The 5TMM3VT murine myeloma cells were donated by Professor Wen Zhou from the State Key Laboratory of Experimental Hematology, Department of Hematology, Xiangya Hospital, Central South University. Velcade (bortezomib for injection) was purchased from Hansoh pharma (H20173307, Jiangsu, China). Acetonitrile and methanol were purchased from Merck, Millipore Ltd (1.00030.4008, 1.06007.4008, Carrigtwohill, Ireland). Formic acid, ethyl acetate, trifluoroacetic acid, ammonium acetate, and L-arginine hydrochloride were purchased from Macklin Biochemical Co., Ltd (F809712, E809174, T818778, A801000, L800291, Shanghai, China). The 2-chloro-L-phenylalanine and dansyl chloride were purchased from Yuanye Bio-Technology Co., Ltd (B25643, S19248, Shanghai, China). The 1,4-Butane-1,1,2,2,3,3,4,4-d8-diamine was purchased from Toronto Research Chemicals (D416027, Canada). Chloroform, sodium carbonate, sodium bicarbonate, and acetone were purchased from Lingfeng Chemical Reagent Co. (Shanghai, China). Ornithine analytical standard was purchased from Solarbio (SO8470, Beijing, China). RPMI-1640 medium without arginine was purchased from Sigma-Aldrich (R1780, USA). Certified fetal bovine serum, RPMI 1640 medium, and dialyzed fetal bovine serum were purchased from Biological Industries (05-065-1A, 04-002-1A, 04-011-1b, Kibbutz Beit Haemek, Israel). Cell counting kit (CCK-8) was purchased from Yeasen Biotechnology Co., Ltd. (40203ES76, Shanghai, China).

All animal procedures were conducted in accordance with government-published recommendations for the Care and Use of Laboratory Animals and approved by the Institutional Ethics Review Boards of Nanjing University of Chinese Medicine. The experimental mice (C57BL/KaLwRij, 6–8 weeks, 18–20 g) were purchased from the institute of model animals of Nanjing University. The experimental mice were housed in the SPF laboratory animal center of Nanjing University of Chinese Medicine with 15–25°C ambient temperature and free access to food and water. After 1 week of adaptive feeding, animal experiments were started. 5TMM3VT murine myeloma cells (1×10⁶) were injected via the tail vein into 6-week-old C57BL/KaLwRij mice. The mice were divided into four groups as follows: model control (C, only modeling, n=10) group, bortezomib (V, n=10) treatment group, acupuncture (A, n=9) treatment group, and VA (n=8) treatment group. One day later, mice in three treatment groups were treated by different methods, containing intraperitoneal injection of 1.2 mg/ml V twice a week or/and electroacupuncture (Model SDZ-II, Suzhou Medical Appliance Factory, Suzhou, China) stimulation of Hegu (29) and Zusanli (30, 31) points (2/100 Hz, 2 mA) three times a week until all the mice were dead.

Blood was taken from tail vein on Tuesday and Wednesday during the experimental period. Blood samples of week 4–6 were mixed in the clean Eppendorf tubes, stored on ice for 2 h, and centrifuged (5,000 rpm, 10 min) at 4°C. The blood was collected for separation of serum. Subsequently, the supernatants were transferred to clean Eppendorf tubes and stored at −80°C.

All serum samples were thawed on the ice. An aliquot of 45 μl serum sample was precipitated by adding 135 μl acetonitrile containing internal standards (IS) [2-Chloro-L-Phenylalanine (plasma sample-acetonitrile: IS=2,000:1)], vortex for 30 s, sonicating for 10 min at 4°C, and then stayed for 3 h on ice. Precipitated protein was removed by centrifugation (13,000 rpm, 10 min) at 4°C. Subsequently, 153 μl supernatant was transferred to glass inserts of LC-MS vials and stored at −80°C for LC-MS analysis.

The LC-300AD LC system (Shimadzu, Japan) coupled to a Triple TOF™5600 mass spectrometer (AB SCIEX, USA) and operated in full scan mode was used for untargeted analysis of serum samples. Each sample was run in duplicate in electron spray ionization^+/− (ESI^+/−) modes. An aliquot of 3 μl extracted plasma sample was injected onto an ACQUITY UPLC HSS T3 C18 (2.1 × 100 mm, 1.8 μm) column (Waters, USA) operating at 40°C. The auto-sampler was conditioned at 4°C. Untargeted metabolomics were detected as described in a previous study (6).

