Bacterial infections exacerbate myeloma bone disease

We found an intractable problem in the treatment of myeloma: osteolytic bone lesions do not heal, even the patient is in complete remission (Fig. 1A, B). Immunohistochemistry examination of bone marrow biopsy samples obtained from myeloma patients (active and remission stage) showed presence of gram-negative (LPS⁺) and gram-positive (LTA⁺, Lipoteichoic acid) bacteria compared to healthy bone marrow (Fig. 1C). To determine the abundance of bacteria in bone marrow under physiological conditions, the bacterial ribosomal 16S rRNA gene was amplified by a real-time quantitative PCR (qPCR) assay with universal primers. We detected much more bacteria in the bone marrow of myeloma patient in active and remission, but less in healthy donors (Fig. 1D). We also found a robust correlation between bone marrow bacteria abundance and bone lesion numbers (Fig. 1E). Therefore, bone marrow bacteria is likely to be a key regulator of bone lesion in myeloma patients.

Bone remodeling is regulated by a balance between osteoclast-mediated resorption and osteoblast-mediated matrix synthesis. To examine whether bone marrow bacteria can regulate this balance, we first assessed their effects on osteoblast and osteoclast differentiation.

Previous study identified that E. coli and Streptococcus pneumoniae (S. pneumoniae) are the most frequent pathogen in myeloma patients bone marrow [15]. We observed more E. coli-but not S. pneumoniae in bone marrow aspirates of myeloma patient in active and remission (Fig. 2A). In osteoblast differentiation assay, we cocultured MSCs (osteoblast progenitors) in osteoblast medium with E. coli. MSCs cultured alone in this medium served as a positive control. We observed significantly lower Alizarin red S staining (Fig. 2B–D), alkaline phosphatase activity (Fig. 2E) and osteoblast marker genes expression [bone gamma-carboxyglutamic acid-containing protein (BGLAP), alkaline phosphatase (ALP), and collagen type I α1 (COL1A1)] (Fig. 2F) in MSCs cultured with E. coli. In osteoclast differentiation assay, we cocultured precursors of osteoclasts (preOCs) with E. coli, which stimulated RANKL-induced TRAP⁺ cell formation (Fig. 3A, B), TRAP 5b secretion in the supernatant (Fig. 3C) and osteoclast marker genes expression [tartrate-resistant acid phosphatase (TRAP), cathepsin K (CTSK) and calcitonin receptor (CALCR)] (Fig. 3D). Our aggregated results demonstrated that E. coli enhances osteoclastogenesis and suppresses osteoblastogenesis.

We next examined the signaling pathways by which E. coli may regulate bone destructions. Previous work has established that bacteria implicated in bone diseases contain or produce molecules with potent effects on bone cells. Bacterial factors can interact with osteoblast or the osteoclast to promote bone resorption and inhibit bone formation [16]. LPS is an important pathogenic factor from gram-negative bacteria cell wall, and LPS-TLR4 (Toll like receptor 4) complex has been identified as one of regulators involved in bone remodeling [17, 18], but the downstream signaling are not very clear. Consistent with previous study, osteoblastogenesis assay indicated that LPS inhibits osteoblast differentiation (Fig. 4A). Runt related transcription factor 2 (RUNX2) is a key regulator of osteoblast differentiation. qPCR and western blot analysis showed that MSCs RUNX2 mRNA and protein levels are lower treated with LPS as compared to untreated cells (Fig. 4B, C). Bone morphogenetic proteins (BMPs) 2 are members of the transforming growth factor β superfamily (TGF-β) that were originally demonstrated to promote bone formation. BMPs activate downstream smad1/5/8 signaling, and the three smads form heteromeric complexes with smad4, which translocate into the nucleus and activate the transcription of target genes, including RUNX2 [19]. To further explore the potential LPS downstream signaling pathways, we treated MSCs with different concentrations of LPS and found that LPS enhanced phosphorylation of NF-κB (nuclear factor κB) p65 signaling (Fig. 4C). Chromatin Immunoprecipitation (ChIP) assay using an anti-p-smad1/5/9 antibody showed LPS-activated NF-κB reduced p-smad1/5/9 binding ability with RUNX2 promoter (Fig. 4D). To show the specificity of NF-κB signaling pathways on RUNX2 inhibition, we treated E. coli and MSCs coculture system with BAY11-7085, a NF-κB signaling inhibitor, resulting in up-regulation of RUNX2 expression (Fig. 4E).

