Genomic c-Myc Quadruplex DNA Selectively Kills Leukemia

This article is featured in Highlights of This Issue, p. 1

The c-myc gene encodes a nuclear phosphoprotein with key regulatory roles in a wide array of cellular processes including the regulation of cell-cycle progression, cell proliferation, differentiation, transformation, angiogenesis, and apoptosis (1–5). Normally, expression of c-myc is tightly regulated and closely correlated with proliferation. When cells are quiescent, c-myc is undetectable, however, upon growth factor stimulation, c-myc levels rapidly increase and sharply decrease as cells progress through the proliferative cycle (6). However, in a variety of human cancers, deregulation and inappropriate activation of c-myc commonly occurs as a consequence of chromosomal translocation, gene amplification, and increased transcription/translation resulting in pronounced c-myc gene amplification (7). In leukemia, aberrant c-myc expression imparts a proliferative advantage over normal cells causing a failure in cellular differentiation. Reducing c-myc expression with oligonucleotides (ODN) may attenuate cell growth and represent a potential antileukemic approach.

Although blocking c-myc expression with ODNs induces differentiation of myelocytes and myeloid leukemia cells, their use is complicated by nuclease degradation in serum and intracellularly, poor uptake into cancer cells, and unwanted effects on normal cells. However, it was found that DNA sequences rich in guanines, capable of forming 4-stranded structures known as quadruplexes, may be inherently stable in biological fluids and sufficiently taken into cells. The use of random G-rich quadruplex-forming ODNs as therapeutic agents have shown impressive antiproliferative activity against a wide range of cancer cells, although being virtually nontoxic to normal cells (8, 9).

Recently, it has been shown that quadruplex sequences are represented disproportionately in biologically important regions of the genome such as telomeres and in the promoters of growth regulatory genes (10). Tumor suppressors tend to possess low quadruplex-forming potential, whereas oncogenes such as c-myc have a high generation of tetrahelical domains (11). The c-myc quadruplex-forming sequence, PU27, is a 27 base pair sequence comprising 5 regions of consecutive runs of guanines within the noncoding strand. It is located −142 to 115 bp upstream of the P₁ and P₂ promoters within the nuclear hypersensitivity element III₁ (NHE III₁), which controls 80% to 90% of c-myc transcription (12, 13). This DNA duplex element can slowly equilibrate between transcriptionally active forms (duplex and single stranded) to a silenced form (14). Single G to A mutations within PU27 destabilize quadruplex formation and induce a 3-fold increase in transcriptional activity, suggesting a role for quadruplexes in gene regulation and that their formation may be critical for transcriptional silencing (15–17).

It has been proposed that formation of quadruplex structure upstream of the c-myc promoter may play a role in modulating c-myc transcription; however, the biological implications of the PU27 sequence are currently unknown. This study characterizes the biological role of the PU27 genomic quadruplex-forming promoter sequence on the regulation of c-myc transcription and proposes that exogenous addition of synthetic PU27 ODNs induces leukemic cell death by downregulating c-myc expression.

U937 (acute myelogenous), HL60 (acute myelogenous), K562 (chronic myelogenous), CCRF-CEM (acute lymphocytic), and MOLT-4 (acute lymphoblastic) leukemia cells and CA46 and Raji (Burkitt's) lymphoma cells (American Type Culture Collection; identity confirmed by short tandem repeat analysis, isoenzymology, and cytochrome C subunit PCR assay) were maintained in RPMI media supplemented with 10% FBS and 100 U penicillin/streptomycin at 5% CO₂ and 37°C. Stromal cells derived from umbilical cord blood were maintained under the same conditions in Dulbecco's Modified Eagle's Medium media with 10% FBS and 100 U penicillin/streptomycin. The following ODNs were used: PU27 5′-TGGGGAGGGTGGGGAGGGTGGGGAAGG-3′ and MutPU27 5′-TGAGTAGCGTGAGCAGAGTGCGTAACG-3′. Synthesized ODNs (Oligos Etc.) were dissolved in RNAse/DNAse-free ultrapure distilled H₂O to a stock concentration of 500 μmol/L and boiled at 95°C for 5 minutes. Cells in logarithmic phase (2 × 10⁵ cells/mL) were treated with 10 μmol/L PU27 (final concentration unless otherwise specified) or the respective mutant control ODN, MutPU27, for various time intervals and used for subsequent biochemical analysis.

