|Home | About | Journals | Submit | Contact Us | Français|
In sea urchin embryos, specification of the secondary (oral-aboral) axis occurs via nodal, expression of which is entirely zygotic and localized to prospective oral ectoderm at blastula stage. The initial source of this spatial anisotropy is not known. Previous studies have shown that oral-aboral (OA) polarity correlates with a mitochondrial gradient, and that nodal activity is dependent both on mitochondrial respiration and p38 stress activated protein kinase. Here we show that the spatial pattern of nodal activity also correlates with the mitochondrial gradient, and that the latter correlates with inhomogeneous levels of intracellular reactive oxygen species. To test whether mitochondrial H2O2 functions as a redox signal to activate nodal, zygotes were injected with mRNA encoding either mitochondrially-targeted catalase, which quenches mitochondrial H2O2 and down-regulates p38, or superoxide dismutase, which augments mitochondrial H2O2 and up-regulates p38. Whereas the former treatment inhibits the initial activation of nodal and entrains OA polarity toward aboral when confined to half of the embryo via 2-cell stage blastomere injections, the latter does not produce the opposite effects. We conclude that mitochondrial H2O2 is rate-limiting for the initial activation of nodal, but that additional rate-limiting factors, likely also involving mitochondria, contribute to the asymmetry in nodal expression.
Axis specification is the symmetry breaking process that initiates pattern formation in a developing embryo. Axial polarities can be specified either cell-autonomously, by spatial localization and subsequent segregation of molecular determinants among dividing cells (‘preformation’), or conditionally, by way of self-organizing processes that depend on stochastic anisotropies and intercellular signaling (‘epigenesis’). Although axis specification in animals involves varying combinations of preformation and epigenesis, many deuterostome embryos are spectacularly regulative, a hallmark of conditional epigenetic specification. A protein that plays a recurring role in deuterostome axis specification is Nodal, a member of the TGFβ family of extracellular signaling ligands. Nodal locally activates both its own expression as well as that of its more diffusible extracellular antagonist Lefty, thus embodying an epigenetic system of short-range activation and long-range inhibition (Juan and Hamada, 2001; Chen and Schier, 2002; Solnica-Krezel, 2003; Duboc et al., 2008). Such systems provide for the self-organization of defined spatial patterns from subtle anisotropies that are present or which arise stochastically within relatively homogeneous precursors (Turing, 1952; Gierer and Meinhardt, 1972).
In the bilaterally symmetric sea urchin embryo, cell differentiation occurs along two orthogonal axes: the maternally determined animal-vegetal axis, along which respectively differentiate ectoderm, endoderm, and mesoderm; and the conditionally specified oral-aboral (OA) axis, along which differentiate oral ectoderm, ciliated band ectoderm, and aboral ectoderm, as well as various mesodermal cell types. Specification of oral ectoderm is initiated by the zygotic activation of nodal (Duboc et al., 2004). Following its initial activation, nodal expression is amplified by way of a positive feedback ‘community effect’ involving Nodal signaling, which also activates lefty, whose expression is required to confine nodal activity to prospective oral ectoderm in the early to mid-blastula stage embryo (Duboc et al., 2004; Nam et al., 2007; Range et al., 2007). Although the spatial information underlying the initial anisotropy in nodal expression has not been fully elucidated, a strong candidate is an asymmetric distribution of mitochondria that prefigures the prospective OA axis, which could regulate the activities of maternal transcription factors required for nodal activity (Coffman and Davidson, 2001; Coffman et al., 2004). This regulation likely involves p38, a stress-activated protein kinase that responds to reactive oxygen species (Torres and Forman, 2003), and whose inhibition has been shown to suppress nodal expression and hence development of oral ectoderm (Bradham and McClay, 2006).
Recent cis-regulatory analyses of the nodal genes from Strongylocentrotus purpuratus and Paracentrotus lividus identified two conserved modules that account for the spatiotemporal pattern of nodal expression in the early embryo (Nam et al., 2007; Range et al., 2007). A conserved sequence module located upstream (5′) of the transcription start site directs initial activation and contributes to auto-activation. A second conserved module within the single intron contains additional sequences involved in auto-activation as well as sites required for spatial repression outside of the oral territory. Sequences within the 5′ cis-regulatory module include consensus target sites for b-Zip transcription factors and other proteins that are known to be regulated by redox state and/or p38 (Nam et al., 2007; Range et al., 2007).
