1. Introduction
Cancer cells usually display a hypermetabolic fingerprint that can be sustained even in the presence of low/normal oxygen supply by adopting a glycolytic nonoxidizing metabolic phenotype [
1]. Reduced dependence on oxygen is not only necessary for activating the so-called Warburg effect, but also has a pivotal role in decreasing the production of reactive oxidant species (ROS), which can be potentially harmful to cancer cell viability. While hypoxia has been extensively studied in cancer, only recent literature has investigated the possible role of hyperoxia in tumor necrosis [
2]. Hyperoxia can in fact be deleterious to cancer cells in different ways. Some authors [
3,
4] have demonstrated that apoptosis can result from hyperoxia on solid tumors, while others have shown cell cycle blocking in carcinoma cells [
5] or anticancer immune-surveillance due to T cell and NK cell stimulation [
6]. Hyperoxia increases the level of ROS inside the cell and has a direct effect on both the cell cycle and cell viability. ROS can induce heavy damage to DNA, thus increasing cell cycle length because of multiple repair needs [
7].
Hypoxia-inducible factors (HIF-1 or perhaps more so HIF-2) [
8,
9,
10,
11,
12] are implicated in the mechanisms of cell apoptosis, both inhibiting [
13] and fostering [
14,
15,
16,
17,
18,
19] them. It is noteworthy that HIF has been shown to be under-expressed during stable hyperoxia and over-expressed just after return to normoxia in noncancer (HUVEC) cells, confirming the fact that transient hyperoxia is perceived as a hypoxic stimulus [
20]. This is an unexpected finding as HIF has been extensively studied only under hypoxic conditions [
21].
It is tempting to speculate that inducing changes in HIF expression by hyperoxia could lead to biological changes in cells with a high metabolic demand.
It has been shown that normobaric oxygen significantly increases endogenous erythropoietin (EPO) production 10, 12, 24, and 36 h after oxygen administration [
22], hence fostering hemoglobin increase in healthy volunteers [
23]. Latency time seems consistent with the time necessary to transcribe, translate, produce, and secrete erythropoietin. Indeed, erythropoietin (EPO) and vascular endothelial growth factor (VEGF) production are under the control of the HIF transcription factor [
24,
25]. Combined with chemotherapy, normobaric oxygen has recently been shown to reduce the tumor load and number in mice with lung cancer [
26]. The mechanism by which hyperoxia could interfere with cell survival lies deep in the fundamental cellular mechanisms of adaptation to hypoxia as proposed in
Figure 1 [
27]. However, being harmful inside cells is not the only way by which hyperoxia could be deleterious to cancer cells. Indeed, another way that hyperoxia could have antitumor activity is by inducing numerous modifications in the adenosine pathway, activating anticancer effects of T cells and NK cells [
6,
28]. Hyperoxia is also known to mobilize stem progenitor cells and change cytokine expression [
29].
The purpose of the present study was to investigate the role of transient hyperoxia on the outcome of nonsolid tumor cells such as leukemia line cells. This was performed using a biological marker (caspase-3), a morphological microscopic analysis, and a nonlinear fractal dimension calculation.
2. Material and Methods
2.1. Cell Lines
CCRF-SB and Jurkat leukemia cell lines (Sigma-Aldrich, St Louis, MO, USA) were maintained in RPMI 1640 containing 10% FCS, 100µg/mL Penicillin, and 100µg/mL Streptomycin in a fully humidified incubator at a concentration of 0.3 × 106 cells/mL under standard conditions of 21% O2, 5% CO2, and 37 °C.
For incubation in an increased oxygen environment, an oxy-concentrator was used (Generator 6000, b-Cat, Tiel, The Netherlands). Cells were incubated in 60% O2, 5% CO2, and 37 °C for 18h and returned to normoxia for the following incubation. Cells were sampled from cultures at various times of incubation for apoptosis and cell cycle determination. In selected experiments, hyperoxia incubation was repeated. The medium was kept unchanged.
2.2. Protein Extraction and Western Blot
For the Western Blot analysis, cells were centrifuged at 1200 rpm for 5 min. The pelleted cells were washed twice in PBS and then lysed (Gibco Lysis Buffer, with the addition of Antiprotease/Antiphosphatase) for 10 min on ice.
