Preclinical evaluation of multiple species of PEGylated recombinant phenylalanine ammonia lyase for the treatment of phenylketonuria

Abstract

Phenylketonuria (PKU) is a metabolic disorder, in which loss of phenylalanine hydroxylase activity results in neurotoxic levels of phenylalanine. We used the Pah^enu2/enu2 PKU mouse model in short- and long-term studies of enzyme substitution therapy with PEGylated phenylalanine ammonia lyase (PEG-PAL conjugates) from 4 different species. The most therapeutically effective PAL (Av, Anabaena variabilis) species was one without the highest specific activity, but with the highest stability; indicating the importance of protein stability in the development of effective protein therapeutics. A PEG-Av-p.C503S/p.C565S-PAL effectively lowered phenylalanine levels in both vascular space and brain tissue over a >90 day trial period, resulting in reduced manifestations associated with PKU, including reversal of PKU-associated hypopigmentation and enhanced animal health. Phenylalanine reduction occurred in a dose- and loading-dependent manner, and PEGylation reduced the neutralizing immune response to the enzyme. Human clinical trials with PEG-Av-p.C503S/p.C565S-PAL as a treatment for PKU are underway.

Phenylketonuria (PKU) and related hyperphenylalaninemias (HPA) are a classic set of autosomal recessive multifactorial metabolic disorders (Online Mendelian Inheritance in Man accession no. 261600) (1, 2), that were the first genetic diseases to respond to treatment. Patients with HPA/PKU have compromised activity of phenylalanine hydroxylase (PAH) (EC 1.14.16.1), the enzyme that catalyzes the irreversible conversion of phenylalanine (Phe) to tyrosine (Tyr). In the absence of treatment, systemic Phe concentration can increase to neurotoxic levels and impair cognitive development. The treatment of HPA/PKU requires life-long selective reduction of Phe intake and an adequate dietary supply of Tyr.

Dietary therapy for HPA/PKU is a successful but difficult treatment. The constant adherence to a restricted diet often leads to reduced compliance in adolescence and beyond (3–6), with potentially negative neurological consequences. Development of an alternate therapy that would permit liberalization of dietary restrictions and simplify disease management would greatly improve treatment of HPA/PKU and the quality of life for the patients; 6-R-L-erythro-5,6,7,8-tetrahydrobiopterin (BH₄) catalytic cofactor therapy, gene therapy, large neutral amino acid (LNAA) supplementation in the diet, and orthotopic liver transplantation are among the different forms of therapy that have been explored to improve the treatment for PKU (1). We report on enzyme substitution with phenylalanine ammonia lyase (PAL, EC 4.3.1.5) (7), a nonmammalian protein providing alternative Phe metabolism, which has the potential to replace dietary treatment. This approach is expected to reduce elevated Phe levels by acting as proxy to the deficient PAH enzyme of the HPA/PKU patient. PAL is a robust protein (8, 9), which is anticipated to reverse HPA by converting excess systemic Phe to trans-cinnamic acid and metabolically insignificant levels of ammonia.

Investigation of PAL treatment for PKU was initiated over two decades ago (10, 11). When taken orally, in a nonabsorbable and protected form, PAL lowers plasma Phe concentration in rodent models of PKU (7, 12, 13) in short-term studies. Subcutaneous administration of PAL reduces plasma Phe in the orthologous mutant PKU mouse model (7, 14), also in short-term studies. Long-term reduction of Phe levels by PAL is hampered by clearance of the enzyme through a neutralizing immune response and proteolysis. Initial attempts to engineer a more efficacious form of PAL by site-directed mutagenesis and chemical modification with polyethylene glycol (PEG) while maintaining specific activity have been successful in reducing its immunogenicity and prolonging plasma half-life in the PKU mouse model (14–17).

