Associations between PON1 enzyme activities in human ovarian follicular fluid and serum specimens.

High-density lipoprotein (HDL) provides the cholesterol substrate for steroid hormone synthesis in the human pre-ovulatory ovarian follicle [ 1 ], but also has well-recognized anti-oxidant and anti-inflammatory properties likely to be of importance for reproduction [ 2 ]. HDL functionality is driven, at least in part, by particle composition [ 3 – 5 ]. The HDL-particle comprises a dynamic micelle in which apolipoproteins frame a core of cholesteryl esters and triglycerides surrounded by a phospholipid monolayer; cholesterol, lipid-soluble micronutrients, and paraoxonase 1 (PON1) are integrated within [ 2 ]. HDL is the exclusive carrier of PON1, and circulating HDL particles cross the blood-follicle barrier and are the major source of PON1 in the ovarian follicle. Furthermore, PON1 is subject to several polymorphisms, the most studied of which is an arginine (R)-glutamine (Q) substitution at the 192 nd amino acid [ 3 ]. This PON1 Q192R polymorphism is functional (meaning it affects enzyme activity) and the mutant R allele is particularly prevalent among Asians [ 6 , 7 ]. Our group previously reported associations between the activity of follicular fluid (FF) PON1 and embryo quality among women undergoing in vitro fertilization (IVF) [ 8 ], presumably due to reduced lipid peroxidation [ 9 ]. The invasive nature of FF collection limits its application to IVF populations and there are limited data available to characterize associations with less invasively acquired serum specimens from non-clinical populations. With growing interest in the relevance of FF lipoproteins to human reproduction it is critical to assess suitability as biomarkers for use in clinical and epidemiologic applications [ 10 ]. To address the pending data gap, we augmented our recent characterization of distributions, sources of variability and reliability characteristics for FF HDL-particle constituents in addition to those measured in serum among 171 IVF patients [ 11 – 13 ]. Here we characterized PON1 enzymes in FF and serum measured as paraoxonase and arylesterase activities, and assessed the impact of the PON1 Q192R polymorphism on associations between PON1 enzymes in FF and serum. Given that the anti-oxidant activity of HDL is, in part, determined by particle size, we further assessed associations between HDL particle size and enzyme activities according to the PON1 Q192R polymorphism.

