A Human Tyrosine Hydroxylase Isoform Associated with Progressive Supranuclear Palsy Shows Altered Enzymatic Activity*

A novel human tyrosine hydroxylase (HTH) messenger RNA subgroup generated by alternative splicing and characterized by the absence of the third exon was recently identified. The corresponding putative protein lacks 74 amino acids including Ser31 and Ser40, two major phosphorylation sites implicated in the regulation of HTH activity. These mRNA species are detected in adrenal medulla and are overexpressed in patients suffering from progressive supranuclear palsy, a neurodegenerative disease mostly affecting catecholaminergic neurons of the basal ganglia. In the present work, an HTH protein isoform lacking exon 3 was identified in human adrenal medulla. For this purpose, an antibody was raised against the HTH exon 3. The effect of the removal of exon 3 on the enzymatic activity of HTH was studied in vitro by comparing a purified recombinant fusion protein without exon 3 (glutathione S-transferase (GST)-HTHΔ3) to the equivalent protein containing exon 3 (GST-HTH3). In initial velocity conditions, GST-HTHΔ3 has 30% of the maximal velocity of GST-HTH3. Moreover, the skipping of exon 3 results in the absence of activation of GST-HTH by heparin and increases by 10-fold the retroinhibition constant for dopamine, demonstrating the involvement of exon 3 in the regulation of HTH enzymatic activity. The identification of a variably expressed HTH isoform that lacks an exon implicated in activity regulation supports the view that HTH alternative splicing contributes to the functional diversity within the catecholaminergic system and may be implicated in some neurological diseases.

A novel human tyrosine hydroxylase (HTH) messenger RNA subgroup generated by alternative splicing and characterized by the absence of the third exon was recently identified. The corresponding putative protein lacks 74 amino acids including Ser 31 and Ser 40 , two major phosphorylation sites implicated in the regulation of HTH activity. These mRNA species are detected in adrenal medulla and are overexpressed in patients suffering from progressive supranuclear palsy, a neurodegenerative disease mostly affecting catecholaminergic neurons of the basal ganglia.
In the present work, an HTH protein isoform lacking exon 3 was identified in human adrenal medulla. For this purpose, an antibody was raised against the HTH exon 3. The effect of the removal of exon 3 on the enzymatic activity of HTH was studied in vitro by comparing a purified recombinant fusion protein without exon 3 (glutathione S-transferase (GST)-HTH⌬3) to the equivalent protein containing exon 3 (GST-HTH3). In initial velocity conditions, GST-HTH⌬3 has 30% of the maximal velocity of GST-HTH3. Moreover, the skipping of exon 3 results in the absence of activation of GST-HTH by heparin and increases by 10-fold the retroinhibition constant for dopamine, demonstrating the involvement of exon 3 in the regulation of HTH enzymatic activity. The identification of a variably expressed HTH isoform that lacks an exon implicated in activity regulation supports the view that HTH alternative splicing contributes to the functional diversity within the catecholaminergic system and may be implicated in some neurological diseases.
Tyrosine hydroxylase (TH) 1 has been the subject of extensive investigation, largely because it catalyzes the rate-limiting step in the biosynthesis of catecholamines. The regulation of the TH level and of its enzymatic activity is thus a major mechanism for controlling the amount of these important amines in blood and synapses. TH is regulated by almost all possible mechanisms of transcriptional and post-transcriptional control including regulation of the transcription rate, alternative splicing of the premessenger, variable stability of the mRNA, translational control, and modulation of the enzymatic activity (1). Recently, it has been shown that some human TH (HTH) gene sequence variants are associated with abnormal TH enzymatic activity and may be involved in some neurodegenerative diseases. Indeed, point mutations in the coding sequence of HTH have been found that decrease the activity of the enzyme in patients with L-DOPA-responsive dystonia (2) and inherited juvenile L-DOPA-responsive parkinsonism (3). There is also emerging evidence that alternative splicing of TH premessenger RNA may have physiological and even pathological consequences (4,5). Initially, four HTH mRNA variants generated by alternative splicing were cloned and characterized by (i) the differential use of two splicing donor sites within exon 1 and (ii) the differential inclusion of exon 2 (4, 6, 7). All four HTH species are present in human catecholaminergic tissues (8,9). Interestingly, the amount of the species containing exon 2 (HTH-3 and HTH-4) is enhanced in pheochromocytoma, a tumor of adrenal medulla (4). The four mRNAs encode proteins that differ in the N-terminal regulatory region and present different enzymatic activities (1,10). These data suggest that alternative splicing of TH premessenger RNA is significant and prompt us to further study the diversity of HTH mRNAs. Recently, we identified previously undetected HTH mRNA species lacking exon 3 in the adrenal medulla (5). These novel variants are produced by the junction of the splicing donor sites of exons 1a-, 1b-, or 2-directly to the acceptor site of exon 4. By RNase mapping experiments, we have shown that exon 3 was skipped in 4 to 6% of the total HTH mRNA population in normal adrenal medulla, a proportion similar to that of the previously described HTH-3 and -4 mRNA species (11,12). These novel mRNA species were unusually abundant (5) in adrenal medulla of patients suffering from progressive supranuclear palsy (PSP), which is a severe neurodegenerative disease predominantly characterized by an alteration of the catecholaminergic neurons within the basal ganglia (13). The novel mRNA species encode putative HTH proteins lacking the 74 amino acids of the exon 3. This region includes Ser 31 and Ser 40 , two phosphorylation sites that participate in the regulation of HTH enzymatic activity (1). Consequently, it is likely that the alternative splicing of exon 3 has consequences for the regulation of catecholamine biosynthesis.
