A null variant in the apolipoprotein L3 gene is associated with non-diabetic nephropathy

Abstract

Background

Inheritance of apolipoprotein L1 gene (APOL1) renal-risk variants in a recessive pattern strongly associates with non-diabetic end-stage kidney disease (ESKD). Further evidence supports risk modifiers in APOL1-associated nephropathy; some studies demonstrate that heterozygotes possess excess risk for ESKD or show earlier age at ESKD, relative to those with zero risk alleles. Nearby loci are also associated with ESKD in non-African Americans.

Methods

We assessed the role of the APOL3 null allele rs11089781 on risk of non-diabetic ESKD. Four cohorts containing 2781 ESKD cases and 2474 controls were analyzed.

Results

Stratifying by APOL1 risk genotype (recessive) and adjusting for African ancestry identified a significant additive association between rs11089781 and ESKD in each stratum and in a meta-analysis [meta-analysis P  =  0.0070; odds ratio (OR) = 1.29]; ORs were consistent across APOL1 risk strata. The biological significance of this association is supported by the finding that the APOL3 gene is co-regulated with APOL1, and that APOL3 protein was able to bind to APOL1 protein.

Conclusions

Taken together, the genetic and biological data support the concept that other APOL proteins besides APOL1 may also influence the risk of non-diabetic ESKD.

INTRODUCTION

The G1 and G2 coding variants in the apolipoprotein L1 gene (APOL1) on chromosome 22q13 are associated with a spectrum of non-diabetic etiologies of chronic kidney disease (CKD) [1–4]. This observation altered our understanding of the pathogenesis of nephropathy in populations with recent African ancestry. The biology of the APOL1 gene product and the mechanisms whereby high-risk APOL1 variants cause kidney disease remain incompletely elucidated. Initial structural characterization of the C-terminal coiled coil domains of G0 (wild-type), G1 and G2 APOL1 proteins has recently been reported [5]. In addition, APOL1 kidney disease risk variant cytotoxicity is accompanied by mitochondrial dysfunction [6], as well as cellular potassium depletion with subsequent induction of stress-activated protein kinases [7].

APOL1-associated nephropathy risk is inherited in an autosomal recessive pattern, such that risk of kidney disease is greatly increased in individuals possessing two copies of the G1 and/or the G2 nephropathy variants [1]. Limou et al. proposed a multimer model for the cellular toxicity of APOL1 protein in order to reconcile toxic gain-of-function mutations with autosomal recessive inheritance [8]. Nonetheless, individuals with a single APOL1 nephropathy variant may have small but significant increases in disease risk and initiate renal replacement therapy at earlier ages than do those without APOL1 nephropathy variants [1, 9–11]. Moreover, most African Americans who inherit two APOL1 nephropathy variants do not develop CKD [1, 12, 13].

Factors in addition to APOL1 G1 and G2 genotypes appear to modulate risk for APOL1-associated kidney disease. Variation at genetic loci near APOL1 may contribute, a hypothesis supported by the association of markers in neighboring genes with CKD in European and Asian ancestry populations [14–17]. Association between end-stage kidney disease (ESKD) and an APOL3 null variant (nonsense mutation rs11089781) was initially described by Tzur et al. in African Americans from New York [2]. Subsequently, Next Generation DNA sequencing in African Americans from the southeastern USA revealed an interaction between APOL1 and APOL3 rs11089781 in individuals with fewer than two APOL1 G1/G2 renal-risk alleles [18]. This association was not statistically significant in African American participants in the Family Investigation of Nephropathy and Diabetes (FIND) study, though the odds ratio (OR) was consistent in direction. To gain a more complete understanding of the complex genetics involved in the contribution of the APOL3 gene to kidney disease, additional cohorts of non-diabetic ESKD cases and controls with APOL1 and APOL3 genotype data were evaluated. Cell-based, gene regulation and protein co-localization studies were also performed in order to gain further insight into the underlying biological mechanisms.

