Nanotoxicology ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/inan20 Interaction of nanoparticles with endotoxin Importance in nanosafety testing and exploitation for endotoxin binding Maria Mangini, Alessandro Verde, Diana Boraschi, Victor F. Puntes, Paola Italiani & Anna Chiara De Luca To cite this article: Maria Mangini, Alessandro Verde, Diana Boraschi, Victor F. Puntes, Paola Italiani & Anna Chiara De Luca (2021) Interaction of nanoparticles with endotoxin Importance�in nanosafety�testing�and�exploitation�for�endotoxin�binding, Nanotoxicology, 15:4, 558-576, DOI: 10.1080/17435390.2021.1898690 To link to this article: https://doi.org/10.1080/17435390.2021.1898690 © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. Published online: 30 Mar 2021. Submit your article to this journal Article views: 2719 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=inan20 NANOTOXICOLOGY 2021, VOL. 15, NO. 4, 558–576 https://doi.org/10.1080/17435390.2021.1898690 REVIEW ARTICLE Interaction of nanoparticles with endotoxin Importance in nanosafety testing and exploitation for endotoxin binding Maria Manginia , Alessandro Verdea, Diana Boraschib, Victor F. Puntesc,d,e, Paola Italianib and Anna Chiara De Lucaa a Laboratory of Biophotonics and Advanced Microscopy, Institute of Biochemistry and Cell Biology (IBBC), National Research Council (CNR), Napoli, Italy; bLaboratory of Innate Immunity, Inflammation and Immuno-nanosafety, IBBC-CNR, Napoli, Italy; cCatalan Institute of Nanoscience and Nanotechnology (ICN2), Barcelona, Spain; dVall d’Hebron Institut de Recerca (VHIR), Barcelona, Spain; eInstitucio Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain ABSTRACT ARTICLE HISTORY The interaction between engineered nanoparticles and the bacterial lipopolysaccharide, or endo- Received 3 August 2020 toxin, is an event that warrants attention. Endotoxin is one of the most potent stimulators of Revised 8 February 2021 inflammation and immune reactions in human beings, and is a very common contaminant in Accepted 27 February 2021 research labs. In nanotoxicology and nanomedicine, the presence of endotoxin on the nanopar- KEYWORDS ticle surface affects their biological properties leading to misinterpretation of results. This review Nanoparticles; endotoxin; discusses the importance of detecting the endotoxin contamination on nanoparticles, focusing inflammation; LPS on the current method of endotoxin detection and their suitability for nanoparticulate materials. sensors; SERS Conversely, the capacity of nanoparticles to bind endotoxin can be enhanced by functionaliza- tion with endotoxin-capturing molecules, opening the way to the development of novel endo- toxin detection assays. Introduction standardized in vitro tests that measure the produc- In the last decades, engineered nanoparticles (NPs) tion of inflammatory cytokines by human cells have been increasingly explored for their use in exposed to NPs (Hartung and Wendel 1995). In nano- industrial applications as well as in medicine. NPs toxicity testing, both in vitro and in vivo, the sterility are used in food, cosmetic, and textile industries. In of NPs is an important requirement, since NPs can be diagnostic technologies, NPs are used as sensors to easily contaminated by bacteria or bacterial endotox- detect analytes at low concentrations or to image ins (lipopolysaccharide, LPS). NP sterilization elimi- biological processes at cellular level. In medicine, nates most biologically active contaminants, which NPs are used in diagnostic imaging and as drug are confounding elements in the evaluation of NP delivery systems (Bobo et al. 2016). toxicity, but is inefficient in eliminating LPS. LPS is a Due to the increasing NP applications in several heat-stable molecule that can be found ubiquitously fields and their increased presence in the environ- in trace amounts, and is a potent elicitor of inflamma- ment, a thorough evaluation of the NP impact on tion in human cells. Thus, when measuring NP- human and environmental health is required, in induced inflammation, one must be aware that the order to promote the development of a safe-by- LPS presence could lead to misinterpretation of the design nanotechnology. The human health hazard of results, by attributing to NPs inflammatory effects NPs is typically assessed as capacity to activate that are instead caused by LPS. Therefore, only by inflammatory responses, mostly by using evaluating the presence and amount of LPS in NPs, CONTACT Paola Italiani paola.italiani@ibbc.cnr.it Laboratory of Innate Immunity, Inflammation, and Immuno-nanosafety, Institute of Biochemistry and Cell Biology (IBBC), National Research Council (CNR), Napoli, Italy; Anna Chiara De Luca annachiara.deluca@ibbc.cnr.it Laboratory of Biophotonics and Advanced Microscopy, IBBC-CNR, Via Pietro Castellino 111, Napoli, Italy Authors contributed equally. This article has been corrected with minor changes. These changes do not impact the academic content of the article. ß 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by- nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way. NANOTOXICOLOGY 559 we will be able to reliably define the intrinsic inflam- for three-quarters of the outer bacterial membrane, mation-inducing capacity of nanomaterials. and is one of the most potent elicitors of the host In this scenario, it is essential to have valid and immune/inflammatory response aimed at eliminating dependable methods to detect LPS contamination the bacterial infection (Holst et al. 1996). in NP preparations. In fact, measuring LPS on NPs is LPS is an amphiphilic lipoglycan with an overall not simple, since NPs may interfere in many ways negative charge. Structurally, it is composed of the with the currently available LPS detection assays, lipid A responsible for the LPS inflammatory proper- leading to false results (Li et al. 2015). ties, the extracellular polysaccharide O-antigen, and On the other hand, we can exploit and enhance the interconnecting core oligosaccharide (Rietschel the capacity of NPs to bind LPS for developing new et al. 1996) (Figure 1). The molecular weight of LPS nanotechnological tools. LPS-binding NPs are being may hugely vary between 10 and 1000 kDa, also used as sensors for setting up novel LPS detection considering that LPS in aqueous solution tends to assays of improved sensitivity, and could be devel- form micelles of variable dimensions depending on oped for capturing LPS in contaminated solutions. the microenvironmental conditions. Therefore, LPS Additionally, LPS-decorated NPs can be used as is conventionally measured in endotoxin units (EU), immunostimulatory devices in medicine, to induce which reflect its biological activity. The degree of therapeutically beneficial enhancement of defensive activity can vary in LPS from different bacterial ori- immunity/inflammation in diseases such as cancer. gins and depending on a number of other factors. This review describes both the traditional and the A rough estimate for the LPS of the bacterium more innovative methods for endotoxin detection Escherichia coli is that 1 EU corresponds to about on NPs, highlighting their advantages and disadvan- 100 pg LPS, which is the amount present on 105 tages and their suitability in testing NPs. We will also bacteria, with each bacterial cell possessing around examine the use of NPs as LPS sensors/capture tools 2106 LPS molecules. for developing new LPS detection assays. The host immune system recognizes LPS as a sig- nal of ‘danger’ and immediately reacts by starting an inflammatory process. Human beings are particu- LPS structure and signaling larly sensitive to LPS, and recognition/reaction can LPS is the major structural component of the Gram- occur upon exposure to minute amounts of LPS negative bacteria cell wall, accounting approximately while, for example, mice are much more resistant. Figure 1. LPS on the cell wall of gram-negative bacteria. Graphic representation of a Gram-negative bacterium cellular membrane together with a scheme of the LPS structure. LPS: lipopolysaccharide; Gln: glucosamine; P: phosphate group. 560 M. MANGINI ET AL. The major mechanism of LPS-induced inflammatory instruments for human use as 0.5 EU/kg, while, if activation is based on the LBP/CD14/TLR4 receptor the product comes in contact with the cerebro- complex, but LPS can trigger inflammation also spinal fluid, the limit is 0.02 EU/kg (Chapter 85, through other recognition systems (Yang et al. Unites States Pharmacopoeia 2007). 1998; Ogata et al. 2000), including TLR2-mediated The approved LPS detection assays, which meas- recognition that is amplified by CD14 and LPB ure the LPS biological activity, are the old rabbit (Yang et al. 1998) and, mainly in neuronal cells, by pyrogen test (RPT) and the more recent limulus membrane perturbation-mediated activation of amoebocyte lysate (LAL) assay. The RPT records the Transient Receptor Potential (TRP) channels that rise in temperature in rabbits after intravenous cause an intracellular increase of calcium and injection of the test solution, i.e. it measures the sodium levels triggering neurogenic inflammation pyrogenic effect of the sample (Hort and Penfold (Boonen et al. 2018). 