Ecotoxicology

, 17:362

What can be inferred from bacterium–nanoparticle interactions about the potential consequences of environmental exposure to nanoparticles?

Authors

    • Centre for Soils and Ecosystem FunctionsRothamsted Research
Article

DOI: 10.1007/s10646-008-0217-x

Cite this article as:
Neal, A.L. Ecotoxicology (2008) 17: 362. doi:10.1007/s10646-008-0217-x

Abstract

This article collates published information regarding the in vitro antibacterial activity of both metal and carbon nanoparticles. The aims are to establish a consensus regarding modes of antibacterial activity, and to evaluate the applicability of current knowledge to prediction of likely effects of nanoparticles upon important microbial processes in environmental exposures. The majority of studies suggest that nanoparticles cause disruption to bacterial membranes, probably by production of reactive oxygen species. Contact between the nanoparticle and bacterial membrane appears necessary for this activity to be manifested. Interfacial forces such as electrostatic interactions are probably important in this respect. However, the toxicity of free metal ions originating from the nanoparticles cannot be discounted. Passage of nanoparticles across intact membranes appears to be unlikely, although accumulation within the cytoplasm, probably after membrane disruption, is often observed. To date, published studies have not been designed to mimic natural systems and therefore provide poor understanding of the likely consequences of intentional or unintentional environmental release. The limited studies currently available fail to identify any significant effects at the microbial level of nanoparticles in more complex systems.

Keywords

NanoparticlesBacteriaReactive oxygen speciesSurface chargeAntibacterial

Introduction

The growth in nanotechnology offers clear technological advances in numerous areas including medicine, manufacturing, electronics, more efficient energy production and utilization, and pollution control and remediation (Biswas and Wu 2005). However, these advances need to be balanced by clear comprehension of potential problems associated with a burgeoning availability of such novel, reactive particles to human health and the environment. Few data are currently available concerning the unforeseen consequences of increased environmental release (probably inevitable as the usage of nanomaterials increases) upon the continued efficient functioning of exposed ecosystems. One potential area of concern is the continued fertility and productivity of soil systems exposed to nanomaterials. Soils provide the key ecological services necessary for efficient food production and for environmental and human health. The general lack of knowledge regarding the fate and reactivity of nanomaterials in complex systems such as soils means that potential areas of concern have not yet been identified.

One area of concern is the bioavailability and toxicity of nanomaterials to bacteria, which may perform many critical roles required for normal ecosystem function and productivity. There is now considerable evidence from clinical isolates, such as Escherichia coli, Pseudomonas aeruginosa and Streptococcus aureus, that nanoparticles exhibit antibacterial activity (Sondi and Salopek-Sondi 2004; Morones et al. 2005; Huang et al. 2005b; see Table 1). This raises the possibility that important biogeochemical processes in soil such as carbon or nitrogen cycling may be affected detrimentally by the release of nanoparticles to soils.
Table 1

Examples of nanoparticles shown to have antibacterial activity in vitro

Nanoparticle

Organism

Reference

Ag

Escherichia coli

Sondi and Salopek-Sondi (2004)

Ag

E. coli, Pseudomonas aeruginosa, Vibrio cholera

Morones et al. (2005)

Ag

E. coli, Bacillus subtilis

Yoon et al. (2007)

Ag

E. coli

Pal et al. (2007)

Cu

E. coli, B. subtilis

Yoon et al. (2007)

Cu

E. coli, Staphylococcus aureus, Lysteria monocytogenes

Yoon et al. (2007)

MgO

E. coli, B. megaterium, B. subtilis

Stoimenov et al. (2002)

MgO

E. coli, S. aureus

Makhluf et al. (2005)

MgO

B. subtilis

Huang et al. (2005a)

MgO

B. subtilis, S. aureus

Huang et al. (2005b)

ZnO

E. coli

Brayner et al. (2006)

ZnO

E. coli

Zhang et al. (2007)

CeO2

E. coli

Thill et al. (2006)

TiO2

E. coli, B. megaterium

Fu et al. (2005)

Ag-doped TiO2

Micrococcus lylae

Zhang et al. (2003)

Pt(IV)-modified TiO2

E. coli, S. aureus, Enterococcus faecalis

Mitoraj et al. (2007)

C-doped TiO2

E. coli, S. aureus, E. faecalis

Mitoraj et al. (2007)

