The administration of high-mobility group box 1 fragment prevents deterioration of cardiac performance by enhancement of bone marrow mesenchymal stem cell homing in the delta-sarcoglycan-deficient hamster

Objectives We hypothesized that systemic administration of high-mobility group box 1 fragment attenuates the progression of myocardial fibrosis and cardiac dysfunction in a hamster model of dilated cardiomyopathy by recruiting bone marrow mesenchymal stem cells thus causing enhancement of a self-regeneration system. Methods Twenty-week-old J2N-k hamsters, which are δ-sarcoglycan-deficient, were treated with systemic injection of high-mobility group box 1 fragment (HMGB1, n = 15) or phosphate buffered saline (control, n = 11). Echocardiography for left ventricular function, cardiac histology, and molecular biology were analyzed. The life-prolonging effect was assessed separately using the HMGB1 and control groups, in addition to a monthly HMGB1 group which received monthly systemic injections of high-mobility group box 1 fragment, 3 times (HMGB1, n = 11, control, n = 9, monthly HMGB1, n = 9). Results The HMGB1 group showed improved left ventricular ejection fraction, reduced myocardial fibrosis, and increased capillary density. The number of platelet-derived growth factor receptor-alpha and CD106 positive mesenchymal stem cells detected in the myocardium was significantly increased, and intra-myocardial expression of tumor necrosis factor α stimulating gene 6, hepatic growth factor, and vascular endothelial growth factor were significantly upregulated after high-mobility group box 1 fragment administration. Improved survival was observed in the monthly HMGB1 group compared with the control group. Conclusions Systemic high-mobility group box 1 fragment administration attenuates the progression of left ventricular remodeling in a hamster model of dilated cardiomyopathy by enhanced homing of bone marrow mesenchymal stem cells into damaged myocardium, suggesting that high-mobility group box 1 fragment could be a new treatment for dilated cardiomyopathy.

a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 All surgeries and sacrifices were performed under deep anesthesia with isoflurane sufficient to minimize animal suffering. All experimental procedures and evaluations were performed in a blinded manner.

Experimental animals
Male J2N-k hamsters, which are deficient in δ-sarcoglycan, were used for this study. Since mutations in δ-sarcoglycan are also detected in DCM patients [15], J2N-k hamsters are a recognized and established animal model of DCM. They exhibit progressive myocardial fibrosis and moderate cardiac dysfunction at 8-9 weeks of age. Accordingly, the average life span of J2N-k hamsters is much shorter (approximately 42 weeks) than control hamsters (approximately 112 weeks) [16,17].

HMGB1 fragment
Mesenchymal stem cell mobilization domain from human HMGB1 was produced as "HMGB1 fragment" by solid-phase synthesis and provided by StemRIM (StemRIM Inc., Osaka, Japan). The HMGB1 fragment was dissolved in phosphate buffered saline (PBS) to a concentration of 1 mg/ml before administration. week-old J2N-k hamsters were purchased from Japan SLC (Shizuoka, Japan). HMGB1 fragment (3mg/kg/day; HMGB1, n = 15) or PBS (3ml/kg/day; control, n = 11) was administered for 4 consecutive days at the age of 20 weeks in the following manner: The external jugular vein was exposed by a median neck skin incision under 1.5% isoflurane anesthesia. Subsequently, HMGB1 fragment or PBS was injected through the external jugular vein. After the complete hemostasis, the skin incision was closed, and the hamsters were housed in a temperature-controlled cage.

Transthoracic echocardiography
Transthoracic echocardiography was performed to assess cardiac function using M-mode echocardiography with Vivid I (GE Healthcare) under isoflurane inhalation (1%). Diastolic and systolic dimensions of the left ventricle (LVDd/Ds), and left ventricular ejection fraction (LVEF) were measured before treatment, and reassessed at 4 and 6 weeks after treatment.
More than 5 sections from the HMGB1 group (n = 15) and the control group (n = 11) were prepared per specimen and 3 low power fields per section were analyzed and averaged. The fibrotic area, the expression of α-sarcoglycan and α-dystroglycan, and the 4-hydroxynonenal positive area were measured using Metamorph image analysis software (Molecular Devices, Inc., Downingtown, PA). BZ-analysis software (Keyence, Tokyo, Japan) was used to measure the capillary density, the number of PDGFRα + and CD106 positive (CD106 + ) cells, the number of CXCR4 + cells and SDF-1 positive area (mm 2 ), the number of CD68 + cells, annexin V + cells, and dihydroethidium positive dots.

