Sustained Adrenergic Activation of YAP1 Induces Anoikis Resistance in Cervical Cancer Cells.

Chronic stress-related hormones modulate tumor pathogenesis at multiple levels; however, the molecular pathways involved in stress and cervical cancer progression are not well understood. We established a preclinical orthotopic mouse model of cervical cancer and used the model to show that daily restraint stress increased tumor growth and metastatic tumor burden. Exposure to norepinephrine significantly protected cervical cancer cells from anoikis. We demonstrated that YAP1 was dephosphorylated and translocated from the cytoplasm to the nucleus by norepinephrine, a process initiated by ADRB2/cAMP/protein kinase A activation. Furthermore, anoikis resistance and YAP1 activation induced by norepinephrine could be rescued by a broad β-adrenergic receptor antagonist, propranolol. Collectively, our results provide a pivotal molecular pathway for disrupting pro-tumor neuroendocrine signaling in cervical cancer.

Introduction

A substantial number of studies have established an important role for biobehavioral stress factors in cancer progression (Antoni and Dhabhar, 2019; Lu et al., 2019; Umamaheswaran et al., 2018). Environmental and psycho-social processes initiate a cascade of signaling in both the central and peripheral nervous systems, which trigger fight-or-flight stress responses and release of stress-related mediators including catecholamines (norepinephrine [NE] and epinephrine), cortisol, and other neurotransmitters via the hypothalamic-pituitary-adrenocortical (HPA) axis or the sympathetic nervous system (SNS) (Antoni et al., 2006). Previously, we demonstrated that SNS activation can promote tumor metastasis-related pathways (Cole et al., 2015) in ovarian cancer. Although research has shown that stress hormones affect tumor pathogenesis at multiple levels, our understanding of the underlying mechanisms is in its infancy and needs to be expanded.

Cervical cancer is the fourth most common malignancy in women worldwide, with an estimated 570,000 new cases in 2018, representing 6.6% of all cancers in women (Bray et al., 2018). Human papillomavirus (HPV) infection is a well-recognized causative factor for cervical cancer (Burd, 2003). It has been reported that severely stressful life events were associated with a 62% increased risk of HPV16 infection, high viral load, and recurrent infection (Lu et al., 2016); psychologic distress had an increased risk of cervical cancer-specific mortality (Lu et al., 2019). Chronic stress is thought to suppress protective immunity that is critical for eliminating immunogenic cancers (e.g., squamous cell and basal cell carcinomas) and virally associated cancers (e.g., HPV-associated cervical, anal, and oral cancers) (Antoni and Dhabhar, 2019). At present, the molecular pathways by which chronic stress affects cervical cancer progression are not well understood.

In this study, we established a preclinical orthotopic mouse model of cervical cancer to determine the physiologic effects of chronic stress in vivo and identified YAP1 activation as a potential stress effector involved in anoikis resistance, which promotes cervical cancer progression.

Results

Chronic Stress Promotes Cervical Cancer Growth In Vivo

Given the lack of established models of cervical cancer for assessing the effects of chronic stress, we sought to establish such a model. First, to identify cell lines that could be affected by stress hormones, we tested multiple cervical (SiHa, CaSki, ME-180, C33A) and ovarian (SKOV3 and A2780) cancer cell lines for β-adrenergic receptors. Most of these cell lines expressed varying levels of β-adrenergic receptors, but the C33A and A2780 cell lines lacked β2-adrenergic receptor (ADRB2) expression (Figure S1).

