INTRODUCTION

-globin gene mutations are of no clinical consequence because Hb¹ A₂ (₂₂) comprises about 2.5% of total adult hemoglobin. These mutations can, however, mask the presence of  thalassemia, which is usually characterized in heterozygotes by low mean cell volume and high Hb A₂. Compound heterozygotes for  thalassemia and a -globin gene mutation may have low mean cell volume and normal Hb A₂ levels and therefore be overlooked as -thalassemia heterozygotes. Identification of these mutations is important in areas where the thalassemia syndromes are prevalent, such as the Mediterranean populations (1, 2, 3).

As with the -thalassemia syndromes, both ⁺- and ^o-thalassemia phenotypes have been observed; the former due to partial and the latter to total suppression of -chain synthesis. In addition, unstable -chain variants may mimic a -thalassemia phenotype, and instability of -globin might also lead to ⁺ and ^o phenotypes.

The molecular basis of  thalassemia has been clarified in only nine instances (4). These include: (a) a transcriptional mutant (5) with a T  C change at a position 77 bp 5 to the cap site that abolishes a GATA-1 binding site (6); (b) three RNA processing mutations including a G  T change at codon 27 Ala  Ser (7), a G  C change at the second position of codon 30 (8), and a T  C substitution at the 5-donor splice site of IVS-1 (7). The G  T change at codon 27 Ala  Ser of the -globin gene activates a cryptic splice site resulting in aberrant transcript processing of some of the pre-mRNA precursors (9). This mutation in the -globin gene produces a variant hemoglobin, Knossos (10). The G  C change at codon 30 interferes with normal -globin mRNA splicing of IVS-1 because it changes the splice consensus sequence from CAGgttggt to CACgttggt. The T  C substitution in the 5-donor splice site of IVS-1 abolishes the splice donor site by changing the invariant GT nucleotide and leads to complete absence of -globin mRNA; (c) two frameshift mutations that include deletion of A at codon 59, which leads to a premature stop at codon 60 (11, 12), and insertion of T at codon 91, which results in a premature stop at codon 94 (13); (d) two unstable hemoglobin variants hemoglobin A₂-Wrens or ₂₂ 98(FG5) Val  Met (14) and hemoglobin A₂-Manzanares or ₂₂ 121(GH4) Glu  Val (15); and, finally, (e) a 7.2-kilobase deletion that starts in the - intergenic region and ends in the IVS-2 of the -globin gene (16, 17).

Twenty-one -globin variants have been characterized so far (18), and all appear to be stable except for the codon 98 and 121 variants. Unstable hemoglobin variants caused by single amino acid replacements within critical regions of the - or -globin chains can lead to congenital Heinz body hemolytic anemia. Precipitated unstable hemoglobins form Heinz bodies that associate with membranes and lead to premature red cell destruction (19). Unstable -globin variants are of little clinical importance because Hb A₂ is such a minor portion of the total hemoglobin.

Mutations that result in globin instability either affect structure of globin and heme subunit or hemoglobin tetramer. Loss of interactions stabilizing structure, disruption of subunit interactions, changes affecting hydrophobic interior, and perturbing secondary structure, such as introduction of proline in a helical region, all can create unstable hemoglobins.

Other mutations in the -chain may not necessarily result in unstable hemoglobins but may promote intermolecular and/or intramolecular disulfide bond formation and thereby decrease -chain levels available for assembly with -chains. Other possibilities include mutations that promote tetramer interaction with red cell membranes and/or those that lead to decreased affinity for assembly with -chains.

In this report, we attempt functional characterization of five -globin gene alleles in the Greek Cypriot population that are associated with reduced Hb A₂. These mutations could affect splicing, such as the G  T change at codon 27 and the AG  GG change at the 3-acceptor of IVS-2, or result in formation of structural variants such as the G  T at codon 27 Ala  Ser, the C  T at codon 116 Arg  Cys, and the T  C at codon 141 Leu  Pro. Finally, we were also interested in defining which of the following changes, a C  T at codon 4 Thr  Ile, a C  T at codon 97, or the AT deletion in IVS-2 found in the same -globin gene, is responsible for reduced Hb A₂.

EXPERIMENTAL PROCEDURES

Oligonucleotides were prepared using phosphoramidite chemistry on a 380B DNA synthesizer (Applied Biosystems, Inc., Foster City, CA) (Table I).