Cell proliferation of all leukemia cell lines and nontransformed stromal cells were assessed by MTT assay. Cells were treated in a 96-well plate with 1 to 15 μmol/L PU27 or MutPU27 for 72 hours (5 × 10³ cells per well) or 6 days (1 × 10³ cells per well). MTT reagent (100 μg/mL PBS) was added and cells were incubated at 37°C and 5% CO₂. After 4 hours, cells were lysed and the formazan product was detected at 570 nm.

PU27 and MutPU27 ODNs were annealed by boiling for 5 minutes, adding physiological buffer to final concentrations of 20 mmol/L KH₂PO₄ dibasic, 120 mmol/L KCl, 5 mmol/L MgCl₂, and slow cooling to room temperature. Annealed ODNs were dissolved in physiologic buffer at a concentration giving an absorbance at 260 nm of 0.8 (about 5 μmol/L). Structures of ODNs in solution were determined by circular dichroism (CD) spectroscopy. Spectra were recorded on a Jasco-810 spectropolarimeter (Jasco), using a quartz cell of 1 mm optical path length, an instrument scanning speed of 200 nm/min, response time of 2 seconds, and over a wavelength range of 340 to 220 nm. The spectra are representative of 3 average scans taken at 25°C and were baseline corrected for signal contributions due to buffer.

Cellular uptake of PU27 or MutPU27 was analyzed by fluorescence-activated cell sorting (FACS), Amnis Image Stream, and confocal microscopy analysis. U937 cells were incubated with 10 μmol/L of fluorescein isothiocyanate (FITC)-labeled PU27 or MutPU27 for 1 to 72 hours, washed with PBS, and analyzed immediately. For Amnis Image Stream analysis, a FACS method for imaging individual cells as they pass through a detector, cells incubated with FITC-labeled PU27 or MutPU27 were fixed with 3% paraformaldehyde, washed with PBS, and stained with the nuclear DRAQ 5 stain. Intensity of intracellular FITC-labeled ODNs and nuclear translocation (colocalization of FITC and DRAQ) were assessed.

ODN uptake was also compared in U937 cells and nontransformed stromal cells by confocal microscopy to determine whether uptake is exclusive to cancer cells. U937 or stromal cells were treated for 72 hours with FITC-labeled PU27 or MutPU27, washed with PBS, fixed in 4% paraformaldehyde, and visualized by confocal microscopy with an Olympus Fluoview FV500 laser scanning confocal microscope.

U937 cells treated with PU27 or MutPU27 for 6 to 72 hours were collected and washed in PBS. Isolated nuclei were stained with propidium iodide (PI; Cycletest Plus DNA Reagent; BD Biosciences) and analyzed by FACS analysis.

PU27 and MutPU27 were end labeled using [γ-³²P]-dATP with T₄ polynucleotide kinase and incubated in RPMI medium with 10% FBS at 37°C or in the presence of U937 nuclear or cytoplasmic (S100) cell extract for 0 to 72 hours. Cold PU27 or MutPU27 were added to give a final ODN concentration of 10 μmol/L. After heating in 98% formamide buffer at 65°C, ODNs were run on a 12% denaturing gel and analyzed by autoradiography.

Total RNA was extracted from U937 cells with TriZol (Invitrogen) reagent according to manufacturer's instructions. RNA purity was determined by A260/A280 ratio and quantified by A260. Preparation of cDNA and forward and reverse primers for c-myc and β-actin were as previously described (18). Real time (RT)-PCR was carried out for a uniform amount of cDNA using the Fast 7500 System (Applied Biosystems). Reactions were diluted 1:2 with SYBR Green I Master Mix (Applied Biosystems) and amplification by PCR was as follows: 1 repetition at 50°C for 2 minutes, 1 repetition at 95°C for 10 minutes, and 40 repetitions of 95°C for 15 seconds and 60°C for 1 minute. A no template control reaction was run for each gene to control for DNA contamination of RNA extracts. A dissociation curve was done to provide evidence for a single reaction product. Message of c-myc in PU27-treated cells was compared with untreated cells collected at the same time point to calculate an expression ratio.

