Combinations of Tyrosine Kinase Inhibitor and ERAD Inhibitor Promote Oxidative Stress–Induced Apoptosis through ATF4 and KLF9 in Medullary Thyroid Cancer

Medullary thyroid cancer (MTC) originates from the neuroendocrine C cells of the thyroid gland, which secrete calcitonin. Activating mutations of the RET proto-oncogene are the most common drivers of sporadic MTC, identified in 40% to 50% of cases (1). Oncogenic RET protein activates a complex network of signal transduction pathways that contribute to cellular transformation, including the RAF/MAPK/ERK and phosphatidylinositol 3-kinase/AKT pathways (2, 3). MTC is resistant to chemotherapy-induced cell death. Currently, the tyrosine kinase inhibitors (TKI) cabozantinib and vandetanib are approved by the U.S. Food and Drug Administration for the treatment of advanced, progressive, or symptomatic MTC (4). However, responses to these agents are incomplete, often unsustainable and resistance develops (5, 6). These TKIs increase mitochondrial membrane potential and cause bioenergetics stress, and their combination with mitochondrial-targeted agents suppresses MTC tumor growth in mice (7). In addition, targeting mitochondrial chaperone such as HSPA9 (GRP75/Mortalin) suppresses human MTC cells in culture and in mouse xenografts (8).

Cancer cells with high secretory function, such as the parafollicular C cells that secrete calcitonin and other peptides, are more dependent on the quality control provided by the ubiquitin-proteasome system. Proteasome inhibitors that target the ubiquitin-proteasome system have been proven to be effective against hematologic malignancies, also active secretory cells (9). Thus, targeting protein homeostasis pathways could be exploited for the treatment of tumors with a high protein synthesis rate. Endoplasmic reticulum (ER)–associated protein degradation (ERAD) controls the quality of glycosylated and nonglycosylated protein in the ER and eliminates misfolded and unassembled proteins (10). Accumulation of unfolded proteins that are not resolved triggers cell death (11). Eeyarestatin I (hereafter “eeyarestatin”) is an ERAD inhibitor that disturbs ER homeostasis and has anticancer activities resembling that of the proteasome inhibitor bortezomib (12). Eeyareatatin directly binds the p97 ATPase, a key regulator of ERAD machinery, to inhibit p97 binding to the ER membrane and thereby induce cell death through ATF4, ATF3, and NOXA (13). One mechanism underlying the chemoresistance of MTC is that RET prevents apoptosis through phosphorylation-dependent inhibition of ATF4 transcriptional activity (14). We recently demonstrated that treatment of MTC cells with TKIs enhances ATF4 protein levels and activation of ATF4 target genes (14). ATF4 promotes the induction of apoptosis under persistent stress conditions, although the mechanism is not clearly understood. We hypothesized that the combination of an ATF4 inducer and a TKI causes excessive cellular oxidative stress resulting in the activation of apoptosis.

In the present study, we demonstrated that combinations of TKIs and eeyarestatin synergistically induced apoptosis in MTC cells in vitro through an increase in reactive oxygen species (ROS) and upregulation of ATF4 and ATF4 target genes. These findings provide a preclinical rationale for further evaluation of combinations of TKIs and ERAD inhibitors for MTC in a clinical trial. This combination therapy requires lower doses of TKIs than are required with single-agent TKI therapy and may increase efficacy and reduce side effects while preventing resistance to TKIs and thereby enhancing the survival of patients with MTC.

Eeyarestatin, sunitinib, and vandetanib were purchased from Tocris Bioscience. Antibodies specific to BBC3 (D30C10) and MCL1 (D35A5) were purchased from Cell Signaling Technology. PMAIP1 (114C307) antibody was purchased from Calbiochem. ATF4 (D4B8) antibody was purchased from Abcam. RET (C-20), SIAH1 (N-15), and MNK2 (S-20) antibodies were purchased from Santa Cruz Biotechnology. Anti-active caspase-3 and anti-cleaved PARP (ASP214) were purchased from BD Biosciences.

