LCN2 secreted by tissue-infiltrating neutrophils induces the ferroptosis and wasting of adipose and muscle tissues in lung cancer cachexia

6–10 week-old male C57BL/6 (Cat# T002040), BALB/cJGpt-Foxn1^nu/Gpt mice (Cat# D000521) were obtained from GemPharmatech (Nanjing, China). The Lcn2^f/+ (Cat# NM-CKO-00134) and LysM^cre mice (Cat# NM-KI-18018) were obtained from ShangHai Model Organisms. Lcn2^f/f mice were crossed with LysM^cre mice to obtain Lcn2^f/f;LysM^cre mice. All mice used were housed under specific pathogen-free conditions. All procedures involving experimental animals were approved by the Ethics Committee of the University of Science and Technology of China and were performed in accordance with the National Guidelines for Animal Usage in Research.

The Lewis lung carcinoma (LLC) and 3T3-L1 cell lines were obtained from The Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Cat# TCM 7 and Cat#SCSP-5038, respectively). Both cell lines were cultured in DMEM supplemented with 10% FBS (HyClone), 1% streptomycin and penicillin, and were maintained at 37 °C and 5% CO₂. The cell lines were routinely tested for mycoplasma using the TransDetect PCR mycoplasma detection kit (Transgen, Cat#FM311).

2 days after the 3T3-L1 cells had reached 70% confluence, the cells were treated with 1 μM dexamethasone (Selleck, Cat#S1322), 5 μg/mL insulin (Novoprotein, Cat#P05019), 0.5 μM isobutylmethylxanthine (IBMX) (Selleck, Cat#S5836), and 1 μM rosiglitazone (Selleck, Cat#S2556). Another 2 days later, the cells were treated with 5 μg/mL insulin and 1 μM rosiglitazone. Starting on day 6, the cells were cultured in DMEM supplemented with 10% FBS, 1% streptomycin and penicillin, and treated with 200 ng/mL recombinant mouse LCN2 protein (Novoprotein, Cat#P11672) for 24 h.

Serum samples from healthy donors and from lung cancer patients with and without cachexia were obtained from the First Affiliated Hospital of the University of Science and Technology of China. Patients were diagnosed with cachexia if they had lost > 5% of their body weight over the past 6 months or had a body mass index (BMI) < 20, plus had three of the following criteria: anorexia, decreased muscle strength, fatigue, low fat-free mass index, or abnormal biochemistry results, including increased levels of inflammatory markers (e.g., CRP and IL-6), anemia, or low serum albumin. All samples were collected with the informed consent of the patients, and the experiments were approved by the Ethics Committee of the University of Science and Technology of China (2020-research-36). Details relating to the patients’ cancer type and cachexia are listed in Additional file 4: Table S5.

LLC-induced cachexia model was established as previous study described [20]. In brief, the male C57BL/6 mice were subcutaneously inoculated with LLC cells (5 × 10⁶ per mouse). Mice were randomly divided into treatment groups while ensuring that the average body weight in each group was roughly the same. Mice which showed significant loss (> 20%) in body weight were defined as having cachexia. Then, they were euthanized and WAT and muscle tissues were harvested to further confirm the cachexia symptom. We comprehensively compared the cachexia phenotype of mice at 1,2,3 weeks and finally determined the analysis of samples on day 21.

For lung cancer patient-derived xenograft (PDX)-induced cachexia, whether or not cachexia occurs is dependent upon the source of the tissue from the tumor patient. We established based on the 3#-Ade PDX of a lung adenocarcinoma. For specific steps and methods, PDX tumors in cold Dulbecco’s Modified Eagle’s Medium (DMEM) were minced into 30–50 mm3 fragments, and each fragment was subcutaneously transplanted into the dorsal flank of 6- to 10-week-old male BALB/cJGpt-Foxn1^nu/Gpt mice. Body weight of these tumor-bearing mice were monitored regularly. Mice who showed significant loss (> 20%) in body weight were defined as having cachexia. Then, they were euthanized and WAT and muscle tissues were harvested to further confirm the cachexia symptom.

