We would like to point out a potential source of data variation with the protocol developed by Rogers et al, which is uneven distribution of mitochondria following centrifugation onto the assay plate. When centrifuging particles onto microplates, we repeatedly observed horizontally skewed distributions of particles over the plate and within wells according to the differences in effective centrifugal force (Fig. 1A; all the figures in this comment are found here http://www.plosone.org/at... ).
This was also seen when centrifuging mitochondria onto the XF24 plate using a Beckman Coulter Avanti^® J-E centrifuge (Fig. 1B). In the Seahorse XF24/XF24-3 Analyzer, sensors for oxygen and pH are placed horizontally off-centre, with the oxygen sensors moved towards the right. Therefore, if uneven distribution of mitochondria occurs, the oxygen sensors will read higher mitochondria concentrations on the right half of the plate (columns 4 – 6) and lower concentrations in columns 1-3. In our experience, this leads to up to ~1.7fold variation in OCR results over the plate (Fig.2), indicating that the oxygen concentration measured by the sensor is largely determined by the local amount of mitochondria directly beneath it. Accordingly, the pH sensor, which is located off-centre towards the left, records the opposite tendency (Figure2 B).
We have not been able to rectify this problem by using smaller volumes of mitochondrial suspension during the centrifugation step (down to 30µl) or by modifying centrifugation speed and time combinations (including the 2,000g for 20 minutes as used by Rogers et al.) . However, variation in OCR between wells in columns 4 -6 was usually below 10%. Using only these wells, we obtained data (Figure 3) well comparable to those by Rogers et al (compare Figure 5C). We propose that checking the homogeneity of distribution across all wells is essential for oxygen consumption analyses in the Seahorse XF24, especially if using isolated mitochondria.
Acknowledgements
We thank Joanne Coulson (Seahorse Bioscience) for discussion and technical support.

Figures
http://www.plosone.org/at...

Figure legends
Fig. 1. Uneven distribution of particles following centrifugation in a microplate rotor. A) Yeast were centrifuged onto a 96 well plate using an IEC Centra-3C centrifuge at 2000rpm for 5 min with the plate in horizontal orientation. Particles are forced horizontally towards the outside of the plate in agreement with the relative centrifugal force that is proportionally to the effective radius. B) Mitochondria (5µg in 50µl suspension per well) were centrifuged onto an XF plate at 2000g for 20 minutes, using a Beckman Coulter Avanti^® J-E centrifuge equipped with a JS-5.3 rotor. Phase contrast images (20x objective) show different mitochondrial densities on the left and right side of well B1.
Fig. 2. OCR reads vary with column. Mouse liver mitochondria (10µg) were suspended in 50µl assay solution and centrifuged into each well (except the blank wells) at 2,500g for 10 minutes at 4°C using a Beckman Coulter Avanti^® J-E centrifuge equipped with a JS-5.3 rotor. The experiment was carried out with the following conditions: succinate (4mM) in the presence of rotenone (2µM) was used as substrate, ADP (2.5 µg/ml), oligomycin (4µM), FCCP (4µM) and antimycin A (4µM) were sequentially added. A) Raw OCR data per column. B) OCRs (top, calculated using maximum values for ADP and FCCP rates) and media alkalinisation during ADP-stimulated respiration (ECAR, bottom) per column. Data are means ± SD for 3 or 4 wells per column.

Fig.3. Respiration data from mouse liver mitochondria using only columns 4 -6 of the XF24 micro plates. Mitochondria were isolated from livers of 24 months old C57Bl6 mice. Data are mean ± SD from 5 separately measured animals. The data are comparable to the Figure 5C in Rogers et al.

Satomi Miwa and Thomas von Zglinicki
Institute for Ageing and Health, Newcastle University, Newcastle upon Tyne, NE4 5PL, U.K.