Renal kallikrein excretion and epigenetics in human acute kidney injury: Expression, mechanisms and consequences

Background Renal kallikrein (KLK1) synthesis and urinary excretion are reportedly diminished during AKI (acute kidney injury) in animal models, and provision of kallikrein abrogates renal injury in this setting, but data in human AKI is limited. Therefore we first examined KLK1 renal excretion in human AKI, and then probed potential endocrine and epigenetic mechanisms for its alterations. Methods KLK1 enzymatic activity excretion was evaluated in urine from patients with established or incipient AKI, versus healthy/non-hospital as well as ICU controls. Endocrine control of KLK1 excretion was then probed by catecholamine and aldosterone measurements in established AKI versus healthy controls. To examine epigenetic control of KLK1 synthesis, we tested blood and urine DNA for changes in promoter CpG methylation of the KLK1 gene, as well as LINE-1 elements, by bisulfite sequencing. Results Patients with early/incipient AKI displayed a modest reduction of KLK1 excretion, but unexpectedly, established AKI displayed substantially elevated urine KLK1 excretion, ~11-fold higher than healthy controls, and ~3-fold greater than ICU controls. We then probed potential mechanisms of the change. Established AKI patients had lower SBP, higher heart rate, and higher epinephrine excretion than healthy controls, though aldosterone excretion was not different. Promoter KLK1 CpG methylation was higher in blood than urine DNA, while KLK1 methylation in blood DNA was significantly higher in established AKI than healthy controls, though KLK1 methylation in urine tended to be higher in AKI, directionally consistent with earlier/incipient but not later/established changes in KLK1 excretion in AKI. On multivariate ANOVA, AKI displayed coordinate changes in KLK1 excretion and promoter methylation, though directionally opposite to expectation. Control (LINE-1 repetitive element) methylation in blood and urine DNA was similar between AKI and controls. Conclusions Unexpectedly, increased KLK1 excretion in AKI patients was found; this increase is likely to be due in part to increments in adrenergic tone during BP depression. Epigenetic changes at KLK1 may also play a role in early changes of KLK1 expression and thus AKI susceptibility or recovery.


Background
The incidence of acute kidney injury (AKI) in hospitalized patients is estimated to be 5-10%, and is much higher in the critically ill [1,2]. Despite the potential for recovery of kidney function, acute kidney injury is associated with substantial morbidity and even mortality. AKI, due to ischemia or nephrotoxic agent exposure, may lead to death or sublethal injury of proximal tubular cells, after which surviving cells may repolarize and/or de-differentiate, proliferate, migrate to denuded areas, re-differentiate, and restore nephron structure (including the tubular epithelium) and function [3,4].
The serine protease kallikrein (KLK1; E.C.-3.4.21.35; OMIM 147910), excreted from kidney into urine, catalyzes the cleavage of low molecular weight kininogen to lysylbradykinin (kallidin), which exhibits both vasodilator and natriuretic pharmacological properties in the kidney; if these properties occur in vivo, the potential of the system for regulating blood pressure is clear [5].
Renal kallikrein levels were markedly reduced in an aminoglycoside-induced AKI animal model [6], and KLK1 gene transfer protected against aminoglycoside-induced nephropathy by diminishing apoptosis and inflammation [7]. In addition, KLK1 infusion during aminoglycoside treatment attenuated drug-induced renal dysfunction, cortical damage, and apoptosis in the rat [8]. Previously, we have reported that urinary KLK1 excretion was diminished in renal allograft recipients with a clinical diagnosis of acute tubular necrosis (ATN) [9]; since urinary KLK1 originates in the kidney, reduced urinary kallikrein levels may reflect impaired renal function. However, this finding has not been pursued in humans.
In mammals, cytosine methylation occurs almost exclusively at CpG dinucleotides, which are enriched at CpG islands, often are located at 5'-/promoter regions of functional genes [10]. Cytosine methylation may result in transcriptional repression either by interfering with transcription factor binding or by inducing a repressive chromatin structure [11]. Apoptotic pathways are targets for such "epigenetic" silencing, and several apoptosis-linked genes that are regulated directly or indirectly by methylation have been described [12].
In this study, we first probed whether KLK1 excretion is altered in human AKI, and if so what mechanisms (endocrine or epigenetic) might be driving the change. Since kidney repair after injury may recapitulate normal morphogenesis, we hypothesized that urinary kallikrein levels would be associated with severity of AKI and with epigenetic changes in the renal kallikrein-1 (KLK1) promoter. We considered that changes in kallikrein excretion or the KLK1 promoter might predict renal functional recovery and thus serve as biomarkers of recovery from AKI, thereby facilitating timely diagnosis and treatment. We therefore examined patients with AKI for urinary expression of KLK1 enzymatic activity, as well as genomic DNA from blood and urine for CpG methylation pattern at the KLK1 gene promoter.

