Utility of siRNA against Keap1 as a strategy to stimulate a cancer chemopreventive phenotype

Abstract

A duplex 21 nucleotide small interfering RNA (siRNA) against human Keap1 is described that represents a unique class of cancer chemopreventive agent. This siRNA can knockdown Keap1 mRNA and thereby relieve negative regulation of nuclear factor erythroid 2 p45-related factor 2 (Nrf2)-mediated gene expression. The siRNA lowered endogenous Keap1 mRNA to <30% of control levels between 24 and 72 h after transfection in human HaCaT keratinocyte cells and was capable of blocking ectopic expression of FLAG-tagged human Keap1 protein but not that of ectopic V5-tagged mouse Keap1 protein. Transfection of human HaCaT cells with Keap1 siRNA markedly enhanced endogenous levels of nuclear factor erythroid 2 p45-related factor 2 (Nrf2) protein and increased transcription of an antioxidant response element-driven reporter gene by 2.3-fold. Furthermore, 48 h after transfection of these cells with Keap1 siRNA, expression of aldo-keto reductase 1C1/2 and the glutamate cysteine ligase catalytic and modifier subunits was elevated between 5- and 14-fold. A modest increase of 3-fold in NAD(P)H:quinone oxidoreductase 1 was also observed. The Keap1 siRNA produced a 1.75-fold increase in intracellular glutathione 48 h after transfection. Thus, antagonism of Keap1 by siRNA can be used to preadapt human cells to oxidative stress without the need to expose them to redox stressors.

Cancer chemoprevention holds the promise of greatly reducing the incidence of neoplastic disease. Many chemopreventive mechanisms have been proposed, including induction of detoxication enzymes, stimulation of apoptosis, promotion of cellular differentiation, and inhibition of inflammation, hormone receptors, angiogenesis, and tissue invasion (1-5). Chemicals that act by increasing detoxication pathways are called chemopreventive blocking agents (1). These chemicals not only up-regulate the expression of genes encoding drug-metabolizing enzymes, such as NAD(P)H:quinone oxidoreductase 1 (NQO1) and glutathione (GSH) transferases, but they also induce many antioxidant genes, including those for glutamate cysteine ligase catalytic (GCLC) and glutamate cysteine ligase modifier (GCLM) subunits that are involved in GSH biosynthesis (6-8).

All chemopreventive blocking agents are thiol-active (9), and their ability to induce gene expression represents a mechanism of cellular adaptation to redox stress (10). Genes that are transcriptionally activated in this fashion contain antioxidant response elements (AREs) in their 5′ upstream regions (10, 11). After treatment of cells with blocking agents, the Cap'n'Collar basic-region leucine zipper transcription factor nuclear factor erythroid 2 p45-related factor 2 (Nrf2) accumulates in the nucleus, where it is recruited as a heterodimer with a small Maf basic-region leucine zipper protein to AREs and stimulates expression of cytoprotective genes (12, 13).

Central to induction of ARE-driven genes is antagonism of Keap1, a protein that negatively regulates Nrf2 by facilitating its proteasomal degradation (14, 15). Keap1 is a redox-sensitive protein that contains a Broad complex, Tramtrack, and Bric-a-Brac/Poxvirus and zinc finger (BTB/POZ) domain in its N-terminal portion, six Kelch repeats in its C-terminal half, and reactive cysteine residues located centrally between the BTB/POZ and Kelch repeat domains (16-18). Humans possess 51 proteins containing both BTB/POZ and Kelch repeat domains; Keap1 is a member of subclass 1 (19). Keap1 controls the half-life of Nrf2 by acting as a substrate adaptor protein for the cullin 3-Roc1 ubiquitin E3 ligase (20-22). Under homeostatic conditions, Keap1 promotes degradation of Nrf2 via the 26S proteasome, but this activity is lost during redox stress. Chemopreventive blocking agents inhibit the interaction between Keap1 and Nrf2, which results in accumulation of the transcription factor (14, 15).

