Fluorescence imaging of ENM biodistribution

The ultimate prerequisite for a harmful effect from any toxicant is exposure. Therefore, tools allowing the visualization of the nanoparticles (NPs) in vivo are crucial for risk evaluation. On the other hand, the growing utilization of ENMs in potential clinical applications ranging from diagnosis to therapeutic treatment of diseases necessitates a need for understanding their in vivo transport, biodistribution, and long-term fate in the organism.^{Reference Liu, Tabakman, Welsher and Dai12} ENMs are known to exhibit a higher uptake by monocytes and macrophages of the mononuclear phagocyte system (a part of the immune system containing phagocytic cells) upon injection into animals, which prevents their rapid clearance through excretion and impairs therapeutic or diagnostic targeting.^{Reference Podila and Brown2,Reference Liu, Tabakman, Welsher and Dai12}

Indeed, the clearance mechanisms for most ENMs remain unresolved due to the existing difficulties in in vivo tracking and monitoring of the ENMs. Fluorescence spectroscopy is one of the important tools in assessing the biological interaction of ENMs. Oxide NPs such as SiO₂, TiO₂, CeO₂, ZnO, and graphene oxide (GO) are often conjugated with visible or near-infrared (NIR) fluorophores to help track their transport in vivo. ^{Reference Liu, Tabakman, Welsher and Dai12–Reference Moore, Podila, Alexis and Rao14} In this regard, fluorescence imaging studies on carbon nanotubes (CNTs) and GO in mice with a U-138 glioblastoma (brain tumor) intracranial xenograft (a surgical graft from a foreign tissue) using Alexa Fluor 647 fluorescent dye are noteworthy since they revealed that the hydroxyl and quinone groups on GO result in a decreased agglomeration relative to CNTs in the tumor due to the polar nature of the hydroxyl/quinone functional groups.^{Reference Moore, Podila, Alexis and Rao14} While fluorescence imaging of ENMs is useful for in vivo tracking over short periods of time (a few hours to a few days), many challenges such as inherent modification of the ENM physicochemical properties following the addition of fluorophore, photobleaching, fluorescent quenching, and fluorophore dissociation from ENMs continue to plague this method. Nonetheless, spectroscopic tools that identify ENMs directly through their intrinsic properties are emerging and could provide an accurate assessment of ENM biodistribution at short (few hours) and long-time scales (many months).^{Reference Liu, Tabakman, Welsher and Dai12}

Raman spectroscopic investigation of the ENM physiological response

Raman spectroscopy provides a spectral fingerprint of ENMs and is based on the principle of inelastic scattering of light by ENMs. When a monochromatic beam of photons (e.g., a laser beam) interacts with atomic vibrations (phonons) in an ENM, it experiences a distinctive shift in its energy known as the Raman shift. The energy and intensity of Raman shifts are uniquely determined by the crystal structure of the ENM. Furthermore, a strong coupling between incident photons, and electronic and phonon energy levels in some ENMs (e.g., CNTs) can result in resonant Raman shifts that are several thousand-fold more intense relative to those in bulk.^{Reference Jorio and Cancado15} Thus, Raman scattering serves as an ideal tool to study the biodistribution of ENMs over long-time scales.^{Reference Liu, Tabakman, Welsher and Dai12,Reference Schipper, Nakayama-Ratchford, Davis, Kam, Chu, Liu, Sun, Dai and Gambhir16–Reference Wingard, Walters, Cathey, Hilderbrand, Katwa, Lin, Ke, Podila, Rao, Lust and Brown18} As an example, we discuss the use of Raman spectroscopy in elucidating the biodistribution and immune response of CeO₂ NPs.^{Reference Wingard, Walters, Cathey, Hilderbrand, Katwa, Lin, Ke, Podila, Rao, Lust and Brown18}

