|Home | About | Journals | Submit | Contact Us | Français|
Cullin4A (Cul4A) is a scaffold protein that assembles cullin-RING ubiquitin ligase (E3) complexes and regulates many cellular events, including cell survival, development, growth, and cell cycle control. Our previous study suggested that Cul4A is oncogenic in vitro, but its oncogenic role in vivo has not been studied. Here, we used a Cul4A transgenic mouse model to study the potential oncogenic role of Cul4A in lung tumor development. After Cul4A overexpression was induced in the lungs for 32 weeks, atypical epithelial cells were observed. After 40 weeks, lung tumors were visible and were characterized as Grade I or II adenocarcinomas. Immunohistochemistry (IHC) revealed decreased levels of Cul4A associated proteins p21CIP1 and tumor suppressor p19ARF in the lung tumors, suggesting Cul4A regulated their expression in these tumors. Increased levels of p27KIP1 and p16INK4a were also detected in these tumors. Moreover, protein level of DNA replication licensing factor CDT1 was decreased. Genomic instability in the lung tumors was further analyzed by the results from pericentrin protein expression and array Comparative Genomic Hybridization analysis. Furthermore, knocking down Cul4A expression in lung cancer H2170 cells increased their sensitivity to the chemotherapy drug cisplatin in vitro, suggesting that Cul4A overexpression is associated with cisplatin resistance in the cancer cells. Our findings indicate that Cul4A is oncogenic in vivo, and this Cul4A mouse model is a tool in understanding the mechanisms of Cul4A in human cancers and for testing experimental therapies targeting Cul4A.
The Cullin4A gene (Cul4A) is overexpressed or amplified in a variety of human cancers [1–4]. Association of Cul4A amplification with aggressive tumor growth and poor prognosis has been suggested [5, 6]. Cul4A depletion reportedly increased the protection of mouse skin from ultraviolet-induced carcinogenesis . We previously demonstrated the amplification of Cul4A copy number and the increased expression of Cul4A protein in malignant pleural mesothelioma cells, suggesting that Cul4A is an oncogene in thoracic cancer . However, the oncogenic role of Cul4A in vivo has not been studied.
Recently, amplification of Cul4A and other genes was detected on chromosome 13q34 using fluorescence in situ hybridization (FISH) and the frequency of occurrence was 3% in human non-small cell lung cancer (NSCLC) . Amplification of Cul4A in lung cancer was further supported by our analysis of the Cancer Genome Atlas (TCGA) data using the Oncomine cancer microarray database (www.oncomine.org). Cul4A copy number increased by 1.4 fold (≥ 3 copies) in 3% of lung adenocarcinomas (n=138) (Supplementary Fig. 1) and in 3% of lung squamous cell carcinomas (n=204) (Supplementary Fig. 2) when compared to normal lung tissues. Our analysis of TCGA data using the cBioPortal for Cancer Genomics database (http://www.cbioportal.org)[11, 12] further revealed the up-regulation of Cul4A in 8.3% of lung squamous cell carcinoma (n=179) and in 6% of lung adenocarcinoma (n=230) (data not shown). We also found a correlation between Cul4A mRNA expression levels and Cul4A copy number in lung adenocarcinoma cases (Supplementary Fig. 3). Collectively, these results suggest that increased expression of Cul4A is associated with the occurrence of lung cancer in humans, but its oncogenic role in vivo remains to be investigated.
To investigate the biology of the ≥ 3% of NSCLC that showed Cul4A overexpression, we developed a Lox–Stop–Lox Cul4A conditional mouse strain (Lox-Cul4A) , in which Cul4A overexpression is activated by the introduction of an engineered adenovirus carrying the Cre recombinase (Ad-Cre) . After Cul4A overexpression was induced, we observed lesions after 24 weeks and visible lung tumors after 40 weeks. Mouse lungs and lung tumors were collected at several time points after Cul4A overexpression and analyzed for the expression of Cul4A protein and its associated proteins. Here, we show that Cul4A is oncogenic in vivo and that overexpression of Cul4A leads to tumorigenesis in the mouse lung.
