All reagents and solvents were LC-MS grade quality unless otherwise noted. [¹³C₆¹⁵N₂]-lysine and [¹³C₆¹⁵N₁]-leucine were from Cambridge Isotope Laboratories (Andover, MA, United States). Anti-frataxin mouse monoclonal antibody [17A11] Ab113691 was from Abcam (Cambridge, MA, United States). Dimethyl pimelimidate dihydrochloride (DMP), Ethylenediaminetetraacetic acid (EDTA), cOmplete™ Mini EDTA-free Easypack protease inhibitor cocktail tablets, endoproteinase Asp-N sequencing grade, DL-dithiothreitol (DTT), bovine serum albumin (BSA), human lysozyme, imidazole, glycerol, phenylmethylsulfonyl fluoride (PMSF), triethanolamine, ethanolamine, and M9, minimal salts, 5X powder, minimal microbial growth medium (M9 media) were purchased from MilliporeSigma (Billerica, MA, United States). Ni-NTA agarose resin was purchased from Qiagen (Germantown, MD, United States). LC grade water and acetonitrile were from Burdick and Jackson (Muskegon, MI, United States). Ammonium bicarbonate and acetic acid were purchased from Fisher Scientific (Pittsburgh, PA, United States). Protein G magnetic beads were obtained from Life Technologies Corporation (Grand Island, NY, United States).

Blood samples were obtained from 11-unaffected healthy control subjects and 100-homozygous FRDA patients (Table 2) and 6-heterozygous FRDA patients (Table 3). A blood pool was made from 6 FRDA patients with > 800 GAA1 repeat (Supplementary Table 1). All were enrolled in parallel in an ongoing natural history study at the Children’s Hospital of Philadelphia (Lynch et al., 2006). Written informed consent was obtained from each donor participating in the study. If subjects were under the age of 18, written informed consent was obtained from a parent and/or legal guardian. The study was approved by the Institutional Review Board (IRB) of the Children Hospital of Philadelphia (IRB Protocol # 01–002609). Venous blood was drawn in 8.5 mL purple cap Vacutainer EDTA tubes and gently invert to mix. All samples were immediately aliquoted to Eppendorf tubes and frozen at −80^°C until analysis.

The expression of unlabeled and SILAC-labeled mature frataxin was performed in Escherichia coli BL21 DE3 as described previously (Guo et al., 2018b). Briefly, the coding sequence of human mature frataxin (81–210) was amplified from FXN cDNA plasmid (pTL1), then cloned into a pET21b plasmid and linked to the 6× histidine (His) sequence. The 6× His-tag fusion of frataxin was expressed in Escherichia coli BL21 DE3 in M9 media containing 1 mM MgSO₄, 10 μM CaCl₂, and 0.5% glucose with 100 mg/L ampicillin. For expressing unlabeled frataxin, the M9 medium was supplemented with 0.025% leucine and lysine. For expressing SILAC-labeled frataxin, the M9 medium was supplemented with 0.025% [¹³C₆¹⁵N₁]-leucine and [¹³C₆¹⁵N₂]-lysine. The cell pellets were collected and lysed in lysis buffer [50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 10 mM imidazole, 10% glycerol, 2 mM β-mercaptoethanol, 2× protease inhibitor mix, 1 mM PMSF] containing 100 μg/mL human lysozyme. The lysate was centrifuged at 20,000 × g for 30 min at 4^°C, and the supernatant was purified with Ni-NTA resin. The purity of the unlabeled mature frataxin and SILAC-labeled mature frataxin was confirmed to be > 95% by SDS-PAGE and Coomassie blue staining.

All blood samples were thawed at room temperature, and 500 μL of each sample was mixed with 750 μL NP-40 lysis buffer (150 mM NaCl, 50 mM Tris/HCl pH 7.5, 0.5% Triton X-100, 0.5% NP-40, 1 mM DTT, 1 mM EDTA) containing protease inhibitor cocktail. The same amount of SILAC-labeled mature frataxin (20 ng) was spiked in each sample (calibrator, QC and whole blood) as an internal standard. Samples were lysed by probe sonication on ice for 30 pulses at power 5 using a sonic dismembranator (Fisher, Pittsburgh, PA, United States), followed by centrifugation at 17,000 g for 15 min at 4^°C. The supernatant was transferred from the pellet and incubated with pre-made DMP-crosslinked anti-frataxin protein G beads for immunoprecipitation.

