1. Introduction

Cell viability assays are used in laboratories around the world to determine the effectiveness of drugs and in the exploration of cell death mechanisms. Traditionally, cellular staining (or labelling) is used to visualise the morphological changes cells undergo during various physiological processes. Staining helps to improve the contrast of the sample under the microscope and allows physical properties of the cell, such as membrane integrity or protein trafficking, to be distinguished. However, there is growing concern about the use of stains and dyes and the potential changes that they may induce in cells. In particular, some studies have shown that their introduction alters the native function of cells which can, in turn, lead to incorrect conclusions being drawn from imaging data [1]. Consequently, the necessity for non-invasive, high contrast techniques that preserve regular functionality is becoming increasingly widely recognised. Such techniques, broadly termed ‘label free imaging’, seek to introduce contrast based on the exploitation of different native photophysical processes without impacting cellular function.

There are a broad number of techniques that meet this definition, from the measurement of polarisation sensitive scattering [2,3], to the detection of Raman spectra corresponding to specific chemical bonds [4]. One of the oldest and most widely used label free approaches is that of measuring the phase shift induced by the sample with respect to the incident light. Examples of this approach include differential interference contrast (DIC) and Zernike phase contrast [5]. There are many applications for quantitative phase measurements in the published literature which include cellular segmentation in tomographic imaging [6], quantification of contents and concentration of the dry mass in red blood cells [7], measurement of cell motility by tracking of intracellular organelles and membranes [8], and the production of diffusion coefficient maps tracking the dynamics of live cells [9].

In this paper we implement ptychography, a quantitative phase-sensitive technique that is able to retrieve the complex wavefront of the light exiting the sample [10–12]. Using ptychography both the amplitude and phase of the sample and the probe are obtained as separate images, resulting in high quality quantitative phase information [13–15]. Ptychography uses multiple diffraction patterns collected from spatially overlapping regions of the sample to form images using iterative algorithms [16]. These algorithms perform a refinement of the amplitude and phase estimate by utilising the diffracted intensities and knowledge of the real-space position of each overlapping point in the scan. The retrieval of the phase information from ptychography is of very high quality and has been successfully used in the X-ray regime to provide additional information such as chemical contrast [17] and mass density maps of cells [18]. In recent years, visible-light ptychography has become increasingly popular with applications including quantitative stress measurement in optically transparent samples [19], label-free imaging of cells [20,21], and as a means of distinguishing between healthy cells and cancer cells [22].

Additionally, the application of ptychography to time-resolved optical imaging is becoming increasingly popular as work to reduce the scan time is making the technique a viable means of phase imaging for a range of biologically relevant timescales. By using continuous sample movement [23,24] known as flyscanning, or by modifying the illumination source, such as by using a LED array [25] or using a shifted illumination [26], scan speed can be greatly increased, in some cases up to video frame-rate [27]. Therefore, although previous ptychographic imaging approaches have been fundamentally time-sensitive, particularly to sample motion between positions, advancements in the technique and in sample preparation methodologies now enable collection of high quality data from time-evolving samples.

Here, ptychography is explored as a means of determining the viability and time of death of Saccharomyces cerevisiae yeast cells in the presence of the plant defensin, NaD1 [28–30]. Moreover, by fixing the cells to the substrate, and tailoring the scan to comparable fluorescence studies of the system, we show that ptychography is able to capture the rearrangement of the cell during death, providing direct evidence of a multi-step death process previously hypothesised by Hayes et al. [29].

2. Methods and materials

2.1 Materials

All materials, unless otherwise stated, were purchased from Sigma Aldrich, including the poly(ethylenimine) (PEI, branched with an average molecular weight of 25,000) with CAS Number 9002-98-6. The yeast extract peptone dextrose (YPD, 1% yeast extract, 2% peptone and 2% dextrose in distilled water), SYTOX green dye, and the wild type Saccharomyces cerevisiae (strain BY4741) were provided by Dr. Mark Bleackley of La Trobe University. SLIDE GRND 90 PLAIN microscope slides (VWR) with dimensions 26 x 76 mm and thickness 1–1.2 mm, and square borosilicate glass cover slips with dimensions 22 x 22 mm and thickness No. 1, were used for all experiments.