In the present work, we analyzed the expression of this novel HTH isoform lacking exon 3 (HTH⌬3) in human adrenals by Western blotting experiments. For that purpose, we raised an antibody against the HTH exon 3. We also tested the effect of exon 3 skipping on the in vitro enzymatic activity of HTH by the use of purified recombinant HTH proteins in fusion with glutathione-S-transferase (GST). It is clearly established that TH activity is positively regulated by phosphorylation and by binding to any of several polyanions such as heparin, polyglutamate, and phosphatidylserine, which change the conformational state of the enzyme (14). Interestingly, the heparin binding site on rat TH (RTH) has been described as being precisely localized within an exon corresponding to HTH exon 3 (15). We therefore tested the effect of exon 3 skipping on HTH activation by heparin and show that GST-HTH⌬3 enzymatic activity is insensitive to this molecule. Feedback inhibition by catecholamines is one of the major mechanisms for modulating HTH enzymatic activity, and the N-terminal part of the enzyme plays an important role in retroinhibition by dopamine (16). Thus we tested the effect of dopamine on GST-HTH⌬3 activity and found that HTH exon 3 is required for feedback inhibition.

EXPERIMENTAL PROCEDURES
Construction of the Expression Vectors-The complete coding sequence of HTH3 was excised from PET3a (Stratagene) (10) by NdeI-BamHI digestion. The fragment was blunted by Klenow polymerase and transferred into the SmaI restriction site of PGEX-2T (Amersham Pharmacia Biotech). This plasmid permits the production of HTH3 fused to Schistosoma japonicum GST. A sequence encoding the corresponding isoform of HTH but lacking exon 3 (HTH⌬3) was obtained by deletion of exon 3 from PGEX-2T-HTH3 as described in Lanièce et al. (17). Briefly, two amplification products of the sequences corresponding to HTH exons 1 ϩ 2 and exons 4 ϩ 5 were joined by polymerase chain reaction, and the final amplification product was used to replace the HTH3 sequence between the BamHI and XhoI sites in PGEX-2T-HTH3. The primers used were (i°) ccataatggatccATGCCCACCCCCGACGCC (capital letters, HTH sequence beginning with initiator codon; bold, BamHI site) and GCTTCAAACGT-CTTGGGGTGGG (junction exon2exon 4, antisense) and (ii°) CCACCCCAAG-ACGTTTGAAGC (junction exon2 -exon 4, sense) and CGGGTGGTCCAAGTCCAG (within exon 5). The PET3a-HTH⌬3 construct was obtained from PET3a-HTH3 by deletion of the third exon using the same method. All amplified sequences were checked by sequencing.
