Preparation and identification of GQDs/Cy5-miR as a fluorescent switch

We first conducted predictive analysis on the target mRNA of miR-193a-3p. 29 genes were predicted to be the target of miR-193a-3p according to database tools, including PicTar, miRDB, TargetScan (Additional file 1: Fig. S1A). To screen this prediction, we then used TCGA and GTEx databases to compare the expression of all 29 miR-193a-3p-targeting mRNAs in GC tissues versus healthy tissues. The results indicated 18 genes among them are highly expressed in GC tissues (i.e., tumor expression level > normal tissue expression level, or T>N) (Additional file 1: Fig. S1B). Then RNA sequencing gene set (GSE77229) was obtained from GEO database. MiR-193a-3p mimics were transfected into SW900 cells (a cell line with deletions of the miR-193a locus) and the transcript abundance was analyzed by RNA sequencing. The downregulated transcripts were next analyzed for KEGG pathways enrichment. As a result, the repressed transcripts were mostly localized in the cell-cycle category and related genes were listed (Additional file 1: Table S1). The predicted target genes with T>N and the cell-cycle gene set were finally intersected to obtain 3-monooxygenase/tryptophan 5-monooxygenase activating protein zeta (YWHAZ) and cyclin D1 (CCND1) (Additional file 1: Fig. S1C). We next examined the gene expression of YWHAZ and CCND1 in human GC cells lines including HGC-27, MKN-28, AGS and SNU-1, and the normal human gastric mucosal epithelial cell line (GES-1) by qRT-PCR. The results displayed the high gene expression of both YWHAZ and CCND1 in the HGC-27 cells line, where CCND1 was most significantly expressed (Additional file 1: Fig. S1D). The combination of miR-193a-3p and its target gene CCND1 was based on the complementary base pairing in seed region (red) (Additional file 1: Fig. S1E). Therefore, through this examination, we selected CCND1 as the target gene of miR-193a-3p in the HGC-27 cells.

We next fabricated the Cy5/miR nanocomplex for miRNA delivery. Firstly, the size of commercially available GQDs was analyzed by high-resolution transmission electron microscopy (HR-TEM). The results showed that the diameter of GQDs was 5 nm, and the lattice fringe was 0.21 nm (Additional file 1: Fig. S2A). The absorbance spectra of GQDs reached a peak at ~ 340 nm (Additional file 1: Fig. S2B) and upon excitation at 360 nm (determined by excitation spectra as the maximal excitation wavelength), the emission peak occurred at 450 nm (Additional file 1: Fig. S2C), in line with the photoluminescent properties of GQDs. Secondly, miR-193a-3p was chemically tagged with either FAM or Cy5 fluorescent dyes. The fluorescent intensity of GQDs (λex/λem = 360/450 nm), Cy5-miR (λex/λem = 645/675 nm) or FAM-miR (λex/λem = 495/520 nm) was individually measured in a series of working solutions, validating the linear range of their fluorescent emission intensity over concentrations (Additional file 1: Fig. S2D–F).

We then evaluated the fluorescence quenching of Cy5-miR in the presence of various concentrations of GQDs at 37 °C. As the aromatic moiety of Cy5 fluorescent tag forms π–π stacking with the planar graphene sheet, the decreasing fluorescence of Cy5 was observed when the constant Cy5-miR was incubated with increasing GQDs in the Tris-HCl buffer (Fig. 2A). As a result, a maximum of (85.0 ± 0.4)% Cy5 fluorescence was quenched when adequate GQDs was added. Thus, the GQDs/Cy5-miR nanocomplexes were prepared with an optimized mass ratio of GQDs:Cy5-miR = 30:1 (Fig. 2B). Then, the complementary target sequence in CCND1 (CTS-CCND1) was added into the solutions. Resultantly, the Cy5 fluorescence was gradually recovered with the increasing concentration of CTS-CCND1, and a final recovery of (82.4 ± 4.1)% Cy5 fluorescence (the ratio of the original Cy5 fluorescence) was attained (Fig. 2C). The optimized molar ratio of required CTS-CCND1 over Cy5-miR was 6:1 (Fig. 2D). By switching the fluorescent moiety from Cy5 to FAM, the fluorescence of FAM-miR showed a resemblance of quenching when adding GQDs due to the increasing π–π stacking and the minimized fluorescence was detected when the mass ratio of GQDs: FAM-miR = 100:1 (Additional file 1: Fig. S3A, B). Similarly, FAM fluorescence was gradually recovered when CTS-CCND1 was further added, and this fluorescence remained unchanged as the molar ratio of CTS-CCND1 over FAM-miR reached 6:1 and above (Additional file 1: Fig. S3C, D).

