Mutagenesis reactions to prepare His-hDAAO R120E and R120L variants were performed on the pET11b-His-hDAAO wild-type expression plasmid (Molla et al., 2006) using the QuickChange site-directed mutagenesis kit (Stratagene, Santa Clara, CA, USA). Previous studies showed that the His-tag at the N-terminal end of hDAAO did not alter its activity or stability (Molla et al., 2006; Romano et al., 2009). Recombinant wild-type and R120 hDAAO variants were expressed in BL21 (DE3) Star E. coli cells and purified as reported in Molla et al. (2006) and Romano et al. (2009); 40 μM of free FAD was present during all purification steps. The final enzyme preparations were stored at −20°C in 20 mM Tris-HCl buffer, pH 8.0, 100 mM NaCl, 10% (v/v) glycerol, 40 μM FAD, and 5 mM 2-mercaptoethanol (added for long-term storage). Final enzyme preparations have a purify degree >95% as judged by SDS-PAGE analysis. The enzyme concentration was determined spectrophotometrically by using the extinction coefficient at 445 nm (12.2 mM⁻¹ cm⁻¹) (Molla et al., 2006).

DAAO activity was assayed with an oxygen electrode at pH 8.5, air saturation, and 25°C, using 28 mM D-alanine as the substrate in the presence of 0.2 mM FAD (Molla et al., 2000, 2006). The kinetic parameters were determined on different D-amino acids, employing increasing substrate concentrations (Molla et al., 2000, 2006). The apparent kinetic parameters were calculated using the initial reaction rate according to a Michaelis–Menten equation using the Kaleidagraph software (Synergy Software, Reading, PA, USA).

Ligand and FAD-binding to hDAAO were investigated by following the quenching of protein fluorescence (0.3 mg protein/mL) (Molla et al., 2000; Caldinelli et al., 2009; Murtas et al., 2017). Protein fluorescence spectra were recorded between 300 and 400 nm, with excitation at 280 nm using a Jasco FP-750 instrument (excitation slit of 10 nm; emission slit of 5 nm) in 10 mM potassium phosphate, pH 8.0, 1% (v/v) glycerol, 5 mM 2-mercaptoethanol, and 40 μM FAD.

Circular dichroism (CD) spectra were recorded in 10 mM Tris-HCl, pH 8.0, 1% (v/v) glycerol, and 40 μM FAD by using a Jasco J-815 spectropolarimeter and analyzed employing a dedicated software (Jasco Co., Cremella, Italy). Cell path length was 1 cm for measurements above 250 nm (at 0.4 mg protein/mL) and 0.1 cm for measurements in the 190- to 250-nm region (at 0.1 mg protein/mL) (Caldinelli et al., 2009, 2010).

Temperature-ramp experiments were performed in 10 mM potassium phosphate, pH 8.0, 1% glycerol, 5 mM 2-mercaptoethanol, and 40 μM FAD. A software-driven, Peltier-equipped fluorimeter was used to measure the fluorescence protein changes at 340 nm and Peltier-equipped CD spectropolarimeter to follow the CD signal at 220 nm: a temperature gradient of 0.5°C/min (Caldinelli et al., 2009, 2010).

The molecular mass in solution of hDAAO variants (0.1–10 mg/mL protein) was investigated by means of gel-permeation chromatography on a Superdex 200 Increase column, using an Akta chromatographic system (GE Healthcare, Uppsala, Sweden), at room temperature and in 20 mM Tris–HCl, pH 8.5, 150 mM NaCl, 5% glycerol, 40 μM FAD, and 5 mM 2-mercaptoethanol.

Human U87 glioblastoma cells (ATCC) were cultured in DMEM supplemented with 10% fetal bovine serum, 1 mM sodium pyruvate, 2 mM L-glutamine, 1% non-essential amino acids, 1% penicillin/streptomycin, and 1% amphotericin B (Euroclone), at 37°C in 5% CO₂. The cells were transfected either with the pEYFP-hDAAO-C3 construct (for the expression of the wild-type enzyme downstream of the enhanced yellow fluorescent protein, EYFP) (Sacchi et al., 2008, 2011), the pEYFP-hDAAO(R120L)-C3, or the pEYFP-hDAAO(R120E)-C3 expression vectors. The latter constructs were obtained by mutagenesis using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent), and the following primers: 5′-ATTTCGGAAGCTGACCCCCGAGGAGCTGGATATGTTCCCAG-3′ and 5′-CTGGGAACATATCCAGCTCCTCGGGGGTCAGCTTCCGAAAT-3′ for the R120E substitution, and 5-TTCGGAAGCTGACCCCCTTAGAGCTGGATATGTTCC-3′ and 5′-GGAACATATCCAGCTCTAAGGGGGTCAGCTTCCGAA-3′ for the R120L substitution.

