Stabilization of tryptophan hydroxylase 2 by l‐phenylalanine‐induced dimerization

Tryptophan hydroxylase 2 (TPH2) catalyses the initial and rate‐limiting step in the biosynthesis of serotonin, which is associated with a variety of disorders such as depression, obsessive compulsive disorder, and schizophrenia. Full‐length TPH2 is poorly characterized due to low purification quantities caused by its inherent instability. Three truncated variants of human TPH2 (rch TPH2; regulatory and catalytic domain, NΔ47‐rch TPH2; truncation of 47 residues in the N terminus of rch TPH2, and ch TPH2; catalytic domain) were expressed, purified, and examined for changes in transition temperature, inactivation rate, and oligomeric state. ch TPH2 displayed 14‐ and 11‐fold higher half‐lives compared to rch TPH2 and NΔ47‐rch TPH2, respectively. Differential scanning calorimetry experiments demonstrated that this is caused by premature unfolding of the less stable regulatory domain. By differential scanning fluorimetry, the unfolding transitions of rch TPH2 and NΔ47‐rch TPH2 are found to shift from polyphasic to apparent two‐state by the addition of l‐Trp or l‐Phe. Analytical gel filtration revealed that rch TPH2 and NΔ47‐rch TPH2 reside in a monomer–dimer equilibrium which is significantly shifted toward dimer in the presence of l‐Phe. The dimerizing effect induced by l‐Phe is accompanied by a stabilizing effect, which resulted in a threefold increase in half‐lives of rch TPH2 and NΔ47‐rch TPH2. Addition of l‐Phe to the purification buffer significantly increases the purification yields, which will facilitate characterization of hTPH2.

