Personalized Genotype-Based Approach for Treatment of Phenylketonuria

Abstract

Extensive studies have examined the clinical manifestations, pathogenic mechanisms, and genetic variations of phenylketonuria (PKU) across different populations, resulting in a substantial collection of molecular genetic data on the phenylalanine hydroxylase (PAH) gene and its variants. However, many genotypes are associated with a range of clinical phenotypes, as well as variable responsiveness to sapropterin, presenting ongoing challenges for effective treatment. To address this, we enhanced the PAH activity landscapes method by incorporating high-throughput techniques, including automated pipetting, integrated data processing via Gaussian modeling of 3D surfaces, and bioinformatics analyses with robust quality control. Using PAH activity landscapes, we visualized PAH enzymatic function across 99 common PAH genotypes under varying metabolic and therapeutic conditions. This deep functional phenotyping approach enabled us to identify distinct genotype subpopulations by using consensus clustering, correlate them with clinical phenotypes, and propose subpopulation-specific treatment protocols. Our findings suggest that clinical phenotypes can be predicted and treatment regimens can be adjusted based on residual PAH function profiles. To further support personalized treatment strategies, we revised our publicly accessible PAH genotype & activity landscapes database to share the latest insights into PAH function and patient phenotypes-namely residual enzyme activity and responsiveness to sapropterin as conveyed by two alleles. This resource underscores the translational significance of functional research in PKU and offers a practical tool to support personalized treatment in clinical settings.

Keywords: misfolding; phenylalanine hydroxylase; phenylketonuria; sapropterin dihydrochloride; tetrahydrobiopterin.

FIGURE 1

Process of data evaluation for the PAH activity landscape of human PAH WT. Data representing PAH enzyme activity at varying [Phe] and [BH₄] were visualized using a color code (with the enzyme activity scale provided at the bottom) to depict activity as a percentage of the activity of the wild type. The sequence of graphs from left to right illustrates the data‐evaluation steps. The pink “X” on the 2D graphs indicates the peak activity position representing maximum activity. The solid pink line delineates 50% of enzyme activity, signifying its optimal working range. (A) Plotting the raw data (left) obtained from the PAH activity assay in a 96‐well format and triangular surface mesh (right) creates a non‐smoothed and non‐modeled 3D surface. (B) Data interpolation with a regular grid‐based algorithm, followed by smoothing via a Gaussian filter, allows for the creation of smoothed non‐modeled 3D (left) and 2D (right) panels. (C) The logarithmic transformation of the data creates a symmetrical surface (left) that can be fitted into a Gaussian model (right). (D) The exponential transformation of the data creates smoothed modeled 3D (left) and 2D (right) panels.

FIGURE 2

Subpopulations of PAH activity landscapes. Characteristics that define subpopulations and their clinical relevance. Box plots display the median, range from minimum to maximum, and the 25th to 75th percentiles. (A) Visual distribution of studied PAH genotypes in PAH activity landscapes based on the parameters of residual peak activity in x‐, y‐, and z‐side views. For better visualization, the z‐axis was log10 transformed. Subpopulation 0 is not shown because of the absence of a residual activity peak. Five defined subpopulations are depicted using a color code. Subpopulation 2 is located in the area of lower × (left‐shifted) and subpopulations 3 and 4 are in the area of higher × (right‐shifted). Subpopulations 3 and 4 differ in z‐coordinates. Subpopulation 5 groups within the highest z. (B) Visual distribution of studied PAH genotypes in PAH activity landscapes based on the parameters of residual peak activity x‐, y‐, and z‐top views. Subpopulation 0 is not shown because of the absence of a residual activity peak. Five defined subpopulations are depicted using a color code. Subpopulation 2 is located in the area of lower × (left‐shifted) and subpopulations 3 and 4 are in the area of higher × (right‐shifted). Subpopulations 3 and 4 differ by y‐coordinate. (C) [Phe] at peak residual activity in various subpopulations. The peak [Phe] for the PAH WT is indicated by a dashed line. Subpopulation 1 demonstrates peaks in a position similar to the WT, subpopulation 2 is characterized by left‐shifted peaks (lower [Phe]), and subpopulations 3 and 4 by right‐shifted peaks (higher [Phe]). The peaks of residual activity for genotypes in subpopulation 5 were located in the area of normal or left‐shifted peaks. (D) Maximum residual activity at the peak in various subpopulations. The two subpopulations with right‐shifted peaks, 3 and 4, differed in their maximum residual activities, with subpopulation 3 demonstrating higher values. Subpopulation 5 exhibited the highest residual peaks among all the genotypes. (E) BH₄‐treatment response in subpopulations (clinical data sourced from BioPKU [9]). Subpopulation 0 was associated with 100% non‐responders. Genotypes from subpopulation 1 were associated with variable response rates owing to significantly reduced residual activity. In subpopulations 2 and 5, response rates close to 100% were observed owing to left‐shifted peaks and high residual peaks, respectively. Subpopulation 3 response rates were more variable owing to the right‐shifted peaks. In comparison, in subpopulation 4, right‐shifted peaks and significantly reduced residual activity led to even lower detection of BH₄ response. (F) Clinical phenotypes in subpopulations (clinical data sourced from BioPKU [9]). Phenotypes presented as classical PKU with blood [Phe] > 1200 μmol/L, mild PKU with blood [Phe] ranging 600–1200 μmol/L, or MHPA with blood [Phe] < 600 μmol/L. Subpopulation 0 was associated with 100% classical PKU. Subpopulations 1, 3, and 4 demonstrated a mixture of possible phenotypes, with subpopulation 4 having the greatest chance of classical PKU. Subpopulations 2 and 5 represent patients with milder phenotypes, whereas subpopulation 5 does not include any patients with classical PKU.

FIGURE 3

The PAH activity landscapes of genotypes corresponding to subpopulations 0–5. One example is shown for every defined subpopulation of PAH activity landscapes, and 2D and 3D panels illustrate the residual PAH activity of the corresponding genotypes normalized to the WT maximum activity. (A) Subpopulation 0: PAH activity landscapes of genotypes exhibiting no residual PAH activity; example genotype p.[Arg408Trp];[Arg408Trp]. The genotypes of this group showed no residual enzyme activity, as the designated PAH peak activity area spanning up to 1500 μM Phe displayed no discernible peaks. A heightened Tyr concentration was observed towards higher [Phe] and [BH₄], which resulted from the natural conversion of Phe into Tyr. However, this artifact does not indicate any residual PAH activity, lies outside the metabolic space existing in human liver cells, and can only be modeled in laboratory settings. The absent peak of residual activity in subpopulation 0 directly correlates with severe clinical presentation and no response to the BH₄‐treatment. (B) Subpopulation 1: PAH activity landscapes of genotypes with a standard peak position and reduced PAH activity, example genotype p.[Ala104Asp];[Ser349Pro]. Typically, this subpopulation displays a PAH residual activity peak that is significantly reduced in comparison with the PAH WT but remains in a similar shape and position. Some activity landscapes of this subpopulation demonstrated discernible peaks of residual activity that were not sufficiently prominent. Genotypes of subpopulation 1 are mostly associated with MHPA or mild PKU phenotypes and a positive response to the BH₄‐treatment. (C) Subpopulation 2: PAH activity landscapes exhibiting left‐shifted peaks of residual PAH activity, example genotype p.[Ile306Val];[Ile306Val]. This subpopulation was defined by left‐shifted, narrowly peaked profiles. Notably, the working range of the enzyme predominantly falls within lower Phe concentrations. Genotypes that belong to subpopulation 2 typically lead to mHPA and a positive response to the BH₄‐treatment. (D) Subpopulation 3: PAH activity landscapes with right‐shifted higher peaks, example genotype p.[Arg297His]; [Arg408Trp]. The PAH activity landscapes for genotypes representing this subpopulation were associated with wide right‐shifted peaks. Despite the relatively high residual peaks, a right shift often leads to undetectable residual activity below 200 μM Phe. Genotypes of subpopulation 3 could lead to more severe clinical presentations; however, the response to the BH₄‐treatment is mostly directly detectable, due to the high residual activity peaks. (E) Subpopulation 4: PAH activity landscapes with right‐shifted lower peaks, example genotype p.[Arg261Gln];[Arg261Gln]. The PAH activity landscapes for genotypes representing this subpopulation were associated with low, wide, right‐shifted residual activity peaks. The rightward shift of both the peak and working range resulted in undetectable enzyme activity below 200 μM Phe. Genotypes grouped to subpopulation 4 are characterized by more frequent severe clinical presentation and a challenging BH₄‐loading test. (F) Subpopulation 5: PAH activity landscapes with high residual activity, example genotype p.[Glu390Gly];[Glu390Gly]. The subpopulation includes activity landscapes of various shapes and peak locations; however, a common characteristic of this subpopulation is high residual activity. In the case of subpopulation 5, very high residual activity peaks allow for mild phenotypes and BH₄‐response.