FRMD3 gene: its role in diabetic kidney disease. A narrative review

The FRMD3 (ID: 257019) is located on chromosome 9, band q21.32. It has 21 exons spanning 2,282 base pairs (http://​www.​ncbi.​nlm.​nih.​gov/​gene/​257019) (Fig. 1). In addition, 7 different mRNA sequence splice variants for the FRMD3 gene are also shown in Fig. 1.

FRMD3 encodes protein 4.1O, a structural protein that is part of the protein 4.1 family [35]. Members of this family act as cytoskeletal proteins, maintaining cell shape and integrity in a variety of cell types including rat nephron cells [36, 37]. However, the role of protein 4.1O has not yet been elucidated [38]. It contains a FERM domain, which contributes to cell integrity by interacting with transmembrane proteins and actin filaments [38, 39]. FRMD3 is detected in different tissues according to the Nephromine database (Fig. 2). Studies also have reported FRMD3 expression in adult ovaries, fetal skeletal muscle, brain, thymus, and human podocytes [35]. In addition, FRMD3 expression seems to be decreased in a DKD mouse model as compared to non-diabetic mouse kidneys (Fig. 3a). FRMD3 is differentially expressed in different regions of murine kidney, showing the highest level in glomeruli (Fig. 3b).

The 4.1 protein family comprises a group of proteins including 4.1R, 4.1G (general type), 4.1B (brain type), and 4.1N (neuron type) [40]. Erythrocyte protein 4.1 (4.1R) is a multifunctional protein essential for maintaining erythrocyte shape, mechanical properties of the membrane, deformability, and stability through interactions with the spectrin and actin skeleton network [40]. Figure 4 represents different domains of the FRMD3 protein. The FERM domain [38] is an N-terminal 30-kDa domain containing binding sites for cytoplasmatic tails of integral membrane proteins such as band 3 (integral membrane protein involved in spectrin-actin interaction) [41, 42], glycophorin C (erythroid membrane proteins) [43], CD44 (transmembrane glycoprotein found in erythroid as well as non-erythroid cells) [44], p55 (abundantly palmitoylated phosphoprotein of the erythroid membrane) [43] and calmodulin (a highly conserved calcium-binding protein, ubiquitously distributed in eukaryotic cells, known to affect a plethora of biological functions) [40]. An internal 8-10 kDa domain contains the spectrin-actin binding activity that is necessary for membrane stability [45], and the 22–24 kDa C-terminal domain has been reported to bind immunophilin FKBP13 (membrane-associated protein thought to function as an endoplasmic reticulum chaperone) [46] and nuclear mitotic apparatus protein (NuMA) [47].

FERM domain proteins have been implicated in nephron function. Some of these proteins were described as important in the composition of specialized cell–cell adhesion complexes. These complexes called slit-diaphragms that form between podocyte foot processes surrounding glomerular blood vessels have an important role in the function of podocytes, establishing the blood filtration barrier in the kidney glomerulus. A reduced expression of some FERM domain proteins in zebrafish, such as nephrin, podocin or band 4.1/FERM domain mosaic eyes resulted in loss of glomerular filtration discrimination and abnormal passage of high molecular weight substances into the glomerular filtrate [48]. Another FERM domain protein that has been recently related to nephron function is kindlin, an adaptor protein that contributes to renal tubulointerstitial fibrosis. This protein regulates renal tubular cell plasticity by activation of Ras and its downstream signaling pathway [49]. Therefore, it is possible that FRMD3 might have related functions in kidney.

In the study of Pezzolesi et al. [25], FRMD3 expression was increased in proximal renal tubular human cells. Expression data for FRMD3 and its co-expressed transcripts suggest that these genes are linked to early development of DKD [50]. Moreover, FRMD3 is predicted to interact with other genes, suggesting alternative pathways in which this gene might act (Fig. 5).

It has been hypothesized that the 9q21.32 locus contributes to glomerular injury early in the pathogenesis of DKD [51]. More recently, the rs1888747 risk allele was shown to generate a transcription factor binding site (TFBS) in a module that is shared by multiple members of the bone morphogenetic protein (BMP) signaling pathway, which has previously been implicated in the development of DKD [50, 52]. Regarding these findings, new pathways are being proposed to explain the role of FRMD3 in DKD physiopathology.

Research has been developed to investigate the association of variants in the FRMD3 gene and risk of DKD. The main variants studied so far are briefly described in Table 1. The polymorphisms rs942278, rs942280, rs942283, rs1535752, rs1535753, rs2378658, rs10867977 and rs10867977 are in strong linkage disequilibrium (LD) with each other. The Hapmap project does not show LD information for the rs1888746, rs1888747, rs10868025 and rs13289150 SNPs (http://​hapmap.​ncbi.​nlm.​nih.​gov).

The first published study regarding variants in FRMD3 was a GWAS with T1DM patients from the GoKinD (Genetics of Kidneys in Diabetes) collection [25]. The SNP rs10868025 had the strongest association with DKD (OR = 1.45, P = 5.0 × 10⁻⁷) [25]. This SNP involves a G for A nucleotide substitution and is located in the long arm of chromosome 9 at position 85.4, near the 5′ end of FRMD3 [25]. The frequency of the risk allele was 0.66/0.56 (cases/controls) in the Joslin Diabetes Center study (JDC study) and 0.66/0.59 in the George Washington University study (GWU study) [25]. Two other FRMD3 SNPs associated with DKD in the GoKinD collection (rs1888746 and rs13289150) were also associated with the development of severe nephropathy in the DCCT/EDIC (Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications) follow-up study [25].

