The atypical Rho GTPase RhoBTB2 is required for ... - Nature

Oncogene (2008) 27, 6856–6865

& 2008 Macmillan Publishers Limited All rights reserved 0950-9232/08 $32.00 www.nature.com/onc

ORIGINAL ARTICLE

The atypical Rho GTPase RhoBTB2 is required for expression of the chemokine CXCL14 in normal and cancerous epithelial cells CM McKinnon1, KA Lygoe1, L Skelton1, R Mitter2 and H Mellor1 1

Department of Biochemistry, School of Medical Sciences, University of Bristol, Bristol, UK and 2Cancer Research UK, London, UK

The Rho family of small GTPases control cell migration, cell invasion and cell cycle. Many of these processes are perturbed in cancer and several family members show altered expression in a number of tumor types. RhoBTB2/ DBC2 is an atypical member of this family of signaling proteins, containing two BTB domains in addition to its conserved Rho GTPase domain. RhoBTB2 is mutated, deleted or silenced in a large percentage of breast and lung cancers; however, the functional consequences of this loss are unclear. Here we use RNA interference in primary human epithelial cells to mimic the loss of RhoBTB2 seen in cancer cells. Through microarray analysis of global gene expression, we show that loss of RhoBTB2 results in downregulation of CXCL14—a chemokine that controls leukocyte migration and angiogenesis, and whose expression is lost through unknown mechanisms in a wide range of epithelial cancers. Loss of RhoBTB2 expression correlates with loss of CXCL14 secretion by head and neck squamous cell carcinoma cell lines, whereas reintroduction of RhoBTB2 restores CXCL14 secretion. Our studies identify CXCL14 as a gene target of RhoBTB2 and support downregulation of CXCL14 as a functional outcome of RhoBTB2 loss in cancer. Oncogene (2008) 27, 6856–6865; doi:10.1038/onc.2008.317; published online 1 September 2008 Keywords: Rho GTPase; BTB domain; microarray; cancer; tumor microenvironment; chemokine

Introduction Rho GTPases are critical regulators of cell migration, but also control a wide range of other cellular processes, including membrane trafficking, cell-cycle progression and transcriptional regulation (Ridley, 2006). Many of these processes become deregulated in cancer and a subset of Rho GTPases have been shown to be upregulated in a range of tumor types (Sahai and Marshall, 2002; Ridley, 2004). The RhoBTB GTPases Correspondence: Dr H Mellor, Department of Biochemistry, School of Medical Sciences, University of Bristol, University Walk, Bristol BS8 1TD, UK. E-mail: [email protected] Received 5 June 2008; accepted 23 July 2008; published online 1 September 2008

comprise a novel subfamily of atypical Rho GTPases that show an unusual structure compared to other family members. In addition to the N-terminal Rho domain, they contain two tandem BTB/POZ (bric-abrac, tramtrack, broad complex/poxvirus and zinc finger) domains and a unique C-terminal region of unknown function (Ramos et al., 2002). Although the N-terminal Rho domain shows a high degree of homology with other Rho GTPases, studies have shown that RhoBTB2 does not show GTPase activity and fails to bind GTP (Chang et al., 2006). The RhoBTB proteins were first identified in Dictyostelium (Rivero et al., 2001) and subsequently shown to be highly conserved in Drosophila and in vertebrates (Ramos et al., 2002). Although the RhoBTB GTPases are comparatively ancient members of the Rho family, we know very little about their function. Recently, RhoBTB2 was isolated in a representational difference analysis screen for novel tumor suppressor genes in breast cancer, and given the alternative nomenclature DBC2 (deleted in breast cancer 2; we refer to RhoBTB2/DBC2 as RhoBTB2 throughout, following the HUGO designation for this gene) (Hamaguchi et al., 2002). RhoBTB2/DBC2 was shown to undergo homologous deletion in a relatively small number (3.5%) of breast tumor samples; however, RhoBTB2 expression was silenced at high frequency (approximately 50%) in breast and lung tumors (Hamaguchi et al., 2002). Sporadic point mutations in RhoBTB2 were also detected in breast tumor samples and breast cancer cell lines (Hamaguchi et al., 2002), and subsequent studies have identified further sporadic mutations of the RhoBTB2 coding region and promoter in bladder (Knowles et al., 2005) and breast (Ohadi et al., 2007) tumor samples. These studies support a role for RhoBTB2 in tumorigenesis; however, it is unclear what this might be. Unlike other Rho GTPases, RhoBTB2 does not affect the actin cytoskeleton (Aspenstrom et al., 2004) and so it seems unlikely that the effects of RhoBTB2 loss in cancer are directly on cell motility, as with RhoC (Clark et al., 2000). One function of the BTB domain is to mediate transcriptional regulation (Bardwell and Treisman, 1994). In this study, we have used RNA interference to mimic the loss of RhoBTB2 seen in cancer and then performed global analysis of gene expression to test the hypothesis that RhoBTB2 might function as a transcriptional regulator. We show that RhoBTB2 controls the expression of the homeostatic

Loss of RhoBTB2 results in downregulation of CXCL14 CM McKinnon et al

6857

chemokine CXCL14, a gene of importance in the progression of a wide range of epithelial tumors.

