
In their classic paper, Mills & Jöbsis (1972) hypothesised that carotid body cells might contain a cytochrome oxidase with a low oxygen affinity. Such a situation could provide an explanation for the data presented above. We therefore sought to try to assay the oxygen sensitivity of cytochrome oxidase in isolation from the rest of the electron transport chain. To do this we inhibited electron transport complexes I and III and the ATP synthase with a combination of rotenone (1 μm), myxothiazol (0.5 μm), antimycin A (1 μm) and oligomycin (2.5 μg ml−1). This cocktail produced a full depolarisation of mitochondria in both Type-1 cells and SCG neurons, i.e. application of 1 μm FCCP had no further effect upon Rh123 fluorescence (see Fig. 6A and C). Subsequent addition of 5 mm ascorbate plus 40 μg ml−1 TMPD, an artificial electron donor capable of reducing cytochrome c, resulted in a rapid repolarisation of ψm (Fig. 6A and C). Under these conditions ψm is assumed to be maintained by proton pumping through complex IV alone, with electrons being passed exclusively from TMPD, via cytochrome c, through complex IV to oxygen. The effects of reduced oxygen levels on ψm were then tested under these conditions. As can be seen from Fig. 6A, both hypoxia and anoxia caused a depolarisation of ψm in type-1 cells. These effects were quantified as percentage of maximal depolarisation (in 1 μm FCCP) with the baseline (0%) measured in the presence of all inhibitors plus TMPD and ascorbate in air-equilibrated solutions. These data are plotted as a function of in Fig. 6B for type-1 cells and Fig. 6D for SCG neurons. Reduction of oxygen level caused a significant depolarisation of ψm in type-1 cells (P < 0.001, n= 14, RM-ANOVA). Post hoc testing against control (Holm–Sidak method) revealed a significant depolarisation at every level of hypoxia tested (P < 0.001). Fitting a rectangular hyperbola to these data returned an R2 value of 0.95 and a P50 of 2.6 mmHg. In SCG neurons reducing oxygen also depolarised ψm (P < 0.001, RM-ANOVA, n= 5). Post hoc testing (Holm–Sidak), however, revealed that only anoxia had a significant effect in SCG neurons. In view of this we did not attempt to fit a rectangular hyperbola to data from SCG neurons; the data points shown in Fig. 6D are joined by straight lines only.
O2 sensitivity of mitochondrial function vs. chemoreception
One of the objectives of this study was to establish whether mitochondrial respiration in type-1 cells had the requisite oxygen sensitivity to allow it to play a role in acute oxygen sensing. Figure 7 shows measurements of the oxygen sensitivity of calcium signalling in type-1 cells. In this, as in previous studies (Buckler & Vaughan Jones, 1994), hypoxia produced a rapid increase in [Ca2+]i in type-1 cells (P < 0.001, n= 11, RM-ANOVA on ranks). Fitting a rectangular hyperbola to these data returned a value for R2 of 0.52 and P50 of 12.5 mm Hg. An equivalent degree of oxygen sensitivity has also been reported for background potassium channel activity in type-1 cells (Buckler, 1997). The [Ca2+]i data were normalised, relative to baseline [Ca2+]i in air/CO2-equilibrated saline (0%) and maximum [Ca2+]i in anoxia (100%), and are re-plotted in Fig 7C. The oxygen sensitivity of type-1 cell NADH, ψm, electron transport and cytochrome oxidase activity are also plotted in Fig. 7C, for comparison.
Figure 7. Oxygen sensitivity of calcium signalling and mitochondrial function A, original recording of intracellular Ca2+ concentration measured using Indo-1 in a small cluster of type-1 cells bathed in normal Tyrode. Cells are exposed to graded hypoxic stimuli from 5% oxygen down to anoxia. B, summary data (mean ± SEM) from 11 recordings as in A. Curve is a rectangular hyperbola fitted to these data (see text for details). C, effects of hypoxia on a relative scale of 0–100% where 0%= baseline [Ca2+]i recorded in 20% O2 and 100%= maximal [Ca2+]i recorded in anoxia. These data (mean ± SEM) are joined by straight lines only. Curves represent oxygen sensitivity of various measures of mitochondrial function in the form of hyperbolic functions fitted to the original data, including: NADH from Fig 2C, ψm from Fig 4C (both in normal Ca2+ Tyrode), electron transport from Fig. 5C and cytochrome oxidase activity from Fig. 6C. Hyperbolas have been rescaled, where necessary, to a 0 (baseline in 20% O2) to 100% (maximum in anoxia) scale.
Download figure to PowerPoint