Abstract

   Most patients with advanced cancer develop cachexia, a multifactorial
   syndrome characterized by progressive skeletal muscle wasting. Despite
   its catastrophic impact on survival, the critical mediators responsible
   for cancer cachexia development remain poorly defined. Here, we show
   that a distinct subset of neutrophil-like monocytes, which we term
   cachexia-inducible monocytes (CiMs), emerges in the advanced cancer
   milieu and promotes skeletal muscle loss. Unbiased transcriptome
   analysis reveals that interleukin 36 gamma (IL36G)-producing CD38^+
   CiMs are induced in chronic monocytic blood cancer characterized by
   prominent cachexia. Notably, the emergence of CiMs and the activation
   of CiM-related gene signatures in monocytes are confirmed in various
   advanced solid cancers. Stimuli of toll-like receptor 4 signaling are
   responsible for the induction of CiMs. Genetic inhibition of
   IL36G-mediated signaling attenuates skeletal muscle loss and rescues
   cachexia phenotypes in advanced cancer models. These findings indicate
   that the IL36G-producing subset of neutrophil-like monocytes could be a
   potential therapeutic target in cancer cachexia.

   Subject terms: Tumour immunology, Tumour immunology, Cancer
   microenvironment
     __________________________________________________________________

   The critical mediators responsible for cancer-associated cachexia
   development remain poorly defined. Here, the authors identify a
   distinct subset of neutrophil-like monocytes that facilitates
   progressive skeletal muscle wasting in cancer pathogenesis.

Introduction

   Cachexia comprises a group of multifactorial inflammatory and metabolic
   syndromes defined by progressive weight loss and skeletal muscle
   wasting independent of whether fat mass loss occurs^[46]1,[47]2.
   Cancer-associated cachexia is often observed in patients with various
   cancers, particularly in advanced stages, with a prevalence of
   approximately 80%^[48]1. However, the mechanisms underlying cachexia
   development in cancer pathogenesis remain unclear. Although cachexia is
   estimated to account for up to 30% of all cancer-related deaths^[49]3,
   limited therapeutic options are currently available.

   Recent advances in cancer biology have uncovered critical pro- and
   anti-tumor roles of immune cells in the generation and progression of
   cancer pathogenesis^[50]4–[51]6. Among these immune cells, myeloid
   innate immune cells, including neutrophils, monocytes, macrophages, and
   dendritic cells, are key components of the complex cancer-associated
   ecosystem^[52]7–[53]15. Accumulating evidence demonstrates that cancer
   cells can reprogram these immune cells toward a tumor-promoting
   phenotype, not only within the tumor microenvironment (TME) but also in
   systemic circulation. This can facilitate tumor growth, invasion,
   metastasis, immune escape, and therapy resistance, leading to disease
   progression and dismal outcomes in patients with cancer^[54]14,[55]16.
   Given that cachexia primarily occurs in advanced cancer^[56]1,
   tumor-educated myeloid innate immune cells likely contribute to the
   pathogenesis of cancer cachexia. Indeed, increased expression of
   several cytokines and chemokines, such as tumor necrosis factor (TNF),
   interleukin 1 (IL1), and IL6, has been reported in patients with cancer
   cachexia^[57]17. Besides tumor cells, many myeloid innate immune cells
   are capable of producing these proinflammatory factors^[58]18. C-C
   motif chemokine ligand (CCL2), a master chemoattractant for monocytes,
   also exhibits high expression levels in serum from patients with
   cancer, particularly those with cachexia^[59]19. Several reports also
   demonstrated the relevance of neutrophils in cachexia^[60]20,[61]21.
   However, the subset of myeloid innate immune cells that play a central
   role in cancer cachexia development remains unclear.

   An increase in circulating monocytes is often associated with a poor
   prognosis in patients with both blood and solid cancers^[62]22–[63]25.
   In blood cancers, an absolute increase in peripheral blood (PB)
   classical monocytes is a characteristic clinical feature of chronic
   myelomonocytic leukemia (CMML), a hematopoietic clonal disorder
   characterized by persistent monocytosis. In addition to the
   hematological phenotype, systemic symptoms such as fatigue, weight
   loss, and cachexia are common clinical features of patients with
   CMML^[64]26,[65]27. These systemic symptoms are prominent in patients
   with CMML, suggesting that a subpopulation of classical monocytes could
   be responsible for these phenomena.

   Here, we identify a distinct subset of inflammatory monocytes that
   exhibits neutrophil-like characteristics and facilitates progressive
   skeletal muscle atrophy in a discrete type of blood cancer
   pathogenesis. Importantly, we find that these monocytes are also
   prominently induced in the context of solid cancer-derived cachexia and
   that inhibition of this monocyte-derived inflammatory axis could blunt
   the development of cachexia.

Results

Cachexia development in monocytic leukemia mice

   To investigate the mechanisms of cachexia development in blood cancer
   with monocytosis, we used a monocytic leukemia (Mo-Leukemia) mouse
   model that was established using a CMML patient-derived NUP98-HBO1
   mutant^[66]28. The characteristic monocytosis seen in CMML is induced
   in this model^[67]28. We retrovirally introduced the NUP98-HBO1 mutant
   into wild-type (WT) murine hematopoietic stem cells and progenitors
   (HSC/Ps) and transplanted them into WT recipient mice (Mo-Leukemia
   mice) (Fig. [68]1a). Mice transplanted with empty vector-transduced
   cells served as the controls. Persistent monocytosis and an increase in
   the subset of CD14^+ CD16^‒ classical monocytes are hallmarks of CMML
   that separate it from other monocytic disorders with high sensitivity
   and specificity^[69]26. Consistent with these clinical findings, the
   Mo-Leukemia mice exhibited prominent monocytosis (Fig. [70]1b–d). The
   subset of CD115^+ Ly6C^high monocytes, an equivalent subpopulation of
   human classical monocytes, was significantly increased in PB from
   Mo-Leukemia mice (Fig. [71]1e). The frequency of bone marrow (BM)
   monocytes was also higher in Mo-Leukemia mice than in control mice
   (Fig. [72]1f). In addition to the hematological phenotypes, the
   Mo-Leukemia mice showed progressive weight loss (Fig. [73]1g). Skeletal
   muscle weight and fiber cross-sectional area (CSA) were significantly
   lower in Mo-Leukemia mice with weight loss than in control mice
   (Fig. [74]1h–j and Supplementary Fig. [75]1). Furthermore, genes
   related to skeletal muscle atrophy, such as Trim63 and Fbxo32 (also
   known as MuRF1 and Atrogin1, respectively)^[76]1, were prominently
   upregulated in the skeletal muscle (quadriceps) of Mo-Leukemia mice
   with weight loss compared to those in control mice (Fig. [77]1k).
   Mo-Leukemia mice with weight loss exhibited lower grip strength than
   control mice (Fig. [78]1l). These results suggest that Mo-Leukemia mice
   develop cancer cachexia, characterized by skeletal muscle wasting and
   weakness.

Fig. 1. Cachexia development in monocytic leukemia mice.

   [79]Fig. 1
   [80]Open in a new tab

   a Schematic of experimental design. b White blood cell (WBC) counts in
   PB from control and Mo-Leukemia mice (n = 7 mice per each). c–e Flow
   cytometric analysis of total PB from control (n = 6 mice) and
   Mo-Leukemia mice (n = 7 mice). Representative flow cytometry plots (c),
   frequencies of total monocytes in PB (d) and Ly6C^+ inflammatory
   monocytes in total monocytes (e) are shown. f Flow cytometric analysis
   of total BM cells from control (n = 3 mice) and Mo-Leukemia (n = 7
   mice) mice. The frequencies of monocytes in BM are shown. g Body weight
   of the control and Mo-Leukemia mice (n = 7 mice per group) after
   transplantation. h Skeletal muscle weights of control (n = 5 mice, at 6
   to 10 weeks after BMT) and Mo-Leukemia mice (n = 7 mice, at the
   endpoint or 10 weeks after BMT). Weights are normalized to the initial
   body weight (IBW) and presented as weight/100 mg IBW. i, j HE staining
   and cross-sectional area (CSA) of quadriceps from control and
   Mo-Leukemia mice (n = 3 mice per group). Representative images (i) and
   average (j) of CSA are shown. Scale bar, 50 μm. k mRNA expression
   levels of muscle atrophy markers in quadriceps from control and
   Mo-Leukemia mice (n = 5 mice per each) at the endpoint. Log2 fold
   change values compared to mean values of control are shown. l Forelimb
   grip strength of control (n = 5) and Mo-Leukemia mice (n = 6 mice) at
   the endpoint. The values were normalized to those of the control group.
   m Schematic of the co-culture assay. n Representative
   immunofluorescence images of C2C12 myotubes. Scale bar, 100 μm. o
   Relative diameter of C2C12 myotubes after co-culture with inflammatory
   monocytes and neutrophils. Representative results from more than three
   independent experiments are shown. The number of muscle fibers analyzed
   in each sample is shown. Statistical analyses were performed using a
   two-tailed unpaired Student’s t-test. Data are presented as the
   mean ± SD. NS is not significant. Source data are provided as a Source
   Data file.

