Responses of glomus cells to hypoxia and acidosis

I read with interest the publication by Lu et al. (2013) in which the authors claim to have found evidence for uncoupling and reciprocity in the responses of carotid body glomus cells to hypoxic and acidic stimuli. By measuring changes in intracellular calcium as an index of chemosensitivity, the authors argue that cells that are highly sensitive to hypoxia are less sensitive to acidosis and vice versa. The authors refer to this as a reciprocal sensitivity although it is not entirely clear if by this they mean that responses to acidosis and hypoxia are negatively correlated (i.e. Y=−aX+b) or inverse (Y=a/X+b).

Whilst this is an intriguing hypothesis I do not believe that it is supported by the data. There are three main issues in particular to which I would draw your attention with regard to the analysis of the data presented.

1. The data set presented on Rat type-1 cells has not been subject to any statistical analysis that would support either the existence of distinct populations of cells (acid sensitive vs. hypoxia sensitive) or any correlation between acid sensitivity and oxygen sensitivity. Conclusions are presented based upon an apparently subjective analysis of the numbers of cells having responses greater or less than some arbitrarily chosen value. Moreover, casual inspection of the histogram in fig. 3E does not immediately suggest the presence of a bimodal distribution for either acid sensitivity or hypoxic sensitivity.

2. The data sets presented for mice (in fig. 6) have been manipulated in an extraordinary manner. Data have been divided into two groups based upon whether the [Ca²⁺]_i response to cyanide (CN, used as a surrogate for hypoxia) is greater than the response to acidosis or not. The two groups are then plotted in different colours and subject to linear regression analysis separately with the regressions forced to pass through the origin. The only possible result from such an analysis is that the regression line fitted to the group of data selected such that CN>acid must have a gradient >1 and that for the group CN<acid must have a gradient <1 and >0. The authors then argue that the (inevitable) differences in the gradients between the two groups ‘suggest two distinct cell populations’. They also argue that ‘the reciprocal nature of the responses was evident’ from this analysis despite the fact that it produced a positive gradient for both groups indicating a positive correlation between CN response and acid response (which is again inevitable as the regression is forced to go through 0 and all values for X and Y are positive). I believe that the segregation of the data into two groups in this manner introduces an undue bias that effectively invalidates any conclusions that might be drawn from this analysis.

3. Data sets appear underpowered. This is particularly true for one of the mouse genotypes (fig. 6C). Even in the rat data set however there are very few cells with ‘high’[Ca²⁺]_i responses to acidosis and many of these may have come from just one cell cluster (that shown in fig. 3C). To test the hypothesis that there are distinct populations of cells within each carotid body it is also necessary to exclude variability in responses which may be due to differences between individuals (due to genetic background, age, environment or health).

In summary I do not believe that any conclusions can be drawn from the analyses presented in this study regarding either correlation between acid sensitivity and oxygen sensitivity or the existence of discrete populations of glomus cells. The only conclusion which would appear to be supported empirically is that most glomus cells are sensitive in some degree to both stimuli (as has been previously reported).