1. NMA Tool Development

In order to develop the tool we used nuclei from the glioblastoma cell line U87 treated with temozolomide (TMZ), which causes DNA damage and G₂ cell cycle arrest, in combination with resveratrol, which blocks this arrest. This combination induced cells to undergo mitosis with increased expression of cyclin B1 and decreased pCDC2(Y15) levels, despite wide DNA damage, as indicated by γH2AX and comet assay (data not shown), thus generating a multitude of nuclear morphology. Nuclei were stained with DAPI and pictures were taken on a fluorescent inverted microscope. Images were analyzed with the Image Pro Plus 6.0 (IPP6) software and 46 parameters of nuclear size, shape and marking were produced for a large population of cells. These parameters were analyzed with principal component analysis (PCA) and the features Aspect (Asp), Area box (Arbx), Radius ratio (Rr) and Roundness (Rou) were combined in an index called Nuclear Irregularity Index (NII = Asp−Arbx+Rr+Rou). Details on the development of the tool are given in the Text S1 file.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Distribution of nuclei in a plot of area versus NII.

N: Normal nuclei (crosses represent nuclei used to establish the reference population and the ellipse that represents the conjoint distribution for area and NII for normal nuclei); I: irregular, LR: Large Regular; LI: Large Irregular; SR: Small and Regular; S: Small; SI: Small and Irregular. Squares denote the averages of the different populations. Pictures (a–r) show examples of nuclei and their localization in the graph.

https://doi.org/10.1371/journal.pone.0042522.g001

In order to permit a wider use of the NMA tool, we developed an Image J plugin (for Image J 1.45 free version) that allows the quantification of the five measurements required for NMA (i.e. those for calculation of NII plus area measurement). A detailed explanation of the plugin features and its use is provided in the Text S2 file. A comparative analysis of data generated with the IPP6 software and the Image J plugin that accompanies the NMA tool showed very similar percentages of nuclei for each population (Text S1), validating the plugin developed for the NMA analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Morphological appearances and putative biological meanings of the different NMA populations presented in Figure 1.

https://doi.org/10.1371/journal.pone.0042522.t001

Data generated with the IPP6 or Image J software were then analyzed in a spreadsheet (NMA.xls or NMA.ods). Data from normal, untreated cells (in which small, large and irregular nuclei are excluded) are used to set the parameters of the normal population. The user can define the number of standard deviations (SD) from the mean that will be considered an acceptable deviation. Next, a large set of data from different treatments are used to set the thresholds of the different populations. These settings (number of SD and the thresholds) are used to analyze the nuclei from different treatments. After entering the data from the nuclei in the data analysis sheets, a plot of area versus NII will be automatically produced (Figure 1), showing the ellipse representing the normal nuclei and the different populations whose thresholds were set in the “Normal Nuclei and Settings” sheet. Detailed use of the spreadsheets is given in the Text S2 file. The Image J plugin, Microsoft Excel® and Open Office Calc files are available at http://www.ufrgs.br/labsinal/nma/.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. NMA of senescent glioma cells.

C6 glioma cells were treated with DMSO as vehicle control (A) or resveratrol (Rsv) and quercetin (Quer) for 12 days (B) for senescence induction [6]. Light blue diamonds are nuclei from β-galactosidase positive cells. Numbers denote β-gal positive and total nuclei in the different categories and the percentage of nuclei in each category. Inserts show examples of nuclear and β-gal staining (a–g positive and h-j negative).

https://doi.org/10.1371/journal.pone.0042522.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. NMA ofcells treated with vincristine.

U87 cells were treated with DMSO as vehicle control (A) or vincristine 50 nM for 24 h (B). Numbers in boxes denote the percentage of nuclei in each category. Inserts show examples of representative nuclei. Images were obtained directly from IPP6 software (with the boundary marking and the number of nuclei in the photo), to exemplify the marking established by the software. All nuclei are from single cells, as defined by direct visualization.

https://doi.org/10.1371/journal.pone.0042522.g003

2. Biological Significance of the different Populations in the NMA Plot

Analyzing the data produced with the nuclear population used to set up the tool we noticed a distribution in four separate populations: normal, large, irregular and small (Text S2). Analysis of a larger set of conditions, including cells treated with classical inducers of senescence, apoptosis and nuclear irregularities, lead to the suggestion of seven different categories of nuclei based on size and NII (Figure 1, Table 1 and Text S1).

Normal nuclei of adhered cells in culture vary in size, but have a very regular shape (Figure 1.h,i,j), with some located at the borderline between N and I (Figure 1.k) and analysis of these nuclei are important for the precise setting of the threshold that separates these populations. Nuclei classified as I, (Figure 1.a,b,c,l,n,r) present high irregularity. Among the large nuclei, we found a clear separation between two groups in three experimental settings (Figure 2, Figure 3 and Figure S1 and S2). On the lower NII range are the well-defined and regular nuclear shapes (Figure 1.e,g), whereas at higher NII range are the nuclei presenting morphological alterations (Figure 1.d,f), which may be undergoing MC.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. NMA of cells in apoptosis.

