8_Testing Metal_Oxide Nanomaterials for Human Safety
Oct 29, 2015
-
Nanomaterials
R. Landsiedel, L. Ma-Hock, A. Kroll,
D. Hahn, J. Schnekenburger, K. Wiench,
W. Wohlleben* ................................xxxxa
Testing Metal-Oxide Nanomaterials for
Human Safety
adma.200902658C
aFinal page numbers not assigned
The novel properties of engineered nanomaterials may alter their interaction with the
human body, especially for inhalation of unintentionally released biopersistent material.
We discuss the characterization of nanoparticles in interaction with biological media
and we review animal inhalation and cell culture studies in comparison to original
results. We establish that an intrinsic size-specific toxicity does not exist and identify
material-specific indicators of concern that help to select safe uses.
REVIEW
-
Testing Metal-Oxide Nanomate
By Robert Landsiedel, Lan Ma-Hock, Alexand
Jurgen Schnekenburger, Karin Wiench, and W
1. Introduction
The intentional generation and application of nanomaterialswith novel properties is one of the centurys key technologydevelopments, offering extraordinary opportunities in varioustechnological fields such as electronics, energy management,structural materials, functional surfaces, construction, andinformation technology, but also in the pharmaceutical andmedical field. Indeed, the appearance of clean-tech, seen as thecapture, storage, and conversion of energy and resource-efficientmaterials, depends critically on nanomaterials, whereof themajority is fabricated by compounding engineered particulatenanomaterials.
Since the miniaturization of materials down to the nanometerscale can change physical and chemical properties, nanomaterials
current developmentsand Canada.[7] Based ofrom potent to haroccupational to consevolves into a case-by(internal) exposure (Sstrategy, e.g., in thknowledge is sufficiebut these have toassessments.[8] The opart of the public awar
In the present coneffects of engineeredto the frequently diAppropriate toxicity tnanomaterial specificbody and possible ncellular level (Scheme 2). The unique nano-specific properties ofnanomaterials require a careful adaptation of the test methods,and the OECD recommends that guidelines be newly developed
y, degradation andphysicochemicalpplicable toxicityas been suggested
[12] dies withrse effectsentail the
REVIEW
www.MaterialsViews.comwww.advmat.de
1
2
3
4
5
6
7
8
9
10
11
12
13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
[*] Dr. W. Wohlleben, Dr. R. Landsiedel, Dr. L. Ma-Hock, Dr. K. Wiench
Westfalische Wilhelms-Universitat Munster
re
ic
bi
fo
ha
za
te
ie
i
tie
d
Zr
d
tion and
production of safe materials.DOI: 10.1002/adma.200902658already by Warheit et al. At present, inhalation stuanimals are the most predictive testing of possible adveof nanomaterials on humans. But inhalation studies
Schloplatz 2, 48149 Munster (Germany)or revised for sample preparation and dosimetrfate, for inhalation and for the majority of thecharacterization methods.[11] A base set of ascreening systems and characterization tools h
BASF SE 67056 Ludwigshafen (Germany)E-mail: [email protected]
Dr. A. Kroll, Dr. D. Hahn, Dr. J. SchnekenburgerGastroenterologische Molekulare Zellbiologie, Medizinische Klinikund Poliklinik Btubes>>CeO2, ZnO> TiO2> functionalized SiO2> SiO2,
black. Enhanced understanding of biophysical properties anNanomaterials can display distinct biological effects compa
bulk materials of the same chemical composition. The phys
characterization of nanomaterials and their interaction with
are essential for reliable studies and are reviewed here with a
used metal oxide and carbon nanomaterials. Available rat in
culture studies compared to original results suggest that ha
not determined by a single physico-chemical property but ins
a combination of material properties. Reactive oxygen spec
fiber shape, size, solubility and crystalline phase are known
nanomaterials biological impact. According to these proper
marized hazard potential decreases in the order multi-walle
results in improved testing strategies and enables the selecAdv. Mater. 2010, 22, 127 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Final page numbers not assignedexposure can be enforced by technicalmeasures.
The scientific community started toevaluate the potential hazard of nanomater-ials since 1992,[36] culminating in theof regulatory frameworks in the EU, USA,n the extreme diversity of hazard potentialmlessand diversity of exposurefromumer settingsthe regulatory framework-case risk assessment. Hazard potential andcheme 1) need to be merged into a testinge REACH Implementation Plan. Currentnt to shape the first regulation approaches,undergo revisions with enhanced risk
utcome of safety research is also an integraleness and confidence in nanotechnology.[9,10]
tribution, we focus on the potential adversemetal oxide nanomaterials, in comparisonsscussed toxicity of carbon nanomaterials.esting requires a thorough understanding ofproperties with regard to distribution in theano-specific effects on the systemic andrials for Human Safety
ra Kroll, Daniela Hahn,
endel Wohlleben*
will presumably also influence biologicalsystemsregardless of a human intentionbehind the materials generation. Thenatural nanomaterials and the unintention-ally man-made nanomaterials by far out-weigh the engineered nanomaterials, butthe exposure scenarios resemble eachother.[1] A systematic risk assessmentrequires the separate determination of boththe hazard potential and the actual exposurelevels resulting in a risk characterization(Scheme 1).[2] Typical consumer productscombine low exposure to free nanostruc-tures and low hazard potential. Materialswith high hazard potential are restricted toprofessional handling, where safe levels of
d with
o-chemical
ological media
cus on widely
lation and cell
rd potential is
ad depends on
s generation,
ndicators of
s the sum-
carbon nano-
O2, carbon
cellular effects1
-
REVIEW
www.advmat.dewww.MaterialsViews.com
enamel are examples of organicinorganic hybrid nano-materials with biopersistence.
(ii) If there is emission of free nanoparticles, are these liquid,soluble or biopersistent? Natural and technological processescan produce ultrafine droplets or nanoemulsions (e.g., milk).The ultrafine state may affect the uptake of a substance inthe body, but inside the body the substance will dissolve orblend in body fluids and only effects different from thoseassociated with nanometer sizes are expectable. In contrast,biopersistent nanomaterials could exhibit general nanome-ter-size-specific effects if internalized. Among the natural(biogenic, geogenic, or pyrogenic) sources of biopersistentnanomaterials, black carbon from incomplete biomasscombustion dominates with 50 to 270 megatons peryear,[16,17] followed by 16 megatons of inorganic dust fromdesert storms.[18] But also human activity releases nanoma-terials as unintended by-products. A typical urban atmo-sphere contains 10mg m3 particulate matter (around105 particles m3); a candle or a cigarette release 10 gm3
particles (around 2 1011 m3).[18] Welding fumes consist of
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
sity and ENS Paris. He earnedhis PhD in 2003 from LMUMunchen with a biophysicalthesis on energy harvesting inphotosynthesis, performed atthe Max-Planck-Institute forQuantum Optics. He thendeveloped chemically selectivemicroscopy in Marburg andjoined BASF polymer physicsresearch in 2005, acting also asinnovation manager for BASFs
nanotechnology activities. His research focuses on prepara-tion, characterization, and self-assembly in complex suspen-sions, especially with regard to the safety of nanomaterials.
2sacrifice of animals and are quite expensive and time-consuming.Traditional methods have to be adapted and in vitro methods[13]
must be improved through better understanding of theirbiophysical mechanisms until the in vitro tests achieve predictivepower.
This paper is organized as follows:Section 2 starts from a wider perspective and discusses
exposure levels and possible routes of internalization in humans.Sections 36 track the physiological effects from biophysical to
cellular to systemic levels.In Section 3 we review the physicochemical properties
of nanomaterials and their characterization with appropriatebiophysical methods. We focus on the biophysical modification ofthe nanomaterials surface and state of agglomeration in cellculture media (Scheme 3).
The in vitro toxicity (cell viability, genotoxicity, inflammation)of metal oxide and carbon nanomaterials is reviewed in Section 4and is complemented by original results from different titaniumdioxide (TiO2) nanomaterials (Scheme 3).
In Section 5 we give an overview on the limited range ofexisting inhalation studies with engineered nanomaterials.Furthermore, we present original data from our inhalationstudies with six metal oxide materials and two carbon materials.These results are excellently comparable due to an identicalexperiment design (Scheme 4).
Section 6 summarizes the correlations between the in vivo andin vitro chapters, leading to a ranking of hazard potential for thematerials tested in Section 7.We identify materials properties andin vitro indicators that should trigger in vivo experiments in afuture testing strategy.
