Probiotic camel milk powder improves glycemic control, dyslipidemia, adipose tissue and skeletal muscle function in T2DM patients: a randomized trial

Background: Due to the close association between gut microbiota and diabetes, probiotic dairy products draw a lot of attention in the development of functional foods with anti-diabetic activity. Methods: 28 type II diabetic patients twice a day received 10 grams of camel milk powder supplemented with BBA6 and camel milk powder (control) over a total of 4 weeks. Results: After the intervention, there was a signicant decrease in fasting blood glucose, serum content of total cholesterol, the cardiovascular risk index (TC/HDL-C), the pro-inammatory cytokines (IL-6, MCP-1) and adipokines (adiponectin, resistin, lipocalin-2, adipsin). Myokines (irisin, osteocrin) increased signicantly, indicating possible improvement in skeletal muscle function. Gut microbiota analysis suggested a signicant enrichment in unclassied_f__Sphingomonadaceae and depletion in Eggerthella after the 4-week intervention with the probiotic camel milk powder, there were also elevated fecal concentrations of elevated fecal concentrations of proline, uracil and galactinol accompanied with a decreased norleucine, glycerol, sedoheptulose, palmitic acid, 5-aminovaleric acid, inositol and γ-aminobutyric acid. Conclusion: Dietary supplement with 10 grams of probiotic camel milk powder twice a day for a consecutive 4 weeks can signicantly decrease fasting blood glucose of type 2 diabetic patients. This functional food also improves dyslipidemia, inammation and functions of adipocytes and skeletal muscle, indicating the possibility of probiotic camel milk powder as a dietary treatment that target metabolic syndrome such as diabetes. of CA. Our study reported a signicant enhanced anti-diabetic activity of camel milk when combined with BBA6, which can be used as a functional food in assisting treatment of type 2 However, the improvement in 2 h postprandial blood glucose, serum insulin level was not signicant. Further research is required for a less dose but longer intervention time in more subjects and tests for glucose tolerance and glucose stimulated insulin response can be took into consideration for the antidiabetic activity.

desert areas of Africa/Middle East or the cooler dry areas of Asia (31), leading to the unavailability of fresh camel milk for people lived in other areas. Moreover, all the exiting clinical trials were based on fresh camel milk. Dairy products are important vectors for the delivery of probiotics to humans, therefore, our lab develop a probiotic camel milk powder product, and in the present study, we evaluate its effect on blood glucose, lipid pro le, inflammatory cytokines, myokines and adipokines in T2DM patients.

Methods And Material
Trial design and sample size This was a randomized, parallel, double-blind trial in type 2 diabetic patients, conducted in Beijing Chinese Medicine Hospital Pinggu Hospital. This study met the CONSORT criteria as recommended elsewhere (32). The study was approved by the local ethics committee of China Agricultural University (CAUHR-2018026), and registered at ClinicalTrials.gov (NCT04296825).
The estimate sample size of 20 was calculated using the parallel clinical trial formula, assuming an alpha error of 0.05 and a power of 80%. Supposing an estimated 10% dropout rate, there were 22~23 patients for each group (45 patients in total). Participants A total of 45 type 2 diabetic patients were recruited from subjects attending to the clinic of Beijing Chinese Medicine Hospital Pinggu Hospital. Inclusion criteria were age 35-68 years, absence of gastrointestinal disease, and willingness to abstain from intake of all kinds of other milk, probiotic food and fermented dairy products during the study but otherwise stick to previous eating habits. Exclusion criteria were pregnancy or lactating in women, cancer, allergy or intolerance to camel milk or cow milk.
These criteria were veri ed during an inclusion visit that included a physical medical examination, dietary and physical activity assessments, standard anthropometrics and fasting glycaemia, insulin and lipid pro le were evaluated. After the veri cation, 40 subjects were eligible for participating in the study. Study procedure was explained for participants, and all participants provided written informed consent.

