The Lifeact-EGFP mouse is a translationally controlled fluorescent reporter of T cell activation

Cytotoxic CD8⁺ T lymphocytes (CTLs) mediate host protection against cancer and infection by patrolling tissues for malignant or pathogen-laden cells, which they can detect and rapidly eliminate (Gerlach et al., 2010). Central to the ability of CTLs to home to and migrate within tissues, and to engage and kill targets, is their capacity for rapid shape change, which relies on the tightly regulated remodelling of cortical networks of filamentous actin (F-actin) (Vicente-Manzanares and Sánchez-Madrid, 2004; Billadeau et al., 2007). Motile leukocytes typically display a polarised morphology characterised by actin-mediated protrusivity at the leading edge and actomyosin-dependent contractility at the rear (Weninger et al., 2014). Live cell microscopy and associated image analysis techniques, together with the development of fluorescent reporters, have enabled the study of the intricacies of leukocyte dynamics. A genetically encoded reporter based on an F-actin binding peptide fused to enhanced green fluorescent protein (Lifeact-EGFP) (Riedl et al., 2008) and its derivative transgenic reporter mouse (Riedl et al., 2010) opened the door to live imaging of cellular F-actin dynamics. Applied to T cells, it led to our understanding that during cytotoxic synapse (CS) formation with target cells, F-actin is cleared from the centre of the CS, through which the delivery of cytotoxic granules is then enabled (Ritter et al., 2015). The tracking of T cell movements and interactions, synapse formation and killing of targets (Mullins and Hansen, 2013) has led to a better understanding of the processes preceding and following T cell activation (Zhao et al., 2012). Upon cognate antigen recognition on the surface of a target cell, T cell receptor (TCR) triggering initiates a proximal signalling cascade via the recruitment of the Src family kinases Lck and Fyn (Chakraborty and Weiss, 2014), resulting in the phosphorylation of the immunoreceptor tyrosine activation motifs (ITAMs) in the cytoplasmic domains of the TCR-CD3 complex (Love and Hayes, 2010). The recruitment of Zap-70 to the ITAMs then induces a series of further phosphorylation events that drive surface receptor and integrin clustering, as well as the membrane and cytoskeletal remodelling that underpin CS formation (Kumari et al., 2014). Subsequent distal TCR signalling, mediated via phosphorylation and nuclear translocation of extracellular signal-regulated kinases (Erk1/2; also known as MAPK3 and MAPK1, respectively), induces a ‘late phase’ of T cell activation characterised by sustained secretion of pro-inflammatory cytokines such as interleukin 2 (IL-2), IL-22 and interferon gamma (IFN-γ) (Yeh et al., 2008). Proximal TCR signalling and early phase T cell activation is completed within the first 15 min of antigen encounter (Varma et al., 2006), whereas late phase activation takes place over several hours (2-6 h) and governs the long-term immune response of effector T cells (Koike et al., 2003).

In addition to phosphorylation-mediated signalling cascades, such as those downstream of TCR triggering, evidence suggests that important facets of leukocyte activation are specifically subject to translational control, which accelerates the delivery of effector functions. For instance, the rapid and targeted translation of cytokines in primed T cells following antigen re-encounter restricts potent effector cytokine production to intended target sites (Piccirillo et al., 2014). Moreover, natural killer (NK) cells acquire cytotoxic functions by translating pre-existing pools of perforin and granzyme B mRNA transcripts (Fehniger et al., 2007). Translational control has recently been shown to be important for timely induction and resolution of inflammation through translatome analyses of macrophage (Schott et al., 2014), naïve T cell (Tan et al., 2017) or effector T cell maturation and activation (Araki et al., 2017), which all reveal complex specificities beyond global protein synthesis in how cellular programs are induced by translational discrimination. For instance, naïve CD8⁺ T cells prioritise translation of transcripts required for ribosome biogenesis to ensure ribosome sufficiency (Tan et al., 2017), whilst CTLs engage both translational upregulation and suppression mechanisms as they progress from the clonal expansion phase to contraction phase (Araki et al., 2017). Distinct subsets of CD4⁺ T regulatory cells can also be distinguished by their translatomes (Bjur et al., 2013).

Here, we reveal that, following activation, CD4⁺ and CD8⁺ T cells isolated from the widely adopted Lifeact-EGFP reporter mouse exhibit a marked increase in fluorescence (hereafter also simply ‘GFP’) that is readily detected by flow cytometry and fluorescence microscopy. The increased fluorescence correlates with interferon gamma (IFN-γ) and tumor necrosis factor (TNF-α; now known as TNF) production and late phase T cell activation, and is translationally controlled via the mTOR pathway. Our results indicate that the Lifeact-EGFP mouse is a valuable simultaneous reporter for F-actin dynamics, T cell activation and ensuing translational activity, allowing researchers to identify and track comprehensive T cell effector activity in situ.

To study the migratory search characteristics and killing kinetics of CTLs, we crossed Lifeact-EGFP mice to OT-I TCR transgenic mice (Hogquist et al., 1994) and co-embedded effector CD8⁺ T cells isolated from these mice (hereafter LGO1) in 3-dimensional (3D) collagen matrices with either cognate or non-cognate EL-4 tumour cells for live imaging, as previously described (Galeano Niño et al., 2016). The addition of propidium iodide (PI) to the bathing medium allowed for the identification of lysed target cells (Jenkins et al., 2015). Combined, these methods constitute a quantitative real-time ‘search-and-kill’ assay, which revealed that CTLs interacting with cognate tumour cells increased in GFP fluorescence over time, but not in a non-cognate context (Fig. 1A,B; Movie 1). Statistical divergence analysis (see Materials and Methods) showed that the gain in GFP signal in T cell populations co-embedded with cognate tumour cells compared with those interacting with non-cognate cells became significant at a population level on an average of 3.7±1.1 h after co-embedding (Fig. 1C; Fig. S1A). The gain in GFP fluorescence was accompanied by increased migratory arrest of the T cell population (Fig. 1D) and specific clearance of cognate tumour targets (Fig. 1E, Fig. S1B, see Materials and Methods).

The increase in GFP fluorescence in LGO1 T cells following clearance of cognate tumour was also confirmed by flow cytometry (Fig. 1F, Fig. S1C), which further revealed increased granularity (side scatter area, SSC-A) and cell size (forward scatter area, FSC-A), indicative of T cell activation (Fig. S1D) (Malik et al., 2009; Pollizzi et al., 2015). No increase in GFP, SSC-A or FSC-A parameters was observed upon co-incubation with non-cognate targets (Fig. S1C,D). LGO1 T cells activated by prior co-incubation with cognate tumour cells maintained their enhanced GFP fluorescence (Fig. 1G) and a state of migratory arrest (Fig. S1E) for at least 14 h after they were sorted and re-embedded in matrices in the absence of tumour cells.

We next monitored the GFP fluorescence of LGO1 T cells during differentiation. Naïve CD8⁺ LGO1 T cells experienced a pronounced (almost 20-fold) increase in GFP fluorescence upon stimulation, reaching a peak 48 h after primary activation by cognate peptide (Fig. 1H). The fluorescence signal then decayed over the following days but was nevertheless maintained at levels 5-fold above that of naïve T cells. At day 6, when re-stimulated with cognate tumour cells, effector LGO1 T cells again increased in GFP fluorescence (∼3-fold from the elevated effector level), which was sustained for a further 48 h (Fig. 1H).

To determine whether the gain in Lifeact-EGFP fluorescence upon activation is a result of proximal or distal TCR signalling, we employed a pan-Src family kinase (Hanke et al., 1996) and an Lck inhibitor (Seo et al., 2017), or an inhibitor of Mek1/2 (also known as MAP2K1/2) (Favata et al., 1998), the immediate upstream kinase of Erk1/2, respectively. Proximal TCR signalling was assessed by measurement of Zap-70 phosphorylation (Wang et al., 2010), which was abrogated by inhibition of Src family kinase and Lck, but not Erk1/2 (Fig. S2A). Distal TCR signalling was assessed by measurement of Erk1/2 phosphorylation, which was abrogated by Src family kinase, Lck and Erk1/2 inhibition (Fig. S2A). Therefore, the Mek1/2 inhibitor affords a specific inhibition of the distal TCR signalling pathway without compromising the proximal component, whereas Lck inhibition results in the abrogation of both.

