Molecular Basis of the Difference in Oxygen Affinity Between Maternal and Foetal Red Blood Cells in the Viviparous Garter Snake Thamnophis Elegans

Viviparity has apparently evolved independently in many lineages of the squamate reptiles (Shine, 1984, 1985). Therefore, a comparative examination of the physiological strategies used by viviparous reptiles should indicate the range of physiological mechanisms which successfully support foetal development in viviparous vertebrates. One important aspect of reproductive physiology involves the transfer of oxygen from mother to foetus and the molecular mechanisms which facilitate that transfer. Although such facilitation has been studied extensively in mammals, relatively little work has focused on the lower vertebrates, especially reptiles.

Manwell (1960) reported that suspensions of foetal red cells from the viviparous garter snake Thamnophis elegans had higher oxygen affinities than did suspensions of maternal red cells. The basis for this difference has not been elucidated. However, almost all viviparous vertebrates show a higher blood oxygen affinity in foetal than in maternal blood. In mammals, this difference appears to facilitate the transfer of oxygen from mother to foetus and promote normal foetal growth and development (Battaglia et al. 1969; Hebbel et al. 1980; Bauer et al. 1981). Molecular mechanisms responsible for this higher foetal blood oxygen affinity vary among vertebrate species, and include the production of high oxygen affinity foetal haemoglobins and/or differing maternal and foetal concentrations of red cell organic phosphates (e.g. Manwell, 1963; Tyuma & Shimizu, 1969; Blunt et al. 1971; Novy et al. 1973; Garlick et al. 1979; Weber & Hartvig, 1984).

Among the viviparous reptiles, the oxygen affinity of maternal blood is lower than that of foetal blood in the lizard, Sphenomorphus quoyii, and the cotton-mouth snake, Agkistrodon piscivoras, although the maternal and foetal haemo-globins are electrophoretically indistinguishable (Grigg & Harlow, 1981; Birchard et al. 1984). The difference in A. piscivoras appears to be mediated by organic phosphate (nucleoside triphosphate) concentrations, which are lower in foetal than in maternal red cells.

A different situation may exist in the garter snake. Pough (1977) found that blood from the neonates of T. sirtalis had a higher oxygen affinity than that of adults. He also found an ontogenetic change in the electrophoretic pattern of the haemoglobin and a decrease in the organic phosphate to haemoglobin ratio with development. (Foetuses were not examined in his study.) These results suggest that the foetal-maternal difference in oxygen affinity noted by Manwell (1960) in T. elegans may be associated with differences in haemoglobin structure rather than differences in organic phosphate concentrations. Consequently, we studied T. elegans with respect to structure and function of haemoglobins from the adult and from foetuses at several developmental stages. We also examined the organic phosphates of maternal and foetal T. elegans red cells, and the influence of adenosine triphosphate (ATP) on haemoglobin function. The concentrations of various divalent cations, mean corpuscular haemoglobin concentration (MCHC), and the level of nonfunctional haemoglobin (methaemoglobin) can influence oxygen-binding properties of haemoglobins. Therefore, we examined these factors in maternal and foetal bloods. The results of our studies are reported in this paper, and a molecular mechanism is proposed to explain why the oxygen affinity is higher in foetal than in maternal red blood cells of Thamnophis elegans.

Adult garter snakes were collected in early summer in Latah Co., Idaho, and identified as Thamnophis elegans according to Nussbaum et al. (1983). Pregnancy was determined by abdominal palpation. The developmental stage of the foetuses was determined according to Zehr (1962). Foetuses of developmental stages 30-36 were examined in this study.

Snakes were bled into ice-cold heparinized 150 mmol 1⁻¹ NaCL All following preparative procedures were performed at 4°C. Red cells were washed three times with 150 mmol 1⁻¹ NaCl by centrifugation at 1000g for 5 min. Cells were lysed with distilled water and a single freeze-thaw cycle. The haemolysate was centrifuged at 5000g for 10min. Subsequently, the supernatant was passed through an 85 cm x 1·6 cm column of Sephadex G100-120 equilibrated with 100 mmol 1⁻¹ NaCl, 10 mmol 1⁻¹ MgCl₂, 50 mmol 1⁻¹ Tris and titrated to pH 8·0 with HC1.

Relative molecular masses of the native haemoglobins were determined on the Sephadex G100-120 column. Elution was monitored spectrophotometrically at 280 nm. Relative molecular mass standards used were bovine serum albumin (M_r= 68000), human haemoglobin (M_r = 64500), ovalbumin (Af_r = 43000) and sperm whale myoglobin (M_r = 17 800). Blue dextran and tryptophan were used to indicate the column void volume and salt peak, respectively.

