CXCL4L1 Promoter Polymorphisms Are Associated with Improved Renal Function in Type 1 Diabetes

The metabolic dysfunctions associated with diabetes mellitus manifest in characteristic long-term complications, such as diabetic nephropathy (DN). Inflammatory factors significantly advance this process, and several single-nucleotide polymorphisms (SNPs) in inflammation-related genes, such as chemokines, have been implicated in the genetic predisposition to DN (1–4). Moreover, the level of kidney function may be linked to a cluster of chemokine genes on chromosome (chr) 4 (5). Chemokines belong to a superfamily of ∼50 small, secreted, structurally related proteins that play different roles in inflammatory and homeostatic processes (6). The two main chemokine families are named CC and CXC chemokines, according to the arrangement of N-terminal cysteine residues. Together with ∼20 different chemokine receptors, they form a complex chemokine system that orchestrates and fine tunes the extent and dynamics of the inflammatory response. Their corresponding genes are located in large clusters at particular chromosomal locations (chr 4 and 2) or separately at unique chromosomal locations. Most clustered chemokines are involved in inflammatory processes.

The platelet factor 4 variant 1 (CXCL4L1) gene is located in the chemokine cluster on chr 4. CXCL4L1 has strong antiangiogenic effects, impairs tumor growth, and protects against blood–retinal barrier breakdown in diabetes patients (7, 8). Endothelial cells cultured in vitro produce significant amounts of CXCL4L1 after stimulation with IL-1β and TNF-α (TNF) (9). Several reports have established the protein structure, subcellular localization, and protein–protein interactions of CXCL4L1; however, the mechanisms underlying the regulation of CXCL4L1 gene expression and the details of its 5′ upstream genomic region have not yet been reported (10–14).

The Genetics of Kidneys in Diabetes (GoKinD) study was initiated to identify genetic determinants of DN in patients with type 1 diabetes (T1D) (15). The GoKinD collection included two subsets of T1D singletons: one with impaired renal status and another with normal renal status despite long-term diabetes. In addition, both parents were included to form family trios. DNA samples and clinical information, including estimates of renal function, are available for each study participant. In clinical practice, renal function is assessed by the glomerular filtration rate, defined as the volume of plasma cleared of creatinine per unit of time (milliliters per minute). For the GoKinD study, estimated glomerular filtration rates (eGFR) were calculated based on the formula by the Modification of Diet in Renal Disease Study Group, including adjustments for age, ethnicity, and sex (15). An eGFR ≥ 90 ml/min/1.73 m² with no other evidence of kidney damage is considered to indicate normal renal function. Thus far, the GoKinD study has led to the identification of a number of genes associated with persistent proteinuria or end-stage renal disease (ESRD) defined by chronic dialysis or kidney transplant (16–19). However, any possible genetic association with measures of kidney function has yet to be reported.

In the current study, we used the GoKinD DNA collection and genotyped 55 genetic variants in 16 candidate chemokine genes that were either located in the chemokine cluster of chr 4, which was linked to measures of kidney function in a genome-wide association study (5) or implicated in renal inflammation (20–22). Three markers located within the promoter of the platelet factor 4 variant 1 (PF4V1/CXCL4L1) gene on chr 4 were associated with kidney function (eGFR) in T1D participants. Moreover, the regulation of CXCL4L1 and its downstream signaling in mesangial cells implied a crucial role for SMAD7 and IκBα, which are key inhibitory factors in renal inflammation. To our knowledge, this study is the first to identify an association between CXCL4L1 variants and kidney function in patients with T1D.

A detailed description of the GoKinD collection has been published previously (15). DNA samples were randomly selected from white non-Hispanic participants from the GoKinD collection, including T1D patients and parental participants without diabetes (noT1D). All the subjects in the GoKinD study provided informed consent to participate in the study. The procedures were carried out in accordance with the Declaration of Helsinki II and approved by the local ethics committees. The material transfer agreements for the cohort of the GoKinD study were signed before the experiments were performed. The demographic and clinical information of the participants are provided in Table I. In the GoKinD study, eGFR was calculated according to the Modification of Diet in Renal Disease Study equation, which applies adjustments for factors, including sex, race, and age and provides the most accurate information possible for renal function.

