J Inherit Metab Dis (2009) 32:506–513 DOI 10.1007/s10545-009-1152-6
ORIGINAL ARTICLE
Metabolic phenotypes of phenylketonuria. Kinetic and molecular evaluation of the Blaskovics protein loading test U. Langenbeck & P. Burgard & U. Wendel & M. Lindner & J. Zschocke & German Collaborative Study on Phenylketonuria (PKU) / Hyperphenylalaninemia (HPA)
Received: 27 January 2009 / Submitted in revised form: 30 April 2009 / Accepted: 19 May 2009 / Published online: 7 July 2009 # SSIEM and Springer 2009
Summary Background: As part of the German Collaborative Study of Children Treated for Phenylketonuria (PKU), a three-day protein loading test was applied to children at 6 months of age. This load elicits three principal types of blood phenylalanine (Phe) response, with types I and III clinically corresponding to classic PKU and mild hyperphenylalaninaemia not requiring diet (MHP), respectively. An intermediate type II, clinically corresponding to mild PKU, is characterized by early decline of blood Phe from above 1200 mmol/L down to levels between 600 and 1200 mmol/L at 72 h. Aims: Unbiased classification and kinetic and molecular characterization of the intermediate Phe response;
Communicating editor: Michael Gibson Competing interests: None declared References to electronic databases: Phenylketonuria: OMIM +261600. Phenylalanine hydroxylase: EC 1.14.16.1. U. Langenbeck (*) Institute of Human Genetics, Frankfurt University Hospital, Theodor-Stern-Kai 7, D-60590 Frankfurt/Main, Germany e-mail:
[email protected] P. Burgard : M. Lindner Centre for Paediatric and Adolescent Medicine, University of Heidelberg, Heidelberg, Germany
estimation of phenotypic variability of Phe disposal. Method: A kinetic model with zero-order protein synthesis and first-order rate of metabolic disposal of Phe is applied to 157 tests. Results: A model of exponentially saturated activation describes the acceleration of Phe disposal from day 1 to 3 in the intermediate type of response. Eleven of 14 p.Y414C functional hemizygotes and two of three p.R261Q homozygotes manifested this kinetic type. The rate estimates of Phe metabolic disposal differ widely in patients with identical PAH genotype, yet are highly correlated with the Phe level at 72 h. Abbreviations AV assigned phenotypic value according to Guldberg et al (1998) HPA hyperphenylalaninaemia Kout first-order kinetic constant of metabolic loss/disposal MHP mild hyperphenylalaninaemia PAH phenylalanine hydroxylase Phe phenylalanine Phe72 blood level of phenylalanine 72 h after start of loading (morning of day 4) PKU phenylketonuria PRA predicted residual activity (mean of in vitro activities, per cent of normal) t1/2 phenylalanine half-life (50% elimination time)
U. Wendel Children_s Hospital, University of Du¨sseldorf, Du¨sseldorf, Germany J. Zschocke Divisions of Human Genetics and Clinical Genetics, Innsbruck Medical University, Innsbruck, Austria
Introduction Being the Fepitome of human biochemical genetics_ (Scriver and Clow 1980), phenylketonuria (PKU) is
J Inherit Metab Dis (2009) 32:506–513
also an early example of clinical and genetic heterogeneity (Berman et al 1969) of inborn errors of metabolism. Clinical manifestations of phenylalanine hydroxylase (PAH) deficiency in untreated individuals range from severe cognitive and neurological impairment (clinically defining classical PKU) to normal intellectual development (defining MHP). In order to determine true incidence figures of the various types of hyperphenylalaninaemia and to validly interpret the results of treatment and the effects of dietary discontinuation (Blaskovics 1976), a Funified and standardized method of assessment_ that included a protein loading test was developed within the realm of the American Collaborative Study of Children Treated for Phenylketonuria (Blaskovics et al 1974). Initially five types of Phe blood level response were distinguished in this test. One of them, the broadly defined intermediate type III with Phe blood levels oscillating between 910 and 1500 mmol/L, was interpreted as possible Fadaptation or enzyme induction as a consequence of high phenylalanine intake_. In a later report from the American Collaborative Study, three types of Phe response were described. Type I represented classical PKU, type III MHP, while the intermediate type II was characterized by an early decline of Phe values to less than 1200 mmol/L by 72 h Fdespite continuation of the procedureF (O_Flynn et al 1980). This classification was confirmed and adopted by the German Collaborative Study of Children Treated for Phenylketonuria (PKU) (Lutz et al 1990). Through analysis of test data from the German Collaborative Study and from routine diagnostic data, the present paper aims at an unbiased classification of the Phe response and at a kinetic and molecular characterization of the intermediate response type. By fitting in parallel the time course of all patients to two different models of Phe disposal, we have been able to better define the intermediate Phe response and to unravel its molecular genetic base. Our kinetic results support the early suggestion of enzyme induction on high Phe intake (Blaskovics et al 1974) and the molecular data of our study recall Fan attractive hypothesis_ that the activity of some PAH mutant enzymes may Fdepend on the concentration of phenylalanine in the liver_ (Gjetting et al 2001).