Here, we report the short- and long-term pharmacodynamic (PD) profiles of engineered, PEG-PAL variants in a PKU mouse model. We tested the effects of different PEGylation formulations and protocols on long-term in vivo PAL efficacy, and examined PEGylated conjugates of WT PALs (PEG-PAL) from 4 different cyanobacterial, parsley, and fungal species [Anabaena variabilis (Av), Nostoc punctiforme (Np), Petroselinum crispum (Pc), and Rhodosporidium toruloides (Rt)], and 19 mutants of Rt- and Av-PALs to identify the most promising candidates for further development as therapeutic agents for HPA/PKU treatment. The most therapeutically efficacious molecule in the PKU mouse was the most thermally stable and protease resistant PAL (PEG-Av-p.C503S/p.C565S-PAL); it produces complete HPA reversal with near total suppression of immune response, and reduces HPA in both vascular space and the brain in the PKU mouse model. Long-term therapy with this PEG-PAL variant reverses PKU-induced hypopigmentation and supports robust health status. Response to PEG-PAL dosing regimens is gender-dependent in the mouse model, suggesting that dosing may need to be different in male patients versus females.

Results

The following results are data extracted from the protocols listed in supporting information (SI) Tables S1–S4.

Short-Term in Vivo Studies.

Previously, we identified a variant of Rt-PAL, Rt-p.R91K-PAL, with a specific activity higher than WT Rt-PAL (17). We used the Rt-p.R91K-PAL-mutant to measure the effect of PEGylation, exploring different PEGylation formulations and protocols, route of administration, and loading dose on Phe levels in the PKU mouse model Pah^enu2/enu2 (ENU2) (18).

The in vivo effects of PAL PEGylation and varied route of administration.

Subcutaneous administration of unPEGylated forms of PAL supports clinically significant clearance of plasma Phe in the mouse model (Fig. 1A), but the effect diminishes after repeat injections >8 days. Plasma Phe levels are then indistinguishable from vehicle controls (P = 0.4375; F = 0.67; df = 1,8).

Fig. 1.

Plasma Phe profile of ENU2 mice during short-term 12-day study with 3 s.c. bolus injections (on days 1, 4, and 8) of 1 I.U. Rt-WT-PAL (■), 1 I.U. Rt-p.R91K-PAL (●), or vehicle (♦) (n = 4) (A); or 1 I.U. PEGylated WT Av (●), Np (×), Pc (■), Rt (▴)-PALs, or vehicle (♦) (n = 5) formulations (B). The time-dependent measures are represented in μM (mean ± 1 SEM). The broken line accounts for the daily plasma Phe fluctuations that are not captured in this figure.

Assuming that loss of PAL efficacy over time reflects inactivation by means of neutralizing antibodies, we tested various PEGylation products and protocols and showed that serum anti-PAL antibody levels in the PKU mouse model are attenuated by PEGylation (Table 1). Of the series tested, the Nippon Oil and Fat (NOF) PEG used as a PEGylated 1:3 NOF-Rt-p.R91K-PAL conjugate was the most effective in suppressing immunogenicity. Reduction in plasma Phe varied inversely with anti-PAL antibody levels, as measured throughout 12 days for each PAL formulation (data not shown). Also, the route of administration (Table S5) did not alter efficacy of the PEG-PAL agent (P = 0.9305; F = 0.15; df = 3,24) measured on day 2. However, single-site administration compromised efficacy as compared with rotating-site, during these short-term studies (P = 0.0015; F = 12.74; df = 1,24) as measured on day 5.

Table 1.

Serum IgG antibody levels in ENU2 mice, 21 days after the first of 3 (days 1, 4, and 8) s.c. bolus injections of 1 I.U. of each of the indicated formulations

Formulation	Antibody (IgG) titer, dilution factor; median/range
Vehicle	50/50–50
Rt-p.R91K-PAL	12,500/6,250–506,250
PEGylated 1:8 Nektar-Rt-p.R91K-PAL	36,450/12150–109,350
PEGylated 1:1 NOF-Rt-p.R91K-PAL	8,100/450–12,150
PEGylated 1:2 NOF-Rt-p.R91K-PAL	900/150–4,050
PEGylated 1:3 NOF-Rt-p.R91K-PAL	100/50–450
PEGylated 1:4 NOF-Rt-p.R91K-PAL	40,198/36,450–109,350
PEGylated 1:8 NOF-Rt-p.R91K-PAL	900/50–1,350

n = 6.

Loading dose effect.

The effect of PEG-Rt-p.R91K-PAL on plasma Phe is altered by dosage frequency; higher dose (1 I.U.) and lower frequency (twice weekly) is more effective than lower, daily dosing (P < 0.0001; F = 32.61; df = 1,14) as measured on day 13 (Table S6).

Screening of Rt-PAL variants.