2. Emission versus Exposure
2.1. Emission Quantities and Possible Routes of
Internalization
Potential human exposure to nanomaterials is as manifold as thepotential applications of different nanomaterials. It is beyond thescope of the present contribution to assess all factors in detail.The following paragraph introduces four questions to guide aprioritization:
(i) Emission of nanoparticles from composites or powders?Touching a composite thermoplastic that was reinforced withsilicon dioxide nanoparticles (Scheme 1) is of less concernthan being exposed to free nanoparticles. Consumerapplications of nanomaterials focus on composite materialsfrom which only the unintended release of fragmentscontaining nanoparticles during use, recycling, or disposalmay raise concerns.[14] Given typical product lifetimes on theorder of years, the dose of release from composites shouldbe vanishingly low, even for a hypothetical completedegradation. First available evidence supports this assump-tion: Abrasion of acrylate coatings containing ZnO nano-particles did not lead to significant release of nanoscaleaerosols.[15] Not intending to banalize the issue, one shouldkeep inmind that evolution itself developedmost remarkablenanostructured materials: Human bones and human tooth 2010 WILEY-VCH Verlag GRobert Landsiedel studiedchemistry, food chemistry, andtoxicology in Kaiserslautern,Mainz, and Leipzig. Afterworking for the state, he earneda PhD with a thesis on themetabolism and mutagenicityof benzylic compounds. After aPostdoc in Potsdam, he joinedBASF in 1999 and worked indifferent functions in Ludwig-shafen, North Carolina, andTokyo. Since 2004 he heads a
unit of several toxicological routine and research labs and isinvolved in projects on alternative methods andnanotoxicology.
Wendel Wohlleben studiedphysics at Heidelberg Univer-mbH & Co. KGaA, Weinheim Adv. Mater. 2010, 22, 127
Final page numbers not assigned
-
REVIEW
www.MaterialsViews.comwww.advmat.de
ments. Measure-ments at workplacesin nanoscale TiO production did not
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
panallletsScheme 1. The generally accepted principle assesses risk as: RiskHazardcontrolled by low hazard or low exposure, ideally both. The examples fromshow the laboratory synthesis of nanostructured battery materials (top left)thermoplastic nanocomposites (bottom left). In contrast to the toxicologicexposure, we show here the external exposure. External exposure does notcases, as demonstrated by the case of nanostructured sun screen pigmen109m3 metal nanoparticles.[19] These values set a frame ofreference, and they justify the use of CB as referencematerialin safety testing. The world production of CB for tires andprinting inks is estimated around 8 megatons per year(in 1996).[20] Emission of CB is relevant in aerosol form, butits quantity is vanishingly small compared to the backgroundof black carbon.[21]
(iii) Is the emission intended or unintended? There are alimited number of applications of biopersistent nanoma-terials with intentional (external) exposure of the humanbody, especially as sunscreens in cosmetics (Scheme 1).The global turnover with engineered nanomaterials[22] canbe converted into very rough estimates for the quantitiesthat were actually produced in 2007: metal oxidenanoparticles: 0.02 megatons (20000 tons); metal nano-particles: 20 tons; carbon nanotubes (CNTs): 100 tons.Graphene catches up with 15 tons in 2009.[23] These valuesare worlds apart from CB, but they still outnumberspecialties in the OECD sponsorship program likequantum dots, dendrimers or fullerenes/C60.
[24] The vastlydominant applications are technically bound: CNTs andgraphene in polymer nanocomposites, metal nanoparticlesin catalysts, electronics, and antimicrobials. Note thatnanoscale silver ranks high only when the number ofmarketed products is counted,[25] but not among theproduction quantities. Metal oxide nanoparticles find broadapplication from coatings and plastics over catalysts to
exposure andother hand, tultrafine parthuman healparticles.
In summary, ththe unintended emay occur mostlyconsumer settingeffects by inhaledtigation of effectials,[31] and amonoxides.
2.2. Approaches t
The US OSHA PIndustry is 5mgunder Inert or NGovernmental InValue states the s(insoluble or poordistinction betwee2001 to PNOS.
Whereas inExposure Limit (O
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
skin damage (bottom right).
Adv. Mater. 2010, 22, 127 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinh
Final page numbers not assignedcomprehensively tested on skin and sev-eral studies demonstrated that the intacthuman skin is an effective barrier for thosenanoparticles.[29,30] The absence of dermalpenetration minimizes human internal
hence minimizes the health risks. On thehere is a wealth of information on the effects oficles in the air[18,19] indicating the concerns forth arising from the inhalation of ultrafine
e highest concern for human health arises fromxposure to biopersistent nanoparticles. Thesein workplaces, and to a much lesser extent ins. The existing knowledge of adverse healthultrafine particles gives priority to the inves-
s caused by inhaled engineered nanomater-g these, the emission quantities prioritize metal
o Regulation
ermissible Exposure Limit (PEL) for Generalm3 time-weighted average (TWA) (PEL listeduisance Dust). The American Conference ofdustrial Hygienists (ACGIH) Threshold Limitame limit value of 5mgm3 TWA for Particlesly soluble) Not Otherwise Specified (PNOS). An inhalable and respirable dust was changed in
Germany the legally binding OccupationalEL) for inhalable dust is 10mgm3, there is2
reveal any significant emission.[2]
(iv) Is the exposure oral, dermal, or byinhalation? Nanotechnology in food pro-cessing focuses on nanostructures forencapsulation, whose degradation in thehuman body is essential to fulfill theirpurpose. Migration of particles larger than1 nm from packaging materials into foodseems to be no concern.[28] This mayexplain why relatively few investigations onthe absorption and effects of nanoparticlesvia the oral route are available. TiO2 andZnO nanoparticles in sun screens were
Exposure. Risk isresent technologyd SiO2-reinforcedy relevant internalad to uptake in all[29,30] that preventsunscreens. Specifically nanoscale TiO2 isused for coatings and sunscreens, with anestimated production of 0.005 megatonsper year,[20] expected to grow to 0.06mega-tons per year until 2025.[26]
Sunscreens represent one of the fewnanomaterial-containing products towhich humans are intentionally exposed.Preliminary scenarios of coating degrada-tion[27] estimate levels of unintendedemission around 102 mg m3 in airand around 10mg L1 in soil compart-
[20]eim 3
-
REVIEW
www.advmat.dewww.MaterialsViews.com
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
4in addition an OEL for respirable dust (
-
REVIEW
www.MaterialsViews.comwww.advmat.de
1
2
3
4
5
6
7
8
9
10
11
12
1
2
3
4
5
6
7
8
9
10
11
12layer of albumin is adsorbed to the surface of the largest andmost hydrophobic particle with an adsorption constant around106 mol1, whereas a sparser layer is associated with the morehydrophilic particles.[38] Over time, albumin has a residence timearound 100 s[43] and is replaced, e.g., by apolipoprotein A-I, aprotein of 30-fold lower abundance, but with higher affinity andslower kinetics.[41] By 1D gel comparison of commercial polymernanoparticles with different chemical surface functionalization,typically 40% of the corona proteins have been found to beconserved between amine, plain, and carboxyl-modified poly-styrene nanoparticles, and around 30% of the corona proteins arespecific to a single functionalization.[42] Preferential conditioning
media (20 to10the nanoparticle su55mV) by protemust be attributedwith isotonic saltmore protein, demmass spectroscopsurface[2] and indthe protein fractnearly the sameSection 5, proteindepletion, but on
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Scheme 3. Workflow of in vitro testing. The nanomaterials are dispersed in a physiologicalnutrient medium that contains proteins and other macromolecules (coils) and low-molar-masscomponents such as salts (dots). The nanomaterial surface changes by differential adsorption ofsome of these components, correlated with changes in the state of agglomeration. Onlyafterwards, the interaction with a multitude of cell species is studied by the (typically optical)readout of a large number of markers and endpoints. Details are shown for the threemarkers thatare essential for the discussion in Section 4.
Adv. Mater. 2010, 22, 127 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinhe
Final page numbers not assignedmethodical advantage of being nicely homo-geneous, spherical, and well dispersible.Corona-related mechanisms are known tomuch less detail for inorganic nanomaterials.First results indicate that the corona is selectivealso for different naked inorganic surfaces: Forinstance, specific pro-inflammatory and anti-inflammatory precursors displayed an up to 30times higher affinity to Ni and Al particles thanalbumines, as demonstrated by isotope label-ing (Fig. 1).[46] Likewise, diamond nanoparti-cles showed a high affinity for vitronectin,which can stimulate tumor necrosis factor a(TNF-a) release from alveolar macrophages.However, even with its low relative affinity,the high concentration of albumin in serumstill represents a significant portion of thebound protein fraction for all nanoparticles.[46]
Quantitatively, albumin adsorption onto car-boxylic-acid functionalized inorganic nano-particles was anti-cooperative and saturatedat serum level concentrations and onemonolayer.[43] Working with semiconductor(quantum dot) particles, Frangioni and co-workers[47,48] showed in a series of experi-ments how surface functionalization controlsbiodelivery: Particles were filtered by renalclearance and urinary excretion only fordiameters below 6nm and with zwitterionicor neutral organic modification to preventprotein adsorption. The significant coronaconservation between different polymer parti-cles[42] is reflected by the uniformity of surfacecharge of various naked metal oxide nanopar-ticles when dispersed in serum-containingmV),[49,50] attributed to a universal coverage ofrface (with zeta-potential ranging from25 toins.[49] Part of the reduction of zeta-potentialto charge-screening in the physiological bufferload.[51] As expected, smaller particles adsorbonstrated directly by UV and secondary iony (SIMS) detection of the colloidal ZnOirectly by BCA assay (bicinchoninic acid) ofion that did not adsorb onto TiO2.
