Two Toxic Lipid Aldehydes, 4-hydroxy-2-hexenal (4-HHE) and 4-hydroxy-2-nonenal (4-HNE), Accumulate in Patients with Chronic Kidney Disease

Lipid aldehydes originating from the peroxidation of n-3 and n-6 polyunsaturated fatty acids are increased in hemodialysis (HD) patients, a process already known to promote oxidative stress. However, data are lacking for patients with chronic kidney disease (CKD) before the initiation of HD. We prospectively evaluated the changes of plasma concentrations of two major lipid aldehydes, 4-HHE and 4-HNE, according to the decrease of glomerular filtration rate (GFR) in 40 CKD and 13 non-CKD participants. GFR was measured by inulin or iohexol clearance. Thus, 4-hydroxy-2-nonenal (4-HNE) and 4-hydroxy-2-hexenal (4-HHE) were quantitated in plasma by gas chromatography coupled with mass spectrometry and their covalent adducts on proteins were quantified by immunoblotting. On the one hand, 4-HHE plasma concentration increased from CKD stage I–II to CKD stage IV–V compared to non-CKD patients (4.5-fold higher in CKD IV–V, p < 0.005). On the other hand, 4-HNE concentration only increased in CKD stage IV–V patients (6.2-fold, p < 0.005). The amount of covalent adducts of 4-HHE on plasma protein was 9.5-fold higher in CKD patients than in controls (p < 0.005), while no difference was observed for 4-HNE protein adducts. Plasma concentrations of 4-HNE and 4-HHE are increased in CKD IV–V patients before the initiation of hemodialysis.


Introduction
Patients with chronic kidney disease (CKD) exhibit a high incidence rate of cardiovascular diseases [1,2]. Oxidative stress, i.e., an imbalance between production of reactive oxygen and nitrogen species and antioxidant systems, is a hallmark of the uremic syndrome [3][4][5]. Oxidative stress further promotes the inflammatory process, accelerates renal injury, and favors cardiovascular dysfunctions [6][7][8][9]. During this process, cell and tissue damages can result from a direct attack by radical species, but also from several oxidation by-products resulting from the oxidative breakdown of biomolecules. In contrast to the radical species, these secondary oxidation by-products can diffuse

Patient Data
Forty patients with CKD stage 1 to end-stage 5 (aged 47 ± 14 years) and 13 non-CKD participants (aged 44 ± 13 years, p = 0.235) were included in the present study. The etiology of renal diseases was nine glomerulonephritis (but none from diabetes), six renovascular diseases (including three nephroangiosclerosis), six solitary kidney, four tubulointerstitial diseases, five cystic kidney disease, and 10 unknown etiologies. Out of the 40 CKD patients, 14 were on antihypertensive drugs (35%, p = 0.045) while eight were on lipid lowering drugs (20%, p = 0.322). The main characteristics of the CKD patients and controls are summarized in Table 1. As expected, mGFR (p < 0.001), plasma creatinine (p < 0.001) and urea (p < 0.001) were the main differences between groups.

Advanced Lipoperoxidation By-Products
It is known that 4-HNE and 4-HHE are prone to react with proteins to form covalent adducts [10]. We therefore used specific antibodies directed against Michael adducts of 4-HNE and 4-HHE to detect covalent adducts on plasma proteins using immunoblotting (i.e., dot blot). Regardless of CKD stage, CKD patients exhibited an increased level of 4-HHE adducts on plasma proteins ( Figure 3A,B). The level of 4-HHE protein adducts was 115 (92-147) arbitrary unit (AU) in non-CKD subjects and 1093 (587-3984) AU in patients with CKD (Mann-Whitney U test, p < 0.005), i.e., a 9.5-fold increase. Surprisingly, no difference was noticed in 4-HNE protein adducts (1617 (1129-2501) and 1863 (1450-2524) AU for non-CKD and CKD patients, respectively ( Figure 4).  On the boxplots, the median is indicated by a horizontal bar, the interquartile range as a box and the 5th to 95th percentile as the "whiskers". GFR was measured using inulin or iohexol clearance. Data are presented as median (interquartile range). * p < 0.05, ** p < 0.01, ns non-significant.

