|Home | About | Journals | Submit | Contact Us | Français|
Aberrant constitutive expression of the NF-κB c-Rel and RelA subunits in breast cancer cells was shown to promote their survival. Recently, we demonstrated that aggressive breast cancers constitutively express high levels of the RelB subunit, which promotes their more invasive phenotype via induction of the BCL2 gene. As these cancers are frequently resistant to therapy, here we tested the hypothesis that RelB promotes their survival. High RelB expressing Hs578T and MDA-MB-231 breast cancer cells were more resistant to γ-radiation than MCF7 and ZR-75 cells, which express lower RelB levels. Knockdown of RelB in Hs578T led to decreased survival in response to γ-irradiation, while conversely ectopic expression of RelB in MCF7 cells protected these cells from radiation. Similar data were obtained upon treatment of Hs578T or MCF7 cells with the chemotherapeutic agent doxorubicin. High serum levels of 25-hydroxyvitamin D are associated with decreased breast cancer risk and mortality, although, the mechanism of its protective action has not been elucidated. Treatment of Hs578T and Her-2/neu-driven NF639 cells with 1,25-dihydroxyvitamin D3 decreased RelB/RELB gene expression and levels of pro-survival targets Survivin, MnSOD and Bcl-2, while increasing their sensitivity to γ-irradiation. Thus, RelB, which promotes survival and a more highly invasive phenotype of breast cancer cells, is a target of 1,25-dihydroxyvitamin D3, providing one mechanism for the observed protective role of 25-hydroxyvitamin D in patients with breast cancer.
NF-κB is a family of dimeric transcription factors that control genes which promote cell survival, proliferation and invasive phenotype (Ghosh et al., 1998; Rayet and Gelinas, 1999; Sonenshein, 1997). Mammals express five NF-κB members: c-Rel, RelB, RelA (also known as p65), p50 and p52. NF-κB factors bind as hetero-or homo-dimers, except for RelB which has a structurally different N-terminus and only binds as a heterodimer typically with either p50 or p52 (Siebenlist et al., 1994). NF-κB complexes differentially regulate gene expression possibly due to various interactions with other transcription factors or kinases and to different DNA binding specificities. For example, RelB is unable to induce transcription of the gene encoding the pro-survival factor Bcl-XL, whereas RelA and c-Rel are potent activators (Jiang et al., 2002; Suhasini et al., 1997). In most untransformed cells, other than B lymphocytes, NF-κB complexes are sequestered in the cytoplasm bound to specific IκB inhibitory proteins. Several years ago, we and others demonstrated that over 90% of breast cancers aberrantly express constitutive nuclear c-Rel, p50, p52 or RelA NF-κB subunits, which promote resistance to chemotherapy (Cogswell et al., 2000; Nakshatri et al., 1997; Sovak et al., 1997; Wang et al., 1996). Little was known about a role for RelB in breast cancer. RelB was thought to function primarily in regulation of adaptive immune responses, specifically during differentiation of dendritic cells (Burkly et al., 1995; Clark et al., 1999). We first noted nuclear RelB in mouse mammary tumors induced by ectopic c-Rel expression (Romieu-Mourez et al., 2003) and then upon exposure to the carcinogen dimethyl[a]benzanthracene (Demicco et al., 2005). Subsequently, we observed an inverse correlation between constitutive RelB expression and estrogen receptor alpha (ERα) levels in breast cancer tissues and cell lines (Wang et al., 2007). Furthermore, RelB was shown to promote a more invasive phenotype via induction of Bcl-2.
