Heat damage of cytoskeleton in erythrocytes increases membrane roughness and cell rigidity

Abstract

The intensity of erythrocyte membrane fluctuations was studied by laser interference microscopy (LIM), which provide information about mechanical properties of the erythrocyte membrane. Atomic force microscopy (AFM) was used to study erythrocyte surface relief; it is related to the cytoskeleton structure of erythrocyte membrane. Intact human erythrocytes and erythrocytes with a destroyed cytoskeleton were used. According to the obtained results, cytoskeleton damage induced by heating up to 50 °С results in a reduced intensity of cell membrane fluctuations compared to non-treated cells (20.6 ± 10.2 vs. 30.5 ± 5.5 nm, correspondingly), while the roughness of the membrane increases (4.5 ± 1.5 vs. 3.4 ± 0.5 nm, correspondingly).

Introduction

Like other live cells, erythrocytes, or red blood cells (RBC), are characterized by local membrane fluctuations called flickering or dynamic membrane fluctuations (low-frequency vibrating movements (0.1–30 Hz) of the cell membrane) [1]. In spite of a long history of studies of such fluctuations and numerous publications devoted to their modeling and spectral analysis (see, e.g., [2, 3]), molecular and cellular mechanisms of this phenomenon are still under discussion. Some authors suppose that the membrane flickering of erythrocytes is caused by a thermal collision of molecules of both extracellular and intracellular media with the plasma membrane and, therefore, represents a passive process occurring without energy costs. In these studies, the observed dependence of a fluctuation amplitude on the intracellular ATP content is connected with changes in the elastic properties of the membrane [2, 4, 5]. Other authors suppose that the membrane flickering is determined by phosphorylation of cytoskeletal or cytoskeleton-associated proteins that cause changes in the spatial configuration of the cytoskeleton and, therefore, cell membrane fluctuations [6–9]. In both cases, it is supposed that the flickering amplitude depends on the cytoskeleton state, i.e., changes in the elastic properties of a membrane-cytoskeleton complex or formation of local membrane bends (protrusions) due to a rearrangement of the spectrin network (e.g., a cleavage of one spectrin filament from a nodal complex with the formation of a nonplanar network configuration).

To date, fluctuations of the RBC membrane are studied using a range of methods (see, e.g., [7, 10, 11]); however, methods based on the measuring of optical path difference (OPD) of biological objects are the most popular and convenient [12–15].

Surface relief (height differences on cell surface) is another important characteristic of erythrocytes. In recent years, it was actively studied using an atomic force microscopy (AFM), which provides a detailed study of the cell surface relief at a nanometer scale [16–18].

Both mentioned parameters (membrane fluctuations and membrane surface relief) may be connected with some important physiological properties of erythrocytes, such as volume regulation, ability to undergo deformations in blood vessels and pass through small capillaries, and, finally, to supply tissues with oxygen. Probably, local oscillations of the “membrane-cytoskeleton” complex may result in changes of such macroscopic RBC parameters as the form and the volume. Many studies devoted to the surface relief of erythrocytes, particularly membrane roughness, proved that this parameter depends mainly on the state of the cytoskeletal spectrin network [17–19]. One can suppose that increase in the amplitude of membrane fluctuations results in the formation of a relief difference on the cell surface; however, there are no studies which would prove this hypothesis under similar impact on a cell. To date, relations between changes in cell membrane fluctuations and the surface relief of the membrane still have not been directly demonstrated.

To study the role of the cytoskeleton in the regulation of the RBC form and functions, scientists often use an approach based on a structural modification of a two-dimensional (2D) cytoskeleton achieved by cell heating to 49–50 °С that causes melting of spectrin heterodimers composing the membrane cytoskeleton [1, 20, 21]. In this case, the rigidity of the membrane-spectrin complex increases, RBC form is changed from discocytes to spheroechinocytes, and spectrin-free vesicles are segregated from membrane protrusions, so the cells become spherocytes; then the hemolytic resistance of erythrocytes decreases, and some proteins are denatured [1]. Increase in membrane rigidity is considered to be connected with the aggregation and immobilization of denatured spectrin molecules [20]. Also the association of some cytosolic proteins (HSP90a, HSP70, catalase, a-enolase, CA (isoforms I and II), Prdx VI, FR, Prdx II, and peptidylprolyl cis-trans isomerase A (PPIA)) with the erythrocyte membrane was found [22]. Experiments with rat erythrocytes showed that the preliminary incubation of cells at 49–50 °С causes inactivation of volume-dependent ionic transporters [23, 24]. Under such conditions, melting of spectrin heterodimers suppresses swelling-induced changes in the folded relief of a membrane [16]. The process of rearrangement of a two-dimensional cytoskeletal network probably represents one of the mechanisms regulating the RBC volume and form.

