The food dye Tartrazine disrupts vascular formation both in zebrafish larvae and in human primary endothelial cells Authors: Dinh Duy Thanh, Nguyen Bich-Ngoc, Cécile Paques, Aurélie Christian, Stéphanie Herkenne, Ingrid Struman, Marc Muller Published: 2024 ABSTRACT AbstractTartrazine (E102) is a controversial coloring agent whose potential impacts on human health are not fully understood. Our study reveals the vascular disrupting effects of tartrazine (TTZ) on developing zebrafish embryos in vivo and on human umbilical vein endothelial cells in vitro. The dye was shown to cause dose-dependent hemorrhages in zebrafish embryos. Analyzing transgenic zebrafish harboring fluorescent endothelial cells revealed that TTZ treatment disrupted cell organization into vessels in both the sub-intestinal vein and the brain area. Assays on human umbilical vein endothelial cells demonstrated that TTZ inhibited endothelial proliferation, tube formation, and migration in a dose-dependent manner. Taken together, our results indicate for the first time that TTZ can affect endothelial cell properties, possibly by disrupting Rho family GTPase pathways which control the cytoskeleton. Our finding provides a credible explanation for many reported human health impacts and offers prospective applications for biomedicine.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-024-82076-5.Keywords: Tartrazine; Blood vessels, Zebrafish; HUVEC; BMP pathways; Rho GTPasesSubject terms: Developmental biology, Angiogenesis, Risk factors Introduction Tartrazine (E102, TTZ), a lemon-yellow monoazo dye, is one of the most widely used coloring agents in food products, medicines, and cosmetics. Despite being extensively studied in recent decades, TTZ remains among the most controversial food additives, with various debates over its potential safety and toxicity1–3. The 2009 EFSA Panel on food additives set its acceptable daily intake (ADI) to 7.5 mg/kg bw based on the “no observed adverse effect level” (NOAEL) at 750 mg/kg bw/day determined in a rat chronic toxicity assay1, while the Joint FAO/WHO expert committee in 2017 approved a higher ADI at 10 mg/kg bw3. However, recent studies have challenged the safety of TTZ even at those ADI levels or lower. For instance, 30-day TTZ feeding to rats at 7.5 mg/kg bw induced liver and kidney histopathology, in addition to leucocyte DNA damage4. Another study described fetal mortality and teratogenicity following 0.45 and 4.5 mg/kg bw TTZ administration to pregnant rats on the 6th −15th days of gestation5. Similarly, neurotoxic6and haemato-immunotoxic7impacts have been reported as the results of TTZ treatment at ADI and below. The main suspect to trigger all these effects is oxidative stress resulting from TTZ metabolites produced by the gut microbiota8,9. In humans, the most controversial health aspect of TTZ is its association with attention-deficit hyperactivity disorder (ADHD). Dated back to the 70s10, that allegation once again drew public attention in 2007 thanks to the RCT study of McCann et al.11, listing TTZ among the infamous “Southampton six” – food additives whose mixture could possibly induce hyperactivity in children. Besides, immune (hypersensitivity) and non-immune (intolerance) reactions to the dye have been long reported, albeit rare3,10,12,13. One notable TTZ-induced symptom in several cases is the occurrence of purpuric spots (subcutaneous hemorrhage)13–15. While most of the studies until now were based on cell culture and rodent models, zebrafish embryo is increasingly used for toxicological investigations16. Generally considered as a non-regulated embryonic in vivo model (until 120 h old)17, this embryo can be used in complement to conventional in vitro and animal models: its transparent body allows visualization of all standard vertebrate systems and organs; its rapid development allows assessment of multiple toxicological targets and processes within a few days; more importantly, research using zebrafish embryos can provide predictive value not only for human health risk but also for ecotoxicity assessment16,18,19. Here, by combining tests on zebrafish embryos and human cells, we show that TTZ displays a disruptive effect on vascular formation, which may help shed light on the possible processes involved in some of