Native polyacrylamide gel electrophoresis (PAGE) was carried out in tubes using the method of Davis (1964). Resolving gels were 7-5% (w/v) acrylamide, pH8-9 and stacking gels were 3-0% acrylamide, pH7-2. The acrylamide to bisacrylamide mass ratio was 30:0-8. Separate electrophoretic experiments were carried out on carboxy-and cyanmethaemoglobin. Haemoglobin samples were converted to carboxyhaemoglobin by addition of sodium dithionite and bubbling with carbon monoxide, and to cyanmethaemoglobin by the addition of potassium cyanide and potassium ferricyanide.

Sodium dodecyl sulphate (SDS)-PAGE of the globin chains was carried out by the method of Laemmli (1970) on 1·5 mm slab gels. Resolving gels were 15% acrylamide, 1·0mmoll⁻¹ dithiothreitol, 0·1% SDS (w/v), pH8·8 with an acryl-amide to bisacrylamide ratio of 30:0·8. Globin chains were denatured by incubation at 40°C for 1-5h in a solution of 62·5mmoll⁻¹ Tris, 2·0% SDS, 5% 2-mercaptoethanol (v/v), l-4mmoll⁻¹ phenylmethylsulphonyl fluoride and titrated to pH 6·8 with HC1. Samples were stored frozen at −20°C for up to 1 month. Relative molecular mass standards used were bovine serum albumin, human haemoglobin, ovalbumin, sperm whale myoglobin and lysozyme (M_r= 14300).

Gel electrophoresis of the globin chains was carried out in the presence of 6·25 mol 1⁻¹ urea (mixed fresh to minimize cyanate formation), 1·0mmoll⁻¹ dithiothreitol, pH 2·2 (established with glacial acetic acid) on 6·0% acrylamide gels having an acrylamide to bisacrylamide ratio of 20:1·2 (Panyim & Chalkley, 1969; Poole et al. 1974). The samples to be electrophoresed were first carboxy-methylated using iodoacetamide as described by Poole et al. (1974) to block sulphhydryl bridge formation.

For oxygen-binding studies, haemoglobins were first purified on the Sephadex G100-120 column as described above. The purified haemoglobin solution was concentrated over sucrose before being further stripped of organic phosphates by passage through a 35 cm x 1·6 cm column of Sephadex G 25-50 equilibrated with 100 mmoll¹ NaCl, 50mmol 1⁻¹ HC1, titrated to pH8·5 with Tris (Garlick et al. 1979). Haemoglobins were dialysed to the desired pH for 2h immediately before binding studies. Oxygen-binding buffers were 100 mmol 1⁻¹ NaCl, 50 mmol 1⁻¹ HC1, titrated to pH 7·0 and below with Bistris and above pH 7 0 with Tris. Oxygen equilibrium measurements were made at 20°C in glass tonometers (from Ryan Scientific Glass, Pullman, WA) by monitoring the change in absorbance at 540, 560, 575 and 580 nm as deoxygenated haemoglobin was reoxygenated (Benesch et al. 1965). Four tonometers were run simultaneously. For these functional studies, the haemoglobin tetramer concentration was 12-18 μmo1l 1⁻¹ determined spectrophotometrically as the cyanmet-derivative using a millimolar haem extinc-tion coefficient of 11-0. The values of P₅₀ (the oxygen tension for half saturation of the haemoglobin) and the Hill coefficient (used as an indicator of haem-haem interaction) were determined from a Hill plot using 3-6 data points corresponding to haemoglobin which was 25-75% saturated with oxygen. The P₅₀ value calculated for each tonometer represents n = 1 in the calculations of Bohr coefficients and in Fig. 6. Oxygen-binding studies were performed at various pH values in the absence and presence of ATP. ATP was added to a final concentration of 2 mmol 1⁻¹ from a concentrated stock solution prepared immedi-ately before use.

For determination of red cell organic phosphate concentrations, snakes were bled into ice-cold heparinized Ringer’s saline (HOmmoll⁻¹ NaCl, l·9mmoll⁻¹ KC1, Tlmmoll⁻¹ CaCl₂, 2·4mmoll⁻¹ NaHCO₃, pH7-6). Red cells were washed three times in this saline by centrifugation at 1000g for 5 min at 4°C. Washed and resuspended cells were extracted with an equal volume of ice-cold 12% trichloro-acetic acid (TCA), within 30 min of blood collection, and centrifuged for 10 min at 5000g. Total red cell nucleoside triphosphate (NTP) concentrations were deter-mined in the supernatant using an enzymatic assay kit from Sigma Chemical Co. (no. 366-UV). Red cell concentrations of 2,3-diphosphoglycerate (2,3-DPG) were also determined using an enzymatic assay kit from Sigma (no. 35-UV) on TCA extracts of red cell suspensions. NTP concentrations are expressed as mole NTP per mole haemoglobin tetramer.

For analysis of ATP to guanosine triphosphate (GTP) ratios, rinsed, packed red cells were extracted with perchloric acid and supernatants were neutralized with concentrated KOH using the method of Bartlett (1978). Extracts were then analysed by HPLC with a Varian AX-10 column using a gradient of O Olmoll⁻¹ KH₂PO₄, pH3·0 to 0·75 mol 1⁻¹ KH₂PO₄, pH5 0.

For analysis of red cell magnesium and calcium concentrations, maternal and foetal cells were collected in ice-cold, heparinized 150 mmol 1⁻¹ NaCl and washed as described above. Washed cell pellets were lysed with distilled water, centrifuged at 20000g for 30 min and the supernatants were used for analysis. Magnesium and calcium levels were determined using inductively coupled plasma atomic emission spectrometry by monitoring atomic emission with an Applied Research Labora-tones 35000 CICP. Results are expressed as mole ion per mole haemoglobin tetramer.

Haematocrit values of red cell suspensions were obtained using 10 /rl micropip-ette capillary tubes; these tubes were centrifuged in an IECMB haematocrit centrifuge for 4min. MCHC values were calculated from red cell suspension haemoglobin concentrations and haematocrit values.

Methaemoglobin concentrations were determined as described by Dubowski (1960), with the modifications suggested by Gruca & Grigg (1980). Red cell suspensions were lysed using approximately one drop of Triton X-100 per 10 ml of suspension. As snake red cells are nucleated, it was necessary first to remove cell debris from the haemolysates to obtain consistent results; this was accomplished by centrifugation at 1000g for 5 min. Methaemoglobin concentrations were then determined spectrophotometrically on the supernatant fractions.

All chemicals and relative molecular mass standards were from Sigma Chemical Co. (St Louis, MO) with the exception of SDS which was from Bio-Rad (Richmond, CA).

The apparent relative molecular mass of haemoglobins from adult and foetal T. elegans was approximately 54000, on the basis of the elution of protein standards from the Sephadex G100-120 column. However, when run together, adult garter snake haemoglobin co-eluted with adult human haemoglobin (Fig. 1), and the haemoglobin from stage 30 and 36 foetuses co-eluted with adult snake haemoglobin.

Electrophoresis of the native cyanmethaemoglobins of adult and stage 30 and 36 foetuses resulted in one relatively broad electrophoretic band for each (Fig. 2). When mixed and run together, these haemoglobins co-electrophoresed. Compar-able results (not shown) were obtained with carboxyhaemoglobin. SDS slab gel electrophoresis of globin chains from adult and stage 30 and 36 foetuses showed a single band for each (Fig. 3) with a relative molecular mass of 15 400 based on the standards. However, when snake and human globins were electrophoresed in the same lane of the gel, the snake globin co-electrophoresed with the human beta-globin chain. Electrophoresis of carboxymethylated globin chains from adults and stage 30 foetuses in the presence of urea at low pH resulted in three electrophoreti-cally separable bands for each (Fig. 4). Globins from stage 30 foetuses co-electrophoresed with those of the adult. Comparable results (not shown) were obtained with stage 36 foetal globins. Human haemoglobin, run as a control, showed two bands.

Oxygen-binding studies yielded very similar functional characteristics for stripped adult and foetal haemoglobins in the absence and presence of 2 mmol 1⁻¹ ATP. In the absence of ATP, the Hill coefficient for each haemoglobin averaged about 3·0 over most of the pH range examined but increased to about 4·0 at pH values between 6·8 and 7·2 (Figs 5 and 6). P₅₀ values ranged from 2·0 to 5·0 mmHg between pH6·5 and 8·5. Adult and foetal haemoglobins showed little Bohr effect with Bohr coefficients, ΔlogP₅₀/ΔpH, of −0 06 (N = 24) and -0·01 (N =12), respectively (with 95% confidence intervals of -0·10 to -0·02 and -0·12 to 0·09, respectively), over all pH values tested (Fig. 6). These Bohr coefficients were significantly different (P<0·05, F-test, Snedecor & Cochran, 1967).

