Galectin-1 mediates radiation-related lymphopenia and attenuates NSCLC radiation response

Gal-1's secretion is enhanced by radiation and induces T cell apoptosis in vitro

Gal-1 secretion increased significantly after RT in several different tumor cell lines, including two murine NSCLC cell lines. This increase of Gal-1 secretion was noted as early as three hours in LLC-1. Intracellular level of Gal-1 did not change, suggesting that the bulk of newly synthesized Gal-1 proteins were secreted after RT (Figure 1A). Similar to others (7, 9-11), we found that rGal-1 induced apoptosis of concanavalin A-activated lymphocytes isolated from mouse spleen in a dose dependent manner (Figure 1B). This effect was abrogated by a Gal-1 specific antibody (anti-Gal1) or Thiodigalactoside (TDG), a Gal-1 competitive inhibitor (Figure 1C) when T cells were exposed to Gal-1 containing conditioned media. In addition to T lymphocytes, rGal-1 can also cause apoptosis in CD19⁺ B cells (assessed by TUNEL) though to a lesser extent (data not shown). Gal-1 is expressed in memory B cells and has been shown to regulate apoptosis through Bcl2 signaling (11).

Tumor secreted Gal-1 reduces circulating helper and cytotoxic T cells after tumor irradiation

On the basis that radiation enhances the secretion of Gal-1, which has pro-apoptotic function in T cells, we evaluated the role of Gal-1 in RT-related lymphopenia. To address this, we examined the levels of circulating Gal-1 and T lymphocytes in the following four host and LLC-1 tumor combinations in order to differentiate between host and tumor Gal-1 contribution to radiation-related lymphopenia: WT/Scr (both host and tumor expressed Gal-1), WT/shGal-1 (only host expressed Gal-1), Gal-1^−/−/ Scr (only tumor expressed Gal-1) and Gal-1^−/−/ shGal-1 (neither host nor tumor expressed Gal-1). Once tumors reached a volume of approximately 150mm³, 20 Gy of radiation was delivered in a single fraction at the tumor site (Supplementary Figure 1).

Local tumor irradiation significantly increased the level of circulating Gal-1 only in Gal-1 expressing tumors (~1.5 in WT host and ~2.9 Gal-1^−/− host) in but not in Gal-1 down-regulated tumors (Figure 2A). These data indicated that RT increased Gal-1 secretion from the tumor that could be measured in the plasma. In parallel to the increase of plasma Gal-1, there was a significant decrease in circulating T cells in WT mice bearing Gal-1 secreting tumors (WT/Scr), 32.9% (P =0.026) for CD3⁺ mature T cells and 31.7% (P = 0.033) for CD3⁺8⁺ cytotoxic T cells, two weeks after tumor irradiation. There was also a decrease in CD4⁺ T helper cells but the reduction did not reach statistical significance (28.2%, P=0.09) (Figure 2, B-D). In contrast, CD4⁺ and CD8⁺ T lymphocyte levels did not decrease significantly after irradiation of shGal-1 tumors implanted in Gal-1 WT or Gal-1^−/− mice (WT/shGal-1 and Gal-1^−/− /shGal-1). Furthermore, we observed a significant decrease in the CD3⁺ and CD8⁺ T cells (29.4%, P = 0.02 and 34.4%, P = 0.02, respectively) but not CD4⁺ T cells (12.9%, P=0.48) in Gal-1^−/− mice bearing Gal-1 secreting tumors (Gal^−/−/Scr), suggesting that RT-related lymphopenia occurred independently of host Gal-1 status (Figure 2, B-D).

To ensure that the noted effect was due to Gal-1 secretion from RT rather than from Gal-1 secretion from growing tumors during the two weeks of observation, we also analyzed circulating T cells in WT mice, implanted with Gal-1 secreting tumors that were not irradiated and allowed to grow for two weeks after reaching approximately 150 mm³. Without tumor irradiation, there was a small decrease in the level of circulating CD3⁺, CD4⁺ and CD8⁺ T cells (18.1%, 18.05% and 6.60%, respectively) but none of the changes reached statistical significance (Figure 2E). Similarly, day 28 terminal bleed from mice in Figure 2E showed a two-fold higher plasma Gal-1 level in the irradiated group compared to the non-irradiated mice, despite larger tumors in the non-irradiated group, suggesting that RT induced Gal-1 secretion into the blood (Figure 2F). These data suggested that Gal-1 containing tumors did release a low basal level of Gal-1 into the blood that caused a small drop in circulating lymphocytes, and that this effect was significantly augmented by tumor irradiation.

We also evaluated the effect of Gal-1 secretion on common lymphoid progenitors or nesting mature T cells in the bone marrow. There was no change in these cell populations with radiation and there was no change in Gal-1 expression on the surface of these cells (Supplementary Figure 2). To ensure that there was no difference in inherent sensitivity to radiation or rGal-1 between lymphocytes from WT and Gal-1^−/− mice, we also tested these effects in vitro; no difference was noted between the two T cell populations (Supplementary Figure 3).

