Talk:Rogers 2011 PLoS One
- 1 Erich Gnaiger: High throughput without high output?
- 2 Christos Chinopoulos: 2011-09-26
Erich Gnaiger: High throughput without high output?
Rogers et al 2011 describe the application of a strongly advertised multiwell system (XF, 'Seahorse Bioscience') for respirometry with isolated mitochondria. Whereas multiwell systems do provide high throughput, it must be critically assessed if this is equivalent to high output of useful data and information.
Based on long-term expertise in high-resolution respirometry (HRR), bioenergetics and mitochondrial physiology (Gnaiger et al 1995; Gnaiger et al 2000; Pesta and Gnaiger 2012), the paper by Rogers et al (2011) is discussed. Most arguments raised in this discussion are not restricted to applications with isolated mitochondria, but address general deficiencies of the multiwell system.
Conflicts of interest:
- Rogers GW and Ferrick DA are employees of Seahorse Bioscience (at the time of publication).
- Gnaiger E is founder and managing director of Oroboros Instruments.
Instrumental specifications of the XF multiwell system are missing
1. No specifications are provided on the limit of detection of respiratory flux, and no information is given on the validation of the large corrections applied for oxygen backdiffusion. The substantial problems involved can be deduced from Figure 2. Fig. 2B shows oxygen pressures dropping to the level of about -10 mmHg (-1.3 kPa; the XF method makes negative oxygen pressures possible), while the corresponding respiratory oxygen flux ('OCR') is high (in the range of 800 pmol/min; Fig. 2A).
- Comparison with HRR: In the Oroboros O2k (O2k for HRR), plastic with high oxygen solubility is strictly avoided as a material in contact with the sample and incubation medium. Thus oxygen backdiffusion is minimized. Routine instrumental background tests are described for HRR as a SOP, to validate instrumental performance and use the resulting parameters for on-line correction of respiratory flux (Gnaiger et al 1995; Gnaiger 2001; Gnaiger 2008). The limit of detection by the O2k of oxygen flux is +- 1 pmol.s-1.ml-1 over the entire experimental oxygen range, and oxygen levels can be measured accurately and be maintained at steady state by auomatic feedback control down to 5 nM or 0.0005 kPa (see O2k-Specifications).
2. The effective volume of the closed multiwell chamber is not given accurately, but is stated to be "~7 µl" (p. 3) or "approximately 7 µl" (p. 10). Any uncertainty of the actual volume has a direct effect on the inaccuracy of the measured respiratory flux.
3. The temperature stability and temperature range are not specified for the multiwell system. How large are the temperature differences between central and peripheral wells?
- Comparison with O2k-Specifications: +- 0.001 °C in the range of 2 °C to 47 °C, with highly homogenous temperature in the chamber surrounded by the constant-temperature copper block with well stirred medium.
4. The oxygen calibration is not described, no information is provided on the integration of barometric pressure or oxygen solubility in the multiwell-calibration procedure. How is the medium stirred for equilibration with air during air calibration? “Ambient oxygen tension” is described as ~158 mmHg (p. 5), but the relevant oxygen partial pressure for air saturation of an aqueous medium is that in air saturated with water vapour pressure, amounting to 147 mmHg (19.63 kPa at a standard barometric pressure of 100 kPa). This lack of accuracy in the calibration of the oxygen sensor may be related to the earlier publications of one of the co-authors of Rogers et al (2011), since Brand et al (1993) assume that air saturated medium at 37 °C contains 220 µM O2, which would even exceed the oxygen solubility in pure water (which contains 207 µM O2 at 37 °C in equilibrium with air at a barometric pressure at sea level of 100 kPa), and this overestimates the solubility in experimental salt solutions to an even higher extent.
- Comparison with HRR: The high scientific standard based on physical chemistry is implemented into the DatLab software supporting the O2k, which receives the signal of the continuously monitored experimental temperature and barometric pressure (electronic barometric pressure transducer), and calculates the oxygen solubility in pure water, corrected for an oxygen solubility factor of the experimental medium, following the protocols and SOP for HRR.