Table I. Oligonucleotide primers for PCR and mutagenesis Bold letters designate mutated nucleotides. N, M, A, or S at the end of primer name designates normal, mutant, antisense, or sense, respectively. Primer Position/name Sequence 5  3 4   +322 to +341 TCTGTCCTCTCCTGATGCTG 5  +758 to +777 TGCATACCAGCTCTCACCTG 6  +845 to +864 CAGTATTCTATGCCTCTCAT 8 +1685 to +1704 CAGGAACCTTCTTACACACC 21 XhoIsNS CCGTCGACCTCGAG 22 pGS1882A AGTCAGATGCACCATGGTTTATTTATGTGTGTTTATTC 23  cDNA5S ACACATAAATAAACCATGGTGCATCTGACTCCTGAG 24  cDNA3A GGGAACAAAGTCGACTCAATGGTACTTGTGAGCCAG 25 pGS188S GTCGACTTTGTTCCCACTGTAC 26 Xho3flNA CCGGGGGGCTCGAGGGGAATTCTCTTAGGATTC 27 Tcd27-MA TGCCCAGGGACTCACCACC 28 Tcdn97MA CAGGATCCACATGCAGCTTGT 29 ATIVS2MA GCATACATATAACATAT--CTATACACACAC 30 GIVS2-MA CCCAAGAGCCGCGGAGAAG 31 BLU3820a CTACAGCGTGAGCTATGAGAAAG 32 BLU3800s CCTGTCCGCCTTTCTCCCTTCG 33 Tcdn4-MS ACACATAAATAAACCATGGTGCATCTGATTCCTGAGGAG 34 TCTc27MA ATCTGCCCAGAGACTCACCACC 35 cd27-2NS CTGGGCAGATTACTGGTGGTC 36 Tcd116MS GTGTGTGTGCTGGCCTGCAACTTTGG 37 Ccd116NA GGCCAGCACACACAC 38 Tcd141NS GGTGTGGCTAATGCCCTGGCTCACAAG 39 Tcd141NA GGGCATTAGCCACAC 40 pSVK35S GCTATTCCAGAAGTAGTGAGGAGG 41 GALPRONS CTCTCCTCCGTGCGTCCTC 42 GS1883UN AGTGAATAGCTATATAAAG 43 +415 to +434 AAGTGCCCTTGAGGTTGTCC

Construction and Characterization of Normal and Mutagenized -Globin Gene Clones

The -genomic SalI to PstI fragment was cloned into the polylinker of pSVK3 (Pharmacia Biotech Inc.), pSVK3 5 SalI-PstI, which places the -globin gene under control of the SV40 early promoter.

The Amersham Site-directed Mutagenesis kit (Amersham Corporation, Arlington Heights, IL) was used to construct G  T at codon 27, C  T at codon 97, the AT deletion at position 722 in IVS-2 and AG  GG in the 3-splice acceptor of IVS-2 in the -globin genomic clone using primers 27-30. Mutagenized clones were identified by PCR followed by digestion with appropriate restriction endonucleases (1, 2): primers 40 and 6 followed by EcoO 109 I digestion for codon 27 change; primers 4 and 6 followed by NlaIII digestion for codon 97 change; and primers 5 and 8 with SacII digestion for AG  GG change. The AT deletion does not alter a restriction pattern and therefore had to be confirmed by DNA sequence analysis. Positive clones were reconfirmed using the PRISM Ready reaction dyedeoxy terminator cycle sequencing kit (Applied Biosystems). Sequence was compared with known -globin gene sequence (20, 21) using SeqEd (Applied Biosystems).

Expression and Characterization of -Globin Gene Transcripts

COS-1 (American Type Culture Collection, Rockville, MD) cells were grown in RPMI 1640 medium supplemented with 10% (v/v) fetal bovine serum, L-glutamine, streptomycin, and penicillin in a 5% CO₂ incubator. Cells were transfected with wild type or mutant plasmids using calcium phosphate precipitation as described (22).

RNA was harvested 48 h after transfection using RNAzol B (Tel-Test, Inc., Friendswood, TX). Radiolabeled riboprobe was made using the Riboprobe Gemini in vitro transcription kit (Promega Corporation, Madison, WI). RNase protection assays were performed using the Ambion RPA II kit (Ambion, Inc., Austin, TX), and protected fragments were electrophoresed on a 6% (w/v) polyacrylamide gel.

Details of RT-PCR assays are described elsewhere (23). Primer 24 was used in a reverse transcription reaction using Moloney murine leukemia virus RT (Life Technologies, Inc.) to synthesize cDNA, which was amplified with Taq Polymerase (Perkin-Elmer Corp.) in a PCR reaction using primers 23 and 43 or 4 and 24 for IVS-1 and IVS-2 regions, respectively. PCR products were run on a 2% (w/v) agarose gel. An Argus-50 Image Processor (Hamamatsu, Photonics Corporation, Bridgewater, NJ) was used to quantitate the density of the 88-bp (normally spliced exon 1) versus the 72-bp (aberrantly spliced exon 1) bands in the codon 27 lane of Fig. 1. The results were normalized for radiolabeled CTP content for the 72- and 88-nucleotide RNase-resistant fragments.

