We monitored ERK activity in individual primary human neonatal keratinocytes (NHKs) via lentiviral expression of an ERK reporter, EKAREV-nls. NHKs were cultured on a 3T3 feeder layer under conditions that support clonal growth of stem cells and differentiation of their progeny, as occurs in skin (Fig. 1a). Within colonies, mean ERK activity was lower in large than small keratinocytes (Fig. 1b) and there was a wider range of mean ERK activity in smaller cells (tails and outliers in Fig.1b). This suggests that the patterns of ERK activity differ in stem cells and differentiated cells, since keratinocytes are known to enlarge as they undergo terminal differentiation.
The temporal patterns of ERK activity varied considerably between cells. Some maintained stable-high or stable-low activity, while others exhibited pulsating ERK activity (Fig. 1c, d). By plotting ERK activity as a function of time, we observed cells with stable-low or stable-high ERK activity (characterized by a low variance) (Fig. 1c, d). We also observed cells with pulsating ERK activity (characterized by a high variance). The pulses were not uniform between cells but differed in amplitude, frequency and duration (Fig. 1c, d). We ruled out the possibility that the pulses were an imaging artifact by using a kinase-dead control FRET biosensor, EKAREV-TA-nls (Extended Data Fig. 1d).
We generated a fluorescent reporter of the differentiation marker, Involucrin, which is expressed by all keratinocytes undergoing terminal differentiation in culture. We used the previously characterised Involucrin promoter to drive expression of mCherry (Fig. 1e). Cells were cultured in low Ca2+ keratinocyte serum-free medium (KSFM), which does not support full terminal differentiation, and then transferred to medium containing high Ca2+ (1.6 mM) or serum. Under those conditions, involucrin-positive cells are known to accumulate and there was a significant increase in mCherry fluorescence (Fig. 1f, g).
Simultaneous monitoring of ERK activity and Involucrin expression in individual cells revealed that ERK pulses are gradually lost, coinciding with the onset of involucrin expression (Fig. 1h). To gain an integrative view of the temporal relationship between ERK activity pulses and Involucrin expression, we computed the variance over a moving time window of ERK activity and a moving mean on the Involucrin time series. This allowed us to build trajectories in the ERK variance/mean Involucrin phase diagram, in order to discover how these qualities relate to each other as a function of time. For this, we incorporated the observations from 780 cells (Fig. 1i, j). The vectors point forward in time, showing how ERK variance and mean Involucrin expression coevolve as cells differentiate. This reveals that ERK activity pulses (measured by the variance) are drastically decreased prior to Involucrin expression, reaching a stable status of low variance-high Involucrin (red square in Fig. 1i). Conversely, the mean ERK activity gradually converges towards a low mean value, as the level of Involucrin increases (see arrows in Fig. 1j). This differential behavior indicates that ERK pulses and mean level are subject to distinct regulation, and that a dramatic decrease in pulsatile behaviour precedes differentiation (Fig. 1k).