The risk of CeO₂ NP exposure is significant since they are widely used in many applications, including solar cells and as fuel additives. Instillation of CeO₂ NPs in the lungs of mice was shown to activate mast cells, which play an important role in allergic immune responses, tissue repair, and the maintenance of homeostasis.^{Reference Wingard, Walters, Cathey, Hilderbrand, Katwa, Lin, Ke, Podila, Rao, Lust and Brown18} Such activation in mice induces pulmonary inflammation, an altered vascular tissue response, and an increase in infarct size within the left ventricle of the heart using a model of myocardial ischemia/reperfusion injury that mimics an episode of myocardial infarction, commonly known as a heart attack. Mice deficient in mast cells had a null effect to CeO₂ instillation, confirming that ENMs’ physiological effects are mediated, in part, by mast-cell activation. In these studies, the biodistribution of CeO₂ was characterized by micro-Raman spectroscopy of sectioned mouse tissue (Figure 2). The crystal structure of CeO₂ consists of a simple cubic O sublattice with Ce ions occupying alternate cube centers. This cubic structure results in a unique Raman signature consisting of a dominant peak at 462 cm⁻¹ (Figure 2b inset) with a typical line width of ∼20–30 cm⁻¹. This peak results from the vibrational motion (breathing mode) of O atoms around the Ce cations and can be used to assess the biodistribution of CeO₂. Consistent with the previously described mast cell mediated response, Wingard et al. observed the 462 cm^–1 Raman peak exclusively in mouse tissue, confirming the uptake of nanostructured CeO₂.^{Reference Wingard, Walters, Cathey, Hilderbrand, Katwa, Lin, Ke, Podila, Rao, Lust and Brown18} An important aspect of Raman spectroscopy is that it can facilitate depth profiling (through confocal Raman spectroscopy) for detecting CeO₂ NPs embedded deep inside the tissue sections (thickness ∼8–10 μm below the surface), unlike electron microscopy.

Figure 2.

Raman spectroscopy of CeO₂ nanoparticles in mouse lung tissue. (a) An optical image of a mouse lung section where CeO₂ was directly instilled into the lung through oropharyngeal aspiration. The inset image (10 × 10 μm) clearly shows black colored CeO₂ at the surface of the lung tissue. (b) The corresponding Raman map of the mouse lung section shown in the inset of (a). The red colored region is a map of the characteristic Raman peak (shown in the inset ∼triply degenerate (F_2g) mode at 462 cm⁻¹) for CeO₂ nanoparticles. The data are taken from Reference 18.

Biophysical phenomena at the nanoscale

ENMs in any physiological medium become inadvertently fouled or coated with proteins and lipids, leading to the formation of a “biocorona” (cf. Figure 1) (also see the Bhattacharya et al. article in this issue).^{Reference Podila and Brown2,Reference Podila, Vedantam, Ke, Brown and Rao19–Reference Shannahan, Lai, Ke, Podila, Brown and Witzmann21} Although it is known that proteins adsorbed on ENM exhibit significant structural changes, the underlying biophysical mechanisms of the ENM-protein interactions are not yet fully understood. It is plausible that a difference in electronegativity between the protein and ENM may induce a charge transfer, which may ensue in hydrogen bond breakage and, ultimately, lead to protein unfolding and a dysfunctional protein. More importantly, the presence of the biocorona influences metal dissolution, reactive oxygen species (ROS) production, and overall ENM interaction with cells to mitigate ENM-induced toxicity. Many groups have suggested that ROS and reactive nitrogen species (RNS) produced in the presence of ENMs are primarily responsible for most of the toxic damage to cells.^{Reference Nel, Xia, Madler and Li22}

While most metal oxide compounds exhibit relatively clean bandgaps (with no electronic states within the gap) in the bulk form, a limited periodicity of crystal planes in metal oxide ENMs in one or more dimensions results in new electronic or surface states within the bandgap. These surface states are often electronically active and allow charge transfer between O₂ and the ENM, resulting in an increased production of superoxide anions and OH radicals. TiO₂ and Al₂O₃ NPs (<50 nm in diameter) exhibit more toxic response per unit mass than their bulk counterparts due to the presence of surface or mid-gap states.^{Reference Oberdorster23–Reference Costa, Guignard, Zalma and Pezerat25} As explained in the proceeding sections, spectroscopic tools (Raman, photoluminescence, and hyperspectral imaging) can illuminate the fundamental mechanisms involved in biocorona formation, assess changes in ENM interaction, and provide facile ways to quantify the net bioavailable metal ions.

