CXCL12 mediates adipose tissue regeneration by recruiting ASCs via migrasomes

To elucidate the mechanism underlying soft tissue regeneration, we utilized our previously established fat grafting model with a donor site [23, 24] (Fig. 1A). This model optimizes the exploration on adipose regeneration by better mimicking the situation of clinical autologous fat grafting which contains fat harvesting (donor site) and fat grafting (recipient site). Adipose tissues at the donor and recipient sites in this model showed different repair and regeneration outcomes, with tissues of the donor site repaired better than that at the recipient site. And the mechanism behind this discrepant regeneration process has been proved to be ascribed to the different expression of CXCL12 post tissue damage [24].

Herein, in same mice model, hematoxylin and eosin (HE) analysis at 90 days after grafting showed that mature adipocytes were regularly arranged, small adipocytes (white arrows) were scattered throughout, and the tissue structure was stable at the donor site, while broken cells, vacuoles (asterisk), and severe tissue fibrosis were observed at the recipient site (Fig. 1B). Western blot analysis showed that protein expression of CXCL12 and CXCR4 was higher at the donor site than at the recipient site as early as Day3 after tissue injury (Fig. 1C-E). Quantitative analysis of immunofluorescence staining of CD34 showed that the number of CD34+ ASCs at the donor site was higher than that at the recipient site soon after injury, increased from Day3 to Day 14, and decreased thereafter. However, infiltration of ASCs was delayed at the recipient site. The number of CD34+ ASCs at the recipient site was lower than that at the donor site during the first 14 days, increased from Day 14 to Day 30, and decreased thereafter (Fig. 1F). Significantly more CD34+ ASCs infiltrated the recipient site than the donor site at the later stage of tissue repair, indicating that regeneration was incomplete. By contrast, there were fewer ASCs at the donor site and perilipin staining showed that the tissue structure was integrated (Fig. 1G).

To investigate whether migrasomes participate in tissue repair, expression of the migrasome markers TSPAN4 and TSPAN7 was tested in tissue samples from Day 3. Immunofluorescence staining showed that vesicles enriched with TSPAN4 and TSPAN7 were present at both the donor and recipient sites. Magnified images showed that some of these vesicles were attached to tubular structures that extended from cells, while others were scattered among cells (Fig. 1H). The diameters of these vesicles were mostly 0.5–3 μm, which is consistent with the characterizations of migrasomes [19,20,21] (Fig. 1I). Furthermore, there were significantly more migrasome-like vesicles at the donor site than at the recipient site soon after fat grafting (Fig. 1J).

These findings showed that the chemokine CXCL12 might mediate infiltration of ASCs during adipose tissue regeneration and that migrasomes might be involved in expression of CXCL12.

Figure 1. Recruitment of ASCs by CXCL12 mediates adipose tissue regeneration. (A) Schematic illustration of the novel mouse fat grafting model with a donor site. N = 7 in each groups. (B) HE staining of adipose tissue obtained from the donor and recipient sites 90 days after surgery. Scale bar, 50 μm. White arrows indicate small adipocytes and asterisks indicate vacuoles. (C) Western blot analysis of CXCL12 and CXCR4 protein expression in adipose tissue from the donor and recipient sites 3 days after surgery. (D-E) Semi-quantitative analysis of CXCL12 and CXCR4 protein expression. *p < 0.05 compared with Donor. (F) Quantitative analysis of infiltrated CD34+ ASCs per field at the donor and recipient site over time. *p < 0.05 compared with Donor. (G) Immunofluorescence staining of CD34 and perilipin in adipose tissue from the donor and recipient site at 60 days after surgery. Scale bar, 20 μm. (H) Immunofluorescence stanning of TSPAN4 and TSPAN7 in adipose tissue from donor and recipient sites at 3 days after surgery. Scale bar, 10 μm. (I) Diameters of the TSPAN4+ and TSPAN7+ vesicles. n = 50 vesicles per group. (J) Quantitative analysis of TSPAN4+ and TSPAN7+ vesicles per field in adipose tissue from the donor and recipient site at 3 days after surgery. *p < 0.05 compared with Donor. The data mean ± SEM. Statistical differences in (D), (E), (I) and (J) were assessed with Student’s t test, (F) was analyzed with One-way ANOVA.