Raw data files from LC-MS were converted by Analyst^®TF 1.7 software and imported into Markview software to match the peaks and the metabolites identified by mass spectral database of Dalian Institute of Chemical Physics, Chinese Academy of Sciences. Mass-to-charge ratio difference less than 0.001 was regarded as the same substance. Then the dataset of normalized peak height intensity, retention time (RT), metabolites names, and sample numbers were analyzed by SIMCA 14.1 software. SIMCA 14.1 conducted a multivariate statistical analysis of the principal component analysis (PCA) and orthogonal partial least-squares discrimination analysis (OPLS-DA) and permutations. The metabolites with P value < 0.05 and variable importance in the projection (VIP) of >1.0 were considered as statistically significant metabolites. The metabolic joint pathway analysis was carried out on the website visualization tools of MetaboAnalyst 5.0 (32).

Five serum samples were randomly selected from group C and group VA, respectively, and prepared for UHPLC-MS analysis. All serum samples were thawed on the ice. An aliquot of 50 μl serum sample was precipitated by adding 5 μl 1,4-Butane-1,1,2,2,3,3,4,4-d8-diamine and 167 μl methanol, vortex for 1 min, then adding 334 μl chloroform and vortex again for 1 min. Supernatants were collected by centrifugation (15,000 rpm, 10 min) at 4°C. Then 100 μl sodium bicarbonate-sodium bicarbonate buffer (pH=9) and 50 μl dansyl chloride solution (dissolved in acetone) were added to the supernatant (33), and stayed for 1 h at room temperature in dark area. Subsequently, the organic phase was extracted with acetic ether twice. Notably, trifluoroacetic acid was added before the second extraction. Finally, the organic phase was transferred to fresh tube and dried in solvent evaporator (Genevac, UK) at 45°C for 2 h. The residue was reconstituted in 100 μl of a mixture of 0.2 mol/L ammonium acetate/acetonitrile (3:7, vol/vol) for UHPLC-MS analysis.

Waters iClass UHPLC system (Waters, USA) coupled with a Triple Quad™ 6500+ (AB SCIEX, USA) was applied for targeted metabolomics analysis. An aliquot of 1 µl sample solution was injected onto Ultimate XB-AQ chromatographic column (100 mm × 2.1 mm, 3 μm) maintained at 40°C. The auto-sampler was conditioned at 4°C. For carrying out analysis, the mobile phase was composed of A (0.1% formic acid in water) and B (acetonitrile acidified by 0.1% formic acid) with different concentration gradient. The flow rate was 0.4 ml/min. Mass spectrometric (MS) parameters were applied as follows: ionization temperature 450°C, ion-source gas 1 pressure 55 psi, ion-source gas 2 pressure 55 psi, curtain gas pressure 40 psi, and ion-source voltage 5,500 V.

Cell growth was evaluated by using CCK8 assay according to the method described in the literature (34). Cells were cultured for 24 h with dialyzed fetal bovine serum and RPMI-1640 medium without arginine, then seeded at a density of 1,500 cells/well in 96-well plates. MM cells were cultured with different concentrations of arginine for 24 and 72 h. And 10 µl CCK8 was added to each well for 3 h before detection. The absorbance was measured at A450 nm with a microplate plate reader (Thermo Fisher Scientific, Inc., USA).

Survival analyses were made by using the Kaplan Meier method. Statistical analyses were performed by using GraphPad Prism 8 software. The statistical results were conducted with Log-rank (Mantel-Cox), and value of P< 0.05 was regarded as a significant difference (*P<0.05, **P<0.01, ***P<0.001).

The MM mouse model was established by tail vein injection of 5TMM3VT murine myeloma cells ( Figure 1A ) and subjected to four different groups: only modeling (group C), bortezomib treatment (group V), acupuncture treatment (group A), and combination therapy (group VA). As Figure 1B shows, the survival rates of myeloma mice in group V (median time of 73.5 days) and A (median time of 67 days) were evidently improved compared with the group C (median survival time of 47 days). Intriguingly, the median survival time of the myeloma mice treated with VA significantly prolonged to 79 days, and in the sixth week after modeling, the survival curves began to show significant differences between the group VA and the group C ( Figure 1C ).

Serum was collected from the myeloma mice in each group, which was used to examine the characteristics of metabolites by LC-MS. The results showed that the peak patterns of total ion current (TIC) obtained in ESI⁺ ( Figures 2A–D ) and ESI⁻ ( Figures 2E–H ) modes were distinctly different. The serum metabolites of the MM mice in each group were well separated under the same detection mode. Within 5~13 min of the injection, there were significant differences between group C and groups A, V, or VA.

