Detection of Microbial Contamination in Nanomaterials Using LAL, rFC and Cell-Based Assays: Implications for Nanotoxicological Hazard Assessment Authors: Peng Lei, Fikirte Debebe Zegeye, Mayes Alswady-Hoff, Chiara Marcolungo, Pernille Høgh Danielsen, Anne Mette Madsen, Håkan Wallin, Ulla Vogel, Shan Zienolddiny-Narui, Johanna Samulin Erdem, Chuanxi Wang Published: 2025 ABSTRACT AbstractAccurate detection of microbial contamination is essential in the assessment of toxicological and immunological responses to various materials, as low-level contaminants can lead to confounding results. Traditional endotoxin testing relies on the Limulus Amebocyte Lysate (LAL) assay, which depends on horseshoe crab blood and raises both ecological and ethical concerns. Sustainable alternatives such as recombinant Factor C (rFC) provide a promising solution, yet validation for the detection of endotoxin in nanomaterials remains incomplete. In this study, we have used rFC alongside Toll-like receptor (TLR) reporter assays to detect both endotoxin and broader microbial contaminants in 31 nanomaterials from diverse classes. Special attention was given to assay interference by nanomaterials to ensure reliable detection. The rFC assay demonstrated a sensitive detection limit of 0.005 EU/mL, equivalent to the LAL assay, and showed that more than 50% of tested nanomaterials contained low-level endotoxin contamination. Additionally, several nanomaterials activated the TLR2 reporter, indicative of microbial contaminants beyond endotoxin. These results suggest that rFC can serve as a sustainable and reliable replacement for LAL in nanomaterial endotoxin testing but also emphasize the limitations of relying solely on endotoxin-specific assays. We recommend that future nanotoxicological evaluations integrate rFC with complementary methods, such as TLR-based approaches, and include thorough interference controls to ensure robust and comprehensive microbial contamination assessment.Keywords: nanotoxicology, nano safety, bacterial contamination, nanoparticles, inflammation, endotoxin, toll-like receptor 1. Introduction The impact of engineered nanomaterials on biological processes and, consequently, their implications for human health and the environment are important to the nanotechnology industry, regulators, and scientists. Occupational exposure to engineered nanomaterials has been associated with adverse health effects, particularly in the respiratory system. Pulmonary inflammation and fibrosis are among the most consistently reported outcomes in both experimental models and exposed workers, with oxidative stress and inflammation recognized as key mechanistic pathways [1,2,3,4,5]. These effects can arise not only from the intrinsic properties of the nanomaterials themselves but also from extrinsic factors, e.g., microbial contamination. Assessment of contamination is critical as even minute amounts of microbial contamination may influence the toxicological responses of nanomaterials, resulting in an over- or underestimation of toxic effects [6]. In occupational hazard assessment, unrecognized contamination can lead to inappropriate hazard classification and risk management decisions. Therefore, risk assessment of engineered nanomaterials produced and handled in occupational environments requires correct evaluation of potential contamination and inclusion of optical-interference controls in alignment with internationally recognized quality-control standards within pharmaceutical manufacturing, food safety, and occupational air monitoring [7,8,9]. Microbiological contamination of nanomaterials is common and may occur during production, prolonged storage, or handling [10]. Microbial components of Gram-negative bacteria are well-recognized contaminants of nanomaterials acknowledged to influence inflammatory and immune responses. Surprisingly, reports have shown that between 30 and 40% of nanomaterials associated with EU-funded nanomedicine-related projects have endotoxin levels above the regulatory requirements for preclinical assessment [11]. Endotoxin is heat stable and cannot be easily removed by common sterilization techniques such as autoclaving and ionization radiation [12]. Keeping this in mind, several recent reviews have stressed