To deplete neutrophils, the male C57BL/6 mice were subcutaneously inoculated with LLC cells (5 × 10⁶ per mouse) on day 0 and then intraperitoneally injected on days − 1, 1, 4, 6, 8, 11, 13, 15, 17, and 20 with a 100 μg dose of an anti-Ly6G (Bio X Cell, Cat# BE0075-1; RRID:AB_1107721) or an isotype IgG (Bio X Cell, Cat# BP0090, RRID: AB_2891360).

To deplete LCN2, male C57BL/6 mice were subcutaneously inoculated with LLC cells (5 × 10⁶ per mouse) on day 0 and then were intraperitoneally injected on days 4, 7, 11, 14, 17, and 20 with a 50 μg dose of an anti-LCN2 (Novus Biologicals, Cat# AF1857, RRID:AB_355022) or an isotype IgG (Bio X Cell, Cat# BP0090, RRID: AB_2891360).

For deferoxamine (DFO) therapy, male C57BL/6 mice were subcutaneously inoculated with LLC cells (5 × 10⁶ per mouse) on day 0 and then intraperitoneally injected with 15 mg/kg DFO (Selleck, Cat#S5742) on days 4, 7, 10, 13, 16, and 19.

For liproxtatin-1 therapy, male C57BL/6 mice were subcutaneously inoculated with LLC cells (5 × 10⁶ per mouse) on day 0 and then intraperitoneally injected with 10 mg/kg liproxtatin-1 (Selleck, Cat#S7699) daily between days 1 and 20.

The pCDH-CMV-MCS-EF1-Lcn2 (Sangon Biotech) or pCDH-CMV-MCS-EF1 vectors (YouBio, Cat# VT1479) were extracted using the Endo-Free Plasmid DNA Mini Kit II (Omega, Cat#D6950) and co-transfected with the pRSV-Rev (YouBio, Cat# VT1445), pLP/VSVG (YouBio, Cat# VT1491), and pNL-GFP-RRE (YouBio) plasmids into 293 T cells using the lipofectamine (Invitrogen, Cat#11668019) transfection method. Supernatants containing the LCN2-expressing or control lentiviruses were collected 48 and 72 h later and centrifuged at 50,000 × g for 2 h at 4 °C to purify the virus. For overexpression of Lcn2, 2 × 10⁹ plaque-forming units (PFUs) of LCN2-expressing lentivirus were injected intravenously into C57BL/6 mice once per week.

Leukocytes were isolated from the epididymal and inguinal white adipose tissue (eWAT and iWAT, respectively), gastrocnemius skeletal muscle (Gast), bone marrow, and blood, as previously described [21, 22]. For intracellular staining, leukocytes were incubated with PMA (50 ng/mL), ionomycin (1 mg/mL), and monensin (10 ng/mL) for 4 h at 37 °C and 5% CO₂, followed by staining for surface markers for 30 min at 4 °C. Cells were fixed and then permeabilized with the Foxp3/Transcription Factor Staining Buffer and incubated with fluorescent antibodies for 30 min at 4 °C. Cells were acquired by flow cytometry (LSR II).

For analysis of human blood samples, blood from cachectic cancer patients was centrifuged and lysed using red blood cell lysis buffer. Cells were then stained for surface markers for 30 min at 4 °C. Cells were fixed and then permeabilized with the Foxp3/Transcription Factor Staining Buffer and incubated with the fluorescent antibodies for 30 min at 4 °C. Cells were acquired by flow cytometry (LSR II). Antibodies and related materials used in this study are listed in Additional file 1: Table S1. Data analysis was performed using FlowJo 10 software.

The concentrations of human IL-6, human CRP, human LCN2, and mouse LCN2 in cell culture supernatants and serum were measured by ELISA, according to the manufacturers’ instructions; the kits used are listed in Additional file 1: Table S1.

Tissues were fixed overnight with 10% neutral-buffered formalin, dehydrated, embedded in paraffin, and sectioned. The 4 μm slices were then stained with hematoxylin and eosin (H&E).

Tissues were collected as described above. Tissues were fixed overnight with 10% neutral-buffered formalin, dehydrated, embedded in paraffin, and sectioned into 4 μm slices. The slides were dewaxed, rehydrated, and subjected to heat-induced epitope retrieval, followed by incubation with 5% goat serum for 1 h at room temperature to block non-specific antibody binding. Next, the sections were incubated with a primary anti-LCN2/NGAL antibody (Abcam, Cat# ab216462) overnight at 4 °C and then with an Alexa-Fluor-546-conjugated goat anti-rabbit IgG (5 μg/mL; Invitrogen, Cat# A-11010, RRID:AB_2534077) secondary antibody. Nuclear staining was performed using DAPI (5 min incubation). The stained sections were imaged using the LSM 880 Confocal Microscope (Zeiss, Jena, Germany) and analyzed with Image J software.