Fig. 2

The interactions between the Cy5-tagged miRNAs and GQDs, together with physicochemical properties of the formed nanocomplexes. A Fluorescent intensity of Cy5-miR (100 nM) in the presence of varying concentrations of GQDs (1, 2, 4, 6, 8, 10, 20, 30 and 40 µg/mL). B The fluorescent quenching efficiency of Cy5-miR (100 nM) and GQDs was calculated. F Fluorescent intensity of the Cy5-miR interacted with varying concentrations of GQDs, F0 fluorescent intensity of Cy5-miR without GQDs. C Fluorescent intensity of GQDs/Cy5-miR in the solution after incubating with different concentrations of CTS-CCND1 (200, 300, 400, 500, 600 and 700 nM). D Fluorescent recovery efficiency calculated to determine the minimum ratio of CTS-CCND1 for miR-193a-3p release in GQDs/Cy5-miR. E Fluorescence intensity of GQDs/Cy5-miR under different conditions: with no target, CTS-CCND1 or scramble sequence. F Zeta potentials of GQDs and GQDs/Cy5-miR were measured. G AFM images and topographic height histograms of GQDs and GQDs/Cy5-miR

To investigate whether GQDs/Cy5-miR or GQDs/FAM-miR can bind CTS-CCND1 specifically, a noncomplementary oligonucleotide strand with random nucleotide sequence was synthesized and described as scramble sequence. GQDs/Cy5-miR or GQDs/FAM-miR were incubated in aqueous solutions with scramble sequence or CTS-CCND1 to confirm their sequence specificity. The Cy5 fluorescence of GQDs/Cy5-miR was enhanced 14.7-folds after incubation with CTS-CCND1, while the fluorescence went through minor change in the presence of scramble RNA. This result suggested that Cy5-miR in the GQDs/Cy5-miR complex failed to bind the scramble sequence, but it recognized and bound CTS-CCND1 with high specificity (Fig. 2E). The similar experiments using GQDs/FAM-miR were also performed, showing its specific recognition and binding of CTS-CCND1 (Additional file 1: Fig. S4). In addition, zeta potentials of all materials in solutions (pH = 7.5) were examined with results shown in Fig. 2F. GQDs demonstrated a positive surface charge of (25.0 ± 0.3) mV, and Cy5-miR showed a negative charge of (− 9.0 ± 0.3) mV, while the particles of GQDs/Cy5-miR owned an overall surface charge of (11.3 ± 0.2) mV. Concurrently, the morphology of GQDs and GQDs/Cy5-miR nanocomplex was examined by atomic force microscope (AFM). GQDs exhibited a typical nano-sized lamella structure with an average height of ~ 2 nm. In contrast, GQDs/Cy5-miR had distinctly increased thickness of ~ 4 nm (Fig. 2G).

Fabrication, characterization and stability of GQDs/Cy5-miR@sEVs

Next, we sought to encapsulate GQDs/Cy5-miR within sEVs to fabricate the dual-functional nanoplatform to induce and visualize their antitumor effects. We first isolated hucMSCs from healthy donors and verified their biological characteristics. Using flow cytometric analysis, we confirmed that the obtained MSCs expressed CD90, CD105, CD73, and CD44, but not CD34 or CD45 (Additional file 1: Fig. S5A). Further analyses using Alizarin Red and Oil Red O staining indicated that these MSCs demonstrated osteogenic and adipogenic differentiation capabilities (Additional file 1: Fig. S5B).

Following the isolation and purification of hucMSCs-derived sEVs, we then optimized the technical parameters to load GQDs/Cy5-miR into sEVs via sonication. Firstly, we examined the morphological structure of sEVs by transmission electron microscopy (TEM) and measured their particle size distribution detected by nanoparticle tracking analysis (NTA) after sEVs were sonicated under various output powers. TEM images showed that the lipid bilayer membrane of sEVs treated with 320 joules sonication output power was damaged, further confirmed by NTA results suggestive of membrane structure cracks in sEVs. By lowering the sonication output power to 250 joules, the treated sEVs remained intact as observed by TEM, and their particle size (peaked at ~ 150 nm) showed no significant difference from untreated sEVs (Additional file 1: Fig. S6A, B). Therefore, we chose 250 joules as the suitable sonication output power to load GQDs/Cy5-miR into sEVs.