We engineered 18 mutants of WT Rt-PAL to improve specific activity, reduce degradation by proteolysis, and reduce immunogencity (data not shown). These modifications did not enhance plasma Phe reduction as measured after the administration of equivalent units of these mutants, day 9 post third treatment (P = 0.0228; F = 2.02; df = 18,57) (Table S7).

Comparison of 4 PAL species.

We also examined WT Av-, Np-, Pc-, and Rt-PAL species, and an Av-p.C503S/p.C565S-PAL double mutant variant that was engineered to reduce aggregation (data not shown). In short-term in vivo studies, we observed reduction of plasma Phe levels by PEG-Av-WT-PAL (100% reduced, P < 0.0001, t = −5.07, df = 23), PEG-Np-WT-PAL (100% reduced, P < 0.0001, t = −4.78, df = 23), PEG-Pc-WT-PAL (89% reduced, P = 0.0001, t = −4.57, df = 23), PEG-Rt-WT-PAL (89% reduced, P < 0.0001, t = −4.86, df = 23) 24 h post administration on day 1 (Fig. 1B). PEG-Av-PAL and PEG-Np-PAL had the highest and most sustained reduction 3 and 4 days post the first, day 1, and second, day 4, injection, respectively) (P < 0.0001; F = 28.81; df = 2,23). After the day 8 injection, the day 9 Phe level reduction by PEG-Rt-PAL was attenuated, compared with the 1-day postinjection level observed after the first injection (Fig. 1B). On day 9, Phe levels of PEG-Av-PAL and PEG-Pc-PAL were below the vehicle, but showed a diminished response to day 8 injection (Fig. 1B).

Dose response.

We measured dose-response profiles for PEG-PAL conjugates of WT Av, Np and Pc (Table S8). We found a uniform dose-response profile with sustained reduction of plasma Phe levels starting from day 2 for the PEG-Av-WT and PEG-Pc-WT-PALs, and from day 4 for PEG-Np-WT-PAL.

Long-Term in Vivo Studies.

We conducted long-term (90 day) tolerance studies with PEG-Np-PAL and unPEGylated and PEG-Rt-p.R91K-PALs in the ENU2 mouse model. The PEGylated conjugates effectively diminished and maintained lower plasma Phe concentrations (Table S9); however, the time to achieve clinically beneficial levels of Phe depended on dosing. Decreasing daily doses of PEG-Rt-p.R91K-PAL and static weekly doses of PEG-Np-WT-PAL lowered and maintained reduced plasma Phe at clinically beneficial levels from the beginning of the study. Escalating daily doses of PEG-Rt-p.R91K-PAL only achieved these levels onward of day 31, when the dosage reached or exceeded 0.6 I.U. UnPEGylated Rt-p.R91K-PAL molecules induced increased antibody production compared with the PEGylated conjugates (Table S10). Short-term 15 day follow-up studies with PEG-Rt-p.R91K-PAL treatment, given to mice preexposed to unPEGylated Rt-p.R91K-PAL for 90 days, resulted in a similar plasma Phe profile as expected with naive animals (data not shown).

Long-term PD response was measured in mice dosed once weekly with PEG-Av-p.C503S/p.C565S-PAL or PEG-Av-WT-PAL conjugates throughout 57 days (Fig. S1). Reduction of plasma Phe was most effective with a 4 I.U. dose. Long-term PD response of mice treated once weekly for 16 weeks, with decreasing doses of PEG-Av-p.C503S/p.C565S-PAL (Fig. 2), showed efficacy with the primary higher doses, 4-days postinitial dosing. Efficacy diminished after the second dosing, and plasma Phe levels became indistinguishable from the vehicle-treated group directly before the third treatment (P = 0.3375; F = 0.94; df = 1,59). Lowered and gradually sustained diminished plasma Phe at clinically beneficial levels were achieved onward of day-29 dosings. The secondary decreased doses also lowered plasma Phe levels; however, the only group that was able to maintain the clinically beneficial reduction was the one receiving 2 I.U. The corresponding serum antibody data (Table 2) indicate that immune responses were decreased with use of the engineered PAL molecule.

Fig. 2.