the importance of sufficient endotoxin testing to avoid misinterpretation of toxicological endpoints [11,13]. Despite this, reporting of endotoxin contamination is often lacking, severely diminishing the potential relevance of nanotoxicological findings. This is exemplified by findings in a recent systematic review on TiO2 nanoparticles evaluating effects such as inflammation, oxidative stress, and genotoxicity, showing that less than 30% of the included studies had considered endotoxin contamination [14]. The assessment of microbial contaminations in nanomaterials remains technically challenging. More robust testing strategies need to be developed, and limitations of existing strategies should be carefully considered when interpreting the results. For endotoxin testing, the chromogenic, turbidimetric, or gel clot Limulus amoebocyte lysate (LAL) assays are most used due to their relatively fast, sensitive, and specific detection. However, due to ethical and sustainability concerns, the U.S. Food and Drug Administration in 2012 [15] published a new guidance recommendation suggesting two new methods, namely the monocyte activation test (MAT) [16] and the recombinant Factor C (rFC) assay [17,18] as alternative methods for endotoxin and pyrogen testing [11]. Notably, the MAT assay measures inflammatory effects by assessing pro-inflammatory cytokine secretion and is thus not specific to endotoxin but detects any inflammatory responses, including potential nanomaterial-specific effects [12]. On the contrary, rFC utilizes a synthetic enzyme for endotoxin detection and can be considered a direct substitute for the LAL assay, making it a more suitable alternative for endotoxin testing of nanomaterials. In addition to these regulatory accepted methods for endotoxin testing, several alternative approaches have been suggested, including the Toll-like receptor 4 (TLR4) reporter assay, high performance liquid chromatography and mass spectrometry (HPLC-MS) methods, and macrophage activation test [11]. Each of these approaches have strengths and limitations when applied to nanomaterials. TLR4 reporter assays offer high biological relevance but rely on living cells and can be affected by cytotoxicity. Furthermore, some evidence suggests that nanomaterials may directly interact with cell surface receptors and elicit inflammatory signaling [19]. HPLC–MS provides structural information yet is limited by extraction efficiency and matrix effects, while the macrophage activation test, similar to MAT, is easily confounded by nanomaterial-induced inflammation. Moreover, biochemical tests, such as LAL and rFC methods, quantify only dispersed endotoxins, while cell-based detection methods sense biologically available, including particle-associated, contaminants, and may provide complementary information to the classical LAL or rFC assays. These methods may prove to be valuable substitutes in the future but need validation and thorough evaluation of aspects such as interference, sensitivity, and specificity. While there is now a general acceptance of the need for endotoxin contamination assessment in nanotoxicological studies, the potential effects of other microbial contaminants have, at large, been underestimated or ignored in the assessment of nanomaterial toxicity. To date, no assessment methods have been thoroughly investigated or validated for the detection of other microbial contaminants, such as Gram-positive contaminants or fungi, on nanomaterials. Furthermore, nanomaterials may interfere with fluorometric and absorbance-based measurements or directly interact with proteins or reagent components, thereby affecting assay readouts, resulting in insufficient, inconclusive, or misleading results. In addition, large interlaboratory variations in endotoxin measurements have been reported [20], and nanomaterials often interfere with traditional endotoxin tests [21]. Altogether, highlighting a need for multiple assessment methods and established standardized protocols for interference control. Thus, improved methods to detect microbial contamination are urgently needed to ensure that observed responses genuinely reflect nanomaterial-related toxicity rather than artifacts of experimental impurity. This study aimed to systematically