Formation of ENM-biocorona

As mentioned earlier, CNTs are known to exhibit strong Raman features due to resonance effects. Notably, the symmetric peak shape and peak frequency arising from the planar vibrations of carbon atoms in a CNT (G band) are highly sensitive to charge transfer (see Figure 3a). Indeed, the peak develops an asymmetric shape (viz., Breit-Wigner-Fano shape^{Reference Podila, Vedantam, Ke, Brown and Rao19}) due to charge transfer between CNTs and adsorbed species. The G band peak frequency upshifts (downshifts) when any acceptor (donor) species interacts with CNTs via hole (electron) transfer, while the peak shape-asymmetry factor (1/q) deviates from 0 (>0 for acceptor and <0 for donor species).^{Reference Podila, Vedantam, Ke, Brown and Rao19} Using micro-Raman spectroscopy studies, Podila et al. observed that the aromatic amino acids present in the proteins (e.g., bovine serum albumin or BSA) could interact with the unhybridized p _z orbitals of the single-walled CNTs (SWNTs) by providing a weak acceptor level in the SWNT electronic density of states to allow partial electron transfer from the SWNT to BSA.^{Reference Podila, Vedantam, Ke, Brown and Rao19} Such an electron transfer is clearly reflected in the asymmetric line shape and an upshift of the G band peak frequency found in the Raman spectrum of protein coated SWNTs (Figure 3a). It is worth noting that the introduction of a relatively weak acceptor level (from proteins) could still affect the Raman spectrum of the CNTs, since the latter possess sharp and discrete energy states. As discussed next, the spectroscopic evidence for the interaction between ENM and proteins go hand-in-hand, with changes in ENM uptake, immune response, and cytotoxicity as deduced from in vitro studies.

Figure 3.

(a) Raman spectra of pristine and bovine serum albumin (BSA)-coated single-wall nanotubes (SWNTs). Pristine SWNTs clearly exhibit a symmetric line shape (1/q = 0, where 1/q is the peak shape-asymmetry factor), while the BSA-coated SWNT spectrum is not only upshifted (as shown by the black arrows) but also shows an asymmetric line shape due to charge transfer. Note that the solid black (red) lines below the top-most (bottom) spectrum indicate the deconvoluted peaks obtained from the peak fits.^{Reference Podila, Vedantam, Ke, Brown and Rao19} (b) The surface plasmon resonance (SPR) shift for a BSA coating on Ag nanoparticles (NPs) occurs due to a change in the local dielectric environment. The presence of AgO on bare Ag NPs results in a thicker protein corona layer and the highest SPR shift.^{Reference Podila, Chen, Ke, Brown and Rao26} (c) Rat lung epithelial cell (RLE) and rat aortic endothelial cells (RAECs) were pretreated with serum-free (SF) media and subsequently exposed to Ag NPs with or without protein corona (HSA, BSA, and HDL-high density lipoprotein). Cells were then washed and assessed by inductively coupled plasma mass spectrometry (ICP-MS) to quantify Ag NP uptake. The presence of scavenger-receptor B1 inhibitor (red bars) reduced the uptake of Ag NPs, suggesting scavenger receptor B1 mediates the uptake of Ag NPs in both cell types and the corresponding inflammatory activation and cytotoxicity.^{Reference Shannahan, Lai, Ke, Podila, Brown and Witzmann21} (d–f) Transmission electron micrographs show a maximum protein coating for AgO/Ag NPs concurring with SPR spectroscopy results shown in (b) (scale bar: 20 nm).^{Reference Podila, Chen, Ke, Brown and Rao26} Note: PVP, polyvinylpyrolidone; HSA, human serum albumin.