   Considering that persistent characteristic monocytosis and cachexia
   development are characteristic clinical features of CMML, it is likely
   that the subpopulation of classical monocytes is responsible for the
   development of cachexia. To test this hypothesis, we collected BM
   CD115^+ Ly6C^high monocytes from Mo-Leukemia or control mice and
   cultured them with differentiated murine C2C12 myotubes in a transwell
   system (Fig. [81]1m and Supplementary Fig. [82]2). We observed a
   significant decrease in the diameter of C2C12 myotubes upon co-culture
   with CD115^+ Ly6C^high inflammatory monocytes from Mo-Leukemia mice
   (Fig. [83]1n, o). In contrast, Ly6G^+ granulocytes from Mo-Leukemia
   mice did not affect the diameter of C2C12 myotubes (Fig. [84]1o). These
   results indicate that the inflammatory monocytes from cachectic
   Mo-Leukemia mice contain cachexia-inducible monocytes that could be
   termed cachexia-inducible monocytes (CiMs).

IL36G-expressing CD38^+ CiMs induce muscle atrophy

   To characterize the gene expression profile of inflammatory monocytes
   in Mo-Leukemia mice with cachexia, we performed bulk RNA sequencing
   (RNA-seq) of BM CD115^+ Ly6C^high inflammatory monocytes obtained from
   Mo-Leukemia mice, with or without weight loss, and control mice
   (Fig. [85]2a). Principal component analysis revealed a unique gene
   expression pattern in inflammatory monocytes derived from Mo-Leukemia
   mice with cachexia (Fig. [86]2b). Many genes were upregulated in
   Mo-Leukemia monocytes compared with control monocytes (Fig. [87]2c and
   Supplementary Data [88]1). We re-clustered these genes and identified
   gene clusters (clusters 7 and 8) that were upregulated in Mo-Leukemia
   mice with cachexia compared to those without weight loss (Fig. [89]2d
   and Supplementary Data [90]2). Genes in cluster 7, in particular, were
   enriched in Gene Ontology (GO); biological processes related to
   granulocyte and neutrophil activation (Fig. [91]2d). Given that
   inflammatory monocytes from cachectic Mo-Leukemia mice induced skeletal
   muscle atrophy even in an indirect co-culture setting (Fig. [92]1m–o),
   we hypothesized that these monocytes might produce signaling molecules
   that could promote skeletal muscle wasting. To identify
   cachexia-inducible molecules derived from monocytes, we intersected
   genes in clusters 7/8 with a GO-term related to receptor ligand
   activity (GO:0048018) (Fig. [93]2e). Among the six genes identified
   using intersection analysis, Il36g (also known as Il1f9) comprised the
   significantly expressed genes in the inflammatory monocytes from
   cachectic Mo-Leukemia mice (but not mice without weight loss)
   (Fig. [94]2f).

Fig. 2. Identification of IL36G-producing CiMs.

   [95]Fig. 2
   [96]Open in a new tab

   a Schematic of sample collection for bulk RNA-Seq (n = 2 independent
   samples per each). b Principal component analysis for bulk RNA-seq data
   of BM CD115^+ Ly6C^high inflammatory monocytes obtained from the
   indicated mice. c HOPACH-generated clusters of differentially expressed
   genes in CD115^+ Ly6C^high inflammatory monocytes from the indicated
   mice. d Heat map of re-clustered genes upregulated in Mo-Leukemia mice
   (HOPACH algorithm) with statistically enriched Gene Ontology (GO) terms
   (left) and enriched GO term-associated genes (right). The upregulated
   genes in Mo-Leukemia mice (c) were re-clustered. e, f
   Receptor-ligand-related genes in DEGs upregulated in CD115^+ Ly6C^high
   inflammatory monocytes from cachectic Mo-Leukemia mice (e). For the six
   selected genes, volcano plots in the indicated comparisons are shown in
   (f). g, h Plasma membrane-related genes among DEGs upregulated in
   CD115^+ Ly6C^high inflammatory monocytes from cachectic Mo-Leukemia
   mice (g). For the selected 95 genes, a volcano plot in the indicated
   comparison is shown in (h). i, j Flow cytometric analysis of total PB
   from the control, Mo-Leukemia mice with cachexia, and Mo-Leukemia mice
   without weight loss (n = 5 mice per group). Representative flow
   cytometry plots (i) and the frequencies of CD38^+ cells in PB Ly6C^+
   inflammatory monocytes (j) are shown. k Diff-Quik staining of CD38^+
   Ly6C^+ inflammatory monocytes from Mo-Leukemia mice with cachexia.
   Scale bar, 10 μm. l Flow cytometric analysis of the total BM from
   control, Mo-Leukemia mice with cachexia, and Mo-Leukemia mice without
   weight loss (n = 4 mice per each). The frequencies of CD38^+ cells in
   BM Ly6C^+ inflammatory monocytes are shown. m Relative diameters of
   C2C12 myotubes after the co-culture assay with the indicated cells.
   Representative results from more than three independent experiments are
   shown. The number of muscle fibers analyzed in each sample is shown. n
   Schematic of sample collection. o mRNA expression levels of Il36g in
   control Ly6C^+ monocytes (n = 4 mice), CD38^+ Ly6C^+ inflammatory
   monocytes from Mo-Leukemia mice (n = 5 mice), and CD38^‒ Ly6C^+
   inflammatory monocytes from Mo-Leukemia mice (n = 5 mice). p Expression
   level of Il36g protein in the culture supernatant of control Ly6C^+
   monocytes (n = 4 mice), CD38^+ Ly6C^+ inflammatory monocytes from
   Mo-Leukemia mice (n = 5 mice), and CD38^‒ Ly6C^+ inflammatory monocytes
   from Mo-Leukemia mice (n = 5 mice). Statistical analyses were performed
   using Fisher’s exact test (d), empirical Bayes moderated t test (f, h),
   and one-way ANOVA followed by Tukey multiple comparison test (j, l, m,
   o, p). Data are presented as the mean ± SD (j, l, m, o, p). Source data
   are provided as a Source Data file.

   Characteristic cell surface molecules enable the isolation of specific
   cell subsets within a heterogeneous cell population using immunological
   techniques, including flow cytometry and fluorescence microscopy. To
   identify surface markers that discriminate CiMs from other monocytes,
   we also intersected genes in clusters 7/8 from the bulk RNA-Seq
   analysis (Fig. [97]2d) with a GO-term related to the plasma membrane
   (GO:0005886) (Fig. [98]2g). Among the genes upregulated in CD115^+
   Ly6C^high inflammatory monocytes from cachectic Mo-Leukemia mice, Cd38
   was the most significantly expressed (Fig. [99]2h and Supplementary
   Data [100]3). We confirmed protein expression using flow cytometry and
   found that CD38 protein was prominently expressed in PB CD115^+
   Ly6C^high inflammatory monocytes from cachectic Mo-Leukemia mice
   compared with those from control or non-cachectic mice (Fig. [101]2i,
   j). These CD38^+ CD115^+ Ly6C^high inflammatory monocytes exhibited
   monocyte/macrophage morphology (Fig. [102]2k). CD38 expression was also
   detected in BM CD115^+ Ly6C^high inflammatory monocytes from cachectic
   Mo-Leukemia mice (but not Mo-Leukemia mice without weight loss)
   (Fig. [103]2l). Furthermore, the diameter of C2C12 myotubes
   significantly decreased upon co-culture with CD38^+ CD115^+ Ly6C^high
   inflammatory monocytes from cachectic Mo-Leukemia mice compared to
   CD115^+ Ly6C^high inflammatory monocytes from control mice or CD38^‒
   CD115^+ Ly6C^high inflammatory monocytes from cachectic Mo-Leukemia
   mice (Fig. [104]2m). The expression levels of Trim63 and Fbxo32 mRNA
   were also increased in C2C12 myotubes upon co-culture with CD38^+
   CD115^+ Ly6C^high inflammatory monocytes from Mo-Leukemia mice
   (Supplementary Fig. [105]3). We confirmed Il36g mRNA expression and
   Il36g protein production in the CD38^+ inflammatory monocytes
   (Fig. [106]2n–p). Taken together, these results suggest that the
   Il36g-producing subset of inflammatory monocytes can function as “CiMs”
   and promote cachexia development in Mo-Leukemia mice. CD38 could serve
   as a useful surface marker for IL36G-producing CiMs.