U87 cells were treated with DMSO as vehicle control (A) (two representative overlaid images of nuclei and annexin are shown - both cells are negative for annexin) or Cisplatin (16.6 µM) (B) for 24 h, a well-established protocol of apoptosis induction [17] with three overlaid images of nuclei (blue staining) and annexin (green staining). Green diamonds on the graph represent annexin-positive cells.

https://doi.org/10.1371/journal.pone.0042522.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Comparison of NMA and annexin apoptosis quantification.

(A) HeLa cells were pre-treated (1 h) with the caspases inhibitor zAsp-CH2-DCB (100 µM), followed by treatment with Cisplatin (40 µM), DAPI staining and NMA analysis. Top: NMA graphs; Middle: phase contrast; Bottom: DAPI fluorescence. Arrows point to cells with apoptotic phenotype. (B) HeLa cells were treated as in B, followed by annexin V-FITC staining and flow cytometry. Left: Flow cytometry plots. Green Fluorescence: annexin V; Red Fluorescence: PI. Region enclosed in red: Annexin-positive, PI-negative cells (i.e. apoptotic cells). Right: comparative percentage of apoptotic cells measured through NMA or annexin analysis.

https://doi.org/10.1371/journal.pone.0042522.g005

Nuclei undergoing mitosis have higher values of NII than normal nuclei because of their elliptic shapes, and a broader range of NII due to the variations in the shapes of the mitotic plate (Figure 1.m,p). However, some small nuclei also present very high irregularity (Figure 1.q), being thus classified as small and irregular (SI). It is also important to stress that in the NMA tool, each mitotic cell will be counted as one before anaphase (Figure 1.o), while after the anaphase the NMA tool will count two nuclei, which are normally located close to each other in the graph (Figure 1.m). Therefore, the NMA is not a good tool for counting mitotic cells, since at anaphase the two nuclei counted will still be part of a single cell, and will be evaluated as two nuclei. Notwithstanding, it provides a broad assessment of mitotic proportion in a cell population, as illustrated by the almost complete absence of mitotic nuclei in the senescent population analyzed in Figure 2.

3. Senescent Nuclear Enlargement in the NMA Plot

For the validation of senescence, we used the rat glioma cell line C6 treated with the combination of resveratrol and quercetin, which we have previously shown to induce high levels of senescence [5]. About 90% (35 of 39) of the large nuclei(LR+LI) were β-galactosidase positive (blue diamonds and examples of positive cells in inserts a–g), whereas only 4 out of 36 nuclei that localized in the normal population were β-gal positive (Figure 2– bottom graph). Additionally, in this population, only 2 mitosis were found in around 90 cells (around 2.2%), whereas in the control population this ratio was around 14% (Figure 2– top graph). These data support the usefulness of the NMA tool in predicting the senescent status of a cell population, with the advantage of being much less subjective than the current standard method of β-gal staining and visual identification of the positive cells.

4. Nuclear Irregularities in the NMA Plot

To test the usefulness of the NMA tool to detect nuclear irregularities, we measured nuclei of a population of U87 glioma cells treated with vincristine (50 nM), which is described to induce several nuclear phenotypes, including MC and/or senescence [11], [12], depending on the cell type and treatment. We observed that the treatment induced an increase of irregular nuclei (I and LI), from 7% in control to 41% (12%+29%) (Figure 3). Moreover, no nuclei were observed in the region were normal mitotic nuclei localized in the NMA graph.

In order to test the applicability of the NMA tool in other cell types, we treated the colorectal carcinoma cell line HCT116 with vincristine 50 nM. As expected, vincristine increased the percentage of both LI and LR nuclei. Average values of area of the LI and LR populations also increased, as did the NII of the LI population of nuclei treated with vincristine when compared to control (Figure S1). It is important to note that induction of senescence-like phenotype in cancer cells usually is accompanied by MC [13]. Thus, our tool allows a dynamic analysis of the development of both events over time.

We have also performed an analysis of data from published papers of Shao et al [14] and Vakifahmetoglu et al [1], which presented enough control nuclei to set the standards for the studied cell line as well as pictures of nuclei undergoing MC that could be measured. Despite the small number of images for control nuclei, NMA provided a well-defined distribution for the control nuclei and a broad distribution of the catastrophic nuclei, mainly in the LI group. All treated nuclei localized in the I or LI regions of the NMA plot. This is expected, since the illustrative pictures probably were selected to clearly show the catastrophic nature of the nuclei (Figure S2).