Mouse monoclonal anti-frataxin antibody (4 μg) was cross-linked to protein G beads (0.5 mg) through DMP as described previously (Guo et al., 2018b). Briefly, mouse monoclonal anti-frataxin antibody was firstly incubated with protein G beads overnight at 4^°C to form the antibody coupled protein G beads. The antibody coupled beads were incubated with 13 mg/mL DMP solution 1 h at room temperature to form the stable cross-linked anti-frataxin protein G beads. The cross-linked protein G beads can be kept in PBS at 4^°C for a week.

The processed whole blood samples (1.25 mL) were added into 0.5 mg anti-frataxin protein G beads to carry out immunoprecipitation at 4^°C overnight under rotary agitation. The beads were washed with PBS with 0.02% Tween-20 three times and frataxin and its isoforms were eluted with 100 μL of elution buffer (90% 100 mM acetic acid/10% acetonitrile). Elutes were transferred to deactivated glass inserts (Waters, Milford, MA, United States) and dried in a vacuum concentrator (Jouan RC 10.22, Fisher, Pittsburgh, PA, United States). Samples were dissolved in 50 μL of 25 mM aqueous ammonium bicarbonate containing 100 ng of Asp-N and incubated at 37 overnight before LC-MS analysis.

A 5% BSA and FRDA whole blood pool from six FRDA patients (Supplementary Table 1) were used for preparation of calibration standards and quality controls (QCs). Calibration standards and QCs were prepared in two types of matrix, 500 μL of 5% BSA and 500 μL of FRDA whole blood pool for the evaluation of matrix effect. In 5% BSA, calibration standards (0.5, 2.0, 4.0, 10.0, 20.0, 40.0, and 80.0 ng/mL) and QCs (1.5, 3.0, 30, and 60 ng/mL) were prepared from spiking in additional stock solutions of mature frataxin at 2, 0.2, or 0.02 ng/mL. In FRDA blood pool, calibration standards (baseline + 0, 2.0, 4.0, 10.0, 20.0, 40.0, and 80.0 ng/mL) and QCs (baseline + 0, 3.0, 30, and 60 ng/mL) were prepared from spiking in additional stock solutions of mature frataxin at 2, 0.2, or 0.02 ng/mL. The extra spiked volume was less than 90% of the matrix. The accuracy and precision were determined on four levels of QCs, lower limit of quantification (LLOQ, 1.5 ng/mL), low QC (LQC, 3.0 ng/mL), middle QC (MQC, 30 ng/mL), and high QC (HQC, 80 ng/mL) in 5% BSA. The QCs were prepared in pooled FRDA blood where the frataxin level had been determined. This value was subtracted from the values that were obtained. The intra-batch accuracy and precision were determined on QCs (n = 5) were performed and analyzed on the same day. The inter-batch accuracy and precision were determined on QCs (n = 3) were performed and analyzed on three different days. The acceptance criteria of ± 15% for calibration stands and QCs (± 20% at lowest QC) at protein level were applied.

Frataxin stability in whole blood at room temperature was evaluated in healthy control whole blood over a 24-h period. After blood sample collection, 500 μL of blood were aliquoted to eight Eppendorf tubes. The aliquots were kept at room temperature on the benchtop for 1, 2, 4, 6, 8, 12, and 24 h, then frozen at −80^°C until analysis. Room temperature stability was evaluated by comparing the incubated samples to freshly prepared sample (T = 0). The freeze-thaw stability was evaluated (−80°C/room temperature) with three replicates at MQC (n = 3, 30 ng/mL).

Mass spectrometry was conducted using a Q Exactive HF Orbitrap high resolution mass spectrometer coupled to Dionex Ultimate 3000 RSLCnano with capillary flowmeter chromatographic systems (Thermo Fisher Scientific, San Jose, CA, United States). The 2D system was setup as a pre-concentration mode which was composed of a 10-port valve, one nanopump and a micropump. The LC trapping column was an Acclaim PepMap C₁₈ cartridge (0.3 μm × 5 mm, 100 Å, Thermo Scientific) and the analytical column was a C₁₈ AQ capillary column with a 10 μm pulled tip (75 μm × 25 cm, 3 μm particle size; Columntip, New Haven, CT, United States). The 2D-nano-UHPLC system was controlled by Xcalibur software from the Q-Exactive mass spectrometer.