[52] Onseries of naked metal oxides as described inadsorption was shown to even induce bufferly at completely unphysiological nanoparticleby immunoglobulin IgG induces clearance byMPS cells, whereas dysopsonins (albumin,IgA) prolong circulation in the bloodstream,[44]
and bovine serum albumin (BSA) conditioningdecreased resorption into lung tumor cells.[45]
In summary of the serum interactions,neutral particles seem to have slower adsorp-tion kinetics than charged particles, dittohydrophobic particles, but these also differin protein identity in the protein corona fromhydrophilic particles.[35]
Polymer nanoparticles have the obviousim 5
-
REVIEW
www.advmat.dewww.MaterialsViews.com
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
inepticno
6concentrations >10mgmL1.[52] Interestingly, the adsorptionwas blocked by pretreatment of the particles in serum, indicatinglonger residence times than on the polymer-functionalizedparticles.[43]
The available results indicate that the protein adsorption and
Figure 1. Detailed qualitative characterization of the nanoparticleproteinprotein abundance in free and bound fraction as estimated from the total p11 most abundant proteins that were common between Ni and Al nanoparprotein corona is specific and selective for the different pristine na(figure redrawn with permission from ref. [46).biokinetics of (stabilized) polymer particles and (polymerstabilized) inorganic particles follow the same mechanisms.
First evidence emerges also on the interaction betweenparticles and the lung lining fluid, the first conditioning contactafter inhalative exposure. Apart from proteins, also phospholipidsfrom lung lining fluid have been shown to adsorb to nanoparticlesurfaces,[53] and we demonstrated by antibody staining that thedominant surfactant protein SP-A does adsorb onto metal oxidenanoparticles.[2] The direct comparison of conditioning CNTs ineither serum-containing medium or in dipalmitoylphosphati-dylcholine (DPPC)-containing medium showed a significantlystronger inflammation potential with the DPPC, demonstratingthe direct impact of the corona on cytotoxicity.[54]
Sometimes, however, minute differences between nanoparti-cle surfaces strongly change the biodistribution: Surfacefunctionalization with poly(ethylene glycol) (PEG) of varyingchain length typically considered an inert molecule resultedin major changes in organ/tissue-selective biodistribution andclearance from the body,[47] although 2D gel electrophoresisshowed that immune-competent proteins (IgG, fibrinogen) bindmuch more than albumins irrespective of PEG chain length.[55]
Verma et al.[56] demonstrated that of two nanoparticle isomerswith identical hydrophobic content, one functionalized withsub-nanometer striations of alternating anionic and hydrophobicgroups, the other with the same moieties in random distribution,only the striated particles penetrated the plasma membranewithout bilayer disruption. Such phenomena make it difficult to
carbide nanoparticIt has been shownsmaller average(Fig. 3b).[51,66] Woconcentration, whtranslocated intoorkers[67] establishcarbide is preventuntil complete ashowed that BSAprocess.[67] Alternstabilization by PEDMEM/FBS.[68] Ahumic acids can act as wetting and dispersing agents fornanoparticles and [61,69]
demonstrating anaddition of serumreported.[70,71] Theof proper charactethis apparent con
There is an e(de)agglomerationidentified in theBronchoalveolar lain which to susnanoparticles, espmost important s
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
2010 WILEY-VCH Verlag GmbH & Co. KGaA, WeinCNTs. However, contradictory resultsincreased agglomeration of nanoparticles byproteins or organic acids has also been
time course of colloidal stability and the choicerization methods may be essential to resolvetradiction, see Section 3.3.ssential need for studies investigating thepotential of the other ligands that have beenprotein corona of conditioned nanoparticles.vage fluid (BALF) was reported to be a vehiclepend organic (soot)[53,72] and metal oxideecially in reduced compositions with only theurfactant protein/phospholipid (phosphatidylNTs, metal nanoparticles, metalles,[63] and metal oxide nanoparticles.[51,61,6467]
that a higher protein concentration leads to aagglomerate size of the nanoparticles
rking with a 50-fold excess of serum proteinich is the relevant range for nanoparticles thatthe human blood stream, Richter and cow-ed that agglomeration of TiO2 and wolframed over more than 40min, compared to 5mingglomeration in DMEM. Furthermore, theyalone is sufficient to prevent the agglomerationatively, suppression of adsorption and stericG functionalization also stabilizes particles inlso natural organic matter such as fulvic orrelate materials properties directly to physio-logical effects without knowing the biophysicalinteractions (compare abstract figure).
We conclude that despite the human riskbeing dominated by inhalation exposure andbymetal oxide nanoparticles, most work on theprotein corona has been devoted to polymericnanoparticles and serum proteins, oftenrestricted to albumin. In future, metal oxidesand lung lining fluid interaction with theirhigher relevance for human safety should be inthe focus.
3.2. State of Agglomeration
Clearly a correlation between the biologicalsurface conditioningcontrolled by the che-mical functionalizationand the colloidalinteraction between the thus coated particlesis to be expected.[24] In good qualitativeagreement, numerous studies established thataqueous suspensions of non-functionalizednanoparticles are stabilized against agglom-eration by the addition of bovine/humanserum albumin (BSA/HSA) and some other
proteins. The effect has also been exploited in production for thedebundling and dispersion of graphene and CNTmaterial beforechemical compounding (Fig. 2).[57,58] Especially albumins inwater or DMEM have dispersed and stabilized a wide variety ofnanomaterials: C [57,5961] [62]
teraction: Relativetide score, for theles. The adsorbedparticle surfacesheim Adv. Mater. 2010, 22, 127
Final page numbers not assigned
-
REVIEW
www.MaterialsViews.comwww.advmat.de
The intrinsic polydispersity and inhomogeneity of nanomaterialsrepresent major obstacles for a biophysical characterization.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31choline) constituents[73] or mixtures of BSA and DPPC.[61] Theuse of extensive ultrasonication in these experiments[61,73,74]
makes it difficult to compare the experimental results tocalculations that find that the interaction energy betweenTiO2agglomerates cannot be overcome by the interactionbetween the particles and DPPC.[75] In a comparison of eightmetal oxide nanomaterials, we showed that the anticipatedinterface activity of surfactant proteins is in general notsufficiently strong to overcome the agglomeration or flocculation
[2]
Figure 2. Qualitative screening of the dispersing action by the proteincorona on CNTs. In varying environments (here: basic, neutral, acidic pH),the different proteins (lower axis) adsorb effectively to CNTs and ensue adispersing action, visualized directly by the black color of dispersed CNTs.Figure reproduced with permission from ref. [57].tendency due to other components in complete BALF.Strikingly, the two particles that were functionalized withsynthetic polymers evaded near-complete agglomeration and atthe same time differentiated by low overall protein adsorption,but strong SP-A interaction.[2] The physicochemical results are ingood agreement with histological studies of lung slices afterinhalation exposure of rodents, in which the particulate materialthat was deposited on the lung surface was found in the form ofagglomerates.[76]
While the deagglomeration potential by natural macromole-cules certainly changes transport and biokinetics, a deagglom-erated nanomaterial is in general not more potent, as demon-strated by the example of polymer-functionalized BaSO4 that stayswell-dispersed in a variety of media, but has virtually no in vitro orin vivo effects.[2]
What is the mechanism of dispersion by interface-activeproteins? Given the rather low zeta-potential of conditionednanoparticles,[49,51] the dispersing effect of the protein corona isnot related to electrostatic repulsion. Instead, the stabilizationmust be a steric mechanism, whereby the entropy decreases if theprotein coronas of approaching nanoparticles start to overlap.Electrostatic stabilization collapses in high ionic strength bufferssuch as DMEM with 0% FBS, then shifts to a steric stabilizationby the adsorbed proteins in 100% FBS. The steric stabilization bybiopolymers has been exploited industrially for a long time since
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Adv. Mater. 2010, 22, 127 2010 WILEY-VCH Verlag Gmb
Final page numbers not assignedTraditional methods already fail to characterize nanomaterials ina controlled environment (e.g., distilled water one surfac-tant).[11] This situation is drastically aggravated under physiolo-gical conditions since at least 40 components add to the colloidaldomain and interact with each other and with the nanomaterial.Recent contributions by Hussain, Tiede, Powers, the OECD andothers stress the need for a conscious characterization beyond thenave application of characterization methods that claim to coverthe relevant parameters of nanotoxicology.[62,64,7982] The mostimportant parameters and appropriate measurement techniquesare summarized in Table 1.
