Abstract
Background: Three-dimensional (3D) in vitro cultures can recapitulate the physiological in vivo microenvironment. 3D Modeling techniques have been investigated and applied in anticancer drug screening. Materials and Methods: A silicate fiber scaffold was used for 3D cell cultures, and used to model the efficacy of anticancer drugs, such as mytomicin C and doxorubicin. Results: A unique 3D structure was observed in 13 human tumor cell lines on scaffold, and these cells exhibited higher drug resistance than cells in two-dimensional (2D) cultures. Furthermore, the production of lactate and expression of the nuclear factor-kappa B (NF-κB)-regulated genes B cell lymphoma-2 (BCL2), cyclooxygenase-2 (COX2), and vascular endothelial growth factor (VEGF) were higher in 3D cultures than in 2D cultures. Conclusion: These findings suggest that a 3D model using a silicate fiber scaffold can mimic features of cancer, and is also a suitable model for the evaluation of anticancer drugs in vitro.
In anticancer drug discovery, dominantinly, screening and initial characterizations of potential anticancer drug candidates is performed using tumor cell monolayer cultures [two-dimensional (2D) cultures]. However, many therapeutics that demonstrate efficiency in vitro fail to present the same effect in vivo (1) because the 2D culture does not accurately reflect the physiological properties of tumors in vivo (2, 3). Most cancer researchers have to depend on 2D cultures at the initial step of drug discovery because they can provide various high-throughput applications (4, 5). However, it is necessary to develop a more biomimetic in vitro model of tumor tissue for prediction of drug efficiency.
Cell–cell and cell-extracellular matrix (ECM) interactions form a complex communication network of biochemical and mechanical signals that are crucial for in vivo tumor physiology. Three-dimensional (3D) cultures are known to closely mimic properties of tumor tissues (6), and many tools are now available to develop 3D cultures in vitro (7). One of the most critical factors in 3D cultures is the ‘scaffold’, which serves as an ECM substrate for cell growth and formation of 3D structures. Scaffolds for 3D culture heve been developed using natural scaffolds, such as collagen (8) and hyaluronic acid (9), synthetic polymers such as poly(lactide-coglycolide) (10), and polymer microparticles formed from poly(D,L-lactic-co-glycolic) acid and polylactide (11). These models are convenient; however, their applications are limited because they do not permit regulation of 3D structure or reproduction of 3D culture (12). The ECM has a predilection to adhere to hydrophobic surfaces; therefore, control of substrate hydrophobicity may be an effective approach for modulating cellular adhesion and proliferation.
Recently, silicate fibers (SNFs) prepared by electrospinning technology via a sol-gel process have been developed as a scaffold for tissue engineering (13-15). Electrospinning using the sol-gel method is useful for generating SNFs, as it enables the control of fiber diameter from several microns down to 100 nm. The resulting SNFs have a large surface area-to-volume ratio, high fraction of void space, and small pore sizes. In SNFs treated at 500°C, the hydroxyl group content decreases to 1% or less, and the surface of the SNFs becomes hydrophobic (15). Therefore, it is expected that SNFs treated in this manner may improve cell adhesion and cell proliferation while facilitating control and in vitro reproduction of the 3D structure. Our hypothesis is that the use of an SNF scaffold-based 3D tumor model may simulate in vivo conditions and thus be more effective than conventional 2D culture for determining the efficacy of anticancer drug candidates.
In the present study, we developed in vitro SNF-scaffold 3D cultures using various types of tumor cell lines. The anticancer effects of chemotherapeutic agents were studied, and their cytotoxicity was compared to that determined in conventional 2D cultures. Furthermore, we evaluated the uptake of anticancer drugs in 3D culture, and compared the effects of anticancer drugs on lactate production (tumor metabolic marker) and alternations of nuclear factor-kappa B (NF-κB)-regulated gene B cell lymphoma-2 (BCL2; antiapoptosis), cyclooxygenase-2 (COX2; cell growth), and vascular endothelial growth factor (VEGF; angiogenic factor) expression in 2D and 3D cultures.
Materials and Methods
Chemicals. Anticancer drugs were purchased as follows: mitomycin C from (Nacalai Tesque, Kyoto, Japan) and doxorubicin from (Sigma, St. Louis, MO, USA). SNFs were supplied from Japan Vilene Company Ltd. (Ibaraki, Japan).
