Measurement of Acylcarnitine Substrate to Product Ratios Specific to Biotin-Dependent Carboxylases Offers a Combination of Indicators of Biotin Status in Humans^,

Abstract

This work describes a novel liquid chromatography tandem MS (LC-MS/MS) method for the determination of ratios of acylcarnitines arising from acyl-CoA substrates and products that reflect metabolic disturbances caused by marginal biotin deficiency. The urinary ratios reflecting reduced activities of biotin-dependent enzymes include the following: 1) the ratio of 3-hydroxyisovalerylcarnitine : 3-methylglutarylcarnitine (3HIAc : MGc) for methylcrotonyl-CoA carboxylase; 2) the ratio of propionylcarnitine:methylmalonylcarnitine (Pc : MMc) for propionyl-CoA carboxylase (PCC); and 3) the ratio of acetylcarnitine : malonylcarnitine (Ac : Mc) for acetyl-CoA carboxylase. To demonstrate the suitability of the LC-MS/MS method for biomonitoring, we measured the 3 ratios for 7 healthy adults at various time points (d 0, 14, and 28) during the induction of marginal biotin through the consumption of egg white. The mean change in the Pc : MMc ratio relative to d 0 was 5.3-fold by d 14 (P = 0.0049) and 8.5-fold by d 28 (P = 0.0042). The mean change in the 3HIAc : MGc ratio was 2.8-fold by d 14 (P = 0.0022) and 3.8-fold by d 28 (P = 0.0001). The mean change in the Ac : Mc ratio was 2.9-fold by d 14 (P = 0.03) and 4.7-fold by d 28 (P = 0.02). The results suggest that simultaneous assessment of ratios of multiple biotin-dependent pathways offers insight into the complex metabolic disturbances caused by marginal biotin deficiency. We hypothesize that one or a combination of the ratios might be more sensitive or robust with respect to other nutrient deficiencies or confounding metabolic processes.

Introduction

Optimal micronutrient nutrition of large populations has been drawing increased attention. However, the limitations of nutrition assessment based on questionnaires and measurement of indices of single micronutrients are well recognized. These include inaccuracy and failure to account for individual variation of micronutrient bioavailability, metabolic requirement, and their interactions.

Biotin is an essential cofactor for 5 known mammalian carboxylases; marginal biotin deficiency causes reduced activity of several of these biotin-dependent carboxylases, which ultimately alters the biochemical flux through these pathways. For example, the reduced activity of one of these, methylcrotonyl-CoA carboxylase (MCC)⁵, causes increased urinary excretion of the organic acid 3-hydroxyisovaleric acid (3HIA), a validated indicator of biotin status in humans (1).

Traditionally, assessment of substrate:product ratios across enzymes that become rate limiting in dedicated metabolic pathways is an excellent way to determine functional enzyme deficiency. Unfortunately, the acyl-CoA substrates and products of the biotin-dependent enzymes are compartmentalized in the mitochondria and are not found in plasma or urine. Fortunately, blocks in acyl-CoA metabolism result in increased urinary excretion of the acylcarnitines corresponding to the accumulating substrates (2).

Here, we describe the development, analytical performance characteristics, and initial clinical investigation of a liquid chromatography tandem MS (LC-MS/MS) method for the simultaneous determination of urinary substrate:product ratios across the following 3 biotin-dependent carboxylases: 1) MCC (EC 6.4.1.4) is located in the mitochondria and converts 3-methylcrotonyl-CoA to 3-methyglutaconyl-CoA. Accumulating 3-methylcrotonyl-CoA is shunted to 3-hydroxyisovaleryl-CoA and transesterified to 3-hydroxyisovalerylcarnitine (3HIAc), which is transported out of the mitochondria by the acylcarnitine translocase and into the plasma and excreted in the urine. Increases and decreases in 3-methyglutaconyl-CoA are reflected in an analogous fashion by urinary 3-methylglutarylcarnitine (MGc). 2) Propionyl-CoA carboxylase (PCC; EC 6.4.1.3) is located in the mitochondria and converts propionyl-CoA to methylmalonyl-CoA. Accumulating propionyl-CoA is transesterified to propionylcarnitine (Pc). Changes in methylmalonyl-CoA are reflected in urinary methylmalonylcarnitine (MMc). 3) Both isoforms of acetyl-CoA carboxylase (ACC; EC 6.4.1.2.) convert acetyl-CoA to malonyl-CoA; the related acylcarnitines are acetylcarnitine (Ac) and malonylcarnitine (Mc).

