Dysfunction of Poly (ADP-Ribose) Glycohydrolase Induces a Synthetic Lethal Effect in Dual Specificity Phosphatase 22-Deficient Lung Cancer Cells

Poly (ADP-ribosylation) is a posttranslational modification by which some PARP family proteins catalyze the transfer of ADP-ribose to target proteins in a nicotinamide adenine dinucleotide (NAD⁺)-dependent manner (1, 2). This reaction is involved in various biological processes, including cell death, chromatin regulation, and DNA repair of single-strand breaks (SSB) and double-strand breaks (DSB; refs. 1, 3). PARP inhibitors were recently developed as a novel anticancer agent based on the concept of synthetic lethality (4, 5). PARP inhibitors selectively induce cell death in homologous recombination repair (HRR)–deficient cancers such as those associated with mutations in BRCA1/2 (6, 7), RAD51 (8), and PTEN (9). The development of novel anticancer agents based on the concept of synthetic lethality is a valuable cancer chemotherapy strategy because these drugs show increased tumor selectivity with reduced adverse effects on normal cells (4).

Poly (ADP-ribose) (PAR) synthesized by PARP is rapidly degraded to ADP-ribose by poly (ADP-ribose) glycohydrolase (PARG; ref. 10) and ADP-ribosyl hydrolase (ARH3; ref. 11). PARG is the main enzyme catabolizing PAR to ADP-ribose through its endo- and exo-glycohydrolase activities (12). As reported previously, PARG is required for the efficient repair of DSBs and SSBs (13). PARG deficiency induces PAR accumulation and a delay of DNA repair (14, 15). PAR accumulation induces cell death (parthanatos) accompanied by the translocation of apoptosis inducing factor from mitochondria to nuclei, leading to fragmentation of large-sized DNA in neuronal cells and cancer cells such as HeLa cells (16, 17). As previously reported, certain human cancer cell lines with PARG knockdown synergistically show higher sensitivity to alkylating agents (14, 18) and cisplatin treatment (18). PARG hypomorphic mouse embryonic stem cells with residual 10% PARG activity did not exhibit growth defect but showed higher sensitivity to alkylating agents, cisplatin, photon, and particle beam irradiation compared with wild-type embryonic stem cells (14, 19, 20). In addition, BRCA2 (21) and Bruton tyrosine kinase (BTK; ref. 22) defects increase PARG inhibition-induced cytotoxicity (21–23). These findings led us to hypothesize that PARG could serve as a novel therapeutic target for anticancer agents in both monotherapy and combination therapy with radiotherapy or other DNA targeting chemotherapeutic agents for particular types of cancers. Recently, PARG inhibitors such as phenolic hydrazide hydrazones (24), rhodanine-based PARG inhibitors (RBPI; ref. 25), xanthene compounds (26), ADP-HPD (27), and PDD00017273 (28), which has an IC₅₀ in the sub-microM range, have been developed. However, specific and potent PARG inhibitors for clinical applications remain to be developed.

Here, we screened genes whose deficiency enhances sensitivity in a synthetic lethal manner to develop a novel anticancer agent targeting PARG. We identified dual specificity phosphatase 22 (DUSP22) as such a novel gene. Synthetic lethality induced by PARG and DUSP22 dysfunction in lung cancer cells led to TP63-dependent apoptosis by upregulating p53 upregulated modulator of apoptosis (PUMA). Double knockdown of PARG and DUSP22 inhibited tumor growth in a mouse xenograft model. These results indicated that alterations in DUSP22 expression levels may serve as a predictive biomarker for PARG inhibitors.

The TRHmPARG#8 cell line is a tetracycline (Tc)-inducible PARG knockdown strain derived from the human T-REx HeLa cell line described previously (29). TRHmPARG#8 and PC14 were cultured in Minimum Essential Medium and DMEM (Thermo Fisher Scientific), respectively. A549 and SBC5 cells were grown in RPMI1640 (Thermo Fisher Scientific). Media were supplemented with 10% FBS (Gibco) and 1% penicillin–streptomycin (Invitrogen) as needed. Cells were maintained in a humidified atmosphere with 5% CO₂ at 37°C. The cell line A549 was obtained from the ATCC. The cell line PC-14 was obtained from Dr. Hayata, Tokyo Medical College (Tokyo Japan). The cell line SBC-5 was obtained from Okayama University in 1994. Cell line authentication of all cell lines was performed by short tandem repeat (STR)-PCR (Promega, August 2018). Mycoplasma testing was carried out using e-Myco plus Mycoplasma PCR Detection Kit (iNtRON Biotechnology) for all cell lines used in this study and all cell lines were mycoplasma free. All cell lines were passaged less than 15 times prior to use.