5. Apoptotic Nuclear Condensation in the NMA Plot

Cells undergoing apoptosis suffer a high and regular condensation of the nucleus [15] which in culture occurs prior to nuclear fragmentation and cell detachment but almost concomitantly with the externalization of phophatidylserine and formation of blebbings in the cell membrane. Due to this high condensation in a near spherical form, we hypothesized that nuclei of cells undergoing apoptosis may appear as small and regular (SR), i. e. on the lower left corner of the NMA plot. In order to validate this hypothesis, we evaluated the induction of apoptosis in U87 cells treated with cisplatin the presence of zAsp, a caspases 3/7 inhibitor. Apoptosis was confirmed by annexin V-FITC staining. Fourteen out of 29 nuclei located in the SR were annexin V positive as visualized with a fluorescence microscope and none of the annexin V-positive cell was located in any region outside SR (Figure 4B), whereas no cell was annexin V-positive in the vehicle-treated (Figure 4A).

A side by side analysis with NMA and annexin analysis through flow cytometry presented a very good correlation (Figure 5B). Inhibition of apoptosis with a caspases inhibitor reduced the apoptotic population detected with NMA to 7% and annexin to 10% in comparison to 49 and 57%, respectively, in the absence of zAsp (Figure 5A and B).

Additionally, NMA was also able to detect etoposide-induced apoptosis in HeLa cells, as well as differentiate apoptotic and mitotic cells (Figure S3), which normally appear in the S region of the NMA graph. Precise separation of S and SR regions may difficult the quantitative analysis of mitotic cells using NMA and additional methods should be employed to confirm mitotic cells. We also performed a kinetic analysis of Hoechst 3342-stained nuclei of HeLa cells treated with cisplatin. NMA graphs with the dynamics of area and NII, together with visible images of the cells undergoing apoptosis support the notion that nuclei located in the SR region of the NMA graph are apoptotic independent of the size and shape of the nuclei before apoptosis (Figure S4).

1. Software Information

Protocol uses Image Pro Plus 6.0 (IPP6 - Media Cybernetics, Bethesda, MD, USA), Image J 14.45 (NHI, Bethesda, MD, USA; http://rsbweb.nih.gov/ij/), PASW Statistics 18.0 (formerly known as SPSS - http://www.spss.com/software/statistics/), Microsoft Excel 2003 or 2007 (Microsoft, Redmond, WA, USA) and Open Office Calc (www.openoffice.org). NMA plugin, spreadsheets and sample images are available at http://www.ufrgs.br/labsinal/nma/.

2. Procedure

Brief instructions for using the NMA procedure (see Text S2 file for detailed instructions) are:

1. Take DAPI images (with at least 300 dpi; preferentially.TIFF format) of at least 100 nuclei in each condition of experiment; do not take images of fields in high confluence;
2. Open an image on IPP6 or Image J (containing the plugin of NMA) and correctly mark the surrounding of nuclei to assess morphometric data to NMA, i.e. area, aspect, area box, radius ratio and roundness;
3. After obtaining raw data, choose a group of nuclei from the control condition and exclude nuclei in mitosis or with clear abnormalities. Use the data from these nuclei to set the parameters for the normal population. Paste data from these nuclei in the spreadsheet named “Normal Nuclei and Settings”, from NMA file from Excel or Open Office; choose the number of SD that best defines the ‘normal ellipse’; if using Excel, don’t forget to erase the formula in blank cells of the NII column;
4. Paste data from all conditions - control and for each treatment - into the columns named “Treated Nuclei (for setting)”, in the “Normal Nuclei and Settings” spreadsheet of NMA file from Excel or Open Office; thereafter, choose the number of SD to define the thresholds that better separate the populations;
5. Create as many spreadsheets as needed (one to each condition) and paste data from analyzed nuclei into them;
6. Analyze the percentage of nuclei in each population as well as the averages of the populations.

3. Cell Lines

U87, C6, HeLa and HCT116 cell lines were purchased from ATCC (American Type Cell Collection, ATCC, Rockville, MD) and were maintained in DMEM supplemented with 10% FCS in 5% of CO₂ at 37°C.

4. Annexin Staining

Briefly, cells were pre-incubated with zAsp or vehicle for 1 h, followed by cisplatin treatment for 24 h. After this, for flow cytometry analysis, supernatant and cells were harvested, washed once with PBS 1x and once with annexin binding buffer, and incubated with a solution containing 6 µM of PI and 0.5 µL per sample of annexin V-FITC (Sigma Chemical; St. Louis, MO, USA) for 30 min, as indicated by the manufacturer, followed by flow cytometry. For microscopy, cells were washed once with PBS 1x and once with annexin binding buffer, and incubated with a solution containing 6 µM of PI and 1 µL of annexin per sample for 30 min, as indicated by the manufacturer. After, cells were fixed with paraformaldehyde 2% in PBS, stained with DAPI and images were acquired and overlaid.