To evaluate the potential role of chronic stress on cervical cancer in vivo, a restraint-stress orthotopic model of cervical cancer was established and characterized. Cervical cancer cells were injected directly into the cervix at the cervical-uterine junction. We found that SiHa and ME-180, but not CaSki cells, had tumorigenic potential. Three days later, nude mice were placed in a movement-restricted space for 2 h daily to mimic chronic stress (Thaker et al., 2006). The mice were subjected to daily restraint stress for 3 weeks, then euthanized 1 week after stress was ceased. Tumor growth was monitored weekly by IVIS bioluminescence imaging for the duration of the experiment. Bioluminescence imaging showed tumor localization at the cervix within the pelvic region (Figure 1A). Further necropsy showed lymph-vascular invasion of tumors and spread into the parametrium and pelvic wall (Figure 1B). Representative hematoxylin and eosin (H&E) staining of the tumor tissues is shown in Figure S2. This model resembles the disease observed in the clinic; thus, it was adopted by our laboratory for further studies. Animals subjected to daily restraint stress had significantly greater tumor weight (p < 0.05) and higher metastatic tumor burden (p < 0.05) than non-stressed control mice that received the same injection of SiHa cells (Figures 1C and 1D). Of note, the broad β-antagonist propranolol inhibited stress-induced increases in both tumor weight (p < 0.01) and metastatic burden (p < 0.01).

Figure 1

Stress Triggers Cervical Tumor Progression and Induces Specific Protein Expression Pattern in Cervical Cancer Cells

(A–D) (A) Representative images of tumor growth in nude mice over the course of 5 weeks obtained by a Xenogen in vivo imaging system (IVIS) and (B) established tumors in the mice at necropsy. Effects of restraint stress and propranolol (Prop) on (C) cervical tumor weight and (D) metastatic nodule formation. n = 10 per group.

(E and F) (E) Heatmap and (F) volcano plot of 52 differentially expressed proteins by reverse phase protein array (RPPA) analysis in SiHa cells treated with 10 μM NE for 12 h compared with no treatment.

(G) Gene Ontology analysis for functional enrichment. Data are presented as mean ± standard deviation (SD). The raw data for the RPPA are shown in Data S1. Differences between treatment groups were determined by orthogonal contrasts and denoted as follows: ∗p < 0.05, and ∗∗p < 0.01.

NE Induces Anoikis Resistance Mediated by Decreased YAP1 Phosphorylation and NF2

To identify potential mechanisms of NE action during cervical cancer progression, we performed reverse phase protein array (RPPA) analysis in SiHa cells treated with or without 10 μM NE for 12 h. As shown in the heatmap (Figure 1E) and volcano plot (Figure 1F), there were 26 upregulated and 26 downregulated proteins in the NE-treated group.

Then, we performed Gene Ontology (GO) analysis for functional enrichment. Among the most prominent GO functions in the differentially expressed genes in the NE treatment group, regulation of apoptosis process was significantly enriched (Figure 1G). Apoptosis can be induced by numerous triggers, including the loss of cell anchorage, or anoikis. Resistance to anoikis is a hallmark of metastasis, affording tumor cells longer survival in the absence of matrix attachment and facilitating migration, reattachment, and colonization of secondary sites (Sood and Lutgendorf, 2011).

To validate whether NE could induce anoikis resistance, we cultured CaSki, ME-180, and C33A cervical cancer cells in ultra-low attachment plates, which allow for anchorage-independent growth. When treated with 10 μM NE for 72 h, CaSki (p < 0.05) and ME-180 (p < 0.0001) cells, which are ADRB2 positive, were significantly protected from anoikis, but the ADRB-negative C33A cells were not (p > 0.05) (Figure 2A). Similarly, SiHa, CaSki, and ME-180 cells had significantly greater migration and invasion when treated with NE, but C33A cells did not (Figure S3). Furthermore, immunohistochemical analysis of apoptosis in tumor sections revealed that cleaved caspase-3-positive cells in mice subjected to 21 days of daily restraint stress as described above were reduced compared with those among controls (p < 0.01; Figures 2B and 2C). These effects were attenuated in the group that was both stressed and treated with propranolol (p < 0.01; Figures 2B and 2C).

Figure 2

Norepinephrine (NE) Induces Anoikis Resistance and Putative Pathways in Cervical Cancer Cells

(A) Number of dead (SYTOX Red-positive, black) and living (SYTOX Red-negative, white) CaSki, ME-180, and C33A cells after 72 h of low attachment and/or co-incubation with 10 μM NE.