MATERIALS AND METHODS

Study populations

Four study sample sets were analyzed and included in the meta-analysis. FIND participants included 807 African American cases with non-diabetic etiologies of ESKD [HIV-associated nephropathy (HIVAN), focal segmental glomerulosclerosis (FSGS), primary or systemic forms of chronic glomerulonephritis, hypertension-attributed and other etiologies] and 634 African American controls [18, 19]. Wake Forest participants included 1559 African American cases with non-diabetic ESKD attributed to FSGS, focal global glomerulosclerosis, HIVAN, chronic glomerulonephritis without a kidney biopsy, hypertension or unknown causes, along with 1332 controls with a serum creatinine  <1.5 mg/dL (men) or <1.3 mg/dL (women), and (when available) urine albumin:creatinine ratio  <30 mg/g [18]. New York Hispanic Americans admixed with recent African ancestry included participants whose families came primarily from the Dominican Republic and Puerto Rico (83 non-diabetic ESKD cases, 365 non-nephropathy controls) [2, 20]. New York African Americans included 332 non-diabetic ESKD cases and 143 controls [2, 20]. Cases with ESKD in both New York ethnic groups included patients with HIVAN, non-monogenic forms of FSGS and kidney disease attributed to hypertension in the absence of other known causes.

All four sample sets were genotyped for nephropathy risk alleles APOL1 G1 (rs73885319; rs60910145) and APOL1 G2 (rs71785313), for APOL3 null allele rs11089781 (Gln58Ter), and for ancestry-informative markers to estimate proportion of African ancestry. Methods and quality control from each study have been reported [1, 2, 18–20].

Statistical analyses

To test for an association of the APOL3 null mutation rs11089781 with ESKD status, a logistic regression model was computed, stratified by APOL1 renal risk genotype status and adjusted for global percentage African ancestry as a covariate. Specifically, the model was stratified into two groups, with the first stratum defined as individuals with 0 or 1 copy of the G1 or G2 APOL1 risk alleles and the second stratum defined as individuals with 2 copies of the G1 and/or G2 APOL1 risk alleles. The effect of rs11089781 was analyzed in an additive model, in which risk increases with the number of APOL3 stop codon variants A. The prevalence of the 2 stop codon (variant A) genotype was extremely low in New York Hispanic samples (only four homozygotes), limiting our ability to analyze a recessive model. The logistic regression models were computed separately in each of the four data sets, as the sample characteristics were different. To test the hypothesis of association across cohorts, inverse normal meta-analyses were computed [21] without or with weighting by the square root of sample size. The corresponding weighted and unweighted OR and 95% confidence intervals (95% CIs) were computed. Finally, the two APOL1 risk strata were combined via the weighted and unweighted inverse normal meta-analysis method. A two-sided P-value is reported.

Cells and reagents

Primary human coronary arterial endothelial cells (HCAEC) were purchased from Lonza (Basel, Switzerland), and experiments were performed at passage 8. HCAEC were cultured in EGM-2 media with MV Bulletkit/5% FBS (Lonza). SV40 T antigen and hTERT virus-immortalized human podocytes were cultured in RPMI/10%FBS/pen-strep/ITS (kind gift of Dr Jeffrey Kopp). Human umbilical vein endothelial cells (HUVEC) were purchased from Lonza, cultured in EGM-2 with Bulletkit/2% FBS (Lonza), and experiments were performed on cells at passage 5–6. Human embryonic kidney (HEK293) cells were purchased from ATCC (Manassas, VA, USA) and cultured in Eagle’s minimal essential media with 10% FBS and pen/strep mix. Interferons were purchased from PBL Interferon Source (Piscataway, NJ, USA) or eBioscience (San Diego, CA, USA).

mRNA expression

mRNA was isolated from cells using the RNeasy kit (Qiagen, Venlo, The Netherlands), reverse transcribed into cDNA using Taqman Reverse Transcription Reagents (Applied Biosystems, Grand Island, NY, USA), and quantified by real-time PCR using an Applied Biosystems 7500 PCR system (Grand Island, NY, USA). Expression of the mRNA expression was normalized to either 18S subunit or β-actin expression. Real-time primer-probes from Applied Biosystems were: APOL1 Hs00358603_g1, 18S 4319413E, β-Actin 4326315E.

Bioinformatic studies

APOL expression data in lymphoblasts was obtained from Stranger et al. [22] and APOL expression data in human glomeruli was obtained from Nephroseq (www.nephroseq.org). Calculations were performed using log2 transformed data.