All in vivo experiments strictly followed the UCSF institutional guidelines (Institutional Animal Care and Use Committee approval number: AN085516-03C). The Cul4A transgenic mice were generated as previously described  and were maintained as heterozygotes on an FVB/N background. The engineered adenovirus (Ad) carrying either Cre-recombinase (Ad5CMVCre) or GFP (Ad5CMVeGFP) was purchased from the Gene Transfer Vector Core of the University of Iowa. The adenoviruses were introduced into the 6–10 week-old mice through inhalation to induce Cul4A overexpression in lung as previously described . Age-matched littermate mice infected with Ad-GFP were used as controls. Transgenic mice were anesthetized with 2.5% Avertin via intraperitoneal injection, after which half of the mice inhaled approximately 106 or 107 particles of Ad-Cre or Ad-GFP directly into the lungs.
Lungs from the transgenic mice were collected at 8, 16, 24, 32, 40, 48, 56, and 64 weeks after infection with the adenovirus. Lung tumors were collected at 40, 48, 56, and 64 weeks after infection with the adenovirus. Histological sections of the mouse lungs and lung tumors were stained with hematoxylin and eosin for general morphology analysis. For IHC analysis, endogenous peroxidase was quenched for 15 min at room temperature with 3% H2O2 in methanol in each lung section. Sections were blocked with 4% normal goat serum in PBS with 0.2% Triton for two hours at room temperature before incubation overnight at 4°C with the properly diluted antibodies: anti-Cul4A (ab34897,Abcam) at 1:400; anti-c-Myc (sc-789, Santa Cruz) at 1:20; anti-PCNA (sc-7907, Santa Cruz) at 1:100; anti- p21CIP1 (sc-471, Santa Cruz) at 1:100; anti- p27KIP1 (sc-528, Santa Cruz) at 1:100; anti- p14/19ARF (251543,Abbiotec) at 1:100; anti-p16 (sc-1207, Santa Cruz) at 1:200; anti-CDT1 (ab70829,Abcam) at 1:250; anti-pericentrin (ab4448,Abcam) at 1:2000.
Three independent researchers blindly scored positivity, and the data represent the samples that were scored positive by all three individuals. The following scoring system was used: −, no stain; +, weak staining (10% or above stained cellularity considered as positive); ++, moderate staining (30% or above stained cellularity considered as positive); +++, strong staining (50% or above stained cellularity considered as positive). All scoring was done under magnification (20×) with a Zeiss Axioscop 2 microscope (Carl Zeiss Inc, Germany) and photomicrographs were obtained with a Carl Zeiss AxioCam MrC5 camera.
The CellTiter-Glo® luminescent cell viability assay (Promega, Madison, WI) was used to evaluate cisplatin cytotoxicity. H2170 cells were transfected with Cul4A siRNA or control siRNA at 100nM, 50nM, or 1nM using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. H838 and H460 cells were transfected with Cul4A siRNA or control siRNA at 50nM. Transfected cells were seeded at 5000 cells/well in 96-well plates and cultured at 37°C supplied with 5% CO2 without antibiotics for 24 hours. Cells were treated with cisplatin (10 fold serial dilution) for 72 hours. Then, 50 μl of the CellTiter-Glo reagent was added into each well. After incubation for 10 minutes, the luminescence signals were read by a GloMax® 96 microplate luminometer (Promega, Madison, WI) and a dose–response curve was plotted. The assay was done in triplicate and data are shown as mean values ± standard deviation (SD). The half maximal inhibitory concentration (IC50) values were determined using GraphPad Prism® log (inhibitor) vs. response (variable slope) software (version 6.01, La Jolla, CA). Paired Student’s t-tests were used to compare cell viability for different treatments against the samples with cisplatin only. Significance was defined as p < 0.05 with two-sided analysis.
To determine whether overexpressing Cul4A resulted in tumor development in mouse lungs, we generated heterozygous transgenic mice overexpressing Cre-inducible Cul4A (Lox-Cul4A), collected mouse lungs after Cul4A overexpression was induced (Fig. 1A), and analyzed tissue sections from mouse lungs by using IHC. We found that the morphology of Ad-GFP induced mouse lung appeared to be normal at all time points (Fig. 1B). In contrast, mice injected with Ad-Cre showed hyperplasia after 8 to 16 weeks of Cul4A overexpression, as was noted previously . Local edema was evident at 24 weeks and notably worse at 40 weeks (Fig. 1C). Multiple aggregations of lymphocytes were observed under the respiratory epithelial lining (Fig. 1D). The number and size of the lymphocyte aggregations increased after 24 weeks of Cul4A overexpression (Fig. 1E). We observed small groups of atypical cells at 32 weeks (Fig. 1F) and larger groups of these cells at 40 weeks (Fig. 1G), whereas the morphology of the surrounding bronchial epithelium appeared normal. The atypical epithelial cells showed enlarged nuclei and cell size, suggesting that these cells may develop into tumor cells at later time points.