Transmission electron microscopy
Cardiac tissue was pre-fixed with Karnovsky fixative containing 2.5% glutaraldehyde and 2% paraformaldehyde in a 0.1 M (pH 7.4) cacodylate buffer for 2 hours at 4˚C and post-fixed with 2% osmium tetroxide for 2 hours at 4˚C. The samples were then immersed in 0.5% uranyl acetate for 3 hours at room temperature, dehydrated in ethanol (50%, 70%, 80%, 90%, 95%, and 100%) and propylene oxide, and embedded in epoxy resin. Semithin sections (0.5μm) were stained with 0.1% toluidine blue solution and examined under a light microscope. Ultrathin sections were made with a Leica ultramicrotome. These sections were counterstained with uranyl acetate and lead citrate, before examination with a Hitachi H-7100 electron microscope at 75 kV. Transmission electron microscopy analysis was conducted using 2 randomly selected samples from each group.

Real-time polymerase chain reaction
Total RNA was extracted from cardiac tissue of HMGB1 group (n = 15) and the control group (n = 11) and reverse transcribed using Omniscript reverse transcriptase (Quiagen, Hilden, Germany). The resulting cDNA was used for real-time polymerase chain reaction with the ABI PRISM 7700 system (Applied Biosystems) and Taqman Universal Master Mix (Applied Biosystems, Division of Life Technologies Corporation, Carlsbad, Calif). Hamster-specific primers were used for tumor necrosis factor-α stimulating gene 6 (TSG-6), vascular endothelial growth factor (VEGF), and CXCR4. Each sample was analyzed in duplicate for each gene studied. Data were normalized to glyceraldehyde-3-phosphate dehydrogenase expression level. For relative expression analysis, the ddCT method was used, and a sample from a control hamster was used as reference. The real-time polymerase reaction was also conducted using Fast SYBR Green Master Mix and primers designed for hepatic growth factor and glyceraldehyde-3-phosphate dehydrogenase as shown in S1 Table. For relative expression analysis, we prepared a 5-fold serial standard curve using a sample from a HMGB1 hamster as reference.

Evaluation of hamster prognosis after treatment
The life-prolonging effect of the HMGB1 fragment on J2N-k hamsters was assessed separately. Twenty-week-old J2N-k hamsters were treated with HMGB1 (HMGB1, n = 11) or PBS (control, n = 9) as described above. An additional treatment group received monthly administration of HMGB1 fragment 3 times, (monthly HMGB1, n = 9) to evaluate the long-term therapeutic effects of HMGB1 fragment. The animals were randomly allocated to each treatment group and housed after the initial treatment. The survival rate in the 3 groups was calculated and the significant difference between the groups was tested at 22 weeks (equal to 42 weeks of age, the average lifespan of J2N-k hamsters) by log-rank analysis.

Statistical analysis
Continuous variables were summarized as means with standard deviations and compared using an unpaired t-test. Survival curves were prepared using the Kaplan-Meier method and compared using the log-rank test. All data were analyzed using JMP Pro 12 (SAS Institute, Cary, NC, USA). Differences were considered statistically significant at P-values < 0.05.

Preserved cardiac performance with HMGB1 fragment administration
The functional effects of HMGB1 fragment on the DCM heart were assessed by transthoracic echocardiography over time. LVDd/Ds and LVEF at 20 weeks of age, just before the treatment, were not significantly different between the HMGB1 group and the control group. After treatment, echocardiography showed that the LVEF was significantly preserved until 6 weeks in the HMGB1 group compared with the control group (4 weeks: 43±8% vs 33±9%, p = 0.01; 6 weeks: 41±7% vs 31±7%, p = 0.0001, HMGB1 vs control, respectively) (Fig 1).

Effect of HMGB1 fragment on myocardial fibrosis
The degree of myocardial fibrosis 6 weeks after HMGB1 fragment treatment was assessed by picrosirius red staining and compared with control group. Quantification of fibrotic area confirmed that the degree of myocardial fibrosis was significantly reduced in the HMGB1 group compared with the control group (16.6±3.8% vs 22.7±5.4%, respectively, p = 0.04) (Fig 2).

