1 Introduction

Small extracellular vesicles (sEVs), including exosomes, are generally nanometer in size (30–150 nm) and carry rich biological substances such as lipids, proteins, and nucleic acids [1,2,3,4,5]. These cell-derived vesicles could implement crosstalks between parental and recipient cells and trigger specific cell responses. This unique ability of sEVs recently aroused huge interest and was proven to be involved in the pathogenesis of various diseases, such as cancers [6,7,8], inflammations [9,10,11], metabolic disorders [12], and diabetes and its complications [13,14,15,16]. Regarding the mechanism of diabetic nephropathy (DN), it is already clear that direct high glucose (HG) stimulation could damage the filtering system inside the kidney and cause kidney injury [17,18,19,20,21]. However, the contribution of sEVs in this injury progression remains to be determined.

To our knowledge, much effort has been directed to figure out the participation of sEVs in kidney injury [22,23,24,25]. Wu et al. [26] validated the promoting effect of exosomes on renal fibrosis and confirmed that exosomes affected epithelial–mesenchymal transition (EMT) and renal fibrosis in DN via the paracrine communication between specific cells [27]. TGF-β1 mRNA, a genetic molecule involved in EMT and fibrosis, is transported by exosomes from injured epithelial cells to activate the fibroblasts [28]. So far, most of these findings were obtained using biological methods and were based on cell population experiments.

Recent encouraging studies showed that sEVs could regulate the biomechanics of cells [29] and even tissues [6]. For kidney fibrosis, cell biomechanics are potential markers at the single-cell level [30,31,32,33,34]. The elasticity and adhesion of tubular cells in vivo are closely related to fibrosis and injury processes [35]. With the use of an atomic force microscope (AFM), the increase in cellular stiffness of NRK52E cells has been identified as a sign of cellular fibrosis [36]. According to the AFM results, the EMT process could be characterized by analyzing the structural changes of cells [37]. In our previous study, the dynamic injury of human proximal tubule kidney (HK-2) cells caused by direct HG stimulation was quantitatively characterized by AFM mapping [38]. Formerly referred to as exosomes according to the MISEV guidelines, sEVs could also be characterized by high-resolution AFM [39]. These results inspired us to further explore the regulation of cell biomechanics by sEVs under indirect HG-induced kidney injury.

In this study, HG-induced sEVs were extracted to stimulate normal HK-2 cells, as shown in Fig. 1A. The biomechanical responses of HK-2 cells to different concentrations of HG-induced sEVs were then examined by AFM in liquid and air as shown in Fig. 1B. Finally, the effects of HG-induced sEVs were compared with those of direct HG injury to help confirm and understand the role of HG-induced sEVs during HG injury and to provide insights into sEV-based therapeutic approaches for the diagnosis and treatment of kidney injury.

Fig. 1
figure 1

Illustration of HG-induced sEVs regulating the biomechanics of HK-2 cells. A Protocol of HG-induced sEVs mediated by indirect HG treatment. HK-2 cell is a cell model for kidney research. The biological structure and contents of a sEV are also shown. B AFM-based characterization of HK-2 cells, with an optical image of AFM scanning and a typical force–distance (FD) curve

Full size image

2 Materials and methods

2.1 Cell culture and treatments.

HK-2 cells were cultured as described in Ref. [38], seeded in 90 mm dishes, and exposed to HG (60 mmol/L) for 72 h after reaching a confluence of 80%. The treatment duration and HG concentration were consistent with the HG conditions reported in Ref. [38] to ensure the induction of cell injury. HG-induced sEVs were collected from the culture mediums of the parental cells, that is, the pretreated HK-2 cells. The untreated HK-2 cells were labeled as the recipient cells; cultured in fresh minimum essential medium with basal glucose (5 mmol/L), 10% EV-depleted FBS (ABW), and different concentrations of HG-induced sEVs (0, 500, 1000, and 2000 μg/mL) for 48 h, and named as HK-2-N, HK-2-E500, HK-2-E1000, and HK-2-E2000, respectively.

2.2 Extraction of HG-induced sEVs

The supernatants of HK-2 cells treated with HG for 72 h were collected, centrifuged at 2000 × g for 10 min, recentrifuged at 10,000 × g for 30 min, filtered with 0.22 μm pore filters (Millpore), and ultracentrifuged (Beckman Coulter, Optima XE-90, USA) at 100,000 × g for 70 min. After the supernatant was removed and resuspended with PBS, a second ultracentrifugation was applied with the same parameters. The HG-induced sEVs were extracted, stored at − 80 °C, and examined with a BCA kit (Sangon Biotech).