evaluate microbial contaminations across 31 different nanomaterials representing the major classical nanomaterial classes, i.e., metal and metal oxides, carbon-based materials, including graphene, carbon black and carbon nanotubes, and nanoclays. The suitability of the rFC and TLR reporter assays as detection methods of microbial contamination was assessed, and nanomaterial interferences were thoroughly investigated. Our data suggest that TLR reporter assays hold promise as complementary tools for assessing microbial contamination of nanomaterials. These findings highlight the urgent need for standardized contamination assessment approaches that employ multiple testing methods and include appropriate interference controls, ensuring accurate interpretation of nanomaterial-induced immunological responses. 2. Materials and Methods 2.1. Materials and Characterization Material specifics and characterization data are summarized in Table 1. The NANOGENOTOX dispersion protocol, with modifications, was applied for particle dispersion for reporter cell exposure [22]. Briefly, 1.28 mg/mL stock solutions were prepared by dispersing 7.68 mg of nanomaterials in 6 mL of 0.05% (v/v) BSA/endotoxin-free water (Merck, St. Louis, MO, USA, and Thermo Fisher Scientific, Waltham, MA, USA). The dispersion was sonicated on ice for 16 min at 10% amplitude using a Branson Sonifier S-450D (Branson Ultrasonics Corp., Danbury, CT, USA), immediately prior to exposure. The hydrodynamic diameter of the particles in dispersion media (DLS stock) and exposure media at the time of exposure (DLS media) and after 24 h (DLS media 24 h) was assessed by dynamic light scattering (DLS) (Zetasizer Nano, Malvern Instruments, Ltd.; Worcestershire, UK) (Table 1). Comprehensive physicochemical datasets for these same materials have been published previously, including CNT surface functionalization/metal content and nanoclay surface modifier characterization [2,23,24,25,26,27,28,29,30,31,32,33,34,35]. Table 1Included nanomaterials, their characteristics, and hydrodynamic diameters.MaterialIDSize (nm) 1Aspect RatioBETDLS (Stock)DLS (Media)DLS (Media 24 h)Z-avePDIZ-avePDIZ-avePDICarbon blackPrintex9014 338106.30.24129.10.20120.30.17Carbon blackXE2B30 1142179.70.22237.80.23217.50.18Reduced graphene oxide 2rGO1.7 × 1500 411547.50.45351.40.54355.40.52MWCNTMitsui-767 × 560083.626672.60.341095.50.41743.70.37MWCNTNM40167 × 404860.418697.90.331303.50.311161.10.40MWCNTNM40312 × 44336.9135198.60.42576.30.52388.50.48MWCNTNRCWE-04022 × 51923.6150180.60.39541.50.49275.30.42MWCNTNRCWE-04127 × 100537.2152172.30.36284.80.46277.10.45MWCNTNRCWE-04230 × 72324.1141190.30.42203.90.35199.60.40MWCNTNRCWE-04356 × 77113.882175.50.30695.80.59308.40.42MWCNTNRCWE-04433 × 133040.374195.60.294870.50314.00.39MWCNTNRCWE-04530 × 155351.8119197.90.27358.20.36312.80.38SWCNTNRCWE-0541 × 17,50017,500370.810250.87374.90.76270.00.53SWCNTNRCWE-0551 × 20002000453.1606.20.89452.80.84550.10.82SWCNTNRCWE-0561 × 20002000373.4502.50.873370.77458.80.86TiO2 anataseMKN-A01531 84.6177.10.38221.60.39211.20.40TiO2 anataseMKN-A10022 73.5149.80.40174.20.40179.40.38TiO2 rutileNRCWE-00110 99115.10.18131.30.17132.50.17TiO2 rutileNRCWE-02538 28.2190.50.17213.90.13222.40.16TiO2 rutileNRCWE-03011 139.1116.40.19133.40.16133.40.17Fe2O3NRCWE-01840 27.7136.40.16151.20.17154.20.17Fe2O3NRCWE-01985 × 4255.027.4253.90.17322.40.18317.50.17NiZnFe4O8NRCWE-02020 104341.90.45248.90.44248.50.45ZnFe2O4NRCWE-02123 87.7406.20.37403.00.48611.90.46NiFe2O4NRCWE-02225 86.9245.60.37246.80.39238.60.36ZnONM11085 142140.13210.40.22189.30.26SiO2DQ12225 10.1430.90.28351.90.54383.80.81Nanoclay 2Bentonite7.8 × 650 73.6284.30.45158.60.43163.10.42Nanoclay 2NanofilSE30001 × 300 2.710770.68102.20.40351.70.49Halloysite nanotubeNatural Nano61 × 2504.125.6266.30.16277.60.28244.00.65Halloysite nanotubeNatural Nano Etched27 × 1003.7128497.20.41219.30.71283.90.291 Size information obtained from [2,23,24,25,26,27,28,29,30,31,32,33,34,35]. 2 Exhibits plate-like morphology. Abbreviations: MWCNT, multi-walled carbon nanotube; SWCNT, single-walled carbon nanotube; DLS, dynamic light scattering; PDI, polydispersity index. Included nanomaterials, their characteristics, and hydrodynamic diameters. 1 Size information obtained from [2,23,24,25,26,27,28,29,30,31,32,33,34,35]. 