To corroborate our previous findings with the shGal-1 inhibition, we also evaluated the effect of inhibiting Gal-1 pharmacologically with the competitive inhibitor, TDG (Figure 2G). TDG is a naturally occurring thioglycoside that is highly stable in vivo, and binds to the carbohydrate recognition domain of Gal-1, thereby preventing ligand interactions. We performed intratumoral TDG injections in Gal-1 WT mice implanted with LLC-1 scramble tumors to inhibit Gal-1 and examined changes in the abundance of circulating CD3⁺, CD4⁺ and CD8⁺ lymphocytes before and after tumor irradiation. TDG treatment before RT resulted in a small, non-significant increase of total, CD4⁺ and CD8⁺ T cells, suggesting that basal Gal-1 release from the tumor caused some circulating T cell death, and this was blocked by TDG injection. More importantly, radiation resulted in a significant drop of circulating CD3⁺ (51.2%, P=0.006), CD4⁺ (42.2%, P=0.005) and CD8⁺ (43.5%, P=0.008) lymphocytes, which was rescued with TDG treatment. Altogether, our data suggest that Gal-1 was secreted into the blood by the tumor after radiation, resulting in lymphopenia that could be abrogated by Gal-1 inhibition.

Gal-1 modulates tumor radiation response through the modulation of tumor T cell apoptosis and angiogenesis

In light of our observation that Gal-1 mediates RT-related lymphopenia, which is associated with worse outcomes in the clinic, we hypothesized that Gal-1 reduced tumor response to RT and enhanced tumor metastasis and that these effects could be minimized with Gal-1 down regulation or inhibition. In vitro, Gal-1 down regulation in LKR-13 lung cancer cells moderately enhanced radiation cell kill under either normoxia or severe hypoxia, as assessed by clonogenic survival curves (Supplementary Figure 4, A and B). To investigate Gal-1 effect on tumor radiosensitivity in vivo, we implanted LLC-1 scramble and shGal-1 tumors in Gal-1 WT C57Bl/6 mice and irradiated the tumors with 20 Gy in a single fraction, after allowing the tumors to reach a volume of approximately 150 mm³. Tumor growth was monitored and upon sacrifice, lung metastases, intratumoral T cell infiltration and apoptosis, as well as, microvessel density count were quantified. As expected, either down regulation of Gal-1 alone or radiation alone resulted in similar decreases in tumor growth (Figure 3A). However, the combination of radiation and Gal-1 down regulation nearly abolished tumor growth (mean 1.16 ± 0.97 fold volume increase when normalized to tumor volume on day of irradiation) during the 2 weeks of observation (P=0.02) (Figure 3A). This observation was also made with LKR13 tumors in vivo (Supplementary Figure 4C). Radiation alone did not affect the number of spontaneous lung metastasis (Figure 3B). Down regulation of Gal-1 significantly decreased the number of lung metastases, which was further reduced with the addition of local field radiotherapy (P<0.05).

Similar results for tumor growth were observed when we inhibited Gal-1 with intratumoral TDG injection. TDG injection alone or radiation alone reduced tumor growth to a similar extent and the combination of radiation and TDG yielded the slowest growth rate (P=0.05)(Figure 3C). Again, tumor irradiation alone did not reduce lung metastases compared to the non-irradiated control (Figure 3D). Intratumoral TDG injection resulted in a non-significant reduction in the number of lung metastasis. However, the combination of TDG and radiation resulted in a significant decrease in lung metastases (approximately 40%) compared to control (P= 0.037).

Mechanistically, down regulation of Gal-1 increased CD8⁺ but not CD4⁺ T cell tumor infiltration (Figure 4, A and B). Gal-1 depletion combined with radiation increased tumor infiltration by CD8⁺ T cells even further (Figure 4B). Notably, Gal-1 knockdown reduced apoptosis of both intratumoral CD4⁺ and CD8⁺ T cell subsets (Figure 4C). Gal-1 inhibition by knockdown and TDG also reduced tumor angiogenesis, which was further attenuated when combined with radiation (Figure 4, D and E).

