The majority of breast cancers are of the luminal subtypes (A and B). These subtypes tend to overexpress the estrogen receptor α (ERα), thus these cancers' growth and expansion are driven by the hormone estrogen. This renders most of these ER+ tumors sensitive to therapy with selective estrogen receptor modulators (SERMs) such as tamoxifen. However, over time, relapses can occur in these patients that are often resistant to this previously effective therapy, frequently leading to death.
To investigate the source of this resistance, Matthew Ellis, MD, PhD, FRCP, Professor and Director, Lester & Sue Smith Breast Center, Baylor College of Medicine, initiated a retrospective study that analyzed the genomes of tumors from more than 600 patients who had received adjuvant tamoxifen monotherapy. The median follow-up period for participants in that study was 10.4 years. The data they obtained showed that loss of function mutations in the NF1 gene was strongly associated with poorer patient outcomes.
“The inactivating NF1 mutations we observed in the tumor cells were specific to those cells; they were not germline mutations from the patients' genomes,” explained Ellis. “We sought to develop an immunohistochemical (IHC) protocol to detect the presence of neurofibromin 1, the protein produced by the NF1 gene, to assess the impact of that protein's loss in ER+ breast cancer patients' outcomes.”
The NF1 gene, which is most commonly associated with the inherited autosomal dominant disorder neurofibromatosis type 1 (von Recklinghausen disease), encodes neurofibromin 1, a protein that serves as a tumor suppressor. This protein acts as a negative regulator of the RAS oncogenes by activating the RAS GTPase function, affording the inactivated GDP-bound form. Genome resequencing studies performed for different cancers have shown that NF1 has frequently undergone deletion or intragenic mutation, thus implicating it as a potential driver in these disease states. NF1 mutations or deletions have already been noted in the following cancers: lung adenocarcinoma, ovarian carcinoma, and glioblastoma multiforme.
The NF1 Gene
In 2012, Schimenti, et al, performed an oncogenomic study that implicated the NF1 gene as a driver in breast cancer (Genetics 2012;192,385-396). Until that point, intragenic mutations and loss of heterozygosity had been noted for the NF1 gene in breast cancer cases. Women with neurofibromatosis type 1 (i.e., those who have an inherited mutation or loss of gene from their germline cells) have a higher association with or risk for breast cancer. When these researchers screened the breast cancer patient data available at the time from The Cancer Genome Atlas, they found 27.7 percent of the patients had NF1 deletions or mutations, and of these, the majority was heterozygous. Additionally, more than 40 percent of the HER2-overexpressed or basal tumor subtypes showed NF1 mutation or loss.
In their study, Schimenti, et al utilized the C3H-Mcm4Chaos3/Chaos3 mouse tumor model (hereafter abbreviated Chaos3) to examine potential drivers because:
- tumors can arise efficiently from this model due to genomic instability induced by a point mutation in the minichromosome maintenance 4 (Mcm4) gene; and
- the mammary tumors in this model were shown via tumor differentiation score analysis to most closely resemble mature human luminal cells, a rare but very important property among mouse models for the study of the most frequent form of human breast cancer.
By using this particular model, the investigators hoped the controlled genetic makeup and single tumor etiology might make it easier to identify recurring mutation events that are likely to be tumorigenesis drivers.
This line of mice has thus found to have more than 80 percent of the nulliparous females develop mammary-based adenocarcinomas exclusively. The Chaos3 tumors isolated in the study had a large number of cells with varying numbers of chromosomes, even within the same tumor (also a trait of human breast cancer tumors). To the researchers' surprise, when partial exomic resequencing of these tumors was done, few somatic point mutations were observed in the targeted exon regions. The calculated mutation rate (~0.25 mutations/Mb) was not higher than the background rates obtained from other breast cancer genomic studies, implying that higher intragenic mutagenesis was not the primary mechanism for Chaos3 tumorigenesis. As a result, the investigators also explored copy number alterations as potential carcinogenesis drivers.
