Depletion of stromal cells expressing fibroblast activation protein-α from skeletal muscle and bone marrow results in cachexia and anemia

Abstract

Fibroblast activation protein-α (FAP) identifies stromal cells of mesenchymal origin in human cancers and chronic inflammatory lesions. In mouse models of cancer, they have been shown to be immune suppressive, but studies of their occurrence and function in normal tissues have been limited. With a transgenic mouse line permitting the bioluminescent imaging of FAP(+) cells, we find that they reside in most tissues of the adult mouse. FAP(+) cells from three sites, skeletal muscle, adipose tissue, and pancreas, have highly similar transcriptomes, suggesting a shared lineage. FAP(+) cells of skeletal muscle are the major local source of follistatin, and in bone marrow they express Cxcl12 and KitL. Experimental ablation of these cells causes loss of muscle mass and a reduction of B-lymphopoiesis and erythropoiesis, revealing their essential functions in maintaining normal muscle mass and hematopoiesis, respectively. Remarkably, these cells are altered at these sites in transplantable and spontaneous mouse models of cancer-induced cachexia and anemia. Thus, the FAP(+) stromal cell may have roles in two adverse consequences of cancer: their acquisition by tumors may cause failure of immunosurveillance, and their alteration in normal tissues contributes to the paraneoplastic syndromes of cachexia and anemia.

Figure 1.

The tissue distribution of FAP⁺ stromal cells in the adult mouse. (A) A Fap-containing BAC was modified by inserting at the start codon a cassette encoding the DTR, Luc2, and mCherry linked by E2A sequences. (B) BAC transgenic mice in which luciferase and the DTR are expressed in FAP⁺ stromal cells and littermate mice were assayed for bioluminescence before and 24 h after two daily doses of 25 ng/g DTX. (C) Organs that had been dissected from DTX- or vehicle-treated BAC transgenic mice from B were assessed for bioluminescence. (B and C) The scale represents relative counts. (D) FAP mRNA levels were measured in different tissues by qRT-PCR. (E) FAP mRNA levels were correlated with those of luciferase, a marker of the expression of the BAC transgene, across different tissues by qRT-PCR. (F) Tissues were enzymatically dissociated, and cells were stained with antibodies to FAP and CD45 for analysis by flow cytometry. Numbers represent the percentage of live cells gated as FAP⁺. (G) CD45⁻FAP⁺, CD45⁻FAP⁻, and CD45⁺ cells were sort purified from skeletal muscle and assessed by qRT-PCR for FAP and luciferase/transgene mRNA. Error bars represent SEM. MOE, main olfactory epithelium. Data are representative of more than three independent analyses (n ≥ 3; B–G).

Figure 2.

Analysis of the transcriptomes of FAP⁺ cells from adipose tissue, skeletal muscle, and pancreas. (A) FAP⁺ cells were assessed by flow cytometry for their expression of the membrane proteins CD90 (Thy1), CD140a (PDGFRα), and Sca-1 (Ly6A/E), which are characteristic of certain mesenchymal cells. Filled histograms represent isotype staining, and the black lines denote the specific stains. (B) The RNA-Seq data for each of the cell types were log transformed and modeled as bimodal distributions. The %int represents the degree of heterogeneity within each population. The density value is indicated in blue. The yellow line represents the bimodal fit for the lowly expressed transcripts and the purple line for the highly expressed transcripts, and the red line indicates the overall density value. (C) The Pearson’s r² and Spearman’s ρ correlations were calculated for the RPKM gene expression levels of FAP⁺ cells, FAP⁻ MEFs, and CD4⁺ T cell subsets after removal of low expressed/nonexpressed genes. (D) PCA analysis of VST (counts) of FAP⁺ cells, CD4⁺ T cell subsets, and MEFs is shown. (E) VST clustering of RPKM levels of replicate samples of FAP⁺ cells, CD4⁺ T cell subsets, and MEFs is shown. (F) The percentage of differentially expressed genes among the different cells was determined by DESeq and edgeR using count data. The highlighted cells here and in C show the correspondence between high levels of similarity in these two analyses of gene expression. Th1, Th2, and iTreg refer to the CD4⁺ T cell subsets. VA, SkM, and Pan refer to FAP⁺ cells from visceral adipose tissue, skeletal muscle, and pancreas, respectively. See also Table S1.

Figure 3.

Depletion of FAP⁺ stromal cells and loss of skeletal muscle mass. (A) BAC transgenic mice and littermates were treated with DTX for 2 d, and total body weight and food intake were measured each day during the ensuing 3 wk. Arrows indicate when DTX was given, and the red highlighting shows the period over which it was being administered. (B) The mass of the quadriceps muscles from BAC transgenic mice and littermates was measured on days 3 and 22 after DTX treatment. (C) The mean cross-sectional area of the fibers of the quadriceps muscles from BAC transgenic mice and littermates taken 22 d after DTX treatment was measured, after fibers were delineated by staining with antibody to the basal lamina. Bar, 100 µm. (D) The number of FAP⁺ cells and the level of FAP mRNA in quadriceps muscles of BAC transgenic mice as a percentage of littermates were measured 3 and 22 d after DTX treatment. (E) The mRNA levels for atrogin-1, MuRF1, Fst, Lama2, and Mstn in quadriceps muscles of BAC transgenic mice and littermates were measured 3 and 22 d after DTX treatment. (F) The levels of Fst protein in quadriceps muscles of BAC transgenic mice and littermates were measured by immunoblot analysis 10 d after DTX treatment. (G) Quadriceps muscle from a normal mouse was stained with antibodies to FAP and laminin and assessed by confocal microscopy. (H) The FAP⁺ subset of enzymatically dissociated cells from quadriceps muscle were sort purified, and the mRNA levels of Lama2, Fst288, and Fst315 in FAP⁺ cells were determined by RNA-Seq. (I) The indicated subsets of enzymatically dissociated cells from quadriceps muscle were sort purified, and the mRNA levels of Fst, Lama2, Mstn, and FAP were determined by qRT-PCR. Error bars represent SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Data are representative of more than three independent analyses (n ≥ 3; A) or are representative of two independent analyses (n ≥ 5; B–F).