After centrifugation at 13,000 rpm for 10 min, the supernatant (protein extract) was immediately frozen to −80 °C. For the assessment of the expression of Bcl-xL protein, protein concentrations were determined, and equal total protein aliquots of 20 µg/10 µL were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred by Western blotting. After blocking, the membranes were incubated with commercially available primary Bcl-xL antibody (Cell Signaling Technology, Danvers, MA, USA). The primary antibody was detected using horseradish peroxidase conjugated secondary antibodies (Cell Signaling Technology, Danvers, MA, USA) and the membranes were subjected to chemoluminescence using the SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific, Dreieich, DE, USA). Exposed films were scanned and intensity of immunoreactivity was measured using NIH ImageJ software (
http://rsb.info.nih.gov/nih-image). Data are expressed as fold increase over control values.
2.3. Caspase-3
Caspases were analyzed by a Western blot analysis. For immunoblot analyses, 40 µg of protein lysates per sample were denatured in SDS-PAGE sample buffer (Tris-HCl 260 mM, pH 8.0, 40% (
v/
v) glycerol, 9.2% (
w/
v) SDS, 0.04% bromophenol blue and 2-mercaptoethanol as a reducing agent) and subjected to SDS-PAGE on 5% acrylamide/bisacrylamide gels. Separated proteins were transferred to nitrocellulose membrane (Hybond-P PVDF, Amersham Biosciences). Residual binding sites on the membrane were blocked by incubation in TBST (10 mM Tris, 100 mM NaCl, 0.1% Tween 20) with 5% (
w/
v). Membranes were then probed with a specific primary antibody, Cleaved Caspase-3 (Cell Signaling Technology, Danvers, MA, USA) (1:200). This was followed by a peroxidase-conjugated secondary antibody, HRP labeled mouse antirabbit Ig (Cell Signaling Technology, Danvers, MA, USA) (1:10000), and visualized with an ECL Plus detection system (Amersham Biosciences). The equivalent loading of proteins in each well was confirmed by Ponceau staining [
30].
2.4. Pictures of Thin Layer Cell Preparation
Using a Cytospin 4 cytocentrifuge (Thermo Scientific, Waltham, MA, USA), 200 µL of cell culture exposed to normobaric hyperoxia or control culture from normoxic conditions were sampled and centrifuged for 5 min at 600 rpm on microscope slides, followed by a standard Giemsa coloration. The slides were then read under a light microscope with 50-fold magnification.
Cell lines cultured in RPMI 1640 supplemented with fetal calf serum were collected. Slides were prepared from cell suspension by cytocentrifugation and stained with May-Grünwald-Giemsa. Randomly chosen fields with non-overlapping cells were captured using a Leica Dialux EB 20 microscope (oil immersion objective ×100) equipped with a DP200 digital camera (DetaPix).
For each slide, classical morphometric parameters such as nuclear area appearance (condensed or uncondensed chromatin, nuclei cytoplasmic ratio, presence of nucleoli, and regularity of nuclei) and cell cytoplasm color and regularity of contour were assessed as well as the presence of apoptotic or mitotic pictures. A scoring system based on FAB criteria [
31] was attributed to each morphologic item (from 0 for absence to 4 for very important). The biologist was blinded as to whether the cells had or had not been submitted to hyperoxia.
2.5. Fractals
For the purpose of the present investigation, determination of the fractal dimension of cells was performed using the Harfa 5.5 program (Faculty of Chemistry, Brno University of technology, Brno, Czech Republic) and by applying the box counting method after appropriate filtering and threshold application. The accepted final result was taken as the fractal dimension with the best fit to the slope described in the slope analysis (
Figure 2).
A fractal (from the Latin ‘fractus’, ‘broken’) is an object with a noninteger dimension that looks exactly the same at every scale. However, the definition of
fractal goes beyond self-similarity per se to exclude trivial self-similarity and include the idea of a detailed pattern repeating itself. Fractal patterns with various degrees of self-similarity have been studied in images, structures, and found in nature and technology [
32].
Euclidean descriptions are not adequate for complex irregular-shaped objects that occur in nature. These “non-Euclidean” objects are better described by fractal geometry, which has the ability to quantify the irregularity and complexity of objects with a measurable value called the fractal dimension [
33].