Two MTC cell lines, TT and MZCRC1, were used. TT cells were purchased from ATCC. MZCRC1 cells were kindly provided by Dr. Alex Knuth (Switzerland) and were previously described (15–17). RET mutations in MTC cell lines were verified by sequencing; the codon 918 in exon 16 of RET gene (methionine is mutated to threonine) RET-M918T in MZCRC1 cells and the codon 634 in exon 11 of RET gene (cysteine is mutated to tryptophan) RET-C634W in TT cells. All cell lines were tested negative for Mycoplasma using the service provided by the core facility at the MD Anderson Cancer Center (MDACC). Cells used for the experiments were between 2 and 5 passages from thawing.

Lentiviral vectors (pLKO.1) containing ATF4-specific shRNAs were purchased from Sigma-Aldrich. Lentiviral ATF4-shRNA plasmids were cotransfected into HEK293T cells along with packaging (VPR8.9) and envelope (VSV-G) plasmids using X-tremeGENE (Roche) for 2 days. The virus particles containing ATF4-shRNA or control shRNA were used to infect TT cells. Transfected cells were selected in media containing 2 μg/mL puromycin (Clontech).

For cell viability assay, 40,000 cells were plated per well in a 96-well plate and treated as indicated with various drug combinations for different time points. Then, the cells were treated with 200 μL of 0.6 mg/mL MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, a tetrazole) in serum-free medium for 4 hours, and then solubilized in dimethyl sulfoxide for 30 minutes and absorbance was measured with a spectrophotometer (Synergy HT; BioTek) at 595 nm. The percentage of apoptotic cells was measured using a BD ApoAlert Annexin V–FITC apoptosis kit (BD Biosciences) according to the manufacturer's instructions, and cells were analyzed on a FACscan system (Becton Dickinson). Additionally, cells were fixed, stained with antibodies against cleaved caspase-3 and cleaved PARP, and analyzed by flow cytometry.

The degree of synergism between the eeyarestatin and TKIs was determined using CompuSyn software, version 1 (ComboSyn, Inc.; ref. 18). The combination index for each combination was calculated at a nonconstant ratio and was used to express synergism (CI< 1), additivity (CI = 1), or antagonism (CI > 1). Normalized isobologram for nonconstant combinations ratios is presented.

Cells were collected after trypsinization, washed with serum-free medium, and resuspended in 500 μL of serum-free medium containing 10 μmol/L of the redox-sensitive fluorescent dye 5-(and 6)-carboxy-2′, 7′-dichlorodihydrofluorescein diacetate (carboxy-H2DCFDA; Sigma). Cells were incubated at 37°C for 30 minutes, pelleted at 1,200 rpm for 5 minutes, and resuspended in 500 μL of serum-free medium. Samples were analyzed on a Beckman Coulter Gallios flow cytometer. Data were analyzed with FlowJo software version 10.3, and mean fluorescence intensity was obtained.

RNA and cDNA were prepared using a Cells-to-CT kit (Invitrogen). Quantitative PCR was performed with a polymerase-activating/DNA-denaturing step of 95°C for 15 seconds and 60°C annealing/elongation step for 1 minute on a StepOne real-time PCR system (Applied Biosystems). TaqMan primer probes were purchased from Applied Biosystems. Gene-expression values were normalized to HPRT mRNA.

MZCRC1 cells were treated with eeyarestatin (5 μmol/L) for 16 hours, and ChIP assay was performed as described previously (14). For ChIP-seq, sequencing libraries are made up of random fragments that represent the entire sample. Libraries are created by shearing DNA (Covaris S220) into 150-bp to 400-bp fragments. These fragments are ligated to specific adapters. Library fragments of the appropriate size are then selected and isolated. Following a sample clean-up step, the resulting library is quantified by qPCR and checked for quality using the Agilent Bioanalyzer.

The libraries were pooled and sequenced on the Illumina Hiseq-2000 platform at the sequencing and microarray facility of The University of Texas MD Anderson Cancer Center. ChIP-seq reads were aligned to the human genome assembly, and the ChIP-seq peaks were called using MACS 2.0. Regions of enrichment comparing ChIP and input control signal exceeding q < 0.01 were called as peaks. The primers used to amplify the indicated genes are shown in Supplementary Table S1.

Western blot analysis and immunoprecipitation were performed as described previously (19).