1912; Seibert 1925). This test is now only applied, Once LPS-bearing bacteria are internalized, cyto- when required, to validate the results of the LAL solic LPS can also activate cells by directly binding assay. The RPT has limited sensitivity, high cost, caspase-4/11, causing the induction of pyroptosis long readout time, and implies the use of animals. and indirectly causing the cleavage/activation of From a regulatory point of view, the RPT advantage, the inflammatory cytokine IL-1b by activation of the as opposed to the LAL assay, is that it can detect NLRP3 inflammasome (Casson et al. 2013; Vigano all types of pyrogens, i.e. substances that cause et al. 2015). fever/inflammation, and is not limited to LPS. The LAL test is based on the use of lysates from amoebocytes (blood cells) of the horseshoe crab Biological effects of endotoxin Limulus polyphemus. Exposure to LPS induces a The main LPS-signal transducer, TLR4, is expressed defensive reaction in the cells, which includes the by several cell types, starting from innate immune LPS-dependent activation of Factor C, an enzyme cells as macrophages and monocytes that are the that initiates a coagulation/clotting cascade aiming first line of defense of our body against infections. In to inhibit the infection (Levin and Bang 1964a, all TLR4-bearing cells, TLR4 participates in responses 1964b). Clotting can also be induced by b-(1,3)-D- associated to stress, inflammation and disease. Upon glucan, a molecule present in fungal pathogens, LPS-induced cell activation, inflammatory molecules which activates another clotting enzyme, Factor G, are produced and secreted. At the local level, the which initiates the same defensive coagulation cas- release of these factors causes vasodilation and cade (Morita et al. 1981). The LAL assay exploits the recruitment of circulating leukocytes (Pober and high reactivity of the Factor C present in the lysate Cotran 1990), which participate to the inflammatory of Limulus amoebocytes for an accurate quantitative defensive reaction by producing other inflammatory detection of the LPS biological activity. Several LAL mediators and microbicidal agents. A balanced assays have been developed and are available com- release of inflammation-related mediators leads to a mercially, from the original clotting and turbidimet- controlled response against invading microorgan- ric assays to the more sensitive assays that use isms. On the contrary, an uncontrolled and massive synthetic chromogenic or otherwise tagged sub- inflammatory reaction can cause severe symptoms as strates. In these assays, LPS-activated Factor C indu- in the case of septic shock, in which the host reac- ces the release of a colored/fluorescent/luminescent tion against bacteria or their toxins (including LPS) molecule from the synthetic substrate, which can causes endothelial damage, hypotension, diffuse be precisely measured. The sensitivity of the LAL intravascular coagulation, organ failure and eventual tests ranges from 0.005 to 50 EU/mL depending on death (Danner et al. 1991). the sample source, bacterial species, strain, and detection method. To avoid the false positives due to the possible presence of b-glucans, many LAL Current methods for LPS detection assay kits available on the market include reagents Regulatory authorities define the maximum toler- that specifically block the Factor G-depend- ated level of endotoxins in drugs or surgical ent reaction. NANOTOXICOLOGY 561 More recently, the regulatory authorities have ratio that increases their reactivity (Jones and included other two LPS detection methods in their Grainger 2009). list of recommended assays, the recombinant Factor LPS was found as a contaminant in different NP C assay and the monocyte activation test (MAT), preparations, such as Au, polystyrene, silica, TiO2, with the general recommendation of using more latex, and AlO3 NPs (Peula-Garcia et al. 2002; Kucki, than one method for a reliable evaluation of the Cavelius, and Kraegeloh 2014; Li et al. 2015; Bianchi endotoxin contamination. et al. 2017; Li et al. 2017). As expected, the pres- The recombinant Factor C assay is a fluorogenic ence of LPS can strongly affect the biological prop- test that uses the recombinant horseshoe crab erties of NPs, as it was shown for AuNPs (Li et al. enzyme, thereby eliminating the need for animals. 2017). In that study, endotoxin-contaminated AuNPs The test is specific for LPS, as b-glucans are unable induced an inflammatory response in human pri- to activate Factor C. The MAT assay is based on the mary monocytes in vitro, assessed as expression capacity of human monocytes coming in contact and production of the inflammatory cytokine IL-1b, with LPS to produce inflammatory cytokines, which while endotoxin-free particles were completely can be detected and quantitated by an enzyme- inactive (Li et al. 2017). Another study reported that linked immunosorbent assay (ELISA). Similar to the LPS-contaminated TiO2NPs could activate inflamma- RPT, the advantage of MAT is that monocytes react tory signaling pathways and increase apoptosis in to all types of pyrogens, not only LPS, providing a murine macrophages (Bianchi et al. 2017). Thus, if functional evaluation (the extent by which contami- unrecognized, the presence of LPS on NPs may be nants can induce an inflammatory reaction in responsible for significant inflammatory/toxic effects humans). The drawback is that the assay does not that would be mistakenly attributed to NPs. This is distinguish between the inflammatory effect of con- true both in vivo and in vitro, and in particular in taminants and those intrinsic to the molecules/ assays that use human cells, which are highly sensi- materials under testing. tive to LPS. For this reason, it is important to exclude LPS contamination during the evaluation of the NP inflammatory/toxic effects. Endotoxin contamination of NPs The procedures for eliminating endotoxin from Endotoxin persists in the environment also in the biomedical materials and devices are defined as absence of living Gram-negative bacteria. In the lab, ‘depyrogenation’ since the original aim is to elimin- it can be found in all chemicals and glassware, and ate exogenous ‘pyrogens,’ i.e. agents able to cause its high thermostability makes it resistant to the fever (LPS being the principal one). Several proto- majority of the sterilization methods usually applied cols of depyrogenation are available, but these pro- (e.g. standard autoclaving). LPS can bind to cationic cedures could be destructive for NPs. One of the surfaces, thanks to its negative charge, and to most used methods is incineration, which consists hydrophobic materials through its lipidic domain. of treating samples with high temperatures for While other inflammation-inducing microbial agents given times, e.g. dry heat at 180  C for 3 h or 250  C can be easily removed with sterilization procedures, for 30 min (Weller et al. 2015). It is known that this the thermostability and the capacity to bind to treatment and others (ethylene oxide treatment, both hydrophilic and hydrophobic materials, formaldehyde treatment, and autoclaving for 5 h at together with its high potency towards human cells, 121  C with a pressure of 20 PSI) are not suitable for make LPS the most important pyrogen to which depyrogenating many NP types, because they can attention should be paid during the synthesis and change the NP physicochemical properties leading the manipulation of pharmaceutic preparations. LPS to aggregation and chemical transformations, can easily bind cationic NPs through electrostatic changes in morphology and size, and consequently interactions, such as Coulomb and van der Waals in their optical properties. Such physicochemical forces (Darkow et al. 1999), and to NPs with a lipo- changes can affect the NP impact on cells by philic surface by hydrophobic interaction with lipid increasing their cytotoxicity, as observed in many A (Bromberg et al. 2010). LPS contamination of NPs studies (Krug 2014). Other methods imply endo- is further facilitated by the high NP surface:volume toxin removal on immobilized LPS-binding 562 M. MANGINI ET AL. molecules (e.g. polymyxin B, PMB). The method can endotoxin negative charge causing aggregation and be used for free endotoxin despite a significant decreasing activity). The interference is usually iden- sample loss, but it is hardly applicable to NPs tified by assessing the recovery of a known amount because LPS bound to PMB does not usually detach of endotoxin spiked into an NP sample. If the spike from NPs, which are therefore retained as well by recovery is less than 50% or more than 200%, the PMB (unpublished results). Therefore, since LPS elim- result of the LAL assay is considered not valid ination requires methods hardly applicable to NPs, because of significant interference (Dobrovolskaia the best way to obtain LPS-free NPs is by synthesiz- et al. 2014). In such cases, NPs can be diluted until ing them in endotoxin-free conditions (Vallhov interference disappears. If extensive dilution is et al. 2006). required for eliminating interference, there is the risk that the NP concentration becomes too low for Assessing endotoxin contamination in NP any kind of realistic LPS testing. Thus, a different preparations assay should be used. Choosing the right type of LAL test to assess The regulatory guidelines about the maximum tol- endotoxin contamination in NP samples depends erated LPS levels do not apply to NPs not intended on the NP characteristics. Factors to take into for medical use. Thus, many NPs used in industrial account include NP turbidity and adsorption spectra applications and consumer products may be conta- (especially at wavelengths 405 and 540 nm, which minated with biologically active levels of LPS. When are the