2.3 Western blot (WB) for HG-induced sEV markers

Lysates of HG-induced sEV extractions were separated by 10% SDS-PAGE, and the proteins were transferred onto a PVDF membrane. After blocking by BSA solution, the membrane was first incubated separately with the mouse polyclonal antibody for CD9 and ALIX, followed by anti-mouse secondary antibodies. Finally, the protein markers of HG-induced sEVs were imaged with the imaging system (Analytik Jena, ChemStudio SA2, Germany) after hypersensitive chemiluminescence.

2.4 AFM characterization of HK-2 cells and HG-induced sEVs

The cells and HG-induced sEVs were characterized by AFM (JPK, Nano Wizard 3, Germany). The living cells, plated on cover glasses in Petri dishes, were investigated under the Quantitative Imaging Mode in liquid using an MLCT cantilever (Bruker, Spring Constant: 0.07 N/m) with the setpoint of 1 nN. The force curves of each pixel and the quantitative images were recorded. At least 10 cells were detected for each sample, and the detection was repeated three times. Hertz fitting was applied to calculate the Young’s modulus of cells.

The HG-induced sEVs, diluted appropriately and drop-casted on the mica, were imaged under the tapping mode in the air using a Tap300Al-G cantilever (Budget Sensors, Spring Constant: 40 N/m) with the setpoint of 2 nN. All the AFM results were processed with the JPK data processing software.

2.5 Scanning electron microscopy (SEM) imaging of HG-induced sEVs

In brief, 5 μL of the diluted HG-induced sEVs was dropped onto the silicon substrates, followed by a drop of 4% glutaraldehyde for fixation. The HG-induced sEVs were dried and then coated with a thin golden film of 1 nm thickness. The HG-induced sEVs were imaged by SEM (FEI, Helios G4 CX, USA).

3 Results and discussion

3.1 Characterization of HG-induced sEVs

Multiple techniques were applied to determine the HG-induced sEV features, such as particle size, distribution, and morphology. High-resolution images of HG-induced sEVs were produced via SEM and AFM. The individual HG-induced sEVs displayed an approximate sphere shape with a diameter of about 90 nm in Fig. 2A. Meanwhile, the HG-induced sEV populations were observed by AFM in a wide range (Fig. 2B) for the mathematical determination of the size distribution. Five AFM images were chosen for plotting the diagram. The histogram showed a unimodal size distribution of the HG-induced sEVs through Gaussian fitting, where 85.7% of the particles were statistically counted to exhibit lateral diameters ranging from 30 to 150 nm (Fig. 2C). The sEV-associated proteins could be specifically identified by WB experiments. The transmembrane proteins (CD9) and cytosolic proteins (ALIX) were expressed and differentiated between the sEVs and cells (Fig. 2D).

Fig. 2
figure 2

Characterization of HG-induced sEVs. A SEM image of HG-induced sEVs. B AFM image of HG-induced sEVs. C Size distribution of the lateral diameters of HG-induced sEVs statistically calculated from the AFM topography images. D Western blot results of the protein markers of HG-induced sEVs, CD9, and ALIX

Full size image

Different from SEM, only simple preparation was required for the AFM experiment. In addition, AFM could provide 3D information for the HG-induced sEVs. In Fig. 3A, B, the HG-induced sEVs showed round shapes similar to those in the SEM images. However, the height values were relatively lower than the lateral diameters (30–150 nm), not exceeding 10 nm. A large ratio of diameter/height (RatioD/H) was observed (Fig. 3D), indicating the nonspecific flattening of the sEV structures. The differences in the images of sEVs are highly related to the methods of sample preparation and imaging techniques [40, 41]. The HG-induced sEV samples were adsorbed onto mica, detected in air, and might have experienced slight dehydration, thus leading to flattening.

Fig. 3
figure 3

Dimensional analysis of HG-induced sEVs. A Height image of HG-induced sEVs. B 3D image of HG-induced sEVs. C Corresponding cross-section lines of a, b, and c in A. D Particle height, diameter (lateral), and ratio of diameter/height (RatioD/H) of HG-induced sEVs. E Height image of HG-induced sEVs. F Phase image of HG-induced sEVs. Squares 1 and 2 are the height images, and squares 3 and 4 are the corresponding phase images of the same area. G and H Magnified phase images of squares 3 and 4. I and J Corresponding cross-section lines d and e in G and H

Full size image

The central areas of HG-induced sEVs were darker than the peripheries in the phase image (Fig. 3G, H), leading to the cupped shape shown in the cross-section lines (d, e) in Fig. 3I, J. However, this unique feature cannot be identified in the height image. Therefore, this collapse was not topographical but viscoelasticity-related. sEV could be described as a soft internal cavity restricted by a stiff membrane with diverse membrane components, as shown in Fig. 1A [2]. These surface molecules, such as proteins and lipids, would bind with the AFM tips. Thus, the surface viscoelasticity of HG-induced sEVs was potentially reflected in the phase images.