Participants and Methods

Chemicals.

Optima LC-MS/MS grade water and formic acid (99% pure, Acros Organics) were purchased from Fisher Scientific. Acetyl-dl-carnitine hydrochloride reference material was purchased from Sigma-Aldrich. Mc (99% pure), MMc (85% pure), and MGc (95% pure) reference materials were purchased from ChemischeLaboratorien. 3HIAc (98% pure), [N-methyl-D3]-3-hydroxyisovalerylcarnitine ([²H₃]-3HIAc, 98% pure), and propionylcarnitine (98% pure) reference materials were all generous gifts from Cambridge Isotope Laboratories.

Preparation of analytical standards and urine samples.

Analytical standard stock solutions were prepared in water and immediately frozen at −70°C and stored until use. The human urine pool used to assess matrix effects was prepared by taking 40-mL aliquots from fresh, untimed urine samples collected from 6 healthy adult volunteers (4 female, 2 male). After pooling, the urine was thoroughly mixed, aliquots were subdivided into 15-mL tubes, immediately frozen at −20°C and stored until use. Prior to the sample preparation for analysis, all human urines were thawed, warmed to 60°C for 30 min, cooled to ambient temperature, and centrifuged at 3000 × g for 10 min to sediment urine precipitates as previously described (3). The urine supernatant was directly sampled without disturbing the precipitate pellet.

Clinical study design.

Analyses were performed on urine samples obtained from participants in a study previously reported (4–6). The Institutional Review Board for the University of Arkansas for Medical Sciences approved this study. Written informed consent was obtained from each participant at enrollment and consent was intermittently assessed throughout the study as part of the informed consent process.

To measure the effect of marginal biotin deficiency on the acylcarnitine substrate:product ratios for Pc:MMc, 3HIAc:MGc, and Ac:Mc, marginal asymptomatic biotin deficiency was induced in 7 healthy adults (3 women) by feeding a diet low in biotin and high in undenatured egg white for 28 d as previously described (4, 5). Leucine intake was controlled by the total protein intake, which was 2.0 g/(kg ⋅ d) with normal energy contribution from protein (∼14%). Dietary odd-chain fatty acids are one of the sources of metabolic flux in the PCC pathway but are a small fraction of total dietary fatty acids. The odd-chain fatty acid intake was not quantitated but was controlled by feeding the same diet from d 0 to 28 on a 3-d rotating schedule (4, 5). Prior to starting the study, all participants underwent a biotin-loading and washout procedure (4). Briefly, participants were supplemented with biotin at 30 μg/d for a period of 1 wk and then allowed to wash out for 2 additional weeks before the study began. Urine was collected from each participant for 24 h prior to initiating the egg white diet (study d 0) and again on study d 14 and 28. Biotin sufficiency on study d 0 and biotin deficiency on study d 14 and 28 was determined by the PCC activity assay in peripheral blood lymphocyte as previously described (6).

Quantitation by LC-MS/MS.

Urine samples were prepared at a 50-fold dilution by mixing 20 μL of sample and 980 μL of water and then vortexed for 10 s. Samples were placed in the autosampler and cooled to 5°C for the duration of analysis. A volume of 2 μL was injected onto a LC-MS/MS system composed of an Agilent 1290 Infinity LC system and an Agilent 6490 triple quadrupole mass spectrometer. Chromatographic separations were performed on an Agilent ZORBAX Eclipse Plus-C18, RRHD 2.1- × 100-mm, 1.8-μm analytical column.

