Galectin-3 (gal-3) is a lectin that participates in many cellular functions, ranging from cell growth, differentiation and proliferation to inflammation, cell-cell and cell-matrix interactions. Similar to other lectins, gal-3 possesses a carbohydrate recognition domain (CRD) that confers to these proteins their typical ability to crosslink glycosylated ligands. However, among the 15 known galectins, gal-3 is the only one with a disordered N-terminal domain that seems to promote its oligomerization and confer unique functions to this lectin. As such, gal-3 can be involved in the onset of many human diseases, including cancer, fibrosis, and inflammation (Sciacchitano et al., 2018). The link between gal-3 and cancer has been mainly studied in thyroid carcinoma (TC). An increased expression of gal-3 has been observed in malignant thyroid tissues, and this correlated with poor prognosis, cancer progression, and metastasis (Paron et al., 2003; Giusti et al., 2008; Song et al., 2012; Li et al., 2019; Wang et al., 2019). Moreover, increased gal-3 serum levels were measured in TC patients compared to controls. For this reason, serum gal-3 was proposed and used as a biomarker for the prognosis of papillary thyroid carcinoma (PTC) (Giusti et al., 2008). More evidence also showed that the over-expression of gal-3 is not the consequence but rather one of the drivers of the malignant transformation of the thyroid (Elad-Sfadia et al., 2004; Levy et al., 2010). Interestingly, single nucleotide polymorphisms (SNPs) within the gal-3 coding gene (LGALS3) have been associated with cancer risk, adding further evidence to the studies suggesting the carcinogenic roles of this protein. In particular, the rs4644, encoding for a proline to histidine substitution at codon 64 of gal-3 [P64 > H64], was associated with variable risks of breast (Balan et al., 2008), prostate, cervical (Meyer et al., 2013; Fang et al., 2017), and differentiated thyroid carcinoma (DTC) (de Boer et al., 2012; Song et al., 2014). Moreover, other studies also reported associations with other types of cancer for SNPs in strong linkage disequilibrium with rs4644, such as the case for rs4652 and the risk of gastric carcinoma (Shi et al., 2017) or the molecular characteristics of colorectal cancer (Korkmaz et al., 2016) suggesting more mechanisms involving gal-3 in tumorigenesis.

Given its high conservation among mammals and vertebrates, rs4644 was predicted to change gal-3 biological activity, prompting in vitro and in vivo studies. Thus, it was found that in gal-3-deprived breast cancer cells, the forced expression of H64 or P64 produced two forms of gal-3 not equally cleaved by matrix metalloproteinases (MMPs) −2 and −9. Moreover, in the same study, H64 and P64 conferred different abilities of chemotaxis, chemo-invasion, angiogenesis, and sensitivity to death receptor-mediated apoptosis (TRAIL-pathway). P64 variant-expressing cells, when xenografted in nude mice, have also shown reduced angiogenesis and tumor progression (Nangia-Makker et al., 2007; Nangia-Makker et al., 2010; Mazurek et al., 2011).

Furthermore, in a large case-control association study, we found that the variant H64 allele of rs4644 was associated with a reduced risk of DTC compared to P64 (Corrado et al., 2021), and the association was also found in the repository offered by the public database “UK BioBank” (Bycroft et al., 2018) following the analysis with Phenoscanner (Kamat et al., 2019) and Open Targets Genetics (Ghoussaini et al., 2021). Thus, we recently evaluated the hypothesis that P64 and H64 gal-3 variants are not functionally equivalent in thyroid cells. By using the CRISPR/Cas9 technique on the non-malignant thyroid follicular epithelial cells Nthy-Ori, we generated two isogenic cell lines that were homozygous either for the CC/gal-3P64 (defined as NC-P64 cells) or the AA/gal-3H64 (NA-H64) genotype at the rs4644 (Corrado et al., 2023). Since nuclear gal-3 is known to bind and cooperate with transcription factors (Guazzi et al., 1990; Paron et al., 2003; Takenaka et al., 2003; Gilbert-Sirieix et al., 2011), in a previous study, we evaluated the transcriptome of these cells as well as that of the original Nthy-Ori (heterozygous for the rs4644) (Corrado et al., 2021).