In the presence of 2 mmol 1⁻¹ ATP, adult and foetal stripped haemoglobins had a lower oxygen affinity at pH values below 8·0 (Fig. 6). Adult and foetal haemoglobins showed significant Bohr effects with Bohr coefficients of -0·37 (N = 24) and -0·39 (N =12), respectively (with 95% confidence intervals of -0·44 to -0·29 and -0·60 to −018, respectively), over pH values of 7·0-8·0. These Bohr coefficients were not significantly different. Over this range of pH, Hill coefficients did not exceed 3·4 (Figs 5 and 6).

Maternal red cells had higher NTP levels than the red cells of their foetuses in all cases, with a maternal-foetal difference of 0·93 ± 0·68 mol NTP mol⁻¹ haemo-globin tetramer (Table 1). As determined by HPLC, the primary NTP in the red cells of both maternal and foetal bloods was ATP (Table 2). No 2,3-DPG was detected in either adult or foetal red cells.

Whereas maternal and foetal haemolysates had similar molar magnesium to haemoglobin ratios, foetal haemolysates had significantly higher calcium to haemoglobin ratios (Table 2). Maternal snakes had significantly higher MCHC values than foetal snakes, with a maternal-foetal difference of 0·76 ± 0·62 mmol 1⁻¹ (Table 2). The methaemoglobin levels in adult and foetal cells were low and not significantly different for centrifuged lysates (Table 2).

In some viviparous vertebrates, maternal-foetal oxygen transfer may be facilitated by the presence of a unique foetal haemoglobin which has an intrinsically higher affinity for oxygen than does the adult haemoglobin. Since Pough (1977) reported ontogenetic changes in the haemoglobins of T. sirtalis, it is possible that different foetal and maternal haemoglobins underlie the difference in red cell oxygen affinities of foetal and maternal T. elegans found by Manwell (1960). The data from the present study, however, do not support this hypothesis.

Adult and foetal haemoglobins were structurally indistinguishable by several forms of electrophoresis and by gel filtration chromatography. Furthermore, based on SDS-PAGE and chromatography, these native molecules appear to be tetrameric with a subunit relative molecular mass of about 16000. [Based on the elution of protein standards from the gel filtration column, the apparent relative molecular mass of these molecules was 54 000. This value appears low for a vertebrate haemoglobin, and may be due to the tendency of some haemoglobins to adsorb to Sephadex gel thereby yielding lower values than expected (Wilkins, 1970; luchi, 1973). However, T. elegans haemoglobin co-eluted with adult human haemoglobin when chromatographed together, suggesting a native relative molecular mass of about 64000.]

The results of the native PAGE and low pH/urea PAGE together suggested that the snake haemoglobin may exist as an asymmetrical tetramer, as has been suggested for the haemoglobins of some salmon and sharks (Tsuyuki & Ronald, 1971; Andersen et al. 1973). However, as vertebrate haemoglobins are generally symmetrical tetramers (Coates, 1975), it is more likely that the snake haemoglobin actually exists as two symmetrical tetramers. This could explain the relatively broad band seen by native PAGE (Fig. 2).

Adult and foetal haemoglobins had relatively high oxygen affinities and low sensitivities to pH in the absence of ATP (Fig. 6). Furthermore, both haemo-globins showed strikingly high Hill coefficients near pH 7·0 in the absence of ATP (Fig. 6). Although similar values have been reported for whole blood from two lizards (Pough, 1969) and a rattlesnake (MacMahon & Hamer, 1975), this is a very unusual finding. Since human sickle cell haemoglobin shows a Hill coefficient of 5-6 owing to haemoglobin polymerization (Gill et al. 1978), our data suggest that T. elegans haemoglobins may also show some tetramer-tetramer interaction at pH 7·0. In the presence of ATP, the haemoglobins from both sources showed an appreciably reduced oxygen affinity, a marked pH sensitivity and a decreased Hill coefficient near pH 7·0 (Fig. 6). Thus, adult and foetal haemoglobins had similar functional characteristics and both were similarly responsive to ATP.

That maternal red cell levels of ATP were appreciably higher than foetal levels (Table 1) suggests that a difference in organic phosphate concentrations is an important molecular strategy underlying the difference in oxygen affinities of maternal and foetal red cells. A doubt about this interpretation arises from a consideration of the stoichiometry of the organic phosphate-haemoglobin interac-tion: one polyanion interacts with one tetrameric haemoglobin. In both foetal and maternal T. elegans red cells, the ATP to haemoglobin ratio was significantly greater than one, suggesting that the haemoglobin in both cell types may be saturated with ATP. Therefore, it becomes necessary to consider intracellular levels of divalent cations.