Two principal models have been proposed to explain how Nrf2 protein is stabilized during induction of ARE-containing genes. In one of these models, Keap1 is modified by blocking agents at Cys-273 and Cys-288, which prevents it from capturing Nrf2 and targeting it for degradation (17). In the other, Nrf2 is phosphorylated by PKC or PKR-like endoplasmic reticulum-resident kinase (PERK) in response to inducing agents, thereby enabling the basic-region leucine zipper protein to either evade Keap1 or dissociate from Keap1 (23-27).

Small interfering RNA (siRNA) can activate RNA interference (28). We hypothesized that it should be possible to use siRNA against Keap1 to activate the ARE gene battery provided that Keap1 protein is relatively short-lived and Nrf2 accumulation is sufficient for gene induction. In the present paper, we report the design of two siRNA species that can knockdown Keap1 and up-regulate cytoprotective genes in human cells.

Materials and Methods

Preparation of siRNA Against Keap1. siRNA were designed by Cenix Bioscience (Dresden, Germany) according to the current guidelines for effective knockdown by this method. The two double-stranded Keap1-targeting siRNAs and the scrambled control with the same GC content were as follows: Target1 for Keap1, 5′-GGCCUUUGGCAUCAUGAACTT-3′ (sense) and 5′-GUUCAUGAUGCCAAAGGCCTG-3′ (antisense); Target2 for Keap1, 5′-GGUCAAGUACCAGGAUGCATT-3′ (sense) and 5′-UGCAUCCUGGUACUUGACCTG-3′ (antisense); and scrambled siRNA, 5′-GACGAGCGGCACGUGCACATT-3′ (sense) and 5′-UGUGCACGUGCCGCUCGUCTT-3′ (antisense). All siRNAs were provided in a freeze-dried, preannealed, HPLC-purified form by Cenix Bioscience, resuspended as a stock solution at a concentration of 100 μmol/liter, and stored at -20°C until required.

Plasmids. A human Keap1 expression construct tagged with an N-terminal FLAG epitope, pCMV-Tag/Keap1 (29), was a gift from Michael L. Freeman (Vanderbilt University, Nashville, TN). The mouse Keap1 expression construct containing a C-terminal V5 epitope tag, pcDNA3.1/V5HisCmKeap1 (14); the mouse nqo1-ARE (nucleotides -454 to -414) cloned into the pGL3-Promoter luciferase reporter vector; and the pRL-TK Renilla luciferase vector (Promega) are described in ref. 13.

Cell Culture and Transfection. Human keratinocyte HaCaT cells were maintained in DMEM supplemented with 10% FBS and 2 mM l-glutamine (Invitrogen). Cells were routinely passaged at 100% confluence by first being rinsed in 0.02% EDTA in PBS (free of Ca²⁺ and Mg²⁺), followed by a 30-min incubation in the same solution. Thereafter, 0.25% (wt/vol) trypsin was added, and digestion was allowed to proceed for 5 min before the single-cell suspension was centrifuged, resuspended in complete media, and split 1:5. Media was replenished every second day.

The cells were seeded in six-well dishes at a density of 2.5 × 10⁵ cells 24 h before each experiment. Transfections were performed with cells at 30-40% confluence by using Lipofectamine 2000 (Invitrogen). When HaCaT cells were transfected with siRNA alone, the two mixtures required for transfection were prepared separately. The first mixture consisted of siRNA oligos dissolved in 250 μl of OptiMEM reagent (Invitrogen); the second mixture consisted of 5 μl of Lipofectamine 2000 in 250 μl of OptiMEM reagent. After 5 min of incubation at 37°C, the two mixtures were combined and incubated for a further 30 min before being applied to cells in a solution of complete media that had a total volume of 2 ml. When HaCaT cells were transfected with siRNA and DNA plasmid vectors, cotransfections were performed by using siRNA and DNA mixtures that were prepared separately, each in 125 μl of OptiMEM reagent. On such occasions, a 10-μl aliquot of Lipofectamine 2000 was added to 250 μl of OptiMEM reagent and incubated for 5 min before 125-μl portions from this solution were added to each of the siRNA and DNA mixes. After a further 30-min incubation, the siRNA- and DNA-containing mixtures were combined and added to the HaCaT cells to a total volume of 2 ml in complete media.