2.2 Cell treatment

For cell growth, 3 ml of yeast extract peptone dextrose (YPD) medium was inoculated with a colony isolated from a fresh YPD agar plate and incubated overnight at 30°C with gentle agitation. The following day a fresh 2 ml YPD culture was prepared by inoculating 2 ml of YPD with 0.1 ml of overnight culture and incubated, with gentle agitation, for 2 hours at 30°C. A 1 mL aliquot of the cell suspension was centrifuged at 5,000 rpm for 5 minutes to pellet the cells. The cell pellet was washed with milliQ water and the sample was centrifuged as above. This process was repeated 3 times. The cell pellet was weighed, and for every 1 mg of pellet, 0.15 mL of millliQ water (pH adjusted to 9.2 with sodium hydroxide) was used to resuspend the cells.

To adhere the yeast to the glass slides, a 30 ${\mu }$L drop of 2% polyethylenimine (PEI) was deposited on a glass slide using a pipette and left to dry in a fume cupboard for 30 minutes. 20 ${\mu }$L of the yeast suspension was then added on top of the PEI and left to adhere for 10 minutes enclosed in a sterile petri-dish on the bench. The excess solution was drawn off with a pipette and the adhered cells were washed once with milliQ water to remove any cell debris. After washing, 20 ${\mu }$L of 1.25 ${\mu }$M SYTOX green solution was immediately added and sealed with a cover slip. Cell death was initiated with the addition of 2 ${\mu }$L of 8.54 ${\mu }$M NaD1 before sealing.

2.3 Fluorescence measurements

Cell viability assays were carried out on the yeast cells by detecting the fluorescence of the SYTOX green nucleic acid stain (Thermofisher) as when the cell dies it becomes permeable, allowing the fluorescent dye to penetrate the cell and a fluorescent signal to be detected. A Nikon Ti-Eclipse optical microscope operating in the FITC channel (495 – 519 nm) was used with a 10x objective (NA 0.3) to collect 1 second exposures over a selected field of view. Images of the alive cells were collected immediately after the addition of the SYTOX and images of the dead cells were collected 90 minutes after the addition of the NaD1 peptide.

2.4 Ptychography measurements

Ptychographic imaging was performed using the configuration described in previous publications [19,31] and shown in Fig. 1. Briefly, an expanded fibre coupled laser source (wavelength 660 nm) was cropped to produce a flat, circular beam with 5 mm diameter. The imaging probe was produced using a reversed 20x Mitutoyo microscope objective (NA = 0.42) with the sample placed 150 ${\mu }$m downstream from the focus and data collected on a sCMOS detector (Andor Zyla 4.2) placed 16 mm from the focus. The sample was moved in a Fermat spiral motion [32] with a step size of 20 ${\mu }$m and an average overlap of 75% between points.

 

Fig. 1. Ptychographic Imaging Configuration. a) A 5 mm beam ($\lambda$ = 660 nm) is incident on an objective lens with the sample placed at a defocused distance of 150 ${\mu }$m and the detector placed 16 mm from the focus. Selected reconstructions of the b) amplitude and c) phase of the recovered probe.

Download Full Size | PDF

Two types of data sets were collected. The first consisted of a large area scan containing 4,096 Fermat spiral points collected as four individual 1,024-point areas with 5% overlap between areas to ensure smooth continuity in the reconstructed images. This region (2.7 mm²) was used to image a large number of cells (>100) to confirm statistical differences between alive and dead cells with high confidence. The second data set was a time-resolved series. A smaller region consisting of a 60-point Fermat spiral (0.4 mm²) was performed. The smaller scan size allowed data to be collected approximately every 40 seconds. For the time-resolved imaging the first scans took place approximately 3 minutes after the addition of the NaD1. This delay arose from the time taken to mount the sample and find a region with sufficient cell density for imaging.

All scans were reconstructed using 100 iterations of the ptychographic engine with probe retrieval (ePIE) [33] implemented in Python 2.7 (Python Software Foundation, https://www.python.org/).

