VEGF and bFGF induction by nitric oxide is associated with hyperbaric oxygen-induced angiogenesis and muscle regeneration

Hyperbaric oxygen (HBO) treatment promotes early recovery from muscle injury. Reactive oxygen species (ROS) upregulation is a key mechanism of HBO, which produces high O2 content in tissues through increased dissolution of oxygen at high pressure. Nitric oxide (NO), a type of ROS, generally stabilizes hypoxia-inducible factor (HIF) 1α and stimulates secretion of vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) from endothelial cells and macrophages, which then induces angiogenesis. The purpose of the present study was to investigate whether HBO could promote angiogenesis via induction of NO and induce muscle regeneration in contused rat skeletal muscles. The HBO protocol consisted of 2.5 atmospheres absolute (ATA) 100% oxygen for 120 minutes, once a day for 5 consecutive days. We also evaluated the effects of a ROS inhibitor (NAC) or NOS-specific inhibitor (L-NAME) on HBO. HBO significantly increased NO3−, VEGF, and bFGF levels and stabilized HIF1α within 1 day. HBO promoted blood vessel formation at 3–7 days and muscle healing at 5–7 days after contusion. Administration of both NAC and L-NAME before HBO suppressed angiogenesis and muscle regeneration even after HBO. HBO thus promoted angiogenesis and muscle regeneration mainly through generation of NO in the early phase after muscle contusion injury.

Next, we analyzed the number of proliferating endothelial cells in the contused muscles. Tie2 and Ki67 double-positive cells, representing endothelial cells and cells not in the quiescent phase, were counted at 6 hours and 1, 3, and 5 days after injury (Fig. 3A,C). Quantitative analysis showed a significant increase at 1 day in the HBO group (NT group: 10.7 ± 5.7 cells/HPF, HBO group: 29.9 ± 7.5 cells/HPF, F(1,8) = 20.500, p = 0.001).
Finally, we counted the number of immature and mature vessels (Fig. 3B,D,E). Generally, vessels are evaluated using vessel-specific tomato lectin, as immature vessels are stained with lectin and do not contain laminin. Mature vessels are stained with lectin and surrounded by laminin (Fig. 3B). Quantitative analysis showed a significant increase of immature vessels in the HBO group at 3 days after injury (3 days-NT group: 14.6 ± 2.8 vessels/HPF, HBO group: 28.1 ± 2.8 vessels/HPF, F(1,8) = 11.539, p = 0.009) (Fig. 3D). Moreover, the number of mature vessels in the HBO group was significantly increased at 3 and 7 days after injury (3 days-NT group: 0.04 ± 0.02 vessels/ We performed HBO immediately after injury. Evaluation was performed at 3 hours, 6 hours, 1 day, 3 days, 5 days, 7 days, and 9 days after muscle contusion injury. (B) Inhibition schedule. ROS and NOS inhibitors were injected every day from one day before contusion injury to the day before measurement. Inhibitors were injected 30 minutes before administration of HBO.
HBO induction of NO mediated HIF1α stabilization and the effects of angiogenic factors in the contused muscle. We hypothesized that upregulation of NO underlies the therapeutics effects of HBO.