3.3.1. Intrinsic Properties
Impurities are an issue especially for CNTs, with catalysts(nanoparticulate Co, Fe, Ni, and Mo) and amorphous carbonbeing present during their synthesis that may impose additionaltoxic effects.[59] Such trace elements were the subject of previousstudies on welding fumes.[83] The distribution of primary particleand aggregate sizes of a pristine nanoparticle powder requiresproper statistics of at least 106 nanoparticles, corresponding tomore scanning electron microscopy (SEM) images than reportedusually. Some nanoparticles are not stable in aqueous solutionsand can release chemical substrates. If particles are designed todissolve in aqueous solutions like water-soluble quantum dots[84]
or show an intrinsic, size-dependent dissolution in aqueousmedia like ZnO,[85] particles will release metal ions whenprotection colloids such as gelatine or starch stabilize organiccomposite particles.[77] Structural models from X-ray diffractionseem to suggest that proteins fold into a single well-definedstructure, which would eliminate the entropy stored in thestructural degrees of freedom, hence disabling steric repulsion.However, most proteins are minimally stabilized mesoscopicbiopolymers whose configuration fluctuates around the time-average structure under physiological conditionsa field thatwas pioneered by Kai Wuthrich (Nobel Prize 2002).[78]
The quantitative degree of deagglomeration is controversial,due to (i) the use of different dispersion protocols and (ii) thedisagreement of measurement techniques, which will bediscussed in more detail in the following subsection. Thedispersion protocol defining shear rate, energy input andduration of conditioning has a drastic influence on the resultingstate of agglomeration as established also in ISO 14887 SamplepreparationDispersing procedures for powders in liquids.One can either mimic the dispersing action that we assume to beactive in the human body, and since the blood stream is laminarwith rather low shear forces, ultrasonication should be omitted.Alternatively, one assumes that only the most dispersed fractionhas a profound effect; then one can try to prepare the totaladministered dose in the most dispersed state, using wettingagents, vortexing, and ultrasonication. Since both approacheshave been pursued, biophysical data published so far are hardlycomparable on the quantitative level.
3.3. Characterization of Nanomaterials for Biological TestingH & Co. KGaA, Weinheim 7
-
REVIEW
www.advmat.dewww.MaterialsViews.com
than particle effects. A high surface/mass ratio of nanoparticulate
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
oo
pa
8materials results in excess surface energy enhancing particlecatalytic activity, depending on the crystalline phase. A variety ofnanoparticles such as metal oxide nanoparticles, fullerenes andsilica (SiO2) particles were reported to produce reactive oxygenspecies (ROS) in cell free systems.[8689] ROS production wassize dependent, e.g., small 24 nm-sized nanoparticles had a1001000-fold increased kinetics compared to 100 nm-sizednanoparticles.[90] Redox-active nanoparticles may cause falsepositive signals in assays based on substrate oxidation or in assaysdetecting cell stress induced ROS production. Few metal oxidenanoparticles like Fe2O3 are magnetic and may generate localmagnetic fields leading to the production of free radicals that inturn may interfere with cytotoxicity test methods based on redoxreactions.[91,92] ROS measurement by electron spin resonance(ESR) is a valid, but not widely available technique. However, ROSinside cells can be tracked as detailed in the materials andmethods.
The characterization of chemical composition and purity,crystalline phase, morphology, and specific surface can beregarded as relatively safe and well establishedthe same doesnot hold true for the (last two properties of Table 1) state ofagglomeration and corona conditioning effects.
3.3.2. State of Agglomerationintroduced into biological media. Cytotoxicity assays that aresensitive to metal ions may then rather reflect metal ion toxicity
Table 1. Most important properties and the appropriate characterization t
Minimal Characterization Needed for Comparability of Studies
Chemical composition and purity (pristine nanoparticles)
Crystalline phase (pristine nanoparticles)
Morphology, primary particle size (pristine nanoparticles)
Specific surface (pristine nanoparticles)
Solubility (in water and after conditioning in the test medium)
Surface chemistry (pristine nanoparticles)
Advanced Characterization for Mechanistic Understanding of Observed Effects
Catalytic activity, esp. ROS generation
Protein corona (in vitro: conditioned nanoparticles)
State of agglomeration and potential of deagglomeration (in vitro: conditioned nanoFor inhalation aerosols, the Scanning Mobility Particle Sizer(SMPS) is widely used to determine size distributions ofsubmicron aerosols, by balancing the electrostatic force onparticles in an electric field with their aerodynamic drag as theycross a laminar gas flow.[93] Aerosols of nanoparticles can begenerated using a dry powder aerosol generator and bynebulization of particle suspensions. The mass concentrationof the particles in the inhalation atmosphere can be determinedgravimetrically, and the particle size using a cascade impactor, anoptical particle counter, or the SMPS. Such dispersion techniquesgenerate fine aerosols with particle size distributions in therespiratory range, but with no more than a few mass percent ofultra-fine material (i.e., agglomerates 0.1wt % agglomerates thatare present in nearly every physiological suspension ofnanoparticles should be considered and second the well-known
ls.
Method
XRD, ICP-MS
XRD
SEM
BET
ICP-MS of supernatant
zeta-Pot., SIMS, XPS
ESR
SDS-PAGE, zeta-Pot., SIMS
rticles//inhalation: aerosol) AUC, cryo-TEM//SMPSfact that retrieving a size distribution from the autocorrelationcurve is a mathematically ill-posed problem[99101] that failsespecially for very broad distributions such as the four orders ofmagnitude in physiological suspension of nanoparticles (Fig. 3a).However, with careful sample preparation and elimination of thevery coarse agglomerates by ultrasonication, Hussain et al.[62]
obtained sub-micrometer average diameters for physiologicalsuspensions of Cu, Al2O3, Al, Ag, TiO2 nanomaterials, and theyconfirm also by DLS that serum-containing media reach thesame, often lower, diameters as in water, thereby drasticallyreducing the agglomeration compared to serum-free cell culturemedia. Themajority of published data from scattering techniquesneglects the very loose structure of nanoparticle agglomerates.The standard Mie routines such as those that retrieve relativeconcentrations and distributions in commercial DLS software
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
mbH & Co. KGaA, Weinheim Adv. Mater. 2010, 22, 127
Final page numbers not assigned
-
REVIEW
www.MaterialsViews.comwww.advmat.de
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24oFigure 3. Colloidal characterization of physiological suspensions of nanassume a solid spherical shape for the StokesEinstein relation.In reality, diffusion-limited colloidal agglomeration leads to afractal morphology, and this has been proposed as the dominatingtransport mechanism.[49] The fractal dimension can be deter-mined explicitly by static (light, X-ray, neutron) scattering, and formany colloids a universal fractal dimension of 2.1 has beenfound.[102] The fractal shape has been incorporated into DLSevaluation only by specialist particle labs.[103] Concluding the DLSdiscussion, dynamic, and static light scattering (DLS, multianglelaser light scattering(MALLS)) as well as Fraunhoffer diffractionprovide complementary information only if it is known fromother sources that size distributions are narrow.
A mighty tool for the characterization of nanocolloids(0.510 000 nm diameter) is the analytical ultracentrifuge(AUC)[104106] especially the universal interference opticsBeckman XLI with widespread use in the proteome busi-ness[107109] and, only to a lesser extent, also the disc centrifuges(Brookhaven Instruments XDC, CPS Instruments DC24000)with their rather limited detection optics and lower speeds.Schlieren, turbidity, interference, UVvis absorption, and X-rayabsorption detection are published.[105,110] The optical AUCmethod detects the time- and radius-dependent concentrationof the solutes simultaneously with the sedimentation at60060 000 rpm. Thereby, we quantify the amount and the
To complementechnique is desirAny optical microsizes. Standardunknown extent bis a compromise,(shock-freezing thpaves the way to a
The disagreemeexemplified for thdiamonds). Ensemdetect only agglomagnitude, whileques (AUC) agree
Hence, it iscompare measureples, such as hscattering.[7982]
field-flow-fractionaparticle tracking.characterization isof the phenomenasensitive methodof deagglomeratio
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
suspension of nanoparticles in serum spans four orders of magnitude in diameter. A fractionatingcharacterization (interference-AUC, dotted blue line; turbidity-AUC, solid blue line) detects allcolloidal components from proteins to agglomerates. The submicrometer fractions aremissed byDLS (magenta line). b) The average diameter of the nanoparticle fraction (X-axis) dropssignificantly with increasing protein concentration in the suspension medium (Y-axis). CeO2(green triangles), TiO2 B (black squares), and an organically modified ZrO2 (red dots) (redrawnwith permission from ref. [51). Diamonds: inter-method comparison, see text Section 3.3. c)Enlarged sub-10 nm-interval with linear axes in order to facilitate the comparison of the proteinsignal to the expected value of the BSA monomer at 66 kDa. d) The metal oxide and carbonnanomaterials of the present study in DMEM/10% FBS (interference-AUC, this data enters intoTable S1, Phys-bio-chem properties of the test materials).
