NAT8 Variants, N-Acetylated Amino Acids, and Progression of CKD

Visual Abstract

Abstract

Background and objectives

Genetic variants in NAT8, a liver- and kidney-specific acetyltransferase encoding gene, have been associated with eGFR and CKD in European populations. Higher circulating levels of two NAT8-associated metabolites, N-δ-acetylornithine and N-acetyl-1-methylhistidine, have been linked to lower eGFR and higher risk of incident CKD in the Black population. We aimed to expand upon prior studies to investigate associations between rs13538, a missense variant in NAT8, N-acetylated amino acids, and kidney failure in multiple, well-characterized cohorts.

Design, setting, participants, & measurements

We conducted analyses among participants with genetic and/or serum metabolomic data in the African American Study of Kidney Disease and Hypertension (AASK; n=962), the Atherosclerosis Risk in Communities (ARIC) study (n=1050), and BioMe, an electronic health record–linked biorepository (n=680). Separately, we evaluated associations between rs13538, urinary N-acetylated amino acids, and kidney failure in participants in the German CKD (GCKD) study (n=1624).

Results

Of 31 N-acetylated amino acids evaluated, the circulating and urinary levels of 14 were associated with rs13538 (P<0.05/31). Higher circulating levels of five of these N-acetylated amino acids, namely, N-δ-acetylornithine, N-acetyl-1-methylhistidine, N-acetyl-3-methylhistidine, N-acetylhistidine, and N2,N5-diacetylornithine, were associated with kidney failure, after adjustment for confounders and combining results in meta-analysis (combined hazard ratios per two-fold higher amino acid levels: 1.48, 1.44, 1.21, 1.65, and 1.41, respectively; 95% confidence intervals: 1.21 to 1.81, 1.22 to 1.70, 1.08 to 1.37, 1.29 to 2.10, and 1.17 to 1.71, respectively; all P values <0.05/14). None of the urinary levels of these N-acetylated amino acids were associated with kidney failure in the GCKD study.

Conclusions

We demonstrate significant associations between an NAT8 gene variant and 14 N-acetylated amino acids, five of which had circulation levels that were associated with kidney failure.

Introduction

The NAT8 gene is specifically expressed in liver and kidney and encodes N-acetyltransferase 8, an enzyme suggested to play a role in detoxification processes by catalyzing the acetylation of cysteine conjugates to form mercapturic acids, which can be excreted in bile and urine (1). Previous genetic studies have identified rs13538, a common missense variant in the NAT8 gene (p.Phe143Ser, NP_003951.3), and other variants within or near NAT8, to be associated with eGFR or incident CKD in European populations (2–5). In Estonians, variants in the promotor region of NAT8 had protective effects against hypertension and kidney failure (6). Several genome-wide association studies (GWAS) of metabolites showed strong associations between NAT8 variants and N-acetylated amino acids (5,7–12). In a community-based cohort of Black participants, higher circulating levels of two N-acetyl amino acids, N-δ-acetylornithine and N-acetyl-1-methylhistidine, were associated with lower eGFR and higher incidence of CKD (13).

The goal of this study was to expand upon prior studies to investigate the associations between NAT8-associated metabolites and kidney failure. Our primary analyses focused on Black participants with CKD, a historically underrepresented group in genetic studies, who had undergone -omics profiling in the African American Study of Kidney Disease and Hypertension (AASK), the Atherosclerosis Risk in Communities (ARIC) study, and BioMe (an electronic health record–linked biorepository). To integrate an additional biologic source, we also evaluated associations between NAT8 variants, urine N-acetyl amino acids, and kidney failure among participants in the German CKD (GCKD) study.

Materials and Methods

Study Population

The study population for serum metabolite analysis consisted of Black participants with blood samples available for metabolomic profiling in AASK (n=962), BioMe (n=680), and ARIC (n=1050) (Supplemental Figure 1). Participants with urine metabolomic data in the GCKD study were included in a secondary analysis of urine metabolites (n=1624).

AASK was a 2×3 randomized trial of BP lowering and antihypertensive medication in Black participants with hypertension-attributed CKD (14). Exclusion criteria included measured GFR <20 or >65 ml/min per 1.73 m², 24-hour urine protein-to-creatinine ratio >2.5 g/g, diabetes mellitus, and known glomerular or autosomal dominant polycystic kidney disease. Randomization occurred between February 1995 and September 1998. Subsequent to trial completion on September 30, 2001, participants without kidney failure were invited to participate in a cohort phase. For our study, a subset of 962 participants who had serum metabolite data at baseline were included. Of these participants, 619 were genotyped (refer to section Identification and Measurement of N-Acetylated Amino Acids).

BioMe is an electronic medical record–linked biobank established in 2007, in which patients receiving health care at the Mount Sinai Health System were enrolled. For our study, we included a sample of 680 Black patients who had an eGFR of 20–75 ml/min per 1.73 m² and serum samples available for metabolomic profiling (15). Genetic data were not available, thus BioMe participants were not included in analyses involving genetics.