Figure 4.

FAP⁺ cells of the bone marrow and hematopoiesis. (A) The cellularity of the femoral bone marrow from BAC transgenic mice and littermates was determined by hematoxylin and eosin staining after 3-d treatment with DTX. (B) Individual populations of hematopoietic cells from the femoral bone marrow of BAC transgenic mice and littermates were analyzed by flow cytometry after 3-d treatment of the mice with DTX. (C) 7 d after DTX administration, samples of peripheral blood from transgenic and littermate mice were assayed for hemoglobin, erythrocyte numbers, mean corpuscular hemoglobin, and mean corpuscular volume. (D) The proportions of FAP⁺ cells in enzyme-dissociated bone and bone marrow preparations from BAC transgenic mice and littermates were assessed by flow cytometry after 3-d treatment of the mice with DTX. (E) The mRNA levels of Cxcl12 and KitL in cells of the femoral bone marrow from BAC transgenic mice and littermates were measured by qRT-PCR after 3-d treatment of the mice with DTX. (F) The protein levels of Cxcl12 and KitL in lysates of bone marrow from BAC transgenic mice and littermates were measured by ELISA after 3-d treatment of the mice with DTX. (G) A cross-section of normal mouse femur was stained with antibodies to FAP, osteopontin, and osteocalcin and assessed by confocal microscopy. (H) Subsets of enzymatically dispersed cells from femoral bone and marrow were FACS purified, and the sorted populations were assessed by qRT-PCR for expression of FAP, Cxcl12, KitL, osteopontin, osteocalcin, and Lepr. Error bars represent SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Data are representative of at least two independent analyses (n ≥ 4; A, B, D, and E) or one confirmatory analysis (n = 6; F). Combined data are from three independent experiments (n ≥ 4; B).

Figure 5.

The transplanted C26 carcinoma, cachexia, and systemic alterations in FAP⁺ stromal cells. (A) Whole body weight and food intake of C57BL/6 × BALB/c F1 mice with and without ectopic C26 tumors were measured during the 3–4-wk post-tumor implantation period. (B) C26 cells were subcutaneously inoculated into BAC transgenic C57BL/6 × BALB/c F1 mice. Whole body bioluminescence was quantified for the entire dorsal surface of the mice at the indicated times. (C) The levels of bioluminescence of the excised C26 tumor and several tissues of tumor-bearing mice and littermate controls were assessed. (B and C) Scale represents relative counts. Error bars represent SEM. Data are representative of three independent analyses (n = 4 for A and B; representative n = 1 for C).

Figure 6.

Alterations of FAP⁺ cells in the quadriceps muscle and bone marrow of cachectic mice bearing C26 tumors. (A and B) The quadriceps muscles from cachectic, C26-bearing and nontumor-bearing littermates were assessed for mass (A) and the mRNA levels of atrogin-1, MuRF1, Mstn, FAP, Fst, and Lama2 (B). (C) The levels of Fst protein in lysates of quadriceps muscles from cachectic, C26-bearing and nontumor-bearing littermates were measured by immunoblot analysis. (D) Individual populations of hematopoietic cells from the femoral bone marrow of cachectic, C26-bearing and nontumor-bearing littermates were assessed by flow cytometry. (E) The levels of Cxcl12 and KitL protein in lysates of cells recovered from the femoral bone marrow of cachectic, C26-bearing and nontumor-bearing littermates were measured by ELISA. Error bars represent SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Combined data are from three independent analyses (n = 1–5).

Figure 7.

Alterations of FAP⁺ cells in the quadriceps muscle and bone marrow of KPC mice with PDA. (A and B) The quadriceps muscles from PDA-bearing KPC mice with and without cachexia and littermates were assessed for mass (A) and the mRNA levels of atrogin-1 MuRF1, FAP, Fst, Lama2, and Mstn (B). (C) Peripheral blood from PDA-bearing mice and nontumor-bearing littermates was assayed for hemoglobin, erythrocyte number, mean corpuscular hemoglobin, and mean corpuscular volume. (D) Individual populations of hematopoietic cells from the femoral bone marrow of PDA-bearing mice and nontumor-bearing littermates were assessed by flow cytometry. (E) The levels of Cxcl12 and KitL protein in lysates of cells recovered from the femoral bone marrow of PDA-bearing KPC mice and nontumor-bearing littermates were measured by ELISA. (F) The mRNA levels of FAP in heart and salivary glands from KPC, PDA-bearing mice with and without cachexia and littermates were assessed by qRT-PCR. Error bars represent SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Combined data are from three independent analyses (n = 1–5).