Tryptophan hydroxylase 2 (TPH2) catalyses the initial and rate-limiting step in the biosynthesis of serotonin, which is associated with a variety of disorders such as depression, obsessive compulsive disorder, and schizophrenia. Full-length TPH2 is poorly characterized due to low purification quantities caused by its inherent instability. Three truncated variants of human TPH2 (rchTPH2; regulatory and catalytic domain, ND47-rchTPH2; truncation of 47 residues in the N terminus of rchTPH2, and chTPH2; catalytic domain) were expressed, purified, and examined for changes in transition temperature, inactivation rate, and oligomeric state. chTPH2 displayed 14-and 11-fold higher half-lives compared to rchTPH2 and ND47-rchTPH2, respectively. Differential scanning calorimetry experiments demonstrated that this is caused by premature unfolding of the less stable regulatory domain. By differential scanning fluorimetry, the unfolding transitions of rchTPH2 and ND47-rchTPH2 are found to shift from polyphasic to apparent two-state by the addition of L-Trp or L-Phe. Analytical gel filtration revealed that rchTPH2 and ND47-rchTPH2 reside in a monomer-dimer equilibrium which is significantly shifted toward dimer in the presence of L-Phe. The dimerizing effect induced by L-Phe is accompanied by a stabilizing effect, which resulted in a threefold increase in halflives of rchTPH2 and ND47-rchTPH2. Addition of L-Phe to the purification buffer significantly increases the purification yields, which will facilitate characterization of hTPH2.
Tryptophan hydroxylase (TPH) catalyses the rate-limiting reaction in the biosynthesis of the hormone and neurotransmitter serotonin (5-HT). TPH uses the cofactor Fe 2+ and the cosubstrates O 2 and tetrahydrobiopterin (BH 4 ) to generate 5-hydroxytryptophan (5-HTP) by hydroxylation of L-tryptophan (L-Trp). 5-HTP is enzymatically converted to 5-HT by aromatic amino acid decarboxylase [1,2]. TPH exists in two isoforms, where isoform 1 (TPH1) is mainly responsible for catalyzing the rate-limiting step in the biosynthesis of peripheral serotonin, and isoform 2 (TPH2) primarily is associated with the biosynthesis of neuronal serotonin [1]. The regulation of serotonin levels is involved in various physiological and psychiatric disorders such as irritable bowel syndrome, depression, obsessive compulsive disorder, and schizophrenia [3].
Of these, depression is associated with decreased levels of neuronal serotonin, whereas some gastrointestinal disorders are associated with increased peripheral serotonin levels [4][5][6]. The serotonergic systems are therefore important targets in treatment of these disorders [7], which makes both TPH isoforms important enzymes to characterize.
All enzymes in the AAAH family form tetramers through dimers of dimers [18]. Each monomeric subunit consists of an N-terminal regulatory domain, a highly conserved catalytic domain, and a C-terminal tetramerization domain [19]. The N-terminal domains of PAH and TH contain a characteristic ACT fold motif [12,20]. From sequence analysis, TPH has also been proposed to contain an ACT domain [21]. Truncations of the regulatory domains have been found not to change the tetrameric assembly of TPH1 or TPH2 [1,22]. In contrast, when the C-terminal tetramerization domain is also removed, monomeric species are observed [1]. Deletions from the C-terminal domain of rabbit TPH1 have been demonstrated to change the macromolecular structure from a tetramer to predominantly monomeric form without compromising catalytic activity [23,24].
The available crystal structures of TPH1 only comprise the catalytic domain, while the crystal structure of TPH2 includes both the catalytic and the tetramerization domains [13,14,[25][26][27]. The lack of structural information of the regulatory domain is partly due to the great instability of full-length TPH [16,28]. In TPH, the regulatory domain, in particular, has been found to cause instability and therefore limiting the quantities purified from Escherichia coli expression systems [1,16]. The regulatory domain of TPH2 has an additional 46 residues compared to TPH1, which are partly responsible for the limited purification quantities [29]. The fundamental role of the regulatory domain and the additional residues in the terminus is, however, poorly understood, as characterization is hampered by the limited purification yields. Therefore, structural insight is obtained from the crystal structure of rat phenylalanine hydroxylase (rnPAH) (PDB ID: 1PHZ -Clustal Omega: 35% sequence identity in the regulatory domain and 57% total sequence identity) which comprises the catalytic and the regulatory domains. This structure lacks interpretable electron density in the first 18 residues of the N terminus, suggesting that this region is flexible [20]. The flexibility of the N terminus has also been confirmed by an NMR study [30]. By alignment of the sequences of rnPAH and hTPH2 (Fig. S1), it is found that residues 20 to 47 of hTPH2 align with the mobile 18 residues of rnPAH, which suggests that these residues of hTPH2 are mobile and might cause instability and insolubility. Therefore, an ND47-rchTPH2 variant was expressed, purified, and characterized.
The current study sheds light on the influence of the regulatory domain on macromolecular structure and stability of TPH2 by characterizing three truncated variants (sequences shown in Supporting information); chTPH2 (catalytic domain), rchTPH2 (regulatory and catalytic domains), and ND47-rchTPH2 (47 residue truncation in the N-terminal domain). The tetramerization domain (residue 460-490) was removed to investigate the role of the regulatory domain when TPH2 was not in a tetramer. As TPH2 is poorly characterized due to low stability, additional efforts were made to identify ligands that could increase the stability, and hence purification yield, of TPH2.