Patients and methods Patients As part of the German Collaborative PKU Study (Lutz et al 1982, 1990; Schmidt et al 1989), a protein loading test was applied to 145 children at 6 months of
507
age. Of these, 134 data sets were available for the present analysis. Another series of tests with 40 children was obtained during routine diagnostic workup and confirmatory analysis. A response type had not been assigned in these children. With a few modifications (Lutz et al 1982), the tests followed the procedure of Blaskovics and colleagues (1974): on each of three successive days, the children received three doses of infant formula plus dissolved milk powder corresponding to 180 mg Phe per kg bodyweight (bw) per day. On day 4 and 5, only 5 mg/kg bw per day was given. Blood Phe was determined twice a day, yielding data sets with Phe concentrations at 0, 8, 24, 32, 48, 56, 72, 80, and 96 h. Kinetic analysis of time course of Phe The kinetic models were built with the Windowsi based simulation and model analysis package ModelMaker version 4 (2000) of Cherwell Scientific Ltd, Oxford, UK, now distributed by ModelKinetix, Wallingford, Oxfordshire, UK. The optimization statistic R2 is calculated by the program from the weighted sums of squares (WSS) as R2 = model WSS / total WSS. For statistical data analysis and graphics, the SYSTAT 11 program package (2004) of Systat Software GmbH, D-40699 Erkrath, Germany, was used. The time course of blood Phe during the protein loading test was analysed as a single compartment model with a single input (alimentary Phe) and two outputs (first-order synthesis of tyrosine plus Phe metabolites = Fmetabolic loss_ = Kout [dayj1], and zero-order net protein synthesis [mg Phe/kg per day]), see Langenbeck et al (2001). In the model, on each of the 5 days, the protein load was taken up in three parts (39%, 22%, and 39%) at 07:00, 11:00, and 15:00, respectively, and blood Phe was determined daily at 09:00 and 17:00. All data, including obvious outliers, were used for modelling, except the measurements at 96 h. With regard to Phe, the model assumes 100% digestability of the ingested load. Fixed model parameters were: body weight (bw) = 7.8 kg, Kin = rate of intestinal uptake = 2.8042/day, perc_w = relative body water = 0.75 L/kg bw, synth_rate = net protein synthesis = 35 mg Phe/kg bw per day1 (the Phe intake needed by 5- to 7-monthold infants with classical PKU to maintain plasma Phe within the 350–700 mmol/L range, see Kindt et al 1984 and van Spronsen et al 2009). The variable parameters Kout and initial Phe blood level were optimized under visual control with 2 or 3 initial Simplex runs (ordinary least squares, 8 convergence steps). A final run with the Marquardt algorithm (without initial parameter estimation) yielded the optimization statistics but left the
508
Simplex parameter estimates unchanged. The modelling work was done without knowledge of the DNA data. Classification of the time course of Phe Response type As first outlined by O_Flynn and colleagues (1980) and also demonstrated in the German Collaborative PKU Study (Lutz et al 1982, 1990; Schmidt et al 1989), three principal types of Phe blood level response, I–III, can be distinguished in the protein loading test. In the German PKU study, the patients_ time course of Phe was classified (Lutz et al 1982, 1990) both by visual clustering (type I, continuous increase of plasma Phe beyond 1200 mmol/L; type II, spontaneous decrease of plasma Phe from above to below 1200 mmol/L on day 2; type III, fluctuations of plasma Phe around 600 mmol/L) and by taking into account the Phe72 blood level (type I >1.200 mmol/L, type II 600–1200 mmol/L, type III 3 ¼ ½5 1 e3x blood level. Thus, the model disregards a possible decrease of Kout from its maximum value at the end of day 3. The term Fmaximum_ refers solely to conditions of the protein loading test. The assignment of kinetic type without (type 1) or with (type 2) activation rested on a statistical comparison of the two optimization statistics R21 and R22 (see above). They were normalized by applying Fisher_s z-transfor 1=2 mation, z ¼0:5 ln ½ð1þrÞ=:ð1rÞ1=ðn3Þ SD, from which the normal variable Z was standard 1 obtained 2 as Z ¼ z1 z2 1 n1 3 þ 1 n2 3 (Sachs 2004) for classical discriminant analysis. This procedure secures an unbiased assignment of kinetic type 2, which is taken to represent the intermediate type II response. The