The ARIC study is a prospective cohort study of 15,792 individuals between 45 and 64 years of age, from four communities in the United States (16). Visit 1 was carried out between 1987 and 1989, and followed by six subsequent in-person visits and annual telephone interviews (semiannual since 2012). Metabolomic profiling was conducted in two phases (2010 and 2014), using stored serum samples from visits 1 and 5, respectively. For this study, we included Black participants without kidney failure at visit 5 (between 2011 and 2013), and followed them for kidney failure through visit 6 (between 2016 and 2017).

The GCKD study is a prospective cohort study of patients with CKD in Germany, with an eGFR of 30–60 ml/min per 1.73 m² or an eGFR>60 ml/min per 1.73 m² in the presence of a urine albumin-to-creatinine ratio >0.3 mg/g, albuminuria >0.3 mg/d, a urine protein-to-creatinine ratio >0.5 mg/g, or proteinuria >0.5 mg/d at enrollment (17). A total of 5217 adult patients were enrolled between 2010 and 2012. Follow-up visits occurred at 2-year intervals, with structured phone interviews conducted in the interim years. A subset of 1624 participants with eGFR<50 ml/min per 1.73 m² and urine albumin-to-creatinine ratio <30 mg/g, urine metabolite profiling, and nonmissing baseline covariate information were included in this study, for a secondary analysis of urinary N-acetylated amino acids.

Identification and Measurement of N-Acetylated Amino Acids

Frozen serum samples from the baseline (G1) visit in AASK, at enrollment in BioMe, at visit 5 in the ARIC study, and urine specimens collected at enrollment in the GCKD study were sent to Metabolon, Inc (Morrisville, NC) for metabolomic profiling, as previously described (11,15,18,19). Briefly, experimental samples were analyzed using separate untargeted mass spectrometry platforms, which included reverse-phase ultraperformance liquid chromatography tandem mass spectrometry methods using positive ion mode electrospray ionization (ESI), reverse-phase ultraperformance liquid chromatography tandem mass spectrometry methods using negative ion mode ESI, and a hydrophilic interaction ultraperformance liquid chromatography tandem mass spectrometry method with negative ion mode ESI. Experimental features were matched to an in-house spectral and chromatographic library of authentic reference standards using a three-criteria match on retention time/index, mass-to-charge ratio, and chromatographic data (including tandem mass spectrometry spectral data). Metabolite levels were quantified using area under the curve of the mass spectrometry peaks after interday normalization. In quality control, serum samples missing >80% of metabolites were excluded. Metabolites were scaled to a median of 1, then log-transformed and scaled to ln (2), so that each unit change would indicate one-fold change in actual metabolite level. We removed outliers by removing any sample in which any principal component deviated by >5 SDs, and capped any metabolite at 5 SDs above the mean. No urine samples in the GCKD study were removed because of missingness or outliers. GCKD values were adjusted for urine dilution, as described previously (7). For the purposes of this study, we focused on 28 N-acetylated amino acids and three amines, namely, N-acetylcadaverine, (N(1)+N(8))-acetylspermidine, and N-acetylputrescine, that are derived from amino acids (Supplemental Table 1).

Genetic Profiling and Identification of NAT8 Variants

Genotyping of participants in AASK, ARIC, and GCKD have been previously described, and for genetic analyses, we used all participants consenting to genetic research who also had available metabolite profiling (10,11). We selected the single nucleotide polymorphism (SNP) rs13538 as the SNP of interest because it was the most frequently reported SNP in NAT8 with statistically significant associations with N-acetylated amino acids, eGFR, or CKD in the existing literature (2–4,9–11). This SNP is imputed with high imputation quality in AASK (r²=0.99, reference panel: TopMed Build 38) and GCKD (r²=1, reference panel: Haplotype Reference Consortium haplotypes version r1.1), and directly genotyped in ARIC. It is in high linkage disequilibrium (r²>0.8) with other associated SNPs in the NAT8 region, including rs13410232, a directly genotyped SNP in AASK.

Covariables, Kidney Function Measures, and Outcomes

Baseline covariables in this study included age, sex, history of smoking, history of coronary heart disease, history of diabetes mellitus, body mass index, GFR (see next paragraph for details), and urine protein-to-creatinine ratio in AASK or urine albumin-to-creatinine ratio in other cohorts. In AASK, ARIC, and GCKD, baseline covariables were recorded per protocol. In BioMe, covariate data were extracted from electronic health records. Values reported within 1 year before and up to 1 day after baseline and closest to baseline were taken.

In AASK, baseline GFR was measured by the urinary clearance of ¹²⁵I iothalamate, and 24-hour urine protein and creatinine were measured using the pyrogallol red technique and the modified Jaffe reaction, respectively. In BioMe, ARIC, and GCKD, baseline eGFR was calculated on the basis of serum creatinine, using the CKD Epidemiology Collaboration equation (20). Urine albumin-to-creatinine ratios were calculated on the basis of spot urine collection. In AASK and GCKD, kidney failure was ascertained per protocol throughout the study periods. In BioMe and ARIC, kidney failure was ascertained through linkage to the United States Renal Data System in September 2016 and June 2017, respectively.