Differential scanning fluorimetry
Very little is known about the regulation and structure of TPH as purification of this enzyme results in limited quantities. The presence of the regulatory domain is known to cause low stability and solubility [1]. To overcome this problem, substrates for the AAAH family (L-phenylalanine, L-tryptophan, and L-tyrosine), other ligands known to bind certain ACT domains (Lvaline [31] and L-serine [32]), as well as D-phenylalanine, 5-HTP, and 5-HT were assayed for changes in the thermal unfolding of the TPH2 variants. The ligands were screened in a broad concentration range (0.1 lM-10 mM). For chTPH2, an almost ideal twostate unfolding behavior was observed, Fig. 1. All ligands, except 5-HTP and 5-HT, were found to have no significant effect on the unfolding or T m value of chTPH2, Fig. 1. 5-HTP and 5-HT appear to change the unfolding of chTPH2 from two-state to continuous; hence, no T m values could be obtained at high ligand concentration.
Containing the regulatory domain, rchTPH2, gave rise to inconsistent and polyphasic unfolding curves from which no T m values could be obtained (Fig. S2). The ligand screen showed that L-Trp and L-Phe induced unfolding transitions with apparent two-state unfolding in a concentration-dependent manner (Fig. 2), which were accompanied by an increase in transition temperatures. The apparent two-state unfolding was gradually acquired with increasing ligand concentration, and robust apparent two-state unfolding was obtained at L-Trp ( Fig. 2A)   10 mM L-Trp. The transition temperature of rchTPH2 was increased from 48.2 AE 1.9 to 51.0 AE 0.9°C by increasing the L-Phe concentration from 1.0 mM to 10.0 mM (Fig. 2D). Similarly, a study by Gersting et al. [33] utilizing differential scanning fluorimetry (DSF) found that 1 mM L-Phe increases the transition temperature of PAH from 47.5 to 50.9°C. None of the other ligands, including D-Phe, were able to induce this change in unfolding (data not shown). The same stabilizing trend for L-Phe and L-Trp was observed for ND47-rchTPH2, Fig. 2E. As both L-Trp and L-Phe gave rise to increased transition temperatures, a combination of the compounds was analyzed. Figure 3 shows that increasing the concentration of either L-Trp or L-Phe in the presence of the other compound results in increased T m values, suggesting that L-Trp and L-Phe are able to increase the transition temperature in an additive fashion. The additive effect is supported by the increase in T m values with increasing L-Phe concentration observed at saturated concentration of L-Trp (10 mM in Fig. 3).