most used in commercially available LAL NPs are subjected to safety testing, such contamin- assays). Generally, gel-clot and turbidimetric assays ation may significantly bias the results. This calls for an accurate evaluation of the LPS contamination of are not suitable for many types of NPs each NP undergoing safety testing. Notably, the LPS (Dobrovolskaia et al. 2010; Smulders et al. 2012; contamination of NPs may need to be lower than Kucki, Cavelius, and Kraegeloh 2014), while the use the limits imposed for in vivo use in humans, of chromogenic and the fluorescent assays must be because the testing conditions (e.g. a limited num- assessed case by case. ber of human cells in vitro in a small volume) make Lipid-based NPs (e.g. liposomes) can pose prob- possible to detect a significant reaction to minute lems in different LAL assays. The gel-clot assay is LPS concentrations (few pg/mL). not suitable because the lipids interfere with the The currently available assays are not generally clot formation (Piluso and Martinez 1999). applicable for assessing the LPS contamination of Moreover, most liposome preparations are milky, NPs, mainly because of the particulate nature of and interfere with turbidity assays. In most cases, NPs and their physicochemical characteristics. The the chromogenic LAL assays (with a final readout at ISO 29701:2010 regulation provides guidelines to 405 nm) work satisfactorily for detecting endotoxin use the LAL tests for monitoring the LPS presence in liposomes after appropriate dilution. in NP preparations. This document generically men- Both classic and modified chromogenic LAL tions that NPs can interfere with the LAL tests. assays can be adapted to endotoxin detection in Indeed, NPs can affect the activity of LAL enzymes metal-oxide NPs (SiO2, TiO2, Fe3O4) (Smulders et al. (e.g. by adsorbing the LAL test components on their 2012; Mano, Kanehira, and Taniguchi 2013; Li et al. surface) resulting in a decrease or increase of the 2015). Likewise, the two commercial chromogenic assay sensitivity. Also, depending on the NP charac- tests turned out to be the most suitable for endo- teristics, NPs can interfere with the final readout of toxin detection in metal NPs (Au and Ag), provided the tests (Smulders et al. 2012; Li et al. 2015; Neun an appropriate NP dilution is used. For instance, Li and Dobrovolskaia 2019) and generate false positive et al. (2015) demonstrated that, with classical or false negative results. For NPs, as for every other chromogenic assays, 14 nm AgNPs can be tested at drug, the use of the LAL assays has limitations in concentrations below 5.4 lg/mL without significant terms of pH (the enzymatic reaction occurs at pH interference, whereas 10 nm AuNPs could be tested between 6.0 and 8.0) and presence and concentra- up to 64 lg/mL. On the other hand, the substantial tion of divalent cations (which can neutralize the quenching of fluorescence by metal NPs prevents NANOTOXICOLOGY 563 the use of all the fluorescence-based assays (includ- TLR4 (the receptor for LPS) and carrying a reporter ing the one with recombinant Factor C). gene whose activation can be quantitatively meas- In the LAL assays, two controls must be per- ured (Smulders et al. 2012). After incubation for 6 formed. An ‘NPs-only’ control will exclude false pos- to 24 h, the concentration of inflammatory cytokines itives due to intrinsic NP absorbance, luminescence, released is measured using cytokine-specific ELISA or fluorescence. Moreover, ‘NPs-plus-readout mol- assays or, in the case of reporter cells, by measuring ecule’ and ‘NPs-plus-assay-reagents-plus-LPS’ con- the fluorescence generated by the products of the trols will identify potential interference with activated reporter gene. In the case of NPs, before endotoxin determination at any level within the the ELISA it is recommended to centrifuge the assay (assay reagents, assay readouts, etc.) (Li supernatant to remove the NPs that could interfere et al. 2015). with the test (Oostingh et al. 2011). Before running It has been shown that the same NP preparation cell-based tests, it is advisable to test the compati- tested simultaneously with different LAL assays can bility of NPs with cell culture medium, since some yield different results (Dobrovolskaia et al. 2010; Li kinds of NPs can corrode releasing toxic cations, or et al. 2015). Thus, the use of at least two LAL for- can aggregate and change their size and morph- mats is recommended in testing NP preparations ology (Oostingh et al. 2011). Disadvantages of these and, when the results differ by more than 25%, functional tests include, in addition to their lack of these should be confirmed with functional tests specificity for LPS, their variability in quantitative (Dobrovolskaia et al. 2010). Carbon-based NPs can terms (e.g. donor-to-donor variations in reactivity), interfere with both gel-clot, turbidimetric and and the fact that they cannot be used with NPs car- chromogenic assays by direct activation of the clot- rying cytotoxic drugs (Dobrovolskaia et al. 2014). ting enzyme present in LAL. As in the case of other For these reasons, cell-based assays should be NPs, it is possible to avoid interference upon appro- intended as complementary to the LAL assays, or priate NP dilution. In a recent study, it was possible when LAL assays are not applicable. For example, to measure LPS contamination in preparations of instead of the unsuitable LAL assay, Mukherjee carbon nanotubes at concentrations below 10 ng/ et al. (2016) used a MAT-like functional assay based mL and 300 ng/mL in the gel-clotting and the on the production of TNF-a by human primary mac- chromogenic assays, respectively (Yang et al. 2017). rophages to detect the presence of LPS in their car- The functional tests include the in vivo induction bon-based NP preparations, exploiting the high of fever in the rabbit (RPT) and the in vitro induc- sensitivity of macrophages to LPS and their ability tion of inflammatory reaction in human leukocytes to produce inflammatory cytokines in response to (MAT and similar assays). The major characteristic of it. Importantly, to overcome at least in part the lack functional assays is the lack of specificity for LPS, as of specificity of the MAT assay, the authors per- all pyrogenic and inflammation-inducing agents are formed the assay in parallel in the presence of PMB, active in such assays. The RPT can be successfully which binds and inhibits LPS with high selectivity. applied to detect endotoxins in NPs but, as already PMB is an amphipathic polypeptide that binds LPS mentioned, it has low sensitivity and a number of leading to the death of Gram-negative bacteria by other drawbacks, including the fact that the rabbit destabilizing their outer membrane. The use of PMB response to pyrogens is not identical to the human in cell cultures should however be considered care- reaction (Hartung and Wendel 1995). The in vitro fully, because of its non-specific activation and tox- assays based on human cells have successfully icity effects. replaced the RPT in pyrogen detection. These tests The methods described above are summarized in are not affected by components that interfere with Table 1, and compared for their ability to detect LAL, and more closely reflect the human immune LPS on NPs. response to the tested sample. The assays consist As mentioned above, the best way to avoid LPS of incubating the NP sample with human blood contamination is working in endotoxin-free condi- cells (whole blood, blood mononuclear cells or iso- tions, and this is more important when NPs are lated monocytes), a human macrophage-like cell intended for drug delivery. In this case, the endo- line, or reporter cells engineered to express human toxin-free conditions are essential not only during Table 1. LPS detection methods useful for NP testing. 564 Measured NPs (Chem-Phys Multiple Required Technique Instrumentation parameter characteristics) Sample preparation Advantages Disadvantages measurements expertise Detection limits LAL assay Plate reader LPS activity All Water suspension, sample Endotoxin-specific NPs may interfere with Yes Yes 0.5 pg/mL (absorbance, (Factor volume about 50 lL, the final readout fluorescence) C activation) NP concentration range 0.1 ng–1 mg/mL depending on the interference controls Rabbit Termo-scanner, LPS activity All Suspension in physiological Detection of all Not specific for No Yes 30–500 pg/mL M. MANGINI ET AL. Pyrogen Test thermometer (pyrogenic saline solution, sample pyrogenic agents endotoxin, long effect volume 1–5 mL/kg readout time, use of in rabbits) intravenously, animals, animal NP concentration range variability depending on NP toxicity (non-lethal concentrations required) MAT assay Plate LPS activity All Suspension in physiological Detection of all Not specific for No Yes 3–10 pg/mL reader (cytokine medium, sample volume pyrogenic agents endotoxin, blood (absorbance) production in 0.05–0.5 mL (depending on donor variability human cells) the culture conditions), NP concentration range depending on NP toxicity (non-toxic concentrations required) DLS Photon NP size, All, Water suspension, liquid Quick response, small Transparent sample Yes No 500 ng/mL correlation superficial dimension range measurements, sample volumes required, polydisperse spectroscope charge, 0.1 nm–1 mm NP concentration range 0.1–500 required (10–100 mL) samples can give agglomeration mono-dispersed mg/mL, sample volume larger radii, >10 mL not specific TEM Transmission NP