Adv. Mater. 2010, 22, 127 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinhe
Final page numbers not assignedseveral orders of magnitude, and (ii) due totheir different absorption spectrum andhigher refractive index. It is possible, but ingeneral not mandatory to use X-ray detectionoptics that are inherently only sensitive to theinorganic components with high electrondensities. It is straightforward to take intoaccount the fractal morphology of nanoparti-cle agglomerates[49] and the hydrodynamicsedimentation of fractals has already beenderived by Lin et al.[102]
t the ensemble methods, an in situ imagingable, but is not generally available at present.scopy does not cover the relevant structuralelectron microscopy introduces artifacts ofy drying and vacuum preparation; cryo-TEMrequiring still a number of preparation stepse liquid, then replicating and etching), buthigh-resolution image of aqueous structures.nt between different measurement methods ise case of TiO2 B nanoparticles in FBS (Fig. 3b,ble techniques (DLS, Fraunhoffer diffraction)merates and disagree by many orders ofimaging (cryo-TEM) and fractionating techni-at least within a factor 4.indispensable to complement and criticallyment techniques of different working princi-ydrodynamic/sedimentation, imaging, andSome complementary techniques may betion (FFF-ICP-MS or FFF-MALLS)[111] orMurdock et al.[97] have mentioned thesesues earlier and gave an excellent discussion, but their preference for the simpler, albeit lessof DLS led them to underestimate the amountn in serum. Hassellov et al.[80] published andiameter of each component indepen-dently.[104] At present, AUC is the only methodthat detects all components from the agglom-erates to the dispersed nanoparticles and thesub-10-nm proteins (Fig. 3a): Note thatinterference-AUC retrieves without priorknowledge the correct molar masses andcorrect concentrations of 33mgmL1 ofBSA monomer and dimer in serum(Fig. 3c). In contrast to light scattering, AUCis a fractionating technique by which adistribution of sizes is determined with highresolution. Furthermore, in contrast to trans-mission electron microscopy (TEM)/SEM, theAUC integrates over 10121014 particles inapproximately 0.5mL of a test substance, sothat statistical relevance even of minor frac-tions is high. If low concentrations ofnanoparticles are present in medium contain-ing high concentrations of proteins, nanopar-ticles are easily discerned from sedimentingproteins (i) due to their much higher densitydifference compared to the surrounding med-ium resulting in faster sedimentation bymaterials. a) Theim 9
-
REVIEW
www.advmat.dewww.MaterialsViews.com
the characterization methods must be adapted to the in situproperties of the nanomaterialsinstead of modifying thedispersion procedure until turbidity[62] is low enough to apply
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
10excellent overview of the pros and cons of measurementtechniques; unfortunately they were unaware of AUC.
3.3.3. Surface Conditioning
Once the nanoparticles are incubated in some physiologicalmedium, the adsorption of proteins requires a combination ofbiochemical (for qualitative identification) with physicochem-ical methods (for quantification of binding mechanisms). Bycentrifugation, harvesting and washing, conditioned nanopar-ticles can be isolated from the medium and their adsorbedcorona can be assessed by SDSPAGE (sodium dodecylsulfatepolyacrylamide gel electrophoresis),[41] ideally by 2D gels forfull characterization, performed up to now only with polymernanoparticles with the exception of an early work on ironoxide.[112] A complementing qualitative method is provided bySIMS, even if the necessity to prepare samples underultra-high vacuum is prone to introduce preparation arti-facts.[113] SIMS records ion fragments from the impact of anenergetic primary ion beam; molecular groups on the surfaceare detected with ppm sensitivity.[113] A purely elementalresolution with 10 nm depth integration such as from X-rayphotoelectron spectroscopy (XPS) is of high relevance for thepurity of the pristine nanoparticle surface, but of less value forthe identification of organic matter. XPS bombards the samplewith X-rays that excite characteristic core electrons, and has theadvantage of quantitative information.[113] Quantitative infor-mation can be drawn from fluorescence correlation spectro-scopy (FCS).[43] Unfortunately, FCS is not generally applicableto industrial metal-oxide nanomaterials due to their lack offluorescence and to their quenching of the fluorescence ofadsorbed dyes.
Surface properties like hydrophobicity and surface chargedetermine the capacity and kinetics of aqueous solutiondispersion and this in turn modulates particle ability to adsorbproteins or assay components.[114] The zeta-potential is closelyrelated to the surface charge density, screened by salts, and isexperimentally accessible in many cases. Any changes inzeta-potential should be observed by step-wise addition of buffercomponents, so that charge-screening cannot be misinterpretedas a surface coating by an organic material.[51] While thezeta-potential records the average surface composition, thedynamic change of the surface can be assessed by gel filtrationof conditioned nanoparticles.[38] The longer a protein is desorbedon average, the longer is its elution delay. Finally, the adsorptionenthalpy can be determined by isothermal titration (also known asmicrocalorimetry).[38,39] Microcalorimetry is a particularly sensi-tive method to measure the heat flow of a sample normally underisothermal conditions at room temperature or at 37 8C or higher.Detectable heat flows range from a few to 3000mW. Due to thehigh baseline stability the dynamics of slow reactions can bestudied over minutes up to several days.
3.3.4. Interferences With In Vitro Test Assays
Classic cytotoxicity or genotoxicity assays are often based on thedetection of fluorescence or light absorption of indicatorsand chemical or enzymatic reactions. Undesired particulateinteractions interfere with the test mechanism and 2010 WILEY-VCH Verlag Gwidespread methods such as DLS. Since the most common invitro assays are pH-dependent and may thus be influenced byacidic or basic nanoparticles, acidity/alkalinity should be testedwhen using nanoparticle concentrations which exceed the buffercapacity of biological media.
3.4. Correlation of Biophysical Properties
To summarize Section 3, the nanometer-sized entity exposed tothe organism is not identical to the pristine nanomaterial, butundergoes dynamic changes of both its surface chemistry and itsstate of agglomeration. The protein corona is partially conserved,and partially selective for specific naked metal or metal oxide ororganically functionalized surfaces. Serum tends to decrease thestate of agglomeration, whereas lung lining fluid in general doesnot. One must abandon the attractively simple picture of a nakedinorganic nanoparticle in the human body; we have to take propercare that the in vitro buffers are nearly identical to human bodyfluids, in order to mimic closely the true corona and state ofagglomeration that develops in vivo. Due to the complexity andpolydispersity of a physiological suspension of nanoparticles, acombination of characterization methods with different physicalmeasurement principles (imaging, hydrodynamic, scattering) ismandatory. Similar statements have been stressed most recentlyby the characterization matters initiative.[124] In the presentcontribution, we fulfill the criteria (Table 1) of minimal charac-terization for comparability of studies, and we additionallyprovide advanced characterization data that may help to elucidatethe mechanisms underlying nanoparticle-effect relationships(Supporting Information Table S1).detection.[13,115117] If undetected by insufficient in situ char-acterization, such interferencesmay lead to amisinterpretation ofresults. Especially CNTs and fullerenes[118] show quite unex-pected interactions with the testing systems that induce artifactsignals. It has been reported that endotoxin tests are lesssensitive,[51] essential nutrients are adsorbed and hence cellsstarved.[115,119] Furthermore, carbon nanomaterials have beenshown to interact with indicator substances (methylthiazolyldi-phenyl-tetrazolium bromide (MTT)).[117] When protein concen-tration or protein activity are read outs of cytotoxicity assays theseparameters can be influenced by particles[120] as well as by assaycomponents used for the detection of cellular activity (e.g.,substrates, dyes)[116,117,121] and proteins (lactate dehydrogenase(LDH)) may be adsorbed and hence misleadingly low concentra-tions detected.
Light absorption or fluorescence emission is used todetermine toxicity by most of the in vitro assay systems(Scheme 3). Optical properties of nanoparticles might thereforedirectly interfere with these detection systems. Metallic nano-particles with light-absorptive or light scattering properties likesodium titanate or TiO2 might influence the readout in cellviability assays.[122] Moreover, close proximity of gold nanopar-ticles and fluorescent dyes, have been shown to result in aquenching of fluorescence signal intensity.[123] We believe thatmbH & Co. KGaA, Weinheim Adv. Mater. 2010, 22, 127
Final page numbers not assigned
-
REVIEW
www.MaterialsViews.comwww.advmat.de
oxides, and some carbon nanomaterial) and we will focus onstudies that have been performed with well-characterized
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57materials and multiple test systems. For the complementarynanomaterial classes of metals (including surface modifications)and quantum dots the reader is referred to the excellent review byLewinski et al.[129]
4.2. Review on the In Vitro Toxicity of Engineered
Nanomaterials
Cultured cells exposed to toxic agents can respond with variousmechanisms that differ in the level of cell damage. Cellularreactions range from reversible short term stress responses toirreversible induction of cell death or the long-term malignanttransformation.[130] Engineered nanomaterials have been studiedmainly with established in vitro toxicity assays that analyze majorcellular parameters such as cell viability and response to variousstress factors.