The concentrations of iron (Fe²⁺) in eWAT, iWAT, and Gast tissues were measured using the Iron Assay Kit (Sigma Aldrich, Cat#MAK025). eWAT, iWAT, and Gast tissues were homogenized in Iron Assay buffer by centrifugation at 16,000 × g for 10 min at 4 °C. The samples were then incubated with an iron reducer in a 96-well plate for 30 min at room temperature, and finally with an iron probe for 60 min at room temperature. Absorbance at 593 nm was measured using a microplate reader.

Adipocyte lipid ROS production was measured using the Lipid Peroxidation Assay Kit (Abcam, Cat#ab243377). Adipocytes purified from eWAT or iWAT were stained with the Lipid Peroxidation Sensor for 30 min at 37 °C and analyzed immediately by flow cytometry (LSR II). The oxidized and non-oxidized lipids were detected on the FITC and PE channels, respectively. The FITC to PE mean fluorescence intensity (MFI) ratio was calculated for each sample. Data analysis was performed using FlowJo 10 software.

RNA was extracted from purified cells or frozen tissue samples using the TRIzol reagent (Invitrogen, Cat#15596018), glycogen (Thermofisher, Cat#AM9515), and sodium acetate (Thermofisher, Cat#R1181). RNA was reverse transcribed into cDNA using the M-MLV Reverse Transcriptase (Invitrogen, Cat#28025013). qPCR was then performed using the SYBR Green Premix Pro Taq HS qPCR Kit (Accurate biology, Cat#AG11701) on a LightCycler 96 instrument (Roche). PCR was performed using the 2 × TransTaq-T PCR SuperMix (Transgen, Cat#AS122). Relative mRNA levels were calculated using the 2^−ΔΔCt method and normalized to actin mRNA levels. Target gene primers are shown in Additional file 1: Table S2.

Total RNA was extracted from the eWAT of cachectic and control mice using the miRNeasy Mini Kit and treated with DEPC-treated water. A total amount of 1 µg RNA per sample was used as input material for the RNA sample preparations. RNA libraries were prepared for sequencing using the NEBNext UltraTM RNA Library Prep Kit (Illumina). Clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumina), according to the manufacturer’s instructions. After cluster generation, the library preparations were sequenced on an Illumina Novaseq platform and 150 bp paired-end reads were generated. Raw data (raw reads in fastq format) were firstly processed through in-house perl scripts. Clean data were obtained by removing reads containing adapter and poly-N sequences, and low-quality reads. The reference genome and gene model annotation files were downloaded from the genome website directly. The reference genome index was built using Hisat2 software, and paired-end clean reads were aligned to the reference genome. FeatureCounts software was used to count the read numbers mapped to each gene. Fragments per kilobase of exon model per million reads mapped (FPKM) were calculated for each gene based on the length of the gene and the read counts mapped to this gene. Differential expression analysis of two groups was performed using the DESeq2 R package (1.16.1). The resulting P-values were adjusted using the Benjamini and Hochberg’s approach to control for the false discovery rate. Genes with a DESeq2-derived adjusted P-value < 0.05 and log2fold change (FC) ≥ 2 were labelled as differentially expressed genes (DEGs; listed in Additional file 5: Table S6). The DEG heatmap was analyzed using MEV software Gene Ontology (GO) enrichment analysis of DEGs was implemented using the clusterProfiler R package. GO terms with a corrected P-value < 0.05 were considered significantly enriched DEGs. We used clusterProfiler R package to test the statistical enrichment of DEGs in various KEGG enrichment pathways. For the Gene Set Enrichment Analysis (GSEA), the genes were ranked according to the degree of differential expression in the two samples being compared. The predefined Gene Sets were then tested to see if they were enriched at the top or bottom of the list. The RNA-seq data generated in this study were deposited in the GEO database repository and can be accessed using with the accession number GSE188479.

