Plasma proteome profiles associated with inflammation, angiogenesis, and cancer

Abstract

Tumor development is accompanied by a complex host systemic response, which includes inflammatory and angiogenic reactions. Both tumor-derived and systemic response proteins are detected in plasma from cancer patients. However, given their non-specific nature, systemic response proteins can confound the detection or diagnosis of neoplasia. Here, we have applied an in-depth quantitative proteomic approach to analyze plasma protein changes in mouse models of subacute irritant-driven inflammation, autoreactive inflammation, and matrix associated angiogenesis and compared results to previously described findings from mouse models of polyoma middle T-driven breast cancer and Pdx1-Cre Kras(G12D) Ink4a/Arf (lox/lox)-induced pancreatic cancer. Among the confounding models, approximately 1/3 of all quantified plasma proteins exhibited a significant change in abundance compared to control mice. Of the proteins that changed in abundance, the majority were unique to each model. Altered proteins included those involved in acute phase response, inflammation, extracellular matrix remodeling, angiogenesis, and TGFβ signaling. Comparison of changes in plasma proteins between the confounder models and the two cancer models revealed proteins that were restricted to the cancer-bearing mice, reflecting the known biology of these tumors. This approach provides a basis for distinguishing between protein changes in plasma that are cancer-related and those that are part of a non-specific host response.

Figure 1. Comparison of plasma proteomic changes in confounding models.

Venn diagrams comparing (A) increased and (B) decreased proteins in plasma from the subacute and chronic inflammation and angiogenesis models as compared to control mice. Diagrams show numbers of proteins, either elevated or reduced in each model, and which are unique or shared between each of the 3 models. The majority of either increased or decreased proteins were unique to each model. (C) Correlation plots for quantified proteins. Case/control (log2) ratios for each quantified protein are plotted on the X- and Y- axes. The plots reflect abundance differences of specific proteins between two models. Proteomic comparisons between chronic versus subacute inflammation comparison are more similar (R = .67) than are comparisons between either inflammation model and the angiogenesis model (R = .49 and .31, respectively).

Figure 2. Heat map of plasma proteins altered in both confounders and cancer models.

(A) Proteins increased in both confounders and cancer models compared to controls (>1.25 fold, p<0.05), (B) proteins increased in confounders and decreased in cancer models, and (C) proteins decreased in confounders and increased in cancer models (red = up, green = down, black = no change, grey = not quantified).

Figure 3. Comparisons of selected proteins in plasma from confounders and cancer models.

Plots show mass spectrometry based log2 case/control ratios for specific proteins in each confounder and cancer model that display differential abundance patterns. Loxl1 represents proteins that are reduced in the confounders and elevated in the cancer models. Fasn shows elevation in all models but more so in PanIN. Prg4 is elevated primarily in the pancreatic cancer model. Lcn2 is elevated in the inflammation and to a greater degree in the cancer models, but not the angiogenesis model.

Figure 4. Quantitative analysis of selected proteins increased in cancer and inflammation mouse models.

ELISA analysis of Pf4, Igf1, Igfbp5, and Lcn2 showing increased plasma concentration in PyMT breast tumor-bearing mice compared to either the subacute or chronic inflammation mice. Plasma from untreated mice was used as the control.

Figure 5. Network analysis of plasma proteins from confounding conditions.

Top networks assigned by Ingenuity Pathway Analysis for both increased and decreased proteins from the subacute inflammation (A, B), chronic inflammation (C, D) and angiogenesis models (E, F). Networks for the subacute model show abundant fibrinogen and ECM proteins, the chronic model gives prominent growth factor and collagen networks, while the angiogenesis network shows chemokine and coagulation proteins. [red = increased, green = decreased, white = no change in abundance in cases vs control mice, >1.25 fold p<0.05.].