The mobile phases were 0.01% formic acid and acetonitrile. The initial mobile phase composition was 0% acetonitrile at the time of injection and was held constant for 2 min. Then the percentage of acetonitrile was linearly increased to 95% for the next 4 min and decreased back to 0% after an additional 0.1 min. The column was reequilibrated over the next 3.9 min to complete the 10-min LC cycle. The flow rate was held constant at 200 μL/min throughout the injection cycle.

Retention times for acylcarnitines ranged between 1.26 and 4.06 min for all analyzed samples (Fig. 1). Ac, Pc, Mc, MMc, 3HIAc, and MGc were acquired in selected reaction monitoring mode monitoring the ion transitions of m/z 204 to 85, m/z 218 to 85, m/z 248 to 85, m/z 262 to 85, m/z 262 to 85, and m/z 290 to 85, respectively. The capillary voltage was 4000 V. The desolvation gas temperature and flow rate were 200°C and 11 L/min, respectively. Sheath gas temperature and flow rate were 100°C and 3 mL/min, respectively. Collisionally induced dissociation for all analytes was performed at a collision energy of 20 eV. The retention times for the acylcarnitines were monitored by injection of a mixture of the acylcarnitines over the course of quantitation of all samples; retention times did not exhibit a trend and were reproducible with a SD of <1%.

FIGURE 1.

Representative extracted ion chromatogram of 50-fold diluted urine samples. 3HIA, 3-hydroxyisovaleric acid; Mc, malonylcarnitine; MGc, 3-methylglutarylcarnitine; MMc, methylmalonylcarnitine; Pc, propionylcarnitine; PCC, propionyl-CoA carboxylase.

Assessment of urine matrix effect.

Matrix effects were assessed at the 50-fold dilution factor by determining the analytical recovery of [²H₃]-3HIAc at concentrations of 12.5, 25, 50, 75, 100, 150, and 200 nmol/L in urine samples and contemporaneously prepared aqueous standards. These measurements were performed on 3 nonsequential days. Urine matrix effects were also assessed at dilution factors of 5, 10, 25, and 50 for the endogenous levels of Ac, Pc, Mc, MMc, 3HIAc, and MGc present in the urine pool.

Assessment of linearity.

The linearity of aqueous standard curves consisting of 6 points for each of the acylcarnitines was measured to assess the dynamic range for this assay. The dynamic range for each curve was prepared as follows: Ac and MMc, 30–1000 nmol/L; Pc and 3HIAc, 3–100 nmol/L; Mc, 6–200 nmol/L; and MGc, 13–400 nmol/L. These concentrations were chosen to span the range of endogenous acylcarnitine concentrations measured in the 50-fold diluted urine samples from biotin-sufficient and -deficient participants.

Assessment of precision.

Intra- and inter-day precision was determined by 8 replicate analyses of the endogenous levels of acylcarnitines present in the 50-fold diluted urine pool both as individual analytes and as the substrate:product ratios of Pc:MMc, 3HIAc:MGc, and Ac:Mc.

Statistical analyses.

Differences in the mean urinary 3HIAc:MGc, Pc:MMc, and Ac:Mc ratios during progressive biotin deficiency were tested for significance by 1-way ANOVA with repeated measures; if found significant, Dunnett’s post hoc test was used to identify significant differences from d 0 using KaliedaGraph (version 4.1.1; Synergy Software) as previously described (7).

