Article activity feed

  1. Review coordinated via ASAPbio’s crowd preprint review

    This review reflects comments and contributions by Anchal Chandra, Luciana Gallo, Joachim Goedhart, Sónia Gomes Pereira, Samuel Lord, Dipika Mishra, Ehssan Moglad, Arthur Molines, Sanjeev Sharma. Review synthesized by Iratxe Puebla.


    The study reports single-cell intracellular pH (pHi) measurements in different cell lines to measure spatiotemporal pHi dynamics during cell cycle progression. The manuscript reports an increase in pHi at the G2/M transition, decreased pHi at the G1/S boundary, S/G2 boundary, and prior to division, and increases during mid-S phase and G2, and suggests that pHi dynamics are necessary for cell cycle progression.

    The reviewers praised the topic of the study, measuring intracellular pH during the cell cycle and looking at the heterogeneity between cells are both important questions. However, there were some questions raised about the methodology as well as the interpretation of the data, as outlined below.

    Comments about methodology

    The pH sensor used in the study has been used previously but the single-cell level use requires new types of control and validations. It would be relevant to report:

    • What is the measurement error?
    • How efficient is the permeabilization protocol?
    • How homogeneous is the expression of the sensor? How does the expression level impact the pHi readout?

    These technical parameters could explain the heterogeneity in pHi reported in Figures 1,2,3, and they are relevant to understand if the fluctuations reported are relevant biologically or at the level of technical variability.

    Recommend providing additional details on the methodology for single cell pHi measurements, to ensure the experiments can be fully reproduced. Please report sample sizes.

    There is apparent intra-cellular heterogeneity present within each cell. The text should highlight whether the cytoplasm is heterogenous in pH. The study uses a single ROI per cell to measure intracellular pHi, however, if the cytoplasm is heterogeneous as some images show, the location of the ROI can influence the readout. It is recommended to use image analysis tools to segment cells and use the whole signal rather than a selected portion.

    There are concerns about the statistical analysis for several figures (including Figs 2I, 3I, and 5), in particular regarding the calculation of p-values based on multiple measurements or cells within each sample. The t-test and ANOVA assume that each measurement is independent, while multiple nuclei within the same sample are not independent. Recommend not reporting p-values or averaging together the values from each sample and then calculate the p-value using those sample-level means. For more information, see https://doi.org/10.1371/journal.pbio.2005282 and https://doi.org/10.1083/jcb.202001064

    The median is used as the reporter of the populations, the context for this choice is unclear. There are concerns about reporting standard deviation to estimate the spread around the median.

    Specific comments

    Introduction ‘In normal cells, intracellular pH (pHi) is near neutral (∼7.2)...’ - Could the text specify the type of cells the statement relates to, does it apply to all eukaryotes, mammalian cells, or even more specific and only demonstrated for human cells?

    Results ‘single-cell standardization is performed using buffers of known pH containing the protonophore nigericin (Fig. 1A, see methods for details).’ - The experiments use two pH extremes (~6.7 and 7.7 per the Materials & Methods)) and assume a linear relationship of the emission ratio between these extremes. Is this linear relationship verified? The supplementary Fig S2. shows an increase in signal across just two points. Suggest presenting an analysis of the biosensor across 4-5 different pH points to demonstrate linearity and dynamic range within the first set of figures. Plotting the ratio as well as the fluorescence intensities of individual channels across these pH ranges would also be relevant.

    Figure 1

    Is there an explanation for the signal from the nucleus? It seems initially more acidic than the cytoplasm and then it does not change as much as the cytoplasm during the nigericin treatment. Is this due to bad permeabilization?

    B) NL20, C) A549, and D) H1299’ - Please indicate which cells are normal and which cancerous.

    E-G) Histograms of single-cell pHi in E) NL20 (n=173, 3 biological replicates), F) A549 (n=424, 4 biological replicates), and G) H1299 (n=315, 3 biological replicates).’ - The distributions are aggregates of 3 or 4 biological replicates. What do the distributions look like in each replicate? Are the differences between conditions visible? If the E, F and G histograms are generated using data pooled from different replicates, recommend separating the replicates and presenting the distributions separately for each replicate experiment.

    We next measured single-cell pHi in individual NL20-mCh-pHl (Fig. 1B), A549-mCh-pHl (Fig. 1C), and H1299-mCh-pHl (Fig. 1D) cells’ - From the methods: "Individual Regions of Interest (ROI) are drawn for each cell in each condition (initial, high pH nigericin, and low pH nigericin), and mCherry aggregates are removed using thresholding holes." From the cells in the image, it appears that the cytoplasmic signal is not homogenous and suggests that the choice of ROI will affect the reading for each cell. In this condition, to do single cell measurements, it is recommended to use the signal from the entire cell (cytoplasm) rather than using an ROI.

