Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Jan 24;13(1):904.
doi: 10.15252/msb.20167258.

Protein abundance of AKT and ERK pathway components governs cell type-specific regulation of proliferation

Lorenz Adlung  1 , Sandip Kar  2   3   4 , Marie-Christine Wagner  1 , Bin She  1 , Sajib Chakraborty  1 , Jie Bao  5 , Susen Lattermann  1 , Melanie Boerries  5   6   7 , Hauke Busch  5   6   7 , Patrick Wuchter  8   9 , Anthony D Ho  8 , Jens Timmer  10 , Marcel Schilling  1 , Thomas Höfer  11   3 , Ursula Klingmüller  12   13
Affiliations

Protein abundance of AKT and ERK pathway components governs cell type-specific regulation of proliferation

Lorenz Adlung et al. Mol Syst Biol. .

Abstract

Signaling through the AKT and ERK pathways controls cell proliferation. However, the integrated regulation of this multistep process, involving signal processing, cell growth and cell cycle progression, is poorly understood. Here, we study different hematopoietic cell types, in which AKT and ERK signaling is triggered by erythropoietin (Epo). Although these cell types share the molecular network topology for pro-proliferative Epo signaling, they exhibit distinct proliferative responses. Iterating quantitative experiments and mathematical modeling, we identify two molecular sources for cell type-specific proliferation. First, cell type-specific protein abundance patterns cause differential signal flow along the AKT and ERK pathways. Second, downstream regulators of both pathways have differential effects on proliferation, suggesting that protein synthesis is rate-limiting for faster cycling cells while slower cell cycles are controlled at the G1-S progression. The integrated mathematical model of Epo-driven proliferation explains cell type-specific effects of targeted AKT and ERK inhibitors and faithfully predicts, based on the protein abundance, anti-proliferative effects of inhibitors in primary human erythroid progenitor cells. Our findings suggest that the effectiveness of targeted cancer therapy might become predictable from protein abundance.

Keywords: MAPK; 32D‐EpoR; BaF3‐EpoR; CFU‐E; PI3K.

PubMed Disclaimer

Figures

Figure 1
Figure 1. Characterization of Epo‐induced proliferation and signaling

  1. DNA content of mCFU‐E and BaF3‐EpoR cells in response to different Epo concentrations. [3H]‐thymidine incorporation was measured after 14 h (mCFU‐E) or 38 h (BaF3‐EpoR). Data represented as mean ± standard deviation, N = 3. Lines represent sigmoidal regression. EC50 values are given.

  2. Cell doubling with time in response to 50 U/ml Epo. Cell numbers were determined by manual counting using trypan blue exclusion assay. Data represented as mean ± standard deviation, N = 3. Lines represent exponential regression. Doubling time is indicated.

  3. Size determinations of mCFU‐E and BaF3‐EpoR cells. Exemplary fluorescence microscopy pictures upon Hoechst staining for nucleus visualization with 60× objective. The bar represents 10 μm distance (upper panel). Cell diameter was measured by imaging flow cytometry. Cytoplasm was stained with calcein, and nuclei were stained with DRAQ5. Probability density function of size distribution with indicated mean diameter of mCFU‐E and BaF3‐EpoR cells. All cells were growth factor‐deprived and unstimulated.

  4. Epo‐induced phosphorylation of EpoR, AKT, and ERK of mCFU‐E and BaF3‐EpoR cells. Above each panel, the number of growth factor‐deprived cells examined per time point is indicated as well as the concentration of Epo applied for stimulation. To the left, the position of the molecular weight marker is indicated in kDa and arrowheads indicate the position of the protein of interest. For the detection of the EpoR, 5 × 106 mCFU‐E cells were stimulated with 2.5 U/ml Epo and 5 × 106 BaF3‐EpoR cells were stimulated with 5 U/ml Epo. Cells were lysed, subjected to immunoprecipitation with anti‐EpoR, and were analyzed by immunoblotting using either anti‐pTyr (pEpoR) or anti‐EpoR (EpoR) antibodies. For the detection of AKT, mCFU‐E cells were stimulated with 2.5 U/ml Epo and BaF3‐EpoR cells were stimulated with 5 U/ml Epo. Per time point, cellular lysates equivalent to 5 × 105 mCFU‐E cells and 1 × 106 BaF3‐EpoR cells were analyzed by immunoblotting using anti‐pAKT and anti‐AKT antibodies. For the detection of ERK, cells were stimulated with 50 U/ml Epo. Per time point, cellular lysates equivalent to 8 × 105 cells were analyzed by immunoblotting. Immunoblot detection was performed with chemiluminescence utilizing a CCD camera device (ImageQuant).