N-doped TiO2

E. coli

Liu et al. (2007)

N-doped ZrO2

E. coli

Liu et al. (2007)

Ag-doped Al2O3

E. coli

Chang et al. (2007)

Ag-loaded polystyrene

S. aureus, Klebsiella pneumoniase

Oh et al. (2007)

C-TiON

E. coli

Li et al. (2007)

Mixed Carbon (soot)

K. pneumoniase

Mohanty et al. (2007)

C nanotubes

E. coli

Kang et al. (2007)

nC60

B. subtilis

Lyon et al. (2006)

C nanohorns

E. coli

Miyako et al. (2007)

Polyethylenimine

S. mutans

Beyth et al. (2006)

A case study: modes of toxicity of silver ions and silver nanoparticles to bacteria

Silver (Ag) provides a good example of developing nanoparticle technology because it has a long history of use and relatively well described modes of action. Prior to 2000, most Ag from anthropogenic sources entering wastewater treatment plants was derived from photographic, printing and plating industries. In medicine, Ag is recognized as especially useful in wound management and has been used since the Eighteenth century for the treatment of ulcers (Chopra 2007). By the Nineteenth century, Ag was recognized for its antimicrobial activity and regulated for wound management in the 1920s by the U.S. Food and Drug Administration (Chopra 2007). Increasing levels of antibiotic resistance in both pathogenic and non-pathogenic bacteria has spurred the search for more effective methods of control of bacterial infections and sterilization of instruments and surfaces; Ag has become more widely used in clinical settings.

Perhaps the principal use of Ag is the topical treatment of burns, where either AgNO3 or Ag-sulfadiazine ointment have traditionally been used (Silver et al. 2006). Wound dressings, such as Silverlon (Argentum Medical) and Actisorb Silver (Johnson and Johnson), contain slow-release Ag compounds. Plastic in-dwelling catheters and heart valves impregnated with Ag+ ions are used to limit the establishment of biofilms (Dasgupta 1994; Gabriel et al. 1996). The noted antimicrobial activity of Ag+, due to poisoning of respiratory electron transport chains (Bard and Holt 2005) and components of DNA replication (Silver and Phung 2005) has also led to incorporation of Ag in non-clinical products. These range from Ag-coated domestic water filters, Microdyn, a Ag-gelatin aggregate available as an antibacterial salad-vegetable wash, to slow release Ag as a preservative in cosmetics and toiletries (Silver 2003). With this increased usage of Ag+-containing products, it is unsurprising that increasing numbers of Ag-resistant bacterial strains are being identified and isolated (Silver et al. 2006).

Bacterial resistance to Ag has been studied best with respect to the pMG101 plasmid, first isolated from a clinical Salmonella strain (McHugh 1975). The 180 kb plasmid harbours nine genes related to Ag-resistance, some homologous to other resistance mechanisms (e.g. Cadmium-Zinc-Cobalt resistance system, Czc, from Cupriavidus (Ralstonia) metallidurans). The silCBA genes encode a three-polypeptide efflux system including an inner membrane H+/cation antiporter, SilA which transfers Ag+ directly to an outer membrane (OM) protein SilC. The third protein, SilB, acts as an anchor in the inner membrane connecting to SilC in the OM (Nies 2003). Other proteins in the silP ORF105 silAB ORF96 silC silSR silE gene cluster have less well-defined functions (Silver 2003). Ag-resistance aside, Ag was identified as the most toxic metal to microbial soil communities among twelve tested (Cornfield 1977). At the molecular level, soil dehydrogenase activity is severely inhibited by addition of soluble Ag salts to soil and soil denitrification is acutely sensitive to Ag (Johansson et al. 1998). Dose-dependent reduction in denitrification activity was also observed in arable soil in response to Ag, the addition of 100 mg Ag kg−1 soil resulting in a significant decrease of copy number of the copper nitrite reductase, nirK (Throbäck et al. 2007).

The constant drive to develop antimicrobial agents to which resistance is unlikely to evolve has led to the study of toxicity of metallic nanoparticles (np) to bacteria and viruses. Of these, Ag-np continue to receive a great deal of interest. As a development of Ag-containing wound dressings used in burns injuries, dressings sputtered with Ag-np are now in use (Acticoat, Smith and Nephew, UK). Topical application of Ag-np to wounds appears to aid healing, not only because of an aggressive antimicrobial effect, but also due to implication in modulation of cytokines involved in healing (Tian et al. 2007). Antimicrobial and antiviral properties of Ag-np have been established (Sondi and Salopek-Sondi 2004; Elechiguerra et al. 2005; Lok et al. 2007). In the case of the HIV-1 virus, Ag-np of 1–10 nm exhibit preferential adhesion to gp120 glycoprotein knobs on the viral surface, inhibiting attachment to CD4 receptor sites of host cells (Elechiguerra et al. 2005). In contrast, antibacterial effects appear more due to physical disruption of cell membranes rather than inhibition of the infection process per se. Studies of the effect of relative antibacterial properties of Ag-np of different size suggest that smaller particles (9 nm) are significantly more effective in their antimicrobial activity towards E. coli than larger particles (62 nm, Lok et al. 2007). There is also evidence suggesting the specific presence of the 111-plane of Ag0 contributes to the effective antimicrobial activity (Pal et al. 2007) as a result of a relatively high Ag-atom density in this crystal plane. Intracellular localization of Ag-np has been identified in compromised cells (Sondi and Salopek-Sonodi 2004), and OM disruption by np repeatedly observed (Pal et al. 2007; Lok et al. 2007). E. coli exposure to 0.4 and 0.8 nM of bovine serum albumin-stabilised Ag-np resulted not only in OM disruption, but also accumulation of a number of precursor OM-proteins including OmpA, OmpC and OmpF, OppA and MetQ. Accumulation of these precursors in the cytoplasm suggests a cessation in function of the inner membrane-associated, ATP-dependant preprotein translocase. In the absence of a membrane potential or ATP, precursor proteins remain untranslocated (Lok et al. 2006). Although similar disruption of the OM is also observed with Ag+, it is observed at concentrations an order of magnitude higher than those with Ag-np (Lok et al. 2006). Ag-doped TiO2-np have also been shown to have effective antibacterial activity against E. coli (Thiel et al. 2007).

Surface reactivity of nanoparticles in relation to antibacterial activity

As the studies of Pal et al. (2007) demonstrate, Ag-np appear to disrupt important membrane activities and are therefore damaging rather than toxic. This is a recurrent theme in nanotoxicological studies where mode of action has been determined: ZnO-np (Brayner et al. 2006), Ag-np (Sondi and Salopek-Sondi 2004; Gorgoi et al. 2006; Pal et al. 2007), MgO-np (Stoimenov et al. 2002; Makhluf et al. 2005), CeO2-np (Thill et al. 2006), single walled carbon nanotubes (Kang et al. 2007) and chitosan-np (Qi et al. 2004) have all been shown to impair membrane architecture. A second theme is that smaller particles of MgO-np (Makhluf et al. 2005; Huang et al. 2005a, b), Ag-np (Morones et al. 2005; Lok et al. 2007) and fullerene (nC60, Lyon et al. 2006), show greater antibacterial activity then larger particles (see Fig. 1). Chemical changes are observed in membranes of nanoparticle-exposed cells (Fang et al. 2007). When exposed to nC60, membranes of the Gram-negative P. putida showed raised levels of cyclopropane fatty acids compared to unexposed cells. For Gram-positive B. subtilis, exposure resulted in an increase in either iso- or anteiso-branched fatty acids or mono-unsaturated fatty acids in an nC60-dependent manner. These changes may provide membranes greater structural stability in response to perturbation.
https://static-content.springer.com/image/art%3A10.1007%2Fs10646-008-0217-x/MediaObjects/10646_2008_217_Fig1_HTML.gif
Fig. 1

Nanoparticles exhibit an inverse relationship between antibacterial activity (here expressed as a percentage reduction in viability after 4 h exposure, compared to unexposed cells) and particle size. Data are shown for the Gram-negative bacteria Escherichia coli and Staphylococcus aureus exposed to 1 mg ml−1 of nanoparticles in nutrient broth. (Data taken from Makhluf et al. 2005)

