Identification of breast cancer targets by functional genetic screening of overexpressed genes for tumor-promoting function and tumor cell dependency

doi:10.21203/rs.3.rs-1620431/v1

Download PDF

Research Article

Identification of breast cancer targets by functional genetic screening of overexpressed genes for tumor-promoting function and tumor cell dependency

https://doi.org/10.21203/rs.3.rs-1620431/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

There are several hundreds of genes overexpressed in a typical breast cancer cell, and their overexpression is often assumed to have a pro-cancer function, but what percentage of these overexpressed genes have a role in promoting and maintaining tumorigenicity isn’t known. To address this question, we screened an unbiased set of seventy-five genes overexpressed in breast cancer for their ability to induce malignant transformation and for breast cancer cell expression dependency. Barcoded open-reading frames under the control of a constitutive promoter were tested by pooled screening for their ability to promote tumor formation of premalignant mammary epithelial cells and growth in 2D and 3D culture. We found that growth in 3D cell culture was more strongly correlated with tumorigenic growth than was 2D growth, validating a long-standing hypothesis. Genes with high activity in promoting both 3D growth and in vivo tumor formation included CEACAM5, CXCL8, NKX2-5, PYGO2 and WWTR1/TAZ. To screen for corresponding tumor dependencies on these six genes, we performed shRNA screening of breast cancer cell lines and found uniform dependency on expression of the Hippo pathway transcriptional activator WWTR1, selective dependencies on CEACAM5 expression, but no consistent effects for the other three genes. Since breast cancer dependency upon WWTR1 is well established, we focused our follow-up studies on CEACAM5. We confirmed that CEACAM5 overexpression conferred tumorigenicity to pre-malignant mammary epithelial cells and that reduced expression selectively decreased the clonogenicity of luminal breast cancer cell lines. Protein analysis indicated that CEACAM5 activated Cdk4/6 and mTOR signaling. We also determined that CEACAM5 overexpression strongly promoted cell number increase in 3D spheroid culture by a novel mechanism that increased the number of epithelial cell layers from a single layer to multiple layers. These results outline a strategy for functional genomic identification of cancer targets, starting with a set of genes overexpressed in cancer, then using pooled approaches to identify those that promote tumorigenicity and cancer cell dependency, followed by validation experiments and exploration of oncogenic mechanisms. We propose that larger screens would substantially increase the number of potential targets for treating breast cancer.

Genome-wide estimates of the number of genes overexpressed in specific cancer types range in the low hundreds to high hundreds (Hibi et al., 1998; Logsdon et al., 2003; Welsh et al., 2001). Depending upon their expression pattern in cancers, these genes can be useful biomarkers for detection, prognosis, or prediction of treatment response. However, determining whether they promote tumorigenicity or evaluating their utility as potential therapeutic targets requires functional analysis. Overexpressed genes may actually be growth-inhibitory, particularly if their overexpression results from a dysregulated negative feedback loop. For example, the HPV oncoprotein E7 binds and induces proteolytic degradation of the Rb protein, which relieves the tumor suppressor gene p16^Ink4a from negative feedback control, leading to its upregulation (Romagosa et al., 2011). Likewise, BRAF-V600E mutant melanomas and KRAS-mutant lung adenocarcinomas both overexpress several inhibitors of the mitogen-activated kinase pathway (Li et al., 2016; Pratilas et al., 2009). In these cases, the primary genetic alteration blocks the negative feedback exerted by the overexpressed inhibitory genes (Pratilas et al., 2009; Romagosa et al., 2011).

Single-gene, small-scale functional analysis of overexpressed genes has led to the generation of hundreds of proposed tumor-promoting genes (Liu et al., 2017). At present, it is difficult to evaluate many of these studies since they often use unique experimental conditions to determine oncogenicity. More systematic approaches that survey the induced oncogenic phenotypes of multiple candidate genes have the advantage of providing several internal comparisons. Using cDNA-based or open-reading-frame (ORF) overexpression, systematic screening has identified oncogenic roles for IKBKE, NECTIN4/PVRL4 and PAK1 in breast cancer (Boehm et al., 2007) (Pavlova et al., 2013; Shrestha et al., 2012) and FGF19 in hepatocellular carcinoma (Sawey et al., 2011). In addition to these systematic screens for genes capable of promoting the tumorigenicity of premalignant cells, overexpression screens have also been used to identify the drug resistance gene COT (Johannessen et al., 2010) and several drivers of metastasis including ACP5, GATAD2B, and PYGO2 (Grzeskowiak et al., 2018; Lu et al., 2018; Scott et al., 2011).

In this study, we first set out to determine what percentage of overexpressed genes in breast cancer were capable of promoting tumorigenicity of premalignant mammary epithelial cells. We performed pooled screening with barcoded ORFs and measured the relative number of mammary epithelial cells expressing each individual ORF both before and after growth in three different conditions, including in vivo tumor formation. Our second step was to determine which if the overexpression of the identified tumor-promoting genes was selectively required in breast cancer cells. For this we used shRNA screening in breast cancer cell lines to characterize growth-dependencies on expression of tumor-promoting genes.

Selection of overexpressed genes and transfection of LentiORFs into both MCF10A and NMuMG