We observed tumors on the surface of 8/25 (32%) mouse lungs after 40 weeks of Cul4A overexpression. One of these eight lungs had two tumors. These nine tumors ranged in size from 2.25 to 25 mm2 (Fig. 2A, T1–T9), and all of the tumor cells showed numerous mitoses and enlarged nuclei and progressed to glandular structures. Three of the nine tumors were characterized as Grade I adenocarcinoma with well-differentiated cell structure (Fig. 2B, D), generally considered the least aggressive tumor cells. Five of the nine tumors were intermediate grade with moderately differentiated tumor cells (Fig. 2C, E). The Grade II tumors also showed breakthrough of the tumor barrier (Fig. 3A) and lymphovascular infiltration (Fig. 3B), suggesting invasive behavior and spread to the lymphatics. In addition, in the tissue surrounding the major tumor, we observed satellite adenocarcinomas with invasion to adjacent stroma after 64 weeks of Cul4A overexpression (Fig. 3C). Multiple lymphoid aggregates surrounded the developed adenocarcinoma (Fig. 3D), suggesting that their appearance is associated with tumor development. These results show that adenocarcinoma growth can be induced in this mouse model and that the tumors have developed invasive behavior and can multiply in the mouse lung. Moreover, increased expression of Cul4A and the c-Myc tag, which was constructed after the Cul4A transgene, was detected in the mouse lung tumors (Supplementary Fig.4A–B). Quantitative analysis of the IHC data for Cul4A, p27KIP1, CDT1 and pericentrin showed similar trend of changes in IHC staining intensity between lung tumors and control lungs (Supplementary Fig. 4C). Copy number of the Cul4A transgene in the transgenic mouse tissues was analyzed by using qPCR (Supplementary Fig. 5A). Similar Cul4A copy number (3–4 copies) was detected in the mouse lungs and in the MEFs whereas the lung tumors showed similar or slightly more Cul4A copy number in the cells.
To further study the possible mechanism of Cul4A, a component of Cul4A-based E3 ubiquitin ligase complexes, on the development of lung adenocarcinoma, we analyzed the expression of several proteins, which are known to associate with Cul4A, in the lungs and tumor samples from mice that overexpressed human Cul4A transgene. The cyclin-dependent kinase (CDK) inhibitor p21CIP1 has been shown to associate with Cul4A-based E3 ubiquitin ligase complex [16, 17]. We previously demonstrated that Cul4A overexpression resulted in downregulation of p21CIP1 protein in mesothelioma cells , suggesting that Cul4A overexpression mediated the degradation of p21CIP1 in vitro. In this Lox-Cul4A mouse model, p21CIP1 protein expression in control lungs was strong at every time point (Fig. 4A, inset) but weak in the six tumors examined (Fig. 4A, main figure). Both chromosomal and transcriptional levels of p21CIP1 measured in the lung tumors and the mouse lungs using qRT-PCR showed similar results (Supplementary Fig. 5B), suggesting the decrease in p21CIP1 expression occurred in the post-translational level. These IHC results suggest that Cul4A overexpression decreased p21CIP1 expression in the lung tumors, an event which may be associated with lung tumor development in this mouse model. We next used the MEFs from transgenic mice to evaluate the observation of p21CIP1 degradation driven by Cul4A overexpression. Primary mouse MEFs generated from the transgenic mouse model were infected with Ad-Cre to induce Cul4A overexpression. Afterward, the cells were co-transfected with HA-tagged p21CIP1 and either Cul4A siRNA or control siRNA. Our results indicate that the expression of p21CIP1 was increased in the cells when Cul4A expression was knocked down (Fig. 4B). To test the possible mechanism that decreased p21CIP1 expression in the lung tumors, we next examined the expression of PCNA, a key co-factor that mediates Cul4A-dependent proteolysis of p21CIP1 [16, 18]. IHC analysis revealed that PCNA expression was strong in the four lung tumors analyzed (Fig. 4C, main figure) but weak in the control lungs (Fig. 4C, inset), suggesting that elevated PCNA expression contributed to Cul4A-mediated p21CIP1 downregulation in lung tumorigenesis.