Within the 3D search-and-kill assay, both the increase in Lifeact-EGFP signal and target clearance were abolished by Lck and Mek1/2 inhibitors (Fig. 2A,B; Fig. S2B; Movie 2), suggesting that these functions are triggered downstream of Erk1/2 phosphorylation during distal TCR signalling, consistent with the observed average lag time of 3.7 h prior to GFP increase. Interestingly, although target clearance is initiated without a lag (Fig. 2B), CTLs stop migrating under Mek1/2 inhibition to an even greater degree than control cells (Fig. S2C), suggesting that cognate contacts are still formed with targets when distal TCR signalling is inhibited. It is unclear if increased migratory arrest was due to direct inhibition of motility and adhesion, or to inhibition of cytotoxic function, leading to prolonged interactions as a result of failed disengagement (Jenkins et al., 2015), or both, resulting in reduced overall serial killing.

In an in vitro cytotoxicity assay where CTLs and target cells were co-pelleted at a 1:1 ratio, the inhibition of proximal TCR signalling abolished the gain in Lifeact-EGFP fluorescence (Fig. 2C) and cytotoxicity (Fig. 2D). However, whilst inhibition of distal TCR signalling significantly impaired these functions, complete abrogation was not observed. Multiple signalling cascades are activated upon TCR engagement, whereby distinct signalosomes and second messengers allow diversification of the proximal TCR signals, enabling T cells to calibrate their activation thresholds and to tune functional and proliferative responses (Gorentla and Zhong, 2012). The differential dependence on Erk1/2 signalling in the two assays could be due to higher effective antigen doses being received per CTL under static conditions, where multiple and sustained contacts are made with each target, compared with when each CTL migrates and interacts with only one or very few targets before effecting a kill in 3D collagen matrices. Nonetheless, these results highlight that the search-and-kill assay can be used to reveal distinct signalling requirements of the discrete cellular events, including migration, polarisation, scanning, migratory arrest and killing, which all contribute to overall efficacy of tumour target clearance.

To determine if the fluorescence intensity of Lifeact-EGFP T cells directly correlates with, and can therefore be used as a quantitative proxy of, the extent of T cell activation, we measured the production of the effector cytokines TNF-α and IFN-γ by activated T cells, which are widely adopted gold standard measures of T cell activation (Brown et al., 2017). In the in vitro cytotoxicity assay, IFN-γ and TNF-α secretion followed the same trend as GFP fluorescence gain and cytotoxicity upon proximal and distal TCR inhibition (Fig. 2E). On a per-cell basis, intracellular cytokine staining revealed that intracellular IFN-γ and TNF-α levels were directly correlated with Lifeact-EGFP fluorescence intensity (Fig. 2F; Fig. S2D,E). We also sorted resting and activated LGO1 CTLs into three subpopulations (high, medium and low) based on their GFP intensity (see Materials and Methods and Fig. S2F) and measured the secretion of IFN-γ and TNF-α from each subpopulation. The GFP fluorescence intensity of each subpopulation of activated LGO1 cells correlated with their levels of secretion of both cytokines (Fig. 2G), with an activated but unsorted population yielding intermediate levels of cytokine secretion (Fig. 2G). By contrast, increases in cell size, as determined by the FSC-A parameter in flow cytometry (Malik et al., 2009; Pollizzi et al., 2015; Jenkins et al., 2015), was indistinguishable between the three subpopulations (Fig. S2F), indicating that the magnitude of the gain in Lifeact-EGFP fluorescence cannot be accounted for simply by an increase in T cell size upon activation (Fig. S2F). Thus, the gain in Lifeact-EGFP fluorescence upon activation: (1) is governed by distal TCR signalling; (2) strongly correlates with IFN-γ and TNF-α production; and (3) serves as a better discriminator of the level of activation in T cell subpopulations within an activated cohort than cell size.

The activation-induced increase in fluorescence is not observable in effector T cells isolated from fluorescent reporters other than Lifeact-EGFP. Neither tdTomato nor GFP intensities, from the TCR-transgenic mG/mT×OT-I and UBC-GFP×gBT-I reporter mice, respectively (see Materials and Methods), increased significantly in CTLs activated by cognate tumour targets compared with their resting counterparts (Fig. 3A). Therefore, overall increase in protein synthesis upon T cell activation cannot solely account for the gain in fluorescence in activated LGO1 T cells. We then generated polyclonal CD4⁺ and CD8⁺ effector T cells from non-TCR transgenic Lifeact-EGFP and Lifeact-mRFPruby (hereafter Lifeact-RFP) mice, and stimulated these with anti-CD3/anti-CD28-coated beads. Both CD4⁺ and CD8⁺ T cells from the Lifeact-EGFP mice displayed a marked increase in fluorescence upon activation, whereas the gain in Lifeact-RFP was far less pronounced (Fig. 3B), despite comparable levels of TCR-mediated activation indicated by the magnitude of TCR-Vα downregulation (Fig. S3A). The stronger activation signal conferred by beads compared with cellular targets likely accounts for the slight increase in fluorescence in the Lifeact-RFP T cells, which remains far lower than in Lifeact-EGFP T cells.

We then investigated whether the increase in Lifeact-EGFP fluorescence could be due to a greater abundance of either total or filamentous actin. Activated T cells have previously been reported to globally assemble more F-actin than resting cells when evaluated by flow cytometry (Kim et al., 2009). We measured no significant increase in either Phalloidin or total β-actin staining in activated T cells compared with resting T cells (Fig. S3B), nor did we detect increased mRNA levels for β- or γ-actin (Fig. S3C). Upon complete depolymerisation of the F-actin network following latrunculin A (LatA) treatment, activated LGO1 T cells retained their enhanced fluorescence, whereas Phalloidin staining was fully abolished (Fig. S3D). We also employed fluorescence lifetime imaging microscopy (FLIM) to determine whether the gain in Lifeact-EGFP fluorescence could be due to a change in the photo-physical characteristics of the fluorophore. FLIM is suitable for detecting dynamic (collisional, FRET) quenching mechanisms, with shortened fluorescence lifetimes indicative of more efficient quenching. In the case of Lifeact-EGFP, FLIM additionally allows the monitoring of changes in the fluorophore microenvironment as the lifetime of GFP depends on the local refractive index, which is related to intracellular crowding (van Manen et al., 2008). By means of a 2D phasor plot of the raw FLIM data (see Materials and Methods), we did not observe any differences in the fluorescence emission decays of resting or activated LGO1 T cells (Fig. 3C). In addition, LatA-mediated disruption of the actin cytoskeleton did not induce changes in the fluorescence lifetime of Lifeact-EGFP (Fig. 3C). These results indicate that the gain in Lifeact-EGFP fluorescence upon activation is independent of the subcellular distribution of F-actin or binding of the reporter thereto, and is not due to quenching or crowding of the fluorophore itself.

By contrast, the GFP signal corresponded directly to an increase in the amount of Lifeact-EGFP protein, as determined by intracellular staining (Fig. 3D) and western blotting (Fig. S3E) using an anti-GFP antibody. Strikingly, we detected no significant increase in eGFP mRNA levels in activated LGO1 CTLs, which expressed high levels of TNF-α and IFN-γ mRNA (Nicolet et al., 2017) (Fig. 3E), suggesting that the gain in Lifeact-EGFP is regulated post-transcriptionally. Post-transcriptional regulation of Lifeact-EGFP is particularly pronounced in naïve T cells, where the marked increase (18.2±1.84 fold change) (Fig. 1H) in GFP signal upon activation was not accompanied by significant increase in mRNA levels (2.51±1.54 fold change) (Fig. 3E). The increase in GFP signal was unaffected by MG-132, a proteasome inhibitor, but was abrogated by cycloheximide (Fig. S3F), further indicating that the increase in GFP signal is not due to its resistance to degradation but is rather a direct consequence of increased Lifeact-EGFP protein synthesis.