Concurrently, the morphological structures of the untreated sEVs and GQDs/Cy5-miR@sEVs were revealed by TEM images (Fig. 3A). The particle sizes of untreated sEVs and GQDs/Cy5-miR@sEVs were determined to be 139.1 ± 19.0 nm and 149.7 ± 3.7 nm (mean ± SD) by NTA, respectively (Fig. 3B). The Western blotting results confirmed the expression of positive biomarker CD9, CD63, and Alix in sEVs, GQDs/Cy5-miR@sEVs and hucMSCs, with calnexin absent in sEVs and GQDs/Cy5-miR@sEVs as the negative biomarker (Fig. 3C). The miRNA contents of GQDs/Cy5-miR@sEVs were further analyzed using qRT-PCR to calculate the drug loading capacity (LC = the ratio of EV-encapsulated amount to the initial amount of cargo). The results showed that the loading capacity for miR-193a-3p was 5.6 times higher in GQDs/Cy5-miR@sEVs than the untreated sEVs (Fig. 3D), which corroborated the substantial inclusion of exogenous miR-193a-3p into GQDs/Cy5-miR@sEVs.

Fig. 3

Characterization and drug loading capacity of GQDs/Cy5-miR@sEVs. A TEM images of sEVs (upper panel) and GQDs/Cy5-miR@sEVs (lower panel). Scale bar = 200 nm. B Size distribution of sEVs and GQDs/Cy5-miR@sEVs. C Expression of Alix, CD63, CD9, Calnexin in sEVs, GQDs/Cy5-miR@sEVs and HucMSCs was detected by Western blot. D MiR-193a-3p in GQDs/Cy5-miR@sEVs and sEVs were detected by qRT-PCR. E The degradation of Cy5-miR, GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs in RNase solutions was detected by agarose gel electrophoresis. F Stability of GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs in PBS in 50 min at room temperature. G Stability of GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs in 50 min after being incubated within mouse serum at 37 °C. Data are expressed as mean ± SD and analyzed by one-way ANOVA. *p < 0.05, ***p < 0.001, ****p < 0.0001

We then evaluated the stability of miRNA in the forms of Cy5-miR, GQDs/Cy5-miR, and GQDs/Cy5-miR@sEVs after RNases treatment. Agarose gel electrophoresis experiments displayed that miRNA contents in Cy5-miR and GQDs/Cy5-miR were degraded by RNases. For the condition containing GQDs/Cy5-miR@sEVs, miRNA band was observed, indicating that the miRNA could be largely protected from RNase degradation once encapsulated inside GQDs/Cy5-miR@sEVs (Fig. 3E).

To assess the stability of the GQDs/Cy5-miR@sEVs in PBS, GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs were added to PBS for ~ 1 h and degradation kinetics of Cy5-miR from GQDs/Cy5-miR@sEVs were evaluated by fluorescence detection of Cy5-miR. As a result, 3.9% fluorescence of GQDs/Cy5-miR@sEVs decayed in 50 min, while 6.9% fluorescence of GQDs/Cy5-miR went off during the same time (Fig. 3F). Next, GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs were added to mouse serum and incubated at 37 °C for ~ 1 h. The stability of GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs in mouse serum was assessed by monitoring Cy5 fluorescence tracer. The results showed that 65.1% fluorescence of GQDs/Cy5-miR decayed in serum while only 12.6% fluorescence of GQDs/Cy5-miR@sEVs vanished, suggesting a higher stability of fluorescence switch when included in sEVs (Fig. 3G). This result confirmed that sEVs effectively protected the encapsulated miRNA from RNase degradation, especially when exposed to physiological milieu.