Magnesium and calcium ions bind to ATP in a 1:1 molar ratio forming a fairly stable complex which binds to haemoglobin more weakly than does ATP alone. Thus, these cations increase the oxygen affinity of haemoglobin by decreasing the ATP-haemoglobin interaction (White et al. 1964; Lehninger, 1975; Weber & Lykkeboe, 1978; Burton, 1980; Ingermann & Terwilliger, 1981). T. elegans foetal and maternal NTP/haemoglobin ratios were about 2·3 and 3·2, respectively, and total divalent cation/haemoglobin ratios were about 1·6 and 1·8, respectively (Table 2). If complete complexing were to occur between the divalent cations and NTP, foetal but not maternal red cells would have less than saturating levels of free NTP available to interact with haemoglobin (0·7 and 1·4 NTP haemoglobin⁻¹, respectively). This should promote a higher oxygen affinity in foetal red cells. Magnesium and calcium ions interact with membrane phospholipids and sedimen-table material (Rose, 1968; Sanui, 1970; Rubin, 1975; Sanui & Rubin, 1979). Since the divalent cation concentrations were determined on haemolysates in this study, the values did not include divalent cations bound to the intra-and extracellular faces of the red cell membranes. Therefore, the divalent cation levels may be underestimates; those cations bound to the extracellular face would probably contribute to the nonmembrane-bound intracellular concentrations in situ. If so, the amount of free NTP should be lowered. This should enhance the functional effect of the maternal-foetal difference in red cell organic phosphate con-centrations.

High concentrations of organic phosphates are associated with high concen-trations of protons in the cell; the relatively low intracellular pH exists because of the Gibbs-Donnan equilibrium (Duhm, 1971,1976). Since there were appreciably higher organic phosphate concentrations in maternal than in foetal red cells, this should result in lower intracellular pH values in maternal than in foetal red cells. In the absence of ATP, T. elegans haemoglobin was relatively insensitive to changes in pH. However, in the presence of ATP, the haemoglobin was much more sensitive to pH (Fig. 6). Therefore, the higher maternal than foetal red cell organic phosphate levels may promote the lower oxygen affinity of maternal red cells via the Bohr effect in the presence of organic phosphate.

MCHC values of maternal and foetal T. elegans red cells were significantly different (Table 2). Such a maternal-foetal difference has also been reported for the teleost, Embiotoca lateralis (Ingermann & Terwilliger, 1982) and the cotton-mouth snake, Agkistrodon piscivoras (Birchard et al. 1984). A decrease in MCHC has been correlated with an increase in red cell oxygen affinity apparently due to a lessened haemoglobin-organic phosphate interaction (Forster, 1972; Sinet et al. 1976; Lykkeboe & Weber, 1978). Since the foetal T. elegans red cell had a lower MCHC than did the maternal cell, the foetal red cell should have a higher affinity because of this difference. This should enhance the difference in oxygen affinity between maternal and foetal bloods.

Although methaemoglobin does not reversibly bind oxygen, it does tend to increase the oxygen affinity of the other, nonmethaemoglobin subunits (Darling & Roughton, 1942; Enoki et al. 1969). Therefore, methaemoglobin tends to increase the overall oxygen affinity but it decreases the oxygen-carrying capacity of the blood. Several reports have suggested that reptilian red cells possess exceptionally high proportions of methaemoglobin (Dessauer, 1970; Wood & Lenfant, 1976; Pough, 1977). This interpretation has been challenged by Gruca & Grigg (1980). As the amount of methaemoglobin in maternal and foetal T. elegans bloods could affect maternal-foetal oxygen transfer, methaemoglobin levels were quantified. Levels of methaemoglobin were relatively low and similar in foetal and adult blood samples (Table 2).

The oxygen affinity difference noted by Manwell (1960) between maternal and foetal T. elegans red cell suspensions does not appear to be due to different maternal and foetal haemoglobins. Rather, it is likely to be due primarily to different red cell concentrations of ATP, and possibly also to different MCHC values. In this respect, the molecular strategies underlying maternal-foetal oxygen transfer in T. elegans are similar to those of the viviparous snake A. piscivoras (Birchard et al. 1984). As ATP levels in nucleated red cells respond quickly to oxygen availability (Greaney & Powers, 1977; Ingermann et al. 1983), this set of molecular strategies may be particularly well-suited to allow the blood of the T. elegans neonate rapidly to assume functional characteristics of adult blood and thus facilitate the onset of independent life.

We are grateful to Randy Balice for help with statistical analyses and Dr Ralph Yount and Carla Robertson of Washington State University for HPLC analysis of red cell ATP to GTP ratios. This research was supported by the Research Council, University of Idaho.