In osteoclastogenesis assay, LPS promotes osteoclast differentiation (Fig. 4F). qPCR and western blot analysis showed that LPS increased preOCs NFATc1 mRNA and protein levels as compared to untreated cells (Fig. 4G, H). Nuclear factor of activated T-cells, cytoplasmic 1 protein (NFATc1) is essential for osteoclast differentiation [2]. Western blot analysis showed that LPS enhanced phosphorylation of NF-κB p65 (Fig. 4H). ChIP assay using an anti-p-p65 antibody showed LPS enhanced phosphorylated NF-κB p65 binding ability with NFATc1 promoter (Fig. 4I) and promotes its transcription. Furthermore, NF-κB signaling inhibitor BAY11-7085 reduced LPS-induced NFATc1 expression (Fig. 4J). Together, these findings indicated that E. coli LPS inhibits osteoblastogenesis and enhances osteoclastogenesis through NF-κB p65 signaling.

According to our previous study [20], we constructed a myeloma mouse model in complete remission with or without E. coli infection. We intravenous injected myeloma cells 5TGM1 with or without E. coli, treated the mice with chemotherapy; mice not injected with myeloma cells but injected with or without E. coli served as controls (Fig. 5A). Two weeks after chemotherapy drugs treatment, M-protein, an indicator of myeloma burden, fell to a very low level, which is consist with our previous results. This suggests that the chemotherapy drugs-treated mice were almost or totally myeloma-free. After 8 weeks, in complete remission myeloma mice with E. coli infection, bone histomorphometric analysis indicated a lower percentages of bone volume/total volume (BV/TV) (Fig. 5B), percentages of osteoid surface (OS/BS) (Fig. 5C) and bone surface covered with osteoblasts (Ob.S/BS) (Fig. 5D); we also found higher percentages of bone surface eroded by osteoclasts (ES/BS) (Fig. 5E) and bone surface covered with osteoclasts (Oc.S/BS) (Fig. 5F). In agreement with these findings, the bone formation rate was decreased in complete remission myeloma mice with E. coli infection (Fig. 5G, H). Thus, E. coli contributes to myeloma patients bone disease in complete remission.

Toward a therapeutic, we asked whether elimination of E. coli in bone marrow helps healing of resorbed bone of myeloma in complete remission. We intravenous injected myeloma cells 5TGM1 with E. coli, 4 weeks later, treated the mice with chemotherapy drugs and with or without ampicillin (Fig. 6A). We observed 16S rRNA gene expression of E. coli in bone marrow aspirates of mice without ampicillin treated group (Fig. 6B), which suggested that the ampicillin -treated mice were almost or totally E. coli free. Bone histomorphometric analysis demonstrated a higher BV/TV (Fig. 6C), OS/BS (Fig. 6D) and Ob.S/BS (Fig. 6E); lower ES/BS (Fig. 6F) and Oc.S/BS (Fig. 6G) in ampicillin-treated mice. These findings indicated that counteracting E. coli may be effective for prevention or treatment of osteolytic bone lesions persistence of myeloma patients in complete remission.

All chemicals were purchased from Sigma-Aldrich, except where specified. BoneTRAP (TRACP 5b human) ELISA kit was purchased from immunodiagnostic systems (IDS) (Cat# SB-TR201R).

MSCs were isolated from the bone marrow of healthy donors and cultured in osteoblast differentiation medium to obtain mature osteoblasts [20]. The bone formation activity of osteoblasts was demonstrated using Alizarin red S staining (Sigma-Aldrich, Cat# A5533). Human monocytes were isolated from peripheral blood mononuclear cells of health donors and cultured in M-CSF (25 ng/ml) (R&D Systems, Cat# 216-MC) for 7 days to obtain the precursors of osteoclasts. The precursors derived from human monocytes were cultured in M-CSF (25 ng/ml) and RANKL (10 ng/ml) (R&D Systems, Cat# 6449-TEC), cocultured with or without E. coli or LPS for 7 days to induce mature osteoclast formation. Leukocyte acid phosphatase kit (Sigma-Aldrich, Cat# 387A) for TRAP staining was used to detect mature osteoclasts.