Advanced Lipoperoxidation By-Products
It is known that 4-HNE and 4-HHE are prone to react with proteins to form covalent adducts [10]. We therefore used specific antibodies directed against Michael adducts of 4-HNE and 4-HHE to detect covalent adducts on plasma proteins using immunoblotting (i.e., dot blot). Regardless of CKD stage, CKD patients exhibited an increased level of 4-HHE adducts on plasma proteins ( Figure 3A  On the boxplots, the median is indicated by a horizontal bar, the interquartile range as a box and the 5th to 95th percentile as the "whiskers". GFR was measured using inulin or iohexol clearance. Data are presented as median (interquartile range). * p < 0.05, ** p < 0.01, ns non-significant. To obtain further insight, ALEs were studied according to the actual stage of CKD which was evaluated with measured GFR. 4-HHE protein adducts gradually increased with the decline of GFR ( Figure 3B) while 4-HNE protein adducts were not correlated with the CKD stage ( Figure 4A,B).

Discussion
Few studies previously reported that plasma lipid aldehydes (among which 4-HNE and 4-HHE) concentrations were increased in chronic hemodialysis patients [29]. However, no data is available in the literature regarding the actual concentration of 4-HNE and 4-HHE in CKD patients before initiation of renal replacement therapy, a process that exacerbates oxidative stress [31,34]. We observed that plasma concentrations of both 4-HNE and 4-HHE dramatically increase in CKD 4-5 patients. Plasma levels of 4-HHE (4.5-fold) and 4-HNE (six-fold) were higher in CKD 4-5 patients than in controls unambiguously evidencing the presence of oxidative stress before the initiation of dialysis. These results are in line with the report of Alhamdani et al. [29] who observed an increase in both 4-HHE (2.5-fold) and 4-HNE (seven-fold) in hemodialysis patients. These data suggest that oxidative stress could rather result from the uremic/inflammatory environment than from the renal replacement therapy (such as HD). Several works suggest that the retention of uremic toxins could contribute to increased oxidative stress in uremia [35][36][37][38][39][40][41][42].
In addition, 4-HHE and 4-HNE exhibited different patterns of accumulation. Indeed, while 4-HHE level gradually increases with the decline of GFR (reaching statistical significance at stage 3), 4-HNE only increased in CKD stage 4-5. Free 4-HHE concentration was positively correlated with the amount of 4-HHE adduct on plasma proteins while free 4-HNE concentration was not correlated with the 4-HNE adducts on plasma proteins (see Table 2). The relative proportion of 4-HHE and 4-HNE accumulated in end-stage renal disease (ESRD) deserves some comments. The proportion of 4-HHE to 4-HNE depends of two main factors: First, the availability of omega 3 PUFAs and omega-6 PUFAs, which largely depends from the dietary habits (e.g., intake of low/large amount of omega-3 PUFAs). Secondly, the relative activity of the various detoxifying enzymes and their respective affinity for 4-HNE and 4-HHE. While plasma level of omega-3 polyunsaturated fatty acids (PUFAs) is much lower than omega-6 PUFAs [43] omega-3 are however more prone to oxidation than omega-6 fatty acids. We recently measured plasma lipid aldehyde concentration in patients with type 2 diabetes [33] and performed simultaneously the assay of omega-3, omega-6 PUFAs (by GC coupled with flame ionization detector -FID) and their aldehyde by-products 4-HHE and 4-HNE (by GC-MS). Our data showed that, in T2D patients, one molecule of omega-3 PUFA out of 12,000 was oxidized into the form of 4-HHE while one molecule of omega 6 PUFA out 78,000 was oxidized into the form of 4-HNE. These data suggest a greater oxidizability of omega-3 PUFAs than omega-6 PUFAs. Detoxification of lipid aldehydes involves several enzymes, such as for instance glutathione S-transferase (GST), fatty aldehyde dehydrogenase (FALD) and/or aldehyde dehydrogenase (ALDH). Hubatsch et al. [44] reported that 4-HHE is a poor substrate for GSTA4-4 compared to HNE. Long et al. [45] demonstrated that HHE is a worse substrate than HNE for the mitochondrial ALDH5A while both of them are detoxified by aldose reductase [46]. Some differences in detoxification and disposition could therefore account for the different levels of 4-HNE and 4-HHE accumulated in ESRD patients. Some data suggest an overexpression of GST in hemodialysis patients [47] but there is no further information about the other detoxifying enzymes.
In an experimental model, increased lipid peroxidation was shown to play a pathophysiological role for glomerulonephritis [48]. A deleterious role of lipid peroxidation was also suggested in CKD [49]. Bae et al. (2011) [50] demonstrated in vitro that 4-HHE induces HK-2 tubular cells apoptosis, suggesting that it could contribute to pathogenesis of kidney injury. Moreover, 4-HHE treatment resulted in a decline of tubular cells viability associated with an over-expression of nuclear factor kappa-B (NF-κB) and a decrease of inhibitor of nuclear factor kappa-B kinase beta (IκBKβ). Nevertheless, the concentrations of 4-HHE used in this study (5-100 µM) were unrealistic, largely exceeding the concentrations found in human plasma from CKD patients [29] (i.e., <1µM, from the present study). However, the lipid peroxidation by-products play an important role in inflammation. 4-HNE and many other aldehydes displayed a chemotactic activity for leukocytes from a concentration of 0.1µM [51,52], suggesting that these compounds could contribute to micro-inflammation commonly observed in CKD.
Furthermore, 4-HHE and 4-HNE are reactive aldehydes that are prone to covalently bind to plasma proteins. Lipid aldehydes can react with proteins through the production of Schiff bases (involving the amino group of lysine residues) and/or Michael adducts (involving thiol or amino groups of cystein, lysine or histidine residues) [53]. This effect is therefore thought to be pivotal in the toxic effect of hydroxy-alkenals. Lipid aldehydes are able to covalently bind to serum albumin [18] and many other plasma proteins, and through change in their tridimensional structures, alter their binding capacity for other ligands. The lipid aldehydes exhibit therefore the double potency to behave as small molecules but also as protein bound uremic toxins. This type of interactions (i.e., covalent) with proteins is very different from more classical protein-bound retention solutes, such as indoxyl sulfate or p-cresyl sulfate. This covalent binding of lipid aldehydes can be regarded as irreversible in contrast to the other ligands, which display a competitive binding with two specific binding sites on albumin. It is worth noting that adducted proteins can actually exert deleterious biological activities [20,54] therefore behaving themselves as secondary toxins. In our study, in contrast to 4-HHE, the increase of plasma 4-HNE was not accompanied with a parallel increase of its protein adducts. This discrepancy may be related to the difference in the biological properties of the two aldehydes that is partially related to their carbon chain length (six and nine carbons for 4-HHE and 4-HNE, respectively). This structural difference may result in large difference of solubility and reactivity with amino acids. For instance, in vitro studies have demonstrated the formation of adducts incorporating multiple 4-HNE molecules [19]. Furthermore, as we only measured plasma concentrations, we cannot exclude a difference in tissue accumulation of 4-HNE adducts conferring intracellular toxicity. Indeed, 1%-8% of 4-HNE that is generated in cells would react with proteins and the majority of the target enzymes for 4-HNE are mitochondrial proteins [55]. Further studies are however needed to investigate the difference in metabolism of these two aldehydes in the context of CKD.
According to the European Uremic Toxin (EuTox) workgroup [30], a uremic toxin can be defined as a compound that (i) is excreted in urine and thus accumulates in patients with CKD as a result of the decline in renal clearance and (ii) that exerts deleterious effects on biological systems. 4-HHE and 4-HNE are partly excreted by the kidney in the form of 4-hydroxy-2-hexenoic acid (4-HHA), 4-hydroxy-2-nonenoic acid (4-HNA), or as mercapturic acid [56], but the kidney is not the major site of their metabolization. Indeed, the main site of hydroxyalkenals biotransformation is the liver [56] and the subsequent metabolites are excreted by the kidneys. Further, 4-HHE and 4-HNE do not accumulate in CKD as the sole result of a decreased renal clearance, and thus according to the EuTox definition, are not strictly speaking uremic toxins.