As described by Table 1 most participants were assigned either the QQ (40.9%) or QR (42.6%) PON1 Q192R phenotype; 16.4% presented with the homozygous RR PON1 Q192R polymorphism. PON1 Q192R phenotype differed by race (P < 0.0001). The QQ polymorphism was less prevalent among Asians (14.9%) than among non-Asians (48.7%). In contrast, the RR polymorphism was common among Asians (31.9%) but infrequent among non-Asians (10.4%). The QQ phenotype conferred higher arylesterase and lower paraoxonase activities, whereas those with the RR phenotype had lower arylesterase and higher paraoxonase activities; the difference was statistically significant only for paraoxonase activity ( Fig 1 ). We detected no differences by age, BMI, cigarette smoking, diagnosis, or COS protocol. a n = 9 missing values; b n = 6 missing values; c includes n = 1 recurrent pregnancy loss; d includes n = 10 endometriosis, n = 10 tubal factor, n = 7 polycystic ovary syndrome and n = 3 anovulation; * P < 0.05 for across PON1 Q192R phenotypes. BMI, body mass index; COS, controlled ovarian stimulation; DOR, diminished ovarian reserve; FF, follicular fluid; PGD, pre-implantation genetic diagnosis. Geometric means and standard deviations (SD) of arylesterase activity (shaded bars) and paraoxonase activity (open bars) presented; differences for enzyme activities in FF and serum are significant (P < 0.0001) within each PON1 Q192R phenotype; * P < 0.0001 for difference across PON1 Q192R phenotypes; † n = 10 missing values; ‡ n = 14 missing values. For all phenotypes, enzyme activities were significantly (P < 0.0001) higher in serum than in FF ( Fig 1 ). Absolute differences in FF and serum arylesterase activities decreased with an increasing number of R alleles (i.e., QQ = 67.6, QR = 59.6, RR = 54.3 kIU/L), whereas paraoxonase activity differences increased with more R alleles (i.e., QQ = 23.9, QR = 53.2, RR = 83.7 IU/L). Still, we detected strong and moderate correlations between enzyme activities measured in FF and serum for paraoxonase (r = 0.80, P < 0.0001) and arylesterase (r = 0.46, P < 0.0001). In Table 2 , sources of variability and reliability indices are described for arylesterase and paraoxonase activities overall and according to PON1 Q192R phenotype. For PON1 activities measured in FF, variability between women was the major contributor to each, with the exception of paraoxonase activity among women with the RR phenotype. For the latter, sources between-follicles was the main contributor (56.04%) to overall variability and this value was significantly higher than for the QR (23.08%) or QQ (21.58%) phenotypes. Variability attributed to analytic factors was uniformly low, although the contribution to paraoxonase was significantly higher in the QQ (3.16%) phenotype than in the QR (1.24%) and RR phenotypes (0.56%). II values were low for arylesterase activity overall (0.26), and similar when stratified by phenotype (0.23–0.30). Although II was also low for overall paraoxonase activity (0.12), values were substantially increased when stratified by phenotype (0.32–1.30). A single specimen collection was sufficient to characterize woman-specific mean values in all scenarios. CVs were mostly below 10%, although 11.51% for paraoxonase activity overall. ICC values approximated or exceeded 0.80 (0.75–0.89) for all groups with the exception of RR phenotype paraoxonase activity (0.43). Variability attributed to analytic factors for serum arylesterase activity was higher for QQ (1.62%) than for QR (0.90%) and RR (0.94%) phenotypes, whereas serum paraoxonase variability attributed to analytic factors was significantly different across all phenotypes. For PON1 activities measured in serum, CVs were below mostly 10%, similar to those measured in FF, and ICC values were close to 1.00 across all PON1 Q192R phenotypes. NOTE: All values were natural log-transformed for the analyses and only women with no missing values were used (i.e., balanced data). a, b, c Different superscript letters indicate P < 0.05 for difference in % σ F 2 and % σ A 2 values between phenotypes for arylesterase or paraoxonase activities. For example, % σ F 2 in FF paraoxonase activity for RR phenotype is significantly different (P < 0.05) from QQ and QR phenotypes. CI, confidence interval; CV, coefficient of variation; ICC, intraclass correlation coefficient; II, index of individuality; k, minimum number of specimens required to estimate the women-specific mean value within 10% of the true value; σ A 2 , variability attributed to analytic factors; σ B 2 , variability between-women; σ F 2 , variability between-follicles; σ T 2 , total variability. As presented in Table 3 , FF HDL particles were associated with FF PON1 enzyme activities regardless of HDL particle size group (e.g., large, medium, and small) or PON1 Q192R phenotype. On the other hand, only medium and small size HDL particles measured in plasma predicted serum arylesterase activity for the QR phenotype, with no impact for the QQ or RR phenotypes. For serum paraoxonase activity, plasma HDL particle predictors varied by phenotype. NOTE: PON enzyme activities and concentrations of HDL particles were natural log-transformed for the analyses. Models were adjusted for age, body mass index, and cigarette smoking. a Generalized estimating equations were used to provide robust standard errors; * indicates P < 0.05; CI, confidence interval. Using 27 size-specific HDL particles measured in FF, we identified that different individual HDL particles sizes predicted different PON1 enzyme activities according to PON1 Q192R phenotypes ( Table 4 ). For the QQ and QR phenotypes, the 8.0–9.4 nm medium-sized FF HDL particles were significant predictors of FF arylesterase activity. On the other hand, for the RR phenotype only the 13.0 nm large-sized HDL particles significantly predicted FF arylesterase activity (21.39% change, 95% CI 2.52–181.62), although this effect estimate was imprecise due to small sample size. Across PON1 Q192R phenotypes, we identified various HDL particle sizes of 8.0 to 9.7 nm diameter as significant, confounder-adjusted predictors of FF paraoxonase activity. NOTE: PON activities and concentrations of specific HDL particle sizes were natural log-transformed for the analyses. Models were adjusted for age, body mass index, and cigarette smoking, and generalized estimating equations were used to provide for robust standard errors. * indicates P < 0.05; CI, confidence interval.

In conclusion, we identified similar reliability characteristics for FF PON1 enzyme activities as serum measures. Stratification by the PON1 Q192R phenotype improved biomarker characteristics for FF PON1 enzyme activities in terms of their likely performance in clinical settings, yet appeared to diminish their suitability for use in population-level epidemiologic studies. Additional experiments about the underlying mechanism are necessary to elucidate reasons why PON1 Q192R turned out to be suitable for clinical investigations. In contrast, serum PON1 enzyme activities appeared suitable for population-level investigations. Although all FF HDL particles determined for large, medium, and small size were positively associated with PON1 activities, specific particle sizes may be important factors for predicting enzyme activities. Still, it is important to recognize the limitations of our results given the promiscuity of the PON1 enzyme substrates employed [ 42 , 44 ]. Given the increasing interest in FF constituents as biomarkers in clinical and research settings and a strong likelihood for PON1 to impact reproductive outcomes, these data should prove useful in guiding clinical use and in designing epidemiologic studies of oxidative stress and IVF endpoints.

