Lewis lung carcinoma (LLC) cells were obtained from National Collection of Authenticated Cell Cultures in China (Shanghai, China) and cultured in DMEM (Dulbecco’s Modified Eagle Medium, Gibco, MA, USA) supplemented with 10% FBS (Fetal Bovin Serum, Gibco) and 100 U/mL Penicillin and 100 μg/mL Streptomycin (Gibco).

All animal experiments were conducted in accordance with the Animal Welfare Committee at Guangzhou Traditional Chinese Medical University. C57BL/6J male mice weighing approximately 20 g were purchased from Vitalstar Biotechnology (Beijing, China) and maintained on a regular dark-light cycle with free access to food and water. LLC cells were trypsinized and washed with Dulbecco’s phosphate-buffered saline (DPBS, Gibco). After centrifugation at 1000 g for 5 min at room temperature, cells were collected and diluted to a final concentration of 1 X 10⁷/mL. The right flanks of mice were subcutaneously inoculated with 1 million LLC cells.

Mice were randomly divided into four groups 10 days after inoculation, and the tumors were radiation treated with 6 MeV electron rays in different split doses under 2.5% isoflurane. The control group mice received no radiation. The 3.6 Gy group mice were treated with 3.6 Gy (Gray) X 21 times, totaling approximately 103 Gy biological effective dose (BED). The 7.2 Gy Group mice were treated with 7.2 Gy X 9 times (BED≈110 Gy). The 14.4 Gy group mice were treated with 14.4 Gy X 3 times (BED≈105 Gy). All the radiation treatments were conducted by using Elekta Infinity linear accelerator (Elekta, Stockholm, Sweden). Forty-eight hours after radiation, the mice were sacrificed under anesthesia, and tumors were collected, weighted, and frozen immediately in liquid nitrogen.

Twenty milligrams of tumors from each sample were grinded in liquid nitrogen, and lipids were extracted in a buffer containing tert-Butyl methyl ether (MTBE, CNW technologies, Germany) and methanol (MeOH, CNW technologies) (v: v = 5: 1) and SPLASH LIPIDOMIX Mass Spec standard (Avanti Polar Lipids, AL, USA). After a brief sonication, samples were centrifuged at 3000 rpm for 15 min in 4 °C. The supernatant was collected, and an equal volume of MTBE was added. Sonication was then repeated twice, and the supernatant was dried by using a vacuum dryer. Samples were then reconstituted under 10 min sonication in a solution containing Dichloromethane (DCM, CNW technologies), MeOH, and water (v: v: v = 60: 30: 4.5). The quality control (QC) sample was prepared by mixing a small and equal aliquots from all samples together. All procedures were carried out on ice.

Samples were analyzed by using an ACQUITY Premier (Waters, MA, USA) ultra-high performance liquid chromatography (UHPLC) system coupled with a SCIEX Triple Quad™ 6500+ mass spectrometer (SCIEX, MA, USA). An ACQUITY UPLC HSS T3 column (Waters) was utilized for UHPLC separation. The mobile phase A consisted of 40% water and 60% acetonitrile, supplemented with 10 mM/L ammonium formate. Mobile phase B consisted of 10% acetonitrile, 90% isopropanol, and 10 mM/L ammonium formate. The elution gradient was set as follows: 80% A at 0min, 40% A at 4min, 2% A at 14 min, 80% A at 16.01 min, and 80% A at 18 min. The flow rate was 0.3 mL/min. The column temperature was maintained at 45°C. The temperature of auto-sampler was set to 10°C. The injection volume was 2 μL per sample.

The ion source parameters for Triple QuadTM 6500+ mass spectrometer were as follows: ion spray voltage: +5500/-4500 V, curtain gas: 40 psi, temperature: 350°C, ion source gas 1: 50 psi, ion source gas 2: 50 psi, and DP: ±80 V.

To identify and quantify compounds, Biobud software (v2.1.4.1, Biotree, Shanghai, China) was used. All peaks were extracted and the peak area was calculated. Peaks with a relative standard deviation (RSD) > 30% across all samples were filtered out. Peaks with more than 50% empty values were further removed. Missing values were filled by multiplying the minimum value by a random number between 0.1 and 0.5. After all these steps, 700 peaks were retained.