Uncoupled flux does not reflect electron transfer system capacity
1. Addition of the uncoupler FCCP at a final concentration of 4 µM (p. 4) is used by Rogers et al (2011) for inducing uncoupled respiration. This uncoupled respiration is not stable, and no comments are provided by the authors on the steep decline of flux always shown after addition of FCCP (Figs. 3A, 4A, 4C). Validation of the optimum FCCP concentration for maximum flux is required (not shown). The fact that uncoupled flux was lower than ADP stimulated OXPHOS capacity is direct proof of an artefact in evaluating electron transfer system (ETS) capacity (Fig. 4A, Fig. 5, Fig. 8).
2. In contrast to the Figures showing ET-capacity (E) that is lower than OXPHOS capacity (P), in the text the RCR (P/L) is lower than E/L (p. 7 and 9), but the discrepancy is not explained.
- Comparison with HRR: Uncoupler is titrated in several steps in each experiment, to quantify ET-capacity (non-coupled respiration at optimum uncoupler concentration for maximum flux), which can be performed manually (Huetter et al 2006), or automatically using the TIP2k (Gnaiger 2008). The optimum FCCP concentration varies as a function of medium, pathophysiological state, and pharmacological state (e.g. in the presence versus absence of oligomycin). At an optimum FCCP concentration in the non-coupled state of ET-capacity, flux is stable over time, declining as a function of time only after inhibitory concentrations of uncoupler are applied (Steinlechner-Maran et al 1996, Huetter et al 2004, Gnaiger 2008).
Lack of validation with high-resolution respirometry
Low-resolution instruments (compare HRR) with Clark-type polarographic oxygen sensors were used for validation of data obtained with the XF. With high variability in all applied systems (Fig. 5, particulary Fig. 5C), it is not surprising (and not informative) that all compared results “were in a similar range” (p. 7). This is the summary of a non-quantitative (no statistics was applied) methodology.
Lack of quality control of isolated mitochondria - lessons in bioenergetics
1. Although the RCR is an accepted parameter for the quality control of isolated mitochondria, injuries of the outer mitochondrial membrane are only poorly reflected by the RCR, except in highly damaged mitochondria (Rasmussen and Rasmussen 2000). Instead, a cytochrome c respiratory test is required, which has not been applied by Rogers et al (2011). Additional wells would be required for such a test with the XF, further reducing the effective throughput and increasing the cost of discharged plates.
- The cytochrome c test (addition of 10 µM cytochrom c is saturating with physiological substrates; Gnaiger and Kuznetsov 2002) can be applied quickly and routinely in HRR with mitochondrial preparations, not only as a quality control of the experimental preparation procedure, but also as a test for pathological injuries (Kuznetsov et al 2004).
2. The authors state that “the electron flow experiment (Fig. 6B, D) is designed to follow and interrogate each complex of the electron transport chain”. This is misleading, since it suggests that the simple protocols applied are comprehensive. The electron transfer system is comprised of several electron transport complexes, including electron transferring flavoprotein (ETF) and glycerolphosphate dehydrogenase. A limited number of Complex I-linked substrates (pyruvate+malate) was applied, not checking if these substrates actually provided maximum Complex I-linked respiratory capacity. In addition, succinate+rotenone provided electron entry through Complex II. The capacity of Complex IV was not estimated correctly, since no correction for chemical autoxidation of TMPD and ascorbate was applied (see Gnaiger and Kuznetsov, 2002, Gnaiger 2008). The additive effect of the combined application of substrates linked to Complex I+II was not studied, but only such a physiological substrate cocktail stimulates maximum OXPHOS and ET-capacity and shifts respiratory control downstream to Complexes III and IV and the phosphorylation system (Gnaiger 2009). Complex III was not interrogated.
- Comparison with HRR - high-output replaces wasteful high-throughput: Complex substrate-uncoupler-inhibitor titration (SUIT) protocols have been developed on the basis of HRR to obtain in a single experiment a large number of substrate and coupling control states in sequential titrations (Kuznetsov et al 2004; Boushel et al 2007; Aragones et al 2008; Lemieux et al 2011; Pesta and Gnaiger 2012; Pesta et al 2001). This HRR approach (with up to 10 chambers operated in parallel) provides high-information content, which in conjunction with the quantitative reliability is based on an economical high-output system - the Power-O2k.