The purpose of this study was to test the hypothesis that H2O2 is the signal through which mitochondria influence the activity of nodal. Toward that end we used confocal fluorescence microscopy and specific molecular probes to examine the spatial correlations between nodal activity, mitochondria and reactive oxygen species (ROS) in the early embryo. We confirmed that there is a significant spatial correlation between nodal activity and mitochondrial distribution, and that the latter also correlates with intracellular levels of ROS, most likely in the form of H2O2. To determine whether this correlation is causal we perturbed mitochondrial H2O2 emissions by overexpressing mitochondrially-targeted catalase (Mt-Cat) and superoxide dismutase 2 (SOD2). Although quenching mitochondrial H2O2 with Mt-Cat down-regulates p38 and inhibits initial activation nodal, augmenting mitochondrial H2O2 with SOD2 activates p38 but not nodal. Moreover, whereas the former treatment entrains OA polarity when localized by 2-cell stage blastomere injections, the latter does not, suggesting that mitochondrial redox signaling via H2O2 is only one of multiple rate-limiting factors that contribute spatial information to the initial activation of nodal.
Strongylocentrotus purpuratus were obtained from Santa Barbara Marine Biologicals (Charles Hollahan, Santa Barbara, CA) or from the Point Loma Marine Invertebrate Lab (Pat Leahy, Coronal del Mar, CA). Gametes were released by vigorous shaking of adult sea urchins. Egg fertilization and embryo culture were carried out in artificial seawater (ASW) using standard methods (Foltz et al., 2004).
Microinjections of zygotes affixed to protamine-sulfate coated dishes were performed using standard methods of timed pressure injection (Cheers and Ettensohn, 2004). For RNA injections, 50–100 ng/μl were used for GFP-OMP25, and 400–1000 ng/μl were used for SOD2 and Mt-Cat. For DNA injections, a nodal-5P-GFP PCR amplicon (Nam et al., 2007; Fig. 1A) was injected at 0.5 ng/μl, in a solution containing 20 ng/μl restriction-digested sea urchin DNA as carrier. This amount of nodal-5P-GFP was determined empirically to be the minimum required to give GFP fluorescence. All injection solutions contained 120 mM KCl. Blastomere injections included 2 mg/ml 10,000 MW Dextran AlexaFluor 647 (Invitrogen Molecular Probes).
Confocal imaging was performed using a Zeiss LSM 510 microscope. For live imaging, embryos were affixed to protamine sulfate-coated glass-bottom MatTek dishes (MatTek Corp., Ashland, MA, USA), injected with RNA or DNA as described above, and imaged on a cooled stage maintained at 12° C. In some cases eggs or embryos were stained with 200 nM MitoTracker™ Deep Red or MitoTracker™ Orange (Invitrogen Molecular Probes) for 20 minutes in the dark. For detection of ROS, embryos were stained in 10 μM CM-H2DCFDA (Invitrogen Molecular Probes) for 20–60 minutes in the dark, and exposed to a minimum amount of fluorescence excitation before being imaged in order to minimize photo-oxidation of the dye.
For antibody staining, MitoTracker™ Deep Red-labeled embryos were fixed with 4% para-formaldehyde in ASW with 10mM EPPS (pH 8.0) supplemented with Roche PhosStop reagent (PS), for 10 min at room temperature (RT) with gentle agitation in the dark. Embryos were then washed twice in ASW +PS, three times in ice cold methanol (MeOH), and stored in MeOH at −20° C. Fixed embryos were washed four times in PBS + 0.1%Tween−20+PS (PBST-PS), blocked for 30 minutes at 4° C in 4% normal goat serum (NGS) in PBST-PS, and stained with rabbit anti-phospho-Smad3 (Rockland #600-401-919) at 1:100 dilution in PBST-PS overnight at 4° C. The embryos were washed four times in PBST-PS, incubated in Alexa-Fluor 488 goat anti-rabbit IgG (Invitrogen #A11008 20ug/mL in PBST-PS) for 3 hours in the dark at RT, and washed again (4×) in PBST-PS. The stained embryos were then transferred to 50% glycerol/PBST-PS and mounted for imaging with DAPI (1 μg/ml).