Total RNAs of tumor samples were prepared by using Trizol reagent (Thermofisher Scientific, MA, USA). 1 μg RNAs from each sample were reverse transcribed to cDNA by using SuperScript™ IV reverse transcriptase (Thermofisher Scientific). The levels of glycerophospholipid metabolism related genes were tested by using SYBR green Real-time PCR mix (Thermofisher Scientific) in QuantStudio™ 5 system (Thermofisher Scientific). The mRNA expression level was quantified and normalized to that of GAPDH by using 2^-ΔΔCT (17). The primers used were as follows: Agpat3, forward: 5’-CTGCTTGCCTACCTGAAGACC-3’, reverse: 5’-GATACGGCGGTATAGGTGCTT-3’; Plpp1, forward: 5’-AGTCTCAGCTAGTCAGTCCTTGA, reverse: GGCTTGAAGATAAAGTGCGACAA; Pld1, forward: 5’-TCGTTTTGTGGACTGAGAACAC-3’, reverse: 5’-GCTGCTGTTGAAACCCAAATC-3’; Scd4, forward: 5’-GCCCACTTGCCACAAGAGAT-3’, reverse: 5’-GTAGCTGGGGTCATACAGATCA-3’; Gpd1, forward: 5’-ATGGCTGGCAAGAAAGTCTG-3’, reverse: 5’-CGTGCTGAGTGTTGATGATCT-3’; Gpcpd1, forward: 5’-TGCCAACACAGGGATGGAGTA-3’, reverse: 5’-TGCTTCTGCCGAACCATTGTA-3’.

Statistical analysis was performed by using IBM SPSS statistics (IBM, NY, USA), Microsoft Excel (Microsoft, WA, USA) and the R language (R Foundation, Vienna, Austria). The lipid expression profile was depicted by principal component analysis (PCA) and partial least-squares discriminant analysis (PLS-DA). Significantly changed lipids were identified by orthogonal projections to latent structure-discriminant analysis (OPLS-DA) and student’s t test. Lipids with a variable importance in projection (VIP) > 1 in OPLS-DA analysis and p-value < 0.05 in student’s t test were considered as significantly changed lipids. Significantly altered lipid classes in response to radiation were assessed by Fisher’s exact test. For multi-group comparisons, one-way ANOVA or two-way ANOVA tests were used. For the relationship between glycerophospholipid unsaturation and tumor weight, linear regression analysis was applied. A p-value < 0.05 was considered significant, and data were presented as mean ± standard error (mean ± SE).

To determine the influence of radiation on tumor growth, the right flanks of mice were injected with LLC cells. All mice started to develop tumors at about one week after inoculation. The mice were randomly divided into four groups: control, 3.6 Gy, 7.2 Gy and 14.4 Gy, which were administrated with different split doses of radiation starting 10 days after injection. Except for the control group, which didn’t receive radiation, all other groups received similar biological effective doses (BED) ( Figure 1A ). Two days after all radiation doses were fully administered, the mice were sacrificed and the tumors were collected. The weight of tumors that received radiation was significantly reduced compared to control mice ( Figure 1B ). This reduction is mostly dependent on the BED the tumors received and is not related to the split dose, as there is no significant difference in tumor weight among the 3.6 Gy, 7.2 Gy and 14.4 Gy groups ( Figure 1B ).

Few studies have reported the impact of radiation on cancer cell lipid metabolism. Therefore, we performed a LC-MS lipid profiling study in LLC tumors to uncover the global lipidome alterations after radiation. A total of 700 lipid species were identified and quantified. PCA analysis, based on all identified lipids, showed a clear separation between the control group and all radiation groups. However, no obvious separation among the three radiation groups was observed ( Figure 2A ). To confirm this, we further conducted PLS-DA analysis, and similar results were obtained ( Figure 2B ). These data suggest that the lipid expression profile was significantly changed in tumor cells in response to radiation, and it was not related to the split doses administered. We thus focused on the differential lipid expression analysis in radiation groups against control group. We found that radiation induced significant expression changes of many lipids (VIP > 1 and p-value < 0.05), regardless of the split doses ( Figure 2C ). These observations indicate a global impact of radiation on tumor cell lipid expression.

