Integrating proteomic data with metabolic modeling provides insight into key pathways of Bordetella pertussis biofilms

Abstract

Pertussis, commonly known as whooping cough is a severe respiratory disease caused by the bacterium, Bordetella pertussis. Despite widespread vaccination, pertussis resurgence has been observed globally. The development of the current acellular vaccine (ACV) has been based on planktonic studies. However, recent studies have shown that B. pertussis readily forms biofilms. A better understanding of B. pertussis biofilms is important for developing novel vaccines that can target all aspects of B. pertussis infection. This study compared the proteomic expression of biofilm and planktonic B. pertussis cells to identify key changes between the conditions. Major differences were identified in virulence factors including an upregulation of toxins (adenylate cyclase toxin and dermonecrotic toxin) and downregulation of pertactin and type III secretion system proteins in biofilm cells. To further dissect metabolic pathways that are altered during the biofilm lifestyle, the proteomic data was then incorporated into a genome scale metabolic model using the Integrative Metabolic Analysis Tool (iMAT). The generated models predicted that planktonic cells utilised the glyoxylate shunt while biofilm cells completed the full tricarboxylic acid cycle. Differences in processing aspartate, arginine and alanine were identified as well as unique export of valine out of biofilm cells which may have a role in inter-bacterial communication and regulation. Finally, increased polyhydroxybutyrate accumulation and superoxide dismutase activity in biofilm cells may contribute to increased persistence during infection. Taken together, this study modeled major proteomic and metabolic changes that occur in biofilm cells which helps lay the groundwork for further understanding B. pertussis pathogenesis.

Keywords: Bordetella pertussis; infectious disease; label free quantification (LFQ); mass spectrometry; metabolic model; proteomics; whooping cough.

Figure 1

Experimental design for iMAT model generation. Proteins were extracted from planktonic and biofilm B. pertussis cells. The expression values were then used to generate context specific metabolic models based on the extensively curated iBP1870 B. pertussis model (322). Flux balance analysis was run on the models and the altered reactions compared between the planktonic and biofilm models.

Figure 2

Volcano plot of total protein expression changes between biofilm and planktonic cells. (A) Expression profile of all proteins identified through the LC–MS/MS analysis of planktonic and biofilm B. pertussis cells. Proteins are plotted on a volcano plot displaying the -log(q-value) on the y-axis and log[fold change (biofilm/planktonic)] on the x-axis. The dashed vertical gray lines mark a fold change of 0.8 and 1.2 and the horizontal line marks the threshold of q-value = 0.05. Highlighted in red are the proteins that were identified in all 6 biological replicates that were incorporated into the iMAT models. (B) Expression profile of the subset of proteins incorporated into the iMAT model. Red markers are proteins that were designated significantly differentially downregulated, and the blue markers are proteins that were significantly upregulated in biofilm cells. The dashed vertical gray lines mark a fold change of 0.8 and 1.2 and the horizontal line marks the threshold of q-value = 0.05. Proteins of interest are labeled.

Figure 3

Proteins up and downregulated in B. pertussis biofilm cells compared to planktonic cells identified using LC–MS/MS. Proteins significantly up and downregulated in biofilm cells were categorised in functional categories based on Bart et al. (2014). Red and blue bars represent the total number of proteins within the functional category significantly up or downregulated, respectively. Asterisk (*) denotes functional categories significantly up or downregulated based on Fisher’s exact test with Benjamini-Hochberg multiple test correction (adjusted p < 0.05).

Figure 4

Comparison of reactions of B. pertussis planktonic and biofilm iMAT models. (A) Venn diagram highlighting common and unique reactions between planktonic and biofilm iMAT models generated by incorporating proteomic expression data. (B) Planktonic and biofilm model reactions grouped into subsystems based on the Escherichia coli iAF1260 metabolic model. Unique reactions identified in one model but not the other are highlighted in different colors on the graph.

Figure 5

Summarized pathways with major changes between planktonic and biofilm iMAT models. Major changed core metabolic reactions from a comparison of iMAT models generated from biofilm and planktonic B. pertussis protein expression data. The model includes the reactions that were unique in either model as well as reactions with altered flux. Red lines indicate reactions that had decreased flux in biofilm cells while green lines indicate reactions with increased flux. The dashed lines indicate unique reactions to each model while the black lines are common reaction with the same flux. Highlighted sections are (A) the Tricarboxylic acid cycle, (B) Arginine metabolism, (C) Aspartate metabolism and (D) Gluconeogenesis and glycerophospholipid metabolism. Pathways for (A–C) are represented in more detail in Figures 6–8, respectively. Ac, Acetate; Ac-CoA, Acetyl-CoA; Akg, α-ketoglutarate; Ala, Alanine; Arg, Arginine; Argos, Argininosuccinate; Asp, Aspartate; BCAA, Branched chain amino acid degradation; Cbp, Carbamoyl phosphate; Cgly, Cysteinylglycine; Cit, Citrate; Citr, Citruline; CO2, Carbon dioxide; Cys, Cysteine; Dhap, Dihydroxyacetone phosphate; Fum, Fumarate; Gln, Glutamine; Glu, Glutamate; Glx, Glyoxylate; Gly, Glycine; Glyc3p, Glycerol 3-phosphate; Glyc, Glycerol; GNG, Gluconeogenesis; GPL, Glycerophospholipid metabolism; Gthrd, Glutathione; Hom, Homoserine; Icit, Isocitrate; Lac, Lactate; Mal, Malate; O2, Oxygen; OAA, Oxaloacetate; Orn, Ornithine; Orot, Orotate; Phb, Polyhydroxybutyrate; Prop, Propanoate metabolism; Ptrc, Putrescine; Pyr, Pyruvate; SO2, Sulfur dioxide; Suc, Succinate; SucCoA, Succinyl-CoA; TCA, Tricarboxcylic acid cycle; Thr, Threonine; Val, Valine; 2obut, 2-oxobutanoate.

Figure 6

Tricarboxylic acid cycle pathways and flux bounds between planktonic and biofilm iMAT models. This figure relates to Figure 5A. Reactions within the tricarboxylic acid (TCA) cycle between planktonic and biofilm B. pertussis iMAT metabolic models generated from proteomic expression data. Reaction names and flux values from the flux balance analysis (FBA) are given for each reaction within the TCA. Each arrow is indicates a reaction and flux values are labeled as (P) representing planktonic model flux and (B) representing biofilm model flux. All flux values are mmol · g_DCW⁻¹ · h⁻¹. Green arrows represent reactions that are upregulated in biofilm cells while red arrows are downregulated reactions. Green dashed arrows are unique reactions to the biofilm model while red dashed arrows are reactions unique to the planktonic model. Black arrows are common reactions. General metabolic pathways are in bold. Inset Range for flux variance analysis and FBA values for the tricarboxylic acid cycle between planktonic and biofilm cells. Flux ranges (min to max) are represented as lines and the FBA value is annotated with a point. Planktonic values are indicated in red and biofilm values in blue. *R_SUCOAS is a reaction that flows from succinate to succinyl-CoA. The reaction runs in reverse in the typical TCA cycle and so these values are negative. For simplicity, these reactions have been annotated as absolute values in the figure. Furthermore, the flux bounds for the planktonic models are at the max value and the max has been omitted from the graph. R_CS, type II citrate synthase; R_ACONTa, citrate hydrolase; R_ACONTb, aconitate hydratase; R_ICDHyr, isocitrate dehydrogenase; R_AKGDH, α-ketoglutarate dehydrogenase; R_SUCOAS, succinyl-CoA synthetase; R_SUCDi, succinate dehydrogenase; R_FUM, fumarate hydratase; R_MDH, malate dehydrogenase; R_ICL, isocitrate lyase; R_MALS, malate synthase.

Figure 7

Arginine metabolism pathways in B. pertussis biofilm and planktonic iMAT models. This figure relates to Figure 5B. Models were generated using proteomic expression data. Green arrows represent reactions that are upregulated in biofilm cells while red arrows are downregulated reactions. Green dashed arrows are unique reactions to the biofilm model while red dashed arrows are reactions unique to the planktonic model. Black arrows are common reactions. General metabolic pathways are in bold. ⍺-ketoglutarate (Akg) and glutamate (Glu) are utilised in many of the reactions and are therefore included multiple times in abbreviated form. N-acetyl-L-glutamate is also included as the abbreviation Acglu for the reaction R_ORNTA. Key sections of the pathways have also been annotated. (A) Both models generate carbamoyl phosphate through the same reactions. (B) The biofilm model has unique reactions to synthesize orotate from carbamoyl phosphate and export it out the cell. (C) The planktonic model completed the arginine biosynthesis pathway and exports urea out of the cell. (D) In addition to the urea, the arginine biosynthesis pathway leads to the production of fumarate which is fed back into the TCA. (E) The pathways from N-acetyl-L-glutamate to N-acetyl-L-glutamate 5 semialdehyde are unique to the biofilm model. (F) The reaction, acetylornithine transaminase reaction (R_ACOTA), runs in opposite directions for planktonic and biofilm models. (G) The reaction, glutamate N-acetyl transferase (R_ORNTA), also runs in opposite directions for biofilm and planktonic models.

Figure 8

Aspartate metabolism pathways in B. pertussis biofilm and planktonic iMAT models. This figure relates to Figure 5C. Models were generated using proteomic expression data. Red arrows represent downregulated reactions in the biofilm model. Green dashed arrows are unique reactions to the biofilm model while red dashed arrows are reactions unique to the planktonic model. Black arrows are common reactions. General metabolic pathways are in bold. Key sections of the pathways have also been annotated. (A) There is a decreased flux from oxaloacetate to L-aspartate in the biofilm model. (B) In both models, 2-oxobutanoate is eventually degraded through the branched chain amino acid degradation pathway specifically, the isoleucine pathway. (C) While biofilm cells convert homoserine through to O-phospho-L-homoserine, planktonic cells convert homoserine to O-acetyl-L-homoserine by utilizing acetyl-CoA.