Depression of Nuclear Transcription and Extension of mRNA Half-Life Under Anoxia in Artemia Franciscana Embryos

While physiological responses to oxygen deprivation among animals are diverse (Hochachka et al., 1993, 1996; Grieshaber et al., 1994; Somero, 1998), the resulting tolerance to this stress depends critically on the degree of metabolic depression achievable (Hochachka and Guppy, 1987; Hand, 1998). The greater the downregulation of ATP turnover, the longer the anoxic bout that can be tolerated. Short-term responses to hypoxia in mammalian cells can involve upregulation of selected suites of genes, such as those encoding for the enzymes of glycolysis (Semenza et al., 1994; Dang and Semenza, 1999). However, animals with the greatest anoxia tolerance are found among invertebrate phyla (Grieshaber et al., 1996; Hand and Hardewig, 1996), and many of these species downregulate metabolism to low levels. Considering the implication for energy-consuming processes, one might predict that nuclear transcription would be arrested and this, if true, would represent a markedly different strategy to cope with anoxia compared with that of vertebrates. In this study, we have employed nuclear run-on assays to evaluate transcriptional activity under anoxia in embryos of the brine shrimp Artemia franciscana, which is one of the most anoxia-tolerant animals known.

In response to anoxia, the encysted gastrula-stage embryos of A. franciscana enter a reversible state of developmental and metabolic quiescence that can last for 4 years or more (Clegg, 1997; for reviews of quiescence, see Hochachka and Guppy, 1987; Storey and Storey, 1990; Hand, 1991, 1998; Guppy and Withers, 1999; Hand and Hardewig, 1996; Hand and Podrabsky, 1999). Many catabolic and anabolic processes are depressed, resulting in energy flows that are very low compared with aerobic values (Hand and Gnaiger, 1988; Hontoria et al., 1993; Hand, 1995). The embryos may even approach an ametabolic state (Clegg, 1997). Considerable experimental evidence shows that the rapid and acute acidification of intracellular pH (from pH 7.7–7.9 to pH6.7) that occurs as these embryos enter anoxia (Busa at al, 1982; Kwast et al., 1995; Clegg et al., 1995) plays a key role in the metabolic arrest. A depression of protein synthesis occurs during anoxia in A. franciscana embryos (Clegg and Jackson, 1989; Hofmann and Hand, 1990, 1994) that is mediated through translational control. However, the fate of transcription during anoxia-induced quiescence is unknown.

Hardewig et al. (1996) demonstrated that the ontogenetic increase in levels of actin mRNA is prevented by exposure to anoxia and that the levels of this message are stable for several hours of anoxia, as is the case for total polyadenylated mRNA (Hofmann and Hand, 1992). One explanation for these patterns could be simultaneous arrest of nucleic acid synthesis and degradation, similar to that documented for protein degradation in these embryos under anoxia (see Anchordoguy et al., 1993; Anchordoguy and Hand, 1994, 1995). Transcription accounts for a modest proportion of the cellular energy budget (typically 1–10 % of basal metabolic rate in mammalian cells; Rolfe and Brown, 1997), but considering the large metabolic depression required to survive prolonged anoxia, even a 1 % drain on cellular energy stores might not be sustainable.

RNA synthesis can be readily detected in A. franciscana embryos during active development under normoxic, fully hydrated conditions (Clegg and Golub, 1969). These gastrula-stage embryos possess predominantly RNA polymerase II activity and some RNA polymerase I activity, but apparently little RNA polymerase III activity (D’Alessio and Bagshaw, 1977; Hentschel and Tata, 1977). In the present study, transcriptional activity was measured in nuclei isolated from both normoxic and anoxic embryos and from those exposed to aerobic acidosis and to carbon monoxide under anoxia. Nuclear run-on assays offer the advantage of providing a snapshot of the levels of in-vivo-initiated transcripts at the moment of nuclear isolation (e.g. Stallcup et al., 1978). Administration of carbon monoxide during hypoxia is known to blunt oxygen-dependent gene expression, an action mediated by the binding of CO to heme groups, which thereby locks the putative hemoprotein sensor (Goldberg et al., 1988; Acker, 1994; Ratcliffe et al., 1998) in the oxy conformation. Our evidence demonstrates that an acute, global arrest of transcriptional initiation occurs during anoxia in A. franciscana embryos, and that this is not reversed by supplying carbon monoxide under oxygen-free conditions.