size, shape, Metallic/ Suspension in buffer solution, High spatial Structural artefacts due Yes Yes Not electron composition semiconductor dry measurements, sample resolution (<2nm) to staining or quantitative microscope (no staining volume >2 mL, dehydration, not required), NP concentration range quantitative, long Organic (negative 100–500 mg/mL, drop sample preparation, staining required), deposition and specific not specific, NP procedures: drying, advanced dimension <1mm negative staining expertise required SERS Confocal Enhanced Raman Metallic, Suspension in buffer solution, Label-free quantitative Interpretation of SERS Yes Yes 50–100 fg/mL Raman spectra of LPS NP dimension and dry or liquid measurements, and qualitative spectra, microscope shape Sample volume <1 mL sensing of LPS SERS enhancement properly selected NP concentration depending on attached to NPs, dependent on NP dimension/shape (10–500 identification LPS size and shape, mg/mL) domains interacting calibration required, with NPs, small endotoxin and NP sample volumes (<10 type-dependent mL), high specificity sensitivity, advanced expertise required NANOTOXICOLOGY 565 NP synthesis but also during the NP loading with Figure 2(A) shows typical hydrodynamic diameter drugs. In the latter case, a potential LPS contamin- and f-potential distribution curves, obtained with a ation could be masked by the NP shell, which will DLS instrument, for 50 nm AuNPs before and after make it undetectable by any LPS detection assay. incubation with LPS. The experiment was performed This could generate the possibility that, upon NP on 1 mL of the NP solution (3.5  105 NPs/mL) incu- uptake by cells and cargo release in the cytoplasm, bated with 5 mg of LPS. By comparing the size of LPS could also be released and activate intracellular both bare and LPS-coated NPs, it was possible to inflammatory pathways. observe an increased size for the LPS-coated NPs of about 20 ± 5%, together with a decrease of the NP f-potential, due to the presence of an LPS layer on Function-independent methods for detecting LPS contamination on NPs the NP surface. The main advantage of the DLS ana- lysis is the ability to detect hundreds of NPs in a The limitations of the traditional LPS detection short time, returning statistically significant results. assays, based on the assessment of its biological The drawback is that the NP hydrodynamic diam- activity, encouraged the development of alternative eter measured by DLS is not the same as the NP LPS detection assays suitable for NPs. An interesting diameter. The conversion from hydrodynamic diam- method was recently proposed, based on the Ultra eter to NP diameter is not direct and depends on High Performance Liquid Chromatography coupled several factors (including issues of random coil vs. with mass spectrometry to detect the LPS presence hard sphere) that could lead to possible errors in on NPs (Giannakou et al. 2019). This method is LPS quantification. Additionally, DLS can measure compatible with some NPs (e.g. TiO2 and CeO2 NPs) different hydrodynamic diameters if the sample is and has a sensitivity between 0.03 and 0.7 EU/mL. polydisperse, implying that a good estimation of Its applicability to different NP types needs case-to- the LPS corona around NPs can be obtained only if case validation. the LPS corona is homogeneous around all NPs in Methods that are traditionally used to character- the sample. Although any kind of NPs can be used, ize NPs have been also applied to detect LPS on this type of assay can be performed only on solu- NPs and may form the basis for the development of tions with a volume >10 mL, and with monodis- new methods for the quantitative assessment of persed NPs with a regular shape. As many other LPS contamination. methods that measure the presence of organic mat- ter at the NP surface, DLS is not specific for LPS. Dynamic light scattering The main characteristics of the DLS are summarized in Table 1. Dynamic Light Scattering (DLS) is widely used to analyze the size and superficial charges (f-potential) of NPs (including nanogold, proteins, colloids, etc.). Transmission electron microscopy Particle size distribution can be determined by Transmission Electron Microscopy (TEM) is usually measuring the random changes in the scattered employed for the morphological characterization of light due to Brownian motion – the random ther- NPs. Based on the interaction of a high energy elec- mal motion – of small particle in solution. By further tron beam with the atoms of the target sample, analyzing the NP mobility across the electric field TEM allows to image NPs with a diameter even generated by two electrodes (electrophoretic light lower than 1 nm (Winey et al. 2014). The LPS pres- scattering), it is possible to estimate the NP superfi- ence on 50 nm AuNPs surface incubated with 5 mg cial charge and, therefore, the NP stability of LPS can be imaged as a light gray layer (about (Bhattacharjee 2016). DLS allows for measuring NP 5 nm-thickness) around the black NP surface, as size between 0.3 nm to 6 mm with a minimum vol- shown in Figure 2(B). Therefore, the imaging ume 10-100 mL of sample suspension, with an accur- approach can be used to visualize and detect the acy of about ±2% and a precision of about ±1%. formation of the LPS corona with a high spatial Li et al. (2017) used DLS for detecting the pres- resolution, to visualize the uniformity of the LPS dis- ence of LPS on AuNPs (size range 40–50 nm). The tribution at different concentrations and to measure 566 M. MANGINI ET AL. Figure 2. Summary of non-conventional techniques for detection and quantitative analysis of LPS on NPs. (A) Dynamic light scat- tering data showing hydrodynamic diameter (d) and f-potential (Z) distributions of 50 nm AuNPs before and after incubation with LPS. The mean values acquired before and after the LPS incubation are reported on the graph. The presence of LPS can be detected by measuring a shift into the NP size and f-potential distributions. (B) TEM images showing 50 nm AuNPs before and after incubation with LPS. From the TEM image acquired after the LPS incubation, it is possible to observe the LPS corona of about 4–5 nm (grey halo) around the NPs (black sphere). (C) SERS spectrum of 50 nm AuNPs, either bare or incubated with LPS, acquired with a Raman microscope equipped with a 785 nm wavelength laser (Manago et al. 2018). The spectral bands allow the biochemical identification of the LPS components, while the signal intensity can be used for LPS quantitation. (D) UV-Vis spectrum showing a decrease of the SPR peak at 525 nm and an increase of the peak at 650 nm related to the agglomeration state of CSH- AuNPs upon addition of increasing LPS concentrations. Inner: Visual colorimetric change of the CSH–AuNPs upon addition of LPS (Sun et al. 2012). The measurement of (A650/A525)1/2 values in the presence of different LPS concentrations allows quantitative analysis. Reprinted with permission. (E) Electrochemical impedance assay showing the changes of the electrode/electrolyte inter- face (AuNPs) by applying an oscillating potential as a function of frequency (Nyquist plot). When the concentration of endotoxin is high, the semi-circles of Nyquist plots increase, due to the rearrangement of LPS molecules on AuNPs and the inherent variation of their electric properties. Inner: schematic representation of the assembly of the LPS-detecting aptasensor (Su et al. 2013). Reprinted with permission. (F) Fluorescence intensity increase as a function of the LPS concentrations for the fluorescence-based sensor proposed by Gao et al. (2017). Reprinted with permission. NANOTOXICOLOGY 567 the particle sizes using a really small sample volume present at low doses (few fg/mL), with a good (>2 mL). Despite its powerful imaging capabilities, reproducibility (variation about 5–8% for the this method is excessively time-consuming, requir- reported concentrations). ing long sample preparation and complex image Furthermore, the spectral analysis allows for the analysis, and it does not provide quantitative and identification of the LPS chemical components. specific information. The NP images should be ana- Indeed, the strong Raman bands at 973, 1256 and lysed one by one, resulting in size measuring of 1610–1640 cm1 can be associated to the lipid A, poor accuracy. In addition, the selection of the while 535 cm1 band and bands between 1300 and image focus and, therefore, the NP dimension 1400 cm1 can be assigned to LPS core or O-antigen strongly depends on the operator. The main charac- (Figure 3). A complete peak assignment, together teristics of the TEM analysis are summarized in with a comparison between the SERS and Raman Table 1. spectra of LPS, is reported in Figure 3(B) (Barrett 1981; Osorio-Roman et al. 2010; Czamara et al. 2015). Surface-enhanced Raman spectroscopy The technique does not require long sample prepar- ation or functionalization procedures, and it can be Surface-enhanced Raman spectroscopy (SERS) can used for the label-free direct biochemical identifica- be exploited for detection and quantitation of LPS tion the LPS domains, as well as for the presence of adsorbed on the surface of metallic NPs. SERS is a different types of endotoxins (high specificity) even surface-sensitive technique where Raman scattering in turbid media (De Luca et al. 2014). On the other signal by molecules adsorbed on metal surfaces hand, SERS enhancement depends on the NP nature (such as metallic NPs), is enhanced up to several (only metallic