Genetic variants of redox enzymes and risk of cytotoxicity in response to oxidative stress invitro: implications for oedematous severe childhood malnutrition

Investigators & Affiliations:

KG Marshall1, RC Landis2, K Hamilton1, S Howell1, H. Fletcher3, TE Forrester1, and CA McKenzie1

1 Tropical Metabolism Research Unit, Tropical Medicine Research Institute, University of the West Indies, 2 Chronic Disease Research Centre, Tropical Medicine Research Institute, University of the West Indies, Cave Hill, Bridgetown, Barbados and 3Department of Obstetrics, Gynaecology, and Child Health, University Hospital of the West Indies


Caribbean Health Research Council


Oedematous Severe Childhood Malnutrition (OSCM) is associated with evidence of oxidative stress and with poorer outcomes than non‐oedematous severe childhood malnutrition (NOSCM); the reasons for variation in risk of oedema in malnutrition are uncertain – geneticallydetermined susceptibility may play a role. Our aim was to develop an in vitro model to investigate whether cellular damage in response to an oxidant stressor is influenced by genetic variation. Blood samples were obtained from routine attenders (n = 103) at the Antenatal Clinic of the University Hospital of the West Indies (UHWI). Peripheral blood mononuclear cells (PBMCs) were exposed to 50 mM hydrogen peroxide for up to four hours. Percentage cytotoxicity was estimated by measuring lactate dehydrogenase (LDH) release. Variants were typed in four glutathione S‐transferase (GST) genes (GSTM1, GSTM3, GSTP1, and GSTT1) and eight other genes involved in redox metabolism and oxidative stress. In univariate analyses the GSTM3 “Deletion” allele was associated with increased cytotoxicity (P = 0.013). In multivariable analyses, the best‐fitting model included only the four GST markers; the GSTM3 and GSTM1 markers were significantly associated with cytotoxicity (P < 0.05) whereas the other two GST markers were not. Our results suggest that genetic variants in the GST superfamily may influence outcomes at the cellular level after exposure to an oxidant stressor, and provide validation of the notion that genetically‐determined susceptibility may influence risk of OSCM. The approach employed here has the potential to be used to investigate other disorders where oxidant stress plays a role.

It remains uncertain why oedema, and other features including impaired antioxidant status, should be found in only some severely malnourished children and not others when the environmental exposures of all severely malnourished children in any particular locale appear to be very similar. Even more difficult to explain, perhaps, is the occurrence of typical features of the clinical syndrome of kwashiorkor in children who do not have other objective signs of undernutrition (Hendrickse, 1991). While there have been improvements in our understanding of the metabolism of amino acids, GSH, and whole body protein turnover in OSCM (Reid et al., 2000; Badaloo et al., 2002); it appears that our understanding of causal pathways in OSCM is still incomplete. A double‐blind randomised trial of antioxidant supplementation failed to demonstrate significant reduction in risk of kwashiorkor (Ciliberto et al., 2005) and a more recent randomised trial of three different fortified spreads for treatment of moderate wasting, where there were significant differences in recovery rates (two of the groups were the same and showed significant improvement compared to the third), reported that the incidence of oedematous malnutrition was the same in all three treatment groups (Matilsky et al., 2009). It seems fairly clear that other approaches to understanding the causal biology of OSCM need to be employed. One hypothesis to explain why, given relatively uniform environmental exposure(s), some at‐risk children end up with OSCM and others do not, is that there is inter‐individual variation in response to the exposure(s); there is evidence that oxidant stress is one such “exposure” (Golden & Ramdath, 1987). The general suggestion that genetic variation might influence differences in clinical appearances in the face of relatively uniform environmental exposure(s) is relatively novel and has not been explored extensively in SCM. The main questions being examined in the present study are whether a working model of “environmental” exposure (H2O2) could be developed and whether that model could be used to filter or evaluate candidate genetic variants identified in retrospective case‐control studies of risk of OSCM. Hence, we examined whether 14 variants in 12 candidate genes were associated with cytotoxicity (percentage cell death, %cyt) in an in vitro model employing PBMCs, from healthy donors, exposed to an oxidant stressor (H2O2).

In univariate analyses, only the GSTM3 marker was associated with %cyt at the 5% level (P = 0.013). We found that homozygosity for the Del allele of GSTM3 was associated with an approximately 40% increase in %cyt when compared to homozygosity for the Ins allele. None of the other markers demonstrated associations with %cyt at the 5% level in univariate analyses and their effects were relatively small (the median change in %cyt per step change from one allele to another was only 2.4%). When we fitted multivariable regression models in order to assess the joint effects of multiple genetic markers (on the basis that resistance to oxidant stressors is almost certain to be “polygenic”) we found suggestive evidence that only four of the 14 markers (ie variants in GSTM1, GSTM3, GSTP1, and GSTT1) had an influence on model fit. That is, excluding any one of these four markers resulted in a worsening of the fit of the linear model whereas excluding any of the other ten markers resulted in an improvement in the fit of the model.

All of the genetic variants analysed in this paper were selected because they have been considered previously or are being considered as potential markers of risk for OSCM in ongoing exploratory case‐control studies (Marshall et al., 2006a; Marshall et al., 2006b; Marshall et al., 2011). The four genetic variants that jointly had an effect on model fit are all members of the GST superfamily. These candidate variants can be considered to be on the "second line of defence" in regard to oxidant stressors because they participate in detoxification reactions only when the primary defences have been breached, catalysing irreversible conjugation of GSH to a wide variety of electrophilic compounds including organic hydroperoxides, unsaturated hydroxyalkenals, and epoxides when required (Commandeur et al., 1995; Wang & Ballatori, 1998). The other "second line" gene variants tested here were not significantly associated with %cyt. For these others it could be argued that they are actually on a "third line" and that these genes would only come into play if the second line (GSTs, say) was breached.

The remainder of the variants that were examined are all in genes coding for enzymes that are either on the "front line" of anti‐oxidant defence or involved in the generation of oxidant species. Catalase, by catalyzing the decomposition of H2O2 into oxygen and water, superoxide dismutase through the dismutation of superoxide into oxygen and H2O2, and microsomal epoxide hydrolase via the hydration of chemically reactive epoxides to dihydrodiol derivatives are involved in the removal of reactive oxygen species. In contrast, NADPH dehydrogenase (quinone) might be a potent source of ROS in the inner mitochondrial membrane (Lin et al., 2003), while NAD(P)H oxidase catalyses the production of superoxide radicals from molecular oxygen (Inoue et al., 1998). None of the variants of these enzymes were associated with cytotoxicity in this study. In retrospect, this might have been expected; natural selection would be unlikely to tolerate serious dysfunction in genes on the front line of defence – variation in genes of the second (or third) line can be tolerated, however, because most of the time these defences are not required, they only come into play, presumably, under circumstances where the first line defences have failed (and in which case, variation in the second line might be associated with a worsening of the situation). Furthermore, many of these front line defences depend on micronutrients for their activity (either in the active site or as cofactors) – genetically‐determined reductions in activity will be difficult or impossible to detect in the presence of the very large effect on activity due to the deficiency itself.

We have reported previously that GSTP1 and GSTT1 genetic variants may have an effect on risk of OSCM (Marshall et al., 2006b); in this paper we have found suggestive evidence that genetic variation in the GST superfamily has an effect on cellular response to H2O2. The results of the current investigation provide a biological validation of the notion that genetic variation may account for cellular response to oxidative stress; it is thus legitimate to look for genetic variants that might explain at least some part of the risk of OSCM (inasmuch as) other than oxidative stress has been shown to be associated with OSCM). What is required now is a more comprehensive and detailed investigation of the problem; a wider variety of cell types need to be subjected to different types of oxidative stressors and different phenotypic markers (e.g. lipid peroxidation, protein carbonylation cell death need to be assayed. These additional read‐outs can be tested against genome‐wide microarrays of sequence variation (tens, to hundreds, of thousands of single nucleotide polymorphisms, SNPs) so that a more comprehensive view of the influence of genes and related pathways involved in these outcomes can be developed. Thus, the approach that we have reported can either be used to identify candidate genes or candidate pathways that can be tested in case‐control studies or, alternatively, as in the present paper, can be used to validate and possibly probe the candidacy of genetic variants that have been identified in case‐control studies.

Back to top