Established AKI cases
Cases were ascertained from a single center prospective non-concurrent observational cohort of inpatients who were identified as having suffered acute kidney injury (AKI) during a hospital admission. The institutional review board of the University of California, San Diego (UCSD) reviewed and approved the study as well as the consent document. Primary providers referred potentially eligible patients to investigators and study coordinators. Patients were eligible for enrollment if they were greater than 18 years of age and met the serum creatinine criteria for acute kidney injury as set out by the Acute Kidney Injury Network (AKIN) [13]: an abrupt rise by ≥0.3 mg/dl within a 48-hour period. Patients were excluded if they received chronic renal replacement therapy (hemodialysis or peritoneal dialysis) within the 6 months prior to admission; had ever been in receipt of a kidney transplant; pregnant or breast feeding; currently incarcerated or otherwise institutionalized (nursing home, rehabilitation); were placed under hospice/comfort care; or did not have reasonable expectation of survival past the present hospitalization. If eligible, the patient or authorized representative was presented the study and consent for participation was obtained. All relevant patient data were derived from the electronic and paper medical records, as well as direct interview of the patient or surrogate. Basic demographic information and co-morbid conditions were recorded on enrollment. Etiology of AKI was determined by chart review of provider diagnosis, and verification by study personnel; diagnosis followed the categories set by the Project to Improve Care of Acute Renal Disease (PICARD) study investigators [1]. Daily assessment included medication review, physical exam (as recorded in medical record or assessed by study personnel when missing), vital signs, intake and output, and labs. Clinical data elements were collected daily and the need for and utilization pattern of renal replacement therapy was also monitored and recorded. Blood and urine samples were collected at time of entry, daily for 7 days maximum, and at hospital discharge. Twenty-four hour urine collections were performed at study entry and hospital discharge for creatinine/urea clearance (in approximation of glomerular filtration rate -GFR) as well as urinary kallikrein activity. One sample of whole blood was collected for genomic DNA preparation (and genetic analysis of the KLK1 locus) at the time of study entry. At the time of discharge, follow-up appointments with either the primary care physician or a nephrologist (not all patients were seen by a nephrologist during the hospitalization), extent of renal recovery, and dialysis dependence (if needed during hospitalization) were ascertained. Recovery: pre-defined study endpoints were 12 months of follow-up, start of maintenance hemodialysis or peritoneal dialysis, receipt of a kidney transplant, or death; recovery of renal function was the primary outcome, defined as a return to within 10% of baseline eGFR or lowest eGFR prior to AKI event. Recovery was assessed at 6 months of follow-up.