NPs can be used), geometrical parame- orders of magnitude thanks to the excitation of ters (size, shape and concentration of the NPs) and localized surface plasmons (Campion and plasmonic mode field distribution (Manago et al. Kambhampati 1998; De Luca et al. 2014; Manago 2018; Quero et al. 2018). Therefore, after the calibra- et al. 2018; Quero et al. 2018). Raman bands are tion of the SERS signal of the metallic NPs, the assay 10–100x narrower than fluorescence bands, and can provide quantitative information on the LPS therefore SERS offers a major advantage over fluor- detection. The main characteristics of the SERS tech- escence for multiplex detection and optical stability. nique are summarized in Table 1. Due to its sensitivity and advantages over other detection techniques, SERS has received much NPs for detection and quantitation of LPS attention as an ultrasensitive analytical technique, particularly for biomedical and environmental appli- Not only in the case of endotoxin detection on NPs cations (Manago et al. 2018). A number of reports but more in general, the LAL assays have a number have used SERS for the direct detection and analysis of limitations. Being based in the detection of a of unlabeled biological molecules and microorgan- specific biological activity (activation of Factor C), isms, including the composition of bacteria cell every condition that interferes with such activity membranes (Zhou et al. 2015). will impact on the assay reliability. One of these We recently measured the SERS signal of LPS limitations is the so called ‘low endotoxin recovery,’ from E. coli O55:B5 adsorbed on 50 nm AuNPs in i.e. the lack of Factor C activation due to the forma- the same experimental conditions reported previ- tion of macromolecular complexes that mask the ously (Li et al. 2017). Compared with the Raman endotoxin Factor C-activating moieties (Chen and spectrum of the LPS powder, the AuNP-LPS SERS Vinther 2013; Schwarz et al. 2017). It has been dem- spectrum showed a remarkable increase of the sig- onstrated that the ‘masked endotoxin’ can still acti- nal intensity, corresponding to about 6–7 orders of vate an inflammatory response in human primary magnitude (Figure 2(C)) (Manago et al. 2018; Quero monocytes inducing the expression of inflammatory et al. 2018). Therefore, due to its sensitivity, the cytokines, indicating that the biological activity of SERS approach could allow for detecting and quan- endotoxin in different biological assay can be very titating the LPS on AuNPs with a minimum volume various (Schwarz et al. 2017). Thus, a significant <1 mL of sample suspension, even when LPS is effort is being devoted to developing new more 568 M. MANGINI ET AL. Figure 3. Intensity comparison between LPS Raman spectrum and LPS-NPs SERS spectrum. (A) SERS signal (NPs þ LPS, upper line) measured using sample concentration conditions of 5 mg LPS on 2 mg (1.4  109 particles) of 50 nm AuNPs. Conditions of acquisi- tion: kex ¼ 785 nm; integration time ¼ 1 s; laser power ¼ 0.1 mW (Manago et al. 2018; Quero et al. 2018). Reference Raman spec- trum of the LPS powder (LPS only, lower line). Condition of acquisition: kex ¼ 785 nm; integration time ¼ 60 s; laser power ¼ 10 mW. The spectra are an average over 30 acquisitions. (B) Assignment of the observed SERS and Raman bands based on the lit- erature (Barrett 1981; Osorio-Roman et al. 2010; Czamara et al. 2015). LPS component band assignment related to Lipid A and O- antigen/Core are indicated on spectra peaks. d: deformation vibrations; m: stretching vibrations. reliable methods for endotoxin determination that absorbance peak, called Surface Plasmon Resonance could replace LAL and other biological assays, and (SPR) peak, in the visible range of light, provides infor- provide a more specific and precise detection. mation about the size and the aggregation state of Some methods based on antibodies, aptamers, or the NPs (Sun et al. 2012). If NPs are agglomerated, recognition ligands have been proposed, but they they will show a shift of their characteristic SPR, which still do not provide sufficient sensitivity and specifi- depends on NP size, shape, and surface functionaliza- city (Su and Ding 2015). tion (Begum et al. 2018). Therefore, colorimetric assays Due to their small size, large surface area, and based on the use of AuNPs can be employed for LPS optical and electrochemical properties, NPs have detection. A key parameter to consider in this kind of been also used in the development of new meth- assay is the capacity of AuNPs to bind LPS. Several ods for LPS detection. Here, new promising NP- NP functionalization protocols have been developed based assays are discussed. The main characteristics for improving the assay sensitivity, using molecules of all the new LPS detection assays described here- that specifically recognize LPS such as monoclonal after are summarized in Table 2. antibodies, aptamers, macromolecules or peptides (Yu et al. 2011; Lei et al. 2016; Zhu et al. 2019). Three examples of AuNP-based colorimetric assays are NP colorimetric assays (optical assays) described below. AuNPs are largely used in colorimetric assays thanks In the pioneering case reported by Sun et al. to their optical properties and ability to change their (2012), 35 nm AuNPs were functionalized with cyste- color when conjugation or agglomeration occurs. amine (CSH), providing a positive surface charge Indeed, UV-Vis spectroscopic observation of the AuNP able to interact with the negative charge of LPS. NANOTOXICOLOGY 569 Upon the addition of increasing concentrations of 29 ag–50 fg/mL Detection 3 ng–1 lg/mL the endotoxin, CSH-AuNPs agglomerated and the limits 10 pg/mL solution changed their color from wine-red to pur- ple and finally blue (Figure 2(D)). UV-Vis spectra of measurements expertise Required CSH-AuNPs alone showed an SPR peak at 525 nm Yes Yes No but, when increasing doses of LPS were added into the solution, CSH-AuNPs started to agglomerate Multiple causing the decrease of the 525 nm peak and the No No No increase of a new peak at 650 nm (Figure 2(D)). The limit of detection (LOD) is reported to be 3.3 ng/mL fluorophore, indirect method, and sensibility dependent on based tests, indirect method, functionalization procedures, required, specificity depends functionalization procedures indirect method, calibration analyte solution, advanced procedures, reproducibility specificity depends on the NPs can interfere with the specificity depends on the Limited sensitivity for UV-Vis- Functionalization procedures, and is evaluated by a linear relationship obtained charge and purity of the Electrode stability, complex on the functionalization between peak amplitude ratio (A650/A525)1/2 and Disadvantages calibration required, expertise required LPS concentration. functionalization Another rapid assay based on UV-Vis measure- procedures ments used AuNPs functionalized with LPS-binding aptamers (Kalita et al. 2015). AuNPs were modified with ethanolamine aptamer (G-probe) or E. coli Suspension in buffer solution, Quick response, high O111:B4 LPS aptamer (S-probe). With S-probe Quantitative, fast quantitative and sample volume dependent on quantitative and qualitative, long Advantages AuNPs in the absence of LPS or with G-probe Quick response, Suspension in buffer solution, Quick response, linear range qualitative specificity, AuNPs, particles were stabilized by the aptamers, and the solution appeared red. In the presence of LPS, the interaction of LPS with the S-probe LPS-fluorophore functionalization non-interfering with NP concentration dependent on NP concentration range 50–100 water, liquid measurements, sample volume about 2 mL, NP concentration range 0.5 aptamer (not with G-probe) caused the aggregation Suspension in buffer, serum, sample volume >10 mL, Sample preparation of AuNPs, inducing a red-to-colorless color change liquid measurements, dimension, shape and liquid measurements, LPS-NP functionalization mg–1 mg/mL, LPS-NP electrochemical cell in solution. In this way the assay showed a double their dimensions, functionalization function, it allowed identifying and quantifing the dimensions, LPS in one biosensing system with a LOD of 1 mg/ mg /mL, mL. Indeed, S-probe aptamers only recognized E. coli O111:B4 and not other LPS types. material selected as Kalita et al. (2015) proposed a colorimetric assay NP dimension and NPs (Chem-Phys characteristics) NP dimension range properly selected Metallic/magnetic, employing AuNPs functionalized with PMB com- fluorophore 10–20 nm bined with the use CMOS sensor to detect the shape Metallic, Metallic, AuNP optical properties with high sensitivity. Serum and water samples spiked with LPS from E. coli agglomeration status O55:B5 were first drop-casted onto a silanized glass NP conjugation and Material impedance parameter substrate and then tagged with PMB-AuNPs. The LPS complexes Measured Number of NP- binding was visualized using a silver halide reduc- Table 2. NP-based LPS detection assays tion process that leads to an increase of the particle size and the deposition of a silver layer where LPS is present. Signal intensities (measured with a spectrophotometer spectrometer (EIS) Instrumentation CMOS sensor on images captured with an optical Flow cytometer, spectroscope Electrochemical Electrochemical fluorescence microscope) increased monotonically with rising impedance LPS concentrations, showing a LOD of 5 and 50 fg/ UV-Vis mL for water and serum, respectively. This device showed an