For transient transfection, 2 ×10⁴ cells were seeded into eight-well chamber slides (Sarstedt). At 24 h after seeding, cells were transfected with 2 μg plasmid DNA using a 4:1 ratio of FuGENE transfection reagent (Promega) to DNA. Then, 12.5 μL of transfection mix was added to 500 μL medium per well. Protein expression levels were monitored by using a fluorescence microscope (Olympus IX51) equipped with a FITC filter and by Western blot analysis using an anti-hDAAO antibody (Davids Biotechnologie) (Sacchi et al., 2008, 2011).

At 24, 48, and 72 h after transfection, cells were washed twice with PBS and fixed with ice-cold 4% p-formaldehyde and 4% sucrose for 10 min at room temperature. After fixation, cells were washed twice with PBS and blocked by incubation in PBS supplemented with 0.2% Triton X-100 and 4% horse serum for 30 min at room temperature. Peroxisomes were stained using rabbit polyclonal anti-PMP70 (peroxisomal membrane protein 70, Sigma), diluted 1:500 in PBS, 0.1% Triton X-100, and 4% horse serum overnight at 4°C; they were then washed twice in PBS and 1% horse serum and incubated for 1 h at room temperature (light protected) with goat anti-rabbit Dylight 550 secondary antibodies (Abcam) diluted 1:400 in PBS, 0.1% Triton X-100, and 1.5% horse serum. Cells were then extensively washed in PBS and nuclei were stained with DRAQ5™ (Thermo Scientific) and diluted 1:500 in PBS for 10 min at room temperature in the dark. Cells were washed three times with PBS and stored in PBS at 4°C until imaging.

Immunostained cells were imaged using an inverted laser scanning confocal microscope (TCS SP5, Leica Microsystems), equipped with a 63.0 ×1.25 NA plan apochromatic oil immersion objective. Confocal image stacks were acquired using the Leica TCS software with a sequential mode to avoid interference between each channel and without saturating any pixel. Cells in 10–15 non-magnified optical fields were analyzed for each condition (i.e., different EYFP-hDAAO expression constructs at different times after transfection). The fluorescence signal corresponding to the overexpressed EYFP-hDAAO variants was monitored by excitation at 514 nm (yellow channel); the peroxisomal and nuclear cellular compartments were imaged by excitation at 546 nm (red channel) and 647 nm (blue channel), respectively.

Analysis of the hDAAO primary sequence for nuclear localization (e.g., by NucPred or cNL Mapper) and for the presence of nuclear localization sequences (e.g., by Moseslab, NLSdb by RostLab, or SeqNLS) does not predict hDAAO targeting to the nucleus. A putative nuclear translocation signal for rDAAO (117T-P-S119) was recently proposed (Luks et al., 2017b): this might facilitate nuclear import through importin 7 after phosphorylation of such a sequence. In hDAAO, this tripeptide corresponds to the 118T-P-R120 sequence.

R120 belongs to α-helix 6 and is involved in electrostatic interactions with the carboxylic group of D73 (α-helix 3) and hydrogen-bonded to the backbone C = O groups of D109 (α-helix 5), T110, and V111 (loop between α-helix 5 and β-strand 4) located at the interface of the opposite monomer (Figure 1C). Indeed, and as stated in the Introduction section, the R120 side chain possesses two alternative orientations, one of which is observed in the benzoate-bound complex only. R120 is located at the monomer-monomer interface (Figure 1B); therefore, a connection between protomer interaction and the binding of the ligand and/or the cofactor is feasible.