    Representative pHluorin and mCherry channels and single-cell standardization lines can be found in Fig. S2’ - The pH probe appears to be comprised of a straight fusion between the pH sensitive GFP (pHluorin) and pH insensitive mCherry. One would expect that the ratio of GFP to mCherry is only determined by pH (and not by expression level or excitation intensity). A question arises around the dynamic range (shown in fig. S2) being different between the different cell lines. For instance, the ratios observed for pH=7 and pH=7.8 are 3 and 8 for NL20, 3 and 5 for A549, and 0.5 and 2 for H1299. Can an explanation be provided for the differences between cell lines? Were the single cell measurements verified with a dye (BCECF/SNARF/SNAFL)? Was the permeabilization protocol validated?

    (NL20-mCh-pHl) (Fig. 1E; 7.42±0.07).’ - The first sentence of the results section indicates "In normal epithelial cells, pHi is near neutral (∼7.2), while cancer cells have a constitutively increased pHi (pHi>7.4)." According to this statement, the NL20 cell line has a pHi corresponding to cancer cells, can this be clarified?

    These data show the advantages of measuring single-cell pHi under physiological culture conditions that match population averages, but also provide pHi distributions lost at the population level.’ - The single cell data reveal the heterogeneity, can further explanation be provided for the advantage gained by these data over bulk measurements?

    ‘These data also show that pHi is heterogeneous even in clonal, genetically identical, cell lines, suggesting pHi may be a biomarker for non-genetic cell phenotype’ - The data show heterogeneity, but do not address how much and what the source of heterogeneity is. It would be helpful to: report the error on the measurement, compare the spread of pHi to something else to get a sense of the normal level of noise in the measurement. Could this be compared to the spread of mCherry intensity, to check if there is more spread in pHi than in expression level of the construct.

    Independent measurement of the heterogeneity of the pH (e.g. with another probe/dye) would shed some light. The heterogeneity (or spread) of basal biosensor distributions could be compared against the distributions achieved after nigericin treatment - to bring out the differences in biological heterogeneity versus measurement error. The results could then further elaborate on whether the biological heterogeneity has relevance in the regulation of cellular processes.

    pHi in physiological environments’ - Can some clarification be provided for how prior studies did not follow physiological conditions, while the current set up would provide such physiological conditions?

    We synchronized H1299-mCh-pHl cells using Palbociclib’ - The study uses H1299 line in most figures hereafter, A549 line in some while not the NL-20 lung cells, can some justification be provided for the selection of cell lines for specific experiments.

    In this representative replicate, we observed single-cell pHi significantly decreased between 0 and 4 h, significantly increased between 4 and 8 h, decreased between 8 and 12 h, and increased again between 12 and 24 h (Fig. 2D).’ - It is not clear whether these data are consistent with the other replicates (Figure S3). For example, another replicate shows a consistent decrease of pHi between 0-4h and 4-8h, which is not the case for the example shown in the main figure. Can some clarification be added about discrepancies between replicates. In Figure S3 the different time points were statistically compared to their previous time point, can the same statistical analysis be applied to the replicate in Figure 2?

    Figure 2

    Box and whisker plots of F) cyclin E1, G) cyclin A2, and H) cyclin B1 immunoblot data across 3 biological replicates’ - There is a concern about the use of boxplots for n=3 as they summarize the data into 5 statistics (2x whiskers, Q1, Q3 and the median): www.nature.com/articles/nmeth.2813. It is recommended to show the individual data with a dotplot.

    Figure 2I. Violin plots of raw pHi across 3 biological replicates’ - A superplot is recommended for identifying the biological vs. technical replicates: https://doi.org/10.1083/jcb.202001064. The significance should be determined based on n=3 (not on the pooled technical replicates).

    Cyclin immunoblots and pHi agreed across 3 biological replicates, and additional blots are shown in Fig. S3.’ -The replicates from Fig. S3 and Fig. 2 do not appear to show a clear behavior. For example at 4h, two replicates show a decrease while the third shows an increase in pHi. Could some clarification be added for this?

    When pHi measurements on Palbociclib-treated cells were compared over three biological replicates, we found that pHi significantly decreased at the G1/S transition (4 h, 7.75±0.15) and in late S phase (12 h, 7.69±0.09), significantly increased at G2/M (24 h, 7.82±0.11) (Fig. 2I), and then significantly decreased once more at the end of the experiment in asynchronous cells (36 h, 7.67±0.10) (Fig. 2I).’ - The population in Fig 1 shows a large spread from around 7.4 to 8. This emcopasses all the distribution shown in Fig 2 and if the individual time points are undersampled, small fluctuations are expected in the mean and the median. Can some comment be provided about the potential influence of undersampling on the fluctuation? If the fluctuations were due to undersampling they would be random and could explain why the replicates are not in very good agreement. Also, can some clarification be added about how many cells were measured in each time point.

