The origin of reactive oxygen species in mouse embryos cultured in vitro

Embryos from most outbred and inbred mouse strains do not develop to blastocysts when cultured in a chemically defined medium but arrest at the 2-cell stage, a phenomenon referred to as ‘the two-cell block’ (Whittingham, 1974). We reported recently that the period of in vitro development at which this block occurs (the mid 2-cell stage to the early 4-cell stage) is associated with a rise in reactive oxygen species such as H₂O₂ or lipid peroxides (Nasr-Esfahani et al. 1990a). This rise only occurred after culture in vitro and regardless of the total time spent in vitro. It was not observed in embryos of the same age recovered directly from the female tract, nor was it observed in embryos at other preimplantation stages that had been cultured in vitro. We concluded that the rise in peroxide levels was a response to in vitro culture dependent on the stage of development. We suggested that the presence of elevated levels of reactive oxygen species under these conditions might be related in some way to the 2-cell block, a conclusion also reached recently by Noda et al. (1991).

Reactive oxygen species such as H₂O₂, in conjunction with superoxide (O₂⁰), can cause cell damage by promoting hydroxy radical formation via the iron-catalysed Haber-Weiss reaction (Halliwell, 1987; Halliwell and Gutteridge, 1987):

Net reaction: (3)

This reaction requires the presence of iron. Iron can also act directly on lipids to magnify peroxidative damage once this has been initiated by free hydroxy radicals (Minotti and Aust, 1989). It is therefore important to note that the addition of bovine apotransferrin and certain other iron chelators to the culture medium to sequester free traces of iron promotes development through the two-cell block (Nasr-Esfahani et al. 1990b), further supporting the idea that the rise in reactive oxygen species may be involved in the block.

In this paper we examine the origin of the rise in reactive oxygen species and its dependence on the first and second cell cycles. We use the technique described previously to measure the rise (Nasr-Esfahani et al. 1990a). The principle underlying this procedure may be described briefly as follows: 2’,7 ’-dichlorodihydrofluor-escein diacetate (DCHFDA), because of its nonionized state, is membrane permeant and therefore is able to diffuse readily into cells. Within the cell, the acetate groups are hydrolyzed by intracellular esterase activity forming 2 ’,7 ’-dichlorodihydrofluorescein (DCHF) which is polar and thus trapped within the cell. DCHF fluoresces when it is oxidised by H₂O₂ or lipid peroxides to yield 2 ’,7 ’-dichlorofluorescein (DCF). The level of DCF produced within the cells is related linearly to that of peroxides present and thus its fluorescent emission provides a measure of the peroxide levels.

MFI female mice (3 –4 weeks; OLAC) and F₁ (3 – 4 weeks; C57BL/10ScSn/OLa females × CBA/Ca/OLa male, bred in laboratory) were superovulated by intraperitoneal injection of 5 or 10 i.u. of pregnant mare’s serum gonadotrophin (PMS; Intervet) and human chorionic gonadotrophin (hCG; Intervet) 48 h apart. To obtain embryos, females were paired individually overnight with HC-CFLP males (Interfauna) and inspected for vaginal plugs the next day as an indication of successful mating.

Unfertilized oocytes were recovered from unmated females at about 13 h post-hCG. Fertilized eggs at the pronuclear stage were recovered from mated females at between 21 and 30 h post-hCG as indicated in the Results. Oocytes and embryos were released from the oviduct into warmed H6+BSA (a Hepes buffered form of modified T6 medium; Nasr-Esfahani et al. 1990b) and then cultured in drops of T6+BSA or BAT6+BSA (Nasr-Esfahani et al. 1990b) under paraffin oil (FSA Suppliers, Loughborough, UK) in Falcon tissue culture dishes in 5 % CO₂ in air. All manipulations were carried out at 37°C on heated stages, pads or in incubators.

In some experiments, oocytes were activated by exposure to 8% ethanol in H6+BSA for 8 – 10min (Kaufman, 1983), rinsed, placed in culture and inspected for activation 4 h later. Non-activated and deformed oocytes were removed, and the remaining activated oocytes were divided into two groups: (i) those that had undergone immediate cleavage to two cells and in which each cell had a single pronucleus – designated ‘immediate cleavage’, and (ii) those with a single pronucleus and a piolar body plus those containing two pronuclei and no second polar body – together designated ‘activated’.

1-cell fertilized oocytes were cultured under the conditions specified in the Results section and inspected at regular intervals as follows: at 47 – 50 h post-hCG the few 1-cell or abnormal eggs were removed and are not included in the totals, the remaining embryos being scored as 2-, 3- or 4-cells; at 69 – 72 h post-hCG embryos were scored as being dead, 2-cell, 3-cell, 4-cell, 5- to 8-cell precompact or compact; at 96 – 100 h post-hCG embryos were scored as being dead, noncompact, compact, early blastocyst or expanded blastocyst (see Chisholm, Johnson, Warren, Fleming and Pickering, 1985, for definitions of blastocyst subtypies); at 116 – 119 h post-hCG embryos were scored as being dead, preblastocyst, early blastocyst or expanded blastocyst.

Stock solutions of DCHFDA (Kodak, Eastman Kodak Company, Rochester, NY, USA) were prepared in acetone at 1 ×10⁻³ M. The stock solutions were diluted in H6+PVP to the required concentration. DCHFDA stock solutions were prepared just before the start of each expieriment and were kept in the dark and used over a maximum pieriod of 48 h. Oocytes, parthenogenotes and embryos were washed thoroughly in H6+PVP after their removal from H6, T6+BSA or the condition specified in the Results section, and then loaded with the dye in cavity blocks for 15 min. The oocytes or embryos were then washed in H6+BSA to remove traces of the dye and were placed in specially designed small chambers containing H6+BSA and covered by a coverslip (Maro et al. 1984).

The fluorescence emissions of the oocytes and embryos were measured immediately by photocytometry, using a Perspex carrying slide for viewing with a long working distance ×32 objective on a Leitz Ortholux II microscope with stabilized HBO100 mercury vapor lamp and filter set L1 for FITC. For quantitation of fluorescence, the photomultiplier housing of a Leitz MPV-1 was fitted to the Ortholux II phototube (McConnell et al. 1990). The phototube contained a variable measuring diaphragm that could be adjusted to surround the periphery of an individual oocyte or embryo, thus excluding background. A 6.25% transmission neutral density filter (Leitz N16) was placed in the path of the exciting light, to minimise any potential damage to the cells. Fluorescent emission was deflected to the amplifier/ discriminator (Model 1140B, SSR Instruments Co, USA) of a quantum photometer (Model 1140A, SSR Instruments) that had been zeroed against background and set to read in a ctss^-1 mode via a deflection meter. The individual oocyte or embryo was positioned within the adjusted diaphragm and exposed to the excitation wavelength for a period of less than 10s and the fluorescent emission recorded (ctss^-1 on the 1M scale to a maximum reading of 10). In some experiments, two neutral density filters were used so that readings higher than 10 could be measured. For each datum point in each experiment, the fluorescent emissions of 10 to 20 oocytes or embryos were measured and their mean values were expressed as the ‘mean reading’.

This set-up, involving short exposure to the exciting light, was designed to avoid damage to the oocytes or embryos, which was only detected with greater than a 40 s exposure to the exciting light or removal of the 6.25 % transmission neutral density filter. Under these latter conditions, a rapid rise in fluorescent emission, reflecting conversion of DCHF to DCF, was detected as a secondary consequence of light-induced lipid peroxidation. This latter property was used by us as a positive control to show that the uptake and deesterification of DCHFDA was not limiting the fluorescent signal that we measured under any given condition. Thus, sample embryos were exposed to a brief pulse of high intensity light and the emission value recorded (see Nasr-Esfahani et al. 1990a).

1-cell fertilized zygotes, treated with nocodazole or aphidicol-in, were sampled at various times, fixed in 4% formaldhyde for 30 min, rinsed in PBS, permeabilised in 0.2% Triton X-100 for 10 min, rinsed in PBS, exposed to ammonium chloride (26 mg ml^-1) for 10 min, rinsed in PBS and stained with Hoeschst 33342 at 0.1 μgml^-1 in PBS for 30min. The samples were rinsed in PBS, mounted in Citiflour (City University, London, UK) and examined on the Leitz Ortholux II microscope.

Embryos were cultured for 1 h in a dilution of 5 μl of [³⁵S]methionine (1000 – 1400 Ci mmol^-1, Amersham International Ltd) in 45 μl T6+BSA with or without drugs. Following the labelling period, groups of 25 embryos were washed thoroughly in H6+PVP and stored at – 70°C before separation of proteins on 10% SDS – polyacrylamide gel as described previously (Flach et al. 1982). Following electrophoresis, gels were treated with Amplify (Amersham International) and exposed to preflashed Fuji X-ray film at – 70°C.

Chemicals used in these experiments were obtained and stored as follows: α-amanitin (Boehringer-Mannheim, FRG) made up as an 11 mg ml^-1 stock in distilled water; Nocodazole (Aldrich) made up as a 10 MM stock solution in DSMO; Hoechst dye 33342 (Sigma) stored as a stock solution at 1 mg ml^-1 saline; anisomycin (gift from Pfizer) was stored as a stock of 10 mM in H6; aphidicolin (ICI) made up as a stock solution of 2mgml^-1 DMSO; phorbol myristate acetate (PMA) was made up as a stock of 25 μgml^-1 DMSO (Sigma); hydroethidine (Polysciences) was made up at 5DIM in N,N- dimethylformamide (BDH); hypoxanthine, allopurinol, oxypurinol, 6-amino nicotinamide, antimycin-A, rotenone, dinitrophenol and azide were obtained from Sigma. Diphenyl-eneiodonium (DPI) was a gift from Dr B. Meier, Chemisches Institut, Hannover. Cyanide and ethanol were obtained from BDH (England).

The rise in peroxides occurs between mid to late G₂ and early M phase of the 2-cell stage, which, in most experimental groups of embryos, covers the period 45 to 53 h post-hCG, although in more slowly developing groups it can be delayed by up to 5 h (discussed in Nasr-Esfahani et al. 1990a). To determine whether the rise at this time depended upon activation of oocytes or simply reflected the length of time since ovulation, two types of comparison were made. First, fertilised and unfertilised oocytes were cultured in vitro and their peroxide levels compared. Only fertilised oocytes showed the rise in H₂O₂, unfertilised oocytes showing declining levels over the same period (Fig. 1). Second, sperm-activated oocytes were compared with oocytes activated parthenogenetically by ethanol. Two such experiments (Fig. 2) revealed a rise in H₂O₂ levels regardless of activation mechanism, although in both cases the parthenotes showed a slightly earlier rise.

In order to determine whether the rise in H₂O₂ levels was dependent upon progress through the first two cell cycles, embryos were arrested by drugs at various points in their progress to the late 2-cell stage, and sampled at time intervals over the period 31 – 54 h post-hCG. To simplify the presentation of results (Fig. 3), sample H₂O₂ readings at 45 h post-hCG (pre-rise in the controls in the experiment shown) and 51 h post-hCG (peak rise) are given.

Embryos arrested at the G₁ – S boundary of the first cell cycle by aphidicolin (2 μgml^-1 added at 23 h post-hCG, activity confirmed by chromatin staining with Hoechst 33342; Howlett, 1986), nonetheless showed an H₂O₂ rise (Fig. 3, compare control values, bar 1, with aphidicolin values bar 2). Likewise, an H₂O₂ rise occurred after arrest in first mitosis (Fig. 3, bars 3 and 4; 10 μM nocodazole applied at 26h post-hCG; control for non-specific effects of nocodazole achieved by adding the drug at 36h post-hCG), arrest being confirmed by Hoechst 33342 staining and analysis of protein biosynthetic profile (Fig. 4, tracks a, d and g; Howlett, 1986).

α-amanitin (11 μgml^-1), applied at 26 h post-hCG to block activation of the embryonic genome (Bolton et al. 1984; confirmed by PAGE analysis of the protein biosynthetic profile; Fig. 4, compare tracks b and f with c and e), nonetheless showed the characteristic rise in H₂O₂ level (Fig. 3, bars 5 and 6). That the synthesis of embryo encoded proteins is not essential for the rise in H₂O₂ was confirmed by exposing newly formed and mid 2-cell embryos to 10 μM anisomycin to inhibit protein synthesis (Levy, Johnson, Goodall and Maro, 1986), after which the rise in H₂O₂ level occurred nonetheless (Fig. 3, bars 7 and 8).

H₂O₂ may be produced either via dismutation of superoxide by the enzyme superoxide dismutase (SOD) or by a direct enzymatic process such as urate oxidase or glucose oxidase. There are two types of SOD enzyme, the cytosolic form, which is known as CuZn-SOD, or the mitochondrial form, which is known as the Mn-SOD. Cyanide is an extremely powerful inhibitor of CuZn-SOD, while it has no effect on the activity of Mn-SOD (Halliwell and Gutteridge, 1989). One-cell zygotes were cultured in vitro to 44.5 h post-hCG and then exposed to T6+BSA+CN (5 MM) at 44.5 h post hCG after which their peroxide levels were measured (Fig. 5A). Although cyanide results in suppression of the H₂O₂ rise at 49.5 h post-hCG, over the 5h of incubation a significant number of embryos developed membrane blebbing and some died. Therefore, in a second experiment, embryos were cultured in vitro until 51 h post-hCG, exposed for 30 min to T6+BSA+CN (5 MM), assayed immediately and found to have suppressed peroxide levels (Fig. 5B). These results are consistent with the possibility, but not proof, that the H₂O₂ rise depends on cytoplasmic dismutation of superoxide.

To test this hypothesis further, the fluorochrome hydroethidine (HE; Kobzik et al. 1990a,b) was used. HE is the reduced, non-fluorescent precursor of ethidium and, when oxidised by either superoxide or H₂O₂, binds to DNA and can be detected by its red fluorescence. In a preliminary experiment, embryos of different ages, in hours post-hCG, were loaded with either DCFHDA or HE (166 μM) for 15 min, washed and assayed immediately for their fluorescence emission (Fig. 6). A similar rise in oxidation of both HE and DCFHDA, was observed. Rothe and Valte (1990) showed recently that oxidation of DCFHDA, but not of HE, was decreased in stimulated neutrophils, which had been loaded simultaneously with HE and DCFHDA as compared to when they had been loaded with either DCFHDA or HE alone. The HE was oxidised by superoxide before it could be converted to H₂O₂, and the lower H₂O₂ level in consequence reduced DCFHDA oxidation. Two-cell embryos (50h post-hCG), which had been cultured overnight, were loaded with DCFHDA at 10 μM, with HE at 250 μM, or with both fluorochromes (Fig. 7). Coloading reduces the signal from DCFHDA oxidation but not from HE oxidation, suggesting that the H₂O₂ rise may depend on dismutation of superoxide in the cytoplasm.

Superoxide can be produced via different pathways, of which the three major ones are: (i) the electron transport chain of mitochondria (Fig. 8), (ii) the NADPH oxidase system, and (iii) the xanthine/ xanthine oxidase system.

Mitochondrial inhibitors or uncouplers were used to analyse whether superoxides might have a mitochondrial source (Fig. 8). 1-cell zygotes were cultured in vitro to 46 h post-hCG, transferred to 0.001 or 0.0001 M azide (uncoupler at site III), 10 μM antimycin A (uncoupler at site II) or 40 μM rotenone (uncoupler at site I), and assayed for their peroxide levels at different time intervals. For simplicity only the results before and during the H₂O₂ rise (48 and 52 h post-hCG in the controls in this experiment) are recorded in Table 1. Although the H₂O₂ rise is inhibited during chronic exposure to rotenone and antimycin A, but not to azide, by 52 h post-hCG some of the embryos had died and most of the remaining embryos showed membrane blebbing. Therefore, in a second set of experiments, exposure to drugs was delayed until 50 h post-hCG and limited to 25 – 45 min total. Acute exposure to antimycin during the period of the H₂O₂ rise was without effect, but rotenone supressed the rise (control reading 5.44±0.4, 25 min exposure 2.56±0.4, 45 min 1.14±0.1), but even with these short exposure periods to rotenone, 40% of embryos showed some membrane blebbing. Finally, we examined whether respiratory inhibitors, such as antimycin A, which block electron flow on the oxygen side of b-type cytochromes, might promote a premature peroxide rise (Flohe et al. 1971). Early 2-cell embryos were exposed for 15 min to azide (0.001M), antimycin A (10 μu), rotenone (40 μM) or dinitrophenol (100 μM) before loading for 15 min with DCFHDA in the continuing presence of the inhibitor. No clear evidence for a premature rise was observed.

Diphenylene iodonium (DPI) is an inhibitor of NADPH-oxidase (Yea, Cross and Jones 1990). Twocell embryos (50 –51 h post-hCG) were incubated in DPI (10 μM) for 0, 7, 15 or 30min and then loaded with DCFHDA in the continuing presence of DPI for 15 min to measure the peroxide level (Fig. 9A). Significant reductions were observed, although after 30 min or more in DPI membrane blebbing occurred. In a second experiment, embryos were incubated in different doses of DPI for 30 min, loaded with DCFHDA in presence of DPI for 15 min (Fig. 9B). A concentration-dependent reduction of H₂O₂ was observed, although embryos in 1 and 10 μM DPI showed some membrane blebbing. More prolonged culture in DPI resulted in death of the embryos.

Glucose is required for NADPH production via the pentose phosphate pathway (PPP). One-cell zygotes were recovered in the absence of glucose, and some were then transferred to medium containing glucose (5 MM) for culture overnight. The two populations of 2-cell embryos were assessed at different time points the next day for their peroxide levels. No significant difference was observed in the rise in peroxide levels in the absence or presence of 5mM glucose.

The NADPH-oxidase activity may be decreased by reducing the activity of the PPP, the second enzyme in this pathway being 6-phosphogluconate dehydrogenase (6PGD), which is inhibited by 6-amino-nicotinamide (AMN). One-cell fertilized embryos were cultured overnight in glucose-free medium containing 1000 μM AMN to limit the availability of NADPH, but no reduction in the peroxide rise was observed (data not shown). In addition, 1-cell zygotes were recovered in the absences of glucose and then were cultured with or without glucose in various doses of ANM until 69 h post-hCG, when they were rinsed and transferred to control medium until 122h post-hCG. They were scored morphologically at 46, 69, 100 and 122 h post-hCG. Only 9 % of the embryos at 10 μM ANM reached the 4-cell stage by 69 h post-hCG, and none of these reached the blastocyst stage, most dying by 76 h post-hCG (data not shown).

NADPH-oxidase is activated by phorbol myristate acetate (PMA), lectins and carcinogens. Embryos (45 h post-hCG and thus pre-H₂O₂ rise) were therefore cultured for 15 min in either control medium or DCFHDA+PMA (2.5ngml^-1), but no significant difference in peroxide profiles was observed, even though the PMA was active, as assessed by its effects on the morphology of the blastomeres (Bloom, 1989).

Recently, purines such as hypoxanthine have been shown to be toxic to preimplantation development in vitro (Loutradis et al. 1987; Nureddin et al. 1990). Hypoxanthine could exert its toxic effect by production of superoxides during degradation to xanthine and then to uric acid by xanthine oxidase. Allopurinol and oxypurinol are known inhibitors of xanthine oxidase (Spector et al. 1989; Spector, 1988). Freshly recovered 1-cell zygotes or 2-cell embryos, developed after overnight culture, were exposed for various periods to allopurinol or oxypurinol before the peroxide levels of these embryos were assessed with or without the continuing presence of the inhibitors (Fig. 10). Both allopurinol and oxypurinol are able to inhibit the peroxide rise without having any immediate obvious adverse effects on the embryos, and the effects are rapid and reversible.