Loading solvent was water/acetonitrile (99.7:0.3; v/v) containing 0.2% formic acid. Solvent A was water/acetonitrile (99.5:0.5; v/v) containing 0.1% formic acid, and solvent B was acetonitrile/water (98:2, v/v) containing 0.1% formic acid. The valve stayed at loading position (1–2) with loading solvent on the trapping column at 10 μL/min for 4 min. Then the valve changed to injection position (1–10) at which the trapping column was connected with the analytical column, and samples were back-flushed into the analytical column. The valve was changed back to the loading position (1–2) at 50 min at which the trapping column was equilibrated with loading solvent and analytical column was equilibrated with 2% B. Samples were eluted with a linear gradient at a flow rate of 0.4 μL/min. The gradient on the analytical column was as follows: 2% B at the start, 5% B at 10 min, 20% B at 15 min, 60% B at 35 min, 80% B at 40–48 min, and 2% B at 48–60 min. Nanospray was conducted using Nanospray Flex™ ion source (Thermo Scientific). MS operating conditions were as follows: spray voltage 2,500 V, ion transfer capillary temperature 275°C, ion polarity positive, S-lens RF level 55, in-source CID 2.0 eV. Both full scan and PRM were used. The full scan parameters were resolution 60,000, AGC target 2 e5, maximum IT 80 ms, scan range 350–1200 m/z. The PRM parameters were resolution 60,000, AGC target 2 e5, maximum IT 80 ms, loop count 5, MSC count 4, isolation window 1.0 m/z, NCE 25. The PRMs were scheduled for 20.5 to 22.5 min for S⁸¹GTLGHPGSL⁹⁰, 24.0 to 28.0 min for D¹²⁴VSFGSGVLTVKLGG¹³⁸, 22.00–24.2 min for D¹⁶⁷WTGKNWVYSH¹⁷⁷, and 21.5–23.5 min for D¹⁹⁹LSSLAYSGK²⁰⁸.

Data analysis was performed using Skyline (MacCoss Laboratory, University of Washington, Seattle, WA, United States). The peak area ratio of each PRM transition for each unlabeled/light (L) peptide to labeled/heavy (H) peptide was calculated by the Skyline software and used for absolute quantification. The peptide ratios were calculated by the average L/H ratios of the three PRM transitions. The total frataxin levels were calculated from the average of the four selected quantitative peptides, the mature frataxin levels were calculated from SGTLGHPGSL peptide, and the frataxin isoform E levels were calculated by subtracting SGTLGHPGSL peptide from total frataxin. Statistical analyses were conducted using Prism 9 for macOS Version 9.3.1 (GraphPad Software, LLC.). Comparisons between groups were made using an unpaired two-tailed t-test. Correlations between GAA repeats and frataxin levels as well as between age of onset and frataxin levels were conducted using linear regression models.

There is an arginine (R⁷⁹) and a lysine (K⁸⁰) before serine (S⁸¹) in the full-length frataxin sequence. Therefore, trypsin would potentially digest other all frataxin isoforms to release the same N-terminal tryptic S⁸¹GTLGHPGSLDETTYER⁹⁷ peptide. In order to obtain the unique N-terminal peptides for differentiating mature frataxin and frataxin isoform E, Asp-N was chosen as the endoproteinase. Mature frataxin and frataxin isoform E release two different N-terminal peptides with Asp-N digestion, S⁸¹GTLGHPGSL⁹⁰ and Ac-M⁷⁶NLRKSGTLGHPGSL⁹⁰ (Figure 1). The expressed and purified mature frataxin (81–210) and SILAC-labeled mature frataxin were digested by Asp-N and the proteolytic peptides were optimized under full scan and PRM/MS modes. The uniqueness of the Asp-N peptides was verified by using the human proteome downloaded from NIST as the background proteome. Initially, seven potentially useful proteolytic peptides with length at 5–12 amino acids were selected, and their precursor and all product ions were assessed. The three most intense Asp-N digested peptides chosen for quantifying total frataxin levels were: D¹²⁴VSFGSGVLTVKLGG¹³⁸, D¹⁵⁷WTGKNWVYSH¹⁷⁷, and D¹⁹⁹LSSLAYSGK²⁰⁸ (Supplementary Table 2). The precursor ions exhibited 10–15-fold higher intensities than their corresponding product ions in the neat solution. However, the precursor peaks had interference peaks around or even not detected due to the high interfering background in the whole blood matrix. In contrast, product ions showed clean peaks as fewer interfering signals were present in the channels of product ions (Figure 3). Therefore, the three most intense product ions for corresponding peptides were used for quantifying both endogenous frataxin and SILAC-labeled mature frataxin in whole blood samples. Direct quantification of frataxin isoform E is not accurate since the oxidation of methionine happens during sample preparation step. Due to a variety of technical challenges (Arnesen, 2011; Aksnes et al., 2016), fully acetylated frataxin isoform E standard is not available. Therefore, isoform E levels were calculated by subtracting the amount of mature frataxin (determined from S⁸¹GTLGHPGSL⁹⁰) from total frataxin determined as the mean of amounts determined from the three common peptides (D¹²⁴VSFGSGVLTVKLGG¹³⁸, D¹⁵⁷WTGKNWVYSH¹⁷⁷, and D¹⁹⁹LSSLAYSGK²⁰⁸).