2 Exhibits plate-like morphology. Abbreviations: MWCNT, multi-walled carbon nanotube; SWCNT, single-walled carbon nanotube; DLS, dynamic light scattering; PDI, polydispersity index. 2.2. Measurement of Endotoxin by Recombinant Factor C The endotoxin levels in all nanomaterials were assessed using PyroGene® recombinant Factor C (rFC) in vitro assay (Endozyme® II GO, bioMérieux, Marcy-l’Étoile, France). The assay uses a pre-filled GOPLATE™ containing the standard curve and positive product controls, ensuring consistent preparation across plates and reducing handling errors. According to the manufacturer, the assay provides lot-to-lot reproducibility, high endotoxin specificity (no Factor G pathway), and a substantially reduced invalid rate compared with classical LAL formats. Plate validity required a standard curve with R2 ≥ 0.99, spike-recovery between 50 and 200%, and an acceptable negative-control background. For rFC analysis, 3 to 4 mg of nanomaterial was dispersed in 6 mL endotoxin-free water using Bandelin Sonorex RK510H ultrasonic bath (BANDELIN electronic GmbH & Co. KG, Berlin, Germany) for 20 min. This step effectively releases surface-associated endotoxin into the aqueous phase before rFC detection. Extracted endotoxin was assessed in nanomaterial dispersions at the final concentrations of approximately 100 µg/mL or 10 µg/mL (only for NM110 and Bentonite). The standard curve was made using a serially diluted lipopolysaccharide (LPS, Lonza, Basel, Switzerland), and the predicted endotoxin concentration of the tested materials was calculated based on the standard curve equation and plotted on the standard curve. All results were normalized with respect to the nanoparticles’ weight. Following the current international standards for endotoxin testing, the rFC assay included inhibition-enhancement controls (IECs), which are spike-recovery controls consisting of a known concentration of endotoxin (0.5 EU/mL) spiked into the test samples to determine their ability to inhibit or enhance the assay readout. An inhibition is considered a spike-recovery < 50% of the concentration introduced, whereas an enhancement is defined as > 200%. In addition, interference was assessed with spiked-in nanomaterials in the absence of rFC enzyme to assess the potential interference with absorbance readouts. The spiked-in nanomaterial concentrations represent a worst-case scenario, assuming complete (100%) deposition and cellular uptake of the administered nanomaterial dose. Each spiked and unspiked sample was measured in duplicates. 2.3. Assessment of Microbial Contamination by Reporter Cell Assays To assess microbial contamination, the HEK-Blue™-hTLR2 and HEK-Blue™-hTLR4 reporter cell lines, together with their parental control HEK-Blue™ Null cell line (InvivoGen, Toulouse, France), were utilized. The cells are derived from human embryonic kidney (HEK293) cells and contain an NF-κB-inducible secreted embryonic alkaline phosphatase (SEAP) gene reporter construct. The HEK-Blue™-hTLR2, HEK-Blue™-hTLR4 cells overexpress TLR2 and TLR4, respectively. The cells were cultured in high-glucose DMEM (Gibco, Thermo Fisher Scientific; Waltham, MA, USA) supplemented with 10% Ultra-Low Endotoxin Fetal Bovine Serum (Biowest, Bradenton, FL, USA), containing 100 units/mL penicillin and 100 µg/mL streptomycin (Gibco), and Normocin (InvivoGen). Zeocin (InvivoGen) was added to the culture medium for HEK-Blue™ Null1 cells, while HEK-Blue™ selection antibiotics (InvivoGen) were used for HEK-Blue™-hTLR2 and HEK-Blue™-hTLR4 cells for plasmid maintenance purposes. All cells were maintained at 37 °C in a humidified incubator with 5% CO2, and experiments were performed using cells within passage 20, in accordance with InvivoGen’s recommendations. Assessment of nanomaterial contamination was performed in 96-well plate format (Sarstedt; Nümbrecht, Germany; growth area: 0.29 cm2). Cells were seeded in triplicate at a density of 280,000 cells per well in a final volume of 180 µL and incubated for 24 h. Cells were then exposed to 20 µL dispersed nanomaterials at the final concentrations of 10 µg/mL or 100 µg/mL for 24 h. For NM110 and Bentonite, only the lower concentration was used as previous data showed reduced cell viability at