Eighty-nine variants in 16 genes associated with inflammation were initially selected using criteria such as their position within a gene and the global major allele frequency (MAF; i.e., MAF ≥ 0.1), among which only 55 SNPs passed quality control to be uniquely amplified (99% confidence) and were selected for final genotyping. Genotyping was performed via medium-throughput genotyping with the iPLEX Gold assay and SpectroCHIP Arrays (Sequenom, San Diego, CA) as previously described (23).

SAS version 9.4 (SAS Institute, Cary, NC) was used for all the statistical analyses presented in Tables I and II. The data are presented as the means ± SDs or as percentages for categorical variables. Nominal p values < 0.05 were considered statistically significant for each single test. The adjusted statistical significance of p values < 0.04149 was calculated using the Sidak method for 55 performed tests. The MIXED procedure and residual plots were used to confirm the primary assumptions of normal distributions and linear means in the data. The MIXED procedure was also used to assess the associations between the genotyped SNPs and eGFR levels. Therefore, the allele variants were set as fixed effects in the statistical model. In cases of significant associations, post hoc tests were performed using least square estimators to detect the relevant alleles and calculate the estimated eGFR difference. The p values from the multiple comparisons were adjusted using the Tukey–Kramer method.

The results of reporter gene and expression assays were analyzed using two-tailed Student t tests and one-way ANOVA with post hoc t tests for multiple comparisons.

The CXCL4L1 promoter sequences were amplified using human high–molecular weight DNA and the following primers: 5′ primer, 5′-TCTTACGCGT GCTAGCCCAA TAATGATTAA AGAG-3′, and 3′ primer, 5′-GATCTCGAGCCCCCCACTGGTCCTTCCAG-3′.

The PCR fragments were precipitated and digested with the NheI and XhoI restriction enzymes. The p-2000 plasmid was constructed by inserting the digested PCR fragment into the NheI and XhoI sites of pTA-Luc (Clontech, Mountain View, CA). The insert was analyzed by performing a restriction analysis. The p-1425 plasmid was constructed by deleting Eco53K1/EcoRV from p-2000. The SNP mutations were introduced to p-1425 using a Q5 Site-Directed Mutagenesis Kit (New England Biolabs, Ipswich, MA). The final constructs were sequence verified. The CXCL4L1 coding region was amplified by RT-PCR using the 5′-ACTGGGATCCTGCTGGAAC-3′ and 5′-GGATCCTAGTAGCTAACTC-3′ primer set. The resulting fragment was digested with BamHI and inserted into the BamHI linearized pCDNA3 vector (Invitrogen, Carlsbad, CA) to obtain pCXCL4L1.

Transformed human HEK293 cells and normal human mesangial cells (NHMCs) were obtained from American Type Culture Collection (Middlesex, U.K.) and Lonza (Walkersville, MA), respectively. NHMCs and HEK293 cells were cultured and transfected according to the manufacturer’s instructions. NHMCs between the sixth and eighth passage were used for all the experiments. Briefly, a total of 10⁵ × NHMCs were seeded in 5 ml of mesangial growth basal medium supplemented with 5% FBS. After 8 h, the cells were incubated overnight in serum-free medium. The following day, the cells were incubated in medium supplemented with 5% FBS under normal glucose (NG; 5.6 mM), high glucose (HG; 30 mM), or osmotic control (OC; 5.6 mM glucose plus 24.4 mM mannitol) conditions. The culture medium was replaced every other day. For the small interfering RNA (siRNA) experiments, the cells were transfected using Lipofectamine RNAiMAX (Life Technologies, Carlsbad, CA) and incubated under NG, HG, or control conditions for an additional 48 h before harvesting. For the stimulation studies, the cells were kept in NG or HG medium for 72 h before stimulation with human 250 U/10 ng IL-1beta/TNF-α (PeproTech, Rocky Hill, NJ). The siRNA targeting CXCL4L1, for which the sequence was 5′-AGCTCTTAAGCATTAC-3′, and the scrambled siRNA, for which the sequence was 5′-GTACTAGCTCATACTA-3′, were designed using siDESIGN (http:/www.dharmacon.com).

Firefly and Renilla luciferase activities were determined using the Dual-Luciferase Reporter System (Promega, Madison, WI), and the results were normalized as previously described (24).