CXCL12 regulates recruitment of ASCs during asymmetric adipose tissue regeneration

To further look into the mechanism behind the different expression of CXCL12 between the donor and recipient site, we sought to diminish the bias brought by physiological environment. Thus, we transferred the adipose regeneration model on the bilateral inguinal fat pads of mice. To further mimic the damage pattern of donor and recipient site, we manipulated the inguinal vessels considering the most prominent difference between the donor and the recipient site of fat grafting is that the vascular honeycomb structure is intact at aspirated sites after liposuction while grafts at recipient sites are avascular [26]. This asymmetric adipose regeneration model was well-established by the published work to evaluate the regeneration of adipose tissue and the activity of ASCs post tissue damage [27, 28]. To this end, the inguinal vessels were left intact on the right side to mimic the donor site (Normal) and cut with scissors on the left side to mimic the recipient site (Ischemic) (Fig. 2A).

As expected, macroscopic observation on Day30 showed that tissues had a natural appearance and soft texture in the Normal group, but appeared atrophic and brittle in the Ischemic group (Fig. 2B). Histological analysis showed that differently sized adipocytes were tightly arranged in the Normal group, whereas large vacuoles and severe fibrosis were observed in the Ischemic group (Fig. 2C). Immunofluorescence staining of CD34 showed that significantly more ASCs infiltrated tissues in the Normal group than in the Ischemic group as early as Day3 post injury, and the number of ASCs in the Normal group kept rising from Day3 to Day7 and decreased thereafter. However, infiltration of ASCs was delayed in the Ischemic group. The number of infiltrated ASCs in the Ischemic group remained relatively low during the first 14 days post-injury and increased sharply from Day 14 to Day 30 (Fig. 2D, E). Perilipin is only detectable in viable adipocytes. Staining of perilipin showed that the adipose tissue structure was relatively stable during the first 7 days post-injury in the Normal group, whereas the perilipin+ area sharply decreased in the Ischemic group. From Day 7 to Day 14, the perilipin+ area decreased in the Normal group but was stably small in the Ischemic group. From Day 14, tissue regeneration was observed in both groups, as shown by the increased perilipin+ area. Nevertheless, Nevertheless, the perilipin+ area was significantly larger, and the number of infiltrated ASCs was lower in the Normal group than in the Ischemic group on Day 30, suggesting that regeneration was better in the Normal group (Fig. 2D, F).

To further investigate the mechanism underlying the asymmetric regeneration between the Normal and Ischemic groups, we tested expression of CXCL12 and CXCR4. Quantitative PCR (qPCR) analysis showed that expression of CXCL12 and CXCR4 was higher in the Normal group than in the Ischemic group soon after injury. Expression of CXCL12 and CXCR4 in the Ischemic group began to increase on Day 7 and surpassed that in the Normal group after Day 14 (Additional file 1: Figure S1). Western blot analysis confirmed that expression of CXCL12 and CXCR4 was significantly higher in the Normal group than in the Ischemic group on Day 7, at which point infiltration of ASCs was evident (Fig. 2G-I). These data provide evidence that CXCL12 induced early infiltration of ASCs via CXCR4 which mediated the asymmetric regeneration of adipose tissue.