Increased vasculature in the heart after HMGB1 fragment administration
Six weeks after the treatment, capillary vascular densities were assessed by CD31 immunostaining. In the HMGB1 group, the number of CD31 positive arterioles and capillaries was significantly increased compared with the control group (654±171 units/mm 2 vs 484±74 units/ mm 2 , respectively, p = 0.02) (Fig 3).

PDGFRα and CD106 positive cells in the hearts
Immunohistochemistry showed that the number of PDGFRα + and CD106 + cells in the heart tissue was significantly greater in HMGB1 group compared with the control group (12±5 cells /field vs 4±2 cells/field, respectively, p<0.001) (Fig 4).

Increased CXCR4 positive cells in the heart after HMGB1 fragment administration
Immunohistochemistry showed significantly increased ratio of the number of CXCR4 + cells to SDF-1 positive area (mm 2 ) in heart tissue in the HMGB1 group than in the control group (1.3 ±1.0 vs 0.3±0.1 cells/mm 2 , respectively, p = 0.02) (Fig 5).

Mitochondrial ultramicrostructure
Transmission electron microscopy of the myocardium showed a relatively regular arrangement of mitochondrial cristae in the HMGB1 group. In contrast, the mitochondrial cristae were disordered in the control group (Fig 7).

Effect of HMGB1 fragment on inflammatory response
Immunohistochemistry showed that the number of CD68 + cells in the heart tissue was significantly decreased in HMGB1 group compared with the control group (568±222 cells /mm 2 vs 293±107 cells /mm 2 , respectively, p = 0.03) (Fig 9).

Effect of HMGB1 fragment on cardiomyocyte apoptosis
The effect of HMGB1 fragment administration on cardiomyocyte apoptosis was assessed by annexin V staining. The number of annexin V + cells was significantly reduced with HMGB1 fragment administration (878±343 cells /mm 2 vs 400±255 cells /mm 2 , respectively, p = 0.02) (Fig 10).

Upregulated TSG-6, VEGF, HGF, and CXCR4 in the heart after HMGB1 fragment administration
Real-time PCR was used to quantitatively assess the expression levels of BMMSC-derived factors, such as VEGF, TSG-6, HGF, and CXCR4. Intramyocardial mRNA levels of VEGF, TSG-6, and HGF were significantly upregulated in HMGB1 group compared with the control group

Survival benefit of monthly HMGB1 administration
Survival of J2N-k hamsters was assessed using the Kaplan-Meier method. There was no significant difference in survival between HMGB1 and control. In contrast, the monthly HMGB1 group all survived to the full 42 weeks, and they showed significantly improved survival rate compared with control group (log-rank p = 0.001) (Fig 12).

