Molecular characterization of G6PD deficiency in Cyprus

Molecular characterization of G6PD deficiencyin Southern Italy:heterogeneity, correlation genotype–phenotypeand description of a new variant (G6PD Neapolis)

FIO RELLA AL FINITO,1,2 AMEL IA CI MMI NO,1,2 FILO ME NA FERRARO,1 MARIA VIT TO RIA CUBE LLIS,3 LUIGI VITAGLIANO,4

MATTE O FRANCESE,5 ADRIANA ZAGARI,4 BRUNO ROTOLI,1 STEFANIA FIL OSA2

AND GIUSE PPE MARTINI2

1Divisione di Ematologia Universita degli Studi Federico II, Naples, 2Istituto Internazionale di Genetica e Biofisica,CNR, Naples, 3Dipartimento di Chimica Organica e Biologica, Universita degli Studi Federico II, Naples,4Centro di Studio di Biocristallografia, CNR and Dipartimento di Chimica, Universita degli Studi Federico II, Naples,and 5Clinica Pediatrica, II Universita’ degli Studi, Naples, Italy

Received 14 November 1996; accepted for publication 1 April 1997

Summary. We report on the molecular basis of glucose-6-phosphate dehydrogenase (G6PD) deficiency in SouthernItaly (Campania region). Thirty-one unrelated G6PD-deficient males were analysed at DNA level for the presenceof G6PD gene mutations. Nine different G6PD variants wereidentified, eight of which have already been described(Mediterranean, Seattle, two different A¹, Santamaria,Cassano, Union and Cosenza). G6PD Mediterranean, Santa-maria, A¹ and Union were associated with haemolyticepisodes. G6PD Seattle, which is polymorphic in several

populations, Cassano and Cosenza appeared to be asympto-matic. A new variant (G6PD Neapolis) is reported here. The467Pro →Arg substitution reponsible for G6PD Neapolis isdiscussed in the light of the current 3D model of humanG6PD and in comparison with other natural mutationswhich occur in the proximity of residue 467.

Keywords: G6PD deficiency, G6PD variants, G6PD molecularheterogeneity, G6PD Neapolis.

G6PD deficiency, the most common human enzymopathy, isan X-linked disorder in which one hundred mutationshave now been characterized at the DNA level (Mason,1996; Beutler et al, 1996). The main clinical manifestationsare neonatal jaundice and acute haemolytic crises triggeredby drugs, infections or fava bean ingestion. In a smallnumber of cases G6PD deficiency produces chronic haemo-lysis (Luzzatto & Metha, 1995) and adversely affectshaemopoiesis in heterozygotes (Filosa et al, 1996).

The main role of G6PD resides in the defence againstoxidative stress, even in nucleated cells which possessalternative defence mechanisms (Pandolfi et al, 1995; Martini& Ursini, 1996). The enzyme is a homotetramer whoseprimary sequence has been highly conserved during evolution(Luzzatto & Metha, 1995). Recently, a 3D structure was

obtained for the enzyme from Leuconostoc mesenteroides(Rowland et al, 1994) which allowed the production of ahomology model for the human enzyme (Naylor et al, 1996).We have performed a study on the molecular heterogeneity ofG6PD deficiency in the Campania region (Southern Italy) andreport on a new mutation in the G6PD gene and its possibleconsequences with respect to the enzyme’s 3D structure.

MATERIALS AND METHODS

Patients. We studied 31 G6PD deficient unrelated malesubjects. Diagnosis was made according to the ICSHrecommendations (Beutler et al, 1977). Patients came toour observation from two different sources: (i) 17 had beenreferred to our anaemia out-patient unit following haemo-lytic crises after ingestion of fava beans or other agents. Noneof these patients showed other haematological abnormalitiesand, after recovering from the haemolytic episode, theyexhibited normal values of haemoglobin concentration,

British Journal of Haematology, 1997, 98, 41–46

41q 1997 Blackwell Science Ltd

Correspondence: Dr Fiorella Alfinito, Divisione di Ematologia Facoltadi Medicina, Universita degli Studi Federico II, Via Pansini 5, 80131Napoli, Italy.

reticulocyte count and bilirubin level; (ii) 14 asymptomaticsubjects were referred to our hospital for confirmation ofG6PD deficiency which had been incidentally discoveredduring a screening of military personnel for specialized corps.A complete haematological screening of these subjectsrevealed no other alteration.

Enzyme characterization. Partial purification and biochem-ical characterization were performed as reported elsewhere(Calabro et al, 1990).