Taking into account the abovementioned facts, the test impact chosen in this study represented thermal damage of the RBC skeleton by a cell preheating up to 50 °С. The purpose of this study was the investigation of fluctuations of erythrocyte membranes and also the relief of their surface using the laser interference microscopy (LIM) and atomic force microscopy (AFM), respectively.

Materials and methods

Object of study

Blood was collected from a healthy human donor, after obtaining informed consent in accordance with the Ethics Committee of Biological Department of Lomonosov Moscow State University approval, in tubes containing 20–30 IU/ml heparin and stored on ice for no more than 2 h. All procedures were performed in accordance with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Erythrocytes isolated from human capillary blood were washed by a triple centrifugation at 4 °C in a buffer containing 145 mM NaCl, 5 mM KCl, 4 mM Na₂HPO₄, 1 mM NaH₂PO₄, 1 mM CaCl₂, 1 mM MgSO₄, and 10 mM glucose (pH 7.4). For the LIM study, the obtained packed RBC was diluted at a ratio of 1:200, then placed onto a glass slide with a mirroring surface, and covered by a cover glass. For the AFM study of fixed erythrocytes, RBC suspension was diluted 200-fold, supplemented with glutaraldehyde at a final concentration of 1% for 1 hour, and washed by a 4× centrifugation in distilled water. The smear of the resulted cell suspension was placed onto an object plate and air-dried. To obtain erythrocytes with a destroyed cytoskeleton, the precipitated cell suspension was diluted tenfold, incubated at 50 °С for 10 min, and then diluted 20-fold in a cooled buffer. The choice of a temperature regime for this experiment was determined by the results of the earlier studies, in which a short-time preincubation of cells at 49–50 °С inactivated transporters sensitive to volume changes [16, 23]. The further measurements were performed at room temperature.

Statistical treatment

The results were statistically processed using Graphpad Prism 7.04 demo software (GraphPad Software, La Jolla California, USA). We estimated not less than 40 cells for specimen (LIM and AFM). All data were normally distributed (D’Agostino-Pearson omnibus normality test, p < 0.05). Statistical differences between control- and heat-treated erythrocytes’ groups were analyzed using unpaired t-test with Welsh correction.

Laser interference microscopy

This type of a microscopic examination provides a quantitative evaluation of an optical path difference at each point of a phase image (PI) [14, 15, 25–28]. Therefore, PI is a 2D projection of the OPD of a 3D object. The study was carried out using an automated MIA-1D interference microscope constructed at the All-Russian Research Institute of Optical and Physical Measurements and based on a MII-4 Linnik interferometer (LOMO, St. Petersburg, Russia) [29]. The microscope was equipped with a 33×/0.65NA objective providing the image size of 180 × 135 μm². Interference images were captured using a Winphast program. Reconstruction of phase images from the obtained interference images was performed by a phase-shifting technique (see [30] for details). The method is based on the recording of a series of interferograms with different relative phase shifts between the reference beam and a beam passing through an object and on the calculation of parameters of an ideal interferogram equation using a linear equation system and the least squares method. A relative phase shift between interferograms was preliminarily calculated by a Fourier transform method [31] using planar mirror interferograms with the recording of a shift equal to ~10 interference fringes. The recording was carried out using a high-speed (300 frames/sec) black-and-white FastVideo 300B CCD-camera (SPA ASTEK, Russia) equipped with a 1/2-inch array (6.33 × 4.75 mm²) and providing a 640 × 480 resolution. The phase image reconstruction was carried out using ten images with interference bands that made it possible to register phase images with a frequency up to 30 Hz. A sample was illuminated with a diode laser (659 nm, 5 mW), so the illumination power per one RBC did not exceed 2 mW. A laser beam was directed transversely to the cell plane. According to the calibration with the standards of different height, the vertical resolution was equal to λ/350; the horizontal resolution was about 0.5 μM.