4.2.1. Oxidative Stress
Evidence is accumulating that oxidative stress induced bynanomaterials is a key route by which these nanomaterialsinduce cell damage.[131] Oxidative stress is often detected using afluorimetric 20,70-dichlorofluorescein (DCF)[132,133] or a colori-metric GSH (reduced L-glutathione) assay[134] (Scheme 3). Anapproximate 50% increase in DCF fluorescence has beenobserved after exposure of cultured human skin fibroblasts toanatase TiO2 nanoparticles (UV irradiated).
[135] However, cellscould be protected against TiO2-induced intracellular ROSformation by encapsulation of particles with NaY zeolites(TiO2@NaY). Sayes et al. reported that the structure of titania4. In Vitro Studies With EngineeredNanomaterials
4.1. Critical Aspects of Nanomaterial Test Systems
In comparison to animal models, cytotoxicity testing allows for asimpler, faster and more cost-efficient determination of definedtoxicity endpoints. Classic cytotoxicity assays were established forsoluble chemicals, not for particles (see Section 3.3.4). Since theyare not sufficient at this time to evaluate toxic nanomaterialeffects in cells, multiple assays have to be employed.[125]
Nanomaterial specific properties are crucial determinants ofbiological effects. Recent in vitro screenings have used a variety ofwell-characterized nanomaterials[126] or different variants oftwo kinds of nanoparticles.[127] In most of the earlier studies,however, nanomaterials were used without prior characterizationregarding their composition and physicochemical properties.Physicochemical properties of nanomaterials such as surfacecharge, size, agglomeration state, and shape have been shown toheavily influence the outcome.[31,49,128] These difficulties mightexplain why numerous in vitro studies dealing with nanomaterialtoxicity have provided confounding results with little or nocorrelation to in vivo data.
Here, we will provide an overview of in vitro toxicityexperiments of engineered nanomaterials (especially metal-Adv. Mater. 2010, 22, 127 2010 WILEY-VCH Verlag Gmb
Final page numbers not assignednanoparticles correlates with toxicity (Fig. 4). In their studies withdermal fibroblasts, rutile TiO2 particles produced two orders ofmagnitude less reactive oxygen species than similarly sizedanatase TiO2 particles.
[136] Using P25 TiO2 (anatase/rutile 80:20),Xia et al.[137] observed TiO2 to generate ROS in a cell-free systembut not in murine macrophages (RAW 264.7). On the contrary,SiO2 nanoparticles doped with 1.6wt % iron, cobalt, manganese,and titanium displayed catalytic oxidative effects inside livingcells.[138] Human lung epithelial cells (A549) were exposed tothoroughly characterized particles of the same morphology,comparable size, shape, and degree of agglomeration todetermine the influence of chemical composition and catalyticactivity on ROS formation. These studies clearly showed that thechemical composition of nanoparticles is a most decisive factorinfluencing ROS formation in lung epithelial cells.[138] The role ofparticle size, shape, and composition to induce oxidative stress inprimary mouse embryo fibroblasts was also evaluated for SiO2,ZnO, CNTs, and CB.[139] Although all four nanomaterials inducedsignificant ROS generation and GSH depletion in a dose-dependent manner, the effects were different with ZnO inducingsignificantly more oxidative damage than the other nanomater-ials. Since SiO2 and ZnO had similar crystal shape and particlesize this further confirms that toxicity diversity can be attributedto their chemical composition.[139] Recently, Park and Park[140]
observed both, ROS formation and an increased level of nitricoxide when macrophages (RAW 264.7) were exposed to SiO2nanoparticles and ROS formation in these cells may triggerproinflammatory responses observed in vitro and in vivo. On thecontrary, Diaz et al.[70] did not always find a positive correlationbetween cytotoxicity of SiO2 nanoparticles and ROS formation inhuman monocytes and mouse peritoneal macrophages. In vitrotoxicity screenings with CeO2 nanoparticles revealed a dose-dependent induction of ROS and a decreased level of intracellularGSH in BEAS-2B as well as in A549 human lung epithelialcells.[141,142]
Commercial SWCNTs and MWCNTs (single-walled andmulti-walled CNTs) were found to induce a dose- and time-dependent increase of intracellular ROSs in rat macrophages(NR8383) and human lung epithelial cells (A549) that might berelated to metal traces present in manufactured nanotubes.[143]
4.2.2. Cell Viability
Different endpoints for cell viability have been used innanomaterial toxicity testing. Metabolic activity, for instance,has been widely determined using the colorimetric MTT assaybased on the reduction of a yellow tetrazolium dye (MTT) to apurple formazan in cells bearing intact mitochondria. Recently,however, the suitability of MTT for toxicity evaluation of CNTs hasbeen doubted since SWCNTs have been shown to deplete freeMTT thereby causing false-negative results.[117] Moreover,numerous cytotoxicity studies are based on the detection ofintact lysosomes via neutral red uptake. Neutral red accumulatesin intact lysosomes of viable cells whereas it is excluded fromdead cells. The uptake of neutral red may be detected viafluorescence or absorption measurement. Cellular necrosis isanother endpoint commonly used in cell viability studies. Uponnecrosis, significant amounts of LDH are released from thecytosol. This LDH release can be easily detected using INT (aH & Co. KGaA, Weinheim 11
-
REVIEW
www.advmat.dewww.MaterialsViews.com
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
12yellow tetrazolin salt) as a substrate since LDH catalyzes itsoxidation to a red formazan (Scheme 3). Nanomaterial toxicityleading to apoptosis is commonly assessed with caspase-3 assays.Caspase-3 is activated in the terminal apoptotic cascade bycleavage and this step can be detected by measuring the cleavageof chromogenic or fluorimetric Caspase-3 substrates.
In a comprehensive study, Simon-Deckers et al. determinedthe cytotoxicity of well-characterized metal oxide nanoparticlesand CNTs using different cell viability assays. Studying theresponse of A549 human lung epithelium cells, Simon-Deckerset al. found metal oxide nanoparticles (rutile or anatase TiO2 and
Figure 4. Doseresponse cellular viability of cultured human cells exposedto nano-TiO2 samples for 48 h. While overall the toxicity in culture was low,different types of nano-TiO2 did exhibit different levels of toxicity. Nano-TiO2 anatase particles were the most cytotoxic to human cells in culture,while nano-TiO2 rutile particles were the least cytotoxic, and two mixedanatase/rutile nano-TiO2 samples were in between. Figure reproduced withpermission from Sayes et al. [136].Al2O3) to be less toxic than CNTs. Although all nanoparticles wereefficiently internalized in A549 cells, their cytotoxicity wasgenerally low with a maximum cell death rate of 25% for TiO2(MTT).[144] Since TiO2 and Al2O3 particles were of similar sizeand shape but of different toxicity (with a maximum cell deathrate of 3% for Al2O3 compared to 25% for TiO2) this studyrevealed again that nanoparticle toxicity can be attributed to theirchemical composition.[138,139,144] In line with nanotoxicity datapreviously published by Sayes et al.[136] Simon-Deckers et al.[144]
reported that anatase TiO2 was slightly more toxic than rutileTiO2.
Redox activity in mouse neuroblastom cells has been shown todecrease significantly when the cells were exposed to ZnOwhereas an exposure to other metal oxide nanoparticles such asFe3O4, TiO2, Al2O3, and CrO3 had no measurable effect on thecells.[145] Similarly, cell viability assays (MTT, LDH) using TiO2and metal nanoparticles (Co, Ni), did not reveal any significanttoxic effect on A549 cells.[146] Nanometer-sized and fine-sizedZnO particles were also found to be more cytotoxic to L2 lungepithelial cells than SiO2 particles in LDH assays by Sayes et al.
[64]
However, a comparison of in vivo and in vitro measurementsdemonstrated little correlation.
Lin et al. reported that SiO2 nanoparticles reduce the viability ofhuman bronchoalveolar carcinoma-derived cells in a dose- and
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
2010 WILEY-VCH Verlag Gtime-dependent manner. The cytotoxicity of 15- and 46-nm SiO2nanoparticles was investigated by using crystalline SiO2 as apositive control. Both SiO2 nanoparticles were more cytotoxicthan the bulk material; however, the cytotoxicities of 15- and46-nm SiO2 nanoparticles were not significantly different.