Brine shrimp (Artemia franciscana) embryos were purchased from Sander’s Brine Shrimp Co. (Ogden, UT, USA). The hatching percentage for these embryos is approximately 80 % (Anchordoguy et al., 1993). Embryos were hydrated for at least 4 h at 0 ° C, washed in a solution of 0.25 mol l⁻¹ NaCl, and then dechorionated as described previously (Kwast and Hand, 1993). The embryos were allowed to develop for 4 h under normoxia (0.25 mol l⁻¹ NaCl solution; air-saturated, 22–23 °C) and then incubated in media equilibrated with one of the desired gas mixtures (anoxia, 100 % Ar or N₂; aerobic acidosis, 60 % CO₂:40 % O₂; carbon monoxide, 80 % CO in oxygen-free nitrogen). The carbon monoxide was obtained pre-mixed from Matheson Gas Products (Cucamonga, CA, USA). Recovery from anoxia was achieved simply by returning embryos to normoxic incubation conditions. At the end of each treatment regime, embryos were chilled to 0 ° C while still immersed in the appropriate gas-equilibrated medium. The embryos were then filtered through Dura Wipe cloth (Johnson and Johnson Advanced Materials Company, New Brunswick, NJ, USA), rapidly transferred to ice-cold homogenization medium (HM) and processed.

RNAase-free conditions were maintained throughout the nuclear isolation and run-on assays. Nuclei were isolated using the method of Acey et al. (1994) with modification. Hydrated and dechorionated embryos (5 g) were homogenized with two passes of a Teflon/glass homogenizer (model 3431; Thomas Scientific, Swedesboro, NJ, USA) in 8 volumes of ice-cold HM consisting of 10 mmol l⁻¹ Tris-HCl buffer, pH 7.5, 10 mmol l⁻¹ MgCl₂, 10 mmol l⁻¹ NaCl, and 0.1 % Nonidet P-40 (NP-40; Igepal CA-630, Sigma Chemical Co., St Louis, MO, USA). The pestle was motor-driven at approximately 500 revs min⁻¹. The homogenate was filtered through one layer of Dura Wipe cloth and centrifuged at 800 g for 10 min at 2 °C. The pellet was gently resuspended in 8 ml of HM without NP-40, and 2 ml samples were layered onto each of four Percoll pads (12 ml of 75 % Percoll containing 150 mmol l⁻¹ NaCl, 10 mmol l⁻¹ MgCl₂ and 10 mmol l⁻¹ Tris-HCl, pH 7.5) contained in 50 ml centrifuge tubes. The samples were centrifuged in a fixed-angle rotor at 15 000 g for 15 min at 2 °C. Nuclei accumulated at the sample/Percoll interface, while yolk platelets sedimented through the Percoll to the bottom of the tubes. Nuclei were collected by aspiration, diluted in approximately 5 volumes of HM without NP-40 and recentrifuged at 850 g for 10 min at 2 °C. The washing procedure was then repeated once to ensure removal of all the Percoll. The final nuclear pellets were resuspended in approximately 1 volume of glycerol storage buffer (50 mmol l⁻¹ Tris-HCl, pH 8.3, 40 % glycerol, 5 mmol l⁻¹ MgCl₂ and 0.1 mmol l⁻¹ EDTA) and frozen at −80 °C until use. To estimate nuclear concentration, nuclei were stained with 0.01 % Methyl Green and quantified with an improved Neubauer hemocytometer (C. A. Hausser & Son, Philadelphia, PA, USA). The typical yield from 5 g of dechorionated embryos was approximately 400×10⁶ to 500×10⁶ nuclei.