Our study was conducted within the usual clinical context of IVF procedures at the University of California at San Francisco (UCSF) Center for Reproductive Health (USA). Sample recruitment and the clinical study protocol were previously described in detail [ 11 ]. Briefly, we enrolled a convenience sample of 180 women undergoing IVF treatment at UCSF from April 10 th , 2010 to June 28 th , 2011. Following controlled gonadotropin-induced ovarian follicle stimulation (COS) according to clinic protocols and subcutaneous administration of human chorionic gonadotropin (hCG), oocytes were retrieved using transvaginal fine needle aspiration and blood specimens were obtained on the same day. We employed a single-follicle design in which the largest follicle (>17 mm diameter) present on each contralateral ovary was aspirated individually and separately from the remaining follicle cohort. Residual, undiluted FF from two separate contralateral follicles and blood specimens were retained and processed for analysis (n = 171) by the Clinical Biochemistry and Oxidative Stress Laboratory at the University of Buffalo, State University of New York (USA). A separate aliquot was sent to LipoScience, Inc. (Chicago, Illinois, USA) for HDL particle size quantification. All participants provided written informed consent prior to study enrollment and the study protocol was approved by the UCSF Committee on Human Research. PON1 arylesterase and paraoxonase activities in FF and serum, and the PON1 Q192R polymorphism phenotype, were determined in duplicate as previously described [ 14 ]. In brief, paraoxonase activity (IU/L) was determined by the rate of formation of p-nitrophenol at 412 nm using 1 mmol/L paraoxon as the substrate in 50 mmol/L glycine buffer, pH 10.5, with 1.0 mmol/L CaCl 2 with and without 1 mol/L NaCl. Arylesterase activity (kIU/L) was determined by the rate of formation of phenol at 270 nm using 4 mmol/L phenyl acetate as the substrate in 20 mM Tris-HCl, pH 8.0, with 1.0 mM CaCl 2 . Arylesterase activity with p-nitrophenyl acetate as substrate was determined only for use in phenotype assignment and was determined as the rate of formation of p-nitrophenol at 405 nm in 25 mmol/L TEA buffer, pH 7.4, with 1.0 mmol/L CaCl 2 with (inhibited arylesterase activity (IA)) or without (non-inhibited arylesterase activity (NIA)) 1 mmol/L phenyl acetate. Water blanks were used to correct for non-enzymatic hydrolysis. The coefficients of variation (CV) were 0.6%-1.4% for PON1 assays. PON1 phenotype was assigned based on the ratio of PON1 enzyme activities using these different analytical conditions and substrates. This activity ratio phenotype has been proven to be 100% accurate in assigning PON1 Q192R phenotype in comparison to Alw1 restriction fragment length polymorphism (i.e., genotype determination) [ 14 ]. We used proton nuclear magnetic resonance spectrometry ( 1 H NMR) to classify HDL particles in serum and FF as large (9.7–13.5 nm), medium (8.3–9.4 nm), and small (7.4–8.2 nm) sizes according to diameter (LipoScience, Inc.) We further quantified size-specific FF HDL particles by identifying NMR signals unique to specific HDL particle sizes according to a previously described method [ 15 ]. We obtained concentrations of 26 HDL particle sizes in FF, ranging from 7.4 nm to 13.5 nm diameter. We normalized PON1 enzyme activities and concentrations of HDL particles and stabilized variances using a natural log transformation prior to data analysis. The distribution of PON1 Q192R phenotypes was characterized by demographic and clinical factors using the χ 2 -test or ANOVA as appropriate. To assess similarities between biologic compartments, we compared PON1 enzyme activities in FF and serum by paired Student-T tests in n = 141 with available data. We also evaluated linear associations between compartments using Pearson correlations. Among n = 118 with no missing values (i.e., balanced data set), we employed two-stage nested ANOVA to characterize sources of variability between-women ( σ B 2 ) and between-follicles ( σ F 2 ) , and due to analytic factors ( σ A 2 ) , which included random variation, laboratory variability, and other factors not captured by σ B 2 or σ F 2 . The relative contribution of each variability source to total measurement variability was calculated and we used overlap of 84% confidence intervals (CI) to evaluate significant differences in geometric mean values [ 16 ]. We characterized the index of individuality (II), to assess the utility of population reference ranges in clinical settings [ 17 ]. We also calculated coefficients of variation (CV%) and intraclass correlation coefficients (ICC), with 95% CIs estimated using the inverse tan transformation of Smith’s variance [ 18 ]. Finally, we determined ‘k,’ the minimum number of specimens required to estimate the woman-specific mean value with 10% error [ 17 ]. For multivariable analysis, we used linear regression models with HDL particles as predictors and PON1 enzyme activities as the outcomes, stratified by PON1 Q192R phenotype. Large, medium, and small size FF and plasma HDL particle groups were simultaneously entered into regression models as independent variables. For HDL particles and enzyme activities measured in FF, we incorporated generalized estimating equations (GEE) to account for the correlated nature of FF measures made within woman [ 19 ]. All models were adjusted for a priori identified confounders, including age in years [ 20 ], body mass index (BMI) in kg/m 2 [ 21 , 22 ], and cigarette smoking as ‘never’ vs. ‘ever’ [ 23 ]. We further assessed 26 size-specific FF HDL particles as predictors of FF PON1 enzyme activities using a forward stepwise selection procedure. Briefly, we screened 26 HDL particle sizes by regressing each on PON1 enzyme activities stratified by PON1 Q192R phenotype and retained only those HDL particles with P < 0.05. We employed a change in quasi-likelihood criterion (QIC) of 1.00 to retain or remove FF HDL particle sizes from the final regression models. Exponentiated regression coefficients from the final models are presented as % change in enzyme activities per ln-transformed μmol/L FF HDL and 95% CIs. SAS v.9.3 (SAS Institute, Cary, NC USA) was used for the analysis and statistical significance was defined as P < 0.05 for a two-tailed test.

(XLSX) Click here for additional data file.