All Chaos3 tumors were analyzed by array comparative genomic hybridization. Of particular interest to the researchers were deletions observed on chromosome 11 of all Chaos3 mammary tumor cells. None of the Chaos3 nonmammary tumors had these same deletions. The genes included in these deletions were NF1, Omg (oligodendrocyte myelin glycoprotein), Ksr1 (kinase suppressor of RAS), and Wsb1 (WD repeat and SOCS Box-containinc Protein). Of this group of genes, the researchers thought NF1 was the most likely candidate as a breast cancer tumor suppressor. Ksr1 is a promoter of oncogenic MAPK and RAS signaling, and thus loss of function would actually inhibit tumor growth, while Omg is required for myelination of the CNS, and consequently is primarily expressed in neuronal tissues.
As previously stated, neurofibromin 1 is a negative regulator of RAS. In turn, RAS is a stimulator for the pro-growth factor mTOR (mechanistic target of rapamycin), thus one might postulate that tumor cells deficient in NF1 may have both higher levels of activated-RAS as well as be sensitive to mTOR inhibitors such as rapamycin. This effect has been observed in acute myelogenous leukemia patients having NF1 deficiency; these patients had tumor cells that displayed both higher levels of activated RAS as well as sensitivity to rapamycin. When Chaos3 mammary tumor cells were analyzed, they too showed exceptionally high levels of activated RAS. Furthermore, Chaos3 mammary tumor cell lines also showed enhanced sensitivity to the mTOR inhibitor rapamycin as well as the MAPK/MEK1 inhibitor PD98059. While this loss of NF1 may suggest a potential therapeutic strategy (targeting of the RAS pathway), there are other implications for NF1, as there are multiple isoforms with unknown functions.
One such potential implication of NF1 inactivation is in the treatment of ER+ patients with the standard therapy of tamoxifen. In a study by Mendes-Pereira, et al, (Proc Natl Acad Sci USA 2011;109:2730-2735), genome-wide functional short hairpin RNA (shRNA) screening was performed with massively-parallel sequencing to map the effects of 56,670 RNA interference agents on 16,487 gene targets. From these data, the researchers showed that the silencing of the following genes induced tamoxifen resistance in MCF7 breast cancer tumor cells: NF1, BAP1, CLPP, GPRC5D, NAE1, NIPBL, NSD1, RAD21, RARG, SMC3, and UBA3. Conversely, sensitivity to tamoxifen therapy was conferred by silencing of the following genes in MCF7 breast cancer tumor cells: C10orf72, C15orf55/NUT, EDF1, ING5, KRAS, NOC3L, PPP1R15B, RRAS2, TMPRSS2, and TPM4.
Finding an IHC Protocol
Ellis, when commenting on this study's findings, noted, “Loss of NF1 was associated with poorer outcomes for those breast cancer patients treated with tamoxifen. “[It] is also associated with treatment resistance in breast cancer patients. It appears that in a NF1 knockout setting, tamoxifen's activity is switched from an ER antagonist to an agonist; this means that as an agonist, the compound could then stimulate breast cancer growth.”
As part of their attempt to gauge the effects of neurofibromin 1 loss, Ellis and his colleagues sought to develop an IHC protocol to detect the presence of the protein. In their initial efforts, commercially-available antibodies were evaluated using cells with reduced amount of neurofibromin as controls, but found that background is a problem. To address this, Ellis and his colleagues attempted to create their own antibody targeting the C-terminus of neurofibromin 1.
When asked about why targeting the C-terminus of neurofibromin, Ellis replied “Because neurofibromin 1 is so large, it is much easier to just check for the C-terminal portion of the protein. If the protein is absent, then clearly the C-terminal portion of that protein will also be absent.” When asked where the project is now, he noted, “The production of both monoclonal and polyclonal antibodies is currently underway.”
Regarding the future of the project, Ellis said, “With an efficient IHC method in place, we will be able to rigorously assess whether neurofibromin 1 loss correlates with poorer patient outcomes; this should then permit us to have a better understanding of the therapeutic roadmap of this disease. We hope that this better understanding will help identify high-risk patients and, ultimately, provide more effective therapies for them.”
Richard Simoneaux is a contributing writer.