Purification of GST-HTH Fusion Proteins-PGEX-2T-HTH3 and -HTH⌬3 constructions were introduced into electrocompetent AX90 Escherichia coli cells. Production of the fusion proteins (GST-HTH3 and GST-HTH⌬3) was induced by isopropyl-1-thio-␤-D-galactopyranoside, and the proteins were affinity-purified on glutathione-agarose beads according to the manufacturer's instructions (Amersham Pharmacia Biotech) with minor modifications. E. coli cells were harvested at 25°C instead of 37°C to avoid formation of inclusion bodies. After elution from the glutathione-agarose beads, fusion proteins were extensively dialyzed against phosphate-buffered saline at 4°C, then to desalt, concentrate, and eliminate low molecular weight protein contaminants, fusion proteins were centrifuged on Microcon-100 microconcentrators (Amicon, 100-kDa molecular mass cut-off columns). Protein purity was confirmed by electrophoresis on a 9% polyacrylamide denaturing gel followed by Coomassie Blue staining. The protein quantities were first estimated with the Bradford method, then precisely measured by comparing the intensity of the Coomassie Blue-stained band to a standard curve obtained with bovine serum albumin (Image 1.59 software).
Measuring of GST-HTH Activities-GST-HTH3 and -HTH⌬3 enzymatic activities were tested by a radioenzymatic assay as described in Alterio et al. (10) with minor modifications. Briefly, the enzymatic activities were measured as the release of tritiated water from L-[ 3 H]tyrosine at 30°C in 0.1 M Hepes, pH 7, 40 M tyrosine, 0.5 mg/ml catalase, 0.1 mM FeSO4, 0.32 Ci of L-[ 3 H]tyrosine (Amersham Pharmacia Biotech) following addition of the cofactor BH4 (250 to 3000 M) diluted in 20 mM dithiothreitol. Fusion proteins were diluted before activity assay in 0.1 M Hepes, pH 7, containing 1 g/l bovine serum albumin. Each activity was measured in triplicate, and further experiments were performed with several independant enzyme batches.
We first established the initial velocity conditions in which the activity increases linearly with time and enzyme quantity. Using different BH4 concentrations (from 250 to 3000 M), Lineweaver-Burk curves were then traced for both isoforms with the Kaleidagraph software. The experimental results were fitted for the inhibition by excess of BH4 substrate (10) with the following equation Dixon and Webb (18), where V m is the maximal velocity of the enzyme, K m is the apparent Michaelis constant for BH4, and K i the inhibition constant for BH4.
To analyze the regulation of GST-HTH3 and -HTH⌬3 by polyanions, heparin (Calciparine Sanofi, France) was added to the assay mix (5 to 15 IU) before the addition of BH4 (500 M). Similarly, final product retroinhibition was tested by the addition of dopamine (3-hydroxytyramin; Sigma) (10 to 100 M) to the reaction before the addition of BH4 (500 M). The inhibition constant for dopamine (K i ) was determined graphically for both isoforms by plotting 1/V as a function of dopamine concentration for 3 different BH4 concentrations (250, 500, and 1000 M). The point of intersection of the 3 lines gives K i directly (Dixon method, see Ref. 18).
Synthesis of anti HTH-exon 3 Antibody-The nucleotide sequence corresponding to the 59 first amino acids of the third exon of HTH was amplified by polymerase chain reaction with oligonucleotides ccataat-ggatccCGGTTCATTGGGCGCAGG and ccacaatgaatCCTCGGGGAG-AAGAGCAG, purified, digested, and introduced between the BamHI and EcoRI restriction sites of pGEX-2T (Amersham Pharmacia Biotech). The amplified products were checked by sequencing. The GST-HTHexon3 fusion protein was produced in AX90 cells and purified according to the manufacturer's instructions (Amersham Pharmacia Biotech). The fusion protein was dialyzed against 20 mM Tris, pH 7.5, 0.1 M NaCl, 0.2 mM EDTA, 1 mM dithiothreitol and sent to Eurogentec (Ougrée, Belgium) for antiserum production in rabbits. The anti-HTHexon 3 antiserum was purified by affinity using the GST-HTHexon3 antigen immobilized on Immobilon-P membranes (Millipore) using standard methods. The immunoreactivity and specificity of this antibody was checked by Western blotting experiments with total lysate from BL21 bacteria (Stratagene) transformed with PET-3a-HTH1 and PET-3a-HTH⌬3.