We also evaluated the effect of the inhibitors on the HBO-mediated induction of VEGF and bFGF. The upregulation of VEGF by HBO was significantly suppressed in the NAC + HBO and L-NAME + HBO groups at 3 Figure 2. HBO promoted VEGF and bFGF synthesis but did not increase HGF and angiopoietin2 production after muscle contusion injury. (A,B) The amount of VEGF and bFGF protein in the injured gastrocnemius muscle was significantly increased at 3 or 6 hours after injury in the HBO group. (C,D) There was no significant difference in the amount of HGF and angiopoietin 2 protein at each phase in the HBO group, n = 6, using total 60 animals ***P < 0.001, *P < 0.05 using multiple-way ANOVA followed by Bonferroni post-tests. Data are the mean ± SEM. Representative images of immature blood vessels (inset) positive for tomato lectin (red) and DAPI (blue) (arrows), and mature vessels (inset) positive for tomato lectin, laminin (green), and DAPI (arrow heads) in injured muscle. Scale bar: 100 µm. One of the immature or mature vessels is enlarged in the upper right of each image. (C) The number of proliferating endothelial cells was significantly increased at 1 day after injury in the HBO group, n = 5, using total 65 animals **P < 0.01, using multiple-way ANOVA followed by Bonferroni post-tests. (D) The number of immature blood vessels was significantly increased at 3 days after injury in the HBO group, n = 5. **P < 0.01 using multiple-way ANOVA followed by Bonferroni posttests. (E) The number of mature blood vessels in the HBO group was considerably increased at 3, 5, and 7 days after injury, n = 5, using 20 animals **P < 0.01, *P < 0.05 using multiple-way ANOVA followed by Bonferroni post-tests. Data are the mean ± SEM. and 6 hours and 3 days after injury compared to the HBO group (3 hours-NAC + HBO group: 150.7 ± 35.8 pg/ ml, p = 0.046, L-NAME + HBO group: 359.5 ± 65.4 pg/ml, F(3,20) = 21.21, p = 0.0015; 6 hours-NAC + HBO group: 114.0 ± 26.0 pg/ml, p = 0.0085, L-NAME + HBO group: 343.4 ± 44.3 pg/ml, F(3,20) = 7.405, p = 0.12; 3 days-NAC + HBO group: 355.5 ± 90.6 pg/ml, p = 0.77, L-NAME + HBO group: 127.4 ± 16.5 pg/ml, F(3,20) = 4.144, p = 0.046) (Fig. 4C). The upregulation of bFGF by HBO was significantly suppressed in the NAC + HBO and L-NAME + HBO groups at 6 hours after injury compared to the HBO group (NAC + HBO group: 1356.3 ± 102.4 pg/ml, L-NAME + HBO group: 922 ± 326.9 pg/ml, F(3,20) = 4.471, p = 0.0015.) (Fig. 4D).

RoS and noS inhibition suppressed HBo-mediated angiogenesis and muscle healing. Tie2
and Ki67 double-positive cells were less frequently observed in the NAC + HBO, L-NAME + HBO, and NT groups than in the HBO group (Fig. 5A). The number of Tie2 and Ki67 double-positive cells in the NAC + HBO and L-NAME + HBO groups was equal to that in the NT group (NAC + HBO group: 3.3 ± 0.6 cells/HPF, p = 0.0015, L-NAME + HBO group: 5.6 ± 0.9 cells/HPF, F(3,16) = 22.97, p < 0.001) at 1 day after injury (Fig. 5B).
We then evaluated muscle regeneration in the inhibitor + HBO groups. We investigated embryonic myosin heavy chain (eMHC) and laminin double-positive fibers as regenerating muscle fibers. Fibers positive for laminin but not eMHC were considered mature muscle fibers. We calculated the ratio of the number of immature to that of mature fibers at 3, 5, 7, and 9 days after injury (Figs. 6A,B). In the HBO group, the ratio was significantly increased at 5 days and decreased at 7 days after injury (5 days-NT group: 0.265 ± 0.063 ratio/HPF, HBO group: 0.651 ± 0.050 ratio/HPF, F(1,8) = 22.99, p = 0.0016, 7 days-NT group: 0.690 ± 0.053 ratio/ HPF, HBO group: 0.402 ± 0.016 ratio/HPF, F(1,8) = 27.02, p = 0.0041) (Fig. 6B). However, the ratio of regenerating to mature muscle fibers in the inhibitor + HBO groups was similar to that in the NT group regardless of HBO induction , the final oxidized