2D cell culture. The following human cancer cell lines were purchased from Health Science Research Resources Bank (Osaka, Japan) or RIKEN (Tsukuba, Japan) or the American Type Culture Collection (Rockville, MD, USA): A549 (lung adenocarcinoma), MCF-7 (breast cancer), MDA-MD-453 (breast cancer cell), HeLa (ovarian carcinoma), NIH:OVCAR-3 (ovarian carcinoma), HT-29 (colon adenocarcinoma), DLD-1 (colon carcinoma), DU145 (prostate cancer), HepG2 (hepatoma), Huh-7 (hepato cellular carcinoma), HLE (hepatoma), SUIT-2 (pancreatic cancer) and MIA PaCa-2 (pancreatic cancer). The cells were cultured in RPMI or Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin at 37°C in humidified atmosphere containing 5% CO2.
2D cell growth inhibition. Briefly, the cells were plated at an appropriate density in 96-well plates in medium with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin and allowed to attach for 24 h. The inhibition of cell growth was assessed by measuring changes in total cellular protein in a culture of each cell line after 72 h of drug treatment by using of a sulforhodamine B (SRB) assay (16). Concentrations that inhibited cell growth by 50% (IC50) and 80% (IC80) after 72 h of treatment were calculated based on the survival rate compared against untreated cells.
3D culture. Three-dimensional culture using SNFs was described by Yamaguchi et al. (13). Briefly, SNFs were sterilized by autoclave, and transferred to 24-well plates. Culture media as same as 2D culture containing human cancer cells were poured into each well with SNFs of 24-well plates. Cells were cultured at 37°C in humidified atmosphere containing 5% CO2. Forty-eight hours later, each SNF with cells was transferred to new 24-well plates, and fresh media were poured into each well. The media were exchanged every other day.
3D cell growth inhibition. The cells were plated at an appropriate density in 24-well plates in medium with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin and allowed to attach for 24 h. Medium containing drugs was exchanged every 24 h for 96 h. Measurement of cell growth inhibition was calculated based on the amount of DNA, which was determined with a DNA Quantity kit (Primary Cell, Hokkaido, Japan) according to the manufacture's instructions.
Scanning electron microscope (SEM). Paraformaldehyde-fixed samples were rinsed in phosphate-buffered saline (PBS) thrice and then post-fixed with osmium tetroxide. Dehydration was accomplished using a graded series of ethanol (10, 50, 60, 70, 80, 90 and 100%), and then the samples were transferred to t-butyl alcohol. After the samples were frozen, dried samples without sputter coating were viewed in an ES-2030 Scanning electron microscope (Hitachi Ltd., Tokyo, Japan) at 15 kV.
Doxorubicin uptake. Seven days after growing, media were removed and 3D cells were incubated with fresh media containing 10 μM doxorubicin. The indicated incubation period (3, 6 and 24 hour), 3D cells were washed with PBS twice to remove unbound free doxorubicin, and were observed with a fluorescence microscope at excitation and emission wavelengths of 534 nm and 602 nm for doxorubicin.
To establish cellular drug uptake profiles, HT-29 cells were plated onto 24-well plates at densities of 5.0×105 cells/well at 37°C. Seven days later, doxorubicin (10 μM) was added to each well to initiate cellular drug accumulation. At different time intervals (3, 6 and 24 h), the supernatant was removed and cells were washed with ice-cold PBS (pH 7.6) and sonicated with PBS containing 2 mM EDTA. Doxorubicin in the cell lysates was measured with an FP-8300 Spectrofluorometer (JASCO, Easton, MD, USA) at an excitation wavelength of 480 nm and an emission wavelength of 560 nm. To adjust for background fluorescence from cellular components or non-specific binding to SNFs, doxorubicin standard curves were also prepared using cells or SNF lysates. Protein concentration was determined by BCA assay according to the manufacture's instructions (Pierce, Rockford, IL, USA).
