Combinatorial Libraries of Bipodal Binders of the Insulin Receptor

The binding process of insulin to its transmembrane receptor entails a sophisticated interplay between two proteins, each possessing two binding sites. Given the difficulties associated with the use of insulin in the treatment of diabetes, despite its remarkable efficacy, there is interest in smaller and more stable compounds than the native hormone that would effectively activate the receptor. Our study adopts a strategy focused on synthesizing extensive combinatorial libraries of bipodal compounds consisting of two distinct peptides linked to a molecular scaffold. These constructs, evaluated in a resin bead-bound format, were designed to assess their binding to the insulin receptor. Despite notable nonspecific binding, our approach successfully generated and tested millions of compounds. Rigorous evaluations via flow cytometry and specific antibodies revealed peptide sequences with specific interactions at either receptor binding Site 1 or 2. Notably, these sequences bear similarity to peptides discovered through phage display by other researchers. This convergence of chemical and biological methods underscores nature's beauty, revealing general principles in peptide binding to the insulin receptor. Overall, our study deepens the understanding of molecular interactions in ligand binding to the insulin receptor, highlighting the challenges of targeting large proteins with small synthetic peptides.

Insulin receptor Arg717 and IGF-1 receptor Arg704 play a key role in ligand binding and in receptor activation

The insulin receptor (IR, with its isoforms IR-A and IR-B) and the insulin-like growth factor 1 receptor (IGF-1R) are related tyrosine kinase receptors. Recently, the portfolio of solved hormone-receptor structures has grown extensively thanks to advancements in cryo-electron microscopy. However, the dynamics of how these receptors transition between their inactive and active state are yet to be fully understood. The C-terminal part of the alpha subunit (αCT) of the receptors is indispensable for the formation of the hormone-binding site. We mutated the αCT residues Arg717 and His710 of IR-A and Arg704 and His697 of IGF-1R. We then measured the saturation binding curves of ligands on the mutated receptors and their ability to become activated. Mutations of Arg704 and His697 to Ala in IGF-1R decreased the binding of IGF-1. Moreover, the number of binding sites for IGF-1 on the His697 IGF-1R mutant was reduced to one-half, demonstrating the presence of two binding sites. Both mutations of Arg717 and His710 to Ala in IR-A inactivated the receptor. We have proved that Arg717 is important for the binding of insulin to its receptor, which suggests that Arg717 is a key residue for the transition to the active conformation.

Non-glycosylated IGF2 prohormones are more mitogenic than native IGF2

Insulin-like Growth Factor-2 (IGF2) is important for the regulation of human embryonic growth and development, and for adults' physiology. Incorrect processing of the IGF2 precursor, pro-IGF2(156), leads to the formation of two IGF2 proforms, big-IGF2(87) and big-IGF2(104). Unprocessed and mainly non-glycosylated IGF2 proforms are found at abnormally high levels in certain diseases, but their mode of action is still unclear. Here, we found that pro-IGF2(156) has the lowest ability to form its inactivating complexes with IGF-Binding Proteins and has higher proliferative properties in cells than IGF2 and other IGF prohormones. We also showed that big-IGF2(104) has a seven-fold higher binding affinity for the IGF2 receptor than IGF2, and that pro-IGF2(87) binds and activates specific receptors and stimulates cell growth similarly to the mature IGF2. The properties of these pro-IGF2 forms, especially of pro-IGF2(156) and big-IGF2(104), indicate them as hormones that may be associated with human diseases related to the accumulation of IGF-2 proforms in the circulation.

Modulation of the antagonistic properties of an insulin mimetic peptide by disulfide bridge modifications

Insulin is a peptide responsible for regulating the metabolic homeostasis of the organism; it elicits its effects through binding to the transmembrane insulin receptor (IR). Insulin mimetics with agonistic or antagonistic effects toward the receptor are an exciting field of research and could find applications in treating diabetes or malignant diseases. We prepared five variants of a previously reported 20-amino acid insulin-mimicking peptide. These peptides differ from each other by the structure of the covalent bridge connecting positions 11 and 18. In addition to the peptide with a disulfide bridge, a derivative with a dicarba bridge and three derivatives with a 1,2,3-triazole differing from each other by the presence of sulfur or oxygen in their staples were prepared. The strongest binding to IR was exhibited by the peptide with a disulfide bridge. All other derivatives only weakly bound to IR, and a relationship between increasing bridge length and lower binding affinity can be inferred. Despite their nanomolar affinities, none of the prepared peptide mimetics was able to activate the insulin receptor even at high concentrations, but all mimetics were able to inhibit insulin-induced receptor activation. However, the receptor remained approximately 30% active even at the highest concentration of the agents; thus, the agents behave as partial antagonists. An interesting observation is that these mimetic peptides do not antagonize insulin action in proportion to their binding affinities. The compounds characterized in this study show that it is possible to modulate the functional properties of insulin receptor peptide ligands using disulfide mimetics.