Randomization, blinding and intervention protocol
The 40 participants were randomly divided into two groups (20 individuals in each group): camel milk powder containing BBA6 at a dose of 2×10 10 viable cells and camel milk powder as control. Both powders were packaged in the same bags (10 gram each bag) and taken twice daily after breakfast and dinner respectively for 4 weeks. All the samples were provided by Xinjiang Jintuo Co., Ltd. (Xinjiang, China) and packaged by Sanhe Fucheng Biological Technology Co. Ltd (Langfang, China). The nutritional contents of camel milk powder used in this study is detailed in Supplementary Table S1.
Patients were given su cient supplies of the two products at the beginning of the intervention.
All participants were asked to maintain their previous diet except all kinds of other milk, probiotic food and fermented dairy products, physical activity, and medications during the study. During the study, participants underwent interviews regarding adverse effects, symptoms, or changes in quality of life and dietary every week. The allocation groups were unrevealed to the participants as well as to researchers who delivered probiotic camel milk or camel milk alone, or to who conducted the weekly follow-ups.

Blood sample collection
Blood samples were collected twice at the beginning (W0) and the end (W4) of the study, respectively. On the day of blood sample collection, patients came to the hospital without breakfast and after the collection of the fasting blood samples, they were given the same breakfast and the 2 h postprandial blood samples were collected after 2 hours of the rst bite of breakfast.
Human peripheral blood was collected in Vacutainer tubes and (Cat # 368921, BD Biosciences) and Vacutainer heparin tubes (Cat # 367886, BD Biosciences), respectively. Blood samples were centrifuged at 1500 ×g for 30 min at room temperature and samples in for fasting glycaemia, 2 hour postprandial glycaemia, insulin, uric acid and lipid measurements within 1 h after blood collection, and serum samples in Vacutainer heparin tubes were carefully removed, aliquoted, snap-frozen in liquid nitrogen, and stored in aliquots at −80°C until further analysis.

Clinical measurements
Serum insulin were measured using the Architect i2000SR analyzer (Abbott Diagnostics, Abbott Park, IL), blood glucose, content of total cholesterol (TC), total triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C) were measured using a Roche cobas ® e 411 analyzer (Roche, Hvidovre, Denmark) according to the manufactures' protocol by the certi ed core clinical laboratory at the Beijing Chinese Medicine Hospital Pinggu Hospital.

Fecal sample collection
Fecal samples were collected in the morning of the day of blood collection by patients themselves at home. Before defecating waterproof paper was rst put into closestool to keep feces away from liquids, then a portion of feces was put into sterile tubes containing RNAlater (Qiagen, Hilden, Germany) and the other portion was put into empty sterile tubes. Some fecal samples were collected the day before blood collection due to a higher defecation frequency. Fecal samples were brought to hospital in ice boxes and then stored at −80°C.
Gut microbiota analysis DNA was extracted from fecal samples using the phenol-chloroform extraction method (33) and quanti ed using a NanoDrop spectrophotometer (OneC, Thermo Fisher Scienti c, Waltham, MA, USA) and stored at −80°C until further analysis. DNA was ampli ed using the universal primers 338F (5'-ACTCCTACGGGAGGCAGCAG-3') and 806R (5'-GGACTACHVGGGTWTCT AAT-3') to target the V3-V4 region of bacterial 16S rRNA. The resulting 468-bp-sized products were assessed, quanti ed, pooled and sequenced on an Illumina Miseq PE300 platform (Illumina, San Diego, CA, USA) at Shanghai Majorbio Bio-pharm Technology Co. Ltd. (Shanghai, China) using a paired-end sequencing strategy. Raw data were spiced, ltered and then used to select the operational taxonomic units (OTUs) with USEARCH software (version 7.0) and a default cutoff of 97% sequence similarity.
OTUs were further subjected to the Ribosomal Database Project classi er software for taxonomic identi cation with an 80% con dence threshold at the phylum, class, order, family, genus and species levels. Further analysis such as ANOSIM/Adonis tests, principal coordinates analysis, abundance heatmap and differences in gut microbiome composition were analyzed on the free online platform of Majorbio I-Sanger Cloud Platform (https://cloud.majorbio.com/) using weighted unifrac distance matrices.