To determine the mechanism of translational regulation, we assessed the proportions of ribosomes engaged in polysomes (Fig. 3F,G) and the relative amounts of translationally active polysome-associated transcripts (Fig. S3G). In resting T cells, about 50% of ribosomes are engaged in polysomes; this fraction was significantly reduced after INK128 treatment (Fig. 3G), an inhibitor of the mammalian target of rapamycin (mTOR) kinase, which is consistent with the known effect of mTOR inhibitors to interfere with the recruitment of ribosomes onto mRNAs, i.e. translation initiation. In activated T cells, a lower proportion of ribosomes was accounted for by polysomes, which might reflect faster elongation (whereby ribosomes run off mRNAs more quickly, leading to a lower steady state proportion in polysomes), or lower rates of initiation, or a combination of both (Fig. 3G). Interestingly, INK128 did not affect the proportion of ribosomes engaged in polysomes in activated T cells, suggesting it has no net effect on translation initiation or elongation (Fig. 3G). The majority of eGFP, IFNγ and β2-microglobulin (β2M) mRNA transcripts are associated with polysomes in both resting and activated T cells (Fig. S3G). INK128 treatment induced a specific redistribution of eGFP and IFNγ, but not β2M, mRNA into smaller polysomes (Fig. S3G), consistent with specific impairment of initiation of translation of the eGFP and IFNγ mRNAs. By contrast, in activated T cells, INK128 treatment induced a shift of all three mRNAs from smaller into larger polysomes (Fig. S3G). Taken together with the observations that Lifeact-EGFP protein levels are higher in activated cells and are markedly decreased by INK128, and the observation that its mRNA is found in larger polysomes upon INK128 treatment, our results suggest that INK128 impairs the rate of translation elongation, thereby reducing the synthesis of Lifeact-EGFP in activated T cells. Given that the other two mRNAs behave similarly, this appears to be a general, rather than a transcript-specific, effect. An alternative explanation for the effect of INK128 to promote polysomal association of mRNAs in activated cells would be that it activates translation initiation; however, there is no known mechanism by which inhibition of mTOR signalling can activate initiation (Proud, 2019).

As activation-induced synthesis of GFP is unique to the Lifeact-EGFP reporter mouse, we sought to identify the site(s) of Lifeact-EGFP transgene insertion by whole genome sequencing (see Materials and Methods). Results indicate that the mouse is heterozygous for the transgene; with a single insertion site comprising a single copy of the Lifeact-EGFP transgene identified on chromosome 7, where a 44 kb deletion including the BAG family molecular chaperone regulator 3 (BAG3) gene occurred (Fig. 3H). This is consistent with the lack of homozygous Lifeact-EGFP littermates in our colony, since BAG3 homozygous deletions or mutations are lethal (Homma et al., 2006). When Lifeact-EGFP and Lifeact-mCherry coding sequences lacking the 5′ and 3′ untranslated regions (UTRs) were introduced into OT-I T cells by retroviral transduction, we did not detect increases in GFP or mCherry fluorescence upon activation (Fig. 3I). However, when we amplified and compared genomic sequences between the Lifeact-EGFP and Lifeact-RFP mice, the immediate 5′ and 3′ UTR sequences that flank the transgenes were found to be common (Fig. S3H). This would be expected, since they were cloned into similar vectors prior to oocyte transfer during generation of the transgenic mice (Riedl et al., 2010). This indicates that the UTRs are unlikely to be involved in the Lifeact-GFP-specific gain of fluorescence. Furthermore, chromosome 7-specific primers paired with primers targeted to shared regions of Lifeact-EGFP and Lifeact-RFP confirmed the insertion site of Lifeact-EGFP but failed to generate any amplicons from Lifeact-RFP genomic DNA, indicating that the Lifeact-RFP transgene was not inserted at the same locus.

Since the gain in GFP fluorescence in LGO1 T cells was reliably detected by flow cytometry and sustained for 48 h, we tested if it was suitable for tracking T cell activation in in vivo experiments. To this end, LGO1 CTLs were adoptively transferred into mice engrafted with tumours expressing cognate antigen or non-cognate tumours on contralateral flanks (Fig. 4A). Whereas TCR Vα2 was downregulated in all LGO1 CTLs isolated from cognate antigen-expressing tumours, suggesting they had all been activated upon antigen re-encounter, only a subset had upregulated IFN-γ production, with a corresponding increase in GFP intensity. (Fig. 4B; Fig. S4A). By contrast, CTLs isolated from the spleen and contralateral non-cognate tumours did not downregulate TCR Vα2; neither did they express IFN-γ or display an increase in GFP fluorescence (Fig. 4B; Fig. S4A), indicating that Lifeact-EGFP gain is a sensitive reporter of T cell activation in vivo.

We next defined a broadly applicable method to identify T cell activation levels in imaging data. Using simple image analysis and a three-sigma interval statistical method – based on Lifeact-EGFP fluorescence intensity and the thereby derived likelihood of a given cell not belonging to the unactivated population – we defined four states that describe both the certainty and extent of activation: indeterminate, low, medium and high (see Materials and Methods and Fig. 4C). By applying these intensity thresholds to LGO1 cells imaged during a cognate search and kill assay, we detected a drop in the indeterminate subpopulation over time, concomitant with an increase in all three activated subpopulations (Fig. 4D; Movie 3). As expected in a non-cognate context, the majority of LGO1 cells remained in the ‘indeterminate’ group, where the cells cannot confidently be identified as activated (Fig. 4D; Movie 4). After 12 h in a cognate environment, approximately half of the tracked LGO1 cells were identifiable as highly activated with near certainty, i.e. with only a 0.14% likelihood of not being activated (Fig. 4E). Thus, this intensity-based statistical thresholding method readily identifies activated T cell subpopulations in imaging data.

We next imaged at high magnification LGO1 CTLs migrating in a 3D collagen matrix and undergoing activation in a temporally controlled manner via acute exposure to monomeric pMHC. Upon addition of cognate pMHC to the bathing medium, the CTL population rapidly stopped moving (Fig. S4B) in contrast to the addition of a control pMHC, which did not result in any detectable change in instantaneous speed (Fig. S4B). By tracking individual cells exposed to cognate pMHC, we observed that LGO1 T cells gradually changed from a polarised migratory state with an F-actin-rich leading edge and F-actin-poor uropod to a rounded morphology with a more uniform distribution of F-actin (Fig. 4G; Movie 5), which was not observable in the presence of control pMHC (Fig. S4C). Furthermore, by applying a similar intensity-based statistical thresholding method as above, we could monitor individual T cell progression through stages of activation (Fig. S4D). Following exposure to cognate pMHC, T cells stopped migrating (Fig. 4F) and sequentially progressed through the four delimited states, from indeterminate to highly activated, as they increased in GFP fluorescence intensity (Fig. 4G; Movie 5). On average, LGO1 T cells reached the highly activated state 5.7±1.4 h following stimulation with cognate pMHC (Fig. 4H), consistent with our results obtained via population-wide divergence analysis of LGO1 CTLs in cognate versus non-cognate search and kill assays (Fig. 1C), notwithstanding the different method of activation.

Real-time ex vivo and intravital imaging approaches are now widely employed to visualise cellular functions and behaviour, complemented by a selection of fluorescent reporters. We show that the widely used Lifeact-EGFP reporter of F-actin dynamics also serves as a real-time indicator of individual CD4⁺ and CD8⁺ T cell activation. A marked increase in GFP intensity is reliably detectable by flow cytometry and fluorescence microscopy, both following naïve T cell priming and antigen re-encounter by effector T cells. The Lifeact-EGFP reporter thus enabled us to track distinct search, arrest and killing functions of primary effector T cells in a search-and-kill assay, and to identify T cell activation on a population and individual level. We were also able to monitor the subcellular F-actin dynamics, motility and morphological characteristics of T cells, extending our ability to study rapid subcellular changes that precede, accompany or follow T cell activation. Since the Lifeact-EGFP fluorescence gain is sustained, we could also sort T cells based on activation levels for downstream experiments and track T cell activation in in vivo experiments, with future potential to monitor in situ activation by intravital imaging.