[147]
Moreover, cell viability of A549, endothelial EAHY926 cells, andJ774 monocyte-macrophages in response to SiO2 particles wasfound to be determined by their total mass, number and surfacearea as well as by their concentration.[148] A time- anddose-dependent effect of 20 nm-sized CeO2 particles on cellviability of A549 cells was reported by Lin et al. In their studies,cell viability decreased to 53.9% when a CeO2 concentration of23.3mg mL1 was used.[141]
Although a variety of cell viability studies using carbonnanomaterials have been published so far, no coherent picturehas emerged yet. Davoren et al.[115,149] found a very low directcytotoxicity of SWCNTs in cell viability assays using A549 cells butthe same group reported later that SWCNTs display an indirectcytotoxicity by depleting cell culture medium components.Cytoxicity of MWCNTs was significantly higher in studies bySimon-Deckers et al.[144] who observed amaximum cell death rateof 40% (determined by LDH assays) neither depending on theirlength, nor on their Fe impurities. Similarly, a dose- andtime-dependent decrease in cell viability of human epidermalkeratinocytes was found in studies conducted byMonteiroRiviere and Imnan[116] who used MWCNTs lackingmetal impurities. In contrast, Pulskamp et al.[143] did not observeany acute toxicity on the viability of A549 cells exposed toSWCNTs or MWCNTs but, as mentioned above, observed a dose-and time-dependent increase of ROS formation presumablyassociated with metal traces found in commercial carbonnanotubes. These confounding findings may be due tointerference of the nanomaterials with the employed test systems.Carbon nanomaterials have been reported to distort lightabsorption and fluorescence measurements due to their opticactivity[2] and to interact with dyes and substrates used in classicalcell viability test systems.[116,117] CNTs in particular adsorb andthereby deplete MTT leading to false negative test results. Toavoid this specific interference MTS was suggested as alternativesubstrate for measuring metabolic activity as it did not interactwith CNTs[125] The interaction of MTS with other nanomaterialsis still to be tested. Further studies using MTS in addition tomultiple other cytotoxicity assays have to be performed for anappropriate assessment of carbon nanomaterial toxicity.
4.2.3. Genotoxicity
For a review dedicated entirely to genotoxicity testing ofnanomaterials, the reader is referred to ref. [150]. In thefollowing, we focus on the most important propertyeffectcorrelations for metal oxide nanomaterials. The classic cometassay based on gel electrophoresis and the detection of in vitromammalian chromosomal aberrations are the most commonlyused test systems to assess genotoxicity. Using comet assays,Wang et al.[151] found genotoxic effects of ultrafine TiO2 particleswhen cells were exposed to high particle concentrations (130mgmL1). In contrast, Warheit et al.[12] reported that ultrafine rutlileTiO2 and P25 TiO2 (anatase/rutile 80:20) particles (of approx.140 nm size) did not induce chromosome aberrations normbH & Co. KGaA, Weinheim Adv. Mater. 2010, 22, 127
Final page numbers not assigned
-
REVIEW
www.MaterialsViews.comwww.advmat.de
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56displayed mutagenicity. Recently, Xu et al. [152] demonstrated thatdifferent anatase TiO2 particles (5 and 40 nm in size, respectively)and fullerenes increased the mutation rate in mouse primaryembryo fibroblasts (MEF) in a dose-dependent manner. Toelucidate the mechanisms underlying TiO2 genotoxicity, thisgroup also conducted studies using a nitric oxide synthaseinhibitor and a chemical inhibitor of cyclooxygenase-2 (COX-2).Both nanomaterials lead to the formation of peroxinitrite anionsand induced kilobase pair deletion mutations that could beprotected by antioxidants. Furthermore, DNA damage could bereduced via suppression of COX-2. COX-2 plays an important rolein cellular inflammation and genomic instability, and the particleinduced oxidative stress may activate the COX-2 signalingpathway.[152] In another detailed study of manufactured nano-particles (ZnO, SiO2, TiO2, CB, and SWCNTs), SWCNTs werefound to be more genotoxic than ZnO.[139] Since ROS productioninduced by ZnO was significantly higher than compared to CNTs,it was assumed that DNA damage induced by carbon nanotubescan be attributed to mechanical injury rather than to an oxidativeeffect. Furthermore, Yang et al.[139] provided evidence that DNAnanoparticle genotoxicity might primarily be due to particle shaperather than to chemical composition. Comet assays performedwith SiO2 nanoparticles in two different laboratories usingcultured 3T3-L1 fibroblasts revealed no significant genotoxicitybut showed that in vitro toxicity testing can be quantitativelyreproducible.[153]
Using comet assays, Jacobsen et al.[154] found different carbonnanomaterials (CB and SWCNTs) to induce significant DNAdamage. However, MWCNTs did not show any mutagenic effectsin chromosome aberration studies using Chinese hamsterlung fibroblasts[155] or in bacterial reverse mutation assays.[156]
Colloidal SiO2 nanoparticles of different sizes (30, 80, 400 nm)did not exert any genotoxicity in 3T3-L1 fibroblasts.[153]
4.2.4. Inflammatory Response
To assess inflammation by nanomaterial immunotoxicity, theproduction of inflammatory markers such as the chemokinesInterleukin-8 (IL-8), TNF-a, or IL-6 are usually measured in cellculture supernatants using enzyme-linked immunosorbant assay(ELISA). In rat model systems, the production of the inflamma-tory cytokine MIP-2 (macrophage-inflammatory protein-2)together with that of TNF-a and/or IL-6 are used as cytotoxicityendpoints. Comparing the toxicity of rutile and anatase TiO2 inA549 cells, Sayes et al.[136] demonstrated an overall greater toxicityof TiO2 anatase nanoparticles (Fig. 4). Anatase TiO2 nanoparticlestriggered a dose-dependent release of Il-8 in human dermalfibroblasts (HDF) and A549 cells that was significantly lowerwhen the cells were exposed to rutile TiO2 nanoparticles.
[136]
Ultrafine (P25 rutile/anatase 80:20) but not fine TiO2 particleswere found to elicite Il-8 release in A549 cells indicating asize-dependent effect of immunotoxicity. However, TiO2 ultrafineparticles remained highly aggregated in cell culture as well asinside the cells.[157] Inflammatory properties of TiO2 particlestherefore appear to be driven by their specific surface area.[157]
In a comprehensive study aimed to determine the importanceof surface area and surface reactivity of particles to induceinflammatory responses, Duffin et al. used a variety ofmanufactured particles, such as TiO2, CB, and metal nanoparticlesAdv. Mater. 2010, 22, 127 2010 WILEY-VCH Verlag Gmb
Final page numbers not assignedAs reviewed above, the available reports of nanomaterial in vitrotesting give a broad overview regarding possible toxicologyeffects. However, a valid testing strategy is not available. Moreovermost of the data are not comparable due to a lack of validated testprotocols and a focus on only a few cell lines. Here we reportexemplarily in vitro data from the Nanocare project in vitroscreening strategy highlighting two critical aspects of reliablenanomaterial in vitro testing: the required number of sensitivecell lines and the selection of essential assays.
Six different stable cell lines (Supporting Information Table S3)were exposed to dispersions of two different types of TiO2nanoparticles and were tested regarding the formation of ROS,their metabolic activity, and cell death. The cell lines representedsix different mammalian organs. A549 and RAW264.7, two of themost commonly used lung derived cell lines in in vitro toxicology,represent the first line of exposure to inhaled nanoparticles.While many studies are restricted to these two cell lines, we alsoincorporated three cell types representative of other routes ofexposure. CaCo2 cells stem from a human colon carcinoma andare characteristic for the colon epithelium, while NRK-52E cellshave been widely used as a model for the mammalian kidneyepithelium and have been established from a healthy rat kidney.Furthermore, the skin is represented byHaCaT, a cell line isolatedfrom spontaneously transformed human epidermal keratino-cytes. A sixth cell line, NIH-3T3, represents the fibroblastphenotype and has been cloned from healthy mouse embryos.NIH 3T3 is a widely used well-known model for sensitive in vitrotoxicology testing.(Ni and Co) both for instillation and for treatment of A549 cells.They observed a correlation between particle surface area dose,specific surface activity, and the proinflammatory effects in vivoand in vitro. Their study also demonstrated the utility of in vitroassays for predicting the ability of nanoparticles to causeinflammation in vivo on the basis of their surface area andreactivity.[146]
Recently, Herzog et al.[158] demonstrated that exposure of A549or normal human bronchial cells to SWCNTs did not induceinflammatory responses but can lead to the suppression of avariety of inflammatory mediators including Il-8, Il-6, and MCP-1(monocyte chemotactic protein-1) in vitro. In contrast, chemicallyunmodified MWCNTs caused a dose-dependent Il-8 increase inHEK cells.[159] Since carbon nanomaterials seem to be capable ofadsorbing a variety of substances including cytokines in theculture medium, classical toxicity assays may not be appropriatefor assessing carbon nanoparticle toxicity.[125,149,160]
In this context it is important to note that Veranth et al.[161] haveobserved a significant change of Il-6 response to nanoparticletreatment, either when different cell types were used or whenthe same cell type was grown in different media. Moreover,inflammatory responses to particles seem to be amplified bycontact-dependent interactions between alveolar macrophagesand epithelial cells.[162] Therefore, future studies determininginflammatory effects of nanoparticles have to be conducted usingco-culture systems with defined cell types and media to generatecomparable data.
4.3. Original Results on the Cytotoxicity of TiO2H & Co. KGaA, Weinheim 13
-
REVIEW
www.advmat.dewww.MaterialsViews.com
We focused on standardized cell lines instead of using primarycells to allow for a sound reproducability of results and highthroughput suitability of the test systems used.
