Paul Pantano1*, Elizabeth I. Braun1, Vasanth S. Murali2, Carole A. Mikoryak2, Rockford K. Draper1,2*
1Department of Chemistry and Biochemistry, 2Department of Biological Sciences, The University of Texas at Dallas
Elemental Analyses of CNMs
Benefits of an Extensive Elemental Analysis of CNMs
Quantifying CNMs Accumulated by Biological Cells
Gel Electrophoresis Method to Quantify CNMs
Gel Electrophoresis Method to Quantify CNMs in Aqueous Samples
Gel Electrophoresis Method to Quantify CNMs Extracted from Living Cells
Benefits and Applications of the Gel Electrophoresis Method
Carbon nanomaterials (CNMs), such as single-walled carbon nanotubes (SWCNTs), multi-walled carbon nanotubes (MWCNTs), and graphene (Figure 1), have diverse commercial applications including lighter and stronger composite materials, improved energy storage devices, more sensitive sensors, and smaller transistors.1-3 CNMs are also under development to improve a wide range of biomedical applications from drug delivery, bio-imaging to cancer therapy.4-7 However, there are many reports that various CNMs can be toxic and a potential hazard in biomedical applications, the workplace, and the environment.8-16 Despite a large increase in the literature on CNM toxicity, a clear understanding of the mechanisms involved and how to predict and reduce toxicity has not emerged. As CNMs become more prevalent, it is increasingly important to accurately assess their potential environmental health and safety (EH&S) risks. Two important components of an accurate assessment are the elemental composition of the CNM sample and the amount of CNMs accumulated in living cells and organisms; both of which are challenging measurements for CNMs, as described next.
Figure 1. (Left)Single-layer graphene is a two-dimensional planar honeycomb structure of 2-hybridized carbon atoms and graphene oxide is graphene that has been oxidized to contain various oxygen functional groups. (Middle) A single-walled carbon nanotube (SWCNT) is a single sheet of pristine graphene rolled into a tube and how the sheet is rolled determines whether the SWCNT will be semi-conducting, semi-metallic, or metallic. (Right) A multi-walled carbon nanotube (MWCNT) contains two or more SWCNTs nested one within another.
There are numerous CNM synthetic methods and the composition of commercially-available CNM products can vary according to manufacturers’ synthetic approaches, post-synthetic treatments, and handling practices. This makes it important for end-users to independently evaluate potential differences between products and the batch-to-batch variability of a single product.17-19 This is especially important for SWCNTs because almost all manufacturing processes generate a heterogeneous powdered soot that may contain a variety of SWCNT chiralities, non-tubular carbons such as amorphous carbon and graphitic nanoparticles, metals and metal oxides encased in these carbon phases, and in some cases, catalyst support material such as silica. Since the chemical and physical characterization of SWCNT soot is challenging, measurement priorities and protocols for working with SWCNT soot have been documented in a number of practical guides that recommend the use of a host of analytical methods (including elemental analyses) for a thorough examination.20-28
Historically, the most common methods of elemental analysis used by manufacturers and end-users have been x-ray photoelectron spectroscopy (XPS) and energy-dispersive x-ray spectroscopy (EDS). The advantage of these techniques lie in the number of elements they can detect; specifically, XPS can detect all elements except for hydrogen and helium,29 and EDS can detect all elements between atomic numbers 4 and 92.30 The disadvantage of using these surface-sensitive techniques for the heterogeneous powder analysis stems from their high spatial resolution. XPS has a depth resolution of <100 Å and a lateral resolution of 10 µM – 2 mm,31, 32 and EDS systems associated with electron microscopes (EMs) have a depth resolution of 0.3 – 5 µM and a lateral resolution of 0.5 µM.33, 34 It is therefore prohibitively expensive and time consuming to obtain enough discrete XPS or EM-EDS spectra to accurately represent the bulk properties of heterogeneous SWCNT powders.
Surprisingly, bulk methods of elemental analyses such as carbon, hydrogen, nitrogen, sulfur, and oxygen (CHNS/O) combustion analyses and inductively coupled plasma-mass spectroscopy (ICP-MS) have only recently gained popularity for generating elemental information for SWCNTs and other CNMs. The strengths of CHNS/O analyses and single quadrupole ICP-MS are that they provide statistically relevant data reflecting the underlying properties of an ensemble sample,23, 25 and that they are rapid, readily accessible, and relatively inexpensive instruments compared to other sensitive elemental analysis techniques such as neutron activation analysis and prompt gamma activation analysis.35 Specific advantages of CHNS/O analyses include low analysis costs and analytical ranges of 0.001 – 3.6 mg for carbon, 0.001 – 1.0 mg for hydrogen, 0.001 – 6.0 mg for nitrogen, 0.001 – 2.0 mg for sulfur, and 0.001 – 2.0 mg for oxygen. ICP-MS advantages include a nine decade analytical working range for much of the periodic table and detection limits that are at or below the part per trillion level; disadvantages include the inability to analyze C, H, N, S, O, and elements without naturally occurring isotopes (i.e., most radioactive elements) and difficulties in determining elements that form negative ions such as halides.36
Since the accurate elemental analysis of CNMs is a requisite for exacting assessments of product quality and EH&S risk, we recently developed a routine laboratory procedure for an extensive elemental analysis using three bulk methods of analysis. Specifically, we used CHNS/O combustion analyses, oxygen flask combustion/anion chromatography (OFC/AC) to determine halide content, and ICP-MS to generate a 77-element analysis of SWCNT soot – essentially all elements on the periodic table that are neither radioactive nor noble gases.37 In brief, the approach involves summing the weight percentages of elements detected by the CHNS/O and OFC/AC analyses, and then using ICP-MS to determine the identity and amount of elements that are not C, H, N, S, O, or halides.
As a demonstration of the procedure, we present here a 75-element analysis of an as-received graphene oxide (GO) powder. Table 1 shows the CHN/O and OFC/AC results and that the combined weight percentage of the five elements detected (C, H, N, O, and Cl) was 90.10%. As expected, carbon and oxygen were the most abundant elements observed, and the carbon/oxygen ratio of roughly 1/1 indicated a highly oxidized graphene material. To determine the identity and amount of elements in the remaining mass (9.90%), ICP-MS was used to assay for 67 other elements. As shown in Table 2, 36 elements were observed above their respective ICP-MS method detection limits with B, Na, Ba, and Ca being the most abundant, most likely stemming from the ubiquitous presence of salts and metals/metal oxides in a manufacturing environment.38 Following the conversion of the ICP-MS concentrations into weight percentages, the calculated percentages for elements detected by ICP-MS that accounted for more than 0.01% of the total mass of the GO sample were combined with the percentages for C, H, N, O, and Cl to yield the complete elemental analysis of this sample (Table 3). Of particular note are the low (<0.05%) levels of Ni, Co, and Cu since these are the most common metal substrates used to grow a graphene layer.
Notes: The CHN analyses were based on the Pregl-Dumas technique and the oxygen analysis was based on the Unterzaucher technique, both capable of a precision of ±0.30% and a limit of detection (LOD) of <0.10%. The OFC/AC analyses were based on the Schöniger flask technique; the precision of the method was ±0.30% with a LOD of <0.10% for fluoride, chloride, and bromide, and a LOD of <0.25% for iodide.
Notes: Elemental percentages are listed in descending order of abundance. The following elements (Ag, As, Au, Be, Bi, Ce, Ga, Ge, Hf, Hg, In, Ir, Lu, Os, Pb, Pd, Pt, Re, Rh, Ru, Sc, Se, Sm, Ta, Te, Th, Tl, Tm, U, V, and Zn) were classified as not detected (ND) because they were observed at or below the respective method detection limit (MDL) for that element.
Notes: Elemental percentages are listed in descending order of abundance for elements that accounted for 0.01% or more of the total mass of the GO sample.
The importance of an extensive analysis to identify expected and unexpected elements in a CNM sample stems from increasing evidence that extremely low levels of impurities can play a critical role in a toxicity assessment; for example, the half-maximal inhibitory concentration (IC50) of cationic yttrium released from SWCNTs that potently interfered with the calcium ion channel function of tsA201 kidney cells was 70 ppb.39 Armed with an extensive elemental analysis, researchers can therefore obtain elemental IC50 data for the particular cell line being studied and compare those values to the elemental analysis results of the CNM sample being analyzed. For example, the ~21-ppm concentration of Co observed in the GO sample is comparable to the 19-ppm IC50 for murine fibroblasts exposed to CoCl2,40 and therefore, a potential cause for concern for this cell line depending on the bioavailability of Co within this sample.39
In summary, an extensive elemental analysis of CNMs is vital not only for assessing product quality and batch-to-batch consistency, but for avoiding the possibility of falsely attributing a biological response to a CNM when the cause could have been an elemental impurity contained within the CNM sample. For these reasons, we developed a routine laboratory procedure for an extensive elemental analysis of CNMs using readily-available bulk methods of analysis.37 Ancillary goals were to keep costs to a minimum to facilitate frequent monitoring of batch-to-batch variability, and to minimize sample size requirements to <100 mg in cases where the amount of sample was limited. For example, a minimum CNM sample size of 20 mg was required for a CHNS analysis, 10 mg for an oxygen analysis , 40 mg (10 mg per non-radioactive halide) for an OFC/AC analysis , and 5 mg for a 67-element ICP-MS analysis respectively.
One of the primary metrics with any in vitro assessment of toxicity is the dose of a substance applied to a population of cultured cells; however, this metric can be difficult to accurately determine for CNMs because sparingly-soluble CNMs suspended in cell growth media can behave differently than soluble chemical substances. For example, CNMs can interact with additives in the medium such as nutrients, growth factors, and vital dyes, and they can interact with themselves and form agglomerates and be subject to gravitational forces.41 Clearly, a more useful and accurate dose-response assessment would involve determining the amount of CNMs actually accumulated by cells, which is routinely performed for metal and metal oxide nanomaterials using ICP-optical emission spectroscopy (OES) or ICP-MS.42 In contrast, CNMs are difficult to quantify in situ because the carbon background in biological and environmental systems is extremely high relative to the typically low concentrations of CNMs present in these samples. As a result, the main analytical approaches for detecting cell-accumulated CNMs have involved optical and spectroscopic imaging techniques such as optical, electron, fluorescence, Raman, and near infrared photoluminescence microscopies.42 Nonetheless, while these high-spatial resolution techniques are indispensable for locating and tracking CNMs in individual cells, their use is impractical in determining an average measure of accumulated CNMs in the thousands of cells required for a statistical analysis of biological variability.
In response to this analytical challenge, we developed a rapid, label-free, and sensitive method to detect and quantify SWCNTs and MWCNTs in biological samples43-47 that also works for graphene oxide. The first step involves polyacrylamide gel electrophoresis with sodium dodecyl sulfate (SDS-PAGE) to concentrate CNMs. CNMs are inherently hydrophobic and bind to the hydrophobic tail of negatively-charged SDS, then migrate towards the anode where they encounter the gel and are trapped because they are too large to enter the pores of the gel (Figure 2). For example, a 50-µL sample results in a thin layer of CNMs at the interface of the running buffer and the gel that approaches a 50-fold increase in the CNM concentration. SDS-PAGE also separates the CNMs from contaminating materials that are negatively charged and small enough to migrate into the gel such as proteins, nucleotides, and other cellular material. The second step is detection and quantification of the CNMs at the gel interface. The simplest method we have used has been scanning the gel with a common white light scanner that images the dark CNM bands. The digitized signal from the bands, less any background, is then compared to standards of known amounts to quantify the amount of CNMs in the sample.
Figure 2.Visualization of SWCNT bands after SDS-PAGE of BSA-SWCNT suspensions. Aliquots (1 μL) of each sample were loaded onto a 4% stacking / 10% resolving gel and electrophoresed at 100 V for 2 h. The scanned image shows that SWCNT-containing materials collected at the interface of the stacking gel and the loading wells as thin dark bands. Lane (1) contains pre-stained protein molecular weight markers, lane (8) is blank, and lanes (2-7) contain increasing amounts of SWCNTs. Image from reference 43, reproduced with permission from the American Chemical Society.
As a demonstration of the SDS-PAGE method, we sonicated well-characterized CoMoCAT® SWCNT powder in water with bovine serum albumin (BSA), followed by centrifugation to produce a suspension of BSA-SWCNTs.43 When BSA-SWCNT suspensions were added to an sodium dodecyl sulfate (SDS) solution then electrophoresed by SDS-PAGE, the SWCNTs accumulated in a sharp band at the interface between the loading well and the stacking gel. As shown in Figure 2 with samples containing increasing amounts of SWCNTs suspended with BSA, it is evident that the intensity of the bands increases with the amount of SWCNTs applied.
To quantify SWCNTs, a BSA-SWCNT standard was made by suspending 1.0 mg of the dry SWCNT powder by sonication in 1.0 mL of 100 mg/mL BSA solution. No centrifugation steps or other procedures to selectively remove material was done, so 100% of the SWCNT material added to the BSA was present in the standard suspension. Different volumes of the BSA-SWCNT standard were analyzed by SDS-PAGE and the digitized band intensities from these suspensions were plotted against SWCNT amounts (obtained from the corrected weight percentage of SWCNTs in the powder) to yield the calibration plot shown in Figure 3. The digitized band intensity was linear up to 575 ng of SWCNTs applied to the gel, and in separate experiments with less material, the typical lower limit of detection of SWCNTs in the gel was 5 ng (Figure 3-inset). The applicability of the SDS-PAGE method was further demonstrated by analysis of CoMoCAT® SWCNTs dispersed using an aqueous sodium dodecylbenzenesulfonate (SDDBS) solution instead of BSA. As shown in Figure 3, the slopes of the standard BSA-SWCNT and SDDBS-SWCNT responses were very similar, which indicates that the choice of surfactant does not affect the quantification of SWCNTs. Nonetheless, it should be emphasized that it is imperative to use the same SWCNT material for preparing a standard curve as the SWCNTs under analysis since different SWCNT materials have different optical properties. It should also be noted that the use of white-light scanning is advantageous in that it enables the detection of all SWCNT structures (i.e., semi-conducting, semi-metallic, and metallic SWCNTs) in a sample regardless if the SWCNTs are pristine, oxidized, and/or covalently functionalized. This is significant because a complete Raman spectroscopic analysis to identify all SWCNT structures in a sample can require the acquisition of Raman spectra with at least four different laser lines, and because near infrared photoluminescence spectroscopy cannot detect metallic SWCNTs nor is it ideally suited to detect SWCNTs that have been heavily oxidized or covalently functionalized.42
Figure 3.Measurement of the SWCNT concentrations in CoMoCAT® SWCNT suspensions. The BSA-SWCNT calibration standard (STD) was prepared by sonicating 1.0 mg SWCNT powder with 1.0 mL of 100 mg/mL BSA for 10 min. The resulting suspension was subjected to no subsequent centrifugation steps, and was corrected by the weight percent of SWCNTs found in the SWCNT powder. The preparation of the SDDBS-SWCNT calibration was identical except that 1.0 mL of 0.15% (w/w) SDDBS was used in place of BSA. Both SWCNT standards were diluted 20-fold in SDS sample loading buffer, and various volumes were dispensed into separate SDS-PAGE loading wells. Following electrophoresis at 100 V for 2 h, the pixel intensities from the scanned image of the dark gel bands were corrected with the pixel intensities imaged from a blank gel lane. Mean and standard deviations were calculated from four independent experiments; the relationship between corrected intensities and SWCNT mass was linear for BSA-SWCNT and SDDBS-SWCNT standards (r2 = 0.9825 and 0.9850, respectively). (Inset) Expanded view of a BSA-SWCNT standard curve where each data point represents the mean obtained from three individual suspensions and the error bars represent standard deviations. Typical detection limits were in the range of 1-5 ng SWCNTs per well. Figure from reference 43, reproduced with permission from the American Chemical Society.
SDS is a widely used detergent for dissolving cellular material and the SDS-PAGE method can also be used to extract and quantify CNMs accumulated by living cells. As a demonstration of the method, normal rat kidney (NRK) cells were cultured in medium containing BSA-SWCNTs with a constant SWCNT concentration of 98 μg/mL for 1, 2, or 3 days. The cells were dissolved with SDS and the concentration of SWCNTs in the extracts was measured by the SDS-PAGE method. As shown in Figure 4, SWCNTs were accumulated by NRK cells as a linear function of incubation time and there was no appreciable signal from control cells not exposed to BSA-SWCNTs. Since the amount of protein applied to each lane of the gel was known and it could be determined that 106 cells contained 94.2 ± 6.5 µg of protein, we could calculate that NRK cells at 37 °C accumulated SWCNTs at a rate of 30 fg/day/cell upon exposure to 98 µg/mL of SWCNTs.43 We also measured the uptake of BSA-SWCNTs as a function of concentration in the medium. NRK cells were incubated continuously for 3 days in BSA-SWCNT suspensions containing 25, 49, 74, and 98 μg/mL SWCNTs. SWCNT accumulation increased linearly with concentration and there was no appreciable signal for control cells not exposed to BSA-SWCNTs, or for cells exposed to SWCNTs and incubated at 4 °C (data not shown). The linear increase in cell-accumulated SWCNTs as a function of time and concentration are characteristics of fluid-phase endocytosis, suggesting that these SWCNTs enter NRK cells by this process.43
Figure 4.Determination of the amount of SWCNTs extracted from NRK cells cultured in media containing BSA-SWCNTs as a function of incubation time. Cell lysate samples were prepared from cells incubated for 1, 2, or 3 d in media with or without BSA-SWCNTs containing 98 μg/mL SWCNTs. Cell lysate samples equivalent to 300 μg of total cellular protein were analyzed to ensure that equal numbers of cells were compared. (Top) Scanned digital image of the control lane showing the dark SWCNT containing bands at the interface of the stacking gel and the loading wells following SDS-PAGE at 100 V for 2 h. Lane (1), 3-d incubation in media containing no SWCNT suspension; lanes (2-4), incubation in media containing BSA-SWCNTs for 1, 2, or 3 d, respectively. (Bottom) Plot of the SWCNT content in cell lysate samples determined from corrected mean pixel intensities of the SWCNT containing bands as a function of incubation time (r2 = 0.9922). Each data point is the mean obtained from three independent experiments and the error bars show standard deviations. Figure from reference 43, reproduced with permission from the American Chemical Society.
In summary, the SDS-PAGE method of quantifying CNMs is sensitive, has low sample volume requirements, and is inexpensive to operate requiring only a gel electrophoresis unit, a flatbed scanner, and image analysis software. Furthermore, it can be easily customized for CNMs of different dimensions by adjusting the porosity of the gel used to concentrate CNMs at the gel interface; for example, the 4% polyacrylamide gel typically used with SWCNTs can be increased to 15% for the analysis of the smallest graphene and graphene oxide nanoflakes. The method can be applied to determine the rates and mechanisms by which CNMs are accumulated by cells making it straight-forward to investigate CNM parameters involved in the accumulation process (e.g., CNM dimensions, concentrations, and surface chemistries), as well as, parameters required to assess the energy dependence of accumulation (e.g., cell incubation times and temperatures). The method is additionally distinctive in that it is particularly useful in studies to optimize the therapeutic effectiveness of CNM constructs, to determine the ADMET profile of CNM constructs, and to compare the toxicity of different CNMs.45 Finally, in pilot studies, we have demonstrated the feasibility of quantifying CNMs in a variety of biological and environmental samples such as SWCNTs spiked in whole blood and MWCNTs spiked in lymph node tissue extracts; and recently, we have validated the method to quantify MWCNTs extracted from zebrafish embryos exposed to MWCNTs.47
Understanding potential EH&S risks of CNMs is vital for the advancement of nanotechnologies that will be used by industries in the coming decades. However, key physical and chemical properties of CNMs that influence their effect upon and the accumulation within cells are not well understood. One reason for this is that elemental impurities that can influence biological responses are not always identified, which is our motivation for developing a routine laboratory procedure for an extensive elemental analysis of CNMs. Another reason has been a lack of methods to directly measure the amount of CNMs that have been accumulated by cells, tissue, and organisms. The SDS-PAGE method for extracting and quantifying CNMs from biological samples was developed to fill this gap to aid in better understanding how CNMs interact with living systems and for generating new knowledge that can be applied in predicting and reducing CNM toxicity.
Follow these links for a list of Single-Walled Carbon Nanotube Products and Graphene Oxide Products offered by Sigma-Aldrich® Materials Science.