3.2 HG-induced sEVs activated the morphological alterations of HK-2 cells

The biomechanics of the HK-2 cells treated with increasing concentrations of HG-induced sEVs were studied by AFM to examine the role of sEVs in HG-injured kidney tubular epithelial cells. Cellular morphological features considered important markers for cellular states, such as cell shapes, pseudopodia, and surface topographies, were detected [42, 43].

The AFM images (Fig. 4A) showed that the HK-2 cells experienced dynamic shape alterations, from the original cobblestone-like shape to irregular shapes, and were ultimately elongated. The ellipticity of normal HK-2-N cells was 32.63% ± 13.24%. Meanwhile, the HG-induced sEVs increased the cell ellipticities, presenting a dose-dependent effect (Fig. 4C1). In addition, the sEVs induced the transitions of pseudopodia in their states and lengths. The ultrastructures of pseudopodia were observed, with the enlargement of detailed images marked with white squares shown in Fig. 4B. Normal HK-2 cells presented the flake-shaped pseudopodia, called lamellipodia, which extended from the cell body and the average length of 5.9 ± 1.8 μm. After the treatment of HG-induced sEVs for 48 h, the scattering lamellipodia around the cell body turned out to be filopodia, which were extended from the cells. The average lengths of pseudopodia increased with the concentrations of HG-induced sEVs (Fig. 4C2). The filopodia of HK-2-E2000 stretched, and the average length doubled compared with that of HK-2-N. Pseudopodia are functional components based on the microtubular cytoskeleton outgrowths of cells [44] and are involved in multiple processes, such as cell spreading, migration, and interactions between cells and external environments. Therefore, the morphologies of pseudopodia are crucial cues for characterizing the cytoskeleton dynamics.

Fig. 4
figure 4

Morphological transition of HK-2 cells caused by HG-induced sEVs. A1–A4 AFM images of HK-2-N, HK-2-E500, HK-2-E1000, and HK-2-E2000 in air. B1–B4 Corresponding zoom-in images of pseudopodia marked squares in A1–A4. C1–C3 Cell ellipticity, pseudopodia length, and surface roughness. The data in C1 and C2 are presented as mean value ± standard deviation (0.01 < *p < 0.05, 0.001 < **p < 0.01, ***p < 0.001)

Full size image

The areas of interest with equal size (9.4 × 9.4 μm2) on the cellular surfaces of individual HK-2 cells from each sample were selected as shown in Fig. S1. Compared with that of control HK-2 cells, the roughness value showed an overall increase and was maximized when the concentration of HG-induced sEVs reached 1000 μg/mL. This trend was consistent with the 3D images. Surface roughness was used as an indicator of membrane micro-irregularities [45]. The changes in roughness suggested the surface modification caused by the interaction between the external sEVs and cells.

Moreover, the average roughness (Ra) values of at least 20 cells from each sample were plotted (Fig. 4C3) and statistically analyzed. The Ra values showed a wide range, suggesting that part of the cells exhibited increased surface roughness due to the treatments of HG-induced sEVs. A slight uptrend was observed in the average roughness in contrast with the Ra of untreated HK-2 cells. The roughness of individual cells and the average for multiple cells were increased in general, demonstrating the roughening responses of the HK-2 cells to the HG-induced sEVs. However, the concentration of HG-induced sEVs that led to the maximum roughness was inconsistent. This phenomenon could be caused by the heterogeneity of cells and the existing differences between individual cell and population analyses [46], suggesting the necessity of further analyzing the morphologies of individual cells and cell groups.

Multiple studies have concluded that HG prompted the morphological phenotype change of HK-2 cells from an oval shape to a spindle shape similar to a fibroblast [47, 48]. Our previous study was in agreement with these results [38]. In the current work, the HG-induced sEVs were derived from the HG-treated HK-2 cells, causing similar morphological conversions of HK-2 cells. The high concentration of HG-induced sEVs drove the changes in cellular shapes and pseudopodia and roughened the cellular surface to some extent. As a result, the effect of sEVs on HG-induced cellular injury was confirmed.

3.3 Quantitative analysis of cellular biomechanics affected by HG-induced sEVs

AFM allowed the quantitative analysis of the mechanical properties of living cells in their physiologically relevant liquid [49]. During the measurement, the interaction between AFM tips and cells was recorded as the FD curves to show the different features that depend on the sample characteristics. In general, the up-and-down motion of the probe would be inevitably resisted by liquid, leading to the distance between the extend and retract curves. Many zig–zag peaks in the retract curves were produced because of the interactions between the tips and glycocalyx coated on the cell surface. The maximum zig–zag peak was defined as the adhesion of this point. Hysteresis between the extend and retract curves was frequently found when scanning the biological samples [42].