Plasma Phe levels in ENU2 mice over 120 days with decreasing dose administrations of PEG-Av-p.C503S/p.C565S-PAL. Primary weeks 1–10; reduced secondary weeks 11–16 doses: 4, 2 (●); 4, 1 (▴) or 2, 1 (■) I.U.; respectively; or vehicle (♦) (weeks 1–16). The time-dependent measures are represented in μM (mean ± 1 SEM); n = 16. The broken line accounts for the daily fluctuations between measures that are not captured in this figure.

Table 2.

Serum antibody (IgG) levels of mice treated with decreasing dose of once weekly PEG-Av-p.C503S/p.C565S-PAL

Day	Antibody (IgG) titer, dilution factor; median/range
	Vehicle	4 I.U., weeks 1–10; 2 I.U., weeks 11–16	4 I.U., weeks 1–10; 1 I.U., weeks 11–16	2 I.U., weeks 1–10; 1 I.U., weeks 11–16
0	50/50–50	50/50–50	50/50–50	50/50–50
12	50/50–50	50/50–150	50/50–150	50/50–50
19	50/50–50	50/50–150	50/50–150	50/50–50
26	50/50–50	50/50–150	50/50–450	50/50–150
33	50/50–50	50/50–450	150/50–450	50/50–150
40	50/50–50	50/50–1,350	50/50–1,350	50/50–150
61	50/50–50	450/50–1,350	450/50–1,350	300/50–1,350
75	50/50–50	150/50–450	450/50–1,350	450/50–450
89	50/50–50	150/150–1,350	450/50–1,350	150/50–150
103	50/50–50	450/150–1,350	450/150–12,150	450/50–450
113	50/50–50	450/50–4,050	450/150–4,050	450/150–4,050
120	50/50–50	150/50–450	150/50–1,350	150/50–450

n = 16.

We observed a gender effect. Although male (Fig. S2) and female (Fig. S3) mice both experienced clinically beneficial reductions in plasma Phe levels, male mice had an overall greater and more sustained response over time and dose. This gender-bias in response was even more apparent at the secondary lower doses. The corresponding serum antibody levels (Table S11) show a similar immune response between males and females.

Additional Endophenotypes.

Reversal of hypopigmentation.

PKU mice treated with PEG-Rt-p.R91K-PAL, PEG-Np-WT-PAL, PEG-Av-WT-PAL, and PEG-Av-p.C503S/p.C565S-PAL displayed darkened dorsal coat color (Fig. 3). This change appeared at 7–10 days around the eyes and in the forehead area, then on the whole body by days 15–20. The effect was reversible within an equivalent period after termination of treatment.

Fig. 3.

PEG-PAL reverses hypopigmentation. (A) PKU mice were photographed on day 124, after vehicle (once weekly for 16 weeks and then on day 123) (Upper) and PEG-Av-p.C503S/p.C565S-PAL [4 I.U. once weekly (1–10) and day 123; 2 I.U. once weekly (11–16)] (Lower) treatments. (B) PKU mouse treated with 0.5 I.U. PEG-Np-WT-PAL on days 1, 4, 8, 11, and 14, photographed on study days 8, 11, and 15.

Weight gain.

A comparison between animals receiving long-term vehicle vs. PEG-Rt-p.R91K, PEG-Np-WT, PEG-Av-WT, or PEG-Av-p.C503S/p.C565S-PALs demonstrates an enhanced and more immediate weight gain with PAL treatment (Table S12).

Health status and mortality.

General health condition, grooming, and behavior did not change during any of the PEG-PAL administration studies. The injection site showed no signs of redness or edema; 5% of vehicle and PEG-PAL treated animals participating in the long term studies died of causes unrelated to drug administration before the study was completed; all remaining animals were in excellent health on study completion.

Brain Phe concentrations.

In acute response studies, i.v. injection of unPEGylated Rt-WT-PAL (0.74 and 3.7 I.U) demonstrated a statistically significant dose-response relationship (P < 0.0001; F = 114.76; df = 1,9) in post treatment (24 h) brain Phe concentrations (Fig. 4A).

Fig. 4.

Plasma (dark column) and corresponding brain (light column) Phe concentration of ENU2 mice 24 h post day 1 i.v. injections of 0 I.U. (vehicle), and 0.74 and 3.7 I.U unPEGylated Rt-WT-PAL (n = 4) (A); or 24 h post day 123 s.c. injections of 0 I.U. (vehicle), and 4, 4 and 2 I.U PEG-Av-p.C503S/p.C565S-PAL (n = 16) (the y axis is logarithmic to better display the variation in the reduced plasma and brain Phe concentrations) (B). Data are represented in μM (mean ± 1 SEM).