Migrasomes are detected during asymmetric adipose tissue regeneration

Migrating cells aggregate tetraspanin proteins, especially TSPAN4, to form a bulbous membrane on the swelling domains of retraction fibers and finally form migrasomes [29]. Migrasomes can integrate spatial and biochemical information, by which cells are recruited to specific location to exert the function of organ development or tissue regeneration [19,20,21]. Thus, if migrasomes indeed mediated the establishment of CXCL12 signal in adipose tissue regeneration, the number of migrasomes should correlated with the expression of CXCL12 and recruited ASCs as well as regeneration outcomes between the two sides of adipose repair model. To verify if migrasomes play a role in adipose tissue regeneration, the presence of TSPAN4+ and TSPAN7+ vesicles was tested in the asymmetric regeneration model. Immunofluorescence staining of tissues detected vesicles enriched with TSPAN4 and TSPAN7 in the extracellular spaces or attached to fibers that projected from cells (Fig. 3A). The average diameter of these vesicles was ∼ 2 μm, which is consistent with the diameter of migrasomes (Fig. 3B). The number of migrasome-like vesicles in the Normal group was higher than that in the Ischemic group as early as Day 1 post-injury, decreased from Day 1 to Day 3, sharply increased until Day 7, and decreased thereafter. However, changes in the number of migrasome-like vesicles were delayed in the Ischemic group. The number of migrasome-like vesicles in the Ischemic group was decreased at Day 5, increased from Day 7, and exceeded that in the Normal group after Day 14 (Fig. 3C).

Fig. 3

Migrasomes are detected during adipose tissue regeneration. (A) Immunofluorescence staining of TSPAN4 and TSPAN7 in adipose tissue from the Normal and Ischemic groups over time. Scale bar, 10 μm. (B) Diameters of the TSPAN4+ and TSPAN7+ vesicles. n = 50 vesicles per group. (C) Quantitative analysis of TSPAN4+ and TSPAN7+ vesicles per field in adipose tissue from the Normal and Ischemic group over time points. *p < 0.05, **p < 0.01 compared with Normal. (D) Immunofluorescent staining of integrin β1 in adipose tissue from Normal and Ischemic groups over time. Scale bar, 10 μm. (E) Diameters of integrin β1+ vesicles. n = 50 vesicles per group. (F) Quantitative analysis of integrin β1+ vesicles per field in adipose tissue from the Normal and Ischemic groups over time. *p < 0.05, **p < 0.01 compared with Normal. (G) Western blot analysis of TSPAN4, TSPAN7 and integrin β1 protein expression in adipose tissue from Normal and Ischemic groups at 7 days after surgery. (H-J) Semiquantitative analysis of TSPAN4, TSPAN7, and integrin β1 protein expression. *p < 0.05 compared with Normal. The data are mean ± SEM. Statistical differences in (B), (E), (H), (I) and (J) were analyzed using Student’s t test. (C) and (F) were assessed by One-way ANOVA

Enrichment of integrin on the bottom of migrasomes enables their tethering to the extracellular matrix so that they do not move away with cell migration and establish a localized signal along the migrating pathway of cells [30]. To further identify the migrasome-like vesicles, integrin β1 was stained. Vesicles enriched with integrin β1 were observed in the extracellular space, and some of these vesicles were attached to cell projections (Fig. 3D). Integrin β1-enriched vesicles had almost identical diameter distributions as TSPAN4+ and TSPAN7+ vesicles, and shared a similar exhibition pattern (Fig. 3E, F).

We further tested expression of migrasome markers in tissues. qPCR analysis showed that expression of TSPAN4, TSPAN7, and integrin β1 was elevated in the Normal group during the first 7 days post injury, but began to increase from Day 7 in the Ischemic group and was higher in the Ischemic group than in the Normal group after Day 14 (Additional file 2: Figure S2). Western blot analysis confirmed that higher protein expression of TSPAN4, TSPAN7, and integrin β1 was significantly higher in the Normal group than in the Ischemic group on Day 7 (Fig. 3G-J).

Hence, we detected the presence of migrasomes during adipose tissue regeneration. As expected, the pattern of migrasomes resembled that of CXCL12 expression in adipose tissue after injury.