Biochemical Analyses. ARE-driven luciferase activity was assayed as reported in ref. 13. Total cellular RNA was extracted by using TRIzol reagent (Invitrogen). The relative levels of mRNA were determined by TaqMan chemistry (30), having first reverse-transcribed the RNA by using Omniscript (Qiagen, Valencia, CA). The sequences of the primers and probes used are listed in Table 1. The primers for measuring aldo-keto reductase (AKR)1C1 mRNA contain one, four, and six mismatches with equivalent regions in RNAs for AKR1C2, AKR1C3, and AKR1C4, respectively; because of concerns of cross hybridization between AKR1C1 and AKR1C2, the data generated for the TaqMan assay are referred to as AKR1C1/2.

Table 1. Oligonucleotide primers and probes for TaqMan analyses.

mRNA	Function*	Sequence
AKR1C1	Forward	5′-CGAGAAGAACCATGGGTGGA-3′
	Reverse	5′-GGCACAAAGGACTGGGTCC-3′
	Probe	5′-CCAAGAGCACCGGGGAGTTCGG-3′
GCLC	Forward	5′-TCTCTAATAAAGAGATGAGCAACATGC-3′
	Reverse	5′-TTGACGATAGATAAAGAGATCTACGAA-3′
	Probe	5′-CAGGAGATGATCAATGCCTTCCTGCAAC-3′
GCLM	Forward	5′-TAGAATCAAACTCTTCATCATCAACTAGAA-3′
	Reverse	5′-TCACAGAATCCAGCTGTGCAA-3′
	Probe	5′-TGCAGTTGACATGGCCTGTTCAGTCC-3′
Keap1	Forward	5′-TTCAAGGCCATGTTCACCAA-3′
	Reverse	5′-TGGATACCCTCAATGGACACC-3′
	Probe	5′-TGCGGGAGCAGGGCATGGA-3′
NQO1	Forward	5′-GGAGAGTTTGCTTACACTTACGC-3′
	Reverse	5′-AGTGGTGATGGAAAGCACTGCCTTC-3′
	Probe	5′-CCATGTATGACAAAGGACCCTTCCGGAG-3′

^*Primers are designated simply as forward or reverse.

Protein and GSH levels were measured as described in ref. 31.

For Western blotting, protein extracts were prepared by harvesting cells in RIPA buffer supplemented with Complete EDTA-free protease inhibitor mixture (Roche Diagnostics). Immunoblotting was performed by using antiserum raised against GCLC, GCLM, AKR1C1, NQO1, Nrf2, FLAG (Sigma), V5 (Invitrogen), or GAPDH (Ambion, Austin, TX) as described in refs. 6 and 32.

Results

Based on data from the keap1^-/- mouse (33), it seemed possible that knockdown of Keap1 by siRNA in human cells may upregulate ARE-driven genes. HaCaT keratinocytes were chosen to examine this possibility because NQO1 is induced in this cell line by treatment with sulforaphane (Sul) (34). Initial experiments indicated that transfection of HaCaT cells with FITC-tagged oligonucleotides resulted in 60-70% of cells staining positively under fluorescence microscopy. This level of transfection efficiency was considered adequate to allow the effects of Keap1 siRNA to be evaluated.

Knockdown of Keap1 mRNA and Protein. Two siRNA sequences, Target1 and Target2, were designed against human Keap1. Each sequence contained a TT overhang at the 3′ end of the sense strand and a TG overhang at the 3′ end of the antisense strand. Target1 and Target2 encompass nucleotides +186 to +204 and +246 to +264, respectively, of Keap1 cDNA.§

Target1 and Target2 siRNAs were separately transfected into HaCaT cells at a dose of 100 nmol/liter. After 48 h, Target1 reduced the steady-state levels of mRNA for Keap1 to ≈30% of that observed in untreated HaCaT cells or those transfected with the scrambled siRNA (Fig. 1A). Target2 reduced Keap1 mRNA to ≈50%. Given a transfection efficiency of 60-70%, Target1 appeared to completely abolish Keap1 mRNA in transfected cells. Because Target1 appeared to be slightly more effective than Target2 in reducing Keap1 mRNA, the former siRNA was studied further.