2.5 Confocal microscopy

S. cerevisiae cells were grown overnight in YPD and diluted to an OD$_{600}$ of 0.3 in half strength potato dextrose broth supplemented with 1.25 ${\mu }$M SYTOX green and then treated with 8.5 ${\mu }$M NaD1. The cells were examined using a Zeiss LSM 780 laser scanning confocal microscope. The cells were observed using the LDC apo 40x water immersion/1.1 Korr M27 objective. The cells were imaged with DIC and fluorescence. SYTOX excitation was at 488 nm with fluorescence detected at 530 nm.

3. Results

3.1 Cell viability

To determine the feasibility of ptychography for performing time-resolved cellular death measurements, two control samples were prepared. The first sample consisted of yeast cells, without the NaD1 peptide, where imaging was performed immediately following deposition of the cells onto the substrate to ensure the majority of cells were alive during the measurement. The second sample consisted of yeast cells mixed with the NaD1 peptide in which imaging was performed after one hour to ensure that the majority of cells were ‘dead’. Both samples were imaged using ptychography, before fluorescence measurements were taken to confirm cell viability using established protocols Fig. 2.

 

Fig. 2. Comparison between the characteristics of alive (a-c) and dead (d-f) cells. Reconstructed amplitude (a,d) and phase (b, e) using ptychography, and optical fluorescence measurements (green) overlaid on bright-field (greyscale) images (c, f), of S. cerevisiae yeast cells. g) Boxplot of the average phase for both the alive and dead cell samples.

Download Full Size | PDF

A total of 273 dead and 248 alive cells were imaged across the two samples with cell viability confirmed via fluorescence with the absence of any fluorescent signal indicating that the cells are still alive. For the ‘alive’sample (ie. no NaD1), fluorescence was observed in 39% of the cells, indicating that although the majority of cells were alive, some cells had died during the deposition and fixation process. For the dead sample, where the peptide was added to induce death, fluorescence was observed in 71% of the cells confirming the majority of cells were dead.

The reconstructed phase shift of the light propagating through the cells, reconstructed using ptychography, was obtained for both the alive and dead cells, with the mean phase shift calculated for each individual cell. The two control samples were compared by taking the average phase shift for all confirmed living and dead cells. The alive cells were calculated to have an average phase of 1.33 $\pm$ 0.36 radians, compared to the dead cells which had an average phase of 1.67 $\pm$ 0.54 radians. The error provided is the standard deviation of average phase values for the selected cells.

In addition to the calculated standard deviation, a box plot is provided showing the statistical separation of the average phases where the median is identified as well as the interquartile range (IQR) (Fig. 2(g)). The plot shows overlap between the IQR for the two cell groups, however, the average and the distribution of values is significantly different, enabling a clear distinction. The larger variance in values for the dead cells, compared to the living cells, suggests a death process where individual cell death occurs at different times and varies between cells.

3.2 Response of the fluorophore

The SYTOX green fluorophore is commonly used to study cell death. It does not cross intact membranes but will easily penetrate compromised membranes (as in the case of dead cells) and bind to nucleic acids within the cell resulting in fluorescence. To study the response of the fluorophore to cell death, and to determine the efficacy of label-free approaches, high-resolution timed fluorescence measurements were performed on 124 yeast cells in the presence of NaD1 and SYTOX.