Negative screening was performed using the Decode RNAi Pooled Lentiviral shRNA Screening Libraries: Annotated Genome Negative Selection Kit (Thermo Fisher Scientific). TRHmPARG#8 cells were infected with a lentiviral siRNA expression library (Thermo Fisher Scientific) using the TransDux reagent (System Biosciences). GFP-positive cells were selected using puromycin for 3 days and divided into 2 populations. Cells were cultured for 6 days in the presence or absence of 40 ng/mL Tc, and genomic DNA was purified from 2 populations using a DNA Purification Kit (Dojindo). Amplification of barcode sequences in genomic DNA and purification of DNA were performed using the Decode shRNA Negative Selection Kit (Thermo Fisher Scientific) and Gene JET PCR Purification Kit (Thermo Fisher Scientific), respectively, as recommended by the manufacturer. Then, the genomic DNA was labeled using a Genomic DNA Enzymatic Labeling Kit (Agilent Technologies) and purified using Amicon Ultra-0.5 mL Centrifugal Filters (Millipore). The labeled barcode sequences were hybridized to microarray slides for 17 hours, and the slides were washed according to the Agilent CpG microarray protocol.

Cells were seeded onto 6-well plates or 24-well plates. Transfection with siRNA was performed using Lipofectamine RNAi MAX (Life Technologies) according to the manufacturer's protocol. Individual siRNAs were used at final concentration of 10 nmol/L in Opti-MEM. siRNAs (PARG#2, DUSP22#2, PUMA, TP63) targeting DUSP22, PARG, PUMA, and TP63 were purchased from Integrated DNA Technologies. The siRNA sequence of DUSP22#1 is based on shRNA sequence of oligo ID:V2LHS_225030 in the Decode RNAi Pooled Lentiviral shRNA Screening Libraries and it was constructed from Integrated DNA Technologies. PARG#1 siRNA was obtained as described previously (14). DS NC1 siRNA (Integrated DNA Technologies) and scrambled siRNA (Ambion/Applied Biosystems) were used as negative controls (N.C.).

RNA was prepared from each individual cell line and reverse transcribed using a High Capacity Reverse Transcription Kit (Thermo Fisher Scientific). The qRT-PCR analysis was performed using SYBR Green with the CFX96 Real-Time System (Bio-Rad). The mRNA levels were normalized to GUSB mRNA. The sequences of primer pairs are listed in Supplementary Table S1.

Cell viability was measured using the Cell Counting Kit-8 (Dojindo Laboratories) according to the manufacturer's instructions. Cells were seeded onto 96-well plates and cultured for 1 week. Cell proliferation rate was determined using the Cell Counting Kit-8 containing water soluble tetrazolium dys (WST-8). Plates were analyzed using a microtiter plate reader at 450 nm with a reference of 600 nm.

Western blotting was performed as described previously (29). Cell extracts were prepared with Laemmli's buffer. Proteins were separated by SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes. The following antibodies were used for immunoblotting: anti-PARG (Millipore), anti-p-p38 (Cell Signaling Technology), anti-β-actin (Sigma-Aldrich), anti-DUSP22 (Gene Tex), anti-AKT (pan; C67E7; Cell Signaling Technology), anti-p-AKT (Ser473; Cell Signaling Technology), anti-p-mTOR (Ser2448; Cell Signaling Technology), anti-p-mTOR (Ser2481; Cell Signaling Technology), anti-mTOR (7C10; Cell Signaling Technology), anti-PTEN (138G6; Cell Signaling Technology), anti-p-PTEN (Ser380; Cell Signaling Technology). Immune complexes were visualized using a horseradish peroxidase-linked secondary antibody and enhanced chemiluminescence (Millipore). Image quantification was performed with the ImageJ software (NIH).