RT-related lymphopenia is associated with reduced survival in early stage NSCLC treated with hypofractionated radiation therapy

The study cohort consisted of 20 patients with stage I-II NSCLC treated definitively with small-field hypofractionated stereotactic ablative radiation therapy (12) alone. Supplementary Table 1 summarizes the patient characteristics. The majority of patients, 15 out of 20 (75%), experienced a drop in absolute lymphocyte count during the first year and the mean maximum decrease for the entire group was 0.4±0.22x10³/ml. Figure 5A shows the mean absolute lymphocyte counts before (pre-SABR) and after SABR (post-SABR), when the counts were at their lowest within the first year. The decrease was statistically significant (P = 0.025). Univariate cox proportional hazards analysis showed that the degree of drop in absolute lymphocyte count (split by quartile) was associated with worse overall survival (OS, P=0.037, HR=1.148) (Figure 5B), disease free survival (DFS, P=0.033, HR=1.134) (Figure 5C), distant progression free survival (DPFS, P=0.023, HR=1.159) (Supplementary Figure 5A) and local progression free survival (LPFS, P=0.049, HR=1.129) (Supplementary Figure 5B). It was also associated with a trend for worse regional progression free survival (RPFS, P=0.056, HR=1.126) (Supplementary Figure 5C). These findings revealed that RT-related lymphopenia also occurred in SABR using very small fields in one to four fractions and was prognostic for tumor control and survival in early stage NSCLC.

Since plasma samples from these 20 NSCLC patients were not available for Gal-1 measurement, we measured circulating Gal-1 and T lymphocytes in a group of head and neck cancer (HNC) patients treated with radiotherapy with either concomitant cetuximab or cisplatin (N=24). Patient characteristics for this group are shown in Supplementary Table 2. We observed a significant decrease in lymphocyte count (0.88 ±0.18x10³/ml, P<0.001) and increase in plasma Gal-1 (5.89±4.3 ng/ml, P=0.05) after treatment (Figure 5, D and E). The degree of RT-induced lymphopenia was not significantly different between HNC treated with RT combined with cisplatin or cetuximab (Supplementary Figure 5D). To date, we only noted three failures in the HNC patients with a median follow up of 11.2 months; such few number of events precludes accurate survival time analysis. However, there was a trend for larger lymphocyte drop (1.2 versus 0.8 x10³/ml) and a higher rise in Gal-1 level (14.1 versus 4.7 ng/ml) in patients who failed compared to those who did not, although the difference did not reach statistical significance due to the small number of patients (Supplementary Figure 5E).

Our findings indicate that in addition to proangiogenic function, Gal-1-mediated intratumoral T cell apoptosis and peripheral T cell depletion, which increase after radiation, facilitate radiation resistance and poor prognosis with SABR treated NSCLC. Figure 6 summarizes our proposed mechanism.

Cell lines and generation of stable Gal-1 knockdown

Lewis lung carcinoma (LLC1),was obtained from American Type Culture Collection (ATCC) (Manassas, VA). LKR13 murine lung adenocarcinoma cell line derived from Kras^LA1 mice (39), was a generous gift from Alejandro Sweet-Cordero. Cells were cultured in DMEM (LLC-1) or RPMI (LKR13) with 10% fetal calf serum, penicillin (100U/ml), streptomycin (0.1mg/ml) and at 37°C and 5% CO2 in air. Cell lines were not independently validated in our laboratory. To generate stable Gal-1 knockdowns, LLC-1 and LKR13 cells were subject to lentivirus transduction with scrambled or Gal-1 pLKo.1 shRNAs, CCTAAGGTTAAGTCGCCCTC GCTCGAGCGAGG GCGACTTAACCTTAGG or CCGGCCTACACTTCAATCCTCGCTTCTCGAGAAGCGAGGATTGAAGTGTAG GTTTTT (Thermoscientific Open Biosystems) as previously described (9).

Western blotting

Lysates were probed with goat anti-Gal-1 primary (1:1000, R&D Systems, Minneapolis, MN), secondary anti-goat-HRP (1:10000, Zymed, South San Francisco, CA), mouse anti-β-actin primary (1:1000, Sigma-Aldrich, St. Louis, MI), and anti-mouse-HRP secondary (1:10000, Invitrogen, Carlsbad, CA) antibodies. Secreted Gal-1 in serum-free media was collected from LLC-1 and LKR13 Scr and shGal-1 cells, concentrated (EMD Millipore, Billerica, MA) and volumes were normalized by cell number.

Mice and Tumor models

C57BL/6 mice were obtained from Jackson Laboratory and bred with Lgals1 null (Gal-1^−/−) mice from the Consortium for Functional Glycomics (Scripps Research Institute) to generate WT littermates (13). Genotyping by PCR WT forward primer: GACCCCATCCCTACACCCCAG; Gal-1 null forward primer: CTATCAGGACATAGCGTTGG; and common antisense primer: AAACTTCAGCCGGGAGAAAGG (9). WT and Gal-1^−/− mice (8-10 weeks old) received subcutaneous inoculation of LLC scramble or shGal-1 cells (2 x 10⁵) on the flank. Once tumors reached approximately 150 mm³, mice were placed in lead shielding jigs and exposed tumors received 20 Gy irradiation using a 250k Vp orthovoltage therapeutic X-ray machine (Philips, Andover, MA). Tumors were caliper measured and volume was calculated as ((width)² x length)/2.

