Abstract

   Osteolytic lesions and pathological fractures are hallmark
   manifestations of multiple myeloma (MM), profoundly influencing the
   quality of life and self-care ability of MM patients. By analyzing
   transcriptome data and single-cell RNA-seq data from our center and the
   GEO database, a subset of MM cells with high expression levels of
   chemokine CXCL7 was identified. This subset of MM cells possesses a
   high capacity for proliferation and activation of osteoclast signaling
   pathway. CXCL7 might be a crucial regulator of osteolytic damage in MM.
   Subsequently, the association between the expression level of CXCL7 and
   pathological fractures in patients was investigated, and the impact of
   CXCL7 on MM proliferation was confirmed both in vivo and in vitro. A
   mouse xenograft tumor model was established through intravenous
   injection of myeloma cell lines based on the homing ability of plasma
   cells. Moreover, the mechanism by which CXCL7 promotes the activation
   of osteoclast signaling pathways in MM was explored. Our findings
   reveal that elevated CXCL7 levels significantly enhance MM cell
   proliferation, increasing the risk of pathological fractures in MM
   patients. Additionally, our mouse xenograft tumor model demonstrated
   that CXCL7 can induce femoral fractures and reduce bone mineral
   density. Concurrently, it was discovered that CXCL7 could activate the
   JAK/STAT3 pathway via CXCR2 and upregulate the expression levels of
   MMP13 and C-myc, facilitating MM cell proliferation and activation of
   the osteoclast signaling pathway. Our study offers novel insights into
   the pathogenic mechanism of osteolytic lesions and implies that
   targeting CXCL7 may present a new therapeutic avenue for MM.

   Subject terms: Myeloma, Oncogenes

Introduction

   Multiple myeloma (MM) is a difficult-to-cure malignant tumor with a
   relatively slow progression and high intra-tumor heterogeneity [[32]1].
   MM presents with diverse clinical manifestations, target organ damage,
   and prognosis among different patients [[33]2]. The emergence of
   osteolytic lesions and the consequent occurrence of pathological
   fractures represent one of the crucial factors affecting the quality of
   life and prognosis of patients with myeloma. Pathological fractures
   also serve as a significant cause for patients with myeloma to lose
   their self-care capacity in the terminal stage [[34]3, [35]4].
   Osteolytic injury in myeloma is driven by two key factors. First, the
   proliferation of myeloma cells suppresses normal bone marrow function.
   Second, the interaction between myeloma cells and the bone marrow
   microenvironment activates osteoclast-related pathways, leading to
   osteolytic damage [[36]5]. Patients with MM often experience relatively
   long overall survival. In fact, as opposed to the complete eradication
   of monoclonal plasma cells, modulating the biological behavior of
   myeloma cells to extend the survival period with tumor and concurrently
   control the degree of target organ damage could be a more efficacious
   therapeutic modality. Identifying key genes that influence myeloma cell
   behavior, particularly those involved in regulating osteoclast
   signaling, is therefore of significant clinical importance.

   Through a comprehensive analysis of RNA-seq and single-cell sequencing
   data of myeloma, we identified that in myeloma, CXCL7 has the capacity
   to facilitate osteolytic destruction and the proliferation of myeloma.
   Chemokine CXCL7 pertains to a category of chemotactic cytokines and
   assumes a crucial role in regulating the migration and functionality of
   immune cells [[37]6]. In tumors, chemokines not only direct the
   migration of immune cells to influence the anti-tumor immune response
   [[38]7] but also regulate the proliferation, invasion, metastasis, and
   angiogenesis of tumor cells by activating diverse signaling pathways
   such as STAT3 and NF-κB [[39]8, [40]9]. However, the exact mechanisms
   by which CXCLs influence tumors remain unclear. Plasma cells, derived
   from B cells, are influenced by chemokines that regulate their
   infiltration, proliferation, and interactions with stromal cells in
   tissues, affecting their overall biological functions. That is to say,
   chemokines might play a decisive role in immune cell tumors. Previous
   studies have demonstrated that CXCL7 exerts a role in promoting
   proliferation and invasion in malignant tumors such as breast cancer,
   lung cancer, and renal cancer, and is associated with an adverse
   prognosis [[41]10–[42]12]. However, the role of CXCL7 in MM remains
   indistinct. This study aims to explore the role and related mechanisms
   of CXCL7 in MM to furnish a theoretical basis for identifying novel
   therapeutic targets.

Results

CXCL7 is one of the key genes associated with pathological fractures in
myeloma

   Our research aimed to identify key genes affecting MM cell
   proliferation and osteoclast signaling. We collected bone marrow
   specimens from 30 newly diagnosed MM (NDMM) patients: 15 with more than
   five osteolytic lesions and pathological fractures, and 15 without
   osteolytic damage. After CD138 magnetic bead sorting, we performed
   high-throughput transcriptome sequencing, identifying 52 differentially
   expressed genes (|logFC | > 1, FDR < 0.01) (via DESeq2) (Fig. [43]1A).

Fig. 1. CXCL7 plays a crucial role in the proliferation of multiple myeloma
and the activation of osteoclast signaling pathways.

   [44]Fig. 1
   [45]Open in a new tab

   A The volcano plot of differentially expressed genes in the plasma
   cells from 15 MM patients with pathological fracture and 15 MM patients
   without pathological fracture (|logFC | > 1, FDR < 0.05). B Results of
   weighted gene co-expression network analysis of RNA-seq data in the TT2
   set. C Correlation between different co-expression modules and
   “Osteolysis signaling“ and “B cell proliferation”. The relationship
   between genes in the dark magenta co-expression module and “Osteolysis
   signaling”(D)/“B cell proliferation”(E) pathway. F A total of 11 hub
   genes related to osteoclast signaling pathways and B cell proliferation
   were discovered in the TT2 cohort. A total of 52 differentially
   expressed genes in the plasma cells of 30 patients with/ without
   pathological fracture were determined. The intersection gene was CXCL7.
   The expression level of CXCL7 and the activation of “Osteolysis
   signaling”(G)/“B cell proliferation”(H) pathway in TT2 set. The
   expression level of CXCL7 and the activation of “Osteolysis
   signaling”(I)/“B cell proliferation”(J) pathway in TT3 set).