Neutrophil-like gene signature in IL36G-producing CiMs

   To confirm that induction of CiMs is not a consequence of the
   progression of severe cachexia, we isolated CD115^+ Ly6C^high BM
   inflammatory monocytes from Mo-Leukemia mice at the time of
   pre-cachexia, when weight loss began to appear, and performed
   single-cell RNA-Seq (scRNA-Seq) analysis (Fig. [107]3a). We then
   compared the results with those of control mice using the Cellranger
   aggr pipeline and identified 10 clusters based on t-distributed
   stochastic neighbor embedding (t-SNE) plots (Fig. [108]3b). Among these
   clusters, clusters 2, 7, and 8 were primarily composed of inflammatory
   monocytes from the Mo-Leukemia mice (Fig. [109]3b). Pathway enrichment
   analysis showed that the distinctively upregulated genes in these three
   clusters were enriched for several signaling pathways related to
   neutrophil regulation and inflammation (Fig. [110]3c and Supplementary
   Data [111]4). Notably, cells expressing Il36g were primarily included
   in these clusters (Fig. [112]3d). Some of Il36g-expressing cells were
   also included in clusters 1 and 3 (Fig. [113]3b, d and Supplementary
   Data [114]4). To further characterize the gene signature of cells
   expressing Il36g at the time of pre-cachexia, we processed the
   scRNA-Seq data of inflammatory monocytes from Mo-Leukemia mice using
   iterative clustering and guide-gene selection (ICGS) algorithm, an
   unsupervised discovery workflow that can resolve rare and transitional
   populations in scRNA-Seq datasets^[115]29. ICGS resolved nine clusters
   and revealed a unique cluster (cluster #4) in which Il36g-expressing
   CiMs were significantly enriched (Fig. [116]3e–g and Supplementary
   Data [117]5). We note that many granulocyte-related genes, such as
   Prtn3, Ctsg, Elane, S100a8, S100a9, and Cebpe, were highly expressed in
   this CiM-enriched cluster (Fig. [118]3e). The upregulated
   differentially expressed genes (DEGs) in Il36g-expressing CiMs in
   comparison to cells without Il36g expression (non-CiMs) included many
   neutrophil-related genes (Fig. [119]3h). GO-term enrichment analysis
   showed that these upregulated DEGs in CiMs were significantly enriched
   for neutrophil regulation and activation (Fig. [120]3i and
   Supplementary Data [121]6). Taken together, these results suggest that
   IL36G-producing CiMs exhibit neutrophil-like characteristics.

Fig. 3. CiMs exhibit neutrophil-like gene signature.

   [122]Fig. 3
   [123]Open in a new tab

   a Schematic of sample collection for scRNA-Seq. b t-SNE plot of 5358
   single cells distributed using a graph-based clustering algorithm (Cell
   Ranger). c Pathway enrichment analysis for differentially expressed
   genes in indicated clusters. d Expression intensity and distribution of
   Il36g are overlaid on the tSNE plot. e, f Heat map and uniform manifold
   approximation and projection (UMAP) of genes delineated by ICGS in
   scRNA-seq data of inflammatory monocytes from Mo-Leukemia mice at the
   time of cachexia onset. g Frequency of Il36g-expressing CiMs in
   individual ICGS clusters. h DEGs upregulated in Il36g-expressing CiMs.
   i GO-term enrichment analysis of DEGs upregulated in Il36g-expressing
   CiMs. Statistical analyses were performed using Fisher’s exact test (c,
   e, i) and empirical Bayes moderated t test (h). Source data are
   provided as a Source Data file.

   Yáñez et al. demonstrated that neutrophil-like inflammatory monocytes
   differentiate from granulocyte-monocyte progenitors (GMPs) upon
   toll-like receptor 4 (TLR4) stimulation^[124]30. Given that the CiMs
   exhibited a neutrophil-like gene signature, they likely belong to the
   subtype of neutrophil-like monocytes and could, therefore, be induced
   in response to TLR4 stimuli. To test this hypothesis, we injected
   lipopolysaccharide (LPS), a TLR4 agonist^[125]31,[126]32, into both
   non-cachectic Mo-Leukemia mice and WT control mice. As expected, we
   observed the emergence of the Il36g-expressing CD38^+ CD115^+ Ly6C^high
   inflammatory monocytes in Mo-Leukemia mice (Supplementary Fig. [127]4a,
   b). In contrast, administration of unmethylated CpG DNA, an agonist of
   TLR9 that could promote inflammatory monocyte differentiation from
   monocyte-dendritic cell progenitors (MDPs)^[128]30, did not induce
   CD38^+ CD115^+ Ly6C^high inflammatory monocytes in Mo-Leukemia mice
   (Supplementary Fig. [129]4a). Furthermore, we tested the upregulated
   DEGs in CiMs in bulk RNA-Seq dataset of GMP- and MDP-derived
   inflammatory monocytes ([130]GSE88982)^[131]30 using gene set
   enrichment analysis (GSEA) and found that these genes were
   significantly enriched in GMP-derived neutrophil-like monocytes
   (Supplementary Fig. [132]4c). Interestingly, some of the
   neutrophil-related genes, such as Ctsg, Elane, Mpo, and Prtn3, were not
   upregulated in CD38^+ CD115^+ Ly6C^high inflammatory monocytes in
   Mo-Leukemia mice after LPS administration. These results suggest the
   heterogeneity of neutrophil-like monocytes and the existence of a
   unique subset of neutrophil-like monocytes in the context of cancer
   pathogenesis.

Emergence of CiMs in solid-tumor mediated cancer cachexia

   A significant association between an increase in circulating monocytes
   and a poor prognosis has been reported in both blood and solid
   cancers^[133]22–[134]25. Given that cancer cachexia is primarily
   observed in patients with advanced stages of cancer, we hypothesized
   that CiMs might also be involved in the pathogenesis of cachexia in
   solid cancers. To test this hypothesis and evaluate the tumor-related
   immune cell dynamics, we generated solid tumor-bearing syngeneic models
   using 4T1 cells (metastatic murine breast cancer cells) and B16 cells
   (murine melanoma cells), both of which are established models of cancer
   cachexia^[135]33–[136]37. By intravenously injecting 4T1 breast cancer
   cells and B16 melanoma cells, we generated mouse models of systemic
   hematogenous metastasis (Fig. [137]4a). As expected, cachexia with
   skeletal muscle atrophy and weakness was observed in all of these mice
   with systemic hematogenous metastasis (Fig. [138]4b–g, Supplementary
   Fig. [139]5 and Supplementary [140]6a–e). This is consistent with
   previous reports of the link between systemic metastasis and cancer
   cachexia development^[141]23. Flow cytometric analysis revealed that
   the frequency of CD38^+ CD115^+ Ly6C^high inflammatory monocytes
   prominently increased in both PB and BM from these cachectic 4T1 and
   B16 metastasis models (Fig. [142]4h–j and Supplementary Fig. [143]6f,
   g). Morphological features of these monocytes were confirmed
   (Fig. [144]4k). Furthermore, these CD38^+ CD115^+ Ly6C^high
   inflammatory monocytes induced a significant decrease in the diameter
   of C2C12 myotubes in the transwell indirect co-culture system
   (Fig. [145]4l), suggesting that CiMs are induced in the pathogenesis of
   solid tumor-mediated cancer cachexia. Consistent with the observation
   in the blood cancer model, we also confirmed the upregulation of mRNA
   expression levels of Il36g and neutrophil-related genes in CD38^+
   CD115^+ Ly6C^high inflammatory monocytes isolated from cachectic solid
   tumor models (Fig. [146]4m, n and Supplementary Fig. [147]6h, i). To
   confirm the TLR4 stimuli-mediated induction of CiMs in the advanced
   solid cancer pathogenesis, we pharmacologically inhibited TLR4 in the
   4T1 and B16 metastasis models using TAK-242, a specific inhibitor of
   TLR4 signaling^[148]38 (Fig. [149]4o and Supplementary Fig. [150]6j).
   The induction of CD38^+ CiMs was significantly suppressed in the
   TAK242-treated metastasis models compared with that in the vehicle
   controls (Fig. [151]4o–q and Supplementary Fig. [152]6k, l), suggesting
   that TLR4 signaling activation is responsible for CiM induction in
   cancer pathogenesis. These results suggest that CiMs are induced in the
   pathogenesis of solid tumor-mediated cancer cachexia.