Construction of Variant -Globin cDNAs and Expression and Characterization of Variant Hb A₂ Tetramers in Yeast

A yeast system was used for expression of wild type and variant soluble hemoglobins. The expression vector, pGS389, contains - and -globin cDNAs under control of galactose-inducible glyceraldehyde 3-phosphate dehydrogenase yeast promoters (24). The -globin cDNA together with the promoter and yeast 3-flanking sequences were excised from pGS389 as an XhoI fragment and was cloned into pBluescript SK⁺ to create the shuttle vector pGS188 in order to facilitate subcloning and replacement of - with -globin cDNA. Wild type -globin cDNA was made employing RT-PCR with primers 23 and 24 using RNA isolated from COS-1 cells transfected with pSVK3 5 SalI-PstI. PCR fragments of the galactose-inducible promoter (primers 21 and 22), full-length -globin cDNA (primers 23 and 24) and yeast mating type -gene 3-untranslated region (primers 25 and 26) were joined using overlap PCR with primers 21 and 26, and the product was cloned into the TA vector (Invitrogen Corporation, San Diego, CA). DNA sequence of the full-length wild type -globin cDNA was determined, and the plasmid was digested with BstBI and KpnI, which cut several nucleotides upstream of the  ATG and downstream of the yeast mating type -gene 3-untranslated region, respectively. This fragment was inserted into a BstBI- and KpnI-digested pGS188  shuttle vector to create pGS188 , which ensures the glyceraldehyde 3-phosphate dehydrogenase yeast promoter driving expression of -globin cDNAs does not contain any PCR-induced mutations. The XhoI fragment from pGS188  containing the glyceraldehyde 3-phosphate dehydrogenase yeast promoter, -globin cDNA, and yeast mating type -gene 3-untranslated region was inserted into pGS389 (+) following digestion with XhoI to create pGS389 (+). Clones carrying the -globin cDNA insert were identified by PCR using primers 4 and 42. Once the different mutations were constructed in pGS188 (see below), insertion of the corresponding XhoI fragments into pGS389 was then repeated, resulting in construction of wild type and variant Hb A₂ expression vectors.

Homologous PCR recombination was used for mutagenesis (25). The initial wild type -globin cDNA amplified by RT-PCR contained an unanticipated PCR-induced T  C at codon 141 Leu  Pro that was identical to one of the -chain variants. This plasmid was used to generate a bona fide wild type -globin cDNA. Normal primers 31 and 39 were used to amplify one region of pGS188, whereas primers 32 and 38 were used to amplify the other half. These two PCR products have complementary ends and were introduced into maximum efficiency DH5 cells (Life Technologies, Inc.). The complementary ends recombine during growth in bacteria to create the wild type -globin cDNA plasmid, pGS188 . Similarly, primer pairs 31/22 and 32/33 were used to introduce the codon 4 change in wild type -globin cDNA and primer pairs 31/34 and 32/35 for the codon 27 change and primer pairs 31/37 and 32/36 for the codon 116 change. DNA sequence analysis was also done for all plasmids to confirm that these were the only PCR-induced changes using the PRISM Ready reaction dyedeoxy terminator cycle sequencing kit (Applied Biosystems).

Yeast expression vectors containing wild type or mutant -globin cDNAs were electroporated (26) into GSY112 yeast cells (27) using a Cell-Porator (Life Technologies, Inc.). Yeast growth, harvesting, and purification of Hb A₂ and variants were performed as described previously (25). Purified Hb A₂ variants were subjected to electrospray mass analysis (Fisons Instruments, VG Biotech, Altricham, UK) using the multiply charged ion peaks from the -globin chain (molecular mass, 15,126.4 Da) as an external reference for mass scale calibrations (28).

