Most subjects who express virtually no plasma PAF-AH activity harbor two F279 alleles; heterozygous subjects express half the normal plasma activity [9]. Molecular studies revealed that the recombinant mutant protein is inactive [9]. The fact that a bulky residue (phenylalanine) replaces a smaller amino acid (valine) near the active site of the enzyme likely accounts for the loss of function associated with this mutation [9]. While the initial discovery of V279F was made in a Japanese cohort, the mutation was subsequently reported in Korean [4], Taiwanese [3], and Chinese [5] populations, and in subjects from Turkey, Azerbaijan, and Kyrgyzstan [11]. Allele frequency varies widely in these populations: the highest prevalence was reported in Japan and in Taiwan, followed by Korea and China (Table 1). Our studies, and those of others [12,13,14], revealed that complete PAF-AH deficiency does not occur in Caucasian subjects, and this has been confirmed in public databases. This variant likely originated in Japanese and Korean ancestors. The clinical consequences associated with complete and partial PAF-AH deficiency have been evaluated in patients suffering from a variety of inflammatory conditions in the vascular, respiratory, renal, and digestive systems (Table 1). These analyses clearly demonstrated that the incidence of the F279 allele correlates with several manifestations of vascular disease in Japanese subjects; multiple studies by Yamada and co-workers and by Unno et al. formed the bases to support this conclusion (Table 1).

A study by Yamamoto and co-workers demonstrated that the intima media thickness in Japanese type 2 diabetics who harbored the F279 allele was significantly greater than that observed in age-matched patients who expressed wild-type PAF-AH [30]. This study suggested that F279 was associated with the development of carotid atherosclerosis in elderly diabetics. Wang and co-workers reported a significant increase in the oxLDL/LDL ratio in homozygous deficient subjects [38], suggesting that complete PAF-AH deficiency increases the susceptibility of LDL to oxidation, as previously shown in purified in vitro systems [39].

Some of these observations, however, were not confirmed in other laboratories. Reports by Liu et al. and by Hou and co-workers in Taiwanese and Chinese patients reported no association of F279 with myocardial infarction or CHD [3,5]. The relatively low frequency of the mutant allele in the Chinese population may explain results by Hou et al., as correctly pointed out by the authors [5]. In addition, the fact that neither study evaluated the results based on gender may have prevented the authors from identifying potential correlations. This is an important issue when one considers that PAF-AH deficiency has been shown to be more common in Japanese male, but not female, patients diagnosed with coronary artery disease, compared to healthy controls [23]. A study in Korean subjects reported that the frequency of F279 in this population is similar to that observed in Japan, and that this allele is less frequent in Korean men diagnosed with cardiovascular disease (Table 1, [4]). The reasons for this discrepancy are unknown, but it is possible that the criteria used for patient selection differed from those utilized in other studies.

Studies in Kei Satoh’s laboratory demonstrated that the incidence of F279 is associated with stroke and cerebral hemorrhage in Japanese subjects [26,27,28]. Interestingly, Zhang and co-workers also reported that F279 may be an independent risk for atherosclerotic cerebral infarction in Chinese subjects [29], suggesting that the clinical impact of F279 is not limited to Japanese populations. Studies in the respiratory system revealed similar trends. We found that the incidence and severity of asthma was higher in Japanese PAF-AH deficient subjects compared to healthy controls [10]. This finding was confirmed by another group [15] but not by N. Satoh et al. [16]. A limitation of the study by N. Satoh et al. is that the prevalence of the F279 allele in control subjects was higher than that detected in most other reports in Japanese populations (Table 1). It is possible that the higher-than-normal frequency of F279 in the controls accounted for the observations reported in this study.