(B and C) (B) Representative immunohistochemistry and (C) quantification staining of apoptosis protein marker in mouse cervical tumors in various conditions. n = 6 mice per group. Scale bar is 200 μm.

(D) Kyoto Encyclopedia of Genes and Genomes (http://www.genome.ad.jp/kegg/) enriched putative pathways for 52 differentially expressed proteins by RPPA. Data are expressed as number of cleaved caspase-3-positive cells per high-power field. Experiments were repeated in triplicate. Data are presented as mean ± standard deviation (SD). Differences between treatment groups were determined by orthogonal contrasts and denoted as follows: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; ns, no significance.

To uncover potential pathways involved in NE-induced anoikis in cervical cancer, additional computational analysis of differentially expressed genes using the Kyoto Encyclopedia of Genes and Genomes (KEGG) showed the top ten enriched pathways (Figure 2D): seven pathways in signal transduction, two in human diseases (EGFR tyrosine kinase inhibitor resistance and endocrine resistance), and one in cellular processes (P53 signaling pathway). Among these, the Hippo signaling pathway, which has been reported to contribute to anoikis evasion (Zhao et al., 2012), has been identified as a candidate for NE-induced signaling. Its well-known downstream effector YAP1 (Howe and Juliano, 2000; Misra and Irvine, 2018) and upstream regulator NF2 (Reginensi et al., 2016) were both indicated in the enriched KEGG network (Figure 2D).

Our reverse phase protein array data indicated that phosphorylation level of YAP1S127 and total NF2 protein were significantly decreased by 82.60% and 90.42%, respectively, in NE-treated SiHa cells (Figures 1E and 1F). Consistent with this finding, when CaSki and ME-180 cells were exposed to 10 μM NE under both normal and low-attachment conditions, phosphorylated YAP1S127 and NF2 were significantly downregulated in CaSki and ME-180 cells (p < 0.01; Figures 3A and 3B) but not in C33A cells (ADRB2-negative; p > 0.05; Figure 3C).

Figure 3

NE Decreases YAP1 Phosphorylation and NF2 Expression in CaSki and ME-180 Cells

Western blot analysis of phosphorylated YAP1S127, total YAP1, and NF2 in CaSki (A), ME-180 (B), and C33A (C) cultured in normal plates (NP) or ultra-low attachment plates (LP) with or without 10 μM NE co-incubation. β-Actin was used as a loading control. The immunoblots are on the left, and quantifications of band intensity relative to β-actin are on the right (n = 3, data represent the mean ± SD). Differences between treatment groups were determined by orthogonal contrasts and denoted as follows: ∗p < 0.05, ∗∗p < 0.01; ns, no significance.

YAP1 Activation Is Required for NE-Induced Anoikis Resistance

Phosphorylation of YAP1S127 generates a 14-3-3-binding motif responsible for YAP1 cytoplasmic retention (Zhao et al., 2007); if dephosphorylated, activated YAP1 can translocate into the nucleus and promote transcription of genes that in turn inhibit apoptosis (Kapoor et al., 2014; Yu and Guan, 2013; Yu et al., 2012a). We used immunofluorescence to visualize the intracellular localization of YAP1 in CaSki and ME-180 cells treated with or without 10 μM NE for 2 h and found a clear shift in YAP1 expression from the cytoplasm to the nucleus after NE treatment (Figures 4A and 4B). Consistently, nuclear YAP1 protein expression was significantly higher in mice exposed to restraint stress than those without restraint stress, and propranolol treatment completely abrogated this effect (Figures 4C and 4D). To determine whether similar findings are noted in human samples, we obtained cervical cancer samples from eleven patients. Levels of depressive scores were measured during the pre-surgical clinic visit 1 to 7 days prior to tumor resection. Based on the established threshold of CESD ≥16, six participants were determined to have high levels of biobehavioral risk factors, whereas five were low risk. Nuclear staining of YAP1 was significantly higher in patients with cervical cancer with high CESD scores than those without (p < 0.05; Figures 4E and 4F).