3.5

226084_s_at 222484_s_at

3.0

Results The silencing of RhoBTB2 causes a decrease in CXCL14 expression Hamaguchi and co-workers have previously examined the effects of silencing RhoBTB2 expression on gene expression in the HeLa cervical cancer cell line and observed changes in the expression of 680 genes (Siripurapu et al., 2005). The HeLa cell line was originally isolated from an aggressive cervical carcinoma and these cells are already fully transformed. We wanted to examine effects of RhoBTB2 silencing in normal primary human cells, where we might expect to be able to identify the specific contribution that loss of RhoBTB2 function makes to tumorigenesis without the masking effects of preexisting changes. RhoBTB2 is relatively highly expressed in lung (Supplementary Figure 1), and expression of RhoBTB2 is also silenced in a high percentage of lung tumors (Hamaguchi et al., 2002), and so we chose primary lung epithelial cells for this screen. Normal human bronchial epithelial (NHBE) cells were transfected with small-interfering RNA (siRNA) targeting RhoBTB2 or control target genes (cyclophilin B, lamin A/C). Specific changes in global gene expression on silencing of RhoBTB2 were detected by microarray screening of Affymetrix U133 Plus 2.0 whole-genome arrays. The use of two different siRNAs to RhoBTB2 allowed us to filter out any possible offtarget effects. Similarly, the use of two independent controls allowed us to filter out any changes that were due to downregulation of either control gene alone. In contrast to the previous screen of HeLa cells, after statistical testing, we saw only 33 Affymetrix probe sets (26 unique target genes) that were significantly changed (a false discovery rate (FDR) P-value of o0.05) on silencing of RhoBTB2 (data deposition: the microarray data referred to in this paper has been deposited in the NCBI GEO database with the accession number GSE8837) (Figure 1). Two of these probe sets showed a greater than twofold increase in target expression. One is annotated by the chip manufacturer as targeting MAP1B, the other targets GenBank sequence AL523076. Further investigation revealed that AL523076 maps to an EST tiling path at the 30 end of the MAP1B gene, indicating that the two probe sets share the same target. Two genes showed significant decrease in expression on silencing of RhoBTB2—RhoBTB2 itself and the chemokine CXCL14/BRAK. CXCL14 showed the greatest change in expression of any gene (>4-fold), greater even than RhoBTB2 (Figure 1). CXCL14 is expressed by epithelial cells (Hromas et al., 1999; Frederick et al., 2000) and is a chemokine for monocytes (Kurth et al., 2001), natural killer cells (Starnes et al., 2006) and immature dendritic cells (Shellenberger et al., 2004; Schaerli et al., 2005). CXCL14 expression is lost at

FDR p−value (−log base 10)

218002_s_at

2.5

2.0

212233_s_at

CXCL14 MAP1B RhoBTB2 CyclophilinB Lamin A/C

1.5

1.0

0.5

0 -2 -1 0 1 2 Fold Change (log base 2): RhoBTB2 Silenced vs Control

Figure 1 Effects of silencing RhoBTB2 on global gene expression. normal human bronchial epithelial (NHBE) cells were transfected with small-interfering RNA (siRNA) duplexes for either cyclophilin B, lamin A/C or one of two siRhoBTB2 duplexes, for 48 h before mRNA extraction. The mRNA was subjected to Affymetrix microarray analysis and analysed using Bioconductor R. The Figure shows a volcano plot of false discovery rate (FDR) P-values against fold changes. Vertical lines indicate the ±2 fold change threshold; horizontal lines indicate FDR P ¼ 0.05. Only two genes underwent significant change (fold change >2, Po0.05). These were CXCL14 (Affymetrix probe set IDs 222484_s_at and 218002_s_at) and MAP1B (Affymetrix probe set IDs 226084_s_at and 212233_s_at). The full data sets have been submitted to the NCBI GEO database with the accession number GSE8837.

high frequency in wide range of epithelial tumors (Hromas et al., 1999; Frederick et al., 2000; Schwarze et al., 2002; Shellenberger et al., 2004; Shurin et al., 2005), although the mechanism for this loss is unknown. Loss of CXCL14 expression is associated with reduced leukocyte infiltration in tumors, leading to the proposal that loss of CXCL14 allows tumors to evade immune surveillance (Shurin et al., 2005). CXCL14 has also been shown to be a potent inhibitor of angiogenesis, suggesting that loss of CXCL14 might also contribute to the angiogenic switch (Shellenberger et al., 2004). Interestingly, CXCL14 was not reported as a RhoBTB2 target in the HeLa cell screen (Siripurapu et al., 2005). The identification of an established cancer gene as a potential target of RhoBTB2 is clearly of interest in terms of understanding the role of RhoBTB2 loss in tumorigenesis. To validate CXCL14 as a RhoBTB2 target, we used quantitative real-time PCR (qPCR) to quantify the levels of CXCL14 expression accurately in NHBE cells treated with or without siRNA to RhoBTB2. Although we were able to silence RhoBTB2 expression by only 40% in these cells, this led to a profound inhibition of CXCL14 expression (Figure 2a). Although CXCL14 was not reported as a target of RhoBTB2 in HeLa cells previously (Siripurapu et al., 2005), we saw the same robust loss of CXCL14 expression in HeLa cells treated with RhoBTB2 siRNAs (Figure 2b). Oncogene


6858 CXCL14

RhoBTB2 1.4 mRNA (normalised to RPII)

mRNA (normalised to RPII)

1.4 1.2 1.0 0.8 0.6 0.4 0.2

1.2 1.0 0.8 0.6 0.4 0.2 0

0 C

siLm

C

siRB2 1 2

RhoBTB2

siRB2 1 2

CXCL14

1.4

1.4 mRNA (normalised to RPII)