All data were expressed as mean ± standard deviation. Data were analyzed with GraphPad Prism7 software using the indicated tests. Statistical significance was indicated as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

ERAD eliminates misfolded or unassembled proteins from the ER. ERAD targets are exported from the ER to the cytoplasm and degraded by the ubiquitin-proteasome system (20). Eeyarestatins are ERAD-specific inhibitors that elicit an integrated stress response program at the ER resulting in activation of ATF4, which subsequently stimulates expression of the apoptotic gene PMAIP1 (NOXA), which results in cell death (12, 13). Because of the secretory nature of MTC cells, we hypothesized that a compound that induces ER stress and overwhelms the hyperactive secretory pathway in MTC cells will induce apoptotic cell death. Eeyarestatin has previously been reported to induce cell death in various other cancers (12). We first tested whether eeyarestatin could induce cell death in the MTC cell lines TT and MZCRC1. The viability of MTC cells treated with eeyarestatin at increasing concentration for 48 hours was measured by a colorometric assay using MTT and annexin V/FITC staining. Eeyarestatin induced cell death in these cells but MZCRC1 cells were less sensitive than TT cells (Fig. 1A and B). Furthermore, in both MTC cell lines, treatment with eeyarestatin increased mRNA levels of genes involved in the UPR and apoptotic pathways, including ATF4, DDIT3 (CHOP), TRIB3, SESN2, BBC3 (PUMA), and PMAIP1 (NOXA) with greater induction in TT cells compared with MZCRC1 cells (Fig. 1C). The mRNA level of RET proto-oncogene was not affected (Fig. 1C). We previously reported that ATF4 overexpression promotes RET degradation (21). Thus, we reasoned that eeyarestatin might promote RET degradation through ATF4 activation. As expected, in eeyrarestatin-treated TT cells, the level of ATF4 protein increased while the level of RET protein completely abrogated (Fig. 1D). We observed similar results in EE-treated MZCRC1 cells but the effect of eeyarestatin on the levels of ATF4 and RET was moderate compared with TT cells (Fig. 1E).

Although the TKIs sunitinib and vandetanib are effective for the treatment of MTC, patients often develop resistance to these drugs (5, 6, 22). We previously reported that vandetanib increases ATF4 transcriptional activity (14). Transcriptional induction of ATF4 and CHOP increases protein synthesis, leading to oxidative stress and cell death (23). ROS accumulation has been reported to trigger ER stress and the UPR, which in turn induce apoptosis through activation of ER stress–mediated apoptotic pathways (24, 25). We hypothesized that eeyarestatin inhibits the degradation of misfolded proteins, leading to severe UPR and that the combination of eeyarestatin with a TKI synergistically enhances ER stress and ROS accumulation, leading to apoptosis. Using MTT assay, we observed decreased viability of MTC cells (TT and MZCRC1) treated with various doses of eeyarestatin alone (0.1–6 μmol/L) or sunitinib or vandetanib alone (0.1–1 μmol/L) and increased cytotoxicity when cells were treated with combinations of eeyarestatin and either sunitinib or vandetanib (Fig. 1F and G). Calculation of the drug combination index showed that the combinations of eeyarestatin with sunitinib or vandetanib had synergistic cytotoxic effects (Supplementary Table S1). Normalized isobologram with nonconstant combination ratios are shown for the combination of eeyarestatin and vandetanib in TT (Fig. 1H and I).

To investigate whether the combination of eeyarestatin and a TKI causes apoptosis, we quantified the fraction of apoptotic cells by counting cleaved caspase-3–positive and cleaved PARP–positive cells by flow cytometry. Cotreatment of TT cells with eeyarestatin and sunitinib or vandetanib triggered substantial apoptosis in a manner that was not observed after treatment with eeyarestatin, sunitinib, or vandetanib alone (Fig. 2A and B). For example, cells treated with 4 μmol/L eeyarestatin, vandetanib or sunitinib alone induced 11.4% ± 1.98% (n = 3); 7.24% ± 3.37% (n = 3); 16.45% ± 5% (n = 3) cell death, respectively (Fig. 2B). Combination treatment with 4 μmol/L of eeyarestatin and 4 μmol/L vandetanib or sunitinib for 24 hours yielded 52% ± 9.2%: 48% ± 0.4% cell death, respectively (Fig. 2B). The MZCRC1 cells showed less sensitivity to the combination of vandetanib with eeyarestatin compared with TT cells (Supplementary Fig. S1). MTC cell lines contain a self-renewing CD133 population that is dependent on RET proto-oncogene. MZCRC1 cells, which harbor an M918RET mutation, have a greater CD133 cell number and sphere-forming ability than the TT cell line, which harbors C634W mutation (15). We sought to test the combination therapy in MZCRC1 spheres. We found that the combination of eeyaresatin and sunitinib caused asynergistic cell death on MZCRC1 spheres more efficiently than monolayer cells (Supplementary Fig. S2).