A geometrically intuitive notion of dimension is as an exponent that expresses the scaling of an object’s bulk with its size:
Here, bulk may correspond to a volume, a mass, or even a measure of information content and size is a linear distance (
Figure 3).
For example, the area (bulk) of a plane figure scales quadratically with its side (size), and so it is two dimensional, meanwhile a volume is related to the cube of its side. By transforming such relationships through the use of logarithms, we obtain a general equation of the form
where size is generally expressed as a fraction of the entire bulk: 1/N = K. This ratio is generally known as homothetia, meaning the operation able to geometrically transform the space without changing its form, i.e., preserving the pattern in between its constitutive elements. Bulk can be divided in N fractions (similar to the entire bulk), and each of those fractions has a length equal to 1/N=K. Then, we obtain D
segment =
= 1, meanwhile for an area we have D
area =
= 2. For a fractal object, like the Koch snowflake (
Figure 3) we have four segments similar to the entire bulk, each one equal to 1/3 of the entire length. Thus, the (fractal) dimension of that object can be calculated as D=
= 1.262. Its fractal dimension (1.262) therefore exceeds its topological dimension ‘1’, providing a quantitative measure of the space-filling capacity of a pattern that tells how a fractal scales differently than the space it is embedded in [
34]. Dimension is mathematically expressed by so-called “power laws”, since the Equation (1) shows that some quantity N can be expressed as some power of another quantity,
s:N
(s) =
s-τTaking the logarithm on both sides of the equation, we find a relationship indistinguishable from (2). By plotting log N(s) versus log s we obtain a straight line (the signature of the power law), being τ (a noninteger number) the slope of the straight line. The scale invariance can be seen from the fact that the straight line looks the same everywhere.
2.6. Statistics
Standard statistical analyses were performed, including mean, standard deviation, and ANOVA for repeated measures to test the between- and within-subject effect after Kolmogorov-Smirnov testing for normality. The Bonferroni or Dunnett’s tests were applied to the experimental and control values. Taking the initial value as 100%, percentual variations were calculated for each parameter, thereby allowing an appreciation of the magnitude of change rather than the absolute values.
Other tests between groups such as t-tests (with Welch correction) or non-parametric analysis were done when appropriate (Mann-Whitney, Wilcoxon). All statistical analyses were performed with the GraphPad Prism version 8.31 for Windows (GraphPad Software, La Jolla, CA, USA).
3. Results
Evident morphological modification of cell shape in JURKAT cells after 18 h of hyperoxic stress can be seen in
Figure 4.
Caspase-3 activity increased after hyperoxia and was even further elevated after return to normoxia when compared to data obtained in leukemia cell lines not submitted to transient hyperoxia (
Figure 5). Bcl-xl activity significantly increased 6 and 12h after hyperoxia in CCRF-CB line cells but remained statistically unchanged in JUKAT cell lines (
Figure 6).
However, morphological changes are generally not always so obvious (see
Figure 2); it is precisely in these situations that the usefulness of a fractal approach emerges in assessing how living structures have changed.
Fractal dimensions significantly increased in CCRF-SB and JURKAT cell lines after hyperoxia.
Figure 7 illustrates the fractal dimension variation in cell lines (JURKAT,
Figure 7A and CCRF-SB,
Figure 7B) up to 48 h after an 18-h hyperoxic period, compared to normoxia. A gradual increase in fractal dimensions of JURKAT (
Figure 7A) and CCRF-SB cells (
Figure 7B) can be seen along the course of the experiment from 0 to 48 h in culture medium following 18 h under hyperoxic conditions. There was no statistical difference between the different groups at 0 and 2 h. Both cells in the standard and hyperoxia group increased their fractal dimensions, but cells submitted to hyperoxia had significantly higher fractal dimensions at 4, 6, and 48 h following 18 h of hyperoxia.
Morphological changes revealed a significantly higher number of pycnotic nuclei (
Figure 8A;
p = 0.0058) in the cells submitted to hyperoxia. Moreover, the visual analysis confirmed the significantly higher number of apoptotic cells in the hyperoxia group (
Figure 8B;
p = 0.036) as previously suggested by the caspase results.