high concentrations [3,36]. Moreover, NRCWE-022 was assessed only at the lower concentration, based on prior internal data from the SmartNanoTox consortium (confidential, unpublished). Media and cell cultures were incubated with nanomaterials at 37 °C for 24 h and examined by light microscopy to confirm the absence of viable bacterial contamination and growth. Standard curves using positive controls included purified lipoteichoic acid (LTA) from Staphylococcus aureus (InvivoGen, tlrl-pslta) and ultrapure LPS from Escherichia coli strain O55:B5 (InvivoGen, tlrl-b5lps) to activate TLR2 and TLR4, respectively. The standard curve was made using respective positive controls, and the predicted contaminant concentration of the tested materials was calculated based on the standard curve equation and plotted on the standard curve. Endotoxin-free water was used as a negative control. Interference of the nanomaterials with the HEK-Blue™ TLR reporter cell assay was assessed using spiked controls as described by Bohmer and colleagues [37]. In short, untreated cells and positive control cells were treated with either media or LTA/LPS. 20 µL of the cell culture supernatant was carefully transferred to a new 96-well plate, and nanomaterials were added to the collected cell culture supernatants at a worst-case scenario of 100 µg/mL. For materials showing interference at the worst-case scenario, the nanomaterials were spiked in using a titration series of 100 µg/mL, 50 µg/mL, and 10 µg/mL to establish a lower, non-interfering threshold for accurate endotoxin assessment. Following the addition of nanomaterials, the media was mixed with 180 µL of QUANTI-Blue™ Solution. The plate was incubated at 37 °C for 3 h, and the SEAP levels were determined by measuring the absorbance at 649 nm using a Synergy Neo2 multi-mode reader (BioTek Instruments, Winooski, VT, USA). Interference was defined as a ±20% change in the readout signal compared to the control. The results were normalized to the weight of the analyzed nanoparticles. Each sample was measured in duplicates. The nanomaterial-free dispersion buffer (0.05% BSA in endotoxin-free water) was included as the negative control in all QUANTI-Blue measurements, ensuring that any background interaction between the buffer and the detection reagent was fully accounted for. 2.4. Data Analysis and Statistical Analysis The limit of detection (LOD) for TLR reporter assays was determined as two times the standard deviation. For the rFC assay, the LOD was determined at 0.05 EU/mL (equivalent to 0.005 ng/mL), based on the calibration data showing that a 0.005 EU/mL standard yielded the same readouts as background (unexposed controls). Measurements above the LOD were interpreted as positive. Statistical analyses were conducted using one-way ANOVA followed by Šídák’s multiple comparisons test (α = 0.05) to correct for multiple testing. Analyses were performed in GraphPad Prism (version 10.5.0.0; GraphPad Software, San Diego, CA, USA). Adjusted p-values < 0.05 were considered statistically significant. Correlation between endotoxin levels and TLR activation was assessed using Pearson correlation. 3. Results 3.1. Recombinant Factor C Assay Assessment of interference by inhibition-enhancement controls (IECs) included in the commercial assay and by additional assessment of spectral interference in the absence of enzyme showed that none of the nanomaterials interfered with the rFC assay (Table S1). Endotoxin measurement resulted in detectable endotoxin levels (LOD: 0.005 ng/mL, equal to 0.05 EU/mL) in 18 of the tested materials, more than half of the total materials tested. While the levels of endotoxin were generally low, the highest contamination values were measured for rGO (0.028 ng/mL), the single-walled carbon nanotubes (SWCNTs) (NRCWE-054: 0.027 ng/mL and NRCWE-056: 0.019 ng/mL), the SiO2 nanomaterial (DQ12: 0.026 ng/mL), and the Fe2O3 nanomaterials (NRCWE-018: 0.0194 ng/mL and NRCWE-019: 0.0145 ng/mL) (Figure 1). Nanomaterials identified as endotoxin positive in the rFC analysis. The standard curve was fitted using a four-parameter logistic (4PL) model (R2 = 0.9994) with a limit of detection (LOD) of 0.005 ng/mL. The limit of detection (LOD) is indicated by the black dashed line. Open circles