   Next, we analyzed the [46]GSE24080 dataset, which includes two cohorts
   of NDMM patients: the Total Therapy 2 [[47]13] (TT2) cohort (340 cases)
   and the Total Therapy 3 [[48]14] (TT3) cohort (214 cases). TT2 cohort
   was designated as a training set, while TT3 cohort was utilized as a
   validation set. WGCNA analysis categorized the 23,517 genes in the TT2
   set into 33 modules (Fig. [49]1B). Subsequently, ssGSEA was adopted to
   quantify the activation of “B cell proliferation” and “osteoclast
   signaling pathway” for all patients. Correlations between the
   co-expressed gene modules from WGCNA and pathway activation were
   examined. The results demonstrated that the dark magenta module
   significantly facilitated the activation of the “B cell proliferation”
   (P = 5.6e−11) and “osteoclast signaling” (P = 9.0e−17) pathways (Fig.
   [50]1C). Then, the correlations of all genes within the dark magenta
   module with these two traits were computed (Fig. [51]1D–E). Based on
   the truncation criterion (|Module membership | > 0.70), 11 genes were
   identified as hub genes. By intersecting the differentially expressed
   genes and the hub genes, we found that CXCL7 was a potential key
   regulator (Fig. [52]1F).

   Further analysis in the TT2 cohort confirmed that CXCL7 expression was
   positively correlated with both “B cell proliferation” (P < 0.0001)
   (Fig. [53]1G) and “osteoclast signaling” (P < 0.0001) (Fig. [54]1H).
   This correlation was also validated in the TT3 cohort (Fig. [55]1I–J).

MM cells featuring a high expression level of CXCL7 possess enhanced
osteoclast signaling pathway activity

   Given the high intratumor heterogeneity of MM, we aim to further
   explore the characteristics of MM cell subsets with elevated CXCL7
   expression. Bone marrow samples were collected from three NDMM patients
   with pathological fractures for single-cell RNAseq. The data from
   individual cells across all patients were integrated and analyzed using
   UMAP downscaling (Fig. [56]2A–C). In addition to high CXCL7 expression
   in monocytes, we observed a significant upregulation of CXCL7 in some
   plasma cells as well (Fig. [57]2D). Further analysis revealed 16
   distinct clonal subpopulations of MM cells (Fig. [58]2E). CXCL7
   expression was most prominent in the mm03 subgroup (Fig. [59]2F).
   Although CXCL7 is expressed only in a subset of MM cells, subgroup
   cells with an elevated CXCL7 expression can be identified in all three
   patients (Fig. [60]2G).

Fig. 2. Single-cell RNA-seq findings confirmed that the activation
intensities of the osteoclast signaling pathway and proliferation signaling
pathway in the MM cell subpopulation with high expression of CXCL7 were more
prominent.

   [61]Fig. 2
   [62]Open in a new tab

   The samples were obtained from the bone marrow of three initially
   diagnosed multiple myeloma (MM) patients with pathological fractures.
   The three patients were respectively designated as MM01, MM02 and MM03.
   A Cell clustering results from single-cell RNA-seq in three newly
   diagnosed MM patients. B The cell subpopulation profiles of bone marrow
   samples from three patients with NDMM. C The distribution of bone
   marrow samples from three NDMM patients. D Expression of CXCL7 in all
   cells. E. Cell clustering results from single-cell RNA-seq of MM cells.
   F CXCL7 was mainly expressed in mm03 subgroup. G The distribution of MM
   cells among different patients. Activation level of the “Osteoclast
   signaling” (H) and “B-cell proliferation” (I) pathway in various MM
   subsets. J The mm03 subset exhibited a strong correlation with the “B
   cell proliferation” and “osteoclast signaling” pathways. K Activation
   status of the “JAK/STAT3” pathway in MM cells. In the mm03
   subpopulation, the JAK/STAT3 pathway was strikingly activated.
   Additionally, a significant co-localization of the expression of CXCL7
   and JAK/STAT3 was discerned).

   We selected “B cell proliferation” and “osteoclast signaling” as target
   pathways and conducted signaling pathway enrichment analysis. The
   results showed that the mm03 subgroup had the highest median activation
   of the “osteoclast signaling” pathway, with “B cell proliferation”
   ranking second (Fig. [63]2H–I). Moreover, the mm03 subgroup
   demonstrated the most significant correlation with both pathways (Fig.
   [64]2J).

CXCL7 is associated with pathological fractures in patients

   We collected bone marrow samples from 108 NDMM patients from 2016 to
   2020 and assessed CXCL7 expression using qPCR. All patients received an
   initial combination treatment regimen of proteasome inhibitors,
   immunomodulators, and dexamethasone. We determined the optimal cut-off
   value for CXCL7 and compared the prognosis and clinicopathological
   disparities among patients with different expression levels of CXCL7
   (Fig. [65]3A–B). The findings revealed that patients in the high CXCL7
   expression group had a higher likelihood of pathological fracture
   (P = 0.048) (Fig. [66]3C) and extramedullary invasion (P = 0.010) (Fig.
   [67]3D). Moreover, elevated levels of CXCL7 were associated with
   reduced overall survival (OS) (P = 0.016) and progression-free survival
   (PFS) (P = 0.008) (Fig. [68]3E–F). However, no significant correlation
   was identified between CXCL7 and ISS stage or high-risk cytogenetics.
   The clinical baseline characteristics of the patients are presented in
   Table [69]1. We then investigated the association between CXCL7
   expression levels and the number of osteolytic lesions in patients.
   Osteolytic lesions were evaluated via PET-CT. Spearman correlation
   analysis revealed a significant positive correlation between CXCL7
   expression and the number of osteolytic lesions (P = 0.034) (Fig.
   [70]3G).

Fig. 3. Prognostic implications and clinicopathological characteristics of
patients stratified based on varying levels of CXCL7 expression.

   [71]Fig. 3
   [72]Open in a new tab

   A The optimal cut-off value of the relative expression level of CXCL7.
   B The scatter plot illustrates the distribution of patient groups in
   relation to the gene expression levels of CXCL7. C Probability of
   pathological fracture in patients with different CXCL7 expression
   levels. D Probability of extramedullary invasion in patients with
   different CXCL7 expression levels. E Survival curves for overall
   survival time of patients in different CXCL7 groups. F Survival curves
   for progression-free-survival time of patients in different CXCL7
   groups. G Correlation between the relative expression level of CXCL7
   and the number of osteolytic bone lesions detected by PET-CT in 108
   patients. H Survival curves for overall survival time of patients in
   different CXCL7 groups in the MMRF database).

Table 1.