Cells were transfected with siRNA against the target gene and cultured in 6-well plates for 9 days. Colonies were fixed with 4% neutralized formalin and stained with 0.02% crystal violet for counting.

Cell-cycle distribution was analyzed by flow cytometry as previously described (30). Cells were fixed with 70% ethanol. Fixed cells were incubated with PBS containing 50 μg/mL propidium iodide (Sigma-Aldrich) and 20 μg/mL RNase A (Sigma-Aldrich) for 2 hours and analyzed using the FACSCalibur system (Becton–Dickinson).

A549 cells were transfected with siRNA against PARG and/or DUSP22 and cultured for 3 days. Cell lysates were analyzed with PathScan Stress and Apoptosis Signaling Antibody Array (Cell Signaling Technology, #12856). This array was performed according to the manufacturer's instructions.

A549 cells were transfected with siRNA against PARG and/or DUSP22 or control siRNA. On the next day, cells (2.9 × 10⁵) were mixed with Growth Factor Reduced Matrigel (BD Biosciences) and injected subcutaneously into both legs of 11-week-old Balb/c-nu/nu nude mice. Tumor diameters were measured every 3 days with micrometer calipers, and tumor volume was calculated using the following formula: (smallest diameter) × (largest diameter) × (height)/2. All animal studies were approved by the Animal Experimental Committee of the National Cancer Center and performed following the Guidelines for Animal Experiments of the National Cancer Center, which meet the ethical guidelines for experimental animals in Japan.

Data were expressed as meant ± SE. Statistical significance was indicated when P value was less than 0.05. In this study, data were analyzed using Tukey test or Mann–Whitney U test. Synergistic effects were analyzed by 2-way ANOVA.

A comprehensive analysis was performed to identify synthetic lethal genes related to PARG dysfunction and understand the function of PARG in cancer cells. For this purpose, a siRNA library was screened using a negative screening strategy. Inducible PARG knockdown T-REx HeLa cells (TRHmPARG#8) were established in which PARG knockdown was induced in the presence of Tc (29). The cells were infected with lentiviral shRNA pools targeting approximately 10,000 genes. The relative abundance of individual shRNAs after PARG knockdown was determined by microarray analysis (Fig. 1A). Seventeen candidate siRNAs that suppressed the growth of PARG knockdown cells were identified at a 4-fold or higher signal rate (Fig. 1B; Table 1). The targeted genes were classified into functional categories, including metabolism, signal transduction, and posttranslational modification (Fig. 1B; Table 1). Among these genes, we focused on DUSP22, because DUSP22 mutations are reported frequently in lung, colon, and other tumors in the CanSAR database (Fig. 1C), and its expression is downregulated in certain cancers (31, 32). To determine whether dysfunction of PARG and DUSP22 exerts a synthetic lethal effect in cancer cells, siRNA against DUSP22 was introduced into the inducible T-REx HeLa cells. As shown in Fig. 1D, a, PARG expression was decreased to approximately 50% of the control in the presence of Tc and transfection of cells with siRNA-targeting DUSP22 decreased DUSP22 mRNA to approximately 10% of control levels (Fig. 1D,b). DUSP22 protein levels decreased to 29% in DUSP22 knockdown cells (Fig. 1D, c). DUSP22 belongs to the DUSP family of proteins, which function in the dephosphorylation of JNK, p38, and ERK (33). Inducible PARG knockdown T-REx HeLa cells were transfected with siRNA against DUSP22 or control siRNA, and the phosphorylation level of p38 was determined using a phospho-p38 antibody. The results showed that p38 phosphorylation levels were 2.3-fold higher in DUSP22 knockdown cells than in control cells (Fig. 1D, c), suggesting that DUSP22 was necessary for the dephosphorylation of p38. To determine whether dysfunction of PARG and DUSP22 induces synthetic lethality, the survival rates of PARG and/or DUSP22 knockdown cells were analyzed using a clonogenic survival assay (Fig. 1E). DUSP22 and PARG double knockdown suppressed the viability of inducible PARG knockdown cells to approximately 40% of that of single knockdown and control cells. This result suggested that DUSP22 deficiency induces synthetic lethality under conditions of PARG inhibition.

