Synthesis and cytotoxicity of silicon nanoparticles with covalently attached organic monolayers

Abstract

A series of highly monodisperse silicon nanoparticles (Si NPs) with either positively (amine), neutral (azide) or negatively (carboxylic acid) charged covalently attached organic monolayers were synthesized and investigated for their cytotoxicity. Infrared data confirmed the presence of these covalently attached surface groups. The Si NPs were characterized by absorption and fluorescence spectroscopy. The cytotoxicity was investigated in Caco-2 cells by determining the cell viability and proliferation. The IC50 values for the Si NPs ranged from 20 μg/l for the amine-terminated Si NPs, via 550–850 μg/l for the azide-terminated Si NPs to non-toxic (no measureable IC50) for the carboxylic acid-terminated Si NPs. These results indicate a trend in cytotoxicity, depending on surface charge, i.e., that positively charged Si NPs are more cytotoxic than negatively charged Si NPs. Interestingly, it appeared that the cytotoxicity of the Si NP-NH₂ depends strongly on the presence of fetal calf serum in the medium.     

A 3D co-culture of three human cell lines to model the inflamed intestinal mucosa for safety testing of nanomaterials

Oral exposure to nanomaterials is a current concern, asking for innovative biological test systems to assess their safety, especially also in conditions of inflammatory disorders. Aim of this study was to develop a 3D intestinal model, consisting of Caco-2 cells and two human immune cell lines, suitable to assess nanomaterial toxicity, in either healthy or diseased conditions. Human macrophages (THP-1) and human dendritic cells (MUTZ-3) were embedded in a collagen scaffold and seeded on the apical side of transwell inserts. Caco-2 cells were seeded on top of this layer, forming a 3D model of the intestinal mucosa. Toxicity of engineered nanoparticles (NM101 TiO₂, NM300 Ag, Au) was evaluated in non-inflamed and inflamed co-cultures, and also compared to non-inflamed Caco-2 monocultures. Inflammation was elicited by IL-1β, and interactions with engineered NPs were addressed by different endpoints. The 3D co-culture showed well preserved ultrastructure and significant barrier properties. Ag NPs were found to be more toxic than TiO₂ or Au NPs. But once inflamed with IL-1β, the co-cultures released higher amounts of IL-8 compared to Caco-2 monocultures. However, the cytotoxicity of Ag NPs was higher in Caco-2 monocultures than in 3D co-cultures. The naturally higher IL-8 production in the co-cultures was enhanced even further by the Ag NPs. This study shows that it is possible to mimic inflamed conditions in a 3D co-culture model of the intestinal mucosa. The fact that it is based on three easily available human cell lines makes this model valuable to study the safety of nanomaterials in the context of inflammation.     

Novel luminescence assay offers new possibilities for the risk assessment of silica nanoparticles

Recent opinions of the Scientific Committee on Consumer Products (SCCP) of the European Commission emphasize the missing validation of in vitro methodologies for nanomaterials and suggest that a review of the safety of insoluble nanomaterials presently used is required. Therefore the influence of Fluospheres® and silica nanoparticles, representing the class of organic and inorganic nanoparticles, respectively, on the lactate dehydrogenase (LDH) assay and the novel luminescence based Vialight® assay was tested. While LDH assay showed strong interactions with the tested silica particles, these problems may be overcome by novel methods like luminescence based assays. These findings suggest that even well characterized assay systems need a careful evaluation of the particle assay interactions when working with nanoparticles. Furthermore, particles based on the same material exhibit different biological properties depending on whether the material is used in micro- or nanometer range.     

Use of simulated intestinal fluid for Caco-2 permeability assay of lipophilic drugs

The most commonly used method to assess intestinal permeability is the measurement of compound flux across a Caco-2 cells monolayer by using Hanks balanced salt solution (HBSS)-like buffers. Nevertheless, lipophilic acid drugs are poorly or not at all soluble in these types of buffers and their adsorption on the transwell plate is commonly observed. To reduce adsorption and increase solubility, permeability assays need to be developed in conditions other than classic conditions for lipophilic compounds. The best model to increase recovery of lipophilic compounds was determined as fasted state simulated intestinal fluid (FaSSIF) in the apical compartment and HBSS with 1% bovine serum albumin (BSA) in basolateral compartment. This model allows a correlation between absorption on Caco-2 cells and absorbed fraction in humans. For 35 compounds, only 2 outliers were observed in the Caco-2 assay using the FaSSIF model. These two outliers were the same outlier compounds as those observed with a classic Caco-2 method. Furthermore, a permeability assay of Pgp substrates evidenced efflux transport in both models and addition of a Pgp inhibitor suppressed Pgp efflux transport. FaSSIF in the apical compartment and HBSS with 1% BSA in the basolateral compartment is the model of choice to predict in vivo absorption for lipophilic acid drugs.     

