SEARCH

SEARCH BY CITATION

Keywords:

  • morphometry;
  • mitochondria;
  • apoptosis;
  • rupture of the outer mitochondrial membrane;
  • mitochondrial permeability transition

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Here we report on the mitochondrial permeability transition (MPT), which refers to the morphology of mitochondria whose inner membrane has lost its selective permeability. In all types of apoptotic cells so far examined, we found outer mitochondrial membranes that had been ruptured. These mitochondria present a swollen matrix covered by an inner membrane herniating into the cytoplasm through the breached outer membrane. Similarly ruptured outer mitochondrial membranes have been reported in studies on mitochondrial fractions induced to undergo MPT, carried out by others. Our observations were made on five types of rat tissue cells and six different cultured cell lines in the early stages of apoptosis. Samples from the cell lines HL-60, HeLa, WEHI-164, and a special batch of PC-12 cells were subjected to various apoptogenic agents and analyzed morphometrically. Nonapoptotic companion cells with unaltered nuclear structure (CUNS) were also analyzed. The mitochondrial volume in μm3 and the volume fraction of the cytoplasm occupied by mitochondria in cells with typical nuclear signs of apoptosis and also in CUNS were evaluated. The volume of the mitochondria with ruptured membrane represents at least 69% (47–89%) of the total mitochondrial volume of the apoptotic cells. Thus, a considerable fraction of the cellular mitochondrial mass is or was in the state of permeability transition and probably involved in enhancement of the apoptotic program. In all samples, a fraction of the cells with normal nuclei possessed mitochondria with breached outer membranes as described above. In these cells, MPT occurred before the appearance of the typical nuclear phenotype of the apoptotic cells. © 2004 Wiley-Liss, Inc.

The mitochondrial permeability transition (MPT) refers to an increase in the permeability of the inner mitochondrial membrane caused by nonselective agents, in apoptotic and necrotic cells (Lemasters et al., 1998). This condition is associated with membrane depolarization and collapse of the transmembrane potential (ΔΨm). Due to the hyperosmolarity of the matrix and the loss of selective permeability of the inner membrane, the ensuing influx of liquid to the mitochondrial matrix expands the matrix, promoting a large swelling of the mitochondria. The loss of selective permeability was initially thought to be caused by defects in the lipid moiety of the membrane, consequent to activation of phospholipase A2 by elevated Ca2+ levels (Pfeiffer et al., 1979; Beatrice et al., 1980). The alternative hypothesis, that MPT results from the opening of a pore or megachannel, referred to as the permeability transition pore (PTP) transversing both mitochondrial membranes (Hunter and Haworth, 1979; Zoratti and Szabo, 1995; Bernardi, 1999), has been extensively investigated and is currently widely accepted.

It was thought that, after the onset of MPT, expansion of the swollen matrix would cause the mitochondrial outer membrane to rupture due to the limited capacity for distension of this membrane. The inner membrane may expand much more because of its continuity with the membrane of the cristae (Petit et al., 1998). The possibility that the rupture of the outer membrane would allow the release of cytochrome c and other mitochondrial intermembrane proteins (Petit et al., 1998) is one of various alternative mechanisms found in the literature to explain how these protein inducers of apoptosis reach the cytoplasm. Another hypothesis is that the proteins are released to the cytoplasm by passing through the outer membrane (Desagher and Martinou, 2000). The idea that the release of mitochondrial proteins occurs either through permeabilization of (Basañez et al., 1999; Belzacq et al., 2002; Ravagnan et al., 2002) or the formation of supramolecular openings in (Antonsson et al., 2001; Kuwana et al., 2002) the outer membrane has recently gained momentum.

In in vitro systems, the proteins of the PTP may interact with proapoptotic proteins, such as Bax and Bid, promoting the permeabilization of the outer mitochondrial membrane to cytochrome c (Zamzami and Kroemer, 2003). MPT has been described in necrotic (Nieminen et al., 1995, 1997; Kim et al., 2003) and the majority of apoptotic cells. There have been reports of cases where the mitochondria of apoptosis-induced cells release cytochrome c without exhibiting permeability transition (PT) (Vander Heiden et al., 1997; Goldstein et al., 2000); however, this is not clearly understood (Tafani et al., 2001). In the majority of reports (Belzacq et al., 2002; Castedo et al., 2002), MPT is directly associated with the initiation of the apoptotic process.

Once in the cytoplasm, cytochrome c assembles with two other proteins, the apoptotic protease-activating factor 1 (Apaf-1) and procaspase 9 to form a complex, the apoptosome. The ensuing activation of caspase 9 leads the cell to the execution phase of apoptosis.

In addition to cytochrome c, other intermembrane mitochondrial proteins, such as Smac/DIABLO and Omi/Htra2, are able to induce or enhance the activation of caspases. The intermembrane proteins AIF and endonuclease G may act independently of caspase activation (Kuwana and Newmeyer, 2003). According to the cell type, caspases 2, 3, and 9 may be added to the list of mitochondrial intermembrane proteins that enhance the apoptotic process when they are liberated into the cytoplasm (Susin et al., 1999a, 1999b).

When apoptosis is induced by the other major route of induction, occupation of the TNF receptors, there are many cases in which the mitochondrial pathway is responsible for the manifestation of apoptosis. This seems to occur when relatively low levels of caspase 8 are activated near the cytoplasmic portion of the occupied TNF receptor. In this instance, procaspase 3 cannot be directly activated by caspase 8 to implement the late stages of the apoptotic program. However, caspase 8 can cleave the cytosolic protein Bid, giving rise to the truncated form of Bid, tBid, which translocates to the outer mitochondrial membrane, where it will interact with Bax and other proteins to promote the release of the mitochondrial intermembrane death-inducing proteins (Scaffidi et al., 1998; Antonsson et al., 2001). These facts emphasize the extensive participation of mitochondria in the enhancement of the apoptotic program.

Examination using transmission electron microscope (TEM) sections from various cell types undergoing programmed cell death identified many mitochondria with ruptured outer membranes. Through the breach, the swollen mitochondrial matrix, covered by an expanding inner membrane, herniates into the cytoplasm. This structural change in the mitochondria is identical to that previously described by Angermüller et al. (1998). The generalized rupture of the outer mitochondrial membrane revealed a wide range of microscopic profiles. Some of these profiles resemble in vitro TEM experiments in which isolated mitochondria swelled to various degrees of induction of the permeability transition. These results will be commented on below. Based on these observations, it seemed of interest to evaluate the magnitude and latitude of mitochondria undergoing the permeability transition in apoptotic cells from various cultured lineages induced with various apoptogenic agents. We have obtained estimates of the volume per cell and the cytoplasmic volume fraction of mitochondria with ruptured and nonruptured outer membranes in apoptotic and companion nonapoptotic cultured cells (CUNS). The cellular volume of mitochondria is the product of the number of mitochondria per cytoplasm and the average individual mitochondrial volume. Thus, by analyzing how these parameters changed, we hope to obtain an indirect insight into how the number of mitochondria was affected by the early stages of apoptosis. In the four cell lines studied morphometrically, we observed that the volume of mitochondria with ruptured outer membrane in apoptotic cells represents from 47% to 89% of the measured cell total mitochondrial volume. In all studied samples, we noticed that a fraction of the CUNS also possessed mitochondria with a breached outer membrane. In these cells, MPT preceded the activation of the caspases that induce the apoptotic nuclear phenotype. Although these results do not prove that the rupture of the outer mitochondrial membrane is the mechanism by which the intermembrane proteins are released into the cytoplasm, it is a most likely possibility.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Electron Microscopy

The organ fragments and cultured cells were processed as previously described (Sesso et al., 1999). Silver sections stained with uranyl acetate and lead citrate (both from Ladd Research Industries) were observed using a Jeol 1010 electron microscope or a Philips 301 at 80 kV. When it was necessary, to check whether the outer mitochondrial membrane was actually in the section but could not be observed due to an unfavorable sectioning angle, we performed a 24° tilt on either side of the normal plane of observation. This plane is at 0° tilt.

Rat Tissue Cells

Cells obtained from rat tissue were as follows: secretory epithelial cells from the ventral lobe of the prostate gland from rats killed daily in the 2–10 days following castration (Kyprianou and Isaacs, 1988); plasma cells from the granulation tissue of an experimentally induced scar in the dorsal skin of adult rats; macrophage, from the same granulation tissue; pancreatic acinar cells from pancreatic glands that had undergone ligature of the excretory ducts (Gukovskaya et al., 1996) 2–4 days previously or from rats maintained on a protein-depletion diet and receiving daily injections (40 mg/100 g of body weight) of dl-ethionine (Sigma) for 5 days (Walker et al., 1993); and secretory mammary cells from female rats subjected to interrupted lactation with daily gland removal from days 1 to 10.

Cultured Cells

PC-12 cells were kindly supplied by Dr. Paulo Lee Ho from the Institute Butantan in São Paulo. These cells were derived from PC-12 cells obtained directly from ATTC (pheochromocytoma; rat; ATCC CRL-172). They were grown in Dulbecco's modified Eagle's medium (DMEM; Life Technologies) supplemented with 10% FCS (Cultilab) at 37°C in a humid atmosphere in 5% CO2. After being stabilized in these growth conditions (Ho and Raw, 1992), it is no longer necessary to add poly-L-lysine to the underlying support as the cells adhere more easily than the original PC-12 cells. The adapted cells used here are referred to as PC-12* cells. The PC-12* cells were serum-deprived for 3 and 8 hr or exposed to brefeldine A (BFA 2.0 μM; Calbiochem; which blocks the anterograde vesicular traffic from the ER to the Golgi, but not the retrograde traffic), to 0.5 μM of staurosporine (STS; Calbiochem; a protein kinase C inhibitor), and to BFA + STS in the same concentrations, with an exposure time of 16 hr.

WEHI-3 cells (myelomonocyte leukemia; mouse; ATCC TIB-68) maintained in RPMI-1640 (Sigma) plus 10% calf serum were exposed for 5 hr to 20 μg/ml of the teneposide VM 26 inhibitor of topoisomerase II; 0.4 μg/ml vimblastine (Calbiochem; which prevents tubulin polymerization; 400 μg/ml of the antibiotic novobiocin (Sigma); 0.5 nM okadaic acid (Sigma), a potent inhibitor of protein phosphatases 1 and 2A; and 0.5 μM STS.

K-562 cells (chronic myelogenous leukemia; human; ATCC (CCL-243) exposed for 5 hr to vimblastine 0.4 μg/ml; oligomycin (Sigma), a highly specific mitochondrial ATP-synthase inhibitor (5 nM); VM 26, 20 μg/ml; novobiocin (Sigma) 400 μg/ml; nigericin (Calbiochem), a proton-ionophore, 10 μM; and BFA 2.0 μM.

HeLa cells (epithelioid carcinoma; cervix; human; ATCC CCL-2) in medium devoid of serum were exposed to 0.5 μM STS plus 2.0 μM BFA or to 0.5 μM STS for 16 hr.

WEHI-164 cells (mouse; methylcholanthrene-induced fibrosarcoma; ATCC CRL-1751) were exposed to BFA, to STS, and to BFA + STS in the same conditions previously employed.

HL-60 cells (human; peripheral blood; promyelocytic leukemia; ATCC CCL-240) were exposed for 16 hr to BFA 2 μM + human TNF 100 ng/ml + 2 μM camptothecin (CAMP; Sigma; this drug is a topoisomerase I drug that acts primarily on cells in the S-phase of the proliferative cycle). WEHI-3, K-562, and HeLa cells were cultured in RPMI-1640 with 10% FBS (Cultilab) in a humid atmosphere at 37°C in 5% CO2.

The apoptogenic agents for all samples were added to the cultures in fresh medium when the cells were about 50–60% confluent. In earlier phases of this study, we examined cultures of WEHI-3, K-562, L 929 (mouse fibrosarcoma; CLL-1.1), LLC-WRC 256 (carcinoma Walker; rat; ATCC CCL-38), WEHI-164, HL-60, and PC-12 cells (pheochromocytoma; rat; CRL-172). The PC-12 cells were obtained from the American Type Culture Collection in Rockville in 1992. Samples from all these cell lineages were analyzed under TEM in the early stages of exponential growth and not exposed to apoptogenic drugs.

Morphometric Procedures

The morphometric study was carried out on the four possible sectional profiles of mitochondria exhibiting rupture of the outer membrane (Fig. 1). Mitochondria with an intact outer membrane in apoptotic and CUNS cells from the same culture are referred to as type 1 mitochondrion and their sections are named type 1 profiles (Fig. 1A). To facilitate the identification of these profiles, we refer to the mitochondrion with a breached outer membrane as a type 2 mitochondrion (Fig. 1B). When it is sectioned unequivocally, we refer to it as a type 2 mitochondrial profile (schematized in Fig. 1D). A type 2 mitochondrion may be sectioned in such a way as to create mitochondrial profiles with both membranes (Fig. 1C) or only one membrane (Fig. 1E and F). The vesicular profiles covered only by the inner mitochondrial membrane are unequivocally recognized as type 3 profiles when they contain the remnants of mitochondrial cristae (Fig. 1E). Mitochondria covered by only the inner membrane and containing no cristae are referred to as type 4 profiles (Fig. 1F). We measured only profile types 1, 2, and 3. When a cell exhibits types 1 and 2 mitochondria, we cannot be sure from which of these two types a given type 1 profile originates (Fig. 1C).

thumbnail image

Figure 1. A schematic drawing illustrating how different angles of the sectioning plane will furnish different profiles in normal type 1 mitochondrion (A) and mitochondrion with a ruptured outer membrane (type 2 mitochondrion; B and D). Type 1 profiles can be generated from type 1 or 2 mitochondrion as demonstrated in C. The generation of profile types 3 and 4 is demonstrated in E (containing cristae) and F (without cristae), respectively. In all micrographs, the bars measure 500 nm.

Download figure to PowerPoint

Estimates of the cytoplasmic volume and the volumes associated with the mitochondrial profile types 1, 2, and 3 in apoptotic cells and the type 1 profile in nonapoptotic cells. These parameters were obtained by point-counting volumetry as indicated by Aherne and Dunnill (1982), Sesso et al. (1999), and Gundersen et al. (1988).

The equivalence of area fraction (AA) and volume fraction (Vv) by which AA = Vv is a fundamental concept of all morphometry (Aherne and Dunnill, 1982). Areas are estimated with a known degree of accuracy by counting how many regularly spaced points of a test system fall inside the area (hits). A given area value at the microscopic magnification chosen is associated with each point. This approach was used to estimate the areas of the cell sections from which we needed to obtain the radii. We will briefly survey the complete morphometric procedures used. The area fraction, AA, and, by extension, the volume fraction, Vv, also referred to as volume density, occupied by the mitochondria in the cytoplasm, is obtained by counting hits over the mitochondrial profiles (be they 1, 2, or 3) and over the cytoplasm including the previous counts over the mitochondrial profiles. The ratio between these two counts is AA = Vv. In order to obtain the cytoplasmic volume, the percentage of the cell volume the cytoplasm represents and its absolute value in μm3 must be calculated. To estimate the cellular volume, the distribution of radii of the cell sections is obtained. The cytoplasmic volume fraction of the cell is obtained by counting hits over the nuclear profiles and over the cytoplasm. The ratio between the number of hits over the cytoplasm and over the entire cell profile (the sum of hits over nuclei and cytoplasm) is the parameter procured. The cellular volume is obtained independently by measuring the radii of the cell sections.

The cellular and cytoplasmic volume of each cultured cell type was estimated using measurements from 10 to 20 enlarged prints (2,000 × 2.5). Each recorded microscopic field possessed 2–8 profiles of CUNS and at least one profile of a cell in explicit apoptosis. To be considered as a cell undergoing apoptosis, in addition to the nuclear phenotype of apoptosis, it had to have most of its perimeter covered by the cell membrane and no signs of advanced proteolysis. Some 15–30 and 70–100 profiles of apoptotic and CUNS cells, respectively, were measured. The test system superimposed over the prints was composed of three types of hit marker, regularly spaced small crosses and the extremities of two different types of regularly spaced segments of straight lines (Gundersen et al., 1988). Estimates of the volume density (Vv) of the nucleus and cytoplasm of the cells were thus obtained [Vv of the cytoplasm (Vvc) = number of hits over the cytoplasm/number of hits over the nucleus plus the number of hits over the cytoplasm]. Since the cell sections are fairly circular, the number of hits over each cell section gives an estimate of the corresponding area (πr2). In each sample, the radii from apoptotic and cells with unaltered nuclear structure were thus obtained. A 10-class distribution of radii was constructed. Employing the procedure of Bach (1963), used either in an HP machine (Arcon et al., 1980) or in a PC, the main parameters, including the mean sphere volume of the corresponding distribution of radii, were obtained. Since the mean sphere volume is an estimate of the cellular volume (vcell), the cytoplasmic volume is vcyt = vcell × Vvcyt.

Volumes of Mitochondria Types 1 and 2

Again, the area fraction was estimated by the point-counting procedure. The volume fraction is more commonly referred to as volume density (Vv). Having the cytoplasmic volume vcyt and, for example, the Vvmit or the AAmit of the mitochondria in the cytoplasm, one obtains the total mitochondrial volume or area (vtm or Atm). Thus, vtm or Atm = Vvmit (or AAmit) × vcyt. Since a randomly sectioned type 2 mitochondrion furnishes profile types 1, 2, 3, and 4, of which only the initial three profiles were measured, the sum of the obtained volumes for profile types 2 and 3 gives the best possible estimate of the volume of type 2 mitochondria. This volume, however, is forcibly underestimated on two accounts. In apoptotic cells, the type 1 profiles derived from type 2 mitochondria cannot be scored as such. Likewise, all type 4 profiles are not counted. Therefore, the mitochondrial volume derived from evaluations carried out on type 1 profiles in apoptotic cells represents the aggregated volume of actual type 1 and some type 2 mitochondria. The estimates of the cytoplasmic volume (or area) associated with mitochondrial types 1, 2, and 3 profiles in apoptotic cells were carried out using prints enlarged to 20,000× (8,000 × 2.5). The cytoplasm from both CUNS and cells undergoing apoptosis were examined for each type of cell culture.

At least 10 micrographs were obtained from each group. In the apoptotic cells, we often found the three types of mitochondrial profiles. In CUNS, particularly from cultures with high apoptotic indexes, in addition to profile type 1, we also observed mitochondrial profile type 2 and/or 3. These mitochondrial profiles, except for the case in line 2 of Table 1, were not point-counted as explained above. In the enlarged (20,000×) micrograph, each type of mitochondrial profile was marked with a fine ink dispenser. The number of hits falling within each profile type (H1, H2, and H3, for mitochondrial profile types 1, 2, and 3, respectively) was scored. The sum of all mitochondrial hits (H = H1 + H2 + H3) and the total number of hits (h) over the cytoplasm was also scored (h = the sum of hits over mitochondrial and nonmitochondrial cytoplasmic structures). The volume density of the total mitochondrial types in the cytoplasm (Vvtm) is Vvtm = H/h; and the Vv of each mitochondria type in apoptotic cells is Vvi = Hi/h, where i refers to mitochondrial profile types 1, 2, and 3 and = 1, 2, and 3. The respective total mitochondrial volume (vtm) is vtm = Vvtm × vcyt and the volume associated with each mitochondrial profile type is vmi = Vvi × vcyt.

Table 1. Estimates of the Volume Density (Vv) and of the Volume (v) of Types 1 and 2 Mitochondria in Apoptotic (AP) Cells and in CUNS*
Cell type and treatmentVolume density (Vv) and volume (v) of type 1 mitochondrial profiles in CUNSVv and v of type 1 mitochondrial profiles in AP cellsVv, v, and % of type 2 mitochondrial profiles in AP cellsaVv and v of type 3 mitochondrial profiles in AP cellsVv, v, and % of summed Types 2 and 3 mitochondrial profiles in AP cellsbVv and v of the total mitochondrial profiles in AP cells
  • *

    In CUNS exempted for the data in line 2 and column 2, only type 1 mitochondrial profiles were sampled.

  • a

    Percentage of the estimated volume in relation to the total mitochondrial volume in AP cells.

  • b

    Percentage of the estimated volume in relation to the total mitochondrial volume in AP cells.

  • c

    In this sample, the volumes of types 2 and 3 mitochondrial profiles were evaluated in CUNS and were 19 and 0.3 μm3, respectively.

  • d

    Extreme percentual values in the columns.

3-sera-deprived PC-12* cellsVv (%) 102.31.52.33.86
v (μm3) 2864.061016
25%cd63%b
HL-60, BFA + TNFα + CAMP/16 hVv (%) 7.94.45.90.46.310.8
v (μm3) 47c273633966
54%59%
PC-12*, BFA 2 μM/16 hVv (%) 5.714.60.65.26.2
v (μm3) 3862843438
70%89%d
PC-12*, BFA + STS, 2 μM and 0.5 μM/16 hVv (%) 70.640.34.34.9
v (μm3) 57317.021922
77%d86%
WEHI-164, BFA 2 μM/16 hrVv (%) 72.41.90.22.14.3
v (μm3) 1626048553113
42%47%d
WEHI-164, STS 0.5 μM/16 hrVv (%) 4.50.52.70.22.83.4
v (μm3) 97124044456
71%79%
WEHI-164, BFA + STS, 2 μM and 0.5 μM/16 hVv (%) 9.21.42.30.734.4
v (μm3) 22432521668100
52%68%
HELA, BFA + STS, 2 μM and 0.5 μM/16 hVv (%) 123.74.70.358.7
v (μm3) 297911158123214
54%57%

Total Mitochondrial Surface Area on a Per-Cell Basis in Apoptotic Cells and CUNS: Evaluation of the Average Surface-to-Volume Ratio of Mitochondria

The hits (hi) made by the extremities of the regularly spaced segments [of length (l) in μm at the given magnification] of the test system, in the interior of the mitochondrial profiles (1, 2, or 3 in apoptotic cells) and also over the remaining cytoplasm, were scored. The number of times the segments from the test system crossed (C) the limiting membranes of the mitochondrial profiles (1, 2, or 3 in apoptotic cells and 1 profiles in CUNS) WAS also detected. The mitochondrial surface density (Sv) is Sv = 4C/l × hi, where C and hi are the total number of times the segments of the test system cross the mitochondrial borders and the number of hits the segment extremities superimpose on the whole cytoplasm, including the mitochondrial profiles. Sv is the mitochondrial surface area in μm2 per μm3 of cytoplasm. The total mitochondrial membrane surface area per cell (TMMSA) is obtained by multiplying Sv by vcyt. The average cellular mitochondrial surface-to-volume ratio (s/v) is s/v = TMMSA/vtm.

Error Associated With Area Estimation

The coefficient of error associated with the estimation of mitochondrial areas, circular or elliptical, was kept below the 0.05 level and was determined as described by Gundersen and Jensen (1987).

Apoptotic Index

With magnifications of 2,000 × 10 and 5,000 × 10, at least 50–100 cells sections were randomly picked and the percentage of apoptotic cells were evaluated. The characteristic nuclear alterations of the apoptotic cells are unmistakably recognized using TEM. The apoptotic cells were selected for morphometric studies if they possessed both a typical apoptotic nucleus and a cell membrane at the cytoplasmic border. For the estimation of the apoptotic indexes (AIs), all sections of cells with the nuclear phenotype of apoptosis were counted. Therefore, cells in various advanced stages of cytoplasmic proteolysis were also counted.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Detection of Apoptotic Cells by Transmission Electron Microscopy

Among the biochemical and structural changes observed in cells undergoing apoptosis, the most prominent and an essential element in the identification of this type of cell death are the structural alterations of the nucleus. Our observations confirm that the various forms of apoptotic nuclear phenotype are represented in both rat tissue cells and in immortalized cultured cells. Nuclear contraction is coincident with hypercondensation of the chromatin (i.e., exceeding that seen in normal heterochromatin) and the clumping of the chromatin into masses that adhere along the nuclear membrane. Such masses vary in appearance from large spherical-like (Fig. 2A) to demilune-like conformations (Fig. 2B and C). The fragmentation of these large masses, which often occupy the entire nuclear profile, into various smaller spheroid bodies also occurs. This particular nuclear phenotype corresponds to what the former cytologists called karyorrhexis. Additional phenotypes of apoptotic cells nuclei as well as of mitochondria in both CUNS and apoptotic cells may be examined at http://www.sebepa.cjb.net/; the password is “sessoimt.”

thumbnail image

Figure 2. A: Apoptotic PC-12* cell treated with BFA (2 μM for 16 hr). The compact mass of hypercondensed chromatin (1) appears detached from the nuclear membrane (2), apparently leaving clear the region where the nuclear lamina dwells (the small black dot at the extremity of the line marked 3 at the opposite end covers the region where the nuclear lamina reside, adherent to the nuclear membrane). Nuclear pores are indicated by arrows. Dilation of the endoplasmic reticulum (ER) cisternae, a common effect of BFA, is also shown (4). B: WEHI-164 cells exposed to staurosporine (0.5 μM for 16 hr). Arrows indicate two apoptotic cells. Chromatin blocks can be visualized adhering to the nuclear membrane (1). One of these blocks, in the cell on the left, has demilune profile (2). C: Apoptotic HeLa cell exposed to BFA (2 μM) plus staurosporine (0.5 μM), both for 16 hr. Peripheral disposition of the chromatin masses in the nucleus (1) and dilation of the ER cisternae (2) can be observed.

Download figure to PowerPoint

Apoptotic Cells in Rat Tissues

The finding of mitochondrial profile types 2 or 3, or both, reveals the presence of a type 2 mitochondrion. In populations with an elevated percentage of apoptotic cells, this finding is frequent. All apoptotic tissue cells we examined possessed type 2 mitochondria (arrows in Figs. 3 and 4B). The low-magnification image of the apoptotic cell in Figure 3A reveals a common structural alteration seen in both apoptotic cells and in CUNS undergoing MPT in a less stretched cytoplasmic form. We observe the absence of organelles in some sectors of the cytoplasm (2 in Fig. 3A) and their clustering in others (3 in Fig. 3A).

thumbnail image

Figure 3. A: An apoptotic cell from the secretory epithelium of the prostate gland of a rat castrated 4 days previously is seen in the central part of the micrograph. The nucleus is in the process of fragmentation (1). The cytoplasm of the region at right is devoid of organelles (2), some of which are well preserved and clustered at the opposite pole of the profile (3). The arrow points to the region enlarged in the inset. B: Apoptotic plasma cell from the granulation tissue of a scar experimentally induced 5 days previously in the dorsal skin of an adult rat. Besides the two mitochondria exhibiting permeability transition (indicated by arrows), a relatively large cluster of ∼ 50 nm microvesicles can be observed (1) (Sesso et al., 1999). C: Apoptotic macrophage from the same granulation tissue referred to in B. A sector of the nucleus can be observed in the upper right corner (1). The type 2 mitochondrion in the lower central part (2) possesses a relatively small breach of the outer mitochondrial membrane, indicated by an arrow.

Download figure to PowerPoint

thumbnail image

Figure 4. A: Nonapoptotic pancreatic acinar cell from a pancreas that had undergone ligature of the excretory ducts 2 days earlier. This cell was close to others that were undergoing generalized cell death. Note the normal texture of the chromatin (1) and the type 2 mitochondria indicated by an arrow. B: Apoptotic pancreatic acinar cell from the same gland referred to in A. Type 2 mitochondrion (indicated by the arrow) can be visualized and, near the nucleus with compacted chromatin (1), a cluster of ∼ 50 nm microvesicles (2) is noted.

Download figure to PowerPoint

Rat prostate secretory cells.

The lack of male hormones induces a marked involution of the prostrate gland consequent to death of the secretory epithelial cells by apoptosis. Ten days after castration, the gland mass is reduced to some 15% of the original weight. We have studied glands removed 4–6 days after orchiectomy more thoroughly (Fig. 3A); these exhibit a high incidence of apoptotic epithelial cells.

Plasma cell and macrophage.

A distinct increase in the number of cells at the granulation tissue occurs 4–6 days after scar formation. At day 5, we found more apoptotic plasma cells (Fig. 2B) and macrophages (Fig. 2C) than at other time intervals. This is coincident with the proliferation time of fibroblasts, plasma cells, and macrophage.

Pancreatic acinar cells.

Both procedures used to induce apoptosis were highly effective. After ligature of the excretory ducts, the pancreas gland regions that suffered a lack of flowing secretion with consequent compression of the cells exhibited generalized apoptosis of the acinar cells. In one of these glands, we observed a nonapoptotic cell with a mitochondrion exhibiting a ruptured outer membrane (arrow in Fig. 4A). In the majority of apoptotic cells, mitochondria undergoing permeability transition could be recognized (Fig. 4B).

Mammary gland secretory epithelial cells.

The abrupt removal of the suction stimulus from the breast promotes a hormonal imbalance that induces apoptosis of the secretory cells (Walker et al., 1989) and consequent involution of the mammary glands. The incidence of apoptosis in the secretory cells is revealed by the rapid increase of intraepithelial macrophage filled with apoptotic bodies containing remnants of the secretory cell cytoplasm. The percentage of macrophage in the epithelium rose by day 3 after the interruption of lactation and remained elevated until day 9. We found type 2 mitochondria in the apoptotic bodies inside epithelial cells from a gland obtained at day 3 after separation of the littermates from the nursing female.

Apoptotic Cultured Cells

Examining the various cell cultures maintained under normal conditions without the addition of apoptogenic agents, in no instance did we find type 2 mitochondria in the nonapoptotic cells. The L 929 and the PC-12 cells from ATCC were also examined after treatment with apoptogenic agents. TNFα was added to cultures of L 929 cells while apoptosis was induced in the PC-12 cells by sera removal. In these particular experiments, no apoptosis was detected in the L 929 cells. To confirm this negative finding, a thorough search for apoptosis as well as type 2 mitochondria in CUNS was carried out with no success. After sera removal from the PC-12 cells, samples were collected at various time intervals between 2 and 48 hr. While the apoptotic cells exhibited type 2 mitochondria in all time intervals, in no instance did we find mitochondria with a ruptured outer membrane in CUNS. We have examined more than 600 CUNS profiles looking for type 2 mitochondria. The sera-deprived PC-12 cells were the most thoroughly examined as we analyzed samples sectioned serially and semiserially for other studies (Sesso et al., 1999). In the modified PC-12* cells from the Butantan Institute, we found mitochondria with a ruptured outer membrane in CUNS and also in apoptotic cells in the very first samples deprived of serum. These observations will be presented and discussed in a follow-up study.

WEHI-3 and K-562 were among the first cell cultures examined and subjected to various apoptogenic substances. We observed type 2 mitochondria in apoptotic WEHI-3 cells exposed to STS, novobiocin, VM 26, and vimblastin and in apoptotic K-562 cells treated with BFA, VM 26, and thapsigargin. All dead cells from cultures subjected to apoptogenic agents so far tested using morphometric studies possessed a cell membrane and nuclei with the typical apoptotic phenotype.

It is common to observe mitochondria with spherical profiles along with the normal elongated mitochondria in CUNS. The incidence of these profiles was influenced by the degree of cell death occurring in the cell culture. In the experiments presented in Table 1, the percentage of cells undergoing apoptosis varied from 6% to 66%. The higher this apoptotic index, the more frequently one could find CUNS with spherical mitochondria, many of which exhibited incipient swelling. From 1% to 17% of the CUNS in the various samples possessed mitochondria undergoing the permeability transition. In these cases, many type 2 or 3 mitochondrial profiles could be found.

The successive morphological changes exhibited by the apoptotic cells in the cultures we have examined suggest that apoptotic cells undergo caspase-orchestrated lysis of all membrane-bound and cytoskeletal cytoplasmic structures. We observed dense chromatin masses, often with circular profiles, in apoptotic nuclei of cells in advanced stages of cytolysis. The cytoplasm of these cells was either almost completely devoid of organelles or contained some dense or swollen type 2 mitochondria as residual organelles. When caspase-driven proteolysis and the action of nucleases was advanced, it was often difficult to be precise about the former morphology of the nucleus and cytoplasmic organelles, all then appearing as remnants. These changes occurred along with the complete rupture of the cell membrane. When this happened, either in an apoptotic cell or in very rare, perhaps necrotic, nonapoptotic ones, the cytoplasm was fragmented and the organelles exposed to the culture medium became swollen and disintegrated.

Two (PC-12* and WEHI-164 cells) of the four cell lines (the two others are HL-60 and HeLa cells) in which the mitochondria were studied morphometrically were also subjected to various apoptogenic agents. In all eight samples, type 2 mitochondria were found in apoptotic cells (Fig. 5) and in some of the cells with normal nuclei. Frequently, one observed type 2 mitochondrial profiles with a relatively small site of rupture at the outer membrane (lower type 2 profile in Fig. 3C and profile with an asterisk in Fig. 5A). In the apoptotic cells, type 2 and 3 mitochondrial profiles may be present in the cytoplasm even in advanced stages of cellular proteolysis. These cell profiles exhibited a few scattered chromatin blocks and scarce type 2 and 3 mitochondrial profiles in a cytoplasm virtually devoid of organelles. The main organelles in PC-12* cells in advanced stages of apoptosis when the cell was being segmented into apoptotic bodies were mitochondrial profile types 2 and 3.

thumbnail image

Figure 5. A and B are from PC-12* CUNS cells treated with BFA (2 μM) and STS (0.5 μM) for 16 hr. The types of mitochondrial profiles are indicated by numbers. In the upper left corner of A, a type 2 profile is sectioned at various levels orthogonally to the plane of the image. Planes AB and CD demonstrate type 2 and 3 profiles, respectively; plane EF, which contains no cristae remnants, presents a type 4 profile. The arrows indicate unimembranous vesicles with homogeneous contents that do not resemble the contents of the swollen part of profiles 2 that are devoid of cristae. These do no look like type 4 profiles. In A, an asterisk marks a type 3 profile with a relatively small breach of the outer membrane.

Download figure to PowerPoint

The configuration of the type 2 mitochondrial profiles may vary considerable with regard to the degree of swelling of the mitochondrial matrix. In Figure 6A and 6C, distinct differences in magnification are presented to demonstrate that, in the smaller images (Fig. 6B and 6D), it is not easy to perceive whether the regions indicated by arrows actually have one or two membranes. In these markedly swollen mitochondrial profiles (arrows in Fig. 6B and D), two apposite poles are visible. One appears as a dense membrane with fragments of cristae nearby. Opposite to this pole, the mitochondrial matrix is limited by a thin covering membrane. The region of the profile with mitochondrial cristae and a dense membrane is where the profile exhibits both mitochondrial membranes. At the opposite pole, the matrix is covered only by the inner membrane.

thumbnail image

Figure 6. Apoptotic HeLa and WEHI-164 cells, both exposed to STS (0.5 μM) for 16 hr. These images illustrate that at relatively low magnifications, it is not possible to perceive that the dense parts of the mitochondrial profiles close to the cristae and pointed by arrows in B and D actually possess two mitochondrial membranes as shown in A and C. The opposite pole of each enlarged profile is covered by a thin inner mitochondrial membrane.

Download figure to PowerPoint

Surface-to-Volume Ratio of Type 1 and 2 Mitochondria in Apoptotic Cells and in Companion Cells With Normal Nuclei

Measurements of the total surface area of the mitochondria carried out in the eight studied samples allowed us to obtain the mitochondrial surface-to-volume ratio in all apoptotic cells and CUNS examined (data not shown). The ratio between the mitochondria of apoptotic cells and in CUNS did not vary significantly (P > 0.05), indicating that no important change in the average mitochondrial volume between the two groups was actually detected. Therefore, in these two groups, the mitocondrial volumes are probably also a measure of the number of mitochondria per cell.

Evaluation of the volume, in μm3, and of the volume fraction Vv, occupied by mitochondria types 1 and 2 in the cytoplasm of apoptotic cells and in the corresponding CUNS of the same cell culture, is presented in Table 1. The data in the second line of Table 1 represent the best estimates of the average volume per cell of types 1 and 2 mitochondria in PC-12* apoptotic cells and CUNS. The mitochondria of the CUNS occupied 10% of the total cytoplasmic area or volume (second line/second column) and had an average global volume per cell of 28 μm3. In apoptotic cells, mitochondrial profile type 1 and the sum of types 2 and 3 occupied 2.3% and 3.8% (second line/columns 3 and 6, respectively) of the cytoplasm, with volumes of 6 and 10 μm3, respectively. In the PC12* apoptotic cells, the volume of type 3 mitochondria is of 6 μm3, representing 37.5% of the total mitochondrial volume of these apoptotic cells (6/16 × 100 = 37.5%); in this case, the value 16 μm3 (second line/column 7). The values of type 1 mitochondrial profiles found in apoptotic cells (column 3) lied in the range of 14% (fifth line 3/22 × 100) to 53% (sixth line 60/113 × 100) of the total mitochondrial volume from these cells. In the various cell lineages analyzed, the volumes associated with type 2 mitochondrial profiles varied from 25% to 77% of the estimated total mitochondrial volume in apoptotic cells (column 4). For the apoptotic PC-12* cells (second line/column 4), this value is 4/16 × 100 = 25%. The percentages of the mitochondrial volume of the apoptotic cells occupied by the sum of mitochondrial profiles types 2 and 3 are presented in column 6. In the apoptotic PC-12* cells of the second line, this percentage is 10/16 × 100 = 63%. In the various samples, this parameter varied from 47% (sixth line) to 89% (fourth line), with an average of 69%. This important information conveyed by the data of column 6 reveals that the majority of the mitochondria of the apoptotic cells when captured under the TEM are or were in the state of permeability transition. It was common to observe some 1–5 type 2 and some 2–6 type 3 mitochondrial profiles in different sections of apoptotic PC-12* cells.

The total mitochondrial volume of the apoptotic cells (v in column 7) in six out of eight cases is lower than that of the CUNS (lines 1, 2, 5, 6, 7, and 8 and columns 2 and 7 of Table 1), while in the other two samples the mitochondrial volume in apoptotic cells and in CUNS are about equal. In the third line, the cited volume in apoptotic cells is 66 μm3, while in the corresponding CUNS (column 2), it is 47 μm3. If the 19 μm3 (see footnote to Table 1) of the CUNS with type 2 mitochondria is included, the total amount is equal to that of the apoptotic cells. In the fourth line, both values of v (columns 2 and 7) for the mitochondrial volume are equal (38 μm3).

As already mentioned, the incidence of type 2 mitochondria in cells with normal nuclei was quite variable (it varied from 1% to 17%) and they would not always be adequately sampled when 10 random (range, 10–20) sections of cells with normal nuclei are taken for measurements. For this reason, with one exemption, presented as an example (line 3 and footnote), we have not included such cells in the nonapoptotic group when estimating mitochondrial volumes. None of the mitochondrial volumes of the various columns of Table 1 correlate significantly with the corresponding AIs. These varied from 6% to 68%.

Cytoplasmic Sectors Devoid of Organelles in Apoptotic Cells and in CUNS

In many of the CUNS possessing a ruptured outer mitochondrial membrane, the cytoplasmic distribution of organelles was altered. The membrane-bound structures were concentrated in certain sectors of the cytoplasm while absent from others. This clustering of organelles is more frequently found in apoptotic cells than in CUNS (Fig. 3A). This observation suggests that the cytoskeleton is undergoing alterations before the nuclear changes appear (data not shown). A literature search reveals that in cells induced to undergo programmed death, alterations of the distribution and morphology of the actin microfilaments is an early apoptotic event.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

We decided to use transmission electron microscopy to identify apoptotic cells as it seems to be the most reliable procedure to accomplish this task (Yasuhara et al., 2003). Morphological data to distinguish between necrotic and apoptotic cells have long since been identified (Kerr et al., 1995). A literature search revealed that the TUNEL reaction gives high false positive rates while DNA ladder assay lacks sensitivity (Yasuhara et al., 2003). The binding of annexin V at the cell surface and positive staining with propidium iodide occur in both apoptotic and necrotic cells. In cultured cells, annexin V marks early stages of apoptosis when the cell is still covered by a continuous membrane that keeps the propidium iodide out of the cell. As apoptosis progresses, various successive portions of this membrane are removed, allowing the entrance of propidium iodide that stains the cell. This apoptotic cell is positive for both annexin V and propidium iodide, as are necrotic cells (Vermes et al., 1995).

The results presented in this article clearly demonstrate the high incidence of morphometric change in the mitochondria of apoptotic cells. From our data, we calculate that at least 47–89% (average 69%) of the mitochondria from these cells have a ruptured outer membrane. Once the outer mitochondrial membrane has ruptured, the inner membrane, covering an expanding swollen matrix, passes through the formed hole and spreads into the surrounding cytoplasm. Since the continuous expansion of the mitochondrial matrix needs to be membrane-bound in order not to rupture, the membrane from the cristae are incorporated into the inner membrane.

It is difficult to evaluate the magnitude of the underestimations of the actual volume of type 2 mitochondria. We would need to know what fraction of the measured type 1 profiles derive from type 2 mitochondria and how the unimembranous vesicles derived from type 2 mitochondria, but devoid of cristae, the type 4 profiles are numerically related to type 3 profiles. It was also evident during the collection of morphometric data that the frequency of the four profile types varied according to the morphological changes of the forming and developing type 2 mitochondria. The change in size of the unimembranous-bound part of type 2 mitochondria will affect the yield of type 3 and 4 profiles when the mitochondria are randomly sectioned.

It is curious that, besides the publications of Angermüller et al. (1998) and of Kwong et al. (1999) in apoptotic hepatocytes and secretory epithelial cells from the prostate gland of castrated rats, respectively, no other reports of rupture of the outer mitochondrial membrane in apoptotic cells have been published.

As mentioned, in six samples out of eight, the mitochondrial volume of the apoptotic cells was less than that of the CUNS. Since we do not know the magnitude of underestimation of the volume of type 2 mitochondria, we cannot be sure of a possible reduction of this volume in apoptotic cells. It must be noted that these results may simply represent the process of organelle disassembly in many of the cells analyzed. It is unclear whether the numerical reduction of mitochondria detected in cells with typical apoptotic nuclei had begun before the changes in nuclear structure.

One of the facts that must have determined the lack of statistical difference between the surface-to-volume ratio of the mitochondria of the apoptotic cells and of the CUNS is that, in cultures subjected to apoptogenic agents, the CUNS often exhibit type 1 mitochondria with a spherical shape (examples in Figs. 10–15; Figs. 36, 72, 73, and 75 of the site http://www.sebepa.cjb.net/). Not rarely a variable part of these profiles, mainly in cultures with high AIs, may appear swollen. It is possible that these swollen mitochondria found in CUNS, namely, those exhibiting ruptured outer membrane, are revealing that such cells are in line to start showing the nuclear signs of apoptosis.

It is as yet unclear what causes the rupture of the outer mitochondrial membrane in CUNS. The involvement of terminal caspases such as caspase 3 can be excluded due to the absence of nuclear alterations; these alterations are a consequence of the activation of this caspase. It is also unclear whether caspase 3 may influence the appearance of type 2 mitochondria once the nuclear alterations have begun. One cannot rule out the possibility that in many of the observed cases an initiatory caspase, such as caspase 8, activates the mitochondrial-dependent apoptotic pathway (Scaffidi et al., 1998, 1999) and is a causative agent of the rupture of the outer mitochondrial membrane in CUNS. In this hypothesis, the type 2 mitochondria in CUNS would be actively involved in causing the explicit structural changes of apoptosis.

We suggest that type 2 mitochondria seems to be an indicator of the MPT since the morphology of these swollen mitochondria could only be attained if the inner mitochondrial membrane had lost its selective permeability. MPT has also been described in necrotic cells. However, we do not know whether rupture of the outer mitochondrial membrane also occurs in necrotic cultured cells. In rare identifiable necrotic cells found in our studies, the cytoplasm was fragmented and the organelles, mitochondria included, in the process of disintegration. An answer to this question may be obtained by performing TEM studies in cells induced to undergo necrosis by an abrupt and severe reduction in the cellular ATP levels (Kim et al., 2003).

To associate the TEM images of type 2 mitochondria undergoing MPT with the concept of a permeability transition pore or megachannel, it is necessary to envisage that once the permeability transition pore is fully open, a massive influx of fluids would occur between both mitochondrial membranes in the region of the pore, i.e., in a restricted sector of the outer membrane. We have microscopic data that will be presented in a follow-up study, supporting this assertion. As the punctual accumulation of liquid progresses, it would promote a small focal rupture of the outer membrane. The depolarization of the inner membrane would then rapidly spread from the initial point where the pore was. In such a manner, the mitochondrial matrix would accumulate fluid from the cytoplasm causing the swelling observed. The frequent finding of type 2 mitochondrial profiles with relatively small breaches in the outer membrane (see also results on our Web site) supports this conjecture.

A clear-cut demonstration of rupture of outer mitochondrial membrane was obtained in mitochondria isolated from cortical neurons and induced to undergo permeability transition. The release of cytochrome c from the intermembranous space appeared to be dependent of the rupture of the outer mitochondrial membrane (Brustovetsky et al., 2002). A second occurrence of ruptured outer mitochondrial membrane in mitochondrial fractions induced to undergo permeability transition may be observed with a magnifying lens in the right superior quadrant of Figure 4B from Petronilli et al. (1993) directly on the journal page. Three extremely swollen and unequivocal type 2 mitochondrial profiles, joined two by two, may be seen. The connection between the mitochondrial configuration in apoptotic cells, and the results in isolated mitochondria expressing permeability transition, substantiates the view that the structurally altered mitochondria we are observing in apoptotic cells and also in companion cultured cells with normal nuclei are actually expressing the state of permeability transition.

Possibly in connection with the just cited observations in isolated mitochondria, it is of relevance to mention that the low-power images of type 2 mitochondrial profiles shown in Figure 6B and 6D resemble transmission electron micrographs, similarly imaged, of pelleted swollen mitochondrial induced to undergo MPT [Figs. 8B–D and 9A, C, and D in Beatrice et al. (1982); Fig. 4B and C in Igbavboa and Pfeiffer (1988); Fig. 4B in Petronilli et al. (1993); Fig. 3B in Jung et al. (1997)]. Ours and these mitochondrial profiles have in common a dense region in one pole of variable curved length. This dense region is rapidly recognized by the presence of mitochondrial cristae close to or contacting its inner surface. In association with a variable amount of cristae, the swollen isolated mitochondria exhibit a more or less empty matrix with a variable amount of cristae. These profiles opposite to their dense region are covered by a thin membrane.

In order to clarify whether the type 2 mitochondria we observe in CUNS and apoptotic cells actually release cytochrome c into the cytoplasm, we will carry out essays in which the time-course incidence of type 2 mitochondria will be correlated to the amount of cytosolic cytochrome c in the cells under analysis. Parallel evaluation of the activity of caspases 8, 9, and 3 will also be performed.

The following observations strongly suggest that the rupture of the outer mitochondrial membrane we describe is the mechanism by which intermembrane mitochondrial proteins are released into the cytoplasm.

One, the morphology of the type 2 mitochondrial profiles reveals exposure of the external surface of the inner membrane and therefore the intermembrane and membranous mitochondrial proteins associated with this surface to the cytoplasm. This indicates that there may be consequences to the apoptotic program if cytochrome c is exposed to cytoplasm containing the Apaf 1. If cytochrome c is indeed loosely attached to the inner membrane (Lemasters et al., 1998), it will be not only exposed, but actually progressively released into the cytoplasm as the inner membrane transverses the breached outer membrane.

Two, various agents can open the permeability transition pore and promote loss of the inner mitochondrial membrane selective permeability. The appearance of the MPT is accompanied by the release of intermembrane mitochondrial proteins that trigger the activation of procaspases (Tafani et al., 2001; de Giorgi et al., 2002; Kokoszka et al., 2004).

Three, the simultaneous occurrence of MPT and altered organelle distribution in the cytoplasm of the CUNS is another circumstantial element, compatible with the view that in these cells the apoptotic program was already beginning to take place.

Four, of the drugs we have employed to induce apoptosis (listed in Table 1), staurosporine (Tafani et al., 2001), BFA (Ito et al., 2001), TNF (Vikhanskaya et al., 2002a), and camptothecin (Stefanis et al., 1999; Sanches-Alcazar et al., 2000) also promote the cytoplasmic release of cytochrome c.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

The authors thank Angela Batista Gomes dos Santos, Maria Cecília dos Santos L. Marcondes and Marcelo Alves Ferreira of the Department of Pathology of the Faculty of Medicini of São Paulo, for their technical support and Miguel da Silva Passos Júnior from the Department of Surgery for doing the photographic work.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED