Female C57BL/6 mice aged 6–8 weeks were purchased from Jackson Laboratories. All mice were used in accordance with the Emory University Institutional Animal Care and Use Committee guidelines. Mice were sacrificed if they becamed sick, lethargic or had >10% wt loss prior to tumor volume defined endpoint. The B16F10 cell line was obtained from American Type Culture Collection (ATCC). A B16F10 cell line expressing the glycoprotein (GP) of the LCMV Armstrong strain was generated by lentiviral transduction as described previously.22 Briefly, the codon-optimized GP was cloned into the bicistronic replication deficient lentiviral vector pLVX-IRES-ZsGreen1 (Takara) followed by production of lentiviral particles in 293 T cells (ATCC) and lentiviral transduction of B16F10 cells. A stable B16F10GP cell line was established by sorting B16F10 cells expressing high levels of the green fluorescent protein ZsGreen1 using a FACSAriaII (BD Biosciences) 2 weeks after transduction. The cell line was grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS, 100 U/mL penicillin and 100 ug/mL streptomycin, 2 mM glutamine. The cells were cultured at 37°C with 5% CO₂.

5×10⁵ B16F10GP cells were injected into bilateral flanks subcutaneously on day 0. After the tumor was palpable (10–12 days), mice were irradiated on one side with a Superflab bolus (0.5 cm tissue equivalent material) placed over the tumor, and thereafter tumor measurements were taken as indicated. During radiation, mice were anesthetized with tribromoethanol. The dose to the tumor was 10 Gy using 320 kVp photons at 10 mA. To avoid scatter radiation, the light field was focused on the tumor with a 5–8 mm margin, and all other regions were shielded with lead.

For tumor-draining LN identification, the tumor-draining LN was initially localized using a contrast micro-CT-scan as well as LN mapping. For the LN mapping, 5% Evans Blue was injected intratumorally or into the right flank of a naïve mouse. Mice were sacrificed at 5 min and 30 min following injection, and at necropsy, blue LN were identified.

For tumor-draining LN irradiation, mice were set up clinically based on anatomic and CT-scan-guided landmarks. To minimize scatter to non-ipsilateral tumor-draining LN sites, all regions including the ipsilateral tumor were lead shielded. To confirm minimal scatter to the ipsilateral tumor and the correct dose to the tumor-draining LN, an optically stimulated luminescent dosimeter was implanted at both sites (Landauer, Glenwood, Illinois, USA). The prescribed dose to the tumor-draining LN were 3 Gy x three for a total of 9 Gy over 6 days. Mice were sacrificed according to the guidelines put forward by Emory University Institutional Animal Care and Use Committee.

For CD8 T-cell depletion, we injected 500 µg of αCD8 (Clone 2.43) injected IP on days 0, 7 and 14.

For FTY720 experiment, FTY720 (Sigma) was placed in drinking water (2 µg/mL) 1 day prior to tumor-directed RT (T-RT). FTY720 treatment was continued throughout the entire experimental course.

For T-cell analysis, mice were sacrificed on Day 17 when tumor, spleen and tumor-draining LN were harvested. CD8⁺ T-cells were purified from tumors using a histopaque gradient (Sigma, St. Louis, MO). Tumors were weighed at time of sacrifice and recorded in grams. For cytokine staining, tumor infiltrating lymphocytes were cultured in RPMI +10% FBS for 5 hours with an LCMV GP peptide pool in the presence of brefeldin A and monensin.

For tumor measurement, tumor diameters were assessed using calipers. Tumor volume was calculated using the formula for ellipse, that is, 4/3π.(l.w.h), where l, w and h represent the three diameters of the tumor taken perpendicular to each other.

Representative flow plots and tumor kinetics for figures 1, 4, 6-7 are taken from one experiment with all three experimental arms included. All experiments were done in duplicate or triplicate, and quantified data represent combined data unless otherwise noted.

Flow cytometric analysis was performed on an FACS Canto II or LSR II (BD Biosciences). Direct ex vivo staining and intracellular cytokine staining were performed as described previously23 with fluorochrome-conjugated antibodies (purchased from BD Bioscience, eBioscience, BioLegend, R&D, Cell Signaling Technology, Vector Laboratories and Invitrogen). To detect tumor-specific CD8⁺ T-cell responses, MHC-I tetramers were prepared as described previously.24 For intracellular detection of transcription factors such as T-cell factor-1 (TCF-1), cells were surface stained for 45 min on ice, fixed and permeabilized using the Foxp3 Permeabilization/Fixation Kit according to manufacturer’s instructions (eBioscience), followed by intracellular staining for 45 min. All staining was performed in a 96 well plate using 1–5 x 10⁶ total cells per well. Following staining, a minimum of 10⁴ CD8⁺ T-cells were collected by flow cytometer. Splenocytes were used for single color controls. TCF-1 gating was determined based on naïve CD8⁺ T-cell expression. Full gating strategy can be seen in (online supplemental figure 1). FACS data were analyzed with FlowJo software (TreeStar). The following antibodies were used: CD8a (Clone 53–6.7, BD), CD4 (Clone GK1.5, BD), CD44 (Clone IM7, BD), PD-1 (Clone 29F.1A12, Biolegend), IFN-γ (Clone XMG1.2, Biolegend), IL-2 (Clone JES6-5H4, BD), TNF-α (Clone MP6-XT22, BD), TCF-1 (S33-966, BD), Tim-3 (Clone RMT3-23, Biolegend), GzmB (Clone GB11, BD), Ki-67 (Clone B56, BD), FoxP3 (Clone R16-715, BD).