The mechanisms by which membranes become compromised in the presence of metal nanoparticles are currently thought to involve lipid peroxidation by reactive oxygen species (ROS) such as superoxide (\( {\text{O}}^{ - }_{2} \)) and hydroxyl radicals (OH), as described for eukaryotic cells by Lovrić et al. (2005). Using TiO2-np as an example, in the presence of UV-radiation and O2, ROS are produced according to the following reactions:

$$ {\text{TiO}}_{{\text{2}}} + hv \Rightarrow {\text{TiO}}_{{\text{2}}} (\hbox{h}^{ + } + e^{ - }) $$
$$ e^{ - } + {\text{O}}_{2} \Rightarrow {\text{O}}^{ - }_{2} $$
$$ {\text{O}}^{ - }_{2} + 2{\text{H}}^{ + } + e^{ - } \Rightarrow {\text{H}}_{2} {\text{O}}_{2} $$
$$ {\text{H}}_{{\text{2}}} {\text{O}}_{{\text{2}}} + {\text{O}}^{ - }_{2} \Rightarrow { }^{\bullet}{\text{OH}} + {\text{OH}}^{ - } + {\text{O}}_{2} $$
$$ {\text{H}}^{{\text{ + }}} + {\text{H}}_{{\text{2}}} {\text{O}} \Rightarrow { }^{\bullet} {\text{OH}} + {\text{H}}^{{\text{ + }}} $$

The idea that ROS are responsible for the observed membrane damage due to metal nanoparticles is supported by the fact that superoxide dismutase, which catalyses the dismutation of \( {\text{O}}^{ - }_{2} \) into O2 and H2O2, and catalase, which catalyses the decomposition of H2O2 to H2O and O2, when added to Ag/Al2O3-np suspensions significantly reduces damage to E. coli cell membranes (Chang et al. 2007). True toxicity mechanisms, as opposed to membrane damage, may also be involved however, such as the release of metal ions into the aqueous phase. TiO2-np presented as a coating on cellulose fibres, still show antibacterial activity, though somewhat reduced, under dark conditions (Daoud et al. 2005). However, it is difficult to discern whether the apparent drop in culturable cells in experiments performed with TiO2-np-coated cellulose in the absence of illumination, is due to toxicity or from a lack of carbon and energy source due to the coating on the cellulose surface. A mechanism by which free radicals may be produced under dark conditions has been proposed for biotin-functionalised CdSe/ZnS quantum dots (cadmium selenide nanoparticles capped with zinc sulphide) which involves, as a first step, the oxidation of the ZnS cap producing soluble SO2 (Green and Howman 2005). In solution, the \( {\text{SO}}^{{ - \bullet }}_{2} \) free radical is formed, which can further oxidise to from the ROS \( {\text{O}}^{ - }_{2} \) and then OH. Formation of ROS in the absence of photons has not been studied for nanoparticles shown to possess antibacterial activity. Toxicity via free ion activity has been reported for Cu-np-containing polymer composites (Cioffi et al. 2005) and Ag-np impregnated silicone (Furno et al. 2004). In fact, there is some evidence to suggest that the toxicity of Ag-np themselves may in fact, in part, be due to chemisorbed Ag+ ions, as only oxidised nanoparticle suspensions exhibit the well described toxicity (Lok et al. 2007). Ag-np synthesised under a N2-atmosphere show no antibacterial activity. Oxidised Ag-np are also not toxic to Ag-resistant E. coli strains (Lok et al. 2007).

To the extent that ROS may be involved in nanoparticle toxicity, direct contact between nanoparticles and the bacterial membrane appears essential (Kang et al. 2007; Thill et al. 2006; Stoimenov et al. 2002). The charge which develops at the interface between a particle (or bacterial cell) and the surrounding medium may arise by any of several mechanisms. Among these are dissociation of ionogenic groups at the particle surface and differential adsorption from solution of ions of different charge into the surface region. The development of a net charge at the cell surface affects the distribution of ions in the neighbouring interfacial region, resulting in an increased concentration of counter ions (ions of opposite charge to that of the particle surface) close to the surface. Thus, an electrical double layer (EDL) is formed in the region of the particle–liquid interface. The EDL consists of an inner region, which includes ions bound relatively strongly to the surface (including specifically adsorbed ions) referred to as the Stern layer, and an outer, or diffuse, layer in which the ion distribution is determined by a balance of electrostatic forces and random thermal motion. The potential in this region, therefore, decays as the distance from the surface increases until, at sufficient distance, typically on the order of nm, it reaches the bulk solution value, conventionally taken to be zero. A boundary becomes established between the strongly sorbed ions and the diffuse layer, the inner potential at the “slipping plane” between these two regions is referred to as the zeta-(ζ) potential, and depends upon cell surface charge density at the slipping plane. The ζ-potential is indicative of the effective charge that a particle will experience as it approaches a surface (Hunter 1993).