We selected genes based on their overexpression in breast cancer from a previously acquired Precision LentiORF expression library containing 3526 ORFs. We applied both outlier and cancer vs. normal differential analysis of breast cancer TCGA data using Oncomine Research tools and chose the top 100 overexpressed genes for functional studies (Supplementary Table 1). We performed a pilot experiment with ten of the chosen ORFs and measured the ability of amplicon sequencing to detect the different ORFs after transfection into MCF10A cells, plasmid integration and selection, and one week of growth before amplifying the ORFs from genomic DNA using universal 5’ and 3’ ORF primers. Based on the uneven amplification of the ORFs, we decided to switch to a 24-bp barcoded ORF system which promised to yield more even amplification of the uniform-length and balanced nucleotide composition barcodes. We tagged each individual ORF with a unique barcode for next-generation sequencing and inserted the tagged-ORF into the pLenti6.3 R1R2 lentiviral expression vector using the production pipeline previously described (Tsang et al., 2016). We successfully transferred and sequence-verified 75 ORFs, and these barcoded-ORF expression vectors were transfected individually (to ensure an initial even representation) into the immortalized human mammary epithelial cell line MCF10A (Debnath et al., 2003b). MCF10A resemble primary breast epithelial cells in that they require substrate attachment, growth factors and hormones for growth, are contact-inhibited, and are not tumorigenic. They have two cancer-related genetic alterations: homozygous deletion of the CDKN2A locus and amplification of MYC, and this allows them to be transformed in vitro by mutant PIK3CA or RAS oncogenes, but acquiring the ability to form carcinomas in vivo requires several additional oncogenic alterations (Croessmann et al., 2017; Kadota et al., 2010; Pires et al., 2013). Mouse cells require fewer genetic alterations to form tumors, thus for in vivo screening we chose the normal murine mammary epithelial cell line NMuMG, which is heterozygous for a missense TP53 mutant (R277C) and forms tumors when transfected with powerful oncogenes such as mutationally activated RAS (Salomon et al., 1987; Termen et al., 2013). In order to allow NMuMG cells to form tumors with weaker oncogenes, we overexpressed MYC, a procedure that we had used previously in our cDNA screen for tumor-promoting genes using TP53 mutant hepatoblasts (Sawey et al., 2011). As we did with MCF10A, the 75 barcoded-ORF expression vectors were transfected individually into NMuMG-Myc cells.

Following selection of transfected cells with blasticidin, the individual transfectants were cultured briefly and then equal cell numbers of each were combined into a pool. Each pool was then divided into three separate cultures to generate biological replicates. After three days, there were sufficient cells from each replicate to use for isolation of genomic DNA, PCR amplification of barcodes, and amplicon sequencing to determine the baseline representation of the different ORF-transfectants. The MCF10A ORF-transfectant pool replicates were then placed into 3D culture (Matrigel) for three weeks to allow for spheroid formation, and in parallel grown for three weeks under standard 2D cell culture conditions (with passaging as needed to avoid confluence). Each of the NMuMG-Myc ORF-transfectant pool replicates were injected subcutaneously into immunodeficient mice and the mice were sacrificed after 8 weeks and the tumors removed. None of the mice injected with control NMuMG-Myc empty vector transfectants formed tumors after 8 weeks.

For each of these growth conditions, genomic DNA was isolated and the barcode read counts determined by amplicon sequencing. Following preprocessing, normalization, and log-transformation of the barcode read counts, the ORF frequencies in the three biological replicates of the initial time point were compared to the ORF frequencies in the latter time points.

Overexpressed genes vary widely in their effects on tumor formation

The seventy-five lentiviral ORF transfectants showed variable degrees of enrichment or depletion during the eight weeks of in vivo growth (Supplementary Table 2). Of the thirteen ORF transfectants that were enriched (p < 0.05), only eight showed more than 2-fold enrichment. In contrast, thirty-three ORF transfectants were depleted (p < 0.05), and twenty-eight showed more than 2-fold depletion, including five that were completely depleted. An additional eight showed a tendency towards depletion (0.05 < p < 0.10), and twenty appeared to be neutral (Fig. 1A).

For the eight lentiviral ORF transfectants that showed more than 2-fold enrichment, six of the corresponding genes have been previously shown to either promote tumorigenicity or tumorigenic properties although not necessarily in breast cancers. These six genes include two oncogenic homeobox domain transcription factors which are activated by chromosomal rearrangements in acute lymphoblastic leukemia, NKX2-1 and NKX2-5 (Homminga et al., 2011; Nagel et al., 2003); the proinflammatory cytokine CXCL8 which has an autocrine function in stimulating the growth of triple-negative breast cancer cells (Hartman et al., 2013); the transcriptional activator WWTR1/TAZ which induces cancer stem-cell phenotypes in breast cancer (Cordenonsi et al., 2011); and CEACAM5 and PYGO2 which have both been shown to promote metastasis when overexpressed (Lu et al., 2018; Powell et al., 2018) (Fig. 1B). ALDH3B1 and ALDH3B2 have not previously associated with oncogenicity, but their known function of detoxifying lipid aldehydes generated by oxidative stress could in principle enable tumor cells to survive or grow faster with higher ROS levels (Kitamura et al., 2015; Marchitti et al., 2010).

Several of the genes with high tumor-inhibitory activity have either known or plausible mechanisms for suppressing tumor growth (Fig. 1C). Placental growth factor (PGF) is a member of the VEGF family that when ectopically overexpressed forms heterodimers with VEGF and inhibits VEGF homodimer formation, angiogenesis, and tumorigenicity (Xu et al., 2006). Expression of retinol dehydrogenase 16 (RDH16), involved in the first and rate-limiting step of converting retinol

into all-trans-retinoic acid (ATRA), antagonizes cancer stem-cell renewal (Hu et al., 2017). IL12RB1 ectopic overexpression has been shown to inhibit the activity of endogenous Jak kinases by a dominant-negative mechanism (Dondi et al., 2001), and breast cancer progression has been shown to be dependent on Jak kinases (Chang et al., 2018; Wehde et al., 2018).

It was initially surprising to find that two of the strongest tumor-inhibitory genes in this screen, NECTIN4/PVRL4 and HMGA1, have been shown to promote tumorigenicity when overexpressed in other systems (Liau et al., 2006; Pavlova et al., 2013). Although our results with these genes appear to be contradictory, we note that extensively validated cancer genes (RUNX1 and NOTCH family members) have been shown to have dual roles of either tumor suppression or survival/tumor promotion depending upon the specific cell type (Goyama et al., 2013; Klinakis et al., 2011). We think it is likely that some overexpressed genes will also have opposing roles that are cell-type dependent.