Verifying the fluorescence switching and tumor inhibition of GQDs/Cy5-miR@sEVs in vitro

Following the fabrication and characterization of GQDs/Cy5-miR@sEVs, we investigated the cellular uptake and “off–on” fluorescence switch of those particles. To assess miRNA delivery and gene regulation of GQDs/FAM-miR@sEVs in vitro, GQDs, GQDs/FAM-miR and GQDs/FAM-miR@sEVs were individually incubated with HGC-27 cells for 1, 2 and 4 h, respectively. Notably, we here adopt GQDs/FAM-miR@sEVs instead of GQDs/Cy5-miR@sEVs because reddot2 dye is used in this experiment to label the nuclei, which concomitantly overlaps with Cy5 fluorescent channel. As a result, images taken by confocal laser scanning microscope (CLSM) showed that HGC-27 cells treated by GQDs/FAM-miR@sEVs exhibited high FAM-miR uptake levels and successful fluorescent “off–on” switch, demonstrated by the observed FAM fluorescence in cells at 2 h that continued to glow inside cytoplasm and nucleus up to 4 h. This observation verified the FAM-miR detached from GQDs and bound to its target genes CCND1 in living cells. Oppositely, the FAM fluorescence in GQDs/FAM-miR treatment remained nearly undetectable in 4 h, suggesting the bindings between miR-193a-3p and its target genes CCND1 rarely occurred on this condition (Fig. 4A).

Fig. 4

In vitro fluorescent “off–on” switch and antitumor activity of GQDs/Cy5-miR@sEVs. A The fluorescence images showing the uptake of FAM-miR-GQDs@sEVs by HGC-27 cells at 1, 2, and 4 h. GQDs (blue), FAM-tagged miRNA (green) and nuclei (red) were shown, where scale bar = 100 μm. B FAM-miRNA (green) successfully escaped from endosomes (red) as evidenced by the separation of green and red fluorescence (indicated by arrows). Endosome/lysosome and nuclei were stained with Lysotracker Red and DAPI, respectively. Scale bar = 20 μm. C MiR-193a-3p was detected in HGC-27 cells by qRT-PCR after they were treated with GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs. D Target gene CCND1 was detected in HGC-27 cells by qRT-PCR after they were treated with sEVs, GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs, respectively. E Protein expression of CCND1 in HGC-27 cells by Western blot after being treated with PBS, GQDs, sEVs, GQDs/Cy5-miR, and GQDs/Cy5-miR@sEVs, respectively. F Western blot was used to detect the protein level of PCNA, Bax, Bcl-2, cleaved caspase-3 as well as β-actin proteins in HGC-27 cells. G Viability of HGC-27 cells and H viability of HUVEC cells were determined by CCK-8 assay. I Cell colony formation assays of HGC-27 cells when treated in different conditions as indicated. Data are expressed as mean ± SD and analyzed by one-way ANOVA. ns, no significance; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

This was supported by subcellular CCND1 mRNA localization studies utilizing RNA fluorescence in situ hybridization (RNA-FISH) technique. CLSM images showed that CCND1 mRNA was distributed in both nuclei and cytoplasm of HGC cells, explaining the spatial “off–on” switch of fluorescence during GQDs/FAM-miR trafficking inside cells (Additional file 1: Fig. S7). These results demonstrated that the sEVs encapsulation assisted the fluorescence switch GQDs/FAM-miR to relocate into cells. In order to track the released miR-193-3p from GQDs/FAM-miR@sEVs following cell internalization, the intracellular localization of FAM-miR was further examined after GQDs/FAM-miR@sEVs were incubated with HGC-27 cells for 2 h. As shown in Fig. 4B, FAM-miR in the GQDs/FAM-miR@sEVs was able to escape from the endosome/lysosome trapping, as evidenced by minimal overlapping of green (FAM-miR) and red (Lyso-tracker Red) fluorescence in the cytoplasm.

Then, the miR-193a-3p miRNA and its target gene CCND1 were detected by qRT-PCR after HGC-27 cells were treated with GQDs/Cy5-miR@sEVs or GQDs/Cy5-miR. The relative expression of miR-193a-3p in GQDs/Cy5-miR@sEVs showed an increase of 20.7-folds when compared with that in GQDs/Cy5-miR (Fig. 4C). Next, the gene expression of CCND1 exhibited a significant decline (> 20%) in GQDs/Cy5-miR@sEVs when compared with that in GQDs/Cy5-miR or in sEVs (Fig. 4D). Taken together, our results validated that GQDs/Cy5-miR@sEVs could effectively deliver miR-193a-3p into HGC-27 cells and substantially downregulate the transcription of its target gene CCND1.