   Baseline characteristics of 108 patients with multiple myeloma.
   Variable Total CXCL7_High CXCL7_Low
   Patinets (No) 108 69 39
   Gender (No,%)
    Male 74 (68.5) 46 (66.7) 28 (71.8)
    Female 34 (31.5) 23 (33.3) 11 (28.2)
   Age (year, (No, %))
    <60 30 (27.8) 15 (21.7) 15 (38.5)
    ≥60 78 (72.2) 54 (78.3) 24 (61.5)
   ISS stage (No,%)
    I 43 (39.8) 29 (42.0) 14 (35.9)
    II 28 (25.9) 16 (23.2) 12 (30.8)
    III 37 (34.3) 24 (34.8) 13 (33.3)
   FISH high-risk (No,%)
    Detected 34 (31.5) 23 (33.3) 11 (28.2)
    Not detected 74 (68.5) 46 (66.6) 28 (71.8)
   Follow-up time (month, (median, range)) 48.7 (2.1–103.0) 46.3
   (2.1–77.6) 52.2 (5.4–103.0)
   Progress (No,%) 53 (49.1) 41 (68.1) 12 (30,8)
   Dead (No,%) 34 (31.5) 27 (39.1) 7 (17.9)
   [73]Open in a new tab

   To further validate the clinical relevance of CXCL7, we assessed its
   prognostic value in the Multiple Myeloma Research Foundation (MMRF)
   database, the largest publicly available dataset of MM patients. A
   total of 784 NDMM patients with complete prognostic data and RNA-seq
   data were included in the study. Survival analysis, based on the
   optimal cutoff value, revealed that higher CXCL7 expression was
   significantly associated with shorter OS time (P = 0.006) (Fig.
   [74]3H).

CXCL7 promotes proliferation of MM cells

   In CCLE database, CXCL7 expression was found to be low in the NCI-H929
   cell line, moderate in the RPMI 8226 cell line, and high in the U266
   cell line (Fig. [75]4A). Lentiviruses carrying CXCL7-overexpression
   (CXCL7-OE) and Normal-Control (NC) were used to infect NCI-H929 and
   RPMI 8226 cell lines. Meanwhile, siRNA was employed for CXCL7 knockdown
   (CXCL7-KD) in U266 and RPMI 8226 cell lines. CXCL7 RNA levels were
   measured by qPCR (Fig. [76]4B–E). Protein expression, both
   intracellular and secreted, was assessed using Western blotting (Fig.
   [77]4F–I) and ELISA (Fig. [78]4J–M).

Fig. 4. CXCL7 facilitates the proliferation, migration and invasion of MM
cells.

   [79]Fig. 4
   [80]Open in a new tab

   A. The expression level of CXCL7 in MM cell lines within the CCLE
   database. The expression level of CXCL7 in the CXCL7-OE and NC H929
   (B)/RPMI 8226 (C) cell lines was evaluated via qPCR. The expression
   level of CXCL7 in the CXCL7-KD and NC U266(D)/RPMI 8226 (E) cell lines
   was evaluated via qPCR. Western blot analysis of the expression level
   of CXCL7 protein in the CXCL7-OE and NC H929(F)/RPMI8226 (G) cell
   lines. Western blot analysis of the expression level of CXCL7 protein
   in the CXCL7-KD and NC U266 (H)/RPMI8226 (I) cell lines. ELISA was
   utilized to assess the levels of CXCL7 extracellular secreted protein
   in the CXCCL7-OE and NC H929 (J)/RPMI8226 (K) cell lines. ELISA was
   utilized to assess the levels of CXCL7 extracellular secreted protein
   in the CXCCL7-KD and NC U266 (L)/RPMI8226 (M) cell lines. N–Q The CCK8
   assay was employed to assess the differences in cell proliferation
   rates among CXCL7-OE, CXCL7-KD, and NC group. R, S Subcutaneous
   xenograft tumors of mice in CXCL7-OE group and NC group. T The mean
   growth rate of subcutaneous xenograft tumors in CXCL7-OE and NC mice. U
   The expression levels of CXCL7 and Ki-67 in the subcutaneous xenograft
   tumors of the CXCL7-OE group and NC group (via immunohistochemical).

   The CCK-8 assay was used to evaluate cell proliferation. The results
   showed a significantly higher proliferation rate in the CXCL7-OE group
   for both H929 and RPMI 8226 cell lines (Fig. [81]4N–O). In contrast,
   the CXCL7-KD group exhibited slower proliferation in U266 and RPMI 8226
   cells (Fig. [82]4P–Q). Subcutaneous xenograft tumor models were
   established in B-NGD mice with CXCL7-OE and NC RPMI 8226 cell lines.
   Analysis revealed accelerated tumor growth in the CXCL7-OE group (Fig.
   [83]4R–T). Immunohistochemistry performed on excised tumors
   demonstrated elevated levels of Ki-67 proliferation index in the
   CXCL7-OE group (Fig. [84]4U).

CXCL7 facilitate incidence of pathological fractures in mouse

   Subsequently, a murine xenograft model was established via intravenous
   infusion of MM cells to scrutinize the colonization of CXCL7-OE MM
   cells in the bone marrow by virtue of the homing effect and its
   ramification on the occurrence of osteolytic lesions (Fig. [85]5A).
   This model mimics the in vivo behavior of plasma cells, allowing us to
   investigate the influence of CXCL7 on myeloma cell activity in a more
   physiologically relevant context. At 7 weeks, three out of five mice in
   the CXCL7-OE group exhibited symptoms of long bone fractures. We
   performed regular venous blood collection to measure serum free light
   chain levels and assess tumor burden, which showed a significantly
   higher tumor burden in the CXCL7-OE group (Fig. [86]5B).

Fig. 5. Impact of CXCL7 on the metastatic potential of MM cells in vivo.

   [87]Fig. 5
   [88]Open in a new tab

   A Schematic diagram of subcutaneous xenograft model and tail vein
   injection xenograft model in mice (the image materials originated from
   Figdraw). B Quantitative levels of free λ light chain in the veins of
   CXCL7-OE and NC group; as well as the ratios of pathological fractures
   in mice of different groups. C The micro-CT scanning profiles of the
   leg bones in distinct groups of mice were presented. We exhibited the
   CT scan images of the femurs, along with the images of bone mineral
   density and trabecular bone of representative mice in the NC group and
   the OE group. The outcomes revealed that the bone mineral density (BMD)
   of the femurs in the CXCL7-OE group of mice was significantly lowered
   (P = 0.007), and the average trabecular thickness (Tb, Th) was
   significantly diminished (P = 0.037). D The presence of tumor
   colonization in the femur was confirmed through HE staining in both the
   CXCL7-OE group and the NC group. The tumor burden within the femur of
   mice in the NC group was more considerable. E HE staining was employed
   to identify the presence of lung, liver, kidney, and subcutaneous tumor
   colonization in mice from the CXCL7-OE group and NC group. There were
   no significant disparities in the lung, liver, and kidney lesions
   between the mice in the OE group and those in the NC group. However,
   subcutaneous soft tissue metastatic foci were identified in the mice of
   the OE group, but not in those of the NC group. F BCMA and CD138 were
   marked by immunohistochemistry for the evaluation of the occurrence of
   monoclonal plasma cells in the femoral bones of mice within the OE and
   NC cohorts. Evident plasma cell infiltrations were exhibited in the
   bone marrow of both cohorts).