Targeting the insulin receptor with hormone and peptide dimers

Insulin is a key hormone involved in the regulation of overall energetic homeostasis of the organism. The dimeric character of the receptor for insulin evokes ideas about its activation or inhibition with peptide dimers that could either trigger or block the structural transition of the insulin receptor, leading to its activation. Herewith, we present the chemical engineering and biological characterization of several series of insulin dimers or dimers of specific peptides that should be able to bind receptors for insulin or insulin growth factor 1. The hormones or peptides in the dimers were interconnected with different linkers, consisting of triazole moieties and 3, 6, 8, 11, or 23 polyethylene glycol units. The prepared dimers were weaker in binding to insulin receptors than human insulin. However, some of the insulin dimers showed preferential binding specificity toward the isoform A of the insulin receptor, and the insulin dimers also stimulated the insulin receptor more strongly than would be consistent with their binding affinities. Our results suggest that designing insulin dimers may be a promising strategy for modulating the ability of the hormone to activate the receptor or to alter its specificity toward insulin receptor isoforms.

Mutations at hypothetical binding site 2 in insulin and insulin-like growth factors 1 and 2

Elucidating how insulin and the related insulin-like growth factors 1 and 2 (IGF-1 and IGF-2) bind to their cellular receptors (IR and IGF-1R) and how the receptors are activated has been the holy grail for generations of scientists. However, deciphering the 3D structure of tyrosine kinase receptors and their hormone-bound complexes has been complicated by the flexible and dimeric nature of the receptors and the dynamic nature of their interaction with hormones. Therefore, mutagenesis of hormones and kinetic studies first became an important tool for studying receptor interactions. It was suggested that hormones could bind to receptors through two binding sites on the hormone surface called site 1 and site 2. A breakthrough in knowledge came with the solution of cryoelectron microscopy (cryoEM) structures of hormone-receptor complexes. In this chapter, we document in detail the mutagenesis of insulin, IGF-1, and IGF-2 with emphasis on modifications of the hypothetical binding site 2 in the hormones, and we discuss the results of structure-activity studies in light of recent cryoEM structures of hormone complexes with IR and IGF-1R.

Recombinant Insulin-Like Growth Factor 1 Dimers: Receptor Binding Affinities and Activation Abilities

Insulin-like growth factor 1 (IGF-1) and its IGF-1 receptor (IGF-1R) belong to an important biological system that is involved in the regulation of normal growth, but that has also been recognized as playing a role in cancer. IGF-1R antagonists could be interesting for the testing of their potential antiproliferative properties as an alternative to IGF-1R tyrosine-kinase inhibitors or anti-IGF-1R monoclonal antibodies. In this study, we were inspired by the successful development of insulin dimers capable of antagonizing insulin effects on the insulin receptor (IR) by simultaneous binding to two separated binding sites and by blocking structural rearrangement of the IR. We designed and produced in Escherichia coli three different IGF-1 dimers in which IGF-1 monomers are interlinked through their N- and C-termini, with linkers having 8, 15 or 25 amino acids. We found that the recombinant products were susceptible to the formation of misfolded or reduced variants, but that some of them were able to bind IGF-1R in low nanomolar affinities and all of them activate IGF-1R proportionally to their binding affinities. Overall, our work can be considered as a pilot study that, although it did not lead to the discovery of new IGF-1R antagonists, explored the possibility of recombinant production of IGF-1 dimers and led to the preparation of active compounds. This work could inspire further studies dealing, for example, with the preparation of IGF-1 conjugates with specific proteins for the study of the hormone and its receptor or for therapeutic applications.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s10989-023-10499-1.