Spectroscopic understanding of in vitro responses to ENM

Indeed, the formation of an oxide layer on many metal NPs (e.g., AgO on Ag NPs) introduces new surface states that result in an increased ENM interaction with proteins,^{Reference Podila, Chen, Ke, Brown and Rao26} leading to (1) a change in the dielectric environment surrounding the ENM upon protein or biocorona formation and a subsequent shift in the surface plasmon resonance (SPR) absorption of ENMs, and (2) quenching of the photoluminescence of aromatic amino acids in adsorbed proteins (particularly in the case of wide bandgap metal oxide ENMs). Previously, it was demonstrated that the shift in SPR (resulting from biocorona formation) is highest for bare Ag NPs with an oxidized surface (AgO), relative to citric acid or polymer-stabilized Ag NPs (Figure 3b, d–f). More importantly, the presence of a biocorona (particularly, human serum albumin or HSA corona) possibly reduces the cytotoxic effects of Ag NPs by inhibiting their rapid dissolution and release of toxic Ag⁺ ions. Inductively coupled plasma mass spectrometry (ICP-MS) measurements in vitro (see Figure 3c) showed that the presence of a biocorona on Ag NPs significantly decreased their uptake, cytotoxicity, and inflammatory response in cells that represent barriers to nanoparticle uptake in the vasculature and pulmonary systems (e.g., rat aortic endothelial cells [RAEC] and rat lung epithelial [RLE] cells).

Furthermore, hyperspectral analysis^{Reference Fairbairn, Christofidou, Kanaras, Newman and Muskens27} revealed that SPR spectra from Ag NPs with biocorona undergo a blueshift (unlike pristine Ag NPs that do not show any shift) upon cellular internalization, suggesting an intracellular loss of the biocorona that, in turn, may plausibly lead to elevated toxic effects via increased dissolution. A combination of SPR spectroscopy, hyperspectral imaging, ICP-MS, and more traditional in vitro cytotoxicity and inflammation assays conclusively suggest that (1) bare Ag NPs (without any capping agents) exhibit the highest protein coating or biocorona due to the presence of an AgO layer, (2) biocorona on Ag NPs reduces Ag⁺ leaching and thus toxicity, (3) biocorona formation decreases the uptake and hence cytotoxicity of Ag NPs, and (4) the protein coating on Ag NPs is lost upon internalization, possibly due to changes in pH and the local intracellular microenvironment.^{Reference Shannahan, Lai, Ke, Podila, Brown and Witzmann21,Reference Podila, Chen, Ke, Brown and Rao26}

The dynamic nature of biocorona formation leads to a complex physiological response to ENM, which warrants a complete thermodynamic understanding of ENM-protein binding and interactions.^{Reference Casals, Pfaller, Duschl, Oostingh and Puntes28} In this regard, isothermal titration calorimetry (ITC) is an important tool that can yield useful binding information, such as the enthalpy, stoichiometry, and binding constants. However, ITC experiments are expensive and time-consuming and need extremely stable suspensions with very accurately defined concentrations. In the case of metal oxide NPs, the need for ITC could be obviated by using NP-induced fluorescence quenching of proteins. Fluorescence quenching of proteins in metal oxide NPs occurs due to an energy transfer (FRET, fluorescence resonance energy transfer) arising from a significant overlap between bandgap absorption of many metal oxides (e.g., 3.37 eV in ZnO) with the emission of aromatic amino acids in proteins (e.g., 3.5 eV in BSA). Thus, one can deduce the binding constant (K) of some proteins on metal oxide ENMs using the traditional fluorophore-quencher binding mechanisms (Figure 4a) instead of ITC.^{Reference Kathiravan, Paramaguru and Renganathan29}

Figure 4.