RNA extraction and quantification were performed using the QIAzol RNeasy and MinElute Cleanup Kit (QIAGEN, Hilden, Germany), the SuperScript III Reverse Transcriptase Kit, a Custom TaqMan Small RNA Kit, TaqMan Gene Expression Assays, and an ABI Prism 7900 HT Detector (Life Technologies). The data were normalized using both PPIA (cyclophilin A) and UBC (ubiquitin C). The results were analyzed with RQ Manager and DataAssist software (Life Technologies).

Protein extraction and Western blotting were performed as previously described (25). The CXCL4L1 protein was detected in NHMC supernatants using the CXCL4L1 ELISA Kit (LifeSpan Biosciences, Seattle, WA). A Smad7 rabbit polyclonal Ab (Invitrogen) and IκBα and GAPDH rabbit polyclonal Abs (Cell Signaling Technology, Leiden, the Netherlands) were used for the detection of Smad7 and IκBα, respectively, in NHMC extracts. In the Western blot analyses, we assessed protein levels using a peroxidase–anti-peroxidase soluble complex Ab and direct autoradiography (Sigma-Aldrich, Munich, Germany). ImageJ 1.38x software (https://imagej.nih.gov/ij/) was used to quantify the detected protein signals.

The study population, whose characteristics are provided in Table I, was extracted from white non-Hispanic participants, including 718 T1D patients and 822 noT1D. We selected 11 characterized chemokine genes from the chemokine cluster on chr 4 and 5, additional chemokines that were previously implicated in renal inflammation (20–22). Among all validated gene markers in the SNP database (https://www.ncbi.nlm.nih.gov/SNP/), we selected 89 variants based on the integrity of the provided information, their position within the gene, and the reported MAF. The assay design software disqualified 31 of the markers, whereas 55 were qualified, passed the quality control criteria, and did not deviate from Hardy–Weinberg equilibrium (Supplemental Fig. 1). The genotyping results were then analyzed using mixed linear models to determine the associations of different genotypes with the level of eGFR in unrelated T1D participants. Among the 55 genotyped markers, we identified statistically significant evidence of associations between three markers (rs872914, rs941757, and rs941758) and the eGFR in T1D patients (Table II, p values 0.0125–0.0044). The p values for these markers are lower than the nominal smallest value of the significance level (0.05) and the corrected value of the significance level for multiple testing (0.04149). The noT1D group, including the unrelated parents of the T1D group, was analyzed independently as an unstructured population. Using the same statistical model as used for the T1D group, we found no significant associations between any of these variants and the level of eGFR in the noT1D group (Table II). In the T1D patients, the highest mean eGFR was associated with rs872914 A/A, rs941757 G/G, and rs941758 A/A (Table II). According to least square estimators, eGFR levels differed significantly between the variant alleles, ranging from 9.3 to 11.1 ml/min/m² (Table II, ΔeGFR). All three of the markers were validated in the HapMap and 1000 Genomes Projects and were located within the 5′ region of the platelet factor 4 variant 1 (PF4V1/CXCL4L1) gene, at positions nt −912 (rs872914), nt −168 (rs941757), and nt −72 (rs941758) relative to the transcription start site, with a linkage disequilibrium of r² > 0.97. Interestingly, two markers, at nt −1428 (rs3775487) and −676 (rs3756072) positions in this region, presented no significant associations, emphasizing the relevance of the −912, −168 and −72 markers. The haplotype with the alleles (−912A, −168G, −72A) was also associated with eGFR in T1D (Table II, p value 0.0105) revealing a higher p value than −912A, but lower than that of −168G or −72A alone (Table II, p values 0.0044–0.0125). These findings prompted us to elucidate the role of the rs872914, rs941757, and rs941758 variants in CXCL4L1 gene regulation. The minor allele frequencies of these variants were similar in both the T1D and noT1D groups (Table II).

