Cancer cell elimination by cytotoxic T cell cooperation and additive damage

Cytotoxic T lymphocytes (CTL) eliminate tumor target cells in an antigen and cell-contact dependent manner. Lethal hit delivery occurs as a rapid and binary, “yes/no” process when immunogenicity is very high1–3, however in vivo CTL often fail to kill solid tumor cells during 1:1 conjugations4–6. Using long-term time-lapse microscopy in three distinct tumor cytotoxicity models and statistical modeling, we here show that migrating CTL transit between target cells and initiate apoptosis by a series of sublethal interactions (‘additive cytotoxicity’), while individual conjugations rarely induced apoptosis. Sublethal damage included perforin-dependent membrane pore formation, nuclear lamina rupture and DNA damage, and these events resolved within minutes to hours. In immunogenic B16F10 melanoma tumors in vivo, frequent serial engagements and sublethal hit delivery of CTL was largely confined to interstitial niches in the invasion front, resulting in eradication of invading tumor cells. Thus, additive cytotoxicity is a probabilistic process achieved by a series of CTL-target cell engagements and sublethal events. The need for additive “hits” has implications for the topographic mechanisms of elimination or immune evasion of tumor cells and microenvironmental regulation of CTL accumulation and cooperation by targeted therapy.


Cytotoxic T lymphocytes (CTL) eliminate tumor target cells in an antigen and cell-contact
Cytotoxic T lymphocytes and NK cells can bind to and attack more than one target cell, termed 39 "serial killing" 7-9 . Estimations from bulk killing assays and mathematical modeling suggest, that (lasting minutes to hours), with lag times between initial CTL binding and target cell death 61 lasting 1.8 ± 1.5 h, and was followed by a subsequent period of ongoing CTL engagement with 62 the dead cell body ('necrophilic phase') (Extended Data Fig. 1 b-d). This extended lag phase 63 until apoptosis differs from lag phases obtained for CTL-mediated killing of target cells from 64 leukocyte lineages, lasting <5 to 25 min 1-3 . In regions of low local CTL density, repetitive contacts resulted in the serial killing of multiple neighboring target cells by a single CTL (Fig. 66 1a; Movie 1). On the population level, 50% of the CTL acted as serial killers (maximum of 11 67 killed target cells/24 h), whereas a small CTL subset (15%) repeatedly contacted target cells 68 without inducing apoptosis (Extended Data Fig. 1e). The percentage of CTL with killing 69 capacity correlated with the surface expression of Lamp-1 by 85% of CTL, indicating 70 recognition of the target cells and lytic vesicle exocytosis by the majority of CTL (Fig. 1b). The 71 lag phase to apoptosis was neither compromised nor accelerated over consecutive killing events 72 (Fig. 1c), which resulted in a consistent eradication frequency of 1 kill every 2 hours (Extended  Fig. 2k) 90% followed by target cell survival (Fig. 2d). Ca 2+ events originated at CTLtumor cell 107 contact regions, and differed from unspecific intracellular Ca 2+ fluctuations by signal intensity 108 and duration (Extended Data Fig. 3 b,c). 109 To test whether the variability of perforin release and consecutive Ca 2+ events in B16F10/OVA 110 cells were a consequence of heterogeneous TCR engagement, we quantified Ca 2+ signals in OT1 111 CTL upon target cell contact. OT1 CTL showed comparably high rates of Ca 2+ signaling when 112 contacting MEC-1/OVA and B16F10/OVA cells (80% to 85%), typically within seconds after 113 contact initiation (Extended Data Fig. 3 d). When co-registered with Ca 2+ events in target cells, 114 40% of Ca 2+ positive-CTL contacts with B16F10/OVA coincided with, or were immediately 115 followed by a Ca 2+ event in the target cell (Extended Data Fig. 3 e,f). In conclusion, while TCR 116 triggering in OT1 CTL occurs reliably, the induction of perforin-events in the target cell varied.

118
To address whether transient transmembrane pores were associated with structural intracellular 119 damage, B16F10/OVA cells were engineered to express NLS-GFP 19 or 53BP1trunc-Apple 20 . 120 NLS-GFP leakage into the cytoplasm was detected in 25% of CTL-target cell contacts and 121 absent when CTL lacked perforin expression (Fig. 2a, panel 2 contacts, in dependence of perforin expression in OT1 CTL (Fig. 2 a,b) and CTL density 127 (Extended Fig. 3 l,m). CTL-induced 53BP1 foci persisted for several hours (median: 4 h) and 128 were resolved in 73% of events (Fig. 2 c,d). These data indicate that CTL contacts induce 129 reversible sublethal damage to the nuclear lamina and DNA.