Total cell lysates were separated by 4% to 15% SDS-Tris PAGE and electroblotted onto polyvinylidene difluoride membranes. The membranes were blocked in 5% nonfat dry milk and incubated overnight at 4°C with c-myc antibodies (Santa Cruz Biotechnology). After washing, the membranes were incubated with a horseradish peroxidase–conjugated secondary antibody. Proteins were visualized by standard chemiluminescence methods (GE Healthcare). Equal loading of proteins was verified by probing the membrane with a mouse monoclonal anti–β-actin primary antibody (Santa Cruz Biotechnology).

DNA strand breaks occurring during cleavage of DNA were measured with DNA Tunnel assay per manufacturer's instructions (Roche Applied Science). DNA strand breaks were labeled with terminal deoxynucleotidyl transferase (TDT), which catalyzes polymerization of labeled nucleotides to free 3′-OH DNA ends in a template-independent manner. Incorporated fluorescein was detected by microscopy. Positive (cells treated with DNAse I) and negative (no TDT treatment) controls were done in parallel.

U937 cells treated with 10 μmol/L PU27 or MutPU27 for 72 hours were collected, washed in cacodylate buffer, and immediately fixed in 3% glutaraldehyde in cacodylate buffer (pH 7.4) overnight at 4°C. After postfixation in 1% OsO₄ in cacodylate buffer, cell pellets were dehydrated in graded ethanol, embedded, and sections were cut (80 μm) with an ultramicrotome. Sections were stained with saturated aqueous uranium acetate and lead citrate and viewed with a Philips CM12 electron microscope operating at 60 KV.

After treatment with PU27 or MutPU27, cells were collected, washed with PBS, and incubated in binding buffer containing Annexin V–FITC conjugate and PI according to manufacturer's instructions (Apoptosis Detection Kit; Biovision). Annexin V binds to phosphatidylserine extruded on the surface of apoptotic cells, although PI is impermeable to viable cells and identifies cells with compromised plasma membranes. Changes in mitochondrial membrane potential were measured with the Mitocapture Mitochondrial Apoptosis Detection kit (Mitocapture; Biovision), which uses a cationic dye fluorescing red in healthy mitochondrial membrane impermeable cells and green in dying mitochondrial membrane permeable cells. All samples were analyzed by FACS.

Reactive oxygen species (ROS) generation was assessed as previously described (18) by 2′,7′-dichlorofluorescin diacetate (DCFH-DA; Sigma Chemical), a lipid-permeable nonfluorescent compound that when oxidized by intracellular ROS forms the fluorescent compound 2′,7′-dichlorofluorescein (DCF). U937 cells were plated in phenol red–free RPMI containing 10 μmol/L DCFH-DA and treated with PU27 or MutPU27. After 24 or 72 hours, the culture media and cells were collected. The medium was saved and the cells were washed with phenol red–free media and lysed in RIPA buffer. The fluorescence of DCF in lysates and media was read in a black plate at 538 nm emission and 485 nm excitation and the values were standardized by lysate protein concentration. Negative controls containing dimethyl sulfoxide instead of DCFH-DA showed negligible fluorescence.

Media ATP was determined by a luminescence assay system according to manufacturer's instructions (Promega). Culture media from cells treated with PU27 for 24 or 72 hours was collected and trichloroacetic acid was added to final concentration of 0.5% and kept at 4°C for 30 minutes. The media were neutralized with 4 volumes of 250 mmol/L Tris acetate (pH 7.75) and mixed with luciferin/luciferase ATP assay reagents. Luminescence was measured with a Zylux Femtometer FB12 luminometer and the resulting relative light units were standardized by cell number and expressed as a percentage of untreated cells.

All values represent mean ± SEM. Differences between treatments were determined by a nonpaired t test (2-tailed) with SigmaStat software 3.5. A probability level of P < 0.05 was used to indicate statistical significance.

To characterize the secondary structure of PU27 and MutPU27, ODNs were analyzed by CD spectroscopy. PU27 formed a quadruplex in physiologic buffer, represented by a peak absorbance at 260 nm and a trough absorbance at 240 nm (Fig. 1B). The mutant PU27 sequence did not form a quadruplex in solution.