Figure 3(a) shows a time-resolved fluorescence study on the uptake of the SYTOX green fluorescence peptide for yeast cells. The earliest fluorescence, from a very small number of cells (3.2%), was detected at 11 minutes, with significant number of cells fluorescing (14.5%) at 18 minutes. The fluorescence signal steadily increases and at 60 minutes most cells (65.3%) show fluorescence. Of the fluorescing cells identified at 60 minutes, 70.4% show a saturated signal over the entirety of the cell, indicating a complete permeabilisation of the membrane due to cell death. Two individual cells selected in Fig. 3(b) highlight the distinction between individual cell fluorescence and cell death. For these cells the first fluorescence is observed at approximately 32 minutes after the SYTOX has been added to the media, where the cell membrane is initially compromised by NaD1 allowing the SYTOX to enter the cell. Between 32 and 43 minutes, a slow uptake of the peptide and/or SYTOX occurs within the cell before saturation is observed at 51 minutes, indicating cell death. The fluorescent signal alone is therefore not a clear measure of cell viability as there is ambiguity between when the cell starts to die (32 minutes) and when the cell is completely dead (51 minutes). At 11 minutes after the addition of SYTOX a few cells show early signs of death and by 18 minutes only 1.6% of cells were observed to have saturated fluorescence (14.5%). Therefore cell death is assumed to be initiated between 11 and 18 minutes with the majority of cell death occurring within 60 minutes.

 

Fig. 3. Confocal fluorescence measurements showing the uptake of SYTOX green by S. cerevisiae cells in the presence of NaD1 over time. a) The distribution of the SYTOX dye, used as an indicator of membrane permeabilisation, is shown for a series of time points for a large number of cells. Shown are (top) Differential interference contrast microscopy (DIC), (middle) confocal microscopy, and (bottom) confocal overlaid on DIC. Inset: An example enlargement of two individual cells highlighting the SYTOX fluorescence distribution over time. Scale bar is 5 $\mu$m.

Download Full Size | PDF

We note that there is a higher confluence of cells in Fig. 3 compared to Fig. 2. To explore this effect, supporting experiments were performed measuring cell death as a function of confluence. The results from these tests revealed that confluence has a negligible effect on the results, which is addressed in the present work by adjusting the peptide concentration.

3.3 Time-resolved cell death

To establish the accuracy of ptychography to accurately determine the time of cell death, a region of cells was selected and their average phase calculated (Fig. 4). Individual scans took approximately 40 seconds and 60 cells were successfully imaged across two slides. One sample consisted of 27 cells, of which 52% died, while the other sample consisted of 43 cells, of which 55% died. Based on the cell viability measurements, the cells were determined to have died when the average phase shift of the cell at the end of one hour of scanning was greater than 20% of the initial phase. The confocal fluorescence imaging results are consistent with this period of cell death (Fig. 3).

 

Fig. 4. Ptychographically reconstructed phase distribution of S. cerevisiae yeast cells over time. Reconstructed phase at a) t=3 mins, b) t=20 mins, and c) t=54 mins. d) A plot of the average phase for each of the four cells shown in a-c. Note that the colour of the lines is matched to the corresponding box around each cell.

Download Full Size | PDF

The reconstructed phase from a region of cells is shown in Fig. 4 at three different times (t=3, 20, and 54 minutes). An increase in the phase characterised by the transition of the cell from low phase shift (blue) to high phase shift (yellow) in Figs. 4(a)–4(c), indicates cell death.

As an example, four cells that died during the imaging have been selected (indicated by the boxes) and their average phase at each time point calculated and plotted (Fig. 4(d)). In these scans cell death was observed between 5 and 25 minutes after the inclusion of the peptide. The starting phase shift (at t=3 minutes) is shown to be between 1.55 and 1.85 radians for the four cells. Between scans the phase shift varies slightly, but largely remains constant until each cell, at different times, experiences a dramatic increase in the reconstructed phase shift of between 1.8 and 2.3 radians. This process occurs over two ptychographic scans, equivalent to approximately two minutes, after which we observe the reconstructed phase to be static (or constant). The increase in phase shift is a clear indication that the cell has undergone a significant chemical or physical change, most likely due to membrane permeabilisation and cellular uptake of the surrounding media which occurs upon cell death. The advantage of this label free technique is it directly measures the state of the cell, while fluorescence is an indirect measure of cell death or cell physiology and may interfere with cellular function.

