Size- and Ligand-Specific Bioresponse of Gold Clusters and Nanoparticles: Challenges and Perspectives

Abstract

This review gives an introduction to the chemical and physical properties of gold clusters and nanoparticles (NPs) and reflects the present understanding how such particles interact with biological systems in vitro and in vivo. It will acquaint the reader with the basic principles of interaction from a chemical point of view and illustrates perspectives that arise for the application of gold nanoparticles (AuNPs) in biological environments.

1 Introduction

The focus of this review article is the discussion of our present understanding of the interaction of gold nanoparticles (AuNPs) with biological materials in vitro and in vivo, called “bioresponse,” as well as the illustration of perspectives that arise for the application of AuNPs in biological environments. The often applied definition of “nanoparticle” – namely to consist of species smaller than 100 nm – is, at least from a scientific viewpoint, too simple. The name nanoparticle should be linked with a characteristic change of properties as is, for instance, the case if AuNPs become smaller than approx. 100 nm when their dispersions in sols or liquids spontaneously become violet or red colored due to the surface plasmon resonance (SPR). Two more precise definitions of nanoscience are therefore cited in Sect. 2.

A prerequisite to understand bioresponse processes is a fundamental knowledge of the influence of size, shape, and surface constitution of the particles. As will be discussed in detail in Sect. 2, the size of a particle is the most important factor influencing predominantly its electronic structure. A gold particle, small enough to restrict the mean free path of the electrons, behaves significantly different from bulk gold. This effect is called “size quantization,” saying that such species no longer follow classical physical laws but instead they obey quantum mechanical rules, traditionally used to describe the electronic nature of atoms and molecules. Metal NPs in general and AuNPs in particular have to become very small to reach that state, usually smaller than 2 nm. A prominent example is the Au₅₅ cluster (the expression cluster is generally used for NPs with an exact chemical formula, see also Sect. 2) with a diameter of 1.4 nm. It exhibits size quantization properties even at room temperature, whereas larger particles reach this state only at very low temperatures and smaller clusters exhibit discrete electronic states of molecular orbitals. Examples are [Au₃₉(PPh₃)₁₄Cl₆]Cl₂ [1], [Au₁₄(PPh₃)₈(NO₃)₄] [2], [Au₁₃(PR₃)₁₀Cl₂]³⁺ [3], or Au₁₁(PR₃)₇X₃ [4–6] (X = halogen, thiolate).

The extraordinary position of Au₅₅ in the transition range between molecule and extended solid has frequently been demonstrated by its physical properties. Among others, it exhibits quantized charging effects associated with adding an extra electron to a small capacitance, which give rise to applications as a single electron switch at room temperature. Apart from its electronic properties, Au₅₅, usually represented in a ligand-protected version, namely Au₅₅(PPh₃)₁₂Cl₆, has a second extraordinary feature: its stability. As will also be discussed in more detail, Au₅₅ is the second of the series of the so-called “full-shell clusters.” It consists of a hexagonal close-packed structure like bulk gold and has a cuboctahedral shape. Full-shell clusters follow the composition 10 n ² + 2 for the number of atoms in the nth shell. Particles following this rule are especially stable, as has been demonstrated in a variety of ways. The stability of AuNPs has indeed an important influence on bioresponse, because of the interaction mechanisms between NP and the relevant biosystems. As ligand-free AuNPs cannot be used because of their instability, only ligand-protected species are used. The interaction with any kind of biomolecule and the metal core can only happen, if the original ligands are either completely or at least partially released or replaced during the chemical processes. This means that the protecting ligand shell must have a special nature. If ligand molecules are not sufficiently labile to be removed from the particle’s surface, the interaction between the particle and the biological environment only occurs via the protecting skin and not by the metal itself. Phosphine ligands in combination with gold usually fulfill these conditions. Au₅₅(Ph₂PC₆H₄SO₃Na)₁₂Cl₆, a water-soluble derivative of the original compound Au₅₅(PPh₃)₁₂Cl₆, therefore plays the dominant role in this article. Larger and smaller particles are introduced for comparisons as well as to highlight the emerging field of applications for AuNPs in diagnostics and therapy.

Size, stability, and electronic properties indeed determine bioresponse: Sect. 3 illustrates in vitro applications by discussing the interaction of AuNPs with proteins and cells. While these applications utilize the size-dependent properties of AuNPs as an analytical probe, the molecular mechanisms occur in the ligand shell of the molecules. In contrast 1.4 nm-sized Au₅₅ with its weak-binding phosphine ligands turned out to be very cytotoxic. This is demonstrated in series of tests with human cancer cell lines. Two reasons for the cytotoxicity have been proposed: (1) the size of the 1.4 nm Au₅₅ clusters fits perfectly to the height of the major grooves of DNA (1.3–1.5 nm) and thus may block transcription of DNA and (2) it induces the formation of reactive oxygen species (ROS) as a consequence of its electronic properties, leading to oxidative damage of neighbored biomolecules and subcellular units. Comparisons with smaller and larger AuNPs, decorated with the same ligand molecules like Au₅₅, clearly show a much less or even no toxicity, supporting the assumption that Au₅₅ has a very special bioresponse.

Properties of AuNPs in vivo are introduced in Sect. 4. These experiments inform about distribution in a living body (exemplified by means of rat data), of course depending on the kind of administration. Again, size dependency plays the dominant role. Among a series of AuNPs from 1.4 up to 200 nm, only the 1.4 nm Au₅₅ species distribute in all relevant organs, whereas larger particles are accumulated up to 97% in the liver. A further aspect of interest is the role of surface charge in relation to biodistribution experiments with positively and negatively charged 2.8 nm Au particles show little differences. Positively charged species are somewhat less assembled in the liver than negatively charged particles. Furthermore, a new model system for in vivo analyses, i.e., the zebrafish, is introduced with its potential to analyze the properties of AuNPs in whole animal tests.

The final section of this article (Sect. 5) deals with perspectives and challenges of the above findings especially in medicine. Diagnostic and therapeutic aspects as well as drug delivery systems are already partially realized or can be prognosticated (Fig. 1).

Fig. 1

Schematic presentation of the size-dependent properties of gold, spanning the size range from individual atoms via ligand-stabilized clusters and colloidal particles up to macroscopic bulk metal. The shaded areas indicate size ranges, where specific properties and features of gold occur. Atoms, molecules, and molecular clusters exhibit an electronic structure with discrete electronic states, while colloidal particles as well as bulk gold exhibit collective electronic states. In the size range of clusters and colloids, biological effects may occur due to complementarity in shape to biomolecules and cellular subunits. In this size range the surface plasmon resonance (SPR) dominates the optical properties and most of the size-dependent catalytic properties of gold nanoparticles occur. The biocompatibility increases with sizes above the typical cluster size. The marked areas are not drawn to scale

2 Size-Dependent Properties of Metal Nanoparticles

The physical and chemical properties of defined materials, for instance elements or chemical compounds in the solid state, depend on size and shape of the particles under consideration. The size dependency of the properties usually becomes obvious in the nanometer regime. Nanotechnology is based on such effects which can have very different appearance. Whereas the bright color of bulk gold is based on relativistic effects [7–9], the red, purple, and blue color of gold nanoparticles (AuNPs) in the size regime of ca. 15 and 50 nm is based on the so-called surface plasmon resonance (SPR). It is quantitatively described by the Mie theory [10]. The reason for the SPR is to be seen in a size- and shape-dependent collective electron oscillation by interaction with visible light. SPR for copper, silver, and gold is in the visible region and superimposes the relativistic effect, covering the typical color of bulk metals. For other metals it appears in the UV region and is therefore not observable with the naked eye.

Numerous other size effects such as changes of melting points, characteristic magnetic properties, and electronic behavior or the kind of interaction with other species are known and play dominant roles in what we call “nanotechnology.” One definition of nanotechnology, considering these different effects, is the following [11]:

Nanotechnology comprises the emerging application of Nanoscience. Nanoscience deals with functional systems either based on the use of subunits with specific size-dependent properties or of individual or combined functionalized subunits.

The so-called scaling effects, i.e., continuous changes of properties during downsizing a material, are excluded because nano-effects appear only below distinct sizes, resulting in properties which characteristically differ from the macroscopic counterparts. Another similar definition, given by the Royal Society and the Royal Academy of Engineering, also considers the use of molecules and macromolecular systems [12]:

Nanoscience is the study of phenomena and manipulation of materials at atomic, molecular and macromolecular scales, where properties differ significantly from those at a larger scale. Nanotechnologies are the design, characterization, production and application of structures, devices and systems by controlling shape and size at the nanometre scale.

These definitions include the development of molecular motors and machines, molecular switches, single-electron memories, and, especially considered in this review article, the interaction of metal nanoparticles with biological systems. Why are size-dependent physical and chemical properties decisive for interactions with biosystems and responsible for bioresponse? These connections shall be briefly discussed in this section.

A single metal atom is not a metal in a classical sense. A huge number of atoms are necessary to reach the metallic state, for instance electric conductivity or metallic luster. Corresponding questions arise for other solid elements, for instance silicon. A single Si atom is not a semiconductor. The basic question therefore arises, namely how many atoms are necessary to reach the metallic state and what happens on the way to this state? In spite of various theoretical calculations, numerous experimental results during the last two to three decades have given us some fundamental answers.

The electronic situation in an atom or molecule is best described by following quantum mechanical rules (size quantization). Bulk systems consisting of an infinite number of atoms can widely be described by the laws of classical physics for bulk materials, based on the statistics of infinite numbers of electrons. Figure 2 illustrates the stepwise transition from a three-dimensional (3D) piece of metal (a) via a two-dimensional quantum well (2D) (b), a quantum wire (1D) (c), and finally a quantum dot (0D) (d), considering the relation between the density of states and the energy.