For quantitation, images were analyzed using ImageJ (NIH, v. 1.38). Using the tools provided in ImageJ, boundaries were drawn around the embryo images, and the average pixel intensity within the boundaries calculated. To obtain the average MitoTracker pixel intensity for half an embryo, the boundary on one half of the embryo was manually moved to form a line down the center of the embryo (thereby bisecting the image, verified by calculation of the resulting area), and the average pixel intensity within the half-embryo thus bounded was calculated and compared to the average pixel intensity for the whole.
Late gastrula stage embryos containing fluorescent lineage tracer were scored by placing the embryos at the time of hatching in an agarose tunnel (Ransick and Davidson, 1995). The embryos were viewed with epifluorescence on a Zeiss Axiovert 200 inverted microscope, scored, and digitally imaged using a Zeiss MRc Axiocam. In general, the lineage tracer exhibited the typical ‘oral-lateral’ and ‘aboral-lateral’ patterns described by Cameron et al. (1989). Embryos exhibiting the oral-lateral pattern of lineage tracer were scored as ‘oral’; embryos exhibiting the aboral-lateral pattern were scored as ‘aboral’. Ambiguous cases that could not be assigned to either the oral or aboral categories were scored as ‘lateral’ or ‘defective’, depending on the phenotype of the embryo.
A nodal-GFP BAC knock-in construct was used as PCR template to amplify linear nodal-5P-GFP DNA as described previously (Nam et al., 2007). The GFP-OMP25 construct consisted of the C-terminal 38 amino acid codons derived from Mus musculusOmp25 (Synj2bp; Genbank accession NM_025292) cloned in-frame into the C-terminal poly-linker site of pEGFP-C1. This region consists of a mitochondrial targeting signal that targets and anchors OMP25 to the mitochondrial outer membrane (Nemoto and De Camilli, 1999). This plasmid was used as a PCR template to amplify a T7 promoter-containing transcription template using the following primers: TAATACGACTCACTATAGGGTTTAGTGAACCGTCAGATCCGCTA (forward) and CCCTTTGACGTTGGAGTCCACGTTCT (reverse). The eGFP-coding mRNA synthesized from this template covers a region from 43 bases upstream of the eGFP start to downstream of the SV40 polyA signal, a total of 1384 nucleotides. Mt-Catalase was constructed by cloning the OMP25 C-terminal sequence from the above construct into the pGloBlu plasmid, a derivative of pBluescript (Stratagene) containing the Xenopus β-globin 3′ UTR and polyadenylation site (Coffman et al., 2004). The OMP25 C-terminal targeting sequence was inserted between the HindIII and XbaI sites of pGloBlu, as a HindIII-XbaI digest of the 114bp PCR amplicon generated from the parent vector with the following primer set: GCCAAGCTTGCTTGCTCATCGAGGTGAAGGAGAGCCAAGTGGAGT (forward) and TCTCTAGATCAAAGCTGCTTTCGGTATCTCACGAA (reverse). Adjustments were included in these primers so that the coding sequence of sea urchin catalase (Genbank Acc. No. CX561177) could be added in-frame relative to the signal sequence as a XhoI and HindIII digest of the 1610 bp PCR amplicon generated from the library clone with the following primers: ACCTCGAGGACGTGGTACTGCTGTTCTCTTTCCCAT (forward) and AGTCGTCCGATTGGACCTTACAGGTT (reverse). The final construct contains 28 bp of 5′UTR, and the catalase coding sequence with the C-terminal 14 codons replaced with 37 codons of OMP25 targeting sequence and ending with an introduced stop codon. This construct was linearized with SacI and used as template to synthesize mRNA using the Ambion T7 mMessage Machine kit. For SOD2 expression, the full-length coding sequence of SOD2 was initially amplified from total RNA extracted from adult testis tissue with Invitrogen one-step PCR kit and the following primers: AGGTACCGCACTTCAAATAACATAATGGCGTCTGCT (forward) and TCTCGAGGAACCTCATAATCCCCAGACCACCA (reverse). The resulting amplicon was inserted into pGloBlu-OMP25 as a KpnI-XhoI fragment maintaining a continuous open reading frame relative to the OMP25 targeting sequence. A second SOD2 expression construct lacking the OMP25 targeting sequence (i.e., wild-type SOD2, which localizes to the mitochondrial matrix) was subsequently inserted in pGloBlu following amplification of the full-length SOD2 coding sequence with the following primers: ACCTCGAGGCAGTGCAGTTACTGTGTATGTCTTTGCA (forward) and CTTCTAGAGCGCATTTCACTTGTCCTCATCTACAA (reverse). Both constructs were linearized with Pst1 and used as template to synthesize mRNA using the T7 mMessage Machine kit from Ambion. Since initial control experiments indicated that the two constructs were similarly effective in augmenting both ROS levels and p38 activity, most of the results reported below were obtained using the unmodified variant.
Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was performed on whole cell extracts from approximately 600 embryos per lane was using the NuPage Bis-Tris PAGE system (Invitrogen). Following transfer to nitrocellulose, immunoblot analysis was performed using the WesternBreeze immunodetection kit (Invitrogen), with rabbit anti-phospho-p38 primary antibody (Cell Signaling Technology) diluted 1:1000. The signal obtained with an anti- trimethyl-K4-histone H3 antibody (H3K4me3; Upstate) diluted 1:20,000 was used for a loading control. The immunoblots were imaged and the signals quantified using a Kodak GelLogic100 Imaging system. For quantification, the background-subtracted band intensities of phospho-p38 were normalized to those of H3K4me3.
RT-PCR was carried out on a Cepheid SmartCyclerII as follows: starting with total RNA isolated from approximately 600 treated embryos (Qiagen RNeasy), cDNA was prepared from typically 250 ng of total RNA using random priming in the SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen). PCR reactions were performed starting with 2 microliters of cDNA in a 25 microliter reaction set up as prescribed for PerfeCta SYBR Green FastMix (Quanta Biosciences). Cycling conditions were as follows: an initial activation step at 95° C for 150 seconds followed by cycles of 95° C for 15 seconds, 58° C for 25 seconds and 72° C for 45 seconds until 9 cycles after Ct. The Ct was reached when the primary curve crossed a manual threshold setting of 30 fluorescence units. Melt curves were inspected following the cycling protocol and all products were analyzed on agarose gels to confirm product specificity. Nodal transcript abundance relative to a given control sample was calculated by the formula 1.9ΔCt, where ΔCt is the difference in Cts obtained for nodal in the treatment and control after normalization to the ΔCts obtained for either ubiquitin or HPRT, which are assumed to be constant between treatments and/or time points. The sequences of primers used to amplify nodal and ubiquitin cDNA have been described (Coffman et al., 2004); the primers used to amplify HPRT cDNA are as follows: ACACATTCTGCGTCCCGAGGCAT (forward) and GGTCGGAGCAGAACTTGTAGCCTCCTT (reverse).
The prospective oral side of the embryo is typically enriched with mitochondria (Czihak, 1963; Coffman et al., 2004), and oral-aboral axis specification is sensitive to treatments that perturb mitochondrial activity or distribution (Coffman and Davidson, 2001; Coffman et al., 2004). Thus, mitochondrial distribution should correlate with the spatial pattern of nodal activity in the early embryo. To confirm that this is the case, we asked whether a spatial correlation exists between mitochondria and expression of a GFP reporter transgene driven by the nodal promoter and 5′ cis-regulatory module (nodal-5P-GFP; Nam et al., 2007; Fig. 1A, B). As noted above, this cis-regulatory module mediates both initiation and auto-activation of nodal expression, and produces an oral ectoderm-specific pattern of expression at gastrula stage (Nam et al., 2007). Because of the mosaic incorporation of the transgene, GFP fluorescence was generally confined to half (or less) of each embryo expressing nodal-5P-GFP. In a modest but significant majority of such embryos (64/100; P=0.005), the average MitoTracker signal intensity on the GFP-expressing half was found to be higher than the average for the whole embryo (Fig. 1B, C).