To the best of our knowledge, the CXCL4L1 5′ untranscribed region has not been characterized before; thus, neither the boundaries nor the function of the CXCL4L1 gene promoter were known. To determine the potential functional consequences of the associated variants, we sought to first delineate the CXCL4L1 promoter region. In HEK293 cells, we tested four reporter plasmids containing 2000, 1425, 608, or 135 bp of the CXCL4L1 5′ gene region fused to the firefly luciferase reporter gene (Fig. 1A, p-2000, p-1425, p-608, and p-135). In contrast to p-608 and p-135, we observed significantly increased reporter gene activity for p-2000 and p-1425 (Fig. 1B). Previously, the combination of IL-1 and TNF-α (IL-1/TNF-α) was shown to increase the endogenous expression of the CXCL4L1 gene in human cell lines (7). Accordingly, the expression of p-2000 and p-1425 was significantly induced when we treated cells with IL-1/TNF-α (Fig. 1B). The uniform responses of p-2000 and p-1425 in unstimulated and IL-1/TNF-α–stimulated cells suggested that p-1425 comprises the entire CXCL4L1 gene promoter. Therefore, we used the p-1425 reporter to investigate the effects of the rs872914, rs941757, and rs941758 markers located at nt −912, −168, and −72 relative to the CXCL4L1 transcription start site, respectively. The SNP rs3775487 was not further analyzed because it is located outside of CXCL4L1 promoter, at position −1438, and showed no association with eGFR (Table II). Another SNP, at position −676 (rs3756072), which had also been genotyped but was not associated with eGFR (Table II), was included in the reporter gene assay because of its location within the functional CXCL4L1 promoter. The p-1425 reporter comprising the nt −912G, −676A, −168C, and −72C variants was set as a control vector (Fig. 1C, control). The reporter variants contained single mutations (−912G to A [p−912/A], −676A to G [p−676/G], −168C to G [p−168/G], or −72C to A [p−72/A]), or a triple mutant (p−912A/−168G/−72A). The −676A to G mutation had no effect on p-1425 reporter activity (Fig. 1C). Compared with the control, the −72A variant increased basal reporter expression (control and p−72/A, white bars), whereas the −168G variant increased IL-1/TNF-α-stimulated reporter activity (control and p−168/G, black bars). The −912A variant increased both basal and IL-1/TNF-α–stimulated reporter gene activities. Despite these differences, compared with the −912/A variant alone (p−912/A), the −912A, −168G, and −72A variants together (p−912A/−168G/−72A) increased CXCL4L1 promoter activity slightly, suggesting a primary role of the −912 SNP (Fig. 1C).

Because glomerular mesangial cells are known to produce a variety of chemokines in response to glucose stimulation, we next investigated CXCL4L1 expression in commercially available NHMCs (26). Because of the limited survival time of NHMCs in vitro, various batches of NHMCs were required to accomplish the following experiments. The CXCL4L1 gene variants of NHMCs were ND in this study; however, they were derived from various healthy donors, who did not suffer from T1D. As shown in Fig. 2A, 2B, CXCL4L1 mRNA and protein levels were increased in response to HG treatment compared with NG or OC conditions. In NHMCs stimulated with IL-1/TNF-α, the mRNA and protein levels of CXCL4L1 increased across all treatment conditions. However, the fold activation was higher under the NG and control conditions than under HG treatment. CXCL4L1 transcript and protein levels displayed similar trends across the tested conditions, suggesting that glucose may regulate gene expression at the transcriptional level (Table III).

The role of CXCL4L1 in renal metabolism remains unknown. To determine the role of this chemokine in cellular signaling, we knocked down CXCL4L1 expression in NHMCs and measured the protein levels of Smad7 and IκBα, which have been demonstrated to confer protection against renal inflammation (3, 27). As shown in Fig. 3A, the addition of CXCL4L1-specific siRNA (CXCL4L1-siR200) effectively downregulated CXCL4L1 protein levels under NG and HG conditions when compared with a scrambled sequence control. The highest degree of CXCL4L1 knockdown was achieved 2 d posttransfection (Fig. 3B). Downregulation of CXCL4L1 expression corresponded to reduced protein levels of Smad7 and IκBα compared with those of the scrambled sequence or no-treatment control conditions (Fig. 4A). The magnitude of the reduction was greater under HG conditions compared with NG conditions. Conversely, treatment with a CXCL4L1-encoding plasmid (pCXCL4L1) increased Smad7 and IκBα transcript levels compared with those of a control plasmid (Fig. 4B). Together, the data suggest that CXCL4L1 regulates the levels of the Smad7 and IκBα proteins in human kidney mesangial cells.