Co-transfection and lysate collection

HEK293T cells were grown to 70–80% confluence prior to transfection with APOL1 and C-terminal myc-FLAG-tagged APOL3 plasmids (Origene RC220066; Rockville, MD, USA). Co-transfection was performed using Lipofectamine^® 2000 and 20 µg of each plasmid, according to the manufacturer’s protocol (Life Technologies #11668027, Grand Island, NY, USA). Opti-MEM^® media was brought to a final concentration of 10% FBS at 4 h post-transfection. Cells were lyzed and collected in Brij99 detergent 24 h post-transfection.

Co-immunoprecipitation

Lysate samples containing 750 µg of protein were pre-cleared in a 1:1 mixture of Protein A and Protein G Dynabeads^® (Life Technologies 10001D & 10003D). Pre-cleared supernatant was isolated and incubated at 4°C overnight with 3 µg of the immunoprecipitation (IP) antibody [anti-DDK Origene TA500011 or anti-APOL1 Sigma Aldrich HPA018885 (St Louis, MO, USA)]. Samples were incubated with DynaBeads^® for 1 h rotating at 4°C to bind the IP antibody, and beads were washed 5× with Brij99 buffer prior to eluting the antibody and its accompanying proteins off of the beads in lithium dodecyl sulfate (LDS) non-reducing buffer (Thermo Scientific #84788, Pittsburgh, PA, USA) at 70°C for 10 min. The vector backbone for the APOL1 clone was pCMV-AN-HA and the vector for APOL3 was pCMV6-Entry (Origene). The APOL1 sequence was generated to match NCBI Reference Sequence NP_001130013.1; genotypes were altered with site-directed mutagenesis. All clones were verified by sequencing.

Western blot detection

IP eluates and lysate control samples were run on denaturing Tris-HCl precast gels and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were blocked in non-fat milk and then incubated in primary antibody overnight at 4°C followed by washing and incubation in secondary antibody for 1 h at room temperature. Proteins were visualized using the SuperSignal West Pico kit (Pierce/ThermoScientific, Pittsburgh, PA, USA). The same antibodies served for immunoblotting and for immunoprecipitation.

Mass spectroscopy

HUVEC cells with either the APOL1 G0/G0 or G1/G1 genotype were stimulated with 10 ng/mL of interferon gamma. After 24 h, the cells were lyzed in Brij99 detergent (1% Brij 99, 20 mM Hepes, 50 mM NaCl, 5 mM EDTA). The lysate was pre-cleared with protein A/G beads, then immunoprecipitation was performed with 3 µg of anti-APOL1 antibody (Sigma). Beads were then washed five times in the Brij99 buffer. Protein was separated by gel electrophoresis and APOL1 pull-down verified by colloidal Coomassie staining. Gel sections were excised and sent to the Harvard University core facility for mass spectroscopy (liquid chromatography-mass spectrometry; LC-MS/MS) using a LTQ Orbitrap Elite (ThermoFisher, Waltham, MA, USA) equipped with a Waters NanoAcuity HPLC system. Assignment of MS/MS spectra was performed using the Sequest algorithm to search the Human Uniprot Database (SwissProt 2014). Data was analyzed using in-house Proteome Browser Suite (PBS) software.

Yeast two-hybrid interactions

DNA constructs coding for amino acids 238–398 (either G0, G1 or G2) were cloned into the pB27 bait vector with the LexA DNA-binding domain upstream (N-terminal) to the APOL1 bait sequence. The prey sequences were derived from a randomly primed human placenta cDNA library for maximum transcript representation. Interaction of bait with prey proteins allowed LexA transcription of His3 (necessary for histidine synthesis) and yeast growth on media lacking histidine. Yeast colonies were sequenced across the prey insert sites to identify the protein fragments encoded.

Immunostaining

HEK293 cells on collagen-coated coverslips were co-transfected with hemagglutinin-tagged APOL1 (transcript variant 4) and FLAG-tagged APOL3 using Lipofectamine 3000. After 16 h, cells were fixed with 4% paraformaldehyde, permeablized with 0.1% Triton X-100, and blocked with goat serum. Cells were probed with mouse anti-FLAG antibody (Sigma) and rabbit anti-HA antibody (Sigma). Secondary antibodies were goat anti-mouse Alexa Fluor 488 Green (ThermoFisher) and goat anti-rabbit Alexa Fluor 555 Red (ThermoFisher). Nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI). Cells were imaged on a Laser Scanning Microscope (LSM) 880 Confocal Laser System.