It is interesting that no two cells show the exact same fluorescence or phase profile. However, in the ptychography case we can link this directly to the mode of action of NaD1. Whereas in the fluorescence study it is uncertain if this variability is a result of the additional effect of the fluorophore. In the case of the cell, highlighted by the blue box in Fig. 4(a), the phase profile clearly reveals an intermediate step between 15 and 25 minutes, where there is a slight increase in phase. Here the phase remains constant for around 5 minutes, followed by a second increase in value. To explore this multi-stage phase increase, as well as to investigate the visual changes to the cells, this cell (marked with the blue box) was isolated from Fig. 4(a) and the phase was investigated at six different time points (Figs. 5(a)–5(f)). Additionally, the phase across the dotted white lines (Fig. 5(f)) is plotted for each scan to show the phase distribution along orthogonal directions (Figs. 5(g) and 5(h)).

 

Fig. 5. Phase distribution of an individual S. cerevisiae cell from Fig. 3(a). The evolution of the phase distribution is shown for the same cell at 6 different time points a) t=3 mins, b) t=10 mins, c) t=20 mins, d) t=30 mins, e) t=40 mins, and f) t=50 mins. The phase distribution along slices taken through the g) horizontal and h) vertical centre of the cell. The approximate point of cell death is indicated with an arrow in g).

Download Full Size | PDF

Figure 5 confirms that the phase shift of yeast cells increases during apoptosis, which is consistent by the work of Chen et al. [34] who identify a density increase during apoptosis. This is consistent with the migration of fluid into the yeast cell due to permeabilisation of the membrane wall. The first measurement (t=3 minutes) shows that the cell predominantly has a low phase value throughout the cell (Fig. 5(a)). Over time, the phase remains constant (Fig. 5(b)), until 20 minutes have elapsed when a region of larger phase shift begins to appear in the centre of the cell (interpreted as the initial stages of cell death) and becomes more intense reaching its maxima at the 30 minute (t=30) incubation time, at which point the cell is considered to be dead (Fig. 5(d)).

When the distribution is viewed along a specific region of the cell (Figs. 5(g) and 5(h)) this process can be seen more clearly. An increase in phase occurs after 10 minutes, before the value reaches its maximum at approximately 23 minutes. This is consistent with the average phase plotted in Fig. 4(d). Figure 5(h) highlights the phase distribution variation, here moving from the left-hand side of the cell towards the centre. It is prudent to note that the confocal fluorescent studies show a similar fluorescence distribution in Fig. 3(b), moving along the edge of the cell, before developing a more intense area in the centre of the cell. Although the fluorescence measurements appear to be less sensitive to the initial onset of cell death the overall concordance of the concentration of fluorescence and increase in the phase confirms that ptychography is a useful, highly sensitive, indicator of yeast cell necrosis.

4. Discussion

The development of label free microscopy techniques for the analysis of biological samples is of significant interest due to the potential for highly-sensitive imaging without interfering with the native cellular physiology. Ptychography is particularly attractive since it has the ability to quantitatively retrieve both the amplitude and the phase shift introduced by a sample as a contrast mechanism for cellular imaging. S. cerevisiae yeast cells and the peptide NaD1 were used in this study to determine the feasibility of ptychography as a cell viability indicator. Cellular imaging was conducted using two types of scan; large area ptychographic imaging with comparative fluorescence imaging, and small area time-resolved ptychographic imaging over the course of one hour.

The large area scans confirmed, using a statistically significant set of data, that ptychography provides a valid means of determining cell viability. A clear distinction in the average phase value was observed for living and dead cells which obviates the need for staining.

Similar to fluorescent studies, ptychography can identify different rates of cell death depending on the chosen cell. In ptychography, whilst cell death began at different times, all cells showed a similar evolution of phase shift throughout apoptosis, with a sudden increase in the phase between each scan just prior to cell death. Therefore, using ptychography studies we can detect cell death as early as two to five minutes after the addition of NaD1 in some cells, making it more sensitive then the corresponding fluorescent studies (11-18 minutes) and implying that ptychography provides a more accurate measure of the onset of cell death. By comparison, the phase signal from ptychography is a direct measurement of the sample changes in response to NaD1, with sensitivity to the membrane disruption resulting in a quicker response.