Hemoglobin concentration was determined spectrophotometrically on a Hitachi U2000 spectrophotometer using a millimolar extinction coefficient of 13.4 at 540 nm for carbonmonoxyhemoglobin and 13.5 at 541 nm for oxyhemoglobin (29). Hemoglobin tetramers were electrophoresed on cellulose acetate at pH 8.6 using Super Heme buffer (Helena Laboratories, Beaumont, TX) and were stained with Ponceau S stain (Helena Laboratories, Beaumont, TX). Heat stabilities of the recombinant oxyhemoglobins (at ~40 µM) were determined in 0.1 M phosphate buffer, pH 7.4, by heating 500-µl samples in a heating block at 45 °C for 10-min intervals. Interdisulfide bond formation of Hb R116C was assessed by gel filtration fast protein liquid chromatography employing Superose-12 HR 10/30 using 0.1 M phosphate buffer, pH 7.0, at 4 °C followed by electrophoresis on SDS-polyacrylamide gel (15% w/v) as described previously (30). Mechanical stabilities of the recombinant hemoglobins were measured in 0.1 M phosphate buffer, pH 7.4, at room temperature as described previously using a TCS shaker Model 250 (TCS Medical Products Co., Huntington Valley, PA) (31). After shaking, the cuvette was centrifuged at 3000 rpm for 5 min to remove denatured, insoluble hemoglobin. Concentration of the soluble hemoglobin was determined spectrophotometrically. Oxygen association curves of hemoglobins were determined in 0.1 M phosphate buffer, pH 7.4, at 20 °C using a Hemox-Analyzer (TCS Medical Products Co.) as described previously (32).

RESULTS

Functional Characterization of -Globin Gene Mutations

Having previously identified five -globin gene alleles (1, 2), we wanted to define how each leads to reduced levels of Hb A₂. We were also interested in determining which of the three changes, C  T at codon 4, C  T at codon 97, or the AT deletion in IVS-2, found in the same -globin gene, was responsible for reduced Hb A₂.

Changes were grouped into two categories: (a) those that could potentially affect splicing such as G  T at codon 27, C  T at codon 97, IVS-2 AT deletion, and AG  GG at the 3-acceptor of IVS-2 and (b) those that result in generation of structural variants such as C  T at codon 4 Thr  Ile, G  T at codon 27 Ala  Ser, C  T at codon 116 Arg  Cys, and T  C at codon 141 Leu  Pro.

We used site-directed mutagenesis to introduce G  T at codon 27, C  T at codon 97, the IVS-2 AT deletion, and AG  GG at the 3-acceptor of IVS-2 in the normal -globin gene. These modified -genes were placed in pSVK3, a mammalian expression vector, under the control of the SV40 early promoter, and plasmids were transfected into COS-1 cells to evaluate effects on splicing using RNase protection and RT-PCR assays.

A radiolabeled riboprobe that detects exon 1, IVS-1, and exon 2 -globin gene sequences was used (Fig. 1). Wild type and codon 97 plasmids produce two sets of doublets at ~237 and 88 bp; therefore, the codon 97 change does not alter splicing of IVS-1. On the other hand, the codon 27 plasmid gives, in addition to those bands, smaller sized bands at ~72 bp, consistent with activation of a cryptic splice site near the mutation in -globin mRNA.

RT-PCR was employed to define the cryptic splice site used. COS-1 transfected RNA was reverse transcribed, and the exon 1/exon 2 boundary was amplified by PCR. The 398-bp product present in all lanes (Fig. 2) corresponds to amplification of either contaminating plasmid DNA or unspliced mRNA, both of which contain IVS-1 sequences. The 271-bp product, also present in all lanes, is the result of correctly spliced mRNA lacking intronic sequences. In the codon 27 RT-PCR lane, there is an additional band at 255 bp. This PCR product, which contains the alternatively spliced message, was excised from the gel, and the DNA sequence was determined to define the location of the cryptic splice. Genomic DNA sequence encompassing the codon 27 change in this individual is shown in Fig. 3 with outlined letters starting with GGTG and dotted line splice to ATT ... indicating the sequence of aberrantly spliced message deduced from the PCR product. The G  T change at codon 27 activates a cryptic splice site (GTGGTGAGG to GTGGTGAGT) within -globin mRNA, whose last six nucleotides match perfectly to the donor splice site consensus listed at the bottom. Use of this cryptic splice results in a smaller mRNA containing a frameshift and premature termination at codon 55. Aberrant splicing of this pre-mRNA accounts for about 80% of total mRNA, as determined by densitometric scans of RNase protection patterns.