Treatment of U937 cells with 1 to 15 μmol/L PU27 for 24 to 144 hours caused a dose- and time-dependent decrease in cell proliferation, culminating to 100% inhibition after 144 hours with 10 and 15 μmol/L and more than 90% inhibition with 5 μmol/L (Fig. 2A). Little change in cell proliferation occurred with MutPU27 treatment. Nontransformed stromal cells derived from human cord blood were treated identically to determine whether the antiproliferative activity of PU27 is cancer cell specific (Fig. 2B). No significant change in stromal cell proliferation was noted in response to PU27 (up to 15 μmol/L) after 72 or 144 hours of treatment. This suggests that growth inhibition by PU27 may be specific for leukemia cells but not nontransformed stromal cells. The effect of PU27 or MutPU27 dose-escalation (1–15 μmol/L) on cell proliferation was also compared in 7 leukemic cell lines (U937, HL60, CA46, K562, Raji, Molt-4, and CCRF-CEM) after 72 and 144 hours (Fig. 2C–F). There was significant dose-dependent inhibition of cell proliferation with PU27 in all 7 leukemic cell lines after 72 and 144 hours compared with untreated and cells treated with MutPU27. IC₅₀ values after 144 hours were U937 <5 μmol/L, HL60 <5 μmol/L, K562 <5 μmol/L, Raji <1 μmol/L, CA46 <1 μmol/L, Molt-4 <5 μmol/L, and CCRF-CEM <3 μmol/L). Little change was noted with MutPU27, however, some cell lines (Raji, CA46, and CCRF-CEM) responded to the highest concentrations.

FITC-PU27 was immediately taken into cells after 1 hour, with increased uptake after 72 hours (Fig. 3A). FITC-MutPU27 was minimally taken into cells after 1 hour, with little change in uptake after 72 hours. Confocal microscopy confirmed these results, showing marked uptake of FITC-PU27 (cytoplasmic and nuclear) into U937 cells after 72 hours, whereas FITC-MutPU27 was virtually undetectable at the same intensity settings (Fig. 3B). Although FITC-PU27 was prominently taken into nontransformed stromal cells, it seemed largely localized in the cytoplasm (Fig. 3C). Thus, PU27 was rapid internalized compared with MutPU27 and seems to be largely localized in the nucleus.

Amnis Image Stream was used to quantify the percentage of cells taking up FITC-PU27 and its subcellular localization. FITC-PU27 uptake was significantly greater than FITC-MutPU27, with 54% of cells taking up FITC-PU27 after 6 hours and 98% of cells after 24 hours compared with 2% and 61%, respectively, of FITC-MutPU27 (Fig. 3D). After 24 hours, 52% of cells had PU27 colocalized in the nucleus whereas only 3% of cells treated with MutPU27 had colocalization. However, after 72 hours, 100% of cells had PU27 or MutPU27 translocated in the nucleus.

To determine whether treatment with PU27 caused a change in cell size, Amnis Image Stream was used to assess the size of U937 cells treated with PU27 or MutPU27. There was no difference in cell size with PU27 treatment after 6 or 24 hours (6 hours, 9.44 μm; 24 hours, 10.66 μm) compared with MutPU27 (6 hours: 8.88 μm; 24 hours: 10.14 μm). However, PU27 treatment significantly increased cell size (13.44 μm) compared with MutPU27 (10.34 μm) after 72 hours. This size change is likely related to cell swelling and induction of cell death by PU27.

Because the therapeutic use of ODNs is limited by their susceptibility to nuclease degradation, we examined the stability of PU27 in serum and intracellularly. PU27 was inherently stable in serum and intracellularly (nucleus and cytoplasm) up to 72 hours compared with MutPU27, which significantly degraded after 4 hours (Fig. 4A). This remarkable increase in stability is undoubtedly related to nuclease resistance of quadruplex DNA.

To examine whether the antiproliferative effect of PU27 was partially due to cell-cycle perturbation, changes in the cell cycle after 6 to 72 hours were determined. PU27 treatment arrested cells in G₀–G₁ after 24 hours, which continued through 72 hours (Fig. 4B). Increased G₀–G₁ arrest was associated with a small shift in FL-2A (PI) staining, coincident with the increase in cell size.

PU27 treatment significantly increased staining of Annexin-positive cells after 24 hours (31.5%) and Annexin- and PI-positive cells (14.4%) compared with untreated and MutPU27 treatment (Fig. 4C). Although 25.9% of cells stained Annexin positive with PU27 after 72 hours, a significantly larger proportion stained Annexin and PI positive (64.1%).

