HbS-Oman Heterozygote: A New Dominant Sickle Syndrome

SICKLE TRAIT red blood cells (RBCs) usually do not sickle at pH 7.4 and the peripheral blood of carriers does not contain dense cells, irreversibly sickled cells, or exhibit increased reticulocyte counts. Furthermore, sickle trait individuals do not have splenomegaly. Hence, it is of particular interest that RBCs containing about 20% (the balance being hemoglobin [Hb] A) of a super polymerizing HbS, which is the result of two mutations (β^S-Oman = β6 Glu → Val; β121 Glu → Lys) in the β-chain, have been observed to contain bizarrely deformed sickled cells (ISCs) in peripheral blood.1 

The increased sickling tendencies of HbS-Oman is the consequence of the pro-sickling effects of the β121 mutation, Glu → Lys, which is also found as a single mutation in HbO_ARAB. The latter hemoglobin, when combined with HbS, produces a syndrome more severe than SC, but easily confused with SC because of the same electrophoretic pattern on cellulose acetate.

Purified S-Oman has a C_SAT (solubility of the deoxy polymer) of 11 g/dL, much lower than HbS alone (C_SAT = 17.8 g/dL).1 This result is interesting because another double mutant (Hb S-Antilles = β6 Glu → Val; 23 Val → Ile), has a similarly low C_SAT and much higher expression (40% to 50%) in the trait form, but has a phenotype that is similar in intensity to the trait form of HbS-Oman.2 Hence, the identical intrinsic tendency to polymerize of these two super HbSs is in contradiction with the phenotype: S-Oman has the same effect as S-Antilles, but with half the level of expression. We need to look for other pathogenic properties of S-Oman that must be cellular in character.

In this report we examine the phenotype of S-Oman trait, in a larger number of individuals, and in more detail than in the original description of this abnormal Hb.1 We have also studied the interaction of this abnormal Hb with the RBC membrane and explored its capacity to increase the density of reticulocytes and to produce dense cells.

Blood counts (RBC, white blood cell [WBC], platelet count, mean cell volume [MCV], mean cell hemoglobin concentration [MCHC], and mean cell hemoglobin [MCH]) were performed with a Technicon 1 (Tarrytown, NY). HbF levels were determined by alkaline denaturation and agar electrophoresis with the standard methods. Deoxygenation, SEM, and TEM were performed also as described elsewhere.3 

C_SAT values were determined in a 0.1 mol/L potassium phosphate buffer, pH 7.35, at 25°C. The hemolysate was first deoxygenated with alternate vacuum and N₂ and then Na dithionite was added to three times the final concentration of Hb. Samples were allowed to gel overnight at 25°C and were centrifuged the next day at 25°C for 2 hours at 35,000 rpm. Purified HbS in the same buffer was run as a control. The supernatants were removed anaerobically and Hb concentrations and deoxy pHs were determined.

The methods involved were conducted as described elsewhere.3 

RBCs were lysed with distilled water and the sample was centrifuged to obtain supernatant 1, following the technique of Klipstein and Ranney.4 The RBC membranes obtained after removing supernatant 1 were extracted with an 8.6 pH buffer solution. The supernatant obtained after centrifugation was labeled supernatant 2. Both supernatants where analyzed by isoelectric focusing and HPLC.

“Leaky” RBC membranes (ghosts) were prepared from a hematologically normal (HbAA) individual as follows: RBCs were pelleted from whole blood by centrifugation at 500g for 10 minutes at 4°C, and washed three times with 5 mmol/L sodium phosphate, pH 7.4, 140 mmol/L sodium chloride (PBS). Each time the small buffy coat was carefully discarded. The washed RBCs were lysed with 5 mmol/L sodium phosphate, pH 8.0 (5P8) and the RBC membranes (ghosts) were collected by centrifugation at 15,000g for 15 minutes at 4°C. The membranes were washed with 5P8 until white. Residual Hb and glycolytic enzymes were stripped by incubating the ghosts with 20 vol of 10 mmol/L phosphate, pH 7.5, 0.5 mol/L sodium chloride for 30 minutes on ice. The ghosts were collected by centrifugation, as described above, and washed once with 10 mmol/L sodium phosphate pH 6.0 (10P6). Under these conditions, the ghosts do not reseal and are thus “leaky.” An aliquot was taken for protein determination using the BCA method (Pierce Chemical Co, Rockford, IL) after solubilization in 1% sodium dodecyl sulfate (SDS).