This previous study showed that the mRNA expression of several genes was dependent on the LGALS3 genotype (Corrado et al., 2021), confirming the gal-3 involvement in gene expression and suggesting a role for the rs4644 in modulating this activity. However, a detailed molecular analysis of protein expression changes in these cells was still missing. In the present work, to assess the effect that the rs4644 may exert on protein expression, we evaluate and compare the proteomic profiles of the engineered NC-P64 and NA-H64 cells. To provide a mechanistic explanation for the observed results, we also put these data in relation to those from our aforementioned transcriptomic study.

In the present work, we added evidence that the common gal-3P64 variant is not functionally equivalent to the alternative gal-3H64 variant. According to public databases, the proline residue at this codon is highly conserved throughout evolution, and the histidine variant is present only in humans. Unlike other galectins, gal-3 can assemble into oligomers whose valency can affect its functions (Nangia-Makker et al., 2018). In particular, dimerization is thought to occur through the reciprocal interaction between a disordered N-terminal domain (NTD) and the carbohydrate-recognition domain (Lin et al., 2017). It has also been shown that the NTD of gal-3H64, but not that of gal-3P64, can be cleaved by the proteolytic activities of MMP2 and MMP9 (Ochieng et al., 1994; Nangia-Makker et al., 2010).

In agreement with this notion, we found that the NA-H64 cells can express a heterodimer composed of a cleaved subunit and an uncleaved one, while, in the NC-P64 cells, only the homodimer of two uncleaved monomers and the monomer were detectable. This fact is strongly indicative that the extent of dimerization is genotype-dependent, and it could explain part of the differential biological effects of the two alleles (Dumic et al., 2006; Chiu et al., 2020; Farhadi et al., 2021).

In addition, our work clearly showed that the proteomic profile in the isogenic cell lines NA-H64 and NC-P64 is different, with 25 proteins overexpressed and 9 downregulated in NA-H64 compared with NC-P64 cells. For 31 of these proteins, we had previous data on the mRNA expression levels. Interestingly, when we compared the protein and the mRNA expression, we found the same trend in 24 out of 31 cases (77%). These results confirmed our previous findings on the transcriptome of these cells (Corrado et al., 2021) and, in turn, can be explained by an rs4644-dependent transcriptional activity of gal-3. In fact, gal-3 is known to function in the nucleus as a co-regulator of many transcription factors, such as TTF-1 and TCF-4 (Guazzi et al., 1990; Paron et al., 2003; Takenaka et al., 2003; Gilbert-Sirieix et al., 2011). Therefore, overall, we can conclude that the rs4644-AA genotype, when compared to the common CC genotype, is associated with or causes deregulation of the transcription of 24 mRNAs, and this altered transcription also causes an altered expression of the encoded proteins.

Among the 24 mRNAs, 21 were upregulated in the AA genetic background, suggesting an enhanced transcriptional nuclear activity of gal-3 associated with this genotype. Interestingly, when an increased nuclear expression of gal-3 was induced in engineered prostate cancer cells, inhibition of malignant behaviors was observed (Califice et al., 2004). In vitro, these cells exhibited reduced Matrigel invasion and limited anchorage-independent growth, while, in vivo, they showed reduced tumor growth and angiogenesis and increased inducible apoptosis (Califice et al., 2004). Thus, we could speculate that the reduced risk of DTC associated with the rs4644-A allele (Corrado et al., 2021) may be ascribed (at least partially) to an increased nuclear activity of gal-3H64.