Results and Discussion

Initially, we determined that assessment of any urine matrix effects on the detection of the 6 acylcarnitines of interest would be problematic because of the presence of unknown quantities of the same endogenous acylcarnitines in urine. We avoided this problem by utilizing [²H₃]-3HIAc in diluted urine and water to determine potential matrix effects on the detection of acylcarnitines at the dilution factor of interest. Specifically, the urine pool described in the “Preparation of analytical standards and urine samples” section was assumed to contain a broad range of representative urine constituents found in a diverse population of participants. Samples were prepared by diluting an aliquot of the pooled urine 50-fold and adding [²H₃]-3HIAc at concentrations described in “Participants and Methods.” Signal responses of the various concentrations of [²H₃]-3HIAc from 12.5 to 200 nmol/L in the diluted urine pool relative to aqueous samples ranged between 100 and 103%; the mean of the CV (%CV) was 3% and all CV were <6%. These observations demonstrate that urine matrix effects on the determination of [²H₃]-3HIAc are negligible at the 50-fold dilution used in this method and suggest that the matrix effects for all 6 analyzed acylcarnitines will also be negligible at this dilution factor.

Urine matrix effects and ion suppression effects were further evaluated for the 6 endogenous acylcarnitines of interest present in the urine pool by analysis of the peak areas at 5-, 10-, 25-, and 50-fold dilution. Signal responses were normalized to 100% of the value at the 50-fold dilution and corrected for the dilution (Table 1). At the 25-fold dilution, the observed matrix effect was negligible; in contrast, at the smaller dilution, moderate matrix effects were observed. We conclude that ion suppression and other matrix effects of human urine are minimal at 25-fold or greater dilution. This dilution is feasible, because the Agilent 6590 triple quadruple mass spectrometer has sufficient sensitivity to allow quantitation over the range of interest (see below) for all 6 acylcarnitines at a 50-fold dilution. Of note, the use of substrate:product ratios rather than individual peak areas for the individual acylcarnitines compensated to a large degree for matrix effects yielding accurate ratios even at a 1:10 dilution and for some ratios at a 1:5 dilution (Table 2).

TABLE 1.

Endogenous acylcarnitine recoveries from pooled human urine at various dilutions¹

	Recovery
Urine dilution factor	MGc	3HIAc	MMc	Mc	Pc	Ac
	%
50	100	100	100	100	100	100
25	101	97	95	95	100	93
10	76	71	67	61	78	61
5	75	71	69	38	91	59

¹Ac, acetylcarnitine; 3HIAc, 3-hydroxyisovalerylcarnitine; Mc, malonylcarnitine; MGc, 3-methylglutarylcarnitine; MMc, methylmalonylcarnitine; Pc, propionylcarnitine.

TABLE 2.

Detectability of acylcarnitines ratios in pooled human urine at various dilutions¹

	Acylcarnitine peak area ratio
Urine dilution factor	3HIAc:MGc	Pc:MMc	Ac:Mc
50	0.78	0.33	23.43
25	0.75	0.35	22.98
10	0.73	0.38	23.23
5	0.74	0.44	36.34

¹Ac, acetylcarnitine; 3HIAc, 3-hydroxyisovalerylcarnitine; Mc, malonylcarnitine; MGc, 3-methylglutarylcarnitine; MMc, methylmalonylcarnitine; Pc, propionylcarnitine.

Standard curves for Ac, Pc, Mc, MMc, 3HIAc, and MGc were generated by linear regression analysis of the peak area compared with the concentration of the analyte in water. The r² for each were > 0.99. Linear regression parameters obtained from standard curves of the 6 acylcarnitines are summarized in Table 3.

TABLE 3.

Linear regression parameters obtained from the calibration curves of acylcarnitines in water¹

Analyte	Slope	Intercept	r2
Ac	308	913	0.99
Pc	493	60	0.99
Mc	126	−81	0.99
MMc	141	−342	0.99
3HIAc	846	949	0.99
MGc	284	158	1.00

¹Ac, acetylcarnitine; 3HIAc, 3-hydroxyisovalerylcarnitine; Mc, malonylcarnitine; MGc, 3-methylglutarylcarnitine; MMc, methylmalonylcarnitine; Pc, propionylcarnitine.