The ghosts (1 mg/mL final concentration) were incubated with purified HbAA or HbS-Oman (20 mg/mL final concentration) in 10P6 (final concentration) for 90 minutes at 4°C in a total volume of 0.2 mL while rotating. The incubations were run in triplicate. After the incubation, ghosts were collected by centrifugation, as described above, and washed vigorously four times with ice-cold 10P6. After the final wash, the ghosts were resuspended in 0.2 mL ice-cold 10P6 and 0.04 mL was removed for SDS-polyacrylamide gel electrophoresis (SDS-PAGE) analysis (see below). The remainder was solubilized by addition of 2 vol of 20 mmol/L sodium phosphate, pH 7.4, 125 mmol/L sodium chloride, 1.5% Triton X-100. Hb in the solubilized ghosts was measured using Drabkin’s reagent (Sigma, St Louis, MO), and compared with an Hb standard. Hb binding results were expressed as moles of Hb dimer bound per mole of ghost band 3. It was assumed that band 3 represents 30% of ghost total protein and has a molecular weight (MW_r) of 90 kD.

Ghosts, 0.02 mg, were solubilized and the proteins separated by SDS-PAGE on 7.5% acrylamide gels using the Fairbanks continuous buffer system. The gels were stained with Coomassie brilliant blue, and the protein bands were quantitated by scanning laser densitometry.

RBCs were either used for determination of density distribution or separated by density gradient centrifugation into three fractions on a Percoll (colloidal silica coated with polyvinyl pyrrolidone; Pharmacia Fine Chemicals, Inc, Piscataway, NJ) and Larex (arabinogalactan polysaccharide; Larex International Inc, Roseville, MN) gradient as previously described.5,6 

The maximal activity of the Ca²⁺-activated K⁺efflux was determined under optimized conditions in the presence of the calcium-ionophore A23187 as previously described by Romero et al7 in the presence and absence of 10 μmol/L clotrimazole (CLT). CLT has been shown to be a potent, high-affinity inhibitor of Ca²⁺-activated K⁺ channels (Gardos Channel) in human RBCs.8 The ionophore A23187 was used to increase cytosolic ionized Ca²⁺ and clamp it at various levels. Net K⁺ efflux was determined by incubating cells in an Na⁺ media containing the A23187 (60 μmol/L of RBCs). The Na⁺ media contained (mmol/L): 140 NaCl, 10 Tris-MOPS pH 7.4 at 37°C, 0.1 ouabain (to inhibit the Na⁺ pump), 10 glucose, 0.15 MgCl₂, 0.1 bumetanide (to inhibit Na⁺:K⁺:2Cl⁻ cotransport), 1 EGTA or 1 Tris-citrate; and total CaCl₂ varied between 0 and 100 μmol/L, A23187 (60 μmol/L of RBCs), at 1% hematocrit. This gives a cytosolic ionized Ca²⁺ that is between <0.1 nmol/L Ca²⁺ with 1 mmol/L EGTA and 20 μmol/L Ca²⁺ with 1 mmol/L citrate. Cytosolic ionized Ca²⁺ was buffered at <0.1 nmol/L Ca²⁺ with 1 mmol/L EGTA or at ≥5 μmol/L Ca²⁺with 1 mmol/L citrate. Initial rates of K⁺ efflux in RBCs were measured by sampling the efflux media in duplicate at 1, 3, 5, and 7 minutes in the presence or absence of 10 μmol/L CLT. In experiments where the Ca²⁺ was reduced to <0.1 nmol/L, the samples were taken at 1, 3.5, and 7 minutes. K⁺ efflux was calculated from the slope of the regression line of K⁺concentration versus time taking into account the volume of RBCs used. The slope ± SE of the linear regression was calculated using Enzfitter software for the personal computer. The CLT-sensitive Ca²⁺-activated K⁺ efflux was estimated by subtracting the flux in the presence of CLT from the total flux at various cytosolic calciums. The CLT-sensitive K⁺efflux was then analyzed as a function of the cytosolic ionized calcium using a nonlinear regression analysis. This allowed for estimation of the V_max and K_m for calcium stimulation of Ca²⁺-activated K⁺ efflux.