Unless indicated otherwise, run-on assays (modified from Acey et al., 1994) were performed under conditions optimized for nuclear concentration, pH and nucleotide concentrations. The optimized reaction mixture (total assay volume 209.5 µl) contained: 100 µl of a buffer solution [100 mmol l⁻¹ Tris, 100 mmol l⁻¹ maleate, 150 mmol l⁻¹ KCl, 40 mmol l⁻¹ (NH₄)₂SO_4, 4.8 mmol l⁻¹ magnesium acetate and 4 mmol l⁻¹ MnCl₂, pH 7.6], 15 µl of a nucleotide mixture [3.3 mmol l⁻¹ ATP, 6.7 mmol l⁻¹ CTP, 6.7 mmol l⁻¹ GTP], 0.5 µl of 20 mmol l⁻¹ UTP, 5 µl of 111 TBq mmol⁻¹ [³²P]UTP (185 MBq; DuPont NEN Life Science Products, Boston, MA, USA), 3 µl of 1 mol l⁻¹ dithiothreitol in 0.01 mol l⁻¹ sodium acetate, pH 5.2, 2 µl of placental RNasin (10 units total; Promega, Madison, WI, USA), 4 µl of water and 80 µl of a nuclear suspension containing 650×10⁶ to 800×10⁶ nuclei ml⁻¹. Assays were initiated by the addition of nuclei, and the reaction mixtures were incubated at 30 °C for the indicated times. Triplicate samples (30 µl each) were removed at the selected time points and applied to Whatman GF/C glass fiber filters (Whatman International Ltd, Maidstone, UK; 2.4 cm). The filters were immediately transferred to ice-cold 10 % trichloroacetic acid (TCA) containing 1 % sodium pyrophosphate and soaked for 1 h with stirring. The filters were then given two successive washes in 5 % TCA with 0.5 % pyrophosphate for 10 min each, followed by three 10 min washes in 95 % ethanol. During all washing procedures, the filters were kept in perforated plastic containers to prevent mechanical damage. The filters were dried in air. Radioactivity was determined by liquid scintillation counting in Scintiverse II fluor (Fisher Scientific, Denver, CO, USA) and expressed as disints min⁻¹. Values at time zero were subtracted from values at later time points to correct for non-specific binding of radioactivity to the filters. Data were analyzed by analysis of variance (ANOVA) or regression analysis where appropriate. Values are presented as means ± S.E.M.

For the determination of apparent Michaelis–Menten constants (K_m), the concentration of one nucleotide was varied while the other nucleotide concentrations were held constant at 240 µmol l⁻¹ ATP, 480 µmol l⁻¹ CTP, 480 µmol l⁻¹ GTP and 48 µmol l⁻¹ UTP. When UTP concentration was varied, the specific radioactivity of UTP was maintained at 185 MBq µmol⁻¹. K_m values were estimated by nonlinear regression analysis using the Leonora program (Cornish-Bowden, 1995). The pH profile for elongation was generated by adjusting the pH of the Tris-maleate reaction buffer and then verifying the pH values of the final reaction mixtures by direct measurement in parallel samples.

Nuclear run-on assays quantify the elongation of in-vivo-initiated transcripts. After cell lysis, initiation of new transcripts is negligible (see Weber et al., 1977; Groudine et al., 1981; Hofer and Darnell, 1981; Lohr and Ide, 1983). Thus, the level of transcriptional initiation present at the time of cell lysis can be compared among organismic treatment groups. As is shown in Fig. 1, incorporation of [³²P]UTP by isolated nuclei from A. franciscana embryos was linear for 15 min. Consequently, subsequent assays were restricted to this period. UTP incorporation in these assays was inhibited by 81 % by adding 50 µg ml⁻¹ actinomycin D (data not shown). When isolated nuclei were serially diluted across a fourfold range of nuclear concentration (Fig. 1, inset), there was no significant difference in transcriptional activity when expressed as a function of nuclear density (the slope is not significantly different from zero, P>0.05).