    Figure 3 - The various replicates provided here and in Sup. Fig. 4 show variability. For example, only the replicate in the main figure shows a decrease at 4h and 12h. The third and fourth replicates are in good agreement and for those pHi stays roughly the same and then drops between 12h and 24h. Should this be reflected in the text?

    From the values on the y axis for each time point and replicate, it seems that the sample size varies between replicates. There is the risk of undersampling, and also that if one replicate contains much more cells than others, it would dominate the distributions once the data are pulled together. Can the sample size for each time point and replicates be reported?

    and decreases at 12 h and 24 h (Fig. S5B-C)’ - The text previously reported that pHi increased between 12 and 24h for H1299 cells, here it reports that there is a decrease at 24h. Please provide a clarification.

    we established a time-lapse approach to track pHi dynamics over an entire cell cycle in a single cell.’ - This is a robust approach to detect pH changes over time. ‘we selected prophase as a “normalization point” for each individual dividing cell’ - Recommend referring to "synchronization point" instead of ‘normalization’.

    Figure 4

    The paper shows that synchronization alters baseline pHi. Could a similar experiment be completed without synchronization?

    A) Representative stills of Video S1 of a dividing H1299-mCh-pHl cell at indicated time (h)’ - It would be good to compare this to the metastatic cells used to establish how much of the pHi fluctuations observed during the cell cycle are "cancer" related.

    Furthermore, the pHluorin increases observed over time in dividing cells are not correlated with increased mCherry fluorescence, which indicates pHluorin increases are not due to increases in biosensor expression (Fig. S8B-C).’ - It is great that this measurement was completed. However, from the plots provided Sup. Fig. 8 B and C, in dividers and non-dividers, it looks like the two signals (mCh intensity and pHluorin) are well correlated (first a decrease for a few hours, then it rises until 10h then it decreases). Could this indicate that the readout is influenced by protein concentration / expression? Suggest plotting the two signals vs each other’s on a scatter plot and formally testing for correlation.

    Figure 5

    For the FUCCI reporter, plotting mVenus and mCherry intensities normalized between the max and min value for each cell allows clear identification of transition between phases. It may be helpful to present example single cell traces from 5-10 cells for each treatment, to more clearly appreciate the cell cycle phase transitions and their durations on panels F,G and H.

    D) Single-cell pHi of H1299-FUCCI cells treated with EIPA and SO859 (E+S, n=233) to lower pHi, untreated (CRL, n=267), or treated with ammonium chloride (NH4Cl, n=202) to raise pHi (see methods for details)‘ - Please clarify how or when delta pHi was calculated for data in Fig. 5D.

    previous work in lower- order organisms’ - "lower-order" has a negative connotation, please consider re-phrasing to include the species or at least family of organisms.

    Discussion - Recommend further discussion about altered progression through cell cycle phases at different pHi and how it could be altered in cancer cells. Is increased intracellular pH in cancer cells related in any way to their increased proliferation? If so, which cell cycle steps are affected? High intracellular pH seems to elongate all phases except the M phase.

    Methods

    ‘Multiple Z-planes were collected with the center focal plane maintained using a Perfect Focus System (PFS).’ - Please report whether pH was analyzed on a projection, a single z-slice, each z-slice?

    Single-cell pHi measurements - Please provide additional detail for the protocol for the single cell pHi measurement. Include information on whether the work involves single image, stacks, projections, etc, and the size and location of the ROIs. Please also provide further context for the "mCherry aggregates", does this mean the construct is cleaved and the mCh aggregate? Does the GFP aggregate too?

    NIS Analysis Software//GraphPad Prism - Please report the version of the software used.

    Individual Regions of Interest (ROI) are drawn for each cell in each condition’ - Could the ROI on a few of the cells be drawn and highlighted in the main figures to show the size and location of the ROI?

    ‘8% laser power for GFP; 700 ms exposure time and 10% laser power for TxRed; and 100 ms exposure time and 5% laser power DAPI’ - Please report the exact wavelengths used to excite the fluorophore, (e.g. 8% power of a 488 laser (GFP excitation)).

    Supplementary figures

    Figure S3 panel A - Should the calibration slope be the same for every cell? Can some explanation be provided for why some cells have a steeper slope than others?

    Figure S4 - Replicates appear to show different trends in pHi and Cyclins, which makes it difficult to interpret the data.

    Figure S8 panel A - This plot shows correlation between the two quantities, they both rise and fall at the same time. Can some clarification be provided.

    Was this evaluation helpful?