Healthy (non-hospital) controls
In addition, we obtained data for a control group of 38 healthy adult subjects. Each healthy control was selected from only one member from each of 38 twin pair sets. Twin pairs were recruited by a population-based twin registry in southern California [14], and by newspaper advertisement. These twins are of European, African-American, and Asian biogeographic ancestry. Ethnicity was established by self-identification. Self-reported zygosity was confirmed by extensive SNP genotyping. There was no clinical evidence for kidney disease or any other cardiovascular disorder in any of the controls. Untimed (spot) urine collections were performed, and one sample of EDTAanticoagulated whole blood was collected for genetic analysis of the KLK1 locus at the time of study entry.
Replication: Incipient (early) AKI cases in an intensive care unit (ICU), with ICU control subjects An additional sample of controls was ascertained for replication of findings, in an ICU setting. In brief, the replication sample consisted of n = 44 subjects ("ICU controls") who did not develop AKI during a 7-day observation period after hospital ICU admission, as well as n = 11 subjects (ICU cases) who did develop AKI, as defined above. Patients were screened at ICU admission for potential study participation at three academic medical centers (University of California San Diego, University of Alabama, and Université de Montréal) between July 2006 and December 2008. Patients were eligible for enrollment if they were age 18 or older and had a life expectancy of at least 48 hours. Controls were excluded if they had AKI according to the AKIN criteria [13], were admitted to the ICU > 48 hours prior to screening, transferred from another ICU, had a serum creatinine > 2 mg/dl ≤ 3 days before ICU admission, were prisoners, received dialysis within the 12 months prior to admission, had a functioning kidney transplant, were on anticoagulants or warfarin within the last 7 days, suffered from decompensated cirrhosis, had CKD stage 5, were anemic (hemoglobin < 9.0 mg/dl or hematocrit < 27%), or were already enrolled in another research project.
The study was approved by the Institutional Review Boards at each institution, and written informed consent was obtained from all participants or their health care surrogates. Following informed consent, data on past medical history were collected once, and clinical, laboratory and process-of-care elements were collected daily. Each institution's local laboratory measured serum creatinine values. AKI was defined as an increase in serum creatinine level of more than 0.3 mg/dL or more than 50% from a reference creatinine within 48 hours (AKIN criteria) [13]. Patients without AKI within the first 4 days had continued blood and urine samples twice daily, for a total study observation period of 7 days.

Biochemical assays
Urine was assayed for kallikrein by an alkaline amidolytic activity assay as previously described (4), using the chromogenic substrate S-2266: [D] Val-Leu-Arg-paranitroanilide (Kabi Pharmacia; Franklin, OH, USA) [5,15]. The activity of kallikrein per liter of urine (units per liter, U/L) is calculated from the formula: U/L = (9.55·A), where A = absorbance, after a 30-min assay incubation of the paranitroanilide product in a spectrophotometer at 405 nm [5,15]. Inclusion or exclusion of the kallikrein inhibitor aprotinin (Trasylol, Miles Inc., West Haven, CT, USA) indicated that a fraction of human urinary S-2266 amidolytic activity in the absence of aprotinin was non-kallikrein (likely urokinase) [15]; thus, aprotinin (20 kallikrein inhibitory U/mL) was systematically included in the assay blank, to assure specificity for kallikrein measurement. The inter-assay coefficient of variation was 18.1%, and activities from n = 20 samples measured on two separate occasions correlated highly (Spearman R = 0.92, P< 0.01). In n = 87 subjects, activity correlated (Spearman R = 0.82, P< 0.0001) when results were compared for kallikrein excretion normalized to time versus creatinine excretion. Specificity of the S-2266 amidolytic assay for glandular (KLK1, including renal, pancreas, and salivary) kallikrein activity in urine arises from two features: first, the substrate S-2266 is cleaved only by a particular subset of serine proteases, including KLK1; and second, the inclusion of aprotinin in the assay blank, which specifically inhibits serine proteases including KLK1; nonetheless, only an immunoassay specific for lysyl-bradykinin (the kinin product of KLK1) generation could provide absolute assurance of specificity for KLK1.
Clinical chemistries (serum or urine, electrolytes or creatinine) were measured by spectrophotometric autoanalyzer. Urine aldosterone concentration was determined by enzyme immunoassay (Alpco Diagnostics, Salem, NH, USA). Urine catecholamine concentration was determined by commercial ELISA kit (Labor Diagnostika Nord GmBH & Co. KG, Nordhorn, Germany). Urine values were normalized to creatinine concentration in the same sample.

