Membrane-proximal binding of STAT3 revealed by cancer-associated receptor variants.

ABSTRACT In cancer biology, somatic mutations in the extracellular (ligand binding) and cytosolic (functional/catalytic) domains are pursued with great interest. However, in our recent publication we report that germline mutations in the membrane-proximal region of type I receptors are able to modulate the amplitude of signal transducer and activator of transcription 3 (STAT3) signaling in cells. This unexpected finding has implications for the prognosis of heritable cancer.

Single-nucleotide polymorphisms (SNPs) are the most common type of genetic variation in the general population. These genetic differences among individuals predict an individual's response to certain drugs, susceptibility to environmental factors, and risk of developing particular diseases. Most importantly, SNPs play a significant role in complex diseases such as autoimmune diseases, metabolic disorders, cardiovascular diseases, and cancer. Approximately 30% of non-synonymous SNPs located in coding regions are estimated to impact protein function 1,2 by affecting the kinetic parameters of enzymes, the DNA-binding properties of transcription factors, the signal transduction activities of transmembrane receptors, and the architectural roles of structural proteins. Recent genome-wide association studies (GWAS) identified thousands of new variants associated with cancer. 3 Some well-known germline mutations associated with the Mendelian predisposition to cancer are found in the following genes: breast cancer 1 (BRCA1) and breast cancer 2 (BRCA2) (breast, ovary); adenomatous polyposis (APC) (colon); mutS homolog 2 (MSH2) and mutL homolog 1 (MLH1) (colon, rectum); cyclin-dependent kinase inhibitor 2A (CDKN2A) (skin); and testicular germ cell tumor susceptibility 1 (TGCT1) (testes). However, these mutations are less commonly found in the general population and do not account for all cancers with a heritable component. Much of the emphasis in the field has been placed on the investigation of somatic mutations and rare germline mutations associated with cancers, and it has become commonplace in such studies to either filter out common germline variants from the cancer genome dataset or ignore them altogether.
In 2002 we identified a common variant of fibroblast growth factor receptor 4 (FGFR4) rs351855-G/A that was present in the genomes of approximately 50% of cancer patients. 4 The minor allele frequency for rs351855-A in the general population is 0.3 and varies across ethnic populations, namely 0.11 in Africans, 0.29 in Europeans, and 0.3 in Americans. Over subsequent years this was revealed to be the only SNP associated with almost all cancer types. Intrigued by its association with poor prognosis and cancer aggressiveness, we performed a systematic investigation at the molecular, cellular, tissue, and organismal levels and uncovered a novel mode of signal transducer and activator of transcription 3 (STAT3) signaling modulatory events in close proximity to inner cell membranes. In our recent publication, we demonstrate a gain-of-function effect for the cancer-associated common SNP rs351855-G/A encoding the FGFR4 p.Gly388Arg variant in humans ( Fig. 1). 5 Rs351855 G/A is a commonly occurring coding region SNP (c.1162G>A) in exon 9 of the FGFR4 gene (also known as cluster of differentiation 334 [CD334]) that results in an amino acid change at codon 388 from glycine to arginine (p.Gly388Arg) in the transmembrane domain (TMD) of the receptor. The a-helical TMDs of membrane proteins traverse the hydrophobic core of the lipid layer and are enriched for hydrophobic amino acid residues. However, the cytosolic flanking regions of TMDs are generally enriched for positively charged amino acid residues and mark the cytosolic edges of the hydrophobic spans ("positive-inside rule"). 6,7 Therefore, we hypothesized that if arginine 388 in the FGFR4 p.Gly388Arg variant shortens the TMD, the tyrosine in the Y 39 RGQ 393 motif 8 will be exposed for the binding of cytoplasmic STAT3 proteins, resulting in recruitment of STAT3 very close to the inner cell membrane where major phosphate transfer reactions are generally carried out and thereby resulting in its activation. Computational analyses of the transmembrane segments of all type I membrane proteins in humans indicated that such motifs (YXXQ/C) occurring proximal to the inner membrane are far from rare. Remarkably, such motifs are particularly enriched in human cluster of differentiation (CD) molecules, which are generally considered surface markers for immune cells. Extensive analysis of coding region variants using publicly available SNP data sets obtained from the National Heart, Lung and Blood Institute (NHLBI) Grand Opportunity (GO) Exome Sequencing Project (ESP) Exome Variant Server, 9 the Ensembl variation database release 79, 10 and the Catalog of Somatic Mutations in Cancer (COS-MIC) database led to the identification of many similar germline variants that introduce cryptic STAT3 binding sites proximal to the membrane. Intriguingly, such germline mutations were found to be co-localized with somatic mutations in the COSMIC cancer dataset, thus questioning the definitive evidence that they are somatically acquired. These results from genomic analyses further encouraged us to test the hypothesis that the FGFR4 p.Gly388Arg variant exposes a membraneproximal STAT3 binding site.