Statistical Analyses

Baseline characteristics were described using summary statistics. To test associations between rs13538 and N-acetylated amino acids, we used linear regression, adjusting for baseline age, sex, GFR, urine protein-to-creatinine ratio or albumin-to-creatinine ratio (log-transformed to achieve approximate normality), history of smoking (categorical, current versus past versus never), history of coronary heart disease, history of diabetes mellitus, body mass index, and the first ten principal component ancestry scores (generated using the EIGENSOFT package) (21). Correlations between metabolites and GFR were assessed using the Spearman rank-order correlation. Cox regression was used to test associations between N-acetylated amino acids and kidney failure, adjusting for the clinical covariates described above. In a sensitivity analysis, Fine and Gray competing risk regression was used to account for death as a competing event (22). Results across cohorts were combined using inverse variance weighted meta-analysis, with a fixed effects model. The chi-squared test was used to test for heterogeneity in meta-analysis. The significance levels for all statistical analyses were Bonferroni adjusted (for example, to test associations of the 31 N-acetylated amino acids with rs13538, those with P<0.05/31 were deemed statistically significant). The extent of missing data for all variables was described using summary statistics (Supplemental Table 2). All missing data were excluded from statistical analyses. All statistical analyses were performed using R (R Foundation, Vienna, Austria) or Stata/IC 14.2 (Stata Corp., College Station, TX).

Results

Baseline Characteristics

The study population with serum metabolites consisted of 962, 1050, and 680 participants from AASK, ARIC, and BioMe, respectively (Table 1). In AASK and ARIC, 619 and 680 had rs13538 genotype data, respectively. Compared with other cohorts, AASK participants had the youngest mean age (55 [SD 11] years), lowest proportion of females (38%), and lowest mean GFR (46 [SD 13] ml/min per 1.73 m²). The median urine protein-to-creatinine ratio was 80 (interquartile range [IQR], 28–359) mg/g in AASK participants, which was higher than the median urine albumin-to-creatinine ratio in the ARIC participants (12 mg/g), but lower than that in BioMe participants (170 mg/g). The A allele frequency for rs13538 was 0.46 and 0.47 in AASK and ARIC participants, respectively, which was similar to that of the general population of African ancestry (Supplemental Table 3).

Table 1.

Participant characteristics at the time of metabolomic profiling

Variable	AASK Study	BioMe	ARIC Study	GCKD Study
N	962	680	1050	1624
Mean age, yr (SD)	55 (11)	60 (13)	75 (5)	64 (10)
Women	374 (39%)	474 (70%)	685 (65%)	733 (45%)
Black race	962 (100%)	680 (100%)	1050 (100%)	0 (0%)
Mean body mass index, kg/m2 (SD)	31 (7)	32 (8)	31 (7)	30 (6)
Current smoker	280 (29%)	103 (18%)	60 (6%)	198 (12%)
Heart disease	497 (52%)	151 (26%)	105 (10%)	380 (23%)
Mean GFR, ml/min per 1.73 m2a	46 (13)	59 (14)	72 (21)	42 (12)
Median urine albumin-to-creatinine ratio, mg/g (interquartile interval)b	80 (28–359)	170 (70–830)	12 (6–31)	12 (6–31)

Data are presented as n (%), unless otherwise indicated. AASK, African American Study of Kidney Disease and Hypertension; ARIC, Atherosclerosis Risk in Communities; GCKD, German CKD.

^aIothalamate-measured GFR was reported for AASK participants. Otherwise, eGFR (CKD Epidemiology Collaboration equation) was reported.

^bUrine protein-to-creatinine ratio is reported for AASK participants.

Associations between rs13538 and All N-Acetylated Amino Acids

Of the 31 N-acetylated amino acids quantified via untargeted serum metabolomic profiling in AASK, 14 were associated with rs13538 (P<0.05/31; Table 2). All of these were significantly associated with rs13538 in ARIC and in meta-analysis (same statistical significance threshold), with all but N-δ-acetylornithine negatively associated with the A allele of rs13538. These 14 metabolites were all negatively associated with GFR (Supplemental Table 4). There were moderate correlations between the 31 N-acetylated amino acids, except with N-δ-acetylornithine and N-acetylcadaverine (Spearman correlation coefficient of approximately 0.3; Figure 1).

Table 2.