AG IVS-2

As evident from both RNase protection and RT-PCR, the codon 97 change does not alter splicing around IVS-1. Because this mutation is at the 3-end of exon 2, it could potentially affect splicing of ISV-2. To test this, a riboprobe that detects exon 2, IVS-2, and exon 3 sequences was hybridized to COS-1 transfected RNAs from three different plasmids containing either the codon 97, IVS-2 AT deletion, or AG  GG change in the 3-splice acceptor of IVS-2 (Fig. 4). Wild type, codon 97, and IVS-2 AT deletion plasmids produce expected 258- and 201-bp bands, indicating that these changes do not alter splicing. An additional doublet is present in the codon 97 lane, which most likely represents an artifact, because it appears in wild type and AT deletion lanes upon longer exposure. On the other hand, the AG  GG in the 3-acceptor splice site of IVS-2 results in aberrant splicing because only the 201-bp band is present in this lane. There is no evidence of a larger protected fragment to indicate use of a cryptic splice site in IVS-2. Because the 201-bp band corresponds to hybridization of riboprobe to exon 2 sequences and the 258 bp band corresponds to hybridization to exon 3 sequences, we conclude that the AG  GG change completely abolishes only the 3-splice acceptor site of IVS-2. The 5-donor site is still used because the 201-bp band is detected. Most likely, a 3-acceptor site in the vector is being used, which cannot be detected by this riboprobe.

AG IVS-2

These results were also confirmed using RT-PCR. COS-1 transfected RNA from wild type, codon 27, codon 97, IVS-2 AT deletion, or mutant IVS-2 3-acceptor plasmids were reverse transcribed, and the exon 2/exon 3 boundary was subsequently PCR amplified. A single band of 320 bp was observed except for the mutant IVS-2 3-acceptor plasmid where no bands were detected (data not shown). The 320-bp band corresponds to normal excision of IVS-2; therefore, absence of this band indicates IVS-2 is not excised normally. Furthermore, the fact that there are no additional bands in either the RNase protection or the RT-PCR assays suggests that there is no cryptic splice site activated in IVS-2 or in the coding region from exon 3. PCR would detect such a cryptic splice because the 3-primer (primer 24) used in the RT-PCR was complementary to the end of the coding region in exon 3.

Of the four changes evaluated in the RNase protection/RT-PCR assays, only the G  T at codon 27 and AG  GG in the 3-splice acceptor of IVS-2 result in aberrant splicing. Because neither codon 97 nor IVS-2 AT deletion changes leads to aberrant splicing, we conclude that the third change in this complex allele, the C  T at codon 4 Thr  Ile, is responsible for decreased Hb A₂.

Expression of -Chain Structural Variants and Functional Characterization of Hemoglobin Tetramers

We were also interested in understanding how the four different -chain structural variants lead to reduced Hb A₂. The C  T at codon 4 Thr  Ile, G  T at codon 27 Ala  Ser, C  T at codon 116 Arg  Cys, and T  C at codon 141 Leu  Pro were introduced separately into normal -globin cDNA using homologous PCR recombination (25). Variant -globin chains were expressed in a yeast system, which results in production of soluble Hb A₂ tetramers, which can be readily isolated. Characterization of oxygen binding properties and stability to heat and mechanical agitation were evaluated in order to monitor functional consequences of these changes.

Normal and variant carbonmonoxy forms of the soluble tetramers were purified from yeast using carboxymethyl-cellulose (CM-52) followed by fast protein liquid chromatography Mono S chromatography (33). Expression of three of the four -chain variants resulted in production of soluble variant Hb A₂ tetramers, which coeluted with wild type Hb A₂. Hb T4I, Hb A27S, and Hb R116C were isolated to homogeneity for further study. No Hb L141P tetramers were observed after the expression of the codon 141 Leu  Pro -chain variant in yeast. Only -globin monomers were detected after expression of this plasmid. This is most likely due to globin instability or inability to form Hb L141P tetramers. A sample of each of the other three purified tetramers was studied by electrophoresis on cellulose acetate at pH 8.9 (data not shown). Hb A₂ migrates close to Hb C in this system. As expected, Hb R116C migrates faster than Hb A₂, similar to Hb S, because the positively charged Arg residue is replaced by a neutral Cys, consistent with the altered electrophoretic mobility of this variant seen in hemolysates from the patient. Both Hb T4I and Hb A27S had similar electrophoretic mobilities and migrated with Hb A₂, consistent with lack of surface charge differences as a result of the amino acid changes (data not shown). Mass spectral analysis of -chains from Hb A₂, Hb A₂ A27S, Hb A₂ R116C, and Hb A₂ T4I showed expected values of 15,924.3, 15,940.3, 15,871.3, and 15,936.3 Da, respectively.

Heat and mechanical stability tests were employed in order to determine whether stability was the cause of decreased expression. Results of heat stability at 45 °C for 10 min are shown in Table II. All three variants and recombinant wild type Hb A₂ were equally stable to heat. The same assay was performed at 37, 55, and 60 °C with identical results (data not shown).