The relative expression levels of human serum proteins in lung cancer patients with and without cachexia and in healthy donors were measured using the G-Series Human Cytokine Antibody Array 440 (RayBiotech). The relative expression levels of mouse serum proteins in cachectic and control mice were determined using the Quantibody® Mouse Cytokine Antibody Array 4000 (RayBiotech). Fluorescence signals were visualized in the Cy3 channel of a laser scanner. Data were extracted using the GAL file from www.​RayBiotech.​com/​Gal-Files.​html. Heatmaps of differentially expressed proteins (DEPs) were analyzed using MEV software.

Purified adipocytes, neutrophils, and macrophages were lysed in RIPA buffer (Beyotime, Cat#P0013B) containing protease inhibitors on ice for 30 min, followed by centrifugation at 16,000 × g for 10 min at 4 °C. Protein concentrations in the supernatants were measured using the BCA Protein Assay Kit (Thermofisher, Cat#23225), and 50 μg aliquots of proteins were incubated at 95 °C for 10 min and separated on SDS-PAGE gels. The proteins were transferred to PVDF membranes, which were blocked in a solution of 5% milk in TBS buffer containing 0.1% Tween-20 for 1 h at room temperature. The membranes were then incubated with primary antibodies overnight at 4 °C and subsequently with HRP-conjugated secondary antibodies for 1 h at room temperature. The protein bands were detected by chemiluminescence autoradiography. The antibodies used are listed in Additional file 1: Table S1.

All statistical analyses were performed using GraphPad Prism 5.0 (GraphPad Software, La Jolla, CA, USA). Results are presented as the mean ± standard error of the mean (SEM) and were compared using unpaired t-tests, one-way ANOVA, or two-way ANOVA. Mouse survival was estimated using the Kaplan–Meier method and compared using the log-rank test. Patient sample data were compared using the Wilcoxon signed rank test and Spearman’s rho was calculated as indicated.

We wished to identify the drivers of lung cancer cachexia in humans. Hence, we used a proteomics approach to evaluate the profiles of serum protein factors in cachectic lung cancer patients vs. non-cachectic lung cancer patients or healthy controls. This approach revealed aberrantly higher LCN2 levels in lung cancer patients with cachexia than in non-cachectic lung cancer patients or healthy controls (Fig. 1A, B). Levels of proinflammatory cytokines (e.g., IL-6 and CRP) were also higher in the serum of cachectic patients than in the other two groups, thereby indicating that cachexia was associated with a highly inflammatory environment (Fig. 1B and Additional file 1: Fig. S1A). Consistent with our data on the profiles of protein factors, ELISA results showed that serum concentrations of LCN2 were higher in cachectic lung cancer patients than in non-cachectic lung cancer patients or healthy controls (Fig. 1C). Moreover, we detected a negative correlation between the serum LCN2 concentration and the BMI and serum albumin level (Fig. 1D). We also detected a positive correlation between the serum LCN2 concentration and the IL-6 level and CRP level (Fig. 1D).

To pursue the basis of these clinical observations experimentally, we followed a previously reported [23] approach to establish a lung cancer PDX mouse model of cachexia (Fig. 1E, F). The tumor-tissue source of the PDX cachexia model was from a patient with lung adenocarcinoma (number: 3#-Ade) and all mice were able to develop cachexia. First, our ELISA data showed that the serum concentration of LCN2 was significantly higher in cachexia model mice than in control mice (i.e., mice without PDX-based induction of cachexia) (Fig. 1G). In addition, after sacrificing the mice, a qPCR analysis showed significantly increased Lcn2 expression in the tissues (eWAT, iWAT, and Gast) of cachexia model mice (Fig. 1H); eWAT, iWAT, and Gast undergo atrophy in cachexia [24, 25].

Subsequently, we induced a classic experimental model of murine cachexia [20] based on the subcutaneous inoculation of mice with LLC cells. As expected, 100% of LLC tumor-bearing mice developed cachexia symptoms. Over the 3-week observation period, tumor-bearing (i.e., cachectic) mice experienced significant weight loss vs. control mice (Fig. 1I). Upon sacrificing the mice, we detected wasting of eWAT, iWAT, and Gast in cachectic mice (Fig. 1J and Additional file 1: Fig. S1B). We also detected dystrophic pathological changes in the iWAT (Fig. 1L, top), eWAT (Fig. 1L, middle), and Gast (Fig. 1L, bottom) of tumor-bearing mice during cachexia progression.