   Micro-CT imaging of femoral bones revealed more pronounced bone
   deterioration in the CXCL7-OE group (Fig. [89]5C). Specifically, bone
   mineral density (BMD) was significantly reduced in the CXCL7-OE group
   (P = 0.007), with a mean decrease of 105 mg/cc compared to the NC
   group. Additionally, the average trabecular thickness (Tb, Th) was
   significantly decreased in the CXCL7-OE group (P = 0.037), indicating
   that CXCL7 overexpression accelerates bone loss and disrupts trabecular
   structure.

   Then, tissue samples from the femur, subcutaneous metastases, livers,
   lungs, and kidneys were collected for histopathological examination via
   hematoxylin and eosin (HE) staining (Fig. [90]5D–E). Subsequently, we
   performed immunohistochemical analysis to detect BCMA and CD138, which
   allowed us to assess the presence of monoclonal plasma cells in the
   bone marrow of mice. Distinct BCMA (+) and CD138 (+) plasma cell
   infiltrates were observed in the femurs of both the OE and NC groups
   (Fig. [91]5F). The femur sections revealed that myeloma cells in the
   CXCL7-OE group showed enhanced colonization and homing abilities within
   the bone marrow.

CXCL7 activates the JAK/STAT3 signaling pathway

   We conducted RNA-seq on CXCL7-OE and NC RPMI 8226 cell lines and
   identified differentially expressed genes (Fig. [92]6A). KEGG pathway
   analysis revealed significant enrichment of the JAK/STAT signaling
   pathway in the CXCL7-OE cells (Fig. [93]6B–C). Among the differentially
   expressed genes, JAK1 and STAT3 were conspicuously upregulated. Prior
   studies have validated that CXCR2 acts as the primary receptor of
   CXCL7, and the JAK/STAT3 is one of the pathways activated following the
   activation of CXCR2 [[94]15, [95]16]. We postulate that CXCL7 activates
   the JAK/STAT3 pathway via CXCR2, thereby influencing the proliferation
   of MM cells and further triggering the expression of downstream
   activating proteins associated with the osteoclast signaling pathway
   (Fig. [96]6D).

Fig. 6. CXCL7 binds to CXCR2 through autocrine of MM cells and activates the
intracellular JAK/STAT3 signaling pathway.

   [97]Fig. 6
   [98]Open in a new tab

   A The volcano plot of the differentially expressed genes revealed by
   RNA-seq profiling of the CXCL7-OE and NC RPMI8226 cell lines
   (|logFC | > 1, p < 0.05). B, C Enrichment analysis of KEGG pathways for
   differentially expressed genes in the CXCL7-OE and NC RPMI8226 cell
   lines. D The hypothesized mechanism schema of how CXCL7 affects
   intracellular signaling pathways in MM cells: CXCL7, autocrined by MM
   cells, binds to CXCR2, thereby activating the JAK/STAT3 pathway and
   further up-regulating MMP13 and C-myc. Western blot analysis was
   conducted to evaluate the protein expression levels of JAK1, p-JAK1,
   STAT3, p-STAT3, MMP13, C-myc and CXCR2 in H929, U266, and RPMI 8226
   cells within the CXCL7-OE group, CXCL7-KD group, and NC group. In the
   CXCL7-OE group, a significant upregulation of the JAK/STAT3 pathway,
   MMP13, C-myc and CXCR2 was observed (E); while in the CXCL7-KD group, a
   notable downregulation was witnessed (F). G Co-immunoprecipitation
   (Co-IP) assay was performed to confirm the combination between CXCL7
   and CXCR2 in H929 and RPMI8226 cells. H In the mouse xenograft model,
   immunohistochemistry results revealed a significant upregulation of
   JAK/STAT3 in the tumors of the CXCL7-OE group).

   To explore the JAK/STAT3 signaling pathway, we carried out WB to
   evaluate the expression of JAK1, phosphorylated JAK1 protein (p-JAK1),
   STAT3, phosphorylated STAT3 (p-STAT3) and CXCR2 in different MM cell
   lines. In both H929 and RPMI 8226 cell lines, the levels of p-JAK1 and
   p-STAT3 were significantly increased in CXCL7-OE group (Fig. [99]6E).
   Conversely, in U266 and RPMI 8226 cell lines, these levels were reduced
   in CXCL7-KD group (Fig. [100]6F). Additionally, the expression level of
   CXCR2 was significantly upregulated in CXCL7-OE H929 and RPMI8226 cells
   (Fig. [101]6E) and significantly downregulated in CXCL7-KD U266 and
   RPMI8226 cells (Fig. [102]6F). Co-immunoprecipitation (Co-IP) confirmed
   that CXCL7 interacts with CXCR2 in both H929 and RPMI8226 cells (Fig.
   [103]6G). These results suggest that increased CXCL7 expression
   enhances its binding to CXCR2 and activates the JAK/STAT3 pathway.

   Immunohistochemical analysis of the mouse xenograft tumor model further
   confirmed significant JAK/STAT3 pathway activation in CXCL7-OE group
   (Fig. [104]6H). Additionally, single-cell RNA sequencing of MM patient
   samples revealed a strong co-expression of CXCL7 and JAK/STAT3 in the
   mm03 subgroup (Fig. [105]2K). JAK/STAT3 forms a crucial oncogenic
   pathway in MM, traditionally believed to be activated by IL-6. However,
   in the absence of IL6, most MM cell lines still have a highly active
   JAK/STAT3 pathway. Our research found that a subset of MM cells
   secreting CXCL7 can activate the JAK/STAT3 pathway through autocrine
   effects. Also, we observed enhanced activation of JAK/STAT3 in other
   cell subgroups (Fig. [106]2K), providing evidence for its potential
   activation via multiple mechanisms.