Fig. 1.

Knockdown of Keap1 with siRNA. HaCaT cells were transfected with Keap1 siRNA and scrambled siRNA. Levels of Keap1 mRNA were quantified by RT-PCR TaqMan using 18S rRNA as an internal standard. Each experiment was performed in triplicate, and data are presented as the mean relative transcriptional units (RTu) ± SE. (A) HaCaT cells were transfected with 100 nM concentrations each of Target1, Target2, or scrambled siRNA 48 h before RNA was isolated and analyzed. (B) Different concentrations (Conc) of Target1 were used to transfect HaCaT cells 48 h before RNA was isolated and analyzed. (C) Cells were transfected with 100 nM Target1, and RNA was analyzed 24, 48, and 72 h after transfection.

Transfection of different amounts of siRNA into HaCaT cells showed that knockdown of Keap1 mRNA occurred at a dose of only 25 nmol/liter Target1 but that a greater reduction was observed with 100 nmol/liter (Fig. 1B). Study of the time-dependency of transient transfection with siRNA revealed that 24 h after transfection with 100 nmol/liter Target1, the level of Keap1 mRNA in HaCaT cells was reduced to ≈30% (Fig. 1C). Knockdown of Keap1 was maintained up to 72 h.

To determine whether the reduction in steady-state levels of Keap1 mRNA translated into a loss of Keap1 protein, Western blotting experiments were conducted. Attempts to detect endogenous Keap1 protein in HaCaT cells met with variable success because of a lack of suitable antibodies. Hence, the effect of siRNA on ectopically expressed FLAG-epitope tagged human Keap1 was investigated. We first demonstrated that N-terminally tagged Keap1 protein produced from pCMVFLAG/hKeap1 was functional in HaCaT cells insofar as it could inhibit ARE-driven luciferase activity (Fig. 2A). This Keap1-mediated inhibition of ARE-reporter activity under homeostatic conditions is probably due to capture of Nrf2 that would normally be nuclear because of constitutive redox stress. Subsequently, Western blotting for FLAG-tagged Keap1 showed that synthesis of the ectopic protein was lowered to below the detection limit by transfection with Target1, whereas the scrambled siRNA did not inhibit synthesis of tagged Keap1 (Fig. 2B).

Fig. 2.

Depletion of ectopically expressed Keap1 protein by siRNA. (A) HaCaT cells were transfected with 0.4 μg of the human pCMV-Tag/Keap1 expression construct or 0.4 μg of pcDNA3.1/V5HisCmKeap1 expression construct, along with 0.2 μg of mouse nqo1-ARE reporter vector and 0.2 μg of the control reporter plasmid, pRL. Cells were harvested 48 h after transfection, and luciferase activity was measured. Data [expressed in relative luciferase units (RLU)] are presented as the mean of two experiments ± SE. (B) HaCaT cells were transfected with 0.4 μg of human pCMV-Tag/Keap1 or 0.4 μg of mouse pcDNA3.1/V5HisCmKeap1 along with 100 nM Target1 or scrambled siRNA. Cells were harvested after 48 h, and lysates (10 μg of protein) were analyzed by Western blotting using antisera raised against the FLAG or V5 epitope tags for the hKeap1-FLAG- or mKeap1-V5-transfected cells, respectively.