product of NO, was significantly increased at 3 and 6 hours after injury in the HBO group, n = 6. ***P < 0.001, **P < 0.01, *P < 0.05 using multiple-way ANOVA followed by Bonferroni post-tests. Data are the mean ± SEM. (B) The amount of HIF1α protein was significantly increased at 3 and 6 hours and 1 day after injury in the HBO group, and this effect was suppressed under administration of both inhibitors, n = 6. ***P < 0.001, using multiple-way ANOVA followed by Bonferroni post-tests. (C,D) With NAC or L-NAME administration, VEGF and bFGF upregulation induced by HBO was suppressed at 3 or 6 hours after injury, n = 6, added total 48 animals ***P < 0.001, **P < 0.01, *P < 0.05 using multiple-way ANOVA followed by Bonferroni post-tests. Data are the mean ± SEM. (2020) 10:2744 | https://doi.org/10.1038/s41598-020-59615-x www.nature.com/scientificreports www.nature.com/scientificreports/ (5 days-NAC + HBO group: 0.367 ± 0.049 ratio/HPF, p = 0.239, L-NAME + HBO group: 0.417 ± 0.081 ratio/ HPF, F(3,16) = 6.89, p = 0.179; 7 days-NAC + HBO group: 0.794 ± 0.102 ratio/HPF, p = 0.160, L-NAME + HBO group: 0.643 ± 0.090 ratio/HPF, F(3,16) = 15.66, p = 0.850) (Fig. 6C). The inhibitors diminished the effect of HBO on muscle regeneration. Next, to evaluate satellite cell proliferation, we counted Pax7 and Ki67 double-positive cells (representative of satellite and non-quiescent cells, respectively) at 3 hours and 1, 3, and 5 days after injury. The number of Pax7+/Ki67 + cells was increased in the HBO group at 3 hours, but did not increase at 3 hours after injury in the NAC + HBO and L-NAME + HBO groups compared to the HBO group  , and DAPI (blue) in injured muscle at 1 day after injury (arrows). Scale bar: 100 µm. One of the double-positive cells is enlarged in the upper right of each image. (B) Administration of NAC or L-NAME reduced the number of Tie2 and Ki67 double-positive cells at 1 day after injury in the HBO group. n = 5, added total 10 animals. **P < 0.01, using one-way ANOVA followed by Bonferroni post-tests. Data are the mean ± SEM. (C) Representative image of immature blood vessels positive for tomato lectin (red) and DAPI (blue), and mature vessels positive for tomato lectin, laminin (green), and DAPI in the injured muscle. Scale bar: 100 µm. One of the immature and/or mature vessels is enlarged on the upper right at each image. (D) The number of immature blood vessels in the NAC + HBO and LNAME + HBO groups was decreased at 3 days after injury. n = 5. (E) The number of mature blood vessels in the NAC + HBO and LNAME + HBO groups was decreased at 3, 5, and 7 days after injury. n = 5, added total 30 animals. **P < 0.01, *P < 0.05, using multiple-way ANOVA followed by Bonferroni post-tests. Data are the mean ± SEM. (2020) 10:2744 | https://doi.org/10.1038/s41598-020-59615-x www.nature.com/scientificreports www.nature.com/scientificreports/ We stained sections with H&E and evaluated the regenerating muscle fibers with a central nucleus by measuring their cross-sectional area (CSA) and counting them at 5 days after injury (Figs. 6E,F). In the HBO group, the CSA of the regenerating muscle fibers was higher than that in the NT, NAC + HBO, and L-NAME + HBO groups at 5 days after injury (NT group: 466.0 ± 17.3 μm 2 /HPF, HBO group: 635.8 ± 19.7 μm 2 /HPF, NAC + HBO group: 407.1 ± 15.9 μm 2 /HPF, LNAME + HBO group: 402.5 ± 16.0 μm 2 /HPF, F(3,16) = 20.7, p < 0.001) (Fig. 6E). The number of regenerating muscle fibers was significantly increased in the HBO group compared to the NT group. However, there was no significant increase in the NAC + HBO and L-NAME + HBO groups at 5 days after injury (NT group: 20.22 ± 2.2 fibers/HPF, HBO group: 34.6 ± 3.2 fibers/HPF, NAC + HBO group: 20.0 ± 2.4 fibers/HPF, p = 0.0031, LNAME + HBO group: 19.4 ± 1.5 fibers/HPF, F(3,16) = 9.261, p = 0.0021) (Fig. 6F).