For image processing, the open-source package FIJI [32] based on ImageJ was used. In the course of the experiment, a series of 512 PIs obtained at 25 Hz was recorded. Afterward a standard deviation of temporal OPD oscillations was calculated for each pixel in projection using the following formula:

$$\sqrt{〈\mathit{OP}{D}^2〉}=\sqrt{\frac{1}{n}\sum _{i=1}^n{\left(\mathit{OP}{D}_i-\overline{\mathit{OPD}}\right)}^2}$$ 1

where OPD_i is the OPD value at the i point of the time series, $\overline{\mathit{OPD}}$ is the arithmetic mean of the OPD values of the time series, and n is the total number of measurements per an image point. Then the average value of root-mean-square amplitude ${\sqrt{〈\mathit{OP}{D}^2〉}}_{\mathit{\text{mean}}}$ was calculated for the whole cell’s projection (Fig. 1).

Fig. 1.

Scheme of a LIM-based measurement procedure. A series of 512 phase images was recorded at 25 Hz. Then a z-projection of the series, representing the projection of stacked images along the axis perpendicular to an image plane and containing a root-mean-square amplitude of the temporal OPD oscillation (see formula in figure) of each pixel of the projection, was calculated. Finally, the mean value of a root-mean-square amplitude of the z-projection was calculated. To exclude the contribution of lateral cell’s fluctuations, this value is calculated excluding values obtained for the erythrocyte boundary

To avoid a contribution of lateral fluctuations of the cell edge, the abovementioned value was determined excluding the erythrocyte edge (2 pixels from the cell border; Fig. 2b, d).

Fig. 2.

Typical LIM images of erythrocytes. a, b, non-treated cell (discocyte); c, d, heat-treated cell with a destroyed cytoskeleton (spherocyte); b and d represent a root-mean-square projection of a series of images for non-treated and heat-treated erythrocytes, respectively

The value of ${\sqrt{〈\mathit{OP}{D}^2〉}}_{\mathit{\text{mean}}}$ depends on both the cell surface fluctuations and a background noise (device vibrations, movements of liquid, camera noise, etc.). Thus, for each point of the cell image, one may calculate a corresponding deviation using the following formula [13]:

$${\sqrt{〈\mathit{OP}{D}^2〉}}_{\mathit{\text{mean}}}=\sqrt{{\left({\sqrt{〈\mathit{OP}{D}^2〉}}_{\mathit{mea}{n}_{\mathit{\text{cell}}}}\right)}^2+{\left({\sqrt{〈\mathit{OP}{D}^2〉}}_{\mathit{mea}{n}_{\mathit{\text{noise}}}}\right)}^2}$$ 2

where ${\sqrt{〈\mathit{OP}{D}^2〉}}_{\mathit{mea}{n}_{\mathit{\text{cell}}}}$ represents temporary cell oscillations, while ${\sqrt{〈\mathit{OP}{D}^2〉}}_{\mathit{mea}{n}_{\mathit{\text{noise}}}}$ represents the oscillation of a phase image area (noise) on the surface of a clear mirror object plate of the same area.

Atomic force microscopy

Cell images were obtained using a NTEGRA SPECTRA AFM complex (NT-MDT Co., Russia) and a Nova registration program (NT-MDT Co., Russia). Measurements were carried out in a semi-contact mode using NSG 10-A cantilevers; the average coefficient of elasticity was 11.8 N/m, and the radius of a tip curvature was 10 nm. The size of a scanned image was 20х20 μm or 1х1 μm (256 × 256 points); the scanning rate was 0.5–1 Hz. Image treatment was performed using a SPIP 6.0 program (Image Metrology A/S, Denmark); the process was described in details in [16]. To evaluate the surface roughness of erythrocytes, we examined areas with the size of 1х1 μm. Image tilting was corrected by a subtraction of a plane approximated by a third-order polynomial. The surface roughness was calculated as

$$S_a=\frac{1}{\mathit{MN}}\sum _{k-0}^{M-1}\sum _{l=0}^{N-1}\mid z\left(x_k,y_l\right)$$ 3

where MN is the number of points on x and y coordinates, k and l are point indices, and z is the amplitude at the (x_k, y_l) point.

