¹Department of Dermatology, University of California, Irvine, Irvine, California, USA

Correspondence: Dr Elizabeth L. Rugg, Department of Dermatology, C340 Medical Sciences 1, University of California, Irvine, Irvine, California 92697. E-mail: erugg@uci.edu

Abstract

The identification of mutations in keratin genes as the cause of several inherited skin disorders raised the possibility that molecular-based therapies might be developed to treat these conditions. In this issue, Smith et al. (2007) have identified small interfering RNAs that specifically and potently silence keratin 6a expression. These molecules have great promise as therapeutic agents for the treatment of pachyonychia congenita.

RNA interference (RNAi) is a highly evolutionally conserved process that is used by cells to selectively regulate and silence gene expression. It is widely used as an experimental tool, and it is increasingly apparent that RNAi also has great potential as a therapeutic strategy for treating many human disorders, including genetic diseases, viral infections, and cancer (Ryther et al., 2005; Akhtar and Benter, 2007). Synthetic small inhibitory RNAs (siRNAs) can be introduced into the cell, where they are incorporated into the RNA-induced silencing complex (RISC), leading to degradation of mRNA with sequence complementary to the siRNA (Kim and Rossi, 2007). Clinical trials are already in progress to evaluate siRNA for the treatment of macular degeneration and lung disease. Although these trials are not complete, early reports suggest that the siRNAs are effective and without side effects (Akhtar and Benter, 2007).

Keratin disorders are a group of inherited conditions caused by mutations in genes that encode proteins of the keratin intermediate filament cytoskeleton. Most of these conditions result from single-nucleotide substitutions in one copy of a keratin gene, leading to amino acid changes that affect keratin filament formation. This presents a challenge for the development of therapies for keratin disorders because they must reduce or eliminate the effects of the mutant protein. Currently, RNAi is the most promising strategy for specifically knocking down expression of disease-causing genes. One of the first steps in drug development is the identification of potent molecules with target specificity and lack of toxicity. Smith et al. (2007, this issue) identified siRNAs that are active in the picomolar range, which brings a step closer the prospect of a therapy for one keratin disorder, pachyonychia congenita (PC).

PC is an autosomal dominant condition characterized by palmoplantar keratoderma and nail dystrophy, which may be accompanied by epidermal cysts and oral lesions. For patients the most painful feature of this condition is plantar keratoderma (Leachman et al., 2005). PC is caused by mutations in keratins K6a, K6b, K16, and K17. Rather than target siRNA to a specific gene mutation, Smith and colleagues (2007) exploited the overlapping expression of a pattern of keratins K6a and K6b and targeted siRNAs to the 3 UTR of the keratin 6a gene. This is a very attractive strategy because a single siRNA can be used to ablate K6a expression irrespective of mutation and gets around the need to develop mutation-specific molecules. Loss of K6a expression is without serious consequences in mice (Wojcik et al., 2000). It is reasonable to assume that ablation of K6a in humans will be without serious side effects, particularly if siRNA treatment is limited to lesion sites such as areas of keratoderma on the sole of the foot. Further support for the lack of effect of reducing K6a expression comes from a recent report on variation of gene copy numbers in the human genome. Around 3% of the population is heterozygous for a deletion that includes the genes for both K6a and K6b (Wong et al., 2007). Smith et al. (2007) speculate that approximately 75,000 individuals in the United States may be null for both K6a and K6b expression. The identification of such individuals and confirmation that they are indeed without a phenotype would provide incontrovertible evidence that silencing of K6 expression is a viable therapeutic strategy for the treatment of PC. In addition, it would indicate important differences between humans and mice because loss of K6a and K6b in mice results in a severe oral phenotype and failure to thrive (Wong et al., 2000).

Although there is considerable redundancy in the expression of K6 genes, this is not true for all keratins. For example, ablation of keratin 14 expression causes the blistering disorder epidermolysis bullosa simplex, and, although there may be compensation in some tissues by K15, the phenotype is broadly similar to that caused by dominant negative K14 mutations (Chan et al., 1994; Rugg et al., 1994). It is theoretically feasible to develop mutation-specific siRNAs; however, this may not be practical given the large number of different keratin mutations and the expense associated with drug development and testing. An alternative, indirect approach would be to target siRNAs to polymorphisms that would result in allele-specific mRNA degradation. However, caution is needed with this approach because the resulting haploinsufficiency may have unwanted side effects. For example, haploinsufficiency in keratin 5 expression is associated with Dowling–Degos disease (Betz et al., 2006). The extent to which reduced levels of expression of other keratins, such as the other PC-associated proteins K16 and K17, would lead to disease phenotypes is not known.

SiRNAs offer the promise of a "magic bullet" for treating many genetic disorders, but there are still some potential problems with their use as therapeutic agents. The target specificity must be validated for each siRNA and off-target effects must be eliminated or minimized. This may not be an easy problem to address because siRNAs, by necessity, are sequence specific and therefore may be ineffective or have different off-target effects in animal and humans. Potential side effects include activation of the interferon system, knockdown of unrelated gene targets due to sequence similarities, and saturation of the RISC, which interferes with processing of endogenous micro RNAs needed for normal cell function (Ryther et al., 2005; Barik, 2006). Evidence to date suggests that such problems can be overcome by avoiding the use of viral expression vectors and taking care in the design of highly specific synthetic siRNA, as is the case in the report by Smith et al. (2007). Finally, it is not known whether there are long-term effects of siRNA treatment. For such molecules to be therapeutically effective, they will need long-term administration.

Perhaps the biggest hurdle to be overcome is that of delivery. siRNAs must cross cell membranes and be incorporated into the RISC. Although the skin offers a large and accessible target, it is also a barrier. Wang et al. (2007) have demonstrated that siRNA can be delivered by injection to keratinocytes in the mouse foot pad, and Smith et al. (2007), using this method of delivery, showed that that K6a gene expression can be specifically inhibited by siRNAs in this model system (Wang et al., 2007). Whether this will prove to be a viable method for treatment of PC patients will depend on a number of factors, including how many injections are needed and how often they will need to be repeated to maintain gene silencing. Several other methods have been shown to be capable of delivering DNA to keratinocytes, including ballistic delivery using a gene gun, ultrasound and iontophoresis, and topical application. It remains to be seen which, if any, of these methods will be most effective and tolerated in patients.

RNAi holds great promise as a therapy for genetic disorders and is particularly attractive for targeting skin disorders. The accessibility of the skin, which allows restricted application, makes it an ideal system for testing the effectiveness of siRNA and minimizing the likelihood of generalized systemic effects. It remains to be established whether RNAi will turn out to be a viable method for treating PC or other keratin disorders. The report by Smith et al. (2007) brings this prospect significantly closer.