Fig. 2

Transition from a three-dimensional (3D) particle to a 0D quantum dot, correlating with the continuous discretion of the energy states

On the way from a 3D system to a 2D quantum well, the relation between density of states and energy develops steps, spikes on the way to a 1D quantum wire and, finally, when small enough (0D), discrete energy levels characterize the relationship like in an atom or molecule. The development of a 0D quantum dot can formally also be considered by downsizing a metal particle as is indicated in Fig. 3.

Fig. 3

Sketch of the transition from a metallic bulk state (a) via a large cluster (b) to a small molecule (c) indicating the changes of the electronic situations

The continuous energy band of s- and d-electrons in the bulk state (a) splits off when the particle is small enough (b, large cluster/nanoparticle) and forms sharp energy levels when a molecular cluster¹ is reached (c). What are the experimental results?

The most relevant physical method for investigating individual metal clusters and nanoparticles electronically is to determine the current (I) voltage (U) characteristic using a scanning tunneling microscopy (STM) tip to contact an individual particle on a conductive surface via the obligatory ligand shell. With this technique, single-electron transitions (SET) between tip, particle, and substrate can be observed, if some conditions are fulfilled. The nonconducting protecting ligand shell plays an important role, since neither the STM tip nor the conductive substrate touches the particle itself. SET can only be observed if the electrostatic energy E _C = e ²/2C (C = capacity) is very large compared to the thermal energy E _T = k _B T (k _B = Boltzmann’s constant). Since C = εε₀ A/d (ε = dielectric constant, ε₀ = electric field constant, d = distance of electrodes from metal core, A = surface area), the thickness of the ligand shell is important for the value of C. At very low temperatures single-electron transitions can be observed using rather large nanoparticles. For instance, the I–U characteristic of a 15 nm Pd particle, protected by a shell of H₂NC₆H₄SO₃Na molecules and investigated at 295 K, clearly shows bulk properties, i.e., a linear I–U relationship following Ohm’s law. However, at 4.2 K a so-called Coulomb blockade (CB) occurs, indicated by an interruption of conductivity between −55 and +55 mV by a single electron [13]. Figure 4 shows the I–U characteristics at both temperatures.

Fig. 4

Current (I)–voltage (U) characteristic of a ligand-protected 15 nm Pd particle at 295 K and 4.2 K. Ohm behavior is observed at 295 K, whereas a Coulomb blockade is existent at 4.2 K. Figure reprinted with kind permission from [13]

Of course, the existence of nanoparticles showing a Coulomb blockade already at room temperature is of high interest for future applications, for instance, as single electron switches. Indeed, such a particle (cluster) has been found with Au₅₅(PPh₃)₁₂Cl₆ [14]. The diameter of the Au₅₅ core is only 1.4 nm. The I–U characteristic of an individual cluster clearly exhibits Coulomb blockade at 295 K between −500 and +500 mV as can be seen in Fig. 5 [15].

Fig. 5

I–U characteristic of Au₅₅ (PPh₃)₁₂Cl₆ at 295 K exhibiting a well-expressed Coulomb blockade between −500 and +500 mV. Figure reprinted with kind permission from [15]

The Au₅₅ cluster has another special property: it belongs to the so-called very stable full-shell clusters, species having an icosahedral or a hexagonal close-packed cuboctahedral shape, consisting of a distinct number of atoms per shell around 1 central atom, namely 10n ² + 2 atoms (n = number of shell). The smallest full-shell Au cluster with n = 1 consists of 1 + 12 atoms, well known since 1981 with icosahedrally structured Au₁₃ cores [3, 16]. Numerous full-shell clusters of other metals have become known in the meantime, for instance the four- and five-shell clusters Pt₃₀₉ [17] and Pd₅₆₁ [18]._. Whereas the ligand-protected Au₁₃ cluster has still typical molecular properties, the Au₅₅ species is just at the borderline between molecule and bulk and is therefore often called as “metal in the embryonic state.”

The answer to the aforementioned question, why electronic and structural properties of clusters or nanoparticles are of relevance with respect to bioresponse, is not very surprising. First of all, the stability of metal particles in a biological medium is of high importance because less stable species would decompose, forming smaller or larger species with different properties. So, a stability guaranteeing structure and size is a condition for reliable results. The electronic structure is responsible for the chemical activity of a particle and so influences fundamentally the interactions with other materials. Some examples of these two factors are given.

Au₅₅ and various smaller and larger gold nanoparticles and even bulk gold surfaces have been investigated with respect to their activity against oxygen in an oxygen plasma by means of X-ray photoelectron spectroscopy (XPS) [19]. Except Au₅₅, all other species became partially or quantitatively oxidized! Why not Au₅₅? This is because of its structure. Oxidation of the cuboctahedral structure would afford too much energy so that oxidation, even under the exceptional conditions in an oxygen plasma, does not happen.

The special stability of the Au₅₅ nucleus becomes additionally visible when again AuNPs, smaller and larger than Au₅₅, are contacted with indium vapor. Bulk gold forms an alloy of the composition AuIn₂. All other NPs react easily with In vapor, but again not Au₅₅. It remains unreacted [19].

Theoretical [20] and experimental results [21] show that very small AuNPs are most active in oxygenation catalysis. Especially particles below ~3.5 nm are the most active species, indicating that electronic properties play a decisive role in chemical processes. Au₅₅(PPh₃)₁₂Cl₆ with a core diameter of 1.4 nm, supported on inert materials, turned out to be the most active catalyst for the selective oxidation of styrene by dioxygen [21]. In this case, particles larger than ~2 nm are completely inactive. This finding is of high importance considering the cell toxicity of Au₅₅ species due to oxidative stress (reactive oxygen species, ROS) [22]. Though larger AuNPs also cause oxidative stress, the 1.4 nm cluster is the most active particle. Oxidative stress is caused by activation of dioxygen when touching the metal surface however, without oxidation of Au atoms. Details will be given in Sect. 3.

Another result indicating the special role of Au₅₅ clusters shall briefly be mentioned. It is the relaxation behavior of excited electrons, compared with two other particle sizes [23]. If 0.7 nm (Au₁₃), 1.4 nm (Au₅₅), and 15 nm AuNPs are treated by femtosecond laser pulses, the corresponding relaxation times are size dependent. They are very different between Au₅₅ and the other two species. The relaxation time is dependent on the electron–phonon coupling and the electron surface collision. In larger particles it is dominated by the weakening of the electron–phonon coupling, resulting in a slowdown of the relaxation. On the way from 15 to 1.4 nm, electron surface collision increases and consequently the relaxation time decreases. On the other side, the 0.7 nm Au₁₃ cluster behaves molecular-like due to the fixed electrons between the Au atoms. Its relaxation is very slow. These results impressively demonstrate Au₅₅ clusters to behave like species just one step before molecular and one step behind metallic state. These correlations can be followed from Fig. 6.

Fig. 6

Relaxation behavior of excited electrons in three differently sized AuNPs. Figure reprinted with kind permission from [23]

Up to here, we have discussed the influence of size, structure, and electronic properties on the relevance concerning bioresponse. However, there is another, very important factor: the protecting ligand shell. As already mentioned above, the use of bare metal clusters and nanoparticles is not possible. Prevention of coalescence affords a “skin” of appropriate molecules on the particles’ surface. The chemical nature of protecting molecules determines not only the solubility in different solvents, but it is decisively responsible for the initial contact between the particle and, in our case, a biosystem. For instance, Au₅₅(PPh₃)₁₂Cl₆ is only soluble in polar organic solvents and can therefore not be used for experiments in living systems. Therefore, a derivative with hydrophilic phosphine ligands is used, namely PPh₂C₆H₄SO₃Na (monosulfonated triphenylphosphine, TPPMS) [24]. The “skin” determines what first happens. This fact can be used for surface recognition or specific targeting [25]. The specific nature of the metal particle becomes only effective if all or at least part of the ligand molecules are removed. Phosphines, coordinated to Au atoms, indeed fulfill this condition as has been shown by ³¹P NMR studies [26] and as can be followed from the different results with respect to bioresponse, as will be demonstrated later. The cell toxicity of Au₅₅(PPh₂C₆H₄SO₃Na)₁₂Cl₆, discussed in more detail in Sect. 3, can clearly be traced back to the Au nucleus itself, since the phosphine ligands have been found to be nontoxic. If the easily removable phosphines are substituted by stronger coordinating thiols, the toxicity of the Au₅₅ clusters is suppressed, since contacts between the clusters and the corresponding cell components happen only via the nontoxic end-groups of the thiols (e.g., glutathione, GSH) [22]. Living cells contain numerous different potential “ligands” that can substitute part or all of the original phosphines and then are directly in contact with the active Au₅₅ cluster core. Unfortunately, these important relations between ligand-protected metal nanoparticles and any other system under investigation are often not sufficiently considered in the literature [27–29].

3 In Vitro Applications: Interaction with Proteins and Cells

AuNPs in general have attracted considerable interest for applications in vitro and in vivo. Since for most applications the particles’ size is in the range of 10–20 nm and stabilizing ligands are thiols, they are typically considered nontoxic, and together with the relatively low amounts, which are applied for different purposes either in vitro or in vivo, it seems to be unlikely that AuNPs cause environmental pollution or health risk due to unintended exposure. Nevertheless, ultrasmall AuNPs can interact with cells and can induce cell death via different pathways, which is related to the size but also to the surface functionalization, as described above. The next two sections will shed light on the state of knowledge about the properties and applications of different classes of nontoxic AuNPs in vitro by means of selected examples. In these examples the AuNPs serve as a scaffold for the binding of (bio)molecules in in vitro applications so that the properties are determined by the binding molecules and by the particle itself. However, the nanoscale features of the AuNPs, e.g., self-fluorescence or SPR, are utilized as probes to study the interaction with biomolecules. These two chapters will be followed by a chapter on in vitro studies on cytotoxic AuNPs, in particular of the 1.4 nm AuNPs, where the gold core itself is biologically active.