The assay precision, expressed as %CV, was determined by 8 replicate analyses of endogenous Ac, Pc, Mc, MMc, 3HIAc, and MGc in the 50-fold diluted urine pool (Table 4). The intra- and inter-day %CV for individual analytes and expressed ratios ranged between 0.4 and 8% (Table 5). These data provide evidence of adequate precision of this method for the individual acylcarnitines and the specified ratios for levels within the expected concentration range. We expect the precision of the method for acylcarnitine levels measured in marginally biotin-deficient individuals will be at least equivalent to previous indicators of biotin deficiency.

TABLE 4.

Summary of intra and inter-day precision of individual analyte and analyte ratios¹

	CV
Analyte:analyte ratio	Intra-day2	Inter-day3
	%
Ac	0.4	4.7
Pc	2.8	4.5
Mc	2.2	8.5
MMc	1.7	3.7
3HIAc	6.7	7
MGc	2	4.5
Ac:Mc	2.2	4.2
Pc:MMc	3.8	3.8
3HIAc”MGc	8.4	7.9

¹Ac, acetylcarnitine; 3HIAc, 3-hydroxyisovalerylcarnitine; Mc, malonylcarnitine; MGc, 3-methylglutarylcarnitine; MMc, methylmalonylcarnitine; Pc, propionylcarnitine.

²n = 8.

³n = 24.

TABLE 5.

Diagnostic sensitivity of indicators of marginal biotin deficiency¹

Indicator of biotin status	Abnormal on d 14	Abnormal on d 28
	%
Urinary 3HIAc:MGc	86	86
Urinary Ac:Mc	71	71
Urinary Pc:MMc	100	100
Lymphocyte PCC activity	86	86
Urinary 3HIA carnitine excretion	100	100
Urinary 3HIA excretion	86	86

¹Defined as the percentage of participants whose values were abnormal (consistent with marginal biotin deficiency) by d 14 and 28. Ac, acetylcarnitine; 3HIA, 3-hydroxyisovaleric acid; 3HIAc, 3-hydroxyisovalerylcarnitine; Mc, malonylcarnitine; MGc, 3-methylglutarylcarnitine; MMc, methylmalonylcarnitine; Pc, propionylcarnitine; PCC, propionyl-CoA carboxylase.

Application to marginal biotin deficiency.

To assess the suitability of the LC-MS/MS method for biomonitoring metabolic changes that are reflected in urine with the development of marginal biotin deficiency, we measured the acylcarnitine substrate:product ratios for specific biochemical pathways in urine from 7 healthy adults during the experimental induction of asymptomatic marginal biotin deficiency by egg white feeding (4, 5). The term marginal is used here as previously published (8, 9) to connote that the induced degree of biotin deficiency was so mild that no signs or symptoms of deficiency developed in any participant (3). Yet marginal deficiency is of considerable biomedical interest, because a similar degree of deficiency induced in pregnant mice causes cleft pallet and limb shortening defects in 100% of the pups (10, 11).