Table II. Heat stability properties of recombinant hemoglobins Oxyhemoglobin solutions in 0.1 M phosphate buffer, pH 7.4, were incubated at 45 °C for 10 min. Precipitated tetramers were pelleted by centrifugation, and the absorbance at 577 nm of the oxyhemoglobin remaining in solution was measured and given as a percentage of the initial hemoglobin concentration. Hb A2 Hb T4I Hb A27S Hb R116C % Oxyhemoglobin in solution 95 93.5 92.5 94.5

Mechanical stabilites after agitation for 60 s are shown in Table III. The Hb T4I variant was less stable to mechanical agitation compared with the other two variants and recombinant wild type Hb A₂. Mechanical stability for Hb T4I is shown in Fig. 5 with those of Hb A, Hb S, and Hb A₂. Native Hb S precipitates rapidly in this assay (34), whereas native Hb A is quite resistant to precipitation. Mechanical stabilities for the various tetramers showed that Hb R116C was similar to Hb A₂, whereas Hb T4I was about 2-fold less stable. Hb A27S showed intermediate stability. Decreased mechanical stability of normal native Hb A₂ as compared with native Hb A is well documented (35). All Hb A₂ variants were less stable to mechanical agitation than Hb A but were more stable than Hb S. 

Table III. Mechanical stability properties of recombinant hemoglobins Oxyhemoglobin solutions in 0.1 M phosphate buffer, pH 7.4, at room temperature were mechanically agitated for 60 s, precipitated tetramers were removed by centrifugation, and the absorbance at 577 nm of the oxyhemoglobin remaining in solution was measured and given as a percentage of the initial hemoglobin concentration. Hb A2 Hb T4I Hb A27S Hb R116C % Oxyhemoglobin in solution 83 50 71 84

Mechanical stability of Hb A₂ and Hb T4I tetramers.

Because one of the variants, Hb A₂ R116C, contained an Arg to Cys change at 116 we tested for levels of dimeric hemoglobins that might be generated by a disulfide bridge. Our results using gel filtration on Superose 12, which readily distinguishes between monomeric and dimeric tetramers, showed the same elution profiles for Hb A₂ R116C, HbA₂, and Hb A (data not shown). Furthermore, lack of disulfide bond formation was also demonstrated by standard SDS-polyacrylamide gel electrophoresis analysis (data not shown).

Finally, oxygen affinities were evaluated to test whether these amino acid changes have any effect on oxygen binding properties of tetramers (Table IV). Oxygen association curve and the Hill plot analysis for wild type recombinant Hb A₂ and Hb R116C are very similar (data not shown). P₅₀ values for all recombinant hemoglobins were similar. Cooperativity values (n_max), a measure of ability to bind subsequent oxygen molecules after initial binding, were slightly lower for Hb A27S, whereas values for the other variants were identical to Hb A₂.

Table IV. Oxygen binding properties of recombinant Hbs Oxygen association curves of recombinant Hbs were determined in 0.1 M phosphate buffer, pH 7.4, at 20 °C. P50 is the partial oxygen pressure required to give 50% saturation of hemoglobin. nmax values were calculated from the Hill plot of oxygen equilibrium curves. Hb A2 Hb T4I Hb A27S Hb R116C P50 3.3 3.8 3.1 3.2 nmax 2.6 2.6 1.9 2.6

DISCUSSION

All five -globin gene alleles are associated with reduced Hb A₂. The last allele described has three base changes present in cis: a C  T change at codon 4 Thr  Ile, a C  T change at codon 97, and an AT deletion at position 722 in IVS-2. Our data suggest the codon 4 amino acid change is responsible for reduced Hb A₂. We propose to name the Hb T4I variant Hb A₂-Mitsero after the village in Cyprus where it was identified.

The G  T Change at Codon 27 Ala  Ser

This mutation was described previously in a Sardinian family (7) and occurs in a region completely homologous between the - and -globin genes. The same change at codon 27 in -globin produces the variant Hb Knossos (10) and also results in ⁺ thalassemia due to aberrant splicing caused by activation of a cryptic donor splice site (9).