siLm

1.2 1.0 0.8 0.6 0.4 0.2

1.2 1.0 0.8 0.6 0.4 0.2

0

0 C

siLm

C

siRB2 1

2

C

siLm

siLm

siRB2 1

siRB2 1

2

siRB1 2

RhoBTB2

tubulin

Figure 2 Effects of silencing RhoBTB2 on CXCL14 expression in normal human bronchial epithelial (NHBE) and HeLa cells. Cells were transfected for 48 h (HeLa) or 72 h (NHBE) with small-interfering RNA (siRNA) duplexes for either lamin A/C (siLm) or one of two non-overlapping siRhoBTB2 duplexes (siRB2). Reverse transcription was then performed and the cDNAs subjected to quantitative real-time PCR using specific primers for RNA polymerase II (RPII), RhoBTB2 or CXCL14, as indicated. Values are expressed relative to the levels of RPII. (a) Levels of RhoBTB2 and CXCL14 mRNA in NHBE cells, comparing untransfected control cells to cells transfected with siRNA duplexes for lamin, or two separate siRNAs to RhoBTB2. Data show the means±s.e.m. for two independent experiments. (b) Levels of RhoBTB2 and CXCL14 mRNA in HeLa cells comparing untransfected cells control to cells transfected with siRNA duplexes for lamin, or two separate siRNAs to RhoBTB2. Data show the means±s.e.m. for three independent experiments. (c) HeLa cells stably expressing RhoBTB2 were treated with siRNA duplexes for lamin A/C, RhoBTB2 or RhoBTB1 (siRB1) for 48 h. Western blot analysis was performed for RhoBTB2 (top) or tubulin (bottom) to determine the efficiency and specificity of the RhoBTB2 siRNAs.

RhoBTB1 and RhoBTB2 control CXCL14 expression in normal keratinocytes Loss of mRNA expression does not necessarily correlate with loss of protein expression. We used a specific CXCL14 enzyme-linked immunosorbent assay (ELISA) assay to measure levels of secreted CXCL14 in cells treated with or without RhoBTB2 siRNA. CXCL14 is expressed by a wide range of epithelial cells, but at Oncogene

different levels (Hromas et al., 1999; Frederick et al., 2000). The level of CXCL14 in medium conditioned by NHBE cells was too low to detect in this assay (o100 pg/ml; data not shown). We therefore examined keratinocytes, which express high levels of CXCL14 (Frederick et al., 2000; Shurin et al., 2005). As with NHBE cells, normal primary human keratinocytes showed dramatic downregulation of CXCL14 mRNA


6859 RhoBTB2

CXCL14



1.2 1.0 0.8 0.6 0.4 0.2 0

1.0 0.8 0.6 0.4 0.2 0

C

siLm

siRB2 siRB1 siRB2 + siRB1

C


Secretion of CXCL14

RhoBTB1

2500

1.2 1.0

2000 [CXCL14] (pg/ml)


siLm

0.8 0.6 0.4 0.2 0

1500 1000 500 0

C

siLm


C

siLm

siRB2

siRB1 siRB2 + siRB1

Figure 3 RhoBTB2 and RhoBTB1 are both required for CXCL14 expression in primary keratinocytes. Cells were transfected for 72 h with small-interfering RNA (siRNA) duplexes for either lamin A/C (siLm), RhoBTB2 (siRB2), RhoBTB1 (siRB1) or a combination of RhoBTB2 and RhoBTB1. Reverse transcription was then performed and the cDNAs subjected to quantitative real-time PCR using specific primers for RNA polymerase II (RPII), RhoBTB2, CXCL14 or RhoBTB1, as indicated. Values are expressed relative to the levels of RPII. Secreted CXCL14 was measured by enzyme-linked immunosorbent assay (ELISA) of the cell media. Data are the means±s.e.m. of three independent experiments.

on silencing of RhoBTB2 (Figure 3). This was matched by a loss of secreted CXCL14 from these cells (Figure 3). Humans have two related RhoBTB genes; RhoBTB1 and RhoBTB2, and the two proteins show a high degree of similarity in sequence (73% identity). We therefore examined the effects of silencing RhoBTB1 expression on CXL14 expression and secretion in keratinocytes. As with RhoBTB2, silencing of RhoBTB1 led to marked inhibition of CXCL14 expression (Figure 3). Interestingly, silencing of either RhoBTB isoform had a broadly equivalent effect, suggesting that both proteins are required for maintenance of CXCL14 expression. Silencing of RhoBTB1 and RhoBTB2 together had a slightly stronger effect on CXCL14 expression than silencing of either isoform alone (Figure 3). In all cases, loss of CXCL14 mRNA correlated with loss of secreted chemokine (Figure 3). Loss of RhoBTB2 expression in HNSCC cell lines correlates with loss of CXCL14 CXCL14 expression is lost at high frequency in head and neck squamous cell carcinoma (HNSCC; Frederick et al., 2000; Shurin et al., 2005), although the mechanism

of loss is unknown. As these tumors derive from keratinocytes, we examined the expression of RhoBTB1 and RhoBTB2 in four different HNSCC keratinocyte clonal cell lines derived from patients with head and neck tumors (Prime et al., 2004). qPCR analysis revealed that all four HNSCC cell lines had markedly reduced levels of RhoBTB2 and CXCL14 mRNA, compared to normal keratinocytes (Figure 4). This was paralleled with a significant decrease in CXCL14 secretion in the HNSCC cell lines compared to normal keratinocytes (Figure 4). Previous studies have shown reduced expression of RhoBTB1 in 37% of HNSCC tumor samples (Beder et al., 2006). Interestingly, the levels of RhoBTB1 were comparable between HNSCC and primary keratinocyte cells in these samples (Figure 4). Reintroduction of RhoBTB2 causes the restoration of CXCL14 expression in HNSCC cells The most straightforward explanation of these data is that loss of RhoBTB2 expression in these HNSCC lines is the cause of loss of CXCL14. It is possible, however, that the two events are simply coincidental. To test this, we used a lentiviral vector to reintroduce RhoBTB2 Oncogene


6860 RhoBTB2

CXCL14 1.2 mRNA (normalised to RPII)


1.2 1.0 0.8 0.6 0.4 0.2 0

1.0 0.8 0.6 0.4 0.2 0

K

H413 H357 H314 H376

K

Secretion of CXCL14

1.4 2500

1.2 [CXCL14] (pg/ml)