Fig. 4. Emergence of CiMs in advanced solid cancers.

   [153]Fig. 4
   [154]Open in a new tab

   a Schematic of experimental design. b Body weight of control and 4T1
   breast cancer mice (n = 8 mice per group). c Forelimb grip strength of
   control (n = 4 mice) and 4T1 breast cancer (n = 6 mice) mice at the
   endpoint. The values were normalized to those of the control group. d
   Skeletal muscle weights of control (n = 4 mice, at day 11 to 13) and
   4T1 breast cancer (n = 6 mice, at day 11 to 13) mice. Weights are
   normalized to the IBW and presented as weight/100 mg IBW. e, f HE
   staining and CSA of quadriceps from control and 4T1 breast cancer mice
   (n = 3 mice per group). Representative images (e) and average (f) of
   CSA are shown. Scale bar, 50 μm. g mRNA expression levels of muscle
   atrophy markers in quadriceps from control (n = 3 mice) and 4T1 breast
   cancer (n = 4 mice) mice at the endpoint. Log2 fold change values
   compared to mean values of control are shown. h, i Flow cytometric
   analysis of total PB from control and 4T1 breast cancer mice (n = 8
   mice per group). Representative flow cytometry plots (h) and the
   frequencies of CD38^+ cells in PB Ly6C^+ inflammatory monocytes (i) are
   shown. j Flow cytometric analysis of the total BM from control and 4T1
   breast cancer mice (n = 8 mice per group). The frequencies of CD38^+
   cells in BM Ly6C^+ inflammatory monocytes are shown. k Diff-Quik
   staining of CD38^+ Ly6C^+ inflammatory PB monocytes from 4T1 mice with
   cachexia. Scale bar, 10 μm. l Relative diameter of C2C12 myotubes after
   co-culture with Ly6C^+ inflammatory monocytes from control mice or
   CD38^+ Ly6C^+ inflammatory monocytes from 4T1 breast cancer mice.
   Representative results from more than three independent experiments are
   shown. The number of muscle fibers analyzed in each sample is shown. m
   mRNA expression levels of Il36g in Ly6C^+ inflammatory monocytes from
   control mice (n = 4 mice) and in CD38^+ Ly6C^+ inflammatory monocytes
   from 4T1 breast cancer mice (n = 5 mice). Log2 fold change values
   compared to mean values of control are shown. n Expression levels of
   neutrophil-related genes in Ly6C^+ inflammatory monocytes from control
   mice (n = 4 mice) and in CD38^+ Ly6C^+ inflammatory monocytes from 4T1
   breast cancer mice (n = 5 mice). Log2 fold change values compared to
   mean values of control are shown. o Schematic of experimental design.
   p, q Flow cytometric analysis of total PB from 4T1 breast cancer mice
   treated with vehicle (n = 4 mice) and TAK-242 (n = 7 mice).
   Representative flow cytometry plots (p) and the frequencies of CD38^+
   cells in PB Ly6C^+ inflammatory monocytes (q) are shown. Statistical
   analyses were performed using a two-tailed unpaired Student’s t test.
   Data are presented as the mean ± SD. Source data are provided as a
   Source Data file.

   To further validate CiM emergence in the context of advanced solid
   tumor-mediated cancer cachexia, we analyzed a dataset of inflammatory
   monocytes obtained from mice with breast cancer bone metastasis
   ([155]GSE152795)^[156]39. Neutrophil-related genes, including Il36g and
   Cd38, were significantly upregulated in inflammatory monocytes from
   mice with advanced breast cancer compared to those from control mice
   (Supplementary Fig. [157]7a, b). The upregulated DEGs in CiMs were
   significantly enriched in the inflammatory monocytes from mice with
   advanced breast cancer (Supplementary Fig. [158]7c). We also tested the
   upregulated DEGs in CiMs in the human datasets of classical monocytes
   obtained from patients with colorectal cancer (CRC) and renal cell
   carcinoma (RCC), in which cachexia is commonly observed ([159]GSE47756
   and [160]GSE38424)^[161]40,[162]41. The CRC cohort included patients
   with non-metastatic CRC (n = 27), patients with metastatic CRC
   (n = 28), and healthy donors (n = 38). The RCC cohort included patients
   with RCC (n = 4) and healthy donors (n = 4). Consistent with the
   findings in multiple mouse models of cancer cachexia, the upregulated
   DEGs in CiMs were significantly enriched in the classical monocytes of
   patients with CRC (particularly in the metastatic CRC group) and RCC
   (Supplementary Fig. [163]8a, b). Notably, the upregulated DEGs in CiMs
   were particularly enriched in the metastatic CRC cohort compared with
   the non-metastatic CRC cohort (Supplementary Fig. [164]8a). This is
   consistent with the fact that cachexia occurs primarily in advanced
   cancer with invasion and metastasis. These results suggest that the
   emergence of neutrophil-like CiMs is a common pathobiological
   phenomenon in cancer cachexia.

Targeting IL36G signaling in cancer cachexia

   Our data obtained from mouse models of blood and solid cancers indicate
   that the IL36G-mediated signaling pathway is responsible for cancer
   cachexia development. To investigate the clinical relevance of IL36G
   levels in patients with cancer cachexia, we measured IL36G levels in
   the serum of patients with advanced cancer with cachexia in the
   palliative care unit (Supplementary Data [165]7). Among the advanced
   cancer patients who met the diagnosis criteria for cachexia, seven out
   of 28 patients (25%) exhibited high levels of serum IL36G, with the
   remaining patients having less than detectable sensitivity
   (Supplementary Fig. [166]9a). We note that the weight loss over the
   preceding 6 months (one of the three criteria for the international
   consensus cachexia definition^[167]2) in cachectic patients with high
   serum IL36G levels was more prominent than in those without increases
   in serum IL36G levels (Supplementary Fig. [168]9b).

   To determine the effect of IL36G signaling inhibition on muscle atrophy
   induction in vitro, we generated CRISPR/Cas9-edited C2C12 cells lacking
   the expression of IL36 receptor (Fig. [169]5a). The knockout of Il36r
   did not affect the viability of C2C12 cells (Supplementary
   Fig. [170]10). We then cultured Il36r-deficient C2C12 myotubes with
   CD38^+ CD115^+ Ly6C^high inflammatory monocytes from cachectic mice
   with blood and solid cancers (Fig. [171]5b, c). Genetic inhibition of
   IL36G signaling significantly attenuated both CiM-mediated reduction in
   myotube diameter and elevation of the muscle atrophy marker
   (Fig. [172]5c and Supplementary Fig. [173]11). To evaluate the effect
   of IL36G signaling inhibition in vivo, we first integrated lentiviral
   transduction strategy and generated CRISPR/Cas9-edited Mo-Leukemia
   cells lacking the expression of Il36g (Fig. [174]5d). Empty
   vector-transduced Mo-Leukemia cells served as control. We then
   transplanted them into WT recipient mice (Fig. [175]5d). Although both
   groups of mice developed monocytosis, a characteristic CMML phenotype,
   the cachexia phenotypes in the monocytic blood cancer model, including
   weight loss, muscle weakness, and muscle loss, were rescued in
   Mo-Leukemia mice with hematopoietic cell-specific deletion of Il36g
   (Fig. [176]5e–g). In addition, we genetically eliminated IL36 receptor
   expression in cachexia mice with both blood and solid cancers. Although
   Il36r-WT Mo-Leukemia mice developed progressive weight loss
   (Fig. [177]5h, i), the cachexia phenotypes in this model were
   attenuated in Il36r-deficient Mo-Leukemia mice (Fig. [178]5i–k). The
   cachexia phenotypes in the syngeneic metastatic breast cancer and
   melanoma models were also rescued by inhibition of IL36G signaling
   (Fig. [179]5l–o and Supplementary Fig. [180]12a–d). Taken together,
   these results suggested that the CiMs/IL36G signaling axis could be a
   novel therapeutic target in cancer cachexia (Fig. [181]5p).