According to lipid classification criteria, all the 700 lipid species were divided into several classes: fatty acyls (FA), glycerolipids (GL), glycerophospholipids (GPL), sterol lipids (SL) and sphingolipids (SPL). To compare the changes in lipid class, the abundance of all lipid species in each class were summarized. Statistical analysis revealed that the level of total SL was significantly elevated in response to radiation in the 3.6 Gy, 7.2 Gy, and 14.4 Gy groups ( Figure 3A ). In contrast, the level of total GL was significantly reduced in the 7.2 Gy and 14.4 Gy groups ( Figure 3A ).

To further understand the influences of radiation on lipid metabolism, all lipids were further divided into sub-classes. For example, GL was divided into diglyceride (DAG), triglyceride (TAG), monogalactosyl diglyceride (MGDG), and digalactosyl diglyceride (DGDG). Interestingly, there were no alterations in DAG, MGDG, and DGDG, but the total level of TAG was significantly decreased in response to radiation ( Figure 3B ). Together with the increase of total cholesterol esters (CE) ( Figure 3B ), these data demonstrate the impact of radiation on lipid storage in tumor cells.

Among all of the identified lipids, GL accounts for majority, which is about 48.43% (339 out of 700). GPL accounts for 36.86% (258 out of 700) ( Figure 4A ). Radiation treatment altered the abundance of many lipids. Specifically, 114 lipids were significantly changed in the 3.6 Gy group, 95 lipids in the 7.2 Gy group, and 106 lipids in the 14.4 Gy group ( Figure 2C ). Among all significantly changed lipids, GPL accounted for a large portion in all three radiation groups: 51.75% in the 3.6 Gy group, 67.37% in the 7.2 Gy group, and 66.98% in the 14.4 Gy group ( Figures 4B–D ). Further analysis using Fisher’s exact test revealed a significant enrichment of GPL among the significantly altered lipids in all three radiation groups ( Figure 4E ), indicating a preferential effect of radiation on GPL metabolism.

We also noticed that very small portion of GLs was significantly changed after radiation ( Figure 4E ), suggesting minimal influence of radiation on GL metabolism.

The composition and properties of GPLs are important for cell membrane structure and transmembrane signaling. Thus, we analyzed the property alterations of GPLs in response to radiation. There were no significant changes in total GPL in any of the three radiation groups ( Figure 5A ). However, we observed a decrease of long acyl chain-containing GPLs after radiation, particularly GPLs containing 36 carbons in the 7.2 Gy and 14.4 Gy groups ( Figure 5B ). Conversely, 38 carbons containing GPLs was observed significantly increased in the 3.6 Gy group ( Figure 5B ).

In addition, a significant increase in unsaturation in GPLs was observed in all three radiation treatment groups ( Figure 5C ). This is most likely due to the significant decrease in GPLs with one and two unsaturations and the significant increase in GPLs with four unsaturations ( Figure 5D ). To fully understand the GPL alterations, we depicted all the significantly changed GPLs by using a heatmap ( Figure 5E ). In the 3.6 Gy group, saturated GPLs were significantly decreased, and unsaturated GPLs were increased. Interestingly, in the 7.2 Gy group, GPLs containing more than 4 unsaturations were significantly increased, while others were significantly decreased. In the 14.4 Gy group, only GPLs containing more than 5 unsaturations were increased. These data suggest a differential influence of different split radiation doses on GPL acyl chain length and unsaturation.

Since radiation significantly altered tumor glycerophospholipid composition, we thus checked the expression of glycerophospholipid metabolism related genes by QPCR. We found that several genes were significantly changed ( Figure 6 ), including Agpat3, Plpp1, Gpcpd1, Gpd1, Pld1 and Scd4. Among them, Scd4 was found to be significantly reduced only in 14.4 Gy group, others were elevated either in two groups or in all three radiation groups. Interestingly, Pld1 and Gpd1 have be reported to be critical for cancer progression and metastasis (18, 19). These data further demonstrate the influence of radiation on glycerophospholipid metabolism.

Cell membrane structural alterations that induce membrane permeabilization is a key phenotype of cells undergoing death (20). Thus, we investigated whether GPL unsaturation alterations due to radiation are correlated with tumor cell death. We analyzed the relationship between tumor weight and the GPL unsaturation index by using linear regression. As shown in Figure 7 , GPL unsaturation is negatively correlated with tumor weight, with a correlation coefficient of -0.6894. These data suggest that radiation induced changes in GPL unsaturation may contribute to lung tumor cell death.