RhoBTB1

H413 H357 H314 H376

1.0 0.8 0.6 0.4 0.2 0

2000 1500 1000 500 0

K

H413 H357 H314 H376

K

H413

H357

H314

H376

Figure 4 Loss of RhoBTB2 expression in head and neck squamous cell carcinoma (HNSCC) cell lines correlates with loss of CXCL14. Primary human keratinocytes (K) or four individual human HNSCC cell lines were seeded a density of 1 105 cells per ml into six-well dishes and cultured for 72 h before total RNA isolation was performed. This was then followed by reverse transcription and quantitative real-time PCR of the resulting cDNAs performed using specific primers for RNA polymerase II (RPII), RhoBTB2, CXCL14 and RhoBTB1, as indicated. Values are expressed relative to the levels of RPII mRNA. Secreted CXCL14 was measured by enzyme-linked immunosorbent assay (ELISA) of the cell media. Data are the means±s.e.m. of three independent experiments.

into these cell lines to see if this was able to restore CXCL14 expression. We were able to return RhoBTB2 expression to near-normal levels in a subset of HNSCC lines (Figure 5). Immunofluorescence microscopy showed that the efficiency of viral infection was much higher in these cells (Supplementary Figure 2). Where we were able to restore RhoBTB2 expression, we saw concomitant restoration of CXCL14 expression (Figure 5). Beder et al. (2006) have reported that RhoBTB1 expression is downregulated in 37% of HNSCC. We had seen that silencing of RhoBTB1 in normal keratinocytes led to downregulation of CXCL14 expression (Figure 4), suggesting that the loss of either RhoBTB isoform may underlie CXCL14 loss in HNSCC. As none of our patient cell lines showed downregulation of RhoBTB1 (Figure 4), we sought to mimic RhoBTB1 loss in HNSCC to examine whether this event could also lead to downregulation of CXCL14. We took the H376 cell line where we had successfully restored RhoBTB2 expression and mimicked RhoBTB1 loss using siRNA. Essentially, this modeled the alternative pathway—loss of RhoBTB1— in the same HNSCC line. As predicted, silencing of RhoBTB1 led to significant downregulation of CXCL14 expression (Figure 6). We conclude that the loss of Oncogene

expression of either RhoBTB2 or RhoBTB1 in HNSCC leads to loss of CXCL14. RhoBTB2 regulation of CXCL14 expression is independent of Cul3 Little is known of the signaling pathways downstream of RhoBTB2. Recently, RhoBTB2 has been shown to bind the E3 ubiquitin ligase Cul3 through its first BTB domain (Wilkins et al., 2004). Several other BTB proteins have also been shown to recruit Cul3, and to use this to target specific proteins for degradation (Perez-Torrado et al., 2006). In some cases these protein targets are transcriptional regulators (Gentleman et al., 2004; Kobayashi et al., 2004). It would seem possible that the regulation of CXCL14 expression by RhoBTB1/RhoBTB2 may occur through a similar mechanism, with the regulation of CXCL14 occurring indirectly through degradation of an inhibitor of CXCL14 expression. To examine this, we looked to see if Cul3 was required for the restoration of CXCL14 expression in HNSCC cell lines by RhoBTB2. We were able to silence Cul3 expression effectively using two independent siRNA oligonucleotide duplexes; however, silencing of Cul3 had no significant effect on the ability of RhoBTB2 to restore CXCL14 expression to the


6861 RhoBTB2


2.5 2.0 1.5 1.0 0.5

0 Virus

K

+

H413

+

H357

-

+

-

+

H314

H376

-

-

CXCL14


-

1.4 1.2 1.0 0.8 0.6 0.4 0.2

0 Virus

K

+

H413

-

+

H357

+

H314

+

H376

Figure 5 Restoration of RhoBTB2 expression in head and neck squamous cell carcinoma (HNSCC) cell lines restores CXCL14 expression. Four HNSCC cell lines were infected with a lentivirus expressing RhoBTB2 48 h before total RNA isolation. Primary human keratinocytes (K) were used as a reference. Reverse transcription was then performed and the cDNAs subjected to quantitative real-time PCR using specific primers for RNA polymerase II (RPII), RhoBTB2 or CXCL14, as indicated. Values are expressed relative to the levels of RPII mRNA. Secreted CXCL14 was measured by enzyme-linked immunosorbent assay (ELISA) of the cell media. Data show the means±s.e.m. of three independent experiments.

HNSCC line (Figure 7). We conclude that the regulation of CXCL14 expression by RhoBTB2 does not involve Cul3-mediated protein ubiquitylation.

Discussion In this study, we have used silencing of RhoBTB2 in primary epithelial cells to identify the homeostatic chemokine CXCL14 as a target of RhoBTB2 regulation in normal and cancerous epithelial cells. CXCL14 expression is highly sensitive to cellular levels of RhoBTB2, and even partial loss of RhoBTB2 expression leads to significant downregulation of the chemokine. This is important in the context of the potential role of RhoBTB2 in tumorigenesis as complete loss of this gene