Fig. 5. Targeting IL36G/IL1RL2 signaling in cancer cachexia.

   [182]Fig. 5
   [183]Open in a new tab

   a Protein expression of Myh1 and Il36r in C2C12 myotubes with or
   without Il36r. Scale bar, 50 μm. Representative image and relative
   expression levels of Il36r are shown (Il36r-WT, n = 18 cells; Il36r-KO,
   n = 15 cells). b Schematic of the co-culture assay. c Relative
   diameters of C2C12 myotubes after the co-culture assay with the
   indicated cells. Representative results of more than three independent
   experiments are shown. The number of muscle fibers analyzed in each
   sample is shown. d Schematic of experimental design. e Body weight of
   Il36g-WT Mo-Leukemia mice (n = 6 mice) and Il36g-KO Mo-Leukemia mice
   (n = 5 mice). f Forelimb grip strength of Il36g-WT Mo-Leukemia mice
   (n = 6 mice) and Il36g-KO Mo-Leukemia mice (n = 5 mice). g Skeletal
   muscle weight of Il36g-WT Mo-Leukemia mice (n = 6 mice, at 5 weeks
   after BMT) and Il36g-KO Mo-Leukemia mice (n = 5 mice, at 5 weeks after
   BMT) at the endpoint. Weights are normalized to the IBW and presented
   as weight/100 mg IBW. h Schematic of experimental design. i Body weight
   of the control (n = 4 mice), Mo-Leukemia mice with Il36r (n = 4 mice),
   and Mo-Leukemia mice without Il36r (n = 8 mice). j Forelimb grip
   strength of Mo-Leukemia mice with or without Il36r (n = 4 mice per
   group) at the endpoint relative to control mice. k Skeletal muscle
   weight of control (n = 3 mice, at 5 weeks after BMT) and Mo-Leukemia
   mice with Il36r (n = 4 mice, at 3 to 5 weeks after BMT) or without
   Il36r (n = 4 mice, at 5 weeks after BMT). Weights are normalized to the
   IBW and presented as weight/100 mg IBW. l Schematic of experimental
   design. m Body weight of the control (n = 6 mice) and 4T1 breast cancer
   mice with or without Il36r (n = 5 mice per group). n Forelimb grip
   strength of 4T1 breast cancer mice with or without Il36r (n = 5 mice
   per group) at the endpoint relative to control mice. o Skeletal muscle
   weight of control (n = 6 mice) and 4T1 breast cancer mice with or
   without Il36r (n = 5 mice per group) at day 15. Weights are normalized
   to the IBW and presented as weight/100 mg IBW. p Proposed mechanism for
   CiM-mediated cancer cachexia development. Statistical analyses were
   performed using two-tailed unpaired Student’s t test (c, e, g, i, j, n)
   and one-way ANOVA followed by Tukey multiple comparison test (k, m, o).
   Data are presented as the mean ± SD. NS is not significant. Source data
   are provided as a Source Data file.

Discussion

   We have identified a distinct cachexia-inducible subset of inflammatory
   monocytes in blood and solid cancers and termed them “CiMs”. CiMs
   exhibit a neutrophil-like gene signature and secrete IL36G, which
   facilitates skeletal muscle wasting and weakness. Importantly,
   inhibition of IL36G signaling attenuated cachexia development in mouse
   models of cancer cachexia, indicating that the CiMs/IL36G signaling
   axis is a potential therapeutic option for cancer cachexia.

   Classical monocytes are differentiated from MDPs^[184]42. Although MDPs
   are thought to emerge from GMPs, recent studies have shown that MDPs
   emerge directly from a subset of common myeloid progenitors (CMPs), not
   from GMPs^[185]30,[186]42. Accumulating evidence has demonstrated that
   GMPs can also give rise to characteristic monocytes with a
   neutrophil-like genetic signature^[187]30. We identified a distinct
   subset of monocytes that emerged in advanced cancer-mediated cachexia
   pathogenesis and found that these monocytes exhibited neutrophil-like
   features. Similar to the neutrophil-like monocytes, which have been
   described by others^[188]30, CiMs were prominently induced upon in vivo
   TLR4 (but not TLR9) stimulation, and the emergence was suppressed by
   inhibition of TLR4, suggesting that tumor-/TME-derived TLR4 ligands are
   responsible for the induction of CiMs in cancer pathogenesis. Increased
   levels of serum damage-associated molecular patterns (DAMPs) have been
   reported in patients with various cancers, particularly if the cancer
   is at an advanced stage^[189]43–[190]45. Notably, several DAMPs that
   have been identified in patients with advanced cancer, such as high
   mobility group box 1 (HMGB1) and the S100 protein family (S100A8 and
   S100A9), could function as TLR4 ligands^[191]46,[192]47. These
   tumor-pathogenesis-related DAMPs are likely to promote the induction of
   CiMs, and thus lead to the development of cachexia in these patients.

   We focused on unbiased transcriptome analysis to determine the distinct
   CiM-derived signaling molecules that could promote skeletal muscle
   atrophy and identified IL36G as a critical mediator of cachexia
   development. Increased expression of serum IL36G has been reported in
   patients with inflammatory diseases, including type-2 diabetes and
   obesity, autoimmune disorders, and many cancers^[193]48,[194]49. IL36G
   is an inflammatory cytokine that belongs to the interleukin-1 (IL1)
   superfamily^[195]48,[196]49. IL36G exhibits high-affinity binding to
   IL36R, recruits interleukin 1 receptor accessory protein (IL1RAP) to
   form the IL36 receptor complex, and activates the intracellular
   toll-like domain, leading to NF-κB signaling
   activation^[197]48,[198]49. It has been well documented that the NF-κB
   signaling pathway can upregulate the expression of TRIM63 and FBXO32,
   which are master E3 ubiquitin ligases responsible for the degradation
   of muscle structural proteins and acceleration of muscle
   atrophy^[199]1. NF-κB signaling also impairs myoblast differentiation
   by suppressing MYOD1 expression^[200]17. This evidence supports the
   critical role of the IL36G/IL36R axis in the development of cancer
   cachexia. Besides cancer, cachexia often occurs in patients with
   chronic inflammation and infection^[201]48,[202]49. A protective role
   of IL36G against obesity has been suggested in patients with type-2
   diabetes^[203]50. Thus, IL36G likely affects skeletal muscle
   homeostasis under chronic inflammation-mediated conditions or diseases.

   Although neutrophil-like inflammatory monocytes have been detected in
   previous scRNA-Seq-based studies^[204]30,[205]42, their surface
   molecules and functional characteristics remain unclear. Our data
   showed that cachexia-related IL36G-producing monocytes exhibited
   neutrophil-like gene signatures and expressed CD38 on their cell
   surface. We demonstrated that CD38^+ inflammatory monocytes increase
   with cachexia progression in both blood and solid cancers, produce
   IL36G, and induce skeletal muscle atrophy. Thus, CD38 may be a
   potential surface marker for a subset of neutrophil-like monocytes.
   CD38 is known to function as a transmembrane enzyme that consumes both
   intracellular and extracellular nicotinamide adenine dinucleotide
   (NAD^+)^[206]51. Growing evidence highlights the critical role of NAD^+
   levels in skeletal muscle homeostasis^[207]52,[208]53. A decline in
   NAD^+ levels could cause skeletal muscle degeneration, whereas
   repletion of NAD^+ rescues skeletal muscle function. Covarrubias, et
   al. recently showed that the senescence-associated secretory phenotype
   (SASP) induces CD38 expression on myeloid innate immune cells and
   promotes tissue NAD^+ decline during aging^[209]54. Senescent
   cell-derived inflammatory SASP is one of the major components of the
   TME that can facilitate tumorigenesis, including cancer cell
   proliferation, invasion, epithelial-mesenchymal transition (EMT),
   metastasis, and immune cell reprograming^[210]55,[211]56. Thus, in
   addition to the role of surface marker, CD38 on the surface of CiMs may
   also directly contribute to cachexia development via its NADase
   activity. However, recent advances in single-cell transcriptomics
   clearly demonstrate the difficulty of precisely characterizing
   functionally discrete cell populations using a few surface
   markers^[212]29,[213]57. Indeed, the gene expression and protein
   production of Il36g in CD38-negative-inflammatory monocytes from some
   of the cachectic mice were also elevated compared with those in control
   cells. Although the majority of inflammatory monocytes from mice with
   prominent cachexia present CD38 on their surface and express Il36g, the
   Il36g-expressing inflammatory monocytes from mice at the time of weight
   loss onset do not always express CD38 antigen on their cell surface.
   These findings suggest that other surface markers, in addition to CD38,
   may also be expressed on the IL36G-producing CiMs.