The main findings of this study were that promoter variants within the CXCL4L1 gene are associated with a higher eGFR in T1D patients from the GoKinD collection and that this effect may be mediated through the activation of Smad7 and IκBα. We also found that endogenous CXCL4L1 expression increased in response to HG levels and inflammatory cytokines. Taken together, to our knowledge, the results of this study provide, the first evidence that CXCL4L1 expression exerts a protective effect on renal function in patients with T1D and that this relationship is modified by genetic factors.

T1D and type 2 diabetes (T2D) are complex diseases characterized by dysregulation of blood glucose levels resulting from various insulin signaling defects. The glucose-mediated effects of CXCL4L1 raise the question of its possible association with nephropathy in T2D. Previous genome-wide linkage analyses have identified multiple loci with small to moderate effects on the risk of kidney disease in T2D, but none of these loci were located on chr 4 (28). Thus, further studies in T2D cohorts will be required to verify the protective effect of CXCL4L1 on renal function in T2D. However, there are various physiological distinctions underlying disease development and complications in T1D or T2D. For instance, the underlying pathophysiology driving an increased risk of cardiovascular complications in T1D is partly related to nephropathy and appears to be distinct from the pathophysiology of cardiovascular complications in T2D (29). Thus, distinguishing between genetic versus secondary renal complications in T2D and T1D will remain a complex study model.

The GoKinD collection has been used in various genetic association studies to address genetic susceptibility to DN, which is diagnosed by persistent proteinuria or ESRD (16–19). None of these studies assessed eGFR levels in particular or identified associations between the risk of DN and the CXCL4L1 gene. In addition, we did not find evidence of an association between CXCL4L1 promoter variants and DN, which appears for T1D participants only (Table III). Thus, the protective effects of CXCL4L1 promoter variants are relevant to the renal function of T1D patients, but they may play a minor role in the late clinical manifestations of DN.

In our study, the eGFR of T1D patients was most significantly associated with three SNPs within the functional CXCL4L1 promoter. Only the variants that increased CXCL4L1 promoter activity were associated with elevated eGFR levels, suggesting a positive effect of CXCL4L1 on renal function. The impact of the CXCL4L1 alleles on renal function ranged from 4.1 to 11.1 ml/min/m² in our study sample (ΔeGFR, Table II). This difference is likely relevant because the mean eGFRs in the T1D and noT1D populations differed by 4.5 ml/min/m² (Table I). The five clinical stages of chronic kidney disease are classified according to eGFR levels, and an eGFR ≥ 90 ml/min/1.73 m² is considered normal. Interestingly, when the eGFR level was set as a qualitative trait, with ≥90 representing normal and <90 representing impaired kidney function, only the SNP nt −912 AA variant was associated with normal kidney function (Table III).

To our knowledge, this study is the first to determine the promoter boundaries of the CXCL4L1 gene and to investigate its expression in human kidney cells. The 1425-nt-long fragment upstream of the transcription start site of CXCL4L1 constitutes the promoter region that regulates the level of CXCL4L1 gene expression. Single mutations in the nt −912, −168, and −72 SNPs showed different enhancing effects on CXCL4L1 promoter. However, the simultaneous nt −912A, −168G, and −72A mutations did not exhibit an additive or synergistic effect beyond that of each single mutation. Moreover, the haplotype containing all three protective alleles together revealed a similar association with the eGFR in T1D compared with single protective alleles. This may suggest that all protective alleles contribute to a common transcription activation mechanism. Initially, we hypothesized that the three SNPs affected the binding sites of different transcription factors. According to an analysis of the CXCL4L1 promoter sequence in the TRANSFAC database, numerous binding sites exist for regulatory factors, such as AP-1, MyoD, and C/EBP (data not shown). However, none of the variants affected the cognate binding sites of known transcription factors. Interestingly, the −676 SNP showed no relevant effect on CXCL4L1 promoter activity in reporter gene assays and was not associated with the eGFR level in T1D. Similarly, the −1428 SNP, located outside of the promoter region, presented no significant association. Both SNPs exhibited smaller MAFs compared with other SNPs in this region. Although speculative, this finding may hint at some possible genetic variability, despite the high linkage disequilibrium across this region. Thus, additional experimental studies are required to define the mechanistic role of these SNP variants in the formation of a transcriptional complex involving the CXCL4L1 gene.