Concerning ACTB, DLST, PSAT1, HNRNPD, and HNRNPC proteins (higher protein expression in NA-H64 than in NC-P64 cells with reduced or unaltered mRNA expression), the effect of the polymorphism is less obvious. However, it has been clearly shown that, in the cytosol, gal-3 binds the heterogeneous nuclear ribonucleoprotein hnRNP-L that, in turn, binds mRNAs (Coppin et al., 2017). This complex is pivotal in regulating mRNA stability and translation rates (Gu et al., 2020) by binding to CAREs (CA-repeated elements) of mRNAs (Coppin et al., 2017; Venkata Subbaiah et al., 2019). When hnRNP-L binds CAREs within the 5′UTR, the complex works as an ITAF (internal ribosome entry site element trans-acting factor), crucial for enhancing translation initiation (Seo et al., 2017). When hnRNP-L binds CAREs within the 3′UTR, the complex can elicit different effects on the intensity of translation by stabilizing the mRNA or cooperating or competing with translation inhibitors such as the RISC-miRNAs or GAIT complexes (Venkata Subbaiah et al., 2019). Since DLST, HNRNPC, and PSAT1 bear CAREs (Venkata Subbaiah et al., 2019), it could be speculated that the gal-3-hnRNPL complex has a differential activity on mRNAs, depending upon the rs4644 genotype.

For ACTB and HNRNPD, we should evoke hitherto unknown mechanisms of rs4644-dependent post-transcriptional regulation. Similar mechanisms could also affect CNN3 (calponin-3) since this protein showed the largest increase of expression (21.8-fold) in association with the rs4644-AA genotype, which is not explainable by the modest increase of mRNA expression (1.1-fold). Interestingly, CNN3 binds actin, another protein strongly upregulated (5.7-fold) in the NA-H64 cells. CNN3 activity is poorly known, but recent studies have highlighted a role in modulating cells’ motility and contractile ability. Moreover, a role in modifying the Yap/Taz-dependent transcriptional activation was suggested. Thus, CNN3 and/or its phosphorylated form could play a role in cell differentiation, proliferation, and migration via stress fiber formation or cytoskeletal remodeling (Shibukawa et al., 2013; Liu and Jin, 2016; Maddala et al., 2020).

Among the proteins whose low expression was associated with the rs4644-AA genotype, PRDX2 (peroxiredoxin 2) caught our attention. This enzyme is involved in the oxy-reductive pathways of thyrocytes for thyroid hormone production, and it has been previously reported as one of the key upregulated enzymes in thyroid carcinoma (Netea-Maier et al., 2008). Moreover, in a proteomic study of thyroid carcinoma cells (Trojanowicz et al., 2010), it has been shown that this enzyme is strongly downregulated when cells are treated with retinoic acid. This change was also associated with differentiation, reduced proliferation, and lower invasive capacities of the cells. Although it is unknown whether these anti-malignant phenotypes are caused by or simply associated with PRDX2 decrease, it is interesting to note that the genotype associated with a reduced PRDX2 protein level was also associated with a reduced risk of DTC (Corrado et al., 2021).

The change of expression of 34 proteins, once analyzed for potentially altered molecular pathways using Reactome, suggested a link between rs4644 and the level of the heat-shock and stress-related proteins (HSPs), likely driven by a higher activity of HSF-1 transcription factor in NA-H64 than NC-P64 cells. These observations suggest that the NA-H64 cells are more constitutively stimulated as they were in a higher stress state compared to NC-P64 cells. HSPs play a pivotal role in controlling the correct folding of newly synthesized proteins, assuring their functional conformation, and preventing the aggregation of damaged proteins. The involvement of rs4644 in post-translational modification and protein folding was also confirmed using an independent algorithm (Ingenuity Pathway Analysis). Among the stress-related proteins, we noticed NDGR1 was highly expressed in NA-H64 cells. This protein is a tumor suppressor in many cell types (Ghafouri-Fard et al., 2023). It is important for p53-mediated caspase activation/apoptosis and mitotic spindle checkpoint. It protects cells from aberrant mitotic spindle formation, helping to maintain euploidy (Ghafouri-Fard et al., 2023). These data suggest that a slightly higher activation of stress pathways associated with the rs4644-AA genotype protects thyrocytes from a malignant transformation.

In summary, the present study provided further information on the biological effects of the rs4644 polymorphism, shedding new light on possible mechanisms of its association with human susceptibility to cancer.