Based on these considerations, and to shed light on benzoate binding, monomer-monomer interaction, and nuclear translocation, we designed two hDAAO variants differing at position 120 aimed at eliminating the positive charge (introducing a Leu) or at mimicking the negative charge of a phosphorylated serine present at the active NTS in rDAAO (introducing a glutamate). A serine at position 120 was not introduced since no evidence is reported of phosphorylation in hDAAO (Murtas et al., 2017).

hDAAO R120E and R120L variants were expressed in E. coli following the same conditions used for expressing the wild-type enzyme and were purified by HiTrap chelating chromatography as previously reported (Molla et al., 2006). The protein preparations have a purity degree of 95% and the overall purification yield is 2.5- to 4-fold lower than the wild-type hDAAO (Table 1).

The R120E and R120L hDAAO variants show the typical absorbance spectrum of FAD-containing flavoenzymes in the oxidized form (Figure 2A): absorbance maxima at 448, 384, and 279 nm and an Abs_279nm/Abs_448nm ratio of 10.4 and 11.2 for R120E and R120L variants, respectively, which are close to the value of 10.6 reported for the wild-type enzyme (Molla et al., 2006).

The far-UV CD spectra show a different signal for hDAAO R120E as compared to the wild-type enzyme and the R120L variant, indicating a lower α-helix content (Figure 2B). Concerning the tertiary structure, the near-UV CD spectra of hDAAO R120E and R120L are slightly altered in the 250- to 290-nm region, corresponding to the signal for aromatic residues, in comparison to the wild-type enzyme (Figure 2C). Accordingly, the protein fluorescence spectra for both of the variants at position 120 show a lower emission intensity of the peak at 330 nm, particularly for the R120E hDAAO (Figure 2D), strengthening the evidence that aromatic residue exposure has been altered.

Overall, the substitution of R120 results in a slight alteration in the conformation (secondary and tertiary structure) of hDAAO, which is most apparent for the R120E variant.

The activity assays performed in the absence of exogenous cofactor or the presence of an excess of FAD result in similar values indicating that purified R120E and R120L hDAAO are present in solution as holoenzyme, this being different from the wild-type enzyme, which is partially present in the apoprotein form in the absence of exogenous FAD (Table 1; see below). Notably, the R120E and R120L hDAAO variants show higher specific activity than the wild-type enzyme in the standard assay (on 28 mM D-alanine, Table 1).

The apparent kinetic parameters were calculated according to a Michaelis–Menten equation using the initial velocity values determined at increasing D-amino acid concentrations and a fixed O₂ concentration (air saturation) at pH 8.5 and 25°C, in the presence of 0.2 mM FAD to avoid apoprotein formation for wild-type hDAAO. Three D-amino acids were used: D-Ala is the reference substrate for DAAOs (Pollegioni et al., 2007), D-Ser is considered the main substrate for hDAAO in the brain (based on the concentration and physiological role) (Sacchi et al., 2012; Murtas et al., 2017; Weatherly et al., 2017), and D-Cys is the putative preferred substrate in specific tissues (Shibuya et al., 2013; Murtas et al., 2017). Both variants show a 2-fold higher apparent kinetic efficiency (k_cat/K_m) for D-Ala than the wild-type hDAAO (due to a higher k_cat; Table 2). On the other hand, using D-Ser as a substrate, all the enzyme variants show similar kinetic efficiency, although both apparent k_cat and K_m values are significantly higher in the R120-substituted variants (2.5- and 3.5-fold for the R120E and R120L, respectively; Table 2). An inhibition effect is observed at very high D-Cys concentration for both R120 variants (K_i > 200 mM), which is absent for wild-type hDAAO. Moreover, the kinetic efficiency of the R120L variant is 2-fold higher than the wild-type hDAAO on this substrate due to an increase in the apparent k_cat value.

In order to assess the effect of the kinetic properties of R120 hDAAO variants on the oxidation of D-Ser under conditions similar to the physiological ones, the enzymatic activity was assayed on 0.3 or 1 mM D-Ser in the absence of free FAD or in the presence of 2.5 μM (corresponding to the maximal estimated concentration in the brain) or 40 μM FAD (a saturating cofactor concentration) (Cappelletti et al., 2015). In all cases, the conditions affected the activity only slightly (Figure 3). At low D-Ser and FAD concentrations higher activity is measured for the R120L variant, while at 1 mM D-Ser and 40 μM FAD, the wild-type hDAAO is the most active variant.

The interaction of R120 hDAAO variants with the classical inhibitor benzoate and with CBIO, the prototype of the new class of hDAAO inhibitors, was investigated following the protein fluorescence signal. Similar to what was observed for the wild-type hDAAO (Murtas et al., 2017), benzoate binding is a biphasic process for R120 variants. The main change observed was a ~3-fold higher K_d for the second phase of binding for the R120E variant, i.e., ~8 vs. 2.6 mM for the wild-type hDAAO (Table 3). Differently from benzoate, the binding of CBIO is a monophasic process for wild-type and variants of hDAAO.

The K_d for FAD was estimated for the R120 hDAAO variants based on the quenching of the apoprotein fluorescence during FAD titration, both in the absence and presence of the active site ligand benzoate. Similar to the wild-type enzyme, FAD-binding is a biphasic process for both R120 variants in the absence of benzoate (Figure 4A): the fluorescence intensity change associated with the first phase accounts for 60–65% of the overall change. The main difference concerning the wild-type hDAAO is a 10-fold lower dissociation constant for the first phase (K_d1) of cofactor binding to the apoprotein of R120 variants (Table 3). Significant alterations are also apparent when FAD binding to hDAAO apoprotein is performed in the presence of a saturating benzoate concentration (i.e., 70 μM, Figure 4B): this is a monophasic process for the wild-type hDAAO and a biphasic event for the R120 variants. Indeed, a tighter binding of the first process is apparent for the R120L variant only, while similar K_d1 values are determined in the presence of benzoate for the wild-type and the R120E variants (Table 3).

Taken together, these findings indicate that R120 substitution modifies the interaction of hDAAO apoprotein with the flavin cofactor.

Wild-type hDAAO is a stable homodimer (Molla et al., 2006). R120 is located at the interface between monomers (Figure 1A) and, therefore, the oligomeric state of hDAAO R120E and R120L holoenzymes was investigated by gel-permeation chromatography. Both variants elute as a dimer in the 0.1–10 mg/mL protein concentration range (elution volume of ~13.7 mL, corresponding to a molecular mass of 80 kDa, not shown).

The stability of hDAAO secondary and tertiary structures was investigated by performing temperature-ramp experiments and analyzing the CD signal at 220 nm and the tryptophan fluorescence at 340 nm, respectively. The melting temperature values determined for the R120E and R120L variants are 3.5–5°C lower than those of wild-type hDAAO when the changes in tertiary structure are analyzed and ~10°C lower when the secondary structure stability is studied (Table 1).

U87 glioblastoma cells were transiently transfected with a pEYFP-hDAAO-C3 plasmid encoding for EYFP-hDAAO chimeric proteins harboring R, E, or L at position 120 of the flavoenzyme. The presence of the fluorescence tag at the N-terminal end was previously demonstrated to not alter the activity, subcellular distribution, or degradation pathway of wild-type hDAAO (Sacchi et al., 2008, 2011; Li et al., 2013). A similar amount of transfected cells (expressing EYFP-hDAAO protein) was obtained for all three constructs (Table 4). The percentage of cells showing a cytosolic, peroxisomal, and nuclear localization of the flavoenzyme was determined by comparing the yellow fluorescence signal of EYFP-hDAAO variants with those of the markers PMP-70 (for peroxisomes) and DRAQ5™ dye (for nuclei, Figure 5). The amount of cytosolic wild-type EYFP-hDAAO decreased with time after transfection (28% of the signal is cytosolic at 72 h vs. 94% at 24 h), as did the amount of nuclear signal (Table 4 and Figure 6). The presence of glutamate at position 120 of hDAAO to mimic a phosphorylation state favors the mistargeting by increasing the number of cells with an EYFP nuclear signal as compared to wild-type hDAAO. This is accompanied by a lower percentage of cells displaying a peroxisomal signal for the EYFP-tagged protein. Introducing an uncharged residue at position 120 (R120L variant) also negatively affects peroxisomal targeting and modifies, to a limited extent, the percentage of cells with an EYFP-hDAAO nuclear signal as compared to the cells expressing the wild-type hDAAO (Figure 5). The substitution of R120 also appears to increase the propensity of hDAAO to form cytosolic aggregates (Table 4). Notably, in all cases, the cells showing nuclear localization of hDAAO also possess a signal largely diffused in the cytosol.

The number of viable cells following transfection is similar for wild-type and R120 hDAAO variants, indicating that mistargeting does not affect cell viability.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.