The total calcium concentration of the flux media was measured by atomic absorption spectrophotometry using calcium standards (EM Sciences, Cherry Hill, NJ) in sodium media. A custom-made computer program was used to calculate the Ca²⁺ concentrations using the dissociation constant and correcting for ionic strength at pH 7.4 and 0.15 mmol/L MgCl₂ as previously described.8CLT was obtained from Sigma (St Louis, MO).

The protocol followed was identical to the ones described in Romero et al.7 

The sample consists of three individuals with 20% to 23% Hb-S Oman (HSR, MSR, ISH) and two with levels of 13% and 14% (SSR, MSK) (Table 1 and pedigree in Fig 1). In the high S-Oman group all have silent carrier status for α-thalassemia, and correspondingly, they have moderately reduced MCV and MCH (70 to 76 fL and 23 to 25 pg). The second group, with low HbS-Oman expression, have concomitantly −α/−α status, and hence significantly lower MCV and MCH (66 to 67 fL and 68 fL and 21.4 to 21.7 pg). The level of anemia was also different between these groups: individuals with higher S-Oman had lower Hb than those with low levels of S-Oman and tended to have enlarged spleens. Correspondingly, reticulocyte counts tend to be lower in the latter than in the former. Nevertheless, the patient ISH is an exception, because she had a low reticulocyte count and the spleen was not palpable, probably due to young age. Finally, the HbF levels are only mildly to slightly elevated.

Clinical records of three patients with high S-Oman expression show the following: MSR, 11 years old, had frequent painful crises and episodes of hypoxic encephalopathy and acute chest syndrome, qualifying for a severe phenotype; ISH, 6 years old, has frequent but mild painful crises; a third patient, HSR, 18 years old, with 22.3% S-Oman, required frequent transfusions between the ages of 3 and 10 and has had mild painful crises afterward. MoSR, a patient not included in the present series, 23 years old, 27% S-Oman, has had acute painful crises, acute chest syndrome, and episodes of pain in the shoulders. The two patients with 14% S-Oman (Table 1), plus another patient, female, 45 years old and 13.3% S-Oman (ZSK), report no clinical symptomatology.

Figure2 depicts the isopycnic gradients for the five carriers. All of the patients had more light cells at the top of gradient, a consequence of the presence of α-thalassemia and hemolysis. The percent dense cells, compared with an AA subject, were increased in two of the high S-Oman expressors and less in the other three individuals. Not surprisingly, the exception was ISH, who was also different because she did not have a high reticulocyte count or a palpable spleen, and was only 6 years old. Dense cells are greatly decreased in early childhood sickle syndromes, most likely due to the presence of a relatively efficient spleen.

The isopycnic gradients were also used preparatively, which allowed separation of the cells into fractions 1, 2, and 3 of increasing density, which were then analyzed for the percent reticulocytes and irreversibly sickled cells. Figure 3depicts the percent reticulocytes as a function of increasing RBC density in both a low S-Oman patient (SSR; 14% S-Oman) and a high S-Oman patient (MSR; 20% S-Oman). Interestingly, the highest percentage in SSR is in the lightest fraction (F1), while in the high S-Oman patient, reticulocytes were located preferentially in the densest fraction (F3).

