Grey Goo on the Skin? Nanotechnology, Cosmetic and Sunscreen Safety

By Nohynek, Gerhard J; Lademann, Jrgen; Ribaud, Christele; Roberts, Michael S

Many modern cosmetic or sunscreen products contain nano-sized components. Nanoemulsions are transparent and have unique tactile and texture properties; nanocapsule, nanosome, noisome, or liposome formulations contain small vesicles (range: 50 to 5000 nm) consisting of traditional cosmetic materials that protect light- or oxygen-sensitive cosmetic ingredients. Transdermal delivery and cosmetic research suggests that vesicle materials may penetrate the stratum corneum (SC) of the human skin, but not into living skin. Depending on the physical/chemical properties of the ingredient and the formulation, nano-sized formulations may enhance or reduce skin penetration, albeit at a limited rate. Modern sunscreens contain insoluble titanium dioxide (TiO^sub 2^) or zinc oxide (ZnO) nanoparticles (NP), which are colorless and reflect/scatter ultraviolet (UV) more efficiently than larger particles. Most available theoretical and experimental evidence suggests that insoluble NP do not penetrate into or through normal as well as compromised human skin. Oral and topical toxicity data suggest that TiO^sub 2^ and ZnO NP have low systemic toxicity and are well tolerated on the skin. In vitro cytotoxicity, genotoxicity, and photogenotoxicity studies on TiO^sub 2^ or other insoluble NP reporting uptake by cells, oxidative cell damage, or genotoxicity should be interpreted with caution, since such toxicities may be secondary to phagocytosis of mammalian cells exposed to high concentrations of insoluble particles. Caution needs to be exercised concerning topical exposure to other NP that either have characteristics enabling some skin penetration and/or have inherently toxic constituents. Studies on wear debris particles from surgical implants and other toxicity studies on insoluble particles support the traditional toxicology view that the hazard of small particles is mainly defined by the intrinsic toxicity of particles, as distinct from their particle size. There is little evidence supporting the principle that smaller particles have greater effects on the skin or other tissues or produce novel toxicities relative to micro-sized materials. Overall, the current weight of evidence suggests that nano-materials such as nano-sized vesicles or TiO^sub 2^ and ZnO nanoparticles currently used in cosmetic preparations or sunscreens pose no risk to human skin or human health, although other NP may have properties that warrant safety evaluation on a case-by-case basis before human use.

Keywords CAS 1314-13-2, CAS 13463-67-7, Cosmetics, Dermal toxicity, Genotoxicity, Nanoparticles, Nanotechnology, Percutaneous penetration, Photogenotoxicity, Sunscreens, Risk Assessment, Titanium dioxide, Zinc oxide

1. INTRODUCTION

Nanotechnology (NT) is a catch-all term for techniques, materials, and devices that operate at the nanometer scale, been defined as the design, characterization, production, and application of structures, devices, and systems by controlling shape and size at the nano-scale (Maynard, 2006), represents one of the most promising technologies of the 21 st century, and has been considered to be a new industrial revolution. Today, nanomaterials are increasingly used in sporting goods, tires, catalysts, electronic components, window sprays, paints, varnishes, coatings, foods, sunscreens, cosmetics, and antimicrobial and antifungal preparations and are expected to be increasingly applied to the medical field in diagnosis, imaging, and drug delivery (SCENIHR, 2005; NeI et al., 2006). The U. S. National Science Foundation estimated that the global market for NTs may reach $1 trillion or more within 20 years (Maynard, 2006).

The arrival of NT also initiated public debate about its potential risks. Michael Crichton’s novel Prey (2002), a tale of rioting and runaway nanobots, popularized Eric Drexler’s earlier Grey Goo Theory (1986), a hypothetical end-of-the-world nanotechdisaster scenario, in which out-of-control self-replicating nanorobots (assemblers) consume all living matter on Earth. Although Grey Goo seems more science fiction than science, the theory has raised concerns in the public and the media as well as prominent persons, such as the Prince of Wales, termed as Royal Nanoangst by the UK media (Telegraph, 2003). The fear of this new technology was neatly described by Ball (2003): Overnight nanoscientists had become the new Frankensteins, modern Prometheuses, contemporary Fausts, dabbling with dangerous forces they cannot control.

A source for concern may also arise from ignorance. Tsuji et al. (2006), for instance, suggest that the skin penetration and toxicological impact of nanoparticles is unclear but there is a potential for a range of local, chronic, metabolic, and photoinduced toxicities. These potential, albeit largely hypothetical, risks to human heath and the environment have resulted in several nongovernmental organizations (NGOs) demanding a ban or greater regulation of NT applications (Greenpeace, 2003; ICTA, 2006; FOE, 2006a). An active discussion about potential risks of NT has been undertaken by regulatory agencies and engaged scientists and has resulted in position papers from various regulatory bodies. Some toxicologists proposed that new properties of nano-materials may require novel approaches for their hazard assessment and a need for the new discipline of nanotoxicology (Oberdrster et al., 2005a, 2005b), whereas others argued that toxicology itself may become a new discipline of nanoscience (Kurath and Maasen, 2006). However, in the view of more traditional toxicologists, the approaches and study protocols for routine toxicological characterization of chemicals are sufficiently robust to provide meaningful characterization ofnanoscale materials (NTP/NIEHS, 2004). During recent years, the potential risks of nano-materials to human health and the environment have been reviewed by numerous national and international expert groups, such as the U. S. NTP/NIEHS (2004), Royal Society of Engineering (2004), European Union (EU) SCENIHR (2005), U. S. Environmental Protection Agency (EPA) (2005a), German BfR (2006), Australian TGA (2006), French AFSSET (2006), Canadian IRRST (2006), and an expert group of the European Chemical Industry’s ECETOC (Borm et al., 2006), as well as individual authors such as Hoet et al. (2004), NeI et al. (2006), Hardman (2006), and Maynard (2006). The general consensus of these reviews was that NTs may pose possible new risks, although the actual nature of these risks remains largely hypothetical.

Nanoparticles (NP) are a subset of nano-materials, and were defined as single particles with a diameter below 100 nm, although their agglomerates may be larger (Maynard, 2006). One of the largest applications of NP is their use in sunscreens where the NP diameter is normally more than 10 nm. The global production of NP for sunscreen products was estimated to be approximately 1000 tons during 2003/2004 (Borm et al., 2006), and principally consists of titanium dioxide (TiO^sub 2^) and zinc oxide (ZnO) particles. Today there is a broad consensus that the principal human health risk may be from inhalation of NP (ECETOC, 2005; Hoet et al., 2004; Maynard, 2006). Human cooking activities (e.g., toasters, oven cooking, frying) or other indoor emissions (e.g., candles) may be the most important sources for human exposure to airborne NP (Olson and Burke, 2006; Afshari et al., 2005). However, concerns have also been raised about potential dangers of the contact of NP with human skin. Recently, the Friends of the Earth warned against NT in cosmetic and sunscreen products, since they may produce a possible uptake of particles by human skin: if nanoparticles penetrate the skin, they can join the bloodstream and circulate around the body with uptake by cells, tissues and organs (FOE, 2006a). Possible human systemic exposure from topically applied NP has also been suggested in toxicological reviews by Hoet et al. (2004), Oberdrster et al. (2005a), and even the US ERA Draft White Paper (2005a), with the last hypothesizing that nanoparticles may penetrate the skin and distribute throughout the body once translocated to the circulatory system.