3.1 Interaction with Proteins: The Corona Effect

The functionality and bioactivity of AuNPs in biological fluids and in particular in cells are critically dependent on the dispersion state. While in well-dispersed AuNPs the properties of the individual particles are retained, aggregating nanoparticles will progressively approach the properties of micron-sized or even bulk material. For any kind of application, it is vital to control the interaction between the AuNPs and the biomolecules in their respective environment, which can cause aggregation and severe change in the surface properties of the nano-objects and, hence, affect their (hydrodynamic) particle size, surface charge, uptake mechanism, intracellular trajectory, and even the toxicity profile.

The most prominent protein which has been studied in its interaction with NPs of different sizes and composition is serum albumin, which is the most abundant globulin in serum. Due to its negative gross charge, it electrostatically binds to positively charged AuNPs, e.g., particles carrying a terminal amine group, and can overcompensate the particle charge turning it from positive to negative. This binding is associated with an increase of the hydrodynamic radius and enhances reabsorption of the AuNPs in tubules via the neonatal Fc receptor–albumin binding.

The nanoparticle protein-binding kinetics has been analyzed by means of several methods, which probe either fluorescence quenching or fluorescence enhancement of the AuNPs [30, 31]. It was shown that the fluorescence from tryptophan, tyrosine, and phenylalanine residues in proteins was quenched upon binding to the AuNP, which indicates a conformation change of the protein. This allowed analysis of the binding kinetics and the analysis of size effects. It turned out that ultrasmall AuNPs exhibit different binding affinities compared to larger ones. For sub-2 nm AuNPs even the strong self-fluorescence emission at 684 nm could be utilized to analyze the interaction with human serum albumin (HSA) protein, apo-transferrin, lysozyme, and apolipoprotein E4. A signal enhancement was observed upon protein adsorption with increasing concentration of the protein (cf. Fig. 7). Since these four proteins do not exhibit any fluorescence in this spectral range, it is assumed that binding of the proteins slows down the fluorescence decay rate and by this enhances the fluorescence signal intensity of the AuNPs by reducing the polarity of the local environment.

Fig. 7

Left: Photoluminescence spectra of HSA (curve A) and sub-2 nm AuNPs in the presence of different concentrations of HSA (curves B–G), taken with excitation at 550 nm. Right: Fluorescence intensity of these AuNPs, plotted as a function of the HSA concentration in the solution. The gray line fits to the data points using the adapted Hill equation. Figure adapted from [30]

This is one example how the formation of a “protein corona” surrounding an AuNP can be analyzed, and it might be assumed that this process will occur either on the surface of the protecting ligand shell of the particles or directly on the particle surface, when weak-binding ligands, such as TPPMS, are replaced by stronger binding or multivalently coordinating molecules. Nevertheless, the mechanistic details of this interplay between AuNPs and proteins are still rather unexplored. A very recent review by Treuel and Nienhaus gives a comprehensive overview about the state of knowledge and the challenges in this particular field of nanoparticle research [32].

3.2 Biofunctionalized Particles: Examples for In Vitro Applications

A very robust form of functionalized AuNPs for application in vivo has been introduced by Mirkin and his group [33–48]. A very recent research perspective gives an excellent overview about the features and perspectives arising from this technology and shall briefly be summarized [49]. They introduced and intensively studied the assembly induced shift of the surface plasmon resonance (SPR) for the detection and sensing of nucleic acids. In this process AuNPs are functionalized with oligonucleotides via terminal thiol groups. These AuNPs, also described as spherical nucleic acids, possess designed recognition properties encoded by the nucleic acid sequence of the respective oligonucleotide ligands. Based on bases sequence complementarity of the oligonucleotides, either direct particle–particle hybridization or assembly via linker strands becomes possible. Hence, in a solution mixture of two kinds of AuNPs with noncomplementary base sequence, addition of a target DNA that is complementary to both sequences of surface-bound DNA (linker strand) will induce AuNP aggregation due to the binding to both kinds of AuNP via DNA hybridization (Fig. 8).

Fig. 8

Top: Schematic illustration of the aggregation and dispersion of DNA−AuNP conjugates. The corresponding SPR shift is indicated by the color of the AuNPs. The solution of dispersed particles appears red, whereas aggregated particles turn the samples purple. Aggregation can be induced via linker strands (as shown here) or by metal ions or any molecule that the oligonucleotide shell has been programmed to recognize and bind. Bottom: Compared to duplexes of free-strand DNA, which dissociates over a broad temperature range, the melting transitions of oligonucleotide functionalized AuNPs are sharp and occur over a very narrow temperature range due to the cooperative binding of the nucleic acids in the ligand shells. Figure adapted from [49]

Aggregation, i.e., polymeric macroscopic assembly of the AuNPs, can quite easily be followed by means of a SPR red shift, which causes a change of the solution color from red to blue. Since the aggregation results from DNA linkages, these DNA–NP conjugates can be disassembled through dehybridization of the duplexes via heating or by lowering the solution salt concentration. As it is well known from “free” DNA duplexes, melting dehybridization occurs when the temperature is raised above the melting point (T _m), whereas T _m is predictable by base sequence design. In contrast to “free” DNA duplexes, which exhibit melting transitions over a broad temperature range (∼20°C), the melting transition in DNA–NP conjugates occurs over a very narrow temperature range (∼2–8°C) and at a temperature higher than the T _m of the particle-free DNA duplex. This reflects cooperative binding in a highly predictable manner, where a single oligonucleotide base-pair mismatch can influence the melting behavior of the aggregate. It was immediately recognized that these striking features could be utilized for high-selectivity detection platforms, which include the quantitative measurement of enzymes, DNA-binding molecules, or bioactive metal ions [50–53].

Figure 9 illustrates a colorimetric screening of DNA-binding molecules with AuNPs functionalized with oligonucleotides [52]. It was developed to determine the binding affinities between potential DNA-binding molecules and duplex DNA in AuNP–DNA networks. The colorimetric readout does not require additional instrumentation, since it relies of the color change via plasmon coupling upon particle aggregation. When the networks are formed in the presence of duplex DNA-binding molecules, such as DAPI (4′,6-diamidino-2-phenylindole) or EB (ethidium bromide), the resulting duplexes connecting the AuNPs become more stable due to the DNA-binding interactions (p–p stacking, electrostatic, or hydrogen bonding). This increases the melting temperature so that the presence of different DNA-binding molecules can be read out from the temperature-dependent color change.

Fig. 9

Top: Schematic representation of the structural and color change of nanoparticle/DNA-binding molecule assemblies at a specific temperature (T1). Bottom: The color change of the nanoassembly (NP-1 and NP-2, each 1.5 nM) in the absence (CON) and presence of DNA-binding molecule (5 μM)-specific temperatures. Figure reprinted with kind permission from [52]

Based on this fundamental properties of DNA-functionalized AuNPs, Mirkin and others have developed more sophisticated techniques, including chip-based scanometric assays, which have been commercialized partially for highly sensitive molecular diagnostic technologies including clinical disease states, such as Alzheimer’s disease or prostate cancer [49]. Furthermore, the unique set of properties is useful for intracellular applications. This includes the high affinity for complementary DNA and RNA [54], the pronounced stability against nuclease degradation [55], and no observed toxicity [56], which allows application as gene regulation agents [57].

Interestingly, despite their large negative charge due to the polyanionic DNA, the AuNP–DNA were found to easily enter cells. It was demonstrated recently that uptake of these particles in tumor cells (HeLa) is mediated by scavenger receptors. Blocking of these receptors with well-known ligands, polyinosinic acid (PolyI) and Fucoidan, has led to inhibition of particle uptake to a high degree, demonstrating competitive binding of AuNP–DNA and the inhibitors to the receptors. The extent of uptake increases with increasing number of DNA ligands per AuNP and is supported by positively charged scavenger proteins. The presence of serum proteins reduced particle uptake as the adsorption of these proteins on the AuNP–DNA is suspected to compete with receptor binding [58]. This suggests that the interactions between AuNP–DNA and scavenger receptors via the scavenger proteins are a charge-mediated process. These unique features hold great promise for applications in gene regulation therapies as well as in diagnostics, since it was shown that also other inorganic particles, in particular iron oxide particles [59] as well as the so-called “coreless spherical nucleic acids,” i.e., cross-linked nucleic acids, which are oriented in the same way as in the case of AuNP–DNA but do not carry an inorganic core [57], exhibit the same uptake properties.

Recently, DNA-functionalized AuNPs have been used as a platform to study the interaction of even weaker-binding molecules in a similar colorimetric approach. Witten el al. introduced glyconanoparticles with thermally addressable DNA ligands, i.e., they bound glycomodified oligonucleotides to complementary functionalized DNA–AuNP to form glyco-DNA–gold nanoparticles (AuNP–DNA-glyco) [60]. These particles provide a multivalent presentation of DNA-glyco ligands that assemble as a result of the binding of carbohydrate-binding proteins, the so-called lectins, which carry a carbohydrate recognition domain (CRD). These carbohydrate–protein interactions play an important role in processes such as cell–cell or cell–matrix interactions and are exploited for the design of glyconanoparticles, e.g., for vaccine development or in vivo imaging [61]. In the example given, the coupling of AuNP–DNA with N-acetyl-d-glucosamine (GlcNAc)-modified single-stranded DNA (DNA-GlcNAc) to form DNA-GlcNAc-functionalized AuNPs (AuNP–DNA-GlcNAc) is demonstrated. The particles assemble in the presence of the tetrameric lectin GS-II with four CRDs with affinity for GlcNAc [62]. Assemblies built up by the GS-II AuNP–DNA-glyco interaction were shown to be equipped with two reversible binding modes that enable the reversible dissociation by two independent external stimuli: temperature-induced DNA duplex melting and displacement of the DNA-glyco ligands from the CRDs of the lectin by competition with free sugar [60] (Fig. 10).