The most striking feature of the smear of the higher expressors of HbS-Oman was the presence of cells which appeared to be the consequence of a single polymerized domain of HbS-Oman traversing the RBC. Although the level of dehydration of these cells varied, they most frequently had little or no dehydration (Figs 4 [see page 4378] and 5). They suggest a “yarn/knitting needle” shape. These forms were very rare or absent in the blood of S-Oman patients with less than 15% S-Oman. These cells, when oxygenated, are devoid of polymer but retain their shape, qualifying for the name irreversibly sickled cells. When deoxygenated, they formed single or double domains, suggesting the presence of single or double nucleation events (Fig 6). Near the membrane, bunches of polymer were observed (Fig 6).

Table 2 reports on the C_SAT of hemolysates of SSR, a low HbS-Oman expressor, and ISH, a high S-Oman expressor, which are significantly different. The patient SSR, who is asymptomatic and has 14% S-Oman, has a C_SAT significantly higher than sickle trait (an increase of ≈4 g/dL) that is similar to 50% mixtures of HbF and HbS, a fact that explains the lack of a pathological phenotype. The results on ISH are particularly interesting: the C_SAT is very close to that of sickle trait. In other words, the phenotype is more severe than the C_SAT predicts. Hence, we have to search for the nonpolymerization-dependent effect of HbS-Oman to explain the phenotype.

Hb retention was studied by comparing the Hb composition of supernatants 1 and 2 (Fig 7). Supernatant 1, generated by distilled water lysis, had percents of HbA and HbS-Oman, similar in proportion to the regular hemolysate. Supernatant 2, obtained by extracting the Hb left in the RBC ghosts after distilled water lysis, was enriched in HbS-Oman. For the patient with 15% S-Oman in the regular hemolysate, quantification of the bands in supernantant 2 (representing the Hb attached to the membrane after distilled water lysis) (Fig 5) gave a percent of HbS-Oman of 73.2% and HbA of 17.3% compared with 14.1% S-Oman and 57% HbA in supernatant 1. The other sample, from a carrier with 20% HbS-Oman in the regular hemolysate, had a very similar distribution, with S-Oman 69.4% and HbA 20.7% compared, numbers that are quite different from the distribution in supernatant 1 (20.8% S-Oman and 58.4% HbA). These numbers do not add up to 100% because IEF separates also Hbs A_1c, S-Oman_Ic, F, and A₂, not relevant to this analysis.

In two separate experiments we compared the binding of HbS-Oman and HbA to RBC ghosts. The results of one of these experiments are depicted in Table 3. The second experiment gave the same results: there is a threefold increase in binding of HbS-Oman to “leaky” ghost compared with HbA. In addition, gel electrophoresis confirmed the significantly increased binding of HbS-Oman to these ghosts (Fig 8 [see this page]).

Human RBCs possess a Ca²⁺-activated, charybdotoxin (CTX) sensitive K⁺ efflux8-10 that is also sensitive to CLT.8 Experiments on RBCs from an SS patient showed that the presence of 20 μmol/L cytosolic Ca²⁺ and A23187 induced a maximal stimulation of the K⁺ efflux (4.27 ± 0.57 mmol/L cell × min). This efflux was reduced to 0.16 ± 0.19 mmol/L cell × min when the RBCs were incubated with 10 μmol/L CLT in the presence of 20 μmol/L cytosolic Ca²⁺and A23187. When cytosolic Ca²⁺ was buffered at <0.1 nmol/L with EGTA, in the presence of A23187, the K⁺ efflux was significantly reduced to 0.07 ± 0.14 mmol/L cell × min. The difference between K⁺ efflux in the presence and absence of CLT yields the Δ-CLT sensitive fraction, which in this experiment was 4.11 mmol/L cell × min. Therefore, the percent inhibition of the total maximal activity of the Ca²⁺-activated K⁺ efflux was approximately 99% for this experiment.