Bacteria typically posses a net negative charge. Reported ζ-potentials for various nanoparticles are variable, as might be expected for particles with widely differing chemistries and surface coatings. Carbon-np exhibit a potential of approximately 0 mV (Mohanty et al. 2007). For 130 nm sized silica-np, ζ-potential ranges from −20 to −50 mV (Singh and Song 2007) dependent upon solution pH. With an isoelectric point (IEP) of pH 8.1, ζ-potentials for zero valent iron-np also vary from 10 mV at pH 7.8 to −30 mV at pH 8.5 (Sun et al. 2007). However, polyvinyl alchohol-co-vinyl-acetate-co-itaconic acid coatings on the nanoparticle surface result in a consistently negative potential across the entire pH range. Oligochitosan stabilised 8 nm Ag-np have a pH-dependant ζ-potential which ranges from 28 mV at pH 2.4 to −56 mV at pH 11 (Long et al. 2007). Freshly prepared Au-np have a ζ-potential of −30 mV at an undisclosed pH (Zhu et al. 2005), but once coated with thioglycolic acid/cetyltrimethyl ammonium bromide the particles possess a charge of 23 mV. Un-capped TiO2-np have an IEP of approximately pH 5.8 and thus at physiological pH of 7 possess a ζ-potential of −20 mV (Liufu et al. 2005). In contrast, un-capped ZnO-np have an IEP of approximately pH 9.4 and are thus positively charged at circumneutral pH (approximately 30 mV at pH 7, Liufu et al. 2004). Polyethylene glycol surface coating of the ZnO-np results in only small changes in ζ-potential. For single-walled carbon nanotubes, an IEP of pH 7 has been observed, which upon hydroxylation shifts, so that ζ-potentials become negative (between −30 and −70 mV) over the entire range of pH studied (Hu et al. 2005).

This small collection of measured potentials clearly demonstrates that nanoparticle surface charge is not only influenced by solution pH, and incidentally also by solution ionic strength (I), but also by the nature of any surface coatings present. Generally, we may assume that positively charged nanoparticles are likely to experience greatest electrostatic attraction to negatively charged bacterial surfaces (see Fig. 2.1), however the veracity of this statement is difficult to gauge since ζ-potential estimates are seldom reported in toxicological studies. For instance, bacterial cellular damage reported by ZnO-np is consistent with the positive charge attributed to the uncoated particles. However, electrostatic repulsion is to be expected between bacterial cell surfaces and un-capped TiO2-np at physiological pH since both are negatively charged, and yet toxicity is still observed. Clearly, detailed studies of the effect of various surface functionalisation of nanoparticles upon ζ-potential, and the subsequent effects upon nanoparticle-bacterial cell interactions and membrane damage are required to demonstrate that close surface contact is indeed required for nanoparticle antibacterial activity.
https://static-content.springer.com/image/art%3A10.1007%2Fs10646-008-0217-x/MediaObjects/10646_2008_217_Fig2_HTML.gif
Fig. 2

A conceptual model for antibacterial activity of nanoparticles and the effect of environmental exposure. The consensus mechanism for the antibacterial activities of nanoparticles is based upon contact-mediated lipid peroxidation via production of reactive oxygen species. Interfacial forces, especially electrostatic, will control contact between nanoparticles and the bacterial membrane. Attractive forces are generated between positively charged nanoparticles and negatively charged bacterial cells (1a). A repulsive force is generated between bacterial cells and negatively charged nanoparticles (1b). Processes which alter the surface charge of nanoparticles (1c) may indirectly alter the interaction between affected nanoparticles and bacterial cells. Once in contact with bacterial membranes, nanoparticles cause lethal cell damage by producing reactive oxygen species (ROS, 2d), eventually allowing ingress of nanoparticles into the periplasm/cytoplasm (2e). Exopolymers (2f)—secreted carbohydrates and proteins, may prevent nanoparticles or ROS contacting the cell membrane thus preventing cell damage. Exopolymers are often associated with attached cells in biofilms, and may result in biofilm-associated cells being more resistant to damage from nanoparticle exposure. Following environmental release of nanoparticles, a number of processes may also reduce the likelihood of membrane–nanoparticle contact: physico-chemical changes in the surrounding medium, such as increases in ionicity may result in the formation of clumps as repulsive forces between the primary particles are reduced, these clumps of primary particles may fall out of suspension (3g); complexation by natural organic matter (3h) such as humic and fulvic acids may also reduce the antibacterial effects of nanoparticles; inorganic surfaces such as mineral particles in soils may also prevent direct nanoparticle-bacterium contact by trapping nanoparticles at the surface (3i)