Overall, the key questions that must be asked for any NP applied to the skin is (1) what is the exposure, (2) is it absorbed and, if so, how much reaches the viable cells, and (3), if so, is it intrinsically toxic? More specific questions that may be raised concerning the safety of NT/NP in cosmetic products and sunscreens include the following:

* Do cosmetic formulations containing nano-sized features (vesicles or droplets) pose new risks when compared with those of traditional cosmetic products?

* Do nano-sized cosmetic formulations enhance the skin penetration of cosmetic ingredients, thereby increasing the risk of human skin sensitization or systemic exposure?

* Are insoluble NP in sunscreens intrinsically more hazardous than larger particles, i.e. micro-particles or bulk material?

* Do topically applied insoluble NP remain on the skin surface or are they able to pass the skin barrier of normal or compromised skin to gain access to systemic compartments of the organism?

To this end, we attempt to summarize the available information on the use of NTs in cosmetic and sunscreen products and to review the evidence on potential adverse effects of NP on or in the skin, \including their potential to penetrate into or through human skin and/or to pose a risk of human systemic exposure and toxicity. In the first part of this article, we examine formulations containing NP and then examine the penetration of NP into and through the skin, addressing ZnO and TiO^sub 2^ penetration, follicular penetration, penetration of other NP, the effect of formulation, and the consequences of the skin integrity being compromised. In the second part, we consider intrinsic, cellular, and phototoxicity of NP, their ingredients, and their coatings. The third part examines in vivo considerations such as the likelihood of toxicity after systemic exposure, topical exposure, and the risk of sensitization. The final part attempts to objectively assess the risks and benefits of topical NP use.

FIG. 1. Examples of nano-sized vesicles or particles in cosmetic formulations or sunscreens.

2. NANOTECHNOLOGY AND NANOPARTICLES IN COSMETICS AND SUNSCREENS

2.1 Cosmetic Formulations Containing Nano-Sized Structures (Figure 1)

2.7.7 Micro- andNanoemulsions

Nano-emulsions are commonly used in certain cosmetic products, such as conditioners or lotions to be applied to the skin and hair (Figure 1). The optical, tactile, and texture properties of nano- emulsions make them highly attractive for cosmetic or consumer products. These emulsions combine traditional cosmetic ingredients, such as water, oils, and surfactants, to manufacture two-phase systems in which droplets of size 50 to 100 nm are dispersed in an external phase. The small droplet size enables nano-emulsions to flow easily, be transparent, and be pleasant to the touch. Their unique texture and Theological properties have yet to be obtained by other formulation methods (SonnevilleAubrun et al., 2004). Nano- emulsions pose no different risk to consumers than traditional emulsions, as both are composed of water and oil droplets, and tend to break down into their constituent ingredients upon application to skin or hair (Van den Berg et al., 1999). Micro- or nano-emulsions can enhance skin penetration of some cosmetic or dermatological ingredients but usually less than that seen with these ingredients in solution (Lehmann et al., 2001; Kreilgard, 2002; Korting et al., 1990; Kogan and Garti, 2006).

2.1.2 Liposomes, Niosomes, Nanosomes, and Nanocapsules

Liposomes and niosomes are globular vesicles with diameters between 25 and 5000 nm. Such vesicles are composed of suitable amphiphilic molecules that associate as a double layer (unilamellar vesicles) or two to four double layers (multilamellar vesicles). In general, the outer and the inner layers of vesicles have a hydrophilic character (Figure 1). Liposomes and niosomes may incorporate hydrophilic or lipophilic substances or drugs. Liposomes are mainly generated from phospholipids; niosomes are composed of nonionic surfactants, such as polyoxyethylene alkyl ethers or – esters (Junginger et al., 1991). Vesicles materials mainly consist of phospholipids, sphingolipids, or ceramides; due to their lipophilic interior they have an enhanced capacity to enclose lipophilic substances (Castor, 2005; Brunke and Charlet, 1991). The ultrastructure of some of these vesicles is quite similar to that of mammalian milk, which contains nano-sized fat droplets surrounded by the milk fat globular membrane (Lopez, 2005). Rigid nano- or microcapsules have rigid walls consisting of sucrose esters, cholesterol or cholesterol sulfate, or biodegradable polymers, such as polycaprolactone (Van den Bergh et al., 1999; Alvarez-Roman et al., 2001).

Vesicle formulations are produced from traditional cosmetic ingredients by techniques such as coacervation or phase separation. These formulations are important in cosmetic applications because they may improve the stability and, possibly, tolerance of ingredients such as unsaturated fatty acids, vitamins, or antioxidants that are encapsulated within the vesicle, but also the efficacy and tolerance of ultraviolet (UV) filters on the skin surface (Padamwar and Pokharkar, 2006; Alvarez-Roman et al., 2001). When applied on skin, vesicles tend to break down into their constituent ingredients, which tend to remain in the upper layers of the stratum corneum (Junginger et al., 1991; Van den Bergh et al., 1999; Choi and Maibach, 2005).

2.2. Insoluble, Mineral-Based Nanoparticles in Sunscreens

Modern sunscreens contain insoluble, mineral-based materials whose performance depends on their particle size. Mineral particles, such as TiO^sub 2^, reflect and scatter UV light most efficiently at a size of 60 to 120 nm (Popov et al., 2005), whereas ZnO has an optimal size of 20-30 nm (Cross et al., 2007; Figures 1 and 3). As such particles scatter UV and not visible light, the resulting sunscreen appears to be clear. Sunscreen-grade nanosized TiO^sub 2^ ranges from an ultrafine particle form with a diameter of 14 nm to micro-sized aggregates (Table 1). ZnO is generally used in the form of particles at 30-200 nm in diameter. The surface of these mineral particles is frequently treated with inert coating materials, such as silicon oils, SiO^sub 2^ or Al^sub 2^O^sub 3^, in order to improve their dispersion in sunscreen formulations (see Table USCCNFP, 2000).

Sunscreen products containing mineral UV filters protect consumers from the harmful effects of UV exposure, including skin aging, skin and lip cancers, and herpes labialis (Pogoda and Preston- Martin, 1996; Nohynek and Schaefer, 2003). Consequently, dermatological associations and national or international health authorities strongly recommend the application of sunscreens before sun exposure (WHO, 1998). Given that nanosized particles of titanium or zinc oxides are transparent, these UV filters are not only more efficient, but also result in better consumer acceptance and, ultimately, improve the protection of human skin against UV-induced damage. Interestingly, it has recently been shown that lead-based traditional hair dyes that have been used since the Greco-Roman period produce their darkening of gray hair by formation of lead sulfide NP (size about 5 nm) on the surface of the hair (Walter et al., 2006). Thus, the use of and human exposure to cosmetic-derived NP appears to have a history of more than 3000 years.