Characterization of insulin crystalline form in isolated β-cell secretory granules

Insulin is stored in vivo inside the pancreatic β-cell insulin secretory granules. In vitro studies have led to an assumption that high insulin and Zn²⁺ concentrations inside the pancreatic β-cell insulin secretory granules should promote insulin crystalline state in the form of Zn²⁺-stabilized hexamers. Electron microscopic images of thin sections of the pancreatic β-cells often show a dense, regular pattern core, suggesting the presence of insulin crystals. However, the structural features of the storage forms of insulin in native preparations of secretory granules are unknown, because of their small size, fragile character and difficult handling. We isolated and investigated the secretory granules from MIN6 cells under near-native conditions, using cryo-electron microscopic (Cryo-EM) techniques. The analysis of these data from multiple intra-granular crystals revealed two different rhomboidal crystal lattices. The minor lattice has unit cell parameters (a ≃ b ≃ 84.0 Å, c ≃ 35.2 Å), similar to in vitro crystallized human 4Zn²⁺-insulin hexamer, whereas the largely prevalent unit cell has more than double c-axis (a ≃ b ≃ c ≃ 96.5 Å) that probably corresponds to two or three insulin hexamers in the asymmetric unit. Our experimental data show that insulin can be present in pancreatic MIN6 cell granules in a microcrystalline form, probably consisting of 4Zn²⁺-hexamers of this hormone.

Insulin Analogues with Altered Insulin Receptor Isoform Binding Specificities and Enhanced Aggregation Stabilities

Insulin is a lifesaver for millions of diabetic patients. There is a need for new insulin analogues with more physiological profiles and analogues that will be thermally more stable than human insulin. Here, we describe the chemical engineering of 48 insulin analogues that were designed to have changed binding specificities toward isoforms A and B of the insulin receptor (IR-A and IR-B). We systematically modified insulin at the C-terminus of the B-chain, at the N-terminus of the A-chain, and at A14 and A18 positions. We discovered an insulin analogue that has Cα-carboxyamidated Glu at B31 and Ala at B29 and that has a more than 3-fold-enhanced binding specificity in favor of the "metabolic" IR-B isoform. The analogue is more resistant to the formation of insulin fibrils at 37 °C and is also more efficient in mice than human insulin. Therefore, [AlaB29,GluB31,amideB31]-insulin may be interesting for further clinical evaluation.

A radioligand receptor binding assay for measuring of insulin secreted by MIN6 cells after stimulation with glucose, arginine, ornithine, dopamine, and serotonin

We adapted a radioligand receptor binding assay for measuring insulin levels in unknown samples. The assay enables rapid and accurate determination of insulin concentrations in experimental samples, such as from insulin-secreting cells. The principle of the method is based on the binding competition of insulin in a measured sample with a radiolabeled insulin for insulin receptor (IR) in IM-9 cells. Both key components, radiolabeled insulin and IM-9 cells, are commercially available. The IR binding assay was used to determine unknown amounts of insulin secreted by MIN6 β cell line after stimulation with glucose, arginine, ornithine, dopamine, and serotonin. The experimental data obtained by the IR binding assay were compared to the results determined by RIA kits and both methods showed a very good agreement of results. We observed the stimulation of glucose-induced insulin secretion from MIN6 cells by arginine, weaker stimulation by ornithine, but inhibitory effects of dopamine. Serotonin effects were either stimulatory or inhibitory, depending on the concentration of serotonin used. The results will require further investigation. The study also clearly revealed advantages of the IR binding assay that allows the measuring of a higher throughput of measured samples, with a broader range of concentrations than in the case of RIA kits. The IR binding assay can provide an alternative to standard RIA and ELISA assays for the determination of insulin levels in experimental samples and can be especially useful in scientific laboratories studying insulin production and secretion by β cells and searching for new modulators of insulin secretion.

The efficiency of insulin production and its content in insulin-expressing model β-cells correlate with their Zn2+ levels

Insulin is produced and stored inside the pancreatic β-cell secretory granules, where it is assumed to form Zn2+-stabilized oligomers. However, the actual storage forms of this hormone and the impact of zinc ions on insulin production in vivo are not known. Our initial X-ray fluorescence experiment on granules from native Langerhans islets and insulinoma-derived INS-1E cells revealed a considerable difference in the zinc content. This led our further investigation to evaluate the impact of the intra-granular Zn2+ levels on the production and storage of insulin in different model β-cells. Here, we systematically compared zinc and insulin contents in the permanent INS-1E and BRIN-BD11 β-cells and in the native rat pancreatic islets by flow cytometry, confocal microscopy, immunoblotting, specific messenger RNA (mRNA) and total insulin analysis. These studies revealed an impaired insulin production in the permanent β-cell lines with the diminished intracellular zinc content. The drop in insulin and Zn2+ levels was paralleled by a lower expression of ZnT8 zinc transporter mRNA and hampered proinsulin processing/folding in both permanent cell lines. To summarize, we showed that the disruption of zinc homeostasis in the model β-cells correlated with their impaired insulin and ZnT8 production. This indicates a need for in-depth fundamental research about the role of zinc in insulin production and storage.

