Background
The Rho GTPase family of signaling proteins are master regulators of cell shape and cell migration. They do this directly through dynamic regulation of the actin cytoskeleton; however, they also have diverse additional cellular roles that contribute to this, including the control of membrane trafficking, cell polarity and gene expression [
1]. The roles of Rho GTPases in cell migration make them important signaling proteins in cancer. While Rho GTPases are generally not direct targets of mutation in cancer, their signaling pathways are frequently deregulated, promoting the switch to cancer cell invasion and metastasis [
2,
3].
The human Rho GTPase family contains 20 members, of which RhoA, Rac1 and Cdc42 are the best characterized [
4]. These are small, globular proteins whose activity is controlled by binding of GTP, which switches them into their active conformation. The Rho family also contains two atypical members – RhoBTB1 and 2. These are larger, multimodular Rho GTPases that have a conserved N-terminal Rho GTPase domain, but also two copies of the BTB (Broad-Complex, Tramtrack and Bric a brac) domain and a carboxyl terminal BACK (BTB and C-terminal Kelch) domain [
5,
6]. Intriguingly, both genes undergo silencing or mutation in human cancer. Hamaguchi and colleagues identified RhoBTB2 in a representational difference analysis screen for novel tumor suppressor genes in breast cancer, and gave it the alternative name DBC2 (deleted in breast cancer 2). The RhoBTB2/DBC2 gene undergoes homologous deletion in a relatively small number of breast tumor samples; however, RhoBTB2 expression is silenced at high frequency (approximately 50%) in breast and lung tumors [
7]. Subsequent studies have reported the silencing of RhoBTB2 expression in a wide range of human tumors, as well as sporadic point mutations of the RhoBTB2 coding region and promoter [
8‐
11]. RhoBTB1 is 73% identical to RhoBTB2 at the protein level. Far less is known about its cellular functions; however, recent studies have shown that it is also downregulated in human cancers. It is subject to loss of heterozygosity at high frequency in head and neck squamous cell (HNSC) carcinomas [
12] and its expression is silenced in colon cancer through the actions of the microRNA miR-31 [
13].
Unlike the majority of members of the Rho GTPase family, RhoBTB1 and 2 do not regulate the actin cytoskeleton directly [
14]. Many proteins with BTB domains function as transcription regulators [
15] and in our previous studies we showed that this is also the case for RhoBTB2 [
16]. To determine transcription targets of RhoBTB2, we silenced its expression in primary lung epithelial cells and then performed whole-genome microarray analysis of gene expression. This allowed us to identify the chemokine CXCL14 as a target of RhoBTB2 regulation [
16]. CXCL14 expression is downregulated in a high percentage of carcinomas, and especially in HNSC carcinomas where its loss is correlated with poor prognosis. Importantly, we found that loss of RhoBTB2 expression is correlated with loss of CXCL14 expression in HNSC cancer cell lines, and that expression of the chemokine is rescued by re-expression of RhoBTB2 [
16].
CXCL14 was the most significant hit in the RhoBTB2 microarray screen; however, several other genes also showed reduced expression upon RhoBTB2 silencing. One of these was METTL7A, a poorly-characterized methyltransferase enzyme. In this study, we investigate the regulation of the METTL7 enzymes by RhoBTB proteins and uncover a pathway controlling Golgi integrity in mammary epithelial cells.
Methods
Materials
Full details of antibodies, oligonucleotides and plasmids used in this study are given in Additional file
1.
Cell culture and transfection
HeLa, HEK293T, MDA-MB-231, MCF7 and T47D cells were cultured in DMEM containing 10% heat-inactivated fetal bovine serum. HMT-S1 and MCF10A cells were cultured as previously described [
17,
18]. HeLa cells were transfected with plasmids and siRNA oligonucleotides using calcium phosphate [
16].
Real-time PCR
RNA was isolated from cells using the TRIzol extraction method (Invitrogen) and 40 μg of purified RNA used for reverse transcription using Omniscript RTase (Qiagen) for 1 h at 37 °C. cDNAs were then subjected to real-time PCR using DyNAmo Flash SYBR Green (Finnzymes). Amplification was performed using an Opticon 2 thermocycler (MJ Research) and data was analyzed using the comparative Ct method.