   Collectively, our data indicate that IL36G-producing neutrophil-like
   monocytes emerge in the context of cancer cachexia and promote skeletal
   muscle wasting and weakness. Our findings provide insights into the
   interplay between dysregulated tumor immunity and the development of
   cancer cachexia. Targeting the CiMs/IL36G signaling axis may be a
   useful therapeutic option for patients with cancer cachexia.

Methods

Ethical statement

   This research complies with all relevant ethical regulations. All
   animal experiments were conducted in accordance with the protocols
   approved by the institutional animal care and use committee at the
   Tokyo University of Pharmacy and Life Sciences. The experiment using
   patients’ samples was approved by the institutional ethics committee at
   Tokyo Metropolitan Cancer and Infectious Diseases Center, Komagome
   Hospital.

Mice

   C57BL/6 J mice and BALB/c mice (7 to 10-week-old) were purchased from
   Tokyo Laboratory Animals Science (Tokyo, Japan). Il36r conventional
   knockout mice (C57BL/6 strain) were purchased from Cyagen Biosciences
   (Santa Clara, CA). For some experiments using 4T1 cells, Il36r
   conventional knockout mice were backcrossed more than 7 times to BALB/c
   mice. Mice were weighed at least twice a week. For ethical reasons,
   when mice developed ≥ 20% weight loss compared to the weight at the
   start of the experiment or tumors approached 20 mm in size (the maximal
   tumor size permitted by the institutional animal care and use
   committee), these mice were euthanized under deep anesthesia
   (endpoint). All animal studies were approved and conducted according to
   the institutional animal care protocol (L18-24, L19-10, L20-07, L21-14,
   L22-08, L23-04, L24-12) and guidelines of the Tokyo University of
   Pharmacy and Life Sciences. All the mice were fed a standard laboratory
   diet (CLEA Rodent Diet CE-2, CLEA Japan) and maintained in a
   pathogen-free environment (23  ±  2 °C, 55  ±  5%) on an 11-h
   light/13-h dark cycle with food and water supplied ad libitum
   throughout the experimental period.

Patient samples

   28 patients with both solid cancer and cachexia (esophageal cancer,
   n = 6 patients; gastric cancer, n = 5 patients; pancreatic cancer,
   n = 4 patients; head and neck cancers, n = 3 patients; lung cancer,
   n = 3 patients; melanoma, n = 2 patients; breast cancer, n = 1 patient;
   renal cancer, n = 1 patient; prostate cancer, n = 1 patient; ovarian
   cancer, n = 1 patient; bladder cancer, n = 1 patient) who were
   hospitalized in the palliative care unit at Tokyo Metropolitan Cancer
   and Infectious Diseases Center, Komagome Hospital (Tokyo, Japan)
   between January 2021 and October 2021 were enrolled in this study. All
   the patients meet the international consensus cachexia definition
   (either “> 5% weight loss over past 6 months” or “> 2% weight loss with
   body mass index (BMI) < 20”). This study was approved by the
   institutional ethics committee, and serum samples were isolated from
   the patients. All patients provided written informed consent for the
   use of anonymized samples and the publication of the clinical
   information in accordance with the Declaration of Helsinki. Detailed
   patients’ data (sex, age, diagnosis, % weight loss over the past 6
   months, BMI, and serum level of IL36G) are provided in Supplementary
   Data [214]7.

Retrovirus and lentivirus generation for mouse BMT experiments

   Plat-E Packaging cells were transfected with retrovirus vectors
   (pMYs-IRES-eGFP, pMYs-NUP98-HBO1-IRES-eGFP) using FuGENE HD (Promega).
   The medium was changed 24 hours after transfection. The supernatant of
   Plat-E Packaging cells containing retroviruses was collected 48 h after
   transfection and concentrated using a polyethylene glycol solution. The
   lentiviral CRISPR-Cas9 vector containing the Cas9-coding sequence and a
   gRNA sequence targeting Il36g (pLV[CRISPR]-hCas9:T2A:Puro-U6
   > mIl36g[gRNA#139]) and Scramble gRNA lentiviral control vector
   (pLV[CRISPR]-hCas9/Puro-U6> Scramble_gRNA1) were cloned by Vector
   Builder. The sequence of Il36g-gRNA is as follows: 5′-
   GGTGTGGATCTTTCGTAATC-3′. 293 T cells were transfected with these
   lentiviral vectors and lentiviral high-titer packaging mix
   (Takara/Clontech) using Lipofectamine 3000 (Thermo Fisher Scientific).
   The supernatant containing the retroviruses and lentiviruses was
   collected 48 hours after transfection and concentrated using a PEG
   solution.

Mouse BMT experiment

   C57BL/6 J donor mice were euthanized via cervical dislocation under
   anesthesia. BM cells were isolated from the femurs and tibias of donor
   mice immediately following euthanasia. BM c-Kit^+ cells were then
   separated using an EasySep mouse cKit positive selection kit (STEMCELL
   Technologies). After incubation with 50 ng/mL mouse stem cell factor
   (SCF), mouse FMS-like tyrosine kinase 3 (FLT3) ligand (FLT3-L), mouse
   interleukin 6 (IL-6), and mouse thrombopoietin (TPO) (PeproTech and
   BioLegend) in Iscove modified Dulbecco medium (IMDM) containing 20% FBS
   and 1% penicillin-streptomycin, cells were transduced with the
   retrovirus for 60 h using RetroNectin (Takara)-coated dishes. Then, 2
   to 5 × 10^6 of the non-sorted cells were injected through the tail vein
   into sublethally irradiated recipient mice. For some experiments, BM
   HSC/Ps isolated from donor mice were resuspended in IMDM containing 20%
   FBS, 50 ng/mL mouse SCF, mouse FLT3-L, mouse IL-6, and mouse TPO and
   immediately transduced with the lentivirus using RetroNectin-coated
   dishes. After 24 h of incubation, the cells were transduced with the
   retrovirus for 48–60 h in a RetroNectin-coated dish under puromycin
   treatment (3 µg/mL). The transduced and puromycin-selected cells were
   injected into the tail vein of sublethally irradiated WT recipient
   mice. NUP98-HBO1-BMT (Mo-Leukemia) mice before the onset of weight loss
   were used as “Mo-Leukemia mice without weight loss”.

Solid tumor models

   To generate syngeneic solid cancer, 4T1 (2.5 to 5.0 × 10^4 cells each)
   and B16 cells (2.5 to 5.0 × 10^4 cells each) were intravenously
   injected into recipient mice. BALB/c mice were used as recipients for
   4T1 cells, and mice of the C57BL/6J strain were used as recipients for
   B16 cells.

Cell culture

   C2C12 (RCB0987) and B16 cells (RCB1283) were purchased from RIKEN BRC
   CELL BANK. 4T1 (CRL-2539) and 293 T cells (CRL-3216) were purchased
   from ATCC. Plat-E cells were kindly provided by Dr. T. Kitamura. C2C12,
   293 T, and Plat-E cells were maintained in Dulbecco’s Modified Eagle
   Medium-low glucose (DMEM) containing 10% FBS and 1%
   penicillin-streptomycin (Fujifilm Wako Pure Chemical Corporation). 4T1
   cells were maintained in RPMI1640 (Fujifilm Wako Pure Chemical
   Corporation) containing 10% FBS and 1% penicillin-streptomycin. B16
   cells were maintained in low-glucose DMEM containing 10% FBS and 1%
   penicillin-streptomycin. DMEM containing 2% horse serum (Thermo Fisher
   Scientific) and 1% penicillin-streptomycin were used for inducing the
   differentiation of C2C12 cells toward myotubes.