Tumor formation is more closely related to 3D cell culture growth than 2D

To test the prediction that tumorigenic growth would be more closely related to 3D growth than 2D growth, we tested for associations between these two in vitro growth measurements with in vivo growth. Since none of these measures had a normal distribution by visual examination of density plots, quantile-quantile plots, or the Shapiro-Wilk test, we did not use Pearson’s correlation coefficient, and instead used the Kendall rank correlation coefficient to compare the associations. The correlation between 3D growth and tumor-promoting activity was significantly higher (τ = 0.44) than was the correlation with 2D growth (τ = 0.22). One of the genes driving this closer association was NKX2-5, expression of which promoted both tumorigenicity and growth in 3D but inhibited standard growth in 2D culture (Fig. 2). These results substantiate the hypothesis that for epithelial cells, 3D culture more closely resembles growth in vivo than does standard 2D cell culture.

shRNA screening reveals breast cancer cell dependencies upon WWTR1 and CEACAM5

The pooled ORF screening results pinpointed five genes with high activity in promoting both 3D growth and in vivo tumor formation: CEACAM5, CXCL8, NKX2-5, PYGO2, and WWTR1 (Fig. 3A). To determine the extent to which breast cancer cells depended upon expression of these five genes, we performed pooled shRNA screening. For each of the five genes, six independent shRNAs were designed and the shRNA pool was used to screen for effects on cell growth in both standard cell culture as well as cells grown in non-adherent culture conditions. Three breast cancer cell lines were employed, including CAL120 (basal-like; triple-negative), MCF7 (luminal, estrogen-receptor positive), and MDA-MB-453 (luminal-like, triple-negative). The p-values of individual shRNA effects were combined into gene scores using Stouffer’s method. Significant dependencies (negative growth effects

of shRNAs) were observed for three genes, CEACAM5, CXCL8, and WWTR1. All three cell lines were dependent on WWTR1 for 2D growth and both luminal cell lines were dependent on CEACAM5 for 2D growth (Fig. 3B). Although MCF7 were dependent on CXCL8 expression for 2D growth, the opposite effect of growth enhancement was observed when cells were grown without attachment. CXCL8-suppressed CAL120 cells also showed growth enhancement in non-adhesive conditions (Fig. 3B). Suppression of PYGO2 expression did not significantly attenuate growth of any cell line but did significantly enhance growth for MDA-MB-453. This latter finding is in agreement with the increased fitness of PYGO2 deficiency in triple-negative breast cancer cell lines reported in the Cancer Dependency Map (https://depmap.org/portal/).

The dependency of all three cell lines on the Hippo transducer WWTR1 is consistent with published reports whereas the dependency on CEACAM5 in luminal breast cancer cells has not been previously reported. Consequently, we focused our follow-up studies on the cell adhesion gene CEACAM5. We first tested the validity of the pooled screening result that CEACAM5 expression promoted the tumorigenicity of NMuMG-Myc cells. NMuMG-Myc cells were transduced with a lentiORF CEACAM5 expression construct and shown by quantitative RT-PCR to express high levels of CEACAM5. We determined that CEACAM5 overexpression induced rapidly growing tumors following subcutaneous injection of 5x10⁶ cells. Over the same time period of four weeks, orthotopic injection of 5x10⁶ cells into mammary fat pads produced slightly smaller tumors. Control NMuMG-Myc cells did not form tumors under either of these conditions (Supplementary Fig. 1).

shRNA mediated suppression of CEACAM5 in reduces clonogenic growth in ER-positive cell lines and enhances tamoxifen-growth inhibition in T47D

Next we tested the pooled shRNA screening results indicating tumor dependency on CEACAM5 in luminal breast cancer cell lines. We constructed two mir-30 based shRNAs plasmids targeting CEACAM5 and determined that each one had on-target activity in suppressing CEACAM5 expression (Fig. 4). Both of these shRNAs suppressed the clonogenic growth of two luminal breast cancer cell lines, T47D and MCF7, but neither shRNA was able to suppress the clonogenic growth of unrelated kidney epithelial 293T cells (Fig. 4). To address whether the selective growth-inhibition of these two shRNAs were due to off-target effects, we performed a RNAi-rescue experiment with a murine Ceacam5 cDNA which unlike the human ORF does not contain corresponding nucleotide sequence matches for the two targeting shRNAs. T47D cells expressing the murine Ceacam5 were then tested to determine whether the two human CEACAM5-targeting shRNAs could suppress clonogenic growth, but they did not, establishing that the clonogenic growth inhibition by the two shRNAs of luminal breast cancer cells was not due to off-target effects (Fig. 5A).

High CEACAM5 protein expression as ascertained by immunohistochemistry has been associated with resistance to endocrine therapy in ER-positive breast cancer patients (Bartlett et al., 2012). Based on this, we examined whether suppression of CEACAM5 expression could modify growth-inhibitory effects of tamoxifen in the two ER-positive cell lines T47D and MCF7. Although suppression of CEACAM5 expression did not modify the response of MCF7 (data not shown), it enhanced growth inhibition by tamoxifen in T47D (Fig. 5B). This result suggests that it is possible that inhibition of CEACAM5 could be of therapeutic benefit in treating a subset of ER-positive breast cancer patients.

RPPA analysis of the effects of CEACAM5 overexpression on the functional proteome

To gain insight into how CEACAM5 overexpression induced tumorigenicity, we used the high-throughput reverse-phase protein analysis (RPPA) pipeline established at MD Anderson to examine how CEACAM5 expression affected levels and post-translational modifications of several key proteins involved in growth control and other cancer-related phenotypes (Akbani et al., 2014). We transfected MCF10A cells with either the CEACAM5 ORF expression construct, the WWTR1 expression construct for comparative purposes, or empty vector as a control. Following selection for stable integration, we cultured the cells briefly in standard medium and prepared biological triplicates for RPPA analysis. Out of a total of 304 protein measurements, 51 were significantly different between CEACAM5 and control empty vector (p < 0.05). The majority of these changes were in common with the effects of overexpression of WWTR1, indicating a substantially shared oncogenic program, including increased phosphorylation-mediated inactivation of the Rb1 negative regulator of cell cycle progression, increased levels of phosphorylation-mediated activation of mTOR, increased levels of IGF1R and SRC tyrosine kinases, decreased levels of the cell cycle inhibitors p27^Kip (CDKN1B) and CHK2, and decreased levels of the protein translation inhibitor Pdcd4 (Figure S2).

Differences between the oncogenic effects of the two genes include WWTR1 expression inducing increased levels of the oncogenic receptor tyrosine kinase Axl, whereas CEACAM5 expression was associated with decreased Axl levels (Figure S2). On the other hand, CEACAM5 expression induced decreased levels of the growth suppressive TP53BP1 protein, whereas WWTR1 did not (Figure S2). Additionally, CEACAM5 expression but not WWTR1 expression was associated with reduced levels of the DNA/RNA helicase Schlafen-11 (SLFN11), indicating that high expression of CEACAM5 may be linked to resistance to diverse DNA-damaging agents (Zoppoli et al., 2012).