Results

Typical LIM images of intact or heat-treated erythrocytes are shown in Fig. 2. Non-treated RBCs appeared as discocytes, whereas cells with the destroyed cytoskeleton were spherocytes.

For nucleus-free RBC, the OPD value depends on the refractive index of a cell (n_cell), cell thickness (z), and refractive index of a medium (n_m):

$$\mathit{OPD}=\left(n_{\mathit{\text{cell}}}-n_m\right)\cdot z.$$ 4

The maximum contribution to the refractive index of a cell is made by the membrane and homogenous hemoglobin-containing cytoplasm [14]. However, the contribution of the cell membrane to OPD is rather small due to a small membrane thickness comparing to the thickness of a cytoplasm (~10 nm vs. 2 μm).

Hemoglobin is the main component of the RBC cytoplasm, so fluctuations of its concentration caused by its diffusion and redistribution within a cell could contribute to the observed OPD fluctuations. However, hemoglobin concentrations at neighboring cell regions are rapidly leveled due to diffusion processes. Based on the diffusion coefficient for hemoglobin D equal to ~4.5 × 10^-8 cm²/sec [33], one can calculate a root-mean-square (RMS) displacement of hemoglobin molecules ($\sqrt{4\mathit{Dt}}$) occurring during the obtaining of one PI (0.04 sec); such displacement makes 0.85 μm. A distance between neighboring points of a phase image is 0.28 μm that means that, due to diffusion processes, the difference between hemoglobin concentrations at these points is leveled within the time of image registration. The refractive index of a cell depends on the hemoglobin concentration in the following way [34]:

$$n_{\mathit{\text{cell}}}=n_m+\mathit{\alpha C}.$$ 5

Thus, in our experiments, we cannot observe local changes in the refractive index connected with fluctuations of a hemoglobin concentration, and the measured OPD fluctuations may be caused only by changes in cell thickness.

To calculate thickness fluctuations, we used the Eq. 4. In this case, temporal oscillations of the thickness are equal to

$${〈z〉}_{\mathit{mea}{n}_{\mathit{\text{cell}}}}=\sqrt{{〈{\left(\frac{\mathit{OPD}}{n_{\mathit{\text{cell}}}-n_m}\right)}^2〉}_{\mathit{mea}{n}_{\mathit{\text{cell}}}}}$$ 6

where n_m is a refractive index of a buffer solution (1.335) and n_cell = 1.404 [14]. Also, using this equation, we can calculate the temporal oscillation of noise, ${〈z〉}_{\mathit{mea}{n}_{\mathit{\text{noise}}}}$, where instead the refractive index of erythrocyte, the refractive index of glasses, 1.52, be applied. This value was approximately equal to 3.0 ± 1,7 nm.

The ${〈z〉}_{\mathit{mea}{n}_{\mathit{\text{cell}}}}$ values of non-treated cells and cells with a destroyed cytoskeleton are shown in Fig. 3. For non-treated RBC, the RMS deviation of a cell height is significantly higher than that for cells with the destroyed cytoskeleton (30.5 ± 5.5 vs. 20.6 ± 10.2 nm, respectively).

Fig. 3.

Temporal oscillation of the z value. 1, non-treated erythrocytes; 2, heat-treated erythrocytes with a destroyed cytoskeleton. Dash line – a mean value of the temporal oscillation of noise. A significant difference between the values is marked with an asterisk

The dependence of mechanical membrane properties on the intensity of membrane fluctuations can be simplistically described by the following formula [12, 30]):

$$k_e=\frac{k_{B}T}{{〈z^2〉}_{\mathit{mea}{n}_{\mathit{\text{cell}}}}}$$ 7

where k_e is the equivalent elastic constant of membrane related with the tension coefficient and the elastic modulus [2], k_B is the Boltzmann constant, and T is the absolute temperature of a sample (298 K). The rigidity of RBC with a destroyed cytoskeleton is higher than in intact cells (k_e values for non-treated and heat-treated RBC were 4.8 × 10^-6 ± 1.7 × 10^-6 and 31.09 × 10^-6 ± 50.49 × 10^-6 J/m², respectively).

Typical AFM images of erythrocytes are shown in Fig. 4. Erythrocytes visualized by LIM and AFM are similar between themselves that confirm the absence of the effect of cell fixation and drying on the cell morphology.

