Toxicological effects of the different substances in tobacco smoke on human embryonic development by a systems chemo-biology approach

Abstract

The physiological and molecular effects of tobacco smoke in adult humans and the development of cancer have been well described. In contrast, how tobacco smoke affects embryonic development remains poorly understood. Morphological studies of the fetuses of smoking pregnant women have shown various physical deformities induced by constant fetal exposure to tobacco components, especially nicotine. In addition, nicotine exposure decreases fetal body weight and bone/cartilage growth in addition to decreasing cranial diameter and tibia length. Unfortunately, the molecular pathways leading to these morphological anomalies are not completely understood. In this study, we applied interactome data mining tools and small compound interaction networks to elucidate possible molecular pathways associated with the effects of tobacco smoke components during embryonic development in pregnant female smokers. Our analysis showed a relationship between nicotine and 50 additional harmful substances involved in a variety of biological process that can cause abnormal proliferation, impaired cell differentiation, and increased oxidative stress. We also describe how nicotine can negatively affect retinoic acid signaling and cell differentiation through inhibition of retinoic acid receptors. In addition, nicotine causes a stress reaction and/or a pro-inflammatory response that inhibits the agonistic action of retinoic acid. Moreover, we show that the effect of cigarette smoke on the developing fetus could represent systemic and aggressive impacts in the short term, causing malformations during certain stages of development. Our work provides the first approach describing how different tobacco constituents affect a broad range of biological process in human embryonic development.

Figure 1. A binary network of chemical-protein and protein-protein interactions (CPI-PPI network) generated by the program Cytoscape 2.8.2.

(A) The main network, showing 49 known substances present in tobacco, 1177 nodes (49 substances, 1128 proteins) and 7522 edges (connections). Proteins were colored to identify the tissue in which they were present: (i) pink indicates fetal tissue; (ii) green, embryonic tissue; and (iii) dark blue, both fetal and embryonic tissues. In addition, each substance was colored according to its solubility: (i) yellow indicates lipophilic and (ii) light blue, hydrophilic. We observed that nicotine resided in a module apart from the major network (A). Therefore, we separated it from the major CPI-PPI network and colored its module purple. (B) The nicotine subnetwork is shown separately from the major CPI-PPI network. It contained proteins related to retinoic acid signaling and retinoic acid (lipophilic molecule). (C) The final major CPI-PPI network after the nicotine module was extracted.

Figure 2. HBs found in the major CPI-PPI network.

Betweenness and node degrees were assessed using the program CentiScaPe. Among the 143 HBs, 53 proteins are present in both tissues, reinforcing the idea of a prolonged effect of CS on embryonic development.

Figure 3. Cluster analysis of the major CPI-PPI network indicating clusters 1, 4, 11, 16 and 20.

Cluster 1 (A) is composed of 83 nodes and 565 edges, with Ci = 6,843. The associated hydrophilic constituents are urethane, N-nitrosoanabasine, N-methylpyrrolidine and pyridine. The lipophilic constituents are toluene, 4-aminobiphenyl, 5-methylcrysene, benz[a]anthracene, chrysene, benzo[b]fluoranthene, 7H-dibenzo[c,g]carbazole and 2-naphthylamine. Related GO terms: oxidation reduction and unsaturated fatty acid metabolic processes. Cluster 4 (B) is composed of 90 nodes and 411 edges, with Ci = 4,567. The associated hydrophilic compounds are urethane, N-nitrosoanabasine, N-methylpyrrolidine, arsenic, selenium and cadmium, and the lipophilic compounds are acrolein, crotonaldehyde, toluene, xylene, ethylbenzene, benz[a]anthracene, styrene and 2-naphthylamine. Related GO term: oxidation reduction. Cluster 11 (C) is composed of 23 nodes and 74 edges, with Ci = 3,217. Only the lipophilic compound isoprene is present in this cluster. Related GO term: steroid biosynthetic processes. Cluster 16 (D) is composed of 15 nodes and 36 edges, with Ci = 2,400. The associated hydrophilic compound is butyraldehyde. Related GO term: lipid modification. Cluster 20 (E) is composed of 29 nodes and 65 edges, with Ci = 2,241. The associated lipophilic compounds are acrolein, 2-naphthylamine, 1,3-butadiene, cyclopentane and 4-aminobiphenyl. Related GO terms: prostaglandin metabolic processes and unsaturated fatty acid metabolic processes. A merge of clusters 1, 4 and 20 (F). Clusters 11 and 16 did not show any proteins overlapping with any other cluster.

Figure 4. Cluster analysis of the major CPI-PPI network and the modules related to cell-cell signaling.

Cluster 2 (A) is composed of 73 nodes and 354 edges, with Ci = 4,849. Cluster 2 contains the two hydrophilic substances cadmium and methylamine. Related GO term: regulation of cell communication. Cluster 18 (B) is composed of 13 nodes and 30 edges, with Ci = 2, 304. Cluster 18 contains one hydrophilic compound, trimethylamine, and one lipophilic compound, ethylbenzene. Related GO term: cell-cell signaling.

Figure 5. A merge of clusters 3, 6, 17 and 21.

In (A), cluster 3 is composed of 14 nodes and 65 edges, with Ci = 4,643. The associated hydrophilic components are urethane, beryllium and polonium-210. Related GO terms: RNA-splicing and nucleobase, nucleoside, nucleotide and nucleic acid metabolic processes. Cluster 6 (B) is composed of 24 nodes and 102 edges, with Ci = 4,250. The associated lipophilic constituents are 1,3-butadiene and chrysene. Related GO term: nucleobase, nucleoside and nucleotide metabolic processes; cluster 17 (C) is composed of 35 nodes and 83 edges, with Ci = 2,371. The hydrophilic constituents present include catechol and arsenic. Related GO term: regulation of nucleobase, nucleoside nucleotide and nucleic acid metabolic processes. Cluster 21 (D) is composed of 22 nodes and 49 edges, with Ci = 2,227. The associated hydrophilic constituent is beryllium, and the lipophilic constituent is crotonaldehyde. Related GO term: regulation of DNA metabolic processes. The union of clusters 3, 17 and 21 (E). Cluster 6 did not show any proteins overlapping with any other cluster.

Figure 6. Subnetworks derived from the merge of clusters 5, 8 and 9.

In (A), Cluster 5 is composed of 69 nodes and 315 edges, with Ci = 4,565. The associated hydrophilic components are vinyl acetate and ethylamine. Related GO terms: response to DNA-damage stimulus and cell cycle. cluster 7 (B) is composed of 13 nodes and 54 edges with Ci = 4,154. The associated hydrophilic compound is dimethylamine. Related GO term: chromatin organization. Cluster 8 (C) is composed of 85 nodes and 338 edges with Ci = 3,976. The associated lipophilic constituents are acrolein and benzo[b]fluoranthene, whereas the hydrophilic constituents are urethane and arsenic. Related GO terms: DNA-damage stimulus and regulation of cell cycle. Cluster 9 (D) is composed of 16 nodes and 58 edges, with Ci = 3,625. The hydrophilic constituent present is trimethylamine. Related GO term: cell cycle processes. The union of clusters 5, 8 and 9 (F). Clusters 7 did not show any proteins overlapping with any other cluster.

Figure 7. A binary network of the interactions between chemical compounds and proteins generated by the program Cytoscape 2.6.3, which contained 330 proteins and 4078 connections.

Nicotine appears in the network as the green node, and RA appears as the blue node. White nodes are connected to both compounds are proteins. A) A subnetwork generated by the program Cytoscape containing 49 nodes and 281 edges and showing the proteins with direct connections with nicotine. B) A subnetwork generated by the program Cytoscape containing 130 nodes and 1,471 edges and showing the proteins that make direct connections with RA.

Figure 8. A molecular model illustrating how nicotine could potentially affects differentiation.

In the first part of the model (I), it can be observed that by generating cellular stresses, nicotine promotes the recruitment of JNK1 through the influx of intracellular Ca²⁺. JNK1, by itself, promotes the inhibition of RARα. Finally, nicotine promotes the inhibition of CYP26A1, which generates an accumulation of RA in the cell and an increase in cell proliferation. In the second part of the model (II), the inhibition of BMP2 and BSP is promoted by nicotine, which results in the negative regulation of bone mineralization and skeletal development. In addition, nicotine promotes a pro-inflammatory reaction that recruits VEGF and placental growth factors, which leads to an impairment of the endovascular trophoblast, resulting in a fetal growth restriction.

Figure 9. A model of the interactions from a systemic view showing how TCs affect development.

In (A), we show that increasing TC levels generate a pro-inflammatory cascade by increasing the levels of PTGS1 and PTGS2. PTGSs are associated with inflammatory responses and are essential for normal pregnancy. Disturbances in PTGS expression could cause impairments in fetal development. TCs are connected to ALOX5 and ALOX15B, which are proteins involved in the synthesis of leukotriene, a molecule that plays pivotal roles in pro-inflammatory responses. The consequence of (A) is low birth weight in newborn infants, abortions and increased proliferation. Moreover, in (B), TCs are linked to BING2 and USP2, which are proteins related to increased activity of MDM2. This MDM2 mediated up-regulation can rapidly down-regulate p53 protein, leaving the cell more susceptible to DNA damage. TCs also down-regulate HPRT1, diminishing purine metabolism. This system exhibits a relationship with increased proliferation. The systems in (C) shows that TCs are associated with the generation of superoxides due to up-regulating NADH oxidase, which increases ROS levels and, consequently, oxidative stress. Increased oxidative stress is known to be related to birth defects. In addition, system (C) is associated with low birth weight and low neutrophil activity. Moreover, (D) shows the relationship between TCs and low hormone synthesis and signaling. Exposure to TCs could have a negative effect on androgen and estrogen solubility due to acting on the UGT cluster. TCs could also be associated with low levels of cholesterol synthesis due to increasing the levels of CYPs and diminishing the levels of FDFT1 and FDPS, which are two enzymes related to cholesterol synthesis. Low cholesterol availability would decrease general hormone synthesis. In addition, TCs affect the transport of cholesterol to the mitochondria by acting on the membrane protein StAR. Finally, in (E), the system shows the relationships between TCs and decreased global gene expression and cellular differentiation and signaling. The activities of the TCs would increase the levels of MBD2, a methylation enzyme. DNA methylation is related to gene silencing. We postulate that TCs could affect the PRC2 complex via its methylation and disturb gene expression, including that of HOX genes. TCs could also have a negative effect on gene expression by increasing YWHAH levels, which would decrease the levels of the master kinase PDPK1 and is linked to AKT activation and SMAD nuclear translocation. In addition, NOTCH signaling could be affected through the action of TCs on APP activation.