RESULTS

Genetic analysis

Table 1 summarizes the APOL3 null variant rs11089781 associations with ESKD in four case–control samples enriched for non-diabetic ESKD: FIND non-diabetic African Americans, Wake Forest African Americans, New York African Americans and New York Hispanic Americans. The New York Hispanic American sample originated largely from the Dominican Republic and Puerto Rico, but contained recent African ancestry due to admixture with African Americans [2].

Table 1.

Association of APOL3 single-nucleotide polymorphism rs11089781 with ESKD in four cohorts, stratified by APOL1 risk genotype status and adjusted for admixture proportions

Sample	ESKD	Number	Allele frequency	Odds ratio	95% CI	P-value
APOL1 low-risk genotype group (0 or 1 copy of renal-risk variants)a
Wake Forest AA	Cases	837	0.21	1.12	0.95–1.31	0.1905
	Controls	1136	0.19
FIND AA	Cases	430	0.21	1.02	0.81–1.28	0.8709
	Controls	561	0.20
New York AA	Cases	199	0.19	1.62	1.02–2.56	0.0410
	Controls	126	0.12
New York HA	Cases	65	0.10	2.10	1.04–4.22	0.0380
	Controls	354	0.05
Unweighted meta-analysis	Cases	1531	0.18	1.40	1.12–1.75	0.0029
	Controls	2177	0.14
Weightedb meta-analysis	Cases	1531	0.20	1.29	1.09–1.53	0.0034
	Controls	2177	0.17
APOL1 high-risk genotype (2 copies of renal-risk variants)c
Wake Forest AA	Cases	722	0.40	1.03	0.82–1.29	0.8056
	Controls	196	0.39
FIND AA	Cases	377	0.40	1.20	0.83–1.75	0.3366
	Controls	73	0.36
New York AA	Cases	133	0.38	1.30	0.61–2.76	0.4980
	Controls	17	0.32
New York HA	Cases	18	0.47	3.44	0.80–14.79	0.9650
	Controls	11	0.27
Unweighted meta-analysis	Cases	1250	0.41	1.53	1.00–2.34	0.0486
	Control	297	0.34
Weightedd meta-analysis	Cases	1250	0.40	1.24	0.98–1.56	0.0740
	Controls	297	0.37

AA, African American; HA, Hispanic American.

^aNo evidence of heterogeneity of the odds ratio for unweighted (P-value  =  0.11) or weighted (P-value  =  0.19) analyses.

^bWeighted analysis weights by the square root of the sample size of the cohort with 0 or 1 copy of APOL1 renal-risk alleles.

^cNo evidence of heterogeneity of the odds ratio for unweighted (P-value  =  0.38) or weighted (P-value  =  0.90) analyses.

^dWeighted analysis weights by the square root of the sample size of the cohort with 2 copies of APOL1 renal-risk alleles.

The APOL3 rs11089781 A allele (null allele) was present at higher frequencies in all four ESKD case sets, compared with their respective controls. Global African ancestry proportion approximated 80% in FIND, Wake Forest and New York African Americans, in contrast to 34% and 31%, respectively, in New York Hispanic American ESKD cases and controls. ORs and 95% CIs for the additive effect of the APOL3 rs11089781 stop codon on risk for non-diabetic ESKD, accounting for global ancestry, are reported, stratified by APOL1 risk genotype (two renal-risk alleles, less than two renal-risk alleles) and by cohort. Meta-analyses using the unweighted and sample size-weighted inverse normal method showed a statistically significant additive effect in those with two APOL1 renal-risk alleles and in those with less than two APOL1 renal-risk alleles (Table 1). The OR was  >1.0 in all cohorts and across both APOL1 risk genotype strata, showing an increased risk based on the number of copies of the minor allele.

Given the consistent pattern of risk across APOL1 risk genotype strata, a combined analysis was computed that adjusted for admixture and APOL1 risk genotype (Table 2). Here, both the weighted and unweighted analyses showed a strong statistical association between rs11089781 and ESKD (OR_unweighted =  1.43, P_unweighted = 0.00027; OR_weighted  = 1.29, P_weighted  =  0.0070). Of note, the magnitude of the effect (OR  =  2.34) and statistical significance (P  =  0.0066) was greatest in the New York Hispanic sample, generalizing results beyond individuals with high proportions of recent African ancestry.