Finally, we measured the muscle tensile isometric strength, twitch force, and tetanic force at 7 days after injury ( Fig. 7A-C). In the HBO group, the twitch force and tetanic force were significantly increased. The HBO-induced Figure 6. HBO-mediated muscle regeneration was also suppressed by NAC and L-NAME inhibition after muscle contusion injury. (A) Representative image of regenerating muscle fibers stained for eMHC and laminin at 1, 3, 5, 7, and 9 days after injury. Scale bar: 100 µm. (B) Relative number of eMHC-and laminin-doublepositive fibers in the HBO group was significantly increased at 5 days after injury, and decreased at 7 days, n = 5, using 20 animals. ***P < 0.001, **P < 0.01, using multiple-way ANOVA followed by Bonferroni posttests. Data are the mean ± SEM. (C) Under inhibition by NAC or L-NAME, relative number of eMHC-and laminin-double-positive fibers were almost equal to those in the NT group, regardless of whether HBO was performed, n = 5, ***P < 0.001, **P < 0.01, using multiple-way ANOVA followed by Bonferroni post-tests. (D) Under inhibition by NAC or L-NAME, the number of Pax7-and Ki67-double-positive cells was significantly suppressed compared to that in the HBO group, n = 5, *P < 0.05, using multiple-way ANOVA followed by Bonferroni post-tests. (E,F) Under inhibition by NAC or L-NAME, CSA and the number of regenerating muscle fibers in the HBO group were suppressed at 5 days after injury. n = 5. ***P < 0.001, using one-way ANOVA followed by Bonferroni post-tests. Data are the mean ± SEM.

Discussion
In this study, we demonstrated that HBO increased VEGF and bFGF levels within 6 hours after contusion injury. After HBO, the concentration of NO in the injured muscle was increased, and consequently HIF1α was stabilized. After these changes occurred, endothelial cells proliferated and blood vessel formation was accelerated, resulting in increased regeneration of muscle fibers and muscle strength. We also investigated whether the regenerative effects of HBO on angiogenesis and muscle regeneration could be attributed to ROS and NO by administering the inhibitors NAC and LNAME.
Skeletal muscle healing after injury is characterized by degeneration, inflammation, and regeneration stages 2,33 . Each of these stages entails a complex sequence of physiological and cellular events, which are considered to depend on an adequate blood supply for transport of cells and metabolites 7 . After severe muscle damage, the vasculature is largely destroyed 2,29 .
Angiogenesis generally occurs in response to expression of HIF1α and VEGF that starts within 6 hours after injury 7 , and upregulation of bFGF, HGF, and angiopoietin 2 from 1 day 6,34 . Endothelial cells are activated and proliferate at 3 days 35,36 , and the peak of angiogenesis with vessel formation occurs at 5-14 days 13,37 . These results in previous reports were consistent with what we observed here in the NT group. In contrast, in the HBO group, the increase of HIF1α, and VEGF production occurred at 3 hours, peak bFGF production was at 6 hours, peak endothelial cell activation was at 1 day, immature vessels were observed at 3 days, and mature vessel formation occurred at 5-7 days after injury. The number of regenerating muscle fibers was increased at 5-7 days, and muscle strength was improved at 7 days. However, HBO did not promote an increase of HGF and angiopoietin 2 Figure 7. HBO recovered muscle tensile strength, and this effects was diminished by NAC and L-NAME. (A) Twitch and tetanic forces were recorded in the NT group at 7 days after injury, and subsequently the ratios of injured leg (If) to non-injured leg forces (Nf) were calculated. (B,C) The ratio of twitch and tetanic forces in the injured leg relative to those in the non-injured leg was calculated at 7 days after injury. Administration of NAC or LNAME eliminated the increase in relative twitch force and tetanic force in the HBO group, n = 6 using 30 animals. ***P < 0.001, **P < 0.01, *P < 0.05, using one-way ANOVA followed by Bonferroni post-tests. Data are the mean ± SEM.