In a meta-analysis, involving three studies including 1052 cases and 2057 controls, the rs10868025 SNP was associated with a lower risk of DKD in European T1DM subjects (OR 0.72 [95 % CI 0.64–0.81]) [29], whereas in a study with T2DM subjects [26], no significant associations between the rs10868025 SNP and DKD were detected in four independent Japanese populations. The G allele frequency for cases vs. controls in the four populations studied was: 0.74 vs. 0.73, 0.72 vs. 0.73, 0.67 vs. 0.73, and 0.70 vs. 0.73 (P > 0.25 in all cases). The same findings were replicated in the Chinese population [53], for which no association between FRMD3 SNPs and DKD was observed (P = 0.249).

The rs1888747 (C/G) SNP is intergenic, located on the long arm of chromosome 9 at position 85.3, near the promoter region of the gene [54]. In the study developed by Pezzolesi et al. [25], a relevant association of rs1888747 SNP was found with DKD (P = 6.3 × 10⁻⁷) in Caucasians with T1DM. The frequency of the risk allele (G) was 0.68/0.73 for controls/cases in the GWU GoKinD and 0.66/0.74 for controls/cases in the JDC GoKinD [25]. These findings were strengthened by a meta-analysis including three studies analyzing the rs1888747 SNP in a total of 1052 cases and 2057 controls [29], which confirmed an association of rs1888747 SNP with DKD (OR 0.74 [95 % CI 0.65–0.83]; P = 0.602) [29], primarily driven by the GoKinD cohort. However, in Japanese individuals with T2DM, no associations between rs1888747 SNP and DKD were demonstrated in four different groups [26]. The frequency of the G allele for cases/controls was 0.81/0.80, 0.80/0.80, 0.70/0.82 and 0.77/0.81 (P > 0.05).

In 2011, Freedman et al. [55] identified that a potential interaction between FRMD3 and the APOL1 (apolipoprotein L1 gene)—MYH9 (non-muscle myosin heavy chain 9 gene) region on chromosome 22 contributes to DKD susceptibility in African Americans. Nevertheless, initially no association was found between FRMD3 SNPs (rs942280 and others, Table 1) and DKD before adjusting for variants on chromosome 20. The results revealed an approximately 25–30 % increase in DKD risk with multiple FRMD3 SNPs in subjects not homozygous for MYH9 risk haplotypes (or APOL1risk variants).

One year later, Williams et al. [30] formed the GEnetics of Nephropathy, an International Effort (GENIE) consortium, to examine reported genetic associations with DKD in T1DM previously observed in the GoKinD collection [25]. The samples were provided by the All Ireland-Warren 3-Genetics of Kidneys in Diabetes UK and Republic of Ireland (UK–ROI) collection and the Finnish DKD Study (FinnDiane). None of the analyzed FRMD3 polymorphisms were associated with DKD (rs1888747 [P = 0.77 and 0.25, UK–ROI and FinnDiane respectively]; rs1086805 [P = 0.52 for UK–ROI and 0.25 for FinnDiane]) [30].

Further evidence of the role of FRMD3 in kidney disease came from the study of Park et al. [56]. In this study, the authors aimed to find candidate genetic determinants of renal function in 1007 individuals from 73 extended families of Mongolian origin [56]. The strongest associations found were with rs17400257 SNP (P = 7.21 × 10⁻⁹) and rs6559725 SNP (P = 9.12 × 10⁻⁷). The rs17400257 SNP is located 45 kb downstream of FRMD3, and rs6559725 SNP is located in the intronic region of FRMD3 [56].

After the initial findings in T1DM in 2009 [25], Pezzolesi et al. [51] proceeded to examine whether the SNPs at these susceptibility loci were associated with DKD in a T2DM Family Collection. The 9q21.32 locus was significantly associated with high microalbuminuria, proteinuria, and ESRD [51]. Among diabetic family members, rs1888747 SNP on chromosome 9q21.32 was associated with advanced nephropathy (P = 0.029). Furthermore, when the definition of DKD was expanded to include individuals with high microalbuminuria, the strength of the association improved significantly [51].

Martini et al. [50] proposed a transcriptional link that might explain how the rs1888747 SNP in FRMD3 influences transcriptional regulation within the bone morphogenetic protein (BMP)-signaling pathway. FRMD3 transcript levels decreased significantly with the progression of DKD (P < 0.02). FRMD3 gene expression was studied by comparing renal biopsies in a group of 22 Pima Indians with T2DM and normal GFR and a cohort of seven participants with T2DM and CKD stage 3 [50]. Hierarchical clustering using the FRMD3 coexpressed transcripts detected two distinct clusters. In cluster 1, presenting higher ∆ACR/year (∆ albumin-to-creatinine ratio per year), gene expressions of seven out of the eight BMP pathway genes (BMPR2, CREB1, KRAS, MAP3K7, PRKAR2B, SMAD5, and XIAP) were lower than the expressions of these genes in cluster 2. These findings suggest a common molecular mechanism responsible for the coregulation of FRMD3 and several BMP pathway members [50]. Additionally, in silico comparison of sequence variants of the risk allele identified a potential homeodomain factor (HOMF) transcription factor binding site (TFBS) covering the SNP position [50]. The rs1888747 SNP affects protein binding, suggesting the generation of a TFBS by this particular SNP [50]. However, the mechanism mediating the connection between FRMD3 and BMP pathway members is still unknown, and no data are available at protein, RNA, or microRNA levels to elucidate this association [50]. It should be noted that Palmer and Freedman [54] suggested that the framework proposed by Martini et al. [50] is limited to genes with known or predicted functional roles.

Recently, a GWAS was performed by Palmer et al. [57] with the objective of evaluating candidate DKD susceptibility genes in African Americans. The SNPs selected were the same previously studied in European ancestries by Pezzolesi et al. [25]. The results showed that SNPs in FRMD3 tended toward association with T2DM-ESRD (P < 0.05).