Table 2.

Association of APOL3 single-nucleotide polymorphism rs11089781 with ESKD in four cohorts, adjusting for APOL1 risk genotype status and admixture proportions

Sample	ESKD	n	Allele frequency	Odds ratio	95% CI	P-valuea
Wake Forest AA	Cases	1559	0.30	1.09	0.95–1.24	0.2171
	Controls	1332	0.22
FIND AA	Cases	807	0.30	1.07	0.88–1.30	0.4940
	Controls	634	0.22
New York AA	Cases	332	0.27	1.52	1.03–2.26	0.0362
	Controls	143	0.15
New York HA	Cases	83	0.18	2.34	1.27–4.33	0.0066
	Controls	365	0.06
Unweighted meta-analysis	Cases	2781	0.26	1.43	1.18–1.72	0.00027
	Control	2474	0.16
Weightedb meta-analysis	Cases	2781	0.28	1.29	1.12–1.48	0.00701
	Controls	2474	0.20

AA, African American; HA, Hispanic American.

^aNo evidence of heterogeneity of the odds ratio for unweighted (P-value  =  0.11) or weighted (P-value  =  0.09) analyses.

^bWeighted analysis weights by the square root of the sample size of the cohort.

Gene regulation of APOL1, APOL2 and APOL3

Bioinformatic analysis of APOL-region gene expression was performed using HapMap lymphoblast data. Results showed very high correlation between APOL1 and APOL2 transcript levels (R² =  0.8, Supplementary data, Figure S1) and more limited correlation of either gene with APOL3 levels (R² =  0.07–0.11) under baseline conditions. APOL1 transcription is highly responsive to interferon stimulation, and APOL2 and APOL3 promoter analysis predicted a similar response. Specifically, unbiased genome-wide chromatin immunoprecipitation assays from the ENCODE project [23] show interferon-induced signal transducer and activator of transcription 1 (STAT1) and STAT2 binding in the regulatory regions of APOL1, APOL2 and APOL3, suggesting the possibility of coordinated gene regulation.

To test this regulation, HCAEC and podocytes were treated with interferon-gamma (IFNγ), and mRNA transcript levels from all three genes were measured (Figure 1). At early time points, expression of the three APOL genes increased to similar degrees, but prolonged interferon exposure resulted in sustained APOL1 expression relative to APOL2 and APOL3. This response suggests that a high ratio of APOL1 to APOL2 or APOL3 may be a signature of the high interferon state.

FIGURE 1.

Co-regulation of APOL genes by interferon. (Left) Podocytes or (right) endothelial cells were treated with 1 ng/mL IFNγ. APOL gene expression was measured 2, 6 or 24 h after interferon stimulation. APOL1, APOL2 and APOL3 were all upregulated by IFNγ in both cell types. In endothelial cells short-term exposure to IFNγ upregulated APOL mRNAs to similar degrees, but APOL1 expression predominated after prolonged IFNγ exposure.

In order to validate the coordinated regulation of these APOL genes in human kidneys, we used Nephroseq gene expression data from isolated glomeruli to analyze the correlation between APOL transcript levels in health and disease (Table 3). In glomeruli from normal kidneys, APOL1 and APOL2 expression showed only minimal correlation, whereas neither was correlated with APOL3. However, in glomeruli from kidneys with a variety of disease states, expression levels of these APOL genes exhibited stronger correlations. These data strengthen the hypothesis that expression of APOL2 and APOL3 is modulated in concert with changes in APOL1 expression, suggesting a coordinated regulatory region with important functional implications.

Table 3.

Correlation between APOL1, APOL2 and APOL3 gene expression in humans

Kidney disorder	Number	APOL1/ APOL2	APOL1/ APOL3	APOL2/APOL3
Normal	41	0.10	0.01	0.02
FSGS	25	0.30	0.20	0.45
Lupus nephritis	32	0.46	0.31	0.64
IgA nephropathy	27	0.61	0.12	0.44
Membranous nephropathy	21	0.46	0.11	0.14
Minimal change disease	13	0.54	0.09	0.47
Diabetic nephropathy	12	0.56	0.40	0.71
RPGN	23	0.33	0.26	0.34

RPGN, rapidly progressive glomerulonephritis. Human glomerular gene expression profiling data from Nephroseq were used to compare APOL correlate transcript abundance between APOL genes. In kidneys from normal donors, expression correlation is low, whereas disease states drive coordinated regulation. Correlation is shown as the correlation coefficient R².