We also investigated the effects of ROS and NO inhibition on HBO-mediated muscle regeneration. ROS without NO are produced by mitochondria from nicotinamide adenine dinucleotide phosphate (NADPH) [17][18][19][20][21][22] . In contrast, NO is produced by NOS in endothelial cells and macrophages 38 . NAC is a precursor of glutathione with a thiol group maintained in a reduced state and good intracellular penetration properties; the thiol group has antioxidant effects for all ROS 39 . L-NAME inhibits all NOS non-selectively by competing with arginine for the active site of NOS, which produces NO 25,40 . As the half-life of NO is relatively short, we measured NO 3 − , the final oxidized product. In this study, we measured the time course of NO 3 − levels and found a significant increase of NO 3 − in the HBO group. Inhibition by NAC and L-NAME before HBO treatment diminished the increase of NO 3 − and the effects of HBO on HIF1α, VEGF, and bFGF levels as well as endothelial cell proliferation, angiogenesis, satellite cell proliferation, muscle fiber regeneration, and muscle tensile strength. The upregulation of VEGF by HBO was suppressed by both inhibitors, and the inhibitory effect was stronger for NAC than for L-NAME. This indicates that VEGF upregulation by HBO was associated with both NO and ROS. The inhibitory effects of NAC and L-NAME were similar for the other factors, suggesting that NOs play a key role in the angiogenesis and muscle regeneration promoted by HBO 38,41 . In contrast, when NAC and L-NAME were administered without performing HBO after contusion injury, angiogenesis and muscle regeneration were not inhibited (Supplement 2B, supplement 3, 4), except for the inhibition of bFGF expression by NAC (Supplement 2C). It has reported that the normal upregulation of bFGF is reduced under NAC inhibition after injury 42,43 . Thus, it is considered that the inhibitors themselves did not affect vessel formation and muscle regeneration during the recovery from muscle injury. Interestingly, at 3 days after injury in the relatively late phase, the VEGF level was significantly decreased in both the L-NAME + HBO group and the L-NAME + NT group. Although NAC inhibits only iNOS, L-NAME inhibits all NOS, including eNOS, nNOS, and iNOS, strongly and non-selectively 44,45 Inhibition of eNOS causes vascular endothelial cell disorders 46,47 . Thus, VEGF production may be decreased by endothelial cell damage caused by L-NAME-mediated eNOS inhibition.
In this study, we evaluated HIF1α, VEGF, bFGF, HGF, and angiopoietin as factors that promote angiogenesis in different ways. HIF1α stabilization stimulates production of VEGF 11,12,14 and activates the VEGF-R2 receptor in endothelial cells. bFGF activates proliferation of endothelial cells via activation of the FGF receptor 48 . HGF promotes mitogenic and morphogenic functions of endothelial cells via activation of c-Met/HGF receptor 49 . Angiopoietin 2 activates the Tie2 receptor in endothelial cells 50 .
Here, HBO increased VEGF and bFGF levels within 6 hours after injury but did not increase the levels of HGF and angiopoietin 2. The increase in VEGF levels during HBO is considered to have occurred through two mechanisms. First, HBO provides oxygen for NO production in injured tissue by increasing oxygen partial pressure gradients between healthy and hypoxic tissues 51 . Second, HBO promotes the stabilization of HIF1 by increasing levels of NO, which decreases prolyl hydroxylase (PHD) activity. The subsequent stabilization of HIF1 induces VEGF expression 14,52 . In our study, mRNA levels of HIF1α did not increase, but the amount of HIF1α protein increased at 3 and 6 hours and 1 day after contusion injury. Under normoxic conditions, the mRNA expression of HIF1α is maintained, but HIF1α protein levels are kept low due to rapid degradation by PHD 14 . HBO stabilizes HIF1α by decreasing PHD activity 11,12,14 . The increase in HIF1α protein levels without a concomitant increase in mRNA levels suggests that HBO stabilized HIF1α in the contused skeletal muscle in the early phase after injury. Stabilization of HIF1α results in an increase in VEGF levels 11,12,14,51 . NO stimulates bFGF production in endothelial cells, which accelerates the proliferation of endothelial cells 53 . In the present study, HBO increased NO levels, which likely increased VEGF levels via HIF1α stabilization and bFGF levels via NO in the injured muscle.
After endothelial cells are stimulated by VEGF, the laminin-containing basement membrane is destroyed, and the endothelial cells start to proliferate abnormally at 5 days after injury 35,36 . Subsequently, the laminin is re-formed, and the abnormal proliferation of blood vessel cells ceases. The blood vessels then become mature, and angiogenesis is completed at 7 to 14 days 13,32 . In the HBO group in the present study, the number of proliferating endothelial cells was increased at 1 day after injury, and then the number of immature vessels increased at 3 days and decreased from 5 days after injury. The number of immature and mature blood vessels was almost equal to that in the NT group at 9 days after injury. HBO thus accelerates angiogenesis by stimulating endothelial cells but does not interfere with the total number of regenerated mature vessels.