We have previously reported that vandetanib treatment to MTC cells led to increased expression of ATF4 (14). We next examined whether ATF4 knockdown in TT cells decreased drug-induced apoptosis. TT-shRNA-ATF4 stable cell lines were previously described (21). We showed that in ATF4-depleted TT cells (shRNA-ATF4 #74 and #76), apoptotic response to either eeyarestatin or the combination of eeyarestatin and sunitinib or vandetanib was significantly decreased compared with the response in parental cells without ATF4 depletion (Fig. 2A and B). Western blot analysis confirmed an increased expression of ATF4 and its proapoptotic targets BBC3 (PUMA) and PMAIP1 (NOXA) in TT cells treated with sunitinib and eeyarestatin than in TT cells treated with a single drug (Fig. 2C). In contrast, expression levels of survival factor MCL1 decreased significantly more in cells treated with the combination than in cells with single-drug treatment (Fig. 2C). These results suggested that the combination of ATF4 inducer and a TKI causes synergistic activation of apoptosis.

We next determined whether eeyarestatin and TKIs cooperatively induced oxidative stress in MTC cells. We detected significantly increased carboxy-H2DFDA fluorescence in sunitinib-treated or eeyarestatin-treated cells compared with control cells (Fig. 3A). The combination of sunitinib and eeyarestatin triggered a greater increase in fluorescence (Fig. 3A and B). We reasoned that ATF4 induction contributes to the effects of this drug combination. Indeed, ATF4 knockdown abolished ROS accumulation, which was induced by sunitinib, eeyarestatin, and their combination (Fig. 3A and B). Enhanced ROS accumulation was also observed in MZCRC1 cells treated with eeyarestatin, sunitinib, or vandetanib and their combination (Fig. 3C). Furthermore, ROS scavenger (n-acetyl cysteine (NAC) decreased ROS accumulation and cell death in MTC cells treated with eeyarestatin or vandetanib (Fig. 3C). Together, these results suggest that excessive oxidative stress, a physiologic response triggered by these drugs, is required for the observed cytotoxicity (Fig. 3C).

To identify the ATF4 transcriptional targets involved in oxidative stress, we performed ChIP-seq assay in MZCRC1 cells treated with eeyarestatin. Chromatins were prepared from treated cells and immunoprecipitated with antibodies specific for ATF4 or H3K9AC, an active histone marker that correlates with gene activation. Precipitated DNA was purified and then used for sequencing with the Illumina Hiseq-2000 platform; the resulting sequences were aligned to the human genome. Peaks containing ATF4 and H3K9AC binding regions were viewed with Integrative Genomics Viewer version 2.3. Relative enrichment and density mapping plots indicated that ATF4 and H3K9AC binding sites were enriched around the promoter regions (Fig. 4A). A total of 512 unique peaks were found after removal of nonspecific genes (26 peaks; P = 1.00E−07 and false discovery rate 0.5). Analysis of distribution of these 512 unique peaks in the genome showed that 18.4% of the peaks were upstream of the transcription start site, 22.1% were at the promoter, 2.9% were in the exon region, 29.1% were in the intronic region, 2.1% were in the transcription end site, 8.6% were downstream of the transcription start site, and 16.8% were in distant regions (Fig. 4B). Moreover, 52% of the ATF4 peaks overlapped with H3K9AC peaks, mostly in the promoter regions, and 55.3% of H3K9AC peaks (a total of 22,339 peaks) lay at the promoter region. Next, we performed motif analyses of the ATF4 binding sites using the MEME-ChIP program. The ATGC frequency in the canonical motifs is shown for ATF4 binding sites to demonstrate motif conservation (Fig. 4C). Taken together, these ChIP-seq findings provided a precise genome-wide view of ATF4 binding upon ER stress induced by eeyarestatin. Ingenuity Pathway Analysis of the ATF4 target genes revealed enrichment of genes involved in the UPR, ER stress pathway, EIF2 signaling, cell death pathway, ERK/MAPK signaling, p38 MAPK signaling, PI3kinase mTOR pathways, TR/RXR activation, and NRF2-mediated oxidative stress response (Supplementary Fig. S3; Supplementary Table S3).