Epidemiological data have demonstrated an association between high levels of serum 25-hydroxyvitamin D [25(OH)D] or vitamin D intake and a decreased risk for development of breast cancer (Bertone-Johnson et al., 2005; Shin et al., 2002). More recently, an inverse correlation between serum levels of 25(OH)D and breast cancer mortality has been observed (Garland et al., 2007). Vitamin D is a group of fat-soluble pro-hormones which includes two major forms, vitamin D2 and vitamin D3 (Hollis, 2008). Hydroxylation of these compounds on carbon 25 of the side chain and then carbon 1 of the A ring results in metabolic activation. In cells in culture, addition of the active 1,25-dihydroxyvitamin D3 [1,25(OH)2D3] form has been implicated in cell cycle arrest, differentiation and apoptosis of breast cancer cells (Eisman et al., 1989; Lazzaro et al., 2000; Narvaez and Welsh, 2001). The mechanism of vitamin D action on breast cancer cells is complex. In ERα positive cells, the anti-proliferative effects of 1,25(OH)2D3 are thought to be at least in part mediated via repression of ERα levels and estrogen-mediated bioresponses (Swami et al., 2000). However, 1,25(OH)2D3 and its analogs are also potent inhibitors of proliferation in ERα negative cells (Elstner et al., 1995), thus indicating additional ERα independent mechanisms of action. Here, we tested the role of RelB in resistance of breast cancer cells to γ-irradiation and to the chemotherapeutic agent doxorubicin. RelB promoted survival of breast cancer cells to these two anti-cancer treatment modalities. Furthermore, 1,25(OH)2D3 inhibited the high RelB levels found in ERα negative breast cancers and thereby decreased their survival. Thus, inhibition of RelB by 1,25(OH)2D3 represents a novel mechanism to explain the association between high levels of serum 25-hydroxyvitamin D [25(OH)D] or vitamin D intake and a decreased risk of breast cancer.
ERα positive MCF7 and ZR-75 cells, and ERα negative Hs578T and MDA-MB-231 cells were maintained as described previously (Wang et al., 2007). The cell line NF639 (kindly provided by P. Leder, Harvard Medical School, Boston MA), which was derived from a mammary gland tumor in a MMTV-Her-2/neu transgenic mouse, was cultured as described previously (Pianetti et al., 2001). MCF7 cell lines containing pcDNA3 or pcDNA3-RelB and Hs578T cell lines containing human pRELB-siRNA or control pRELB sense (si-Control) were established as described (Wang et al., 2007) and grown in the presence of 400 μg/ml G418 (Sigma) or 1 μg/ml puromycin (Sigma), respectively. Where indicated, cells were treated with 10−5 M Doxorubicin (Sigma) or an equivalent volume of water. An initial dose response curve of the effects of Doxorubicin on survival indicated a significant difference between Hs578T-siRelB and Hs578T-Con cells was achieved at 10−5 M (not shown). For 1,25(OH)2D3 treatments, cells were exposed to 10−8 M 1,25(OH)2D3 (a generous gift from Dr. M. Uskokovic, Hofmann LaRoche, Nutley, NJ) or an equivalent volume of ethanol. An initial dose response curve of the effects of 10−12, 10−10, 10−8 and 10−6 M 1,25(OH)2D3 on RelB levels in Hs578T cells indicated effective inhibition was achieved at 10−8 M (not shown).
Cells cultures, at 90% confluency, were transfected with 10 μg of the RelB expression vector pMexNeo-RelB (a kind gift of Rodrigo Bravo, Bristol-Myers Squibb Co.) or the corresponding pMexNeo control empty vector (EV) using Lipofectamine™ 2000 Reagent (Invitrogen, Carlsbad, CA), in Optimem serum-free medium, according to the manufacturer’s protocol. After overnight transfection, Optimem was replaced with fresh complete medium and cells allowed to recover for 8 hours. The resulting NF639-EV and NF639-RelB cells were used immediately in experiments evaluating the effects of 1,25(OH)2D3 and γ-irradiation.
Whole cell, nuclear and cytoplasmic extracts were prepared and immunoblotting performed as previously described (Wang et al., 2007). Antibodies against RelB NF-κB (sc-226) and the pro-survival factors Survivin (sc-10811) and Bcl-2 (sc-492) were purchased from Santa Cruz Biotechnology. Antibodies against Manganese superoxide dismutase (MnSOD; catalog no. 06–984) and β-actin were purchased from Upstate and Sigma, respectively. For quantitation of protein expression, blots from three separate experiments were scanned, subjected to densitometry and values normalized to β-actin. Average fold change ± standard deviation (SD) in protein levels relative to levels in control untreated cells (set at 1) are given.