DNA analysis. Genomic DNA was obtained using standardtechniques (Sykes, 1983). All samples were investigated bythe PCR technique and enzyme digestion specific for themost common variants in our country, as previouslyreported (Calabro et al, 1993). G6PD A variants wereinvestigated for the presence of further mutations leadingto A¹ variants according to Beutler et al (1989). TheSantamaria mutation was analysed by direct sequencing ofexon VI of G6PD A samples.

Samples that were not defined by the above procedureswere submitted to amplification of all exons and thenprocessed using the SSCP technique. Direct sequencing(Sequenase PCR Product Sequencing Kit by USB) wasperformed when indicated by the SSCP investigation.

To confirm the G6PD 467 Pro → Arg mutation, a PCR ampli-fication followed by HinfI and PvuII digestion was performed.

The oligonucleotides 5 0GCAGGCAGTGGCATCAGCAAG30

in exon XI and 5 0GCTCAATCTGGTGCAGGAGT30 in exon XIIwere employed. The latter introduces a point mutation(underlined) creating a new HinfI site in cooperation with1400 C → G.

3D protein model analysis. The protein structural motifswere analysed using the program PROMOTIF (Hutchinson& Thornton, 1996); the replacement of Pro by Arg atposition 467 and the graphical analysis of the modelwere carried out using the program O ( Jones et al, 1991).Protein structure models were drawn using the programMOLSCRIPT (Kraulis, 1991).

RESULTS

DNA analysisIn our series of 31 G6PD deficient subjects, PCR amplificationfollowed by restriction analysis identified nine G6PD variants.

The commonest variant was G6PD Mediterranean (45%),followed by G6PD Seattle (25.8%), G6PD A¹ (12.9%) andG6PD Cassano (6.45%); single cases of G6PD Maewo and ofG6PD Cosenza were also found (Table I).

q 1997 Blackwell Science Ltd, British Journal of Haematology 98: 41–46

42 Fiorella Alfinito et al

Table I. Some clinical features of the G6PD variants identified in the present study.

Triggering agentAsymptomatic Patients with a history Neonatal

Variant subjects of haemolytic crises jaundice Transfusions Fava beans Other agents

Mediterranean 2 12 2 2 11 1Santamaria 2 2A¹ 2 1 1 1Union 1 1 1Seattle 8Cassano 2Cosenza 1Neapolis 1

Table II. Biochemical characteristics of G6PD Neapolis.

G6PD Neapolis Normal values

Electrophoretic mobility (%) 85 100Km G6P (mmol) 55 55–80Km NADP (mmol) 22 25–30dG6P (%) 4 3–4Thermostability (%) 27 28–35

Thermostability was evaluated as residual activity after 40 minincubation at 508C.

Fig 1. (A) SSCP analysis of amplified fragment encompassing exonsXI–XIII. (B) Products of double digestion HinfI/PvuII of amplifiedfragments of exons XI–XIII. Lanes 1: G6PD Neapolis; lanes 2:normal control.

43G6PD Neapolis


The subjects analysed in this study belong to twocategories: patients studied for symptoms of paroxysmalhaemolytic crises, and totally asymptomatic subjects whohad incidentally been found to be G6PD deficient (seeMaterials and Methods). The distribution of G6PD mutationsand some clinical data are listed in Table I, although in oldersubjects neonatal jaundice might have been overlooked ornot recorded. Molecular lesions in the group of patients whohad suffered from haemolytic crises showed a predominanceof G6PD Mediterranean (70%). In the group of asympto-matic G6PD-deficient subjects the molecular variant mostfrequently found was G6PD Seattle (56%) followed by G6PDMediterranean, Cassano, Cosenza and Neapolis.

A new G6PD variantOne sample from the group of asymptomatic subjectsappeared not to correspond to any known variant.

Biochemical enzyme characterization showed reduced activ-ity (30%) and electrophoretic mobility (85%) whereas otherproperties appeared within or close to the normal rangeestablished in our laboratory (Table II). By SSCP analysis weobserved an abnormal shift in the amplified DNA fragmentencompassing exons XI, XII and XIII (Fig 1A). Directsequencing revealed a point mutation (C → G) at nucleotide1400 producing a Pro → Arg aminoacid substitution atposition 476. To confirm the 1400 C → G mutation, wedesigned an oligonucleotide which, when used as a PCRprimer, introduces a point mutation creating a new Hinf Isite in the presence of 1400 C → G. When the PCR productswere digested with Hinf I and PvuII, two fragments, one of150 and another of 127 bp, were produced in normalsamples. In contrast, in DNA carrying the 1400 C → Gmutation, three fragments (of 130, 127 and 20 bp) weredetected after the double digestion, as expected (Fig 1B).