Due to the time it takes for a sufficient amount of fluorescent signal to build up inside the cell there is thus a significant delay between the initial point of cell death and the detection of fluorescence under the microscope. For example, 18 minutes after the addition of SYTOX only 14.5% of cells which were later confirmed to have died were detected. This may be contrasted with ptychography whereby 20 minutes all of the cells which had died had been identified via an optical phase shift.

Additionally, a multi-step process was observed in our ptychography measurements where some cells undergo a smaller increase in phase shift which stabilised before a second increase would occur, indicating cell death. This multi-stage mechanism has been previously reported in the literature for NaD1 by VanDerWeerden et al. [28,29,35], where cell membranes were shown to semi-permeabilise without cell death immediately occurring. This is consistent with the results shown here, wherein the cell initially increases in phase, which may be an indication of NaD1 starting to enter the yeast cell, before the phase reaches a maximum and cell death is observed. This is the first time that a direct observation of this process, without labelling, has been possible and serves to further develop the model of NaD1’s mechanism of action.

Although these results indicate the benefits of quantitative phase imaging to explore cell death, the current achievable temporal resolution is a constraint. A more detailed understanding of the death process would require a time sensitivity which is less, or comparable, to that of the process being measured. Unfortunately, due to constraints with our current ptychographic microscope, this was not possible. However, there are many published results that highlight the feasibility of quantitative phase imaging with ptychography for high quality, high speed measurements [23–27]. In addition, the simultaneous acquisition of both fluorescence and quantitative phase information from ptychography could potentially yield improvements in both the location and timing of cell death. This will be considered as part of the future development of the ptychographic microscope.

Even though there is clear evidence relating the phase change to cell death, the physical origin of the phase shift is open to conjecture. In our attempts to understand this process we consider a number of different physical and chemical changes occurring within the cell. These changes can introduce a delay in the optical wavefield passing through the sample, and hence introduce a phase shift. There are a number of known physical and chemical changes within a yeast cell during endocytosis reported in the literature [28]. The activation mechanism of NaD1, or the permeabilisation of the cell membrane, is primarily a physical change by which the tension in the membrane is disrupted. Hence, this can be perceived as the initial phase change we detect since our measurements are sensitive to mechanical changes within the cell. The fluorescence results confirm this. In both cases signal first appears along one edge of the cell before progressing towards, and increasing in, the centre of the cell. Fluorescence from the SYTOX green occurs when it interacts with nucleic acids within the cell and is an indicator of membrane permeabilisation. The hypothesis is therefore that the initial phase change reflects the physical changes within the cell, directly related to the membrane permeabilisation. Secondly, a measurable chemical shift where there is transport of cytoplasmic contents between the cell and surrounding media has also been identified in the yeast cells which is predicted to occur once the NaD1 has entered the cells [35,36]. The transport of metal ions from within the vacuole to the cell, would likely result in a change in the complex refractive index and hence a phase change [17], which could explain the second phase shift identified in some cells before death. However, this process is not the only change happening within the vacuole as it is also likely undergoing a physical change and is only one possible explanation of what may be happening within the cell [37]. A detailed interpretation of the phase change in relation to the physical or chemical changes within the cell, however, still requires further understanding of the processes involved with greater temporal sensitivity.

Our results show, for the first time, how phase information with temporal sensitivity can be used to observe variations in cell structure, not only for cell viability measurements but as a source of additional information about the processes taking place during cell death. The phase sensitivity of ptychography allows for a label free method of detecting the initial onset of cell death within as little as 2 minutes and provides further insight into the cell death process of S. cerevisiae cells when exposed to the NaD1 antimicrobial peptide.

5. Conclusions

We have demonstrated, for the first time, cell viability measurements of S. cerevisiae cells to the NaD1 peptide using quantitative phase imaging via ptychography, whereby dead cells were observed to have a higher phase shift than living cells. Additionally, time-resolved measurements identified trends in the phase distribution within the cells during death that support the previously published hypothesis of a two-stage cell death mechanism. Comparison of ptychographic and fluorescent images demonstrate the sensitivity of ptychography as a technique capable of detecting membrane permeabilisation earlier than the corresponding fluorescent studies. Ptychography also offers the significant advantage that it does not directly influence the native cellular function.