The features of cells could be reflected by the FD curves, so their mechanical properties could be extracted from these curves [20]. Here, single-cell and multicell average analyses [50] were adopted for the quantitative evaluation of cellular biomechanics. Through the analyses of cross-section lines of Young’s modulus and adhesion from the central areas of 20 cells in each group, the quantitative changes in cellular biomechanics were observed, as shown in Fig. 5A and B. With the increasing concentrations of HG-induced sEVs, Young’s modulus increased, and adhesion decreased.

Fig. 5
figure 5

Quantitative analysis of the biomechanics of HK-2 cells treated with different concentrations of HG-induced sEVs. A1–A4 and B1–B4 Cross-section lines from the images of adhesion and Young’s modulus. The central areas of 20 cells were selected from each group. C1 Analysis of Young’s moduli and adhesions based on a single cell. C2 and C3 Data distributions of Young’s moduli and adhesions based on cell population

Full size image

Young’s modulus and adhesion were two crucial parameters for characterizing biomechanics, with the former derived from the fitting mathematical models of contact mechanics and the latter directly obtained from the FD curves. The Hertz model was the simplest and most used contact mechanics model for calculating the elasticity of living cells [51,52,53]. Although it did not take surface adhesion into account, it was still an effective one-parameter model and was suitable for the relative quantifications because the varying adhesions of the cellular surface would lead to a high level of inaccuracy. The Young’s modulus and adhesion for a single cell are shown in Fig. 5C1.

Young’s moduli and adhesions were box-plotted in Fig. 5C2 and C3. Multicell level analysis revealed the general mechanical changes of HK-2 cells responding to the different concentrations of HG-induced sEVs. The Young’s moduli of HK-2-E500 and HK-2-E1000 showed minor differences to those of HK-2-N. Only partial cells from HK-2-E2000 displayed sharply increased Young’s moduli. The adhesion showed a downward tendency and was remarkably reduced with the increasing sEV concentrations.

The mechanical properties were found to be related to the concentrations of the HG-induced sEVs. At the maximum concentration of 2000 μg/mL, the cellular biomechanics exhibited differences. In particular, the HG-induced sEVs affected the cellular surface adhesion more than Young’s modulus because only a few HK-2 cells showed sharply increased Young’s moduli. The average Young’s modulus increased by 26.2%, and the average surface adhesion decreased by 36.8% (from 1.071 to 0.685 nN).

The elasticity and surface adhesion were tightly linked to the state of cells. Together with the morphological changes mentioned above, the decreased elasticity (increased Young’s moduli) and surface adhesion were regarded as the biomechanical cues of the cells in response to the HG treatments [54, 55]. Our results demonstrated similar morphological and mechanical alterations on cells caused by HG-induced sEVs but to different extents. This phenomenon was likely due to the sources and capabilities of HG-induced sEVs that enabled them to elicit specific cellular responses similar to direct HG stimulation. sEVs delivered the HG injury-related signals and mediated the recipient cells through intercellular crosstalks.

The cellular mechanical properties under direct HG and indirect HG-induced sEV stimulations were compared to evaluate the indirect effect of HG through sEV mediation. In our previous work (Ref. 38), HG treatment for 72 h increased the average Young’s modulus by over 70%, which was higher than the 26.2% recorded for the cells treated with HG-induced sEVs. This finding indicated that the direct HG stimulation resulted in a more significant change in the cellular elasticity compared with the indirect stimulation. In terms of surface adhesion, the reduction caused by the direct HG stimulation was only 5.9%, which was smaller than the 36.8% caused by the indirect HG-induced sEVs. In short, HG and indirect HG-induced sEV stimulations caused consistent mechanical trends, suggesting that both treatments induced the same physiological changes in HK-2 cells. However, differences existed in the variations of the cellular elasticity and adhesion, which could be related to the differences between glucose–cell and sEV–cell interactions.

4 Summary

This work investigated the morphologies and biomechanics of HK-2 cells responding to HG-induced sEVs via AFM-based single-cell force microscopy in air and liquid. Results showed that treatment with increasing sEV concentrations induced the morphological transformations of decreased cellular elasticity and surface adhesion. Thus, the indirect role of HG-induced sEVs in transmitting HG injury between cells was confirmed. The differences between the two stimulations were also studied by comparing the induced biomechanical changes. This work provides new insights into the research into cellular injury.