Previous studies of the Hpa I cleavage site-sickle cell hemoglobin gene linkage in various African populations suggested that the sickle gene arose independently more than once. In the present study we have performed restriction endonuclease haplotype analysis for the beta-globin-like gene cluster from four separate geographic areas in Africa, all of which possess the sickle gene. In Benin (Central West Africa) and Algeria (Arab North Africa) all chromosomes carrying the sickle gene possess an identical haplotype as defined by 11 different polymorphic restriction endonuclease sites within the 60-kilobase region of the beta-globin-like gene cluster. In the Central African Republic (Bantu-speaking Africa) and in Senegal (Atlantic West Africa) a very large proportion of the sickle gene chromosomes were associated with a haplotype specific for each country. Thus, three different haplotypes are shown to be associated with the sickle gene in Africa, and each is present at a very high frequency in geographically separate regions. Since the three haplotypes differ from each other by at least three sites residing both 5' and 3' to a putative hot spot for recombination, it is most likely that the sickle gene arose at least three times on separate preexisting chromosomal haplotypes. This may have implications for a better understanding of the variable nature of the expression of sickle cell anemia, because clinically relevant sequences (for example, gamma-globin gene regulatory sequences responsive to anemia) might be linked polymorphically to these haplotypes.

Patients with sickle cell anemia vary in the hematologic and clinical features of their disease, in part because of variability in the presence of linked and unlinked genes that modify the expression of the disease. The hemoglobin S gene is strongly linked to three different haplotypes of polymorphic endonuclease-restriction sites of the beta-like gene cluster (genes in the vicinity of the beta-globin gene)--one prevalent in Atlantic West Africa, another in central West Africa, and yet another in Bantu-speaking Africa (equatorial, East, and southern Africa). We have studied the differences in the hematologic characteristics of patients with sickle cell anemia from the first two geographical areas. We find that the Senegalese (Atlantic West Africa) patients have higher levels of hemoglobin F, a preponderance of G gamma chains in hemoglobin F, a lower proportion of very dense red cells, and a lower percentage of irreversibly sickled cells than those from Benin (central West Africa). We interpret these data to mean that the gamma-chain composition and the hemoglobin F level are haplotype linked and that the decrease in the percentage of dense cells and irreversibly sickled cells is secondary to the elevation in the hemoglobin F level. Patients with sickle cell anemia in the New World probably correspond to various combinations of these types, in addition to the still hematologically undefined haplotype associated with sickle cell anemia in the Bantu-speaking areas of Africa.

We have studied 42 homozygous 3-thalassemia patients from Algeria and 34 sickle cell anemia patients from Senegal and Benin, determinin* the relationship between haplotypes, Hb F, and G7yglobin/ -globin ratios. Populations selected have a high frequency of haplotype homozygotes because of consanguinity (Algeria) and geographic homogeneity (West Africa). We find in (-thalassemia In addition, haplotypes IX, m, and Senegalese sickle cell anemia patients exhibit hematological amelioration of their disease. Conversely, haplotypes I, V, and A in thalassemia pa-

We have previously found that the inhibitory effect of hemoglobin F (Hb F) on the polymerization of Hb S proceeds via the formation of asymmetrical hybrid tetramers of the type a2ts. Examination of the gelling properties of binary mixtures of Hb S and several Hb variants now shows that, among the y chain amino acid residues that differ from those of the P chain, residues y80 (EF4) and 'y87 (F3) are at least partly responsible for this inhibition. Furthermore, we find that mixing Hb A2(a2h2) with Hb S strongly inhibits gelling to an extent similar to that seen with Hb S/Hb F mixtures; this inhibition is attributable to amino acid differences between the 5 and P chain sequences at positions 522 (B4) and 587 (F3). Therefore, residues 22, 80, and 87 of the f chain appear to be involved in intermolecular contact sites that stabilize the deoxy Hb S polymers. The capacity of some Hb variants to facilitate or impair the polymerization of Hb S is now well known (1). It has long been recognized that these effects correlate with the clinical severity of genetic variants of sickle cell disease in which two major Hb components coexist in the erythrocytes (2). Our laboratory has developed the use of binary Hb mixtures (Hb X + Hb S) to map the residues involved in the interactions between Hb S tetramers in the polymer (3). The inhibitory effect of Hb F on the polymerization of Hb S has been well documented (1). More recently, it has been shown that this inhibition depends on the formation, in Hb S/Hb F mixtures, of asymmetrical hybrid tetramers of the type a2lS3y. Such an inhibitory effect did not occur with the formation of a2t3S#fA hybrids (4,5). Thus, the replacement of one of the Os chains of Hb S by a y chain appeared to have a strong inhibitory effect on polymerization. In the present article we will define part of the structural basis of this inhibition. In addition, we report a prominant inhibitory effect of Hb A2, on polymerization in mixtures with Hb S and provide data concerning the structural basis of this inhibition. METHODSHemolysates were made from blood' samples obtained after informed consent from donors with sickle cell anemia (SS) documented in our Heredity Clinic. The erythrocytes from these donors contained less than 5% (mean) Hb F and were used without further purification. (MGCs) were determined at 240C as described (6). It should be noted that we perform this assay in a manner such that the MGC values closely approximate the values we obtain by measuring the supernatant Hb concentration after highspeed centrifugation of the gel: the Hb mixture is fully deoxygenated for at least 40 min before its concentration is allowed to rise very slowly (by the evaporation of the solvent), and the determination is considered reliable only if the final Hb concentration is less than 2 g/dl higher than the starting concentration and if the total duration of the procedure from the time evaporation is begun is at least 30 min but less than 2 hr. For each mixture, repeat MGC determinations differed from the mean...

Our data support the efficacy of HU in reducing painful events in children with sickle cell disease. Short- and middle-term tolerances are good. Thus, we think that HU can be given to children affected by frequent and severe painful crises. We recommend, however, very cautious use of this drug, because its long-term effects in children are still unknown.