The gain in Lifeact-EGFP fluorescence correlates with late phase activation and effector cytokine production. Basal GFP expression enables tracking of T cells before they are activated, providing an advantage over reporters that fluoresce only upon cytokine production [including TNF-α (Shebzukhov et al., 2014) and IL-2 (Naramura et al., 1998)] and hence require additional vital dye labelling; or reporters that monitor the translocation of nuclear factor of activated T cells (NFATs) (Lodygin et al., 2013) that require an additional constitutive nuclear label. It will be interesting to compare the Lifeact-EGFP sensitivity and specificity with the Nur77^EGFP reporter (Au-Yeung et al., 2014), in which GFP is expressed from the immediate early gene Nr4a1 (Nur77) locus and can rapidly and specifically be induced upon antigen receptor triggering. The Lifeact-EGFP reporter is unsuitable for monitoring proximal and transient TCR signalling, events for which fluorescently-tagged TCR (Friedman et al., 2010; Neve-Oz et al., 2015), Lat (Azar et al., 2010) and Zap-70 (Neve-Oz et al., 2015) FRET-based systems or genetically encoded calcium indicators (GECIs) (Mues et al., 2013; Thestrup et al., 2014) have been employed. The Lifeact-EGFP reporter is likely to prove more valuable in in vivo and clinical settings, where sustained signals are more favourable than transient ones for identification and selection of activated cell cohorts or individuals.

The activation-induced gain in Lifeact-EGFP fluorescence is translationally controlled. Polysome analyses in this study showed that global translation, translation discrimination and the mTOR-dependent regulation of translation initiation and elongation phases are distinct in resting and activated T cells, consistent with the increasingly appreciated complexity of translational regulation of immune functions. The Lifeact-EGFP reporter may well prove valuable for investigating other leukocytes and non-immune cell types that employ translational regulation upon activation, maturation or differentiation.

Lifeact-EGFP mice (Riedl et al., 2010) (a gift from E. Hardeman and R. Wedlich-Söldner) and mT/mG (membrane tdTomato) reporter mice (Muzumdar et al., 2007) were crossed to OT-I mice (a gift from E. Deenick) and the resultant Lifeact-EGFP×OT-I (LGO1) and mT/mG.OT-I strains bred and housed at Australian BioResources (Moss Vale, NSW, Australia). gBT-1 TCR transgenic mice with TCR specific for H-2K^b/SSIEFARL (Coles et al., 2003) that had been crossed to UBI-GFP mice (expressing GFP under control of the human Ubiquitin C promoter) (Schaefer et al., 2001) (gBT-I.EGFP) were a kind gift from W. R. Heath, and Lifeact-mRFPruby mice (Riedl et al., 2010) were a kind gift from E. Hardeman.

Mouse spleens were dissociated in a 70 µm cell strainer then washed with 10 ml T cell medium (TCM) (TCM: RPMI 1640; 10% heat-inactivated foetal calf serum (HI-FCS); 1 mM sodium pyruvate; 10 mM HEPES; 100 U/ml penicillin; 100 µg/ml streptomycin; 50 µM β-mercaptoethanol (Gibco, Thermo Fisher Scientific, Waltham, MA, USA). Splenocytes were centrifuged at 227 g for 5 min and resuspended in 1 ml red cell ACK lysis buffer (Gibco, Thermo Fisher Scientific) at 4°C for 1 min. A further 10 ml of TCM was added to the cell suspension and passed through the same strainer before centrifugation, as above. For OT-I strains, 3×10⁶ splenocytes were seeded in 10 ml TCM per 25 cm³ flask (incubated upright) and pulsed with 1 µg/ml SIINFEKL or SSIEFARL peptide (Auspep, Melbourne, VIC, Australia) for OT-I or gBT-I, respectively, for 4 h at 37°C and 5% CO₂, washed and returned to 10 ml warm, fresh TCM for further incubation at 37°C, 5% CO₂. The following day (day 1), cells were washed and resuspended in fresh TCM supplemented with 100 ng/ml mouse IL-2 (R&D Systems, Minneapolis, MN, USA). Cells were then cultured to day 6, expanded based on confluency and replenished with fresh TCM and IL-2 every 1-2 days.

To generate polyclonal effector T cells from non-TCR transgenic Lifeact-EGFP or Lifeact-mRFPruby mice, splenocytes were stimulated for 24 h in vitro with 1 µg/ml anti-CD3 (clone: 145-2C11; BioLegend, San Diego, CA, USA), 1 µg/ml anti-CD28 (clone: 37.51; BioLegend) and 100 ng/ml IL-2 in 10 ml of TCM before washing and expansion in TCM and IL-2 until day 6 as above. In some experiments, cells were frozen on day 3 at 3×10⁶ cells/ml in 10% dimethyl sulfoxide (DMSO) in heat-inactivated FCS as described (Galeano Niño et al., 2016), and thawed for expansion until day 6 before use.

Effector CD8⁺ T cells were cultured to day 6 as described above, whereas EL-4 mouse lymphoma cells (a gift from G. Logan) were maintained in TCM and pulsed overnight with 1 µg/ml SIINFEKL to generate cognate target cells. Un-pulsed EL-4 cells were used as non-cognate controls. EL-4 cells were stained with 1 µM CellTracker Deep Red dye (Thermo Fisher Scientific) as per manufacturer's instructions and allowed to recover for 30 min at 37°C before use. T cells and stained EL-4 cells (cognate or non-cognate) were counted and the following was performed on ice: 1.5×10⁵ T cells and 1.5×10⁵ EL-4 cells were dispensed into the same 1.5 ml Eppendorf tube and centrifuged at 227 g for 5 min. The co-pelleted cells were resuspended in 40 µl ice-cold TCM and 10 µl of 10× PBS and 1.14 µl of 1 M NaOH added. Next, 50 µl of ice-cold rat tail collagen type I (Corning, One Riverfront Plaza, NY, USA) was added, quickly mixed and 70 µl of cell-collagen mixture immediately transferred to a single well of a 96-well imaging plate (Greiner Bio-One, Kremsmünster, Austria). The gels were set at 37°C for 10 min prior to addition of 200 µl pre-warmed, Phenol Red-free TCM containing 1 µg/ml propidium iodide (PI) (Thermo Fisher Scientific). Four-dimensional imaging data was collected through a 10× dry objective with a 0.30 numerical aperture on a Leica TCS SP5 confocal microscope (Leica Microsystems, Wetzlar, Germany). Lifeact-EGFP was excited at 488 nm, PI at 561 nm and CellTracker Deep Red Dye at 630 nm using a tuneable white light laser (Leica Microsystems) and emitted light collected within 498-550 nm, 570-620 nm and 640-690 nm, respectively. Data were acquired from the x, y and z planes, with a total z depth of 70 µm and a step size of 1.8 µm every 80 s for 12 h.

The resultant intensity difference I_Diff was then used to determine the provenance of the PI foci as follows: I_Diff>10: PI focus ascribed to a dying tumour cell; I_Diff<−10: PI focus ascribed to a dying T cell; −10≤I_Diff≤10: PI focus of unknown origin. This last ‘unknown’ category enables the identification of nuclear debris diffusing into the field of view that ought not to be counted as a ‘kill’ within the analysed volume.

Effector cell and EL-4 target cell co-cultures were performed in either collagen gels or in suspension (see figure legends). To prepare co-cultures within collagen gels, 1.5×10⁵ T cells and 1.5×10⁵ EL-4 cells were resuspended in 70 µl ice-cold TCM, 30 µl of which was transferred to a separate tube on ice to establish the input at 0 h. The remaining cell mixture was embedded in collagen as described for the search-and-kill assay above (MatTek, Ashland, MA, USA). To prepare cells for flow cytometry or sorting, cells were released by gel dissociation in 2 ml of warm TCM containing 2% collagenase type IV (Sigma-Aldrich, St Louis, MO, USA) for 25 min at 37°C, washed in 10 ml FACS wash buffer and centrifuged (227 g, 5 min). To prepare co-cultures in suspension, T cells were incubated at a 1:1 ratio with either cognate or non-cognate EL-4 target cells in TCM in 6-well (up to 1×10⁶ cells in 3 ml) or 96-well plates (up to 2×10⁵ cells in 0.2 ml).