The occupancy of ATF4 at the promoter of several known (DDIT3, BBC3) and previously unknown target genes (SIAH1, MKNK2, LZTFL1) was further validated by manual ChIP in MZCRC1 cells treated with eeyarestatin, vandetanib, or their combination (Fig. 4D; Supplementary Table S3 and S4). Leucine zipper transcription factor-like 1 (LZTFL1) has been shown to inhibit TGFβ-activated MAPK, hedgehog signaling, and β-catenin–Wnt signaling pathway, acting as a tumor suppressor gene in lung and gastric cancers (26). MKNK2 is downregulated in human cancers and is a tumor suppressor that activates p38-MAPK, leading to p38-mediated cell death (27). SIAH1 is a p53-inducible mediator of cell-cycle arrest, tumor suppression, and apoptosis (28). We found that the combination of eeyarestatin and vandetanib increased the occupancy of ATF4 at the promoter of DDIT3, BBC3, SIAH1, MKNK2, and LZTFL1 genes more than either single treatment did (Fig. 4D). Gene-expression analyses showed that in TT or MZCRC1 cells treated with eeyarestatin, sunitinib, or vandetanib, as well as their combination, the mRNA levels of ATF4 and ATF4 target genes were more induced with the combination of these drugs than sunitinib, vandetanib, and eeyarestatin alone (Fig. 5A–C)). Western blot analysis further confirmed increased protein levels of LZTFL1, SIAH1, and MKNK2 in MZCRC1 cells treated with eeyarestatin, sunitinib, and the two drugs in combination (Fig. 5D).

ChIP-seq analysis revealed that ATF4 binds to the promoter and enhancer region of KLF9 (Fig. 6A–C). KLF9 is a member of the SP1 C2H2-type zinc finger family of transcription factors and is ubiquitously expressed in various tissues (29). KLF9 is upregulated by oxidative stress, which causes accumulation of ROS and promotes oxidative stress–induced cell death (30). Using ChIP assay, we confirmed the ATF4 binding to the promoter of KLF9 using the specific primers flanking the ATF4 peaks within the promoter and enhancer regions. We found that ATF4 occupancy at the KLF9 promoter or enhancer regions was significantly increased in MZCRC1 cells treated with eeyarestatin alone or vandetanib alone compared with untreated cells (Fig. 6B and C). The combination increased the ATF4 occupancy compared with single-drug treatment (Fig. 6B and C). Expression analysis by quantitative RT-PCR revealed that the combination treatment had enhanced expression levels of the KLF9 gene compared with single-drug treatment in MZCRC1 cells (Fig. 6D). To determine whether upregulation of ATF4 plays a functional role in the induction of KLF9 upon treatment, ATF4 was stably downregulated in TT and MZCRC1 cells by shRNA, and then cells were treated with eeyarestatin (Fig. 6E–G). ATF4 knockdown prevented EE-induced upregulation of KLF9 gene (Fig. 6F and G) and oxidative stress–induced cell death (Fig. 3B). Western blot analysis confirmed that KLF9 expression was upregulated by eeyarestatin and sunitinib alone and further upregulated by the combination (Fig. 6H). Altogether, our results indicate that ATF4 induction activates the transcription of KLF9 gene and induces oxidative stress–mediated cell death.

In the work reported herein, we identified a promising drug combination that functions by triggering irresolvable oxidative stress and apoptosis in response to ER stress. Our data demonstrate that in MTC cells, combinations of the ERAD inhibitor eeyarestatin and a TKI, either sunitinib or vandetanib, synergistically induced a high level of ATF4 expression. Among the ATF4 target genes activated by this drug combination, we identified KLF9, which has been reported to regulate intracellular ROS (31), as a novel ATF4 target gene.