CXCL7 upregulates the expression levels of MMP13 and C-myc via JAK/STAT3
pathway

   We next investigated the downstream genes activated by the JAK/STAT3
   pathway. Transcriptome sequencing of CXCL7-OE cells revealed
   significant upregulation of matrix metalloproteinase 13 (MMP13) and
   C-myc. Both MMP13 and C-myc are crucial effector proteins in response
   to JAK/STAT3 activation and are closely associated with osteoclast
   signaling and MM cell proliferation. Consequently, we proceeded with
   the validation of MMP13 and C-myc expression levels. The results showed
   a significant increase in both MMP13 and C-myc in CXCL7-OE cells (Fig.
   [107]6E), whereas their expression was significantly reduced in
   CXCL7-KD cells (Fig. [108]6F).

   Meanwhile, we utilized TRAP as a marker for osteoclasts and observed a
   significant increase in osteoclast numbers in the bone tissue of
   CXCL7-OE mice. Furthermore, we assessed the expression levels of MMP2,
   MMP9, and MMP13, which serve as markers for proteins associated with
   the activation of the osteoclast signaling pathway. Our findings
   revealed a significant elevation in the expression of all three MMPs in
   the bone tissue of the CXCL7-OE mice, with MMP13 showing the highest
   expression (Fig. [109]7A–B).

Fig. 7. CXCL7 activates the JAK/STAT3 pathway, resulting in the upregulation
of MMP13 and C-myc, further promoting osteoclast signaling pathway activation
and tumor proliferation.

   [110]Fig. 7
   [111]Open in a new tab

   A, B Immunohistochemical analyses were performed to assess the
   expression levels of TRAP, MMP2, MMP9, and MMP13 in the femurs of mice
   within the OE group and the NC group. TRAP was employed for labeling
   osteoclasts, and the results indicated a substantial increase in the
   quantity of osteoclasts in the OE group. MMP2, MMP9, and MMP13 were
   used to mark the relevant proteins related to the activation of
   osteoclast signaling pathways. The observations also revealed that the
   expression levels of MMP2, MMP9, and MMP13 in the bone tissue of mice
   in the OE group were significantly elevated. In the CXCL7-OE RPMI8226
   (C) and H929 (D) cell lines, the incorporation of CXCR2 inhibitors
   incurs a significant downregulation of the JAK/STAT3 pathway,
   concurrently accompanied by a significant reduction in the expression
   magnitudes of MMP13 and C-myc. In the CXCL7-OE RPMI8226 (E) and H929
   (F) cell lines, the administration of CXCR2 inhibitors elicited a
   significant attenuation of the cell proliferative capacity. In the
   CXCL7-OE RPMI8226 (G) and H929 (H) cell lines, the incorporation of
   JAK1 inhibitors gave rise to a significant reduction in the expression
   levels of p-STAT3, MMP13, and C-myc. In the CXCL7-OE RPMI8226 (I) and
   H929 (J) cell lines, the administration of JAK1 inhibitors elicited a
   significant attenuation of the cell proliferative capacity. K TRAP
   staining of CXCL7-OE/NC RPMI8226 cell lines co-cultured with primary
   mouse BMDMs. After 6 days, the CXCL7-OE group showed a higher number of
   multinucleated osteoclasts, indicating that CXCL7 promotes osteoclast
   differentiation. L Immunofluorescence was implemented for the precise
   detection of TRAP and MMP13, where TRAP acted as a definitive marker of
   osteoclasts. The obtained results manifested that in the CXCL7-OE
   group, TRAP and MMP13 presented conspicuous co-localization, inferring
   that MMP13 assumes a crucial role in the activation of osteoclasts.

   To further substantiate that the upregulation of JAK/STAT3 is
   attributable to CXCR2, we treated MM cells with SB225002, a CXCR2
   inhibitor. This treatment led to a significant reduction in JAK/STAT3
   activation (Fig. [112]7C–D) and cell proliferation (Fig. [113]7E–F). We
   further treated CXCL7-OE cells with Ruxolitinib (JAK inhibitor) (Fig.
   [114]7G–H). The results demonstrated that both p-STAT3 expression and
   cell proliferative capacity were significantly suppressed (Fig.
   [115]7G–J). Furthermore, treatment of MM cells with CXCR2 inhibitors
   and JAK inhibitors resulted in downregulation of MMP13 and C-myc
   expression (Fig. [116]7C–D & [117]G–[118]H).

   We further explored the relationship between CXCL7 and the expression
   of CXCR2 and MMP13 using data from the MMRF database. Our analysis
   revealed a significant positive correlation between CXCL7 and both
   CXCR2 (P < 0.001) and MMP13 (P < 0.001) (Fig. [119]S1).

Elevated expression of CXCL7 in MM cells significantly facilitates the
generation of osteoclasts

   To further validate that high CXCL7 expression in MM cells promotes
   osteoclastogenesis, we co-cultured RPMI8226 CXCL7-OE/NC cell lines with
   bone marrow-derived macrophages (BMDMs) isolated from C57BL/6 mice.
   M-CSF and RANKL were added to the BMDMs to induce osteoclast
   differentiation. After 6 days of co-culture, TRAP staining was
   performed to assess osteoclast formation. The results showed that BMDMs
   co-cultured with CXCL7-OE cells generated a significantly higher number
   of multinucleated osteoclasts compared to the NC group, indicating that
   elevated CXCL7 expression enhances osteoclastogenesis (Fig. [120]7K).

   Subsequently, immunofluorescence was employed to examine both MMP13 and
   TRAP (a marker for osteoclasts) within femoral bone marrow from mouse
   xenograft models (Fig. [121]7L). The results revealed a marked
   co-localization of MMP13 around TRAP-positive multinucleated
   osteoclasts, suggesting a strong interaction between MMP13 and
   osteoclasts. In contrast, no such co-expression was observed in the NC
   group, further supporting the role of MMP13 secreted by myeloma cells
   in facilitating osteoclast activity.

Discussion

   MM is a heterogeneous tumor characterized by diverse subpopulations of
   myeloma cells with distinct biological behaviors [[122]5, [123]17,
   [124]18]. Osteolytic damage and pathological fractures are hallmark
   clinical symptoms of myeloma, leading to restricted mobility and loss
   of self-care ability in MM patients [[125]4]. Our research identifies
   CXCL7 as a key gene contributing to pathological fractures in MM.
   Subpopulations of MM cells with elevated CXCL7 expression show enhanced
   activation of the osteoclast signaling pathway. Notably, CXCL7 is a
   member of the chemokine family, which plays a critical role in
   regulating immune cell function [[126]19, [127]20]. Given that MM
   originates from lymphocytes, chemokines like CXCL7 are likely key
   regulators of myeloma cell behavior.