To stimulate effector cells with beads, OT-I cells were co-pelleted with streptavidin beads coated with biotinylated H-2K^b/SIINFEKL peptide (Tetramer Synthesis Service, John Curtin School of Medical Research, Australia National University, ACT, Australia, a gift from K. Gaus). After the indicated incubation times, cells were collected and centrifuged (227 g, 5 min). To stimulate polyclonal effector cells, 5×10⁵ cells were incubated for 20 h with streptavidin beads coated with 1 µg/ml biotinylated anti-CD3 (clone: 145-2C11; Thermo Fisher Scientific, Waltham, MA, USA) and 1 µg/ml biotinylated anti-CD28 (clone: 37.51; Thermo Fisher Scientific) monoclonal antibodies or with 1 µg/ml biotinylated IgG2a kappa isotype control (clone: eBR2a; Thermo Fisher Scientific).

The following inhibitors were used: 10 µM Lck-specific inhibitor, 7-Cyclopentyl-5-(4-phenoxyphenyl)-7H-pyrrolo[2,3-d]pyrimidin-4-ylamine (Calbiochem, Darmstadt, Germany), 10 µM PP2 Src-family kinases inhibitor (Abcam, Cambridge, UK), 10 µM PP3 Src-family kinases control (Abcam), 5 µM UO126 (Sigma-Aldrich) Mek1/2 inhibitor, 1 µM latrunculin-A (Abcam), 10 µM cycloheximide (Sigma-Aldrich), 5 µM MG-132 (Sigma-Aldrich), 1 µM INK128 or 0.01% DMSO (Lck inhibitor vehicle). T cells and EL-4 cells were pre-treated for 1 h in the inhibitors, which were also added to the gels or co-cultures for the duration of the experiment.

Where indicated, cells were surface stained with 1.25 µg/ml anti-TCR-β (conjugated with Alexa Fluor 700; clone: H57-597; BioLegend, cat. no. 109223), 1.25 µg/ml⁻¹ anti-CD4 (conjugated with fluorescein or phycoerythrin; clone: RM4-4; BioLegend, cat. no. 100511) or 1.25 µg/ml anti-CD8 (conjugated with Pacific Blue; clone: 53-6.7, BioLegend, cat. no. 100728) for 30 min in FACS wash buffer (2% HI-FCS, 2 mM EDTA and 0.02% sodium azide in 1× PBS). Final cell suspensions were prepared in 200 µl cold FACS wash buffer containing 0.5 µg/ml 4′,6-diamidino-2-phenylindole (DAPI) and acquired on the BD Fortessa x20 flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA). Flow cytometry data were analysed with FlowJo software (Treestar Inc., Ashland, OR, USA).

For intracellular staining, cells were first incubated with LIVE/DEAD™ Fixable Aqua Dead Cell Stain (Thermo Fisher Scientific) for 30 min on ice, washed twice, then fixed and permeabilised using the BD Bioscienes Cytofix/Cytoperm™ Plus kit. Permeabilised cells were stained for 45 min with 2.0 µg/ml allophycocyanin-conjugated anti-TNF-α (Clone: MP6-XT22; eBioscience, cat. No. 17-7321-81, San Diego, CA, USA), 2.0 µg/ml phycoerythrin-conjugated anti-IFN-γ (Clone: XMG1.2; BioLegend, cat. no. 505807), rabbit anti-phospho-Zap-70 (Tyr319)/Syk (Tyr352) (1:1000, Cell Signaling Technologies, cat. no. 2717S, Danvers, MA, USA), phospho-p44/42 MAPK (Erk 1/2) (Thr202/Tyr204) (1:200, Cell Signaling Technology, cat. no. 9101S), anti-beta Actin (1:100, Abcam, cat. no. AB8227, Cambridge, UK), 5 U Alexa Fluor 647-conjugated Phalloidin (Invitrogen, cat. no. A22287, Thermo Fisher Scientific) or 1.0 µg/ml Alexa Fluor 647-conjugated anti-GFP (Bioss Antibodies Inc, cat. no. bs-0890R-A647, Woburn, MA, USA) in 1× BD Biosciences Perm/Wash buffer. Where secondary antibody staining is required, cells were centrifuged (227 g, 5 min) and stained for 45 min in goat anti-rabbit IgG conjugated to either 1.0 µg/ml Alexa Fluor 594 (Invitrogen, cat. no. A11037, Thermo Fisher Scientific) or to phycoerythrin (1:100, Invitrogen, cat. no. P-2771MP, cat. no. A11037, Thermo Fisher Scientific). Cells were then centrifuged (227 g, 5 min) and resuspended in 400 µl FACS wash for acquisition on the BD Fortessa x20 flow cytometer (BD Biosciences) and analysis with FlowJo software (Treestar Inc.).

EL-4 non-cognate and cognate cells were interchangeably stained with CellTracker Deep Red as described above, mixed in a 1:1 ratio and 1×10⁵ of each cell type dispensed into 24W plates in TCM. Effector LGO1 CTLs were pre-treated with 10 µM Lck-specific inhibitor, 5 µM UO126 (Mek1/2 inhibitor) or 0.01% DMSO (vehicle control) for 1 h and 1×10⁵ of CTL was added to the mixture of EL-4 cells. TCM containing each inhibitor was further added to achieve final concentrations of 10 µM Lck-specific inhibitor, 5 µM UO126 or 0.01% DMSO in a final volume of 0.5 ml for 12 h. Prior to flow cytometric acquisition, plates were transferred to ice, and 50 µl cold FACS wash buffer containing 0.5 µg/ml DAPI and a reference population of mScarlet-I-expressing EL-4 equivalent to 1×10⁴ Spherotech AccuCount Blank Particles (Spherotech, Chicago, IL, USA) was added before acquisition on the BD Fortessa x20 flow cytometer. Absolute cell numbers were calculated and the cytotoxic index calculated as previously described (Galeano Niño et al., 2016) where input and output are numbers at the start and end of the cytotoxic assay, respectively.

To prepare cells for RNA extraction, 2×10⁶ LGO1 cells were incubated in 3 ml TCM in 6-well plates with equal amounts of either cognate or non-cognate EL-4 cells for 20 h. Cells were then collected and centrifuged (227 g, 5 min) and washed 2× with PBS before being re-suspended in 2 ml cold TCM. A total of 1×10⁶ EGFP-expressing LGO1 cells were sorted on the BD FACS Aria III flow sorter (BD Biosciences). Cells were then centrifuged (227 g, 5 min) before RNA isolation using the RNeasy Mini Kit (Qiagen, Hilden, Germany), according to manufacturer's instructions. 1.5 µg of RNA were reverse transcribed to cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Thermo Fisher Scientific) according to the manufacturer's instructions. Real-time quantitative PCR was carried out using the following primers predesigned and synthesised by Sigma-Aldrich (KiCqStart^® SYBR^® Green Primers). eGFP-Fwd: 5′-GTAAACGGCCACAAGTTCAGC-3′; eGFP-Rev: 5′-TGGTGCAGATGAACTTCAGGG-3′; FM1_Rpl13a: 5′-CCTATGACAAGAAAAAGCGG-3′; RM1_Rpl13a: 5′-CAGGTAAGCAAACTTTCTGG-3′; FM1_B2m: 5′-GTATGCTATCCAGAAAACCC-3′; RM1_B2m: 5′-CTGAAGGACATATCTGACATC-3′; FM1_Tnf: 5′-CTATGTCTCAGCCTCTTCTC-3′; RM1_Tnf: 5′-CATTTGGGAACTTCTCATCC-3′; FM1_Ifng: 5-TGAGTATTGCCAAGTTTGAG-3′; RM1_Ifng: 5′-CTTATTGGGACAATCTCTTCC-3′; FM1_Actb: 5′-GATGTATGAAGGCTTTGGTC-3′; RM1_Actb: 5′-TGTGCACTTTTATTGGTCTC-3′; FM1_Actg1: 5′-GTATCCATGAGACCACTTTC-3′; RM1-Actg1: 5′-CAATGATCTTAATCTTCATCGTG-3′. 20 ng cDNA was added to each well of a 96-well PCR plate with 1 µM forward and reverse primer for each gene. 40 cycles were performed, with denaturing temperature at 95°C for 15 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s. The amount of amplicon was measured using SYBR Green and detected in a BIO-RAD CFX96 Real time system (Bio-Rad Laboratories, Hercules, CA, USA). The expression of each gene was normalised to the expression of the housekeeping genes β2-microglobulin (β2M) and ribosomal protein L13A (RPL13A).