Circulating N-acetylated amino acids with statistically significant associations with rs13538 in the African American Study of Kidney Disease and Hypertension study (additive model, coded allele =A, allele frequency =0.46), and their associations with rs13538 in the Atherosclerosis Risk in Communities study (additive model, coded allele =A, allele frequency =0.47) and meta-analysis, after adjustment for GFR and other covariables

Study	AASK Study			ARIC Study			Meta-Analysisa
Variable	N	β Coefficient (95% CI)	P Value	N	β Coefficient (95% CI)	P Value	β Coefficient (95% CI)	P Value
N-δ-acetylornithine	619	0.69 (0.61, 0.77)	3.18E-49	709	0.78 (0.69, 0.87)	9.94E-52	0.73 (0.67, 0.79)	<0.001
N-acetylcitrulline	619	−0.48 (−0.54, −0.42)	3.20E-46	736	−0.90 (−1.00, −0.80)	6.25E-55	−0.59 (−0.64, −0.54)	<0.001
N-acetylasparagine	619	−0.83 (−0.94, −0.72)	9.86E-44	759	−0.41 (−0.46, −0.36)	5.92E-51	−0.48 (−0.53, −0.44)	<0.001
N-acetylarginine	619	−0.72 (−0.82, −0.61)	1.90E-36	759	−0.47 (−0.53, −0.41)	5.38E-46	−0.53 (−0.58, −0.48)	<0.001
N-acetylglutamine	619	−0.64 (−0.74, −0.54)	8.41E-33	757	−0.42 (−0.47, −0.36)	5.25E-45	−0.47 (−0.52, −0.42)	<0.001
N-acetyl-1-methylhistidine	609	−0.39 (−0.46, −0.32)	1.01E-31	756	−0.74 (−0.83, −0.65)	7.33E-48	−0.52 (−0.58, −0.47)	<0.001
N2-acetyllysine	619	−0.46 (−0.54, −0.40)	1.13E-28	553	−0.63 (−0.72, −0.53)	2.08E-30	−0.52 (−0.58, −0.46)	<0.001
N-acetylleucine	619	−0.74 (−0.89, −0.59)	1.95E-20	752	−0.32 (−0.37, −0.28)	2.19E-40	−0.36 (−0.40, −0.31)	<0.001
N-acetylkynurenine	619	−0.20 (−0.25, −0.14)	1.06E-11	NA	NA	NA	NA	NA
N-acetylphenylalanine	619	−0.32 (−0.44, −0.21)	7.44E-08	759	−0.34 (−0.40, −0.29)	8.85E-31	−0.34 (−0.39, −0.29)	<0.001
N-acetyl-3-methylhistidine	518	−0.15 (−0.19, −0.11)	2.16E-12	685	−0.42 (−0.57, −0.27)	3.28E-08	−0.17 (−0.21, −0.13)	<0.001
N-acetyltyrosine	543	−0.15 (−0.21, −0.09)	2.23E-06	726	−0.31 (−0.37, −0.25)	3.75E-25	−0.23 (0.27, −0.19)	<0.001
N-acetylhistidine	619	−0.30 (−0.42, −0.17)	3.99E-06	721	−0.16 (−0.23, −0.09)	1.01E-05	−0.19 (−0.25, −0.13)	<0.001
N2,N5-diacetylornithine	619	−0.16 (−0.25, −0.07)	0.001	739	−0.22 (−0.30, −0.15)	7.36E-09	−0.20 (−0.25, −0.14)	<0.001

Bonferroni-adjusted significance level: P<0.05/31. Each unit change in β coefficient represents one-fold change (2^β–1=100%, when β=1) in the level of N-acetylated amino acids per copy of rs13538 A allele. Covariables in linear regression models: baseline age, sex, measured GFR, urine protein-to-creatinine ratio in AASK or urine albumin-to-creatinine ratio in ARIC, the first ten principal component ancestry scores, history of diabetes (ARIC only, AASK participants were nondiabetic at baseline), history of smoking, history of coronary heart disease, and body mass index. AASK, African American Study of Kidney Disease and Hypertension; ARIC, Atherosclerosis Risk in Communities; NA, metabolite not available in the study; 95% CI, 95% confidence interval.

^aHeterogeneity statistic was statistically significant for all of the N-acetylated amino acids evaluated in this meta-analysis, except for N-δ-acetylornithine (P=0.14), N-acetylphenylalanine (P=0.76), and N2,N5-diacetylornihtine (P=0.32).

Figure 1.

Circulating levels of many of the 31 N-acetylated amino acids were moderately correlated in the African American Study of Kidney Disease and Hypertension (scale represents Spearman correlations).

Associations of rs13538 and N-Acetylated Amino Acids with Kidney Failure

There were 274, 13, and 26 kidney failure events in AASK, ARIC, and BioMe participants, with median follow-up periods of 8.8 (IQR, 4.7–10.4), 5.5 (IQR, 5.1–6.0), and 6.0 (IQR, 5.3–7.4) years, respectively. Of the 14 N-acetylated amino acids associated with rs13538 in AASK, higher levels of N-δ-acetylornithine, N-acetyl-1-methylhistidine, N-acetyl-3-methylhistidine, N-acetylhistidine, and N2,N5-diacetylornithine were associated with kidney failure in meta-analysis, after adjusting for baseline GFR, proteinuria/albuminuria, and other potential confounders (P<0.05/14; Table 3). Results were similar in competing risk regression analyses, accounting for death as a competing event (Supplemental Table 5). The direct association between rs13538 and kidney failure was not statistically significant in either AASK or ARIC participants, but samples sizes were only 696 and 891, respectively (Supplemental Table 6).

Table 3.