Lactate production assay. To determine the glycolytic activity in 2D and 3D culture, the amount of produced lactate was measured. The cells were cultured until confluent or 3D form in regular medium. The medium was then changed, and the cells were incubated in Hank's balanced salt solution for 2 h at 37°C in humidified atmosphere containing 5% CO2. The lactate in each sample was measured by EnzyChrom Lactate Assay Kit (BioAssay Systems, Hayward, CA, USA).
Total RNA extraction and reverse transcription (RT). Isogen RNA extraction kit was purchased from Nippon Gene (Toyama, Japan). The quality of RNA was verified by the OD260:OD280 absorption ratio. Total RNA (0.5 μg) isolated from 2D or 3D cells was converted to single-strand cDNA using PrimeScript RTase (Takara, Kyoto, Japan) using 50 pM random hexamer primers (TaKaRa, Kyoto, Japan) according to the manufacturer's instructions.
LightCycler real-time polymerase chain reaction (PCR). For LightCycler reaction, a master mix of the following reaction components was prepared to the indicated end-concentration: 3 μl water (PCR grade), 1 μl forward primer (10 μM), 1 μl reverse primer (10 μM) and 10 μl LightCyler FastStart Essential DNA Green Master (Roche, Tokyo, Japan). cDNA (5 μl) was amplified by using a standardized RT-PCR protocol with FastStart Essential DNA Green Master kit (Roche) in a LightCycler instrument (Roche). The melting curve analysis program of the LightCycler was used to identify specific PCR products. The expression levels were normalized by that of 28S. The primers were as follows: BCL2: F, 5’-GGCTGGGATGCCTTTGTG-3’ and R, 5’-GCCAGGAGAAATCAA ACAGAGG-3’; COX-2: F, 5’-CTTCACGCATCAGTTTTTCAAG-3’ and R, 5’-TCACCGTAAATATGATTTAAGTCCAC-3’; VEGF: F, 5’-CCGCAGACGTGTAAATGTTCCT-3’ and R, 5’-CGGCTTGTCA CATCTGCAAGTA-3’; 28S: F, 5’-CCGCTGCGGTGAGCCTTGAA-3’ and R, 5’-TCTCCGGGATCGGTCGCGTT-3’. The presence of the expected PCR products after quantitative real-time RT-PCR reactions (64 bp for BCL2, 96 bp for COX2, 95 bp for VEGF and 312 bp for 28S) was confirmed by an agarose gel electrophoresis (data not shown).
3D structure formation of HT-29 cells on silicate fibers (SNFs). A: Generation of 3D structure of HT-29 cells on SNFs. Time-lapse scanning electron micrograph of HT-29 cells growing on SNFs. B: Cross section of 3D structure of HT-29 cells on SNFs at day 7. C: Amount of DNA in growing cultures (reflective of cell number).
Statistical analysis. All data are presented as the mean±SD for at least three replicate experiments and were analyzed with Student's t-test.
Results
Generation of unique 3D structures by tumor cell lines on SNFs. To determine whether tumor cells form 3D structures on SNF scaffolds, HT-29 colorectal adenocarcinoma cells were used. SEM showed that HT-29 cells grew gradually and had tight cell-cell adhesions where individual cell margins were hardly detectable at day 10 (Figure 1A). The SEM cross-section showed that HT-29 cells on SNFs formed tight aggregates (Figure 1B). As shown in Figure 1C, the amount of DNA (used as a proxy for cell number) in HT-29 cells on SNFs increased commensurately with the formation of the 3D structure, reaching a plateau at day 8. Twelve other tumor cell lines were also found to form unique 3D structures on SNFs (Table I), and these 3D cultures were grouped into four categories: tight aggregate, loose aggregate, small spheroidal aggregate, and loose small spheroidal aggregate. As shown in Figure 2, HT-29, DLD-1, and MCF-7 cells on SNFs were tightly aggregated without cell margins, whereas cell margins were observed in A549 cells and ovarian, prostate, and liver cancer cells. The aggregates appeared to be undistinguished rod aggregates (SUIT-2 and MIA PaCa-2) with loose small spheroidal aggregates (MDA-MB-453 and HeLa). Cross-sections showed that in rod aggregates, small spheroidal units were linearly-arrayed in a vertical direction whereas these units were completely randomly distributed in the loose small spheroidal aggregates (Figure 2G and H).
Morphology of 3D cancer cell lines on silicate fibers (SNFs).
