Naïve T cells continuously exit the circulation to crawl though lymph nodes (LNs), scanning antigen-presenting cells (APCs), primarily dendritic cells (DCs), for peptides presented in the groove of MHC molecules (pMHC complexes). Several hours of TCR stimulation are required for efficient T cell activation and commitment to proliferation 
. It is not clear, however, whether the hours of long-term contacts between T cells and DCs need necessarily be continuous. Recent in vitro studies have suggested that T cells are equipped with mechanisms that allow them to integrate, in an additive manner, the signals received during periods of interrupted stimulation 
. Using proliferation as a readout, one study demonstrated that naïve T cells can sum up signals delivered during two 7 hour-long periods of stimulation separated by a rest period, even one lasting nearly a day 
. In a complementary study, the production of IFN-γ by T cell clones was found to be proportional to the duration of the TCR stimulation even when the intervals of stimulation were alternated with intervals of rest 
. The ability of T cells to thus recall short, suboptimal stimulations has been termed “biochemical T cell memory” to distinguish it from classical T cell memory 
Whether this mode of activation also operates in vivo is an open question, but studies using multi-photon laser microscopy have provided clues that it might be the case 
. These studies, involving real-time observation of interactions of T cells with DCs within explanted LNs or in the LNs of anesthetized mice, have revealed that T cells often have the opportunity to engage in successive and diverse interactions with Ag-bearing DCs. In some settings, early during antigen exposure the interactions may be predominantly transient, lasting several minutes, while later on during priming they progress to become stable contacts lasting several hours 
. Activated T cells can also re-interact with Ag-bearing DCs in the late phase of priming 
. It remains to be firmly established whether signals received at these different stages are additive. In this respect, we have previously shown that T cells integrate signals received during late interactions with DCs to sustain CD25 expression and become IFN-γ producers 
. Furthermore, T cells stimulated in vitro for 24 h respond more rapidly upon adoptive transfer than their naïve counterparts, suggesting that their initial activation program was ‘memorized’ 
. However, it remains unclear whether T cells integrate short, suboptimal signals during serial contacts in vivo, in particular prior to commitment to proliferation.
Another important question relates to the molecular basis underlying such biochemical memory. A number of non-mutually exclusive mechanisms could potentially endow the T cell with this ability. While most proximal signaling intermediates are likely to be lost within seconds of cessation of TCR engagement, it has recently been shown that the feedback regulation of SOS (Son of Sevenless) results in hysteresis in Ras activation, a mechanism that conferred lymphocytes primed by a first stimulation with the ability to respond to weak stimulation even 5–10 min after removal of the initial stimulus 
. Early epigenetic modifications are also likely to participate in such T cell biochemical memory 
. Recent computational modeling has suggested that accumulation of transcription factors regulated by a positive feedback loop might help T cells memorize past activation signals 
. One such transcription factor is AP-1, which plays an important role in the T cell activation program and cell cycle progression 
. C-fos, a component of this dimeric transcription factor, is an immediate early gene (IEG) product rapidly induced upon T cell activation 
. Whether c-fos could act as a molecular counter of intermittent TCR stimulation is an intriguing possibility.
Here we provide new evidence that T cells have the ability to sum up interrupted suboptimal stimulations both in vitro and in vivo. In addition, we show that the total c-fos pool remains elevated in the nucleus of activated T cells for a few hours following signal withdrawal, while a key phosphorylated form remains stable for at least one day. Finally, we provide evidence that c-fos increases incrementally upon serial in vivo antigenic stimulations. We propose that phosphorylated c-fos may serve as a biochemical fingerprint of previous suboptimal stimulation, leaving the T cell poised to rapidly resume its activation program upon its next encounter with an antigen-bearing DC.