Analytical gel filtration
It has previously been shown in DSF assays that under stabilizing conditions multi-component complexes can change from polyphasic to almost two-state unfolding and that this is indicative of monodispersity and stability [34]. This was investigated in the case of L-Phe and L-Trp as unfolding of rchTPH2 and ND47-rchTPH2 was shifted toward two-state with the addition of these compounds. The oligomeric states of the TPH2 variants were analyzed utilizing analytical size exclusion chromatography (SEC). In consensus with the findings of D'Sa et al. [22], chTPH2 was found to elute at a volume corresponding to the molecular weight of a monomer (36.2 kDa) [35]. In the loading concentration range of 2 to 60 lM, chTPH2 was found only to reside as a monomer, Fig. 4. Additionally, 3 mM L-Phe did not induce any change in the elution pattern of the monodisperse solution complementing the observations from the DSF experiments.
A study by Mockus et al. [23] on rabbit TPH1 has demonstrated that truncation of the C-terminal tetramerization domain resulted in a disruption of the tetrameric assembly. This was also found to be the case for hTPH2, as SEC of ND47-rchTPH2 yielded two overlapping peaks with elution volumes corresponding to molecular weights of a monomer (47.4 kDa) and a dimer (94.9 kDa) (Fig. 5). The gradual shift of the elution peaks from a monomer to a dimer shows that ND47-rchTPH2 is found to be in a concentration-dependent monomer-dimer equilibrium. Within experimental error, the peak widths at half height were constant over the range of TPH concentrations, and the peak heights were also found to be related directly to the concentration of TPH injected. The dilution factor was found to be 1.90 AE 0.13 and constant within the elution range of the TPH variants. From the SEC results, the equilibrium constant, K d , of the dissociation of a dimer into monomers was calculated (equation (7) in Materials and methods). A K d value of 1.3 AE 0.1 lM was found based on the six concentrations of ND47-rchTPH2 in the absence of L-Phe. This value is lower than the dissociation constant of 46 AE 35 lM determined for the dimerization of the regulatory domain of PAH [36]. A plot of fraction dimer, f D , as a function of ND47-rchTPH2 concentration gave a hyperbolic-like curve as expected for a monomer-dimer equilibrium, Fig. 6. The difference in oligomeric states of chTPH2 and rchTPH2/ND47-rchTPH2 shows that dimerization is caused by the regulatory domain. This finding is supported by a model structure proposed by Jiang et al. [37] in which the regulatory domain of one monomer forms intersubunit interactions to an adjacent monomer. The involvement of the regulatory domain in structural assembly is further supported by results from Yohrling et al. [38] on rabbit TPH1, where ND41 and ND90 truncations resulted in monomers indicating that the regulatory domain is involved in the formation of the tetrameric assembly. D'Sa et al. [22], however, found that the tetrameric assembly was retained upon truncation of the regulatory domain of human TPH2. Collectively, this might suggest that both the C-terminal tetramerization domain and the N-terminal regulatory domain contribute to the formation of a stable tetramer.
The effect of L-Phe on the monomer-dimer equilibrium was analyzed by performing SEC on ND47-rchTPH2 samples with and without the addition of L-Phe. In Fig. 5, the black curves represent elution of ND47-rchTPH2 without the addition of L-Phe, and the orange curves represent the elution of protein when the sample and running buffer contained 3 mM L-Phe. It is evident that L-Phe shifts the equilibrium toward dimer. At a loading concentration of 2 lM ND47-rchTPH2, the addition of L-Phe was found to shift the equilibrium from predominately monomer to almost exclusively dimer. The same shift in monomerdimer equilibrium was observed for rchTPH2 (Fig. 7A). This result correlates well with the observations in the DSF experiments, where the unfolding curves change from polyphasic toward two-state with the addition of L-Phe. The influence of L-Trp and D-Phe on the monomerdimer equilibrium was additionally analyzed to investigate the specificity of L-Phe. Figure 5C,D shows the monomer-dimer distribution of ND47-rchTPH2 with 0.5 mM L-Trp or 3 mM D-Phe in a protein concentrations range from 2 to 60 lM. From the SEC results in  Fig. 5B,D that neither the addition of 0.5 mM L-Trp nor 3 mM D-Phe induced the shift in the monomer-dimer equilibrium toward dimer as observed for L-Phe. L-Trp absorption introduced high background noise and was hence only added to concentrations of 0.5 and 1.0 mM. As a consequence of the background noise, chromatograms of ND47-rchTPH2 with loading concentrations below 5 lM could not be obtained at 1.0 mM L-Trp. However, at low protein concentrations (5-20 lM), the monomer-dimer equilibrium does not seem to be influenced by 1.0 mM L-Trp as monomer is present (Fig. 7B), which is not observed in the presence of L-Phe (Fig. 5A,B). The addition of 1 mM L-Trp is compared with 3 mM L-Phe as these concentrations were found in the DSF experiments to change the unfolding from polyphasic to apparent two-state ( Fig. 2A,C). The fact that L-Trp did not induce the same shift in equilibrium suggests that the stabilization observed in the DSF experiments occurs through a different mechanism than for L-Phe. Furthermore, L-Phe binding is found to be specific as D-Phe did not induce dimerization. That L-Trp does not induce dimerization in rchTPH2 extends recent results obtained by Patel et al., who showed that among the regulatory domains of TH, PAH, and TPH1, only PAH dimerizes in the presence of its natural substrate [39].

