From surface to intracellular non-invasive nanoscale study of living cells impairments

Studying the intricate structures of a cell constitutes a major challenge of non-destructive label-free metrology in physiological medium. As recent reports shed light on the importance of the nanomechanical properties [1–3] of living cells in cancer or drug delivery, there is a rising urge to access the information below the surface [4–6] and resolve the inner features of cells in liquid environment [7–10]. The availability of such a nanometrology tool will likely impact the current understanding of cellular behavior. Characterizing the effects  of  the cytoskeleton dynamics on the cell elasticity and intra-cellular conformational changes represents a prime example of what liquid-mode synthesizing atomic force microscopy (l-MSAFM) can achieve, with direct applications in medicine. Due to the infancy of l-MSAFM, we use AFM and confocal imaging to get some insight of the mechanisms at play in the cell and establish a connection with the features observed in l-MSAFM.

While in-liquid imaging of biological systems is key to a better understanding of the cellular dynamics, high spatial and time resolution intracellular techniques capable of operating in buffer media for label-free characterization are non-existent [7–10]. Moreover, high resolution of HeLa cells is also achieved with the use of extended nanoneedle tip, which can counteract the hydrodynamic damping due to the liquid during AFM operations [11, 12]. Recent reports highlight the growing interest in such capabilities with the development of environmental scanning electron microscopes [8], nanowire based endoscopy [13], multi-harmonic atomic force microscopy (AFM) [14], scanning electrochemical microscopy [15]. The potential of scanning ion-conductance microscopy [16, 17] has also been illustrated by imaging fibroblasts without any damage of the sample observed. Besides advances in multi-dimensional mechanical characterization, the temporal resolution has recently been increased by High-Speed AFM and draws great interest to explore the dynamic mechanisms of biological systems such as bacteria [18]. Furthermore, multifrequency AFM-based techniques are emerging as a powerful platform for exploring material properties [19–22]. The l-MSAFM presents the unique advantage of overcoming the various problems of diffraction, and thus the loss of resolution in conventional acoustic microscopy, by taking advantage of the nanometer sized tip of the probe allowing measurements of the nanoscale response of the material [20, 22]. Importantly, the flexibility of the AFM also makes it possible to expose the sample to external stimuli [23] and study their effects on subsystems of interest. Here we mechanically excite the cells as required for l-MSAFM imaging and chemically expose the cells to stress (pesticide). Specifically, oxidative-stress on keratinocytes (HaCaT cells) was induced by exposure to glyphosate, a common herbicide and an accepted oxidative stress induction model [24]. Glyphosate is known to impair HaCaT cells by inducing an intracellular oxidative cascade, which diminishes the HaCaT cell integrity. Thus despite the natural resistance of epidermal cells to various external potential physical or chemical threats, glyphosate causes harmful effects to biochemical, morphological and histological properties of the HaCaT cells [25, 26].