The specificity of the knockdown was investigated further by using a V5-tagged mouse Keap1 expression vector. Within the region that Target1 was designed against, two mismatches occur between the human 5′-GGCCTTTGGCATCATGAAC-3′ and mouse 5′-GGCTTTTGGCGTCATGAAC-3′ sequences (with differences in italics). It was therefore expected that if the RNA interference process was specific, Target1 would not inhibit the ectopic expression of mouse Keap1 in HaCaT cells. As shown in Fig. 2B, this prediction was confirmed. In rescue experiments, in which either human Keap1 or mouse Keap1 was cotransfected with nqo1-ARE into HaCaT cells, treatment with Target1 prevented human Keap1 but not mouse Keap1 from repressing luciferase reporter activity (Fig. 6, which is published as supporting information on the PNAS web site).

Transfection with Keap1 siRNA Increases Intracellular Nrf2 Protein and ARE-Driven Reporter Gene Activity. Immunoblotting was performed to determine whether knockdown of Keap1 is sufficient to increase the intracellular level of Nrf2. Transfection of HaCaT cells with Target1 caused a substantial increase in endogenous Nrf2 protein (Fig. 3A). The increase in Nrf2 was apparent 24 h after transfection with Target1. In these experiments, treatment with Sul was used as a positive control. The immunoreactive Nrf2 polypeptide observed after transfection with Keap1 siRNA comigrated with that observed after treatment with 5 μM Sul.

Fig. 3.

Knockdown of Keap1 increases Nrf2 protein and induces ARE-driven gene expression. (A) HaCaT cells were transfected with 100 nM Target1 and were harvested at various times thereafter. Cell lysates (10 μg of protein) were analyzed in duplicate by Western blotting using antiserum raised against mouse Nrf2. (B) HaCaT cells were transfected with 100 nM Target1 or scrambled siRNA. Cotransfections were performed with 0.2 μg of the nqo1-ARE reporter vector and 0.2 μg of pRL. Cells were harvested 48 h later, and luciferase activity was measured. Data [expressed in relative luciferase units (RLU)] are presented as the mean of duplicate experiments ± SE.

To determine whether the increase in Nrf2 protein elicited by the siRNA was sufficient to stimulate an increase in ARE-driven transcription, reporter gene assays were carried out. Cotransfection of HaCaT cells with the nqo1-ARE vector and either Target1 or scrambled siRNA showed a 2.3-fold increase in luciferase reporter gene activity in those cells exposed to Target1 but not in cells treated with scrambled siRNA (Fig. 3B).

Induction of Antioxidant and Detoxication Genes by Keap1 siRNA. We next explored whether knockdown of Keap1 allows Nrf2 to activate transcription of endogenous ARE-driven genes. The target genes examined included AKR1C1, which is inducible by Sul and other chemopreventive agents (35, 36), as well as NQO1, GCLC, and GCLM because they contain AREs (37-39).

HaCaT cells were either mock-transfected, transfected with Target1, or treated with Sul. TaqMan chemistry and Western blotting showed that by comparison with mock transfection, the Target1 siRNA produced a 14-fold increase in AKR1C1/2 mRNA and a 5-fold increase in AKR1C protein (Fig. 4 A and B). Transfection with scrambled siRNA did not influence levels of AKR1C1/2 mRNA or AKR1C protein. The amount of AKR1C protein in mock-transfected HaCaT cells appeared to be slightly elevated when compared with the level of AKR1C in DMSO-treated HaCaT cells. This result may be due to the ability of DMSO to scavenge reactive oxygen species, which in turn causes a repression of homeostatic gene expression.

Fig. 4.

Keap1 siRNA causes induction of antioxidant and detoxication proteins. HaCaT cells were transfected with either Lipofectamine alone (Mock), 100 nM Target1, or 100 nM scrambled (Scrm) siRNA and then harvested 48 h later. In addition, nontransfected HaCaT cells were treated with 0.01% DMSO or 5 μM Sul (in DMSO to a final concentration of 0.01%) before being harvested 24 h later. (A) Western blotting was performed on duplicate 10-μg protein portions of cell lysates. (B-D) TaqMan chemistry was performed on total RNA to quantify message levels for AKR1C1/2 (B), GCLC (C), and NQO1 (D); the data are presented as relative transcriptional units (RTu), and each is the mean of duplicate experiments ± SE.