DNA extraction and bisulfite treatment of CpG sites
A sample of EDTA-anticoagulated whole blood was obtained from participants and stored at -70°C prior to DNA extraction. Timed and spot urine samples were obtained and frozen at -70°C before assay. Blood DNA was prepared from blood leukocytes with a commercial kit (QIAamp ® DNA Mini Kit; Qiagen, USA). Urine DNA was prepared from the urine pellet with spun columns (urine DNA isolation kit; Norgen, Canada). Both DNAs were subjected to sodium metabisulfite (Na 2 S 2 O 5 ) treatment (Imprint™ DNA Modification Kit; Sigma, USA), and then eluted in 20 μL elution buffer. Bisulfite converts cytosine residues to uracil, but leaves 5-methylcytosine residues unaffected.
As a control, global CpG methylation analysis was completed using PyroMark LINE-1 reagents (Pyrosequencing; Qiagen, USA). We thus determined the methylation status of three CpG sites in LINE-1 repetitive (LTR-like) elements, wherein methylation levels of CpG sites represent global methylation status across the genome, because of the repetitive nature of LINE-1 elements [17,18].
PCR amplification was performed using 10X PCR buffer, 3.0 mM MgCl 2 , 200 μM of each dNTP, 0.2 μM each of forward and reverse primers, HotStar DNA polymerase (Qiagen, USA) 1.25 U, and~10 ng of bisulfiteconverted DNA per 50 μl reaction. PCR cycling conditions were: 94°C 15 min; 45 cycles of 94°C 30 s, 56°C 30 s, 72°C 30 s; 72°C 5 min; and then products were held at 4°C. The PCR was performed with one of the PCR primers biotinylated to convert the PCR product to single-stranded DNA template. PCR products (each 10 μl) were sequenced by Pyrosequencing PSQ96 HS System (Pyrosequencing, Qiagen, USA). The methylation status of each locus was analyzed individually as a T/C SNP using Pyro-QCpG™ software (Pyrosequencing).
AKI case samples were evaluated at study entry (baseline, time of diagnosis). After bisulfite modification and PCR amplification, KLK1 blood DNA promoter methylation data from 13 AKI patients and 30 controls were obtained. KLK1 urine DNA promoter methylation data from 9 AKI and 22 controls were available. LINE-1 blood DNA methylation data from 14 AKI patients and 32 controls were available for evaluation. LINE-1 urine DNA methylation data from 15 AKI patients and 32 controls were obtained.

Statistical analyses
Results are expressed as the mean value ± one standard error of the mean (SEM) for continuous variables. For comparisons of two groups, unpaired two-sided t-tests or one-way ANOVA (enabling adjustment for covariates of age, sex, and ethnicity) were performed. Non-parametric Wilcoxon Rank Sum test was used to confirm parametric tests in the face of relatively small sample sizes. Proportions were evaluated by Fisher's exact test (2 × 2 tables) or chi-square test (3 × 2 tables). Statistical analyses were performed in R2.10.1 <http://www.r-project.org/ > or SPSS-17 (Statistical Package for the Social Sciences; Chicago, IL, USA). A P value of < 0.05 was considered significant. Multiple linear regression was performed with default criteria of entry (p < 0.05) and exit (p > 0.10) from the multivariate regression model, using stepwise or forward options. Recovery from AKI was pre-defined as return (within 6 months follow-up) to within 10% of baseline eGFR, or lowest eGFR prior to AKI event.

Results
Renal KLK1 excretion and eGFR in the 4 subject groups: AKI cases and controls Demographic and anthropometric description of the study samples is presented in Table 1. Baseline demographic characteristics (age, sex) were similar across groups.

Established AKI: Kallikrein, catecholamines, and aldosterone
Here we probed potential hormonal mechanisms whereby KLK1 excretion was elevated in established AKI, focusing on such known KLK1 stimulators as catecholamines [19] and aldosterone.
Since black and white subjects differ in reported KLK1 excretion [5,15], we evaluated the role of ethnicity in our samples (Table 1). Although cases and controls each included several biogeographic ancestries, KLK1 excretion did not differ significantly in black versus white cases (p = 0.26) or black versus white controls (p = 0.69), perhaps reflecting the relatively small sample sizes. Disease analyses were adjusted for biogeographic ancestry as a covariate.