Immunofluorescence microscopy
Cells were fixed for 15 min in 4% paraformaldehyde in PBS and then permeabilized in 0.2% Triton X-100 in PBS for 5 min. Cells were then incubated with 0.1% sodium borohydride for 10 min. Primary antibodies were incubated with cells in 1% BSA for 1 h followed by secondary antibodies for 45 min. The cells were stained with 2 μg/ml DAPI for 10 min and mounted over MOWIOL 4–88 (Calbiochem) containing 0.6% 1,4-diazabicyclo-(2.2.2) octane as an anti-photobleaching agent. Confocal microscopy was performed using a Leica TCS-NT confocal laser-scanning microscope with an attached Leica DMRBE upright epifluorescence microscope under a PlanApo x63/1.32 oil-immersion objective. A series of images were taken at 0.5 μm intervals through the Z-plane of the cells and processed to form a projected image.
Analysis of Golgi fragmentation
Fragmentation of the Golgi ribbon was scored blind in cells stained for the Golgi marker giantin. Fifty cells were scored from each condition to give the percentage of cells with a fragmented Golgi. Data from three independent experiments were processed to give the mean.
Re-expression of RhoBTB1
Expression of RhoBTB1 was restored in T47D cells by stable integration of a lentiviral RhoBTB1 construct. mCherry-tagged RhoBTB1 was subcloned into pHR’SIN-cPPT-SEW [
19]. Virus was generated by transfection of HEK293T cells as described [
19]. Briefly, 40 μg of RhoBTB1 vector was mixed with 10 μg of envelope plasmid pMDG, 30 μg of packaging plasmid psPAX2 and 1 μl of 1 mM polyethylenimine (Sigma) in OptiMEM (Invitrogen). This transfection mixture was replaced 4 h later with 15 ml of normal culture medium. The cells were then incubated for 48 h to allow virus production. After the incubation, the medium was removed and centrifuged for 10 min at 2,600 x
g. The supernatant was then passed through a 0.45 μm filter and this filtrate was used as the virus stock. T47D cells were transduced with virus stock by overnight incubation.
Cell migration assays
T47D cells were grown to confluence in chamber slides (Ibidi). The confluent monolayer was scratched with a sterile pipette tip and migration was followed by brightfield imaging at 37 °C for 14 h using a Leica AF6000 live cell imaging workstation and x40 objective. Multipoint revisiting allowed for parallel imaging of samples. Average cell velocity was calculated using ImageJ (NIH). For quantification of Golgi polarization, cells were fixed 1 h after making the scratch and stained for giantin. Cell polarization was assessed by dividing cells into quadrants centered on the nucleus. Cells were judged to have polarized if the majority of their Golgi apparatus resided within the quadrant facing the scratch edge.
Cell invasion assays
Invasion of T47D cells was quantified using BD BioCoat Matrigel Invasion Chambers (BD Biosciences) according to the manufacturer’s protocol. Briefly, T47D cells in serum-free medium were plated at 1 × 105 cells per chamber. Medium containing 10% fetal bovine serum was used as the chemoattractant in the lower chamber. After 24 h, non-invaded cells were removed from the upper chamber using a cotton swab. Cells that invaded through the Matrigel to the bottom of the insert were fixed and stained by incubation with Diff-Quick (BD Biosciences) for 10 min. These stained cells were washed with PBS, air-dried and counted.
Statistical analysis
Analysis of statistical significance was performed using GraphPad Prism software. Comparison of two sample experiments was by paired, two-tailed Student’s t-test. Analysis of multiple samples was made using one-way ANOVA. For comparisons to a control, Dunnett’s post hoc test was used, whereas comparisons between all samples employed Tukey’s post hoc test.
Discussion
RhoBTB proteins are highly unusual members of the RhoGTPase family. Their multidomain structure makes them atypical; however, they are present in a wide range of organisms, including the social amoeba
Dictyostelium discoideum, where they were first identified [
31]. We still know very little about the cellular functions of RhoBTB1 and 2. Most progress has come from investigating the function of their twin BTB domains [
6]. In many proteins, the BTB domain functions to recruit the Cul3 ubiquitin ligase [
32]. RhoBTB2 can bind Cul3 through its first BTB domain [
33,
34]. RhoBTB2 itself is a target of ubiquitylation by this complex, leading to its subsequent degradation by the proteasome [
34]. This raises the possibility of a role for the RhoBTB2/Cul3 complex in regulating the turnover of other cellular proteins, although no targets of RhoBTB2/Cul3 mediated ubiquitylation have yet been identified.