indicate predicted endotoxin concentrations of tested nanomaterials. Materials listed in the frame were identified as endotoxin-positive based on responses above the LOD. 3.2. Particle Characterization by DLS To investigate the stability of the nanomaterials used in cell-based assays, DLS was assessed in the dispersion and cell culture media at 0 and 24 h (Table 1). Most nanomaterials showed relatively high dispersion stability with minor changes in Z-average values in the media over the 24 h incubation time, with PDI values around 0.4, indicating moderate aggregation. In contrast, several high-aspect-ratio CNTs (including Mitsui-7, NM401, NM403, NRCWE-040, -043, -044, and -045) had reduced Z-averages and PDIs after 24 h. The apparent decrease in Z-average and PDI after 24 h is likely an artifact of sedimentation of larger CNT agglomerates out of the DLS detection limit rather than true re-dispersion of the materials. 3.3. TLR4 Reporter Assay For the TLR4 reporter assay, significant interference was observed for all the included multi-walled carbon nanotubes (MWCNTs) as well as the carbon black materials, namely NM401, NM403, Mitsui-7, NRCWE-040, NRCWE-041, NRCWE-042, NRCWE-043, NRCWE-044, NRCWE-045, Printex90, and XE2B (Figure 2A,B). Both material types significantly increased the baseline absorbance in the absence of the receptor agonist, and most of them further enhanced the assay response upon agonist stimulation. The interference was consistent even at diluted nanomaterial concentrations (MWCNTs, Figure S1), though the higher concentration of materials exhibited stronger interference, suggesting that intrinsic optical or chemical properties of these carbon-based materials interfere with the assay measurement, potentially confounding endotoxin quantification. No interference or limited interference was observed for the remaining nanomaterial classes (Figure S2), suggesting that the TLR4 reporter assay is suitable for assessment of microbial contamination for SWCNT, metal-oxide, nanoclay, and graphene oxide materials. Interference of carbon-based nanomaterials with the TLR4 reporter assay. Interference was assessed for (A) MWCNTs and (B) carbon black materials at a concentration of 100 µg/mL. Con Null and Int Null represent negative control cells without a receptor agonist, in the absence or presence of nanomaterials, respectively. Con R4 and Int R4 represent TLR4 reporter cells stimulated with agonist, without or with nanomaterials, respectively. Data are presented as mean ± SD (n = 3). Only statistically significant changes that also exceeded a 20% difference relative to the respective control are shown in the figure. Statistical significance compared with respective controls is indicated (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). Con = Control, Int = Interference. For the TLR4 reporter assay, ten materials had activation levels above the detection limit (LOD: 0.00032 ng/mL), indicating the presence of endotoxin contamination. The highest contamination was detected in Bentonite (0.0026 ng/mL), DQ12 (0.0022 ng/mL), NanofilSE3000 (0.0012 ng/mL), NRCWE-20 (0.0012 ng/mL), and NM110 (0.001 ng/mL), while the rest of the tested materials had endotoxin values below 0.001 ng/mL (Figure 3). Notably, all the detected contaminations were below the LOD for the rFC assay (0.005 ng/mL). Nanomaterials identified as endotoxin positive in the TLR4 reporter assay (LOD: 0.00032 ng/mL). The standard curve for the TLR4 (LPS) reporter assay was fitted using a four-parameter logistic (4PL) regression model: y = 2.16/(1 + (x/0.1427)−0.81) with R2 = 0.9992. The limit of detection (LOD) is indicated by a black dashed line. Open circles indicate predicted endotoxin concentrations. Nanomaterials listed in the box were classified as TLR4 positive. 3.4. Correlation Between rFC and TLR4 Measurements Comparison of the measured rFC and TLR4 data showed that, out of the materials with the highest rFC values, DQ12 and rGO were consistently positive for endotoxin across both assays. Moreover, historical LAL data also identified rGO as one of the most contaminated materials (Table 2). The other materials with the highest rFC values showed minimal or no response in the TLR4 assay, i.e., NRCWE-054 (rFC: 0.027 ng/mL; TLR4: