**4. Discussion**

The goal of the present study was to determine whether peroxiredoxins play a role in the survival of neonatal male germ cells, as it is the case in spermatozoa. We hypothesized that neonatal germ cells and spermatozoa would likely require different antioxidant machineries, in view of their considerable molecular, morphological and functional differences, as well as their distinct cellular environments. Indeed, the most abundant antioxidants genes in PND3 gonocytes were *Txn1*, *Prdx1*, *Sod1*, *Prdx2, Prdx6*, and *Prdx5*, whereas a mixed population of adult germ cells expressed *Sod1* at a similar level as these genes, followed by *mGst1*, *Prdx1*, *Prdx6*, *Txn1*, *Prdx2* and *Prdx5* at lower levels. These data indicate that neonatal gonocytes express very high levels of antioxidant genes relative to adult germ cells, although one cannot exclude that some of these genes might be highly expressed in discreet subsets of adult spermatogonia, spermatocytes, or spermatids. The requirement for high levels of antioxidant genes in neonatal gonocytes might reflect the production of ROS during their multiple functions, including DNA methylation, cell proliferation, migration, and differentiation, all processes requiring energy and the generation of ROS [13,14].

Indeed, our finding that the direct inhibition of 2-Cys PRDX with conoidin A and the specific inhibition of GSTP1, enzyme necessary for the re-activation of PRDX6 peroxidase activity, with ezatiostat [28–31] induce rapid and extensive cell death in PND3 gonocytes implies that high levels of ROS are formed in these cells in physiological conditions. This further suggests that PRDXs play an essential role in maintaining ROS at levels required for physiological functions, but not high enough to induce cell damage. Our results showing that lipid peroxidation is greatly increased by these inhibitors and is associated with increased cell death sugges<sup>t</sup> that lipid peroxides are in part responsible for the adverse effects on gonocytes. In this context, high levels of antioxidant genes, in particular PRDXs, protect neonatal gonocytes from damaging levels of ROS. Moreover, the data clearly show that other antioxidant genes are not capable of rescuing gonocytes from the deleterious effects of ROS accumulation in the absence of PRDXs. In this regard, PRDXs are clearly essential to the maintenance on ROS homeostasis and gonocytes survival, as they are in spermatozoa [4,6].

Recently, it was reported that as human PRXD6, targeted to the yeas<sup>t</sup> mitochondrial matrix, elicited glutathione disulfide (GSSG) formation upon treatment of cells with H2O2 [32]. Because yeas<sup>t</sup> lack GSTP1, it was suggested that other enzymes may re-activate the peroxidase activity of PRDX6. Although we observed that glutaredoxin 1 (GRX1) and other GSTs were present in gonocytes, the level

of their expression was lower than that of GSTP1 (Figure 1; Table S1). This finding and the decrease in gonocyte survival due to the inhibition of GSTP1 by ezatiostat, indicate that the re-activation of PRDX6 peroxidase activity is probably accomplished by GSTP1 and GSH in gonocytes, as described in other mammalian cells [33].

The finding, that conoidin A and ezatiostat both greatly aggravated H2O2 adverse e ffects, suggests that 2-Cys PRDXs, likely PRDX1 and 2 which were the most abundant at the protein level, as well as PRDX6 are the peroxidases that actively remove ROS such as H2O2 from neonatal gonocytes exposed to oxidative stress. Taken together with the fact that Sertoli cells produce H2O2 [20], these data sugges<sup>t</sup> that the exposure of neonatal gonocytes to ROS produced by Sertoli cells might be part of a physiological process, as it happens in various tissues and cell types [8,34–36]. However, as shown by the e ffects of PRDX inhibitors, ROS levels need to be tightly regulated in order to maintain ROS at non-toxic levels in gonocytes.

The fact that the PRDX inhibitors conoidin A and ezatiostat both decrease RA-induced gonocyte di fferentiation suggests that PRDXs are essential to control ROS levels, preventing cell death and allowing some of the cells to undergo gonocyte di fferentiation. Moreover, the finding that PRDX inhibition promoted high levels of ROS, which prevented gonocyte di fferentiation without impairing viability, suggests that one way by which the Sertoli cells could control the timing of gonocyte di fferentiation would be by producing ROS at levels that would prevent or allow gonocyte di fferentiation. Further studies are necessary to confirm this possibility. Considering recent studies showing that gonocytes are heterogeneous, similarly to spermatogonia [14,37,38], it would be interesting to study whether neonatal testes contain sub-sets of gonocytes with di fferent PRDX activities, corresponding to di fferent functional states.

One of the main di fferences we found with our studies on PRDXs in spermatozoa is related to the role of the PRDX6 iPLA2 activity. While this activity is critical for the survival and fertilizing ability of spermatozoa by repairing oxidized membranes [10–12], our results with MJ33 sugges<sup>t</sup> that the PLA2 activity of PRDX6 plays a role in gonocyte survival mainly in the presence of exogenous oxidative stress. Yet, the increased cell death observed after treating gonocytes for 18 h with 50 μM MJ33 suppressing PRDX6 iPLA2 activity implies that impairing the repair of oxidized membranes in gonocytes can jeopardize PND3 gonocyte survival over time. We recently reported that exogenous addition of arachidonic acid or lysophosphatidic acid (LPA) prevented cell death in spermatozoa treated with MJ33 [39]. The addition of H2O2 to cells where PRDX6 iPLA2 is inactive and possibly arachidonic acid levels were depleted is su fficient to induce a large increase in cell death in only two hours. Thus, excessive levels of ROS combined with an increased lipid peroxidation and depletion of arachidonic acid and/or LPA appear to aggravate the fate of the cells.

Interestingly, we recently reported that neonatal gonocytes express COX2 and other enzymes of the prostaglandin pathway, and produce prostaglandins (PG) E2 and PGF2a [19]. Moreover, blocking PGE2 and PGF2a synthesis with ibuprofen for 24 h correlated with a partial decrease in RA-induced di fferentiation in PND3 gonocytes [19]. However, inhibiting the PRDX6 iPLA2 activity for 2 h does not seem to a ffect gonocyte di fferentiation. Since arachidonic acid is the precursor of prostaglandins, the present data sugges<sup>t</sup> that a short two-hour treatment with MJ33 may not deplete arachidonic acid su fficiently to compromise cell di fferentiation, or that this process involves other PLA2 enzymes expressed in gonocytes [19].