TTZ’s adverse effects. Methods Materials The zebrafish wild type (WT) AB strain, and transgenic strains Tg(fli1:EGFP) (y1Tg)20 and Tg(kdrl:mCherry)(y171Tg)21were maintained within the Zebrafish Facility in GIGA-Research, University of Liège. Fish were reared in a Techniplast (Buguggiate, Italy) recirculating system under a 14:10-h light/dark photocycle. The isolation and maintenance of human umbilical vein endothelial cells (HUVECs) were previously described22,23. The cells were cultured in EGM2 (Lonza, Verviers, Belgium) supplemented with 5% FCS (Gibco, Thermo Fisher Scientific, Waltham, MA, USA). Tartrazine (#86310, Sigma Aldrich, Hoeilaart, Belgium) 100 g/L stock solution was prepared and diluted as appropriate in E3 (5 mM NaCl, 0.17 mM KCl, 0.4 mM CaCl2, and 0.16 mM MgSO4) for zebrafish experiments and in PBS for cell experiments. Zebrafish embryonic assays General toxicity evaluation The zebrafish embryonic toxicity (ZET) assay was adapted from the OECD Test No.236 (Fish Embryo Toxicity Test)19. After breeding, eggs were collected into E3 medium. At 3–4 h post-fertilization (hpf), fertilized and healthy embryos were selected and distributed into 6-well plates at 25 embryos/well containing 5 mL of E3 medium supplemented or not with TTZ. Embryos were raised until four days post-fertilization (4 dpf), solutions were renewed and dead embryos were removed daily. Lethality and malformation rates were recorded. Phenotypes were compared with those described previously by Kimmel et al.24 Blood vessel assessment: Transgenic embryos were obtained by crossing heterozygous Tg(fli1:EGFP) or Tg(kdrl:mCherry) with WT and exposed to 5 and 10 g/L TTZ from 3-4hpf. Visual analysis was performed at 3 dpf using a Leica (Wetzlar, Germany) M165FC fluorescent dissecting microscope. Both transgenic lines are vasculogenic reporters whose endothelial cells (ECs) are labeled with respectively green or red fluorescence. Despite displaying a brighter fluorescence signal, the Tg(fli1:EGFP) was used solely for the sub-intestinal vein (SIV) assessment due to the intense fluorescence in the pharyngeal arch cranial neural crest cells. Therefore, the pan-vasculature observation was performed on the Tg(kdrl:mCherry) line. All experiments were performed in triplicate, n = 50. Human umbilical vein endothelial cell (HUVEC) assays Human umbilical vein endothelial cells were obtained from Lonza (Verviers, Belgium) and used between passages 3 and 8. All experiments were performed in triplicate, and each included five wells per TTZ concentration (including control), each time cell viability was ensured by staining with trypan blue before seeding. Cell proliferation Cells were seeded into a 96-well plate for 48 h at 8000 cells/well. Selected concentrations of TTZ (0–1000 ug/mL) were then added, and proliferation was analyzed after 24 h using the Cell Proliferation ELISA BrdU Colorimetric kit (Roche, Mannheim, Germany) following the manufacturer’s instructions. Absorbance at 355 nm was measured on a VICTOR X3 Multilabel Reader (PerkinElmer, Waltham, USA). Tubulogenesis HUVECs were seeded onto a Matrigel-coated 96-well plate at 3000 cells/well into 100 µl of medium supplemented with selected concentrations of TTZ (0–1000 µg/ml), then incubated at 37 °C for 6 h. Tubulogenic pictures were taken under a CKX41 microscope (Olympus, Tokyo, Japan) and analyzed using ImageJ. Migration HUVECs were plated into 48-well plates and cultured to confluency in EGM2. A scratch wound was then created in the monolayers using a pipette tip. The scratched wells were washed with PBS and incubated in EBM2 0.5% DBS for 6 h. The wound size (i.e., the distance between two sides of the wound) was measured at time points 0 h and 6 h with a CKX41 microscope (Olympus, Tokyo, Japan). The difference between wound sizes at 6 h and 0 h (i.e., wound closure) was calculated and represented as relative to the respective control per replicate. TargetNet prediction The online platform TargetNet (http://targetnet.scbdd.com, accessed January 2022) was used to predict potential protein targets of TTZ using its molecular fingerprint. TTZ’s SMILES string was obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov). TargetNet parameters were set at default values (AUC ≥ 0.7; Fingerprint type ECFP4). Potential target hits were defined as having a probability score > 0. Data and statistical analysis All data analysis and graphs were performed using R 4.3.2. All scripts and raw data are available upon request from the authors. Concerning zebrafish embryo development assessment, toxicity indices (i.e. LC50 and EC50) of TTZ were estimated by fitting dose response models with two-parameter log-logistic functions using package “drc”25. The teratogenic index (TI) was then calculated by dividing the LC50 by EC50. Mixed effect models (using package “lme4”26) were fitted to analyze the observed survival and malformation rates at different treatment conditions due to the multilevel nature of the experiments (including both biological and technical replications). The average survival and malformation rates and their 95% prediction intervals (error bars) were obtained using the function predictInterval()in packages “merTools”27 and then plotted using Prism. A similar approach was also employed for ectopic sprouting data. The same analytical pipeline and R packages (“drc”, “lme4”, and “merTools”) were used for the results from HUVEC cell experiments. The IC50 of TTZ on HUVEC proliferation was determined using the same method as the ZET assay. For tube formation data, since both tube length and branching results are bounded by zero and not normally distributed, they were first log-transformed before analyzing and model fitting. Results In vivo effect of TTZ on the developing zebrafish embryos Hemorrhage as a remarkable malformation Following a preliminary range-finding experiment, wild type zebrafish larvae were exposed to TTZ concentrations ranging from 0 to 50 g/L in order to identify toxicity indices (including LC50, EC50, and TI - teratogenic index) as well as the types of malformations caused by TTZ exposure. Compared to controls, the test revealed TTZ to be non-aquatoxic with an LC5096hpf of 9.31 g/L and an EC5096hpf of 5.49 g/L (Fig. 1A) (more description below). Thus, with a TI (teratogenic index, defined as the ratio between LC50 and EC50) of 1.7, the compound can be considered slightly teratogenic for zebrafish embryonic development (Fig. 1A). ZET results for TTZ. (A) Dose-response curves showing survival and occurrence of any malformation induced by 96 h TTZ exposure, X-axis is logarithmically scaled. (B) A control embryo at 4 dpf showing normal morphology. (C) Hemorrhagic sites (arrowheads) induced by TTZ exposure. (D) Percentages of surviving embryos with different phenotypes at 4 dpf (Standard Deviations were excluded for visual clarity). (E) Severe deformities at high TTZ concentration. Compared to the control (Fig. 1B), one remarkable malformation caused by TTZ exposure was the occurrence of multiple hemorrhagic sites across the embryonic body, starting at 2 dpf. The severity of hemorrhage was dose-dependent: at lower concentrations, the sites mainly appeared around the duct of Cuvier (i.e., the common cardinal vein) region (Fig. 1C) and then expanded to other body parts at higher concentrations, including brain hemorrhage. Other notable defects included pericardial edema, yolk edema, yolk necrosis, small eye, and whole-body deformities (Fig. 1D-E). However, the percentage of TTZ-induced bleeding larvae did not follow the typical dose-response relationship (Fig. 1D, thick red line). The decrease in the percentage of larvae with specific defects (hemorrhage, small eye, yolk edema and necrosis, Fig. 1D) at high concentrations suggests that these treatments lead to multiple, more severe defects16. Also, the biphasic curves of these specific defects, and the transitions of dominant phenotypes, suggest multiple modes of action involved16. The hemorrhage phenotype, which reached 14% at 5 g/L then declined at 10 g/L, was genuinely remarkable, thus follow-up investigations were conducted. TTZ treatment of transgenic reporter fish lines revealed impairment of in vivo angiogenesis Upon discovering the TTZ-induced hemorrhagic phenotype in wild type zebrafish embryos, further experiments were conducted using transgenic reporter zebrafish lines to verify whether TTZ could affect blood vessel formation. The exposure doses were selected at 5 and 10 g/L as these concentrations produced the most hemorrhagic phenotypes without inducing too much mortality or other deformities (Fig. 1A, E). Firstly, whole vascular observation was conducted on the Tg(kdrl:mCherry) embryos at 3 dpf. The results revealed that TTZ exposure could disrupt endothelial cell (EC) organization into the vascular system to various extents in the larvae. The effect was more evident at 10 g/L TTZ, where many larvae had impaired branching and segmentation in both the cephalic vasculature and the caudal vein plexus (CVP) (Fig. 2A). Notably, the dorsal lateral anastomosis vessels (DLAV) and intersegmental vessels (ISV) are only marginally affected. Vascularization analysis on transgenic fluorescent zebrafish lines. (A) Cephalic vessel and caudal venous plexus vascularization are dose-dependently affected by TTZ; Vessels appear to be increasingly merged/clumped together instead of forming distinguishable fine structures; Notably , the three-longitudinal-vessel pattern in the CVP gradually merged (5 g/L) into a two-vessel pattern (10 g/L). (B) Normal (basket-shaped) and ectopically sprouted (arrowheads) SIV formation. (C) Percentages of ectopic sprouting at different TTZ concentrations. Next, to quantify the inhibitory effect of TTZ on the zebrafish blood vessel formation, further experiments were conducted with emphasis on SIV formation18,28 using the Tg(fli1:GFP). At 3 dpf, a normally developed SIV forms a smooth, basket-shaped structure on each side of the embryonic yolk (Fig. 2B). In our experiments, TTZ exposure at 5 and 10 g/L respectively induced a smaller basket size and abnormal SIV ectopic sprouting in 50% and 14% of 3-dpf embryos (Fig. 2C) – following the same trend of hemorrhage occurrences at 4-dpf larvae (Fig. 1E) but at higher percentage. TTZ inhibits HUVEC tubulogenesis, migration, and proliferation After confirming the vascular disrupting potential of TTZ on in vivo zebrafish embryos, the subsequent step is to verify if it was due to the direct effect of the food dye on ECs or possibly another indirect mechanism. Furthermore, to verify whether these effects would be observed in human cells, several in vitro experiments were carried out using primary Human Umbilical Vascular Endothelial Cells (HUVECs). BrdU incorporation assays revealed that TTZ could inhibit HUVEC proliferation with an IC50 of 30 mg/L (56 µM) (Fig. 3A). Tubulogenesis assays showed that the compound also inhibited HUVEC tube formation (Fig. 3B-D). Practically, tubes and branches stopped forming at 267 mg/L (500 µM) (Fig. 3C-D). Additionally, wound healing assays (Fig. 3E-F) revealed that HUVEC cell migration was substantially inhibited by TTZ. All tested effects of TTZ on HUVEC activities (proliferation, tubulogenesis, and migration) were dose-dependent, while statistical analyses indicated that the observed inhibition was significant even at the lowest tested dose (10.7 mg/L or 20 µM, p < 0.005). TTZ inhibits HUVEC angiogenic activities in a dose-dependent manner. (A) TTZ inhibits HUVEC proliferation in the BrdU assay. (B-D) TTZ treatment leads to reduced tube extension and branching during tubulogenesis. (E-F) Wound healing assay showing anti-migratory effect of TTZ. Concentrations are log-scaled , and biological replicates are represented using different colors. Asterisks denote statistical significance: *p < 0.01; **p < 0.001. Discussion TTZ affects blood vessel formation, possibly through interaction with the BMP signaling pathway Throughout decades of research on the food dye tartrazine with numerous investigations, this is the first study reporting that TTZ has vascular disrupting properties. We have shown that the dye can impair angiogenesis by disrupting blood vessel formation in developing zebrafish embryos and by inhibiting human primary ECs from proliferating, migrating, and forming tubular networks using an array of simple experiments. Combining tests using zebrafish embryos and mammalian ECs has previously been used as a cost-effective tool in studying angiogenic factors29–32, here it helped reveal that the vascular disrupting effect of TTZ is universally conserved across multiple species, possibly pan-vertebrate. In zebrafish, the formation of both the DLAV and the ISV are mainly regulated by the vascular endothelial growth factor (VEGF) signaling pathway (Vegf-A and Vegf-C)33. On the other hand, SIV development is dependent on both the bone morphogenic protein (BMP) and the VEGF pathway33–36, while brain vascularization37and EC migration during CVP formation require mainly BMP signaling34,38. Thus, the effects of TTZ observed on zebrafish embryos predominantly on SIV, CVP, and cranial vessels, while DLAV and ISV are unaffected (Table 1), strongly suggest that this compound may affect vascularization by interacting with the BMP pathway, rather than the more popular VEGF-A signaling. Table 1Summary of TTZ’s angiogenic effects on zebrafish embryos and HUVECs.Zebrafish embryosHUVECs Hemorrhage SIV CVP Cephalic vessels DLAV ISV Proliferation Migration Tube formation YesEctopic sproutingUndeveloped networkN/AN/AInhibitedPartially impairedInhibited Summary of TTZ’s angiogenic effects on zebrafish embryos and HUVECs. Rho GTPases as the main suspect The action of BMP signaling in zebrafish blood vessel formation is closely linked with activities of the Rho GTPase family, which regulates cell migration by controlling the formation of cytoskeleton processes32,39–41. Wakayama et al. (2015) reported BMP action mediated by Cdc42, a Rho GTPase family member whose inhibition leads to a fused CVP network without clear spaces between capillaries34. Cdc42 has been shown to control EC directional sensing through projection of filopodia39, whose disruption could lead to inhibition of the wound healing process31. It also regulates microtubule alignment and EC elongation29,42; hence its loss-of-function would facilitate EC lumen collapse, which, together with the impaired migration, could further contribute to tubulogenic failure. Another family member is RAC1, which plays multiple roles in endothelial cell activities and vascular development43, contributing to EC survival41, migration39, and zebrafish brain vascular pruning44. Interestingly, human RAC1 protein was predicted to interact with TTZ by the TargetNet platform (Supplementary Table S1). The third family member RHOA also controls EC directional migration45,46and tube collapse47. A recent paper also suggested key roles of Rac1 and RhoA in the zebrafish BMP angiogenic pathways32. While no direct relationship between Rho GTPases and zebrafish SIV ectopic formation was found in the literature, indirect evidence exists, such as via elements of the lipid metabolism pathways48– for example, treatment with the cholesterol-reducing drug atorvastatin led to both SIV ectopic sprouting in zebrafish49and an increase of GTP loading of all Rho GTPases in HUVECs50. Another indirect evidence is the family of cerebral cavernous malformations genes (CCMs), whose pathways regulate Rho GTPases in ECs51,52. Loss-of-function of CCM orthologs in zebrafish causes ectopic SIV sprouting and fused CVP53,54, and induce diverse effects on human EC proliferation, migration, and tube formation depending on both EC type (HUVEC or brain EC)55and the CCM gene involved (CCM1, 2, or 3)52,55,56. Noteworthily, variants of the CCM genes are also responsible for many cases of brain hemorrhage51,52,54,55. On the other hand, TTZ treatment in vitro (in human hepatoma cells)57and in vivo (in CD1 mice)58led to changes in expression of several genes participating in RhoA GTPase signaling pathways: JUN, MAPK8 (JNK1)59,60, MAP2K6 (MKK6)60,61, and PIK3C2A62,63. Re-analyzing the transcriptomic changes following TTZ treatment in human hepatoma cells57 further revealed regulation of genes related to Rho GTPase-activated formins, Rho GTPase effectors, and signaling by Rho GTPases. In addition to RAC1, TargetNet also suggests other Rho GTPase signaling-related proteins capable of interacting with TTZ (Supplementary Table S1). They include HDAC4, which could antagonize KLF2-induced RHOA activation and promote random cell motility64, or PTGS2, which is downstream of RHOA/RAC1/CDC42 in endothelia65and fibroblasts66. These observations further support the possible interactions between TTZ and Rho family GTPase activities. Other possible interactions Aside from the possibility of interacting with Rho GTPases, TTZ may act via multiple pathways, as the compound was reported to have both estrogenic67and oxidative68potentials – both of which are related to the angiogenesis pathway69,70and GTPase activities69,71. Indeed, our ZET results corroborate the involvement of multiple toxicological mechanisms beside vascularization (Fig. 1E). Still, further investigations are required to comprehensively understand TTZ’s vascular disruptive effect and other reported bioactivities. In this study, one striking observation is the difference in effective concentrations of TTZ in the two systems: in zebrafish larvae, substantial effects on vascular development are seen at 5–10 g/L, while HUVECs display decreased activities at 10.7 mg/L (20 µM) and completely malfunction at 267 mg/L (0.5 mM). It is important to note here that TTZ is a hydrophilic compound, dissolved in water, thus the chorionic or other cell membranes may protect the larvae from its influence, while HUVECs are directly exposed to it. In addition, the detoxifying organs (i.e., liver and kidney) may be involved, as well as the affinity of serum albumin to TTZ1,3. While the exact mechanism of TTZ’s vascular disrupting feature is yet to be determined, our discovery may already offer several applied implications. Tartrazine is a common, yet controversial food additive that is alleged to cause multiple adverse effects, including genotoxicity, oxidative stress, and behavioral alteration1,2,4,6–10,12,13,15,67,68,72. However, no report has mentioned its effect on the vascular system or endothelium, even though hemorrhagic sites can be observed in previous studies73,74. This newly found anti-angiogenic feature may shed light on the presence of purpura experienced by some individuals14,15, which has long been attributed to an immune response12,14,15. Our discovery may offer a straightforward explanation for these cases: ECs exposed to TTZ will malfunction, leading to vascular defect and internal hemorrhage. The reason why TTZ-induced purpura rarely happens could be explained by either a high individual sensitivity to its effects, or the fact that more often TTZ does not reach an effective concentration within the blood vessels: less than 5% of TTZ is absorbed after oral uptake, then predominantly excreted via urine1, whereas a large proportion of the chemical may have been metabolized by the gut microflora2. Regarding the emerging recognition of gut microbiota in food allergy cases75, it is reasonable to hypothesize a link between the microbiota and TTZ hypersensitivity. Another health aspect of TTZ, perhaps the most controversial, is its alleged link with ADHD in children3,11. If the link is valid, one possible explanation is the indirect impact of TTZ consumption on brain function via blood vessel function76. The discovered vascular effect of TTZ may also contribute to the safety regulation of the colorant, especially in food and medical applications. Methods have been published using TTZ as a photoabsorber for light-based hydrogel bioprinting to be used as scaffolds for endothelial cell seeding and tissue engineering77–80. Even though there is a washing step between scaffold printing and cell seeding, the risk of TTZ residues affecting HUVECs behaviors should not be neglected. More importantly, TTZ as a common drug excipient and food additive may interfere with proangiogenic therapy such as in coronary and peripheral artery disease or ischemic stroke81. On the other hand, this TTZ’s function could lead to potential applications in treating conditions such as cancer or macular degeneration28,81,82. Given the extensive information on TTZ toxicity and safety, its status as a food additive, and its abundant and cheap production, the chemical may serve as an affordable alternative for existing anti-angiogenic drugs. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements The authors would like to thank the GIGA zebrafish facility for providing zebrafish adults for spawning and the GIGA imaging platform for their help and support with microscopy. Author contributions D.D.T. and M.M. conceived the study; D.D.T. carried out the zebrafish experiments; N.B.N. performed data analysis and visualization; C.P., A.C., S.H., and I.S. carried out the HUVEC experiments; D.D.T., M.M., N.B.N., S.H., and I.S. contributed to writing the paper. Data availability All scripts and raw data are available upon request from the authors - namely M.M., D.D.T., and N.B.N. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement All scripts and raw data are available upon request from the authors - namely M.M., D.D.T., and N.B.N.