In further experiments, 1-cell embryos were cultured in medium containing hypoxanthine and on the following day their peroxide level was assayed in the presence or absence of oxypurinol (Fig. 11). The hypoxanthine-treated group produced higher levels of H₂O₂ compared with the control group and this rise is inhibited in the presence of oxypurinol.

1-cell MFI embryos were cultured in allopurinol or oxypurinol at various doses to 70 h post-hCG, by which time 40 – 60% had developed to the 4-cell stage, and they were then transferred to T6 medium, in which they progressed no further. If incubation in allopurinol or oxypurinol was restricted to shorter periods, encompassing only the period of the peroxide rise, no improvement in blastocyst rate was seen. MF1 2-cell embryos placed in allopurinol or oxypurinol at 48 – 52 h post-hCG developed to blastocysts whether removed from the drugs or not.

The rise in reactive oxygen species observed after culture of mouse embryos in vitro coincides with the period over which the developmental block occurs. This period is also marked by a number of important developmental events, notably the activation of transcription by the embryonic genome and the inactivation or destruction of much of the preexisting maternal mRNA, both of which are dependent on progress through the first two cell cycles. (Flach et al. 1982; Bolton et al. 1984; reviewed by Telford et al. 1990). In this paper, we demonstrate that the rise in reactive oxygen species is dependent on the activation of the oocyte (Fig. 1 and 2). The observation that partheno-genetically activated oocytes show a slightly earlier rise in peroxides may reflect their slightly accelerated passage through to the 4-cell stage (Kaufman, 1983). However, although the rise in reactive oxygen species occurs characteristically over the end of the second cell cycle, it does not depend on passage through the first and second DNA synthetic phases of development, the first mitotic division or G₁ of the second cell cycle (Fig. 3). Neither is it dependent upon the transcription of the embryonic genes or the translation of their mRNAs (Fig. 3). Thus, the rise in reactive oxygen species appears to involve the setting of a timing mechanism within the oocyte itself at or shortly after activation, which then can interact with in vitro culture conditions to produce the observed and appropriately timed rise in peroxides independently of continuing progress through the cell cycle. Presumably the timing mechanism involves some sort of metabolic maturation or cascade.

Although the rise in reactive oxygen species requires the activation of the oocyte, it is not dependent upon a contribution from the spermatozoon. Interestingly, it is known that the genotype of the oocyte alone, regardless of the fertilising spermatozoon, determines whether embryos will block at the 2-cell stage when cultured in vitro (Goddard and Pratt, 1983). Thus, both the peroxide rise and the in vitro block share a common underlying feature. Might then the two be related? In principle, a rise in reactive oxygen species might be part of the mechanism by which the block is caused or might simply reflect underlying metabolic problems, themselves responsible for establishing the block. Alternatively, the rise in reactive oxygen species and the block might merely be coincidental but independent events. Indeed, it is possible under certain circumstances to separate the in vitro block from the rise in H2O2. Thus, (i) embryos from different strains of mice show a rise in H₂O₂ production in vitro whether or not their embryos also block in vitro, and (ii) conditions of improved in vitro culture, such as the addition of iron chelating agents (Nasr-Esfahani et al. 1990a), may allow otherwise blocking embryos to develop to blastocysts more effectively, but appear nonetheless to be associated with a rise in reactive oxygen species at the late 2-cell stage (Nasr-Esfahani et al. 1990b). These observations do not prove that the rise in peroxides is unrelated to the block, since arrest of embryonic development by peroxidative damage to cells may require further interactions, for example with superoxide to produce reactive oxygen species such as hydroxy radicals. It is possible that the properties of non-blocking embryos, as well as those of culture conditions that can to some extent overcome the block in vitro, include the inhibition of superoxide production, the inhibition of conversion of superoxide and H₂O₂ to hydroxy radicals and/or the presence of scavenging activity to prevent damage by free radicals generated. Indeed, as we pointed out previously (Nasr-Esfahani et al. 1990b), the positive effects of iron chelators on in vitro culture could result from the protection that they offer against both free radical generation and the extension of free radical peroxidative damage to lipids by free iron. Recent evidence suggests that there are strain differences in the sensitivity of cardiac lipids to peroxidation, Fi strains (which characteristically produce nonblocking embryos) being more resistant to peroxidative damage than are those strains that generate blocking embryos (Baird and Hough, 1990). Clearly, it will be important to measure the dynamics of production of superoxide, hydroxy radical and peroxidised lipid in blocking and non-blocking embryos under different culture conditions.

In this context, it was important to try to identify the likely origin of the reactive oxygen species. The finding that the peroxide rise was sensitive to cyanide suggested that H₂O₂ might be produced via cytoplasmic dismutation of superoxide, a conclusion supported by use of hydroethidine, which reacts with superoxide to deplete its availability for forming H₂O₂, so reducing the level of DCF formation by embryos.

Superoxide can be produced via electron leakage from the electron transport chain of the mitochondria and by various cytoplasmic systems. It is difficult to investigate the origin of superoxides in embryos by conventional biochemical assays, because of the quantitative limitations of the material, and so use of intact cells with indicator dyes is necessary. Nonetheless, we were able to draw some tentative conclusions. First, we could not provide clear evidence to favour a mitochondrial origin for superoxides. Thus, although acute and chronic exposure to the mitochondrial poison rotenone, and chronic exposure to antimycin A, prevented the rise in peroxide levels, acute exposure to antimycin A - and chronic exposure to azide had no effect on the H₂O₂ level. The depressant effect of these poisons seemed to be related more to their toxic effects than to a specific effect on electron leakage. This conclusion is supported by the fact that respiratory inhibitors blocking electron flow on the substrate site of the b-cytochrome (site II in Fig. 8) might be expected to suppress H₂O₂ formation, while antimycin A, blocking electron flow on the oxygen side of b-type cytochromes between the complex II and III, is expected to enhance peroxide production (Flohe et al. 1977). Yet a premature rise in H₂O₂ was not obtained after exposure to antimycin A, azide or cyanide. Thus, it seems unlikely that the peroxide rise is due to electron leakage or superoxide production at sites I and II of the electron transport chain. It is also known that oxidative phosphorylation activity via the Kreb’s TCA cycle is very low at pre-8-cell stages of mouse development.

In contrast, the pentose phosphate pathway (PPP), which produces NADPH, shows relatively high activity during the 2-cell stage (O’Fallon and Wright, 1986). NADPH-oxidase is a membrane enzyme system, well defined in neutrophils and known as the oxygen burst system (Lambeth, 1988). It consists of a flavo-haem-protein, which oxidizes NADPH in the cytosol into NADP⁺. The electron produced by oxidation of NADPH is used to reduce oxygen to the superoxide radical. NADPH is known to be involved in H₂O₂ production via NADPH-oxidase in sea urchin oocytes at fertilization (Heinecke and Shapiro, 1989). Diphenylene iodonium (DPI), an analogue of NADPH, is known to inhibit NADPH-oxidase activity in neutrophils during the oxygen burst. Low concentrations of DPI (0.001 μM) inhibit the peroxide rise in mouse 2-cell embryos. However, brief exposure to DPI proved to be quite toxic to mouse embryos, which raises concern about the specificity and mechanism of its effects. Glucose, which is required for NADPH production via the PPP, has been reported to retard the development of embryos through the 2-cell stage in vitro (Chatot et al. 1989). However, omission of glucose from the medium did not prevent the H₂O₂ rise. Were NADPH-oxidase to be the source of peroxide production, removal of glucose on its own might not reduce the H₂O₂ level since the embryos are likely to obtain glucose from their glycogen stores. Therefore, in order to reduce the intracellular formation of NADPH, embryos were cultured free of glucose in 6-amino-nicotinamide, which partially inhibits the production of NADPH (White et al. 1988). These conditions neither reduced the peroxide level nor improved the embryo culture. Finally, NADPH-oxidase activity in neutrophils and other biological systems is stimulated with phorbol myristate acetate (PMA). However, PM A did not trigger a premature H₂O₂ rise in the early two-cell embryos, although evidently capable of affecting embryos at this stage (Bloom, 1989). This evidence, taken together, suggests that NADPH-oxidase activity is unlikely to be involved in causing the peroxide to rise. However, the situation is complex since, NADPH production via the PPP is also essential for the conversion of H₂O₂ to H₂O by the glutathione anti-oxidant system, and so its neutralisation might elevate peroxide levels and enhance developmental damage caused by H₂O₂ from other sources.

Purines such as hypoxanthine have been shown to retard development through the second cell cycle of embryos from most strains of mice and to do so by a mechanism that does not involve inhibition of phosphodiesterase (Nureddin et al. 1990). Superoxide is a byproduct of the degradation of purines to uric acid. Indeed, embryos cultured in medium supplemented with hypoxanthine produced high and premature levels of H₂O₂ compared with control embryos. Moreover, both the natural and hypoxanthine-induced peroxide elevations were prevented by the xanthine oxidase inhibitors allopurinol and oxypurinol. This inhibition was reversible upon the removal of the inhibitor. The fact that 1-cell embryos cultured in medium supplemented with allopurinol and oxypurinol showed no improved rates of blastocyst formation compared with controls may be due to other adverse effects of these inihibitors (Miyazaki et al. 1989; Miyazaki et al. 1991). Both allopurinol and oxypurinol are also known to have antioxidant properties (Moorhouse et al. 1987), but their ability to prevent the peroxide rise is more likely to be due to the inhibition of xanthine oxidase since the hypoxanthine-induced premature rise is inhibited by oxypurinol whereas UV-induced lipid peroxidation damage is not (data not shown). Therefore, it seems possible that the superoxide precursor of H₂O₂ in early mouse embryos may be produced via the xanthine oxidase system. A possible explanation for the rise in superoxide output at the 2-cell stage is the accumulation of substrates for xanthine oxidase as a result of the breakdown of maternal mRNA occurring at about this time in activated oocytes (see Telford, et al 1990 for review). What is not clear is why peroxides produced by this system should accumulate after in vitro culture but not in vivo. It also remains to be determined whether the accumulation is due to more active formation of reactive oxygen species or to their less effective removal.

Finally, it is important to stress that we have not proved that xanthine oxidase is the only system generating peroxides in vitro, and their accumulation from a number of sources remains a possibility. Thus, the recovery and culture of 1-cell zygotes from the oviduct involves their exposure to light, high oxygen levels, traces of transitional elements and disturbed concentrations of metabolic substrates. Exposure to visible light has been shown to be deleterious to development (Daniel, 1964; Hirao and Yanagimachi, 1978; Hegele-Hartung et al. 1988; Fischer et al. 1988), possibly via an effect on photosensitizer pigments, such as the mitochondrial flavins, NADPH-oxidase or blood pigments carried over into culture’ on BSA, so generating singlet oxygen species which then lead to superoxides, lipid peroxides and H₂O₂. The elevated oxygen in vitro will promote these reactions as will traces of free iron, to which embryos are unlikely to be exposed in vivo because of the presence of iron-binding transferrin in the genital tract secretions (Nasr-Esfahani et al. 1990b). It is possible that the accumulated insults of mRNA turnover, light, high oxygen tension and trace metals push endogenous peroxide production to higher levels in vitro than in vivo and that some strains of embryo are protected less adequately from their effects by scavengers and endogenous reducing agents than are others.

We wish to thank John Aitken for valuable advice and Martin George and Brendan Doe for technical assistance. This work was supported by a grant from the Medical Research Council to MHJ and Dr PR Braude.