RNase protection and RT-PCR assays for the -gene variant show the change leads to activation of a cryptic splice site and aberrant splicing, which accounts for ~80% of the total -globin mRNA. This is the same cryptic site used in mutations at codons 24, 26 (^E), and 27 (^Knossos) in the -globin gene (Refs. 36, 37, and 9 and Fig. 3), with transient assays showing 75, 5-8, and 5-7%, respectively, abnormally spliced message. Of the three -globin gene mutations, the silent T  A change in codon 24 creates the strongest cryptic splice site. An A is present at position 2 of the splice site consensus in 52% of mammalian gene splice donor sites reported in the GenBank^TM data base, whereas a T at position 2 is only present in 18% of donor sites (38). The G  A change at codon 26, in addition to the alternatively spliced mRNAs, appears to increase steady-state levels of unspliced precursors in transient expression assays. An A is present at position +3 of the splice site consensus in 42% of mammalian gene splice donor sites reported in the GenBank^TM data base, and a G at position +3 is present in 52% of donor sites (38). Thus, an A or a G at this position does not greatly affect the strength of this cryptic splice site. Shapiro and Senapathy (38) predict that the G  T change at codon 27 alters splice-site strength because a T is present at position +6 in 56% of splice sites reported, whereas a G at this position is only present in 13% of splice sites (38). This prediction fits well with our data showing that 80% of the  G  T codon 27 message is abnormally spliced but does not match the 5-7% abnormally spliced message seen in the  codon 27 transient expression assays.

Despite the fact that the region surrounding the codon 27 change is identical in the - and -globin genes, the same change leads to different levels of abnormal splicing, 5-7% versus 80%. Of more importance is the observation that ^Knossos carriers express about 35% variant tetramers (9), which reflects about 30% abnormally spliced message, a much higher amount than the 5-7% reported in expression assays.

Our results showing about 80% abnormally spliced message are consistent with the 0.2% Hb A₂ seen in a compound heterozygote for the  codon 27/ IVS-2 3-acceptor mutation (1) (see also under the AG  GG change at the 3-acceptor site of IVS-2) and the 1.7% Hb A₂ seen in a heterozygote for the codon 27 change (1) (Family B, Individuals II-3 and I-2). In fact, protein levels would correspond to an even higher percent (85%) of abnormally spliced message, if one assumes 1.3% Hb A₂ as the output of a normal -globin gene. This difference between the same mutation in - and -globin genes could be reconciled if the alanine to serine change would affect tetramer folding, assembly, or stability of the variant. Heat stability of the variant was similar to Hb A₂, whereas mechanical stability was slightly less. Whether tetramer folding and/or assembly are affected is not known. Oxygen equilibrium curves show no significant differences compared with wild type Hb A₂ suggesting minimal if any effects on tetramer structure-function.

The AG  GG Change at the 3-Acceptor Site of IVS-2

The AG  GG change in the 3-acceptor site of IVS-2 of the -globin gene is responsible for total absence of Hb A₂ in homozygotes for this mutation (1) (Family C, Individual II-1). The same change in the -globin gene results in a ^o-thalassemia mutation, because no normally spliced transcripts are produced (39, 40). The 3-acceptor site in IVS-2 of the  gene is also abolished, and a new cryptic acceptor site at position 579 of IVS-2 is activated. This alternate splicing produces transcripts in which IVS-2 is only partially excised. The resulting mRNA cannot encode for -globin chain (39). We did not expect to see the same cryptic splice site in IVS-2 activated in the AG  GG change of the  IVS-2 because the  and  IVS-2 sequences diverge considerably.

In the RNase protection assay (Fig. 4), there was no protected fragment corresponding to exon 3 sequences, confirming that the AG  GG change abolishes the IVS-2 acceptor site in the -globin gene. However, there was protection for exon 2 sequences, suggesting the IVS-2 donor splice site was utilized and spliced to another acceptor site that was neither in IVS-2 nor in exon 3, because no other smaller or larger protected fragments were seen in the RNase protection assay. These data definitively show the AG  GG change completely abolishes the 3-acceptor site.

The C  T Change at Codon 97 and the AT Deletion at Position 722 in IVS-2

Because the C  T at codon 97 change is close to the end of exon 2, we wanted to determine whether it activated a cryptic splice site at that position. The deleted AT at position 722 of IVS-2 is one of three ATs that are far from the branch point, which is critical for splicing. In the -globin gene, a mutation at nucleotide 745 of IVS-2 creates a 5-splice site and activates a cryptic 3-splice site at IVS-2 nucleotide 579 (41). This -thalassemia gene produces an aberrant transcript containing 165 nucleotides of IVS-2 inserted between exons 2 and 3, in addition to a small amount of normally spliced -globin RNA. It would be unlikely that the AT deletion causes the same post-transcriptional defect in this -globin gene, because  and  IVS-2 contain minimal if any sequence homology. We found that neither the codon 97 nor the AT deletion changes alter splicing in the RNase protection/RT-PCR assays (Fig. 4).