A radioligand binding assay for the insulin-like growth factor 2 receptor

Insulin-like growth factors 2 and 1 (IGF2 and IGF1) and insulin are closely related hormones that are responsible for the regulation of metabolic homeostasis, development and growth of the organism. Physiological functions of insulin and IGF1 are relatively well-studied, but information about the role of IGF2 in the body is still sparse. Recent discoveries called attention to emerging functions of IGF2 in the brain, where it could be involved in processes of learning and memory consolidation. It was also proposed that these functions could be mediated by the receptor for IGF2 (IGF2R). Nevertheless, little is known about the mechanism of signal transduction through this receptor. Here we produced His-tagged domain 11 (D11), an IGF2-binding element of IGF2R; we immobilized it on the solid support through a well-defined sandwich, consisting of neutravidin, biotin and synthetic anti-His-tag antibodies. Next, we prepared specifically radiolabeled [125I]-monoiodotyrosyl-Tyr2-IGF2 and optimized a sensitive and robust competitive radioligand binding assay for determination of the nanomolar binding affinities of hormones for D11 of IGF2. The assay will be helpful for the characterization of new IGF2 mutants to study the functions of IGF2R and the development of new compounds for the treatment of neurological disorders.

Cross-Linking/Mass Spectrometry Uncovers Details of Insulin-Like Growth Factor Interaction With Insect Insulin Binding Protein Imp-L2

Structural details of changes accompanying interaction between insulin-related hormones and their binding partners are often enigmatic. Here, cross-linking/mass spectrometry could complement structural techniques and reveal details of these protein-protein interfaces. We used such approach to clarify missing structural description of the interface in human insulin-like growth factor (IGF-1): Drosophila melanogaster imaginal morphogenesis protein-late 2 protein (Imp-L2) complex which we studied previously by X-ray crystallography. We crosslinked these proteins by heterobifunctional cross-linker sulfosuccinimidyl 4,4'-azidopentanoate (Sulfo-SDA) for the subsequent mass spectrometry (MS) analysis. The MS analysis revealed IGF-1:Imp-L2 interactions which were not resolved in the crystal structure of this assembly, and they converged with X-ray results, indicating the importance of the IGF-1 N-terminus interaction with the C-terminal (185-242) part of the Imp-L2 for stability of this complex. Here, we also showed the advantage and reliability of MS approach in solving details of protein-protein interactions that are too flexible for solid state structural methods.

Mutations at hypothetical binding site 2 in insulin and insulin-like growth factors 1 and 2 result in receptor- and hormone-specific responses

Information on how insulin and insulin-like growth factors 1 and 2 (IGF-1 and -2) activate insulin receptors (IR-A and -B) and the IGF-1 receptor (IGF-1R) is crucial for understanding the difference in the biological activities of these peptide hormones. Cryo-EM studies have revealed that insulin uses its binding sites 1 and 2 to interact with IR-A and have identified several critical residues in binding site 2. However, mutagenesis studies suggest that Ile-A10, Ser-A12, Leu-A13, and Glu-A17 also belong to insulin's site 2. Here, to resolve this discrepancy, we mutated these insulin residues and the equivalent residues in IGFs. Our findings revealed that equivalent mutations in the hormones can result in differential biological effects and that these effects can be receptor-specific. We noted that the insulin positions A10 and A17 are important for its binding to IR-A and IR-B and IGF-1R and that A13 is important only for IR-A and IR-B binding. The IGF-1/IGF-2 positions 51/50 and 54/53 did not appear to play critical roles in receptor binding, but mutations at IGF-1 position 58 and IGF-2 position 57 affected the binding. We propose that IGF-1 Glu-58 interacts with IGF-1R Arg-704 and belongs to IGF-1 site 1, a finding supported by the NMR structure of the less active Asp-58–IGF-1 variant. Computational analyses indicated that the aforementioned mutations can affect internal insulin dynamics and inhibit adoption of a receptor-bound conformation, important for binding to receptor site 1. We provide a molecular model and alternative hypotheses for how the mutated insulin residues affect activity.