Western Blotting Experiments-Frozen post-mortem human adrenal glands from controls and PSP patients were obtained from the brain bank of Inserm U 289 (Hôpital de la Pitié-Salpêtrière, Paris) and from Dr. P. F. Plouin (Hôpital Broussais, Paris). PSP was clinically characterized, and the diagnosis was confirmed as described in Dumas et al. (5). Tissues were lyzed in 10 mM Tris, pH 7.5, 5 mM EDTA supplemented with 1 mM phenylmethyl sulfonyl fluoride; Sigma) and centrifuged for 10 min at 4°C. The total protein concentration was estimated by the Bradford method. Western blotting experiments were performed by standard procedures with an anti-RTH polyclonal antibody provided by Dr. J. F. Reinhard (19) and 125 I-protein A. The membrane was stripped in 0.1 M glycine, pH 2.8, 0.05% Tween 20 at 80°C and reprobed with the anti-HTHexon3-purified antiserum. The amount of the HTH⌬3 isoform in the adrenal samples was estimated with the Image 1.59 software.

Identification of an HTH Isoform Lacking Exon 3 in Human
Adrenals-To test human tissues for the presence of an HTH protein isoform lacking exon 3, we raised and purified an antiserum against the third exon of HTH. We ascertained the immunoreactivity and specificity of affinity-purified anti-HTHexon3 antibodies by Western blotting experiments with HTH⌬3 (containing exons 1a,2,4 -14) and HTH1 (containing exons 1a, 3,4 -14) produced in E. coli by the PET-3a expression system. As expected, this anti-HTHexon 3 antibody only recognized the HTH1 isoform (Fig. 1B), whereas a polyclonal anti-RTH antibody (19) recognized both HTH isoforms (Fig. 1A), attesting to the specificity of the anti-HTHexon3 antibody.
Then, to determine the expression profile of the HTH⌬3 isoform in human tissues, total protein extracts from control and PSP adrenal medullas were probed with both antibodies by Western blotting experiments. Recombinant HTH proteins expressed from the PET-3A system in E. coli were used as size controls. With the polyclonal anti-rat TH antibody (19), several bands were detected in these tissue samples ( Fig. 2A). The two slower migrating bands correspond, respectively, to HTH 3 and 4 and HTH 1 and 2, as described previously (20). Two other isoforms of approximately 55 and 50 kDa were found in all tissue samples; they were also detected in a noncatecholaminergic rat cell line (ST14A) stably transfected with HTH1-cDNA, showing that were not the result of alternative splicing.
Interestingly, as seen on Fig. 2A, the polyclonal anti-rat TH antibody detected a novel protein of the same size as recombinant HTH⌬3 in the PSP adrenal medulla sample. To confirm the identity of this isoform, the membrane was stripped and reincubated with the anti-HTHexon3 antibody (Fig. 2B). The novel protein was not recognized by this antibody, evidencing that it lacks exon 3 and is certainly HTH⌬3. This isoform is present in a substantial amount in PSP adrenal medulla and is present in much lower amounts in normal adrenal medullas (Fig. 2).
GST-HTH⌬3 Presents 30% of GST-HTH3 Maximal Velocity in Vitro-To analyze the involvement of exon 3 in HTH enzymatic activity, the cDNAs encoding HTH⌬3 and HTH3 (the equivalent isoform containing exon 3) were inserted into PGEX-2T (Amersham Pharmacia Biotech), allowing the synthesis of GST fusion proteins. Both GST-HTH⌬3 and GST-HTH3 fusion proteins were produced and purified from total bacterial lysate on an affinity column. The integrity and purity of both fusion proteins were ascertained by SDS-polyacrylamide gel electrophoresis and Coomassie Blue staining (Fig. 3). The activities of the recombinant GST-HTH proteins were measured by a radioenzymatic activity assay. The absence of activity presented by GST alone in this assay was ascertained (data not shown). The initial velocity conditions in which product formation increases linearly with time and enzyme concentration were determined (Fig. 4), and all further experiments were performed in the following conditions: 50 ng of fusion protein, 5 min reaction time. The GST-HTH activities were measured with various BH4 concentrations from 400 to 3000 M, and the Lineweaver-Burk plots were traced, showing typical excess substrate inhibition (Fig. 5). The kinetic parameters were calculated according to the classical analysis of substrate inhibition kinetics, as described under "Experimental Procedures." The maximal velocities of GST-HTH3 and -HTH⌬3 were, respectively, 1400 (Ϯ70) and 450 (Ϯ 40) pmols of DOPA/ min/g of enzyme, which shows that GST-HTH⌬3 has approximately 30% GST-HTH3 maximal velocity. The absence of exon 3 did not significantly change the Michaëlis constant (K m ) of the enzyme for the cofactor BH4 (174 Ϯ 24 M for GST-HTH3, 180 Ϯ 50 M for GST-HTH⌬3) nor the inhibition constant (K i ) for BH4 (4100 Ϯ 500 M for GST-HTH3, 2800 Ϯ 500 M for GST-HTH⌬3).