CRISPR-Cas9 knockout of Il36r in C2C12 cells

   The lentiviral CRISPR-Cas9 vector containing the Cas9-coding sequence
   and dual gRNA sequences targeting Il36r
   (pLV-2CRISPR-hCas9-T2A-Puro-U6>mIl1rl2-gRNA#2722-U6>mIl1rl2- gRNA#1877)
   were cloned by Vector Builder. The sequences of Il36r-gRNA are as
   follows: 5′-CGTCCTTTCACGTGTCATGC-3′ and 5′-TCTTGTGAACAACGTTGCTG-3′.
   293 T cells were transfected with these lentivirus vectors and
   Lentiviral High Titer Packaging Mix (Takara/Clontech) using
   Lipofectamine 3000 (Thermo Fisher Scientific). The supernatant
   containing lentivirus was collected 48 hours following transfection and
   concentrated using PEG solution. C2C12 cells were then transduced with
   the lentivirus by spin-infection at 2000 × g, 32 °C for 2 h. The
   transduced cells were selected with puromycin (1 µg/mL).

Co-culture assay and Immunofluorescence staining

   FACS-sorted murine CD115^+Ly6C^high monocytes, CD115^+Ly6C^highCD38^+
   monocytes, CD115^+Ly6C^highCD38^‒ monocytes, or CD115^‒Ly6G^+
   neutrophils (5 × 10^5 cells per well) were co-cultured with C2C12
   myotubes for 48 h in 6 well- or 24 well-tissue culture treated plate
   using ThinCert transwell (0.4 μm pore size, Greiner). C2C12 cells were
   then fixed with 4% paraformaldehyde in PBS for 20 min, and then
   15 mmol/L glycine in PBS was added. Following the treatment with 0.2%
   TritonX-100 in PBS, cells were washed with PBS. Following the blocking
   with 2% BSA in PBS, cells were stained with the following antibodies:
   anti- Myosin Heavy Chain (clone, MF20; catalog number, MAB4470; R&D;
   dilution, 5 μg/mL) and Alexa Fluor 488–conjugated anti-mouse IgG
   (H + L) (a21235; Invitrogen; 2 μg/mL). Cells were then washed three
   times with PBS, mounted with Hoechst, and examined using an Olympus
   FV3000-IX83 confocal fluorescence microscope (Olympus) and a BZ-X700
   fluorescence microscope (Keyence). Myotube diameter was measured using
   ImageJ. The number of muscle fibers analyzed is indicated in each
   figure. Representative results from more than three independent
   experiments are shown in the figures.

   For confirmation of Il36r knockout (KO), C2C12 cells with or without
   Il36r KO were fixed and stained with the following antibodies: anti-
   Myosin Heavy Chain (MF20; MAB4470; R&D; 5 μg/mL), anti-Il36r (ab180894;
   Abcam; 10 μg/mL), Alexa Fluor 488–conjugated anti-mouse IgG (H + L;
   a11001; Invitrogen; 2 μg/mL), and Alexa Fluor 647–conjugated
   anti-rabbit IgG (H + L) (a21245; Invitrogen; 2 μg/mL). Cells were then
   washed three times with PBS, mounted with Hoechst, and examined using
   an Olympus FV3000-IX83 confocal fluorescence microscope (Olympus). The
   integrated density of Il36r and the area of myotube was measured using
   ImageJ (NIH, Bethesda, MD).

Flow cytometry

   Following red blood cell lysis with Pharm Lyse (BD Biosciences), murine
   PB and BM cells were stained in FACS buffer (Phosphate-buffered saline
   (PBS) containing 2% fetal bovine serum (FBS) and 2 mM EDTA) with the
   following anti-mouse antibodies: CD115 (clone, AFS98; catalog number:
   135524; Biolegend; dilution: 1:500), Ly6C (HK1.4; 128026, 128008;
   Biolegend; 1:500), Ly6G (1A8; 127612; Biolegend; 1:500), Gr1 (RB6-8C5;
   108408; Biolegend; 1:500), and CD38 (90; 102708, 102712; Biolegend;
   1:500). Cells were then washed with FACS buffer and resuspended in FACS
   buffer containing 7AAD (BioLegend). Annexin V apoptosis detection kit
   (Biolegend) was used to evaluate cell viability. Flow cytometric
   analysis was performed using FACSCanto (BD Biosciences), FACSCelesta
   (BD Biosciences), and CytoFLEX (Beckman Coulter). Cell sorting was
   performed using FACSAriaIII (BD Biosciences) or SH800 (SONY). Data were
   analyzed using FlowJo software (BD Biosciences, RRID:SCR_008520).
   Nucleated cells were gated based on forward scatter area and side
   scatter area (SSC-A), followed by SSC-A/side scatter height gates to
   exclude cell multiplets. After the morphology gates, 7AAD^+ dead cells
   were excluded. Unless otherwise noted, all nucleated cells that appear
   following the regular gating for live singlets. Representative plots
   showing the sequential gating strategy for flow cytometric analysis are
   shown in Supplementary Fig. [215]13. The Gating strategy for collecting
   Ly6C^high monocytes is shown in Supplementary Fig. [216]2.

Bulk RNA-Seq

   BM cells were obtained from control mice transplanted with empty vector
   transduced cells and the Mo-Leukemia mice with/without weight loss.
   Total BM cells were immunostained, and CD115^+ Ly6C^+ cells were
   FACS-sorted. Total RNA was isolated using RNeasy Mini Kit (Qiagen). RNA
   degradation was determined on 1% agarose gel. RNA purity was determined
   using a NanoPhotometer spectrophotometer (IMPLEN). RNA integrity and
   quantitation were assessed using the RNA Nano 6000 Assay Kit and the
   Bioanalyzer 2100 (Agilent Technologies). RNA sequencing was performed
   as described before^[217]58. FASTQ files were processed in AltAnalyze
   (version 2.1.4.4)^[218]29. Transcript per million (TPM) estimates were
   calculated via a built-in call to Kallisto for read pseudoalignment and
   transcript quantification (Ensembl 72, mm10), and differentially
   expressed genes (DEGs; Log2FC > 1.0 or 0.5, P < 0.05) were selected in
   AltAnalyze (any comparison). DEGs were used as input for clustering and
   heatmap visualization. DEG log2 folds relative to the row mean were
   used in generating heatmaps and hierarchical clustering in AltAnalyze.

Single-cell RNA-Seq

   FACS sorted Ly6C^+CD115^+ inflammatory monocytes were washed with
   RPMI1640 containing 10% FBS, followed by filtration with a 40 µm
   strainer and then resuspended in an appropriate amount of the media.
   Approximately 5000 cells with > 90% cell survival rate for each sample
   were used for GEM (Gel Bead in the emulsion) generation using Chromium
   Controller (10x Genomics), where RNA molecules are tagged in each cell
   separately. The molecular-tagged RNA fragments were converted into
   cDNA, followed by library establishment for next-generation sequencing
   using Chromium Next GEM Single Cell 3’ Reagent Kits v3.1 (10x
   Genomics). The resulting libraries were quantified by Qubit dsDNA Assay
   (Thermo Fisher Scientific) and TapeStation D1000 (Agilent). The
   libraries were sequenced by the Illumina HiSeq platform with the
   following configuration: read 1: 28 bp, read 2: 91, and i7 index: 8 bp,
   yielding approximately 400 M paired-end reads per sample. Image
   analysis and base calling were conducted using software on the HiSeq
   instrument. Raw FASTQ reads were imported into the Cell Ranger pipeline
   (v3.0.2, 10x Genomics) for mapping and gene expression count analyses.
   The single-cell gene expression study mentioned above was conducted by
   Azenta Life Sciences (formerly Genewiz). Using the
   filtered_feature_bc_matrix, additional unsupervised ICGS was performed
   in the AltAnalyze pipeline (version 2.1.4.4). A moderated t test from
   AltAnalyze was used with P < 0.05 and Log2FC > 1.0. The minimum number
   of changed genes was three.