Overexpression of CEACAM5 activates Akt and induces independence from added insulin

MCF10A cells require both insulin and EGF growth factors for optimal growth (Soule et al., 1990). We tested the ability of CEACAM5 overexpression to substitute for these two growth factors. CEACAM5 overexpressing cells were completely independent of insulin as a growth factor, and although less dependent upon EGF than control MCF10A cells, they still showed EGF dependence (Fig. 6). Previous studies that examined the biochemical effects of CEACAM5 overexpression in L6 myoblasts showed activation of Akt (Camacho-Leal and Stanners, 2008). To determine if this held true in normal mammary epithelial cells and provided a mechanism for insulin independent growth, we examined the activated phosphorylation status of Akt in MCF10A-CEACAM5 and MCF10A-control cells under normal growth conditions as well as growth-factor depleted conditions. CEACAM5 overexpression induced high levels of Akt activation in both growth conditions compared to control MCF10A cells (Fig. 6).

Overexpression of CEACAM5 promotes spheroid growth by increasing the number of cell layers

The pooled ORF screening results indicated that overexpression of CEACAM5 greatly increased the growth of MCF10A cells in 3D culture. To test this, we examined MCF10A-CEACAM5 after growth in a 3D matrigel matrix and determined that CEACAM5 overexpression induced a modest but significant increase in the average diameter of spheroids formed in matrigel (Fig. 7). This modest increase could not explain the stronger growth predicted from the screening results. However, examination of spheroids after they were fixed, sectioned, and visualized by staining with DAPI revealed that CEACAM5 overexpression increased growth significantly within the spheroid itself. Unlike the oncogenes ERBB2, PIK3CA, or WWTR1 previously tested in the MCF10A spheroid formation assay, CEACAM5 overexpression did not alter the spheroid morphology or prevent hollowing out of the inner mass of cells (Debnath et al., 2003a), but instead induced a notable increase in growth in the spheroid periphery, increasing the epithelial cell layers (Fig. 7). This effect is similar to the hyperproliferation of epithelial cells lining acini seen in early precursor lesions of breast cancer (Allred et al., 2001). A similar effect was observed in the colon of transgenic mice with multiple copies of the human locus containing CEACAM5 and its close homolog CEACAM6. In addition to hyperproliferation, the colon of these mice showed increased crypt fission, a process that increases the number of epithelial cell layers (Chan et al., 2007).

A major impetus for undertaking this project was to determine whether functional analysis of overexpressed genes in breast cancer could lead to the identification of robust therapeutic targets, potentially extending the successful targeting of genetically altered oncoproteins paradigm (amplified ERBB2, mutationally activated PIK3CA) to oncoproteins that are activated transcriptionally either through epigenetics or some other mechanism. Such an approach, if successful, could lead to a coupled therapeutic/diagnostic where overexpression of the target protein could pinpoint patients who would respond to treatment with the corresponding therapeutic. A variety of alternative strategies for identifying breast cancer cell targets that bypass the genetically altered oncoproteins paradigm have been developed. These include synthetic lethal screening, and targeting proteins involved in either cancer stem cell maintenance, the epithelial-to-mesenchymal transition (EMT), or cancer cell metabolism (Hahn et al., 2021). However, many of these alternative strategies have been explored for over twenty years without any clinical impact. As a result, we believe it is incumbent upon us to take a critical look at the results of our project to fairly evaluate whether this approach could ultimately lead to clinical success.

One of the key positive findings of our study was that 10% of overexpressed genes increased fitness. By studying one in depth, CEACAM5, we found that tumor cells with overexpression of CEACAM were dependent upon it for optimal growth and tumorigenicity. If generalizable, breast cancers may be at least partially dependent upon a much larger set of genes that currently believed. This clearly has implications for breast cancer target discovery and needs to be explored further.

However, we also found that two of the strongest tumor-inhibitory genes of our screen, NECTIN4/PVRL4 and HMGA1, were shown to promote tumorigenicity when overexpressed in other systems (Liau et al., 2006; Pavlova et al., 2013). Therefore some overexpressed genes will have opposing roles that are context or cell-type dependent, similar to how RUNX1 and NOTCH family members have been shown to have dual roles of either tumor suppression or survival/tumor promotion depending upon the specific cell type (Goyama et al., 2013; Klinakis et al., 2011).

Along these lines, a limitation of our study is that we did not utilize a more diverse set of normal mammary cells. The two cell lines that we used, MCF10A and NMuMG, are both estrogen-receptor negative yet form luminal mammospheres when placed into three dimensional culture. MCF10A expresses both luminal and basal markers (Qu et al., 2015), and there is uncertainty as to what normal mammary cell type it represents. Clearly, our approach would be more robust if there was a wider variety of normal mammary epithelial cells to screen, particularly ones that are estrogen-receptor positive.

Another key finding of our study was three dimensional growth in Matrigel was much better correlated with tumorigenicity than standard 2D tissue culture. The vast majority of cancer screens are still being carried out using standard cell culture conditions, even though the deficiency of standard cell culture growth assays has been well known for many years and can be improved considerably by relatively straightforward modifications of cell culture conditions such as anchorage-independent growth or growth without complete serum or growth factors (Rotem et al., 2015; Zhao et al., 2021).

In summary, we believe that although our approach has potential, a major roadblock is the lack of a panel of normal mammary epithelial cells by which to perform the screens. This roadblock affects many other strategies aimed at discovering new breast cancer targets, and we hope that the cancer community can overcome this shortcoming in the near future.

Barcoded-ORF library and shRNA library generation

The barcoded-ORF library was constructed by Dr. Kenneth Scott’s laboratory at Baylor College of Medicine with the human ORFeome version 8.1 Entry Clone Collection (hORFeome V8) as donors and pLenti6.3/V5 as destination vector using the Gateway recombination-based barcoding pipeline as described (Tsang et al., 2016). Pooling was done post-lentiviral transduction as described below. The shRNA library pool was designed and constructed at Dr. Michael Hemann’s laboratory at the Koch Institute at MIT with six independently designed shRNA for each genefollowing published proceduusing oligonucleotides bulk synthesized by LC Sciences (LC Sciences OligoMix) that were PCR amplified as described (Meacham et al., 2015). The amplified shRNAs were then batch cloned into the mir30-retroviral vector LMP (Dickins et al. 2005) and then 20 clones of the library were Sanger sequenced to validate the library.

