**3. Discussion**

We undertook the study to explore the potential of hydroxychloroquine as a novel targeted therapy addressing key pathophysiological pathways in preeclampsia. We demonstrated that this antimalarial drug affords no apparent protection against hypoxia or oxidative stress in placental explants but that it does have endothelial protective properties. These observations sugges<sup>t</sup> that hydroxychloroquine is a potential therapy for women with established preeclampsia but is unlikely to be useful as a preventative treatment.

We hypothesized that hydroxychloroquine would protect placental tissue from hypoxia-induced injury ex vivo. Specifically, we sought to show that hydroxychloroquine could mitigate the effects of hypoxia and hyperoxia on the placental release of the anti-angiogenic factors sFlt-1 and sEng, as well as on the release of the pro-inflammatory cytokines, TNF-<sup>α</sup>, and activin A. However, we found this not to be the case. Hydroxychloroquine had no effect on modulating hypoxia-induced placental injury. These findings support those of others who tested hydroxychloroquine in a trophoblast-derived cell line exposed to antiphospholipid antibodies as a model of antiphospholipid syndrome [26]. They found that although hydroxychloroquine was able to mitigate trophoblast secretion of IL-6, it had no effect on sEng release [26]. Collectively, this suggests that in an established diagnosis of preeclampsia, the use of hydroxychloroquine may not confer any beneficial effects.

The maternal symptoms of preeclampsia are largely due to widespread maternal endothelial dysfunction [27,28]. Lupus shares this feature as the key mechanism underlying hypertension, renal dysfunction, and other organ injury [29]. Indeed, the endothelial dysfunction in both preeclampsia and lupus have also been shown to be due, at least in part, to excessive oxidative stress secondary to NOX activation [12,30,31]. Recently, in murine models of lupus, hydroxychloroquine was shown to reverse endothelial dysfunction via the downregulation of NOX, and subsequently, oxidative stress [24,32]. Here, we showed that hydroxychloroquine may have similar effects in an in vitro model of preeclampsia-like endothelial dysfunction. Specifically, hydroxychloroquine was able to prevent the TNF-α induction of NOX2 and subsequent oxidative stress in HUVECs but, importantly, was not able to block similar effects induced by sera from preeclamptic women. Interestingly, apocynin, a NOX inhibitor, was able to prevent the effects of both TNF-α and sera from preeclamptic women on NOX2. This confirms the pro-oxidative effects of the sera of preeclamptic women are mediated via NOX2 [12]. Müller-Calleja et al. demonstrated the inhibition of reactive oxygen species (ROS) generated by endosomal NOX in human monocytic cells [33]. However, the concentration used was much higher (10 μM) but still within therapeutic range as compared to ours (3.6 μM). It is apparent that the concentration of hydroxychloroquine differs in various cell types in in vitro experiments. Although several isoforms of NOX family members including NOX1, NOX2 (also called gp91phox), NOX3, NOX 4, and NOX 5 have been reported to date, in endothelial cells, NOX1, NOX2, and NOX4 isoforms are reported to be involved in the inflammatory response and cytokine expression, triggered by angiotensin II treatment, through different mitogen activated protein kinases (MAPK) pathway activation (phosphorylated form of p38 MAPK, extracellular signal regulated kinases, ERK-1/2 and stress-activated protein kinases and c-jun N-terminal kinase, SAPK/JNK) [34].

We have shown before that follistatin, an activin binding protein, can block the endothelial effects of sera from preeclamptic women [12,35]. Compared to women with a normal pregnancy, maternal circulating levels of activin are increased approximately 10-fold in women with preeclampsia [36]. We have not ye<sup>t</sup> explored whether hydroxychloroquine can block activin-mediated effects. However, the current study suggests that sera from preeclamptic women contains factors capable of inducing NOX, which cannot be mitigated by hydroxychloroquine.

Intriguingly, hydroxychloroquine was able to mitigate the effects of both TNF-α and sera from preeclamptic women on the loss of endothelial ZO-1 and integrity. Endothelial ZO-1 is a protein present in endothelial cell–cell junction known to regulate cellular permeability [37]. Any changes in ZO-1 protein, such as induced by a response to inflammatory cytokines, will alter the endothelial cell permeability. The loss of ZO-1 induced by TNF-α is known to be mediated via the activation of NOX [38,39]. Hydroxychloroquine was able to prevent the loss of ZO-1 and the subsequently increased endothelial permeability induced by both TNF-α and sera from preeclamptic women, suggesting these effects may be mediated through a TNF-α-dependent upregulation of NOX. Further evaluation is required to verify this theory, perhaps using TNF-α receptor antagonists co-incubated with the sera of preeclamptic women.

Circulating levels of TNF-α increase in normal pregnan<sup>t</sup> women and are further raised in preeclamptic women [40–42]. These levels are much lower than the concentrations tested in the present study (15 pg/mL vs. 100 ng/mL, respectively). Although in the in vitro model of acute exposure of cells to single high dose may not be a good representation of the in vivo situation, previous studies have reported that 100 ng/mL of TNFα in cultured HUVEC enhanced endothelial cell activation, similar to that observed in preeclampsia [43].

The e ffect of hydroxychloroquine on other pathogenic pathways of preeclampsia have not been explored in this study. For example, it is now thought that a key mechanism of action of antimalarial drugs is the antagonism of toll-like receptor (TLR) signaling and subsequent downstream activation of pro-inflammatory cytokines [16,44]. In preeclampsia, placental expression of TLR3, TLR7, and TLR8 are upregulated [45,46]. The treatment of pregnan<sup>t</sup> rodents with TLR agonists induces a preeclampsia-like phenotype, providing further evidence of other mechanistic pathways that hydroxychloroquine treatment could also target.

In addition to the e ffects of antimalarial agents on TLRs, these drugs have other benefits, such as inhibition of phospholipase A2 (PLA2) enzyme. PLA2 has been implicated in the pathogenesis of preeclampsia and is found to be elevated in both decidual tissue and the sera of preeclamptic women [47,48]. Similarly, in patients with active SLE, there is a 4.6-fold increase in the mean activity of PLA2 [49]. Lipid peroxidation occurs because of oxidative stress induced by the elevated levels of reactive oxygen species. This leads to membrane phospholipid degradation and hence release of arachidonic acid [50]. Zabul et al. have described the potential role of arachidonic acid hydroperoxide underlying the molecular mechanism of oxidative stress in preeclampsia [51]. Arachidonic acid stimulates release of superoxide from neutrophils and macrophages [52]. Antimalarial drugs have been shown to inhibit PLA2 activity and therefore reduce the generation of superoxide, which will be beneficial for improving endothelial dysfunction in preeclamptic patients [50,53].