J Inherit Metab Dis (2009) 32:506–513
kinetic type 1 is differentiated by Phe72 (> or A]
[D84Y] + [R158Q] [P281L] + [R408W] [L311P] + [IVS7 + 5G>A] [R261Q] + [R261Q] (2) [R68S] + [c.165delT] [R408Q] + [P281L] [Y414C] + [P281L] [Y414C] + [S295X] [Y414C] + [R408W] (6) [Y414C] + [IVS12 + 1G>A] (2)
1 0 2 1
[Y414C] + [R261Q] [V190A] + [R408W] [E390G] + [G272X] [V177L] + [F39del]
The mutations are sorted by sum of assigned value (AV; Guldberg et al 1998) and kinetic type (1 without activation, 2 with exponentially saturated activation). N: number of cases; n, number of multiple genotypes. The genotype is known for only 21 of the 29 patients with kinetic type 2. There is no excess of BH4-responsive mutants among kinetic type 2 (see Blau 2005).
512
By testing the patient data in parallel for the models with (kinetic type 2) and without activation (kinetic type 1), activation is inferred in 29 of 157 patients (18.5%). Kinetic type 2 data can also be modelled by assuming a protein load of 48 h only (data not shown). However, our findings (i) of a conspicuous association with the functionally hemizygous p.Y414C genotype and (ii) of a positive correlation between the AV sum and the degree of activation as deduced from the Z statistic both indicate that in complete loading tests a biologically true effect is observed. Because the wildtype PAH enzyme is allosterically activated by Phe (Tho´ro´lfsson et al 2002), it may be assumed that activation by Phe is possible also in some mutant PAH enzymes. Almost all mutations associated with severe PKU lack evidence of activation, and it cannot be excluded that type 2 response patterns in the small number of patients with severe PKU represent misclassifications. In contrast, evidence of activation of the mutation is found in at least half of patients with mild PKU or MHP. Activation has been shown previously in vitro for the mutation p.R68S (Pey and Martinez 2005) supporting our corresponding in vivo finding. Further in vitro studies are needed to finally confirm our results. Interestingly, there appear to be some mild mutations such as p.L48S that are consistently typed as nonactivatable. Patients with homozygosity or functional hemizygosity for such mutations may display a mild PKU phenotype with limited rise of Phe concentrations on high Phe intake, but not the intermediate type of Phe disposal by O_Flynn et al (1980). Our data allow no conclusions with regard to alternative mechanisms such as stabilization of mRNA, stabilization of a labile transcription factor, stabilization of the target enzyme, and transcriptional activation, i.e., increased enzyme synthesis (Waxman and Azaroff 1992). The t1/2 parameter estimates of our kinetic models are highly congruent with the Phe72 value in the Blaskovics protein loading test (see Fig. 2) and less so with the AV and PRA parameters of genotype– phenotype correlation (see Fig. 3a and b). Thus, the Phe72 value is validated by our calculations as a robust predictor of the PKU metabolic phenotype during the protein load. Yet, the kinetic estimates of metabolic disposal of Phe (i.e. Kout and t1/2) translate the Phe response during protein load into a universal constant which, in contrast to the Blaskovics test Phe72 value, (i) allows comparison of all types of loadings, and (ii) by applying the equilibrium equation 5 of Langenbeck and colleagues (2001), may allow predictions of response to diet, including evolution of Phe tolerance at different ages (van Spronsen et al 2009), if recent empirical, age-
J Inherit Metab Dis (2009) 32:506–513