excellent recognition of O55:B5 LPS, and Fluorescence it can be used to detect other LPS serogroups Colorimetric Technique assays assays assays thanks to the high binding affinity of PMB to any kind of LPS. 570 M. MANGINI ET AL. These three examples demonstrate that colori- the potentiostat, and converted by a software into metric assays are reliable methods to detect LPS an impedance value with a real and imaginary com- based on its interaction with AuNPs and their ponent (impedance spectra). Real and imaginary optical properties, with several advantages. These impedance components are generally plotted tests are cost-effective, quick and direct, as they against one another in Nyquist plots (Figure 2(E)), allow for a visual detection of the LPS presence. which have to be interpreted properly to deduce However, they require special instrumentation able analyte molecule resistances and, then, concentra- to translate a band intensity and spectral position tions (Randviir and Banks 2013). This type of elec- to a given LPS concentration. Furthermore, the sen- trochemical sensors has been used to measure the sitivity is variable, depending on the LPS affinity for LPS concentration with the aid of NPs. Three exam- the ligand on NPs, and the system does not allow ples of LPS electrochemical sensors with different for detecting and quantitating LPS concentrations immobilization procedures are described below. that cause little NP aggregation. The main disadvan- An electrochemical biosensor for LPS, based on tages are related to limited detection sensitivity of the use of CramoLL and cysteine-coated AuNPs transmitted light (i.e. low concentrations of dis- (AuNPCys) was reported (Oliveira et al. 2011). persed AuNPs cannot be detected and do not sig- CramoLL is a lectin isolated from Cratylia mollis nificantly affect optical density) by instruments such seeds that can bind to the carbohydrate moieties of as plate readers, optical microscopes and UV-Vis LPS. The sensor is formed by a gold electrode spectrometers (Table 2). coated with the poly(vinylchloride-co-vinyl acetate- Consequently, despite the use of efficient ligands co-maleic acid) (PVM) on which AuNPCys are such as aptamers, the colorimetric methods show absorbed and CramoLL added to them. The biosen- poor sensitivity, with a LOD of 1 mg/mL (Zhu et al. sor was tested with LPS from E. coli, Serratia marces- 2019). This sensitivity slightly improved in the case cens, Salmonella enterica, and Klebsiella pneumoniae, of the cysteamine, but the method described can- by measuring the charge transfer resistance not detect LPS concentrations lower that 3.3 ng/mL, changes upon LPS binding. Depending on the LPS roughly corresponding to 33 EU/mL, i.e. still a very composition, the biosensor showed different charge high biological activity. The highest sensitivity transfer responses. The biosensor showed a good achieved with image analysis (5–50 fg/mL, corre- response for 200 mg/mL of LPS of each tested spe- sponding to 0.05–0.5 mEU/mL) is excellent for cies, meaning that its sensitivity for LPS recognition detecting bioactive LPS levels, but it implies a com- is unsatisfactory. In addition, the PVM–AuNpCys plex elaboration of images that makes it difficult to CramoLL system cannot discriminate between LPS apply in quantitative experiments. and other molecules with glycosyl residues. Aptamer-assisted electrochemical biosensors for NP electrochemical assays LPS detection, which exploit aptamers able to bind Electrochemical sensors measure the conductivity LPS, showed improved sensitivity and selectivity (Su changes of an electrode due to the selective et al. 2013). Su et al. proposed the use of AuNPs to absorption of the analyte on the conductive surfa- immobilize a thiol-modified aptamer that recognizes ces. Since the electrical signal is concentration- LPS. A scheme of the sensor is reported in Figure dependent, it is used to determine the concentra- 2(E). The biosensor sensitivity was evaluated by tion of the analyte. In Electrochemical Impedance measuring the Nyquist plot at different endotoxin Spectroscopy (EIS), a sinusoidal voltage is applied concentrations. The semi-circle of the Nyquist plot by a potentiostat across an electrode cell, and the increases linearly with the LPS concentration due to analyte molecules are placed on the electrode sur- the attraction between the aptamers and their tar- face (Randviir and Banks 2013). The current flows gets (Figure 2(E)). The presence of AuNPs dramatic- through the electrochemical device and gains infor- ally increased the conducting capacity of the mation about its impedance properties, being con- electrode, ensuring an excellent linear response in a ductivity of the system affected by the presence of concentration range from 0.1 pg to 1 ng/mL of LPS, analyte molecules. The current is then recorded by with a LOD of about 0.05–0.1 pg/mL. NANOTOXICOLOGY 571 Recently, the performance of the aptamer An interesting method is based on the use of assisted electrochemical biosensors has been further magnetic NPs coated with an LPS-capture aptamer improved (Posha and Sandhyarani 2020). A novel and an LPS-detection aptamer functionalized with a nano-probe for LPS detection was designed, based fluorophore (Zuo et al. 2014). Flow cytometry was on an AuNP-loaded, water-soluble layered molyb- used to measure and quantify the green fluores- denum disulfide-polyacrylic acid (Au/MoS2-PAA) cence emission (at 520 nm) of the endotoxin-conju- nanocomposite. Basically, AuNPs loaded over MoS2- gated sandwich complex when detecting endotoxin PPA substrate enable the immobilization of the with magnetic aptasensor in the range from 10 fg thiol-terminated LPS binding aptamers (LBA) on to 1 mg/mL. This assay requires a small amount of MoS2 nanosheets, with strong affinity (nanomolar LPS solution (only 2 ll), shows a good reproducibil- range). The presence of the substrate and NPs, ity (5.5% variations after several regenerations), and amplify the current response of the electrode, fast response (detection can be performed in less allowing the detection of the LPS with an excellent than a minute). linear response in a concentration range from 100 Another LPS fluorescence sensor was developed ag/mL to 100 pg/mL of LPS. Compared with the based on the use of liposomes and AuNPs (Gao existing approaches, this approch showed a broad et al. 2017) (Figure 2(F)). The authors synthesized detection range, a low limit of detection (LOD of 29 liposomes functionalized with Rhodamine B and ag/mL), and high reproducibility. This sensor was exploited the ability of both LPS and AuNPs to also tested to determine the LPS concentration in bind liposomes. Their strategy consists in incubat- clinical-grade insulin, showing results within an ing the analyzed sample first with liposomes and error of 2%. then with AuNPs. If LPS is present in the analyzed Despite excellent sensitivity, the use of EIS needs sample it will insert into the liposome phospho- advanced expertise, and therefore it is not access- lipid bilayer, due to its amphiphilic nature, com- ible by all users (Table 2). Real and imaginary pete and limit the binding of AuNPs to liposomes impedance plots need to be interpreted properly to (Figure 2(F)). AuNPs will not be able to completely deduce solution resistances, charge transfer resist- cover liposomes because their phospholipid bilayer ance, and other parameters. Moreover, these sen- strengthened by LPS can limit diffusion and pene- sors can be difficult to fabricate, and can be made tration of hydrophobic substances. This will avoid only with NPs that are able to conduct properly the the fluorescence quenching of the fluorophore- current (i.e. metallic NPs) and that allow measuring labelled liposomes caused by the binding of impedance changes of the electrochemical sensor. AuNPs. Conversely, if the LPS is not present AuNPs As for the colorimetric assays, the LPS detection is will be free to bind to fluorescent liposomes, and indirect and the immobilization of the LPS on NPs the fluorescence will be effectively quenched. The has direct implications for the development and sensor was successfully used to detect LPS dis- properties of the proposed biosensors. Finally, elec- solved in cell culture medium and showed a LOD trode stability, charge and purity of the analyte of 6.5 ng/mL. solution, may affect the reproducibility and selectiv- NP-based sensors that exploit fluorescence, ity of the biosensor. although providing fast response, are indirect meth- ods because of the difficulties in directly tagging LPS with a fluorophore. The selection of the fluoro- NP fluorescence assays phore or fluorescent dye is a key element of the NPs can also be used to set up new LPS detection assay, and particular attention should be paid to strategies based on fluorescence. In these type of selecting NPs that do not interfere with the fluores- biosensors different NP types (magnetic, metallic) cent dyes during measurements (Table 2). The fluor- can be employed, the selectivity strongly depend- escence assays require a series of functionalization ing on the LPS immobilization strategy, and a fluo- and sample preparation steps, as well as an accur- rophore is used to identify and quantify the ate choice of NPs, which must be individually tested presence of LPS. Two examples are reported below. for validating the sensor performance. On the other 572 M. MANGINI ET AL. hand, it requires a small amount of solution, the Conclusions and future perspective