A proteomics-based analysis of the profiles of serum protein factors in lung cancer cachectic mice showed significantly higher LCN2 levels in cachectic mice than in control animals (Fig. 1M, N). ELISA results further supported the significantly increased serum LCN2 concentrations in cachectic mice compared with those in control mice (Fig. 1O). In addition, we observed aberrantly increased LCN2 (mRNA and protein) levels during cachexia progression in the eWAT, iWAT, and Gast of cachectic mice (Fig. 1K and Additional file 1: Fig. S1E), as well as stronger LCN2 protein signals (immunofluorescence staining) in the iWAT and Gast of cachectic mice (Additional file 1: Fig. S1C, D). Collectively, these findings implied that the pathological changes which affected the wasting tissues of lung cancer cachexia models were related to the increased LCN2 level observed in lung cancer patients suffering from cachexia.

To assess the pathological transformations that occur in the microenvironment of local wasting tissue during lung cancer cachexia, we assessed changes in the wasting tissues of LLC-implanted cancer cachectic mice at the transcriptomic level. Among the 3,624 DEGs identified in our comparison of the eWAT from cachectic mice and control mice, 1770 showed upregulated expression and 1,854 had downregulated expression in cachectic mice (Fig. 2A). Enrichment analysis using the KEGG database revealed that the DEGs with upregulated expression were enriched in pathways such as “fatty acid degradation”, “inflammation”, and “ferroptosis” (Fig. 2B).

Ferroptosis is a form of cell death that results from iron-dependent accumulation of lipid peroxide and ROS production [26, 27]. GSEA showed enrichment of the ROS signature (Fig. 2D). Pursuing ferroptosis-related genes specifically, RNA-seq data and follow-up qPCR analysis showed that expression of Ncoa4, Slc39a8, Slc3a2, Pcbp2, Slc39a14, and Sat1 [28] was significantly higher in the wasting eWAT, iWAT, and Gast of cachectic mice than in the equivalent tissues in control mice (Fig. 2C, E). Ptgs2 expression [29] was significantly higher in the wasting eWAT of tumor-bearing mice than in the eWAT of control animals (Fig. 2C).

Tissues were harvested 21 days after tumor inoculation. We also assessed the known biochemical features of ferroptosis [26]. Levels of Fe²⁺, MDA, and lipid ROS were significantly higher in wasting eWAT, iWAT, and Gast than in the equivalent normal tissues (Fig. 2F, G and Additional file 1: Fig. S2B). Finally, immunofluorescence and flow-cytometry analyses of fresh adipocytes purified from wasting eWAT and normal eWAT showed that wasting eWAT exhibited increased lipid peroxidation (Additional file 1: Fig. S2C) and had a higher proportion of adipocytes (Additional file 1: Fig. S2A) than those in the eWAT of controls.

We also examined a lung cancer PDX-induced mouse model of cachexia, and found that Fe²⁺ and MDA levels were significantly higher in wasting eWAT, iWAT, and Gast than in normal tissues (Additional file 1: Fig. S2D, E). We also detected significantly higher expression of ferroptosis-related genes in the wasting eWAT, iWAT, and Gast of cachectic mice than in the equivalent normal tissues of control animals (Additional file 1: Fig. S2F). These results suggested that, in lung cancer cachexia, wasting tissues undergo ferroptosis.

To investigate the potential impact of the LCN2 level on ferroptosis, we conducted in vitro experiments with 3T3-L1 adipocytes [30]. Briefly, addition of the recombinant murine Lcn2 protein to 3T3-L1 cells increased the extent of cell death, lipid peroxidation (Fig. 3A and Additional file 1: Fig. S3E), and expression of ferroptosis-related genes (Ncoa4, Slc3a2, Slc39a8, Slc39a14, Pcbp2, Ptgs2, and Sat1) significantly (Fig. 3B and Additional file 1: Fig. S3H). We also treated 3T3-L1 cells with a ferroptosis inhibitor, liproxstatin-1 (after treatment with recombinant Lcn2 protein), and found that it alleviated the increase in MDA or lipid ROS levels triggered by Lcn2 (Additional file 1: Fig. S3F, G). These results indicated that LCN2 could induce ferroptosis in cultured adipocytes.