Representative unique 3D cell morphology of four types of tumor cell lines. Unique 3D structures were generated from 5.0×105 cells (HT-29, HepG2 and MDA-MB-453) and 2.5×105 cells (Suit-2) on silicate fibers (SNFs) and cultured for seven days. 3D cell morphology: HT-29 (A), HepG2 (B), SUIT-2 (C), MDA-MB-453 (D), cross section shows 3D cell morphology: HT-29 (E), HepG2 (F), SUIT-2 (G), and MDA-MB-453 (H) cells.
Antiproliferative activity of anticancer drugs in 2D culture vs. 3D culture. The response to mitomycin C was evaluated in 2D and 3D cultures of HT-29, HepG2, SUIT-2, and MDA-MB-453 cells. The IC50 determined in initial 2D experiments was then applied to 3D culture; however, cytotoxicity was not observed in 3D culture with SNFs (data not shown). Similar effects were also observed for other anticancer drugs (camptothecin, doxorubicin, vinblastine, and etoposide; data not shown). We therefore determined the IC80 in 2D cultures exposed to pulses every 24 h (Figure 3A). Cell viability was similar in 2D and 3D cultures at the time of treatment, relative to the untreated control. Treatment of 3D-cultured cells with mitomycin C resulted in some disruption of 3D structure relative to the untreated control by SEM (Figure 3B), particularly in MDA-MD-453 cells on SNFs. A cell viability assay showed a significant difference between 2D and 3D cultures with regard to anticancer drug sensitivitiy: resistance to mitomycin C in 3D cultures was increased by 4.1-, 2.2-, 4.5-, and 1.8-fold for HT-29, HepG2, SUIT-2, and MDA-MB-453 cells, respectively (Figure 3C). Increased drug resistance to camptothecin was also observed for HT-29, HepG2, and MDA-MB-453 cells, and to sorafenib for Huh-7 cells (data not shown).
Anti-proliferative activity of anticancer drugs in 3D model of HT-29, HepG2, SUIT-2 and MDA-MB-453 cells. A: Anticancer drug administration protocol for 3D culture. Arrowhead shows time of anticancer drug administration. B: Images by scanning electron microscopy of HepG2, Suit-2, MDA-MB-453, and HT-29 cells growing on SNF cultures and the effect of mitomycin C (MMC). C: Comparative cytotoxicity of MMC in 2D and 3D cultures. **p<0.001.
Doxorubicin uptake by HT-29 cells on SNFs. It is possible that the aggregated morphology may have restricted drug penetration and reduced drug exposure in cells in the interior of the 3D structure; therefore, the penetration of doxorubicin into 3D cultures was evaluated. The fluorescence intensity of the cultures was measured at 3, 6, and 24 h after drug exposure. Over time, the fluorescence intensity of doxorubicin increased, with maximum fluorescence being observed at 24 h (Figure 4A). The cells were still alive after 24 h of incubation with doxorubicin (data not shown). Only 7.5% of the total doxorubicin added (10 μM) penetrated into HT-29 cells on SNFs (Figure 4B).
Glycolysis in 2D culture vs. 3D culture. Cancer cells generally exhibit increased glycolysis for ATP generation, which is known as the Warburg effect (17). This increased glycolysis is achieved in part by mitochondrial respiration injury and hypoxia, which are frequently associated with resistance to anticancer drugs. It has been shown that inhibition of glycolysis in tumor cells leads to increased drug sensitivity (18, 19). Lactate production was evaluated to determine the glycolytic activity in 2D and 3D cultures. As anticipated, lactate levels were significantly higher in 3D cultures relative to 2D cultures (Figure 5). Moreover, a similar increase in lactate production was observed in HepG2, SUIT-2, and MDA-MB-453 cells on SNFs (data not shown).
Genes regulated by the transcriptional factor NF-κB in 2D culture vs. 3D culture. NF-κB is a central transcriptional factor that influences oncogenesis, apoptosis, and anticancer drug sensitivity in colorectal cancer. NF-κB is involved in cell survival, proliferation, and angiogenesis in multiple tumor types (20, 21). To evaluate the molecular mechanism underlying anticancer drug resistance in 3D culture, the expression of NF-κB-regulated genes was studied. Expression of BCL2 (survival), COX2 (proliferation), and VEGF (angiogenesis) was significantly higher in 3D than in 2D cultures (Figure 6).