   Current research on the role of CXCL7 in tumors is limited. Previous
   studies have shown that in glioma, tumor-induced macrophages secrete
   CXCL7 to enhance glioma stem cell functionality [[128]21]. In breast
   cancer, targeting CXCL7 inhibits tumor growth and metastasis [[129]10],
   while in cholangiocarcinoma, CXCL7 promotes tumor proliferation and
   invasion via the AKT signaling pathway [[130]22]. In MM, CXCL7 has been
   identified as the predominant activator of MMP13. CXCL7 can be secreted
   by MSCs, thereby stimulating MM cells to produce a significant amount
   of MMP13, which constitutes a crucial means for activating the
   osteoclast signaling pathway [[131]23, [132]24]. Our research has
   disclosed that in MM, not only stromal cells but also a subset of
   myeloma cells secrete CXCL7 in an autocrine manner. CXCL7 activates the
   JAK/STAT3 pathway within MM cells via the CXCR2 receptor and
   upregulates the expression of MMP13.

   Through multiple analyses, we confirmed that MM cells with elevated
   CXCL7 expression secrete high levels of MMP13. A strong correlation was
   observed between the gene expression of CXCL7, CXCR2, and MMP13 in the
   MMRF database. However, no significant correlation was found between
   CXCL7 and MYC gene expression. We hypothesize that this lack of
   correlation arises because MMP13 in myeloma cells is primarily induced
   by CXCL7, whereas MYC, involved in various cancer pathways, is
   regulated by multiple mechanisms. As only a subset of myeloma cells
   secretes CXCL7, the potential association between CXCL7 and MYC
   expression may be diluted in the broader plasma cell RNA-seq data.

   The JAK/STAT3 pathway is widely recognized as a crucial oncogenic
   pathway in MM. It was formerly hypothesized that it was primarily
   activated via IL6 [[133]25]. Nevertheless, the activation of the
   JAK/STAT3 pathway is ubiquitous in MM [[134]26]; numerous MM cells
   still exhibit a significantly activated JAK/STAT3 pathway in the
   absence of IL6, suggesting the involvement of alternative activation
   mechanisms. Our Single-cell RNA sequencing analysis revealed widespread
   activation of the JAK/STAT3 pathway across various MM subgroups, with a
   notable co-localization of JAK/STAT3 and CXCL7 in one subgroup. This
   concurrently implies that there are additional pathways for the
   activation of STAT3.

   STAT3 is a key transcription factor involved in the initiation and
   progression of MM [[135]27]. In our study, we observed that the
   upregulation of STAT3 significantly increased the protein levels of
   MMP13 and C-myc. MMPs play a critical role in osteolytic damage in MM
   and promote tumor growth by enhancing angiogenesis and facilitating the
   migration and invasion of MM cells [[136]28–[137]30]. Interestingly,
   previous studies have shown that IL6 can lead to the upregulation of
   MMP13, but CXCL7 is the most significant factor in upregulating MMP13
   in MM [[138]23, [139]24]. Our study explains this phenomenon, that is,
   CXCL7 leads to the upregulation of MMP13 through the activation of the
   JAK/STAT3 pathway.

   Our research examined the role of CXCL7 in promoting osteolytic lesions
   in myeloma using various methods. In animal studies, we measured femur
   BMD using Micro-CT. BMD was calculated as Bone Mineral Content (BMC)
   divided by Bone Volume (BV). As BV was computed using a computerized
   method, fractured portions of the femur, which were disconnected from
   the rest of the bone, were excluded from the analysis. Given that the
   CXCL7-OE group exhibited more pathological fractures, this exclusion
   likely minimized differences in BDM between OE and NC groups. In the NC
   group, implantation of myeloma cells into the femurs also resulted in a
   decrease in BMD compared to normal mice. However, the OE group
   displayed a significantly greater reduction in BMD, with a decrease of
   over 100 mg/cc compared to the NC group, reflecting substantial bone
   loss.

   Our findings highlight that a subset of myeloma cells secretes CXCL7
   via an autocrine mechanism, playing a pivotal role in tumor progression
   and the development of osteolytic lesions. Targeting CXCL7 emerges as a
   promising therapeutic strategy, as inhibiting this autocrine loop could
   help prevent bone deterioration. Despite CXCL7 being expressed in only
   a fraction of myeloma cells, the high intra-tumor heterogeneity of
   myeloma suggests that targeting these particularly aggressive
   subpopulations could provide substantial clinical benefit. Blocking
   CXCL7 may reduce MMP13 production, decrease osteoclast activation, and
   mitigate bone resorption, ultimately lowering fracture risk and
   improving patient outcomes.

   While research on CXCL7 inhibitors for MM is still in its early stages
   and no targeted drugs have been developed, the potential to treat
   patients with severe bone lesions is considerable. The development of
   CXCL7-targeted therapies could open new avenues for managing myeloma
   and its complications. A deeper understanding of CXCL7’s mechanisms,
   coupled with the development of effective inhibitors, will be essential
   for advancing therapeutic options and improving patient care.

Methods and Materials

Weighted correlation network analysis (WGCNA)

   First, we utilized the gene expression profiles of patients in the GEO
   database. We chose the [140]GSE24080 cohort and designated the TT2
   patient group as the training set, while assigning the TT3 patient
   group as the validation set. In the TT2 training set, we separately
   calculated the Median Absolute Deviation (MAD) for each gene. To
   exclude the genes with the smallest MAD, we excluded the top 50% of
   genes. The R package “goodSamplesGenes” was utilized for the removal of
   outlier genes and samples, and WGCNA was further used to construct a
   scale-free co-expression network. Initially, we performed both
   Pearson’s correlation matrices and the average linkage method for all
   pairwise genes. After selecting the power of 5, the adjacency was
   transformed into a topological overlap matrix (TOM) to measure the
   network connectivity of a gene by defining it as the sum of its
   adjacency with all other genes for network gene ration, and the
   corresponding dissimilarity (1-TOM) was calculated. To classify genes
   with similar expression profiles into gene modules, we conducted
   average linkage hierarchical clustering according to the TOM-based
   dissimilarity measure with a minimum gene group size of 30 for the
   genes dendrogram. Additionally, we merged modules with a distance of
   less than 0.25.