ROS production in cancer cells is one of the mechanisms underlying synergistic cytotoxicity of combination therapies (32, 33). Cancer cells exhibit higher levels of ROS than normal cells, and these higher levels of ROS increase the sensitivity of cancer cells to drugs that further increase ROS levels and oxidative stress (34). In response to oxidative stress, accumulation of unfolded or misfolded protein triggers ER stress, thereby enabling amplification loops that may contribute to the switch from adaptive to fatal UPR (35). Our findings showed that both vandetanib and eeyarestatin increased ATF4 recruitment to the KLF9 promoter and that recruitment was even greater when the two drugs were combined, which consequently results in induction of KLF9 at both mRNA and proteins levels and ROS production, leading to increased apoptosis. It has been reported that KLF9 causes intracellular ROS accumulation partly by suppressing the transcription of the thioredoxin reductase 2 gene (TXNRD2), which plays a critical role in defense against oxidative damage (30). It is shown that cytotoxicity mediated by the stress-inducing agents, the histone deacetylase inhibitor panobinostat and the proteasome inhibitor bortezomib is through increased expression of KLF9 (36). Our data suggest a model of ATF4-dependent regulation of ROS through upregulation of KLF9 that is sufficient to cause cell death. The similar feed-forward regulatory loop is previously reported in response to DNA damage or oxidative stress (37, 38).

It has been shown that ATF4 mediates ER stress–induced apoptosis in response to compounds triggering ER stress, such as an inhibitor of ATPase p97 (39); HA15 (a thiazole benzenesulfonamide), which targets GRP78/Bip (40); fenretinide; and the proteasome inhibitor bortezomib (41). Furthermore, ATF4 knockdown reduced vemurafenib-induced apoptosis (42).

ERAD inhibitor eeyarestatin directly binds to the p97 ATPase complex, a key regulator of protein homeostasis, and inhibits deubiquitination of substrates, leading to accumulation of polyubiquitinated proteins in cells, which induces ER stress and subsequent cell death (39). P97 inhibitor exerts strong antitumor activity in several solid tumors by inducing UPR and irresolvable proteotoxic stress leading to cancer cell death (39). Interestingly, eeyarestatin seems to induce moderate ER stress in normal cells without inducing cell death, supporting the idea that cancer cells are more sensitive than normal cells to perturbations of ER homeostasis (43). Because MTC originates from the C cells of the thyroid gland, which secrete calcitonin, MTC cells are highly reliant on their UPR and ERAD pathways as a result of their high protein synthesis burden; this may explain why MTC cells are more sensitive to ER stress inducers than are cancer cells with no secretory function. Thus, increasing ATF4 expression in MTC tumors may constitute a therapeutic strategy for inhibiting tumor growth through increasing the expression of tumor suppressor genes.

In summary, as increased ER stress is common in most human tumors, inducing ER stress and oxidative stress sufficiently to induce death of cancer cells but not normal cells could be a useful new approach to cancer treatment.

S.I. Sherman is a consultant/advisory board member for LOXO Oncology and Genzyme. R.F. Gagel reports receiving a commercial research grant from Genzyme. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R. Bagheri-Yarmand, K.M. Sinha, R.F. Gagel

Development of methodology: R. Bagheri-Yarmand, K.M. Sinha, L. Li, R.F. Gagel

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Bagheri-Yarmand, K.M. Sinha, G.J. Cote, R.F. Gagel

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Bagheri-Yarmand, K.M. Sinha, Y. Lu, G.J. Cote, R.F. Gagel

Writing, review, and/or revision of the manuscript: R. Bagheri-Yarmand, K.M. Sinha, G.J. Cote, S.I. Sherman, R.F. Gagel

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Bagheri-Yarmand, R.F. Gagel

Study supervision: R. Bagheri-Yarmand, S.I. Sherman, R.F. Gagel

This work was supported by the NIH through grants P50CA168505/DRP (to R. Bagheri-Yarmand), Cancer Center Support Grant (CA016672), Kosberg Foundation (to R.F. Gagel). We thank the Department of Scientific Publication at MD Anderson for editing the manuscript.

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.