Pathway and gene cluster enrichment analysis

   GO-term and pathway enrichment analyses of differentially expressed
   genes were performed using ToppFun in the ToppGene Suite
   ([219]https://toppgene.cchmc.org/)^[220]59. The following parameters
   were used: P < 0.05, 15 < gene size < 500. Enrichment analysis of
   cluster genes was performed using GO-term-gene-set and
   biomarker-gene-set of the GO-Elite algorithm in AltAnalyze.

Gene set enrichment analysis (GSEA)

   The data set of upregulated DEGs in CiMs was tested for enrichment in
   bulk RNA-Seq data sets of murine inflammatory monocytes obtained from
   Mo-Leukemia mice with/without weight loss, GMP- and MDP-derived
   inflammatory monocytes ([221]GSE88982), inflammatory monocytes obtained
   from mice with breast cancer bone metastasis ([222]GSE152795), and
   inflammatory monocytes obtained from patients with advanced colorectal
   cancer ([223]GSE47756) and renal cell carcinoma ([224]GSE38424) using
   GSEA version v4.2.3 (RRID:SCR_003199)^[225]60. The following parameters
   were used: 1000 gene_set permutations, Diff_of_Classes ranking metric
   and descending order real mode for gene sorting for Supplementary
   Fig. [226]4c; 1000 gene set permutations, Signal2Noise ranking metric,
   and descending order real mode for gene sorting for Supplementary
   Figs. [227]7c and [228]8a, b.

Complete blood counts

   PocH-100iV Diff (Sysmex) was used for the measurement of blood cell
   counts.

HE staining

   Quadriceps femoris muscle was fixed using a 10% formalin neutral buffer
   solution (Fujifilm Wako Pure Chemical Corporation) and
   paraffin-embedded for hematoxylin-eosin (HE). Staining was performed at
   Genostaff (Tokyo, Japan). Cross-sectional area (CSA) was quantified
   from > 200 fibers within randomly chosen fields for each mouse using
   ImageJ (RRID:SCR_003070).

Cytospin and Diff-Quik staining

   FACS-sorted PB and BM cells were resuspended in 100 μL of 100% FBS and
   spun onto Matsunami SUPERFROST slides (Matsunami Glass) for 5 min at
   500 rpm using a Cyto-Tek 2500 cytocentrifuge (Sakura Finetek). Slides
   of mice PB and BM cells were then fixed and stained using Diff-Quik
   (Sysmex).

Grip strength test

   Forelimb grip strength was measured using the MK-380Si grip strength
   meter (Muromachi Kikai). Mice were placed on a metal mesh and allowed
   to grip it with forelimbs. The tail was horizontally pulled backward.
   The peak tension was recorded at the time the mice released the mesh.
   Measurement was repeated ten times and the mean was calculated.

Quantitative RT-PCR

   The total RNA of hematopoietic cells and C2C12 cells was isolated using
   RNeasy Mini or Micro kit (Qiagen) and reverse transcribed to cDNA using
   RevertraAce (Toyobo). C2C12 cells for Quantitative RT-PCR were
   collected 24 h after the start of the co-culture assay. Murine skeletal
   muscle was homogenized using Biomasher III (Nippi) for RNA extraction.
   The total RNA of muscle tissue was isolated using RNeasy Fibrous Tissue
   Mini Kit (Qiagen) and reverse transcribed to cDNA using RevertraAce
   (Toyobo). Quantitative PCR was performed using Applied Biosystems Step
   One Plus thermal cycler (Applied Biosystems), Power SYBR Green PCR
   master mix (Applied Biosystems), and the following primers: for mouse
   Il36g, 5’-AGAGTAACCCCAGTCAGCGTG-3’ and 5’-AGGGTGGTGGTACAAATCCAA-3’; for
   mouse Trim63, 5’-GTGTGAGGTGCCTACTTGCTC-3’ and
   5’-GCTCAGTCTTCTGTCCTTGGA-3’; for mouse Fbxo32,
   5’-CAGCTTCGTGAGCGACCTC-3’ and 5’-GGCAGTCGAGCCGTCCAGTC-3’; for mouse
   Lcn2, 5’-CTGAATGGGTGGTGAGTGTG-3’ and 5’-GCTCTCTGGCAACAGGAAAG-3’; for
   mouse S100a8, 5’-CCTTTGTCAGCTCCGTCTTC-3’ and
   5’-TAGAGGGCATGGTGATTTCC-3’; for mouse Chil3, 5’-GAAGGAGCCACTGAGGTCTG-3’
   and 5’-CACGGCACCTCCTAAATTGT-3’; for mouse Elane,
   5’-ACTCTGGCTGCCATGCTACT-3’ and 5’-GCCACCAACAATCTCTGAGG-3’; for mouse
   Ctsg, 5’-CAACGGTTCTGGAAAGATGC-3’ and 5’-CTTCTCGGCCTCCAATGAT-3’; for
   mouse Prtn3, 5’-AGCTACCCATCCCCCAAG-3’ and 5’-TCGTGCCCACCTACAATCTT-3’;
   for mouse Actb, 5’-ACACCCGCCACCAGTTC-3’ and 5’-TACAGCCCGGGGAGCAT-3’;
   and for mouse Hprt, 5’-TTGTATACCTAATCATTATGCCGAG-3’ and
   5’-CATCTCGAGCAAGTCTTTCA-3’.

Administration of LPS and unmethylated CpG DNA

   LPS (L6529, Sigma; 2 µg/mouse) was injected intraperitoneally once
   every other day for a total of 2 injections. CpG oligonucleotide (D35,
   CN-65001, GeneDesign; 12 µg/mouse) was injected intraperitoneally once
   every other day up to 14 times. Mice were analyzed 24 hours after final
   administration.

TAK-242 Treatment

   From day 10 after tumor cell injection, B16 and 4T1 metastasis models
   were treated with 3 mg/kg TAK-242 (S7455, Selleck) or DMSO vehicle
   intraperitoneally daily up to 15 times.

ELISA

   FACS-sorted 1 × 10^5 indicated cells were seeded on a 96-well flat
   bottom plate and incubated in 100 μL RPMI1640 for 24 h. The culture
   medium was then immediately snap-frozen in liquid nitrogen and stored
   at − 80 °C until usage. Il36g concentrations were measured using a
   Mouse Interleukin-36 gamma ELISA Kit (MyBioSource) according to the
   manufacturer’s instructions. Absorbance was measured at 450 nm using
   EnSpire (Perkin Elmer). For serum from patients with cachexia, IL36G
   concentrations were measured using a Human IL36G ELISA kit
   (Invitrogen). SpectraMax ABS (Molecular Device) was used for the
   measurement of absorbance at 450 nm.

Statistics and reproducibility

   GraphPad Prism (v8) (GraphPad, RRID:SCR_002798) was used. Statistical
   analyses for the comparison of two independent, normally distributed
   samples were compared using the two-tailed Student’s t test. For
   multiple pairwise comparisons, the data were analyzed by one-way ANOVA
   followed by the Tukey multiple comparison test. For pathway or GO-term
   enrichment analysis, Fisher’s exact test was used. An empirical Bayes
   moderated t-test was used for the identification of differentially
   expressed genes in AltAnalyze. Statistical significance was set at
   P < 0.05. No statistical method was used to predetermine the sample
   size. No data were excluded from the analyses. The experiments were not
   randomized. The Investigators were not blinded to allocation during
   experiments and outcome assessment.

Reporting summary

   Further information on research design is available in the [229]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [230]Supplementary Information^ (3.9MB, pdf)
   [231]Peer Review File^ (2.8MB, pdf)
   [232]41467_2024_51873_MOESM3_ESM.docx^ (13.1KB, docx)

   Description of Additional Supplementary Information
   [233]Supplementary Data 1^ (130.2KB, xlsx)
   [234]Supplementary Data 2^ (49.7KB, xlsx)
   [235]Supplementary Data 3^ (13.7KB, xlsx)
   [236]Supplementary Data 4^ (16.9KB, xlsx)
   [237]Supplementary Data 5^ (10MB, xlsx)
   [238]Supplementary Data 6^ (10.1KB, xlsx)
   [239]Supplementary Data 7^ (12.4KB, xlsx)
   [240]Reporting Summary^ (2.9MB, pdf)

Source data

   [241]Source Data^ (76.9KB, xlsx)

Acknowledgements