l-MSAFM is introduced (figure 1) to non-destructively study the heterogeneities within living cells, specifically the intracellular changes induced by an oxidative-stress in keratinocytes skin cells in their physiological environment. The parameters of glyphosate exposure used (0.5 h exposure at 53 mM glyphosate) correspond to an early oxidative stress configuration for the keratinocytes skin cells [27]. In l-MSAFM, the sample is immersed in liquid as depicted in figure 1. A customized liquid cell was developed in order to host the piezoelectric components at the base of the cantilever and below the sample–substrate interface, while preserving the isolation features of the fluid cell. Liquid imaging implementation constitutes a major challenge in nanoscale metrology, especially in AFM-based techniques [28, 29]. Indeed, the dynamical attributes resulting from the liquid–cantilever interaction modify the frequency content of the tip–sample interaction. With a particular attention given to the design of the acoustic liquid cell, retrofitting additional PZTs, and modifying the probe tilt angle for optimum reflectometry of the probe's position, we have achieved in-liquid high resolution subsurface measurement capability. In l-MSAFM, two mechanical actuators (PZT) were used to generate elastic waves: (1) at the base of the cantilever probe at a frequency ${{f}_{|1p\,0s>}}$ (in the custom fluid cell) and (2) at the base of the sample at a frequency ${{f}_{|0p\,1s>}}$. Through this excitation scheme, a multi-harmonic forcing can be measured from the cantilever motion due to the forcing on the probe and at ${{f}_{|1p\,0s>}}$ on the sample at ${{f}_{|0p\,1s>}}$. The liquid interaction increases the damping and spurious resonant modes and a reduced sensitivity. By monitoring the behavior of the acoustic response S(t) (figure 1) with respect to the amplitude and phase of the resulting frequency mixing components, l-MSAFM renders an imaging signal, which is independent of possible spurious signals. As the tip of the probe and the sample enter the contact regime, a difference and sum frequency generation takes place in the tip-sample system as a result of the nonlinear nature of their interaction. This process can be viewed as the mechanical counterpart of the parametric process of sum and difference frequency generation in nonlinear optics. The resulting dynamical signal S(t) is then analyzed by lock-in detection using the newly formed difference frequency mode ${{f}_{|-1p\,1s>}}$ as reference (see figure 1). Following the notation introduced in the case of tip–sample interaction in air [21], we denote each Fourier component of the mechanical oscillations of the coupled probe-sample system with $|{{n}_{q}},{{m}_{q}}>$. The effect of liquid interactions is bypassed here due to the lock-in amplifier measurements.

Zoom In Zoom Out Reset image size

2.1. Electronic apparatus

2.2. Frequency mixing in l-MSAFM

As reported previously for the multi-frequency AFM [21], the partial differential equation of the probe, considered as a mechanical beam, can be expressed in the form:

$$\begin{eqnarray*}&&\begin{array}{lllllllllllllll} \rho \partial _{t}^{(2)}\left[ u(x,t) \right]+\rho \partial _{t}^{(2)}\left[ h(t) \right]+EI\partial _{x}^{(4)}\left[ u(x,t) \right] \\ \quad +c{{\partial }_{x}}{{\partial }_{t}}\left[ u(x,t) \right]=0, \\ \end{array}\end{eqnarray*}$$

where u(x,t) represents the transverse displacement of a point P of coordinate x on the cantilever at a given time t, r the density, E the Young's modulus, I the moment of inertia and c the structural damping. The boundary conditions of the system require:

$$\begin{eqnarray*}&&u(0,t)={{\partial }_{x}}\left[ u(0,t) \right]=\partial _{x}^{(2)}\left[ u(0,t) \right],\end{eqnarray*}$$

and

$$\begin{eqnarray*}&&\begin{array}{lllllllllllllll} {{m}_{e}}\partial _{t}^{(2)}\left[ u(L,t) \right]+{{m}_{e}}\partial _{t}^{(2)}\left[ h(t) \right]-EI\partial _{x}^{(4)}\left[ u(L,t) \right] \\ \quad =\Gamma (t),\forall t \\ \end{array}\end{eqnarray*}$$

where ${{m}_{e}}$ in the tip-mass, and $L$ the cantilever length.

In the case of l-MSAFM, the interaction force $\Gamma (t)$ is time dependent. Assuming a harmonic dynamics of the sample (at ${{f}_{|0{\rm p}\,1{\rm s}>}}$), the tip-sample distance $d(t)$ becomes modulated in time, and a spatial component of the interaction should be considered $\Gamma (y,t)$. By means of analytical calculations, we have previously demonstrated that by imposing a forcing on the probe and a forcing on the sample, the modulation of $d(t)$ engenders a discontinuous response of the system, which could be observed in the amplitude of the difference mode ${{a}_{|-1{\rm p}\,1{\rm s}>}}$. We consider here the optimized case of the force, corresponding to the volume integrated Lennard–Jones potential.