Suspensions of nanoemulsions (400–800 nm) of either 64% soy bean oil, 8% tri-η-butyl phosphate, 8% Triton X-100 and 20% water (8N8), or 1% Tween-60, 0.1% halogen-containing cetylpyridinium chloride, 3% glycerol monooleate, 1% soya sterols, 20% soya bean oil and 75% water (W60C) have antibacterial effects consistent with their surface charge (Hamouda and Baker 2000). The positively charged W60C (ζ-potential = 32 mV) has a clear antibacterial effect when diluted in water, manifested by membrane disruption. Negatively charged 8N8 (ζ-potential = −13 mV) has no effect upon membrane structure. In support of the relationship between electrostatic attraction and membrane disruption, no toxicity is observed for W60C when diluted in a complex bacterial growth medium (Hamouda and Baker 2000), consistent with collapse of the EDL at high ionic strengths and thus loss of electrostatic attractive forces between the nanoemulsion and the bacterial surface. The requirement for close surface contact between cell and nanoparticle for antibacterial activity is also supported by studies with bacterial cells attached to surfaces as biofilms. Planktonic cells, typically associated with little extracellular material (secreted carbohydrates and proteins), are readily susceptible to membrane damage by nanoparticles. On the other hand, attached biofilm cells, which often secrete copious extracellular material, are protected from direct contact with nanoparticles and do not exhibit typical sensitivity to nanoparticles (Liu et al. 2007). It appears that the exopolymers secreted by biofilm-associated cells may either absorb harmful ROS, or prevent nanoparticles contacting the cell surface in the first place, either way preventing membrane damage (Fig. 2.2).

Studies of both MgO-np and ZnO-np have observed the presence of nanoparticles in the cytoplasm of damaged cells and inferred transport of particles across both the outer and inner membranes (Stoimenov et al. 2002; Brayner et al. 2006; Sondi and Salopek-Sondi 2004). Pinocytosis, the mechanism by which nanoparticles are taken up by eukaryotic cells (see for example Shukla et al. 2005), cannot be responsible for uptake in bacteria. One possible route for nanoparticle ingress is via various pores in the OM. The largest pores in the bacterial OM are probably those used to secrete mature proteins, either onto the OM surface or into the wider milieu. For example, the secretin (GspD) family of proteins are associated with the OM of various Gram-negative bacteria, forming ring-shaped pore structures (see Filloux 2004). In P. aeruginosa, elastase is secreted in its folded form with a maximum width of 6 nm via GspD (Thayer et al. 1991). Estimates put the diameter of the pore to be as large as 9.5 nm (Bitter et al. 1998). Such pores are theoretically large enough to allow passage of 1–9 nm nanoparticles. However due to its large size, the pore is probably gated (closed) to prevent loss of cytoplasm. Effectively, the channel is only open when it is required to transport proteins across the OM (Filloux 2004).

Thus, it is likely that nanoparticles are prevented from crossing even these largest of OM pores. Observation of nanoparticles in the cytoplasm most probably arises from Brownian diffusion after the membrane is disrupted. In support of this, Ag-np accumulation within P. aeruginosa has been observed in cells with membranes disrupted by monocyclic β-lactam and chloramphenicol antibiotics (Kyriacou et al. 2004; Xu et al. 2004). The antibiotics act to increase the permeability and porosity of bacterial membranes. Ag-np as large as 80 nm have been identified within the cytoplasm of antibiotic-treated cells. The incidence of such large particles within cells is positively correlated with antibiotic concentration. Presence of low numbers of cells containing Ag-np in the absence of antibiotics probably reflects the natural incidence of cells with disrupted membranes (i.e. moribund) in batch culture.