in tumors is comparatively rare—far more common is reduced expression or mutation of one allele (Hamaguchi et al., 2002). Similarly, the most common derangement of RhoBTB1 in HNSCC is allelic loss accompanied by partial loss of expression, leading to the proposal that haploinsufficiency of RhoBTB1 may underlie its role in tumorigenesis (Beder et al., 2006). RhoBTB1 and RhoBTB2 would therefore fit the classification of type II tumor suppressor genes—those where partial loss of function is sufficient for effect. Also important is the finding that loss of either RhoBTB1 or RhoBTB2 leads to loss of CXCL14 expression. In many proteins BTB domains mediate protein–protein dimerization (Perez-Torrado et al., 2006). It is therefore tempting to speculate that the functional regulator of CXCL14 expression is a heterodimer of RhoBTB1/ RhoBTB2. In this way, loss of expression of either isoform would lead to loss of CXCL14 expression. In many proteins, the BTB domain mediates transcriptional repression by recruiting co-repressor complexes (Perez-Torrado et al., 2006). Several BTBcontaining proteins have proven links to cancer (Kelly and Daniel, 2006). PLZF is a translocation partner of retinoic acid receptor-a (RAR-a) in a subset of acute promyelocytic leukemia. The resultant fusion protein targets the transcriptional co-repressor SMRT to RAR-aregulated genes, leading to transcriptional repression (Redner, 2002). Overexpression of the BTB-domaincontaining protein BCL6 in large B-cell lymphoma leads to an inhibition of apoptotic cell death. The normal function of BCL6 is to repress transcription of p53 and p21, and its overexpression allows B cells to escape from normal controls of cell survival (Pasqualucci et al., 2003). Our initial expectation was that loss of RhoBTB2 would similarly lead to de-repression of target genes and an upregulation of their expression. Instead, the gene that showed the most significant change on silencing of RhoBTB2 displayed marked downregulation of expression. PLZF and BCL6 are part of the BTB-ZF category of BTB-domain-containing proteins and contain zincfinger DNA-binding domains that allow them to target transcriptional co-repressors to their specific gene targets (Perez-Torrado et al., 2006). RhoBTB1 and RhoBTB2 lack a zinc-finger or any other obvious DNAbinding motif, and so do not fit into this category. Recently, RhoBTB2 has been shown to bind the E3 ubiquitin ligase Cul3 through its first BTB domain (Wilkins et al., 2004). The targets for RhoBTB2-Cul3mediated protein degradation are unknown; however, we were able to show that Cul3 is not required for the regulation of CXCL14 by RhoBTB2 in HNSCC. CXCL14 expression is lost at high frequency in wide range of epithelial tumors (Hromas et al., 1999; Frederick et al., 2000; Schwarze et al., 2002; Shellenberger et al., 2004; Shurin et al., 2005) and especially in HNSCC, where CXCL14 expression is lost in the majority of tumors (Frederick et al., 2000; Shurin et al., 2005). Head and neck tumors that do not express CXCL14 have low levels of infiltrating dendritic cells, and this can be reversed by reintroduction of CXCL14 expression (Shurin et al., 2005). Dendritic cells are Oncogene


6862 RhoBTB2

RhoBTB1

CXCL14

Secretion of CXCL14 2000

1.0 [CXCL14] (pg/ml)


1.2

0.8 0.6 0.4

1500

1000

500

0.2 0

0 -

+

-

+

-

+

siRB1

K

H376-2 + siRB1

Figure 6 Loss of RhoBTB1 expression is an alternative route to loss of CXCL14 expression in head and neck squamous cell carcinoma (HNSCC) cells. The HNSCC cell line H376 was stably transfected with RhoBTB2 to restore expression. This cell line (H376-2) was then transfected for 72 h with small-interfering RNA (siRNA) duplexes targeting RhoBTB1 (siRB1) or with a lamin A/C control. Total RNA isolation was performed as before. This was followed by a reverse transcription to produce cDNA and quantitative real-time PCR performed using specific primers for RNA polymerase II (RPII), RhoBTB2, CXCL14 and RhoBTB1, as indicated. Values are expressed relative to the levels of RPII mRNA. CXCL14 secretion was measured by enzyme-linked immunosorbent assay (ELISA). Data show the means±s.e.m. of three independent experiments.

important in cancer immunosurveillance (Yang and Carbone, 2004) and loss of CXCL14 expression correlates with increased tumor growth (Shurin et al., 2005). CXCL14 has also been shown to be a potent inhibitor of angiogenesis, suggesting that loss of CXCL14 might also contribute to the neovascularization of tumors (Shellenberger et al., 2004). The mechanisms controlling the expression of CXCL14 in normal and cancerous tissue are completely unknown. Our studies show that CXCL14 expression is potently regulated by the atypical Rho GTPases RhoBTB1 and RhoBTB2 in normal epithelial cells, and that the loss of RhoBTB2 in HNSCC cell lines leads to loss of CXCL14 expression. Importantly, the percentage of epithelial tumors showing reduced expression of RhoBTB2 (Hamaguchi et al., 2002) or RhoBTB1 (Beder et al., 2006) is comparable to the percentage of tumors showing loss of CXCL14 expression. Taken together, these data suggest that loss of RhoBTB1/2 is the major factor controlling CXCL14 expression in tumorigenesis.

Materials and methods Reagents and chemicals Lipofectamine 2000 was from Invitrogen (Paisley, UK). GeneFECTOR was from Venn Nova (FL, USA). siRNA oligonucleotides were from MWG (Germany). siRNA SmartPools were from Dharmacon (CO, USA). Rabbit polyclonal DBC2 (C20) antibody was from Delta Biolabs (CA, USA). Mouse monoclonal Cul3 (C18) antibody was from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Horseradish peroxidase (HRP)-conjugated secondary antibodies were from Jackson Laboratories (Suffolk, UK). Donkey anti-rabbit antibody conjugated to Alexa Fluor 488 was from Molecular Probes (Invitrogen). Oncogene