2.3. Fluid cell for acoustic imaging

2.4. Cell culture

2.5. Cytotoxicity assays

Figures 2 and 3 show a sequence of the resulting l-MSAFM phase maps ${{\Phi }_{|-1{\rm p}\,1{\rm s}>}}$ obtained on epidermal cells, in their physiological conditions, before and after oxidative treatments. Series of measurements on control cells are presented in figure 2 and illustrate the subsurface capability of l-MSAFM. The large size of the cell (up to 40 μm laterally, and up to 3 μm in height as seen in figures 2(a), (b)), and subsequent displacements of the actuators and the piezoelectric tube occasionally induce the presence of irregular kinks on the phase response of the l-MSAFM image. As the observed disturbances can affect tip–sample interaction, we decreased the size of the scanned area (figure 2(d)), to focus on the area of the cell including the nucleus. For the first time, the images obtained using l-MSAFM reveal details of the cytoskeleton and inner structures of the cell that could not be resolved with the common state of AFM. We focused here on the phase lag signal, which has been linked to mechanical properties in similar technique such as tapping mode and amplitude modulation mode imaging [22]. The phase lag in the AFM tip–sample interaction is a measure of the energy dissipation, which is commonly linked to the samples mechanical properties [30, 31]. In figure 2, we highlight the unique subsurface mapping capability of l-MSAFM in a composite set of phase response representative of eight different frequency component of the tip–sample interaction. The l-MSAFM phase maps presented in figures 2(e)–(l) are associated with the following generated difference frequencies ${{f}_{|-1p\,1s>}}$: 35 (l), 37 (e), 38 (f), 39 (g), 40 (h), 45 (i), 50 (j) and 55 kHz (k). The nucleus, which was not observable in the topography images figures 2(c), (d), can clearly be observed (circled in light blue) in figures 2(e)–(g), (i), (k), with an exhibited diameter close to 5 μm. Strikingly, the l-MSAFM images expose the presence of filamentous structures, a few of which are indicated by the green arrows in figures 2(e)–(g), (i), (l). Measurements on the filaments indicate a length of 4 μm and width of 0.5 μm and further analysis suggests that the structures could potentially be mitochondria, although only localized chemical analysis could validate such a claim. The entities observed with diameters ranging from 500 nm to 1 μm (red circles in figure 2(i)) present the characteristics of vesicles. While the exact mechanism behind the low signal associated with the observed vesicles may not be directly available from the measured contrast, it may be linked to scattering events as in the 'absorption' of signal by the cavities, analogous to an echo effect. Note that an inversion of contrast occurs between ${{f}_{|-1p\,1s>}}$ = 40 kHz and 45 kHz [32], as can be seen from the inversed scale bars in figures 2(i)–(l), which can be explained by one of the resonances of the cantilever at 40 kHz. Due to the lack of complementary tool for high resolution subsurface imaging of cells in liquid, the interpretation of l-MSAFM images is based on: (1) the morphology of the cell that is available from AFM images, and (2) fluorescence imaging (figures 2(b) and 3(b)) of the same cells, in culture, with special staining to visualize the actin filaments (red) and the nucleus (blue). A more ideal control that would consist of imaging the same cell by AFM and fluorescence on a single platform is not yet possible due to the early stage of developments of l-MSAFM. However, the correlation between the l-MSAFM images and the changes in actin networks can be appreciated in figures 2 and 3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The effect of the pesticide on the HaCaT cell line was later investigated using l-MSAFM, under the same buffer conditions. As seen in figures 3(a)–(c), the oxidative stress caused by the glyphosate is clearly observable from the morphological damages and the inflation of the cells. In fact, from single profile measurements on the control cell (figure 2(c)) and the stressed cell (figure 3(c)), we observed an increase in height from 1.5 μm for the control cell to 2.1 μm for the stressed cell. Interestingly, the lateral dimensions rose from 12.7 × 23.1 μm² to 19.5 × 22.2 μm² as a consequence of the stress, causing an approximated two fold global volume expansion, i.e. 462 μm³ for the control cell and 906 μm³ for the stressed cell, which was the trend observed throughout the cell culture. The volume variation was estimated numerically by assuming an ellipsoidal form for both control and stressed cells. The cross sections of a control and treated cell toward of the length and the width are illustrated on the figure 4.