A 5% BSA and a whole blood pool from six FRDA patients (Supplementary Table 1) were used as a surrogate matrix and the biological matrix respectively in method validation. Calibration curves were constructed at six different concentrations ranging from 0.5 to 80.0 ng/mL in 5% BSA solution, and from baseline to baseline with spiked-in 80 ng/mL frataxin standard in the FRDA blood pool. Linear standard curves were obtained for each of the four peptides with r² values range between 0.9986 to 0.9998 in 5% BSA matrix and between 0.9865 to 0.9983 in FRDA blood pool (Supplementary Figure 1). The LLOQ was set at 1.5 ng/mL in 5% BSA, which is below the mean concentration of total frataxin reported in whole blood from healthy controls and FRDA patients determined previously (Table 1).

Method accuracy and precision were obtained for the analysis of four levels of QCs in whole blood for the LLOQ (1.5 ng/mL), LQC (3.0 ng/mL), MQC (30 ng/mL), and HQC (60 ng/mL). The intra-day precision (N = 5) for LLOQ, LQC, MQC, and HQC (Supplementary Table 3) and inter-day precision for LQC, MQC, and HQC (Supplementary Table 4) precision was better than ± 15% for total frataxin. The accuracy was better than 90–110%. Endogenous frataxin isoforms in the FRDA whole blood that was used to prepare QCs ere determined and subtracted from the determined amounts.

A matrix effect could alter the absolute responses of the analyte and internal standard but would not change the L/H ratio of the response. Calibration curves generated from 5% BSA and from FRDA whole blood were compared (Supplementary Figure 1 and Supplementary Table 5). Two sets of calibration curves were parallel to each other. However, the matrix effect did not affect assay accuracy and precision of the assay.

The frataxin levels in the whole blood were stable over a 24-h period at room temperature before sample preparation (Supplementary Table 6). The frataxin levels were the same after two freeze-thaw cycles in three replicates of whole blood (Supplementary Table 7). LQC, MQC, and HQC samples that were re-analyzed after 48 h standing in the autosampler (4^°C) gave essentially identical data to that obtained from the original analyses (data not shown).

Twelve FRDA whole blood samples were re-analyzed. The deviations of the first analysis and repeat analysis for each individual for mature frataxin, isoform E and total frataxin ranged from −26.6 to 18.7% (Supplementary Table 8). The mean value of the first analysis and repeat analysis for mature frataxin, frataxin isoform E and total frataxin was −2.2, −13.3, and −10.8%, respectively (Supplementary Table 8). This demonstrated that data were reproducible in the whole blood samples.

The demographic information for the homozygous and heterogenous FRDA patients enrolled in this study are summarized in Tables 2, 3. All frataxin levels were within the linear range of standard curves. The mean concentrations (± SD) of mature frataxin in healthy controls and homozygous FRDA cases in blood samples taken at year-1 were significantly different (p < 0.0001) at 7.5 ± 1.5 ng/mL and 2.1 ± 1.2 ng/mL, respectively (Figure 4A). The mean concentrations (± SD) of isoform E in healthy controls and homozygous FRDA cases in blood samples taken at year-1 were significantly different (p < 0.0001) at 26.8 ± 4.1 ng/mL and 4.7 ± 3.3 ng/mL, respectively (Figure 4B). The mean concentrations (± SD) of total frataxin in healthy controls and homozygous FRDA cases in blood samples taken at year-1 were significantly different (p < 0.001) at 34.2 ± 4.2 ng/mL and 6.8 ± 4.0 ng/mL, respectively (Figure 4C).