Doxorubicin uptake in a 3D model (HT-29 cells) on silicate fibers (SNFs). A: Representative images of doxorubicin uptake in HT-29 cells growing on SNFs. Images were taken at 3-, 6-, and 24-h time points. B: Relative fluorescence intensitiy of doxorubicin uptake.
Discussion
Three-dimensional culture platforms present different cell differentiation and behavior relative to conventional 2D culture (1, 3, 22-23). Given the benefit of 3D culture models, particularly with regard to anticancer drug discovery, we have developed an in vitro model using various types of tumor cells with SNFs prepared by electrospinning via a sol-gel process. 3D Culture systems facilitate the expression of hallmark features of cancer cells, such as unlimited proliferation, self-sufficiency in growth signals, resistance to anti-growth signals, anti-apoptosis, invasion, and angiogenesis (12, 24). Conventionally, naturally-occurring or organic synthetic scaffolds formed from collagen, hyaluronic acid poly(lactide-coglycolide), and poly(D,L-lactic-co-glycolic), or polylactide polymer microparticles have been used in 3D culture systems (8-11).
In the present study, we found that an inorganic scaffold generated using SNFs could be adapted to 3D culture. Although SNF scaffolds have been used for the 3D culture of several cell lines, such as MG63, CHO-K1, and HepG2 (13-15), they have not yet been used for other human tumor cell lines. In the present study, we found that at least 13 tumor cell lines can be cultured in 3D culture using SNF scaffolds. The effective formation of 3D structures and subsequent culture of these cancer cells depend to a large extent on the features of the SNFs, including hydrophobicity. Previous studies have shown that various factors, including the affinity of the cell to the scaffold surface, contribute to the formation of 3D structures, and that scaffold hydrophobicity is a particularly important factor in this process, likely because the ECM facilitates adhesion to scaffold surfaces with hydrophobic composition (25, 26). In the present study, the cultured cells formed four classes of unique 3D structures, partly according to their original tissues. The reason for such different 3D structures in culture of cancer cell lines remains unclear. Among the cell lines, 3D structural differences were not observed in colon, liver, and pancreas cancer cells. In contrast, breast cancer cell lines had a completely different 3D structure. However, other cell lines must be evaluated to exclude the possibility that these differences may reflect the individual characteristics of each cancer cell line.
Glycolysis in 2D and 3D culture on silicate fibers (SNFs). The amount of lactate produced by 2D and 3D tumor models was measured by lactate assay. Representative data (mean±SD, n=3) were normalised to their DNA content. Experiments were performed in triplicate.
Analysis of nuclear factor-kappa B (NF-κB)-regulated genes in 2D and 3D culture on silicate fibers (SNFs). Relative expression of NF-κB-regulated genes, B cell lymphoma-2 (BCL2), cyclooxygenase-2 (COX2), and vascular endothelial growth factor (VEGF) in 2D and 3D culture. All data (mean±SD, n=4-6) were normalized by 28S RNA. *p<0.005, **p<0.001.
It has been reported that MCF-7 cells form spheroids in Matrigel (27), whereas they induce the formation of tight aggregates on SNFs. Spheroid cultures have been predominantly used in in vitro 3D culture systems but are associated with limitations, such as those related to culture handling, control of spheroid size, and adaptation of the re-circulation experiment, for example, with respect to the bioreactor system which is for assessment of the penetration of anticancer drugs through a solid tissue environment. The use of SNFs can overcome these drawbacks and may thus enable the development of a novel bioreactor system. In addition, regarding the four different unique 3D conformations, human colon adenocarcinoma HT-29 and DLD-1 cells formed the same type of 3D structure, that is, a tight aggregate, not demonstrated by other cell lines.