Considering that the INK4a/ARF locus encodes tumor suppressor genes that play a pivotal role in controlling the RB/p53 network, we next analyzed the expression of p19ARF (p14ARF in human) and p16INK4a in the lung tumors overexpressing Cul4A. IHC analysis revealed that p19ARF expression was weak in the seven analyzed tumors (Fig. 4D, main figure) when compared to that in the control lungs (Fig. 4D, inset). In contrast, p16INK4a expression was strong in the lung tumors (Fig. 4E, main figure) but weak in the control lungs (Fig. 4E, inset).
Protein p27KIP1 is a key regulator of cell-cycle progression, and can regulate cell-cycle arrest by inhibiting CDK activity . Protein p27KIP1 may be a prognostic factor in human cancers [20–22] and its depletion in mice results in tumor growth . Moreover, Cul4A-based E3 ubiquitin ligase was shown to mediate the expression of p27KIP1 . In our mouse model, p27KIP1 expression was strong in the lung tumors overexpressing Cul4A (Fig. 4F, main figure) but moderate in the control lungs (Fig. 4F, inset). These results suggest that Cul4A overexpression elevated p27KIP1 expression in lung tumorigenesis.
As a key factor in regulating genetic stability, DNA replication licensing factor CDT1 can also associate with Cul4A-based E3 ubiquitin ligase complexes [25–27]. To determine whether overexpression of Cul4A resulted in loss of CDT1 expression, we compared CDT1 expression in lung tumors and control lungs by using IHC analysis. Expression of CDT1 was negative in the six lung tumors (Fig. 5A, main figure) but moderate in the control lungs (Fig. 5A, inset). Both chromosomal and transcriptional levels of CDT1 measured in the lung tumors and the mouse lungs by using qRT-PCR showed similar results (Supplementary Fig.5C), suggesting the decrease of CDT1 expression occurred in the post-translational level. We next measured the expression of CDT1 in primary MEFs using western blotting and our results showed that knocking down Cul4A expression using siRNA decreased Cul4A expression but increased CDT1 expression in the cells (Fig. 5B). Together, the results suggested that CDT1 expression was decreased in the lung tumors overexpressing Cul4A. To test whether genomic instability occurred in the lung tumors, the expression of pericentrin, an indicator of genomic instability , was analyzed using IHC. Our data showed that pericentrin expression was strong in the lung tumors (Fig. 5C, main figure) but weak in control lungs (Fig. 5C, inset), suggesting genomic instability occurred in the lung tumors overexpressing Cul4A.
The genomic instability in the tumors that overexpress Cul4A was further evaluated using array Comparative Genomic Hybridization (aCGH) to visualize DNA copy number alterations across the lung tumor genomes. The most frequent amplification was on distal chromosome 8 and the most frequent deletion was on proximal chromosome 18 (Fig. 5D). Kras transgenic mouse model of lung cancer was developed and genomic instability in the lung tumors was analyzed using aCGH . Increase in DNA copy number was detected on multiple chromosomes in the aCGH results collected from Kras overexpressed lung tumors. Similar to our observation of the aCGH data from Cul4A overexpressed lung tumors, increase in DNA copy number was also detected on the chromosome 8 of Kras overexpressed tumors. Genomic instability was also shown in human NSCLC cell lines H2170 (4 copies) and H226 (3 copies)(Fig. 6A) analyzed by using aCGH . High level of amplification at 8q24 was identified in both H2170 and H226 cells. Additional amplification was also identified in H2170 cells at 14q13. These results suggest that recurring DNA copy alterations occurred during tumorigenesis in this Cul4A mouse model. However, further study with a larger sample size of tumors is needed to confirm this hypothesis.