Because the frequency of DUSP22 mutation is higher in lung cancer than in other types of tumors (Fig. 1C), we examined whether DUSP22 and PARG dysfunction induced synthetic lethality in the lung cancer cell lines A549, PC14, and SBC5. To exclude the possibility of off-target effects of PARG and DUSP22 knockdown, the effect of 2 different siRNA sets (#1 and #2) against PARG and DUSP22 was tested (Fig. 2A,a and b; Fig. 2B, a and b) using a clonogenic survival assay (Fig. 2C, a–c). Double knockdown of PARG and DUSP22 in A549 showed synergistic growth inhibition in comparison with single knockdown and N.C. cells (Fig. 2C, a). In PC14 and SBC5 cells, one of each siRNA set showed a synergistic effect but the other showed an additive effect, respectively (Fig. 2C, b and c). These results suggested that dysfunction of PARG and DUSP22 efficiently induced synthetic lethality in particular lung cancer cell lines.

To examine the mechanism underlying lethality in DUSP22 and PARG double knockdown cells, the expression levels of apoptosis-related and cell-cycle–related genes were analyzed by qRT-PCR. As shown in Fig. 3A, a–c, PUMA mRNA levels were higher in double knockdown cells than in single knockdown cells in these lung cancer cell lines. In PC14 cells, DUSP22 single knockdown and DUSP22 and PARG double knockdown upregulated the expression of NOXA, a gene involved in apoptosis induction, compared with the levels in control cells (Fig. 3A, b). The expression level of CDKN2A, a cell-cycle negative regulator and cell senescence-related factor, was also elevated in response to DUSP22 single knockdown and double knockdown conditions in SBC5 cells (Fig. 3A, c). To determine whether apoptosis induction was involved in the decreased cell viability caused by PARG and DUSP22 double knockdown in lung cancer cell lines, we performed cell-cycle distribution analysis. In all cell lines tested, PARG and DUSP22 double knockdown increased the sub-G₁ population compared with that in PARG or DUSP22 single knockdown cells (Fig. 3B, a–c). To determine whether PUMA induction was responsible for the suppression of cell viability, siRNA-targeting PUMA was introduced into PARG and DUSP22 double knockdown A549 cells (Fig. 4A). As shown in Fig. 4B, PARG/DUSP22 double knockdown reduced cell viability compared with N.C. (P < 0.05), whereas the additional PUMA knockdown moderately recovered cell viability of PARG/DUSP22 double knockdown condition at 9 days after transfection. To identify the cell death pathway involved in synthetic lethality in A549 cells, a protein array analysis was performed using PathScan Stress and an Apoptosis Signaling Antibody Array Kit (Cell Signaling Technology). As shown in Supplementary Fig. S1, cleaved caspase 3 and cleaved PARP1 were upregulated at 3 days after transfection in response to DUSP22 and PARG double knockdown compared with their expression in the single knockdown condition in A549 cells. Taken together, these data suggested that synthetic lethality was induced by the promotion of apoptosis through the upregulation of PUMA under conditions of PARG and DUSP22 double knockdown in lung cancer cells.

TP63 and TP73 are both involved in apoptosis induction (34). To determine the mechanism underlying DUSP22 and PARG double knockdown-induced synthetic lethality in A549 cells, the mRNA levels of TP63 and TP73 were analyzed by qRT-PCR. As shown in Fig. 4C, a and b, TP63 levels were synergistically increased in double knockdown A549 cells than in PARG and DUSP22 single knockdown cells at 4 days after transfection, whereas TP73 levels showed an increase compared with N.C. and DUSP22 knockdown cells but did not differ between double knockdown and PARG single knockdown cells. Next, we examined whether TP63 dysfunction affected cell viability in DUSP22 and PARG knockdown A549 cells. The results showed that TP63 knockdown rescued cell growth in double knockdown cells (Fig. 4D and E). These suggest that TP63 expression is necessary for apoptosis induction in the double knockdown condition. To determine whether DUSP22 and PARG knockdown-induced apoptosis in A549 cells is dependent on reduced dephosphorylation of p38 MAPK (Fig. 1D, c), double knockdown cells were cultured in the presence or absence of the p38 MAPK-specific inhibitor SB203580, and cell growth was analyzed by the cell proliferation assay. As shown in Supplementary Fig. S2, SB203580 treatment restored cell growth in DUSP22 and PARG knockdown A549 cells. These results suggested that the upregulation of phosphorylated p38 level under DUSP22 dysfunction is involved in promoting PUMA-mediated apoptosis in double knockdown A549 cells.