Circulating N-acetylated amino acids that are associated with rs13538 in AASK and their associations with kidney failure in AASK, BioMe, ARIC, and meta-analysis

Study	AASK Study			BioMe			ARIC Study			Meta-Analysisa
Exposure	N (Cases/Controls)	Hazard Ratio	P Value	N (Cases/Controls)	Hazard Ratio	P Value	N (Cases/Controls)	Hazard Ratio	P Value	Hazard Ratio	P Value
N-δ-acetylornithine	274/688	1.37 (1.10, 1.71)	0.005	26/654	2.36 (1.17, 4.79)	0.02	13/696	2.13 (1.01, 4.48)	0.05	1.48 (1.21, 1.81)b	<0.001b
N-acetylcitrulline	274/688	1.17 (1.00, 1.37)	0.06	26/649	1.15 (0.71, 1.87)	0.57	13/723	1.35 (0.66, 2.76)	0.41	1.18 (1.02, 1.36)	0.04
N-acetylasparagine	274/688	1.23 (0.94, 1.61)	0.13	26/654	0.81 (0.47, 1.42)	0.47	13/746	1.53 (0.41, 5.72)	0.53	1.15 (0.90, 1.45)	0.26
N-acetylarginine	274/688	1.26 (0.97, 1.64)	0.08	26/654	1.05 (0.51, 2.16)	0.90	13/759	1.09 (0.35, 3.41)	0.88	1.23 (0.96, 1.56)	0.10
N-acetylglutamine	274/688	1.23 (0.97, 1.57)	0.09	26/651	1.39 (0.68, 2.85)	0.37	13/743	3.10 (0.71, 13.54)	0.13	1.27 (1.02, 1.59)	0.04
N-acetyl-1-methylhistidine	274/672	1.43 (1.20, 1.71)b	6.79E-05b	26/644	1.40 (0.80, 2.47)	0.24	13/743	1.92 (0.79, 4.68)	0.15	1.44 (1.22, 1.70)b	<0.001b
N2-acetyllysine	274/688	1.07 (0.87, 1.31)	0.52	23/367	1.41 (0.83, 2.41)	0.20	12/541	0.93 (0.36, 2.39)	0.88	1.10 (0.91, 1.33)	0.31
N-acetylleucine	274/688	1.40 (0.98, 1.99)	0.06	26/654	1.06 (0.58, 1.82)	0.31	13/739	1.28 (0.28, 5.74)	0.75	1.30 (0.96, 1.74)	0.09
N-acetylkynurenine	274/688	0.97 (0.82, 1.15)	0.75	26/654	0.99 (0.70, 1.41)	0.97	NA	NA	NA	0.97 (0.84, 1.13)	0.73
N-acetylphenylalanine	274/688	1.05 (0.83, 1.33)	0.71	26/654	0.95 (0.35, 2.59)	0.92	13/746	1.94 (0.53, 7.14)	0.32	1.06 (0.85, 1.33)	0.59
N-acetyl-3-methylhistidine	258/552	1.20 (1.06, 1.37)	0.005	26/654	1.31 (0.84, 2.02)	0.23	13/672	1.36 (0.75, 2.48)	0.31	1.21 (1.08, 1.37)b	0.002b
N-acetyltyrosine	236/593	1.20 (1.00, 1.43)	0.05	26/654	0.74 (0.33, 1.65)	0.46	13/713	1.77 (0.48, 6.49)	0.39	1.18 (0.99, 1.40)	0.06
N-acetylhistidine	274/684	1.52 (1.17, 1.97)b	0.002b	26/649	3.19 (1.43, 7.12)	0.005	13/708	2.03 (0.64, 6.41)	0.23	1.65 (1.29, 2.10)b	<0.001b
N2,N5-diacetylornithine	274/688	1.34 (1.10, 1.64)	0.004	26/645	1.68 (0.89, 3.15)	0.11	13/726	7.98 (2.04, 31.20)b	0.003b	1.41 (1.17, 1.71)b	<0.001b

Bonferroni-adjusted significance level: P<0.05/14. Hazard ratio indicates change in risk per one-fold change in the level of N-acetylated amino acids, accounting for death as a competing event. Covariables in Cox models: baseline age, sex, measured GFR, urine protein-to-creatinine ratio in AASK or urine albumin-to-creatinine ratio in ARIC or BioMe, history of diabetes (BioMe and ARIC only, AASK participants were nondiabetic at baseline), history of smoking, history of coronary heart disease, and body mass index. AASK, African American Study of Kidney Disease and Hypertension; ARIC, Atherosclerosis Risk in Communities; NA, metabolite not available in the study.

^aHeterogeneity statistic was not statistically significant for any of the N-acetylated amino acids evaluated in meta-analysis, except for N2,N5-diacetylornihtine (P=0.03).

^bStatistically significant.

