Drosophila modeling can be effectively used for comprehensive and refined analysis of tumorigenesis, including the discovery of therapeutic anti-cancer drugs and their testing [1]. The Drosophila lgl gene was the first animal tumor suppressor found and the first tumor-associated gene encoding cytoplasmic protein. We compared the gene ontology of two cancer  associated cytoplasmic proteins, Lgl and DFak56 (ortholog of human focal adhesion kinase, FAK). On the molecular level, both Lgl and FAK are involved in protein binding and the formation of macromolecular complexes mediated by phosphorylation in specific sites. On the cellular level, Lgl and FAK participate in cytoskeletal structure/dynamics, cell/ESM and cell to cell adhesion, and cell/tissue polarity. The biological processes of both genes comprise protein transport, cell signaling, cell motility and proliferation. Surprisingly, we found that diverse lgl*- null variants are widespread in lgl*-/lgl+ haplozygotic state in distant populations. To address this paradox, we found that under permanent thermal stress the developmental viability and the life span of lgl*-/lgl+ heterozygotes increased compared to the control flies. The stress-protective haplo-adaptive effect was maternally mediated and sex-specific, as males are more sensitive. The exposure of virgin haplozygotic females with one functional lgl allele to pulse thermal stress at successive stages of oogenesis showed that the germ line - early oocyte stage appeared most sensitive. Pulse heating of this stage in the parental females with one lgl dose resulted in a transgenerational haplo-adaptive effect on the viability and life span of the next generation animals. These data are important for a comprehensive knowledge of cancer-associated gene effects and evaluation of the aftermath of cancer therapy.