Chicken nucleated blood cells as a cellular model for genotoxicity testing using the comet assay.

Mycotoxins can frequently occur in animal feed and human food. T-2 toxin, as the most toxic trichothecene, has been implicated as the causative agent in a variety of animal diseases and is associated with some human diseases. The comet assay was performed as a test for detection of DNA damage caused by T-2 toxin in peripheral blood cells of chicken. The suitability of the comet assay as a biomarker for genotoxic analysis has been applied in studies using human white blood cells. It can be applied to any tissue from which a single cell suspension can be obtained. The method has already been applied to chicken as a foodstuff for detection of irradiation of food containing DNA. However, application of the method on chicken blood cells has not been set up yet. The aim of this research was to develop a protocol for detection of DNA damage induced by T-2 toxin in chicken blood cells. Chickens were administered orally with T-2 toxin and the samples of whole blood were collected at 24 h post treatment. The DNA damage was determined by an increase in the comet parameters in tested animals. Our results show that T-2 toxin had induced significant DNA damage in treated chicken as compared with control animals, indicating that the assay can be used for the assessment of primary DNA damage caused by mycotoxins.     

In vitro evaluation of chitosan-EDTA conjugate polyplexes as a nanoparticulate gene delivery system

It was the purpose of this study to evaluate the potential of different molecular-weight chitosan-EDTA conjugates as a carrier matrix for nanoparticulate gene delivery systems. Covalent binding of EDTA to more than one chitosan chain provides a cross-linked polymer that is anticipated to produce stabilized particles. pDNA/chitosan-EDTA particles, generated via coazervation, were characterized in size and zeta potential by electrophoretic light scattering and electron microscopy. Stability was investigated at different pH values by enzymatic degradation and subsequent gel retardation assay. Lactate dehydrogenase assay was performed to determine toxicity. Furthermore, transfection efficiency into Caco-2 cells was assessed using a beta-galactosidase reporter gene. Chitosan-EDTA produced from low-viscous chitosan with 68% amino groups being modified by the covalent attachment of EDTA showed the highest complexing efficacy resulting in nanoparticles of 43 nm mean size and exhibiting a zeta potential of +6.3 mV. These particles were more stable at pH 8 than chitosan control particles. The cytotoxicity of chitosan-EDTA particles was below 1% over a time period of 4 hours. These new nanoplexes showed 35% improved in vitro transfection efficiency compared with unmodified chitosan nanoparticles. According to these results, the chitosan-EDTA conjugate may be a promising polymer for gene transfer.     

Cytotoxicity and biological effects of functional nanomaterials delivered to various cell lines

Functional nanomaterials that included gold, silver nanoparticles and single wall carbon nanotubes were delivered to two cell lines (MLO‐Y4 osteocytic cells and HeLa cervical cancer cells) in various concentrations. The cells were found to uptake the nanomaterials in a relatively short time, a process that significantly affected the shape and the size of the cells. The percentage of cellular death, due to the delivery of these nanomaterials, was found to be the highest for carbon nanotubes and increased gradually with the concentration of these nanostructures. Moreover, when the nanomaterials were delivered to the cells combined with commonly used chemotherapeutic agents such as etoposide or dexamethasone, the number of the cells that died increased significantly (100–300%) as compared with the case when only the nanomaterials or the chemotherapeutic agents were delivered. The experimental results were confirmed by Caspase 3 studies, indicating a strong interaction between the nanomaterials used in this study and the protein structure of the cells, which allowed a more effective action of the apoptotic agents. These findings could be the foundation of a new class of cancer therapies that are composed of both chemotherapeutic agents and nanomaterials. Copyright © 2009 John Wiley & Sons, Ltd.     