Fig 2. (A) Overall fold of human G6PD subunit,with a helices and b strands shown as a coiledribbon and flattened arrows, respectively.In the aþb domain (red) helix an is colouredyellow; in the coenzyme domain (green) helixae is coloured blue. (B) Close-up view of panelA, showing the inter-domain boundary formedby helix an (residues 455–473), and helix ae(residues 177–190), followed by the loop191–193. The structure refers to the Neapolisvariant. Only arginines at positions 459, 463and 467, and residues interacting with them,are shown. (C) Spiral representation of thehelix an drawn in panel B; the residues arecolour-coded for hydrophobic (green), polar(blue) and charged (red) amino acid types.Residues span from Ser455 to Glu473.

Enzyme structureBased on the three-dimensional structure of G6PD fromLeuconostoc mesenteroides (Rowland et al, 1994), a homologymodel of the human enzyme has recently been built(Naylor et al, 1996). Each subunit (Fig 2A) is formed by asmaller a/b coenzyme domain and a larger bþa domain.The NADPþ binding site is located in the coenzyme domainand the substrate binding site is at the boundary betweenthe two domains. According to the human model, theG6PD Neapolis mutation affects the alpha helix n(residues 455–473). This helix is located on the surfaceof the larger bþa domain and faces the coenzymedomain. As shown in Fig 2C, helix an is amphipathic;the hydrophobic and the polar sides interact with theinterior of the bþa domain, whereas the polar-chargedside interacts with the helix ae of the a/b domain (Figs 2Aand 2B).

Besides the Neapolis mutation, four other naturallyoccurring mutations, 459Arg →Leu, 459Arg →Pro,463Arg →Cys and 463Arg →His (Table III) map to thesame structure (Fig 2B). All these residues are far more than10 A away from either the active site (His 201) or thecoenzyme binding site (Gly 38); in fact, residues Gly 38 andHis 201 are confidently considered close to the cofactor andsubstrate binding site respectively (Rowland et al, 1994). Thesteric interactions with neighbouring residues have beenanalysed for each point mutation in helix an and aredescribed as follows.

Arg459 interacts with a residue of helix ae (residues 177–190) which is located on the surface of the coenzymedomain. In fact, the Arg459 side chain forms weak ion-pair interactions with both Asp181 and Glu193 (Fig 2B).

The positively charged side chain of Arg463 favourablyinteracts with the negatively charged Glu460 alonghelix an (Fig 2B). We verified that either Cys or His canbe accommodated at position 463 without steric overlaps.

Finally, the residue at position 467 is on the exterior ofthe molecule and therefore can be easily subjected to localchanges without significant alterations of the enzymeproperties. We replaced Pro467 by Arg and verifiedthat this substitution does not cause defective contactsin the model of human G6PD, but gives rise to afavourable ion-pair interaction with Glu297 of a near loop(Fig 2B).

DISCUSSION

The occurrence of nine different DNA variants in this studyindicates extensive heterogeneity of G6PD deficiency inCampania, as already suggested by biochemical studies(Colonna-Romano et al, 1985). Similar findings werereported for other regions of continental Italy (Calabro etal, 1993; Cappellini et al, 1996), whereas in Sardinia a singleG6PD mutation (188 Ser →Phe: G6PD Mediterranean) isprevalent (De Vita et al, 1989). The G6PD variant namedSantamaria was previously described in two unrelatedpatients from Costa Rica (Beutler et al, 1991) and morerecently in Italy (Ninfali et al, 1993). To our knowledge, aG6PDA¹ mutation characterized by 323 Leu →Pro hasnever been reported in Europe. The two patients in our seriescarrying G6PD Santamaria were not related, and therefore itis likely that this variant is not sporadic in Campania. A newG6PD mutation was isolated in this study and has beennamed G6PD Neapolis for the city where the patient wasborn. The complex historical background of the geographicarea analysed here has probably played an important role increating such a genetic admixture. Our study clearlydemonstrates that the mutation 188 Ser →Phe (G6PDMediterranean) is strongly associated with haemolytic eventswhereas the mutation 282 Asp →Hys (Seattle) is alwaysassociated with a non-haemolytic phenotype (Table I). Thisis in agreement with a study of nine unrelated Spanishsubjects with favism, none of whom was found to bear theG6PD Seattle variant, suggesting that this variant, which ispolymorphic in Spain, is asymptomatic (Rovira et al, 1995).As for other variants, the small number of cases does notallow us to draw meaningful conclusions regarding thehaemolytic phenotype.