For clonogenic cell survival studies, cells (200–500 cells/well) were seeded in 6-well plates in triplicate and after overnight incubation, subjected to the indicated doses of ionizing radiation in a 130 kV X-ray machine (Faxitron X-ray Corporation) at a dose rate of 89.7 cGy/min. For 1,25(OH)2D3 experiments, cells were either pre-treated with 1,25(OH)2D3 or a corresponding volume of vehicle ethanol for 24 h, and then irradiated. Cells were allowed to form colonies over a period of 5–10 days after irradiation. Colonies were counted if they contained 50 or more cells. Survival fraction (in log10 scale) for the different radiation doses was calculated as a ratio of the number of colonies in each treatment condition over the total number of colonies in the corresponding untreated control. Values shown are averages of triplicate samples ± SD. Statistical significance between curves was determined using a paired Student’s t-test and for a specific dose between the curves using a Student’s t-test for samples with equal variance. P-values of <0.05 were considered statistically significant.
Cell viability was assessed with an APTlite 1step luminescence ATP detection assay system (Perkin Elmer), performed as recommended by the manufacturer. Briefly, samples (3000 or 8000 cells/100 μl for Hs578T or MCF7 cells, respectively) were plated in triplicate in 96-well plates and treated with the indicated dose of Doxorubicin for 24 h. An equal volume of APTlite 1step luminescence reagent was added to each well and luciferase activity measured in a Luminoskan Ascent 96-well luminometer. Viability (% Survival) was calculated as a ratio of the relative luminescence values for each treatment condition over those in corresponding untreated samples multiplied by 100. Values shown are averages of triplicate samples ± SD. The statistical significance was determined as described above.
For NF639 cell counting experiments, samples (4000 cells/well) were plated in triplicate in 6-well plates. After overnight incubation, cells were treated with the indicated dose of 1,25(OH)2D3 for 24 h, then exposed to 2 Grays of γ-irradiation and allowed to recover for 3 days. Cells numbers were counted in a Beckman Coulter Z2 Counter. Viability (% Survival) was calculated as a ratio of the number of cells in each treatment condition over the number of cells in the corresponding untreated control samples multiplied by 100. Values shown are averages of triplicate samples ± SD. The statistical significance was determined as described above.
RNA was isolated using Trizol reagent (Invitrogen) and quantified by measuring A260. cDNA was prepared using SuperScript III reverse transcriptase (Invitrogen) according to the manufacturer’s protocol. RNA expression for Vdr/VDR, RelB/RELB, Survivin/SURVIVIN, MnSOD/MNSOD, Bcl-2/BCL-2 and glyceraldehyde-3-phosphate dehydrogenase (Gapdh/GAPDH), used as a loading control, was assessed by reverse transcription (RT)-PCR. PCR was performed in a thermal cycler as follows: for mouse Bcl-2, 30 cycles at 95 °C for 30 s, 62 °C for 30 s, and 72 °C for 45 s; for human MNSOD, 24 cycles at 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 45 s; and for mouse Vdr and human VDR, 33 cycles at 95 °C for 30 s, 59 °C for 30 s, and 72 °C for 45 s. The following primer sets were used: for mouse Bcl-2, 5′-TATAAGCTGTCACAGAGGGGCTA-3′ a n d 5′-CATGCTGGGGCCATATAGTT-3′; for mouse Vdr, 5′-CCAGGAATACCAAGGCTGAA-3′ a n d 5′-TGGATAGGCGGTCCTGAATGGC-3′; for human MNSOD, 5′-GCACTAGCAGCATGTTGAGC-3′ and 5′-CAACGCCTCCTGGTACTTCT-3′; and for human VDR 5′-ATGGCCATCTGCATCGTCTC-3′ and 5′-CTTCCTCTTCGGCCTTTTCT-3′. The PCR conditions and primer sets used for mouse RelB, MnSOD, Survivin and Gapdh, and for human RELB, SURVIVIN, BCL2 and GAPDH have been described previously (Mineva et al., 2007; Wang et al., 2007). For quantitation of RNA levels, gels from triplicate experiments were subjected to densitometry and normalized to Gapdh/GAPDH. Average fold change ± SD in mRNA levels relative to those in control untreated cells (set to 1) are presented.
Previously, we demonstrated an inverse correlation between ERα status and RelB levels in breast cancer cells (Wang et al., 2007). ERα positive ZR-75 and MCF7 expressed low RelB levels, while ERα negative Hs578T and MDA-MB-231 and ERα low NF639 breast cancer cells expressed high levels of RelB. To begin to assess the involvement of RelB in resistance of breast cancer cells to γ-irradiation, ZR-75, MCF7, Hs578T and MDA-MB-231 cell cultures were subjected to a gradient dose of γ-radiation followed by a clonogenic survival assay. As seen in Figure 1, high RelB-expressing Hs578T and MDA-MB-231 cells were more resistant to radiation-induced cell death compared to low RelB-expressing ZR-75 and MCF7 cells.