In summary, the death of bacteria in the presence of various nanoparticles appears to be due to membrane disruption, possibly arising from the formation of ROS. Close contact is necessary for membrane disruption to occur, and it is unlikely that nanoparticles cross into the cytoplasm until membranes become sufficiently porous due to peroxidation. The toxicity of free ions may be important, however, since very little direct comparison is made between free ions and nanoparticles this remains difficult to assess. The challenge for estimating the impact of nanoparticles on the wider environment is to place this model of nanoparticle antibacterial activity in the context of a greatly increased complexity.

Environmental effects of released nanoparticles

Based on the examples given above, there is clear evidence that under certain circumstances, both metal and carbon nanoparticles can cause damage and toxicity to planktonic bacteria. What is still lacking is knowledge of the effect of nanoparticles in complex systems; published experiments to date have not been designed to examine such questions.

The same fundamental processes which determine the transport of colloids through sedimentary systems—size exclusion, where particles are prevented from entering pores of certain sizes due to physical constraints and filtration, where particles are removed from the aqueous phase by surface capture, diffusion and sedimentation, will also affect the transport of nanoparticles (Fig. 2.3). Nanoparticles released into the environment are likely to have relatively short residence times due to their efficient transport, via Brownian diffusion, to surfaces. Once at a surface, interfacial forces such as electrostatic and hydrophobic interactions ultimately determine a particle’s mobility. Water soluble fullerol (C60[OH]m, m = 22–26) exhibits the greatest mobility, TiO2-np and ferroxane (FeOOH-np) the least (Biswas and Wu 2005). Aggregation of nanoparticles will tend to limit mobility and is caused by a number of factors, not least, increasing I of the surrounding aqueous medium (see for example Sano et al. 2001). As already discussed, this will lead to a collapse of the EDL around particles and a loss of charge. Under conditions of reduced electrostatic charge, and therefore electrostatic repulsion, particles tend to aggregate. Carbon nanotubes, which express little surface charge (Mohanty et al. 2007) and are extremely hydrophobic tend to aggregate in aqueous suspension, but this may be overcome by sorption of natural organic matter (humic and fulvic acids) to the particle surface (Hyung et al. 2007). Zero-valent iron nanoparticles are also rapidly coated by humic acids and other organic molecules (Klupinski et al. 2004) also preventing aggregation (Fig. 2.3). Surface modification upon environmental exposure of these highly reactive particles will therefore modify their behaviour with implications for both transport and reactivity (e.g. ROS formation).

A second issue concerning toxicity of environmentally released nanoparticles, concerns the likelihood of close contact between bacterial membranes and nanoparticles, surface modified or not. Clearly, our current knowledge of the damage caused to bacterial membranes by contact with nanoparticles is derived from experimental systems—typically in aqueous systems with high bacterial cell densities and a relatively low surface area to volume ratio, in which contact between nanoparticles and the bacterial membrane is guaranteed. In natural systems, with a high surface area and many reactive particles (surfaces), the interaction between a nanoparticle and a bacterial cell is likely to be a less common event. Add to this the fact that bacteria have a tendency to attach to surfaces in aquatic (Kjelleberg 1984) as well as subsurface and soil environments (Ekendahl and Pedersen 1994; Pedersen 2001; Pedersen et al. 1996) suggests that attached bacteria (biofilms) may constitute a significant proportion of the subsurface bacterial community. Furthermore, estimates of cell activities suggest that attached bacteria exhibit a higher activity per cell compared to unattached bacteria (Pedersen 2001). Thus, in natural systems, bacterial communities will be dominated both numerically, and metabolically, by attached cells. It is likely therefore, that biofilm communities are a more pertinent model system for consideration of nanoparticle toxicity in environmental systems than planktonic cells: as we have already seen, nanoparticles may exhibit quite different effects upon biofilms cells compared to planktonic cells (Liu et al. 2007).

There are therefore a number of questions that need to be addressed if we are to understand potential threats posed by the either willful or accidental release of nanoparticles in the environment: are surface-modified nanoparticles equally as reactive towards target bacterial systems (biofilms), as laboratory-based experiments with planktonic cells suggest; and what are the critical issues associated with the transport of nanoparticles to biofilms in complex, porous systems?

