Triple negative breast cancer (TNBC) is a heterogeneous disease with more aggressive clinical courses than other subtypes of breast cancer. To data, functional proteomic analysis provides complementary information that can be integrated with the genomic and transcriptomic data to explore the clinic-pathological characteristic differences as well as the accurate classification for breast cancer subtypes. We performed a highly sensitive proteomic approach of iTRAQ-labeling coupled LC-MS/MS to obtain the global proteome and unravel protein signatures in tumor tissues and corresponding para-tumor tissues from 24 patients with Grade I-II and Grade III primary TNBC. Totally, 5,401 unique proteins were identified and quantified in different stage of TNBC. 865 proteins were changed in patients with Grade I or II TNBC, among which 309 were up-regulated and 556 were down-regulated. Meanwhile, for patients with Grade III TNBC, 359 proteins were increased and 672 proteins were decreased. Differentially expressed proteins were further analyzed by bioinformatic analyses, including GO function classification annotation, ingenuity pathway analysis and KEGG enrichment analysis. Comparing to para-cancerous tissues, various signaling pathways and metabolic processes, including PPAR pathways, PI3K-Akt pathway, one-carbon metabolism, amino acid synthesis, and lipid metabolism were activated in TNBC cancer tissues. Interestingly, death receptor signaling was significantly activated in Grade I-II TNBC, however, remarkably inhibited in Grade III TNBC. Our proteomic data presented precise quantification of potential signatures, signaling pathways, regulatory networks and characteristic differences in each clinic-pathological subgroup. The proteome provides complementary information for TNBC accurate subtype classification and therapeutic targets research.