The Exon 3 Participates in the Regulation of HTH by Dopamine and Heparin-HTH enzymatic activity is tightly controlled in vivo, subject to down-regulation by catecholamines as the end products of the biosynthetic chain, and activated by phosphorylation and polyanion binding (1).
Because retroinhibition by catecholamines is one of the major mechanisms for controlling TH activity, we tested the effect of exon 3 skipping on the regulation of HTH activity by dopamine. Both GST-HTH3 and GST-HTH⌬3 activities were analyzed with different concentrations of dopamine from 10 to 100 M. As seen on Fig. 6B, GST-HTH3 was more strongly inhibited; its activity was reduced to 50% in the presence of 10 M dopamine, whereas GST-HTH⌬3 was only decreased by 10%. To further characterize the kinetic parameters of the inhibition of GST-HTH proteins by dopamine, we measured GST-HTH3 and -HTH⌬3 activities at various dopamine concentrations in the presence of three different quantities of BH4 (Fig. 7). We graphically determined the inhibition constant (K i ) of both GST-HTH isoforms for dopamine according to the Dixon method (18). The calculated K i for dopamine was increased from 1 to 10 M by skipping exon 3, demonstrating that the exon 3 is implicated in the retroinhibition of TH by dopamine.
We also studied the effect of exon 3 skipping on the activation of HTH by heparin. GST-HTH⌬3 and -HTH3 activities were measured in the presence of a series of concentrations of heparin (Fig. 6A). GST-HTH3 activity increased gradually with heparin concentration (up to 20%), whereas no significant activation of GST-HTH⌬3 was observed, strongly implicating exon 3 in the activation of HTH by heparin.
Altogether these observations provide the first demonstration that exon 3 is implicated in the HTH overall catalytic activity and, most importantly, in the regulation of HTH activity by polyanions and by catecholamines. DISCUSSION A great diversity of mechanisms controlling TH activity has been described, and among these, alternative splicing of HTH pre-mRNA is thought to be significant (25). In the present study, we evidenced the presence in adrenal medulla of a novel HTH isoform, HTH⌬3, that lacks the third exon comprising two major phosphorylation sites, Ser 31 and Ser 40 . This isoform cor-responds to the translation of a previously unsuspected HTH mRNA generated by alternative splicing recently identified in adrenal medulla and overexpressed in adrenals of patients suffering from PSP, a neurodegenerative disease affecting predominantly catecholaminergic neurons of the basal ganglia (5). We also established that the exon 3 participates in the regulation of HTH enzymatic activity in vitro.
The presence of an HTH protein isoform lacking the third exon in vivo was evidenced by Western blotting experiments. This HTH isoform, having the size of recombinant HTH⌬3, was not recognized by an antibody specifically raised against HTH exon 3, whereas it was clearly detected by a polyclonal antibody directed against whole rat TH recognizing all HTH isoforms. This protein isoform was clearly detected in PSP adrenal medulla and was about six times less abundant in control adrenal medullas. Substantial amounts of the novel isoform were also found in all other PSP adrenals tested (data not shown), in accordance with the higher levels of HTH⌬3 mRNA previously observed in PSP adrenals (5). This variable expression of the HTH⌬3 isoform indicate that alternative splicing of HTH exon 3 may be significant. The increased amount of HTH⌬3 in PSP suggests that the splicing alteration may be a cause of the disease. This schema involving a poorly expressed truncated enzyme isoform generated by alternative splicing has already been proposed by Hirano et al (42). Indeed, some familial patients with DOPA-responsive dystonia have been shown to express a small proportion of incorrectly spliced mRNA encoding GTP-cyclohydrolase I (the limiting enzyme in the synthesis of BH4), and the amount correlates with the severity of the symptoms. Because several familial cases of PSP have been described (39), we searched by sequencing for mutations at exon 3 splice junctions within HTH gene on genomic DNA extracted from the cerebellum of 10 PSP patients. Among these, only three adrenal medullas were available, and they show an expression of the HTH⌬3 protein (data not shown). No mutation was found in a region of 100 bases around the donor and acceptor splice junctions of exon 3 (data not shown), indicating that the modification of the splicing of exon 3 is not because of a mutation of the HTH sequence in these patients. Alternatively, there may be a modification in the amount of FIG. 4. Enzymatic activity of GST-HTH3 and GST-HTH⌬3 fusion proteins: initial velocity conditions. TH activity was measured with 25 to 100 ng of purified fusion protein by a radioenzymatic assay using L-[ 3 H]tyrosine as described under "Experimental Procedures" with various incubation times. The activity is expressed as pmols of DOPA according to the radioactive counts. A, pmols of DOPA increase linearly with time for both isoforms; B, the slopes of the curves obtained in A were plotted against enzyme quantity, showing that pmols of DOPA/min increases linearly with the amount of enzyme. splicing factors influencing the choice of splice junctions and the inclusion/exclusion of alternative exons, such as SF2/ASF and heterogeneous nuclear ribonucleoprotein A1 (40).
Interestingly, in Western blot experiments, an unidentified 55-kDa HTH isoform that was not the result of alternative splicing was detected in all tissue samples. This isoform has already been reported in previous studies (8,20) but was interpreted as a degradation product. However, it may rather be a genuine HTH isoform. Indeed, its size corresponds to (i) the use of a second AUG initiator codon located at the end of exon 1 that matches perfectly with a translation initiation sequence (21) and/or to (ii) the result of a cleavage of HTH by calpain (Ca 2ϩ -activated neutral protease) (22)(23)(24). The possibility that this 55-kDa isoform is the proteolysis calpain cleavage product is of particular interest. Indeed, calpain is present in catecholaminergic tissues and activates TH in vitro (22). The 55-kDa isoform seems to be particularly abundant in PSP adrenals (Fig. 2) but further experiments are needed to confirm its identity and any link with PSP disease.
To evaluate the role of exon 3 specifically, the enzymatic activity of the purified fusion proteins GST-HTH3 and GST-HTH⌬3, which differ only by the presence of exon 3, were compared. This work was carefully performed in conditions in which such a comparison is meaningful, e.g. the two fusion proteins were (i) purified in the same way and (ii) tested precisely in initial velocity conditions with exactly the same protocol. In these conditions, the GST-HTH proteins displayed some of the enzymatic properties characteristic of TH, in particular inhibition by excess BH4, heparin activation, and dopamine retroinhibition. Moreover, the two fusion proteins displayed kinetic parameters (K m and K i for BH4) of the same magnitude. However, because of the possible influence of GST on HTH activity, the kinetic parameters obtained can only be used to compare GST-HTH3 and GST-HTH⌬3 proteins and thereby to evaluate the role of exon 3. Our results show that GST-HTH⌬3 presents in vitro 30% of the maximal velocity of GST-HTH3, the equivalent isoform containing exon 3. Interestingly, this lower GST-HTH⌬3 maximal velocity is not associated with a modification of the Michaelis constant (K m ) for the cofactor BH4 nor of the inhibition constant (K i ) for BH4. Thus, exon 3 may influence the accessibility of some other components such as tyrosine and/or ferrous iron, possibly by modifying the conformation of the enzyme. It is unlikely that the skipping of exon 3 simply affects the catalytic process itself, as exon 3 is not localized within the catalytic part of the enzyme.
The third exon of HTH, absent from the novel isoform, is located in the N-terminal regulatory part of the enzyme (26). This N-terminal region has been implicated in the action of numerous allosteric regulatory substances, including polyanions like heparin. Little is known about possible direct actions of heparin on enzymatic activities within the cell. However, it has been hypothetized that a direct interaction can occur between intracellular proteins and heparin in the Golgi complex or that heparin might mimic the effects of intracellular heparin-like factors (27). Concerning TH, it has been shown that heparin reversibly activates the enzyme probably by changing its conformation (14,28). Here, we show that exon 3 is involved in the activation of HTH by heparin because GST-HTH3 is activated by heparin, whereas GST-HTH⌬3 is not. This result is consistent with other studies evidencing that in rat TH, the heparin binding site is precisely localized within an exon homologous to HTH exon 3 (15). As the interaction of the HTH N-terminal part with heparin involves a similar mechanism as several polyanions (1), it is probable that the absence of exon 3 also affects the activation of HTH by other physiological polyanions such as phospholipids, polyglutamic acid, and nucleic acids (14,28).