Pooled Screening

The same set of ORFs were lentivirally transduced into both murine non-malignant NMuMG-Myc mammary epithelial cells and human MCF10A cells. To prepare lentivirus, each individual ORF plasmid was transfected separately along with helper plasmids into 293T cells using jetPRIME transfection reagent (Polyplus Transfection) and virus was collected after 3 days. MCF10A and NMuMG-Myc cells were individually infected for 24 hours, followed by addition of fresh media and 48 hours before adding blasticidin to select for stable transfectants. After 8 days the individually transfected cells were then pooled and split into four biological replicates which were cultured for an additional 3 days before dividing them into the four conditions: (1) time zero, from which genomic DNA was isolated and used for PCR amplification of the barcodes; (2) 2D cell culture for 30 days (with periodic splitting to avoid reaching confluence); (3) 3D culture of single cells in matrigel for 30 days; and (4) aliquots comprised of 2x10⁶ cells were injected subcutaneously into nude mice for in vivo tumor formation for 8 weeks. Following genomic DNA isolation and PCR amplification of the barcodes, each PCR amplicon product was ligated to a unique index adaptors for construction of a multiplexed Illumina library using the TrueSeq DNA LT Sample Prep Kit (Illumina) for MiSeq sequencing. After mapping reads to the set of utilized barcode sequences, counts for each ORF were normalized by the number of total number of mapped reads per amplicon. Following log2 transformation, the four values from the endpoint samples were compared to the corresponding initial values from the zero time point samples as described (Tsang et al., 2016).

For shRNA screening, the LMP shRNA plasmid library and helper plasmid was transfected into Bing packaging cells with Profection Mammalian Transfection System. After 48 hours, the viral supernatant was collected and filtered. Breast cancer cell lines MCF7, Cal120, MDA-MB-453 were infected at a MOI of 0.3 using diluted viral supernatant supplemented with 8 μg/ml polybrene. 72 hours after the infection, infected cells were selected using puromycin for 3 days. The cells were then split into separate cultures for generating quadruple replicates for zero time point cultures, quadruple replicates for two-weeks growth using normal tissue culture plates, and quadruple replicates for two-weeks growth on non-adhesive plates coated with poly-HEMA. The cells were collected and genomic DNA was extracted using E.Z.N.A Tissue DNA Kit (Omega Bio-Tek). PCR reactions was done using specific primers: half forward (5’-TAGTGAAGCCACAGATGTA) and bc1-reverse (5’-AAAGCGCATGTCCAGACTGCC) to amplify hairpin sequence. Each PCR amplicon was prepared for Illumina Miseq library using the TrueSeq DNA LT Sample Prep Kit (Illumina), where each amplicon was ligated to unique index Adaptors for MiSeq sequencing. Following normalization and log2 transformation, the quadruple values from the endpoint samples were compared to the corresponding initial values from the zero time point samples as described (Meacham et al., 2015).

Animals

All studies utilizing mice were approved by and in accordance with Stony Brook University Institutional Animal Care and Use Committee. Four to five week old female nude mice (NCR nu/nu; Envigo Inc., Wilmington, MA) One million mammary epithelial cells were trypsinized, resuspended in 100 µl DMEM and injected subcutaneously or orthotopically into mammary pads. Growth was followed over time by taking caliper measurements at indicated time points. Tumor volume was measured as 0.52×length×width2. Tumors were excised six-eight weeks post injections or when one of the measurements reached 2 cm.

Tissue culture and lentiviral infection

MCF10A, T47D, HEK293T, MCF7, MCF10A, NMuMG cells were obtained from the American Type Culture Collection (ATCC). HEK293T and T47D were cultured in DMEM with 10% fetal bovine serum (FBS) and 100 U/ml penicillin streptomycin. MCF10A cells were cultured in DMEM/F12 supplemented with 5% donor horse serum, EGF 20ng/mL, insulin 10ug/mL, hydrocortisone 100ug/mL cholera toxin 10ng/mL and 100 U/ml penicillin streptomycin. MCF7 and NMuMG-myc cells were cultured in DMEM with 10% fetal bovine serum (FBS), insulin 10ug/mL and 100 U/ml penicillin streptomycin. To produce lentiviral particles, HEK293T cells were transiently transfected with transfer, envelope, and packaging plasmids using Profection Mammalian Transfection System (Promega). The viral supernatant was collected 24 hours later. For lentiviral infection, MCF10A cells were grown in a 1:4 mixture of the growth medium and the viral supernatant for 24 hours. Two days post infection, blasticidin (Invivogen) selection was performed for 8 days to select for positive clones and these cells were used for various assays. Colony formation assays were performed by plating 1000 cells in triplicate on 6-well plates. Medium was changed every 3 days. After 3 weeks culture, cells were methanol fixed and stained with 0.5% crystal violet. After pictures were taken, crystal violet was dissolved with 0.1% SDS overnight. Dissolved crystal violet staining was read at 595 nm using the Victor 3 machine (Perkin Elmer). 3D culture was performed as previously reported (Debnath et al., 2003) using growth factor-reduced matrigel (BD Bioscience).

RPPA and immunoblotting

RPPA was performed by the MD Anderson RPPA Core Facility using protein lysates. For each RPPA replicate, cells were washed with cold PBS twice and lysed with 100 μl lysis buffer (1% Triton X-100, 50 mM HEPES, pH 7.4, 150 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, 100 mM NaF, 10 mM Na pyrophosphate, 1 mM Na3VO4 and 10% glycerol, containing freshly added protease and phosphatase inhibitors from Roche Applied Science Cat. # 05056489001 and 04906837001, respectively), and mixed with 4 × SDS sample buffer (40% Glycerol, 8% SDS, 0.25 M Tris-HCl, pH 6.8). Immunoblotting was carried out using antibodies against Ceacam5 (Sigma), Akt, phospho AKT (Cell Signaling and Actin (Abcam) or GAPDH (Santa Cruz) as loading control. Cells were lysed in lysis buffer (25mM Tris, pH 7, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X 100) and cocktail protease inhibitors (Roche) and centrifuged to clear the cell lysates. The cell lysates were then loaded on a 10% SDS-PAGE, and the proteins were subsequently transferred onto a nitrocelluose membrane. The membrane was blocked with 5% milk in TBST (20 mM Tris-HCl, pH7.5, 150 mM NaCl, 0.1% Tween-20) for 30 mins and then incubated with antibodies in 5% milk in TBST overnight. After three extensive washes for 10 minutes each in TBST, the membrane was incubated with the secondary antibody, horseradish peroxidase (HRP)-conjugated anti-rabbit (Santa Cruz) or anti-mouse IgG antibody (Abcam) for 1 hour, followed by three washes in TBST. The membrane was developed with the chemiluminascent kit (Biorad) and exposed to X-ray films (Kodak).