The 3D model of the human enzyme enabled Naylor et al(1996) to provide evidence that weakened dimer interac-tions are responsible for a large class of G6PD mutationsgiving rise to chronic non-spherocytic haemolytic anaemia.Along this line we have attempted to rationalize theproperties of all known mutations of helix an, using as areference the same model of human G6PD. It is worthpointing out that the current model of the human enzyme isbased on sequence homology and the sequence of alpha helixn shows little evolutionary conservation (only two aminoacids are identical in Leuconostoc and man). Therefore any


44 Fiorella Alfinito et alTable III. G6PD mutations affecting alpha helix n.

W.H.O.Variant AA substitution class Activity (%) Reference

Taiwan Hakka 459 Arg →Leu 2 4.9 Chiu et al (1991)Cosenza 459 Arg →Pro 2 2 Calabro et al (1993)Kaiping 463 Arg →His 2 8.0 Chiu et al (1991)Karniube 463 Arg →Cys 3 NA Chiu et al (1993)Neapolis 467 Pro →Arg 3 30 This reportG6PD B Wild type 100

NA: not available.

45G6PD Neapolis


conclusion based on the model must wait for experimentaldata on the 3D structure of the human enzyme to becomedefinitive. However, considering that the amphipathiccharacter of helix an is quite similar in Leuconostoc andhuman enzyme, we analysed the 3D-structure of all naturalvariants affecting the helix an: the substitution of Arg459for Leu or Pro destroys the favourable contacts with Asp181and Glu193, because a charged residue is replaced by ahydrophobic residue on the polar side of helix an whichinteracts with the coenzyme domain. As a consequence,these mutations can indirectly affect the inter-domainorientation which undergoes a pronounced change uponsubstrate binding. This can significantly affect the enzymaticactivity. On the other hand, the mutations at position 463are not expected to be so deleterious as those found atposition 459, because either Cys or His can be easilyaccommodated at position 463. Finally, since the residue atposition 467 is on the exterior of the molecule, it is notsurprising that G6PD Neapolis mutation affects the enzymeactivity to a lesser extent than the other mutations detectedin alpha helix n (Tables II and III). Therefore our analysisconfirms that the mutation 467Pro →Arg can be welltolerated by the enzyme and produces a slight effect on theactivity of human G6PD.

In summary, we have predicted the properties of all knownmutations of alpha helix n and speculated that the inter-helical ae–an interactions contribute to determining thereciprocal orientation of the two enzyme domains which, inanalogy with other dehydrogenases, might dynamicallychange during catalysis.

ACKNOWLEDGMENTS

We are indebted to L. Luzzatto for his kind suggestions, M. V.Ursini and A. M. Franze’ for their help on many occasionsand to C. E. Naylor, P. Rowland and M. J. Adams for sendingus the atomic coordinates of the human G6PD 3D modelbefore publication. The skilful technical assistance of MariaTerracciano is gratefully acknowledged. Graphic and com-puting systems were made available by CE.IN.GE (Naples).

REFERENCES

Beutler, E., Blume, K.G., Kaplan, P.C., Lohr, G.W., Ramot, B. &Valentine, W.M. (1977) International Committee for Standardiza-tion in Haematology. Recommended methods for red cell enzymeanalysis. British Journal of Haematology, 35, 331–340.

Beutler, E., Kuhl, W., Saenz R., German, F. & Rodriguez, W.R. (1991)Mutation analysis of glucose-6-phosphate dehydrogenase (G6PD)variants in Costa Rica. Human Genetics, 87, 462–464.

Beutler, E., Kuhl, W., Vives-Corrons, J.L. & Prchal, J.T. (1989)Molecular heterogeneity of glucose-6-phosphate dehydrogenaseA¹. Blood, 74, 2550–2555.

Beutler, E., Vulliamy, T. & Luzzatto, L. (1996) Hematologicalimportant mutations: glucose-6-phosphate dehydrogenase. BloodCells, Molecules, and Diseases, 22, 49–56.

Calabro, V., Giacobbe, A., Vallone, D., Montanaro, V., Cascone, A.,Filosa, S. & Battistuzzi, G. (1990) Genetic heterogeneity at theglucose-6-phosphate dehydrogenase locus in Southern Italy: a

study on a population from the Matera district. Human Genetics,69, 49–53.

Calabro, V., Mason, P.J., Filosa, S., Civitelli, D., Cittadella, R.,Tagarelli, A., Martini, G., Brancati, C. & Luzzatto, L. (1993)Genetic heterogeneity of glucose 6-phosphate dehydrogenasedeficiency revealed by single-strand conformation and sequenceanalysis. American Journal of Human Genetics, 52, 527–536.