Feedback inhibition of TH enzymatic activity by catecholamines is considered as the primary mechanism by which this amine biosynthesis is regulated (29). The regulation of TH activity by dopamine has been extensively studied (1,30,31) and shown to involve a portion of the N-terminal part of the enzyme close to Ser 40 (16). Thus, we tested if HTH exon 3 played a role in the retroinhibition process. We found that the skipping of the HTH third exon increases the inhibition constant (K i ) of the GST-HTH protein for dopamine by approximately 10-fold, demonstrating that exon 3 is implicated in the inhibition of HTH activity by dopamine. Generally, phosphorylation of TH and retroinhibition by catecholamines act in coordination to regulate TH activity according to the cellular requirement. For example, phosphorylation of Ser 40 , mediated by most second messenger systems, activates TH and partially relieves a tonical inhibition by increasing the K i of the enzyme for dopamine (32,33). The novel HTH⌬3 isoform lacks Ser 40 , and this may explain its escape from dopamine regulation. It also lacks Ser 31 , the phosphorylation of which by two mitogenactivated protein kinases (ERK1 and 2) has been shown to activate TH in vitro (34,35). However, it seems clear that the functions carried by exon 3 are more complex than those carried by Ser 40 and Ser 31 alone, because the skipping of exon 3 (i) also abolishes activation by heparin and (ii) decreases the activity of TH in vitro, whereas the directed mutagenesis of Ser 40 results in an activated TH (32). Thus, the HTH⌬3 protein, lacking such an important exon, probably exhibits particular properties in vivo and has a genuine physiological role. The remaining phosphorylation site for CamPKII, Ser 19 (35), may be the only site at which this isoform can be regulated.
Native TH assembles into tetramers by a stochastic process (9), and tetramerization is required for maximal activity (36). The N-terminal region of the protein, including exon 3, is not necessary for the formation of the tetramer (37); thus, it is unlikely that skipping exon 3 would prevent tetramer formation. However, if we assume that the various HTH isoforms could be incorporated into heterotetramers, a large number of combinations than previously described are possible, and the regulation of HTH enzymatic activity is more complex than suspected. It would also be of importance to establish the existence and the mechanisms of any possible dominant effect of HTH⌬3 on the regulation of the enzymatic activity of the heterotetramer. Any such dominant effect would result in the enzyme activity being constantly maximal, in particular in PSP patients where HTH⌬3 is abundant. As TH activity generates reactive free radicals in vitro (43), such a decrease of HTH regulability could result in oxidative stress and contribute to the degeneration of catecholaminergic cells in these patients.
The novel HTH⌬3 mRNA was found in adrenal medulla but was not detected in substantia nigra from a control (5). Possibly, the differential expression of the HTH⌬3 isoform between central and peripheral nervous system contributes to the functional diversity within the catecholaminergic system. Indeed, adrenomedullary chromaffin cells are terminally differentiated secretory cells of neural crest origin dedicated to the synthesis, storage, and release of adrenaline and noradrenaline and are phenotypically and functionally distinct from catecholaminergic neurons of the central nervous system (38). Another argument for the possible link between alternative splicing of TH premessenger RNA and phenotypic diversity within the catecholaminergic system is the increased amount of HTH isoforms containing exon 2 (HTH3 and 4) in the adrenal medulla compared with the central nervous system (4). This is also probably significant because the isoform 3 has recently been shown to present specific regulatory properties after cAMPdependant protein kinase phosphorylation (10).
It is clear that exon 3 plays a role in the regulation of HTH enzymatic activity. Interestingly, this part of TH may also have a role in other species. Indeed, the region homologous to the HTH exon 3 is also alternatively spliced in rat (17) and in Drosophila (41) (Fig. 8). Moreover, in Drosophila, the TH isoform that differs in the region corresponding to exon 3 is specifically expressed in the hypoderm and has been proposed to be hyperactive and contribute to the synthesis of the large amounts of dopamine within the cuticle.
Finally, the identification of a variably expressed HTH isoform that lacks an exon implicated in enzymatic activity regulation supports the view that HTH alternative splicing may be involved in some neurological diseases and/or in the functional diversity of the catecholaminergic system. Ackowledgements-We are grateful to Dr. J. F. Reinhard for providing the anti-TH antibody. Thanks to O. Corti for providing the HTH1transfected cell line. Special thanks to V. Berthelier for technical advice and helpful discussions.