2×10⁶ LGO1 cells were cultured overnight together with equal amounts of either cognate or non-cognate EL-4 cells in 6-well plates. LGO1 cells cultured with non-cognate EL-4 cells are considered ‘resting’, and cells cultured with cognate EL-4 cells are considered ‘activated’. Cells were then centrifuged (227 g, 5 min), washed twice with cold PBS and resuspended in 2 ml cold TCM for sorting by flow cytometry. LGO1 cells from non-cognate culture conditions were then used to establish three gates based on Lifeact-EGFP fluorescence intensity in the cognate LGO1 cells: the high gate corresponds to the GFP signal that does not overlap with the intensity of non-cognate LGO1 cells; medium encompasses a portion of the GFP intensity from non-cognate LGO1 cells and low corresponds with the GFP signal that entirely overlaps with the GFP intensity of non-cognate LGO1 cells. Similar gates were drawn in the non-cognate LGO1 cells. 2×10⁵ LGO1 cell non-cognate, and 2×10⁵ LGO1 cognate cells were then sorted into each of these gates, to give 6 cell populations. After two washes with PBS, sorted cells were cultured into wells of a round-bottomed 96-well plate with 250 µl serum-free TCM. After 4 h, plates were centrifuged (227 g, 5 min) and supernatants collected. Supernatants were then filtered through a 0.22 µm pore filter (Corning, Corning, NY, USA) and IFN-γ and TNF-α cytometric bead array carried out using the LEGENDplex™ Multi-Analyte Flow Assay Kit (BioLegend) as per manufacturer's instructions. 25 µl of each supernatant was mixed with 25 µl of captured beads against TNF-α and IFN-γ for 2 h in a V-bottom plate. After two washes, the samples were incubated with 25 µl of the detection antibody for 1 h followed by the addition of 25 µl of PE-conjugated secondary antibody, both supplied with the kit. Cytokine levels from the supernatants were interpolated from standard curves generated using recombinant proteins from the kit. The data were analysed by FlowJo software.

Cells were lysed in RIPA buffer (25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) supplemented with protease inhibitors (Halt™ Protease Inhibitor Cocktail (100×) (Thermo Fisher Scientific). Cell debris was pelleted by centrifugation at 12,000 g for 20 min at 4°C, and the supernatant was used for western blotting. Total protein concentration was determined with Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific). 30 µg of protein per sample was separated by Bolt™ 4-12% Bis-Tris gel (Thermo Fisher Scientific) using SeeBlue™ Plus2 Pre-stained Protein Standard (Thermo Fisher Scientific) as a size marker and transferred onto nitrocellulose membranes (Thermo Fisher Scientific). Membranes were probed with primary antibodies against GFP (Abcam, polyclonal), BAG3 (Origene, polyclonal) and Tubulin (Abcam, YL1/2) at 1:1000, 1:1000 and 1:5000 dilutions, respectively. Followed by incubation with secondary antibodies anti-rabbit IgG (Amersham, 1:2000) and goat anti-rat (Thermo Fisher Scientific, 1:10,000). ECL™ Prime Western Blotting Substrate (Merck) was used for detection. Image acquisition of blots was performed on ChemiDoc™ Imaging System (Bio-Rad) and densitometric quantification of bands performed using Image Lab Software (Bio-Rad).

The following coding sequences were cloned into the murine stem cell virus (MSCV)-based retroviral expression vector and are available upon request: Lifeact-EGFP, Lifeact-mCherry, mTagBFP2-F2A-OVA-Ag85b [which includes the coding sequence for OVA257-264 (SIINFEKL)]. Ecotropic retrovirus was produced by calcium phosphate-mediated transfection of Eco-Pack 2-293 cells (a gift from Y. Wang) using 15 µg of rMSCV-Lifeact-EGFP or 15 µg of rMSCV-Lifeact-EGFP per 4×10⁶ cells. Calcium phosphate precipitates were prepared by dropwise addition of 15 µg plasmid DNA (in 0.5 ml of 1 mM Tris, 0.1 mM EDTA, pH 8.0, 0.25 M CaCl₂) to 2× HEPES-buffered saline (0.5 ml of 280 mM NaCl, 50 mM HEPES, pH 7.0, 1.5 mM Na₂HPO₄). Precipitates were formed by incubation at room temperature for 20 min and 1 ml of precipitate was added to cells for 16 h. At 48 h after transfection, viral supernatants from 8 dishes were concentrated by high-speed centrifugation (33,403 g, 90 min, 4°C, in a Hitachi R18A fixed angle rotor) and used to transduce 3×10⁶ day 1 OT-I T cells activated by SIINFEKL and IL-2 as described above. Transduction was performed in 6-well dishes in the presence of 4 µg/µl Polybrene (Sigma-Aldrich) and centrifuged for 1 h at 2000 g at 32°C before sorting at 72 h post-transduction (BD FACS Aria III flow sorter, BD Biosciences). Lifeact-EGFP transduced OT-I cells expressing comparable GFP intensities to day 6 LGO1 cells were identified (∼30% of the population) and sorted. As transduction efficiencies were comparable for both Lifeact-EGFP and Lifeact-mCherry, the sort gates were also set on a ∼30% of mCherry-expressing OT-I, excluding high and low-expressors. Sorted cells were recovered for 48 h in TCM before use.

To transduce EL-4 cells, retrovirus pseudotyped with the vesicular stomatitis virus (VSV-G) envelope were produced by polyethylenimine (PEI, molecular weight 4000, PolySciences, cat. no. 24885-2, Warrington, PA, USA) transfection of GP2-293 cells (Clontech, Palo Alto, CA, USA). NaOH-neutralised PEI (1 mg/ml) was complexed with 6.8 µg rMSCV-mTagBFP2-F2A-OVA-Ag85b and 3.2 µg of pMD2.G plasmid (VSVG coding sequence expressed from the CMV promoter) for 30 min at room temperature before addition to 7×10⁶ GP2-293 cells. At 72 h after transfection, viral supernatant was used to transduce EL-4 and mTagBFP2-expressing EL-4 were sorted (BD FACS Aria III) 72 h after transduction. Eco-Pack and GP2-293 cells were maintained in DMEM (Gibco) containing 4.5 g/l glucose, 4 mM L-glutamine and 1 mM sodium pyruvate supplemented with 10% heat-inactivated FBS.

Genomic DNA from a heterozygous Lifeact-EGFP mouse was extracted using the Isolate II Genomic DNA Kit (Bioline, Meridian Life Sciences). The sequencing library was prepared using TruSeq DNA Nano Library Prep (Illumina Inc.), according to the manufacturer's protocols. Paired end sequencing (2×150 bp) was performed on a HiSeq X sequencer (Illumina Inc.), at the KCCG sequencing laboratory, Garvan Institute of Medical Research. We obtained a total yield of 67.47 GB. The pCAG-promoter-Lifeact-EGFP transgene sequence of the Lifeact-EGFP mouse was inferred from the description of the transgene preparation (Niwa et al., 1991; Riedl et al., 2010) without access to the true transgene sequence. The Mouse Dec. 2011 (GRCm38/mm10) assembly reference genome was obtained from the UCSC genome browser data (ftp://hgdownload.soe.ucsc.edu/goldenPath/mm10/bigZips/chromFa.tar.gz) provided by the GRC Mouse Genome (https://www.ncbi.nlm.nih.gov/grc). A reference genome was prepared by combining the inferred pCAG-promoter-Lifeact-EGFP transgene sequence with the mm10 reference genome using STAR Aligner (version 2.5.2b; Dobin et al., 2013) genomeGenerate. Mouse paired end reads were aligned to the mm10/pCAG-promoter-Lifeact-EGFP reference genome using STAR Aligner alignReads. To identify any potential integration sites, chimeric regions joining the pCAG-promoter-Lifeact-EGFP transgene to the mm10 genome segments were identified from the STAR aligner output. Only one integration site was identified. Aligned reads were visualised using the Integrative Genomics Viewer (IGV) (Robinson et al., 2011). A mm10/pCAG-promoter-Lifeact-EGFP reference genome for use with IGV was prepared by concatenating the inferred pCAG-promoter-Lifeact-EGFP reference genome sequence to the mm10 reference genome as an additional contig with indexing using Samtools faidx (version 1.4; Li et al., 2009). Genome alignments in bam format were indexed using bwa (version 0.7.3a-r367; Li and Durbin, 2009). A deletion was suspected at the transgene integration site based on visualisation of reduction of read depth spanning the region between the chimeric regions from both ends of the transgene. The deletion was assessed using CNVNator (version 0.3.3; Abyzov et al., 2011) with root (version 6.14.04; https://root.cern.ch) with analysis of chr7 only. Additional sequences from the pCAG-promoter-Lifeact-EGFP transgene were manually collected from soft clipped reads at the integration site, then manually aligned to generate an additional transgene sequence. As the transgene sequence obtained differed from the inferred sequence, we repeated the above method (genome generate through to IGV visualisation) across 4 additional iterations using the updated transgene sequence each iteration, clarifying and extending additional novel transgene sequence until a near complete transgene sequence was obtained.