In addition, we examined whether PARG and DUSP22 knockdown suppressed the protein expression of cell proliferation-related genes in A549 cells. PTEN, phospho-PTEN (Ser380), and phospho-mTOR (Ser2448) levels were decreased in double knockdown A549 but not in single knockdown cells (Fig. 5). Phospho-AKT (Thr308 and Ser473) and phospho-mTOR (Ser2481) were downregulated in both double knockdown and PARG or DUSP22 single knockdown cells (Fig. 5). These results suggested that the decreased survival rate of DUSP22 and PARG double knockdown cells was induced in part by the inhibition of the PI3K/AKT/mTOR pathway.

Based on the in vitro data showing the induction of apoptosis in DUSP22 and PARG double knockdown cells, we examined whether PARG and DUSP22 knockdown exerted a synthetic lethal effect in A549-derived xenograft tumors. Mice were injected with A549 cells transfected with PARG siRNA and/or DUSP22 siRNA, and effect on tumor growth was observed (Fig. 6A). As shown in Fig. 6B, a and B, b, double knockdown of DUSP22 and PARG in A549 cells suppressed tumor growth compared with that of tumors transfected with control siRNA (Fig. 6B, b, P < 0.05). Although statistical differences between single and double knockdown groups were not observed, a tendency of decreased tumor volume in double knockdown condition was observed. This observation thus suggested that PARG and DUSP22 double knockdown suppressed tumor xenograft growth, possibly in a synergistic manner.

PARP inhibitors were recently shown to induce synthetic lethality in HRR-deficient cancer cells (5). Olaparib, a PARP inhibitor, was approved for the treatment of ovarian cancer harboring BRCA1/2 mutations, and this novel type of anticancer agent is effective as monotherapy against BRCA1/2-mutated cancers (5, 35). These drugs are expected to provide an effective cure for cancer with few adverse effects on normal cells. By contrast, little is known about synthetic lethal targets of PARG inhibition. PARG inhibition by gallotannin and siRNA-mediated silencing of PARG induce a weak synthetic lethal effect in BRCA2-mutated breast cancer cells (21). In addition, ibrutinib, an inhibitor of BTK, enhances the lethal effects of PARG inhibition by ethacridine (22). Despite these findings, the potential of PARG as a therapeutic target for anticancer drugs based on synthetic lethality remains unclear. In this study, we searched for novel synthetic lethal targets of PARG inhibition by performing a comprehensive analysis of synthetic lethal genes using inducible PARG knockdown cells and a shRNA library. Among the candidate genes identified, we focused on DUSP22, because it is frequently mutated in various types of cancer (Fig. 1C).

DUSP22 is a member of the DUSP subfamily of protein tyrosine phosphatases. Its dephosphorylation substrate remains unclear, whereas other DUSP family proteins function in the dephosphorylation of JNK, p38, and/or ERK (33). Here, we showed that DUSP22 directly or indirectly dephosphorylated p38 by silencing DUSP22 in inducible PARG knockdown T-REx HeLa cells (Fig. 1D, c). The p38 MAPK pathway is activated by phosphorylation of p38 and is involved in the induction of cell death through apoptosis (36). Colony formation assays showed that double knockdown of DUSP22 and PARG in the lung cancer cell lines A549, PC14, and SBC5 efficiently induced cell death compared with the effect of single knockdown. Cell-cycle analysis showed that double knockdown increased the sub-G₁ population, and treatment with the p38 MAPK inhibitor SB203580 restored cell survival in double knockdown A549 cells. Taken together, these results suggest that the synthetic lethal effect of dysfunction of DUSP22 and PARG was mediated in part by the induction of apoptosis via the p38 MAPK pathway in A549 cells. Under oxidative stress conditions, PARP1 promotes the phosphorylation of p38 in association with the downregulation of MAPK phosphatase-1, which is involved in the dephosphorylation of JNK and p38 MAP kinases, resulting in increased cell death (37). In this study, we demonstrated that increased p38 phosphorylation induced by DUSP22 knockdown and PAR accumulation induced by PARG knockdown (Supplementary Fig. S3) exerted a synthetic lethal effect by promoting apoptotic cell death in lung cancer cells.