Next, we evaluated the potential functional contributions of LCN2 to ferroptosis and tissue wasting in vivo by developing mice overexpressing LCN2. To achieve this, we injected (i.v.) wild-type C57BL/6 mice with 2 × 10⁹ PFUs of an LCN2-expressing lentivirus [31]. mRNA expression of Lcn2 was increased significantly in the eWAT, iWAT, and Gast of mice injected with the LCN2-expressing lentivirus compared with that in the tissues of mice injected with the empty control lentivirus (Fig. 3C); protein expression of LCN2 was also increased significantly, considerably exceeding the serum concentration of endogenous LCN2 (Fig. 3E). In addition, the concentrations of MDA (Fig. 3D) and Fe²⁺ (Fig. 3F) were significantly higher in the eWAT, iWAT, and Gast of LCN2-overexpressing mice than in those of control mice. Moreover, expression of ferroptosis-related genes was increased significantly in the eWAT, iWAT, and Gast of LCN2-overexpressing mice than in those of control mice (Additional file 1: Fig. S3D).

LCN2-overexpressing mice also showed a significant reduction in body weight (Fig. 3G), as well as wasting of adipose tissues (eWAT and iWAT) and skeletal muscle (Gast) (Fig. 3H, I). Histology revealed the clearly aberrant morphology of the iWAT and eWAT of LCN2-overexpressing mice (Fig. 3J); the skeletal muscle of these mice was also pathologically altered markedly (Fig. 3J). In addition, the eWAT and iWAT of mice injected with the LCN2-expressing lentivirus expressed higher levels of thermogenic and lipolytic genes (Additional file 1: Fig. S3A, B), whereas the Gast of these mice expressed higher levels of genes involved in protein degradation (Additional file 1: Fig. S3C). Collectively, our results from these narrowly focused experimental models revealed that LCN2: (i) induced ferroptosis in adipocytes; (ii) contributed specifically and functionally to the observed wasting of WAT and muscle observed initially in cachexia model mice.

Next, we investigated the source of LCN2 in wasting tissues. Notably, LCN2 is a neutrophil gelatinase-associated lipocalin, and has been reported to act as a pleiotropic mediator in several inflammatory and metabolic diseases [32]. The eWAT, iWAT, and Gast of LLC lung cancer cachectic mice had significantly higher numbers of TI-Neu cells (Fig. 4A and Additional file 1:Fig. S4B, E), macrophages, and myeloid-derived suppressor cells (MDSCs) that the equivalent tissues of control mice (Fig. 4A and Additional file 1:sFig.4C, D, E).

We also conducted flow-cytometry analyses of the eWAT, iWAT, and Gast from LLC lung cancer cachectic model mice: TI-Neu cells, macrophages, and MDSCs all expressed LCN2 protein. Moreover, the LCN2 level was higher in the TI-Neu cells derived from the wasting eWAT, iWAT, and Gast of cachectic mice than from control animals. In contrast, the LCN2 levels in eWAT-, iWAT-, and Gast-derived MDSCs and macrophages did not differ between the experimental lung cancer cachexia mice and controls (Fig. 4C). Notably, the LCN2 level was very low in the lymphocytes (e.g., T, B, and natural killer [NK] cells) of cachectic mice and control mice (Additional file 1: Fig. S4G). Furthermore, neutrophils were the largest population of LCN2-positive immune cells (LCN2 + CD45 + cells) (Additional file 1: Fig. S4H). Co-localization of LCN2 and neutrophils (MPO +) in the adipose tissue of cachectic mice was observed (Additional file 1: Fig. S4I), indicating that TI-Neu cells were the major source of LCN2. To garner additional support for the conclusions inferred from the LLC cancer cachexia model, we examined a lung cancer PDX-induced mouse model of cachexia. The two models elicited similar results (Fig. 4B, D).