Urinary aldosterone and catecholamine excretions
We measured urinary aldosterone, epinephrine and norepinephrine excretions (Table 3), since these hormones are known to increase urinary kallikrein. Established AKI subjects exhibited substantially elevated epinephrine excretion,~2.7 times higher than that of healthy controls (Table 3, Figure 3; 20.1 ± 2.4 vs. 7.48 ± 1.07 ng/mg creatinine; ANOVA p = 0.00016). Urinary kallikrein/epinephrine ratio did not differ between groups (Table 3); parallel elevations of KLK1 and epinephrine suggest that the epinephrine excess in AKI may be sufficient to account for the increased KLK1. To exclude the possibility that  Figure 3 illustrates parallel elevations of renal kallikrein and epinephrine excretion in the AKI group. Urinary aldosterone excretion did not differ between established AKI and healthy controls (Table 3) (9269.9 ± 2652.31 vs. 12982.4 ± 2619.78 pg/mg creatinine; ANOVA p = 0.65). Since urinary kallikrein/aldosterone ratio was also elevated in AKI (Table 3), increments in aldosterone could not explain the rise in KLK1 excretion in AKI. Urinary norepinephrine excretion did not differ in AKI (Table 3) (37.15 ± 8.14 vs. 26.85 ± 3.20 ng/mg creatinine; ANOVA p = 0.16); nor did urine kallikrein/ norepinephrine excretion ratios.

Urine albumin excretion
Quantitative urine albumin excretion was evaluated in established AKI cases and healthy controls. Albumin values in cases ranged from 2.0-4490 mg/gm creatinine  (mean, 1090 mg/gm), but kallikrein and albumin excretions did not correlate (Pearson r = 0.006, p = 0.54), rendering it unlikely that elevated kallikrein activity in AKI arose simply from pathological excretion of plasma proteases. In healthy controls, albumin excretion was 6.27 ± 0.39 mg/gm creatinine.

ICU controls: Urinary kallikrein activity and clinical findings
44 "ICU controls" were available to evaluate the specificity of urinary kallikrein elevation in AKI. Table 1 shows demographic, laboratory and clinical findings of these ICU controls. The kallikrein increment in AKI persisted when studied with ICU controls (6.74 ± 1.92 vs. 2.04 ± 0.47 U/gm creatinine; p = 0.028), though kallikrein excretion was modestly elevated in ICU-compared to healthy controls (2.04 ± 0.47 vs. 0.63 ± 0.08 U/ gm creatinine; p = 0.005).
Urine KLK1 promoter CpG methylation did not predict KLK1 activity excretion. However, in a multivariate analysis, disease status jointly predicted both KLK1 promoter methylation and enzyme activity excretion, with higher values for each in cases (multivariate p = 0.004; Figure 5. Since increased KLK1 promoter methylation would be predicted to decrease gene expression, this joint effect cannot explain (and indeed runs counter to) the elevated KLK1 excretion observed in our AKI cases ( Figure 5).