Genomic DNA was extracted from 3-5 million T cells isolated from the Lifeact-EGFP×OT-1 and Lifeact-mRFPruby mouse using the Isolate II isolation kit (Bioline) and subject to Touchdown PCR using a combination of the following primer pairs (100 ng of primer per 50 µl reaction) with PfU-Ultra polymerase (Agilent Technologies, now Integrated Sciences, NSW, Australia): GFP-Fwd 5′-GTCCTGCTGGAGTTCGTGAC-3′; Ruby-Fwd 5′-GGACTACACCATCGTGGAACA-3′; Rabbit-Rev 5′-CCCATATGTCCTTCCGAGTG-3′; Ch7-Fwd 5′-TAAGCAAGATCTTAGCTTCAAACTCCCAGTCTATGG-3′;Ch7-Rev 5′-TAAGCAGTCGACACTTAGAGTCACCGTTCAAACACA-3′; Rabbit-Fwd 5′-CACTCGGAAGGACATATGGG-3′; CAG-Fwd 5′-TCCGCGTTACATAACTTACGG-3′; CAG-Rev 5′-GGGCGTACTTGGCATATGAT-3′. Additional primers as per Fig. S3H were as follows: A: 5′-ACGGGGTCATTAGTTCATAGCC-3′ and 5′-GGGCGTACTTGGCATATGAT-3′; B: 5′-ACGGGGTCATTAGTTCATAGCC-3′ and 5′-AGATGGGGAGAGTGAAGCAGA-3′; C: 5′-TCCGCGTTACATAACTTACGG-3′ and 5′-GGGCGTACTTGGCATATGAT-3′; D: 5′-TCCGCGTTACATAACTTACGG-3′ and 5′-AGATGGGGAGAGTGAAGCAGA-3′; E: 5′-TCTGCTTCACTCTCCCCATCT-3′ and 5′-GCGCTAATTACAGCCCGGA-3′; F: 5′-CGCTGCGTTGCCTTCGC-3′ and 5′-TCGCACGATTACCATAAAAGGCA-3′; G: 5′-CGTGTGTGTGTGCGTGGG-3′ and 5′-TCGCACGATTACCATAAAAGGCA-3′; H: 5′-TGCCTTTTATGGTAATCGTGCGA-3′ and 5′-GCTTCCCTCCATCTTGACCTTAA-3′; I: 5′-CCGAAATCTGGGAGGCGC-3′ and 5′-CAATCTCGAACTCGTGTCCGT-3′; J: 5′-GGACTACACCATCGTGGAACA-3′ and 5′- CCCATATGTCCTTCCGAGTG-3′; K: 5′-CACTCGGAAGGACATATGGG-3′ and 5′-GAAGAGGGACAGCTATGACTGG-3′; L: 5′-CACTCGGAAGGACATATGGG-3′ and 5′-CAGGTCGAGGGATCTTCATA-3′. After an initial denature step (2 min, 95°C), 10 touchdown PCR cycles [denature (30 s, 95°C), anneal (30 s, reducing at 1°C per cycle from 69°C to 59°C), extension (3 min 19 s, 72°C)] followed by 25 standard cycles [denature (30 s, 95°C), anneal (30 s, 59°C), extension (3 min 19 s, 72°C)] were performed before final extension (5 min, 72°C). Products were separated on 1-2% agarose gels, excised and purified (Wizard SV Gel Clean Up Kit, Promega, Madison, WI) before sequencing using 32 pmol of the amplification primers (Ramaciotti Centre for Genomics, UNSW, Sydney, Australia).

A total of 1×10⁶ EL-4 cells transduced to express mTagBFP2 and OVA-Ag85 bicistronically (cognate EL-4) or non-cognate untransduced EL-4 were separately injected subcutaneously into each flank of 8-week-old Rag-deficient B6.SVJ129-Rag1^tm1Bal/Arc mice (‘RAG1N10’, Australian BioResources, NSW, Australia). The tumours were allowed to grow for 6 days before 1×10⁷ LGO1 day 6 effector cells were adoptively transferred to the mice via tail vein injection in 200 μl PBS. Twenty-four hours post T cell transfer, mice were euthanised by CO₂ asphyxiation. The spleens and both tumours were collected and dissociated with 1 mg/ml collagenase IV (Sigma-Aldrich) for 30 min at 37°C (shaking at 800 rpm). The samples were filtered through 70 µm cell strainers to obtain single cell suspensions. Tumours or spleens were resuspended in final volumes of 2 ml or 5 ml FACS wash buffer, respectively. 100 µl of cells from these suspensions were mixed with 100 µl of counting beads equivalent to 2×10⁴ beads (Spherotech AccuCount Blank Particles; Chicago, IL, USA) to determine the absolute number of infiltrating cells. Remaining cells were stained with 2.0 µg/ml anti-TCRVα2 (conjugated with APC/Cy7; Clone: B20.1; BioLegend, cat. no. 127818) antibody and LIVE/DEAD™ Fixable Aqua Dead Cell Stain (Invitrogen). Cells were fixed and permeabilised using the BD Cytofix/Cytoperm™ Plus kit for IFN-γ intracellular staining as described above before acquisition on a BD Bioscience Fortessa x20 flow cytometer (BD Biosciences).

LGO1 effector T cells were incubated at a 1:1 ratio with cognate EL-4 target cells in T25 flasks, upright (up to 12×10⁶ cells each in 24 ml TCM) or were incubated in TCM only for 20 h, in the presence of INK128 (1 μM) or vehicle. Live cells were enriched by layering over 14 ml of Ficoll-Paque™ PLUS (density 1.077 g/ml; GE Healthcare, Pittsburgh, PA) and centrifugation at 400 g without brakes at room temperature. Live cells collected from the buffy layer were washed twice in TCM and pelleted at 277 g. An aliquot of cells analysed by DAPI staining and flow cytometry demonstrated efficient removal of EL4 tumour debris (90% purity). Cell pellets were resuspended in polysome lysis buffer [1% (v/v) Triton X-100, 10 mM Tris-HCl (pH 7.5), 10 mM NaCl, 10 mM MgCl, 3.3 mM dithiothreitol (DTT), sodium deocycholate 10 μg/ml and 0.13 Units/μl recombinant RNAsin ribonuclease inhibitor (Promega, Madison, WI)], immediately snap frozen in liquid nitrogen and stored at −80°C until analysis. A549 (human lung carcinoma) cells were cultured in pH 6.7 buffered DMEM (Dulbecco's modified Eagle medium, Sigma-Aldrich, Castle Hill, NSW, Australia) supplemented with 10% (v/v) fetal bovine serum (FBS) and 100 U/ml penicillin and 0.1 mg/ml streptomycin, and maintained at 37°C for 16 h in humidified air with 5% (v/v) CO₂. Primary mouse cortical neurons or A549 cells were collected by trypsinization, washed twice with PBS (phosphate-buffered saline, Thermo Fisher Scientific, North Ryde, NSW, Australia) and lysed in ice-cold lysis buffer [1% (v/v) Triton X-100, 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 2.5 mM NaH₂P₂O₇, 1 mM β-glycerophosphate, 1 mM dithiothreitol (DTT), 1 mM Na₃VO₄ and protease inhibitor cocktail (1×)]. Lysates were centrifuged at 16,000 g and 4°C for 10 min and total protein concentrations in the supernatants quantified by Bradford assay. Where indicated, insoluble pellets were dissolved in 1× Laemmli sample buffer [62.5 μM Tris-HCl pH 6.8, 1% (w/v) SDS, 10% (v/v) glycerol, 10 μM DTT and 5 μg/ml Bromophenol Blue].