In general, investigation of the mechanistic basis of any such cancer-associated SNP remains a challenge and can only be addressed using a SNP knock-in transgenic mouse model for cancers. In our recent publication, we used our rs351855-G/A SNP knock-in transgenic mouse models for lung and breast cancers (Fgfr4 G/G , homozygous for the rs351855-G allele and Fgfr4 A/A , homozygous for the rs351855-A allele) and gathered several lines of evidence demonstrating that the membraneproximal STAT3 binding site in FGFR4 leads to enhanced recruitment of STAT3 proteins to the inner cell membrane. This in turn leads to enhanced STAT3 tyrosine phosphorylation and enhanced STAT3 signaling. 5 We observed enhanced activation of STAT3 in Fgfr4 A/A (mouse) and FGFR4 A/A (human) genotyped cells when compared to heterozygous or wild-type cells, which led to an increase in STAT3-dependent promoter activity. The FGFR4 p. Gly388Arg protein variant-induced upregulation of phosphorylated tyrosine-705 STAT3 [pSTAT3 (Y705)] was nullified upon knockdown of endogenous FGFR4 in human FGFR4 A/A cells, but not in FGFR4 G/G genotyped cells. Upon knockdown of STAT3, the enhanced proliferative capacity of Fgfr4 A/A mouse embryonic fibroblasts (MEFs) was lost. These cells were indeed more susceptible to cell death induced by inhibition of the STAT3 signaling pathway, either by small molecular inhibitors or small interfering RNA, but not to that induced by agents that inhibited the mitogen-activated protein kinase (MAPK) signaling pathway. These results evidently present an opportunity for personalized therapy of patients with rs351855-A SNP in their germline genome with treatment using STAT3 inhibitors as well as the stratification of volunteers for clinical trials of STAT3 drugs.
Enhanced STAT3 activation dependent on the FGFR4 p. Gly388Arg variant results from a direct interaction with the Y 390 RGQ 393 motif proximal to inner cell membranes and is not due to increased tyrosine kinase activity. Several independent assays led us to this conclusion. First, truncated FGFR4 constructs (in which the cytoplasmic segment from the 398 th to 802 nd amino acids was replaced by YFP [FGFR4 Gly388DYFP & FGFR4 Arg388DYFP]) co-localized with STAT3. Second, using mass spectrometry we identified the phosphate-modified tyrosine-390 (Y-390), a prerequisite for an SH2 domain interaction, in purified recombinant full-length and truncated FGFR4 Arg388 protein variants. Third, we observed a significant increase in fluorescence resonance energy transfer (FRET) signals in cell membranes co-expressing STAT3-CFP and FGFR4 Arg388DYFP compared to cells co-expressing STAT3-CFP and FGFR4 Gly388DYFP. Fourth, the FGFR4 Arg388 variant lacking the tyrosine kinase domain was still able to enhance endogenous STAT3 activation. Finally, pull-down of biotinconjugated transmembrane peptide sequences corresponding to FGFR4 p.Gly388Arg (rs351855-G/A) and macrophage stimulating 1 receptor (MST1R) p.Arg983Gln (rs375697146-C/T) germline receptor variants from cell membrane extracts of peptide transfectants exhibited increased binding of YXXQ motif containing peptides to endogenous STAT3 compared to peptides lacking the motif, although similar amounts of EGFR were associated with membrane extracts. In the transfectants, YXXQ-containing peptide variants not only colocalized with STAT3, but also enhanced tyrosine phosphorylation and STAT3-dependent promoter activity. Thus, we established the biological function of the membrane-proximal STAT3 binding site in type I receptor proteins independent of its extracellular or intracellular kinase domains. Finally, we validated our mechanistic findings in vivo by assessing the pSTAT3 (Y705) levels The germline coding region variation rs351855-G/A encoding the FGFR4 p. Gly388Arg protein variant occurs at a high frequency (pink individuals) in the general population and varies across ethnic populations. This heritable mutation, which is frequently associated with multiple cancer types, exposes a STAT3 binding site in close proximity to the inner cell membrane leading to enhanced recruitment of STAT3 proteins, enhanced STAT3 tyrosine phosphorylation, and enhanced STAT3 signaling. Our findings suggest that germline variants of cell-surface proteins that recruit STAT3 to the inner cell membrane are a significant risk factor for cancer prognosis and disease progression.
in whole organs and tumors extracted from Fgfr4 G/G and Fgfr4 A/A mice.
We also addressed the eminent question of whether mere recruitment of STAT3 to the inner membrane is sufficient to mediate STAT3 phosphorylation and enhance STAT3 signaling activity. Upon membrane targeting of STAT3 to the plasma membrane using lipid-modified sequence motifs, the tyrosine-705 phosphorylation and activation of STAT3 dramatically increased. Intriguingly, membrane-proximal STAT3 tyrosine phosphorylation events depended on the function of EGFR, although the direct components involved in membrane-proximal phosphate transfer reactions remain to be identified. The preponderance of genetic variations affecting the membraneproximal STAT3 binding site in immune cell surface markers suggests a function of such germline variations in modulating immune cell responses. Thus, our work demonstrates that membrane-proximal STAT3 interaction sites modulate the amplitude of the STAT3 signaling pathway.

Disclosure of potential conflicts of interest
No potential conflicts of interest were disclosed.