Conclusions
We reported in the present study that plasma levels of two lipids aldehydes, namely 4-HNE and 4-HHE, displayed increased level in CKD 4-5 patients. These data confirmed that the oxidative stress preexists in CKD prior to the initiation of renal replacement therapy (e.g., dialysis). Lipid aldehydes accumulated in ESRD as well as advanced lipoperoxidation end products (e.g., protein adducts) could, as has already been documented in many other diseases (e.g., diabetes), exerting a deleterious effect, and can therefore be regarded as a "uremic toxin like" compounds.

Chemicals
Unless otherwise specified, all solvents were from Carlo Erba (Peypin, France) and all chemicals from Sigma Aldrich (Saint Quentin Fallavier, France).

Subjects
A total of 53 subjects (32 men), including 13 non-CKD subjects, were recruited from department of Nephrology of Edouard Herriot University Hospital (Lyon, France). This study was proposed to CKD patients referred for GFR measurement excluding renal transplant recipients. Exclusion criteria were age >65 years and the main pathologies other than CKD, known to be associated with oxidative stress (diabetes, obesity defined as a body mass index >30 kg/m 2 , systemic inflammatory, autoimmune diseases or known active cancer). Non-CKD patients were candidates for kidney donation.

Blood Sampling
After an overnight fast, blood samples were obtained by venipuncture and centrifuged at 3500× g for 10 min. Plasma was isolated, snap frozen in liquid nitrogen and stored at −80 • C until use.

Glomerular Filtration Rate Measurements (mGFR)
GFR was measured by inulin or iohexol clearances (expressed as mL/min per 1.73 m 2 ). Urinary inulin clearance was performed for 32 subjects. After a priming dose, inulin (Polyfructosan [inutest], Laevosan, Linz, Austria) was perfused at constant flow for 3 h and urines were collected every 30 min by spontaneous voiding. Blood samples were obtained in the middle of each period of urine collection (3 to 4 collection periods of 30 min each). Inulin concentration was assayed by an enzymatic method. Inulin clearance was calculated using the following equation: where U is urine concentration of inulin, V is urine volume and P is the plasma concentration of inulin. Plasma iohexol clearance was performed for 21 subjects. Iohexol (300 mg; Omnipaque; GE Healthcare SAS, Vélizy-Villacoublay, France) was injected intravenously and blood was drawn at 120, 180, and 240 min. The plasma concentration of iohexol was measured by HPLC. The GFR was calculated using the following equation: where Slope refers to the slope of the plasma decay of iohexol concentration and C 0 the plasma iohexol concentration at time 0 (corrected with the Bröchner-Mortensen equation). CKD stages were determined according to the K/DOQI (Kidney Disease Outcomes Quality Initiative) guidelines.

Other Biochemical Measurements
Plasma urea, bicarbonate, hemoglobin, plasma proteins and serum albumin were measured by standard laboratory.

Statistical Analysis
Data were analysed using Graphpad Prism (GraphPad softwares, La Jolla, CA, USA) and R (https://www.r-project.org/) softwares. The data are expressed as mean ± 1 standard deviation (SD) or as median (interquartile range-IQR) when variables were not normally distributed. Distributions were tested for normality using Shappiro-Wilk test. Differences between groups were assessed by Kruskall-Wallis tests followed when appropriate by Dunn tests. Sex ratio and medications between groups were compared using Fisher exact test. Univariate analysis was performed using the Spearman rank correlation method. A p < 0.05 was considered as statistically significant in all analysis.