Other methods

For real-time RT-PCR, total mRNA was isolated from the indicated cells using the RNeasy Mini Kit (Qiagen). cDNAs were synthesized using the cDNA Synthesis Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. The transcript level of GAPDH was used as an internal control. The amplification steps consisted of 10 minutes at 50°C and 5 minutes at 95°C, followed by 40 cycle of denaturation for 10 seconds at 95°C and annealing/extension for 30 seconds at 60°C. All samples were analyzed in triplicate, and the relative gene expression was calculated according to the comparative threshold cycle (ΔΔCt) method (Livak and Schmittgen, 2001).

Student’s t-test was used to determine the significance of difference of means measurements, using the standard p < 0.05 cutoff. Error bars represent the standard deviation of data.

For analysis of 3D Matrigel cultures, spheroids were photographed and their diameters were assessed using Leica Q500 MC Qwin software. For immunofluorescence staining, acini were fixed in 4% paraformaldehyde, washed in PBS and embedded in optimum cutting temperature (OCT) compound (Tissue-Tek) and then processed by regular histological procedures. 4’, 6-diamidino-2-phenylindole (DAPI) (Molecular Probes) was used as DNA stain.

Acknowledgements

Research reported in this publication was supported by the NIH under award numbers R01CA217206 and U01CA168409, a Peter Rowley Breast Cancer Research Award from the New York State Department of Health, the Stony Brook Cancer Center, the Babylon Breast Cancer Coalition, the Carol M. Baldwin Breast Cancer Research Fund, and the Walk-for-Beauty Foundation Award. We acknowledge the Stony Brook Cancer Center Biostatistics & Bioinformatics Shared Resource for expert assistance with bioinformatic analysis. The Functional Proteomics RPPA Core facility is supported by MD Anderson Cancer Center Support Grant # 5 P30 CA016672-40.

Akbani, R., Ng, P. K., Werner, H. M., Shahmoradgoli, M., Zhang, F., Ju, Z., Liu, W., Yang, J. Y., Yoshihara, K., Li, J., et al. (2014). A pan-cancer proteomic perspective on The Cancer Genome Atlas. Nat Commun 5, 3887.

Allred, D. C., Mohsin, S. K., and Fuqua, S. A. (2001). Histological and biological evolution of human premalignant breast disease. Endocr Relat Cancer 8, 47-61.

Bartlett, J. M., Bloom, K. J., Piper, T., Lawton, T. J., van de Velde, C. J., Ross, D. T., Ring, B. Z., Seitz, R. S., Beck, R. A., Hasenburg, A., et al. (2012). Mammostrat as an immunohistochemical multigene assay for prediction of early relapse risk in the tamoxifen versus exemestane adjuvant multicenter trial pathology study. J Clin Oncol 30, 4477-4484.

Boehm, J. S., Zhao, J. J., Yao, J., Kim, S. Y., Firestein, R., Dunn, I. F., Sjostrom, S. K., Garraway, L. A., Weremowicz, S., Richardson, A. L., et al. (2007). Integrative genomic approaches identify IKBKE as a breast cancer oncogene. Cell 129, 1065-1079.

Camacho-Leal, P., and Stanners, C. P. (2008). The human carcinoembryonic antigen (CEA) GPI anchor mediates anoikis inhibition by inactivation of the intrinsic death pathway. Oncogene 27, 1545-1553.

Chan, C. H., Camacho-Leal, P., and Stanners, C. P. (2007). Colorectal hyperplasia and dysplasia due to human carcinoembryonic antigen (CEA) family member expression in transgenic mice. PLoS One 2, e1353.

Chang, R., Song, L., Xu, Y., Wu, Y., Dai, C., Wang, X., Sun, X., Hou, Y., Li, W., Zhan, X., and Zhan, L. (2018). Loss of Wwox drives metastasis in triple-negative breast cancer by JAK2/STAT3 axis. Nat Commun 9, 3486.

Cordenonsi, M., Zanconato, F., Azzolin, L., Forcato, M., Rosato, A., Frasson, C., Inui, M., Montagner, M., Parenti, A. R., Poletti, A., et al. (2011). The Hippo transducer TAZ confers cancer stem cell-related traits on breast cancer cells. Cell 147, 759-772.

Croessmann, S., Wong, H. Y., Zabransky, D. J., Chu, D., Rosen, D. M., Cidado, J., Cochran, R. L., Dalton, W. B., Erlanger, B., Cravero, K., et al. (2017). PIK3CA mutations and TP53 alterations cooperate to increase cancerous phenotypes and tumor heterogeneity. Breast Cancer Res Treat 162, 451-464.

Debnath, J., Muthuswamy, S. K., and Brugge, J. S. (2003a). Morphogenesis and oncogenesis of MCF-10A mammary epithelial acini grown in three-dimensional basement membrane cultures. Methods 30, 256-268.

Debnath, J., Muthuswamy, S. K., and Brugge, J. S. (2003b). Morphogenesis and oncogenesis of MCF-10A mammary epithelial acini grown in three-dimensional basement membrane cultures. Methods 30, 256-268.

Dondi, E., Pattyn, E., Lutfalla, G., Van Ostade, X., Uze, G., Pellegrini, S., and Tavernier, J. (2001). Down-modulation of type 1 interferon responses by receptor cross-competition for a shared Jak kinase. J Biol Chem 276, 47004-47012.