We next examined whether Cul4A overexpression is associated with drug resistance to the chemotherapy drug cisplatin in NSCLC. To test this hypothesis, we knocked down Cul4A expression in NSCLC using RNA interference and then measured the sensitivity of these NSCLC cells to cisplatin treatment. Among the six lung cancer cell lines expressing moderate to high levels of Cul4A compared to the weak Cul4A expression in the normal lung as shown by IHC analysis, the H2170 cell line had the highest copy number of Cul4A when analyzed using FISH (Fig. 6A). We next compared the inhibition effects of cisplatin on H2170 with or without Cul4A siRNA. The inhibition efficiency of RNA interference on Cul4A mRNA transcription was over 94% in four selected cancer cell lines (Supplementary Fig. 5A). Viability of H2170 cells was measured after 72 hours of cisplatin treatment (Fig. 6B, C). From the dose response curve, IC50 values were calculated in H2170 cells treated with cisplatin and either 100nM Cul4A siRNA (0.91 ± 5.8 μM), 100nM control siRNA (2.34 ± 5.5 μM), 1nM Cul4A siRNA (2.48 ± 5.1 μM), or 1nM control siRNA (2.31 ± 6.5 μM) (Fig. 6D). IC50 values in H2170 cells treated with cisplatin only averaged 3.35 ± 5.3 μM. We further compared the effect of the combination treatment in NSCLC cell lines H460, H838, and H2170 using the combination treatment of 50 nM Cul4A siRNA and cisplatin (Supplementary Fig. 6B–D), and all the NSCLC cell lines showed increased in their sensitivity to cisplatin when the Cul4A expression was knocked down. From the dose response curve, the IC50 values in H2170 cells (higher Cul4A copy number) decreased ~20% when Cul4A was knocked down, and the IC50 values in H460 or H838 cell lines (lower Cul4A copy number) was decreased ≤ 3% when Cul4A was knocked down (Supplementary Fig. 6E). Together, these results suggest that the combination treatment of Cul4A knock down and cisplatin showed a synergistic effect in inhibiting NSCLC cells.
Our study demonstrates that Cul4A is oncogenic in vivo and that the development of lung tumors can be induced by Cul4A overexpression. After Cul4A was overexpressed in lung tumors, the protein levels of p21CIP1, p19ARF and CDT1 decreased, whereas those of p27KIP1, p16INK4a, PCNA, and pericentrin increased. Our results suggest that these cellular events contribute to the initiation and development of the lung tumors in the mouse model of Cul4A overexpression, and that the expression of DNA replication licensing factor CDT1 may contribute to genomic instability and tumor development. Our study also showed that in addition to the oncogenic role of Cul4A expression in vivo, knocking down Cul4A expression in lung cancer H2170 cells increased their sensitivity to the chemotherapy drug cisplatin.
In IHC analysis of six human NSCLC cell lines, H2170 cells with four Cul4A copy numbers analyzed using FISH showed strong Cul4A staining score (+++)(Fig 6A). H226 and H345 cells with three Cul4A copy numbers showed strong to moderate IHC staining score (+++ and ++, respectively). H838, A549 and H460 cells with two Cul4A copy numbers showed moderate IHC staining score (++). In contract, normal lung tissue showed weak Cul4A staining score (+/−) (data not shown). We believe that our mouse model represents, at least in part, a fraction of human NSCLC with amplified Cul4A copy number.
Our IHC analysis indicated that the level of p27KIP1 expression was increased in lung tumors overexpressing Cul4A, suggesting that p27KIP1 is probably not proteolyzed by Cul4A-based E3 ubiquitin ligase in the lung tumors. This finding is supported by previous studies suggesting that p27KIP1 is not a target degraded by the Cul4A-based E3 ubiquitin ligase [31, 32]. For instance, depletion of the Cul4A-RING ubiquitin ligase (E3) complexes did not immediately increase p27KIP1 expression in the cells [31, 32]. In addition, p27KIP1 expression was elevated in response to UV-irradiation, regardless the presence of functional Cul4A-based E3 ubiquitin ligase complex in the cells . Together, these findings and ours all suggest that p27KIP1 is not a target degraded by Cul4A-based E3 ubiquitin ligase. Because p27KIP1 functions as a tumor suppressor and was strongly expressed in our Cul4A mouse model, knocking down the expression of p27KIP1 in our mouse model could provide further insights into the biology of human cancers that overexpress Cul4A.