3. Too low resolution to detect residual oxygen consumption (ROX): The authors state that “respiration stops” after injection of rotenone (p. 7). It is well established that rotenone induces ROS production, which constitutes part of uncoupled (decoupled) respiration (electron leak). With a sensitive instrument (HRR), therefore, it is easy to detect ROX after inhibition of Complex I with rotenone.
4. Respiratory capacity of the isolated mitochondria was not compared with improved isolation media (Brewer et al 2004), nor with improved storage and respiratory incubation media (Gnaiger et al 2000). The detrimental isolation conditions are apparent from the observation that even a partial step towards a high-quality medium (addition of substrate in the initial dilution; compare with Mitochondrial Preservation Medium) improved the respiratory control ratios (p. 3).
5. The difference or ratio between OXPHOS and ET-capacity is not interpreted and is not accurately measured. A low-throughput multiwell approach with many parallel wells receiving different concentrations of uncoupler would be required: time consuming and expensive.
- Valuable information can be obtained from the quantitative determination of OXPHOS (P) and ET-capacity (E). At a P/E ratio of 1.0, there is no limitation of OXPHOS capacity by the phosphorylation system, whereas the phosphorylation system exerts control on flux when P/E<1.0. This important information is obtained with HRR (Gnaiger 2009; Lemieux et al 2011; Pesta and Gnaiger 2011; Pesta et al 2012), but it is obscured in a multiwell high-throughput approach that yields artefacts (i.e. P/E>1.0) since it does not allow multi-step titrations of uncoupler. More information: MitoPedia: Respiratory states.
It is concluded that the multiwell format yields apparent high-throughput, but
1. without providing experimental flexibility: serial injections are restricted to 2 or 4 per well, in the 96 or 24 well system; no substrate kinetic data can be obtained at high throughput; no uncoupler titrations can be performed; oxygen and substrate gradients are not well defined in the unstirred well; the concentration of lipid-soluble compounds is not well defined due to solution in the plastic well and the high surface/volume ratio;
2. without facilitating the quantity of information: a large number of parallel replica is required to compensate for the poor reproducibility (contradictory results in Fig. 5C with P/E=1, but P/E>1.0 in Fig. 7B for mouse liver mitochondria respiring on succinate);
3. without minimizing the experimental time: an immense number of quality control tests is required before a high-throughput experimental series can even be started, and more quality control tests have to be performed for each experimental variation of pathophysiological states: narrow range of mitochondrial or cell density for reliable respiratory data, saturating substrate and inhibitor concentrations, optimum uncoupler concentrations, instrumental background correction, correction for autooxidation of TMPD and ascorbate, check for loss of mitochondria or cells to the supernatant and sides of the wells (<10% of total adhered protein was detectable in the supernatant [p. 7], but from such a high suspended mitochondrial fraction there may be continuous adherence to the sides of the wells); check for dilution effect in a typical sequence of four injections (50+55+60+65 µl per well).
4. High-resolution respirometry (HRR) offers a flexible and cost-effective high-output technology which may replace an expensive and wasteful high-throughput approach.
- Further information: HRR versus multiwell
- For a Comparison of common bioenergetic analysis methods see Zhang 2012 Nat Protoc
- In the spirit of Gentle Science.
--Gnaiger Erich 22:24, 24 September 2011 (CEST)
Christos Chinopoulos: 2011-09-26
- I am pleased to see your comments about the XF multiwell system, and its numerous deficiencies, combined to its extremely high costs. I agree with you 100%. I was wondering if you were willing to share their responses to your email, in case they send them to you directly, instead of openly in the MIG.
Best regards, Christos - firstname.lastname@example.org
- Reply: If the individual authors submitting their response agree, we will add the comments to this discussion page. Thanks, Erich
- But there was no response from the authors Rogers et al --Gnaiger Erich 19:26, 6 March 2012 (CET).