The 3 ratios tended to increase with time consuming the egg white diet relative to d 0 (Fig. 2A–C). When calculated for each participant relative to that person’s value on d 0, the mean change in the Pc:MMc ratio was 5.3-fold by d 14 (P = 0.0049) and 8.5-fold by d 28 (P = 0.0042). The mean change in the 3HIAc:MGc ratio was 2.8-fold by d 14 (P = 0.0022) and 3.8-fold by d 28 (P = 0.0001). The mean change in the individual Ac:Mc ratio was 2.9-fold by d 14 (P = 0.03) and 4.7-fold by d 28 (P = 0.02). For comparison, data are provided for the PCC activity measured in lymphocytes isolated from peripheral blood for each participant (Fig. 2D) and the diagnostic sensitivity at d 14 and d 28 (percent abnormal) is provided for these 4 indices plus 2 others (urinary 3HIA and 3HIAc) in Table 5. For interest, we included the data from a participant who admitted noncompliance on questioning at the end of the study (6). That participant's PCC activity did not decrease to less than the lower limit of normal and that participant's 3 ratios all remained within the normal range at d 14 but increased to greater than the upper limit of normal by d 28. One interpretation is that PCC activity will prove to be the most robust for detecting marginal biotin deficiency; an alternative interpretation is that the noncompliant participant did become biotin deficient, but to a lesser degree, and that the abnormal ratios reflect this subtle metabolic block. Because the biotin-dependent enzymes whose substrate:product ratios were examined here (MCC, PCC, and ACC-1 or ACC-2) are at the heart of intermediary metabolism, the fluxes down any one of these pathways are determined by multiple other processes (e.g., dietary intake of several macronutrients, protein turnover, demand for energy from fatty acyl-CoA, flux to energy storage under conditions of high substrate availability such as fed state, etc.). This complexity may be an important source of the large intra-individual variability. Characterization of the ratios in a larger number of participants and more diverse circumstances is needed to establish and validate the sensitivity, specificity, and overall suitability of these substrate:product ratios.

FIGURE 2.

Urinary 3HIAc:MGc (A), Ac:Mc (B), and Pc:MMc (C) ratios and lymphocyte PCC activity (D) in participants (n = 7) who consumed raw egg whites for 28 d. Bars depict group mean values. Differences of d 14 and 28 from d 0 were tested by 1-way ANOVA with repeated measures. Asterisks indicate different from d 0: *P < 0.03, **P < 0.005, ***P < 0.002, ****P < 0.0004. 3HIA, 3-hydroxyisovaleric acid; Mc, malonylcarnitine; MGc, 3-methylglutarylcarnitine; MMc, methylmalonylcarnitine; Pc, propionylcarnitine; PCC, propionyl-CoA carboxylase.

These data provide additional evidence for the effects of marginal biotin deficiency on the metabolic flux in pathways whose essential steps are catalyzed by 3 or 4 biotin-dependent carboxylases. These data also provide preliminary evidence that Ac:Mc, Pc:MMc, and 3HIAc:MGc ratios are biomarkers for marginal biotin deficiency in humans. The ratios each correlated negatively and significantly for 2 of the 3, but only weakly, with lymphocyte PCC activity. The 3 correlation coefficients were −0.57 for Pc:MMc (P = 0.007); −0.51 for 3HIAc:MGc (P = 0.018); and −0.42 for Ac:Mc (P = 0.061). The 3HIAc:MGc ratio correlated positively with urinary 3HIA; R = 0.86 (P < 0.001). For Pc:MMc, R was 0.49 (P = 0.028). For Ac:Mc, there was no correlation with urinary 3HIA; R = 0.02 (P = 0.929).

These ratios are of particular interest, because one or a combination might have advantages in assessing biotin status during pregnancy (8, 9, 12, 13). For example, urinary 3HIAc is a marker for marginal biotin deficiency, at least in healthy, carnitine-sufficient adults. However, production and excretion of urinary 3HIAc might be reduced by deficiency of carnitine that develops in pregnancy (14–16), thus masking biotin deficiency. In theory, determining these ratios might effectively normalize for differences in carnitine status. Moreover, determination of lymphocyte PCC activity is demanding with respect to sample acquisition and assay (3, 17). The unavoidable variability of timing of sample availability for population studies has the potential to produce artifactual decreases in absolute PCC activity, making PCC measurement problematic for human biomonitoring. In contrast, urinary acylcarnitine substrate:product ratios can be measured without time-consuming sample processing, internal standards, or calibration curves required for the individual acylcarnitines (1, 18, 19) by virtue of the fact that the peak area ratios of closely related molecules are being determined. Accordingly, this approach improves assay speed and reduces cost and is well suited for the high-throughput analyses needed for large population studies. Although the need for this LC-MS/MS currently prevents broad access, LC-MS/MS instrumentation is becoming more widely available and services are available commercially and at most major institutions (20).