The kinetic parameters of the CLT-sensitive Ca²⁺-activated K⁺ efflux measurements in RBCs from SS and heterozygous S-Oman patients is of interest. In both cell types, the Ca²⁺-activated K⁺ efflux was already maximally stimulated by approximately 10 μmol/L cytosolic calcium. In the presence of CLT, Ca²⁺-activated K⁺ efflux values were not significantly different in both cell types. Therefore, the ΔCTX-sensitive K⁺ efflux was not significantly different in these cells. However, the K_m for stimulation of the Ca²⁺-activated K⁺ efflux was significantly lower in S-Oman than in SS RBCs. The RBCs from HSR (S-Oman 22%) had a K_m of 0.487 μm and a V_maxof 6.22 FU while the RBCs of MSK (S-Oman 13%) had a K_m of 1.81 μm and V_max of 6.05 FU. This is in clear contrast to the values of K_m 7.53 μm and 5.5 FU obtained for SS blood.

Three patients were studied: two with less than 14% S-Oman (RHK, SSR) and one with 20% HbS-Oman (MSR). The volume-stimulated K:Cl cotransports were 5.6 and 1.2 mmol/L cell × h, respectively, and the NEM-stimulated K:Cl cotransports were 25.6 and 24.0 mmol/L cell × h in the low S-Oman RBCs. The high S-Oman RBCs exhibited a volume stimulated K:Cl efflux of 14.7 mmol/L cell × h and an NEM-stimulated K:Cl cotransport of 56.7 mmol/L cell × h. For comparison, the average values for the Cl-dependent part of volume stimulated K:Cl cotransport for AA, SS, and CC cells are 2.5 ± 0.5, 15.8 ±1.5, 17.3 ± 1.5 mmol/L cells × h, respectively. The average value for NEM stimulation for SS RBCs has greater variability (32.7 ± 10.6 mmol/L cells × h).

The pedigree presented here corresponds to a new sickle syndrome characterized by the presence of anemia, high reticulocyte count, the presence of “yarn/knitting-needle” shaped RBCs, splenomegaly, painful crises, and, in one case, central nervous system (CNS) symptoms and acute chest syndrome. Because all the symptomatic individuals studied were concomitantly heterozygous for the African-type α-thalassemia (−α/αα), due to the extremely high incidence of α-thalassemia in Oman,11 we suspect that we are describing a milder form of this syndrome than would be the case in the absence of α-thalassemia. The clinical syndrome is only observed in individuals with hemolysates having more than 20% HbS-Oman. Individuals with a lower percent of the doubly mutated HbS (around 14%) and concomitant homozygosity for the African-type α-thalassemia (−α/−α) are asymptomatic clinically and hematologically have only slight to moderate RBC changes.

The failure of the original investigators who first described HbS-Oman1 to recognize the clinical syndrome stems from the fact that they had access to a single carrier who happened to have a concomitant −α/−α haplotype and, hence, a low expression (14%) of S-Oman.

Several factors contribute to the pathogenic effect of this double mutation. It is clear that the second mutation (β121 Glu → Lys) has an enhancing effect on polymerization, as shown by the C_SAT of the hemolysate (not S-Oman purified) of a symptomatic patient, which is the same as that of a sickle trait (AS) despite the relatively low expression of 22% for S-Oman compared with 40% HbS in sickle trait. S-Oman shows that the effect of β121 Glu → Lys is also apparent in cis (that is, cis to the active β6Val site).

Nevertheless, the pro-polymerizing effects of β121 (Glu → Val) cannot entirely explain the phenotype. Pure S-Oman has a C_SAT of 11 g/dL, which is almost identical to the 11.8 g/dL C_SAT of S-Antilles, another super-hemoglobin S that is the product of a double mutation in the β-chain. The moderate sickle cell anemia phenotype of S-Antilles trait is the consequence of the enhancing effect of the β23 Val → Ile mutation on polymerization and a yet undefined effect of the right shift of the oxygen equilibrium curve observed both in HbS-Antilles as well as the recombinant hemoglobin α₂β₂^{23Val → Ile}.2,12 Considering the differences in expression between the two, of almost 40% to 50% for S-Antilles compared with the 20% to 22% of S-Oman, the phenotype of the latter should be quite mild or nonexistent. In fact, the phenotypes are very similar. Finally, the hemolysate C_SAT’s lend further support to this analysis. These features suggest that there is an additional pathogenic effect due to the presence of the β121 (Glu → Lys) mutation.