ASCs generate migrasomes

We noticed that the pattern of migrasomes resembled that of ASCs in the Normal and Ischemic groups (Fig. 2), with infiltration of migrasomes and ASCs delayed in the Ischemic group compared the Normal group. ASCs have been documented to secret CXCL12 under inflammatory conditions [31]. We thus postulated that ASCs generate migrasomes. Fluorescently tagged wheat-germ agglutinin (WGA) is a probe for rapid detection of migrasomes in living cells [32]. We first observed cultured ASCs stained with WGA-Texas Red. Confocal microscopy showed that numerous vesicles enriched with WGA with a diameter of ∼ 2 μm were scattered around ASCs or attached to projections that extended from these cells (Fig. 4A). These vesicles were oval shaped, with diameter that fits the definition of migarsomes and clustered on one side of the cells, proving these vesicles are migrasomes. Live cell time-lapse confocal microscopy also revealed the growth of migrasomes on the tips of or along the retraction fibers generated by ASCs (Additional file 3: Figure S3; Supplementary Video 1). Scanning electron microscopy showed that ASCs generated membrane-bound vesicles with diameters of 0.5–3 μm that were attached to the tips or intersections of retraction fibers, a typical characteristic of migrasomes (Fig. 4B). Transmission electron microscopy (TEM) confirmed that the vesicles were connected or in close proximity to retraction fibers of ASCs. Most of these vesicles were single-membraned, oval, with the diameter of 0.5–3 μm and contained numerous small vesicle (Fig. 4C), which is agreed with the definition of migrasomes. Thus, we showed that ASCs can generate migrasomes.

Fig. 4

ASCs generate migrasomes. (A) Confocal microscopy images of cultured ASCs stained with WGA-Texas Red. Scale bar, 10 μm. White arrows indicated vesicles enriched with WGA that are highly resembled migrasomes, with diameter of ∼ 2 μm and were scattered around ASCs or attached to projections that extended from these cells. (B) Scanning electron microscopy images of cultured ASCs. Scale bar, 100 μm. (C) TEM images of ASCs cluster. Scale bar, 2 μm. (D) Schematic illustration of the centrifugation procedures used to isolate migrasomes. (E) TEM image of isolated migrasomes. Scale bar, 1000 nm. (F) Western blot analysis of TSPAN4, integrin β1, actin and CXCL12 protein expression in isolated migrasomes. (G) Western blot analysis of TSPAN4, integrin β1 and CXCL12 protein expression in isolated migrasomes from ASCs or ASCs treated with siRNA. (H-J) Semiquantitative analysis of TSPAN4, Integrin β1, and CXCL12 protein expression. *p < 0.05. The data are mean ± SEM. Statistical differences were analyzed using One-way ANOVA followed by Bonferroni posttest

To confirm the role of ASC-derived migrasomes in adipose tissue regeneration, we first isolated migrasomes from cultured ASCs (Fig. 4D). The isolated migrasomes were analyzed by TEM, which showed a characteristic morphological feature identical to those previously reported, with round shape, attachment to retraction fibers and containing luminal vesicles [19,20,21, 25, 33] (Fig. 4E). Moreover, the preparations were highly enriched with the migrasome markers TSPAN4 and integrin β1, while expressed lower levels of actin compared with cell body (Fig. 4E). Studies have documented the enrichment of CXCL12 within migrasomes [19,20,21]. Consistently, migrasomes isolated from ASCs contained significantly higher levels of CXCL12 than the cell bodies (Fig. 4F). In summary, we showed that ASCs can generate migrasomes enriched with CXCL12.