To test directly whether RelB is responsible for the observed resistance to radiation, we used paired stable cultures of Hs578T cells expressing siRNA RELB or control sense RELB RNA (Hs578T-siRELB vs Hs578T-Con) and MCF7 cells ectopically expressing RelB or empty vector control DNA (MCF7-RelB vs MCF7-EV cells), that had been prepared and characterized previously (Wang et al., 2007). Cells were subjected to a gradient dose of γ-irradiation followed by a clonogenic survival assay. Knockdown of RelB in Hs578T-siRELB cells led to increased killing following radiation compared to Hs578T-Con cells (Fig. 2A). Conversely, enforced RelB expression in MCF7-RelB cells led to increased survival upon irradiation as compared to MCF7-EV cells (Fig. 2B). Thus, RelB promotes resistance of breast cancer cells to ionizing radiation.
Doxorubicin (or Adriamycin) is one of the most important drugs available for the treatment of metastatic breast cancer and is highly effective in treatment of invasive HER-2 positive and ERα negative breast cancers (Paik et al., 1998; Paik et al., 2000; Paridaens et al., 2000). Thus, we next tested the ability of RelB to protect Hs578T and MCF7 breast cancer cells from death induced by treatment with doxorubicin. Cultures were treated with 10−5 M doxorubicin or with an equal volume of vehicle water as control for 24 h. Doxorubicin treatment induced a decrease in survival of both Hs578T-Con and Hs578T-siRELB cells (Fig. 3A), and the killing was further enhanced by knockdown of RelB. A moderate, statistically significant increase in the loss of cell viability by doxorubicin treatment was seen in the Hs578T-siRELB vs Hs578T-Con cells (Fig. 3A). Similarly, treatment of MCF7-EV with doxorubicin resulted in robust killing of these cells (with only 29.8% viable cells remaining) (Fig. 4A). RelB overexpression in MCF7 cells profoundly increased survival following doxorubicin treatment (Fig. 4A). We next asked whether doxorubin effects RelB expression, and subjected whole cell extracts from treated cultures to immunoblot analysis for RelB (Fig. 3B). Unexpectedly we observed that doxorubicin treatment profoundly decreased RelB protein levels in the control Hs578T cells, such that they were comparable to those in untreated Hs578T-siRELB. Doxorubicin treatment of Hs578T-siRELB cells further reduced RelB expression to levels slightly lower than treated Hs578T-Con cells (Fig. 3B). Immunoblotting analysis demonstrated that while the ectopic RelB overexpression in MCF7-RelB was substantially reduced by doxorubicin treatment, consistent with a post-transcriptional site of regulation, the levels were substantially above those seen in the control MCF7-EV cells (Fig. 4B). Thus, RelB promotes survival of breast cancer cells to treatment with doxorubicin, which robustly represses the levels of this NF-κB subunit.
Recently, RelB has been implicated in sensitivity of prostate cancer cells to treatment with 1,25(OH)2D3 (Xu et al., 2007). Thus, the effects of 1,25(OH)2D3 on RelB levels were assessed in Hs578T breast cancer cells, and in a second line NF639, which also expresses high levels of RelB (Wang et al., 2007). Cultures were treated with 10−8 M 1,25(OH)2D3 or vehicle ethanol, as control, and total RNA and whole cell extracts isolated. Treatment of Hs578T cells with 1,25(OH)2D3 resulted in a robust decrease in RELB mRNA (Fig. 5A) and in a commensurate reduction in RelB protein levels (Fig. 5B). Similarly, treatment of NF639 cells with 1,25(OH)2D3 decreased RelB RNA and RelB protein levels (Figs. 6A and 6B). When the effects of 1,25(OH)2D3 treatment on RelB pro-survival target genes was examined, a substantial decrease in mRNA and protein expression of Survivin, MnSOD, and Bcl2 was noted in the NF639 cells (Figs. 6A and 6B). Similarly, MNSOD, SURVIVIN and BCL2 gene expression in Hs578T cells was reduced by 1,25(OH)2D3 treatment (Figs. 5A and 5B). The presence of 1,25(OH)2D3 receptor RNA was confirmed in these lines, as well as in the MDA-MB-231, MCF7, and ZR-75 cells (Supplementary Fig. 1), as expected. Thus 1,25(OH)2D3 treatment inhibits functional RelB levels in breast cancer cells.