Physical interaction of APOL1 with APOL3

In parallel experiments, a yeast two-hybrid screen was performed to identify APOL1 binding partners using as bait the C-terminal aa 238–398 of APOL1. APOL3 was identified as an APOL1 binding partner, with the APOL1 C-terminus binding to a polypeptide fragment of APOL3 (APOL3 isoform 1 aa 100–287; Supplementary data, Figure S2). To confirm this interaction, APOL1 and APOL3-FLAG expression plasmids were transfected into HEK293 cells. Different splice variant isoforms of the APOL1 gene product alter the signal peptide and its cleavage and are thus expected to have different intracellular localization and secretory pathway patterns [24]. Accordingly, we used an APOL1 transcript variant (tv)4 lacking a signal peptide in order to express both proteins in the same cellular compartments. APOL1 pull-down co-immunoprecipitated APOL3-FLAG and, conversely, pull-down of APOL3-FLAG co-immunoprecipitated APOL1 (Figure 2). No apparent differences were observed in APOL3 affinity among APOL1 polypeptides of different risk genotypes [G0 (wild-type), G1 and G2].

FIGURE 2.

Physical interaction between APOL1 and APOL3 proteins. (a, b) Co-transfection of APOL1 and flag-tagged APOL3 into HEK293 cells was followed by immunoprecipitation and Western blotting. (a) Pull-down performed on lysates using anti-FLAG antibody (APOL3-FLAG) followed by Western blotting with α-APOL1 antibody (Sigma rabbit polyclonal) demonstrated that APOL3 binds G0, G1 and G2 APOL1. (b) The same lysates and antibodies were used for pull-downs with anti-APOL1 antibody followed by Western blotting with anti-FLAG antibody. APOL1 G0, G1 and G2 all pulled down FLAG-tagged APOL3.

Potential interactions were tested under more physiologic conditions by performing immunoprecipitations in IFNγ-treated endothelial cells. In the absence of antibodies that reliably and specifically bind native APOL3, APOL1 was pulled down with anti-APOL1 antibody and mass spectroscopy was used to detect APOL3 (Table 4). Endothelial cells of APOL1 G0/G0 and G1/G1 genotypes were tested (we were unable to obtain endothelial cells with the less common G2/G2 genotype). Pull-down fractions of both G0 and G1 variants of APOL1 contained peptide spectra unique to APOL3. Thus, a physical interaction was supported in three different experimental systems: yeast two-hybrid, plasmid-transfected HEK cells and IFNγ- induced primary endothelial cells.

Table 4.

Peptide spectra uniquely mapping to APOL3 with APOL1 pull-down

Unique APOL3 peptide spectra
APOL1 WT allele	APOL1 G1 variant
ALANGIEEVHR	ALANGIEEVHR
ARLPVTTWR	ARLPVTTWR
LTATSIDR	LTATSIDR
TYAAIEDEYVQQK	TYAAIEDEYVQQK
	LTATSIDRLK
	VFKEVMR
	DITPNLLSL
	FTEEATKYFR
	IQESIEKLR

To begin probing the nature of the APOL1/APOL3 interaction, we tested whether altering APOL3 levels influenced APOL1-mediated cytotoxicity. We were unable either to rescue APOL1 toxicity in HEK293 cells by co-transfection of APOL3 or to exacerbate APOL1 toxicity in interferon-stimulated endothelial cells via siRNA knockdown of APOL3 (data not shown), possibly due to the need for other upregulated proteins in a larger complex. We performed immunostaining on HEK293 cells co-transfected with hemagglutinin-tagged APOL1 and FLAG-tagged APOL3 in order to identify potential sites of interaction within the cell (Supplementary data, Figure S3). While much of APOL3 trafficked to a perinuclear aggresome-like organelle, both APOL1 and APOL3 are widely distributed throughout the cell and demonstrate broad areas of physical overlap, supporting an interaction taking place in the cell rather than following lysis.