  5. Absolute concentrations of pEpoR, pAKT, and ppERK in mCFU‐E and BaF3‐EpoR cells with time in response to 50 U/ml Epo. Experimental data of a representative experiment are depicted with filled circles, and dashed lines represent splines. Error bars indicate standard deviation estimated by an error model. N = 1.

  6. Expression of cell cycle indicator genes of mCFU‐E and BaF3‐EpoR cells in response to 5 U/ml Epo. Genes were selected based on microarray analysis. Experimental data are shown as fold change to unstimulated cells with mean ± standard deviation, N = 3. Welch modified two‐sample t‐test, *P < 0.05.

  7. Fold change of fractions of cells in S/G2/M phase of the cell cycle with respect to 0 U/ml Epo. Growth factor‐deprived cells were stimulated with indicated Epo doses for given time. Fractions in sub‐G1, G1, S, G2/M were determined by propidium iodide (PI) staining for DNA content. Data represented as mean ± standard deviation, N = 3. Welch modified two‐sample t‐test, n.s. = not significant, *P < 0.05, **P < 0.01, ***P < 0.005. The percentage of cells in S/G2/M phase in unstimulated cells is additionally indicated.

Source data are available online for this figure.
Figure 2
Figure 2. Mathematical modeling of the Epo‐induced AKT, ERK, and S6 activation
  1. A

    The model is represented as a process diagram and reactions are modulated by enzyme catalysis (circle‐headed lines) or inhibition (bar‐headed lines). Prefix “p” represents phosphorylated species. Bold framed species were experimentally measured. White capsules represent inhibitors. Binding of the ligand Epo to its cognate receptor results in the phosphorylation of EpoR. Receptor and associated complexes are depicted in yellow, the AKT pathway is red, the ERK pathway is blue, and cell cycle genes and the S6 network are shown in green.

  2. B–G

    Model calibration with time‐resolved quantitative immunoblot data of mCFU‐E cells in blue and BaF3‐EpoR cells in black. Growth factor‐deprived mCFU‐E cells (5 × 106 cells per condition) and BaF3‐EpoR cells (1 × 107 cells per condition) were stimulated with different Epo doses, and absolute concentrations were determined for pEpoR (B), pAKT (C), ppERK (D). The scale for pS6 (E) was estimated in arbitrary units. GTP‐Ras (F) and ppERK were determined upon stimulation with indicated, color‐coded Epo doses. pEpoR was analyzed by immunoprecipitation followed by immunoblotting, GTP‐Ras was analyzed after pulldown using a fusion protein harboring GST fused to the Ras binding domain of Raf‐1 followed by detection by quantitative immunoblotting. For pAkt and ppERK, cellular lysates were subjected to quantitative immunoblotting. Calibrator proteins were used for EpoR, AKT, GTP‐Ras, and ERK to facilitate the conversion to nM concentrations. Experimental data are represented by filled circles. Error bars represent standard deviation estimated by an error model. Solid lines represent model trajectories. N = 1.

Source data are available online for this figure.
Figure 3
Figure 3. Model predictions and experimental validations for negative regulators of AKT and ERK signaling

  1. Model calibration with mCFU‐E wild‐type cells, and mCFU‐E cells overexpressing SHIP1 or PTEN, and BaF3‐EpoR wild‐type cells. Experimental data are represented by filled circles. Error bars represent standard deviation estimated by an error model. Solid lines represent model trajectories. N = 1.

