Genetically identical cells are known to differ in many physiological parameters such as growth rate and drug tolerance, but the source of such heterogeneity is often insufficiently understood. Exchange interactions between metabolite producing and consuming cells are believed to be one possible cause, but detecting metabolically divergent subpopulations remains technically challenging. We developed a proteomics-based technology, termed differential isotope labelling by amino acids (DILAC), which monitors amino acid incorporation into peptides with multiple occurrences of the same amino acid. DILAC is used to differentiate producer and consumer cells of a particular amino acid within an isogenic cell population. We applied DILAC to young, morphologically undifferentiated yeast colonies and reveal that they contain sub-populations of lysine producers and consumers which emerge due to nutrient gradients. DILAC can deconvolute the proteome of subpopulations from bulk measurements which indicated an in situ cross feeding situation where fast growing cells ferment and provide the slower growing, respiring cells with ethanol as substrate. Finally, by combining DILAC with FACS, we show that the metabolic states that differ between isogenic cells, confer resistance to the antifungal drug amphotericin B. Overall, this novel and broadly applicable methodological approach captures previously unnoticed metabolic heterogeneity, providing experimental evidence for the role of metabolic specialisation and cross-feeding interactions as a source of phenotypic heterogeneity in isogenic cell populations.