All experiments were analyzed using Prism 6 (GraphPad Software). Statistical differences were assessed using a Student’s t-test or as detailed in the figure legend. P values of <0.05 were considered statistically significant (*).

To assess whether tumor-directed RT (T-RT) alone can improve distant tumor control, we implanted syngeneic B16F10GP tumor cells into the bilateral flanks of immunocompetent mice. The use of the B16F10GP cell line expressing LCMV GP allows us to identify and track tumor-specific T-cells with MHC-I tetramers in various organs. In mice treated with T-RT, the tumor received 10 Gy x 1 with a 0.5 cm bolus to ensure appropriate surface dose (figure 1A). Following T-RT, we observed slowing of tumor growth of the irradiated (tumor 1) (figure 1B,C top) and a reproducible abscopal effect at the unirradiated site (tumor 2) (figure 1B C bottom). To characterize and quantify the intraumoral T-cell response, we evaluated tumor-specific (Gp33⁺) CD8⁺ T-cells. Our gating strategy is shown in online supplemental figure 1. There was an increase in the number of tumor-specific CD8⁺ T-cells per gram in the unirradiated tumor (tumor 2) with a trend towards an increase in the irradiated tumor (Tumor 1) (figure 1D, online supplemental figure 2A). Increased proliferation of tumor-specific CD8⁺ T-cells in tumor 1 was also observed (figure 1D, right). Both tumor 1 and tumor 2 had an increased frequency and tumor 1 showed increased number with a trend for tumor 2 of IFN-γ⁺ CD8⁺ T-cells following T-RT (figure 1E, left, middle). There was also an increased frequency of polyfunctional CD8⁺ T-cells in Tumor 2 (figure 1E, right) which represent a less exhausted T-cell subset. To verify that the observed effect was CD8⁺ T-cell dependent, we depleted CD8⁺ T-cells prior to tumor irradiation (figure 2A,E). This significantly reduced the abscopal effect (figure 2B,C,D).

To determine whether this abscopal response is dependent on the strong antigen (Gp33), we performed the same experiment with the B16F10 parental line, and we observed improved local control with tumor irradiation, but no abscopal effect (online supplemental figure 2B) with minimal increase in the density of CD8+PD-1+T cells in the unirradiated tumor (online supplemental figure 2C).

We then performed a dose escalation experiment from 0, 10 and 20 Gy to assess whether increasing dose would increase the magnitude of the abscopal effect using B16F10GP tumors. We found that both 10 and 20 Gy had similar local control (figure 2F,G top) and distant control (figure 2F,G bottom). We also found a similar trend towards increased tumor-specific CD8⁺ T-cells per gram in the 10 and 20 Gy groups with a trend towards a higher T-cell frequency in the 10 Gy cohort (figure 2H).

Next, we wished to characterize the tumor-specific T-cell population in the tumor and tumor-draining LN in more detail. We found that the majority (>85%) of the Gp33⁺ T-cells in the tumor were terminally differentiated effectors (figure 3A). Given the importance of stem-like CD8⁺ T-cells (Tim-3^- TCF-1⁺) in response to αPD-L1,21 we next evaluated the stem-like CD8⁺ T-cell population in the tumor. After gating on the tumor-specific CD8⁺ T-cells, we found CD8⁺ T-cells with a stem-like phenotype in the tumor although at low frequencies (figure 3A, top). In the chronic LCMV infection, stem-like CD8⁺ T-cells were found at greatest frequencies in secondary lymphoid organs, therefore, we next evaluated the tumor-draining LN. In the LN, we found tumor-specific CD8⁺ T-cells with intermediate expression levels of PD-1 (figure 3A, bottom left). After gating on the tumor-specific CD8⁺ T-cells, we found that >75% of these cells have a stem-like phenotype (figure 3A, bottom middle). We confirmed that the GzmB expression was much higher in the terminally differentiated effectors both in the tumor and LN than the stem-like CD8⁺ T-cells as has been shown previously21 (figure 3A, right). Finally, we quantified the frequency differences between the stem-like CD8⁺ T-cells in both the LN and tumor and the terminally differentiated effectors (figure 3B). These data clearly show that the frequency of stem-like CD8⁺ T-cells is dramatically higher in the LN than the tumor. A similar difference in stem-like T-cell frequency between the tumor and LN was observed in the parental B16F10 cell line (online supplemental figure 2D).

Next, we evaluate the impact of tumor irradiation on the T-cell subsets in both the tumor and the tumor-draining LN. We found that T-RT increased the number of stem-like CD8⁺ T-cells, but not their frequency in tumor 1 with a similar trend for tumor 2 and for terminal effectors (figure 4A). In the parental B16F10 cell line, no significant increase in stem-like CD8⁺ T-cells was observed following T-RT in tumor 1 or 2 (data not shown).