Cells were exposed to 0.1, 1, and 10mg cm2 anatase/rutileTiO2 nanoparticles that originate from opposed synthesis routes.TiO2 A is precipitated in a wet chemical process (solgel),while TiO2 B is formed in flame pyrolysis of titanate salts.Concentrations of nanoparticles above 10mg cm2 interferedstrongly with the assay systems which were based on opticaldetection and were therefore neglected.
Dispersions of TiO2 A nanoparticles did neither induce asignificant change in any of the three parameters studied nor inany of the cell lines investigated (Figs. 5a, 6a, and SupportingInformation Fig. S3). TiO2 B, on the other hand, provoked theformation of ROS in all cell lines tested in three or moreindependent experiments (Fig. 5b). The percent increase in ROSformation was dependent on the cell line and the concentration ofTiO2 applied. The mouse fibroblast cell line NIH 3T3 showed thestrongest increase in ROS formation after exposition to 10mgcm2 TiO2 B. The metabolic activity and the incidence of celldeath remained unaffected by TiO2 B in all cell lines tested (Fig. 6Figs. 6b, S3b). The responses described above are summarized inTable 2.
It has been suggested that inhaled particles excert their adverseeffects primarily by triggering an inflammatory response which isin turn mostly elicited by the formation of ROS by the particles
themselves and byROS formation ispotential of a givebeen shown to catherefore trigger a
It has beennanoparticles maynanoparticles trigsimilarly sized a(Fig. 4).[136] Howeparameters testedTiO2 B. Both partiorganic modificatifor the differentnanoparticles.
In contrast to thfollowing sectionan exemplary instypes and test sysnanoparticles. Thindividual degreeB nanoparticles.RAW264.7, behavsensitive to the exour laboratory shodependent on the
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
ixceltly
ed
mdi
14Figure 5. Oxidative stress (expressed as mean DCF fluorescence [%]) in supon exposition to dispersions of (10, 1, and 0.1mg cm2) or stirred(control). Standard deviations are indicated. a) TiO2 A; b) TiO2 B; * significancontrol at the 0.05 level; NA549 32, NRAW 21, Nother 28.
Figure 6. Cell death (measured by LDH activity, expressed as mean INTreducsix different cell lines upon exposition to dispersions (10, 1, and 0.1mg cculture medium (DMEM/10% FBS, control). Standard deviations are inb) TiO2 B. 2010 WILEY-VCH Verlag Ginto the body (e.g., ref. [164), the need for theinvestigation of cell types representing otherorgans than the lung becomes evident.
As described, concentrations of nanopar-ticles above 10mg cm2 interfered with thequantification of the chosen endpoints.Consequently, in vitro methods and especiallythose based on optical detection have to beadapted with respect to interference withnanoparticles and are limited regarding themaximum applicable dose. A comparison toadverse effects of high doses used in inhala-tion studies is therefore impossible. Invarious studies, higher concentrations ofnanoparticles (e.g., P25 and other TiO2particles) than those reported here have beenapplied and found to induce strong effects.However, it remains questionable if theapplication of for instance 100mg cm2 yieldsmeasurement artifacts or reliable results.Based on our findings we argue that theinvestigation of several parameters at lowerparticle concentrations is preferable over theapplication of high doses.
Taken together, our results and recentlypublished data[125] demonstrate that it is
different cell linesl culture mediumdifferent from the
absorption [%]) in2) or stirred cellcated. a) TiO2 A;mbH & Co. KGaA, Weinthe cellular stress response. The observation oftherefore a good indication of the inflammatoryn particle.[163] In the present study, TiO2 B hasuse the formation of ROS in vitro and mayn inflammatory response in vivo.shown that the crystal structure of TiO2influence there in vitro toxicity. Rutile TiO2
gered two orders of magnitude less ROS thannatase TiO2 particles in dermal fibroblastsver, TiO2 A, which did not influence any of thein vitro, consists mostly of anatase TiO2 likecles are in a similar size range which leaves theon detected on TiO2 A as possible explanationbiological activity of the two types of TiO2
e extensive inhalation studies presented in the, this investigation was designed to provideight into the necessity of using different celltems when assessing the in vitro toxicity ofe six cell types presented here displayeds of ROS formation in the presence of TiO2While some cell lines, such as HaCaT anded more robust, NIH-3T3 seem to be moreposition with TiO2 B. Unpublished results ofw that the cell type specific sensitivity is alsonanoparticle applied. In line with this, Veranthet al.[161] haven shown that inflammatoryresponses to TiO2 nanoparticles are influencedby the cell type and culture conditions applied.Furthermore, our results show that cell typesof the other routes of exposure may also beaffected by nanoparticles. As it has been shownthat inhaled nanoparticles may be translocatedheim Adv. Mater. 2010, 22, 127
Final page numbers not assigned
-
REVIEW
www.MaterialsViews.comwww.advmat.de
necessary to use a minimum set of sensitive cell lines andto consider several test systems as nanoparticles may exertparticle type specific adverse effects which will arise in differentendpoints.
4.4. Correlation of In Vitro Toxicity Data
A number of studies conducted with physicochemical character-ization and multiple cytotoxicity assays showed that nanoparticletoxicity can be attributed to size,[33,147] chemical composi-tion,[138,139] surface,[146,165] and structure.[136]
Currently, howadverse effects ofproperties are stillof test results, itproperties of nananoparticles withlevel of toxic effmaterials and valisuspensions shoutoxicity.
An increasingmechanisms underecently and evidderive from oxidaThe ROS generatiwith their potentidamage.[64,136,140,152] Therefore, measurement of oxidative stresspotential can be rcomponent of aassessment. Hownanoparticles ma
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Table 2. Summary of cellular reactions upon the in vitro exposure todispersions of TiO2 nanoparticles.
TiO2 A TiO2 B
Reactive oxygen species None Metabolic activity None None
Cell death None None46
47
48
49
50
51
52
53
54
Scheme 4. Work flow of in vivo inhalation studies for nanomaterials. Aerosfrom the nanomaterials (Supporting Information Fig. S2) and monitored (Tinhalation chamber, typically with head-only exposure. The study designinhalation test developed by the authors [76] is shown on the bottom. X, heato aerosols for 6 h day1 on five consecutive days; R, post-exposure time; H, hiorgans (especially lungs slices, as shown bottom right) including cell proliferatie, examinations of blood and bronchoalveolar lavage fluid (as shown on bot
Adv. Mater. 2010, 22, 127 2010 WILEY-VCH Verlag Gmb
Final page numbers not assignedhardly possible. For an appropriate design of invivo experiments, standardized in vitro testingwill be of considerable value.
5. In Vivo Studies WithEngineered Nanomaterials
5.1. Review of Pulmonary Toxicity Studies
With Engineered Nanomaterials
Adverse health effects of air pollution havebeen recognized in epidemiological studies.Part of the pollution is Particulate Matter,mostly black carbon (see Section 2), and hasbeen linked with cardiovascular effects andpulmonary toxicity.[166168]
Here we focus on pulmonary toxicity ofengineered nanomaterials, and since studydesigns are not standardized yet, we report
ols are generatedable S2) [93] in anof the short-termdnose exposurestology of selectedon and apoptosis;tom left).H & Co. KGaA, Weinheegarded as an important and highly sensitivescreening strategy for nanoparticle toxicityever, intracellular ROS formation induced byy not be predictive of all possible cytotoxiceffects. For SiO2 particles and carbon nano-materials, for instance, a positive correlationbetween cytotoxicity and ROS formation couldnot be found.[70,139] Multiple tests shouldtherefore be used in a comparative mannerto enable an appropriate evaluation of nano-particle cytotoxicity.
Taken together, the presented in vitrotesting strategy may be suitable for predictingthe in vivo effects of nanomaterials. Currently,however, there is little correlation betweenqualitative in vitro data generated in differentlaboratories which might result from a lack ofadapted in vitro test systems. Furthermore,in vitro test systems display a lower complexitythan living organisms and the transfer of dosesapplied in vitro to in vivo exposure scenarios isever, sufficient data enabling to predictnanoparticles based on their physicochemicalmissing. To allow an appropriate interpretationis not sufficient to characterize the intrinsicnoparticles only since the interaction ofphysiological media will also influence the
ects.[51,115] Furthermore, appropriate controldated protocols for the preparation of particleld be used in future studies of nanoparticle
number of studies designed to analyze therlying nanoparticle toxicity has been publishedence is accumulating that many toxic effectstive stress initiated by the formation of ROS.ng capacity of nanoparticles seems to correlateal to induce cellular inflammation and DNAim 15
-
REVIEW
www.advmat.dewww.MaterialsViews.com
ions need to be:
e lung?g to other tissues
e best possiblerespiratory tract
il the sacrifice ofg of test materialsol preparation.and replaces thels in suspensiontes of agglomera-ed to nanomater-to consideration,ation, pharyngealard identification.are organ-specificproliferation, andent of damage to
ies and literatureconclusion that
nano
lung burdens and rapid clearance of par