In the apoptosis pathway, the tumor suppressor p53 activates apoptosis-related factors, and p53 missense mutations are present in various types of cancer including lung cancer (38). However, the p53 homolog TP63 is rarely mutated (39). TP63 has common transcription targets with p53 and promotes the expression of PUMA to induce p53-independent apoptosis (40). In this study, we showed that knockdown of DUSP22 and PARG in A549 cells upregulated TP63 (Fig. 4C) and PUMA expression. In addition, despite the fact that PC14 and SBC5 cells bear p53 mutations (41, 42), double knockdown of DUSP22 and PARG in these cells induced apoptosis by upregulating PUMA (Fig. 3A, a–c). These results indicated that the synthetic lethal effect of DUSP22 and PARG dysfunction was mediated by the induction of apoptosis through the TP63 pathway. The induction of apoptosis through a p53-independent pathway has important implications because many cancers have p53 pathway mutations.

The present results suggested that dysfunction of PARG and DUSP22 in A549 cells affects the PI3K/AKT/mTOR pathway, which is frequently involved in cancer cell proliferation (43). As shown in Fig. 5, the level of phospho-mTOR (Ser-2448) was lower in double knockdown cells than in control cells despite the downregulation of PTEN and phospho-PTEN, a negative regulator of the PI3K/AKT pathway, in double knockdown cells. The phosphorylation level of AKT (Thr308 and Ser473) and mTOR (Ser2481) was reduced in both double knockdown A549 cells and single knockdown cells. Overall, these results indicated that the promotion of TP53-independent apoptosis and suppression of PI3K/AKT/mTOR pathway activity induced synthetic lethality in DUSP22 and PARG double knockdown A549 cells.

As shown in Fig. 1C, various cancers including lung and colon cancers occasionally bear DUSP22 gene mutations. The complete genome sequence analysis from a patient with lung cancer showed that DUSP22 was inactivated by loss of heterozygosity and point mutations (31). DUSP22 expression is also downregulated in breast cancer and anaplastic lymphoma kinase-negative anaplastic large cell lymphoma (32). In this study, the double deficiency of DUSP22 and PARG suppressed lung tumor growth in a xenograft model (Fig. 6), suggesting that DUSP22 could be useful as a biomarker for monitoring the synthetic lethal effect of PARG inhibitors. PARG deficiency sensitizes cancer cells to alkylating agents (14, 18), cisplatin (18) and γ-irradiation (18, 19). Cancers with low expression levels of DUSP22 may increase sensitivity to combination therapy with these drugs and PARG inhibitors. The development of PARG inhibitors for clinical application is awaited. Both inhibition of BTK and BRCA2 in combination with PARG inhibition has a moderate synthetic lethal effect (21, 22). This study suggested that PARG-specific inhibitors could be useful for cancer therapy by exerting a synthetic lethal effect on cancers with DUSP22 deficiency.

No potential conflicts of interest were disclosed.

Conception and design: Y. Sasaki, H. Fujimori, M. Hozumi, T. Nozaki, K. Inoue, F. Koizumi, M. Masutani

Development of methodology: H. Fujimori, M. Masutani

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Sasaki, H. Fujimori, M. Hozumi, M. Masutani

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Sasaki, H. Fujimori, M. Hozumi, T. Onodera, K. Ashizawa, F. Koizumi, M. Masutani

Writing, review, and/or revision of the manuscript: Y. Sasaki, H. Fujimori, T. Onodera, T. Nozaki, Y. Murakami, K. Ashizawa, K. Inoue, M. Masutani

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Fujimori, T. Nozaki, M. Masutani

Study supervision: H. Fujimori, K. Inoue, M. Masutani

We are thankful for kind support by Dr. Toshio Imai of Central Animal Division, National Cancer Center and technical assistance by Hiromi Harada. This research is partially supported by the Practical Research for Innovative Cancer Control from Japan Agency for Medical Research and Development, AMED (15Ack0106021, 17ck0106286), and Grant-in-Aid for Scientific Research [KibanB 22300343, H23-Jitsuyoka(Gan)-004] to M. Masutani.

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.