Fecal metabolomics analysis
Fifty milligrams of feces were mixed with 40 μL internal standard (0.3 mg/mL, 2-chloro-L-phenylalanine dissolved in methanol) and ultrasonically extracted with 360 μL methanol then with 200 μL chloroform and 400 μL ddH 2 O in ice bath for 30 min, respectively. After extraction, samples were centrifuged at 12000 rpm at 4 °C for 10 min. Then, 400 μL supernatant was volatilized and oximated with 80 μL methoxyamine hydrochloride in pyridine (15 mg/mL) and shaking them for 90 min at 37 °C, after which they were trimethylsilylated by adding 30 μL BSTFA (containing 1% TMCS) and 20 μL n-hexane incubating for 1 h at 70 °C. After left at room temperature for 30 min, samples were then subjected to GC-MS analysis.
Metabolic pro ling of fecal samples was acquired by an Agilent 7890 A/5975C GC-MS (Agilent Technologies, Santa Clara, CA, USA) using a HP-5MS fused silica capillary column (30 m × 0.25mm × 0.25 μm, Agilent J&W Scienti c, Folsom, CA, USA). 1 μL sample was injected in a non-split mode. The injector, ion source and quadrupole rod temperatures were 260 °C, 230 °C and 150 °C respectively. Highpurity helium (>99.999%) was used as the carrier gas with a ow rate of 1.0 mL/min. The GC oven temperature program consisted of 60 °C for 2 min, after which the temperature ramped to 310 °C at 8 °C /min, and held steady for 6 min. Mass spectra were acquired and the mass scan range was set at m/z 50-600. Fecal samples were analyzed randomly.
Raw GC-MS mass spectra were converted to CDF format les by ChemStation (version E.02.02.1431, Agilent, CA, USA) and subsequently preprocessed using Chroma TOF (version 4.34, LECO, St Joseph, MI), including raw signal extraction, data baseline ltering, peak identi cation, and integration. After alignment with the statistical comparison component, the ".CSV" le was obtained with three-dimension data sets including sample information, retention time, the mass-to-charge ratio and peak intensity.
Identi cation of metabolites was conducted using the Automatic Mass Spectral Deconvolution and Identi cation System, which was searched against commercially available databases such as the National Institute of Standards and Technology and Fiehn libraries. The signal integration area of each metabolite was normalized to the internal standard (2-chloro-L-phenylalanine) for each sample.
The normalized data were transformed using SIMCA-P 14.0 software (Umetrics AB, Umea, Sweden) for principal component analysis and partial least Squares-discriminant analysis (PLS-DA). The variable importance in projection (VIP) values of all the metabolites from the PLS-DA model was taken as criteria to nd the variable importance of differential metabolites, and variables with a VIP >1.0 and a p-value < 0.05 were considered relevant for group discrimination. The statistical signi cance between two groups was evaluated by a univariate Student's t-test.

Statistical analysis
Data entry was performed twice by two separate persons. Differences between W0 and W4 of each group were evaluated by paired two-tailed Student's t-tests using GraphPad Prism version 7.0 software (San Diego, CA, USA). Differences between the two groups at the same time point (W0 or W4) were compared by unpaired two-tailed Student's t-tests using GraphPad Prism. Statistical signi cance was evaluated at an alpha level of 0.05.

Study population
As shown in Figure 1, of the 40 participants that were randomized, 5 persons did not come to pick up the intervention products, 2 persons were lost to follow-up due to go out for travel and 5 poor compliance (took other dairy products and probiotics). At the end of the 4-week intervention, 28 participants completed the experiment and subjected to the analysis, 14 received camel milk powder supplemented with BBA6 (CA) and 14 received camel milk powder (C). None of the participants reported any adverse effects including gastrointestinal disorders. Baseline comparison showed no signi cant differences in blood glucose, insulin and lipid pro les between different groups (p>0.05, Table 1).