Goyama, S., Schibler, J., Cunningham, L., Zhang, Y., Rao, Y., Nishimoto, N., Nakagawa, M., Olsson, A., Wunderlich, M., Link, K. A., et al. (2013). Transcription factor RUNX1 promotes survival of acute myeloid leukemia cells. J Clin Invest 123, 3876-3888.

Grzeskowiak, C. L., Kundu, S. T., Mo, X., Ivanov, A. A., Zagorodna, O., Lu, H., Chapple, R. H., Tsang, Y. H., Moreno, D., Mosqueda, M., et al. (2018). In vivo screening identifies GATAD2B as a metastasis driver in KRAS-driven lung cancer. Nat Commun 9, 2732.

Hahn, W. C., Bader, J. S., Braun, T. P., Califano, A., Clemons, P. A., Druker, B. J., Ewald, A. J., Fu, H., Jagu, S., Kemp, C. J., et al. (2021). An expanded universe of cancer targets. Cell 184, 1142-1155.

Hartman, Z. C., Poage, G. M., den Hollander, P., Tsimelzon, A., Hill, J., Panupinthu, N., Zhang, Y., Mazumdar, A., Hilsenbeck, S. G., Mills, G. B., and Brown, P. H. (2013). Growth of triple-negative breast cancer cells relies upon coordinate autocrine expression of the proinflammatory cytokines IL-6 and IL-8. Cancer Res 73, 3470-3480.

Hibi, K., Liu, Q., Beaudry, G. A., Madden, S. L., Westra, W. H., Wehage, S. L., Yang, S. C., Heitmiller, R. F., Bertelsen, A. H., Sidransky, D., and Jen, J. (1998). Serial analysis of gene expression in non-small cell lung cancer. Cancer Res 58, 5690-5694.

Homminga, I., Pieters, R., Langerak, A. W., de Rooi, J. J., Stubbs, A., Verstegen, M., Vuerhard, M., Buijs-Gladdines, J., Kooi, C., Klous, P., et al. (2011). Integrated transcript and genome analyses reveal NKX2-1 and MEF2C as potential oncogenes in T cell acute lymphoblastic leukemia. Cancer Cell 19, 484-497.

Hu, P. S., Xia, Q. S., Wu, F., Li, D. K., Qi, Y. J., Hu, Y., Wei, Z. Z., Li, S. S., Tian, N. Y., Wei, Q. F., et al. (2017). NSPc1 promotes cancer stem cell self-renewal by repressing the synthesis of all-trans retinoic acid via targeting RDH16 in malignant glioma. Oncogene 36, 4706-4718.

Johannessen, C. M., Boehm, J. S., Kim, S. Y., Thomas, S. R., Wardwell, L., Johnson, L. A., Emery, C. M., Stransky, N., Cogdill, A. P., Barretina, J., et al. (2010). COT drives resistance to RAF inhibition through MAP kinase pathway reactivation. Nature 468, 968-972.

Kadota, M., Yang, H. H., Gomez, B., Sato, M., Clifford, R. J., Meerzaman, D., Dunn, B. K., Wakefield, L. M., and Lee, M. P. (2010). Delineating genetic alterations for tumor progression in the MCF10A series of breast cancer cell lines. PLoS One 5, e9201.

Kitamura, T., Takagi, S., Naganuma, T., and Kihara, A. (2015). Mouse aldehyde dehydrogenase ALDH3B2 is localized to lipid droplets via two C-terminal tryptophan residues and lipid modification. Biochem J 465, 79-87.

Klinakis, A., Lobry, C., Abdel-Wahab, O., Oh, P., Haeno, H., Buonamici, S., van De Walle, I., Cathelin, S., Trimarchi, T., Araldi, E., et al. (2011). A novel tumour-suppressor function for the Notch pathway in myeloid leukaemia. Nature 473, 230-233.

Li, J., Sordella, R., and Powers, S. (2016). Effectors and potential targets selectively upregulated in human KRAS-mutant lung adenocarcinomas. Scientific Reports 6.

Liau, S. S., Jazag, A., and Whang, E. E. (2006). HMGA1 is a determinant of cellular invasiveness and in vivo metastatic potential in pancreatic adenocarcinoma. Cancer Res 66, 11613-11622.

Liu, Y., Sun, J., and Zhao, M. (2017). ONGene: A literature-based database for human oncogenes. J Genet Genomics 44, 119-121.

Livak, K. J., and Schmittgen, T. D. (2001). Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25, 402-408.

Logsdon, C. D., Simeone, D. M., Binkley, C., Arumugam, T., Greenson, J. K., Giordano, T. J., Misek, D. E., Kuick, R., and Hanash, S. (2003). Molecular profiling of pancreatic adenocarcinoma and chronic pancreatitis identifies multiple genes differentially regulated in pancreatic cancer. Cancer Res 63, 2649-2657.

Lu, X., Pan, X., Wu, C. J., Zhao, D., Feng, S., Zang, Y., Lee, R., Khadka, S., Amin, S. B., Jin, E. J., et al. (2018). An In Vivo Screen Identifies PYGO2 as a Driver for Metastatic Prostate Cancer. Cancer Res 78, 3823-3833.

Marchitti, S. A., Brocker, C., Orlicky, D. J., and Vasiliou, V. (2010). Molecular characterization, expression analysis, and role of ALDH3B1 in the cellular protection against oxidative stress. Free Radic Biol Med 49, 1432-1443.

Meacham, C. E., Lawton, L. N., Soto-Feliciano, Y. M., Pritchard, J. R., Joughin, B. A., Ehrenberger, T., Fenouille, N., Zuber, J., Williams, R. T., Young, R. A., and Hemann, M. T. (2015). A genome-scale in vivo loss-of-function screen identifies Phf6 as a lineage-specific regulator of leukemia cell growth. Genes Dev 29, 483-488.

Nagel, S., Kaufmann, M., Drexler, H. G., and MacLeod, R. A. (2003). The cardiac homeobox gene NKX2-5 is deregulated by juxtaposition with BCL11B in pediatric T-ALL cell lines via a novel t(5;14)(q35.1;q32.2). Cancer Res 63, 5329-5334.

Pavlova, N. N., Pallasch, C., Elia, A. E., Braun, C. J., Westbrook, T. F., Hemann, M., and Elledge, S. J. (2013). A role for PVRL4-driven cell-cell interactions in tumorigenesis. Elife 2, e00358.