Comparative cytotoxicity of nanosilver in human liver HepG2 and colon Caco2 cells in culture

The use of silver nanoparticles in food, food contact materials, dietary supplements and cosmetics has increased significantly owing to their antibacterial and antifungal properties. As a consequence, the need for validated rapid screening methods to assess their toxicity is necessary to ensure consumer safety. This study evaluated two widely used in vitro cell culture models, human liver HepG2 cells and human colon Caco2 cells, as tools for assessing the potential cytotoxicity of food‐ and cosmetic‐related nanoparticles. The two cell culture models were utilized to compare the potential cytotoxicity of 20‐nm silver. The average size of the silver nanoparticle determined by our transmission electron microscopy (TEM) analysis was 20.4 nm. The dynamic light scattering (DLS) analysis showed no large agglomeration of the silver nanoparticles. The concentration of the 20‐nm silver solution determined by our inductively coupled plasma–mass spectrometry (ICP‐MS) analysis was 0.962 mg ml^–1. Our ICP‐MS and TEM analysis demonstrated the uptake of 20‐nm silver by both HepG2 and Caco2 cells. Cytotoxicity, determined by the Alamar Blue reduction assay, was evaluated in the nanosilver concentration range of 0.1 to 20 µg ml^–1. Significant concentration‐dependent cytotoxicity of the nanosilver in HepG2 cells was observed in the concentration range of 1 to 20 µg ml^–1 and at a higher concentration range of 10 to 20 µg ml^–1 in Caco2 cells compared with the vehicle control. A concentration‐dependent decrease in dsDNA content was observed in both cell types exposed to nanosilver but not controls, suggesting an increase in DNA damage. The DNA damage was observed in the concentration range of 1 to 20 µg ml^–1. Nanosilver‐exposed HepG2 and Caco2 cells showed no cellular oxidative stress, determined by the dichlorofluorescein assay, compared with the vehicle control in the concentration range used in this study. A concentration‐dependent decrease in mitochondria membrane potential in both nanosilver exposed cell types suggested increased mitochondria injury compared with the vehicle control. The mitochondrial injury in HepG2 cells was significant in the concentration range of 1 to 20 µg ml^–1, but in Caco2 cells it was significant at a higher concentration range of 10 to 20 µg ml^–1. These results indicated that HepG2 cells were more sensitive to nanosilver exposure than Caco2 cells. It is generally believed that cellular oxidative stress induces cytotoxicity of nanoparticles. However, in this study we did not detect any nanosilver‐induced oxidative stress in either cell type at the concentration range used in this study. Our results suggest that cellular oxidative stress did not play a major role in the observed cytotoxicity of nanosilver in HepG2 and Caco2 cells and that a different mechanism of nanosilver‐induced mitochondrial injury leads to the cytotoxicity. The HepG2 and Caco2 cells used this study appear to be targets for silver nanoparticles. The results of this study suggest that the differences in the mechanisms of toxicity induced by nanosilver may be largely as a consequence of the type of cells used. This differential rather than universal response of different cell types exposed to nanoparticles may play an important role in the mechanism of their toxicity. In summary, the results of this study indicate that the widely used in vitro models, HepG2 and Caco2 cells in culture, are excellent systems for screening cytotoxicity of silver nanoparticles. These long established cell culture models and simple assays used in this study can provide useful toxicity and mechanistic information that can help to better inform safety assessments of food‐ and cosmetic‐related silver nanoparticles. Published 2014. This article is a U.S. Government work and is in the public domain in the USA.     

In vitro cytotoxicity of superparamagnetic iron oxide nanoparticles on neuronal and glial cells. Evaluation of nanoparticle interference with viability tests