Discussion
Overview Several previous lines of investigation link alterations in KLK1 expression to AKI, in both experimental animals and humans. A decrease in urinary kallikrein excretion occurs in rodents with AKI after methemoglobin [21]; rats treated with aminoglycosides have dramatically reduced levels of urinary kallikrein [6], and a transient decrease in urinary kallikrein excretion occurs during Urine DNA KLK1 versus urine LINE-1 methylation: AKI p=0.003, and controls p<0.0001 Figure 4 CpG methylation analyzed by bisulfite sequencing: Results at the KLK1 promoter, as well as a global control (LINE-1 repetitive elements), in genomic DNA from urine or blood (mononuclear cells). Results (shown as mean ± one SEM), from established AKI cases or healthy controls, were analyzed by ANOVA, factoring for age, sex and race. LINE-1 reagents (Pyromark, Biotage) were used to analyze the 3 CpG sites in LINE-1 repetitive elements, while the KLK1 gene was analyzed by a separate PCR covering 4 promoter CpGs. Promoter KLK1 specific methylation was substantially higher in blood than urine DNA (blood 66.38 ± 1.00 vs. urine 33.43 ± 4.67%; ANOVA p < 0.0001). Promoter KLK1 methylation in blood DNA was higher in AKI than controls (AKI, 70.32 ± 2.27 vs. controls, 65.36 ± 1.05%; ANOVA p = 0.011). Promoter KLK1 methylation in urine DNA did not differ in AKI versus controls (AKI, 40.95 ± 7.06 vs. controls, 30.35 ± 5.88%; ANOVA p = 0.22). Global LINE-1 methylation was greater in both blood than urine DNA (blood 73.11 ± 0.38 vs. urine 69.37 ± 0.86%, p = 0.0002). LINE-1 methylation in blood DNA did not differ in AKI and controls (AKI, 71.71 ± 0.44 vs. controls, 73.67 ± 0.41%; ANOVA p = 0.08). LINE-1 methylation in urine DNA did not differ in cases/controls (AKI, 69.53 ± 1.54 vs. controls, 69.29 ± 1.05%; ANOVA p = 0.79).
chromate-induced AKI in the rat [22]. Renal kallikrein mRNA expression was specifically reduced in the postischemic rodent kidney, with persistently altered expression even after functional recovery from ischemic acute renal failure [23]. In humans, we reported previously that urinary kallikrein excretion was diminished in acute tubular necrosis (ATN) after renal transplantation [9].
Studies in rodents suggest beneficial effects of exogenous KLK1 replacement in the setting of experimental AKI. KLK1 gene transfer protected against aminoglycoside-induced nephropathy, with inhibition of apoptosis and inflammation [7]. KLK1 replacement after gentamicin attenuated drug-induced renal dysfunction, cortical damage, and apoptosis in the rat [8]. Furthermore, KLK1 reduced gentamicin-induced renal dysfunction and fibrosis, with decreased myofibroblast and collagen accumulation [8]. These findings indicate that the renal kallikrein/ kinin system prevents and promotes recovery of aminoglycoside-induced renal injury by inhibiting apoptosis, inflammatory cell recruitment, and fibrotic lesions.
Thus, we expected that AKI patients would have diminished urinary kallikrein excretion, since urinary kallikrein originates in the kidney; further, we anticipated that kallikrein increments might be associated with superior outcomes in AKI. Unexpectedly our established AKI subjects displayed substantially elevated (by as much as~11-fold; ANOVA p = 0.00029; Figure 2; Tables 1,3) kallikrein excretion.

Origin of increased KLK1 excretion in established AKI
Although previous reports of KLK1 excretion in AKI indicated diminished excretion in both rodent models [6,21,22] and AKI/ATN in the setting of human renal transplantation [9], we noted a~11-fold elevation of KLK1 excretion in our established AKI subjects (Tables 1-3; Figure 2). Why might KLK1 excretion be elevated in human AKI? Here we examined hormonal factors known to increase KLK1 excretion: catecholamines and aldosterone [5,19,20,[24][25][26]. We found that an elevation in epinephrine excretion paralleled that for kallikrein ( Figure 3), while norepinephrine and aldosterone excretions were unchanged (Table 3). In experimental animals, kallikrein excretion is regulated by adrenergic receptors, with stimulation by β-receptors and inhibition by α-receptors [19]. While diuretics can also elevate kallikrein excretion [20,27], the KLK1 excretion increment in AKI persisted after exclusion of the 3 subjects on diuretics.
Why was epinephrine elevated in our AKI subjects? AKI patients had lower systolic BP than controls (119.8 ± 4.4 vs. 131.4 ± 1.7 mmHg; p = 0.02) and higher heart rate (89.3 ± 3.6 vs. 68.0 ± 1.6 bpm; p = 1.73E-05) ( Table 2). While the mechanism cannot be readily probed in the setting of acute human illness, we suspect that that lower BP in AKI may stimulate baroreceptors, with resulting increase in endogenous production of epinephrine, and consequently increased heart rate and kallikrein excretion ( Figure 6).