To confirmed the enrichment of CXCL12 within migrasomes and the increased CXCL12 protein expression of migrasomes was indeed mediated by the transferring of CXCL12 from ASCs, we ought to deplete the endogenous CXCL12 expression of ASCs. Using siRNA significantly decreased the protein expression of CXCL12 in ASCs (Additional file 4: Figure S4). Isolated migrasomes from both ASCs and siRNA treated ASCs expressed higher level of migrasome marker TSPAN4 and Integrin β1 compared with cell body (Fig. 4G-I). Migrasomes contained significantly higher level of CXCL12 compared with cell body whereas depleting the expression of CXCL12 within ASCs significantly decreased the CXCL12 level within migrasomes, substantiating the direct relationship between migrasomes and CXCL12 as well as between migrasomes and ASCs (Fig. 4G-J). Furthermore, immunofluorescent stained ASCs with WGA and CXCL12 showed the colocalized enrichment of CXCL12 with WGA, which directly revealed the delivery of CXCL12 by adipose-derived stem cells via migrasomes (Additional file 5: Figure S5).

Migrasomes promote ASC recruitment and tissue regeneration via CXCR4/RhoA

Next, we investigated the role of ASC-derived migrasomes in adipose tissue regeneration by adding isolated migrasomes to poorly vascularized adipose tissue post-injury. Migrasomes (+ Mig group) or phosphate-buffered saline (PBS, PBS group) were focally injected into inguinal fat pads (Ischemic repair model) subjected to the same procedure as the Ischemic group for six consecutive days post-injury (Fig. 5A). Migrasomes, including stem cell-derived migrasomes, are enriched with CXCL12 [19, 21, 22]. Consistently, western blot analysis showed that expression of CXCL12 was significantly higher in the + Mig group than in the PBS group at Day 7 after injury (Fig. 5B, C). Macroscopic observation on Day 30 showed that tissues in the + Mig group had a more natural appearance and soft texture while those in the PBS group appeared atrophic and felt rigid (Fig. 5D). HE staining of adipose tissues on Day 30 indicated that adipose tissue regeneration was better in the + Mig group than in the PBS group. Large vacuoles, oil cyst-like structure, and severe fibrosis were observed in the PBS group, whereas regularly arranged round adipocytes of different sizes were observed in the + Mig group (Fig. 5E).

Fig. 5

Migrasomes promote adipose tissue regeneration. (A) Schematic illustration of injection of poorly vascularized Ischemic adipose tissue with migrasomes (+ Mig group) or PBS (PBS group). N = 7 in each groups. (B) Western blot analysis of CXCL12 protein expression in adipose tissue from the + Mig and PBS groups at 7 days after surgery. (C) Semiquantitative analysis of CXCL12 protein expression. *p < 0.05 compared with PBS group. (D) Macroscopic image of adipose tissue from the + Mig and PBS groups at 30 days after surgery. Scale bar, 50 mm. (E) HE staining of adipose tissue from the + Mig and PBS groups at 30 days after surgery. Scale bar, 200 μm. (F) Immunofluorescence staining of CD34 and perilipin in adipose tissue from the + Mig and PBS groups at 30 days after surgery. Scale bar, 20 μm. (G) Quantitative analysis of infiltrated CD34+ ASCs per field in adipose tissue from the + Mig and PBS groups over time. (H) Quantitative analysis of the perilipin+ area per field in adipose tissue from the + Mig and PBS groups over time. *p < 0.05 compared with PBS group. (I) Western blot analysis of CXCR4, RhoA, and PPARγ protein expression in adipose tissue from the + Mig and PBS groups 7 days after surgery. (J-L) Semiquantitative analysis of CXCR4, RhoA, and PPARγ protein expression. *p < 0.05 compared with PBS group. The data are mean ± SEM. Statistical differences in (C) were tested by nonparametric Mann-Whitney test. (G), (H) and (J) to (L) were analyzed by One-way ANOVA