BTB domains are also frequently found in transcriptional regulators, where they have a number of roles, including the recruitment of transcriptional co-repressors [
35]. Previously, we showed that RhoBTB2 controls the expression of the CXCL14 chemokine in epithelial cells [
16]. Here were show that the METTL7 enzymes are targets of transcriptional control by RhoBTB1 and 2. CXCL14 expression requires both RhoBTB1 and 2 [
16], whereas the two RhoBTBs had specific roles in regulating the METTL7 enzymes, with RhoBTB2 regulating expression of METTL7A and RhoBTB1 regulating METTL7B. The RhoBTBs dimerize through their BTB domains, and can form both homo- and heterodimers [
33]. It seems likely these different dimer pairs have different transcriptional targets. This becomes important when considering the effects of the downregulation of RhoBTB1 and 2 in cancer cells, where loss of one isoform could potentially increase homodimer concentration of the other, in addition to the simpler effects of reducing its own concentration.
We show here that RhoBTB1 regulates Golgi integrity. Interestingly, the third RhoBTB family member, RhoBTB3, has also been shown to regulate Golgi function. RhoBTB3 is only weakly-related to RhoBTB1 and 2, but shares a similar domain organization [
6]. RhoBTB3 is a Golgi-localized protein and binds to directly to the secretory pathway regulator Rab9 to mediate trafficking from the endoplasmic reticulum to the Golgi [
36]. RhoBTB3 silencing has been reported to lead to enlargement of the Golgi [
36] and in a later study to cause Golgi fragmentation[
37] The mechanism by which RhoBTB3 affects Golgi integrity has not been reported. We show here that RhoBTB1 controls Golgi integrity through regulation of the expression of METTL7B. The downstream targets of the poorly-characterized METTL7 enzymes are unknown. Both contain an N-terminal transmembrane domain and a C-terminal methyltransferase domain, predicted to lie on the cytoplasmic face. Interestingly, METTL7B undergoes arginine dimethylation [
20], raising the possibility of automethylation. Depletion of either METTL7A or B caused Golgi fragmentation (Fig.
3c) and METTL7A overexpression was able to rescue the effects of depletion of METTL7B (Fig.
3e). This suggests that the two proteins have overlapping functions in regulating Golgi integrity. Silencing of RhoBTB2 did not cause significant Golgi fragmentation (Fig.
2b), despite a reduction in METTL7A expression (Fig.
1a). The effects of RhoBTB2 on METTL7A are less pronounced than those of RhoBTB1 on METTL7B, at least in HeLa cells (Fig.
1). It is possible that RhoBTB2 might affect Golgi integrity in other cell types, perhaps where METTL7A is the more dominant isoform.
In vertebrates, the Golgi apparatus is organized into a compact ribbon structure, typically located next to the nucleus [
38]. This organization depends on the constant action of the motor protein dynein and on the functions of Golgi structural proteins [
29,
38]. Surprisingly, Golgi ribbon structure is not requisite for the functioning of the secretory pathway. In plants, individual Golgi stacks are scattered throughout the cytosol, and in the yeast
Saccharomyces cerevisiae, the Golgi is composed of dispersed, isolated cisternae [
38,
39]. In vertebrate cells, treatments that fragment the Golgi do not affect the rate of protein secretion [
40,
41]. Instead, the organized Golgi ribbon structure seems to be important for vertebrate cell polarization. In polarized epithelial cells, the Golgi is orientated towards the apical surface [
28]. In most migrating vertebrate cells, the Golgi is orientated towards the leading edge [
29,
42]. These observations suggest that a polarized Golgi ribbon may facilitate polarized secretion. In polarized epithelial cells, this would support apical/basolateral polarity, and in migrating cells, it would contribute to directional movement. In keeping with this, recent work supports a role for the Golgi in the polarized delivery of active Cdc42 to the leading edge of migrating cells [
43]. Fragmentation of the Golgi has previously been shown to inhibit the migration of HeLa cells [
27]. Here we find that the loss of expression of RhoBTB1 in T47D breast cancer cells is linked to fragmentation of the Golgi. This can be rescued by re-expression of RhoBTB1; however, this has no effect on migration of these cells in 2D (Fig.
6e). Instead, we found that re-expression of RhoBTB1 in T47D breast cancer cells strongly inhibited their invasive capacity in 3D (Fig.
6f). The initial stages of breast cancer progression involve loss of apical-basolateral polarity and invasion through the basement lamina. We propose that loss of RhoBTB1 early in breast cancer development promotes loss of normal epithelial polarity through reduced METTL7B expression and Golgi fragmentation. We propose that this then contributes to the switch to an invasive phenotype.