Mice treated (s.c.) post long-term (120 day) study with PEG-Av-p.C503S/p.C565S-PAL experienced reduced brain Phe levels; values correlate directly with reduced plasma Phe concentrations (Fig. 4B). The brain Phe levels of the treated animals were reduced to levels indistinguishable from their untreated WT BTBR/Pas counterparts (data not shown). There was no gender effect on brain Phe levels.

Discussion

Treatment of HPA/PKU through dietary restrictions to reduce Phe intake (19–21) can be difficult to follow. Development of an alternate treatment such as enzyme therapy has been explored as a means to reduce reliance on dietary restriction. Enzyme therapy with PAL (a nonmammalian protein) was selected as a substitute (7) for the native PAH, because PAL, unlike PAH, is inherently stable and does not require a cofactor for activity. PAL converts the excess systemic Phe to trans-cinnamic acid with trace amounts of ammonia (11, 22). Trans-cinnamate has no embryotoxic effects in laboratory animals (10), and is converted in the liver to benzoic acid, which is excreted in the urine as hippurate (23). Small amounts of cinnamate and benzoic acid are also excreted (11).

Here, we show that PEG-PAL corrects the metabolic phenotype in the PKU mouse model. Short-term (12 day) studies with s.c. administration of unmodified PAL show that the protein supports clearance of plasma Phe to levels compatible with treatment, achieving euphenylalaninemia (the designated endpoint to evaluate the response to PAL) (24). However, the efficacy of PAL diminishes after repeated injections (Fig. 1A). Appearance of IgG immune response is loosely associated with loss of PAL efficacy. Unmodified PAL augments immune response (Table S10), and permanently eliminates efficacy post initial 7-day plasma Phe reduction. We have already shown that PEGylation prolongs in vivo activity of the Rt-WT-PAL through suppression of immunogenicity (14). Optimization of PEGylation demonstrates maximum suppression of immunogenicity in ENU2 mice with NOF PEG conjugated at a 1:3 ratio (moles PAL lysine: moles PEG; see Tables 1 and 2, and Table S10). Another example is present in animals formerly treated over the long-term with unmodified PAL molecules, followed by short-term exposure to PEG-PAL treatment. The latter regimen showed the plasma Phe profile of naive animals exposed to PEG-PAL for the first time. IgM antibody levels are very low, transient and not dose-dependent (data not shown).

We attempted to improve the therapeutic properties of Rt-PAL in the PKU mouse model by further engineering its structure (15). Conserving surface-exposed lysine residues to ensure maximum PEGylation at these sites (17), we introduced lysine residues in regions that were previously identified as immunogenic (15), and removed others to redirect PEGylation (p.Q41K, p.H345K, p.H359K, p.Q558K, p.T565K, p.G566K, p.S614K, p.S616K, and p.K132R) (data not shown). One previously mutated PEGylation site (p.R91K) increased activity both in vitro and in vivo in the PKU mouse (15, 17). The p.R91 residue, partially exposed in the PAL structure, is located in the helix that connects with loop 102–124 (25); it is not directly involved in the PAL active site, but mutant p.R91K may stabilize the interaction of helix 86–101 with loop 102–124; thus, increasing enzymatic activity of p.R91K PAL. However, combining this mutation as a double or triple mutant with the above mutations did not improve in vivo efficacy relative to Rt-p.R91K-PAL, as measured by plasma Phe reduction in the PKU mouse model (Table S7).

We also tested wild-type PAL proteins from 4 different species in the PKU mice, looking for improved efficacy over Rt-PAL (Fig. 1B). Although Rt-PAL has the highest specific activity, other properties, including pH optimum, increased protease resistance, thermal stability, and the lower K_m found in the Av-WT-PAL and its last-stage engineered Av-p.C503S/p.C565S-PAL double mutant (altered in 2 places to reduce aggregation; data not shown), make the Av-PAL variants better therapeutic molecules than Rt-PAL (Fig. 1B; Table S8 and Fig. S1). The absence of a C-terminal domain insertion in the Np- and Av-PALs reduces the number of lysines per monomer to 18, compared with Rt- and Pc-WT-PALs, which have 29 and 44 lysines per monomer, respectively. The lack of the insertion domain in Av-PAL also results in a molecule with a more globular shape than Rt-PAL (25). Perhaps this shape and the reduced numbers of lysines per monomer in the Np- and Av-PALs results in a more uniform and protective distribution of PEG molecules on the protein surface, leading to higher and more sustained in vivo Phe reductions than for the Rt- and Pc-PALs. This result emphasizes the importance of considering and optimizing characteristics other than specific activity when designing effective protein therapeutics. It also indicates the importance of in vivo screening to find the optimal molecule for therapeutic purposes.

In our long-term studies in PKU mice, we demonstrate efficacy of PAL under conditions that would mimic patient treatment. Our PD studies show that the PEG-Av-p.C503S/p.C565S-PAL double mutant best demonstrates reduction of plasma Phe concentrations, with decreasing weekly dose (initial 4 I.U. for 10 weeks, followed by 2 I.U. for the remaining period) as the most effective protocol for retaining a prolonged and sustained clinically significant effect in PKU mice (Fig. 2). The prolonged efficacy of higher doses, combined with less frequent administration, is of particular interest, because fewer injections of PAL would facilitate treatment. The transient loss of efficacy, 8 days post first treatment, likely attributed to a primary immune response, is overcome after repeated PAL administrations. Diminished plasma Phe levels were achieved and sustained once the animals had pushed through the initial immune response (≈30 days post primary dose administration).

Hypopigmentation associated with HPA/PKU stems from impaired Tyr metabolism, which affects the production of melanin. Thus, elevated levels of Phe inhibit tyrosinase, reducing melanin production (26), as well as Tyr uptake by melanocytes (27). We observed reversal of hypopigmentation in the PKU mouse after long-term treatment with PEG-Np-WT-PAL and PEG-Av-p.C503S/p.C565S-PAL (Fig. 3), similar to what has been seen after low-Phe dietary treatment or gene therapy (1).

PEGylated and unmodified PAL therapies upheld the health status of ENU2 mice and did not cause excess mortality to the animals (data not shown). Also, PEG-PAL enhanced more immediate weight gain consistent with a therapeutic effect (Table S12), and repeated and long-term administration of unmodified PAL had no secondary or toxic effects.

Reduction of brain Phe levels was also observed on short- and long-term treatment of PKU mice with PEG-Rt-WT-PAL and PEG-Av-p.C503S/p.C565S-PAL. The untreated PKU mouse has an ≈10-fold increase in brain Phe level, which correlates with their elevated plasma Phe levels (28). This endophenotype was altered by acute and long-term PEG-PAL treatment; brain Phe levels were corrected in a dose-dependent manner, regardless of the route of PAL administration (Fig. 4 A and B).

We observed a gender effect (Fig. S2 and Fig. S3) after long-term exposure to PEG-PAL similar to that observed in mice undergoing genome-targeted PAH gene therapy (29). In the latter, the Phe levels in the gene-treated females were reduced to normal after gonadectomy, and although estradiol-treated sterile males developed HPA, the dihydrotestosterone-treated sterile females remained euphenylalaninemic. Estrogen may be affecting PAL activity and/or turnover/clearance in a similar manner in our studies. These findings in the PKU mouse suggest that the long-term PAL treatment of female patients may potentially require a different dosing regimen than males.

In conclusion, our findings suggest that an injectable species of PEG-PAL has the potential to diminish the neurotoxic HPA of the PKU phenotype over the long-term, and is anticipated to overcome the handicaps of dietary therapy. This approach is expected to allow PKU patients (many of whom will be committed to life-long therapy) access to a better quality of life.

On the bases of the work reported here, a Phase I clinical study of PEG-PAL for the treatment of PKU has been initiated (05/20/2008; http://www.biomarinpharm.com).

Methods

PAL Species and Mutant Variants.

We tested 4 different species of PAL protein isolated from Av, Np, Pc, and Rt, 1 variant of Av WT; and 18 variants of Rt WT derived by site-directed mutagenesis: Av-p.C503S/p.C565S, Rt-p.R91K, Rt-p.R91K/p.H345K, Rt-p.R91K/p.R359K, Rt-p.R91K/p.E403Q, Rt-p.R91K/p.Q558K, Rt-p.R91K/p.T565K, Rt-p.R91K/p.G566K, Rt-p.R91K/p.S614K, Rt-p.R91K/p.S616K, Rt-p.R91K/p.Q41K/p.H345K, Rt-p.R91K/p.Q41K/p.Q558K, Rt-p.R91K/p.Q41K/p.S616K, Rt-p.R91K/p.H345K/p.Q558K, Rt-p.R91K/p.H345K/p.T565K, Rt-p.R91K/p.H345K/p.G566K, Rt-p.R91K/p.H345K/p.S614K, Rt-p.R91K/p.E403Q/p.K132R, and Rt-p.R91K/p.E403Q/p.H598Q.

Synthesis of Recombinant PAL.

Synthesis of the Rt coding sequence for PAL has been described (7). Cloning of Np, Av, and Pc-PALs were as follows: Np genomic DNA was purchased from ATCC (29133D), and the histidine ammonia-lyase (HAL) gene (ZP_00105927), which is in fact a PAL (30), was PCR amplified from this genomic DNA. Av cells were purchased from ATCC (29413) and the PAL gene (YP_324488) was amplified by PCR. Pc PAL 1 gene (P24481) was synthesized by PCR assembly of oligonucleotides as described by Baedeker and Schulz (31). All 3 resulting PCR products were ligated into pET-28a(+) and pET-30a(+) (Novagen) for N-His tagged and untagged constructs, respectively.

Bacterial strains and culture conditions.

An Rt-PAL expressing strain was constructed as previously described (14). For Np-PAL as well as for Pc-PAL, Escherichia coli BL21(DE3) cells (Stratagene) were transformed with pGro7, which contains the genes for groES and groEL (TaKaRa). BL21(DE3)/pGro7 cells were transformed with either pET-28-Np-PAL or pET-28-Pc-PAL and cultured in 25 mL of LB with 50 mg/L kanamycin and 20 mg/L chloramphenicol overnight at 37 °C; 20 mL of the overnight culture were diluted into 1 L of LB/kanamycin/chloramphenicol, with 500 mg/L L-arabinose to induce chaperones, and grown at 37 °C to an OD₆₀₀ of 0.6. The culture was chilled on ice for 5 min, and PAL expression induced by addition of 0.3 mM IPTG. Cells were grown at 20 °C for 16 h and harvested by centrifugation.

Av-PAL was produced by using 2 different strains. BL21(DE3)pLysS cells (Stratagene) were transformed with pET-28-Av-PAL and cultured as above, but without arabinose induction and chloramphenicol selection. Alternatively, Av-PAL cloned into the pIBX7 vector was introduced by transformation in BLR (DE3)pLysS (Novagen) and cultured as above.

Mutagenesis.

Point mutations of the PAL sequence for correction of the Pc synthetic gene, and mutagenesis of all PAL sequences, were introduced by PCR standard methods or site-directed mutagenesis by using the QuikChange site-directed mutagenesis kit (Stratagene) (17).

PAL Purification, PEGylation, and Preparation for Dosing.

Protein was purified as described previously (see supplemental information in ref. 14). PAL activity was measured as reported earlier (7).

PAL–PEG conjugates were produced by using 2 different PEG molecules and protocols: (i) mPEG-SPA (Nektar Therapeutics) conjugates were produced by coupling linear 20-kDa methoxy-PEG-SPA to PAL (at ratio of 1:8 moles PAL lysine: moles PEG) by using an established reaction protocol (14); (ii) 20-kDa linear PEG, NOF catalog number ME-200HS (Nippon Oil and Fat) at various ratios (1:1, 1:2, 1:3, 1:4 and 1:8 moles PAL lysines: moles PEG), under optimized reaction conditions derived from the earlier protocol (14).

All conjugates were cleared of endotoxin by passage through a Mustang E Acrodisc (Pall Filtron), and sterilized by 0.22-μm filtration (Ultra free MC, Millipore) before s.c. injection.

PKU Mouse Model.

The Pah^enu2/enu2 (ENU2) homozygous mutant mouse (18, 32), classified as an orthologous counterpart for human PKU (see http://www.pahdb.mcgill.ca; see Information/Mouse links), was used for the in vivo studies. Animals at 3–6 months of age, starting at ≈25-g body weight, were housed individually, and were supplied with food (JE Mondou lab chow no. 5001) and water ad libitum during the studies.

In Vivo Evaluations of PAL Species.

Short-term protocol.

Short-term efficacy was assessed in ENU2 mice by measuring plasma Phe concentrations before and after scheduled injections of PAL (Table S1). The immune response of mice was measured over a 3-week period. Two controls were included in each study: 1 positive with an enzyme known to lower Phe levels, and 1 negative (buffer vehicle; either 10 mM Na-phosphate/150 mM NaCl, pH 7.4 or 10 mM Tris/150 mM NaCl, pH 7.5, depending on the buffer used for the test article). Animals received test enzyme at 10:00 h on a set schedule. Blood samples were collected from the tail vein before initial dosing and on subsequent scheduled days (at 09:00 h) (Table S1). Protocols were designed to test: (i) PEGylation products and PEGylated vs. unPEGylated PAL molecules, (ii) route of administration, (iii) loading dose, and (iv) dose-response effects.

Long-term protocol.

Three protocols were used to measure long-term PD effect and immunogenicity of the most promising PAL molecules in the ENU2 mice. (i) A chronic (90 day) tolerance study profiled unPEGylated and PEG-Rt-p.R91K and PEG-Np-WT-PALs at various doses and schedules; again, as with the short-term studies, animals were treated on set schedules for enzyme dosing (10:00 h) and bleeds (09:00 h) (Table S2). Two substudies further evaluated administration of the same species of PAL with varied PEGylation, dose and/or frequency, in animals that participated in the first part of the long-term protocol (Table S3). (ii) We measured responses to PEG-Av-WT-PAL and the double mutant PEG-Av-p.C503S/p.C565S-PAL over an 8-week period. (iii) Last, we measured responses to decreasing doses of PEG-Av-p.C503S/p.C565S-PAL over a 16-week period (Table S2). Animals were dosed at 15:30 h on a set schedule, and blood samples were collected from the tail vain before initial dosing and on subsequent scheduled days at 13:30 h for this protocol only. Controls, as described above, were included in each component.

Weight change.

Weights were recorded to monitor animal well-being; changes were calculated midexperiment and on the final day.

Health status and mortality.

General health condition, grooming, and behavior for all animals were monitored daily, and injection sites checked for signs of redness or edema. All mortalities were recorded and cause of death examined.

Brain Phe.

Both brain and plasma Phe levels were measured in response to (i) unPEGylated Rt-WT PAL (i.v.), and (ii) dose-response to PEG-Av-p.C503S/p.C565S-PAL (s.c.; Table S4).

Analytical.

Plasma Phe concentrations were measured by fluorometric microtiter plate assay (33), based on the method of McCaman and Robins (34). To measure brain L-Phe concentration, brain samples were prepared according to the method of Diomede et al. (35), and analyzed by standard HPLC amino acid measurement protocol for physiological samples (36).

Anti-PAL antibody concentration was measured by ELISA.

Statistical Methods.

Separate analyses were conducted for the data summarized in each of (i) days 0, 8, 9, and 12 (Fig. 1A), (ii) all days (Table S5), (iii) days 0 and 9 (Table S7), (iv) all days (Table S8), (v) days 0 and 15 (Fig. 2), (vi) 24 h (Fig. 4A), and (vii) all days (Table S6). With the exception of vi, data were analyzed by using a repeated measures ANCOVA with plasma Phe as the response variable, treatment group as a fixed effect, animal within treatment group as a random effect, and baseline (starting value at day 0) as a covariate. The analyses of ii, iv, and vii also included terms for assessment day and treatment group by assessment day interaction; these additional terms were included to account for differences in variability between assessments made immediately postdose and those made at other times. The response variable in the analysis of the data in iii was change from baseline to Day 9 in plasma Phe. The data in vi were analyzed by using a single factor ANOVA with brain Phe as the response variable and treatment group as a fixed effect.

Overall comparisons between treatment groups were taken from the relevant ANCOVA/ANOVA, whereas specific between group comparisons were made by using contrasts. Residuals were examined postanalysis and examined for departures from normality; none was detected. To guard against errors of false discovery, a significance level of 0.01 was used.