Transfection with Target1 and treatment with Sul produced increases of between 5- and 7-fold in mRNA for GCLC (Fig. 4C). Western blotting showed that Target1 produced similar increases in GCLC protein (Fig. 4A). Likewise, Target1 and Sul induced GCLM between 2.5- and 3.5-fold (Fig. 4A, only Western blot data shown). By contrast, neither mock transfection nor transfection with scrambled siRNA or treatment with DMSO produced increases in these proteins. A modest maximal 3-fold induction in mRNA and protein levels of NQO1 in HaCaT cells was observed after treatment with Target1 (Fig. 4 A and D).

Keap1 siRNA Can Augment Intracellular GSH Levels. The hypothesis that increases in GCLC and GCLM protein produced by Target1 would result in an elevation in intracellular GSH was tested. Transfection of HaCaT cells with Target1 siRNA increased total GSH levels in a time-dependent fashion. At 24 h after transfection, the level of GSH was increased by ≈20% (data not shown). At 48 h after transfection, GSH was increased by 75% (Fig. 5). Mock transfection or transfection with scrambled siRNA did not alter the concentration of GSH (Fig. 5). The increase in GSH achieved by Target1 siRNA is similar to that produced after treatment with Sul (data not shown).

Fig. 5.

Keap1 siRNA can elevate intracellular GSH levels. HaCaT cells were either mock-transfected or transfected with 100 nM Target1 or scrambled siRNA. The cells were harvested 48 h later, and total GSH levels were measured. Data are presented as the mean of duplicate experiments ± SE.

Discussion

We have shown that human Keap1 is amenable to knockdown by using RNA interference. Two siRNA species have been designed that reduce human Keap1 mRNA levels. One of these siRNA, Target1, has been shown to increase Nrf2 protein in HaCaT keratinocyte cells, resulting in up-regulation of endogenous human antioxidant genes and an elevation in intracellular GSH. While the present study was being undertaken, two groups reported using plasmid-delivered short hairpin RNA against mouse Keap1 in 293T cells to demonstrate that Keap1 functions as an adaptor protein for the ubiquitination of Nrf2 by the Cul3-Roc1 ligase (21, 22). In neither instance, however, was the effect of RNA interference on the expression of endogenous target genes investigated.

By contrast with conventional cancer chemopreventive blocking agents, the use of siRNA has the advantage that it should specifically target the Keap1-Nrf2 pathway. Blocking agents can have pleiotropic effects on cells and are likely to modulate other pathways besides Keap1-Nrf2 (2, 5). Of potential concern is the fact that all blocking agents are thiol-active redox stressors (9) and can presumably modify intracellular proteins other than Keap1. In the case of Sul, it is expected that thiocarbamylation by the isothiocyanate and/or oxidation of cysteine residues in essential proteins will be harmful to the cell. Whereas low doses of Sul induce ARE-driven gene expression, high doses of the isothiocyanate cause cell cycle arrest and apoptosis (40). It seems unlikely that siRNA against Keap1 will suffer from the same problems of toxicity as can occur after exposure to isothiocyanates. Certainly, by using 100 nmol/liter Target1, we have found no evidence from flow cytometry that it causes apoptosis or cell cycle arrest (data not shown).

The ability of siRNA reagents to interfere only with the intended target is clearly important because lack of specificity can give rise to spurious results (41). We believe that Target1 siRNA is highly selective in its action based on the observation that it can block ectopic expression of human Keap1 but is unable to block ectopic expression of mouse Keap1. Furthermore, in rescue experiments, Target1 could not prevent mouse Keap1, heterologously expressed in human HaCaT cells, from inhibiting ARE-driven luciferase activity. Comparison of cDNAs for human and mouse Keap1 reveal that nucleotides +186 to +204 contain mismatches at +189 and +196, suggesting that failure of the Target1 siRNA to inhibit expression and function of the murine protein is due to loss of homology in the mRNA. We found by searching the human DNA database using blast that the most homologous cDNA sequence to Target1 contained four mismatches. It therefore seems unlikely that Target1 siRNA elicits off-target effects.

The fact that Target1 siRNA could induce ARE-containing genes in the absence of stress suggests that accumulation of Nrf2 protein is necessary and sufficient for activation of the gene battery. As induction of AKR1C1/2 by Target1 siRNA was quantitatively similar to that obtained by using Sul, it appears unlikely that a regulated limiting step exists downstream of Keap1. These results are consistent with the observed up-regulation of NQO1, GST, GCLC, and GCLM in the keap1^-/- mouse (33). However, our findings and those from the keap1^-/- mouse are surprising given the reports that release of Nrf2 from Keap1 and/or nuclear translocation of Nrf2 are controlled during oxidative stress by PKC (24, 25) and during endoplasmic reticulum stress by PKR-like endoplasmic reticulum-resident kinase (PERK) (27). We conclude that phosphorylation of Nrf2 is primarily involved in antagonizing the interaction between the basic-region leucine zipper protein and Keap1 during stress and is not essential for nuclear import of the transcription factor or its ability per se to transactivate ARE-driven gene expression. Furthermore, mitogen-activated protein kinases (42), c-Jun N-terminal kinase 1 (43), and phosphatidylinositol 3-kinase (44) have been reported to up-regulate AR E-driven gene expression in an Nrf2-dependent fashion. Our results suggest that activation of these kinases is not necessary for induction of the ARE-gene battery.

Our current data provide evidence that down-regulation of Keap1 in human cells can alter expression of endogenous ARE-driven genes. This advance is significant because it will allow the battery of genes regulated by Nrf2 to be defined in this species. The question of identifying genes regulated by the Keap1-Nrf2 pathway in humans is an important issue because species-specific differences exist in inducible proteins. For example, AKR1C1/2 is highly inducible in human cell lines (35, 36), yet the homologous dihydrodiol dehydrogenase genes in rats and mice appear to be relatively unresponsive to xenobiotics. Conversely, GSH transferases do not appear to be particularly inducible in human cell lines, yet they are highly inducible in rodents (45).

The human AKR1C dihydrodiol dehydrogenases share 80-98% sequence identity (46). It is unlikely that our antisera distinguish between AKR1C1, AKR1C2, and AKR1C3, although they show minimal crossreactivity with AKR1C4 (32). It is probable that our TaqMan assay for AKR1C1 also measures mRNA for AKR1C2 because there is only one mismatch in the region covered by the forward primer and no mismatches in the reverse primer. It is however unlikely that the AKR1C1 TaqMan assay measures mRNA for either AKR1C3 or AKR1C4 because the forward and reverse primers possess four and six mismatches with the corresponding regions of AKR1C3 and AKR1C4, respectively. Previously, RNase protection assays have been used by Burczynski et al. (35) to show that AKR1C1 is the principal human dihydrodiol dehydrogenase up-regulated by tert-butylhydroquinone, a prototypic monofunctional inducer that works through the ARE (47, 48). It is therefore highly likely that Target1 induces AKR1C1 but not AKR1C2. Consistent with the notion that AKR1C1 is highly inducible, we have found by bioinformatic searching that the 15-kb 5′ upstream region of AKR1C1 contains seven sequences that conform to the “core” ARE consensus, 5′-TGACnnnGC-3′ (data not shown). It will be important to demonstrate which of these putative AREs are functional and whether they can recruit Nrf2.

An interesting observation from this study is that treatment of HaCaT cells with Target1 siRNA resulted in a 1.75-fold increase in total GSH levels. Our results suggest that this increase is achieved by up-regulation of glutamate cysteine ligase. However, the cysteine/glutamate exchange transporter gene in mouse is regulated through an ARE (49), and it may also contribute to the increase in GSH because it is responsible for the uptake of cysteine into the cell (50).

Concluding Comments

Knockdown of Keap1 is a potentially valuable strategy to allow preadaptation to electrophile and oxidative stress in human cells. The present study provides proof of principle that Keap1 siRNA can up-regulate ARE-containing genes in human and may therefore be of therapeutic value in combating degenerative disease.