Many solid tumors are poorly-vascularized, with variable rates of blood flow and larger intercapillary distances than in normal tissues. The requirement for anticancer drugs to penetrate multiple layers of tissue may pose a barrier to the effective therapy of solid tumors, and penetration of anticancer drugs into tumor cells distant from the vascular system is necessary for efficacy of cancer chemotherapy against solid tumors. Recently, an in vitro model was developed for direct quantitative assessment of the penetration of anticancer drugs into solid tumors (28, 29). The structural features of HT-29 and DLD-1 cells on SNFs should allow their adaptation to such an assay system, thereby resulting in a novel model for assessment of anticancer drug penetration. Moreover, although the 3D structure of HT-29 and DLD-1 cells on SNFs is similar to that of multicellular spheroids (30), their planar structure allows flux through the cultures to be more easily measured (31-32). Consequently, the 3D SNF system will likely provide more predictive data, allow for reduced animal testing, and reduce the costs and time for required drug discovery, in addition to enabling a faster time to the markets.
In this study, we observed a critical difference in the response to anticancer drugs in 2D and 3D culture systems. To investigate the disparity in the effective drug dose in our tumor model, we compared drug uptake and gene expression in conventional 2D and 3D culture systems. Firstly, we studied the penetration efficiency of doxorubicin into 3D-cultured cells on SNFs. Only 7.5% of the doxorubicin penetrated inside the 3D cells, which is consistent with the well-known feature of insufficient penetration of drugs in 3D tumor models (12, 28, 33). Moreover, the poor penetration of anticancer drugs into an in vitro tumor may explain clinically-observed anticancer drug resistance, as a similar environment is present in vivo (34). Poor drug penetration results in a requirement for high doses of anticancer drugs during therapy. We speculate that the poor drug penetration observed in a 3D culture may be why many anticancer drug candidates screened in 2D tumor models fail in in vivo tests or clinical trials.
Another possible factor underlying anticancer drug resistance in 3D culture is increased glycolytic activity. Persistent glycolysis is a characteristic of cancer cells and a hallmark of advanced cancer. Lactate is associated with tumor progression (35), and lactate accumulation in tumors is associated with metastasis and poor survival (36, 37). We observed that lactate accumulation was higher in 3D cultures than in 2D cultures. In addition, in the media of the 3D cultures, the color of the phenol red dye changed from red to yellow within only 24 h, i.e. between pre-confluency and post-confluency, whereas this change was not observed in the media of 2D cultures, even after confluency (data not shown). The rapid acidification of the 3D culture likely resulted from the secretion of lactate by the cells, as supported by the higher lactate production in 3D cultures relative to that in 2D cultures. These findings further suggest that 3D culture on SNFs may be better representative of the in vivo tumor environment.
Furthermore, the disparity of anticancer drug sensitivity in 2D and 3D cultures may be associated with altered gene expression. We found that gene expression of BCL2, COX2, and VEGF was higher in 3D cultures than in 2D cultures. All these genes are known to be regulated by NF-κB (20, 21). The up-regulation of the anti-apoptic marker BCL2 in 3D cultures is consistent with findings from other 3D culture systems showing that a decrease in apoptosis alters gene expression (38, 39). It was previously found that COX2 is up-regulated in colorectal adenoma and colorectal cancer (40, 41). Considering that COX2 inhibitors can induce apoptosis and enhance the cytotoxicity of anticancer drugs (42), COX2 may play a role in colorectal cancer. COX2 overexpression has been reported to confer resistance to apoptosis by up-regulation of BCL2 (43, 44). VEGF as a survival factor for tumor cells is also regulated by NF-κB, and it has been reported that COX2 expression in cancer cells stimulates the production of VEGF (45). In addition, VEGF up-regulates BCL2 and inhibits apoptosis of human cancer cells (46). Up-regulation of BCL2 may be mediated by COX2 and VEGF, thereby inducing anticancer drug resistance.
In conclusion, our study demonstrates that an in vitro 3D tumor model using SNF scaffolds prepared by electrospinning technology via a sol-gel process can be used as a novel system for assessment of the efficacy of potential anticancer drugs. This method allows for direct quantitative assessment of the penetration of anticancer drugs through a solid-tissue environment. Furthermore, this 3D tumor model can be used in screening angiogenic factor-targeted therapies. In addition, the fact that SNF scaffolds can create different unique 3D structures, with each type being tumor-dependent, can help to understand tumor cellular morphological evolution more precisely.
Footnotes
-
↵* These Authors contributed equally to this work.
- Received October 22, 2013.
- Revision received November 7, 2013.
- Accepted November 8, 2013.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.