3. LOCAL AND SYSTEMIC EXPOSURE FOLLOWING DERMAL APPLICATION OF NANOMATERIALS

3.1 General

Mammalian skin is structured in several layers: the stratum corneum (SC), epidermis, dermis, and the subcutaneous layer. For most substances, the SC is the rate-limiting barrier against the percutaneous penetration of topically applied substances (Schaefer et al., 2003). The rate of skin absorption and penetration of cosmetic ingredients may be measured in vivo or in vitro under realistic conditions of product use. However, in vitmand in vivo skin absorption data should be interpreted with caution. Currently used in vitro models, such as pig and human skin, may yield comparable results, provided the total quantity of ingredients present in skin sample is taken into consideration. However, the distribution of a substance in the different compartments may vary due to difference in the relative thicknesses of the SC, epidermis, and dermis of human and pig skin. Differences in relative distribution between pig and human skin may be large or small depending on the substance under evaluation (Diembeck et al., 1999). The proportion of the substance absorbed or penetrated may also vary as a function of the quantity applied. This variability depends on the nature of the substance under evaluation and is not easily predicted. For this reason, it is crucial to use experimental protocols where the quantity applied is representative of the actual use conditions of the product.

TABLE 1

Commercial coated and noncoated sunscreen-grade titanium dioxide particles: Results of in vitro phototoxicity, genotoxicity, and photo-genotoxicity tests (unpublished data included in the industry safety dossier, summarized in the EU SCCNFP opinion on TiO^sub 2^; SCCNFP, 2000)

Experimental protocols that produce partial or total occlusion tend to favor the penetration of substances into the SC by increasing its level of hydration, which reduces its barrier function (Zhai and Maibach, 2001). Therefore, when determining the localization of a substance in an in vitro absorption model, it is important to consider swelling phenomena that may produce artifacts.

Some percutaneous penetration studies used isolated human or animal epidermis or stratum comeum (SC), which are available in the form of thin and fragile films, although such materials will yield higher absorption or penetration rates than full-thickness skin (Surberet al., 1990; Potts and Guy, 1992). Similar to occlusion, immersion may produce substantial swelling of the SC; therefore, substances may penetrate into spaces between swollen corneocytes, whereas skin absorption may be low or absent after topical application of substance to intact skin (Zhai and Maibach, 2001). Destructive methods for studying the distribution of a substance in the skin and stripping of the SC can also lead to artifacts in that hair follicle or skin furrow material may be assayed together with the epidermis or the living skin. In reality, particles readily penetrate into the hair follicle opening (its ostium), which is several tens of micrometers deep. sectioning or stripping of the epidermis by heat treatment may result in material stored in hair follicles or skin furrows being collected together with sections of the epidermis so that a potential incorrect conclusion that may be reached that a substance had penetrated into the epidermis. In vitro, NP or micro-particles are known to aggregate in the hair follicle ostium or skin furrows without further penetrating into or through the living skin (Lademann et al., 1999; Mavon et al., 2007). We have used multiphoton and confocal microscopy to confirm that many NP > 15 nm do not penet\rate the human stratum corneum beyond the superficial layers (M. Roberts, unpublished observations, 2006).

In vivo skin penetration studies are often carried out in a range of rodent or farm animals as well as in human subjects (Schaefer and Redelmeier, 1996; Walters and Roberts, 1993). Interpretation of in vivo skin penetration studies between different species must be undertaken with caution as the permeability can vary widely depending both on the nature of the species and of the compound being studied. In general, the penetration of rabbit skin > rat > pig > monkey > human, with the pig being about 4 times or more and the rat up about 9 times more permeable than human skin for certain compounds (Magnusson et al., 2001).

FIG. 2. Section of the hair follicle and routes of possible penetration pathways of externally applied substances into and through skin (adapted from Lademann et al., 1999, 2006).

3.2 Passive Penetration of Insoluble ZnO and TiO^sub 2^ NP Into or Through the Skin (Figure 2)

Skin penetration studies on TiO^sub 2^ and ZnO micro- or nanoparticles are summarized in Table 2. The maximum flux into or through the skin of molecules in solution or as pure solutions falls exponentially with molecular weight to have an upper limit of 10^sup -12^ mol/cm^sup 2^/h at 800 Da (Magnusson et al., 2004). Brown et al. (2006) have suggested that solutes that are >500 Da, have high melting points and have insufficient amphiphilicity show little or negligible tendency for passive penetration into or through intact human skin. Simulations show that a solute with a molecular volume equivalent to insoluble NP and that is one to two orders of magnitude larger than a solute with a molecular mass of ~500 Da results in epidermal concentrations of the order of 10^sup -18^ nmol/ ml after a safety factor of 100 is included and becomes even lower when desquamation is taken into account (Roberts, 2006). Accordingly, based on the available skin permeability data, it is predicted that there will be no passive penetration of solid, insoluble particles into or through the skin. This absence of skin penetration has also been corroborated by numerous experimental data, summarized later in this article. However, a number of exceptions have also been reported.

The penetration of micro-fine zinc and titanium oxide particles into animal and human skin appears to be the most studied of all NP. Most studies have reported that NP applied to the skin only penetrate into hair follicle openings and skin furrows, with minimal material being found below the stratum corneum surface. Landsdown and Taylor (1997) measured the penetration of micro-fine zinc and titanium oxide particles into rabbit skin. Most of the applied material remained on the skin surface, the outer layers of the SC, or the outer aspects of the hair follicles, and no deposits were found in deeper aspects of the epidermis or the dermis. Another study on the skin penetration of nano-sized, sunscreen-grade TiO^sub 2^ and ZnO particles using transmission and scanning electron microscopy as well as x-ray diffraction techniques detected particles only on the surface of the stratum corneum. No skin or intracellular penetration of particles was found (Dussert and Gooris, 1997). A series of in vivo and in vitro studies were performed on the percutaneous penetration of nano-sized TiO^sub 2^ pigments. The results of more than 10 different published or unpublished studies in vitro (human and pig skin) or in vivo (rat model, or biopsies of human skin) were summarized in a relatively recent opinion of the EU Scientific Committee on Cosmetics and Non- Food Products (SCCNFP) on TiO^sub 2^ (2000). All studies included in the SCCNFP opinion as well as published investigations concluded that micro- or nanosized TiO^sub 2^ particles remain on the skin surface or the outer layers of the SC and do not penetrate into or through the living skin (Tan et al., 1996; Pflucker et al., 2001; Lademann et al., 1999; Schulz et al., 2002; Gontier et al., 2004; Gamer et al., 2006). Gottbrath and Miiller-Goymann (2003) showed that the penetration of TiO^sub 2^ (20 nm and 100 nm) into human stratum corneum, studied by tape stripping, was limited to the surface and valleys between the corneocytes. In contrast, Menzel et al. (2004) suggested that TiO^sub 2^ applied in formulations to pig skin penetrates into the stratum granulosum but not the stratum spinosum. Penetration via the follicles was discounted. The very recent study by Mavon et al. (2007) on the percutaneous absorption of TiO^sub 2^ NP (20 nm) in a sunscreen formulation in human skin in vitro as well as in human subjects showed that penetration was limited to the upper layer of the stratum corneum and confirmed once again that these particles do not penetrate into or through living human skin. In addition, the study revealed that small amounts of particles may be found in the epidermal compartment in vitro, which corresponds to sunscreen located in the skin furrows or the infundibulum, but not in the living epidermis.