A versatile insulin analog with high potency for both insulin and insulin-like growth factor 1 receptors: Structural implications for receptor binding

Insulin and insulin-like growth factor 1 (IGF-1) are closely related hormones involved in the regulation of metabolism and growth. They elicit their functions through activation of tyrosine kinase–type receptors: insulin receptors (IR-A and IR-B) and IGF-1 receptor (IGF-1R). Despite similarity in primary and three-dimensional structures, insulin and IGF-1 bind the noncognate receptor with substantially reduced affinity. We prepared [d-His^B24, Gly^B31, Tyr^B32]-insulin, which binds all three receptors with high affinity (251 or 338% binding affinity to IR-A respectively to IR-B relative to insulin and 12.4% binding affinity to IGF-1R relative to IGF-1). We prepared other modified insulins with the aim of explaining the versatility of [d-His^B24, Gly^B31, Tyr^B32]-insulin. Through structural, activity, and kinetic studies of these insulin analogs, we concluded that the ability of [d-His^B24, Gly^B31, Tyr^B32]-insulin to stimulate all three receptors is provided by structural changes caused by a reversed chirality at the B24 combined with the extension of the C terminus of the B chain by two extra residues. We assume that the structural changes allow the directing of the B chain C terminus to some extra interactions with the receptors. These unusual interactions lead to a decrease of dissociation rate from the IR and conversely enable easier association with IGF-1R. All of the structural changes were made at the hormones' Site 1, which is thought to interact with the Site 1 of the receptors. The results of the study suggest that merely modifications of Site 1 of the hormone are sufficient to change the receptor specificity of insulin.

Synthesis and Evaluation of a Library of Trifunctional Scaffold-Derived Compounds as Modulators of the Insulin Receptor

We designed a combinatorial library of trifunctional scaffold-derived compounds, which were derivatized with 30 different in-house-made azides. The compounds were proposed to mimic insulin receptor (IR)-binding epitopes in the insulin molecule and bind to and activate this receptor. This work has enabled us to test our synthetic and biological methodology and to prove its robustness and reliability for the solid-phase synthesis and testing of combinatorial libraries of the trifunctional scaffold-derived compounds. Our effort resulted in the discovery of two compounds, which were able to weakly induce the autophosphorylation of IR and weakly bind to this receptor at a 0.1 mM concentration. Despite these modest biological results, which well document the well-known difficulty in modulating protein-protein interactions, this study represents a unique example of targeting the IR with a set of nonpeptide compounds that were specifically designed and synthesized for this purpose. We believe that this work can open new perspectives for the development of next-generation insulin mimetics based on the scaffold structure.

Rational steering of insulin binding specificity by intra-chain chemical crosslinking

Insulin is a key hormone of human metabolism with major therapeutic importance for both types of diabetes. New insulin analogues with more physiological profiles and better glycemic control are needed, especially analogues that preferentially bind to the metabolic B-isoform of insulin receptor (IR-B). Here, we aimed to stabilize and modulate the receptor-compatible conformation of insulin by covalent intra-chain crosslinking within its B22-B30 segment, using the Cu(I)-catalyzed Huisgen 1,3-dipolar cycloaddition reaction of azides and alkynes. This approach resulted in 14 new, systematically crosslinked insulin analogues whose structures and functions were extensively characterized and correlated. One of the analogues, containing a B26-B29 triazole bridge, was highly active in binding to both IR isoforms, with a significant preference for IR-B. Our results demonstrate the potential of chemistry-driven modulation of insulin function, also shedding new light on the functional importance of hormone's B-chain C-terminus for its IR-B specificity.

Targeted Metabolomics for Homocysteine-Related Metabolites in Primary Hepatocytes

Liquid chromatography-tandem mass spectrometry has become the most convenient method to identify and quantify low molecular weight metabolites from various sources. Metabolomics studies of hepatocytes hold promise for the identification of the mechanisms of toxicant-related disease processes. In this chapter, we present a rapid and sensitive liquid chromatography-tandem mass spectrometry method for the quantification of intracellular concentrations of nine homocysteine-based metabolites, namely homocysteine, methionine, cysteine, dimethylglycine, cystathionine, S-adenosylmethionine, S-adenosylhomocysteine, choline, and betaine. The method is specifically designed for the analysis of cultured primary hepatocytes.