Simultaneously, the Western blotting analysis suggested that the expression of CCND1 protein in HGC-27 cells declined after being treated with GQDs/Cy5-miR@sEVs for 48 h, while the cells treatment with GQDs, sEVs or GQDs/Cy5-miR revealed no noticeable changes in the CCND1 protein expression (Fig. 4E). CCND1 protein is well known as a key regulator in arresting cell cycle and inducing cell apoptosis. Consequently, we monitored the apoptosis markers in HGC-27 cells treated with GQDs/Cy5-miR@sEVs and other indicated groups. GQDs/Cy5-miR@sEVs significantly decreased the proliferating cell nuclear antigen (PCNA) protein expression level in HGC-27 cells. The level of the anti-apoptosis protein B-cell lymphoma-2 (Bcl-2) was significantly reduced, whilst the pro-apoptosis protein Bcl-2 associated X (Bax) and cleaved caspase 3 were increased (Fig. 4F).

Cell viability tests showed that the viability of GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs treated cells deceased to (32.4 ± 5.9) % and (18.1 ± 1.7) % (p < 0.001) after 48 h, respectively, whereas the treatment with GQDs or sEVs showed no statistical difference from the untreated (Fig. 4G). The same experiments were further performed on HUVEC cells, and the results revealed that GQDs/Cy5-miR@sEVs had no adverse effect on non-cancerous HUVEC cells (Fig. 4H).

To further assess the growth inhibition of tumor cells in vitro, HGC-27 cells were treated with different concentrations of GQDs/Cy5-miR@sEVs from 10 to 200 µg/mL (based on the total protein mass concentration of sEVs). Results revealed that GQDs/Cy5-miR@sEVs suppressed the tumor cell growth in a dose-dependent manner at 48 h when its concentration was within 100 µg/mL (Additional file 1: Fig. S9). Moreover, the cell colony formation assay demonstrated that GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs induced the fewest colony formations and exhibited the most durable therapeutic effect among all groups (Fig. 4I). Transwell migration and invasion tests also revealed that GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs most significantly prevented the migration and invasion of HGC-27 cells among all conditions (Additional file 1: Fig. S10). These results verified that both GQDs/Cy5-miR and GQDs/Cy5-miR@sEVs could promote miRNA delivery to inhibit the growth and proliferation of HGC-27 cells in vitro.

The anticancer effect of GQDs/Cy5-miR@sEVs in tumor-bearing mice visualized by their fluorescence switch

We next evaluated the fluorescence switching effect of GQDs/Cy5-miR@sEVs in a xenograft murine tumor model established by subcutaneous injection of HGC-27 cells in the flanks of BALB/c nude mice. Each group (n = 5) of mice was i.v. injected through tail vein with Cy5-miR, GQDs/Cy5-miR, and GQDs/Cy5-miR@sEVs, respectively (based on the equivalent dosage of Cy5-miR). Tumor accumulation and biodistribution of the Cy5-miR were monitored within 24 h p.i. As a result, attenuated fluorescence signals were detected at the tumor sites for the GQDs/Cy5-miR or Cy5-miR treated mice. In contrast, much higher fluorescence signals in tumors of mice injected with GQDs/Cy5-miR@sEVs were noticed, reaching a peak at 1 h p.i. and sustaining up to 6 h p.i. (Fig. 5A, B). After 24 h p.i., the mice were sacrificed before the major organs and tumors were harvested to examine the fluorescence biodistribution. As shown in Fig. 5C, D, the Cy5 fluorescence was mainly accumulated in the kidneys and tumors for each substance administered. In comparison, GQDs/Cy5-miR@sEVs showed higher accumulation than Cy5-miR or GQDs/Cy5-miR in tumors.