Fig. 10

Top: Schematic illustration of the reversible GS-II- and GlcNAc-DNA-mediated assembly of AuNP–DNA. Lectin–DNA–AuNP networks are dissociated by external stimuli: temperature-induced DNA melting or competition with free GlcNAc. After competition with free GlcNAc, the particles are separated by centrifugation enabling reassembly by the new addition of GS-II. Note that the scheme is not drawn to scale. ssDNA refers to single-stranded DNA. Bottom: GS-II-mediated assembly of AuNP–DNA-GlcNAc and subsequent competition with GlcNAc or d-lactose. Photographs of the samples: AuNP–DNA-GlcNAc, AuNP–DNA-GlcNAc 2 days after the addition of GS-II, AuNP–DNA-GlcNAc 2 days after addition of GS-II, and subsequent addition of GlcNAc or d-lactose (from left to right). Figure reprinted with kind permission from ref. [60]

3.3 Interaction with Cells

3.3.1 Cyto- and Genotoxicity

Numerous reports about toxicity and nontoxicity of metal nanoparticles can be found in literature [29, 63–80]. Most of the reports are dominated by results concerning AuNPs. Silver, known for its antibacterial function for more than 100 years, also plays an important role. However, gold absolutely dominates the scene. Silver (15–25 nm), gold (5–6 nm), and silver–gold alloy nanoparticles (10–12 nm), protected either by poly(vinylpyrrolidone) (PVP) or tris(3-sulfonatophenyl)phosphine (TPPTS), have recently been tested toward human mesenchymal stem cells (hMSC). As expected, AgNPs showed a significant influence on the cell viability in contrast to the AuNPs with only a small effect. Surprisingly, the alloy particles, though containing silver, had no significant influence on the viability of the cells [81]. Most of the numerous reports on bioresponse of AuNPs are characterized by a disadvantage: the conditions of the various studies all differ so that comparisons are difficult to be drawn. First of all, the experiments differ not only by the particles’ size and shape but, even more important, by their protecting ligand shell. As has already been discussed in Sect. 2, variations of the particles’ skin may change toxicity into nontoxicity or vice versa. Another problem to compare literature data is the use of different cell lines, doses, time of observation, etc. If the metal-specific toxicity is of interest, contact between cell components and metal surface atoms is a condition, but not between ligand molecules and the corresponding biomolecules. Nontoxicity of cystein- and citrate-capped 4 nm AuNPs, glucose-covered 12 nm Au particles, or citrate-, biotin-, and cetylammonium bromide (CTAB)-modified 18 nm particles in human K562 Leukemia cells has been observed [82, 83], whereas toxicity of poly(ethyleneglycol) (PEG)-coated 13 nm AuNPs causes acute inflammation and apoptosis in the liver [27]. A review article with many details describes results of in vitro and in vivo toxicity studies up to 2010 [29].

As already mentioned in Sect. 2, reliable results regarding size dependency can only be gained if identically coated AuNPs are used and which, most importantly, quantitatively or at least partially lose their ligand shell in contact with cell species to allow direct contact between the gold surface and the relevant biomolecules. Phosphines, coordinated to the surface of AuNPs, fulfill this condition.

Au₅₅(PPh₂C₆H₄SO₃Na)₁₂Cl₆ has exclusively been applied to study toxicity in a series of human cancer cell lines. Preceding experiments with this cluster compound and B-DNA showed that phosphine molecules are substituted by DNA sections comprising the major grooves. The reason for this result is to be seen in the size conditions: the diameter of the Au₅₅ nucleus is 1.4 nm and the height of the major grooves is 1.3–1.5 nm. As molecular calculations of that system showed, there are strong chemical interactions between the Au₅₅ core and the DNA, supported by the polydentate character of the major grooves [84]. At least so many ligands are removed from the original cluster that Au₅₅ fits into the groove to allow interactions with relevant DNA components. The coverage of B-DNA with Au₅₅ clusters has already been supposed by a preliminary experiment, namely when B-DNA in aqueous solution was contacted with the coated clusters and investigated by atomic force microscopy (AFM) [85]. Cross sections show the correct difference in height between unloaded and cluster-containing DNA sections. The observed 1.8 nm sections correspond with the sum of the DNA plus about half of the cluster, possibly still having some ligand molecules outside. Figure 11 shows the AFM image and the cross section indicating the difference between DNA with and without Au₅₅ clusters. Figure 12 shows the result of the molecular modeling calculations.

Fig. 11

(a) AFM image of DNA sections partially linked with Au₅₅ clusters. (b) Height profile of unloaded and loaded parts, respectively. Figure reprinted with kind permission from [84]

Fig. 12

Energy-minimized structure of B-DNA with Au₅₅ clusters along the phosphate backbone of the major grooves. Figure reprinted with kind permission from [84]

The Au₅₅ clusters are bound in a stable manner to the major grove of B-DNA as various investigations showed [85].These results clearly indicate that the original phosphine ligands can be removed by biological materials without any problem. An important consequence of this finding is that a DNA, damaged in this manner, can no longer be active in a cell, synonymous with cell death. In other words, 1.4 nm Au₅₅ clusters must finally be toxic.

In a series of experiments the toxicity of Au₅₅ has been investigated using 11 different types of human cancer cells in comparison with cisplatin, a commonly used anticancer drug. Table 1 informs on the IC₅₀ values of cisplatin and Au₅₅, i.e., the amount of material necessary to kill 50% of the cells. In each case, Au₅₅ was more effective, especially considering the difference in time to reach IC₅₀, namely 24 h for Au₅₅ and 72 h for cisplatin.

Table 1 IC₅₀ values of 11 human cancer cell lines treated with cisplatin (72 h) and Au₅₅ (24 h). Table reprinted with kind permission from ref. [85]

The difference between cisplatin and Au₅₅ becomes also visible from Fig. 13a, b. BLM cells were treated with cisplatin and Au₅₅(PPh₂C₆H₄SO₃Na)₁₂Cl₆. Whereas the IC₅₀ value for cisplatin is only reached with 50 μM, theAu₅₅ cluster leads to the viability 0 with 0.4 μM [85].

Fig. 13

Comparison of the viability of BLM cells treating them with cisplatin (a) and Au₅₅(PPh₂C₆H₄SO₃Na)₁₂Cl₆ (b). Figure reprinted with kind permission from [85]

In addition to the interaction with DNA, Au₅₅ causes another reason for a fast cell death: oxidative stress. Necrosis and mitochondrial damage have been observed by incubation of HeLa cells, registered by flow cytometry [22, 86]. It is known that larger AuNPs also cause oxidative stress. However, Au₅₅ is, due to its structure and its special electronic properties, a much better source for oxidative stress. Its catalytic properties, based on the ability to excite O₂, but not to be oxidized itself (see Sect. 2, p. 9), contribute decisively to its toxicity. If the phosphine ligands are substituted by GSH, the toxicity is significantly reduced [22].

The question, if there is a size-dependent cytotoxicity of AuNPs, is still not answered by the experiments discussed above. A series of experiments with smaller and larger AuNPs, all equipped with PPh₂C₆H₄SO₃Na molecules in the coating shell, give a clear answer. The cell lines HeLa, SK-Mel-28, L929, and J774A1 have been incubated with 0.8 (Au₉), 1.2, 1.4 (Au₅₅), 1.8, and 15 nm Au species [86]. Tauredon, a commercially available Au thiomalate complex, has been applied additionally. The result can be seen from Fig. 14.

Fig. 14

(a) IC₅₀ values (logarithmic scale) of four different types of cell lines in dependence of particle size. (b) Results for HeLa cells only in a linear scale (MS = monosulfonated triphenylphosphine, PPh₂C₆H₄SO₃Na). Figures reprinted with kind permission from [86]

As can be followed from Fig. 14a, b, the micromolar amounts for reaching the IC₅₀ values for all four cell lines are a minimum for the 1.4 nm Au₅₅ cluster. There is a lower toxicity for slightly smaller or larger particles, but without any doubt, Au₅₅ is the most toxic one, as can better be seen from the linearly scaled Fig. 14b. The question, if such smaller or larger particles cause less toxicity for not ideally fitting into the major grooves or if reduced oxidative stress is responsible, cannot be answered yet. Some additional experiments concerning reactive oxygen species (ROS) shall be mentioned [22]. Oxidative stress in HeLa cells can be observed by using flow cytometry. However, 100 μM amounts of Au₅₅ clusters cause heavy oxidative stress, increasing with increasing time (12–48 h). On the contrary, even 1,000 μM of 15 nm AuNPs do not cause any oxidative stress. As could be expected, 1.1 nm Au particles, coated with GSH, are also not able to generate oxidative stress, due to the much stronger bound thiol ligands that prevent formation of bare surfaces.

As already discussed, GSH-treated 1.4 nm Au₅₅ clusters restored cell survival in contrast to the original phosphine-coated particles due to the existence of very strong Au–S bonds. Other sulfur-containing molecules such as N-acetylcysteine (NAC) on the cluster surface also reduce cytotoxicity. If the dissociation of TPPMS from the surface is hindered by excessive phosphine, cytotoxicity is also drastically reduced, since the equilibrium of dissociation is shifted. However, GSH, NAC, and phosphines can also be considered as antioxidants. In order to exclude reducing properties of these compounds as reason for the restricted cytotoxicity, ascorbic acid as a strong reducer was tested. As could be expected, the cytotoxicity of Au₅₅(Ph₂PC₆H₄SO₃Na)₁₂Cl₆ was unchanged compared with experiments without ascorbic acid [22].These experiments clearly indicate the repeatedly mentioned condition that bare Au₅₅ surfaces are necessary for the interaction with vital biological targets to observe cytotoxicity.