reproducibility is high, the analysis is compatible The early identification of bioactive LPS in NP prep- with the use of solvents (PBS, DMEM etc.) generally arations would avoid misinterpretation of toxicity used in cell culture experiments, and could be eas- data and inappropriate attribution to NPs of ily adapted to 96-well plates for high-throughput LPS-dependent inflammatory/toxic effects, thereby and automated analysis. promoting a knowledge-based development of Figure 4. Decision tree. (A) In order to detect LPS contamination in NP samples, the first choice is generally the LAL assay. The LAL assay format is chosen depending on NP characteristics and behaviour in the assay conditions. Here we reported a synthetic guide on how to choose the most suitable LAL assay. Note that in general the suitability of a specific LAL assay for NP samples should be verified empirically case by case. If NPs interfere with LAL assay components or final readout, diluting NPs until the interference disappear can be an efficient solution. When NPs are not suitable for any LAL assay format, other functional assays (RPT, MAT or WBA) are used, which however are not specific for LPS and cannot be used with cytotoxic NPs. (B) If NPs are conta- minated by LPS, the contamination can be quantified by LAL (if you are interested to determine the LPS biological activity) or by DLS or SERS, to determine the number of LPS molecules, the choice of technique depending on the NP physico-chemical charac- teristcs. The presence of the LPS on NPs can be visualized by TEM. Note that in this figure the NP-based LPS detection assays have not been included since they are intended for different types of samples and cannot be applied to NP-containing samples. Information on such assays is reported in the paragraph ‘NPs for Detection and Quantitation of LPS’ and in the Table 2. LAL: Limulus amoebocyte lysate; exc/em: excitation/emission; RPT: rabbit pyrogen test; MAT: monocyte activation test; WBA: whole blood assay; TEM: transmission electron microscopy: DLS: dynamic light scattering; SERS: surface-enhanced Raman spectroscopy. NANOTOXICOLOGY 573 safe-by-design nanotechnology. Since NPs can inter- grant [21420] of the Associazione Italiana Ricerca sul Cancro fere with the current LPS detection methods and (AIRC), the project NeON [ARS01_00769] of the Program PRIN, and the projects CIRO- Campania Infrastructure for each NP type/size/shape can have different effects, Research in Oncology and PLATT of the Programma special attention should be paid to selecting the Operativo Regionale of Regione Campania [POR FERS appropriate assay and running the appropriate con- 2014/2020]. trols, in order to obtain reliable results. In addition to adapting the currently available LPS detection assays, a number of new methods that can be used in sup- ORCID port. We propose here a ‘decision tree’ (Figure 4) Maria Mangini http://orcid.org/0000-0002-4524-6317 that can help researchers in selecting the most suit- able detection assays for their NP preparations. Future perspectives go in two directions. One, References partly addressed above, consists in the exploitation Barrett, T. W. 1981. “Laser Raman Spectra of Mono-, Oligo- of the NP characteristics for developing more pre- and Polysaccharides in Solution.” Spectrochimica Acta Part cise and robust LPS detection assays, which are a A: Molecular Spectroscopy 37: 7. general need. The current attempts still have limita- Begum, R., Z. H. Farooqi, K. Naseem, F. Ali, M. Batool, J. Xiao, and A. Irfan. 2018. “Applications of UV/Vis Spectroscopy in tions either in sensitivity or affordability, feasibility, Characterization and Catalytic Activity of Noble Metal reproducibility, specificity. It is also notable that Nanoparticles Fabricated in Responsive Polymer Microgels: these assays do not measure the LPS biological A Review.” Critical Reviews in Analytical Chemistry 48 (6): activity on human cells, which not necessarily corre- 503–516. doi:10.1080/10408347.2018.1451299. lates with LPS amount. On the other hand, these Bhattacharjee, S. 2016. “DLS and Zeta Potential – What They systems of NP-based LPS recognition and binding Are and What They Are Not?” Journal of Controlled Release 235: 337–351. doi:10.1016/j.jconrel.2016.06.017. can be the basis for other promising applications Bianchi, M. G., M. Allegri, M. Chiu, A. L. Costa, M. Blosi, S. both in the area of decontamination (from solutions Ortelli, O. Bussolati, and E. Bergamaschi. 2017. and samples of clinical/experimental use to environ- “Lipopolysaccharide Adsorbed to the Bio-Corona of TiO2 mental applications). Lastly, new nanomedical appli- Nanoparticles Powerfully Activates Selected Pro- cations can be envisaged for LPS-carrying NPs to Inflammatory Transduction Pathways.” Frontiers in achieve a controlled localized inflammatory/immune Immunology 8: 866 doi:10.3389/fimmu.2017.00866. Bobo, D., K. J. Robinson, J. Islam, K. J. Thurecht, and S. R. response with high efficacy and limited side effects. Corrie. 2016. “Nanoparticle-Based Medicines: A Review of This would be particularly useful in the implementa- FDA-Approved Materials and Clinical Trials to Date.” tion of novel immunotherapeutic approaches Pharmaceutical Research 33 (10): 2373–2387. doi:10.1007/ to cancer, but it could also be of great value in s11095-016-1958-5. obtaining limited activation of innate/inflammatory Boonen, B.,. Y. A. Alpizar, V. M. Meseguer, and K. Talavera. reactions in immunostimulatory and vaccination 2018. “TRP Channels as Sensors of Bacterial Endotoxins.” Toxins 10 (8): 326. doi:10.3390/toxins10080326. strategies. Bromberg, Lev, Emily P. Chang, Carmen Alvarez-Lorenzo, Beatriz Magarinos, Angel Concheiro, and T. Alan Hatton. Acknowledgements 2010. “Binding of Functionalized Paramagnetic Nanoparticles to Bacterial Lipopolysaccharides and DNA.” The authors aknowledge DiSTABiF and Universita della Langmuir 26 (11): 8829–8835. doi:10.1021/la904589p. Campania Luigi Vanvitelli that coordinates the PhD program Campion, A., and P. Kambhampati. 1998. “Surface-Enhanced of AV. Raman Scattering.” Chemical Society Reviews 27 (4): 241–250. doi:10.1039/a827241z. Casson, C. N., A. M. Copenhaver, E. E. Zwack, H. T. Nguyen, Disclosure statement T. Strowig, B. Javdan, W. P. Bradley, et al. 2013. “Caspase- The authors declare no conflict of interest. 11 Activation in Response to Bacterial Secretion Systems That Access the Host Cytosol.” PLOS Pathogens 9 (6): e1003400 doi:10.1371/journal.ppat.1003400. Funding Chen J, A. V. 2013. “Low Endotoxin Recovery in Common The authors were supported by the EU H2020 grants Biologics Products.” Paper presented at Proceedings of ENDONANO [GA 812661] and PANDORA [GA 671881], the IG the PDA Annual Meeting, Orlando, FL, USA, 3. 574 M. MANGINI ET AL. Czamara, K., K. Majzner, M. Z. Pacia, K. Kochan, A. Kaczor, English) 53 (46): 12304–12319. doi:10.1002/anie. and M. Baranska. 2015. “Raman Spectroscopy of Lipids: A 201403367. Review.” Journal of Raman Spectroscopy 46 (1): 4–20. doi: Kucki, M., C. Cavelius, and A. Kraegeloh. 2014. “Interference 10.1002/jrs.4607. of Silica Nanoparticles with the Traditional Limulus Danner, R. L., R. J. Elin, J. M. Hosseini, R. A. Wesley, J. M. Amebocyte Lysate Gel Clot Assay.” Innate Immunity 20 (3): Reilly, and J. E. Parillo. 1991. “Endotoxemia in Human 327–336. doi:10.1177/1753425913492833. Septic Shock.” Chest 99 (1): 169–175. doi:10.1378/chest.99. Lei, C., Z. Qiao, Y. Fu, and Y. Li. 2016. “Colorimetric Detection 1.169. of Lipopolysaccharides Based on a Lipopolysaccharide- Darkow, R., T. Groth, W. Albrecht, K. Lutzow, and D. Paul. Binding Peptide and AuNPs.” Analytical Methods 8 (45): 1999. “Functionalized Nanoparticles for Endotoxin Binding 8079–8083. doi:10.1039/C6AY02371A. in Aqueous Solutions.” Biomaterials 20 (14): 1277–1283. Levin, J., and F. B. Bang. 1964a. “A Description of Cellular doi:10.1016/S0142-9612(99)00022-8. Coagulation in the Limulus.” Bulletin of the Johns Hopkins De Luca, A. C., P. Reader-Harris, M. Mazilu, S. Mariggio, D. Hospital 115: 337–345. Corda, and A. Di Falco. 2014. “Reproducible Surface- Levin, J., and F. B. Bang. 1964b. “The Role of Endotoxin in Enhanced Raman Quantification of Biomarkers in the Extracellular Coagulation of Limulus Blood.” Bulletin of Multicomponent Mixtures.” ACS Nano 8 (3): 2575–2583. the Johns Hopkins Hospital 115: 265–274. doi:10.1021/nn406200y. Li, Y., P. Italiani, E. Casals, N. Tran, V. F. Puntes, and D. Dobrovolskaia, M. A., B. W. Neun, J. D. Clogston, H. Ding, J. Boraschi. 2015. “Optimising the Use of Commercial LAL Ljubimova, and S. E. Mcneil. 2010. “Ambiguities in Assays for the Analysis of Endotoxin Contamination in Applying Traditional Limulus Amebocyte Lysate Tests to Metal Colloids and Metal Oxide Nanoparticles.” Nanotoxicology 9 (4): 462–473. doi:10.3109/17435390.2014. Quantify Endotoxin in Nanoparticle Formulations.” 948090. Nanomedicine 5 (4): 555–562. doi:10.2217/nnm.10.29. Li, Y., Z. Shi, I. Radauer-Preiml, A. Andosch, E. Casals, U. Dobrovolskaia, M. A., B. W. Neun, J. D. Clogston, J. H. Luetz-Meindl, M. Cobaleda, et al. 2017. “Bacterial Grossman, and S. E. Mcneil. 