Elongation of the transcripts was dependent on ATP, CTP, GTP and UTP concentration. For all four precursors of RNA synthesis, transcription followed Michaelis–Menten saturation kinetics (Fig. 2A–D). Apparent K_m values (µmol l⁻¹ ± S.E.M. of kinetic parameter) were 5.0±0.8, 2.9±0.1, 5.9±0.5 and 6.9±1.3 µmol l⁻¹ for ATP, CTP, GTP and UTP, respectively. Our apparent K_m values for nuclei from A. franciscana embryos closely match other published values for isolated nuclei, which fall within the range 1–10 µmol l⁻¹ (Furuichi, 1981; Job et al., 1988; Klumpp et al., 1998).

Nuclei isolated from normoxic embryos were used to determine the pH sensitivity for transcriptional elongation (Fig. 3). When pH was reduced from 7.9, the value approximating the intracellular pH (pHi) of aerobic embryos, to pH 6.7–6.8 (reflective of the value during early quiescence; Busa et al., 1982; Kwast et al., 1995), elongation was inhibited by 50 %. If assay pH was lowered further to 6.3, elongation rate was inhibited by 65 %. The latter pH value is physiologically relevant because it is observed in vivo after an overnight exposure of embryos to anoxia (Busa et al., 1982).

Transcriptional activity in A. franciscana embryos during pre-emergence development was unchanged over 10 h at room temperature (22–23 °C) under normoxia (Fig. 4). The slope of the relationship between UTP incorporation rate by isolated nuclei versus time of embryonic development was not statistically different from zero (P>0.05). However, when aerobic development was interrupted by exposing embryos to 4 h of anoxia (Fig. 5), transcriptional activity was reduced by 79.3±1 % (mean ± S.E.M., N=3; analysis of covariance, P<0.05). When the duration of the anoxic bout was extended to 24 h, transcriptional activity was depressed a maximum of 88.2±0.7 % (mean ± S.E.M., N=3). Transcriptional activity was fully restored after anoxia by allowing the embryos 1 h of normoxic recovery (Fig. 5).

Incubation of embryos under the artificial condition of aerobic acidosis (60 % CO₂:40 % O₂) provides insight into the role of pHi acidification in the arrest of transcription. This treatment is known to depress the pHi of embryos to 6.8 (Busa and Crowe, 1983) under fully aerobic conditions, a value comparable with that measured in embryos after 1 h of anoxia. Intracellular ATP concentration remains high and unchanged for several hours of aerobic acidosis, segregating this factor from pHi (Carpenter and Hand, 1986; Anchordoguy and Hand, 1995). When nuclei were isolated from embryos given 4 h of normoxic incubation followed by an additional 4 h exposure to aerobic acidosis, the initiation of transcription was depressed by 55.2±3.6 % compared with the normoxic value (Fig. 6A; ANOVA, P<0.05). Hyperoxia experienced by embryos during the aerobic acidosis treatment had no significant effect on transcription; UTP incorporation was identical for embryos incubated under normoxia for 4 h (air-saturated medium) and in those incubated in medium equilibrated with 60 % N₂:40 % O₂ (ANOVA, P>0.05; data not shown). Thus, over half of the transcriptional arrest under anoxia can be ascribed to acidification of pHi.

To evaluate the involvement of a heme-based oxygen sensor in the global depression of transcription seen under anoxic exposure, embryos were incubated in the presence of 80 % CO:20 % N₂. Ligation of the porphyrin heme with CO is sufficient to convert hemoproteins to the oxy conformation in the absence of oxygen (Goldberg et al., 1988). Thus, if a heme-based oxygen sensor were involved in the global depression of transcription under anoxia, the addition of 80 % carbon monoxide to anoxic embryos would be expected to reverse at least some of the transcriptional depression. However, the addition of carbon monoxide to anoxic embryos was without effect on transcription, i.e. it did not reverse the depression caused by anoxia alone (Fig. 6B; ANOVA, P>0.05).