Cappellini, M.D., Martinez di Montemuros, F., De Bellis, G.,Debernardi, S., Dotti, C. & Fiorelli, G. (1996) Multiple G6PDmutations are associated with a clinical and biochemicalphenotype similar to that of G6PD Mediterranean. Blood, 87,3953–3958.

Chiu, D.T.-Y., Zuo, L., Chao, L., Chen, E., Louie, E., Lubin, B., Liu, T.Z.& Du, C.-S. (1993) Molecular characterization of glucose-6-phosphate dehydrogenase (G6PD) deficiency in patients of Chinesedescent and identification of new base substitutions in the humanG6PD gene. Blood, 81, 2150–2156.

Chiu, D.T.-Y., Zuo, L., Chen, E., Chao, L., Louie, E., Lubin, B., Liu, T.Z.& Du, C.-S. (1991) Two commonly occurring nucleotide basesubstitutions in Chinese G6PD variants. Biochemical and Biophy-sical Research Communications, 180, 988–993.

Colonna-Romano, S., Iolascon, A., Lippo, S., Pinto, L., Cutillo, S. &Battistuzzi, G. (1985) Genetic heterogeneity at the glucose-6-phoshate dehydrogenase locus in Southern Italy: a study on thepopulation of Naples. Human Genetics, 69, 228–232.

De Vita, G., Alcalay, M., Sampietro, M., Cappellini, D., Fiorelli, G. &Toniolo, D. (1989) Two point mutations are responsible for G6PDpolymorphism in Sardinia. American Journal of Human Genetics,44, 233–240.

Filosa, S., Giacometti, N., Cai, W., De Mattia, D., Pagnini, D.,Alfinito, F., Schettini, F., Luzzatto, L. & Martini, G. (1996)Somatic cell selection is a major determinant of the blood cellphenotype in heterozygotes G6PD mutations causing severeenzyme deficiency. American Journal of Human Genetics, 59, 887–895.

Hutchinson, E.G. & Thornton, J.M. (1996) PROMOTIF: a program toidentify and analyze structural motifs in proteins. Protein Science,5, 212–220.

Jones, T.A., Zou, J.-Y., Cowan, S.W. & Kjeldgaard, M. (1991)Improved methods for building protein models in electron-densitymaps and the location of errors in these models. ActaCrystallographica, A47, 110–119.

Kraulis, P.J. (1991) MOLSCRIPT: a program to produce both detailedand schematic plots of protein structures. Journal of AppliedCrystallography, 24, 946–950.

Luzzatto, L. & Metha, A. (1995) Glucose 6-phosphate dehydrogenasedeficiency. The Metabolic Basis of Inherited Diseases, pp. 3367–3398. McGraw Hill, London.

Martini, G. & Ursini, M.V. (1996) A new lease of life for an oldenzyme. BioEssays, 18, 631–637.

Mason, J.P. (1996) New insights into G6PD deficiency. British Journalof Haematology, 94, 585–591.

Naylor, C.E., Rowland, P., Basak, A.K., Gover, S., Mason, P.J.,Bautista, J.M., Vulliamy, T.J., Luzzatto, L. & Adams, M.J. (1996)Glucose 6-phosphate dehydrogenase mutations causing enzymedeficiency in a model of tertiary structure of the human enzyme.Blood, 87, 2974–2982.

Ninfali, P., Baronciani, L., Ruzzo, A., Fortini, C., Amadori, E.,Dall’Ara, G., Magnani, M. & Beutler, E. (1993) Molecular analysisof G6PD variants in Northern Italy: a study on the populationfrom Ferrara district. Human Genetics, 92, 139–142.

Pandolfi, P.P., Sonati, F., Rivi, R., Mason, P., Grosveld, F. & Luzzatto, L.(1995) Targeted disruption of the housekeeping gene encodingglucose 6-phosphate dehydrogenase (G6PD): G6PD is dispensable

for pentose synthesis but essential for defense against oxidativestress. EMBO Journal, 14, 5209–5215.

Rovira, A., Vulliamy, T., Pujades, M.A., Luzzatto, L. & Vives Corrons,J. (1995) Molecular genetics of glucose-6-phosphate dehydrogen-ase (G6PD) deficiency in Spain: identification of two new pointmutations in the G6PD gene. British Journal of Haematology, 91,66–71.

Rowland, P., Basak, A.K., Gover, S., Levy, H.R. & Adams, M. (1994)The three-dimensional structure of glucose 6-phosphate dehy-drogenase from Leuconostoc mesenteroides refined at 2.0 Aresolution. Structure, 2, 1073–1087.

Sykes, B. (1983) DNA in heritable disease. Lancet, ii, 787–788.


46 Fiorella Alfinito et al

Documents

Molecular characterization of G6PD deficiency in Cyprus