To verify the mechanism by which migrasomes promote tissue regeneration, immunofluorescence staining of CD34 and perilipin was performed. Quantification of CD34+ cells showed that migrasomes promote the infiltration of ASCs as early as Day 3 post-injury. The number of ASCs sharply increased from Day 3 to Day 14 in the + Mig group but remained relatively low in the PBS group. The number of ASCs decreased in the + Mig group but increased in the PBS group from Day 14 to Day 30, indicating ASCs are required for regenerative events in PBS group (Fig. 5F, G). The perilipin+ area remained low in both groups in the first 7 days. The perilipin+ area in the + Mig group increased from Day 7 to Day 30. By contrast, the perilipin+ area in the PBS group dropped from Day 7 to Day 14 and slightly increased to Day 30 (Fig. 5F, H). Tissue structures remained broken with significantly more infiltrated ASCs in the PBS group, while more complete tissue structures with fewer ASCs were observed in the + Mig group (Fig. 5F). In summary, we showed that migrasomes can promote adipose tissue regeneration by facilitating early recruitment of ASCs.

CXCL12 binds to its receptor CXCR4 and activates RhoA through activation of the small G proteins, Gi and Gα13, which leads to directional cell migration [34, 35]. Western blot analysis of tissues on Day 7 confirmed that protein expression of CXCR4 and RhoA was significantly higher in the + Mig group than in the PBS group (Fig. 5I-K), indicating that migrasomes activate CXCR4/RhoA signaling. Furthermore, addition of migrasomes promoted expression of the adipogenesis-associated protein PPAR-γ (Fig. 5I, L). These data show that migrasomes enriched with CXCL12 promote adipose tissue regeneration by recruiting ASCs through CXCR4/RhoA signaling. The recruited ASCs augment tissue regeneration probably by promoting adipogenesis as shown by elevated expression of PPAR-γ.

Blockade of ASC infiltration reduces the number of migrasomes

To prove the origin of migrasomes in vivo, we blocked infiltration of ASCs using the CXCR4 inhibitor AMD3100 [24]. AMD3100 (+ AMD group) or PBS (PBS group) was focally injected to the inguinal fat pads subjected to the same procedure as the Normal group (Fig. 6A). The in vivo use of AMD3100 prevent the infiltration of ASCs, which worsen the regeneration of adipose tissue. Macroscopic observation on Day 30 post operation showed adipose tissue from + AMD group with decreased tissue mass and atrophy appearance and brittle texture while that from PBS group with natural appearance and soft texture (Fig. 6B). HE staining showed tissue structure on Day 30 from PBS group was stable, with round adipocytes arranged tightly and regularly whereas large vacuoles, severe cell infiltration with disorganized structure could be witnessed from adipose tissue of + AMD group (Fig. 6C). Furthermore, western blot analysis of tissue from Day 7 showed expression of CXCL12 was significantly higher in the PBS group than in the + AMD group (Fig. 6D-E). Quantitative analysis of CD34+ cells confirmed that AMD3100 blocked infiltration of ASCs, with significantly lower levels of CD34+ cells in the + AMD group than in the PBS group over the first 7 days (Fig. 6B, C). Immunofluorescence staining of TSPAN4 and TSPAN7 confirmed that blockade of ASC infiltrations significantly decreased the number of migrasomes during tissue regeneration (Fig. 6D, E). These data provide evidence that ASCs can generate migrasomes enriched with CXCL12 to promote ASC infiltration during adipose tissue regeneration.