Associations of Urinary Levels of N-Acetylated Amino Acids with rs13538 and Kidney Failure in Persons of European Ancestry

All of the 1624 participants in the GCKD study were of European ancestry. The mean age was 64 (SD 10), 45% were female, mean eGFR was 42 (SD 12) ml/min per 1.73 m², and median urine albumin-to-creatinine ratio was 12 mg/g (IQR, 6–31). The A allele frequency of rs13538 in GCKD was 0.79. There were 61 kidney failure events over 4 years of follow-up. The SNP rs13538 was associated with urinary levels of all 14 N-acetyl amino acids that were significant in AASK, each with the same direction of association as that with serum levels in AASK, BioMe, and ARIC (Table 4). However, none of these urinary metabolites were significantly associated with kidney failure.

Table 4.

Associations between urinary N-acetylated amino acid levels and rs13538 and kidney failure in the German CKD study, among those that were associated with rs13538 in blood

Variable	N	β Coefficient for Association with rs13538	P Value	Hazard Ratio for Kidney Failure	P Value
N-δ-acetylornithine	1478	0.97 (0.89, 1.05)	5.88E-53	1.08 (0.83, 1.40)	0.58
N-acetylcitrulline	1598	−0.91 (−0.96, −0.86)	6.56E-47	1.02 (0.74, 1.39)	0.92
N-acetylasparagine	1598	−0.78 (−0.83, −0.72)	5.27E-45	0.67 (0.47, 0.96)	0.03
N-acetylarginine	1598	−0.53 (−0.57, −0.49)	2.14E-37	0.80 (0.51, 1.26)	0.34
N-acetylglutamine	1597	−0.83 (−0.89, −0.76)	5.29E-34	0.71 (0.54, 0.92)	0.01
N-acetyl-1-methylhistidine	1598	−0.75 (−0.81, −0.68)	1.01E-31	1.20 (0.92, 1.58)	0.18
N2-acetyllysine	1597	−0.89 (−0.93, −0.85)	1.13E-28	0.89 (0.62, 1.26)	0.50
N-acetylleucine	1561	−0.41 (−0.47, −0.35)	7.58E-20	0.79 (0.56, 1.12)	0.18
N-acetylkynurenine	1368	−0.48 (−0.60, −0.36)	3.20E-08	0.96 (0.79, 1.16)	0.67
N-acetylphenylalanine	1597	−0.51 (−0.58, −0.44)	4.20E-08	0.96 (0.72, 1.27)	0.77
N-acetyl-3-methylhistidine	1591	−0.86 (−0.98, −0.73)	1.57E-07	1.03 (0.87, 1.23)	0.70
N-acetyltyrosine	1598	−0.66 (−0.73, −0.58)	3.22E-06	0.98 (0.75, 1.27)	0.87
N-acetylhistidine	1598	−0.20 (−0.24, −0.16)	7.54E-06	0.64 (0.38, 1.09)	0.10
N2,N5-diacetylornithine	1595	−0.17 (−0.22, −0.12)	0.001	1.52 (1.04, 2.22)	0.03

Each unit change in β coefficient represents one-fold change (2^β–1=100%, when β=1) in the level of N-acetylated amino acids per copy of rs13538 A allele. Hazard ratio indicates change in risk per one-fold change in the level of N-acetylated amino acids. Covariables: baseline age, sex, eGFR, urine albumin-to-creatinine ratio, the first ten principal component ancestry scores, history of diabetes, history of smoking, history of coronary heart disease, and body mass index.

Discussion

Among participants from well-characterized cohorts with concomitant metabolite and/or genetic profiling, we confirmed that rs13538, a common missense SNP in the NAT8 gene, was significantly associated with the concentrations of multiple N-acetyl amino acids in blood. Associations were present in two cohorts of Black participants using serum metabolites, and one cohort of Europeans using urine metabolites. Higher circulating levels of several rs13538-associated N-acetylated amino acids, including N-δ-acetylornithine, N-acetyl-1-methylhistidine, N-acetyl-3-methylhistidine, N-acetylhistidine, and N2,N5-diacetylornithine, were associated with kidney failure independently of baseline GFR. These findings suggest that the NAT8 gene may have broad functionality, and that higher circulating levels of specific N-acetylated amino acids may portend worse CKD outcomes.

Rs13538 is one of the most commonly studied NAT8 variants in previous genetic studies (2–5). This missense variant causes a nonconservative amino acid change (p.Phe143Ser, NP_003951.3) within the NAT8 acetyl-CoA binding site, which affects NAT8, ALMS1, ALMS1P, and TPRKB gene expression in a variety of tissues, including subcutaneous adipocytes, blood, and the gastrointestinal tract. The associations between rs13538 and several N-acetylated amino acids have already been identified in previous studies involving both European and Black populations (5,10,12). In our study, the A allele of rs13538 was associated with lower circulating levels of 13 associated N-acetylated amino acids, except N-δ-acetylornithine. To explain this observation, we conjecture that the A allele is associated with lower acetylation activity of the enzyme. Because N-terminal acetylation results in transfer of an acetyl group to the α-amino group of an amino acid, N-δ-acetylornithine may represent a substrate rather than a product of NAT8 activity, explaining the opposite direction of associations. In fact, although N-δ-acetylornithine was higher with the NAT8 genetic variant, N2,N5-diacetylornithine was lower, which may signal an accumulation of N-δ-acetylornithine in the setting of reduced enzymatic ability.