LCN2 secretion from bone-marrow neutrophils has been reported to mediate appetite suppression during pancreatic cancer cachexia [12]. Similarly, we found that the LCN2 levels in neutrophils derived from the bone marrow and blood were higher in cachectic mice than in control animals (Additional file 1: Fig. S4F). Our evaluation of cachectic mice also showed that the MFI of LCN2 was much higher in WAT neutrophils than in neutrophils originating from the bone marrow or blood (Additional file 1: Fig. S4F). Next, we measured the LCN2 levels in adipocytes, TI-Neu cells, and macrophages purified from the wasting adipose tissue of cachectic mice. The LCN2 level was markedly higher in the TI-Neu cells than in the adipocytes or macrophages of wasting eWAT (Fig. 4E). Then, we cultured these three cell types individually. ELISA of the culture supernatants showed that the LCN2 level was obviously increased in TI-Neu cells but not in the other two cell types (Fig. 4F).

Analysis of the peripheral blood from lung cancer cachectic patients revealed two main phenomena. First, the proportion and absolute number of circulating neutrophils were higher in cancer patients with cachexia than in cancer patients not suffering from cachexia or in healthy controls (Fig. 4G). Second, the neutrophils of cancer patients with cachexia had significantly higher LCN2 levels than other immune cells (monocytes, T cells, NK cells, and B cells) (Fig. 4H). Furthermore, the number of neutrophils in the blood of cancer patients suffering from cachexia correlated positively with their LCN2 level, but negatively with their BMI (Fig. 4I). Taken together, these results showed that TI-Neu cells from wasting tissues were a major source of LCN2.

To determine if elimination of TI-Neu cells from mice with lung cancer cachexia could alleviate tissue wasting and ferroptosis, we treated model mice with a neutrophil-depleting, anti-Ly6G antibody (Additional file 1: Fig. S5A). As expected, treatment with the anti-Ly6G antibody reduced the number of neutrophils in the blood and adipose tissue of model mice significantly compared with treatment with the control (IgG) (Additional file 1: Fig. S5B–E). Strikingly, the body weight of neutrophil-depleted lung cancer cachectic model mice was significantly higher than that of IgG control-treated animals; there was no significant difference between the body weight of control mice and neutrophil-depleted cachexia model animals (Fig. 5A).

Neutrophil depletion alleviated the wasting of eWAT and iWAT significantly (Fig. 5B, C). Moreover, histology showed that neutrophil depletion alleviated the pathological morphology of iWAT, eWAT, and Gast observed in cachexia model mice (Fig. 5D) and also reduced the LCN2 levels in the eWAT, iWAT, Gast, and serum significantly (Fig. 5E, F). Neutrophil depletion reduced the MDA concentrations in iWAT, eWAT, and Gast significantly (Fig. 5G, H) and alleviated the increase in Fe²⁺ concentrations in iWAT and Gast compared with those in IgG control-treated mice (Additional file 1: Fig. S5F). Next, we measured mRNA expression of ferroptosis-related genes. The increase in expression of these genes was alleviated significantly in the eWAT, iWAT, and Gast of anti-Ly6G-antibody-treated mice compared with those in IgG control-treated animals (Additional file 1: Fig. S5G). Collectively, these results showed that the targeted elimination of neutrophils alleviated tissue wasting in lung cancer cachexia model mice significantly.

Next, we examined the impact of knocking out LCN2 by generating myeloid-specific Lcn2 knockout (Lcn2^f/fLysM^cre) mice (Additional file 1: Fig. S6A) and inducing lung cancer cachexia in these mice and their control (Lcn2^f/+LysM^cre) littermates. Induction of lung cancer cachexia led to the expected increase in protein expression of LCN2 in the serum of Lcn2^f/+LysM^cre mice, but not in Lcn2^f/fLysM^cre mice (Additional file 1: Fig. S6B). As expected, Lcn2^f/+LysM^cre, but not Lcn2^f/fLysM^cre mice, exhibited weight loss (Fig. 6A), significant wasting of eWAT, iWAT, and Gast (Fig. 6B–D), and pathological morphology of eWAT, iWAT, and Gast (Fig. 6E). Examination of the biochemical features of ferroptosis showed that Fe²⁺ and MDA concentrations were increased significantly in the eWAT, iWAT, and Gast of Lcn2^f/+LysM^cre, but not in those of Lcn2^f/fLysM^cre mice (Fig. 6F, G). Moreover, analyses of the lipid ROS of fresh adipocytes purified from eWAT showed lipid ROS levels to be increased significantly in Lcn2^f/+LysM^cre mice, but not in Lcn2^f/fLysM^cre mice (Additional file 1: Fig. S6H). Taken together, these findings suggested that myeloid-derived LCN2 contributed specifically to the ferroptotic tissue wasting observed in lung cancer cachexia model mice.