Figure 4

YAP1 Is Activated by NE and Is Indispensable for NE-Induced Anoikis Resistance

(A–J) (A) Representative immunofluorescence staining and (B) quantification of YAP1 in CaSki and ME-180 cells after 2 h under low-attachment conditions with (bottom) or without (top) 10 μM NE treatment. Representative immunohistochemical staining and quantification of YAP1 in cervical cancer tissues from orthotopic mice (C and D, n = 6) and patients (E and F) (×400 magnification). Western blot analysis (G, left part) in CaSki and ME-180 (H, left part) cells showing knockdown efficiency of YAP1 at the protein level using two siRNAs. (Right part of G and H) Quantifications of band intensity relative to β-actin. Bar graphs showing number of dead (STYOX Red-positive, black) and living (SYTOX Red-negative, white) CaSki (I) and ME-180 (J) cells after 72 h of low attachment. Bars and error bars represent mean ± SD.∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001. CTRL, control. Arrow：YAP1 expression in the nucleus；arrow head: YAP1 expression in the cytoplasm.

Next, we used two siRNAs to knock down YAP1 gene expression in CaSki cells (p < 0.01 and p < 0.05; respectively; Figure 4G) and ME-180 cells (p < 0.05 and p < 0.01, respectively; Figure 4H). NE 10 μM for 72 h inhibited anoikis in cervical cancer cells transfected with control siRNAs, whereas YAP1 siRNAs abrogated this NE-induced anoikis resistance in both CaSki (Figure 4I) and ME-180 (Figure 4J) cells under anchorage-independent conditions. In addition, when expression of NF2 was downregulated by two siRNAs (Figures S4A and S4B), phosphorylated YAP1S127 was decreased (Figures S4C and S4D), and NE-induced anoikis resistance was completely reversed (p < 0.05; Figures S4E and S4F).

NE Regulates Hippo-YAP1 Pathway via ADRB2-Mediated Signaling

We next delineated the signaling pathway involved in NE-mediated Hippo-YAP1 activation. Following NE treatment for 1–3 h, in addition to expression changes in NF2 and pYAP1S127, two core kinases in the Hippo pathway, mammalian Ste20-like kinase 1 (MST1) and large tumor suppressor kinase 1 (LATS1), were both dephosphorylated and inactivated (Figure 5A). When cells were pretreated with 10 μM propranolol, 10 μM atenolol (ADRB1 antagonist), or 10 μM ICI-118,551 (ADRB2 antagonist) for 1 h before NE exposure, NE-induced anoikis resistance was abrogated completely by propranolol and ICI-118,551 (Figures 5B and 5C). Furthermore, treatment with ICI-118,551 or propranolol resulted in abrogation of the NE-mediated decrease in YAP1 phosphorylation (Figure 5D); ADRB2 silencing by siRNAs in CaSki and ME-180 cells (Figures S5A and S5B) resulted in abrogation of NE-mediated YAP1 dephosphorylation (Figures S5C–S5E). When ADRB2 was overexpressed using the p6596 ADRB2 plasmid in C33A cells, YAP1 was dephosphorylated by NE (Figure S6).

Figure 5

NE Regulates Hippo-YAP1 Pathway in CaSki and ME-180 Cells

(A–C) (A) Western blot analysis of YAP1, NF2, and two core kinases in the Hippo pathway were analyzed in CaSki and ME-180 cells with or without 10 μM NE co-incubation. β-Actin was used as a loading control. The immunoblots are on the left, and quantifications of band intensity relative to β-actin are on the right. Effects of atenolol (Ate; ADRB1 antagonist), ICI-118,551 (ICI; ADRB2 antagonist), or propranolol (Prop; non-specific ADRB antagonist) on anoikis in CaSki (B) and ME-180 (C) cells treated with 10 μM NE for 72 h.

(D) Effects of ICI-118,551 or propranolol on phosphorylated YAP1S127 and total YAP1 in CaSki and ME-180 cells treated with 10 μM NE for 3 h. The immunoblot is at the top, and quantification of band intensity relative to β-actin is below. Data represent the mean ± SD. ∗p < 0.05, and ∗∗p < 0.01. CTRL, control.