Changes in glycemic indices and serum insulin
Fasting blood glucose, 2 h postprandial blood glucose and fasting serum insulin of patients before and after 4-week intervention were shown in Figure 2. At baseline, there were both no signi cant differences between the four groups (p>0.05). The hypoglycemic effect of camel milk has been proved in type I (20)(21)(22)(23)(24)(25)(26) and type II diabetic patients (27)(28)(29)(30), in this study, patients in the CA group exhibited a signi cant decrease in fasting blood glucose after the intervention (p = 0.0458, Figure 2A) and a more effective hypoglycemic activity than camel milk powder alone (p = 0.0441, Figure 2B). However, there were no signi cant changes in 2 h postprandial blood glucose of patients either before and after the intervention ( Figure 2C) or between the two groups ( Figure 2D).
Serum content of insulin was also not affected ( Figure 2E, p>0.05) and the intervention did not improve the insulin resistance of the patients ( Figure 2F, p>0.05). Previous studies about camel milk found a consistent unchanged insulin level in type I diabetic patients (20)(21)(22)(23)(24)(25)(26)(27), but in type II diabetic patients, the existing results were inconsistent (20,31,34) which may be due to the complicate mechanism in type II diabetes mellitus.

Changes in lipid pro le and cardiovascular risk
The relationship between diabetes and atherosclerotic cardiovascular disease are well established, with a signi cantly elevated risk for cardiovascular disease in diabetic patients (35), therefore we also measured serum content of TC, TG and the indicators of vascular risk (LDL/HDL cholesterol ratio and TC/HDL-C, Figure 3). Previous clinical studies seldomly reported the effects of camel milk on lipid pro le, although animal studies reported a consistent decrease in TC (27,28,36). In the limited studies in type II diabetic patients, one study (27) was in accordance with ours while another one reported that there were no changes in lipid pro le (29). As we can see from Figure 3, both at baseline (W0) and post-intervention (W4), there were no signi cant differences between the two groups (CA-W0 vs C-W0 and CA-W4 vs C-W4, p>0.05), whereas after the intervention, TC content of patients in CA and C group decreased compared with values at baseline (CA-W0 vs CA-W4, p = 0.0697 and C-W0 vs C-W4, p = 0.0225, Figure 3A). Furthermore, although there was no change in TG ( Figure 3B) and the decreased TC in group CA was nonsigni cant (p = 0.0697, Figure 3A), intervention of CA resulted a signi cant decrease in the ratio of TC and HDL-C (TC/HDL-C, p = 0.0364, Figure 3D), indicating a decreased vascular risk.
It was reported that there was a chronic in ammation in diabetes (37), and a greater antioxidant and immunomodulatory activity of camel milk protein than bovine and other whey proteins (8,38). Previous studies found that camel milk or camel milk whey proteins was shown to reduce the proin ammatory IL-1β, IL-6, and TNFα in diabetic rats (28,39,40). As shown in Figure 4, there were no signi cant differences in serum contents of in ammatory markers (TNF-α, IL-6, MCP-1) between groups both at baseline and post-intervention (p>0.05). Within group comparisons (W0 vs W4) suggested that the decreased TNF-α (p>0.05), IL-6 (p = 0.0103) and MCP-1 (p = 0.0814) contents all in CA group were more obvious than those in C group.