Additional work in patients afflicted by a variety of diseases revealed potential contributions of F279 to the severity, but not the incidence, of a number of syndromes. In these analyses, F279 was shown to contribute to disease progression rather than onset. First, while F279 was reportedly unrelated to either susceptibility or severity of conventional multiple sclerosis, it increased the severity of female opticospinal multiple sclerosis [36]. Second, Nakamura and co-workers found no significant differences in genotypic frequency of ulcerative colitis patients compared to controls [34]. However, unresponsiveness to steroid therapy was significantly higher in patients who harbored one F279 allele compared to those who did not (66.7 vs. 27.6 percent, p = 0.019). The authors concluded that F279 could be a useful marker to predict responsiveness to steroid therapy [34]. Third, no significant differences have been detected in genotype frequency between Kawasaki Disease patients and controls [37]. However, patients who harbored one or two F279 alleles required additional intravenous immunoglobulin administration (52% vs 14%, p = 0.001). Three studies reported that F279 correlates with the severity of renal diseases (Table 1). Tanaka and co-workers showed that IgA nephritic children who carried the F279 allele excreted more protein in the urine compared to patients who harbored only the wild-type allele [31]. In addition, the percentage of glomeruli with mesangial cell proliferation was significantly greater in F279 carriers, suggesting that the mutation may increase the severity of childhood IgA nephropathy [31]. Second, patients diagnosed with steroid-responsive nephritic syndrome and who harbored one F279 allele had more relapses during the first year after disease onset compared to patients who carried only the wild-type allele. This observation suggests that genotyping at this locus could facilitate identification of children likely to have a disease relapse [32]. Finally, the mean duration of oligoanuria was significantly longer in heterozygous compared to homozygous wild- type patients diagnosed with E. coli O157-associated hemolytic uremic syndrome (p = 0.012, [33]). Most (75%) heterozygous subjects required dialysis, while most (63%) patients who harbored only the wild-type allele did not. Interestingly, aberrant PAF production has been reported in hemolytic uremic syndrome patients. A study in Caucasians who did not harbor F279, showed a rise in PAF levels following infection with E. coli O157, and decreased levels of the mediator during development of the disease [14]. It is possible that the initial increase in PAF levels is followed by resolution without full disease development in subjects who can rapidly and effectively inactivate PAF. Unfortunately, this study did not assess the levels of PAF-AH activity, so it is not possible to draw firm conclusions at this point. It is tempting to speculate, however, that subjects deficient in PAF-AH activity may have more severe forms of hemolytic uremic syndrome owing to impaired abilities to inactivate PAF. In summary, results from independent studies indicate that the presence of F279 increases the severity of several renal disorders and that identification of subjects who harbor this mutation could have significant clinical value.

A relatively uncommon mutation (Q281R) located two amino acids downstream of the most prevalent inactivating PAF-AH mutation was shown to result in severe, but not complete, loss of enzymatic activity ([6,10], Figure 1). The precise incidence of this variant is likely quite low, even in Japanese populations. Previous studies suggested an allele frequency of approximately 1.0 % for this variant [7,40]. The Q281R mutation is near the active site of PAF-AH and is therefore likely to affect proper folding in the region, thus impairing enzymatic activity. A woman who harbored two R281 alleles suffered from coronary artery disease and hypertension [6]. In addition, a patient harboring F279 and R281 alleles had severe asthma [10]. Finally, Kujiraoka et al. identified a diabetic subject who was heterozygous for the Q281R mutation [40]. These isolated observations are consistent with findings in subjects harboring the F279 allele and they suggest that impaired PAF-AH function correlates with the incidence of several human diseases.

Ishihara and co-workers described two rare mutations in the coding region of PAF-AH. An insertion of adenine at nucleotide 191 in exon 3 (Ins191) creates a premature termination at codon 63 and impairs PAF-AH function ([7], Figure 1). In addition, a novel mutation in which adenine is substituted for thymine at nucleotide 950 in exon 10, results in substitution of N for I at codon 317 (I317N, Figure 1). This mutation could theoretically create a new N-linked glycosylation site on PAF-AH, but this has not been confirmed experimentally. Ins191 and I317N impair PAF-AH secretion by macrophages isolated from subjects carrying the mutant alleles [7].

An issue that merits consideration is related to the functional consequences associated with partial versus complete deficiency of PAF-AH. In certain settings, partial deficiency of the enzymatic activity may result in a disease phenotype intermediate between that of subjects harboring two wild-type alleles and that of individuals harboring two mutant alleles. However, this may not always be the case. In fact, a 50% reduction in PAF-AH levels such as that typical of subjects harboring one F279 allele, may not significantly affect the extent of substrate hydrolysis [41]. When substrate levels are relatively low (i.e., nanomolar range), the rate-limiting step of the reaction is likely determined by factors other than total enzyme activity levels. Instead, the type of lipoprotein with which the enzyme associates, substrate levels, accessibility of substrates to the active site, and/or their solubility may determine the extent of hydrolysis [41]. A corollary of this observation is that partial deficiency of PAF-AH due to genetic, pharmacologic, or other causes may have limited functional consequences in vivo. Wang and co-workers correctly pointed out that it may not be appropriate to jointly analyze heterozygous and homozygous deficient subjects in clinical studies aimed at assessing the consequences of PAF-AH deficiency [38]. To rigorously determine the impact of aberrant PAF-AH expression in human diseases, it will be essential to independently compare cohorts of subjects harboring 0, 1, and 2 functional PAF-AH alleles.

The presence of arginine at codon 92 as the major allele in plasma PAF-AH is observed only in humans and in mice (Figure 2). This indicates that absolute conservation of this residue is not essential for proper function. The role played by the amino acid present at this position has been analyzed in a variety of subjects from different populations. The largest study conducted thus far (3234 subjects) detected modest but statistically significant decreases in plasma activity levels in German subjects harboring 1 or 2 H92 alleles [43]. Similar results were recently reported by Li et al. who analyzed a cohort of ARDS patients [44]. In contrast, studies in Chinese and European subjects revealed no effects of H92 on total plasma PAF-AH activity [5,45]. These observations are consistent with kinetic studies by Kruse et al. [42] who expressed, purified, and tested the recombinant mutant protein. These investigators found that purified preparations of R92H and wild-type PAF-AH displayed similar affinities for PAF and hydrolyzed the substrate at comparable rates. While susceptibility of the mutant to substrate inhibition was much higher than that of the wild-type protein, it is clear that this was not a reflection of increased substrate affinity. Our own studies revealed that the presence of histidine at position 92 does not affect binding of PAF-AH to LDL or HDL in vitro (Figure 2, [46]), and does not alter substrate affinity (Table 2).

These combined observations indicate that the presence of histidine at position 92 of PAF-AH has modest or no impact on optimized plasma activity levels, does not significantly affect classic biochemical properties of the protein, and does not seem to alter location. However, it would be of importance to complement these studies with lipoprotein distribution profiles in fresh plasma samples from subjects that harbor 0-1-2 H92 alleles as these analyses would definitively establish whether H92 contributes to the ability of PAF-AH to bind to lipoprotein particles in vivo.

Genetic studies revealed that the allele frequency of H92 varies among populations. A method of base-quenched probe for polymerase chain reaction genotyping that requires only a pair of primers and one fluorescent probe was recently described for the detection of this polymorphism [47]. In public databases, the lowest frequency (13.5%) has been reported in a Japanese cohort and the highest prevalence (35.0%) was observed in a primarily Caucasian cohort (http://www.ncbi.nlm.nih.gov/). Clinical studies centered on the contribution of H92 to disease incidence revealed similar frequencies of this polymorphism in healthy subjects compared to atopic and asthmatic patients (Table 3, [42]) and in patients with schizophrenia (Table 3, [12]). In addition, there seem to be no changes in the frequency of this polymorphism in patients with coronary artery diseases or myocardial infarction, compared to healthy subjects.

Two exceptions include a study by Ninio et al. describing increased risk of coronary artery disease associated with H92 [45] and a report by Sutton et al. [8] that found the opposite trend, that is, a protective effect of the minor allele (Table 3). The suitability of control cohorts utilized in these two studies has been questioned owing to the fact that the frequency of the H92 allele (22.4 and 34%, respectively) differed from that observed in other studies in similar populations (Table 3) and in an unaffected cohort (n=718) of GENECARD [43]. In addition, the report by Sutton et al. indicated that no differential gene expression was observed in diseased arteries and showed no influence of the polymorphism on disease burden in the aorta [8]. Given that a larger recent study [43] failed to confirm the contradictory associations reported in these analyses [8,45], it is tempting to conclude that H92 does not modulate, or has a modest impact, on vascular disease.

Recent work by Li and co-workers [44] showed that ARDS patients carrying the H92 allele were mechanically ventilated for a longer time period when compared to non-carriers. This patient group expressed 25.8% lower plasma PAF-AH activity compared to patients harboring the wild-type allele. This interesting study also reported increased survival in patients who expressed higher plasma PAF-AH levels seven days following diagnosis. The results indicate that inter-individual variability in plasma PAF-AH activity may have prognostic value and suggest that PAF-AH genotyping and assessment of plasma activity levels in critically ill patients may help in the selection of therapeutic approaches that can be tailored to the needs of individual patients.

Individuals harboring a threonine at position 198 of PAF-AH were first described by Bell and co-workers [12]. The degree of conservation of this residue across species is moderate (Figure 3). Expression of the recombinant variant revealed that enzymatic activity was preserved [42]. However, the mutant protein displayed decreased affinity for PAF (Table 2, [42]). Our studies revealed that the presence of T198 did not affect binding to LDL or HDL in vitro (Figure 3, [46]). As in the case of R92H, it would be of interest to complement these studies with lipoprotein distribution profiles in fresh plasma samples from subjects that harbor 0-1-2 T198 alleles, as these analyses would establish whether T198 contributes to the ability of PAF-AH to associate with lipoprotein particles in vivo.

Individuals homozygous for the I198T polymorphism are rare, so most of the work in humans has been conducted in heterozygous subjects. There is controversy on the impact of T198 on the levels of expression of plasma PAF-AH protein and activity. Three investigations revealed no statistically significant effect of T198 on enzymatic activity [43,44,45]. However, a recent study by Hou et al. showed that subjects harboring this variant expressed substantially lower levels of PAF-AH activity [5]. The haplotypes of these subjects differed from those of the control subjects in more than the I198T polymorphism [5] and it is not clear whether these additional genetic variations contributed to the observed decreases in activity. This is particularly important as PAF-AH risk alleles may have an additive effect that leads to the generation of dysfunctional protein products [8]. Sutton et al. reported a trend towards decreased PAF-AH expression in the aorta of subjects harboring T198 [8]. These observations leave open the possibility that at least in some populations, T198 affects activity and/or protein levels, or protein stability, in addition to its effect on substrate affinity.

Public databases (http://www.ncbi.nlm.nih.gov/) report that the lowest T198 allele frequency occurs in Caucasian (4.2–12.0%) and Asian (4.3–11.0%, excluding Japanese) cohorts. The frequency is much higher in Japanese subjects (19.7%), in African Americans (23.5%), and in sub-Saharan African populations (28.5%).

Kruse and co-workers found the T198 variant to be highly associated with total IgE concentrations in an atopic population and with atopic asthma in an asthmatic population (Table 4, [42]). Interestingly, T198 seems to be somewhat more prevalent in ARDS patients [44] compared to control populations with similar ethnic backgrounds (Table 4), suggesting that additional, well-controlled studies related to the impact of T198 on ARDS incidence and severity may be warranted.

The prevalence of the T198 polymorphism has been found to be weakly associated with the incidence of schizophrenia ([12], Table 4). Mechanistically, it is possible that the variant contributes to these syndromes as a consequence of decreased protein expression, at least in some populations [5]. Alternatively, since substrate concentration and/or accessibility to the active site tend to be rate limiting in vivo [41], decreased substrate affinity (i.e, increased Km [42]) may account for the observed effects.

A variety of studies have assessed the incidence of T198 in patients with cardiovascular disease (Table 4). These analyses have led to conflicting results possibly owing to the size and heterogeneity of the patient populations and/or the type of statistical analyses utilized. Sutton et al. used logistic regression modeling and found no significant association of T198 with CAD outcome [8]. Similarly, Ninio et al. reported that the incidence of cardiovascular events was not significantly associated with survival outcome by univariate analyses [45]. In contrast, Hoffmann and co-workers reported lower incidence of T198 in a CAD cohort. The larger size of the populations studied by Hoffmann et al. may explain why this correlation did not become evident in other studies. Kokubo and co-workers reported that the presence of T198 was associated with hypertension in women when the data were adjusted for several factors and analyzed by multivariate logistic regression analyses [49]. In summary, while there is no absolute agreement on the precise role of the somewhat rare T198 polymorphism on activity and/or disease incidence, it seems that additional large-scale studies should further evaluate whether, and how, T198 contributes to inflammatory and vascular disorders. These analyses should include activity measurements as at least one study [5] showed a remarkably high inhibitory effect of this polymorphism on PAF-AH levels. It seems that analyses in Japanese cohorts would be ideal for this purpose, owing to the reportedly high prevalence of the polymorphism in this population.

Individuals harboring a valine at position 379 of PAF-AH were described a few years ago [12]. While humans express predominantly an alanine residue at this position, most species harbor a valine residue (Figure 4). This feature forecasts limited impact of V379 on enzyme function in humans. Purified recombinant A379V displays increased maximal velocity and a two-fold decrease in substrate affinity compared to the wild-type protein ([42], Table 2). We found that the A379V variant associated with LDL and HDL in fashions comparable to those of the human protein (Figure 4, [46]). However, a systematic study to characterize the lipoprotein distribution of PAF-AH in subjects homozygous for the V379 polymorphism has not, to my knowledge, been performed.

As in the case of other polymorphisms, there is controversy on the impact of V379 on plasma PAF-AH protein and activity levels. Hoffmann and co-workers reported increases in PAF-AH plasma levels with a clear gene-dose effect [43]. Similarly, Ninio and collaborators found that the presence of V379 was significantly associated with increased PAF-AH activity in a co-dominant fashion (p = 0.02, [45]). Schnabel and co-workers reported higher PAF-AH activity in subjects carrying two V379 alleles and extended the observation to a large community-based cohort [50]. While the observed differences were statistically significant in these two instances, the increases in activity were modest (<10 %). Three unrelated studies found no statistically significant effect of V379 on enzymatic activity in ARDS patients (Table 5), in a Chinese cohort, and in a Caucasian control group [5,44,52]. An interesting study in a Taiwanese population showed dramatic decreases in activity in subjects homozygous for the V379 allele [3]. The fact that this is the only study reporting such large effects suggests that an additional factor(s) peculiar to this group of subjects likely contributed to the results. In summary, the presence of V379 may confer a very modest increase in plasma PAF-AH activity levels in some populations. These effects are likely mediated by the enzyme itself, because studies that evaluated properties of the purified recombinant A379V variant in a lipoprotein-free assay system recapitulated the increased enzymatic activity that was observed in some clinical studies [42].

The prevalence of the V379 allele varies among populations. The lowest frequency (9.2–9.5%) has been reported in Japanese populations and the highest prevalence (26.7–32.6%) was observed in Sub-Saharan African subjects. A number of groups have investigated whether V379 is associated with the incidence of various diseases and/or physiological responses (Table 4). Bell and co-workers found no significant differences in allele frequency between patients with schizophrenia and control subjects, although a trend (p = 0.057) towards increased incidence in the patients was noted [12]. In contrast, V379 was found to be highly associated with specific sensitization in a German atopic population and with atopic asthma in a British cohort (Table 5, [42]). Wootton and co-workers reported that after ten weeks of intensive physical exercise, individuals homozygous for the V379 allele showed decreased percentage of adipose tissue compared to AV and AA genotype groups (p = 0.01, [52]). A potentially interesting observation by Li and co-workers is that ARDS patients seem to have a lower incidence of the V379 polymorphism compared to similar Caucasian cohorts [13.4% versus 19.6–25.8% in control Caucasian subjects, excluding reference [42], Table 5]. Unfortunately, this analysis only evaluated ARDS patients, so a comparison with healthy control subjects can only be made across studies. Nonetheless, the fact that ARDS survivors expressed higher plasma PAF-AH activity [44], combined with the fact that the prevalence of the V379 allele was lower in at least one ARDS cohort, suggests that V379 may offer some protection from this disease.

Eight studies investigated whether V379 contributes to vascular disease [3,4,5,8,43,45,52,53]. Four of these analyses reported no statistically significant differences in allele frequency of control subjects compared to patients with coronary heart disease or myocardial infarction (Table 5). Three studies found statistically significant, but opposite, relationships between these parameters [3,8,45]. The suitability of control cohorts utilized in two of these reports [8,45] was questioned by others [43], as indicated above. The lack of consensus among these analyses could be due to the criteria utilized for patient selection, combined with the size and nature of the patient populations studied. The mechanism by which potential V379-mediated effects contribute to inflammatory diseases could include direct effects of the polymorphism on enzymatic activity, but this contribution is likely to be relatively modest [45]. The possibility that V379 is in linkage disequilibrium with an un-identified etiological variant cannot be ruled out.

These variants have only been reported in public databases and they occur with various frequencies in human populations. One of the variants (rs 35142331) results in no amino acid change and is thus unlikely to impact protein function. The frequency of the remaining variants is either unknown or very rare in Caucasians. The functional consequences associated with these polymorphisms are currently unknown.

We previously showed that the kinetics of PAF hydrolysis are dependent on the ratio between enzyme and substrate, and that at low substrate levels the rate of hydrolysis deviates from that predicted by Michaelis-Menten kinetics [41]. We investigated the effect of sequentially introducing naturally occurring PAF-AH polymorphisms on the kinetics of PAF hydrolysis at sub-saturating substrate levels. This analysis revealed that single, double, and triple mutants hydrolyzed PAF with similar efficiencies compared to the wild-type protein (Figure 5, left panel).

In these assays, we utilized equal amounts of PAF-AH activity as determined by our standard, optimized, enzyme assay. Thus, the results were qualitatively comparable in both assay types, that is, we observed no major differences in the rate of PAF hydrolysis by PAF-AH variants. However, the experimental design in these studies did not recapitulate the environment in which the enzyme functions in vivo. This is an important issue because lipoproteins alter the efficiency of hydrolysis of phospholipid substrates, presumably by modulating access to the active site, altering substrate presentation, or changing solubility. To investigate this, we incorporated wild-type and PAF-AH variants into LDL, and then assessed the ability of the lipoprotein-associated variants to hydrolyze sub-saturating levels of PAF. Our studies showed that sequential introduction of each polymorphism increased the half–life of PAF in LDL (Figure 5, right panel). These results suggested that in experimental settings modeled to mimic conditions likely to occur in vivo, that is, when substrate is limited, enzyme levels are in excess, and the enzyme functions in its natural environment, factors other than absolute enzyme levels impact the rate of hydrolysis. These studies revealed potentially important functional consequences associated with naturally occurring polymorphisms in PAF-AH. In addition, the results illustrate the importance of complementing traditional biochemical analyses with assays that take into consideration properties peculiar to the gene product and reaction being evaluated.

Previous work demonstrated that PAF-AH is susceptible to inactivation by oxidants, and that key methionine and tyrosine residues confer susceptibility to oxidant attack [57]. We investigated whether naturally occurring PAF-AH variants were differentially affected by oxidants, as this could provide additional mechanistic insight into the functional consequences of genetic alterations. We expressed single, double, and triple mutants harboring R92H, I198T, and A379V, and then investigated if these polymorphic forms of PAF-AH were differentially affected by SIN-1, a peroxynitrite-generating agent [57]. We found that sequential replacement of these residues increased susceptibility to oxidative inactivation (Figure 6). These findings suggested that subjects harboring these polymorphisms may accumulate PAF-AH substrates in settings of elevated oxidant stress that are characteristic of acute inflammation and other syndromes.