Because cAMP is an important component of the ADRB2 signaling pathway, we examined intracellular cAMP levels after NE treatment. Relative to controls, treatment with 10 μM NE for 30 min increased cAMP levels by 2.92-fold in CaSki (p < 0.05), 3.32-fold in SiHa (p < 0.05), and 3.97-fold in ME-180 (p < 0.05) cells, but no change was noted in C33A cells (p > 0.05) (ADRB2-negative; Figure 6A). Treatment with 10 μM forskolin (cAMP activator) induced similar anoikis resistance in CaSki (p < 0.05) and ME-180 (p < 0.05) cells (Figure 6B). Exposure to 10 μM forskolin for various durations decreased pYAP1S127 expression, much like NE treatment did (Figure 6C). Protein kinase A (PKA) is an important protein downstream of cAMP, and inhibition of PKA (using 10 μM H-89) markedly protected against NE-induced anoikis resistance (Figures 6D and 6E).

Figure 6

cAMP and Protein Kinase A (PKA) Mediate NE-Induced Anoikis Resistance and YAP1 Activation

(A) Baseline and NE-induced levels of intracellular cyclic adenosine monophosphate (cAMP) in cervical cancer cells measured by enzyme-linked immunosorbent assay (n = 3).

(B) Effects of 10 μM forskolin (cAMP activator) on anoikis in CaSki and ME-180 cells.

(C–E) (C) Expression pattern of pYAP1S127 and total YAP1 at different time points on western blot analysis (n = 2). The immunoblot is at the top, and quantification of band intensity relative to β-actin is below. Effect of H-89 (PKA antagonist, 10 μM or 1 μM incubated for 1 h before NE exposure) on anoikis in CaSki (D) and ME-180 (E) cells.

∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ns, no significance.

Discussion

The key finding of this study is that chronic stress protects cervical cancer cells from anoikis and this protection promotes malignant tumor progression. Here, we developed a preclinical orthotopic mouse model of cervical cancer to provide a new understanding of the effects of chronic stress on cervical cancer growth and metastasis. HPV-16 is the most common HPV genotype in cervical cancer; ME-180 cells are HPV-68 positive; the oncogenic ability is lower than HPV-16, so ME-180 was not selected for orthotopic model. Even though Caski cell and SiHa cells both are HPV-16 positive, we found that CaSki did not have orthotopic tumorigenic potential in the mouse cervix, partially attributed to the fact that CaSki cells are derived from a metastatic site (small intestine). Thus, SiHa cells (HPV16 positive) were used for the orthotopic mouse model. Our data obtained using this model indicate that the neuroendocrine stress response by NE directly induces anoikis resistance through a signaling pathway mediated by Hippo-YAP1 that is initiated by ADRB2/cAMP/PKA activation. Anoikis is a process by which normal cells undergo apoptosis when detached from the surrounding extracellular matrix. Avoidance of anoikis is an essential prerequisite for tumor metastasis; it provides a selective advantage that allows metastatic cancer cells to transit to new sites for attachment (Frisch and Screaton, 2001). Previously, we demonstrated an important role for YAP1 signaling in blocking anoikis (Haemmerle et al., 2017). Similarly, other groups have demonstrated Hippo pathway-mediated anoikis inhibition in hepatocellular carcinoma (Cheng et al., 2018); YAP/TAZ activation by ASPP1 leads to anoikis resistance (Vigneron et al., 2010). In the present study, we implicate sustained adrenergic stimulation in inducing cervical cancer anoikis resistance by inhibition of the tumor suppressive Hippo-YAP1 pathway.

The Hippo-YAP1 pathway plays an important role in regulating cell proliferation, death, and differentiation (Yu and Guan, 2013). The tight control of this pathway and its cross talk with other signaling pathways is critical for tumorigenesis and cancer progression (Yu et al., 2013). Several modulators of the Hippo-YAP1/TAZ pathway have been identified via extensive genetic and biochemical analyses (Yu and Guan, 2013; Zhao et al., 2010, 2011); the Hippo-YAP1/TAZ pathway is robustly regulated by a wide range of signals and their corresponding G protein-coupled receptors (Yu et al., 2012b). Here, we found that ADRB2, a classic G protein-coupled receptor, mediates NE-induced YAP1 activation in cervical carcinogenesis, which indicates that neuroendocrine stress signaling takes part in non-canonical regulation of the Hippo-YAP1 pathway.

Activation of ADRB2 by epinephrine, NE, or specific agonists typically results in Gs-dependent activation of adenylate cyclase and a subsequent increase in intracellular cAMP (Tan et al., 2007). This increase in cAMP stimulates PKA to phosphorylate multiple target proteins, including transcription factors of the CREB/ATF and GATA families (Cole and Sood, 2012; Rockman et al., 2002). Similarly, we found that intracellular cAMP levels in cervical cancer cells were increased by NE stimulation and NE-induced effects could be mimicked by cAMP activation and blocked by a PKA antagonist. Neurofibromin 2 (NF2, also known as merlin), a tumor suppressor and an upstream component of the Hippo pathway, is a direct target of PKA (Alfthan et al., 2004). Our study identified a new functional role for NF2 as a key molecular effector that links adrenergic signaling to the downstream Hippo-YAP1 pathway and tumor progression.

Considering the role of stress-induced suppression of protective immune responses in infection-related cancer (Antoni and Dhabhar, 2019), HPV-driven cervical cancer is potentially much more sensitive to effects of chronic stress than other cancers. Lu et al. found that major life events, including bereavement, severe illness of a family member, divorce, and being between jobs, were prevalent among patients with cervical cancer (37.4%) (Lu et al., 2019). We previously reported that intra-tumoral NE levels in primary ovarian carcinomas are linked to both disease severity and patient psychosocial characteristics (Lutgendorf et al., 2009), but very little data about NE levels and β-adrenergic signaling in cervical cancer have been reported to date. In our study, chronic stress induced YAP1 nuclear translocation and anoikis resistance in cervical cancer, and these effects were attenuated by propranolol. These data indicate that pharmacologic inhibition of adrenergic receptors may have therapeutic relevance in cervical cancer. Emerging clinical data also link the use of non-selective β-adrenergic receptor blockers with reduced cancer progression (Barron et al., 2011; De Giorgi et al., 2011; Melhem-Bertrandt et al., 2011). Of note, results of retrospective studies are prone to immortal time bias and need to be further validated in prospective studies (Weberpals et al., 2016).

In summary, our data represent a new understanding of YAP1 activation in response to sustained adrenergic signaling in cervical cancer models (Figure 7). Protective interventions targeting the neuroendocrine system may provide a biologically plausible method to prevent cervical cancer progression.

Figure 7

Working Model of NE-Induced YAP1 Activation and Anoikis Resistance in Cervical Cancer Cells

Limitations of the Study

Although our results demonstrate that norepinephrine induces anoikis resistance in cervical cancer, it is possible that other pathways (e.g., cortisol) could have similar or broader effects. To what extent these stress-related hormones are important in clinical context would require further work. Moreover, in the absence of available syngeneic mouse models, we used cross-species cell line-derived tumor xenograft (CDX) mouse models in this study to understand the role of stress in mediating cancer cell-driven mechanisms of cervical cancer progression. However, it is possible that the tumor microenvironment may also play an important role; immune-competent models would be required for such work.

Resource Availability

Lead Contact

Anil K. Sood, MD Department of Gynecologic Oncology and Reproductive Medicine, The University of Texas MD Anderson Cancer Center, 1155 Herman Pressler, Houston, TX 77030, Tel 713-745-5266, Fax 713-792-7586 (Email: asood@mdanderson.org).

Materials Availability

No new materials were generated in this study.

Data and Code Availability

The raw data that support the findings of this study are available from the corresponding authors, upon request.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.