Similarly, as shown in recent in vitro percutaneous penetration studies, ZnO NP showed negligible penetration into pig (Gamer et al., 2006) and human skin (Cross et al., 2006). These findings confirmed the results of a number in vitro or in vivo percutaneous penetration studies on ZnO particles that were reviewed in the recent SCCNFP opinion (2003a). None of these studies suggested significant penetration into or through living human or animal skin. However, although TiO^sub 2^ NP can have particle diameters as low as 10 nm, ZnO NP are typically >20 nm in order to have transparency in visible light (Figure 3).

TABLE 2

Overview of TiO^sub 2^ and ZnO skin absorption/penetration studies

TABLE 2

Overview of TiO^sub 2^ and ZnO skin absorption/penetration studies

3.3 Penetration of NP Into the Stratum Corneum and Follicular Penetration

Hair follicles represent a potential target for transdermal drug delivery, since they are surrounded by a tight network of capillaries which may be an important target for drug uptake (Figure 2). However, the presence of TiO^sub 2^ NP in the hair follicle reported by Lademann et al. (1999, 2001) has frequently been misquoted or misinterpreted as penetration into the living skin, although the particles remained outside the living epidermis or dermis were shown be eliminated by sebum flow. Lademann et al. (1999) and Weigmann et al. (1999), in studying the penetration of sunscreen-grade TiO^sub 2^ nanoparticles into the skin by means of tape stripping, showed that particles were mainly located on the skin surface and in the lipid layers around the corneocytes of the first cell layers of the stratum corneum (Figure 4). In deeper parts of the stratum corneum, TiO^sub 2^ was absent. Significant amounts of TiO^sub 2^ particles were detected in the orifices of hair follicles; their presence was confirmed by skin biopsy in human subjects. Interestingly, not all hair follicle orifices appeared to act as a reservoir for topically applied nanoparticles, there were open or closed follicle orifices (Lademann et al., 1999).

FIG. 3. ZnO nanoparticles and topical absorption: (A) TEM of coated particles, (B) PCS size distribution of micronized particles, (C) spectral transmittance of ZnO particles in aqueous solution, and (D) TEM showing ZnO particles present on the skin surface and around desquamating corneocytes (no penetration into the underlying stratum corneum was observed (from Cross et al., 2006).

FIG. 4. TiO^sub 2^ penetration into the stratum corneum of human subjects after in vivo sunscreen application for 4 days (adapted from Lademann, 1999).

Later it was shown that the phenomenon of open and closed follicle orifices is not limited to particles; it may be relevant for many topically applied nonparticular substances (Otberg et al., 2004). Results of subsequent investigations suggested that hair follicle orifices are open during sebum production and/or active hair growth (Lademann et al., 2001). The phenomenon of closed follicles may be due to a mixture of desquamated corneocytes and dry sebum, which form a protective cover during the resting phase of the follicle. This cover is opened by sebum flow or hair growth and may easily be removed by washing or skin peeling (Otberg et al., 2004).

Collection of topically applied particles or substances in the follicular sink does not mean that they pass the skin barrier. Insoluble particles collected in the follicular orifices do not penetrate into the living epidermis, but remain in the follicular orifices for some time and are slowly removed by sebum flow. There may be an optimal size of particles concerning their collection and storage in the follicular orifice. For example, Toll, et al. (2004) investigated the tendency of particles at different sizes (600 to 2,500 nm) to collect in the follicular orifice of hair follicles of excised human skin. They found that the smallest particles (600 nm) had the greatest tendency for follicular storage.

Teichmann et al. (2006) compared in pig-ear skin the penetration and storage of the fluorescent dye fluorescein in a particle (320 nm) formulation as well as in solution. The dye formulations were applied with and without massage, which was performed using a commercial massage applicator for 1 min following the application of the formulations. In the absence of massage, an identical penetration depth of the particle and nonparticle formulations was observed. However, subsequent to massage, particles moved five times deeper into the hair follicles when compared with the depth of penetration of the dye solution. It was suggested that the cuticula of moving hair shafts could act like a geared pump, pushing the particles deeper into the hair follicles by mechanical force. This process seems to be particularly efficient when the size of applied particles is in the same order of magnitude as the cuticula of \human hair, that is, 400 to 700 nm. Interestingly, nano-sized particles (< 100 nm) showed no tendency of enhanced movement into the follicular orifice. Although massage resulted in a greater depth of particle distribution into the follicular orifice, the particles remained outside the living epidermis.

Lademann et al. (2007) repeated the experiments with fluorescein in a particle formulation and in solution in human skin using differential cyanoacrylate skin stripping (Teichmann et al., 2006), which allows a noninvasive removal of the hair follicle content (Figure 5). After the follicle contents were removed by cyanoacrylate surface biopsies, the amount of nanoparticles in the tape strips and the cyanoacrylate biopsies was determined and storage of the particles in the stratum corneum and the hair follicle orifice was compared. The results suggested that particles located in the stratum corneum were nearly quantitatively removed after one day, whereas particles in the hair follicles remained for more than 10 days. Comparison of the storage of particle- and solution-based formulations in hair follicle orifices showed quantitative disappearance of the dye solution after 6 days, whereas dye particles remained in the hair follicles for more than 10 days before being eliminated.

FIG. 5. Hair follicle content removed with a cyanoacrylate biopsy (laser scanning microscopy). Adapted from Lademann et al. (2006).

Summarizing the current knowledge on the follicular sink, it could be established that the orifice of hair follicles may represent a long-term reservoir for topically applied substances and particles. Dermal application of substances targeted for follicular orifices may open new routes in drug delivery. However, although insoluble NP, such as TiO^sub 2^, were shown to be present in the hair follicle orifices, they remained outside the living skin and no evidence for local (living skin) or systemic exposure via follicular penetration was found.

3.4 Penetration of Other NP Into the Skin

It has been argued that there is some evidence suggesting that nanoparticles may penetrate into or through human skin (Hoet et al., 2004; Oberdrster, 2005a). However, most of this work has been undertaken using NP other than ZnO or TiO^sub 2^ and, in general, equivocal results have been found. For instance, on application of covalently bound, fluorescent nanocapsules to pig skin, fluorescence was detected in open follicles and in furrows of the skin, but no penetration of these particles into the living epidermis or dermis was found (Alvarez-Roman et al., 2004a, 2004b), and Stracke et al. (2006) found no penetration of insoluble, polymeric nanoparticles into or through the living skin. Gopee et al. (2006) suggested that quantum dots only penetrated into intact mouse skin after dermabrasion (removal of the entire SC by tape-stripping). In contrast, Ryman-Rasmussen et al. (2006) suggested that quantum dots (QD; semi-conductor nanocrystals, spherical and ellipsoid shape, particle size 4.6 or 12 nm) with neutral or cationic coatings may penetrate into the epidermis or dermis of intact porcine skin, whereas QD with anionic coating penetrated to a small extent into the epidermis after 24 h of exposure. However, these studies were conducted with the QDs being applied in quite alkaline solutions.

FIG. 6. Incidence (%) of chromosome aberrations in in vitro photo- genotoxicity tests on ZnO (particle size: <200 nm) in Chinese hamster ovary (CHO) cells relative to ZnO-induced cytotoxicity: (a) in the dark, (b) after irradiation with UV at 700 mJ and simultaneous treatment of cells with ZnO (standard photogenotoxicity protocol), and (c) after pre-irradiation of cells with UV at 700 ml, followed by treatment with ZnO in the dark (from Dufour et al., 2006).

Another study frequently quoted as evidence for particle penetration into the skin actually described presence of soil particles (0.4-0.5 m) in the dermis of a limited number of patients with endemic elephantiasis (affecting their feet) who have walked barefoot in African rift valleys and elsewhere (Corcachan et al., 1988; Blundell et al., 1989). It is likely that chronic exposure, pressure, excessive skin hydration arising from the underlying edema, and impaired skin permeability may have contributed to skin penetration.

A number of studies have suggested that penetration of NP through pig skin may be dependent on the surface charge and size of the particles. Kohli and Alpar (2004) reported that negatively charged 50-nm and 500-nm fluorescent particles permeated pig skin, but that negatively charged 100- or 200-nm particles did not, and also that neutral and positively charge particles did not penetrate. Shim et al. (2004) showed that neutral PEG coated 40-nm NP penetrated into the epidermis of hairless guinea pig skin. The earlier quoted work of Ryman-Rasmussen et al. (2006) reported that quantum dots could not be detected in the perfusate of dermatomed pig skin mounted in flow through diffusion cells. However, significant penetration of neutral, positively charged, and negatively charged nanoparticles (diameters: spherical 4.6 nm, ellipsoid 6 12 nm) into the epidermis and, for cations, also into the dermis was found. While these findings suggest that very small NP may have a capacity for passive penetration into intact skin, the relevance of pig skin and of the alkaline pHs used to defining likely human exposure is not clear. Accordingly, these data require confirmation in humans using in-use conditions.

It has been suggested that NP after penetrating the skin may be taken up by local lymph nodes and transported into the blood circulation; a hypothetical uptake of NP by skin nerves or sweat glands was also suggested (Oberdrster, 2005a). While it is likely that NP administered directly to the viable epidermis or dermis will be redistributed to the lymph nodes, it is noted that small, inert surgical implant wear debris particles tend to remain localized and result in little, if any, systemic exposure (see section 4.1). Further, nano- or micro-particles in NP- or nanovesicle-containing intravenous formulations are rapidly cleared from the circulation by the so-called phagocytic barrier, that is, monocytes in the blood, macrophages in the spleen, or Kupffer cells in the liver (Moghimi et al., 2001).

Overall, although a gray zone may exist concerning the passive skin penetration capacity of extremely small NP with sizes comparable to that of large molecules, the current evidence indicates that the far larger and insoluble NP used in sunscreens do not show significant skin penetration or systemic exposure. Given the work of Ryman-Rasmussen et al (2006), more research is needed on the passive dermal absorption of small (< 10 nm) NP, such as quantum dots across human skin under appropriate exposure conditions.

3.5 Skin Penetration from Nano-Sized Vesicle-Type or Other Formulations (Nanosomes, Liposomes, Niosomes, Nanoemulsions, Nanocapsules, Solid Lipid Nanoparticles)

Most of our knowledge on the skin penetration of ingredients in vesicle-type and other nano- or micro-sized formulations has been gained by the research on transdermal drug delivery (TDD) techniques. TDD has many advantages over traditional routes of drug administration, including avoidance of drug inactivation in the gastrointestinal tract, hepatic first-pass effects, and providing continuous systemic delivery. TDD systems include passive (skin penetration by diffusion alone) and active systems (drug delivery enhanced by application of external energy).

In 2004, about 35 passive TDD preparations, all of them used under occlusive patches, were approved in the United States or the EU for a wide variety of conditions, including hypertension, motion sickness, postmenopausal problems, and, recently, contraception and urinary incontinence (Thomas and Finnin, 2004). Examples of drug substances that are currently used in passive TDD systems are shown in Table 3. During recent decades, huge TDD research efforts investigated drug formulations to enhance or facilitate the percutaneous penetration of topically applied substances. Skin penetration enhancement techniques used in TDD were recently reviewed (Benson, 2005). Drug formulations containing nano- or micro- sized vesicles, such as emulsions, liposomes, niosomes, or nanosomes, have been the object of intensive research, since they were considered to be highly promising to carry drugs into or through human skin. TDD using novel formulations and techniques resulted in numerous innovative applications and indications of traditional drug substances, which has recently been termed a transdermal revolution (Thomas and Finnin, 2004).

TABLE 3

Examples of drug substances used in commercial passive transdermal drug delivery systems (occlusive patches), molecular weight, drug structures, and calculated molecular sizes (adapted from Benson, 2005)

An in-depth review on the TDD of various lipophilic and hydrophilic drug substances (retinoic acid, 5-fluorouracil, triptolide, ascorbic acid, diclofenac, lidocaine, prilocaine) formulated as different microemulsions on the basis of several fatty acids as oil phases, phospholipid-type and anionic surfactants, short-chain alcohols cosurfactants, and penetration-enhancing substances has recently been published. The results in animal and human skin suggested that there are many factors affecting the delivery of topically applied drugs. Although some microemulsions were capable to enhance the skin penetration of drug substances (max. 2 to 3 times), their delivery rate depended on the physical/ chemical properties and concentration of the drug, structure and ingredients of the carrier and the type of the skin membrane used for penetration studies. When compared to standard formulations (gels, creams, solutions), drug delivery enhancement by microemulsions appeared more efficient for hydrophilic than for lipophilic drug s\ubstances (Kogan and Garti, 2006).

The use of liposome and niosome formulations in TDD including their skin penetration characteristics has been reviewed by Choi and Maibach (2005), who concluded that vesicle formulations may significantly enhance the percutaneous penetration of drug substances, but only for certain and suitable molecules. For drug substances that have a high intrinsic capacity for skin penetration, such as steroid hormones, caffeine, or nicotine, liposome formulations may result in less skin penetration when compared with that from solutions, but may result in higher drug concentrations in the stratum corneum. For more poorly penetrating drug substances, liposomes or other vesicle-type formulations may enhance their skin penetration, although the magnitude of penetration enhancement tends to be moderate. These reports are consistent with the experience of our own laboratory, which found that liposome or nanosomes vesicle formulation may enhance or reduce skin penetration of cosmetic ingredients, although at a limited scale, that is, maximally two- to threefold (C. Ribaud, unpublished data, 2006).

Encapsulation of lipophilic ingredients in rigid nanocapsules, such as polycaprolactone capsules, was shown to enhance the skin penetration of some molecules, such as OMC, a lipophilic ultraviolet filter, but also increased its skin protection against UV light, which would permit the use of lower concentrations (Alvarez-Roman et al., 2004a). In contrast, in an earlier study, the same UV filter formulated in nanocapsules consisting of the same polymer after application to pig skin was only found in the SC, but was not detected in the deeper layers of the skin (Alvarez-Roman et al., 2001). Similarly, comparison in excised human skin of the penetration of the drug flufenamic acid in poly(lactide- coglycolide) nanoparticles with that of the drug in solution showed less penetration of the NP-formulation over 12 h when compared with that of the drug solution; longer incubation times resulted in a somewhat greater penetration of the drug from the NP-formulation, presumably due to protection of the active ingredient. NP were only detected in the upper layers of the SC and no penetration into the living skin was observed (Luengo et al., 2006).

The degree and magnitude of skin penetration of drugs substances formulated in vesicles does not appear to be related or proportional to a smaller particle size of vesicles. For example, it was shown that the skin penetration of triamcinolone acetonide in rat skin of mono- or multilamellar vesicles at 200 or 1000 nm diameter was similar (Yu and Liao, 1996). An in-depth study in mouse, hamster, and pig skin on the skin penetration of cyclosporine A in liposomes of various particle sizes (60,300, and 1000 nm) showed the greatest penetration into or through the skin for the 300 nm vesicles and suggested that skin penetration of intact vesicles or their materials does not occur (Du Plessis et al., 1994). Another study on the penetration in human abdominal skin of hydrophilic or lipophilic fluorescent dyes formulated in liposomes of 5 different particle sizes (73 to 810 nm) found an enhanced penetration of the hydrophilic dye from liposomes of 120 or 190 nm, whereas the lipophilic dye penetrated most from vesicles at a size of 71 nm (Verma et al., 2003).

Formulation is, however, important. Bennat and Mller-Goymann (2000) had previously reported that oily microfine TiO^sub 2^ (20 nm) penetrates deeper into stratum corneum from an oily emulsion than an aqueous one and liposomes increases the penetration depth of TiO^sub 2^ into the stratum corneum. This result was confirmed by Gottbrath and Mller-Goymann (2003). Another large study investigated the skin penetration of tretinoin in positive- or negative-charged liposomes of different particle sizes (135 to 1163 nm) and types (multi- and monolamellar), as well as in water/ethanol, oil, and a dermatological cream formulation. The results suggested that skin penetration from negatively charged vesicles was superior, when compared with that of traditional (cream, oil, ethanol/water) formulations (maximal twofold increase), but was unaffected by vesicle size or lamellarity and primarily depended on the vesicle material. No evidence for intact skin penetration of vesicles or their materials was found in the study (Sinico et al., 2005).

Overall, these data suggests that vesicle materials as well as vesicle size may affect the skin penetration of liposome- or niosome- encapsulated drugs, whereas the degree of the penetration of the active ingredient is not proportional to a reduced (<100 nm) vesicle size. No evidence was found that vesicles at nano-range produce greater skin penetration of encapsulated drugs when compared with that of micro-sized vesicles. Possibly, an ideal vesicle size exists depending on the individual vesicle material and drug substance (Verma et al., 2003). The results of reports on skin penetration of vesicles and their materials consistently concluded that the vesicles or their materials do not penetrate beyond the most superficial layers of the SC (Ganesan et al., 1984; Schreier and Bouwstra, 1994; Honeywell-Nguyen et al., 2004). Similar results were found in an in-depth investigation on the skin penetration of formulations containing elastic or rigid vesicles (particle size: 100 to 150 nm), confirming that that vesicle materials did not penetrate further than the stratum corneum (Van den Bergh et al., 1999).

The only TDD system that resembles solid NP consists of solid- lipid nanoparticle (SLNP) systems that were investigated for skin penetration enhancement for a wide variety of substances. Although SLNP were initially reported to improve the skin absorption of some drugs, sunscreens or vitamins (Santos et al., 2002), the enhancement has been recognized to be mainly related to an increase in skin hydration produced by an occlusive lipid film formed on the surface of the skin, and not by actual skin penetration of the SLNP themselves (Wissing and Mller, 2003; reviewed by Benson, 2005).

In general, the penetration-enhancing capacity of vesicle- containing formulations appears to be limited; such formulations are not able to perform miracles and drive any compound into or through human skin. Today it has been recognized that passive TDD, that is, delivery by occlusive patches, vesicletype formulations, gels, or creams, may only yield significant skin penetration and therapeutic effects of drug substances that combine a high pharmacological potency, a suitable log octanolwater partition coefficient (log P) value (1 to 3), a melting point below 200C, and a molecular mass of <500 Da, a combination of parameters similar to Lipinski's Rule of Five predicting bioavailability of oral drugs (Lipinski et al., 2001). Drug substances marketed today in passive TDD systems all fit into these limits, they have molecular weights between 160 and 360 and a molecular size between 0.75 to 1.6 nm (Benson, 2005; Table 3).

For larger molecules and/or physical characteristics outside this range, such as large molecules, peptides, proteins, or nucleotides, significant percutaneous penetration may only be achieved by active delivery methods, such as skin abrasion, skin erosion (suction blisters), electrical (ionophoretic) and mechanical (skin stretching) methods, or other energy-related techniques, such as ultrasound or needleless injection, or by the combination of active and/or passive delivery methods (Benson, 2005; Thomas and Finnin, 2004).

3.6 Nanoparticles and Altered, Compromised, or Diseased Skin

In the absence of the stratum corneum barrier, there is a potential for certain NP to lead to immune modulation. Kim et al. (2004) have shown that quantum dots may be translocated to regional lymph nodes, possibly via macrophages and Langerhans cells after application to the dermis. In general, however, other than with lacerations and burns, some epidermal barrier function normally exists in compromised skin. Further, most cosmetic products, with perhaps the exception of sunscreens or after-sun products that are applied to sunburned, that is, inflamed skin, cosmetic products are designed to be used on healthy skin. In contrast, certain topical dermatological drugs are applied to diseased skin. This position is reflected by current EU Notes of Guidance for Testing of Cosmetics and their Ingredients that indicate that key dermal safety studies, such as irritation, percutaneous penetration, or sensitization testing in vitro, in vivo or in human subjects should be performed on normal, healthy skin, and not on compromised or diseased skin (SCCNFP, 2003b).

A review of the literature suggests that there is little evidence suggesting a general rule that slightly compromised skin has a greater susceptibility to skin penetration by small particles (Schfer-Korting et al., 1994), although it has been recognized that certain pathological skin conditions may affect skin penetration of topically applied substances. Overall, it is uncertain whether all conditions of compromised skin enhance percutaneous penetration of topically applied substances or particles. For example, it has been shown that the percutaneous absorption and systemic bioavailability following application of a [^sup 14^C]methylprednisolone aceponate- containing lotion to intact and inflamed skin (UVB-induced sunburn) of human subjects were identical or lower for inflamed skin, whereas removal of the stratum corneum by tape stripping significantly enhanced the penetration of the drug into normal and inflamed skin (Gunther et al., 1998). This is not surprising, given that skin inflammation produces a thickening of the epidermis and thereby may enhance, rather than reduce, the barrier function of the skin (Walker et al., 2003).

Some skin diseases, such as psoriasis vulgaris, produce hyperkeratosis, which may also result in a reduced penetration o\f topically applied substances, whereas other skin diseases, such as eczema or podoconiosis (elephanatiosis), may produce rupture of the stratum corneum and thereby reduce the barrier function of the skin, resulting in increased penetration of topically applied substances (Korting et al., 1990). Low-frequency ultrasound promotes the penetration of 20-nm quantum dots through pig skin in a heterogeneous manner to an epidermal depth of ~60 nm (Paliwal et al., 2006). This penetration was considerably increased by the presence of sodium lauryl sulfate. Recently, Upadhyay (2006) has suggested that mild hyperthermia results in higher mass transport dot quantum dot antigen-labeled NP. A range of immune responses in vivo followed, leading to the suggestion that NP may be used for transdermal immunization. However, quantum dots are often cell- impermeable and require transporters to facilitate crossing over cell membranes (Zhang et al., 2006).

Work by Tinkle et al. (2003) also suggests that fluorescent dextran beads penetrate into flexed human skin with NP being found in the epidermis or dermis. However, the skin sections flexed were relatively thin (300-400 m), and the findings may alternatively be explained by penetration of particles into hair follicles (Lademann et al., 1999), which is known to be enhanced by mechanical movement (Cormier et al., 2001; Teichmann et al., 2006). In addition, the human skin was stored in a tissue medium at 4C for 24 h before the studies. Further, one should take into account the possibility that mechanical flexing of mammalian skin with a sufficiently large NP may create sufficient pressure to cause epidermal penetration (Tinkle, personal communication to M. Roberts, 2006). A lack of a diffuse fluorescence in the skin treated with fluorescent-marked NP suggests that a detachment of dye molecules from particles with subsequent penetration of the dye into the skin has not occurred (Tinkle, personal communication to M. Roberts, 2006). However, such a detachment and other sample preparation artifacts have caused interpretation issues in other studies. For example, in two recent, independent inhalation studies on ^sup 99m^Tc-marked carbon NP it has been recognized that marker substances may become detached from inhaled NP, thereby leaching into the tissue and resulting in the semblance of systemic exposure to NP, although the particles remained localized in the lungs (Wiebert et al., 2006a and 2006b). Our own unpublished work (Roberts, 2006), has shown that ZnO NP are retained at the stratum corneum surface and in follicular openings after flexing of human skin.

The fact that topical application of metals such as beryllium or nickel or their salts may lead to skin sensitization has been suggested as supporting evidence for skin penetration of NP (Tinkle et al., 2003). However, this phenomenon may not represent genuine evidence for skin penetration of particles or NP, since metals are principally thought to penetrate the skin by diffusion of their ionized, that is, atomic or molecular forms (Hostynek, 2003).

4. LOCAL AND SYSTEMIC TOXICITY OF INSOLUBLE NP USED IN SUNSCREENS

4.1 Intrinsic Toxicity

The likely toxicity of NP after topical application depends on the likely exposure of viable cells, the concentrations present, the toxicity of NP as particles, and the intrinsic toxicity of NP components or surfaces. Evidence of the low intrinsic toxicity of TiO^sub 2^ and ZnO NP is provided by their large market share (~70% and 30%, respectively) and the fact that they have been on the market since ~ 1990 and ~ 1999, respectively, without producing adverse skin or systemic effects. Zn is an essential nutrient with an average intake of 15 mg and a no-observable-effect level (NOEL) of approximately 50 mg/kg/day (Maita et al., 1981). The solubility of zinc oxide (1.6 ppm) is pH dependent and higher at the acidic skin surface. In vitro pig skin studies showed that between 1.5 and 2.3% of the topically applied ZnO NP and nil of TiO^sub 2^ was recovered in the receptor solution (Gamer et al., 2006). A recent study on skin absorption of ZnO NP reported levels of zinc over 24 h in diffusion cells using human skin that were slightly, but not significantly, different from those for a placebo formulation (Cross et al., 2007).These absorption values were an order of magnitude less than reported in previous human work (Pirot et al., 1996) and two orders less than values found by Gamer et al. in pig skin (2006). The recovery of Zn in receptor phases also reflects background Zn levels in both the placebo formulations and in the skin.

Whereas ZnO and TiO^sub 2^ appear relatively nontoxic, some carbon nanotubes, fullerene derivatives and quantum dots may have intrinsic biological activity (Hardman, 2006). Quantum dots cores can consist of metal complexes such as semiconductors, noble metals, and magnetic transition metals. Of these, indium arsenate, cadmium tellurium, and gallium arsenate appear to be quantum dots with potential intrinsic toxicity sufficient to warrant appropriate exposure limits. In addition, dissolution of the core to the elemental metal, such as cadmium, may also be associated with toxicity.

Intrinsic toxicity may also be modified by the type of particle coating used. In principle, coating with polyethylene glycol makes NP less hydrophobic and more biocompatible. Other coatings, however, may be intrinsically toxic, such as mercaptoacetic acid or immunogenic, for example, covalent serum albumin, coatings (Hardman, 2006). Ryman-Rasmussen et al. (2006) showed carboxylic acid or PEG- amine coatings of quantum dot NP produced cytotoxicity, whereas PEG- coated NP were nontoxic. The study concluded that quantum dot surface coating is a primary determinant of cytotoxicity and immunogenicity in human epidermal keratinocytes.

4.2 In Vitro Toxicity Studies on Insoluble NP in Mammalian Cells

A wealth of information on the behavior in the organism and adverse effects of small, insoluble, and foreign particles is available from studies on adverse effects of surgical implants, such as artificial hip or knee joints. Such devices may produce wear debris consisting of nano- and micro-sized particles on their bearing parts and at the interface between the device and bone (Breme and Helsen, 1998). Very high concentrations of metallic, polymeric, or ceramic debris particles (sizes: 100 nm to 10 m) have been observed in human tissues surrounding hip and knee replacements at very high densities ranging from 8.5 10^sup 8^ to 5.7 10^sup 13^ per gram of tissue. It has been recognized that the presence of high concentrations of wear debris particles in the periarticular tissues may result in inflammation and, ultimately, osteolysis (Amstutz et al., 1992; Hirakawa et al., 1996). Since these particles were only found in periarticular tissues, these findings also suggest that small, inert particles tend to remain localized and result in little, if any, systemic exposure.

The interaction with and adverse effects of debris particles on living tissue have been thoroughly investigated in relevant mammalian cell cultures. A recent study evaluated in murine fibroblasts and macrophages the role of particle size and shape in the cytotoxicity of micro- and nano-sized, insoluble ceramic particles, including TiO^sub 2^, Al^sub 2^O^sub 3^, ZrO^sub 2^, Si^sub 3^N^sub 4^, or SiC. The results suggested that larger particles of TiO^sub 2^ (1600 nm), ZrO^sub 2^ (530 nm), Al^sub 2^O^sub 3^ (590 nm), or Si^sub 3^N^sub 4^ (700 nm) were more cytotoxic to fibroblasts and macrophages than smaller TiO^sub 2^ (90 or 130 nm) or SiC (180 nm) particles, whereas dendritic particles were more cytotoxic to macrophages than spherical particles. The study results also suggested that the volume of the total amount of the particles phagocytosed by cells, and not their particle size is the decisive factor in the particle-mediated inhibition of cell proliferation and cytotoxicity (Yamamoto et al., 2004).

Another study compared the toxicity of micro- and nanosized particles of CdO (1000 nm), Ag (15 and 100 nm) MoO^sub 3^ (30 and 150 nm), Fe^sub 3^O^sub 4^ (30 and 47 nm), Al (103 nm), MnO^sub 2^ (200 nm), and W (tungsten, 27 m) in rat liver cells. The results showed no correlation of cytotoxicity with the particle size, but with the chemical nature of the test substance. Materials of known toxicity (CdO and Ag) were highly cytotoxic at all particle sizes, whereas the remaining materials showed a similar degree of cytotoxicity independently of particle size (Hussain et al., 2005). These findings are also supported by the results of Olivier et al. (2003), who showed that the cytotoxicity of 0.45and 3.45-m polystyrene or 0.43- and 2.81-m alumina particles to macrophages or fibroblasts was unrelated to their particle size. Similar results were found in an investigation on the in vivo or in vitro toxicity of metallic titanium particles in osteoblasts (Choi et al., 2005).

The toxicity of insoluble and inert particles to mammalian cells is directly correlated to their cell uptake. Many mammalian cells have a capacity for endocytosis or phagocytosis of small, insoluble particles; that is, they may actively ingest small particles. During endocytosis, materials ingested are progressively enclosed by the cell membrane, which eventually detaches to form an endocytic vesicle, whereas phagocytosis is a receptor-mediated characteristic for neutrophiles, macrophages, and dendritic cells and may result in active ingestion of insoluble particles up to 3 m (Garnett and Kallinteri, 2006). Cells that phagocytose small insoluble particles release reactive oxygen species and lysosomal enzymes in order to destroy or degrade the ingested, insoluble particles, attempting to convert them to an ineffective, safer form (Yoshikawa, 1991). When phagocytosed particles cannot be degraded, they may accumulate in the cell, resulting in oxidative cel\l damage, inhibition of cell proliferation, and, ultimately, cytotoxicity, and provoking a physiological response termed activation; the nonproliferating cell may release numerous inflammatory factors, produce inflammatory responses in adjacent tissues, and/or stimulate fibroblasts for fibrogenesis (Yamamoto et al., 2004).

This well-known sequel, a normal physiological response of phagocytosing cells to an excessive amount of insoluble particles, may result in oxidative cell damage, such as lipid peroxidation or DNA damage and cytotoxicity (Grg et al., 1988). Ultimately, oxidative cell damage may also produce genotoxic effects (Yoshikawa et al., 1991), the mechanisms of which were recently reviewed by Schins (2002). Consequently, in order to avoid false positive findings, international guidelines for in vitro genotoxicity testing in mammalian cell cultures recommend testing of insoluble compounds only up to the lowest precipitating concentration (ICH, 1996). Cerium oxide NP uptake into fibroblast cell cultures is strongly inversely dependent on particle size, and uptake is faster than would be predicted by diffusion, suggesting a biological uptake process at the cell surface (Limbach et al., 2005). Agglomeration at the cell surface is a prerequisite for uptake and emphasizes the need to relate the environmental conditions used in cell culture studies to that actually likely to be seen by tissues in vivo.

Accordingly, reports claiming the discovery of active penetration of nanoparticles into mammalian cells in vitro need to be treated with caution, especially since such findings may often be more convincingly explained by the phagocytic activity of treated cells. Given that human keratinocytes in culture are known to have considerable phagocytic capacity (Korting, 1993), reports on uptake of NP or carbon nanotubes by cultured keratinocytes (Monteiro- Riviere et al., 2005a) or other cells do not necessarily predict a potential toxicity or a risk to intact human skin or the human organism, but may simply reflect the intrinsic biological response of cultured mammalian cells to treatment with an insoluble, foreign material. This view is supported by a recent report suggesting that the principal route of uptake of quantum dots by mammalian cell cultures was via endocytosis (Hardman, 2006).

Therefore, effects such as oxidative stress, antioxidant depletion, and cytotoxicity in keratinocytes treated with carbon nanotube materials, such as reported by Shvedova et al. (2003) or Monteiro-Riviere et al. (2005a), are not surprising and do not necessarily suggest a risk for human skin, but may correspond to a normal response of mammalian cells to foreign and insoluble materials. Similarly, findings suggesting in vitro genotoxic effects of small, insoluble particles in mammalian cells do not necessarily suggest an intrinsic genotoxic activity of these materials, but may be due to the same mechanism. For example, a recent report describing a genotoxic effect of dust storm fine particles in human lymphocytes (Wei and Meng, 2006) does not necessarily suggest a genotoxic risk of desert dust to human health, but the observed effects may be more consistent with the response of mammalian cells to an excessive uptake of insoluble particles.

Similarly, the results of a recent report on the uptake of TiO^sub 2^ NP by murine brain microglia cells and subsequent generation of reactive oxygen species (Long et al., 2006), do not necessarily mean that TiO^sub 2^ particles in sunscreens are potentially neurotoxic or may cause brain damage in people, as suggested by media or NGOs (Anonymous, 2006; FOE, 2006b), but should most likely be attributed to the well-known phagocytic capacity of microglia, which has been characterized by Schilling et al. (2005). The same mechanism may also explain the findings of a recent report on oxidative damage in human bronchial epithelial cells following exposure to rutile and anatase-type TiO^sub 2^ (10,20 and 200 nm) nanoparticles (Gurr et al., 2005). In today’s climate, it is surprising that media and NGOs have not paid attention to a recent, curious report describing a high capacity for antioxidant depletion in an in vitro model of the human respiratory tract of nano-sized combustion particulates generated by burning cow dung (Mudway et al., 2005)-another, albeit hypothetical, risk of NP to human health?

It has been recently been shown that both cytotoxicity and cellular interleukin (IL)-8 release produced by the interaction of multiwall carbon nanotubes (MWCNT) with human keratinocytes were affected by the type and concentration of surfactants used to keep the hydrophobic test materials in solution, secondary to complex interactions of the surfactant with the physical shape of MWCNT as well as the viability/activity of the cells (Monteiro-Riviere et al., 2005b). A critical appraisal of in vitro assays measuring possible toxicological effects of carbon nanomaterials recognized that in vitro tests may yield conflicting results due to interference of test and control carbon materials with dye markers used in cell cytotoxicity assays (Monteiro-Riviere and Inman, 2006). Conflicting results of in vitro/in vivo test results have also been recently been described for the pulmonary toxicity of TiO^sub 2^ particles (Sayes et al., 2006; Warheit et al., 2006).

However, it is unclear whether all mechanisms of uptake of NP by mammalian cells occur via phagocytosis or endocytosis. For example, a recent report described uptake of very small TiO^sub 2^ particles (22-nm agglomerates of smaller particles) by lung macrophages and human blood cells via a nonphagocytotic mechanism (Geiser et al., 2005). Clearly more research is needed to clarify the mechanisms of cellular entry and subsequent adverse effects of NP in cultured mammalian cells.

Given these uncertainties and confounding factors, reports of in vitro studies that claim discovery of intracellular penetration and oxidative stress