Fig. 6

ASCs generate migrasomes in vivo. (A) Schematic illustration of inhibition of ASC infiltration in Normal adipose tissue using CXCR4 inhibitor AMD3100 (+ AMD). N = 7 in each groups. (B) Macroscopic image of adipose tissue from the PBS and + AMD groups at 30 days after surgery. Scale bar, 50 mm. (C) HE staining of adipose tissue from the PBS and + AMD groups at 30 days after surgery. Scale bar, 200 μm. (D) Western blot analysis of CXCL12 protein expression in adipose tissue from the PBS and + AMD groups at 7 days after surgery. (E) Semiquantitative analysis of CXCL12 protein expression. *p < 0.05 compared with PBS group. (F) Immunofluorescence staining of CD34 and perilipin in adipose tissue from the + AMD and PBS groups at 7 days after surgery. Scale bar, 20 μm. (G) Quantitative analysis of infiltrated CD34+ ASCs per field in adipose tissue from the + AMD and PBS groups over time. *p < 0.05 compared with PBS. (H) Immunofluorescence staining of TSPAN4 and TSPAN7 in adipose tissue from the + AMD and PBS groups at 7 days after surgery. Scale bar, 10 μm. (I) Quantitative analysis of migrasomes per field in adipose tissue from the + AMD and PBS groups over time. *p < 0.05, **p < 0.01 compared with PBS. The data are mean ± SEM. Statistical differences were analyzed using One-way ANOVA

Migrasomes enhance migration of ASCs via CXCR4/RhoA in vitro

To verify the capacity of migrasomes to promote migration of ASCs, we performed transwell assay with migrasomes, AMD3100, and migrasomes plus AMD3100 (migrasome + AMD3100 group) (Fig. 7A). Migration capacity of ASCs was enhanced by migrasomes, diminished by AMD3100, and restored in the migrasome + AMD3100 group (Fig. 7B, C). Western blot and qPCR analyses of cultured cells showed that expression of CXCR4 and RhoA was upregulated by migrasomes and inhibited by AMD3100 (Fig. 7D-F, Additional file 6: Figure S6). Expression of CXCR4 and RhoA was partially restored in the migrasome + AMD3100 group, suggesting that migrasomes promote migration of ASCs via CXCR4/RhoA signaling.

Fig. 7

Migrasomes promote migration of ASCs in vitro via activation of CXCR4/RhoA signaling by CXCL12. (A) Schematic illustration of the transwell assay of ASCs with the control, migrasomes, AMD3100, and Migrasome + AMD3100 groups. Data are obtained from 3 independent reduplicated experiments. (B) Images of migrated ASCs after incubation for 12 and 24 h. Scale bar, 200 μm. (C) Quantitative analysis of the number of migrated ASCs per field. *p < 0.05 compared with control. (D) Western blot analysis of CXCR4 and RhoA protein expression. (E-F) Semiquantitative analysis of CXCR4 and RhoA protein expression. *p < 0.05 compared with control. (G) Schematic illustration of the transwell assay of ASCs with the control, migrasomes, CCG-1423 or Migrasomes + CCG-1423 groups. Data are obtained from 3 independent reduplicated experiments. (H) Images of migrated ASCs after incubation for 12 and 24 h. Scale bar, 200 μm. (I) Quantitative analysis of the number of migrated ASCs per field. *p < 0.05 compared with control. (J) Western blot analysis of CXCR4 and RhoA protein expression. (K-L) Semiquantitative analysis of of CXCR4 and RhoA protein expression. *p < 0.05 compared with control. The data are mean ± SEM. Statistical differences in (C) was assessed by Kruskal-Wallis. (E), (F), (I), (K) and (L) were analyzed using One-way ANOVA followed by Bonferroni posttest

The transwell assay was performed with control, migrasomes, CCG-1423 (a RhoA inhibitor), and migrasomes plus CCG-1423 (migrasome + CCG-1423 group) (Fig. 7G). Migration of ASCs was promoted by migrasomes, diminished by CCG-1423, and restored in the migrasome + CCG-1423 group (Fig. 7H, I). Western blot and qPCR analyses showed that migrasomes increased expression of RhoA, while CCG-1423 inhibited expression of RhoA and had no influence on expression of CXCR4 (Fig. 7J-L, Additional file 7: Figure S7). Expression of RhoA was partially rescued while expression of CXCR4 was not significantly altered in the migrasome + CCG-1423 group (Fig. 7J-L, Figure S7). In summary, we demonstrated that migrasomes enriched with CXCL12 promote migration of ASCs in vitro via CXCR4/RhoA signaling.