Fig 1.
PPI quality control pipeline for LRRK2 interactome construction.
Human PPI data downloaded from PINOT, HIPPIE and MIST databases were merged after ID conversion using HGNC gene symbols. Merged data underwent interaction detection method reassignment using an in-house dictionary. Publication score (PS) was defined as the number of papers in which a PPI was reported, while method score (MS) was defined as the number of different methods by which a PPI was detected. Final score (FS) was calculated as PS + MS. PPIs with FS ≤ 2 were excluded from further analysis.
Fig 2.
The general LRRK2 interactome.
A) Nodes (n = 407) represent LRRK2 interactors that were with FS > 2 and were reviewed by the UniprotKB. Node fill colour and node size are weighted on the final score (FS). Larger size and darker colour indicate higher FS. B) Family classification of LRRK2 interactors (of note 176 LRRK2 interactors were not classified into families).
Fig 3.
Differential Expression Analysis (DEA) on the LRRK2int.
A) DEA was performed to compare the expression levels of each LRRK2 interactor across different tissues. Tissues were ranked based on the significant comparison results. The bar graph shows the distribution of the ranks of different tissues. Each bar in the graph represents a tissue, and segments in the bar represent ranks of that tissue at three levels: AVERAGE (≤ 4); MIDDLE (between 5 and 9); HIGH (≥ 10). Of note, an AVERAGE rank suggests that for interactor I, the expression level in a given tissue is lower/not significantly higher than in other tissues, or only higher than in ≤ 4 tissues (< 27% of all tissues), while a HIGH rank means the expression level of interactor I is significantly higher in a given tissue than in ≥ 10 other tissues (> 67% of all tissues). B) The network graph shows the LRRK2 interactors with significantly high expression in certain tissues (tissue ranks ≥ 12), suggesting that the expression levels of these interactors in a specific tissue are significantly higher than in ≥ 12 other tissues (86% of all tissues). Tissues are represented as rectangular nodes, while interactors are represented as round nodes. Different colour indicates different tissues. C) Heatmap_DEA was generated from normalised read counts (log2 transformed) of LRRK2 interactors in different tissues derived from DEA. Darker colour represents higher expression levels. The horizonal dendrogram of Heatmap_DEA was extracted as Den_DEA1. It shows the hierarchical clustering of tissues in which the LRRK2int exhibits similar expression patterns. The vertical dendrogram of Heatmap_DEA was extracted as Den_DEA2. It shows the hierarchical clustering of LRRK2 interactors based on the similarity of their expression figures across different tissues. Den_DEA2 was cut to generate 4 clusters of LRRK2 interactors (Cluster 1–4, marked in green, red, blue and yellow, respectively). The cluster containing LRRK2 (marked in red) is defined as DEA_ClusterLRRK2, in which the interactors presented similar overall expression distribution across tissues as LRRK2. Abbreviations: ACC: Anterior Cingulate Cortex; AMYG: Amygdala; CAU: caudate; CR: cerebellum; FC: frontal cortex; HP: hippocampus; HYPT: hypothalamus; NAc: nucleus accumbens; PUT: putamen; SN: substantia nigra; SPC: spinal cord c-1; Kidney_c: kidney cortex.
Table 1.
Top terms in the GO:BP functional enrichment of the DEA ClusterLRRK2.
Fig 4.
Co-expression analysis on the LRRK2int.
A) Pair-wise Tukey’s test was performed to compare the co-expression coefficients (interactors vs LRRK2) across different tissues. Tissues were ranked according to the results. The bar graph shows that putamen, nucleus accumbens, caudate and hypothalamus are tissues with the highest ranks. Liver presents a rank of 0, meaning the co-expression coefficients of LRRK2 interactors are the lowest in comparison with any other tissues analysed. B) The heatmap was generated from the coefficient matrix (2^coefficient = 2 raised to the power of the coefficient) derived from the co-expression analysis (Heatmap_Co-ex). Darker colour represents higher co-expression coefficient. The horizonal dendrogram of Heatmap_Co-ex was extracted as Den_Co-ex1, which shows the hierarchical clustering of tissues in which the LRRK2 interactors exhibited similar co-expression patterns with LRRK2. The vertical dendrogram of Heatmap_Co-ex was extracted as Den_Co-ex2, which shows the hierarchical clustering of interactors based on the similarity of their co-expression figures with LRRK2 across different tissues. Den_Co-ex2 was cut to generate 6 clusters of LRRK2 interactors (Cluster A-F, marked in green, blue, yellow, red, purple and turquoise, respectively). Interactors in Cluster D presents the highest level of overall co-expression behaviour with LRRK2 across different tissues (referred as Co-ex_ClusterLRRK2). Abbreviations: ACC: Anterior Cingulate Cortex; AMYG: Amygdala; CAU: caudate; CR: cerebellum; FC: frontal cortex; HP: hippocampus; HYPT: hypothalamus; NAc: nucleus accumbens; PUT: putamen; SN: substantia nigra; SPC: spinal cord c-1; Kidney_c: kidney cortex.
Table 2.
Top terms in the GO:BP functional enrichment of the Co-ex_ClusterLRRK2.
Fig 5.
A) The network graphs represent the LRRK2ints with nodes color-coded for DEA ranks and dimension-coded based on the co-expression coefficients. The darker the colour, the higher the expression level of the interactor in a given tissue. The larger the node, the higher the co-expression coefficient calculated between LRRK2 and the interactor in a given tissue. B) For each of the 4 brain regions the LRRK2 interactors with a DEA rank > 8 were extracted and the distribution of the DEA ranks was visualized. C) For each of the 4 brain regions the LRRK2 interactors with a co-expression coefficient ≥ 0.7 were extracted and the distribution of the co-expression coefficients was visualized. D) Functional enrichment results for the tissue-specific LRRK2 interactors; GO:BPs were filtered for term size ≤ 80 to keep only specific terms and GO:BPs were grouped based on semantic similarity. The semantic class is reported on the x-axis, the p-value of the most significant term in the semantic class is reported on the y-axis, the dimension of the circle represent the number of GO:BPs terms grouped in the semantic class. For simplicity, the semantic classes containing only 1 GO:BP term were omitted from the graph. -log10(p-value) = p-value in logarithmic scale, base 10, reverse order.
Fig 6.
Functional roles of Rab interactors of LRRK2.
The heatmap shows the functional groups that included the Rab proteins presented in the LRRK2int. Blue squares represent the presence of a certain Rab interactor in a given functional group identified in the functional enrichment analysis for the general LRRK2int.
Fig 7.
Tissue specific LRRK2:Rab interactomes (LRRK2ints_Rab).
The network graphs show the different expression attributes of the 12 Rab proteins in the general LRRK2int. Nodes represent Rab interactors of LRRK2 (in grey is represented the scheme with the order of the individual Rab proteins in the graphs). Node colour represents the DEA rank scores of Rab proteins. The darker the colour, the higher the DEA rank of a certain Rab interactor of LRRK2 for a given tissue. Node size represents co-expression coefficients. The larger the node, the higher the co-expression coefficient between LRRK2 and the Rab protein in a given tissue.