SHP2 decreases IL-6-induced STAT3 activation
Cellular heterogeneity e.g. in protein expression and activation is a central feature of multicellular organisms. We have recently shown that the negative feedback inhibitor SOCS3 increases robustness of late IL-6-induced STAT3 activation against differential STAT3 protein expression in individual cells. In contrast, SOCS3 does not affect robustness of early IL-6-induced STAT3 activation or basal cytokine-independent STAT3 activation. Additionally, STAT3-Y705 phosphorylation is robust when cytokine doses are low. In accordance, STAT3-S727 phosphorylation, which results in a reduction of STAT3-Y705 phosphorylation, increases robustness of IL-6-induced STAT3 activation [20]. These observations led us to the overarching hypothesis that negative regulation increases robustness of IL-6-induced STAT3-Y705 phosphorylation against varying STAT3 expression in individual cells. However, the influence of other negative regulatory proteins such as phosphatases on IL-6 signalling in heterogeneous cell populations has not been addressed so far. Consequently, we ask here whether and how the protein-tyrosine-phosphatase SHP2 controls robustness of IL-6-induced STAT3 activation.
We made use of murine embryonal fibroblasts (MEF) expressing a mutant SHP2 protein lacking 65 amino acids within the N-terminal SH2-domain (ΔEx3) [21]. This mutation prevents recruitment of SHP2 to the phosphorylated tyrosine motif 759 within gp130, thereby mimicking a SHP2 knock-out. Expression of SHP2 ΔEx3 enhances IL-6-induced activation of STAT3 and increased activation of STAT3-responsive promoter elements [13].
MEF cells, like most other cells, do not express transmembrane IL-6Rα. Hence, IL-6-induced STAT3 activation in MEF cells depends on trans-signalling enabled by IL-6:soluble IL-6 receptor α (sIL-6Rα) complexes. IL-6 and sIL-6Rα form noncovalent complexes whose equilibrium concentration is not trivially predictable. As a remedy for analyses of dose-dependent effects of IL-6 trans-signalling, Hyper-IL-6 (Hy-IL-6), a fusion protein of IL-6 and sIL-6Rα [30], is used.
In accordance with previously published results [13], Hy-IL-6-induced STAT3-Y705 phosphorylation is increased in MEF SHP2 ΔEx3 cells compared to MEF wildtype (wt) cells (Fig. 1a). Mutant cells reconstituted by stable expression of wt SHP (MEF SHP2 ΔEx3 + SHP2) do not show enhanced Hy-IL-6-induced STAT3 phosphorylation. This clearly supports that functional SHP2 is a negative regulator of IL-6-induced STAT3 activation.
Hy-IL-6-induced STAT3-Y705 phosphorylation is transient (Fig. 1b), which underlines the importance of negative inhibitors in shaping signalling and cellular answers in response to IL-6. SOCS3 is not expressed in the first 15 min but after 30 min of Hy-IL-6 stimulation, which coincides with reduced STAT3-Y705 phosphorylation. In contrast to SOCS3, SHP2 is expressed constitutively, while it is phosphorylated transiently in response to Hy-IL-6. Based on these results JAK/STAT signalling can be divided into a pre-stimulation phase, which is independent of IL-6, an early phase with strong STAT3-Y705 phosphorylation and a late phase with low steady state STAT3 activation and expression of SOCS3. Of note, SHP2 is expressed in all three phases.
SHP2 increases robustness of basal and early IL-6-induced STAT3 activation
Previous analyses revealed that mechanisms enabling robustness of IL-6-induced STAT3 phosphorylation operate in a time- and cytokine-dose-dependent manner [20]. We therefore also analysed the contribution of SHP2 to robustness of STAT3 activation in unstimulated cells and stimulated cells at early and late stages of signalling.
In a first step, we tested whether SHP2 contributes to robustness of basal and early IL-6-induced STAT3 activation. We applied a flow cytometric assay that allows for simultaneous analysis of STAT3 expression and STAT3-Y705 phosphorylation in single cells. MEF wt, MEF SHP2 ΔEx3 cells, and MEF SHP2 ΔEx3 + SHP2 cells were left untreated or were stimulated with increasing amounts of Hy-IL-6 for 15 min. Subsequently, cells were fixed and stained with differentially labelled antibodies against STAT3 and Y705-phosphorylated STAT3.
As shown earlier [20] single cell flow cytometry analyses of STAT3 reveal considerable differences in STAT3 expression within the MEF wt cell population, indicating that individual cells differ strongly in respect to STAT3 protein expression (Fig. 2a). The expression and distribution of STAT3 within the cell population are however independent from Hy-IL-6. A 15 min stimulation with Hy-IL-6 results in an increase in STAT3-Y705 phosphorylation (Fig. 2b). Median STAT3 phosphorylation increases dose-dependently up to 75 ng/ml Hy-IL-6. As seen for STAT3 expression STAT3-Y705 phosphorylation varies strongly within the cell population (Fig. 2b). Likewise, in MEF SHP2 ΔEx3 cells and MEF SHP2 ΔEx3 + SHP2 cells STAT3 expression (Fig. 2c, e) and dose-dependent Hy-IL-6-induced STAT3-Y705 phosphorylation (Fig. 2d, f) are heterogeneous.
We next asked to what extent an individual cell´s STAT3 amount affects STAT3-Y705 phosphorylation.
The multiplexed single cell analysis performed, enables us to correlate STAT3 expression and STAT3 activation in single cells. Scatter plots of STAT3 expression and STAT3-Y705 phosphorylation in either unstimulated MEF wt cells (Fig. 3a) or MEF wt cells treated for 15 min with a saturating dose of Hy-IL-6 (Fig. 3b) indicate a positive correlation between the amount of STAT3 and the strength of STAT3 phosphorylation. Of note, the positive correlation is much larger in cells treated with Hy-IL-6 (Pearson correlation = 0.718, p-value < 10–16) than in unstimulated cells (Pearson correlation = 0.464, p-value < 10–16). To formally challenge the difference between unstimulated cells and cells stimulated with a saturating dose of Hy-IL-6, we employed a two-sample Student’s t-test between Pearson’s correlation coefficients obtained from n = 3 biological replicates. Indeed, Hy-IL-6-induced STAT3 phosphorylation correlates statistically stronger with STAT3 expression than basal cytokine-independent STAT3 phosphorylation (Student’s t-test, β = 0.254, p-value = 0.006).
Calculation of linear regression to quantify the dependency between STAT3 expression and activation can fail to detect non-linear associations and thereby might prevent from discovering critical dynamical features of STAT3 activation. To make sure that all types of correlations and not only linear correlation are captured by our analysis, we use the information theoretic measure MI to quantify the dependency between STAT3 and STAT3-Y705 phosphorylation [20]. MI is known to be a sensitive method to detect different patterns of interdependence [31]. In principle, a lower MI means that two variables are less dependent on each other compared to variables with higher MI. Consequently, we interpret a low MI between STAT3 expression and STAT3-Y705 phosphorylation as high robustness of STAT3 activation to varying STAT3 expression, because the strength of STAT3 phosphorylation in this case is not dependent on the STAT3 amount in an individual cell (Fig. 3c).
In unstimulated MEF wt cells MI between STAT3 expression and STAT3-Y705 phosphorylation is low, indicating that the magnitude of basal STAT3 activation is indeed independent from the magnitude of STAT3 protein expression (Fig. 3d). In unstimulated MEF SHP2 ΔEx3 cells the MI between STAT3 expression and STAT3-Y705 phosphorylation is significantly increased compared to MEF wt cells, indicating that SHP2 increases robustness of basal STAT3 phosphorylation in the pre-stimulation phase. Consequently, reconstitution of mutant cells with wildtype SHP2 (SHP2 ΔEx3 + SHP2) restores robustness of STAT3 activation against varying STAT3 expression. This is in line with the hypothesis that regulatory mechanisms, that reduce STAT3 phosphorylation, increase robustness. In summary, SHP2 increases robustness of basal STAT3 phosphorylation against cell-to-cell heterogeneity in STAT3 expression.
Next, we addressed robustness of early IL-6-induced STAT3 activation against varying STAT3 protein copy number. Hy-IL-6-induced STAT3 phosphorylation in MEF wt cells is robust for low cytokine concentrations. With increasing Hy-IL-6 amounts resulting in stronger phosphorylation of STAT3 (Fig. 2b), MI is significantly increased compared to MI in unstimulated MEF wt cells (Fig. 3d, e, blue boxes and *) [20]. Also, in MEF SHP2 ΔEx3 cells (green boxes and *) and MEF SHP2 ΔEx3 + SHP2 cells (violet boxes and *) robustness of early STAT3 activation is reduced for high Hy-IL-6 concentrations compared to the corresponding unstimulated cells. This indicates that for low cytokine concentrations STAT3 phosphorylation is robust against heterogeneous STAT3 expression.