Because NAT8 is believed to be involved in the detoxification of products in the mercapturic pathway by acetylating cysteine conjugates to allow for kidney excretion, the hypothesized diminished activity of the A allele variant may result in the accumulation of toxic substrates (1,23,24). The A variant has also been associated with higher risk of CKD and lower GFR in several GWAS in European populations (2–4). In ex vivo experiments, exposure to nephrotoxic species resulted in upregulated expression of NAT8 (25). However, the positive association between the N-acetylated amino acids and adverse kidney outcomes suggests a more complex physiology because one might expect that only higher levels of N-δ-acetylornithine, our hypothesized substrate, would be associated with higher risk of kidney failure. The associations of amino acids acetylated at the α position (N-acetyl-1-methylhistidine, N-acetyl-3-methylhistidine, N-acetylhistidine, and N2,N5-diacetylornithine) and kidney failure might simply reflect their strong correlation with GFR and imply their role as a biomarker rather than a causal determinant of disease.

Genetic variability in protein acetylation capacity has long been recognized, with pathologic implications in many diseases (26). Various alterations to post-translational protein modification occur in the setting of chronic diseases, with acetylation being one of the most common (27,28). For example, the N- and O-acetylation of surface tumor antigens may be involved in the pathogenesis of melanoma and leukemia, and many acetoproteins have been implicated in cognitive disorders (28). Disturbances of acetylation have also been implicated in drug- and toxin-induced kidney injury (29). Different genotypes of NAT2, a related gene that encodes N-acetyltransferase 2, exhibit different acetylator phenotypes (rapid, intermediate, and slow), with different rates of metabolism for specific drugs, such as sulfamethazine and caffeine (30). The acetylation activity by NAT8 may have pathologic implications in kidney disease, either directly through the build-up of endogenous substances or indirectly by diminished therapeutics of drugs or other exogenous substances undergoing N-acetylation. That said, to our knowledge, no existing therapeutics are known to undergo acetylation by NAT8.

The kidney excretion of acetylated amino acids is incompletely understood. On the basis of molecular size, they should readily undergo glomerular filtration, which is also evidenced by the strong negative correlations between these metabolites and GFR. What little is known about their subsequent kidney handling is derived largely from analysis of patients with mutations in ACY1 (aminoacylase 1). This enzyme, expressed almost exclusively in kidney tubular epithelial cells, the liver, and the gastrointestinal tract, removes the acetyl group from acetylated amino acids, yielding “free” amino acids (31). Individuals with an inborn error in this enzyme have marked elevations of acetylated amino acids (including N-acetylasparagine, N-acetylglutamine, N-acetylleucine, and N-acetylphenylalanine) in both blood and urine (32–36). In vitro, kidney epithelial cells robustly express ACY1 and can take up and convert acetylated amino acids into free amino acids. Taken together, these findings lead to the view that circulating acetylated amino acids, derived from systemic degradation of N-acetylated proteins, are filtered and then reabsorbed by the kidney tubular epithelium, where they are then converted back to free amino acids (37). Interestingly, the urinary levels of N-acetylated amino acids were associated with the A allele of rs13538, but not associated with kidney failure in GCKD participants, which may imply that N-acetylated amino acids in urine are biomarkers of reduced kidney function, rather than directly related to kidney failure risk.

Strengths of this study include a study population consisting of multiple, well-defined cohorts, and the availability of both genetic and untargeted serum metabolomics data, enabling inferences about pathophysiology underlying genetic effects. The long follow-up periods allow assessment of associations of metabolites and kidney failure. This study also has limitations. First, because of its observational nature, we can generate only biologic hypotheses, which warrant experimental validation. We caution against overinterpretation of the associations with kidney failure because of small sample size. We acknowledge that, despite a known association of rs13538 with CKD and GFR, the causal effect may not be through the N-acetyl amino acid pathways. Second, the number of kidney failure events was relatively small in the BioMe, ARIC, and GCKD cohorts. Our use of meta-analysis to combine results allowed for greater power, but the results from AASK participants, who were relatively young and had advanced nondiabetic CKD and comorbid cardiovascular diseases, had the greatest weight in meta-analyses, thus potentially limiting generalizability. Nonetheless, the directions of associations in the other cohorts were mostly consistent with those in AASK, suggesting an absence of substantial heterogeneity across cohorts. Third, no study had concomitant urine and serum metabolites, precluding analysis of correlations between serum and urine metabolite levels and fractional excretion of metabolites.

In conclusion, combining analyses of genetic and metabolomic data from multiple cohorts, we redemonstrate significant associations between an NAT8 gene variant and 14 N-acetylated amino acids. We identified five NAT8-associated N-acetylated amino acids, namely, N-δ-acetylornithine, N-acetyl-1-methylhistidine, N-acetyl-3-methylhistidine, N-acetylhistidine, and N2,N5-diacetylornithine, with circulating levels that were associated with kidney failure. These findings suggest pathways of kidney disease pathophysiology involving the NAT8 gene, N-acetylated amino acids, and progression of CKD.