Superparamagnetic iron oxide nanoparticles (ION) have attracted great interest for use in several biomedical fields. In general, they are considered biocompatible, but little is known of their effects on the human nervous system. The main objective of this work was to evaluate the cytotoxicity of two ION (magnetite), coated with silica and oleic acid, previously determining the possible interference of the ION with the methodological procedures to assure the reliability of the results obtained. Human neuroblastoma SHSY5Y and glioblastoma A172 cells were exposed to different concentrations of ION (5–300 µg ml^–1), prepared in complete and serum‐free cell culture medium for three exposure times (3, 6 and 24 h). Cytotoxicity was evaluated by means of the MTT, neutral red uptake and alamar blue assays. Characterization of the main physical–chemical properties of the ION tested was also performed. Results demonstrated that both ION could significantly alter absorbance readings. To reduce these interferences, protocols were modified by introducing additional washing steps and cell‐free systems. Significant decreases in cell viability were observed for both cell lines in specific conditions by all assays. In general, oleic acid‐coated ION were less cytotoxic than silica‐coated ION; besides, a serum‐protective effect was observed for both ION studied and cell lines. These results contribute to increase the knowledge of the potential harmful effects of ION on the human nervous system. Understanding these effects is essential to establish satisfactory regulatory policies on the safe use of magnetite nanoparticles in biomedical applications. Copyright © 2015 John Wiley & Sons, Ltd.     

检测白喉毒素特异性细胞毒性的Vero细胞/MTT法的建立

目的 建立检测白喉毒素（diphtheria toxin,DT）特异性细胞毒性的Vero细胞/MTT法.方法 用MEM培养液对DT进行不同倍数的预稀释,用目测法确定DT的最适预稀释倍数.用建立的方法检测以最适预稀释倍数稀释的DT,以确定该法的灵敏度和细胞病变半数抑制浓度（IC50）,同时绘制剂量-反应曲线以验证该法的重复性.用建立的方法于不同pH、盐浓度或蔗糖浓度条件下检测DT,计算DT回收率以验证该法的耐用性.结果 目测法确定的DT最适预稀释倍数为106倍.建立的方法的平均检测灵敏度为（6.01±1.44）×10^-5 lf/ml DT,细胞病变半数抑制率对数值（log IC50）范围为5.02 ～ 5.82,变异系数为5.82％,该法具有较好的重复性.用建立的方法检测不同条件下的DT显示,DT的平均回收率为93.7％ ～ 110.4％,该法具有较好的耐用性.结论 Vero细胞/MTr法可用于检测DT特异性细胞毒性.     

Non-invasive monitoring of cytotoxicity based on kinetic changes of cellular autofluorescence

A quantitative, non-destructive cellular autofluorescence based in vitro imaging assay has been developed and applied to study the cytotoxicity of Sodium Lauryl Sulfate (SLS) and HgCl₂ on Balb/c 3T3 cells. A phenomenological double logistic model was proposed to quantify and relate the observed kinetic changes of fluorescence to the toxic potency of chemical compounds. This work forms the basis for cellular autofluorescence measurements in in vitro toxicity screening assays.     

9种中药提取物细胞毒性及其与人结肠腺癌细胞系单层细胞旁通透性的相关性研究

目的了解9种中药提取物对Caco-2细胞毒性与对其细胞单层细胞-细胞间通透性的相关性。方法用MTT法测定9种中药提取物24 h及48 h的细胞毒,并计算IC50值。采用TranswellTM培养板构建Caco-2细胞单层,细胞电阻仪测定受试物对其TEER值的影响,找到影响细胞单层紧密连接的临界剂量,以此剂量与细胞毒参数行相关分析。结果 24、48 h细胞毒参数(IC50)与临界剂量(最大无毒剂量、最小有毒剂量)相关系数均>0.84,相关性检验P值均<0.01。最大无毒剂量与IC50的比值列阵显示,《小品方》甘草饮最小,三七醇提物最大。结论 9种中药提取物对Caco-2细胞毒与该细胞单层细胞旁通透性高度相关,采用细胞毒参数来预估中药提取物在Caco-2单层细胞模型供侧起始最大浓度具可行性。     

Cell detachment and cloning efficiency as parameters for cytotoxicity

Cell detachment and cloning efficiency of Baby Hamster Kidney cells (BHK-21 C13) were used as parameters to quantify cytotoxicity in vitro of 3 endogenous chemicals (glutathione, L-methionine, L-cysteine HCl) and 4 organotin compounds (tributyltinoxide, tributylchloride, tetrabutyltin, tetraphenyltin). IC₅₀ values (inhibitory concentration at which the cloning efficiency was reduced to 50%) were estimated to be larger than 10^?3 M for all 3 endogenous substances, which served as a calibration of the cell culture system for non-toxic chemicals. For the 2 tributyltin salts the IC₅₀ values were estimated to be near 10^?6 M and for the 2 tetraalkyltin compounds near 10^?5 M.The estimated CD₅₀ values (concentration at whicb 50% of the cells detach) were at least twice as large as the corresponding IC₅₀ values for all 7 chemicals tested. For the toxic tributyltin salts the cell detachment assay was 30–60 times less sensitive than the cloning efficiency assay. However, both assays rank all compounds tested in the same sequence of toxicity as that known from in vivo studies.     