Zoom In Zoom Out Reset image size

Furthermore, a redistribution of the protrusions on the cell surfaces could also be observed from the AFM topography images shown in figures 2(c) and 3(c). Analyzing a 20 × 20 μm² area centered at the highest point of each cell (see figures 2(d), 3(d)) reveals that the control cell exhibits greater number of protrusions than the stressed cell, which presents a smoother surface, with a roughness R_a  = 501 nm for the control cell and R_a  = 429 nm for the treated cell (see figures 2(c), 3(c)). The curves plotted show also the difference of roughness of each cell, less important on the stressed cell (figures 4(c) and (d)), due to the infection corresponding to a volume expansion. These observations of protrusions redistribution correlate well with previous results [23].

After exposing the cells to glyphosate, none of the inner features of the cells resolved in figure 2 could be observed as can be seen in figures 3(d), (e), (g), (h), despite systematic tuning of the frequency ${{f}_{|-1p\,1s>}}$ for phase imaging ${{\Phi }_{|-1p\,1s>}}$. However, the presence of large entities with lower phase suggests that vesicles have grown in size and number (dark areas in figures 3(d), (e), (g), (h)). These observations are in agreement with previous confocal microscopy studies of the same cells, carried out with green tubulin label to track the structure of the cells, which demonstrated a severe alteration of the inner cell conditions [25]. We determined that the mechanical properties of the disorganized structures (red stained actin structures observed in the confocal images in figure 3(b)) differ from the properties of their assembled state (observed in the confocal images in figure 2(b)) as the contrast in the l-MSAFM phase images exhibit very different features. The resulting l-MSAFM images in figure 3 indicate inner damages of the cells, and support the understanding of the modifications of cellular and mitochondrial functions. For low concentrations of glyphosate, as it is the case in this study, oxidative stress inducing an increased production of actin fibers in the filamentous network has already been previously reported [26]. Such observations corroborate our hypothesis of an increased stiffness and cell rigidity, linked to the reorganization of the actin filaments, as a protection mechanism of the HaCaT cells against chemical attacks [33]. As discussed by Heu et al [26], oxidative stress incurred by the exposure to glyphosate triggers some of the cell's defense mechanisms including reorganization of the actin filaments and subsequent increase in stiffness while the volume and morphology of the cell surface undergo important changes.

The introduction of subsurface imaging for live cell characterization holds great potential to bridge the gap between nanoscience and single cell analysis. Using the emerging multifrequency nanometrology in fluid l-MSAFM, we resolved for the first time the nanometer size inner structures of live HaCaT cells, in their physiological medium, without the need for staining or labeling. By using the spectral properties of the synthesized difference mode ${{f}_{|-1p\,1s>}}$ to acquire series of complementary phase maps, we showed that the l-MSAFM could resolve the cytoskeleton, the nucleus and mitochondrial morphology in live HaCaT. In addition, our quantification of the effect of chemically stimulated stress in terms of volume increase and variation in roughness provided the association with the redistribution of the protrusions on the cell membrane. We demonstrated that glyphosate has a significant impact on the inner structure of the epidermal cells. We envision that this capability will have a broad impact in cellular biology, nanomedecine, toxicology and environmental studies.

This work has been supported by the Labex ACTION project. Dr. Laurene Tetard would like to acknowledge partial support from ORNL's. Research performed as a Eugene P Wigner Fellow and staff member at the Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the US Department of Energy under Contract DE-AC05-00OR22725. Other authors report no other relationships or activities that could appear to have influenced the submitted work.