Thiodigalactoside treatment

Thiodigalactoside (TDG) was injected intratumorally in mice (120mg/kg) one week before radiation and every 2-3 days thereafter.

Isolation of Splenocyte and whole blood T cells

Splenocytes were isolated from spleen of 8-10-week old C57BL/6 WT and Gal-1^−/− mice. After filtration and red blood cell lysis (Sigma, St. Louis, MO), splenocytes underwent density centrifugation (Cedarlane, Ontario, Canada) and resuspension at 1-2X10⁶ cells/ml in RPMI containing 5ug/ml conconavalin A (Sigma, St. Louis MO),55uM betamercaptoethanol (Gibco, Grand Island, NY) and Dithiolthreitol (1.1mM). Peripheral blood from retro-orbital eye bleed was collected in EDTA tubes and RBCs were lysed (Sigma Aldrich, St. Louis, MI).

In vitro T cell apoptosis and TUNEL staining

Lymphocytes were plated at 1-2x10⁶ cells/ml for 24 hours followed by treatment with rGal-1 (10ug/ml) (Sigma, St. Louis, MO) for 24 to 48 hours. TUNEL-FITC staining was performed using the Apo-Direct Kit (BD Biosciences, San Jose, CA).

Flow cytometry

Lymphocyte subsets were stained with Pacific Blue anti-mouse CD3e rat IgG (clone 500A2), PE anti-mouse CD4 rat IgG (clone RM4-5), and Alexa Fluor 647 anti-mouse CD8a rat IgG (clone 53-6.7), at 4°C for 45 minutes in staining buffer (BD Biosciences San Jose, CA). Cell acquisition was performed with FACSDiva software on a LSR II flow cytometer (BD Biosciences, San Jose, CA) and analyzed with Flowjo software (Tree Star Inc., Ashland, OR).

Immunohistochemistry

Lung metastasis was quantified by H&E staining (15). Two sections 100μm apart were used for metastasis quantification, with at least five mice per group. Fixed sections were stained with goat anti-CD31 (1:100, clone M-20) (Santa Cruz Biotechnology Inc., Santa Cruz, CA) followed by secondary anti-goat detection by DAB (Vector Laboratories, Burlingame, CA) (9).

Immunofluorescence

Frozen tissue sections were incubated with CD4 and CD8 primary (1:100, BD Biosciences, San Jose CA) and Alexa Fluor 594 anti-rat secondary antibodies (1:200, Invitrogen, Carlsbad, CA) (13). Slides were labeled with TUNEL-FITC (Roche Applied Science, Indianapolis, IN) and mounted in DAPI (Vector Laboratories, Burlingame, CA). Immunofluorescence images were acquired using a Zeiss LSM510 laser scanning confocal microscope, using 40x/0.95 NA and 63x/1.4 NA Plan Apochromat objectives.

Galectin-1 ELISA

Whole blood from retro-orbital bleeds were collected in BD Microtainer tubes with EDTA and centrifuged at 3,000 rpm for 10 minutes. Plasma was diluted and subject to mouse Galectin-1 ELISA per manufacturer instructions (R&D Systems, Minneapolis, MN).

Patient population

Early-stage NSCLC patients included in our analysis were diagnosed between 1999 and 2013 with inoperable disease, received only SABR (no chemotherapy) and had pre- and post-radiation lymphocyte values up to 38 months after treatment. Head and neck cancer patients in our study were diagnosed between 2007 and 2013 and received radiation with cetuximab or cisplatin. Because some patients had multiple post-radiation lymphocyte counts, we calculated the maximum lymphocyte change as difference between the pre-RT value and the lowest post-RT value.

Statistical analysis

Patient survival analysis was performed using Cox proportional hazards model with the maximum drop in absolute lymphocyte counts between before and after radiotherapy as the only predictor. Four quartiles of the maximum drop in absolute lymphocyte counts before and after radiotherapy were plotted for the indicated endpoints. Overall survival (OS) was defined as time from treatment date to death. Distant progression-free survival (DPFS) was defined as time from treatment to documented first distant relapse or death, local progression-free survival (LPFS) as time from treatment to documented first local relapse or death, and regional progression-free survival (RPFS) as time from treatment to documented first regional nodal relapse or death. In all other cases, quantifications were subject to two-tailed, unpaired Student's t test to analyze statistical differences between groups. Data are expressed as mean ± SE. A p-value of <0.05 was considered significant.

Declaration of ethical approval for experiments

Patient data collection was approved by the Stanford Institutional Review Board (IRB protocol 13535). Written informed consent was received from participants prior to inclusion in our study. Animals were housed in pathogen-free conditions and all procedures were conducted in accordance to the Institutional Animal Care and Use Committee at Stanford University (APLAC protocol 15106).