DISCUSSION

We report the detection of a genetic association between APOL3 null variant rs11089781 and non-diabetic ESKD in populations with recent African ancestry. This finding was present regardless of APOL1 risk genotype status and degree of African ancestry, and was found in both African Americans and Hispanic Americans with recent African ancestry from New York. Subsequent experiments revealed physical interaction between APOL1 and APOL3 proteins in kidney cells. Multimerization of APOL proteins influencing nephropathy susceptibility has been proposed [8]. Although APOL3 genetic association may be important in explaining the weaker autosomal dominant inheritance of APOL1 G1 and G2 risk alleles in some reports (compared with the major autosomal recessive inheritance) and the earlier age at ESKD in subjects who possess a single APOL1 G1 or G2 risk allele, the current study also supports an independent effect of this allele, including in individuals with two APOL1 renal-risk alleles.

While the genetic association may assist in refining risk prediction, perhaps its most important implication is for understanding potential biological mechanisms of APOL1-associated kidney disease. Several reports demonstrate enhanced toxicity of the APOL1 G1 and G2 risk alleles, relative to the G0 wild-type APOL1 allele [25–27]. However, these reports show that wild-type APOL1 may not be entirely benign with respect to tissue injury. The small excess risk of disease in APOL1 risk allele heterozygotes from genetic studies is also consistent with gain-of-function toxicity of risk variants. APOL1 protects humans and a few other primate species from pathogens, while APOL1 risk variants appear to protect against a wider range of pathogens, explaining the rapid rise in APOL1 renal-risk variant allele frequency in Africa [1, 27]. However, this protection against an extended spectrum of pathogens may come at the expense of greater propensity to APOL1-mediated tissue injury. We propose that co-expression of other members of the APOL gene family may mitigate the toxic effects of APOL1, those of G0 and of both risk variants [2, 18]. Complete absence of the full-length isoform 1 of APOL3 may render individuals more vulnerable to APOL1-mediated tissue injury (the APOL3 Q58ter variant theoretically allows possible translation of apparently minor APOL3 transcript isoforms 2 and 3), but even functional APOL3 alleles may offer only limited protection when both APOL1 alleles carry G1 or G2 renal-risk variants. The present analyses strengthen the formulation of possible multimerization of APOL proteins in kidney cells, as an explanation for the risk inheritance mode [7]. On the other hand, since we did not observe rescue of APOL1-induced toxicity by APOL3 co-transfection, it is possible that additional proteins are necessary for APOL3-based protection as part of a larger complex, possibly involving one or more of the many other genes that respond to interferon. Another possibility is that APOL3 acts through other mechanisms to increase the risk of CKD and the genetic interaction observed might be attributed to additive effects in an already sensitized population.

Several kidney diseases associated with APOL1 risk alleles, such as HIVAN, FSGS and lupus nephritis, are high interferon states [26]. In fact, interferon itself appears to promote kidney disease in individuals with two APOL1 risk alleles [26]. The data herein show that IFNγ coordinately upregulates the transcription of APOL1, APOL2 and APOL3. This suggests that these genes operate as a functional cassette in response to stimuli characteristic of pathogen attack. Direct physical interaction of APOL1 with APOL3 protein within transfected HEK cells and interferon-stimulated endothelial cells provides additional evidence for functional coordination. Mapping of the interaction interfaces and mechanism may be helpful in future efforts to design therapies which attenuate APOL1-mediated human disease.

Similar to the APOL1 G1 variant, the APOL3 nonsense variant is found only in individuals with recent African ancestry. Although the APOL3 nonsense allele is physically quite distant from the APOL1 risk alleles, it is possible that the APOL3 nonsense allele was carried to high frequency by the same selective sweep that drove the G1 allele to high frequency in Africa [1]. Subsequent recombination appears to have weakened linkage disequilibrium between these loci (D′  =  0.57; R²  = 0.12). However, the APOL3 null allele may have been co-selected with G1, perhaps because G1 may be more efficient in pathogen protection when not bound by APOL3. Evolutionary forces would have governed the balance between such pathogen protection on one hand, and human cell and tissue injury on the other. Further functional studies will be needed to understand the full ramifications of the biological interactions among APOL family members. The results of such studies have the potential to inform treatment strategies directed against APOL1-mediated nephrotoxicity.

In conclusion, human genetic and functional experiments support a modulatory role of the APOL3 gene in nephropathy susceptibility. We postulate that in the presence of a functional APOL3 gene and abundant APOL3 protein, APOL1 G1 or G2 alleles may be less likely to cause kidney disease.

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/ndt.