Additionally, in the present study, HBO accelerated muscle healing from the early phase, and L-NAME and NAC administration suppressed this effect. We measured muscle healing in the group that received only 1 HBO treatment at 3 days after injury, i.e., at a later time point. This group did not show any difference compared to the NT group. Thus, later induction of HBO seems to be not effective (Supplement 5). Thus, HBO appears to act at early time points and accelerate physiological angiogenesis and muscle regeneration.
There are several limitations in the present study. First, we did not use an inhibitor that inhibits ROS without inhibiting NO. There are several NAPDH inhibitors that are selective for ROS without NO. However, these inhibitors also inhibit MAPK and COX pathways, which are also involved in angiogenesis. Thus, these inhibitors could not be used to evaluate the effects of ROS in the present study because of the confounding effects on angiogenesis. Instead, we used NAC, which inhibits all ROS, and L-NAME, which inhibits only NO. Although the effect of ROS can thus be determined by subtracting the effect of L-NAME from that of NAC, the measurement is still indirect. Additionally, the muscle contusion injury model in this study may not accurately represent most occupation-and sports-related muscle injuries. Even though the model that we used is not the same as human injury, it is widely adopted as a skeletal muscle injury model to study the muscle-healing process [34][35][36]54 . Furthermore, we did not investigate the most effective treatment condition for angiogenesis and muscle regeneration. Further studies that evaluate the most effective treatment protocol with regard to oxygen dose, pressure, timing, and frequency are needed. (2020) 10:2744 | https://doi.org/10.1038/s41598-020-59615-x www.nature.com/scientificreports www.nature.com/scientificreports/ In conclusion, HBO increased NO, VEGF, and bFGF levels and stabilized HIF1α, and subsequently accelerated endothelial and satellite cell proliferation and promoted angiogenesis and muscle regeneration after skeletal muscle contusion injury. This injury model is considered to simulate clinical skeletal muscle injury in terms of effects on not only skeletal muscle but also blood vessels. By using this injury model, we were able to evaluate both muscle regeneration and angiogenesis. Administration of NAC and L-NAME before HBO diminished the effects of HBO. The upregulation of NO is crucial for the effects of HBO on the muscle contusion injury. These findings provide basic evidence to support the clinical indication of HBO for skeletal muscle contusion injury.

Materials and Methods
Animals and contusion model. All animal experiments were performed under approved protocols and in accordance with the recommendations of the Institutional Animal Care and Use Committee of Tokyo Medical and Dental University. There is no conflict of interests in terms of the affiliations of researchers. Male 10-week-old Wistar rats weighing 250-300 g were kept in standard cages under a constant temperature and light/dark cycle of 12 hours each, with the light/dark hours changing automatically, and given water and food (MF; Oriental Yeast, Tokyo, Japan) ad libitum throughout the experimental period. The experiments were conducted mainly during the day from 6:00 AM to 6:00 PM, and the total number of rats used was 445 animals. The animals were anesthetized with an intraperitoneal injection of chloral hydrate (280 mg/kg) or gas anesthesia (1.5% to 3% isoflurane, 1.5 L/min flow). Muscle contusion was induced by the modified mass-drop method 55 under gas anesthesia in the right calf muscle of the rats and subsequently in the left calf muscle to reduce the sample number of the rats. We used contralateral controls in the same animal to calculate the ratio of the strength of injured muscle to that of non-injured muscle when measuring muscle strength. First, the targeted hind limb was fixed distally at the Achilles' tendon and proximally at the gastrocnemius muscle using percutaneous needles. The calf was placed on silicon clay (Therapy Putty; AliMed, Massachusetts, US). A solid aluminum cylinder (640 g) was dropped from a height of 250 mm onto the impactor (diameter 10 mm, hemispherical surface), which was placed on the belly of the medial calf of the rat. Subsequently, rats were randomly assigned to either no treatment (NT) or HBO treatment (HBO) after muscle contusion. If pain was apparent in the contused rats, we applied an analgesic agent such as felbinac gel.