This finding is further corroborated by studies on the interaction of 1.4 nm Au clusters with ion channels [87]. The human ether-à-go-go-related potassium channel (hERG) is intensely studied and is known to interact with multiple molecular ligands. By patch clamp technique in hERG channel-transfected HEK293 cells, it was shown that 1.4 nm phosphine-stabilized AuNP irreversibly blocked hERG channels, whereas thiol-stabilized AuNP of similar size had no effect.

4 Properties In Vivo

4.1 Biodistribution

The toxicity phenomena discussed before have all been observed in cell cultures, i.e., in vitro. However, from experience we know that toxicity problems in a living being can be characteristically different depending on the kind of administration and distribution in the body. Radiolabeling by ¹⁹⁸Au isotopes allows very precise determination of the distribution in the different organs, though very low concentrations have to be used in order to avoid toxicity effects. In a series of experiments with 1.4, 5, 18, 80, and 200 nm AuNPs, all protected with the proven phosphine ligand Ph₂PC₆H₄SO₃Na, the influence of size on the biodistribution in healthy female Wistar–Kyoto rats in the course of 24 h has been studied [88–90]. γ-Spectroscopy has been applied to determine the gold content in the various organs. The administration was performed by intravenous injection (i.v., tail vein). As can be followed from Fig. 15, the 5, 18, 80, and 200 nm AuNPs are accumulated with 92–97% in the liver, in contrast to the 1.4 nm species which accumulates only with ca. 50% in this organ.

Fig. 15

Accumulation of 1.4, 5, 18, 80, and 200 nm AuNPs, all protected by Ph₂PC₆H₄SO₃Na, in the liver of rats after 24 h (i.v. injection). Figure reprinted with kind permission from [88]

A detailed comparison of the biodistribution of 1.4 and 18 nm AuNPs is given in Fig. 16. A more or less quantitative accumulation of the 18 nm species in the liver can be registered, whereas 50% of the 1.4 nm clusters are distributed over the lung, spleen, kidney, blood, and urine [89, 90].

Fig. 16

Biodistribution of phosphine-coated 18 and 1.4 nm AuNPs after i.v. injection and 24 h retention. Figure adapted from [89]

Considering the findings that traces of the 1.4 nm Au₅₅ cluster could be found in the heart, brain, and uterus (not shown in Fig. 16), it clearly turns out that biodistribution of this species dominates, compared with larger AuNPs. Regarding its cell toxicity, Au₅₅ particles must be considered as very dangerous if present in the blood stream.

There is still another question to be answered: which influence has the surface charge on biodistribution, if at all? Assuming a quantitative dissociation of the Na⁺ ions, Au₅₅(Ph₂PC₆H₁₂SO₃Na)₁₂Cl₆ would be a 12-fold negatively charged particle; however, as we know, in a biological environment, substitution of the phosphines happens quickly. To study the influence of surface charge, strongly coordinated ligands have to be used, best guaranteed by Au–S bonds. Thioglycolic acid (TGA) is an ideal ligand system to investigate the influence of charge, because the carboxylic functions (COO^-) can be reacted with cysteamine (CA) in order to generate positive charges (NH₃ ⁺). AuNPs (2.8 nm) coated with TGA and its positively charged derivative were used to find an answer [88]. As can already be followed from Fig. 15, phosphine-protected 2.8 nm particles accumulate less much in the liver compared with the larger particles by about 20%. The TGA-coated 2.8 nm particles accumulate with 81.6% in the liver, in good agreement with the also negatively charged phosphine-protected species. The positively charged CA-protected 2.8 nm particles assembled only with 72% in the liver with a higher concentration in other organs, for instance 11% in the spleen. As a conclusion it can be stated that biodistribution of positively charged AuNPs is slightly increased compared to negatively charged species.

An investigation, studying the absorption across intestinal barriers by intra-oesophageal instillation of phosphine-stabilized 1.4, 5, 18, 80, and 200 nm AuNPs as well as of TGA- and CA-covered 2.8 nm particles into healthy female rats showed again that biodistribution of the 1.4 nm clusters in secondary organs is maximum [91]. In contrast to the above-discussed situation, biodistribution of negatively charged TGA-coated 2.8 nm species was more expressed than that of the positively charged CA-protected particles of same size. Fig. 17 shows details.

Fig. 17

Biodistribution of negatively and positively charged 2.8 nm AuNPs after intra-oesophageal application. Figure adapted from [91]

4.2 Zebrafish: A New In Vivo Model

Due to the exponential growth of new nanomaterials and to the lack of screening assays and putative standards, there is a shortfall of correlative and predictive models to assess acute and chronic toxicities. This limits rapid preclinical development of new therapeutics and, therefore, it is of critical need to develop in vivo models which ideally maintain the biological representativeness. Small mammalian models are the most common method used to assess possible toxicities and biodistribution of nanomaterials in humans. Alternatively, Danio rerio, commonly known as zebrafish, has attracted much interest because of its unique features, manipulation accessibility, and short reproduction time [92–94]. Many fundamental cellular and molecular pathways involved in the response to toxicants or stress are highly conserved between the zebrafish and mammals [95, 96].

Harper et al. tested AuNPs to determine the effect of core size, surface functionalization and charge on uptake rates, and the biological consequences of exposure [97]. AuNPs with core sizes of 0.8, 1.5, and 15 nm and positive, negative, and neutral surface charges were used (see Fig. 18). Chorions of fish embryos were chemically removed prior to the exposure of nanomaterials since the chorions may act as a physical barrier for AuNPs.

Fig. 18

AuNPs synthesis, structure, and properties. (a) Schematic scheme of the ligand exchange reaction starting from triphenylphosphine-stabilized AuNPs and leading to thiol-stabilized ones. (b) Charge, structure, and name for each ligand tested. (c) Chemical composition of the different AuNPs. Dose response results of embryonic zebrafish screen for tested AuNPs. (d) Percentage of zebrafish exhibiting either mortality or any malformation for each size and ligand tested across a spectrum of concentrations compared to control. (e) Examples of malformations seen upon exposure to 0.8 nm MES–AuNPs scored in the malformation category compared to control embryo. (f) Percentage of zebrafish dying after exposure to the AuNPs. (g) Percentage of zebrafish that survived but showing some malformation or abnormal behavior. Figure reprinted with kind permission from [97]

The 5-day exposure assay resulted in a clear surface charge-dependent toxicity. AuNPs with no charge do not adversely impact biological systems across a broad range of sizes. While positively charged TMAT–AuNPs primarily caused mortality, the negatively charged MES–AuNPs induced malformations (see Fig. 18). The uptake of the AuNPs was further investigated by instrumental neutron activation analysis (INAA), and it was shown that TMAT–AuNPs were not eliminated as rapidly as the other ones [97].

Pan et al. treated zebrafish embryos with TPPMS-capped 1.4 nm AuNPs (Au1.4MS), which were previously analyzed in HeLa cells [98]. In the cell test was shown that Au1.4MS was toxic while smaller and larger TPPMS-stabilized AuNPs were much less toxic [22, 86]. By the addition of GSH to Au1.4MS particles, the toxicity was inhibited due to replacement of the weak-binding TPPMS by the GSH hence resultant in a stronger gold thiolate bond [22]. They showed that the toxicity in zebrafish model like in the cell culture depends on the size and ligand chemistry. At similar size AuNPs carrying ligands with high affinity to the gold atomic core were less toxic than AuNPs with more labile ligands. Along with the cytotoxicity of Au1.4MS-treated HeLa cells, the robust heat shock protein (HSP) stress response was observed. Therefore Pan et al. deemed transgenic heat shock reporter zebrafish ideal to test toxicity, teratogenicity, and upregulation of defense pathways. These transgenic zebrafish had similar responses to Au1.4MS-like wild-type zebrafish in terms of teratogenicity but were 20-fold more sensitive than wild type in reporting hepatotoxicity of Au1.4MS (Fig. 19).

Fig. 19

Rapid induction of green fluorescent protein (GFP) expression in transgenic zebrafish triggered by toxic doses of Au1.4MS. Transgenic zebrafish embryos expressing GFP under the control of the heat shock protein were exposed to (a, e) embryo medium, (b, d) 200 μM Au1.4MS, and (c) 400 μM Au1.4MS. Spotty GFP signal was seen at 48 hpf in the presence of 200 μM Au1.4MS (b). The GFP signal increased with the toxic dose (c) and with time (d). In addition to the spotty GFP signal observed in the head region, strong GFP expression was present along the trunk at 48 hpf and was further enhanced at 72 hpf (d). Figure reprinted with kind permission from [98]

Razansky et al. used PAI to image in vivo the expression of the fluorescence protein mCherry in the head of an adult transgenic zebrafish with a cross-sectional diameter of ~6 mm. The imaging results demonstrate the ability of the process to reveal many morphological features which matches well with the corresponding histology. Moreover, multispectral reconstructions accurately resolved fluorescence protein expression in the hindbrain of an intact living animal in high consistence with the corresponding epifluorescence image of the dissected brain [99].

5 Perspectives and Challenges

5.1 Cell Targeting

The examples given above illustrate some of the versatile applications of AuNPs in biological environments, addressing different fields in therapy and diagnosis and several recent reviews reflect that this is a rapidly evolving interdisciplinary research area [100]. However, any type of in vivo application should ideally follow the concepts of targeted delivery, which combines minimization of systemic exposure and, hence, decreased side effect with enhanced local concentration of the therapeutic or diagnostic agent for maximum efficacy. Cell targeting can be achieved along two conceptually different ways, i.e., passive and active targeting, respectively. While passive targeting utilizes, e.g., the enhanced permeability and retention (EPR) effect [101], by which small nanoparticles can accumulate in tumor tissue due to the disordered vasculature in tumors, active targeting relies on recognition molecules or molecular subunits, which can selectively bind to certain tumor cells and thus enhance the accumulation of the “targeting” compound at tumor sites. Although the passive targeting can lead to high uptake rates into the tumor tissue, the penetration into tumor interstitium is rather limited and varies the size or density of tumor tissue and is thus difficult to control. This may be enhanced by active targeting, as it was introduced for several types of “recognition molecules” bound to the surface of AuNPs, including small peptides [102, 103] and antibodies [104, 105]. One of the first promising examples was given in a study on targeted photothermal ablation of murine melanomas with targeted hollow gold nanospheres (HAuNs) of approx. 40 nm in diameter. It was found that the AuNPs were specifically taken up by melanoma cells associated with enhanced extravasation and dispersion into tumor matrix of targeted HAuNs compared to untargeted ones of the same size [105] (Fig. 20).