Cell culture and maintenance HeLa cells were obtained from the American Type Culture Collection (Rockville, MD, USA). HeLa and HEK293T cells were maintained in Dulbecco’s modified Eagle’s medium containing 10% heat-inactivated bovine fetal calf serum. Primary NHBE were from Lonza (Wokingham, UK) and were maintained in NHBE media supplemented with the corresponding BulletKit (Lonza). Primary human keratinocyte cells (Lonza) were maintained in keratinocyte basal media supplemented with the corresponding BulletKit and subcultured using the keratinocyte subculturing kit (Lonza), as per the manufacturer’s instructions. Human HNSCC lines were a generous gift from Ian Patterson (Bristol Dental School) and were cultured as described previously (Prime et al., 2004). RhoBTB2 expression patterns A human Multiple Tissue Expression (MTE) Array (Clontech, Saint-Germain-en-Laye, France) was probed with a 793 bp PCR fragment from the 30 untranslated region of RhoBTB2. Probes were labeled with [32P]dCTP (ICN) by random primer extension using the Rapid Mutiprime DNA labeling kit (Amersham, Buckinghamshire, UK). Unincorporated 32P was removed by gel filtration on a Nick column (Pharmacia Biotech, Buckinghamshire, UK). The MTE array was prehybridized for 30 min at 65 1C with continuous agitation in ExpressHyb solution (Clontech), and hybridized overnight in the same solution, containing 1 ng/ml of the probe at 109 c.p.m. per mg, under the same conditions. The membrane was washed five times in 2 SCC, 1% SDS at 65 1C and twice in 0.1% SCC, 0.5% SDS. The MTE array was then exposed to X-ray film. siRNA introduction into cells siRNA oligonucleotides were introduced into HeLa cells using calcium phosphate. Briefly, cells were cultured in 10 cm dishes until 70% confluent (approximately 7 107 cells). Transfection solutions were prepared by mixing 4 ml of 20 mM siRNA in 100 ml 0.25 M CaCl2 solution with 100 ml 50 mM BES (pH 6.95), 280 mM NaCl, 1.5 mM Na2HPO4. After 15 min incubation, this suspension was added dropwise to the dish. Samples were


6863 C

siLa

and incubated at 37 1C/5% CO2 for 4 h. After 4 h, the siRNA/ GeneFECTOR mix was removed and replaced with 2 ml of complete media. Cells were then incubated for a further 72 h. siRNA sequences are detailed in Supplementary Table 1.

siCL3 1

2

Cul3 tubulin

RhoBTB2 mRNA (Normalised to RPII)

3.0 2.5 2.0 1.5 1.0 0.5 0 - +

- +

- +

- +

C

siLm

siCL3-1

siCL3-2

RhoBTB2

CXCL14 mRNA (Normalised to RPII)

2.5 2.0 1.5 1.0 0.5 0 - +

- +

- +

- +

C

siLm

siCL3-1

siCL3-2

RhoBTB2

Figure 7 RhoBTB2 regulation of CXCL14 expression is not by the Cul3 ubiquitin ligase. The head and neck squamous cell carcinoma (HNSCC) cell line H376 was stably transfected with RhoBTB2 to restore expression. This cell line (H376-2; solid bars), and the parent line (open bars) were then transfected for 72 h with small-interfering RNA (siRNA) duplexes targeting lamin A/C (siLm), or two independent siRNAs to Cul3 (siCL3-1 and siCL3-2). (a) Western blot analysis of H376-2 cell extracts was performed to ensure the expression of Cul3 siRNAs resulted in a sufficient decrease in Cul3 protein. (b) Total RNA was extracted and reverse transcription performed as previously described. The cDNAs were then subjected to quantitative real-time PCR using specific primers for RNA polymerase II (RPII), RhoBTB2 or CXCL14 as indicated. Values are expressed relative to the levels of RPII. Data show the means±s.e.m. of three independent experiments.

incubated at 37 1C/3% CO2 overnight. The media was then replaced twice and samples returned to 37 1C/5% CO2 for a further 48 h. siRNA oligonucleotides were introduced into primary human cells using GeneFECTOR. Cells were seeded in six-well dishes (2 105 cells per dish) on the day before transfection. siRNA oligonucleotide (4 mg) and GeneFECTOR (4 ml) in 200 ml OptiMEM (Invitrogen) were added to each well

Microarray screening and analysis NHBE cells were transfected with siRNA duplexes targeting cyclophilin B, lamin A/C or RhoBTB2. siRNA sequences are detailed in Supplementary Table 1. Cells were transfected using GeneFECTOR, as described previously, except that the cells were incubated for 48 h. RNA isolation was performed using the RNeasy kit (Qiagen, Crawley, UK), as per the manufacturer’s instructions. cDNA was then produced using SuperScript First-Strand Synthesis System (Invitrogen). cDNA samples were sent to the microarray facility at Cancer Research UK (Patterson Institute, Manchester, UK) and used to screen Affymetrix U133 Plus 2.0 human whole-genome arrays. Microarray data analysis was performed using Bioconductor (Gentleman et al., 2004), an open source software project for the analysis of genomic data. Data from three independent experiments were combined to identify candidate genes regulated by RhoBTB2. Two separate siRNA duplexes against RhoBTB2 were used in each experiment to allow for elimination of possible off-target effects. Similarly, two separate control target genes were used (lamin A/C and cyclophilin B) to allow filtering of any changes due to silencing of either control. Probeset expression measures were calculated using the ‘affy’ (Gautier et al., 2004) package’s Robust Multichip Average (Irizarry et al., 2003) default method. Linear models were fitted to the data for each probe set using the ‘limma’ (Smyth, 2005) package’s lmFit function to model experiment-specific effects and remove them from further analysis. Differential gene expression was assessed between RhoBTB2 silenced and control replicate groups using an empirical Bayes t-test as implemented in the ‘limma’ package. The resultant P-values were adjusted for multiple testing using the Benjamini and Hochberg method (Benjamini and Hochberg, 1995). Any probe sets that exhibited an FDR-controlled P-value o0.05 and absolute fold change >2 were called differentially expressed. Data sets were submitted to the NCBI GEO database, with the accession number GSE8837. Quantitative real-time PCR RNA was isolated from adherent cells using the TRIzol extraction method (Gibco-BRL Life Technologies, Paisley, UK), quantified by spectrophotometry, and 50 mg of purified RNA used for reverse transcription using Omniscript RTase (Qiagen) for 1 h at 37 1C. cDNAs were then subjected to realtime PCR using DyNAmo Flash SYBR Green (Finnzymes, Espoo, Finland). A full list of oligonucleotide primers is given in Supplementary Table 1. Amplification was performed and detected using an Opticon 2 (MJ Research, MA, USA) thermocycler and data were analysed using the comparative Ct method (ABI Biosystems, CA, USA). Measurement of secreted CXCL14 by ELISA The amount of CXCL14 secreted from cells was measured using a sandwich ELISA assay kit from R&D Systems (MN, USA), as per the manufacturer’s instructions except for the following changes: the capture antibody was diluted to a working solution of 2 mg/ml and incubated on the plate for 18 h at 4 1C. Immobilized HRP-coupled secondary antibody was quantified by using QuantaBlu Fluorogenic Peroxidase Substrate (Pierce, IL, USA) and measuring the fluorescence on a SpectraMAX M2 fluorimeter (Molecular Devices, CA, USA) using an excitation wavelength of 325 nm and an Oncogene