  2. Model prediction of pAKT dynamics in BaF3‐EpoR cells overexpressing PTEN or SHIP1. Model predictions are represented by solid lines. Experimental validation data obtained by quantitative immunoblotting are represented by filled circles. Error bars represent standard deviation estimated by an error model. A representative experiment is shown. N = 1.

  3. Model prediction for the basal expression level of the dual‐specific phosphatase (DUSP). DUSP abundance ratio between BaF3‐EpoR and mCFU‐E cells was identified by the mathematical model. The solid line indicates the profile likelihood. The dashed red line indicates the threshold to assess point‐wise 95% confidence interval. The asterisk indicates the optimal parameter value.

  4. Experimental validation of basal DUSP expression in mCFU‐E and BaF3‐EpoR. Quantitative RT–PCR was performed with all samples on the same plate, allowing a direct comparison of mRNA levels between mCFU‐E and BaF3‐EpoR cells. Data are normalized to the Rpl32 gene. Ratios of expression in BaF3‐EpoR cells compared with mCFU‐E cells are shown as box plots. Boxes indicate the interquartile range and whiskers extend to 1.5 × interquartile range. N = 10. For details, see Appendix K.

Source data are available online for this figure.
Figure 4
Figure 4. Validation of Epo‐induced signaling dynamics in 32D‐EpoR cells under RSK wild‐type conditions and upon overexpression of RSK

  1. Model prediction and experimental validation for pAKT, ppERK, and pS6 dynamics of 32D‐EpoR cells in response to 50 U/ml Epo. Simulations are based on measured, cell type‐specific protein abundance of 32D‐EpoR cells and global kinetic rates estimated from mCFU‐E and BaF3‐EpoR cells. Model predictions are represented by solid lines. Experimental validation data obtained by quantitative immunoblotting are represented by filled circles. Error bars represent standard deviation estimated by an error model, N = 1.

  2. Sensitivity analysis for integrated pS6 in mCFU‐E, 32D‐EpoR, and BaF3‐EpoR cells. Measured, cell type‐specific protein abundance and estimated global kinetic rates were taken into account. Proteins were grouped according to the network modules. Integrated pS6 exhibited high sensitivity toward RSK in 32D‐EpoR and BaF3‐EpoR cells but not in mCFU‐E cells.

  3. Model prediction and experimental validation for RSK, AKT, ERK, and S6 activation upon RSK overexpression (oe) and 5 U/ml Epo stimulation in 32D‐EpoR cells. Simulations are based on measured, cell type‐specific protein abundance of 32D‐EpoR cells and global kinetic rates estimated from mCFU‐E and BaF3‐EpoR cells. Model predictions are represented by solid lines. Experimental validation data obtained by quantitative immunoblotting are represented by filled circles. Error bars represent standard deviation estimated by an error model. Integrated pS6 was significantly higher upon RSK overexpression as compared to 32D‐EpoR wild‐type (wt) cells, N = 4 (lower right panel). Two‐sample t‐test, ***P < 0.005.

  4. Impact of AKT, Ras, or PTEN overexpression on Epo‐induced pS6 dynamics in mCFU‐E and BaF3‐EpoR cells. Quantitative immunoblotting upon overexpression of PTEN, AKT, a constitutively active Ras protein, or the empty vector control in mCFU‐E cells or BaF3‐EpoR cells. Growth factor‐deprived mCFU‐E cells (5 × 106 cells per condition) and BaF3‐EpoR cells (1 × 107 cells per condition) were stimulated with 5 U/ml Epo for indicated time points. Cellular lysates were analyzed by immunoblotting employing sequential reprobing anti‐pAKT, anti‐ppERK, anti‐pS6, anti‐S6, and to ensure equal loading with anti‐β‐actin antibodies. Detection was performed with chemiluminescence using a CCD camera device (ImageQuant). Quantification of pS6 on the right is depicted as fold change to wild‐type samples at 30 min after Epo stimulation. Error bars represent standard deviation. oe: overexpression. N = 3. Welch modified two‐sample t‐test, n.s. = not significant, *P < 0.05.