Fig. 20

Biodistribution of HAuNs targeted with NDP-MSH-PEG–HAuNs and untargeted, i.e., functionalized with PEG. Data are plotted as % of injected dose per gram of tissue. Mean FSD (n = 5). *, P < 0.01. Figure adapted from [105]

This control over biodistribution and localization of AuNPs gives rise to applications in nanomedicine, where AuNPs are utilized as efficient contrast agents for different imaging techniques, i.e., X-ray computed tomography (CT) and photoacoustic imaging (PAI). Furthermore therapeutic applications can be envisaged in the same manner, if an external stimulus such as light or X-ray irradiation selectively addresses the AuNPs based on their size-dependent physical properties, such as SPR or X-ray radiosensitization. The potential applications that arise from these opportunities will be discussed in the following sections by means of selected examples.

5.2 AuNPs as Contrast Agents in Molecular Imaging and as Therapeutic Agents

Molecular imaging is an emerging field integrating molecular biology, chemistry, and radiology in order to gain understanding about biological processes and to identify diseases based on molecular markers, which appear earlier than the clinical symptoms [106]. The application of AuNPs in this area is increasing rapidly and offers excellent prospects for the development of new strategies for the diagnosis and treatment of cancer [107]. The reasons for this increasing attention to gold nanoconstructs are many: their size-dependent and shape-dependent plasmonic properties enable them to absorb and scatter light in the visible to NIR region, which may render them suitable for image-guided therapy and photothermal therapy (PTT). Significant synthetic advances now allow the design of AuNPs with highly controlled geometry, surface charge, physicochemical properties, and the decoration of their surfaces with polymers and bioactive molecules in order to improve biocompatibility and to achieve active targeting. This is stimulating the development of a diverse range of nanometer-sized objects that can recognize cancer tissue [107]. The use of the unique optical properties of AuNPs presents a new opportunity for noninvasive imaging and therapy for a variety of diseases without exposing the rest of the body. Here, we introduce some imaging applications including X-ray computed tomography (CT) and photoacoustic imaging (PAI) and the opportunities of PTT as well as radiotherapy based on AuNP systems.

5.2.1 AuNPs for CT Imaging

X-ray computed tomography (CT) is clinically important because of its affordable price, high spatial resolution, unrestricted depth, and accurate anatomical information with reconstructed three-dimensional imaging [108]. The limitations of commercial iodine-based CT contrast agents such as rapid renal clearance, poor sensitivity, and toxicity have to be overcome, and AuNPs are interesting tools for that [109]. Their ability to absorb large amounts of X-ray radiation can be used to increase imaging contrast in diagnostic CT scans at lower radiation doses, which is based on the atomic weight of Au relative to atoms present in biological tissue. AuNPs have shown in vivo functionality as CT contrast agents for blood pool [110–116] and cancer imaging [117–124].

For passive targeting AuNPs are modified with PEG and its derivatives to enhance the circulation period in the blood. Kim et al. reinforced that by showing a 5.7 times higher attenuation of PEGylated AuNPs (~30 nm) than that of the current iodine-based CT contrast agent Ultravist [124]. Wang and his coworkers found that dendrimer-entrapped AuNPs can be detected through the attenuation of X-rays [117]. After either an intratumoral or an intraperitoneal administration of these particles, a xenograft tumor model could be imaged via CT.

The conjugation of antibodies or peptides onto the NP surface allows for active targeting. AuNPs with active tumor-targeting anti-HER2 antibodies enhance the visibility of millimeter-sized human breast tumors in mice in a 1.6-fold more efficient way than passive targeting AuNPs [123]. Chanda et al. reported enhanced CT attenuation of bombesin-functionalized AuNPs that selectively targeted cancer receptor sites that are overexpressed in the prostate, breast, and small-cell lung carcinoma [119]. Reuveni et al. demonstrated the in vivo feasibility of cancer diagnosis based on AuNPs as molecular markers using CT [118]. AuNPs with a diameter of 30 nm were functionalized with anti-epidermal growth factor receptor molecules and intravenously injected into nude mice implanted with human squamous cell carcinoma head and neck cancer. It is demonstrated that a small tumor, which is currently undetectable through anatomical CT, is enhanced and becomes clearly visible by the molecularly targeted AuNPs. As can be seen in Fig. 21, it is further shown that active tumor targeting is more efficient and specific than passive targeting with anti-rabbit IgG antibody-coated AuNPs with the same size.

Fig. 21

In vivo X-ray CT volume-rendered images of mouse (a) before injection of AuNPs, (b) 6 h postinjection of nonspecific anti-rabbit IgG-coated AuNPs as a passive targeting experiment, and (c) 6 h postinjection of anti-epidermal growth factor receptor (EGFR)-coated AuNPs that specifically targeted the squamous cell carcinoma head and neck tumor. The anti-EGFR-targeted AuNPs show a clear contrast enhancement of the tumor (c, yellow arrow), which was undetectable without the AuNPs as contrast agents (a, yellow arrow). Figure adapted from [118]

Besides the imaging of cancer, Ghann et al. described a preparation of AuNPs with a core size of 10.7 nm that are stabilized with lisinopril via amine, disulfide, and thiol attachments to the AuNP surface to image the blood pool [115]. A stability study showed that the thiol lisinopril-coated AuNPs were the most stable ones and therefore chosen for further studies to assess the targeting of angiotensin-converting enzyme (ACE) via X-ray CT. The resulting images displayed a high contrast in the region of the lungs and heart which indicates the targeting of ACE (see Fig. 22). The overexpression of ACE is associated with the development of cardiac and pulmonary fibrosis, and because of this the AuNPs could be a useful tool for the visualizing of cardiovascular pathophysiologies using CT imaging.

Fig. 22

3D subtraction CT image (measured at 80 and 140 kVp) of mice treated with thiol lisinopril-coated AuNPs resulting in a specific targeting of the lungs and heart (enlargement of the lungs and heart area on the right side). Figure reprinted with kind permission from [115]

Cai and his coworkers injected PEG–GNPs with a size of 38 nm into adult mice via subcutaneous application [113]. Compared to the CT image contrast before PEG–GNP injection, the CT can produce clear images of the vascularity even 24 h after injection, indicating that the PEG–AuNPs have great potential as a blood-pool agent for CT imaging. In another study dendrimer-entrapped AuNPs with sizes in a range of 2–4 nm were injected subcutaneously into mice [110]. The vascular system could be imaged 5 and 20 min after injection and the urinary system could be imaged after 60 min. The feasibility for blood-pool CT imaging of PEGylated dendrimer-entrapped AuNPs after intravenous injection in rats and mice was explored as well resulting in an efficient blood-pool CT imaging of both rats and mice [116].

5.2.2 AuNPs for PAI Imaging

Photoacoustic imaging (PAI) is a technique which combines the advantages of optical and ultrasound imaging methods to visualize cancer tumor spreading and growth in vivo [125, 126] resulting in the ability to detect metal NPs with nonionizing radiation relative deep in the tissue without the loss of high spatial resolution [126–128]. A temperature increase is induced by short pulses of electromagnetic irradiation in the range of strong absorbance. This causes a local increase in pressure via thermoelastic expansion leading to a generation of acoustic waves which can be detected with an ultrasound transducer [125–129].

AuNPs have gained immense interest for PAI since they combine size-dependent optical properties with a versatile surface chemistry [130–132]. As compared to the established group of organic dyes with its standard PAI contrast agent methylene blue (MB), recent research has demonstrated the superiority of AuNPs as they show high absorption cross sections combined with a resistance toward photobleaching. By controlling the particle size and geometry, the extinction characteristics of AuNPs can be tuned. Numerous groups have found that nanospheres [133–135], nanorods [136–143], nanoshells [132], hollow nanospheres [127], and nanocages [144–146] are eminently suitable.

The first studies with gold nanomaterials were done by Sokolov et al. who showed that 12 nm AuNPs conjugated with anti-epidermal growth factor receptor (anti-EGFR) antibodies specifically bound to EGFR proteins that are overexpressed on the surfaces of cervical cancer cells [147]. The receptor-mediated aggregation of AuNPs causes plasmon coupling of the clustered NPs, leading to an optical redshift of the plasmon resonances. Based on these results, the group of Emelianov demonstrated in ex vivo mouse tissue that multiwavelength photoacoustic imaging can detect cancer with high selectivity and sensitivity based on the plasmon resonance coupling effect of the EGFR-targeted AuNPs [148, 149]. In addition to that plasmon resonance coupling AuNPs have been suggested for imaging of macrophages in atherosclerotic plaques for an early detection of cancer [133].

In a recent study, Gutrath et al. showed that the molar extinction of spherical AuNPs with core diameters of 1.0, 1.4, and 11.2 nm, gold nanorods (AuNRs) with longitudinal/transversal elongation of 38/9 and 44/12 nm, and HAuNs’ with outer/inner diameters of 33/19, 57/30, 68/45, and 85/56 nm can be correlated with the molar PA signal intensity of the respective AuNPs [150]. For increasing lateral particle size, increasing molar photoacoustic signal intensity is observed, which is shown in Fig. 23. The studied AuNPs exhibit significantly larger molar PA intensities as compared to the commonly used dye MB, except for spherical Au1.0NPs, which exhibited results comparable to the values of MB. The particle-size-dependent evaluation of the PA signal intensities is of great interest because larger particles are useful to obtain molecular imaging of post-angiogenesis processes, whereas only small NPs are able to infiltrate the nonvasculated parts of tumors for efficient targeting. These results were confirmed by Popović et al. who observed that particles within the 10–150 nm-size range show a particle-size-dependent uptake mechanism after injection into a mouse bearing a Mu89 human melanoma xenograft [151]. When a mixture of 12, 60, and 125 nm particles is injected in vivo the 12 nm particles extravasated easily and diffused from the vessels with minimal hindrance, the 60 nm particles extravasated but do not leave the immediate perivascular space, while the 125 nm particles did not appreciably extravasate.