In contrast to recent work with mammalian systems in which upregulation of selected genes during oxygen-limitation is prevalent (see Bunn and Poyton, 1996; Semenza et al., 1994; Dang and Semenza, 1999), exposure of A. franciscana to anoxia elicits a global arrest of nuclear transcription. Instead of activating compensatory mechanisms to cope with anoxia, this animal enters a state of deep metabolic depression during which gene expression represents a cellular cost that is apparently not compatible with survival. When embryos were exposed to anoxia, transcriptional initiation was reduced by approximately 90 % relative to normoxic controls. This in vivo downregulation of initiation occurs rapidly and, to our knowledge, is the largest transcriptional arrest reported to date in response to oxygen limitation. The elongation of these transcripts was performed in our experiments under optimal assay conditions (i.e. high pH and nucleotide availability) but, as discussed below, the actual transcriptional activity under anoxia may be even lower than estimated if pausing of the polymerase during elongation is also promoted because of the low pH and ATP availability characteristic of this state in these embryos.

Run-on assays performed with nuclei isolated from aerobic acidotic embryos revealed a 55 % reduction in the initiation of transcription relative to normoxic controls (Fig. 6A), thereby indicating a clear role for acidification in the inhibition of initiation. In addition to a direct influence on the transcriptional machinery, the proximal effect of pH acidification on initiation could involve the recently reported pH-dependent translocation of p26, a small heat-shock protein, to the nuclear compartment of A. franciscana embryos (Clegg et al., 1994; Liang et al., 1997a; Liang and MacRae, 1999). Under anoxia, approximately half the total embryonic p26 is transported from the cytoplasm to the nucleus, and this process is reversed during aerobic recovery. Acidic pH favors the accumulation of p26 in the nucleus, and realkalization of embryos favors the removal of p26 from the nucleus. While the protein clearly possesses molecular chaperone activity (Liang et al., 1997b), it may well be that this protein is involved in some manner in the transcriptional arrest reported in the present study (see Clegg et al., 1995). It is appropriate to note that the p26 is accumulated in the encysted embryo to levels that represent 15 % of the total non-yolk protein (Clegg et al., 1994). Regardless of the mechanism by which pHi affects transcription, the results obtained under aerobic acidosis indicate that other factors are required to explain fully the depression of transcriptional initiation seen in vivo during anoxia.

Another hallmark change in A. franciscana embryos known to occur sufficiently rapidly to be of relevance for the transcriptional arrest observed in this study is the level of cellular ATP. While ATP concentration is well-regulated in most mammalian systems and major declines are usually only associated with cell death (Atkinson, 1977), anoxic A. franciscana embryos experience a large reduction in ATP concentration (over 80 %) during the first hour of anoxia (Stocco et al., 1972; Carpenter and Hand, 1986; Rees et al., 1989; Anchordoguy and Hand, 1994). Indeed, part of the acidification in pHi observed during early quiescence is caused by net ATP hydrolysis (see Kwast et al., 1995 and references therein). Data for bacterial transcription suggest that initiation at the rRNA promoter is dependent on high nucleoside triphosphate concentrations, such that growth-rate-dependent changes in ATP and GTP concentrations may eventually determine rates of ribosome synthesis (Gaal et al., 1997). Furthermore, RNA synthesis in Ehrlich ascites cells is also apparently influenced by ATP and GTP levels (Grummt and Grummt, 1976). On the basis of these considerations, it is likely that the low ATP levels in the absence of oxygen explain a sizable proportion of the in vivo arrest of transcriptional initiation in A. franciscana embryos.