Polysome analysis was performed as previously described (Xie et al., 2019b). For RNA extraction, 1% (w/v) SDS and 0.15 mg/ml proteinase K were added to each fractions, 1:3 (v/v) phenol:chloroform pH 4.5 was then added to the samples to extract RNA, RNA were precipitated from the aqueous phase by the addition of 70% (v/v) isopropanol. RNA pellets were washed once with 80% (v/v) ethanol, before dissolving in RNase/DNase-free water for further analysis.

Total RNA was extracted using TRIzol (Life Technologies). cDNA was produced using the ImProm-II reverse transcription (RT) system (Promega) with oligo(dT)15. For qPCR analysis of sucrose gradient fractions, 1.2 kb kanamycin RNA provided by the RT kit was used as an internal control. qPCR was performed using the primers indicated in the RT-qPCR section above. Samples were analysed with SYBR green dye mix (Life Technologies) on an ABI Step One Plus qPCR instrument (Applied Biosystems, Cheshire, UK). For total RNA analysis, B2M was used as the normalisation control. The comparative threshold cycle (C_T) method was applied to quantify mRNAs present in each sample.

Statistical analyses were performed using a one-way or two-way ANOVA with an unpaired Student's t-test with the means of three to four biological replicates as specified. Data are presented as means±s.d., GraphPad Prism software was used to calculate P-values. *0.01≤P<0.05; **0.001≤P<0.01; ***P<0.001.

For high-resolution tiled 40× (dry objective with 0.75 NA, Leica Microsystems, Wetzlar, Germany) time-lapse imaging, LGO1 cells were embedded in collagen gels as described above, but in the absence of EL-4 cells. Gels were imaged for 1 h before the addition of 1 µg/ml biotinylated H-2K^b/SIINFEKL monomers (Tetramer Synthesis Service, John Curtin School of Medical Research, Australia National University, ACT, Australia, a gift from K. Gaus) in 50 µl to the bathing medium, after which gels were imaged for a further 10 h. As a control, H2K^b/RGYVYQGL biotinylated monomers (Tetramer Synthesis Service, John Curtin School of Medical Research, Australia National University, ACT, Australia, a gift from K. Gaus) were added to the medium. Data was analysed using the Imaris software as described above.

We examined cell fluorescence intensity values from imaging data at t=0 h, when cells are still considered to be unactivated. As the fluorescence intensity values display a Gaussian distribution (D'Agostino & Pearson omnibus K2 test; P=0.1885 with a null hypothesis that the data are sampled from a Gaussian distribution) 99.7% (∼100%) of values will fall within 3 standard deviations (σ) of the mean (µ). Therefore, we defined a fluorescence intensity threshold of μ_t=0+3σ_t=0, where at any given time point, cells displaying a fluorescence intensity above this can be considered with near-certainty as not belonging to the unactivated population, i.e. are certainly activated, and were designated as belonging to the ‘high’ level of activation cohort (magenta). We then applied this sigma-based selection criteria to reflect three further levels of activation; cells exhibiting a fluorescence intensity (GFP) of GFP<μ_t=0+1σ_t=0, considered ‘indeterminate’ (grey), cells where μ_t=0+1σ_t=0≤GFP<μ_t=0+2σ_t=0 were assigned a ‘low’ level of activation (yellow), and cells exhibiting a fluorescence intensity of μ_t=0+2σ_t=0≤GFP<μ_t=0+3σ_t=0 were assigned a ‘medium’ level of activation (cyan). This method is easily implementable for the determination of activation state and certainty where an unactivated reference population is available (either at t=0 in a temporal experiment or as a separate condition).

All analysis steps in this section were performed with the help of custom MATLAB (MathWorks, MA, USA) routines, available upon request.

Time-series of fluorescence intensity distributions of Lifeact-EGFP fluorescence signals (3 independent series for cognate environment, 3 independent series for non-cognate environment) were normalised to a median of 1 at t=0 in order to account for differences in laser powers and systematic variations. Non-cognate condition time-series were pooled, and a mono-exponential decay function [f(t)=A · e^τ·t] was fitted to the median of the pooled data. Cognate and pooled non-cognate time-series were divided by the obtained exponential decay function to correct for photo-bleaching. Only normalised and bleach-corrected data sets were used for further analysis.

To determine the time of divergence, the non-parametric right-tailed Wilcoxon rank sum test was used to compare the cognate and pooled non-cognate intensity distributions at each time point. Briefly, the null hypothesis that Lifeact-EGFP fluorescence from non-cognate and cognate intensity distributions are samples from the same continuous distribution is tested against the alternative hypothesis that intensity values of the cognate distribution tend to be larger than the ones of the non-cognate distribution. Three different test significance levels were considered (0.1587, 0.0227, 0.0014 corresponding to a probability of finding values above the mean plus 1, 2 or 3 standard deviations in a normal distribution), stating the probability that the null-hypothesis is falsely rejected. The time of divergence was determined as the first time-point after which all subsequent P-values were P<0.0014. It should be noted that the maximal temporal difference between matching cognate and non-cognate pairs was 39.6 s, which was considered negligible taking the relevant time-scale (12 h) of the experiment into account.

All analysis steps in this section were performed with the help of custom MATLAB (MathWorks, MA, USA) routines, available upon request.

Time-series of fluorescence intensity distributions of Lifeact-EGFP signals before (‘PRE’, 0-52 min) and after (‘POST’, 52 min signal ending) the addition of cognate H-2K^b/SIINFEKL were determined. ‘PRE’ single-cell intensity traces were re-aligned to start at t=0 and a mono-exponential decay model function [f(t)=A · e^τ·t] was fitted to the re-aligned data set. Re-aligned ‘PRE’ distribution and ‘POST’ intensity traces of five individual cells were divided by the obtained decay to correct for photo-bleaching. Only bleach-corrected data sets were used for further analysis.

‘PRE’ fluorescence intensities were pooled and a non-parametric estimation of the underlying probability distribution function (pdf) was obtained by fitting a kernel distribution with a normal kernel smoothing function to the data set. By minimising with respect to x_lim, intensity thresholds corresponding to specific error probability levels α (i.e. α₁=15.87%, α₂=2.27%, α₃=0.14%) were determined. In other words, the probability to find an intensity value of the ‘PRE’ intensity distribution greater than equals 15.87%, 2.27%, 0.14%, respectively. By determining the time points corresponding to ⁠, three activation times with increasing significance were obtained for each ‘POST’ intensity trace (Fig. 4F; Fig. S4E).

2×10⁶ LGO1 cells were incubated for 20 h with equal numbers of either cognate or non-cognate EL-4 cells. Subsequently, half of each population remained untreated (cognate and non-cognate) whereas the other half was treated with LatA for 30 min [cognate (LatA), non-cognate (LatA)]. Prior to imaging, cells were transferred to untreated glass-bottom dishes (Mattek). FLIM data was acquired with a Picoquant Microtime200 setup (Picoquant, Berlin, Germany), using a 60× UPlanSApo NA 1.2 water immersion objective (Olympus, Tokyo, Japan) and a laser pulse frequency of 20 MHz at a wavelength of 470 nm. For each of the four conditions, three different fields of view were imaged.

Statistical parameters are reported in the figures and legends. Mann–Whitney U-tests were used to compare medians between two groups and Kruskal–Wallis test to compare medians between more than two groups followed by Dunn's multiple comparison tests. One-sample t-tests were used to compare the mean of the log₁₀ fold change to a hypothetical mean of 0 (no difference between the groups). Statistical analyses were performed using Prism 7.0 (GraphPad Software, La Jolla, CA, USA).

Animal breeding and experimentation was conducted in accordance with New South Wales state and Australian federal laws and animal ethics protocols overseen and approved by the University of New South Wales Animal Care and Ethics Committee.

The authors thank Sandra Cooper for helpful discussions, Mark Read for technical advice and the BioMedical Imaging Facility and Biological Resources Imaging Laboratory of UNSW for assistance with imaging and flow cytometry, respectively. M.B. acknowledges Bitplane AG for an Imaris Developer licence.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.238014.reviewer-comments.pdf.