Single sample Gene Set Enrichment Analysis (GSEA)

   Single sample GSEA (ssGSEA) analysis was performed to determine the
   degree of activation for each pathway, which was recorded as the
   Enrichment Score (ES). The score ranged from 0 to 1, indicating minimal
   to maximal activation. The activation scores for the “B cell tumor
   proliferation” pathway and the “osteoclast signaling activation”
   pathway were calculated using ssGSEA. The gene sets for these pathways
   were obtained from the MsiGDB database.

The expression level of CXCL7 and clinicopathological features of samples

   A total of 108 clinical samples of newly diagnosed MM patients were
   included in this study, which was collected from patients who visited
   Zhongshan Hospital of Fudan University from 2016 to 2020. In order to
   further observe the relationship between CXCL7 and the prognosis and
   drug treatment response of patients, only patients who had not
   undergone autologous hematopoietic stem cell transplantation were
   included. All patients received an initial treatment of a three-drug
   combination of protease inhibitor, lenalidomide, and dexamethasone. The
   bone marrow samples of patients were collected at the time of
   diagnosis. All bone marrow fluid samples were enriched with nucleated
   cells and sorted with CD138 magnetic beads. The expression level of
   CXCL7 is detected by the qPCR.

Single cell mRNA sequencing

   A diverse array of bone marrow samples was collected from Zhongshan
   Hospital affiliated with Fudan University, following approval from the
   Institutional Review Board of Zhongshan Hospital. Bone marrow
   mononuclear cells (BMMCs) were isolated and employed for single-cell
   RNA sequencing (scRNAseq). Single-cell libraries were generated from
   unsorted BMMCs using a 10× Genomics Chromium Single Cell 5′ Library Kit
   and Chromium instrument, followed by sequencing on an Illumina
   sequencer. The obtained FASTQ data was aligned to the reference genome
   (GRCh38) using Cell Ranger (v5.0.1), subsequently undergoing read
   filtering, barcode counting, and unique molecular identifier (UMI)
   counting. Seurat v4.0.1 software was utilized for cell clustering and
   dimension reduction. Single-cell data from different patients were
   integrated and analyzed via the UMAP downscaling approach. For pathway
   activation analysis in different cell subsets, we employed the irGSEA
   package along with AUCell and UCell methods to score pathway activation
   levels based on gene sets obtained from the MSigDB database.

Cell culture

   U266 and NCI-H929 cell lines are purchased from the Institute of Basic
   Medical Sciences, Chinese Academy of Medical Sciences, and RPMI-8226
   cell line is obtained from the cell bank of the Chinese Academy of
   Sciences for use in this investigation. Cells are grown in RPMI 1640
   media (Gibco, USA) with 10% fetal bovine serum (FBS) (Gibco, USA) under
   37 °C, saturated humidity, and 5% CO2 conditions.

Establishment of lentiviral stabilized cell lines

   The CXCL7 overexpression and control lentiviruses were obtained from
   Genechem Biotechnology Co., Ltd. (Shanghai, China). The vector
   components are arranged in the following sequence:
   Ubi-MCS-3FLAG-CBh-gcGFP-IRES-puromycin. H929 and RPMI8226 cells were
   subjected to infection at a multiplicity of infection (MOI) of 100 for
   12 h. Seventy-two hours post-transduction, the culture medium was
   amended with 2 μg/ml puromycin dihydrochloride (GeneChem, China) to
   facilitate the selection of stably transfected cells. The successful
   overexpression of CXCL7 was validated using Western blot analysis and
   quantitative PCR.

Small interfering RNA transfection

   Small interfering RNA (siRNA) targeting CXCL7 (Genomeditech, China),
   with the sequence provided in Supplementary Table [141]1, was used for
   transfection. U266 and RPMI8226 cells in the logarithmic growth phase
   were seeded into a six-well cell culture plate. The transfection
   complex was prepared by mixing 125 μL of RPMI 1640 medium (Gibco, USA),
   3.75 μL of Lipo3000 reagent (Thermo Fisher Scientific, USA), and 10 μL
   of siRNA. After allowing the transfection complex to stand for 20 min,
   it was added to the six-well plate and gently mixed with the cell
   suspension. The plate was then incubated at the appropriate conditions
   for 24-48 h before changing to a fresh medium.

Enzyme-linked immunosorbent assay (ELISA assay)

   The concentration of human CXCL7 cytokine in the culture medium was
   measured using an ELISA kit (Abcam, England). All procedures were
   performed according to the manufacturer’s instructions.

Quantitative RT-PCR analysis

   Quantitative RT-PCR analyses were performed using the EZ Bioscience RNA
   Extraction Kit to extract RNA from control or treated cells, followed
   by cDNA synthesis with the Takara cDNA Synthesis Kit and real-time PCR
   using the Takara TB Green Premix Ex Taq (Tli RNaseH Plus) quantitative
   PCR kit. All primer sequences used in this study are shown in
   Supplementary Table [142]2.

Western blot analysis

   To lyse the cells, use 1X RIPA buffer containing Tris buffer (50 mM, pH
   7.4), NaCl (150 mM), Triton X-100 (1%), PMSF (1 mM), SDS (0.1%), and
   protease inhibitor mixture from Beyotime Biotechnology. Cell lysates
   were mixed with Loading Buffer, separated by electrophoresis in a
   10-15% SDS-PAGE gel, and transferred to a PVDF membrane from Millipore.
   Membranes were blocked with QuickBlock Blocking Buffer for Western Blot
   from Biotechnology. Membranes were washed three times with PBST and
   incubated with enzyme-linked secondary antibodies for 1 h at room
   temperature. Following PBST cleaning, luminescence was recorded with
   Lumin4000 from GE, USA, and ECL reagent from BioSharp. All antibodies
   used are listed in Supplementary Table [143]3.

Subcutaneous Transplantation Tumor

   CXCL7-OE and NC RPMI8226 cells were harvested during logarithmic
   growth, washed with PBS, and resuspended in the appropriate volume of
   PBS to achieve a density of f 5×10^6/100 μL. The cells were then
   subcutaneously injected into B-NDG mice within 2.5 h. Five randomly
   assigned 5-week-old female mice were divided into NC group and CXCL7-OE
   group. After alcohol disinfection, the mice were inoculated with 100 μL
   of cell suspension on the outside of the right upper limb and returned
   to their cages for further observation. The mice were monitored
   regularly for tumor growth, and tumor volumes were measured twice
   weekly using calipers. Mice were euthanized at the experimental
   endpoint, and tumors were excised for further analysis, including
   histopathological and molecular investigations.