Pires, M. M., Hopkins, B. D., Saal, L. H., and Parsons, R. E. (2013). Alterations of EGFR, p53 and PTEN that mimic changes found in basal-like breast cancer promote transformation of human mammary epithelial cells. Cancer Biol Ther 14, 246-253.

Powell, E., Shao, J., Picon, H. M., Bristow, C., Ge, Z., Peoples, M., Robinson, F., Jeter-Jones, S. L., Schlosberg, C., Grzeskowiak, C. L., et al. (2018). A functional genomic screen in vivo identifies CEACAM5 as a clinically relevant driver of breast cancer metastasis. NPJ Breast Cancer 4, 9.

Pratilas, C. A., Taylor, B. S., Ye, Q., Viale, A., Sander, C., Solit, D. B., and Rosen, N. (2009). V600EBRAF is associated with disabled feedback inhibition of RAF–MEK signaling and elevated transcriptional output of the pathway. Proceedings of the National Academy of Sciences 106, 4519-4524.

Qu, Y., Han, B., Yu, Y., Yao, W., Bose, S., Karlan, B. Y., Giuliano, A. E., and Cui, X. (2015). Evaluation of MCF10A as a Reliable Model for Normal Human Mammary Epithelial Cells. PLoS One 10, e0131285.

Romagosa, C., Simonetti, S., Lopez-Vicente, L., Mazo, A., Lleonart, M., Castellvi, J., and y Cajal, S. R. (2011). p16Ink4a overexpression in cancer: a tumor suppressor gene associated with senescence and high-grade tumors. Oncogene 30, 2087-2097.

Rotem, A., Janzer, A., Izar, B., Ji, Z., Doench, J. G., Garraway, L. A., and Struhl, K. (2015). Alternative to the soft-agar assay that permits high-throughput drug and genetic screens for cellular transformation. Proc Natl Acad Sci U S A 112, 5708-5713.

Salomon, D. S., Perroteau, I., Kidwell, W. R., Tam, J., and Derynck, R. (1987). Loss of growth responsiveness to epidermal growth factor and enhanced production of alpha-transforming growth factors in ras-transformed mouse mammary epithelial cells. J Cell Physiol 130, 397-409.

Sawey, E. T., Chanrion, M., Cai, C., Wu, G., Zhang, J., Zender, L., Zhao, A., Busuttil, R. W., Yee, H., and Stein, L. (2011). Identification of a therapeutic strategy targeting amplified FGF19 in liver cancer by Oncogenomic screening. Cancer cell 19, 347-358.

Scott, K. L., Nogueira, C., Heffernan, T. P., van Doorn, R., Dhakal, S., Hanna, J. A., Min, C., Jaskelioff, M., Xiao, Y., Wu, C. J., et al. (2011). Proinvasion metastasis drivers in early-stage melanoma are oncogenes. Cancer Cell 20, 92-103.

Shrestha, Y., Schafer, E. J., Boehm, J. S., Thomas, S. R., He, F., Du, J., Wang, S., Barretina, J., Weir, B. A., Zhao, J. J., et al. (2012). PAK1 is a breast cancer oncogene that coordinately activates MAPK and MET signaling. Oncogene 31, 3397-3408.

Soule, H. D., Maloney, T. M., Wolman, S. R., Peterson, W. D., Jr., Brenz, R., McGrath, C. M., Russo, J., Pauley, R. J., Jones, R. F., and Brooks, S. C. (1990). Isolation and characterization of a spontaneously immortalized human breast epithelial cell line, MCF-10. Cancer Res 50, 6075-6086.

Termen, S., Tan, E. J., Heldin, C. H., and Moustakas, A. (2013). p53 regulates epithelial-mesenchymal transition induced by transforming growth factor beta. J Cell Physiol 228, 801-813.

Tsang, Y. H., Dogruluk, T., Tedeschi, P. M., Wardwell-Ozgo, J., Lu, H., Espitia, M., Nair, N., Minelli, R., Chong, Z., and Chen, F. (2016). Functional annotation of rare gene aberration drivers of pancreatic cancer. Nature communications 7.

Wehde, B. L., Radler, P. D., Shrestha, H., Johnson, S. J., Triplett, A. A., and Wagner, K. U. (2018). Janus Kinase 1 Plays a Critical Role in Mammary Cancer Progression. Cell Rep 25, 2192-2207 e2195.

Welsh, J. B., Sapinoso, L. M., Su, A. I., Kern, S. G., Wang-Rodriguez, J., Moskaluk, C. A., Frierson, H. F., Jr., and Hampton, G. M. (2001). Analysis of gene expression identifies candidate markers and pharmacological targets in prostate cancer. Cancer Res 61, 5974-5978.

Xu, L., Cochran, D. M., Tong, R. T., Winkler, F., Kashiwagi, S., Jain, R. K., and Fukumura, D. (2006). Placenta growth factor overexpression inhibits tumor growth, angiogenesis, and metastasis by depleting vascular endothelial growth factor homodimers in orthotopic mouse models. Cancer Res 66, 3971-3977.

Zhao, X., Li, J., Liu, Z., and Powers, S. (2021). Combinatorial CRISPR/Cas9 Screening Reveals Epistatic Networks of Interacting Tumor Suppressor Genes and Therapeutic Targets in Human Breast Cancer. Cancer Res 81, 6090-6105.

Zoppoli, G., Regairaz, M., Leo, E., Reinhold, W. C., Varma, S., Ballestrero, A., Doroshow, J. H., and Pommier, Y. (2012). Putative DNA/RNA helicase Schlafen-11 (SLFN11) sensitizes cancer cells to DNA-damaging agents. Proc Natl Acad Sci U S A 109, 15030-15035.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Identification of breast cancer targets by functional genetic screening of overexpressed genes for tumor-promoting function and tumor cell dependency

Status:

Version 1

Abstract

Figures

Introduction

Results

Selection of overexpressed genes and transfection of LentiORFs into both MCF10A and NMuMG

Overexpressed genes vary widely in their effects on tumor formation

Tumor formation is more closely related to 3D cell culture growth than 2D

Discussion

Methods

Declarations

Acknowledgements

References

Additional Declarations

Supplementary Files

Status:

Version 1