dependent rates of protein synthesis and estimates of total body water (Fomon et al 1982) are applied. A further result of our study is the wide range of Phe half-life estimates (and of their matching Phe72 values) in patients with severe PAH mutations and identical genotype (see Table 1). The urinary excretion during the loading test of Phe metabolites has not been documented in greater detail yet (Mo¨nch et al 1990). Such data could indicate contributions of alternative enzymes, e.g. Phe transaminase (Treacy et al 1996) or tyrosine hydroxylase (Fukami et al 1990; Thompson et al 1990). A more likely cause, however, is variability in effective protein intake and resorption or net protein synthesis, which was assumed to be constant in the present calculations (see Methods) for methodological reasons: If zero-order net protein synthesis and firstorder Phe metabolism are optimized simultaneously by the simplex algorithm, the system, under the conditions of the protein loading test, is dominated by the latter, and reliable, uncorrelated estimates are not obtained (data not shown). Repeated protein loading tests would be required to clarify this issue. In conclusion, our analysis lends support to earlier suggestions of enzyme induction (Blaskovics et al 1974) in intermediate Phe response and to the Finteresting hypothesis_ (Gjetting et al 2001) that some PAH mutant enzymes are activated at higher Phe blood levels. The great variability of Phe half-life as determined during the Blaskovics test both within groups of identical AV sum and in patients with identical PAH genotype again demonstrates that the monogenic PAH deficiency is justly considered a complex disorder (Scriver and Waters 1999). Acknowledgement The German 1978 to 1995 Collaborative Study of Children Treated for Phenylketonuria (headed until 1989 by the late Professor Horst Bickel, thereafter by Professor Hans Joachim Bremer) received financial support from Stiftung Volkswagenwerk and Bundesministerium fu¨r Forschung und Technologie (BMBF). Eight paediatric centres participated in the study: Berlin (E. Mo¨nch), Du¨sseldorf (Hildegard Przyrembel, U. Wendel), Go¨ttingen (A.W. Behbehani, W. Voss), Hamburg (P. Koepp, P. Clemens), Heidelberg (Hildgund Schmidt, P. Lutz, K. Bartholome´, F.K. Trefz), Mu¨nchen (J. Schaub, W. Endres), Mu¨nster (H. Gro¨be, K. Ullrich), and Ulm (Dorothea Leupold). Additional acknowledgements appeared 1990 in Eur J Pediatr 149 (Supplement 1): S3–S4. Thanks are due to Sylvia Koerner for excellent administrative work and data handling, and to Elfriede Quak, Rainer Bielen and Verena Wahl for expert technical assistance with genotype analysis.
References Berman JL, Cunningham GC, Day RW, Ford R, Hsia DY (1969): Causes for high phenylalanine with normal tyrosine in newborn screening programs. Am J Dis Child 117: 54–65.
J Inherit Metab Dis (2009) 32:506–513 Blaskovics ME (1976): Diagnostic considerations in phenylalaninemic subjects before and after dietary therapy. J Irish Med 69: 410–414. Blaskovics ME, Schaeffler GE, Hack S (1974): Phenylalaninemia. Differential diagnosis. Arch Dis Child 49: 835–843. doi:10.1136/adc.49.11.835 Blau N (2005): BIOPKU: International database of patients and mutations causing BH4-responsive HPA/PKU. http://www. bh4.org/BH4DatabasesBioPKU.asp Bortz J, Lienert GA (1998) Kurzgefasste Statistik fu¨ r die klinische Forschung. Springer-Verlag, Berlin. Burgard P, Rupp A, Konecki DS, et al (1996): Phenylalanine hydroxylase genotypes, predicted residual enzyme activity and phenotypic parameters of diagnosis and treatment of phenylketonuria. Eur J Pediatr 155([Suppl 1]):S11–S15. doi:10. 1007/PL00014222 Fomon SJ, Haschke F, Ziegler EE, Nelson SE (1982) Body composition of reference children from birth to age 10 years. Am J Clin Nutr 35: 1169–1175. Fukami MH, Haavik J, Flatmark T (1990): Phenylalanine as substrate for tyrosine hydroxylase in bovine adrenal chromaffin cells. Biochem J 268: 525–528. Gjetting T, Petersen M, Guldberg P, Gu¨ttler F (2001): Missense mutations in the N-terminal domain of human phenylalanine hydroxylase interfere with binding of regulatory phenylalanine. Am J Hum Genet 68: 1353–1360. doi:10.1086/320604 Gramer G, Burgard P, Garbrade SF, Lindner M (2007): Effects and clinical significance of