Source data are available online for this figure.
Figure 5
Figure 5. Evaluation of the effects of AKT inhibitor AKT VIII and MEK inhibitor U0126 on signaling and cell cycle

  1. Epo‐induced signaling upon AKT or MEK inhibitor treatment. Growth factor‐deprived cells were pretreated for half an hour with AKT VIII or U0126, respectively, and subsequently stimulated with 5 U/ml Epo. PDI served as loading control.

  2. Model calibration with integrated pAKT, ppERK, and pS6 data upon inhibitor treatment and 5 U/ml Epo stimulation of mCFU‐E, 32D‐EpoR, and BaF3‐EpoR cells. Area under curve from time‐resolved, quantitative immunoblotting data was calculated and is represented by filled circles. N = 3. Error bars represent standard deviation estimated by an error model. Solid lines represent model trajectories.

  3. Analysis of cell cycle indicator genes upon AKT VIII and U0126 treatment. Growth factor‐deprived cells were pretreated for half an hour with indicated doses of a single inhibitor, followed by stimulation with 5 U/ml Epo for 0 and 3 h. The expression of cyclinD2, cyclinG2, and p27 was measured by quantitative RT–PCR and normalized to the Rpl32 gene. Genes were selected based on microarray analysis. Experimental data are shown as fold change to unstimulated cells with mean ± standard deviation, N = 3. Welch modified two‐sample t‐test, n.s. not significant, *P < 0.05, **P < 0.01, ***P < 0.005.

Source data are available online for this figure.
Figure 6
Figure 6. Linking cell cycle and integrated pS6 to proliferation

  1. Scheme of the mathematical model. Integrated pAKT and ppERK obtained from our kinetic model were linked by linear regression analysis to measured cell cycle indicator. Similarly, cell cycle indicator and integrated pS6 were linked to measured proliferation by linear regression analysis and model selection.

  2. Linear regression fit for the respective cell cycle indicator of experimental data. Data points represent mean ± standard deviation, N = 3. Solid lines represent the linear regression fit.

  3. Simulated integrated pS6 response versus simulated cell cycle indicator of mCFU‐E, BaF3‐EpoR, and 32D‐EpoR cells upon 5 U/ml Epo stimulation and single inhibitor treatment at indicated doses.

  4. Linear regression modeling and model selection revealed that while proliferation of mCFU‐E cells is best described by integrated pS6 only, for 32D‐EpoR and BaF3‐EpoR cells proliferation correlates mainly with cell cycle indicator. Proliferation was measured, and integrated pS6 and cell cycle indicator were simulated with our mathematical model. The linear regression model selection is based on Akaike's information criterion. The respective correlation is given. For details, see Appendix N and O.

Source data are available online for this figure.
Figure 7
Figure 7. Prediction of cell type‐specific proliferation

  1. Upper panel: Model prediction of Epo‐dependent proliferation in mCFU‐E and BaF3‐EpoR cells overexpressing SHIP1 or PTEN. Solid lines represent model trajectories. Lower panel: Experimental validation of PTEN and SHIP1 overexpression effects on Epo‐dependent proliferation. Proliferation was assessed using [3H]‐thymidine incorporation 14 h (mCFU‐E) or 38 h (BaF3‐EpoR) after retroviral transduction with PTEN or SHIP1 construct for overexpression. Data are represented as mean ± standard deviation, N = 3. EC50 values are given.

  2. Upper panel: Model prediction of proliferation upon combined inhibitor treatment and 5 U/ml Epo stimulation. Maximum proliferation was scaled to 1. Lower panel: Experimental validation of proliferation upon combined inhibitor treatment and 5 U/ml Epo stimulation. Proliferation was measured as cell numbers after 14 h (mCFU‐E) or 38 h (BaF3‐EpoR, 32D‐EpoR) with Coulter Counter. Maximum was scaled to 1.

Source data are available online for this figure.
Figure 8
Figure 8. Predicting combined effects of inhibitor treatment on Epo‐induced proliferation of hCFU‐E cells solely based on protein abundance
Human CFU‐E cells were prepared from CD34+ cells mobilized into the peripheral blood of three independent stem cell donors.