KLK1 epigenetics
Promoter KLK1 specific CpG methylation was higher in blood than urine DNA (blood 66.38 ± 1.00 vs. urine 33.43 ± 4.67%; ANOVA p < 0.0001; Figure 4), consistent with kidney-specific expression of the KLK1 gene, in that renal kallikrein is synthesized in the distal tubule and released into urine and peritubular interstitium [20], and cytosine methylation results in transcriptional repression either by interfering with transcription factor binding or by inducing a repressive chromatin structure [11].  Figure 6 Hypothetical schema integrating experimental findings in this study of KLK1 in AKI. This diagram is presented not as established fact, but rather to generate hypotheses for further investigation. Endogenous factors may influence KLK1 synthesis and renal excretion: lower BP in AKI may activate baroreceptors, thus increasing endogenous secretion of epinephrine, thereby increasing both heart rate and urinary kallikrein excretion. KLK1 promoter genetic variants or CpG methylation can influence renal kallikrein production. Finally, exogenous factors such as adrenergic pressor infusions or diuretic treatment can also increase renal kallikrein production; indeed, since a subset of our AKI cases received such treatments (Table 1), we cannot exclude this possibility.

Hypotension
KLK1 promoter methylation in blood DNA was higher in AKI than controls (70.32 ± 2.27 vs. 65.36 ± 1.05%; p = 0.011), and there was also a trend towards higher urine KLK1 methylation in AKI than controls, but the difference was not significant (40.95 ± 7.06 vs. 30.35 ± 5.88%; p = 0.22). While a multivariate analysis indicated a joint effect of AKI on both KLK1 excretion and urine KLK1 CpG methylation ( Figure 5), elevated KLK1 methylation would be predicted to decrease KLK1 excretion in AKI, as occurs in early/incipient AKI (Figure 2). Increased KLK1 excretion in later/established AKI (Figure 2) thus highlights the influence of epinephrine (Figure 3) to elevate KLK1, even in the face of opposing epigenetic influence.
We investigated the 4 most proximal consecutive CpG sites in KLK1 promoter (Figure 1). This proximal promoter region is unusually polymorphic, containing a polyguanine length polymorphism coupled with multiple basesubstitution variants that constitute at least ten different alleles or haplotypes [28]. Functional/transfection analysis of several alleles in this region suggests that different variants lead to alterations in expression of the KLK1 gene [29]. Since genetic variation may contribute to AKI susceptibility [30], this hypothesis warrants future studies of KLK1 promoter variants in larger cohorts, assessing the effects of such variants on both susceptibility and recovery in AKI, since exogenous KLK1 does exert protective effects against aminoglycoside-induced AKI [7,8].
It should be noted that the sources of DNA for these epigenetic studies in blood and urine are likely to be heterogeneous -blood DNA could emerge from any leukocyte subpopulation, while DNA in urine can emerge from cell types other than renal. Nor have we established whether the promoter CpG methylation events we observed have functional consequences for transcription.

Advantages and limitations of this study
Our conclusions are derived from analysis of four subject groups ( Figure 2): two degrees of AKI (established versus incipient/early), and two kinds of controls (ICU versus healthy/non-hospital). Furthermore, we were able to probe clinical and biochemical characteristics of cases and controls to identify elevated epinephrine as a likely driver of increased KLK1 excretion (Figure 3, Table 3). While we were able to evaluate epigenetic factors in control of KLK1 excretion in the form of promoter CpG methylation ( Figures 1, 4), we did not study other epigenetic mechanisms, such as histone modifications, nor could we probe gene expression more directly by evaluating transcript abundance in tissue, since biopsies would have been hazardous. Furthermore, the established AKI cases were ascertained at a later time point than either AKI cases or controls in the ICU cohorts. Finally, the results would benefit from replication, given the numbers of subjects studied ( Figure 2, Table 1), as well as evaluation of additional mediators of AKI risk and repair.

Conclusions
In conclusion, human patients with established AKI display an unexpected increase in urinary KLK1 enzymatic activity excretion ( Figure 2); the effect is reproducible across control groups, and seems to be driven by epinephrine excess in the setting of hemodynamic instability (Figures 3, 6). AKI and controls differed in KLK1 promoter CpG methylation in blood DNA (AKI > controls), and KLK1 CpG methylation differed systematically from global control (LINE-1 element) methylation, suggesting a potential role of epigenetic factors in AKI susceptibility (Figure 4).