Changes in adipokines and myokines pro le
Recent evidence has identi ed skeletal muscle and adipocytes as secretory organs, which communicate with each other to regulate energy homeostasis and insulin sensitivity though the cytokines called myokines and adipokines, respectively (41,42). Therefore, we measured serum contents of adipokines (adiponectin, resistin, lipocalin-2, adipsin) and myokines (FGF-21, irisin, osteocrin, osteonectin) in patients before and after 4-week intervention, and the results were shown in Figure 5. There were no signi cant differences between different groups at baseline (W0) and after intervention (W4, p>0.05). Intervention with camel milk powder supplemented with BBA6 signi cantly decreased the content of adipokines (adiponectin, resistin and lipocalin-2 and adipsin) and increased myokines (irisin and osteocrin) levels. Although increased level of adiponectin ( Figure 5A) and adipsin ( Figure 5D) were found to be associated with a lower risk of type 2 diabetes (43,44) in human and improvement in pancreatic beta-cell function in mice (45,46), respectively, we found a signi cant decrease in patients with a signi cant decrease in fasting blood glucose (group CA). The other two adipokines, resistin ( Figure 5B) and lipocalin-2 ( Figure 5C), which decreased signi cantly in patients intervened with camel milk powder alone and in combination, was reported to be good for the improving of diabetes. Elevated serum lipocalin-2 is closely and independently associated with impaired glucose regulation and type 2 diabetes in Chinese people (47), and the lipocalin-2 de ciency attenuates insulin resistance associated with obesity in mice (48). Resistin promotes insulin resistance in mice, whereas whether it does so in humans is unclear (49,50) because it was synthesized in adipocytes in mice whereas in humans it is generated by macrophages and monocytes, but not adipocytes (51).
As for the myokines, there was a signi cant decrease in resistin after the intervention accompanied with the decreased fasting blood glucose, it maybe also positively correlated with the hyperglycemia. Furthermore, the signi cant and speci c decrease in irisin (p = 0.0079, Figure 5F) and osteocrin (p = 0.0033, Figure 5G) was also an indicator for the improvement in diabetes. Circulating irisin levels were reported to be associated negatively with the risk of the metabolic syndrome in individuals from China (52), and the signi cant increase in patients intervened with camel milk powder supplemented with BBA6 indicated an improvement in diabetes. Osteocrin is also a regulator of bone growth as a novel vitamin Dregulated bone-speci c protein (53), suggesting an enhanced effect on bone growth after supplemented with BBA6.

Changes in gut microbiota
More and more evidence suggest a close relationship between gut microbiota and diabetes (54) and since there is a component of probiotics in our study, we analyzed gut microbiota before and after the intervention using the two-tailed Student's t-test ( Figure 6). There was a signi cant enrichment in the relative abundance of unclassi ed_f__Sphingomonadaceae (p = 0.02477) accompanied with a depletion in Eggerthella (p = 0.04577, Figure 6A) in group CA at the genus level. It was found that members of the Eggerthella genus possess particular pathogenic potential, and the signi cant decrease in its abundance after the intervention with CA may be related to the improvement in in ammation (55). There were no different genera before and after the intervention in group C. In addition, since group CA contained a dietary supplement with BBA6, we analyzed the relative abundance of Bi dobacterium animalis at the species level, which was enriched in group CA (p = 0.02754, Figure 6B). We also analyzed the different genera between group CA and C after the 4-week intervention (p = 0.0498, Figure 6C). There was a signi cant elevation in the relative abundance of Holdemania, which was not due to the difference between groups before the intervention ( Figure 6D). It was reported that the Holdemania genus was found to be associated with being lean in Japanese men (56) but also correlated with clinical indicators of impaired lipid and glucose metabolism (57).

Changes in fecal metabolites
Fecal metabolites concentrations between the two groups were compared and the top ten abundant signi cantly changed metabolites were shown in Figure 7. We previously compared the fecal metabolites from diabetic patients intervened with camel milk and cow milk, among the top ten abundant signi cantly changed metabolites, six are amino acids or the metabolites of amino acids (Supplementary Figure S1A) and the other two are FFAs (Supplementary Figure S1B), indicating the anti-diabetic activity of camel milk may be related to the different amino acid composition. However, when compared fecal metabolites in group CA and C, among the top ten abundant fecal metabolites, norleucine, glycerol, sedoheptulose, palmitic acid, 5-aminovaleric acid, inositol and γ-aminobutyric acid decreased signi cantly, whereas proline, uracil and galactinol increased signi cantly in group CA (p<0.05).
Diabetic individuals had elevated serum proline levels (58, 59) and our study found an elevated fecal proline concentration, but lower fecal levels of uracil was found in type 2 diabetes (60) and high-fat dietinduced pre-obese individuals (61). Besides uracil, it was also worth noting that the decreased fecal metabolites in group CA were kind of good for glycemic control or energy metabolism. For example, sedoheptulose was found to decrease serum levels of glucose, total cholesterol, TNF-α, IL-6, resistin in type 2 diabetic db/db mice (62), γ-aminobutyric acid can be used to treat diabetes due to the protective effects on β-cell survival and function (63,64), as well as the promotion of the conversion of α-cells to βcells (65). Supplement with norleucine in rats stimulated postprandial protein synthesis in adipose tissue, skeletal muscle, and liver (66), and inositol was shown to reduce the risk of metabolic disease in people with PCOS (67), the lysine degradation product, 5-aminovaleric acid decreases β-oxidation of fatty acids in mouse cardiomyocytes (68). Since oligosaccharides in camel milk (69) could function as prebiotics for BBA6 and gut commensal bacteria (70), we speculated that the decreased concentrations of sedoheptulose, 5-aminovaleric acid, and γ-aminobutyric acid may be the direct effect of BBA6, because there were no signi cant differences between patients received camel milk powder containing BBA6 and BBA6 alone (Supplementary Figure S2B). Combined these results, it was suggested that camel milk powder supplemented with BBA6 specially decreased fecal concentrations of glycerol and inositol and increased proline and uracil levels, which was not altered by camel milk powder alone or BBA6 alone.