The principal component analysis (PCA) was used to reflect the degree of dispersion among the four groups. Differences and changes in metabolic profiles of MM mouse serum from each group were evaluated by PCA in ESI⁺ ( Figure 3A ) and ESI⁻ ( Figure 3B ) modes. The results displayed a significant separation of serum samples from mice in the four groups with good clustering of samples within groups ( Figures 3A, B ), as well as the three-dimensional (3D) scatter plot ( Figures 3C, D ).

The orthogonal partial least-squares discrimination analysis (OPLS-DA) model of serum metabolomics from myeloma mice showed the significant differences in group V, A, or VA compared with group C in both ESI⁺ ( Figures 4A, C, E ) and ESI⁻ ( Figures 4G, I, K ) modes. In addition, all the permutation test results indicated that the fitted model was reliable ( Figures 4B, D, F, H, J, L ). The differential metabolites that satisfied the criterion (VIP >1.0 and P value <0.05) were considered as significantly different substances. There were 97 different substances in the serum of group V compared with group C, including 64 upregulated and 33 downregulated substances ( Figures 4A, B, G, H ). There were 151 different serum substances between group A and group C, with 113 upregulated and 28 downregulated substances ( Figures 4C, D, I, J ). Importantly, we found 174 different substances in the serum of group VA in comparison with group C, including 102 upregulated and 72 downregulated substances ( Figures 4E, F, K, L ).

To narrow down the potential therapeutic targets, all the significantly different substances from each comparison groups were collected to plot Venn diagrams. Excluding the intersection, there were 20 upregulated ( Figure 5A and Table 1 ) and 32 downregulated ( Figure 5B and Table 2 ) distinct metabolites in the serum of group VA. Subsequently, MetaboAnalyst 5.0 was used to analyze the joint pathways of differential metabolites in ESI⁺ ( Figure 5C ) and ESI⁻ ( Figure 5D ) modes, respectively. In Figure 5D , the main pathway of arginine and proline metabolism was involved in the serum of group VA with impact 0.20964 (−log(P)=4.6259). Ornithine and arginine were major metabolites in arginine and proline metabolism pathway.

Cluster analysis and heatmap showed that both ornithine (median of m/z=133.1057307) and arginine (median of m/z=347.2197103) were significantly decreased in the serum of group VA compared with group C ( Figure 5E ). To a large degree, these data illustrated that VA treatment inhibited arginine and proline metabolism pathway, thus causing arginine and ornithine reduction. Additionally, ornithine was also involved in the regulation of the glutathione metabolic pathway (−log(P)=41.4122) in Figure 5D . These results suggested that ornithine might be a therapeutic target of VA treatment in MM.

To further prove the above data, we conducted targeted metabolomics to detect ornithine concentration in the serum of MM mice. The chromatogram revealed a characteristic peak of ornithine standard at 3.76 min after injection ( Figure 6A ). In Figure 6B , according to the linear standard curve (r=0.99796), ornithine content in the serum samples of group VA (Average concentration of 7,333.33 ng/ml) was decreased by 73.36% compared with the group C (Average concentration of 1,953.33 ng/ml) ( Figure 6C ); however, it didn’t reach statistical difference due to the relatively small sample size in each group and large individual variation. In agreement with previous results of untargeted metabolomics, these data confirmed that VA treatment decreased the level of serum ornithine.

Arginine is a semi-essential amino acid that can be metabolized into ornithine, which is a non-essential amino acid ( Figure 6D ). We further assessed the effect of supplying extra arginine on MM cell proliferation by using CCK8 assay. As shown in Figures 7A–D , the viability of human ARP1, H929, OCI and mouse 5TMM3VT cells was significantly increased upon serial concentration of arginine (5 nM~5 μM) treatment for 72 h, suggesting that VA treatment could regulate arginine and its metabolites to promote MM cell proliferation.

To gain further insights into the deregulated ornithine, we also explored the relationship between ODC1 known as the coding gene encoding ornithine decarboxylase ( Figure 6D ) and the prognosis of MM patients. GEP analysis showed that increased ODC1 expression was associated with poor overall survival (OS) in MM patients (TT2, GSE2658) (p=0.0002; Figure 7E ). This result was also verified in the APEX phase III clinical trial with relapsed MM patients (p=0.0009; Figure 7F ). Furthermore, analyses of two gene expression omnibus (GEO) databases, GSE5900 (p<0.0001; Figure 7G ) and GSE6477 (p=0.0350; Figure 7H ), demonstrated that ODC1 mRNA was significantly increased in MM patients compared with smoldering myeloma (SMM), monoclonal gammopathy of undetermined significance (MGUS), and normal plasma (NP).