Although no DNA fragmentation [terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) staining] was measured after 24 hours, PU27 treatment for 72 hours caused significant fragmentation compared with untreated or MutPU27 treatment (Fig. 4D).

PU27 treatment for 72 hours caused diffuse cell swelling, blister formation, condensation of chromatin into irregular clumps, and a compromised cell membrane, all characteristic of oncosis (Fig. 4E). Ultrastructural damage included significant mitochondrial swelling, widespread cytoplasmic vacuolization, and the appearance of autophagic fluid-filled vacuoles.

Treatment with PU27 for 24 hours increased mitochondrial membrane permeability, which indicated dissipation of the transmembrane potential (Fig. 5A). This was shown by a pronounced shift from red to green fluorescence of the cationic dye with PU27 compared with untreated or MutPU27 treatment.

The PU27-induced decrease in mitochondrial transmembrane potential after 24 hours corresponded with reduced cellular ATP production and greater total ROS production after 72 hours, indicated by greater oxidation of DCFH-DA (Fig. 5B and C).

PU27 treatment significantly decreased c-myc mRNA after 6 hours compared with untreated cells. These levels continued to decrease through 48 hours, culminating to a 90% decrease after 72 hours (Fig 6A). This is in marked contrast to cells treated with MutPU27, which showed little change in c-myc mRNA from 6 to 48 hours and a slight increase at 72 hours.

In general, c-myc protein expression paralleled c-myc mRNA levels. No changes in c-myc expression were measured after 6 hours of PU27 treatment, however, a significant decrease in c-myc occurred after 48 and 72 hours compared with untreated and MutPU27, coincident with the decrease in c-myc mRNA (Fig. 6B). This is in contrast to untreated and MutPU27-treated cells, which showed an increase in c-myc expression over time.

Summary of the proposed mechanism of PU27-mediated cell death in leukemia cells (Fig. 6C).

The promoters of several cancer-related genes, such as c-myc, contain a disproportionate number of sequences within nuclease hypersensitivity regions that are capable of forming quadruplex (4-stranded) DNA (10). There is strong evidence that the quadruplex-forming sequence within the c-myc promoter, PU27, plays an important role in downregulation of c-myc transcription, however, the biological significance of PU27 is currently unknown.

Although inhibition of gene expression has proven effective in promoting tumor regression in animal models, many biological barriers exist that prevent this approach from being used therapeutically. Conventional ODNs can be rapidly degraded by DNases and serum nucleases in biological fluids (19), are poorly transported into human cells (20), and have unwanted effects on normal cells (21). ODN uptake into hematologic cells is especially poor, as leukemic cells are often more resistant to uptake enhancement strategies (22). This study introduces a unique strategy for abrogating c-myc expression in leukemia that addresses all of the current limitations of ODN therapy. We used the genomic c-myc promoter sequence to treat leukemia cells in the absence of any uptake enhancement strategy and have shown that it is nontoxic to normal cells, abundantly taken up by leukemia cells, is remarkably stable in serum and intracellularly, and has potent antileukemic activity in a variety of leukemia cells.

We have shown that PU27 has potent biological activity in multiple leukemia cell lines, causing significant dose-dependent inhibition of cell proliferation after 72 hours and 6 days. PU27 is remarkably resistant to nuclease degradation in serum and in the presence of cellular extracts and is abundantly taken into cells after 1 hour in comparison with the mutant sequence, colocalizing in both the nucleus and cytoplasm. The cellular internalization and stability of PU27 is further shown by the prominent uptake and visualization of FITC-labeled PU27 after more than 14 days (unpublished results). Although PU27 was taken up by nontransformed stromal cells, no change in cell growth occurred after 6 days, suggesting that the antiproliferative effects of PU27 may be specific for transformed cells and have no adverse effects on normal cells. These results are consistent with previous studies, which showed that quadruplex-forming ODNs often show enhanced cellular uptake into cells, form inherent nuclease resistant structures, and have minimal effects on normal cells (23, 9). In general, G-quadruplex DNA has also been shown to accumulate and have greater retention in the nucleus compared with unstructured ODNs (24), which may be due to binding of specific targets inside the nucleus and/or G-quadruplex–binding proteins (25).

