Phenylalanine Monitoring via Aptamer-Field-Effect Transistor Sensors

Abstract

Determination of the amino acid phenylalanine is important for lifelong disease management in patients with phenylketonuria, a genetic disorder in which phenylalanine accumulates and persists at levels that alter brain development and cause permanent neurological damage and cognitive dysfunction. Recent approaches for treating phenylketonuria focus on injectable medications that efficiently break down phenylalanine but sometimes result in detrimentally low phenylalanine levels. We have identified new DNA aptamers for phenylalanine in two formats, initially as fluorescent sensors and then, incorporated with field-effect transistors (FETs). Aptamer-FET sensors detected phenylalanine over a wide range of concentrations (fM-mM). para-Chlorophenylalanine, which inhibits the enzyme that converts phenylalanine to tyrosine, was used to induce hyperphenylalaninemia during brain development in mice. Aptamer-FET sensors were specific for phenylalanine vs para-chlorophenylalanine and differentiated changes in mouse serum phenylalanine at levels expected in patients. Aptamer-FETs can be used to investigate models of hyperphenylalanemia in the presence of structurally related enzyme inhibitors, as well as naturally occurring amino acids. Nucleic acid-based receptors that discriminate phenylalanine analogs, some that differ by a single substituent, indicate a refined ability to identify aptamers with binding pockets tailored for high affinity and specificity. Aptamers of this type integrated into field-effect transistors enable rapid, electronic, label-free phenylalanine sensing.

Graphical Abstract

INTRODUCTION

Phenylketonuria (PKU) is an autosomal recessive genetic disorder involving mutations in the gene that encodes phenylalanine hydroxylase (PAH), a liver enzyme that converts the essential amino acid phenylalanine to tyrosine (Figure 1A).^1–7 In the United States, PKU occurs in ~1/10,000–15,000 babies yearly. Diagnosis at birth is critical.^1,3,4,7,8 This ‘inborn error of metabolism’ leads to hyperphenylalaninemia in the blood and brain.^1–5,7,9 Elevated phenylalanine causes abnormalities in brain development associated with permanent intellectual impairment. Screening newborns for PKU involves laboratory testing by a bacterial inhibition assay, i.e., the Guthrie test,¹⁰ or more recently, tandem mass spectrometry.¹¹ Although sufficient for diagnosis, these methods involve turnaround times of days to inform treatment providers and families.¹²

Figure 1. Aptamers for phenylalanine.

(A) In humans and mice, phenylalanine is para-hydroxylated to form tyrosine by the liver enzyme phenylalanine hydroxylase (PAH). The genetic disorder phenylketonuria is caused by mutations in the PAH gene, which result in high blood and brain levels of phenylalanine. (B) Phenylalanine analogs para-chlorophenylalanine (PCPA) and para-ethynylphenylalanine (PEPA). (C-E) Three phenylalanine-specific aptamer sequences (Phe 1, Phe 2, and Phe 3) were isolated for sensor development. All three aptamers showed concentration-dependent responses towards phenylalanine determined via competitive fluorescence assays. Responses were measured for other aromatic amino acids (tyrosine and tryptophan) and the phenylalanine analogs (PCPA and PEPA). Fluorescence concentration curves enabled determination of solution dissociation constants (K_d) for Phe 1 (10 μM), Phe 2 (7 μM), and Phe 3 (16 μM). Here, N=6 for each phenylalanine concentration; N=3 for each nonspecific target concentration. Standard errors of the means for each datum were too small to be displayed in some cases; RFU is relative fluorescence units.

Phenylketonuria is primarily managed through strict avoidance of phenylalanine-containing foods.^{1,3,4,7,13,14} Early-life dietary management largely prevents the damaging effects of PKU on brain development.^15–17 However, even modestly uncontrolled blood phenylalanine levels in children and adults have been correlated with cognitive and psychiatric sequelae.^1,7,18–21 Nevertheless, studies on this topic are sparse, partly because technologies for straightforward monitoring of blood phenylalanine levels in clinical studies are not readily available.^1,18

Phenylalanine serum concentrations in healthy individuals are 60±30 μM.¹⁷ Current treatment guidelines for individuals with PKU are to maintain blood phenylalanine levels between 120 and 360 μM. Moderate hyperphenylalaninemia is associated with blood levels ranging from 360 to 600 μM, while untreated PKU is characterized by phenylalanine levels >1000 μM (with concentrations >3000 μM having been reported).^1,3,7,22

Emerging treatments for PKU are based on enzyme replacement with pegylated versions of bacterial phenylalanine ammonia lyase.²³ Enzyme substitution decreases phenylalanine levels in patients with PKU and reduces the need for dietary restrictions, which can be exceedingly difficult to maintain over a lifetime.²³ In 2017, the U.S. Food and Drug Administration approved the first enzyme substitution therapy, pegvaliase (Palynziq™).²³ However, pegvaliase and related treatment strategies can result in hypophenylalaninemia, which negatively affects protein synthesis, and thus, growth in children, and neurotransmitter synthesis in children and adults.^17,23 In light of the present clinical picture, improvements in phenylalanine monitoring are indicated to provide insights into the pharmacokinetics²⁴ and efficacy of current and future PKU treatment strategies and their combinations.²⁵ Enabling patients to determine their own blood phenylalanine levels will be empowering. Regardless, there are no at-home or point-of-care options for measuring phenylalanine.^26–28

Field-effect transistors (FETs) modified with protein receptors have been developed for rapid electronic, reagentless detection of biological targets.^29–37 Receptor recognition of charged targets and/or target-induced receptor reorientation gate the semiconductor channels of FETs to modulate transconductance. However, bioFETs based on protein receptors or antibodies are not feasible for direct sensing in biological fluids because large receptors (~10 nm) are too far away from semiconductor channels relative to the Debye screening length. At physiological ion concentrations, the Debye length is on the order of 1 nm.^30,38

Recently, we developed thin-film metal oxide FETs functionalized with oligonucleotide stem-loop receptors (aptamers).^39–42 We discovered that target-mediated reorganization of a portion of the highly negatively charged backbones of small oligonucleotides (and rearrangement of associated counter ions) occurs near enough to the semiconductor surfaces to circumvent Debye-length limitations under biological conditions.^43–46 Aptamer-FETs have enabled sensitive and selective detection of small molecules, including neutral targets, in physiological buffers and fluids, and complex biological matrices.³⁰

Here, previously unreported aptamers for direct phenylalanine detection were coupled with nanometer-thin In₂O₃ FETs. Phenylalanine was detected over many orders of magnitude in physiological solutions. Selectivity for phenylalanine over closely structured endogenous and exogenous aromatic amino acid analogs (Figure 1B) and metabolites was excellent, particularly for one of the phenylalanine aptamers. Phenylalanine sensing was carried out in serum from mice with induced hyperphenylalaninemia, demonstrating the ability of aptamer-FETs to detect biologically important differences in phenylalanine concentrations. Aptamer sequences were rationally modified to shift device sensitivities towards physiologically relevant concentrations. These findings portend the use of aptamer-functionalized FETs for phenylalanine monitoring in PKU patients and other relevant populations.

Experimental Section

No unexpected or unusually high safety hazards were encountered.

Materials

All materials were purchased from Sigma-Aldrich Co. (St. Louis, MO), unless otherwise noted. The amino acids used throughout were the L-forms, i.e., L-phenylalanine (Sigma P5482 and P2126), L-tyrosine (Sigma T3754), and L-tryptophan (Sigma 93659 and T0254). Oligonucleotides were obtained from Integrated DNA Technologies (Coralville, IA). The SYLGARD 184 used to make polydimethylsiloxane (PDMS) wells was from Dow Corning Corporation (Midland, MI). Water was deionized before use (18.2 MΩ) with a Milli-Q system (Millipore, Billerica, MA).

para-Ethynylphenylalanine (PEPA) was synthesized by a previously reported route⁴⁷ with minor modifications. Trimethylsilyl acetylene was coupled under palladium catalysis to to N-acetyl 4-iodophenylalanine methyl ester,⁴⁸ followed by deprotection and purification. The nuclear magnetic resonance spectroscopy (NMR) data matched spectra reported in the literature (Figure S1).

Phenylalanine aptamer selection

Phenylalanine aptamer selection was carried out as per a previously reported strategy.⁴¹ Initial selection resulted in isolation of an aptamer for phenylalanine complexed with pentamethylcyclopentadienyl rhodium(III) (Cp*Rh). This previously reported aptamer, which we designate Phe-Cp*Rh 1, showed cross-reactivity with the analogous tryptophan-Cp*Rh complex (Trp-Cp*Rh), limiting its use in practical applications. To reduce cross-reactivity, additional selections for Phe-Cp*Rh were performed with Trp-Cp*Rh counter-selection,^41,42 which resulted in the isolation of two new Phe-Cp*Rh aptamers, which we designate Phe-Cp*Rh 2 and Phe-Cp*Rh 3 (Table S1). These previously unreported Phe-Cp*Rh aptamers were not cross-reactive with Trp-Cp*Rh (Figure S2C,S3). However, Phe-Cp*Rh 2 showed cross reactivity with Tyr-Cp*Rh (Figure S2C).

Unexpectedly, as byproducts of the selections for additional Phe-Cp*Rh aptamers, we identified three aptamers that directly recognize phenylalanine, rather than the Phe-Cp*Rh complex. These previously unreported, direct-binding aptamers are designated Phe 1, Phe 2, and Phe 3 (Table S1). An excess of phenylalanine (1 mM) was used in the selection procedures to produce the Phe-Cp*Rh complex such that Cp*Rh was the limiting reagent. Under these conditions, a large excess of free phenylalanine, which also interacted with the oligonucleotide sequences, was present along with the Phe-Cp*Rh complex in the target solutions. We compared the responses of Phe 1 to phenylalanine with and without Cp*Rh (Figure S4). In the presence of Cp*Rh, phenylalanine prefers to complex with Cp*Rh, however the remaining free phenylalanine in solution was detected by Phe 1.

Fluorescence assays were carried out in 20 mM HEPES, 1 M NaCl, 10 mM MgCl₂, and 5 mM KCl (pH 7.5) using a Victor II microplate reader (PerkinElmer). Each aptamer sequence was modified with fluorescein at the 5’ end. Complementary strands were 3’-modified with dabcyl for solution quenching determination of dissociation constants (K_d) (Table S1) and for selectivity testing. Concentrations of aptamers and complementary strands were empirically determined and aptamer K_d values were calculated as per previously published methods.^30,40,41 Putative aptamer secondary structures shown were predicted using Mfold (http://unafold.rna.albany.edu/?q=mfold).

Aptamer-functionalized field-effect transistors

Field-effect transistors were fabricated using ~4 nm In₂O₃ semiconductor films as channel materials with high surface-to-volume ratios. Aqueous solutions (0.1 M) of indium(III) nitrate hydrate (In(NO₃)₃•xH₂O, 99.999%) were spin-coated at 3000 rpm for 30 s onto heavily doped silicon wafers (University Wafer, Boston, MA or WaferPro, San Jose, CA) with 100-nm thermally grown oxide layers. After coating, substrates were pre-heated at 150 °C for 10 min followed by 3 h of annealing at 350 °C.^29,49 Source and drain electrodes (10 nm Ti/30 nm Au) were deposited by electron-beam evaporation and patterned via standard photolithography.

To functionalize thin-film FETs with thiolated phenylalanine aptamers, we first self-assembled mixed monolayers of (3-aminopropyl)trimethoxysilane and trimethoxy(propyl)silane (1:9 v/v ratio) using vapor-phase deposition on In₂O₃ surfaces at 40 °C for 1 h. Substrates were then incubated with a 1 mM ethanolic solution of 1-dodecanethiol for 1 h to passivate Au electrodes via alkanethiol monolayer formation. After rinsing with ethanol, substrates were incubated with a 1 mM solution of 3-maleimidobenzoic acid N-hydroxysuccinimide ester (MBS) in 1:9 (v/v) dimethyl sulfoxide and phosphate-buffered saline (1× PBS, pH 7.4) for 30 min. Thiolated DNA aptamers (Table S2) were diluted to 1 μM in nuclease-free water and heated for 5 min at 95 °C followed by rapid cooling in an ice bath. Substrates were then immersed in aptamer solutions for 24 h, rinsed with deionized water, and dried under N₂ gas.

A scrambled aptamer sequence for Phe 3 was designed using mfold to have a secondary structure that was predicted to differ from the correct phenylalanine aptamer sequence, while maintaining the same numbers and types of nucleotides (Figure S5B and Table S2).

Field-effect transistor measurements

Polydimethylsiloxane wells were sealed on individual FETs to hold sensing solutions. Substrates had inter-FET distances (~2 mm) that were large enough to enable isolation of single FET devices within the PDMS wells (Figure S6). Ringer’s solution (147 mM NaCl, 4 mM KCl, 2.25 mM CaCl₂) was used as the electrolyte solution. The Ag/AgCl reference electrodes (World Precision Instruments, Inc., Sarasota, FL) were placed in sensing solutions in a top-gate (solution-gate) device configuration. Measurements were performed using a manual analytical probe station (Signatone, Gilroy, CA) equipped with a Keithley 4200A-SCS (Tektronix, Beaverton, OR) semiconductor parameter analyzer.

Source-drain current (I_DS) transfer curves were obtained wherein gate voltages (V_GS) were varied from 0 to 400 mV with a step voltage of 5 mV. The drain voltage (V_D) was held at 10 mV throughout. Five sweeps were averaged for each transfer curve. Calibrated responses were calculated by dividing the absolute sensor response (ΔI), which takes into account baseline subtraction, by the change in source-drain current with voltage sweep (ΔI_DS/ΔV_G).⁵⁰ Aptamer-FET responses at V_G=375 mV were used to calculate mean calibrated responses.

Mouse serum

Mice were generated at the University of California, Los Angeles (UCLA) from a core colony of a serotonin transporter (SERT)-deficient lineage maintained on a mixed CD1 × 129S6/SvEv background via heterozygous SERT-deficient (SERT+/−) pairings. For this study, three pairs of SERT wildtype (SERT+/+) mice from the core colony were bred to produce 18 wildtype pups. All mice were maintained on a 12-h light/dark cycle (lights on at 0600 h (Zeitgeber time 0)) with ad libitum food and water. The Association for Assessment and Accreditation of Laboratory Animal Care International has fully accredited UCLA. All animal care and use met the requirements of the NIH Guide for the Care and Use of Laboratory Animals, revised 2011. The UCLA Chancellor’s Animal Research Committee (Institutional Animal Care and Use Committee) preapproved all procedures.

For postnatal treatment, pups from each litter were randomly assigned to one of three groups: (1) Saline (vehicle control); (2) 100 mg/kg para-chlorophenylalanine (Sigma #C3635);⁵¹ or (3) 10 mg/kg para-ethynylphenylalanine.⁵² Doses were calculated based on the free base form of each compound. The pH values of PEPA and PCPA saline were adjusted to pH 7.4 prior to injection. Dosing solutions were filtered via 0.22 μm filters for sterilization.

Each litter contained all treatment groups. Each pup received a subcutaneous injection of the assigned treatment daily during ZT 6–8 on postnatal days (P)4–21. The injection volume was 10 mL/kg during P4–11, and 5 mL/kg during P12–21. A total of three of the 18 pups were excluded from this study: 1/6 saline-treated and 1/6 PCPA-treated subjects died during the postnatal period. In addition, 1/6 PEPA-treated subjects stopped receiving injections at P17 due to body weight loss for two continuous days. The data from the remaining N=5 mice per treatment group are reported.

Pups were weaned after the last injection on P21 and housed with their same-sex siblings. Two hours after the last injection, subjects were euthanized by decapitation under deep anesthesia with isoflurane. Whole blood samples were collected via cardiac puncture and placed in microcentrifuge tubes pre-chilled on ice for 30–60 min. Following coagulation, blood samples were centrifuged at 16,000 g for 15 min at 4 °C twice. After each centrifugation, supernatants were removed and transferred to clean microcentrifuge tubes on ice.

Serum samples were aliquoted and stored at −80 °C until analysis. Aptamer-FET and high performance liquid chromatography (HPLC) measurements were carried out by investigators blind to the treatment group identification of each sample. All serum samples for aptamer-FET measurements were diluted to ~10 pM phenylalanine in 1× Ringer’s buffer, using the same serial dilution strategy, i.e., first 1:10, then 1:100, and finally 1:1.5, which was based on the average phenylalanine concentration determined in mouse serum by HPLC.

Circular dichroism spectroscopy

Intensities and positions of positive and negative peaks for oligonucleotides in circular dichroism (CD) spectra correspond to exciton interactions induced by stacking of hydrophobic bases in asymmetric helices.^45,46,53 Aptamer and target concentrations were 2 μM in 1× Ringer’s buffer. Thiolated aptamers were heated at 95 °C for 5 min and cooled to room temperature slowly to relax DNA molecules into extended conformations. Spectra were collected using a JASCO J-715 circular dichroism spectrophotometer (Oklahoma City, OK). Four scans with 0.5-nm resolution, 1.0-nm bandwidth, a 4-s response time, and a 20 nm/min scan rate were acquired per sample. Scans of 1× Ringers solution were subtracted as background.

Statistics

Data for fluorescence assays and FET calibrated responses are reported as means ± standard errors of the means and were analyzed using GraphPad Prism 7.04 (GraphPad Software Inc., San Diego, CA) via one-way analysis of variance followed by Tukey’s multiple comparisons post hoc tests. Cross-validation data for phenylalanine levels were analyzed by linear regression analysis. In all cases, P<0.05 was considered statistically significant.

RESULTS AND DISCUSSION

Three previously unreported DNA aptamer sequences that directly recognize the amino acid phenylalanine were identified via solution-phase, in vitro systematic evolution of ligands by exponential enrichment (SELEX) (Figure 1C–E).^39,41,42 Dissociation constants were determined using competitive fluorescence assays.^30,40,41 Quencher-labeled complementary sequences (Table S1 and Figure S7) were displaced from fluorescently labeled aptamer sequences upon phenylalanine binding resulting in increases in fluorescence intensities, as shown in Figure 1C–E.^30,40,41 Solution dissociation constants (K_d) were 10 μM, 7 μM, and 16 μM for Phe 1, Phe 2, and Phe 3, respectively. Dissociation constants measured for Phe 1 and 2 via dye displacement assays were consistent with those measured via competitive binding assays (Figure S8). Dye displacement did not work for the Phe 3 aptamer. This result is sometimes the case for dye displacement, which is why we do not use this method as a primary means of determining dissociation constants.

Selections initially had been carried out using a strategy to try to increase aptamer selectivity towards low epitope targets by associating targets with metal complexes.⁴¹ In addition to a previously reported aptamer sequence recognizing phenylalanine complexed with pentamethylcyclopentadienyl rhodium(III) (Cp*Rh) (sequence herein referred to as Phe-Cp*Rh 1),⁴¹ two previously unreported sequences, Phe-Cp*Rh 2 and Phe-Cp*Rh 3 were characterized (Figures S2,S3).

Selectivity testing for the three direct-detection phenylalanine aptamers using competitive florescence assays showed reduced (Phe 1 and Phe 2) or negligible (Phe 3) responses towards the endogenous aromatic amino acids tyrosine and tryptophan (Figure 1C–E). We also investigated selectivity of the direct-detection phenylalanine aptamers for two phenylalanine analogs, para-chlorophenylalanine (PCPA) and para-ethynylphenylalanine (PEPA) (Figure 1B), which potentially induce hyperphenylalaninemia in animal models (vide infra).^54–58 The Phe 3 aptamer showed minimal responses to PCPA or PEPA via competitive fluorescence assays (Figure 1E), in contrast to Phe 1 and Phe 2 (Figure 1C–D), and the Phe-Cp*Rh aptamers (Figure S9), all of which had appreciable responses to PCPA.

Next, each of the phenylalanine-specific aptamers was attached to the semiconducting channels of FETs for electronic detection of phenylalanine (Figure 2A).^29,30 As in our prior aptamer-FET studies, shortened,^59,60 thiolated aptamers were attached to In₂O₃ surfaces via covalent modification to self-assembled silanes using m-maleimidobenzoyl-N-hydroxysuccinimide as a crosslinker.^29,30 Field-effect transistors were operated in a top-gate setup. The source-drain current (I_DS) was measured while sweeping the gate voltage (V_GS) during target exposure (Figure 2A).^29,30 Reorganization of surface-tethered negatively charged oligonucleotides occurred in close proximity to metal-oxide semiconducting surfaces upon target capture to gate FET transconductances resulting in target-concentration-dependent current changes under physiological ionic conditions (Figure 2B).³⁰

Figure 2. Detection of phenylalanine via aptamer-field-effect transistor sensors.

(A) Schematic of the FET platform and surface chemistry. Here, FETs were composed of 4-nm thin-film In₂O₃ as the channel material, with a 10-nm Ti adhesion layer and a 30-nm top Au layer patterned as interdigitated electrodes over the semiconductor layer. Sensing was performed by applying a source-drain bias voltage, sweeping the gate voltage with respect to a Ag/AgCl reference electrode in a solution-gated configuration, and measuring changes in source-drain currents. Thiolated aptamers were tethered to semiconductor surfaces using m-maleimidobenzoyl-N-hydroxysuccinimide ester to crosslink thiol groups to amine-terminated silanes, co-self-assembled with methyl-terminated silanes, which served as spacer molecules to optimize aptamer surface densities for target recognition. (B) Each of three phenylalanine aptamers attached to FETs (Phe 1, Phe 2, Phe 3) produced concentration-dependent responses in 1× Ringer’s solution. (C) Representative transfer (I_DS-V_GS) curves for Phe 3 aptamer-FETs upon increasing phenylalanine concentrations. (D) Circular dichroism spectra of Phe 3 in 1× Ringer’s solution before and after introduction of phenylalanine (2 μM). Spectra shown are averages of N=3 spectra each. (E) The Phe 3 aptamer, when incorporated into FETs, had negligible responses to nonspecific targets, including tryptophan (Trp), tyrosine (Tyr), phenylypyruvic acid (PPA), 2-phenylethylamine (2-PEA), para-chlorophenylalanine (PCPA), or para-ethynylphenylalanine (PEPA) compared to phenylalanine (all targets at 100 μM) [F (6, 10)=76; P<0.001]. Error bars, are standard errors of the means with N=3 for B, and N=3 for phenylalanine, PCPA, and PEPA and N=2 for Trp, Tyr, PPA, and 2-PEA in E. ***P<0.001 vs. all nonspecific targets.

Phenylalanine-aptamer-FETs showed a wide range of concentration-dependent responses to phenylalanine (fM detection limit) in 1× Ringer’s solution, which mimics the ionic composition of the plasma fraction of human blood (Figure 2B). Sensing in physiological buffers that have similar ion concentrations to the target biological matrix is important for evaluating oligonucleotide receptors because interactions with different types and concentrations of solution ions may result in alternate aptamer secondary structures and thus, differences in sensor sensitivities and selectivities.^61,62

We also tested one of the aptamers recognizing phenylalanine complexed with Cp*Rh when attached to FETs, as an example of the use of this type of aptamer. The Phe-Cp*Rh 2 aptamer showed similar concentration-dependent responses (Figure S2D) to those of the direct detection sequences (Figure 2B), suggesting that metal complexation to increase sensitivity, which would complicate point-of-care or at-home monitoring approaches, is unnecessary⁴¹ and direct phenylalanine detection is sufficient in the context of FET sensors.

Target-concentration-dependent decreases in current were observed for FET transfer curves (I_DS-V_GS sweeps) for each of the three direct sensing aptamers (Figures 2C,S10). We attribute decreases in FET transfer characteristics on n-type semiconductors to a dominant effect of negatively charged DNA aptamer backbones moving closer to semiconductor channel surfaces to increase electrostatic repulsion of charge carriers (band bending) and to gate transconductance upon target binding.³⁰ We previously determined that aptamer-FET small-molecule sensing is due to gating associated with the reorganization of charged DNA aptamer backbones and not to target charge. Target-aptamer interaction can reorient aptamer charge toward or away from semiconductor channels, based on specific aptamers.³⁰

We performed circular dichroism (CD) spectroscopy to investigate target-induced changes in aptamer secondary structures for the three phenylalanine direct-sensing aptamers, as we have been able to associate CD spectral changes with individual aptamer-FET responses.³⁰ The spectra for Phe 3 showed a small but reproducible decrease in peak intensity at 280 nm after target capture, potentially corresponding to target-induced formation of hairpin motifs (Figure 2D).⁵³ Spectra showed negligible changes in peak positions or intensities for Phe 1 or Phe 2 upon association with phenylalanine (Figure S11). These findings suggest that the latter two aptamers do not form new secondary structural motifs upon target recognition and that all three aptamers primarily undergo adaptive target binding largely involving pre-formed secondary structures.³⁰

Of the three direct-detection phenylalanine aptamers, Phe 3 showed the largest target-related responses and the smallest replication variability when integrated with FETs (Figure 2B). The Phe 3 aptamer also showed the highest selectivity towards nonspecific targets compared to Phe 1 and Phe 2 in competitive fluorescence assays (Figure 1C–E); thus, we focused on investigating this sequence further. To test selectivity on FETs, we measured the responses of Phe 3-aptamer-FETs to the aromatic amino acid tryptophan, the phenylalanine metabolites tyrosine, phenylpyruvic acid, and 2-phenylethylamine, and the two phenylalanine analogs (Figure 2E). Sensor responses upon exposure to 100 μM solutions of five of the nonspecific targets were <10% of the average response to an equivalent concentration of phenylalanine; responses to PCPA were 15% of the average phenylalanine response. Concentrations of nonspecific targets were selected based on their physiologically relevant concentrations.^63–65 As a further indication of selectivity, responses of FETs functionalized with a scrambled version of the Phe 3 sequence having the same numbers and types of nucleotides as the correct Phe 3 aptamer sequence but with a different predicted secondary structure were negligible (Figure S5).

To investigate Phe 3-aptamer-FET detection of clinically relevant phenylalanine levels, we measured phenylalanine in serum (diluted with 1× Ringers, see Experimental Section) from mice injected daily with PCPA, PEPA, or saline during postnatal days 4–21 (Figure 3A). This postnatal period in mice is the human developmental equivalent of the last trimester of pregnancy and the first postnatal year.^66,67 We selected this treatment period because it is important for determining the impact of elevated phenylalanine levels on cortical axon development.⁶⁸ Both para-substituted phenylalanine analogs inhibit tryptophan hydroxylase (TPH), which converts dietary tryptophan to 5-hydroxytryptophan. The latter is decarboxylated to produce the neurotransmitter serotonin (5-hydroxytryptamine).^52,69–71 However, whereas PCPA has been reported to inhibit PAH activity, in addition to TPH,^54–58 PEPA has been suggested to lack inhibitory effects on PAH.⁵²

Figure 3. Aptamer-field-effect transistors (FETs) differentiate serum phenylalanine levels.

(A) Schematic illustration of the in vivo experimental design. Mice were treated once per day with para-chlorophenylalanine (PCPA) or para-ethynylphenylalanine (PEPA) during postnatal days (P)4–21. These phenylalanine analogs inhibit tryptophan hydroxylase (TPH). Some studies suggest that PCPA may also inhibit phenylalanine hydroxylase (PAH). Serum samples were collected 2 h after the final injection of each analog on P21. Phenylalanine concentrations were measured by Phe 3-aptamer-FETs and cross-correlated with high-performance liquid chromatography (HPLC) as a reference method (Figure S12). (B) Mice treated with PCPA had modestly elevated serum phenylalanine concentrations compared to mice treated with PEPA or saline [F(2,12)=17.3; P<0.01]. Data points depict levels from individual animals. Error bars are standard errors of the means with N=5 for each treatment group. **P<0.01 vs. PCPA; ***P<0.001 vs. PCPA. (C) Correlation of phenylalanine concentrations measured in individual mouse serum samples via aptamer-FETs vs. HPLC with the corresponding linearity index (R²) and regression slope (m). P=0.762 (Run’s test), deviation from linearity is not significant.

Mice receiving PCPA showed a doubling of the concentration of serum phenylalanine levels (individual levels ranging from ~90 to 160 μM) compared to mice exposed to PEPA or saline (Figure 3B). Phenylalanine concentrations in mouse serum samples were cross-validated via HPLC with electrochemical detection (Figure S12 and Supplemental Methods). Individual phenylalanine serum sample levels determined using aptamer-FETs were highly correlated when compared to those measured by HPLC (R²=0.95; Figure 3C). The slope of the regression line correlating phenylalanine levels determined by both methods was ~1 indicating that Phe 3-aptamer-FETs accurately reported phenylalanine levels.

Phenylalanine levels detected by aptamer-FETs in PCPA-treated mice (~120 μM) were at the low end of the range of serum levels in humans with modestly elevated phenylalanine, i.e., PKU patients adhering to dietary restrictions, and possibly, phenylalanine levels in some PKU carriers, i.e., individuals with one mutant PAH allele (ca. one in 50 Caucasians)^72,73. Thus, aptamer-FETs can be used to differentiate modest yet physiologically relevant changes in phenylalanine levels.

The Phe 3 aptamer-FETs were sufficient to detect biologically relevant differences in phenylalanine levels in diluted serum (Figure 3). Removing base pairs from aptamer stem regions destabilizes aptamer secondary structures, particularly in stem-loop aptamers isolated by solution SELEX where stem closure plays an important role in identifying target-specific sequences.^59,74–76 We reasoned that stem truncation could be used to shift aptamer-FET sensitivity ranges closer to the therapeutic range of PKU patients, thereby reducing the need for serum dilution.

Strategically deleting complementary C-G base pairs (C-G base pairs form three hydrogen bonds vs. A-T base pairs, which are weaker in forming only two H-bonds) in the stem regions of Phe 1 and Phe 3 (Table S2) shifted the corresponding aptamer-FET concentration curves to the right, i.e., towards higher phenylalanine concentrations (Figure 4). These modifications also increased the steepness of the concentration curves such that small differences in phenylalanine levels in the steepest regions of these curves would be more highly differentiable.

Figure 4. Tuning aptamer-field-effect transistor (FET) responses via aptamer stem truncation.

(A-B) Aptamer-FET sensing of phenylalanine with truncated versions of Phe 1 and Phe 3 aptamers, respectively (truncated sequences are shown in the corresponding insets). Normalized response data for phenylalanine sensing on FETs using unmodified Phe 1 and Phe 3 sequences are presented for comparison and are reproduced from Figure 2B. Error bars are standard errors of the means with N=3 for all data.

Notably, the serum phenylalanine concentration range in PKU patients is ~10–1000 μM.¹ Thus, narrowing the sensitivity range of aptamer-FET sensors, in addition to reducing the binding affinity, i.e., shifting to higher K_d, advances this strategy toward an at-home monitoring scenario wherein phenylalanine might be determined in whole blood from finger or heel sticks without the need for dilution. Stem destabilization, in addition to decreasing aptamer densities,³⁰ and/or tuning FET characteristics⁷⁷ and measurement parameters can eventually be combined to achieve phenylalanine detection in undiluted blood.

There are numerous benefits and opportunities for at-home phenylalanine monitoring. In a National Institutes of Health (NIH) consensus panel on “PKU: screening and management”, the NIH recommended the development of reliable home testing methods for measuring phenylalanine to increase adherence to dietary treatments.⁷⁸ There is consensus from PKU patients that at-home measurement of phenylalanine will make disease management, i.e., following a PKU diet and/or enzyme substitution dosing, easier to achieve and to maintain throughout life.²⁶ National Institutes of Health guidelines for monitoring frequency in PKU patients are once weekly during the first year of life, twice per month for individuals 1–12 years of age, and at least monthly thereafter. In women with PKU prior to and during pregnancy and lactation, frequency should increase to at least weekly.

Phenylalanine has been determined using chromatographic,^79,80 plasmonic,⁸¹ and fluorimetric techniques.^79,82–84 These methods involve specialized laboratory instrumentation and sometimes, complex extractions. As such, these methods are largely incompatible with point-of-care or at-home monitoring. Colorimetric, paper-based detection platforms show promise for PKU diagnosis and monitoring in low-resource settings,^85,86 but use enzyme-based recognition elements, which can lose activity over time due to denaturation.⁸⁷ Electrochemical signal transduction has been used for phenylalanine sensing^88,89 or a conductive electrode design involving graphene oxide, which is intrinsically defect prone and synthetically heterogeneous.⁹⁰ Hasanzadeh and coworkers^91,92 reported an electrochemical strategy involving phenylalanine capture via a previously reported RNA phenylalanine aptamer (K_d~120 μM)⁹³ followed by phenylalanine oxidation at gold electrodes. Compared to RNA, aptamers comprised of DNA are advantageous in terms of higher stability when exposed to biological matrices, such as serum.⁹⁴

CONCLUSIONS AND PROSPECTS

We have identified three new DNA-based receptors that recognize the biochemically and medically important target phenylalanine. We also report two new aptamers that recognize a phenylalanine-organometallic complex (Phe-Cp*Rh). The direct-binding phenylalanine aptamers (and one of the Phe-Cp*Rh aptamers) were integrated with thin-film metal-oxide field-effect transistors (FETs). Phenylalanine-FET sensors detected phenylalanine under physiological ionic conditions over six orders of magnitude with fM detection limits. Sensors incorporating the Phe 3 direct-detection aptamer showed exceptional selectivity for phenylalanine vs. similarly structured aromatic amino acids and metabolites. The accuracy and precision of Phe 3 aptamer-FET sensors were excellent compared to a gold-standard laboratory method.⁶²

The ability to differentiate clinically relevant differences in serum phenylalanine levels with a minimal dilution step demonstrates the capability of aptamer-FETs for future use in electronic point-of-care devices for PKU diagnosis and management. In principle, aptamer-FET sensors can also be used for in vivo monitoring to investigate transiently induced or permanently maintained (through continuous administration of PCPA or via mice with constitutive reductions in PAH)⁹⁵ elevations in phenylalanine. Rapid monitoring in PKU animal models will enable investigating how changes in PAH activity or diet impact temporally resolved phenylalanine levels.

In sum, DNA aptamers are readily synthesized⁹⁶ and the FET fabrication methods used here are straightforward and easily scaled up. We have miniaturized²⁹ and nanostructured⁷⁷ In₂O₃ thin-film FETs using low-cost soft-lithographic methods.^97–99 We have also fabricated In₂O₃ FETs on flexible substrates¹⁰⁰ demonstrating capabilities to tailor device performance and architectures for specific biomedical applications.

Aptamer-FET sensors can be rapidly deployed, with minor adjustments, at the point-of-care level, e.g., in clinical laboratories. While not the focus of this work, we have developed prototype instrumentation and analysis software that is low-cost, has a relatively small footprint, and can perform multiplexed FET sensing. Translation for at-home monitoring will require additional development of FET-measurement technology to reduce size for portability, i.e., hand-held devices, and validation in clinical samples.¹⁰¹ We anticipate that phenylalanine sensing via aptamer-FETs will improve phenylketonuria disease management and monitoring in other at-risk populations.