(a) Photoluminescence spectra of tryptophan and tyrosine residues of bovine serum albumin (BSA: 3 g/dL) display quenching (i.e., decrease in emission intensity as shown by the black arrow) with increasing ZnO nanoparticle concentration. Such spectra could be used in conjunction with fluorophore-quencher binding mechanisms to obtain binding constant and stoichiometry similar to isothermal titration calorimetry.^{Reference Kathiravan, Paramaguru and Renganathan29} (b) Concentration of bioavailable Cu²⁺ in aqueous suspensions of nano CuO and bulk CuO as measured by luminescence of recombinant Escherichia coli MC1061 Cu-sensor. The inset shows concentration of bioavailable Zn²⁺ in aqueous suspensions of nano ZnO and bulk ZnO as measured using bioluminescence of recombinant Escherichia coli MC1061 Zn-sensor. The data originate from Reference 34. The bioavailable concentration of nano CuO is contrastingly different from the bulk, unlike the case of ZnO.

Biochemical toxicity of metal oxide nanomaterials

Although the toxic effects of metallic NPs reported for mammalian cells in vitro can be linked to oxidative stress,^{Reference Ivask, Juganson, Bondarenko, Mortimer, Aruoja, Kasemets, Blinova, Heinlaan, Slaveykova and Kahru30} for a comprehensive understanding of biochemical toxicity of metal oxide ENMs, the release of (toxic) metal ions should be taken into account, which can be achieved using novel spectroscopic tools. Indeed, Brunner et al.^{Reference Brunner, Wick, Manser, Spohn, Grass, Limbach, Bruinink and Stark31} showed that the solubility of oxide NPs strongly influences their cytotoxicity. Through in vitro studies, CuO, ZnO, and Ag NPs have been shown to release metal ions (Cu²⁺, Zn²⁺, and Ag⁺)^{Reference Bondarenko, Juganson, Ivask, Kasemets, Mortimer and Kahru32} that can increase oxidative stress and possibly cause cell apoptosis (i.e., programmed cell death).^{Reference Ivask, Juganson, Bondarenko, Mortimer, Aruoja, Kasemets, Blinova, Heinlaan, Slaveykova and Kahru30}

Traditionally, the metal ion content in natural or biological matrices is determined using ICP-MS after the digestion of samples with strong acids. While ICP-MS exhibits high sensitivity (parts per trillion) in metal detection, it does not allow a clear distinction between bioavailable (potentially hazardous) and non-available (potentially non-hazardous) fractions of metals to living organisms. Use of bioluminescent metal-specific recombinant bacteria is an excellent spectroscopic tool for assessing the true bioavailable content of metal ions. In these metal-specific bacterial sensors, the expression of a reporter gene (e.g., encoding luciferase) is controlled by a genetic regulatory unit (or receptor), which specifically responds to a particular bioavailable metal ion.^{Reference Ivask, Rõlova and Kahru33} For example, zntA and zntR genes introduced by genetic manipulations into Escherichia coli (a host bacterium) transform the bacterium into a living Zn-sensor that recognizes bioavailable Zn²⁺ ions and subsequently controls the expression of luciferase—a reporter protein. Thus, the amount of bioavailable Zn²⁺ that has entered the bacterial cell could be quantified. Kahru and co-workers have used recombinant bacterial sensors to determine the influence of nanosizing on the bioavailability of metal ions in CuO and ZnO NPs.^{Reference Heinlaan, Ivask, Blinova, Dubourguier and Kahru34} Interestingly, the bioavailable content of Zn²⁺ ions did not significantly differ between bulk and nanosized ZnO, while there was a stark contrast between Cu²⁺ release from CuO NPs relative to its bulk form (Figure 4b). Such observations could be attributed to an increased surface area at the nanoscale and a positive reduction potential for Cu (Cu/Cu²⁺: 0.34 V) in contrast to Zn (Zn/Zn²⁺: –0.76 V). Coherent with the data on bioavailable fraction, the toxicity of CuO NPs in bacteria Vibrio fischeri and water fleas Daphnia magna was 48–52 times higher than that of the bulk formulation. On the other hand, the toxicity of bulk and nanosized ZnO was comparable.^{Reference Heinlaan, Ivask, Blinova, Dubourguier and Kahru34} These observations reaffirm the hypothesis that ENMs can exhibit an “action-at-a-distance” (cf. Figure 1) toxic response without necessarily entering the cells. The release of metal ions may also cause the toxicity through increased generation of cellular ROS.