We found no statistically significant or trends in changes in the frequency of stem-like or terminal effectors in the LN 7 days following T-RT (figure 4B). We did, however, see an increase in the frequency of proliferating tumor-specific CD8⁺ T-cells (figure 4C, left and middle), stem-like CD8⁺ T-cells (figure 4C, right) and a more modest increase in terminal effectors (online supplemental figure 2E). As this proliferative response suggests the LN may play a role in the abscopal effect, we next focused our attention on the LN.

To evaluate the importance of the tumor-draining LN in mediating the abscopal effect, we depleted tumor-draining LN lymphocytes by targeted RT (LN-RT). To first confirm that we were able to identify the first echelon tumor-draining LN anatomically, we injected Evans Blue at the site of flank tumor inoculation and sacrificed the mice at 5 min postinjection. At that time, the only LN that demonstrated visible uptake of the dye was the ipsilateral inguinal node (figure 5A). To confirm that this node could be accurately targeted with kilovoltage X-rays, we acquired a contrasted micro CT (figure 5B) and performed dosimetric calculations based on planned field size, and depth of the inguinal nodal basin. We then confirmed prescribed dose delivery by dosimeter implantation at the LN site and tumor site demonstrating minimal scatter to the tumor with appropriate shielding (not shown). As a further confirmation, we evaluated for lymphocyte death in the targeted node following RT by sacrificing mice 36 hours after the first RT fraction (figure 5C) and performing Annexin-V and 7-AAD staining (figure 5D). We observed increased cell death in the irradiated node (figure 5D). To confirm that there was actual depletion of T-cells from the targeted tumor-draining LN, mice were sacrificed on day 7 following 3 Gy x 3 fractions to the CT-assessed location. Fractionated LN-RT was selected after 10 Gy x 1 showed no depletion of lymphocytes in the irradiated LN at day 7 post-RT (data not shown). This was likely due to rapid LN repopulation from the blood. The inguinal LN in the prescribed radiation field demonstrated a consistent two log decrease in the number of CD4⁺ and CD8⁺ T-cells (figure 5E). No concomitant decrease in the number of splenic lymphocytes was observed (figure 5F) demonstrating that the radiation is specifically targeting the LN and not causing a global decrease in the circulating lymphocyte population. We next performed a time course and found the nadir of the LN T-cells was at day 10 following the first radiation fraction (figure 5G). However, day 7 was selected for sacrifice as there was a significant decrease in LN cell numbers at this time point and it represented the apex of the T-cell infiltration into the tumor.

Next, we evaluated the impact of LN irradiation on tumor growth. We irradiated both the tumor-draining LN of tumor 1 with fractionated RT (3 Gy x 3 fractions) and tumor 1 with the single 10 Gy dose (figure 6). Irradiating the tumor-draining LN and the tumor (T-RT+LN RT) did not significantly impair local tumor control (figure 6B,C top), however, there was a reduction in the abscopal effect (figure 6B,C bottom).

For T-cell analysis, mice were sacrificed 1 day following the last fraction of tumor-draining LN-directed RT (day 17) (figure 6A). The increase in tumor-specific CD8⁺ T-cells seen in tumor 2 (unirradiated tumor) with tumor 1-directed RT was decreased in the T-RT+LN RT group (figure 6D, right). Tumor-draining LN-directed RT also reduced the frequency of proliferating tumor-specific CD8⁺ T-cells in the irradiated tumor (Tumor 1) (figure 6D, bottom). Frequencies of both IFN-γ⁺ CD8⁺ T-cells and IFN-γ⁺ TNF-α⁺ CD8⁺ T-cells were reduced in tumor 2 in the T-RT+LN RT compared with the T-RT group (figure 6E).

We next evaluated whether tumor-draining LN-directed RT reduced the stem-like CD8⁺ T-cell numbers in the LN. We found a significant decrease in the T-RT+LN RT group in the irradiated (tumor 1 side) as well as the un-irradiated (tumor 2 side) tumor-draining LN (figure 7A, left). The magnitude of the decrease was much greater in the irradiated compared with the unirradiated tumor-draining LN. Next, we looked in the tumor and found that in both tumor 1 and tumor 2 of the T-RT+LN RT group there was a reduction in stem-like CD8⁺ T-cell numbers relative to the T-RT group (figure 7B, left). There was similar trend towards a reduction for terminally differentiated effectors in Tumor 1, but not for tumor 2 (figure 7B, right). No significant difference was observed in the frequency of tumor-infiltrating regulatory T-cells with T-RT or T-RT+LN RT in either tumor 1 or tumor 2 (online supplemental figure 2F).

To further confirm the importance of the LN in the abscopal response, mice were treated with FTY720, which sequesters lymphocytes in secondary lymphoid tissue, prior to T-RT (online supplemental figure 3A,B). A significant decrease in the abscopal effect at tumor 2 was observed with T-RT+FTY720 (online supplemental figure 3C,D bottom) relative to T-RT alone.