tetrahydrobiopterin supplementation in phenylalanine hydroxylase-deficient hyperphenylalaninaemia. J Inherit Metab Dis 30: 556–562. doi:10.1007/ s10545-007-0651-6 Guldberg P, Rey F, Zschocke J et al (1998): A European multicenter study of phenylalanine hydroxylase deficiency: Classification of 105 mutations and a general system for genotype-based prediction of metabolic phenotype. Am J Hum Genet 63: 71–79. doi:10.1086/301920 Keen RE, Spain JD (1992): Computer Simulation in Biology. A BASIC Introduction. Wiley-Liss, New York. Kindt E, Motzfeldt K, Halvorsen S, et al (1984) Is phenylalanine requirement in infants and children related to protein intake? Br J Nutr 51: 435–442. doi:10.1079/BJN19840049 Langenbeck U (2008): Classifying tetrahydrobiopterin responsiveness in the hyperphenylalaninaemias. J Inherit Metab Dis 31: 67–72. doi:10.1007/s10545-007-0572-4 Langenbeck U, Zschocke J, Wendel U, Ho¨ nig V (2001): Modelling the phenylalanine blood level response during treatment of phenylketonuria. J Inherit Metab Dis 24: 805– 814. doi:10.1023/A:1013946006155 Lutz P, Schmidt H, Frey G, Bickel H (1982): Standardized loading test with protein for the differentiation of phenylketonuria from hyperphenylalaninaemia. J Inherit Metab Dis 5: 29–35. doi:10.1007/BF01799751 Lutz P, Schmidt H, Batzler U (1990): Study design and description of patients. Eur J Pediatr 149(Supplement 1): S5–S12. doi:10.1007/BF02126292
513 Mo¨nch E, Kneer J, Jakobs C, et al (1990): Examination of urine metabolites in the newborn period and during protein loading tests at 6 months of age. Eur J Pediatr 149 (Supplement 1): S17–S24. doi:10.1007/BF02126294 O_Flynn ME, Holtzman NA, Blaskovics M, et al (1980): The diagnosis of phenylketonuria. A report from the collaborative study of children treated for phenylketonuria. Am J Dis Child 134: 769–774. Okano Y, Eisensmith RC, Gu¨ttler F, et al (1991): Molecular basis of phenotypic heterogeneity in phenylketonuria. N Engl J Med 324: 1232–1238. Pey AL, Martinez A (2005): The activity of wild-type and mutant phenylalanine hydroxylase and its regulation by phenylalanine and tetrahydrobiopterin at physiological and pathological concentrations: An isothermal titration calorimetry study. Mol Genet Metab 86(Supplement 1): S43– S53. doi:10.1016/j.ymgme.2005.04.008 Sachs L (2004) Angewandte Statistik, 11th edn. Springer-Verlag, Berlin Schmidt H, Lutz P, Batzler U (1989): Differentialdiagnose des erho¨ hten Phenylalanin-Blutspiegels im Sa¨ uglingsalter. Ergebnisse der deutschen Verbundstudie u¨ber Phenylketonurie (PKU) / Hyperphenylalanina¨mie (HPA). Monatsschr Kinderheilkd 137: 86–92. Scriver CR, Clow CL (1980): Phenylketonuria: epitome of human biochemical genetics. N Engl J Med 303: 1336– 1342, 1394–1400. Scriver CR, Waters PJ (1999): Monogenic traits are not simple: lessons from phenylketonuria. Trends Genet 15: 267–272. doi:10.1016/S0168-9525(99)01761-8 Thompson GN, Walter JH, Leonard JV, Halliday D (1990) In vivo enzyme activity in inborn errors of metabolism. Metabolism 39: 799–807. doi:10.1016/0026-0495(90)90122-S Tho´ro´lfsson M, Ibarra-Molero B, Fojan P, et al (2002): Lphenylalanine binding and domain organization in human phenylalanine hydroxylase: a differential scanning calorimetry study. Biochemistry 41: 7573–7585. doi:10.1021/ bi0160720 Treacy E, Pitt JJ, Seller K, et al (1996): In vivo disposal of phenylalanine in phenylketonuria: a study of two siblings. J Inher Metab Dis 19: 595–602. doi:10.1007/ BF01799832 van Spronsen FJ, van Rijn M, Dorgelo B, et al (2009) Phenylalanine tolerance can reliably be assessed at the age of 2 years in patients with PKU. J Inherit Metab Dis 32: 27– 31. doi:10.1007/s10545-008-0937-3 Waxman DJ, Azaroff L (1992): Phenobarbital induction of cytochrome P-450 gene expression. Biochem J 281: 577–592. Zschocke J, Hoffmann GF (1999): Phenylketonuria mutations in Germany. Hum Genet 104: 390–398. doi:10.1007/s00439 0050973 Zurflu¨h MR, Zschocke J, Lindner M (2008) Molecular genetics of tetrahydrobiopterin-responsive phenylalanine hydroxylase deficiency. Hum Mutat 29: 167–175. doi:10.1002/ humu.20637