We next tested whether treatment of these breast cancer cells with 1,25(OH)2D3 enhances their sensitivity to γ-irradiation. Hs578T cells were pre-treated with 10−8 M 1,25(OH)2D3 for 24 h, exposed to a gradient dose of γ-radiation and subjected to a clonogenic survival assay. As seen in Figure 5C, pre-treatment with 1,25(OH)2D3 resulted in a dose-dependent increase in the sensitivity of Hs578T cells to γ-irradiation. Statistically significant differences in sensitivity were seen at doses of 3, 4, 5, and 6 Gray. Similarly, NF639 cells displayed an increased sensitivity to γ-irradiation following pre-treatment with 1,25(OH)2D3 (Fig. 6C). Thus, 1,25(OH)2D3 treatment enhances sensitivity to radiation treatment.
To confirm that the increased sensitivity to γ-irradiation mediated by 1,25(OH)2D3 was due to inhibition of RelB, NF639 cells were transiently transfected with the pMexNeo-RelB vector (NF639-RelB) or the corresponding control EV (NF639-EV) DNA. Immunoblot analysis of whole cell extracts confirmed that NF639-RelB cells expressed higher levels of RelB as compared to NF639-EV cells (Fig. 7A). The NF639-RelB and NF639-EV cells were either left untreated, pre-treated with 10−8 M 1,25(OH)2D3 for 24 h, exposed to 2 Grays of γ-irradiation, or a combination of both and allowed to recover for 3 days and RelB protein expression and cell survival assessed. Immunoblot analysis of whole cell extracts isolated following these treatments revealed that exposure to 1,25(OH)2D3 alone led to substantial decreases in RelB levels in NF639-RelB vs NF639-EV cells (Fig. 7B). In the NF639-EV cells, the effects of irradiation alone were more modest and the combination appeared similar to 1,25(OH)2D3 alone. In the NF639-RelB, irradiation did not result in a decrease in RelB levels and RelB levels in the cells given the combination of irradiation and 1,25(OH)2D3 were maintained at almost twice that of the 1,25(OH)2D3 alone (Fig. 7B). The effects of the treatments with 1,25(OH)2D3 and irradiation on survival of the NF639-RelB and NF639-EV cultures were then assessed. NF639-RelB cells were found to be significantly less sensitive to treatment with a combination of 1,25(OH)2D3 and γ-irradiation or to γ-irradiation alone as compared to control NF639-EV cells, consistent with retention of higher RelB levels during these treatments (Fig. 7C). Furthermore, treatment with 1,25(OH)2D3 alone decreased cell survival equally in NF639-RelB and NF639-EV cells, consistent with a lack of difference in RelB levels seen above (Fig. 7B). Thus, ectopic RelB impairs the killing by a combination of 1,25(OH)2D3 and radiation treatments in these breast cancer cells.
Here, RelB was shown to promote survival of breast cancer cells to γ-irradiation and doxorubicin, two commonly used therapeutic treatments. Furthermore, exposure to 1,25(OH)2D3 was found to repress RELB gene expression and thereby to increase sensitivity to irradiation. The ability of RelB to promote survival can be attributed to its ability to induce the anti-apoptotic proteins Survivin, MnSOD, and Bcl-2. In particular, we showed that the reduction in levels of RelB by 1,25(OH)2D3 treatment, sensitized the ERα negative Hs578T cells and Her-2/neu-driven NF639 cells, which are ERα low, to γ-irradiation treatment. An adjuvant effect of 1,25(OH)2D3 and analogs on doxorubicin and irradiation has been reported for ERα positive breast cancer cells (Chaudhry et al., 2001). To our knowledge, this is the first report of similar effects of 1,25(OH)2D3 in the response of ERα negative or Her-2/neu-driven breast cancer cells to irradiation. Recent work suggests a similar scenario involving RelB and 1,25(OH)2D3 occurs in prostate cancer. Induction of MnSOD by RelB played an important radioprotective role in aggressive PC-3 prostate cancer cells (Josson et al., 2006). RelB was identified as a potential prognostic marker in prostate cancer, as nuclear RelB levels statistically correlated with high Gleason score (Lessard et al., 2005). More recently, Xu and co-workers demonstrated a drop in RelB levels and enhanced radiosensitivity of aggressive prostate cancer cells following treatment with 1,25(OH)2D3 (Xu et al., 2007). Overall, these results suggest a significant role for repression of RelB in the observed correlation of serum 25(OH)D levels and decreased risk of breast cancer.