HBo protocol. Approximately 15 minutes after contusion injury, the rats in the HBO group were placed in a hyperbaric experimental chamber in which 100% oxygen was administered at 2.5 ATA pressure for 2 h, with 15 minutes for compression, 120 minutes of exposure at 2.5 ATA, and 15 minutes of decompression under 100% oxygen. The compression and decompression speeds were 0.1 ATA/min. In the clinic, HBO treatment is performed five times a week (except on weekends); thus, we also performed the same HBO protocol in this study, where HBO was performed once a day for 5 consecutive days (Fig. 1).
Histological evaluation. We sacrificed all rats by overdose of isoflurane at each time point. When collecting the tissues, we separated the gastrocnemius and soleus muscles and collected the gastrocnemius, and froze the samples in liquid nitrogen-cooled 2-methylbutane. The muscles were stored at −80 °C until further analysis. Transverse sections of the calf muscles were cut at 20 µm using a cryostat (CM 300; Leica Japan, Tokyo, Japan) at 2, 4, 6, 8, 10, and 12 mm proximal from the end of the Achilles' tendon, and stored at −30 °C. The sections were stained with hematoxylin and eosin (H&E). Images were obtained on a microscope (Olympus BX51; Olympus). In the evaluation of regenerated fibers, myofibers with centrally located nuclei were defined as regenerating fibers 31,32,56 . The injured area was determined as the cell-invaded area. We randomly selected 10 high-power fields (HPFs) from the injured area and counted the number of regenerating myofibers in the injured area. Moreover, we randomly selected 250 myofibers with regenerating myofibers and measured the cross-sectional areas (CSA) using the ImageJ software (ImageJ; National Institutes of Health, Bethesda, MD) (n = 5). immunohistochemical analysis. Transverse sections of the gastrocnemius muscle were immersed in blocking solution (5% normal goat serum in PBS with 0.5% Triton X-100) for 30 minutes at 1, 3, 5, 7 and 9 days after injury, and then incubated overnight at 4 °C with the primary antibody (Tie2, mouse  Finally, the sections were incubated for 1 min with DAPI (Life Technologies), washed with PBS, and mounted in mounting solution (PermaFluor; Thermo Fisher Scientific Japan, Yokohama, Japan). Positively stained cells were counted in 10 high-power fields (HPFs) using the histological procedures described above (n = 5).
ELISA and nitrite measurement. Gastrocnemius muscles were collected from the hind limbs before injury, 3 and 6 hours after injury, and 1 and 3 days after injury. The samples were trimmed, frozen, and crushed using a cell crusher. The samples were homogenized with 1,000 µl of lysis/extraction reagent (CelLytick; Sigma-Aldrich, St. Louis, Mo.) and centrifuged at 14,000 rpm for 10 minutes at 4 °C, and the supernatant was then extracted for assay. The concentration of VEGF, bFGF, HGF, angiopoietin2, and HIF1 was measured using an enzyme-linked immunoassay (ELISA) kit (HIF1, Cell Biolabs, San Diego, USA; VEGF/bFGF/HGF/angiopoietin 2, R&D Systems, Minneapolis, MN) according to the manufacturer's protocol (n = 6). The concentration of nitrite www.nature.com/scientificreports www.nature.com/scientificreports/ was measured using the OxiSelect Nitrite Assay Kit (Cell Biolabs, San Diego, USA) according to the provided protocol (n = 6).

Quantitative reverse transcription polymerase chain reaction (qRt-pcR). Gastrocnemius mus-
cles were snap-frozen in liquid nitrogen and stored at −80 °C at each time point after the injury. Total RNA including the RNA fraction was isolated by lysis in 1 ml of Qiazol Total RNA Isolation Reagent (Qiagen, Hilden, Germany) using the RNeasy Mini Kit (Qiagen). One microgram of total RNA was reverse transcribed using the PrimeScript RT reagent Kit (TAKARA, Tokyo, Japan). Quantitative PCR was performed using GoTaq qPCR Master Mix (Promega), a qPCR System (Mx3000P; Agilent Technologies), 20 ng of cDNA, and specific primers according to the manufacturer's instructions. The primer sequences were as follows: HIF1, 5′-GATTAGCCGAGTGCTCAGAATCAAG-3′ (forward) and 5′ = CAGGGCCAACCACTGTTTCATA-3′ (reverse); GAPDH, 5′-GGCACAGTCAAGGCTGAGAATG-3′ (forward) and 5′-ATGGTGGTGAAGACG CCAGTA-3′ (reverse). The expression of the target mRNA was normalized to that of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA. Relative quantification of gene expression was performed based on the comparative CT (threshold cycle value) method (ΔCT = CT gene of target -CT GAPDH gene). A comparison of gene expression in different samples was performed based on differences in the ΔCT of individual samples (ΔΔCT) (n = 4).

RoS and noS inhibitor administration.
The ROS inhibitor N-acetyl cysteine (NAC, Wako, Tokyo, Japan) and NOS inhibitor N-nitro-L-arginine methyl ester hydrochloride (L-NAME, Sigma Aldrich, St. Louis, MO, USA) were peritoneally injected daily from the day before injury. Subsequently, rats were randomly divided into no treatment (NAC + NT, LNAME + NT) or HBO treatment (NAC + HBO, LNAME + HBO) groups after muscle contusion. On the day of HBO treatment, the inhibitor was administered at a dose of 0.018 mg/g 30 minutes before HBO as previously described 57 (Fig. 1).

Measurement of muscle tension isometric strength.
Physiological testing was performed using a modified muscle tensile strength test at 7 days after contusion 58 . Rats were anesthetized intraperitoneally with chloral hydrate (280 mg/kg), placed on the platform in a prone position, and maintained at 37 °C using a heating pad. The Achilles tendon was sectioned 1 cm distal to the end of the muscle and sutured securely with 4-0 nylon using the Kessler method 59 . The exposed tendons and muscles were kept moist by periodically applying isotonic saline. A transducer (TB-653T; Nihon Koden, Tokyo, Japan), sensor interface (Power lab; AD Instruments Japan, Nagoya, Japan), and software (Power lab software; AD Instruments Japan) were used to measure the tension of the tendon. All data were displayed and stored on a computer using a custom-made program in LabView (National Instruments). General stimulation of the tibial nerve at 1 Hz (twitch) or 50 Hz (tetanus) was performed using surface electrodes (UL2-2020, Unique Medical Co., Ltd., Tokyo, Japan), and the maximum strength of Neurostimulus (Neuropack μ; Nihon Koden) was recorded 31,32 . The stimulation pulse was performed at the minimum voltage that caused the gastrocnemius muscle to reach its maximum contraction. The maximum crimp and tetanus isometric tensile strength of the injured (If) and non-injured (Nf) legs were measured. Before injury, there was no difference between the right and left sides (Supplement 1D). The strength ratio of the injured muscle to the non-injured muscle (ratio of If to Nf) was calculated (n = 6).
Statistics. Data are presented as the mean ± standard error of the mean (SEM). All analyses were two-sided, with a significance level of 5%. Statistical analysis was performed using Windows SPSS version 23.0 (IBM Japan, Ltd., Tokyo, Japan). Data were analyzed for normality using the Shapiro-Wilk test and for homogeneity using the Levine test. We investigated longitudinal effects, the effect of HBO treatment, and the interaction between longitudinal effects and treatment using two-way ANOVA, with the number of proliferating endothelial cells, immature and mature vessels, proliferating satellite cells, muscle fibers stained with eMHC and laminin, amount of nitrite, relative expression ratio, and ELISA data under inhibition as the outcome means. The number of proliferating endothelial cells, number of regenerating muscle fibers with H&E staining, muscle tensile strength under inhibition, and CSA of regenerating muscle fibers with H&E staining under inhibition were compared across five or six groups (NT, HBO, NAC + HBO, LNAME + HBO, NAC + NT, LNAME + NT or NT, HBO, NAC + HBO, LNAME + HBO, non-injured) by one-way ANOVA followed by Bonferroni post-tests based on the normal distribution and homoscedasticity of the data. The twitch and tetanic force of the right and left gastrocnemius muscle before injury, which had only a normal distribution and without homoscedasticity, were subjected to Welch's t-test for comparison between two groups.