Conclusion
As a traditional milk in Africa, Asia and Middle East, it was believed that regular consumption of fresh camel milk may aid in prevention and control of diabetes. In this study, we evaluated the anti-diabetic activity of a probiotic camel milk product, camel milk powder supplemented with BBA6, a strain of Bi dobacterium animalis isolated by our lab in type II diabetic patients. It was found that a 4-week intervention of this probiotic camel milk powder can signi cantly decrease fasting blood glucose, serum content of TC and the cardiovascular risk (TC/HDL-C). Meanwhile, patients intervened with camel milk powder supplemented with BBA6 also exhibited a decrease in in ammatory cytokines (IL-6, MCP-1) and adipokines (adiponectin, resistin, lipocalin-2, adipsin), as well as an improvement in myokines (irisin, osteocrin). Camel milk powder containing BBA6 signi cantly enriched the relative abundance of unclassi ed_f__Sphingomonadaceae and depleted Eggerthella after the 4-week intervention, and patients in this group exhibited a gut microbiota with an enrichement in the relative abundance of Holdemania when compared with patients supplemented with camel milk powder alone. Furthermore, elevated fecal concentrations of proline, uracil and galactinol accompanied with a decreased norleucine, glycerol, sedoheptulose, palmitic acid, 5-aminovaleric acid, inositoland γ-aminobutyric acid was found in patients of group CA. Our study reported a signi cant enhanced anti-diabetic activity of camel milk powder when combined with BBA6, which can be used as a functional food in assisting treatment of type 2 diabetes. However, the improvement in 2 h postprandial blood glucose, serum insulin level was not signi cant.
Further research is required for a less dose but longer intervention time in more subjects and tests for glucose tolerance and glucose stimulated insulin response can be took into consideration for the antidiabetic activity. Abbreviations BBA6, Bi dobacterium animalis A6; W0, the beginning of the study; W4, the end of the study; group CA, camel milk supplemented with BBA6; group C, camel milk; TC, total cholesterol; TG, total triglyceride; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TNF-α, tumor necrosis factor-α; IL-6, interleukin-6; MCP-1, monocyte chemotactic protein-1; FGF-21, broblast growth factor-21; OTUs, operational taxonomic units; PLS-DA, squares-discriminant analysis; VIP, variable importance in projection.        Different gut bacteria at the genus level in patients intervened with cow milk (P, placebo) and camel milk supplemented with BBA6 (CA) before (W0) and after (W4) the intervention analyzed by two-tailed Student's t-test. C and D, relative proportion of Bi dobacterium animalis in patients intervened with camel milk supplemented with BBA6 (CA) and BBA6 alone (A). Different gut bacteria at the genus level in each group before (W0) and after (W4) the intervention analyzed by two-tailed Student's t-test. Patients intervened with A, cow milk (group P); B, camel milk supplemented with BBA6 (group CA); C, camel milk alone (group C) and D, BBA6 alone (group A).