The importance of CXCL4L1 in renal function in T1D was further supported by the experiments using NHMCs, which exhibit only a limited survival time in vitro but play an important role in triggering glomerular inflammation (30). First, endogenous CXCL4L1 expression was elevated under HG conditions and was markedly induced by the inflammatory cytokine IL-1/TNF-α. The HG-mediated fold increase was lower than the increase mediated by IL-1/TNF-α stimulation alone, suggesting that two distinct activation mechanisms are involved. CXCL4L1 mRNA is highly stable in human platelets, and the secretion of the CXCL4L1 protein is regulated in lymphoid cells (31, 32). In NHMCs, however, CXCL4L1 mRNA and secreted protein levels present comparable expression patterns, excluding the possibility of differential translational regulation of CXCL4L1 mRNA. Second, CXCL4L1 increase in NHMCs supernatant elevated the expression levels of two prominent inhibitors of kidney inflammation: Smad7 and IκBα. Human mesangial cells were previously shown to express CXCR3, which is the most investigated receptor of CXCL4L1 (21, 33, 34). Although CXCR3 signaling remains to be investigated in NHMCs, the activation of Smad7 and IκBα by CXCL4L1 was not markedly high in these cells or depending on phosphorylation of the TAK1 or IKK signaling pathways, which are known to induce Smad7 and IκBα highly and transiently (Supplemental Figs. 2, 3) (3, 35). Thus, HG and CXCL4L1 overexpression likely induces a weak but prolonged activation of the Smad7 and IκBα genes in NHMCs. Compared with the limited cultivation of primary cells in vitro, however, the cumulative expression of CXCL4L1 and its anti-inflammatory effects could be more evident in vivo. In a parallel set of experiments, CXCL4L1 expression was not found to be significantly regulated by glucose treatment of renal proximal tubule or cortical epithelial cells (data not shown). However, chemokine/cytokine cross-talk is likely to occur based on observations that podocyte injury frequently results in mesangial cell proliferation, whereas mesangial cell injury leads to proteinuria (36). Previously, CXCL4L1 was found to inhibit the diabetes-induced proliferation of retinal microvascular endothelial cells (7). Thus, depending on the cell type and function, CXCL4L1 may stimulate disparate biological pathways.

Although Smad7 and IκBα levels were increased by CXCL4L1 under NG conditions, the CXCL4L1-enhancing variants did not exhibit a strong association with eGFR levels in the participants without T1D. This finding may partially be due to the statistical model’s stringency and limited representation of the study participants without T1D. For instance, the association between eGFR and the protective haplotype almost reached statistical significance in the noT1D subpopulation (Table II, p value 0.0636). It is conceivable that the anti-inflammatory function of CXCL4L1 in mesangial cells is less prominent or absent in other renal cells under NG conditions, which could contribute to a moderate protective effect of CXCL4L1 in the absence of T1D. Therefore, a larger population-based study may be able to reveal a protective role of CXCL4L1 against renal inflammation in the noT1D population.

In the current study, we provide insights regarding the function of CXCL4L1 in NHMCs. We conclude that CXCL4L1 can restrain the activation of inflammatory genes via Smad7 and IκBα. Despite the strong association between the CXCL4L1 gene and renal function in T1D, the function of mesangial cells may represent only a part of the protective effect of CXCL4L1 expression on renal function. Recently, CXCL4L1 has been shown to induce monocytes to differentiate into distinct macrophage phenotypes different from M1 and M2 in vitro (34). Thus, CXCL4L1 expression is also likely to affect the response of resident immune cells in the kidney. Because species other than primates lack a homologous CXCL4L1 gene, investigating the systemic impact of CXCL4L1 on the progression of DN in vivo is challenging.

We thank Kimberly Yeatts for assisting with the initial Illumina genotyping experiments and Dr. Christopher M. Kingsley for supporting the power calculations for the study sample size. We also thank Birgit Ritter and Regina Ax-Smolarski for their excellent technical assistance.

The authors have no financial conflicts of interest.