Thermal inactivation
To investigate if the dimerizing effect had an impact on TPH2 stability, rates of inactivation at 30°C in the presence or absence of 3 mM L-Phe were examined. The data were fitted with one-exponential decay curves, and the rate constants, k, were calculated using an exponential function, E t = E 0 e Àkt , where E 0 is the initial enzyme activity, and E t is the activity after time t at 30°C (Fig. 8). From the decay rate constants, half-lives (t ½ ) of the hTPH2 variants were calculated using t ½ = ln(2)/ k. At 30°C and a concentration of 5 lM, rchTPH2, and ND47-rchTPH2 displayed t ½ values of only 15 AE 2 and 18 AE 3 min, respectively. Hence, truncation of the N terminus slightly increased the t ½ value (P = 0.18, Ttest). With a t ½ value of 203 AE 40 min, chTPH2 displayed a 14-fold higher half-life compared to rchTPH2, which is in agreement with the results of Carkaci-Salli et al. [1] and confirms that the presence of the regulatory domain causes significant destabilization. With addition of 3 mM L-Phe, t ½ values of 41 AE 3, 49 AE 5, and 193 AE 29 min were obtained for rchTPH2, ND47-rchTPH2, and chTPH2, respectively. Variants containing the regulatory domain displayed nearly 3-fold increase in half-lives whereas chTPH2 was not significantly influenced by L-Phe.  That the stabilizing effect is exclusively observed for rchTPH2 and ND47-rchTPH2 implies that L-Phe is stabilizing through binding to the regulatory domain. Further, as L-Phe induces a stabilizing effect in both rchTPH2 and ND47-rchTPH2, binding of L-Phe does not seem to involve the 47 N-terminal residues. L-Phe induced stability and shift in equilibrium toward dimer imply that TPH2 variants containing the regulatory domain are more stable as dimers.
A stabilizing effect of L-Trp has previously been observed by McKinney et al. [16] who hypothesized that L-Trp might stabilize through binding in the active site, which might tighten up flexible regions, and hereby protect the active site. The increase in T m (DSF) upon the addition of L-Phe or L-Trp might, analogously to the hypothesis of McKinney et al. [16], occur through binding in the active site which in turn induces a more closed conformation, as it has been seen for binding of tryptophan in chicken TPH [25]. However, a shift in the monomer-dimer equilibrium is only observed for L-Phe, suggesting that the increase in T m and half-life occur through different mechanisms. Alternatively, L-Phe binds in an allosteric site which stabilizes the regulatory domain through improved interactions with an adjacent monomer. Such an allosteric site in the regulatory domain has been identified in PAH, which has been found to increase PAH activity and stability and induce large conformational changes [20,[39][40][41]. This is consistent with the binding of amino acids to ACT domains which often occurs at domain interfaces and results in conformational changes [42]. This hypothesis is supported by the observed shift in monomer-dimer equilibrium and the additive effect of L-Trp and L-Phe observed in the DSF experiments. Such an allosteric site might have relevance in vivo functioning as an allosteric modulating site that stabilizes a dimer in the native tetramer (dimer of dimers). Stabilization of the tetramer is important as disruption of the native tetramer results in decreased enzymatic activity [43].

Differential scanning calorimetry
The transition temperatures, T m , of the TPH2 variants were measured utilizing differential scanning calorimetry (DSC), to investigate if the low half-life of rchTPH2 is caused by premature unfolding of the regulatory domain. The DSC experiments were performed in the presence of L-Phe to observe domain unfolding in a monodisperse solution. All variants displayed irreversible unfolding upon reheating and hence, only T m values were extracted from the thermograms. Illustrative examples of DSC thermograms of the three variants are presented in Fig. 9.
The unfolding transitions of the three variants seem equivalent, however, by comparison of the thermograms of chTPH2 and ND47-rchTPH2 (Fig. 9A,B), it is evident that truncation of the regulatory domain resulted in a reduction in the heat capacity of the low temperature transition (T m (low)). This suggests that the lowest temperature transitions are partly due to the unfolding of the regulatory domain, as suggested for hPAH by Th or olfsson et al. [44]. The remaining transitions, therefore, originate from the unfolding of the more stable catalytic domain, quantified by the transition temperature of the main peak in the thermograms (T m (high)). Transition temperatures obtained from the thermograms are presented in Table 1. The regulatory and catalytic domains of rchTPH2 unfold at 40.1 AE 1.2°C and 46.8 AE 0.3°C, respectively. These results are, despite variations in the buffer systems, in the same range as previously reported transition temperatures for AAAHs (47.5-55.5°C) [33,45,46]. The DSC data relate to the inactivation measurements, as unfolding of the variants containing the regulatory domain is initiated at around 30°C (Fig. 9C,B), which is the temperature of the inactivation measurements, and around 35°C for chTPH2 (Fig. 9A). This might explain the significantly higher t ½ values observed for chTPH2 at 30°C. Unfolding of rchTPH2 shows a less profound low temperature transition compared to ND47-rchTPH2. This might be explained by the mobile N terminus causing a less defined regulatory domain which seems to unfold more continuously initiated at lower temperature than observed for chTPH2. Furthermore, the presence of the regulatory domain causes an earlier unfolding of the catalytic domain; this signifies that attempts to stabilize TPH should occur through the regulatory domain.