The C  T Change at Codon 4 Thr  Ile

Because the codon 97 and AT deletion changes appear to be neutral polymorphisms, the codon 4 Thr  Ile change, also present in cis, remains the likely cause for decreased Hb A₂. However, it is not immediately obvious how this change results in decreased expression. Thus far, this position in the -globin chain is not implicated in any important function. A  mutation nearby that has clinical implications is the substitution of Glu at position 6 by Val, which results in ^S-globin (42). A hydrophobic amino acid on the surface of the tetramer, such as Ile at codon 4, might significantly increase surface hydrophobicity. In fact, site-directed mutagenesis studies show decreased stability to mechanical agitation as 6 amino acid hydrophobicity increases (43). The Thr  Ile change at codon 4 in the -chain significantly decreases mechanical stability of this variant, whereas heat stability was not affected, similar to Hb S.

Hb A₂ has high affinity for red cell membranes (44). Recent studies in transgenic mice show expression of high levels of - and ^S-globin result in severe red cell shape abnormalities, suggesting increased interaction of - or ^S-chains with red cell membranes (45). The Thr to Ile change at codon 4 is on the surface and should increase hydrophobicity as reflected by the observed increased instability of tetramers to mechanical agitation. Increased hydrophobicity could also promote interaction with membranes, thereby leading to reduced levels of Hb T4I.

The C  T Change at Codon 116 Arg  Cys

This change results in a variant with altered electrophoretic mobility, suggesting that 116 is exposed and the charge difference can be manifested. Position 116, as well as 108, 112, and 115 are ₁₁ contact points (46). Of the six  variants reported at these positions (18), three are stable and three are unstable. Furthermore, residue 116 is one of the 10 amino acid differences between - and -globin chains (His versus Arg, respectively). Arg at this position is thought to provide an additional contact between - and -chains by hydrogen bonding to residue 114 Pro (47). Higher thermal stability of Hb A₂ compared with Hb A is considered due to this extra bond that forms between - and -chains. We proposed substitution of Arg 116 by Cys would have detrimental effects on Hb A₂ stability. However, our present results show no increased instability of Hb R116C. Unlike the ₁₂ interphase, the ₁₁ interphase does not undergo much shifting when hemoglobin changes from oxy to deoxy form. Therefore, we do not expect this Arg to Cys change to affect oxygen binding properties of the protein.

We previously suggested the change to Cys at 116 might promote disulfide bond formation (1). Hb Porto Alegre 9 Ser  Cys (48) and Hb Ta-Li 83 Gly  Cys (49) changes occur on the outside surface, which allows intermolecular disulfide bonding. In Hb Rainier 145 Tyr  Cys (50) the change promotes intramolecular hydrogen bonding with the normal 93 Cys. Interdisulfide bond formation could inhibit heme/heme interactions and impair tetramer function. Our results with Hb R116C show no evidence for intermolecular disulfide bond formation and no effect on function compared with normal Hb A₂. These results suggest that the 116 Arg to Cys change does not lead to intra- or interdisulfide bond formation like Hb Porto Alegre, Hb Ta-Li, or Hb Rainier.

Interestingly, our recent expression studies of soluble -globin chains show that in the absence of -chains the -chains form dimers because of disulfide bond formation involving Cys-112.² The Cys-112 position is also an ₁₁ interaction site, and disulfide bond formation of --chains at this site inhibits assembly with -chains. These dimers do not dissociate to monomers and do not interact with -chains to form Hb A. The R116C change might therefore facilitate intra- or interdisulfide bonded -chain dimer formation, so that overall Hb R116C levels would be reduced. The variant -chains, which did productively assemble with -chains, would, however, result in a normally functioning tetramer. We are currently testing this hypothesis by expressing normal and variant -chains and then monitoring their ability to assemble in vitro with -chains.

The T  C Change at Codon 141 Leu  Pro

This change is a nonconservative substitution, and 24 -chain variants have been reported that change an amino acid to proline; 21 result in unstable hemoglobins (18), because proline interferes with -helix formation (46). In addition, 141 is in the interior of hemoglobin, and its hydrophobic side chain is in direct contact with heme (51). Therefore, a change to Arg at codon 141 results in marked instability of Hb A tetramers (52), whereas an Ala to Pro change at 142 (H20) results in instability and heme loss (53). The Leu  Pro substitution in codon 141 (H19) is at a homologous site in the -chain variant Hb Bibba 136 (H19) Leu  Pro, which is also a very unstable variant (54). From these data, our observation that no Hb A₂ was detected in two siblings homozygous for this mutation, and our inability to obtain Hb  L141P using the yeast expression system, it appears that the Leu  Pro change results in either marked instability or lack of assembly.