Fig. 5

In vivo fluorescent “off–on” switch effect of GQDs/Cy5-miR@sEVs. A The fluorescent images of tumor-bearing mice after i.v. injection of GQDs/Cy5-miR@sEVs at time points as indicated. B The fluorescent intensity of tumor sites in vivo at indicated time points. C Ex vivo fluorescent signals in major organs (i.e., liver, spleen, lung, heart, kidney, stomach, intestines, pancreas, and ovary) and tumors 24 h p.i. D Ex vivo fluorescent intensity of major organs and tumors 24 h p.i

We further evaluated the therapeutic effect of GQDs/Cy5-miR@sEVs on the subcutaneous tumor model established by xenograft HGC-27 cells injection in BALB/c nude mice. Notably, as miRNA was the therapeutic segment in GQDs/Cy5-miR@sEVs, we here included Cy5-miR@sEVs for a comparison. The fabrication of Cy5-miR@sEVs followed the same procedure as preparing GQDs/Cy5-miR@sEVs, using the equivalent amount of Cy5-miR. Then, the miRNA loading capacity of Cy5-miR@sEVs and GQDs/Cy5-miR@sEVs was compared. Results indicated that the content of miRNA in GQDs/Cy5-miR@sEVs was nearly fourfold of that in Cy5-miR@sEVs based on the same total protein mass of sEVs (Fig. 6A).

Fig. 6

Visualized treatment of GQDs/Cy5-miR@sEVs in the tumor-bearing murine model. A The miR-193a-3p in GQDs/Cy5-miR@sEVs and Cy5-miR@sEVs was detected by qRT-PCR. B Scheme illustrating GQDs/Cy5-miR@sEVs treatment in subcutaneous xenograft tumor model established by HGC-27 cells in BALB/c nude mice. C The in vivo tumor volumes were monitored over time during four injection cycles. D, E The fluorescence intensity of tumor region was monitored and plotted over time. F Images of subcutaneous xenograft tumors in mice in each group (n = 5) when the tumor sizes were measured ex vivo. G Tumor weights of mice in each group were measured. H Western blot assays detected the expression of CCND1 protein in different treatments. I Ki-67 and J TUNEL staining of tumors from mice in different groups. Scale bars = 50 μm. One-way ANOVA for multiple groups was applied for statistical analysis. ns, no significance; **p < 0.01 and ***p < 0.001

Then, each group of mice (n = 5) was i.v. administered with PBS, sEVs, Cy5-miR, and Cy5-miR@sEVs, and GQDs/Cy5-miR@sEVs (based on the equivalent dosage of Cy5-miR) for 4 cycles (3 days per cycle) (Fig. 6B) and the tumors growth in mice was monitored (Fig. 6C). During a total of 4 injections, individually at 12, 15, 18, and 21 days after tumor inoculation, the fluorescent intensities of tumors were monitored following 24 h p.i. (Fig. 6D, E). Substantial tumor growth was detected in PBS group, while mice with Cy5-miR, sEVs, or GQDs/Cy5-miR treatment revealed moderate tumor suppression. In contrast, the tumor growth in mice treated with GQDs/Cy5-miR@sEVs was significantly inhibited (Fig. 6F, G).

Concomitantly, the Western blotting assays demonstrated a distinctive decrease of CCND1 expression in mice treated with GQDs/Cy5-miR@sEVs compared to any of other treatments (Fig. 6H). Furthermore, the number of Ki-67 positive cells was significantly decreased, and the number of terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling (TUNEL) positive cells was notably increased after treatment with GQDs/Cy5-miR@sEVs, in comparison to that under other condition (Fig. 6I, J). These results confirmed that GQDs/Cy5-miR@sEVs could effectively induce cell apoptosis in tumors, showing their prominent capability on tumor inhibition in vivo. Simultaneously, for GQDs/Cy5-miR@sEVs treatment, there was no significant change in mouse body weight (Fig. 7A), nor pathological changes in main organs (Fig. 7B), nor any abnormality noticed in major hepatic and renal functions (Fig. 7C). These results reflected that GQDs/Cy5-miR@sEVs possessed a tumor-specific toxicity while bearing a very good biocompatibility.

Fig. 7

The biosafety evaluation of GQDs/Cy5-miR@sEVs in vivo. BALB/c nude mice were sacrificed after administration with PBS, sEVs, Cy5-miR@sEVs and GQDs/Cy5-miR@sEVs. A Body weights of mice subject to different treatments as indicated. B The major organs of mice in each group including liver, spleen, heart, lung, and kidney were collected for H&E staining. C The measured levels of biochemical indicators in the mouse blood samples, including the indices for liver (ALT and AST) and kidney (BUN and CREA). Data are expressed as mean ± SD and analyzed by one-way ANOVA. ns, no significance