Fig. 4.

Typical AFM imaging of fixed erythrocytes. a, b, non-treated cell; c, d, heat-treated cell with a destroyed cytoskeleton; a and c represent erythrocytes, while b and d show small areas of their membrane surface (indicated by white squares) at a higher resolution with the subtraction of a third-order approximation surface

The obtained AFM images of erythrocytes revealed a folded relief of a cell membrane with considerable height difference and also small globular structures on the membrane surface. Earlier we calculated the roughness (S_am) of such a surface [16] and showed that the value of this parameter is significantly higher in RBCs with a destroyed cytoskeleton than in intact RBCs (4.5 ± 1.5 vs. 3.4±0.5 nm, respectively) (see Fig. 5). According to the existing data, temperature increase to 50 °C causes irreversible melting of spectrin dimers in erythrocytes [24, 35], but no denaturation of other proteins, such as bands 2.1, 4.1, and 4.2, and the cytoplasmic domain of band 3 [21] was shown, so we can suppose that the observed relief changes are connected with changes in the cytoskeleton structure.

Fig. 5.

Membrane roughness of fixed erythrocytes. 1, non-treated cells; 2, heat-treated cells with a destroyed cytoskeleton measured by AFM. A significant difference between the values is indicated with an asterisk

Discussion

In the study [16], we supposed that the surface relief of erythrocytes is directly connected with membrane fluctuations: the more expressed fluctuations, the larger the surface relief difference (and the roughness parameter) of the fixed cells. However, in this study, we demonstrated that heat-induced cytoskeleton damage resulted not only in the increase in surface relief differences but also in the reduction of fluctuation amplitude and in the increase in membrane rigidity. Authors of other studies, who used heat-induced cytoskeleton damage at 50 °C, also observed increased rigidity (see, e.g., [20]) and reduced membrane fluctuations [1]. This fact probably means that the static surface relief of erythrocyte membranes does not represent an instantaneous “frozen” membrane state, and the presence of protrusions and considerable height difference is not connected with dynamic fluctuations. Probably, the reason for the increase of membrane stiffness and increased relief differences is the association of cytosolic proteins with membranes or cytoskeleton of erythrocytes, as was shown in [22]. It occurs simultaneously with denaturation of spectrin in erythrocytes and changes in the mechanical properties of the membrane-spectrin complex. In some publications (see, e.g., [21]), it was shown that thermal denaturation of the cytoplasmic domain of the band 3 protein results in the detachment of the spectrin network from the ankyrin complex. However, this was observed at higher temperatures than in our study (about 63 °C, see [21]). Therefore, we can assume that in this case there is no total detachment of the spectrin cytoskeleton from the membrane. At the same time, it is possible that a local decrease in ATP concentration with increasing temperature leads to the cleavage of individual spectrin molecules from transmembrane complexes and the formation of sections of the nonplanar cytoskeleton configuration according to the mechanism described in [8]. Besides, membrane fluctuations depend not only on mechanical properties of a cell surface but also on the viscosity of the extracellular and intracellular medium and the cell shape. Both parameters may vary during heating of cells to 50 °C; it was shown that in this case, the activity of ion transporters [16] and the morphology of erythrocytes [1] have also undergone some changes. Authors of the study [3] note that cell shape plays an important role in the modeling of flicker: the transition to a spherical form results in a decreased fluctuation intensity in a low-frequency range and an increased fluctuation intensity in a high-frequency range. Fluctuations observed in this study were mainly in the low-frequency range (up to 10 Hz), so the observed reduction of their intensity corresponded to the behavior predicted in [2].

Conclusion

In this study, we showed that the damage of a cytoskeleton results in a reduced amplitude of membrane fluctuations, accompanied by increased relief differences. It seems that spectrin denaturation causes an increase in both membrane rigidity and relief differences, which may be connected with structural changes of the cytoskeleton in the course of its denaturation and with a formation of nonplanar cytoskeleton configuration with protrusions and hollows. Association of cytosolic proteins with the membrane and cytoskeleton is also possible when heated. At the same time, intensity of membrane fluctuations may be also controlled by some other factors, such as the viscosity of the extracellular and intracellular medium, the cell shape, or transmembrane protein movements.