Cells were collected and lysed with lysis buffer (Cell Signaling Technology, Cat# 9803S). Cell lysates were subjected to SDS-PAGE assay, transferred to a Polyvinylidene fluoride (PVDF) membrane (Millipore, Cat# IPVH00010), and immunoblotted with antibodies against GAPDH (Cat# 2118), RUNX2 (Cat# 12556), p65 (Cat# 8242), phosphorylated p65 (Cat# 3033), phosphorylated smad1/5/9 (Cat# 13820) and NFATc1 (Cat# 8032) (Cell Signaling Technology).

Total RNA was isolated using a Total-RNA isolation kit (Thermo Fisher Scientific, Cat# 12183025). An aliquot of 0.5 μg of total RNA was subjected to reverse transcription (RT) with a SuperScript II RT-PCR kit (Invitrogen, Cat# 18064014). Quantitative PCR was performed using SYBR Green Master Mix (Life Technologies, Cat# A25780) with the 7500 Real-Time PCR System (Life Technologies). The primers used are listed in Additional file 1: Table S1.

Cells were fixed in 4% formaldehyde and sonicated to prepare chromatin fragments using SimpleChIP Enzymatic Chromatin IP Kit (Cell Signaling Technology, Cat# 9002). Chromatin samples were immunoprecipitated with antibodies against phosphorylated p65 and phosphorylated smad1/5/9 and control IgG (Cell Signaling Technology, Cat# 2729S) at 4 °C for 4 h. Immunoprecipitates and total chromatin inputs were reverse cross-linked. DNA was isolated and analyzed using quantitative real-time PCR. The primer sequences used are listed in Additional file 1: Table S2. Relative fold enrichment was calculated by determining the immunoprecipitation efficiency (ratio of the amount of immunoprecipitated DNA to that of the input sample).

Genomic DNA (gDNA) was extracted from fresh bone marrow aspirates with the QIAamp DNA Mini Kit (QIAGEN, Cat# 51304). gDNA from each specimen was subjected to qPCR to determine the amounts of bacteria by detecting the 16S genes. qPCR was performed using SYBR Green Master Mix (Life Technologies, Cat# A25780) with the 7500 Real-Time PCR System (Life Technologies). The RNase P copy number assay was utilized as a basis to compare bacterial DNA content to host (human) DNA content by performing a logarithmic ratio of the cycle thresholds obtained for bacterial and host DNA. The primers used in RT-PCR and qPCR are in Additional file 1: Table S3.

Formalin-fixed, paraffin-embedded sections of bone marrow biopsy samples obtained from patients with myeloma were deparaffinized and stained. HE staining was performed according to standard protocols [24]. Slides were stained with anti-LPS (Cat# LS-C75640) and LTA (Cat# LS-C202488) (LifeSpan BioSciences) antibody using an EnVision System (DAKO, Cat# K5361) following the protocols and nuclei were counterstained with hematoxylin.

C57BL/KaLwRij mice purchased from Charles River Labs, Beijing, China, were maintained in Xiamen University Animal Care-accredited facilities. 5TGM1 cells (5 × 10⁵ cells per mouse) were intravenous injected into 8-week-old C57BL/KaLwRij mice with or without E. coli (1 × 10⁴ cells per mouse). After 4 weeks, 1 mg/kg bortezomib (Cayman Chemicals, Cat# 10008822) and 2 mg/kg melphalan (Cayman Chemicals, Cat# 16665) were injected intraperitoneally into the mice three times a week for 2 weeks. Serum samples were collected from the mice weekly and tested for myeloma-secreted M proteins using ELISA analysis to monitor the tumor burden (Thermo Fisher Scientific, Cat # 88-50430-88). To examine the lytic bone lesions, radiographs were scanned with a Bruker In-Vivo Xtreme imaging system. Bone tissues were fixed in 4% paraformaldehyde and decalcified, and sections of them were stained with toluidine blue or TRAP following standard protocols. Both analyses were done using the BIOQUANT OSTEO software program (BIOQUANT Image Analysis Corporation).

Statistical significance was analyzed using the Graphpad 9.0 program with Student t-tests for comparison of two groups, and one-way ANOVA with Tukey’s test for comparison of more than two groups. P values less than 0.05 were considered statistically significant. All results were reproduced in at least three independent experiments.