Disclosures

D.E. Arking reports serving on the Association for the Eradication of Heart Attach (AEHA) Scientific Advisory Board. E. Boerwinkle reports ownership interest in Codified Genomics. J. Coresh reports employment at Welch Center for Prevention Epidemiology & Clinical Research; consultancy agreements with Healthy.io, Kaleido, and Ultragenyx; ownership interest in Health.io; receiving research funding from National Institutes of Health and National Kidney Foundation, which receives industry support; PCT/US2015/044567 provisional patent (Coresh, Inker, and Levey), filed 8/15/2014, “Precise Estimation of Glomerular Filtration Rate from Multiple Biomarkers” (the technology is not licensed in whole or in part to any company); and serving as a scientific advisor or member of Healthy.io and National Kidney Foundation. M.E. Grams reports receiving honoraria from academic institutions for giving grand rounds and American Society of Nephrology for a Young Investigator Award; serving as a scientific advisor or member of American Journal of Kidney Diseases, CJASN, JASN Editorial Fellowship Committee, Kidney Disease Improving Global Outcomes Executive Committee, National Kidney Foundation Scientific Advisory Board, and United States Renal Data System Scientific Advisory Board; and other interests/relationships with National Kidney Foundation, which receives funding from Abbvie, Relypsa, and Thrasos; received travel support from Dialysis Clinic, Inc. to speak at the annual meeting, and Kidney Disease Improving Global Outcomes for participation in scientific meetings and the executive committee. A.M. Hung reports employment by Vanderbilt University and Veterans Affairs; receiving research funding from Department of Veterans Affairs; and serving as a scientific advisor or member of Journal of Renal Nutrition, section editor of Clinical Nephrology, standing member of Scientific Review Committee (SRC) Health Services Research and Development bioinformatics, ad hoc SRC National Heart, Lung, and Blood Institute; ad hoc Scientific Review Committee Clinical Science Research and Development, and ad hoc Scientific Review Committee The Kidney, Nutrition, Obesity and Diabetes Study. A. Kottgen reports serving as a scientific advisor or member of American Journal of Kidney Diseases, JASN, Kidney International, and Nature Reviews Nephrology. G.N. Nadkarni reports receiving grants, personal fees, and nonfinancial support and having equity in Renalytix AI, as member of Scientific Advisory Board and cofounder; receiving personal fees from AstraZeneca for serving on advisory board for SGLT2 inhibitors; receiving personal fees from Reata for serving on virtual advisory board; and receiving personal fees from BioVie and GLG Consulting, during the conduct of the study. E.P. Rhee reports research funding from Elysium Pharmaceuticals. All remaining authors have nothing to disclose.

Funding

M.E. Grams and S. Luo receive support from the National Institute of Diabetes and Digestive and Kidney Diseases (grant R01DK108803); M.E. Grams and J. Coresh also receive support from grant R01DK124399. J. Coresh and E.P. Rhee also receive support from the National Institute of Diabetes and Digestive and Kidney Diseases (grants U01DK085689 and U01DK106981). The work of A. Köttgen was funded by the Deutsche Forschungsgemeinschaft (grant KO 3598/5-1). The ARIC study has been funded, in whole or in part, with Federal funds from the National Heart, Lung, and Blood Institute (grants R01HL087641, R01HL059367, and R01HL086694), National Institutes of Health (contract HHSN268200625226C), US Department of Health and Human Services (contract numbers HHSN268201700001I, HHSN268201700002I, HHSN268201700003I, HHSN268201700004I, and HHSN268201700005I), and National Human Genome Research Institute (contract U01HG004402). The authors thank the staff and participants of the ARIC study for their important contributions. Infrastructure was partly supported by the National Institutes of Health and the National Institutes of Health Roadmap for Medical Research (grant UL1RR025005). Metabolomics measures in the ARIC study were supported by the JLH Foundation (Houston, Texas) and National Heart, Lung, and Blood Institute (grant R01HL141824). The GCKD study was/is funded by grants from the Bundesministerium für Bildung und Forschung (grant number 01ER0804), the KfH-Stiftung Präventivmedizin, and several industry partners.

Supplemental Material

This article contains the following supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.08600520/-/DCSupplemental.

Supplemental Table 1. All N-acetyl amino acids quantified via untargeted serum metabolomic profiling (N=31) in AASK.

Supplemental Table 2. Missingness for all covariables of interest.

Supplemental Table 3. rs13538 in other populations.

Supplemental Table 4. Circulating N-acetylated amino acids that are associated with rs13538 in AASK and their correlations with GFR.

Supplemental Table 5. Circulating N-acetylated amino acids that are associated with rs13538 in AASK and their associations with ESKD in AASK, ARIC, and meta-analysis, accounting for death as the competing event.

Supplemental Table 6. rs13538 and ESKD.

Supplemental Figure 1. Overview of study design.