It has been shown that Cul4A-based E3 ubiquitin ligase complex is required for p16INK4a activation in human lung cells . In clinical studies, inactivation of p16INK4a was reported in 40% to 70% of human NSCLC cases [34, 35]. Interestingly, p16INK4a expression was not decreased in the lung tumors of our Cul4A mouse model, suggesting that downregulation of p16INK4a expression in this model could provide further insights into the biology of Cul4A-mediated tumorigenesis. Therefore, by crossing this mouse model with a p16INK4a knockout strain, we generated a new mouse model that has p16INK4a depletion and oncogenic Cul4A overexpression. This new model provides an excellent opportunity to improve our understanding of the biology of cancers in which Cul4A is overexpressed, including NSCLC, mesothelioma, breast cancer, and skin carcinomas.
The results of our IHC analysis showed decreased expression of p19ARF in the lung tumors overexpressing Cul4A. This finding suggests that degradation of p19ARF protein occurred after Cul4A overexpression and that the loss of p19ARF may contribute to tumor development in this mouse model.
p21CIP1 deficient mice have been shown to develop different tumors in the mouse organs, including lung carcinomas . Because Cul4A can target multiple different proteins for degradation (Fig. 7), the phenotype of Cul4A transgenic mice is expected to be more tumorigenic than that in the p21CIP1 deficient mice.
The role Cul4A plays in the initiation and development of lung tumors in our mouse model has implications for understanding cancer biology and introduces new avenues for therapeutic development targeting Cul4A. These possibilities are supported by recent findings that MLN4924, a small molecule that blocks the activation of Cullin-based E3 ubiquitin ligase, suppresses tumor cell growth . The effect of MLN4924 in controlling the growth of lung tumors can be further investigated in our mouse model. To date, cisplatin, a platinum-based therapy, is one of the most commonly used chemotherapeutic agents in the treatment of NSCLC, but about 20% of patients experience a partial response to cisplatin [38, 39]. Our study found that the IC50 of NSCLC cell line H2170 decreased more than two-fold when Cul4A transcription was knocked down before the cisplatin treatment, suggesting that Cul4A overexpression is associated with cisplatin resistance in the cancer cells. This possibility is supported by the fact that a combination therapy consisting of MLN4924 and cisplatin significantly reduces tumor burden in mice, a finding that also suggests a mechanism to induce cancer cell apoptosis by trigging ROS production to induce DNA damage in cancer cells . Whether this mechanism is involved in the decreased cell viability in the H2170 cells treated with both Cul4A siRNA and cisplatin remains to be examined.
In summary, the Cul4A mouse model described here has many advantages over existing mouse models for cancer research. The ability to target specific organs through localized Ad-Cre infection can be useful to generate mouse models for studying the development of organ-specific tumors. Using this mouse model, we demonstrated for the first time that Cul4A is oncogenic in vivo and that the development of lung tumors can be induced by Cul4A overexpression. From these findings, we conclude that overexpression of Cul4A plays a role in tumor development in this mouse model (Fig. 7). A better understanding of the mechanisms of Cul4A-mediated lung tumorigenesis in this mouse model could provide new insights into the biology of human cancers that overexpress Cul4A and could contribute to future experimental therapies.
The present work was supported by NIH grant R01 CA140654-01A1 (LY). We are grateful for support from the Kazan, McClain, Abrams, Fernandez, Lyons, Greenwood, Harley & Oberman Foundation, Inc; the Estate of Robert Griffiths; the Jeffrey and Karen Peterson Family Foundation; Paul and Michelle Zygielbaum; the Estate of Norman Mancini; and the Barbara Isackson Lung Cancer Research Fund. We specially thank Osamu Tetsu, Ph.D. from UCSF Cancer Center for kindly provide the HA-p21CIP1-pcDNA3 plasmid as a gift. We thank Loretta Chan in the UCSF Cancer Center Tissue Core for her help. We also thank Pamela Derish in the UCSF Department of Surgery for editorial assistance with the manuscript.
Conflict of Interest Statement: The authors have no declared conflicts of interest.
Author ContributionsYY, MH, YW, DH and ZX performed the research; YY, MH, YW, NJ, DH, AA, AK, DQ, LF, CY, CP, ZX, DMJ, and LY analyzed data; YY, MH, YW and LY designed the research and wrote the paper.