While many, perhaps most, changes in transcriptional rate are due to depressed initiation (see Egan et al., 1994; Lassam and Jay, 1989; Spencer and Kilvert, 1993), recent work has shown that pausing of the polymerase during elongation may limit the expression of many genes (e.g. β-actin, c-fos, c-myc, Hsp70; for a review, see Uptain et al., 1997). Elongational arrest within coding regions can effectively depress mRNA synthesis from the affected genes (Shilatiford, 1998) and, in principle, arrest of elongation and initiation are not mutually exclusive. In A. franciscana embryos, the large changes in pHi and ATP availability under anoxia may work together to downregulate both processes. If re-entry of paused polymerases into productive elongation occurs when isolated nuclei are placed in optimized conditions for run-on assays, then elongational pause would go undetected and the in vivo arrest of transcription would be even greater than suggested by the results presented in Fig. 5.

In attempts to identify other factors involved in the arrest of transcription in quiescent embryos, we also considered the activity of heme-based oxygen sensors. Kwast and Hand (1996a,b) provided evidence for a mitochondrially located oxygen sensor that may mediate translational arrest within this organelle during anoxia in A. franciscana embryos. A sizable literature now suggests that the nuclear transcription of many hypoxia-inducible genes is regulated via heme-dependent oxygen sensing (Bunn and Poyton, 1996; Dang and Semenza, 1999; Land and Hochachka, 1995). One test for the involvement of an oxygen sensor is the ability to reverse the effects of hypoxia/anoxia on transcription by administering carbon monoxide (Goldberg et al., 1988). In the case of A. franciscana embryos under anoxia, CO-induced blunting of the anoxic effect on transcription would mean a deinhibition of the process. However, the addition of 80 % carbon monoxide to anoxic embryos was without effect and did not rescue transcriptional initiation (Fig. 6B). Thus, it seems unlikely that a heme-based oxygen sensor is pivotal in the global inhibition of nuclear transcription observed during quiescence. One might argue, however, that the CO treatment was ineffective under anoxia because pHi and adenylate levels were simply too low in vivo to permit activation of transcription even if a putative oxygen sensor had been shifted to the oxy conformation by CO binding. Perhaps it is under hypoxia rather than anoxia where oxygen sensing may play an observable role in A. franciscana embryos.

An additional conclusion that emerges from this study, when combined with our results that show unchanged mRNA concentrations for at least the first 6 h of anoxia (Hofmann and Hand, 1992; Hardewig et al., 1996), is that mRNA half-life (t_1/2) is extended in the anoxic embryo by at least 8.5-fold compared with the normoxic embryo [where t_1/2=[RNA]×log_e2×(synthesis rate)⁻¹]. mRNA pools in the embryo would decline markedly if message stability were not increased simultaneously in the face of the transcriptional arrest during anoxia. One biological advantage of stable mRNA pools under anoxia would be the quick resumption of protein synthesis as soon as oxygen is returned to the embryo. In summary, the depression of transcription and the extension of mRNA half-life are concordant with limited energy availability. Transcriptional downregulation is facilitated by the decreases in pHi and ATP concentration that occur under anoxia and their resulting influence on initiation and perhaps elongation. At this juncture, the available evidence does not indicate the involvement of a hemoprotein oxygen sensor that downregulates nuclear transcription. In vertebrates, the upregulation of critical suites of genes serves a homeostatic role during anoxic bouts of limited duration. In contrast, the global arrest seen in A. franciscana precludes normal metabolic functioning, but it extends the tolerance to anoxia dramatically, which may be more critical when the variable physical environment of these animals is considered. Multiple cellular solutions for coping with oxygen limitation, each with inherent biological tradeoffs, are available across animal phyla.

This work was supported by National Science Foundation grant IBN-9723746 to S.H. and by an EPO Biology Fellowship and Graduate School Dean’s Small Grant Award to F.v.B. We would like to thank Dr Mike Grant for assistance with statistical analyses and Dr Roger Acey for advice on protocols for nuclear isolation and run-on assays.