Furthermore, mutation of SHP2 (MEF SHP2 ΔEx3) significantly reduces robustness of STAT3 activation independently of Hy-IL-6 dose (Fig. 3e, compare blue and green) (Additional file 2: Table 1). Reconstitution with wt SHP2 (MEF SHP2 ΔEx3 + SHP2 cells) restores robustness (Fig. 3e, compare green and violet) (Additional file 2 Table 1). Thus, SHP2 not only affects robustness of cytokine-independent STAT3 activation but also robustness of early Hy-IL-6-induced STAT3 activation.
In summary, SHP2 contributes to robustness of basal, cytokine independent STAT3 activation and to robustness of early IL-6-induced STAT3 activation. This seems to complement the function of SOCS3, which controls robustness of late IL-6-induced STAT3 activation [20].
SHP2 does not affect robustness of late IL-6-induced STAT3 activation
We next tested whether SHP2 affects robustness of late IL-6-induced STAT3-Y705 phosphorylation. MEF wt (Fig. 4a, b), MEF SHP2 ΔEx3 (Fig. 4c, d), and MEF SHP2 ΔEx3 + SHP2 (Fig. 4e, f) cells were treated with increasing amounts of Hy-IL-6 for 90 min and STAT3 expression and activation were analysed by intracellular multiplex flow cytometry as described above. As seen for short stimulation periods STAT3 expression is heterogeneous in all three cell lines and not affected by Hy-IL-6 (Fig. 4a, c, e). Late Hy-IL-6-induced STAT3-phosphorylation is dose-dependent and heterogenous (Fig. 4b, d, f) but as also shown in Fig. 1B weaker than early STAT3 activation.
In contrast to early STAT3 activation (Fig. 3e) the robustness of late STAT3 phosphorylation is not affected by the concentration of Hy-IL-6 (Fig. 4g), supporting the hypothesis that late STAT3 phosphorylation, which is weaker than early STAT3 phosphorylation (Fig. 1b), is independent from STAT3 expression and thus robust. Interestingly, expression of SHP2 ΔEx3 does not significantly affect robustness of late Hy-IL-6-induced STAT3 activation in contrast to early signalling (Fig. 4g, Additional file 3: Table 2).
In summary, this raises a scenario in which SHP2 and SOCS3 enable robustness of STAT3 activation in a timed manner. SHP2 is expressed constitutively, which enables it to act on basal and early IL-6-induced STAT3 activation. In contrast SOCS3 is not expressed in early signalling (Fig. 1b). When SOCS3 is expressed it reduces STAT3 phosphorylation and increases robustness [20] while SHP2 no longer contributes to robustness.
Activation of MAPK does not affect robustness of JAK/STAT signalling
SHP2 reduces robustness of cytokine-independent and early IL-6-induced JAK/STAT signalling against differential STAT3 expression. SHP2 has a dual function in IL-6-induced signalling. While it reduces JAK/STAT signalling it is indispensable for IL-6-induced MAPK pathway activation [32, 33]. It is possible, that SHP2-dependent MAPK activation influences IL-6-induced STAT3 activation in heterogenous cell populations and consequently the robustness of STAT3 activation. Hence, we next asked whether activation of MAPK contributes to robustness of cytokine-independent or early IL-6-induced STAT3 activation. MEF wt cells were treated with the MEK inhibitor U0126 alone or pre-treated with U0126 before stimulation with Hy-IL-6. Activation of MAPK and JAK/STAT signalling was analysed by Western Blotting. IL-6-induced phosphorylation of ERK1/2 was efficiently blocked by U0126, while STAT3 activation was seemingly unaffected (Fig. 5a). Next, U0126 treated and control MEF wt cells were stimulated with increasing amounts of Hy-IL-6. STAT3 expression and phosphorylation were analysed by multiplex intracellular flow cytometry as described before (Fig. 5b). In support of Fig. 5a, the strength of STAT3 phosphorylation induced by both high and low amounts of Hy-IL-6 is independent of the inhibition of the MAPK pathway. Cytokine-independent STAT3 phosphorylation is also not affected by MAPK inhibition (Fig. 5b, Additional file 4: Table 3). Of note, also robustness, as measured by MI between STAT3 expression and phosphorylation, of cytokine-independent and IL-6-induced STAT3 activation against varying STAT3 expression is not affected by inhibition of MAPK (Fig. 5c, compare blue and orange, Additional file 5: Table 4). As shown earlier (Fig. 3e) MI increases significantly with increasing amounts of Hy-IL-6.