It is possible that the relative weakness of the spatial correlation between nodal-5P-GFP expression and mitochondrial density results from irregularities associated with microinjection and/or expression of the exogenous transgene reporter. In addition, nodal appears ab initio to be expressed broadly in a gradient (Duboc et al., 2004; Flowers et al., 2004; Range et al., 2007), and given the long half-life of GFP, this would tend to randomize the mosaic expression pattern of the reporter at early blastula stage. In light of these possibilities we used phospho-Smad2/3 immunofluorescence staining (Yaguchi et al., 2007) to further examine the correlation between nodal activity and mitochondrial distribution. As shown previously (Yaguchi et al., 2007), phospho-Smad2/3 signal was found to be confined to nuclei on one side of early blastula stage (13 hr) embryos (Fig. 1D). In a significant majority of embryos (74/102; P=5×10−6) the side containing nuclear phospho-Smad2/3 displayed a higher than average MitoTracker intensity compared to the embryo as a whole (Fig. 1D, E). These data further indicate that mitochondrial density and nodal activity are positively correlated in the early embryo.
One potential mechanism through which mitochondria could influence nodal activity is via the production of reactive oxygen species (ROS) such as H2O2, which are known to activate p38, a stress-activated protein kinase required for nodal expression (Bradham and McClay, 2006). To determine whether mitochondrial density correlates with ROS levels, embryos were co-stained with MitoTracker and CM-H2DCFDA (DCF), a fluorescent ROS indicator. Images obtained by confocal fluorescence microscopy of the labeled embryos revealed a significant spatial correlation between DCF and MitoTracker signal intensities (Fig. 1F, G), as underscored by the occasional instances of a single mitochondria-rich blastomeres that invariably displayed elevated DCF signal compared to their mitochondria-poor neighbors (Fig. 1F, arrow). On average, the DCF signal intensity differed by a factor of ~10% between mitochondria-rich and mitochondria-poor halves of late cleavage stage embryos.
If mitochondrial ROS contribute to activation of p38 and/or nodal, we reasoned that we should be able to perturb OA axis specification by targeting enzymatic antioxidants to mitochondria. Toward this end we made use of the C-terminal targeting sequence of the mammalian outer mitochondrial membrane protein OMP25 (Nemoto and De Camilli, 1999; Horie et al., 2002). To determine whether this sequence would target proteins to mitochondria in sea urchin embryos, we injected mRNA encoding a GFP-OMP25 fusion protein into zygotes and used epifluorescence microscopy to observe GFP distribution at blastula stage. Unlike untargeted GFP, which is diffusely distributed throughout the embryo, GFP-OMP25 has a punctate, perinuclear distribution, typically with a higher density on one side of the embryo, consistent with a mitochondrial localization (Fig. 2A). To confirm that the GFP-OMP25 is localized to mitochondria, GFP-OMP25-injected embryos were stained with MitoTracker and imaged confocally (Fig. 2B). The two fluorescent signals display considerable spatial correlation (Fig. 2C), the GFP co-localizing with (or closely apposing) a subset of the MitoTracker-labeled structures (e.g. Fig. 2B, arrows). This suggests either that OMP25 partitions to a specific subregion within the mitochondrial compartment (note that OMP25 localizes to the outer mitochondrial membrane, whereas MitoTracker accumulates in the matrix), or that it targets to a specific subpopulation of mitochondria. In either case, the data indicate that the C-terminal sequence of OMP25 effectively targets GFP to mitochondria in sea urchin embryos.
Hydrogen peroxide (H2O2) is the major mitochondrial ROS known to function as a signaling intermediate and to activate p38 (Robinson et al., 1999). To investigate the function of mitochondrial H2O2, we engineered a construct (Mt-Cat) that fuses the OMP25 C-terminal targeting sequence to catalase, an enzyme normally associated with peroxisomes that reduces H2O2 to H2O and molecular oxygen. We also made use of SOD2, a mitochondrial enzyme that converts superoxide to H2O2, thereby increasing mitochondrial production of the latter (Connor et al., 2005). To test the efficacy of these mitochondrially-targeted enzymes in modulating H2O2 levels, we injected single blastomeres at the 2-cell stage with mRNA encoding either Mt-Cat or SOD2 together with a far-red fluorescent dextran lineage tracer and allowed the embryos to develop to late cleavage (7 hr) stage, at which point the embryos were stained with DCF and imaged by confocal fluorescence microscopy. In a majority (14/18 = 78%; χ2 = 5.56, P =.018) of Mt-Cat mRNA-injected embryos, the injected half of the embryo (marked by the lineage tracer) displayed a DCF signal that was lower than that of the uninjected half (Fig. 3A). Conversely, a proportionate majority (13/17 = 76%; χ2 = 4.76, P=.029) of embryo halves injected with SOD2 mRNA displayed elevated DCF fluorescence compared to their uninjected counterparts (Fig. 3A). Similar results were obtained at blastula stage (data not shown). Quantification of pixel intensities showed that each perturbation produced an effect of similar average magnitude, corresponding at 7 hrs to an 8% (+/− 3% SEM) difference in DCF fluorescence between injected and uninjected halves (Fig. 3B). This differential in DCF signal intensity is comparable to that observed along the secondary axis of unperturbed 7 hr embryos (see above). The fact that the perturbations are less than 100% effective suggests that they do not always override the endogenous mitochondrial gradient. Nevertheless, the results strongly suggest that over-expression of Mt-Cat quenches, whereas over-expression of SOD2 augments, mitochondrial H2O2.
To assess the effect of these perturbations on p38 activity, immunoblot analyses was used to measure relative P-p38 levels in embryos injected with Mt-Cat or SOD2 mRNA compared to uninjected controls. In duplicate experiments, P-p38 signal was decreased in embryos expressing Mt-Cat mRNA, and increased in embryos expressing SOD2 mRNA (Fig. 3C). Quantification of the immunoblot signals showed that the effects produced by the two perturbations were of similar average magnitude, corresponding to an average ~2-fold change in P-p38 signal intensity compared to controls (Fig. 3D). These data suggest that mitochondrial H2O2 is both necessary and sufficient for activating p38.
Quantitative (real-time) RT-PCR was used to examine the effects of these perturbations on nodal expression. At the 32–60 cell stage (7 hpf), nodal transcript levels were found to be three-fold lower in embryos injected with Mt-Cat mRNA than in controls, whereas they were not significantly affected in embryos injected with SOD2 mRNA (Fig. 3E). This result was duplicated in 60-cell embryos from a second experiment (data not shown). However, the effect produced by Mt-Cat is transient; by the 60–120 cell stage (9 hpf) in the same injection experiment, nodal transcript abundance did not differ significantly between the perturbed embryos and controls (Fig. 3E). At this stage nodal expression has reached (or nearly reached) its ultimate per-embryo steady state level, attributable in large part to positive feedback, with initial inputs accounting for only (at most) 10% of the total level of expression (Nam et al., 2007). The results shown in Figure 3E therefore suggest that mitochondrial H2O2 constitutes an initial input that is rate-limiting but not sufficient for nodal activation, and furthermore, which is not required for the positive feedback activation.
The majority of embryos injected with either Mt-Cat or SOD2 mRNA developed into normal plutei, although a fraction (20–25%) of the former embryos displayed mild axial defects such as shortened skeletal rods. The otherwise normal development of an OA axis in these embryos is probably attributable to the fact that the enzymatic perturbation of mitochondrial redox signaling affects neither the ultimate levels of nodal expression, nor the spatial asymmetry in ROS levels associated with anisotropic mitochondrial distribution (data not shown). However, since nodal activation is suppressed in Mt-Cat-injected embryos, we reasoned that it should be possible to entrain OA polarity by localized application of this perturbation, providing an initial spatial bias in nodal expression that suffices to specify the axis. To test this, Mt-Cat mRNA was co-injected with a fluorescent dextran lineage tracer into single blastomeres of 2-cell stage embryos. As shown in Figure 3A and B this leads to reduction of ROS levels in the injected half compared to the uninjected half in a majority of embryos. Given the correlation between the first cleavage plane and the OA axis in S. purpuratus (Cameron, 1989), late gastrula stage descendents of 2-cell stage blastomeres can be scored as predominantly aboral or oral (Fig. 4A, B; see Materials and Methods). In late gastrula stage embryos developed from zygotes injected with Mt-Cat at the 2-cell stage, the marked lineage was scored as aboral in a significant majority of cases (Table 1; Fig. 4C), an effect proportionate to the effectiveness of Mt-Cat in decreasing ROS levels relative to the uninjected half (see above). In contrast, in control embryos injected with lineage tracer alone, or with lineage tracer plus SOD2 mRNA, the marked lineage was scored as oral and aboral in equal proportion (Table 1; Fig. 4C). These experiments indicate that OA polarity can be entrained by localized quenching, but not by localized augmentation, of mitochondrial H2O2.
Our results (Fig. 1) provide a direct confirmation of the correlation between mitochondrial distribution and the spatial pattern of nodal activity in the early embryo, a correlation that was previously inferred indirectly from fate mapping studies (Coffman et al., 2004). This correlation is demonstrated via the activity of an exogenous nodal-GFP reporter gene (Fig. 1A–C), and by the distribution of endogenous phospho-SMAD2/3 (Fig. 1D, E), both of which report the spatial pattern of Nodal activity (Nam et al., 2007; Yaguchi et al., 2007). As with our previous studies however, the correlation is not absolute, suggesting that the relationship between mitochondrial distribution and nodal activity is probabilistic rather than deterministic.
Mitochondrial distribution also correlates with ROS levels (Fig. 1F), most likely in the form of H2O2. The expression of nodal requires the activity of p38 stress-activated protein kinase (Bradham and McClay, 2006), which is known to be activated by H2O2 in mammalian cells cultured in vitro (Robinson et al., 1999). Our results show that the activity (as indicated by the phosphorylation state) of sea urchin p38 also responds to mitochondrial H2O2, decreasing when H2O2 is quenched and increasing when it is augmented (Fig. 3C). However, whereas the former treatment both suppresses the initial activation of nodal and entrains OA polarity toward aboral when localized by blastomere injections, the latter does not have the opposite effects (Figs. 3 and and4).4). Note that although p38 was shown to be necessary for nodal expression, its sufficiency was not tested (Bradham and McClay, 2006). Our data indicate that mitochondrial redox signaling via H2O2 (and by extension, p38) is rate-limiting for the initial activation of nodal, but that additional rate-limiting factors must also be involved.
What other factors might be at play? One likely possibility is that the spatial information provided by the mitochondrial gradient is complex and physiologically integrated, and hence not reducible to a single factor such as H2O2. For example, mitochondria influence intracellular redox state not only by emitting H2O2, but also by modulating NADH/NAD+, which is known to regulate transcription factor activity (Rutter et al., 2001; Zheng et al., 2003). Mitochondrial ATP production is rate-limiting for protein synthesis and numerous other aspects of cell physiology. Finally, mitochondrial density and activity strongly influence calcium signaling (Jacobson and Duchen, 2004), which has been shown to play a role in OA axis specification (Akasaka et al., 1997). Thus, mitochondria probably provide multiple, parallel signals that impinge upon the nodalcis -regulatory system (Fig. 5). Moreover, unlike the redox signaling tested here, some of these might affect the positive and/or negative feedback regulation of nodal, and thus have a bigger effect on axis specification.
It is also likely that mitochondria are only one source of spatial information among multiple alternatives capable of producing anisotropic nodal activity. As noted above the evidence gathered to date suggests that OA axis specification correlates significantly but not absolutely with asymmetric respiratory activity and/or distribution of mitochondria (Coffman and Davidson, 2001; Coffman et al., 2004, and Fig. 1E of this work). We have observed substantial variability in the steepness of the mitochondrial gradient, particularly among embryos developed from different batches of eggs. Given that short-range activation coupled with long-range inhibition provides a robust means of organizing specific patterns in response to subtle anisotropies, it is plausible that sources of spatial information other than mitochondria are capable of producing an initial anisotropy in nodal expression that suffices to specify an axis via Nodal-Lefty mediated epigenesis. Thus, rather than being mechanically dependent on a single source of spatial information, anisotropic nodal expression could be established by a variety of stochastically varying sources, any one of which might predominate in a given embryo.
We thank the two anonymous reviewers for providing comments and suggestions that improved the manuscript. Funding for this work was provided by grants from the NIH (ES016722 and RR016463).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.