Metformin lowers HMGA1a and RBM3 protein in breast cancer cells, and its effect on ESR1 splicing depends on cell density and exposure time

Abstract

The ESR1 gene produces ER46, an N-terminally truncated isoform generated by skipping of exon 1, which is reduced in tamoxifen-resistant breast cancer cells. We have reported that metformin moves the ratio of full-length ER to ER46 in opposite directions in MCF-7 cells and their tamoxifen-resistant subline TAMR1, and that metformin reduces TAMR1 xenograft growth (Harada et al., Am J Pharm Health Res 2022;10(11)). That report leaves two questions open: whether the effect is dosedependent and stable across culture conditions, and what splicing factor might carry it. Here we report the experiments that bear on both. Across a full dose range from 0.25 to 5 mM at high seeding density, metformin produced no consistent change in the full-length share of ESR1 transcript in either line: densitometry of the gel gave 42%, 43%, 42%, 43%, 45% and 48% in MCF-7 cells and 55%, 44%, 52%, 47%, 68% and 43% in TAMR1 cells. The reported shift therefore depends on the seeding density used, and the comparison that produced it is not reproduced at high density. At low density the MCF-7 response persisted to 48 h (52%, 62% and 68% at 0, 2.5 and 5 mM). Metformin lowered HMGA1a protein in MCF-7 cells and showed the same tendency in TAMR1 cells, at both seeding densities, and lowered RBM3 protein in TAMR1 cells at 24 and 48 h. HMGA1a is the factor we previously showed to generate ER46 by trapping U1 snRNP at an upstream pseudo-5' splice site, so a fall in HMGA1a predicts less ER46, the opposite of what was seen in TAMR1 cells at low density. The fall in RBM3, the protein through which metformin alters splicing in myotonic dystrophy, is the more plausible candidate. RT-PCR of the xenograft tumours from the earlier study detected full-length ESR1 transcript in the untreated and tamoxifen-plus-metformin groups but not in the singleagent groups, with amplification failing in several lanes; these panels do not separate the treatments. Taken together, the direction of the metformin effect on ESR1 splicing is conditional on culture density and does not survive a dose-response test at high density, and HMGA1a is unlikely to be the factor that carries it.