Fig. 23

Molar PA signal intensities as a function of the lateral particle size for the four laser excitation wavelengths. Error bars indicate the standard deviations evaluated with error propagation calculations resulting from statistical analysis of STEM micrographs. Lines are drawn to guide the eyes. Figure adapted from [150]

By using anti-HER2-conjugated AuNRs, Agarwal et al. detected prostate cancer with PAI [139]. This method can also be used to detect and localize AuNRs with an approximate size of 50 × 15 nm at a very low concentration deep within tissue [152]. AuNRs of varying aspect ratios (3.7 and 5.9) and functionalized with different molecules to target MBT2 and HepG2 cells were used by Li et al. [153]. Jokerst et al. varied the aspect ratios of AuNRs in the range of 2.4–3.5, and those with an aspect ratio of 3.5 showed the highest ex vivo and in vivo PA signal [154]. Therefore these particles were used to image subcutaneous xenografts of the 2008, HEY and SKOV3 ovarian cancer cell lines in living mice. A linear relationship between the PA signal and the concentration of the injected AuNRs was observed. Another study demonstrated multiple selective targeting on oral cancer cells with HER2-antibody- (aspect ratio of 3.7) and EGFR-antibody-conjugated AuNRs (aspect ratio of 5.9) [155]. Huang et al. demonstrated a combined cancer imaging method using X-ray and PAI with PEG–AuNRs as contrast agents [156]. Figure 24 shows that the combination of X-ray and PAI can provide more comprehensive details of the tumor such as position, size, and vascular network. This is of great interest because vasculature regulates the metabolic and hemodynamic states of biological tissues wherefore the visualizing of blood vessels enables the tracking of cancer metastasis and monitoring of tumor angiogenesis.

Fig. 24

Combined X-ray and PA images of a mouse tumor in vivo, which clearly show the location and the blood vessels of the tumor. (a) Overlay of X-ray image and PA image. (b) Enlargement of the corresponding image area highlighted by an inset in panel a. Figure reprinted with kind permission from [156]

The application of PEGylated hollow gold nanospheres (HAuNs) with a diameter of ~40 nm in nude mice showed the brain vasculature with great clarity and more detail structures. The images shown in Fig. 25 depicted blood vessels as small as ~100 mm in diameter which could be clearly seen with PEG–HAuNs as PAI system [127].

Fig. 25

Noninvasive PA images of a mouse brain in vivo employing PEG–HAuNs and NIR light at a wavelength of 800 nm. PA images received (a) before (the arrow pointed on the middle cerebral artery), (b) 5 min after, and (c) 2 h after the intravenous injection of PEG–HAuNs. (d, e) Differential images obtained by subtracting the preinjection image from the postinjection images (Image d = Image b − Image a; Image e = Image c − Image a). (f) Open-skull photograph of the mouse brain cortex obtained after the data acquisition for PAT. Scale bar = 2 mm. Figure reprinted with kind permission from [127]

Beneath AuNPs, AuNRs, and HAuNs, also gold nanocages (AuNCs) have shown great promise in PAI. In one report, 50 nm AuNCs were intradermally injected on the left forepaw pads of Sprague–Dawley rats to image sentinel lymph nodes (SLN) noninvasively [157]. With the passing of time, the contrast at the SLN gradually and the PA signal within the SLN were increased. It was also demonstrated that the total image depth below the skin surface of 33 mm could be reached with good contrast which is substantial because this depth is greater than the mean depth of SLNs in human beings [157]. PEGylated AuNCs were used by Yang et al. to confirm the contrast enhancement for PA imaging in vivo [146]. The particles were injected into Sprague–Dawley rats via the tail vein and better resolved images of the cerebral cortex after administration were recorded. For the visualizing of this layer in the brain of Sprague–Dawley rats, also PEGylated gold nanoshells were used by injection via the tail vein as well [132]. The nanoshells were administrated three times successively, and after the final injection the rat brain was imaged sequentially 10 times for more than about 6 h. The PA signal enhancement after the injection as a function of time was quantified with a maximal signal increase of about 63%.

In a recent report Bao et al. used PEGylated gold nanoprisms (AuNRs) with a uniform thickness of ca. 10 nm, with three congruent edge lengths of ca. 120 nm in in vivo settings [158]. The mice were treated with this AuNRs in order to visualize tumor angiogenesis in gastrointestinal cancer cells. It is shown that the AuNRs have the capacity to penetrate tumors and provide a high-resolution signal amplifier for PAI.

5.2.3 AuNPs as Agents for Photothermal Therapy

As compared to the diagnostic approach followed by PAI during photothermal experiments, a therapy approach is proposed by photothermia. Due to their SPR properties AuNPs are able to absorb light from incident radiation with high efficiency (extinction coefficient ~10⁹ M⁻¹ cm⁻¹) [159] in the near-infrared (NIR) region of the electromagnetic spectrum and convert it into heat [83, 160, 161]. The generated heat is subsequently delivered to the immediate surroundings of the AuNPs and allows a highly specific thermal ablation of diseased or infected tissue [162–165]. The advantage of the use of light in the NIR region is that the tissue damage and attenuation by biological fluids and tissues are minimal.

In 2003 Pitsillides et al. were the first who demonstrated that AuNPs can be used for PTT [166]. Since that time, several other studies have examined the use of AuNPs in PTT. Nam and coworkers designed AuNPs with a size of 10 nm with a pH-dependent agglomeration behavior [167]. The incubation of HeLa and B16 F10 mouse melanoma cells with these AuNPs shows an aggregation of the AuNPs in a typical tumor intercellular pH, leading to a shift of their absorption to the far- and near-IR spectral regions and thus to a utilization in PTT.

Nevertheless, AuNRs [168], HAuNs [105], AuNCs [104, 144, 169–171], and gold/silica nanoshells [172] with SPR frequencies in the NIR range are more promising agents for PTT because the absorption of the NPs can be tuned by synthetically varying the aspect ratio and shell thickness : core radius. The latter were some of the first applied in PTT by Halas and West [172–175]. Based on efficiently destroying breast carcinoma cells with PEGylated silica/gold nanoshells, the particles were injected into the tumor interstitium of SCID mice bearing sarcoma xenografts. A subsequent NIR light exposure demonstrated a 4–6 mm depth of thermal damage [172]. Later on NIR PTT was demonstrated using systemically administered PEGylated nanoshells in a colon cancer mouse model. All tumors in nanoshell-treated mice were completely ablated after a single PTT treatment and the animals appeared healthy and tumor free for more than 90 days posttreatment. In contrast the tumors of the control animals and the additional sham-treated animals (laser treatment without nanoshell injection) continued to grow, with nearly 50% mortality at day 10 [174].

For vascular-targeted PTT of glioma nanoshells were conjugated to VEGF and/or PEG to thermally ablate VEGF receptor-2-positive endothelial cells upon NIR laser irradiation [176]. It was observed that VEGF-coated but not PEG-coated nanoshells bound to VEGF receptor-2-positive cells in vitro to enable targeted photothermal ablation. Subsequent in vivo studies in mice-bearing intracerebralglioma tumors showed that VEGF targeting could double the proportion of nanoshells bound to tumor vessels and vasculature was disrupted following laser exposure.

AuNPs with an interior composition of gold sulfide, or gold/gold sulfide composite structures, were first produced by self-assembly by Zhou et al. and were shown to have strong NIR-absorbing properties [177]. Gobin et al. compared gold/gold sulfide AuNPs with an overall diameter of 35–55 nm that provides higher absorption as well as potentially better tumor penetration, with gold/silica nanoshells with an overall diameter of 120–160 nm [178]. At relatively low concentrations and laser powers, the heating profiles of gold/gold sulfide AuNPs showed temperatures high enough to effect tumor ablation by hyperthermia. The smaller size of the gold/gold sulfide NPs yields a particle with higher absorbing cross-sectional area ratio than the gold/silica nanoshell.

Hu et al. noted that AuNRs and AuNCs have much larger absorption and scattering cross sections than nanoshells wherefore these are better candidates for PTT [179]. The use of AuNRs, which can more efficiently absorb photons to generate heat, as PTT agent is based on the pioneering work of Catherine Murphy’s group [180]. These particles have been extensively studied for PTT applications in vitro [168, 181, 182] and in vivo [183, 184]. Fundamental work by El-Sayed and coworkers has provided insights into how to optimize metallic NP-based PTT, including how shape influences heat generation efficiency [185] and the critical temperatures associated with therapy (typically 70–80°C) [186]. They also found that AuNRs labeled with an EGFR-targeting ligand were effective for PTT under NIR illumination in vitro [168]. The feasibility of in vivo PTT treatment of squamous cell carcinoma xenografts with PEGylated AuNRs was demonstrated by the inhibition of average tumor growth for direct and intravenous delivery methods [183].

Von Maltzahn et al. described an approach to improve plasmonic therapy by using computational simulations to understand the effect of NP concentration on heating in vivo. Therefore an X-ray CT scan was utilized to characterize the distribution of intratumorally and intravenously administered PEGylated AuNRs in human tumor xenograft bearing mice. The high absorption efficiency of the AuNRs in both the X-ray and NIR regions enabled real-time visualization and therapy [187]. To assess the utility of the high X-ray absorption of PEG–NRs for detection of in vivo, PEG–AuNRs were directly injected into the tumors of mice bearing bilateral human MDA-MB-435 tumors, implanted either in the mammary fat pad or the rear flank. It was found that X-ray CT rapidly detailed the three-dimensional distribution of PEG–NRs in tumors, showing clear distinction between AuNRs and soft tissues, as seen in Fig. 26.