A bioluminescent cytotoxicity assay for assessment of membrane integrity using a proteolytic biomarker

Measurement of cell membrane integrity has been widely used to assess chemical cytotoxity. Several assays are available for determining cell membrane integrity including differential labeling techniques using neutral red and trypan blue dyes or fluorescent compounds such as propidium iodide. Other common methods for assessing cytotoxicity are enzymatic “release” assays which measure the extra-cellular activities of lactate dehydrogenase (LDH), adenylate kinase (AK), or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in culture medium. However, all these assays suffer from several practical limitations, including multiple reagent additions, scalability, low sensitivity, poor linearity, or requisite washes and medium exchanges. We have developed a new cytotoxicity assay which measures the activity of released intracellular proteases as a result of cell membrane impairment. It allows for a homogenous, one-step addition assay with a luminescent readout. We have optimized and miniaturized this assay into a 1536-well format, and validated it by screening a library of known compounds from the National Toxicology Program (NTP) using HEK 293 and human renal mesangial cells by quantitative high-throughput screening (qHTS). Several known and novel membrane disrupters were identified from the library, which indicates that the assay is robust and suitable for large scale library screening. This cytotoxicity assay, combined with the qHTS platform, allowed us to quickly and efficiently evaluate compound toxicities related to cell membrane integrity.     

Incompatibility of silver nanoparticles with lactate dehydrogenase leakage assay for cellular viability test is attributed to protein binding and reactive oxygen species generation

A growing number of studies report that conventional cytotoxicity assays are incompatible with certain nanoparticles (NPs) due to artifacts caused by the distinctive characteristics of NPs. Lactate dehydrogenase (LDH) leakage assays have inadequately detected cytotoxicity of silver nanoparticles (AgNPs), leading to research into the underlying mechanism. When ECV304 endothelial-like umbilical cells were treated with citrate-capped AgNPs (cAgNPs) or bare AgNPs (bAgNPs), the plasma membrane was disrupted, but the LDH leakage assay failed to detect cytotoxicity, indicating interference with the assay by AgNPs. Both cAgNPs and bAgNPs inactivated LDH directly when treated to cell lysate as expected. AgNPs adsorbed LDH and thus LDH, together with AgNPs, was removed from assay reactants during sample preparation, with a resultant underestimation of LDH leakage from cells. cAgNPs, but not bAgNPs, generated reactive oxygen species (ROS), which were successfully scavenged by N-acetylcysteine or ascorbic acid. LDH inhibition by cAgNPs could be restored partially by simultaneous treatment with those antioxidants, suggesting the contribution of ROS to LDH inactivation. Additionally, the composition of the protein corona surrounding AgNPs was identified employing liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. In sum, the LDH leakage assay, a conventional cell viability test method, should be employed with caution when assessing cytotoxicity of AgNPs.     

Evaluation of the toxicity of food additive silica nanoparticles on gastrointestinal cells

Silica nanoparticles (NPs) have been widely used in food products as an additive; however, their toxicity and safety to the human body and the environment still remain unclear. As a food additive, silica NPs firstly enter the human gastrointestinal tract along with food, thus their gastrointestinal toxicity deserves thorough study. Herein, we evaluated the toxicity of food additive silica NPs to cells originating from the gastrointestinal tract. Four silica NP samples were introduced to human gastric epithelial cell GES‐1 and colorectal adenocarcinoma cell Caco‐2 to investigate the effect of silica sample, exposure dose and exposure period on the morphology, viability and membrane integrity of cells. The cell uptake, cellular reactive oxygen species (ROS) level, cell cycle and apoptosis were determined to reveal the toxicity mechanism. The results indicate that all four silica NPs are safe for both GES‐1 and Caco‐2 cells after 24‐h exposure at a concentration lower than 100 µg ml^–1. At a higher concentration and longer exposure period, silica NPs do not induce the apoptosis/necrosis of cells, but arrest cell cycle and inhibit the cell growth. Notably, silica NPs do not pass through the Caco‐2 cell monolayer after 4‐h contact, indicating the low potential of silica NPs to cross the gastrointestinal tract in vivo. Our findings indicate that silica NPs could be used as a safe food additive, but more investigations, such as long‐term in vivo exposure, are necessary in future studies. Copyright © 2013 John Wiley & Sons, Ltd.     