Purification yield
Characterization of hTPH2 variants containing the regulatory domain has been hampered by the low quantities obtained from E. coli, partially due to the poor stability. Figure 10 presents the purification yields obtained for the three variants with and without L-Phe added to the purification buffer. Without L-Phe, very low purification quantities were obtained for rchTPH2 (0.9 AE 0.2 mgÁL À1 ) compared to chTPH2 (12.0 AE 2.4 mgÁL À1 ). The low quantities for rchTPH2 were overcome by truncation of the N terminus which resulted in an 11-fold increase in yield (10.4 AE 3.1 mgÁL À1 ), which was similar to that of chTPH2.
The stabilizing effect of L-Phe was reflected in the purification yields, as they were increased seven-and twofold for rchTPH2 (6.4 AE 1.2 mgÁL À1 ) and ND47-rchTPH2 (25.1 AE 2.9 mgÁL À1 ), respectively (Fig. 10). In line with the results observed in the inactivation experiment, no significant change in purification yield was observed for chTPH2 (9.9 AE 3.1 mgÁL À1 ).

Conclusion
The current results demonstrate that L-Trp and L-Phe change the unfolding mechanism of hTPH2, but only when the regulatory domain is present. Deletion of the C-terminal tetramerization domain results in a monomer-dimer equilibrium which is shifted to predominately dimer with the addition of L-Phe. In the presence of L-Phe, the dimer displayed significantly increased half-life which in turn resulted in significantly increased purification yields of hTPH2 variants containing the regulatory domain. These findings will facilitate future characterization of hTPH2.

Materials
All used chemicals were of analytical grade, and all solutions were prepared using water from an 18.2 MΩÁcm Milli-Q synthesis A10 Q-Gard system which was filtered through a 0.22lm filter. Protein purification was performed on an € AKTA purifier 100 from GE Healthcare. Utilized GE Healthcare column variants: HiLoad Superdex 200 26/60 pg, Superdex 200 10/300 GL, and a XK 16/20 column packed with 25 mL Dextrin Sepharose High Performance media. During purifications, all TPH-containing solutions were kept in ice water, except during the chromatographic steps, which were performed at room temperature. Protein solutions were concentrated using an Amicon ultrafiltration cell with an Ultracell PL-3 membrane. Protein concentrations were determined by measuring the absorbance at 280 nm on an ND-1000 NanoDrop Spectrophotometer from Saveen Werner (Limhamn, Sweden).

Cloning and expression
Full-length human TPH2 cDNA optimized for expression in E. coli was obtained from GenScript (Piscataway, NJ, USA). All proteins were expressed as maltose binding protein fusion proteins from the pET26 expression vector in E. coli BL21 (DE3) (Novagen, Merck Millipore, Darmstadt, Germany) cells. The sequences of the proteins expressed are given in the Supporting information. The recombinant fusion proteins contain a cleavage recognition site for human rhinovirus 3C protease (3CP) [47]. The construct encoding the different protein variants was obtained by PCR. The primers used are listed in Table S1. All DNA sequences were verified by sequencing (Eurofins). Proteins were expressed at 20°C for 14 h, as previously described [17]. MBP-3CP was cloned and expressed in-house in a similar manner.

Purification
Escherichia coli cells from 650 mL cultures were thawed from À80°C and resuspended in buffer containing 20 mM HEPES/NH 4 OH, 300 mM (NH 4 ) 2 SO 4 , pH 7.0 (standard buffer, Std buffer) or 20 mM HEPES/NH 4 OH, 300 mM (NH 4 ) 2 SO 4 , pH 7.0, and 3 mM L-Phe (Phe buffer), to a volume of 40 mL. (NH 4 ) 2 SO 4 was included in the purification buffer, as it has been found that the purification yield of recombinant catalytic core of rabbit TPH1 was enhanced by the addition of ammonium sulfate [24]. The resuspended cell culture was lysed by sonication for 3 9 30 s using a Satorius Labsonic at 80% amplitude, while kept in ice water. The lysed sample was centrifuged at 4°C and 18 000 g for 20 min. The supernatant was decanted to another tube and centrifuged a second time at 4°C and 18 000 g for 20 min, while the pellet was discarded. The supernatant was collected and filtered through a 0.45-lm filter. A volume of approximately 35 mL filtered supernatant was loaded with a flow rate of 5 mLÁmin À1 onto a Dextrin Sepharose column, which was equilibrated with five column volumes of Std or Phe buffer. Following sample loading, an MBP-3CP solution was prepared by diluting MBP-3CP from stock with Std or Phe buffer to a concentration of 2.2 lM. About 30-mL MBP-3CP solution was loaded onto the column with a flow rate of 5 mLÁmin À1 , and the column was incubated at room temperature for 1 h. Once the flow (5 mLÁmin À1 ) was resumed, 10 mL of the protein-containing eluate was collected. The collected solution, containing target protein, was filtered through a 0.45-lm filter prior to loading on a HiLoad Superdex 200 prep grade column, which had been equilibrated with two column volumes of Std or Phe buffer. Std or Phe buffer was used as mobile phase using a flow rate of 2.5 mLÁmin À1 , and UV-detected (280 nm) peaks containing the target protein (verified by SDS/PAGE analysis -Figs S3-S5) were collected. The concentration of the collected protein was determined by UV-Vis absorption at 280 nm utilizing theoretical extinction coefficients obtained from Expasy [48]. The samples were either concentrated by ultrafiltration prior to freezing in liquid nitrogen, or if the concentration of target protein was high enough in the eluate, the ultrafiltration was bypassed, and the protein solution was frozen in liquid nitrogen and stored at À80°C.

SDS/PAGE
Evaluation of molecular weights and purity were conducted by SDS/PAGE. Proteins were resolved on Mini-PRO-TEAN TGX gels (7.5%) from Bio-Rad (Hercules, CA, USA) run at 100 V for 75 min with a protein standard from Bio-Rad (no. 161-0304). Gels were stained with Coomassie Blue to visualize the proteins.

Differential scanning fluorimetry
The unfolding of the TPH2 variants was recorded with an Agilent Technologies Stratagene MX3005 P RT-PCR machine (Santa Clara, CA, USA). The ligand screen was performed with a total volume of 25 lL in 96-well plates (polypropylene plates from Agilent Technologies). Each well was composed of protein at a concentration of 1 lM, SYPRO orange at a concentration of 29 (diluted from SYPRO Ò 50009 stock from Sigma, St. Louis, MO, USA), and ligand (diluted in purification buffer) in a concentration range of 0.1 lM to 10 mM. Each plate contained control wells with the purification buffer with and without protein and ligand. Scans were carried out using a scan rate of 1°CÁmin À1 , going from 20°C to 95°C. The thermograms were baseline corrected with MXPRO QPCR Software (Agilent Technologies, Santa Clara, CA, USA) and analyzed for transition temperatures with GRAPHPAD PRISM 6 (GraphPad Software, Inc, La Jolla, CA, USA) utilizing a Boltzmann sigmoid fit: where LL and UL are the values of minimum and maximum intensities, respectively, and a denotes the slope of the unfolding curve at T m [49].

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Appendix S1. Sequences of hTPH2 protein variants. Table S1. Primers for cloning the recombinant truncated hTPH2 protein variants. Fig. S1. Alignment of the regulatory domains of hTPH2 and rnPAH. Fig. S2. Representative example of raw data from differential scanning fluorimetry. Fig. S3. SDS-PAGE results of collected peaks from purification of rchTPH2 using Phe buffer. Fig. S4. SDS-PAGE results of collected peaks from purification of ND47-rchTPH2 using Phe buffer.