Fig. 26

Photothermal heating of passively targeted AuNRs in tumors. PEG–AuNRs or saline was given i.v. to mice bearing MDA-MB-435 tumors on opposing flanks. After AuNRs had cleared from circulation (72 h after injection), the right flank was irradiated (beam size indicated by dotted circle). Figure reprinted with kind permission from [187]

To prolong the circulation time, optimize the tumor targeting, and decrease the liver uptake, Choi et al. prepared AuNR-loaded, chitosan-conjugated, pluronic-based nanocarriers which could serve as imaging agents for cancer cells and as a very effective hyperthermia agent for PTT [184]. By delivering the nanocarriers via an intravenous injection followed by NIR laser irradiation to the tumor site resulted in a very efficient thermolysis in vivo, achieving a complete tumor resorption without damage to the surrounding tissue.

Xia and coworkers studied AuNCs as photothermal transducers for therapeutic applications [104, 144, 169–171]. They utilized PEGylated AuNCs with an edge length of 45 nm for selective destruction of neoplastic tissue using a bilateral tumor model. The particles accumulated in tumors with a relatively high efficiency leading to a slightly higher amount of AuNCs in the tumor periphery than in the inner core. With an infrared camera the temperature increase during photothermal treatment was monitored which is shown in Fig. 27.

Fig. 27

(a) Photograph of a tumor-bearing mouse under photothermal treatment. PEGylated nanocages or saline was administrated intravenously through the tail vein as indicated by an arrow. After the NCs had been cleared from the circulation (72 h after injection), the tumor on the right flank was irradiated with a beam size indicated by the dashed circle. (b–g) Thermographic images of nanocage-injected (b–e) and saline-injected (f–i) tumor-bearing mice at different time points: (b, e) 1 min, (c, f) 3 min, (d, g) 5 min, and (e, i) 10 min. (j) Plots of average temperature within the tumors (dashed circle) as a function of irradiation time. All scale bars are 1 cm. Figure reprinted with kind permission from [170]

The effect of PTT on cells by larger, more rapid temperature increases, often referred to as “ablative” treatments, is considered to be necrosis with the corresponding melting of cell membranes and organelles [188]. In contrast to that mild temperature increases (known as hyperthermia), mostly proceed via apoptotic pathways, and are known to perturb a variety of normal cellular functions [183]. Huang et al. designed AuNR elastin-like polypeptide matrices loaded with the heat shock protein (HSP) inhibitor 17-(allylamino)-17-demethoxygeldanamycin (17-AAG) to give insights into the release of heat shock proteins (HSPs), which help to restore normal processes to the cell, and their relation to PTT [181]. They demonstrated that AuNRs in combination with 17-AAG improve significantly (>90%) death of cancer cells, while “single treatments” (i.e., hyperthermia alone and 17-AAG alone) demonstrated minimal loss of cancer cell viability (<10%). This result seems to imply that in addition to the local thermal effects, significant improvements in NP-mediated PTT can also take advantage of limiting cellular responses.

5.2.4 AuNPs in Radiotherapy

Another method which is widely used to help reduce tumor size is through radiofrequency. Since the early 1990s, it has been utilized as a treatment for destroying liver tumors [189]. However, this method does have its drawbacks, such as accuracy, invasive needle placement, and toxic effects, on the tissue surrounding the tumor and also on internal organs [190, 191]. These side effects have a drastic limitation on the amount of radiation used during treatment.

Several studies have focused on the use of AuNPs as novel high-Z radiosensitizing agent to increase the sensitivity of tumors irradiated with clinical X-ray beams [192, 193]. Numerous experimental [194–196] and theoretical studies [197–199] have shown that clinically significant enhancements are achievable with AuNPs-treated tumors in combination with X-ray therapy.

In vitro studies showed AuNP radiosensitization by irradiation of cells and plasmid DNA [200–204]. Irradiating plasmid DNA in the presence of 5 nm AuNPs leads to single-strand and double-strand break enhancement [202]. Rahman et al. observed a biological dose enhancement factor of up to 24.6 when irradiating bovine aortic endothelial cells in the presence of 1.9 nm AuNPs [203]. Roa et al. were the first who reported that 10.8 nm glucose-capped AuNPs trigger the activation of CDK kinases leading to cell cycle accumulation in the G2/M phase and acceleration in the G0/G1 phase [205]. This leads to the suggestion that p53 and the CDK kinases are targets for AuNPs. A striking sensitization to ionizing radiation is achieved as well. In another study AuNPs with an approximate size of 45 nm showed an increase of the effectiveness of proton radiotherapy for the killing of prostate tumor cells of about 15–20% [206].

Despite the rapid increase of in vivo studies investigating the uptake and distribution of AuNPs, there remains a lack of studies of in vivo radiosensitization with these particles. In 2004, the suitability of AuNPs for radiotherapy was examined by Hainfeld et al. in vivo [192]. Nontargeted 1.9 nm AuNPs (AuroVist) in combination with 250 kV radiation were shown to prolong survival in Balb/C mice bearing EMT-6 murine breast cancer tumors. One month after the treatment, a dramatic reduction in tumor growth was observed. Based on these results, a second experiment with a longer follow-up study in mice which received a slightly lower radiation dose alone or with AuNPs was pursued. A remarkable tumor regression in mammary tumors and long-term survival without any significant toxicity were demonstrated thus indicating the utility of these particles for radiotherapy. In a recent study by Hainfeld et al., a highly radioresistant murine squamous cell carcinoma was used in mice [207]. A significant tumor growth delay and a long-term tumor control were observed by combining AuNPs with irradiation. Chang et al. used 13 nm citrate-stabilized AuNPs in a mouse model with B16F10 murine melanoma cells. A significant in vivo tumor growth delay and increased survival were noted when AuNPs were injected 24 h before irradiation [195]. The median survival of the mice with 65 days was shorter than in the Hainfeld study but the AuNP concentration used is much lower.

While demonstrating the potential efficacy of AuNP radiosensitization, the large variations in these studies revealed AuNP radiosensitization to be highly sensitive to a number of physics and pharmacological parameters such as irradiation energy, AuNP size and concentration, and intracellular localization [192, 202, 203]. Toward understanding and predicting the effects of these parameters, there have been a number of Monte Carlo simulation studies exploring AuNP dose enhancement [197, 198, 208–213].

5.3 Drug Delivery

If AuNPs shall be used as therapeutic agents based on their cytotoxicity in the future, a fundamental challenge has to be faced. One of the most common drawbacks of the present treatment of tumors is the small specificity of the respective drugs that means that the number of damaged tumor cells compared with the number of damaged healthy cells is far too small. Therefore, worldwide efforts are being made to improve the situation, among others by using the so-called drug delivery (drug release) systems. The goal is to transport a drug specifically into a tumor in order to avoid damage of healthy cells as far as possible. One promising attempt is the use of micro- or nano-sized hollow containers which are filled with chemotherapeutics or cytotoxins. The advantage of this technique is the possibility to modify the surface of the capsules by tumor-specific antibodies without the need to modify the drug itself. Accumulated in the tumor, the containers open themselves or by stimuli from outside, for instance, by photoinduced heating of AuNPs, also present in the containers [163, 166, 214–216]. The capsules usually consist of polymers which can easily be modified by tumor-specific antibodies. Poly(alkylcyanoacrylate) even degrades rapidly in vivo without any photoinduction of AuNPs [217]. Figure 28 elucidates the artificial opening process using AuNPs.

Fig. 28

Sketch of the opening process of a polymer container filled with the drug and additionally containing AuNPs for photoinduced heating process. Figure adapted from [90]

The use of AuNPs for photoinduced heating is only of importance for the opening of the polymer containers. Another chance is the direct use of AuNPs as a drug. The cell toxicity of Au₅₅ with its ultrasmall 1.4 nm diameter has already been discussed in Sect. 3. Experiments to transport 1.4 nm Au particles in polymer spheres are presently under investigation (Simon U, Mayer C, private communication).

Though not yet tested for the use of AuNPs, mesoporous silica nanoparticles are a promising alternative to hollow nanospheres. They are intensively under investigation, especially to study single particle tracking to living cells by fluorescence microscopy allowing the discrimination between active transport and random motion [218–220].

Also still under investigation for use as drug delivery systems are the so-called polymersomes. Polymersomes consist of a double layer of polymers with a hydrophilic and a hydrophobic function, cross-linked by various polymers. The combinations form a hollow sphere, comparable with the structure of liposomes from which the name polymersome derives. Their size can be varied between 100 nm and 100 μm [221]. They are able to transport bioactive molecules to the relevant cells and to release them there [222].

6 Summary

This review has summarized relevant aspects on the properties and the bioresponse of AuNPs in the context of a rapidly growing field of applications, i.e., the application in vivo and in vitro for medical purposes. It has been shown by a number of selected examples that most of the AuNPs synthesized so far are considered nontoxic, making them suited for in vivo applications, e.g., for imaging and therapy. Special focus is laid on AuNPs in the sub-2 nm range, which under certain conditions can become toxic. This sheds light on how adverse effects of AuNPs in biological systems may occur and, at the same time, suggests how such toxic effect may be used in therapy as well. The present state of knowledge is summarized on the mechanism how such small AuNPs may interact with biomolecules and cells, addressing the effect of shape complementarity in the sense that AuNPs approach the size of biological molecules and subcellular structures and thus may even behave as molecular ligands.

It becomes evident that the high degree of control over particle size and ligand chemistry opens up an avenue of novel applications in a biological environment.