In summary, these observations contradict the hypothesis that SHP2-dependent MAPK activation increases robustness of IL-6-induced STAT3 activation in heterogeneous cell populations. SHP2 most probably directly increases robustness of STAT3 activation, independent of MAPK activation.
SHP2 increases Channel Capacity of IL-6-induced JAK/STAT signalling
Channel Capacity is an information theoretic measure for the maximal number of input values - referred here to cytokine concentrations - that can be discriminated by a receiver – referred here to STAT3-Y705 phosphorylation. The negative feedback inhibitor SOCS3 has opposing functions in regulating robustness of STAT3 activation and defining the amount of information transferred through IL-6-induced JAK/STAT signalling. While it increases robustness of STAT3-Y705 phosphorylation it reduces Channel Capacity of late JAK/STAT signalling [20]. We therefore addressed whether the phosphatase SHP2 also affects the amount of information transmitted through IL-6-induced JAK/STAT signalling. To do so we calculated Channel Capacity of early and late IL-6-induced STAT3 activation in MEF wt, MEF SHP2 ΔEx3, and MEF SHP2 ΔEx3 + SHP2 cells based on the data presented in Figs. 2 and 4. Channel Capacity of early Hy-IL-6-induced JAK/STAT signalling in MEF wt cells is approximately 0.7 bit (Fig. 6a, blue). When SHP2 is mutated (green) Channel Capacity is significantly reduced to 0.3 bit. This reduction is partly restored by expression of wt SHP2 (violet), which suggests that SHP2 in contrast to SOCS3 increases information transfer of early IL-6-induced JAK/STAT signalling.
As shown earlier [20] information transfer through the JAK/STAT pathway is strongly reduced at late timepoints, which reflects reduced activation of STAT3 at late time points. Mutation of SHP2 does not affect late Channel Capacity (Fig. 6a).
Interestingly, SHP2 and SOCS3 act opposingly on Channel Capacity. We hypothesize, that SHP2 increases information transfer because it reduces basal STAT3 phosphorylation in the pre-stimulation phase. As a consequence, it would extend the STAT3 phosphorylation range in which cells can operate by increasing their sensitivity to lower stimulation doses. To test this hypothesis, we analysed basal cytokine-independent STAT3 phosphorylation in MEF wt, MEF SHP2 ΔEx3, and MEF SHP2 ΔEx3 + SHP2 cells by intracellular flow cytometry. In line with our hypothesis, basal STAT3-Y705 phosphorylation is increased, when SHP2 is mutated (Fig. 6b).
To test whether SHP2-dependent MAPK affects Channel Capacity of IL-6-induced Jak/STAT signalling, MEF wt cells were pre-treated with the MEK inhibitor U0126 before stimulation with increasing amounts of Hy-IL-6 for 15 min. Expression and phosphorylation of STAT3 were analysed by multiplexed intracellular flow cytometry. Based on these data Channel Capacity of early JAK/STAT signalling was calculated (Fig. 6c, Additional file 7: Table 6) and compared to Channel Capacity of Hy-IL-6 treated MEF wt cells. Inhibition of MAPK does not affect Channel Capacity of Hy-IL-6-induced JAK/STAT signalling, indicating that the amount of information transferred through IL-6-induced JAK/STAT signalling is independent of SHP2-induced MAPK activation.
In summary, our data highlight new functions of the phosphatase SHP2 in ensuring robust STAT3 activity despite heterogeneous STAT3 expression. SHP2 increases robustness and information transfer of early IL-6-induced STAT3-Y705 phosphorylation and ensures independence of basal STAT3 phosphorylation from varying STAT3 expression. However, SHP2 does not affect robustness and information transfer of late IL-6-induced JAK/STAT signalling, indicating a timely orchestration of mechanisms that enable cells to cope with cellular heterogeneity. These effects are most probably independent of SHP2-induced MAPK activation and hence cross-talk of MAPK and JAK/STAT signalling.