Xenograft tumor model in mice through intravenous injection

   CXCL7-OE and NC RPMI8226 cells, harvested during logarithmic growth,
   were washed with PBS and resuspended in PBS at a concentration of 10^7
   cells per 100 µL. The cell suspension was injected into the tail vein
   of 5-week-old female B-NDG mice within 2.5 h of preparation. Mice were
   randomly assigned into two groups: the NC group and the CXCL7-OE group,
   with five mice per group. Prior to injection, the tail veins were
   sterilized with cotton balls soaked in a 75% alcohol solution to
   promote vein dilation. A total volume of 100 µL of the cell suspension
   was injected in the direction of blood flow. Tumor growth and
   development were monitored regularly throughout the experiment. At the
   experimental endpoint, mice were euthanized, and tissues, including
   lungs, liver, and bone, were harvested for histological and molecular
   analyses.

Immunohistochemistry (IHC) analysis

   Mouse tumors were sampled and then embedded in paraffin and sectioned
   by a microtome. Immunohistochemistry was performed according to the
   steps provided by the manufacturer of the immunohistochemistry kit
   (Sevier, China), in which the primary antibodies used are also listed
   in Supplementary Table [144]3.

Co-Immunoprecipitation (CO-IP)

   To isolate the protein supernatant, cell lysates were subjected to
   centrifugation for 10–15 min. The resulting supernatant was incubated
   with the specific antibody overnight at 4 °C. Following this, the
   immune complexes were captured by magnetic beads through a 1-hour
   incubation. To eliminate nonspecific binding, the beads underwent three
   washes with IP buffer and one wash with distilled water. The
   immunoprecipitated proteins were then analyzed via WB.

Co-culture of MM cells and mouse bone marrow-derived macrophages (BMDMs)

   The co-culture system of MM cells and mouse BMDMs was established using
   a Transwell co-culture system. Logarithmically growing CXCL7-OE/NC
   RPMI8226 cells were seeded into the insert wells of the Transwell at a
   density of 1×10^5 cells per well, while BMDMs were plated in the lower
   chamber at a density of 5 × 10^5 cells per well. To extract BMDMs,
   8-week-old C57BL/6 mice were euthanized by cervical dislocation
   following deep anesthesia. Both femoral bones were exposed and
   carefully cut open, and the bone marrow was collected using sterile
   scissors and forceps. The extracted bone marrow was placed into a dish
   containing sterile PBS, and the marrow was gently squeezed out using a
   syringe piston. The resulting cell suspension was filtered through a 70
   μm cell sieve to remove debris and bone fragments. The filtered
   suspension was transferred into a culture flask with a complete medium,
   where it was allowed to settle for 4 h to remove non-adherent cells.
   The supernatant was then transferred into a new flask containing medium
   supplemented with M-CSF (2 ng/μL) and RANKL (20 ng/μL). After 6 days,
   osteoclast differentiation and morphology in the lower chamber were
   assessed using TRAP staining.

Statistical analysis

   Continuous variables were analyzed using Student’s t-test for normally
   distributed data or the Mann-Whitney U test for non-normally
   distributed data. PFS and OS probabilities were estimated using the
   Kaplan-Meier method, and differences in statistical significance were
   evaluated using the two-sided log-rank test. Categorical data were
   assessed using the Chi-square test. Statistical analyses were performed
   using R, version 4.3.3 ([145]http://cran.r-project.org). A p-value less
   than 0.05 was considered statistically significant unless otherwise
   specified. All confidence intervals (CIs) were calculated at a 95%
   level. *P < 0.05, **P < 0.01 and ***P < 0.001 were noted.

Reporting summary

   Further information on research design is available in the [146]Nature
   Research Reporting Summary linked to this article.

Supplementary information

   [147]Supplementary Table 1^ (16KB, docx)
   [148]Supplementary Table 2^ (16.2KB, docx)
   [149]Supplementary Table 3^ (17.6KB, docx)
   [150]Supplementary Figure S1^ (1.2MB, tif)
   [151]Original Data File^ (2MB, pdf)
   [152]Reporting Summary^ (1.8MB, pdf)

Author contributions

   The conceptualization of this study was accomplished by YW, TL, and PL.
   YW and TL were responsible for developing the methodology. CZ played
   the role of the assistant. Data collection was jointly carried out by
   YW, TL, and CZ. The determination of the pathological sections and
   immunohistochemical results was accomplished by QZ. The original draft
   of the writing was produced by YW and TL. Finally, the review and
   editing of the writing were meticulously handled by PL.

Funding

   Science and Technology Innovation Plan of Shanghai Science and
   Technology Commission (21YF1406300). Natural Science Foundation of
   Shanghai Municipality (22ZR1411400). Youth Foundation of Zhongshan
   hospital Fudan University (2021ZSQNN61).

Data availability

   The single-cell RNA-seq processed gene expression data reported in this
   paper have been deposited into the CNGB Sequence Archive (CNSA)
   [[153]31] of the China National GeneBank DataBase (CNGBdb) [[154]32]
   with accession number CNP0005613(CSE0000445), and raw sequencing data
   have been deposited in the OMIX, China National Center for
   Bioinformation, Chinese Academy of Sciences (accession no. OMIX007530)
   [[155]33]. To request access to raw sequencing data, please apply at
   the Human Genetic Resources Service System of the Ministry of Science
   and Technology ([156]https://apply.hgrg.net/) in accordance with the
   Regulations on the Management of Human Genetic Resources of China. The
   multiple myeloma RNA-seq data [157]GSE24080 were downloaded from the
   GEO database.

Competing interests

   The authors declare no competing interests.

Ethics approval and consent to participate

   The study was approved by the Ethics Committee of Fudan University
   Zhong Hospital (Observation Research Cohort of Plasma Cell Diseases in
   Zhongshan Hospital, Registration Number: B2017-031R)) and complied with
   the principles of the Helsinki Accord. Bone marrow samples were
   collected from all subjects, with patient consent, and stored in the
   tissue bank at Fudan University Zhongshan Hospital. The animal
   experiments were approved by the Animal Ethics Committee of Zhongshan
   Hospital Fudan University.

Footnotes

   Edited by Anastasis Stephanou

   Publisher’s note Springer Nature remains neutral with regard to
   jurisdictional claims in published maps and institutional affiliations.

   These authors contributed equally: Yue Wang, Tianwei Lan.

Supplementary information

   The online version contains supplementary material available at
   10.1038/s41419-025-07413-6.

References