Discussion
In the present study we have shown that, first, systemic administration of HMGB1 fragment leads to the accumulation of PDGFRα + and CD106 + cells in damaged myocardium possibly The HMGB1 group showed significantly higher CXCR4 + cells to SDF-1 positive area (mm 2 ) in heart tissue than the control group. HMGB1, high-mobility group box 1; SDF-1, stromal derived factor-1.
https://doi.org/10.1371/journal.pone.0202838.g005 Fig 6. Immunostaining for α-sarcoglycan and α-dystroglycan in cardiomyocytes. (a), Representative photomicrographs (×600, scale bar = 50μm) of immunostaining of α-sarcoglycan and α-dystroglycan in cardiomyocytes. (b), Quantitative analysis of immunostaining showed significantly increased staining of both αsarcoglycan and α-dystroglycan in the HMGB1 group than the control group. HMGB1, high-mobility group box 1. through the SDF-1/CXCR4 axis and upregulated expression of cardioprotective factors such as TSG-6, VEGF, and HGF in the heart tissue of J2N-k hamsters. Second, the myocardial histology in the HMGB1 group demonstrated significantly decreased fibrosis, increased capillary vascular density, decreased oxidative stress, decreased cardimyocyte apoptosis, decreased inflammatory response, and well-organized cytoskeletal proteins compared with the control group. Finally, cardiac function was significantly preserved after HMGB1 fragment administration and the survival benefit was shown with monthly HMGB1 fragment treatment.
The present study demonstrates the feasibility of "drug-induced endogenous regenerative therapy" using an HMGB1 fragment in a hamster model of DCM. While the precise mechanism remains unclear, it is well known that HMGB1 acts as a chemoattractant for MSCs [13,14,22]. Systemic HMGB1 administration has been reported to induce accumulation of  Tissue sections were stained with dihydroethidium to estimate superoxide production, and 4-hydroxynonenal to estimate lipid peroxidation. The HMGB1 group showed significantly reduced production of superoxide (c) and a trend towards reduced lipid peroxidation (d) compared with the control group. HMGB1, high-mobility group box 1. PDGFRα + cells around blood vessels in the bone marrow and significant increases in these cells in the peripheral blood [13]. In addition, the enhancement of CXCR4 expression with HMGB1 treatment promotes the local migration to damaged tissue through the SDF-1/ CXCR4 axis, which might be essential in DCM [23] as well as ischemic cardiomyopathy [24][25][26].
While PDGFRα + BMMSCs might be the predominant cell population mobilized by administration of HMGB1 fragment and therefore exerting therapeutic effects on damaged myocardium, it has been suggested that PDGFRα + MSCs include other defined subpopulations with distinct functions [27]. As HMGB1 is also reported to induce other cell types [28], the accumulated cells in damaged heart tissue after HMGB1 administration might be highly heterogeneous and it will therefore be important to identify in the future, specific PDGFRα + subpopulations induced by HMGB1 which have therapeutic benefits.
Paracrine signaling is a well-investigated mechanism of protective effects exhibited by BMMSCs on surrounding cells [29][30][31][32][33]. TSG-6 plays a key role in the anti-inflammatory effects of BMMSCs [32,34]. TSG-6 attenuates oxidative stress through activation of CD44 [35,36], and downregulates TGF-β by suppressing plasmin activity [34], which could result in decreased myocardial fibrosis. Since increased oxidative stress is one of the essential factors in High-mobility group box 1 fragment in dilated cardiomyopathy the pathogenesis of myocardial fibrotic changes in J2N-k hamsters [17], our results suggest that HMGB1 fragment administration could become a substantial therapy for DCM. VEGF has been known to promote angiogenesis in ischemic conditions [30,37,38], which might have a beneficial effect on the defective vascularization within the left ventricle, which is associated with the pathophysiology of DCM [39,40]. HGF is known to be a putative paracrine mediator in cardiac repair mechanisms of BMMSCs [41]. Our group previously reported several roles of HGF in damaged myocardium, including reduction of cardiac fibrosis and myocyte apoptosis [42,43], and enhancement of cytoskeletal protein expression [44].
No significant difference in survival was observed between the HMGB1 and control groups, however, animals that received monthly HMGB1 treatment showed significantly better survival compared with control. The therapeutic benefits of HMGB1 fragment might be sustained by repeated administration in J2N-k hamsters and further investigation concerning the optimal dose and interval of administration of HMGB1 fragment will be needed for the clinical use of HMGB1 fragment in DCM patients. Intramyocardial mRNA levels of TSG-6, VEGF, and HGF were significantly higher in HMGB1 group compared with the control group. PCR, polymerase chain reaction, TSG-6, tumor necrosis factor-α stimulating gene 6, VEGF, vascular endothelial growth factor, HGF, hepatic growth factor, HMGB1, high-mobility group box 1.
https://doi.org/10.1371/journal.pone.0202838.g011 There was no significant difference between the single HMGB1 treatment (n = 11) group and the control group (n = 9), whereas the monthly HMGB1 group (n = 9) showed a significantly greater survival rate than control (p = 0.01). HMGB1, high-mobility group box 1. The present study has several limitations. First, it should be noted that the J2N-k hamster is not the only available animal model of DCM. Further studies using other DCM models such as transgenic murine models lacking γ-sarcoglycan [45], dystrophin and MyoD [46], or animal models which have undergone transverse aortic constriction to induce DCM would be required to assess the potential therapeutic benefits of HMGB1 fragment for DCM patients. Second, BMMSCs have been reported to repair damaged tissue thorough their differentiation potential and paracrine signaling [47]. The differentiation of the immigrated BMMSCs was not examined in the present study.

Conclusion
Systemic HMGB1 fragment administration attenuates the progression of left ventricular remodeling in a hamster model of DCM by enhanced homing of BMMSCs into damaged myocardium, suggesting that HMGB1 fragment could be beneficial in the treatment of DCM.
Supporting information S1