  1. Venn diagram representing overlap of hCFU‐E proteome from three independent healthy donors. In total, 6,925 proteins were quantified.

  2. Size determinations of hCFU‐E compared to mCFU‐E cells. Cell diameter was measured by imaging flow cytometry. Cytoplasm was stained with calcein, and nuclei were stained with DRAQ5. Probability density function of size distribution with indicated mean diameter of hCFU‐E and mCFU‐E cells. All cells were growth factor‐deprived and unstimulated.

  3. Cytoplasmic volumes of BaF3‐EpoR, 32D‐EpoR, mCFU‐E, and the hCFU‐E cells. All volumes were determined using imaging flow cytometry as described in (B).

  4. Correlation between protein molecules per cell for key signaling components determined by quantitative immunoblotting or quantitative mass spectrometry and the “proteomic ruler” method (Wiśniewski et al, 2014) in BaF3‐EpoR and mCFU‐E cells. Diagonal line as guide to the eye. Spearman's rank‐based coefficient of correlation ρ = 0.88.

  5. Protein abundance of key signaling components was determined by mass spectrometry of whole cell lysates of hCFU‐E cells. Copy numbers of proteins per cell were obtained by the “proteomic ruler” method and converted to cytoplasmic concentrations with the volumes calculated from (B) and shown in (C).

  6. Model prediction and experimental validation of proliferation upon single or combined inhibitor treatment with AKT VIII and U0126 upon 5 U/ml Epo stimulation. Maximum proliferation was scaled to 1. Proliferation was measured as cell numbers, counted with a hemocytometer after 96 h by trypan blue exclusion assay. Solid lines represent model trajectories. Experimental data are represented as mean ± standard deviation. N = 3.

  7. Comparison of the effect of single or combinatorial inhibitor treatment on Epo‐induced proliferation of murine and human CFU‐E cells. Proliferation was measured upon single or combined inhibitor treatment with AKT VIII and U0126 upon 5 U/ml Epo stimulation. Numbers of mCFU‐E cells were determined with the Coulter Counter after 14 h, whereas numbers of hCFU‐E cells were determined with a hemocytometer after 96 h by the trypan blue exclusion assay. Maximum proliferation was scaled to 1. Data are represented as mean (N = 3), and error bars indicate standard deviation. Data of mCFU‐E cells are presented in Figs 6D and 7B, and data of hCFU‐E cells are taken from Fig 8F. Tukey multiple comparison of means, n.s. = not significant, *P < 0.05.

Source data are available online for this figure.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. van den Akker E, van Dijk T, Parren‐van Amelsvoort M, Grossmann KS, Schaeper U, Toney‐Earley K, Waltz SE, Löwenberg B, von Lindern M (2004) Tyrosine kinase receptor RON functions downstream of the erythropoietin receptor to induce expansion of erythroid progenitors. Blood 103: 4457–4465 - PubMed
    1. Alessi DR, Andjelkovic M, Caudwell B, Cron P, Morrice N, Cohen P, Hemmings BA (1996) Mechanism of activation of protein kinase B by insulin and IGF‐1. EMBO J 15: 6541–6551 - PMC - PubMed
    1. Alfieri R, Barberis M, Chiaradonna F, Gaglio D, Milanesi L, Vanoni M, Klipp E, Alberghina L (2009) Towards a systems biology approach to mammalian cell cycle: modeling the entrance into S phase of quiescent fibroblasts after serum stimulation. BMC Bioinformatics 10(Suppl. 1): S16 - PMC - PubMed
    1. AlQuraishi M, Koytiger G, Jenney A, MacBeath G, Sorger PK (2014) A multiscale statistical mechanical framework integrates biophysical and genomic data to assemble cancer networks. Nat Genet 46: 1363–1371 - PMC - PubMed
    1. Bachmann J, Raue A, Schilling M, Böhm ME, Kreutz C, Kaschek D, Busch H, Gretz N, Lehmann WD, Timmer J, Klingmüller U (2011) Division of labor by dual feedback regulators controls JAK2/STAT5 signaling over broad ligand range. Mol Syst Biol 7: 516 - PMC - PubMed

Publication types

Substances