2014. “Choice of Method for Endotoxin (Lipopolysaccharide) Binds to the Surface of Endotoxin Detection Depends on Nanoformulation.” Gold Nanoparticles, Interferes with Biocorona Formation Nanomedicine 9 (12): 1847–1856. doi:10.2217/nnm.13.157. and Induces Human Monocyte Inflammatory Activation.” Gao, J., Z. Li, O. Zhang, C. Wu, and Y. Zhao. 2017. “Tunable Nanotoxicology 11 (9-10): 1157–1175., doi:10.1080/ Accessibility of Dye-Doped Liposomes Towards Gold 17435390.2017.1401142. Nanoparticles for Fluorescence Sensing of Manago, S., N. Migliaccio, M. Terracciano, M. Napolitano, N. Lipopolysaccharide.” The Analyst 142 (7): 1084–1090. doi: M. Martucci, L. De Stefano, I. Rendina, A. C. De Luca, A. 10.1039/c7an00019g. Lamberti, and I. Rea. 2018. “Internalization Kinetics and Giannakou, C., K. Aimonen, L. V. Bloois, J. Catalan, R. E. Cytoplasmic Localization of Functionalized Diatomite Geertsma, E. R. Gremmer, W. H. De Jong, et al. 2019. Nanoparticles in Cancer Cells by Raman Imaging.” Journal “Sensitive Method for Endotoxin Determination in of Biophotonics 11 (4): e201700207 doi:10.1002/jbio. Nanomedicinal Product Samples.” Nanomedicine 14 (10): 201700207. 1231–1246., doi:10.2217/nnm-2018-0339. Mano, S. S., K. Kanehira, and A. Taniguchi. 2013. Hartung, T., and A. Wendel. 1995. “Detection of Pyrogens “Comparison of Cellular Uptake and Inflammatory Using Human Whole Blood.” ALTEX 12 (2): 70–75. Response via Toll-like Receptor 4 to Lipopolysaccharide Holst, O., A. J. Ulmer, H. Brade, H. D. Flad, and E. T. Rietschel. and Titanium Dioxide nanoparticles.” International Journal 1996. “Biochemistry and Cell Biology of Bacterial of Molecular Sciences 14 (7): 13154–13170. doi:10.3390/ Endotoxins.” FEMS Immunology and Medical Microbiology ijms140713154. 16 (2): 83–104. doi:10.1111/j.1574-695X.1996.tb00126.x. Morita, T., S. Tanaka, T. Nakamura, and S. Iwanaga. 1981. “A Hort, E. C., and W. J. Penfold. 1912. “Microorganisms and New (1 ! 3)-b-D-Glucan-Mediated Coagulation Pathway Their Relation to Fever: Preliminary Communication.” The Found in Limulus Amebocytes.” FEBS Letters 129 (2): Journal of Hygiene 12 (3): 361–390. doi:10.1017/ 318–321. doi:10.1016/0014-5793(81)80192-5. s0022172400005052. Mukherjee, S. P., N. Lozano, M. Kucki, A. E. Del Rio-Castillo, L. Jones, C. F., and D. W. Grainger. 2009. “In Vitro Assessments Newman, E. Vazquez, K. Kostarelos, P. Wick, and B. Fadeel. of Nanomaterial Toxicity.” Advanced Drug Delivery Reviews 2016. “Detection of Endotoxin Contamination of 61 (6): 438–456. doi:10.1016/j.addr.2009.03.005. Graphene Based Materials Using the TNF-a Expression Kalita, P., A. Dasgupta, V. Sritharan, and S. Gupta. 2015. Test and Guidelines for Endotoxin-Free Graphene Oxide “Nanoparticle-Drug Bioconjugate as Dual Functional Production.” PLOS One 11 (11): e0166816 doi:10.1371/jour- Affinity Ligand for Rapid Point-of-Care Detection of nal.pone.0166816. Endotoxin in Water and Serum.” Analytical Chemistry 87 Neun, B. W., and M. A. Dobrovolskaia. 2019. “Detection of (21): 11007–11012. doi:10.1021/acs.analchem.5b02957. Endotoxin in Nano-Formulations Using Limulus Krug, H. F. 2014. “Nanosafety Research – Are We on the Amoebocyte Lysate (LAL) Assays.” Journal of Visualized Right Track??” Angewandte Chemie (International ed. in Experiments NANOTOXICOLOGY 575 Ogata, H., I. Su, K. Miyake, Y. Nagai, S. Akashi, I. Schwarz, H., J. Gornicec, T. Neuper, M. A. Parigiani, M. Mecklenbrauker, K. Rajewsky, M. Kimoto, and A. Wallner, A. Duschl, and J. Horejs-Hoeck. 2017. “Biological Tarakhovsky. 2000. “The Toll-like Receptor Protein RP105 Activity of Masked Endotoxin.” Scientific Reports 7: 44750. Regulates Lipopolysaccharide Signaling in B cells.” Journal doi:10.1038/srep44750. of Experimental Medicine 192 (1): 23–29. doi:10.1084/jem. Seibert, F. B. 1925. “The Cause of Many Febrile Reactions fol- 192.1.23. lowing Intravenous Injections. I.” American Journal of Oliveira, M. D., C. A. Andrade, M. T. Correia, L. C. Coelho, P. Physiology-Legacy Content 71 (3): 621–651. doi:10.1152/ R. Singh, and X. Zeng. 2011. “Impedimetric Biosensor ajplegacy.1925.71.3.621. Based on Self-Assembled Hybrid Cystein-Gold Smulders, S.,. J. P. Kaiser, S. Zuin, K. L. Van Landuyt, L. Nanoparticles and CramoLL Lectin for Bacterial Golanski, J. Vanoirbeek, P. Wick, and P. H. Hoet. 2012. Lipopolysaccharide Recognition.” Journal of Colloid and “Contamination of Nanoparticles by Endotoxin: Evaluation Interface Science 362 (1): 194–201. doi:10.1016/j.jcis.2011. of Different Test Methods.” Particle and Fibre Toxicology 9: 06.042. 41 doi:10.1186/1743-8977-9-41. Oostingh, G. J., E. Casals, P. Italiani, R. Colognato, R. Su, W., M. Cho, J.-D. Nam, W.-S. Choe, and Y. Lee. 2013. Stritzinger, J. Ponti, T. Pfaller, et al. 2011. “Problems and “Aptamer-Assisted Gold Nanoparticles/PEDOT Platform for Challenges in the Development and Validation of Human Ultrasensitive Detection of LPS.” Electroanalysis 25 (2): Cell-Based Assays to Determine Nanoparticle-Induced 380–386. doi:10.1002/elan.201200453. Immunomodulatory Effects.” Particle and Fibre Toxicology Su, W., and X. Ding. 2015. “Methods of Endotoxin 8 (1): 8 doi:10.1186/1743-8977-8-8. Detection.” Journal of Laboratory Automation 20 (4): Osorio-Roman, I. O., R. F. Aroca, J. Astudillo, B. Matsuhiro, C. 354–364. doi:10.1177/2211068215572136. Vasquez, and J. M. Perez. 2010. “Characterization of Sun, J., J. Ge, W. Liu, X. Wang, Z. Fan, W. Zhao, H. Zhang, P. Bacteria Using Its O-antigen with surface-enhanced Wang, and S.-T. Lee. 2012. “A Facile Assay for Direct Raman scattering.” The Analyst 135 (8): 1997–2001. doi:10. Colorimetric Visualization of Lipopolysaccharides at Low 1039/c0an00061b. Nanomolar Level.” Nano Research 5 (7): 486–493. doi:10. Peula-Garcia, J. M., J. A. Molina-Bolivar, J. Velasco, A. Rojas, 1007/s12274-012-0234-1. and F. Galisteo-Gonzalez. 2002. “Interaction of Bacterial Unites States Pharmacopoeia. 2007. Chapter 85 Bacterial Endotoxine (Lipopolysaccharide) with Latex Particles: Endotoxin Test. application to Latex Agglutination Immunoassays.” Journal Vallhov, H., J. Qin, S. M. Johansson, N. Ahlborg, M. A. of Colloid and Interface Science 245 (2): 230–236. doi:10. Muhammed, A. Scheynius, and S. Gabrielsson. 2006. “The 1006/jcis.2001.7958. Importance of an Endotoxin-Free Environment during the Piluso, L. G., and M. Y. Martinez. 1999. “Resolving Liposomal Production of Nanoparticles Used in Medical Inhibition of Quantitative LAL Methods.” PDA Journal of Applications.” Nano Letters 6 (8): 1682–1686. doi:10.1021/ Pharmaceutical Science and Technology 53 (5): 260–263. nl060860z. Pober, J. S., and R. S. Cotran. 1990. “Cytokines and Vigano, E., C. E. Diamond, R. Spreafico, A. Balachander, R. M. Endothelial Cell Biology.” Physiological Reviews 70 (2): Sobota, and A. Mortellaro. 2015. “Human Caspase-4 and 427–451. doi:10.1152/physrev.1990.70.2.427. Posha, B., and N. Sandhyarani. 2020. “Highly Sensitive Caspase-5 Regulate the One-Step Non-Canonical Endotoxin Detection Using a Gold Nanoparticle Loaded Inflammasome Activation in Monocytes.” Nature Layered Molybdenum Disulfide-Polyacrylic Acid Communications 6: 8761 doi:10.1038/ncomms9761. Nanocomposite.” The Analyst 145 (11): 3939–3947. doi:10. Weller, T., R. Kragseth, R. Dullinger, H. Illum, and A. Perry. 1039/D0AN00567C. 2015. “Dry-Heat Depyrogenation Ovens for Pharmaceutical Quero, G., G. Zito, S. Manago, F. Galeotti, M. Pisco, A. C. De Compounding Facilities.” International Journal of Luca, and A. Cusano. 2018. “Nanosphere Lithography on Pharmaceutical Compounding 19: 182–192. Fiber: Towards Engineered Lab-On-Fiber SERS Optrodes.” Winey, M., J. B. Meehl, E. T. O’toole, and T. H. Giddings. Jr., Sensors 18 (3): 680. doi:10.3390/s18030680. 2014. “Conventional Transmission Electron microscopy.” Randviir, E. P., and C. E. Banks. 2013. “Electrochemical Molecular Biology of the Cell 25 (3): 319–323. doi:10.1091/ Impedance Spectroscopy: An Overview of Bioanalytical mbc.E12-12-0863. Applications.” Analytical Methods 5 (5): 1098–1115. doi:10. Yang, R. B., M. R. Mark, A. Gray, A. Huang, M. H. Xie, M. 1039/c3ay26476a. Zhang, A. Goddard, W. I. Wood, A. L. Gurney, and P. J. Rietschel, E. T., H. Brade, O. Holst, L. Brade, S. Muller- Godowski. 1998. “Toll-like Receptor-2 Mediates Loennies, U. Mamat, U. Zahringer, et al. 1996. “Bacterial Lipopolysaccharide-Induced Cellular Signalling.” Nature Endotoxin: Chemical Constitution, Biological Recognition, 395 (6699): 284–288. doi:10.1038/26239. Host Response, and Immunological Detoxification.” Yang, M., X. Nie, J. Meng, J. Liu, Z. Sun, and H. Xu. 2017. Current Topics in Microbiology and Immunology 216: “Carbon Nanotubes Activate Limulus Amebocyte Lysate 39–81. doi:10.1007/978-3-642-80186-0_3. Coagulation by Interface Adsorption.” ACS Applied 576 M. MANGINI ET AL. Materials and Interfaces 9 (10): 8450–8454. doi:10.1021/ Enhanced Raman Scattering.” Analytical Chemistry 87 (13): acsami.7b00543. 6553–6561. doi:10.1021/acs.analchem.5b01271. Yu, C. Y., G. Y. Ang, A. L. Chua, E. H. Tan, S. Y. Lee, G. Falero- Zhu, L., S. Li, X. Shao, Y. Feng, P. Xie, Y. Luo, K. Huang, and Diaz, O. Otero, et al. 2011. “Dry-Reagent Gold W. Xu. 2019. “Colorimetric Detection and Typing of E. coli Nanoparticle-Based Lateral Flow Biosensor for the Lipopolysaccharides Based on a Dual Aptamer-functional- Simultaneous Detection of Vibrio Cholerae Serogroups O1 ized Gold nanoparticle probe .” Mikrochimica Acta 186 (2): and O139.” Journal of Microbiological Methods 86 (3): 111 doi:10.1007/s00604-018-3212-9. 277–282., doi:10.1016/j.mimet.2011.04.020. Zuo, M. Y., L. J. Chen, H. Jiang, L. Tan, Z. F. Luo, and Y. M. Zhou, H., D. Yang, N. P. Ivleva, N. E. Mircescu, S. Schubert, R. Wang. 2014. “Detecting Endotoxin with a Flow Cytometry- Niessner, A. Wieser, and C. Haisch. 2015. “Label-Free in Based Magnetic Aptasensor.” Analytical Biochemistry 466: Situ Discrimination of Live and Dead Bacteria by Surface- 38–43. doi:10.1016/j.ab.2014.08.018.