Previously, we demonstrated that RelB promotes a more invasive phenotype of breast cancer cells (Wang et al., 2007). Consistent with our findings, Kattan and coworkers recently demonstrated that suppression of MnSOD in breast cancer cells results in a decrease in colony forming ability and invasive properties (Kattan et al., 2007). Bcl-2 and Survivin have been associated in breast cancer patients with a low probability of response to chemotherapy and poor prognosis, respectively (Bonetti et al., 1998; Span et al., 2004). Reduced apoptotic indices and poorer survival rates were noted in breast cancer cases that were positive for both Survivin and Bcl-2 (Tanaka et al., 2000). These findings suggest Survivin and Bcl-2 may cooperate to promote breast cancer. Of note, in our current study, RelB expression led to concurrent activation of MnSOD, Bcl-2 and Survivin and resistance to chemotherapy and radiation. Preliminary data from microarray analysis (data not shown) suggest that RelB regulates the expression of a larger set of genes involved in control of survival. However, further studies are necessary to elucidate their specific roles in RelB-mediated breast cancer cell survival and response to therapy.
Doxorubicin is a common anticancer treatment, which has been found to function via several cytostatic and cytotoxic mechanisms of action. These include (1) intercalation into DNA, which results in inhibition of macromolecular biosynthesis; (2) free radical formation; and (3) inhibition of topoisomerase II, which leads to induction of DNA damage (Gewirtz, 1999). While early studies showed that doxorubicin, and other topoisomerase inhibitors increase nuclear NF-κB levels (Bottero et al., 2001; Wang et al., 1996), these have more recently been shown to be functionally inactive by Ho and coworkers (2005). Specifically, classical NF-κB p50/p65 binding complexes induced by doxorubicin treatment in breast cancer cells are inactive due to lack of post-transcriptional modifications, resulting in decreased NF-κB target gene expression (Ho et al., 2005). In the present study, doxorubicin was shown to act as a potent RelB inhibitor. The decrease in classical NF-κB activity may contribute to the reduced RelB levels, as its promoter has two functional NF-κB elements (Bren et al., 2001; Wang and Sonenshein, 2005). Consistent with the ability of doxorubicin to inhibit RelB levels, no additive effects of 1,25(OH)2D3 and doxorubicin on RelB levels or survival were observed in NF639 or Hs578T cells (data not shown). Doxorubicin is a standard treatment for advanced breast cancers and has been found particularly effective against aggressive Her-2 positive and ERα negative subtypes (Gianni et al., 2005; Kuerer et al., 1999; Mortimer et al., 1985; Paik et al., 1998; Paik et al., 2000). Importantly, our previous studies demonstrated that these aggressive cancers express high levels of RelB. Overall, these findings suggest that doxorubicin is an effective treatment against these breast cancers, in part, because of its ability to reduce their high RelB levels, and therefore survival of these cells.
In summary, our data demonstrate that the RelB NF-κB subunit promotes survival of breast cancer cells and that 1,25(OH)2D3 reduces RelB levels and therefore survival. Together these observations provide one mechanism to explain the recently reported association between decreased breast cancer risk and mortality and serum levels of 25(OH)D (Bertone-Johnson et al., 2005; Garland et al., 2007). Our studies suggest that further analysis of the potential beneficial role of 1,25(OH)2D3 as an adjuvant in radiation treatment of breast cancers with high RelB levels is warranted.
Contract grant sponsor: NIH;
Contract grant numbers: R01 CA129129 and P01 ES11624.
Contract grant sponsor: AICR (with funds from the Derx Foundation);
Contract grant number: 07A-151.
GES is supported by grants R01 CA129129 and P01 ES11624. NDM is supported by a fellowship from the AICR (with funds from the Derx Foundation). We thank Tai Cheng Chen for his providing reagents.