Coincident with the antiproliferative effect of PU27, treatment of U937 cells with PU27 induced prominent G₀–G₁ cell-cycle arrest beginning at 24 hours concurrent with significant attenuation of c-myc mRNA and protein expression. This is consistent with previous studies that showed that treatment of U937 cells with a small-molecule c-myc inhibitor markedly reduced c-myc protein expression after 24 hours, inducing G₀–G₁ cell-cycle arrest (26). Inhibition of HL60 acute myeloblastic leukemia cells with antisense c-myc fragments also effectively induced G₀–G₁ cell-cycle arrest after 48 hours, accompanied by reduced c-myc mRNA and protein expression (3). Therefore, downregulation of c-myc is an important early event related to growth arrest and occurs in a variety of cell types (27, 28).

In conjunction with G₀–G₁ cell-cycle arrest, treatment with PU27 induced cell death, which was accompanied by a significant increase in staining of Annexin-positive cells and a larger proportion of both Annexin- and PI-positive cells as well as significant DNA damage (29–31). Although these results initially suggest an apoptotic cell death, exposure of phosphatidylserine on the cell surface (recognized by Annexin V) and TUNEL-positive staining for DNA damage also occurs in oncotic necrosis (32). Oncotic necrosis is a proinflammatory pathway accompanied by cell and organelle swelling (oncosis) and significant membrane permeability (necrosis), a process of cell death related to oxidative stress and energy depletion that is morphologically and biochemically distinct from apoptosis. Apoptosis and oncosis can coexist or cells may undergo transition from apoptosis to oncosis, depending on their threshold for sustaining ion homeostasis and ATP production (32). Inhibition of glycolysis and ATP synthesis in mitochondria and increased plasma membrane permeability inactivates Na⁺/K⁺-ATPase causing accumulation of intracellular Na⁺ accompanied by Cl⁻ and water and increased intracellular Ca²⁺ from decreased Na⁺/Ca²⁺ exchange causing significant cell swelling (33). Analysis of PU27-treated cells by electron microscopy showed diffuse cell and organelle swelling, a compromised cell membrane with surface blebs and blisters, and significant ultrastructural damage including widespread cytoplasmic vacuolization, characteristic of oncotic necrosis, and autophagic fluid-filled vacuoles. This is much different from a noninflammatory apoptotic cell death, which is characterized by nuclear and cytoplasmic condensation, formation of apoptotic bodies, and maintenance of an intact plasma membrane (34).

In this study, induction of oncotic necrosis corresponded with significant changes in cellular energetics as shown by early disruption of mitochondrial membrane potential leading to suppression of ATP synthesis and ROS formation. This is consistent with c-myc's role in mitochondrial biogenesis and regulation of glucose metabolism. c-Myc transcriptionally upregulates target glycolytic genes involved in aerobically converting glucose to lactate (35, 36). Cell death by oncosis has previously been shown to occur in U937 cells in response to NaN₃, a chemical blocking ATP production at complex IV of the mitochondria inducing oxidative stress (32) as well as in K562 chronic myelogenous leukemia cells in response to sanguinarine (37) or arsenic trioxide (38).

Therefore, our studies suggest that treatment of U937 leukemia cells with exogenous addition of the genomic 27 bp PU27 sequence induces G₀–G₁ growth arrest and cell death by oncosis through a mechanism involving inhibition of c-myc mRNA and protein expression. Significant c-myc suppression corresponded with prominent mitochondrial dysfunction leading to loss of ATP production and enhanced oxidative stress. The nature of PU27-mediated inhibition of c-myc and its affiliation with the endogenous c-myc promoter is currently under investigation and may be related to either binding of PU27 to the quadruplex-forming homologous sequence upstream of the c-myc promoter or binding to the nonquadruplex-forming, complementary strand. This study represents a promising therapy for the treatment of multiple forms of leukemia and is the first report of selective cancer cell killing by a genomic DNA sequence.

No potential conflicts of interest were disclosed.

The authors thank Cathie G. Caple for assistance with electron microscopy sample preparation, Katherine Richardson Hobbing for assistance with mitochondrial potential measurements, and Magdelena Dailey and Nichola Garbett for assistance with CD spectroscopy and also thank Magdalena Kucia for assistance with stromal cell experiments.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.