Few studies are currently available which have attempted to deal with such complex issues. While the caecum and duodenum of Coturnix coturnix japonica, the Japanese Quail, are clearly not an adequate model for the wider environment, studies of the effects of Ag-np upon gut microflora (Sawosz et al. 2007) do encompass greater levels of complexity than is achieved in vitro. Studies where C. c. japonica were provided 2–5 nm Ag-np in drinking water at varying concentrations for 12 days, suggest that no significant changes in culturable enterobacteriaceae numbers were evident in the caecum when compared to animals which were not exposed to Ag-np, even at Ag-np drinking water concentrations of 25 mg kg−1, higher than the in vitro minimal inhibitory concentration of 10 mg kg−1. A slight increase in numbers of culturable Gram-positive bacteria including Lactobacilli was observed at 5 mg kg−1. Increased environmental complexity does therefore appear to have an influence upon the perceived antibacterial activity of nanoparticles. It is also interesting to note that gut microflora typically exists as biofilms attached to the gut wall, possibly protected from contact with nanoparticles by secreted exopolymers.

When applied to silty clay loam surface soil nC60, toxic (minimal inhibitory concentration, in vitro of approximately 0.5 mg l−1) to the common soil bacterium B. subtilis (Lyon et al. 2006; Fang et al. 2007), is reported to have no influence upon soil bacterial communities (Tong et al. 2007). Employing soil basal and glucose-induced respirometry, and phospholipid fatty acid (PLFA) and PCR-DGGE 16S rRNA gene-based community profiling, no significant effect of adding nC60 at a rate of 1 μg g−1 soil could be established. Even the addition of a granular C60 at a rate calculated to saturate sorption sites and salt complexation in the soil (1 mg g−1 soil) failed to produce any observable effects. However, clear changes in the PLFA-profile of soils exposed to both nC60 and granular C60 were observed, the key differences being due to changes in the amount of 16:0 saturated PLFA present (cf. changes in lipid composition of P. putida and B. subtilis exposed to nC60 reported by Fang et al. 2007). In support of these empirical studies, risk assessment of the impact of Ag-np release from consumer products into freshwater aquatic environments suggests microbial communities in wastewater treatment plants are at little risk from the increase in the Ag-np burden of wastewater (Blaser et al. 2008). The lack of effect of nanoparticle exposure in complex systems is in contrast to the effect of Ag+ addition to a Swedish surface soil at a maximum rate of 100 μg g−1 (Throbäck et al. 2007). Significant reductions in the rate of denitrification were observed, as was reduction in the copy number of the copper nitrate-reductase-encoding nirK. A clear population shift to novel, presumably Ag-resistant nirK expressing strains was also identified.

Conclusion

Nanoparticles of various chemistries can cause damage to bacterial membranes and exhibit clear antibacterial properties as a consequence. Figure 2 addresses the issues discussed in this review which appear critical in determining antibacterial activities of nanoparticles. In simple experimental systems, interfacial forces probably exert significant influence upon bacterium–nanoparticle interactions (Fig. 2.1). Unfortunately, it remains extremely difficult to predict the likely consequences of introduction of nanoparticles into the wider environment based upon the limited scope of available data. If a robust, predictive model describing the likely effects of environmental exposure to nanoparticles upon activity and function of microbial communities is to be developed, a number of complex issues need to be addressed. These include the likely probability that nanoparticles of varying chemistry will be brought into close contact with the correct target, probably bacterial biofilms (Fig. 2.2), and that if nanoparticles are brought into contact, that they retain some reactive capability, that is to say that the surface is not already passivated and unreactive. At present these complex issues remain uninvestigated and the studies that are available, suggest that in complex systems little antibacterial effect of nanoparticles can be determined. Critical to our improved understanding of the interaction between bacterial cells and nanoparticles, is the establishment of a relationship between the surface charge of uncapped and surface-functionalised nanoparticles and observed antibacterial activity. Also, it is important from the point of view of predicting the relative risk in exposure of the wider environment to nanoparticles, to compare antibacterial effects of nanoparticles to, in the instance of metal nanoparticles, the corresponding free metal ion. Finally, it will be necessary to use the experience gained from relatively simple in vitro systems to perform increasingly complex experiments, ultimately using methods designed to trace nanoparticles in porous media (Fig. 2.3), such as isotopic labeling perhaps, to begin to understand the processes controlling transport of nanoparticles to target organisms.

Acknowledgements

The author wishes to thank Nadine Kabengi, Keith Goulding and Steve McGrath for constructive comments on an early version of the manuscript. Rothamsted Research receives grant-aided support from the UK Biotechnology and Biological Sciences Research Council.

Copyright information

© Springer Science+Business Media, LLC 2008