131
Across all tested tumor models, sequential or simultaneous contacts by multiple CTL with the 132 same target cell occurred before target cell death (Fig. 3a). In the B16F10 model, >90% of 133 successful kills were preceded by 2-9 CTL encounters by distinct CTL (Extended Data Fig. 4a). 134 When Ca 2+ events in target cells were recorded, death induction was preceded by serial Ca 2+

143
Additive cytotoxicity 144 We then addressed whether apoptosis could have been induced by rare deadly CTL hits 145 ('stochastic killing') instead of sublethal contacts that add up over time ('additive cytotoxicity').

146
Therefore, we analyzed whether the lethality of the final hit was enhanced by, or independent of, 147 previous Ca 2+ events, by plotting the lag time to apoptosis and target cell survival probability in 148 relation to the number of pre-final Ca 2+ events. Target cells which received two or more hits 149 prior to the lethal one showed accelerated apoptosis induction, together with a sharp decrease in 150 survival probability ( Fig. 3c; Extended Data Fig. 5a,b). The dependence of the lag time to 151 apoptosis on prior CTL hits is expected when additive effects exists, but it is inconsistent with 152 the stochastic killing hypothesis. To explore stochastic killing directly, we performed the same 153 analysis on simulated data, using randomly swapped times between hits and between the last hit 154 and apoptosis. Here, cell death induction was gradual, and neither the lag time to apoptosis nor 155 the survival probability was dependent on the number of prior hits (Fig. 3d). 156 To address how long sublethal events remain relevant, we estimated the time required to repair 157 the damage caused by a single Ca 2+ hit, using a Cox regression model based on additive killing.

158
This resulted in an estimated damage half-life of 56.7 minutes (95% CI: 33.1-112.2) of Ca 2+ 159 events to contribute to lethal outcome. This interval was consistent with the recovery kinetics of 160 nuclear lamina defects (Fig. 2c). Using the Bayesian Information Criterion (BIC) to compare 161 model fits, showed that the model which integrates serial damage and decay explained the data 162 better (BIC difference >10) than a model based on the number of Ca 2+ events alone.  (Fig. 3e), whereas at low CTL density (ET 1:16) infrequent apoptotic 169 events were near-exclusively preceded by single-contact engagements (Fig. 3e). The CTL 170 density effect was not enhanced by the addition of perforin-deficient CTL (Extended Data Fig.   171 2l). This indicates that high CTL density ("swarming") enables efficient apoptosis induction by 172 favoring serial perforin-dependent hits, whereas the poor killing at low CTL density largely 173 relies upon single encounters. 175 To address whether multiple encounters by CTL mediate anti-tumor cytotoxicity in vivo, 176 activated OT1 CTL were adoptively transferred into C57BL/6 J mice bearing intradermal 177 B16F10 melanoma and monitored longitudinally by intravital microscopy through a skin 178 window (Extended Data Fig. 6 a, b). A single application of OT1 CTL caused a transient, dose-179 dependent growth delay of the OVA antigen-expressing but not of control tumors (Extended  (Fig. 4a). This resulted in local ET ratios of 1:1 along the tumor-stroma 183 interface, whereas ET ratios in the tumor core remained below 1:250 (Extended Data Fig. 6 184 d,e). To identify where and by which contact mechanism eradication of tumor cells was 185 achieved, CTL contacts and outcome were mapped using histone-2B/mCherry (H2B/mCherry) 186 to detect nuclear fragmentation in B16F10/OVA cells 23 . Despite comparable ET ratios, high 187 fragmentation rates occurred in the invasion niche but not the tumor rim (Fig. 4b). In either zone, 188 >95% of CTL contacts were transient, short-lived (median duration 15 min) and non-lethal ( Fig.   189 4c). When aggregated, >86% of apoptotic events were preceded by multiple CTL contacts, and a 190 minority (<14%) by individual CTL conjugation (Fig. 4d). In the tumor invasion niche, high              Extended data figure 5. Statistical analysis. a, Time-points of Ca 2+ -positive events (dots color-coded for CTL causing the event) in individual B16F10/OVA cells contacted by OT-1 CTL. Outcome of the interaction is indicated as lethal (filled black dot) or non-lethal (open circle) at the end of each trajectory. b, Survival probability of B16F10/OVA cells having received increasing numbers of Ca 2+ events prior to the terminal event after filtering redundant Ca 2+ events. p-values, log rank test comparing all groups. c, Estimation of damage recovery half-life in B16F10/OVA after one single Ca 2+ event by a statistical model that assumes additive killing (see Methods). Point and error bars: damage half-life that is most consistent with the data and 95% CI. CTL transfer on B16F10 tumor volume based on measurements obtained from epifluorescence overviews. Error bars, means ± SD (5 independent tumors). d, Tumor morphology and distribution of CTL monitored by epifluorescence detection through the skin window (left image) and multi-photon microscopy images (right images) recorded in different tumor subregions. Red, tumor nuclei (H2B/mCherry); green, OT1 CTL (eGFP); cyan, collagen fibers (SHG); blue, blood vessels (70kDa-dextran/Alexa750). e, Sub-regional quantification of ET ratios over time. Error bars, means ± SD (8 independent tumors).