Isolation and purification of enniatins A, A1, B, B1, produced by Fusarium tricinctum in solid culture, and cytotoxicity effects on Caco-2 cells

Enniatins (ENs) are antibiotic compounds of hexadepsipeptidic structure produced by several strains of Fusarium spp. The ENs A, A₁, B, B₁ were purified from extracts of Fusarium tricinctum grown on a solid medium of corn, by a low pressure liquid chromatography (LPLC) on reverse phase of Amberlite XAD-7 followed by semipreparative LC. The purity and the structure of the isolated compounds were confirmed by LC-MS/MS.The technique of the purification of the fungal extract enabled complete separation of the ENs A, A₁, B, B₁ with a mean purity of 97% for all the compounds.The cytoxicity of the ENs was tested in the cell lines of human origin (epithelial colorectal adenocarcinoma cells, Caco-2) by MTT assays. Only EN A₁ and B₁ evoked toxicity at the tested concentrations. The inhibitory concentration (IC₅₀) for EN A₁ on Caco-2 cells was 12.3 μM, whereas the IC₅₀ produced by the EN B₁ was 19.5 μM.This study indicates that ENs, fungal metabolites that are commonly found in corn and in general in product composed by corn, may have a toxic potential for human health.     

BODIPY-conjugated chitosan nanoparticles as a fluorescent probe

Recently, development of fluorescent nanoparticle-based probes for various bioimaging applications has attracted great attention. This work aims to develop a new type fluorescent nanoparticle conjugate and evaluate its cytotoxic effects on A549 and BEAS 2B cell lines. Throughout the study, ionically crosslinked chitosan nanoparticles (CNs) were conjugated with carboxylated 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene (BODIPY-COOH). The results of conjugates (BODIPY-CNs) were investigated with regard to their physic-chemical, optical, cytotoxic properties and cellular internalization. The morphology of BODIPY-CNs was found to be spherical in shape and quite uniform having average diameter of 70.25?±?11.99?nm. Cytotoxicty studies indicated that although BODIPY-COOH itself was quite toxic on both A549- and BEAS 2B-treated cells, CNs increased the cell viability of both cell lines via conjugation to BODIPY-COOH fluorescent molecule up to 67% for A549 and 74% for BEAS 2B cells. These results may suggest a possible utilization of the new fluorescent nanoparticle-based probe for bioimaging in biology and medicine.     

高效液相色谱法测定U87细胞内外液中的阿霉素含量

目的:建立U87肿瘤细胞中阿霉素的高效液相色谱荧光检测方法,为进一步研究化疗药对肿瘤细胞功能和活性的影响及逆转耐药研究提供方法学基础。方法:U87细胞以1×106·mL-1密度与不同浓度阿霉素孵育24h为药物处理条件。将细胞分为空白对照组和加药组,采用HPLC法测定细胞内外阿霉素的含量。色谱柱为Inertsil ODS-3(250mm×4.6mm,5μm);流动相为甲醇-乙腈-0.01mol·L-1磷酸二氢铵(40∶15∶45)(以0.1%冰醋酸调整pH至3.52);柱温35℃,流速1.4mL·min-1,荧光检测器检测波长λex=495nm,λem=560nm,以盐酸柔红霉素为内标。结果:细胞内阿霉素在0.025～7.5mg·L-1范围内线性关系好,r=0.9999;细胞外阿霉素在0.1～10.0mg·L-1范围内线性关系好,r=0.9999。细胞内外液阿霉素的检测限均为0.005mg·L-1,细胞外液的定量限为0.1mg·L-1。日内、日间RSD均小于10%,样品稳定性好。结论:采用HPLC荧光法测定U87细胞内外液中阿霉素的含量,方法简单、准确、线形范围宽、灵敏度高、精密度好,可用于对培养的肿瘤细胞内外...