Given our findings regarding LCN2 knockout, next we explored if treating mice with an anti-LCN2 antibody to achieve targeted elimination of LCN2 conferred similarly protective effects in lung cancer cachexia model mice (Additional file 1: Fig. S6C). Upon administration of the anti-LCN2 antibody (or IgG control; both 50 μg per mouse), we observed the expected reductions in eWAT, iWAT, and serum LCN2 level (Additional file 1: Fig. S6D–E). Furthermore, treatment with the anti-LCN2 antibody alleviated the pathogenic phenotypes of lung cancer cachexia model mice, including the reduction in body weight (Fig. 6H) and eWAT/iWAT wasting (Fig. 6I, J). Histology showed that treatment with anti-LCN2 antibody also alleviated the pathological morphology of iWAT and Gast of cachexia model mice (Additional file 1: Fig. S6F, G). Biochemical analysis showed that lung cancer cachexia model mice treated with anti-LCN2 antibody had significantly lower Fe²⁺ and MDA concentrations in their eWAT, iWAT, and Gast than those in controls (Fig. 6K, L). Moreover, anti-LCN2 antibody-treated mice lived significantly longer (Fig. 6M). Collectively, the findings from studies on myeloid-specific knockout and therapy with anti-LCN2 antibody demonstrated that targeted elimination of LCN2 alleviated ferroptosis and tissue wasting significantly in mice with lung cancer cachexia.

Until this point, we had focused on the nature of the aberrantly high LCN2 level in lung cancer cachexia model mice and LCN2-mediated induction of ferroptosis in wasting tissues. We were also interested in whether disrupting ferroptosis could protect mice directly from developing lung cancer cachexia. To this end, we first treated lung cancer cachectic model mice with a ferroptosis inhibitor: liproxstatin-1 (10 mg per kg body weight, delivered by intraperitoneal injection daily between days 1 and 20). Histology showed that liproxstatin-1 treatment alleviated the pathological morphology of the iWAT, eWAT, and Gast of cachexia model mice (Fig. 7A). Liproxstatin-1 treatment also alleviated the loss in body weight of cachectic mice (Additional file 1: Fig. S7D). As expected, liproxstatin-1 treatment reduced the Fe²⁺ concentration in eWAT, iWAT, and Gast (Fig. 7B). Liproxstatin-1 treatment also led to a significant reduction in the MDA and lipid ROS levels in eWAT and iWAT (Fig. 7C, D), as well as a significant reduction in expression of ferroptosis-related genes (Ncoa4, Slc39a8, Slc3a2, Slc39a14, Pcbp2, and Sat1) in eWAT, iWAT, and Gast (Additional file 1: Fig. S7C).

We also treated lung cancer cachexia model mice with an alternative ferroptosis inhibitor: DFO (15 mg per kg body weight, delivered by intraperitoneal injection) [33] (Additional file 1: Fig. S7A). Beyond the expected reduction in the Fe²⁺ concentration in eWAT (Fig. 7E), DFO treatment induced four phenotypes in lung cancer cachexia model mice: (i) reduced MDA and lipid ROS levels in eWAT (Fig. 7F, G); (ii) alleviation of weight loss and wasting of eWAT and iWAT (Fig. 7I, J); (iii) alleviation of the pathological morphology of iWAT (Fig. 7K); (iv) reduced expression of ferroptosis-related genes (Ncoa4, Slc39a8, Slc3a2, Slc39a14, Pcbp2, and Sat1) (Additional file 1: Fig. S7B). These four phenotypes were very similar to those of mice subjected to myeloid-specific LCN2 knockout or treatment with anti-LCN2 antibody. DFO treatment prolonged the survival of lung cancer cachexia model mice significantly (Fig. 7L) and reduced the percentage of dead (7AAD⁺) adipocytes in their eWAT significantly (Fig. 7H). The results of experiments on DFO treatment showed that disrupting ferroptosis protected mice from cachexia directly. Thus, the antibody-based disruption of LCN2-induced ferroptosis or chemical inhibition of ferroptosis represent potentially promising strategies for treating the tissue wasting and multiple other deleterious aspects of cachexia.