Quantification of homocysteine-related metabolites and the role of betaine-homocysteine S-methyltransferase in HepG2 cells

We optimized and validated a rapid and sensitive liquid chromatography-tandem mass spectrometry (LC-MS/MS) method for the quantification of six metabolites of homocysteine metabolism: homocysteine, methionine, cysteine, S-adenosylmethionine, S-adenosylhomocysteine and betaine. The detection limits for these metabolites were in the nanomolar range, and the intra- and inter-day precisions were lower than 20% of the relative standard deviations. The method was specifically designed for the determination of the intracellular concentrations of the metabolites in cultured cells. To study the role of betaine-homocysteine S-methyltransferase (BHMT), HepG2 cells and HepG2 cells that were stably transfected with BHMT ((BHMT) HepG2) were treated with homocysteine or with a specific inhibitor of BHMT, and metabolite levels were subsequently measured. Severely compromised methyl group metabolism in the HepG2 cells, which is typical of cancer-derived cells, prevented clear evaluation of the changes caused by the external manipulations of homocysteine metabolism. However, the ease of handling these cells and the almost unlimited source of experimental material supplied by cells in permanent culture allowed us to develop a reliable methodology. The precautions concerning intracellular metabolite determinations using LC-MS/MS in cultured cells that are expressed in this work will have global validity for future metabolomics studies.

Two-dimensional electrophoretic comparison of metastatic and non-metastatic human breast tumors using in vitro cultured epithelial cells derived from the cancer tissues

BACKGROUND

Breast carcinomas represent a heterogeneous group of tumors diverse in behavior, outcome, and response to therapy. Identification of proteins resembling the tumor biology can improve the diagnosis, prediction, treatment selection, and targeting of therapy. Since the beginning of the post-genomic era, the focus of molecular biology gradually moved from genomes to proteins and proteomes and to their functionality. Proteomics can potentially capture dynamic changes in protein expression integrating both genetic and epigenetic influences.

METHODS

We prepared primary cultures of epithelial cells from 23 breast cancer tissue samples and performed comparative proteomic analysis. Seven patients developed distant metastases within three-year follow-up. These samples were included into a metastase-positive group, the others formed a metastase-negative group. Two-dimensional electrophoretical (2-DE) gels in pH range 4-7 were prepared. Spot densities in 2-DE protein maps were subjected to statistical analyses (R/maanova package) and data-mining analysis (GUHA). For identification of proteins in selected spots, liquid chromatography-tandem mass spectrometry (LC-MS/MS) was employed.

RESULTS

Three protein spots were significantly altered between the metastatic and non-metastatic groups. The correlations were proven at the 0.05 significance level. Nucleophosmin was increased in the group with metastases. The levels of 2,3-trans-enoyl-CoA isomerase and glutathione peroxidase 1 were decreased.

CONCLUSION

We have performed an extensive proteomic study of mammary epithelial cells from breast cancer patients. We have found differentially expressed proteins between the samples from metastase-positive and metastase-negative patient groups.

2-DE analysis of a new human cell line EM-G3 derived from breast cancer progenitor cells and comparison with normal mammary epithelial cells

We performed a 2-DE analysis of proteins of the newly established spontaneously immortalized clonal cell line EM-G3 derived from a primary lesion of infiltrating ductal breast carcinoma. EM-G3 cells may represent progenitors of the mammary epithelial cells spontaneously immortalized in early phase of cancerogenesis. We compared the protein profile of EM-G3 line with proteins from populations of normal mammary epithelial cells (NME), and determined the phenotype of both types of cells. NME cells are a mixture of both main cell types in breast epithelia, myoepithelial and luminal cells. The EM-G3 breast cancer cell line has a unique basal-like phenotype. We identified proteins that are differently expressed in these cells. Cytokeratin 16, cytokeratin 19, squamous cell carcinoma antigen 1, caphepsin B and caspase 14 were predominantly expressed by NME cells. Cytokeratin 13, isoelectric variant of annexin 5, isoelectric variant of chloride intracellular channel protein 1, glyoxalase 1 and glutamine synthetase were predominantly expressed by EM-G3 cells. The proteins up-regulated in EM-G3 cells may represent potential protein markers of mammary epithelial cells progenitors and may be important in early phase of carcinogenesis.