(1) Background: methionine cycle is not only essential for cancer cell proliferation but is also critical for metabolic reprogramming, a cancer hallmark. Hepatic and extrahepatic tissues methionine adenosyltransferases (MATs) are products of two genes, MAT1A and MAT2A that catalyze the formation of S-adenosylmethionine (SAM), the principal biological methyl donor. Glycine N-methyltransferase (GNMT) further utilizes SAM for sarcosine formation, thus it regulates the ratio of SAM:S-adenosylhomocysteine (SAH). (2) Methods: by analyzing the TCGA/GTEx datasets available within GEPIA2, we discovered that breast cancer patients with higher MAT2A had worse survival rate (p = 0.0057). Protein expression pattern of MAT1AA, MAT2A and GNMT were investigated in the tissue microarray in our own cohort (n = 252) by immunohistochemistry. MAT2A C/N expression ratio and cell invasion activity were further investigated in a panel of breast cancer cell lines. (3) Results: GNMT and MAT1A were detected in the cytoplasm, whereas MAT2A showed both cytoplasmic and nuclear immunoreactivity. Neither GNMT nor MAT1A protein expression was associated with patient survival rate in our cohort. Kaplan–Meier survival curves showed that a higher cytoplasmic/nuclear (C/N) MAT2A protein expression ratio correlated with poor overall survival (5 year survival rate: 93.7% vs. 83.3%, C/N ratio ≥ 1.0 vs. C/N ratio < 1.0, log-rank p = 0.004). Accordingly, a MAT2A C/N expression ratio ≥ 1.0 was determined as an independent risk factor by Cox regression analysis (hazard ratio = 2.771, p = 0.018, n = 252). In vitro studies found that breast cancer cell lines with a higher MAT2A C/N ratio were more invasive. (4) Conclusions: the subcellular localization of MAT2A may affect its functions, and elevated MAT2A C/N ratio in breast cancer cells is associated with increased invasiveness. MAT2A C/N expression ratio determined by IHC staining could serve as a novel independent prognostic marker for breast cancer.