6864 emission wavelength of 420 nm. Assays were calibrated using serial dilutions of recombinant human CXCL14 (R&D Systems). Lentiviral cloning and production A 2.4 kbp fragment corresponding to full-length human RhoBTB2 was cloned into the BamHI and Sbf1 sites of the pSEWsin lentiviral vector backbone (a generous gift of Adrian Thrasher, ICH, London). Virus was generated by transfection of HEK293T cells as described (Demaison et al., 2002). Briefly, 40 mg of pSEWsin or pSEWsin–RhoBTB2 were mixed with 10 mg of packaging plasmid pMDG2 and 30 mg of packaging plasmid p8.91 and 1 ml of 1 mM polyethylenimine (Sigma, Poole, UK) in OptiMEM. 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 amplification. After the incubation, the media were removed and centrifuged for 10 min at 2600 g. The supernatant was then passed through a 0.45 mm filter and this filtrate was used as the virus stock. Immunofluorescence microscopy HNSCC cells were plated onto acid-washed glass coverslips and allowed to adhere overnight. After 48 h of viral infection, cells were fixed for 15 min in 4% paraformaldehyde in phosphate-buffered saline (PBS), washed in PBS and then permeabilized in 0.2% Triton X-100 in PBS for 5 min. The cells were then washed again in PBS and incubated with 0.1% sodium borohydride for 10 min. The cells were washed three times in PBS then incubated with a 1:200 dilution of antiRhoBTB2 antibody (C20) in 1% bovine serum albumin for 1 h. The cells were washed three times in PBS then incubated for 45 min with anti-rabbit Alexa Fluor 488 antibody in PBS. The cells were washed three times in PBS and mounted over

MOWIOL 4-88 (Calbiochem, Darmstadt, Germany) containing 0.6% 1,4-diazabicyclo-(2.2.2) octane (Sigma) as an antiphotobleaching 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 63/1.32 oil-immersion objective. Fluorophores were excited using the 405 nm line of a diode laser (46-diamidino-2-phenyl indole), or the 488 nm line of a krypton-argon laser (Alexa 488). A series of images were taken at 0.5 mm intervals through the Z-plane of the cells and processed to form a projected image. Immunoblotting Cells grown in six-well tissue culture plates were washed three times with PBS and scraped into 800 ml SDS–polyacrylamide gel electrophoresis (PAGE) sample buffer. Samples were heated to 95 1C for 5 min. Proteins were resolved by SDS–PAGE and transferred onto polyvinylidene fluoride membrane for immunoblotting. Statistical analysis Unless where otherwise stated, data are given as means±s.e.m. of three entirely independent experiments. Comparisons between means were performed using the Student’s t-test. Acknowledgements We thank Ian Paterson for providing the HNSCC cell lines and Adrian Thrasher for the provision of lentiviral reagents. We particularly thank James Doherty, who provided initial analysis of the microarray data. This work was supported by a grant awarded to HM by Cancer Research UK.

References Aspenstrom P, Fransson A, Saras J. (2004). Rho GTPases have diverse effects on the organization of the actin filament system. Biochem J 377: 327–337. Bardwell VJ, Treisman R. (1994). The POZ domain: a conserved protein–protein interaction motif. Genes Dev 8: 1664–1677. Beder LB, Gunduz M, Ouchida M, Gunduz E, Sakai A, Fukushima K et al. (2006). Identification of a candidate tumor suppressor gene RHOBTB1 located at a novel allelic loss region 10q21 in head and neck cancer. J Cancer Res Clin Oncol 132: 19–27. Benjamini Y, Hochberg Y. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Statis Soc Ser B 57: 289–300. Chang FK, Sato N, Kobayashi-Simorowski N, Yoshihara T, Meth JL, Hamaguchi M. (2006). DBC2 is essential for transporting vesicular stomatitis virus glycoprotein. J Mol Biol 364: 302–308. Clark EA, Golub TR, Lander ES, Hynes RO. (2000). Genomic analysis of metastasis reveals an essential role for RhoC. Nature 406: 532–535. Demaison C, Parsley K, Brouns G, Scherr M, Battmer K, Kinnon C et al. (2002). High-level transduction and gene expression in hematopoietic repopulating cells using a human immunodeficiency (correction of imunodeficiency) virus type 1-based lentiviral vector containing an internal spleen focus forming virus promoter. Hum Gene Ther 13: 803–813. Frederick MJ, Henderson Y, Xu X, Deavers MT, Sahin AA, Wu H et al. (2000). In vivo expression of the novel CXC chemokine BRAK in normal and cancerous human tissue. Am J Pathol 156: 1937–1950. Gautier L, Cope L, Bolstad BM, Irizarry RA. (2004). affy—analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 20: 307–315. Oncogene

Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S et al. (2004). Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5: R80. Hamaguchi M, Meth JL, von Klitzing C, Wei W, Esposito D, Rodgers L et al. (2002). DBC2, a candidate for a tumor suppressor gene involved in breast cancer. Proc Natl Acad Sci USA 99: 13647–13652. Hromas R, Broxmeyer HE, Kim C, Nakshatri H, Christopherson K, Azam M et al. (1999). Cloning of BRAK, a novel divergent CXC chemokine preferentially expressed in normal versus malignant cells. Biochem Biophys Res Commun 255: 703–706. Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U et al. (2003). Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4: 249–264. Kelly KF, Daniel JM. (2006). POZ for effect—POZ-ZF transcription factors in cancer and development. Trends Cell Biol 16: 578–587. Knowles MA, Aveyard JS, Taylor CF, Harnden P, Bass S. (2005). Mutation analysis of the 8p candidate tumour suppressor genes DBC2 (RHOBTB2) and LZTS1 in bladder cancer. Cancer Lett 225: 121–130. Kobayashi A, Kang MI, Okawa H, Ohtsuji M, Zenke Y, Chiba T et al. (2004). Oxidative stress sensor Keap1 functions as an adaptor for Cul3-based E3 ligase to regulate proteasomal degradation of Nrf2. Mol Cell Biol 24: 7130–7139. Kurth I, Willimann K, Schaerli P, Hunziker T, Clark-Lewis I, Moser B. (2001). Monocyte selectivity and tissue localization suggests a role for breast and kidney-expressed chemokine (BRAK) in macrophage development. J Exp Med 194: 855–861. Ohadi M, Totonchi M, Maguire P, Lindblom A, Habibi R, Alavi BA et al. (2007). Mutation analysis of the DBC2 gene in sporadic and familial breast cancer. Acta Oncol 46: 770–772.


6865 Pasqualucci L, Bereschenko O, Niu H, Klein U, Basso K, Guglielmino R et al. (2003). Molecular pathogenesis of non-Hodgkin’s lymphoma: the role of Bcl-6. Leuk Lymphoma 44: S5–S12. Perez-Torrado R, Yamada D, Defossez P-A. (2006). Born to bind: the BTB protein–protein interaction domain. BioEssays 28: 1194–1202. Prime SS, Eveson JW, Stone AM, Huntley SP, Davies M, Paterson IC et al. (2004). Metastatic dissemination of human malignant oral keratinocyte cell lines following orthotopic transplantation reflects response to TGF-beta 1. J Pathol 203: 927–932. Ramos S, Khademi F, Somesh BP, Rivero F. (2002). Genomic organization and expression profile of the small GTPases of the RhoBTB family in human and mouse. Gene 298: 147–157. Redner RL. (2002). Variations on a theme: the alternate translocations in APL. Leukemia 16: 1927–1932. Ridley A. (2004). Rho proteins and cancer. Breast Cancer Res Treat 84: 13–19. Ridley AJ. (2006). Rho GTPases and actin dynamics in membrane protrusions and vesicle trafficking. Trends Cell Biol 16: 522–529. Rivero F, Dislich H, Glockner G, Noegel AA. (2001). The Dictyostelium discoideum family of Rho-related proteins. Nucleic Acids Res 29: 1068–1079. Sahai E, Marshall CJ. (2002). RHO-GTPases and cancer. Nat Rev Cancer 2: 133–142. Schaerli P, Willimann K, Ebert LM, Walz A, Moser B. (2005). Cutaneous CXCL14 targets blood precursors to epidermal niches for Langerhans cell differentiation. Immunity 23: 331–342. Schwarze R, DePrimo SE, Grabert LM, Fu VX, Brooks JD, Jarrard DF. (2002). Novel pathways associated with bypassing cellular

senescence in human prostate epithelial cells. J Biol Chem 277: 14877–14883. Shellenberger TD, Wang M, Gujrati M, Jayakumar A, Strieter RM, Burdick MD et al. (2004). BRAK/CXCL14 is a potent inhibitor of angiogenesis and a chemotactic factor for immature dendritic cells. Cancer Res 64: 8262–8270. Shurin GV, Ferris RL, Tourkova IL, Perez L, Lokshin A, Balkir L et al. (2005). Loss of new chemokine CXCL14 in tumor tissue is associated with low infiltration by dendritic cells (DC), while restoration of human CXCL14 expression in tumor cells causes attraction of DC both in vitro and in vivo. J Immunol 174: 5490–5498. Siripurapu V, Meth J, Kobayashi N, Hamaguchi M. (2005). DBC2 significantly influences cell-cycle, apoptosis, cytoskeleton and membrane-trafficking pathways. J Mol Biol 346: 83–89. Smyth GK. (2005). Limma: linear models for microarray data. In: Gentleman R, Carey V, Dudoit S, Irizarry R, Huber W (eds). Bioinformatics and Computational Biology Solutions using R and Bioconductor. Springer: New York. pp 397–420. Starnes T, Rasila KK, Robertson MJ, Brahmi Z, Dahl R, Christopherson K et al. (2006). The chemokine CXCL14 (BRAK) stimulates activated NK cell migration: implications for the downregulation of CXCL14 in malignancy. Exp Hematol 34: 1101–1105. Wilkins A, Ping Q, Carpenter CL. (2004). RhoBTB2 is a substrate of the mammalian Cul3 ubiquitin ligase complex. Genes Dev 18: 856–861. Yang L, Carbone DP. (2004). Tumor–host immune interactions and dendritic cell dysfunction. Adv Cancer Res 92: 13–27.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene