Multi-omics after O-GlcNAc alteration identified cellular processes promoting aneuploidy after loss of O-GlcNAc transferase

Abstract

Objective: Pharmacologic or genetic manipulation of O-GlcNAcylation, an intracellular, single sugar post-translational modification, are difficult to interpret due to the pleotropic nature of O-GlcNAc and the vast signaling pathways it regulates.

Method: To address the pleotropic nature of O-GlcNAc, we employed either OGT (O-GlcNAc transferase), OGA (O-GlcNAcase) liver knockouts, or pharmacological inhibition of OGA coupled with multi-Omics analysis and bioinformatics.

Results: We identified numerous genes, proteins, phospho-proteins, or metabolites that were either inversely or equivalently changed between conditions. Moreover, we identified pathways in OGT knockout samples associated with increased aneuploidy. To test and validate these pathways, we induced liver growth in OGT knockouts by partial hepatectomy. OGT knockout livers showed a robust aneuploidy phenotype with disruptions in mitosis, nutrient sensing, protein metabolism/amino acid metabolism, stress response, and HIPPO signaling demonstrating how OGT is essential in controlling aneuploidy pathways.

Conclusion: These data show how a multi-Omics platform can disentangle the pleotropic nature of O-GlcNAc to discern how OGT fine-tunes multiple cellular pathways involved in aneuploidy.

Keywords: Aneuploidy; Aurora kinase B; Cell cycle; O-GlcNAc; OGA; OGT; mTOR.

Figure 1

Multi-omic based approach to understand the effects of O-GlcNAcylation on liver: A: Schematic of multi-omic approach, floxed-OGT or OGA mice livers were treated with AAV-cre via IP injection; alternatively, mice were treated with OGA inhibitor TMG via IP injection for 1 or 2 weeks. At 1 or 2 weeks, animals were sacrificed and livers harvest. Liver samples were used for transcriptomics, proteomics, phospho-proteomic, and metabolomics analysis. Bioinformatics was performed with the AMEND algorithm to integrate the data longitudinally or heterogeneously (Figure designed in Biorender). B: Western blot analysis for O-GlcNAcylation, OGT, and OGA expression from OGT knockout livers at 1 and 2 weeks post-knockout. Actin is used as a load control (n = 3). C–E: Densitometry of O-GlcNAc, OGA, and OGT western blots. ∗ = p value less than 0.05. F: Western blot analysis for O-GlcNAcylation, OGT, and OGA expression from OGA knockout livers at 1 and 2 weeks post-knockout. Actin is used as a load control (n = 3). G–I: Densitometry of O-GlcNAc, OGA, and OGT western blots. ∗ = p value less than 0.05. J: Western blot analysis for O-GlcNAcylation, OGT, and OGA expression from mice at 1 and 2 weeks TMG treatment. Actin is used as a load control (n = 3). K–M: Densitometry of O-GlcNAc, OGA, and OGT western blots. ∗ = p value less than 0.05.

Figure 2

Identification of transcriptomic subnetworks using AMEND and Overrepresentation Analysis (ORA) pathways from longitudinal sampling: “T” = transcriptomic, “+” = AMEND selects for positive ECI nodes (equivalent change), “-” = AMEND selects for negative ECI nodes (inverse change). Clusters within each subnetwork are colored corresponding to the pathway associated with the nodes in that cluster. Treatment 1 and treatment 2 correspond to the first and second treatment listed in the panel description, respectively. Arrows in the ORA networks show the direction of nestedness (i.e., the source node is nested within the target node) within the directed acyclic graph and the sizes of the pathways are represented by the relative sizes of the nodes. The color of the node reflects the AMEND protein interaction network. Only the top 15 pathways, ranked by adjusted p-value, are shown. A: AMEND module and ORA pathways, OGT-KO 1 week vs. 2 week (T) (+); B: AMEND module and ORA pathways, OGT-KO 1 week vs. 2 week (T) (−); C: AMEND module and ORA pathways, OGA-KO 1 week vs. 2 week (T) (+); D: AMEND module and ORA pathways; OGA-KO 1 week vs. 2 week (T) (−). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Figure 3

Identification of transcriptomic subnetworks using AMEND and Overrepresentation Analysis (ORA) pathways from heterogeneous sampling: “T” = transcriptomic, “+” = AMEND selects for positive ECI nodes (equivalent change), “-” = AMEND selects for negative ECI nodes (inverse change). Clusters within each subnetwork are colored corresponding to the pathway associated with the nodes in that cluster. Treatment 1 and treatment 2 correspond to the first and second treatment listed in the panel description, respectively. Arrows in the ORA networks show the direction of nestedness (i.e., the source node is nested within the target node) within the directed acyclic graph and the sizes of the pathways are represented by the relative sizes of the nodes. The color of the node reflects the AMEND protein interaction network. Only the top 15 pathways, ranked by adjusted p-value, are shown. A: AMEND module and ORA pathways, OGT-KO vs. OGA-KO at 2 weeks (T) (+); B: AMEND module and ORA pathways, OGT-KO vs. TMG at 2 weeks (T) (−); C: AMEND module and ORA pathways, OGT-KO vs. TMG at 2 weeks (T) (+); D: AMEND module and ORA pathways, OGT-KO vs. TMG at 2 weeks (T) (−). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Figure 4

Identification of proteomic subnetworks using AMEND and Overrepresentation Analysis (ORA) pathways from OGT KO longitudinal sampling and OGT-OGA KO heterogeneous sampling: “P” = proteomic, “+” = AMEND selects for positive ECI nodes (equivalent change), “-” = AMEND selects for negative ECI nodes (inverse change). Clusters within each subnetwork are colored corresponding to the pathway associated with the nodes in that cluster. Treatment 1 and treatment 2 correspond to the first and second treatment listed in the panel description, respectively. Arrows in the ORA networks show the direction of nestedness within the directed acyclic graph and the sizes of the pathways are represented by the relative sizes of the nodes. The color of the node reflects the AMEND protein interaction network. Only the top 15 pathways, ranked by adjusted p-value, are shown. A: AMEND module and ORA pathways, OGT-KO 1 week vs. 2 week (P) (+); B: AMEND module and ORA pathways, OGT-KO 1 week vs 2 weeks (P) (−); C: AMEND module and ORA pathways, OGT-KO vs OGA-KO at 2 weeks (P) (+); D: AMEND module and ORA pathways, OGT-KO vs OGA-KO at 2 weeks (P) (−). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Figure 5

Identification of proteomic subnetworks using AMEND and Overrepresentation Analysis (ORA) pathways from heterogeneous sampling: “P” = proteomic, “+” = AMEND selects for positive ECI nodes (equivalent change), “-” = AMEND selects for negative ECI nodes (inverse change). Clusters within each subnetwork are colored corresponding to the pathway associated with the nodes in that cluster. Treatment 1 and treatment 2 correspond to the first and second treatment listed in the panel description, respectively. Arrows in the ORA networks show the direction of nestedness within the directed acyclic graph and the sizes of the pathways are represented by the relative sizes of the nodes. The color of the node reflects the AMEND protein interaction network. Only the top 15 pathways, ranked by adjusted p-value, are shown. A: AMEND module and ORA pathways, OGT-KO vs. TMG at 2 weeks (P) (+); B: AMEND module and ORA pathways, OGT-KO vs. TMG at 2 weeks (P) (−); C: AMEND module and ORA pathways, OGA-KO vs. TMG at 2 weeks (P) (+); G: AMEND module and ORA pathways, OGA-KO vs. TMG at 2 weeks (P) (−). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Figure 6

OGT KO Livers have increased aneuploidy after partial hepatectomy: A: Flow cytometry was performed on fixed and propidium iodine stained livers. Control livers after PHX were polyploid (blue is no PHX, red is after PHX). B: OGT KO livers had greater then 4n and aberrant DNA content after PHX (blue is no PHX, red is after PHX) (n = 3). I: Western blot analysis on mitotic proteins. Actin is used as a load control (n = 3). C–H: Quantitation of % Polyploidy between wildtype and OGT-KO after PHX gated to the wildtype polyploid peak (n = 3). J–Q: Densitometry of western blot samples. ∗ = p value less than 0.05. R: Western blot of HIPPO signaling pathway proteins (n = 3). Actin is used as a load control (n = 3). S–V: Densitometry of western blot samples. ∗ = p value less than 0.05. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Figure 7

Loss of OGT activates several pathways involved in aneuploidy: A: Western blot of mTOR signaling pathway proteins (n = 3). GAPDH is used as a load control (n = 3). B–H: Densitometry of western blot samples. ∗ = p value less than 0.05. I: Western blot of AMPK signaling pathway proteins (n = 3). GAPDH is used as a load control (n = 3). J–M: Densitometry of western blot samples. ∗ = p value less than 0.05. N: Western blot of p53 signaling pathway proteins (n = 3). GAPDH is used as a load control (n = 3). O–Q: Densitometry of western blot samples. ∗ = p value less than 0.05. R: Summary: Multi-Omic approached coupled with genetic and pharmacologic manipulation of O-GlcNAc identified several pathways linked to increased aneuploidy changed in OGT KO livers. Flow cytometry of OGT KO livers after PHX showed increased aneuploidy and validation western blots displayed changes consistent with aneuploidy (Figure designed in Biorender).