We propose here that the β121 mutation, by increasing the charge difference by three, produces an abnormal interaction of the HbS-Oman tetramers containing the doubly mutated β-chain with the RBC membrane. A strong independent confirmation of this phenomena is the presence of moderate hemolytic anemia in the homozygotes of HbO_ARAB.12,13 Abnormal interaction with the membrane has been described for HbC (a two-charge difference)4,4,15 and for HbA₂ (a two-charge difference).4,14 Evidence for an abnormal interaction of S-Oman with the membrane can be summarized as follows:

(1)

The presence of dense reticulocytes, of the type seen only in individuals with SC and CC genotypes,16 which is the result of abnormal volume regulation of the reticulocytes leaving the marrow.

(2)

The significant decrease in the K_m of the Ca²⁺-activated K⁺ efflux, which suggests that this important channel requires less Ca²⁺ than it does in SS RBCs to extrude K⁺. The K:Cl cotransport activity seems to follow the extent of S-Oman expression; hence it seems to be sickling dependent, but a contribution of the nonsickling hemolysis induced by the second mutation might be operating through the increase in reticulocytes. Nevertheless, the effect of α-thalassemia needs to be further investigated.

(3)

The increased association of HbS-Oman, over HbA and HbS, with the RBC membrane, and the over-representation of HbS-Oman in membrane extracts. In addition, binding of HbS-Oman to normal “leaky” Hb-stripped ghosts is increased threefold to fourfold compared with HbA.

(4)

Presence of dense cells by isopycnic gradients.

(5)

The presence of abundant nidus of sickle polymerization.

(6)

Generation of folded RBCs similar to those found in CC, SC, SO_ARAB, and homozygous HbO_ARAB..13,16 

The presence of “yarn/knitting needle” cells, which is characteristic of the syndrome described here and most likely the consequence of the polymerization of S-Oman in a single domain, is limited to patients expressing HbS-Oman at the 20% to 22% level. Electron microscopy of in vitro deoxygenated S-Oman RBCs shows that cells already deformed to the “yarn/knitting needle” shape, due to previous events of polymerization that surpassed the yield coefficient of the membrane, exhibit one or two domains of polymerization. This represents a low number of nucleation events, which is not surprising considering the relatively low concentration of HbS-Oman.

We conclude that the pathophysiology of the sickle syndrome S-Oman heterozygote is the result of the following gene-product effects: (1) The sickle mutation (β6 Val) induces sickling; (2) The β121 Lys mutation enhances sickling; and (3) the β121 Lys mutation induces hemolysis and RBC shape changes.

We know that Hbs with increased positive charge (HbC, HbO_ARAB) are by themselves15-17 capable of reducing the volume of the RBC (with corresponding increase in MCHC), reducing the RBC life span, and producing morphological changes (folded cells). In the homozygous form, both of these Hbs produce hemolytic anemias.16,17 S-Oman heterozygotes have folded cells and reduced RBC life span, very much like CC and SC cells, although they do not have the uniformly increased density18 of these genotypes.

Finally, there is the matter at the low percentages of S-Oman found in the heterozygotes describe here. The difference in percentage of S-Oman are associated with the degree of α-thalassemia and the numbers are still low in absolute terms, suggesting that other factors are involved. Message instability or transcription abnormalities might be present and need to be excluded. Nevertheless, another interesting possibility is that the assembly of HbS-Oman is decreased by the very high positive charge of the β^S-Oman, as Bunn19 has suggested for other positively charged Hbs (C,E,O_ARAB).

We postulate that the pathogenic mechanisms mediated by S-Oman, in addition to the facilitation of Hb polymerization, include a direct or indirect charge-dependent interaction between S-Oman and the RBC membrane, resulting in changes of the intracellular Hb concentration or increased cell density, by altering transport mechanisms and increasing RBC destruction.

We acknowledge the help of Nazim Fataliev’s competent purification of hemoglobin S-Oman, and to all those in the Sultan Qaboos Royal Hospital who helped in securing many valuable samples.